diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/mgd_dit_core_only3.yaml b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/mgd_dit_core_only3.yaml new file mode 100644 index 0000000000000000000000000000000000000000..556f58eac01641ee22d6369f95fdeaa817202059 --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/mgd_dit_core_only3.yaml @@ -0,0 +1,30 @@ +experiment_name: mgd_sweep_from_trained_base +policy_type: groot_mgd +base_model_path: ${HOME}/groot_robocasa/robocasa_v2/MGD_Processing_line_Only/mgd_token_mask_ratio_${sweep.mgd_token_mask_ratio}/phase2 +output_root: ${HOME}/groot_robocasa/robocasa_v2/MGD_Processing_line_Only/dit_core_only +sweep: + mgd_token_mask_ratio: + - 0.3 +dataset: + dataset_soup: my_atomic26_human +training: + num_gpus: 2 + batch_size: 64 +model: null +phases: +- name: phase3_mgd_fm + max_steps: 30000 + save_steps: 0 + trainable: + preset: freeze_processing_line + tune_llm: false + tune_visual: false + tune_projector: true + tune_diffusion_model: true + losses: + mgd_enabled: false + mgd_fm_loss_weight: 1.0 + mgd_loss_weight: 0.0 + mgd_token_mask_ratio: ${sweep.mgd_token_mask_ratio} + mgd_use_cosine_loss: true + mgd_use_mse_loss: false diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/phase3/experiment_cfg/metadata.json b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/phase3/experiment_cfg/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..300eb070963337016e16465f50ba8351ea2eb25c --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/phase3/experiment_cfg/metadata.json @@ -0,0 +1,431 @@ +{ + "new_embodiment": { + "statistics": { + "state": { + "base_position": { + "max": [ + 7.3139495849609375, + 0.4876587688922882, + 0.7196521759033203 + ], + "min": [ + -4.94293737411499, + -6.890198230743408, + 0.6996827125549316 + ], + "mean": [ + 2.749288365724937, + -2.0455216239909983, + 0.700532551020533 + ], + "std": [ + 1.6786295897046262, + 1.312897969434244, + 0.00133751758906258 + ], + "q01": [ + -1.4241907881148037, + -5.1716547935950885, + 0.6999670898768614 + ], + "q99": [ + 6.198111545831555, + -0.5873531334104489, + 0.7038833014242587 + ] + }, + "base_rotation": { + "max": [ + 0.0, + 0.0, + 1.0, + 1.0 + ], + "min": [ + 0.0, + 0.0, + -1.0, + 0.0 + ], + "mean": [ + 0.0, + 0.0, + 0.23084999587450247, + 0.6238431088581847 + ], + "std": [ + 0.0, + 0.0, + 0.6556404371644047, + 0.3572252105732042 + ], + "q01": [ + 0.0, + 0.0, + -0.9999999999999999, + 1.7482354593273349e-06 + ], + "q99": [ + 0.0, + 0.0, + 0.9999999999999999, + 0.9999999999999999 + ] + }, + "end_effector_position_relative": { + "max": [ + 0.9014528393745422, + 0.8003877401351929, + 0.9829942584037781 + ], + "min": [ + -0.37471598386764526, + -0.8472502827644348, + -0.25070279836654663 + ], + "mean": [ + 0.28667332601863704, + -0.038315473170876704, + 0.45974658906777444 + ], + "std": [ + 0.17067058828305154, + 0.23569952037784195, + 0.21950702354181487 + ], + "q01": [ + 0.015172948963784228, + -0.44450502063220615, + 0.23314444103329338 + ], + "q99": [ + 0.5609069689572412, + 0.38454179128009547, + 0.7028102736473852 + ] + }, + "end_effector_rotation_relative": { + "max": [ + 0.9999998807907104, + 0.9984455108642578, + 0.9434727430343628, + 0.9062229990959167 + ], + "min": [ + -0.9999930262565613, + -0.9988337755203247, + -0.9618149995803833, + 2.1872274658107926e-07 + ], + "mean": [ + -0.25672297401336, + 0.024425335806687216, + -0.08740977164483847, + 0.16608739644018397 + ], + "std": [ + 0.7932585208392383, + 0.3102260195580416, + 0.3772472576315148, + 0.17451959300158773 + ], + "q01": [ + -0.99379006987882, + -0.5478102050038828, + -0.6101777952971636, + 0.002274085014134748 + ], + "q99": [ + 0.8804856038896288, + 0.5910611092873037, + 0.5216058042993562, + 0.5231965974012255 + ] + }, + "gripper_qpos": { + "max": [ + 0.055869169533252716, + 0.010916369967162609 + ], + "min": [ + -0.011436971835792065, + -0.05664053186774254 + ], + "mean": [ + 0.031651551516052555, + -0.03161481387920421 + ], + "std": [ + 0.013119769222217526, + 0.01306078225192402 + ], + "q01": [ + 0.0060999246446777336, + -0.04062601653964198 + ], + "q99": [ + 0.04054891621517283, + -0.0059163567113456345 + ] + } + }, + "action": { + "base_motion": { + "max": [ + 1.0, + 1.0, + 1.0, + 0.0 + ], + "min": [ + -1.0, + -1.0, + -1.0, + 0.0 + ], + "mean": [ + 0.008144896753718373, + -0.00018893135146078263, + -0.0008739062727221845, + 0.0 + ], + "std": [ + 0.11030045424364411, + 0.10148082570313594, + 0.08975698373467506, + 0.0 + ], + "q01": [ + -0.07287373067581518, + -0.09446994199569948, + -0.07702818259249572, + 0.0 + ], + "q99": [ + 0.04485485414407898, + 0.09626941605965177, + 0.07610665906122742, + 0.0 + ] + }, + "control_mode": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.9221974313425939 + ], + "std": [ + 0.38670194342612674 + ], + "q01": [ + -1.0 + ], + "q99": [ + -0.384693946883099 + ] + }, + "end_effector_position": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + 0.01976129923017491, + -0.020870631344490034, + -0.05993476517349181 + ], + "std": [ + 0.4347024990804053, + 0.4221662976569997, + 0.3845506843044066 + ], + "q01": [ + -0.8835792939767807, + -0.9280919251713778, + -0.783361071470956 + ], + "q99": [ + 0.7622084438449307, + 0.8625125252609077, + 0.7470974753250706 + ] + }, + "end_effector_rotation": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + 0.008089673068544335, + -0.026498254627927348, + 0.0029083551743059005 + ], + "std": [ + 0.11216734503733773, + 0.13206002613798629, + 0.12615870412692864 + ], + "q01": [ + -0.2814627972846672, + -0.4342386516115439, + -0.34489755628340574 + ], + "q99": [ + 0.31879353266939386, + 0.28411797683220563, + 0.3597718671669894 + ] + }, + "gripper_close": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.3824228874769045 + ], + "std": [ + 0.9240015096469786 + ], + "q01": [ + -1.0 + ], + "q99": [ + 0.5597272305813592 + ] + } + } + }, + "modalities": { + "video": { + "robot0_eye_in_hand": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + }, + "robot0_agentview_left": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + }, + "robot0_agentview_right": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + } + }, + "state": { + "base_position": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "base_rotation": { + "absolute": true, + "rotation_type": "quaternion", + "shape": [ + 4 + ], + "continuous": true + }, + "end_effector_position_relative": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "end_effector_rotation_relative": { + "absolute": true, + "rotation_type": "quaternion", + "shape": [ + 4 + ], + "continuous": true + }, + "gripper_qpos": { + "absolute": true, + "rotation_type": null, + "shape": [ + 2 + ], + "continuous": true + } + }, + "action": { + "base_motion": { + "absolute": true, + "rotation_type": null, + "shape": [ + 4 + ], + "continuous": true + }, + "control_mode": { + "absolute": true, + "rotation_type": null, + "shape": [ + 1 + ], + "continuous": true + }, + "end_effector_position": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "end_effector_rotation": { + "absolute": true, + "rotation_type": "axis_angle", + "shape": [ + 3 + ], + "continuous": true + }, + "gripper_close": { + "absolute": true, + "rotation_type": null, + "shape": [ + 1 + ], + "continuous": true + } + } + }, + "embodiment_tag": "new_embodiment" + } +} \ No newline at end of file diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.7/mgd_dit_core_only2.yaml b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.7/mgd_dit_core_only2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..d02e4e6dc817d5181a963964a3d2068a103f9266 --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.7/mgd_dit_core_only2.yaml @@ -0,0 +1,30 @@ +experiment_name: mgd_sweep_from_trained_base +policy_type: groot_mgd +base_model_path: ${HOME}/groot_robocasa/robocasa_v2/MGD_Processing_line_Only/mgd_token_mask_ratio_${sweep.mgd_token_mask_ratio}/phase2 +output_root: ${HOME}/groot_robocasa/robocasa_v2/MGD_Processing_line_Only/dit_core_only +sweep: + mgd_token_mask_ratio: + - 0.7 +dataset: + dataset_soup: my_atomic26_human +training: + num_gpus: 1 + batch_size: 128 +model: null +phases: +- name: phase3_mgd_fm + max_steps: 30000 + save_steps: 0 + trainable: + preset: freeze_processing_line + tune_llm: false + tune_visual: false + tune_projector: true + tune_diffusion_model: true + losses: + mgd_enabled: false + mgd_fm_loss_weight: 1.0 + mgd_loss_weight: 0.0 + mgd_token_mask_ratio: ${sweep.mgd_token_mask_ratio} + mgd_use_cosine_loss: true + mgd_use_mse_loss: false diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.7/phase3/model.safetensors.index.json b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.7/phase3/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..579ccf56b744c24c73f571cce697c12343122f56 --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.7/phase3/model.safetensors.index.json @@ -0,0 +1,916 @@ +{ + "metadata": { + "total_size": 7642572676 + }, + "weight_map": { + "action_head.action_decoder.layer1.W": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer1.b": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer2.W": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer2.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W1.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W1.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W2.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W2.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W3.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W3.b": "model-00002-of-00002.safetensors", + "action_head.future_tokens.weight": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_1.bias": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_1.weight": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_2.bias": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_2.weight": "model-00002-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_1.bias": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_1.weight": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_2.bias": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.position_embedding.weight": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer1.W": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer1.b": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer2.W": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer2.b": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vlln.bias": "model-00002-of-00002.safetensors", + "action_head.vlln.weight": "model-00002-of-00002.safetensors", + "action_target_projector.projector.bias": "model-00002-of-00002.safetensors", + "action_target_projector.projector.weight": "model-00002-of-00002.safetensors", + "backbone.eagle_model.language_model.lm_head.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.embed_tokens.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.mlp1.0.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.mlp1.0.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.patch_embedding.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.patch_embedding.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.position_embedding.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.in_proj_bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.in_proj_weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.layernorm.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.probe": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.post_layernorm.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.post_layernorm.weight": "model-00001-of-00002.safetensors", + "sequence_mgd_head.attn_score.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.attn_score.weight": "model-00002-of-00002.safetensors", + "sequence_mgd_head.out_proj.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.out_proj.weight": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.0.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.0.weight": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.2.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.2.weight": "model-00002-of-00002.safetensors" + } +} diff --git a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.7/resolved_config.yaml b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.7/resolved_config.yaml index b592e64975df122f9af92a62fa77cee9a0ba4cbd..e429054459622743fbd9487f055a8585ea30aab7 100644 --- a/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.7/resolved_config.yaml +++ b/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.7/resolved_config.yaml @@ -8,8 +8,8 @@ sweep: dataset: dataset_soup: my_atomic26_human training: - num_gpus: 2 - batch_size: 64 + num_gpus: 1 + batch_size: 128 model: null phases: - name: phase3_mgd_fm diff --git a/MGD_Processing_line_Only/dit_core_only/train.log b/MGD_Processing_line_Only/dit_core_only/train.log new file mode 100644 index 0000000000000000000000000000000000000000..e99ff940236fd8f2d1ca9ab80830ec08d1945db8 --- /dev/null +++ b/MGD_Processing_line_Only/dit_core_only/train.log @@ -0,0 +1,3354 @@ +[robosuite WARNING] No private macro file found! (macros.py:57) +[robosuite WARNING] It is recommended to use a private macro file (macros.py:58) +[robosuite WARNING] To setup, run: python /home/ext_minje/clvla/benchmarks/robocasa_v2/robosuite/robosuite/scripts/setup_macros.py (macros.py:59) +[robosuite WARNING] Could not import robosuite_models. Some robots may not be available. If you want to use these robots, please install robosuite_models from source (https://github.com/ARISE-Initiative/robosuite_models) or through pip install. (__init__.py:30) +[robosuite WARNING] Could not load the mink-based whole-body IK. Make sure you install related import properly (e.g. pip install mink==0.0.5), otherwise you will not be able to use the default IK controller setting for GR1 robot. (__init__.py:40) +/home/ext_minje/miniforge3/envs/robocasa/lib/python3.11/site-packages/albumentations/__init__.py:13: UserWarning: A new version of Albumentations is available: 2.0.8 (you have 1.4.18). Upgrade using: pip install -U albumentations. To disable automatic update checks, set the environment variable NO_ALBUMENTATIONS_UPDATE to 1. + check_for_updates() +`use_fast` is set to `True` but the image processor class does not have a fast version. Falling back to the slow version. + +***************************************** +Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +***************************************** +[robosuite WARNING] No private macro file found! (macros.py:57) +[robosuite WARNING] It is recommended to use a private macro file (macros.py:58) +[robosuite WARNING] To setup, run: python /home/ext_minje/clvla/benchmarks/robocasa_v2/robosuite/robosuite/scripts/setup_macros.py (macros.py:59) +[robosuite WARNING] No private macro file found! (macros.py:57) +[robosuite WARNING] It is recommended to use a private macro file (macros.py:58) +[robosuite WARNING] To setup, run: python /home/ext_minje/clvla/benchmarks/robocasa_v2/robosuite/robosuite/scripts/setup_macros.py (macros.py:59) +[robosuite WARNING] Could not import robosuite_models. Some robots may not be available. If you want to use these robots, please install robosuite_models from source (https://github.com/ARISE-Initiative/robosuite_models) or through pip install. (__init__.py:30) +[robosuite WARNING] Could not load the mink-based whole-body IK. Make sure you install related import properly (e.g. pip install mink==0.0.5), otherwise you will not be able to use the default IK controller setting for GR1 robot. (__init__.py:40) +[robosuite WARNING] Could not import robosuite_models. Some robots may not be available. If you want to use these robots, please install robosuite_models from source (https://github.com/ARISE-Initiative/robosuite_models) or through pip install. (__init__.py:30) +[robosuite WARNING] Could not load the mink-based whole-body IK. Make sure you install related import properly (e.g. pip install mink==0.0.5), otherwise you will not be able to use the default IK controller setting for GR1 robot. (__init__.py:40) +WARNING: mimicgen environments not imported since mimicgen is not installed! +WARNING: mimicgen environments not imported since mimicgen is not installed! +/home/ext_minje/miniforge3/envs/robocasa/lib/python3.11/site-packages/albumentations/__init__.py:13: UserWarning: A new version of Albumentations is available: 2.0.8 (you have 1.4.18). Upgrade using: pip install -U albumentations. To disable automatic update checks, set the environment variable NO_ALBUMENTATIONS_UPDATE to 1. + check_for_updates() +/home/ext_minje/miniforge3/envs/robocasa/lib/python3.11/site-packages/albumentations/__init__.py:13: UserWarning: A new version of Albumentations is available: 2.0.8 (you have 1.4.18). Upgrade using: pip install -U albumentations. To disable automatic update checks, set the environment variable NO_ALBUMENTATIONS_UPDATE to 1. + check_for_updates() +`use_fast` is set to `True` but the image processor class does not have a fast version. Falling back to the slow version. +`use_fast` is set to `True` but the image processor class does not have a fast version. Falling back to the slow version. + +================================================== +GR00T FINE-TUNING CONFIGURATION: +================================================== +config: /home/ext_minje/clvla/benchmarks/robocasa_v2/experiment_cfg/processing_line_only/mgd_dit_core_only3.yaml +dataset_soup: None +output_dir: /tmp/gr00t +output_root: None +data_config: panda_omron +batch_size: 64 +max_steps: 300000 +num_gpus: 2 +save_steps: 20000 +run_name: None +save_total_limit: 100 +seed: 42 +base_model_path: nvidia/GR00T-N1.5-3B +tune_llm: False +tune_visual: False +tune_projector: True +tune_diffusion_model: True +resume: False +learning_rate: 3e-05 +weight_decay: 1e-05 +warmup_ratio: 0.05 +lora_rank: 0 +lora_alpha: 16 +lora_dropout: 0.1 +lora_full_model: False +dataloader_num_workers: 8 +report_to: wandb +embodiment_tag: new_embodiment +video_backend: opencv +balance_dataset_weights: True +balance_trajectory_weights: True +ds_weights_alpha: 0.4 +================================================== + +Using 2 GPUs + +================================================================================ +Starting sweep branch: mgd_token_mask_ratio_0.3 +Sweep vars: {'mgd_token_mask_ratio': 0.3} +================================================================================ + +-------------------------------------------------------------------------------- +Running phase 1: phase3_mgd_fm +Policy type: groot_mgd +Base model path: /home/ext_minje/groot_robocasa/robocasa_v2/MGD_Processing_line_Only/mgd_token_mask_ratio_0.3/phase2 +Output dir: /home/ext_minje/groot_robocasa/robocasa_v2/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/phase3 +Trainable preset: freeze_processing_line +Policy overrides: {'mgd_enabled': False, 'mgd_fm_loss_weight': 1.0, 'mgd_loss_weight': 0.0, 'mgd_token_mask_ratio': 0.3, 'mgd_use_cosine_loss': True, 'mgd_use_mse_loss': False} +-------------------------------------------------------------------------------- + +[{'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/CloseBlenderLid/20250822/lerobot', 'horizon': 900, 'filter_key': '100_demos', 'task': 'CloseBlenderLid', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/CloseFridge/20250819/lerobot', 'horizon': 900, 'filter_key': '100_demos', 'task': 'CloseFridge', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/CloseToasterOvenDoor/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'CloseToasterOvenDoor', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/CoffeeSetupMug/20250819/lerobot', 'horizon': 600, 'filter_key': '100_demos', 'task': 'CoffeeSetupMug', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/NavigateKitchen/20250821/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'NavigateKitchen', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/OpenCabinet/20250819/lerobot', 'horizon': 1050, 'filter_key': '100_demos', 'task': 'OpenCabinet', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/OpenDrawer/20250819/lerobot', 'horizon': 750, 'filter_key': '100_demos', 'task': 'OpenDrawer', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/OpenStandMixerHead/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'OpenStandMixerHead', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceCounterToCabinet/20250819/lerobot', 'horizon': 750, 'filter_key': '100_demos', 'task': 'PickPlaceCounterToCabinet', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceCounterToStove/20250819/lerobot', 'horizon': 600, 'filter_key': '100_demos', 'task': 'PickPlaceCounterToStove', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceDrawerToCounter/20250820/lerobot', 'horizon': 750, 'filter_key': '100_demos', 'task': 'PickPlaceDrawerToCounter', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceSinkToCounter/20250819/lerobot', 'horizon': 900, 'filter_key': '100_demos', 'task': 'PickPlaceSinkToCounter', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceToasterToCounter/20250819/lerobot', 'horizon': 600, 'filter_key': '100_demos', 'task': 'PickPlaceToasterToCounter', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceCabinetToCounter/20250819/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'PickPlaceCabinetToCounter', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOnElectricKettle/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'TurnOnElectricKettle', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOnMicrowave/20250819/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'TurnOnMicrowave', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOffMicrowave/20250819/lerobot', 'horizon': 300, 'filter_key': '100_demos', 'task': 'TurnOffMicrowave', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/OpenElectricKettleLid/20250820/lerobot', 'horizon': 300, 'filter_key': '100_demos', 'task': 'OpenElectricKettleLid', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/SlideDishwasherRack/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'SlideDishwasherRack', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOnSinkFaucet/20250819/lerobot', 'horizon': 600, 'filter_key': '100_demos', 'task': 'TurnOnSinkFaucet', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnSinkSpout/20250820/lerobot', 'horizon': 300, 'filter_key': '100_demos', 'task': 'TurnSinkSpout', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOffSinkFaucet/20250819/lerobot', 'horizon': 300, 'filter_key': '100_demos', 'task': 'TurnOffSinkFaucet', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/AdjustWaterTemperature/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'AdjustWaterTemperature', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOffStove/20250819/lerobot', 'horizon': 750, 'filter_key': '100_demos', 'task': 'TurnOffStove', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOnStove/20250819/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'TurnOnStove', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/AdjustToasterOvenTemperature/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'AdjustToasterOvenTemperature', 'split': 'pretrain', 'source': 'human'}] +Using 100 subset demos for filter_key: 100_demos +/home/ext_minje/clvla/benchmarks/robocasa_v2/Isaac-GR00T/gr00t/data/transform/state_action.py:257: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.detach().clone() or sourceTensor.detach().clone().requires_grad_(True), rather than torch.tensor(sourceTensor). + self.statistics[key] = torch.tensor(value) +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos + +================================================== +GR00T FINE-TUNING CONFIGURATION: +================================================== +config: /home/ext_minje/clvla/benchmarks/robocasa_v2/experiment_cfg/processing_line_only/mgd_dit_core_only3.yaml +dataset_soup: None +output_dir: /tmp/gr00t +output_root: None +data_config: panda_omron +batch_size: 64 +max_steps: 300000 +num_gpus: 2 +save_steps: 20000 +run_name: None +save_total_limit: 100 +seed: 42 +base_model_path: nvidia/GR00T-N1.5-3B +tune_llm: False +tune_visual: False +tune_projector: True +tune_diffusion_model: True +resume: False +learning_rate: 3e-05 +weight_decay: 1e-05 +warmup_ratio: 0.05 +lora_rank: 0 +lora_alpha: 16 +lora_dropout: 0.1 +lora_full_model: False +dataloader_num_workers: 8 +report_to: wandb +embodiment_tag: new_embodiment +video_backend: opencv +balance_dataset_weights: True +balance_trajectory_weights: True +ds_weights_alpha: 0.4 +================================================== + +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +Using 100 subset demos for filter_key: 100_demos +Initialized dataset lerobot with EmbodimentTag.NEW_EMBODIMENT +dataset weights: [1. 0.88494669 0.76443286 0.83888607 0.73785464 1.00501644 + 0.80029956 0.66065525 0.83967491 0.83515002 0.95047759 0.86791692 + 0.88329477 0.78510653 0.64381843 0.67404766 0.69697218 0.60443684 + 0.78448103 0.83447486 0.61146215 0.64384058 0.79545025 0.93845856 + 0.75517122 0.7973985 ] +Loaded 26 datasets +Loading pretrained dual brain from /home/ext_minje/groot_robocasa/robocasa_v2/MGD_Processing_line_Only/mgd_token_mask_ratio_0.3/phase2 +Tune backbone vision tower: False +Tune backbone LLM: False +Tune action head projector: True +Tune action head DiT: True +Model not found or avail in the huggingface hub. Loading from local path: /home/ext_minje/groot_robocasa/robocasa_v2/MGD_Processing_line_Only/mgd_token_mask_ratio_0.3/phase2 +Using 2 GPUs + +================================================================================ +Starting sweep branch: mgd_token_mask_ratio_0.3 +Sweep vars: {'mgd_token_mask_ratio': 0.3} +================================================================================ + +-------------------------------------------------------------------------------- +Running phase 1: phase3_mgd_fm +Policy type: groot_mgd +Base model path: /home/ext_minje/groot_robocasa/robocasa_v2/MGD_Processing_line_Only/mgd_token_mask_ratio_0.3/phase2 +Output dir: /home/ext_minje/groot_robocasa/robocasa_v2/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/phase3 +Trainable preset: freeze_processing_line +Policy overrides: {'mgd_enabled': False, 'mgd_fm_loss_weight': 1.0, 'mgd_loss_weight': 0.0, 'mgd_token_mask_ratio': 0.3, 'mgd_use_cosine_loss': True, 'mgd_use_mse_loss': False} +-------------------------------------------------------------------------------- + +[{'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/CloseBlenderLid/20250822/lerobot', 'horizon': 900, 'filter_key': '100_demos', 'task': 'CloseBlenderLid', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/CloseFridge/20250819/lerobot', 'horizon': 900, 'filter_key': '100_demos', 'task': 'CloseFridge', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/CloseToasterOvenDoor/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'CloseToasterOvenDoor', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/CoffeeSetupMug/20250819/lerobot', 'horizon': 600, 'filter_key': '100_demos', 'task': 'CoffeeSetupMug', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/NavigateKitchen/20250821/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'NavigateKitchen', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/OpenCabinet/20250819/lerobot', 'horizon': 1050, 'filter_key': '100_demos', 'task': 'OpenCabinet', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/OpenDrawer/20250819/lerobot', 'horizon': 750, 'filter_key': '100_demos', 'task': 'OpenDrawer', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/OpenStandMixerHead/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'OpenStandMixerHead', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceCounterToCabinet/20250819/lerobot', 'horizon': 750, 'filter_key': '100_demos', 'task': 'PickPlaceCounterToCabinet', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceCounterToStove/20250819/lerobot', 'horizon': 600, 'filter_key': '100_demos', 'task': 'PickPlaceCounterToStove', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceDrawerToCounter/20250820/lerobot', 'horizon': 750, 'filter_key': '100_demos', 'task': 'PickPlaceDrawerToCounter', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceSinkToCounter/20250819/lerobot', 'horizon': 900, 'filter_key': '100_demos', 'task': 'PickPlaceSinkToCounter', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceToasterToCounter/20250819/lerobot', 'horizon': 600, 'filter_key': '100_demos', 'task': 'PickPlaceToasterToCounter', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/PickPlaceCabinetToCounter/20250819/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'PickPlaceCabinetToCounter', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOnElectricKettle/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'TurnOnElectricKettle', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOnMicrowave/20250819/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'TurnOnMicrowave', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOffMicrowave/20250819/lerobot', 'horizon': 300, 'filter_key': '100_demos', 'task': 'TurnOffMicrowave', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/OpenElectricKettleLid/20250820/lerobot', 'horizon': 300, 'filter_key': '100_demos', 'task': 'OpenElectricKettleLid', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/SlideDishwasherRack/20250820/lerobot', 'horizon': 450, 'filter_key': '100_demos', 'task': 'SlideDishwasherRack', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOnSinkFaucet/20250819/lerobot', 'horizon': 600, 'filter_key': '100_demos', 'task': 'TurnOnSinkFaucet', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnSinkSpout/20250820/lerobot', 'horizon': 300, 'filter_key': '100_demos', 'task': 'TurnSinkSpout', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/benchmarks/robocasa_v2/robocasa/datasets/v1.0/pretrain/atomic/TurnOffSinkFaucet/20250819/lerobot', 'horizon': 300, 'filter_key': '100_demos', 'task': 'TurnOffSinkFaucet', 'split': 'pretrain', 'source': 'human'}, {'path': '/home/ext_minje/clvla/b 0%| | 32/30000 [01:08<18:54:38, 2.27s/it] 0%| | 33/30000 [01:09<16:12:47, 1.95s/it] 0%| | 34/30000 [01:10<14:14:03, 1.71s/it] 0%| | 35/30000 [01:11<12:50:53, 1.54s/it] 0%| | 36/30000 [01:12<11:55:09, 1.43s/it] 0%| | 37/30000 [01:14<11:15:06, 1.35s/it] 0%| | 38/30000 [01:15<11:02:38, 1.33s/it] 0%| | 39/30000 [01:16<10:43:39, 1.29s/it] 0%| | 40/30000 [01:20<18:06:39, 2.18s/it] 0%| | 40/30000 [01:20<18:06:39, 2.18s/it] 0%| | 41/30000 [01:21<15:42:06, 1.89s/it] 0%| | 42/30000 [01:23<13:54:03, 1.67s/it] 0%| | 43/30000 [01:24<12:39:48, 1.52s/it] 0%| | 44/30000 [01:25<11:49:58, 1.42s/it] 0%| | 45/30000 [01:26<11:52:13, 1.43s/it] 0%| | 46/30000 [01:29<14:24:34, 1.73s/it] 0%| | 47/30000 [01:31<16:11:39, 1.95s/it] 0%| | 48/30000 [01:34<18:30:32, 2.22s/it] 0%| | 49/30000 [01:36<17:28:40, 2.10s/it] 0%| | 50/30000 [01:38<17:57:54, 2.16s/it] 0%| | 50/30000 [01:38<17:57:54, 2.16s/it] 0%| | 51/30000 [01:41<18:24:01, 2.21s/it] 0%| | 52/30000 [01:43<18:30:51, 2.23s/it] 0%| | 53/30000 [01:45<18:40:13, 2.24s/it] 0%| | 54/30000 [01:47<18:52:54, 2.27s/it] 0%| | 55/30000 [01:50<18:53:36, 2.27s/it] 0%| | 56/30000 [01:52<18:59:24, 2.28s/it] 0%| | 57/30000 [01:54<19:06:32, 2.30s/it] 0%| | 58/30000 [01:57<19:03:06, 2.29s/it] 0%| | 59/30000 [01:59<18:54:56, 2.27s/it] 0%| | 60/30000 [02:01<18:53:22, 2.27s/it] 0%| | 60/30000 [02:01<18:53:22, 2.27s/it] 0%| | 61/30000 [02:04<19:03:39, 2.29s/it] 0%| | 62/30000 [02:06<18:55:50, 2.28s/it] 0%| | 63/30000 [02:08<18:51:34, 2.27s/it] 0%| | 64/30000 [02:10<19:02:55, 2.29s/it] 0%| | 65/30000 [02:13<19:05:12, 2.30s/it] 0%| | 66/30000 [02:15<18:59:35, 2.28s/it] 0%| | 67/30000 [02:17<18:54:29, 2.27s/it] 0%| | 68/30000 [02:19<19:01:22, 2.29s/it] 0%| | 69/30000 [02:22<18:54:09, 2.27s/it] 0%| | 70/30000 [02:24<19:00:19, 2.29s/it] 0%| | 70/30000 [02:24<19:00:19, 2.29s/it] 0%| | 71/30000 [02:26<19:11:35, 2.31s/it] 0%| | 72/30000 [02:29<19:16:12, 2.32s/it] 0%| | 73/30000 [02:31<19:05:52, 2.30s/it] 0%| | 74/30000 [02:33<19:07:49, 2.30s/it] 0%| | 75/30000 [02:36<19:12:27, 2.31s/it] 0%| | 76/30000 [02:38<19:11:45, 2.31s/it] 0%| | 77/30000 [02:40<19:01:50, 2.29s/it] 0%| | 78/30000 [02:42<18:59:37, 2.29s/it] 0%| | 79/30000 [02:45<19:10:05, 2.31s/it] 0%| | 80/30000 [02:47<19:15:59, 2.32s/it] 0%| | 80/30000 [02:47<19:15:59, 2.32s/it] 0%| | 81/30000 [02:49<19:11:38, 2.31s/it] 0%| | 82/30000 [02:52<19:15:49, 2.32s/it] 0%| | 83/30000 [02:54<19:13:28, 2.31s/it] 0%| | 84/30000 [02:56<19:08:21, 2.30s/it] 0%| | 85/30000 [02:59<19:03:03, 2.29s/it] 0%| | 86/30000 [03:01<19:12:40, 2.31s/it] 0%| | 87/30000 [03:03<19:04:59, 2.30s/it] 0%| | 88/30000 [03:06<19:06:24, 2.30s/it] 0%| | 89/30000 [03:08<18:59:50, 2.29s/it] 0%| | 90/30000 [03:10<19:11:35, 2.31s/it] 0%| | 90/30000 [03:10<19:11:35, 2.31s/it] 0%| | 91/30000 [03:13<19:15:56, 2.32s/it] 0%| | 92/30000 [03:15<19:09:21, 2.31s/it] 0%| | 93/30000 [03:17<19:03:39, 2.29s/it] 0%| | 94/30000 [03:19<19:08:43, 2.30s/it] 0%| | 95/30000 [03:22<19:01:25, 2.29s/it] 0%| | 96/30000 [03:24<18:58:30, 2.28s/it] 0%| | 97/30000 [03:26<19:06:36, 2.30s/it] 0%| | 98/30000 [03:29<19:05:05, 2.30s/it] 0%| | 99/30000 [03:31<19:00:24, 2.29s/it] 0%| | 100/30000 [03:33<18:59:50, 2.29s/it] 0%| | 100/30000 [03:33<18:59:50, 2.29s/it] 0%| | 101/30000 [03:35<19:13:11, 2.31s/it] 0%| | 102/30000 [03:38<19:07:52, 2.30s/it] 0%| | 103/30000 [03:40<19:04:26, 2.30s/it] 0%| | 104/30000 [03:42<18:59:05, 2.29s/it] 0%| | 105/30000 [03:45<19:08:42, 2.31s/it] 0%| | 106/30000 [03:47<18:56:44, 2.28s/it] 0%| | 107/30000 [03:49<18:55:22, 2.28s/it] 0%| | 108/30000 [03:51<19:05:01, 2.30s/it] 0%| | 109/30000 [03:54<19:05:03, 2.30s/it] 0%| | 110/30000 [03:56<19:06:40, 2.30s/it] 0%| | 110/30000 [03:56<19:06:40, 2.30s/it] 0%| | 111/30000 [03:58<19:03:47, 2.30s/it] 0%| | 112/30000 [04:01<19:12:33, 2.31s/it] 0%| | 113/30000 [04:03<19:16:44, 2.32s/it] 0%| | 114/30000 [04:05<19:11:35, 2.31s/it] 0%| | 115/30000 [04:08<19:01:54, 2.29s/it] 0%| | 116/30000 [04:10<19:15:30, 2.32s/it] 0%| | 117/30000 [04:12<19:16:07, 2.32s/it] 0%| | 118/30000 [04:15<19:05:08, 2.30s/it] 0%| | 119/30000 [04:17<19:10:42, 2.31s/it] 0%| | 120/30000 [04:19<19:09:27, 2.31s/it] 0%| | 120/30000 [04:19<19:09:27, 2.31s/it] 0%| | 121/30000 [04:21<19:07:28, 2.30s/it] 0%| | 122/30000 [04:24<19:00:05, 2.29s/it] 0%| | 123/30000 [04:26<19:09:20, 2.31s/it] 0%| | 124/30000 [04:28<19:14:10, 2.32s/it] 0%| | 125/30000 [04:31<19:07:40, 2.30s/it] 0%| | 126/30000 [04:33<19:06:32, 2.30s/it] 0%| | 127/30000 [04:35<19:04:49, 2.30s/it] 0%| | 128/30000 [04:38<19:03:05, 2.30s/it] 0%| | 129/30000 [04:40<18:59:09, 2.29s/it] 0%| | 130/30000 [04:42<19:08:25, 2.31s/it] 0%| | 130/30000 [04:42<19:08:25, 2.31s/it] 0%| | 131/30000 [04:45<19:07:22, 2.30s/it] 0%| | 132/30000 [04:47<18:56:27, 2.28s/it] 0%| | 133/30000 [04:49<19:02:17, 2.29s/it] 0%| | 134/30000 [04:51<18:57:38, 2.29s/it] 0%| | 135/30000 [04:54<18:55:22, 2.28s/it] 0%| | 136/30000 [04:56<18:53:27, 2.28s/it] 0%| | 137/30000 [04:58<19:04:22, 2.30s/it] 0%| | 138/30000 [05:01<19:04:01, 2.30s/it] 0%| | 139/30000 [05:03<19:03:33, 2.30s/it] 0%| | 140/30000 [05:05<19:05:43, 2.30s/it] 0%| | 140/30000 [05:05<19:05:43, 2.30s/it] 0%| | 141/30000 [05:07<19:01:35, 2.29s/it] 0%| | 142/30000 [05:10<18:51:26, 2.27s/it] 0%| | 143/30000 [05:12<18:49:05, 2.27s/it] 0%| | 144/30000 [05:14<18:53:55, 2.28s/it] 0%| | 145/30000 [05:16<18:49:43, 2.27s/it] 0%| | 146/30000 [05:19<18:48:57, 2.27s/it] 0%| | 147/30000 [05:21<19:01:57, 2.30s/it] 0%| | 148/30000 [05:23<19:04:46, 2.30s/it] 0%| | 149/30000 [05:26<19:02:22, 2.30s/it] 0%| | 150/30000 [05:28<19:01:19, 2.29s/it] 0%| | 150/30000 [05:28<19:01:19, 2.29s/it] 1%| | 151/30000 [05:30<19:05:37, 2.30s/it] 1%| | 152/30000 [05:33<19:05:05, 2.30s/it] 1%| | 153/30000 [05:35<18:56:24, 2.28s/it] 1%| | 154/30000 [05:37<19:04:57, 2.30s/it] 1%| | 155/30000 [05:39<19:02:42, 2.30s/it] 1%| | 156/30000 [05:42<18:59:07, 2.29s/it] 1%| | 157/30000 [05:44<18:55:37, 2.28s/it] 1%| | 158/30000 [05:46<19:02:45, 2.30s/it] 1%| | 159/30000 [05:49<19:04:57, 2.30s/it] 1%| | 160/30000 [05:51<19:00:30, 2.29s/it] 1%| | 160/30000 [05:51<19:00:30, 2.29s/it] 1%| | 161/30000 [05:53<19:01:40, 2.30s/it] 1%| | 162/30000 [05:55<19:00:00, 2.29s/it] 1%| | 163/30000 [05:58<18:55:06, 2.28s/it] 1%| | 164/30000 [06:00<18:54:03, 2.28s/it] 1%| | 165/30000 [06:02<19:01:39, 2.30s/it] 1%| | 166/30000 [06:05<18:55:46, 2.28s/it] 1%| | 167/30000 [06:07<18:56:01, 2.28s/it] 1%| | 168/30000 [06:09<19:01:44, 2.30s/it] 1%| | 169/30000 [06:12<19:03:18, 2.30s/it] 1%| | 170/30000 [06:14<18:55:47, 2.28s/it] 1%| | 170/30000 [06:14<18:55:47, 2.28s/it] 1%| | 171/30000 [06:16<18:49:37, 2.27s/it] 1%| | 172/30000 [06:18<18:55:37, 2.28s/it] 1%| | 173/30000 [06:21<18:56:22, 2.29s/it] 1%| | 174/30000 [06:23<18:58:15, 2.29s/it] 1%| | 175/30000 [06:25<19:03:20, 2.30s/it] 1%| | 176/30000 [06:28<19:00:33, 2.29s/it] 1%| | 177/30000 [06:30<18:59:10, 2.29s/it] 1%| | 178/30000 [06:32<18:51:16, 2.28s/it] 1%| | 179/30000 [06:34<19:00:33, 2.29s/it] 1%| | 180/30000 [06:37<19:07:31, 2.31s/it] 1%| | 180/30000 [06:37<19:07:31, 2.31s/it] 1%| | 181/30000 [06:39<19:05:06, 2.30s/it] 1%| | 182/30000 [06:41<19:06:46, 2.31s/it] 1%| | 183/30000 [06:44<19:03:54, 2.30s/it] 1%| | 184/30000 [06:46<19:01:25, 2.30s/it] 1%| | 185/30000 [06:48<18:54:35, 2.28s/it] 1%| | 186/30000 [06:51<19:02:30, 2.30s/it] 1%| | 187/30000 [06:53<18:58:04, 2.29s/it] 1%| | 188/30000 [06:55<18:53:20, 2.28s/it] 1%| | 189/30000 [06:57<19:00:23, 2.30s/it] 1%| | 190/30000 [07:00<18:57:15, 2.29s/it] 1%| | 190/30000 [07:00<18:57:15, 2.29s/it] 1%| | 191/30000 [07:02<18:53:23, 2.28s/it] 1%| | 192/30000 [07:04<18:49:02, 2.27s/it] 1%| | 193/30000 [07:07<19:02:53, 2.30s/it] 1%| | 194/30000 [07:09<19:02:33, 2.30s/it] 1%| | 195/30000 [07:11<18:58:04, 2.29s/it] 1%| | 196/30000 [07:13<18:56:08, 2.29s/it] 1%| | 197/30000 [07:16<19:03:47, 2.30s/it] 1%| | 198/30000 [07:18<18:58:30, 2.29s/it] 1%| | 199/30000 [07:20<18:55:45, 2.29s/it] 1%| | 200/30000 [07:23<19:04:15, 2.30s/it] 1%| | 200/30000 [07:23<19:04:15, 2.30s/it] 1%| | 201/30000 [07:25<19:07:28, 2.31s/it] 1%| | 202/30000 [07:27<18:52:43, 2.28s/it] 1%| | 203/30000 [07:29<19:00:03, 2.30s/it] 1%| | 204/30000 [07:32<19:01:09, 2.30s/it] 1%| | 205/30000 [07:34<18:57:37, 2.29s/it] 1%| | 206/30000 [07:36<18:52:37, 2.28s/it] 1%| | 207/30000 [07:39<19:02:32, 2.30s/it] 1%| | 208/30000 [07:41<18:56:32, 2.29s/it] 1%| | 209/30000 [07:43<18:52:31, 2.28s/it] 1%| | 210/30000 [07:46<19:02:45, 2.30s/it] 1%| | 210/30000 [07:46<19:02:45, 2.30s/it] 1%| | 211/30000 [07:48<19:03:07, 2.30s/it] 1%| | 212/30000 [07:50<18:57:34, 2.29s/it] 1%| | 213/30000 [07:52<19:03:04, 2.30s/it] 1%| | 214/30000 [07:55<19:07:24, 2.31s/it] 1%| | 215/30000 [07:57<18:59:13, 2.29s/it] 1%| | 216/30000 [07:59<18:58:08, 2.29s/it] 1%| | 217/30000 [08:02<19:09:34, 2.32s/it] 1%| | 218/30000 [08:04<19:08:26, 2.31s/it] 1%| | 219/30000 [08:06<19:03:27, 2.30s/it] 1%| | 220/30000 [08:08<18:52:36, 2.28s/it] 1%| | 220/30000 [08:09<18:52:36, 2.28s/it] 1%| | 221/30000 [08:11<19:03:26, 2.30s/it] 1%| | 222/30000 [08:13<18:58:48, 2.29s/it] 1%| | 223/30000 [08:15<19:00:30, 2.30s/it] 1%| | 224/30000 [08:18<19:00:54, 2.30s/it] 1%| | 225/30000 [08:20<19:00:43, 2.30s/it] 1%| | 226/30000 [08:22<18:55:00, 2.29s/it] 1%| | 227/30000 [08:25<18:58:10, 2.29s/it] 1%| | 228/30000 [08:27<18:55:56, 2.29s/it] 1%| | 229/30000 [08:29<18:51:37, 2.28s/it] 1%| | 230/30000 [08:31<18:46:00, 2.27s/it] 1%| | 230/30000 [08:31<18:46:00, 2.27s/it] 1%| | 231/30000 [08:34<18:56:15, 2.29s/it] 1%| | 232/30000 [08:36<18:56:48, 2.29s/it] 1%| | 233/30000 [08:38<18:52:49, 2.28s/it] 1%| | 234/30000 [08:41<19:03:09, 2.30s/it] 1%| | 235/30000 [08:43<18:58:49, 2.30s/it] 1%| | 236/30000 [08:45<18:53:00, 2.28s/it] 1%| | 237/30000 [08:47<18:52:53, 2.28s/it] 1%| | 238/30000 [08:50<18:58:37, 2.30s/it] 1%| | 239/30000 [08:52<18:55:03, 2.29s/it] 1%| | 240/30000 [08:54<18:47:00, 2.27s/it] 1%| | 240/30000 [08:54<18:47:00, 2.27s/it] 1%| | 241/30000 [08:57<18:59:07, 2.30s/it] 1%| | 242/30000 [08:59<18:59:17, 2.30s/it] 1%| | 243/30000 [09:01<18:52:07, 2.28s/it] 1%| | 244/30000 [09:03<18:58:25, 2.30s/it] 1%| | 245/30000 [09:06<19:02:31, 2.30s/it] 1%| | 246/30000 [09:08<19:00:03, 2.30s/it] 1%| | 247/30000 [09:10<18:52:31, 2.28s/it] 1%| | 248/30000 [09:13<19:02:52, 2.30s/it] 1%| | 249/30000 [09:15<19:06:41, 2.31s/it] 1%| | 250/30000 [09:17<18:59:21, 2.30s/it] 1%| | 250/30000 [09:17<18:59:21, 2.30s/it] 1%| | 251/30000 [09:20<19:03:30, 2.31s/it] 1%| | 252/30000 [09:22<18:59:52, 2.30s/it] 1%| | 253/30000 [09:24<19:01:14, 2.30s/it] 1%| | 254/30000 [09:26<18:57:11, 2.29s/it] 1%| | 255/30000 [09:29<19:04:07, 2.31s/it] 1%| | 256/30000 [09:31<18:54:35, 2.29s/it] 1%| | 257/30000 [09:33<18:47:20, 2.27s/it] 1%| | 258/30000 [09:36<19:00:44, 2.30s/it] 1%| | 259/30000 [09:38<19:02:24, 2.30s/it] 1%| | 260/30000 [09:40<18:55:31, 2.29s/it] 1%| | 260/30000 [09:40<18:55:31, 2.29s/it] 1%| | 261/30000 [09:43<18:58:09, 2.30s/it] 1%| | 262/30000 [09:45<19:07:15, 2.31s/it] 1%| | 263/30000 [09:47<19:12:29, 2.33s/it] 1%| | 264/30000 [09:50<19:01:22, 2.30s/it] 1%| | 265/30000 [09:52<19:06:28, 2.31s/it] 1%| | 266/30000 [09:54<19:06:42, 2.31s/it] 1%| | 267/30000 [09:56<18:57:38, 2.30s/it] 1%| | 268/30000 [09:59<18:55:40, 2.29s/it] 1%| | 269/30000 [10:01<18:58:12, 2.30s/it] 1%| | 270/30000 [10:03<18:50:49, 2.28s/it] 1%| | 270/30000 [10:03<18:50:49, 2.28s/it] 1%| | 271/30000 [10:06<18:46:27, 2.27s/it] 1%| | 272/30000 [10:08<18:59:10, 2.30s/it] 1%| | 273/30000 [10:10<18:56:30, 2.29s/it] 1%| | 274/30000 [10:12<18:54:06, 2.29s/it] 1%| | 275/30000 [10:15<18:59:25, 2.30s/it] 1%| | 276/30000 [10:17<18:56:58, 2.30s/it] 1%| | 277/30000 [10:19<18:50:00, 2.28s/it] 1%| | 278/30000 [10:22<18:47:32, 2.28s/it] 1%| | 279/30000 [10:24<18:56:40, 2.29s/it] 1%| | 280/30000 [10:26<18:56:05, 2.29s/it] 1%| | 280/30000 [10:26<18:56:05, 2.29s/it] 1%| | 281/30000 [10:28<18:53:42, 2.29s/it] 1%| | 282/30000 [10:31<19:00:59, 2.30s/it] 1%| | 283/30000 [10:33<19:01:42, 2.31s/it] 1%| | 284/30000 [10:35<19:00:22, 2.30s/it] 1%| | 285/30000 [10:38<18:53:04, 2.29s/it] 1%| | 286/30000 [10:40<19:01:46, 2.31s/it] 1%| | 287/30000 [10:42<19:06:44, 2.32s/it] 1%| | 288/30000 [10:45<18:59:07, 2.30s/it] 1%| | 289/30000 [10:47<19:09:23, 2.32s/it] 1%| | 290/30000 [10:49<19:01:10, 2.30s/it] 1%| | 290/30000 [10:49<19:01:10, 2.30s/it] 1%| | 291/30000 [10:52<18:58:32, 2.30s/it] 1%| | 292/30000 [10:54<18:54:56, 2.29s/it] 1%| | 293/30000 [10:56<18:59:57, 2.30s/it] 1%| | 294/30000 [10:58<18:54:27, 2.29s/it] 1%| | 295/30000 [11:01<18:51:04, 2.28s/it] 1%| | 296/30000 [11:03<19:02:04, 2.31s/it] 1%| | 297/30000 [11:05<19:01:32, 2.31s/it] 1%| | 298/30000 [11:08<18:56:58, 2.30s/it] 1%| | 299/30000 [11:10<18:57:50, 2.30s/it] 1%| | 300/30000 [11:12<18:59:14, 2.30s/it] 1%| | 300/30000 [11:12<18:59:14, 2.30s/it] 1%| | 301/30000 [11:15<18:56:58, 2.30s/it] 1%| | 302/30000 [11:17<18:47:20, 2.28s/it] 1%| | 303/30000 [11:19<18:54:30, 2.29s/it] 1%| | 304/30000 [11:21<18:51:38, 2.29s/it] 1%| | 305/30000 [11:24<18:46:22, 2.28s/it] 1%| | 306/30000 [11:26<18:54:36, 2.29s/it] 1%| | 307/30000 [11:28<18:59:35, 2.30s/it] 1%| | 308/30000 [11:30<18:47:16, 2.28s/it] 1%| | 309/30000 [11:33<18:47:53, 2.28s/it] 1%| | 310/30000 [11:35<18:48:40, 2.28s/it] 1%| | 310/30000 [11:35<18:48:40, 2.28s/it] 1%| | 311/30000 [11:37<18:47:59, 2.28s/it] 1%| | 312/30000 [11:40<18:42:41, 2.27s/it] 1%| | 313/30000 [11:42<18:58:12, 2.30s/it] 1%| | 314/30000 [11:44<18:57:43, 2.30s/it] 1%| | 315/30000 [11:46<18:51:33, 2.29s/it] 1%| | 316/30000 [11:49<18:45:47, 2.28s/it] 1%| | 317/30000 [11:51<18:55:10, 2.29s/it] 1%| | 318/30000 [11:53<18:51:59, 2.29s/it] 1%| | 319/30000 [11:56<18:54:50, 2.29s/it] 1%| | 320/30000 [11:58<18:55:51, 2.30s/it] 1%| | 320/30000 [11:58<18:55:51, 2.30s/it] 1%| | 321/30000 [12:00<18:53:32, 2.29s/it] 1%| | 322/30000 [12:03<18:52:00, 2.29s/it] 1%| | 323/30000 [12:05<18:48:20, 2.28s/it] 1%| | 324/30000 [12:07<18:52:26, 2.29s/it] 1%| | 325/30000 [12:09<18:54:50, 2.29s/it] 1%| | 326/30000 [12:12<18:49:14, 2.28s/it] 1%| | 327/30000 [12:14<18:45:15, 2.28s/it] 1%| | 328/30000 [12:16<18:53:57, 2.29s/it] 1%| | 329/30000 [12:19<18:57:52, 2.30s/it] 1%| | 330/30000 [12:21<18:54:45, 2.29s/it] 1%| | 330/30000 [12:21<18:54:45, 2.29s/it] 1%| | 331/30000 [12:23<18:48:42, 2.28s/it] 1%| | 332/30000 [12:25<18:52:49, 2.29s/it] 1%| | 333/30000 [12:28<18:44:47, 2.27s/it] 1%| | 334/30000 [12:30<18:40:54, 2.27s/it] 1%| | 335/30000 [12:32<18:52:46, 2.29s/it] 1%| | 336/30000 [12:35<18:56:09, 2.30s/it] 1%| | 337/30000 [12:37<18:49:18, 2.28s/it] 1%| | 338/30000 [12:39<18:58:03, 2.30s/it] 1%| | 339/30000 [12:41<19:02:16, 2.31s/it] 1%| | 340/30000 [12:44<18:57:46, 2.30s/it] 1%| | 340/30000 [12:44<18:57:46, 2.30s/it] 1%| | 341/30000 [12:46<18:54:59, 2.30s/it] 1%| | 342/30000 [12:48<18:59:09, 2.30s/it] 1%| | 343/30000 [12:51<19:02:08, 2.31s/it] 1%| | 344/30000 [12:53<18:53:55, 2.29s/it] 1%| | 345/30000 [12:55<18:54:54, 2.30s/it] 1%| | 346/30000 [12:58<19:01:20, 2.31s/it] 1%| | 347/30000 [13:00<18:59:45, 2.31s/it] 1%| | 348/30000 [13:02<18:51:50, 2.29s/it] 1%| | 349/30000 [13:04<18:48:11, 2.28s/it] 1%| | 350/30000 [13:07<18:47:36, 2.28s/it] 1%| | 350/30000 [13:07<18:47:36, 2.28s/it] 1%| | 351/30000 [13:09<18:38:32, 2.26s/it] 1%| | 352/30000 [13:11<18:30:05, 2.25s/it] 1%| | 353/30000 [13:13<18:44:42, 2.28s/it] 1%| | 354/30000 [13:16<18:39:42, 2.27s/it] 1%| | 355/30000 [13:18<18:37:36, 2.26s/it] 1%| | 356/30000 [13:20<18:44:56, 2.28s/it] 1%| | 357/30000 [13:23<18:50:33, 2.29s/it] 1%| | 358/30000 [13:25<18:46:03, 2.28s/it] 1%| | 359/30000 [13:27<18:49:33, 2.29s/it] 1%| | 360/30000 [13:29<18:56:46, 2.30s/it] 1%| | 360/30000 [13:30<18:56:46, 2.30s/it] 1%| | 361/30000 [13:32<18:50:04, 2.29s/it] 1%| | 362/30000 [13:34<18:49:43, 2.29s/it] 1%| | 363/30000 [13:36<18:50:36, 2.29s/it] 1%| | 364/30000 [13:39<18:44:58, 2.28s/it] 1%| | 365/30000 [13:41<18:43:41, 2.28s/it] 1%| | 366/30000 [13:43<18:42:04, 2.27s/it] 1%| | 367/30000 [13:45<18:51:43, 2.29s/it] 1%| | 368/30000 [13:48<18:52:32, 2.29s/it] 1%| | 369/30000 [13:50<18:46:21, 2.28s/it] 1%| | 370/30000 [13:52<18:51:25, 2.29s/it] 1%| | 370/30000 [13:52<18:51:25, 2.29s/it] 1%| | 371/30000 [13:55<18:51:33, 2.29s/it] 1%| | 372/30000 [13:57<18:47:25, 2.28s/it] 1%| | 373/30000 [13:59<18:41:11, 2.27s/it] 1%| | 374/30000 [14:01<18:48:32, 2.29s/it] 1%|▏ | 375/30000 [14:04<18:43:39, 2.28s/it] 1%|▏ | 376/30000 [14:06<18:39:18, 2.27s/it] 1%|▏ | 377/30000 [14:08<18:45:45, 2.28s/it] 1%|▏ | 378/30000 [14:11<18:48:35, 2.29s/it] 1%|▏ | 379/30000 [14:13<18:45:04, 2.28s/it] 1%|▏ | 380/30000 [14:15<18:52:32, 2.29s/it] 1%|▏ | 380/30000 [14:15<18:52:32, 2.29s/it] 1%|▏ | 381/30000 [14:17<18:47:23, 2.28s/it] 1%|▏ | 382/30000 [14:20<18:48:54, 2.29s/it] 1%|▏ | 383/30000 [14:22<18:46:06, 2.28s/it] 1%|▏ | 384/30000 [14:24<18:55:03, 2.30s/it] 1%|▏ | 385/30000 [14:27<18:45:54, 2.28s/it] 1%|▏ | 386/30000 [14:29<18:36:37, 2.26s/it] 1%|▏ | 387/30000 [14:31<18:45:39, 2.28s/it] 1%|▏ | 388/30000 [14:33<18:42:30, 2.27s/it] 1%|▏ | 389/30000 [14:36<18:42:05, 2.27s/it] 1%|▏ | 390/30000 [14:38<18:53:22, 2.30s/it] 1%|▏ | 390/30000 [14:38<18:53:22, 2.30s/it] 1%|▏ | 391/30000 [14:40<18:58:56, 2.31s/it] 1%|▏ | 392/30000 [14:43<18:54:34, 2.30s/it] 1%|▏ | 393/30000 [14:45<19:00:27, 2.31s/it] 1%|▏ | 394/30000 [14:47<18:57:10, 2.30s/it] 1%|▏ | 395/30000 [14:49<18:51:03, 2.29s/it] 1%|▏ | 396/30000 [14:52<18:48:12, 2.29s/it] 1%|▏ | 397/30000 [14:54<18:55:47, 2.30s/it] 1%|▏ | 398/30000 [14:56<18:49:56, 2.29s/it] 1%|▏ | 399/30000 [14:59<18:48:13, 2.29s/it] 1%|▏ | 400/30000 [15:01<18:54:42, 2.30s/it] 1%|▏ | 400/30000 [15:01<18:54:42, 2.30s/it] 1%|▏ | 401/30000 [15:03<18:56:13, 2.30s/it] 1%|▏ | 402/30000 [15:06<18:54:55, 2.30s/it] 1%|▏ | 403/30000 [15:08<18:49:41, 2.29s/it] 1%|▏ | 404/30000 [15:10<18:55:36, 2.30s/it] 1%|▏ | 405/30000 [15:12<18:49:42, 2.29s/it] 1%|▏ | 406/30000 [15:15<18:42:31, 2.28s/it] 1%|▏ | 407/30000 [15:17<18:50:40, 2.29s/it] 1%|▏ | 408/30000 [15:19<18:55:37, 2.30s/it] 1%|▏ | 409/30000 [15:22<18:49:39, 2.29s/it] 1%|▏ | 410/30000 [15:24<18:53:23, 2.30s/it] 1%|▏ | 410/30000 [15:24<18:53:23, 2.30s/it] 1%|▏ | 411/30000 [15:26<18:55:38, 2.30s/it] 1%|▏ | 412/30000 [15:28<18:55:08, 2.30s/it] 1%|▏ | 413/30000 [15:31<18:51:06, 2.29s/it] 1%|▏ | 414/30000 [15:33<18:57:00, 2.31s/it] 1%|▏ | 415/30000 [15:35<19:01:54, 2.32s/it] 1%|▏ | 416/30000 [15:38<18:55:08, 2.30s/it] 1%|▏ | 417/30000 [15:40<19:00:24, 2.31s/it] 1%|▏ | 418/30000 [15:42<18:58:01, 2.31s/it] 1%|▏ | 419/30000 [15:45<18:52:42, 2.30s/it] 1%|▏ | 420/30000 [15:47<18:48:38, 2.29s/it] 1%|▏ | 420/30000 [15:47<18:48:38, 2.29s/it] 1%|▏ | 421/30000 [15:49<18:53:46, 2.30s/it] 1%|▏ | 422/30000 [15:51<18:48:12, 2.29s/it] 1%|▏ | 423/30000 [15:54<18:45:31, 2.28s/it] 1%|▏ | 424/30000 [15:56<18:58:23, 2.31s/it] 1%|▏ | 425/30000 [15:58<18:51:00, 2.29s/it] 1%|▏ | 426/30000 [16:01<18:47:26, 2.29s/it] 1%|▏ | 427/30000 [16:03<18:55:09, 2.30s/it] 1%|▏ | 428/30000 [16:05<18:55:01, 2.30s/it] 1%|▏ | 429/30000 [16:08<18:48:04, 2.29s/it] 1%|▏ | 430/30000 [16:10<18:58:59, 2.31s/it] 1%|▏ | 430/30000 [16:10<18:58:59, 2.31s/it] 1%|▏ | 431/30000 [16:12<19:00:36, 2.31s/it] 1%|▏ | 432/30000 [16:15<18:56:28, 2.31s/it] 1%|▏ | 433/30000 [16:17<18:53:58, 2.30s/it] 1%|▏ | 434/30000 [16:19<18:58:15, 2.31s/it] 1%|▏ | 435/30000 [16:21<18:57:36, 2.31s/it] 1%|▏ | 436/30000 [16:24<18:52:54, 2.30s/it] 1%|▏ | 437/30000 [16:26<18:58:00, 2.31s/it] 1%|▏ | 438/30000 [16:28<18:52:30, 2.30s/it] 1%|▏ | 439/30000 [16:31<18:51:40, 2.30s/it] 1%|▏ | 440/30000 [16:33<18:44:47, 2.28s/it] {'loss': 0.1196, 'grad_norm': 0.6391264796257019, 'learning_rate': 1.8e-07, 'flow_matching_loss': 0.11957852989435196, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.0} +{'loss': 0.109, 'grad_norm': 0.5143718719482422, 'learning_rate': 3.8e-07, 'flow_matching_loss': 0.10896031707525253, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.01} +{'loss': 0.1051, 'grad_norm': 0.5473178625106812, 'learning_rate': 5.800000000000001e-07, 'flow_matching_loss': 0.10506509095430375, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.01} +{'loss': 0.0942, 'grad_norm': 0.4443802237510681, 'learning_rate': 7.799999999999999e-07, 'flow_matching_loss': 0.09418077915906906, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.01} +{'loss': 0.0838, 'grad_norm': 0.334856241941452, 'learning_rate': 9.8e-07, 'flow_matching_loss': 0.08383444249629975, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.01} +{'loss': 0.0766, 'grad_norm': 0.26298630237579346, 'learning_rate': 1.18e-06, 'flow_matching_loss': 0.07655692845582962, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.02} +{'loss': 0.066, 'grad_norm': 0.2860617935657501, 'learning_rate': 1.38e-06, 'flow_matching_loss': 0.06602602563798428, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.02} +{'loss': 0.0629, 'grad_norm': 0.21859042346477509, 'learning_rate': 1.5800000000000001e-06, 'flow_matching_loss': 0.06293820664286613, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.02} +{'loss': 0.0528, 'grad_norm': 0.24048349261283875, 'learning_rate': 1.7800000000000001e-06, 'flow_matching_loss': 0.05282442457973957, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.03} +{'loss': 0.0503, 'grad_norm': 0.1738506406545639, 'learning_rate': 1.98e-06, 'flow_matching_loss': 0.05028784424066544, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.03} +{'loss': 0.0486, 'grad_norm': 0.2052169293165207, 'learning_rate': 2.1800000000000003e-06, 'flow_matching_loss': 0.048641151189804076, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.03} +{'loss': 0.049, 'grad_norm': 0.1721595823764801, 'learning_rate': 2.38e-06, 'flow_matching_loss': 0.04896783232688904, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.03} +{'loss': 0.0428, 'grad_norm': 0.1367134302854538, 'learning_rate': 2.58e-06, 'flow_matching_loss': 0.04284493736922741, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.04} +{'loss': 0.0397, 'grad_norm': 0.11662483215332031, 'learning_rate': 2.78e-06, 'flow_matching_loss': 0.039705104380846026, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.04} +{'loss': 0.0418, 'grad_norm': 0.16507379710674286, 'learning_rate': 2.98e-06, 'flow_matching_loss': 0.04180329218506813, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.04} +{'loss': 0.0367, 'grad_norm': 0.13730306923389435, 'learning_rate': 3.18e-06, 'flow_matching_loss': 0.03667588848620653, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.05} +{'loss': 0.0411, 'grad_norm': 0.17700107395648956, 'learning_rate': 3.38e-06, 'flow_matching_loss': 0.04107833988964558, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.05} +{'loss': 0.0378, 'grad_norm': 0.1482573300600052, 'learning_rate': 3.58e-06, 'flow_matching_loss': 0.037821179069578645, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.05} +{'loss': 0.0337, 'grad_norm': 0.11779645830392838, 'learning_rate': 3.7800000000000002e-06, 'flow_matching_loss': 0.033722754940390585, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.06} +{'loss': 0.0342, 'grad_norm': 0.1451074630022049, 'learning_rate': 3.98e-06, 'flow_matching_loss': 0.034226584061980245, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.06} +{'loss': 0.0343, 'grad_norm': 0.1266142725944519, 'learning_rate': 4.18e-06, 'flow_matching_loss': 0.034252094104886055, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.06} +{'loss': 0.035, 'grad_norm': 0.12142464518547058, 'learning_rate': 4.3799999999999996e-06, 'flow_matching_loss': 0.03504473008215427, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.06} +{'loss': 0.0315, 'grad_norm': 0.12539684772491455, 'learning_rate': 4.58e-06, 'flow_matching_loss': 0.03148159850388765, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.07} +{'loss': 0.0309, 'grad_norm': 0.0804213136434555, 'learning_rate': 4.78e-06, 'flow_matching_loss': 0.030874002538621426, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.07} +{'loss': 0.0313, 'grad_norm': 0.11944226920604706, 'learning_rate': 4.980000000000001e-06, 'flow_matching_loss': 0.03129255045205355, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.07} +{'loss': 0.03, 'grad_norm': 0.10760107636451721, 'learning_rate': 5.18e-06, 'flow_matching_loss': 0.02998856231570244, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.08} +{'loss': 0.032, 'grad_norm': 0.14510394632816315, 'learning_rate': 5.38e-06, 'flow_matching_loss': 0.031979494355618955, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.08} +{'loss': 0.0276, 'grad_norm': 0.0853445753455162, 'learning_rate': 5.58e-06, 'flow_matching_loss': 0.027600772306323053, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.08} +{'loss': 0.0297, 'grad_norm': 0.12685206532478333, 'learning_rate': 5.7800000000000006e-06, 'flow_matching_loss': 0.029718828946352006, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.08} +{'loss': 0.0289, 'grad_norm': 0.11411578953266144, 'learning_rate': 5.98e-06, 'flow_matching_loss': 0.028926638327538966, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.09} +{'loss': 0.0288, 'grad_norm': 0.10206004232168198, 'learning_rate': 6.18e-06, 'flow_matching_loss': 0.0288217443972826, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.09} +{'loss': 0.0291, 'grad_norm': 0.1455816924571991, 'learning_rate': 6.38e-06, 'flow_matching_loss': 0.02908130194991827, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.09} +{'loss': 0.029, 'grad_norm': 0.10206829011440277, 'learning_rate': 6.58e-06, 'flow_matching_loss': 0.028966113366186617, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.1} +{'loss': 0.0261, 'grad_norm': 0.1263730227947235, 'learning_rate': 6.78e-06, 'flow_matching_loss': 0.026145823672413825, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.1} +{'loss': 0.0251, 'grad_norm': 0.13555730879306793, 'learning_rate': 6.98e-06, 'flow_matching_loss': 0.02507718987762928, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.1} +{'loss': 0.0276, 'grad_norm': 0.08892174810171127, 'learning_rate': 7.180000000000001e-06, 'flow_matching_loss': 0.027567699924111365, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.1} +{'loss': 0.0273, 'grad_norm': 0.10770472139120102, 'learning_rate': 7.3800000000000005e-06, 'flow_matching_loss': 0.027296516858041287, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.11} +{'loss': 0.0282, 'grad_norm': 0.12191107124090195, 'learning_rate': 7.5799999999999994e-06, 'flow_matching_loss': 0.02820488251745701, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.11} +{'loss': 0.0283, 'grad_norm': 0.0956963449716568, 'learning_rate': 7.780000000000002e-06, 'flow_matching_loss': 0.028291431628167628, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.11} +{'loss': 0.0258, 'grad_norm': 0.13158927857875824, 'learning_rate': 7.98e-06, 'flow_matching_loss': 0.02576776556670666, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.12} +{'loss': 0.024, 'grad_norm': 0.10368754714727402, 'learning_rate': 8.18e-06, 'flow_matching_loss': 0.02399444915354252, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.12} +{'loss': 0.0251, 'grad_norm': 0.10552521049976349, 'learning_rate': 8.38e-06, 'flow_matching_loss': 0.025058373250067233, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.12} +{'loss': 0.0275, 'grad_norm': 0.13808134198188782, 'learning_rate': 8.58e-06, 'flow_matching_loss': 0.027486556395888328, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.13} +{'loss': 0.0276, 'grad_norm': 0.12872029840946198, 'learning_rate': 8.78e-06, 'flow_matching_loss': 0.027584309317171575, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.13} 1%|▏ | 440/30000 [16:33<18:44:47, 2.28s/it] 1%|▏ | 441/30000 [16:35<18:55:49, 2.31s/it] 1%|▏ | 442/30000 [16:37<18:46:24, 2.29s/it] 1%|▏ | 443/30000 [16:40<18:43:27, 2.28s/it] 1%|▏ | 444/30000 [16:42<18:51:08, 2.30s/it] 1%|▏ | 445/30000 [16:44<18:52:11, 2.30s/it] 1%|▏ | 446/30000 [16:47<18:52:14, 2.30s/it] 1%|▏ | 447/30000 [16:49<18:55:33, 2.31s/it] 1%|▏ | 448/30000 [16:51<18:52:13, 2.30s/it] 1%|▏ | 449/30000 [16:54<18:50:20, 2.30s/it] 2%|▏ | 450/30000 [16:56<18:46:22, 2.29s/it] 2%|▏ | 450/30000 [16:56<18:46:22, 2.29s/it] 2%|▏ | 451/30000 [16:58<18:58:28, 2.31s/it] 2%|▏ | 452/30000 [17:00<18:53:28, 2.30s/it] 2%|▏ | 453/30000 [17:03<18:50:18, 2.30s/it] 2%|▏ | 454/30000 [17:05<18:55:50, 2.31s/it] 2%|▏ | 455/30000 [17:07<18:56:47, 2.31s/it] 2%|▏ | 456/30000 [17:10<18:51:33, 2.30s/it] 2%|▏ | 457/30000 [17:12<18:56:54, 2.31s/it] 2%|▏ | 458/30000 [17:14<18:53:25, 2.30s/it] 2%|▏ | 459/30000 [17:17<18:44:42, 2.28s/it] 2%|▏ | 460/30000 [17:19<18:43:13, 2.28s/it] 2%|▏ | 460/30000 [17:19<18:43:13, 2.28s/it] 2%|▏ | 461/30000 [17:21<18:52:44, 2.30s/it] 2%|▏ | 462/30000 [17:23<18:53:33, 2.30s/it] 2%|▏ | 463/30000 [17:26<18:49:51, 2.30s/it] 2%|▏ | 464/30000 [17:28<18:55:10, 2.31s/it] 2%|▏ | 465/30000 [17:30<18:50:36, 2.30s/it] 2%|▏ | 466/30000 [17:33<18:45:13, 2.29s/it] 2%|▏ | 467/30000 [17:35<18:40:37, 2.28s/it] 2%|▏ | 468/30000 [17:37<18:49:08, 2.29s/it] 2%|▏ | 469/30000 [17:39<18:40:03, 2.28s/it] 2%|▏ | 470/30000 [17:42<18:40:45, 2.28s/it] 2%|▏ | 470/30000 [17:42<18:40:45, 2.28s/it] 2%|▏ | 471/30000 [17:44<18:53:35, 2.30s/it] 2%|▏ | 472/30000 [17:46<18:50:52, 2.30s/it] 2%|▏ | 473/30000 [17:49<18:49:02, 2.29s/it] 2%|▏ | 474/30000 [17:51<18:50:47, 2.30s/it] 2%|▏ | 475/30000 [17:53<18:42:23, 2.28s/it] 2%|▏ | 476/30000 [17:55<18:43:35, 2.28s/it] 2%|▏ | 477/30000 [17:58<18:53:18, 2.30s/it] 2%|▏ | 478/30000 [18:00<18:54:18, 2.31s/it] 2%|▏ | 479/30000 [18:02<18:54:49, 2.31s/it] 2%|▏ | 480/30000 [18:05<18:54:43, 2.31s/it] 2%|▏ | 480/30000 [18:05<18:54:43, 2.31s/it] 2%|▏ | 481/30000 [18:07<18:55:20, 2.31s/it] 2%|▏ | 482/30000 [18:09<18:46:54, 2.29s/it] 2%|▏ | 483/30000 [18:12<18:44:59, 2.29s/it] 2%|▏ | 484/30000 [18:14<18:54:08, 2.31s/it] 2%|▏ | 485/30000 [18:16<18:52:14, 2.30s/it] 2%|▏ | 486/30000 [18:19<18:52:53, 2.30s/it] 2%|▏ | 487/30000 [18:21<18:51:01, 2.30s/it] 2%|▏ | 488/30000 [18:23<18:55:09, 2.31s/it] 2%|▏ | 489/30000 [18:25<18:47:15, 2.29s/it] 2%|▏ | 490/30000 [18:28<18:52:59, 2.30s/it] 2%|▏ | 490/30000 [18:28<18:52:59, 2.30s/it] 2%|▏ | 491/30000 [18:30<18:51:48, 2.30s/it] 2%|▏ | 492/30000 [18:32<18:49:16, 2.30s/it] 2%|▏ | 493/30000 [18:35<18:52:34, 2.30s/it] 2%|▏ | 494/30000 [18:37<18:54:58, 2.31s/it] 2%|▏ | 495/30000 [18:39<18:49:41, 2.30s/it] 2%|▏ | 496/30000 [18:42<18:43:46, 2.29s/it] 2%|▏ | 497/30000 [18:44<18:46:31, 2.29s/it] 2%|▏ | 498/30000 [18:46<18:35:26, 2.27s/it] 2%|▏ | 499/30000 [18:48<18:36:30, 2.27s/it] 2%|▏ | 500/30000 [18:51<18:48:04, 2.29s/it] 2%|▏ | 500/30000 [18:51<18:48:04, 2.29s/it] 2%|▏ | 501/30000 [18:53<18:51:34, 2.30s/it] 2%|▏ | 502/30000 [18:55<18:38:07, 2.27s/it] 2%|▏ | 503/30000 [18:58<18:48:28, 2.30s/it] 2%|▏ | 504/30000 [19:00<18:48:38, 2.30s/it] 2%|▏ | 505/30000 [19:02<18:43:17, 2.29s/it] 2%|▏ | 506/30000 [19:04<18:48:27, 2.30s/it] 2%|▏ | 507/30000 [19:07<18:47:05, 2.29s/it] 2%|▏ | 508/30000 [19:09<18:44:35, 2.29s/it] 2%|▏ | 509/30000 [19:11<18:47:29, 2.29s/it] 2%|▏ | 510/30000 [19:14<18:49:20, 2.30s/it] 2%|▏ | 510/30000 [19:14<18:49:20, 2.30s/it] 2%|▏ | 511/30000 [19:16<18:42:42, 2.28s/it] 2%|▏ | 512/30000 [19:18<18:38:45, 2.28s/it] 2%|▏ | 513/30000 [19:20<18:44:32, 2.29s/it] 2%|▏ | 514/30000 [19:23<18:37:07, 2.27s/it] 2%|▏ | 515/30000 [19:25<18:38:17, 2.28s/it] 2%|▏ | 516/30000 [19:27<18:50:22, 2.30s/it] 2%|▏ | 517/30000 [19:30<18:54:41, 2.31s/it] 2%|▏ | 518/30000 [19:32<18:51:17, 2.30s/it] 2%|▏ | 519/30000 [19:34<18:57:56, 2.32s/it] 2%|▏ | 520/30000 [19:37<18:55:45, 2.31s/it] 2%|▏ | 520/30000 [19:37<18:55:45, 2.31s/it] 2%|▏ | 521/30000 [19:39<18:51:16, 2.30s/it] 2%|▏ | 522/30000 [19:41<18:55:15, 2.31s/it] 2%|▏ | 523/30000 [19:43<18:50:05, 2.30s/it] 2%|▏ | 524/30000 [19:46<18:45:37, 2.29s/it] 2%|▏ | 525/30000 [19:48<18:45:27, 2.29s/it] 2%|▏ | 526/30000 [19:50<18:44:47, 2.29s/it] 2%|▏ | 527/30000 [19:53<18:42:23, 2.28s/it] 2%|▏ | 528/30000 [19:55<18:41:45, 2.28s/it] 2%|▏ | 529/30000 [19:57<18:42:33, 2.29s/it] 2%|▏ | 530/30000 [19:59<18:35:21, 2.27s/it] 2%|▏ | 530/30000 [19:59<18:35:21, 2.27s/it] 2%|▏ | 531/30000 [20:02<18:38:59, 2.28s/it] 2%|▏ | 532/30000 [20:04<18:41:16, 2.28s/it] 2%|▏ | 533/30000 [20:06<18:37:49, 2.28s/it] 2%|▏ | 534/30000 [20:08<18:36:55, 2.27s/it] 2%|▏ | 535/30000 [20:11<18:43:48, 2.29s/it] 2%|▏ | 536/30000 [20:13<18:36:51, 2.27s/it] 2%|▏ | 537/30000 [20:15<18:34:09, 2.27s/it] 2%|▏ | 538/30000 [20:18<18:45:05, 2.29s/it] 2%|▏ | 539/30000 [20:20<18:42:53, 2.29s/it] 2%|▏ | 540/30000 [20:22<18:39:32, 2.28s/it] 2%|▏ | 540/30000 [20:22<18:39:32, 2.28s/it] 2%|▏ | 541/30000 [20:24<18:43:53, 2.29s/it] 2%|▏ | 542/30000 [20:27<18:42:37, 2.29s/it] 2%|▏ | 543/30000 [20:29<18:47:39, 2.30s/it] 2%|▏ | 544/30000 [20:31<18:54:02, 2.31s/it] 2%|▏ | 545/30000 [20:34<18:55:19, 2.31s/it] 2%|▏ | 546/30000 [20:36<18:49:08, 2.30s/it] 2%|▏ | 547/30000 [20:38<18:53:42, 2.31s/it] 2%|▏ | 548/30000 [20:41<18:53:00, 2.31s/it] 2%|▏ | 549/30000 [20:43<18:45:18, 2.29s/it] 2%|▏ | 550/30000 [20:45<18:42:38, 2.29s/it] 2%|▏ | 550/30000 [20:45<18:42:38, 2.29s/it] 2%|▏ | 551/30000 [20:48<18:48:14, 2.30s/it] 2%|▏ | 552/30000 [20:50<18:42:25, 2.29s/it] 2%|▏ | 553/30000 [20:52<18:41:28, 2.29s/it] 2%|▏ | 554/30000 [20:54<18:48:10, 2.30s/it] 2%|▏ | 555/30000 [20:57<18:41:44, 2.29s/it] 2%|▏ | 556/30000 [20:59<18:38:14, 2.28s/it] 2%|▏ | 557/30000 [21:01<18:46:11, 2.29s/it] 2%|▏ | 558/30000 [21:04<18:48:23, 2.30s/it] 2%|▏ | 559/30000 [21:06<18:42:19, 2.29s/it] 2%|▏ | 560/30000 [21:08<18:43:50, 2.29s/it] 2%|▏ | 560/30000 [21:08<18:43:50, 2.29s/it] 2%|▏ | 561/30000 [21:10<18:48:00, 2.30s/it] 2%|▏ | 562/30000 [21:13<18:43:25, 2.29s/it] 2%|▏ | 563/30000 [21:15<18:42:58, 2.29s/it] 2%|▏ | 564/30000 [21:17<18:42:54, 2.29s/it] 2%|▏ | 565/30000 [21:20<18:40:20, 2.28s/it] 2%|▏ | 566/30000 [21:22<18:38:25, 2.28s/it] 2%|▏ | 567/30000 [21:24<18:48:17, 2.30s/it] 2%|▏ | 568/30000 [21:26<18:40:27, 2.28s/it] 2%|▏ | 569/30000 [21:29<18:39:58, 2.28s/it] 2%|▏ | 570/30000 [21:31<18:50:34, 2.30s/it] 2%|▏ | 570/30000 [21:31<18:50:34, 2.30s/it] 2%|▏ | 571/30000 [21:33<18:56:04, 2.32s/it] 2%|▏ | 572/30000 [21:36<18:49:20, 2.30s/it] 2%|▏ | 573/30000 [21:38<18:51:32, 2.31s/it] 2%|▏ | 574/30000 [21:40<18:42:41, 2.29s/it] 2%|▏ | 575/30000 [21:42<18:33:38, 2.27s/it] 2%|▏ | 576/30000 [21:45<18:42:08, 2.29s/it] 2%|▏ | 577/30000 [21:47<18:41:12, 2.29s/it] 2%|▏ | 578/30000 [21:49<18:33:21, 2.27s/it] 2%|▏ | 579/30000 [21:52<18:42:43, 2.29s/it] 2%|▏ | 580/30000 [21:54<18:47:07, 2.30s/it] 2%|▏ | 580/30000 [21:54<18:47:07, 2.30s/it] 2%|▏ | 581/30000 [21:56<18:39:12, 2.28s/it] 2%|▏ | 582/30000 [21:58<18:33:56, 2.27s/it] 2%|▏ | 583/30000 [22:01<18:47:17, 2.30s/it] 2%|▏ | 584/30000 [22:03<18:51:23, 2.31s/it] 2%|▏ | 585/30000 [22:05<18:43:46, 2.29s/it] 2%|▏ | 586/30000 [22:08<18:45:55, 2.30s/it] 2%|▏ | 587/30000 [22:10<18:42:48, 2.29s/it] 2%|▏ | 588/30000 [22:12<18:37:05, 2.28s/it] 2%|▏ | 589/30000 [22:15<18:37:57, 2.28s/it] 2%|▏ | 590/30000 [22:17<18:41:53, 2.29s/it] 2%|▏ | 590/30000 [22:17<18:41:53, 2.29s/it] 2%|▏ | 591/30000 [22:19<18:39:13, 2.28s/it] 2%|▏ | 592/30000 [22:21<18:39:12, 2.28s/it] 2%|▏ | 593/30000 [22:24<18:48:29, 2.30s/it] 2%|▏ | 594/30000 [22:26<18:43:56, 2.29s/it] 2%|▏ | 595/30000 [22:28<18:45:22, 2.30s/it] 2%|▏ | 596/30000 [22:31<18:49:16, 2.30s/it] 2%|▏ | 597/30000 [22:33<18:47:23, 2.30s/it] 2%|▏ | 598/30000 [22:35<18:41:18, 2.29s/it] 2%|▏ | 599/30000 [22:37<18:40:29, 2.29s/it] 2%|▏ | 600/30000 [22:40<18:43:59, 2.29s/it] 2%|▏ | 600/30000 [22:40<18:43:59, 2.29s/it] 2%|▏ | 601/30000 [22:42<18:38:04, 2.28s/it] 2%|▏ | 602/30000 [22:44<18:38:49, 2.28s/it] 2%|▏ | 603/30000 [22:47<18:50:25, 2.31s/it] 2%|▏ | 604/30000 [22:49<18:49:34, 2.31s/it] 2%|▏ | 605/30000 [22:51<18:40:14, 2.29s/it] 2%|▏ | 606/30000 [22:54<18:43:57, 2.29s/it] 2%|▏ | 607/30000 [22:56<18:42:51, 2.29s/it] 2%|▏ | 608/30000 [22:58<18:41:03, 2.29s/it] 2%|▏ | 609/30000 [23:00<18:38:50, 2.28s/it] 2%|▏ | 610/30000 [23:03<18:48:47, 2.30s/it] 2%|▏ | 610/30000 [23:03<18:48:47, 2.30s/it] 2%|▏ | 611/30000 [23:05<18:49:47, 2.31s/it] 2%|▏ | 612/30000 [23:07<18:44:46, 2.30s/it] 2%|▏ | 613/30000 [23:10<18:52:56, 2.31s/it] 2%|▏ | 614/30000 [23:12<18:48:20, 2.30s/it] 2%|▏ | 615/30000 [23:14<18:41:35, 2.29s/it] 2%|▏ | 616/30000 [23:16<18:39:55, 2.29s/it] 2%|▏ | 617/30000 [23:19<18:46:43, 2.30s/it] 2%|▏ | 618/30000 [23:21<18:40:49, 2.29s/it] 2%|▏ | 619/30000 [23:23<18:37:20, 2.28s/it] 2%|▏ | 620/30000 [23:26<18:47:20, 2.30s/it] 2%|▏ | 620/30000 [23:26<18:47:20, 2.30s/it] 2%|▏ | 621/30000 [23:28<18:52:00, 2.31s/it] 2%|▏ | 622/30000 [23:30<18:44:52, 2.30s/it] 2%|▏ | 623/30000 [23:33<18:47:00, 2.30s/it] 2%|▏ | 624/30000 [23:35<18:45:14, 2.30s/it] 2%|▏ | 625/30000 [23:37<18:39:23, 2.29s/it] 2%|▏ | 626/30000 [23:39<18:45:05, 2.30s/it] 2%|▏ | 627/30000 [23:42<18:39:06, 2.29s/it] 2%|▏ | 628/30000 [23:44<18:35:39, 2.28s/it] 2%|▏ | 629/30000 [23:46<18:35:27, 2.28s/it] 2%|▏ | 630/30000 [23:49<18:41:40, 2.29s/it] 2%|▏ | 630/30000 [23:49<18:41:40, 2.29s/it] 2%|▏ | 631/30000 [23:51<18:46:09, 2.30s/it] 2%|▏ | 632/30000 [23:53<18:44:40, 2.30s/it] 2%|▏ | 633/30000 [23:56<18:51:15, 2.31s/it] 2%|▏ | 634/30000 [23:58<18:45:08, 2.30s/it] 2%|▏ | 635/30000 [24:00<18:40:15, 2.29s/it] 2%|▏ | 636/30000 [24:02<18:40:49, 2.29s/it] 2%|▏ | 637/30000 [24:05<18:38:23, 2.29s/it] 2%|▏ | 638/30000 [24:07<18:36:25, 2.28s/it] 2%|▏ | 639/30000 [24:09<18:34:50, 2.28s/it] 2%|▏ | 640/30000 [24:12<18:47:22, 2.30s/it] 2%|▏ | 640/30000 [24:12<18:47:22, 2.30s/it] 2%|▏ | 641/30000 [24:14<18:45:22, 2.30s/it] 2%|▏ | 642/30000 [24:16<18:42:32, 2.29s/it] 2%|▏ | 643/30000 [24:18<18:49:39, 2.31s/it] 2%|▏ | 644/30000 [24:21<18:47:23, 2.30s/it] 2%|▏ | 645/30000 [24:23<18:41:56, 2.29s/it] 2%|▏ | 646/30000 [24:25<18:37:25, 2.28s/it] 2%|▏ | 647/30000 [24:28<18:42:03, 2.29s/it] 2%|▏ | 648/30000 [24:30<18:39:17, 2.29s/it] 2%|▏ | 649/30000 [24:32<18:36:49, 2.28s/it] 2%|▏ | 650/30000 [24:34<18:46:22, 2.30s/it] 2%|▏ | 650/30000 [24:35<18:46:22, 2.30s/it] 2%|▏ | 651/30000 [24:37<18:41:19, 2.29s/it] 2%|▏ | 652/30000 [24:39<18:39:14, 2.29s/it] 2%|▏ | 653/30000 [24:41<18:46:07, 2.30s/it] 2%|▏ | 654/30000 [24:44<18:49:43, 2.31s/it] 2%|▏ | 655/30000 [24:46<18:43:26, 2.30s/it] 2%|▏ | 656/30000 [24:48<18:44:08, 2.30s/it] 2%|▏ | 657/30000 [24:51<18:40:06, 2.29s/it] 2%|▏ | 658/30000 [24:53<18:35:39, 2.28s/it] 2%|▏ | 659/30000 [24:55<18:41:12, 2.29s/it] 2%|▏ | 660/30000 [24:57<18:43:28, 2.30s/it] 2%|▏ | 660/30000 [24:57<18:43:28, 2.30s/it] 2%|▏ | 661/30000 [25:00<18:38:26, 2.29s/it] 2%|▏ | 662/30000 [25:02<18:39:23, 2.29s/it] 2%|▏ | 663/30000 [25:04<18:44:19, 2.30s/it] 2%|▏ | 664/30000 [25:07<18:36:24, 2.28s/it] 2%|▏ | 665/30000 [25:09<18:35:54, 2.28s/it] 2%|▏ | 666/30000 [25:11<18:44:24, 2.30s/it] 2%|▏ | 667/30000 [25:14<18:49:52, 2.31s/it] 2%|▏ | 668/30000 [25:16<18:47:23, 2.31s/it] 2%|▏ | 669/30000 [25:18<18:52:00, 2.32s/it] 2%|▏ | 670/30000 [25:20<18:46:29, 2.30s/it] 2%|▏ | 670/30000 [25:20<18:46:29, 2.30s/it] 2%|▏ | 671/30000 [25:23<18:46:48, 2.31s/it] 2%|▏ | 672/30000 [25:25<18:41:43, 2.29s/it] 2%|▏ | 673/30000 [25:27<18:46:10, 2.30s/it] 2%|▏ | 674/30000 [25:30<18:38:25, 2.29s/it] 2%|▏ | 675/30000 [25:32<18:36:35, 2.28s/it] 2%|▏ | 676/30000 [25:34<18:41:51, 2.30s/it] 2%|▏ | 677/30000 [25:36<18:41:14, 2.29s/it] 2%|▏ | 678/30000 [25:39<18:33:20, 2.28s/it] 2%|▏ | 679/30000 [25:41<18:35:18, 2.28s/it] 2%|▏ | 680/30000 [25:43<18:35:46, 2.28s/it] 2%|▏ | 680/30000 [25:43<18:35:46, 2.28s/it] 2%|▏ | 681/30000 [25:46<18:30:11, 2.27s/it] 2%|▏ | 682/30000 [25:48<18:30:14, 2.27s/it] 2%|▏ | 683/30000 [25:50<18:38:27, 2.29s/it] 2%|▏ | 684/30000 [25:52<18:30:02, 2.27s/it] 2%|▏ | 685/30000 [25:55<18:26:38, 2.27s/it] 2%|▏ | 686/30000 [25:57<18:35:24, 2.28s/it] 2%|▏ | 687/30000 [25:59<18:43:13, 2.30s/it] 2%|▏ | 688/30000 [26:02<18:37:54, 2.29s/it] 2%|▏ | 689/30000 [26:04<18:41:38, 2.30s/it] 2%|▏ | 690/30000 [26:06<18:38:38, 2.29s/it] 2%|▏ | 690/30000 [26:06<18:38:38, 2.29s/it] 2%|▏ | 691/30000 [26:08<18:37:48, 2.29s/it] 2%|▏ | 692/30000 [26:11<18:36:12, 2.29s/it] 2%|▏ | 693/30000 [26:13<18:45:18, 2.30s/it] 2%|▏ | 694/30000 [26:15<18:38:10, 2.29s/it] 2%|▏ | 695/30000 [26:18<18:37:23, 2.29s/it] 2%|▏ | 696/30000 [26:20<18:45:44, 2.30s/it] 2%|▏ | 697/30000 [26:22<18:44:59, 2.30s/it] 2%|▏ | 698/30000 [26:25<18:41:51, 2.30s/it] 2%|▏ | 699/30000 [26:27<18:34:51, 2.28s/it] 2%|▏ | 700/30000 [26:29<18:44:29, 2.30s/it] 2%|▏ | 700/30000 [26:29<18:44:29, 2.30s/it] 2%|▏ | 701/30000 [26:31<18:46:48, 2.31s/it] 2%|▏ | 702/30000 [26:34<18:37:16, 2.29s/it] 2%|▏ | 703/30000 [26:36<18:47:58, 2.31s/it] 2%|▏ | 704/30000 [26:38<18:42:59, 2.30s/it] 2%|▏ | 705/30000 [26:41<18:34:20, 2.28s/it] 2%|▏ | 706/30000 [26:43<18:38:03, 2.29s/it] 2%|▏ | 707/30000 [26:45<18:46:49, 2.31s/it] 2%|▏ | 708/30000 [26:47<18:36:36, 2.29s/it] 2%|▏ | 709/30000 [26:50<18:35:05, 2.28s/it] 2%|▏ | 710/30000 [26:52<18:45:22, 2.31s/it] 2%|▏ | 710/30000 [26:52<18:45:22, 2.31s/it] 2%|▏ | 711/30000 [26:54<18:47:32, 2.31s/it] 2%|▏ | 712/30000 [26:57<18:40:40, 2.30s/it] 2%|▏ | 713/30000 [26:59<18:38:13, 2.29s/it] 2%|▏ | 714/30000 [27:01<18:47:45, 2.31s/it] 2%|▏ | 715/30000 [27:04<18:41:53, 2.30s/it] 2%|▏ | 716/30000 [27:06<18:40:20, 2.30s/it] 2%|▏ | 717/30000 [27:08<18:41:53, 2.30s/it] 2%|▏ | 718/30000 [27:11<18:48:54, 2.31s/it] 2%|▏ | 719/30000 [27:13<18:41:04, 2.30s/it] 2%|▏ | 720/30000 [27:15<18:45:03, 2.31s/it] 2%|▏ | 720/30000 [27:15<18:45:03, 2.31s/it] 2%|▏ | 721/30000 [27:17<18:44:35, 2.30s/it] 2%|▏ | 722/30000 [27:20<18:38:33, 2.29s/it] 2%|▏ | 723/30000 [27:22<18:28:25, 2.27s/it] 2%|▏ | 724/30000 [27:24<18:34:01, 2.28s/it] 2%|▏ | 725/30000 [27:26<18:29:51, 2.27s/it] 2%|▏ | 726/30000 [27:29<18:26:45, 2.27s/it] 2%|▏ | 727/30000 [27:31<18:40:28, 2.30s/it] 2%|▏ | 728/30000 [27:33<18:43:50, 2.30s/it] 2%|▏ | 729/30000 [27:36<18:35:19, 2.29s/it] 2%|▏ | 730/30000 [27:38<18:36:26, 2.29s/it] 2%|▏ | 730/30000 [27:38<18:36:26, 2.29s/it] 2%|▏ | 731/30000 [27:40<18:32:08, 2.28s/it] 2%|▏ | 732/30000 [27:42<18:29:04, 2.27s/it] 2%|▏ | 733/30000 [27:45<18:30:09, 2.28s/it] 2%|▏ | 734/30000 [27:47<18:43:33, 2.30s/it] 2%|▏ | 735/30000 [27:49<18:34:56, 2.29s/it] 2%|▏ | 736/30000 [27:52<18:33:29, 2.28s/it] 2%|▏ | 737/30000 [27:54<18:43:37, 2.30s/it] 2%|▏ | 738/30000 [27:56<18:40:14, 2.30s/it] 2%|▏ | 739/30000 [27:59<18:34:12, 2.28s/it] 2%|▏ | 740/30000 [28:01<18:27:54, 2.27s/it] 2%|▏ | 740/30000 [28:01<18:27:54, 2.27s/it] 2%|▏ | 741/30000 [28:03<18:37:35, 2.29s/it] 2%|▏ | 742/30000 [28:05<18:33:32, 2.28s/it] 2%|▏ | 743/30000 [28:08<18:34:21, 2.29s/it] 2%|▏ | 744/30000 [28:10<18:43:11, 2.30s/it] 2%|▏ | 745/30000 [28:12<18:45:10, 2.31s/it] 2%|▏ | 746/30000 [28:15<18:38:08, 2.29s/it] 2%|▏ | 747/30000 [28:17<18:35:06, 2.29s/it] 2%|▏ | 748/30000 [28:19<18:41:53, 2.30s/it] 2%|▏ | 749/30000 [28:21<18:41:57, 2.30s/it] 2%|▎ | 750/30000 [28:24<18:39:06, 2.30s/it] 2%|▎ | 750/30000 [28:24<18:39:06, 2.30s/it] 3%|▎ | 751/30000 [28:26<18:43:07, 2.30s/it] 3%|▎ | 752/30000 [28:28<18:32:33, 2.28s/it] 3%|▎ | 753/30000 [28:31<18:32:07, 2.28s/it] 3%|▎ | 754/30000 [28:33<18:37:00, 2.29s/it] 3%|▎ | 755/30000 [28:35<18:32:18, 2.28s/it] 3%|▎ | 756/30000 [28:37<18:34:17, 2.29s/it] 3%|▎ | 757/30000 [28:40<18:26:34, 2.27s/it] 3%|▎ | 758/30000 [28:42<18:30:40, 2.28s/it] 3%|▎ | 759/30000 [28:44<18:32:28, 2.28s/it] 3%|▎ | 760/30000 [28:47<18:30:54, 2.28s/it] 3%|▎ | 760/30000 [28:47<18:30:54, 2.28s/it] 3%|▎ | 761/30000 [28:49<18:37:28, 2.29s/it] 3%|▎ | 762/30000 [28:51<18:36:28, 2.29s/it] 3%|▎ | 763/30000 [28:53<18:33:06, 2.28s/it] 3%|▎ | 764/30000 [28:56<18:29:48, 2.28s/it] 3%|▎ | 765/30000 [28:58<18:37:17, 2.29s/it] 3%|▎ | 766/30000 [29:00<18:33:04, 2.28s/it] 3%|▎ | 767/30000 [29:03<18:28:59, 2.28s/it] 3%|▎ | 768/30000 [29:05<18:39:12, 2.30s/it] 3%|▎ | 769/30000 [29:07<18:36:01, 2.29s/it] 3%|▎ | 770/30000 [29:09<18:33:34, 2.29s/it] 3%|▎ | 770/30000 [29:09<18:33:34, 2.29s/it] 3%|▎ | 771/30000 [29:12<18:29:16, 2.28s/it] 3%|▎ | 772/30000 [29:14<18:38:15, 2.30s/it] 3%|▎ | 773/30000 [29:16<18:37:39, 2.29s/it] 3%|▎ | 774/30000 [29:19<18:33:53, 2.29s/it] 3%|▎ | 775/30000 [29:21<18:33:49, 2.29s/it] 3%|▎ | 776/30000 [29:23<18:36:08, 2.29s/it] 3%|▎ | 777/30000 [29:26<18:39:30, 2.30s/it] 3%|▎ | 778/30000 [29:28<18:41:56, 2.30s/it] 3%|▎ | 779/30000 [29:30<18:38:22, 2.30s/it] 3%|▎ | 780/30000 [29:32<18:37:43, 2.30s/it] 3%|▎ | 780/30000 [29:32<18:37:43, 2.30s/it] 3%|▎ | 781/30000 [29:35<18:32:59, 2.29s/it] 3%|▎ | 782/30000 [29:37<18:28:42, 2.28s/it] 3%|▎ | 783/30000 [29:39<18:34:18, 2.29s/it] 3%|▎ | 784/30000 [29:42<18:33:11, 2.29s/it] 3%|▎ | 785/30000 [29:44<18:33:18, 2.29s/it] 3%|▎ | 786/30000 [29:46<18:30:19, 2.28s/it] 3%|▎ | 787/30000 [29:48<18:34:49, 2.29s/it] 3%|▎ | 788/30000 [29:51<18:37:58, 2.30s/it] 3%|▎ | 789/30000 [29:53<18:30:00, 2.28s/it] 3%|▎ | 790/30000 [29:55<18:30:40, 2.28s/it] 3%|▎ | 790/30000 [29:55<18:30:40, 2.28s/it] 3%|▎ | 791/30000 [29:58<18:33:03, 2.29s/it] 3%|▎ | 792/30000 [30:00<18:29:27, 2.28s/it] 3%|▎ | 793/30000 [30:02<18:30:32, 2.28s/it] 3%|▎ | 794/30000 [30:04<18:36:49, 2.29s/it] 3%|▎ | 795/30000 [30:07<18:33:00, 2.29s/it] 3%|▎ | 796/30000 [30:09<18:26:54, 2.27s/it] 3%|▎ | 797/30000 [30:11<18:29:34, 2.28s/it] 3%|▎ | 798/30000 [30:14<18:41:31, 2.30s/it] 3%|▎ | 799/30000 [30:16<18:41:51, 2.31s/it] 3%|▎ | 800/30000 [30:18<18:35:21, 2.29s/it] 3%|▎ | 800/30000 [30:18<18:35:21, 2.29s/it] 3%|▎ | 801/30000 [30:20<18:35:28, 2.29s/it] 3%|▎ | 802/30000 [30:23<18:38:07, 2.30s/it] 3%|▎ | 803/30000 [30:25<18:32:09, 2.29s/it] 3%|▎ | 804/30000 [30:27<18:32:01, 2.29s/it] 3%|▎ | 805/30000 [30:30<18:32:24, 2.29s/it] 3%|▎ | 806/30000 [30:32<18:31:47, 2.28s/it] 3%|▎ | 807/30000 [30:34<18:32:46, 2.29s/it] 3%|▎ | 808/30000 [30:36<18:38:15, 2.30s/it] 3%|▎ | 809/30000 [30:39<18:43:57, 2.31s/it] 3%|▎ | 810/30000 [30:41<18:39:40, 2.30s/it] 3%|▎ | 810/30000 [30:41<18:39:40, 2.30s/it] 3%|▎ | 811/30000 [30:43<18:38:02, 2.30s/it] 3%|▎ | 812/30000 [30:46<18:37:02, 2.30s/it] 3%|▎ | 813/30000 [30:48<18:30:53, 2.28s/it] 3%|▎ | 814/30000 [30:50<18:28:19, 2.28s/it] 3%|▎ | 815/30000 [30:53<18:45:07, 2.31s/it] 3%|▎ | 816/30000 [30:55<18:43:07, 2.31s/it] 3%|▎ | 817/30000 [30:57<18:40:21, 2.30s/it] 3%|▎ | 818/30000 [30:59<18:40:18, 2.30s/it] 3%|▎ | 819/30000 [31:02<18:45:47, 2.31s/it] 3%|▎ | 820/30000 [31:04<18:34:01, 2.29s/it] 3%|▎ | 820/30000 [31:04<18:34:01, 2.29s/it] 3%|▎ | 821/30000 [31:06<18:30:34, 2.28s/it] 3%|▎ | 822/30000 [31:09<18:36:36, 2.30s/it] 3%|▎ | 823/30000 [31:11<18:39:59, 2.30s/it] 3%|▎ | 824/30000 [31:13<18:35:12, 2.29s/it] 3%|▎ | 825/30000 [31:15<18:33:26, 2.29s/it] 3%|▎ | 826/30000 [31:18<18:42:01, 2.31s/it] 3%|▎ | 827/30000 [31:20<18:42:18, 2.31s/it] 3%|▎ | 828/30000 [31:22<18:36:53, 2.30s/it] 3%|▎ | 829/30000 [31:25<18:34:44, 2.29s/it] 3%|▎ | 830/30000 [31:27<18:37:15, 2.30s/it] 3%|▎ | 830/30000 [31:27<18:37:15, 2.30s/it] 3%|▎ | 831/30000 [31:29<18:37:28, 2.30s/it] 3%|▎ | 832/30000 [31:32<18:30:51, 2.29s/it] 3%|▎ | 833/30000 [31:34<18:35:39, 2.30s/it] 3%|▎ | 834/30000 [31:36<18:39:40, 2.30s/it] 3%|▎ | 835/30000 [31:39<18:37:15, 2.30s/it] 3%|▎ | 836/30000 [31:41<18:31:34, 2.29s/it] 3%|▎ | 837/30000 [31:43<18:38:05, 2.30s/it] 3%|▎ | 838/30000 [31:45<18:33:20, 2.29s/it] 3%|▎ | 839/30000 [31:48<18:30:52, 2.29s/it] 3%|▎ | 840/30000 [31:50<18:34:59, 2.29s/it] 3%|▎ | 840/30000 [31:50<18:34:59, 2.29s/it] 3%|▎ | 841/30000 [31:52<18:42:02, 2.31s/it] 3%|▎ | 842/30000 [31:55<18:38:29, 2.30s/it] 3%|▎ | 843/30000 [31:57<18:31:47, 2.29s/it] 3%|▎ | 844/30000 [31:59<18:38:08, 2.30s/it] 3%|▎ | 845/30000 [32:01<18:39:20, 2.30s/it] 3%|▎ | 846/30000 [32:04<18:30:25, 2.29s/it] 3%|▎ | 847/30000 [32:06<18:35:56, 2.30s/it] 3%|▎ | 848/30000 [32:08<18:38:50, 2.30s/it] 3%|▎ | 849/30000 [32:11<18:33:55, 2.29s/it] 3%|▎ | 850/30000 [32:13<18:35:39, 2.30s/it] 3%|▎ | 850/30000 [32:13<18:35:39, 2.30s/it] 3%|▎ | 851/30000 [32:15<18:40:49, 2.31s/it] 3%|▎ | 852/30000 [32:18<18:43:23, 2.31s/it] 3%|▎ | 853/30000 [32:20<18:36:42, 2.30s/it] 3%|▎ | 854/30000 [32:22<18:35:59, 2.30s/it] 3%|▎ | 855/30000 [32:24<18:42:41, 2.31s/it] 3%|▎ | 856/30000 [32:27<18:41:49, 2.31s/it] 3%|▎ | 857/30000 [32:29<18:32:21, 2.29s/it] 3%|▎ | 858/30000 [32:31<18:32:55, 2.29s/it] 3%|▎ | 859/30000 [32:34<18:37:38, 2.30s/it] 3%|▎ | 860/30000 [32:36<18:30:26, 2.29s/it] 3%|▎ | 860/30000 [32:36<18:30:26, 2.29s/it] 3%|▎ | 861/30000 [32:38<18:28:38, 2.28s/it] 3%|▎ | 862/30000 [32:41<18:36:12, 2.30s/it] 3%|▎ | 863/30000 [32:43<18:36:54, 2.30s/it] 3%|▎ | 864/30000 [32:45<18:36:51, 2.30s/it] 3%|▎ | 865/30000 [32:47<18:27:29, 2.28s/it] 3%|▎ | 866/30000 [32:50<18:32:20, 2.29s/it] 3%|▎ | 867/30000 [32:52<18:27:31, 2.28s/it] 3%|▎ | 868/30000 [32:54<18:27:33, 2.28s/it] 3%|▎ | 869/30000 [32:57<18:32:02, 2.29s/it] 3%|▎ | 870/30000 [32:59<18:31:01, 2.29s/it] +{'loss': 0.0261, 'grad_norm': 0.12289468944072723, 'learning_rate': 8.98e-06, 'flow_matching_loss': 0.02612177673727274, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.13} +{'loss': 0.0247, 'grad_norm': 0.11251690238714218, 'learning_rate': 9.18e-06, 'flow_matching_loss': 0.024657025188207626, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.13} +{'loss': 0.0259, 'grad_norm': 0.11474744975566864, 'learning_rate': 9.38e-06, 'flow_matching_loss': 0.025887068547308444, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.14} +{'loss': 0.025, 'grad_norm': 0.11526566743850708, 'learning_rate': 9.580000000000001e-06, 'flow_matching_loss': 0.02497375961393118, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.14} +{'loss': 0.0248, 'grad_norm': 0.09640290588140488, 'learning_rate': 9.780000000000001e-06, 'flow_matching_loss': 0.024761319160461426, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.14} +{'loss': 0.0258, 'grad_norm': 0.10522473603487015, 'learning_rate': 9.980000000000001e-06, 'flow_matching_loss': 0.025818328000605108, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.15} +{'loss': 0.0259, 'grad_norm': 0.1299685835838318, 'learning_rate': 1.0179999999999999e-05, 'flow_matching_loss': 0.025872234813869, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.15} +{'loss': 0.022, 'grad_norm': 0.10535416007041931, 'learning_rate': 1.0379999999999999e-05, 'flow_matching_loss': 0.021965932659804822, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.15} +{'loss': 0.0212, 'grad_norm': 0.0916435495018959, 'learning_rate': 1.058e-05, 'flow_matching_loss': 0.021229392476379873, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.15} +{'loss': 0.0249, 'grad_norm': 0.11862961947917938, 'learning_rate': 1.078e-05, 'flow_matching_loss': 0.024864858854562043, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.16} +{'loss': 0.0227, 'grad_norm': 0.08918412029743195, 'learning_rate': 1.098e-05, 'flow_matching_loss': 0.022710948903113603, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.16} +{'loss': 0.0216, 'grad_norm': 0.06377934664487839, 'learning_rate': 1.118e-05, 'flow_matching_loss': 0.021623315662145613, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.16} +{'loss': 0.0255, 'grad_norm': 0.1168038472533226, 'learning_rate': 1.1380000000000001e-05, 'flow_matching_loss': 0.02545337565243244, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.17} +{'loss': 0.0223, 'grad_norm': 0.0990239679813385, 'learning_rate': 1.1580000000000001e-05, 'flow_matching_loss': 0.022344311326742174, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.17} +{'loss': 0.026, 'grad_norm': 0.0860186517238617, 'learning_rate': 1.178e-05, 'flow_matching_loss': 0.02600078023970127, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.17} +{'loss': 0.0237, 'grad_norm': 0.10180020332336426, 'learning_rate': 1.198e-05, 'flow_matching_loss': 0.023707061633467676, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.17} +{'loss': 0.0217, 'grad_norm': 0.10598171502351761, 'learning_rate': 1.2180000000000002e-05, 'flow_matching_loss': 0.02173575069755316, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.18} +{'loss': 0.023, 'grad_norm': 0.10324563831090927, 'learning_rate': 1.238e-05, 'flow_matching_loss': 0.023042372614145278, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.18} +{'loss': 0.0227, 'grad_norm': 0.1065848097205162, 'learning_rate': 1.258e-05, 'flow_matching_loss': 0.022666584141552447, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.18} +{'loss': 0.0232, 'grad_norm': 0.10722683370113373, 'learning_rate': 1.278e-05, 'flow_matching_loss': 0.023166095651686192, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.19} +{'loss': 0.023, 'grad_norm': 0.11428894847631454, 'learning_rate': 1.298e-05, 'flow_matching_loss': 0.022956989146769045, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.19} +{'loss': 0.0196, 'grad_norm': 0.11275286972522736, 'learning_rate': 1.3180000000000001e-05, 'flow_matching_loss': 0.019625614769756793, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.19} +{'loss': 0.0258, 'grad_norm': 0.11304866522550583, 'learning_rate': 1.338e-05, 'flow_matching_loss': 0.025806786492466927, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.19} +{'loss': 0.0225, 'grad_norm': 0.11920002847909927, 'learning_rate': 1.358e-05, 'flow_matching_loss': 0.022468149475753308, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.2} +{'loss': 0.0216, 'grad_norm': 0.1312607377767563, 'learning_rate': 1.378e-05, 'flow_matching_loss': 0.02155338190495968, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.2} +{'loss': 0.0207, 'grad_norm': 0.11397044360637665, 'learning_rate': 1.3980000000000002e-05, 'flow_matching_loss': 0.020712326094508172, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.2} +{'loss': 0.0225, 'grad_norm': 0.09693106263875961, 'learning_rate': 1.4180000000000001e-05, 'flow_matching_loss': 0.022502155043184757, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.21} +{'loss': 0.0202, 'grad_norm': 0.08406130969524384, 'learning_rate': 1.438e-05, 'flow_matching_loss': 0.02017089566215873, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.21} +{'loss': 0.0229, 'grad_norm': 0.11150752753019333, 'learning_rate': 1.458e-05, 'flow_matching_loss': 0.022927170246839525, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.21} +{'loss': 0.0213, 'grad_norm': 0.09255439043045044, 'learning_rate': 1.4779999999999999e-05, 'flow_matching_loss': 0.021252985298633575, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.22} +{'loss': 0.0215, 'grad_norm': 0.10574609786272049, 'learning_rate': 1.498e-05, 'flow_matching_loss': 0.02149612121284008, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.22} +{'loss': 0.0215, 'grad_norm': 0.09783576428890228, 'learning_rate': 1.518e-05, 'flow_matching_loss': 0.021513335593044757, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.22} +{'loss': 0.0222, 'grad_norm': 0.1033083125948906, 'learning_rate': 1.5380000000000002e-05, 'flow_matching_loss': 0.02217772863805294, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.22} +{'loss': 0.0221, 'grad_norm': 0.10406232625246048, 'learning_rate': 1.558e-05, 'flow_matching_loss': 0.022127522714436056, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.23} +{'loss': 0.0222, 'grad_norm': 0.12757836282253265, 'learning_rate': 1.578e-05, 'flow_matching_loss': 0.022202673368155956, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.23} +{'loss': 0.0222, 'grad_norm': 0.11735603958368301, 'learning_rate': 1.598e-05, 'flow_matching_loss': 0.0221654886379838, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.23} +{'loss': 0.0202, 'grad_norm': 0.11844217032194138, 'learning_rate': 1.618e-05, 'flow_matching_loss': 0.020244254916906356, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.24} +{'loss': 0.0218, 'grad_norm': 0.10603825747966766, 'learning_rate': 1.6380000000000002e-05, 'flow_matching_loss': 0.021805757470428944, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.24} +{'loss': 0.0192, 'grad_norm': 0.12730568647384644, 'learning_rate': 1.658e-05, 'flow_matching_loss': 0.019202703423798085, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.24} +{'loss': 0.0218, 'grad_norm': 0.0999128445982933, 'learning_rate': 1.6780000000000002e-05, 'flow_matching_loss': 0.021799416095018388, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.24} +{'loss': 0.0243, 'grad_norm': 0.11548066139221191, 'learning_rate': 1.698e-05, 'flow_matching_loss': 0.024295849446207284, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.25} +{'loss': 0.0211, 'grad_norm': 0.14502982795238495, 'learning_rate': 1.718e-05, 'flow_matching_loss': 0.02110443785786629, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.25} +{'loss': 0.0206, 'grad_norm': 0.0936955064535141, 'learning_rate': 1.7380000000000003e-05, 'flow_matching_loss': 0.020604066736996173, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.25} 3%|▎ | 870/30000 [32:59<18:31:01, 2.29s/it] 3%|▎ | 871/30000 [33:01<18:27:39, 2.28s/it] 3%|▎ | 872/30000 [33:03<18:24:52, 2.28s/it] 3%|▎ | 873/30000 [33:06<18:33:22, 2.29s/it] 3%|▎ | 874/30000 [33:08<18:34:15, 2.30s/it] 3%|▎ | 875/30000 [33:10<18:32:47, 2.29s/it] 3%|▎ | 876/30000 [33:13<18:34:38, 2.30s/it] 3%|▎ | 877/30000 [33:15<18:32:35, 2.29s/it] 3%|▎ | 878/30000 [33:17<18:29:56, 2.29s/it] 3%|▎ | 879/30000 [33:19<18:24:24, 2.28s/it] 3%|▎ | 880/30000 [33:22<18:30:12, 2.29s/it] 3%|▎ | 880/30000 [33:22<18:30:12, 2.29s/it] 3%|▎ | 881/30000 [33:24<18:24:08, 2.28s/it] 3%|▎ | 882/30000 [33:26<18:22:39, 2.27s/it] 3%|▎ | 883/30000 [33:29<18:33:50, 2.30s/it] 3%|▎ | 884/30000 [33:31<18:40:02, 2.31s/it] 3%|▎ | 885/30000 [33:33<18:33:53, 2.30s/it] 3%|▎ | 886/30000 [33:35<18:41:13, 2.31s/it] 3%|▎ | 887/30000 [33:38<18:41:56, 2.31s/it] 3%|▎ | 888/30000 [33:40<18:37:06, 2.30s/it] 3%|▎ | 889/30000 [33:42<18:37:41, 2.30s/it] 3%|▎ | 890/30000 [33:45<18:34:19, 2.30s/it] 3%|▎ | 890/30000 [33:45<18:34:19, 2.30s/it] 3%|▎ | 891/30000 [33:47<18:24:59, 2.28s/it] 3%|▎ | 892/30000 [33:49<18:29:21, 2.29s/it] 3%|▎ | 893/30000 [33:52<18:30:55, 2.29s/it] 3%|▎ | 894/30000 [33:54<18:23:46, 2.28s/it] 3%|▎ | 895/30000 [33:56<18:29:28, 2.29s/it] 3%|▎ | 896/30000 [33:58<18:27:28, 2.28s/it] 3%|▎ | 897/30000 [34:01<18:25:11, 2.28s/it] 3%|▎ | 898/30000 [34:03<18:31:42, 2.29s/it] 3%|▎ | 899/30000 [34:05<18:32:19, 2.29s/it] 3%|▎ | 900/30000 [34:08<18:30:31, 2.29s/it] 3%|▎ | 900/30000 [34:08<18:30:31, 2.29s/it] 3%|▎ | 901/30000 [34:10<18:41:59, 2.31s/it] 3%|▎ | 902/30000 [34:12<18:39:21, 2.31s/it] 3%|▎ | 903/30000 [34:14<18:32:14, 2.29s/it] 3%|▎ | 904/30000 [34:17<18:31:26, 2.29s/it] 3%|▎ | 905/30000 [34:19<18:38:17, 2.31s/it] 3%|▎ | 906/30000 [34:21<18:30:24, 2.29s/it] 3%|▎ | 907/30000 [34:24<18:27:49, 2.28s/it] 3%|▎ | 908/30000 [34:26<18:37:10, 2.30s/it] 3%|▎ | 909/30000 [34:28<18:34:38, 2.30s/it] 3%|▎ | 910/30000 [34:30<18:27:16, 2.28s/it] 3%|▎ | 910/30000 [34:31<18:27:16, 2.28s/it] 3%|▎ | 911/30000 [34:33<18:37:49, 2.31s/it] 3%|▎ | 912/30000 [34:35<18:33:01, 2.30s/it] 3%|▎ | 913/30000 [34:37<18:30:39, 2.29s/it] 3%|▎ | 914/30000 [34:40<18:42:47, 2.32s/it] 3%|▎ | 915/30000 [34:42<18:45:41, 2.32s/it] 3%|▎ | 916/30000 [34:44<18:37:03, 2.30s/it] 3%|▎ | 917/30000 [34:47<18:43:21, 2.32s/it] 3%|▎ | 918/30000 [34:49<18:39:01, 2.31s/it] 3%|▎ | 919/30000 [34:51<18:32:48, 2.30s/it] 3%|▎ | 920/30000 [34:54<18:40:37, 2.31s/it] 3%|▎ | 920/30000 [34:54<18:40:37, 2.31s/it] 3%|▎ | 921/30000 [34:56<18:37:03, 2.30s/it] 3%|▎ | 922/30000 [34:58<18:30:05, 2.29s/it] 3%|▎ | 923/30000 [35:00<18:29:24, 2.29s/it] 3%|▎ | 924/30000 [35:03<18:29:23, 2.29s/it] 3%|▎ | 925/30000 [35:05<18:24:35, 2.28s/it] 3%|▎ | 926/30000 [35:07<18:19:59, 2.27s/it] 3%|▎ | 927/30000 [35:10<18:29:32, 2.29s/it] 3%|▎ | 928/30000 [35:12<18:25:27, 2.28s/it] 3%|▎ | 929/30000 [35:14<18:26:30, 2.28s/it] 3%|▎ | 930/30000 [35:16<18:37:20, 2.31s/it] 3%|▎ | 930/30000 [35:17<18:37:20, 2.31s/it] 3%|▎ | 931/30000 [35:19<18:39:46, 2.31s/it] 3%|▎ | 932/30000 [35:21<18:34:12, 2.30s/it] 3%|▎ | 933/30000 [35:23<18:43:24, 2.32s/it] 3%|▎ | 934/30000 [35:26<18:38:53, 2.31s/it] 3%|▎ | 935/30000 [35:28<18:35:32, 2.30s/it] 3%|▎ | 936/30000 [35:30<18:33:40, 2.30s/it] 3%|▎ | 937/30000 [35:33<18:34:46, 2.30s/it] 3%|▎ | 938/30000 [35:35<18:31:11, 2.29s/it] 3%|▎ | 939/30000 [35:37<18:27:56, 2.29s/it] 3%|▎ | 940/30000 [35:40<18:37:00, 2.31s/it] 3%|▎ | 940/30000 [35:40<18:37:00, 2.31s/it] 3%|▎ | 941/30000 [35:42<18:40:00, 2.31s/it] 3%|▎ | 942/30000 [35:44<18:34:01, 2.30s/it] 3%|▎ | 943/30000 [35:46<18:39:10, 2.31s/it] 3%|▎ | 944/30000 [35:49<18:38:29, 2.31s/it] 3%|▎ | 945/30000 [35:51<18:33:52, 2.30s/it] 3%|▎ | 946/30000 [35:53<18:32:55, 2.30s/it] 3%|▎ | 947/30000 [35:56<18:37:03, 2.31s/it] 3%|▎ | 948/30000 [35:58<18:31:44, 2.30s/it] 3%|▎ | 949/30000 [36:00<18:27:34, 2.29s/it] 3%|▎ | 950/30000 [36:03<18:34:18, 2.30s/it] 3%|▎ | 950/30000 [36:03<18:34:18, 2.30s/it] 3%|▎ | 951/30000 [36:05<18:38:54, 2.31s/it] 3%|▎ | 952/30000 [36:07<18:31:18, 2.30s/it] 3%|▎ | 953/30000 [36:09<18:37:41, 2.31s/it] 3%|▎ | 954/30000 [36:12<18:37:29, 2.31s/it] 3%|▎ | 955/30000 [36:14<18:31:57, 2.30s/it] 3%|▎ | 956/30000 [36:16<18:26:19, 2.29s/it] 3%|▎ | 957/30000 [36:19<18:27:37, 2.29s/it] 3%|▎ | 958/30000 [36:21<18:21:33, 2.28s/it] 3%|▎ | 959/30000 [36:23<18:23:27, 2.28s/it] 3%|▎ | 960/30000 [36:25<18:33:41, 2.30s/it] 3%|▎ | 960/30000 [36:26<18:33:41, 2.30s/it] 3%|▎ | 961/30000 [36:28<18:34:39, 2.30s/it] 3%|▎ | 962/30000 [36:30<18:29:51, 2.29s/it] 3%|▎ | 963/30000 [36:32<18:35:07, 2.30s/it] 3%|▎ | 964/30000 [36:35<18:29:06, 2.29s/it] 3%|▎ | 965/30000 [36:37<18:28:56, 2.29s/it] 3%|▎ | 966/30000 [36:39<18:35:18, 2.30s/it] 3%|▎ | 967/30000 [36:42<18:34:21, 2.30s/it] 3%|▎ | 968/30000 [36:44<18:28:17, 2.29s/it] 3%|▎ | 969/30000 [36:46<18:34:25, 2.30s/it] 3%|▎ | 970/30000 [36:48<18:28:06, 2.29s/it] 3%|▎ | 970/30000 [36:48<18:28:06, 2.29s/it] 3%|▎ | 971/30000 [36:51<18:22:52, 2.28s/it] 3%|▎ | 972/30000 [36:53<18:28:36, 2.29s/it] 3%|▎ | 973/30000 [36:55<18:27:46, 2.29s/it] 3%|▎ | 974/30000 [36:58<18:22:21, 2.28s/it] 3%|▎ | 975/30000 [37:00<18:22:51, 2.28s/it] 3%|▎ | 976/30000 [37:02<18:28:44, 2.29s/it] 3%|▎ | 977/30000 [37:04<18:24:18, 2.28s/it] 3%|▎ | 978/30000 [37:07<18:18:30, 2.27s/it] 3%|▎ | 979/30000 [37:09<18:25:24, 2.29s/it] 3%|▎ | 980/30000 [37:11<18:24:01, 2.28s/it] 3%|▎ | 980/30000 [37:11<18:24:01, 2.28s/it] 3%|▎ | 981/30000 [37:14<18:22:48, 2.28s/it] 3%|▎ | 982/30000 [37:16<18:34:47, 2.31s/it] 3%|▎ | 983/30000 [37:18<18:35:03, 2.31s/it] 3%|▎ | 984/30000 [37:20<18:34:19, 2.30s/it] 3%|▎ | 985/30000 [37:23<18:28:41, 2.29s/it] 3%|▎ | 986/30000 [37:25<18:26:25, 2.29s/it] 3%|▎ | 987/30000 [37:27<18:20:38, 2.28s/it] 3%|▎ | 988/30000 [37:30<18:15:16, 2.27s/it] 3%|▎ | 989/30000 [37:32<18:23:26, 2.28s/it] 3%|▎ | 990/30000 [37:34<18:19:44, 2.27s/it] 3%|▎ | 990/30000 [37:34<18:19:44, 2.27s/it] 3%|▎ | 991/30000 [37:36<18:19:24, 2.27s/it] 3%|▎ | 992/30000 [37:39<18:31:13, 2.30s/it] 3%|▎ | 993/30000 [37:41<18:31:45, 2.30s/it] 3%|▎ | 994/30000 [37:43<18:25:10, 2.29s/it] 3%|▎ | 995/30000 [37:46<18:29:40, 2.30s/it] 3%|▎ | 996/30000 [37:48<18:36:36, 2.31s/it] 3%|▎ | 997/30000 [37:50<18:36:44, 2.31s/it] 3%|▎ | 998/30000 [37:52<18:26:28, 2.29s/it] 3%|▎ | 999/30000 [37:55<18:25:05, 2.29s/it] 3%|▎ | 1000/30000 [37:57<18:33:14, 2.30s/it] 3%|▎ | 1000/30000 [37:57<18:33:14, 2.30s/it] 3%|▎ | 1001/30000 [37:59<18:31:15, 2.30s/it] 3%|▎ | 1002/30000 [38:02<18:30:23, 2.30s/it] 3%|▎ | 1003/30000 [38:04<18:32:15, 2.30s/it] 3%|▎ | 1004/30000 [38:06<18:29:50, 2.30s/it] 3%|▎ | 1005/30000 [38:09<18:25:17, 2.29s/it] 3%|▎ | 1006/30000 [38:11<18:23:10, 2.28s/it] 3%|▎ | 1007/30000 [38:13<18:29:30, 2.30s/it] 3%|▎ | 1008/30000 [38:15<18:32:13, 2.30s/it] 3%|▎ | 1009/30000 [38:18<18:25:17, 2.29s/it] 3%|▎ | 1010/30000 [38:20<18:21:30, 2.28s/it] 3%|▎ | 1010/30000 [38:20<18:21:30, 2.28s/it] 3%|▎ | 1011/30000 [38:22<18:29:26, 2.30s/it] 3%|▎ | 1012/30000 [38:25<18:33:21, 2.30s/it] 3%|▎ | 1013/30000 [38:27<18:29:06, 2.30s/it] 3%|▎ | 1014/30000 [38:29<18:26:56, 2.29s/it] 3%|▎ | 1015/30000 [38:32<18:40:25, 2.32s/it] 3%|▎ | 1016/30000 [38:34<18:31:59, 2.30s/it] 3%|▎ | 1017/30000 [38:36<18:31:38, 2.30s/it] 3%|▎ | 1018/30000 [38:39<18:38:50, 2.32s/it] 3%|▎ | 1019/30000 [38:41<18:35:15, 2.31s/it] 3%|▎ | 1020/30000 [38:43<18:27:34, 2.29s/it] 3%|▎ | 1020/30000 [38:43<18:27:34, 2.29s/it] 3%|▎ | 1021/30000 [38:45<18:24:50, 2.29s/it] 3%|▎ | 1022/30000 [38:48<18:22:50, 2.28s/it] 3%|▎ | 1023/30000 [38:50<18:16:17, 2.27s/it] 3%|▎ | 1024/30000 [38:52<18:12:30, 2.26s/it] 3%|▎ | 1025/30000 [38:54<18:24:28, 2.29s/it] 3%|▎ | 1026/30000 [38:57<18:30:02, 2.30s/it] 3%|▎ | 1027/30000 [38:59<18:22:43, 2.28s/it] 3%|▎ | 1028/30000 [39:01<18:18:27, 2.27s/it] 3%|▎ | 1029/30000 [39:04<18:27:56, 2.29s/it] 3%|▎ | 1030/30000 [39:06<18:35:04, 2.31s/it] 3%|▎ | 1030/30000 [39:06<18:35:04, 2.31s/it] 3%|▎ | 1031/30000 [39:08<18:28:16, 2.30s/it] 3%|▎ | 1032/30000 [39:10<18:21:51, 2.28s/it] 3%|▎ | 1033/30000 [39:13<18:30:50, 2.30s/it] 3%|▎ | 1034/30000 [39:15<18:24:11, 2.29s/it] 3%|▎ | 1035/30000 [39:17<18:26:05, 2.29s/it] 3%|▎ | 1036/30000 [39:20<18:27:17, 2.29s/it] 3%|▎ | 1037/30000 [39:22<18:29:10, 2.30s/it] 3%|▎ | 1038/30000 [39:24<18:20:07, 2.28s/it] 3%|▎ | 1039/30000 [39:26<18:13:17, 2.27s/it] 3%|▎ | 1040/30000 [39:29<18:23:29, 2.29s/it] 3%|▎ | 1040/30000 [39:29<18:23:29, 2.29s/it] 3%|▎ | 1041/30000 [39:31<18:26:42, 2.29s/it] 3%|▎ | 1042/30000 [39:33<18:24:40, 2.29s/it] 3%|▎ | 1043/30000 [39:36<18:27:46, 2.30s/it] 3%|▎ | 1044/30000 [39:38<18:32:05, 2.30s/it] 3%|▎ | 1045/30000 [39:40<18:32:01, 2.30s/it] 3%|▎ | 1046/30000 [39:43<18:26:35, 2.29s/it] 3%|▎ | 1047/30000 [39:45<18:34:41, 2.31s/it] 3%|▎ | 1048/30000 [39:47<18:36:59, 2.31s/it] 3%|▎ | 1049/30000 [39:49<18:28:11, 2.30s/it] 4%|▎ | 1050/30000 [39:52<18:22:08, 2.28s/it] 4%|▎ | 1050/30000 [39:52<18:22:08, 2.28s/it] 4%|▎ | 1051/30000 [39:54<18:32:29, 2.31s/it] 4%|▎ | 1052/30000 [39:56<18:35:20, 2.31s/it] 4%|▎ | 1053/30000 [39:59<18:29:54, 2.30s/it] 4%|▎ | 1054/30000 [40:01<18:34:31, 2.31s/it] 4%|▎ | 1055/30000 [40:03<18:30:05, 2.30s/it] 4%|▎ | 1056/30000 [40:06<18:25:23, 2.29s/it] 4%|▎ | 1057/30000 [40:08<18:25:20, 2.29s/it] 4%|▎ | 1058/30000 [40:10<18:34:22, 2.31s/it] 4%|▎ | 1059/30000 [40:13<18:34:16, 2.31s/it] 4%|▎ | 1060/30000 [40:15<18:26:57, 2.29s/it] 4%|▎ | 1060/30000 [40:15<18:26:57, 2.29s/it] 4%|▎ | 1061/30000 [40:17<18:24:59, 2.29s/it] 4%|▎ | 1062/30000 [40:19<18:32:16, 2.31s/it] 4%|▎ | 1063/30000 [40:22<18:36:34, 2.32s/it] 4%|▎ | 1064/30000 [40:24<18:32:09, 2.31s/it] 4%|▎ | 1065/30000 [40:26<18:27:59, 2.30s/it] 4%|▎ | 1066/30000 [40:29<18:36:35, 2.32s/it] 4%|▎ | 1067/30000 [40:31<18:34:49, 2.31s/it] 4%|▎ | 1068/30000 [40:33<18:32:15, 2.31s/it] 4%|▎ | 1069/30000 [40:36<18:23:55, 2.29s/it] 4%|▎ | 1070/30000 [40:38<18:20:03, 2.28s/it] 4%|▎ | 1070/30000 [40:38<18:20:03, 2.28s/it] 4%|▎ | 1071/30000 [40:40<18:19:55, 2.28s/it] 4%|▎ | 1072/30000 [40:42<18:16:56, 2.28s/it] 4%|▎ | 1073/30000 [40:45<18:28:27, 2.30s/it] 4%|▎ | 1074/30000 [40:47<18:31:37, 2.31s/it] 4%|▎ | 1075/30000 [40:49<18:21:30, 2.28s/it] 4%|▎ | 1076/30000 [40:51<18:15:36, 2.27s/it] 4%|▎ | 1077/30000 [40:54<18:24:25, 2.29s/it] 4%|▎ | 1078/30000 [40:56<18:21:45, 2.29s/it] 4%|▎ | 1079/30000 [40:58<18:19:58, 2.28s/it] 4%|▎ | 1080/30000 [41:01<18:20:00, 2.28s/it] 4%|▎ | 1080/30000 [41:01<18:20:00, 2.28s/it] 4%|▎ | 1081/30000 [41:03<18:22:21, 2.29s/it] 4%|▎ | 1082/30000 [41:05<18:22:04, 2.29s/it] 4%|▎ | 1083/30000 [41:07<18:18:23, 2.28s/it] 4%|▎ | 1084/30000 [41:10<18:28:16, 2.30s/it] 4%|▎ | 1085/30000 [41:12<18:21:40, 2.29s/it] 4%|▎ | 1086/30000 [41:14<18:15:53, 2.27s/it] 4%|▎ | 1087/30000 [41:17<18:11:01, 2.26s/it] 4%|▎ | 1088/30000 [41:19<18:19:27, 2.28s/it] 4%|▎ | 1089/30000 [41:21<18:17:10, 2.28s/it] 4%|▎ | 1090/30000 [41:23<18:10:24, 2.26s/it] 4%|▎ | 1090/30000 [41:23<18:10:24, 2.26s/it] 4%|▎ | 1091/30000 [41:26<18:14:50, 2.27s/it] 4%|▎ | 1092/30000 [41:28<18:12:18, 2.27s/it] 4%|▎ | 1093/30000 [41:30<18:07:30, 2.26s/it] 4%|▎ | 1094/30000 [41:32<18:08:08, 2.26s/it] 4%|▎ | 1095/30000 [41:35<18:16:17, 2.28s/it] 4%|▎ | 1096/30000 [41:37<18:14:00, 2.27s/it] 4%|▎ | 1097/30000 [41:39<18:09:36, 2.26s/it] 4%|▎ | 1098/30000 [41:42<18:18:11, 2.28s/it] 4%|▎ | 1099/30000 [41:44<18:22:49, 2.29s/it] 4%|▎ | 1100/30000 [41:46<18:17:02, 2.28s/it] 4%|▎ | 1100/30000 [41:46<18:17:02, 2.28s/it] 4%|▎ | 1101/30000 [41:48<18:24:00, 2.29s/it] 4%|▎ | 1102/30000 [41:51<18:24:23, 2.29s/it] 4%|▎ | 1103/30000 [41:53<18:17:40, 2.28s/it] 4%|▎ | 1104/30000 [41:55<18:15:30, 2.27s/it] 4%|▎ | 1105/30000 [41:58<18:24:27, 2.29s/it] 4%|▎ | 1106/30000 [42:00<18:24:29, 2.29s/it] 4%|▎ | 1107/30000 [42:02<18:21:48, 2.29s/it] 4%|▎ | 1108/30000 [42:04<18:12:34, 2.27s/it] 4%|▎ | 1109/30000 [42:07<18:23:22, 2.29s/it] 4%|▎ | 1110/30000 [42:09<18:19:02, 2.28s/it] 4%|▎ | 1110/30000 [42:09<18:19:02, 2.28s/it] 4%|▎ | 1111/30000 [42:11<18:17:24, 2.28s/it] 4%|▎ | 1112/30000 [42:14<18:24:44, 2.29s/it] 4%|▎ | 1113/30000 [42:16<18:28:53, 2.30s/it] 4%|▎ | 1114/30000 [42:18<18:16:49, 2.28s/it] 4%|▎ | 1115/30000 [42:20<18:14:06, 2.27s/it] 4%|▎ | 1116/30000 [42:23<18:23:17, 2.29s/it] 4%|▎ | 1117/30000 [42:25<18:27:38, 2.30s/it] 4%|▎ | 1118/30000 [42:27<18:22:56, 2.29s/it] 4%|▎ | 1119/30000 [42:30<18:13:45, 2.27s/it] 4%|▎ | 1120/30000 [42:32<18:21:19, 2.29s/it] 4%|▎ | 1120/30000 [42:32<18:21:19, 2.29s/it] 4%|▎ | 1121/30000 [42:34<18:13:21, 2.27s/it] 4%|▎ | 1122/30000 [42:36<18:08:27, 2.26s/it] 4%|▎ | 1123/30000 [42:39<18:19:11, 2.28s/it] 4%|▎ | 1124/30000 [42:41<18:21:52, 2.29s/it] 4%|▍ | 1125/30000 [42:43<18:18:03, 2.28s/it] 4%|▍ | 1126/30000 [42:46<18:15:01, 2.28s/it] 4%|▍ | 1127/30000 [42:48<18:25:04, 2.30s/it] 4%|▍ | 1128/30000 [42:50<18:19:41, 2.29s/it] 4%|▍ | 1129/30000 [42:52<18:16:10, 2.28s/it] 4%|▍ | 1130/30000 [42:55<18:22:02, 2.29s/it] 4%|▍ | 1130/30000 [42:55<18:22:02, 2.29s/it] 4%|▍ | 1131/30000 [42:57<18:25:36, 2.30s/it] 4%|▍ | 1132/30000 [42:59<18:21:24, 2.29s/it] 4%|▍ | 1133/30000 [43:02<18:22:05, 2.29s/it] 4%|▍ | 1134/30000 [43:04<18:27:36, 2.30s/it] 4%|▍ | 1135/30000 [43:06<18:28:19, 2.30s/it] 4%|▍ | 1136/30000 [43:09<18:26:08, 2.30s/it] 4%|▍ | 1137/30000 [43:11<18:21:53, 2.29s/it] 4%|▍ | 1138/30000 [43:13<18:30:24, 2.31s/it] 4%|▍ | 1139/30000 [43:15<18:28:07, 2.30s/it] 4%|▍ | 1140/30000 [43:18<18:25:53, 2.30s/it] 4%|▍ | 1140/30000 [43:18<18:25:53, 2.30s/it] 4%|▍ | 1141/30000 [43:20<18:23:24, 2.29s/it] 4%|▍ | 1142/30000 [43:22<18:26:24, 2.30s/it] 4%|▍ | 1143/30000 [43:25<18:22:03, 2.29s/it] 4%|▍ | 1144/30000 [43:27<18:19:54, 2.29s/it] 4%|▍ | 1145/30000 [43:29<18:26:21, 2.30s/it] 4%|▍ | 1146/30000 [43:31<18:15:46, 2.28s/it] 4%|▍ | 1147/30000 [43:34<18:13:29, 2.27s/it] 4%|▍ | 1148/30000 [43:36<18:16:58, 2.28s/it] 4%|▍ | 1149/30000 [43:38<18:19:16, 2.29s/it] 4%|▍ | 1150/30000 [43:41<18:15:32, 2.28s/it] 4%|▍ | 1150/30000 [43:41<18:15:32, 2.28s/it] 4%|▍ | 1151/30000 [43:43<18:14:12, 2.28s/it] 4%|▍ | 1152/30000 [43:45<18:22:18, 2.29s/it] 4%|▍ | 1153/30000 [43:47<18:25:01, 2.30s/it] 4%|▍ | 1154/30000 [43:50<18:17:49, 2.28s/it] 4%|▍ | 1155/30000 [43:52<18:25:43, 2.30s/it] 4%|▍ | 1156/30000 [43:54<18:27:42, 2.30s/it] 4%|▍ | 1157/30000 [43:57<18:23:05, 2.29s/it] 4%|▍ | 1158/30000 [43:59<18:21:48, 2.29s/it] 4%|▍ | 1159/30000 [44:01<18:26:17, 2.30s/it] 4%|▍ | 1160/30000 [44:04<18:22:38, 2.29s/it] 4%|▍ | 1160/30000 [44:04<18:22:38, 2.29s/it] 4%|▍ | 1161/30000 [44:06<18:16:24, 2.28s/it] 4%|▍ | 1162/30000 [44:08<18:28:17, 2.31s/it] 4%|▍ | 1163/30000 [44:10<18:28:28, 2.31s/it] 4%|▍ | 1164/30000 [44:13<18:24:11, 2.30s/it] 4%|▍ | 1165/30000 [44:15<18:29:51, 2.31s/it] 4%|▍ | 1166/30000 [44:17<18:29:11, 2.31s/it] 4%|▍ | 1167/30000 [44:20<18:25:44, 2.30s/it] 4%|▍ | 1168/30000 [44:22<18:29:02, 2.31s/it] 4%|▍ | 1169/30000 [44:24<18:30:19, 2.31s/it] 4%|▍ | 1170/30000 [44:27<18:20:20, 2.29s/it] 4%|▍ | 1170/30000 [44:27<18:20:20, 2.29s/it] 4%|▍ | 1171/30000 [44:29<18:15:48, 2.28s/it] 4%|▍ | 1172/30000 [44:31<18:27:15, 2.30s/it] 4%|▍ | 1173/30000 [44:33<18:21:16, 2.29s/it] 4%|▍ | 1174/30000 [44:36<18:16:33, 2.28s/it] 4%|▍ | 1175/30000 [44:38<18:29:37, 2.31s/it] 4%|▍ | 1176/30000 [44:40<18:31:32, 2.31s/it] 4%|▍ | 1177/30000 [44:43<18:25:38, 2.30s/it] 4%|▍ | 1178/30000 [44:45<18:23:43, 2.30s/it] 4%|▍ | 1179/30000 [44:47<18:16:15, 2.28s/it] 4%|▍ | 1180/30000 [44:49<18:16:53, 2.28s/it] 4%|▍ | 1180/30000 [44:50<18:16:53, 2.28s/it] 4%|▍ | 1181/30000 [44:52<18:20:47, 2.29s/it] 4%|▍ | 1182/30000 [44:54<18:19:28, 2.29s/it] 4%|▍ | 1183/30000 [44:56<18:16:24, 2.28s/it] 4%|▍ | 1184/30000 [44:59<18:17:43, 2.29s/it] 4%|▍ | 1185/30000 [45:01<18:22:50, 2.30s/it] 4%|▍ | 1186/30000 [45:03<18:15:42, 2.28s/it] 4%|▍ | 1187/30000 [45:05<18:15:09, 2.28s/it] 4%|▍ | 1188/30000 [45:08<18:22:54, 2.30s/it] 4%|▍ | 1189/30000 [45:10<18:23:37, 2.30s/it] 4%|▍ | 1190/30000 [45:12<18:16:06, 2.28s/it] 4%|▍ | 1190/30000 [45:12<18:16:06, 2.28s/it] 4%|▍ | 1191/30000 [45:15<18:22:16, 2.30s/it] 4%|▍ | 1192/30000 [45:17<18:21:44, 2.29s/it] 4%|▍ | 1193/30000 [45:19<18:18:32, 2.29s/it] 4%|▍ | 1194/30000 [45:22<18:19:46, 2.29s/it] 4%|▍ | 1195/30000 [45:24<18:29:31, 2.31s/it] 4%|▍ | 1196/30000 [45:26<18:17:41, 2.29s/it] 4%|▍ | 1197/30000 [45:28<18:18:45, 2.29s/it] 4%|▍ | 1198/30000 [45:31<18:25:51, 2.30s/it] 4%|▍ | 1199/30000 [45:33<18:26:49, 2.31s/it] 4%|▍ | 1200/30000 [45:35<18:15:49, 2.28s/it] 4%|▍ | 1200/30000 [45:35<18:15:49, 2.28s/it] 4%|▍ | 1201/30000 [45:38<18:17:43, 2.29s/it] 4%|▍ | 1202/30000 [45:40<18:22:37, 2.30s/it] 4%|▍ | 1203/30000 [45:42<18:19:10, 2.29s/it] 4%|▍ | 1204/30000 [45:45<18:25:30, 2.30s/it] 4%|▍ | 1205/30000 [45:47<18:25:56, 2.30s/it] 4%|▍ | 1206/30000 [45:49<18:23:18, 2.30s/it] 4%|▍ | 1207/30000 [45:51<18:23:23, 2.30s/it] 4%|▍ | 1208/30000 [45:54<18:27:32, 2.31s/it] 4%|▍ | 1209/30000 [45:56<18:22:28, 2.30s/it] 4%|▍ | 1210/30000 [45:58<18:21:16, 2.30s/it] 4%|▍ | 1210/30000 [45:58<18:21:16, 2.30s/it] 4%|▍ | 1211/30000 [46:01<18:30:12, 2.31s/it] 4%|▍ | 1212/30000 [46:03<18:32:25, 2.32s/it] 4%|▍ | 1213/30000 [46:05<18:25:47, 2.30s/it] 4%|▍ | 1214/30000 [46:08<18:29:40, 2.31s/it] 4%|▍ | 1215/30000 [46:10<18:34:45, 2.32s/it] 4%|▍ | 1216/30000 [46:12<18:27:58, 2.31s/it] 4%|▍ | 1217/30000 [46:15<18:24:05, 2.30s/it] 4%|▍ | 1218/30000 [46:17<18:25:32, 2.30s/it] 4%|▍ | 1219/30000 [46:19<18:14:59, 2.28s/it] 4%|▍ | 1220/30000 [46:21<18:12:14, 2.28s/it] 4%|▍ | 1220/30000 [46:21<18:12:14, 2.28s/it] 4%|▍ | 1221/30000 [46:24<18:20:52, 2.30s/it] 4%|▍ | 1222/30000 [46:26<18:13:02, 2.28s/it] 4%|▍ | 1223/30000 [46:28<18:10:01, 2.27s/it] 4%|▍ | 1224/30000 [46:31<18:22:40, 2.30s/it] 4%|▍ | 1225/30000 [46:33<18:19:27, 2.29s/it] 4%|▍ | 1226/30000 [46:35<18:18:40, 2.29s/it] 4%|▍ | 1227/30000 [46:37<18:23:52, 2.30s/it] 4%|▍ | 1228/30000 [46:40<18:26:58, 2.31s/it] 4%|▍ | 1229/30000 [46:42<18:25:08, 2.30s/it] 4%|▍ | 1230/30000 [46:44<18:30:13, 2.32s/it] 4%|▍ | 1230/30000 [46:44<18:30:13, 2.32s/it] 4%|▍ | 1231/30000 [46:47<18:31:51, 2.32s/it] 4%|▍ | 1232/30000 [46:49<18:24:07, 2.30s/it] 4%|▍ | 1233/30000 [46:51<18:20:00, 2.29s/it] 4%|▍ | 1234/30000 [46:54<18:21:42, 2.30s/it] 4%|▍ | 1235/30000 [46:56<18:17:43, 2.29s/it] 4%|▍ | 1236/30000 [46:58<18:13:53, 2.28s/it] 4%|▍ | 1237/30000 [47:00<18:18:26, 2.29s/it] 4%|▍ | 1238/30000 [47:03<18:12:18, 2.28s/it] 4%|▍ | 1239/30000 [47:05<18:11:32, 2.28s/it] 4%|▍ | 1240/30000 [47:07<18:23:26, 2.30s/it] 4%|▍ | 1240/30000 [47:07<18:23:26, 2.30s/it] 4%|▍ | 1241/30000 [47:10<18:26:56, 2.31s/it] 4%|▍ | 1242/30000 [47:12<18:19:04, 2.29s/it] 4%|▍ | 1243/30000 [47:14<18:25:57, 2.31s/it] 4%|▍ | 1244/30000 [47:17<18:29:01, 2.31s/it] 4%|▍ | 1245/30000 [47:19<18:24:44, 2.31s/it] 4%|▍ | 1246/30000 [47:21<18:26:43, 2.31s/it] 4%|▍ | 1247/30000 [47:23<18:26:47, 2.31s/it] 4%|▍ | 1248/30000 [47:26<18:18:23, 2.29s/it] 4%|▍ | 1249/30000 [47:28<18:19:29, 2.29s/it] 4%|▍ | 1250/30000 [47:30<18:24:47, 2.31s/it] 4%|▍ | 1250/30000 [47:30<18:24:47, 2.31s/it] 4%|▍ | 1251/30000 [47:33<18:21:01, 2.30s/it] 4%|▍ | 1252/30000 [47:35<18:17:45, 2.29s/it] 4%|▍ | 1253/30000 [47:37<18:26:50, 2.31s/it] 4%|▍ | 1254/30000 [47:40<18:28:33, 2.31s/it] 4%|▍ | 1255/30000 [47:42<18:20:19, 2.30s/it] 4%|▍ | 1256/30000 [47:44<18:23:27, 2.30s/it] 4%|▍ | 1257/30000 [47:46<18:20:25, 2.30s/it] 4%|▍ | 1258/30000 [47:49<18:10:27, 2.28s/it] 4%|▍ | 1259/30000 [47:51<18:12:35, 2.28s/it] 4%|▍ | 1260/30000 [47:53<18:18:03, 2.29s/it] 4%|▍ | 1260/30000 [47:53<18:18:03, 2.29s/it] 4%|▍ | 1261/30000 [47:56<18:16:04, 2.29s/it] 4%|▍ | 1262/30000 [47:58<18:16:25, 2.29s/it] 4%|▍ | 1263/30000 [48:00<18:28:43, 2.31s/it] 4%|▍ | 1264/30000 [48:03<18:28:02, 2.31s/it] 4%|▍ | 1265/30000 [48:05<18:18:26, 2.29s/it] 4%|▍ | 1266/30000 [48:07<18:27:50, 2.31s/it] 4%|▍ | 1267/30000 [48:09<18:28:57, 2.32s/it] 4%|▍ | 1268/30000 [48:12<18:21:37, 2.30s/it] 4%|▍ | 1269/30000 [48:14<18:23:22, 2.30s/it] 4%|▍ | 1270/30000 [48:16<18:26:00, 2.31s/it] 4%|▍ | 1270/30000 [48:16<18:26:00, 2.31s/it] 4%|▍ | 1271/30000 [48:19<18:19:47, 2.30s/it] 4%|▍ | 1272/30000 [48:21<18:18:13, 2.29s/it] 4%|▍ | 1273/30000 [48:23<18:22:25, 2.30s/it] 4%|▍ | 1274/30000 [48:26<18:27:14, 2.31s/it] 4%|▍ | 1275/30000 [48:28<18:22:18, 2.30s/it] 4%|▍ | 1276/30000 [48:30<18:31:16, 2.32s/it] 4%|▍ | 1277/30000 [48:32<18:28:27, 2.32s/it] 4%|▍ | 1278/30000 [48:35<18:21:19, 2.30s/it] 4%|▍ | 1279/30000 [48:37<18:21:46, 2.30s/it] 4%|▍ | 1280/30000 [48:39<18:23:00, 2.30s/it] 4%|▍ | 1280/30000 [48:39<18:23:00, 2.30s/it] 4%|▍ | 1281/30000 [48:42<18:20:24, 2.30s/it] 4%|▍ | 1282/30000 [48:44<18:19:21, 2.30s/it] 4%|▍ | 1283/30000 [48:46<18:21:00, 2.30s/it] 4%|▍ | 1284/30000 [48:49<18:17:28, 2.29s/it] 4%|▍ | 1285/30000 [48:51<18:14:51, 2.29s/it] 4%|▍ | 1286/30000 [48:53<18:15:05, 2.29s/it] 4%|▍ | 1287/30000 [48:55<18:07:35, 2.27s/it] 4%|▍ | 1288/30000 [48:58<18:09:11, 2.28s/it] 4%|▍ | 1289/30000 [49:00<18:16:57, 2.29s/it] 4%|▍ | 1290/30000 [49:02<18:19:16, 2.30s/it] 4%|▍ | 1290/30000 [49:02<18:19:16, 2.30s/it] 4%|▍ | 1291/30000 [49:05<18:14:22, 2.29s/it] 4%|▍ | 1292/30000 [49:07<18:22:49, 2.30s/it] 4%|▍ | 1293/30000 [49:09<18:16:46, 2.29s/it] 4%|▍ | 1294/30000 [49:11<18:14:21, 2.29s/it] 4%|▍ | 1295/30000 [49:14<18:14:17, 2.29s/it] 4%|▍ | 1296/30000 [49:16<18:19:08, 2.30s/it] 4%|▍ | 1297/30000 [49:18<18:15:15, 2.29s/it] 4%|▍ | 1298/30000 [49:21<18:10:21, 2.28s/it] 4%|▍ | 1299/30000 [49:23<18:20:13, 2.30s/it] 4%|▍ | 1300/30000 [49:25<18:19:16, 2.30s/it] +{'loss': 0.022, 'grad_norm': 0.09237299114465714, 'learning_rate': 1.758e-05, 'flow_matching_loss': 0.022041736356914043, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.26} +{'loss': 0.0213, 'grad_norm': 0.07801612466573715, 'learning_rate': 1.778e-05, 'flow_matching_loss': 0.021292009763419627, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.26} +{'loss': 0.0228, 'grad_norm': 0.09062657505273819, 'learning_rate': 1.798e-05, 'flow_matching_loss': 0.022785918414592744, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.26} +{'loss': 0.0208, 'grad_norm': 0.09298733621835709, 'learning_rate': 1.818e-05, 'flow_matching_loss': 0.02075223308056593, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.26} +{'loss': 0.0217, 'grad_norm': 0.11384975165128708, 'learning_rate': 1.838e-05, 'flow_matching_loss': 0.021669294498860837, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.27} +{'loss': 0.0219, 'grad_norm': 0.1148248165845871, 'learning_rate': 1.8579999999999998e-05, 'flow_matching_loss': 0.021932714246213438, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.27} +{'loss': 0.0219, 'grad_norm': 0.13721060752868652, 'learning_rate': 1.878e-05, 'flow_matching_loss': 0.021899109706282616, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.27} +{'loss': 0.0218, 'grad_norm': 0.11574936658143997, 'learning_rate': 1.898e-05, 'flow_matching_loss': 0.021844013407826424, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.28} +{'loss': 0.0222, 'grad_norm': 0.08913720399141312, 'learning_rate': 1.918e-05, 'flow_matching_loss': 0.022226464096456768, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.28} +{'loss': 0.0211, 'grad_norm': 0.0918000116944313, 'learning_rate': 1.938e-05, 'flow_matching_loss': 0.021070724166929723, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.28} +{'loss': 0.0207, 'grad_norm': 0.11576782912015915, 'learning_rate': 1.958e-05, 'flow_matching_loss': 0.020725973695516587, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.29} +{'loss': 0.0217, 'grad_norm': 0.08966407924890518, 'learning_rate': 1.978e-05, 'flow_matching_loss': 0.021653497219085695, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.29} +{'loss': 0.022, 'grad_norm': 0.12441512942314148, 'learning_rate': 1.9980000000000002e-05, 'flow_matching_loss': 0.02201499305665493, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.29} +{'loss': 0.0222, 'grad_norm': 0.13032536208629608, 'learning_rate': 2.018e-05, 'flow_matching_loss': 0.022179529070854187, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.29} +{'loss': 0.0216, 'grad_norm': 0.13288572430610657, 'learning_rate': 2.038e-05, 'flow_matching_loss': 0.02164872195571661, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.3} +{'loss': 0.0213, 'grad_norm': 0.14666040241718292, 'learning_rate': 2.0580000000000003e-05, 'flow_matching_loss': 0.021323128789663314, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.3} +{'loss': 0.0211, 'grad_norm': 0.11167845129966736, 'learning_rate': 2.078e-05, 'flow_matching_loss': 0.02106987340375781, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.3} +{'loss': 0.0223, 'grad_norm': 0.09753567725419998, 'learning_rate': 2.0980000000000002e-05, 'flow_matching_loss': 0.022299323789775373, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.31} +{'loss': 0.0199, 'grad_norm': 0.13520494103431702, 'learning_rate': 2.118e-05, 'flow_matching_loss': 0.019890730641782282, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.31} +{'loss': 0.021, 'grad_norm': 0.10593137890100479, 'learning_rate': 2.1380000000000002e-05, 'flow_matching_loss': 0.02100431462749839, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.31} +{'loss': 0.0204, 'grad_norm': 0.144779771566391, 'learning_rate': 2.1580000000000003e-05, 'flow_matching_loss': 0.02043596040457487, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.31} +{'loss': 0.0188, 'grad_norm': 0.11020537465810776, 'learning_rate': 2.178e-05, 'flow_matching_loss': 0.01877970974892378, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.32} +{'loss': 0.0218, 'grad_norm': 0.15087948739528656, 'learning_rate': 2.198e-05, 'flow_matching_loss': 0.021806272957473994, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.32} +{'loss': 0.0223, 'grad_norm': 0.1395636796951294, 'learning_rate': 2.2179999999999998e-05, 'flow_matching_loss': 0.02225855067372322, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.32} +{'loss': 0.0195, 'grad_norm': 0.12086892127990723, 'learning_rate': 2.238e-05, 'flow_matching_loss': 0.019476784858852625, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.33} +{'loss': 0.0211, 'grad_norm': 0.11054575443267822, 'learning_rate': 2.258e-05, 'flow_matching_loss': 0.021062810719013215, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.33} +{'loss': 0.0188, 'grad_norm': 0.12821991741657257, 'learning_rate': 2.278e-05, 'flow_matching_loss': 0.018803967535495757, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.33} +{'loss': 0.02, 'grad_norm': 0.13383738696575165, 'learning_rate': 2.298e-05, 'flow_matching_loss': 0.020003264583647252, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.33} +{'loss': 0.0216, 'grad_norm': 0.15301638841629028, 'learning_rate': 2.3179999999999998e-05, 'flow_matching_loss': 0.021575142815709113, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.34} +{'loss': 0.0211, 'grad_norm': 0.09102558344602585, 'learning_rate': 2.338e-05, 'flow_matching_loss': 0.0210911032743752, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.34} +{'loss': 0.0227, 'grad_norm': 0.08975644409656525, 'learning_rate': 2.358e-05, 'flow_matching_loss': 0.022730602882802486, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.34} +{'loss': 0.0199, 'grad_norm': 0.08419822156429291, 'learning_rate': 2.378e-05, 'flow_matching_loss': 0.019879920966923238, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.35} +{'loss': 0.0207, 'grad_norm': 0.1551811397075653, 'learning_rate': 2.398e-05, 'flow_matching_loss': 0.020650241617113353, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.35} +{'loss': 0.0212, 'grad_norm': 0.0947592556476593, 'learning_rate': 2.4180000000000002e-05, 'flow_matching_loss': 0.02122237253934145, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.35} +{'loss': 0.0207, 'grad_norm': 0.1302022635936737, 'learning_rate': 2.438e-05, 'flow_matching_loss': 0.02069281041622162, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.35} +{'loss': 0.0201, 'grad_norm': 0.08868718892335892, 'learning_rate': 2.4580000000000002e-05, 'flow_matching_loss': 0.020118600130081175, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.36} +{'loss': 0.02, 'grad_norm': 0.09383807331323624, 'learning_rate': 2.478e-05, 'flow_matching_loss': 0.01999882245436311, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.36} +{'loss': 0.0216, 'grad_norm': 0.10378488898277283, 'learning_rate': 2.498e-05, 'flow_matching_loss': 0.021627848595380784, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.36} +{'loss': 0.0211, 'grad_norm': 0.10800045728683472, 'learning_rate': 2.5180000000000003e-05, 'flow_matching_loss': 0.02108713462948799, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.37} +{'loss': 0.0198, 'grad_norm': 0.0819505825638771, 'learning_rate': 2.538e-05, 'flow_matching_loss': 0.01979217482730746, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.37} +{'loss': 0.023, 'grad_norm': 0.10492303222417831, 'learning_rate': 2.5580000000000002e-05, 'flow_matching_loss': 0.02296731136739254, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.37} +{'loss': 0.0187, 'grad_norm': 0.12010547518730164, 'learning_rate': 2.578e-05, 'flow_matching_loss': 0.018664024583995343, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.38} +{'loss': 0.0202, 'grad_norm': 0.12036475539207458, 'learning_rate': 2.5980000000000002e-05, 'flow_matching_loss': 0.020160029642283916, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.38} 4%|▍ | 1300/30000 [49:25<18:19:16, 2.30s/it] 4%|▍ | 1301/30000 [49:27<18:17:56, 2.30s/it] 4%|▍ | 1302/30000 [49:30<18:23:15, 2.31s/it] 4%|▍ | 1303/30000 [49:32<18:19:30, 2.30s/it] 4%|▍ | 1304/30000 [49:34<18:12:49, 2.28s/it] 4%|▍ | 1305/30000 [49:37<18:05:06, 2.27s/it] 4%|▍ | 1306/30000 [49:39<18:13:35, 2.29s/it] 4%|▍ | 1307/30000 [49:41<18:11:12, 2.28s/it] 4%|▍ | 1308/30000 [49:43<18:12:33, 2.28s/it] 4%|▍ | 1309/30000 [49:46<18:18:34, 2.30s/it] 4%|▍ | 1310/30000 [49:48<18:20:51, 2.30s/it] 4%|▍ | 1310/30000 [49:48<18:20:51, 2.30s/it] 4%|▍ | 1311/30000 [49:50<18:15:38, 2.29s/it] 4%|▍ | 1312/30000 [49:53<18:09:58, 2.28s/it] 4%|▍ | 1313/30000 [49:55<18:16:12, 2.29s/it] 4%|▍ | 1314/30000 [49:57<18:06:28, 2.27s/it] 4%|▍ | 1315/30000 [49:59<18:05:13, 2.27s/it] 4%|▍ | 1316/30000 [50:02<18:17:34, 2.30s/it] 4%|▍ | 1317/30000 [50:04<18:23:40, 2.31s/it] 4%|▍ | 1318/30000 [50:06<18:17:31, 2.30s/it] 4%|▍ | 1319/30000 [50:09<18:20:23, 2.30s/it] 4%|▍ | 1320/30000 [50:11<18:16:04, 2.29s/it] 4%|▍ | 1320/30000 [50:11<18:16:04, 2.29s/it] 4%|▍ | 1321/30000 [50:13<18:12:44, 2.29s/it] 4%|▍ | 1322/30000 [50:16<18:11:59, 2.28s/it] 4%|▍ | 1323/30000 [50:18<18:13:35, 2.29s/it] 4%|▍ | 1324/30000 [50:20<18:12:13, 2.29s/it] 4%|▍ | 1325/30000 [50:22<18:15:01, 2.29s/it] 4%|▍ | 1326/30000 [50:25<18:19:01, 2.30s/it] 4%|▍ | 1327/30000 [50:27<18:19:16, 2.30s/it] 4%|▍ | 1328/30000 [50:29<18:14:05, 2.29s/it] 4%|▍ | 1329/30000 [50:32<18:08:29, 2.28s/it] 4%|▍ | 1330/30000 [50:34<18:19:45, 2.30s/it] 4%|▍ | 1330/30000 [50:34<18:19:45, 2.30s/it] 4%|▍ | 1331/30000 [50:36<18:25:39, 2.31s/it] 4%|▍ | 1332/30000 [50:38<18:10:57, 2.28s/it] 4%|▍ | 1333/30000 [50:41<18:18:57, 2.30s/it] 4%|▍ | 1334/30000 [50:43<18:19:36, 2.30s/it] 4%|▍ | 1335/30000 [50:45<18:14:54, 2.29s/it] 4%|▍ | 1336/30000 [50:48<18:10:22, 2.28s/it] 4%|▍ | 1337/30000 [50:50<18:18:36, 2.30s/it] 4%|▍ | 1338/30000 [50:52<18:08:14, 2.28s/it] 4%|▍ | 1339/30000 [50:54<18:06:35, 2.27s/it] 4%|▍ | 1340/30000 [50:57<18:17:49, 2.30s/it] 4%|▍ | 1340/30000 [50:57<18:17:49, 2.30s/it] 4%|▍ | 1341/30000 [50:59<18:17:50, 2.30s/it] 4%|▍ | 1342/30000 [51:01<18:11:49, 2.29s/it] 4%|▍ | 1343/30000 [51:04<18:06:19, 2.27s/it] 4%|▍ | 1344/30000 [51:06<18:16:32, 2.30s/it] 4%|▍ | 1345/30000 [51:08<18:19:54, 2.30s/it] 4%|▍ | 1346/30000 [51:11<18:13:40, 2.29s/it] 4%|▍ | 1347/30000 [51:13<18:19:03, 2.30s/it] 4%|▍ | 1348/30000 [51:15<18:17:27, 2.30s/it] 4%|▍ | 1349/30000 [51:17<18:09:52, 2.28s/it] 4%|▍ | 1350/30000 [51:20<18:05:37, 2.27s/it] 4%|▍ | 1350/30000 [51:20<18:05:37, 2.27s/it] 5%|▍ | 1351/30000 [51:22<18:20:56, 2.31s/it] 5%|▍ | 1352/30000 [51:24<18:21:52, 2.31s/it] 5%|▍ | 1353/30000 [51:27<18:16:31, 2.30s/it] 5%|▍ | 1354/30000 [51:29<18:05:38, 2.27s/it] 5%|▍ | 1355/30000 [51:31<18:14:52, 2.29s/it] 5%|▍ | 1356/30000 [51:34<18:22:08, 2.31s/it] 5%|▍ | 1357/30000 [51:36<18:19:57, 2.30s/it] 5%|▍ | 1358/30000 [51:38<18:18:07, 2.30s/it] 5%|▍ | 1359/30000 [51:40<18:24:51, 2.31s/it] 5%|▍ | 1360/30000 [51:43<18:25:24, 2.32s/it] 5%|▍ | 1360/30000 [51:43<18:25:24, 2.32s/it] 5%|▍ | 1361/30000 [51:45<18:15:05, 2.29s/it] 5%|▍ | 1362/30000 [51:47<18:10:13, 2.28s/it] 5%|▍ | 1363/30000 [51:50<18:17:06, 2.30s/it] 5%|▍ | 1364/30000 [51:52<18:09:47, 2.28s/it] 5%|▍ | 1365/30000 [51:54<18:10:01, 2.28s/it] 5%|▍ | 1366/30000 [51:56<18:13:13, 2.29s/it] 5%|▍ | 1367/30000 [51:59<18:13:52, 2.29s/it] 5%|▍ | 1368/30000 [52:01<18:02:34, 2.27s/it] 5%|▍ | 1369/30000 [52:03<18:04:02, 2.27s/it] 5%|▍ | 1370/30000 [52:06<18:11:06, 2.29s/it] 5%|▍ | 1370/30000 [52:06<18:11:06, 2.29s/it] 5%|▍ | 1371/30000 [52:08<18:16:14, 2.30s/it] 5%|▍ | 1372/30000 [52:10<18:07:52, 2.28s/it] 5%|▍ | 1373/30000 [52:12<18:08:57, 2.28s/it] 5%|▍ | 1374/30000 [52:15<18:12:08, 2.29s/it] 5%|▍ | 1375/30000 [52:17<18:10:59, 2.29s/it] 5%|▍ | 1376/30000 [52:19<18:10:30, 2.29s/it] 5%|▍ | 1377/30000 [52:22<18:15:54, 2.30s/it] 5%|▍ | 1378/30000 [52:24<18:13:56, 2.29s/it] 5%|▍ | 1379/30000 [52:26<18:11:19, 2.29s/it] 5%|▍ | 1380/30000 [52:28<18:10:50, 2.29s/it] 5%|▍ | 1380/30000 [52:28<18:10:50, 2.29s/it] 5%|▍ | 1381/30000 [52:31<18:21:54, 2.31s/it] 5%|▍ | 1382/30000 [52:33<18:12:47, 2.29s/it] 5%|▍ | 1383/30000 [52:35<18:11:21, 2.29s/it] 5%|▍ | 1384/30000 [52:38<18:18:50, 2.30s/it] 5%|▍ | 1385/30000 [52:40<18:18:17, 2.30s/it] 5%|▍ | 1386/30000 [52:42<18:10:56, 2.29s/it] 5%|▍ | 1387/30000 [52:44<18:05:02, 2.28s/it] 5%|▍ | 1388/30000 [52:47<18:11:44, 2.29s/it] 5%|▍ | 1389/30000 [52:49<18:04:10, 2.27s/it] 5%|▍ | 1390/30000 [52:51<18:02:24, 2.27s/it] 5%|▍ | 1390/30000 [52:51<18:02:24, 2.27s/it] 5%|▍ | 1391/30000 [52:54<18:10:23, 2.29s/it] 5%|▍ | 1392/30000 [52:56<18:11:10, 2.29s/it] 5%|▍ | 1393/30000 [52:58<18:07:03, 2.28s/it] 5%|▍ | 1394/30000 [53:01<18:12:40, 2.29s/it] 5%|▍ | 1395/30000 [53:03<18:08:19, 2.28s/it] 5%|▍ | 1396/30000 [53:05<17:57:11, 2.26s/it] 5%|▍ | 1397/30000 [53:07<18:05:38, 2.28s/it] 5%|▍ | 1398/30000 [53:10<18:05:40, 2.28s/it] 5%|▍ | 1399/30000 [53:12<18:02:22, 2.27s/it] 5%|▍ | 1400/30000 [53:14<18:03:47, 2.27s/it] 5%|▍ | 1400/30000 [53:14<18:03:47, 2.27s/it] 5%|▍ | 1401/30000 [53:16<18:11:40, 2.29s/it] 5%|▍ | 1402/30000 [53:19<18:10:14, 2.29s/it] 5%|▍ | 1403/30000 [53:21<18:13:26, 2.29s/it] 5%|▍ | 1404/30000 [53:23<18:18:53, 2.31s/it] 5%|▍ | 1405/30000 [53:26<18:19:20, 2.31s/it] 5%|▍ | 1406/30000 [53:28<18:15:05, 2.30s/it] 5%|▍ | 1407/30000 [53:30<18:13:23, 2.29s/it] 5%|▍ | 1408/30000 [53:33<18:19:20, 2.31s/it] 5%|▍ | 1409/30000 [53:35<18:18:10, 2.30s/it] 5%|▍ | 1410/30000 [53:37<18:14:47, 2.30s/it] 5%|▍ | 1410/30000 [53:37<18:14:47, 2.30s/it] 5%|▍ | 1411/30000 [53:40<18:22:57, 2.31s/it] 5%|▍ | 1412/30000 [53:42<18:15:41, 2.30s/it] 5%|▍ | 1413/30000 [53:44<18:06:17, 2.28s/it] 5%|▍ | 1414/30000 [53:46<18:09:54, 2.29s/it] 5%|▍ | 1415/30000 [53:49<18:12:30, 2.29s/it] 5%|▍ | 1416/30000 [53:51<18:04:24, 2.28s/it] 5%|▍ | 1417/30000 [53:53<18:04:18, 2.28s/it] 5%|▍ | 1418/30000 [53:55<18:13:10, 2.29s/it] 5%|▍ | 1419/30000 [53:58<18:12:37, 2.29s/it] 5%|▍ | 1420/30000 [54:00<18:08:38, 2.29s/it] 5%|▍ | 1420/30000 [54:00<18:08:38, 2.29s/it] 5%|▍ | 1421/30000 [54:02<18:15:26, 2.30s/it] 5%|▍ | 1422/30000 [54:05<18:16:39, 2.30s/it] 5%|▍ | 1423/30000 [54:07<18:13:40, 2.30s/it] 5%|▍ | 1424/30000 [54:09<18:10:02, 2.29s/it] 5%|▍ | 1425/30000 [54:12<18:16:39, 2.30s/it] 5%|▍ | 1426/30000 [54:14<18:21:19, 2.31s/it] 5%|▍ | 1427/30000 [54:16<18:19:49, 2.31s/it] 5%|▍ | 1428/30000 [54:19<18:23:24, 2.32s/it] 5%|▍ | 1429/30000 [54:21<18:21:16, 2.31s/it] 5%|▍ | 1430/30000 [54:23<18:11:51, 2.29s/it] 5%|▍ | 1430/30000 [54:23<18:11:51, 2.29s/it] 5%|▍ | 1431/30000 [54:25<18:11:18, 2.29s/it] 5%|▍ | 1432/30000 [54:28<18:17:51, 2.31s/it] 5%|▍ | 1433/30000 [54:30<18:12:35, 2.29s/it] 5%|▍ | 1434/30000 [54:32<18:08:32, 2.29s/it] 5%|▍ | 1435/30000 [54:35<18:17:58, 2.31s/it] 5%|▍ | 1436/30000 [54:37<18:21:05, 2.31s/it] 5%|▍ | 1437/30000 [54:39<18:11:27, 2.29s/it] 5%|▍ | 1438/30000 [54:42<18:17:01, 2.30s/it] 5%|▍ | 1439/30000 [54:44<18:14:01, 2.30s/it] 5%|▍ | 1440/30000 [54:46<18:08:47, 2.29s/it] 5%|▍ | 1440/30000 [54:46<18:08:47, 2.29s/it] 5%|▍ | 1441/30000 [54:48<18:12:36, 2.30s/it] 5%|▍ | 1442/30000 [54:51<18:13:22, 2.30s/it] 5%|▍ | 1443/30000 [54:53<18:05:22, 2.28s/it] 5%|▍ | 1444/30000 [54:55<18:02:58, 2.28s/it] 5%|▍ | 1445/30000 [54:57<18:08:00, 2.29s/it] 5%|▍ | 1446/30000 [55:00<18:07:28, 2.29s/it] 5%|▍ | 1447/30000 [55:02<18:06:17, 2.28s/it] 5%|▍ | 1448/30000 [55:04<18:15:06, 2.30s/it] 5%|▍ | 1449/30000 [55:07<18:05:32, 2.28s/it] 5%|▍ | 1450/30000 [55:09<17:59:10, 2.27s/it] 5%|▍ | 1450/30000 [55:09<17:59:10, 2.27s/it] 5%|▍ | 1451/30000 [55:11<18:12:19, 2.30s/it] 5%|▍ | 1452/30000 [55:14<18:11:51, 2.29s/it] 5%|▍ | 1453/30000 [55:16<18:09:46, 2.29s/it] 5%|▍ | 1454/30000 [55:18<18:24:00, 2.32s/it] 5%|▍ | 1455/30000 [55:20<18:22:48, 2.32s/it] 5%|▍ | 1456/30000 [55:23<18:15:38, 2.30s/it] 5%|▍ | 1457/30000 [55:25<18:14:58, 2.30s/it] 5%|▍ | 1458/30000 [55:27<18:21:29, 2.32s/it] 5%|▍ | 1459/30000 [55:30<18:18:41, 2.31s/it] 5%|▍ | 1460/30000 [55:32<18:09:20, 2.29s/it] 5%|▍ | 1460/30000 [55:32<18:09:20, 2.29s/it] 5%|▍ | 1461/30000 [55:34<18:15:00, 2.30s/it] 5%|▍ | 1462/30000 [55:37<18:07:23, 2.29s/it] 5%|▍ | 1463/30000 [55:39<18:03:55, 2.28s/it] 5%|▍ | 1464/30000 [55:41<18:10:36, 2.29s/it] 5%|▍ | 1465/30000 [55:43<18:08:13, 2.29s/it] 5%|▍ | 1466/30000 [55:46<18:01:13, 2.27s/it] 5%|▍ | 1467/30000 [55:48<18:08:53, 2.29s/it] 5%|▍ | 1468/30000 [55:50<18:14:19, 2.30s/it] 5%|▍ | 1469/30000 [55:53<18:07:44, 2.29s/it] 5%|▍ | 1470/30000 [55:55<18:14:07, 2.30s/it] 5%|▍ | 1470/30000 [55:55<18:14:07, 2.30s/it] 5%|▍ | 1471/30000 [55:57<18:10:38, 2.29s/it] 5%|▍ | 1472/30000 [55:59<18:03:01, 2.28s/it] 5%|▍ | 1473/30000 [56:02<18:09:12, 2.29s/it] 5%|▍ | 1474/30000 [56:04<18:15:23, 2.30s/it] 5%|▍ | 1475/30000 [56:06<18:08:46, 2.29s/it] 5%|▍ | 1476/30000 [56:09<18:07:21, 2.29s/it] 5%|▍ | 1477/30000 [56:11<18:17:04, 2.31s/it] 5%|▍ | 1478/30000 [56:13<18:09:54, 2.29s/it] 5%|▍ | 1479/30000 [56:15<18:07:50, 2.29s/it] 5%|▍ | 1480/30000 [56:18<18:13:11, 2.30s/it] 5%|▍ | 1480/30000 [56:18<18:13:11, 2.30s/it] 5%|▍ | 1481/30000 [56:20<18:15:11, 2.30s/it] 5%|▍ | 1482/30000 [56:22<18:10:26, 2.29s/it] 5%|▍ | 1483/30000 [56:25<18:09:20, 2.29s/it] 5%|▍ | 1484/30000 [56:27<18:18:29, 2.31s/it] 5%|▍ | 1485/30000 [56:29<18:14:01, 2.30s/it] 5%|▍ | 1486/30000 [56:32<18:09:31, 2.29s/it] 5%|▍ | 1487/30000 [56:34<18:13:53, 2.30s/it] 5%|▍ | 1488/30000 [56:36<18:11:02, 2.30s/it] 5%|▍ | 1489/30000 [56:38<18:06:46, 2.29s/it] 5%|▍ | 1490/30000 [56:41<18:05:39, 2.28s/it] 5%|▍ | 1490/30000 [56:41<18:05:39, 2.28s/it] 5%|▍ | 1491/30000 [56:43<18:08:54, 2.29s/it] 5%|▍ | 1492/30000 [56:45<18:05:56, 2.29s/it] 5%|▍ | 1493/30000 [56:48<18:00:20, 2.27s/it] 5%|▍ | 1494/30000 [56:50<18:10:17, 2.29s/it] 5%|▍ | 1495/30000 [56:52<18:06:30, 2.29s/it] 5%|▍ | 1496/30000 [56:54<18:03:32, 2.28s/it] 5%|▍ | 1497/30000 [56:57<18:15:46, 2.31s/it] 5%|▍ | 1498/30000 [56:59<18:14:08, 2.30s/it] 5%|▍ | 1499/30000 [57:01<18:06:20, 2.29s/it] 5%|▌ | 1500/30000 [57:04<18:03:58, 2.28s/it] 5%|▌ | 1500/30000 [57:04<18:03:58, 2.28s/it] 5%|▌ | 1501/30000 [57:06<18:08:35, 2.29s/it] 5%|▌ | 1502/30000 [57:08<18:02:54, 2.28s/it] 5%|▌ | 1503/30000 [57:10<18:06:15, 2.29s/it] 5%|▌ | 1504/30000 [57:13<18:14:24, 2.30s/it] 5%|▌ | 1505/30000 [57:15<18:18:28, 2.31s/it] 5%|▌ | 1506/30000 [57:17<18:12:55, 2.30s/it] 5%|▌ | 1507/30000 [57:20<18:04:31, 2.28s/it] 5%|▌ | 1508/30000 [57:22<18:12:52, 2.30s/it] 5%|▌ | 1509/30000 [57:24<18:11:08, 2.30s/it] 5%|▌ | 1510/30000 [57:27<18:05:06, 2.29s/it] 5%|▌ | 1510/30000 [57:27<18:05:06, 2.29s/it] 5%|▌ | 1511/30000 [57:29<18:06:08, 2.29s/it] 5%|▌ | 1512/30000 [57:31<18:10:23, 2.30s/it] 5%|▌ | 1513/30000 [57:33<18:00:37, 2.28s/it] 5%|▌ | 1514/30000 [57:36<17:58:00, 2.27s/it] 5%|▌ | 1515/30000 [57:38<18:11:12, 2.30s/it] 5%|▌ | 1516/30000 [57:40<18:13:49, 2.30s/it] 5%|▌ | 1517/30000 [57:43<18:09:04, 2.29s/it] 5%|▌ | 1518/30000 [57:45<18:06:35, 2.29s/it] 5%|▌ | 1519/30000 [57:47<18:12:52, 2.30s/it] 5%|▌ | 1520/30000 [57:50<18:17:47, 2.31s/it] 5%|▌ | 1520/30000 [57:50<18:17:47, 2.31s/it] 5%|▌ | 1521/30000 [57:52<18:11:55, 2.30s/it] 5%|▌ | 1522/30000 [57:54<18:06:23, 2.29s/it] 5%|▌ | 1523/30000 [57:56<18:14:27, 2.31s/it] 5%|▌ | 1524/30000 [57:59<18:15:14, 2.31s/it] 5%|▌ | 1525/30000 [58:01<18:08:36, 2.29s/it] 5%|▌ | 1526/30000 [58:03<18:02:58, 2.28s/it] 5%|▌ | 1527/30000 [58:06<18:12:37, 2.30s/it] 5%|▌ | 1528/30000 [58:08<18:13:18, 2.30s/it] 5%|▌ | 1529/30000 [58:10<18:04:33, 2.29s/it] 5%|▌ | 1530/30000 [58:12<18:07:44, 2.29s/it] 5%|▌ | 1530/30000 [58:13<18:07:44, 2.29s/it] 5%|▌ | 1531/30000 [58:15<18:06:30, 2.29s/it] 5%|▌ | 1532/30000 [58:17<18:03:47, 2.28s/it] 5%|▌ | 1533/30000 [58:19<18:03:08, 2.28s/it] 5%|▌ | 1534/30000 [58:22<18:12:24, 2.30s/it] 5%|▌ | 1535/30000 [58:24<18:11:34, 2.30s/it] 5%|▌ | 1536/30000 [58:26<18:04:57, 2.29s/it] 5%|▌ | 1537/30000 [58:28<18:02:18, 2.28s/it] 5%|▌ | 1538/30000 [58:31<18:04:22, 2.29s/it] 5%|▌ | 1539/30000 [58:33<17:59:21, 2.28s/it] 5%|▌ | 1540/30000 [58:35<17:51:42, 2.26s/it] 5%|▌ | 1540/30000 [58:35<17:51:42, 2.26s/it] 5%|▌ | 1541/30000 [58:38<17:59:58, 2.28s/it] 5%|▌ | 1542/30000 [58:40<18:04:48, 2.29s/it] 5%|▌ | 1543/30000 [58:42<18:04:11, 2.29s/it] 5%|▌ | 1544/30000 [58:44<18:00:30, 2.28s/it] 5%|▌ | 1545/30000 [58:47<18:04:52, 2.29s/it] 5%|▌ | 1546/30000 [58:49<18:10:18, 2.30s/it] 5%|▌ | 1547/30000 [58:51<18:06:36, 2.29s/it] 5%|▌ | 1548/30000 [58:54<18:02:36, 2.28s/it] 5%|▌ | 1549/30000 [58:56<18:08:01, 2.29s/it] 5%|▌ | 1550/30000 [58:58<18:06:59, 2.29s/it] 5%|▌ | 1550/30000 [58:58<18:06:59, 2.29s/it] 5%|▌ | 1551/30000 [59:00<18:02:50, 2.28s/it] 5%|▌ | 1552/30000 [59:03<18:07:06, 2.29s/it] 5%|▌ | 1553/30000 [59:05<18:13:15, 2.31s/it] 5%|▌ | 1554/30000 [59:07<18:06:56, 2.29s/it] 5%|▌ | 1555/30000 [59:10<18:05:03, 2.29s/it] 5%|▌ | 1556/30000 [59:12<18:10:06, 2.30s/it] 5%|▌ | 1557/30000 [59:14<18:09:03, 2.30s/it] 5%|▌ | 1558/30000 [59:17<18:09:47, 2.30s/it] 5%|▌ | 1559/30000 [59:19<18:05:08, 2.29s/it] 5%|▌ | 1560/30000 [59:21<18:10:46, 2.30s/it] 5%|▌ | 1560/30000 [59:21<18:10:46, 2.30s/it] 5%|▌ | 1561/30000 [59:23<18:10:11, 2.30s/it] 5%|▌ | 1562/30000 [59:26<18:03:48, 2.29s/it] 5%|▌ | 1563/30000 [59:28<17:53:38, 2.27s/it] 5%|▌ | 1564/30000 [59:30<18:02:49, 2.28s/it] 5%|▌ | 1565/30000 [59:33<18:07:36, 2.29s/it] 5%|▌ | 1566/30000 [59:35<18:09:01, 2.30s/it] 5%|▌ | 1567/30000 [59:37<18:10:20, 2.30s/it] 5%|▌ | 1568/30000 [59:40<18:14:40, 2.31s/it] 5%|▌ | 1569/30000 [59:42<18:07:45, 2.30s/it] 5%|▌ | 1570/30000 [59:44<18:04:28, 2.29s/it] 5%|▌ | 1570/30000 [59:44<18:04:28, 2.29s/it] 5%|▌ | 1571/30000 [59:46<18:11:09, 2.30s/it] 5%|▌ | 1572/30000 [59:49<18:11:24, 2.30s/it] 5%|▌ | 1573/30000 [59:51<18:05:48, 2.29s/it] 5%|▌ | 1574/30000 [59:53<18:06:30, 2.29s/it] 5%|▌ | 1575/30000 [59:56<18:02:34, 2.29s/it] 5%|▌ | 1576/30000 [59:58<18:00:14, 2.28s/it] 5%|▌ | 1577/30000 [1:00:00<18:01:20, 2.28s/it] 5%|▌ | 1578/30000 [1:00:02<18:06:58, 2.29s/it] 5%|▌ | 1579/30000 [1:00:05<18:00:35, 2.28s/it] 5%|▌ | 1580/30000 [1:00:07<17:59:17, 2.28s/it] 5%|▌ | 1580/30000 [1:00:07<17:59:17, 2.28s/it] 5%|▌ | 1581/30000 [1:00:09<18:08:11, 2.30s/it] 5%|▌ | 1582/30000 [1:00:12<18:12:14, 2.31s/it] 5%|▌ | 1583/30000 [1:00:14<18:06:18, 2.29s/it] 5%|▌ | 1584/30000 [1:00:16<18:09:20, 2.30s/it] 5%|▌ | 1585/30000 [1:00:18<18:06:27, 2.29s/it] 5%|▌ | 1586/30000 [1:00:21<17:59:19, 2.28s/it] 5%|▌ | 1587/30000 [1:00:23<18:01:59, 2.28s/it] 5%|▌ | 1588/30000 [1:00:25<18:03:58, 2.29s/it] 5%|▌ | 1589/30000 [1:00:28<17:58:58, 2.28s/it] 5%|▌ | 1590/30000 [1:00:30<17:59:24, 2.28s/it] 5%|▌ | 1590/30000 [1:00:30<17:59:24, 2.28s/it] 5%|▌ | 1591/30000 [1:00:32<18:08:13, 2.30s/it] 5%|▌ | 1592/30000 [1:00:34<18:06:48, 2.30s/it] 5%|▌ | 1593/30000 [1:00:37<17:59:17, 2.28s/it] 5%|▌ | 1594/30000 [1:00:39<18:08:38, 2.30s/it] 5%|▌ | 1595/30000 [1:00:41<18:08:37, 2.30s/it] 5%|▌ | 1596/30000 [1:00:44<18:01:48, 2.29s/it] 5%|▌ | 1597/30000 [1:00:46<18:09:37, 2.30s/it] 5%|▌ | 1598/30000 [1:00:48<18:08:14, 2.30s/it] 5%|▌ | 1599/30000 [1:00:50<17:59:41, 2.28s/it] 5%|▌ | 1600/30000 [1:00:53<18:08:05, 2.30s/it] 5%|▌ | 1600/30000 [1:00:53<18:08:05, 2.30s/it] 5%|▌ | 1601/30000 [1:00:55<18:09:22, 2.30s/it] 5%|▌ | 1602/30000 [1:00:57<18:01:59, 2.29s/it] 5%|▌ | 1603/30000 [1:01:00<18:08:07, 2.30s/it] 5%|▌ | 1604/30000 [1:01:02<18:10:12, 2.30s/it] 5%|▌ | 1605/30000 [1:01:04<18:00:06, 2.28s/it] 5%|▌ | 1606/30000 [1:01:07<18:11:37, 2.31s/it] 5%|▌ | 1607/30000 [1:01:09<18:11:14, 2.31s/it] 5%|▌ | 1608/30000 [1:01:11<18:06:38, 2.30s/it] 5%|▌ | 1609/30000 [1:01:14<18:11:53, 2.31s/it] 5%|▌ | 1610/30000 [1:01:16<18:11:15, 2.31s/it] 5%|▌ | 1610/30000 [1:01:16<18:11:15, 2.31s/it] 5%|▌ | 1611/30000 [1:01:18<18:07:05, 2.30s/it] 5%|▌ | 1612/30000 [1:01:20<18:10:26, 2.30s/it] 5%|▌ | 1613/30000 [1:01:23<18:06:03, 2.30s/it] 5%|▌ | 1614/30000 [1:01:25<17:57:04, 2.28s/it] 5%|▌ | 1615/30000 [1:01:27<18:04:28, 2.29s/it] 5%|▌ | 1616/30000 [1:01:30<18:02:19, 2.29s/it] 5%|▌ | 1617/30000 [1:01:32<18:01:09, 2.29s/it] 5%|▌ | 1618/30000 [1:01:34<17:59:30, 2.28s/it] 5%|▌ | 1619/30000 [1:01:36<18:05:17, 2.29s/it] 5%|▌ | 1620/30000 [1:01:39<18:06:48, 2.30s/it] 5%|▌ | 1620/30000 [1:01:39<18:06:48, 2.30s/it] 5%|▌ | 1621/30000 [1:01:41<18:02:20, 2.29s/it] 5%|▌ | 1622/30000 [1:01:43<18:09:54, 2.30s/it] 5%|▌ | 1623/30000 [1:01:46<18:07:19, 2.30s/it] 5%|▌ | 1624/30000 [1:01:48<17:58:30, 2.28s/it] 5%|▌ | 1625/30000 [1:01:50<18:04:29, 2.29s/it] 5%|▌ | 1626/30000 [1:01:53<18:05:26, 2.30s/it] 5%|▌ | 1627/30000 [1:01:55<18:01:48, 2.29s/it] 5%|▌ | 1628/30000 [1:01:57<18:06:17, 2.30s/it] 5%|▌ | 1629/30000 [1:01:59<18:04:40, 2.29s/it] 5%|▌ | 1630/30000 [1:02:02<18:03:00, 2.29s/it] 5%|▌ | 1630/30000 [1:02:02<18:03:00, 2.29s/it] 5%|▌ | 1631/30000 [1:02:04<18:12:17, 2.31s/it] 5%|▌ | 1632/30000 [1:02:06<18:03:36, 2.29s/it] 5%|▌ | 1633/30000 [1:02:09<17:59:21, 2.28s/it] 5%|▌ | 1634/30000 [1:02:11<17:57:32, 2.28s/it] 5%|▌ | 1635/30000 [1:02:13<18:08:41, 2.30s/it] 5%|▌ | 1636/30000 [1:02:15<18:02:43, 2.29s/it] 5%|▌ | 1637/30000 [1:02:18<17:59:26, 2.28s/it] 5%|▌ | 1638/30000 [1:02:20<18:06:48, 2.30s/it] 5%|▌ | 1639/30000 [1:02:22<18:12:06, 2.31s/it] 5%|▌ | 1640/30000 [1:02:25<18:08:40, 2.30s/it] 5%|▌ | 1640/30000 [1:02:25<18:08:40, 2.30s/it] 5%|▌ | 1641/30000 [1:02:27<18:10:43, 2.31s/it] 5%|▌ | 1642/30000 [1:02:29<18:10:41, 2.31s/it] 5%|▌ | 1643/30000 [1:02:32<18:09:28, 2.31s/it] 5%|▌ | 1644/30000 [1:02:34<18:07:50, 2.30s/it] 5%|▌ | 1645/30000 [1:02:36<18:18:36, 2.32s/it] 5%|▌ | 1646/30000 [1:02:39<18:17:49, 2.32s/it] 5%|▌ | 1647/30000 [1:02:41<18:12:16, 2.31s/it] 5%|▌ | 1648/30000 [1:02:43<18:13:22, 2.31s/it] 5%|▌ | 1649/30000 [1:02:45<18:12:19, 2.31s/it] 6%|▌ | 1650/30000 [1:02:48<18:05:50, 2.30s/it] 6%|▌ | 1650/30000 [1:02:48<18:05:50, 2.30s/it] 6%|▌ | 1651/30000 [1:02:50<18:10:34, 2.31s/it] 6%|▌ | 1652/30000 [1:02:52<18:11:01, 2.31s/it] 6%|▌ | 1653/30000 [1:02:55<18:06:14, 2.30s/it] 6%|▌ | 1654/30000 [1:02:57<18:01:03, 2.29s/it] 6%|▌ | 1655/30000 [1:02:59<17:57:20, 2.28s/it] 6%|▌ | 1656/30000 [1:03:01<17:58:21, 2.28s/it] 6%|▌ | 1657/30000 [1:03:04<18:01:44, 2.29s/it] 6%|▌ | 1658/30000 [1:03:06<18:07:27, 2.30s/it] 6%|▌ | 1659/30000 [1:03:08<18:07:43, 2.30s/it] 6%|▌ | 1660/30000 [1:03:11<18:01:44, 2.29s/it] 6%|▌ | 1660/30000 [1:03:11<18:01:44, 2.29s/it] 6%|▌ | 1661/30000 [1:03:13<18:07:22, 2.30s/it] 6%|▌ | 1662/30000 [1:03:15<18:05:17, 2.30s/it] 6%|▌ | 1663/30000 [1:03:18<18:01:43, 2.29s/it] 6%|▌ | 1664/30000 [1:03:20<18:02:58, 2.29s/it] 6%|▌ | 1665/30000 [1:03:22<18:01:24, 2.29s/it] 6%|▌ | 1666/30000 [1:03:24<17:58:16, 2.28s/it] 6%|▌ | 1667/30000 [1:03:27<17:58:30, 2.28s/it] 6%|▌ | 1668/30000 [1:03:29<17:53:48, 2.27s/it] 6%|▌ | 1669/30000 [1:03:31<18:02:12, 2.29s/it] 6%|▌ | 1670/30000 [1:03:34<17:58:14, 2.28s/it] 6%|▌ | 1670/30000 [1:03:34<17:58:14, 2.28s/it] 6%|▌ | 1671/30000 [1:03:36<17:56:59, 2.28s/it] 6%|▌ | 1672/30000 [1:03:38<18:02:44, 2.29s/it] 6%|▌ | 1673/30000 [1:03:40<17:58:01, 2.28s/it] 6%|▌ | 1674/30000 [1:03:43<17:56:31, 2.28s/it] 6%|▌ | 1675/30000 [1:03:45<17:55:31, 2.28s/it] 6%|▌ | 1676/30000 [1:03:47<18:04:21, 2.30s/it] 6%|▌ | 1677/30000 [1:03:50<18:04:19, 2.30s/it] 6%|▌ | 1678/30000 [1:03:52<17:59:59, 2.29s/it] 6%|▌ | 1679/30000 [1:03:54<17:55:32, 2.28s/it] 6%|▌ | 1680/30000 [1:03:56<18:05:33, 2.30s/it] 6%|▌ | 1680/30000 [1:03:56<18:05:33, 2.30s/it] 6%|▌ | 1681/30000 [1:03:59<18:08:23, 2.31s/it] 6%|▌ | 1682/30000 [1:04:01<18:05:05, 2.30s/it] 6%|▌ | 1683/30000 [1:04:03<18:01:28, 2.29s/it] 6%|▌ | 1684/30000 [1:04:06<18:06:27, 2.30s/it] 6%|▌ | 1685/30000 [1:04:08<18:10:12, 2.31s/it] 6%|▌ | 1686/30000 [1:04:10<18:09:39, 2.31s/it] 6%|▌ | 1687/30000 [1:04:13<18:03:32, 2.30s/it] 6%|▌ | 1688/30000 [1:04:15<18:09:13, 2.31s/it] 6%|▌ | 1689/30000 [1:04:17<18:10:41, 2.31s/it] 6%|▌ | 1690/30000 [1:04:19<18:02:18, 2.29s/it] 6%|▌ | 1690/30000 [1:04:20<18:02:18, 2.29s/it] 6%|▌ | 1691/30000 [1:04:22<17:57:55, 2.28s/it] 6%|▌ | 1692/30000 [1:04:24<18:06:09, 2.30s/it] 6%|▌ | 1693/30000 [1:04:26<18:03:54, 2.30s/it] 6%|▌ | 1694/30000 [1:04:29<17:59:42, 2.29s/it] 6%|▌ | 1695/30000 [1:04:31<18:05:38, 2.30s/it] 6%|▌ | 1696/30000 [1:04:33<18:07:39, 2.31s/it] 6%|▌ | 1697/30000 [1:04:35<16:28:35, 2.10s/it] 6%|▌ | 1698/30000 [1:04:37<16:53:36, 2.15s/it] 6%|▌ | 1699/30000 [1:04:39<17:19:44, 2.20s/it] 6%|▌ | 1700/30000 [1:04:42<17:34:39, 2.24s/it] 6%|▌ | 1700/30000 [1:04:42<17:34:39, 2.24s/it] 6%|▌ | 1701/30000 [1:04:44<17:40:12, 2.25s/it] 6%|▌ | 1702/30000 [1:04:46<17:44:29, 2.26s/it] 6%|▌ | 1703/30000 [1:04:49<17:55:33, 2.28s/it] 6%|▌ | 1704/30000 [1:04:51<17:58:04, 2.29s/it] 6%|▌ | 1705/30000 [1:04:53<17:49:12, 2.27s/it] 6%|▌ | 1706/30000 [1:04:56<18:00:04, 2.29s/it] 6%|▌ | 1707/30000 [1:04:58<17:56:51, 2.28s/it] 6%|▌ | 1708/30000 [1:05:00<17:58:47, 2.29s/it] 6%|▌ | 1709/30000 [1:05:02<18:08:03, 2.31s/it] 6%|▌ | 1710/30000 [1:05:05<18:05:36, 2.30s/it] 6%|▌ | 1710/30000 [1:05:05<18:05:36, 2.30s/it] 6%|▌ | 1711/30000 [1:05:07<18:01:08, 2.29s/it] 6%|▌ | 1712/30000 [1:05:09<18:06:14, 2.30s/it] 6%|▌ | 1713/30000 [1:05:12<18:06:49, 2.31s/it] 6%|▌ | 1714/30000 [1:05:14<18:03:35, 2.30s/it] 6%|▌ | 1715/30000 [1:05:16<18:02:04, 2.30s/it] 6%|▌ | 1716/30000 [1:05:19<18:00:08, 2.29s/it] 6%|▌ | 1717/30000 [1:05:21<17:57:39, 2.29s/it] 6%|▌ | 1718/30000 [1:05:23<17:57:29, 2.29s/it] 6%|▌ | 1719/30000 [1:05:25<18:05:23, 2.30s/it] 6%|▌ | 1720/30000 [1:05:28<18:00:30, 2.29s/it] +{'loss': 0.0202, 'grad_norm': 0.10722903907299042, 'learning_rate': 2.618e-05, 'flow_matching_loss': 0.020206746086478232, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.38} +{'loss': 0.021, 'grad_norm': 0.11839815229177475, 'learning_rate': 2.6379999999999998e-05, 'flow_matching_loss': 0.020989946834743024, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.38} +{'loss': 0.0215, 'grad_norm': 0.10365170985460281, 'learning_rate': 2.658e-05, 'flow_matching_loss': 0.021519336104393005, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.39} +{'loss': 0.0201, 'grad_norm': 0.11617409437894821, 'learning_rate': 2.678e-05, 'flow_matching_loss': 0.020084141939878463, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.39} +{'loss': 0.0217, 'grad_norm': 0.09898514300584793, 'learning_rate': 2.698e-05, 'flow_matching_loss': 0.021676960028707982, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.39} +{'loss': 0.0201, 'grad_norm': 0.12194157391786575, 'learning_rate': 2.718e-05, 'flow_matching_loss': 0.020147782191634177, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.4} +{'loss': 0.0196, 'grad_norm': 0.10781208425760269, 'learning_rate': 2.738e-05, 'flow_matching_loss': 0.01957361977547407, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.4} +{'loss': 0.0182, 'grad_norm': 0.09992589056491852, 'learning_rate': 2.758e-05, 'flow_matching_loss': 0.018152609560638668, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.4} +{'loss': 0.0202, 'grad_norm': 0.09094756096601486, 'learning_rate': 2.778e-05, 'flow_matching_loss': 0.02023630989715457, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.4} +{'loss': 0.0197, 'grad_norm': 0.10377799719572067, 'learning_rate': 2.798e-05, 'flow_matching_loss': 0.019676712807267904, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.41} +{'loss': 0.0224, 'grad_norm': 0.13727609813213348, 'learning_rate': 2.818e-05, 'flow_matching_loss': 0.022427573427557947, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.41} +{'loss': 0.0193, 'grad_norm': 0.09035764634609222, 'learning_rate': 2.838e-05, 'flow_matching_loss': 0.019265454821288587, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.41} +{'loss': 0.021, 'grad_norm': 0.11919813603162766, 'learning_rate': 2.858e-05, 'flow_matching_loss': 0.021049279905855655, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.42} +{'loss': 0.0196, 'grad_norm': 0.10342154651880264, 'learning_rate': 2.8780000000000002e-05, 'flow_matching_loss': 0.019644626323133706, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.42} +{'loss': 0.0207, 'grad_norm': 0.09389200806617737, 'learning_rate': 2.898e-05, 'flow_matching_loss': 0.020715662091970444, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.42} +{'loss': 0.0209, 'grad_norm': 0.12466754019260406, 'learning_rate': 2.9180000000000002e-05, 'flow_matching_loss': 0.020945943519473077, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.42} +{'loss': 0.0225, 'grad_norm': 0.17169566452503204, 'learning_rate': 2.938e-05, 'flow_matching_loss': 0.02250794190913439, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.43} +{'loss': 0.0196, 'grad_norm': 0.11155330389738083, 'learning_rate': 2.958e-05, 'flow_matching_loss': 0.019605210702866315, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.43} +{'loss': 0.0189, 'grad_norm': 0.11541755497455597, 'learning_rate': 2.9780000000000003e-05, 'flow_matching_loss': 0.018920379504561425, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.43} +{'loss': 0.0193, 'grad_norm': 0.12945598363876343, 'learning_rate': 2.998e-05, 'flow_matching_loss': 0.019303594809025527, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.44} +{'loss': 0.0193, 'grad_norm': 0.11383146047592163, 'learning_rate': 2.9999992618302022e-05, 'flow_matching_loss': 0.01928238393738866, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.44} +{'loss': 0.02, 'grad_norm': 0.15020206570625305, 'learning_rate': 2.999996710133069e-05, 'flow_matching_loss': 0.02000527232885361, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.44} +{'loss': 0.0192, 'grad_norm': 0.12204582989215851, 'learning_rate': 2.9999923357984923e-05, 'flow_matching_loss': 0.01922872643917799, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.45} +{'loss': 0.0193, 'grad_norm': 0.10549847036600113, 'learning_rate': 2.9999861388317874e-05, 'flow_matching_loss': 0.019270368199795485, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.45} +{'loss': 0.0216, 'grad_norm': 0.13807307183742523, 'learning_rate': 2.9999781192404847e-05, 'flow_matching_loss': 0.021559222601354122, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.45} +{'loss': 0.0194, 'grad_norm': 0.13433963060379028, 'learning_rate': 2.9999682770343283e-05, 'flow_matching_loss': 0.01943699046969414, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.45} +{'loss': 0.0213, 'grad_norm': 0.10999911278486252, 'learning_rate': 2.9999566122252777e-05, 'flow_matching_loss': 0.02126980908215046, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.46} +{'loss': 0.0185, 'grad_norm': 0.11278800666332245, 'learning_rate': 2.9999431248275065e-05, 'flow_matching_loss': 0.0185450810007751, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.46} +{'loss': 0.0182, 'grad_norm': 0.10676553100347519, 'learning_rate': 2.9999278148574027e-05, 'flow_matching_loss': 0.01823292300105095, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.46} +{'loss': 0.0196, 'grad_norm': 0.09251893311738968, 'learning_rate': 2.9999106823335705e-05, 'flow_matching_loss': 0.01963558318093419, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.47} +{'loss': 0.0196, 'grad_norm': 0.14447124302387238, 'learning_rate': 2.999891727276827e-05, 'flow_matching_loss': 0.019634858518838883, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.47} +{'loss': 0.0173, 'grad_norm': 0.11315406858921051, 'learning_rate': 2.9998709497102037e-05, 'flow_matching_loss': 0.01726821893826127, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.47} +{'loss': 0.0208, 'grad_norm': 0.12515220046043396, 'learning_rate': 2.9998483496589485e-05, 'flow_matching_loss': 0.020811807550489902, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.47} +{'loss': 0.0204, 'grad_norm': 0.10874652117490768, 'learning_rate': 2.9998239271505217e-05, 'flow_matching_loss': 0.020409489050507545, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.48} +{'loss': 0.0222, 'grad_norm': 0.09445799142122269, 'learning_rate': 2.9997976822145995e-05, 'flow_matching_loss': 0.022171822655946017, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.48} +{'loss': 0.0199, 'grad_norm': 0.11820948868989944, 'learning_rate': 2.9997696148830714e-05, 'flow_matching_loss': 0.019888615421950816, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.48} +{'loss': 0.0198, 'grad_norm': 0.08661482483148575, 'learning_rate': 2.9997397251900423e-05, 'flow_matching_loss': 0.019783002324402332, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.49} +{'loss': 0.0187, 'grad_norm': 0.1063147559762001, 'learning_rate': 2.999708013171831e-05, 'flow_matching_loss': 0.018650043196976183, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.49} +{'loss': 0.0197, 'grad_norm': 0.11096528172492981, 'learning_rate': 2.9996744788669705e-05, 'flow_matching_loss': 0.019747700728476047, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.49} +{'loss': 0.0206, 'grad_norm': 0.11614083498716354, 'learning_rate': 2.999639122316208e-05, 'flow_matching_loss': 0.020599117409437893, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.49} +{'loss': 0.0165, 'grad_norm': 0.11915189772844315, 'learning_rate': 2.999601943562505e-05, 'flow_matching_loss': 0.01648150905966759, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.5} +{'loss': 0.0215, 'grad_norm': 0.12210331857204437, 'learning_rate': 2.9995629426510377e-05, 'flow_matching_loss': 0.0214836566708982, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.5} 6%|▌ | 1720/30000 [1:05:28<18:00:30, 2.29s/it] 6%|▌ | 1721/30000 [1:05:30<17:54:31, 2.28s/it] 6%|▌ | 1722/30000 [1:05:32<18:02:36, 2.30s/it] 6%|▌ | 1723/30000 [1:05:35<17:51:05, 2.27s/it] 6%|▌ | 1724/30000 [1:05:37<17:47:22, 2.26s/it] 6%|▌ | 1725/30000 [1:05:39<17:56:45, 2.28s/it] 6%|▌ | 1726/30000 [1:05:41<17:54:52, 2.28s/it] 6%|▌ | 1727/30000 [1:05:44<17:56:46, 2.29s/it] 6%|▌ | 1728/30000 [1:05:46<18:05:05, 2.30s/it] 6%|▌ | 1729/30000 [1:05:48<18:05:29, 2.30s/it] 6%|▌ | 1730/30000 [1:05:51<17:58:53, 2.29s/it] 6%|▌ | 1730/30000 [1:05:51<17:58:53, 2.29s/it] 6%|▌ | 1731/30000 [1:05:53<18:04:44, 2.30s/it] 6%|▌ | 1732/30000 [1:05:55<18:06:00, 2.31s/it] 6%|▌ | 1733/30000 [1:05:58<18:05:21, 2.30s/it] 6%|▌ | 1734/30000 [1:06:00<18:10:34, 2.31s/it] 6%|▌ | 1735/30000 [1:06:02<18:08:02, 2.31s/it] 6%|▌ | 1736/30000 [1:06:04<18:02:22, 2.30s/it] 6%|▌ | 1737/30000 [1:06:07<17:58:40, 2.29s/it] 6%|▌ | 1738/30000 [1:06:09<18:05:21, 2.30s/it] 6%|▌ | 1739/30000 [1:06:11<18:00:02, 2.29s/it] 6%|▌ | 1740/30000 [1:06:14<17:58:22, 2.29s/it] 6%|▌ | 1740/30000 [1:06:14<17:58:22, 2.29s/it] 6%|▌ | 1741/30000 [1:06:16<18:10:10, 2.31s/it] 6%|▌ | 1742/30000 [1:06:18<18:06:44, 2.31s/it] 6%|▌ | 1743/30000 [1:06:20<17:59:36, 2.29s/it] 6%|▌ | 1744/30000 [1:06:23<18:09:39, 2.31s/it] 6%|▌ | 1745/30000 [1:06:25<18:09:41, 2.31s/it] 6%|▌ | 1746/30000 [1:06:27<18:01:14, 2.30s/it] 6%|▌ | 1747/30000 [1:06:30<18:01:18, 2.30s/it] 6%|▌ | 1748/30000 [1:06:32<17:57:17, 2.29s/it] 6%|▌ | 1749/30000 [1:06:34<17:55:37, 2.28s/it] 6%|▌ | 1750/30000 [1:06:37<17:53:16, 2.28s/it] 6%|▌ | 1750/30000 [1:06:37<17:53:16, 2.28s/it] 6%|▌ | 1751/30000 [1:06:39<18:01:58, 2.30s/it] 6%|▌ | 1752/30000 [1:06:41<17:53:54, 2.28s/it] 6%|▌ | 1753/30000 [1:06:43<17:50:56, 2.27s/it] 6%|▌ | 1754/30000 [1:06:46<18:00:23, 2.29s/it] 6%|▌ | 1755/30000 [1:06:48<17:55:38, 2.28s/it] 6%|▌ | 1756/30000 [1:06:50<17:53:37, 2.28s/it] 6%|▌ | 1757/30000 [1:06:53<18:03:25, 2.30s/it] 6%|▌ | 1758/30000 [1:06:55<17:55:59, 2.29s/it] 6%|▌ | 1759/30000 [1:06:57<17:52:18, 2.28s/it] 6%|▌ | 1760/30000 [1:06:59<17:57:27, 2.29s/it] 6%|▌ | 1760/30000 [1:06:59<17:57:27, 2.29s/it] 6%|▌ | 1761/30000 [1:07:02<18:09:07, 2.31s/it] 6%|▌ | 1762/30000 [1:07:04<18:08:06, 2.31s/it] 6%|▌ | 1763/30000 [1:07:06<17:59:34, 2.29s/it] 6%|▌ | 1764/30000 [1:07:09<18:04:47, 2.31s/it] 6%|▌ | 1765/30000 [1:07:11<17:58:59, 2.29s/it] 6%|▌ | 1766/30000 [1:07:13<17:54:01, 2.28s/it] 6%|▌ | 1767/30000 [1:07:15<17:54:16, 2.28s/it] 6%|▌ | 1768/30000 [1:07:18<18:05:17, 2.31s/it] 6%|▌ | 1769/30000 [1:07:20<18:02:04, 2.30s/it] 6%|▌ | 1770/30000 [1:07:22<17:53:56, 2.28s/it] 6%|▌ | 1770/30000 [1:07:22<17:53:56, 2.28s/it] 6%|▌ | 1771/30000 [1:07:25<18:02:48, 2.30s/it] 6%|▌ | 1772/30000 [1:07:27<18:02:37, 2.30s/it] 6%|▌ | 1773/30000 [1:07:29<17:59:18, 2.29s/it] 6%|▌ | 1774/30000 [1:07:32<17:57:33, 2.29s/it] 6%|▌ | 1775/30000 [1:07:34<18:04:55, 2.31s/it] 6%|▌ | 1776/30000 [1:07:36<18:03:51, 2.30s/it] 6%|▌ | 1777/30000 [1:07:38<17:56:45, 2.29s/it] 6%|▌ | 1778/30000 [1:07:41<18:01:03, 2.30s/it] 6%|▌ | 1779/30000 [1:07:43<17:57:00, 2.29s/it] 6%|▌ | 1780/30000 [1:07:45<17:54:27, 2.28s/it] 6%|▌ | 1780/30000 [1:07:45<17:54:27, 2.28s/it] 6%|▌ | 1781/30000 [1:07:48<17:55:10, 2.29s/it] 6%|▌ | 1782/30000 [1:07:50<18:04:15, 2.31s/it] 6%|▌ | 1783/30000 [1:07:52<18:03:30, 2.30s/it] 6%|▌ | 1784/30000 [1:07:55<18:00:26, 2.30s/it] 6%|▌ | 1785/30000 [1:07:57<17:55:23, 2.29s/it] 6%|▌ | 1786/30000 [1:07:59<18:02:42, 2.30s/it] 6%|▌ | 1787/30000 [1:08:01<17:59:55, 2.30s/it] 6%|▌ | 1788/30000 [1:08:04<17:52:33, 2.28s/it] 6%|▌ | 1789/30000 [1:08:06<18:00:00, 2.30s/it] 6%|▌ | 1790/30000 [1:08:08<18:01:34, 2.30s/it] 6%|▌ | 1790/30000 [1:08:08<18:01:34, 2.30s/it] 6%|▌ | 1791/30000 [1:08:11<17:59:25, 2.30s/it] 6%|▌ | 1792/30000 [1:08:13<17:56:13, 2.29s/it] 6%|▌ | 1793/30000 [1:08:15<18:01:44, 2.30s/it] 6%|▌ | 1794/30000 [1:08:17<17:54:34, 2.29s/it] 6%|▌ | 1795/30000 [1:08:20<17:51:22, 2.28s/it] 6%|▌ | 1796/30000 [1:08:22<17:59:41, 2.30s/it] 6%|▌ | 1797/30000 [1:08:24<17:59:06, 2.30s/it] 6%|▌ | 1798/30000 [1:08:27<17:57:04, 2.29s/it] 6%|▌ | 1799/30000 [1:08:29<17:51:18, 2.28s/it] 6%|▌ | 1800/30000 [1:08:31<18:01:38, 2.30s/it] 6%|▌ | 1800/30000 [1:08:31<18:01:38, 2.30s/it] 6%|▌ | 1801/30000 [1:08:34<17:54:40, 2.29s/it] 6%|▌ | 1802/30000 [1:08:36<17:46:20, 2.27s/it] 6%|▌ | 1803/30000 [1:08:38<17:53:03, 2.28s/it] 6%|▌ | 1804/30000 [1:08:40<17:50:24, 2.28s/it] 6%|▌ | 1805/30000 [1:08:43<17:48:50, 2.27s/it] 6%|▌ | 1806/30000 [1:08:45<17:49:24, 2.28s/it] 6%|▌ | 1807/30000 [1:08:47<18:01:40, 2.30s/it] 6%|▌ | 1808/30000 [1:08:50<18:06:06, 2.31s/it] 6%|▌ | 1809/30000 [1:08:52<18:01:44, 2.30s/it] 6%|▌ | 1810/30000 [1:08:54<18:01:17, 2.30s/it] 6%|▌ | 1810/30000 [1:08:54<18:01:17, 2.30s/it] 6%|▌ | 1811/30000 [1:08:56<18:01:43, 2.30s/it] 6%|▌ | 1812/30000 [1:08:59<17:54:43, 2.29s/it] 6%|▌ | 1813/30000 [1:09:01<17:51:58, 2.28s/it] 6%|▌ | 1814/30000 [1:09:03<17:55:31, 2.29s/it] 6%|▌ | 1815/30000 [1:09:06<18:00:28, 2.30s/it] 6%|▌ | 1816/30000 [1:09:08<17:55:42, 2.29s/it] 6%|▌ | 1817/30000 [1:09:10<17:52:58, 2.28s/it] 6%|▌ | 1818/30000 [1:09:12<17:59:19, 2.30s/it] 6%|▌ | 1819/30000 [1:09:15<17:52:34, 2.28s/it] 6%|▌ | 1820/30000 [1:09:17<17:54:23, 2.29s/it] 6%|▌ | 1820/30000 [1:09:17<17:54:23, 2.29s/it] 6%|▌ | 1821/30000 [1:09:19<17:57:57, 2.30s/it] 6%|▌ | 1822/30000 [1:09:22<18:06:32, 2.31s/it] 6%|▌ | 1823/30000 [1:09:24<18:09:01, 2.32s/it] 6%|▌ | 1824/30000 [1:09:26<18:00:08, 2.30s/it] 6%|▌ | 1825/30000 [1:09:28<17:48:21, 2.28s/it] 6%|▌ | 1826/30000 [1:09:31<17:55:54, 2.29s/it] 6%|▌ | 1827/30000 [1:09:33<17:50:53, 2.28s/it] 6%|▌ | 1828/30000 [1:09:35<17:52:02, 2.28s/it] 6%|▌ | 1829/30000 [1:09:38<17:52:09, 2.28s/it] 6%|▌ | 1830/30000 [1:09:40<17:52:03, 2.28s/it] 6%|▌ | 1830/30000 [1:09:40<17:52:03, 2.28s/it] 6%|▌ | 1831/30000 [1:09:42<17:49:53, 2.28s/it] 6%|▌ | 1832/30000 [1:09:44<17:49:01, 2.28s/it] 6%|▌ | 1833/30000 [1:09:47<17:54:04, 2.29s/it] 6%|▌ | 1834/30000 [1:09:49<17:44:54, 2.27s/it] 6%|▌ | 1835/30000 [1:09:51<17:42:06, 2.26s/it] 6%|▌ | 1836/30000 [1:09:54<17:52:40, 2.29s/it] 6%|▌ | 1837/30000 [1:09:56<17:53:01, 2.29s/it] 6%|▌ | 1838/30000 [1:09:58<17:51:05, 2.28s/it] 6%|▌ | 1839/30000 [1:10:00<17:53:03, 2.29s/it] 6%|▌ | 1840/30000 [1:10:03<17:52:16, 2.28s/it] 6%|▌ | 1840/30000 [1:10:03<17:52:16, 2.28s/it] 6%|▌ | 1841/30000 [1:10:05<17:44:28, 2.27s/it] 6%|▌ | 1842/30000 [1:10:07<17:42:08, 2.26s/it] 6%|▌ | 1843/30000 [1:10:10<17:51:17, 2.28s/it] 6%|▌ | 1844/30000 [1:10:12<17:47:58, 2.28s/it] 6%|▌ | 1845/30000 [1:10:14<17:51:14, 2.28s/it] 6%|▌ | 1846/30000 [1:10:16<18:01:10, 2.30s/it] 6%|▌ | 1847/30000 [1:10:19<17:59:37, 2.30s/it] 6%|▌ | 1848/30000 [1:10:21<17:54:47, 2.29s/it] 6%|▌ | 1849/30000 [1:10:23<17:53:27, 2.29s/it] 6%|▌ | 1850/30000 [1:10:26<17:55:30, 2.29s/it] 6%|▌ | 1850/30000 [1:10:26<17:55:30, 2.29s/it] 6%|▌ | 1851/30000 [1:10:28<17:52:42, 2.29s/it] 6%|▌ | 1852/30000 [1:10:30<17:48:50, 2.28s/it] 6%|▌ | 1853/30000 [1:10:32<17:55:22, 2.29s/it] 6%|▌ | 1854/30000 [1:10:35<17:50:02, 2.28s/it] 6%|▌ | 1855/30000 [1:10:37<17:46:31, 2.27s/it] 6%|▌ | 1856/30000 [1:10:39<17:55:47, 2.29s/it] 6%|▌ | 1857/30000 [1:10:42<18:01:59, 2.31s/it] 6%|▌ | 1858/30000 [1:10:44<17:56:45, 2.30s/it] 6%|▌ | 1859/30000 [1:10:46<18:01:53, 2.31s/it] 6%|▌ | 1860/30000 [1:10:49<17:56:54, 2.30s/it] 6%|▌ | 1860/30000 [1:10:49<17:56:54, 2.30s/it] 6%|▌ | 1861/30000 [1:10:51<17:52:59, 2.29s/it] 6%|▌ | 1862/30000 [1:10:53<17:51:57, 2.29s/it] 6%|▌ | 1863/30000 [1:10:55<17:51:28, 2.28s/it] 6%|▌ | 1864/30000 [1:10:58<17:48:31, 2.28s/it] 6%|▌ | 1865/30000 [1:11:00<17:47:01, 2.28s/it] 6%|▌ | 1866/30000 [1:11:02<17:57:57, 2.30s/it] 6%|▌ | 1867/30000 [1:11:04<17:49:37, 2.28s/it] 6%|▌ | 1868/30000 [1:11:07<17:46:44, 2.28s/it] 6%|▌ | 1869/30000 [1:11:09<17:56:05, 2.30s/it] 6%|▌ | 1870/30000 [1:11:11<17:50:02, 2.28s/it] 6%|▌ | 1870/30000 [1:11:11<17:50:02, 2.28s/it] 6%|▌ | 1871/30000 [1:11:14<17:43:53, 2.27s/it] 6%|▌ | 1872/30000 [1:11:16<17:57:41, 2.30s/it] 6%|▌ | 1873/30000 [1:11:18<18:04:35, 2.31s/it] 6%|▌ | 1874/30000 [1:11:21<18:02:56, 2.31s/it] 6%|▋ | 1875/30000 [1:11:23<18:08:37, 2.32s/it] 6%|▋ | 1876/30000 [1:11:25<17:59:40, 2.30s/it] 6%|▋ | 1877/30000 [1:11:27<17:49:35, 2.28s/it] 6%|▋ | 1878/30000 [1:11:30<17:53:17, 2.29s/it] 6%|▋ | 1879/30000 [1:11:32<17:51:27, 2.29s/it] 6%|▋ | 1880/30000 [1:11:34<17:49:16, 2.28s/it] 6%|▋ | 1880/30000 [1:11:34<17:49:16, 2.28s/it] 6%|▋ | 1881/30000 [1:11:37<17:50:15, 2.28s/it] 6%|▋ | 1882/30000 [1:11:39<18:00:19, 2.31s/it] 6%|▋ | 1883/30000 [1:11:41<17:57:11, 2.30s/it] 6%|▋ | 1884/30000 [1:11:43<17:55:10, 2.29s/it] 6%|▋ | 1885/30000 [1:11:46<18:03:21, 2.31s/it] 6%|▋ | 1886/30000 [1:11:48<18:04:52, 2.32s/it] 6%|▋ | 1887/30000 [1:11:50<17:57:23, 2.30s/it] 6%|▋ | 1888/30000 [1:11:53<18:00:00, 2.31s/it] 6%|▋ | 1889/30000 [1:11:55<17:57:04, 2.30s/it] 6%|▋ | 1890/30000 [1:11:57<17:54:47, 2.29s/it] 6%|▋ | 1890/30000 [1:11:57<17:54:47, 2.29s/it] 6%|▋ | 1891/30000 [1:12:00<17:56:59, 2.30s/it] 6%|▋ | 1892/30000 [1:12:02<18:02:08, 2.31s/it] 6%|▋ | 1893/30000 [1:12:04<17:53:36, 2.29s/it] 6%|▋ | 1894/30000 [1:12:07<17:54:52, 2.29s/it] 6%|▋ | 1895/30000 [1:12:09<17:54:33, 2.29s/it] 6%|▋ | 1896/30000 [1:12:11<17:46:15, 2.28s/it] 6%|▋ | 1897/30000 [1:12:13<17:55:19, 2.30s/it] 6%|▋ | 1898/30000 [1:12:16<18:00:12, 2.31s/it] 6%|▋ | 1899/30000 [1:12:18<18:01:02, 2.31s/it] 6%|▋ | 1900/30000 [1:12:20<17:55:10, 2.30s/it] 6%|▋ | 1900/30000 [1:12:20<17:55:10, 2.30s/it] 6%|▋ | 1901/30000 [1:12:23<17:57:33, 2.30s/it] 6%|▋ | 1902/30000 [1:12:25<17:57:44, 2.30s/it] 6%|▋ | 1903/30000 [1:12:27<17:56:39, 2.30s/it] 6%|▋ | 1904/30000 [1:12:30<18:00:10, 2.31s/it] 6%|▋ | 1905/30000 [1:12:32<17:53:18, 2.29s/it] 6%|▋ | 1906/30000 [1:12:34<17:51:18, 2.29s/it] 6%|▋ | 1907/30000 [1:12:36<18:07:52, 2.32s/it] 6%|▋ | 1908/30000 [1:12:39<18:02:50, 2.31s/it] 6%|▋ | 1909/30000 [1:12:41<17:56:56, 2.30s/it] 6%|▋ | 1910/30000 [1:12:43<18:02:48, 2.31s/it] 6%|▋ | 1910/30000 [1:12:43<18:02:48, 2.31s/it] 6%|▋ | 1911/30000 [1:12:46<18:02:43, 2.31s/it] 6%|▋ | 1912/30000 [1:12:48<17:56:47, 2.30s/it] 6%|▋ | 1913/30000 [1:12:50<18:02:21, 2.31s/it] 6%|▋ | 1914/30000 [1:12:53<17:56:45, 2.30s/it] 6%|▋ | 1915/30000 [1:12:55<17:52:53, 2.29s/it] 6%|▋ | 1916/30000 [1:12:57<17:51:34, 2.29s/it] 6%|▋ | 1917/30000 [1:12:59<17:57:12, 2.30s/it] 6%|▋ | 1918/30000 [1:13:02<17:46:55, 2.28s/it] 6%|▋ | 1919/30000 [1:13:04<17:46:56, 2.28s/it] 6%|▋ | 1920/30000 [1:13:06<17:53:51, 2.29s/it] 6%|▋ | 1920/30000 [1:13:06<17:53:51, 2.29s/it] 6%|▋ | 1921/30000 [1:13:09<17:59:01, 2.31s/it] 6%|▋ | 1922/30000 [1:13:11<17:52:46, 2.29s/it] 6%|▋ | 1923/30000 [1:13:13<17:57:32, 2.30s/it] 6%|▋ | 1924/30000 [1:13:15<17:53:14, 2.29s/it] 6%|▋ | 1925/30000 [1:13:18<17:47:45, 2.28s/it] 6%|▋ | 1926/30000 [1:13:20<17:48:35, 2.28s/it] 6%|▋ | 1927/30000 [1:13:22<17:57:35, 2.30s/it] 6%|▋ | 1928/30000 [1:13:25<17:49:40, 2.29s/it] 6%|▋ | 1929/30000 [1:13:27<17:47:44, 2.28s/it] 6%|▋ | 1930/30000 [1:13:29<17:56:05, 2.30s/it] 6%|▋ | 1930/30000 [1:13:29<17:56:05, 2.30s/it] 6%|▋ | 1931/30000 [1:13:32<17:59:33, 2.31s/it] 6%|▋ | 1932/30000 [1:13:34<17:53:05, 2.29s/it] 6%|▋ | 1933/30000 [1:13:36<17:52:31, 2.29s/it] 6%|▋ | 1934/30000 [1:13:38<17:55:02, 2.30s/it] 6%|▋ | 1935/30000 [1:13:41<17:50:09, 2.29s/it] 6%|▋ | 1936/30000 [1:13:43<17:44:02, 2.27s/it] 6%|▋ | 1937/30000 [1:13:45<17:55:50, 2.30s/it] 6%|▋ | 1938/30000 [1:13:48<17:55:59, 2.30s/it] 6%|▋ | 1939/30000 [1:13:50<17:45:15, 2.28s/it] 6%|▋ | 1940/30000 [1:13:52<17:46:28, 2.28s/it] 6%|▋ | 1940/30000 [1:13:52<17:46:28, 2.28s/it] 6%|▋ | 1941/30000 [1:13:54<17:46:48, 2.28s/it] 6%|▋ | 1942/30000 [1:13:57<17:44:00, 2.28s/it] 6%|▋ | 1943/30000 [1:13:59<17:39:27, 2.27s/it] 6%|▋ | 1944/30000 [1:14:01<17:46:12, 2.28s/it] 6%|▋ | 1945/30000 [1:14:03<17:42:33, 2.27s/it] 6%|▋ | 1946/30000 [1:14:06<17:39:31, 2.27s/it] 6%|▋ | 1947/30000 [1:14:08<17:40:27, 2.27s/it] 6%|▋ | 1948/30000 [1:14:10<17:39:23, 2.27s/it] 6%|▋ | 1949/30000 [1:14:13<17:43:21, 2.27s/it] 6%|▋ | 1950/30000 [1:14:15<17:41:04, 2.27s/it] 6%|▋ | 1950/30000 [1:14:15<17:41:04, 2.27s/it] 7%|▋ | 1951/30000 [1:14:17<17:53:52, 2.30s/it] 7%|▋ | 1952/30000 [1:14:19<17:56:19, 2.30s/it] 7%|▋ | 1953/30000 [1:14:22<17:50:04, 2.29s/it] 7%|▋ | 1954/30000 [1:14:24<17:52:33, 2.29s/it] 7%|▋ | 1955/30000 [1:14:26<17:59:58, 2.31s/it] 7%|▋ | 1956/30000 [1:14:29<17:54:30, 2.30s/it] 7%|▋ | 1957/30000 [1:14:31<17:50:30, 2.29s/it] 7%|▋ | 1958/30000 [1:14:33<17:49:41, 2.29s/it] 7%|▋ | 1959/30000 [1:14:36<17:49:16, 2.29s/it] 7%|▋ | 1960/30000 [1:14:38<17:46:38, 2.28s/it] 7%|▋ | 1960/30000 [1:14:38<17:46:38, 2.28s/it] 7%|▋ | 1961/30000 [1:14:40<17:45:19, 2.28s/it] 7%|▋ | 1962/30000 [1:14:42<17:53:18, 2.30s/it] 7%|▋ | 1963/30000 [1:14:45<18:01:17, 2.31s/it] 7%|▋ | 1964/30000 [1:14:47<17:54:08, 2.30s/it] 7%|▋ | 1965/30000 [1:14:49<17:50:36, 2.29s/it] 7%|▋ | 1966/30000 [1:14:52<17:55:21, 2.30s/it] 7%|▋ | 1967/30000 [1:14:54<17:55:39, 2.30s/it] 7%|▋ | 1968/30000 [1:14:56<17:49:17, 2.29s/it] 7%|▋ | 1969/30000 [1:14:58<17:47:54, 2.29s/it] 7%|▋ | 1970/30000 [1:15:01<17:48:14, 2.29s/it] 7%|▋ | 1970/30000 [1:15:01<17:48:14, 2.29s/it] 7%|▋ | 1971/30000 [1:15:03<17:52:58, 2.30s/it] 7%|▋ | 1972/30000 [1:15:05<17:48:14, 2.29s/it] 7%|▋ | 1973/30000 [1:15:08<17:39:15, 2.27s/it] 7%|▋ | 1974/30000 [1:15:10<17:44:52, 2.28s/it] 7%|▋ | 1975/30000 [1:15:12<17:54:12, 2.30s/it] 7%|▋ | 1976/30000 [1:15:14<17:47:24, 2.29s/it] 7%|▋ | 1977/30000 [1:15:17<17:53:19, 2.30s/it] 7%|▋ | 1978/30000 [1:15:19<17:55:51, 2.30s/it] 7%|▋ | 1979/30000 [1:15:21<17:51:00, 2.29s/it] 7%|▋ | 1980/30000 [1:15:24<17:49:00, 2.29s/it] 7%|▋ | 1980/30000 [1:15:24<17:49:00, 2.29s/it] 7%|▋ | 1981/30000 [1:15:26<17:51:21, 2.29s/it] 7%|▋ | 1982/30000 [1:15:28<17:52:17, 2.30s/it] 7%|▋ | 1983/30000 [1:15:31<17:47:12, 2.29s/it] 7%|▋ | 1984/30000 [1:15:33<17:45:13, 2.28s/it] 7%|▋ | 1985/30000 [1:15:35<17:54:12, 2.30s/it] 7%|▋ | 1986/30000 [1:15:37<18:00:47, 2.31s/it] 7%|▋ | 1987/30000 [1:15:40<17:55:06, 2.30s/it] 7%|▋ | 1988/30000 [1:15:42<17:53:39, 2.30s/it] 7%|▋ | 1989/30000 [1:15:44<17:59:59, 2.31s/it] 7%|▋ | 1990/30000 [1:15:47<18:02:22, 2.32s/it] 7%|▋ | 1990/30000 [1:15:47<18:02:22, 2.32s/it] 7%|▋ | 1991/30000 [1:15:49<17:56:35, 2.31s/it] 7%|▋ | 1992/30000 [1:15:51<17:49:04, 2.29s/it] 7%|▋ | 1993/30000 [1:15:54<17:46:35, 2.28s/it] 7%|▋ | 1994/30000 [1:15:56<17:37:40, 2.27s/it] 7%|▋ | 1995/30000 [1:15:58<17:33:57, 2.26s/it] 7%|▋ | 1996/30000 [1:16:00<17:44:52, 2.28s/it] 7%|▋ | 1997/30000 [1:16:03<17:48:31, 2.29s/it] 7%|▋ | 1998/30000 [1:16:05<17:45:14, 2.28s/it] 7%|▋ | 1999/30000 [1:16:07<17:45:51, 2.28s/it] 7%|▋ | 2000/30000 [1:16:09<17:49:28, 2.29s/it] 7%|▋ | 2000/30000 [1:16:10<17:49:28, 2.29s/it] 7%|▋ | 2001/30000 [1:16:12<17:44:44, 2.28s/it] 7%|▋ | 2002/30000 [1:16:14<17:41:58, 2.28s/it] 7%|▋ | 2003/30000 [1:16:16<17:45:35, 2.28s/it] 7%|▋ | 2004/30000 [1:16:19<17:46:42, 2.29s/it] 7%|▋ | 2005/30000 [1:16:21<17:42:53, 2.28s/it] 7%|▋ | 2006/30000 [1:16:23<17:40:57, 2.27s/it] 7%|▋ | 2007/30000 [1:16:25<17:47:42, 2.29s/it] 7%|▋ | 2008/30000 [1:16:28<17:40:53, 2.27s/it] 7%|▋ | 2009/30000 [1:16:30<17:38:00, 2.27s/it] 7%|▋ | 2010/30000 [1:16:32<17:48:38, 2.29s/it] 7%|▋ | 2010/30000 [1:16:32<17:48:38, 2.29s/it] 7%|▋ | 2011/30000 [1:16:35<17:52:58, 2.30s/it] 7%|▋ | 2012/30000 [1:16:37<17:51:00, 2.30s/it] 7%|▋ | 2013/30000 [1:16:39<17:56:11, 2.31s/it] 7%|▋ | 2014/30000 [1:16:41<17:45:10, 2.28s/it] 7%|▋ | 2015/30000 [1:16:44<17:42:59, 2.28s/it] 7%|▋ | 2016/30000 [1:16:46<17:44:56, 2.28s/it] 7%|▋ | 2017/30000 [1:16:48<17:39:58, 2.27s/it] 7%|▋ | 2018/30000 [1:16:51<17:38:14, 2.27s/it] 7%|▋ | 2019/30000 [1:16:53<17:40:33, 2.27s/it] 7%|▋ | 2020/30000 [1:16:55<17:45:02, 2.28s/it] 7%|▋ | 2020/30000 [1:16:55<17:45:02, 2.28s/it] 7%|▋ | 2021/30000 [1:16:57<17:40:31, 2.27s/it] 7%|▋ | 2022/30000 [1:17:00<17:43:23, 2.28s/it] 7%|▋ | 2023/30000 [1:17:02<17:51:33, 2.30s/it] 7%|▋ | 2024/30000 [1:17:04<17:48:27, 2.29s/it] 7%|▋ | 2025/30000 [1:17:07<17:47:09, 2.29s/it] 7%|▋ | 2026/30000 [1:17:09<17:55:41, 2.31s/it] 7%|▋ | 2027/30000 [1:17:11<17:58:26, 2.31s/it] 7%|▋ | 2028/30000 [1:17:13<17:51:16, 2.30s/it] 7%|▋ | 2029/30000 [1:17:16<17:57:10, 2.31s/it] 7%|▋ | 2030/30000 [1:17:18<18:03:54, 2.33s/it] 7%|▋ | 2030/30000 [1:17:18<18:03:54, 2.33s/it] 7%|▋ | 2031/30000 [1:17:20<17:56:01, 2.31s/it] 7%|▋ | 2032/30000 [1:17:23<17:50:31, 2.30s/it] 7%|▋ | 2033/30000 [1:17:25<17:58:32, 2.31s/it] 7%|▋ | 2034/30000 [1:17:27<17:47:44, 2.29s/it] 7%|▋ | 2035/30000 [1:17:30<17:42:28, 2.28s/it] 7%|▋ | 2036/30000 [1:17:32<17:51:32, 2.30s/it] 7%|▋ | 2037/30000 [1:17:34<17:44:45, 2.28s/it] 7%|▋ | 2038/30000 [1:17:36<17:43:29, 2.28s/it] 7%|▋ | 2039/30000 [1:17:39<17:57:11, 2.31s/it] 7%|▋ | 2040/30000 [1:17:41<17:46:03, 2.29s/it] 7%|▋ | 2040/30000 [1:17:41<17:46:03, 2.29s/it] 7%|▋ | 2041/30000 [1:17:43<17:46:03, 2.29s/it] 7%|▋ | 2042/30000 [1:17:46<17:58:57, 2.32s/it] 7%|▋ | 2043/30000 [1:17:48<17:59:35, 2.32s/it] 7%|▋ | 2044/30000 [1:17:50<17:51:08, 2.30s/it] 7%|▋ | 2045/30000 [1:17:53<17:55:38, 2.31s/it] 7%|▋ | 2046/30000 [1:17:55<17:50:02, 2.30s/it] 7%|▋ | 2047/30000 [1:17:57<17:44:31, 2.28s/it] 7%|▋ | 2048/30000 [1:17:59<17:42:45, 2.28s/it] 7%|▋ | 2049/30000 [1:18:02<17:45:56, 2.29s/it] 7%|▋ | 2050/30000 [1:18:04<17:42:39, 2.28s/it] 7%|▋ | 2050/30000 [1:18:04<17:42:39, 2.28s/it] 7%|▋ | 2051/30000 [1:18:06<17:43:13, 2.28s/it] 7%|▋ | 2052/30000 [1:18:09<17:58:21, 2.32s/it] 7%|▋ | 2053/30000 [1:18:11<17:58:17, 2.31s/it] 7%|▋ | 2054/30000 [1:18:13<17:51:25, 2.30s/it] 7%|▋ | 2055/30000 [1:18:16<17:55:12, 2.31s/it] 7%|▋ | 2056/30000 [1:18:18<17:51:45, 2.30s/it] 7%|▋ | 2057/30000 [1:18:20<17:47:33, 2.29s/it] 7%|▋ | 2058/30000 [1:18:22<17:45:11, 2.29s/it] 7%|▋ | 2059/30000 [1:18:25<17:46:51, 2.29s/it] 7%|▋ | 2060/30000 [1:18:27<17:43:54, 2.28s/it] 7%|▋ | 2060/30000 [1:18:27<17:43:54, 2.28s/it] 7%|▋ | 2061/30000 [1:18:29<17:45:59, 2.29s/it] 7%|▋ | 2062/30000 [1:18:32<17:59:35, 2.32s/it] 7%|▋ | 2063/30000 [1:18:33<15:29:12, 2.00s/it] 7%|▋ | 2064/30000 [1:18:34<13:36:24, 1.75s/it] 7%|▋ | 2065/30000 [1:18:35<12:15:19, 1.58s/it] 7%|▋ | 2066/30000 [1:18:37<11:24:38, 1.47s/it] 7%|▋ | 2067/30000 [1:18:38<10:44:49, 1.39s/it] 7%|▋ | 2068/30000 [1:18:39<10:33:53, 1.36s/it] 7%|▋ | 2069/30000 [1:18:41<12:51:56, 1.66s/it] 7%|▋ | 2070/30000 [1:18:44<14:21:41, 1.85s/it] 7%|▋ | 2070/30000 [1:18:44<14:21:41, 1.85s/it] 7%|▋ | 2071/30000 [1:18:46<15:19:35, 1.98s/it] 7%|▋ | 2072/30000 [1:18:48<16:04:21, 2.07s/it] 7%|▋ | 2073/30000 [1:18:51<16:44:25, 2.16s/it] 7%|▋ | 2074/30000 [1:18:53<17:07:52, 2.21s/it] 7%|▋ | 2075/30000 [1:18:55<17:18:40, 2.23s/it] 7%|▋ | 2076/30000 [1:18:57<17:26:05, 2.25s/it] 7%|▋ | 2077/30000 [1:19:00<17:35:13, 2.27s/it] 7%|▋ | 2078/30000 [1:19:02<17:40:39, 2.28s/it] 7%|▋ | 2079/30000 [1:19:04<17:41:45, 2.28s/it] 7%|▋ | 2080/30000 [1:19:07<17:38:41, 2.28s/it] 7%|▋ | 2080/30000 [1:19:07<17:38:41, 2.28s/it] 7%|▋ | 2081/30000 [1:19:09<17:48:58, 2.30s/it] 7%|▋ | 2082/30000 [1:19:11<17:47:17, 2.29s/it] 7%|▋ | 2083/30000 [1:19:14<17:42:16, 2.28s/it] 7%|▋ | 2084/30000 [1:19:16<17:36:53, 2.27s/it] 7%|▋ | 2085/30000 [1:19:18<17:47:03, 2.29s/it] 7%|▋ | 2086/30000 [1:19:20<17:42:43, 2.28s/it] 7%|▋ | 2087/30000 [1:19:23<17:38:05, 2.27s/it] 7%|▋ | 2088/30000 [1:19:25<17:40:28, 2.28s/it] 7%|▋ | 2089/30000 [1:19:27<17:45:22, 2.29s/it] 7%|▋ | 2090/30000 [1:19:29<17:40:28, 2.28s/it] 7%|▋ | 2090/30000 [1:19:30<17:40:28, 2.28s/it] 7%|▋ | 2091/30000 [1:19:32<17:40:01, 2.28s/it] 7%|▋ | 2092/30000 [1:19:34<17:44:26, 2.29s/it] 7%|▋ | 2093/30000 [1:19:36<17:42:50, 2.29s/it] 7%|▋ | 2094/30000 [1:19:39<17:37:54, 2.27s/it] 7%|▋ | 2095/30000 [1:19:41<17:42:46, 2.29s/it] 7%|▋ | 2096/30000 [1:19:43<17:40:21, 2.28s/it] 7%|▋ | 2097/30000 [1:19:45<17:40:32, 2.28s/it] 7%|▋ | 2098/30000 [1:19:48<17:40:01, 2.28s/it] 7%|▋ | 2099/30000 [1:19:50<17:47:13, 2.30s/it] 7%|▋ | 2100/30000 [1:19:52<17:50:44, 2.30s/it] 7%|▋ | 2100/30000 [1:19:52<17:50:44, 2.30s/it] 7%|▋ | 2101/30000 [1:19:55<17:45:14, 2.29s/it] 7%|▋ | 2102/30000 [1:19:57<17:44:51, 2.29s/it] 7%|▋ | 2103/30000 [1:19:59<17:44:04, 2.29s/it] 7%|▋ | 2104/30000 [1:20:01<17:39:20, 2.28s/it] 7%|▋ | 2105/30000 [1:20:04<17:38:09, 2.28s/it] 7%|▋ | 2106/30000 [1:20:06<17:48:41, 2.30s/it] 7%|▋ | 2107/30000 [1:20:08<17:50:15, 2.30s/it] 7%|▋ | 2108/30000 [1:20:11<17:44:27, 2.29s/it] 7%|▋ | 2109/30000 [1:20:13<17:50:34, 2.30s/it] 7%|▋ | 2110/30000 [1:20:15<17:49:43, 2.30s/it] 7%|▋ | 2110/30000 [1:20:15<17:49:43, 2.30s/it] 7%|▋ | 2111/30000 [1:20:18<17:46:48, 2.30s/it] 7%|▋ | 2112/30000 [1:20:20<17:41:17, 2.28s/it] 7%|▋ | 2113/30000 [1:20:22<17:45:58, 2.29s/it] 7%|▋ | 2114/30000 [1:20:24<17:40:38, 2.28s/it] 7%|▋ | 2115/30000 [1:20:27<17:37:15, 2.27s/it] 7%|▋ | 2116/30000 [1:20:29<17:48:40, 2.30s/it] 7%|▋ | 2117/30000 [1:20:31<17:46:26, 2.29s/it] 7%|▋ | 2118/30000 [1:20:34<17:44:21, 2.29s/it] 7%|▋ | 2119/30000 [1:20:36<17:52:25, 2.31s/it] 7%|▋ | 2120/30000 [1:20:38<17:53:15, 2.31s/it] 7%|▋ | 2120/30000 [1:20:38<17:53:15, 2.31s/it] 7%|▋ | 2121/30000 [1:20:41<17:47:38, 2.30s/it] 7%|▋ | 2122/30000 [1:20:43<17:44:10, 2.29s/it] 7%|▋ | 2123/30000 [1:20:45<17:47:40, 2.30s/it] 7%|▋ | 2124/30000 [1:20:47<17:47:00, 2.30s/it] 7%|▋ | 2125/30000 [1:20:50<17:47:08, 2.30s/it] 7%|▋ | 2126/30000 [1:20:52<17:54:46, 2.31s/it] 7%|▋ | 2127/30000 [1:20:54<17:58:31, 2.32s/it] 7%|▋ | 2128/30000 [1:20:57<17:48:34, 2.30s/it] 7%|▋ | 2129/30000 [1:20:59<17:48:03, 2.30s/it] 7%|▋ | 2130/30000 [1:21:01<17:45:20, 2.29s/it] +{'loss': 0.0209, 'grad_norm': 0.13143905997276306, 'learning_rate': 2.9995221196291952e-05, 'flow_matching_loss': 0.02088084127753973, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.5} +{'loss': 0.019, 'grad_norm': 0.12355560809373856, 'learning_rate': 2.999479474546582e-05, 'flow_matching_loss': 0.019048825930804016, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.51} +{'loss': 0.0191, 'grad_norm': 0.10883723944425583, 'learning_rate': 2.9994350074550153e-05, 'flow_matching_loss': 0.019093702919781207, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.51} +{'loss': 0.019, 'grad_norm': 0.12529560923576355, 'learning_rate': 2.9993887184085273e-05, 'flow_matching_loss': 0.01897429833188653, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.51} +{'loss': 0.0204, 'grad_norm': 0.11655668914318085, 'learning_rate': 2.9993406074633628e-05, 'flow_matching_loss': 0.02036708313971758, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.51} +{'loss': 0.0196, 'grad_norm': 0.12911736965179443, 'learning_rate': 2.9992906746779824e-05, 'flow_matching_loss': 0.019621549919247628, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.52} +{'loss': 0.0199, 'grad_norm': 0.10910261422395706, 'learning_rate': 2.9992389201130576e-05, 'flow_matching_loss': 0.01990507422015071, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.52} +{'loss': 0.0208, 'grad_norm': 0.07461588084697723, 'learning_rate': 2.999185343831476e-05, 'flow_matching_loss': 0.02075831890106201, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.52} +{'loss': 0.0201, 'grad_norm': 0.11058427393436432, 'learning_rate': 2.999129945898338e-05, 'flow_matching_loss': 0.02014348916709423, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.53} +{'loss': 0.0188, 'grad_norm': 0.09275602549314499, 'learning_rate': 2.999072726380957e-05, 'flow_matching_loss': 0.01883417833596468, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.53} +{'loss': 0.0203, 'grad_norm': 0.10750189423561096, 'learning_rate': 2.9990136853488603e-05, 'flow_matching_loss': 0.0203109884634614, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.53} +{'loss': 0.0185, 'grad_norm': 0.09932545572519302, 'learning_rate': 2.9989528228737878e-05, 'flow_matching_loss': 0.01849959949031472, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.54} +{'loss': 0.0189, 'grad_norm': 0.10849635303020477, 'learning_rate': 2.9988901390296935e-05, 'flow_matching_loss': 0.018869185727089642, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.54} +{'loss': 0.0202, 'grad_norm': 0.07934395223855972, 'learning_rate': 2.9988256338927442e-05, 'flow_matching_loss': 0.020195501390844584, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.54} +{'loss': 0.0189, 'grad_norm': 0.08885929733514786, 'learning_rate': 2.9987593075413198e-05, 'flow_matching_loss': 0.018912001233547925, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.54} +{'loss': 0.0195, 'grad_norm': 0.10698315501213074, 'learning_rate': 2.9986911600560128e-05, 'flow_matching_loss': 0.019494645390659572, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.55} +{'loss': 0.0185, 'grad_norm': 0.09530776739120483, 'learning_rate': 2.998621191519629e-05, 'flow_matching_loss': 0.01851522019132972, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.55} +{'loss': 0.0202, 'grad_norm': 0.10968459397554398, 'learning_rate': 2.9985494020171873e-05, 'flow_matching_loss': 0.020170450396835803, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.55} +{'loss': 0.0184, 'grad_norm': 0.09446757286787033, 'learning_rate': 2.9984757916359176e-05, 'flow_matching_loss': 0.018374338280409575, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.56} +{'loss': 0.0184, 'grad_norm': 0.11678633093833923, 'learning_rate': 2.998400360465264e-05, 'flow_matching_loss': 0.018384858034551144, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.56} +{'loss': 0.0187, 'grad_norm': 0.14418740570545197, 'learning_rate': 2.998323108596883e-05, 'flow_matching_loss': 0.018652132991701365, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.56} +{'loss': 0.0192, 'grad_norm': 0.10495369136333466, 'learning_rate': 2.9982440361246422e-05, 'flow_matching_loss': 0.019151432253420354, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.56} +{'loss': 0.0179, 'grad_norm': 0.10260157287120819, 'learning_rate': 2.998163143144622e-05, 'flow_matching_loss': 0.01788393370807171, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.57} +{'loss': 0.0224, 'grad_norm': 0.12304387986660004, 'learning_rate': 2.9980804297551155e-05, 'flow_matching_loss': 0.022401940356940032, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.57} +{'loss': 0.0181, 'grad_norm': 0.08812795579433441, 'learning_rate': 2.997995896056627e-05, 'flow_matching_loss': 0.018053389620035887, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.57} +{'loss': 0.0194, 'grad_norm': 0.124569371342659, 'learning_rate': 2.997909542151873e-05, 'flow_matching_loss': 0.01937941098585725, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.58} +{'loss': 0.0224, 'grad_norm': 0.16409650444984436, 'learning_rate': 2.9978213681457823e-05, 'flow_matching_loss': 0.022355991415679455, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.58} +{'loss': 0.0211, 'grad_norm': 0.13631121814250946, 'learning_rate': 2.9977313741454933e-05, 'flow_matching_loss': 0.02111262958496809, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.58} +{'loss': 0.0193, 'grad_norm': 0.12480654567480087, 'learning_rate': 2.997639560260358e-05, 'flow_matching_loss': 0.01932366006076336, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.58} +{'loss': 0.0183, 'grad_norm': 0.10486571490764618, 'learning_rate': 2.9975459266019387e-05, 'flow_matching_loss': 0.01828473322093487, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.59} +{'loss': 0.0176, 'grad_norm': 0.10991387814283371, 'learning_rate': 2.9974504732840092e-05, 'flow_matching_loss': 0.017612296901643277, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.59} +{'loss': 0.0188, 'grad_norm': 0.095576211810112, 'learning_rate': 2.997353200422555e-05, 'flow_matching_loss': 0.018814872298389674, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.59} +{'loss': 0.0192, 'grad_norm': 0.09007420390844345, 'learning_rate': 2.99725410813577e-05, 'flow_matching_loss': 0.019239315204322337, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.6} +{'loss': 0.0187, 'grad_norm': 0.12215665727853775, 'learning_rate': 2.997153196544063e-05, 'flow_matching_loss': 0.018721137940883637, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.6} +{'loss': 0.019, 'grad_norm': 0.08025005459785461, 'learning_rate': 2.997050465770049e-05, 'flow_matching_loss': 0.019014161452651022, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.6} +{'loss': 0.02, 'grad_norm': 0.12162505090236664, 'learning_rate': 2.9969459159385562e-05, 'flow_matching_loss': 0.020035501383244992, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.61} +{'loss': 0.0193, 'grad_norm': 0.09975284337997437, 'learning_rate': 2.9968395471766237e-05, 'flow_matching_loss': 0.019283664971590044, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.61} +{'loss': 0.0179, 'grad_norm': 0.10556112229824066, 'learning_rate': 2.996731359613498e-05, 'flow_matching_loss': 0.017876154370605947, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.61} +{'loss': 0.0181, 'grad_norm': 0.11613994091749191, 'learning_rate': 2.996621353380638e-05, 'flow_matching_loss': 0.018078568018972875, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.61} +{'loss': 0.0192, 'grad_norm': 0.14407341182231903, 'learning_rate': 2.9965095286117113e-05, 'flow_matching_loss': 0.019239773973822594, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.62} +{'loss': 0.018, 'grad_norm': 0.11327681690454483, 'learning_rate': 2.996395885442596e-05, 'flow_matching_loss': 0.018044556118547918, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.62} 7%|▋ | 2130/30000 [1:21:01<17:45:20, 2.29s/it] 7%|▋ | 2131/30000 [1:21:03<17:38:48, 2.28s/it] 7%|▋ | 2132/30000 [1:21:06<17:36:36, 2.27s/it] 7%|▋ | 2133/30000 [1:21:08<17:44:07, 2.29s/it] 7%|▋ | 2134/30000 [1:21:10<17:38:13, 2.28s/it] 7%|▋ | 2135/30000 [1:21:13<17:39:05, 2.28s/it] 7%|▋ | 2136/30000 [1:21:15<17:47:37, 2.30s/it] 7%|▋ | 2137/30000 [1:21:17<17:50:34, 2.31s/it] 7%|▋ | 2138/30000 [1:21:20<17:44:41, 2.29s/it] 7%|▋ | 2139/30000 [1:21:22<17:48:40, 2.30s/it] 7%|▋ | 2140/30000 [1:21:24<17:48:29, 2.30s/it] 7%|▋ | 2140/30000 [1:21:24<17:48:29, 2.30s/it] 7%|▋ | 2141/30000 [1:21:26<17:44:48, 2.29s/it] 7%|▋ | 2142/30000 [1:21:29<17:45:24, 2.29s/it] 7%|▋ | 2143/30000 [1:21:31<17:42:47, 2.29s/it] 7%|▋ | 2144/30000 [1:21:33<17:39:49, 2.28s/it] 7%|▋ | 2145/30000 [1:21:36<17:39:30, 2.28s/it] 7%|▋ | 2146/30000 [1:21:38<17:39:25, 2.28s/it] 7%|▋ | 2147/30000 [1:21:40<17:36:56, 2.28s/it] 7%|▋ | 2148/30000 [1:21:42<17:35:24, 2.27s/it] 7%|▋ | 2149/30000 [1:21:45<17:44:48, 2.29s/it] 7%|▋ | 2150/30000 [1:21:47<17:47:15, 2.30s/it] 7%|▋ | 2150/30000 [1:21:47<17:47:15, 2.30s/it] 7%|▋ | 2151/30000 [1:21:49<17:45:18, 2.30s/it] 7%|▋ | 2152/30000 [1:21:52<17:52:06, 2.31s/it] 7%|▋ | 2153/30000 [1:21:54<17:49:06, 2.30s/it] 7%|▋ | 2154/30000 [1:21:56<17:43:08, 2.29s/it] 7%|▋ | 2155/30000 [1:21:58<17:39:56, 2.28s/it] 7%|▋ | 2156/30000 [1:22:01<17:44:08, 2.29s/it] 7%|▋ | 2157/30000 [1:22:03<17:38:15, 2.28s/it] 7%|▋ | 2158/30000 [1:22:05<17:37:13, 2.28s/it] 7%|▋ | 2159/30000 [1:22:08<17:46:59, 2.30s/it] 7%|▋ | 2160/30000 [1:22:10<17:52:09, 2.31s/it] 7%|▋ | 2160/30000 [1:22:10<17:52:09, 2.31s/it] 7%|▋ | 2161/30000 [1:22:12<17:47:04, 2.30s/it] 7%|▋ | 2162/30000 [1:22:15<17:51:13, 2.31s/it] 7%|▋ | 2163/30000 [1:22:17<17:47:58, 2.30s/it] 7%|▋ | 2164/30000 [1:22:19<17:42:46, 2.29s/it] 7%|▋ | 2165/30000 [1:22:21<17:44:18, 2.29s/it] 7%|▋ | 2166/30000 [1:22:24<17:41:03, 2.29s/it] 7%|▋ | 2167/30000 [1:22:26<17:42:02, 2.29s/it] 7%|▋ | 2168/30000 [1:22:28<17:39:00, 2.28s/it] 7%|▋ | 2169/30000 [1:22:31<17:42:27, 2.29s/it] 7%|▋ | 2170/30000 [1:22:33<17:35:37, 2.28s/it] 7%|▋ | 2170/30000 [1:22:33<17:35:37, 2.28s/it] 7%|▋ | 2171/30000 [1:22:35<17:41:04, 2.29s/it] 7%|▋ | 2172/30000 [1:22:37<17:46:17, 2.30s/it] 7%|▋ | 2173/30000 [1:22:40<17:47:29, 2.30s/it] 7%|▋ | 2174/30000 [1:22:42<17:41:39, 2.29s/it] 7%|▋ | 2175/30000 [1:22:44<17:35:25, 2.28s/it] 7%|▋ | 2176/30000 [1:22:47<17:44:20, 2.30s/it] 7%|▋ | 2177/30000 [1:22:49<17:49:03, 2.31s/it] 7%|▋ | 2178/30000 [1:22:51<17:43:55, 2.29s/it] 7%|▋ | 2179/30000 [1:22:53<17:37:30, 2.28s/it] 7%|▋ | 2180/30000 [1:22:56<17:44:23, 2.30s/it] 7%|▋ | 2180/30000 [1:22:56<17:44:23, 2.30s/it] 7%|▋ | 2181/30000 [1:22:58<17:39:27, 2.29s/it] 7%|▋ | 2182/30000 [1:23:00<17:37:44, 2.28s/it] 7%|▋ | 2183/30000 [1:23:03<17:39:29, 2.29s/it] 7%|▋ | 2184/30000 [1:23:05<17:43:28, 2.29s/it] 7%|▋ | 2185/30000 [1:23:07<17:40:15, 2.29s/it] 7%|▋ | 2186/30000 [1:23:09<17:37:32, 2.28s/it] 7%|▋ | 2187/30000 [1:23:12<17:44:31, 2.30s/it] 7%|▋ | 2188/30000 [1:23:14<17:44:46, 2.30s/it] 7%|▋ | 2189/30000 [1:23:16<17:39:44, 2.29s/it] 7%|▋ | 2190/30000 [1:23:19<17:35:01, 2.28s/it] 7%|▋ | 2190/30000 [1:23:19<17:35:01, 2.28s/it] 7%|▋ | 2191/30000 [1:23:21<17:45:43, 2.30s/it] 7%|▋ | 2192/30000 [1:23:23<17:46:50, 2.30s/it] 7%|▋ | 2193/30000 [1:23:26<17:42:46, 2.29s/it] 7%|▋ | 2194/30000 [1:23:28<17:40:23, 2.29s/it] 7%|▋ | 2195/30000 [1:23:30<17:47:12, 2.30s/it] 7%|▋ | 2196/30000 [1:23:32<17:44:52, 2.30s/it] 7%|▋ | 2197/30000 [1:23:35<17:38:37, 2.28s/it] 7%|▋ | 2198/30000 [1:23:37<17:36:01, 2.28s/it] 7%|▋ | 2199/30000 [1:23:39<17:38:41, 2.28s/it] 7%|▋ | 2200/30000 [1:23:42<17:32:53, 2.27s/it] 7%|▋ | 2200/30000 [1:23:42<17:32:53, 2.27s/it] 7%|▋ | 2201/30000 [1:23:44<17:31:18, 2.27s/it] 7%|▋ | 2202/30000 [1:23:46<17:35:29, 2.28s/it] 7%|▋ | 2203/30000 [1:23:48<17:33:05, 2.27s/it] 7%|▋ | 2204/30000 [1:23:51<17:31:04, 2.27s/it] 7%|▋ | 2205/30000 [1:23:53<17:31:17, 2.27s/it] 7%|▋ | 2206/30000 [1:23:55<17:37:51, 2.28s/it] 7%|▋ | 2207/30000 [1:23:57<17:32:07, 2.27s/it] 7%|▋ | 2208/30000 [1:24:00<17:32:38, 2.27s/it] 7%|▋ | 2209/30000 [1:24:02<17:43:24, 2.30s/it] 7%|▋ | 2210/30000 [1:24:04<17:42:38, 2.29s/it] 7%|▋ | 2210/30000 [1:24:04<17:42:38, 2.29s/it] 7%|▋ | 2211/30000 [1:24:07<17:40:58, 2.29s/it] 7%|▋ | 2212/30000 [1:24:09<17:38:16, 2.29s/it] 7%|▋ | 2213/30000 [1:24:11<17:46:27, 2.30s/it] 7%|▋ | 2214/30000 [1:24:14<17:48:28, 2.31s/it] 7%|▋ | 2215/30000 [1:24:16<17:44:17, 2.30s/it] 7%|▋ | 2216/30000 [1:24:18<17:39:55, 2.29s/it] 7%|▋ | 2217/30000 [1:24:20<17:45:16, 2.30s/it] 7%|▋ | 2218/30000 [1:24:23<17:39:32, 2.29s/it] 7%|▋ | 2219/30000 [1:24:25<17:38:16, 2.29s/it] 7%|▋ | 2220/30000 [1:24:27<17:46:31, 2.30s/it] 7%|▋ | 2220/30000 [1:24:27<17:46:31, 2.30s/it] 7%|▋ | 2221/30000 [1:24:30<17:46:15, 2.30s/it] 7%|▋ | 2222/30000 [1:24:32<17:41:06, 2.29s/it] 7%|▋ | 2223/30000 [1:24:34<17:40:52, 2.29s/it] 7%|▋ | 2224/30000 [1:24:37<17:47:50, 2.31s/it] 7%|▋ | 2225/30000 [1:24:39<17:44:33, 2.30s/it] 7%|▋ | 2226/30000 [1:24:41<17:36:33, 2.28s/it] 7%|▋ | 2227/30000 [1:24:43<17:43:33, 2.30s/it] 7%|▋ | 2228/30000 [1:24:46<17:42:44, 2.30s/it] 7%|▋ | 2229/30000 [1:24:48<17:40:01, 2.29s/it] 7%|▋ | 2230/30000 [1:24:50<17:41:51, 2.29s/it] 7%|▋ | 2230/30000 [1:24:50<17:41:51, 2.29s/it] 7%|▋ | 2231/30000 [1:24:53<17:48:44, 2.31s/it] 7%|▋ | 2232/30000 [1:24:55<17:45:15, 2.30s/it] 7%|▋ | 2233/30000 [1:24:57<17:44:36, 2.30s/it] 7%|▋ | 2234/30000 [1:25:00<17:50:06, 2.31s/it] 7%|▋ | 2235/30000 [1:25:02<17:49:04, 2.31s/it] 7%|▋ | 2236/30000 [1:25:04<17:41:20, 2.29s/it] 7%|▋ | 2237/30000 [1:25:06<17:45:31, 2.30s/it] 7%|▋ | 2238/30000 [1:25:09<17:45:27, 2.30s/it] 7%|▋ | 2239/30000 [1:25:11<17:36:32, 2.28s/it] 7%|▋ | 2240/30000 [1:25:13<17:33:49, 2.28s/it] 7%|▋ | 2240/30000 [1:25:13<17:33:49, 2.28s/it] 7%|▋ | 2241/30000 [1:25:16<17:42:55, 2.30s/it] 7%|▋ | 2242/30000 [1:25:18<17:43:05, 2.30s/it] 7%|▋ | 2243/30000 [1:25:20<17:38:28, 2.29s/it] 7%|▋ | 2244/30000 [1:25:22<17:44:38, 2.30s/it] 7%|▋ | 2245/30000 [1:25:25<17:44:49, 2.30s/it] 7%|▋ | 2246/30000 [1:25:27<17:40:21, 2.29s/it] 7%|▋ | 2247/30000 [1:25:29<17:49:24, 2.31s/it] 7%|▋ | 2248/30000 [1:25:32<17:49:52, 2.31s/it] 7%|▋ | 2249/30000 [1:25:34<17:43:08, 2.30s/it] 8%|▊ | 2250/30000 [1:25:36<17:43:58, 2.30s/it] 8%|▊ | 2250/30000 [1:25:36<17:43:58, 2.30s/it] 8%|▊ | 2251/30000 [1:25:39<17:47:54, 2.31s/it] 8%|▊ | 2252/30000 [1:25:41<17:40:53, 2.29s/it] 8%|▊ | 2253/30000 [1:25:43<17:34:46, 2.28s/it] 8%|▊ | 2254/30000 [1:25:45<17:40:58, 2.29s/it] 8%|▊ | 2255/30000 [1:25:48<17:40:37, 2.29s/it] 8%|▊ | 2256/30000 [1:25:50<17:36:54, 2.29s/it] 8%|▊ | 2257/30000 [1:25:52<17:39:55, 2.29s/it] 8%|▊ | 2258/30000 [1:25:55<17:42:00, 2.30s/it] 8%|▊ | 2259/30000 [1:25:57<17:39:14, 2.29s/it] 8%|▊ | 2260/30000 [1:25:59<17:45:54, 2.31s/it] 8%|▊ | 2260/30000 [1:25:59<17:45:54, 2.31s/it] 8%|▊ | 2261/30000 [1:26:02<17:45:38, 2.31s/it] 8%|▊ | 2262/30000 [1:26:04<17:38:48, 2.29s/it] 8%|▊ | 2263/30000 [1:26:06<17:38:28, 2.29s/it] 8%|▊ | 2264/30000 [1:26:08<17:36:33, 2.29s/it] 8%|▊ | 2265/30000 [1:26:11<17:38:48, 2.29s/it] 8%|▊ | 2266/30000 [1:26:13<17:37:37, 2.29s/it] 8%|▊ | 2267/30000 [1:26:15<17:46:38, 2.31s/it] 8%|▊ | 2268/30000 [1:26:18<17:47:45, 2.31s/it] 8%|▊ | 2269/30000 [1:26:20<17:39:43, 2.29s/it] 8%|▊ | 2270/30000 [1:26:22<17:50:24, 2.32s/it] 8%|▊ | 2270/30000 [1:26:22<17:50:24, 2.32s/it] 8%|▊ | 2271/30000 [1:26:25<17:52:56, 2.32s/it] 8%|▊ | 2272/30000 [1:26:27<17:46:16, 2.31s/it] 8%|▊ | 2273/30000 [1:26:29<17:43:53, 2.30s/it] 8%|▊ | 2274/30000 [1:26:31<17:43:09, 2.30s/it] 8%|▊ | 2275/30000 [1:26:34<17:37:37, 2.29s/it] 8%|▊ | 2276/30000 [1:26:36<17:39:41, 2.29s/it] 8%|▊ | 2277/30000 [1:26:38<17:44:45, 2.30s/it] 8%|▊ | 2278/30000 [1:26:41<17:34:35, 2.28s/it] 8%|▊ | 2279/30000 [1:26:43<17:32:57, 2.28s/it] 8%|▊ | 2280/30000 [1:26:45<17:41:12, 2.30s/it] 8%|▊ | 2280/30000 [1:26:45<17:41:12, 2.30s/it] 8%|▊ | 2281/30000 [1:26:47<17:38:28, 2.29s/it] 8%|▊ | 2282/30000 [1:26:50<17:36:54, 2.29s/it] 8%|▊ | 2283/30000 [1:26:52<17:46:59, 2.31s/it] 8%|▊ | 2284/30000 [1:26:54<17:42:09, 2.30s/it] 8%|▊ | 2285/30000 [1:26:57<17:37:56, 2.29s/it] 8%|▊ | 2286/30000 [1:26:59<17:32:20, 2.28s/it] 8%|▊ | 2287/30000 [1:27:01<17:32:54, 2.28s/it] 8%|▊ | 2288/30000 [1:27:03<17:33:40, 2.28s/it] 8%|▊ | 2289/30000 [1:27:06<17:34:30, 2.28s/it] 8%|▊ | 2290/30000 [1:27:08<17:43:48, 2.30s/it] 8%|▊ | 2290/30000 [1:27:08<17:43:48, 2.30s/it] 8%|▊ | 2291/30000 [1:27:10<17:51:35, 2.32s/it] 8%|▊ | 2292/30000 [1:27:13<17:43:46, 2.30s/it] 8%|▊ | 2293/30000 [1:27:15<17:41:37, 2.30s/it] 8%|▊ | 2294/30000 [1:27:17<17:51:24, 2.32s/it] 8%|▊ | 2295/30000 [1:27:20<17:49:09, 2.32s/it] 8%|▊ | 2296/30000 [1:27:22<17:42:57, 2.30s/it] 8%|▊ | 2297/30000 [1:27:24<17:40:04, 2.30s/it] 8%|▊ | 2298/30000 [1:27:26<17:37:35, 2.29s/it] 8%|▊ | 2299/30000 [1:27:29<17:34:52, 2.28s/it] 8%|▊ | 2300/30000 [1:27:31<17:31:43, 2.28s/it] 8%|▊ | 2300/30000 [1:27:31<17:31:43, 2.28s/it] 8%|▊ | 2301/30000 [1:27:33<17:41:04, 2.30s/it] 8%|▊ | 2302/30000 [1:27:36<17:37:38, 2.29s/it] 8%|▊ | 2303/30000 [1:27:38<17:34:27, 2.28s/it] 8%|▊ | 2304/30000 [1:27:40<17:41:37, 2.30s/it] 8%|▊ | 2305/30000 [1:27:43<17:36:46, 2.29s/it] 8%|▊ | 2306/30000 [1:27:45<17:33:00, 2.28s/it] 8%|▊ | 2307/30000 [1:27:47<17:30:34, 2.28s/it] 8%|▊ | 2308/30000 [1:27:49<17:37:48, 2.29s/it] 8%|▊ | 2309/30000 [1:27:52<17:39:52, 2.30s/it] 8%|▊ | 2310/30000 [1:27:54<17:33:36, 2.28s/it] 8%|▊ | 2310/30000 [1:27:54<17:33:36, 2.28s/it] 8%|▊ | 2311/30000 [1:27:56<17:33:48, 2.28s/it] 8%|▊ | 2312/30000 [1:27:59<17:37:36, 2.29s/it] 8%|▊ | 2313/30000 [1:28:01<17:29:32, 2.27s/it] 8%|▊ | 2314/30000 [1:28:03<17:27:04, 2.27s/it] 8%|▊ | 2315/30000 [1:28:05<17:37:14, 2.29s/it] 8%|▊ | 2316/30000 [1:28:08<17:39:09, 2.30s/it] 8%|▊ | 2317/30000 [1:28:10<17:34:38, 2.29s/it] 8%|▊ | 2318/30000 [1:28:12<17:26:05, 2.27s/it] 8%|▊ | 2319/30000 [1:28:14<17:35:13, 2.29s/it] 8%|▊ | 2320/30000 [1:28:17<17:32:04, 2.28s/it] 8%|▊ | 2320/30000 [1:28:17<17:32:04, 2.28s/it] 8%|▊ | 2321/30000 [1:28:19<17:33:26, 2.28s/it] 8%|▊ | 2322/30000 [1:28:21<17:34:06, 2.29s/it] 8%|▊ | 2323/30000 [1:28:24<17:38:56, 2.30s/it] 8%|▊ | 2324/30000 [1:28:26<17:33:26, 2.28s/it] 8%|▊ | 2325/30000 [1:28:28<17:30:51, 2.28s/it] 8%|▊ | 2326/30000 [1:28:31<17:39:02, 2.30s/it] 8%|▊ | 2327/30000 [1:28:33<17:47:36, 2.31s/it] 8%|▊ | 2328/30000 [1:28:35<17:41:13, 2.30s/it] 8%|▊ | 2329/30000 [1:28:37<17:37:41, 2.29s/it] 8%|▊ | 2330/30000 [1:28:40<17:39:20, 2.30s/it] 8%|▊ | 2330/30000 [1:28:40<17:39:20, 2.30s/it] 8%|▊ | 2331/30000 [1:28:42<17:32:30, 2.28s/it] 8%|▊ | 2332/30000 [1:28:44<17:23:24, 2.26s/it] 8%|▊ | 2333/30000 [1:28:47<17:36:23, 2.29s/it] 8%|▊ | 2334/30000 [1:28:49<17:40:47, 2.30s/it] 8%|▊ | 2335/30000 [1:28:51<17:36:46, 2.29s/it] 8%|▊ | 2336/30000 [1:28:53<17:31:41, 2.28s/it] 8%|▊ | 2337/30000 [1:28:56<17:37:19, 2.29s/it] 8%|▊ | 2338/30000 [1:28:58<17:32:48, 2.28s/it] 8%|▊ | 2339/30000 [1:29:00<17:28:07, 2.27s/it] 8%|▊ | 2340/30000 [1:29:03<17:36:29, 2.29s/it] 8%|▊ | 2340/30000 [1:29:03<17:36:29, 2.29s/it] 8%|▊ | 2341/30000 [1:29:05<17:40:33, 2.30s/it] 8%|▊ | 2342/30000 [1:29:07<17:37:32, 2.29s/it] 8%|▊ | 2343/30000 [1:29:09<17:40:00, 2.30s/it] 8%|▊ | 2344/30000 [1:29:12<17:37:30, 2.29s/it] 8%|▊ | 2345/30000 [1:29:14<17:32:02, 2.28s/it] 8%|▊ | 2346/30000 [1:29:16<17:27:57, 2.27s/it] 8%|▊ | 2347/30000 [1:29:19<17:34:45, 2.29s/it] 8%|▊ | 2348/30000 [1:29:21<17:40:29, 2.30s/it] 8%|▊ | 2349/30000 [1:29:23<17:35:42, 2.29s/it] 8%|▊ | 2350/30000 [1:29:26<17:40:49, 2.30s/it] 8%|▊ | 2350/30000 [1:29:26<17:40:49, 2.30s/it] 8%|▊ | 2351/30000 [1:29:28<17:41:38, 2.30s/it] 8%|▊ | 2352/30000 [1:29:30<17:36:02, 2.29s/it] 8%|▊ | 2353/30000 [1:29:32<17:41:13, 2.30s/it] 8%|▊ | 2354/30000 [1:29:35<17:42:05, 2.31s/it] 8%|▊ | 2355/30000 [1:29:37<17:37:54, 2.30s/it] 8%|▊ | 2356/30000 [1:29:39<17:39:02, 2.30s/it] 8%|▊ | 2357/30000 [1:29:42<17:41:23, 2.30s/it] 8%|▊ | 2358/30000 [1:29:44<17:36:41, 2.29s/it] 8%|▊ | 2359/30000 [1:29:46<17:26:53, 2.27s/it] 8%|▊ | 2360/30000 [1:29:48<17:41:06, 2.30s/it] 8%|▊ | 2360/30000 [1:29:49<17:41:06, 2.30s/it] 8%|▊ | 2361/30000 [1:29:51<17:46:00, 2.31s/it] 8%|▊ | 2362/30000 [1:29:53<17:39:17, 2.30s/it] 8%|▊ | 2363/30000 [1:29:55<17:42:49, 2.31s/it] 8%|▊ | 2364/30000 [1:29:58<17:38:05, 2.30s/it] 8%|▊ | 2365/30000 [1:30:00<17:33:46, 2.29s/it] 8%|▊ | 2366/30000 [1:30:02<17:41:17, 2.30s/it] 8%|▊ | 2367/30000 [1:30:05<17:40:38, 2.30s/it] 8%|▊ | 2368/30000 [1:30:07<17:35:28, 2.29s/it] 8%|▊ | 2369/30000 [1:30:09<17:40:59, 2.30s/it] 8%|▊ | 2370/30000 [1:30:11<17:40:43, 2.30s/it] 8%|▊ | 2370/30000 [1:30:12<17:40:43, 2.30s/it] 8%|▊ | 2371/30000 [1:30:14<17:33:22, 2.29s/it] 8%|▊ | 2372/30000 [1:30:16<17:40:23, 2.30s/it] 8%|▊ | 2373/30000 [1:30:18<17:41:59, 2.31s/it] 8%|▊ | 2374/30000 [1:30:21<17:39:20, 2.30s/it] 8%|▊ | 2375/30000 [1:30:23<17:38:11, 2.30s/it] 8%|▊ | 2376/30000 [1:30:25<17:41:22, 2.31s/it] 8%|▊ | 2377/30000 [1:30:28<17:36:53, 2.30s/it] 8%|▊ | 2378/30000 [1:30:30<17:36:54, 2.30s/it] 8%|▊ | 2379/30000 [1:30:32<17:46:08, 2.32s/it] 8%|▊ | 2380/30000 [1:30:35<17:45:38, 2.31s/it] 8%|▊ | 2380/30000 [1:30:35<17:45:38, 2.31s/it] 8%|▊ | 2381/30000 [1:30:37<17:41:30, 2.31s/it] 8%|▊ | 2382/30000 [1:30:39<17:47:14, 2.32s/it] 8%|▊ | 2383/30000 [1:30:41<17:45:00, 2.31s/it] 8%|▊ | 2384/30000 [1:30:44<17:41:00, 2.31s/it] 8%|▊ | 2385/30000 [1:30:46<17:38:57, 2.30s/it] 8%|▊ | 2386/30000 [1:30:48<17:40:56, 2.31s/it] 8%|▊ | 2387/30000 [1:30:51<17:33:47, 2.29s/it] 8%|▊ | 2388/30000 [1:30:53<17:30:15, 2.28s/it] 8%|▊ | 2389/30000 [1:30:55<17:33:58, 2.29s/it] 8%|▊ | 2390/30000 [1:30:57<17:31:57, 2.29s/it] 8%|▊ | 2390/30000 [1:30:58<17:31:57, 2.29s/it] 8%|▊ | 2391/30000 [1:31:00<17:28:20, 2.28s/it] 8%|▊ | 2392/30000 [1:31:02<17:39:59, 2.30s/it] 8%|▊ | 2393/30000 [1:31:04<17:43:55, 2.31s/it] 8%|▊ | 2394/30000 [1:31:07<17:35:32, 2.29s/it] 8%|▊ | 2395/30000 [1:31:09<17:31:38, 2.29s/it] 8%|▊ | 2396/30000 [1:31:11<17:39:06, 2.30s/it] 8%|▊ | 2397/30000 [1:31:14<17:33:57, 2.29s/it] 8%|▊ | 2398/30000 [1:31:16<17:36:06, 2.30s/it] 8%|▊ | 2399/30000 [1:31:18<17:44:26, 2.31s/it] 8%|▊ | 2400/30000 [1:31:20<17:36:56, 2.30s/it] 8%|▊ | 2400/30000 [1:31:21<17:36:56, 2.30s/it] 8%|▊ | 2401/30000 [1:31:23<17:34:37, 2.29s/it] 8%|▊ | 2402/30000 [1:31:25<17:30:42, 2.28s/it] 8%|▊ | 2403/30000 [1:31:27<17:42:35, 2.31s/it] 8%|▊ | 2404/30000 [1:31:30<17:42:22, 2.31s/it] 8%|▊ | 2405/30000 [1:31:32<17:38:57, 2.30s/it] 8%|▊ | 2406/30000 [1:31:34<17:40:30, 2.31s/it] 8%|▊ | 2407/30000 [1:31:37<17:37:06, 2.30s/it] 8%|▊ | 2408/30000 [1:31:39<17:33:00, 2.29s/it] 8%|▊ | 2409/30000 [1:31:41<17:28:40, 2.28s/it] 8%|▊ | 2410/30000 [1:31:43<17:39:31, 2.30s/it] 8%|▊ | 2410/30000 [1:31:44<17:39:31, 2.30s/it] 8%|▊ | 2411/30000 [1:31:46<17:38:12, 2.30s/it] 8%|▊ | 2412/30000 [1:31:48<17:33:27, 2.29s/it] 8%|▊ | 2413/30000 [1:31:50<17:39:35, 2.30s/it] 8%|▊ | 2414/30000 [1:31:53<17:39:25, 2.30s/it] 8%|▊ | 2415/30000 [1:31:55<17:33:21, 2.29s/it] 8%|▊ | 2416/30000 [1:31:57<17:30:59, 2.29s/it] 8%|▊ | 2417/30000 [1:32:00<17:39:53, 2.31s/it] 8%|▊ | 2418/30000 [1:32:02<17:42:56, 2.31s/it] 8%|▊ | 2419/30000 [1:32:04<17:36:58, 2.30s/it] 8%|▊ | 2420/30000 [1:32:06<17:29:34, 2.28s/it] 8%|▊ | 2420/30000 [1:32:06<17:29:34, 2.28s/it] 8%|▊ | 2421/30000 [1:32:09<17:42:19, 2.31s/it] 8%|▊ | 2422/30000 [1:32:11<17:38:03, 2.30s/it] 8%|▊ | 2423/30000 [1:32:13<17:34:06, 2.29s/it] 8%|▊ | 2424/30000 [1:32:16<17:35:36, 2.30s/it] 8%|▊ | 2425/30000 [1:32:18<17:35:15, 2.30s/it] 8%|▊ | 2426/30000 [1:32:20<17:31:39, 2.29s/it] 8%|▊ | 2427/30000 [1:32:22<17:32:08, 2.29s/it] 8%|▊ | 2428/30000 [1:32:25<17:38:46, 2.30s/it] 8%|▊ | 2429/30000 [1:32:27<17:37:21, 2.30s/it] 8%|▊ | 2430/30000 [1:32:29<17:35:15, 2.30s/it] 8%|▊ | 2430/30000 [1:32:29<17:35:15, 2.30s/it] 8%|▊ | 2431/30000 [1:32:32<17:32:11, 2.29s/it] 8%|▊ | 2432/30000 [1:32:34<17:41:58, 2.31s/it] 8%|▊ | 2433/30000 [1:32:36<17:46:31, 2.32s/it] 8%|▊ | 2434/30000 [1:32:39<17:43:59, 2.32s/it] 8%|▊ | 2435/30000 [1:32:41<17:39:37, 2.31s/it] 8%|▊ | 2436/30000 [1:32:43<17:46:27, 2.32s/it] 8%|▊ | 2437/30000 [1:32:46<17:39:26, 2.31s/it] 8%|▊ | 2438/30000 [1:32:48<17:31:39, 2.29s/it] 8%|▊ | 2439/30000 [1:32:50<17:36:06, 2.30s/it] 8%|▊ | 2440/30000 [1:32:52<17:34:44, 2.30s/it] 8%|▊ | 2440/30000 [1:32:53<17:34:44, 2.30s/it] 8%|▊ | 2441/30000 [1:32:55<17:27:45, 2.28s/it] 8%|▊ | 2442/30000 [1:32:57<17:29:46, 2.29s/it] 8%|▊ | 2443/30000 [1:32:59<17:42:21, 2.31s/it] 8%|▊ | 2444/30000 [1:33:02<17:42:15, 2.31s/it] 8%|▊ | 2445/30000 [1:33:04<17:36:21, 2.30s/it] 8%|▊ | 2446/30000 [1:33:06<17:33:37, 2.29s/it] 8%|▊ | 2447/30000 [1:33:09<17:42:29, 2.31s/it] 8%|▊ | 2448/30000 [1:33:11<17:46:28, 2.32s/it] 8%|▊ | 2449/30000 [1:33:13<17:39:10, 2.31s/it] 8%|▊ | 2450/30000 [1:33:16<17:36:01, 2.30s/it] 8%|▊ | 2450/30000 [1:33:16<17:36:01, 2.30s/it] 8%|▊ | 2451/30000 [1:33:18<17:40:12, 2.31s/it] 8%|▊ | 2452/30000 [1:33:20<17:35:54, 2.30s/it] 8%|▊ | 2453/30000 [1:33:22<17:33:58, 2.30s/it] 8%|▊ | 2454/30000 [1:33:25<17:33:43, 2.30s/it] 8%|▊ | 2455/30000 [1:33:27<17:32:39, 2.29s/it] 8%|▊ | 2456/30000 [1:33:29<17:25:45, 2.28s/it] 8%|▊ | 2457/30000 [1:33:32<17:25:10, 2.28s/it] 8%|▊ | 2458/30000 [1:33:34<17:33:00, 2.29s/it] 8%|▊ | 2459/30000 [1:33:36<17:37:53, 2.30s/it] 8%|▊ | 2460/30000 [1:33:38<17:35:55, 2.30s/it] 8%|▊ | 2460/30000 [1:33:38<17:35:55, 2.30s/it] 8%|▊ | 2461/30000 [1:33:41<17:33:53, 2.30s/it] 8%|▊ | 2462/30000 [1:33:43<17:31:43, 2.29s/it] 8%|▊ | 2463/30000 [1:33:45<17:25:47, 2.28s/it] 8%|▊ | 2464/30000 [1:33:48<17:27:44, 2.28s/it] 8%|▊ | 2465/30000 [1:33:50<17:31:39, 2.29s/it] 8%|▊ | 2466/30000 [1:33:52<17:32:25, 2.29s/it] 8%|▊ | 2467/30000 [1:33:54<17:27:45, 2.28s/it] 8%|▊ | 2468/30000 [1:33:57<17:19:25, 2.27s/it] 8%|▊ | 2469/30000 [1:33:59<17:26:43, 2.28s/it] 8%|▊ | 2470/30000 [1:34:01<17:32:48, 2.29s/it] 8%|▊ | 2470/30000 [1:34:01<17:32:48, 2.29s/it] 8%|▊ | 2471/30000 [1:34:04<17:29:49, 2.29s/it] 8%|▊ | 2472/30000 [1:34:06<17:24:42, 2.28s/it] 8%|▊ | 2473/30000 [1:34:08<17:32:56, 2.30s/it] 8%|▊ | 2474/30000 [1:34:10<17:25:37, 2.28s/it] 8%|▊ | 2475/30000 [1:34:13<17:23:19, 2.27s/it] 8%|▊ | 2476/30000 [1:34:15<17:32:13, 2.29s/it] 8%|▊ | 2477/30000 [1:34:17<17:29:34, 2.29s/it] 8%|▊ | 2478/30000 [1:34:20<17:24:56, 2.28s/it] 8%|▊ | 2479/30000 [1:34:22<17:29:37, 2.29s/it] 8%|▊ | 2480/30000 [1:34:24<17:30:41, 2.29s/it] 8%|▊ | 2480/30000 [1:34:24<17:30:41, 2.29s/it] 8%|▊ | 2481/30000 [1:34:26<17:30:27, 2.29s/it] 8%|▊ | 2482/30000 [1:34:29<17:25:26, 2.28s/it] 8%|▊ | 2483/30000 [1:34:31<17:32:12, 2.29s/it] 8%|▊ | 2484/30000 [1:34:33<17:32:44, 2.30s/it] 8%|▊ | 2485/30000 [1:34:36<17:30:07, 2.29s/it] 8%|▊ | 2486/30000 [1:34:38<17:37:44, 2.31s/it] 8%|▊ | 2487/30000 [1:34:40<17:40:18, 2.31s/it] 8%|▊ | 2488/30000 [1:34:43<17:35:04, 2.30s/it] 8%|▊ | 2489/30000 [1:34:45<17:32:43, 2.30s/it] 8%|▊ | 2490/30000 [1:34:47<17:40:54, 2.31s/it] 8%|▊ | 2490/30000 [1:34:47<17:40:54, 2.31s/it] 8%|▊ | 2491/30000 [1:34:49<17:31:50, 2.29s/it] 8%|▊ | 2492/30000 [1:34:52<17:29:23, 2.29s/it] 8%|▊ | 2493/30000 [1:34:54<17:31:16, 2.29s/it] 8%|▊ | 2494/30000 [1:34:56<17:29:09, 2.29s/it] 8%|▊ | 2495/30000 [1:34:59<17:32:23, 2.30s/it] 8%|▊ | 2496/30000 [1:35:01<17:35:07, 2.30s/it] 8%|▊ | 2497/30000 [1:35:03<17:30:23, 2.29s/it] 8%|▊ | 2498/30000 [1:35:05<17:27:29, 2.29s/it] 8%|▊ | 2499/30000 [1:35:08<17:34:50, 2.30s/it] 8%|▊ | 2500/30000 [1:35:10<17:36:13, 2.30s/it] 8%|▊ | 2500/30000 [1:35:10<17:36:13, 2.30s/it] 8%|▊ | 2501/30000 [1:35:12<17:30:58, 2.29s/it] 8%|▊ | 2502/30000 [1:35:15<17:36:03, 2.30s/it] 8%|▊ | 2503/30000 [1:35:17<17:35:42, 2.30s/it] 8%|▊ | 2504/30000 [1:35:19<17:27:39, 2.29s/it] 8%|▊ | 2505/30000 [1:35:22<17:31:03, 2.29s/it] 8%|▊ | 2506/30000 [1:35:24<17:40:21, 2.31s/it] 8%|▊ | 2507/30000 [1:35:26<17:30:52, 2.29s/it] 8%|▊ | 2508/30000 [1:35:28<17:26:11, 2.28s/it] 8%|▊ | 2509/30000 [1:35:31<17:32:52, 2.30s/it] 8%|▊ | 2510/30000 [1:35:33<17:24:35, 2.28s/it] 8%|▊ | 2510/30000 [1:35:33<17:24:35, 2.28s/it] 8%|▊ | 2511/30000 [1:35:35<17:20:26, 2.27s/it] 8%|▊ | 2512/30000 [1:35:38<17:30:09, 2.29s/it] 8%|▊ | 2513/30000 [1:35:40<17:30:46, 2.29s/it] 8%|▊ | 2514/30000 [1:35:42<17:28:09, 2.29s/it] 8%|▊ | 2515/30000 [1:35:45<17:37:18, 2.31s/it] 8%|▊ | 2516/30000 [1:35:47<17:36:45, 2.31s/it] 8%|▊ | 2517/30000 [1:35:49<17:33:09, 2.30s/it] 8%|▊ | 2518/30000 [1:35:51<17:35:23, 2.30s/it] 8%|▊ | 2519/30000 [1:35:54<17:36:08, 2.31s/it] 8%|▊ | 2520/30000 [1:35:56<17:26:55, 2.29s/it] 8%|▊ | 2520/30000 [1:35:56<17:26:55, 2.29s/it] 8%|▊ | 2521/30000 [1:35:58<17:26:05, 2.28s/it] 8%|▊ | 2522/30000 [1:36:01<17:31:43, 2.30s/it] 8%|▊ | 2523/30000 [1:36:03<17:27:52, 2.29s/it] 8%|▊ | 2524/30000 [1:36:05<17:23:02, 2.28s/it] 8%|▊ | 2525/30000 [1:36:07<17:31:02, 2.30s/it] 8%|▊ | 2526/30000 [1:36:10<17:32:28, 2.30s/it] 8%|▊ | 2527/30000 [1:36:12<17:23:14, 2.28s/it] 8%|▊ | 2528/30000 [1:36:14<17:31:21, 2.30s/it] 8%|▊ | 2529/30000 [1:36:17<17:33:09, 2.30s/it] 8%|▊ | 2530/30000 [1:36:19<17:30:11, 2.29s/it] 8%|▊ | 2530/30000 [1:36:19<17:30:11, 2.29s/it] 8%|▊ | 2531/30000 [1:36:21<17:27:02, 2.29s/it] 8%|▊ | 2532/30000 [1:36:24<17:36:15, 2.31s/it] 8%|▊ | 2533/30000 [1:36:26<17:39:06, 2.31s/it] 8%|▊ | 2534/30000 [1:36:28<17:30:01, 2.29s/it] 8%|▊ | 2535/30000 [1:36:30<17:37:16, 2.31s/it] 8%|▊ | 2536/30000 [1:36:33<17:33:06, 2.30s/it] 8%|▊ | 2537/30000 [1:36:35<17:22:51, 2.28s/it] 8%|▊ | 2538/30000 [1:36:37<17:22:52, 2.28s/it] 8%|▊ | 2539/30000 [1:36:40<17:31:51, 2.30s/it] 8%|▊ | 2540/30000 [1:36:42<17:31:15, 2.30s/it] +{'loss': 0.0182, 'grad_norm': 0.09858638048171997, 'learning_rate': 2.996280424011379e-05, 'flow_matching_loss': 0.018167013954371213, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.62} +{'loss': 0.0197, 'grad_norm': 0.1150483638048172, 'learning_rate': 2.9961631444583565e-05, 'flow_matching_loss': 0.01969497688114643, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.63} +{'loss': 0.019, 'grad_norm': 0.10183852910995483, 'learning_rate': 2.9960440469260347e-05, 'flow_matching_loss': 0.018967631552368403, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.63} +{'loss': 0.0192, 'grad_norm': 0.11118480563163757, 'learning_rate': 2.995923131559128e-05, 'flow_matching_loss': 0.01922054812312126, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.63} +{'loss': 0.0185, 'grad_norm': 0.09912718832492828, 'learning_rate': 2.99580039850456e-05, 'flow_matching_loss': 0.018541223369538784, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.63} +{'loss': 0.0197, 'grad_norm': 0.10524135082960129, 'learning_rate': 2.9956758479114635e-05, 'flow_matching_loss': 0.019695677421987058, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.64} +{'loss': 0.0196, 'grad_norm': 0.11900739371776581, 'learning_rate': 2.9955494799311787e-05, 'flow_matching_loss': 0.019582299143075944, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.64} +{'loss': 0.0194, 'grad_norm': 0.09284541010856628, 'learning_rate': 2.995421294717254e-05, 'flow_matching_loss': 0.019446412939578293, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.64} +{'loss': 0.0191, 'grad_norm': 0.12375140190124512, 'learning_rate': 2.995291292425448e-05, 'flow_matching_loss': 0.019145808089524506, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.65} +{'loss': 0.0178, 'grad_norm': 0.10429172217845917, 'learning_rate': 2.9951594732137248e-05, 'flow_matching_loss': 0.017825152166187762, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.65} +{'loss': 0.0188, 'grad_norm': 0.13928836584091187, 'learning_rate': 2.995025837242257e-05, 'flow_matching_loss': 0.018849579617381096, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.65} +{'loss': 0.0193, 'grad_norm': 0.12333313375711441, 'learning_rate': 2.994890384673426e-05, 'flow_matching_loss': 0.019289367739111184, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.65} +{'loss': 0.0164, 'grad_norm': 0.10467875748872757, 'learning_rate': 2.9947531156718174e-05, 'flow_matching_loss': 0.016385636944323777, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.66} +{'loss': 0.0187, 'grad_norm': 0.1073923408985138, 'learning_rate': 2.994614030404228e-05, 'flow_matching_loss': 0.01866155294701457, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.66} +{'loss': 0.0191, 'grad_norm': 0.0848807618021965, 'learning_rate': 2.994473129039659e-05, 'flow_matching_loss': 0.01914365654811263, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.66} +{'loss': 0.0184, 'grad_norm': 0.09243578463792801, 'learning_rate': 2.9943304117493183e-05, 'flow_matching_loss': 0.018379815109074115, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.67} +{'loss': 0.0197, 'grad_norm': 0.07869881391525269, 'learning_rate': 2.9941858787066207e-05, 'flow_matching_loss': 0.019703232031315566, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.67} +{'loss': 0.0197, 'grad_norm': 0.06921471655368805, 'learning_rate': 2.9940395300871888e-05, 'flow_matching_loss': 0.019653220474720002, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.67} +{'loss': 0.0185, 'grad_norm': 0.10677073895931244, 'learning_rate': 2.993891366068849e-05, 'flow_matching_loss': 0.01848568581044674, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.68} +{'loss': 0.0177, 'grad_norm': 0.08929085731506348, 'learning_rate': 2.993741386831634e-05, 'flow_matching_loss': 0.017731867823749782, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.68} +{'loss': 0.0187, 'grad_norm': 0.10084296017885208, 'learning_rate': 2.9935895925577837e-05, 'flow_matching_loss': 0.018697912525385618, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.68} +{'loss': 0.0208, 'grad_norm': 0.10279800742864609, 'learning_rate': 2.9934359834317425e-05, 'flow_matching_loss': 0.020753286965191364, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.68} +{'loss': 0.0195, 'grad_norm': 0.1161407083272934, 'learning_rate': 2.993280559640159e-05, 'flow_matching_loss': 0.019533946178853512, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.69} +{'loss': 0.0187, 'grad_norm': 0.08206971734762192, 'learning_rate': 2.9931233213718887e-05, 'flow_matching_loss': 0.018656597286462785, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.69} +{'loss': 0.0203, 'grad_norm': 0.09677283465862274, 'learning_rate': 2.992964268817991e-05, 'flow_matching_loss': 0.020320266485214233, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.69} +{'loss': 0.0195, 'grad_norm': 0.11219554394483566, 'learning_rate': 2.9928034021717286e-05, 'flow_matching_loss': 0.01947930259630084, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.7} +{'loss': 0.0183, 'grad_norm': 0.1100105494260788, 'learning_rate': 2.9926407216285707e-05, 'flow_matching_loss': 0.018299046996980907, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.7} +{'loss': 0.0214, 'grad_norm': 0.10984180867671967, 'learning_rate': 2.9924762273861894e-05, 'flow_matching_loss': 0.021406875737011433, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.7} +{'loss': 0.0201, 'grad_norm': 0.10705562680959702, 'learning_rate': 2.992309919644461e-05, 'flow_matching_loss': 0.020138508174568415, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.7} +{'loss': 0.0185, 'grad_norm': 0.08153270930051804, 'learning_rate': 2.9921417986054633e-05, 'flow_matching_loss': 0.018548611644655467, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.71} +{'loss': 0.0194, 'grad_norm': 0.09491351246833801, 'learning_rate': 2.9919718644734815e-05, 'flow_matching_loss': 0.01939406953752041, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.71} +{'loss': 0.0184, 'grad_norm': 0.08720671385526657, 'learning_rate': 2.991800117455e-05, 'flow_matching_loss': 0.018421135656535626, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.71} +{'loss': 0.0201, 'grad_norm': 0.09193934500217438, 'learning_rate': 2.9916265577587085e-05, 'flow_matching_loss': 0.02005987484008074, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.72} +{'loss': 0.0195, 'grad_norm': 0.09338230639696121, 'learning_rate': 2.9914511855954986e-05, 'flow_matching_loss': 0.019460596051067112, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.72} +{'loss': 0.0196, 'grad_norm': 0.07846778631210327, 'learning_rate': 2.9912740011784626e-05, 'flow_matching_loss': 0.01961137177422643, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.72} +{'loss': 0.0195, 'grad_norm': 0.12197179347276688, 'learning_rate': 2.991095004722898e-05, 'flow_matching_loss': 0.01948198825120926, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.72} +{'loss': 0.0194, 'grad_norm': 0.12225307524204254, 'learning_rate': 2.9909141964463017e-05, 'flow_matching_loss': 0.01944761462509632, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.73} +{'loss': 0.0197, 'grad_norm': 0.1040320172905922, 'learning_rate': 2.990731576568372e-05, 'flow_matching_loss': 0.019709926564246416, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.73} +{'loss': 0.0187, 'grad_norm': 0.11408799141645432, 'learning_rate': 2.99054714531101e-05, 'flow_matching_loss': 0.0187241074629128, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.73} +{'loss': 0.0184, 'grad_norm': 0.12192201614379883, 'learning_rate': 2.9903609028983178e-05, 'flow_matching_loss': 0.018447735998779535, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.74} +{'loss': 0.0178, 'grad_norm': 0.11018020659685135, 'learning_rate': 2.990172849556596e-05, 'flow_matching_loss': 0.01776484688743949, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.74} 8%|▊ | 2540/30000 [1:36:42<17:31:15, 2.30s/it] 8%|▊ | 2541/30000 [1:36:44<17:30:32, 2.30s/it] 8%|▊ | 2542/30000 [1:36:46<17:33:08, 2.30s/it] 8%|▊ | 2543/30000 [1:36:49<17:30:00, 2.29s/it] 8%|▊ | 2544/30000 [1:36:51<17:29:44, 2.29s/it] 8%|▊ | 2545/30000 [1:36:53<17:29:37, 2.29s/it] 8%|▊ | 2546/30000 [1:36:56<17:41:34, 2.32s/it] 8%|▊ | 2547/30000 [1:36:58<17:37:22, 2.31s/it] 8%|▊ | 2548/30000 [1:37:00<17:34:42, 2.31s/it] 8%|▊ | 2549/30000 [1:37:03<17:27:41, 2.29s/it] 8%|▊ | 2550/30000 [1:37:05<17:28:41, 2.29s/it] 8%|▊ | 2550/30000 [1:37:05<17:28:41, 2.29s/it] 9%|▊ | 2551/30000 [1:37:07<17:22:31, 2.28s/it] 9%|▊ | 2552/30000 [1:37:09<17:19:52, 2.27s/it] 9%|▊ | 2553/30000 [1:37:12<17:28:41, 2.29s/it] 9%|▊ | 2554/30000 [1:37:14<17:29:50, 2.30s/it] 9%|▊ | 2555/30000 [1:37:16<17:20:57, 2.28s/it] 9%|▊ | 2556/30000 [1:37:18<17:17:51, 2.27s/it] 9%|▊ | 2557/30000 [1:37:21<17:21:11, 2.28s/it] 9%|▊ | 2558/30000 [1:37:23<17:18:20, 2.27s/it] 9%|▊ | 2559/30000 [1:37:25<17:16:44, 2.27s/it] 9%|▊ | 2560/30000 [1:37:28<17:16:48, 2.27s/it] 9%|▊ | 2560/30000 [1:37:28<17:16:48, 2.27s/it] 9%|▊ | 2561/30000 [1:37:30<17:14:40, 2.26s/it] 9%|▊ | 2562/30000 [1:37:32<17:15:11, 2.26s/it] 9%|▊ | 2563/30000 [1:37:34<17:15:32, 2.26s/it] 9%|▊ | 2564/30000 [1:37:37<17:20:39, 2.28s/it] 9%|▊ | 2565/30000 [1:37:39<17:27:40, 2.29s/it] 9%|▊ | 2566/30000 [1:37:41<17:22:21, 2.28s/it] 9%|▊ | 2567/30000 [1:37:43<17:18:33, 2.27s/it] 9%|▊ | 2568/30000 [1:37:46<17:25:41, 2.29s/it] 9%|▊ | 2569/30000 [1:37:48<17:23:31, 2.28s/it] 9%|▊ | 2570/30000 [1:37:50<17:21:47, 2.28s/it] 9%|▊ | 2570/30000 [1:37:50<17:21:47, 2.28s/it] 9%|▊ | 2571/30000 [1:37:53<17:22:01, 2.28s/it] 9%|▊ | 2572/30000 [1:37:55<17:30:12, 2.30s/it] 9%|▊ | 2573/30000 [1:37:57<17:22:46, 2.28s/it] 9%|▊ | 2574/30000 [1:37:59<17:23:17, 2.28s/it] 9%|▊ | 2575/30000 [1:38:02<17:31:15, 2.30s/it] 9%|▊ | 2576/30000 [1:38:04<17:28:51, 2.29s/it] 9%|▊ | 2577/30000 [1:38:06<17:27:04, 2.29s/it] 9%|▊ | 2578/30000 [1:38:09<17:22:32, 2.28s/it] 9%|▊ | 2579/30000 [1:38:11<17:31:04, 2.30s/it] 9%|▊ | 2580/30000 [1:38:13<17:28:51, 2.30s/it] 9%|▊ | 2580/30000 [1:38:13<17:28:51, 2.30s/it] 9%|▊ | 2581/30000 [1:38:16<17:24:44, 2.29s/it] 9%|▊ | 2582/30000 [1:38:18<17:33:30, 2.31s/it] 9%|▊ | 2583/30000 [1:38:20<17:27:42, 2.29s/it] 9%|▊ | 2584/30000 [1:38:22<17:21:35, 2.28s/it] 9%|▊ | 2585/30000 [1:38:25<17:21:48, 2.28s/it] 9%|▊ | 2586/30000 [1:38:27<17:21:05, 2.28s/it] 9%|▊ | 2587/30000 [1:38:29<17:15:40, 2.27s/it] 9%|▊ | 2588/30000 [1:38:31<17:16:47, 2.27s/it] 9%|▊ | 2589/30000 [1:38:34<17:24:28, 2.29s/it] 9%|▊ | 2590/30000 [1:38:36<17:18:55, 2.27s/it] 9%|▊ | 2590/30000 [1:38:36<17:18:55, 2.27s/it] 9%|▊ | 2591/30000 [1:38:38<17:17:50, 2.27s/it] 9%|▊ | 2592/30000 [1:38:41<17:27:22, 2.29s/it] 9%|▊ | 2593/30000 [1:38:43<17:17:06, 2.27s/it] 9%|▊ | 2594/30000 [1:38:45<17:16:06, 2.27s/it] 9%|▊ | 2595/30000 [1:38:47<17:23:24, 2.28s/it] 9%|▊ | 2596/30000 [1:38:50<17:20:02, 2.28s/it] 9%|▊ | 2597/30000 [1:38:52<17:22:40, 2.28s/it] 9%|▊ | 2598/30000 [1:38:54<17:36:18, 2.31s/it] 9%|▊ | 2599/30000 [1:38:57<17:34:01, 2.31s/it] 9%|▊ | 2600/30000 [1:38:59<17:26:24, 2.29s/it] 9%|▊ | 2600/30000 [1:38:59<17:26:24, 2.29s/it] 9%|▊ | 2601/30000 [1:39:01<17:29:02, 2.30s/it] 9%|▊ | 2602/30000 [1:39:04<17:27:49, 2.29s/it] 9%|▊ | 2603/30000 [1:39:06<17:25:08, 2.29s/it] 9%|▊ | 2604/30000 [1:39:08<17:31:53, 2.30s/it] 9%|▊ | 2605/30000 [1:39:10<17:33:26, 2.31s/it] 9%|▊ | 2606/30000 [1:39:13<17:28:48, 2.30s/it] 9%|▊ | 2607/30000 [1:39:15<17:26:25, 2.29s/it] 9%|▊ | 2608/30000 [1:39:17<17:31:55, 2.30s/it] 9%|▊ | 2609/30000 [1:39:20<17:26:52, 2.29s/it] 9%|▊ | 2610/30000 [1:39:22<17:25:52, 2.29s/it] 9%|▊ | 2610/30000 [1:39:22<17:25:52, 2.29s/it] 9%|▊ | 2611/30000 [1:39:24<17:32:48, 2.31s/it] 9%|▊ | 2612/30000 [1:39:27<17:31:20, 2.30s/it] 9%|▊ | 2613/30000 [1:39:29<17:26:37, 2.29s/it] 9%|▊ | 2614/30000 [1:39:31<17:34:32, 2.31s/it] 9%|▊ | 2615/30000 [1:39:33<17:35:55, 2.31s/it] 9%|▊ | 2616/30000 [1:39:36<17:32:08, 2.31s/it] 9%|▊ | 2617/30000 [1:39:38<17:44:24, 2.33s/it] 9%|▊ | 2618/30000 [1:39:41<17:43:23, 2.33s/it] 9%|▊ | 2619/30000 [1:39:43<17:36:10, 2.31s/it] 9%|▊ | 2620/30000 [1:39:45<17:31:38, 2.30s/it] 9%|▊ | 2620/30000 [1:39:45<17:31:38, 2.30s/it] 9%|▊ | 2621/30000 [1:39:47<17:29:35, 2.30s/it] 9%|▊ | 2622/30000 [1:39:50<17:22:22, 2.28s/it] 9%|▊ | 2623/30000 [1:39:52<17:27:14, 2.30s/it] 9%|▊ | 2624/30000 [1:39:54<17:29:51, 2.30s/it] 9%|▉ | 2625/30000 [1:39:57<17:27:11, 2.30s/it] 9%|▉ | 2626/30000 [1:39:59<17:22:56, 2.29s/it] 9%|▉ | 2627/30000 [1:40:01<17:31:03, 2.30s/it] 9%|▉ | 2628/30000 [1:40:03<17:26:04, 2.29s/it] 9%|▉ | 2629/30000 [1:40:06<17:24:33, 2.29s/it] 9%|▉ | 2630/30000 [1:40:08<17:30:43, 2.30s/it] 9%|▉ | 2630/30000 [1:40:08<17:30:43, 2.30s/it] 9%|▉ | 2631/30000 [1:40:10<17:36:33, 2.32s/it] 9%|▉ | 2632/30000 [1:40:13<17:31:40, 2.31s/it] 9%|▉ | 2633/30000 [1:40:15<17:37:33, 2.32s/it] 9%|▉ | 2634/30000 [1:40:17<17:37:30, 2.32s/it] 9%|▉ | 2635/30000 [1:40:20<17:29:11, 2.30s/it] 9%|▉ | 2636/30000 [1:40:22<17:25:23, 2.29s/it] 9%|▉ | 2637/30000 [1:40:24<17:33:07, 2.31s/it] 9%|▉ | 2638/30000 [1:40:26<17:26:35, 2.29s/it] 9%|▉ | 2639/30000 [1:40:29<17:25:49, 2.29s/it] 9%|▉ | 2640/30000 [1:40:31<17:35:53, 2.32s/it] 9%|▉ | 2640/30000 [1:40:31<17:35:53, 2.32s/it] 9%|▉ | 2641/30000 [1:40:33<17:39:12, 2.32s/it] 9%|▉ | 2642/30000 [1:40:36<17:29:37, 2.30s/it] 9%|▉ | 2643/30000 [1:40:38<17:29:53, 2.30s/it] 9%|▉ | 2644/30000 [1:40:40<17:27:32, 2.30s/it] 9%|▉ | 2645/30000 [1:40:43<17:23:24, 2.29s/it] 9%|▉ | 2646/30000 [1:40:45<17:24:43, 2.29s/it] 9%|▉ | 2647/30000 [1:40:47<17:35:42, 2.32s/it] 9%|▉ | 2648/30000 [1:40:50<17:35:47, 2.32s/it] 9%|▉ | 2649/30000 [1:40:52<17:29:55, 2.30s/it] 9%|▉ | 2650/30000 [1:40:54<17:35:50, 2.32s/it] 9%|▉ | 2650/30000 [1:40:54<17:35:50, 2.32s/it] 9%|▉ | 2651/30000 [1:40:56<17:26:49, 2.30s/it] 9%|▉ | 2652/30000 [1:40:59<17:24:15, 2.29s/it] 9%|▉ | 2653/30000 [1:41:01<17:25:24, 2.29s/it] 9%|▉ | 2654/30000 [1:41:03<17:29:37, 2.30s/it] 9%|▉ | 2655/30000 [1:41:06<17:24:17, 2.29s/it] 9%|▉ | 2656/30000 [1:41:08<17:27:42, 2.30s/it] 9%|▉ | 2657/30000 [1:41:10<17:32:55, 2.31s/it] 9%|▉ | 2658/30000 [1:41:13<17:30:32, 2.31s/it] 9%|▉ | 2659/30000 [1:41:15<17:21:13, 2.28s/it] 9%|▉ | 2660/30000 [1:41:17<17:19:01, 2.28s/it] 9%|▉ | 2660/30000 [1:41:17<17:19:01, 2.28s/it] 9%|▉ | 2661/30000 [1:41:19<17:27:52, 2.30s/it] 9%|▉ | 2662/30000 [1:41:22<17:24:12, 2.29s/it] 9%|▉ | 2663/30000 [1:41:24<17:23:23, 2.29s/it] 9%|▉ | 2664/30000 [1:41:26<17:22:48, 2.29s/it] 9%|▉ | 2665/30000 [1:41:29<17:26:49, 2.30s/it] 9%|▉ | 2666/30000 [1:41:31<17:22:16, 2.29s/it] 9%|▉ | 2667/30000 [1:41:33<17:12:05, 2.27s/it] 9%|▉ | 2668/30000 [1:41:35<17:23:03, 2.29s/it] 9%|▉ | 2669/30000 [1:41:38<17:26:21, 2.30s/it] 9%|▉ | 2670/30000 [1:41:40<17:23:00, 2.29s/it] 9%|▉ | 2670/30000 [1:41:40<17:23:00, 2.29s/it] 9%|▉ | 2671/30000 [1:41:42<17:20:20, 2.28s/it] 9%|▉ | 2672/30000 [1:41:45<17:29:13, 2.30s/it] 9%|▉ | 2673/30000 [1:41:47<17:20:57, 2.29s/it] 9%|▉ | 2674/30000 [1:41:49<17:28:11, 2.30s/it] 9%|▉ | 2675/30000 [1:41:51<17:25:24, 2.30s/it] 9%|▉ | 2676/30000 [1:41:54<17:32:11, 2.31s/it] 9%|▉ | 2677/30000 [1:41:56<17:33:43, 2.31s/it] 9%|▉ | 2678/30000 [1:41:58<17:34:13, 2.32s/it] 9%|▉ | 2679/30000 [1:42:01<17:27:23, 2.30s/it] 9%|▉ | 2680/30000 [1:42:03<17:31:21, 2.31s/it] 9%|▉ | 2680/30000 [1:42:03<17:31:21, 2.31s/it] 9%|▉ | 2681/30000 [1:42:05<17:31:47, 2.31s/it] 9%|▉ | 2682/30000 [1:42:08<17:25:32, 2.30s/it] 9%|▉ | 2683/30000 [1:42:10<17:20:27, 2.29s/it] 9%|▉ | 2684/30000 [1:42:12<17:14:33, 2.27s/it] 9%|▉ | 2685/30000 [1:42:14<17:09:47, 2.26s/it] 9%|▉ | 2686/30000 [1:42:17<17:07:09, 2.26s/it] 9%|▉ | 2687/30000 [1:42:19<17:17:50, 2.28s/it] 9%|▉ | 2688/30000 [1:42:21<17:23:29, 2.29s/it] 9%|▉ | 2689/30000 [1:42:23<17:18:25, 2.28s/it] 9%|▉ | 2690/30000 [1:42:26<17:16:10, 2.28s/it] 9%|▉ | 2690/30000 [1:42:26<17:16:10, 2.28s/it] 9%|▉ | 2691/30000 [1:42:28<17:30:02, 2.31s/it] 9%|▉ | 2692/30000 [1:42:30<17:28:35, 2.30s/it] 9%|▉ | 2693/30000 [1:42:33<17:22:18, 2.29s/it] 9%|▉ | 2694/30000 [1:42:35<17:32:50, 2.31s/it] 9%|▉ | 2695/30000 [1:42:37<17:25:37, 2.30s/it] 9%|▉ | 2696/30000 [1:42:40<17:20:40, 2.29s/it] 9%|▉ | 2697/30000 [1:42:42<17:18:34, 2.28s/it] 9%|▉ | 2698/30000 [1:42:44<17:27:05, 2.30s/it] 9%|▉ | 2699/30000 [1:42:46<17:26:39, 2.30s/it] 9%|▉ | 2700/30000 [1:42:49<17:21:53, 2.29s/it] 9%|▉ | 2700/30000 [1:42:49<17:21:53, 2.29s/it] 9%|▉ | 2701/30000 [1:42:51<17:18:16, 2.28s/it] 9%|▉ | 2702/30000 [1:42:53<17:26:53, 2.30s/it] 9%|▉ | 2703/30000 [1:42:56<17:25:53, 2.30s/it] 9%|▉ | 2704/30000 [1:42:58<17:18:28, 2.28s/it] 9%|▉ | 2705/30000 [1:43:00<17:16:08, 2.28s/it] 9%|▉ | 2706/30000 [1:43:02<17:21:10, 2.29s/it] 9%|▉ | 2707/30000 [1:43:05<17:17:04, 2.28s/it] 9%|▉ | 2708/30000 [1:43:07<17:15:40, 2.28s/it] 9%|▉ | 2709/30000 [1:43:09<17:19:18, 2.28s/it] 9%|▉ | 2710/30000 [1:43:12<17:19:24, 2.29s/it] 9%|▉ | 2710/30000 [1:43:12<17:19:24, 2.29s/it] 9%|▉ | 2711/30000 [1:43:14<17:13:56, 2.27s/it] 9%|▉ | 2712/30000 [1:43:16<17:11:26, 2.27s/it] 9%|▉ | 2713/30000 [1:43:18<17:20:13, 2.29s/it] 9%|▉ | 2714/30000 [1:43:21<17:23:58, 2.30s/it] 9%|▉ | 2715/30000 [1:43:23<17:22:19, 2.29s/it] 9%|▉ | 2716/30000 [1:43:25<17:27:27, 2.30s/it] 9%|▉ | 2717/30000 [1:43:28<17:21:29, 2.29s/it] 9%|▉ | 2718/30000 [1:43:30<17:19:46, 2.29s/it] 9%|▉ | 2719/30000 [1:43:32<17:14:54, 2.28s/it] 9%|▉ | 2720/30000 [1:43:34<17:20:33, 2.29s/it] 9%|▉ | 2720/30000 [1:43:35<17:20:33, 2.29s/it] 9%|▉ | 2721/30000 [1:43:37<17:15:01, 2.28s/it] 9%|▉ | 2722/30000 [1:43:39<17:10:50, 2.27s/it] 9%|▉ | 2723/30000 [1:43:41<17:20:58, 2.29s/it] 9%|▉ | 2724/30000 [1:43:44<17:25:58, 2.30s/it] 9%|▉ | 2725/30000 [1:43:46<17:16:55, 2.28s/it] 9%|▉ | 2726/30000 [1:43:48<17:23:01, 2.29s/it] 9%|▉ | 2727/30000 [1:43:50<17:23:02, 2.29s/it] 9%|▉ | 2728/30000 [1:43:53<17:18:33, 2.28s/it] 9%|▉ | 2729/30000 [1:43:55<17:25:13, 2.30s/it] 9%|▉ | 2730/30000 [1:43:57<17:22:58, 2.29s/it] 9%|▉ | 2730/30000 [1:43:57<17:22:58, 2.29s/it] 9%|▉ | 2731/30000 [1:44:00<17:17:32, 2.28s/it] 9%|▉ | 2732/30000 [1:44:02<17:16:35, 2.28s/it] 9%|▉ | 2733/30000 [1:44:04<17:23:36, 2.30s/it] 9%|▉ | 2734/30000 [1:44:07<17:19:18, 2.29s/it] 9%|▉ | 2735/30000 [1:44:09<17:21:41, 2.29s/it] 9%|▉ | 2736/30000 [1:44:11<17:32:01, 2.32s/it] 9%|▉ | 2737/30000 [1:44:13<17:32:21, 2.32s/it] 9%|▉ | 2738/30000 [1:44:16<17:24:18, 2.30s/it] 9%|▉ | 2739/30000 [1:44:18<17:32:55, 2.32s/it] 9%|▉ | 2740/30000 [1:44:20<17:30:52, 2.31s/it] 9%|▉ | 2740/30000 [1:44:20<17:30:52, 2.31s/it] 9%|▉ | 2741/30000 [1:44:23<17:28:36, 2.31s/it] 9%|▉ | 2742/30000 [1:44:25<17:32:14, 2.32s/it] 9%|▉ | 2743/30000 [1:44:27<17:30:20, 2.31s/it] 9%|▉ | 2744/30000 [1:44:30<17:23:46, 2.30s/it] 9%|▉ | 2745/30000 [1:44:32<17:21:07, 2.29s/it] 9%|▉ | 2746/30000 [1:44:34<17:27:19, 2.31s/it] 9%|▉ | 2747/30000 [1:44:36<17:21:55, 2.29s/it] 9%|▉ | 2748/30000 [1:44:39<17:20:22, 2.29s/it] 9%|▉ | 2749/30000 [1:44:41<17:32:02, 2.32s/it] 9%|▉ | 2750/30000 [1:44:43<17:31:32, 2.32s/it] 9%|▉ | 2750/30000 [1:44:44<17:31:32, 2.32s/it] 9%|▉ | 2751/30000 [1:44:46<17:27:31, 2.31s/it] 9%|▉ | 2752/30000 [1:44:48<17:33:27, 2.32s/it] 9%|▉ | 2753/30000 [1:44:50<17:33:10, 2.32s/it] 9%|▉ | 2754/30000 [1:44:53<17:27:16, 2.31s/it] 9%|▉ | 2755/30000 [1:44:55<17:23:49, 2.30s/it] 9%|▉ | 2756/30000 [1:44:57<17:28:49, 2.31s/it] 9%|▉ | 2757/30000 [1:45:00<17:28:22, 2.31s/it] 9%|▉ | 2758/30000 [1:45:02<17:29:21, 2.31s/it] 9%|▉ | 2759/30000 [1:45:04<17:35:38, 2.33s/it] 9%|▉ | 2760/30000 [1:45:07<17:36:47, 2.33s/it] 9%|▉ | 2760/30000 [1:45:07<17:36:47, 2.33s/it] 9%|▉ | 2761/30000 [1:45:09<17:34:36, 2.32s/it] 9%|▉ | 2762/30000 [1:45:11<17:37:27, 2.33s/it] 9%|▉ | 2763/30000 [1:45:14<17:30:49, 2.31s/it] 9%|▉ | 2764/30000 [1:45:16<17:28:58, 2.31s/it] 9%|▉ | 2765/30000 [1:45:18<17:28:21, 2.31s/it] 9%|▉ | 2766/30000 [1:45:21<17:33:44, 2.32s/it] 9%|▉ | 2767/30000 [1:45:23<17:31:28, 2.32s/it] 9%|▉ | 2768/30000 [1:45:25<17:29:03, 2.31s/it] 9%|▉ | 2769/30000 [1:45:27<17:35:03, 2.32s/it] 9%|▉ | 2770/30000 [1:45:30<17:34:46, 2.32s/it] 9%|▉ | 2770/30000 [1:45:30<17:34:46, 2.32s/it] 9%|▉ | 2771/30000 [1:45:32<17:30:32, 2.31s/it] 9%|▉ | 2772/30000 [1:45:34<17:27:52, 2.31s/it] 9%|▉ | 2773/30000 [1:45:37<17:28:50, 2.31s/it] 9%|▉ | 2774/30000 [1:45:39<17:22:49, 2.30s/it] 9%|▉ | 2775/30000 [1:45:41<17:21:55, 2.30s/it] 9%|▉ | 2776/30000 [1:45:44<17:30:15, 2.31s/it] 9%|▉ | 2777/30000 [1:45:46<17:26:38, 2.31s/it] 9%|▉ | 2778/30000 [1:45:48<17:21:57, 2.30s/it] 9%|▉ | 2779/30000 [1:45:51<17:25:07, 2.30s/it] 9%|▉ | 2780/30000 [1:45:53<17:31:32, 2.32s/it] 9%|▉ | 2780/30000 [1:45:53<17:31:32, 2.32s/it] 9%|▉ | 2781/30000 [1:45:55<17:28:56, 2.31s/it] 9%|▉ | 2782/30000 [1:45:58<17:34:53, 2.33s/it] 9%|▉ | 2783/30000 [1:46:00<17:35:44, 2.33s/it] 9%|▉ | 2784/30000 [1:46:02<17:24:44, 2.30s/it] 9%|▉ | 2785/30000 [1:46:04<17:21:11, 2.30s/it] 9%|▉ | 2786/30000 [1:46:07<17:25:09, 2.30s/it] 9%|▉ | 2787/30000 [1:46:09<17:21:27, 2.30s/it] 9%|▉ | 2788/30000 [1:46:11<17:15:10, 2.28s/it] 9%|▉ | 2789/30000 [1:46:14<17:26:14, 2.31s/it] 9%|▉ | 2790/30000 [1:46:16<17:20:17, 2.29s/it] 9%|▉ | 2790/30000 [1:46:16<17:20:17, 2.29s/it] 9%|▉ | 2791/30000 [1:46:18<17:21:29, 2.30s/it] 9%|▉ | 2792/30000 [1:46:21<17:27:56, 2.31s/it] 9%|▉ | 2793/30000 [1:46:23<17:28:21, 2.31s/it] 9%|▉ | 2794/30000 [1:46:25<17:23:58, 2.30s/it] 9%|▉ | 2795/30000 [1:46:27<17:30:11, 2.32s/it] 9%|▉ | 2796/30000 [1:46:30<17:31:40, 2.32s/it] 9%|▉ | 2797/30000 [1:46:32<17:28:53, 2.31s/it] 9%|▉ | 2798/30000 [1:46:34<17:26:08, 2.31s/it] 9%|▉ | 2799/30000 [1:46:37<17:34:48, 2.33s/it] 9%|▉ | 2800/30000 [1:46:39<17:30:05, 2.32s/it] 9%|▉ | 2800/30000 [1:46:39<17:30:05, 2.32s/it] 9%|▉ | 2801/30000 [1:46:41<17:26:28, 2.31s/it] 9%|▉ | 2802/30000 [1:46:44<17:34:40, 2.33s/it] 9%|▉ | 2803/30000 [1:46:46<17:34:49, 2.33s/it] 9%|▉ | 2804/30000 [1:46:48<17:29:50, 2.32s/it] 9%|▉ | 2805/30000 [1:46:51<17:28:35, 2.31s/it] 9%|▉ | 2806/30000 [1:46:53<17:26:23, 2.31s/it] 9%|▉ | 2807/30000 [1:46:55<17:21:03, 2.30s/it] 9%|▉ | 2808/30000 [1:46:57<17:17:59, 2.29s/it] 9%|▉ | 2809/30000 [1:47:00<17:24:33, 2.30s/it] 9%|▉ | 2810/30000 [1:47:02<17:19:52, 2.29s/it] 9%|▉ | 2810/30000 [1:47:02<17:19:52, 2.29s/it] 9%|▉ | 2811/30000 [1:47:04<17:16:02, 2.29s/it] 9%|▉ | 2812/30000 [1:47:07<17:19:44, 2.29s/it] 9%|▉ | 2813/30000 [1:47:09<17:17:54, 2.29s/it] 9%|▉ | 2814/30000 [1:47:11<17:11:52, 2.28s/it] 9%|▉ | 2815/30000 [1:47:13<17:07:38, 2.27s/it] 9%|▉ | 2816/30000 [1:47:16<17:14:13, 2.28s/it] 9%|▉ | 2817/30000 [1:47:18<17:10:29, 2.27s/it] 9%|▉ | 2818/30000 [1:47:20<17:06:48, 2.27s/it] 9%|▉ | 2819/30000 [1:47:23<17:20:18, 2.30s/it] 9%|▉ | 2820/30000 [1:47:25<17:26:15, 2.31s/it] 9%|▉ | 2820/30000 [1:47:25<17:26:15, 2.31s/it] 9%|▉ | 2821/30000 [1:47:27<17:20:33, 2.30s/it] 9%|▉ | 2822/30000 [1:47:29<17:15:26, 2.29s/it] 9%|▉ | 2823/30000 [1:47:32<17:25:37, 2.31s/it] 9%|▉ | 2824/30000 [1:47:34<17:17:53, 2.29s/it] 9%|▉ | 2825/30000 [1:47:36<17:12:12, 2.28s/it] 9%|▉ | 2826/30000 [1:47:39<17:19:52, 2.30s/it] 9%|▉ | 2827/30000 [1:47:41<17:20:43, 2.30s/it] 9%|▉ | 2828/30000 [1:47:43<17:15:09, 2.29s/it] 9%|▉ | 2829/30000 [1:47:46<17:17:11, 2.29s/it] 9%|▉ | 2830/30000 [1:47:48<17:25:08, 2.31s/it] 9%|▉ | 2830/30000 [1:47:48<17:25:08, 2.31s/it] 9%|▉ | 2831/30000 [1:47:50<17:20:25, 2.30s/it] 9%|▉ | 2832/30000 [1:47:52<17:15:24, 2.29s/it] 9%|▉ | 2833/30000 [1:47:55<17:13:00, 2.28s/it] 9%|▉ | 2834/30000 [1:47:57<17:17:39, 2.29s/it] 9%|▉ | 2835/30000 [1:47:59<17:16:53, 2.29s/it] 9%|▉ | 2836/30000 [1:48:02<17:13:26, 2.28s/it] 9%|▉ | 2837/30000 [1:48:04<17:20:05, 2.30s/it] 9%|▉ | 2838/30000 [1:48:06<17:24:01, 2.31s/it] 9%|▉ | 2839/30000 [1:48:09<17:21:21, 2.30s/it] 9%|▉ | 2840/30000 [1:48:11<17:09:21, 2.27s/it] 9%|▉ | 2840/30000 [1:48:11<17:09:21, 2.27s/it] 9%|▉ | 2841/30000 [1:48:13<17:18:06, 2.29s/it] 9%|▉ | 2842/30000 [1:48:15<17:16:37, 2.29s/it] 9%|▉ | 2843/30000 [1:48:18<17:15:15, 2.29s/it] 9%|▉ | 2844/30000 [1:48:20<17:14:04, 2.28s/it] 9%|▉ | 2845/30000 [1:48:22<17:16:16, 2.29s/it] 9%|▉ | 2846/30000 [1:48:24<17:08:46, 2.27s/it] 9%|▉ | 2847/30000 [1:48:27<17:11:10, 2.28s/it] 9%|▉ | 2848/30000 [1:48:29<17:14:44, 2.29s/it] 9%|▉ | 2849/30000 [1:48:31<17:10:59, 2.28s/it] 10%|▉ | 2850/30000 [1:48:34<17:12:12, 2.28s/it] 10%|▉ | 2850/30000 [1:48:34<17:12:12, 2.28s/it] 10%|▉ | 2851/30000 [1:48:36<17:12:20, 2.28s/it] 10%|▉ | 2852/30000 [1:48:38<17:19:04, 2.30s/it] 10%|▉ | 2853/30000 [1:48:41<17:21:06, 2.30s/it] 10%|▉ | 2854/30000 [1:48:43<17:13:40, 2.28s/it] 10%|▉ | 2855/30000 [1:48:45<17:10:34, 2.28s/it] 10%|▉ | 2856/30000 [1:48:47<17:15:43, 2.29s/it] 10%|▉ | 2857/30000 [1:48:50<17:12:01, 2.28s/it] 10%|▉ | 2858/30000 [1:48:52<17:04:13, 2.26s/it] 10%|▉ | 2859/30000 [1:48:54<17:17:04, 2.29s/it] 10%|▉ | 2860/30000 [1:48:56<17:15:47, 2.29s/it] 10%|▉ | 2860/30000 [1:48:56<17:15:47, 2.29s/it] 10%|▉ | 2861/30000 [1:48:59<17:10:22, 2.28s/it] 10%|▉ | 2862/30000 [1:49:01<17:14:35, 2.29s/it] 10%|▉ | 2863/30000 [1:49:03<17:15:10, 2.29s/it] 10%|▉ | 2864/30000 [1:49:06<17:14:02, 2.29s/it] 10%|▉ | 2865/30000 [1:49:08<17:09:03, 2.28s/it] 10%|▉ | 2866/30000 [1:49:10<17:19:35, 2.30s/it] 10%|▉ | 2867/30000 [1:49:13<17:24:26, 2.31s/it] 10%|▉ | 2868/30000 [1:49:15<17:24:27, 2.31s/it] 10%|▉ | 2869/30000 [1:49:17<17:21:43, 2.30s/it] 10%|▉ | 2870/30000 [1:49:19<17:29:47, 2.32s/it] 10%|▉ | 2870/30000 [1:49:20<17:29:47, 2.32s/it] 10%|▉ | 2871/30000 [1:49:22<17:28:18, 2.32s/it] 10%|▉ | 2872/30000 [1:49:24<17:21:42, 2.30s/it] 10%|▉ | 2873/30000 [1:49:26<17:17:12, 2.29s/it] 10%|▉ | 2874/30000 [1:49:29<17:16:49, 2.29s/it] 10%|▉ | 2875/30000 [1:49:31<17:13:46, 2.29s/it] 10%|▉ | 2876/30000 [1:49:33<17:13:30, 2.29s/it] 10%|▉ | 2877/30000 [1:49:35<17:15:00, 2.29s/it] 10%|▉ | 2878/30000 [1:49:38<17:17:30, 2.30s/it] 10%|▉ | 2879/30000 [1:49:40<17:10:00, 2.28s/it] 10%|▉ | 2880/30000 [1:49:42<17:16:00, 2.29s/it] 10%|▉ | 2880/30000 [1:49:42<17:16:00, 2.29s/it] 10%|▉ | 2881/30000 [1:49:45<17:13:22, 2.29s/it] 10%|▉ | 2882/30000 [1:49:47<17:13:11, 2.29s/it] 10%|▉ | 2883/30000 [1:49:49<17:11:00, 2.28s/it] 10%|▉ | 2884/30000 [1:49:52<17:17:07, 2.29s/it] 10%|▉ | 2885/30000 [1:49:54<17:10:23, 2.28s/it] 10%|▉ | 2886/30000 [1:49:56<17:10:22, 2.28s/it] 10%|▉ | 2887/30000 [1:49:58<17:18:05, 2.30s/it] 10%|▉ | 2888/30000 [1:50:01<17:20:30, 2.30s/it] 10%|▉ | 2889/30000 [1:50:03<17:12:27, 2.28s/it] 10%|▉ | 2890/30000 [1:50:05<17:19:52, 2.30s/it] 10%|▉ | 2890/30000 [1:50:05<17:19:52, 2.30s/it] 10%|▉ | 2891/30000 [1:50:08<17:18:59, 2.30s/it] 10%|▉ | 2892/30000 [1:50:10<17:14:24, 2.29s/it] 10%|▉ | 2893/30000 [1:50:12<17:15:50, 2.29s/it] 10%|▉ | 2894/30000 [1:50:14<17:17:14, 2.30s/it] 10%|▉ | 2895/30000 [1:50:17<17:13:57, 2.29s/it] 10%|▉ | 2896/30000 [1:50:19<17:12:38, 2.29s/it] 10%|▉ | 2897/30000 [1:50:21<17:17:54, 2.30s/it] 10%|▉ | 2898/30000 [1:50:24<17:21:12, 2.31s/it] 10%|▉ | 2899/30000 [1:50:26<17:17:20, 2.30s/it] 10%|▉ | 2900/30000 [1:50:28<17:21:24, 2.31s/it] 10%|▉ | 2900/30000 [1:50:28<17:21:24, 2.31s/it] 10%|▉ | 2901/30000 [1:50:31<17:26:04, 2.32s/it] 10%|▉ | 2902/30000 [1:50:33<17:19:09, 2.30s/it] 10%|▉ | 2903/30000 [1:50:35<17:17:01, 2.30s/it] 10%|▉ | 2904/30000 [1:50:37<17:23:51, 2.31s/it] 10%|▉ | 2905/30000 [1:50:40<17:15:03, 2.29s/it] 10%|▉ | 2906/30000 [1:50:42<17:13:05, 2.29s/it] 10%|▉ | 2907/30000 [1:50:44<17:22:01, 2.31s/it] 10%|▉ | 2908/30000 [1:50:47<17:23:25, 2.31s/it] 10%|▉ | 2909/30000 [1:50:49<17:18:52, 2.30s/it] 10%|▉ | 2910/30000 [1:50:51<17:25:13, 2.32s/it] 10%|▉ | 2910/30000 [1:50:51<17:25:13, 2.32s/it] 10%|▉ | 2911/30000 [1:50:54<17:23:29, 2.31s/it] 10%|▉ | 2912/30000 [1:50:56<17:17:41, 2.30s/it] 10%|▉ | 2913/30000 [1:50:58<17:18:55, 2.30s/it] 10%|▉ | 2914/30000 [1:51:00<17:19:40, 2.30s/it] 10%|▉ | 2915/30000 [1:51:03<17:13:21, 2.29s/it] 10%|▉ | 2916/30000 [1:51:05<17:18:48, 2.30s/it] 10%|▉ | 2917/30000 [1:51:07<17:17:06, 2.30s/it] 10%|▉ | 2918/30000 [1:51:10<17:12:05, 2.29s/it] 10%|▉ | 2919/30000 [1:51:12<17:16:15, 2.30s/it] 10%|▉ | 2920/30000 [1:51:14<17:10:41, 2.28s/it] 10%|▉ | 2920/30000 [1:51:14<17:10:41, 2.28s/it] 10%|▉ | 2921/30000 [1:51:16<17:10:36, 2.28s/it] 10%|▉ | 2922/30000 [1:51:19<17:07:04, 2.28s/it] 10%|▉ | 2923/30000 [1:51:21<17:07:46, 2.28s/it] 10%|▉ | 2924/30000 [1:51:23<17:05:11, 2.27s/it] 10%|▉ | 2925/30000 [1:51:26<17:02:55, 2.27s/it] 10%|▉ | 2926/30000 [1:51:28<17:12:29, 2.29s/it] 10%|▉ | 2927/30000 [1:51:30<17:08:25, 2.28s/it] 10%|▉ | 2928/30000 [1:51:32<17:07:12, 2.28s/it] 10%|▉ | 2929/30000 [1:51:35<17:15:43, 2.30s/it] 10%|▉ | 2930/30000 [1:51:37<17:15:54, 2.30s/it] 10%|▉ | 2930/30000 [1:51:37<17:15:54, 2.30s/it] 10%|▉ | 2931/30000 [1:51:39<17:13:32, 2.29s/it] 10%|▉ | 2932/30000 [1:51:42<17:18:36, 2.30s/it] 10%|▉ | 2933/30000 [1:51:44<17:20:22, 2.31s/it] 10%|▉ | 2934/30000 [1:51:46<17:15:48, 2.30s/it] 10%|▉ | 2935/30000 [1:51:49<17:16:30, 2.30s/it] 10%|▉ | 2936/30000 [1:51:51<17:20:46, 2.31s/it] 10%|▉ | 2937/30000 [1:51:53<17:12:46, 2.29s/it] 10%|▉ | 2938/30000 [1:51:55<17:14:17, 2.29s/it] 10%|▉ | 2939/30000 [1:51:58<17:18:53, 2.30s/it] 10%|▉ | 2940/30000 [1:52:00<17:20:56, 2.31s/it] 10%|▉ | 2940/30000 [1:52:00<17:20:56, 2.31s/it] 10%|▉ | 2941/30000 [1:52:02<17:18:34, 2.30s/it] 10%|▉ | 2942/30000 [1:52:05<17:21:54, 2.31s/it] 10%|▉ | 2943/30000 [1:52:07<17:18:30, 2.30s/it] 10%|▉ | 2944/30000 [1:52:09<17:15:57, 2.30s/it] 10%|▉ | 2945/30000 [1:52:12<17:15:56, 2.30s/it] 10%|▉ | 2946/30000 [1:52:14<17:22:59, 2.31s/it] 10%|▉ | 2947/30000 [1:52:16<17:17:32, 2.30s/it] 10%|▉ | 2948/30000 [1:52:18<17:13:52, 2.29s/it] 10%|▉ | 2949/30000 [1:52:21<17:25:21, 2.32s/it] 10%|▉ | 2950/30000 [1:52:23<17:21:27, 2.31s/it] 10%|▉ | 2950/30000 [1:52:23<17:21:27, 2.31s/it] 10%|▉ | 2951/30000 [1:52:25<17:12:17, 2.29s/it] 10%|▉ | 2952/30000 [1:52:28<17:09:26, 2.28s/it] 10%|▉ | 2953/30000 [1:52:30<17:14:53, 2.30s/it] 10%|▉ | 2954/30000 [1:52:32<17:09:28, 2.28s/it] 10%|▉ | 2955/30000 [1:52:34<17:05:22, 2.27s/it] 10%|▉ | 2956/30000 [1:52:37<17:13:55, 2.29s/it] 10%|▉ | 2957/30000 [1:52:39<17:17:14, 2.30s/it] 10%|▉ | 2958/30000 [1:52:41<17:09:28, 2.28s/it] 10%|▉ | 2959/30000 [1:52:44<17:13:24, 2.29s/it] 10%|▉ | 2960/30000 [1:52:46<17:13:47, 2.29s/it] +{'loss': 0.0199, 'grad_norm': 0.11644422262907028, 'learning_rate': 2.989982985514348e-05, 'flow_matching_loss': 0.01990019017830491, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.74} +{'loss': 0.0182, 'grad_norm': 0.07951903343200684, 'learning_rate': 2.9897913110022766e-05, 'flow_matching_loss': 0.018186735361814498, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.74} +{'loss': 0.02, 'grad_norm': 0.13351088762283325, 'learning_rate': 2.989597826253284e-05, 'flow_matching_loss': 0.019962443970143795, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.75} +{'loss': 0.02, 'grad_norm': 0.10336884111166, 'learning_rate': 2.989402531502473e-05, 'flow_matching_loss': 0.019971180334687234, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.75} +{'loss': 0.0175, 'grad_norm': 0.1109415739774704, 'learning_rate': 2.989205426987144e-05, 'flow_matching_loss': 0.017464050464332102, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.75} +{'loss': 0.0189, 'grad_norm': 0.09036893397569656, 'learning_rate': 2.9890065129467988e-05, 'flow_matching_loss': 0.018906620889902116, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.76} +{'loss': 0.0182, 'grad_norm': 0.11626835912466049, 'learning_rate': 2.9888057896231364e-05, 'flow_matching_loss': 0.018178500607609747, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.76} +{'loss': 0.0189, 'grad_norm': 0.10080782324075699, 'learning_rate': 2.988603257260054e-05, 'flow_matching_loss': 0.018914533499628305, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.76} +{'loss': 0.02, 'grad_norm': 0.11216042190790176, 'learning_rate': 2.988398916103648e-05, 'flow_matching_loss': 0.01995226228609681, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.77} +{'loss': 0.0186, 'grad_norm': 0.09995784610509872, 'learning_rate': 2.988192766402212e-05, 'flow_matching_loss': 0.018609867244958878, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.77} +{'loss': 0.0187, 'grad_norm': 0.08435508608818054, 'learning_rate': 2.9879848084062378e-05, 'flow_matching_loss': 0.018661878257989883, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.77} +{'loss': 0.018, 'grad_norm': 0.07912006974220276, 'learning_rate': 2.9877750423684132e-05, 'flow_matching_loss': 0.017954350262880326, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.77} +{'loss': 0.0174, 'grad_norm': 0.09589036554098129, 'learning_rate': 2.987563468543624e-05, 'flow_matching_loss': 0.017441228777170182, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.78} +{'loss': 0.0176, 'grad_norm': 0.11459698528051376, 'learning_rate': 2.9873500871889533e-05, 'flow_matching_loss': 0.017584876623004675, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.78} +{'loss': 0.0175, 'grad_norm': 0.10205350071191788, 'learning_rate': 2.9871348985636778e-05, 'flow_matching_loss': 0.017514295037835838, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.78} +{'loss': 0.0192, 'grad_norm': 0.10688585788011551, 'learning_rate': 2.986917902929273e-05, 'flow_matching_loss': 0.01924346461892128, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.79} +{'loss': 0.0193, 'grad_norm': 0.10314787924289703, 'learning_rate': 2.9866991005494095e-05, 'flow_matching_loss': 0.01929849525913596, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.79} +{'loss': 0.0167, 'grad_norm': 0.09474411606788635, 'learning_rate': 2.986478491689952e-05, 'flow_matching_loss': 0.016679305117577314, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.79} +{'loss': 0.0166, 'grad_norm': 0.07706945389509201, 'learning_rate': 2.986256076618961e-05, 'flow_matching_loss': 0.016576969437301158, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.79} +{'loss': 0.018, 'grad_norm': 0.11411388218402863, 'learning_rate': 2.986031855606692e-05, 'flow_matching_loss': 0.017964319605380297, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.8} +{'loss': 0.0201, 'grad_norm': 0.12172304838895798, 'learning_rate': 2.9858058289255954e-05, 'flow_matching_loss': 0.02012732233852148, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.8} +{'loss': 0.0181, 'grad_norm': 0.095218226313591, 'learning_rate': 2.985577996850314e-05, 'flow_matching_loss': 0.01814053561538458, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.8} +{'loss': 0.0179, 'grad_norm': 0.105841726064682, 'learning_rate': 2.9853483596576856e-05, 'flow_matching_loss': 0.017857825849205256, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.81} +{'loss': 0.021, 'grad_norm': 0.10726771503686905, 'learning_rate': 2.9851169176267412e-05, 'flow_matching_loss': 0.02097136788070202, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.81} +{'loss': 0.0202, 'grad_norm': 0.13044854998588562, 'learning_rate': 2.9848836710387044e-05, 'flow_matching_loss': 0.020247659645974636, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.81} +{'loss': 0.0204, 'grad_norm': 0.08266475796699524, 'learning_rate': 2.9846486201769917e-05, 'flow_matching_loss': 0.020375383924692868, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.81} +{'loss': 0.0204, 'grad_norm': 0.10482478141784668, 'learning_rate': 2.9844117653272122e-05, 'flow_matching_loss': 0.020383878983557223, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.82} +{'loss': 0.0182, 'grad_norm': 0.10711780190467834, 'learning_rate': 2.9841731067771676e-05, 'flow_matching_loss': 0.018226347956806423, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.82} +{'loss': 0.0187, 'grad_norm': 0.12391617894172668, 'learning_rate': 2.9839326448168492e-05, 'flow_matching_loss': 0.018655291851609947, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.82} +{'loss': 0.019, 'grad_norm': 0.12120126187801361, 'learning_rate': 2.983690379738442e-05, 'flow_matching_loss': 0.01902230428531766, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.83} +{'loss': 0.0177, 'grad_norm': 0.07339368015527725, 'learning_rate': 2.9834463118363204e-05, 'flow_matching_loss': 0.017729255277663468, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.83} +{'loss': 0.0193, 'grad_norm': 0.11923196911811829, 'learning_rate': 2.98320044140705e-05, 'flow_matching_loss': 0.019286504853516816, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.83} +{'loss': 0.0201, 'grad_norm': 0.08904014527797699, 'learning_rate': 2.9829527687493868e-05, 'flow_matching_loss': 0.020085631869733334, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.84} +{'loss': 0.0172, 'grad_norm': 0.10706599056720734, 'learning_rate': 2.9827032941642765e-05, 'flow_matching_loss': 0.01717304652556777, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.84} +{'loss': 0.0184, 'grad_norm': 0.11302343755960464, 'learning_rate': 2.9824520179548535e-05, 'flow_matching_loss': 0.01837529828771949, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.84} +{'loss': 0.0174, 'grad_norm': 0.09746816754341125, 'learning_rate': 2.9821989404264428e-05, 'flow_matching_loss': 0.01735926577821374, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.84} +{'loss': 0.0189, 'grad_norm': 0.0711943730711937, 'learning_rate': 2.9819440618865572e-05, 'flow_matching_loss': 0.018904269952327014, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.85} +{'loss': 0.0167, 'grad_norm': 0.0913877859711647, 'learning_rate': 2.981687382644898e-05, 'flow_matching_loss': 0.01671912157908082, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.85} +{'loss': 0.0186, 'grad_norm': 0.07872047275304794, 'learning_rate': 2.9814289030133556e-05, 'flow_matching_loss': 0.018588849902153017, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.85} +{'loss': 0.0179, 'grad_norm': 0.08371059596538544, 'learning_rate': 2.9811686233060058e-05, 'flow_matching_loss': 0.017932650446891785, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.86} +{'loss': 0.0168, 'grad_norm': 0.08201240003108978, 'learning_rate': 2.9809065438391138e-05, 'flow_matching_loss': 0.016837144922465085, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.86} +{'loss': 0.0187, 'grad_norm': 0.11473485827445984, 'learning_rate': 2.9806426649311304e-05, 'flow_matching_loss': 0.018669214565306902, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.86} 10%|▉ | 2960/30000 [1:52:46<17:13:47, 2.29s/it] 10%|▉ | 2961/30000 [1:52:48<17:07:53, 2.28s/it] 10%|▉ | 2962/30000 [1:52:51<17:08:04, 2.28s/it] 10%|▉ | 2963/30000 [1:52:53<17:06:45, 2.28s/it] 10%|▉ | 2964/30000 [1:52:55<17:04:02, 2.27s/it] 10%|▉ | 2965/30000 [1:52:57<17:01:24, 2.27s/it] 10%|▉ | 2966/30000 [1:53:00<17:10:39, 2.29s/it] 10%|▉ | 2967/30000 [1:53:02<17:08:15, 2.28s/it] 10%|▉ | 2968/30000 [1:53:04<16:51:00, 2.24s/it] 10%|▉ | 2969/30000 [1:53:06<17:05:21, 2.28s/it] 10%|▉ | 2970/30000 [1:53:09<17:15:39, 2.30s/it] 10%|▉ | 2970/30000 [1:53:09<17:15:39, 2.30s/it] 10%|▉ | 2971/30000 [1:53:11<17:14:22, 2.30s/it] 10%|▉ | 2972/30000 [1:53:13<17:10:06, 2.29s/it] 10%|▉ | 2973/30000 [1:53:16<17:16:48, 2.30s/it] 10%|▉ | 2974/30000 [1:53:18<17:18:08, 2.30s/it] 10%|▉ | 2975/30000 [1:53:20<17:14:05, 2.30s/it] 10%|▉ | 2976/30000 [1:53:22<17:09:08, 2.28s/it] 10%|▉ | 2977/30000 [1:53:25<17:13:23, 2.29s/it] 10%|▉ | 2978/30000 [1:53:27<17:08:45, 2.28s/it] 10%|▉ | 2979/30000 [1:53:29<16:56:31, 2.26s/it] 10%|▉ | 2980/30000 [1:53:32<17:05:46, 2.28s/it] 10%|▉ | 2980/30000 [1:53:32<17:05:46, 2.28s/it] 10%|▉ | 2981/30000 [1:53:34<17:17:32, 2.30s/it] 10%|▉ | 2982/30000 [1:53:36<17:11:30, 2.29s/it] 10%|▉ | 2983/30000 [1:53:38<17:05:17, 2.28s/it] 10%|▉ | 2984/30000 [1:53:41<17:14:39, 2.30s/it] 10%|▉ | 2985/30000 [1:53:43<17:25:30, 2.32s/it] 10%|▉ | 2986/30000 [1:53:45<17:09:41, 2.29s/it] 10%|▉ | 2987/30000 [1:53:48<17:17:10, 2.30s/it] 10%|▉ | 2988/30000 [1:53:50<17:14:12, 2.30s/it] 10%|▉ | 2989/30000 [1:53:52<17:08:16, 2.28s/it] 10%|▉ | 2990/30000 [1:53:55<17:06:34, 2.28s/it] 10%|▉ | 2990/30000 [1:53:55<17:06:34, 2.28s/it] 10%|▉ | 2991/30000 [1:53:57<17:17:53, 2.31s/it] 10%|▉ | 2992/30000 [1:53:59<17:16:18, 2.30s/it] 10%|▉ | 2993/30000 [1:54:01<17:13:24, 2.30s/it] 10%|▉ | 2994/30000 [1:54:04<17:05:15, 2.28s/it] 10%|▉ | 2995/30000 [1:54:06<17:17:53, 2.31s/it] 10%|▉ | 2996/30000 [1:54:08<17:07:37, 2.28s/it] 10%|▉ | 2997/30000 [1:54:11<16:58:58, 2.26s/it] 10%|▉ | 2998/30000 [1:54:13<17:06:34, 2.28s/it] 10%|▉ | 2999/30000 [1:54:15<17:16:17, 2.30s/it] 10%|█ | 3000/30000 [1:54:17<17:08:40, 2.29s/it] 10%|█ | 3000/30000 [1:54:17<17:08:40, 2.29s/it] 10%|█ | 3001/30000 [1:54:20<17:01:52, 2.27s/it] 10%|█ | 3002/30000 [1:54:22<16:58:36, 2.26s/it] 10%|█ | 3003/30000 [1:54:24<16:59:28, 2.27s/it] 10%|█ | 3004/30000 [1:54:26<16:53:17, 2.25s/it] 10%|█ | 3005/30000 [1:54:29<17:07:51, 2.28s/it] 10%|█ | 3006/30000 [1:54:31<17:03:59, 2.28s/it] 10%|█ | 3007/30000 [1:54:33<17:01:37, 2.27s/it] 10%|█ | 3008/30000 [1:54:36<17:10:14, 2.29s/it] 10%|█ | 3009/30000 [1:54:38<17:12:29, 2.30s/it] 10%|█ | 3010/30000 [1:54:40<17:06:58, 2.28s/it] 10%|█ | 3010/30000 [1:54:40<17:06:58, 2.28s/it] 10%|█ | 3011/30000 [1:54:43<17:18:10, 2.31s/it] 10%|█ | 3012/30000 [1:54:45<17:20:33, 2.31s/it] 10%|█ | 3013/30000 [1:54:47<17:11:48, 2.29s/it] 10%|█ | 3014/30000 [1:54:49<17:09:50, 2.29s/it] 10%|█ | 3015/30000 [1:54:52<17:20:25, 2.31s/it] 10%|█ | 3016/30000 [1:54:54<17:23:15, 2.32s/it] 10%|█ | 3017/30000 [1:54:56<17:13:05, 2.30s/it] 10%|█ | 3018/30000 [1:54:59<17:11:34, 2.29s/it] 10%|█ | 3019/30000 [1:55:01<17:12:33, 2.30s/it] 10%|█ | 3020/30000 [1:55:03<17:11:13, 2.29s/it] 10%|█ | 3020/30000 [1:55:03<17:11:13, 2.29s/it] 10%|█ | 3021/30000 [1:55:06<17:08:22, 2.29s/it] 10%|█ | 3022/30000 [1:55:08<17:15:13, 2.30s/it] 10%|█ | 3023/30000 [1:55:10<17:16:14, 2.30s/it] 10%|█ | 3024/30000 [1:55:12<17:14:52, 2.30s/it] 10%|█ | 3025/30000 [1:55:15<17:20:26, 2.31s/it] 10%|█ | 3026/30000 [1:55:17<17:23:51, 2.32s/it] 10%|█ | 3027/30000 [1:55:19<17:18:44, 2.31s/it] 10%|█ | 3028/30000 [1:55:22<17:15:40, 2.30s/it] 10%|█ | 3029/30000 [1:55:24<17:20:04, 2.31s/it] 10%|█ | 3030/30000 [1:55:26<16:55:52, 2.26s/it] 10%|█ | 3030/30000 [1:55:26<16:55:52, 2.26s/it] 10%|█ | 3031/30000 [1:55:28<16:57:53, 2.26s/it] 10%|█ | 3032/30000 [1:55:31<17:09:48, 2.29s/it] 10%|█ | 3033/30000 [1:55:33<17:03:11, 2.28s/it] 10%|█ | 3034/30000 [1:55:35<17:04:53, 2.28s/it] 10%|█ | 3035/30000 [1:55:38<17:13:14, 2.30s/it] 10%|█ | 3036/30000 [1:55:40<17:05:51, 2.28s/it] 10%|█ | 3037/30000 [1:55:42<17:11:00, 2.29s/it] 10%|█ | 3038/30000 [1:55:45<17:15:40, 2.30s/it] 10%|█ | 3039/30000 [1:55:47<17:25:02, 2.33s/it] 10%|█ | 3040/30000 [1:55:49<17:18:24, 2.31s/it] 10%|█ | 3040/30000 [1:55:49<17:18:24, 2.31s/it] 10%|█ | 3041/30000 [1:55:52<17:19:36, 2.31s/it] 10%|█ | 3042/30000 [1:55:54<17:11:03, 2.29s/it] 10%|█ | 3043/30000 [1:55:56<17:09:31, 2.29s/it] 10%|█ | 3044/30000 [1:55:58<16:59:20, 2.27s/it] 10%|█ | 3045/30000 [1:56:01<17:17:58, 2.31s/it] 10%|█ | 3046/30000 [1:56:03<17:15:28, 2.30s/it] 10%|█ | 3047/30000 [1:56:05<17:19:05, 2.31s/it] 10%|█ | 3048/30000 [1:56:08<17:15:06, 2.30s/it] 10%|█ | 3049/30000 [1:56:10<17:03:27, 2.28s/it] 10%|█ | 3050/30000 [1:56:12<17:16:56, 2.31s/it] 10%|█ | 3050/30000 [1:56:12<17:16:56, 2.31s/it] 10%|█ | 3051/30000 [1:56:15<17:13:44, 2.30s/it] 10%|█ | 3052/30000 [1:56:17<17:13:22, 2.30s/it] 10%|█ | 3053/30000 [1:56:19<17:15:18, 2.31s/it] 10%|█ | 3054/30000 [1:56:21<17:15:25, 2.31s/it] 10%|█ | 3055/30000 [1:56:24<17:32:10, 2.34s/it] 10%|█ | 3056/30000 [1:56:26<17:30:08, 2.34s/it] 10%|█ | 3057/30000 [1:56:29<17:32:23, 2.34s/it] 10%|█ | 3058/30000 [1:56:31<17:30:05, 2.34s/it] 10%|█ | 3059/30000 [1:56:33<17:15:21, 2.31s/it] 10%|█ | 3060/30000 [1:56:35<17:21:30, 2.32s/it] 10%|█ | 3060/30000 [1:56:35<17:21:30, 2.32s/it] 10%|█ | 3061/30000 [1:56:38<17:23:13, 2.32s/it] 10%|█ | 3062/30000 [1:56:40<17:17:00, 2.31s/it] 10%|█ | 3063/30000 [1:56:42<17:18:03, 2.31s/it] 10%|█ | 3064/30000 [1:56:45<17:17:48, 2.31s/it] 10%|█ | 3065/30000 [1:56:47<17:06:32, 2.29s/it] 10%|█ | 3066/30000 [1:56:49<17:06:21, 2.29s/it] 10%|█ | 3067/30000 [1:56:52<17:09:26, 2.29s/it] 10%|█ | 3068/30000 [1:56:54<17:03:56, 2.28s/it] 10%|█ | 3069/30000 [1:56:56<17:02:50, 2.28s/it] 10%|█ | 3070/30000 [1:56:58<17:25:20, 2.33s/it] 10%|█ | 3070/30000 [1:56:59<17:25:20, 2.33s/it] 10%|█ | 3071/30000 [1:57:01<17:28:06, 2.34s/it] 10%|█ | 3072/30000 [1:57:03<17:15:00, 2.31s/it] 10%|█ | 3073/30000 [1:57:05<17:06:51, 2.29s/it] 10%|█ | 3074/30000 [1:57:08<17:01:34, 2.28s/it] 10%|█ | 3075/30000 [1:57:10<17:06:01, 2.29s/it] 10%|█ | 3076/30000 [1:57:12<17:07:04, 2.29s/it] 10%|█ | 3077/30000 [1:57:14<17:09:14, 2.29s/it] 10%|█ | 3078/30000 [1:57:17<17:26:05, 2.33s/it] 10%|█ | 3079/30000 [1:57:19<17:18:25, 2.31s/it] 10%|█ | 3080/30000 [1:57:22<17:21:45, 2.32s/it] 10%|█ | 3080/30000 [1:57:22<17:21:45, 2.32s/it] 10%|█ | 3081/30000 [1:57:24<17:19:56, 2.32s/it] 10%|█ | 3082/30000 [1:57:26<17:15:38, 2.31s/it] 10%|█ | 3083/30000 [1:57:28<17:16:25, 2.31s/it] 10%|█ | 3084/30000 [1:57:31<17:14:04, 2.31s/it] 10%|█ | 3085/30000 [1:57:33<17:04:25, 2.28s/it] 10%|█ | 3086/30000 [1:57:35<17:14:01, 2.31s/it] 10%|█ | 3087/30000 [1:57:38<17:15:35, 2.31s/it] 10%|█ | 3088/30000 [1:57:40<17:10:34, 2.30s/it] 10%|█ | 3089/30000 [1:57:42<17:12:30, 2.30s/it] 10%|█ | 3090/30000 [1:57:44<17:10:06, 2.30s/it] 10%|█ | 3090/30000 [1:57:45<17:10:06, 2.30s/it] 10%|█ | 3091/30000 [1:57:47<17:20:28, 2.32s/it] 10%|█ | 3092/30000 [1:57:49<17:17:48, 2.31s/it] 10%|█ | 3093/30000 [1:57:51<17:20:37, 2.32s/it] 10%|█ | 3094/30000 [1:57:54<17:15:08, 2.31s/it] 10%|█ | 3095/30000 [1:57:56<17:04:39, 2.29s/it] 10%|█ | 3096/30000 [1:57:58<16:53:46, 2.26s/it] 10%|█ | 3097/30000 [1:58:00<16:44:10, 2.24s/it] 10%|█ | 3098/30000 [1:58:03<16:55:03, 2.26s/it] 10%|█ | 3099/30000 [1:58:05<17:08:25, 2.29s/it] 10%|█ | 3100/30000 [1:58:07<17:05:08, 2.29s/it] 10%|█ | 3100/30000 [1:58:07<17:05:08, 2.29s/it] 10%|█ | 3101/30000 [1:58:10<17:06:58, 2.29s/it] 10%|█ | 3102/30000 [1:58:12<17:21:12, 2.32s/it] 10%|█ | 3103/30000 [1:58:14<17:18:19, 2.32s/it] 10%|█ | 3104/30000 [1:58:17<17:18:52, 2.32s/it] 10%|█ | 3105/30000 [1:58:19<17:10:02, 2.30s/it] 10%|█ | 3106/30000 [1:58:21<17:18:19, 2.32s/it] 10%|█ | 3107/30000 [1:58:24<17:05:11, 2.29s/it] 10%|█ | 3108/30000 [1:58:26<17:02:34, 2.28s/it] 10%|█ | 3109/30000 [1:58:28<17:10:08, 2.30s/it] 10%|█ | 3110/30000 [1:58:30<17:11:34, 2.30s/it] 10%|█ | 3110/30000 [1:58:30<17:11:34, 2.30s/it] 10%|█ | 3111/30000 [1:58:33<17:10:28, 2.30s/it] 10%|█ | 3112/30000 [1:58:35<17:08:34, 2.30s/it] 10%|█ | 3113/30000 [1:58:37<17:12:13, 2.30s/it] 10%|█ | 3114/30000 [1:58:40<17:05:58, 2.29s/it] 10%|█ | 3115/30000 [1:58:42<16:52:06, 2.26s/it] 10%|█ | 3116/30000 [1:58:44<17:05:39, 2.29s/it] 10%|█ | 3117/30000 [1:58:46<17:17:11, 2.31s/it] 10%|█ | 3118/30000 [1:58:49<17:11:41, 2.30s/it] 10%|█ | 3119/30000 [1:58:51<17:04:50, 2.29s/it] 10%|█ | 3120/30000 [1:58:53<17:09:35, 2.30s/it] 10%|█ | 3120/30000 [1:58:53<17:09:35, 2.30s/it] 10%|█ | 3121/30000 [1:58:56<17:08:40, 2.30s/it] 10%|█ | 3122/30000 [1:58:58<17:01:52, 2.28s/it] 10%|█ | 3123/30000 [1:59:00<17:04:45, 2.29s/it] 10%|█ | 3124/30000 [1:59:02<17:00:34, 2.28s/it] 10%|█ | 3125/30000 [1:59:05<16:47:30, 2.25s/it] 10%|█ | 3126/30000 [1:59:07<17:00:58, 2.28s/it] 10%|█ | 3127/30000 [1:59:09<17:00:59, 2.28s/it] 10%|█ | 3128/30000 [1:59:12<16:59:23, 2.28s/it] 10%|█ | 3129/30000 [1:59:14<17:04:14, 2.29s/it] 10%|█ | 3130/30000 [1:59:16<17:07:27, 2.29s/it] 10%|█ | 3130/30000 [1:59:16<17:07:27, 2.29s/it] 10%|█ | 3131/30000 [1:59:18<17:06:11, 2.29s/it] 10%|█ | 3132/30000 [1:59:21<16:55:11, 2.27s/it] 10%|█ | 3133/30000 [1:59:23<16:59:45, 2.28s/it] 10%|█ | 3134/30000 [1:59:25<16:53:14, 2.26s/it] 10%|█ | 3135/30000 [1:59:28<17:05:13, 2.29s/it] 10%|█ | 3136/30000 [1:59:30<17:02:44, 2.28s/it] 10%|█ | 3137/30000 [1:59:32<17:05:44, 2.29s/it] 10%|█ | 3138/30000 [1:59:34<17:10:10, 2.30s/it] 10%|█ | 3139/30000 [1:59:37<17:09:12, 2.30s/it] 10%|█ | 3140/30000 [1:59:39<17:07:07, 2.29s/it] 10%|█ | 3140/30000 [1:59:39<17:07:07, 2.29s/it] 10%|█ | 3141/30000 [1:59:41<17:12:34, 2.31s/it] 10%|█ | 3142/30000 [1:59:44<17:07:56, 2.30s/it] 10%|█ | 3143/30000 [1:59:46<16:58:49, 2.28s/it] 10%|█ | 3144/30000 [1:59:48<17:04:00, 2.29s/it] 10%|█ | 3145/30000 [1:59:50<17:02:57, 2.29s/it] 10%|█ | 3146/30000 [1:59:53<17:06:06, 2.29s/it] 10%|█ | 3147/30000 [1:59:55<17:07:06, 2.29s/it] 10%|█ | 3148/30000 [1:59:57<17:08:35, 2.30s/it] 10%|█ | 3149/30000 [2:00:00<17:01:36, 2.28s/it] 10%|█ | 3150/30000 [2:00:02<16:54:25, 2.27s/it] 10%|█ | 3150/30000 [2:00:02<16:54:25, 2.27s/it] 11%|█ | 3151/30000 [2:00:04<17:05:32, 2.29s/it] 11%|█ | 3152/30000 [2:00:07<17:14:24, 2.31s/it] 11%|█ | 3153/30000 [2:00:09<17:07:30, 2.30s/it] 11%|█ | 3154/30000 [2:00:11<17:06:34, 2.29s/it] 11%|█ | 3155/30000 [2:00:13<17:07:31, 2.30s/it] 11%|█ | 3156/30000 [2:00:16<16:56:38, 2.27s/it] 11%|█ | 3157/30000 [2:00:18<16:53:20, 2.27s/it] 11%|█ | 3158/30000 [2:00:20<16:50:19, 2.26s/it] 11%|█ | 3159/30000 [2:00:22<16:49:31, 2.26s/it] 11%|█ | 3160/30000 [2:00:25<16:52:36, 2.26s/it] 11%|█ | 3160/30000 [2:00:25<16:52:36, 2.26s/it] 11%|█ | 3161/30000 [2:00:27<16:57:06, 2.27s/it] 11%|█ | 3162/30000 [2:00:29<16:53:50, 2.27s/it] 11%|█ | 3163/30000 [2:00:31<16:57:45, 2.28s/it] 11%|█ | 3164/30000 [2:00:34<17:18:06, 2.32s/it] 11%|█ | 3165/30000 [2:00:36<17:16:56, 2.32s/it] 11%|█ | 3166/30000 [2:00:38<17:09:50, 2.30s/it] 11%|█ | 3167/30000 [2:00:41<17:08:59, 2.30s/it] 11%|█ | 3168/30000 [2:00:43<17:12:09, 2.31s/it] 11%|█ | 3169/30000 [2:00:45<17:13:01, 2.31s/it] 11%|█ | 3170/30000 [2:00:48<17:06:23, 2.30s/it] 11%|█ | 3170/30000 [2:00:48<17:06:23, 2.30s/it] 11%|█ | 3171/30000 [2:00:50<17:25:20, 2.34s/it] 11%|█ | 3172/30000 [2:00:52<17:23:33, 2.33s/it] 11%|█ | 3173/30000 [2:00:55<17:16:27, 2.32s/it] 11%|█ | 3174/30000 [2:00:57<17:08:00, 2.30s/it] 11%|█ | 3175/30000 [2:00:59<17:14:33, 2.31s/it] 11%|█ | 3176/30000 [2:01:02<17:08:17, 2.30s/it] 11%|█ | 3177/30000 [2:01:04<16:56:00, 2.27s/it] 11%|█ | 3178/30000 [2:01:06<16:53:22, 2.27s/it] 11%|█ | 3179/30000 [2:01:08<16:58:03, 2.28s/it] 11%|█ | 3180/30000 [2:01:11<16:55:16, 2.27s/it] 11%|█ | 3180/30000 [2:01:11<16:55:16, 2.27s/it] 11%|█ | 3181/30000 [2:01:13<17:07:25, 2.30s/it] 11%|█ | 3182/30000 [2:01:15<16:58:55, 2.28s/it] 11%|█ | 3183/30000 [2:01:17<16:56:01, 2.27s/it] 11%|█ | 3184/30000 [2:01:20<17:07:02, 2.30s/it] 11%|█ | 3185/30000 [2:01:22<17:12:03, 2.31s/it] 11%|█ | 3186/30000 [2:01:24<17:11:32, 2.31s/it] 11%|█ | 3187/30000 [2:01:27<17:12:59, 2.31s/it] 11%|█ | 3188/30000 [2:01:29<17:07:59, 2.30s/it] 11%|█ | 3189/30000 [2:01:31<17:00:14, 2.28s/it] 11%|█ | 3190/30000 [2:01:34<17:05:33, 2.30s/it] 11%|█ | 3190/30000 [2:01:34<17:05:33, 2.30s/it] 11%|█ | 3191/30000 [2:01:36<17:01:40, 2.29s/it] 11%|█ | 3192/30000 [2:01:38<17:01:46, 2.29s/it] 11%|█ | 3193/30000 [2:01:41<17:08:15, 2.30s/it] 11%|█ | 3194/30000 [2:01:43<17:10:02, 2.31s/it] 11%|█ | 3195/30000 [2:01:45<17:06:40, 2.30s/it] 11%|█ | 3196/30000 [2:01:47<17:05:01, 2.29s/it] 11%|█ | 3197/30000 [2:01:50<17:13:15, 2.31s/it] 11%|█ | 3198/30000 [2:01:52<17:06:40, 2.30s/it] 11%|█ | 3199/30000 [2:01:54<17:02:07, 2.29s/it] 11%|█ | 3200/30000 [2:01:57<17:06:59, 2.30s/it] 11%|█ | 3200/30000 [2:01:57<17:06:59, 2.30s/it] 11%|█ | 3201/30000 [2:01:59<17:12:52, 2.31s/it] 11%|█ | 3202/30000 [2:02:01<17:06:48, 2.30s/it] 11%|█ | 3203/30000 [2:02:04<17:14:50, 2.32s/it] 11%|█ | 3204/30000 [2:02:06<17:11:53, 2.31s/it] 11%|█ | 3205/30000 [2:02:08<17:08:29, 2.30s/it] 11%|█ | 3206/30000 [2:02:11<17:16:13, 2.32s/it] 11%|█ | 3207/30000 [2:02:13<17:15:33, 2.32s/it] 11%|█ | 3208/30000 [2:02:15<17:14:03, 2.32s/it] 11%|█ | 3209/30000 [2:02:17<17:11:09, 2.31s/it] 11%|█ | 3210/30000 [2:02:20<17:09:11, 2.31s/it] 11%|█ | 3210/30000 [2:02:20<17:09:11, 2.31s/it] 11%|█ | 3211/30000 [2:02:22<17:07:15, 2.30s/it] 11%|█ | 3212/30000 [2:02:24<17:01:22, 2.29s/it] 11%|█ | 3213/30000 [2:02:27<17:15:25, 2.32s/it] 11%|█ | 3214/30000 [2:02:29<17:05:13, 2.30s/it] 11%|█ | 3215/30000 [2:02:31<17:01:04, 2.29s/it] 11%|█ | 3216/30000 [2:02:34<17:13:21, 2.31s/it] 11%|█ | 3217/30000 [2:02:36<17:18:34, 2.33s/it] 11%|█ | 3218/30000 [2:02:38<17:22:14, 2.33s/it] 11%|█ | 3219/30000 [2:02:41<17:21:03, 2.33s/it] 11%|█ | 3220/30000 [2:02:43<17:17:29, 2.32s/it] 11%|█ | 3220/30000 [2:02:43<17:17:29, 2.32s/it] 11%|█ | 3221/30000 [2:02:45<16:49:07, 2.26s/it] 11%|█ | 3222/30000 [2:02:47<17:01:58, 2.29s/it] 11%|█ | 3223/30000 [2:02:50<17:05:00, 2.30s/it] 11%|█ | 3224/30000 [2:02:52<16:58:48, 2.28s/it] 11%|█ | 3225/30000 [2:02:54<17:22:34, 2.34s/it] 11%|█ | 3226/30000 [2:02:57<17:17:38, 2.33s/it] 11%|█ | 3227/30000 [2:02:59<17:30:06, 2.35s/it] 11%|█ | 3228/30000 [2:03:01<17:19:53, 2.33s/it] 11%|█ | 3229/30000 [2:03:04<17:27:05, 2.35s/it] 11%|█ | 3230/30000 [2:03:06<17:37:15, 2.37s/it] 11%|█ | 3230/30000 [2:03:06<17:37:15, 2.37s/it] 11%|█ | 3231/30000 [2:03:08<17:24:34, 2.34s/it] 11%|█ | 3232/30000 [2:03:11<17:36:17, 2.37s/it] 11%|█ | 3233/30000 [2:03:13<17:26:48, 2.35s/it] 11%|█ | 3234/30000 [2:03:15<17:11:52, 2.31s/it] 11%|█ | 3235/30000 [2:03:18<17:15:51, 2.32s/it] 11%|█ | 3236/30000 [2:03:20<17:09:54, 2.31s/it] 11%|█ | 3237/30000 [2:03:22<17:06:10, 2.30s/it] 11%|█ | 3238/30000 [2:03:25<16:55:08, 2.28s/it] 11%|█ | 3239/30000 [2:03:27<16:57:23, 2.28s/it] 11%|█ | 3240/30000 [2:03:29<16:50:58, 2.27s/it] 11%|█ | 3240/30000 [2:03:29<16:50:58, 2.27s/it] 11%|█ | 3241/30000 [2:03:31<16:55:56, 2.28s/it] 11%|█ | 3242/30000 [2:03:34<17:01:21, 2.29s/it] 11%|█ | 3243/30000 [2:03:36<16:52:23, 2.27s/it] 11%|█ | 3244/30000 [2:03:38<16:52:16, 2.27s/it] 11%|█ | 3245/30000 [2:03:41<16:57:27, 2.28s/it] 11%|█ | 3246/30000 [2:03:43<17:02:43, 2.29s/it] 11%|█ | 3247/30000 [2:03:45<17:09:06, 2.31s/it] 11%|█ | 3248/30000 [2:03:47<17:06:16, 2.30s/it] 11%|█ | 3249/30000 [2:03:50<17:07:33, 2.30s/it] 11%|█ | 3250/30000 [2:03:52<17:00:05, 2.29s/it] 11%|█ | 3250/30000 [2:03:52<17:00:05, 2.29s/it] 11%|█ | 3251/30000 [2:03:54<16:56:10, 2.28s/it] 11%|█ | 3252/30000 [2:03:57<16:54:51, 2.28s/it] 11%|█ | 3253/30000 [2:03:59<17:04:11, 2.30s/it] 11%|█ | 3254/30000 [2:04:01<17:06:44, 2.30s/it] 11%|█ | 3255/30000 [2:04:04<17:17:55, 2.33s/it] 11%|█ | 3256/30000 [2:04:06<17:15:08, 2.32s/it] 11%|█ | 3257/30000 [2:04:08<17:11:54, 2.32s/it] 11%|█ | 3258/30000 [2:04:11<17:14:05, 2.32s/it] 11%|█ | 3259/30000 [2:04:13<17:25:41, 2.35s/it] 11%|█ | 3260/30000 [2:04:15<17:13:23, 2.32s/it] 11%|█ | 3260/30000 [2:04:15<17:13:23, 2.32s/it] 11%|█ | 3261/30000 [2:04:17<17:06:39, 2.30s/it] 11%|█ | 3262/30000 [2:04:20<17:10:23, 2.31s/it] 11%|█ | 3263/30000 [2:04:22<17:07:20, 2.31s/it] 11%|█ | 3264/30000 [2:04:24<16:55:35, 2.28s/it] 11%|█ | 3265/30000 [2:04:27<17:09:36, 2.31s/it] 11%|█ | 3266/30000 [2:04:29<17:13:40, 2.32s/it] 11%|█ | 3267/30000 [2:04:31<16:59:25, 2.29s/it] 11%|█ | 3268/30000 [2:04:34<17:10:41, 2.31s/it] 11%|█ | 3269/30000 [2:04:36<17:15:03, 2.32s/it] 11%|█ | 3270/30000 [2:04:38<17:10:05, 2.31s/it] 11%|█ | 3270/30000 [2:04:38<17:10:05, 2.31s/it] 11%|█ | 3271/30000 [2:04:40<16:51:18, 2.27s/it] 11%|█ | 3272/30000 [2:04:43<16:47:16, 2.26s/it] 11%|█ | 3273/30000 [2:04:45<16:39:43, 2.24s/it] 11%|█ | 3274/30000 [2:04:47<16:44:40, 2.26s/it] 11%|█ | 3275/30000 [2:04:49<16:50:17, 2.27s/it] 11%|█ | 3276/30000 [2:04:52<16:49:49, 2.27s/it] 11%|█ | 3277/30000 [2:04:54<16:47:07, 2.26s/it] 11%|█ | 3278/30000 [2:04:56<16:57:40, 2.29s/it] 11%|█ | 3279/30000 [2:04:59<16:57:58, 2.29s/it] 11%|█ | 3280/30000 [2:05:01<16:45:07, 2.26s/it] 11%|█ | 3280/30000 [2:05:01<16:45:07, 2.26s/it] 11%|█ | 3281/30000 [2:05:03<16:52:50, 2.27s/it] 11%|█ | 3282/30000 [2:05:05<16:54:14, 2.28s/it] 11%|█ | 3283/30000 [2:05:08<16:51:35, 2.27s/it] 11%|█ | 3284/30000 [2:05:10<16:55:11, 2.28s/it] 11%|█ | 3285/30000 [2:05:12<16:59:16, 2.29s/it] 11%|█ | 3286/30000 [2:05:15<17:02:44, 2.30s/it] 11%|█ | 3287/30000 [2:05:17<17:07:08, 2.31s/it] 11%|█ | 3288/30000 [2:05:19<17:11:56, 2.32s/it] 11%|█ | 3289/30000 [2:05:22<17:16:30, 2.33s/it] 11%|█ | 3290/30000 [2:05:24<17:17:47, 2.33s/it] 11%|█ | 3290/30000 [2:05:24<17:17:47, 2.33s/it] 11%|█ | 3291/30000 [2:05:26<17:09:20, 2.31s/it] 11%|█ | 3292/30000 [2:05:28<17:01:31, 2.29s/it] 11%|█ | 3293/30000 [2:05:31<17:11:02, 2.32s/it] 11%|█ | 3294/30000 [2:05:33<17:16:10, 2.33s/it] 11%|█ | 3295/30000 [2:05:36<17:16:41, 2.33s/it] 11%|█ | 3296/30000 [2:05:38<17:07:21, 2.31s/it] 11%|█ | 3297/30000 [2:05:40<17:09:47, 2.31s/it] 11%|█ | 3298/30000 [2:05:42<17:13:52, 2.32s/it] 11%|█ | 3299/30000 [2:05:45<17:07:03, 2.31s/it] 11%|█ | 3300/30000 [2:05:47<16:55:34, 2.28s/it] 11%|█ | 3300/30000 [2:05:47<16:55:34, 2.28s/it] 11%|█ | 3301/30000 [2:05:49<16:58:26, 2.29s/it] 11%|█ | 3302/30000 [2:05:51<16:54:07, 2.28s/it] 11%|█ | 3303/30000 [2:05:54<16:52:06, 2.27s/it] 11%|█ | 3304/30000 [2:05:56<16:57:16, 2.29s/it] 11%|█ | 3305/30000 [2:05:58<17:00:42, 2.29s/it] 11%|█ | 3306/30000 [2:06:01<16:55:55, 2.28s/it] 11%|█ | 3307/30000 [2:06:03<16:49:31, 2.27s/it] 11%|█ | 3308/30000 [2:06:05<16:57:30, 2.29s/it] 11%|█ | 3309/30000 [2:06:07<16:52:44, 2.28s/it] 11%|█ | 3310/30000 [2:06:10<16:51:50, 2.27s/it] 11%|█ | 3310/30000 [2:06:10<16:51:50, 2.27s/it] 11%|█ | 3311/30000 [2:06:12<16:51:35, 2.27s/it] 11%|█ | 3312/30000 [2:06:14<17:06:00, 2.31s/it] 11%|█ | 3313/30000 [2:06:17<17:22:10, 2.34s/it] 11%|█ | 3314/30000 [2:06:19<17:05:48, 2.31s/it] 11%|█ | 3315/30000 [2:06:21<17:15:04, 2.33s/it] 11%|█ | 3316/30000 [2:06:24<17:16:26, 2.33s/it] 11%|█ | 3317/30000 [2:06:26<17:02:48, 2.30s/it] 11%|█ | 3318/30000 [2:06:28<17:04:40, 2.30s/it] 11%|█ | 3319/30000 [2:06:31<17:38:00, 2.38s/it] 11%|█ | 3320/30000 [2:06:33<17:25:03, 2.35s/it] 11%|█ | 3320/30000 [2:06:33<17:25:03, 2.35s/it] 11%|█ | 3321/30000 [2:06:35<17:22:39, 2.34s/it] 11%|█ | 3322/30000 [2:06:38<17:18:11, 2.33s/it] 11%|█ | 3323/30000 [2:06:40<17:16:31, 2.33s/it] 11%|█ | 3324/30000 [2:06:42<17:11:12, 2.32s/it] 11%|█ | 3325/30000 [2:06:45<17:06:28, 2.31s/it] 11%|█ | 3326/30000 [2:06:47<17:03:13, 2.30s/it] 11%|█ | 3327/30000 [2:06:49<17:08:39, 2.31s/it] 11%|█ | 3328/30000 [2:06:52<17:11:54, 2.32s/it] 11%|█ | 3329/30000 [2:06:54<17:02:05, 2.30s/it] 11%|█ | 3330/30000 [2:06:56<17:01:25, 2.30s/it] 11%|█ | 3330/30000 [2:06:56<17:01:25, 2.30s/it] 11%|█ | 3331/30000 [2:06:58<17:03:47, 2.30s/it] 11%|█ | 3332/30000 [2:07:01<17:13:28, 2.33s/it] 11%|█ | 3333/30000 [2:07:03<17:01:25, 2.30s/it] 11%|█ | 3334/30000 [2:07:05<17:07:56, 2.31s/it] 11%|█ | 3335/30000 [2:07:08<17:06:38, 2.31s/it] 11%|█ | 3336/30000 [2:07:10<17:03:53, 2.30s/it] 11%|█ | 3337/30000 [2:07:12<16:55:20, 2.28s/it] 11%|█ | 3338/30000 [2:07:15<16:59:48, 2.29s/it] 11%|█ | 3339/30000 [2:07:17<16:57:05, 2.29s/it] 11%|█ | 3340/30000 [2:07:19<16:57:23, 2.29s/it] 11%|█ | 3340/30000 [2:07:19<16:57:23, 2.29s/it] 11%|█ | 3341/30000 [2:07:22<17:16:00, 2.33s/it] 11%|█ | 3342/30000 [2:07:24<17:17:12, 2.33s/it] 11%|█ | 3343/30000 [2:07:26<17:13:45, 2.33s/it] 11%|█ | 3344/30000 [2:07:29<17:12:05, 2.32s/it] 11%|█ | 3345/30000 [2:07:31<17:14:24, 2.33s/it] 11%|█ | 3346/30000 [2:07:33<17:10:16, 2.32s/it] 11%|█ | 3347/30000 [2:07:35<16:56:42, 2.29s/it] 11%|█ | 3348/30000 [2:07:38<16:49:06, 2.27s/it] 11%|█ | 3349/30000 [2:07:40<16:53:02, 2.28s/it] 11%|█ | 3350/30000 [2:07:42<16:51:08, 2.28s/it] 11%|█ | 3350/30000 [2:07:42<16:51:08, 2.28s/it] 11%|█ | 3351/30000 [2:07:44<16:45:25, 2.26s/it] 11%|█ | 3352/30000 [2:07:47<16:53:53, 2.28s/it] 11%|█ | 3353/30000 [2:07:49<16:58:01, 2.29s/it] 11%|█ | 3354/30000 [2:07:51<16:49:03, 2.27s/it] 11%|█ | 3355/30000 [2:07:54<16:46:35, 2.27s/it] 11%|█ | 3356/30000 [2:07:56<16:51:37, 2.28s/it] 11%|█ | 3357/30000 [2:07:58<16:51:40, 2.28s/it] 11%|█ | 3358/30000 [2:08:00<16:53:59, 2.28s/it] 11%|█ | 3359/30000 [2:08:03<16:55:37, 2.29s/it] 11%|█ | 3360/30000 [2:08:05<16:57:55, 2.29s/it] 11%|█ | 3360/30000 [2:08:05<16:57:55, 2.29s/it] 11%|█ | 3361/30000 [2:08:07<16:57:07, 2.29s/it] 11%|█ | 3362/30000 [2:08:10<17:03:44, 2.31s/it] 11%|█ | 3363/30000 [2:08:12<17:03:41, 2.31s/it] 11%|█ | 3364/30000 [2:08:14<16:58:29, 2.29s/it] 11%|█ | 3365/30000 [2:08:17<16:55:34, 2.29s/it] 11%|█ | 3366/30000 [2:08:19<17:09:11, 2.32s/it] 11%|█ | 3367/30000 [2:08:21<17:02:58, 2.30s/it] 11%|█ | 3368/30000 [2:08:23<16:58:26, 2.29s/it] 11%|█ | 3369/30000 [2:08:26<17:02:50, 2.30s/it] 11%|█ | 3370/30000 [2:08:28<17:09:45, 2.32s/it] 11%|█ | 3370/30000 [2:08:28<17:09:45, 2.32s/it] 11%|█ | 3371/30000 [2:08:30<17:01:50, 2.30s/it] 11%|█ | 3372/30000 [2:08:33<16:58:42, 2.30s/it] 11%|█ | 3373/30000 [2:08:35<17:02:34, 2.30s/it] 11%|█ | 3374/30000 [2:08:37<17:02:43, 2.30s/it] 11%|█▏ | 3375/30000 [2:08:40<16:56:47, 2.29s/it] 11%|█▏ | 3376/30000 [2:08:42<17:04:29, 2.31s/it] 11%|█▏ | 3377/30000 [2:08:44<16:57:58, 2.29s/it] 11%|█▏ | 3378/30000 [2:08:46<16:56:46, 2.29s/it] 11%|█▏ | 3379/30000 [2:08:49<16:57:02, 2.29s/it] 11%|█▏ | 3380/30000 [2:08:51<16:55:25, 2.29s/it] +{'loss': 0.0194, 'grad_norm': 0.10103511810302734, 'learning_rate': 2.980376986902694e-05, 'flow_matching_loss': 0.019387588649988175, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.86} +{'loss': 0.0183, 'grad_norm': 0.09916017204523087, 'learning_rate': 2.9801095100766276e-05, 'flow_matching_loss': 0.01833098204806447, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.87} +{'loss': 0.0181, 'grad_norm': 0.09585433453321457, 'learning_rate': 2.979840234777942e-05, 'flow_matching_loss': 0.01812260625883937, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.87} +{'loss': 0.0185, 'grad_norm': 0.0918707624077797, 'learning_rate': 2.979569161333831e-05, 'flow_matching_loss': 0.01848535854369402, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.87} +{'loss': 0.0195, 'grad_norm': 0.10949733108282089, 'learning_rate': 2.9792962900736744e-05, 'flow_matching_loss': 0.019470172561705114, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.88} +{'loss': 0.0183, 'grad_norm': 0.08958007395267487, 'learning_rate': 2.9790216213290372e-05, 'flow_matching_loss': 0.018322648573666812, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.88} +{'loss': 0.018, 'grad_norm': 0.0973551943898201, 'learning_rate': 2.9787451554336672e-05, 'flow_matching_loss': 0.017954500485211612, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.88} +{'loss': 0.0197, 'grad_norm': 0.1024533361196518, 'learning_rate': 2.9784668927234972e-05, 'flow_matching_loss': 0.019681660644710063, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.88} +{'loss': 0.0173, 'grad_norm': 0.0963742583990097, 'learning_rate': 2.9781868335366418e-05, 'flow_matching_loss': 0.017294289357960225, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.89} +{'loss': 0.0203, 'grad_norm': 0.1114104837179184, 'learning_rate': 2.9779049782134e-05, 'flow_matching_loss': 0.020310640707612036, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.89} +{'loss': 0.0191, 'grad_norm': 0.10190286487340927, 'learning_rate': 2.9776213270962523e-05, 'flow_matching_loss': 0.01913429582491517, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.89} +{'loss': 0.0172, 'grad_norm': 0.10903666913509369, 'learning_rate': 2.977335880529861e-05, 'flow_matching_loss': 0.0171782236546278, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.9} +{'loss': 0.0162, 'grad_norm': 0.08125019818544388, 'learning_rate': 2.977048638861072e-05, 'flow_matching_loss': 0.01622651591897011, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.9} +{'loss': 0.0188, 'grad_norm': 0.08775831013917923, 'learning_rate': 2.97675960243891e-05, 'flow_matching_loss': 0.018778041750192643, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.9} +{'loss': 0.0183, 'grad_norm': 0.13157816231250763, 'learning_rate': 2.976468771614582e-05, 'flow_matching_loss': 0.018334154784679414, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.9} +{'loss': 0.0178, 'grad_norm': 0.09423839300870895, 'learning_rate': 2.9761761467414736e-05, 'flow_matching_loss': 0.01778433732688427, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.91} +{'loss': 0.0166, 'grad_norm': 0.07193324714899063, 'learning_rate': 2.9758817281751538e-05, 'flow_matching_loss': 0.016575857251882552, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.91} +{'loss': 0.0164, 'grad_norm': 0.10706868022680283, 'learning_rate': 2.975585516273367e-05, 'flow_matching_loss': 0.016401616297662257, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.91} +{'loss': 0.0183, 'grad_norm': 0.07814232259988785, 'learning_rate': 2.975287511396039e-05, 'flow_matching_loss': 0.01829546596854925, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.92} +{'loss': 0.0172, 'grad_norm': 0.09994473308324814, 'learning_rate': 2.9749877139052745e-05, 'flow_matching_loss': 0.017188655305653812, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.92} +{'loss': 0.0176, 'grad_norm': 0.10922437906265259, 'learning_rate': 2.9746861241653555e-05, 'flow_matching_loss': 0.017632752005010844, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.92} +{'loss': 0.0197, 'grad_norm': 0.11382201313972473, 'learning_rate': 2.9743827425427416e-05, 'flow_matching_loss': 0.019671374931931496, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.93} +{'loss': 0.0172, 'grad_norm': 0.11926373094320297, 'learning_rate': 2.9740775694060704e-05, 'flow_matching_loss': 0.017244889214634896, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.93} +{'loss': 0.0197, 'grad_norm': 0.09741265326738358, 'learning_rate': 2.973770605126156e-05, 'flow_matching_loss': 0.01974462904036045, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.93} +{'loss': 0.0201, 'grad_norm': 0.09367163479328156, 'learning_rate': 2.9734618500759886e-05, 'flow_matching_loss': 0.02007892346009612, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.93} +{'loss': 0.0184, 'grad_norm': 0.12068048119544983, 'learning_rate': 2.9731513046307356e-05, 'flow_matching_loss': 0.018431354593485594, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.94} +{'loss': 0.0181, 'grad_norm': 0.12101586163043976, 'learning_rate': 2.972838969167738e-05, 'flow_matching_loss': 0.018067772407084702, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.94} +{'loss': 0.0166, 'grad_norm': 0.08586813509464264, 'learning_rate': 2.972524844066513e-05, 'flow_matching_loss': 0.016623569931834937, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.94} +{'loss': 0.0192, 'grad_norm': 0.10989166796207428, 'learning_rate': 2.9722089297087524e-05, 'flow_matching_loss': 0.01916105281561613, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.95} +{'loss': 0.0185, 'grad_norm': 0.08806995302438736, 'learning_rate': 2.971891226478322e-05, 'flow_matching_loss': 0.018497587833553552, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.95} +{'loss': 0.0166, 'grad_norm': 0.11021354794502258, 'learning_rate': 2.9715717347612615e-05, 'flow_matching_loss': 0.016648671496659516, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.95} +{'loss': 0.0184, 'grad_norm': 0.10511526465415955, 'learning_rate': 2.971250454945783e-05, 'flow_matching_loss': 0.018388393986970186, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.95} +{'loss': 0.0174, 'grad_norm': 0.09247571229934692, 'learning_rate': 2.970927387422272e-05, 'flow_matching_loss': 0.01740486789494753, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.96} +{'loss': 0.0192, 'grad_norm': 0.10625152289867401, 'learning_rate': 2.9706025325832858e-05, 'flow_matching_loss': 0.019228547997772693, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.96} +{'loss': 0.0192, 'grad_norm': 0.10350123047828674, 'learning_rate': 2.9702758908235534e-05, 'flow_matching_loss': 0.019225964974611997, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.96} +{'loss': 0.0203, 'grad_norm': 0.0902726873755455, 'learning_rate': 2.9699474625399765e-05, 'flow_matching_loss': 0.020313945598900318, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.97} +{'loss': 0.0193, 'grad_norm': 0.12675310671329498, 'learning_rate': 2.9696172481316246e-05, 'flow_matching_loss': 0.019294144678860904, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.97} +{'loss': 0.0178, 'grad_norm': 0.11727787554264069, 'learning_rate': 2.9692852479997407e-05, 'flow_matching_loss': 0.017797700595110655, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.97} +{'loss': 0.0177, 'grad_norm': 0.11914455145597458, 'learning_rate': 2.9689514625477364e-05, 'flow_matching_loss': 0.01769801564514637, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.97} +{'loss': 0.021, 'grad_norm': 0.11208023130893707, 'learning_rate': 2.968615892181192e-05, 'flow_matching_loss': 0.02099292604252696, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.98} +{'loss': 0.018, 'grad_norm': 0.1126713976264, 'learning_rate': 2.9682785373078567e-05, 'flow_matching_loss': 0.01800288539379835, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.98} +{'loss': 0.0202, 'grad_norm': 0.1046132817864418, 'learning_rate': 2.9679393983376493e-05, 'flow_matching_loss': 0.02023499784991145, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.98} 11%|█▏ | 3380/30000 [2:08:51<16:55:25, 2.29s/it] 11%|█▏ | 3381/30000 [2:08:53<16:54:38, 2.29s/it] 11%|█▏ | 3382/30000 [2:08:56<16:55:39, 2.29s/it] 11%|█▏ | 3383/30000 [2:08:58<16:59:48, 2.30s/it] 11%|█▏ | 3384/30000 [2:09:00<16:56:13, 2.29s/it] 11%|█▏ | 3385/30000 [2:09:03<16:56:38, 2.29s/it] 11%|█▏ | 3386/30000 [2:09:05<16:57:37, 2.29s/it] 11%|█▏ | 3387/30000 [2:09:07<16:56:07, 2.29s/it] 11%|█▏ | 3388/30000 [2:09:09<16:56:02, 2.29s/it] 11%|█▏ | 3389/30000 [2:09:12<16:55:41, 2.29s/it] 11%|█▏ | 3390/30000 [2:09:14<16:46:37, 2.27s/it] 11%|█▏ | 3390/30000 [2:09:14<16:46:37, 2.27s/it] 11%|█▏ | 3391/30000 [2:09:16<16:48:55, 2.28s/it] 11%|█▏ | 3392/30000 [2:09:19<17:09:30, 2.32s/it] 11%|█▏ | 3393/30000 [2:09:21<17:06:20, 2.31s/it] 11%|█▏ | 3394/30000 [2:09:23<16:55:09, 2.29s/it] 11%|█▏ | 3395/30000 [2:09:25<17:00:33, 2.30s/it] 11%|█▏ | 3396/30000 [2:09:28<17:03:20, 2.31s/it] 11%|█▏ | 3397/30000 [2:09:30<17:12:08, 2.33s/it] 11%|█▏ | 3398/30000 [2:09:32<17:04:42, 2.31s/it] 11%|█▏ | 3399/30000 [2:09:35<16:59:57, 2.30s/it] 11%|█▏ | 3400/30000 [2:09:37<17:02:58, 2.31s/it] 11%|█▏ | 3400/30000 [2:09:37<17:02:58, 2.31s/it] 11%|█▏ | 3401/30000 [2:09:39<16:43:42, 2.26s/it] 11%|█▏ | 3402/30000 [2:09:41<16:39:43, 2.26s/it] 11%|█▏ | 3403/30000 [2:09:44<16:40:44, 2.26s/it] 11%|█▏ | 3404/30000 [2:09:46<16:47:46, 2.27s/it] 11%|█▏ | 3405/30000 [2:09:48<16:48:08, 2.27s/it] 11%|█▏ | 3406/30000 [2:09:51<16:45:51, 2.27s/it] 11%|█▏ | 3407/30000 [2:09:53<17:11:37, 2.33s/it] 11%|█▏ | 3408/30000 [2:09:55<17:05:54, 2.31s/it] 11%|█▏ | 3409/30000 [2:09:58<17:01:37, 2.31s/it] 11%|█▏ | 3410/30000 [2:10:00<16:58:57, 2.30s/it] 11%|█▏ | 3410/30000 [2:10:00<16:58:57, 2.30s/it] 11%|█▏ | 3411/30000 [2:10:02<16:59:49, 2.30s/it] 11%|█▏ | 3412/30000 [2:10:04<16:54:16, 2.29s/it] 11%|█▏ | 3413/30000 [2:10:07<16:48:03, 2.27s/it] 11%|█▏ | 3414/30000 [2:10:09<17:01:51, 2.31s/it] 11%|█▏ | 3415/30000 [2:10:11<17:04:14, 2.31s/it] 11%|█▏ | 3416/30000 [2:10:14<16:52:23, 2.28s/it] 11%|█▏ | 3417/30000 [2:10:16<17:02:12, 2.31s/it] 11%|█▏ | 3418/30000 [2:10:18<17:00:50, 2.30s/it] 11%|█▏ | 3419/30000 [2:10:21<17:04:48, 2.31s/it] 11%|█▏ | 3420/30000 [2:10:23<17:02:05, 2.31s/it] 11%|█▏ | 3420/30000 [2:10:23<17:02:05, 2.31s/it] 11%|█▏ | 3421/30000 [2:10:25<17:07:10, 2.32s/it] 11%|█▏ | 3422/30000 [2:10:28<17:07:04, 2.32s/it] 11%|█▏ | 3423/30000 [2:10:30<16:56:13, 2.29s/it] 11%|█▏ | 3424/30000 [2:10:32<16:51:25, 2.28s/it] 11%|█▏ | 3425/30000 [2:10:34<17:12:53, 2.33s/it] 11%|█▏ | 3426/30000 [2:10:37<17:12:18, 2.33s/it] 11%|█▏ | 3427/30000 [2:10:39<17:02:37, 2.31s/it] 11%|█▏ | 3428/30000 [2:10:41<16:50:44, 2.28s/it] 11%|█▏ | 3429/30000 [2:10:44<16:50:01, 2.28s/it] 11%|█▏ | 3430/30000 [2:10:46<16:39:14, 2.26s/it] 11%|█▏ | 3430/30000 [2:10:46<16:39:14, 2.26s/it] 11%|█▏ | 3431/30000 [2:10:48<16:48:09, 2.28s/it] 11%|█▏ | 3432/30000 [2:10:50<16:48:50, 2.28s/it] 11%|█▏ | 3433/30000 [2:10:53<16:48:43, 2.28s/it] 11%|█▏ | 3434/30000 [2:10:55<16:45:35, 2.27s/it] 11%|█▏ | 3435/30000 [2:10:57<17:04:42, 2.31s/it] 11%|█▏ | 3436/30000 [2:11:00<17:07:51, 2.32s/it] 11%|█▏ | 3437/30000 [2:11:01<14:14:33, 1.93s/it] 11%|█▏ | 3438/30000 [2:11:17<46:48:51, 6.34s/it] 11%|█▏ | 3439/30000 [2:11:20<37:47:32, 5.12s/it] 11%|█▏ | 3440/30000 [2:11:22<31:33:49, 4.28s/it] 11%|█▏ | 3440/30000 [2:11:22<31:33:49, 4.28s/it] 11%|█▏ | 3441/30000 [2:11:24<27:02:46, 3.67s/it] 11%|█▏ | 3442/30000 [2:11:26<24:02:51, 3.26s/it] 11%|█▏ | 3443/30000 [2:11:29<21:52:47, 2.97s/it] 11%|█▏ | 3444/30000 [2:11:31<20:22:06, 2.76s/it] 11%|█▏ | 3445/30000 [2:11:33<19:30:18, 2.64s/it] 11%|█▏ | 3446/30000 [2:11:36<18:43:10, 2.54s/it] 11%|█▏ | 3447/30000 [2:11:38<18:07:31, 2.46s/it] 11%|█▏ | 3448/30000 [2:11:40<17:51:05, 2.42s/it] 11%|█▏ | 3449/30000 [2:11:43<17:37:01, 2.39s/it] 12%|█▏ | 3450/30000 [2:11:45<17:19:13, 2.35s/it] 12%|█▏ | 3450/30000 [2:11:45<17:19:13, 2.35s/it] 12%|█▏ | 3451/30000 [2:11:47<17:10:18, 2.33s/it] 12%|█▏ | 3452/30000 [2:11:49<17:05:34, 2.32s/it] 12%|█▏ | 3453/30000 [2:11:52<16:58:18, 2.30s/it] 12%|█▏ | 3454/30000 [2:11:54<16:53:54, 2.29s/it] 12%|█▏ | 3455/30000 [2:11:56<16:52:59, 2.29s/it] 12%|█▏ | 3456/30000 [2:11:59<17:00:19, 2.31s/it] 12%|█▏ | 3457/30000 [2:12:01<16:55:20, 2.30s/it] 12%|█▏ | 3458/30000 [2:12:03<16:58:14, 2.30s/it] 12%|█▏ | 3459/30000 [2:12:06<17:04:15, 2.32s/it] 12%|█▏ | 3460/30000 [2:12:08<16:55:37, 2.30s/it] 12%|█▏ | 3460/30000 [2:12:08<16:55:37, 2.30s/it] 12%|█▏ | 3461/30000 [2:12:10<16:55:12, 2.30s/it] 12%|█▏ | 3462/30000 [2:12:12<16:52:34, 2.29s/it] 12%|█▏ | 3463/30000 [2:12:15<16:41:08, 2.26s/it] 12%|█▏ | 3464/30000 [2:12:17<16:46:02, 2.27s/it] 12%|█▏ | 3465/30000 [2:12:19<16:43:01, 2.27s/it] 12%|█▏ | 3466/30000 [2:12:21<16:49:44, 2.28s/it] 12%|█▏ | 3467/30000 [2:12:24<16:57:48, 2.30s/it] 12%|█▏ | 3468/30000 [2:12:26<17:00:09, 2.31s/it] 12%|█▏ | 3469/30000 [2:12:28<16:57:37, 2.30s/it] 12%|█▏ | 3470/30000 [2:12:31<17:05:38, 2.32s/it] 12%|█▏ | 3470/30000 [2:12:31<17:05:38, 2.32s/it] 12%|█▏ | 3471/30000 [2:12:33<16:58:33, 2.30s/it] 12%|█▏ | 3472/30000 [2:12:35<16:57:08, 2.30s/it] 12%|█▏ | 3473/30000 [2:12:38<16:55:29, 2.30s/it] 12%|█▏ | 3474/30000 [2:12:40<16:52:57, 2.29s/it] 12%|█▏ | 3475/30000 [2:12:42<16:52:02, 2.29s/it] 12%|█▏ | 3476/30000 [2:12:44<16:56:53, 2.30s/it] 12%|█▏ | 3477/30000 [2:12:47<17:15:56, 2.34s/it] 12%|█▏ | 3478/30000 [2:12:49<17:11:32, 2.33s/it] 12%|█▏ | 3479/30000 [2:12:51<17:01:56, 2.31s/it] 12%|█▏ | 3480/30000 [2:12:54<17:19:30, 2.35s/it] 12%|█▏ | 3480/30000 [2:12:54<17:19:30, 2.35s/it] 12%|█▏ | 3481/30000 [2:12:56<17:14:36, 2.34s/it] 12%|█▏ | 3482/30000 [2:12:58<17:01:56, 2.31s/it] 12%|█▏ | 3483/30000 [2:13:01<17:00:14, 2.31s/it] 12%|█▏ | 3484/30000 [2:13:03<16:58:44, 2.31s/it] 12%|█▏ | 3485/30000 [2:13:05<17:08:52, 2.33s/it] 12%|█▏ | 3486/30000 [2:13:08<17:02:45, 2.31s/it] 12%|█▏ | 3487/30000 [2:13:10<17:04:57, 2.32s/it] 12%|█▏ | 3488/30000 [2:13:12<17:00:28, 2.31s/it] 12%|█▏ | 3489/30000 [2:13:15<16:57:54, 2.30s/it] 12%|█▏ | 3490/30000 [2:13:17<17:03:40, 2.32s/it] 12%|█▏ | 3490/30000 [2:13:17<17:03:40, 2.32s/it] 12%|█▏ | 3491/30000 [2:13:19<17:01:02, 2.31s/it] 12%|█▏ | 3492/30000 [2:13:22<17:05:54, 2.32s/it] 12%|█▏ | 3493/30000 [2:13:24<17:02:19, 2.31s/it] 12%|█▏ | 3494/30000 [2:13:26<16:57:56, 2.30s/it] 12%|█▏ | 3495/30000 [2:13:28<16:53:25, 2.29s/it] 12%|█▏ | 3496/30000 [2:13:31<16:56:15, 2.30s/it] 12%|█▏ | 3497/30000 [2:13:33<16:54:08, 2.30s/it] 12%|█▏ | 3498/30000 [2:13:35<16:54:41, 2.30s/it] 12%|█▏ | 3499/30000 [2:13:38<16:58:48, 2.31s/it] 12%|█▏ | 3500/30000 [2:13:40<17:07:01, 2.33s/it] 12%|█▏ | 3500/30000 [2:13:40<17:07:01, 2.33s/it] 12%|█▏ | 3501/30000 [2:13:42<16:41:03, 2.27s/it] 12%|█▏ | 3502/30000 [2:13:45<16:47:55, 2.28s/it] 12%|█▏ | 3503/30000 [2:13:47<16:55:04, 2.30s/it] 12%|█▏ | 3504/30000 [2:13:49<16:41:25, 2.27s/it] 12%|█▏ | 3505/30000 [2:13:51<16:52:35, 2.29s/it] 12%|█▏ | 3506/30000 [2:13:54<16:53:13, 2.29s/it] 12%|█▏ | 3507/30000 [2:13:56<16:52:04, 2.29s/it] 12%|█▏ | 3508/30000 [2:13:58<16:48:49, 2.28s/it] 12%|█▏ | 3509/30000 [2:14:01<16:58:16, 2.31s/it] 12%|█▏ | 3510/30000 [2:14:03<16:52:14, 2.29s/it] 12%|█▏ | 3510/30000 [2:14:03<16:52:14, 2.29s/it] 12%|█▏ | 3511/30000 [2:14:05<16:34:34, 2.25s/it] 12%|█▏ | 3512/30000 [2:14:07<16:46:40, 2.28s/it] 12%|█▏ | 3513/30000 [2:14:10<16:47:26, 2.28s/it] 12%|█▏ | 3514/30000 [2:14:12<16:47:13, 2.28s/it] 12%|█▏ | 3515/30000 [2:14:14<17:02:35, 2.32s/it] 12%|█▏ | 3516/30000 [2:14:17<17:28:21, 2.38s/it] 12%|█▏ | 3517/30000 [2:14:19<17:23:59, 2.37s/it] 12%|█▏ | 3518/30000 [2:14:21<17:11:07, 2.34s/it] 12%|█▏ | 3519/30000 [2:14:24<17:01:46, 2.32s/it] 12%|█▏ | 3520/30000 [2:14:26<17:07:38, 2.33s/it] 12%|█▏ | 3520/30000 [2:14:26<17:07:38, 2.33s/it] 12%|█▏ | 3521/30000 [2:14:28<17:12:39, 2.34s/it] 12%|█▏ | 3522/30000 [2:14:31<17:10:15, 2.33s/it] 12%|█▏ | 3523/30000 [2:14:33<16:55:06, 2.30s/it] 12%|█▏ | 3524/30000 [2:14:35<17:02:43, 2.32s/it] 12%|█▏ | 3525/30000 [2:14:38<17:06:50, 2.33s/it] 12%|█▏ | 3526/30000 [2:14:40<16:57:18, 2.31s/it] 12%|█▏ | 3527/30000 [2:14:42<16:59:02, 2.31s/it] 12%|█▏ | 3528/30000 [2:14:45<16:57:40, 2.31s/it] 12%|█▏ | 3529/30000 [2:14:47<16:48:45, 2.29s/it] 12%|█▏ | 3530/30000 [2:14:49<16:47:08, 2.28s/it] 12%|█▏ | 3530/30000 [2:14:49<16:47:08, 2.28s/it] 12%|█▏ | 3531/30000 [2:14:51<16:53:38, 2.30s/it] 12%|█▏ | 3532/30000 [2:14:54<16:38:35, 2.26s/it] 12%|█▏ | 3533/30000 [2:14:56<16:40:11, 2.27s/it] 12%|█▏ | 3534/30000 [2:14:58<16:56:28, 2.30s/it] 12%|█▏ | 3535/30000 [2:15:01<16:54:45, 2.30s/it] 12%|█▏ | 3536/30000 [2:15:03<16:41:25, 2.27s/it] 12%|█▏ | 3537/30000 [2:15:05<16:46:09, 2.28s/it] 12%|█▏ | 3538/30000 [2:15:08<17:10:31, 2.34s/it] 12%|█▏ | 3539/30000 [2:15:10<16:57:47, 2.31s/it] 12%|█▏ | 3540/30000 [2:15:12<16:53:58, 2.30s/it] 12%|█▏ | 3540/30000 [2:15:12<16:53:58, 2.30s/it] 12%|█▏ | 3541/30000 [2:15:14<17:05:35, 2.33s/it] 12%|█▏ | 3542/30000 [2:15:17<17:02:49, 2.32s/it] 12%|█▏ | 3543/30000 [2:15:19<16:58:07, 2.31s/it] 12%|█▏ | 3544/30000 [2:15:21<16:45:32, 2.28s/it] 12%|█▏ | 3545/30000 [2:15:24<16:49:51, 2.29s/it] 12%|█▏ | 3546/30000 [2:15:26<16:47:28, 2.29s/it] 12%|█▏ | 3547/30000 [2:15:28<16:44:29, 2.28s/it] 12%|█▏ | 3548/30000 [2:15:30<16:46:32, 2.28s/it] 12%|█▏ | 3549/30000 [2:15:33<16:41:06, 2.27s/it] 12%|█▏ | 3550/30000 [2:15:35<16:25:26, 2.24s/it] 12%|█▏ | 3550/30000 [2:15:35<16:25:26, 2.24s/it] 12%|█▏ | 3551/30000 [2:15:37<16:34:21, 2.26s/it] 12%|█▏ | 3552/30000 [2:15:39<16:40:40, 2.27s/it] 12%|█▏ | 3553/30000 [2:15:42<16:42:25, 2.27s/it] 12%|█▏ | 3554/30000 [2:15:44<16:40:39, 2.27s/it] 12%|█▏ | 3555/30000 [2:15:46<16:44:23, 2.28s/it] 12%|█▏ | 3556/30000 [2:15:49<16:55:28, 2.30s/it] 12%|█▏ | 3557/30000 [2:15:51<16:51:36, 2.30s/it] 12%|█▏ | 3558/30000 [2:15:53<16:46:25, 2.28s/it] 12%|█▏ | 3559/30000 [2:15:56<17:00:50, 2.32s/it] 12%|█▏ | 3560/30000 [2:15:58<16:56:52, 2.31s/it] 12%|█▏ | 3560/30000 [2:15:58<16:56:52, 2.31s/it] 12%|█▏ | 3561/30000 [2:16:00<16:52:54, 2.30s/it] 12%|█▏ | 3562/30000 [2:16:02<16:55:55, 2.31s/it] 12%|█▏ | 3563/30000 [2:16:05<16:54:22, 2.30s/it] 12%|█▏ | 3564/30000 [2:16:07<16:54:49, 2.30s/it] 12%|█▏ | 3565/30000 [2:16:09<16:49:58, 2.29s/it] 12%|█▏ | 3566/30000 [2:16:12<16:54:11, 2.30s/it] 12%|█▏ | 3567/30000 [2:16:14<17:01:49, 2.32s/it] 12%|█▏ | 3568/30000 [2:16:16<16:51:04, 2.30s/it] 12%|█▏ | 3569/30000 [2:16:19<16:50:26, 2.29s/it] 12%|█▏ | 3570/30000 [2:16:21<16:51:57, 2.30s/it] 12%|█▏ | 3570/30000 [2:16:21<16:51:57, 2.30s/it] 12%|█▏ | 3571/30000 [2:16:23<16:45:44, 2.28s/it] 12%|█▏ | 3572/30000 [2:16:25<16:39:18, 2.27s/it] 12%|█▏ | 3573/30000 [2:16:28<16:50:59, 2.30s/it] 12%|█▏ | 3574/30000 [2:16:30<16:40:24, 2.27s/it] 12%|█▏ | 3575/30000 [2:16:32<16:37:10, 2.26s/it] 12%|█▏ | 3576/30000 [2:16:34<16:40:09, 2.27s/it] 12%|█▏ | 3577/30000 [2:16:37<16:39:54, 2.27s/it] 12%|█▏ | 3578/30000 [2:16:39<16:41:50, 2.28s/it] 12%|█▏ | 3579/30000 [2:16:41<16:52:20, 2.30s/it] 12%|█▏ | 3580/30000 [2:16:44<16:54:30, 2.30s/it] 12%|█▏ | 3580/30000 [2:16:44<16:54:30, 2.30s/it] 12%|█▏ | 3581/30000 [2:16:46<16:56:20, 2.31s/it] 12%|█▏ | 3582/30000 [2:16:48<16:58:11, 2.31s/it] 12%|█▏ | 3583/30000 [2:16:51<16:59:28, 2.32s/it] 12%|█▏ | 3584/30000 [2:16:53<16:51:12, 2.30s/it] 12%|█▏ | 3585/30000 [2:16:55<16:57:01, 2.31s/it] 12%|█▏ | 3586/30000 [2:16:58<17:01:32, 2.32s/it] 12%|█▏ | 3587/30000 [2:17:00<16:56:47, 2.31s/it] 12%|█▏ | 3588/30000 [2:17:02<16:49:57, 2.29s/it] 12%|█▏ | 3589/30000 [2:17:04<16:58:55, 2.31s/it] 12%|█▏ | 3590/30000 [2:17:07<16:54:24, 2.30s/it] 12%|█▏ | 3590/30000 [2:17:07<16:54:24, 2.30s/it] 12%|█▏ | 3591/30000 [2:17:09<16:42:02, 2.28s/it] 12%|█▏ | 3592/30000 [2:17:11<16:46:18, 2.29s/it] 12%|█▏ | 3593/30000 [2:17:14<16:45:36, 2.28s/it] 12%|█▏ | 3594/30000 [2:17:16<16:43:52, 2.28s/it] 12%|█▏ | 3595/30000 [2:17:18<16:44:43, 2.28s/it] 12%|█▏ | 3596/30000 [2:17:20<16:55:40, 2.31s/it] 12%|█▏ | 3597/30000 [2:17:23<16:52:21, 2.30s/it] 12%|█▏ | 3598/30000 [2:17:25<16:44:58, 2.28s/it] 12%|█▏ | 3599/30000 [2:17:27<16:53:57, 2.30s/it] 12%|█▏ | 3600/30000 [2:17:30<17:00:36, 2.32s/it] 12%|█▏ | 3600/30000 [2:17:30<17:00:36, 2.32s/it] 12%|█▏ | 3601/30000 [2:17:32<16:58:03, 2.31s/it] 12%|█▏ | 3602/30000 [2:17:34<17:01:47, 2.32s/it] 12%|█▏ | 3603/30000 [2:17:37<17:16:00, 2.35s/it] 12%|█▏ | 3604/30000 [2:17:39<17:08:07, 2.34s/it] 12%|█▏ | 3605/30000 [2:17:41<16:56:19, 2.31s/it] 12%|█▏ | 3606/30000 [2:17:44<16:57:59, 2.31s/it] 12%|█▏ | 3607/30000 [2:17:46<16:59:11, 2.32s/it] 12%|█▏ | 3608/30000 [2:17:48<17:03:12, 2.33s/it] 12%|█▏ | 3609/30000 [2:17:51<16:57:34, 2.31s/it] 12%|█▏ | 3610/30000 [2:17:53<17:03:08, 2.33s/it] 12%|█▏ | 3610/30000 [2:17:53<17:03:08, 2.33s/it] 12%|█▏ | 3611/30000 [2:17:55<16:50:38, 2.30s/it] 12%|█▏ | 3612/30000 [2:17:57<16:44:11, 2.28s/it] 12%|█▏ | 3613/30000 [2:18:00<16:50:16, 2.30s/it] 12%|█▏ | 3614/30000 [2:18:02<16:56:39, 2.31s/it] 12%|█▏ | 3615/30000 [2:18:04<17:04:59, 2.33s/it] 12%|█▏ | 3616/30000 [2:18:07<16:53:50, 2.31s/it] 12%|█▏ | 3617/30000 [2:18:09<16:48:17, 2.29s/it] 12%|█▏ | 3618/30000 [2:18:11<16:45:57, 2.29s/it] 12%|█▏ | 3619/30000 [2:18:14<16:44:36, 2.28s/it] 12%|█▏ | 3620/30000 [2:18:16<16:57:44, 2.31s/it] 12%|█▏ | 3620/30000 [2:18:16<16:57:44, 2.31s/it] 12%|█▏ | 3621/30000 [2:18:18<16:58:56, 2.32s/it] 12%|█▏ | 3622/30000 [2:18:21<16:53:21, 2.31s/it] 12%|█▏ | 3623/30000 [2:18:23<16:56:35, 2.31s/it] 12%|█▏ | 3624/30000 [2:18:25<16:52:40, 2.30s/it] 12%|█▏ | 3625/30000 [2:18:27<16:50:00, 2.30s/it] 12%|█▏ | 3626/30000 [2:18:30<16:53:39, 2.31s/it] 12%|█▏ | 3627/30000 [2:18:32<16:53:12, 2.31s/it] 12%|█▏ | 3628/30000 [2:18:34<16:47:11, 2.29s/it] 12%|█▏ | 3629/30000 [2:18:37<16:47:51, 2.29s/it] 12%|█▏ | 3630/30000 [2:18:39<16:50:13, 2.30s/it] 12%|█▏ | 3630/30000 [2:18:39<16:50:13, 2.30s/it] 12%|█▏ | 3631/30000 [2:18:41<16:47:16, 2.29s/it] 12%|█▏ | 3632/30000 [2:18:43<16:42:41, 2.28s/it] 12%|█▏ | 3633/30000 [2:18:46<16:50:37, 2.30s/it] 12%|█▏ | 3634/30000 [2:18:48<16:42:09, 2.28s/it] 12%|█▏ | 3635/30000 [2:18:50<16:39:19, 2.27s/it] 12%|█▏ | 3636/30000 [2:18:53<16:45:08, 2.29s/it] 12%|█▏ | 3637/30000 [2:18:55<16:41:42, 2.28s/it] 12%|█▏ | 3638/30000 [2:18:57<16:47:15, 2.29s/it] 12%|█▏ | 3639/30000 [2:19:00<17:07:38, 2.34s/it] 12%|█▏ | 3640/30000 [2:19:02<17:02:48, 2.33s/it] 12%|█▏ | 3640/30000 [2:19:02<17:02:48, 2.33s/it] 12%|█▏ | 3641/30000 [2:19:04<16:57:31, 2.32s/it] 12%|█▏ | 3642/30000 [2:19:07<16:59:47, 2.32s/it] 12%|█▏ | 3643/30000 [2:19:09<16:55:24, 2.31s/it] 12%|█▏ | 3644/30000 [2:19:11<16:53:37, 2.31s/it] 12%|█▏ | 3645/30000 [2:19:13<16:51:46, 2.30s/it] 12%|█▏ | 3646/30000 [2:19:16<16:56:59, 2.32s/it] 12%|█▏ | 3647/30000 [2:19:18<16:55:40, 2.31s/it] 12%|█▏ | 3648/30000 [2:19:20<16:42:58, 2.28s/it] 12%|█▏ | 3649/30000 [2:19:23<16:43:56, 2.29s/it] 12%|█▏ | 3650/30000 [2:19:25<16:45:05, 2.29s/it] 12%|█▏ | 3650/30000 [2:19:25<16:45:05, 2.29s/it] 12%|█▏ | 3651/30000 [2:19:27<16:45:48, 2.29s/it] 12%|█▏ | 3652/30000 [2:19:30<16:50:36, 2.30s/it] 12%|█▏ | 3653/30000 [2:19:32<16:53:11, 2.31s/it] 12%|█▏ | 3654/30000 [2:19:34<16:51:06, 2.30s/it] 12%|█▏ | 3655/30000 [2:19:37<17:05:21, 2.34s/it] 12%|█▏ | 3656/30000 [2:19:39<17:02:05, 2.33s/it] 12%|█▏ | 3657/30000 [2:19:41<17:07:43, 2.34s/it] 12%|█▏ | 3658/30000 [2:19:43<16:57:41, 2.32s/it] 12%|█▏ | 3659/30000 [2:19:46<17:00:02, 2.32s/it] 12%|█▏ | 3660/30000 [2:19:48<16:51:26, 2.30s/it] 12%|█▏ | 3660/30000 [2:19:48<16:51:26, 2.30s/it] 12%|█▏ | 3661/30000 [2:19:50<16:50:52, 2.30s/it] 12%|█▏ | 3662/30000 [2:19:53<16:56:37, 2.32s/it] 12%|█▏ | 3663/30000 [2:19:55<16:56:29, 2.32s/it] 12%|█▏ | 3664/30000 [2:19:57<16:47:12, 2.29s/it] 12%|█▏ | 3665/30000 [2:20:00<16:50:08, 2.30s/it] 12%|█▏ | 3666/30000 [2:20:02<16:33:35, 2.26s/it] 12%|█▏ | 3667/30000 [2:20:04<16:06:45, 2.20s/it] 12%|█▏ | 3668/30000 [2:20:06<16:20:34, 2.23s/it] 12%|█▏ | 3669/30000 [2:20:08<16:23:45, 2.24s/it] 12%|█▏ | 3670/30000 [2:20:11<16:21:22, 2.24s/it] 12%|█▏ | 3670/30000 [2:20:11<16:21:22, 2.24s/it] 12%|█▏ | 3671/30000 [2:20:13<16:38:07, 2.27s/it] 12%|█▏ | 3672/30000 [2:20:15<16:39:47, 2.28s/it] 12%|█▏ | 3673/30000 [2:20:18<16:33:17, 2.26s/it] 12%|█▏ | 3674/30000 [2:20:20<16:42:18, 2.28s/it] 12%|█▏ | 3675/30000 [2:20:22<17:04:56, 2.34s/it] 12%|█▏ | 3676/30000 [2:20:25<16:57:08, 2.32s/it] 12%|█▏ | 3677/30000 [2:20:27<16:54:24, 2.31s/it] 12%|█▏ | 3678/30000 [2:20:29<17:01:05, 2.33s/it] 12%|█▏ | 3679/30000 [2:20:32<16:53:32, 2.31s/it] 12%|█▏ | 3680/30000 [2:20:34<16:47:47, 2.30s/it] 12%|█▏ | 3680/30000 [2:20:34<16:47:47, 2.30s/it] 12%|█▏ | 3681/30000 [2:20:36<16:59:25, 2.32s/it] 12%|█▏ | 3682/30000 [2:20:39<17:00:47, 2.33s/it] 12%|█▏ | 3683/30000 [2:20:41<16:46:51, 2.30s/it] 12%|█▏ | 3684/30000 [2:20:43<16:50:00, 2.30s/it] 12%|█▏ | 3685/30000 [2:20:45<16:49:09, 2.30s/it] 12%|█▏ | 3686/30000 [2:20:48<16:42:42, 2.29s/it] 12%|█▏ | 3687/30000 [2:20:50<16:49:53, 2.30s/it] 12%|█▏ | 3688/30000 [2:20:52<16:49:25, 2.30s/it] 12%|█▏ | 3689/30000 [2:20:54<16:44:13, 2.29s/it] 12%|█▏ | 3690/30000 [2:20:57<16:47:47, 2.30s/it] 12%|█▏ | 3690/30000 [2:20:57<16:47:47, 2.30s/it] 12%|█▏ | 3691/30000 [2:20:59<16:38:45, 2.28s/it] 12%|█▏ | 3692/30000 [2:21:01<16:32:58, 2.26s/it] 12%|█▏ | 3693/30000 [2:21:04<16:37:48, 2.28s/it] 12%|█▏ | 3694/30000 [2:21:06<16:36:12, 2.27s/it] 12%|█▏ | 3695/30000 [2:21:08<16:36:32, 2.27s/it] 12%|█▏ | 3696/30000 [2:21:10<16:37:08, 2.27s/it] 12%|█▏ | 3697/30000 [2:21:13<16:45:08, 2.29s/it] 12%|█▏ | 3698/30000 [2:21:15<16:42:10, 2.29s/it] 12%|█▏ | 3699/30000 [2:21:17<16:28:38, 2.26s/it] 12%|█▏ | 3700/30000 [2:21:20<16:38:53, 2.28s/it] 12%|█▏ | 3700/30000 [2:21:20<16:38:53, 2.28s/it] 12%|█▏ | 3701/30000 [2:21:22<16:35:07, 2.27s/it] 12%|█▏ | 3702/30000 [2:21:24<16:54:33, 2.31s/it] 12%|█▏ | 3703/30000 [2:21:26<16:49:34, 2.30s/it] 12%|█▏ | 3704/30000 [2:21:29<16:49:22, 2.30s/it] 12%|█▏ | 3705/30000 [2:21:31<16:44:45, 2.29s/it] 12%|█▏ | 3706/30000 [2:21:33<16:29:44, 2.26s/it] 12%|█▏ | 3707/30000 [2:21:36<16:38:45, 2.28s/it] 12%|█▏ | 3708/30000 [2:21:38<16:39:35, 2.28s/it] 12%|█▏ | 3709/30000 [2:21:40<16:42:16, 2.29s/it] 12%|█▏ | 3710/30000 [2:21:43<16:58:29, 2.32s/it] 12%|█▏ | 3710/30000 [2:21:43<16:58:29, 2.32s/it] 12%|█▏ | 3711/30000 [2:21:45<16:58:16, 2.32s/it] 12%|█▏ | 3712/30000 [2:21:47<16:48:44, 2.30s/it] 12%|█▏ | 3713/30000 [2:21:49<16:45:27, 2.29s/it] 12%|█▏ | 3714/30000 [2:21:52<16:46:08, 2.30s/it] 12%|█▏ | 3715/30000 [2:21:54<16:41:52, 2.29s/it] 12%|█▏ | 3716/30000 [2:21:56<16:36:44, 2.28s/it] 12%|█▏ | 3717/30000 [2:21:59<16:56:30, 2.32s/it] 12%|█▏ | 3718/30000 [2:22:01<16:57:34, 2.32s/it] 12%|█▏ | 3719/30000 [2:22:03<16:43:21, 2.29s/it] 12%|█▏ | 3720/30000 [2:22:05<16:39:47, 2.28s/it] 12%|█▏ | 3720/30000 [2:22:05<16:39:47, 2.28s/it] 12%|█▏ | 3721/30000 [2:22:08<16:44:35, 2.29s/it] 12%|█▏ | 3722/30000 [2:22:10<16:38:27, 2.28s/it] 12%|█▏ | 3723/30000 [2:22:12<16:49:59, 2.31s/it] 12%|█▏ | 3724/30000 [2:22:15<17:02:08, 2.33s/it] 12%|█▏ | 3725/30000 [2:22:17<16:58:09, 2.33s/it] 12%|█▏ | 3726/30000 [2:22:19<16:49:50, 2.31s/it] 12%|█▏ | 3727/30000 [2:22:22<16:41:15, 2.29s/it] 12%|█▏ | 3728/30000 [2:22:24<16:47:54, 2.30s/it] 12%|█▏ | 3729/30000 [2:22:26<16:44:16, 2.29s/it] 12%|█▏ | 3730/30000 [2:22:28<16:27:51, 2.26s/it] 12%|█▏ | 3730/30000 [2:22:28<16:27:51, 2.26s/it] 12%|█▏ | 3731/30000 [2:22:31<16:42:29, 2.29s/it] 12%|█▏ | 3732/30000 [2:22:33<16:45:08, 2.30s/it] 12%|█▏ | 3733/30000 [2:22:35<16:28:30, 2.26s/it] 12%|█▏ | 3734/30000 [2:22:38<16:37:21, 2.28s/it] 12%|█▏ | 3735/30000 [2:22:40<16:41:29, 2.29s/it] 12%|█▏ | 3736/30000 [2:22:42<16:39:40, 2.28s/it] 12%|█▏ | 3737/30000 [2:22:44<16:36:48, 2.28s/it] 12%|█▏ | 3738/30000 [2:22:47<16:33:18, 2.27s/it] 12%|█▏ | 3739/30000 [2:22:49<16:46:49, 2.30s/it] 12%|█▏ | 3740/30000 [2:22:51<16:42:46, 2.29s/it] 12%|█▏ | 3740/30000 [2:22:51<16:42:46, 2.29s/it] 12%|█▏ | 3741/30000 [2:22:54<16:48:51, 2.31s/it] 12%|█▏ | 3742/30000 [2:22:56<16:50:38, 2.31s/it] 12%|█▏ | 3743/30000 [2:22:58<16:44:02, 2.29s/it] 12%|█▏ | 3744/30000 [2:23:01<16:49:40, 2.31s/it] 12%|█▏ | 3745/30000 [2:23:03<16:57:39, 2.33s/it] 12%|█▏ | 3746/30000 [2:23:05<16:47:43, 2.30s/it] 12%|█▏ | 3747/30000 [2:23:07<16:42:33, 2.29s/it] 12%|█▏ | 3748/30000 [2:23:10<16:50:53, 2.31s/it] 12%|█▏ | 3749/30000 [2:23:12<16:49:54, 2.31s/it] 12%|█▎ | 3750/30000 [2:23:14<16:43:19, 2.29s/it] 12%|█▎ | 3750/30000 [2:23:14<16:43:19, 2.29s/it] 13%|█▎ | 3751/30000 [2:23:17<16:40:15, 2.29s/it] 13%|█▎ | 3752/30000 [2:23:19<16:45:34, 2.30s/it] 13%|█▎ | 3753/30000 [2:23:21<16:34:23, 2.27s/it] 13%|█▎ | 3754/30000 [2:23:23<16:33:50, 2.27s/it] 13%|█▎ | 3755/30000 [2:23:26<16:43:50, 2.29s/it] 13%|█▎ | 3756/30000 [2:23:28<16:48:13, 2.31s/it] 13%|█▎ | 3757/30000 [2:23:30<16:45:50, 2.30s/it] 13%|█▎ | 3758/30000 [2:23:33<16:40:04, 2.29s/it] 13%|█▎ | 3759/30000 [2:23:35<16:46:08, 2.30s/it] 13%|█▎ | 3760/30000 [2:23:37<16:42:25, 2.29s/it] 13%|█▎ | 3760/30000 [2:23:37<16:42:25, 2.29s/it] 13%|█▎ | 3761/30000 [2:23:40<16:39:19, 2.29s/it] 13%|█▎ | 3762/30000 [2:23:42<16:46:45, 2.30s/it] 13%|█▎ | 3763/30000 [2:23:44<16:56:52, 2.33s/it] 13%|█▎ | 3764/30000 [2:23:46<16:42:19, 2.29s/it] 13%|█▎ | 3765/30000 [2:23:49<16:45:17, 2.30s/it] 13%|█▎ | 3766/30000 [2:23:51<16:53:30, 2.32s/it] 13%|█▎ | 3767/30000 [2:23:53<16:47:23, 2.30s/it] 13%|█▎ | 3768/30000 [2:23:56<16:43:52, 2.30s/it] 13%|█▎ | 3769/30000 [2:23:58<16:50:34, 2.31s/it] 13%|█▎ | 3770/30000 [2:24:00<16:50:02, 2.31s/it] 13%|█▎ | 3770/30000 [2:24:00<16:50:02, 2.31s/it] 13%|█▎ | 3771/30000 [2:24:03<16:48:53, 2.31s/it] 13%|█▎ | 3772/30000 [2:24:05<16:49:07, 2.31s/it] 13%|█▎ | 3773/30000 [2:24:07<16:42:46, 2.29s/it] 13%|█▎ | 3774/30000 [2:24:09<16:26:44, 2.26s/it] 13%|█▎ | 3775/30000 [2:24:12<16:40:05, 2.29s/it] 13%|█▎ | 3776/30000 [2:24:14<16:42:57, 2.29s/it] 13%|█▎ | 3777/30000 [2:24:16<16:37:48, 2.28s/it] 13%|█▎ | 3778/30000 [2:24:19<16:40:48, 2.29s/it] 13%|█▎ | 3779/30000 [2:24:21<16:42:15, 2.29s/it] 13%|█▎ | 3780/30000 [2:24:23<16:37:24, 2.28s/it] 13%|█▎ | 3780/30000 [2:24:23<16:37:24, 2.28s/it] 13%|█▎ | 3781/30000 [2:24:25<16:37:51, 2.28s/it] 13%|█▎ | 3782/30000 [2:24:28<16:22:47, 2.25s/it] 13%|█▎ | 3783/30000 [2:24:30<16:26:45, 2.26s/it] 13%|█▎ | 3784/30000 [2:24:32<16:34:34, 2.28s/it] 13%|█▎ | 3785/30000 [2:24:35<16:40:05, 2.29s/it] 13%|█▎ | 3786/30000 [2:24:37<16:44:17, 2.30s/it] 13%|█▎ | 3787/30000 [2:24:39<16:46:50, 2.30s/it] 13%|█▎ | 3788/30000 [2:24:41<16:41:51, 2.29s/it] 13%|█▎ | 3789/30000 [2:24:44<16:36:04, 2.28s/it] 13%|█▎ | 3790/30000 [2:24:46<16:34:26, 2.28s/it] +{'loss': 0.019, 'grad_norm': 0.14581292867660522, 'learning_rate': 2.9675984756826545e-05, 'flow_matching_loss': 0.01896311677992344, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.99} +{'loss': 0.0182, 'grad_norm': 0.10426869988441467, 'learning_rate': 2.9672557697571273e-05, 'flow_matching_loss': 0.01821328653022647, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.99} +{'loss': 0.021, 'grad_norm': 0.11001290380954742, 'learning_rate': 2.966911280977486e-05, 'flow_matching_loss': 0.021029244549572466, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 0.99} +{'loss': 0.018, 'grad_norm': 0.09236831963062286, 'learning_rate': 2.966565009762318e-05, 'flow_matching_loss': 0.018004207126796245, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.0} +{'loss': 0.0181, 'grad_norm': 0.09382433444261551, 'learning_rate': 2.966216956532375e-05, 'flow_matching_loss': 0.018140034284442663, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.0} +{'loss': 0.0185, 'grad_norm': 0.10031302273273468, 'learning_rate': 2.9658671217105744e-05, 'flow_matching_loss': 0.01851024329662323, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.0} +{'loss': 0.018, 'grad_norm': 0.12977974116802216, 'learning_rate': 2.965515505721999e-05, 'flow_matching_loss': 0.018028015177696942, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.0} +{'loss': 0.0198, 'grad_norm': 0.08032368868589401, 'learning_rate': 2.9651621089938948e-05, 'flow_matching_loss': 0.019814448338001965, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.01} +{'loss': 0.0167, 'grad_norm': 0.08694867789745331, 'learning_rate': 2.9648069319556725e-05, 'flow_matching_loss': 0.016713718045502902, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.01} +{'loss': 0.018, 'grad_norm': 0.0849858820438385, 'learning_rate': 2.9644499750389057e-05, 'flow_matching_loss': 0.018012770265340806, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.01} +{'loss': 0.0189, 'grad_norm': 0.07064961642026901, 'learning_rate': 2.9640912386773305e-05, 'flow_matching_loss': 0.018896611500531436, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.02} +{'loss': 0.0171, 'grad_norm': 0.08646975457668304, 'learning_rate': 2.9637307233068455e-05, 'flow_matching_loss': 0.017089961376041175, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.02} +{'loss': 0.0177, 'grad_norm': 0.09774674475193024, 'learning_rate': 2.9633684293655108e-05, 'flow_matching_loss': 0.017704608850181103, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.02} +{'loss': 0.0173, 'grad_norm': 0.07876400649547577, 'learning_rate': 2.963004357293548e-05, 'flow_matching_loss': 0.017291402257978917, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.02} +{'loss': 0.0165, 'grad_norm': 0.07524371892213821, 'learning_rate': 2.962638507533339e-05, 'flow_matching_loss': 0.016510333213955165, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.03} +{'loss': 0.0178, 'grad_norm': 0.07280609756708145, 'learning_rate': 2.962270880529425e-05, 'flow_matching_loss': 0.01775587061420083, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.03} +{'loss': 0.0176, 'grad_norm': 0.08158526569604874, 'learning_rate': 2.9619014767285084e-05, 'flow_matching_loss': 0.017568336520344018, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.03} +{'loss': 0.0176, 'grad_norm': 0.07003781199455261, 'learning_rate': 2.9615302965794488e-05, 'flow_matching_loss': 0.017553476337343454, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.04} +{'loss': 0.0182, 'grad_norm': 0.07239402830600739, 'learning_rate': 2.961157340533266e-05, 'flow_matching_loss': 0.018184498883783817, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.04} +{'loss': 0.0171, 'grad_norm': 0.10114186257123947, 'learning_rate': 2.9607826090431364e-05, 'flow_matching_loss': 0.017108201421797274, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.04} +{'loss': 0.0182, 'grad_norm': 0.09527178108692169, 'learning_rate': 2.9604061025643936e-05, 'flow_matching_loss': 0.01818251721560955, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.04} +{'loss': 0.0186, 'grad_norm': 0.08926309645175934, 'learning_rate': 2.9600278215545292e-05, 'flow_matching_loss': 0.018562502786517145, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.05} +{'loss': 0.0166, 'grad_norm': 0.08077188581228256, 'learning_rate': 2.95964776647319e-05, 'flow_matching_loss': 0.016624342650175095, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.05} +{'loss': 0.0191, 'grad_norm': 0.07033433765172958, 'learning_rate': 2.9592659377821787e-05, 'flow_matching_loss': 0.019081162102520467, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.05} +{'loss': 0.0169, 'grad_norm': 0.07533829659223557, 'learning_rate': 2.9588823359454528e-05, 'flow_matching_loss': 0.016871089302003383, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.06} +{'loss': 0.0185, 'grad_norm': 0.08127737045288086, 'learning_rate': 2.958496961429126e-05, 'flow_matching_loss': 0.018510181363672017, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.06} +{'loss': 0.0185, 'grad_norm': 0.12658695876598358, 'learning_rate': 2.9581098147014628e-05, 'flow_matching_loss': 0.018526287842541934, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.06} +{'loss': 0.0191, 'grad_norm': 0.07055851817131042, 'learning_rate': 2.9577208962328843e-05, 'flow_matching_loss': 0.01906443266198039, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.06} +{'loss': 0.0171, 'grad_norm': 0.08822812885046005, 'learning_rate': 2.9573302064959633e-05, 'flow_matching_loss': 0.01712263636291027, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.07} +{'loss': 0.0194, 'grad_norm': 0.10270032286643982, 'learning_rate': 2.9569377459654235e-05, 'flow_matching_loss': 0.01943627716973424, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.07} +{'loss': 0.0191, 'grad_norm': 0.11913636326789856, 'learning_rate': 2.9565435151181424e-05, 'flow_matching_loss': 0.019090043753385542, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.07} +{'loss': 0.0186, 'grad_norm': 0.09321325272321701, 'learning_rate': 2.9561475144331473e-05, 'flow_matching_loss': 0.018570251110941172, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.08} +{'loss': 0.0162, 'grad_norm': 0.08255013823509216, 'learning_rate': 2.9557497443916162e-05, 'flow_matching_loss': 0.01618027873337269, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.08} +{'loss': 0.0177, 'grad_norm': 0.10429845005273819, 'learning_rate': 2.9553502054768774e-05, 'flow_matching_loss': 0.017737521417438985, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.08} +{'loss': 0.0184, 'grad_norm': 0.10847841203212738, 'learning_rate': 2.954948898174409e-05, 'flow_matching_loss': 0.018365352880209684, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.09} +{'loss': 0.0185, 'grad_norm': 0.08747928589582443, 'learning_rate': 2.954545822971836e-05, 'flow_matching_loss': 0.018504612054675816, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.09} +{'loss': 0.0175, 'grad_norm': 0.09519334882497787, 'learning_rate': 2.9541409803589335e-05, 'flow_matching_loss': 0.017504666186869144, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.09} +{'loss': 0.0182, 'grad_norm': 0.0996343269944191, 'learning_rate': 2.9537343708276235e-05, 'flow_matching_loss': 0.018213493190705775, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.09} +{'loss': 0.0179, 'grad_norm': 0.08563773334026337, 'learning_rate': 2.953325994871975e-05, 'flow_matching_loss': 0.017857956886291503, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.1} +{'loss': 0.0189, 'grad_norm': 0.1031501367688179, 'learning_rate': 2.9529158529882023e-05, 'flow_matching_loss': 0.018901527021080257, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.1} +{'loss': 0.016, 'grad_norm': 0.1090025007724762, 'learning_rate': 2.9525039456746677e-05, 'flow_matching_loss': 0.015986056718975305, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.1} 13%|█▎ | 3790/30000 [2:24:46<16:34:26, 2.28s/it] 13%|█▎ | 3791/30000 [2:24:48<16:32:10, 2.27s/it] 13%|█▎ | 3792/30000 [2:24:50<16:29:44, 2.27s/it] 13%|█▎ | 3793/30000 [2:24:53<16:35:47, 2.28s/it] 13%|█▎ | 3794/30000 [2:24:55<16:38:53, 2.29s/it] 13%|█▎ | 3795/30000 [2:24:57<16:41:28, 2.29s/it] 13%|█▎ | 3796/30000 [2:25:00<16:36:56, 2.28s/it] 13%|█▎ | 3797/30000 [2:25:02<16:42:22, 2.30s/it] 13%|█▎ | 3798/30000 [2:25:04<16:40:43, 2.29s/it] 13%|█▎ | 3799/30000 [2:25:07<16:36:09, 2.28s/it] 13%|█▎ | 3800/30000 [2:25:09<16:44:37, 2.30s/it] 13%|█▎ | 3800/30000 [2:25:09<16:44:37, 2.30s/it] 13%|█▎ | 3801/30000 [2:25:11<16:43:49, 2.30s/it] 13%|█▎ | 3802/30000 [2:25:13<16:34:50, 2.28s/it] 13%|█▎ | 3803/30000 [2:25:16<16:43:05, 2.30s/it] 13%|█▎ | 3804/30000 [2:25:18<16:41:27, 2.29s/it] 13%|█▎ | 3805/30000 [2:25:20<16:38:06, 2.29s/it] 13%|█▎ | 3806/30000 [2:25:23<16:41:51, 2.29s/it] 13%|█▎ | 3807/30000 [2:25:25<16:38:46, 2.29s/it] 13%|█▎ | 3808/30000 [2:25:27<16:32:38, 2.27s/it] 13%|█▎ | 3809/30000 [2:25:29<16:32:19, 2.27s/it] 13%|█▎ | 3810/30000 [2:25:32<16:40:48, 2.29s/it] 13%|█▎ | 3810/30000 [2:25:32<16:40:48, 2.29s/it] 13%|█▎ | 3811/30000 [2:25:34<16:40:49, 2.29s/it] 13%|█▎ | 3812/30000 [2:25:36<16:18:57, 2.24s/it] 13%|█▎ | 3813/30000 [2:25:38<16:29:50, 2.27s/it] 13%|█▎ | 3814/30000 [2:25:41<16:26:30, 2.26s/it] 13%|█▎ | 3815/30000 [2:25:43<16:26:56, 2.26s/it] 13%|█▎ | 3816/30000 [2:25:45<16:36:37, 2.28s/it] 13%|█▎ | 3817/30000 [2:25:48<16:39:40, 2.29s/it] 13%|█▎ | 3818/30000 [2:25:50<16:33:08, 2.28s/it] 13%|█▎ | 3819/30000 [2:25:52<16:43:24, 2.30s/it] 13%|█▎ | 3820/30000 [2:25:55<16:44:38, 2.30s/it] 13%|█▎ | 3820/30000 [2:25:55<16:44:38, 2.30s/it] 13%|█▎ | 3821/30000 [2:25:57<16:34:04, 2.28s/it] 13%|█▎ | 3822/30000 [2:25:59<16:38:19, 2.29s/it] 13%|█▎ | 3823/30000 [2:26:01<16:45:07, 2.30s/it] 13%|█▎ | 3824/30000 [2:26:04<16:42:06, 2.30s/it] 13%|█▎ | 3825/30000 [2:26:06<16:37:50, 2.29s/it] 13%|█▎ | 3826/30000 [2:26:08<16:38:22, 2.29s/it] 13%|█▎ | 3827/30000 [2:26:11<16:46:37, 2.31s/it] 13%|█▎ | 3828/30000 [2:26:13<16:46:52, 2.31s/it] 13%|█▎ | 3829/30000 [2:26:15<16:33:30, 2.28s/it] 13%|█▎ | 3830/30000 [2:26:17<16:40:29, 2.29s/it] 13%|█▎ | 3830/30000 [2:26:17<16:40:29, 2.29s/it] 13%|█▎ | 3831/30000 [2:26:20<16:41:48, 2.30s/it] 13%|█▎ | 3832/30000 [2:26:22<16:37:27, 2.29s/it] 13%|█▎ | 3833/30000 [2:26:24<16:35:30, 2.28s/it] 13%|█▎ | 3834/30000 [2:26:27<16:42:29, 2.30s/it] 13%|█▎ | 3835/30000 [2:26:29<16:37:57, 2.29s/it] 13%|█▎ | 3836/30000 [2:26:31<16:28:21, 2.27s/it] 13%|█▎ | 3837/30000 [2:26:33<16:35:53, 2.28s/it] 13%|█▎ | 3838/30000 [2:26:36<16:38:25, 2.29s/it] 13%|█▎ | 3839/30000 [2:26:38<16:39:48, 2.29s/it] 13%|█▎ | 3840/30000 [2:26:40<16:39:58, 2.29s/it] 13%|█▎ | 3840/30000 [2:26:40<16:39:58, 2.29s/it] 13%|█▎ | 3841/30000 [2:26:43<16:47:40, 2.31s/it] 13%|█▎ | 3842/30000 [2:26:45<16:42:51, 2.30s/it] 13%|█▎ | 3843/30000 [2:26:47<16:38:32, 2.29s/it] 13%|█▎ | 3844/30000 [2:26:49<16:38:16, 2.29s/it] 13%|█▎ | 3845/30000 [2:26:52<17:06:28, 2.35s/it] 13%|█▎ | 3846/30000 [2:26:54<16:57:56, 2.34s/it] 13%|█▎ | 3847/30000 [2:26:57<16:46:49, 2.31s/it] 13%|█▎ | 3848/30000 [2:26:59<16:38:55, 2.29s/it] 13%|█▎ | 3849/30000 [2:27:01<16:40:37, 2.30s/it] 13%|█▎ | 3850/30000 [2:27:03<16:33:42, 2.28s/it] 13%|█▎ | 3850/30000 [2:27:03<16:33:42, 2.28s/it] 13%|█▎ | 3851/30000 [2:27:06<16:39:19, 2.29s/it] 13%|█▎ | 3852/30000 [2:27:08<16:44:18, 2.30s/it] 13%|█▎ | 3853/30000 [2:27:10<16:50:23, 2.32s/it] 13%|█▎ | 3854/30000 [2:27:13<16:37:53, 2.29s/it] 13%|█▎ | 3855/30000 [2:27:15<16:42:12, 2.30s/it] 13%|█▎ | 3856/30000 [2:27:17<16:47:54, 2.31s/it] 13%|█▎ | 3857/30000 [2:27:20<16:53:27, 2.33s/it] 13%|█▎ | 3858/30000 [2:27:22<16:45:12, 2.31s/it] 13%|█▎ | 3859/30000 [2:27:24<16:39:27, 2.29s/it] 13%|█▎ | 3860/30000 [2:27:26<16:45:28, 2.31s/it] 13%|█▎ | 3860/30000 [2:27:26<16:45:28, 2.31s/it] 13%|█▎ | 3861/30000 [2:27:29<16:38:29, 2.29s/it] 13%|█▎ | 3862/30000 [2:27:31<16:28:52, 2.27s/it] 13%|█▎ | 3863/30000 [2:27:33<16:28:18, 2.27s/it] 13%|█▎ | 3864/30000 [2:27:35<16:32:37, 2.28s/it] 13%|█▎ | 3865/30000 [2:27:38<16:21:29, 2.25s/it] 13%|█▎ | 3866/30000 [2:27:40<16:33:26, 2.28s/it] 13%|█▎ | 3867/30000 [2:27:42<16:41:26, 2.30s/it] 13%|█▎ | 3868/30000 [2:27:45<16:37:57, 2.29s/it] 13%|█▎ | 3869/30000 [2:27:47<16:40:24, 2.30s/it] 13%|█▎ | 3870/30000 [2:27:49<16:41:45, 2.30s/it] 13%|█▎ | 3870/30000 [2:27:49<16:41:45, 2.30s/it] 13%|█▎ | 3871/30000 [2:27:52<16:38:03, 2.29s/it] 13%|█▎ | 3872/30000 [2:27:54<16:34:56, 2.28s/it] 13%|█▎ | 3873/30000 [2:27:56<16:38:35, 2.29s/it] 13%|█▎ | 3874/30000 [2:27:58<16:36:34, 2.29s/it] 13%|█▎ | 3875/30000 [2:28:01<16:33:59, 2.28s/it] 13%|█▎ | 3876/30000 [2:28:03<16:54:09, 2.33s/it] 13%|█▎ | 3877/30000 [2:28:05<16:51:49, 2.32s/it] 13%|█▎ | 3878/30000 [2:28:08<16:42:06, 2.30s/it] 13%|█▎ | 3879/30000 [2:28:10<16:44:22, 2.31s/it] 13%|█▎ | 3880/30000 [2:28:12<16:47:26, 2.31s/it] 13%|█▎ | 3880/30000 [2:28:12<16:47:26, 2.31s/it] 13%|█▎ | 3881/30000 [2:28:15<16:44:09, 2.31s/it] 13%|█▎ | 3882/30000 [2:28:17<16:43:06, 2.30s/it] 13%|█▎ | 3883/30000 [2:28:19<16:34:17, 2.28s/it] 13%|█▎ | 3884/30000 [2:28:21<16:26:30, 2.27s/it] 13%|█▎ | 3885/30000 [2:28:24<16:37:56, 2.29s/it] 13%|█▎ | 3886/30000 [2:28:26<16:35:54, 2.29s/it] 13%|█▎ | 3887/30000 [2:28:28<16:29:59, 2.27s/it] 13%|█▎ | 3888/30000 [2:28:31<16:39:29, 2.30s/it] 13%|█▎ | 3889/30000 [2:28:33<16:45:59, 2.31s/it] 13%|█▎ | 3890/30000 [2:28:35<16:41:16, 2.30s/it] 13%|█▎ | 3890/30000 [2:28:35<16:41:16, 2.30s/it] 13%|█▎ | 3891/30000 [2:28:37<16:36:44, 2.29s/it] 13%|█▎ | 3892/30000 [2:28:40<16:47:38, 2.32s/it] 13%|█▎ | 3893/30000 [2:28:42<16:43:26, 2.31s/it] 13%|█▎ | 3894/30000 [2:28:44<16:48:53, 2.32s/it] 13%|█▎ | 3895/30000 [2:28:47<16:54:48, 2.33s/it] 13%|█▎ | 3896/30000 [2:28:49<16:53:31, 2.33s/it] 13%|█▎ | 3897/30000 [2:28:51<16:47:30, 2.32s/it] 13%|█▎ | 3898/30000 [2:28:54<16:47:29, 2.32s/it] 13%|█▎ | 3899/30000 [2:28:56<16:48:02, 2.32s/it] 13%|█▎ | 3900/30000 [2:28:58<16:42:19, 2.30s/it] 13%|█▎ | 3900/30000 [2:28:58<16:42:19, 2.30s/it] 13%|█▎ | 3901/30000 [2:29:01<16:46:23, 2.31s/it] 13%|█▎ | 3902/30000 [2:29:03<16:50:47, 2.32s/it] 13%|█▎ | 3903/30000 [2:29:05<16:53:34, 2.33s/it] 13%|█▎ | 3904/30000 [2:29:08<16:47:36, 2.32s/it] 13%|█▎ | 3905/30000 [2:29:10<16:52:11, 2.33s/it] 13%|█▎ | 3906/30000 [2:29:12<16:47:08, 2.32s/it] 13%|█▎ | 3907/30000 [2:29:15<16:43:51, 2.31s/it] 13%|█▎ | 3908/30000 [2:29:17<16:38:29, 2.30s/it] 13%|█▎ | 3909/30000 [2:29:19<16:39:59, 2.30s/it] 13%|█▎ | 3910/30000 [2:29:21<16:34:42, 2.29s/it] 13%|█▎ | 3910/30000 [2:29:21<16:34:42, 2.29s/it] 13%|█▎ | 3911/30000 [2:29:24<16:33:48, 2.29s/it] 13%|█▎ | 3912/30000 [2:29:26<16:36:36, 2.29s/it] 13%|█▎ | 3913/30000 [2:29:28<16:32:24, 2.28s/it] 13%|█▎ | 3914/30000 [2:29:31<16:32:24, 2.28s/it] 13%|█▎ | 3915/30000 [2:29:33<16:41:15, 2.30s/it] 13%|█▎ | 3916/30000 [2:29:35<16:58:51, 2.34s/it] 13%|█▎ | 3917/30000 [2:29:38<16:48:54, 2.32s/it] 13%|█▎ | 3918/30000 [2:29:40<16:54:21, 2.33s/it] 13%|█▎ | 3919/30000 [2:29:42<16:46:05, 2.31s/it] 13%|█▎ | 3920/30000 [2:29:45<16:36:24, 2.29s/it] 13%|█▎ | 3920/30000 [2:29:45<16:36:24, 2.29s/it] 13%|█▎ | 3921/30000 [2:29:47<16:46:41, 2.32s/it] 13%|█▎ | 3922/30000 [2:29:49<16:41:48, 2.30s/it] 13%|█▎ | 3923/30000 [2:29:52<16:46:25, 2.32s/it] 13%|█▎ | 3924/30000 [2:29:54<16:49:39, 2.32s/it] 13%|█▎ | 3925/30000 [2:29:56<16:48:38, 2.32s/it] 13%|█▎ | 3926/30000 [2:29:58<16:41:06, 2.30s/it] 13%|█▎ | 3927/30000 [2:30:01<16:46:40, 2.32s/it] 13%|█▎ | 3928/30000 [2:30:03<16:45:07, 2.31s/it] 13%|█▎ | 3929/30000 [2:30:05<16:38:33, 2.30s/it] 13%|█▎ | 3930/30000 [2:30:08<16:38:33, 2.30s/it] 13%|█▎ | 3930/30000 [2:30:08<16:38:33, 2.30s/it] 13%|█▎ | 3931/30000 [2:30:10<16:41:24, 2.30s/it] 13%|█▎ | 3932/30000 [2:30:12<16:34:58, 2.29s/it] 13%|█▎ | 3933/30000 [2:30:15<16:39:56, 2.30s/it] 13%|█▎ | 3934/30000 [2:30:17<16:46:47, 2.32s/it] 13%|█▎ | 3935/30000 [2:30:19<16:27:52, 2.27s/it] 13%|█▎ | 3936/30000 [2:30:21<16:37:43, 2.30s/it] 13%|█▎ | 3937/30000 [2:30:24<16:35:09, 2.29s/it] 13%|█▎ | 3938/30000 [2:30:26<16:31:56, 2.28s/it] 13%|█▎ | 3939/30000 [2:30:28<16:39:25, 2.30s/it] 13%|█▎ | 3940/30000 [2:30:31<16:42:20, 2.31s/it] 13%|█▎ | 3940/30000 [2:30:31<16:42:20, 2.31s/it] 13%|█▎ | 3941/30000 [2:30:33<16:44:46, 2.31s/it] 13%|█▎ | 3942/30000 [2:30:35<16:40:53, 2.30s/it] 13%|█▎ | 3943/30000 [2:30:38<16:42:27, 2.31s/it] 13%|█▎ | 3944/30000 [2:30:40<16:36:06, 2.29s/it] 13%|█▎ | 3945/30000 [2:30:42<16:31:37, 2.28s/it] 13%|█▎ | 3946/30000 [2:30:45<16:54:18, 2.34s/it] 13%|█▎ | 3947/30000 [2:30:47<16:43:04, 2.31s/it] 13%|█▎ | 3948/30000 [2:30:49<16:38:59, 2.30s/it] 13%|█▎ | 3949/30000 [2:30:52<16:59:39, 2.35s/it] 13%|█▎ | 3950/30000 [2:30:54<16:54:40, 2.34s/it] 13%|█▎ | 3950/30000 [2:30:54<16:54:40, 2.34s/it] 13%|█▎ | 3951/30000 [2:30:56<16:49:23, 2.32s/it] 13%|█▎ | 3952/30000 [2:30:58<16:50:04, 2.33s/it] 13%|█▎ | 3953/30000 [2:31:01<16:45:23, 2.32s/it] 13%|█▎ | 3954/30000 [2:31:03<16:53:09, 2.33s/it] 13%|█▎ | 3955/30000 [2:31:05<16:43:36, 2.31s/it] 13%|█▎ | 3956/30000 [2:31:08<16:49:58, 2.33s/it] 13%|█▎ | 3957/30000 [2:31:10<16:42:59, 2.31s/it] 13%|█▎ | 3958/30000 [2:31:12<16:36:20, 2.30s/it] 13%|█▎ | 3959/30000 [2:31:15<16:40:52, 2.31s/it] 13%|█▎ | 3960/30000 [2:31:17<16:37:18, 2.30s/it] 13%|█▎ | 3960/30000 [2:31:17<16:37:18, 2.30s/it] 13%|█▎ | 3961/30000 [2:31:19<16:34:11, 2.29s/it] 13%|█▎ | 3962/30000 [2:31:22<16:53:42, 2.34s/it] 13%|█▎ | 3963/30000 [2:31:24<16:51:35, 2.33s/it] 13%|█▎ | 3964/30000 [2:31:26<16:39:04, 2.30s/it] 13%|█▎ | 3965/30000 [2:31:28<16:42:13, 2.31s/it] 13%|█▎ | 3966/30000 [2:31:31<16:39:01, 2.30s/it] 13%|█▎ | 3967/30000 [2:31:33<16:37:53, 2.30s/it] 13%|█▎ | 3968/30000 [2:31:36<16:59:48, 2.35s/it] 13%|█▎ | 3969/30000 [2:31:38<16:54:33, 2.34s/it] 13%|█▎ | 3970/30000 [2:31:40<16:44:34, 2.32s/it] 13%|█▎ | 3970/30000 [2:31:40<16:44:34, 2.32s/it] 13%|█▎ | 3971/30000 [2:31:42<16:42:13, 2.31s/it] 13%|█▎ | 3972/30000 [2:31:45<16:44:44, 2.32s/it] 13%|█▎ | 3973/30000 [2:31:47<16:37:45, 2.30s/it] 13%|█▎ | 3974/30000 [2:31:49<16:31:18, 2.29s/it] 13%|█▎ | 3975/30000 [2:31:52<16:38:08, 2.30s/it] 13%|█▎ | 3976/30000 [2:31:54<16:35:41, 2.30s/it] 13%|█▎ | 3977/30000 [2:31:56<16:30:27, 2.28s/it] 13%|█▎ | 3978/30000 [2:31:58<16:38:26, 2.30s/it] 13%|█▎ | 3979/30000 [2:32:01<16:33:27, 2.29s/it] 13%|█▎ | 3980/30000 [2:32:03<16:29:18, 2.28s/it] 13%|█▎ | 3980/30000 [2:32:03<16:29:18, 2.28s/it] 13%|█▎ | 3981/30000 [2:32:05<16:41:03, 2.31s/it] 13%|█▎ | 3982/30000 [2:32:08<16:44:40, 2.32s/it] 13%|█▎ | 3983/30000 [2:32:10<16:34:56, 2.29s/it] 13%|█▎ | 3984/30000 [2:32:12<16:55:23, 2.34s/it] 13%|█▎ | 3985/30000 [2:32:15<16:52:23, 2.33s/it] 13%|█▎ | 3986/30000 [2:32:17<16:57:36, 2.35s/it] 13%|█▎ | 3987/30000 [2:32:19<16:51:32, 2.33s/it] 13%|█▎ | 3988/30000 [2:32:22<16:57:57, 2.35s/it] 13%|█▎ | 3989/30000 [2:32:24<16:56:39, 2.35s/it] 13%|█▎ | 3990/30000 [2:32:26<16:39:21, 2.31s/it] 13%|█▎ | 3990/30000 [2:32:26<16:39:21, 2.31s/it] 13%|█▎ | 3991/30000 [2:32:29<16:49:09, 2.33s/it] 13%|█▎ | 3992/30000 [2:32:31<16:43:17, 2.31s/it] 13%|█▎ | 3993/30000 [2:32:33<16:42:31, 2.31s/it] 13%|█▎ | 3994/30000 [2:32:36<16:41:37, 2.31s/it] 13%|█▎ | 3995/30000 [2:32:38<16:31:08, 2.29s/it] 13%|█▎ | 3996/30000 [2:32:40<16:28:59, 2.28s/it] 13%|█▎ | 3997/30000 [2:32:42<16:31:01, 2.29s/it] 13%|█▎ | 3998/30000 [2:32:45<16:53:36, 2.34s/it] 13%|█▎ | 3999/30000 [2:32:47<16:51:14, 2.33s/it] 13%|█▎ | 4000/30000 [2:32:49<16:41:46, 2.31s/it] 13%|█▎ | 4000/30000 [2:32:49<16:41:46, 2.31s/it] 13%|█▎ | 4001/30000 [2:32:52<16:43:04, 2.31s/it] 13%|█▎ | 4002/30000 [2:32:54<16:42:27, 2.31s/it] 13%|█▎ | 4003/30000 [2:32:56<16:35:10, 2.30s/it] 13%|█▎ | 4004/30000 [2:32:59<16:27:14, 2.28s/it] 13%|█▎ | 4005/30000 [2:33:01<16:39:01, 2.31s/it] 13%|█▎ | 4006/30000 [2:33:03<16:36:09, 2.30s/it] 13%|█▎ | 4007/30000 [2:33:05<16:32:31, 2.29s/it] 13%|█▎ | 4008/30000 [2:33:08<16:40:34, 2.31s/it] 13%|█▎ | 4009/30000 [2:33:10<16:38:38, 2.31s/it] 13%|█▎ | 4010/30000 [2:33:12<16:40:52, 2.31s/it] 13%|█▎ | 4010/30000 [2:33:13<16:40:52, 2.31s/it] 13%|█▎ | 4011/30000 [2:33:15<16:42:46, 2.32s/it] 13%|█▎ | 4012/30000 [2:33:17<16:43:52, 2.32s/it] 13%|█▎ | 4013/30000 [2:33:19<16:37:41, 2.30s/it] 13%|█▎ | 4014/30000 [2:33:22<16:35:31, 2.30s/it] 13%|█▎ | 4015/30000 [2:33:24<16:39:18, 2.31s/it] 13%|█▎ | 4016/30000 [2:33:26<16:25:40, 2.28s/it] 13%|█▎ | 4017/30000 [2:33:28<16:25:25, 2.28s/it] 13%|█▎ | 4018/30000 [2:33:31<16:37:05, 2.30s/it] 13%|█▎ | 4019/30000 [2:33:33<16:31:00, 2.29s/it] 13%|█▎ | 4020/30000 [2:33:35<16:14:36, 2.25s/it] 13%|█▎ | 4020/30000 [2:33:35<16:14:36, 2.25s/it] 13%|█▎ | 4021/30000 [2:33:38<16:28:41, 2.28s/it] 13%|█▎ | 4022/30000 [2:33:40<16:39:11, 2.31s/it] 13%|█▎ | 4023/30000 [2:33:42<16:25:24, 2.28s/it] 13%|█▎ | 4024/30000 [2:33:45<16:36:57, 2.30s/it] 13%|█▎ | 4025/30000 [2:33:47<16:35:31, 2.30s/it] 13%|█▎ | 4026/30000 [2:33:49<16:30:44, 2.29s/it] 13%|█▎ | 4027/30000 [2:33:51<16:32:39, 2.29s/it] 13%|█▎ | 4028/30000 [2:33:54<16:35:35, 2.30s/it] 13%|█▎ | 4029/30000 [2:33:56<16:29:43, 2.29s/it] 13%|█▎ | 4030/30000 [2:33:58<16:23:09, 2.27s/it] 13%|█▎ | 4030/30000 [2:33:58<16:23:09, 2.27s/it] 13%|█▎ | 4031/30000 [2:34:01<16:33:47, 2.30s/it] 13%|█▎ | 4032/30000 [2:34:03<16:31:01, 2.29s/it] 13%|█▎ | 4033/30000 [2:34:05<16:31:35, 2.29s/it] 13%|█▎ | 4034/30000 [2:34:08<16:40:08, 2.31s/it] 13%|█▎ | 4035/30000 [2:34:10<16:42:02, 2.32s/it] 13%|█▎ | 4036/30000 [2:34:12<16:29:55, 2.29s/it] 13%|█▎ | 4037/30000 [2:34:14<16:29:44, 2.29s/it] 13%|█▎ | 4038/30000 [2:34:17<16:35:17, 2.30s/it] 13%|█▎ | 4039/30000 [2:34:19<16:33:00, 2.30s/it] 13%|█▎ | 4040/30000 [2:34:21<16:31:46, 2.29s/it] 13%|█▎ | 4040/30000 [2:34:21<16:31:46, 2.29s/it] 13%|█▎ | 4041/30000 [2:34:24<16:37:46, 2.31s/it] 13%|█▎ | 4042/30000 [2:34:26<16:36:14, 2.30s/it] 13%|█▎ | 4043/30000 [2:34:28<16:33:26, 2.30s/it] 13%|█▎ | 4044/30000 [2:34:30<16:26:36, 2.28s/it] 13%|█▎ | 4045/30000 [2:34:33<16:14:56, 2.25s/it] 13%|█▎ | 4046/30000 [2:34:35<16:14:59, 2.25s/it] 13%|█▎ | 4047/30000 [2:34:37<16:11:58, 2.25s/it] 13%|█▎ | 4048/30000 [2:34:39<16:23:41, 2.27s/it] 13%|█▎ | 4049/30000 [2:34:42<16:20:23, 2.27s/it] 14%|█▎ | 4050/30000 [2:34:44<16:09:08, 2.24s/it] 14%|█▎ | 4050/30000 [2:34:44<16:09:08, 2.24s/it] 14%|█▎ | 4051/30000 [2:34:46<16:09:53, 2.24s/it] 14%|█▎ | 4052/30000 [2:34:48<16:09:08, 2.24s/it] 14%|█▎ | 4053/30000 [2:34:51<16:16:54, 2.26s/it] 14%|█▎ | 4054/30000 [2:34:53<16:27:52, 2.28s/it] 14%|█▎ | 4055/30000 [2:34:55<16:29:09, 2.29s/it] 14%|█▎ | 4056/30000 [2:34:58<16:28:58, 2.29s/it] 14%|█▎ | 4057/30000 [2:35:00<16:29:47, 2.29s/it] 14%|█▎ | 4058/30000 [2:35:02<16:31:11, 2.29s/it] 14%|█▎ | 4059/30000 [2:35:04<16:30:50, 2.29s/it] 14%|█▎ | 4060/30000 [2:35:07<16:24:15, 2.28s/it] 14%|█▎ | 4060/30000 [2:35:07<16:24:15, 2.28s/it] 14%|█▎ | 4061/30000 [2:35:09<16:36:01, 2.30s/it] 14%|█▎ | 4062/30000 [2:35:11<16:34:13, 2.30s/it] 14%|█▎ | 4063/30000 [2:35:14<16:27:56, 2.29s/it] 14%|█▎ | 4064/30000 [2:35:16<16:20:19, 2.27s/it] 14%|█▎ | 4065/30000 [2:35:18<16:34:43, 2.30s/it] 14%|█▎ | 4066/30000 [2:35:21<16:44:04, 2.32s/it] 14%|█▎ | 4067/30000 [2:35:23<16:41:42, 2.32s/it] 14%|█▎ | 4068/30000 [2:35:25<16:32:25, 2.30s/it] 14%|█▎ | 4069/30000 [2:35:27<16:36:31, 2.31s/it] 14%|█▎ | 4070/30000 [2:35:30<16:33:34, 2.30s/it] 14%|█▎ | 4070/30000 [2:35:30<16:33:34, 2.30s/it] 14%|█▎ | 4071/30000 [2:35:32<16:28:51, 2.29s/it] 14%|█▎ | 4072/30000 [2:35:34<16:20:49, 2.27s/it] 14%|█▎ | 4073/30000 [2:35:37<16:31:01, 2.29s/it] 14%|█▎ | 4074/30000 [2:35:39<16:29:33, 2.29s/it] 14%|█▎ | 4075/30000 [2:35:41<16:37:27, 2.31s/it] 14%|█▎ | 4076/30000 [2:35:43<16:32:50, 2.30s/it] 14%|█▎ | 4077/30000 [2:35:46<16:36:17, 2.31s/it] 14%|█▎ | 4078/30000 [2:35:48<15:45:30, 2.19s/it] 14%|█▎ | 4079/30000 [2:35:49<13:42:25, 1.90s/it] 14%|█▎ | 4080/30000 [2:35:50<12:18:30, 1.71s/it] 14%|█▎ | 4080/30000 [2:35:50<12:18:30, 1.71s/it] 14%|█▎ | 4081/30000 [2:35:51<11:17:00, 1.57s/it] 14%|█▎ | 4082/30000 [2:35:53<10:28:48, 1.46s/it] 14%|█▎ | 4083/30000 [2:35:54<9:54:24, 1.38s/it] 14%|█▎ | 4084/30000 [2:35:55<9:54:52, 1.38s/it] 14%|█▎ | 4085/30000 [2:35:58<11:59:13, 1.67s/it] 14%|█▎ | 4086/30000 [2:36:00<13:22:04, 1.86s/it] 14%|█▎ | 4087/30000 [2:36:02<14:20:50, 1.99s/it] 14%|█▎ | 4088/30000 [2:36:04<14:50:15, 2.06s/it] 14%|█▎ | 4089/30000 [2:36:07<15:22:40, 2.14s/it] 14%|█▎ | 4090/30000 [2:36:09<15:59:51, 2.22s/it] 14%|█▎ | 4090/30000 [2:36:09<15:59:51, 2.22s/it] 14%|█▎ | 4091/30000 [2:36:11<16:04:45, 2.23s/it] 14%|█▎ | 4092/30000 [2:36:14<16:19:17, 2.27s/it] 14%|█▎ | 4093/30000 [2:36:16<16:25:03, 2.28s/it] 14%|█▎ | 4094/30000 [2:36:18<16:26:39, 2.29s/it] 14%|█▎ | 4095/30000 [2:36:21<16:23:03, 2.28s/it] 14%|█▎ | 4096/30000 [2:36:23<16:32:28, 2.30s/it] 14%|█▎ | 4097/30000 [2:36:25<16:25:34, 2.28s/it] 14%|█▎ | 4098/30000 [2:36:27<16:21:38, 2.27s/it] 14%|█▎ | 4099/30000 [2:36:30<16:26:35, 2.29s/it] 14%|█▎ | 4100/30000 [2:36:32<16:32:57, 2.30s/it] 14%|█▎ | 4100/30000 [2:36:32<16:32:57, 2.30s/it] 14%|█▎ | 4101/30000 [2:36:34<16:26:03, 2.28s/it] 14%|█▎ | 4102/30000 [2:36:37<16:32:20, 2.30s/it] 14%|█▎ | 4103/30000 [2:36:39<16:29:41, 2.29s/it] 14%|█▎ | 4104/30000 [2:36:41<16:36:01, 2.31s/it] 14%|█▎ | 4105/30000 [2:36:44<16:40:40, 2.32s/it] 14%|█▎ | 4106/30000 [2:36:46<16:39:29, 2.32s/it] 14%|█▎ | 4107/30000 [2:36:48<16:34:33, 2.30s/it] 14%|█▎ | 4108/30000 [2:36:50<16:30:45, 2.30s/it] 14%|█▎ | 4109/30000 [2:36:53<16:35:22, 2.31s/it] 14%|█▎ | 4110/30000 [2:36:55<16:32:36, 2.30s/it] 14%|█▎ | 4110/30000 [2:36:55<16:32:36, 2.30s/it] 14%|█▎ | 4111/30000 [2:36:57<16:30:47, 2.30s/it] 14%|█▎ | 4112/30000 [2:37:00<16:31:02, 2.30s/it] 14%|█▎ | 4113/30000 [2:37:02<16:35:33, 2.31s/it] 14%|█▎ | 4114/30000 [2:37:04<16:37:51, 2.31s/it] 14%|█▎ | 4115/30000 [2:37:07<16:37:56, 2.31s/it] 14%|█▎ | 4116/30000 [2:37:09<16:41:37, 2.32s/it] 14%|█▎ | 4117/30000 [2:37:11<16:30:51, 2.30s/it] 14%|█▎ | 4118/30000 [2:37:14<16:25:40, 2.29s/it] 14%|█▎ | 4119/30000 [2:37:16<16:21:01, 2.27s/it] 14%|█▎ | 4120/30000 [2:37:18<16:12:27, 2.25s/it] 14%|█▎ | 4120/30000 [2:37:18<16:12:27, 2.25s/it] 14%|█▎ | 4121/30000 [2:37:20<16:21:30, 2.28s/it] 14%|█▎ | 4122/30000 [2:37:23<16:23:24, 2.28s/it] 14%|█▎ | 4123/30000 [2:37:25<16:14:41, 2.26s/it] 14%|█▎ | 4124/30000 [2:37:27<16:16:44, 2.26s/it] 14%|█▍ | 4125/30000 [2:37:30<16:38:42, 2.32s/it] 14%|█▍ | 4126/30000 [2:37:32<16:30:48, 2.30s/it] 14%|█▍ | 4127/30000 [2:37:34<16:28:53, 2.29s/it] 14%|█▍ | 4128/30000 [2:37:36<16:32:30, 2.30s/it] 14%|█▍ | 4129/30000 [2:37:39<16:48:13, 2.34s/it] 14%|█▍ | 4130/30000 [2:37:41<16:40:42, 2.32s/it] 14%|█▍ | 4130/30000 [2:37:41<16:40:42, 2.32s/it] 14%|█▍ | 4131/30000 [2:37:43<16:39:51, 2.32s/it] 14%|█▍ | 4132/30000 [2:37:46<16:26:39, 2.29s/it] 14%|█▍ | 4133/30000 [2:37:48<16:15:22, 2.26s/it] 14%|█▍ | 4134/30000 [2:37:50<16:23:04, 2.28s/it] 14%|█▍ | 4135/30000 [2:37:52<16:27:22, 2.29s/it] 14%|█▍ | 4136/30000 [2:37:55<16:40:19, 2.32s/it] 14%|█▍ | 4137/30000 [2:37:57<16:39:21, 2.32s/it] 14%|█▍ | 4138/30000 [2:37:59<16:38:14, 2.32s/it] 14%|█▍ | 4139/30000 [2:38:02<16:32:55, 2.30s/it] 14%|█▍ | 4140/30000 [2:38:04<16:31:34, 2.30s/it] 14%|█▍ | 4140/30000 [2:38:04<16:31:34, 2.30s/it] 14%|█▍ | 4141/30000 [2:38:06<16:38:06, 2.32s/it] 14%|█▍ | 4142/30000 [2:38:09<16:38:38, 2.32s/it] 14%|█▍ | 4143/30000 [2:38:11<16:41:20, 2.32s/it] 14%|█▍ | 4144/30000 [2:38:13<16:40:02, 2.32s/it] 14%|█▍ | 4145/30000 [2:38:16<16:32:40, 2.30s/it] 14%|█▍ | 4146/30000 [2:38:18<16:26:23, 2.29s/it] 14%|█▍ | 4147/30000 [2:38:20<16:24:25, 2.28s/it] 14%|█▍ | 4148/30000 [2:38:22<16:30:17, 2.30s/it] 14%|█▍ | 4149/30000 [2:38:25<16:29:22, 2.30s/it] 14%|█▍ | 4150/30000 [2:38:27<16:28:15, 2.29s/it] 14%|█▍ | 4150/30000 [2:38:27<16:28:15, 2.29s/it] 14%|█▍ | 4151/30000 [2:38:29<16:37:03, 2.31s/it] 14%|█▍ | 4152/30000 [2:38:32<16:32:30, 2.30s/it] 14%|█▍ | 4153/30000 [2:38:34<16:28:41, 2.30s/it] 14%|█▍ | 4154/30000 [2:38:36<16:26:41, 2.29s/it] 14%|█▍ | 4155/30000 [2:38:39<16:33:23, 2.31s/it] 14%|█▍ | 4156/30000 [2:38:41<16:27:51, 2.29s/it] 14%|█▍ | 4157/30000 [2:38:43<16:22:42, 2.28s/it] 14%|█▍ | 4158/30000 [2:38:45<16:31:13, 2.30s/it] 14%|█▍ | 4159/30000 [2:38:48<16:27:28, 2.29s/it] 14%|█▍ | 4160/30000 [2:38:50<16:22:38, 2.28s/it] 14%|█▍ | 4160/30000 [2:38:50<16:22:38, 2.28s/it] 14%|█▍ | 4161/30000 [2:38:52<16:20:45, 2.28s/it] 14%|█▍ | 4162/30000 [2:38:55<16:28:42, 2.30s/it] 14%|█▍ | 4163/30000 [2:38:57<16:19:27, 2.27s/it] 14%|█▍ | 4164/30000 [2:38:59<16:22:29, 2.28s/it] 14%|█▍ | 4165/30000 [2:39:01<16:22:58, 2.28s/it] 14%|█▍ | 4166/30000 [2:39:04<16:19:50, 2.28s/it] 14%|█▍ | 4167/30000 [2:39:06<16:17:49, 2.27s/it] 14%|█▍ | 4168/30000 [2:39:08<16:26:36, 2.29s/it] 14%|█▍ | 4169/30000 [2:39:11<16:30:39, 2.30s/it] 14%|█▍ | 4170/30000 [2:39:13<16:25:15, 2.29s/it] 14%|█▍ | 4170/30000 [2:39:13<16:25:15, 2.29s/it] 14%|█▍ | 4171/30000 [2:39:15<16:30:34, 2.30s/it] 14%|█▍ | 4172/30000 [2:39:18<16:37:13, 2.32s/it] 14%|█▍ | 4173/30000 [2:39:20<16:31:26, 2.30s/it] 14%|█▍ | 4174/30000 [2:39:22<16:30:16, 2.30s/it] 14%|█▍ | 4175/30000 [2:39:24<16:32:33, 2.31s/it] 14%|█▍ | 4176/30000 [2:39:27<16:25:06, 2.29s/it] 14%|█▍ | 4177/30000 [2:39:29<16:22:59, 2.28s/it] 14%|█▍ | 4178/30000 [2:39:31<16:27:10, 2.29s/it] 14%|█▍ | 4179/30000 [2:39:34<16:38:58, 2.32s/it] 14%|█▍ | 4180/30000 [2:39:36<16:27:38, 2.30s/it] 14%|█▍ | 4180/30000 [2:39:36<16:27:38, 2.30s/it] 14%|█▍ | 4181/30000 [2:39:38<16:25:13, 2.29s/it] 14%|█▍ | 4182/30000 [2:39:40<16:29:12, 2.30s/it] 14%|█▍ | 4183/30000 [2:39:43<16:29:06, 2.30s/it] 14%|█▍ | 4184/30000 [2:39:45<16:24:39, 2.29s/it] 14%|█▍ | 4185/30000 [2:39:47<16:33:18, 2.31s/it] 14%|█▍ | 4186/30000 [2:39:50<16:34:35, 2.31s/it] 14%|█▍ | 4187/30000 [2:39:52<16:44:54, 2.34s/it] 14%|█▍ | 4188/30000 [2:39:54<16:37:10, 2.32s/it] 14%|█▍ | 4189/30000 [2:39:57<16:28:23, 2.30s/it] 14%|█▍ | 4190/30000 [2:39:59<16:33:03, 2.31s/it] 14%|█▍ | 4190/30000 [2:39:59<16:33:03, 2.31s/it] 14%|█▍ | 4191/30000 [2:40:01<16:35:47, 2.31s/it] 14%|█▍ | 4192/30000 [2:40:04<16:32:00, 2.31s/it] 14%|█▍ | 4193/30000 [2:40:06<16:23:32, 2.29s/it] 14%|█▍ | 4194/30000 [2:40:08<16:28:00, 2.30s/it] 14%|█▍ | 4195/30000 [2:40:10<16:22:33, 2.28s/it] 14%|█▍ | 4196/30000 [2:40:13<16:17:59, 2.27s/it] 14%|█▍ | 4197/30000 [2:40:15<16:26:05, 2.29s/it] 14%|█▍ | 4198/30000 [2:40:17<16:24:32, 2.29s/it] 14%|█▍ | 4199/30000 [2:40:19<16:18:16, 2.27s/it] 14%|█▍ | 4200/30000 [2:40:22<16:24:29, 2.29s/it] +{'loss': 0.0167, 'grad_norm': 0.10821790993213654, 'learning_rate': 2.952090273431877e-05, 'flow_matching_loss': 0.01665734425187111, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.11} +{'loss': 0.0169, 'grad_norm': 0.08292470872402191, 'learning_rate': 2.9516748367624808e-05, 'flow_matching_loss': 0.01686789495870471, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.11} +{'loss': 0.0185, 'grad_norm': 0.08133069425821304, 'learning_rate': 2.951257636171274e-05, 'flow_matching_loss': 0.01849607201293111, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.11} +{'loss': 0.02, 'grad_norm': 0.09021090716123581, 'learning_rate': 2.9508386721651944e-05, 'flow_matching_loss': 0.019985859096050263, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.11} +{'loss': 0.0183, 'grad_norm': 0.08560869842767715, 'learning_rate': 2.9504179452533228e-05, 'flow_matching_loss': 0.0183102079667151, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.12} +{'loss': 0.0185, 'grad_norm': 0.13800564408302307, 'learning_rate': 2.9499954559468826e-05, 'flow_matching_loss': 0.018546116072684526, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.12} +{'loss': 0.0174, 'grad_norm': 0.08593789488077164, 'learning_rate': 2.9495712047592378e-05, 'flow_matching_loss': 0.01735885962843895, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.12} +{'loss': 0.0175, 'grad_norm': 0.08763536810874939, 'learning_rate': 2.949145192205893e-05, 'flow_matching_loss': 0.01754489177837968, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.13} +{'loss': 0.0187, 'grad_norm': 0.10825508832931519, 'learning_rate': 2.9487174188044953e-05, 'flow_matching_loss': 0.01871220227330923, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.13} +{'loss': 0.0188, 'grad_norm': 0.09162640571594238, 'learning_rate': 2.9482878850748276e-05, 'flow_matching_loss': 0.018772626109421254, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.13} +{'loss': 0.0176, 'grad_norm': 0.08956382423639297, 'learning_rate': 2.9478565915388157e-05, 'flow_matching_loss': 0.017574034724384548, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.13} +{'loss': 0.0159, 'grad_norm': 0.09416811168193817, 'learning_rate': 2.9474235387205207e-05, 'flow_matching_loss': 0.015855283103883268, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.14} +{'loss': 0.0182, 'grad_norm': 0.08198630809783936, 'learning_rate': 2.946988727146143e-05, 'flow_matching_loss': 0.01824382385239005, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.14} +{'loss': 0.0175, 'grad_norm': 0.09703607857227325, 'learning_rate': 2.9465521573440196e-05, 'flow_matching_loss': 0.01752053163945675, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.14} +{'loss': 0.0178, 'grad_norm': 0.10077141970396042, 'learning_rate': 2.946113829844624e-05, 'flow_matching_loss': 0.017832505516707897, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.15} +{'loss': 0.018, 'grad_norm': 0.10195383429527283, 'learning_rate': 2.945673745180566e-05, 'flow_matching_loss': 0.01803307430818677, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.15} +{'loss': 0.0182, 'grad_norm': 0.0788196325302124, 'learning_rate': 2.945231903886589e-05, 'flow_matching_loss': 0.018184073269367218, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.15} +{'loss': 0.018, 'grad_norm': 0.08540042489767075, 'learning_rate': 2.9447883064995728e-05, 'flow_matching_loss': 0.018025304563343524, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.16} +{'loss': 0.019, 'grad_norm': 0.09530279785394669, 'learning_rate': 2.9443429535585288e-05, 'flow_matching_loss': 0.01903387373313308, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.16} +{'loss': 0.0206, 'grad_norm': 0.11037687212228775, 'learning_rate': 2.9438958456046044e-05, 'flow_matching_loss': 0.020587613712996243, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.16} +{'loss': 0.0176, 'grad_norm': 0.10139672458171844, 'learning_rate': 2.9434469831810765e-05, 'flow_matching_loss': 0.017603846732527018, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.16} +{'loss': 0.0164, 'grad_norm': 0.07711735367774963, 'learning_rate': 2.9429963668333566e-05, 'flow_matching_loss': 0.01641847686842084, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.17} +{'loss': 0.0186, 'grad_norm': 0.1207059696316719, 'learning_rate': 2.9425439971089845e-05, 'flow_matching_loss': 0.018554951902478932, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.17} +{'loss': 0.0169, 'grad_norm': 0.073297418653965, 'learning_rate': 2.9420898745576336e-05, 'flow_matching_loss': 0.016874324530363083, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.17} +{'loss': 0.0177, 'grad_norm': 0.12059706449508667, 'learning_rate': 2.9416339997311044e-05, 'flow_matching_loss': 0.017677681427448987, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.18} +{'loss': 0.0187, 'grad_norm': 0.10837530344724655, 'learning_rate': 2.941176373183329e-05, 'flow_matching_loss': 0.018725256156176328, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.18} +{'loss': 0.0183, 'grad_norm': 0.1228313073515892, 'learning_rate': 2.9407169954703653e-05, 'flow_matching_loss': 0.018330416921526194, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.18} +{'loss': 0.0172, 'grad_norm': 0.09575636684894562, 'learning_rate': 2.9402558671504018e-05, 'flow_matching_loss': 0.017222365736961363, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.18} +{'loss': 0.0187, 'grad_norm': 0.0976257175207138, 'learning_rate': 2.9397929887837524e-05, 'flow_matching_loss': 0.018668787088245153, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.19} +{'loss': 0.0176, 'grad_norm': 0.10228557884693146, 'learning_rate': 2.939328360932858e-05, 'flow_matching_loss': 0.017625673115253447, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.19} +{'loss': 0.0188, 'grad_norm': 0.08967453241348267, 'learning_rate': 2.938861984162285e-05, 'flow_matching_loss': 0.018832901399582623, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.19} +{'loss': 0.0164, 'grad_norm': 0.08499201387166977, 'learning_rate': 2.938393859038726e-05, 'flow_matching_loss': 0.016413830872625112, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.2} +{'loss': 0.0173, 'grad_norm': 0.08433124423027039, 'learning_rate': 2.9379239861309964e-05, 'flow_matching_loss': 0.017256331630051136, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.2} +{'loss': 0.0174, 'grad_norm': 0.06862207502126694, 'learning_rate': 2.937452366010036e-05, 'flow_matching_loss': 0.017418409790843727, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.2} +{'loss': 0.0172, 'grad_norm': 0.11601414531469345, 'learning_rate': 2.936978999248908e-05, 'flow_matching_loss': 0.017175285890698434, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.2} +{'loss': 0.0179, 'grad_norm': 0.11593836545944214, 'learning_rate': 2.936503886422798e-05, 'flow_matching_loss': 0.017895869445055723, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.21} +{'loss': 0.0171, 'grad_norm': 0.09781889617443085, 'learning_rate': 2.9360270281090122e-05, 'flow_matching_loss': 0.017054905649274588, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.21} +{'loss': 0.0186, 'grad_norm': 0.09650484472513199, 'learning_rate': 2.935548424886979e-05, 'flow_matching_loss': 0.018594330083578824, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.21} +{'loss': 0.0177, 'grad_norm': 0.11229865998029709, 'learning_rate': 2.935068077338246e-05, 'flow_matching_loss': 0.017697219830006362, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.22} +{'loss': 0.0168, 'grad_norm': 0.07744954526424408, 'learning_rate': 2.934585986046481e-05, 'flow_matching_loss': 0.016808687802404167, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.22} +{'loss': 0.0182, 'grad_norm': 0.07977195829153061, 'learning_rate': 2.93410215159747e-05, 'flow_matching_loss': 0.018246331345289947, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.22} 14%|█▍ | 4200/30000 [2:40:22<16:24:29, 2.29s/it] 14%|█▍ | 4201/30000 [2:40:24<16:30:48, 2.30s/it] 14%|█▍ | 4202/30000 [2:40:26<16:32:46, 2.31s/it] 14%|█▍ | 4203/30000 [2:40:29<16:28:17, 2.30s/it] 14%|█▍ | 4204/30000 [2:40:31<16:32:31, 2.31s/it] 14%|█▍ | 4205/30000 [2:40:33<16:35:05, 2.31s/it] 14%|█▍ | 4206/30000 [2:40:36<16:35:15, 2.32s/it] 14%|█▍ | 4207/30000 [2:40:38<16:29:36, 2.30s/it] 14%|█▍ | 4208/30000 [2:40:40<16:35:10, 2.32s/it] 14%|█▍ | 4209/30000 [2:40:43<16:37:02, 2.32s/it] 14%|█▍ | 4210/30000 [2:40:45<16:36:20, 2.32s/it] 14%|█▍ | 4210/30000 [2:40:45<16:36:20, 2.32s/it] 14%|█▍ | 4211/30000 [2:40:47<16:25:41, 2.29s/it] 14%|█▍ | 4212/30000 [2:40:49<16:23:12, 2.29s/it] 14%|█▍ | 4213/30000 [2:40:52<16:26:58, 2.30s/it] 14%|█▍ | 4214/30000 [2:40:54<16:27:53, 2.30s/it] 14%|█▍ | 4215/30000 [2:40:56<16:25:57, 2.29s/it] 14%|█▍ | 4216/30000 [2:40:59<16:25:56, 2.29s/it] 14%|█▍ | 4217/30000 [2:41:01<16:22:14, 2.29s/it] 14%|█▍ | 4218/30000 [2:41:03<16:21:21, 2.28s/it] 14%|█▍ | 4219/30000 [2:41:05<16:14:03, 2.27s/it] 14%|█▍ | 4220/30000 [2:41:08<16:19:17, 2.28s/it] 14%|█▍ | 4220/30000 [2:41:08<16:19:17, 2.28s/it] 14%|█▍ | 4221/30000 [2:41:10<16:19:50, 2.28s/it] 14%|█▍ | 4222/30000 [2:41:12<16:12:29, 2.26s/it] 14%|█▍ | 4223/30000 [2:41:15<16:14:02, 2.27s/it] 14%|█▍ | 4224/30000 [2:41:17<16:34:49, 2.32s/it] 14%|█▍ | 4225/30000 [2:41:19<16:28:34, 2.30s/it] 14%|█▍ | 4226/30000 [2:41:22<16:29:39, 2.30s/it] 14%|█▍ | 4227/30000 [2:41:24<16:33:24, 2.31s/it] 14%|█▍ | 4228/30000 [2:41:26<16:30:50, 2.31s/it] 14%|█▍ | 4229/30000 [2:41:28<16:27:57, 2.30s/it] 14%|█▍ | 4230/30000 [2:41:31<16:32:13, 2.31s/it] 14%|█▍ | 4230/30000 [2:41:31<16:32:13, 2.31s/it] 14%|█▍ | 4231/30000 [2:41:33<16:37:19, 2.32s/it] 14%|█▍ | 4232/30000 [2:41:35<16:31:02, 2.31s/it] 14%|█▍ | 4233/30000 [2:41:38<16:24:07, 2.29s/it] 14%|█▍ | 4234/30000 [2:41:40<16:35:03, 2.32s/it] 14%|█▍ | 4235/30000 [2:41:42<16:32:04, 2.31s/it] 14%|█▍ | 4236/30000 [2:41:45<16:21:34, 2.29s/it] 14%|█▍ | 4237/30000 [2:41:47<16:23:25, 2.29s/it] 14%|█▍ | 4238/30000 [2:41:49<16:27:03, 2.30s/it] 14%|█▍ | 4239/30000 [2:41:51<16:23:41, 2.29s/it] 14%|█▍ | 4240/30000 [2:41:54<16:25:15, 2.29s/it] 14%|█▍ | 4240/30000 [2:41:54<16:25:15, 2.29s/it] 14%|█▍ | 4241/30000 [2:41:56<16:30:28, 2.31s/it] 14%|█▍ | 4242/30000 [2:41:58<16:26:19, 2.30s/it] 14%|█▍ | 4243/30000 [2:42:01<16:27:12, 2.30s/it] 14%|█▍ | 4244/30000 [2:42:03<16:21:48, 2.29s/it] 14%|█▍ | 4245/30000 [2:42:05<16:16:48, 2.28s/it] 14%|█▍ | 4246/30000 [2:42:08<16:21:43, 2.29s/it] 14%|█▍ | 4247/30000 [2:42:10<16:31:00, 2.31s/it] 14%|█▍ | 4248/30000 [2:42:12<16:29:30, 2.31s/it] 14%|█▍ | 4249/30000 [2:42:14<16:17:21, 2.28s/it] 14%|█▍ | 4250/30000 [2:42:17<16:27:55, 2.30s/it] 14%|█▍ | 4250/30000 [2:42:17<16:27:55, 2.30s/it] 14%|█▍ | 4251/30000 [2:42:19<16:33:16, 2.31s/it] 14%|█▍ | 4252/30000 [2:42:21<16:18:01, 2.28s/it] 14%|█▍ | 4253/30000 [2:42:24<16:27:32, 2.30s/it] 14%|█▍ | 4254/30000 [2:42:26<16:31:32, 2.31s/it] 14%|█▍ | 4255/30000 [2:42:28<16:25:34, 2.30s/it] 14%|█▍ | 4256/30000 [2:42:31<16:30:35, 2.31s/it] 14%|█▍ | 4257/30000 [2:42:33<16:26:33, 2.30s/it] 14%|█▍ | 4258/30000 [2:42:35<16:21:04, 2.29s/it] 14%|█▍ | 4259/30000 [2:42:37<16:20:33, 2.29s/it] 14%|█▍ | 4260/30000 [2:42:40<16:20:45, 2.29s/it] 14%|█▍ | 4260/30000 [2:42:40<16:20:45, 2.29s/it] 14%|█▍ | 4261/30000 [2:42:42<16:24:03, 2.29s/it] 14%|█▍ | 4262/30000 [2:42:44<16:19:10, 2.28s/it] 14%|█▍ | 4263/30000 [2:42:47<16:28:49, 2.31s/it] 14%|█▍ | 4264/30000 [2:42:49<16:22:40, 2.29s/it] 14%|█▍ | 4265/30000 [2:42:51<16:26:35, 2.30s/it] 14%|█▍ | 4266/30000 [2:42:54<16:29:24, 2.31s/it] 14%|█▍ | 4267/30000 [2:42:56<16:30:07, 2.31s/it] 14%|█▍ | 4268/30000 [2:42:58<16:29:47, 2.31s/it] 14%|█▍ | 4269/30000 [2:43:00<16:34:26, 2.32s/it] 14%|█▍ | 4270/30000 [2:43:03<16:32:20, 2.31s/it] 14%|█▍ | 4270/30000 [2:43:03<16:32:20, 2.31s/it] 14%|█▍ | 4271/30000 [2:43:05<16:29:09, 2.31s/it] 14%|█▍ | 4272/30000 [2:43:07<16:28:38, 2.31s/it] 14%|█▍ | 4273/30000 [2:43:10<16:30:19, 2.31s/it] 14%|█▍ | 4274/30000 [2:43:12<16:25:02, 2.30s/it] 14%|█▍ | 4275/30000 [2:43:14<16:18:24, 2.28s/it] 14%|█▍ | 4276/30000 [2:43:16<16:00:10, 2.24s/it] 14%|█▍ | 4277/30000 [2:43:19<15:52:37, 2.22s/it] 14%|█▍ | 4278/30000 [2:43:21<16:06:28, 2.25s/it] 14%|█▍ | 4279/30000 [2:43:23<16:14:55, 2.27s/it] 14%|█▍ | 4280/30000 [2:43:25<16:14:59, 2.27s/it] 14%|█▍ | 4280/30000 [2:43:25<16:14:59, 2.27s/it] 14%|█▍ | 4281/30000 [2:43:28<16:15:12, 2.28s/it] 14%|█▍ | 4282/30000 [2:43:30<16:25:20, 2.30s/it] 14%|█▍ | 4283/30000 [2:43:32<16:20:38, 2.29s/it] 14%|█▍ | 4284/30000 [2:43:35<16:16:25, 2.28s/it] 14%|█▍ | 4285/30000 [2:43:37<16:32:04, 2.31s/it] 14%|█▍ | 4286/30000 [2:43:39<16:20:57, 2.29s/it] 14%|█▍ | 4287/30000 [2:43:42<16:22:48, 2.29s/it] 14%|█▍ | 4288/30000 [2:43:44<16:21:43, 2.29s/it] 14%|█▍ | 4289/30000 [2:43:46<16:19:28, 2.29s/it] 14%|█▍ | 4290/30000 [2:43:48<16:04:10, 2.25s/it] 14%|█▍ | 4290/30000 [2:43:48<16:04:10, 2.25s/it] 14%|█▍ | 4291/30000 [2:43:51<16:15:36, 2.28s/it] 14%|█▍ | 4292/30000 [2:43:53<16:19:51, 2.29s/it] 14%|█▍ | 4293/30000 [2:43:55<16:17:24, 2.28s/it] 14%|█▍ | 4294/30000 [2:43:57<16:14:58, 2.28s/it] 14%|█▍ | 4295/30000 [2:44:00<16:22:28, 2.29s/it] 14%|█▍ | 4296/30000 [2:44:02<16:22:59, 2.29s/it] 14%|█▍ | 4297/30000 [2:44:04<16:30:51, 2.31s/it] 14%|█▍ | 4298/30000 [2:44:07<16:30:35, 2.31s/it] 14%|█▍ | 4299/30000 [2:44:09<16:33:35, 2.32s/it] 14%|█▍ | 4300/30000 [2:44:11<16:35:25, 2.32s/it] 14%|█▍ | 4300/30000 [2:44:11<16:35:25, 2.32s/it] 14%|█▍ | 4301/30000 [2:44:14<16:27:08, 2.30s/it] 14%|█▍ | 4302/30000 [2:44:16<16:21:34, 2.29s/it] 14%|█▍ | 4303/30000 [2:44:18<16:29:22, 2.31s/it] 14%|█▍ | 4304/30000 [2:44:21<16:27:01, 2.30s/it] 14%|█▍ | 4305/30000 [2:44:23<16:20:10, 2.29s/it] 14%|█▍ | 4306/30000 [2:44:25<16:21:27, 2.29s/it] 14%|█▍ | 4307/30000 [2:44:27<16:28:33, 2.31s/it] 14%|█▍ | 4308/30000 [2:44:30<16:29:17, 2.31s/it] 14%|█▍ | 4309/30000 [2:44:32<16:21:56, 2.29s/it] 14%|█▍ | 4310/30000 [2:44:34<16:26:29, 2.30s/it] 14%|█▍ | 4310/30000 [2:44:34<16:26:29, 2.30s/it] 14%|█▍ | 4311/30000 [2:44:37<16:28:59, 2.31s/it] 14%|█▍ | 4312/30000 [2:44:39<16:23:26, 2.30s/it] 14%|█▍ | 4313/30000 [2:44:41<16:14:20, 2.28s/it] 14%|█▍ | 4314/30000 [2:44:44<16:23:01, 2.30s/it] 14%|█▍ | 4315/30000 [2:44:46<16:26:34, 2.30s/it] 14%|█▍ | 4316/30000 [2:44:48<16:23:35, 2.30s/it] 14%|█▍ | 4317/30000 [2:44:50<16:25:42, 2.30s/it] 14%|█▍ | 4318/30000 [2:44:53<16:35:46, 2.33s/it] 14%|█▍ | 4319/30000 [2:44:55<16:35:26, 2.33s/it] 14%|█▍ | 4320/30000 [2:44:57<16:26:49, 2.31s/it] 14%|█▍ | 4320/30000 [2:44:57<16:26:49, 2.31s/it] 14%|█▍ | 4321/30000 [2:45:00<16:21:41, 2.29s/it] 14%|█▍ | 4322/30000 [2:45:02<16:26:48, 2.31s/it] 14%|█▍ | 4323/30000 [2:45:04<16:23:02, 2.30s/it] 14%|█▍ | 4324/30000 [2:45:07<16:14:36, 2.28s/it] 14%|█▍ | 4325/30000 [2:45:09<16:21:53, 2.29s/it] 14%|█▍ | 4326/30000 [2:45:11<16:25:12, 2.30s/it] 14%|█▍ | 4327/30000 [2:45:13<16:19:32, 2.29s/it] 14%|█▍ | 4328/30000 [2:45:16<16:16:56, 2.28s/it] 14%|█▍ | 4329/30000 [2:45:18<16:27:02, 2.31s/it] 14%|█▍ | 4330/30000 [2:45:20<16:23:41, 2.30s/it] 14%|█▍ | 4330/30000 [2:45:20<16:23:41, 2.30s/it] 14%|█▍ | 4331/30000 [2:45:23<16:23:39, 2.30s/it] 14%|█▍ | 4332/30000 [2:45:25<16:20:52, 2.29s/it] 14%|█▍ | 4333/30000 [2:45:27<16:28:13, 2.31s/it] 14%|█▍ | 4334/30000 [2:45:30<16:28:21, 2.31s/it] 14%|█▍ | 4335/30000 [2:45:32<16:35:42, 2.33s/it] 14%|█▍ | 4336/30000 [2:45:34<16:26:18, 2.31s/it] 14%|█▍ | 4337/30000 [2:45:37<16:42:12, 2.34s/it] 14%|█▍ | 4338/30000 [2:45:39<16:43:42, 2.35s/it] 14%|█▍ | 4339/30000 [2:45:41<16:39:56, 2.34s/it] 14%|█▍ | 4340/30000 [2:45:44<16:31:18, 2.32s/it] 14%|█▍ | 4340/30000 [2:45:44<16:31:18, 2.32s/it] 14%|█▍ | 4341/30000 [2:45:46<16:23:51, 2.30s/it] 14%|█▍ | 4342/30000 [2:45:48<16:32:16, 2.32s/it] 14%|█▍ | 4343/30000 [2:45:50<16:27:49, 2.31s/it] 14%|█▍ | 4344/30000 [2:45:53<16:24:24, 2.30s/it] 14%|█▍ | 4345/30000 [2:45:55<16:28:16, 2.31s/it] 14%|█▍ | 4346/30000 [2:45:57<16:26:59, 2.31s/it] 14%|█▍ | 4347/30000 [2:46:00<16:23:40, 2.30s/it] 14%|█▍ | 4348/30000 [2:46:02<16:18:54, 2.29s/it] 14%|█▍ | 4349/30000 [2:46:04<16:26:10, 2.31s/it] 14%|█▍ | 4350/30000 [2:46:07<16:25:54, 2.31s/it] 14%|█▍ | 4350/30000 [2:46:07<16:25:54, 2.31s/it] 15%|█▍ | 4351/30000 [2:46:09<16:21:33, 2.30s/it] 15%|█▍ | 4352/30000 [2:46:11<16:14:36, 2.28s/it] 15%|█▍ | 4353/30000 [2:46:13<16:11:54, 2.27s/it] 15%|█▍ | 4354/30000 [2:46:16<16:11:56, 2.27s/it] 15%|█▍ | 4355/30000 [2:46:18<16:10:02, 2.27s/it] 15%|█▍ | 4356/30000 [2:46:20<16:21:16, 2.30s/it] 15%|█▍ | 4357/30000 [2:46:23<16:23:37, 2.30s/it] 15%|█▍ | 4358/30000 [2:46:25<16:17:11, 2.29s/it] 15%|█▍ | 4359/30000 [2:46:27<16:19:14, 2.29s/it] 15%|█▍ | 4360/30000 [2:46:29<16:15:17, 2.28s/it] 15%|█▍ | 4360/30000 [2:46:29<16:15:17, 2.28s/it] 15%|█▍ | 4361/30000 [2:46:32<16:04:18, 2.26s/it] 15%|█▍ | 4362/30000 [2:46:34<16:02:55, 2.25s/it] 15%|█▍ | 4363/30000 [2:46:36<16:10:03, 2.27s/it] 15%|█▍ | 4364/30000 [2:46:38<16:07:59, 2.27s/it] 15%|█▍ | 4365/30000 [2:46:41<16:13:37, 2.28s/it] 15%|█▍ | 4366/30000 [2:46:43<16:18:23, 2.29s/it] 15%|█▍ | 4367/30000 [2:46:45<16:16:04, 2.28s/it] 15%|█▍ | 4368/30000 [2:46:48<16:15:16, 2.28s/it] 15%|█▍ | 4369/30000 [2:46:50<16:13:19, 2.28s/it] 15%|█▍ | 4370/30000 [2:46:52<16:22:25, 2.30s/it] 15%|█▍ | 4370/30000 [2:46:52<16:22:25, 2.30s/it] 15%|█▍ | 4371/30000 [2:46:55<16:22:06, 2.30s/it] 15%|█▍ | 4372/30000 [2:46:57<16:18:06, 2.29s/it] 15%|█▍ | 4373/30000 [2:46:59<16:28:25, 2.31s/it] 15%|█▍ | 4374/30000 [2:47:01<16:23:19, 2.30s/it] 15%|█▍ | 4375/30000 [2:47:04<16:35:05, 2.33s/it] 15%|█▍ | 4376/30000 [2:47:06<16:22:07, 2.30s/it] 15%|█▍ | 4377/30000 [2:47:08<16:33:26, 2.33s/it] 15%|█▍ | 4378/30000 [2:47:11<16:31:17, 2.32s/it] 15%|█▍ | 4379/30000 [2:47:13<16:22:41, 2.30s/it] 15%|█▍ | 4380/30000 [2:47:15<16:17:08, 2.29s/it] 15%|█▍ | 4380/30000 [2:47:15<16:17:08, 2.29s/it] 15%|█▍ | 4381/30000 [2:47:18<16:14:30, 2.28s/it] 15%|█▍ | 4382/30000 [2:47:20<16:12:53, 2.28s/it] 15%|█▍ | 4383/30000 [2:47:22<16:11:20, 2.28s/it] 15%|█▍ | 4384/30000 [2:47:24<16:06:50, 2.26s/it] 15%|█▍ | 4385/30000 [2:47:27<16:04:11, 2.26s/it] 15%|█▍ | 4386/30000 [2:47:29<16:03:02, 2.26s/it] 15%|█▍ | 4387/30000 [2:47:31<16:09:55, 2.27s/it] 15%|█▍ | 4388/30000 [2:47:33<16:08:12, 2.27s/it] 15%|█▍ | 4389/30000 [2:47:36<16:09:51, 2.27s/it] 15%|█▍ | 4390/30000 [2:47:38<16:15:44, 2.29s/it] 15%|█▍ | 4390/30000 [2:47:38<16:15:44, 2.29s/it] 15%|█▍ | 4391/30000 [2:47:40<16:20:50, 2.30s/it] 15%|█▍ | 4392/30000 [2:47:43<16:18:41, 2.29s/it] 15%|█▍ | 4393/30000 [2:47:45<16:19:57, 2.30s/it] 15%|█▍ | 4394/30000 [2:47:47<16:23:19, 2.30s/it] 15%|█▍ | 4395/30000 [2:47:49<16:13:33, 2.28s/it] 15%|█▍ | 4396/30000 [2:47:52<16:24:06, 2.31s/it] 15%|█▍ | 4397/30000 [2:47:54<16:30:38, 2.32s/it] 15%|█▍ | 4398/30000 [2:47:56<16:25:18, 2.31s/it] 15%|█▍ | 4399/30000 [2:47:59<16:15:55, 2.29s/it] 15%|█▍ | 4400/30000 [2:48:01<16:23:53, 2.31s/it] 15%|█▍ | 4400/30000 [2:48:01<16:23:53, 2.31s/it] 15%|█▍ | 4401/30000 [2:48:03<16:26:08, 2.31s/it] 15%|█▍ | 4402/30000 [2:48:06<16:22:17, 2.30s/it] 15%|█▍ | 4403/30000 [2:48:08<16:19:56, 2.30s/it] 15%|█▍ | 4404/30000 [2:48:10<16:22:54, 2.30s/it] 15%|█▍ | 4405/30000 [2:48:12<16:14:55, 2.29s/it] 15%|█▍ | 4406/30000 [2:48:15<16:12:18, 2.28s/it] 15%|█▍ | 4407/30000 [2:48:17<16:20:21, 2.30s/it] 15%|█▍ | 4408/30000 [2:48:19<16:18:23, 2.29s/it] 15%|█▍ | 4409/30000 [2:48:22<16:16:06, 2.29s/it] 15%|█▍ | 4410/30000 [2:48:24<16:22:19, 2.30s/it] 15%|█▍ | 4410/30000 [2:48:24<16:22:19, 2.30s/it] 15%|█▍ | 4411/30000 [2:48:26<16:21:56, 2.30s/it] 15%|█▍ | 4412/30000 [2:48:29<16:14:11, 2.28s/it] 15%|█▍ | 4413/30000 [2:48:31<16:19:56, 2.30s/it] 15%|█▍ | 4414/30000 [2:48:33<16:18:37, 2.29s/it] 15%|█▍ | 4415/30000 [2:48:35<16:15:55, 2.29s/it] 15%|█▍ | 4416/30000 [2:48:38<16:20:43, 2.30s/it] 15%|█▍ | 4417/30000 [2:48:40<16:23:25, 2.31s/it] 15%|█▍ | 4418/30000 [2:48:42<16:17:49, 2.29s/it] 15%|█▍ | 4419/30000 [2:48:45<16:29:49, 2.32s/it] 15%|█▍ | 4420/30000 [2:48:47<16:30:47, 2.32s/it] 15%|█▍ | 4420/30000 [2:48:47<16:30:47, 2.32s/it] 15%|█▍ | 4421/30000 [2:48:49<16:27:49, 2.32s/it] 15%|█▍ | 4422/30000 [2:48:52<16:19:22, 2.30s/it] 15%|█▍ | 4423/30000 [2:48:54<16:24:26, 2.31s/it] 15%|█▍ | 4424/30000 [2:48:56<16:24:25, 2.31s/it] 15%|█▍ | 4425/30000 [2:48:58<16:11:53, 2.28s/it] 15%|█▍ | 4426/30000 [2:49:01<16:18:03, 2.29s/it] 15%|█▍ | 4427/30000 [2:49:03<16:26:34, 2.31s/it] 15%|█▍ | 4428/30000 [2:49:05<16:18:49, 2.30s/it] 15%|█▍ | 4429/30000 [2:49:08<16:21:54, 2.30s/it] 15%|█▍ | 4430/30000 [2:49:10<16:24:06, 2.31s/it] 15%|█▍ | 4430/30000 [2:49:10<16:24:06, 2.31s/it] 15%|█▍ | 4431/30000 [2:49:12<16:18:41, 2.30s/it] 15%|█▍ | 4432/30000 [2:49:15<16:19:35, 2.30s/it] 15%|█▍ | 4433/30000 [2:49:17<16:23:26, 2.31s/it] 15%|█▍ | 4434/30000 [2:49:19<16:17:28, 2.29s/it] 15%|█▍ | 4435/30000 [2:49:21<16:17:24, 2.29s/it] 15%|█▍ | 4436/30000 [2:49:24<16:21:18, 2.30s/it] 15%|█▍ | 4437/30000 [2:49:26<16:20:30, 2.30s/it] 15%|█▍ | 4438/30000 [2:49:28<16:15:16, 2.29s/it] 15%|█▍ | 4439/30000 [2:49:31<16:21:31, 2.30s/it] 15%|█▍ | 4440/30000 [2:49:33<16:31:35, 2.33s/it] 15%|█▍ | 4440/30000 [2:49:33<16:31:35, 2.33s/it] 15%|█▍ | 4441/30000 [2:49:35<16:22:33, 2.31s/it] 15%|█▍ | 4442/30000 [2:49:38<16:29:55, 2.32s/it] 15%|█▍ | 4443/30000 [2:49:40<16:23:50, 2.31s/it] 15%|█▍ | 4444/30000 [2:49:42<16:17:43, 2.30s/it] 15%|█▍ | 4445/30000 [2:49:45<16:31:34, 2.33s/it] 15%|█▍ | 4446/30000 [2:49:47<16:25:23, 2.31s/it] 15%|█▍ | 4447/30000 [2:49:49<16:25:42, 2.31s/it] 15%|█▍ | 4448/30000 [2:49:52<16:21:04, 2.30s/it] 15%|█▍ | 4449/30000 [2:49:54<16:25:22, 2.31s/it] 15%|█▍ | 4450/30000 [2:49:56<16:20:13, 2.30s/it] 15%|█▍ | 4450/30000 [2:49:56<16:20:13, 2.30s/it] 15%|█▍ | 4451/30000 [2:49:58<16:15:45, 2.29s/it] 15%|█▍ | 4452/30000 [2:50:01<16:20:13, 2.30s/it] 15%|█▍ | 4453/30000 [2:50:03<16:14:25, 2.29s/it] 15%|█▍ | 4454/30000 [2:50:05<16:00:30, 2.26s/it] 15%|█▍ | 4455/30000 [2:50:07<16:02:04, 2.26s/it] 15%|█▍ | 4456/30000 [2:50:10<15:54:20, 2.24s/it] 15%|█▍ | 4457/30000 [2:50:12<16:00:06, 2.26s/it] 15%|█▍ | 4458/30000 [2:50:14<16:11:11, 2.28s/it] 15%|█▍ | 4459/30000 [2:50:17<16:08:43, 2.28s/it] 15%|█▍ | 4460/30000 [2:50:19<16:07:46, 2.27s/it] 15%|█▍ | 4460/30000 [2:50:19<16:07:46, 2.27s/it] 15%|█▍ | 4461/30000 [2:50:21<16:19:13, 2.30s/it] 15%|█▍ | 4462/30000 [2:50:23<16:18:40, 2.30s/it] 15%|█▍ | 4463/30000 [2:50:26<16:13:17, 2.29s/it] 15%|█▍ | 4464/30000 [2:50:28<16:11:57, 2.28s/it] 15%|█▍ | 4465/30000 [2:50:30<16:20:15, 2.30s/it] 15%|█▍ | 4466/30000 [2:50:33<16:16:55, 2.30s/it] 15%|█▍ | 4467/30000 [2:50:35<16:01:43, 2.26s/it] 15%|█▍ | 4468/30000 [2:50:37<16:11:19, 2.28s/it] 15%|█▍ | 4469/30000 [2:50:39<16:06:21, 2.27s/it] 15%|█▍ | 4470/30000 [2:50:42<16:06:21, 2.27s/it] 15%|█▍ | 4470/30000 [2:50:42<16:06:21, 2.27s/it] 15%|█▍ | 4471/30000 [2:50:44<16:04:13, 2.27s/it] 15%|█▍ | 4472/30000 [2:50:46<16:07:23, 2.27s/it] 15%|█▍ | 4473/30000 [2:50:48<16:08:41, 2.28s/it] 15%|█▍ | 4474/30000 [2:50:51<16:07:30, 2.27s/it] 15%|█▍ | 4475/30000 [2:50:53<16:16:54, 2.30s/it] 15%|█▍ | 4476/30000 [2:50:55<16:26:30, 2.32s/it] 15%|█▍ | 4477/30000 [2:50:58<16:15:24, 2.29s/it] 15%|█▍ | 4478/30000 [2:51:00<16:11:51, 2.28s/it] 15%|█▍ | 4479/30000 [2:51:02<16:32:14, 2.33s/it] 15%|█▍ | 4480/30000 [2:51:05<16:28:45, 2.32s/it] 15%|█▍ | 4480/30000 [2:51:05<16:28:45, 2.32s/it] 15%|█▍ | 4481/30000 [2:51:07<16:27:32, 2.32s/it] 15%|█▍ | 4482/30000 [2:51:09<16:22:22, 2.31s/it] 15%|█▍ | 4483/30000 [2:51:12<16:16:49, 2.30s/it] 15%|█▍ | 4484/30000 [2:51:14<16:18:37, 2.30s/it] 15%|█▍ | 4485/30000 [2:51:16<16:17:56, 2.30s/it] 15%|█▍ | 4486/30000 [2:51:18<16:07:20, 2.27s/it] 15%|█▍ | 4487/30000 [2:51:21<16:26:24, 2.32s/it] 15%|█▍ | 4488/30000 [2:51:23<16:26:00, 2.32s/it] 15%|█▍ | 4489/30000 [2:51:25<16:18:42, 2.30s/it] 15%|█▍ | 4490/30000 [2:51:28<16:15:23, 2.29s/it] 15%|█▍ | 4490/30000 [2:51:28<16:15:23, 2.29s/it] 15%|█▍ | 4491/30000 [2:51:30<16:23:40, 2.31s/it] 15%|█▍ | 4492/30000 [2:51:32<16:32:57, 2.34s/it] 15%|█▍ | 4493/30000 [2:51:35<16:21:33, 2.31s/it] 15%|█▍ | 4494/30000 [2:51:37<16:17:43, 2.30s/it] 15%|█▍ | 4495/30000 [2:51:39<16:24:44, 2.32s/it] 15%|█▍ | 4496/30000 [2:51:42<16:26:21, 2.32s/it] 15%|█▍ | 4497/30000 [2:51:44<16:20:45, 2.31s/it] 15%|█▍ | 4498/30000 [2:51:46<16:17:23, 2.30s/it] 15%|█▍ | 4499/30000 [2:51:49<16:26:03, 2.32s/it] 15%|█▌ | 4500/30000 [2:51:51<16:27:11, 2.32s/it] 15%|█▌ | 4500/30000 [2:51:51<16:27:11, 2.32s/it] 15%|█▌ | 4501/30000 [2:51:53<16:09:59, 2.28s/it] 15%|█▌ | 4502/30000 [2:51:55<16:20:52, 2.31s/it] 15%|█▌ | 4503/30000 [2:51:58<16:16:45, 2.30s/it] 15%|█▌ | 4504/30000 [2:52:00<16:26:05, 2.32s/it] 15%|█▌ | 4505/30000 [2:52:02<16:22:07, 2.31s/it] 15%|█▌ | 4506/30000 [2:52:05<16:25:32, 2.32s/it] 15%|█▌ | 4507/30000 [2:52:07<16:42:10, 2.36s/it] 15%|█▌ | 4508/30000 [2:52:09<16:36:22, 2.35s/it] 15%|█▌ | 4509/30000 [2:52:12<16:31:03, 2.33s/it] 15%|█▌ | 4510/30000 [2:52:14<16:21:11, 2.31s/it] 15%|█▌ | 4510/30000 [2:52:14<16:21:11, 2.31s/it] 15%|█▌ | 4511/30000 [2:52:16<16:24:46, 2.32s/it] 15%|█▌ | 4512/30000 [2:52:19<16:27:51, 2.33s/it] 15%|█▌ | 4513/30000 [2:52:21<16:13:03, 2.29s/it] 15%|█▌ | 4514/30000 [2:52:23<16:09:49, 2.28s/it] 15%|█▌ | 4515/30000 [2:52:26<16:18:15, 2.30s/it] 15%|█▌ | 4516/30000 [2:52:28<16:20:34, 2.31s/it] 15%|█▌ | 4517/30000 [2:52:30<16:00:09, 2.26s/it] 15%|█▌ | 4518/30000 [2:52:32<16:08:12, 2.28s/it] 15%|█▌ | 4519/30000 [2:52:35<16:12:44, 2.29s/it] 15%|█▌ | 4520/30000 [2:52:37<16:11:23, 2.29s/it] 15%|█▌ | 4520/30000 [2:52:37<16:11:23, 2.29s/it] 15%|█▌ | 4521/30000 [2:52:39<16:09:03, 2.28s/it] 15%|█▌ | 4522/30000 [2:52:41<16:06:15, 2.28s/it] 15%|█▌ | 4523/30000 [2:52:44<16:10:31, 2.29s/it] 15%|█▌ | 4524/30000 [2:52:46<15:59:55, 2.26s/it] 15%|█▌ | 4525/30000 [2:52:48<16:06:38, 2.28s/it] 15%|█▌ | 4526/30000 [2:52:51<16:00:00, 2.26s/it] 15%|█▌ | 4527/30000 [2:52:53<16:00:34, 2.26s/it] 15%|█▌ | 4528/30000 [2:52:55<15:58:58, 2.26s/it] 15%|█▌ | 4529/30000 [2:52:57<16:05:52, 2.28s/it] 15%|█▌ | 4530/30000 [2:53:00<16:03:10, 2.27s/it] 15%|█▌ | 4530/30000 [2:53:00<16:03:10, 2.27s/it] 15%|█▌ | 4531/30000 [2:53:02<16:09:13, 2.28s/it] 15%|█▌ | 4532/30000 [2:53:04<16:11:24, 2.29s/it] 15%|█▌ | 4533/30000 [2:53:07<16:12:49, 2.29s/it] 15%|█▌ | 4534/30000 [2:53:09<16:11:28, 2.29s/it] 15%|█▌ | 4535/30000 [2:53:11<16:17:20, 2.30s/it] 15%|█▌ | 4536/30000 [2:53:13<16:22:25, 2.31s/it] 15%|█▌ | 4537/30000 [2:53:16<16:15:56, 2.30s/it] 15%|█▌ | 4538/30000 [2:53:18<16:11:41, 2.29s/it] 15%|█▌ | 4539/30000 [2:53:20<16:19:04, 2.31s/it] 15%|█▌ | 4540/30000 [2:53:23<16:05:35, 2.28s/it] 15%|█▌ | 4540/30000 [2:53:23<16:05:35, 2.28s/it] 15%|█▌ | 4541/30000 [2:53:25<16:04:23, 2.27s/it] 15%|█▌ | 4542/30000 [2:53:27<16:05:25, 2.28s/it] 15%|█▌ | 4543/30000 [2:53:29<16:04:53, 2.27s/it] 15%|█▌ | 4544/30000 [2:53:32<16:11:11, 2.29s/it] 15%|█▌ | 4545/30000 [2:53:34<16:19:03, 2.31s/it] 15%|█▌ | 4546/30000 [2:53:36<16:33:58, 2.34s/it] 15%|█▌ | 4547/30000 [2:53:39<16:26:58, 2.33s/it] 15%|█▌ | 4548/30000 [2:53:41<16:26:43, 2.33s/it] 15%|█▌ | 4549/30000 [2:53:43<16:27:41, 2.33s/it] 15%|█▌ | 4550/30000 [2:53:46<16:14:56, 2.30s/it] 15%|█▌ | 4550/30000 [2:53:46<16:14:56, 2.30s/it] 15%|█▌ | 4551/30000 [2:53:48<16:23:32, 2.32s/it] 15%|█▌ | 4552/30000 [2:53:50<16:37:50, 2.35s/it] 15%|█▌ | 4553/30000 [2:53:53<16:31:13, 2.34s/it] 15%|█▌ | 4554/30000 [2:53:55<16:18:26, 2.31s/it] 15%|█▌ | 4555/30000 [2:53:57<16:37:39, 2.35s/it] 15%|█▌ | 4556/30000 [2:54:00<16:31:12, 2.34s/it] 15%|█▌ | 4557/30000 [2:54:02<16:20:43, 2.31s/it] 15%|█▌ | 4558/30000 [2:54:04<16:18:43, 2.31s/it] 15%|█▌ | 4559/30000 [2:54:07<16:20:42, 2.31s/it] 15%|█▌ | 4560/30000 [2:54:09<16:00:06, 2.26s/it] 15%|█▌ | 4560/30000 [2:54:09<16:00:06, 2.26s/it] 15%|█▌ | 4561/30000 [2:54:11<16:05:55, 2.28s/it] 15%|█▌ | 4562/30000 [2:54:14<16:23:42, 2.32s/it] 15%|█▌ | 4563/30000 [2:54:16<16:21:18, 2.31s/it] 15%|█▌ | 4564/30000 [2:54:18<16:16:58, 2.30s/it] 15%|█▌ | 4565/30000 [2:54:20<16:20:33, 2.31s/it] 15%|█▌ | 4566/30000 [2:54:23<16:20:00, 2.31s/it] 15%|█▌ | 4567/30000 [2:54:25<16:11:29, 2.29s/it] 15%|█▌ | 4568/30000 [2:54:27<16:11:34, 2.29s/it] 15%|█▌ | 4569/30000 [2:54:30<16:14:08, 2.30s/it] 15%|█▌ | 4570/30000 [2:54:32<16:02:58, 2.27s/it] 15%|█▌ | 4570/30000 [2:54:32<16:02:58, 2.27s/it] 15%|█▌ | 4571/30000 [2:54:34<16:11:29, 2.29s/it] 15%|█▌ | 4572/30000 [2:54:36<16:10:02, 2.29s/it] 15%|█▌ | 4573/30000 [2:54:39<16:12:13, 2.29s/it] 15%|█▌ | 4574/30000 [2:54:41<16:05:56, 2.28s/it] 15%|█▌ | 4575/30000 [2:54:43<16:10:01, 2.29s/it] 15%|█▌ | 4576/30000 [2:54:46<16:07:02, 2.28s/it] 15%|█▌ | 4577/30000 [2:54:48<16:12:15, 2.29s/it] 15%|█▌ | 4578/30000 [2:54:50<16:16:37, 2.30s/it] 15%|█▌ | 4579/30000 [2:54:53<16:19:35, 2.31s/it] 15%|█▌ | 4580/30000 [2:54:55<16:13:17, 2.30s/it] 15%|█▌ | 4580/30000 [2:54:55<16:13:17, 2.30s/it] 15%|█▌ | 4581/30000 [2:54:57<16:19:07, 2.31s/it] 15%|█▌ | 4582/30000 [2:55:00<16:30:48, 2.34s/it] 15%|█▌ | 4583/30000 [2:55:02<16:25:49, 2.33s/it] 15%|█▌ | 4584/30000 [2:55:04<16:19:53, 2.31s/it] 15%|█▌ | 4585/30000 [2:55:06<16:15:06, 2.30s/it] 15%|█▌ | 4586/30000 [2:55:09<16:07:22, 2.28s/it] 15%|█▌ | 4587/30000 [2:55:11<16:11:44, 2.29s/it] 15%|█▌ | 4588/30000 [2:55:13<16:07:39, 2.28s/it] 15%|█▌ | 4589/30000 [2:55:15<16:03:47, 2.28s/it] 15%|█▌ | 4590/30000 [2:55:18<16:12:09, 2.30s/it] 15%|█▌ | 4590/30000 [2:55:18<16:12:09, 2.30s/it] 15%|█▌ | 4591/30000 [2:55:20<16:07:21, 2.28s/it] 15%|█▌ | 4592/30000 [2:55:22<16:07:23, 2.28s/it] 15%|█▌ | 4593/30000 [2:55:25<16:15:09, 2.30s/it] 15%|█▌ | 4594/30000 [2:55:27<15:59:15, 2.27s/it] 15%|█▌ | 4595/30000 [2:55:29<16:02:45, 2.27s/it] 15%|█▌ | 4596/30000 [2:55:31<16:05:43, 2.28s/it] 15%|█▌ | 4597/30000 [2:55:34<16:06:15, 2.28s/it] 15%|█▌ | 4598/30000 [2:55:36<15:58:07, 2.26s/it] 15%|█▌ | 4599/30000 [2:55:38<16:11:13, 2.29s/it] 15%|█▌ | 4600/30000 [2:55:41<16:14:43, 2.30s/it] 15%|█▌ | 4600/30000 [2:55:41<16:14:43, 2.30s/it] 15%|█▌ | 4601/30000 [2:55:43<16:11:32, 2.30s/it] 15%|█▌ | 4602/30000 [2:55:45<16:11:09, 2.29s/it] 15%|█▌ | 4603/30000 [2:55:48<16:08:22, 2.29s/it] 15%|█▌ | 4604/30000 [2:55:50<16:06:37, 2.28s/it] 15%|█▌ | 4605/30000 [2:55:52<16:07:32, 2.29s/it] 15%|█▌ | 4606/30000 [2:55:54<16:06:43, 2.28s/it] 15%|█▌ | 4607/30000 [2:55:57<16:04:51, 2.28s/it] 15%|█▌ | 4608/30000 [2:55:59<16:05:46, 2.28s/it] 15%|█▌ | 4609/30000 [2:56:01<16:15:36, 2.31s/it] 15%|█▌ | 4610/30000 [2:56:04<16:09:45, 2.29s/it] 15%|█▌ | 4610/30000 [2:56:04<16:09:45, 2.29s/it] 15%|█▌ | 4611/30000 [2:56:06<16:10:05, 2.29s/it] 15%|█▌ | 4612/30000 [2:56:08<16:23:35, 2.32s/it] 15%|█▌ | 4613/30000 [2:56:11<16:17:43, 2.31s/it] 15%|█▌ | 4614/30000 [2:56:13<16:26:37, 2.33s/it] 15%|█▌ | 4615/30000 [2:56:15<16:17:43, 2.31s/it] 15%|█▌ | 4616/30000 [2:56:18<16:22:48, 2.32s/it] 15%|█▌ | 4617/30000 [2:56:20<16:17:21, 2.31s/it] 15%|█▌ | 4618/30000 [2:56:22<16:12:47, 2.30s/it] 15%|█▌ | 4619/30000 [2:56:24<16:22:46, 2.32s/it] 15%|█▌ | 4620/30000 [2:56:27<16:14:41, 2.30s/it] +{'loss': 0.0172, 'grad_norm': 0.06756410002708435, 'learning_rate': 2.933616574579118e-05, 'flow_matching_loss': 0.017191207874566315, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.22} +{'loss': 0.0175, 'grad_norm': 0.11626248806715012, 'learning_rate': 2.933129255581447e-05, 'flow_matching_loss': 0.01754891499876976, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.23} +{'loss': 0.0197, 'grad_norm': 0.07443170994520187, 'learning_rate': 2.9326401951965954e-05, 'flow_matching_loss': 0.019722874648869036, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.23} +{'loss': 0.0171, 'grad_norm': 0.0938928872346878, 'learning_rate': 2.9321493940188165e-05, 'flow_matching_loss': 0.017063596099615098, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.23} +{'loss': 0.018, 'grad_norm': 0.0905243456363678, 'learning_rate': 2.931656852644482e-05, 'flow_matching_loss': 0.017973087914288043, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.24} +{'loss': 0.0171, 'grad_norm': 0.08731282502412796, 'learning_rate': 2.9311625716720748e-05, 'flow_matching_loss': 0.01711770324036479, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.24} +{'loss': 0.0185, 'grad_norm': 0.08525966852903366, 'learning_rate': 2.9306665517021927e-05, 'flow_matching_loss': 0.018538788426667452, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.24} +{'loss': 0.017, 'grad_norm': 0.09485193341970444, 'learning_rate': 2.9301687933375483e-05, 'flow_matching_loss': 0.017017473839223383, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.25} +{'loss': 0.02, 'grad_norm': 0.09641290456056595, 'learning_rate': 2.929669297182963e-05, 'flow_matching_loss': 0.019961184356361628, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.25} +{'loss': 0.0183, 'grad_norm': 0.09967435151338577, 'learning_rate': 2.929168063845373e-05, 'flow_matching_loss': 0.018259262572973965, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.25} +{'loss': 0.0181, 'grad_norm': 0.08038952201604843, 'learning_rate': 2.9286650939338236e-05, 'flow_matching_loss': 0.018136469274759294, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.25} +{'loss': 0.0178, 'grad_norm': 0.12716582417488098, 'learning_rate': 2.9281603880594706e-05, 'flow_matching_loss': 0.017794671840965746, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.26} +{'loss': 0.0187, 'grad_norm': 0.11686268448829651, 'learning_rate': 2.9276539468355796e-05, 'flow_matching_loss': 0.01874416312202811, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.26} +{'loss': 0.0188, 'grad_norm': 0.14414700865745544, 'learning_rate': 2.9271457708775244e-05, 'flow_matching_loss': 0.01879497552290559, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.26} +{'loss': 0.0159, 'grad_norm': 0.09816278517246246, 'learning_rate': 2.9266358608027863e-05, 'flow_matching_loss': 0.01588035998865962, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.27} +{'loss': 0.0187, 'grad_norm': 0.12927688658237457, 'learning_rate': 2.9261242172309543e-05, 'flow_matching_loss': 0.0186996141448617, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.27} +{'loss': 0.0175, 'grad_norm': 0.08532265573740005, 'learning_rate': 2.925610840783723e-05, 'flow_matching_loss': 0.01746635101735592, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.27} +{'loss': 0.0166, 'grad_norm': 0.0956910103559494, 'learning_rate': 2.9250957320848946e-05, 'flow_matching_loss': 0.01657203268259764, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.27} +{'loss': 0.0184, 'grad_norm': 0.09720522165298462, 'learning_rate': 2.924578891760374e-05, 'flow_matching_loss': 0.018387085013091563, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.28} +{'loss': 0.0172, 'grad_norm': 0.09337832033634186, 'learning_rate': 2.924060320438171e-05, 'flow_matching_loss': 0.017194008361548184, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.28} +{'loss': 0.0179, 'grad_norm': 0.12469319999217987, 'learning_rate': 2.9235400187483978e-05, 'flow_matching_loss': 0.017932888865470887, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.28} +{'loss': 0.0185, 'grad_norm': 0.09238400310277939, 'learning_rate': 2.9230179873232716e-05, 'flow_matching_loss': 0.01848166957497597, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.29} +{'loss': 0.0172, 'grad_norm': 0.142038494348526, 'learning_rate': 2.922494226797108e-05, 'flow_matching_loss': 0.017232669051736593, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.29} +{'loss': 0.0185, 'grad_norm': 0.07484936714172363, 'learning_rate': 2.9219687378063275e-05, 'flow_matching_loss': 0.018479018844664095, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.29} +{'loss': 0.0184, 'grad_norm': 0.11738834530115128, 'learning_rate': 2.921441520989447e-05, 'flow_matching_loss': 0.01841676440089941, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.29} +{'loss': 0.0181, 'grad_norm': 0.10958316177129745, 'learning_rate': 2.9209125769870854e-05, 'flow_matching_loss': 0.018146234937012194, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.3} +{'loss': 0.02, 'grad_norm': 0.11122437566518784, 'learning_rate': 2.92038190644196e-05, 'flow_matching_loss': 0.020034205354750156, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.3} +{'loss': 0.0192, 'grad_norm': 0.1333494484424591, 'learning_rate': 2.9198495099988843e-05, 'flow_matching_loss': 0.019206910766661166, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.3} +{'loss': 0.0178, 'grad_norm': 0.0669863149523735, 'learning_rate': 2.9193153883047712e-05, 'flow_matching_loss': 0.017751287575811146, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.31} +{'loss': 0.0188, 'grad_norm': 0.1233728751540184, 'learning_rate': 2.9187795420086287e-05, 'flow_matching_loss': 0.018790558725595475, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.31} +{'loss': 0.0199, 'grad_norm': 0.0998883992433548, 'learning_rate': 2.9182419717615602e-05, 'flow_matching_loss': 0.019940192345529795, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.31} +{'loss': 0.0189, 'grad_norm': 0.11406343430280685, 'learning_rate': 2.917702678216765e-05, 'flow_matching_loss': 0.018871088698506355, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.32} +{'loss': 0.0169, 'grad_norm': 0.08203241229057312, 'learning_rate': 2.9171616620295347e-05, 'flow_matching_loss': 0.016867879033088683, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.32} +{'loss': 0.0185, 'grad_norm': 0.11034803837537766, 'learning_rate': 2.9166189238572554e-05, 'flow_matching_loss': 0.018540493119508027, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.32} +{'loss': 0.0173, 'grad_norm': 0.06641562283039093, 'learning_rate': 2.9160744643594052e-05, 'flow_matching_loss': 0.017267818655818702, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.32} +{'loss': 0.0171, 'grad_norm': 0.0827474445104599, 'learning_rate': 2.915528284197554e-05, 'flow_matching_loss': 0.01710999021306634, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.33} +{'loss': 0.0172, 'grad_norm': 0.07506970316171646, 'learning_rate': 2.914980384035362e-05, 'flow_matching_loss': 0.0171516721136868, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.33} +{'loss': 0.0179, 'grad_norm': 0.08414442092180252, 'learning_rate': 2.9144307645385787e-05, 'flow_matching_loss': 0.017941261269152166, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.33} +{'loss': 0.0172, 'grad_norm': 0.08578333258628845, 'learning_rate': 2.9138794263750454e-05, 'flow_matching_loss': 0.01724579920992255, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.34} +{'loss': 0.0183, 'grad_norm': 0.08672690391540527, 'learning_rate': 2.913326370214688e-05, 'flow_matching_loss': 0.018318196292966604, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.34} +{'loss': 0.0176, 'grad_norm': 0.0760151743888855, 'learning_rate': 2.9127715967295234e-05, 'flow_matching_loss': 0.017566915322095155, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.34} +{'loss': 0.0176, 'grad_norm': 0.09774770587682724, 'learning_rate': 2.912215106593653e-05, 'flow_matching_loss': 0.01760668670758605, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.34} 15%|█▌ | 4620/30000 [2:56:27<16:14:41, 2.30s/it] 15%|█▌ | 4621/30000 [2:56:29<16:11:41, 2.30s/it] 15%|█▌ | 4622/30000 [2:56:31<16:21:19, 2.32s/it] 15%|█▌ | 4623/30000 [2:56:34<16:24:29, 2.33s/it] 15%|█▌ | 4624/30000 [2:56:36<16:18:09, 2.31s/it] 15%|█▌ | 4625/30000 [2:56:38<16:23:37, 2.33s/it] 15%|█▌ | 4626/30000 [2:56:41<16:14:37, 2.30s/it] 15%|█▌ | 4627/30000 [2:56:43<16:04:12, 2.28s/it] 15%|█▌ | 4628/30000 [2:56:45<16:07:37, 2.29s/it] 15%|█▌ | 4629/30000 [2:56:47<16:10:23, 2.29s/it] 15%|█▌ | 4630/30000 [2:56:50<16:12:35, 2.30s/it] 15%|█▌ | 4630/30000 [2:56:50<16:12:35, 2.30s/it] 15%|█▌ | 4631/30000 [2:56:52<16:19:37, 2.32s/it] 15%|█▌ | 4632/30000 [2:56:54<16:23:23, 2.33s/it] 15%|█▌ | 4633/30000 [2:56:57<16:19:02, 2.32s/it] 15%|█▌ | 4634/30000 [2:56:59<16:00:22, 2.27s/it] 15%|█▌ | 4635/30000 [2:57:01<16:03:43, 2.28s/it] 15%|█▌ | 4636/30000 [2:57:03<16:04:05, 2.28s/it] 15%|█▌ | 4637/30000 [2:57:06<16:04:53, 2.28s/it] 15%|█▌ | 4638/30000 [2:57:08<16:11:39, 2.30s/it] 15%|█▌ | 4639/30000 [2:57:10<16:05:56, 2.29s/it] 15%|█▌ | 4640/30000 [2:57:13<16:05:51, 2.29s/it] 15%|█▌ | 4640/30000 [2:57:13<16:05:51, 2.29s/it] 15%|█▌ | 4641/30000 [2:57:15<16:19:02, 2.32s/it] 15%|█▌ | 4642/30000 [2:57:17<16:18:33, 2.32s/it] 15%|█▌ | 4643/30000 [2:57:20<16:14:17, 2.31s/it] 15%|█▌ | 4644/30000 [2:57:22<16:13:42, 2.30s/it] 15%|█▌ | 4645/30000 [2:57:24<16:13:30, 2.30s/it] 15%|█▌ | 4646/30000 [2:57:27<16:11:22, 2.30s/it] 15%|█▌ | 4647/30000 [2:57:29<16:09:21, 2.29s/it] 15%|█▌ | 4648/30000 [2:57:31<16:13:52, 2.30s/it] 15%|█▌ | 4649/30000 [2:57:33<16:15:01, 2.31s/it] 16%|█▌ | 4650/30000 [2:57:36<16:14:02, 2.31s/it] 16%|█▌ | 4650/30000 [2:57:36<16:14:02, 2.31s/it] 16%|█▌ | 4651/30000 [2:57:38<16:20:46, 2.32s/it] 16%|█▌ | 4652/30000 [2:57:40<16:22:27, 2.33s/it] 16%|█▌ | 4653/30000 [2:57:43<16:12:49, 2.30s/it] 16%|█▌ | 4654/30000 [2:57:45<16:21:15, 2.32s/it] 16%|█▌ | 4655/30000 [2:57:47<16:14:59, 2.31s/it] 16%|█▌ | 4656/30000 [2:57:50<16:10:29, 2.30s/it] 16%|█▌ | 4657/30000 [2:57:52<16:12:55, 2.30s/it] 16%|█▌ | 4658/30000 [2:57:54<16:11:15, 2.30s/it] 16%|█▌ | 4659/30000 [2:57:56<15:58:19, 2.27s/it] 16%|█▌ | 4660/30000 [2:57:59<15:49:43, 2.25s/it] 16%|█▌ | 4660/30000 [2:57:59<15:49:43, 2.25s/it] 16%|█▌ | 4661/30000 [2:58:01<15:56:08, 2.26s/it] 16%|█▌ | 4662/30000 [2:58:03<15:55:57, 2.26s/it] 16%|█▌ | 4663/30000 [2:58:06<16:05:29, 2.29s/it] 16%|█▌ | 4664/30000 [2:58:08<16:05:20, 2.29s/it] 16%|█▌ | 4665/30000 [2:58:10<16:07:35, 2.29s/it] 16%|█▌ | 4666/30000 [2:58:12<16:06:41, 2.29s/it] 16%|█▌ | 4667/30000 [2:58:15<16:13:15, 2.31s/it] 16%|█▌ | 4668/30000 [2:58:17<16:17:44, 2.32s/it] 16%|█▌ | 4669/30000 [2:58:19<16:09:36, 2.30s/it] 16%|█▌ | 4670/30000 [2:58:22<16:18:19, 2.32s/it] 16%|█▌ | 4670/30000 [2:58:22<16:18:19, 2.32s/it] 16%|█▌ | 4671/30000 [2:58:24<16:17:32, 2.32s/it] 16%|█▌ | 4672/30000 [2:58:26<15:59:49, 2.27s/it] 16%|█▌ | 4673/30000 [2:58:28<16:02:01, 2.28s/it] 16%|█▌ | 4674/30000 [2:58:31<16:00:07, 2.27s/it] 16%|█▌ | 4675/30000 [2:58:33<16:00:32, 2.28s/it] 16%|█▌ | 4676/30000 [2:58:35<15:56:50, 2.27s/it] 16%|█▌ | 4677/30000 [2:58:38<15:57:17, 2.27s/it] 16%|█▌ | 4678/30000 [2:58:40<15:57:37, 2.27s/it] 16%|█▌ | 4679/30000 [2:58:42<15:56:00, 2.27s/it] 16%|█▌ | 4680/30000 [2:58:44<15:55:36, 2.26s/it] 16%|█▌ | 4680/30000 [2:58:44<15:55:36, 2.26s/it] 16%|█▌ | 4681/30000 [2:58:47<15:56:25, 2.27s/it] 16%|█▌ | 4682/30000 [2:58:49<16:00:09, 2.28s/it] 16%|█▌ | 4683/30000 [2:58:51<15:58:09, 2.27s/it] 16%|█▌ | 4684/30000 [2:58:53<16:05:46, 2.29s/it] 16%|█▌ | 4685/30000 [2:58:56<16:07:01, 2.29s/it] 16%|█▌ | 4686/30000 [2:58:58<16:02:15, 2.28s/it] 16%|█▌ | 4687/30000 [2:59:00<16:01:41, 2.28s/it] 16%|█▌ | 4688/30000 [2:59:03<16:02:19, 2.28s/it] 16%|█▌ | 4689/30000 [2:59:05<16:15:33, 2.31s/it] 16%|█▌ | 4690/30000 [2:59:07<16:20:02, 2.32s/it] 16%|█▌ | 4690/30000 [2:59:07<16:20:02, 2.32s/it] 16%|█▌ | 4691/30000 [2:59:10<16:13:10, 2.31s/it] 16%|█▌ | 4692/30000 [2:59:12<16:19:40, 2.32s/it] 16%|█▌ | 4693/30000 [2:59:14<16:19:39, 2.32s/it] 16%|█▌ | 4694/30000 [2:59:17<16:22:01, 2.33s/it] 16%|█▌ | 4695/30000 [2:59:19<16:14:57, 2.31s/it] 16%|█▌ | 4696/30000 [2:59:21<16:24:39, 2.33s/it] 16%|█▌ | 4697/30000 [2:59:24<16:20:44, 2.33s/it] 16%|█▌ | 4698/30000 [2:59:26<16:13:00, 2.31s/it] 16%|█▌ | 4699/30000 [2:59:28<16:08:22, 2.30s/it] 16%|█▌ | 4700/30000 [2:59:30<16:11:25, 2.30s/it] 16%|█▌ | 4700/30000 [2:59:30<16:11:25, 2.30s/it] 16%|█▌ | 4701/30000 [2:59:33<16:14:54, 2.31s/it] 16%|█▌ | 4702/30000 [2:59:35<16:08:57, 2.30s/it] 16%|█▌ | 4703/30000 [2:59:37<16:05:19, 2.29s/it] 16%|█▌ | 4704/30000 [2:59:40<16:10:50, 2.30s/it] 16%|█▌ | 4705/30000 [2:59:42<16:29:38, 2.35s/it] 16%|█▌ | 4706/30000 [2:59:44<16:19:04, 2.32s/it] 16%|█▌ | 4707/30000 [2:59:47<16:17:44, 2.32s/it] 16%|█▌ | 4708/30000 [2:59:49<16:24:52, 2.34s/it] 16%|█▌ | 4709/30000 [2:59:51<16:09:05, 2.30s/it] 16%|█▌ | 4710/30000 [2:59:53<16:00:18, 2.28s/it] 16%|█▌ | 4710/30000 [2:59:54<16:00:18, 2.28s/it] 16%|█▌ | 4711/30000 [2:59:56<15:57:10, 2.27s/it] 16%|█▌ | 4712/30000 [2:59:58<16:03:20, 2.29s/it] 16%|█▌ | 4713/30000 [3:00:00<16:10:50, 2.30s/it] 16%|█▌ | 4714/30000 [3:00:03<16:05:08, 2.29s/it] 16%|█▌ | 4715/30000 [3:00:05<16:10:14, 2.30s/it] 16%|█▌ | 4716/30000 [3:00:07<16:12:22, 2.31s/it] 16%|█▌ | 4717/30000 [3:00:10<15:56:55, 2.27s/it] 16%|█▌ | 4718/30000 [3:00:12<15:53:47, 2.26s/it] 16%|█▌ | 4719/30000 [3:00:14<16:15:32, 2.32s/it] 16%|█▌ | 4720/30000 [3:00:16<15:59:57, 2.28s/it] 16%|█▌ | 4720/30000 [3:00:16<15:59:57, 2.28s/it] 16%|█▌ | 4721/30000 [3:00:19<16:01:31, 2.28s/it] 16%|█▌ | 4722/30000 [3:00:21<16:05:49, 2.29s/it] 16%|█▌ | 4723/30000 [3:00:23<16:04:29, 2.29s/it] 16%|█▌ | 4724/30000 [3:00:26<16:01:43, 2.28s/it] 16%|█▌ | 4725/30000 [3:00:28<16:01:18, 2.28s/it] 16%|█▌ | 4726/30000 [3:00:30<16:03:06, 2.29s/it] 16%|█▌ | 4727/30000 [3:00:32<16:15:27, 2.32s/it] 16%|█▌ | 4728/30000 [3:00:35<16:22:45, 2.33s/it] 16%|█▌ | 4729/30000 [3:00:37<16:21:01, 2.33s/it] 16%|█▌ | 4730/30000 [3:00:40<16:21:43, 2.33s/it] 16%|█▌ | 4730/30000 [3:00:40<16:21:43, 2.33s/it] 16%|█▌ | 4731/30000 [3:00:42<16:12:00, 2.31s/it] 16%|█▌ | 4732/30000 [3:00:44<16:06:21, 2.29s/it] 16%|█▌ | 4733/30000 [3:00:46<16:09:17, 2.30s/it] 16%|█▌ | 4734/30000 [3:00:49<16:08:57, 2.30s/it] 16%|█▌ | 4735/30000 [3:00:51<16:05:14, 2.29s/it] 16%|█▌ | 4736/30000 [3:00:53<16:04:39, 2.29s/it] 16%|█▌ | 4737/30000 [3:00:56<16:07:50, 2.30s/it] 16%|█▌ | 4738/30000 [3:00:58<16:06:44, 2.30s/it] 16%|█▌ | 4739/30000 [3:01:00<16:10:41, 2.31s/it] 16%|█▌ | 4740/30000 [3:01:02<16:14:13, 2.31s/it] 16%|█▌ | 4740/30000 [3:01:03<16:14:13, 2.31s/it] 16%|█▌ | 4741/30000 [3:01:05<16:26:41, 2.34s/it] 16%|█▌ | 4742/30000 [3:01:07<16:26:04, 2.34s/it] 16%|█▌ | 4743/30000 [3:01:09<16:11:09, 2.31s/it] 16%|█▌ | 4744/30000 [3:01:12<16:15:58, 2.32s/it] 16%|█▌ | 4745/30000 [3:01:14<16:18:37, 2.32s/it] 16%|█▌ | 4746/30000 [3:01:16<15:58:58, 2.28s/it] 16%|█▌ | 4747/30000 [3:01:19<16:00:10, 2.28s/it] 16%|█▌ | 4748/30000 [3:01:21<16:08:26, 2.30s/it] 16%|█▌ | 4749/30000 [3:01:23<16:01:32, 2.28s/it] 16%|█▌ | 4750/30000 [3:01:25<16:01:02, 2.28s/it] 16%|█▌ | 4750/30000 [3:01:26<16:01:02, 2.28s/it] 16%|█▌ | 4751/30000 [3:01:28<16:09:35, 2.30s/it] 16%|█▌ | 4752/30000 [3:01:30<16:07:48, 2.30s/it] 16%|█▌ | 4753/30000 [3:01:32<16:05:40, 2.29s/it] 16%|█▌ | 4754/30000 [3:01:35<16:06:01, 2.30s/it] 16%|█▌ | 4755/30000 [3:01:37<16:11:00, 2.31s/it] 16%|█▌ | 4756/30000 [3:01:39<16:02:49, 2.29s/it] 16%|█▌ | 4757/30000 [3:01:42<16:16:14, 2.32s/it] 16%|█▌ | 4758/30000 [3:01:44<16:13:27, 2.31s/it] 16%|█▌ | 4759/30000 [3:01:46<16:11:09, 2.31s/it] 16%|█▌ | 4760/30000 [3:01:49<16:02:55, 2.29s/it] 16%|█▌ | 4760/30000 [3:01:49<16:02:55, 2.29s/it] 16%|█▌ | 4761/30000 [3:01:51<16:01:24, 2.29s/it] 16%|█▌ | 4762/30000 [3:01:53<16:09:27, 2.30s/it] 16%|█▌ | 4763/30000 [3:01:55<16:12:05, 2.31s/it] 16%|█▌ | 4764/30000 [3:01:58<16:07:45, 2.30s/it] 16%|█▌ | 4765/30000 [3:02:00<16:10:11, 2.31s/it] 16%|█▌ | 4766/30000 [3:02:02<16:10:49, 2.31s/it] 16%|█▌ | 4767/30000 [3:02:05<16:08:21, 2.30s/it] 16%|█▌ | 4768/30000 [3:02:07<16:09:01, 2.30s/it] 16%|█▌ | 4769/30000 [3:02:09<16:00:54, 2.29s/it] 16%|█▌ | 4770/30000 [3:02:12<16:03:30, 2.29s/it] 16%|█▌ | 4770/30000 [3:02:12<16:03:30, 2.29s/it] 16%|█▌ | 4771/30000 [3:02:14<16:00:20, 2.28s/it] 16%|█▌ | 4772/30000 [3:02:16<16:07:35, 2.30s/it] 16%|█▌ | 4773/30000 [3:02:19<16:21:38, 2.33s/it] 16%|█▌ | 4774/30000 [3:02:21<16:08:25, 2.30s/it] 16%|█▌ | 4775/30000 [3:02:23<16:07:23, 2.30s/it] 16%|█▌ | 4776/30000 [3:02:25<16:12:57, 2.31s/it] 16%|█▌ | 4777/30000 [3:02:28<16:14:44, 2.32s/it] 16%|█▌ | 4778/30000 [3:02:30<16:12:48, 2.31s/it] 16%|█▌ | 4779/30000 [3:02:32<16:06:50, 2.30s/it] 16%|█▌ | 4780/30000 [3:02:35<16:09:00, 2.31s/it] 16%|█▌ | 4780/30000 [3:02:35<16:09:00, 2.31s/it] 16%|█▌ | 4781/30000 [3:02:37<16:09:53, 2.31s/it] 16%|█▌ | 4782/30000 [3:02:39<16:00:47, 2.29s/it] 16%|█▌ | 4783/30000 [3:02:42<16:07:56, 2.30s/it] 16%|█▌ | 4784/30000 [3:02:44<16:07:09, 2.30s/it] 16%|█▌ | 4785/30000 [3:02:46<16:07:53, 2.30s/it] 16%|█▌ | 4786/30000 [3:02:48<16:13:13, 2.32s/it] 16%|█▌ | 4787/30000 [3:02:51<16:19:52, 2.33s/it] 16%|█▌ | 4788/30000 [3:02:53<16:18:37, 2.33s/it] 16%|█▌ | 4789/30000 [3:02:56<16:26:28, 2.35s/it] 16%|█▌ | 4790/30000 [3:02:58<16:17:01, 2.33s/it] 16%|█▌ | 4790/30000 [3:02:58<16:17:01, 2.33s/it] 16%|█▌ | 4791/30000 [3:03:00<16:20:06, 2.33s/it] 16%|█▌ | 4792/30000 [3:03:03<16:17:52, 2.33s/it] 16%|█▌ | 4793/30000 [3:03:05<16:13:27, 2.32s/it] 16%|█▌ | 4794/30000 [3:03:07<16:06:01, 2.30s/it] 16%|█▌ | 4795/30000 [3:03:09<16:15:19, 2.32s/it] 16%|█▌ | 4796/30000 [3:03:12<16:32:45, 2.36s/it] 16%|█▌ | 4797/30000 [3:03:14<16:22:10, 2.34s/it] 16%|█▌ | 4798/30000 [3:03:16<16:05:44, 2.30s/it] 16%|█▌ | 4799/30000 [3:03:19<16:12:24, 2.32s/it] 16%|█▌ | 4800/30000 [3:03:21<16:14:24, 2.32s/it] 16%|█▌ | 4800/30000 [3:03:21<16:14:24, 2.32s/it] 16%|█▌ | 4801/30000 [3:03:23<16:09:40, 2.31s/it] 16%|█▌ | 4802/30000 [3:03:26<16:02:44, 2.29s/it] 16%|█▌ | 4803/30000 [3:03:28<16:18:57, 2.33s/it] 16%|█▌ | 4804/30000 [3:03:30<16:15:16, 2.32s/it] 16%|█▌ | 4805/30000 [3:03:33<16:04:55, 2.30s/it] 16%|█▌ | 4806/30000 [3:03:35<16:05:46, 2.30s/it] 16%|█▌ | 4807/30000 [3:03:37<16:09:18, 2.31s/it] 16%|█▌ | 4808/30000 [3:03:39<16:03:17, 2.29s/it] 16%|█▌ | 4809/30000 [3:03:42<16:03:32, 2.29s/it] 16%|█▌ | 4810/30000 [3:03:44<16:00:45, 2.29s/it] 16%|█▌ | 4810/30000 [3:03:44<16:00:45, 2.29s/it] 16%|█▌ | 4811/30000 [3:03:46<15:56:50, 2.28s/it] 16%|█▌ | 4812/30000 [3:03:49<16:06:47, 2.30s/it] 16%|█▌ | 4813/30000 [3:03:51<16:03:53, 2.30s/it] 16%|█▌ | 4814/30000 [3:03:53<16:04:46, 2.30s/it] 16%|█▌ | 4815/30000 [3:03:56<16:04:55, 2.30s/it] 16%|█▌ | 4816/30000 [3:03:58<16:16:00, 2.33s/it] 16%|█▌ | 4817/30000 [3:04:00<16:05:20, 2.30s/it] 16%|█▌ | 4818/30000 [3:04:03<16:16:00, 2.33s/it] 16%|█▌ | 4819/30000 [3:04:05<16:17:17, 2.33s/it] 16%|█▌ | 4820/30000 [3:04:07<16:13:11, 2.32s/it] 16%|█▌ | 4820/30000 [3:04:07<16:13:11, 2.32s/it] 16%|█▌ | 4821/30000 [3:04:09<16:06:54, 2.30s/it] 16%|█▌ | 4822/30000 [3:04:12<16:09:11, 2.31s/it] 16%|█▌ | 4823/30000 [3:04:14<16:11:47, 2.32s/it] 16%|█▌ | 4824/30000 [3:04:16<15:44:37, 2.25s/it] 16%|█▌ | 4825/30000 [3:04:19<15:53:39, 2.27s/it] 16%|█▌ | 4826/30000 [3:04:21<15:54:38, 2.28s/it] 16%|█▌ | 4827/30000 [3:04:23<15:57:15, 2.28s/it] 16%|█▌ | 4828/30000 [3:04:25<15:58:30, 2.28s/it] 16%|█▌ | 4829/30000 [3:04:28<16:07:15, 2.31s/it] 16%|█▌ | 4830/30000 [3:04:30<16:03:53, 2.30s/it] 16%|█▌ | 4830/30000 [3:04:30<16:03:53, 2.30s/it] 16%|█▌ | 4831/30000 [3:04:32<15:57:34, 2.28s/it] 16%|█▌ | 4832/30000 [3:04:35<15:56:37, 2.28s/it] 16%|█▌ | 4833/30000 [3:04:37<15:38:40, 2.24s/it] 16%|█▌ | 4834/30000 [3:04:39<15:44:25, 2.25s/it] 16%|█▌ | 4835/30000 [3:04:41<15:48:00, 2.26s/it] 16%|█▌ | 4836/30000 [3:04:44<16:01:58, 2.29s/it] 16%|█▌ | 4837/30000 [3:04:46<15:59:23, 2.29s/it] 16%|█▌ | 4838/30000 [3:04:48<15:55:46, 2.28s/it] 16%|█▌ | 4839/30000 [3:04:51<16:05:04, 2.30s/it] 16%|█▌ | 4840/30000 [3:04:53<16:05:58, 2.30s/it] 16%|█▌ | 4840/30000 [3:04:53<16:05:58, 2.30s/it] 16%|█▌ | 4841/30000 [3:04:55<16:08:53, 2.31s/it] 16%|█▌ | 4842/30000 [3:04:57<16:01:49, 2.29s/it] 16%|█▌ | 4843/30000 [3:05:00<16:08:57, 2.31s/it] 16%|█▌ | 4844/30000 [3:05:02<16:25:22, 2.35s/it] 16%|█▌ | 4845/30000 [3:05:04<16:10:04, 2.31s/it] 16%|█▌ | 4846/30000 [3:05:07<16:10:24, 2.31s/it] 16%|█▌ | 4847/30000 [3:05:09<16:08:54, 2.31s/it] 16%|█▌ | 4848/30000 [3:05:11<16:05:05, 2.30s/it] 16%|█▌ | 4849/30000 [3:05:14<16:03:13, 2.30s/it] 16%|█▌ | 4850/30000 [3:05:16<16:07:28, 2.31s/it] 16%|█▌ | 4850/30000 [3:05:16<16:07:28, 2.31s/it] 16%|█▌ | 4851/30000 [3:05:18<16:05:02, 2.30s/it] 16%|█▌ | 4852/30000 [3:05:20<15:59:13, 2.29s/it] 16%|█▌ | 4853/30000 [3:05:23<15:55:10, 2.28s/it] 16%|█▌ | 4854/30000 [3:05:25<15:55:14, 2.28s/it] 16%|█▌ | 4855/30000 [3:05:27<15:58:32, 2.29s/it] 16%|█▌ | 4856/30000 [3:05:30<15:58:48, 2.29s/it] 16%|█▌ | 4857/30000 [3:05:32<16:07:28, 2.31s/it] 16%|█▌ | 4858/30000 [3:05:34<16:06:32, 2.31s/it] 16%|█▌ | 4859/30000 [3:05:37<15:58:05, 2.29s/it] 16%|█▌ | 4860/30000 [3:05:39<16:03:48, 2.30s/it] 16%|█▌ | 4860/30000 [3:05:39<16:03:48, 2.30s/it] 16%|█▌ | 4861/30000 [3:05:41<16:03:55, 2.30s/it] 16%|█▌ | 4862/30000 [3:05:43<15:58:12, 2.29s/it] 16%|█▌ | 4863/30000 [3:05:46<16:02:50, 2.30s/it] 16%|█▌ | 4864/30000 [3:05:48<15:56:48, 2.28s/it] 16%|█▌ | 4865/30000 [3:05:50<15:59:26, 2.29s/it] 16%|█▌ | 4866/30000 [3:05:53<15:58:16, 2.29s/it] 16%|█▌ | 4867/30000 [3:05:55<16:02:50, 2.30s/it] 16%|█▌ | 4868/30000 [3:05:57<15:56:26, 2.28s/it] 16%|█▌ | 4869/30000 [3:05:59<15:52:59, 2.28s/it] 16%|█▌ | 4870/30000 [3:06:02<15:56:16, 2.28s/it] 16%|█▌ | 4870/30000 [3:06:02<15:56:16, 2.28s/it] 16%|█▌ | 4871/30000 [3:06:04<16:00:12, 2.29s/it] 16%|█▌ | 4872/30000 [3:06:06<15:54:11, 2.28s/it] 16%|█▌ | 4873/30000 [3:06:09<15:59:33, 2.29s/it] 16%|█▌ | 4874/30000 [3:06:11<16:03:30, 2.30s/it] 16%|█▋ | 4875/30000 [3:06:13<16:03:21, 2.30s/it] 16%|█▋ | 4876/30000 [3:06:16<16:25:49, 2.35s/it] 16%|█▋ | 4877/30000 [3:06:18<16:15:17, 2.33s/it] 16%|█▋ | 4878/30000 [3:06:20<16:09:41, 2.32s/it] 16%|█▋ | 4879/30000 [3:06:23<16:04:28, 2.30s/it] 16%|█▋ | 4880/30000 [3:06:25<16:08:24, 2.31s/it] 16%|█▋ | 4880/30000 [3:06:25<16:08:24, 2.31s/it] 16%|█▋ | 4881/30000 [3:06:27<16:04:48, 2.30s/it] 16%|█▋ | 4882/30000 [3:06:30<16:15:27, 2.33s/it] 16%|█▋ | 4883/30000 [3:06:32<16:20:55, 2.34s/it] 16%|█▋ | 4884/30000 [3:06:34<16:13:17, 2.33s/it] 16%|█▋ | 4885/30000 [3:06:36<16:03:00, 2.30s/it] 16%|█▋ | 4886/30000 [3:06:39<16:03:04, 2.30s/it] 16%|█▋ | 4887/30000 [3:06:41<16:11:40, 2.32s/it] 16%|█▋ | 4888/30000 [3:06:43<16:00:26, 2.29s/it] 16%|█▋ | 4889/30000 [3:06:46<15:56:29, 2.29s/it] 16%|█▋ | 4890/30000 [3:06:48<16:04:38, 2.31s/it] 16%|█▋ | 4890/30000 [3:06:48<16:04:38, 2.31s/it] 16%|█▋ | 4891/30000 [3:06:50<16:03:58, 2.30s/it] 16%|█▋ | 4892/30000 [3:06:52<15:58:52, 2.29s/it] 16%|█▋ | 4893/30000 [3:06:55<16:00:46, 2.30s/it] 16%|█▋ | 4894/30000 [3:06:57<16:00:50, 2.30s/it] 16%|█▋ | 4895/30000 [3:06:59<15:58:21, 2.29s/it] 16%|█▋ | 4896/30000 [3:07:02<16:06:09, 2.31s/it] 16%|█▋ | 4897/30000 [3:07:04<16:02:10, 2.30s/it] 16%|█▋ | 4898/30000 [3:07:06<15:53:53, 2.28s/it] 16%|█▋ | 4899/30000 [3:07:09<16:08:22, 2.31s/it] 16%|█▋ | 4900/30000 [3:07:11<16:11:56, 2.32s/it] 16%|█▋ | 4900/30000 [3:07:11<16:11:56, 2.32s/it] 16%|█▋ | 4901/30000 [3:07:13<16:17:58, 2.34s/it] 16%|█▋ | 4902/30000 [3:07:16<16:14:26, 2.33s/it] 16%|█▋ | 4903/30000 [3:07:18<16:15:28, 2.33s/it] 16%|█▋ | 4904/30000 [3:07:20<16:14:16, 2.33s/it] 16%|█▋ | 4905/30000 [3:07:23<16:07:16, 2.31s/it] 16%|█▋ | 4906/30000 [3:07:25<16:15:15, 2.33s/it] 16%|█▋ | 4907/30000 [3:07:27<16:11:09, 2.32s/it] 16%|█▋ | 4908/30000 [3:07:30<16:05:04, 2.31s/it] 16%|█▋ | 4909/30000 [3:07:32<16:09:15, 2.32s/it] 16%|█▋ | 4910/30000 [3:07:34<16:04:47, 2.31s/it] 16%|█▋ | 4910/30000 [3:07:34<16:04:47, 2.31s/it] 16%|█▋ | 4911/30000 [3:07:36<15:59:03, 2.29s/it] 16%|█▋ | 4912/30000 [3:07:39<16:04:09, 2.31s/it] 16%|█▋ | 4913/30000 [3:07:41<16:03:04, 2.30s/it] 16%|█▋ | 4914/30000 [3:07:43<15:59:52, 2.30s/it] 16%|█▋ | 4915/30000 [3:07:46<15:55:20, 2.29s/it] 16%|█▋ | 4916/30000 [3:07:48<16:07:23, 2.31s/it] 16%|█▋ | 4917/30000 [3:07:50<16:02:15, 2.30s/it] 16%|█▋ | 4918/30000 [3:07:53<16:00:01, 2.30s/it] 16%|█▋ | 4919/30000 [3:07:55<16:02:02, 2.30s/it] 16%|█▋ | 4920/30000 [3:07:57<15:53:55, 2.28s/it] 16%|█▋ | 4920/30000 [3:07:57<15:53:55, 2.28s/it] 16%|█▋ | 4921/30000 [3:07:59<15:57:17, 2.29s/it] 16%|█▋ | 4922/30000 [3:08:02<16:03:52, 2.31s/it] 16%|█▋ | 4923/30000 [3:08:04<16:06:29, 2.31s/it] 16%|█▋ | 4924/30000 [3:08:06<15:58:29, 2.29s/it] 16%|█▋ | 4925/30000 [3:08:09<16:07:20, 2.31s/it] 16%|█▋ | 4926/30000 [3:08:11<16:09:19, 2.32s/it] 16%|█▋ | 4927/30000 [3:08:13<16:03:26, 2.31s/it] 16%|█▋ | 4928/30000 [3:08:16<16:03:57, 2.31s/it] 16%|█▋ | 4929/30000 [3:08:18<16:15:02, 2.33s/it] 16%|█▋ | 4930/30000 [3:08:20<15:55:09, 2.29s/it] 16%|█▋ | 4930/30000 [3:08:20<15:55:09, 2.29s/it] 16%|█▋ | 4931/30000 [3:08:22<15:58:57, 2.30s/it] 16%|█▋ | 4932/30000 [3:08:25<16:01:36, 2.30s/it] 16%|█▋ | 4933/30000 [3:08:27<15:57:10, 2.29s/it] 16%|█▋ | 4934/30000 [3:08:29<15:58:09, 2.29s/it] 16%|█▋ | 4935/30000 [3:08:32<16:00:35, 2.30s/it] 16%|█▋ | 4936/30000 [3:08:34<16:05:43, 2.31s/it] 16%|█▋ | 4937/30000 [3:08:36<16:05:05, 2.31s/it] 16%|█▋ | 4938/30000 [3:08:39<16:08:18, 2.32s/it] 16%|█▋ | 4939/30000 [3:08:41<16:06:26, 2.31s/it] 16%|█▋ | 4940/30000 [3:08:43<15:40:41, 2.25s/it] 16%|█▋ | 4940/30000 [3:08:43<15:40:41, 2.25s/it] 16%|█▋ | 4941/30000 [3:08:45<15:48:34, 2.27s/it] 16%|█▋ | 4942/30000 [3:08:48<15:46:49, 2.27s/it] 16%|█▋ | 4943/30000 [3:08:50<15:46:37, 2.27s/it] 16%|█▋ | 4944/30000 [3:08:52<15:57:34, 2.29s/it] 16%|█▋ | 4945/30000 [3:08:55<16:01:39, 2.30s/it] 16%|█▋ | 4946/30000 [3:08:57<15:52:28, 2.28s/it] 16%|█▋ | 4947/30000 [3:08:59<16:09:49, 2.32s/it] 16%|█▋ | 4948/30000 [3:09:02<16:11:55, 2.33s/it] 16%|█▋ | 4949/30000 [3:09:04<16:04:10, 2.31s/it] 16%|█▋ | 4950/30000 [3:09:06<15:58:14, 2.30s/it] 16%|█▋ | 4950/30000 [3:09:06<15:58:14, 2.30s/it] 17%|█▋ | 4951/30000 [3:09:08<15:59:15, 2.30s/it] 17%|█▋ | 4952/30000 [3:09:11<15:57:05, 2.29s/it] 17%|█▋ | 4953/30000 [3:09:13<15:54:50, 2.29s/it] 17%|█▋ | 4954/30000 [3:09:15<15:46:30, 2.27s/it] 17%|█▋ | 4955/30000 [3:09:18<15:54:21, 2.29s/it] 17%|█▋ | 4956/30000 [3:09:20<15:47:15, 2.27s/it] 17%|█▋ | 4957/30000 [3:09:22<15:55:28, 2.29s/it] 17%|█▋ | 4958/30000 [3:09:24<15:50:13, 2.28s/it] 17%|█▋ | 4959/30000 [3:09:27<15:41:44, 2.26s/it] 17%|█▋ | 4960/30000 [3:09:29<15:46:12, 2.27s/it] 17%|█▋ | 4960/30000 [3:09:29<15:46:12, 2.27s/it] 17%|█▋ | 4961/30000 [3:09:31<15:51:45, 2.28s/it] 17%|█▋ | 4962/30000 [3:09:33<15:52:05, 2.28s/it] 17%|█▋ | 4963/30000 [3:09:36<16:05:26, 2.31s/it] 17%|█▋ | 4964/30000 [3:09:38<16:18:36, 2.35s/it] 17%|█▋ | 4965/30000 [3:09:41<16:07:53, 2.32s/it] 17%|█▋ | 4966/30000 [3:09:43<16:01:27, 2.30s/it] 17%|█▋ | 4967/30000 [3:09:45<15:55:23, 2.29s/it] 17%|█▋ | 4968/30000 [3:09:47<15:59:01, 2.30s/it] 17%|█▋ | 4969/30000 [3:09:50<15:57:15, 2.29s/it] 17%|█▋ | 4970/30000 [3:09:52<15:58:29, 2.30s/it] 17%|█▋ | 4970/30000 [3:09:52<15:58:29, 2.30s/it] 17%|█▋ | 4971/30000 [3:09:54<15:56:22, 2.29s/it] 17%|█▋ | 4972/30000 [3:09:56<15:49:59, 2.28s/it] 17%|█▋ | 4973/30000 [3:09:59<16:03:08, 2.31s/it] 17%|█▋ | 4974/30000 [3:10:01<16:05:01, 2.31s/it] 17%|█▋ | 4975/30000 [3:10:03<16:05:49, 2.32s/it] 17%|█▋ | 4976/30000 [3:10:06<15:42:15, 2.26s/it] 17%|█▋ | 4977/30000 [3:10:08<15:57:28, 2.30s/it] 17%|█▋ | 4978/30000 [3:10:10<15:58:42, 2.30s/it] 17%|█▋ | 4979/30000 [3:10:13<15:51:01, 2.28s/it] 17%|█▋ | 4980/30000 [3:10:15<15:52:56, 2.29s/it] 17%|█▋ | 4980/30000 [3:10:15<15:52:56, 2.29s/it] 17%|█▋ | 4981/30000 [3:10:17<15:55:52, 2.29s/it] 17%|█▋ | 4982/30000 [3:10:19<15:47:53, 2.27s/it] 17%|█▋ | 4983/30000 [3:10:22<15:48:04, 2.27s/it] 17%|█▋ | 4984/30000 [3:10:24<15:43:30, 2.26s/it] 17%|█▋ | 4985/30000 [3:10:26<15:07:54, 2.18s/it] 17%|█▋ | 4986/30000 [3:10:28<15:07:56, 2.18s/it] 17%|█▋ | 4987/30000 [3:10:29<13:32:24, 1.95s/it] 17%|█▋ | 4988/30000 [3:10:32<14:22:17, 2.07s/it] 17%|█▋ | 4989/30000 [3:10:34<14:46:55, 2.13s/it] 17%|█▋ | 4990/30000 [3:10:36<15:12:31, 2.19s/it] 17%|█▋ | 4990/30000 [3:10:36<15:12:31, 2.19s/it] 17%|█▋ | 4991/30000 [3:10:39<15:31:25, 2.23s/it] 17%|█▋ | 4992/30000 [3:10:41<15:30:20, 2.23s/it] 17%|█▋ | 4993/30000 [3:10:43<15:41:36, 2.26s/it] 17%|█▋ | 4994/30000 [3:10:46<15:45:41, 2.27s/it] 17%|█▋ | 4995/30000 [3:10:49<18:47:35, 2.71s/it] 17%|█▋ | 4996/30000 [3:10:53<20:08:04, 2.90s/it] 17%|█▋ | 4997/30000 [3:10:54<16:39:50, 2.40s/it] 17%|█▋ | 4998/30000 [3:10:56<16:26:46, 2.37s/it] 17%|█▋ | 4999/30000 [3:10:59<16:23:31, 2.36s/it] 17%|█▋ | 5000/30000 [3:11:01<16:11:12, 2.33s/it] 17%|█▋ | 5000/30000 [3:11:01<16:11:12, 2.33s/it] 17%|█▋ | 5001/30000 [3:11:03<16:12:49, 2.33s/it] 17%|█▋ | 5002/30000 [3:11:05<15:09:04, 2.18s/it] 17%|█▋ | 5003/30000 [3:11:10<22:05:11, 3.18s/it] 17%|█▋ | 5004/30000 [3:11:14<23:12:00, 3.34s/it] 17%|█▋ | 5005/30000 [3:11:16<20:20:30, 2.93s/it] 17%|█▋ | 5006/30000 [3:11:17<16:43:23, 2.41s/it] 17%|█▋ | 5007/30000 [3:11:20<17:24:24, 2.51s/it] 17%|█▋ | 5008/30000 [3:11:22<16:47:45, 2.42s/it] 17%|█▋ | 5009/30000 [3:11:25<16:36:02, 2.39s/it] 17%|█▋ | 5010/30000 [3:11:26<15:27:43, 2.23s/it] 17%|█▋ | 5010/30000 [3:11:27<15:27:43, 2.23s/it] 17%|█▋ | 5011/30000 [3:11:33<23:44:20, 3.42s/it] 17%|█▋ | 5012/30000 [3:11:37<25:52:28, 3.73s/it] 17%|█▋ | 5013/30000 [3:11:38<20:38:01, 2.97s/it] 17%|█▋ | 5014/30000 [3:11:40<18:14:40, 2.63s/it] 17%|█▋ | 5015/30000 [3:11:42<16:43:51, 2.41s/it] 17%|█▋ | 5016/30000 [3:11:44<16:30:37, 2.38s/it] 17%|█▋ | 5017/30000 [3:11:47<16:15:02, 2.34s/it] 17%|█▋ | 5018/30000 [3:11:49<16:06:11, 2.32s/it] 17%|█▋ | 5019/30000 [3:11:54<21:34:58, 3.11s/it] 17%|█▋ | 5020/30000 [3:11:58<23:42:25, 3.42s/it] 17%|█▋ | 5020/30000 [3:11:58<23:42:25, 3.42s/it] 17%|█▋ | 5021/30000 [3:12:00<21:38:24, 3.12s/it] 17%|█▋ | 5022/30000 [3:12:02<18:52:52, 2.72s/it] 17%|█▋ | 5023/30000 [3:12:04<16:50:48, 2.43s/it] 17%|█▋ | 5024/30000 [3:12:06<16:36:26, 2.39s/it] 17%|█▋ | 5025/30000 [3:12:09<16:35:01, 2.39s/it] 17%|█▋ | 5026/30000 [3:12:11<16:13:40, 2.34s/it] 17%|█▋ | 5027/30000 [3:12:16<21:35:13, 3.11s/it] 17%|█▋ | 5028/30000 [3:12:20<23:09:12, 3.34s/it] 17%|█▋ | 5029/30000 [3:12:22<20:52:56, 3.01s/it] 17%|█▋ | 5030/30000 [3:12:24<18:58:53, 2.74s/it] 17%|█▋ | 5030/30000 [3:12:24<18:58:53, 2.74s/it] 17%|█▋ | 5031/30000 [3:12:26<16:30:52, 2.38s/it] 17%|█▋ | 5032/30000 [3:12:28<16:09:26, 2.33s/it] 17%|█▋ | 5033/30000 [3:12:29<14:52:34, 2.15s/it] 17%|█▋ | 5034/30000 [3:12:31<14:25:15, 2.08s/it] 17%|█▋ | 5035/30000 [3:12:39<26:53:25, 3.88s/it] 17%|█▋ | 5036/30000 [3:12:43<26:56:36, 3.89s/it] 17%|█▋ | 5037/30000 [3:12:46<23:35:55, 3.40s/it] 17%|█▋ | 5038/30000 [3:12:48<21:19:39, 3.08s/it] 17%|█▋ | 5039/30000 [3:12:49<17:34:45, 2.54s/it] 17%|█▋ | 5040/30000 [3:12:52<17:06:46, 2.47s/it] +{'loss': 0.0179, 'grad_norm': 0.07294467836618423, 'learning_rate': 2.911656900483265e-05, 'flow_matching_loss': 0.01794936563819647, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.35} +{'loss': 0.0185, 'grad_norm': 0.07920266687870026, 'learning_rate': 2.9110969790766322e-05, 'flow_matching_loss': 0.018472930509597064, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.35} +{'loss': 0.0178, 'grad_norm': 0.0815187469124794, 'learning_rate': 2.910535343054112e-05, 'flow_matching_loss': 0.01784668294712901, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.35} +{'loss': 0.0167, 'grad_norm': 0.11634989827871323, 'learning_rate': 2.909971993098145e-05, 'flow_matching_loss': 0.016710187681019305, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.36} +{'loss': 0.0196, 'grad_norm': 0.09780547767877579, 'learning_rate': 2.9094069298932552e-05, 'flow_matching_loss': 0.019587599392980336, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.36} +{'loss': 0.0185, 'grad_norm': 0.10789816826581955, 'learning_rate': 2.908840154126047e-05, 'flow_matching_loss': 0.01849818490445614, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.36} +{'loss': 0.0153, 'grad_norm': 0.062022749334573746, 'learning_rate': 2.9082716664852063e-05, 'flow_matching_loss': 0.01533505255356431, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.36} +{'loss': 0.0182, 'grad_norm': 0.0765312984585762, 'learning_rate': 2.9077014676615e-05, 'flow_matching_loss': 0.018212305568158626, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.37} +{'loss': 0.0168, 'grad_norm': 0.09144172817468643, 'learning_rate': 2.9071295583477724e-05, 'flow_matching_loss': 0.016794453840702772, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.37} +{'loss': 0.0164, 'grad_norm': 0.08697270601987839, 'learning_rate': 2.906555939238948e-05, 'flow_matching_loss': 0.016384642105549574, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.37} +{'loss': 0.0179, 'grad_norm': 0.08140653371810913, 'learning_rate': 2.905980611032028e-05, 'flow_matching_loss': 0.017887113709002735, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.38} +{'loss': 0.018, 'grad_norm': 0.0906154066324234, 'learning_rate': 2.9054035744260904e-05, 'flow_matching_loss': 0.018021827936172484, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.38} +{'loss': 0.0165, 'grad_norm': 0.08482995629310608, 'learning_rate': 2.9048248301222885e-05, 'flow_matching_loss': 0.016529027745127677, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.38} +{'loss': 0.018, 'grad_norm': 0.07993477582931519, 'learning_rate': 2.904244378823852e-05, 'flow_matching_loss': 0.017980134487152098, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.38} +{'loss': 0.0158, 'grad_norm': 0.09419599175453186, 'learning_rate': 2.9036622212360836e-05, 'flow_matching_loss': 0.015756020601838828, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.39} +{'loss': 0.017, 'grad_norm': 0.08274397999048233, 'learning_rate': 2.9030783580663596e-05, 'flow_matching_loss': 0.017025388684123753, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.39} +{'loss': 0.0173, 'grad_norm': 0.12208238989114761, 'learning_rate': 2.902492790024129e-05, 'flow_matching_loss': 0.0172802759334445, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.39} +{'loss': 0.0185, 'grad_norm': 0.09283638745546341, 'learning_rate': 2.901905517820912e-05, 'flow_matching_loss': 0.01849865010008216, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.4} +{'loss': 0.0165, 'grad_norm': 0.08133002370595932, 'learning_rate': 2.9013165421703e-05, 'flow_matching_loss': 0.0165410072542727, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.4} +{'loss': 0.0186, 'grad_norm': 0.08000727742910385, 'learning_rate': 2.9007258637879534e-05, 'flow_matching_loss': 0.01856268374249339, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.4} +{'loss': 0.0167, 'grad_norm': 0.0753961056470871, 'learning_rate': 2.9001334833916025e-05, 'flow_matching_loss': 0.016700625605881213, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.41} +{'loss': 0.0167, 'grad_norm': 0.08640418946743011, 'learning_rate': 2.899539401701046e-05, 'flow_matching_loss': 0.016661651153117418, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.41} +{'loss': 0.0171, 'grad_norm': 0.09864164888858795, 'learning_rate': 2.8989436194381474e-05, 'flow_matching_loss': 0.017137428838759662, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.41} +{'loss': 0.0177, 'grad_norm': 0.08597861230373383, 'learning_rate': 2.8983461373268404e-05, 'flow_matching_loss': 0.017694329656660557, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.41} +{'loss': 0.0175, 'grad_norm': 0.10314324498176575, 'learning_rate': 2.8977469560931205e-05, 'flow_matching_loss': 0.01748534757643938, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.42} +{'loss': 0.0168, 'grad_norm': 0.07918404042720795, 'learning_rate': 2.8971460764650502e-05, 'flow_matching_loss': 0.01680967016145587, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.42} +{'loss': 0.0177, 'grad_norm': 0.07928013056516647, 'learning_rate': 2.8965434991727545e-05, 'flow_matching_loss': 0.01772168409079313, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.42} +{'loss': 0.0173, 'grad_norm': 0.08083848655223846, 'learning_rate': 2.895939224948422e-05, 'flow_matching_loss': 0.01727563962340355, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.43} +{'loss': 0.0151, 'grad_norm': 0.06903116405010223, 'learning_rate': 2.895333254526303e-05, 'flow_matching_loss': 0.01505086775869131, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.43} +{'loss': 0.0174, 'grad_norm': 0.09588684141635895, 'learning_rate': 2.8947255886427084e-05, 'flow_matching_loss': 0.017383875139057637, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.43} +{'loss': 0.0179, 'grad_norm': 0.08520248532295227, 'learning_rate': 2.8941162280360095e-05, 'flow_matching_loss': 0.017861897218972446, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.43} +{'loss': 0.018, 'grad_norm': 0.058098725974559784, 'learning_rate': 2.8935051734466368e-05, 'flow_matching_loss': 0.018031459674239157, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.44} +{'loss': 0.0181, 'grad_norm': 0.08408401161432266, 'learning_rate': 2.8928924256170792e-05, 'flow_matching_loss': 0.01812290884554386, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.44} +{'loss': 0.0184, 'grad_norm': 0.09920460730791092, 'learning_rate': 2.8922779852918836e-05, 'flow_matching_loss': 0.018396809231489897, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.44} +{'loss': 0.019, 'grad_norm': 0.09478297829627991, 'learning_rate': 2.8916618532176526e-05, 'flow_matching_loss': 0.019034580700099467, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.45} +{'loss': 0.0168, 'grad_norm': 0.10513750463724136, 'learning_rate': 2.891044030143045e-05, 'flow_matching_loss': 0.01682168859988451, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.45} +{'loss': 0.016, 'grad_norm': 0.1039203405380249, 'learning_rate': 2.890424516818774e-05, 'flow_matching_loss': 0.01601092265918851, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.45} +{'loss': 0.017, 'grad_norm': 0.0710081160068512, 'learning_rate': 2.8898033139976057e-05, 'flow_matching_loss': 0.017031166702508926, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.45} +{'loss': 0.0169, 'grad_norm': 0.12970642745494843, 'learning_rate': 2.889180422434362e-05, 'flow_matching_loss': 0.016943196207284926, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.46} +{'loss': 0.0165, 'grad_norm': 0.08096943795681, 'learning_rate': 2.8885558428859136e-05, 'flow_matching_loss': 0.016532422881573437, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.46} +{'loss': 0.0163, 'grad_norm': 0.08339137583971024, 'learning_rate': 2.8879295761111835e-05, 'flow_matching_loss': 0.01627649664878845, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.46} +{'loss': 0.0184, 'grad_norm': 0.10386183857917786, 'learning_rate': 2.8873016228711452e-05, 'flow_matching_loss': 0.01844171164557338, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.47} 17%|█▋ | 5040/30000 [3:12:52<17:06:46, 2.47s/it] 17%|█▋ | 5041/30000 [3:12:54<16:42:25, 2.41s/it] 17%|█▋ | 5042/30000 [3:12:55<14:47:31, 2.13s/it] 17%|█▋ | 5043/30000 [3:13:01<22:25:34, 3.23s/it] 17%|█▋ | 5044/30000 [3:13:06<26:24:01, 3.81s/it] 17%|█▋ | 5045/30000 [3:13:08<22:53:10, 3.30s/it] 17%|█▋ | 5046/30000 [3:13:11<20:44:33, 2.99s/it] 17%|█▋ | 5047/30000 [3:13:13<19:11:23, 2.77s/it] 17%|█▋ | 5048/30000 [3:13:15<17:29:46, 2.52s/it] 17%|█▋ | 5049/30000 [3:13:17<17:04:59, 2.46s/it] 17%|█▋ | 5050/30000 [3:13:19<16:35:11, 2.39s/it] 17%|█▋ | 5050/30000 [3:13:19<16:35:11, 2.39s/it] 17%|█▋ | 5051/30000 [3:13:22<16:48:34, 2.43s/it] 17%|█▋ | 5052/30000 [3:13:27<23:17:21, 3.36s/it] 17%|█▋ | 5053/30000 [3:13:30<21:04:13, 3.04s/it] 17%|█▋ | 5054/30000 [3:13:31<18:23:48, 2.65s/it] 17%|█▋ | 5055/30000 [3:13:34<17:28:11, 2.52s/it] 17%|█▋ | 5056/30000 [3:13:36<16:57:12, 2.45s/it] 17%|█▋ | 5057/30000 [3:13:38<16:13:44, 2.34s/it] 17%|█▋ | 5058/30000 [3:13:40<16:05:16, 2.32s/it] 17%|█▋ | 5059/30000 [3:13:43<17:04:21, 2.46s/it] 17%|█▋ | 5060/30000 [3:13:48<23:02:49, 3.33s/it] 17%|█▋ | 5060/30000 [3:13:49<23:02:49, 3.33s/it] 17%|█▋ | 5061/30000 [3:13:51<20:50:55, 3.01s/it] 17%|█▋ | 5062/30000 [3:13:53<19:15:45, 2.78s/it] 17%|█▋ | 5063/30000 [3:13:54<16:26:42, 2.37s/it] 17%|█▋ | 5064/30000 [3:13:57<16:08:28, 2.33s/it] 17%|█▋ | 5065/30000 [3:13:59<15:56:30, 2.30s/it] 17%|█▋ | 5066/30000 [3:14:01<15:17:15, 2.21s/it] 17%|█▋ | 5067/30000 [3:14:06<20:40:08, 2.98s/it] 17%|█▋ | 5068/30000 [3:14:11<25:05:47, 3.62s/it] 17%|█▋ | 5069/30000 [3:14:12<20:39:18, 2.98s/it] 17%|█▋ | 5070/30000 [3:14:15<19:18:25, 2.79s/it] 17%|█▋ | 5070/30000 [3:14:15<19:18:25, 2.79s/it] 17%|█▋ | 5071/30000 [3:14:17<18:14:58, 2.64s/it] 17%|█▋ | 5072/30000 [3:14:19<17:27:13, 2.52s/it] 17%|█▋ | 5073/30000 [3:14:21<17:01:27, 2.46s/it] 17%|█▋ | 5074/30000 [3:14:24<16:28:34, 2.38s/it] 17%|█▋ | 5075/30000 [3:14:26<15:58:53, 2.31s/it] 17%|█▋ | 5076/30000 [3:14:28<16:05:32, 2.32s/it] 17%|█▋ | 5077/30000 [3:14:30<15:52:47, 2.29s/it] 17%|█▋ | 5078/30000 [3:14:33<15:48:13, 2.28s/it] 17%|█▋ | 5079/30000 [3:14:35<15:55:15, 2.30s/it] 17%|█▋ | 5080/30000 [3:14:37<15:57:11, 2.30s/it] 17%|█▋ | 5080/30000 [3:14:37<15:57:11, 2.30s/it] 17%|█▋ | 5081/30000 [3:14:39<15:33:18, 2.25s/it] 17%|█▋ | 5082/30000 [3:14:42<15:41:35, 2.27s/it] 17%|█▋ | 5083/30000 [3:14:45<18:05:55, 2.61s/it] 17%|█▋ | 5084/30000 [3:14:51<23:54:16, 3.45s/it] 17%|█▋ | 5085/30000 [3:14:53<20:55:19, 3.02s/it] 17%|█▋ | 5086/30000 [3:14:55<19:06:18, 2.76s/it] 17%|█▋ | 5087/30000 [3:14:57<17:58:27, 2.60s/it] 17%|█▋ | 5088/30000 [3:14:59<16:08:35, 2.33s/it] 17%|█▋ | 5089/30000 [3:15:01<15:11:28, 2.20s/it] 17%|█▋ | 5090/30000 [3:15:03<15:26:02, 2.23s/it] 17%|█▋ | 5090/30000 [3:15:03<15:26:02, 2.23s/it] 17%|█▋ | 5091/30000 [3:15:09<23:14:07, 3.36s/it] 17%|█▋ | 5092/30000 [3:15:12<22:24:08, 3.24s/it] 17%|█▋ | 5093/30000 [3:15:14<20:30:04, 2.96s/it] 17%|█▋ | 5094/30000 [3:15:16<18:33:06, 2.68s/it] 17%|█▋ | 5095/30000 [3:15:19<19:44:05, 2.85s/it] 17%|█▋ | 5096/30000 [3:15:22<18:24:28, 2.66s/it] 17%|█▋ | 5097/30000 [3:15:24<16:54:30, 2.44s/it] 17%|█▋ | 5098/30000 [3:15:26<16:35:17, 2.40s/it] 17%|█▋ | 5099/30000 [3:15:30<20:02:00, 2.90s/it] 17%|█▋ | 5100/30000 [3:15:35<23:51:35, 3.45s/it] 17%|█▋ | 5100/30000 [3:15:35<23:51:35, 3.45s/it] 17%|█▋ | 5101/30000 [3:15:36<19:10:56, 2.77s/it] 17%|█▋ | 5102/30000 [3:15:38<17:17:01, 2.50s/it] 17%|█▋ | 5103/30000 [3:15:42<21:39:55, 3.13s/it] 17%|█▋ | 5104/30000 [3:15:45<19:49:58, 2.87s/it] 17%|█▋ | 5105/30000 [3:15:47<18:40:37, 2.70s/it] 17%|█▋ | 5106/30000 [3:15:49<17:46:01, 2.57s/it] 17%|█▋ | 5107/30000 [3:15:51<17:11:58, 2.49s/it] 17%|█▋ | 5108/30000 [3:15:56<20:51:35, 3.02s/it] 17%|█▋ | 5109/30000 [3:15:58<19:21:09, 2.80s/it] 17%|█▋ | 5110/30000 [3:16:00<18:31:59, 2.68s/it] 17%|█▋ | 5110/30000 [3:16:00<18:31:59, 2.68s/it] 17%|█▋ | 5111/30000 [3:16:03<18:03:10, 2.61s/it] 17%|█▋ | 5112/30000 [3:16:05<17:17:03, 2.50s/it] 17%|█▋ | 5113/30000 [3:16:07<16:52:46, 2.44s/it] 17%|█▋ | 5114/30000 [3:16:09<15:51:29, 2.29s/it] 17%|█▋ | 5115/30000 [3:16:13<18:05:44, 2.62s/it] 17%|█▋ | 5116/30000 [3:16:17<22:20:39, 3.23s/it] 17%|█▋ | 5117/30000 [3:16:20<20:20:12, 2.94s/it] 17%|█▋ | 5118/30000 [3:16:21<17:05:38, 2.47s/it] 17%|█▋ | 5119/30000 [3:16:25<20:54:22, 3.02s/it] 17%|█▋ | 5120/30000 [3:16:28<19:21:19, 2.80s/it] 17%|█▋ | 5120/30000 [3:16:28<19:21:19, 2.80s/it] 17%|█▋ | 5121/30000 [3:16:29<17:14:47, 2.50s/it] 17%|█▋ | 5122/30000 [3:16:31<15:54:33, 2.30s/it] 17%|█▋ | 5123/30000 [3:16:33<15:16:51, 2.21s/it] 17%|█▋ | 5124/30000 [3:16:40<24:14:32, 3.51s/it] 17%|█▋ | 5125/30000 [3:16:41<20:22:53, 2.95s/it] 17%|█▋ | 5126/30000 [3:16:43<18:22:14, 2.66s/it] 17%|█▋ | 5127/30000 [3:16:48<21:28:27, 3.11s/it] 17%|█▋ | 5128/30000 [3:16:50<19:48:26, 2.87s/it] 17%|█▋ | 5129/30000 [3:16:52<18:31:16, 2.68s/it] 17%|█▋ | 5130/30000 [3:16:54<17:40:16, 2.56s/it] 17%|█▋ | 5130/30000 [3:16:54<17:40:16, 2.56s/it] 17%|█▋ | 5131/30000 [3:16:57<17:08:01, 2.48s/it] 17%|█▋ | 5132/30000 [3:16:59<16:48:57, 2.43s/it] 17%|█▋ | 5133/30000 [3:17:01<16:25:25, 2.38s/it] 17%|█▋ | 5134/30000 [3:17:03<16:12:30, 2.35s/it] 17%|█▋ | 5135/30000 [3:17:07<19:08:06, 2.77s/it] 17%|█▋ | 5136/30000 [3:17:10<18:12:05, 2.64s/it] 17%|█▋ | 5137/30000 [3:17:12<17:26:09, 2.52s/it] 17%|█▋ | 5138/30000 [3:17:14<15:51:15, 2.30s/it] 17%|█▋ | 5139/30000 [3:17:15<13:34:35, 1.97s/it] 17%|█▋ | 5140/30000 [3:17:21<22:30:02, 3.26s/it] 17%|█▋ | 5140/30000 [3:17:21<22:30:02, 3.26s/it] 17%|█▋ | 5141/30000 [3:17:23<19:24:31, 2.81s/it] 17%|█▋ | 5142/30000 [3:17:25<18:23:29, 2.66s/it] 17%|█▋ | 5143/30000 [3:17:31<25:27:50, 3.69s/it] 17%|█▋ | 5144/30000 [3:17:33<22:28:08, 3.25s/it] 17%|█▋ | 5145/30000 [3:17:36<20:31:07, 2.97s/it] 17%|█▋ | 5146/30000 [3:17:38<19:10:24, 2.78s/it] 17%|█▋ | 5147/30000 [3:17:40<18:03:15, 2.62s/it] 17%|█▋ | 5148/30000 [3:17:43<17:22:26, 2.52s/it] 17%|█▋ | 5149/30000 [3:17:45<16:48:55, 2.44s/it] 17%|█▋ | 5150/30000 [3:17:47<16:24:31, 2.38s/it] 17%|█▋ | 5150/30000 [3:17:47<16:24:31, 2.38s/it] 17%|█▋ | 5151/30000 [3:17:49<16:13:27, 2.35s/it] 17%|█▋ | 5152/30000 [3:17:52<16:08:24, 2.34s/it] 17%|█▋ | 5153/30000 [3:17:54<16:05:17, 2.33s/it] 17%|█▋ | 5154/30000 [3:17:56<15:58:20, 2.31s/it] 17%|█▋ | 5155/30000 [3:17:59<15:53:08, 2.30s/it] 17%|█▋ | 5156/30000 [3:18:01<16:00:28, 2.32s/it] 17%|█▋ | 5157/30000 [3:18:03<15:51:15, 2.30s/it] 17%|█▋ | 5158/30000 [3:18:05<15:48:20, 2.29s/it] 17%|█▋ | 5159/30000 [3:18:08<15:52:28, 2.30s/it] 17%|█▋ | 5160/30000 [3:18:10<15:51:49, 2.30s/it] 17%|█▋ | 5160/30000 [3:18:10<15:51:49, 2.30s/it] 17%|█▋ | 5161/30000 [3:18:12<15:59:30, 2.32s/it] 17%|█▋ | 5162/30000 [3:18:15<15:51:41, 2.30s/it] 17%|█▋ | 5163/30000 [3:18:17<15:54:45, 2.31s/it] 17%|█▋ | 5164/30000 [3:18:19<15:51:47, 2.30s/it] 17%|█▋ | 5165/30000 [3:18:22<15:50:42, 2.30s/it] 17%|█▋ | 5166/30000 [3:18:24<15:52:01, 2.30s/it] 17%|█▋ | 5167/30000 [3:18:26<15:50:18, 2.30s/it] 17%|█▋ | 5168/30000 [3:18:28<15:44:18, 2.28s/it] 17%|█▋ | 5169/30000 [3:18:31<15:47:49, 2.29s/it] 17%|█▋ | 5170/30000 [3:18:33<16:05:56, 2.33s/it] 17%|█▋ | 5170/30000 [3:18:33<16:05:56, 2.33s/it] 17%|█▋ | 5171/30000 [3:18:35<16:01:44, 2.32s/it] 17%|█▋ | 5172/30000 [3:18:38<15:55:27, 2.31s/it] 17%|█▋ | 5173/30000 [3:18:40<15:49:15, 2.29s/it] 17%|█▋ | 5174/30000 [3:18:42<15:54:04, 2.31s/it] 17%|█▋ | 5175/30000 [3:18:45<15:53:45, 2.31s/it] 17%|█▋ | 5176/30000 [3:18:47<15:49:36, 2.30s/it] 17%|█▋ | 5177/30000 [3:18:49<15:49:45, 2.30s/it] 17%|█▋ | 5178/30000 [3:18:52<15:52:14, 2.30s/it] 17%|█▋ | 5179/30000 [3:18:54<15:56:17, 2.31s/it] 17%|█▋ | 5180/30000 [3:18:56<15:51:50, 2.30s/it] 17%|█▋ | 5180/30000 [3:18:56<15:51:50, 2.30s/it] 17%|█▋ | 5181/30000 [3:18:58<15:49:24, 2.30s/it] 17%|█▋ | 5182/30000 [3:19:01<15:55:31, 2.31s/it] 17%|█▋ | 5183/30000 [3:19:03<15:48:17, 2.29s/it] 17%|█▋ | 5184/30000 [3:19:05<15:47:18, 2.29s/it] 17%|█▋ | 5185/30000 [3:19:08<15:53:24, 2.31s/it] 17%|█▋ | 5186/30000 [3:19:10<15:52:04, 2.30s/it] 17%|█▋ | 5187/30000 [3:19:12<15:38:06, 2.27s/it] 17%|█▋ | 5188/30000 [3:19:14<15:41:27, 2.28s/it] 17%|█▋ | 5189/30000 [3:19:17<15:41:47, 2.28s/it] 17%|█▋ | 5190/30000 [3:19:19<15:34:36, 2.26s/it] 17%|█▋ | 5190/30000 [3:19:19<15:34:36, 2.26s/it] 17%|█▋ | 5191/30000 [3:19:21<15:37:36, 2.27s/it] 17%|█▋ | 5192/30000 [3:19:24<15:45:49, 2.29s/it] 17%|█▋ | 5193/30000 [3:19:26<15:49:26, 2.30s/it] 17%|█▋ | 5194/30000 [3:19:28<15:50:24, 2.30s/it] 17%|█▋ | 5195/30000 [3:19:30<15:43:18, 2.28s/it] 17%|█▋ | 5196/30000 [3:19:33<15:44:19, 2.28s/it] 17%|█▋ | 5197/30000 [3:19:35<15:45:54, 2.29s/it] 17%|█▋ | 5198/30000 [3:19:37<15:46:52, 2.29s/it] 17%|█▋ | 5199/30000 [3:19:40<15:50:03, 2.30s/it] 17%|█▋ | 5200/30000 [3:19:42<15:48:19, 2.29s/it] 17%|█▋ | 5200/30000 [3:19:42<15:48:19, 2.29s/it] 17%|█▋ | 5201/30000 [3:19:44<15:47:33, 2.29s/it] 17%|█▋ | 5202/30000 [3:19:47<15:51:48, 2.30s/it] 17%|█▋ | 5203/30000 [3:19:49<15:57:29, 2.32s/it] 17%|█▋ | 5204/30000 [3:19:51<15:51:36, 2.30s/it] 17%|█▋ | 5205/30000 [3:19:53<15:56:07, 2.31s/it] 17%|█▋ | 5206/30000 [3:19:56<15:50:36, 2.30s/it] 17%|█▋ | 5207/30000 [3:19:58<15:48:34, 2.30s/it] 17%|█▋ | 5208/30000 [3:20:00<15:44:01, 2.28s/it] 17%|█▋ | 5209/30000 [3:20:03<15:53:54, 2.31s/it] 17%|█▋ | 5210/30000 [3:20:05<15:52:00, 2.30s/it] 17%|█▋ | 5210/30000 [3:20:05<15:52:00, 2.30s/it] 17%|█▋ | 5211/30000 [3:20:07<15:47:33, 2.29s/it] 17%|█▋ | 5212/30000 [3:20:10<15:49:44, 2.30s/it] 17%|█▋ | 5213/30000 [3:20:12<15:51:00, 2.30s/it] 17%|█▋ | 5214/30000 [3:20:14<15:49:08, 2.30s/it] 17%|█▋ | 5215/30000 [3:20:16<15:46:51, 2.29s/it] 17%|█▋ | 5216/30000 [3:20:19<15:51:49, 2.30s/it] 17%|█▋ | 5217/30000 [3:20:21<15:47:41, 2.29s/it] 17%|█▋ | 5218/30000 [3:20:23<15:47:31, 2.29s/it] 17%|█▋ | 5219/30000 [3:20:26<15:51:04, 2.30s/it] 17%|█▋ | 5220/30000 [3:20:28<15:49:31, 2.30s/it] 17%|█▋ | 5220/30000 [3:20:28<15:49:31, 2.30s/it] 17%|█▋ | 5221/30000 [3:20:30<15:46:53, 2.29s/it] 17%|█▋ | 5222/30000 [3:20:32<15:45:43, 2.29s/it] 17%|█▋ | 5223/30000 [3:20:35<15:51:52, 2.31s/it] 17%|█▋ | 5224/30000 [3:20:37<15:46:12, 2.29s/it] 17%|█▋ | 5225/30000 [3:20:39<15:52:52, 2.31s/it] 17%|█▋ | 5226/30000 [3:20:42<15:55:41, 2.31s/it] 17%|█▋ | 5227/30000 [3:20:44<15:47:30, 2.29s/it] 17%|█▋ | 5228/30000 [3:20:46<15:44:40, 2.29s/it] 17%|█▋ | 5229/30000 [3:20:49<15:43:58, 2.29s/it] 17%|█▋ | 5230/30000 [3:20:51<15:48:29, 2.30s/it] 17%|█▋ | 5230/30000 [3:20:51<15:48:29, 2.30s/it] 17%|█▋ | 5231/30000 [3:20:53<15:43:37, 2.29s/it] 17%|█▋ | 5232/30000 [3:20:55<15:46:15, 2.29s/it] 17%|█▋ | 5233/30000 [3:20:58<15:56:11, 2.32s/it] 17%|█▋ | 5234/30000 [3:21:00<15:53:35, 2.31s/it] 17%|█▋ | 5235/30000 [3:21:02<15:58:43, 2.32s/it] 17%|█▋ | 5236/30000 [3:21:05<15:53:22, 2.31s/it] 17%|█▋ | 5237/30000 [3:21:07<15:51:00, 2.30s/it] 17%|█▋ | 5238/30000 [3:21:09<15:45:43, 2.29s/it] 17%|█▋ | 5239/30000 [3:21:12<15:43:56, 2.29s/it] 17%|█▋ | 5240/30000 [3:21:14<15:49:42, 2.30s/it] 17%|█▋ | 5240/30000 [3:21:14<15:49:42, 2.30s/it] 17%|█▋ | 5241/30000 [3:21:16<15:53:53, 2.31s/it] 17%|█▋ | 5242/30000 [3:21:19<15:51:12, 2.31s/it] 17%|█▋ | 5243/30000 [3:21:21<15:54:03, 2.31s/it] 17%|█▋ | 5244/30000 [3:21:23<15:58:48, 2.32s/it] 17%|█▋ | 5245/30000 [3:21:25<15:51:15, 2.31s/it] 17%|█▋ | 5246/30000 [3:21:28<15:46:19, 2.29s/it] 17%|█▋ | 5247/30000 [3:21:30<15:50:49, 2.30s/it] 17%|█▋ | 5248/30000 [3:21:32<15:46:16, 2.29s/it] 17%|█▋ | 5249/30000 [3:21:35<15:45:56, 2.29s/it] 18%|█▊ | 5250/30000 [3:21:37<15:49:33, 2.30s/it] 18%|█▊ | 5250/30000 [3:21:37<15:49:33, 2.30s/it] 18%|█▊ | 5251/30000 [3:21:39<15:43:51, 2.29s/it] 18%|█▊ | 5252/30000 [3:21:41<15:40:25, 2.28s/it] 18%|█▊ | 5253/30000 [3:21:44<15:51:26, 2.31s/it] 18%|█▊ | 5254/30000 [3:21:46<15:51:33, 2.31s/it] 18%|█▊ | 5255/30000 [3:21:48<15:47:02, 2.30s/it] 18%|█▊ | 5256/30000 [3:21:51<15:51:06, 2.31s/it] 18%|█▊ | 5257/30000 [3:21:53<15:48:23, 2.30s/it] 18%|█▊ | 5258/30000 [3:21:55<15:53:44, 2.31s/it] 18%|█▊ | 5259/30000 [3:21:58<15:51:44, 2.31s/it] 18%|█▊ | 5260/30000 [3:22:00<15:57:13, 2.32s/it] 18%|█▊ | 5260/30000 [3:22:00<15:57:13, 2.32s/it] 18%|█▊ | 5261/30000 [3:22:02<15:55:01, 2.32s/it] 18%|█▊ | 5262/30000 [3:22:04<15:35:58, 2.27s/it] 18%|█▊ | 5263/30000 [3:22:07<15:42:49, 2.29s/it] 18%|█▊ | 5264/30000 [3:22:09<15:39:59, 2.28s/it] 18%|█▊ | 5265/30000 [3:22:11<15:43:07, 2.29s/it] 18%|█▊ | 5266/30000 [3:22:14<15:49:45, 2.30s/it] 18%|█▊ | 5267/30000 [3:22:16<15:51:43, 2.31s/it] 18%|█▊ | 5268/30000 [3:22:18<15:46:33, 2.30s/it] 18%|█▊ | 5269/30000 [3:22:21<15:48:11, 2.30s/it] 18%|█▊ | 5270/30000 [3:22:23<15:44:11, 2.29s/it] 18%|█▊ | 5270/30000 [3:22:23<15:44:11, 2.29s/it] 18%|█▊ | 5271/30000 [3:22:25<15:42:52, 2.29s/it] 18%|█▊ | 5272/30000 [3:22:27<15:44:53, 2.29s/it] 18%|█▊ | 5273/30000 [3:22:30<15:43:42, 2.29s/it] 18%|█▊ | 5274/30000 [3:22:32<15:40:26, 2.28s/it] 18%|█▊ | 5275/30000 [3:22:34<15:37:59, 2.28s/it] 18%|█▊ | 5276/30000 [3:22:37<15:44:18, 2.29s/it] 18%|█▊ | 5277/30000 [3:22:39<15:34:05, 2.27s/it] 18%|█▊ | 5278/30000 [3:22:41<15:31:18, 2.26s/it] 18%|█▊ | 5279/30000 [3:22:43<15:42:53, 2.29s/it] 18%|█▊ | 5280/30000 [3:22:46<15:48:16, 2.30s/it] 18%|█▊ | 5280/30000 [3:22:46<15:48:16, 2.30s/it] 18%|█▊ | 5281/30000 [3:22:48<15:48:39, 2.30s/it] 18%|█▊ | 5282/30000 [3:22:50<15:44:58, 2.29s/it] 18%|█▊ | 5283/30000 [3:22:53<15:49:01, 2.30s/it] 18%|█▊ | 5284/30000 [3:22:55<15:42:04, 2.29s/it] 18%|█▊ | 5285/30000 [3:22:57<15:36:13, 2.27s/it] 18%|█▊ | 5286/30000 [3:23:00<15:49:12, 2.30s/it] 18%|█▊ | 5287/30000 [3:23:02<15:46:08, 2.30s/it] 18%|█▊ | 5288/30000 [3:23:04<15:43:08, 2.29s/it] 18%|█▊ | 5289/30000 [3:23:06<15:43:14, 2.29s/it] 18%|█▊ | 5290/30000 [3:23:09<15:49:26, 2.31s/it] 18%|█▊ | 5290/30000 [3:23:09<15:49:26, 2.31s/it] 18%|█▊ | 5291/30000 [3:23:11<15:46:39, 2.30s/it] 18%|█▊ | 5292/30000 [3:23:13<15:42:54, 2.29s/it] 18%|█▊ | 5293/30000 [3:23:16<15:49:52, 2.31s/it] 18%|█▊ | 5294/30000 [3:23:18<15:46:27, 2.30s/it] 18%|█▊ | 5295/30000 [3:23:20<15:45:58, 2.30s/it] 18%|█▊ | 5296/30000 [3:23:22<15:41:13, 2.29s/it] 18%|█▊ | 5297/30000 [3:23:25<15:48:47, 2.30s/it] 18%|█▊ | 5298/30000 [3:23:27<15:45:38, 2.30s/it] 18%|█▊ | 5299/30000 [3:23:29<15:42:40, 2.29s/it] 18%|█▊ | 5300/30000 [3:23:32<15:43:54, 2.29s/it] 18%|█▊ | 5300/30000 [3:23:32<15:43:54, 2.29s/it] 18%|█▊ | 5301/30000 [3:23:34<15:44:23, 2.29s/it] 18%|█▊ | 5302/30000 [3:23:36<15:41:57, 2.29s/it] 18%|█▊ | 5303/30000 [3:23:38<15:29:59, 2.26s/it] 18%|█▊ | 5304/30000 [3:23:41<15:38:54, 2.28s/it] 18%|█▊ | 5305/30000 [3:23:43<15:38:12, 2.28s/it] 18%|█▊ | 5306/30000 [3:23:45<15:39:25, 2.28s/it] 18%|█▊ | 5307/30000 [3:23:48<15:38:49, 2.28s/it] 18%|█▊ | 5308/30000 [3:23:50<15:44:07, 2.29s/it] 18%|█▊ | 5309/30000 [3:23:52<15:42:47, 2.29s/it] 18%|█▊ | 5310/30000 [3:23:54<15:38:56, 2.28s/it] 18%|█▊ | 5310/30000 [3:23:55<15:38:56, 2.28s/it] 18%|█▊ | 5311/30000 [3:23:57<15:35:19, 2.27s/it] 18%|█▊ | 5312/30000 [3:23:59<15:37:41, 2.28s/it] 18%|█▊ | 5313/30000 [3:24:01<15:39:13, 2.28s/it] 18%|█▊ | 5314/30000 [3:24:04<15:37:43, 2.28s/it] 18%|█▊ | 5315/30000 [3:24:06<15:41:45, 2.29s/it] 18%|█▊ | 5316/30000 [3:24:08<15:42:34, 2.29s/it] 18%|█▊ | 5317/30000 [3:24:10<15:38:25, 2.28s/it] 18%|█▊ | 5318/30000 [3:24:13<15:36:24, 2.28s/it] 18%|█▊ | 5319/30000 [3:24:15<15:40:38, 2.29s/it] 18%|█▊ | 5320/30000 [3:24:17<15:44:13, 2.30s/it] 18%|█▊ | 5320/30000 [3:24:17<15:44:13, 2.30s/it] 18%|█▊ | 5321/30000 [3:24:20<15:42:37, 2.29s/it] 18%|█▊ | 5322/30000 [3:24:22<15:40:47, 2.29s/it] 18%|█▊ | 5323/30000 [3:24:24<15:49:16, 2.31s/it] 18%|█▊ | 5324/30000 [3:24:27<15:53:04, 2.32s/it] 18%|█▊ | 5325/30000 [3:24:29<15:47:25, 2.30s/it] 18%|█▊ | 5326/30000 [3:24:31<15:44:14, 2.30s/it] 18%|█▊ | 5327/30000 [3:24:33<15:50:05, 2.31s/it] 18%|█▊ | 5328/30000 [3:24:36<15:45:31, 2.30s/it] 18%|█▊ | 5329/30000 [3:24:38<15:41:53, 2.29s/it] 18%|█▊ | 5330/30000 [3:24:40<15:39:57, 2.29s/it] 18%|█▊ | 5330/30000 [3:24:40<15:39:57, 2.29s/it] 18%|█▊ | 5331/30000 [3:24:43<15:44:40, 2.30s/it] 18%|█▊ | 5332/30000 [3:24:45<15:46:32, 2.30s/it] 18%|█▊ | 5333/30000 [3:24:47<15:41:49, 2.29s/it] 18%|█▊ | 5334/30000 [3:24:49<15:40:30, 2.29s/it] 18%|█▊ | 5335/30000 [3:24:52<15:42:21, 2.29s/it] 18%|█▊ | 5336/30000 [3:24:54<15:36:21, 2.28s/it] 18%|█▊ | 5337/30000 [3:24:56<15:35:39, 2.28s/it] 18%|█▊ | 5338/30000 [3:24:59<15:34:59, 2.27s/it] 18%|█▊ | 5339/30000 [3:25:01<15:39:15, 2.29s/it] 18%|█▊ | 5340/30000 [3:25:03<15:37:35, 2.28s/it] 18%|█▊ | 5340/30000 [3:25:03<15:37:35, 2.28s/it] 18%|█▊ | 5341/30000 [3:25:05<15:38:03, 2.28s/it] 18%|█▊ | 5342/30000 [3:25:08<15:41:41, 2.29s/it] 18%|█▊ | 5343/30000 [3:25:10<15:41:09, 2.29s/it] 18%|█▊ | 5344/30000 [3:25:12<15:48:55, 2.31s/it] 18%|█▊ | 5345/30000 [3:25:15<16:00:18, 2.34s/it] 18%|█▊ | 5346/30000 [3:25:17<15:59:19, 2.33s/it] 18%|█▊ | 5347/30000 [3:25:19<15:53:27, 2.32s/it] 18%|█▊ | 5348/30000 [3:25:22<15:47:19, 2.31s/it] 18%|█▊ | 5349/30000 [3:25:24<15:44:21, 2.30s/it] 18%|█▊ | 5350/30000 [3:25:26<15:49:56, 2.31s/it] 18%|█▊ | 5350/30000 [3:25:26<15:49:56, 2.31s/it] 18%|█▊ | 5351/30000 [3:25:29<15:50:35, 2.31s/it] 18%|█▊ | 5352/30000 [3:25:31<15:34:00, 2.27s/it] 18%|█▊ | 5353/30000 [3:25:33<15:38:42, 2.29s/it] 18%|█▊ | 5354/30000 [3:25:35<15:39:49, 2.29s/it] 18%|█▊ | 5355/30000 [3:25:38<15:36:58, 2.28s/it] 18%|█▊ | 5356/30000 [3:25:40<15:33:09, 2.27s/it] 18%|█▊ | 5357/30000 [3:25:42<15:39:29, 2.29s/it] 18%|█▊ | 5358/30000 [3:25:44<15:32:48, 2.27s/it] 18%|█▊ | 5359/30000 [3:25:47<15:34:32, 2.28s/it] 18%|█▊ | 5360/30000 [3:25:49<15:38:33, 2.29s/it] 18%|█▊ | 5360/30000 [3:25:49<15:38:33, 2.29s/it] 18%|█▊ | 5361/30000 [3:25:51<15:35:14, 2.28s/it] 18%|█▊ | 5362/30000 [3:25:54<15:31:42, 2.27s/it] 18%|█▊ | 5363/30000 [3:25:56<15:33:06, 2.27s/it] 18%|█▊ | 5364/30000 [3:25:58<15:39:41, 2.29s/it] 18%|█▊ | 5365/30000 [3:26:00<15:35:40, 2.28s/it] 18%|█▊ | 5366/30000 [3:26:03<15:32:30, 2.27s/it] 18%|█▊ | 5367/30000 [3:26:05<15:42:32, 2.30s/it] 18%|█▊ | 5368/30000 [3:26:07<15:42:19, 2.30s/it] 18%|█▊ | 5369/30000 [3:26:10<15:29:00, 2.26s/it] 18%|█▊ | 5370/30000 [3:26:12<15:38:09, 2.29s/it] 18%|█▊ | 5370/30000 [3:26:12<15:38:09, 2.29s/it] 18%|█▊ | 5371/30000 [3:26:14<15:37:26, 2.28s/it] 18%|█▊ | 5372/30000 [3:26:16<15:37:26, 2.28s/it] 18%|█▊ | 5373/30000 [3:26:19<15:35:42, 2.28s/it] 18%|█▊ | 5374/30000 [3:26:21<15:38:11, 2.29s/it] 18%|█▊ | 5375/30000 [3:26:23<15:34:27, 2.28s/it] 18%|█▊ | 5376/30000 [3:26:25<15:24:48, 2.25s/it] 18%|█▊ | 5377/30000 [3:26:28<15:34:02, 2.28s/it] 18%|█▊ | 5378/30000 [3:26:30<15:29:13, 2.26s/it] 18%|█▊ | 5379/30000 [3:26:32<15:31:18, 2.27s/it] 18%|█▊ | 5380/30000 [3:26:35<15:39:29, 2.29s/it] 18%|█▊ | 5380/30000 [3:26:35<15:39:29, 2.29s/it] 18%|█▊ | 5381/30000 [3:26:37<15:39:58, 2.29s/it] 18%|█▊ | 5382/30000 [3:26:39<15:35:18, 2.28s/it] 18%|█▊ | 5383/30000 [3:26:42<15:41:00, 2.29s/it] 18%|█▊ | 5384/30000 [3:26:44<15:39:22, 2.29s/it] 18%|█▊ | 5385/30000 [3:26:46<15:37:53, 2.29s/it] 18%|█▊ | 5386/30000 [3:26:48<15:45:52, 2.31s/it] 18%|█▊ | 5387/30000 [3:26:51<15:47:29, 2.31s/it] 18%|█▊ | 5388/30000 [3:26:53<15:40:15, 2.29s/it] 18%|█▊ | 5389/30000 [3:26:55<15:41:59, 2.30s/it] 18%|█▊ | 5390/30000 [3:26:58<15:46:15, 2.31s/it] 18%|█▊ | 5390/30000 [3:26:58<15:46:15, 2.31s/it] 18%|█▊ | 5391/30000 [3:27:00<15:48:46, 2.31s/it] 18%|█▊ | 5392/30000 [3:27:02<15:44:58, 2.30s/it] 18%|█▊ | 5393/30000 [3:27:05<15:49:26, 2.32s/it] 18%|█▊ | 5394/30000 [3:27:07<15:53:02, 2.32s/it] 18%|█▊ | 5395/30000 [3:27:09<15:43:48, 2.30s/it] 18%|█▊ | 5396/30000 [3:27:11<15:38:32, 2.29s/it] 18%|█▊ | 5397/30000 [3:27:14<15:43:04, 2.30s/it] 18%|█▊ | 5398/30000 [3:27:16<15:35:27, 2.28s/it] 18%|█▊ | 5399/30000 [3:27:18<15:40:15, 2.29s/it] 18%|█▊ | 5400/30000 [3:27:21<15:43:32, 2.30s/it] 18%|█▊ | 5400/30000 [3:27:21<15:43:32, 2.30s/it] 18%|█▊ | 5401/30000 [3:27:23<15:44:11, 2.30s/it] 18%|█▊ | 5402/30000 [3:27:25<15:40:06, 2.29s/it] 18%|█▊ | 5403/30000 [3:27:27<15:34:12, 2.28s/it] 18%|█▊ | 5404/30000 [3:27:30<15:32:03, 2.27s/it] 18%|█▊ | 5405/30000 [3:27:32<15:28:01, 2.26s/it] 18%|█▊ | 5406/30000 [3:27:34<15:34:55, 2.28s/it] 18%|█▊ | 5407/30000 [3:27:37<15:30:47, 2.27s/it] 18%|█▊ | 5408/30000 [3:27:39<15:31:09, 2.27s/it] 18%|█▊ | 5409/30000 [3:27:41<15:35:37, 2.28s/it] 18%|█▊ | 5410/30000 [3:27:43<15:41:27, 2.30s/it] 18%|█▊ | 5410/30000 [3:27:43<15:41:27, 2.30s/it] 18%|█▊ | 5411/30000 [3:27:46<15:30:09, 2.27s/it] 18%|█▊ | 5412/30000 [3:27:48<15:21:32, 2.25s/it] 18%|█▊ | 5413/30000 [3:27:50<15:28:10, 2.27s/it] 18%|█▊ | 5414/30000 [3:27:52<15:26:14, 2.26s/it] 18%|█▊ | 5415/30000 [3:27:55<15:17:15, 2.24s/it] 18%|█▊ | 5416/30000 [3:27:57<15:25:48, 2.26s/it] 18%|█▊ | 5417/30000 [3:27:59<15:31:19, 2.27s/it] 18%|█▊ | 5418/30000 [3:28:01<15:28:08, 2.27s/it] 18%|█▊ | 5419/30000 [3:28:04<15:29:10, 2.27s/it] 18%|█▊ | 5420/30000 [3:28:06<15:35:55, 2.28s/it] 18%|█▊ | 5420/30000 [3:28:06<15:35:55, 2.28s/it] 18%|█▊ | 5421/30000 [3:28:08<15:35:27, 2.28s/it] 18%|█▊ | 5422/30000 [3:28:11<15:34:57, 2.28s/it] 18%|█▊ | 5423/30000 [3:28:13<15:39:51, 2.29s/it] 18%|█▊ | 5424/30000 [3:28:15<15:39:54, 2.29s/it] 18%|█▊ | 5425/30000 [3:28:18<15:39:52, 2.29s/it] 18%|█▊ | 5426/30000 [3:28:20<15:37:26, 2.29s/it] 18%|█▊ | 5427/30000 [3:28:22<15:43:15, 2.30s/it] 18%|█▊ | 5428/30000 [3:28:24<15:41:16, 2.30s/it] 18%|█▊ | 5429/30000 [3:28:27<15:38:23, 2.29s/it] 18%|█▊ | 5430/30000 [3:28:29<15:33:22, 2.28s/it] 18%|█▊ | 5430/30000 [3:28:29<15:33:22, 2.28s/it] 18%|█▊ | 5431/30000 [3:28:31<15:41:50, 2.30s/it] 18%|█▊ | 5432/30000 [3:28:34<15:43:56, 2.31s/it] 18%|█▊ | 5433/30000 [3:28:36<15:21:17, 2.25s/it] 18%|█▊ | 5434/30000 [3:28:38<15:29:42, 2.27s/it] 18%|█▊ | 5435/30000 [3:28:40<15:34:01, 2.28s/it] 18%|█▊ | 5436/30000 [3:28:43<15:29:02, 2.27s/it] 18%|█▊ | 5437/30000 [3:28:45<15:28:38, 2.27s/it] 18%|█▊ | 5438/30000 [3:28:47<15:32:44, 2.28s/it] 18%|█▊ | 5439/30000 [3:28:49<15:27:53, 2.27s/it] 18%|█▊ | 5440/30000 [3:28:52<15:28:40, 2.27s/it] 18%|█▊ | 5440/30000 [3:28:52<15:28:40, 2.27s/it] 18%|█▊ | 5441/30000 [3:28:54<15:29:54, 2.27s/it] 18%|█▊ | 5442/30000 [3:28:56<15:33:27, 2.28s/it] 18%|█▊ | 5443/30000 [3:28:59<15:32:09, 2.28s/it] 18%|█▊ | 5444/30000 [3:29:01<15:24:40, 2.26s/it] 18%|█▊ | 5445/30000 [3:29:03<15:31:53, 2.28s/it] 18%|█▊ | 5446/30000 [3:29:05<15:26:03, 2.26s/it] 18%|█▊ | 5447/30000 [3:29:08<15:28:32, 2.27s/it] 18%|█▊ | 5448/30000 [3:29:10<15:32:54, 2.28s/it] 18%|█▊ | 5449/30000 [3:29:12<15:34:07, 2.28s/it] 18%|█▊ | 5450/30000 [3:29:14<15:30:58, 2.28s/it] 18%|█▊ | 5450/30000 [3:29:14<15:30:58, 2.28s/it] 18%|█▊ | 5451/30000 [3:29:17<15:27:37, 2.27s/it] 18%|█▊ | 5452/30000 [3:29:19<15:40:07, 2.30s/it] 18%|█▊ | 5453/30000 [3:29:21<15:39:51, 2.30s/it] 18%|█▊ | 5454/30000 [3:29:24<15:36:40, 2.29s/it] 18%|█▊ | 5455/30000 [3:29:26<15:31:54, 2.28s/it] 18%|█▊ | 5456/30000 [3:29:28<15:37:37, 2.29s/it] 18%|█▊ | 5457/30000 [3:29:30<15:34:46, 2.29s/it] 18%|█▊ | 5458/30000 [3:29:33<15:33:05, 2.28s/it] 18%|█▊ | 5459/30000 [3:29:35<15:38:24, 2.29s/it] 18%|█▊ | 5460/30000 [3:29:37<15:37:09, 2.29s/it] +{'loss': 0.0183, 'grad_norm': 0.1413814276456833, 'learning_rate': 2.8866719839288215e-05, 'flow_matching_loss': 0.018290637247264385, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.47} +{'loss': 0.017, 'grad_norm': 0.10948627442121506, 'learning_rate': 2.886040660049283e-05, 'flow_matching_loss': 0.016997105814516543, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.47} +{'loss': 0.0163, 'grad_norm': 0.09320095181465149, 'learning_rate': 2.8854076519996467e-05, 'flow_matching_loss': 0.016289071924984456, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.48} +{'loss': 0.015, 'grad_norm': 0.07379238307476044, 'learning_rate': 2.884772960549078e-05, 'flow_matching_loss': 0.01503765545785427, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.48} +{'loss': 0.0188, 'grad_norm': 0.09777366369962692, 'learning_rate': 2.8841365864687867e-05, 'flow_matching_loss': 0.018831222504377364, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.48} +{'loss': 0.0159, 'grad_norm': 0.0771479606628418, 'learning_rate': 2.8834985305320276e-05, 'flow_matching_loss': 0.01591841457411647, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.48} +{'loss': 0.0164, 'grad_norm': 0.07898079603910446, 'learning_rate': 2.882858793514098e-05, 'flow_matching_loss': 0.016431513521820307, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.49} +{'loss': 0.0171, 'grad_norm': 0.1075642928481102, 'learning_rate': 2.8822173761923398e-05, 'flow_matching_loss': 0.017069196980446576, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.49} +{'loss': 0.0159, 'grad_norm': 0.08737059682607651, 'learning_rate': 2.8815742793461343e-05, 'flow_matching_loss': 0.01585097620263696, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.49} +{'loss': 0.0166, 'grad_norm': 0.09448487311601639, 'learning_rate': 2.880929503756906e-05, 'flow_matching_loss': 0.01661896863952279, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.5} +{'loss': 0.0174, 'grad_norm': 0.11666061729192734, 'learning_rate': 2.880283050208117e-05, 'flow_matching_loss': 0.017361583095043896, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.5} +{'loss': 0.0165, 'grad_norm': 0.06508040428161621, 'learning_rate': 2.8796349194852704e-05, 'flow_matching_loss': 0.016455515287816525, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.5} +{'loss': 0.0172, 'grad_norm': 0.09014807641506195, 'learning_rate': 2.8789851123759052e-05, 'flow_matching_loss': 0.01716578146442771, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.5} +{'loss': 0.0178, 'grad_norm': 0.09673328697681427, 'learning_rate': 2.8783336296695985e-05, 'flow_matching_loss': 0.017792343720793725, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.51} +{'loss': 0.0162, 'grad_norm': 0.08644711226224899, 'learning_rate': 2.877680472157964e-05, 'flow_matching_loss': 0.01618854282423854, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.51} +{'loss': 0.0186, 'grad_norm': 0.07266746461391449, 'learning_rate': 2.8770256406346483e-05, 'flow_matching_loss': 0.01862327391281724, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.51} +{'loss': 0.017, 'grad_norm': 0.07603176683187485, 'learning_rate': 2.8763691358953348e-05, 'flow_matching_loss': 0.017012276593595742, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.52} +{'loss': 0.0161, 'grad_norm': 0.08934494853019714, 'learning_rate': 2.8757109587377377e-05, 'flow_matching_loss': 0.016088925302028656, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.52} +{'loss': 0.0163, 'grad_norm': 0.07086964696645737, 'learning_rate': 2.8750511099616047e-05, 'flow_matching_loss': 0.01628910694271326, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.52} +{'loss': 0.0167, 'grad_norm': 0.09887402504682541, 'learning_rate': 2.8743895903687143e-05, 'flow_matching_loss': 0.0166813756339252, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.52} +{'loss': 0.019, 'grad_norm': 0.10313404351472855, 'learning_rate': 2.8737264007628753e-05, 'flow_matching_loss': 0.019000981654971837, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.53} +{'loss': 0.0156, 'grad_norm': 0.0866900384426117, 'learning_rate': 2.8730615419499247e-05, 'flow_matching_loss': 0.015572401788085699, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.53} +{'loss': 0.0196, 'grad_norm': 0.13202694058418274, 'learning_rate': 2.8723950147377303e-05, 'flow_matching_loss': 0.019560024421662093, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.53} +{'loss': 0.0176, 'grad_norm': 0.09223472326993942, 'learning_rate': 2.871726819936184e-05, 'flow_matching_loss': 0.017563256621360778, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.54} +{'loss': 0.017, 'grad_norm': 0.0844714343547821, 'learning_rate': 2.8710569583572068e-05, 'flow_matching_loss': 0.017023221775889396, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.54} +{'loss': 0.0161, 'grad_norm': 0.07358773052692413, 'learning_rate': 2.870385430814742e-05, 'flow_matching_loss': 0.016103014629334212, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.54} +{'loss': 0.0168, 'grad_norm': 0.09081678837537766, 'learning_rate': 2.8697122381247605e-05, 'flow_matching_loss': 0.01680693160742521, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.54} +{'loss': 0.0184, 'grad_norm': 0.0926685482263565, 'learning_rate': 2.8690373811052545e-05, 'flow_matching_loss': 0.01836457820609212, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.55} +{'loss': 0.0161, 'grad_norm': 0.07901865243911743, 'learning_rate': 2.8683608605762378e-05, 'flow_matching_loss': 0.016051399521529675, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.55} +{'loss': 0.0181, 'grad_norm': 0.09506777673959732, 'learning_rate': 2.8676826773597485e-05, 'flow_matching_loss': 0.01812496967613697, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.55} +{'loss': 0.0179, 'grad_norm': 0.09674572199583054, 'learning_rate': 2.8670028322798413e-05, 'flow_matching_loss': 0.01791005441918969, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.56} +{'loss': 0.017, 'grad_norm': 0.12036973237991333, 'learning_rate': 2.8663213261625932e-05, 'flow_matching_loss': 0.0170031463727355, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.56} +{'loss': 0.0182, 'grad_norm': 0.0873119980096817, 'learning_rate': 2.865638159836098e-05, 'flow_matching_loss': 0.018219267297536135, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.56} +{'loss': 0.0174, 'grad_norm': 0.08233197778463364, 'learning_rate': 2.8649533341304677e-05, 'flow_matching_loss': 0.017394870799034835, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.57} +{'loss': 0.015, 'grad_norm': 0.06513916701078415, 'learning_rate': 2.8642668498778297e-05, 'flow_matching_loss': 0.014961966685950756, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.57} +{'loss': 0.0176, 'grad_norm': 0.07748878747224808, 'learning_rate': 2.8635787079123268e-05, 'flow_matching_loss': 0.017580207064747812, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.57} +{'loss': 0.017, 'grad_norm': 0.08867599815130234, 'learning_rate': 2.8628889090701175e-05, 'flow_matching_loss': 0.01698612906038761, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.57} +{'loss': 0.0157, 'grad_norm': 0.09579691290855408, 'learning_rate': 2.862197454189371e-05, 'flow_matching_loss': 0.01570911183953285, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.58} +{'loss': 0.0169, 'grad_norm': 0.06944193691015244, 'learning_rate': 2.8615043441102708e-05, 'flow_matching_loss': 0.01685804072767496, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.58} +{'loss': 0.0173, 'grad_norm': 0.1117226630449295, 'learning_rate': 2.860809579675011e-05, 'flow_matching_loss': 0.017252314370125532, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.58} +{'loss': 0.0167, 'grad_norm': 0.10219967365264893, 'learning_rate': 2.860113161727796e-05, 'flow_matching_loss': 0.016684918198734522, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.59} +{'loss': 0.0169, 'grad_norm': 0.10607108473777771, 'learning_rate': 2.8594150911148393e-05, 'flow_matching_loss': 0.016875190660357475, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.59} 18%|█▊ | 5460/30000 [3:29:37<15:37:09, 2.29s/it] 18%|█▊ | 5461/30000 [3:29:40<15:33:32, 2.28s/it] 18%|█▊ | 5462/30000 [3:29:42<15:31:37, 2.28s/it] 18%|█▊ | 5463/30000 [3:29:44<15:37:28, 2.29s/it] 18%|█▊ | 5464/30000 [3:29:46<15:32:37, 2.28s/it] 18%|█▊ | 5465/30000 [3:29:49<15:33:28, 2.28s/it] 18%|█▊ | 5466/30000 [3:29:51<15:38:39, 2.30s/it] 18%|█▊ | 5467/30000 [3:29:53<15:36:59, 2.29s/it] 18%|█▊ | 5468/30000 [3:29:56<15:33:38, 2.28s/it] 18%|█▊ | 5469/30000 [3:29:58<15:36:45, 2.29s/it] 18%|█▊ | 5470/30000 [3:30:00<15:38:39, 2.30s/it] 18%|█▊ | 5470/30000 [3:30:00<15:38:39, 2.30s/it] 18%|█▊ | 5471/30000 [3:30:03<15:35:29, 2.29s/it] 18%|█▊ | 5472/30000 [3:30:05<15:40:41, 2.30s/it] 18%|█▊ | 5473/30000 [3:30:07<15:37:26, 2.29s/it] 18%|█▊ | 5474/30000 [3:30:09<15:34:46, 2.29s/it] 18%|█▊ | 5475/30000 [3:30:12<15:38:22, 2.30s/it] 18%|█▊ | 5476/30000 [3:30:14<15:40:50, 2.30s/it] 18%|█▊ | 5477/30000 [3:30:16<15:18:52, 2.25s/it] 18%|█▊ | 5478/30000 [3:30:18<15:23:02, 2.26s/it] 18%|█▊ | 5479/30000 [3:30:21<15:31:21, 2.28s/it] 18%|█▊ | 5480/30000 [3:30:23<15:29:59, 2.28s/it] 18%|█▊ | 5480/30000 [3:30:23<15:29:59, 2.28s/it] 18%|█▊ | 5481/30000 [3:30:25<15:30:32, 2.28s/it] 18%|█▊ | 5482/30000 [3:30:28<15:37:05, 2.29s/it] 18%|█▊ | 5483/30000 [3:30:30<15:31:35, 2.28s/it] 18%|█▊ | 5484/30000 [3:30:32<15:24:37, 2.26s/it] 18%|█▊ | 5485/30000 [3:30:34<15:29:52, 2.28s/it] 18%|█▊ | 5486/30000 [3:30:37<15:27:18, 2.27s/it] 18%|█▊ | 5487/30000 [3:30:39<15:31:18, 2.28s/it] 18%|█▊ | 5488/30000 [3:30:41<15:39:09, 2.30s/it] 18%|█▊ | 5489/30000 [3:30:44<15:35:33, 2.29s/it] 18%|█▊ | 5490/30000 [3:30:46<15:33:59, 2.29s/it] 18%|█▊ | 5490/30000 [3:30:46<15:33:59, 2.29s/it] 18%|█▊ | 5491/30000 [3:30:48<15:28:04, 2.27s/it] 18%|█▊ | 5492/30000 [3:30:50<15:31:52, 2.28s/it] 18%|█▊ | 5493/30000 [3:30:53<15:29:18, 2.28s/it] 18%|█▊ | 5494/30000 [3:30:55<15:28:58, 2.27s/it] 18%|█▊ | 5495/30000 [3:30:57<15:39:43, 2.30s/it] 18%|█▊ | 5496/30000 [3:31:00<15:38:57, 2.30s/it] 18%|█▊ | 5497/30000 [3:31:02<15:35:19, 2.29s/it] 18%|█▊ | 5498/30000 [3:31:04<15:31:44, 2.28s/it] 18%|█▊ | 5499/30000 [3:31:06<15:38:14, 2.30s/it] 18%|█▊ | 5500/30000 [3:31:09<15:36:06, 2.29s/it] 18%|█▊ | 5500/30000 [3:31:09<15:36:06, 2.29s/it] 18%|█▊ | 5501/30000 [3:31:11<15:28:23, 2.27s/it] 18%|█▊ | 5502/30000 [3:31:13<15:37:06, 2.30s/it] 18%|█▊ | 5503/30000 [3:31:16<15:44:17, 2.31s/it] 18%|█▊ | 5504/30000 [3:31:18<15:38:30, 2.30s/it] 18%|█▊ | 5505/30000 [3:31:20<15:34:39, 2.29s/it] 18%|█▊ | 5506/30000 [3:31:23<15:39:45, 2.30s/it] 18%|█▊ | 5507/30000 [3:31:25<15:42:12, 2.31s/it] 18%|█▊ | 5508/30000 [3:31:27<15:35:34, 2.29s/it] 18%|█▊ | 5509/30000 [3:31:29<15:45:24, 2.32s/it] 18%|█▊ | 5510/30000 [3:31:32<15:42:58, 2.31s/it] 18%|█▊ | 5510/30000 [3:31:32<15:42:58, 2.31s/it] 18%|█▊ | 5511/30000 [3:31:34<15:39:46, 2.30s/it] 18%|█▊ | 5512/30000 [3:31:36<15:42:57, 2.31s/it] 18%|█▊ | 5513/30000 [3:31:39<15:39:19, 2.30s/it] 18%|█▊ | 5514/30000 [3:31:41<15:42:10, 2.31s/it] 18%|█▊ | 5515/30000 [3:31:43<15:34:39, 2.29s/it] 18%|█▊ | 5516/30000 [3:31:46<15:35:02, 2.29s/it] 18%|█▊ | 5517/30000 [3:31:48<15:31:56, 2.28s/it] 18%|█▊ | 5518/30000 [3:31:50<15:29:58, 2.28s/it] 18%|█▊ | 5519/30000 [3:31:52<15:40:10, 2.30s/it] 18%|█▊ | 5520/30000 [3:31:55<15:36:18, 2.29s/it] 18%|█▊ | 5520/30000 [3:31:55<15:36:18, 2.29s/it] 18%|█▊ | 5521/30000 [3:31:57<15:37:12, 2.30s/it] 18%|█▊ | 5522/30000 [3:31:59<15:37:21, 2.30s/it] 18%|█▊ | 5523/30000 [3:32:02<15:43:29, 2.31s/it] 18%|█▊ | 5524/30000 [3:32:04<15:44:48, 2.32s/it] 18%|█▊ | 5525/30000 [3:32:06<15:38:59, 2.30s/it] 18%|█▊ | 5526/30000 [3:32:09<15:36:20, 2.30s/it] 18%|█▊ | 5527/30000 [3:32:11<15:35:14, 2.29s/it] 18%|█▊ | 5528/30000 [3:32:13<15:32:33, 2.29s/it] 18%|█▊ | 5529/30000 [3:32:15<15:32:15, 2.29s/it] 18%|█▊ | 5530/30000 [3:32:18<15:42:28, 2.31s/it] 18%|█▊ | 5530/30000 [3:32:18<15:42:28, 2.31s/it] 18%|█▊ | 5531/30000 [3:32:20<15:41:26, 2.31s/it] 18%|█▊ | 5532/30000 [3:32:22<15:36:13, 2.30s/it] 18%|█▊ | 5533/30000 [3:32:25<15:42:24, 2.31s/it] 18%|█▊ | 5534/30000 [3:32:27<15:43:58, 2.31s/it] 18%|█▊ | 5535/30000 [3:32:29<15:46:54, 2.32s/it] 18%|█▊ | 5536/30000 [3:32:32<15:43:20, 2.31s/it] 18%|█▊ | 5537/30000 [3:32:34<15:39:21, 2.30s/it] 18%|█▊ | 5538/30000 [3:32:36<15:49:26, 2.33s/it] 18%|█▊ | 5539/30000 [3:32:39<15:47:29, 2.32s/it] 18%|█▊ | 5540/30000 [3:32:41<15:42:28, 2.31s/it] 18%|█▊ | 5540/30000 [3:32:41<15:42:28, 2.31s/it] 18%|█▊ | 5541/30000 [3:32:43<15:31:53, 2.29s/it] 18%|█▊ | 5542/30000 [3:32:45<15:36:55, 2.30s/it] 18%|█▊ | 5543/30000 [3:32:48<15:36:32, 2.30s/it] 18%|█▊ | 5544/30000 [3:32:50<15:31:07, 2.28s/it] 18%|█▊ | 5545/30000 [3:32:52<15:32:25, 2.29s/it] 18%|█▊ | 5546/30000 [3:32:55<15:38:14, 2.30s/it] 18%|█▊ | 5547/30000 [3:32:57<15:35:16, 2.29s/it] 18%|█▊ | 5548/30000 [3:32:59<15:36:36, 2.30s/it] 18%|█▊ | 5549/30000 [3:33:02<15:38:15, 2.30s/it] 18%|█▊ | 5550/30000 [3:33:04<15:39:44, 2.31s/it] 18%|█▊ | 5550/30000 [3:33:04<15:39:44, 2.31s/it] 19%|█▊ | 5551/30000 [3:33:06<15:35:49, 2.30s/it] 19%|█▊ | 5552/30000 [3:33:08<15:30:56, 2.28s/it] 19%|█▊ | 5553/30000 [3:33:11<15:32:48, 2.29s/it] 19%|█▊ | 5554/30000 [3:33:13<15:37:29, 2.30s/it] 19%|█▊ | 5555/30000 [3:33:15<15:29:41, 2.28s/it] 19%|█▊ | 5556/30000 [3:33:18<15:38:14, 2.30s/it] 19%|█▊ | 5557/30000 [3:33:20<16:06:07, 2.37s/it] 19%|█▊ | 5558/30000 [3:33:22<16:03:21, 2.36s/it] 19%|█▊ | 5559/30000 [3:33:25<15:46:48, 2.32s/it] 19%|█▊ | 5560/30000 [3:33:27<15:39:29, 2.31s/it] 19%|█▊ | 5560/30000 [3:33:27<15:39:29, 2.31s/it] 19%|█▊ | 5561/30000 [3:33:29<15:47:59, 2.33s/it] 19%|█▊ | 5562/30000 [3:33:32<15:42:13, 2.31s/it] 19%|█▊ | 5563/30000 [3:33:34<15:34:34, 2.29s/it] 19%|█▊ | 5564/30000 [3:33:36<15:38:42, 2.30s/it] 19%|█▊ | 5565/30000 [3:33:38<15:39:45, 2.31s/it] 19%|█▊ | 5566/30000 [3:33:41<15:33:40, 2.29s/it] 19%|█▊ | 5567/30000 [3:33:43<15:32:55, 2.29s/it] 19%|█▊ | 5568/30000 [3:33:45<15:39:15, 2.31s/it] 19%|█▊ | 5569/30000 [3:33:48<15:41:42, 2.31s/it] 19%|█▊ | 5570/30000 [3:33:50<15:29:10, 2.28s/it] 19%|█▊ | 5570/30000 [3:33:50<15:29:10, 2.28s/it] 19%|█▊ | 5571/30000 [3:33:52<15:30:34, 2.29s/it] 19%|█▊ | 5572/30000 [3:33:55<15:36:23, 2.30s/it] 19%|█▊ | 5573/30000 [3:33:57<15:40:23, 2.31s/it] 19%|█▊ | 5574/30000 [3:33:59<15:34:40, 2.30s/it] 19%|█▊ | 5575/30000 [3:34:01<15:37:02, 2.30s/it] 19%|█▊ | 5576/30000 [3:34:04<15:40:46, 2.31s/it] 19%|█▊ | 5577/30000 [3:34:06<15:39:46, 2.31s/it] 19%|█▊ | 5578/30000 [3:34:08<15:33:02, 2.29s/it] 19%|█▊ | 5579/30000 [3:34:11<15:29:24, 2.28s/it] 19%|█▊ | 5580/30000 [3:34:13<15:40:35, 2.31s/it] 19%|█▊ | 5580/30000 [3:34:13<15:40:35, 2.31s/it] 19%|█▊ | 5581/30000 [3:34:19<22:27:01, 3.31s/it] 19%|█▊ | 5582/30000 [3:34:21<20:28:11, 3.02s/it] 19%|█▊ | 5583/30000 [3:34:23<18:59:42, 2.80s/it] 19%|█▊ | 5584/30000 [3:34:26<17:57:41, 2.65s/it] 19%|█▊ | 5585/30000 [3:34:28<17:21:08, 2.56s/it] 19%|█▊ | 5586/30000 [3:34:30<16:51:18, 2.49s/it] 19%|█▊ | 5587/30000 [3:34:33<16:27:09, 2.43s/it] 19%|█▊ | 5588/30000 [3:34:35<16:19:18, 2.41s/it] 19%|█▊ | 5589/30000 [3:34:41<23:59:25, 3.54s/it] 19%|█▊ | 5590/30000 [3:34:43<20:42:54, 3.06s/it] 19%|█▊ | 5590/30000 [3:34:43<20:42:54, 3.06s/it] 19%|█▊ | 5591/30000 [3:34:45<19:17:48, 2.85s/it] 19%|█▊ | 5592/30000 [3:34:47<17:38:37, 2.60s/it] 19%|█▊ | 5593/30000 [3:34:50<17:00:04, 2.51s/it] 19%|█▊ | 5594/30000 [3:34:52<16:41:48, 2.46s/it] 19%|█▊ | 5595/30000 [3:34:54<16:25:06, 2.42s/it] 19%|█▊ | 5596/30000 [3:34:57<16:11:32, 2.39s/it] 19%|█▊ | 5597/30000 [3:35:02<22:02:52, 3.25s/it] 19%|█▊ | 5598/30000 [3:35:04<20:01:08, 2.95s/it] 19%|█▊ | 5599/30000 [3:35:06<18:13:10, 2.69s/it] 19%|█▊ | 5600/30000 [3:35:08<16:22:07, 2.42s/it] 19%|█▊ | 5600/30000 [3:35:08<16:22:07, 2.42s/it] 19%|█▊ | 5601/30000 [3:35:10<16:01:16, 2.36s/it] 19%|█▊ | 5602/30000 [3:35:12<15:36:30, 2.30s/it] 19%|█▊ | 5603/30000 [3:35:15<15:40:50, 2.31s/it] 19%|█▊ | 5604/30000 [3:35:17<15:46:52, 2.33s/it] 19%|█▊ | 5605/30000 [3:35:23<23:08:44, 3.42s/it] 19%|█▊ | 5606/30000 [3:35:25<20:56:16, 3.09s/it] 19%|█▊ | 5607/30000 [3:35:28<19:16:46, 2.85s/it] 19%|█▊ | 5608/30000 [3:35:30<17:34:53, 2.59s/it] 19%|█▊ | 5609/30000 [3:35:31<15:18:24, 2.26s/it] 19%|█▊ | 5610/30000 [3:35:33<15:20:38, 2.26s/it] 19%|█▊ | 5610/30000 [3:35:33<15:20:38, 2.26s/it] 19%|█▊ | 5611/30000 [3:35:36<15:27:48, 2.28s/it] 19%|█▊ | 5612/30000 [3:35:38<15:30:38, 2.29s/it] 19%|█▊ | 5613/30000 [3:35:42<19:39:03, 2.90s/it] 19%|█▊ | 5614/30000 [3:35:45<18:30:18, 2.73s/it] 19%|█▊ | 5615/30000 [3:35:47<17:32:47, 2.59s/it] 19%|█▊ | 5616/30000 [3:35:49<16:49:51, 2.48s/it] 19%|█▊ | 5617/30000 [3:35:52<16:31:13, 2.44s/it] 19%|█▊ | 5618/30000 [3:35:54<16:18:54, 2.41s/it] 19%|█▊ | 5619/30000 [3:35:55<14:01:24, 2.07s/it] 19%|█▊ | 5620/30000 [3:35:58<14:35:19, 2.15s/it] 19%|█▊ | 5620/30000 [3:35:58<14:35:19, 2.15s/it] 19%|█▊ | 5621/30000 [3:36:04<22:25:08, 3.31s/it] 19%|█▊ | 5622/30000 [3:36:05<19:03:29, 2.81s/it] 19%|█▊ | 5623/30000 [3:36:07<17:00:27, 2.51s/it] 19%|█▊ | 5624/30000 [3:36:09<16:34:14, 2.45s/it] 19%|█▉ | 5625/30000 [3:36:12<16:17:52, 2.41s/it] 19%|█▉ | 5626/30000 [3:36:14<16:09:34, 2.39s/it] 19%|█▉ | 5627/30000 [3:36:16<16:03:39, 2.37s/it] 19%|█▉ | 5628/30000 [3:36:19<15:56:20, 2.35s/it] 19%|█▉ | 5629/30000 [3:36:24<22:18:57, 3.30s/it] 19%|█▉ | 5630/30000 [3:36:25<18:20:23, 2.71s/it] 19%|█▉ | 5630/30000 [3:36:25<18:20:23, 2.71s/it] 19%|█▉ | 5631/30000 [3:36:28<17:29:27, 2.58s/it] 19%|█▉ | 5632/30000 [3:36:30<15:51:49, 2.34s/it] 19%|█▉ | 5633/30000 [3:36:32<15:52:35, 2.35s/it] 19%|█▉ | 5634/30000 [3:36:34<15:43:40, 2.32s/it] 19%|█▉ | 5635/30000 [3:36:36<15:32:23, 2.30s/it] 19%|█▉ | 5636/30000 [3:36:39<15:26:37, 2.28s/it] 19%|█▉ | 5637/30000 [3:36:47<28:40:42, 4.24s/it] 19%|█▉ | 5638/30000 [3:36:49<22:28:51, 3.32s/it] 19%|█▉ | 5639/30000 [3:36:50<18:20:30, 2.71s/it] 19%|█▉ | 5640/30000 [3:36:52<17:31:48, 2.59s/it] 19%|█▉ | 5640/30000 [3:36:52<17:31:48, 2.59s/it] 19%|█▉ | 5641/30000 [3:36:55<16:58:45, 2.51s/it] 19%|█▉ | 5642/30000 [3:36:57<16:27:24, 2.43s/it] 19%|█▉ | 5643/30000 [3:36:59<16:05:48, 2.38s/it] 19%|█▉ | 5644/30000 [3:37:01<15:08:43, 2.24s/it] 19%|█▉ | 5645/30000 [3:37:08<24:22:03, 3.60s/it] 19%|█▉ | 5646/30000 [3:37:10<21:36:57, 3.20s/it] 19%|█▉ | 5647/30000 [3:37:12<19:45:38, 2.92s/it] 19%|█▉ | 5648/30000 [3:37:15<18:29:44, 2.73s/it] 19%|█▉ | 5649/30000 [3:37:17<17:42:00, 2.62s/it] 19%|█▉ | 5650/30000 [3:37:19<17:02:02, 2.52s/it] 19%|█▉ | 5650/30000 [3:37:19<17:02:02, 2.52s/it] 19%|█▉ | 5651/30000 [3:37:21<16:29:26, 2.44s/it] 19%|█▉ | 5652/30000 [3:37:24<16:21:05, 2.42s/it] 19%|█▉ | 5653/30000 [3:37:28<19:07:58, 2.83s/it] 19%|█▉ | 5654/30000 [3:37:30<17:50:12, 2.64s/it] 19%|█▉ | 5655/30000 [3:37:32<17:06:26, 2.53s/it] 19%|█▉ | 5656/30000 [3:37:34<16:41:37, 2.47s/it] 19%|█▉ | 5657/30000 [3:37:37<16:20:15, 2.42s/it] 19%|█▉ | 5658/30000 [3:37:39<16:03:26, 2.37s/it] 19%|█▉ | 5659/30000 [3:37:41<15:59:31, 2.37s/it] 19%|█▉ | 5660/30000 [3:37:44<16:00:34, 2.37s/it] 19%|█▉ | 5660/30000 [3:37:44<16:00:34, 2.37s/it] 19%|█▉ | 5661/30000 [3:37:49<21:41:20, 3.21s/it] 19%|█▉ | 5662/30000 [3:37:51<19:47:04, 2.93s/it] 19%|█▉ | 5663/30000 [3:37:53<18:22:37, 2.72s/it] 19%|█▉ | 5664/30000 [3:37:56<17:30:44, 2.59s/it] 19%|█▉ | 5665/30000 [3:37:58<16:43:44, 2.47s/it] 19%|█▉ | 5666/30000 [3:38:00<16:20:09, 2.42s/it] 19%|█▉ | 5667/30000 [3:38:02<16:10:38, 2.39s/it] 19%|█▉ | 5668/30000 [3:38:04<15:15:56, 2.26s/it] 19%|█▉ | 5669/30000 [3:38:09<20:11:28, 2.99s/it] 19%|█▉ | 5670/30000 [3:38:11<17:47:00, 2.63s/it] 19%|█▉ | 5670/30000 [3:38:11<17:47:00, 2.63s/it] 19%|█▉ | 5671/30000 [3:38:13<17:06:27, 2.53s/it] 19%|█▉ | 5672/30000 [3:38:15<16:32:46, 2.45s/it] 19%|█▉ | 5673/30000 [3:38:18<16:22:47, 2.42s/it] 19%|█▉ | 5674/30000 [3:38:20<15:31:06, 2.30s/it] 19%|█▉ | 5675/30000 [3:38:22<14:17:20, 2.11s/it] 19%|█▉ | 5676/30000 [3:38:23<12:46:22, 1.89s/it] 19%|█▉ | 5677/30000 [3:38:30<23:27:53, 3.47s/it] 19%|█▉ | 5678/30000 [3:38:32<21:09:36, 3.13s/it] 19%|█▉ | 5679/30000 [3:38:34<18:23:18, 2.72s/it] 19%|█▉ | 5680/30000 [3:38:36<17:25:31, 2.58s/it] 19%|█▉ | 5680/30000 [3:38:36<17:25:31, 2.58s/it] 19%|█▉ | 5681/30000 [3:38:39<16:47:34, 2.49s/it] 19%|█▉ | 5682/30000 [3:38:41<16:30:32, 2.44s/it] 19%|█▉ | 5683/30000 [3:38:43<16:03:59, 2.38s/it] 19%|█▉ | 5684/30000 [3:38:45<15:50:43, 2.35s/it] 19%|█▉ | 5685/30000 [3:38:48<15:51:40, 2.35s/it] 19%|█▉ | 5686/30000 [3:38:50<15:49:40, 2.34s/it] 19%|█▉ | 5687/30000 [3:38:52<15:41:56, 2.32s/it] 19%|█▉ | 5688/30000 [3:38:55<15:43:11, 2.33s/it] 19%|█▉ | 5689/30000 [3:38:57<15:38:51, 2.32s/it] 19%|█▉ | 5690/30000 [3:38:59<15:34:14, 2.31s/it] 19%|█▉ | 5690/30000 [3:38:59<15:34:14, 2.31s/it] 19%|█▉ | 5691/30000 [3:39:02<15:33:29, 2.30s/it] 19%|█▉ | 5692/30000 [3:39:04<15:39:41, 2.32s/it] 19%|█▉ | 5693/30000 [3:39:06<15:37:50, 2.31s/it] 19%|█▉ | 5694/30000 [3:39:09<15:32:25, 2.30s/it] 19%|█▉ | 5695/30000 [3:39:11<15:36:16, 2.31s/it] 19%|█▉ | 5696/30000 [3:39:13<15:31:56, 2.30s/it] 19%|█▉ | 5697/30000 [3:39:15<15:29:27, 2.29s/it] 19%|█▉ | 5698/30000 [3:39:18<15:33:56, 2.31s/it] 19%|█▉ | 5699/30000 [3:39:20<15:31:07, 2.30s/it] 19%|█▉ | 5700/30000 [3:39:22<15:26:32, 2.29s/it] 19%|█▉ | 5700/30000 [3:39:22<15:26:32, 2.29s/it] 19%|█▉ | 5701/30000 [3:39:25<15:29:11, 2.29s/it] 19%|█▉ | 5702/30000 [3:39:27<15:27:36, 2.29s/it] 19%|█▉ | 5703/30000 [3:39:29<15:23:29, 2.28s/it] 19%|█▉ | 5704/30000 [3:39:31<15:19:39, 2.27s/it] 19%|█▉ | 5705/30000 [3:39:34<15:24:40, 2.28s/it] 19%|█▉ | 5706/30000 [3:39:36<15:22:53, 2.28s/it] 19%|█▉ | 5707/30000 [3:39:38<15:24:13, 2.28s/it] 19%|█▉ | 5708/30000 [3:39:41<15:35:45, 2.31s/it] 19%|█▉ | 5709/30000 [3:39:43<15:38:37, 2.32s/it] 19%|█▉ | 5710/30000 [3:39:45<15:28:04, 2.29s/it] 19%|█▉ | 5710/30000 [3:39:45<15:28:04, 2.29s/it] 19%|█▉ | 5711/30000 [3:39:48<15:35:52, 2.31s/it] 19%|█▉ | 5712/30000 [3:39:50<15:35:34, 2.31s/it] 19%|█▉ | 5713/30000 [3:39:52<15:35:15, 2.31s/it] 19%|█▉ | 5714/30000 [3:39:55<15:31:15, 2.30s/it] 19%|█▉ | 5715/30000 [3:39:57<15:36:21, 2.31s/it] 19%|█▉ | 5716/30000 [3:39:59<15:32:15, 2.30s/it] 19%|█▉ | 5717/30000 [3:40:01<15:27:46, 2.29s/it] 19%|█▉ | 5718/30000 [3:40:04<15:38:04, 2.32s/it] 19%|█▉ | 5719/30000 [3:40:06<15:33:07, 2.31s/it] 19%|█▉ | 5720/30000 [3:40:08<15:27:14, 2.29s/it] 19%|█▉ | 5720/30000 [3:40:08<15:27:14, 2.29s/it] 19%|█▉ | 5721/30000 [3:40:11<15:27:41, 2.29s/it] 19%|█▉ | 5722/30000 [3:40:13<15:34:08, 2.31s/it] 19%|█▉ | 5723/30000 [3:40:15<15:29:12, 2.30s/it] 19%|█▉ | 5724/30000 [3:40:18<15:28:24, 2.29s/it] 19%|█▉ | 5725/30000 [3:40:20<15:32:23, 2.30s/it] 19%|█▉ | 5726/30000 [3:40:22<15:28:55, 2.30s/it] 19%|█▉ | 5727/30000 [3:40:24<15:26:26, 2.29s/it] 19%|█▉ | 5728/30000 [3:40:27<15:24:24, 2.29s/it] 19%|█▉ | 5729/30000 [3:40:29<15:29:32, 2.30s/it] 19%|█▉ | 5730/30000 [3:40:31<15:28:07, 2.29s/it] 19%|█▉ | 5730/30000 [3:40:31<15:28:07, 2.29s/it] 19%|█▉ | 5731/30000 [3:40:34<15:27:05, 2.29s/it] 19%|█▉ | 5732/30000 [3:40:36<15:34:01, 2.31s/it] 19%|█▉ | 5733/30000 [3:40:38<15:35:44, 2.31s/it] 19%|█▉ | 5734/30000 [3:40:41<15:29:23, 2.30s/it] 19%|█▉ | 5735/30000 [3:40:43<15:25:48, 2.29s/it] 19%|█▉ | 5736/30000 [3:40:45<15:33:20, 2.31s/it] 19%|█▉ | 5737/30000 [3:40:47<15:16:01, 2.27s/it] 19%|█▉ | 5738/30000 [3:40:50<15:15:16, 2.26s/it] 19%|█▉ | 5739/30000 [3:40:52<15:26:53, 2.29s/it] 19%|█▉ | 5740/30000 [3:40:54<15:28:38, 2.30s/it] 19%|█▉ | 5740/30000 [3:40:54<15:28:38, 2.30s/it] 19%|█▉ | 5741/30000 [3:40:57<15:23:56, 2.29s/it] 19%|█▉ | 5742/30000 [3:40:59<15:24:47, 2.29s/it] 19%|█▉ | 5743/30000 [3:41:01<15:23:14, 2.28s/it] 19%|█▉ | 5744/30000 [3:41:03<15:22:40, 2.28s/it] 19%|█▉ | 5745/30000 [3:41:06<15:23:24, 2.28s/it] 19%|█▉ | 5746/30000 [3:41:08<15:28:56, 2.30s/it] 19%|█▉ | 5747/30000 [3:41:10<15:31:43, 2.31s/it] 19%|█▉ | 5748/30000 [3:41:13<15:29:04, 2.30s/it] 19%|█▉ | 5749/30000 [3:41:15<15:24:33, 2.29s/it] 19%|█▉ | 5750/30000 [3:41:17<15:29:29, 2.30s/it] 19%|█▉ | 5750/30000 [3:41:17<15:29:29, 2.30s/it] 19%|█▉ | 5751/30000 [3:41:19<15:31:55, 2.31s/it] 19%|█▉ | 5752/30000 [3:41:22<15:28:34, 2.30s/it] 19%|█▉ | 5753/30000 [3:41:24<15:32:17, 2.31s/it] 19%|█▉ | 5754/30000 [3:41:26<15:36:06, 2.32s/it] 19%|█▉ | 5755/30000 [3:41:29<15:35:35, 2.32s/it] 19%|█▉ | 5756/30000 [3:41:31<15:28:30, 2.30s/it] 19%|█▉ | 5757/30000 [3:41:33<15:24:31, 2.29s/it] 19%|█▉ | 5758/30000 [3:41:36<15:24:34, 2.29s/it] 19%|█▉ | 5759/30000 [3:41:38<15:23:36, 2.29s/it] 19%|█▉ | 5760/30000 [3:41:40<15:19:30, 2.28s/it] 19%|█▉ | 5760/30000 [3:41:40<15:19:30, 2.28s/it] 19%|█▉ | 5761/30000 [3:41:42<15:28:20, 2.30s/it] 19%|█▉ | 5762/30000 [3:41:45<15:31:35, 2.31s/it] 19%|█▉ | 5763/30000 [3:41:47<15:27:32, 2.30s/it] 19%|█▉ | 5764/30000 [3:41:49<15:26:18, 2.29s/it] 19%|█▉ | 5765/30000 [3:41:52<15:28:28, 2.30s/it] 19%|█▉ | 5766/30000 [3:41:54<15:33:42, 2.31s/it] 19%|█▉ | 5767/30000 [3:41:56<15:30:25, 2.30s/it] 19%|█▉ | 5768/30000 [3:41:59<15:25:31, 2.29s/it] 19%|█▉ | 5769/30000 [3:42:01<15:34:41, 2.31s/it] 19%|█▉ | 5770/30000 [3:42:03<15:30:55, 2.31s/it] 19%|█▉ | 5770/30000 [3:42:03<15:30:55, 2.31s/it] 19%|█▉ | 5771/30000 [3:42:05<15:22:23, 2.28s/it] 19%|█▉ | 5772/30000 [3:42:08<15:17:42, 2.27s/it] 19%|█▉ | 5773/30000 [3:42:10<15:22:31, 2.28s/it] 19%|█▉ | 5774/30000 [3:42:12<15:22:51, 2.29s/it] 19%|█▉ | 5775/30000 [3:42:14<15:17:12, 2.27s/it] 19%|█▉ | 5776/30000 [3:42:17<15:25:27, 2.29s/it] 19%|█▉ | 5777/30000 [3:42:19<15:26:29, 2.29s/it] 19%|█▉ | 5778/30000 [3:42:21<15:22:27, 2.28s/it] 19%|█▉ | 5779/30000 [3:42:24<15:21:22, 2.28s/it] 19%|█▉ | 5780/30000 [3:42:26<15:24:33, 2.29s/it] 19%|█▉ | 5780/30000 [3:42:26<15:24:33, 2.29s/it] 19%|█▉ | 5781/30000 [3:42:28<15:27:50, 2.30s/it] 19%|█▉ | 5782/30000 [3:42:31<15:22:55, 2.29s/it] 19%|█▉ | 5783/30000 [3:42:33<15:25:18, 2.29s/it] 19%|█▉ | 5784/30000 [3:42:35<15:24:18, 2.29s/it] 19%|█▉ | 5785/30000 [3:42:37<15:20:52, 2.28s/it] 19%|█▉ | 5786/30000 [3:42:40<15:21:06, 2.28s/it] 19%|█▉ | 5787/30000 [3:42:42<15:27:01, 2.30s/it] 19%|█▉ | 5788/30000 [3:42:44<15:28:59, 2.30s/it] 19%|█▉ | 5789/30000 [3:42:47<15:26:16, 2.30s/it] 19%|█▉ | 5790/30000 [3:42:49<15:28:45, 2.30s/it] 19%|█▉ | 5790/30000 [3:42:49<15:28:45, 2.30s/it] 19%|█▉ | 5791/30000 [3:42:51<15:32:02, 2.31s/it] 19%|█▉ | 5792/30000 [3:42:54<15:25:53, 2.29s/it] 19%|█▉ | 5793/30000 [3:42:56<15:23:45, 2.29s/it] 19%|█▉ | 5794/30000 [3:42:58<15:26:13, 2.30s/it] 19%|█▉ | 5795/30000 [3:43:00<15:20:09, 2.28s/it] 19%|█▉ | 5796/30000 [3:43:03<15:18:05, 2.28s/it] 19%|█▉ | 5797/30000 [3:43:05<15:27:18, 2.30s/it] 19%|█▉ | 5798/30000 [3:43:07<15:25:38, 2.29s/it] 19%|█▉ | 5799/30000 [3:43:10<15:23:38, 2.29s/it] 19%|█▉ | 5800/30000 [3:43:12<15:31:13, 2.31s/it] 19%|█▉ | 5800/30000 [3:43:12<15:31:13, 2.31s/it] 19%|█▉ | 5801/30000 [3:43:14<15:31:59, 2.31s/it] 19%|█▉ | 5802/30000 [3:43:16<15:28:27, 2.30s/it] 19%|█▉ | 5803/30000 [3:43:19<15:30:04, 2.31s/it] 19%|█▉ | 5804/30000 [3:43:21<15:30:45, 2.31s/it] 19%|█▉ | 5805/30000 [3:43:23<15:26:47, 2.30s/it] 19%|█▉ | 5806/30000 [3:43:26<15:25:19, 2.29s/it] 19%|█▉ | 5807/30000 [3:43:28<15:32:04, 2.31s/it] 19%|█▉ | 5808/30000 [3:43:30<15:35:25, 2.32s/it] 19%|█▉ | 5809/30000 [3:43:33<15:36:10, 2.32s/it] 19%|█▉ | 5810/30000 [3:43:35<15:36:39, 2.32s/it] 19%|█▉ | 5810/30000 [3:43:35<15:36:39, 2.32s/it] 19%|█▉ | 5811/30000 [3:43:37<15:32:09, 2.31s/it] 19%|█▉ | 5812/30000 [3:43:40<15:27:02, 2.30s/it] 19%|█▉ | 5813/30000 [3:43:42<15:27:46, 2.30s/it] 19%|█▉ | 5814/30000 [3:43:44<15:29:08, 2.30s/it] 19%|█▉ | 5815/30000 [3:43:46<15:25:39, 2.30s/it] 19%|█▉ | 5816/30000 [3:43:49<15:23:13, 2.29s/it] 19%|█▉ | 5817/30000 [3:43:51<15:31:19, 2.31s/it] 19%|█▉ | 5818/30000 [3:43:53<15:20:05, 2.28s/it] 19%|█▉ | 5819/30000 [3:43:56<15:16:00, 2.27s/it] 19%|█▉ | 5820/30000 [3:43:58<15:25:10, 2.30s/it] 19%|█▉ | 5820/30000 [3:43:58<15:25:10, 2.30s/it] 19%|█▉ | 5821/30000 [3:44:00<15:32:22, 2.31s/it] 19%|█▉ | 5822/30000 [3:44:03<15:23:59, 2.29s/it] 19%|█▉ | 5823/30000 [3:44:05<15:31:32, 2.31s/it] 19%|█▉ | 5824/30000 [3:44:07<15:28:54, 2.31s/it] 19%|█▉ | 5825/30000 [3:44:09<15:23:30, 2.29s/it] 19%|█▉ | 5826/30000 [3:44:12<15:29:15, 2.31s/it] 19%|█▉ | 5827/30000 [3:44:14<15:29:32, 2.31s/it] 19%|█▉ | 5828/30000 [3:44:16<15:10:45, 2.26s/it] 19%|█▉ | 5829/30000 [3:44:19<15:34:46, 2.32s/it] 19%|█▉ | 5830/30000 [3:44:21<15:31:00, 2.31s/it] 19%|█▉ | 5830/30000 [3:44:21<15:31:00, 2.31s/it] 19%|█▉ | 5831/30000 [3:44:23<15:26:21, 2.30s/it] 19%|█▉ | 5832/30000 [3:44:26<15:29:56, 2.31s/it] 19%|█▉ | 5833/30000 [3:44:28<15:31:37, 2.31s/it] 19%|█▉ | 5834/30000 [3:44:30<15:25:49, 2.30s/it] 19%|█▉ | 5835/30000 [3:44:32<15:23:22, 2.29s/it] 19%|█▉ | 5836/30000 [3:44:35<15:28:03, 2.30s/it] 19%|█▉ | 5837/30000 [3:44:37<15:21:19, 2.29s/it] 19%|█▉ | 5838/30000 [3:44:39<15:17:46, 2.28s/it] 19%|█▉ | 5839/30000 [3:44:42<15:24:27, 2.30s/it] 19%|█▉ | 5840/30000 [3:44:44<15:22:10, 2.29s/it] 19%|█▉ | 5840/30000 [3:44:44<15:22:10, 2.29s/it] 19%|█▉ | 5841/30000 [3:44:46<15:22:34, 2.29s/it] 19%|█▉ | 5842/30000 [3:44:49<15:27:32, 2.30s/it] 19%|█▉ | 5843/30000 [3:44:51<15:28:24, 2.31s/it] 19%|█▉ | 5844/30000 [3:44:53<15:25:34, 2.30s/it] 19%|█▉ | 5845/30000 [3:44:55<15:24:54, 2.30s/it] 19%|█▉ | 5846/30000 [3:44:58<15:20:10, 2.29s/it] 19%|█▉ | 5847/30000 [3:45:00<15:17:06, 2.28s/it] 19%|█▉ | 5848/30000 [3:45:02<15:18:31, 2.28s/it] 19%|█▉ | 5849/30000 [3:45:05<15:28:18, 2.31s/it] 20%|█▉ | 5850/30000 [3:45:07<15:28:08, 2.31s/it] 20%|█▉ | 5850/30000 [3:45:07<15:28:08, 2.31s/it] 20%|█▉ | 5851/30000 [3:45:09<15:25:03, 2.30s/it] 20%|█▉ | 5852/30000 [3:45:12<15:29:12, 2.31s/it] 20%|█▉ | 5853/30000 [3:45:14<15:29:06, 2.31s/it] 20%|█▉ | 5854/30000 [3:45:16<15:23:10, 2.29s/it] 20%|█▉ | 5855/30000 [3:45:18<15:29:31, 2.31s/it] 20%|█▉ | 5856/30000 [3:45:21<15:25:58, 2.30s/it] 20%|█▉ | 5857/30000 [3:45:23<15:24:10, 2.30s/it] 20%|█▉ | 5858/30000 [3:45:25<15:22:39, 2.29s/it] 20%|█▉ | 5859/30000 [3:45:28<15:31:45, 2.32s/it] 20%|█▉ | 5860/30000 [3:45:30<15:25:02, 2.30s/it] 20%|█▉ | 5860/30000 [3:45:30<15:25:02, 2.30s/it] 20%|█▉ | 5861/30000 [3:45:32<15:22:29, 2.29s/it] 20%|█▉ | 5862/30000 [3:45:35<15:28:35, 2.31s/it] 20%|█▉ | 5863/30000 [3:45:37<15:30:27, 2.31s/it] 20%|█▉ | 5864/30000 [3:45:39<15:25:19, 2.30s/it] 20%|█▉ | 5865/30000 [3:45:41<15:29:23, 2.31s/it] 20%|█▉ | 5866/30000 [3:45:44<15:23:11, 2.30s/it] 20%|█▉ | 5867/30000 [3:45:46<15:21:18, 2.29s/it] 20%|█▉ | 5868/30000 [3:45:48<15:19:53, 2.29s/it] 20%|█▉ | 5869/30000 [3:45:51<15:26:45, 2.30s/it] 20%|█▉ | 5870/30000 [3:45:53<15:24:09, 2.30s/it] +{'loss': 0.0181, 'grad_norm': 0.10895755141973495, 'learning_rate': 2.8587153686843626e-05, 'flow_matching_loss': 0.018059553019702433, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.59} +{'loss': 0.0161, 'grad_norm': 0.10499639809131622, 'learning_rate': 2.858013995286594e-05, 'flow_matching_loss': 0.016144927684217693, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.59} +{'loss': 0.0181, 'grad_norm': 0.09541905671358109, 'learning_rate': 2.857310971773769e-05, 'flow_matching_loss': 0.01809315672144294, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.6} +{'loss': 0.0178, 'grad_norm': 0.10060036182403564, 'learning_rate': 2.8566062990001274e-05, 'flow_matching_loss': 0.01778232790529728, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.6} +{'loss': 0.0174, 'grad_norm': 0.09510762989521027, 'learning_rate': 2.8558999778219126e-05, 'flow_matching_loss': 0.017414864525198935, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.6} +{'loss': 0.0162, 'grad_norm': 0.08502375334501266, 'learning_rate': 2.8551920090973722e-05, 'flow_matching_loss': 0.01616199463605881, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.61} +{'loss': 0.0176, 'grad_norm': 0.07633312791585922, 'learning_rate': 2.854482393686754e-05, 'flow_matching_loss': 0.017623922694474457, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.61} +{'loss': 0.0177, 'grad_norm': 0.09117305278778076, 'learning_rate': 2.8537711324523086e-05, 'flow_matching_loss': 0.017661013826727866, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.61} +{'loss': 0.0164, 'grad_norm': 0.08659131079912186, 'learning_rate': 2.8530582262582845e-05, 'flow_matching_loss': 0.016406095679849388, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.61} +{'loss': 0.0175, 'grad_norm': 0.09384981542825699, 'learning_rate': 2.8523436759709306e-05, 'flow_matching_loss': 0.017479505948722364, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.62} +{'loss': 0.018, 'grad_norm': 0.0737551599740982, 'learning_rate': 2.8516274824584926e-05, 'flow_matching_loss': 0.018003450240939856, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.62} +{'loss': 0.0152, 'grad_norm': 0.08111722022294998, 'learning_rate': 2.8509096465912136e-05, 'flow_matching_loss': 0.015221840143203736, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.62} +{'loss': 0.018, 'grad_norm': 0.07046273350715637, 'learning_rate': 2.8501901692413315e-05, 'flow_matching_loss': 0.017960367631167172, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.63} +{'loss': 0.0182, 'grad_norm': 0.07960425317287445, 'learning_rate': 2.8494690512830788e-05, 'flow_matching_loss': 0.01822669245302677, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.63} +{'loss': 0.0162, 'grad_norm': 0.06337831914424896, 'learning_rate': 2.8487462935926823e-05, 'flow_matching_loss': 0.016237550415098667, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.63} +{'loss': 0.0187, 'grad_norm': 0.09085789322853088, 'learning_rate': 2.8480218970483606e-05, 'flow_matching_loss': 0.018720805179327726, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.64} +{'loss': 0.0172, 'grad_norm': 0.08947379887104034, 'learning_rate': 2.847295862530324e-05, 'flow_matching_loss': 0.01724295550957322, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.64} +{'loss': 0.0156, 'grad_norm': 0.08142384141683578, 'learning_rate': 2.846568190920773e-05, 'flow_matching_loss': 0.015557727962732314, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.64} +{'loss': 0.015, 'grad_norm': 0.08896787464618683, 'learning_rate': 2.8458388831038965e-05, 'flow_matching_loss': 0.014966677408665419, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.64} +{'loss': 0.0161, 'grad_norm': 0.09310554713010788, 'learning_rate': 2.8451079399658725e-05, 'flow_matching_loss': 0.016115553118288516, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.65} +{'loss': 0.0168, 'grad_norm': 0.13455618917942047, 'learning_rate': 2.8443753623948665e-05, 'flow_matching_loss': 0.01682263296097517, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.65} +{'loss': 0.0196, 'grad_norm': 0.1290634125471115, 'learning_rate': 2.8436411512810288e-05, 'flow_matching_loss': 0.019589664414525032, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.65} +{'loss': 0.0182, 'grad_norm': 0.11339693516492844, 'learning_rate': 2.842905307516495e-05, 'flow_matching_loss': 0.018188603036105633, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.66} +{'loss': 0.0182, 'grad_norm': 0.0911894291639328, 'learning_rate': 2.8421678319953854e-05, 'flow_matching_loss': 0.018213918153196573, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.66} +{'loss': 0.0167, 'grad_norm': 0.08298409730195999, 'learning_rate': 2.841428725613801e-05, 'flow_matching_loss': 0.016742206830531358, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.66} +{'loss': 0.0163, 'grad_norm': 0.09323465079069138, 'learning_rate': 2.8406879892698272e-05, 'flow_matching_loss': 0.016323442477732895, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.66} +{'loss': 0.0167, 'grad_norm': 0.06632257252931595, 'learning_rate': 2.8399456238635274e-05, 'flow_matching_loss': 0.016671096812933682, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.67} +{'loss': 0.0159, 'grad_norm': 0.07370807230472565, 'learning_rate': 2.8392016302969457e-05, 'flow_matching_loss': 0.015876694489270447, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.67} +{'loss': 0.0155, 'grad_norm': 0.0818537175655365, 'learning_rate': 2.8384560094741055e-05, 'flow_matching_loss': 0.01548408530652523, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.67} +{'loss': 0.0182, 'grad_norm': 0.10679682344198227, 'learning_rate': 2.8377087623010054e-05, 'flow_matching_loss': 0.018204374704509974, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.68} +{'loss': 0.0192, 'grad_norm': 0.09296752512454987, 'learning_rate': 2.8369598896856213e-05, 'flow_matching_loss': 0.019166892766952513, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.68} +{'loss': 0.0188, 'grad_norm': 0.0971725657582283, 'learning_rate': 2.836209392537904e-05, 'flow_matching_loss': 0.018760529905557634, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.68} +{'loss': 0.0171, 'grad_norm': 0.09415116161108017, 'learning_rate': 2.8354572717697787e-05, 'flow_matching_loss': 0.017140307836234568, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.68} +{'loss': 0.0173, 'grad_norm': 0.09866904467344284, 'learning_rate': 2.8347035282951428e-05, 'flow_matching_loss': 0.017298791743814945, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.69} +{'loss': 0.0202, 'grad_norm': 0.10787111520767212, 'learning_rate': 2.8339481630298653e-05, 'flow_matching_loss': 0.020154422149062157, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.69} +{'loss': 0.0193, 'grad_norm': 0.09033793956041336, 'learning_rate': 2.833191176891787e-05, 'flow_matching_loss': 0.01932883858680725, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.69} +{'loss': 0.0173, 'grad_norm': 0.09649154543876648, 'learning_rate': 2.832432570800717e-05, 'flow_matching_loss': 0.017310749925673007, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.7} +{'loss': 0.0157, 'grad_norm': 0.08289851248264313, 'learning_rate': 2.8316723456784332e-05, 'flow_matching_loss': 0.015698580630123617, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.7} +{'loss': 0.0167, 'grad_norm': 0.10048871487379074, 'learning_rate': 2.8309105024486808e-05, 'flow_matching_loss': 0.01670493567362428, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.7} +{'loss': 0.0173, 'grad_norm': 0.0845789685845375, 'learning_rate': 2.830147042037172e-05, 'flow_matching_loss': 0.017347517609596252, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.7} +{'loss': 0.0182, 'grad_norm': 0.07265578955411911, 'learning_rate': 2.8293819653715816e-05, 'flow_matching_loss': 0.0181811167858541, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.71} 20%|█▉ | 5870/30000 [3:45:53<15:24:09, 2.30s/it] 20%|█▉ | 5871/30000 [3:45:55<15:19:44, 2.29s/it] 20%|█▉ | 5872/30000 [3:45:58<15:25:37, 2.30s/it] 20%|█▉ | 5873/30000 [3:46:00<15:25:55, 2.30s/it] 20%|█▉ | 5874/30000 [3:46:02<15:22:26, 2.29s/it] 20%|█▉ | 5875/30000 [3:46:04<15:20:09, 2.29s/it] 20%|█▉ | 5876/30000 [3:46:07<15:26:49, 2.31s/it] 20%|█▉ | 5877/30000 [3:46:09<15:22:03, 2.29s/it] 20%|█▉ | 5878/30000 [3:46:11<15:18:17, 2.28s/it] 20%|█▉ | 5879/30000 [3:46:14<15:24:00, 2.30s/it] 20%|█▉ | 5880/30000 [3:46:16<15:24:40, 2.30s/it] 20%|█▉ | 5880/30000 [3:46:16<15:24:40, 2.30s/it] 20%|█▉ | 5881/30000 [3:46:18<15:22:04, 2.29s/it] 20%|█▉ | 5882/30000 [3:46:20<15:19:43, 2.29s/it] 20%|█▉ | 5883/30000 [3:46:23<15:27:51, 2.31s/it] 20%|█▉ | 5884/30000 [3:46:25<15:21:10, 2.29s/it] 20%|█▉ | 5885/30000 [3:46:27<15:19:40, 2.29s/it] 20%|█▉ | 5886/30000 [3:46:30<15:27:30, 2.31s/it] 20%|█▉ | 5887/30000 [3:46:32<15:26:20, 2.30s/it] 20%|█▉ | 5888/30000 [3:46:34<15:19:57, 2.29s/it] 20%|█▉ | 5889/30000 [3:46:37<15:20:35, 2.29s/it] 20%|█▉ | 5890/30000 [3:46:39<15:25:08, 2.30s/it] 20%|█▉ | 5890/30000 [3:46:39<15:25:08, 2.30s/it] 20%|█▉ | 5891/30000 [3:46:41<15:25:04, 2.30s/it] 20%|█▉ | 5892/30000 [3:46:43<15:23:49, 2.30s/it] 20%|█▉ | 5893/30000 [3:46:46<15:28:01, 2.31s/it] 20%|█▉ | 5894/30000 [3:46:48<15:26:32, 2.31s/it] 20%|█▉ | 5895/30000 [3:46:50<15:20:33, 2.29s/it] 20%|█▉ | 5896/30000 [3:46:53<15:21:42, 2.29s/it] 20%|█▉ | 5897/30000 [3:46:55<15:26:31, 2.31s/it] 20%|█▉ | 5898/30000 [3:46:57<15:21:57, 2.30s/it] 20%|█▉ | 5899/30000 [3:46:59<15:16:29, 2.28s/it] 20%|█▉ | 5900/30000 [3:47:02<15:24:42, 2.30s/it] 20%|█▉ | 5900/30000 [3:47:02<15:24:42, 2.30s/it] 20%|█▉ | 5901/30000 [3:47:04<15:22:10, 2.30s/it] 20%|█▉ | 5902/30000 [3:47:06<15:17:51, 2.29s/it] 20%|█▉ | 5903/30000 [3:47:09<15:24:13, 2.30s/it] 20%|█▉ | 5904/30000 [3:47:11<15:19:38, 2.29s/it] 20%|█▉ | 5905/30000 [3:47:13<15:19:02, 2.29s/it] 20%|█▉ | 5906/30000 [3:47:16<15:15:00, 2.28s/it] 20%|█▉ | 5907/30000 [3:47:18<15:23:43, 2.30s/it] 20%|█▉ | 5908/30000 [3:47:20<15:27:24, 2.31s/it] 20%|█▉ | 5909/30000 [3:47:22<15:23:24, 2.30s/it] 20%|█▉ | 5910/30000 [3:47:25<15:18:06, 2.29s/it] 20%|█▉ | 5910/30000 [3:47:25<15:18:06, 2.29s/it] 20%|█▉ | 5911/30000 [3:47:27<15:25:49, 2.31s/it] 20%|█▉ | 5912/30000 [3:47:29<15:17:30, 2.29s/it] 20%|█▉ | 5913/30000 [3:47:32<15:16:21, 2.28s/it] 20%|█▉ | 5914/30000 [3:47:34<15:26:50, 2.31s/it] 20%|█▉ | 5915/30000 [3:47:36<15:28:37, 2.31s/it] 20%|█▉ | 5916/30000 [3:47:39<15:24:32, 2.30s/it] 20%|█▉ | 5917/30000 [3:47:41<15:20:28, 2.29s/it] 20%|█▉ | 5918/30000 [3:47:43<15:24:10, 2.30s/it] 20%|█▉ | 5919/30000 [3:47:46<15:29:47, 2.32s/it] 20%|█▉ | 5920/30000 [3:47:48<15:20:03, 2.29s/it] 20%|█▉ | 5920/30000 [3:47:48<15:20:03, 2.29s/it] 20%|█▉ | 5921/30000 [3:47:50<15:16:06, 2.28s/it] 20%|█▉ | 5922/30000 [3:47:52<15:21:34, 2.30s/it] 20%|█▉ | 5923/30000 [3:47:55<15:21:59, 2.30s/it] 20%|█▉ | 5924/30000 [3:47:57<15:15:52, 2.28s/it] 20%|█▉ | 5925/30000 [3:47:59<15:16:48, 2.28s/it] 20%|█▉ | 5926/30000 [3:48:01<15:14:25, 2.28s/it] 20%|█▉ | 5927/30000 [3:48:04<15:15:49, 2.28s/it] 20%|█▉ | 5928/30000 [3:48:06<15:12:32, 2.27s/it] 20%|█▉ | 5929/30000 [3:48:08<15:21:34, 2.30s/it] 20%|█▉ | 5930/30000 [3:48:11<15:16:48, 2.29s/it] 20%|█▉ | 5930/30000 [3:48:11<15:16:48, 2.29s/it] 20%|█▉ | 5931/30000 [3:48:13<15:17:15, 2.29s/it] 20%|█▉ | 5932/30000 [3:48:15<15:10:03, 2.27s/it] 20%|█▉ | 5933/30000 [3:48:17<15:15:48, 2.28s/it] 20%|█▉ | 5934/30000 [3:48:20<15:19:24, 2.29s/it] 20%|█▉ | 5935/30000 [3:48:22<15:21:34, 2.30s/it] 20%|█▉ | 5936/30000 [3:48:24<15:20:28, 2.30s/it] 20%|█▉ | 5937/30000 [3:48:27<15:25:23, 2.31s/it] 20%|█▉ | 5938/30000 [3:48:29<15:26:07, 2.31s/it] 20%|█▉ | 5939/30000 [3:48:31<15:20:00, 2.29s/it] 20%|█▉ | 5940/30000 [3:48:34<15:20:38, 2.30s/it] 20%|█▉ | 5940/30000 [3:48:34<15:20:38, 2.30s/it] 20%|█▉ | 5941/30000 [3:48:36<15:23:39, 2.30s/it] 20%|█▉ | 5942/30000 [3:48:38<15:20:06, 2.29s/it] 20%|█▉ | 5943/30000 [3:48:40<15:16:50, 2.29s/it] 20%|█▉ | 5944/30000 [3:48:43<15:15:37, 2.28s/it] 20%|█▉ | 5945/30000 [3:48:45<15:22:47, 2.30s/it] 20%|█▉ | 5946/30000 [3:48:47<15:26:15, 2.31s/it] 20%|█▉ | 5947/30000 [3:48:50<15:24:38, 2.31s/it] 20%|█▉ | 5948/30000 [3:48:52<15:17:55, 2.29s/it] 20%|█▉ | 5949/30000 [3:48:54<15:22:24, 2.30s/it] 20%|█▉ | 5950/30000 [3:48:57<15:22:09, 2.30s/it] 20%|█▉ | 5950/30000 [3:48:57<15:22:09, 2.30s/it] 20%|█▉ | 5951/30000 [3:48:59<15:17:37, 2.29s/it] 20%|█▉ | 5952/30000 [3:49:01<15:14:12, 2.28s/it] 20%|█▉ | 5953/30000 [3:49:03<15:17:46, 2.29s/it] 20%|█▉ | 5954/30000 [3:49:06<15:12:28, 2.28s/it] 20%|█▉ | 5955/30000 [3:49:08<15:14:28, 2.28s/it] 20%|█▉ | 5956/30000 [3:49:10<15:19:41, 2.30s/it] 20%|█▉ | 5957/30000 [3:49:13<15:23:08, 2.30s/it] 20%|█▉ | 5958/30000 [3:49:15<15:19:59, 2.30s/it] 20%|█▉ | 5959/30000 [3:49:17<15:19:23, 2.29s/it] 20%|█▉ | 5960/30000 [3:49:19<15:22:42, 2.30s/it] 20%|█▉ | 5960/30000 [3:49:19<15:22:42, 2.30s/it] 20%|█▉ | 5961/30000 [3:49:22<15:16:56, 2.29s/it] 20%|█▉ | 5962/30000 [3:49:24<15:21:01, 2.30s/it] 20%|█▉ | 5963/30000 [3:49:26<15:24:01, 2.31s/it] 20%|█▉ | 5964/30000 [3:49:29<15:25:06, 2.31s/it] 20%|█▉ | 5965/30000 [3:49:31<15:16:32, 2.29s/it] 20%|█▉ | 5966/30000 [3:49:33<15:15:04, 2.28s/it] 20%|█▉ | 5967/30000 [3:49:36<15:20:36, 2.30s/it] 20%|█▉ | 5968/30000 [3:49:38<15:16:41, 2.29s/it] 20%|█▉ | 5969/30000 [3:49:40<15:16:44, 2.29s/it] 20%|█▉ | 5970/30000 [3:49:42<15:13:29, 2.28s/it] 20%|█▉ | 5970/30000 [3:49:42<15:13:29, 2.28s/it] 20%|█▉ | 5971/30000 [3:49:45<15:20:43, 2.30s/it] 20%|█▉ | 5972/30000 [3:49:47<15:22:09, 2.30s/it] 20%|█▉ | 5973/30000 [3:49:49<15:16:37, 2.29s/it] 20%|█▉ | 5974/30000 [3:49:52<15:24:10, 2.31s/it] 20%|█▉ | 5975/30000 [3:49:54<15:19:09, 2.30s/it] 20%|█▉ | 5976/30000 [3:49:56<15:16:36, 2.29s/it] 20%|█▉ | 5977/30000 [3:49:58<15:12:55, 2.28s/it] 20%|█▉ | 5978/30000 [3:50:01<15:21:08, 2.30s/it] 20%|█▉ | 5979/30000 [3:50:03<15:24:06, 2.31s/it] 20%|█▉ | 5980/30000 [3:50:05<15:21:39, 2.30s/it] 20%|█▉ | 5980/30000 [3:50:05<15:21:39, 2.30s/it] 20%|█▉ | 5981/30000 [3:50:08<15:15:32, 2.29s/it] 20%|█▉ | 5982/30000 [3:50:10<15:15:59, 2.29s/it] 20%|█▉ | 5983/30000 [3:50:12<15:10:22, 2.27s/it] 20%|█▉ | 5984/30000 [3:50:14<15:07:06, 2.27s/it] 20%|█▉ | 5985/30000 [3:50:17<15:19:12, 2.30s/it] 20%|█▉ | 5986/30000 [3:50:19<15:23:25, 2.31s/it] 20%|█▉ | 5987/30000 [3:50:21<15:19:51, 2.30s/it] 20%|█▉ | 5988/30000 [3:50:24<15:22:27, 2.31s/it] 20%|█▉ | 5989/30000 [3:50:26<15:25:01, 2.31s/it] 20%|█▉ | 5990/30000 [3:50:28<15:19:32, 2.30s/it] 20%|█▉ | 5990/30000 [3:50:28<15:19:32, 2.30s/it] 20%|█▉ | 5991/30000 [3:50:31<15:16:19, 2.29s/it] 20%|█▉ | 5992/30000 [3:50:33<15:22:13, 2.30s/it] 20%|█▉ | 5993/30000 [3:50:35<15:18:21, 2.30s/it] 20%|█▉ | 5994/30000 [3:50:37<15:14:57, 2.29s/it] 20%|█▉ | 5995/30000 [3:50:40<15:21:32, 2.30s/it] 20%|█▉ | 5996/30000 [3:50:42<15:23:31, 2.31s/it] 20%|█▉ | 5997/30000 [3:50:44<15:21:03, 2.30s/it] 20%|█▉ | 5998/30000 [3:50:47<15:19:05, 2.30s/it] 20%|█▉ | 5999/30000 [3:50:49<15:21:22, 2.30s/it] 20%|██ | 6000/30000 [3:50:51<15:14:11, 2.29s/it] 20%|██ | 6000/30000 [3:50:51<15:14:11, 2.29s/it] 20%|██ | 6001/30000 [3:50:54<15:17:03, 2.29s/it] 20%|██ | 6002/30000 [3:50:56<15:24:46, 2.31s/it] 20%|██ | 6003/30000 [3:50:58<15:27:53, 2.32s/it] 20%|██ | 6004/30000 [3:51:01<15:22:55, 2.31s/it] 20%|██ | 6005/30000 [3:51:03<15:17:35, 2.29s/it] 20%|██ | 6006/30000 [3:51:05<15:22:35, 2.31s/it] 20%|██ | 6007/30000 [3:51:07<15:25:09, 2.31s/it] 20%|██ | 6008/30000 [3:51:10<15:24:14, 2.31s/it] 20%|██ | 6009/30000 [3:51:12<15:29:40, 2.33s/it] 20%|██ | 6010/30000 [3:51:14<15:24:43, 2.31s/it] 20%|██ | 6010/30000 [3:51:14<15:24:43, 2.31s/it] 20%|██ | 6011/30000 [3:51:17<15:22:49, 2.31s/it] 20%|██ | 6012/30000 [3:51:19<15:21:33, 2.31s/it] 20%|██ | 6013/30000 [3:51:21<15:19:54, 2.30s/it] 20%|██ | 6014/30000 [3:51:24<15:17:31, 2.30s/it] 20%|██ | 6015/30000 [3:51:26<15:20:39, 2.30s/it] 20%|██ | 6016/30000 [3:51:28<15:22:36, 2.31s/it] 20%|██ | 6017/30000 [3:51:31<15:23:36, 2.31s/it] 20%|██ | 6018/30000 [3:51:33<15:18:19, 2.30s/it] 20%|██ | 6019/30000 [3:51:35<15:21:57, 2.31s/it] 20%|██ | 6020/30000 [3:51:37<15:25:14, 2.32s/it] 20%|██ | 6020/30000 [3:51:37<15:25:14, 2.32s/it] 20%|██ | 6021/30000 [3:51:40<15:19:03, 2.30s/it] 20%|██ | 6022/30000 [3:51:42<15:24:29, 2.31s/it] 20%|██ | 6023/30000 [3:51:44<15:18:22, 2.30s/it] 20%|██ | 6024/30000 [3:51:47<15:10:00, 2.28s/it] 20%|██ | 6025/30000 [3:51:49<15:18:16, 2.30s/it] 20%|██ | 6026/30000 [3:51:51<15:18:41, 2.30s/it] 20%|██ | 6027/30000 [3:51:53<15:16:29, 2.29s/it] 20%|██ | 6028/30000 [3:51:56<15:19:25, 2.30s/it] 20%|██ | 6029/30000 [3:51:58<15:17:50, 2.30s/it] 20%|██ | 6030/30000 [3:52:00<15:14:24, 2.29s/it] 20%|██ | 6030/30000 [3:52:00<15:14:24, 2.29s/it] 20%|██ | 6031/30000 [3:52:03<15:21:15, 2.31s/it] 20%|██ | 6032/30000 [3:52:05<15:22:49, 2.31s/it] 20%|██ | 6033/30000 [3:52:07<15:19:55, 2.30s/it] 20%|██ | 6034/30000 [3:52:10<15:19:39, 2.30s/it] 20%|██ | 6035/30000 [3:52:12<15:18:29, 2.30s/it] 20%|██ | 6036/30000 [3:52:14<15:15:36, 2.29s/it] 20%|██ | 6037/30000 [3:52:16<15:16:20, 2.29s/it] 20%|██ | 6038/30000 [3:52:19<15:19:46, 2.30s/it] 20%|██ | 6039/30000 [3:52:21<15:16:00, 2.29s/it] 20%|██ | 6040/30000 [3:52:23<15:13:05, 2.29s/it] 20%|██ | 6040/30000 [3:52:23<15:13:05, 2.29s/it] 20%|██ | 6041/30000 [3:52:26<15:22:48, 2.31s/it] 20%|██ | 6042/30000 [3:52:28<15:20:26, 2.31s/it] 20%|██ | 6043/30000 [3:52:30<15:14:51, 2.29s/it] 20%|██ | 6044/30000 [3:52:33<15:22:58, 2.31s/it] 20%|██ | 6045/30000 [3:52:35<15:25:11, 2.32s/it] 20%|██ | 6046/30000 [3:52:37<15:22:21, 2.31s/it] 20%|██ | 6047/30000 [3:52:40<15:23:59, 2.31s/it] 20%|██ | 6048/30000 [3:52:42<15:21:12, 2.31s/it] 20%|██ | 6049/30000 [3:52:44<15:24:04, 2.31s/it] 20%|██ | 6050/30000 [3:52:46<15:20:39, 2.31s/it] 20%|██ | 6050/30000 [3:52:47<15:20:39, 2.31s/it] 20%|██ | 6051/30000 [3:52:49<15:24:23, 2.32s/it] 20%|██ | 6052/30000 [3:52:51<15:17:29, 2.30s/it] 20%|██ | 6053/30000 [3:52:53<15:15:03, 2.29s/it] 20%|██ | 6054/30000 [3:52:56<15:14:26, 2.29s/it] 20%|██ | 6055/30000 [3:52:58<15:17:58, 2.30s/it] 20%|██ | 6056/30000 [3:53:00<15:15:25, 2.29s/it] 20%|██ | 6057/30000 [3:53:03<15:24:05, 2.32s/it] 20%|██ | 6058/30000 [3:53:05<15:20:30, 2.31s/it] 20%|██ | 6059/30000 [3:53:07<15:20:57, 2.31s/it] 20%|██ | 6060/30000 [3:53:09<15:19:14, 2.30s/it] 20%|██ | 6060/30000 [3:53:10<15:19:14, 2.30s/it] 20%|██ | 6061/30000 [3:53:12<15:23:47, 2.32s/it] 20%|██ | 6062/30000 [3:53:14<15:18:31, 2.30s/it] 20%|██ | 6063/30000 [3:53:16<15:11:30, 2.28s/it] 20%|██ | 6064/30000 [3:53:19<15:16:32, 2.30s/it] 20%|██ | 6065/30000 [3:53:21<15:20:25, 2.31s/it] 20%|██ | 6066/30000 [3:53:23<15:14:50, 2.29s/it] 20%|██ | 6067/30000 [3:53:26<15:15:12, 2.29s/it] 20%|██ | 6068/30000 [3:53:28<15:21:24, 2.31s/it] 20%|██ | 6069/30000 [3:53:30<15:15:45, 2.30s/it] 20%|██ | 6070/30000 [3:53:32<15:14:30, 2.29s/it] 20%|██ | 6070/30000 [3:53:33<15:14:30, 2.29s/it] 20%|██ | 6071/30000 [3:53:34<13:55:14, 2.09s/it] 20%|██ | 6072/30000 [3:53:35<12:07:08, 1.82s/it] 20%|██ | 6073/30000 [3:53:36<10:53:02, 1.64s/it] 20%|██ | 6074/30000 [3:53:38<10:01:47, 1.51s/it] 20%|██ | 6075/30000 [3:53:39<9:27:10, 1.42s/it] 20%|██ | 6076/30000 [3:53:40<9:03:42, 1.36s/it] 20%|██ | 6077/30000 [3:53:42<10:25:27, 1.57s/it] 20%|██ | 6078/30000 [3:53:45<12:02:30, 1.81s/it] 20%|██ | 6079/30000 [3:53:47<13:06:44, 1.97s/it] 20%|██ | 6080/30000 [3:53:49<13:43:54, 2.07s/it] 20%|██ | 6080/30000 [3:53:49<13:43:54, 2.07s/it] 20%|██ | 6081/30000 [3:53:51<14:09:49, 2.13s/it] 20%|██ | 6082/30000 [3:53:54<14:37:35, 2.20s/it] 20%|██ | 6083/30000 [3:53:56<14:51:28, 2.24s/it] 20%|██ | 6084/30000 [3:53:58<14:57:52, 2.25s/it] 20%|██ | 6085/30000 [3:54:01<15:09:19, 2.28s/it] 20%|██ | 6086/30000 [3:54:03<15:09:53, 2.28s/it] 20%|██ | 6087/30000 [3:54:05<15:08:08, 2.28s/it] 20%|██ | 6088/30000 [3:54:08<15:06:59, 2.28s/it] 20%|██ | 6089/30000 [3:54:10<15:14:52, 2.30s/it] 20%|██ | 6090/30000 [3:54:12<15:11:54, 2.29s/it] 20%|██ | 6090/30000 [3:54:12<15:11:54, 2.29s/it] 20%|██ | 6091/30000 [3:54:15<15:11:09, 2.29s/it] 20%|██ | 6092/30000 [3:54:17<15:17:11, 2.30s/it] 20%|██ | 6093/30000 [3:54:19<15:12:31, 2.29s/it] 20%|██ | 6094/30000 [3:54:21<15:04:33, 2.27s/it] 20%|██ | 6095/30000 [3:54:24<15:07:53, 2.28s/it] 20%|██ | 6096/30000 [3:54:26<15:15:28, 2.30s/it] 20%|██ | 6097/30000 [3:54:28<15:14:06, 2.29s/it] 20%|██ | 6098/30000 [3:54:31<15:11:09, 2.29s/it] 20%|██ | 6099/30000 [3:54:33<15:21:21, 2.31s/it] 20%|██ | 6100/30000 [3:54:35<15:21:58, 2.31s/it] 20%|██ | 6100/30000 [3:54:35<15:21:58, 2.31s/it] 20%|██ | 6101/30000 [3:54:38<15:15:44, 2.30s/it] 20%|██ | 6102/30000 [3:54:40<15:11:00, 2.29s/it] 20%|██ | 6103/30000 [3:54:42<15:15:25, 2.30s/it] 20%|██ | 6104/30000 [3:54:44<15:11:56, 2.29s/it] 20%|██ | 6105/30000 [3:54:47<15:09:54, 2.28s/it] 20%|██ | 6106/30000 [3:54:49<15:16:17, 2.30s/it] 20%|██ | 6107/30000 [3:54:51<15:11:22, 2.29s/it] 20%|██ | 6108/30000 [3:54:54<15:11:27, 2.29s/it] 20%|██ | 6109/30000 [3:54:56<15:06:56, 2.28s/it] 20%|██ | 6110/30000 [3:54:58<15:07:52, 2.28s/it] 20%|██ | 6110/30000 [3:54:58<15:07:52, 2.28s/it] 20%|██ | 6111/30000 [3:55:00<15:06:43, 2.28s/it] 20%|██ | 6112/30000 [3:55:03<15:06:22, 2.28s/it] 20%|██ | 6113/30000 [3:55:05<15:10:47, 2.29s/it] 20%|██ | 6114/30000 [3:55:07<15:11:30, 2.29s/it] 20%|██ | 6115/30000 [3:55:09<15:09:42, 2.29s/it] 20%|██ | 6116/30000 [3:55:12<15:07:12, 2.28s/it] 20%|██ | 6117/30000 [3:55:14<15:15:28, 2.30s/it] 20%|██ | 6118/30000 [3:55:16<15:16:40, 2.30s/it] 20%|██ | 6119/30000 [3:55:19<15:14:59, 2.30s/it] 20%|██ | 6120/30000 [3:55:21<15:09:37, 2.29s/it] 20%|██ | 6120/30000 [3:55:21<15:09:37, 2.29s/it] 20%|██ | 6121/30000 [3:55:23<15:21:27, 2.32s/it] 20%|██ | 6122/30000 [3:55:26<15:22:00, 2.32s/it] 20%|██ | 6123/30000 [3:55:28<15:13:43, 2.30s/it] 20%|██ | 6124/30000 [3:55:30<15:18:30, 2.31s/it] 20%|██ | 6125/30000 [3:55:33<15:14:50, 2.30s/it] 20%|██ | 6126/30000 [3:55:35<15:10:54, 2.29s/it] 20%|██ | 6127/30000 [3:55:37<15:11:24, 2.29s/it] 20%|██ | 6128/30000 [3:55:39<15:16:12, 2.30s/it] 20%|██ | 6129/30000 [3:55:42<15:16:39, 2.30s/it] 20%|██ | 6130/30000 [3:55:44<15:15:17, 2.30s/it] 20%|██ | 6130/30000 [3:55:44<15:15:17, 2.30s/it] 20%|██ | 6131/30000 [3:55:46<15:16:56, 2.30s/it] 20%|██ | 6132/30000 [3:55:49<15:20:52, 2.31s/it] 20%|██ | 6133/30000 [3:55:51<15:12:43, 2.29s/it] 20%|██ | 6134/30000 [3:55:53<15:08:19, 2.28s/it] 20%|██ | 6135/30000 [3:55:56<15:18:11, 2.31s/it] 20%|██ | 6136/30000 [3:55:58<15:21:22, 2.32s/it] 20%|██ | 6137/30000 [3:56:00<15:15:45, 2.30s/it] 20%|██ | 6138/30000 [3:56:02<15:13:06, 2.30s/it] 20%|██ | 6139/30000 [3:56:05<15:16:22, 2.30s/it] 20%|██ | 6140/30000 [3:56:07<15:13:39, 2.30s/it] 20%|██ | 6140/30000 [3:56:07<15:13:39, 2.30s/it] 20%|██ | 6141/30000 [3:56:09<15:13:46, 2.30s/it] 20%|██ | 6142/30000 [3:56:12<15:13:44, 2.30s/it] 20%|██ | 6143/30000 [3:56:14<15:17:05, 2.31s/it] 20%|██ | 6144/30000 [3:56:16<15:13:10, 2.30s/it] 20%|██ | 6145/30000 [3:56:19<15:11:16, 2.29s/it] 20%|██ | 6146/30000 [3:56:21<15:09:15, 2.29s/it] 20%|██ | 6147/30000 [3:56:23<15:13:55, 2.30s/it] 20%|██ | 6148/30000 [3:56:25<15:08:11, 2.28s/it] 20%|██ | 6149/30000 [3:56:28<15:12:21, 2.30s/it] 20%|██ | 6150/30000 [3:56:30<15:14:45, 2.30s/it] 20%|██ | 6150/30000 [3:56:30<15:14:45, 2.30s/it] 21%|██ | 6151/30000 [3:56:32<15:18:25, 2.31s/it] 21%|██ | 6152/30000 [3:56:35<15:09:09, 2.29s/it] 21%|██ | 6153/30000 [3:56:37<15:11:52, 2.29s/it] 21%|██ | 6154/30000 [3:56:39<15:11:52, 2.29s/it] 21%|██ | 6155/30000 [3:56:41<15:13:51, 2.30s/it] 21%|██ | 6156/30000 [3:56:44<15:09:03, 2.29s/it] 21%|██ | 6157/30000 [3:56:46<15:06:33, 2.28s/it] 21%|██ | 6158/30000 [3:56:48<15:10:29, 2.29s/it] 21%|██ | 6159/30000 [3:56:51<15:15:36, 2.30s/it] 21%|██ | 6160/30000 [3:56:53<15:17:01, 2.31s/it] 21%|██ | 6160/30000 [3:56:53<15:17:01, 2.31s/it] 21%|██ | 6161/30000 [3:56:55<15:16:24, 2.31s/it] 21%|██ | 6162/30000 [3:56:58<15:20:18, 2.32s/it] 21%|██ | 6163/30000 [3:57:00<15:26:11, 2.33s/it] 21%|██ | 6164/30000 [3:57:02<15:18:08, 2.31s/it] 21%|██ | 6165/30000 [3:57:05<15:19:17, 2.31s/it] 21%|██ | 6166/30000 [3:57:07<15:20:30, 2.32s/it] 21%|██ | 6167/30000 [3:57:09<15:14:49, 2.30s/it] 21%|██ | 6168/30000 [3:57:11<15:09:11, 2.29s/it] 21%|██ | 6169/30000 [3:57:14<15:18:11, 2.31s/it] 21%|██ | 6170/30000 [3:57:16<15:17:17, 2.31s/it] 21%|██ | 6170/30000 [3:57:16<15:17:17, 2.31s/it] 21%|██ | 6171/30000 [3:57:18<15:19:32, 2.32s/it] 21%|██ | 6172/30000 [3:57:21<15:19:55, 2.32s/it] 21%|██ | 6173/30000 [3:57:23<15:17:39, 2.31s/it] 21%|██ | 6174/30000 [3:57:25<15:14:43, 2.30s/it] 21%|██ | 6175/30000 [3:57:28<15:10:47, 2.29s/it] 21%|██ | 6176/30000 [3:57:30<15:12:31, 2.30s/it] 21%|██ | 6177/30000 [3:57:32<15:11:30, 2.30s/it] 21%|██ | 6178/30000 [3:57:34<15:07:38, 2.29s/it] 21%|██ | 6179/30000 [3:57:37<15:14:14, 2.30s/it] 21%|██ | 6180/30000 [3:57:39<15:16:10, 2.31s/it] 21%|██ | 6180/30000 [3:57:39<15:16:10, 2.31s/it] 21%|██ | 6181/30000 [3:57:41<15:12:21, 2.30s/it] 21%|██ | 6182/30000 [3:57:44<15:12:49, 2.30s/it] 21%|██ | 6183/30000 [3:57:46<15:13:25, 2.30s/it] 21%|██ | 6184/30000 [3:57:48<15:13:31, 2.30s/it] 21%|██ | 6185/30000 [3:57:51<15:09:05, 2.29s/it] 21%|██ | 6186/30000 [3:57:53<15:16:02, 2.31s/it] 21%|██ | 6187/30000 [3:57:55<15:22:19, 2.32s/it] 21%|██ | 6188/30000 [3:57:58<15:15:34, 2.31s/it] 21%|██ | 6189/30000 [3:58:00<15:18:13, 2.31s/it] 21%|██ | 6190/30000 [3:58:02<15:15:05, 2.31s/it] 21%|██ | 6190/30000 [3:58:02<15:15:05, 2.31s/it] 21%|██ | 6191/30000 [3:58:04<15:13:41, 2.30s/it] 21%|██ | 6192/30000 [3:58:07<15:09:59, 2.29s/it] 21%|██ | 6193/30000 [3:58:09<15:17:23, 2.31s/it] 21%|██ | 6194/30000 [3:58:11<15:09:34, 2.29s/it] 21%|██ | 6195/30000 [3:58:14<15:10:57, 2.30s/it] 21%|██ | 6196/30000 [3:58:16<15:15:51, 2.31s/it] 21%|██ | 6197/30000 [3:58:18<15:14:48, 2.31s/it] 21%|██ | 6198/30000 [3:58:21<15:10:45, 2.30s/it] 21%|██ | 6199/30000 [3:58:23<15:11:49, 2.30s/it] 21%|██ | 6200/30000 [3:58:25<15:12:10, 2.30s/it] 21%|██ | 6200/30000 [3:58:25<15:12:10, 2.30s/it] 21%|██ | 6201/30000 [3:58:27<15:08:52, 2.29s/it] 21%|██ | 6202/30000 [3:58:30<15:03:55, 2.28s/it] 21%|██ | 6203/30000 [3:58:32<15:10:07, 2.29s/it] 21%|██ | 6204/30000 [3:58:34<15:11:56, 2.30s/it] 21%|██ | 6205/30000 [3:58:37<15:06:35, 2.29s/it] 21%|██ | 6206/30000 [3:58:39<15:14:53, 2.31s/it] 21%|██ | 6207/30000 [3:58:41<15:19:26, 2.32s/it] 21%|██ | 6208/30000 [3:58:44<15:15:44, 2.31s/it] 21%|██ | 6209/30000 [3:58:46<15:14:03, 2.31s/it] 21%|██ | 6210/30000 [3:58:48<15:16:25, 2.31s/it] 21%|██ | 6210/30000 [3:58:48<15:16:25, 2.31s/it] 21%|██ | 6211/30000 [3:58:50<15:08:46, 2.29s/it] 21%|██ | 6212/30000 [3:58:53<15:06:13, 2.29s/it] 21%|██ | 6213/30000 [3:58:55<15:14:58, 2.31s/it] 21%|██ | 6214/30000 [3:58:57<15:21:31, 2.32s/it] 21%|██ | 6215/30000 [3:59:00<15:17:07, 2.31s/it] 21%|██ | 6216/30000 [3:59:02<15:16:51, 2.31s/it] 21%|██ | 6217/30000 [3:59:04<15:16:22, 2.31s/it] 21%|██ | 6218/30000 [3:59:07<15:12:06, 2.30s/it] 21%|██ | 6219/30000 [3:59:09<15:11:37, 2.30s/it] 21%|██ | 6220/30000 [3:59:11<15:16:35, 2.31s/it] 21%|██ | 6220/30000 [3:59:11<15:16:35, 2.31s/it] 21%|██ | 6221/30000 [3:59:14<15:15:57, 2.31s/it] 21%|██ | 6222/30000 [3:59:16<15:07:43, 2.29s/it] 21%|██ | 6223/30000 [3:59:18<15:11:41, 2.30s/it] 21%|██ | 6224/30000 [3:59:20<15:12:20, 2.30s/it] 21%|██ | 6225/30000 [3:59:23<15:08:04, 2.29s/it] 21%|██ | 6226/30000 [3:59:25<15:07:13, 2.29s/it] 21%|██ | 6227/30000 [3:59:27<15:11:43, 2.30s/it] 21%|██ | 6228/30000 [3:59:30<15:01:32, 2.28s/it] 21%|██ | 6229/30000 [3:59:32<15:02:21, 2.28s/it] 21%|██ | 6230/30000 [3:59:34<15:11:38, 2.30s/it] 21%|██ | 6230/30000 [3:59:34<15:11:38, 2.30s/it] 21%|██ | 6231/30000 [3:59:36<15:13:03, 2.30s/it] 21%|██ | 6232/30000 [3:59:39<15:07:59, 2.29s/it] 21%|██ | 6233/30000 [3:59:41<15:15:20, 2.31s/it] 21%|██ | 6234/30000 [3:59:43<15:12:51, 2.30s/it] 21%|██ | 6235/30000 [3:59:46<15:09:58, 2.30s/it] 21%|██ | 6236/30000 [3:59:48<15:08:46, 2.29s/it] 21%|██ | 6237/30000 [3:59:50<15:09:08, 2.30s/it] 21%|██ | 6238/30000 [3:59:53<15:05:36, 2.29s/it] 21%|██ | 6239/30000 [3:59:55<15:02:08, 2.28s/it] 21%|██ | 6240/30000 [3:59:57<15:12:57, 2.31s/it] 21%|██ | 6240/30000 [3:59:57<15:12:57, 2.31s/it] 21%|██ | 6241/30000 [3:59:59<15:14:11, 2.31s/it] 21%|██ | 6242/30000 [4:00:02<15:10:04, 2.30s/it] 21%|██ | 6243/30000 [4:00:04<15:18:16, 2.32s/it] 21%|██ | 6244/30000 [4:00:06<15:21:39, 2.33s/it] 21%|██ | 6245/30000 [4:00:09<15:18:01, 2.32s/it] 21%|██ | 6246/30000 [4:00:11<15:18:51, 2.32s/it] 21%|██ | 6247/30000 [4:00:13<15:17:44, 2.32s/it] 21%|██ | 6248/30000 [4:00:16<15:15:46, 2.31s/it] 21%|██ | 6249/30000 [4:00:18<15:15:30, 2.31s/it] 21%|██ | 6250/30000 [4:00:20<15:18:41, 2.32s/it] 21%|██ | 6250/30000 [4:00:20<15:18:41, 2.32s/it] 21%|██ | 6251/30000 [4:00:23<15:23:11, 2.33s/it] 21%|██ | 6252/30000 [4:00:25<15:14:15, 2.31s/it] 21%|██ | 6253/30000 [4:00:27<15:17:26, 2.32s/it] 21%|██ | 6254/30000 [4:00:30<15:17:56, 2.32s/it] 21%|██ | 6255/30000 [4:00:32<15:12:24, 2.31s/it] 21%|██ | 6256/30000 [4:00:34<15:12:34, 2.31s/it] 21%|██ | 6257/30000 [4:00:37<15:15:50, 2.31s/it] 21%|██ | 6258/30000 [4:00:39<15:11:38, 2.30s/it] 21%|██ | 6259/30000 [4:00:41<15:11:43, 2.30s/it] 21%|██ | 6260/30000 [4:00:43<15:15:39, 2.31s/it] 21%|██ | 6260/30000 [4:00:43<15:15:39, 2.31s/it] 21%|██ | 6261/30000 [4:00:46<15:12:10, 2.31s/it] 21%|██ | 6262/30000 [4:00:48<15:06:35, 2.29s/it] 21%|██ | 6263/30000 [4:00:50<15:10:21, 2.30s/it] 21%|██ | 6264/30000 [4:00:53<15:14:21, 2.31s/it] 21%|██ | 6265/30000 [4:00:55<15:10:12, 2.30s/it] 21%|██ | 6266/30000 [4:00:57<15:15:04, 2.31s/it] 21%|██ | 6267/30000 [4:01:00<15:16:09, 2.32s/it] 21%|██ | 6268/30000 [4:01:02<15:13:24, 2.31s/it] 21%|██ | 6269/30000 [4:01:04<15:09:42, 2.30s/it] 21%|██ | 6270/30000 [4:01:06<15:12:06, 2.31s/it] 21%|██ | 6270/30000 [4:01:07<15:12:06, 2.31s/it] 21%|██ | 6271/30000 [4:01:09<15:07:52, 2.30s/it] 21%|██ | 6272/30000 [4:01:11<15:02:33, 2.28s/it] 21%|██ | 6273/30000 [4:01:13<15:15:35, 2.32s/it] 21%|██ | 6274/30000 [4:01:16<15:18:11, 2.32s/it] 21%|██ | 6275/30000 [4:01:18<15:14:26, 2.31s/it] 21%|██ | 6276/30000 [4:01:20<15:14:43, 2.31s/it] 21%|██ | 6277/30000 [4:01:23<15:13:41, 2.31s/it] 21%|██ | 6278/30000 [4:01:25<15:09:54, 2.30s/it] 21%|██ | 6279/30000 [4:01:27<15:10:19, 2.30s/it] 21%|██ | 6280/30000 [4:01:29<15:01:49, 2.28s/it] +{'loss': 0.0175, 'grad_norm': 0.12034767866134644, 'learning_rate': 2.828615273381551e-05, 'flow_matching_loss': 0.01749199368059635, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.71} +{'loss': 0.0165, 'grad_norm': 0.10932344943284988, 'learning_rate': 2.827846966998683e-05, 'flow_matching_loss': 0.01654358869418502, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.71} +{'loss': 0.0176, 'grad_norm': 0.08873020857572556, 'learning_rate': 2.8270770471565422e-05, 'flow_matching_loss': 0.01763467360287905, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.72} +{'loss': 0.017, 'grad_norm': 0.0922614261507988, 'learning_rate': 2.826305514790654e-05, 'flow_matching_loss': 0.016993628069758415, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.72} +{'loss': 0.0188, 'grad_norm': 0.11580375581979752, 'learning_rate': 2.825532370838503e-05, 'flow_matching_loss': 0.018817862402647732, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.72} +{'loss': 0.0176, 'grad_norm': 0.0897599458694458, 'learning_rate': 2.824757616239532e-05, 'flow_matching_loss': 0.017556344904005528, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.73} +{'loss': 0.0172, 'grad_norm': 0.09729140996932983, 'learning_rate': 2.8239812519351404e-05, 'flow_matching_loss': 0.01715866168960929, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.73} +{'loss': 0.0178, 'grad_norm': 0.07686830312013626, 'learning_rate': 2.823203278868685e-05, 'flow_matching_loss': 0.01777117932215333, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.73} +{'loss': 0.0179, 'grad_norm': 0.08235519379377365, 'learning_rate': 2.8224236979854752e-05, 'flow_matching_loss': 0.017890595458447935, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.73} +{'loss': 0.0199, 'grad_norm': 0.0808558464050293, 'learning_rate': 2.821642510232776e-05, 'flow_matching_loss': 0.01989372055977583, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.74} +{'loss': 0.0192, 'grad_norm': 0.0824810117483139, 'learning_rate': 2.8208597165598046e-05, 'flow_matching_loss': 0.019226994179189206, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.74} +{'loss': 0.0181, 'grad_norm': 0.11257638037204742, 'learning_rate': 2.8200753179177285e-05, 'flow_matching_loss': 0.018084902595728636, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.74} +{'loss': 0.0174, 'grad_norm': 0.11582588404417038, 'learning_rate': 2.819289315259666e-05, 'flow_matching_loss': 0.017362800147384404, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.75} +{'loss': 0.0188, 'grad_norm': 0.10098263621330261, 'learning_rate': 2.8185017095406845e-05, 'flow_matching_loss': 0.01877316916361451, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.75} +{'loss': 0.0166, 'grad_norm': 0.07611538469791412, 'learning_rate': 2.8177125017177998e-05, 'flow_matching_loss': 0.016554318461567165, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.75} +{'loss': 0.0164, 'grad_norm': 0.08150531351566315, 'learning_rate': 2.8169216927499727e-05, 'flow_matching_loss': 0.01637625163421035, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.75} +{'loss': 0.0162, 'grad_norm': 0.08863822370767593, 'learning_rate': 2.816129283598112e-05, 'flow_matching_loss': 0.01620745100080967, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.76} +{'loss': 0.0169, 'grad_norm': 0.07439881563186646, 'learning_rate': 2.815335275225068e-05, 'flow_matching_loss': 0.01692084800451994, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.76} +{'loss': 0.0169, 'grad_norm': 0.0853864997625351, 'learning_rate': 2.8145396685956374e-05, 'flow_matching_loss': 0.016922676749527455, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.76} +{'loss': 0.0172, 'grad_norm': 0.08500128984451294, 'learning_rate': 2.813742464676556e-05, 'flow_matching_loss': 0.017206930555403233, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.77} +{'loss': 0.0165, 'grad_norm': 0.1160147488117218, 'learning_rate': 2.8129436644365025e-05, 'flow_matching_loss': 0.01654683379456401, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.77} +{'loss': 0.0167, 'grad_norm': 0.06942442804574966, 'learning_rate': 2.8121432688460938e-05, 'flow_matching_loss': 0.016722906846553087, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.77} +{'loss': 0.018, 'grad_norm': 0.09589885175228119, 'learning_rate': 2.8113412788778867e-05, 'flow_matching_loss': 0.017990529723465442, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.77} +{'loss': 0.0168, 'grad_norm': 0.0750080943107605, 'learning_rate': 2.8105376955063744e-05, 'flow_matching_loss': 0.016799407079815865, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.78} +{'loss': 0.0172, 'grad_norm': 0.07226303219795227, 'learning_rate': 2.8097325197079862e-05, 'flow_matching_loss': 0.017199462000280618, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.78} +{'loss': 0.0161, 'grad_norm': 0.10195241868495941, 'learning_rate': 2.808925752461087e-05, 'flow_matching_loss': 0.016121744737029077, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.78} +{'loss': 0.0167, 'grad_norm': 0.06571152061223984, 'learning_rate': 2.808117394745975e-05, 'flow_matching_loss': 0.016745352186262608, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.79} +{'loss': 0.0174, 'grad_norm': 0.08149969577789307, 'learning_rate': 2.807307447544881e-05, 'flow_matching_loss': 0.017415294609963893, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.79} +{'loss': 0.0173, 'grad_norm': 0.09414371848106384, 'learning_rate': 2.806495911841967e-05, 'flow_matching_loss': 0.017331584822386503, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.79} +{'loss': 0.0174, 'grad_norm': 0.08183462172746658, 'learning_rate': 2.8056827886233255e-05, 'flow_matching_loss': 0.017356385476887226, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.8} +{'loss': 0.0174, 'grad_norm': 0.08893103152513504, 'learning_rate': 2.8048680788769783e-05, 'flow_matching_loss': 0.017430007550865413, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.8} +{'loss': 0.0157, 'grad_norm': 0.0707133412361145, 'learning_rate': 2.8040517835928742e-05, 'flow_matching_loss': 0.015677034854888916, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.8} +{'loss': 0.016, 'grad_norm': 0.08220255374908447, 'learning_rate': 2.803233903762889e-05, 'flow_matching_loss': 0.015951901953667403, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.8} +{'loss': 0.0148, 'grad_norm': 0.07485468685626984, 'learning_rate': 2.8024144403808235e-05, 'flow_matching_loss': 0.014752757921814918, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.81} +{'loss': 0.0175, 'grad_norm': 0.07357978075742722, 'learning_rate': 2.8015933944424037e-05, 'flow_matching_loss': 0.017506221681833266, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.81} +{'loss': 0.0159, 'grad_norm': 0.07209871709346771, 'learning_rate': 2.800770766945277e-05, 'flow_matching_loss': 0.015915028750896454, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.81} +{'loss': 0.0169, 'grad_norm': 0.0774451270699501, 'learning_rate': 2.7999465588890135e-05, 'flow_matching_loss': 0.016900410316884516, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.82} +{'loss': 0.017, 'grad_norm': 0.07061811536550522, 'learning_rate': 2.799120771275105e-05, 'flow_matching_loss': 0.01702676070854068, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.82} +{'loss': 0.0165, 'grad_norm': 0.08194387704133987, 'learning_rate': 2.7982934051069597e-05, 'flow_matching_loss': 0.01653643324971199, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.82} +{'loss': 0.0165, 'grad_norm': 0.07407239079475403, 'learning_rate': 2.7974644613899063e-05, 'flow_matching_loss': 0.016485044546425343, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.82} +{'loss': 0.0172, 'grad_norm': 0.10221776366233826, 'learning_rate': 2.796633941131189e-05, 'flow_matching_loss': 0.017232183087617158, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.83} 21%|██ | 6280/30000 [4:01:30<15:01:49, 2.28s/it] 21%|██ | 6281/30000 [4:01:32<15:01:34, 2.28s/it] 21%|██ | 6282/30000 [4:01:34<15:06:38, 2.29s/it] 21%|██ | 6283/30000 [4:01:36<15:15:24, 2.32s/it] 21%|██ | 6284/30000 [4:01:39<15:03:49, 2.29s/it] 21%|██ | 6285/30000 [4:01:41<15:05:14, 2.29s/it] 21%|██ | 6286/30000 [4:01:43<15:10:23, 2.30s/it] 21%|██ | 6287/30000 [4:01:46<15:10:36, 2.30s/it] 21%|██ | 6288/30000 [4:01:48<15:03:41, 2.29s/it] 21%|██ | 6289/30000 [4:01:50<15:13:47, 2.31s/it] 21%|██ | 6290/30000 [4:01:52<15:10:35, 2.30s/it] 21%|██ | 6290/30000 [4:01:53<15:10:35, 2.30s/it] 21%|██ | 6291/30000 [4:01:55<15:07:40, 2.30s/it] 21%|██ | 6292/30000 [4:01:57<15:07:26, 2.30s/it] 21%|██ | 6293/30000 [4:01:59<15:12:46, 2.31s/it] 21%|██ | 6294/30000 [4:02:02<15:04:37, 2.29s/it] 21%|██ | 6295/30000 [4:02:04<15:04:40, 2.29s/it] 21%|██ | 6296/30000 [4:02:06<15:09:03, 2.30s/it] 21%|██ | 6297/30000 [4:02:09<15:08:34, 2.30s/it] 21%|██ | 6298/30000 [4:02:11<14:59:59, 2.28s/it] 21%|██ | 6299/30000 [4:02:13<15:07:22, 2.30s/it] 21%|██ | 6300/30000 [4:02:15<15:10:07, 2.30s/it] 21%|██ | 6300/30000 [4:02:15<15:10:07, 2.30s/it] 21%|██ | 6301/30000 [4:02:18<15:06:29, 2.30s/it] 21%|██ | 6302/30000 [4:02:20<15:10:04, 2.30s/it] 21%|██ | 6303/30000 [4:02:22<15:09:36, 2.30s/it] 21%|██ | 6304/30000 [4:02:25<15:04:01, 2.29s/it] 21%|██ | 6305/30000 [4:02:27<15:02:32, 2.29s/it] 21%|██ | 6306/30000 [4:02:29<15:09:43, 2.30s/it] 21%|██ | 6307/30000 [4:02:32<15:09:41, 2.30s/it] 21%|██ | 6308/30000 [4:02:34<15:05:36, 2.29s/it] 21%|██ | 6309/30000 [4:02:36<15:13:11, 2.31s/it] 21%|██ | 6310/30000 [4:02:38<15:09:49, 2.30s/it] 21%|██ | 6310/30000 [4:02:38<15:09:49, 2.30s/it] 21%|██ | 6311/30000 [4:02:41<15:09:20, 2.30s/it] 21%|██ | 6312/30000 [4:02:43<15:08:51, 2.30s/it] 21%|██ | 6313/30000 [4:02:45<15:15:57, 2.32s/it] 21%|██ | 6314/30000 [4:02:48<15:18:18, 2.33s/it] 21%|██ | 6315/30000 [4:02:50<15:12:49, 2.31s/it] 21%|██ | 6316/30000 [4:02:52<15:15:10, 2.32s/it] 21%|██ | 6317/30000 [4:02:55<15:12:16, 2.31s/it] 21%|██ | 6318/30000 [4:02:57<15:05:12, 2.29s/it] 21%|██ | 6319/30000 [4:02:59<15:02:31, 2.29s/it] 21%|██ | 6320/30000 [4:03:01<15:06:28, 2.30s/it] 21%|██ | 6320/30000 [4:03:02<15:06:28, 2.30s/it] 21%|██ | 6321/30000 [4:03:04<15:09:18, 2.30s/it] 21%|██ | 6322/30000 [4:03:06<15:05:03, 2.29s/it] 21%|██ | 6323/30000 [4:03:08<15:13:22, 2.31s/it] 21%|██ | 6324/30000 [4:03:11<15:12:13, 2.31s/it] 21%|██ | 6325/30000 [4:03:13<15:09:23, 2.30s/it] 21%|██ | 6326/30000 [4:03:15<15:04:55, 2.29s/it] 21%|██ | 6327/30000 [4:03:18<15:02:15, 2.29s/it] 21%|██ | 6328/30000 [4:03:20<14:59:30, 2.28s/it] 21%|██ | 6329/30000 [4:03:22<15:01:46, 2.29s/it] 21%|██ | 6330/30000 [4:03:25<15:10:31, 2.31s/it] 21%|██ | 6330/30000 [4:03:25<15:10:31, 2.31s/it] 21%|██ | 6331/30000 [4:03:27<15:16:10, 2.32s/it] 21%|██ | 6332/30000 [4:03:29<15:09:17, 2.31s/it] 21%|██ | 6333/30000 [4:03:31<15:06:06, 2.30s/it] 21%|██ | 6334/30000 [4:03:34<15:05:39, 2.30s/it] 21%|██ | 6335/30000 [4:03:36<15:03:02, 2.29s/it] 21%|██ | 6336/30000 [4:03:38<14:57:19, 2.28s/it] 21%|██ | 6337/30000 [4:03:41<15:08:29, 2.30s/it] 21%|██ | 6338/30000 [4:03:43<15:06:55, 2.30s/it] 21%|██ | 6339/30000 [4:03:45<15:05:09, 2.30s/it] 21%|██ | 6340/30000 [4:03:47<15:01:06, 2.29s/it] 21%|██ | 6340/30000 [4:03:47<15:01:06, 2.29s/it] 21%|██ | 6341/30000 [4:03:50<15:04:58, 2.30s/it] 21%|██ | 6342/30000 [4:03:52<15:13:09, 2.32s/it] 21%|██ | 6343/30000 [4:03:54<15:08:12, 2.30s/it] 21%|██ | 6344/30000 [4:03:57<15:05:22, 2.30s/it] 21%|██ | 6345/30000 [4:03:59<15:12:32, 2.31s/it] 21%|██ | 6346/30000 [4:04:01<15:08:15, 2.30s/it] 21%|██ | 6347/30000 [4:04:04<15:08:54, 2.31s/it] 21%|██ | 6348/30000 [4:04:06<15:12:18, 2.31s/it] 21%|██ | 6349/30000 [4:04:08<15:08:35, 2.30s/it] 21%|██ | 6350/30000 [4:04:10<15:01:59, 2.29s/it] 21%|██ | 6350/30000 [4:04:11<15:01:59, 2.29s/it] 21%|██ | 6351/30000 [4:04:13<15:01:04, 2.29s/it] 21%|██ | 6352/30000 [4:04:15<15:07:55, 2.30s/it] 21%|██ | 6353/30000 [4:04:17<15:05:52, 2.30s/it] 21%|██ | 6354/30000 [4:04:20<15:00:25, 2.28s/it] 21%|██ | 6355/30000 [4:04:22<15:03:04, 2.29s/it] 21%|██ | 6356/30000 [4:04:24<15:05:00, 2.30s/it] 21%|██ | 6357/30000 [4:04:27<15:01:04, 2.29s/it] 21%|██ | 6358/30000 [4:04:29<14:58:14, 2.28s/it] 21%|██ | 6359/30000 [4:04:31<15:04:36, 2.30s/it] 21%|██ | 6360/30000 [4:04:33<15:06:05, 2.30s/it] 21%|██ | 6360/30000 [4:04:33<15:06:05, 2.30s/it] 21%|██ | 6361/30000 [4:04:36<15:02:41, 2.29s/it] 21%|██ | 6362/30000 [4:04:38<14:59:24, 2.28s/it] 21%|██ | 6363/30000 [4:04:40<15:04:28, 2.30s/it] 21%|██ | 6364/30000 [4:04:43<15:00:35, 2.29s/it] 21%|██ | 6365/30000 [4:04:45<14:55:23, 2.27s/it] 21%|██ | 6366/30000 [4:04:47<15:05:12, 2.30s/it] 21%|██ | 6367/30000 [4:04:49<15:04:57, 2.30s/it] 21%|██ | 6368/30000 [4:04:52<14:59:25, 2.28s/it] 21%|██ | 6369/30000 [4:04:54<14:58:35, 2.28s/it] 21%|██ | 6370/30000 [4:04:56<15:04:31, 2.30s/it] 21%|██ | 6370/30000 [4:04:56<15:04:31, 2.30s/it] 21%|██ | 6371/30000 [4:04:59<15:05:30, 2.30s/it] 21%|██ | 6372/30000 [4:05:01<15:05:49, 2.30s/it] 21%|██ | 6373/30000 [4:05:03<15:01:09, 2.29s/it] 21%|██ | 6374/30000 [4:05:06<15:07:22, 2.30s/it] 21%|██▏ | 6375/30000 [4:05:08<15:06:24, 2.30s/it] 21%|██▏ | 6376/30000 [4:05:10<15:02:20, 2.29s/it] 21%|██▏ | 6377/30000 [4:05:12<14:58:55, 2.28s/it] 21%|██▏ | 6378/30000 [4:05:15<15:07:59, 2.31s/it] 21%|██▏ | 6379/30000 [4:05:17<15:12:34, 2.32s/it] 21%|██▏ | 6380/30000 [4:05:19<15:08:28, 2.31s/it] 21%|██▏ | 6380/30000 [4:05:19<15:08:28, 2.31s/it] 21%|██▏ | 6381/30000 [4:05:22<15:02:34, 2.29s/it] 21%|██▏ | 6382/30000 [4:05:24<15:07:20, 2.31s/it] 21%|██▏ | 6383/30000 [4:05:26<15:02:47, 2.29s/it] 21%|██▏ | 6384/30000 [4:05:28<15:00:00, 2.29s/it] 21%|██▏ | 6385/30000 [4:05:31<15:08:16, 2.31s/it] 21%|██▏ | 6386/30000 [4:05:33<15:05:18, 2.30s/it] 21%|██▏ | 6387/30000 [4:05:35<15:02:54, 2.29s/it] 21%|██▏ | 6388/30000 [4:05:38<14:58:29, 2.28s/it] 21%|██▏ | 6389/30000 [4:05:40<15:04:11, 2.30s/it] 21%|██▏ | 6390/30000 [4:05:42<15:04:06, 2.30s/it] 21%|██▏ | 6390/30000 [4:05:42<15:04:06, 2.30s/it] 21%|██▏ | 6391/30000 [4:05:45<14:59:10, 2.29s/it] 21%|██▏ | 6392/30000 [4:05:47<14:57:26, 2.28s/it] 21%|██▏ | 6393/30000 [4:05:49<15:08:16, 2.31s/it] 21%|██▏ | 6394/30000 [4:05:51<15:09:27, 2.31s/it] 21%|██▏ | 6395/30000 [4:05:54<15:09:26, 2.31s/it] 21%|██▏ | 6396/30000 [4:05:56<15:02:13, 2.29s/it] 21%|██▏ | 6397/30000 [4:05:58<15:07:31, 2.31s/it] 21%|██▏ | 6398/30000 [4:06:01<15:06:35, 2.30s/it] 21%|██▏ | 6399/30000 [4:06:03<14:59:54, 2.29s/it] 21%|██▏ | 6400/30000 [4:06:05<15:01:00, 2.29s/it] 21%|██▏ | 6400/30000 [4:06:05<15:01:00, 2.29s/it] 21%|██▏ | 6401/30000 [4:06:08<15:09:35, 2.31s/it] 21%|██▏ | 6402/30000 [4:06:10<15:09:33, 2.31s/it] 21%|██▏ | 6403/30000 [4:06:12<15:05:48, 2.30s/it] 21%|██▏ | 6404/30000 [4:06:14<15:04:12, 2.30s/it] 21%|██▏ | 6405/30000 [4:06:17<15:09:18, 2.31s/it] 21%|██▏ | 6406/30000 [4:06:19<15:06:03, 2.30s/it] 21%|██▏ | 6407/30000 [4:06:21<15:03:16, 2.30s/it] 21%|██▏ | 6408/30000 [4:06:24<15:02:43, 2.30s/it] 21%|██▏ | 6409/30000 [4:06:26<15:09:38, 2.31s/it] 21%|██▏ | 6410/30000 [4:06:28<15:03:19, 2.30s/it] 21%|██▏ | 6410/30000 [4:06:28<15:03:19, 2.30s/it] 21%|██▏ | 6411/30000 [4:06:31<15:01:04, 2.29s/it] 21%|██▏ | 6412/30000 [4:06:33<15:07:32, 2.31s/it] 21%|██▏ | 6413/30000 [4:06:35<15:07:19, 2.31s/it] 21%|██▏ | 6414/30000 [4:06:37<15:01:35, 2.29s/it] 21%|██▏ | 6415/30000 [4:06:40<14:57:03, 2.28s/it] 21%|██▏ | 6416/30000 [4:06:42<15:03:11, 2.30s/it] 21%|██▏ | 6417/30000 [4:06:44<15:01:01, 2.29s/it] 21%|██▏ | 6418/30000 [4:06:47<15:01:29, 2.29s/it] 21%|██▏ | 6419/30000 [4:06:49<15:03:30, 2.30s/it] 21%|██▏ | 6420/30000 [4:06:51<15:05:17, 2.30s/it] 21%|██▏ | 6420/30000 [4:06:51<15:05:17, 2.30s/it] 21%|██▏ | 6421/30000 [4:06:54<15:01:58, 2.30s/it] 21%|██▏ | 6422/30000 [4:06:56<15:02:28, 2.30s/it] 21%|██▏ | 6423/30000 [4:06:58<15:07:30, 2.31s/it] 21%|██▏ | 6424/30000 [4:07:00<15:05:50, 2.31s/it] 21%|██▏ | 6425/30000 [4:07:03<15:03:36, 2.30s/it] 21%|██▏ | 6426/30000 [4:07:05<15:03:46, 2.30s/it] 21%|██▏ | 6427/30000 [4:07:07<15:11:02, 2.32s/it] 21%|██▏ | 6428/30000 [4:07:10<15:13:44, 2.33s/it] 21%|██▏ | 6429/30000 [4:07:12<15:07:24, 2.31s/it] 21%|██▏ | 6430/30000 [4:07:14<15:01:33, 2.30s/it] 21%|██▏ | 6430/30000 [4:07:14<15:01:33, 2.30s/it] 21%|██▏ | 6431/30000 [4:07:17<15:08:07, 2.31s/it] 21%|██▏ | 6432/30000 [4:07:19<15:04:36, 2.30s/it] 21%|██▏ | 6433/30000 [4:07:21<15:06:03, 2.31s/it] 21%|██▏ | 6434/30000 [4:07:24<15:01:27, 2.30s/it] 21%|██▏ | 6435/30000 [4:07:26<15:09:58, 2.32s/it] 21%|██▏ | 6436/30000 [4:07:28<15:08:57, 2.31s/it] 21%|██▏ | 6437/30000 [4:07:30<15:02:16, 2.30s/it] 21%|██▏ | 6438/30000 [4:07:33<14:56:02, 2.28s/it] 21%|██▏ | 6439/30000 [4:07:35<15:01:00, 2.29s/it] 21%|██▏ | 6440/30000 [4:07:37<15:04:54, 2.30s/it] 21%|██▏ | 6440/30000 [4:07:37<15:04:54, 2.30s/it] 21%|██▏ | 6441/30000 [4:07:40<15:03:21, 2.30s/it] 21%|██▏ | 6442/30000 [4:07:42<14:59:19, 2.29s/it] 21%|██▏ | 6443/30000 [4:07:44<15:07:46, 2.31s/it] 21%|██▏ | 6444/30000 [4:07:47<15:09:35, 2.32s/it] 21%|██▏ | 6445/30000 [4:07:49<15:04:06, 2.30s/it] 21%|██▏ | 6446/30000 [4:07:51<14:59:40, 2.29s/it] 21%|██▏ | 6447/30000 [4:07:53<15:03:42, 2.30s/it] 21%|██▏ | 6448/30000 [4:07:56<15:00:57, 2.30s/it] 21%|██▏ | 6449/30000 [4:07:58<15:00:29, 2.29s/it] 22%|██▏ | 6450/30000 [4:08:00<15:00:16, 2.29s/it] 22%|██▏ | 6450/30000 [4:08:00<15:00:16, 2.29s/it] 22%|██▏ | 6451/30000 [4:08:03<15:06:34, 2.31s/it] 22%|██▏ | 6452/30000 [4:08:05<15:01:07, 2.30s/it] 22%|██▏ | 6453/30000 [4:08:07<14:58:42, 2.29s/it] 22%|██▏ | 6454/30000 [4:08:10<15:02:38, 2.30s/it] 22%|██▏ | 6455/30000 [4:08:12<15:05:31, 2.31s/it] 22%|██▏ | 6456/30000 [4:08:14<14:59:19, 2.29s/it] 22%|██▏ | 6457/30000 [4:08:16<14:59:56, 2.29s/it] 22%|██▏ | 6458/30000 [4:08:19<15:07:38, 2.31s/it] 22%|██▏ | 6459/30000 [4:08:21<15:05:21, 2.31s/it] 22%|██▏ | 6460/30000 [4:08:23<14:59:11, 2.29s/it] 22%|██▏ | 6460/30000 [4:08:23<14:59:11, 2.29s/it] 22%|██▏ | 6461/30000 [4:08:26<14:59:08, 2.29s/it] 22%|██▏ | 6462/30000 [4:08:28<15:04:07, 2.30s/it] 22%|██▏ | 6463/30000 [4:08:30<15:06:08, 2.31s/it] 22%|██▏ | 6464/30000 [4:08:33<15:05:06, 2.31s/it] 22%|██▏ | 6465/30000 [4:08:35<15:05:14, 2.31s/it] 22%|██▏ | 6466/30000 [4:08:37<15:11:00, 2.32s/it] 22%|██▏ | 6467/30000 [4:08:40<15:11:55, 2.33s/it] 22%|██▏ | 6468/30000 [4:08:42<15:05:20, 2.31s/it] 22%|██▏ | 6469/30000 [4:08:44<15:00:25, 2.30s/it] 22%|██▏ | 6470/30000 [4:08:46<15:02:49, 2.30s/it] 22%|██▏ | 6470/30000 [4:08:46<15:02:49, 2.30s/it] 22%|██▏ | 6471/30000 [4:08:49<15:06:40, 2.31s/it] 22%|██▏ | 6472/30000 [4:08:51<15:00:15, 2.30s/it] 22%|██▏ | 6473/30000 [4:08:53<14:59:18, 2.29s/it] 22%|██▏ | 6474/30000 [4:08:56<15:01:54, 2.30s/it] 22%|██▏ | 6475/30000 [4:08:58<14:56:52, 2.29s/it] 22%|██▏ | 6476/30000 [4:09:00<14:54:24, 2.28s/it] 22%|██▏ | 6477/30000 [4:09:02<14:58:05, 2.29s/it] 22%|██▏ | 6478/30000 [4:09:05<15:00:12, 2.30s/it] 22%|██▏ | 6479/30000 [4:09:07<14:58:21, 2.29s/it] 22%|██▏ | 6480/30000 [4:09:09<14:55:41, 2.28s/it] 22%|██▏ | 6480/30000 [4:09:09<14:55:41, 2.28s/it] 22%|██▏ | 6481/30000 [4:09:12<15:03:25, 2.30s/it] 22%|██▏ | 6482/30000 [4:09:14<14:59:05, 2.29s/it] 22%|██▏ | 6483/30000 [4:09:16<14:58:07, 2.29s/it] 22%|██▏ | 6484/30000 [4:09:19<15:03:50, 2.31s/it] 22%|██▏ | 6485/30000 [4:09:21<15:04:14, 2.31s/it] 22%|██▏ | 6486/30000 [4:09:23<15:00:03, 2.30s/it] 22%|██▏ | 6487/30000 [4:09:25<14:55:45, 2.29s/it] 22%|██▏ | 6488/30000 [4:09:28<15:01:49, 2.30s/it] 22%|██▏ | 6489/30000 [4:09:30<15:05:27, 2.31s/it] 22%|██▏ | 6490/30000 [4:09:32<15:06:24, 2.31s/it] 22%|██▏ | 6490/30000 [4:09:32<15:06:24, 2.31s/it] 22%|██▏ | 6491/30000 [4:09:35<15:07:50, 2.32s/it] 22%|██▏ | 6492/30000 [4:09:37<15:07:57, 2.32s/it] 22%|██▏ | 6493/30000 [4:09:39<15:03:27, 2.31s/it] 22%|██▏ | 6494/30000 [4:09:42<15:00:32, 2.30s/it] 22%|██▏ | 6495/30000 [4:09:44<15:02:16, 2.30s/it] 22%|██▏ | 6496/30000 [4:09:46<15:06:25, 2.31s/it] 22%|██▏ | 6497/30000 [4:09:49<15:00:52, 2.30s/it] 22%|██▏ | 6498/30000 [4:09:51<15:01:55, 2.30s/it] 22%|██▏ | 6499/30000 [4:09:53<15:06:37, 2.31s/it] 22%|██▏ | 6500/30000 [4:09:55<15:01:22, 2.30s/it] 22%|██▏ | 6500/30000 [4:09:55<15:01:22, 2.30s/it] 22%|██▏ | 6501/30000 [4:09:58<15:02:32, 2.30s/it] 22%|██▏ | 6502/30000 [4:10:00<15:06:34, 2.31s/it] 22%|██▏ | 6503/30000 [4:10:02<15:07:24, 2.32s/it] 22%|██▏ | 6504/30000 [4:10:05<15:02:21, 2.30s/it] 22%|██▏ | 6505/30000 [4:10:07<14:58:55, 2.30s/it] 22%|██▏ | 6506/30000 [4:10:09<15:08:42, 2.32s/it] 22%|██▏ | 6507/30000 [4:10:12<15:09:51, 2.32s/it] 22%|██▏ | 6508/30000 [4:10:14<15:02:26, 2.30s/it] 22%|██▏ | 6509/30000 [4:10:16<14:59:15, 2.30s/it] 22%|██▏ | 6510/30000 [4:10:18<14:54:53, 2.29s/it] 22%|██▏ | 6510/30000 [4:10:19<14:54:53, 2.29s/it] 22%|██▏ | 6511/30000 [4:10:21<14:53:27, 2.28s/it] 22%|██▏ | 6512/30000 [4:10:23<14:51:33, 2.28s/it] 22%|██▏ | 6513/30000 [4:10:25<14:56:23, 2.29s/it] 22%|██▏ | 6514/30000 [4:10:28<14:59:44, 2.30s/it] 22%|██▏ | 6515/30000 [4:10:30<14:59:42, 2.30s/it] 22%|██▏ | 6516/30000 [4:10:32<14:59:02, 2.30s/it] 22%|██▏ | 6517/30000 [4:10:35<14:59:58, 2.30s/it] 22%|██▏ | 6518/30000 [4:10:37<14:54:12, 2.28s/it] 22%|██▏ | 6519/30000 [4:10:39<14:51:47, 2.28s/it] 22%|██▏ | 6520/30000 [4:10:41<15:03:04, 2.31s/it] 22%|██▏ | 6520/30000 [4:10:41<15:03:04, 2.31s/it] 22%|██▏ | 6521/30000 [4:10:44<15:05:29, 2.31s/it] 22%|██▏ | 6522/30000 [4:10:46<15:00:00, 2.30s/it] 22%|██▏ | 6523/30000 [4:10:48<14:59:33, 2.30s/it] 22%|██▏ | 6524/30000 [4:10:51<14:54:39, 2.29s/it] 22%|██▏ | 6525/30000 [4:10:53<14:53:29, 2.28s/it] 22%|██▏ | 6526/30000 [4:10:55<14:52:54, 2.28s/it] 22%|██▏ | 6527/30000 [4:10:58<15:02:28, 2.31s/it] 22%|██▏ | 6528/30000 [4:11:00<15:02:58, 2.31s/it] 22%|██▏ | 6529/30000 [4:11:02<15:01:37, 2.30s/it] 22%|██▏ | 6530/30000 [4:11:04<15:04:17, 2.31s/it] 22%|██▏ | 6530/30000 [4:11:04<15:04:17, 2.31s/it] 22%|██▏ | 6531/30000 [4:11:07<14:58:15, 2.30s/it] 22%|██▏ | 6532/30000 [4:11:09<14:54:42, 2.29s/it] 22%|██▏ | 6533/30000 [4:11:11<15:02:08, 2.31s/it] 22%|██▏ | 6534/30000 [4:11:14<15:01:27, 2.30s/it] 22%|██▏ | 6535/30000 [4:11:16<14:53:27, 2.28s/it] 22%|██▏ | 6536/30000 [4:11:18<14:51:10, 2.28s/it] 22%|██▏ | 6537/30000 [4:11:20<15:00:08, 2.30s/it] 22%|██▏ | 6538/30000 [4:11:23<14:54:44, 2.29s/it] 22%|██▏ | 6539/30000 [4:11:25<14:55:35, 2.29s/it] 22%|██▏ | 6540/30000 [4:11:27<15:00:10, 2.30s/it] 22%|██▏ | 6540/30000 [4:11:27<15:00:10, 2.30s/it] 22%|██▏ | 6541/30000 [4:11:30<15:01:41, 2.31s/it] 22%|██▏ | 6542/30000 [4:11:32<14:57:04, 2.29s/it] 22%|██▏ | 6543/30000 [4:11:34<15:05:34, 2.32s/it] 22%|██▏ | 6544/30000 [4:11:37<15:05:22, 2.32s/it] 22%|██▏ | 6545/30000 [4:11:39<15:03:10, 2.31s/it] 22%|██▏ | 6546/30000 [4:11:41<14:59:23, 2.30s/it] 22%|██▏ | 6547/30000 [4:11:44<15:03:18, 2.31s/it] 22%|██▏ | 6548/30000 [4:11:46<14:55:03, 2.29s/it] 22%|██▏ | 6549/30000 [4:11:48<14:57:32, 2.30s/it] 22%|██▏ | 6550/30000 [4:11:50<15:02:34, 2.31s/it] 22%|██▏ | 6550/30000 [4:11:50<15:02:34, 2.31s/it] 22%|██▏ | 6551/30000 [4:11:53<15:04:29, 2.31s/it] 22%|██▏ | 6552/30000 [4:11:55<14:59:30, 2.30s/it] 22%|██▏ | 6553/30000 [4:11:57<15:04:41, 2.32s/it] 22%|██▏ | 6554/30000 [4:12:00<15:04:02, 2.31s/it] 22%|██▏ | 6555/30000 [4:12:02<14:59:44, 2.30s/it] 22%|██▏ | 6556/30000 [4:12:04<14:57:33, 2.30s/it] 22%|██▏ | 6557/30000 [4:12:07<15:05:40, 2.32s/it] 22%|██▏ | 6558/30000 [4:12:09<15:01:36, 2.31s/it] 22%|██▏ | 6559/30000 [4:12:11<14:54:50, 2.29s/it] 22%|██▏ | 6560/30000 [4:12:14<15:01:36, 2.31s/it] 22%|██▏ | 6560/30000 [4:12:14<15:01:36, 2.31s/it] 22%|██▏ | 6561/30000 [4:12:16<15:04:09, 2.31s/it] 22%|██▏ | 6562/30000 [4:12:18<14:59:13, 2.30s/it] 22%|██▏ | 6563/30000 [4:12:20<15:01:48, 2.31s/it] 22%|██▏ | 6564/30000 [4:12:23<15:02:38, 2.31s/it] 22%|██▏ | 6565/30000 [4:12:25<14:53:29, 2.29s/it] 22%|██▏ | 6566/30000 [4:12:27<14:55:57, 2.29s/it] 22%|██▏ | 6567/30000 [4:12:30<15:01:49, 2.31s/it] 22%|██▏ | 6568/30000 [4:12:32<15:02:03, 2.31s/it] 22%|██▏ | 6569/30000 [4:12:34<15:00:44, 2.31s/it] 22%|██▏ | 6570/30000 [4:12:37<15:03:49, 2.31s/it] 22%|██▏ | 6570/30000 [4:12:37<15:03:49, 2.31s/it] 22%|██▏ | 6571/30000 [4:12:39<15:03:37, 2.31s/it] 22%|██▏ | 6572/30000 [4:12:41<15:01:59, 2.31s/it] 22%|██▏ | 6573/30000 [4:12:43<14:55:45, 2.29s/it] 22%|██▏ | 6574/30000 [4:12:46<15:00:31, 2.31s/it] 22%|██▏ | 6575/30000 [4:12:48<14:56:12, 2.30s/it] 22%|██▏ | 6576/30000 [4:12:50<14:52:12, 2.29s/it] 22%|██▏ | 6577/30000 [4:12:53<15:01:20, 2.31s/it] 22%|██▏ | 6578/30000 [4:12:55<14:55:37, 2.29s/it] 22%|██▏ | 6579/30000 [4:12:57<14:57:26, 2.30s/it] 22%|██▏ | 6580/30000 [4:13:00<15:04:18, 2.32s/it] 22%|██▏ | 6580/30000 [4:13:00<15:04:18, 2.32s/it] 22%|██▏ | 6581/30000 [4:13:02<15:04:30, 2.32s/it] 22%|██▏ | 6582/30000 [4:13:04<14:54:21, 2.29s/it] 22%|██▏ | 6583/30000 [4:13:06<14:58:39, 2.30s/it] 22%|██▏ | 6584/30000 [4:13:09<14:54:46, 2.29s/it] 22%|██▏ | 6585/30000 [4:13:11<14:57:29, 2.30s/it] 22%|██▏ | 6586/30000 [4:13:13<14:56:01, 2.30s/it] 22%|██▏ | 6587/30000 [4:13:16<14:57:50, 2.30s/it] 22%|██▏ | 6588/30000 [4:13:18<14:53:14, 2.29s/it] 22%|██▏ | 6589/30000 [4:13:20<14:49:55, 2.28s/it] 22%|██▏ | 6590/30000 [4:13:23<14:58:05, 2.30s/it] 22%|██▏ | 6590/30000 [4:13:23<14:58:05, 2.30s/it] 22%|██▏ | 6591/30000 [4:13:25<14:58:01, 2.30s/it] 22%|██▏ | 6592/30000 [4:13:27<14:54:26, 2.29s/it] 22%|██▏ | 6593/30000 [4:13:29<15:00:20, 2.31s/it] 22%|██▏ | 6594/30000 [4:13:32<15:05:36, 2.32s/it] 22%|██▏ | 6595/30000 [4:13:34<15:00:59, 2.31s/it] 22%|██▏ | 6596/30000 [4:13:36<15:05:51, 2.32s/it] 22%|██▏ | 6597/30000 [4:13:39<15:03:06, 2.32s/it] 22%|██▏ | 6598/30000 [4:13:41<14:58:40, 2.30s/it] 22%|██▏ | 6599/30000 [4:13:43<14:55:25, 2.30s/it] 22%|██▏ | 6600/30000 [4:13:46<14:54:06, 2.29s/it] 22%|██▏ | 6600/30000 [4:13:46<14:54:06, 2.29s/it] 22%|██▏ | 6601/30000 [4:13:48<14:56:07, 2.30s/it] 22%|██▏ | 6602/30000 [4:13:50<14:52:49, 2.29s/it] 22%|██▏ | 6603/30000 [4:13:52<14:55:02, 2.30s/it] 22%|██▏ | 6604/30000 [4:13:55<14:51:16, 2.29s/it] 22%|██▏ | 6605/30000 [4:13:57<14:47:52, 2.28s/it] 22%|██▏ | 6606/30000 [4:13:59<14:55:58, 2.30s/it] 22%|██▏ | 6607/30000 [4:14:02<14:52:46, 2.29s/it] 22%|██▏ | 6608/30000 [4:14:04<14:52:41, 2.29s/it] 22%|██▏ | 6609/30000 [4:14:06<14:57:48, 2.30s/it] 22%|██▏ | 6610/30000 [4:14:09<14:59:41, 2.31s/it] 22%|██▏ | 6610/30000 [4:14:09<14:59:41, 2.31s/it] 22%|██▏ | 6611/30000 [4:14:11<14:57:53, 2.30s/it] 22%|██▏ | 6612/30000 [4:14:13<14:54:13, 2.29s/it] 22%|██▏ | 6613/30000 [4:14:15<15:00:46, 2.31s/it] 22%|██▏ | 6614/30000 [4:14:18<14:56:35, 2.30s/it] 22%|██▏ | 6615/30000 [4:14:20<14:51:22, 2.29s/it] 22%|██▏ | 6616/30000 [4:14:22<14:58:26, 2.31s/it] 22%|██▏ | 6617/30000 [4:14:25<14:59:40, 2.31s/it] 22%|██▏ | 6618/30000 [4:14:27<14:52:16, 2.29s/it] 22%|██▏ | 6619/30000 [4:14:29<14:57:23, 2.30s/it] 22%|██▏ | 6620/30000 [4:14:32<14:57:58, 2.30s/it] 22%|██▏ | 6620/30000 [4:14:32<14:57:58, 2.30s/it] 22%|██▏ | 6621/30000 [4:14:34<14:53:50, 2.29s/it] 22%|██▏ | 6622/30000 [4:14:36<14:50:09, 2.28s/it] 22%|██▏ | 6623/30000 [4:14:38<14:58:47, 2.31s/it] 22%|██▏ | 6624/30000 [4:14:41<15:06:02, 2.33s/it] 22%|██▏ | 6625/30000 [4:14:43<15:00:29, 2.31s/it] 22%|██▏ | 6626/30000 [4:14:45<14:55:20, 2.30s/it] 22%|██▏ | 6627/30000 [4:14:48<14:53:11, 2.29s/it] 22%|██▏ | 6628/30000 [4:14:50<14:53:03, 2.29s/it] 22%|██▏ | 6629/30000 [4:14:52<14:50:10, 2.29s/it] 22%|██▏ | 6630/30000 [4:14:55<14:56:25, 2.30s/it] 22%|██▏ | 6630/30000 [4:14:55<14:56:25, 2.30s/it] 22%|██▏ | 6631/30000 [4:14:57<14:54:29, 2.30s/it] 22%|██▏ | 6632/30000 [4:14:59<14:52:36, 2.29s/it] 22%|██▏ | 6633/30000 [4:15:01<14:53:14, 2.29s/it] 22%|██▏ | 6634/30000 [4:15:04<14:55:53, 2.30s/it] 22%|██▏ | 6635/30000 [4:15:06<14:55:41, 2.30s/it] 22%|██▏ | 6636/30000 [4:15:08<14:48:19, 2.28s/it] 22%|██▏ | 6637/30000 [4:15:11<14:54:50, 2.30s/it] 22%|██▏ | 6638/30000 [4:15:13<14:58:49, 2.31s/it] 22%|██▏ | 6639/30000 [4:15:15<14:53:01, 2.29s/it] 22%|██▏ | 6640/30000 [4:15:17<14:51:07, 2.29s/it] 22%|██▏ | 6640/30000 [4:15:18<14:51:07, 2.29s/it] 22%|██▏ | 6641/30000 [4:15:20<15:02:43, 2.32s/it] 22%|██▏ | 6642/30000 [4:15:22<14:59:23, 2.31s/it] 22%|██▏ | 6643/30000 [4:15:24<14:57:24, 2.31s/it] 22%|██▏ | 6644/30000 [4:15:27<14:54:08, 2.30s/it] 22%|██▏ | 6645/30000 [4:15:29<14:56:53, 2.30s/it] 22%|██▏ | 6646/30000 [4:15:31<14:50:30, 2.29s/it] 22%|██▏ | 6647/30000 [4:15:34<14:44:06, 2.27s/it] 22%|██▏ | 6648/30000 [4:15:36<14:51:09, 2.29s/it] 22%|██▏ | 6649/30000 [4:15:38<14:54:18, 2.30s/it] 22%|██▏ | 6650/30000 [4:15:40<14:51:08, 2.29s/it] 22%|██▏ | 6650/30000 [4:15:40<14:51:08, 2.29s/it] 22%|██▏ | 6651/30000 [4:15:43<14:46:55, 2.28s/it] 22%|██▏ | 6652/30000 [4:15:45<14:51:00, 2.29s/it] 22%|██▏ | 6653/30000 [4:15:47<14:50:55, 2.29s/it] 22%|██▏ | 6654/30000 [4:15:50<14:42:09, 2.27s/it] 22%|██▏ | 6655/30000 [4:15:52<14:48:34, 2.28s/it] 22%|██▏ | 6656/30000 [4:15:54<14:53:24, 2.30s/it] 22%|██▏ | 6657/30000 [4:15:56<14:52:06, 2.29s/it] 22%|██▏ | 6658/30000 [4:15:59<14:47:48, 2.28s/it] 22%|██▏ | 6659/30000 [4:16:01<14:55:56, 2.30s/it] 22%|██▏ | 6660/30000 [4:16:03<14:55:34, 2.30s/it] 22%|██▏ | 6660/30000 [4:16:03<14:55:34, 2.30s/it] 22%|██▏ | 6661/30000 [4:16:06<14:53:05, 2.30s/it] 22%|██▏ | 6662/30000 [4:16:08<14:50:24, 2.29s/it] 22%|██▏ | 6663/30000 [4:16:10<14:55:05, 2.30s/it] 22%|██▏ | 6664/30000 [4:16:13<14:54:23, 2.30s/it] 22%|██▏ | 6665/30000 [4:16:15<14:50:28, 2.29s/it] 22%|██▏ | 6666/30000 [4:16:17<14:49:11, 2.29s/it] 22%|██▏ | 6667/30000 [4:16:19<14:59:54, 2.31s/it] 22%|██▏ | 6668/30000 [4:16:22<15:00:27, 2.32s/it] 22%|██▏ | 6669/30000 [4:16:24<14:55:32, 2.30s/it] 22%|██▏ | 6670/30000 [4:16:26<14:52:28, 2.30s/it] 22%|██▏ | 6670/30000 [4:16:26<14:52:28, 2.30s/it] 22%|██▏ | 6671/30000 [4:16:29<14:57:14, 2.31s/it] 22%|██▏ | 6672/30000 [4:16:31<14:49:36, 2.29s/it] 22%|██▏ | 6673/30000 [4:16:33<14:51:23, 2.29s/it] 22%|██▏ | 6674/30000 [4:16:35<14:49:16, 2.29s/it] 22%|██▏ | 6675/30000 [4:16:38<14:51:09, 2.29s/it] 22%|██▏ | 6676/30000 [4:16:40<14:46:25, 2.28s/it] 22%|██▏ | 6677/30000 [4:16:42<14:47:21, 2.28s/it] 22%|██▏ | 6678/30000 [4:16:45<14:51:38, 2.29s/it] 22%|██▏ | 6679/30000 [4:16:47<14:58:28, 2.31s/it] 22%|██▏ | 6680/30000 [4:16:49<14:51:13, 2.29s/it] 22%|██▏ | 6680/30000 [4:16:49<14:51:13, 2.29s/it] 22%|██▏ | 6681/30000 [4:16:52<14:50:12, 2.29s/it] 22%|██▏ | 6682/30000 [4:16:54<14:55:04, 2.30s/it] 22%|██▏ | 6683/30000 [4:16:56<14:54:02, 2.30s/it] 22%|██▏ | 6684/30000 [4:16:58<14:51:17, 2.29s/it] 22%|██▏ | 6685/30000 [4:17:01<14:50:39, 2.29s/it] 22%|██▏ | 6686/30000 [4:17:03<14:53:20, 2.30s/it] 22%|██▏ | 6687/30000 [4:17:05<14:45:15, 2.28s/it] 22%|██▏ | 6688/30000 [4:17:08<14:38:17, 2.26s/it] 22%|██▏ | 6689/30000 [4:17:10<14:47:55, 2.29s/it] 22%|██▏ | 6690/30000 [4:17:12<14:48:12, 2.29s/it] +{'loss': 0.0175, 'grad_norm': 0.0904102623462677, 'learning_rate': 2.795801845339969e-05, 'flow_matching_loss': 0.017522669583559036, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.83} +{'loss': 0.0162, 'grad_norm': 0.09380383044481277, 'learning_rate': 2.7949681750273208e-05, 'flow_matching_loss': 0.016194858588278294, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.83} +{'loss': 0.0153, 'grad_norm': 0.07979407906532288, 'learning_rate': 2.7941329312062322e-05, 'flow_matching_loss': 0.015287991892546416, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.84} +{'loss': 0.0173, 'grad_norm': 0.09106510877609253, 'learning_rate': 2.7932961148916037e-05, 'flow_matching_loss': 0.017290185578167437, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.84} +{'loss': 0.0154, 'grad_norm': 0.07037898153066635, 'learning_rate': 2.7924577271002456e-05, 'flow_matching_loss': 0.015430368855595589, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.84} +{'loss': 0.0161, 'grad_norm': 0.05681585147976875, 'learning_rate': 2.7916177688508782e-05, 'flow_matching_loss': 0.01609450001269579, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.84} +{'loss': 0.0171, 'grad_norm': 0.09189490973949432, 'learning_rate': 2.7907762411641302e-05, 'flow_matching_loss': 0.01707817576825619, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.85} +{'loss': 0.0183, 'grad_norm': 0.10156885534524918, 'learning_rate': 2.7899331450625367e-05, 'flow_matching_loss': 0.018344346061348916, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.85} +{'loss': 0.0178, 'grad_norm': 0.08959219604730606, 'learning_rate': 2.7890884815705397e-05, 'flow_matching_loss': 0.017775400914251803, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.85} +{'loss': 0.0186, 'grad_norm': 0.0971435010433197, 'learning_rate': 2.788242251714484e-05, 'flow_matching_loss': 0.018578748032450677, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.86} +{'loss': 0.0172, 'grad_norm': 0.11674308031797409, 'learning_rate': 2.787394456522619e-05, 'flow_matching_loss': 0.017189202923327686, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.86} +{'loss': 0.0185, 'grad_norm': 0.0848507434129715, 'learning_rate': 2.7865450970250965e-05, 'flow_matching_loss': 0.018492572568356992, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.86} +{'loss': 0.0163, 'grad_norm': 0.09937480092048645, 'learning_rate': 2.7856941742539676e-05, 'flow_matching_loss': 0.01633382961153984, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.86} +{'loss': 0.0162, 'grad_norm': 0.08764338493347168, 'learning_rate': 2.784841689243184e-05, 'flow_matching_loss': 0.016189302131533623, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.87} +{'loss': 0.0177, 'grad_norm': 0.08089357614517212, 'learning_rate': 2.783987643028595e-05, 'flow_matching_loss': 0.017714053113013505, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.87} +{'loss': 0.0169, 'grad_norm': 0.10263040661811829, 'learning_rate': 2.783132036647948e-05, 'flow_matching_loss': 0.016881211940199137, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.87} +{'loss': 0.0175, 'grad_norm': 0.11991552263498306, 'learning_rate': 2.7822748711408852e-05, 'flow_matching_loss': 0.017497071344405413, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.88} +{'loss': 0.0154, 'grad_norm': 0.08834826201200485, 'learning_rate': 2.7814161475489428e-05, 'flow_matching_loss': 0.015395635645836592, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.88} +{'loss': 0.0171, 'grad_norm': 0.07557949423789978, 'learning_rate': 2.780555866915552e-05, 'flow_matching_loss': 0.017094421107321976, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.88} +{'loss': 0.0181, 'grad_norm': 0.08473949879407883, 'learning_rate': 2.7796940302860338e-05, 'flow_matching_loss': 0.018055094685405493, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.89} +{'loss': 0.0164, 'grad_norm': 0.07579516619443893, 'learning_rate': 2.7788306387076018e-05, 'flow_matching_loss': 0.01640403354540467, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.89} +{'loss': 0.0173, 'grad_norm': 0.09215551614761353, 'learning_rate': 2.7779656932293584e-05, 'flow_matching_loss': 0.017296364065259695, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.89} +{'loss': 0.0171, 'grad_norm': 0.07609230279922485, 'learning_rate': 2.777099194902293e-05, 'flow_matching_loss': 0.017104742862284184, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.89} +{'loss': 0.0158, 'grad_norm': 0.09556776285171509, 'learning_rate': 2.776231144779283e-05, 'flow_matching_loss': 0.01584559753537178, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.9} +{'loss': 0.0164, 'grad_norm': 0.09666924923658371, 'learning_rate': 2.775361543915091e-05, 'flow_matching_loss': 0.016443815547972918, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.9} +{'loss': 0.0169, 'grad_norm': 0.09146234393119812, 'learning_rate': 2.774490393366365e-05, 'flow_matching_loss': 0.01686391718685627, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.9} +{'loss': 0.0169, 'grad_norm': 0.08843562006950378, 'learning_rate': 2.7736176941916346e-05, 'flow_matching_loss': 0.016857012733817102, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.91} +{'loss': 0.0176, 'grad_norm': 0.09796661883592606, 'learning_rate': 2.772743447451311e-05, 'flow_matching_loss': 0.01760939313098788, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.91} +{'loss': 0.0157, 'grad_norm': 0.07968240231275558, 'learning_rate': 2.7718676542076872e-05, 'flow_matching_loss': 0.015726991835981606, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.91} +{'loss': 0.0159, 'grad_norm': 0.06395082175731659, 'learning_rate': 2.770990315524934e-05, 'flow_matching_loss': 0.015904546342790128, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.91} +{'loss': 0.0171, 'grad_norm': 0.08119373023509979, 'learning_rate': 2.7701114324691015e-05, 'flow_matching_loss': 0.017076310887932777, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.92} +{'loss': 0.0162, 'grad_norm': 0.0991370677947998, 'learning_rate': 2.7692310061081145e-05, 'flow_matching_loss': 0.016194525733590127, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.92} +{'loss': 0.0163, 'grad_norm': 0.10749704390764236, 'learning_rate': 2.7683490375117742e-05, 'flow_matching_loss': 0.016300099436193705, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.92} +{'loss': 0.0166, 'grad_norm': 0.06251049041748047, 'learning_rate': 2.7674655277517556e-05, 'flow_matching_loss': 0.016560526192188264, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.93} +{'loss': 0.017, 'grad_norm': 0.12784874439239502, 'learning_rate': 2.7665804779016072e-05, 'flow_matching_loss': 0.01695288047194481, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.93} +{'loss': 0.0156, 'grad_norm': 0.09243197739124298, 'learning_rate': 2.765693889036747e-05, 'flow_matching_loss': 0.015560354571789503, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.93} +{'loss': 0.0164, 'grad_norm': 0.08910775184631348, 'learning_rate': 2.764805762234465e-05, 'flow_matching_loss': 0.01643256563693285, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.93} +{'loss': 0.0181, 'grad_norm': 0.0979708731174469, 'learning_rate': 2.7639160985739182e-05, 'flow_matching_loss': 0.018131051398813724, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.94} +{'loss': 0.0176, 'grad_norm': 0.07146275788545609, 'learning_rate': 2.7630248991361317e-05, 'flow_matching_loss': 0.017591935582458973, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.94} +{'loss': 0.0183, 'grad_norm': 0.09048923850059509, 'learning_rate': 2.762132165003998e-05, 'flow_matching_loss': 0.018344883993268012, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.94} +{'loss': 0.0159, 'grad_norm': 0.09246297180652618, 'learning_rate': 2.7612378972622722e-05, 'flow_matching_loss': 0.015901629626750947, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.95} 22%|██▏ | 6690/30000 [4:17:12<14:48:12, 2.29s/it] 22%|██▏ | 6691/30000 [4:17:14<14:42:02, 2.27s/it] 22%|██▏ | 6692/30000 [4:17:17<14:47:10, 2.28s/it] 22%|██▏ | 6693/30000 [4:17:19<14:47:36, 2.29s/it] 22%|██▏ | 6694/30000 [4:17:21<14:46:58, 2.28s/it] 22%|██▏ | 6695/30000 [4:17:24<14:43:49, 2.28s/it] 22%|██▏ | 6696/30000 [4:17:26<14:47:56, 2.29s/it] 22%|██▏ | 6697/30000 [4:17:28<14:48:03, 2.29s/it] 22%|██▏ | 6698/30000 [4:17:30<14:43:22, 2.27s/it] 22%|██▏ | 6699/30000 [4:17:33<14:51:29, 2.30s/it] 22%|██▏ | 6700/30000 [4:17:35<14:52:37, 2.30s/it] 22%|██▏ | 6700/30000 [4:17:35<14:52:37, 2.30s/it] 22%|██▏ | 6701/30000 [4:17:37<14:51:16, 2.30s/it] 22%|██▏ | 6702/30000 [4:17:40<14:49:18, 2.29s/it] 22%|██▏ | 6703/30000 [4:17:42<14:57:40, 2.31s/it] 22%|██▏ | 6704/30000 [4:17:44<14:58:16, 2.31s/it] 22%|██▏ | 6705/30000 [4:17:47<14:54:46, 2.30s/it] 22%|██▏ | 6706/30000 [4:17:49<14:58:34, 2.31s/it] 22%|██▏ | 6707/30000 [4:17:51<15:02:05, 2.32s/it] 22%|██▏ | 6708/30000 [4:17:53<14:53:19, 2.30s/it] 22%|██▏ | 6709/30000 [4:17:56<14:50:42, 2.29s/it] 22%|██▏ | 6710/30000 [4:17:58<15:00:47, 2.32s/it] 22%|██▏ | 6710/30000 [4:17:58<15:00:47, 2.32s/it] 22%|██▏ | 6711/30000 [4:18:00<15:02:34, 2.33s/it] 22%|██▏ | 6712/30000 [4:18:03<14:57:57, 2.31s/it] 22%|██▏ | 6713/30000 [4:18:05<14:56:31, 2.31s/it] 22%|██▏ | 6714/30000 [4:18:07<14:55:40, 2.31s/it] 22%|██▏ | 6715/30000 [4:18:10<14:52:02, 2.30s/it] 22%|██▏ | 6716/30000 [4:18:12<14:47:11, 2.29s/it] 22%|██▏ | 6717/30000 [4:18:14<14:51:24, 2.30s/it] 22%|██▏ | 6718/30000 [4:18:17<14:56:18, 2.31s/it] 22%|██▏ | 6719/30000 [4:18:19<14:49:46, 2.29s/it] 22%|██▏ | 6720/30000 [4:18:21<14:48:46, 2.29s/it] 22%|██▏ | 6720/30000 [4:18:21<14:48:46, 2.29s/it] 22%|██▏ | 6721/30000 [4:18:23<14:55:44, 2.31s/it] 22%|██▏ | 6722/30000 [4:18:26<14:54:18, 2.31s/it] 22%|██▏ | 6723/30000 [4:18:28<14:47:00, 2.29s/it] 22%|██▏ | 6724/30000 [4:18:30<14:52:13, 2.30s/it] 22%|██▏ | 6725/30000 [4:18:33<14:55:50, 2.31s/it] 22%|██▏ | 6726/30000 [4:18:35<14:55:22, 2.31s/it] 22%|██▏ | 6727/30000 [4:18:37<14:57:18, 2.31s/it] 22%|██▏ | 6728/30000 [4:18:40<14:54:55, 2.31s/it] 22%|██▏ | 6729/30000 [4:18:42<14:52:55, 2.30s/it] 22%|██▏ | 6730/30000 [4:18:44<14:47:52, 2.29s/it] 22%|██▏ | 6730/30000 [4:18:44<14:47:52, 2.29s/it] 22%|██▏ | 6731/30000 [4:18:46<14:56:50, 2.31s/it] 22%|██▏ | 6732/30000 [4:18:49<14:58:11, 2.32s/it] 22%|██▏ | 6733/30000 [4:18:51<14:51:54, 2.30s/it] 22%|██▏ | 6734/30000 [4:18:53<14:54:05, 2.31s/it] 22%|██▏ | 6735/30000 [4:18:56<14:52:00, 2.30s/it] 22%|██▏ | 6736/30000 [4:18:58<14:48:29, 2.29s/it] 22%|██▏ | 6737/30000 [4:19:00<14:49:15, 2.29s/it] 22%|██▏ | 6738/30000 [4:19:03<14:54:01, 2.31s/it] 22%|██▏ | 6739/30000 [4:19:05<14:55:16, 2.31s/it] 22%|██▏ | 6740/30000 [4:19:07<14:48:29, 2.29s/it] 22%|██▏ | 6740/30000 [4:19:07<14:48:29, 2.29s/it] 22%|██▏ | 6741/30000 [4:19:10<14:55:54, 2.31s/it] 22%|██▏ | 6742/30000 [4:19:12<14:54:02, 2.31s/it] 22%|██▏ | 6743/30000 [4:19:14<14:53:55, 2.31s/it] 22%|██▏ | 6744/30000 [4:19:16<14:50:22, 2.30s/it] 22%|██▏ | 6745/30000 [4:19:19<14:55:50, 2.31s/it] 22%|██▏ | 6746/30000 [4:19:21<14:51:19, 2.30s/it] 22%|██▏ | 6747/30000 [4:19:23<14:49:20, 2.29s/it] 22%|██▏ | 6748/30000 [4:19:26<14:54:14, 2.31s/it] 22%|██▏ | 6749/30000 [4:19:28<14:53:40, 2.31s/it] 22%|██▎ | 6750/30000 [4:19:30<14:48:05, 2.29s/it] 22%|██▎ | 6750/30000 [4:19:30<14:48:05, 2.29s/it] 23%|██▎ | 6751/30000 [4:19:32<14:48:26, 2.29s/it] 23%|██▎ | 6752/30000 [4:19:35<14:53:06, 2.30s/it] 23%|██▎ | 6753/30000 [4:19:37<14:46:48, 2.29s/it] 23%|██▎ | 6754/30000 [4:19:39<14:40:57, 2.27s/it] 23%|██▎ | 6755/30000 [4:19:42<14:49:38, 2.30s/it] 23%|██▎ | 6756/30000 [4:19:44<14:51:34, 2.30s/it] 23%|██▎ | 6757/30000 [4:19:46<14:46:13, 2.29s/it] 23%|██▎ | 6758/30000 [4:19:49<14:47:22, 2.29s/it] 23%|██▎ | 6759/30000 [4:19:51<14:50:04, 2.30s/it] 23%|██▎ | 6760/30000 [4:19:53<14:49:14, 2.30s/it] 23%|██▎ | 6760/30000 [4:19:53<14:49:14, 2.30s/it] 23%|██▎ | 6761/30000 [4:19:55<14:50:03, 2.30s/it] 23%|██▎ | 6762/30000 [4:19:58<14:54:15, 2.31s/it] 23%|██▎ | 6763/30000 [4:20:00<14:56:53, 2.32s/it] 23%|██▎ | 6764/30000 [4:20:02<14:52:03, 2.30s/it] 23%|██▎ | 6765/30000 [4:20:05<14:58:27, 2.32s/it] 23%|██▎ | 6766/30000 [4:20:07<14:57:35, 2.32s/it] 23%|██▎ | 6767/30000 [4:20:09<14:52:20, 2.30s/it] 23%|██▎ | 6768/30000 [4:20:12<14:57:42, 2.32s/it] 23%|██▎ | 6769/30000 [4:20:14<14:58:32, 2.32s/it] 23%|██▎ | 6770/30000 [4:20:16<14:51:45, 2.30s/it] 23%|██▎ | 6770/30000 [4:20:16<14:51:45, 2.30s/it] 23%|██▎ | 6771/30000 [4:20:19<14:57:48, 2.32s/it] 23%|██▎ | 6772/30000 [4:20:21<14:53:15, 2.31s/it] 23%|██▎ | 6773/30000 [4:20:23<14:48:55, 2.30s/it] 23%|██▎ | 6774/30000 [4:20:25<14:45:38, 2.29s/it] 23%|██▎ | 6775/30000 [4:20:28<14:53:08, 2.31s/it] 23%|██▎ | 6776/30000 [4:20:30<14:49:49, 2.30s/it] 23%|██▎ | 6777/30000 [4:20:32<14:49:10, 2.30s/it] 23%|██▎ | 6778/30000 [4:20:35<14:51:58, 2.30s/it] 23%|██▎ | 6779/30000 [4:20:37<14:54:42, 2.31s/it] 23%|██▎ | 6780/30000 [4:20:39<14:50:00, 2.30s/it] 23%|██▎ | 6780/30000 [4:20:39<14:50:00, 2.30s/it] 23%|██▎ | 6781/30000 [4:20:42<14:55:33, 2.31s/it] 23%|██▎ | 6782/30000 [4:20:44<14:48:29, 2.30s/it] 23%|██▎ | 6783/30000 [4:20:46<14:47:27, 2.29s/it] 23%|██▎ | 6784/30000 [4:20:48<14:50:28, 2.30s/it] 23%|██▎ | 6785/30000 [4:20:51<14:54:08, 2.31s/it] 23%|██▎ | 6786/30000 [4:20:53<14:51:13, 2.30s/it] 23%|██▎ | 6787/30000 [4:20:55<14:52:06, 2.31s/it] 23%|██▎ | 6788/30000 [4:20:58<14:55:58, 2.32s/it] 23%|██▎ | 6789/30000 [4:21:00<14:53:29, 2.31s/it] 23%|██▎ | 6790/30000 [4:21:02<14:47:57, 2.30s/it] 23%|██▎ | 6790/30000 [4:21:02<14:47:57, 2.30s/it] 23%|██▎ | 6791/30000 [4:21:05<14:54:51, 2.31s/it] 23%|██▎ | 6792/30000 [4:21:07<14:52:11, 2.31s/it] 23%|██▎ | 6793/30000 [4:21:09<14:51:22, 2.30s/it] 23%|██▎ | 6794/30000 [4:21:12<14:52:27, 2.31s/it] 23%|██▎ | 6795/30000 [4:21:14<15:00:16, 2.33s/it] 23%|██▎ | 6796/30000 [4:21:16<14:50:23, 2.30s/it] 23%|██▎ | 6797/30000 [4:21:18<14:46:43, 2.29s/it] 23%|██▎ | 6798/30000 [4:21:21<14:51:04, 2.30s/it] 23%|██▎ | 6799/30000 [4:21:23<14:46:17, 2.29s/it] 23%|██▎ | 6800/30000 [4:21:25<14:44:45, 2.29s/it] 23%|██▎ | 6800/30000 [4:21:25<14:44:45, 2.29s/it] 23%|██▎ | 6801/30000 [4:21:28<14:54:27, 2.31s/it] 23%|██▎ | 6802/30000 [4:21:30<14:51:04, 2.30s/it] 23%|██▎ | 6803/30000 [4:21:32<14:53:18, 2.31s/it] 23%|██▎ | 6804/30000 [4:21:35<14:48:28, 2.30s/it] 23%|██▎ | 6805/30000 [4:21:37<14:55:00, 2.32s/it] 23%|██▎ | 6806/30000 [4:21:39<14:48:47, 2.30s/it] 23%|██▎ | 6807/30000 [4:21:41<14:39:03, 2.27s/it] 23%|██▎ | 6808/30000 [4:21:44<14:49:16, 2.30s/it] 23%|██▎ | 6809/30000 [4:21:46<14:54:05, 2.31s/it] 23%|██▎ | 6810/30000 [4:21:48<14:47:35, 2.30s/it] 23%|██▎ | 6810/30000 [4:21:48<14:47:35, 2.30s/it] 23%|██▎ | 6811/30000 [4:21:51<14:48:58, 2.30s/it] 23%|██▎ | 6812/30000 [4:21:53<14:49:19, 2.30s/it] 23%|██▎ | 6813/30000 [4:21:55<14:44:11, 2.29s/it] 23%|██▎ | 6814/30000 [4:21:57<14:38:34, 2.27s/it] 23%|██▎ | 6815/30000 [4:22:00<14:45:50, 2.29s/it] 23%|██▎ | 6816/30000 [4:22:02<14:39:41, 2.28s/it] 23%|██▎ | 6817/30000 [4:22:04<14:38:45, 2.27s/it] 23%|██▎ | 6818/30000 [4:22:07<14:44:29, 2.29s/it] 23%|██▎ | 6819/30000 [4:22:09<14:48:21, 2.30s/it] 23%|██▎ | 6820/30000 [4:22:11<14:44:49, 2.29s/it] 23%|██▎ | 6820/30000 [4:22:11<14:44:49, 2.29s/it] 23%|██▎ | 6821/30000 [4:22:14<14:43:03, 2.29s/it] 23%|██▎ | 6822/30000 [4:22:16<14:53:20, 2.31s/it] 23%|██▎ | 6823/30000 [4:22:18<14:49:16, 2.30s/it] 23%|██▎ | 6824/30000 [4:22:20<14:44:46, 2.29s/it] 23%|██▎ | 6825/30000 [4:22:23<14:48:43, 2.30s/it] 23%|██▎ | 6826/30000 [4:22:25<14:46:21, 2.29s/it] 23%|██▎ | 6827/30000 [4:22:27<14:43:53, 2.29s/it] 23%|██▎ | 6828/30000 [4:22:30<14:42:00, 2.28s/it] 23%|██▎ | 6829/30000 [4:22:32<14:45:45, 2.29s/it] 23%|██▎ | 6830/30000 [4:22:34<14:41:48, 2.28s/it] 23%|██▎ | 6830/30000 [4:22:34<14:41:48, 2.28s/it] 23%|██▎ | 6831/30000 [4:22:36<14:39:44, 2.28s/it] 23%|██▎ | 6832/30000 [4:22:39<14:47:33, 2.30s/it] 23%|██▎ | 6833/30000 [4:22:41<14:50:33, 2.31s/it] 23%|██▎ | 6834/30000 [4:22:43<14:47:10, 2.30s/it] 23%|██▎ | 6835/30000 [4:22:46<14:45:50, 2.29s/it] 23%|██▎ | 6836/30000 [4:22:48<14:48:47, 2.30s/it] 23%|██▎ | 6837/30000 [4:22:50<14:49:02, 2.30s/it] 23%|██▎ | 6838/30000 [4:22:53<14:42:52, 2.29s/it] 23%|██▎ | 6839/30000 [4:22:55<14:38:21, 2.28s/it] 23%|██▎ | 6840/30000 [4:22:57<14:38:56, 2.28s/it] 23%|██▎ | 6840/30000 [4:22:57<14:38:56, 2.28s/it] 23%|██▎ | 6841/30000 [4:22:59<14:38:59, 2.28s/it] 23%|██▎ | 6842/30000 [4:23:02<14:42:08, 2.29s/it] 23%|██▎ | 6843/30000 [4:23:04<14:42:09, 2.29s/it] 23%|██▎ | 6844/30000 [4:23:06<14:42:51, 2.29s/it] 23%|██▎ | 6845/30000 [4:23:09<14:44:29, 2.29s/it] 23%|██▎ | 6846/30000 [4:23:11<14:40:22, 2.28s/it] 23%|██▎ | 6847/30000 [4:23:13<14:41:14, 2.28s/it] 23%|██▎ | 6848/30000 [4:23:15<14:37:30, 2.27s/it] 23%|██▎ | 6849/30000 [4:23:18<14:47:14, 2.30s/it] 23%|██▎ | 6850/30000 [4:23:20<14:46:29, 2.30s/it] 23%|██▎ | 6850/30000 [4:23:20<14:46:29, 2.30s/it] 23%|██▎ | 6851/30000 [4:23:22<14:45:15, 2.29s/it] 23%|██▎ | 6852/30000 [4:23:25<14:46:28, 2.30s/it] 23%|██▎ | 6853/30000 [4:23:27<14:43:16, 2.29s/it] 23%|██▎ | 6854/30000 [4:23:29<14:41:02, 2.28s/it] 23%|██▎ | 6855/30000 [4:23:31<14:39:58, 2.28s/it] 23%|██▎ | 6856/30000 [4:23:34<14:43:04, 2.29s/it] 23%|██▎ | 6857/30000 [4:23:36<14:43:19, 2.29s/it] 23%|██▎ | 6858/30000 [4:23:38<14:41:25, 2.29s/it] 23%|██▎ | 6859/30000 [4:23:41<14:40:08, 2.28s/it] 23%|██▎ | 6860/30000 [4:23:43<14:44:41, 2.29s/it] 23%|██▎ | 6860/30000 [4:23:43<14:44:41, 2.29s/it] 23%|██▎ | 6861/30000 [4:23:45<14:42:14, 2.29s/it] 23%|██▎ | 6862/30000 [4:23:47<14:44:45, 2.29s/it] 23%|██▎ | 6863/30000 [4:23:50<14:44:19, 2.29s/it] 23%|██▎ | 6864/30000 [4:23:52<14:42:52, 2.29s/it] 23%|██▎ | 6865/30000 [4:23:54<14:42:18, 2.29s/it] 23%|██▎ | 6866/30000 [4:23:57<14:38:39, 2.28s/it] 23%|██▎ | 6867/30000 [4:23:59<14:47:03, 2.30s/it] 23%|██▎ | 6868/30000 [4:24:01<14:47:58, 2.30s/it] 23%|██▎ | 6869/30000 [4:24:03<14:42:14, 2.29s/it] 23%|██▎ | 6870/30000 [4:24:06<14:41:45, 2.29s/it] 23%|██▎ | 6870/30000 [4:24:06<14:41:45, 2.29s/it] 23%|██▎ | 6871/30000 [4:24:08<14:42:49, 2.29s/it] 23%|██▎ | 6872/30000 [4:24:10<14:41:22, 2.29s/it] 23%|██▎ | 6873/30000 [4:24:13<14:44:46, 2.30s/it] 23%|██▎ | 6874/30000 [4:24:14<12:09:39, 1.89s/it] 23%|██▎ | 6875/30000 [4:24:29<37:58:40, 5.91s/it] 23%|██▎ | 6876/30000 [4:24:31<31:00:29, 4.83s/it] 23%|██▎ | 6877/30000 [4:24:34<26:10:31, 4.08s/it] 23%|██▎ | 6878/30000 [4:24:36<22:46:28, 3.55s/it] 23%|██▎ | 6879/30000 [4:24:38<20:17:21, 3.16s/it] 23%|██▎ | 6880/30000 [4:24:40<18:40:33, 2.91s/it] 23%|██▎ | 6880/30000 [4:24:40<18:40:33, 2.91s/it] 23%|██▎ | 6881/30000 [4:24:43<17:31:25, 2.73s/it] 23%|██▎ | 6882/30000 [4:24:45<16:43:49, 2.61s/it] 23%|██▎ | 6883/30000 [4:24:47<16:09:57, 2.52s/it] 23%|██▎ | 6884/30000 [4:24:50<15:43:28, 2.45s/it] 23%|██▎ | 6885/30000 [4:24:52<15:30:02, 2.41s/it] 23%|██▎ | 6886/30000 [4:24:54<15:22:44, 2.40s/it] 23%|██▎ | 6887/30000 [4:24:57<15:04:25, 2.35s/it] 23%|██▎ | 6888/30000 [4:24:59<15:02:32, 2.34s/it] 23%|██▎ | 6889/30000 [4:25:01<14:58:13, 2.33s/it] 23%|██▎ | 6890/30000 [4:25:03<14:48:07, 2.31s/it] 23%|██▎ | 6890/30000 [4:25:03<14:48:07, 2.31s/it] 23%|██▎ | 6891/30000 [4:25:06<14:51:05, 2.31s/it] 23%|██▎ | 6892/30000 [4:25:08<14:47:26, 2.30s/it] 23%|██▎ | 6893/30000 [4:25:10<14:37:15, 2.28s/it] 23%|██▎ | 6894/30000 [4:25:13<14:42:42, 2.29s/it] 23%|██▎ | 6895/30000 [4:25:15<14:44:19, 2.30s/it] 23%|██▎ | 6896/30000 [4:25:17<14:44:11, 2.30s/it] 23%|██▎ | 6897/30000 [4:25:19<14:41:06, 2.29s/it] 23%|██▎ | 6898/30000 [4:25:22<14:48:33, 2.31s/it] 23%|██▎ | 6899/30000 [4:25:24<14:44:41, 2.30s/it] 23%|██▎ | 6900/30000 [4:25:26<14:44:09, 2.30s/it] 23%|██▎ | 6900/30000 [4:25:26<14:44:09, 2.30s/it] 23%|██▎ | 6901/30000 [4:25:29<14:53:18, 2.32s/it] 23%|██▎ | 6902/30000 [4:25:31<14:53:07, 2.32s/it] 23%|██▎ | 6903/30000 [4:25:33<14:47:26, 2.31s/it] 23%|██▎ | 6904/30000 [4:25:36<14:51:35, 2.32s/it] 23%|██▎ | 6905/30000 [4:25:38<14:54:47, 2.32s/it] 23%|██▎ | 6906/30000 [4:25:40<14:48:56, 2.31s/it] 23%|██▎ | 6907/30000 [4:25:43<14:45:40, 2.30s/it] 23%|██▎ | 6908/30000 [4:25:45<14:48:49, 2.31s/it] 23%|██▎ | 6909/30000 [4:25:47<14:48:12, 2.31s/it] 23%|██▎ | 6910/30000 [4:25:50<14:49:01, 2.31s/it] 23%|██▎ | 6910/30000 [4:25:50<14:49:01, 2.31s/it] 23%|██▎ | 6911/30000 [4:25:52<14:50:15, 2.31s/it] 23%|██▎ | 6912/30000 [4:25:54<14:49:16, 2.31s/it] 23%|██▎ | 6913/30000 [4:25:56<14:45:43, 2.30s/it] 23%|██▎ | 6914/30000 [4:25:59<14:49:38, 2.31s/it] 23%|██▎ | 6915/30000 [4:26:01<14:46:02, 2.30s/it] 23%|██▎ | 6916/30000 [4:26:03<14:43:21, 2.30s/it] 23%|██▎ | 6917/30000 [4:26:06<14:47:18, 2.31s/it] 23%|██▎ | 6918/30000 [4:26:08<14:48:20, 2.31s/it] 23%|██▎ | 6919/30000 [4:26:10<14:45:18, 2.30s/it] 23%|██▎ | 6920/30000 [4:26:13<14:45:18, 2.30s/it] 23%|██▎ | 6920/30000 [4:26:13<14:45:18, 2.30s/it] 23%|██▎ | 6921/30000 [4:26:15<14:49:18, 2.31s/it] 23%|██▎ | 6922/30000 [4:26:17<14:45:42, 2.30s/it] 23%|██▎ | 6923/30000 [4:26:19<14:42:24, 2.29s/it] 23%|██▎ | 6924/30000 [4:26:22<14:44:31, 2.30s/it] 23%|██▎ | 6925/30000 [4:26:24<14:49:03, 2.31s/it] 23%|██▎ | 6926/30000 [4:26:26<14:46:54, 2.31s/it] 23%|██▎ | 6927/30000 [4:26:29<14:46:24, 2.31s/it] 23%|██▎ | 6928/30000 [4:26:31<14:53:50, 2.32s/it] 23%|██▎ | 6929/30000 [4:26:33<14:44:41, 2.30s/it] 23%|██▎ | 6930/30000 [4:26:36<14:52:36, 2.32s/it] 23%|██▎ | 6930/30000 [4:26:36<14:52:36, 2.32s/it] 23%|██▎ | 6931/30000 [4:26:38<14:48:16, 2.31s/it] 23%|██▎ | 6932/30000 [4:26:40<14:41:34, 2.29s/it] 23%|██▎ | 6933/30000 [4:26:42<14:37:18, 2.28s/it] 23%|██▎ | 6934/30000 [4:26:45<14:38:32, 2.29s/it] 23%|██▎ | 6935/30000 [4:26:47<14:34:56, 2.28s/it] 23%|██▎ | 6936/30000 [4:26:49<14:37:38, 2.28s/it] 23%|██▎ | 6937/30000 [4:26:52<14:38:29, 2.29s/it] 23%|██▎ | 6938/30000 [4:26:54<14:35:39, 2.28s/it] 23%|██▎ | 6939/30000 [4:26:56<14:36:03, 2.28s/it] 23%|██▎ | 6940/30000 [4:26:59<14:48:04, 2.31s/it] 23%|██▎ | 6940/30000 [4:26:59<14:48:04, 2.31s/it] 23%|██▎ | 6941/30000 [4:27:01<14:48:49, 2.31s/it] 23%|██▎ | 6942/30000 [4:27:03<14:46:33, 2.31s/it] 23%|██▎ | 6943/30000 [4:27:05<14:49:08, 2.31s/it] 23%|██▎ | 6944/30000 [4:27:08<14:50:23, 2.32s/it] 23%|██▎ | 6945/30000 [4:27:10<14:45:20, 2.30s/it] 23%|██▎ | 6946/30000 [4:27:12<14:39:29, 2.29s/it] 23%|██▎ | 6947/30000 [4:27:15<14:45:15, 2.30s/it] 23%|██▎ | 6948/30000 [4:27:17<14:40:47, 2.29s/it] 23%|██▎ | 6949/30000 [4:27:19<14:37:47, 2.28s/it] 23%|██▎ | 6950/30000 [4:27:22<14:46:11, 2.31s/it] 23%|██▎ | 6950/30000 [4:27:22<14:46:11, 2.31s/it] 23%|██▎ | 6951/30000 [4:27:24<14:48:04, 2.31s/it] 23%|██▎ | 6952/30000 [4:27:26<14:47:12, 2.31s/it] 23%|██▎ | 6953/30000 [4:27:28<14:39:40, 2.29s/it] 23%|██▎ | 6954/30000 [4:27:31<14:46:18, 2.31s/it] 23%|██▎ | 6955/30000 [4:27:33<14:39:26, 2.29s/it] 23%|██▎ | 6956/30000 [4:27:35<14:37:15, 2.28s/it] 23%|██▎ | 6957/30000 [4:27:38<14:45:28, 2.31s/it] 23%|██▎ | 6958/30000 [4:27:40<14:47:35, 2.31s/it] 23%|██▎ | 6959/30000 [4:27:42<14:44:00, 2.30s/it] 23%|██▎ | 6960/30000 [4:27:45<14:43:23, 2.30s/it] 23%|██▎ | 6960/30000 [4:27:45<14:43:23, 2.30s/it] 23%|██▎ | 6961/30000 [4:27:47<14:48:20, 2.31s/it] 23%|██▎ | 6962/30000 [4:27:49<14:40:57, 2.29s/it] 23%|██▎ | 6963/30000 [4:27:51<14:38:53, 2.29s/it] 23%|██▎ | 6964/30000 [4:27:54<14:43:00, 2.30s/it] 23%|██▎ | 6965/30000 [4:27:56<14:45:48, 2.31s/it] 23%|██▎ | 6966/30000 [4:27:58<14:42:19, 2.30s/it] 23%|██▎ | 6967/30000 [4:28:01<14:35:31, 2.28s/it] 23%|██▎ | 6968/30000 [4:28:03<14:43:06, 2.30s/it] 23%|██▎ | 6969/30000 [4:28:05<14:34:18, 2.28s/it] 23%|██▎ | 6970/30000 [4:28:07<14:36:17, 2.28s/it] 23%|██▎ | 6970/30000 [4:28:08<14:36:17, 2.28s/it] 23%|██▎ | 6971/30000 [4:28:10<14:46:10, 2.31s/it] 23%|██▎ | 6972/30000 [4:28:12<14:42:04, 2.30s/it] 23%|██▎ | 6973/30000 [4:28:14<14:40:21, 2.29s/it] 23%|██▎ | 6974/30000 [4:28:17<14:38:28, 2.29s/it] 23%|██▎ | 6975/30000 [4:28:19<14:44:55, 2.31s/it] 23%|██▎ | 6976/30000 [4:28:21<14:44:43, 2.31s/it] 23%|██▎ | 6977/30000 [4:28:24<14:39:53, 2.29s/it] 23%|██▎ | 6978/30000 [4:28:26<14:37:12, 2.29s/it] 23%|██▎ | 6979/30000 [4:28:28<14:44:30, 2.31s/it] 23%|██▎ | 6980/30000 [4:28:31<14:45:51, 2.31s/it] 23%|██▎ | 6980/30000 [4:28:31<14:45:51, 2.31s/it] 23%|██▎ | 6981/30000 [4:28:33<14:40:33, 2.30s/it] 23%|██▎ | 6982/30000 [4:28:35<14:40:08, 2.29s/it] 23%|██▎ | 6983/30000 [4:28:37<14:46:16, 2.31s/it] 23%|██▎ | 6984/30000 [4:28:40<14:50:13, 2.32s/it] 23%|██▎ | 6985/30000 [4:28:42<14:43:39, 2.30s/it] 23%|██▎ | 6986/30000 [4:28:44<14:41:13, 2.30s/it] 23%|██▎ | 6987/30000 [4:28:47<14:40:13, 2.29s/it] 23%|██▎ | 6988/30000 [4:28:49<14:38:07, 2.29s/it] 23%|██▎ | 6989/30000 [4:28:51<14:37:25, 2.29s/it] 23%|██▎ | 6990/30000 [4:28:54<14:45:27, 2.31s/it] 23%|██▎ | 6990/30000 [4:28:54<14:45:27, 2.31s/it] 23%|██▎ | 6991/30000 [4:28:56<14:44:04, 2.31s/it] 23%|██▎ | 6992/30000 [4:28:58<14:41:00, 2.30s/it] 23%|██▎ | 6993/30000 [4:29:00<14:38:52, 2.29s/it] 23%|██▎ | 6994/30000 [4:29:03<14:45:49, 2.31s/it] 23%|██▎ | 6995/30000 [4:29:05<14:45:07, 2.31s/it] 23%|██▎ | 6996/30000 [4:29:07<14:40:32, 2.30s/it] 23%|██▎ | 6997/30000 [4:29:10<14:38:35, 2.29s/it] 23%|██▎ | 6998/30000 [4:29:12<14:42:57, 2.30s/it] 23%|██▎ | 6999/30000 [4:29:14<14:43:18, 2.30s/it] 23%|██▎ | 7000/30000 [4:29:17<14:41:03, 2.30s/it] 23%|██▎ | 7000/30000 [4:29:17<14:41:03, 2.30s/it] 23%|██▎ | 7001/30000 [4:29:19<14:36:12, 2.29s/it] 23%|██▎ | 7002/30000 [4:29:21<14:39:09, 2.29s/it] 23%|██▎ | 7003/30000 [4:29:23<14:40:31, 2.30s/it] 23%|██▎ | 7004/30000 [4:29:26<14:35:09, 2.28s/it] 23%|██▎ | 7005/30000 [4:29:28<14:31:30, 2.27s/it] 23%|██▎ | 7006/30000 [4:29:30<14:39:52, 2.30s/it] 23%|██▎ | 7007/30000 [4:29:33<14:36:07, 2.29s/it] 23%|██▎ | 7008/30000 [4:29:35<14:37:07, 2.29s/it] 23%|██▎ | 7009/30000 [4:29:37<14:39:56, 2.30s/it] 23%|██▎ | 7010/30000 [4:29:39<14:38:57, 2.29s/it] 23%|██▎ | 7010/30000 [4:29:39<14:38:57, 2.29s/it] 23%|██▎ | 7011/30000 [4:29:42<14:34:04, 2.28s/it] 23%|██▎ | 7012/30000 [4:29:44<14:32:00, 2.28s/it] 23%|██▎ | 7013/30000 [4:29:46<14:41:24, 2.30s/it] 23%|██▎ | 7014/30000 [4:29:49<14:44:51, 2.31s/it] 23%|██▎ | 7015/30000 [4:29:51<14:42:16, 2.30s/it] 23%|██▎ | 7016/30000 [4:29:53<14:38:58, 2.29s/it] 23%|██▎ | 7017/30000 [4:29:56<14:45:24, 2.31s/it] 23%|██▎ | 7018/30000 [4:29:58<14:42:55, 2.31s/it] 23%|██▎ | 7019/30000 [4:30:00<14:37:45, 2.29s/it] 23%|██▎ | 7020/30000 [4:30:02<14:42:45, 2.30s/it] 23%|██▎ | 7020/30000 [4:30:02<14:42:45, 2.30s/it] 23%|██▎ | 7021/30000 [4:30:05<14:41:34, 2.30s/it] 23%|██▎ | 7022/30000 [4:30:07<14:41:21, 2.30s/it] 23%|██▎ | 7023/30000 [4:30:09<14:41:48, 2.30s/it] 23%|██▎ | 7024/30000 [4:30:12<14:45:14, 2.31s/it] 23%|██▎ | 7025/30000 [4:30:14<14:42:19, 2.30s/it] 23%|██▎ | 7026/30000 [4:30:16<14:37:52, 2.29s/it] 23%|██▎ | 7027/30000 [4:30:19<14:40:46, 2.30s/it] 23%|██▎ | 7028/30000 [4:30:21<14:42:11, 2.30s/it] 23%|██▎ | 7029/30000 [4:30:23<14:43:14, 2.31s/it] 23%|██▎ | 7030/30000 [4:30:25<14:43:59, 2.31s/it] 23%|██▎ | 7030/30000 [4:30:26<14:43:59, 2.31s/it] 23%|██▎ | 7031/30000 [4:30:28<14:48:34, 2.32s/it] 23%|██▎ | 7032/30000 [4:30:30<14:50:47, 2.33s/it] 23%|██▎ | 7033/30000 [4:30:32<14:40:00, 2.30s/it] 23%|██▎ | 7034/30000 [4:30:35<14:40:52, 2.30s/it] 23%|██▎ | 7035/30000 [4:30:37<14:40:54, 2.30s/it] 23%|██▎ | 7036/30000 [4:30:39<14:35:43, 2.29s/it] 23%|██▎ | 7037/30000 [4:30:42<14:36:24, 2.29s/it] 23%|██▎ | 7038/30000 [4:30:44<14:42:47, 2.31s/it] 23%|██▎ | 7039/30000 [4:30:46<14:42:08, 2.31s/it] 23%|██▎ | 7040/30000 [4:30:48<14:41:03, 2.30s/it] 23%|██▎ | 7040/30000 [4:30:49<14:41:03, 2.30s/it] 23%|██▎ | 7041/30000 [4:30:51<14:48:01, 2.32s/it] 23%|██▎ | 7042/30000 [4:30:53<14:43:58, 2.31s/it] 23%|██▎ | 7043/30000 [4:30:55<14:42:37, 2.31s/it] 23%|██▎ | 7044/30000 [4:30:58<14:36:58, 2.29s/it] 23%|██▎ | 7045/30000 [4:31:00<14:40:35, 2.30s/it] 23%|██▎ | 7046/30000 [4:31:02<14:36:14, 2.29s/it] 23%|██▎ | 7047/30000 [4:31:05<14:37:20, 2.29s/it] 23%|██▎ | 7048/30000 [4:31:07<14:43:05, 2.31s/it] 23%|██▎ | 7049/30000 [4:31:09<14:45:50, 2.32s/it] 24%|██▎ | 7050/30000 [4:31:12<14:45:42, 2.32s/it] 24%|██▎ | 7050/30000 [4:31:12<14:45:42, 2.32s/it] 24%|██▎ | 7051/30000 [4:31:14<14:49:08, 2.32s/it] 24%|██▎ | 7052/30000 [4:31:16<14:47:51, 2.32s/it] 24%|██▎ | 7053/30000 [4:31:18<14:41:29, 2.30s/it] 24%|██▎ | 7054/30000 [4:31:21<14:36:49, 2.29s/it] 24%|██▎ | 7055/30000 [4:31:23<14:42:57, 2.31s/it] 24%|██▎ | 7056/30000 [4:31:25<14:43:18, 2.31s/it] 24%|██▎ | 7057/30000 [4:31:28<14:37:54, 2.30s/it] 24%|██▎ | 7058/30000 [4:31:30<14:40:59, 2.30s/it] 24%|██▎ | 7059/30000 [4:31:32<14:37:12, 2.29s/it] 24%|██▎ | 7060/30000 [4:31:35<14:30:10, 2.28s/it] 24%|██▎ | 7060/30000 [4:31:35<14:30:10, 2.28s/it] 24%|██▎ | 7061/30000 [4:31:37<14:37:14, 2.29s/it] 24%|██▎ | 7062/30000 [4:31:39<14:33:10, 2.28s/it] 24%|██▎ | 7063/30000 [4:31:41<14:30:14, 2.28s/it] 24%|██▎ | 7064/30000 [4:31:44<14:41:32, 2.31s/it] 24%|██▎ | 7065/30000 [4:31:46<14:39:16, 2.30s/it] 24%|██▎ | 7066/30000 [4:31:48<14:34:19, 2.29s/it] 24%|██▎ | 7067/30000 [4:31:51<14:39:22, 2.30s/it] 24%|██▎ | 7068/30000 [4:31:53<14:33:51, 2.29s/it] 24%|██▎ | 7069/30000 [4:31:55<14:30:53, 2.28s/it] 24%|██▎ | 7070/30000 [4:31:57<14:40:37, 2.30s/it] 24%|██▎ | 7070/30000 [4:31:58<14:40:37, 2.30s/it] 24%|██▎ | 7071/30000 [4:32:00<14:38:56, 2.30s/it] 24%|██▎ | 7072/30000 [4:32:02<14:38:23, 2.30s/it] 24%|██▎ | 7073/30000 [4:32:04<14:31:11, 2.28s/it] 24%|██▎ | 7074/30000 [4:32:07<14:37:33, 2.30s/it] 24%|██▎ | 7075/30000 [4:32:09<14:38:27, 2.30s/it] 24%|██▎ | 7076/30000 [4:32:11<14:37:19, 2.30s/it] 24%|██▎ | 7077/30000 [4:32:14<14:43:40, 2.31s/it] 24%|██▎ | 7078/30000 [4:32:16<14:40:41, 2.31s/it] 24%|██▎ | 7079/30000 [4:32:18<14:34:32, 2.29s/it] 24%|██▎ | 7080/30000 [4:32:20<14:35:38, 2.29s/it] 24%|██▎ | 7080/30000 [4:32:20<14:35:38, 2.29s/it] 24%|██▎ | 7081/30000 [4:32:23<14:37:25, 2.30s/it] 24%|██▎ | 7082/30000 [4:32:25<14:33:33, 2.29s/it] 24%|██▎ | 7083/30000 [4:32:27<14:35:26, 2.29s/it] 24%|██▎ | 7084/30000 [4:32:30<14:41:08, 2.31s/it] 24%|██▎ | 7085/30000 [4:32:32<14:43:11, 2.31s/it] 24%|██▎ | 7086/30000 [4:32:34<14:43:51, 2.31s/it] 24%|██▎ | 7087/30000 [4:32:37<14:47:14, 2.32s/it] 24%|██▎ | 7088/30000 [4:32:39<14:44:47, 2.32s/it] 24%|██▎ | 7089/30000 [4:32:41<14:38:40, 2.30s/it] 24%|██▎ | 7090/30000 [4:32:43<14:32:18, 2.28s/it] 24%|██▎ | 7090/30000 [4:32:44<14:32:18, 2.28s/it] 24%|██▎ | 7091/30000 [4:32:46<14:38:04, 2.30s/it] 24%|██▎ | 7092/30000 [4:32:48<14:34:49, 2.29s/it] 24%|██▎ | 7093/30000 [4:32:50<14:30:23, 2.28s/it] 24%|██▎ | 7094/30000 [4:32:53<14:37:17, 2.30s/it] 24%|██▎ | 7095/30000 [4:32:55<14:36:43, 2.30s/it] 24%|██▎ | 7096/30000 [4:32:57<14:36:20, 2.30s/it] 24%|██▎ | 7097/30000 [4:33:00<14:33:45, 2.29s/it] 24%|██▎ | 7098/30000 [4:33:02<14:33:05, 2.29s/it] 24%|██▎ | 7099/30000 [4:33:04<14:31:21, 2.28s/it] 24%|██▎ | 7100/30000 [4:33:06<14:28:11, 2.27s/it] +{'loss': 0.0169, 'grad_norm': 0.10724200308322906, 'learning_rate': 2.7603420969975743e-05, 'flow_matching_loss': 0.016860281769186258, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.95} +{'loss': 0.018, 'grad_norm': 0.09994521737098694, 'learning_rate': 2.7594447652983862e-05, 'flow_matching_loss': 0.018016140814870595, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.95} +{'loss': 0.0183, 'grad_norm': 0.10873758792877197, 'learning_rate': 2.7585459032550497e-05, 'flow_matching_loss': 0.01829990539699793, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.96} +{'loss': 0.0154, 'grad_norm': 0.10149592161178589, 'learning_rate': 2.7576455119597684e-05, 'flow_matching_loss': 0.015357098821550608, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.96} +{'loss': 0.0167, 'grad_norm': 0.0814293622970581, 'learning_rate': 2.7567435925066012e-05, 'flow_matching_loss': 0.016721124574542046, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.96} +{'loss': 0.0171, 'grad_norm': 0.06600968539714813, 'learning_rate': 2.7558401459914665e-05, 'flow_matching_loss': 0.01712523363530636, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.96} +{'loss': 0.0164, 'grad_norm': 0.06897532939910889, 'learning_rate': 2.7549351735121362e-05, 'flow_matching_loss': 0.01638155207037926, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.97} +{'loss': 0.0173, 'grad_norm': 0.11103256046772003, 'learning_rate': 2.7540286761682377e-05, 'flow_matching_loss': 0.017254160810261966, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.97} +{'loss': 0.018, 'grad_norm': 0.09817861020565033, 'learning_rate': 2.75312065506125e-05, 'flow_matching_loss': 0.017970383726060392, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.97} +{'loss': 0.0184, 'grad_norm': 0.09734764695167542, 'learning_rate': 2.7522111112945052e-05, 'flow_matching_loss': 0.018382926005870104, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.98} +{'loss': 0.0177, 'grad_norm': 0.10186561942100525, 'learning_rate': 2.751300045973184e-05, 'flow_matching_loss': 0.01765251439064741, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.98} +{'loss': 0.0156, 'grad_norm': 0.0832919180393219, 'learning_rate': 2.750387460204317e-05, 'flow_matching_loss': 0.015632458962500095, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.98} +{'loss': 0.0159, 'grad_norm': 0.08892107754945755, 'learning_rate': 2.7494733550967827e-05, 'flow_matching_loss': 0.015918474271893503, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.98} +{'loss': 0.0163, 'grad_norm': 0.08667997270822525, 'learning_rate': 2.748557731761304e-05, 'flow_matching_loss': 0.01633006604388356, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.99} +{'loss': 0.018, 'grad_norm': 0.07802540808916092, 'learning_rate': 2.74764059131045e-05, 'flow_matching_loss': 0.018007144890725614, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.99} +{'loss': 0.0171, 'grad_norm': 0.08695584535598755, 'learning_rate': 2.7467219348586322e-05, 'flow_matching_loss': 0.017149584274739026, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 1.99} +{'loss': 0.0168, 'grad_norm': 0.07436618953943253, 'learning_rate': 2.7458017635221056e-05, 'flow_matching_loss': 0.016799836698919535, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.0} +{'loss': 0.0179, 'grad_norm': 0.07455611228942871, 'learning_rate': 2.7448800784189652e-05, 'flow_matching_loss': 0.017902564816176893, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.0} +{'loss': 0.0178, 'grad_norm': 0.0956747904419899, 'learning_rate': 2.7439568806691448e-05, 'flow_matching_loss': 0.017821810860186815, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.0} +{'loss': 0.0152, 'grad_norm': 0.09084337949752808, 'learning_rate': 2.7430321713944167e-05, 'flow_matching_loss': 0.015173871815204621, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.0} +{'loss': 0.0154, 'grad_norm': 0.08174826949834824, 'learning_rate': 2.7421059517183906e-05, 'flow_matching_loss': 0.015359226893633605, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.01} +{'loss': 0.0167, 'grad_norm': 0.08032640069723129, 'learning_rate': 2.7411782227665097e-05, 'flow_matching_loss': 0.016740091610699892, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.01} +{'loss': 0.0187, 'grad_norm': 0.08388732373714447, 'learning_rate': 2.7402489856660528e-05, 'flow_matching_loss': 0.018670203164219856, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.01} +{'loss': 0.0154, 'grad_norm': 0.08058849722146988, 'learning_rate': 2.7393182415461307e-05, 'flow_matching_loss': 0.015359473507851362, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.02} +{'loss': 0.0173, 'grad_norm': 0.10241051018238068, 'learning_rate': 2.7383859915376853e-05, 'flow_matching_loss': 0.01730056619271636, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.02} +{'loss': 0.0169, 'grad_norm': 0.07475873827934265, 'learning_rate': 2.7374522367734875e-05, 'flow_matching_loss': 0.016866291780024766, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.02} +{'loss': 0.0166, 'grad_norm': 0.08385030180215836, 'learning_rate': 2.7365169783881378e-05, 'flow_matching_loss': 0.01663920246064663, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.03} +{'loss': 0.0148, 'grad_norm': 0.08441538363695145, 'learning_rate': 2.735580217518064e-05, 'flow_matching_loss': 0.014750133268535137, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.03} +{'loss': 0.0173, 'grad_norm': 0.08109751343727112, 'learning_rate': 2.7346419553015175e-05, 'flow_matching_loss': 0.01725252689793706, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.03} +{'loss': 0.0177, 'grad_norm': 0.10495557636022568, 'learning_rate': 2.733702192878577e-05, 'flow_matching_loss': 0.017660737317055464, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.03} +{'loss': 0.0159, 'grad_norm': 0.09454198181629181, 'learning_rate': 2.732760931391141e-05, 'flow_matching_loss': 0.01593604953959584, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.04} +{'loss': 0.0165, 'grad_norm': 0.07170224189758301, 'learning_rate': 2.731818171982932e-05, 'flow_matching_loss': 0.01654547341167927, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.04} +{'loss': 0.0155, 'grad_norm': 0.10838077962398529, 'learning_rate': 2.730873915799491e-05, 'flow_matching_loss': 0.015454737935215235, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.04} +{'loss': 0.017, 'grad_norm': 0.07772323489189148, 'learning_rate': 2.7299281639881783e-05, 'flow_matching_loss': 0.016959916707128285, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.05} +{'loss': 0.0148, 'grad_norm': 0.06683853268623352, 'learning_rate': 2.728980917698172e-05, 'flow_matching_loss': 0.014778068009763955, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.05} +{'loss': 0.0162, 'grad_norm': 0.08981888741254807, 'learning_rate': 2.7280321780804655e-05, 'flow_matching_loss': 0.01618355754762888, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.05} +{'loss': 0.0174, 'grad_norm': 0.08273127675056458, 'learning_rate': 2.7270819462878666e-05, 'flow_matching_loss': 0.017350093089044095, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.05} +{'loss': 0.016, 'grad_norm': 0.07325915992259979, 'learning_rate': 2.7261302234749965e-05, 'flow_matching_loss': 0.01599160796031356, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.06} +{'loss': 0.0152, 'grad_norm': 0.10310603678226471, 'learning_rate': 2.7251770107982888e-05, 'flow_matching_loss': 0.015203469712287188, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.06} +{'loss': 0.0161, 'grad_norm': 0.07476403564214706, 'learning_rate': 2.7242223094159856e-05, 'flow_matching_loss': 0.016120374388992786, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.06} +{'loss': 0.0165, 'grad_norm': 0.07918092608451843, 'learning_rate': 2.7232661204881408e-05, 'flow_matching_loss': 0.016491062846034764, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.07} 24%|██▎ | 7100/30000 [4:33:06<14:28:11, 2.27s/it] 24%|██▎ | 7101/30000 [4:33:09<14:37:23, 2.30s/it] 24%|██▎ | 7102/30000 [4:33:11<14:42:53, 2.31s/it] 24%|██▎ | 7103/30000 [4:33:13<14:33:10, 2.29s/it] 24%|██▎ | 7104/30000 [4:33:16<14:39:33, 2.30s/it] 24%|██▎ | 7105/30000 [4:33:18<14:40:33, 2.31s/it] 24%|██▎ | 7106/30000 [4:33:20<14:36:15, 2.30s/it] 24%|██▎ | 7107/30000 [4:33:22<14:28:18, 2.28s/it] 24%|██▎ | 7108/30000 [4:33:25<14:28:40, 2.28s/it] 24%|██▎ | 7109/30000 [4:33:27<14:31:44, 2.28s/it] 24%|██▎ | 7110/30000 [4:33:29<14:27:34, 2.27s/it] 24%|██▎ | 7110/30000 [4:33:29<14:27:34, 2.27s/it] 24%|██▎ | 7111/30000 [4:33:32<14:30:35, 2.28s/it] 24%|██▎ | 7112/30000 [4:33:34<14:34:11, 2.29s/it] 24%|██▎ | 7113/30000 [4:33:36<14:31:14, 2.28s/it] 24%|██▎ | 7114/30000 [4:33:38<14:28:49, 2.28s/it] 24%|██▎ | 7115/30000 [4:33:41<14:35:07, 2.29s/it] 24%|██▎ | 7116/30000 [4:33:43<14:32:21, 2.29s/it] 24%|██▎ | 7117/30000 [4:33:45<14:31:11, 2.28s/it] 24%|██▎ | 7118/30000 [4:33:48<14:30:42, 2.28s/it] 24%|██▎ | 7119/30000 [4:33:50<14:35:05, 2.29s/it] 24%|██▎ | 7120/30000 [4:33:52<14:40:03, 2.31s/it] 24%|██▎ | 7120/30000 [4:33:52<14:40:03, 2.31s/it] 24%|██▎ | 7121/30000 [4:33:55<14:38:37, 2.30s/it] 24%|██▎ | 7122/30000 [4:33:57<14:32:51, 2.29s/it] 24%|██▎ | 7123/30000 [4:33:59<14:36:25, 2.30s/it] 24%|██▎ | 7124/30000 [4:34:01<14:29:03, 2.28s/it] 24%|██▍ | 7125/30000 [4:34:04<14:24:05, 2.27s/it] 24%|██▍ | 7126/30000 [4:34:06<14:33:48, 2.29s/it] 24%|██▍ | 7127/30000 [4:34:08<14:34:05, 2.29s/it] 24%|██▍ | 7128/30000 [4:34:10<14:31:33, 2.29s/it] 24%|██▍ | 7129/30000 [4:34:13<14:28:37, 2.28s/it] 24%|██▍ | 7130/30000 [4:34:15<14:31:48, 2.29s/it] 24%|██▍ | 7130/30000 [4:34:15<14:31:48, 2.29s/it] 24%|██▍ | 7131/30000 [4:34:17<14:30:21, 2.28s/it] 24%|██▍ | 7132/30000 [4:34:20<14:27:04, 2.28s/it] 24%|██▍ | 7133/30000 [4:34:22<14:33:00, 2.29s/it] 24%|██▍ | 7134/30000 [4:34:24<14:35:04, 2.30s/it] 24%|██▍ | 7135/30000 [4:34:26<14:29:09, 2.28s/it] 24%|██▍ | 7136/30000 [4:34:29<14:29:30, 2.28s/it] 24%|██▍ | 7137/30000 [4:34:31<14:43:00, 2.32s/it] 24%|██▍ | 7138/30000 [4:34:33<14:41:05, 2.31s/it] 24%|██▍ | 7139/30000 [4:34:36<14:35:09, 2.30s/it] 24%|██▍ | 7140/30000 [4:34:38<14:34:26, 2.30s/it] 24%|██▍ | 7140/30000 [4:34:38<14:34:26, 2.30s/it] 24%|██▍ | 7141/30000 [4:34:40<14:34:49, 2.30s/it] 24%|██▍ | 7142/30000 [4:34:43<14:29:56, 2.28s/it] 24%|██▍ | 7143/30000 [4:34:45<14:29:41, 2.28s/it] 24%|██▍ | 7144/30000 [4:34:47<14:37:32, 2.30s/it] 24%|██▍ | 7145/30000 [4:34:50<14:43:02, 2.32s/it] 24%|██▍ | 7146/30000 [4:34:52<14:33:25, 2.29s/it] 24%|██▍ | 7147/30000 [4:34:54<14:31:06, 2.29s/it] 24%|██▍ | 7148/30000 [4:34:56<14:32:27, 2.29s/it] 24%|██▍ | 7149/30000 [4:34:59<14:30:56, 2.29s/it] 24%|██▍ | 7150/30000 [4:35:01<14:28:44, 2.28s/it] 24%|██▍ | 7150/30000 [4:35:01<14:28:44, 2.28s/it] 24%|██▍ | 7151/30000 [4:35:03<14:37:20, 2.30s/it] 24%|██▍ | 7152/30000 [4:35:06<14:42:33, 2.32s/it] 24%|██▍ | 7153/30000 [4:35:08<14:44:07, 2.32s/it] 24%|██▍ | 7154/30000 [4:35:10<14:39:16, 2.31s/it] 24%|██▍ | 7155/30000 [4:35:13<14:43:33, 2.32s/it] 24%|██▍ | 7156/30000 [4:35:15<14:44:33, 2.32s/it] 24%|██▍ | 7157/30000 [4:35:17<14:38:39, 2.31s/it] 24%|██▍ | 7158/30000 [4:35:19<14:37:49, 2.31s/it] 24%|██▍ | 7159/30000 [4:35:22<14:43:22, 2.32s/it] 24%|██▍ | 7160/30000 [4:35:24<14:45:10, 2.33s/it] 24%|██▍ | 7160/30000 [4:35:24<14:45:10, 2.33s/it] 24%|██▍ | 7161/30000 [4:35:26<14:43:48, 2.32s/it] 24%|██▍ | 7162/30000 [4:35:29<14:42:38, 2.32s/it] 24%|██▍ | 7163/30000 [4:35:31<14:37:08, 2.30s/it] 24%|██▍ | 7164/30000 [4:35:33<14:32:50, 2.29s/it] 24%|██▍ | 7165/30000 [4:35:36<14:27:20, 2.28s/it] 24%|██▍ | 7166/30000 [4:35:38<14:37:54, 2.31s/it] 24%|██▍ | 7167/30000 [4:35:40<14:40:03, 2.31s/it] 24%|██▍ | 7168/30000 [4:35:43<14:34:10, 2.30s/it] 24%|██▍ | 7169/30000 [4:35:45<14:35:12, 2.30s/it] 24%|██▍ | 7170/30000 [4:35:47<14:34:16, 2.30s/it] 24%|██▍ | 7170/30000 [4:35:47<14:34:16, 2.30s/it] 24%|██▍ | 7171/30000 [4:35:49<14:32:32, 2.29s/it] 24%|██▍ | 7172/30000 [4:35:52<14:34:48, 2.30s/it] 24%|██▍ | 7173/30000 [4:35:54<14:36:21, 2.30s/it] 24%|██▍ | 7174/30000 [4:35:56<14:37:53, 2.31s/it] 24%|██▍ | 7175/30000 [4:35:59<14:38:11, 2.31s/it] 24%|██▍ | 7176/30000 [4:36:01<14:42:33, 2.32s/it] 24%|██▍ | 7177/30000 [4:36:03<14:46:12, 2.33s/it] 24%|██▍ | 7178/30000 [4:36:06<14:45:57, 2.33s/it] 24%|██▍ | 7179/30000 [4:36:08<14:36:25, 2.30s/it] 24%|██▍ | 7180/30000 [4:36:10<14:34:24, 2.30s/it] 24%|██▍ | 7180/30000 [4:36:10<14:34:24, 2.30s/it] 24%|██▍ | 7181/30000 [4:36:12<14:32:18, 2.29s/it] 24%|██▍ | 7182/30000 [4:36:15<14:28:02, 2.28s/it] 24%|██▍ | 7183/30000 [4:36:17<14:26:21, 2.28s/it] 24%|██▍ | 7184/30000 [4:36:19<14:33:54, 2.30s/it] 24%|██▍ | 7185/30000 [4:36:22<14:27:06, 2.28s/it] 24%|██▍ | 7186/30000 [4:36:24<14:29:54, 2.29s/it] 24%|██▍ | 7187/30000 [4:36:26<14:36:33, 2.31s/it] 24%|██▍ | 7188/30000 [4:36:28<14:30:44, 2.29s/it] 24%|██▍ | 7189/30000 [4:36:31<14:30:35, 2.29s/it] 24%|██▍ | 7190/30000 [4:36:33<14:33:58, 2.30s/it] 24%|██▍ | 7190/30000 [4:36:33<14:33:58, 2.30s/it] 24%|██▍ | 7191/30000 [4:36:35<14:38:42, 2.31s/it] 24%|██▍ | 7192/30000 [4:36:38<14:37:53, 2.31s/it] 24%|██▍ | 7193/30000 [4:36:40<14:37:50, 2.31s/it] 24%|██▍ | 7194/30000 [4:36:42<14:40:40, 2.32s/it] 24%|██▍ | 7195/30000 [4:36:45<14:37:13, 2.31s/it] 24%|██▍ | 7196/30000 [4:36:47<14:29:37, 2.29s/it] 24%|██▍ | 7197/30000 [4:36:49<14:31:22, 2.29s/it] 24%|██▍ | 7198/30000 [4:36:52<14:35:22, 2.30s/it] 24%|██▍ | 7199/30000 [4:36:54<14:30:29, 2.29s/it] 24%|██▍ | 7200/30000 [4:36:56<14:30:10, 2.29s/it] 24%|██▍ | 7200/30000 [4:36:56<14:30:10, 2.29s/it] 24%|██▍ | 7201/30000 [4:36:58<14:39:34, 2.31s/it] 24%|██▍ | 7202/30000 [4:37:01<14:36:00, 2.31s/it] 24%|██▍ | 7203/30000 [4:37:03<14:32:59, 2.30s/it] 24%|██▍ | 7204/30000 [4:37:05<14:37:16, 2.31s/it] 24%|██▍ | 7205/30000 [4:37:08<14:34:05, 2.30s/it] 24%|██▍ | 7206/30000 [4:37:10<14:34:40, 2.30s/it] 24%|██▍ | 7207/30000 [4:37:12<14:31:48, 2.29s/it] 24%|██▍ | 7208/30000 [4:37:15<14:39:24, 2.32s/it] 24%|██▍ | 7209/30000 [4:37:17<14:39:33, 2.32s/it] 24%|██▍ | 7210/30000 [4:37:19<14:32:08, 2.30s/it] 24%|██▍ | 7210/30000 [4:37:19<14:32:08, 2.30s/it] 24%|██▍ | 7211/30000 [4:37:22<14:36:10, 2.31s/it] 24%|██▍ | 7212/30000 [4:37:24<14:34:04, 2.30s/it] 24%|██▍ | 7213/30000 [4:37:26<14:30:56, 2.29s/it] 24%|██▍ | 7214/30000 [4:37:28<14:26:33, 2.28s/it] 24%|██▍ | 7215/30000 [4:37:31<14:30:46, 2.29s/it] 24%|██▍ | 7216/30000 [4:37:33<14:25:22, 2.28s/it] 24%|██▍ | 7217/30000 [4:37:35<14:25:14, 2.28s/it] 24%|██▍ | 7218/30000 [4:37:38<14:32:41, 2.30s/it] 24%|██▍ | 7219/30000 [4:37:40<14:35:46, 2.31s/it] 24%|██▍ | 7220/30000 [4:37:42<14:32:45, 2.30s/it] 24%|██▍ | 7220/30000 [4:37:42<14:32:45, 2.30s/it] 24%|██▍ | 7221/30000 [4:37:44<14:39:21, 2.32s/it] 24%|██▍ | 7222/30000 [4:37:47<14:40:30, 2.32s/it] 24%|██▍ | 7223/30000 [4:37:49<14:36:29, 2.31s/it] 24%|██▍ | 7224/30000 [4:37:51<14:34:30, 2.30s/it] 24%|██▍ | 7225/30000 [4:37:54<14:38:02, 2.31s/it] 24%|██▍ | 7226/30000 [4:37:56<14:29:21, 2.29s/it] 24%|██▍ | 7227/30000 [4:37:58<14:24:00, 2.28s/it] 24%|██▍ | 7228/30000 [4:38:01<14:34:18, 2.30s/it] 24%|██▍ | 7229/30000 [4:38:03<14:36:59, 2.31s/it] 24%|██▍ | 7230/30000 [4:38:05<14:34:57, 2.31s/it] 24%|██▍ | 7230/30000 [4:38:05<14:34:57, 2.31s/it] 24%|██▍ | 7231/30000 [4:38:08<14:43:32, 2.33s/it] 24%|██▍ | 7232/30000 [4:38:10<14:34:47, 2.31s/it] 24%|██▍ | 7233/30000 [4:38:12<14:35:16, 2.31s/it] 24%|██▍ | 7234/30000 [4:38:14<14:34:02, 2.30s/it] 24%|██▍ | 7235/30000 [4:38:17<14:32:25, 2.30s/it] 24%|██▍ | 7236/30000 [4:38:19<14:26:42, 2.28s/it] 24%|██▍ | 7237/30000 [4:38:21<14:21:23, 2.27s/it] 24%|██▍ | 7238/30000 [4:38:24<14:29:59, 2.29s/it] 24%|██▍ | 7239/30000 [4:38:26<14:28:58, 2.29s/it] 24%|██▍ | 7240/30000 [4:38:28<14:31:41, 2.30s/it] 24%|██▍ | 7240/30000 [4:38:28<14:31:41, 2.30s/it] 24%|██▍ | 7241/30000 [4:38:30<14:37:04, 2.31s/it] 24%|██▍ | 7242/30000 [4:38:33<14:34:29, 2.31s/it] 24%|██▍ | 7243/30000 [4:38:35<14:25:09, 2.28s/it] 24%|██▍ | 7244/30000 [4:38:37<14:33:47, 2.30s/it] 24%|██▍ | 7245/30000 [4:38:40<14:31:27, 2.30s/it] 24%|██▍ | 7246/30000 [4:38:42<14:30:29, 2.30s/it] 24%|██▍ | 7247/30000 [4:38:44<14:29:34, 2.29s/it] 24%|██▍ | 7248/30000 [4:38:47<14:37:45, 2.31s/it] 24%|██▍ | 7249/30000 [4:38:49<14:35:37, 2.31s/it] 24%|██▍ | 7250/30000 [4:38:51<14:31:50, 2.30s/it] 24%|██▍ | 7250/30000 [4:38:51<14:31:50, 2.30s/it] 24%|██▍ | 7251/30000 [4:38:54<14:38:19, 2.32s/it] 24%|██▍ | 7252/30000 [4:38:56<14:35:41, 2.31s/it] 24%|██▍ | 7253/30000 [4:38:58<14:31:26, 2.30s/it] 24%|██▍ | 7254/30000 [4:39:00<14:37:15, 2.31s/it] 24%|██▍ | 7255/30000 [4:39:03<14:38:23, 2.32s/it] 24%|██▍ | 7256/30000 [4:39:05<14:29:52, 2.29s/it] 24%|██▍ | 7257/30000 [4:39:07<14:39:05, 2.32s/it] 24%|██▍ | 7258/30000 [4:39:10<14:28:20, 2.29s/it] 24%|██▍ | 7259/30000 [4:39:12<14:27:47, 2.29s/it] 24%|██▍ | 7260/30000 [4:39:14<14:28:57, 2.29s/it] 24%|██▍ | 7260/30000 [4:39:14<14:28:57, 2.29s/it] 24%|██▍ | 7261/30000 [4:39:16<14:27:57, 2.29s/it] 24%|██▍ | 7262/30000 [4:39:19<14:28:20, 2.29s/it] 24%|██▍ | 7263/30000 [4:39:21<14:26:58, 2.29s/it] 24%|██▍ | 7264/30000 [4:39:23<14:35:26, 2.31s/it] 24%|██▍ | 7265/30000 [4:39:26<14:33:07, 2.30s/it] 24%|██▍ | 7266/30000 [4:39:28<14:32:20, 2.30s/it] 24%|██▍ | 7267/30000 [4:39:30<14:34:21, 2.31s/it] 24%|██▍ | 7268/30000 [4:39:33<14:30:20, 2.30s/it] 24%|██▍ | 7269/30000 [4:39:35<14:25:50, 2.29s/it] 24%|██▍ | 7270/30000 [4:39:37<14:33:02, 2.30s/it] 24%|██▍ | 7270/30000 [4:39:37<14:33:02, 2.30s/it] 24%|██▍ | 7271/30000 [4:39:39<14:32:52, 2.30s/it] 24%|██▍ | 7272/30000 [4:39:42<14:29:07, 2.29s/it] 24%|██▍ | 7273/30000 [4:39:44<14:30:35, 2.30s/it] 24%|██▍ | 7274/30000 [4:39:46<14:36:01, 2.31s/it] 24%|██▍ | 7275/30000 [4:39:49<14:42:22, 2.33s/it] 24%|██▍ | 7276/30000 [4:39:51<14:38:25, 2.32s/it] 24%|██▍ | 7277/30000 [4:39:53<14:34:39, 2.31s/it] 24%|██▍ | 7278/30000 [4:39:56<14:43:00, 2.33s/it] 24%|██▍ | 7279/30000 [4:39:58<14:36:07, 2.31s/it] 24%|██▍ | 7280/30000 [4:40:00<14:31:59, 2.30s/it] 24%|██▍ | 7280/30000 [4:40:00<14:31:59, 2.30s/it] 24%|██▍ | 7281/30000 [4:40:03<14:44:53, 2.34s/it] 24%|██▍ | 7282/30000 [4:40:05<14:38:43, 2.32s/it] 24%|██▍ | 7283/30000 [4:40:07<14:34:03, 2.31s/it] 24%|██▍ | 7284/30000 [4:40:10<14:35:45, 2.31s/it] 24%|██▍ | 7285/30000 [4:40:12<14:34:53, 2.31s/it] 24%|██▍ | 7286/30000 [4:40:14<14:34:47, 2.31s/it] 24%|██▍ | 7287/30000 [4:40:17<14:31:34, 2.30s/it] 24%|██▍ | 7288/30000 [4:40:19<14:37:15, 2.32s/it] 24%|██▍ | 7289/30000 [4:40:21<14:36:12, 2.31s/it] 24%|██▍ | 7290/30000 [4:40:23<14:30:18, 2.30s/it] 24%|██▍ | 7290/30000 [4:40:23<14:30:18, 2.30s/it] 24%|██▍ | 7291/30000 [4:40:26<14:37:01, 2.32s/it] 24%|██▍ | 7292/30000 [4:40:28<14:37:36, 2.32s/it] 24%|██▍ | 7293/30000 [4:40:30<14:28:41, 2.30s/it] 24%|██▍ | 7294/30000 [4:40:33<14:23:10, 2.28s/it] 24%|██▍ | 7295/30000 [4:40:35<14:29:17, 2.30s/it] 24%|██▍ | 7296/30000 [4:40:37<14:23:21, 2.28s/it] 24%|██▍ | 7297/30000 [4:40:39<14:25:18, 2.29s/it] 24%|██▍ | 7298/30000 [4:40:42<14:34:07, 2.31s/it] 24%|██▍ | 7299/30000 [4:40:44<14:31:28, 2.30s/it] 24%|██▍ | 7300/30000 [4:40:46<14:24:30, 2.29s/it] 24%|██▍ | 7300/30000 [4:40:46<14:24:30, 2.29s/it] 24%|██▍ | 7301/30000 [4:40:49<14:24:35, 2.29s/it] 24%|██▍ | 7302/30000 [4:40:51<14:24:37, 2.29s/it] 24%|██▍ | 7303/30000 [4:40:53<14:25:45, 2.29s/it] 24%|██▍ | 7304/30000 [4:40:56<14:25:56, 2.29s/it] 24%|██▍ | 7305/30000 [4:40:58<14:31:48, 2.30s/it] 24%|██▍ | 7306/30000 [4:41:00<14:34:19, 2.31s/it] 24%|██▍ | 7307/30000 [4:41:02<14:26:32, 2.29s/it] 24%|██▍ | 7308/30000 [4:41:05<14:22:21, 2.28s/it] 24%|██▍ | 7309/30000 [4:41:07<14:29:33, 2.30s/it] 24%|██▍ | 7310/30000 [4:41:09<14:33:05, 2.31s/it] 24%|██▍ | 7310/30000 [4:41:09<14:33:05, 2.31s/it] 24%|██▍ | 7311/30000 [4:41:12<14:29:43, 2.30s/it] 24%|██▍ | 7312/30000 [4:41:14<14:32:43, 2.31s/it] 24%|██▍ | 7313/30000 [4:41:16<14:38:36, 2.32s/it] 24%|██▍ | 7314/30000 [4:41:19<14:27:02, 2.29s/it] 24%|██▍ | 7315/30000 [4:41:21<14:25:08, 2.29s/it] 24%|██▍ | 7316/30000 [4:41:23<14:32:10, 2.31s/it] 24%|██▍ | 7317/30000 [4:41:25<14:29:36, 2.30s/it] 24%|██▍ | 7318/30000 [4:41:28<14:24:50, 2.29s/it] 24%|██▍ | 7319/30000 [4:41:30<14:25:27, 2.29s/it] 24%|██▍ | 7320/30000 [4:41:32<14:30:31, 2.30s/it] 24%|██▍ | 7320/30000 [4:41:32<14:30:31, 2.30s/it] 24%|██▍ | 7321/30000 [4:41:35<14:29:23, 2.30s/it] 24%|██▍ | 7322/30000 [4:41:37<14:24:43, 2.29s/it] 24%|██▍ | 7323/30000 [4:41:39<14:30:27, 2.30s/it] 24%|██▍ | 7324/30000 [4:41:42<14:27:36, 2.30s/it] 24%|██▍ | 7325/30000 [4:41:44<14:24:11, 2.29s/it] 24%|██▍ | 7326/30000 [4:41:46<14:25:02, 2.29s/it] 24%|██▍ | 7327/30000 [4:41:48<14:36:52, 2.32s/it] 24%|██▍ | 7328/30000 [4:41:51<14:39:28, 2.33s/it] 24%|██▍ | 7329/30000 [4:41:53<14:32:53, 2.31s/it] 24%|██▍ | 7330/30000 [4:41:55<14:28:20, 2.30s/it] 24%|██▍ | 7330/30000 [4:41:55<14:28:20, 2.30s/it] 24%|██▍ | 7331/30000 [4:41:58<14:34:05, 2.31s/it] 24%|██▍ | 7332/30000 [4:42:00<14:32:51, 2.31s/it] 24%|██▍ | 7333/30000 [4:42:02<14:27:57, 2.30s/it] 24%|██▍ | 7334/30000 [4:42:05<14:25:25, 2.29s/it] 24%|██▍ | 7335/30000 [4:42:07<14:29:22, 2.30s/it] 24%|██▍ | 7336/30000 [4:42:09<14:29:11, 2.30s/it] 24%|██▍ | 7337/30000 [4:42:11<14:29:15, 2.30s/it] 24%|██▍ | 7338/30000 [4:42:14<14:26:09, 2.29s/it] 24%|██▍ | 7339/30000 [4:42:16<14:30:34, 2.31s/it] 24%|██▍ | 7340/30000 [4:42:18<14:26:45, 2.30s/it] 24%|██▍ | 7340/30000 [4:42:18<14:26:45, 2.30s/it] 24%|██▍ | 7341/30000 [4:42:21<14:23:12, 2.29s/it] 24%|██▍ | 7342/30000 [4:42:23<14:26:16, 2.29s/it] 24%|██▍ | 7343/30000 [4:42:25<14:25:12, 2.29s/it] 24%|██▍ | 7344/30000 [4:42:28<14:25:01, 2.29s/it] 24%|██▍ | 7345/30000 [4:42:30<14:21:24, 2.28s/it] 24%|██▍ | 7346/30000 [4:42:32<14:25:15, 2.29s/it] 24%|██▍ | 7347/30000 [4:42:34<14:22:05, 2.28s/it] 24%|██▍ | 7348/30000 [4:42:37<14:18:53, 2.27s/it] 24%|██▍ | 7349/30000 [4:42:39<14:16:47, 2.27s/it] 24%|██▍ | 7350/30000 [4:42:41<14:25:33, 2.29s/it] 24%|██▍ | 7350/30000 [4:42:41<14:25:33, 2.29s/it] 25%|██▍ | 7351/30000 [4:42:44<14:31:03, 2.31s/it] 25%|██▍ | 7352/30000 [4:42:46<14:26:13, 2.29s/it] 25%|██▍ | 7353/30000 [4:42:48<14:25:50, 2.29s/it] 25%|██▍ | 7354/30000 [4:42:50<14:25:54, 2.29s/it] 25%|██▍ | 7355/30000 [4:42:53<14:20:36, 2.28s/it] 25%|██▍ | 7356/30000 [4:42:55<14:20:17, 2.28s/it] 25%|██▍ | 7357/30000 [4:42:57<14:21:25, 2.28s/it] 25%|██▍ | 7358/30000 [4:43:00<14:26:48, 2.30s/it] 25%|██▍ | 7359/30000 [4:43:02<14:28:05, 2.30s/it] 25%|██▍ | 7360/30000 [4:43:04<14:24:04, 2.29s/it] 25%|██▍ | 7360/30000 [4:43:04<14:24:04, 2.29s/it] 25%|██▍ | 7361/30000 [4:43:06<14:28:10, 2.30s/it] 25%|██▍ | 7362/30000 [4:43:09<14:34:26, 2.32s/it] 25%|██▍ | 7363/30000 [4:43:11<14:31:24, 2.31s/it] 25%|██▍ | 7364/30000 [4:43:13<14:29:29, 2.30s/it] 25%|██▍ | 7365/30000 [4:43:16<14:22:45, 2.29s/it] 25%|██▍ | 7366/30000 [4:43:18<14:27:26, 2.30s/it] 25%|██▍ | 7367/30000 [4:43:20<14:29:45, 2.31s/it] 25%|██▍ | 7368/30000 [4:43:23<14:26:12, 2.30s/it] 25%|██▍ | 7369/30000 [4:43:25<14:26:57, 2.30s/it] 25%|██▍ | 7370/30000 [4:43:27<14:32:30, 2.31s/it] 25%|██▍ | 7370/30000 [4:43:27<14:32:30, 2.31s/it] 25%|██▍ | 7371/30000 [4:43:30<14:34:08, 2.32s/it] 25%|██▍ | 7372/30000 [4:43:32<14:31:49, 2.31s/it] 25%|██▍ | 7373/30000 [4:43:34<14:26:57, 2.30s/it] 25%|██▍ | 7374/30000 [4:43:36<14:33:49, 2.32s/it] 25%|██▍ | 7375/30000 [4:43:39<14:33:51, 2.32s/it] 25%|██▍ | 7376/30000 [4:43:41<14:29:03, 2.30s/it] 25%|██▍ | 7377/30000 [4:43:43<14:24:26, 2.29s/it] 25%|██▍ | 7378/30000 [4:43:46<14:30:47, 2.31s/it] 25%|██▍ | 7379/30000 [4:43:48<14:33:23, 2.32s/it] 25%|██▍ | 7380/30000 [4:43:50<14:26:01, 2.30s/it] 25%|██▍ | 7380/30000 [4:43:50<14:26:01, 2.30s/it] 25%|██▍ | 7381/30000 [4:43:53<14:26:07, 2.30s/it] 25%|██▍ | 7382/30000 [4:43:55<14:34:34, 2.32s/it] 25%|██▍ | 7383/30000 [4:43:57<14:29:05, 2.31s/it] 25%|██▍ | 7384/30000 [4:44:00<14:29:59, 2.31s/it] 25%|██▍ | 7385/30000 [4:44:02<14:32:41, 2.32s/it] 25%|██▍ | 7386/30000 [4:44:04<14:33:03, 2.32s/it] 25%|██▍ | 7387/30000 [4:44:06<14:26:58, 2.30s/it] 25%|██▍ | 7388/30000 [4:44:09<14:23:46, 2.29s/it] 25%|██▍ | 7389/30000 [4:44:11<14:26:29, 2.30s/it] 25%|██▍ | 7390/30000 [4:44:13<14:26:54, 2.30s/it] 25%|██▍ | 7390/30000 [4:44:13<14:26:54, 2.30s/it] 25%|██▍ | 7391/30000 [4:44:16<14:30:17, 2.31s/it] 25%|██▍ | 7392/30000 [4:44:18<14:25:09, 2.30s/it] 25%|██▍ | 7393/30000 [4:44:20<14:35:16, 2.32s/it] 25%|██▍ | 7394/30000 [4:44:23<14:31:18, 2.31s/it] 25%|██▍ | 7395/30000 [4:44:25<14:27:30, 2.30s/it] 25%|██▍ | 7396/30000 [4:44:27<14:24:42, 2.30s/it] 25%|██▍ | 7397/30000 [4:44:29<14:28:03, 2.30s/it] 25%|██▍ | 7398/30000 [4:44:32<14:33:38, 2.32s/it] 25%|██▍ | 7399/30000 [4:44:34<14:31:55, 2.31s/it] 25%|██▍ | 7400/30000 [4:44:36<14:29:15, 2.31s/it] 25%|██▍ | 7400/30000 [4:44:36<14:29:15, 2.31s/it] 25%|██▍ | 7401/30000 [4:44:39<14:37:29, 2.33s/it] 25%|██▍ | 7402/30000 [4:44:41<14:31:43, 2.31s/it] 25%|██▍ | 7403/30000 [4:44:43<14:23:15, 2.29s/it] 25%|██▍ | 7404/30000 [4:44:46<14:21:44, 2.29s/it] 25%|██▍ | 7405/30000 [4:44:48<14:27:40, 2.30s/it] 25%|██▍ | 7406/30000 [4:44:50<14:20:59, 2.29s/it] 25%|██▍ | 7407/30000 [4:44:53<14:22:44, 2.29s/it] 25%|██▍ | 7408/30000 [4:44:55<14:26:25, 2.30s/it] 25%|██▍ | 7409/30000 [4:44:57<14:29:03, 2.31s/it] 25%|██▍ | 7410/30000 [4:44:59<14:23:28, 2.29s/it] 25%|██▍ | 7410/30000 [4:44:59<14:23:28, 2.29s/it] 25%|██▍ | 7411/30000 [4:45:02<14:19:24, 2.28s/it] 25%|██▍ | 7412/30000 [4:45:04<14:24:07, 2.30s/it] 25%|██▍ | 7413/30000 [4:45:06<14:26:56, 2.30s/it] 25%|██▍ | 7414/30000 [4:45:09<14:24:59, 2.30s/it] 25%|██▍ | 7415/30000 [4:45:11<14:24:00, 2.30s/it] 25%|██▍ | 7416/30000 [4:45:13<14:27:29, 2.30s/it] 25%|██▍ | 7417/30000 [4:45:16<14:27:10, 2.30s/it] 25%|██▍ | 7418/30000 [4:45:18<14:22:54, 2.29s/it] 25%|██▍ | 7419/30000 [4:45:20<14:23:00, 2.29s/it] 25%|██▍ | 7420/30000 [4:45:22<14:28:57, 2.31s/it] 25%|██▍ | 7420/30000 [4:45:22<14:28:57, 2.31s/it] 25%|██▍ | 7421/30000 [4:45:25<14:25:04, 2.30s/it] 25%|██▍ | 7422/30000 [4:45:27<14:25:17, 2.30s/it] 25%|██▍ | 7423/30000 [4:45:29<14:27:37, 2.31s/it] 25%|██▍ | 7424/30000 [4:45:32<14:25:02, 2.30s/it] 25%|██▍ | 7425/30000 [4:45:34<14:23:13, 2.29s/it] 25%|██▍ | 7426/30000 [4:45:36<14:20:57, 2.29s/it] 25%|██▍ | 7427/30000 [4:45:38<14:24:14, 2.30s/it] 25%|██▍ | 7428/30000 [4:45:41<14:25:18, 2.30s/it] 25%|██▍ | 7429/30000 [4:45:43<14:17:13, 2.28s/it] 25%|██▍ | 7430/30000 [4:45:45<14:18:56, 2.28s/it] 25%|██▍ | 7430/30000 [4:45:45<14:18:56, 2.28s/it] 25%|██▍ | 7431/30000 [4:45:48<14:27:34, 2.31s/it] 25%|██▍ | 7432/30000 [4:45:50<14:33:35, 2.32s/it] 25%|██▍ | 7433/30000 [4:45:52<14:31:48, 2.32s/it] 25%|██▍ | 7434/30000 [4:45:55<14:27:06, 2.31s/it] 25%|██▍ | 7435/30000 [4:45:57<14:32:12, 2.32s/it] 25%|██▍ | 7436/30000 [4:45:59<14:31:37, 2.32s/it] 25%|██▍ | 7437/30000 [4:46:02<14:25:33, 2.30s/it] 25%|██▍ | 7438/30000 [4:46:04<14:24:03, 2.30s/it] 25%|██▍ | 7439/30000 [4:46:06<14:29:04, 2.31s/it] 25%|██▍ | 7440/30000 [4:46:08<14:24:37, 2.30s/it] 25%|██▍ | 7440/30000 [4:46:08<14:24:37, 2.30s/it] 25%|██▍ | 7441/30000 [4:46:11<14:22:07, 2.29s/it] 25%|██▍ | 7442/30000 [4:46:13<14:19:09, 2.29s/it] 25%|██▍ | 7443/30000 [4:46:15<14:20:35, 2.29s/it] 25%|██▍ | 7444/30000 [4:46:18<14:17:07, 2.28s/it] 25%|██▍ | 7445/30000 [4:46:20<14:16:52, 2.28s/it] 25%|██▍ | 7446/30000 [4:46:22<14:20:14, 2.29s/it] 25%|██▍ | 7447/30000 [4:46:24<14:21:46, 2.29s/it] 25%|██▍ | 7448/30000 [4:46:27<14:17:53, 2.28s/it] 25%|██▍ | 7449/30000 [4:46:29<14:21:14, 2.29s/it] 25%|██▍ | 7450/30000 [4:46:31<14:22:41, 2.30s/it] 25%|██▍ | 7450/30000 [4:46:31<14:22:41, 2.30s/it] 25%|██▍ | 7451/30000 [4:46:34<14:20:47, 2.29s/it] 25%|██▍ | 7452/30000 [4:46:36<14:17:31, 2.28s/it] 25%|██▍ | 7453/30000 [4:46:38<14:14:43, 2.27s/it] 25%|██▍ | 7454/30000 [4:46:40<14:20:44, 2.29s/it] 25%|██▍ | 7455/30000 [4:46:43<14:21:19, 2.29s/it] 25%|██▍ | 7456/30000 [4:46:45<14:24:26, 2.30s/it] 25%|██▍ | 7457/30000 [4:46:47<14:25:16, 2.30s/it] 25%|██▍ | 7458/30000 [4:46:50<14:24:09, 2.30s/it] 25%|██▍ | 7459/30000 [4:46:52<14:19:41, 2.29s/it] 25%|██▍ | 7460/30000 [4:46:54<14:19:36, 2.29s/it] 25%|██▍ | 7460/30000 [4:46:54<14:19:36, 2.29s/it] 25%|██▍ | 7461/30000 [4:46:57<14:26:34, 2.31s/it] 25%|██▍ | 7462/30000 [4:46:59<14:33:05, 2.32s/it] 25%|██▍ | 7463/30000 [4:47:01<14:24:33, 2.30s/it] 25%|██▍ | 7464/30000 [4:47:04<14:27:34, 2.31s/it] 25%|██▍ | 7465/30000 [4:47:06<14:35:01, 2.33s/it] 25%|██▍ | 7466/30000 [4:47:08<14:33:09, 2.32s/it] 25%|██▍ | 7467/30000 [4:47:10<14:25:09, 2.30s/it] 25%|██▍ | 7468/30000 [4:47:13<14:25:38, 2.31s/it] 25%|██▍ | 7469/30000 [4:47:15<14:19:43, 2.29s/it] 25%|██▍ | 7470/30000 [4:47:17<14:16:12, 2.28s/it] 25%|██▍ | 7470/30000 [4:47:17<14:16:12, 2.28s/it] 25%|██▍ | 7471/30000 [4:47:20<14:19:09, 2.29s/it] 25%|██▍ | 7472/30000 [4:47:22<14:27:25, 2.31s/it] 25%|██▍ | 7473/30000 [4:47:24<14:30:48, 2.32s/it] 25%|██▍ | 7474/30000 [4:47:27<14:27:05, 2.31s/it] 25%|██▍ | 7475/30000 [4:47:29<14:24:17, 2.30s/it] 25%|██▍ | 7476/30000 [4:47:31<14:25:51, 2.31s/it] 25%|██▍ | 7477/30000 [4:47:33<14:21:38, 2.30s/it] 25%|██▍ | 7478/30000 [4:47:36<14:24:11, 2.30s/it] 25%|██▍ | 7479/30000 [4:47:38<14:29:41, 2.32s/it] 25%|██▍ | 7480/30000 [4:47:40<14:29:21, 2.32s/it] 25%|██▍ | 7480/30000 [4:47:40<14:29:21, 2.32s/it] 25%|██▍ | 7481/30000 [4:47:43<14:26:45, 2.31s/it] 25%|██▍ | 7482/30000 [4:47:45<14:24:20, 2.30s/it] 25%|██▍ | 7483/30000 [4:47:47<14:28:24, 2.31s/it] 25%|██▍ | 7484/30000 [4:47:50<14:21:58, 2.30s/it] 25%|██▍ | 7485/30000 [4:47:52<14:21:25, 2.30s/it] 25%|██▍ | 7486/30000 [4:47:54<14:30:05, 2.32s/it] 25%|██▍ | 7487/30000 [4:47:57<14:29:40, 2.32s/it] 25%|██▍ | 7488/30000 [4:47:59<14:27:04, 2.31s/it] 25%|██▍ | 7489/30000 [4:48:01<14:20:34, 2.29s/it] 25%|██▍ | 7490/30000 [4:48:03<14:24:34, 2.30s/it] 25%|██▍ | 7490/30000 [4:48:04<14:24:34, 2.30s/it] 25%|██▍ | 7491/30000 [4:48:06<14:25:32, 2.31s/it] 25%|██▍ | 7492/30000 [4:48:08<14:18:19, 2.29s/it] 25%|██▍ | 7493/30000 [4:48:10<14:24:18, 2.30s/it] 25%|██▍ | 7494/30000 [4:48:13<14:27:34, 2.31s/it] 25%|██▍ | 7495/30000 [4:48:15<14:22:14, 2.30s/it] 25%|██▍ | 7496/30000 [4:48:17<14:21:57, 2.30s/it] 25%|██▍ | 7497/30000 [4:48:20<14:27:32, 2.31s/it] 25%|██▍ | 7498/30000 [4:48:22<14:19:10, 2.29s/it] 25%|██▍ | 7499/30000 [4:48:24<14:16:10, 2.28s/it] 25%|██▌ | 7500/30000 [4:48:26<14:24:31, 2.31s/it] 25%|██▌ | 7500/30000 [4:48:27<14:24:31, 2.31s/it] 25%|██▌ | 7501/30000 [4:48:29<14:23:09, 2.30s/it] 25%|██▌ | 7502/30000 [4:48:31<14:22:09, 2.30s/it] 25%|██▌ | 7503/30000 [4:48:33<14:25:20, 2.31s/it] 25%|██▌ | 7504/30000 [4:48:36<14:25:49, 2.31s/it] 25%|██▌ | 7505/30000 [4:48:38<14:22:05, 2.30s/it] 25%|██▌ | 7506/30000 [4:48:40<14:18:28, 2.29s/it] 25%|██▌ | 7507/30000 [4:48:43<14:21:15, 2.30s/it] 25%|██▌ | 7508/30000 [4:48:45<14:24:01, 2.30s/it] 25%|██▌ | 7509/30000 [4:48:47<14:19:24, 2.29s/it] 25%|██▌ | 7510/30000 [4:48:49<14:24:15, 2.31s/it] +{'loss': 0.0151, 'grad_norm': 0.09591260552406311, 'learning_rate': 2.7223084451766123e-05, 'flow_matching_loss': 0.015142364241182805, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.07} +{'loss': 0.017, 'grad_norm': 0.10381370037794113, 'learning_rate': 2.721349284645067e-05, 'flow_matching_loss': 0.016970221512019636, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.07} +{'loss': 0.0171, 'grad_norm': 0.09185776859521866, 'learning_rate': 2.7203886400589752e-05, 'flow_matching_loss': 0.017051411606371403, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.07} +{'loss': 0.0171, 'grad_norm': 0.10021667927503586, 'learning_rate': 2.7194265125856105e-05, 'flow_matching_loss': 0.017098691128194333, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.08} +{'loss': 0.0168, 'grad_norm': 0.11007888615131378, 'learning_rate': 2.718462903394049e-05, 'flow_matching_loss': 0.016829704586416482, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.08} +{'loss': 0.0175, 'grad_norm': 0.1074223592877388, 'learning_rate': 2.7174978136551662e-05, 'flow_matching_loss': 0.017500566504895686, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.08} +{'loss': 0.0181, 'grad_norm': 0.1136183813214302, 'learning_rate': 2.716531244541637e-05, 'flow_matching_loss': 0.018054713401943444, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.09} +{'loss': 0.0165, 'grad_norm': 0.08568757772445679, 'learning_rate': 2.715563197227934e-05, 'flow_matching_loss': 0.01648937426507473, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.09} +{'loss': 0.0154, 'grad_norm': 0.07440062612295151, 'learning_rate': 2.714593672890327e-05, 'flow_matching_loss': 0.015366742014884948, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.09} +{'loss': 0.0175, 'grad_norm': 0.10559970140457153, 'learning_rate': 2.7136226727068783e-05, 'flow_matching_loss': 0.01750759230926633, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.09} +{'loss': 0.0155, 'grad_norm': 0.07742292433977127, 'learning_rate': 2.712650197857445e-05, 'flow_matching_loss': 0.015514460671693087, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.1} +{'loss': 0.0176, 'grad_norm': 0.07515960931777954, 'learning_rate': 2.7116762495236763e-05, 'flow_matching_loss': 0.017607845552265643, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.1} +{'loss': 0.0159, 'grad_norm': 0.09743686765432358, 'learning_rate': 2.7107008288890106e-05, 'flow_matching_loss': 0.015851312689483164, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.1} +{'loss': 0.0154, 'grad_norm': 0.06548064947128296, 'learning_rate': 2.709723937138676e-05, 'flow_matching_loss': 0.015351446624845267, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.11} +{'loss': 0.0166, 'grad_norm': 0.07743366807699203, 'learning_rate': 2.7087455754596883e-05, 'flow_matching_loss': 0.016595125664025545, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.11} +{'loss': 0.0161, 'grad_norm': 0.09489785134792328, 'learning_rate': 2.7077657450408498e-05, 'flow_matching_loss': 0.016051541734486818, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.11} +{'loss': 0.0158, 'grad_norm': 0.08436472713947296, 'learning_rate': 2.7067844470727456e-05, 'flow_matching_loss': 0.01579417735338211, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.12} +{'loss': 0.0188, 'grad_norm': 0.09410113841295242, 'learning_rate': 2.7058016827477455e-05, 'flow_matching_loss': 0.0188221450895071, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.12} +{'loss': 0.0157, 'grad_norm': 0.09739621728658676, 'learning_rate': 2.7048174532600018e-05, 'flow_matching_loss': 0.01571461632847786, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.12} +{'loss': 0.0161, 'grad_norm': 0.09729532152414322, 'learning_rate': 2.7038317598054455e-05, 'flow_matching_loss': 0.016110573802143334, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.12} +{'loss': 0.0165, 'grad_norm': 0.06990286707878113, 'learning_rate': 2.702844603581787e-05, 'flow_matching_loss': 0.016490278113633395, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.13} +{'loss': 0.0164, 'grad_norm': 0.10305049270391464, 'learning_rate': 2.701855985788514e-05, 'flow_matching_loss': 0.016419134754687547, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.13} +{'loss': 0.0162, 'grad_norm': 0.10823473334312439, 'learning_rate': 2.7008659076268912e-05, 'flow_matching_loss': 0.01617075828835368, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.13} +{'loss': 0.0174, 'grad_norm': 0.09205766767263412, 'learning_rate': 2.6998743702999562e-05, 'flow_matching_loss': 0.017440353985875845, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.14} +{'loss': 0.016, 'grad_norm': 0.1336563527584076, 'learning_rate': 2.6988813750125202e-05, 'flow_matching_loss': 0.0160367825999856, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.14} +{'loss': 0.0169, 'grad_norm': 0.09867194294929504, 'learning_rate': 2.6978869229711668e-05, 'flow_matching_loss': 0.016865062713623046, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.14} +{'loss': 0.0169, 'grad_norm': 0.08355677127838135, 'learning_rate': 2.6968910153842484e-05, 'flow_matching_loss': 0.01690604668110609, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.14} +{'loss': 0.0157, 'grad_norm': 0.08776777982711792, 'learning_rate': 2.6958936534618874e-05, 'flow_matching_loss': 0.015676092728972434, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.15} +{'loss': 0.0161, 'grad_norm': 0.10372947156429291, 'learning_rate': 2.694894838415971e-05, 'flow_matching_loss': 0.01612120782956481, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.15} +{'loss': 0.0157, 'grad_norm': 0.09617681056261063, 'learning_rate': 2.6938945714601555e-05, 'flow_matching_loss': 0.01571722310036421, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.15} +{'loss': 0.0153, 'grad_norm': 0.09277535229921341, 'learning_rate': 2.6928928538098587e-05, 'flow_matching_loss': 0.015271288715302945, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.16} +{'loss': 0.0176, 'grad_norm': 0.08877278119325638, 'learning_rate': 2.6918896866822615e-05, 'flow_matching_loss': 0.017560992669314147, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.16} +{'loss': 0.0153, 'grad_norm': 0.09108344465494156, 'learning_rate': 2.690885071296308e-05, 'flow_matching_loss': 0.01533149303868413, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.16} +{'loss': 0.0178, 'grad_norm': 0.10470380634069443, 'learning_rate': 2.6898790088726993e-05, 'flow_matching_loss': 0.017757116910070182, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.16} +{'loss': 0.0172, 'grad_norm': 0.10377341508865356, 'learning_rate': 2.6888715006338968e-05, 'flow_matching_loss': 0.017198972310870884, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.17} +{'loss': 0.0166, 'grad_norm': 0.09103534370660782, 'learning_rate': 2.6878625478041172e-05, 'flow_matching_loss': 0.01663223747164011, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.17} +{'loss': 0.0195, 'grad_norm': 0.114688441157341, 'learning_rate': 2.6868521516093342e-05, 'flow_matching_loss': 0.019528802204877137, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.17} +{'loss': 0.0144, 'grad_norm': 0.06969216465950012, 'learning_rate': 2.6858403132772745e-05, 'flow_matching_loss': 0.014382893871515989, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.18} +{'loss': 0.017, 'grad_norm': 0.10866544395685196, 'learning_rate': 2.6848270340374162e-05, 'flow_matching_loss': 0.016963027883321047, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.18} +{'loss': 0.0149, 'grad_norm': 0.0828837901353836, 'learning_rate': 2.68381231512099e-05, 'flow_matching_loss': 0.014895957242697478, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.18} +{'loss': 0.017, 'grad_norm': 0.06893998384475708, 'learning_rate': 2.6827961577609745e-05, 'flow_matching_loss': 0.017039693985134365, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.19} 25%|██▌ | 7510/30000 [4:48:50<14:24:15, 2.31s/it] 25%|██▌ | 7511/30000 [4:48:52<14:26:45, 2.31s/it] 25%|██▌ | 7512/30000 [4:48:54<14:23:25, 2.30s/it] 25%|██▌ | 7513/30000 [4:48:56<14:22:47, 2.30s/it] 25%|██▌ | 7514/30000 [4:48:59<14:24:17, 2.31s/it] 25%|██▌ | 7515/30000 [4:49:01<14:18:40, 2.29s/it] 25%|██▌ | 7516/30000 [4:49:03<14:15:07, 2.28s/it] 25%|██▌ | 7517/30000 [4:49:06<14:21:08, 2.30s/it] 25%|██▌ | 7518/30000 [4:49:08<14:27:50, 2.32s/it] 25%|██▌ | 7519/30000 [4:49:10<14:26:42, 2.31s/it] 25%|██▌ | 7520/30000 [4:49:13<14:32:28, 2.33s/it] 25%|██▌ | 7520/30000 [4:49:13<14:32:28, 2.33s/it] 25%|██▌ | 7521/30000 [4:49:15<14:30:17, 2.32s/it] 25%|██▌ | 7522/30000 [4:49:17<14:24:12, 2.31s/it] 25%|██▌ | 7523/30000 [4:49:19<14:15:32, 2.28s/it] 25%|██▌ | 7524/30000 [4:49:22<14:21:04, 2.30s/it] 25%|██▌ | 7525/30000 [4:49:24<14:18:41, 2.29s/it] 25%|██▌ | 7526/30000 [4:49:26<14:15:23, 2.28s/it] 25%|██▌ | 7527/30000 [4:49:29<14:22:03, 2.30s/it] 25%|██▌ | 7528/30000 [4:49:31<14:16:03, 2.29s/it] 25%|██▌ | 7529/30000 [4:49:33<14:15:39, 2.28s/it] 25%|██▌ | 7530/30000 [4:49:35<14:23:12, 2.30s/it] 25%|██▌ | 7530/30000 [4:49:36<14:23:12, 2.30s/it] 25%|██▌ | 7531/30000 [4:49:38<14:24:56, 2.31s/it] 25%|██▌ | 7532/30000 [4:49:40<14:21:53, 2.30s/it] 25%|██▌ | 7533/30000 [4:49:42<14:23:46, 2.31s/it] 25%|██▌ | 7534/30000 [4:49:45<14:18:23, 2.29s/it] 25%|██▌ | 7535/30000 [4:49:47<14:18:12, 2.29s/it] 25%|██▌ | 7536/30000 [4:49:49<14:19:51, 2.30s/it] 25%|██▌ | 7537/30000 [4:49:52<14:25:30, 2.31s/it] 25%|██▌ | 7538/30000 [4:49:54<14:20:11, 2.30s/it] 25%|██▌ | 7539/30000 [4:49:56<14:18:46, 2.29s/it] 25%|██▌ | 7540/30000 [4:49:59<14:27:04, 2.32s/it] 25%|██▌ | 7540/30000 [4:49:59<14:27:04, 2.32s/it] 25%|██▌ | 7541/30000 [4:50:01<14:27:57, 2.32s/it] 25%|██▌ | 7542/30000 [4:50:03<14:22:32, 2.30s/it] 25%|██▌ | 7543/30000 [4:50:05<14:24:38, 2.31s/it] 25%|██▌ | 7544/30000 [4:50:08<14:22:45, 2.31s/it] 25%|██▌ | 7545/30000 [4:50:10<14:21:24, 2.30s/it] 25%|██▌ | 7546/30000 [4:50:12<14:20:50, 2.30s/it] 25%|██▌ | 7547/30000 [4:50:15<14:26:08, 2.31s/it] 25%|██▌ | 7548/30000 [4:50:17<14:26:32, 2.32s/it] 25%|██▌ | 7549/30000 [4:50:19<14:22:28, 2.30s/it] 25%|██▌ | 7550/30000 [4:50:22<14:25:59, 2.31s/it] 25%|██▌ | 7550/30000 [4:50:22<14:25:59, 2.31s/it] 25%|██▌ | 7551/30000 [4:50:24<14:24:59, 2.31s/it] 25%|██▌ | 7552/30000 [4:50:26<14:20:16, 2.30s/it] 25%|██▌ | 7553/30000 [4:50:28<14:16:35, 2.29s/it] 25%|██▌ | 7554/30000 [4:50:31<14:22:14, 2.30s/it] 25%|██▌ | 7555/30000 [4:50:33<14:19:03, 2.30s/it] 25%|██▌ | 7556/30000 [4:50:35<14:16:20, 2.29s/it] 25%|██▌ | 7557/30000 [4:50:38<14:24:51, 2.31s/it] 25%|██▌ | 7558/30000 [4:50:40<14:27:21, 2.32s/it] 25%|██▌ | 7559/30000 [4:50:42<14:23:13, 2.31s/it] 25%|██▌ | 7560/30000 [4:50:45<14:21:06, 2.30s/it] 25%|██▌ | 7560/30000 [4:50:45<14:21:06, 2.30s/it] 25%|██▌ | 7561/30000 [4:50:47<14:34:31, 2.34s/it] 25%|██▌ | 7562/30000 [4:50:49<14:31:39, 2.33s/it] 25%|██▌ | 7563/30000 [4:50:52<14:24:35, 2.31s/it] 25%|██▌ | 7564/30000 [4:50:54<14:26:56, 2.32s/it] 25%|██▌ | 7565/30000 [4:50:56<14:26:45, 2.32s/it] 25%|██▌ | 7566/30000 [4:50:59<14:27:19, 2.32s/it] 25%|██▌ | 7567/30000 [4:51:01<14:27:11, 2.32s/it] 25%|██▌ | 7568/30000 [4:51:03<14:34:46, 2.34s/it] 25%|██▌ | 7569/30000 [4:51:06<14:31:48, 2.33s/it] 25%|██▌ | 7570/30000 [4:51:08<14:24:26, 2.31s/it] 25%|██▌ | 7570/30000 [4:51:08<14:24:26, 2.31s/it] 25%|██▌ | 7571/30000 [4:51:10<14:22:25, 2.31s/it] 25%|██▌ | 7572/30000 [4:51:13<14:23:38, 2.31s/it] 25%|██▌ | 7573/30000 [4:51:15<14:18:10, 2.30s/it] 25%|██▌ | 7574/30000 [4:51:17<14:18:15, 2.30s/it] 25%|██▌ | 7575/30000 [4:51:19<14:24:01, 2.31s/it] 25%|██▌ | 7576/30000 [4:51:22<14:25:39, 2.32s/it] 25%|██▌ | 7577/30000 [4:51:24<14:23:23, 2.31s/it] 25%|██▌ | 7578/30000 [4:51:26<14:23:54, 2.31s/it] 25%|██▌ | 7579/30000 [4:51:29<14:22:46, 2.31s/it] 25%|██▌ | 7580/30000 [4:51:31<14:17:18, 2.29s/it] 25%|██▌ | 7580/30000 [4:51:31<14:17:18, 2.29s/it] 25%|██▌ | 7581/30000 [4:51:33<14:15:38, 2.29s/it] 25%|██▌ | 7582/30000 [4:51:36<14:19:59, 2.30s/it] 25%|██▌ | 7583/30000 [4:51:38<14:14:50, 2.29s/it] 25%|██▌ | 7584/30000 [4:51:40<14:13:59, 2.29s/it] 25%|██▌ | 7585/30000 [4:51:42<14:24:02, 2.31s/it] 25%|██▌ | 7586/30000 [4:51:45<14:19:12, 2.30s/it] 25%|██▌ | 7587/30000 [4:51:47<14:15:30, 2.29s/it] 25%|██▌ | 7588/30000 [4:51:49<14:22:32, 2.31s/it] 25%|██▌ | 7589/30000 [4:51:52<14:23:33, 2.31s/it] 25%|██▌ | 7590/30000 [4:51:54<14:21:14, 2.31s/it] 25%|██▌ | 7590/30000 [4:51:54<14:21:14, 2.31s/it] 25%|██▌ | 7591/30000 [4:51:56<14:19:20, 2.30s/it] 25%|██▌ | 7592/30000 [4:51:59<14:24:07, 2.31s/it] 25%|██▌ | 7593/30000 [4:52:01<14:21:36, 2.31s/it] 25%|██▌ | 7594/30000 [4:52:03<14:15:41, 2.29s/it] 25%|██▌ | 7595/30000 [4:52:05<14:19:52, 2.30s/it] 25%|██▌ | 7596/30000 [4:52:08<14:19:06, 2.30s/it] 25%|██▌ | 7597/30000 [4:52:10<14:16:00, 2.29s/it] 25%|██▌ | 7598/30000 [4:52:12<14:17:30, 2.30s/it] 25%|██▌ | 7599/30000 [4:52:15<14:23:46, 2.31s/it] 25%|██▌ | 7600/30000 [4:52:17<14:18:07, 2.30s/it] 25%|██▌ | 7600/30000 [4:52:17<14:18:07, 2.30s/it] 25%|██▌ | 7601/30000 [4:52:19<14:14:27, 2.29s/it] 25%|██▌ | 7602/30000 [4:52:22<14:19:42, 2.30s/it] 25%|██▌ | 7603/30000 [4:52:24<14:14:48, 2.29s/it] 25%|██▌ | 7604/30000 [4:52:26<14:12:19, 2.28s/it] 25%|██▌ | 7605/30000 [4:52:28<14:09:10, 2.28s/it] 25%|██▌ | 7606/30000 [4:52:31<14:17:07, 2.30s/it] 25%|██▌ | 7607/30000 [4:52:33<14:22:03, 2.31s/it] 25%|██▌ | 7608/30000 [4:52:35<14:13:20, 2.29s/it] 25%|██▌ | 7609/30000 [4:52:38<14:14:45, 2.29s/it] 25%|██▌ | 7610/30000 [4:52:40<14:18:27, 2.30s/it] 25%|██▌ | 7610/30000 [4:52:40<14:18:27, 2.30s/it] 25%|██▌ | 7611/30000 [4:52:42<14:17:46, 2.30s/it] 25%|██▌ | 7612/30000 [4:52:44<14:18:37, 2.30s/it] 25%|██▌ | 7613/30000 [4:52:47<14:22:57, 2.31s/it] 25%|██▌ | 7614/30000 [4:52:49<14:21:19, 2.31s/it] 25%|██▌ | 7615/30000 [4:52:51<14:18:54, 2.30s/it] 25%|██▌ | 7616/30000 [4:52:54<14:18:08, 2.30s/it] 25%|██▌ | 7617/30000 [4:52:56<14:27:03, 2.32s/it] 25%|██▌ | 7618/30000 [4:52:58<14:24:02, 2.32s/it] 25%|██▌ | 7619/30000 [4:53:01<14:18:18, 2.30s/it] 25%|██▌ | 7620/30000 [4:53:03<14:12:49, 2.29s/it] 25%|██▌ | 7620/30000 [4:53:03<14:12:49, 2.29s/it] 25%|██▌ | 7621/30000 [4:53:05<14:17:52, 2.30s/it] 25%|██▌ | 7622/30000 [4:53:08<14:16:43, 2.30s/it] 25%|██▌ | 7623/30000 [4:53:10<14:13:09, 2.29s/it] 25%|██▌ | 7624/30000 [4:53:12<14:17:04, 2.30s/it] 25%|██▌ | 7625/30000 [4:53:14<14:13:13, 2.29s/it] 25%|██▌ | 7626/30000 [4:53:17<14:11:24, 2.28s/it] 25%|██▌ | 7627/30000 [4:53:19<14:07:37, 2.27s/it] 25%|██▌ | 7628/30000 [4:53:21<14:11:35, 2.28s/it] 25%|██▌ | 7629/30000 [4:53:24<14:18:46, 2.30s/it] 25%|██▌ | 7630/30000 [4:53:26<14:15:58, 2.30s/it] 25%|██▌ | 7630/30000 [4:53:26<14:15:58, 2.30s/it] 25%|██▌ | 7631/30000 [4:53:28<14:14:16, 2.29s/it] 25%|██▌ | 7632/30000 [4:53:30<14:18:13, 2.30s/it] 25%|██▌ | 7633/30000 [4:53:33<14:13:58, 2.29s/it] 25%|██▌ | 7634/30000 [4:53:35<14:12:55, 2.29s/it] 25%|██▌ | 7635/30000 [4:53:37<14:10:54, 2.28s/it] 25%|██▌ | 7636/30000 [4:53:40<14:13:47, 2.29s/it] 25%|██▌ | 7637/30000 [4:53:42<14:10:50, 2.28s/it] 25%|██▌ | 7638/30000 [4:53:44<14:10:16, 2.28s/it] 25%|██▌ | 7639/30000 [4:53:46<14:08:12, 2.28s/it] 25%|██▌ | 7640/30000 [4:53:49<14:14:10, 2.29s/it] 25%|██▌ | 7640/30000 [4:53:49<14:14:10, 2.29s/it] 25%|██▌ | 7641/30000 [4:53:51<14:21:19, 2.31s/it] 25%|██▌ | 7642/30000 [4:53:53<14:15:16, 2.30s/it] 25%|██▌ | 7643/30000 [4:53:56<14:12:02, 2.29s/it] 25%|██▌ | 7644/30000 [4:53:58<14:14:28, 2.29s/it] 25%|██▌ | 7645/30000 [4:54:00<14:09:52, 2.28s/it] 25%|██▌ | 7646/30000 [4:54:02<14:06:33, 2.27s/it] 25%|██▌ | 7647/30000 [4:54:05<14:14:28, 2.29s/it] 25%|██▌ | 7648/30000 [4:54:07<14:16:39, 2.30s/it] 25%|██▌ | 7649/30000 [4:54:09<14:15:21, 2.30s/it] 26%|██▌ | 7650/30000 [4:54:12<14:12:53, 2.29s/it] 26%|██▌ | 7650/30000 [4:54:12<14:12:53, 2.29s/it] 26%|██▌ | 7651/30000 [4:54:14<14:14:50, 2.29s/it] 26%|██▌ | 7652/30000 [4:54:16<14:08:01, 2.28s/it] 26%|██▌ | 7653/30000 [4:54:18<14:07:47, 2.28s/it] 26%|██▌ | 7654/30000 [4:54:21<14:15:17, 2.30s/it] 26%|██▌ | 7655/30000 [4:54:23<14:17:28, 2.30s/it] 26%|██▌ | 7656/30000 [4:54:25<14:12:41, 2.29s/it] 26%|██▌ | 7657/30000 [4:54:28<14:10:26, 2.28s/it] 26%|██▌ | 7658/30000 [4:54:30<14:15:16, 2.30s/it] 26%|██▌ | 7659/30000 [4:54:32<14:17:18, 2.30s/it] 26%|██▌ | 7660/30000 [4:54:35<14:17:46, 2.30s/it] 26%|██▌ | 7660/30000 [4:54:35<14:17:46, 2.30s/it] 26%|██▌ | 7661/30000 [4:54:37<14:13:10, 2.29s/it] 26%|██▌ | 7662/30000 [4:54:39<14:20:15, 2.31s/it] 26%|██▌ | 7663/30000 [4:54:41<14:21:25, 2.31s/it] 26%|██▌ | 7664/30000 [4:54:44<14:19:49, 2.31s/it] 26%|██▌ | 7665/30000 [4:54:46<14:19:00, 2.31s/it] 26%|██▌ | 7666/30000 [4:54:48<14:19:45, 2.31s/it] 26%|██▌ | 7667/30000 [4:54:51<14:14:14, 2.30s/it] 26%|██▌ | 7668/30000 [4:54:53<14:14:11, 2.29s/it] 26%|██▌ | 7669/30000 [4:54:55<14:21:21, 2.31s/it] 26%|██▌ | 7670/30000 [4:54:58<14:22:08, 2.32s/it] 26%|██▌ | 7670/30000 [4:54:58<14:22:08, 2.32s/it] 26%|██▌ | 7671/30000 [4:55:00<14:20:59, 2.31s/it] 26%|██▌ | 7672/30000 [4:55:02<14:17:12, 2.30s/it] 26%|██▌ | 7673/30000 [4:55:05<14:25:59, 2.33s/it] 26%|██▌ | 7674/30000 [4:55:07<14:22:29, 2.32s/it] 26%|██▌ | 7675/30000 [4:55:09<14:16:44, 2.30s/it] 26%|██▌ | 7676/30000 [4:55:12<14:19:36, 2.31s/it] 26%|██▌ | 7677/30000 [4:55:14<14:19:46, 2.31s/it] 26%|██▌ | 7678/30000 [4:55:16<14:16:33, 2.30s/it] 26%|██▌ | 7679/30000 [4:55:18<14:13:34, 2.29s/it] 26%|██▌ | 7680/30000 [4:55:21<14:19:42, 2.31s/it] 26%|██▌ | 7680/30000 [4:55:21<14:19:42, 2.31s/it] 26%|██▌ | 7681/30000 [4:55:23<14:20:07, 2.31s/it] 26%|██▌ | 7682/30000 [4:55:25<14:18:45, 2.31s/it] 26%|██▌ | 7683/30000 [4:55:28<14:21:09, 2.32s/it] 26%|██▌ | 7684/30000 [4:55:30<14:19:15, 2.31s/it] 26%|██▌ | 7685/30000 [4:55:32<14:10:36, 2.29s/it] 26%|██▌ | 7686/30000 [4:55:34<14:07:30, 2.28s/it] 26%|██▌ | 7687/30000 [4:55:37<14:17:42, 2.31s/it] 26%|██▌ | 7688/30000 [4:55:39<14:20:20, 2.31s/it] 26%|██▌ | 7689/30000 [4:55:41<14:18:30, 2.31s/it] 26%|██▌ | 7690/30000 [4:55:44<14:14:45, 2.30s/it] 26%|██▌ | 7690/30000 [4:55:44<14:14:45, 2.30s/it] 26%|██▌ | 7691/30000 [4:55:46<14:15:58, 2.30s/it] 26%|██▌ | 7692/30000 [4:55:48<14:11:38, 2.29s/it] 26%|██▌ | 7693/30000 [4:55:51<14:08:16, 2.28s/it] 26%|██▌ | 7694/30000 [4:55:53<14:16:03, 2.30s/it] 26%|██▌ | 7695/30000 [4:55:55<14:09:44, 2.29s/it] 26%|██▌ | 7696/30000 [4:55:57<14:08:35, 2.28s/it] 26%|██▌ | 7697/30000 [4:56:00<14:20:22, 2.31s/it] 26%|██▌ | 7698/30000 [4:56:02<14:23:54, 2.32s/it] 26%|██▌ | 7699/30000 [4:56:04<14:17:06, 2.31s/it] 26%|██▌ | 7700/30000 [4:56:07<14:13:40, 2.30s/it] 26%|██▌ | 7700/30000 [4:56:07<14:13:40, 2.30s/it] 26%|██▌ | 7701/30000 [4:56:09<14:17:01, 2.31s/it] 26%|██▌ | 7702/30000 [4:56:11<14:20:40, 2.32s/it] 26%|██▌ | 7703/30000 [4:56:14<14:15:00, 2.30s/it] 26%|██▌ | 7704/30000 [4:56:16<14:22:05, 2.32s/it] 26%|██▌ | 7705/30000 [4:56:18<14:20:56, 2.32s/it] 26%|██▌ | 7706/30000 [4:56:21<14:13:02, 2.30s/it] 26%|██▌ | 7707/30000 [4:56:23<14:17:42, 2.31s/it] 26%|██▌ | 7708/30000 [4:56:25<14:17:47, 2.31s/it] 26%|██▌ | 7709/30000 [4:56:27<14:13:29, 2.30s/it] 26%|██▌ | 7710/30000 [4:56:30<14:20:50, 2.32s/it] 26%|██▌ | 7710/30000 [4:56:30<14:20:50, 2.32s/it] 26%|██▌ | 7711/30000 [4:56:32<14:21:40, 2.32s/it] 26%|██▌ | 7712/30000 [4:56:34<14:16:46, 2.31s/it] 26%|██▌ | 7713/30000 [4:56:37<14:17:14, 2.31s/it] 26%|██▌ | 7714/30000 [4:56:39<14:19:57, 2.32s/it] 26%|██▌ | 7715/30000 [4:56:41<14:14:21, 2.30s/it] 26%|██▌ | 7716/30000 [4:56:44<14:09:53, 2.29s/it] 26%|██▌ | 7717/30000 [4:56:46<14:14:48, 2.30s/it] 26%|██▌ | 7718/30000 [4:56:48<14:14:35, 2.30s/it] 26%|██▌ | 7719/30000 [4:56:51<14:16:25, 2.31s/it] 26%|██▌ | 7720/30000 [4:56:53<14:26:38, 2.33s/it] 26%|██▌ | 7720/30000 [4:56:53<14:26:38, 2.33s/it] 26%|██▌ | 7721/30000 [4:56:55<14:31:13, 2.35s/it] 26%|██▌ | 7722/30000 [4:56:58<14:20:30, 2.32s/it] 26%|██▌ | 7723/30000 [4:57:00<14:15:19, 2.30s/it] 26%|██▌ | 7724/30000 [4:57:02<14:19:37, 2.32s/it] 26%|██▌ | 7725/30000 [4:57:05<14:17:54, 2.31s/it] 26%|██▌ | 7726/30000 [4:57:07<14:12:29, 2.30s/it] 26%|██▌ | 7727/30000 [4:57:09<14:19:01, 2.31s/it] 26%|██▌ | 7728/30000 [4:57:11<14:19:12, 2.31s/it] 26%|██▌ | 7729/30000 [4:57:14<14:15:32, 2.30s/it] 26%|██▌ | 7730/30000 [4:57:16<14:19:12, 2.31s/it] 26%|██▌ | 7730/30000 [4:57:16<14:19:12, 2.31s/it] 26%|██▌ | 7731/30000 [4:57:18<14:24:20, 2.33s/it] 26%|██▌ | 7732/30000 [4:57:21<14:17:17, 2.31s/it] 26%|██▌ | 7733/30000 [4:57:23<14:20:15, 2.32s/it] 26%|██▌ | 7734/30000 [4:57:25<14:20:48, 2.32s/it] 26%|██▌ | 7735/30000 [4:57:28<14:14:55, 2.30s/it] 26%|██▌ | 7736/30000 [4:57:30<14:15:09, 2.30s/it] 26%|██▌ | 7737/30000 [4:57:32<14:22:55, 2.33s/it] 26%|██▌ | 7738/30000 [4:57:35<14:16:32, 2.31s/it] 26%|██▌ | 7739/30000 [4:57:37<14:11:26, 2.29s/it] 26%|██▌ | 7740/30000 [4:57:39<14:18:39, 2.31s/it] 26%|██▌ | 7740/30000 [4:57:39<14:18:39, 2.31s/it] 26%|██▌ | 7741/30000 [4:57:42<14:19:06, 2.32s/it] 26%|██▌ | 7742/30000 [4:57:44<14:15:52, 2.31s/it] 26%|██▌ | 7743/30000 [4:57:46<14:22:45, 2.33s/it] 26%|██▌ | 7744/30000 [4:57:49<14:23:27, 2.33s/it] 26%|██▌ | 7745/30000 [4:57:51<14:19:28, 2.32s/it] 26%|██▌ | 7746/30000 [4:57:53<14:16:36, 2.31s/it] 26%|██▌ | 7747/30000 [4:57:55<14:17:55, 2.31s/it] 26%|██▌ | 7748/30000 [4:57:58<14:11:52, 2.30s/it] 26%|██▌ | 7749/30000 [4:58:00<14:09:16, 2.29s/it] 26%|██▌ | 7750/30000 [4:58:02<14:17:28, 2.31s/it] 26%|██▌ | 7750/30000 [4:58:02<14:17:28, 2.31s/it] 26%|██▌ | 7751/30000 [4:58:05<14:19:46, 2.32s/it] 26%|██▌ | 7752/30000 [4:58:07<14:16:03, 2.31s/it] 26%|██▌ | 7753/30000 [4:58:09<14:26:16, 2.34s/it] 26%|██▌ | 7754/30000 [4:58:12<14:22:42, 2.33s/it] 26%|██▌ | 7755/30000 [4:58:14<14:15:56, 2.31s/it] 26%|██▌ | 7756/30000 [4:58:16<14:12:42, 2.30s/it] 26%|██▌ | 7757/30000 [4:58:19<14:15:42, 2.31s/it] 26%|██▌ | 7758/30000 [4:58:21<14:13:47, 2.30s/it] 26%|██▌ | 7759/30000 [4:58:23<14:11:58, 2.30s/it] 26%|██▌ | 7760/30000 [4:58:25<14:17:22, 2.31s/it] 26%|██▌ | 7760/30000 [4:58:25<14:17:22, 2.31s/it] 26%|██▌ | 7761/30000 [4:58:28<14:19:16, 2.32s/it] 26%|██▌ | 7762/30000 [4:58:30<14:14:09, 2.30s/it] 26%|██▌ | 7763/30000 [4:58:32<14:15:34, 2.31s/it] 26%|██▌ | 7764/30000 [4:58:35<14:15:07, 2.31s/it] 26%|██▌ | 7765/30000 [4:58:37<14:11:04, 2.30s/it] 26%|██▌ | 7766/30000 [4:58:39<14:08:59, 2.29s/it] 26%|██▌ | 7767/30000 [4:58:42<14:15:47, 2.31s/it] 26%|██▌ | 7768/30000 [4:58:44<14:09:39, 2.29s/it] 26%|██▌ | 7769/30000 [4:58:46<14:13:06, 2.30s/it] 26%|██▌ | 7770/30000 [4:58:48<14:16:35, 2.31s/it] 26%|██▌ | 7770/30000 [4:58:49<14:16:35, 2.31s/it] 26%|██▌ | 7771/30000 [4:58:51<14:18:53, 2.32s/it] 26%|██▌ | 7772/30000 [4:58:53<14:13:49, 2.30s/it] 26%|██▌ | 7773/30000 [4:58:55<14:19:43, 2.32s/it] 26%|██▌ | 7774/30000 [4:58:58<14:23:38, 2.33s/it] 26%|██▌ | 7775/30000 [4:59:00<14:11:29, 2.30s/it] 26%|██▌ | 7776/30000 [4:59:02<14:09:18, 2.29s/it] 26%|██▌ | 7777/30000 [4:59:05<14:14:25, 2.31s/it] 26%|██▌ | 7778/30000 [4:59:07<14:08:10, 2.29s/it] 26%|██▌ | 7779/30000 [4:59:09<14:07:05, 2.29s/it] 26%|██▌ | 7780/30000 [4:59:12<14:14:48, 2.31s/it] 26%|██▌ | 7780/30000 [4:59:12<14:14:48, 2.31s/it] 26%|██▌ | 7781/30000 [4:59:14<14:13:34, 2.30s/it] 26%|██▌ | 7782/30000 [4:59:16<14:10:11, 2.30s/it] 26%|██▌ | 7783/30000 [4:59:18<14:17:59, 2.32s/it] 26%|██▌ | 7784/30000 [4:59:21<14:17:48, 2.32s/it] 26%|██▌ | 7785/30000 [4:59:23<14:12:43, 2.30s/it] 26%|██▌ | 7786/30000 [4:59:25<14:13:45, 2.31s/it] 26%|██▌ | 7787/30000 [4:59:28<14:10:32, 2.30s/it] 26%|██▌ | 7788/30000 [4:59:30<14:07:57, 2.29s/it] 26%|██▌ | 7789/30000 [4:59:32<14:04:36, 2.28s/it] 26%|██▌ | 7790/30000 [4:59:35<14:14:52, 2.31s/it] 26%|██▌ | 7790/30000 [4:59:35<14:14:52, 2.31s/it] 26%|██▌ | 7791/30000 [4:59:37<14:19:08, 2.32s/it] 26%|██▌ | 7792/30000 [4:59:39<14:17:23, 2.32s/it] 26%|██▌ | 7793/30000 [4:59:42<14:20:46, 2.33s/it] 26%|██▌ | 7794/30000 [4:59:44<14:23:17, 2.33s/it] 26%|██▌ | 7795/30000 [4:59:46<14:15:56, 2.31s/it] 26%|██▌ | 7796/30000 [4:59:48<14:12:02, 2.30s/it] 26%|██▌ | 7797/30000 [4:59:51<14:17:04, 2.32s/it] 26%|██▌ | 7798/30000 [4:59:53<14:16:31, 2.31s/it] 26%|██▌ | 7799/30000 [4:59:55<14:12:07, 2.30s/it] 26%|██▌ | 7800/30000 [4:59:58<14:18:27, 2.32s/it] 26%|██▌ | 7800/30000 [4:59:58<14:18:27, 2.32s/it] 26%|██▌ | 7801/30000 [5:00:00<14:17:43, 2.32s/it] 26%|██▌ | 7802/30000 [5:00:02<14:12:53, 2.31s/it] 26%|██▌ | 7803/30000 [5:00:05<14:16:30, 2.32s/it] 26%|██▌ | 7804/30000 [5:00:07<14:12:43, 2.31s/it] 26%|██▌ | 7805/30000 [5:00:09<14:09:07, 2.30s/it] 26%|██▌ | 7806/30000 [5:00:11<14:07:04, 2.29s/it] 26%|██▌ | 7807/30000 [5:00:14<14:10:59, 2.30s/it] 26%|██▌ | 7808/30000 [5:00:16<14:08:46, 2.29s/it] 26%|██▌ | 7809/30000 [5:00:18<14:12:31, 2.31s/it] 26%|██▌ | 7810/30000 [5:00:21<14:20:40, 2.33s/it] 26%|██▌ | 7810/30000 [5:00:21<14:20:40, 2.33s/it] 26%|██▌ | 7811/30000 [5:00:23<14:15:47, 2.31s/it] 26%|██▌ | 7812/30000 [5:00:25<14:12:07, 2.30s/it] 26%|██▌ | 7813/30000 [5:00:28<14:15:03, 2.31s/it] 26%|██▌ | 7814/30000 [5:00:30<14:18:16, 2.32s/it] 26%|██▌ | 7815/30000 [5:00:32<14:13:09, 2.31s/it] 26%|██▌ | 7816/30000 [5:00:35<14:08:46, 2.30s/it] 26%|██▌ | 7817/30000 [5:00:37<14:10:17, 2.30s/it] 26%|██▌ | 7818/30000 [5:00:39<14:07:15, 2.29s/it] 26%|██▌ | 7819/30000 [5:00:41<14:03:23, 2.28s/it] 26%|██▌ | 7820/30000 [5:00:44<14:10:45, 2.30s/it] 26%|██▌ | 7820/30000 [5:00:44<14:10:45, 2.30s/it] 26%|██▌ | 7821/30000 [5:00:46<14:14:18, 2.31s/it] 26%|██▌ | 7822/30000 [5:00:48<14:12:24, 2.31s/it] 26%|██▌ | 7823/30000 [5:00:51<14:13:14, 2.31s/it] 26%|██▌ | 7824/30000 [5:00:53<14:12:42, 2.31s/it] 26%|██▌ | 7825/30000 [5:00:55<14:12:18, 2.31s/it] 26%|██▌ | 7826/30000 [5:00:58<14:10:37, 2.30s/it] 26%|██▌ | 7827/30000 [5:01:00<14:17:42, 2.32s/it] 26%|██▌ | 7828/30000 [5:01:02<14:16:01, 2.32s/it] 26%|██▌ | 7829/30000 [5:01:05<14:09:21, 2.30s/it] 26%|██▌ | 7830/30000 [5:01:07<14:08:12, 2.30s/it] 26%|██▌ | 7830/30000 [5:01:07<14:08:12, 2.30s/it] 26%|██▌ | 7831/30000 [5:01:09<14:12:00, 2.31s/it] 26%|██▌ | 7832/30000 [5:01:11<14:08:45, 2.30s/it] 26%|██▌ | 7833/30000 [5:01:14<14:04:33, 2.29s/it] 26%|██▌ | 7834/30000 [5:01:16<14:11:01, 2.30s/it] 26%|██▌ | 7835/30000 [5:01:18<14:18:08, 2.32s/it] 26%|██▌ | 7836/30000 [5:01:21<14:14:02, 2.31s/it] 26%|██▌ | 7837/30000 [5:01:23<14:08:19, 2.30s/it] 26%|██▌ | 7838/30000 [5:01:25<14:15:48, 2.32s/it] 26%|██▌ | 7839/30000 [5:01:28<14:18:34, 2.32s/it] 26%|██▌ | 7840/30000 [5:01:30<14:12:47, 2.31s/it] 26%|██▌ | 7840/30000 [5:01:30<14:12:47, 2.31s/it] 26%|██▌ | 7841/30000 [5:01:32<14:11:54, 2.31s/it] 26%|██▌ | 7842/30000 [5:01:35<14:16:00, 2.32s/it] 26%|██▌ | 7843/30000 [5:01:37<14:10:39, 2.30s/it] 26%|██▌ | 7844/30000 [5:01:39<14:06:32, 2.29s/it] 26%|██▌ | 7845/30000 [5:01:41<14:09:07, 2.30s/it] 26%|██▌ | 7846/30000 [5:01:44<14:15:21, 2.32s/it] 26%|██▌ | 7847/30000 [5:01:46<14:11:27, 2.31s/it] 26%|██▌ | 7848/30000 [5:01:48<14:11:17, 2.31s/it] 26%|██▌ | 7849/30000 [5:01:51<14:15:04, 2.32s/it] 26%|██▌ | 7850/30000 [5:01:53<14:09:43, 2.30s/it] 26%|██▌ | 7850/30000 [5:01:53<14:09:43, 2.30s/it] 26%|██▌ | 7851/30000 [5:01:55<14:08:40, 2.30s/it] 26%|██▌ | 7852/30000 [5:01:58<14:13:12, 2.31s/it] 26%|██▌ | 7853/30000 [5:02:00<14:12:43, 2.31s/it] 26%|██▌ | 7854/30000 [5:02:02<14:12:19, 2.31s/it] 26%|██▌ | 7855/30000 [5:02:05<14:07:10, 2.30s/it] 26%|██▌ | 7856/30000 [5:02:07<14:16:25, 2.32s/it] 26%|██▌ | 7857/30000 [5:02:09<14:17:50, 2.32s/it] 26%|██▌ | 7858/30000 [5:02:11<14:12:33, 2.31s/it] 26%|██▌ | 7859/30000 [5:02:14<14:10:14, 2.30s/it] 26%|██▌ | 7860/30000 [5:02:16<14:14:07, 2.31s/it] 26%|██▌ | 7860/30000 [5:02:16<14:14:07, 2.31s/it] 26%|██▌ | 7861/30000 [5:02:18<14:06:20, 2.29s/it] 26%|██▌ | 7862/30000 [5:02:21<14:05:55, 2.29s/it] 26%|██▌ | 7863/30000 [5:02:23<14:13:05, 2.31s/it] 26%|██▌ | 7864/30000 [5:02:25<14:14:40, 2.32s/it] 26%|██▌ | 7865/30000 [5:02:28<14:12:42, 2.31s/it] 26%|██▌ | 7866/30000 [5:02:30<14:02:42, 2.28s/it] 26%|██▌ | 7867/30000 [5:02:32<14:08:15, 2.30s/it] 26%|██▌ | 7868/30000 [5:02:34<14:06:44, 2.30s/it] 26%|██▌ | 7869/30000 [5:02:37<14:03:41, 2.29s/it] 26%|██▌ | 7870/30000 [5:02:39<14:06:38, 2.30s/it] 26%|██▌ | 7870/30000 [5:02:39<14:06:38, 2.30s/it] 26%|██▌ | 7871/30000 [5:02:41<14:09:16, 2.30s/it] 26%|██▌ | 7872/30000 [5:02:44<14:07:24, 2.30s/it] 26%|██▌ | 7873/30000 [5:02:46<14:06:28, 2.30s/it] 26%|██▌ | 7874/30000 [5:02:48<14:12:15, 2.31s/it] 26%|██▋ | 7875/30000 [5:02:51<14:12:30, 2.31s/it] 26%|██▋ | 7876/30000 [5:02:53<14:03:48, 2.29s/it] 26%|██▋ | 7877/30000 [5:02:55<14:03:04, 2.29s/it] 26%|██▋ | 7878/30000 [5:02:57<14:10:17, 2.31s/it] 26%|██▋ | 7879/30000 [5:03:00<14:15:05, 2.32s/it] 26%|██▋ | 7880/30000 [5:03:02<14:09:08, 2.30s/it] 26%|██▋ | 7880/30000 [5:03:02<14:09:08, 2.30s/it] 26%|██▋ | 7881/30000 [5:03:04<14:11:16, 2.31s/it] 26%|██▋ | 7882/30000 [5:03:07<14:06:18, 2.30s/it] 26%|██▋ | 7883/30000 [5:03:09<14:04:07, 2.29s/it] 26%|██▋ | 7884/30000 [5:03:11<13:57:54, 2.27s/it] 26%|██▋ | 7885/30000 [5:03:14<14:04:20, 2.29s/it] 26%|██▋ | 7886/30000 [5:03:16<14:10:07, 2.31s/it] 26%|██▋ | 7887/30000 [5:03:18<14:07:11, 2.30s/it] 26%|██▋ | 7888/30000 [5:03:20<14:07:06, 2.30s/it] 26%|██▋ | 7889/30000 [5:03:23<14:07:52, 2.30s/it] 26%|██▋ | 7890/30000 [5:03:25<14:02:58, 2.29s/it] 26%|██▋ | 7890/30000 [5:03:25<14:02:58, 2.29s/it] 26%|██▋ | 7891/30000 [5:03:27<14:02:35, 2.29s/it] 26%|██▋ | 7892/30000 [5:03:30<14:09:22, 2.31s/it] 26%|██▋ | 7893/30000 [5:03:32<14:08:09, 2.30s/it] 26%|██▋ | 7894/30000 [5:03:34<14:04:45, 2.29s/it] 26%|██▋ | 7895/30000 [5:03:37<14:03:48, 2.29s/it] 26%|██▋ | 7896/30000 [5:03:39<14:07:43, 2.30s/it] 26%|██▋ | 7897/30000 [5:03:41<14:10:40, 2.31s/it] 26%|██▋ | 7898/30000 [5:03:43<14:05:30, 2.30s/it] 26%|██▋ | 7899/30000 [5:03:46<14:02:25, 2.29s/it] 26%|██▋ | 7900/30000 [5:03:48<14:06:39, 2.30s/it] 26%|██▋ | 7900/30000 [5:03:48<14:06:39, 2.30s/it] 26%|██▋ | 7901/30000 [5:03:50<14:04:49, 2.29s/it] 26%|██▋ | 7902/30000 [5:03:53<14:02:52, 2.29s/it] 26%|██▋ | 7903/30000 [5:03:55<14:01:21, 2.28s/it] 26%|██▋ | 7904/30000 [5:03:57<14:09:58, 2.31s/it] 26%|██▋ | 7905/30000 [5:04:00<14:12:28, 2.31s/it] 26%|██▋ | 7906/30000 [5:04:02<14:08:13, 2.30s/it] 26%|██▋ | 7907/30000 [5:04:04<14:02:35, 2.29s/it] 26%|██▋ | 7908/30000 [5:04:06<14:05:51, 2.30s/it] 26%|██▋ | 7909/30000 [5:04:09<14:05:05, 2.30s/it] 26%|██▋ | 7910/30000 [5:04:11<14:02:01, 2.29s/it] 26%|██▋ | 7910/30000 [5:04:11<14:02:01, 2.29s/it] 26%|██▋ | 7911/30000 [5:04:13<13:56:11, 2.27s/it] 26%|██▋ | 7912/30000 [5:04:16<14:06:41, 2.30s/it] 26%|██▋ | 7913/30000 [5:04:18<14:05:11, 2.30s/it] 26%|██▋ | 7914/30000 [5:04:20<14:00:52, 2.28s/it] 26%|██▋ | 7915/30000 [5:04:22<13:59:27, 2.28s/it] 26%|██▋ | 7916/30000 [5:04:25<14:01:09, 2.29s/it] 26%|██▋ | 7917/30000 [5:04:27<13:58:14, 2.28s/it] 26%|██▋ | 7918/30000 [5:04:29<13:58:46, 2.28s/it] 26%|██▋ | 7919/30000 [5:04:31<14:00:10, 2.28s/it] 26%|██▋ | 7920/30000 [5:04:34<14:01:29, 2.29s/it] 26%|██▋ | 7920/30000 [5:04:34<14:01:29, 2.29s/it] 26%|██▋ | 7921/30000 [5:04:36<13:56:01, 2.27s/it] 26%|██▋ | 7922/30000 [5:04:38<13:54:54, 2.27s/it] 26%|██▋ | 7923/30000 [5:04:41<14:01:39, 2.29s/it] 26%|██▋ | 7924/30000 [5:04:43<13:56:26, 2.27s/it] 26%|██▋ | 7925/30000 [5:04:45<13:56:48, 2.27s/it] 26%|██▋ | 7926/30000 [5:04:47<13:54:09, 2.27s/it] 26%|██▋ | 7927/30000 [5:04:50<14:02:26, 2.29s/it] 26%|██▋ | 7928/30000 [5:04:52<14:11:01, 2.31s/it] 26%|██▋ | 7929/30000 [5:04:54<13:45:37, 2.24s/it] 26%|██▋ | 7930/30000 [5:04:56<13:49:34, 2.26s/it] +{'loss': 0.0162, 'grad_norm': 0.07202392816543579, 'learning_rate': 2.681778563192097e-05, 'flow_matching_loss': 0.01618116656318307, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.19} +{'loss': 0.0165, 'grad_norm': 0.09681548178195953, 'learning_rate': 2.6807595326508308e-05, 'flow_matching_loss': 0.016526246815919875, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.19} +{'loss': 0.018, 'grad_norm': 0.10063367336988449, 'learning_rate': 2.679739067375394e-05, 'flow_matching_loss': 0.0179961153306067, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.19} +{'loss': 0.0183, 'grad_norm': 0.08261393010616302, 'learning_rate': 2.6787171686057487e-05, 'flow_matching_loss': 0.01832523727789521, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.2} +{'loss': 0.017, 'grad_norm': 0.10449963063001633, 'learning_rate': 2.677693837583597e-05, 'flow_matching_loss': 0.01695553045719862, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.2} +{'loss': 0.0159, 'grad_norm': 0.08271070569753647, 'learning_rate': 2.6766690755523836e-05, 'flow_matching_loss': 0.015904280357062817, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.2} +{'loss': 0.0158, 'grad_norm': 0.10787206143140793, 'learning_rate': 2.675642883757291e-05, 'flow_matching_loss': 0.01575610777363181, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.21} +{'loss': 0.0166, 'grad_norm': 0.09200070798397064, 'learning_rate': 2.674615263445239e-05, 'flow_matching_loss': 0.016596919670701028, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.21} +{'loss': 0.0181, 'grad_norm': 0.0885273888707161, 'learning_rate': 2.673586215864882e-05, 'flow_matching_loss': 0.01805265499278903, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.21} +{'loss': 0.0165, 'grad_norm': 0.09317810833454132, 'learning_rate': 2.6725557422666113e-05, 'flow_matching_loss': 0.01647864207625389, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.21} +{'loss': 0.0158, 'grad_norm': 0.1048714816570282, 'learning_rate': 2.6715238439025493e-05, 'flow_matching_loss': 0.015808716975152494, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.22} +{'loss': 0.0166, 'grad_norm': 0.09508407860994339, 'learning_rate': 2.6704905220265493e-05, 'flow_matching_loss': 0.016592751350253822, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.22} +{'loss': 0.0172, 'grad_norm': 0.08996113389730453, 'learning_rate': 2.6694557778941947e-05, 'flow_matching_loss': 0.01720146369189024, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.22} +{'loss': 0.0167, 'grad_norm': 0.10598383098840714, 'learning_rate': 2.6684196127627977e-05, 'flow_matching_loss': 0.01671403869986534, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.23} +{'loss': 0.0151, 'grad_norm': 0.0661490336060524, 'learning_rate': 2.6673820278913963e-05, 'flow_matching_loss': 0.015097912773489952, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.23} +{'loss': 0.0152, 'grad_norm': 0.07215132564306259, 'learning_rate': 2.6663430245407544e-05, 'flow_matching_loss': 0.01522701745852828, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.23} +{'loss': 0.0172, 'grad_norm': 0.08673103898763657, 'learning_rate': 2.6653026039733585e-05, 'flow_matching_loss': 0.017201013490557672, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.23} +{'loss': 0.0173, 'grad_norm': 0.0706494078040123, 'learning_rate': 2.6642607674534176e-05, 'flow_matching_loss': 0.01728763971477747, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.24} +{'loss': 0.0174, 'grad_norm': 0.07108838111162186, 'learning_rate': 2.663217516246862e-05, 'flow_matching_loss': 0.017370521556586027, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.24} +{'loss': 0.0159, 'grad_norm': 0.10340288281440735, 'learning_rate': 2.66217285162134e-05, 'flow_matching_loss': 0.015931936725974084, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.24} +{'loss': 0.0173, 'grad_norm': 0.09005376696586609, 'learning_rate': 2.6611267748462175e-05, 'flow_matching_loss': 0.01726842364296317, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.25} +{'loss': 0.0158, 'grad_norm': 0.0758841335773468, 'learning_rate': 2.6600792871925764e-05, 'flow_matching_loss': 0.01579492585733533, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.25} +{'loss': 0.0156, 'grad_norm': 0.08643247187137604, 'learning_rate': 2.659030389933213e-05, 'flow_matching_loss': 0.015560926869511604, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.25} +{'loss': 0.0172, 'grad_norm': 0.09427540004253387, 'learning_rate': 2.6579800843426358e-05, 'flow_matching_loss': 0.017243430763483048, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.25} +{'loss': 0.0168, 'grad_norm': 0.07737462222576141, 'learning_rate': 2.656928371697067e-05, 'flow_matching_loss': 0.01680444134399295, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.26} +{'loss': 0.0166, 'grad_norm': 0.07463520020246506, 'learning_rate': 2.6558752532744346e-05, 'flow_matching_loss': 0.016614085622131826, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.26} +{'loss': 0.0179, 'grad_norm': 0.09467581659555435, 'learning_rate': 2.654820730354377e-05, 'flow_matching_loss': 0.017885309271514417, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.26} +{'loss': 0.0178, 'grad_norm': 0.12022972851991653, 'learning_rate': 2.6537648042182406e-05, 'flow_matching_loss': 0.017834890447556973, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.27} +{'loss': 0.016, 'grad_norm': 0.09629964083433151, 'learning_rate': 2.6527074761490735e-05, 'flow_matching_loss': 0.015965427830815317, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.27} +{'loss': 0.0152, 'grad_norm': 0.08877071738243103, 'learning_rate': 2.6516487474316303e-05, 'flow_matching_loss': 0.01523082023486495, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.27} +{'loss': 0.0151, 'grad_norm': 0.06224657595157623, 'learning_rate': 2.650588619352365e-05, 'flow_matching_loss': 0.015103575401008129, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.28} +{'loss': 0.016, 'grad_norm': 0.08615821599960327, 'learning_rate': 2.649527093199434e-05, 'flow_matching_loss': 0.015976173616945744, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.28} +{'loss': 0.0145, 'grad_norm': 0.06663165241479874, 'learning_rate': 2.6484641702626914e-05, 'flow_matching_loss': 0.014502124860882759, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.28} +{'loss': 0.016, 'grad_norm': 0.10475897789001465, 'learning_rate': 2.647399851833689e-05, 'flow_matching_loss': 0.016006383392959834, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.28} +{'loss': 0.0165, 'grad_norm': 0.08100198209285736, 'learning_rate': 2.6463341392056735e-05, 'flow_matching_loss': 0.016521460376679897, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.29} +{'loss': 0.0152, 'grad_norm': 0.0940919816493988, 'learning_rate': 2.6452670336735867e-05, 'flow_matching_loss': 0.015193694457411765, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.29} +{'loss': 0.0171, 'grad_norm': 0.08626744896173477, 'learning_rate': 2.644198536534063e-05, 'flow_matching_loss': 0.017112257797271012, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.29} +{'loss': 0.0169, 'grad_norm': 0.11797182261943817, 'learning_rate': 2.6431286490854255e-05, 'flow_matching_loss': 0.01693006381392479, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.3} +{'loss': 0.0181, 'grad_norm': 0.09922970831394196, 'learning_rate': 2.6420573726276903e-05, 'flow_matching_loss': 0.01807980826124549, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.3} +{'loss': 0.0169, 'grad_norm': 0.10104074329137802, 'learning_rate': 2.6409847084625575e-05, 'flow_matching_loss': 0.016918319277465343, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.3} +{'loss': 0.0159, 'grad_norm': 0.08861827105283737, 'learning_rate': 2.6399106578934167e-05, 'flow_matching_loss': 0.0159209243953228, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.3} +{'loss': 0.0182, 'grad_norm': 0.12202727049589157, 'learning_rate': 2.63883522222534e-05, 'flow_matching_loss': 0.018168943468481302, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.31} 26%|██▋ | 7930/30000 [5:04:57<13:49:34, 2.26s/it] 26%|██▋ | 7931/30000 [5:04:59<13:49:23, 2.25s/it] 26%|██▋ | 7932/30000 [5:05:01<13:54:48, 2.27s/it] 26%|██▋ | 7933/30000 [5:05:03<14:04:38, 2.30s/it] 26%|██▋ | 7934/30000 [5:05:06<14:09:55, 2.31s/it] 26%|██▋ | 7935/30000 [5:05:08<14:03:20, 2.29s/it] 26%|██▋ | 7936/30000 [5:05:10<13:58:41, 2.28s/it] 26%|██▋ | 7937/30000 [5:05:13<14:01:45, 2.29s/it] 26%|██▋ | 7938/30000 [5:05:15<13:56:41, 2.28s/it] 26%|██▋ | 7939/30000 [5:05:17<13:54:22, 2.27s/it] 26%|██▋ | 7940/30000 [5:05:19<14:02:36, 2.29s/it] 26%|██▋ | 7940/30000 [5:05:19<14:02:36, 2.29s/it] 26%|██▋ | 7941/30000 [5:05:22<14:11:27, 2.32s/it] 26%|██▋ | 7942/30000 [5:05:24<14:07:52, 2.31s/it] 26%|██▋ | 7943/30000 [5:05:26<14:09:27, 2.31s/it] 26%|██▋ | 7944/30000 [5:05:29<14:11:28, 2.32s/it] 26%|██▋ | 7945/30000 [5:05:31<14:07:14, 2.30s/it] 26%|██▋ | 7946/30000 [5:05:33<14:04:59, 2.30s/it] 26%|██▋ | 7947/30000 [5:05:36<14:01:45, 2.29s/it] 26%|██▋ | 7948/30000 [5:05:38<14:03:23, 2.29s/it] 26%|██▋ | 7949/30000 [5:05:40<14:04:43, 2.30s/it] 26%|██▋ | 7950/30000 [5:05:42<14:03:52, 2.30s/it] 26%|██▋ | 7950/30000 [5:05:42<14:03:52, 2.30s/it] 27%|██▋ | 7951/30000 [5:05:45<14:04:01, 2.30s/it] 27%|██▋ | 7952/30000 [5:05:47<14:10:54, 2.32s/it] 27%|██▋ | 7953/30000 [5:05:49<14:09:10, 2.31s/it] 27%|██▋ | 7954/30000 [5:05:52<14:05:49, 2.30s/it] 27%|██▋ | 7955/30000 [5:05:54<14:03:02, 2.29s/it] 27%|██▋ | 7956/30000 [5:05:56<14:03:30, 2.30s/it] 27%|██▋ | 7957/30000 [5:05:59<13:59:10, 2.28s/it] 27%|██▋ | 7958/30000 [5:06:01<13:59:02, 2.28s/it] 27%|██▋ | 7959/30000 [5:06:03<14:06:00, 2.30s/it] 27%|██▋ | 7960/30000 [5:06:05<14:06:08, 2.30s/it] 27%|██▋ | 7960/30000 [5:06:05<14:06:08, 2.30s/it] 27%|██▋ | 7961/30000 [5:06:08<14:03:45, 2.30s/it] 27%|██▋ | 7962/30000 [5:06:10<14:02:16, 2.29s/it] 27%|██▋ | 7963/30000 [5:06:12<14:02:52, 2.29s/it] 27%|██▋ | 7964/30000 [5:06:15<13:55:21, 2.27s/it] 27%|██▋ | 7965/30000 [5:06:17<13:58:04, 2.28s/it] 27%|██▋ | 7966/30000 [5:06:19<14:07:43, 2.31s/it] 27%|██▋ | 7967/30000 [5:06:22<14:09:37, 2.31s/it] 27%|██▋ | 7968/30000 [5:06:24<14:02:30, 2.29s/it] 27%|██▋ | 7969/30000 [5:06:26<14:08:45, 2.31s/it] 27%|██▋ | 7970/30000 [5:06:28<14:06:51, 2.31s/it] 27%|██▋ | 7970/30000 [5:06:28<14:06:51, 2.31s/it] 27%|██▋ | 7971/30000 [5:06:31<14:02:04, 2.29s/it] 27%|██▋ | 7972/30000 [5:06:33<14:02:01, 2.29s/it] 27%|██▋ | 7973/30000 [5:06:35<14:08:22, 2.31s/it] 27%|██▋ | 7974/30000 [5:06:38<14:07:24, 2.31s/it] 27%|██▋ | 7975/30000 [5:06:40<14:04:49, 2.30s/it] 27%|██▋ | 7976/30000 [5:06:42<14:08:26, 2.31s/it] 27%|██▋ | 7977/30000 [5:06:45<14:12:30, 2.32s/it] 27%|██▋ | 7978/30000 [5:06:47<14:06:06, 2.31s/it] 27%|██▋ | 7979/30000 [5:06:49<14:03:24, 2.30s/it] 27%|██▋ | 7980/30000 [5:06:51<14:04:32, 2.30s/it] 27%|██▋ | 7980/30000 [5:06:51<14:04:32, 2.30s/it] 27%|██▋ | 7981/30000 [5:06:54<14:09:41, 2.32s/it] 27%|██▋ | 7982/30000 [5:06:56<14:01:39, 2.29s/it] 27%|██▋ | 7983/30000 [5:06:58<14:06:35, 2.31s/it] 27%|██▋ | 7984/30000 [5:07:01<14:13:59, 2.33s/it] 27%|██▋ | 7985/30000 [5:07:03<14:09:20, 2.31s/it] 27%|██▋ | 7986/30000 [5:07:05<14:06:54, 2.31s/it] 27%|██▋ | 7987/30000 [5:07:08<14:10:13, 2.32s/it] 27%|██▋ | 7988/30000 [5:07:10<14:09:44, 2.32s/it] 27%|██▋ | 7989/30000 [5:07:12<14:03:05, 2.30s/it] 27%|██▋ | 7990/30000 [5:07:15<14:06:30, 2.31s/it] 27%|██▋ | 7990/30000 [5:07:15<14:06:30, 2.31s/it] 27%|██▋ | 7991/30000 [5:07:17<14:06:15, 2.31s/it] 27%|██▋ | 7992/30000 [5:07:19<14:01:30, 2.29s/it] 27%|██▋ | 7993/30000 [5:07:21<14:01:09, 2.29s/it] 27%|██▋ | 7994/30000 [5:07:24<14:05:52, 2.31s/it] 27%|██▋ | 7995/30000 [5:07:26<14:02:13, 2.30s/it] 27%|██▋ | 7996/30000 [5:07:28<13:56:39, 2.28s/it] 27%|██▋ | 7997/30000 [5:07:31<14:03:55, 2.30s/it] 27%|██▋ | 7998/30000 [5:07:33<14:09:14, 2.32s/it] 27%|██▋ | 7999/30000 [5:07:35<14:05:03, 2.30s/it] 27%|██▋ | 8000/30000 [5:07:38<14:00:37, 2.29s/it] 27%|██▋ | 8000/30000 [5:07:38<14:00:37, 2.29s/it] 27%|██▋ | 8001/30000 [5:07:40<14:07:55, 2.31s/it] 27%|██▋ | 8002/30000 [5:07:42<14:06:01, 2.31s/it] 27%|██▋ | 8003/30000 [5:07:44<14:02:29, 2.30s/it] 27%|██▋ | 8004/30000 [5:07:47<14:10:57, 2.32s/it] 27%|██▋ | 8005/30000 [5:07:49<14:11:05, 2.32s/it] 27%|██▋ | 8006/30000 [5:07:51<14:08:31, 2.31s/it] 27%|██▋ | 8007/30000 [5:07:54<14:06:30, 2.31s/it] 27%|██▋ | 8008/30000 [5:07:56<14:16:35, 2.34s/it] 27%|██▋ | 8009/30000 [5:07:58<14:16:00, 2.34s/it] 27%|██▋ | 8010/30000 [5:08:01<14:08:50, 2.32s/it] 27%|██▋ | 8010/30000 [5:08:01<14:08:50, 2.32s/it] 27%|██▋ | 8011/30000 [5:08:03<14:03:59, 2.30s/it] 27%|██▋ | 8012/30000 [5:08:05<14:09:58, 2.32s/it] 27%|██▋ | 8013/30000 [5:08:08<14:05:41, 2.31s/it] 27%|██▋ | 8014/30000 [5:08:10<14:02:14, 2.30s/it] 27%|██▋ | 8015/30000 [5:08:12<14:10:06, 2.32s/it] 27%|██▋ | 8016/30000 [5:08:15<14:17:22, 2.34s/it] 27%|██▋ | 8017/30000 [5:08:17<14:12:53, 2.33s/it] 27%|██▋ | 8018/30000 [5:08:19<14:04:55, 2.31s/it] 27%|██▋ | 8019/30000 [5:08:22<14:11:10, 2.32s/it] 27%|██▋ | 8020/30000 [5:08:24<14:06:13, 2.31s/it] 27%|██▋ | 8020/30000 [5:08:24<14:06:13, 2.31s/it] 27%|██▋ | 8021/30000 [5:08:26<13:59:23, 2.29s/it] 27%|██▋ | 8022/30000 [5:08:28<14:05:37, 2.31s/it] 27%|██▋ | 8023/30000 [5:08:31<14:05:30, 2.31s/it] 27%|██▋ | 8024/30000 [5:08:33<14:00:50, 2.30s/it] 27%|██▋ | 8025/30000 [5:08:35<14:00:53, 2.30s/it] 27%|██▋ | 8026/30000 [5:08:38<14:03:11, 2.30s/it] 27%|██▋ | 8027/30000 [5:08:40<14:01:47, 2.30s/it] 27%|██▋ | 8028/30000 [5:08:42<13:58:40, 2.29s/it] 27%|██▋ | 8029/30000 [5:08:45<13:57:31, 2.29s/it] 27%|██▋ | 8030/30000 [5:08:47<14:04:37, 2.31s/it] 27%|██▋ | 8030/30000 [5:08:47<14:04:37, 2.31s/it] 27%|██▋ | 8031/30000 [5:08:49<14:09:21, 2.32s/it] 27%|██▋ | 8032/30000 [5:08:52<14:07:37, 2.32s/it] 27%|██▋ | 8033/30000 [5:08:54<14:02:57, 2.30s/it] 27%|██▋ | 8034/30000 [5:08:56<14:06:20, 2.31s/it] 27%|██▋ | 8035/30000 [5:08:58<13:57:30, 2.29s/it] 27%|██▋ | 8036/30000 [5:09:01<13:53:59, 2.28s/it] 27%|██▋ | 8037/30000 [5:09:03<14:01:02, 2.30s/it] 27%|██▋ | 8038/30000 [5:09:05<14:05:15, 2.31s/it] 27%|██▋ | 8039/30000 [5:09:08<14:04:27, 2.31s/it] 27%|██▋ | 8040/30000 [5:09:10<14:06:33, 2.31s/it] 27%|██▋ | 8040/30000 [5:09:10<14:06:33, 2.31s/it] 27%|██▋ | 8041/30000 [5:09:12<14:04:18, 2.31s/it] 27%|██▋ | 8042/30000 [5:09:14<13:59:25, 2.29s/it] 27%|██▋ | 8043/30000 [5:09:17<13:54:09, 2.28s/it] 27%|██▋ | 8044/30000 [5:09:19<13:59:53, 2.30s/it] 27%|██▋ | 8045/30000 [5:09:21<14:05:34, 2.31s/it] 27%|██▋ | 8046/30000 [5:09:24<14:07:05, 2.32s/it] 27%|██▋ | 8047/30000 [5:09:26<14:04:06, 2.31s/it] 27%|██▋ | 8048/30000 [5:09:28<14:10:03, 2.32s/it] 27%|██▋ | 8049/30000 [5:09:31<14:18:00, 2.35s/it] 27%|██▋ | 8050/30000 [5:09:33<14:14:42, 2.34s/it] 27%|██▋ | 8050/30000 [5:09:33<14:14:42, 2.34s/it] 27%|██▋ | 8051/30000 [5:09:35<14:09:15, 2.32s/it] 27%|██▋ | 8052/30000 [5:09:38<14:11:20, 2.33s/it] 27%|██▋ | 8053/30000 [5:09:40<14:03:20, 2.31s/it] 27%|██▋ | 8054/30000 [5:09:42<14:02:19, 2.30s/it] 27%|██▋ | 8055/30000 [5:09:45<14:09:58, 2.32s/it] 27%|██▋ | 8056/30000 [5:09:47<14:08:25, 2.32s/it] 27%|██▋ | 8057/30000 [5:09:49<14:02:22, 2.30s/it] 27%|██▋ | 8058/30000 [5:09:51<13:53:25, 2.28s/it] 27%|██▋ | 8059/30000 [5:09:54<13:59:43, 2.30s/it] 27%|██▋ | 8060/30000 [5:09:56<14:04:31, 2.31s/it] 27%|██▋ | 8060/30000 [5:09:56<14:04:31, 2.31s/it] 27%|██▋ | 8061/30000 [5:09:58<14:03:55, 2.31s/it] 27%|██▋ | 8062/30000 [5:10:01<14:03:06, 2.31s/it] 27%|██▋ | 8063/30000 [5:10:03<14:01:24, 2.30s/it] 27%|██▋ | 8064/30000 [5:10:05<14:00:30, 2.30s/it] 27%|██▋ | 8065/30000 [5:10:08<14:03:40, 2.31s/it] 27%|██▋ | 8066/30000 [5:10:10<14:06:44, 2.32s/it] 27%|██▋ | 8067/30000 [5:10:12<14:04:15, 2.31s/it] 27%|██▋ | 8068/30000 [5:10:15<13:59:48, 2.30s/it] 27%|██▋ | 8069/30000 [5:10:17<13:59:43, 2.30s/it] 27%|██▋ | 8070/30000 [5:10:19<14:05:41, 2.31s/it] 27%|██▋ | 8070/30000 [5:10:19<14:05:41, 2.31s/it] 27%|██▋ | 8071/30000 [5:10:22<14:09:34, 2.32s/it] 27%|██▋ | 8072/30000 [5:10:24<14:09:12, 2.32s/it] 27%|██▋ | 8073/30000 [5:10:26<14:07:47, 2.32s/it] 27%|██▋ | 8074/30000 [5:10:28<13:05:09, 2.15s/it] 27%|██▋ | 8075/30000 [5:10:29<11:21:48, 1.87s/it] 27%|██▋ | 8076/30000 [5:10:30<10:07:44, 1.66s/it] 27%|██▋ | 8077/30000 [5:10:32<9:16:35, 1.52s/it] 27%|██▋ | 8078/30000 [5:10:33<8:42:34, 1.43s/it] 27%|██▋ | 8079/30000 [5:10:34<8:18:26, 1.36s/it] 27%|██▋ | 8080/30000 [5:10:36<9:32:03, 1.57s/it] 27%|██▋ | 8080/30000 [5:10:36<9:32:03, 1.57s/it] 27%|██▋ | 8081/30000 [5:10:38<10:55:19, 1.79s/it] 27%|██▋ | 8082/30000 [5:10:41<11:47:44, 1.94s/it] 27%|██▋ | 8083/30000 [5:10:43<12:24:12, 2.04s/it] 27%|██▋ | 8084/30000 [5:10:45<12:56:47, 2.13s/it] 27%|██▋ | 8085/30000 [5:10:47<13:15:45, 2.18s/it] 27%|██▋ | 8086/30000 [5:10:50<13:30:04, 2.22s/it] 27%|██▋ | 8087/30000 [5:10:52<13:44:38, 2.26s/it] 27%|██▋ | 8088/30000 [5:10:55<13:55:09, 2.29s/it] 27%|██▋ | 8089/30000 [5:10:57<13:54:04, 2.28s/it] 27%|██▋ | 8090/30000 [5:10:59<13:53:13, 2.28s/it] 27%|██▋ | 8090/30000 [5:10:59<13:53:13, 2.28s/it] 27%|██▋ | 8091/30000 [5:11:01<13:59:57, 2.30s/it] 27%|██▋ | 8092/30000 [5:11:04<14:03:19, 2.31s/it] 27%|██▋ | 8093/30000 [5:11:06<13:58:14, 2.30s/it] 27%|██▋ | 8094/30000 [5:11:08<14:05:10, 2.31s/it] 27%|██▋ | 8095/30000 [5:11:11<14:05:55, 2.32s/it] 27%|██▋ | 8096/30000 [5:11:13<14:01:15, 2.30s/it] 27%|██▋ | 8097/30000 [5:11:15<14:04:22, 2.31s/it] 27%|██▋ | 8098/30000 [5:11:18<14:08:26, 2.32s/it] 27%|██▋ | 8099/30000 [5:11:20<14:09:56, 2.33s/it] 27%|██▋ | 8100/30000 [5:11:22<14:02:50, 2.31s/it] 27%|██▋ | 8100/30000 [5:11:22<14:02:50, 2.31s/it] 27%|██▋ | 8101/30000 [5:11:25<14:00:47, 2.30s/it] 27%|██▋ | 8102/30000 [5:11:27<14:12:56, 2.34s/it] 27%|██▋ | 8103/30000 [5:11:29<14:12:49, 2.34s/it] 27%|██▋ | 8104/30000 [5:11:32<14:04:09, 2.31s/it] 27%|██▋ | 8105/30000 [5:11:34<14:03:40, 2.31s/it] 27%|██▋ | 8106/30000 [5:11:36<14:09:46, 2.33s/it] 27%|██▋ | 8107/30000 [5:11:39<14:11:38, 2.33s/it] 27%|██▋ | 8108/30000 [5:11:41<14:06:09, 2.32s/it] 27%|██▋ | 8109/30000 [5:11:43<14:00:11, 2.30s/it] 27%|██▋ | 8110/30000 [5:11:45<14:07:43, 2.32s/it] 27%|██▋ | 8110/30000 [5:11:46<14:07:43, 2.32s/it] 27%|██▋ | 8111/30000 [5:11:48<14:05:14, 2.32s/it] 27%|██▋ | 8112/30000 [5:11:50<14:02:09, 2.31s/it] 27%|██▋ | 8113/30000 [5:11:52<14:03:59, 2.31s/it] 27%|██▋ | 8114/30000 [5:11:55<14:03:17, 2.31s/it] 27%|██▋ | 8115/30000 [5:11:57<13:58:03, 2.30s/it] 27%|██▋ | 8116/30000 [5:11:59<13:54:52, 2.29s/it] 27%|██▋ | 8117/30000 [5:12:02<14:01:20, 2.31s/it] 27%|██▋ | 8118/30000 [5:12:04<14:06:24, 2.32s/it] 27%|██▋ | 8119/30000 [5:12:06<13:57:27, 2.30s/it] 27%|██▋ | 8120/30000 [5:12:09<14:04:15, 2.32s/it] 27%|██▋ | 8120/30000 [5:12:09<14:04:15, 2.32s/it] 27%|██▋ | 8121/30000 [5:12:11<14:02:22, 2.31s/it] 27%|██▋ | 8122/30000 [5:12:13<13:57:30, 2.30s/it] 27%|██▋ | 8123/30000 [5:12:15<13:56:14, 2.29s/it] 27%|██▋ | 8124/30000 [5:12:18<14:01:01, 2.31s/it] 27%|██▋ | 8125/30000 [5:12:20<13:55:09, 2.29s/it] 27%|██▋ | 8126/30000 [5:12:22<13:56:58, 2.30s/it] 27%|██▋ | 8127/30000 [5:12:25<14:07:05, 2.32s/it] 27%|██▋ | 8128/30000 [5:12:27<14:04:06, 2.32s/it] 27%|██▋ | 8129/30000 [5:12:29<13:58:39, 2.30s/it] 27%|██▋ | 8130/30000 [5:12:32<13:55:16, 2.29s/it] 27%|██▋ | 8130/30000 [5:12:32<13:55:16, 2.29s/it] 27%|██▋ | 8131/30000 [5:12:34<13:58:26, 2.30s/it] 27%|██▋ | 8132/30000 [5:12:36<13:56:51, 2.30s/it] 27%|██▋ | 8133/30000 [5:12:38<13:53:50, 2.29s/it] 27%|██▋ | 8134/30000 [5:12:41<13:55:45, 2.29s/it] 27%|██▋ | 8135/30000 [5:12:43<14:02:52, 2.31s/it] 27%|██▋ | 8136/30000 [5:12:45<14:03:25, 2.31s/it] 27%|██▋ | 8137/30000 [5:12:48<14:06:16, 2.32s/it] 27%|██▋ | 8138/30000 [5:12:50<14:02:53, 2.31s/it] 27%|██▋ | 8139/30000 [5:12:52<13:57:16, 2.30s/it] 27%|██▋ | 8140/30000 [5:12:55<13:55:47, 2.29s/it] 27%|██▋ | 8140/30000 [5:12:55<13:55:47, 2.29s/it] 27%|██▋ | 8141/30000 [5:12:57<14:01:03, 2.31s/it] 27%|██▋ | 8142/30000 [5:12:59<13:59:13, 2.30s/it] 27%|██▋ | 8143/30000 [5:13:01<13:57:52, 2.30s/it] 27%|██▋ | 8144/30000 [5:13:04<14:02:35, 2.31s/it] 27%|██▋ | 8145/30000 [5:13:06<14:04:54, 2.32s/it] 27%|██▋ | 8146/30000 [5:13:08<14:00:51, 2.31s/it] 27%|██▋ | 8147/30000 [5:13:11<13:59:46, 2.31s/it] 27%|██▋ | 8148/30000 [5:13:13<14:01:22, 2.31s/it] 27%|██▋ | 8149/30000 [5:13:15<13:55:05, 2.29s/it] 27%|██▋ | 8150/30000 [5:13:18<13:57:18, 2.30s/it] 27%|██▋ | 8150/30000 [5:13:18<13:57:18, 2.30s/it] 27%|██▋ | 8151/30000 [5:13:20<14:00:26, 2.31s/it] 27%|██▋ | 8152/30000 [5:13:22<14:03:08, 2.32s/it] 27%|██▋ | 8153/30000 [5:13:25<13:58:48, 2.30s/it] 27%|██▋ | 8154/30000 [5:13:27<14:02:53, 2.32s/it] 27%|██▋ | 8155/30000 [5:13:29<14:00:44, 2.31s/it] 27%|██▋ | 8156/30000 [5:13:31<13:58:16, 2.30s/it] 27%|██▋ | 8157/30000 [5:13:34<14:00:02, 2.31s/it] 27%|██▋ | 8158/30000 [5:13:36<13:56:24, 2.30s/it] 27%|██▋ | 8159/30000 [5:13:38<13:55:51, 2.30s/it] 27%|██▋ | 8160/30000 [5:13:41<13:54:32, 2.29s/it] 27%|██▋ | 8160/30000 [5:13:41<13:54:32, 2.29s/it] 27%|██▋ | 8161/30000 [5:13:43<13:59:16, 2.31s/it] 27%|██▋ | 8162/30000 [5:13:45<13:57:43, 2.30s/it] 27%|██▋ | 8163/30000 [5:13:48<13:52:44, 2.29s/it] 27%|██▋ | 8164/30000 [5:13:50<13:59:26, 2.31s/it] 27%|██▋ | 8165/30000 [5:13:52<14:01:34, 2.31s/it] 27%|██▋ | 8166/30000 [5:13:54<13:56:45, 2.30s/it] 27%|██▋ | 8167/30000 [5:13:57<14:02:09, 2.31s/it] 27%|██▋ | 8168/30000 [5:13:59<14:10:05, 2.34s/it] 27%|██▋ | 8169/30000 [5:14:02<14:04:54, 2.32s/it] 27%|██▋ | 8170/30000 [5:14:04<14:02:33, 2.32s/it] 27%|██▋ | 8170/30000 [5:14:04<14:02:33, 2.32s/it] 27%|██▋ | 8171/30000 [5:14:06<14:03:11, 2.32s/it] 27%|██▋ | 8172/30000 [5:14:08<13:58:21, 2.30s/it] 27%|██▋ | 8173/30000 [5:14:11<13:54:45, 2.29s/it] 27%|██▋ | 8174/30000 [5:14:13<14:04:31, 2.32s/it] 27%|██▋ | 8175/30000 [5:14:15<14:04:23, 2.32s/it] 27%|██▋ | 8176/30000 [5:14:18<14:00:58, 2.31s/it] 27%|██▋ | 8177/30000 [5:14:20<14:05:30, 2.32s/it] 27%|██▋ | 8178/30000 [5:14:22<14:00:13, 2.31s/it] 27%|██▋ | 8179/30000 [5:14:25<13:53:57, 2.29s/it] 27%|██▋ | 8180/30000 [5:14:27<13:53:14, 2.29s/it] 27%|██▋ | 8180/30000 [5:14:27<13:53:14, 2.29s/it] 27%|██▋ | 8181/30000 [5:14:29<13:56:04, 2.30s/it] 27%|██▋ | 8182/30000 [5:14:31<13:54:37, 2.30s/it] 27%|██▋ | 8183/30000 [5:14:34<13:52:33, 2.29s/it] 27%|██▋ | 8184/30000 [5:14:36<14:02:36, 2.32s/it] 27%|██▋ | 8185/30000 [5:14:38<14:05:41, 2.33s/it] 27%|██▋ | 8186/30000 [5:14:41<13:58:01, 2.31s/it] 27%|██▋ | 8187/30000 [5:14:43<14:05:44, 2.33s/it] 27%|██▋ | 8188/30000 [5:14:45<14:02:09, 2.32s/it] 27%|██▋ | 8189/30000 [5:14:48<13:59:11, 2.31s/it] 27%|██▋ | 8190/30000 [5:14:50<14:00:00, 2.31s/it] 27%|██▋ | 8190/30000 [5:14:50<14:00:00, 2.31s/it] 27%|██▋ | 8191/30000 [5:14:52<14:04:48, 2.32s/it] 27%|██▋ | 8192/30000 [5:14:55<14:03:52, 2.32s/it] 27%|██▋ | 8193/30000 [5:14:57<14:02:30, 2.32s/it] 27%|██▋ | 8194/30000 [5:14:59<14:07:58, 2.33s/it] 27%|██▋ | 8195/30000 [5:15:02<14:05:56, 2.33s/it] 27%|██▋ | 8196/30000 [5:15:04<14:01:16, 2.31s/it] 27%|██▋ | 8197/30000 [5:15:06<13:56:53, 2.30s/it] 27%|██▋ | 8198/30000 [5:15:09<13:57:14, 2.30s/it] 27%|██▋ | 8199/30000 [5:15:11<13:55:01, 2.30s/it] 27%|██▋ | 8200/30000 [5:15:13<13:55:37, 2.30s/it] 27%|██▋ | 8200/30000 [5:15:13<13:55:37, 2.30s/it] 27%|██▋ | 8201/30000 [5:15:15<14:05:35, 2.33s/it] 27%|██▋ | 8202/30000 [5:15:18<14:05:16, 2.33s/it] 27%|██▋ | 8203/30000 [5:15:20<13:59:11, 2.31s/it] 27%|██▋ | 8204/30000 [5:15:22<14:01:55, 2.32s/it] 27%|██▋ | 8205/30000 [5:15:25<14:02:55, 2.32s/it] 27%|██▋ | 8206/30000 [5:15:27<13:55:27, 2.30s/it] 27%|██▋ | 8207/30000 [5:15:29<13:57:29, 2.31s/it] 27%|██▋ | 8208/30000 [5:15:32<14:01:58, 2.32s/it] 27%|██▋ | 8209/30000 [5:15:34<13:59:43, 2.31s/it] 27%|██▋ | 8210/30000 [5:15:36<13:59:06, 2.31s/it] 27%|██▋ | 8210/30000 [5:15:36<13:59:06, 2.31s/it] 27%|██▋ | 8211/30000 [5:15:39<13:57:15, 2.31s/it] 27%|██▋ | 8212/30000 [5:15:41<13:55:30, 2.30s/it] 27%|██▋ | 8213/30000 [5:15:43<13:54:51, 2.30s/it] 27%|██▋ | 8214/30000 [5:15:45<13:57:15, 2.31s/it] 27%|██▋ | 8215/30000 [5:15:48<13:51:33, 2.29s/it] 27%|██▋ | 8216/30000 [5:15:50<13:48:42, 2.28s/it] 27%|██▋ | 8217/30000 [5:15:52<13:58:24, 2.31s/it] 27%|██▋ | 8218/30000 [5:15:55<13:58:17, 2.31s/it] 27%|██▋ | 8219/30000 [5:15:57<13:57:06, 2.31s/it] 27%|██▋ | 8220/30000 [5:15:59<14:01:20, 2.32s/it] 27%|██▋ | 8220/30000 [5:15:59<14:01:20, 2.32s/it] 27%|██▋ | 8221/30000 [5:16:02<13:57:22, 2.31s/it] 27%|██▋ | 8222/30000 [5:16:04<13:51:39, 2.29s/it] 27%|██▋ | 8223/30000 [5:16:06<13:47:40, 2.28s/it] 27%|██▋ | 8224/30000 [5:16:08<13:55:48, 2.30s/it] 27%|██▋ | 8225/30000 [5:16:11<13:51:11, 2.29s/it] 27%|██▋ | 8226/30000 [5:16:13<13:47:40, 2.28s/it] 27%|██▋ | 8227/30000 [5:16:15<13:55:00, 2.30s/it] 27%|██▋ | 8228/30000 [5:16:18<13:53:33, 2.30s/it] 27%|██▋ | 8229/30000 [5:16:20<13:52:11, 2.29s/it] 27%|██▋ | 8230/30000 [5:16:22<13:57:43, 2.31s/it] 27%|██▋ | 8230/30000 [5:16:22<13:57:43, 2.31s/it] 27%|██▋ | 8231/30000 [5:16:25<13:58:12, 2.31s/it] 27%|██▋ | 8232/30000 [5:16:27<13:54:06, 2.30s/it] 27%|██▋ | 8233/30000 [5:16:29<13:58:52, 2.31s/it] 27%|██▋ | 8234/30000 [5:16:31<13:56:01, 2.30s/it] 27%|██▋ | 8235/30000 [5:16:34<13:51:18, 2.29s/it] 27%|██▋ | 8236/30000 [5:16:36<13:50:57, 2.29s/it] 27%|██▋ | 8237/30000 [5:16:38<13:48:20, 2.28s/it] 27%|██▋ | 8238/30000 [5:16:41<13:47:23, 2.28s/it] 27%|██▋ | 8239/30000 [5:16:43<13:44:24, 2.27s/it] 27%|██▋ | 8240/30000 [5:16:45<13:52:54, 2.30s/it] 27%|██▋ | 8240/30000 [5:16:45<13:52:54, 2.30s/it] 27%|██▋ | 8241/30000 [5:16:47<13:55:19, 2.30s/it] 27%|██▋ | 8242/30000 [5:16:50<13:51:04, 2.29s/it] 27%|██▋ | 8243/30000 [5:16:52<13:54:46, 2.30s/it] 27%|██▋ | 8244/30000 [5:16:54<13:56:28, 2.31s/it] 27%|██▋ | 8245/30000 [5:16:57<13:49:17, 2.29s/it] 27%|██▋ | 8246/30000 [5:16:59<13:49:09, 2.29s/it] 27%|██▋ | 8247/30000 [5:17:01<13:46:59, 2.28s/it] 27%|██▋ | 8248/30000 [5:17:03<13:45:18, 2.28s/it] 27%|██▋ | 8249/30000 [5:17:06<13:45:53, 2.28s/it] 28%|██▊ | 8250/30000 [5:17:08<13:56:53, 2.31s/it] 28%|██▊ | 8250/30000 [5:17:08<13:56:53, 2.31s/it] 28%|██▊ | 8251/30000 [5:17:10<13:51:57, 2.30s/it] 28%|██▊ | 8252/30000 [5:17:13<13:52:20, 2.30s/it] 28%|██▊ | 8253/30000 [5:17:15<13:58:21, 2.31s/it] 28%|██▊ | 8254/30000 [5:17:17<13:52:10, 2.30s/it] 28%|██▊ | 8255/30000 [5:17:20<13:51:41, 2.29s/it] 28%|██▊ | 8256/30000 [5:17:22<13:52:07, 2.30s/it] 28%|██▊ | 8257/30000 [5:17:24<13:58:41, 2.31s/it] 28%|██▊ | 8258/30000 [5:17:27<13:55:00, 2.30s/it] 28%|██▊ | 8259/30000 [5:17:29<13:52:21, 2.30s/it] 28%|██▊ | 8260/30000 [5:17:31<13:53:22, 2.30s/it] 28%|██▊ | 8260/30000 [5:17:31<13:53:22, 2.30s/it] 28%|██▊ | 8261/30000 [5:17:33<13:52:00, 2.30s/it] 28%|██▊ | 8262/30000 [5:17:36<13:49:13, 2.29s/it] 28%|██▊ | 8263/30000 [5:17:38<13:48:55, 2.29s/it] 28%|██▊ | 8264/30000 [5:17:40<14:01:07, 2.32s/it] 28%|██▊ | 8265/30000 [5:17:43<14:00:41, 2.32s/it] 28%|██▊ | 8266/30000 [5:17:45<13:52:59, 2.30s/it] 28%|██▊ | 8267/30000 [5:17:47<13:50:38, 2.29s/it] 28%|██▊ | 8268/30000 [5:17:50<13:54:21, 2.30s/it] 28%|██▊ | 8269/30000 [5:17:52<13:56:57, 2.31s/it] 28%|██▊ | 8270/30000 [5:17:54<13:54:20, 2.30s/it] 28%|██▊ | 8270/30000 [5:17:54<13:54:20, 2.30s/it] 28%|██▊ | 8271/30000 [5:17:56<13:51:34, 2.30s/it] 28%|██▊ | 8272/30000 [5:17:59<13:59:51, 2.32s/it] 28%|██▊ | 8273/30000 [5:18:01<14:03:51, 2.33s/it] 28%|██▊ | 8274/30000 [5:18:03<13:48:45, 2.29s/it] 28%|██▊ | 8275/30000 [5:18:06<13:48:37, 2.29s/it] 28%|██▊ | 8276/30000 [5:18:08<13:48:32, 2.29s/it] 28%|██▊ | 8277/30000 [5:18:10<13:45:20, 2.28s/it] 28%|██▊ | 8278/30000 [5:18:12<13:41:53, 2.27s/it] 28%|██▊ | 8279/30000 [5:18:15<13:50:33, 2.29s/it] 28%|██▊ | 8280/30000 [5:18:17<13:55:49, 2.31s/it] 28%|██▊ | 8280/30000 [5:18:17<13:55:49, 2.31s/it] 28%|██▊ | 8281/30000 [5:18:19<13:51:23, 2.30s/it] 28%|██▊ | 8282/30000 [5:18:22<13:49:53, 2.29s/it] 28%|██▊ | 8283/30000 [5:18:24<13:56:48, 2.31s/it] 28%|██▊ | 8284/30000 [5:18:26<14:01:03, 2.32s/it] 28%|██▊ | 8285/30000 [5:18:29<13:52:09, 2.30s/it] 28%|██▊ | 8286/30000 [5:18:31<13:49:27, 2.29s/it] 28%|██▊ | 8287/30000 [5:18:33<13:53:57, 2.30s/it] 28%|██▊ | 8288/30000 [5:18:36<13:58:16, 2.32s/it] 28%|██▊ | 8289/30000 [5:18:38<13:54:44, 2.31s/it] 28%|██▊ | 8290/30000 [5:18:40<13:50:48, 2.30s/it] 28%|██▊ | 8290/30000 [5:18:40<13:50:48, 2.30s/it] 28%|██▊ | 8291/30000 [5:18:42<13:55:39, 2.31s/it] 28%|██▊ | 8292/30000 [5:18:45<13:50:58, 2.30s/it] 28%|██▊ | 8293/30000 [5:18:47<13:47:51, 2.29s/it] 28%|██▊ | 8294/30000 [5:18:49<13:49:10, 2.29s/it] 28%|██▊ | 8295/30000 [5:18:52<13:50:36, 2.30s/it] 28%|██▊ | 8296/30000 [5:18:54<13:46:48, 2.29s/it] 28%|██▊ | 8297/30000 [5:18:56<13:42:43, 2.27s/it] 28%|██▊ | 8298/30000 [5:18:58<13:48:32, 2.29s/it] 28%|██▊ | 8299/30000 [5:19:01<13:53:12, 2.30s/it] 28%|██▊ | 8300/30000 [5:19:03<13:48:06, 2.29s/it] 28%|██▊ | 8300/30000 [5:19:03<13:48:06, 2.29s/it] 28%|██▊ | 8301/30000 [5:19:05<13:46:14, 2.28s/it] 28%|██▊ | 8302/30000 [5:19:08<13:52:28, 2.30s/it] 28%|██▊ | 8303/30000 [5:19:10<13:57:32, 2.32s/it] 28%|██▊ | 8304/30000 [5:19:12<13:53:50, 2.31s/it] 28%|██▊ | 8305/30000 [5:19:15<13:55:06, 2.31s/it] 28%|██▊ | 8306/30000 [5:19:17<13:57:45, 2.32s/it] 28%|██▊ | 8307/30000 [5:19:19<13:55:54, 2.31s/it] 28%|██▊ | 8308/30000 [5:19:21<13:48:42, 2.29s/it] 28%|██▊ | 8309/30000 [5:19:24<13:44:44, 2.28s/it] 28%|██▊ | 8310/30000 [5:19:26<13:52:41, 2.30s/it] 28%|██▊ | 8310/30000 [5:19:26<13:52:41, 2.30s/it] 28%|██▊ | 8311/30000 [5:19:28<13:55:31, 2.31s/it] 28%|██▊ | 8312/30000 [5:19:31<13:51:33, 2.30s/it] 28%|██▊ | 8313/30000 [5:19:33<13:47:15, 2.29s/it] 28%|██▊ | 8314/30000 [5:19:35<13:51:49, 2.30s/it] 28%|██▊ | 8315/30000 [5:19:38<13:42:46, 2.28s/it] 28%|██▊ | 8316/30000 [5:19:40<13:39:16, 2.27s/it] 28%|██▊ | 8317/30000 [5:19:42<13:46:56, 2.29s/it] 28%|██▊ | 8318/30000 [5:19:44<13:48:39, 2.29s/it] 28%|██▊ | 8319/30000 [5:19:47<13:51:13, 2.30s/it] 28%|██▊ | 8320/30000 [5:19:49<13:47:36, 2.29s/it] 28%|██▊ | 8320/30000 [5:19:49<13:47:36, 2.29s/it] 28%|██▊ | 8321/30000 [5:19:51<13:53:04, 2.31s/it] 28%|██▊ | 8322/30000 [5:19:54<13:45:12, 2.28s/it] 28%|██▊ | 8323/30000 [5:19:56<13:42:35, 2.28s/it] 28%|██▊ | 8324/30000 [5:19:58<13:45:10, 2.28s/it] 28%|██▊ | 8325/30000 [5:20:00<13:39:58, 2.27s/it] 28%|██▊ | 8326/30000 [5:20:03<13:41:36, 2.27s/it] 28%|██▊ | 8327/30000 [5:20:05<13:44:29, 2.28s/it] 28%|██▊ | 8328/30000 [5:20:07<13:51:04, 2.30s/it] 28%|██▊ | 8329/30000 [5:20:10<13:55:23, 2.31s/it] 28%|██▊ | 8330/30000 [5:20:12<13:51:19, 2.30s/it] 28%|██▊ | 8330/30000 [5:20:12<13:51:19, 2.30s/it] 28%|██▊ | 8331/30000 [5:20:14<13:52:12, 2.30s/it] 28%|██▊ | 8332/30000 [5:20:17<13:55:22, 2.31s/it] 28%|██▊ | 8333/30000 [5:20:19<13:48:36, 2.29s/it] 28%|██▊ | 8334/30000 [5:20:21<13:47:10, 2.29s/it] 28%|██▊ | 8335/30000 [5:20:23<13:54:29, 2.31s/it] 28%|██▊ | 8336/30000 [5:20:26<13:52:13, 2.30s/it] 28%|██▊ | 8337/30000 [5:20:28<13:48:36, 2.30s/it] 28%|██▊ | 8338/30000 [5:20:30<13:50:08, 2.30s/it] 28%|██▊ | 8339/30000 [5:20:33<13:49:09, 2.30s/it] 28%|██▊ | 8340/30000 [5:20:35<13:45:40, 2.29s/it] +{'loss': 0.0158, 'grad_norm': 0.08724018931388855, 'learning_rate': 2.6377584027650827e-05, 'flow_matching_loss': 0.015838419552892447, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.31} +{'loss': 0.0173, 'grad_norm': 0.11001388728618622, 'learning_rate': 2.636680200821083e-05, 'flow_matching_loss': 0.017250462993979455, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.31} +{'loss': 0.0166, 'grad_norm': 0.10062267631292343, 'learning_rate': 2.635600617703457e-05, 'flow_matching_loss': 0.016607213579118252, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.32} +{'loss': 0.0165, 'grad_norm': 0.1028779000043869, 'learning_rate': 2.6345196547240007e-05, 'flow_matching_loss': 0.016539415810257196, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.32} +{'loss': 0.0161, 'grad_norm': 0.09629010409116745, 'learning_rate': 2.6334373131961854e-05, 'flow_matching_loss': 0.016135498136281966, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.32} +{'loss': 0.0159, 'grad_norm': 0.08179789781570435, 'learning_rate': 2.6323535944351583e-05, 'flow_matching_loss': 0.015930989291518927, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.32} +{'loss': 0.0152, 'grad_norm': 0.06850016117095947, 'learning_rate': 2.63126849975774e-05, 'flow_matching_loss': 0.01520235175266862, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.33} +{'loss': 0.0162, 'grad_norm': 0.08197231590747833, 'learning_rate': 2.6301820304824234e-05, 'flow_matching_loss': 0.01624268973246217, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.33} +{'loss': 0.0152, 'grad_norm': 0.10971414297819138, 'learning_rate': 2.6290941879293703e-05, 'flow_matching_loss': 0.015210281778126954, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.33} +{'loss': 0.0169, 'grad_norm': 0.09198158979415894, 'learning_rate': 2.628004973420412e-05, 'flow_matching_loss': 0.016861541103571652, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.34} +{'loss': 0.0172, 'grad_norm': 0.10308793932199478, 'learning_rate': 2.626914388279047e-05, 'flow_matching_loss': 0.017163799982517956, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.34} +{'loss': 0.0153, 'grad_norm': 0.07291965186595917, 'learning_rate': 2.6258224338304394e-05, 'flow_matching_loss': 0.015297060925513506, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.34} +{'loss': 0.0163, 'grad_norm': 0.08919917047023773, 'learning_rate': 2.6247291114014166e-05, 'flow_matching_loss': 0.016296430490911007, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.35} +{'loss': 0.0157, 'grad_norm': 0.09717602282762527, 'learning_rate': 2.6236344223204683e-05, 'flow_matching_loss': 0.015698900818824767, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.35} +{'loss': 0.0168, 'grad_norm': 0.08262206614017487, 'learning_rate': 2.6225383679177453e-05, 'flow_matching_loss': 0.016794399172067643, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.35} +{'loss': 0.016, 'grad_norm': 0.07528749853372574, 'learning_rate': 2.6214409495250568e-05, 'flow_matching_loss': 0.015983247291296722, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.35} +{'loss': 0.0195, 'grad_norm': 0.08688241988420486, 'learning_rate': 2.620342168475869e-05, 'flow_matching_loss': 0.01953901518136263, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.36} +{'loss': 0.0177, 'grad_norm': 0.10984530299901962, 'learning_rate': 2.619242026105306e-05, 'flow_matching_loss': 0.017705923598259687, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.36} +{'loss': 0.0158, 'grad_norm': 0.0604446604847908, 'learning_rate': 2.618140523750143e-05, 'flow_matching_loss': 0.015780732687562703, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.36} +{'loss': 0.0167, 'grad_norm': 0.07224846631288528, 'learning_rate': 2.6170376627488103e-05, 'flow_matching_loss': 0.0166504867374897, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.37} +{'loss': 0.0171, 'grad_norm': 0.09462336450815201, 'learning_rate': 2.6159334444413873e-05, 'flow_matching_loss': 0.0170687647536397, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.37} +{'loss': 0.0188, 'grad_norm': 0.10542982071638107, 'learning_rate': 2.614827870169604e-05, 'flow_matching_loss': 0.01878861226141453, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.37} +{'loss': 0.0148, 'grad_norm': 0.09903089702129364, 'learning_rate': 2.6137209412768367e-05, 'flow_matching_loss': 0.014758850354701281, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.37} +{'loss': 0.0176, 'grad_norm': 0.10778328776359558, 'learning_rate': 2.612612659108109e-05, 'flow_matching_loss': 0.017610883247107266, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.38} +{'loss': 0.0162, 'grad_norm': 0.10709578543901443, 'learning_rate': 2.611503025010088e-05, 'flow_matching_loss': 0.01624194998294115, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.38} +{'loss': 0.0157, 'grad_norm': 0.07311269640922546, 'learning_rate': 2.6103920403310832e-05, 'flow_matching_loss': 0.01568342950195074, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.38} +{'loss': 0.016, 'grad_norm': 0.07027493417263031, 'learning_rate': 2.6092797064210465e-05, 'flow_matching_loss': 0.01595008047297597, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.39} +{'loss': 0.0171, 'grad_norm': 0.08756469935178757, 'learning_rate': 2.6081660246315685e-05, 'flow_matching_loss': 0.017098479252308608, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.39} +{'loss': 0.0152, 'grad_norm': 0.07350801676511765, 'learning_rate': 2.6070509963158774e-05, 'flow_matching_loss': 0.015159698110073805, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.39} +{'loss': 0.0153, 'grad_norm': 0.0750538781285286, 'learning_rate': 2.6059346228288375e-05, 'flow_matching_loss': 0.015296839829534292, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.39} +{'loss': 0.0163, 'grad_norm': 0.07158420979976654, 'learning_rate': 2.604816905526948e-05, 'flow_matching_loss': 0.016342896874994038, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.4} +{'loss': 0.0155, 'grad_norm': 0.06751872599124908, 'learning_rate': 2.603697845768341e-05, 'flow_matching_loss': 0.015469567757099867, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.4} +{'loss': 0.0157, 'grad_norm': 0.08411609381437302, 'learning_rate': 2.6025774449127792e-05, 'flow_matching_loss': 0.01574276462197304, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.4} +{'loss': 0.0156, 'grad_norm': 0.088570237159729, 'learning_rate': 2.6014557043216562e-05, 'flow_matching_loss': 0.015600107423961163, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.41} +{'loss': 0.0152, 'grad_norm': 0.09559714794158936, 'learning_rate': 2.6003326253579913e-05, 'flow_matching_loss': 0.015241039637476205, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.41} +{'loss': 0.016, 'grad_norm': 0.09286770224571228, 'learning_rate': 2.599208209386432e-05, 'flow_matching_loss': 0.016027992777526378, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.41} +{'loss': 0.0173, 'grad_norm': 0.12336403876543045, 'learning_rate': 2.5980824577732502e-05, 'flow_matching_loss': 0.01734726456925273, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.41} +{'loss': 0.0151, 'grad_norm': 0.11323237419128418, 'learning_rate': 2.5969553718863395e-05, 'flow_matching_loss': 0.015072564221918583, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.42} +{'loss': 0.0174, 'grad_norm': 0.0942811444401741, 'learning_rate': 2.595826953095216e-05, 'flow_matching_loss': 0.0173687182366848, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.42} +{'loss': 0.016, 'grad_norm': 0.0802612379193306, 'learning_rate': 2.594697202771014e-05, 'flow_matching_loss': 0.015977287106215952, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.42} +{'loss': 0.0166, 'grad_norm': 0.07412584125995636, 'learning_rate': 2.5935661222864888e-05, 'flow_matching_loss': 0.016625703126192094, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.43} 28%|██▊ | 8340/30000 [5:20:35<13:45:40, 2.29s/it] 28%|██▊ | 8341/30000 [5:20:37<13:46:52, 2.29s/it] 28%|██▊ | 8342/30000 [5:20:39<13:50:38, 2.30s/it] 28%|██▊ | 8343/30000 [5:20:42<13:46:21, 2.29s/it] 28%|██▊ | 8344/30000 [5:20:44<13:39:44, 2.27s/it] 28%|██▊ | 8345/30000 [5:20:46<13:48:58, 2.30s/it] 28%|██▊ | 8346/30000 [5:20:49<13:50:44, 2.30s/it] 28%|██▊ | 8347/30000 [5:20:51<13:46:05, 2.29s/it] 28%|██▊ | 8348/30000 [5:20:53<13:50:11, 2.30s/it] 28%|██▊ | 8349/30000 [5:20:56<13:49:08, 2.30s/it] 28%|██▊ | 8350/30000 [5:20:58<13:45:03, 2.29s/it] 28%|██▊ | 8350/30000 [5:20:58<13:45:03, 2.29s/it] 28%|██▊ | 8351/30000 [5:21:00<13:47:20, 2.29s/it] 28%|██▊ | 8352/30000 [5:21:02<13:49:12, 2.30s/it] 28%|██▊ | 8353/30000 [5:21:05<13:48:53, 2.30s/it] 28%|██▊ | 8354/30000 [5:21:07<13:47:33, 2.29s/it] 28%|██▊ | 8355/30000 [5:21:09<13:50:25, 2.30s/it] 28%|██▊ | 8356/30000 [5:21:12<13:47:46, 2.29s/it] 28%|██▊ | 8357/30000 [5:21:14<13:48:06, 2.30s/it] 28%|██▊ | 8358/30000 [5:21:16<13:52:10, 2.31s/it] 28%|██▊ | 8359/30000 [5:21:19<13:50:55, 2.30s/it] 28%|██▊ | 8360/30000 [5:21:21<13:48:21, 2.30s/it] 28%|██▊ | 8360/30000 [5:21:21<13:48:21, 2.30s/it] 28%|██▊ | 8361/30000 [5:21:23<13:53:54, 2.31s/it] 28%|██▊ | 8362/30000 [5:21:25<13:53:20, 2.31s/it] 28%|██▊ | 8363/30000 [5:21:28<13:49:08, 2.30s/it] 28%|██▊ | 8364/30000 [5:21:30<13:46:27, 2.29s/it] 28%|██▊ | 8365/30000 [5:21:32<13:52:35, 2.31s/it] 28%|██▊ | 8366/30000 [5:21:35<13:53:03, 2.31s/it] 28%|██▊ | 8367/30000 [5:21:37<13:52:39, 2.31s/it] 28%|██▊ | 8368/30000 [5:21:39<13:58:42, 2.33s/it] 28%|██▊ | 8369/30000 [5:21:42<13:59:09, 2.33s/it] 28%|██▊ | 8370/30000 [5:21:44<13:53:55, 2.31s/it] 28%|██▊ | 8370/30000 [5:21:44<13:53:55, 2.31s/it] 28%|██▊ | 8371/30000 [5:21:46<13:49:51, 2.30s/it] 28%|██▊ | 8372/30000 [5:21:49<13:54:00, 2.31s/it] 28%|██▊ | 8373/30000 [5:21:51<13:53:14, 2.31s/it] 28%|██▊ | 8374/30000 [5:21:53<13:49:48, 2.30s/it] 28%|██▊ | 8375/30000 [5:21:56<13:57:20, 2.32s/it] 28%|██▊ | 8376/30000 [5:21:58<14:01:01, 2.33s/it] 28%|██▊ | 8377/30000 [5:22:00<13:55:47, 2.32s/it] 28%|██▊ | 8378/30000 [5:22:02<13:53:27, 2.31s/it] 28%|██▊ | 8379/30000 [5:22:05<13:57:44, 2.32s/it] 28%|██▊ | 8380/30000 [5:22:07<13:59:17, 2.33s/it] 28%|██▊ | 8380/30000 [5:22:07<13:59:17, 2.33s/it] 28%|██▊ | 8381/30000 [5:22:09<13:54:37, 2.32s/it] 28%|██▊ | 8382/30000 [5:22:12<13:50:13, 2.30s/it] 28%|██▊ | 8383/30000 [5:22:14<13:55:39, 2.32s/it] 28%|██▊ | 8384/30000 [5:22:16<13:58:21, 2.33s/it] 28%|██▊ | 8385/30000 [5:22:19<13:49:32, 2.30s/it] 28%|██▊ | 8386/30000 [5:22:21<13:53:50, 2.31s/it] 28%|██▊ | 8387/30000 [5:22:23<13:50:53, 2.31s/it] 28%|██▊ | 8388/30000 [5:22:26<13:47:07, 2.30s/it] 28%|██▊ | 8389/30000 [5:22:28<13:50:20, 2.31s/it] 28%|██▊ | 8390/30000 [5:22:30<13:55:14, 2.32s/it] 28%|██▊ | 8390/30000 [5:22:30<13:55:14, 2.32s/it] 28%|██▊ | 8391/30000 [5:22:33<13:48:36, 2.30s/it] 28%|██▊ | 8392/30000 [5:22:35<13:47:22, 2.30s/it] 28%|██▊ | 8393/30000 [5:22:37<13:55:13, 2.32s/it] 28%|██▊ | 8394/30000 [5:22:39<13:51:27, 2.31s/it] 28%|██▊ | 8395/30000 [5:22:42<13:46:45, 2.30s/it] 28%|██▊ | 8396/30000 [5:22:44<13:50:16, 2.31s/it] 28%|██▊ | 8397/30000 [5:22:46<13:47:33, 2.30s/it] 28%|██▊ | 8398/30000 [5:22:49<13:44:23, 2.29s/it] 28%|██▊ | 8399/30000 [5:22:51<13:40:23, 2.28s/it] 28%|██▊ | 8400/30000 [5:22:53<13:48:29, 2.30s/it] 28%|██▊ | 8400/30000 [5:22:53<13:48:29, 2.30s/it] 28%|██▊ | 8401/30000 [5:22:55<13:46:52, 2.30s/it] 28%|██▊ | 8402/30000 [5:22:58<13:45:14, 2.29s/it] 28%|██▊ | 8403/30000 [5:23:00<13:52:57, 2.31s/it] 28%|██▊ | 8404/30000 [5:23:02<13:54:25, 2.32s/it] 28%|██▊ | 8405/30000 [5:23:05<13:50:52, 2.31s/it] 28%|██▊ | 8406/30000 [5:23:07<13:52:39, 2.31s/it] 28%|██▊ | 8407/30000 [5:23:09<13:51:47, 2.31s/it] 28%|██▊ | 8408/30000 [5:23:12<13:47:36, 2.30s/it] 28%|██▊ | 8409/30000 [5:23:14<13:44:44, 2.29s/it] 28%|██▊ | 8410/30000 [5:23:16<13:45:44, 2.29s/it] 28%|██▊ | 8410/30000 [5:23:16<13:45:44, 2.29s/it] 28%|██▊ | 8411/30000 [5:23:18<13:39:28, 2.28s/it] 28%|██▊ | 8412/30000 [5:23:21<13:37:06, 2.27s/it] 28%|██▊ | 8413/30000 [5:23:23<13:39:32, 2.28s/it] 28%|██▊ | 8414/30000 [5:23:25<13:40:08, 2.28s/it] 28%|██▊ | 8415/30000 [5:23:28<13:42:00, 2.28s/it] 28%|██▊ | 8416/30000 [5:23:30<13:49:53, 2.31s/it] 28%|██▊ | 8417/30000 [5:23:32<13:54:43, 2.32s/it] 28%|██▊ | 8418/30000 [5:23:35<13:49:16, 2.31s/it] 28%|██▊ | 8419/30000 [5:23:37<13:54:59, 2.32s/it] 28%|██▊ | 8420/30000 [5:23:39<13:52:37, 2.32s/it] 28%|██▊ | 8420/30000 [5:23:39<13:52:37, 2.32s/it] 28%|██▊ | 8421/30000 [5:23:42<13:48:43, 2.30s/it] 28%|██▊ | 8422/30000 [5:23:44<13:52:16, 2.31s/it] 28%|██▊ | 8423/30000 [5:23:46<13:54:28, 2.32s/it] 28%|██▊ | 8424/30000 [5:23:48<13:49:10, 2.31s/it] 28%|██▊ | 8425/30000 [5:23:51<13:46:33, 2.30s/it] 28%|██▊ | 8426/30000 [5:23:53<13:51:11, 2.31s/it] 28%|██▊ | 8427/30000 [5:23:55<13:47:58, 2.30s/it] 28%|██▊ | 8428/30000 [5:23:58<13:46:58, 2.30s/it] 28%|██▊ | 8429/30000 [5:24:00<13:53:51, 2.32s/it] 28%|██▊ | 8430/30000 [5:24:02<13:53:44, 2.32s/it] 28%|██▊ | 8430/30000 [5:24:02<13:53:44, 2.32s/it] 28%|██▊ | 8431/30000 [5:24:05<13:50:54, 2.31s/it] 28%|██▊ | 8432/30000 [5:24:07<13:48:07, 2.30s/it] 28%|██▊ | 8433/30000 [5:24:09<13:54:59, 2.32s/it] 28%|██▊ | 8434/30000 [5:24:12<13:46:48, 2.30s/it] 28%|██▊ | 8435/30000 [5:24:14<13:43:06, 2.29s/it] 28%|██▊ | 8436/30000 [5:24:16<13:48:10, 2.30s/it] 28%|██▊ | 8437/30000 [5:24:18<13:52:44, 2.32s/it] 28%|██▊ | 8438/30000 [5:24:21<13:49:02, 2.31s/it] 28%|██▊ | 8439/30000 [5:24:23<13:53:24, 2.32s/it] 28%|██▊ | 8440/30000 [5:24:25<13:53:57, 2.32s/it] 28%|██▊ | 8440/30000 [5:24:25<13:53:57, 2.32s/it] 28%|██▊ | 8441/30000 [5:24:28<13:50:31, 2.31s/it] 28%|██▊ | 8442/30000 [5:24:30<13:49:48, 2.31s/it] 28%|██▊ | 8443/30000 [5:24:32<13:44:20, 2.29s/it] 28%|██▊ | 8444/30000 [5:24:35<13:38:16, 2.28s/it] 28%|██▊ | 8445/30000 [5:24:37<13:39:12, 2.28s/it] 28%|██▊ | 8446/30000 [5:24:39<13:45:17, 2.30s/it] 28%|██▊ | 8447/30000 [5:24:41<13:46:25, 2.30s/it] 28%|██▊ | 8448/30000 [5:24:44<13:42:04, 2.29s/it] 28%|██▊ | 8449/30000 [5:24:46<13:49:38, 2.31s/it] 28%|██▊ | 8450/30000 [5:24:48<13:47:33, 2.30s/it] 28%|██▊ | 8450/30000 [5:24:48<13:47:33, 2.30s/it] 28%|██▊ | 8451/30000 [5:24:51<13:43:16, 2.29s/it] 28%|██▊ | 8452/30000 [5:24:53<13:43:22, 2.29s/it] 28%|██▊ | 8453/30000 [5:24:55<13:45:44, 2.30s/it] 28%|██▊ | 8454/30000 [5:24:57<13:40:05, 2.28s/it] 28%|██▊ | 8455/30000 [5:25:00<13:49:33, 2.31s/it] 28%|██▊ | 8456/30000 [5:25:02<13:48:51, 2.31s/it] 28%|██▊ | 8457/30000 [5:25:04<13:45:43, 2.30s/it] 28%|██▊ | 8458/30000 [5:25:07<13:50:48, 2.31s/it] 28%|██▊ | 8459/30000 [5:25:09<13:51:51, 2.32s/it] 28%|██▊ | 8460/30000 [5:25:11<13:46:54, 2.30s/it] 28%|██▊ | 8460/30000 [5:25:11<13:46:54, 2.30s/it] 28%|██▊ | 8461/30000 [5:25:14<13:46:57, 2.30s/it] 28%|██▊ | 8462/30000 [5:25:16<13:46:11, 2.30s/it] 28%|██▊ | 8463/30000 [5:25:18<13:43:11, 2.29s/it] 28%|██▊ | 8464/30000 [5:25:21<13:42:07, 2.29s/it] 28%|██▊ | 8465/30000 [5:25:23<13:49:34, 2.31s/it] 28%|██▊ | 8466/30000 [5:25:25<13:44:56, 2.30s/it] 28%|██▊ | 8467/30000 [5:25:27<13:41:48, 2.29s/it] 28%|██▊ | 8468/30000 [5:25:30<13:44:17, 2.30s/it] 28%|██▊ | 8469/30000 [5:25:32<13:44:25, 2.30s/it] 28%|██▊ | 8470/30000 [5:25:34<13:41:30, 2.29s/it] 28%|██▊ | 8470/30000 [5:25:34<13:41:30, 2.29s/it] 28%|██▊ | 8471/30000 [5:25:37<13:49:26, 2.31s/it] 28%|██▊ | 8472/30000 [5:25:39<13:50:22, 2.31s/it] 28%|██▊ | 8473/30000 [5:25:41<13:45:53, 2.30s/it] 28%|██▊ | 8474/30000 [5:25:44<13:42:20, 2.29s/it] 28%|██▊ | 8475/30000 [5:25:46<13:43:22, 2.30s/it] 28%|██▊ | 8476/30000 [5:25:48<13:39:16, 2.28s/it] 28%|██▊ | 8477/30000 [5:25:50<13:36:29, 2.28s/it] 28%|██▊ | 8478/30000 [5:25:53<13:45:05, 2.30s/it] 28%|██▊ | 8479/30000 [5:25:55<13:46:35, 2.30s/it] 28%|██▊ | 8480/30000 [5:25:57<13:45:51, 2.30s/it] 28%|██▊ | 8480/30000 [5:25:57<13:45:51, 2.30s/it] 28%|██▊ | 8481/30000 [5:26:00<13:52:48, 2.32s/it] 28%|██▊ | 8482/30000 [5:26:02<13:54:03, 2.33s/it] 28%|██▊ | 8483/30000 [5:26:04<13:48:02, 2.31s/it] 28%|██▊ | 8484/30000 [5:26:07<13:45:46, 2.30s/it] 28%|██▊ | 8485/30000 [5:26:09<13:48:05, 2.31s/it] 28%|██▊ | 8486/30000 [5:26:11<13:41:22, 2.29s/it] 28%|██▊ | 8487/30000 [5:26:13<13:37:46, 2.28s/it] 28%|██▊ | 8488/30000 [5:26:16<13:47:00, 2.31s/it] 28%|██▊ | 8489/30000 [5:26:18<13:42:42, 2.29s/it] 28%|██▊ | 8490/30000 [5:26:20<13:39:28, 2.29s/it] 28%|██▊ | 8490/30000 [5:26:20<13:39:28, 2.29s/it] 28%|██▊ | 8491/30000 [5:26:23<13:44:08, 2.30s/it] 28%|██▊ | 8492/30000 [5:26:25<13:42:34, 2.29s/it] 28%|██▊ | 8493/30000 [5:26:27<13:39:23, 2.29s/it] 28%|██▊ | 8494/30000 [5:26:29<13:33:59, 2.27s/it] 28%|██▊ | 8495/30000 [5:26:32<13:43:19, 2.30s/it] 28%|██▊ | 8496/30000 [5:26:34<13:46:41, 2.31s/it] 28%|██▊ | 8497/30000 [5:26:36<13:41:57, 2.29s/it] 28%|██▊ | 8498/30000 [5:26:39<13:41:37, 2.29s/it] 28%|██▊ | 8499/30000 [5:26:41<13:42:36, 2.30s/it] 28%|██▊ | 8500/30000 [5:26:43<13:39:23, 2.29s/it] 28%|██▊ | 8500/30000 [5:26:43<13:39:23, 2.29s/it] 28%|██▊ | 8501/30000 [5:26:46<13:38:59, 2.29s/it] 28%|██▊ | 8502/30000 [5:26:48<13:44:54, 2.30s/it] 28%|██▊ | 8503/30000 [5:26:50<13:43:40, 2.30s/it] 28%|██▊ | 8504/30000 [5:26:52<13:39:51, 2.29s/it] 28%|██▊ | 8505/30000 [5:26:55<13:38:07, 2.28s/it] 28%|██▊ | 8506/30000 [5:26:57<13:45:28, 2.30s/it] 28%|██▊ | 8507/30000 [5:26:59<13:45:40, 2.30s/it] 28%|██▊ | 8508/30000 [5:27:02<13:42:22, 2.30s/it] 28%|██▊ | 8509/30000 [5:27:04<13:38:34, 2.29s/it] 28%|██▊ | 8510/30000 [5:27:06<13:43:52, 2.30s/it] 28%|██▊ | 8510/30000 [5:27:06<13:43:52, 2.30s/it] 28%|██▊ | 8511/30000 [5:27:09<13:48:20, 2.31s/it] 28%|██▊ | 8512/30000 [5:27:11<13:44:59, 2.30s/it] 28%|██▊ | 8513/30000 [5:27:13<13:42:32, 2.30s/it] 28%|██▊ | 8514/30000 [5:27:15<13:43:48, 2.30s/it] 28%|██▊ | 8515/30000 [5:27:18<13:41:04, 2.29s/it] 28%|██▊ | 8516/30000 [5:27:20<13:39:00, 2.29s/it] 28%|██▊ | 8517/30000 [5:27:22<13:36:43, 2.28s/it] 28%|██▊ | 8518/30000 [5:27:25<13:42:33, 2.30s/it] 28%|██▊ | 8519/30000 [5:27:27<13:44:18, 2.30s/it] 28%|██▊ | 8520/30000 [5:27:29<13:41:26, 2.29s/it] 28%|██▊ | 8520/30000 [5:27:29<13:41:26, 2.29s/it] 28%|██▊ | 8521/30000 [5:27:31<13:40:16, 2.29s/it] 28%|██▊ | 8522/30000 [5:27:34<13:43:11, 2.30s/it] 28%|██▊ | 8523/30000 [5:27:36<13:41:47, 2.30s/it] 28%|██▊ | 8524/30000 [5:27:38<13:38:20, 2.29s/it] 28%|██▊ | 8525/30000 [5:27:41<13:39:39, 2.29s/it] 28%|██▊ | 8526/30000 [5:27:43<13:39:49, 2.29s/it] 28%|██▊ | 8527/30000 [5:27:45<13:40:40, 2.29s/it] 28%|██▊ | 8528/30000 [5:27:48<13:40:14, 2.29s/it] 28%|██▊ | 8529/30000 [5:27:50<13:44:05, 2.30s/it] 28%|██▊ | 8530/30000 [5:27:52<13:43:55, 2.30s/it] 28%|██▊ | 8530/30000 [5:27:52<13:43:55, 2.30s/it] 28%|██▊ | 8531/30000 [5:27:54<13:40:28, 2.29s/it] 28%|██▊ | 8532/30000 [5:27:57<13:39:08, 2.29s/it] 28%|██▊ | 8533/30000 [5:27:59<13:42:20, 2.30s/it] 28%|██▊ | 8534/30000 [5:28:01<13:37:50, 2.29s/it] 28%|██▊ | 8535/30000 [5:28:04<13:34:10, 2.28s/it] 28%|██▊ | 8536/30000 [5:28:06<13:35:02, 2.28s/it] 28%|██▊ | 8537/30000 [5:28:08<13:39:53, 2.29s/it] 28%|██▊ | 8538/30000 [5:28:10<13:37:01, 2.28s/it] 28%|██▊ | 8539/30000 [5:28:13<13:35:21, 2.28s/it] 28%|██▊ | 8540/30000 [5:28:15<13:35:05, 2.28s/it] 28%|██▊ | 8540/30000 [5:28:15<13:35:05, 2.28s/it] 28%|██▊ | 8541/30000 [5:28:17<13:40:11, 2.29s/it] 28%|██▊ | 8542/30000 [5:28:20<13:38:18, 2.29s/it] 28%|██▊ | 8543/30000 [5:28:22<13:39:57, 2.29s/it] 28%|██▊ | 8544/30000 [5:28:24<13:48:15, 2.32s/it] 28%|██▊ | 8545/30000 [5:28:27<13:51:09, 2.32s/it] 28%|██▊ | 8546/30000 [5:28:29<13:46:05, 2.31s/it] 28%|██▊ | 8547/30000 [5:28:31<13:42:11, 2.30s/it] 28%|██▊ | 8548/30000 [5:28:33<13:48:11, 2.32s/it] 28%|██▊ | 8549/30000 [5:28:36<13:44:56, 2.31s/it] 28%|██▊ | 8550/30000 [5:28:38<13:39:55, 2.29s/it] 28%|██▊ | 8550/30000 [5:28:38<13:39:55, 2.29s/it] 29%|██▊ | 8551/30000 [5:28:40<13:35:53, 2.28s/it] 29%|██▊ | 8552/30000 [5:28:43<13:44:45, 2.31s/it] 29%|██▊ | 8553/30000 [5:28:45<13:48:24, 2.32s/it] 29%|██▊ | 8554/30000 [5:28:47<13:40:48, 2.30s/it] 29%|██▊ | 8555/30000 [5:28:50<13:40:58, 2.30s/it] 29%|██▊ | 8556/30000 [5:28:52<13:38:49, 2.29s/it] 29%|██▊ | 8557/30000 [5:28:54<13:35:46, 2.28s/it] 29%|██▊ | 8558/30000 [5:28:56<13:28:20, 2.26s/it] 29%|██▊ | 8559/30000 [5:28:59<13:38:44, 2.29s/it] 29%|██▊ | 8560/30000 [5:29:01<13:33:10, 2.28s/it] 29%|██▊ | 8560/30000 [5:29:01<13:33:10, 2.28s/it] 29%|██▊ | 8561/30000 [5:29:03<13:33:45, 2.28s/it] 29%|██▊ | 8562/30000 [5:29:06<13:42:28, 2.30s/it] 29%|██▊ | 8563/30000 [5:29:08<13:43:31, 2.30s/it] 29%|██▊ | 8564/30000 [5:29:10<13:39:43, 2.29s/it] 29%|██▊ | 8565/30000 [5:29:12<13:35:03, 2.28s/it] 29%|██▊ | 8566/30000 [5:29:15<13:41:44, 2.30s/it] 29%|██▊ | 8567/30000 [5:29:17<13:46:10, 2.31s/it] 29%|██▊ | 8568/30000 [5:29:19<13:40:03, 2.30s/it] 29%|██▊ | 8569/30000 [5:29:22<13:38:41, 2.29s/it] 29%|██▊ | 8570/30000 [5:29:24<13:42:28, 2.30s/it] 29%|██▊ | 8570/30000 [5:29:24<13:42:28, 2.30s/it] 29%|██▊ | 8571/30000 [5:29:26<13:39:20, 2.29s/it] 29%|██▊ | 8572/30000 [5:29:28<13:36:53, 2.29s/it] 29%|██▊ | 8573/30000 [5:29:31<13:43:40, 2.31s/it] 29%|██▊ | 8574/30000 [5:29:33<13:40:08, 2.30s/it] 29%|██▊ | 8575/30000 [5:29:35<13:39:53, 2.30s/it] 29%|██▊ | 8576/30000 [5:29:38<13:39:52, 2.30s/it] 29%|██▊ | 8577/30000 [5:29:40<13:46:16, 2.31s/it] 29%|██▊ | 8578/30000 [5:29:42<13:48:15, 2.32s/it] 29%|██▊ | 8579/30000 [5:29:45<13:43:56, 2.31s/it] 29%|██▊ | 8580/30000 [5:29:47<13:41:34, 2.30s/it] 29%|██▊ | 8580/30000 [5:29:47<13:41:34, 2.30s/it] 29%|██▊ | 8581/30000 [5:29:49<13:39:32, 2.30s/it] 29%|██▊ | 8582/30000 [5:29:51<13:35:40, 2.29s/it] 29%|██▊ | 8583/30000 [5:29:54<13:34:51, 2.28s/it] 29%|██▊ | 8584/30000 [5:29:56<13:43:52, 2.31s/it] 29%|██▊ | 8585/30000 [5:29:58<13:43:53, 2.31s/it] 29%|██▊ | 8586/30000 [5:30:01<13:37:51, 2.29s/it] 29%|██▊ | 8587/30000 [5:30:03<13:37:10, 2.29s/it] 29%|██▊ | 8588/30000 [5:30:05<13:42:28, 2.30s/it] 29%|██▊ | 8589/30000 [5:30:08<13:36:35, 2.29s/it] 29%|██▊ | 8590/30000 [5:30:10<13:35:05, 2.28s/it] 29%|██▊ | 8590/30000 [5:30:10<13:35:05, 2.28s/it] 29%|██▊ | 8591/30000 [5:30:12<13:45:07, 2.31s/it] 29%|██▊ | 8592/30000 [5:30:15<13:50:32, 2.33s/it] 29%|██▊ | 8593/30000 [5:30:17<13:40:54, 2.30s/it] 29%|██▊ | 8594/30000 [5:30:19<13:38:03, 2.29s/it] 29%|██▊ | 8595/30000 [5:30:21<13:38:26, 2.29s/it] 29%|██▊ | 8596/30000 [5:30:24<13:32:15, 2.28s/it] 29%|██▊ | 8597/30000 [5:30:26<13:34:13, 2.28s/it] 29%|██▊ | 8598/30000 [5:30:28<13:40:11, 2.30s/it] 29%|██▊ | 8599/30000 [5:30:31<13:46:47, 2.32s/it] 29%|██▊ | 8600/30000 [5:30:33<13:43:32, 2.31s/it] 29%|██▊ | 8600/30000 [5:30:33<13:43:32, 2.31s/it] 29%|██▊ | 8601/30000 [5:30:35<13:45:35, 2.31s/it] 29%|██▊ | 8602/30000 [5:30:38<13:44:39, 2.31s/it] 29%|██▊ | 8603/30000 [5:30:40<13:35:50, 2.29s/it] 29%|██▊ | 8604/30000 [5:30:42<13:33:41, 2.28s/it] 29%|██▊ | 8605/30000 [5:30:44<13:37:54, 2.29s/it] 29%|██▊ | 8606/30000 [5:30:47<13:32:32, 2.28s/it] 29%|██▊ | 8607/30000 [5:30:49<13:33:28, 2.28s/it] 29%|██▊ | 8608/30000 [5:30:51<13:41:12, 2.30s/it] 29%|██▊ | 8609/30000 [5:30:54<13:43:43, 2.31s/it] 29%|██▊ | 8610/30000 [5:30:56<13:38:13, 2.30s/it] 29%|██▊ | 8610/30000 [5:30:56<13:38:13, 2.30s/it] 29%|██▊ | 8611/30000 [5:30:58<13:44:21, 2.31s/it] 29%|██▊ | 8612/30000 [5:31:00<13:41:49, 2.31s/it] 29%|██▊ | 8613/30000 [5:31:03<13:36:27, 2.29s/it] 29%|██▊ | 8614/30000 [5:31:05<13:43:16, 2.31s/it] 29%|██▊ | 8615/30000 [5:31:07<13:45:52, 2.32s/it] 29%|██▊ | 8616/30000 [5:31:10<13:43:05, 2.31s/it] 29%|██▊ | 8617/30000 [5:31:12<13:41:07, 2.30s/it] 29%|██▊ | 8618/30000 [5:31:14<13:43:48, 2.31s/it] 29%|██▊ | 8619/30000 [5:31:17<13:40:59, 2.30s/it] 29%|██▊ | 8620/30000 [5:31:19<13:38:23, 2.30s/it] 29%|██▊ | 8620/30000 [5:31:19<13:38:23, 2.30s/it] 29%|██▊ | 8621/30000 [5:31:21<13:47:09, 2.32s/it] 29%|██▊ | 8622/30000 [5:31:24<13:49:05, 2.33s/it] 29%|██▊ | 8623/30000 [5:31:26<13:46:15, 2.32s/it] 29%|██▊ | 8624/30000 [5:31:28<13:45:26, 2.32s/it] 29%|██▉ | 8625/30000 [5:31:31<13:42:23, 2.31s/it] 29%|██▉ | 8626/30000 [5:31:33<13:39:58, 2.30s/it] 29%|██▉ | 8627/30000 [5:31:35<13:36:12, 2.29s/it] 29%|██▉ | 8628/30000 [5:31:37<13:38:22, 2.30s/it] 29%|██▉ | 8629/30000 [5:31:40<13:34:15, 2.29s/it] 29%|██▉ | 8630/30000 [5:31:42<13:31:15, 2.28s/it] 29%|██▉ | 8630/30000 [5:31:42<13:31:15, 2.28s/it] 29%|██▉ | 8631/30000 [5:31:44<13:41:38, 2.31s/it] 29%|██▉ | 8632/30000 [5:31:47<13:46:56, 2.32s/it] 29%|██▉ | 8633/30000 [5:31:49<13:43:00, 2.31s/it] 29%|██▉ | 8634/30000 [5:31:51<13:42:17, 2.31s/it] 29%|██▉ | 8635/30000 [5:31:54<13:38:29, 2.30s/it] 29%|██▉ | 8636/30000 [5:31:56<13:35:07, 2.29s/it] 29%|██▉ | 8637/30000 [5:31:58<13:32:06, 2.28s/it] 29%|██▉ | 8638/30000 [5:32:00<13:40:32, 2.30s/it] 29%|██▉ | 8639/30000 [5:32:03<13:40:19, 2.30s/it] 29%|██▉ | 8640/30000 [5:32:05<13:38:42, 2.30s/it] 29%|██▉ | 8640/30000 [5:32:05<13:38:42, 2.30s/it] 29%|██▉ | 8641/30000 [5:32:07<13:45:47, 2.32s/it] 29%|██▉ | 8642/30000 [5:32:10<13:44:49, 2.32s/it] 29%|██▉ | 8643/30000 [5:32:12<13:37:57, 2.30s/it] 29%|██▉ | 8644/30000 [5:32:14<13:43:15, 2.31s/it] 29%|██▉ | 8645/30000 [5:32:17<13:43:46, 2.31s/it] 29%|██▉ | 8646/30000 [5:32:19<13:38:37, 2.30s/it] 29%|██▉ | 8647/30000 [5:32:21<13:43:34, 2.31s/it] 29%|██▉ | 8648/30000 [5:32:24<13:43:22, 2.31s/it] 29%|██▉ | 8649/30000 [5:32:26<13:39:21, 2.30s/it] 29%|██▉ | 8650/30000 [5:32:28<13:36:33, 2.29s/it] 29%|██▉ | 8650/30000 [5:32:28<13:36:33, 2.29s/it] 29%|██▉ | 8651/30000 [5:32:30<13:41:45, 2.31s/it] 29%|██▉ | 8652/30000 [5:32:33<13:45:04, 2.32s/it] 29%|██▉ | 8653/30000 [5:32:35<13:41:24, 2.31s/it] 29%|██▉ | 8654/30000 [5:32:37<13:46:06, 2.32s/it] 29%|██▉ | 8655/30000 [5:32:40<13:49:32, 2.33s/it] 29%|██▉ | 8656/30000 [5:32:42<13:45:46, 2.32s/it] 29%|██▉ | 8657/30000 [5:32:44<13:42:14, 2.31s/it] 29%|██▉ | 8658/30000 [5:32:47<13:44:44, 2.32s/it] 29%|██▉ | 8659/30000 [5:32:49<13:44:37, 2.32s/it] 29%|██▉ | 8660/30000 [5:32:51<13:39:59, 2.31s/it] 29%|██▉ | 8660/30000 [5:32:51<13:39:59, 2.31s/it] 29%|██▉ | 8661/30000 [5:32:54<13:44:34, 2.32s/it] 29%|██▉ | 8662/30000 [5:32:56<13:39:07, 2.30s/it] 29%|██▉ | 8663/30000 [5:32:58<13:39:09, 2.30s/it] 29%|██▉ | 8664/30000 [5:33:01<13:42:55, 2.31s/it] 29%|██▉ | 8665/30000 [5:33:03<13:45:22, 2.32s/it] 29%|██▉ | 8666/30000 [5:33:05<13:37:35, 2.30s/it] 29%|██▉ | 8667/30000 [5:33:07<13:33:59, 2.29s/it] 29%|██▉ | 8668/30000 [5:33:10<13:35:44, 2.29s/it] 29%|██▉ | 8669/30000 [5:33:12<13:31:23, 2.28s/it] 29%|██▉ | 8670/30000 [5:33:14<13:32:51, 2.29s/it] 29%|██▉ | 8670/30000 [5:33:14<13:32:51, 2.29s/it] 29%|██▉ | 8671/30000 [5:33:17<13:42:08, 2.31s/it] 29%|██▉ | 8672/30000 [5:33:19<13:44:36, 2.32s/it] 29%|██▉ | 8673/30000 [5:33:21<13:39:00, 2.30s/it] 29%|██▉ | 8674/30000 [5:33:24<13:43:43, 2.32s/it] 29%|██▉ | 8675/30000 [5:33:26<13:41:21, 2.31s/it] 29%|██▉ | 8676/30000 [5:33:28<13:38:14, 2.30s/it] 29%|██▉ | 8677/30000 [5:33:30<13:41:07, 2.31s/it] 29%|██▉ | 8678/30000 [5:33:33<13:37:39, 2.30s/it] 29%|██▉ | 8679/30000 [5:33:35<13:35:47, 2.30s/it] 29%|██▉ | 8680/30000 [5:33:37<13:36:17, 2.30s/it] 29%|██▉ | 8680/30000 [5:33:37<13:36:17, 2.30s/it] 29%|██▉ | 8681/30000 [5:33:40<13:42:31, 2.31s/it] 29%|██▉ | 8682/30000 [5:33:42<13:37:26, 2.30s/it] 29%|██▉ | 8683/30000 [5:33:44<13:31:03, 2.28s/it] 29%|██▉ | 8684/30000 [5:33:47<13:41:12, 2.31s/it] 29%|██▉ | 8685/30000 [5:33:49<13:42:25, 2.32s/it] 29%|██▉ | 8686/30000 [5:33:51<13:39:28, 2.31s/it] 29%|██▉ | 8687/30000 [5:33:54<13:43:04, 2.32s/it] 29%|██▉ | 8688/30000 [5:33:56<13:47:06, 2.33s/it] 29%|██▉ | 8689/30000 [5:33:58<13:44:59, 2.32s/it] 29%|██▉ | 8690/30000 [5:34:00<13:36:13, 2.30s/it] 29%|██▉ | 8690/30000 [5:34:00<13:36:13, 2.30s/it] 29%|██▉ | 8691/30000 [5:34:03<13:38:45, 2.31s/it] 29%|██▉ | 8692/30000 [5:34:05<13:31:17, 2.28s/it] 29%|██▉ | 8693/30000 [5:34:07<13:27:30, 2.27s/it] 29%|██▉ | 8694/30000 [5:34:10<13:32:10, 2.29s/it] 29%|██▉ | 8695/30000 [5:34:12<13:31:23, 2.29s/it] 29%|██▉ | 8696/30000 [5:34:14<13:30:05, 2.28s/it] 29%|██▉ | 8697/30000 [5:34:16<13:36:34, 2.30s/it] 29%|██▉ | 8698/30000 [5:34:19<13:41:25, 2.31s/it] 29%|██▉ | 8699/30000 [5:34:21<13:35:09, 2.30s/it] 29%|██▉ | 8700/30000 [5:34:23<13:38:19, 2.31s/it] 29%|██▉ | 8700/30000 [5:34:23<13:38:19, 2.31s/it] 29%|██▉ | 8701/30000 [5:34:26<13:41:22, 2.31s/it] 29%|██▉ | 8702/30000 [5:34:28<13:37:13, 2.30s/it] 29%|██▉ | 8703/30000 [5:34:30<13:36:14, 2.30s/it] 29%|██▉ | 8704/30000 [5:34:33<13:44:40, 2.32s/it] 29%|██▉ | 8705/30000 [5:34:35<13:45:16, 2.33s/it] 29%|██▉ | 8706/30000 [5:34:37<13:38:55, 2.31s/it] 29%|██▉ | 8707/30000 [5:34:40<13:40:04, 2.31s/it] 29%|██▉ | 8708/30000 [5:34:42<13:38:57, 2.31s/it] 29%|██▉ | 8709/30000 [5:34:44<13:34:03, 2.29s/it] 29%|██▉ | 8710/30000 [5:34:46<13:36:57, 2.30s/it] 29%|██▉ | 8710/30000 [5:34:46<13:36:57, 2.30s/it] 29%|██▉ | 8711/30000 [5:34:49<13:34:57, 2.30s/it] 29%|██▉ | 8712/30000 [5:34:51<13:32:49, 2.29s/it] 29%|██▉ | 8713/30000 [5:34:53<13:34:31, 2.30s/it] 29%|██▉ | 8714/30000 [5:34:56<13:39:51, 2.31s/it] 29%|██▉ | 8715/30000 [5:34:58<13:33:19, 2.29s/it] 29%|██▉ | 8716/30000 [5:35:00<13:31:30, 2.29s/it] 29%|██▉ | 8717/30000 [5:35:03<13:38:46, 2.31s/it] 29%|██▉ | 8718/30000 [5:35:05<13:37:17, 2.30s/it] 29%|██▉ | 8719/30000 [5:35:07<13:32:38, 2.29s/it] 29%|██▉ | 8720/30000 [5:35:09<13:36:48, 2.30s/it] 29%|██▉ | 8720/30000 [5:35:09<13:36:48, 2.30s/it] 29%|██▉ | 8721/30000 [5:35:12<13:40:22, 2.31s/it] 29%|██▉ | 8722/30000 [5:35:14<13:34:25, 2.30s/it] 29%|██▉ | 8723/30000 [5:35:16<13:33:41, 2.29s/it] 29%|██▉ | 8724/30000 [5:35:19<13:39:31, 2.31s/it] 29%|██▉ | 8725/30000 [5:35:21<13:39:37, 2.31s/it] 29%|██▉ | 8726/30000 [5:35:23<13:34:40, 2.30s/it] 29%|██▉ | 8727/30000 [5:35:26<13:33:05, 2.29s/it] 29%|██▉ | 8728/30000 [5:35:28<13:40:10, 2.31s/it] 29%|██▉ | 8729/30000 [5:35:30<13:38:59, 2.31s/it] 29%|██▉ | 8730/30000 [5:35:32<13:34:29, 2.30s/it] 29%|██▉ | 8730/30000 [5:35:33<13:34:29, 2.30s/it] 29%|██▉ | 8731/30000 [5:35:35<13:43:24, 2.32s/it] 29%|██▉ | 8732/30000 [5:35:37<13:40:56, 2.32s/it] 29%|██▉ | 8733/30000 [5:35:39<13:37:47, 2.31s/it] 29%|██▉ | 8734/30000 [5:35:42<13:35:41, 2.30s/it] 29%|██▉ | 8735/30000 [5:35:44<13:42:48, 2.32s/it] 29%|██▉ | 8736/30000 [5:35:46<13:41:20, 2.32s/it] 29%|██▉ | 8737/30000 [5:35:49<13:37:37, 2.31s/it] 29%|██▉ | 8738/30000 [5:35:51<13:40:04, 2.31s/it] 29%|██▉ | 8739/30000 [5:35:53<13:36:35, 2.30s/it] 29%|██▉ | 8740/30000 [5:35:56<13:33:01, 2.29s/it] 29%|██▉ | 8740/30000 [5:35:56<13:33:01, 2.29s/it] 29%|██▉ | 8741/30000 [5:35:58<13:32:21, 2.29s/it] 29%|██▉ | 8742/30000 [5:36:00<13:32:13, 2.29s/it] 29%|██▉ | 8743/30000 [5:36:02<13:32:10, 2.29s/it] 29%|██▉ | 8744/30000 [5:36:05<13:35:03, 2.30s/it] 29%|██▉ | 8745/30000 [5:36:07<13:40:04, 2.31s/it] 29%|██▉ | 8746/30000 [5:36:09<13:41:15, 2.32s/it] 29%|██▉ | 8747/30000 [5:36:12<13:38:26, 2.31s/it] 29%|██▉ | 8748/30000 [5:36:14<13:41:15, 2.32s/it] 29%|██▉ | 8749/30000 [5:36:16<13:37:46, 2.31s/it] 29%|██▉ | 8750/30000 [5:36:19<13:35:54, 2.30s/it] +{'loss': 0.0177, 'grad_norm': 0.09889943152666092, 'learning_rate': 2.592433713016008e-05, 'flow_matching_loss': 0.017746695782989264, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.43} +{'loss': 0.0139, 'grad_norm': 0.08268271386623383, 'learning_rate': 2.5912999763355553e-05, 'flow_matching_loss': 0.013939985074102879, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.43} +{'loss': 0.0151, 'grad_norm': 0.0791548639535904, 'learning_rate': 2.590164913622729e-05, 'flow_matching_loss': 0.015081868972629309, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.44} +{'loss': 0.0169, 'grad_norm': 0.09906744956970215, 'learning_rate': 2.589028526256737e-05, 'flow_matching_loss': 0.016892579011619092, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.44} +{'loss': 0.0162, 'grad_norm': 0.06402256339788437, 'learning_rate': 2.587890815618397e-05, 'flow_matching_loss': 0.01623848723247647, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.44} +{'loss': 0.0168, 'grad_norm': 0.11139404028654099, 'learning_rate': 2.5867517830901343e-05, 'flow_matching_loss': 0.016782090160995723, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.44} +{'loss': 0.016, 'grad_norm': 0.07841439545154572, 'learning_rate': 2.5856114300559812e-05, 'flow_matching_loss': 0.015975214634090663, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.45} +{'loss': 0.0169, 'grad_norm': 0.11154390871524811, 'learning_rate': 2.584469757901574e-05, 'flow_matching_loss': 0.01692858589813113, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.45} +{'loss': 0.0143, 'grad_norm': 0.061450738459825516, 'learning_rate': 2.5833267680141526e-05, 'flow_matching_loss': 0.014338553790003062, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.45} +{'loss': 0.0155, 'grad_norm': 0.09279503673315048, 'learning_rate': 2.5821824617825564e-05, 'flow_matching_loss': 0.015503286756575108, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.46} +{'loss': 0.0159, 'grad_norm': 0.08564874529838562, 'learning_rate': 2.5810368405972265e-05, 'flow_matching_loss': 0.01585025144740939, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.46} +{'loss': 0.0162, 'grad_norm': 0.07401000708341599, 'learning_rate': 2.5798899058501996e-05, 'flow_matching_loss': 0.016200677119195463, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.46} +{'loss': 0.0144, 'grad_norm': 0.07762070000171661, 'learning_rate': 2.578741658935111e-05, 'flow_matching_loss': 0.014444562513381243, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.46} +{'loss': 0.0154, 'grad_norm': 0.07920286059379578, 'learning_rate': 2.5775921012471875e-05, 'flow_matching_loss': 0.015426216647028923, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.47} +{'loss': 0.0153, 'grad_norm': 0.0873798355460167, 'learning_rate': 2.5764412341832505e-05, 'flow_matching_loss': 0.015282637905329466, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.47} +{'loss': 0.0167, 'grad_norm': 0.11981558799743652, 'learning_rate': 2.5752890591417125e-05, 'flow_matching_loss': 0.016729251109063627, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.47} +{'loss': 0.0177, 'grad_norm': 0.07959409803152084, 'learning_rate': 2.5741355775225744e-05, 'flow_matching_loss': 0.01765403039753437, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.48} +{'loss': 0.0163, 'grad_norm': 0.1161184161901474, 'learning_rate': 2.5729807907274254e-05, 'flow_matching_loss': 0.016284286603331567, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.48} +{'loss': 0.0165, 'grad_norm': 0.09296166896820068, 'learning_rate': 2.5718247001594403e-05, 'flow_matching_loss': 0.016473979409784078, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.48} +{'loss': 0.0151, 'grad_norm': 0.07755027711391449, 'learning_rate': 2.5706673072233776e-05, 'flow_matching_loss': 0.01506030298769474, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.48} +{'loss': 0.0159, 'grad_norm': 0.07190908491611481, 'learning_rate': 2.5695086133255795e-05, 'flow_matching_loss': 0.015932923182845116, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.49} +{'loss': 0.017, 'grad_norm': 0.07529281824827194, 'learning_rate': 2.5683486198739687e-05, 'flow_matching_loss': 0.017024032678455113, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.49} +{'loss': 0.0166, 'grad_norm': 0.0828472226858139, 'learning_rate': 2.5671873282780452e-05, 'flow_matching_loss': 0.01660259123891592, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.49} +{'loss': 0.0167, 'grad_norm': 0.08625348657369614, 'learning_rate': 2.5660247399488887e-05, 'flow_matching_loss': 0.016667408775538205, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.5} +{'loss': 0.016, 'grad_norm': 0.07573306560516357, 'learning_rate': 2.564860856299154e-05, 'flow_matching_loss': 0.01603578059002757, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.5} +{'loss': 0.0156, 'grad_norm': 0.09372639656066895, 'learning_rate': 2.5636956787430693e-05, 'flow_matching_loss': 0.015590169373899699, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.5} +{'loss': 0.0171, 'grad_norm': 0.07568460702896118, 'learning_rate': 2.5625292086964346e-05, 'flow_matching_loss': 0.017080802656710147, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.51} +{'loss': 0.0174, 'grad_norm': 0.1157645732164383, 'learning_rate': 2.5613614475766212e-05, 'flow_matching_loss': 0.017413893528282643, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.51} +{'loss': 0.0163, 'grad_norm': 0.07187263667583466, 'learning_rate': 2.5601923968025694e-05, 'flow_matching_loss': 0.016290823742747308, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.51} +{'loss': 0.015, 'grad_norm': 0.08803549408912659, 'learning_rate': 2.5590220577947862e-05, 'flow_matching_loss': 0.014991325419396162, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.51} +{'loss': 0.0164, 'grad_norm': 0.10450978577136993, 'learning_rate': 2.5578504319753427e-05, 'flow_matching_loss': 0.01642779540270567, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.52} +{'loss': 0.0163, 'grad_norm': 0.07677729427814484, 'learning_rate': 2.556677520767876e-05, 'flow_matching_loss': 0.01629939591512084, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.52} +{'loss': 0.015, 'grad_norm': 0.06766960769891739, 'learning_rate': 2.5555033255975835e-05, 'flow_matching_loss': 0.014951254427433013, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.52} +{'loss': 0.0156, 'grad_norm': 0.08217111974954605, 'learning_rate': 2.5543278478912228e-05, 'flow_matching_loss': 0.015571688674390316, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.53} +{'loss': 0.017, 'grad_norm': 0.09228511899709702, 'learning_rate': 2.55315108907711e-05, 'flow_matching_loss': 0.01702773626893759, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.53} +{'loss': 0.0173, 'grad_norm': 0.09963161498308182, 'learning_rate': 2.551973050585119e-05, 'flow_matching_loss': 0.01727589061483741, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.53} +{'loss': 0.0155, 'grad_norm': 0.08121296018362045, 'learning_rate': 2.5507937338466768e-05, 'flow_matching_loss': 0.015458132140338421, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.53} +{'loss': 0.0144, 'grad_norm': 0.08168414235115051, 'learning_rate': 2.549613140294764e-05, 'flow_matching_loss': 0.014396661799401045, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.54} +{'loss': 0.015, 'grad_norm': 0.09581513702869415, 'learning_rate': 2.548431271363914e-05, 'flow_matching_loss': 0.015005306154489518, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.54} +{'loss': 0.0171, 'grad_norm': 0.07747106999158859, 'learning_rate': 2.5472481284902092e-05, 'flow_matching_loss': 0.017050480004400015, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.54} +{'loss': 0.0161, 'grad_norm': 0.10051104426383972, 'learning_rate': 2.5460637131112788e-05, 'flow_matching_loss': 0.016053544357419014, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.55} 29%|██▉ | 8750/30000 [5:36:19<13:35:54, 2.30s/it] 29%|██▉ | 8751/30000 [5:36:21<13:36:05, 2.30s/it] 29%|██▉ | 8752/30000 [5:36:23<13:45:55, 2.33s/it] 29%|██▉ | 8753/30000 [5:36:26<13:47:35, 2.34s/it] 29%|██▉ | 8754/30000 [5:36:28<13:38:54, 2.31s/it] 29%|██▉ | 8755/30000 [5:36:30<13:34:14, 2.30s/it] 29%|██▉ | 8756/30000 [5:36:33<13:37:04, 2.31s/it] 29%|██▉ | 8757/30000 [5:36:35<13:32:09, 2.29s/it] 29%|██▉ | 8758/30000 [5:36:37<13:32:06, 2.29s/it] 29%|██▉ | 8759/30000 [5:36:39<13:42:52, 2.32s/it] 29%|██▉ | 8760/30000 [5:36:42<13:43:40, 2.33s/it] 29%|██▉ | 8760/30000 [5:36:42<13:43:40, 2.33s/it] 29%|██▉ | 8761/30000 [5:36:44<13:35:51, 2.30s/it] 29%|██▉ | 8762/30000 [5:36:46<13:36:01, 2.31s/it] 29%|██▉ | 8763/30000 [5:36:49<13:33:02, 2.30s/it] 29%|██▉ | 8764/30000 [5:36:51<13:29:43, 2.29s/it] 29%|██▉ | 8765/30000 [5:36:53<13:24:06, 2.27s/it] 29%|██▉ | 8766/30000 [5:36:55<13:26:23, 2.28s/it] 29%|██▉ | 8767/30000 [5:36:58<13:26:07, 2.28s/it] 29%|██▉ | 8768/30000 [5:37:00<13:25:01, 2.27s/it] 29%|██▉ | 8769/30000 [5:37:02<13:33:54, 2.30s/it] 29%|██▉ | 8770/30000 [5:37:05<13:35:08, 2.30s/it] 29%|██▉ | 8770/30000 [5:37:05<13:35:08, 2.30s/it] 29%|██▉ | 8771/30000 [5:37:07<13:31:07, 2.29s/it] 29%|██▉ | 8772/30000 [5:37:09<13:35:35, 2.31s/it] 29%|██▉ | 8773/30000 [5:37:12<13:34:12, 2.30s/it] 29%|██▉ | 8774/30000 [5:37:14<13:30:20, 2.29s/it] 29%|██▉ | 8775/30000 [5:37:16<13:27:47, 2.28s/it] 29%|██▉ | 8776/30000 [5:37:18<13:34:09, 2.30s/it] 29%|██▉ | 8777/30000 [5:37:21<13:34:09, 2.30s/it] 29%|██▉ | 8778/30000 [5:37:23<13:31:15, 2.29s/it] 29%|██▉ | 8779/30000 [5:37:25<13:36:14, 2.31s/it] 29%|██▉ | 8780/30000 [5:37:28<13:34:53, 2.30s/it] 29%|██▉ | 8780/30000 [5:37:28<13:34:53, 2.30s/it] 29%|██▉ | 8781/30000 [5:37:30<13:32:29, 2.30s/it] 29%|██▉ | 8782/30000 [5:37:32<13:29:44, 2.29s/it] 29%|██▉ | 8783/30000 [5:37:35<13:38:28, 2.31s/it] 29%|██▉ | 8784/30000 [5:37:37<13:38:47, 2.32s/it] 29%|██▉ | 8785/30000 [5:37:39<13:36:12, 2.31s/it] 29%|██▉ | 8786/30000 [5:37:41<13:28:21, 2.29s/it] 29%|██▉ | 8787/30000 [5:37:44<13:33:59, 2.30s/it] 29%|██▉ | 8788/30000 [5:37:46<13:28:21, 2.29s/it] 29%|██▉ | 8789/30000 [5:37:48<13:24:27, 2.28s/it] 29%|██▉ | 8790/30000 [5:37:51<13:31:54, 2.30s/it] 29%|██▉ | 8790/30000 [5:37:51<13:31:54, 2.30s/it] 29%|██▉ | 8791/30000 [5:37:53<13:37:43, 2.31s/it] 29%|██▉ | 8792/30000 [5:37:55<13:38:32, 2.32s/it] 29%|██▉ | 8793/30000 [5:37:58<13:30:12, 2.29s/it] 29%|██▉ | 8794/30000 [5:38:00<13:40:34, 2.32s/it] 29%|██▉ | 8795/30000 [5:38:02<13:30:13, 2.29s/it] 29%|██▉ | 8796/30000 [5:38:04<13:24:47, 2.28s/it] 29%|██▉ | 8797/30000 [5:38:07<13:31:35, 2.30s/it] 29%|██▉ | 8798/30000 [5:38:09<13:31:30, 2.30s/it] 29%|██▉ | 8799/30000 [5:38:11<13:30:22, 2.29s/it] 29%|██▉ | 8800/30000 [5:38:14<13:26:31, 2.28s/it] 29%|██▉ | 8800/30000 [5:38:14<13:26:31, 2.28s/it] 29%|██▉ | 8801/30000 [5:38:16<13:37:19, 2.31s/it] 29%|██▉ | 8802/30000 [5:38:18<13:41:51, 2.33s/it] 29%|██▉ | 8803/30000 [5:38:21<13:36:18, 2.31s/it] 29%|██▉ | 8804/30000 [5:38:23<13:31:25, 2.30s/it] 29%|██▉ | 8805/30000 [5:38:25<13:36:03, 2.31s/it] 29%|██▉ | 8806/30000 [5:38:27<13:31:45, 2.30s/it] 29%|██▉ | 8807/30000 [5:38:30<13:26:38, 2.28s/it] 29%|██▉ | 8808/30000 [5:38:32<13:28:02, 2.29s/it] 29%|██▉ | 8809/30000 [5:38:34<13:33:18, 2.30s/it] 29%|██▉ | 8810/30000 [5:38:37<13:32:34, 2.30s/it] 29%|██▉ | 8810/30000 [5:38:37<13:32:34, 2.30s/it] 29%|██▉ | 8811/30000 [5:38:39<13:29:14, 2.29s/it] 29%|██▉ | 8812/30000 [5:38:41<13:31:27, 2.30s/it] 29%|██▉ | 8813/30000 [5:38:44<13:31:25, 2.30s/it] 29%|██▉ | 8814/30000 [5:38:46<13:24:23, 2.28s/it] 29%|██▉ | 8815/30000 [5:38:48<13:23:14, 2.27s/it] 29%|██▉ | 8816/30000 [5:38:50<13:34:35, 2.31s/it] 29%|██▉ | 8817/30000 [5:38:53<13:34:14, 2.31s/it] 29%|██▉ | 8818/30000 [5:38:55<13:29:01, 2.29s/it] 29%|██▉ | 8819/30000 [5:38:57<13:27:00, 2.29s/it] 29%|██▉ | 8820/30000 [5:39:00<13:29:46, 2.29s/it] 29%|██▉ | 8820/30000 [5:39:00<13:29:46, 2.29s/it] 29%|██▉ | 8821/30000 [5:39:02<13:30:08, 2.30s/it] 29%|██▉ | 8822/30000 [5:39:04<13:25:53, 2.28s/it] 29%|██▉ | 8823/30000 [5:39:06<13:27:05, 2.29s/it] 29%|██▉ | 8824/30000 [5:39:09<13:34:08, 2.31s/it] 29%|██▉ | 8825/30000 [5:39:11<13:40:08, 2.32s/it] 29%|██▉ | 8826/30000 [5:39:13<13:33:41, 2.31s/it] 29%|██▉ | 8827/30000 [5:39:16<13:33:22, 2.30s/it] 29%|██▉ | 8828/30000 [5:39:18<13:38:00, 2.32s/it] 29%|██▉ | 8829/30000 [5:39:20<13:37:13, 2.32s/it] 29%|██▉ | 8830/30000 [5:39:23<13:35:16, 2.31s/it] 29%|██▉ | 8830/30000 [5:39:23<13:35:16, 2.31s/it] 29%|██▉ | 8831/30000 [5:39:25<13:33:30, 2.31s/it] 29%|██▉ | 8832/30000 [5:39:27<13:37:05, 2.32s/it] 29%|██▉ | 8833/30000 [5:39:30<13:36:09, 2.31s/it] 29%|██▉ | 8834/30000 [5:39:32<13:32:11, 2.30s/it] 29%|██▉ | 8835/30000 [5:39:34<13:29:59, 2.30s/it] 29%|██▉ | 8836/30000 [5:39:36<13:31:36, 2.30s/it] 29%|██▉ | 8837/30000 [5:39:39<13:31:03, 2.30s/it] 29%|██▉ | 8838/30000 [5:39:41<13:26:36, 2.29s/it] 29%|██▉ | 8839/30000 [5:39:43<13:24:47, 2.28s/it] 29%|██▉ | 8840/30000 [5:39:46<13:30:28, 2.30s/it] 29%|██▉ | 8840/30000 [5:39:46<13:30:28, 2.30s/it] 29%|██▉ | 8841/30000 [5:39:48<13:32:39, 2.30s/it] 29%|██▉ | 8842/30000 [5:39:50<13:26:27, 2.29s/it] 29%|██▉ | 8843/30000 [5:39:52<13:24:09, 2.28s/it] 29%|██▉ | 8844/30000 [5:39:55<13:28:01, 2.29s/it] 29%|██▉ | 8845/30000 [5:39:57<13:29:30, 2.30s/it] 29%|██▉ | 8846/30000 [5:39:59<13:25:47, 2.29s/it] 29%|██▉ | 8847/30000 [5:40:02<13:27:31, 2.29s/it] 29%|██▉ | 8848/30000 [5:40:04<13:31:42, 2.30s/it] 29%|██▉ | 8849/30000 [5:40:06<13:28:44, 2.29s/it] 30%|██▉ | 8850/30000 [5:40:08<13:23:16, 2.28s/it] 30%|██▉ | 8850/30000 [5:40:09<13:23:16, 2.28s/it] 30%|██▉ | 8851/30000 [5:40:11<13:31:11, 2.30s/it] 30%|██▉ | 8852/30000 [5:40:13<13:31:51, 2.30s/it] 30%|██▉ | 8853/30000 [5:40:15<13:29:14, 2.30s/it] 30%|██▉ | 8854/30000 [5:40:18<13:27:03, 2.29s/it] 30%|██▉ | 8855/30000 [5:40:20<13:33:43, 2.31s/it] 30%|██▉ | 8856/30000 [5:40:22<13:33:55, 2.31s/it] 30%|██▉ | 8857/30000 [5:40:25<13:30:33, 2.30s/it] 30%|██▉ | 8858/30000 [5:40:27<13:31:00, 2.30s/it] 30%|██▉ | 8859/30000 [5:40:29<13:28:46, 2.30s/it] 30%|██▉ | 8860/30000 [5:40:31<13:26:13, 2.29s/it] 30%|██▉ | 8860/30000 [5:40:32<13:26:13, 2.29s/it] 30%|██▉ | 8861/30000 [5:40:34<13:23:46, 2.28s/it] 30%|██▉ | 8862/30000 [5:40:36<13:30:27, 2.30s/it] 30%|██▉ | 8863/30000 [5:40:38<13:35:38, 2.32s/it] 30%|██▉ | 8864/30000 [5:40:41<13:32:15, 2.31s/it] 30%|██▉ | 8865/30000 [5:40:43<13:32:07, 2.31s/it] 30%|██▉ | 8866/30000 [5:40:45<13:36:43, 2.32s/it] 30%|██▉ | 8867/30000 [5:40:48<13:31:02, 2.30s/it] 30%|██▉ | 8868/30000 [5:40:50<13:29:49, 2.30s/it] 30%|██▉ | 8869/30000 [5:40:52<13:29:34, 2.30s/it] 30%|██▉ | 8870/30000 [5:40:55<13:32:07, 2.31s/it] 30%|██▉ | 8870/30000 [5:40:55<13:32:07, 2.31s/it] 30%|██▉ | 8871/30000 [5:40:57<13:28:24, 2.30s/it] 30%|██▉ | 8872/30000 [5:40:59<13:28:51, 2.30s/it] 30%|██▉ | 8873/30000 [5:41:01<13:31:21, 2.30s/it] 30%|██▉ | 8874/30000 [5:41:04<13:28:15, 2.30s/it] 30%|██▉ | 8875/30000 [5:41:06<13:26:26, 2.29s/it] 30%|██▉ | 8876/30000 [5:41:08<13:25:22, 2.29s/it] 30%|██▉ | 8877/30000 [5:41:11<13:29:24, 2.30s/it] 30%|██▉ | 8878/30000 [5:41:13<13:26:39, 2.29s/it] 30%|██▉ | 8879/30000 [5:41:15<13:31:03, 2.30s/it] 30%|██▉ | 8880/30000 [5:41:18<13:40:27, 2.33s/it] 30%|██▉ | 8880/30000 [5:41:18<13:40:27, 2.33s/it] 30%|██▉ | 8881/30000 [5:41:20<13:37:49, 2.32s/it] 30%|██▉ | 8882/30000 [5:41:22<13:31:10, 2.30s/it] 30%|██▉ | 8883/30000 [5:41:24<13:30:25, 2.30s/it] 30%|██▉ | 8884/30000 [5:41:27<13:33:08, 2.31s/it] 30%|██▉ | 8885/30000 [5:41:29<13:31:10, 2.30s/it] 30%|██▉ | 8886/30000 [5:41:31<13:30:10, 2.30s/it] 30%|██▉ | 8887/30000 [5:41:34<13:25:16, 2.29s/it] 30%|██▉ | 8888/30000 [5:41:36<13:31:50, 2.31s/it] 30%|██▉ | 8889/30000 [5:41:38<13:32:53, 2.31s/it] 30%|██▉ | 8890/30000 [5:41:41<13:28:33, 2.30s/it] 30%|██▉ | 8890/30000 [5:41:41<13:28:33, 2.30s/it] 30%|██▉ | 8891/30000 [5:41:43<13:28:17, 2.30s/it] 30%|██▉ | 8892/30000 [5:41:45<13:31:40, 2.31s/it] 30%|██▉ | 8893/30000 [5:41:47<13:26:31, 2.29s/it] 30%|██▉ | 8894/30000 [5:41:50<13:23:21, 2.28s/it] 30%|██▉ | 8895/30000 [5:41:52<13:25:45, 2.29s/it] 30%|██▉ | 8896/30000 [5:41:54<13:28:37, 2.30s/it] 30%|██▉ | 8897/30000 [5:41:57<13:25:12, 2.29s/it] 30%|██▉ | 8898/30000 [5:41:59<13:23:31, 2.28s/it] 30%|██▉ | 8899/30000 [5:42:01<13:26:42, 2.29s/it] 30%|██▉ | 8900/30000 [5:42:04<13:30:06, 2.30s/it] 30%|██▉ | 8900/30000 [5:42:04<13:30:06, 2.30s/it] 30%|██▉ | 8901/30000 [5:42:06<13:27:21, 2.30s/it] 30%|██▉ | 8902/30000 [5:42:08<13:24:32, 2.29s/it] 30%|██▉ | 8903/30000 [5:42:10<13:32:37, 2.31s/it] 30%|██▉ | 8904/30000 [5:42:13<13:36:04, 2.32s/it] 30%|██▉ | 8905/30000 [5:42:15<13:30:39, 2.31s/it] 30%|██▉ | 8906/30000 [5:42:17<13:28:07, 2.30s/it] 30%|██▉ | 8907/30000 [5:42:20<13:29:09, 2.30s/it] 30%|██▉ | 8908/30000 [5:42:22<13:26:01, 2.29s/it] 30%|██▉ | 8909/30000 [5:42:24<13:22:38, 2.28s/it] 30%|██▉ | 8910/30000 [5:42:27<13:25:48, 2.29s/it] 30%|██▉ | 8910/30000 [5:42:27<13:25:48, 2.29s/it] 30%|██▉ | 8911/30000 [5:42:29<13:31:08, 2.31s/it] 30%|██▉ | 8912/30000 [5:42:31<13:29:37, 2.30s/it] 30%|██▉ | 8913/30000 [5:42:33<13:28:46, 2.30s/it] 30%|██▉ | 8914/30000 [5:42:36<13:29:01, 2.30s/it] 30%|██▉ | 8915/30000 [5:42:38<13:23:18, 2.29s/it] 30%|██▉ | 8916/30000 [5:42:40<13:17:52, 2.27s/it] 30%|██▉ | 8917/30000 [5:42:43<13:22:15, 2.28s/it] 30%|██▉ | 8918/30000 [5:42:45<13:24:26, 2.29s/it] 30%|██▉ | 8919/30000 [5:42:47<13:23:22, 2.29s/it] 30%|██▉ | 8920/30000 [5:42:49<13:24:06, 2.29s/it] 30%|██▉ | 8920/30000 [5:42:49<13:24:06, 2.29s/it] 30%|██▉ | 8921/30000 [5:42:52<13:32:01, 2.31s/it] 30%|██▉ | 8922/30000 [5:42:54<13:25:46, 2.29s/it] 30%|██▉ | 8923/30000 [5:42:56<13:23:01, 2.29s/it] 30%|██▉ | 8924/30000 [5:42:59<13:27:26, 2.30s/it] 30%|██▉ | 8925/30000 [5:43:01<13:25:47, 2.29s/it] 30%|██▉ | 8926/30000 [5:43:03<13:25:35, 2.29s/it] 30%|██▉ | 8927/30000 [5:43:06<13:25:30, 2.29s/it] 30%|██▉ | 8928/30000 [5:43:08<13:29:47, 2.31s/it] 30%|██▉ | 8929/30000 [5:43:10<13:27:12, 2.30s/it] 30%|██▉ | 8930/30000 [5:43:12<13:26:00, 2.30s/it] 30%|██▉ | 8930/30000 [5:43:12<13:26:00, 2.30s/it] 30%|██▉ | 8931/30000 [5:43:15<13:30:30, 2.31s/it] 30%|██▉ | 8932/30000 [5:43:17<13:27:55, 2.30s/it] 30%|██▉ | 8933/30000 [5:43:19<13:23:47, 2.29s/it] 30%|██▉ | 8934/30000 [5:43:22<13:26:55, 2.30s/it] 30%|██▉ | 8935/30000 [5:43:24<13:29:55, 2.31s/it] 30%|██▉ | 8936/30000 [5:43:26<13:27:18, 2.30s/it] 30%|██▉ | 8937/30000 [5:43:29<13:27:35, 2.30s/it] 30%|██▉ | 8938/30000 [5:43:31<13:32:40, 2.32s/it] 30%|██▉ | 8939/30000 [5:43:33<13:29:16, 2.31s/it] 30%|██▉ | 8940/30000 [5:43:35<13:23:53, 2.29s/it] 30%|██▉ | 8940/30000 [5:43:35<13:23:53, 2.29s/it] 30%|██▉ | 8941/30000 [5:43:38<13:27:32, 2.30s/it] 30%|██▉ | 8942/30000 [5:43:40<13:31:08, 2.31s/it] 30%|██▉ | 8943/30000 [5:43:42<13:31:35, 2.31s/it] 30%|██▉ | 8944/30000 [5:43:45<13:34:07, 2.32s/it] 30%|██▉ | 8945/30000 [5:43:47<13:28:29, 2.30s/it] 30%|██▉ | 8946/30000 [5:43:49<13:24:40, 2.29s/it] 30%|██▉ | 8947/30000 [5:43:52<13:24:02, 2.29s/it] 30%|██▉ | 8948/30000 [5:43:54<13:29:19, 2.31s/it] 30%|██▉ | 8949/30000 [5:43:56<13:22:10, 2.29s/it] 30%|██▉ | 8950/30000 [5:43:58<13:21:36, 2.28s/it] 30%|██▉ | 8950/30000 [5:43:58<13:21:36, 2.28s/it] 30%|██▉ | 8951/30000 [5:44:01<13:28:56, 2.31s/it] 30%|██▉ | 8952/30000 [5:44:03<13:27:32, 2.30s/it] 30%|██▉ | 8953/30000 [5:44:05<13:28:35, 2.31s/it] 30%|██▉ | 8954/30000 [5:44:08<13:35:15, 2.32s/it] 30%|██▉ | 8955/30000 [5:44:10<13:32:11, 2.32s/it] 30%|██▉ | 8956/30000 [5:44:12<13:28:30, 2.31s/it] 30%|██▉ | 8957/30000 [5:44:15<13:26:50, 2.30s/it] 30%|██▉ | 8958/30000 [5:44:17<13:26:35, 2.30s/it] 30%|██▉ | 8959/30000 [5:44:19<13:23:36, 2.29s/it] 30%|██▉ | 8960/30000 [5:44:21<13:22:43, 2.29s/it] 30%|██▉ | 8960/30000 [5:44:22<13:22:43, 2.29s/it] 30%|██▉ | 8961/30000 [5:44:24<13:32:44, 2.32s/it] 30%|██▉ | 8962/30000 [5:44:26<13:28:36, 2.31s/it] 30%|██▉ | 8963/30000 [5:44:28<13:22:06, 2.29s/it] 30%|██▉ | 8964/30000 [5:44:31<13:27:04, 2.30s/it] 30%|██▉ | 8965/30000 [5:44:33<13:22:12, 2.29s/it] 30%|██▉ | 8966/30000 [5:44:35<13:21:03, 2.29s/it] 30%|██▉ | 8967/30000 [5:44:38<13:24:49, 2.30s/it] 30%|██▉ | 8968/30000 [5:44:40<13:30:28, 2.31s/it] 30%|██▉ | 8969/30000 [5:44:42<13:26:24, 2.30s/it] 30%|██▉ | 8970/30000 [5:44:45<13:30:41, 2.31s/it] 30%|██▉ | 8970/30000 [5:44:45<13:30:41, 2.31s/it] 30%|██▉ | 8971/30000 [5:44:47<13:32:28, 2.32s/it] 30%|██▉ | 8972/30000 [5:44:49<13:28:59, 2.31s/it] 30%|██▉ | 8973/30000 [5:44:51<13:26:54, 2.30s/it] 30%|██▉ | 8974/30000 [5:44:54<13:28:55, 2.31s/it] 30%|██▉ | 8975/30000 [5:44:56<13:25:35, 2.30s/it] 30%|██▉ | 8976/30000 [5:44:58<13:25:10, 2.30s/it] 30%|██▉ | 8977/30000 [5:45:01<13:33:15, 2.32s/it] 30%|██▉ | 8978/30000 [5:45:03<13:35:15, 2.33s/it] 30%|██▉ | 8979/30000 [5:45:05<13:32:29, 2.32s/it] 30%|██▉ | 8980/30000 [5:45:08<13:34:33, 2.33s/it] 30%|██▉ | 8980/30000 [5:45:08<13:34:33, 2.33s/it] 30%|██▉ | 8981/30000 [5:45:10<13:34:37, 2.33s/it] 30%|██▉ | 8982/30000 [5:45:12<13:28:26, 2.31s/it] 30%|██▉ | 8983/30000 [5:45:15<13:32:06, 2.32s/it] 30%|██▉ | 8984/30000 [5:45:17<13:32:30, 2.32s/it] 30%|██▉ | 8985/30000 [5:45:19<13:30:04, 2.31s/it] 30%|██▉ | 8986/30000 [5:45:21<13:23:19, 2.29s/it] 30%|██▉ | 8987/30000 [5:45:24<13:29:43, 2.31s/it] 30%|██▉ | 8988/30000 [5:45:26<13:23:53, 2.30s/it] 30%|██▉ | 8989/30000 [5:45:28<13:25:23, 2.30s/it] 30%|██▉ | 8990/30000 [5:45:31<13:30:58, 2.32s/it] 30%|██▉ | 8990/30000 [5:45:31<13:30:58, 2.32s/it] 30%|██▉ | 8991/30000 [5:45:33<13:33:35, 2.32s/it] 30%|██▉ | 8992/30000 [5:45:35<13:27:49, 2.31s/it] 30%|██▉ | 8993/30000 [5:45:38<13:27:41, 2.31s/it] 30%|██▉ | 8994/30000 [5:45:40<13:29:41, 2.31s/it] 30%|██▉ | 8995/30000 [5:45:42<13:25:57, 2.30s/it] 30%|██▉ | 8996/30000 [5:45:45<13:24:01, 2.30s/it] 30%|██▉ | 8997/30000 [5:45:47<13:28:17, 2.31s/it] 30%|██▉ | 8998/30000 [5:45:49<13:21:33, 2.29s/it] 30%|██▉ | 8999/30000 [5:45:51<13:24:07, 2.30s/it] 30%|███ | 9000/30000 [5:45:54<13:32:56, 2.32s/it] 30%|███ | 9000/30000 [5:45:54<13:32:56, 2.32s/it] 30%|███ | 9001/30000 [5:45:56<13:33:35, 2.32s/it] 30%|███ | 9002/30000 [5:45:58<13:32:02, 2.32s/it] 30%|███ | 9003/30000 [5:46:01<13:32:26, 2.32s/it] 30%|███ | 9004/30000 [5:46:03<13:28:08, 2.31s/it] 30%|███ | 9005/30000 [5:46:05<13:23:55, 2.30s/it] 30%|███ | 9006/30000 [5:46:08<13:22:01, 2.29s/it] 30%|███ | 9007/30000 [5:46:10<13:24:44, 2.30s/it] 30%|███ | 9008/30000 [5:46:12<13:24:09, 2.30s/it] 30%|███ | 9009/30000 [5:46:15<13:28:25, 2.31s/it] 30%|███ | 9010/30000 [5:46:17<13:27:42, 2.31s/it] 30%|███ | 9010/30000 [5:46:17<13:27:42, 2.31s/it] 30%|███ | 9011/30000 [5:46:19<13:21:10, 2.29s/it] 30%|███ | 9012/30000 [5:46:21<13:18:07, 2.28s/it] 30%|███ | 9013/30000 [5:46:24<13:25:46, 2.30s/it] 30%|███ | 9014/30000 [5:46:26<13:26:15, 2.31s/it] 30%|███ | 9015/30000 [5:46:28<13:23:59, 2.30s/it] 30%|███ | 9016/30000 [5:46:31<13:29:12, 2.31s/it] 30%|███ | 9017/30000 [5:46:33<13:27:45, 2.31s/it] 30%|███ | 9018/30000 [5:46:35<13:25:22, 2.30s/it] 30%|███ | 9019/30000 [5:46:38<13:25:06, 2.30s/it] 30%|███ | 9020/30000 [5:46:40<13:28:45, 2.31s/it] 30%|███ | 9020/30000 [5:46:40<13:28:45, 2.31s/it] 30%|███ | 9021/30000 [5:46:42<13:20:14, 2.29s/it] 30%|███ | 9022/30000 [5:46:44<13:16:56, 2.28s/it] 30%|███ | 9023/30000 [5:46:47<13:25:24, 2.30s/it] 30%|███ | 9024/30000 [5:46:49<13:30:46, 2.32s/it] 30%|███ | 9025/30000 [5:46:51<13:26:56, 2.31s/it] 30%|███ | 9026/30000 [5:46:54<13:31:00, 2.32s/it] 30%|███ | 9027/30000 [5:46:56<13:31:10, 2.32s/it] 30%|███ | 9028/30000 [5:46:58<13:25:18, 2.30s/it] 30%|███ | 9029/30000 [5:47:01<13:23:36, 2.30s/it] 30%|███ | 9030/30000 [5:47:03<13:21:43, 2.29s/it] 30%|███ | 9030/30000 [5:47:03<13:21:43, 2.29s/it] 30%|███ | 9031/30000 [5:47:05<13:25:57, 2.31s/it] 30%|███ | 9032/30000 [5:47:08<13:25:03, 2.30s/it] 30%|███ | 9033/30000 [5:47:10<13:31:05, 2.32s/it] 30%|███ | 9034/30000 [5:47:12<13:23:17, 2.30s/it] 30%|███ | 9035/30000 [5:47:14<13:21:17, 2.29s/it] 30%|███ | 9036/30000 [5:47:17<13:26:41, 2.31s/it] 30%|███ | 9037/30000 [5:47:19<13:26:25, 2.31s/it] 30%|███ | 9038/30000 [5:47:21<13:18:17, 2.28s/it] 30%|███ | 9039/30000 [5:47:24<13:27:38, 2.31s/it] 30%|███ | 9040/30000 [5:47:26<13:28:36, 2.31s/it] 30%|███ | 9040/30000 [5:47:26<13:28:36, 2.31s/it] 30%|███ | 9041/30000 [5:47:28<13:27:06, 2.31s/it] 30%|███ | 9042/30000 [5:47:31<13:27:37, 2.31s/it] 30%|███ | 9043/30000 [5:47:33<13:22:51, 2.30s/it] 30%|███ | 9044/30000 [5:47:35<13:19:25, 2.29s/it] 30%|███ | 9045/30000 [5:47:37<13:21:22, 2.29s/it] 30%|███ | 9046/30000 [5:47:40<13:24:36, 2.30s/it] 30%|███ | 9047/30000 [5:47:42<13:21:41, 2.30s/it] 30%|███ | 9048/30000 [5:47:44<13:20:30, 2.29s/it] 30%|███ | 9049/30000 [5:47:47<13:30:35, 2.32s/it] 30%|███ | 9050/30000 [5:47:49<13:31:19, 2.32s/it] 30%|███ | 9050/30000 [5:47:49<13:31:19, 2.32s/it] 30%|███ | 9051/30000 [5:47:51<13:23:45, 2.30s/it] 30%|███ | 9052/30000 [5:47:54<13:25:54, 2.31s/it] 30%|███ | 9053/30000 [5:47:56<13:24:50, 2.31s/it] 30%|███ | 9054/30000 [5:47:58<13:16:22, 2.28s/it] 30%|███ | 9055/30000 [5:48:00<13:20:21, 2.29s/it] 30%|███ | 9056/30000 [5:48:03<13:24:56, 2.31s/it] 30%|███ | 9057/30000 [5:48:05<13:21:03, 2.29s/it] 30%|███ | 9058/30000 [5:48:07<13:20:27, 2.29s/it] 30%|███ | 9059/30000 [5:48:10<13:26:18, 2.31s/it] 30%|███ | 9060/30000 [5:48:12<13:23:35, 2.30s/it] 30%|███ | 9060/30000 [5:48:12<13:23:35, 2.30s/it] 30%|███ | 9061/30000 [5:48:14<13:23:56, 2.30s/it] 30%|███ | 9062/30000 [5:48:17<13:27:34, 2.31s/it] 30%|███ | 9063/30000 [5:48:19<13:26:56, 2.31s/it] 30%|███ | 9064/30000 [5:48:21<13:24:33, 2.31s/it] 30%|███ | 9065/30000 [5:48:24<13:24:25, 2.31s/it] 30%|███ | 9066/30000 [5:48:26<13:31:57, 2.33s/it] 30%|███ | 9067/30000 [5:48:28<13:31:56, 2.33s/it] 30%|███ | 9068/30000 [5:48:31<13:25:29, 2.31s/it] 30%|███ | 9069/30000 [5:48:33<13:23:28, 2.30s/it] 30%|███ | 9070/30000 [5:48:35<13:21:28, 2.30s/it] 30%|███ | 9070/30000 [5:48:35<13:21:28, 2.30s/it] 30%|███ | 9071/30000 [5:48:37<13:24:23, 2.31s/it] 30%|███ | 9072/30000 [5:48:40<13:28:27, 2.32s/it] 30%|███ | 9073/30000 [5:48:42<13:33:51, 2.33s/it] 30%|███ | 9074/30000 [5:48:44<13:31:55, 2.33s/it] 30%|███ | 9075/30000 [5:48:47<13:26:29, 2.31s/it] 30%|███ | 9076/30000 [5:48:49<13:20:43, 2.30s/it] 30%|███ | 9077/30000 [5:48:51<13:27:43, 2.32s/it] 30%|███ | 9078/30000 [5:48:54<13:22:42, 2.30s/it] 30%|███ | 9079/30000 [5:48:56<13:24:37, 2.31s/it] 30%|███ | 9080/30000 [5:48:58<13:23:52, 2.31s/it] 30%|███ | 9080/30000 [5:48:58<13:23:52, 2.31s/it] 30%|███ | 9081/30000 [5:49:01<13:27:15, 2.32s/it] 30%|███ | 9082/30000 [5:49:03<13:19:21, 2.29s/it] 30%|███ | 9083/30000 [5:49:05<13:18:24, 2.29s/it] 30%|███ | 9084/30000 [5:49:07<13:22:16, 2.30s/it] 30%|███ | 9085/30000 [5:49:10<13:26:16, 2.31s/it] 30%|███ | 9086/30000 [5:49:12<13:20:01, 2.30s/it] 30%|███ | 9087/30000 [5:49:14<13:25:45, 2.31s/it] 30%|███ | 9088/30000 [5:49:17<13:27:34, 2.32s/it] 30%|███ | 9089/30000 [5:49:19<13:21:21, 2.30s/it] 30%|███ | 9090/30000 [5:49:21<13:22:19, 2.30s/it] 30%|███ | 9090/30000 [5:49:21<13:22:19, 2.30s/it] 30%|███ | 9091/30000 [5:49:24<13:26:48, 2.32s/it] 30%|███ | 9092/30000 [5:49:26<13:22:22, 2.30s/it] 30%|███ | 9093/30000 [5:49:28<13:18:10, 2.29s/it] 30%|███ | 9094/30000 [5:49:30<13:14:54, 2.28s/it] 30%|███ | 9095/30000 [5:49:33<13:19:25, 2.29s/it] 30%|███ | 9096/30000 [5:49:35<13:23:46, 2.31s/it] 30%|███ | 9097/30000 [5:49:37<13:20:09, 2.30s/it] 30%|███ | 9098/30000 [5:49:40<13:16:33, 2.29s/it] 30%|███ | 9099/30000 [5:49:42<13:21:20, 2.30s/it] 30%|███ | 9100/30000 [5:49:44<13:19:21, 2.29s/it] 30%|███ | 9100/30000 [5:49:44<13:19:21, 2.29s/it] 30%|███ | 9101/30000 [5:49:47<13:17:26, 2.29s/it] 30%|███ | 9102/30000 [5:49:49<13:22:18, 2.30s/it] 30%|███ | 9103/30000 [5:49:51<13:27:00, 2.32s/it] 30%|███ | 9104/30000 [5:49:53<13:20:38, 2.30s/it] 30%|███ | 9105/30000 [5:49:56<13:18:40, 2.29s/it] 30%|███ | 9106/30000 [5:49:58<13:23:07, 2.31s/it] 30%|███ | 9107/30000 [5:50:00<13:13:02, 2.28s/it] 30%|███ | 9108/30000 [5:50:03<13:08:40, 2.27s/it] 30%|███ | 9109/30000 [5:50:05<13:16:07, 2.29s/it] 30%|███ | 9110/30000 [5:50:07<13:19:23, 2.30s/it] 30%|███ | 9110/30000 [5:50:07<13:19:23, 2.30s/it] 30%|███ | 9111/30000 [5:50:10<13:21:54, 2.30s/it] 30%|███ | 9112/30000 [5:50:12<13:16:19, 2.29s/it] 30%|███ | 9113/30000 [5:50:14<13:24:09, 2.31s/it] 30%|███ | 9114/30000 [5:50:16<13:21:33, 2.30s/it] 30%|███ | 9115/30000 [5:50:19<13:22:03, 2.30s/it] 30%|███ | 9116/30000 [5:50:21<13:23:59, 2.31s/it] 30%|███ | 9117/30000 [5:50:23<13:26:13, 2.32s/it] 30%|███ | 9118/30000 [5:50:26<13:22:03, 2.30s/it] 30%|███ | 9119/30000 [5:50:28<13:21:48, 2.30s/it] 30%|███ | 9120/30000 [5:50:30<13:22:24, 2.31s/it] 30%|███ | 9120/30000 [5:50:30<13:22:24, 2.31s/it] 30%|███ | 9121/30000 [5:50:33<13:29:17, 2.33s/it] 30%|███ | 9122/30000 [5:50:35<13:28:30, 2.32s/it] 30%|███ | 9123/30000 [5:50:37<13:24:22, 2.31s/it] 30%|███ | 9124/30000 [5:50:39<13:19:52, 2.30s/it] 30%|███ | 9125/30000 [5:50:42<13:23:37, 2.31s/it] 30%|███ | 9126/30000 [5:50:44<13:26:27, 2.32s/it] 30%|███ | 9127/30000 [5:50:46<13:19:44, 2.30s/it] 30%|███ | 9128/30000 [5:50:49<13:15:38, 2.29s/it] 30%|███ | 9129/30000 [5:50:51<13:25:36, 2.32s/it] 30%|███ | 9130/30000 [5:50:53<13:21:52, 2.31s/it] 30%|███ | 9130/30000 [5:50:53<13:21:52, 2.31s/it] 30%|███ | 9131/30000 [5:50:56<13:19:08, 2.30s/it] 30%|███ | 9132/30000 [5:50:58<13:21:13, 2.30s/it] 30%|███ | 9133/30000 [5:51:00<13:21:11, 2.30s/it] 30%|███ | 9134/30000 [5:51:03<13:19:19, 2.30s/it] 30%|███ | 9135/30000 [5:51:05<13:20:40, 2.30s/it] 30%|███ | 9136/30000 [5:51:07<13:22:04, 2.31s/it] 30%|███ | 9137/30000 [5:51:09<13:24:50, 2.31s/it] 30%|███ | 9138/30000 [5:51:12<13:20:20, 2.30s/it] 30%|███ | 9139/30000 [5:51:14<13:15:19, 2.29s/it] 30%|███ | 9140/30000 [5:51:16<13:22:57, 2.31s/it] 30%|███ | 9140/30000 [5:51:16<13:22:57, 2.31s/it] 30%|███ | 9141/30000 [5:51:19<13:24:50, 2.32s/it] 30%|███ | 9142/30000 [5:51:21<13:20:13, 2.30s/it] 30%|███ | 9143/30000 [5:51:23<13:20:27, 2.30s/it] 30%|███ | 9144/30000 [5:51:26<13:27:27, 2.32s/it] 30%|███ | 9145/30000 [5:51:28<13:26:20, 2.32s/it] 30%|███ | 9146/30000 [5:51:30<13:21:54, 2.31s/it] 30%|███ | 9147/30000 [5:51:33<13:19:20, 2.30s/it] 30%|███ | 9148/30000 [5:51:35<13:22:32, 2.31s/it] 30%|███ | 9149/30000 [5:51:37<13:26:55, 2.32s/it] 30%|███ | 9150/30000 [5:51:39<13:20:24, 2.30s/it] 30%|███ | 9150/30000 [5:51:40<13:20:24, 2.30s/it] 31%|███ | 9151/30000 [5:51:42<13:16:06, 2.29s/it] 31%|███ | 9152/30000 [5:51:44<13:24:09, 2.31s/it] 31%|███ | 9153/30000 [5:51:46<13:24:11, 2.31s/it] 31%|███ | 9154/30000 [5:51:49<13:21:13, 2.31s/it] 31%|███ | 9155/30000 [5:51:51<13:18:48, 2.30s/it] 31%|███ | 9156/30000 [5:51:53<13:15:28, 2.29s/it] 31%|███ | 9157/30000 [5:51:56<13:14:09, 2.29s/it] 31%|███ | 9158/30000 [5:51:58<13:11:10, 2.28s/it] 31%|███ | 9159/30000 [5:52:00<13:20:39, 2.31s/it] 31%|███ | 9160/30000 [5:52:02<13:22:14, 2.31s/it] +{'loss': 0.0162, 'grad_norm': 0.10861380398273468, 'learning_rate': 2.5448780266662995e-05, 'flow_matching_loss': 0.016183933429419994, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.55} +{'loss': 0.0176, 'grad_norm': 0.07638845592737198, 'learning_rate': 2.5436910705959926e-05, 'flow_matching_loss': 0.017637730110436678, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.55} +{'loss': 0.0161, 'grad_norm': 0.07043460756540298, 'learning_rate': 2.542502846342621e-05, 'flow_matching_loss': 0.01613210644572973, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.55} +{'loss': 0.0171, 'grad_norm': 0.09041484445333481, 'learning_rate': 2.5413133553499892e-05, 'flow_matching_loss': 0.017090037930756807, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.56} +{'loss': 0.0179, 'grad_norm': 0.0740489810705185, 'learning_rate': 2.540122599063442e-05, 'flow_matching_loss': 0.017914092727005483, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.56} +{'loss': 0.0156, 'grad_norm': 0.07219716906547546, 'learning_rate': 2.538930578929859e-05, 'flow_matching_loss': 0.015627609007060526, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.56} +{'loss': 0.0183, 'grad_norm': 0.10164990276098251, 'learning_rate': 2.537737296397658e-05, 'flow_matching_loss': 0.018298513442277908, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.57} +{'loss': 0.0185, 'grad_norm': 0.08901666104793549, 'learning_rate': 2.5365427529167903e-05, 'flow_matching_loss': 0.018487638793885708, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.57} +{'loss': 0.015, 'grad_norm': 0.09261022508144379, 'learning_rate': 2.5353469499387375e-05, 'flow_matching_loss': 0.015038092061877251, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.57} +{'loss': 0.0171, 'grad_norm': 0.07215695828199387, 'learning_rate': 2.534149888916514e-05, 'flow_matching_loss': 0.017097650188952686, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.57} +{'loss': 0.0142, 'grad_norm': 0.09334525465965271, 'learning_rate': 2.5329515713046616e-05, 'flow_matching_loss': 0.014233542513102293, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.58} +{'loss': 0.0155, 'grad_norm': 0.07499765604734421, 'learning_rate': 2.531751998559249e-05, 'flow_matching_loss': 0.015516739059239625, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.58} +{'loss': 0.0152, 'grad_norm': 0.08905529975891113, 'learning_rate': 2.5305511721378703e-05, 'flow_matching_loss': 0.015188798680901528, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.58} +{'loss': 0.0157, 'grad_norm': 0.10457529127597809, 'learning_rate': 2.529349093499642e-05, 'flow_matching_loss': 0.015663477778434753, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.59} +{'loss': 0.0156, 'grad_norm': 0.08079066872596741, 'learning_rate': 2.528145764105205e-05, 'flow_matching_loss': 0.015633022971451284, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.59} +{'loss': 0.0155, 'grad_norm': 0.07249439507722855, 'learning_rate': 2.5269411854167166e-05, 'flow_matching_loss': 0.015501982904970646, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.59} +{'loss': 0.0183, 'grad_norm': 0.07552612572908401, 'learning_rate': 2.5257353588978534e-05, 'flow_matching_loss': 0.01829608529806137, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.6} +{'loss': 0.0164, 'grad_norm': 0.1026063933968544, 'learning_rate': 2.5245282860138083e-05, 'flow_matching_loss': 0.016371575929224492, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.6} +{'loss': 0.0155, 'grad_norm': 0.08165484666824341, 'learning_rate': 2.5233199682312894e-05, 'flow_matching_loss': 0.015549315325915813, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.6} +{'loss': 0.015, 'grad_norm': 0.055332690477371216, 'learning_rate': 2.5221104070185163e-05, 'flow_matching_loss': 0.015000383090227842, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.6} +{'loss': 0.0165, 'grad_norm': 0.08178418129682541, 'learning_rate': 2.5208996038452204e-05, 'flow_matching_loss': 0.016515743173658847, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.61} +{'loss': 0.0152, 'grad_norm': 0.0980820506811142, 'learning_rate': 2.5196875601826403e-05, 'flow_matching_loss': 0.015181224979460239, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.61} +{'loss': 0.0158, 'grad_norm': 0.07630413770675659, 'learning_rate': 2.518474277503525e-05, 'flow_matching_loss': 0.015769525058567525, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.61} +{'loss': 0.0159, 'grad_norm': 0.07979630678892136, 'learning_rate': 2.5172597572821266e-05, 'flow_matching_loss': 0.01592537760734558, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.62} +{'loss': 0.0151, 'grad_norm': 0.07829853147268295, 'learning_rate': 2.5160440009942015e-05, 'flow_matching_loss': 0.015098175685852765, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.62} +{'loss': 0.0154, 'grad_norm': 0.09768007695674896, 'learning_rate': 2.514827010117009e-05, 'flow_matching_loss': 0.015383510012179613, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.62} +{'loss': 0.0151, 'grad_norm': 0.08862598240375519, 'learning_rate': 2.5136087861293068e-05, 'flow_matching_loss': 0.015145967528223992, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.62} +{'loss': 0.0156, 'grad_norm': 0.09620499610900879, 'learning_rate': 2.5123893305113525e-05, 'flow_matching_loss': 0.015580752026289701, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.63} +{'loss': 0.0156, 'grad_norm': 0.0745769813656807, 'learning_rate': 2.5111686447448997e-05, 'flow_matching_loss': 0.01559119364246726, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.63} +{'loss': 0.0161, 'grad_norm': 0.07894081622362137, 'learning_rate': 2.5099467303131966e-05, 'flow_matching_loss': 0.016066277492791416, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.63} +{'loss': 0.0154, 'grad_norm': 0.07842984795570374, 'learning_rate': 2.508723588700984e-05, 'flow_matching_loss': 0.015427187085151672, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.64} +{'loss': 0.0162, 'grad_norm': 0.06613453477621078, 'learning_rate': 2.5074992213944952e-05, 'flow_matching_loss': 0.01619447683915496, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.64} +{'loss': 0.0155, 'grad_norm': 0.07790035754442215, 'learning_rate': 2.506273629881451e-05, 'flow_matching_loss': 0.015470034163445234, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.64} +{'loss': 0.0177, 'grad_norm': 0.08576702326536179, 'learning_rate': 2.5050468156510617e-05, 'flow_matching_loss': 0.01765516111627221, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.64} +{'loss': 0.015, 'grad_norm': 0.07592269778251648, 'learning_rate': 2.5038187801940215e-05, 'flow_matching_loss': 0.01496027149260044, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.65} +{'loss': 0.0161, 'grad_norm': 0.06679565459489822, 'learning_rate': 2.5025895250025088e-05, 'flow_matching_loss': 0.01610537599772215, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.65} +{'loss': 0.0168, 'grad_norm': 0.0946272611618042, 'learning_rate': 2.501359051570186e-05, 'flow_matching_loss': 0.016818077117204667, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.65} +{'loss': 0.0168, 'grad_norm': 0.08136779814958572, 'learning_rate': 2.500127361392193e-05, 'flow_matching_loss': 0.016765352711081503, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.66} +{'loss': 0.0158, 'grad_norm': 0.0812956914305687, 'learning_rate': 2.49889445596515e-05, 'flow_matching_loss': 0.015825557336211205, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.66} +{'loss': 0.0162, 'grad_norm': 0.07115481793880463, 'learning_rate': 2.497660336787154e-05, 'flow_matching_loss': 0.016228922177106142, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.66} +{'loss': 0.0157, 'grad_norm': 0.0937192440032959, 'learning_rate': 2.496425005357775e-05, 'flow_matching_loss': 0.015736792329698802, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.67} 31%|███ | 9160/30000 [5:52:03<13:22:14, 2.31s/it] 31%|███ | 9161/30000 [5:52:05<13:20:25, 2.30s/it] 31%|███ | 9162/30000 [5:52:07<13:19:47, 2.30s/it] 31%|███ | 9163/30000 [5:52:09<13:21:09, 2.31s/it] 31%|███ | 9164/30000 [5:52:12<13:20:33, 2.31s/it] 31%|███ | 9165/30000 [5:52:14<13:17:00, 2.30s/it] 31%|███ | 9166/30000 [5:52:16<13:16:38, 2.29s/it] 31%|███ | 9167/30000 [5:52:19<13:14:56, 2.29s/it] 31%|███ | 9168/30000 [5:52:21<13:23:38, 2.31s/it] 31%|███ | 9169/30000 [5:52:23<13:30:40, 2.33s/it] 31%|███ | 9170/30000 [5:52:26<13:24:28, 2.32s/it] 31%|███ | 9170/30000 [5:52:26<13:24:28, 2.32s/it] 31%|███ | 9171/30000 [5:52:28<13:21:29, 2.31s/it] 31%|███ | 9172/30000 [5:52:30<13:25:33, 2.32s/it] 31%|███ | 9173/30000 [5:52:32<13:18:45, 2.30s/it] 31%|███ | 9174/30000 [5:52:35<13:14:55, 2.29s/it] 31%|███ | 9175/30000 [5:52:37<13:17:13, 2.30s/it] 31%|███ | 9176/30000 [5:52:39<13:19:30, 2.30s/it] 31%|███ | 9177/30000 [5:52:42<13:19:27, 2.30s/it] 31%|███ | 9178/30000 [5:52:44<13:13:13, 2.29s/it] 31%|███ | 9179/30000 [5:52:46<13:18:23, 2.30s/it] 31%|███ | 9180/30000 [5:52:49<13:23:12, 2.31s/it] 31%|███ | 9180/30000 [5:52:49<13:23:12, 2.31s/it] 31%|███ | 9181/30000 [5:52:51<13:16:44, 2.30s/it] 31%|███ | 9182/30000 [5:52:53<13:15:49, 2.29s/it] 31%|███ | 9183/30000 [5:52:55<13:19:13, 2.30s/it] 31%|███ | 9184/30000 [5:52:58<13:23:16, 2.32s/it] 31%|███ | 9185/30000 [5:53:00<13:27:24, 2.33s/it] 31%|███ | 9186/30000 [5:53:02<13:21:50, 2.31s/it] 31%|███ | 9187/30000 [5:53:05<13:19:55, 2.31s/it] 31%|███ | 9188/30000 [5:53:07<13:25:22, 2.32s/it] 31%|███ | 9189/30000 [5:53:09<13:20:32, 2.31s/it] 31%|███ | 9190/30000 [5:53:12<13:18:22, 2.30s/it] 31%|███ | 9190/30000 [5:53:12<13:18:22, 2.30s/it] 31%|███ | 9191/30000 [5:53:14<13:24:30, 2.32s/it] 31%|███ | 9192/30000 [5:53:16<13:23:42, 2.32s/it] 31%|███ | 9193/30000 [5:53:19<13:20:58, 2.31s/it] 31%|███ | 9194/30000 [5:53:21<13:19:41, 2.31s/it] 31%|███ | 9195/30000 [5:53:23<13:21:26, 2.31s/it] 31%|███ | 9196/30000 [5:53:26<13:20:42, 2.31s/it] 31%|███ | 9197/30000 [5:53:28<13:18:43, 2.30s/it] 31%|███ | 9198/30000 [5:53:30<13:15:00, 2.29s/it] 31%|███ | 9199/30000 [5:53:32<13:21:01, 2.31s/it] 31%|███ | 9200/30000 [5:53:35<13:23:52, 2.32s/it] 31%|███ | 9200/30000 [5:53:35<13:23:52, 2.32s/it] 31%|███ | 9201/30000 [5:53:37<13:21:45, 2.31s/it] 31%|███ | 9202/30000 [5:53:39<13:18:23, 2.30s/it] 31%|███ | 9203/30000 [5:53:42<13:18:07, 2.30s/it] 31%|███ | 9204/30000 [5:53:44<13:11:32, 2.28s/it] 31%|███ | 9205/30000 [5:53:46<13:05:48, 2.27s/it] 31%|███ | 9206/30000 [5:53:48<13:16:14, 2.30s/it] 31%|███ | 9207/30000 [5:53:51<13:15:38, 2.30s/it] 31%|███ | 9208/30000 [5:53:53<13:15:15, 2.29s/it] 31%|███ | 9209/30000 [5:53:55<13:15:13, 2.29s/it] 31%|███ | 9210/30000 [5:53:58<13:20:41, 2.31s/it] 31%|███ | 9210/30000 [5:53:58<13:20:41, 2.31s/it] 31%|███ | 9211/30000 [5:54:00<13:19:48, 2.31s/it] 31%|███ | 9212/30000 [5:54:02<13:17:31, 2.30s/it] 31%|███ | 9213/30000 [5:54:05<13:13:24, 2.29s/it] 31%|███ | 9214/30000 [5:54:07<13:17:02, 2.30s/it] 31%|███ | 9215/30000 [5:54:09<13:21:46, 2.31s/it] 31%|███ | 9216/30000 [5:54:12<13:17:15, 2.30s/it] 31%|███ | 9217/30000 [5:54:14<13:17:00, 2.30s/it] 31%|███ | 9218/30000 [5:54:16<13:23:51, 2.32s/it] 31%|███ | 9219/30000 [5:54:18<13:21:40, 2.31s/it] 31%|███ | 9220/30000 [5:54:21<13:19:31, 2.31s/it] 31%|███ | 9220/30000 [5:54:21<13:19:31, 2.31s/it] 31%|███ | 9221/30000 [5:54:23<13:17:31, 2.30s/it] 31%|███ | 9222/30000 [5:54:25<13:26:55, 2.33s/it] 31%|███ | 9223/30000 [5:54:28<13:29:11, 2.34s/it] 31%|███ | 9224/30000 [5:54:30<13:23:46, 2.32s/it] 31%|███ | 9225/30000 [5:54:32<13:19:19, 2.31s/it] 31%|███ | 9226/30000 [5:54:35<13:24:08, 2.32s/it] 31%|███ | 9227/30000 [5:54:37<13:22:07, 2.32s/it] 31%|███ | 9228/30000 [5:54:39<13:17:35, 2.30s/it] 31%|███ | 9229/30000 [5:54:42<13:14:33, 2.30s/it] 31%|███ | 9230/30000 [5:54:44<13:16:05, 2.30s/it] 31%|███ | 9230/30000 [5:54:44<13:16:05, 2.30s/it] 31%|███ | 9231/30000 [5:54:46<13:18:56, 2.31s/it] 31%|███ | 9232/30000 [5:54:49<13:17:53, 2.31s/it] 31%|███ | 9233/30000 [5:54:51<13:20:21, 2.31s/it] 31%|███ | 9234/30000 [5:54:53<13:15:37, 2.30s/it] 31%|███ | 9235/30000 [5:54:55<13:17:21, 2.30s/it] 31%|███ | 9236/30000 [5:54:58<13:10:06, 2.28s/it] 31%|███ | 9237/30000 [5:55:00<13:09:14, 2.28s/it] 31%|███ | 9238/30000 [5:55:02<13:10:42, 2.29s/it] 31%|███ | 9239/30000 [5:55:05<13:13:52, 2.29s/it] 31%|███ | 9240/30000 [5:55:07<13:14:30, 2.30s/it] 31%|███ | 9240/30000 [5:55:07<13:14:30, 2.30s/it] 31%|███ | 9241/30000 [5:55:09<13:13:28, 2.29s/it] 31%|███ | 9242/30000 [5:55:11<13:17:42, 2.31s/it] 31%|███ | 9243/30000 [5:55:14<13:18:09, 2.31s/it] 31%|███ | 9244/30000 [5:55:16<13:12:56, 2.29s/it] 31%|███ | 9245/30000 [5:55:18<13:10:53, 2.29s/it] 31%|███ | 9246/30000 [5:55:21<13:18:14, 2.31s/it] 31%|███ | 9247/30000 [5:55:23<13:17:07, 2.30s/it] 31%|███ | 9248/30000 [5:55:25<13:05:43, 2.27s/it] 31%|███ | 9249/30000 [5:55:27<13:06:38, 2.27s/it] 31%|███ | 9250/30000 [5:55:30<13:13:38, 2.29s/it] 31%|███ | 9250/30000 [5:55:30<13:13:38, 2.29s/it] 31%|███ | 9251/30000 [5:55:32<13:12:23, 2.29s/it] 31%|███ | 9252/30000 [5:55:34<13:09:06, 2.28s/it] 31%|███ | 9253/30000 [5:55:37<13:09:55, 2.28s/it] 31%|███ | 9254/30000 [5:55:39<13:11:43, 2.29s/it] 31%|███ | 9255/30000 [5:55:41<13:10:00, 2.28s/it] 31%|███ | 9256/30000 [5:55:44<13:15:22, 2.30s/it] 31%|███ | 9257/30000 [5:55:46<13:20:50, 2.32s/it] 31%|███ | 9258/30000 [5:55:48<13:19:40, 2.31s/it] 31%|███ | 9259/30000 [5:55:50<13:12:51, 2.29s/it] 31%|███ | 9260/30000 [5:55:53<13:07:54, 2.28s/it] 31%|███ | 9260/30000 [5:55:53<13:07:54, 2.28s/it] 31%|███ | 9261/30000 [5:55:55<13:14:43, 2.30s/it] 31%|███ | 9262/30000 [5:55:57<13:14:37, 2.30s/it] 31%|███ | 9263/30000 [5:56:00<13:11:02, 2.29s/it] 31%|███ | 9264/30000 [5:56:02<13:10:30, 2.29s/it] 31%|███ | 9265/30000 [5:56:04<13:10:52, 2.29s/it] 31%|███ | 9266/30000 [5:56:06<13:09:05, 2.28s/it] 31%|███ | 9267/30000 [5:56:09<13:08:06, 2.28s/it] 31%|███ | 9268/30000 [5:56:11<13:13:17, 2.30s/it] 31%|███ | 9269/30000 [5:56:13<13:14:35, 2.30s/it] 31%|███ | 9270/30000 [5:56:16<13:08:04, 2.28s/it] 31%|███ | 9270/30000 [5:56:16<13:08:04, 2.28s/it] 31%|███ | 9271/30000 [5:56:18<13:16:26, 2.31s/it] 31%|███ | 9272/30000 [5:56:20<13:18:47, 2.31s/it] 31%|███ | 9273/30000 [5:56:23<13:16:26, 2.31s/it] 31%|███ | 9274/30000 [5:56:25<13:09:00, 2.28s/it] 31%|███ | 9275/30000 [5:56:27<13:13:36, 2.30s/it] 31%|███ | 9276/30000 [5:56:29<13:15:01, 2.30s/it] 31%|███ | 9277/30000 [5:56:32<13:11:30, 2.29s/it] 31%|███ | 9278/30000 [5:56:34<13:14:49, 2.30s/it] 31%|███ | 9279/30000 [5:56:36<13:14:50, 2.30s/it] 31%|███ | 9280/30000 [5:56:39<13:13:23, 2.30s/it] 31%|███ | 9280/30000 [5:56:39<13:13:23, 2.30s/it] 31%|███ | 9281/30000 [5:56:41<13:08:56, 2.28s/it] 31%|███ | 9282/30000 [5:56:43<13:14:04, 2.30s/it] 31%|███ | 9283/30000 [5:56:46<13:15:47, 2.30s/it] 31%|███ | 9284/30000 [5:56:48<13:15:00, 2.30s/it] 31%|███ | 9285/30000 [5:56:50<13:14:11, 2.30s/it] 31%|███ | 9286/30000 [5:56:52<13:14:51, 2.30s/it] 31%|███ | 9287/30000 [5:56:55<13:13:24, 2.30s/it] 31%|███ | 9288/30000 [5:56:57<13:08:51, 2.29s/it] 31%|███ | 9289/30000 [5:56:59<13:15:44, 2.31s/it] 31%|███ | 9290/30000 [5:57:02<13:14:58, 2.30s/it] 31%|███ | 9290/30000 [5:57:02<13:14:58, 2.30s/it] 31%|███ | 9291/30000 [5:57:04<13:12:35, 2.30s/it] 31%|███ | 9292/30000 [5:57:06<13:16:53, 2.31s/it] 31%|███ | 9293/30000 [5:57:09<13:17:00, 2.31s/it] 31%|███ | 9294/30000 [5:57:11<13:10:32, 2.29s/it] 31%|███ | 9295/30000 [5:57:13<13:11:55, 2.29s/it] 31%|███ | 9296/30000 [5:57:15<13:12:46, 2.30s/it] 31%|███ | 9297/30000 [5:57:18<13:14:06, 2.30s/it] 31%|███ | 9298/30000 [5:57:20<13:11:27, 2.29s/it] 31%|███ | 9299/30000 [5:57:22<13:16:52, 2.31s/it] 31%|███ | 9300/30000 [5:57:25<13:21:02, 2.32s/it] 31%|███ | 9300/30000 [5:57:25<13:21:02, 2.32s/it] 31%|███ | 9301/30000 [5:57:27<13:14:54, 2.30s/it] 31%|███ | 9302/30000 [5:57:29<13:17:02, 2.31s/it] 31%|███ | 9303/30000 [5:57:32<13:16:22, 2.31s/it] 31%|███ | 9304/30000 [5:57:34<13:16:06, 2.31s/it] 31%|███ | 9305/30000 [5:57:36<13:18:56, 2.32s/it] 31%|███ | 9306/30000 [5:57:39<13:23:11, 2.33s/it] 31%|███ | 9307/30000 [5:57:41<13:18:52, 2.32s/it] 31%|███ | 9308/30000 [5:57:43<13:15:01, 2.31s/it] 31%|███ | 9309/30000 [5:57:45<13:09:26, 2.29s/it] 31%|███ | 9310/30000 [5:57:48<13:12:51, 2.30s/it] 31%|███ | 9310/30000 [5:57:48<13:12:51, 2.30s/it] 31%|███ | 9311/30000 [5:57:50<13:09:29, 2.29s/it] 31%|███ | 9312/30000 [5:57:52<13:08:18, 2.29s/it] 31%|███ | 9313/30000 [5:57:55<13:13:00, 2.30s/it] 31%|███ | 9314/30000 [5:57:57<13:17:22, 2.31s/it] 31%|███ | 9315/30000 [5:57:59<13:15:19, 2.31s/it] 31%|███ | 9316/30000 [5:58:02<13:23:14, 2.33s/it] 31%|███ | 9317/30000 [5:58:04<13:18:05, 2.32s/it] 31%|███ | 9318/30000 [5:58:06<13:15:34, 2.31s/it] 31%|███ | 9319/30000 [5:58:08<13:12:04, 2.30s/it] 31%|███ | 9320/30000 [5:58:11<13:18:13, 2.32s/it] 31%|███ | 9320/30000 [5:58:11<13:18:13, 2.32s/it] 31%|███ | 9321/30000 [5:58:13<13:15:05, 2.31s/it] 31%|███ | 9322/30000 [5:58:15<13:11:00, 2.30s/it] 31%|███ | 9323/30000 [5:58:18<13:14:58, 2.31s/it] 31%|███ | 9324/30000 [5:58:20<13:15:37, 2.31s/it] 31%|███ | 9325/30000 [5:58:22<13:14:20, 2.31s/it] 31%|███ | 9326/30000 [5:58:25<13:16:51, 2.31s/it] 31%|███ | 9327/30000 [5:58:27<13:15:53, 2.31s/it] 31%|███ | 9328/30000 [5:58:29<13:13:33, 2.30s/it] 31%|███ | 9329/30000 [5:58:32<13:18:11, 2.32s/it] 31%|███ | 9330/30000 [5:58:34<13:21:39, 2.33s/it] 31%|███ | 9330/30000 [5:58:34<13:21:39, 2.33s/it] 31%|███ | 9331/30000 [5:58:36<13:15:21, 2.31s/it] 31%|███ | 9332/30000 [5:58:38<13:09:17, 2.29s/it] 31%|███ | 9333/30000 [5:58:41<13:08:29, 2.29s/it] 31%|███ | 9334/30000 [5:58:43<13:04:42, 2.28s/it] 31%|███ | 9335/30000 [5:58:45<13:05:06, 2.28s/it] 31%|███ | 9336/30000 [5:58:48<13:16:16, 2.31s/it] 31%|███ | 9337/30000 [5:58:50<13:16:52, 2.31s/it] 31%|███ | 9338/30000 [5:58:52<13:11:51, 2.30s/it] 31%|███ | 9339/30000 [5:58:55<13:17:27, 2.32s/it] 31%|███ | 9340/30000 [5:58:57<13:14:11, 2.31s/it] 31%|███ | 9340/30000 [5:58:57<13:14:11, 2.31s/it] 31%|███ | 9341/30000 [5:58:59<13:09:24, 2.29s/it] 31%|███ | 9342/30000 [5:59:01<13:07:59, 2.29s/it] 31%|███ | 9343/30000 [5:59:04<13:12:02, 2.30s/it] 31%|███ | 9344/30000 [5:59:06<13:08:31, 2.29s/it] 31%|███ | 9345/30000 [5:59:08<13:09:39, 2.29s/it] 31%|███ | 9346/30000 [5:59:11<13:14:58, 2.31s/it] 31%|███ | 9347/30000 [5:59:13<13:06:38, 2.29s/it] 31%|███ | 9348/30000 [5:59:15<13:03:23, 2.28s/it] 31%|███ | 9349/30000 [5:59:18<13:12:30, 2.30s/it] 31%|███ | 9350/30000 [5:59:20<13:11:33, 2.30s/it] 31%|███ | 9350/30000 [5:59:20<13:11:33, 2.30s/it] 31%|███ | 9351/30000 [5:59:22<13:11:22, 2.30s/it] 31%|███ | 9352/30000 [5:59:24<13:16:45, 2.32s/it] 31%|███ | 9353/30000 [5:59:27<13:15:30, 2.31s/it] 31%|███ | 9354/30000 [5:59:29<13:07:24, 2.29s/it] 31%|███ | 9355/30000 [5:59:31<13:11:52, 2.30s/it] 31%|███ | 9356/30000 [5:59:34<13:08:24, 2.29s/it] 31%|███ | 9357/30000 [5:59:36<13:07:30, 2.29s/it] 31%|███ | 9358/30000 [5:59:38<13:04:35, 2.28s/it] 31%|███ | 9359/30000 [5:59:41<13:13:16, 2.31s/it] 31%|███ | 9360/30000 [5:59:43<13:14:52, 2.31s/it] 31%|███ | 9360/30000 [5:59:43<13:14:52, 2.31s/it] 31%|███ | 9361/30000 [5:59:45<13:13:40, 2.31s/it] 31%|███ | 9362/30000 [5:59:47<13:14:36, 2.31s/it] 31%|███ | 9363/30000 [5:59:50<13:17:09, 2.32s/it] 31%|███ | 9364/30000 [5:59:52<13:07:48, 2.29s/it] 31%|███ | 9365/30000 [5:59:54<13:06:50, 2.29s/it] 31%|███ | 9366/30000 [5:59:57<13:11:30, 2.30s/it] 31%|███ | 9367/30000 [5:59:59<13:14:39, 2.31s/it] 31%|███ | 9368/30000 [6:00:01<13:12:27, 2.30s/it] 31%|███ | 9369/30000 [6:00:04<13:11:53, 2.30s/it] 31%|███ | 9370/30000 [6:00:06<13:09:52, 2.30s/it] 31%|███ | 9370/30000 [6:00:06<13:09:52, 2.30s/it] 31%|███ | 9371/30000 [6:00:08<13:07:09, 2.29s/it] 31%|███ | 9372/30000 [6:00:10<13:04:35, 2.28s/it] 31%|███ | 9373/30000 [6:00:13<13:09:19, 2.30s/it] 31%|███ | 9374/30000 [6:00:15<13:06:56, 2.29s/it] 31%|███▏ | 9375/30000 [6:00:17<13:05:00, 2.28s/it] 31%|███▏ | 9376/30000 [6:00:20<13:12:01, 2.30s/it] 31%|███▏ | 9377/30000 [6:00:22<13:15:31, 2.31s/it] 31%|███▏ | 9378/30000 [6:00:24<13:10:57, 2.30s/it] 31%|███▏ | 9379/30000 [6:00:26<13:07:02, 2.29s/it] 31%|███▏ | 9380/30000 [6:00:29<13:09:53, 2.30s/it] 31%|███▏ | 9380/30000 [6:00:29<13:09:53, 2.30s/it] 31%|███▏ | 9381/30000 [6:00:31<13:05:53, 2.29s/it] 31%|███▏ | 9382/30000 [6:00:33<13:08:03, 2.29s/it] 31%|███▏ | 9383/30000 [6:00:36<13:11:08, 2.30s/it] 31%|███▏ | 9384/30000 [6:00:38<13:14:21, 2.31s/it] 31%|███▏ | 9385/30000 [6:00:40<13:07:27, 2.29s/it] 31%|███▏ | 9386/30000 [6:00:43<13:07:14, 2.29s/it] 31%|███▏ | 9387/30000 [6:00:45<13:10:03, 2.30s/it] 31%|███▏ | 9388/30000 [6:00:47<13:05:42, 2.29s/it] 31%|███▏ | 9389/30000 [6:00:49<13:02:25, 2.28s/it] 31%|███▏ | 9390/30000 [6:00:52<13:09:16, 2.30s/it] 31%|███▏ | 9390/30000 [6:00:52<13:09:16, 2.30s/it] 31%|███▏ | 9391/30000 [6:00:54<13:10:22, 2.30s/it] 31%|███▏ | 9392/30000 [6:00:56<13:08:37, 2.30s/it] 31%|███▏ | 9393/30000 [6:00:59<13:07:52, 2.29s/it] 31%|███▏ | 9394/30000 [6:01:01<13:15:02, 2.31s/it] 31%|███▏ | 9395/30000 [6:01:03<13:16:32, 2.32s/it] 31%|███▏ | 9396/30000 [6:01:06<13:09:11, 2.30s/it] 31%|███▏ | 9397/30000 [6:01:08<13:06:42, 2.29s/it] 31%|███▏ | 9398/30000 [6:01:10<13:07:52, 2.29s/it] 31%|███▏ | 9399/30000 [6:01:12<13:08:05, 2.30s/it] 31%|███▏ | 9400/30000 [6:01:15<13:09:16, 2.30s/it] 31%|███▏ | 9400/30000 [6:01:15<13:09:16, 2.30s/it] 31%|███▏ | 9401/30000 [6:01:17<13:07:14, 2.29s/it] 31%|███▏ | 9402/30000 [6:01:19<13:13:20, 2.31s/it] 31%|███▏ | 9403/30000 [6:01:22<13:23:30, 2.34s/it] 31%|███▏ | 9404/30000 [6:01:24<13:17:35, 2.32s/it] 31%|███▏ | 9405/30000 [6:01:26<13:15:09, 2.32s/it] 31%|███▏ | 9406/30000 [6:01:29<13:13:39, 2.31s/it] 31%|███▏ | 9407/30000 [6:01:31<13:13:55, 2.31s/it] 31%|███▏ | 9408/30000 [6:01:33<13:08:07, 2.30s/it] 31%|███▏ | 9409/30000 [6:01:36<13:13:01, 2.31s/it] 31%|███▏ | 9410/30000 [6:01:38<13:09:04, 2.30s/it] 31%|███▏ | 9410/30000 [6:01:38<13:09:04, 2.30s/it] 31%|███▏ | 9411/30000 [6:01:40<13:07:59, 2.30s/it] 31%|███▏ | 9412/30000 [6:01:42<13:05:01, 2.29s/it] 31%|███▏ | 9413/30000 [6:01:45<13:10:03, 2.30s/it] 31%|███▏ | 9414/30000 [6:01:47<13:07:46, 2.30s/it] 31%|███▏ | 9415/30000 [6:01:49<13:07:38, 2.30s/it] 31%|███▏ | 9416/30000 [6:01:52<13:06:57, 2.29s/it] 31%|███▏ | 9417/30000 [6:01:54<13:06:54, 2.29s/it] 31%|███▏ | 9418/30000 [6:01:56<13:03:35, 2.28s/it] 31%|███▏ | 9419/30000 [6:01:58<12:59:12, 2.27s/it] 31%|███▏ | 9420/30000 [6:02:01<13:04:12, 2.29s/it] 31%|███▏ | 9420/30000 [6:02:01<13:04:12, 2.29s/it] 31%|███▏ | 9421/30000 [6:02:03<13:09:55, 2.30s/it] 31%|███▏ | 9422/30000 [6:02:05<13:10:22, 2.30s/it] 31%|███▏ | 9423/30000 [6:02:08<13:09:41, 2.30s/it] 31%|███▏ | 9424/30000 [6:02:10<13:10:42, 2.31s/it] 31%|███▏ | 9425/30000 [6:02:12<13:05:24, 2.29s/it] 31%|███▏ | 9426/30000 [6:02:15<13:02:27, 2.28s/it] 31%|███▏ | 9427/30000 [6:02:17<13:06:32, 2.29s/it] 31%|███▏ | 9428/30000 [6:02:19<13:07:26, 2.30s/it] 31%|███▏ | 9429/30000 [6:02:21<13:04:11, 2.29s/it] 31%|███▏ | 9430/30000 [6:02:24<12:58:12, 2.27s/it] 31%|███▏ | 9430/30000 [6:02:24<12:58:12, 2.27s/it] 31%|███▏ | 9431/30000 [6:02:26<13:06:48, 2.30s/it] 31%|███▏ | 9432/30000 [6:02:28<13:12:51, 2.31s/it] 31%|███▏ | 9433/30000 [6:02:31<13:15:09, 2.32s/it] 31%|███▏ | 9434/30000 [6:02:33<13:10:41, 2.31s/it] 31%|███▏ | 9435/30000 [6:02:35<13:11:24, 2.31s/it] 31%|███▏ | 9436/30000 [6:02:38<13:10:13, 2.31s/it] 31%|███▏ | 9437/30000 [6:02:40<13:01:44, 2.28s/it] 31%|███▏ | 9438/30000 [6:02:42<13:01:59, 2.28s/it] 31%|███▏ | 9439/30000 [6:02:44<13:06:10, 2.29s/it] 31%|███▏ | 9440/30000 [6:02:47<13:11:03, 2.31s/it] 31%|███▏ | 9440/30000 [6:02:47<13:11:03, 2.31s/it] 31%|███▏ | 9441/30000 [6:02:49<13:05:21, 2.29s/it] 31%|███▏ | 9442/30000 [6:02:51<13:04:14, 2.29s/it] 31%|███▏ | 9443/30000 [6:02:54<13:11:38, 2.31s/it] 31%|███▏ | 9444/30000 [6:02:56<13:14:56, 2.32s/it] 31%|███▏ | 9445/30000 [6:02:58<13:05:24, 2.29s/it] 31%|███▏ | 9446/30000 [6:03:00<13:03:38, 2.29s/it] 31%|███▏ | 9447/30000 [6:03:03<13:06:55, 2.30s/it] 31%|███▏ | 9448/30000 [6:03:05<13:05:47, 2.29s/it] 31%|███▏ | 9449/30000 [6:03:07<13:06:35, 2.30s/it] 32%|███▏ | 9450/30000 [6:03:10<13:07:37, 2.30s/it] 32%|███▏ | 9450/30000 [6:03:10<13:07:37, 2.30s/it] 32%|███▏ | 9451/30000 [6:03:12<13:14:24, 2.32s/it] 32%|███▏ | 9452/30000 [6:03:14<13:12:56, 2.32s/it] 32%|███▏ | 9453/30000 [6:03:17<13:04:12, 2.29s/it] 32%|███▏ | 9454/30000 [6:03:19<13:01:13, 2.28s/it] 32%|███▏ | 9455/30000 [6:03:21<13:07:48, 2.30s/it] 32%|███▏ | 9456/30000 [6:03:23<13:05:38, 2.29s/it] 32%|███▏ | 9457/30000 [6:03:26<13:02:18, 2.28s/it] 32%|███▏ | 9458/30000 [6:03:28<13:06:36, 2.30s/it] 32%|███▏ | 9459/30000 [6:03:30<13:07:46, 2.30s/it] 32%|███▏ | 9460/30000 [6:03:33<13:02:20, 2.29s/it] 32%|███▏ | 9460/30000 [6:03:33<13:02:20, 2.29s/it] 32%|███▏ | 9461/30000 [6:03:35<13:03:32, 2.29s/it] 32%|███▏ | 9462/30000 [6:03:37<13:09:51, 2.31s/it] 32%|███▏ | 9463/30000 [6:03:40<13:10:15, 2.31s/it] 32%|███▏ | 9464/30000 [6:03:42<13:05:33, 2.30s/it] 32%|███▏ | 9465/30000 [6:03:44<13:07:46, 2.30s/it] 32%|███▏ | 9466/30000 [6:03:47<13:12:28, 2.32s/it] 32%|███▏ | 9467/30000 [6:03:49<13:11:48, 2.31s/it] 32%|███▏ | 9468/30000 [6:03:51<13:07:38, 2.30s/it] 32%|███▏ | 9469/30000 [6:03:53<13:02:13, 2.29s/it] 32%|███▏ | 9470/30000 [6:03:56<13:06:02, 2.30s/it] 32%|███▏ | 9470/30000 [6:03:56<13:06:02, 2.30s/it] 32%|███▏ | 9471/30000 [6:03:58<13:07:33, 2.30s/it] 32%|███▏ | 9472/30000 [6:04:00<13:05:42, 2.30s/it] 32%|███▏ | 9473/30000 [6:04:03<13:03:53, 2.29s/it] 32%|███▏ | 9474/30000 [6:04:05<13:08:08, 2.30s/it] 32%|███▏ | 9475/30000 [6:04:07<13:03:31, 2.29s/it] 32%|███▏ | 9476/30000 [6:04:09<13:01:31, 2.28s/it] 32%|███▏ | 9477/30000 [6:04:12<13:11:39, 2.31s/it] 32%|███▏ | 9478/30000 [6:04:14<13:08:01, 2.30s/it] 32%|███▏ | 9479/30000 [6:04:16<13:07:54, 2.30s/it] 32%|███▏ | 9480/30000 [6:04:19<13:04:13, 2.29s/it] 32%|███▏ | 9480/30000 [6:04:19<13:04:13, 2.29s/it] 32%|███▏ | 9481/30000 [6:04:21<13:10:35, 2.31s/it] 32%|███▏ | 9482/30000 [6:04:23<13:12:33, 2.32s/it] 32%|███▏ | 9483/30000 [6:04:26<13:10:52, 2.31s/it] 32%|███▏ | 9484/30000 [6:04:28<13:04:30, 2.29s/it] 32%|███▏ | 9485/30000 [6:04:30<13:08:47, 2.31s/it] 32%|███▏ | 9486/30000 [6:04:33<13:04:43, 2.30s/it] 32%|███▏ | 9487/30000 [6:04:35<13:08:36, 2.31s/it] 32%|███▏ | 9488/30000 [6:04:37<13:06:33, 2.30s/it] 32%|███▏ | 9489/30000 [6:04:39<13:10:42, 2.31s/it] 32%|███▏ | 9490/30000 [6:04:42<13:06:39, 2.30s/it] 32%|███▏ | 9490/30000 [6:04:42<13:06:39, 2.30s/it] 32%|███▏ | 9491/30000 [6:04:44<13:04:57, 2.30s/it] 32%|███▏ | 9492/30000 [6:04:46<13:10:02, 2.31s/it] 32%|███▏ | 9493/30000 [6:04:49<13:06:55, 2.30s/it] 32%|███▏ | 9494/30000 [6:04:51<13:00:20, 2.28s/it] 32%|███▏ | 9495/30000 [6:04:53<13:01:53, 2.29s/it] 32%|███▏ | 9496/30000 [6:04:55<13:03:41, 2.29s/it] 32%|███▏ | 9497/30000 [6:04:58<13:00:59, 2.29s/it] 32%|███▏ | 9498/30000 [6:05:00<12:55:30, 2.27s/it] 32%|███▏ | 9499/30000 [6:05:02<13:04:37, 2.30s/it] 32%|███▏ | 9500/30000 [6:05:05<13:02:50, 2.29s/it] 32%|███▏ | 9500/30000 [6:05:05<13:02:50, 2.29s/it] 32%|███▏ | 9501/30000 [6:05:07<13:02:19, 2.29s/it] 32%|███▏ | 9502/30000 [6:05:09<13:07:12, 2.30s/it] 32%|███▏ | 9503/30000 [6:05:12<13:03:53, 2.29s/it] 32%|███▏ | 9504/30000 [6:05:14<13:02:48, 2.29s/it] 32%|███▏ | 9505/30000 [6:05:16<13:03:49, 2.29s/it] 32%|███▏ | 9506/30000 [6:05:18<13:07:41, 2.31s/it] 32%|███▏ | 9507/30000 [6:05:21<13:11:23, 2.32s/it] 32%|███▏ | 9508/30000 [6:05:23<13:07:37, 2.31s/it] 32%|███▏ | 9509/30000 [6:05:25<13:08:19, 2.31s/it] 32%|███▏ | 9510/30000 [6:05:28<13:06:53, 2.30s/it] 32%|███▏ | 9510/30000 [6:05:28<13:06:53, 2.30s/it] 32%|███▏ | 9511/30000 [6:05:30<13:05:28, 2.30s/it] 32%|███▏ | 9512/30000 [6:05:32<13:05:38, 2.30s/it] 32%|███▏ | 9513/30000 [6:05:35<13:12:41, 2.32s/it] 32%|███▏ | 9514/30000 [6:05:37<13:12:26, 2.32s/it] 32%|███▏ | 9515/30000 [6:05:39<13:06:54, 2.30s/it] 32%|███▏ | 9516/30000 [6:05:42<13:10:32, 2.32s/it] 32%|███▏ | 9517/30000 [6:05:44<13:08:19, 2.31s/it] 32%|███▏ | 9518/30000 [6:05:46<13:05:23, 2.30s/it] 32%|███▏ | 9519/30000 [6:05:49<13:13:05, 2.32s/it] 32%|███▏ | 9520/30000 [6:05:51<13:16:33, 2.33s/it] 32%|███▏ | 9520/30000 [6:05:51<13:16:33, 2.33s/it] 32%|███▏ | 9521/30000 [6:05:53<13:07:26, 2.31s/it] 32%|███▏ | 9522/30000 [6:05:55<12:57:47, 2.28s/it] 32%|███▏ | 9523/30000 [6:05:58<13:01:29, 2.29s/it] 32%|███▏ | 9524/30000 [6:06:00<13:03:14, 2.30s/it] 32%|███▏ | 9525/30000 [6:06:02<13:06:05, 2.30s/it] 32%|███▏ | 9526/30000 [6:06:05<13:08:52, 2.31s/it] 32%|███▏ | 9527/30000 [6:06:07<13:11:45, 2.32s/it] 32%|███▏ | 9528/30000 [6:06:09<13:06:28, 2.31s/it] 32%|███▏ | 9529/30000 [6:06:12<13:06:35, 2.31s/it] 32%|███▏ | 9530/30000 [6:06:14<13:10:41, 2.32s/it] 32%|███▏ | 9530/30000 [6:06:14<13:10:41, 2.32s/it] 32%|███▏ | 9531/30000 [6:06:16<13:09:58, 2.32s/it] 32%|███▏ | 9532/30000 [6:06:18<13:03:57, 2.30s/it] 32%|███▏ | 9533/30000 [6:06:21<13:04:12, 2.30s/it] 32%|███▏ | 9534/30000 [6:06:23<13:01:22, 2.29s/it] 32%|███▏ | 9535/30000 [6:06:25<13:00:36, 2.29s/it] 32%|███▏ | 9536/30000 [6:06:28<13:03:19, 2.30s/it] 32%|███▏ | 9537/30000 [6:06:30<13:05:27, 2.30s/it] 32%|███▏ | 9538/30000 [6:06:32<13:10:05, 2.32s/it] 32%|███▏ | 9539/30000 [6:06:35<13:06:58, 2.31s/it] 32%|███▏ | 9540/30000 [6:06:37<13:08:50, 2.31s/it] 32%|███▏ | 9540/30000 [6:06:37<13:08:50, 2.31s/it] 32%|███▏ | 9541/30000 [6:06:39<13:06:04, 2.31s/it] 32%|███▏ | 9542/30000 [6:06:41<13:03:08, 2.30s/it] 32%|███▏ | 9543/30000 [6:06:44<13:05:05, 2.30s/it] 32%|███▏ | 9544/30000 [6:06:46<13:04:18, 2.30s/it] 32%|███▏ | 9545/30000 [6:06:48<12:56:41, 2.28s/it] 32%|███▏ | 9546/30000 [6:06:51<12:56:42, 2.28s/it] 32%|███▏ | 9547/30000 [6:06:53<13:02:28, 2.30s/it] 32%|███▏ | 9548/30000 [6:06:55<12:57:25, 2.28s/it] 32%|███▏ | 9549/30000 [6:06:57<12:52:47, 2.27s/it] 32%|███▏ | 9550/30000 [6:07:00<12:56:58, 2.28s/it] 32%|███▏ | 9550/30000 [6:07:00<12:56:58, 2.28s/it] 32%|███▏ | 9551/30000 [6:07:02<13:03:19, 2.30s/it] 32%|███▏ | 9552/30000 [6:07:04<12:55:20, 2.28s/it] 32%|███▏ | 9553/30000 [6:07:07<13:02:03, 2.29s/it] 32%|███▏ | 9554/30000 [6:07:09<13:06:50, 2.31s/it] 32%|███▏ | 9555/30000 [6:07:11<13:00:22, 2.29s/it] 32%|███▏ | 9556/30000 [6:07:14<13:07:52, 2.31s/it] 32%|███▏ | 9557/30000 [6:07:16<13:10:53, 2.32s/it] 32%|███▏ | 9558/30000 [6:07:18<13:06:28, 2.31s/it] 32%|███▏ | 9559/30000 [6:07:21<13:10:26, 2.32s/it] 32%|███▏ | 9560/30000 [6:07:23<13:08:54, 2.32s/it] 32%|███▏ | 9560/30000 [6:07:23<13:08:54, 2.32s/it] 32%|███▏ | 9561/30000 [6:07:25<13:08:36, 2.32s/it] 32%|███▏ | 9562/30000 [6:07:27<13:05:11, 2.31s/it] 32%|███▏ | 9563/30000 [6:07:30<13:07:57, 2.31s/it] 32%|███▏ | 9564/30000 [6:07:32<12:59:06, 2.29s/it] 32%|███▏ | 9565/30000 [6:07:34<12:55:00, 2.28s/it] 32%|███▏ | 9566/30000 [6:07:37<12:56:42, 2.28s/it] 32%|███▏ | 9567/30000 [6:07:39<12:54:37, 2.27s/it] 32%|███▏ | 9568/30000 [6:07:41<12:57:09, 2.28s/it] 32%|███▏ | 9569/30000 [6:07:43<13:07:18, 2.31s/it] 32%|███▏ | 9570/30000 [6:07:46<13:08:54, 2.32s/it] +{'loss': 0.0175, 'grad_norm': 0.10165403038263321, 'learning_rate': 2.4951884631780578e-05, 'flow_matching_loss': 0.017475182097405194, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.67} +{'loss': 0.0155, 'grad_norm': 0.06392738968133926, 'learning_rate': 2.493950711750518e-05, 'flow_matching_loss': 0.01548903388902545, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.67} +{'loss': 0.0183, 'grad_norm': 0.10442104190587997, 'learning_rate': 2.49271175257914e-05, 'flow_matching_loss': 0.01827444154769182, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.67} +{'loss': 0.0167, 'grad_norm': 0.09201356023550034, 'learning_rate': 2.4914715871693757e-05, 'flow_matching_loss': 0.01667042775079608, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.68} +{'loss': 0.0175, 'grad_norm': 0.11368203908205032, 'learning_rate': 2.4902302170281438e-05, 'flow_matching_loss': 0.01748319622129202, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.68} +{'loss': 0.0168, 'grad_norm': 0.10431733727455139, 'learning_rate': 2.4889876436638252e-05, 'flow_matching_loss': 0.016765958070755003, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.68} +{'loss': 0.0164, 'grad_norm': 0.09550170600414276, 'learning_rate': 2.4877438685862643e-05, 'flow_matching_loss': 0.016432017739862203, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.69} +{'loss': 0.015, 'grad_norm': 0.07246292382478714, 'learning_rate': 2.486498893306765e-05, 'flow_matching_loss': 0.014995853323489428, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.69} +{'loss': 0.016, 'grad_norm': 0.0898195207118988, 'learning_rate': 2.485252719338089e-05, 'flow_matching_loss': 0.016036652494221926, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.69} +{'loss': 0.0169, 'grad_norm': 0.09035919606685638, 'learning_rate': 2.4840053481944563e-05, 'flow_matching_loss': 0.01691264696419239, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.69} +{'loss': 0.0157, 'grad_norm': 0.09571180492639542, 'learning_rate': 2.4827567813915398e-05, 'flow_matching_loss': 0.015686024725437165, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.7} +{'loss': 0.0171, 'grad_norm': 0.07744565606117249, 'learning_rate': 2.481507020446466e-05, 'flow_matching_loss': 0.01712434496730566, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.7} +{'loss': 0.0163, 'grad_norm': 0.0672847181558609, 'learning_rate': 2.4802560668778122e-05, 'flow_matching_loss': 0.01630834210664034, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.7} +{'loss': 0.0157, 'grad_norm': 0.0879279226064682, 'learning_rate': 2.479003922205605e-05, 'flow_matching_loss': 0.015662869717925788, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.71} +{'loss': 0.0159, 'grad_norm': 0.06991494446992874, 'learning_rate': 2.4777505879513183e-05, 'flow_matching_loss': 0.01594325704500079, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.71} +{'loss': 0.0152, 'grad_norm': 0.09269137680530548, 'learning_rate': 2.4764960656378714e-05, 'flow_matching_loss': 0.015211998857557774, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.71} +{'loss': 0.0141, 'grad_norm': 0.08727968484163284, 'learning_rate': 2.4752403567896274e-05, 'flow_matching_loss': 0.014068638160824775, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.71} +{'loss': 0.0144, 'grad_norm': 0.09582386165857315, 'learning_rate': 2.47398346293239e-05, 'flow_matching_loss': 0.014367791917175054, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.72} +{'loss': 0.0162, 'grad_norm': 0.10427840054035187, 'learning_rate': 2.4727253855934055e-05, 'flow_matching_loss': 0.016246006544679403, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.72} +{'loss': 0.0162, 'grad_norm': 0.06124483048915863, 'learning_rate': 2.4714661263013548e-05, 'flow_matching_loss': 0.0161787249147892, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.72} +{'loss': 0.0161, 'grad_norm': 0.10400816798210144, 'learning_rate': 2.4702056865863575e-05, 'flow_matching_loss': 0.016100141033530236, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.73} +{'loss': 0.0163, 'grad_norm': 0.08667192608118057, 'learning_rate': 2.4689440679799672e-05, 'flow_matching_loss': 0.016270245891064405, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.73} +{'loss': 0.0152, 'grad_norm': 0.09287340193986893, 'learning_rate': 2.4676812720151678e-05, 'flow_matching_loss': 0.015190940909087658, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.73} +{'loss': 0.0156, 'grad_norm': 0.06336111575365067, 'learning_rate': 2.466417300226377e-05, 'flow_matching_loss': 0.015566894691437483, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.73} +{'loss': 0.016, 'grad_norm': 0.07335017621517181, 'learning_rate': 2.4651521541494394e-05, 'flow_matching_loss': 0.01602795282378793, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.74} +{'loss': 0.0165, 'grad_norm': 0.0796826034784317, 'learning_rate': 2.4638858353216267e-05, 'flow_matching_loss': 0.016490754764527084, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.74} +{'loss': 0.0167, 'grad_norm': 0.0941772311925888, 'learning_rate': 2.4626183452816355e-05, 'flow_matching_loss': 0.016662957798689603, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.74} +{'loss': 0.0147, 'grad_norm': 0.07162295281887054, 'learning_rate': 2.4613496855695855e-05, 'flow_matching_loss': 0.01469989325851202, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.75} +{'loss': 0.0144, 'grad_norm': 0.07710712403059006, 'learning_rate': 2.4600798577270183e-05, 'flow_matching_loss': 0.014436995424330235, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.75} +{'loss': 0.0161, 'grad_norm': 0.09488355368375778, 'learning_rate': 2.4588088632968944e-05, 'flow_matching_loss': 0.016104507073760033, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.75} +{'loss': 0.0155, 'grad_norm': 0.08227191120386124, 'learning_rate': 2.4575367038235916e-05, 'flow_matching_loss': 0.015470258146524429, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.76} +{'loss': 0.0154, 'grad_norm': 0.092167928814888, 'learning_rate': 2.4562633808529043e-05, 'flow_matching_loss': 0.015377181861549617, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.76} +{'loss': 0.0169, 'grad_norm': 0.11207162588834763, 'learning_rate': 2.4549888959320396e-05, 'flow_matching_loss': 0.016913816425949336, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.76} +{'loss': 0.0157, 'grad_norm': 0.05276364088058472, 'learning_rate': 2.4537132506096162e-05, 'flow_matching_loss': 0.015725202020257713, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.76} +{'loss': 0.0143, 'grad_norm': 0.10448216646909714, 'learning_rate': 2.452436446435664e-05, 'flow_matching_loss': 0.014307956118136645, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.77} +{'loss': 0.0176, 'grad_norm': 0.09501653909683228, 'learning_rate': 2.451158484961621e-05, 'flow_matching_loss': 0.017557715345174075, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.77} +{'loss': 0.0177, 'grad_norm': 0.09256640821695328, 'learning_rate': 2.4498793677403304e-05, 'flow_matching_loss': 0.017688268143683673, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.77} +{'loss': 0.0172, 'grad_norm': 0.0798371285200119, 'learning_rate': 2.4485990963260398e-05, 'flow_matching_loss': 0.017225933074951173, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.78} +{'loss': 0.0164, 'grad_norm': 0.10337930917739868, 'learning_rate': 2.4473176722744e-05, 'flow_matching_loss': 0.016357009205967186, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.78} +{'loss': 0.0182, 'grad_norm': 0.08487426489591599, 'learning_rate': 2.446035097142463e-05, 'flow_matching_loss': 0.018202284071594476, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.78} +{'loss': 0.0169, 'grad_norm': 0.08879834413528442, 'learning_rate': 2.444751372488677e-05, 'flow_matching_loss': 0.016866580676287414, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.78} 32%|███▏ | 9570/30000 [6:07:46<13:08:54, 2.32s/it] 32%|███▏ | 9571/30000 [6:07:48<13:05:34, 2.31s/it] 32%|███▏ | 9572/30000 [6:07:50<13:03:23, 2.30s/it] 32%|███▏ | 9573/30000 [6:07:53<13:03:52, 2.30s/it] 32%|███▏ | 9574/30000 [6:07:55<12:58:41, 2.29s/it] 32%|███▏ | 9575/30000 [6:07:57<12:59:21, 2.29s/it] 32%|███▏ | 9576/30000 [6:08:00<13:09:07, 2.32s/it] 32%|███▏ | 9577/30000 [6:08:02<13:10:15, 2.32s/it] 32%|███▏ | 9578/30000 [6:08:04<13:04:55, 2.31s/it] 32%|███▏ | 9579/30000 [6:08:07<13:08:03, 2.32s/it] 32%|███▏ | 9580/30000 [6:08:09<13:06:58, 2.31s/it] 32%|███▏ | 9580/30000 [6:08:09<13:06:58, 2.31s/it] 32%|███▏ | 9581/30000 [6:08:11<13:01:50, 2.30s/it] 32%|███▏ | 9582/30000 [6:08:13<13:08:17, 2.32s/it] 32%|███▏ | 9583/30000 [6:08:16<13:08:37, 2.32s/it] 32%|███▏ | 9584/30000 [6:08:18<13:04:41, 2.31s/it] 32%|███▏ | 9585/30000 [6:08:20<13:03:04, 2.30s/it] 32%|███▏ | 9586/30000 [6:08:23<13:06:25, 2.31s/it] 32%|███▏ | 9587/30000 [6:08:25<13:01:35, 2.30s/it] 32%|███▏ | 9588/30000 [6:08:27<12:59:39, 2.29s/it] 32%|███▏ | 9589/30000 [6:08:30<13:04:04, 2.30s/it] 32%|███▏ | 9590/30000 [6:08:32<13:04:00, 2.30s/it] 32%|███▏ | 9590/30000 [6:08:32<13:04:00, 2.30s/it] 32%|███▏ | 9591/30000 [6:08:34<13:01:42, 2.30s/it] 32%|███▏ | 9592/30000 [6:08:36<13:02:06, 2.30s/it] 32%|███▏ | 9593/30000 [6:08:39<13:03:52, 2.30s/it] 32%|███▏ | 9594/30000 [6:08:41<12:59:13, 2.29s/it] 32%|███▏ | 9595/30000 [6:08:43<12:56:59, 2.28s/it] 32%|███▏ | 9596/30000 [6:08:46<13:04:17, 2.31s/it] 32%|███▏ | 9597/30000 [6:08:48<13:04:32, 2.31s/it] 32%|███▏ | 9598/30000 [6:08:50<13:02:25, 2.30s/it] 32%|███▏ | 9599/30000 [6:08:53<13:00:36, 2.30s/it] 32%|███▏ | 9600/30000 [6:08:55<13:06:44, 2.31s/it] 32%|███▏ | 9600/30000 [6:08:55<13:06:44, 2.31s/it] 32%|███▏ | 9601/30000 [6:08:57<13:03:14, 2.30s/it] 32%|███▏ | 9602/30000 [6:08:59<12:59:20, 2.29s/it] 32%|███▏ | 9603/30000 [6:09:02<13:03:39, 2.31s/it] 32%|███▏ | 9604/30000 [6:09:04<13:03:34, 2.31s/it] 32%|███▏ | 9605/30000 [6:09:06<12:59:20, 2.29s/it] 32%|███▏ | 9606/30000 [6:09:09<12:50:19, 2.27s/it] 32%|███▏ | 9607/30000 [6:09:11<13:00:39, 2.30s/it] 32%|███▏ | 9608/30000 [6:09:13<13:00:57, 2.30s/it] 32%|███▏ | 9609/30000 [6:09:16<13:00:42, 2.30s/it] 32%|███▏ | 9610/30000 [6:09:18<13:01:29, 2.30s/it] 32%|███▏ | 9610/30000 [6:09:18<13:01:29, 2.30s/it] 32%|███▏ | 9611/30000 [6:09:20<13:03:25, 2.31s/it] 32%|███▏ | 9612/30000 [6:09:22<13:00:08, 2.30s/it] 32%|███▏ | 9613/30000 [6:09:25<12:58:00, 2.29s/it] 32%|███▏ | 9614/30000 [6:09:27<13:04:25, 2.31s/it] 32%|███▏ | 9615/30000 [6:09:29<13:06:37, 2.32s/it] 32%|███▏ | 9616/30000 [6:09:32<13:05:47, 2.31s/it] 32%|███▏ | 9617/30000 [6:09:34<13:03:13, 2.31s/it] 32%|███▏ | 9618/30000 [6:09:36<13:10:24, 2.33s/it] 32%|███▏ | 9619/30000 [6:09:39<13:08:13, 2.32s/it] 32%|███▏ | 9620/30000 [6:09:41<13:00:51, 2.30s/it] 32%|███▏ | 9620/30000 [6:09:41<13:00:51, 2.30s/it] 32%|███▏ | 9621/30000 [6:09:43<12:57:22, 2.29s/it] 32%|███▏ | 9622/30000 [6:09:46<13:01:50, 2.30s/it] 32%|███▏ | 9623/30000 [6:09:48<12:57:35, 2.29s/it] 32%|███▏ | 9624/30000 [6:09:50<12:56:32, 2.29s/it] 32%|███▏ | 9625/30000 [6:09:52<12:59:34, 2.30s/it] 32%|███▏ | 9626/30000 [6:09:55<13:02:22, 2.30s/it] 32%|███▏ | 9627/30000 [6:09:57<13:01:11, 2.30s/it] 32%|███▏ | 9628/30000 [6:09:59<13:00:45, 2.30s/it] 32%|███▏ | 9629/30000 [6:10:02<13:07:58, 2.32s/it] 32%|███▏ | 9630/30000 [6:10:04<13:04:51, 2.31s/it] 32%|███▏ | 9630/30000 [6:10:04<13:04:51, 2.31s/it] 32%|███▏ | 9631/30000 [6:10:06<13:01:54, 2.30s/it] 32%|███▏ | 9632/30000 [6:10:08<12:55:48, 2.29s/it] 32%|███▏ | 9633/30000 [6:10:11<13:02:59, 2.31s/it] 32%|███▏ | 9634/30000 [6:10:13<13:01:13, 2.30s/it] 32%|███▏ | 9635/30000 [6:10:15<12:58:05, 2.29s/it] 32%|███▏ | 9636/30000 [6:10:18<12:56:01, 2.29s/it] 32%|███▏ | 9637/30000 [6:10:20<13:06:43, 2.32s/it] 32%|███▏ | 9638/30000 [6:10:22<13:01:56, 2.30s/it] 32%|███▏ | 9639/30000 [6:10:25<13:01:28, 2.30s/it] 32%|███▏ | 9640/30000 [6:10:27<13:05:48, 2.32s/it] 32%|███▏ | 9640/30000 [6:10:27<13:05:48, 2.32s/it] 32%|███▏ | 9641/30000 [6:10:29<13:11:03, 2.33s/it] 32%|███▏ | 9642/30000 [6:10:32<13:04:00, 2.31s/it] 32%|███▏ | 9643/30000 [6:10:34<12:58:29, 2.29s/it] 32%|███▏ | 9644/30000 [6:10:36<13:01:37, 2.30s/it] 32%|███▏ | 9645/30000 [6:10:38<13:03:10, 2.31s/it] 32%|███▏ | 9646/30000 [6:10:41<12:57:53, 2.29s/it] 32%|███▏ | 9647/30000 [6:10:43<12:56:11, 2.29s/it] 32%|███▏ | 9648/30000 [6:10:45<13:01:15, 2.30s/it] 32%|███▏ | 9649/30000 [6:10:48<12:58:05, 2.29s/it] 32%|███▏ | 9650/30000 [6:10:50<12:52:03, 2.28s/it] 32%|███▏ | 9650/30000 [6:10:50<12:52:03, 2.28s/it] 32%|███▏ | 9651/30000 [6:10:52<12:54:28, 2.28s/it] 32%|███▏ | 9652/30000 [6:10:54<12:55:28, 2.29s/it] 32%|███▏ | 9653/30000 [6:10:57<12:53:44, 2.28s/it] 32%|███▏ | 9654/30000 [6:10:59<12:54:03, 2.28s/it] 32%|███▏ | 9655/30000 [6:11:01<12:59:32, 2.30s/it] 32%|███▏ | 9656/30000 [6:11:04<13:02:37, 2.31s/it] 32%|███▏ | 9657/30000 [6:11:06<12:54:54, 2.29s/it] 32%|███▏ | 9658/30000 [6:11:08<12:55:55, 2.29s/it] 32%|███▏ | 9659/30000 [6:11:11<12:58:16, 2.30s/it] 32%|███▏ | 9660/30000 [6:11:13<12:59:19, 2.30s/it] 32%|███▏ | 9660/30000 [6:11:13<12:59:19, 2.30s/it] 32%|███▏ | 9661/30000 [6:11:15<12:54:05, 2.28s/it] 32%|███▏ | 9662/30000 [6:11:17<12:52:36, 2.28s/it] 32%|███▏ | 9663/30000 [6:11:20<12:58:27, 2.30s/it] 32%|███▏ | 9664/30000 [6:11:22<13:00:34, 2.30s/it] 32%|███▏ | 9665/30000 [6:11:24<12:56:43, 2.29s/it] 32%|███▏ | 9666/30000 [6:11:27<12:53:50, 2.28s/it] 32%|███▏ | 9667/30000 [6:11:29<12:59:23, 2.30s/it] 32%|███▏ | 9668/30000 [6:11:31<13:00:23, 2.30s/it] 32%|███▏ | 9669/30000 [6:11:33<13:01:12, 2.31s/it] 32%|███▏ | 9670/30000 [6:11:36<12:59:25, 2.30s/it] 32%|███▏ | 9670/30000 [6:11:36<12:59:25, 2.30s/it] 32%|███▏ | 9671/30000 [6:11:38<13:05:33, 2.32s/it] 32%|███▏ | 9672/30000 [6:11:40<13:05:51, 2.32s/it] 32%|███▏ | 9673/30000 [6:11:43<13:00:51, 2.30s/it] 32%|███▏ | 9674/30000 [6:11:45<12:58:34, 2.30s/it] 32%|███▏ | 9675/30000 [6:11:47<13:00:47, 2.30s/it] 32%|███▏ | 9676/30000 [6:11:50<12:53:35, 2.28s/it] 32%|███▏ | 9677/30000 [6:11:52<12:55:14, 2.29s/it] 32%|███▏ | 9678/30000 [6:11:54<12:57:53, 2.30s/it] 32%|███▏ | 9679/30000 [6:11:57<13:00:19, 2.30s/it] 32%|███▏ | 9680/30000 [6:11:59<12:57:13, 2.29s/it] 32%|███▏ | 9680/30000 [6:11:59<12:57:13, 2.29s/it] 32%|███▏ | 9681/30000 [6:12:01<12:57:22, 2.30s/it] 32%|███▏ | 9682/30000 [6:12:03<13:02:34, 2.31s/it] 32%|███▏ | 9683/30000 [6:12:06<13:01:23, 2.31s/it] 32%|███▏ | 9684/30000 [6:12:08<12:56:49, 2.29s/it] 32%|███▏ | 9685/30000 [6:12:10<12:52:59, 2.28s/it] 32%|███▏ | 9686/30000 [6:12:13<12:57:49, 2.30s/it] 32%|███▏ | 9687/30000 [6:12:15<13:01:55, 2.31s/it] 32%|███▏ | 9688/30000 [6:12:17<12:51:42, 2.28s/it] 32%|███▏ | 9689/30000 [6:12:19<12:48:23, 2.27s/it] 32%|███▏ | 9690/30000 [6:12:22<12:53:04, 2.28s/it] 32%|███▏ | 9690/30000 [6:12:22<12:53:04, 2.28s/it] 32%|███▏ | 9691/30000 [6:12:24<12:53:18, 2.28s/it] 32%|███▏ | 9692/30000 [6:12:26<12:50:57, 2.28s/it] 32%|███▏ | 9693/30000 [6:12:29<12:58:09, 2.30s/it] 32%|███▏ | 9694/30000 [6:12:31<12:58:26, 2.30s/it] 32%|███▏ | 9695/30000 [6:12:33<12:57:08, 2.30s/it] 32%|███▏ | 9696/30000 [6:12:35<12:56:52, 2.30s/it] 32%|███▏ | 9697/30000 [6:12:38<13:00:13, 2.31s/it] 32%|███▏ | 9698/30000 [6:12:40<12:55:14, 2.29s/it] 32%|███▏ | 9699/30000 [6:12:42<12:52:13, 2.28s/it] 32%|███▏ | 9700/30000 [6:12:45<12:58:40, 2.30s/it] 32%|███▏ | 9700/30000 [6:12:45<12:58:40, 2.30s/it] 32%|███▏ | 9701/30000 [6:12:47<12:57:52, 2.30s/it] 32%|███▏ | 9702/30000 [6:12:49<12:53:51, 2.29s/it] 32%|███▏ | 9703/30000 [6:12:52<12:55:53, 2.29s/it] 32%|███▏ | 9704/30000 [6:12:54<13:05:02, 2.32s/it] 32%|███▏ | 9705/30000 [6:12:56<13:02:00, 2.31s/it] 32%|███▏ | 9706/30000 [6:12:58<12:58:57, 2.30s/it] 32%|███▏ | 9707/30000 [6:13:01<13:02:52, 2.31s/it] 32%|███▏ | 9708/30000 [6:13:03<13:00:01, 2.31s/it] 32%|███▏ | 9709/30000 [6:13:05<12:53:07, 2.29s/it] 32%|███▏ | 9710/30000 [6:13:08<12:50:33, 2.28s/it] 32%|███▏ | 9710/30000 [6:13:08<12:50:33, 2.28s/it] 32%|███▏ | 9711/30000 [6:13:10<12:58:49, 2.30s/it] 32%|███▏ | 9712/30000 [6:13:12<13:01:01, 2.31s/it] 32%|███▏ | 9713/30000 [6:13:15<12:57:17, 2.30s/it] 32%|███▏ | 9714/30000 [6:13:17<12:56:16, 2.30s/it] 32%|███▏ | 9715/30000 [6:13:19<12:56:27, 2.30s/it] 32%|███▏ | 9716/30000 [6:13:21<12:53:33, 2.29s/it] 32%|███▏ | 9717/30000 [6:13:24<12:52:34, 2.29s/it] 32%|███▏ | 9718/30000 [6:13:26<12:56:48, 2.30s/it] 32%|███▏ | 9719/30000 [6:13:28<13:01:08, 2.31s/it] 32%|███▏ | 9720/30000 [6:13:31<12:57:04, 2.30s/it] 32%|███▏ | 9720/30000 [6:13:31<12:57:04, 2.30s/it] 32%|███▏ | 9721/30000 [6:13:33<12:56:58, 2.30s/it] 32%|███▏ | 9722/30000 [6:13:35<13:00:23, 2.31s/it] 32%|███▏ | 9723/30000 [6:13:38<12:59:02, 2.31s/it] 32%|███▏ | 9724/30000 [6:13:40<12:53:41, 2.29s/it] 32%|███▏ | 9725/30000 [6:13:42<12:59:14, 2.31s/it] 32%|███▏ | 9726/30000 [6:13:44<12:59:13, 2.31s/it] 32%|███▏ | 9727/30000 [6:13:47<12:56:38, 2.30s/it] 32%|███▏ | 9728/30000 [6:13:49<12:57:51, 2.30s/it] 32%|███▏ | 9729/30000 [6:13:51<13:01:24, 2.31s/it] 32%|███▏ | 9730/30000 [6:13:54<12:59:37, 2.31s/it] 32%|███▏ | 9730/30000 [6:13:54<12:59:37, 2.31s/it] 32%|███▏ | 9731/30000 [6:13:56<12:54:56, 2.29s/it] 32%|███▏ | 9732/30000 [6:13:58<12:55:15, 2.30s/it] 32%|███▏ | 9733/30000 [6:14:01<12:56:56, 2.30s/it] 32%|███▏ | 9734/30000 [6:14:03<12:52:29, 2.29s/it] 32%|███▏ | 9735/30000 [6:14:05<12:53:26, 2.29s/it] 32%|███▏ | 9736/30000 [6:14:07<12:54:10, 2.29s/it] 32%|███▏ | 9737/30000 [6:14:10<12:52:15, 2.29s/it] 32%|███▏ | 9738/30000 [6:14:12<12:49:22, 2.28s/it] 32%|███▏ | 9739/30000 [6:14:14<12:56:40, 2.30s/it] 32%|███▏ | 9740/30000 [6:14:17<12:59:49, 2.31s/it] 32%|███▏ | 9740/30000 [6:14:17<12:59:49, 2.31s/it] 32%|███▏ | 9741/30000 [6:14:19<12:55:54, 2.30s/it] 32%|███▏ | 9742/30000 [6:14:21<12:52:52, 2.29s/it] 32%|███▏ | 9743/30000 [6:14:24<12:58:34, 2.31s/it] 32%|███▏ | 9744/30000 [6:14:26<13:02:39, 2.32s/it] 32%|███▏ | 9745/30000 [6:14:28<13:01:46, 2.32s/it] 32%|███▏ | 9746/30000 [6:14:30<13:01:20, 2.31s/it] 32%|███▏ | 9747/30000 [6:14:33<12:59:16, 2.31s/it] 32%|███▏ | 9748/30000 [6:14:35<12:53:39, 2.29s/it] 32%|███▏ | 9749/30000 [6:14:37<12:50:06, 2.28s/it] 32%|███▎ | 9750/30000 [6:14:40<12:56:55, 2.30s/it] 32%|███▎ | 9750/30000 [6:14:40<12:56:55, 2.30s/it] 33%|███▎ | 9751/30000 [6:14:42<12:59:55, 2.31s/it] 33%|███▎ | 9752/30000 [6:14:44<12:59:18, 2.31s/it] 33%|███▎ | 9753/30000 [6:14:47<13:00:38, 2.31s/it] 33%|███▎ | 9754/30000 [6:14:49<12:57:53, 2.31s/it] 33%|███▎ | 9755/30000 [6:14:51<12:55:08, 2.30s/it] 33%|███▎ | 9756/30000 [6:14:53<12:52:35, 2.29s/it] 33%|███▎ | 9757/30000 [6:14:56<12:56:18, 2.30s/it] 33%|███▎ | 9758/30000 [6:14:58<12:56:24, 2.30s/it] 33%|███▎ | 9759/30000 [6:15:00<12:53:38, 2.29s/it] 33%|███▎ | 9760/30000 [6:15:03<12:58:42, 2.31s/it] 33%|███▎ | 9760/30000 [6:15:03<12:58:42, 2.31s/it] 33%|███▎ | 9761/30000 [6:15:05<12:53:03, 2.29s/it] 33%|███▎ | 9762/30000 [6:15:07<12:53:55, 2.29s/it] 33%|███▎ | 9763/30000 [6:15:10<12:52:45, 2.29s/it] 33%|███▎ | 9764/30000 [6:15:12<12:56:49, 2.30s/it] 33%|███▎ | 9765/30000 [6:15:14<12:55:47, 2.30s/it] 33%|███▎ | 9766/30000 [6:15:16<12:52:26, 2.29s/it] 33%|███▎ | 9767/30000 [6:15:19<12:59:09, 2.31s/it] 33%|███▎ | 9768/30000 [6:15:21<13:02:38, 2.32s/it] 33%|███▎ | 9769/30000 [6:15:23<12:57:24, 2.31s/it] 33%|███▎ | 9770/30000 [6:15:26<12:55:12, 2.30s/it] 33%|███▎ | 9770/30000 [6:15:26<12:55:12, 2.30s/it] 33%|███▎ | 9771/30000 [6:15:28<12:57:44, 2.31s/it] 33%|███▎ | 9772/30000 [6:15:30<12:55:29, 2.30s/it] 33%|███▎ | 9773/30000 [6:15:33<12:50:43, 2.29s/it] 33%|███▎ | 9774/30000 [6:15:35<12:55:36, 2.30s/it] 33%|███▎ | 9775/30000 [6:15:37<13:00:04, 2.31s/it] 33%|███▎ | 9776/30000 [6:15:39<12:54:39, 2.30s/it] 33%|███▎ | 9777/30000 [6:15:42<13:00:00, 2.31s/it] 33%|███▎ | 9778/30000 [6:15:44<12:59:44, 2.31s/it] 33%|███▎ | 9779/30000 [6:15:46<12:56:33, 2.30s/it] 33%|███▎ | 9780/30000 [6:15:49<12:57:15, 2.31s/it] 33%|███▎ | 9780/30000 [6:15:49<12:57:15, 2.31s/it] 33%|███▎ | 9781/30000 [6:15:51<12:57:03, 2.31s/it] 33%|███▎ | 9782/30000 [6:15:53<12:52:50, 2.29s/it] 33%|███▎ | 9783/30000 [6:15:56<12:53:37, 2.30s/it] 33%|███▎ | 9784/30000 [6:15:58<12:58:17, 2.31s/it] 33%|███▎ | 9785/30000 [6:16:00<13:01:23, 2.32s/it] 33%|███▎ | 9786/30000 [6:16:03<12:59:42, 2.31s/it] 33%|███▎ | 9787/30000 [6:16:05<13:06:15, 2.33s/it] 33%|███▎ | 9788/30000 [6:16:07<13:01:35, 2.32s/it] 33%|███▎ | 9789/30000 [6:16:10<12:54:49, 2.30s/it] 33%|███▎ | 9790/30000 [6:16:12<12:54:34, 2.30s/it] 33%|███▎ | 9790/30000 [6:16:12<12:54:34, 2.30s/it] 33%|███▎ | 9791/30000 [6:16:14<13:03:02, 2.32s/it] 33%|███▎ | 9792/30000 [6:16:17<13:02:36, 2.32s/it] 33%|███▎ | 9793/30000 [6:16:19<13:00:01, 2.32s/it] 33%|███▎ | 9794/30000 [6:16:21<13:02:28, 2.32s/it] 33%|███▎ | 9795/30000 [6:16:23<12:58:48, 2.31s/it] 33%|███▎ | 9796/30000 [6:16:26<12:55:12, 2.30s/it] 33%|███▎ | 9797/30000 [6:16:28<13:00:13, 2.32s/it] 33%|███▎ | 9798/30000 [6:16:30<12:59:27, 2.31s/it] 33%|███▎ | 9799/30000 [6:16:33<12:58:41, 2.31s/it] 33%|███▎ | 9800/30000 [6:16:35<13:01:38, 2.32s/it] 33%|███▎ | 9800/30000 [6:16:35<13:01:38, 2.32s/it] 33%|███▎ | 9801/30000 [6:16:37<13:05:35, 2.33s/it] 33%|███▎ | 9802/30000 [6:16:40<13:00:19, 2.32s/it] 33%|███▎ | 9803/30000 [6:16:42<12:53:46, 2.30s/it] 33%|███▎ | 9804/30000 [6:16:44<12:58:39, 2.31s/it] 33%|███▎ | 9805/30000 [6:16:47<12:55:07, 2.30s/it] 33%|███▎ | 9806/30000 [6:16:49<12:48:33, 2.28s/it] 33%|███▎ | 9807/30000 [6:16:51<12:58:08, 2.31s/it] 33%|███▎ | 9808/30000 [6:16:53<12:56:10, 2.31s/it] 33%|███▎ | 9809/30000 [6:16:56<12:56:34, 2.31s/it] 33%|███▎ | 9810/30000 [6:16:58<12:56:56, 2.31s/it] 33%|███▎ | 9810/30000 [6:16:58<12:56:56, 2.31s/it] 33%|███▎ | 9811/30000 [6:17:00<13:00:00, 2.32s/it] 33%|███▎ | 9812/30000 [6:17:03<12:47:35, 2.28s/it] 33%|███▎ | 9813/30000 [6:17:05<12:48:13, 2.28s/it] 33%|███▎ | 9814/30000 [6:17:07<12:55:54, 2.31s/it] 33%|███▎ | 9815/30000 [6:17:10<12:50:28, 2.29s/it] 33%|███▎ | 9816/30000 [6:17:12<12:49:24, 2.29s/it] 33%|███▎ | 9817/30000 [6:17:14<12:57:14, 2.31s/it] 33%|███▎ | 9818/30000 [6:17:16<12:53:29, 2.30s/it] 33%|███▎ | 9819/30000 [6:17:19<12:48:13, 2.28s/it] 33%|███▎ | 9820/30000 [6:17:21<12:54:45, 2.30s/it] 33%|███▎ | 9820/30000 [6:17:21<12:54:45, 2.30s/it] 33%|███▎ | 9821/30000 [6:17:23<12:58:12, 2.31s/it] 33%|███▎ | 9822/30000 [6:17:26<12:54:43, 2.30s/it] 33%|███▎ | 9823/30000 [6:17:28<12:55:40, 2.31s/it] 33%|███▎ | 9824/30000 [6:17:30<12:57:15, 2.31s/it] 33%|███▎ | 9825/30000 [6:17:33<12:54:36, 2.30s/it] 33%|███▎ | 9826/30000 [6:17:35<12:51:21, 2.29s/it] 33%|███▎ | 9827/30000 [6:17:37<12:56:46, 2.31s/it] 33%|███▎ | 9828/30000 [6:17:39<12:51:14, 2.29s/it] 33%|███▎ | 9829/30000 [6:17:42<12:50:01, 2.29s/it] 33%|███▎ | 9830/30000 [6:17:44<12:56:22, 2.31s/it] 33%|███▎ | 9830/30000 [6:17:44<12:56:22, 2.31s/it] 33%|███▎ | 9831/30000 [6:17:46<13:01:08, 2.32s/it] 33%|███▎ | 9832/30000 [6:17:49<12:57:21, 2.31s/it] 33%|███▎ | 9833/30000 [6:17:51<13:02:28, 2.33s/it] 33%|███▎ | 9834/30000 [6:17:53<12:58:29, 2.32s/it] 33%|███▎ | 9835/30000 [6:17:56<12:54:45, 2.31s/it] 33%|███▎ | 9836/30000 [6:17:58<12:54:10, 2.30s/it] 33%|███▎ | 9837/30000 [6:18:00<12:55:17, 2.31s/it] 33%|███▎ | 9838/30000 [6:18:03<12:50:35, 2.29s/it] 33%|███▎ | 9839/30000 [6:18:05<12:50:56, 2.29s/it] 33%|███▎ | 9840/30000 [6:18:07<12:58:32, 2.32s/it] 33%|███▎ | 9840/30000 [6:18:07<12:58:32, 2.32s/it] 33%|███▎ | 9841/30000 [6:18:10<13:01:07, 2.32s/it] 33%|███▎ | 9842/30000 [6:18:12<12:56:17, 2.31s/it] 33%|███▎ | 9843/30000 [6:18:14<12:59:44, 2.32s/it] 33%|███▎ | 9844/30000 [6:18:17<13:00:40, 2.32s/it] 33%|███▎ | 9845/30000 [6:18:19<12:57:46, 2.32s/it] 33%|███▎ | 9846/30000 [6:18:21<12:54:36, 2.31s/it] 33%|███▎ | 9847/30000 [6:18:23<13:00:52, 2.32s/it] 33%|███▎ | 9848/30000 [6:18:26<12:54:58, 2.31s/it] 33%|███▎ | 9849/30000 [6:18:28<12:49:40, 2.29s/it] 33%|███▎ | 9850/30000 [6:18:30<12:57:05, 2.31s/it] 33%|███▎ | 9850/30000 [6:18:30<12:57:05, 2.31s/it] 33%|███▎ | 9851/30000 [6:18:33<12:52:44, 2.30s/it] 33%|███▎ | 9852/30000 [6:18:35<12:52:35, 2.30s/it] 33%|███▎ | 9853/30000 [6:18:37<12:58:58, 2.32s/it] 33%|███▎ | 9854/30000 [6:18:40<12:57:08, 2.31s/it] 33%|███▎ | 9855/30000 [6:18:42<12:52:37, 2.30s/it] 33%|███▎ | 9856/30000 [6:18:44<12:59:46, 2.32s/it] 33%|███▎ | 9857/30000 [6:18:47<12:58:45, 2.32s/it] 33%|███▎ | 9858/30000 [6:18:49<12:55:02, 2.31s/it] 33%|███▎ | 9859/30000 [6:18:51<12:57:02, 2.31s/it] 33%|███▎ | 9860/30000 [6:18:53<12:55:46, 2.31s/it] 33%|███▎ | 9860/30000 [6:18:54<12:55:46, 2.31s/it] 33%|███▎ | 9861/30000 [6:18:56<12:53:28, 2.30s/it] 33%|███▎ | 9862/30000 [6:18:58<12:49:30, 2.29s/it] 33%|███▎ | 9863/30000 [6:19:00<12:55:20, 2.31s/it] 33%|███▎ | 9864/30000 [6:19:03<12:49:52, 2.29s/it] 33%|███▎ | 9865/30000 [6:19:05<12:46:44, 2.28s/it] 33%|███▎ | 9866/30000 [6:19:07<12:53:17, 2.30s/it] 33%|███▎ | 9867/30000 [6:19:10<12:54:10, 2.31s/it] 33%|███▎ | 9868/30000 [6:19:12<12:50:26, 2.30s/it] 33%|███▎ | 9869/30000 [6:19:14<12:53:16, 2.30s/it] 33%|███▎ | 9870/30000 [6:19:16<12:53:16, 2.30s/it] 33%|███▎ | 9870/30000 [6:19:16<12:53:16, 2.30s/it] 33%|███▎ | 9871/30000 [6:19:19<12:52:42, 2.30s/it] 33%|███▎ | 9872/30000 [6:19:21<12:53:48, 2.31s/it] 33%|███▎ | 9873/30000 [6:19:23<12:57:28, 2.32s/it] 33%|███▎ | 9874/30000 [6:19:26<12:53:40, 2.31s/it] 33%|███▎ | 9875/30000 [6:19:28<12:47:18, 2.29s/it] 33%|███▎ | 9876/30000 [6:19:30<12:50:04, 2.30s/it] 33%|███▎ | 9877/30000 [6:19:33<12:47:38, 2.29s/it] 33%|███▎ | 9878/30000 [6:19:35<12:44:33, 2.28s/it] 33%|███▎ | 9879/30000 [6:19:37<12:52:24, 2.30s/it] 33%|███▎ | 9880/30000 [6:19:39<12:52:21, 2.30s/it] 33%|███▎ | 9880/30000 [6:19:39<12:52:21, 2.30s/it] 33%|███▎ | 9881/30000 [6:19:42<12:50:52, 2.30s/it] 33%|███▎ | 9882/30000 [6:19:44<12:52:58, 2.31s/it] 33%|███▎ | 9883/30000 [6:19:46<12:52:04, 2.30s/it] 33%|███▎ | 9884/30000 [6:19:49<12:49:54, 2.30s/it] 33%|███▎ | 9885/30000 [6:19:51<12:47:45, 2.29s/it] 33%|███▎ | 9886/30000 [6:19:53<12:49:45, 2.30s/it] 33%|███▎ | 9887/30000 [6:19:55<12:47:54, 2.29s/it] 33%|███▎ | 9888/30000 [6:19:58<12:50:06, 2.30s/it] 33%|███▎ | 9889/30000 [6:20:00<12:58:57, 2.32s/it] 33%|███▎ | 9890/30000 [6:20:02<12:57:20, 2.32s/it] 33%|███▎ | 9890/30000 [6:20:03<12:57:20, 2.32s/it] 33%|███▎ | 9891/30000 [6:20:05<12:50:54, 2.30s/it] 33%|███▎ | 9892/30000 [6:20:07<12:56:13, 2.32s/it] 33%|███▎ | 9893/30000 [6:20:09<12:53:29, 2.31s/it] 33%|███▎ | 9894/30000 [6:20:12<12:49:01, 2.29s/it] 33%|███▎ | 9895/30000 [6:20:14<12:47:16, 2.29s/it] 33%|███▎ | 9896/30000 [6:20:16<12:51:43, 2.30s/it] 33%|███▎ | 9897/30000 [6:20:19<12:50:34, 2.30s/it] 33%|███▎ | 9898/30000 [6:20:21<12:47:30, 2.29s/it] 33%|███▎ | 9899/30000 [6:20:23<12:49:40, 2.30s/it] 33%|███▎ | 9900/30000 [6:20:25<12:49:28, 2.30s/it] 33%|███▎ | 9900/30000 [6:20:25<12:49:28, 2.30s/it] 33%|███▎ | 9901/30000 [6:20:28<12:45:30, 2.29s/it] 33%|███▎ | 9902/30000 [6:20:30<12:43:57, 2.28s/it] 33%|███▎ | 9903/30000 [6:20:32<12:49:33, 2.30s/it] 33%|███▎ | 9904/30000 [6:20:35<12:50:41, 2.30s/it] 33%|███▎ | 9905/30000 [6:20:37<12:45:17, 2.29s/it] 33%|███▎ | 9906/30000 [6:20:39<12:48:08, 2.29s/it] 33%|███▎ | 9907/30000 [6:20:41<12:46:40, 2.29s/it] 33%|███▎ | 9908/30000 [6:20:44<12:41:09, 2.27s/it] 33%|███▎ | 9909/30000 [6:20:46<12:43:16, 2.28s/it] 33%|███▎ | 9910/30000 [6:20:48<12:49:27, 2.30s/it] 33%|███▎ | 9910/30000 [6:20:48<12:49:27, 2.30s/it] 33%|███▎ | 9911/30000 [6:20:51<12:49:48, 2.30s/it] 33%|███▎ | 9912/30000 [6:20:53<12:45:45, 2.29s/it] 33%|███▎ | 9913/30000 [6:20:55<12:49:44, 2.30s/it] 33%|███▎ | 9914/30000 [6:20:58<12:51:14, 2.30s/it] 33%|███▎ | 9915/30000 [6:21:00<12:46:09, 2.29s/it] 33%|███▎ | 9916/30000 [6:21:02<12:47:19, 2.29s/it] 33%|███▎ | 9917/30000 [6:21:04<12:46:41, 2.29s/it] 33%|███▎ | 9918/30000 [6:21:07<12:45:19, 2.29s/it] 33%|███▎ | 9919/30000 [6:21:09<12:48:37, 2.30s/it] 33%|███▎ | 9920/30000 [6:21:11<12:55:56, 2.32s/it] 33%|███▎ | 9920/30000 [6:21:11<12:55:56, 2.32s/it] 33%|███▎ | 9921/30000 [6:21:14<12:54:57, 2.32s/it] 33%|███▎ | 9922/30000 [6:21:16<12:50:53, 2.30s/it] 33%|███▎ | 9923/30000 [6:21:18<12:48:18, 2.30s/it] 33%|███▎ | 9924/30000 [6:21:21<12:52:04, 2.31s/it] 33%|███▎ | 9925/30000 [6:21:23<12:42:41, 2.28s/it] 33%|███▎ | 9926/30000 [6:21:25<12:41:09, 2.28s/it] 33%|███▎ | 9927/30000 [6:21:27<12:47:15, 2.29s/it] 33%|███▎ | 9928/30000 [6:21:30<12:49:00, 2.30s/it] 33%|███▎ | 9929/30000 [6:21:32<12:43:38, 2.28s/it] 33%|███▎ | 9930/30000 [6:21:34<12:42:02, 2.28s/it] 33%|███▎ | 9930/30000 [6:21:34<12:42:02, 2.28s/it] 33%|███▎ | 9931/30000 [6:21:37<12:48:36, 2.30s/it] 33%|███▎ | 9932/30000 [6:21:39<12:49:43, 2.30s/it] 33%|███▎ | 9933/30000 [6:21:41<12:46:48, 2.29s/it] 33%|███▎ | 9934/30000 [6:21:43<12:44:11, 2.29s/it] 33%|███▎ | 9935/30000 [6:21:46<12:52:59, 2.31s/it] 33%|███▎ | 9936/30000 [6:21:48<12:48:13, 2.30s/it] 33%|███▎ | 9937/30000 [6:21:50<12:46:18, 2.29s/it] 33%|███▎ | 9938/30000 [6:21:53<12:41:56, 2.28s/it] 33%|███▎ | 9939/30000 [6:21:55<12:44:29, 2.29s/it] 33%|███▎ | 9940/30000 [6:21:57<12:45:52, 2.29s/it] 33%|███▎ | 9940/30000 [6:21:57<12:45:52, 2.29s/it] 33%|███▎ | 9941/30000 [6:21:59<12:42:49, 2.28s/it] 33%|███▎ | 9942/30000 [6:22:02<12:43:32, 2.28s/it] 33%|███▎ | 9943/30000 [6:22:04<12:47:22, 2.30s/it] 33%|███▎ | 9944/30000 [6:22:06<12:42:45, 2.28s/it] 33%|███▎ | 9945/30000 [6:22:09<12:41:04, 2.28s/it] 33%|███▎ | 9946/30000 [6:22:11<12:44:54, 2.29s/it] 33%|███▎ | 9947/30000 [6:22:13<12:43:57, 2.29s/it] 33%|███▎ | 9948/30000 [6:22:15<12:42:39, 2.28s/it] 33%|███▎ | 9949/30000 [6:22:18<12:40:46, 2.28s/it] 33%|███▎ | 9950/30000 [6:22:20<12:45:49, 2.29s/it] 33%|███▎ | 9950/30000 [6:22:20<12:45:49, 2.29s/it] 33%|███▎ | 9951/30000 [6:22:22<12:43:45, 2.29s/it] 33%|███▎ | 9952/30000 [6:22:25<12:39:33, 2.27s/it] 33%|███▎ | 9953/30000 [6:22:27<12:36:13, 2.26s/it] 33%|███▎ | 9954/30000 [6:22:29<12:42:12, 2.28s/it] 33%|███▎ | 9955/30000 [6:22:31<12:36:00, 2.26s/it] 33%|███▎ | 9956/30000 [6:22:34<12:35:49, 2.26s/it] 33%|███▎ | 9957/30000 [6:22:36<12:43:16, 2.28s/it] 33%|███▎ | 9958/30000 [6:22:38<12:41:23, 2.28s/it] 33%|███▎ | 9959/30000 [6:22:40<12:43:05, 2.28s/it] 33%|███▎ | 9960/30000 [6:22:43<12:44:57, 2.29s/it] 33%|███▎ | 9960/30000 [6:22:43<12:44:57, 2.29s/it] 33%|███▎ | 9961/30000 [6:22:45<12:49:57, 2.31s/it] 33%|███▎ | 9962/30000 [6:22:47<12:42:09, 2.28s/it] 33%|███▎ | 9963/30000 [6:22:50<12:43:21, 2.29s/it] 33%|███▎ | 9964/30000 [6:22:52<12:50:22, 2.31s/it] 33%|███▎ | 9965/30000 [6:22:54<12:49:36, 2.30s/it] 33%|███▎ | 9966/30000 [6:22:57<12:49:07, 2.30s/it] 33%|███▎ | 9967/30000 [6:22:59<12:46:44, 2.30s/it] 33%|███▎ | 9968/30000 [6:23:01<12:52:29, 2.31s/it] 33%|███▎ | 9969/30000 [6:23:04<12:52:14, 2.31s/it] 33%|███▎ | 9970/30000 [6:23:06<12:50:20, 2.31s/it] 33%|███▎ | 9970/30000 [6:23:06<12:50:20, 2.31s/it] 33%|███▎ | 9971/30000 [6:23:08<12:53:31, 2.32s/it] 33%|███▎ | 9972/30000 [6:23:10<12:53:01, 2.32s/it] 33%|███▎ | 9973/30000 [6:23:13<12:46:57, 2.30s/it] 33%|███▎ | 9974/30000 [6:23:15<12:48:58, 2.30s/it] 33%|███▎ | 9975/30000 [6:23:17<12:55:03, 2.32s/it] 33%|███▎ | 9976/30000 [6:23:20<12:54:31, 2.32s/it] 33%|███▎ | 9977/30000 [6:23:22<12:53:42, 2.32s/it] 33%|███▎ | 9978/30000 [6:23:24<12:51:21, 2.31s/it] 33%|███▎ | 9979/30000 [6:23:27<12:53:27, 2.32s/it] 33%|███▎ | 9980/30000 [6:23:29<12:47:28, 2.30s/it] +{'loss': 0.0186, 'grad_norm': 0.05794541910290718, 'learning_rate': 2.443466499872889e-05, 'flow_matching_loss': 0.01857468169182539, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.79} +{'loss': 0.0143, 'grad_norm': 0.07364710420370102, 'learning_rate': 2.4421804808563413e-05, 'flow_matching_loss': 0.014288917556405068, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.79} +{'loss': 0.0187, 'grad_norm': 0.08846189826726913, 'learning_rate': 2.440893317001667e-05, 'flow_matching_loss': 0.018690633680671455, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.79} +{'loss': 0.0148, 'grad_norm': 0.08696884661912918, 'learning_rate': 2.4396050098728925e-05, 'flow_matching_loss': 0.01478832308202982, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.8} +{'loss': 0.0156, 'grad_norm': 0.06989570707082748, 'learning_rate': 2.4383155610354314e-05, 'flow_matching_loss': 0.015587813965976239, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.8} +{'loss': 0.0185, 'grad_norm': 0.10138457268476486, 'learning_rate': 2.437024972056087e-05, 'flow_matching_loss': 0.01854343991726637, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.8} +{'loss': 0.0155, 'grad_norm': 0.0890192985534668, 'learning_rate': 2.4357332445030452e-05, 'flow_matching_loss': 0.015503582637757063, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.8} +{'loss': 0.0169, 'grad_norm': 0.09498883038759232, 'learning_rate': 2.434440379945878e-05, 'flow_matching_loss': 0.016867614164948464, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.81} +{'loss': 0.0169, 'grad_norm': 0.12455708533525467, 'learning_rate': 2.4331463799555366e-05, 'flow_matching_loss': 0.016877350118011235, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.81} +{'loss': 0.0151, 'grad_norm': 0.07714275270700455, 'learning_rate': 2.4318512461043544e-05, 'flow_matching_loss': 0.015054069552570581, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.81} +{'loss': 0.0152, 'grad_norm': 0.12289591878652573, 'learning_rate': 2.4305549799660397e-05, 'flow_matching_loss': 0.015206705499440431, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.82} +{'loss': 0.0163, 'grad_norm': 0.08894221484661102, 'learning_rate': 2.429257583115679e-05, 'flow_matching_loss': 0.016265198402106763, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.82} +{'loss': 0.0172, 'grad_norm': 0.083670474588871, 'learning_rate': 2.4279590571297316e-05, 'flow_matching_loss': 0.017174868565052747, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.82} +{'loss': 0.0161, 'grad_norm': 0.09961961209774017, 'learning_rate': 2.4266594035860288e-05, 'flow_matching_loss': 0.016140423901379108, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.83} +{'loss': 0.0162, 'grad_norm': 0.09243503212928772, 'learning_rate': 2.425358624063773e-05, 'flow_matching_loss': 0.016245375387370586, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.83} +{'loss': 0.0142, 'grad_norm': 0.08667387813329697, 'learning_rate': 2.424056720143532e-05, 'flow_matching_loss': 0.014185183309018612, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.83} +{'loss': 0.0151, 'grad_norm': 0.08263567090034485, 'learning_rate': 2.422753693407244e-05, 'flow_matching_loss': 0.015114926639944315, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.83} +{'loss': 0.0149, 'grad_norm': 0.09814136475324631, 'learning_rate': 2.421449545438209e-05, 'flow_matching_loss': 0.0148558066226542, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.84} +{'loss': 0.018, 'grad_norm': 0.0857098400592804, 'learning_rate': 2.4201442778210887e-05, 'flow_matching_loss': 0.017969620134681463, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.84} +{'loss': 0.0161, 'grad_norm': 0.07886181771755219, 'learning_rate': 2.418837892141907e-05, 'flow_matching_loss': 0.016136865131556988, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.84} +{'loss': 0.0159, 'grad_norm': 0.07093358039855957, 'learning_rate': 2.4175303899880458e-05, 'flow_matching_loss': 0.01588539471849799, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.85} +{'loss': 0.0159, 'grad_norm': 0.06196693331003189, 'learning_rate': 2.4162217729482427e-05, 'flow_matching_loss': 0.015897923056036233, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.85} +{'loss': 0.0145, 'grad_norm': 0.12211130559444427, 'learning_rate': 2.4149120426125918e-05, 'flow_matching_loss': 0.014537472557276488, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.85} +{'loss': 0.0147, 'grad_norm': 0.0765785351395607, 'learning_rate': 2.413601200572538e-05, 'flow_matching_loss': 0.014733227156102657, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.85} +{'loss': 0.0155, 'grad_norm': 0.07804866135120392, 'learning_rate': 2.4122892484208794e-05, 'flow_matching_loss': 0.015478320326656103, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.86} +{'loss': 0.0165, 'grad_norm': 0.09049414098262787, 'learning_rate': 2.4109761877517603e-05, 'flow_matching_loss': 0.016539573855698108, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.86} +{'loss': 0.017, 'grad_norm': 0.08025554567575455, 'learning_rate': 2.4096620201606737e-05, 'flow_matching_loss': 0.017017710953950882, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.86} +{'loss': 0.0142, 'grad_norm': 0.06290726363658905, 'learning_rate': 2.4083467472444573e-05, 'flow_matching_loss': 0.014235160406678915, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.87} +{'loss': 0.0146, 'grad_norm': 0.0755898728966713, 'learning_rate': 2.4070303706012912e-05, 'flow_matching_loss': 0.014623530395328998, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.87} +{'loss': 0.0159, 'grad_norm': 0.0824689120054245, 'learning_rate': 2.4057128918306975e-05, 'flow_matching_loss': 0.01587908938527107, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.87} +{'loss': 0.0151, 'grad_norm': 0.08085981756448746, 'learning_rate': 2.4043943125335373e-05, 'flow_matching_loss': 0.015053980238735675, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.87} +{'loss': 0.0169, 'grad_norm': 0.09555201232433319, 'learning_rate': 2.403074634312008e-05, 'flow_matching_loss': 0.016866143513470887, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.88} +{'loss': 0.0165, 'grad_norm': 0.09441875666379929, 'learning_rate': 2.4017538587696442e-05, 'flow_matching_loss': 0.016470300406217574, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.88} +{'loss': 0.0149, 'grad_norm': 0.06815000623464584, 'learning_rate': 2.4004319875113118e-05, 'flow_matching_loss': 0.014890693034976721, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.88} +{'loss': 0.0159, 'grad_norm': 0.08270730078220367, 'learning_rate': 2.3991090221432096e-05, 'flow_matching_loss': 0.01593627892434597, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.89} +{'loss': 0.0161, 'grad_norm': 0.07483769208192825, 'learning_rate': 2.397784964272865e-05, 'flow_matching_loss': 0.01612103981897235, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.89} +{'loss': 0.0159, 'grad_norm': 0.08344195038080215, 'learning_rate': 2.3964598155091335e-05, 'flow_matching_loss': 0.015913822036236525, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.89} +{'loss': 0.0164, 'grad_norm': 0.06890333443880081, 'learning_rate': 2.3951335774621952e-05, 'flow_matching_loss': 0.016424527578055858, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.89} +{'loss': 0.0169, 'grad_norm': 0.07427181303501129, 'learning_rate': 2.3938062517435548e-05, 'flow_matching_loss': 0.01693023629486561, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.9} +{'loss': 0.0183, 'grad_norm': 0.08422599732875824, 'learning_rate': 2.392477839966038e-05, 'flow_matching_loss': 0.018289812467992307, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.9} +{'loss': 0.0159, 'grad_norm': 0.08955325186252594, 'learning_rate': 2.3911483437437904e-05, 'flow_matching_loss': 0.015879007149487735, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.9} 33%|███▎ | 9980/30000 [6:23:29<12:47:28, 2.30s/it] 33%|███▎ | 9981/30000 [6:23:31<12:46:07, 2.30s/it] 33%|███▎ | 9982/30000 [6:23:34<12:57:48, 2.33s/it] 33%|███▎ | 9983/30000 [6:23:36<12:58:48, 2.33s/it] 33%|███▎ | 9984/30000 [6:23:38<12:50:30, 2.31s/it] 33%|███▎ | 9985/30000 [6:23:40<12:43:52, 2.29s/it] 33%|███▎ | 9986/30000 [6:23:43<12:49:14, 2.31s/it] 33%|███▎ | 9987/30000 [6:23:45<12:47:49, 2.30s/it] 33%|███▎ | 9988/30000 [6:23:47<12:46:37, 2.30s/it] 33%|███▎ | 9989/30000 [6:23:50<12:49:07, 2.31s/it] 33%|███▎ | 9990/30000 [6:23:52<12:48:46, 2.31s/it] 33%|███▎ | 9990/30000 [6:23:52<12:48:46, 2.31s/it] 33%|███▎ | 9991/30000 [6:23:54<12:49:52, 2.31s/it] 33%|███▎ | 9992/30000 [6:23:57<12:47:38, 2.30s/it] 33%|███▎ | 9993/30000 [6:23:59<12:55:43, 2.33s/it] 33%|███▎ | 9994/30000 [6:24:01<12:56:41, 2.33s/it] 33%|███▎ | 9995/30000 [6:24:04<12:53:02, 2.32s/it] 33%|███▎ | 9996/30000 [6:24:06<12:49:53, 2.31s/it] 33%|███▎ | 9997/30000 [6:24:08<12:54:24, 2.32s/it] 33%|███▎ | 9998/30000 [6:24:11<12:51:46, 2.32s/it] 33%|███▎ | 9999/30000 [6:24:13<12:48:47, 2.31s/it] 33%|███▎ | 10000/30000 [6:24:15<12:54:52, 2.32s/it] 33%|███▎ | 10000/30000 [6:24:15<12:54:52, 2.32s/it] 33%|███▎ | 10001/30000 [6:24:18<12:53:14, 2.32s/it] 33%|███▎ | 10002/30000 [6:24:20<12:49:50, 2.31s/it] 33%|███▎ | 10003/30000 [6:24:22<12:47:30, 2.30s/it] 33%|███▎ | 10004/30000 [6:24:24<12:52:19, 2.32s/it] 33%|███▎ | 10005/30000 [6:24:27<12:54:01, 2.32s/it] 33%|███▎ | 10006/30000 [6:24:29<12:49:42, 2.31s/it] 33%|███▎ | 10007/30000 [6:24:31<12:47:28, 2.30s/it] 33%|███▎ | 10008/30000 [6:24:34<12:53:36, 2.32s/it] 33%|███▎ | 10009/30000 [6:24:36<12:50:15, 2.31s/it] 33%|███▎ | 10010/30000 [6:24:38<12:46:05, 2.30s/it] 33%|███▎ | 10010/30000 [6:24:38<12:46:05, 2.30s/it] 33%|███▎ | 10011/30000 [6:24:41<12:53:49, 2.32s/it] 33%|███▎ | 10012/30000 [6:24:43<12:49:18, 2.31s/it] 33%|███▎ | 10013/30000 [6:24:45<12:47:06, 2.30s/it] 33%|███▎ | 10014/30000 [6:24:48<12:48:13, 2.31s/it] 33%|███▎ | 10015/30000 [6:24:50<12:50:37, 2.31s/it] 33%|███▎ | 10016/30000 [6:24:52<12:49:05, 2.31s/it] 33%|███▎ | 10017/30000 [6:24:54<12:47:36, 2.30s/it] 33%|███▎ | 10018/30000 [6:24:57<12:53:27, 2.32s/it] 33%|███▎ | 10019/30000 [6:24:59<12:51:32, 2.32s/it] 33%|███▎ | 10020/30000 [6:25:01<12:44:33, 2.30s/it] 33%|███▎ | 10020/30000 [6:25:01<12:44:33, 2.30s/it] 33%|███▎ | 10021/30000 [6:25:04<12:44:09, 2.29s/it] 33%|███▎ | 10022/30000 [6:25:06<12:48:42, 2.31s/it] 33%|███▎ | 10023/30000 [6:25:08<12:44:34, 2.30s/it] 33%|███▎ | 10024/30000 [6:25:11<12:42:02, 2.29s/it] 33%|███▎ | 10025/30000 [6:25:13<12:49:23, 2.31s/it] 33%|███▎ | 10026/30000 [6:25:15<12:50:23, 2.31s/it] 33%|███▎ | 10027/30000 [6:25:17<12:45:32, 2.30s/it] 33%|███▎ | 10028/30000 [6:25:20<12:43:56, 2.30s/it] 33%|███▎ | 10029/30000 [6:25:22<12:42:52, 2.29s/it] 33%|███▎ | 10030/30000 [6:25:24<12:41:27, 2.29s/it] 33%|███▎ | 10030/30000 [6:25:24<12:41:27, 2.29s/it] 33%|███▎ | 10031/30000 [6:25:27<12:39:24, 2.28s/it] 33%|███▎ | 10032/30000 [6:25:29<12:47:17, 2.31s/it] 33%|███▎ | 10033/30000 [6:25:31<12:50:58, 2.32s/it] 33%|███▎ | 10034/30000 [6:25:34<12:47:59, 2.31s/it] 33%|███▎ | 10035/30000 [6:25:36<12:46:52, 2.30s/it] 33%|███▎ | 10036/30000 [6:25:38<12:43:38, 2.30s/it] 33%|███▎ | 10037/30000 [6:25:40<12:41:07, 2.29s/it] 33%|███▎ | 10038/30000 [6:25:43<12:40:46, 2.29s/it] 33%|███▎ | 10039/30000 [6:25:45<12:45:21, 2.30s/it] 33%|███▎ | 10040/30000 [6:25:47<12:45:21, 2.30s/it] 33%|███▎ | 10040/30000 [6:25:47<12:45:21, 2.30s/it] 33%|███▎ | 10041/30000 [6:25:50<12:42:56, 2.29s/it] 33%|███▎ | 10042/30000 [6:25:52<12:49:09, 2.31s/it] 33%|███▎ | 10043/30000 [6:25:54<12:48:51, 2.31s/it] 33%|███▎ | 10044/30000 [6:25:57<12:44:43, 2.30s/it] 33%|███▎ | 10045/30000 [6:25:59<12:47:33, 2.31s/it] 33%|███▎ | 10046/30000 [6:26:01<12:46:32, 2.30s/it] 33%|███▎ | 10047/30000 [6:26:04<12:46:41, 2.31s/it] 33%|███▎ | 10048/30000 [6:26:06<12:47:11, 2.31s/it] 33%|███▎ | 10049/30000 [6:26:08<12:53:58, 2.33s/it] 34%|███▎ | 10050/30000 [6:26:10<12:46:34, 2.31s/it] 34%|███▎ | 10050/30000 [6:26:10<12:46:34, 2.31s/it] 34%|███▎ | 10051/30000 [6:26:13<12:44:17, 2.30s/it] 34%|███▎ | 10052/30000 [6:26:15<12:47:06, 2.31s/it] 34%|███▎ | 10053/30000 [6:26:17<12:49:30, 2.31s/it] 34%|███▎ | 10054/30000 [6:26:20<12:44:40, 2.30s/it] 34%|███▎ | 10055/30000 [6:26:22<12:51:09, 2.32s/it] 34%|███▎ | 10056/30000 [6:26:24<12:51:30, 2.32s/it] 34%|███▎ | 10057/30000 [6:26:27<12:49:36, 2.32s/it] 34%|███▎ | 10058/30000 [6:26:29<12:48:20, 2.31s/it] 34%|███▎ | 10059/30000 [6:26:31<12:50:34, 2.32s/it] 34%|███▎ | 10060/30000 [6:26:34<12:43:35, 2.30s/it] 34%|███▎ | 10060/30000 [6:26:34<12:43:35, 2.30s/it] 34%|███▎ | 10061/30000 [6:26:36<12:41:54, 2.29s/it] 34%|███▎ | 10062/30000 [6:26:38<12:46:54, 2.31s/it] 34%|███▎ | 10063/30000 [6:26:41<12:50:59, 2.32s/it] 34%|███▎ | 10064/30000 [6:26:43<12:49:07, 2.31s/it] 34%|███▎ | 10065/30000 [6:26:45<12:52:29, 2.33s/it] 34%|███▎ | 10066/30000 [6:26:47<12:50:30, 2.32s/it] 34%|███▎ | 10067/30000 [6:26:50<12:44:08, 2.30s/it] 34%|███▎ | 10068/30000 [6:26:52<12:40:34, 2.29s/it] 34%|███▎ | 10069/30000 [6:26:54<12:48:06, 2.31s/it] 34%|███▎ | 10070/30000 [6:26:57<12:47:15, 2.31s/it] 34%|███▎ | 10070/30000 [6:26:57<12:47:15, 2.31s/it] 34%|███▎ | 10071/30000 [6:26:59<12:48:37, 2.31s/it] 34%|███▎ | 10072/30000 [6:27:01<12:52:55, 2.33s/it] 34%|███▎ | 10073/30000 [6:27:04<12:53:48, 2.33s/it] 34%|███▎ | 10074/30000 [6:27:06<12:48:45, 2.31s/it] 34%|███▎ | 10075/30000 [6:27:08<12:45:37, 2.31s/it] 34%|███▎ | 10076/30000 [6:27:11<12:46:13, 2.31s/it] 34%|███▎ | 10077/30000 [6:27:13<12:44:08, 2.30s/it] 34%|███▎ | 10078/30000 [6:27:15<12:42:03, 2.30s/it] 34%|███▎ | 10079/30000 [6:27:17<12:47:42, 2.31s/it] 34%|███▎ | 10080/30000 [6:27:20<12:47:48, 2.31s/it] 34%|███▎ | 10080/30000 [6:27:20<12:47:48, 2.31s/it] 34%|███▎ | 10081/30000 [6:27:22<12:46:42, 2.31s/it] 34%|███▎ | 10082/30000 [6:27:24<12:45:08, 2.30s/it] 34%|███▎ | 10083/30000 [6:27:27<12:45:26, 2.31s/it] 34%|███▎ | 10084/30000 [6:27:29<12:08:34, 2.19s/it] 34%|███▎ | 10085/30000 [6:27:30<10:31:43, 1.90s/it] 34%|███▎ | 10086/30000 [6:27:31<9:20:24, 1.69s/it] 34%|███▎ | 10087/30000 [6:27:32<8:34:04, 1.55s/it] 34%|███▎ | 10088/30000 [6:27:33<7:59:43, 1.45s/it] 34%|███▎ | 10089/30000 [6:27:35<7:41:15, 1.39s/it] 34%|███▎ | 10090/30000 [6:27:36<8:11:11, 1.48s/it] 34%|███▎ | 10090/30000 [6:27:36<8:11:11, 1.48s/it] 34%|███▎ | 10091/30000 [6:27:39<9:32:53, 1.73s/it] 34%|███▎ | 10092/30000 [6:27:41<10:29:30, 1.90s/it] 34%|███▎ | 10093/30000 [6:27:43<11:14:09, 2.03s/it] 34%|███▎ | 10094/30000 [6:27:46<11:38:26, 2.11s/it] 34%|███▎ | 10095/30000 [6:27:48<11:53:46, 2.15s/it] 34%|███▎ | 10096/30000 [6:27:50<12:17:10, 2.22s/it] 34%|███▎ | 10097/30000 [6:27:53<12:27:39, 2.25s/it] 34%|███▎ | 10098/30000 [6:27:55<12:29:06, 2.26s/it] 34%|███▎ | 10099/30000 [6:27:57<12:35:32, 2.28s/it] 34%|███▎ | 10100/30000 [6:27:59<12:37:29, 2.28s/it] 34%|███▎ | 10100/30000 [6:28:00<12:37:29, 2.28s/it] 34%|███▎ | 10101/30000 [6:28:02<12:38:48, 2.29s/it] 34%|███▎ | 10102/30000 [6:28:04<12:36:01, 2.28s/it] 34%|███▎ | 10103/30000 [6:28:06<12:42:54, 2.30s/it] 34%|███▎ | 10104/30000 [6:28:09<12:41:29, 2.30s/it] 34%|███▎ | 10105/30000 [6:28:11<12:39:59, 2.29s/it] 34%|███▎ | 10106/30000 [6:28:13<12:45:05, 2.31s/it] 34%|███▎ | 10107/30000 [6:28:16<12:40:48, 2.29s/it] 34%|███▎ | 10108/30000 [6:28:18<12:37:00, 2.28s/it] 34%|███▎ | 10109/30000 [6:28:20<12:40:44, 2.29s/it] 34%|███▎ | 10110/30000 [6:28:22<12:38:02, 2.29s/it] 34%|███▎ | 10110/30000 [6:28:22<12:38:02, 2.29s/it] 34%|███▎ | 10111/30000 [6:28:25<12:37:17, 2.28s/it] 34%|███▎ | 10112/30000 [6:28:27<12:37:29, 2.29s/it] 34%|███▎ | 10113/30000 [6:28:29<12:43:35, 2.30s/it] 34%|███▎ | 10114/30000 [6:28:32<12:44:53, 2.31s/it] 34%|███▎ | 10115/30000 [6:28:34<12:42:14, 2.30s/it] 34%|███▎ | 10116/30000 [6:28:36<12:41:26, 2.30s/it] 34%|███▎ | 10117/30000 [6:28:39<12:44:14, 2.31s/it] 34%|███▎ | 10118/30000 [6:28:41<12:42:22, 2.30s/it] 34%|███▎ | 10119/30000 [6:28:43<12:40:46, 2.30s/it] 34%|███▎ | 10120/30000 [6:28:45<12:47:31, 2.32s/it] 34%|███▎ | 10120/30000 [6:28:46<12:47:31, 2.32s/it] 34%|███▎ | 10121/30000 [6:28:48<12:49:08, 2.32s/it] 34%|███▎ | 10122/30000 [6:28:50<12:41:31, 2.30s/it] 34%|███▎ | 10123/30000 [6:28:52<12:38:48, 2.29s/it] 34%|███▎ | 10124/30000 [6:28:55<12:38:19, 2.29s/it] 34%|███▍ | 10125/30000 [6:28:57<12:36:39, 2.28s/it] 34%|███▍ | 10126/30000 [6:28:59<12:32:34, 2.27s/it] 34%|███▍ | 10127/30000 [6:29:02<12:41:07, 2.30s/it] 34%|███▍ | 10128/30000 [6:29:04<12:38:58, 2.29s/it] 34%|███▍ | 10129/30000 [6:29:06<12:42:02, 2.30s/it] 34%|███▍ | 10130/30000 [6:29:08<12:43:20, 2.31s/it] 34%|███▍ | 10130/30000 [6:29:08<12:43:20, 2.31s/it] 34%|███▍ | 10131/30000 [6:29:11<12:44:09, 2.31s/it] 34%|███▍ | 10132/30000 [6:29:13<12:37:49, 2.29s/it] 34%|███▍ | 10133/30000 [6:29:15<12:33:50, 2.28s/it] 34%|███▍ | 10134/30000 [6:29:18<12:42:03, 2.30s/it] 34%|███▍ | 10135/30000 [6:29:20<12:47:04, 2.32s/it] 34%|███▍ | 10136/30000 [6:29:22<12:45:42, 2.31s/it] 34%|███▍ | 10137/30000 [6:29:25<12:42:04, 2.30s/it] 34%|███▍ | 10138/30000 [6:29:27<12:42:55, 2.30s/it] 34%|███▍ | 10139/30000 [6:29:29<12:35:54, 2.28s/it] 34%|███▍ | 10140/30000 [6:29:31<12:36:04, 2.28s/it] 34%|███▍ | 10140/30000 [6:29:31<12:36:04, 2.28s/it] 34%|███▍ | 10141/30000 [6:29:34<12:41:15, 2.30s/it] 34%|███▍ | 10142/30000 [6:29:36<12:43:50, 2.31s/it] 34%|███▍ | 10143/30000 [6:29:38<12:41:59, 2.30s/it] 34%|███▍ | 10144/30000 [6:29:41<12:40:46, 2.30s/it] 34%|███▍ | 10145/30000 [6:29:43<12:47:45, 2.32s/it] 34%|███▍ | 10146/30000 [6:29:45<12:47:33, 2.32s/it] 34%|███▍ | 10147/30000 [6:29:48<12:40:32, 2.30s/it] 34%|███▍ | 10148/30000 [6:29:50<12:34:38, 2.28s/it] 34%|███▍ | 10149/30000 [6:29:52<12:42:16, 2.30s/it] 34%|███▍ | 10150/30000 [6:29:54<12:45:42, 2.31s/it] 34%|███▍ | 10150/30000 [6:29:55<12:45:42, 2.31s/it] 34%|███▍ | 10151/30000 [6:29:57<12:41:28, 2.30s/it] 34%|███▍ | 10152/30000 [6:29:59<12:37:49, 2.29s/it] 34%|███▍ | 10153/30000 [6:30:01<12:41:50, 2.30s/it] 34%|███▍ | 10154/30000 [6:30:04<12:33:39, 2.28s/it] 34%|███▍ | 10155/30000 [6:30:06<12:33:05, 2.28s/it] 34%|███▍ | 10156/30000 [6:30:08<12:37:43, 2.29s/it] 34%|███▍ | 10157/30000 [6:30:10<12:35:21, 2.28s/it] 34%|███▍ | 10158/30000 [6:30:13<12:31:54, 2.27s/it] 34%|███▍ | 10159/30000 [6:30:15<12:33:17, 2.28s/it] 34%|███▍ | 10160/30000 [6:30:17<12:38:11, 2.29s/it] 34%|███▍ | 10160/30000 [6:30:17<12:38:11, 2.29s/it] 34%|███▍ | 10161/30000 [6:30:20<12:41:06, 2.30s/it] 34%|███▍ | 10162/30000 [6:30:22<12:35:05, 2.28s/it] 34%|███▍ | 10163/30000 [6:30:24<12:31:38, 2.27s/it] 34%|███▍ | 10164/30000 [6:30:26<12:35:59, 2.29s/it] 34%|███▍ | 10165/30000 [6:30:29<12:34:01, 2.28s/it] 34%|███▍ | 10166/30000 [6:30:31<12:34:37, 2.28s/it] 34%|███▍ | 10167/30000 [6:30:33<12:40:59, 2.30s/it] 34%|███▍ | 10168/30000 [6:30:36<12:38:22, 2.29s/it] 34%|███▍ | 10169/30000 [6:30:38<12:36:03, 2.29s/it] 34%|███▍ | 10170/30000 [6:30:40<12:35:33, 2.29s/it] 34%|███▍ | 10170/30000 [6:30:40<12:35:33, 2.29s/it] 34%|███▍ | 10171/30000 [6:30:42<12:35:54, 2.29s/it] 34%|███▍ | 10172/30000 [6:30:45<12:37:27, 2.29s/it] 34%|███▍ | 10173/30000 [6:30:47<12:32:51, 2.28s/it] 34%|███▍ | 10174/30000 [6:30:49<12:32:39, 2.28s/it] 34%|███▍ | 10175/30000 [6:30:52<12:41:04, 2.30s/it] 34%|███▍ | 10176/30000 [6:30:54<12:47:36, 2.32s/it] 34%|███▍ | 10177/30000 [6:30:56<12:42:57, 2.31s/it] 34%|███▍ | 10178/30000 [6:30:59<12:40:26, 2.30s/it] 34%|███▍ | 10179/30000 [6:31:01<12:41:15, 2.30s/it] 34%|███▍ | 10180/30000 [6:31:03<12:37:54, 2.29s/it] 34%|███▍ | 10180/30000 [6:31:03<12:37:54, 2.29s/it] 34%|███▍ | 10181/30000 [6:31:05<12:36:02, 2.29s/it] 34%|███▍ | 10182/30000 [6:31:08<12:34:30, 2.28s/it] 34%|███▍ | 10183/30000 [6:31:10<12:42:44, 2.31s/it] 34%|███▍ | 10184/30000 [6:31:12<12:39:13, 2.30s/it] 34%|███▍ | 10185/30000 [6:31:15<12:36:25, 2.29s/it] 34%|███▍ | 10186/30000 [6:31:17<12:33:04, 2.28s/it] 34%|███▍ | 10187/30000 [6:31:19<12:32:47, 2.28s/it] 34%|███▍ | 10188/30000 [6:31:21<12:32:53, 2.28s/it] 34%|███▍ | 10189/30000 [6:31:24<12:31:12, 2.28s/it] 34%|███▍ | 10190/30000 [6:31:26<12:36:39, 2.29s/it] 34%|███▍ | 10190/30000 [6:31:26<12:36:39, 2.29s/it] 34%|███▍ | 10191/30000 [6:31:28<12:39:54, 2.30s/it] 34%|███▍ | 10192/30000 [6:31:31<12:35:35, 2.29s/it] 34%|███▍ | 10193/30000 [6:31:33<12:37:05, 2.29s/it] 34%|███▍ | 10194/30000 [6:31:35<12:34:14, 2.28s/it] 34%|███▍ | 10195/30000 [6:31:37<12:34:15, 2.29s/it] 34%|███▍ | 10196/30000 [6:31:40<12:32:51, 2.28s/it] 34%|███▍ | 10197/30000 [6:31:42<12:37:25, 2.29s/it] 34%|███▍ | 10198/30000 [6:31:44<12:36:17, 2.29s/it] 34%|███▍ | 10199/30000 [6:31:47<12:38:43, 2.30s/it] 34%|███▍ | 10200/30000 [6:31:49<12:45:46, 2.32s/it] 34%|███▍ | 10200/30000 [6:31:49<12:45:46, 2.32s/it] 34%|███▍ | 10201/30000 [6:31:51<12:50:52, 2.34s/it] 34%|███▍ | 10202/30000 [6:31:54<12:46:24, 2.32s/it] 34%|███▍ | 10203/30000 [6:31:56<12:40:51, 2.31s/it] 34%|███▍ | 10204/30000 [6:31:58<12:43:42, 2.31s/it] 34%|███▍ | 10205/30000 [6:32:01<12:42:46, 2.31s/it] 34%|███▍ | 10206/30000 [6:32:03<12:39:57, 2.30s/it] 34%|███▍ | 10207/30000 [6:32:05<12:38:10, 2.30s/it] 34%|███▍ | 10208/30000 [6:32:08<12:43:35, 2.31s/it] 34%|███▍ | 10209/30000 [6:32:10<12:46:20, 2.32s/it] 34%|███▍ | 10210/30000 [6:32:12<12:39:10, 2.30s/it] 34%|███▍ | 10210/30000 [6:32:12<12:39:10, 2.30s/it] 34%|███▍ | 10211/30000 [6:32:14<12:43:27, 2.31s/it] 34%|███▍ | 10212/30000 [6:32:17<12:42:03, 2.31s/it] 34%|███▍ | 10213/30000 [6:32:19<12:33:55, 2.29s/it] 34%|███▍ | 10214/30000 [6:32:21<12:31:45, 2.28s/it] 34%|███▍ | 10215/30000 [6:32:24<12:37:32, 2.30s/it] 34%|███▍ | 10216/30000 [6:32:26<12:35:37, 2.29s/it] 34%|███▍ | 10217/30000 [6:32:28<12:31:08, 2.28s/it] 34%|███▍ | 10218/30000 [6:32:30<12:35:14, 2.29s/it] 34%|███▍ | 10219/30000 [6:32:33<12:35:07, 2.29s/it] 34%|███▍ | 10220/30000 [6:32:35<12:34:30, 2.29s/it] 34%|███▍ | 10220/30000 [6:32:35<12:34:30, 2.29s/it] 34%|███▍ | 10221/30000 [6:32:37<12:40:40, 2.31s/it] 34%|███▍ | 10222/30000 [6:32:40<12:40:23, 2.31s/it] 34%|███▍ | 10223/30000 [6:32:42<12:38:42, 2.30s/it] 34%|███▍ | 10224/30000 [6:32:44<12:41:22, 2.31s/it] 34%|███▍ | 10225/30000 [6:32:47<12:41:34, 2.31s/it] 34%|███▍ | 10226/30000 [6:32:49<12:36:10, 2.29s/it] 34%|███▍ | 10227/30000 [6:32:51<12:32:37, 2.28s/it] 34%|███▍ | 10228/30000 [6:32:53<12:38:15, 2.30s/it] 34%|███▍ | 10229/30000 [6:32:56<12:37:24, 2.30s/it] 34%|███▍ | 10230/30000 [6:32:58<12:37:20, 2.30s/it] 34%|███▍ | 10230/30000 [6:32:58<12:37:20, 2.30s/it] 34%|███▍ | 10231/30000 [6:33:00<12:45:47, 2.32s/it] 34%|███▍ | 10232/30000 [6:33:03<12:43:44, 2.32s/it] 34%|███▍ | 10233/30000 [6:33:05<12:39:32, 2.31s/it] 34%|███▍ | 10234/30000 [6:33:07<12:43:19, 2.32s/it] 34%|███▍ | 10235/30000 [6:33:10<12:40:38, 2.31s/it] 34%|███▍ | 10236/30000 [6:33:12<12:38:15, 2.30s/it] 34%|███▍ | 10237/30000 [6:33:14<12:35:54, 2.29s/it] 34%|███▍ | 10238/30000 [6:33:17<12:37:06, 2.30s/it] 34%|███▍ | 10239/30000 [6:33:19<12:33:47, 2.29s/it] 34%|███▍ | 10240/30000 [6:33:21<12:34:43, 2.29s/it] 34%|███▍ | 10240/30000 [6:33:21<12:34:43, 2.29s/it] 34%|███▍ | 10241/30000 [6:33:23<12:41:36, 2.31s/it] 34%|███▍ | 10242/30000 [6:33:26<12:42:18, 2.31s/it] 34%|███▍ | 10243/30000 [6:33:28<12:36:39, 2.30s/it] 34%|███▍ | 10244/30000 [6:33:30<12:41:00, 2.31s/it] 34%|███▍ | 10245/30000 [6:33:33<12:37:09, 2.30s/it] 34%|███▍ | 10246/30000 [6:33:35<12:32:58, 2.29s/it] 34%|███▍ | 10247/30000 [6:33:37<12:34:28, 2.29s/it] 34%|███▍ | 10248/30000 [6:33:39<12:35:08, 2.29s/it] 34%|███▍ | 10249/30000 [6:33:42<12:34:12, 2.29s/it] 34%|███▍ | 10250/30000 [6:33:44<12:32:13, 2.29s/it] 34%|███▍ | 10250/30000 [6:33:44<12:32:13, 2.29s/it] 34%|███▍ | 10251/30000 [6:33:46<12:39:00, 2.31s/it] 34%|███▍ | 10252/30000 [6:33:49<12:33:37, 2.29s/it] 34%|███▍ | 10253/30000 [6:33:51<12:34:45, 2.29s/it] 34%|███▍ | 10254/30000 [6:33:53<12:41:56, 2.32s/it] 34%|███▍ | 10255/30000 [6:33:56<12:43:42, 2.32s/it] 34%|███▍ | 10256/30000 [6:33:58<12:41:48, 2.32s/it] 34%|███▍ | 10257/30000 [6:34:00<12:43:11, 2.32s/it] 34%|███▍ | 10258/30000 [6:34:03<12:42:47, 2.32s/it] 34%|███▍ | 10259/30000 [6:34:05<12:37:44, 2.30s/it] 34%|███▍ | 10260/30000 [6:34:07<12:37:00, 2.30s/it] 34%|███▍ | 10260/30000 [6:34:07<12:37:00, 2.30s/it] 34%|███▍ | 10261/30000 [6:34:10<12:41:02, 2.31s/it] 34%|███▍ | 10262/30000 [6:34:12<12:37:26, 2.30s/it] 34%|███▍ | 10263/30000 [6:34:14<12:36:41, 2.30s/it] 34%|███▍ | 10264/30000 [6:34:16<12:42:58, 2.32s/it] 34%|███▍ | 10265/30000 [6:34:19<12:41:42, 2.32s/it] 34%|███▍ | 10266/30000 [6:34:21<12:37:10, 2.30s/it] 34%|███▍ | 10267/30000 [6:34:23<12:37:34, 2.30s/it] 34%|███▍ | 10268/30000 [6:34:26<12:34:04, 2.29s/it] 34%|███▍ | 10269/30000 [6:34:28<12:32:19, 2.29s/it] 34%|███▍ | 10270/30000 [6:34:30<12:31:38, 2.29s/it] 34%|███▍ | 10270/30000 [6:34:30<12:31:38, 2.29s/it] 34%|███▍ | 10271/30000 [6:34:33<12:37:57, 2.31s/it] 34%|███▍ | 10272/30000 [6:34:35<12:43:07, 2.32s/it] 34%|███▍ | 10273/30000 [6:34:37<12:41:11, 2.32s/it] 34%|███▍ | 10274/30000 [6:34:39<12:41:43, 2.32s/it] 34%|███▍ | 10275/30000 [6:34:42<12:37:21, 2.30s/it] 34%|███▍ | 10276/30000 [6:34:44<12:32:02, 2.29s/it] 34%|███▍ | 10277/30000 [6:34:46<12:30:09, 2.28s/it] 34%|███▍ | 10278/30000 [6:34:49<12:32:38, 2.29s/it] 34%|███▍ | 10279/30000 [6:34:51<12:31:15, 2.29s/it] 34%|███▍ | 10280/30000 [6:34:53<12:27:47, 2.28s/it] 34%|███▍ | 10280/30000 [6:34:53<12:27:47, 2.28s/it] 34%|███▍ | 10281/30000 [6:34:55<12:33:33, 2.29s/it] 34%|███▍ | 10282/30000 [6:34:58<12:38:05, 2.31s/it] 34%|███▍ | 10283/30000 [6:35:00<12:33:48, 2.29s/it] 34%|███▍ | 10284/30000 [6:35:02<12:34:34, 2.30s/it] 34%|███▍ | 10285/30000 [6:35:05<12:35:23, 2.30s/it] 34%|███▍ | 10286/30000 [6:35:07<12:32:01, 2.29s/it] 34%|███▍ | 10287/30000 [6:35:09<12:31:26, 2.29s/it] 34%|███▍ | 10288/30000 [6:35:12<12:37:13, 2.30s/it] 34%|███▍ | 10289/30000 [6:35:14<12:33:36, 2.29s/it] 34%|███▍ | 10290/30000 [6:35:16<12:32:29, 2.29s/it] 34%|███▍ | 10290/30000 [6:35:16<12:32:29, 2.29s/it] 34%|███▍ | 10291/30000 [6:35:18<12:31:23, 2.29s/it] 34%|███▍ | 10292/30000 [6:35:21<12:39:29, 2.31s/it] 34%|███▍ | 10293/30000 [6:35:23<12:38:34, 2.31s/it] 34%|███▍ | 10294/30000 [6:35:25<12:34:21, 2.30s/it] 34%|███▍ | 10295/30000 [6:35:28<12:35:32, 2.30s/it] 34%|███▍ | 10296/30000 [6:35:30<12:31:20, 2.29s/it] 34%|███▍ | 10297/30000 [6:35:32<12:37:26, 2.31s/it] 34%|███▍ | 10298/30000 [6:35:35<12:38:26, 2.31s/it] 34%|███▍ | 10299/30000 [6:35:37<12:35:49, 2.30s/it] 34%|███▍ | 10300/30000 [6:35:39<12:39:12, 2.31s/it] 34%|███▍ | 10300/30000 [6:35:39<12:39:12, 2.31s/it] 34%|███▍ | 10301/30000 [6:35:41<12:37:34, 2.31s/it] 34%|███▍ | 10302/30000 [6:35:44<12:34:33, 2.30s/it] 34%|███▍ | 10303/30000 [6:35:46<12:35:10, 2.30s/it] 34%|███▍ | 10304/30000 [6:35:48<12:39:06, 2.31s/it] 34%|███▍ | 10305/30000 [6:35:51<12:37:22, 2.31s/it] 34%|███▍ | 10306/30000 [6:35:53<12:33:48, 2.30s/it] 34%|███▍ | 10307/30000 [6:35:55<12:38:28, 2.31s/it] 34%|███▍ | 10308/30000 [6:35:58<12:38:29, 2.31s/it] 34%|███▍ | 10309/30000 [6:36:00<12:34:10, 2.30s/it] 34%|███▍ | 10310/30000 [6:36:02<12:36:35, 2.31s/it] 34%|███▍ | 10310/30000 [6:36:02<12:36:35, 2.31s/it] 34%|███▍ | 10311/30000 [6:36:03<10:19:09, 1.89s/it] 34%|███▍ | 10312/30000 [6:36:19<32:47:19, 6.00s/it] 34%|███▍ | 10313/30000 [6:36:21<26:41:45, 4.88s/it] 34%|███▍ | 10314/30000 [6:36:23<22:33:12, 4.12s/it] 34%|███▍ | 10315/30000 [6:36:26<19:29:55, 3.57s/it] 34%|███▍ | 10316/30000 [6:36:28<17:22:46, 3.18s/it] 34%|███▍ | 10317/30000 [6:36:30<15:56:46, 2.92s/it] 34%|███▍ | 10318/30000 [6:36:33<15:02:30, 2.75s/it] 34%|███▍ | 10319/30000 [6:36:35<14:17:04, 2.61s/it] 34%|███▍ | 10320/30000 [6:36:37<13:43:32, 2.51s/it] 34%|███▍ | 10320/30000 [6:36:37<13:43:32, 2.51s/it] 34%|███▍ | 10321/30000 [6:36:39<13:27:08, 2.46s/it] 34%|███▍ | 10322/30000 [6:36:42<13:14:55, 2.42s/it] 34%|███▍ | 10323/30000 [6:36:44<13:01:07, 2.38s/it] 34%|███▍ | 10324/30000 [6:36:46<13:01:05, 2.38s/it] 34%|███▍ | 10325/30000 [6:36:49<12:52:48, 2.36s/it] 34%|███▍ | 10326/30000 [6:36:51<12:48:39, 2.34s/it] 34%|███▍ | 10327/30000 [6:36:53<12:39:38, 2.32s/it] 34%|███▍ | 10328/30000 [6:36:56<12:38:30, 2.31s/it] 34%|███▍ | 10329/30000 [6:36:58<12:34:02, 2.30s/it] 34%|███▍ | 10330/30000 [6:37:00<12:30:51, 2.29s/it] 34%|███▍ | 10330/30000 [6:37:00<12:30:51, 2.29s/it] 34%|███▍ | 10331/30000 [6:37:03<12:35:53, 2.31s/it] 34%|███▍ | 10332/30000 [6:37:05<12:39:03, 2.32s/it] 34%|███▍ | 10333/30000 [6:37:07<12:34:12, 2.30s/it] 34%|███▍ | 10334/30000 [6:37:10<12:43:29, 2.33s/it] 34%|███▍ | 10335/30000 [6:37:12<12:40:09, 2.32s/it] 34%|███▍ | 10336/30000 [6:37:14<12:33:53, 2.30s/it] 34%|███▍ | 10337/30000 [6:37:16<12:34:53, 2.30s/it] 34%|███▍ | 10338/30000 [6:37:19<12:35:07, 2.30s/it] 34%|███▍ | 10339/30000 [6:37:21<12:31:01, 2.29s/it] 34%|███▍ | 10340/30000 [6:37:23<12:29:45, 2.29s/it] 34%|███▍ | 10340/30000 [6:37:23<12:29:45, 2.29s/it] 34%|███▍ | 10341/30000 [6:37:26<12:33:50, 2.30s/it] 34%|███▍ | 10342/30000 [6:37:28<12:31:41, 2.29s/it] 34%|███▍ | 10343/30000 [6:37:30<12:30:56, 2.29s/it] 34%|███▍ | 10344/30000 [6:37:33<12:39:14, 2.32s/it] 34%|███▍ | 10345/30000 [6:37:35<12:35:22, 2.31s/it] 34%|███▍ | 10346/30000 [6:37:37<12:33:30, 2.30s/it] 34%|███▍ | 10347/30000 [6:37:39<12:40:19, 2.32s/it] 34%|███▍ | 10348/30000 [6:37:42<12:41:03, 2.32s/it] 34%|███▍ | 10349/30000 [6:37:44<12:36:16, 2.31s/it] 34%|███▍ | 10350/30000 [6:37:46<12:38:41, 2.32s/it] 34%|███▍ | 10350/30000 [6:37:46<12:38:41, 2.32s/it] 35%|███▍ | 10351/30000 [6:37:49<12:43:34, 2.33s/it] 35%|███▍ | 10352/30000 [6:37:51<12:40:40, 2.32s/it] 35%|███▍ | 10353/30000 [6:37:53<12:34:51, 2.31s/it] 35%|███▍ | 10354/30000 [6:37:56<12:36:52, 2.31s/it] 35%|███▍ | 10355/30000 [6:37:58<12:31:17, 2.29s/it] 35%|███▍ | 10356/30000 [6:38:00<12:27:26, 2.28s/it] 35%|███▍ | 10357/30000 [6:38:03<12:34:42, 2.31s/it] 35%|███▍ | 10358/30000 [6:38:05<12:28:41, 2.29s/it] 35%|███▍ | 10359/30000 [6:38:07<12:27:32, 2.28s/it] 35%|███▍ | 10360/30000 [6:38:09<12:33:52, 2.30s/it] 35%|███▍ | 10360/30000 [6:38:09<12:33:52, 2.30s/it] 35%|███▍ | 10361/30000 [6:38:12<12:35:28, 2.31s/it] 35%|███▍ | 10362/30000 [6:38:14<12:34:34, 2.31s/it] 35%|███▍ | 10363/30000 [6:38:16<12:33:56, 2.30s/it] 35%|███▍ | 10364/30000 [6:38:19<12:36:50, 2.31s/it] 35%|███▍ | 10365/30000 [6:38:21<12:34:35, 2.31s/it] 35%|███▍ | 10366/30000 [6:38:23<12:31:15, 2.30s/it] 35%|███▍ | 10367/30000 [6:38:26<12:36:19, 2.31s/it] 35%|███▍ | 10368/30000 [6:38:28<12:31:11, 2.30s/it] 35%|███▍ | 10369/30000 [6:38:30<12:26:55, 2.28s/it] 35%|███▍ | 10370/30000 [6:38:32<12:31:53, 2.30s/it] 35%|███▍ | 10370/30000 [6:38:32<12:31:53, 2.30s/it] 35%|███▍ | 10371/30000 [6:38:35<12:35:14, 2.31s/it] 35%|███▍ | 10372/30000 [6:38:37<12:33:34, 2.30s/it] 35%|███▍ | 10373/30000 [6:38:39<12:34:12, 2.31s/it] 35%|███▍ | 10374/30000 [6:38:42<12:33:23, 2.30s/it] 35%|███▍ | 10375/30000 [6:38:44<12:30:20, 2.29s/it] 35%|███▍ | 10376/30000 [6:38:46<12:28:54, 2.29s/it] 35%|███▍ | 10377/30000 [6:38:49<12:34:29, 2.31s/it] 35%|███▍ | 10378/30000 [6:38:51<12:29:46, 2.29s/it] 35%|███▍ | 10379/30000 [6:38:53<12:28:04, 2.29s/it] 35%|███▍ | 10380/30000 [6:38:55<12:36:07, 2.31s/it] 35%|███▍ | 10380/30000 [6:38:55<12:36:07, 2.31s/it] 35%|███▍ | 10381/30000 [6:38:58<12:36:55, 2.31s/it] 35%|███▍ | 10382/30000 [6:39:00<12:33:17, 2.30s/it] 35%|███▍ | 10383/30000 [6:39:02<12:30:51, 2.30s/it] 35%|███▍ | 10384/30000 [6:39:05<12:27:42, 2.29s/it] 35%|███▍ | 10385/30000 [6:39:07<12:24:43, 2.28s/it] 35%|███▍ | 10386/30000 [6:39:09<12:24:03, 2.28s/it] 35%|███▍ | 10387/30000 [6:39:11<12:28:17, 2.29s/it] 35%|███▍ | 10388/30000 [6:39:14<12:24:01, 2.28s/it] 35%|███▍ | 10389/30000 [6:39:16<12:27:44, 2.29s/it] 35%|███▍ | 10390/30000 [6:39:18<12:31:25, 2.30s/it] +{'loss': 0.0153, 'grad_norm': 0.090792715549469, 'learning_rate': 2.3898177646922757e-05, 'flow_matching_loss': 0.015280448831617832, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.91} +{'loss': 0.0145, 'grad_norm': 0.07854966074228287, 'learning_rate': 2.3884861044282722e-05, 'flow_matching_loss': 0.014453674945980311, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.91} +{'loss': 0.0147, 'grad_norm': 0.10502863675355911, 'learning_rate': 2.3871533645698732e-05, 'flow_matching_loss': 0.01468636654317379, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.91} +{'loss': 0.0155, 'grad_norm': 0.07289799302816391, 'learning_rate': 2.3858195467364833e-05, 'flow_matching_loss': 0.015492429211735725, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.92} +{'loss': 0.0137, 'grad_norm': 0.07094506919384003, 'learning_rate': 2.3844846525488166e-05, 'flow_matching_loss': 0.013739456050097942, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.92} +{'loss': 0.0148, 'grad_norm': 0.07667036354541779, 'learning_rate': 2.3831486836288957e-05, 'flow_matching_loss': 0.014847536757588386, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.92} +{'loss': 0.0151, 'grad_norm': 0.08083973079919815, 'learning_rate': 2.381811641600048e-05, 'flow_matching_loss': 0.01514466879889369, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.92} +{'loss': 0.0159, 'grad_norm': 0.08967342972755432, 'learning_rate': 2.3804735280869073e-05, 'flow_matching_loss': 0.015857574995607138, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.93} +{'loss': 0.0154, 'grad_norm': 0.07347968220710754, 'learning_rate': 2.3791343447154064e-05, 'flow_matching_loss': 0.015380233153700828, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.93} +{'loss': 0.0153, 'grad_norm': 0.07243114709854126, 'learning_rate': 2.37779409311278e-05, 'flow_matching_loss': 0.015328123327344656, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.93} +{'loss': 0.0163, 'grad_norm': 0.07957731932401657, 'learning_rate': 2.3764527749075596e-05, 'flow_matching_loss': 0.01633472153916955, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.94} +{'loss': 0.0158, 'grad_norm': 0.09710264950990677, 'learning_rate': 2.3751103917295735e-05, 'flow_matching_loss': 0.015757478773593903, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.94} +{'loss': 0.0175, 'grad_norm': 0.08406912535429001, 'learning_rate': 2.373766945209944e-05, 'flow_matching_loss': 0.01745665296912193, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.94} +{'loss': 0.0171, 'grad_norm': 0.0887690931558609, 'learning_rate': 2.3724224369810856e-05, 'flow_matching_loss': 0.017084371577948333, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.94} +{'loss': 0.0164, 'grad_norm': 0.11708499491214752, 'learning_rate': 2.3710768686767015e-05, 'flow_matching_loss': 0.016416219621896745, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.95} +{'loss': 0.0153, 'grad_norm': 0.0643615648150444, 'learning_rate': 2.3697302419317856e-05, 'flow_matching_loss': 0.015348739642649888, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.95} +{'loss': 0.016, 'grad_norm': 0.12805472314357758, 'learning_rate': 2.3683825583826146e-05, 'flow_matching_loss': 0.016016786731779576, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.95} +{'loss': 0.0152, 'grad_norm': 0.07232817262411118, 'learning_rate': 2.3670338196667528e-05, 'flow_matching_loss': 0.01521172933280468, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.96} +{'loss': 0.0144, 'grad_norm': 0.08671021461486816, 'learning_rate': 2.3656840274230443e-05, 'flow_matching_loss': 0.014402250479906797, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.96} +{'loss': 0.0156, 'grad_norm': 0.09499255567789078, 'learning_rate': 2.3643331832916133e-05, 'flow_matching_loss': 0.015610834211111068, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.96} +{'loss': 0.0179, 'grad_norm': 0.09580863267183304, 'learning_rate': 2.362981288913863e-05, 'flow_matching_loss': 0.01793188862502575, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.96} +{'loss': 0.016, 'grad_norm': 0.08713854849338531, 'learning_rate': 2.3616283459324737e-05, 'flow_matching_loss': 0.01597502799704671, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.97} +{'loss': 0.0144, 'grad_norm': 0.08800195902585983, 'learning_rate': 2.3602743559913973e-05, 'flow_matching_loss': 0.014374320860952139, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.97} +{'loss': 0.0147, 'grad_norm': 0.09330254048109055, 'learning_rate': 2.35891932073586e-05, 'flow_matching_loss': 0.014742946065962315, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.97} +{'loss': 0.0157, 'grad_norm': 0.12108491361141205, 'learning_rate': 2.357563241812357e-05, 'flow_matching_loss': 0.015739009622484444, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.98} +{'loss': 0.0156, 'grad_norm': 0.07671075314283371, 'learning_rate': 2.3562061208686522e-05, 'flow_matching_loss': 0.015564217418432235, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.98} +{'loss': 0.016, 'grad_norm': 0.09530340135097504, 'learning_rate': 2.354847959553776e-05, 'flow_matching_loss': 0.01604644991457462, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.98} +{'loss': 0.0151, 'grad_norm': 0.08848127722740173, 'learning_rate': 2.353488759518022e-05, 'flow_matching_loss': 0.015127280447632074, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.99} +{'loss': 0.0154, 'grad_norm': 0.07028676569461823, 'learning_rate': 2.352128522412946e-05, 'flow_matching_loss': 0.015365494973957538, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.99} +{'loss': 0.0152, 'grad_norm': 0.07755360007286072, 'learning_rate': 2.3507672498913653e-05, 'flow_matching_loss': 0.015208727028220893, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.99} +{'loss': 0.016, 'grad_norm': 0.10339933633804321, 'learning_rate': 2.3494049436073537e-05, 'flow_matching_loss': 0.015964175574481487, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 2.99} +{'loss': 0.0158, 'grad_norm': 0.07640321552753448, 'learning_rate': 2.348041605216242e-05, 'flow_matching_loss': 0.015841569844633342, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.0} +{'loss': 0.0156, 'grad_norm': 0.08482292294502258, 'learning_rate': 2.3466772363746146e-05, 'flow_matching_loss': 0.01562760816887021, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.0} +{'loss': 0.0151, 'grad_norm': 0.08816058933734894, 'learning_rate': 2.345311838740309e-05, 'flow_matching_loss': 0.015086142625659705, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.0} +{'loss': 0.0146, 'grad_norm': 0.09501238912343979, 'learning_rate': 2.343945413972411e-05, 'flow_matching_loss': 0.014634321257472039, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.01} +{'loss': 0.0151, 'grad_norm': 0.07990912348031998, 'learning_rate': 2.3425779637312568e-05, 'flow_matching_loss': 0.015055603347718716, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.01} +{'loss': 0.0165, 'grad_norm': 0.09662150591611862, 'learning_rate': 2.3412094896784264e-05, 'flow_matching_loss': 0.01646934598684311, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.01} +{'loss': 0.0172, 'grad_norm': 0.08356604725122452, 'learning_rate': 2.339839993476745e-05, 'flow_matching_loss': 0.01717370990663767, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.01} +{'loss': 0.0157, 'grad_norm': 0.07903715968132019, 'learning_rate': 2.3384694767902804e-05, 'flow_matching_loss': 0.015665557608008383, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.02} +{'loss': 0.0133, 'grad_norm': 0.07235628366470337, 'learning_rate': 2.3370979412843384e-05, 'flow_matching_loss': 0.013255877234041692, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.02} +{'loss': 0.0149, 'grad_norm': 0.0669739842414856, 'learning_rate': 2.3357253886254648e-05, 'flow_matching_loss': 0.014915210753679275, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.02} 35%|███▍ | 10390/30000 [6:39:18<12:31:25, 2.30s/it] 35%|███▍ | 10391/30000 [6:39:21<12:31:33, 2.30s/it] 35%|███▍ | 10392/30000 [6:39:23<12:26:31, 2.28s/it] 35%|███▍ | 10393/30000 [6:39:25<12:20:05, 2.26s/it] 35%|███▍ | 10394/30000 [6:39:27<12:30:04, 2.30s/it] 35%|███▍ | 10395/30000 [6:39:30<12:31:54, 2.30s/it] 35%|███▍ | 10396/30000 [6:39:32<12:30:14, 2.30s/it] 35%|███▍ | 10397/30000 [6:39:34<12:33:35, 2.31s/it] 35%|███▍ | 10398/30000 [6:39:37<12:33:43, 2.31s/it] 35%|███▍ | 10399/30000 [6:39:39<12:31:15, 2.30s/it] 35%|███▍ | 10400/30000 [6:39:41<12:27:42, 2.29s/it] 35%|███▍ | 10400/30000 [6:39:41<12:27:42, 2.29s/it] 35%|███▍ | 10401/30000 [6:39:44<12:32:24, 2.30s/it] 35%|███▍ | 10402/30000 [6:39:46<12:30:19, 2.30s/it] 35%|███▍ | 10403/30000 [6:39:48<12:27:05, 2.29s/it] 35%|███▍ | 10404/30000 [6:39:50<12:28:08, 2.29s/it] 35%|███▍ | 10405/30000 [6:39:53<12:34:41, 2.31s/it] 35%|███▍ | 10406/30000 [6:39:55<12:36:42, 2.32s/it] 35%|███▍ | 10407/30000 [6:39:57<12:34:16, 2.31s/it] 35%|███▍ | 10408/30000 [6:40:00<12:31:43, 2.30s/it] 35%|███▍ | 10409/30000 [6:40:02<12:33:24, 2.31s/it] 35%|███▍ | 10410/30000 [6:40:04<12:32:30, 2.30s/it] 35%|███▍ | 10410/30000 [6:40:04<12:32:30, 2.30s/it] 35%|███▍ | 10411/30000 [6:40:07<12:27:23, 2.29s/it] 35%|███▍ | 10412/30000 [6:40:09<12:27:46, 2.29s/it] 35%|███▍ | 10413/30000 [6:40:11<12:26:11, 2.29s/it] 35%|███▍ | 10414/30000 [6:40:13<12:22:09, 2.27s/it] 35%|███▍ | 10415/30000 [6:40:16<12:23:40, 2.28s/it] 35%|███▍ | 10416/30000 [6:40:18<12:27:16, 2.29s/it] 35%|███▍ | 10417/30000 [6:40:20<12:28:53, 2.29s/it] 35%|███▍ | 10418/30000 [6:40:23<12:26:36, 2.29s/it] 35%|███▍ | 10419/30000 [6:40:25<12:24:23, 2.28s/it] 35%|███▍ | 10420/30000 [6:40:27<12:31:41, 2.30s/it] 35%|███▍ | 10420/30000 [6:40:27<12:31:41, 2.30s/it] 35%|███▍ | 10421/30000 [6:40:30<12:38:21, 2.32s/it] 35%|███▍ | 10422/30000 [6:40:32<12:35:56, 2.32s/it] 35%|███▍ | 10423/30000 [6:40:34<12:31:53, 2.30s/it] 35%|███▍ | 10424/30000 [6:40:36<12:33:52, 2.31s/it] 35%|███▍ | 10425/30000 [6:40:39<12:34:02, 2.31s/it] 35%|███▍ | 10426/30000 [6:40:41<12:26:26, 2.29s/it] 35%|███▍ | 10427/30000 [6:40:43<12:26:03, 2.29s/it] 35%|███▍ | 10428/30000 [6:40:46<12:31:33, 2.30s/it] 35%|███▍ | 10429/30000 [6:40:48<12:33:26, 2.31s/it] 35%|███▍ | 10430/30000 [6:40:50<12:31:12, 2.30s/it] 35%|███▍ | 10430/30000 [6:40:50<12:31:12, 2.30s/it] 35%|███▍ | 10431/30000 [6:40:52<12:27:32, 2.29s/it] 35%|███▍ | 10432/30000 [6:40:55<12:31:16, 2.30s/it] 35%|███▍ | 10433/30000 [6:40:57<12:28:11, 2.29s/it] 35%|███▍ | 10434/30000 [6:40:59<12:26:51, 2.29s/it] 35%|███▍ | 10435/30000 [6:41:02<12:27:32, 2.29s/it] 35%|███▍ | 10436/30000 [6:41:04<12:33:27, 2.31s/it] 35%|███▍ | 10437/30000 [6:41:06<12:27:27, 2.29s/it] 35%|███▍ | 10438/30000 [6:41:09<12:23:49, 2.28s/it] 35%|███▍ | 10439/30000 [6:41:11<12:29:30, 2.30s/it] 35%|███▍ | 10440/30000 [6:41:13<12:28:35, 2.30s/it] 35%|███▍ | 10440/30000 [6:41:13<12:28:35, 2.30s/it] 35%|███▍ | 10441/30000 [6:41:15<12:24:59, 2.29s/it] 35%|███▍ | 10442/30000 [6:41:18<12:22:04, 2.28s/it] 35%|███▍ | 10443/30000 [6:41:20<12:28:56, 2.30s/it] 35%|███▍ | 10444/30000 [6:41:22<12:31:22, 2.31s/it] 35%|███▍ | 10445/30000 [6:41:25<12:30:47, 2.30s/it] 35%|███▍ | 10446/30000 [6:41:27<12:28:20, 2.30s/it] 35%|███▍ | 10447/30000 [6:41:29<12:31:56, 2.31s/it] 35%|███▍ | 10448/30000 [6:41:32<12:33:27, 2.31s/it] 35%|███▍ | 10449/30000 [6:41:34<12:28:35, 2.30s/it] 35%|███▍ | 10450/30000 [6:41:36<12:30:42, 2.30s/it] 35%|███▍ | 10450/30000 [6:41:36<12:30:42, 2.30s/it] 35%|███▍ | 10451/30000 [6:41:38<12:28:36, 2.30s/it] 35%|███▍ | 10452/30000 [6:41:41<12:28:17, 2.30s/it] 35%|███▍ | 10453/30000 [6:41:43<12:24:58, 2.29s/it] 35%|███▍ | 10454/30000 [6:41:45<12:30:00, 2.30s/it] 35%|███▍ | 10455/30000 [6:41:48<12:23:25, 2.28s/it] 35%|███▍ | 10456/30000 [6:41:50<12:18:13, 2.27s/it] 35%|███▍ | 10457/30000 [6:41:52<12:21:08, 2.28s/it] 35%|███▍ | 10458/30000 [6:41:54<12:25:24, 2.29s/it] 35%|███▍ | 10459/30000 [6:41:57<12:23:19, 2.28s/it] 35%|███▍ | 10460/30000 [6:41:59<12:29:34, 2.30s/it] 35%|███▍ | 10460/30000 [6:41:59<12:29:34, 2.30s/it] 35%|███▍ | 10461/30000 [6:42:01<12:29:14, 2.30s/it] 35%|███▍ | 10462/30000 [6:42:04<12:25:22, 2.29s/it] 35%|███▍ | 10463/30000 [6:42:06<12:33:02, 2.31s/it] 35%|███▍ | 10464/30000 [6:42:08<12:37:05, 2.33s/it] 35%|███▍ | 10465/30000 [6:42:11<12:32:21, 2.31s/it] 35%|███▍ | 10466/30000 [6:42:13<12:29:35, 2.30s/it] 35%|███▍ | 10467/30000 [6:42:15<12:34:08, 2.32s/it] 35%|███▍ | 10468/30000 [6:42:18<12:33:04, 2.31s/it] 35%|███▍ | 10469/30000 [6:42:20<12:30:41, 2.31s/it] 35%|███▍ | 10470/30000 [6:42:22<12:34:18, 2.32s/it] 35%|███▍ | 10470/30000 [6:42:22<12:34:18, 2.32s/it] 35%|███▍ | 10471/30000 [6:42:24<12:33:28, 2.31s/it] 35%|███▍ | 10472/30000 [6:42:27<12:28:50, 2.30s/it] 35%|███▍ | 10473/30000 [6:42:29<12:30:05, 2.30s/it] 35%|███▍ | 10474/30000 [6:42:31<12:28:40, 2.30s/it] 35%|███▍ | 10475/30000 [6:42:34<12:25:46, 2.29s/it] 35%|███▍ | 10476/30000 [6:42:36<12:30:09, 2.31s/it] 35%|███▍ | 10477/30000 [6:42:38<12:37:52, 2.33s/it] 35%|███▍ | 10478/30000 [6:42:41<12:38:10, 2.33s/it] 35%|███▍ | 10479/30000 [6:42:43<12:28:33, 2.30s/it] 35%|███▍ | 10480/30000 [6:42:45<12:33:33, 2.32s/it] 35%|███▍ | 10480/30000 [6:42:45<12:33:33, 2.32s/it] 35%|███▍ | 10481/30000 [6:42:48<12:33:24, 2.32s/it] 35%|███▍ | 10482/30000 [6:42:50<12:30:25, 2.31s/it] 35%|███▍ | 10483/30000 [6:42:52<12:30:27, 2.31s/it] 35%|███▍ | 10484/30000 [6:42:54<12:27:10, 2.30s/it] 35%|███▍ | 10485/30000 [6:42:57<12:24:17, 2.29s/it] 35%|███▍ | 10486/30000 [6:42:59<12:20:35, 2.28s/it] 35%|███▍ | 10487/30000 [6:43:01<12:27:19, 2.30s/it] 35%|███▍ | 10488/30000 [6:43:04<12:24:18, 2.29s/it] 35%|███▍ | 10489/30000 [6:43:06<12:24:20, 2.29s/it] 35%|███▍ | 10490/30000 [6:43:08<12:28:47, 2.30s/it] 35%|███▍ | 10490/30000 [6:43:08<12:28:47, 2.30s/it] 35%|███▍ | 10491/30000 [6:43:10<12:28:06, 2.30s/it] 35%|███▍ | 10492/30000 [6:43:13<12:25:33, 2.29s/it] 35%|███▍ | 10493/30000 [6:43:15<12:31:42, 2.31s/it] 35%|███▍ | 10494/30000 [6:43:17<12:33:45, 2.32s/it] 35%|███▍ | 10495/30000 [6:43:20<12:33:08, 2.32s/it] 35%|███▍ | 10496/30000 [6:43:22<12:35:21, 2.32s/it] 35%|███▍ | 10497/30000 [6:43:24<12:31:36, 2.31s/it] 35%|███▍ | 10498/30000 [6:43:27<12:25:15, 2.29s/it] 35%|███▍ | 10499/30000 [6:43:29<12:29:09, 2.31s/it] 35%|███▌ | 10500/30000 [6:43:31<12:28:26, 2.30s/it] 35%|███▌ | 10500/30000 [6:43:31<12:28:26, 2.30s/it] 35%|███▌ | 10501/30000 [6:43:34<12:25:41, 2.29s/it] 35%|███▌ | 10502/30000 [6:43:36<12:23:51, 2.29s/it] 35%|███▌ | 10503/30000 [6:43:38<12:27:56, 2.30s/it] 35%|███▌ | 10504/30000 [6:43:40<12:23:38, 2.29s/it] 35%|███▌ | 10505/30000 [6:43:43<12:20:39, 2.28s/it] 35%|███▌ | 10506/30000 [6:43:45<12:28:33, 2.30s/it] 35%|███▌ | 10507/30000 [6:43:47<12:25:18, 2.29s/it] 35%|███▌ | 10508/30000 [6:43:50<12:25:40, 2.30s/it] 35%|███▌ | 10509/30000 [6:43:52<12:38:26, 2.33s/it] 35%|███▌ | 10510/30000 [6:43:54<12:35:28, 2.33s/it] 35%|███▌ | 10510/30000 [6:43:54<12:35:28, 2.33s/it] 35%|███▌ | 10511/30000 [6:43:57<12:30:27, 2.31s/it] 35%|███▌ | 10512/30000 [6:43:59<12:26:12, 2.30s/it] 35%|███▌ | 10513/30000 [6:44:01<12:26:57, 2.30s/it] 35%|███▌ | 10514/30000 [6:44:03<12:19:46, 2.28s/it] 35%|███▌ | 10515/30000 [6:44:06<12:22:09, 2.29s/it] 35%|███▌ | 10516/30000 [6:44:08<12:31:33, 2.31s/it] 35%|███▌ | 10517/30000 [6:44:10<12:31:20, 2.31s/it] 35%|███▌ | 10518/30000 [6:44:13<12:27:26, 2.30s/it] 35%|███▌ | 10519/30000 [6:44:15<12:30:12, 2.31s/it] 35%|███▌ | 10520/30000 [6:44:17<12:25:43, 2.30s/it] 35%|███▌ | 10520/30000 [6:44:17<12:25:43, 2.30s/it] 35%|███▌ | 10521/30000 [6:44:20<12:25:29, 2.30s/it] 35%|███▌ | 10522/30000 [6:44:22<12:24:47, 2.29s/it] 35%|███▌ | 10523/30000 [6:44:24<12:25:24, 2.30s/it] 35%|███▌ | 10524/30000 [6:44:26<12:22:48, 2.29s/it] 35%|███▌ | 10525/30000 [6:44:29<12:21:25, 2.28s/it] 35%|███▌ | 10526/30000 [6:44:31<12:28:01, 2.30s/it] 35%|███▌ | 10527/30000 [6:44:33<12:27:38, 2.30s/it] 35%|███▌ | 10528/30000 [6:44:36<12:22:49, 2.29s/it] 35%|███▌ | 10529/30000 [6:44:38<12:27:07, 2.30s/it] 35%|███▌ | 10530/30000 [6:44:40<12:24:07, 2.29s/it] 35%|███▌ | 10530/30000 [6:44:40<12:24:07, 2.29s/it] 35%|███▌ | 10531/30000 [6:44:43<12:23:49, 2.29s/it] 35%|███▌ | 10532/30000 [6:44:45<12:24:27, 2.29s/it] 35%|███▌ | 10533/30000 [6:44:47<12:32:31, 2.32s/it] 35%|███▌ | 10534/30000 [6:44:50<12:34:50, 2.33s/it] 35%|███▌ | 10535/30000 [6:44:52<12:29:52, 2.31s/it] 35%|███▌ | 10536/30000 [6:44:54<12:28:46, 2.31s/it] 35%|███▌ | 10537/30000 [6:44:56<12:23:07, 2.29s/it] 35%|███▌ | 10538/30000 [6:44:59<12:18:22, 2.28s/it] 35%|███▌ | 10539/30000 [6:45:01<12:23:00, 2.29s/it] 35%|███▌ | 10540/30000 [6:45:03<12:22:16, 2.29s/it] 35%|███▌ | 10540/30000 [6:45:03<12:22:16, 2.29s/it] 35%|███▌ | 10541/30000 [6:45:05<12:21:30, 2.29s/it] 35%|███▌ | 10542/30000 [6:45:08<12:19:57, 2.28s/it] 35%|███▌ | 10543/30000 [6:45:10<12:26:46, 2.30s/it] 35%|███▌ | 10544/30000 [6:45:12<12:27:06, 2.30s/it] 35%|███▌ | 10545/30000 [6:45:15<12:23:08, 2.29s/it] 35%|███▌ | 10546/30000 [6:45:17<12:22:02, 2.29s/it] 35%|███▌ | 10547/30000 [6:45:19<12:15:33, 2.27s/it] 35%|███▌ | 10548/30000 [6:45:21<12:17:10, 2.27s/it] 35%|███▌ | 10549/30000 [6:45:24<12:15:43, 2.27s/it] 35%|███▌ | 10550/30000 [6:45:26<12:24:37, 2.30s/it] 35%|███▌ | 10550/30000 [6:45:26<12:24:37, 2.30s/it] 35%|███▌ | 10551/30000 [6:45:28<12:29:08, 2.31s/it] 35%|███▌ | 10552/30000 [6:45:31<12:22:37, 2.29s/it] 35%|███▌ | 10553/30000 [6:45:33<12:22:18, 2.29s/it] 35%|███▌ | 10554/30000 [6:45:35<12:29:16, 2.31s/it] 35%|███▌ | 10555/30000 [6:45:38<12:31:42, 2.32s/it] 35%|███▌ | 10556/30000 [6:45:40<12:26:59, 2.31s/it] 35%|███▌ | 10557/30000 [6:45:42<12:26:06, 2.30s/it] 35%|███▌ | 10558/30000 [6:45:45<12:30:32, 2.32s/it] 35%|███▌ | 10559/30000 [6:45:47<12:26:05, 2.30s/it] 35%|███▌ | 10560/30000 [6:45:49<12:24:32, 2.30s/it] 35%|███▌ | 10560/30000 [6:45:49<12:24:32, 2.30s/it] 35%|███▌ | 10561/30000 [6:45:51<12:26:11, 2.30s/it] 35%|███▌ | 10562/30000 [6:45:54<12:24:58, 2.30s/it] 35%|███▌ | 10563/30000 [6:45:56<12:20:32, 2.29s/it] 35%|███▌ | 10564/30000 [6:45:58<12:20:23, 2.29s/it] 35%|███▌ | 10565/30000 [6:46:01<12:26:45, 2.31s/it] 35%|███▌ | 10566/30000 [6:46:03<12:31:07, 2.32s/it] 35%|███▌ | 10567/30000 [6:46:05<12:21:25, 2.29s/it] 35%|███▌ | 10568/30000 [6:46:07<12:19:15, 2.28s/it] 35%|███▌ | 10569/30000 [6:46:10<12:24:07, 2.30s/it] 35%|███▌ | 10570/30000 [6:46:12<12:21:16, 2.29s/it] 35%|███▌ | 10570/30000 [6:46:12<12:21:16, 2.29s/it] 35%|███▌ | 10571/30000 [6:46:14<12:21:14, 2.29s/it] 35%|███▌ | 10572/30000 [6:46:17<12:21:14, 2.29s/it] 35%|███▌ | 10573/30000 [6:46:19<12:19:11, 2.28s/it] 35%|███▌ | 10574/30000 [6:46:21<12:20:10, 2.29s/it] 35%|███▌ | 10575/30000 [6:46:23<12:17:02, 2.28s/it] 35%|███▌ | 10576/30000 [6:46:26<12:20:56, 2.29s/it] 35%|███▌ | 10577/30000 [6:46:28<12:23:46, 2.30s/it] 35%|███▌ | 10578/30000 [6:46:30<12:20:23, 2.29s/it] 35%|███▌ | 10579/30000 [6:46:33<12:18:42, 2.28s/it] 35%|███▌ | 10580/30000 [6:46:35<12:25:06, 2.30s/it] 35%|███▌ | 10580/30000 [6:46:35<12:25:06, 2.30s/it] 35%|███▌ | 10581/30000 [6:46:37<12:24:06, 2.30s/it] 35%|███▌ | 10582/30000 [6:46:40<12:23:17, 2.30s/it] 35%|███▌ | 10583/30000 [6:46:42<12:19:10, 2.28s/it] 35%|███▌ | 10584/30000 [6:46:44<12:25:32, 2.30s/it] 35%|███▌ | 10585/30000 [6:46:46<12:27:25, 2.31s/it] 35%|███▌ | 10586/30000 [6:46:49<12:22:24, 2.29s/it] 35%|███▌ | 10587/30000 [6:46:51<12:15:55, 2.27s/it] 35%|███▌ | 10588/30000 [6:46:53<12:21:23, 2.29s/it] 35%|███▌ | 10589/30000 [6:46:56<12:20:16, 2.29s/it] 35%|███▌ | 10590/30000 [6:46:58<12:19:11, 2.28s/it] 35%|███▌ | 10590/30000 [6:46:58<12:19:11, 2.28s/it] 35%|███▌ | 10591/30000 [6:47:00<12:25:11, 2.30s/it] 35%|███▌ | 10592/30000 [6:47:02<12:22:15, 2.29s/it] 35%|███▌ | 10593/30000 [6:47:05<12:19:24, 2.29s/it] 35%|███▌ | 10594/30000 [6:47:07<12:18:43, 2.28s/it] 35%|███▌ | 10595/30000 [6:47:09<12:23:35, 2.30s/it] 35%|███▌ | 10596/30000 [6:47:12<12:23:03, 2.30s/it] 35%|███▌ | 10597/30000 [6:47:14<12:13:40, 2.27s/it] 35%|███▌ | 10598/30000 [6:47:16<12:15:27, 2.27s/it] 35%|███▌ | 10599/30000 [6:47:18<12:21:14, 2.29s/it] 35%|███▌ | 10600/30000 [6:47:21<12:21:29, 2.29s/it] 35%|███▌ | 10600/30000 [6:47:21<12:21:29, 2.29s/it] 35%|███▌ | 10601/30000 [6:47:23<12:18:15, 2.28s/it] 35%|███▌ | 10602/30000 [6:47:25<12:14:30, 2.27s/it] 35%|███▌ | 10603/30000 [6:47:28<12:20:12, 2.29s/it] 35%|███▌ | 10604/30000 [6:47:30<12:28:26, 2.32s/it] 35%|███▌ | 10605/30000 [6:47:32<12:27:07, 2.31s/it] 35%|███▌ | 10606/30000 [6:47:35<12:23:59, 2.30s/it] 35%|███▌ | 10607/30000 [6:47:37<12:30:51, 2.32s/it] 35%|███▌ | 10608/30000 [6:47:39<12:31:53, 2.33s/it] 35%|███▌ | 10609/30000 [6:47:42<12:23:17, 2.30s/it] 35%|███▌ | 10610/30000 [6:47:44<12:22:59, 2.30s/it] 35%|███▌ | 10610/30000 [6:47:44<12:22:59, 2.30s/it] 35%|███▌ | 10611/30000 [6:47:46<12:30:41, 2.32s/it] 35%|███▌ | 10612/30000 [6:47:48<12:27:21, 2.31s/it] 35%|███▌ | 10613/30000 [6:47:51<12:26:23, 2.31s/it] 35%|███▌ | 10614/30000 [6:47:53<12:19:04, 2.29s/it] 35%|███▌ | 10615/30000 [6:47:55<12:25:03, 2.31s/it] 35%|███▌ | 10616/30000 [6:47:58<12:26:53, 2.31s/it] 35%|███▌ | 10617/30000 [6:48:00<12:18:35, 2.29s/it] 35%|███▌ | 10618/30000 [6:48:02<12:21:11, 2.29s/it] 35%|███▌ | 10619/30000 [6:48:05<12:22:04, 2.30s/it] 35%|███▌ | 10620/30000 [6:48:07<12:21:33, 2.30s/it] 35%|███▌ | 10620/30000 [6:48:07<12:21:33, 2.30s/it] 35%|███▌ | 10621/30000 [6:48:09<12:18:29, 2.29s/it] 35%|███▌ | 10622/30000 [6:48:11<12:25:48, 2.31s/it] 35%|███▌ | 10623/30000 [6:48:14<12:25:47, 2.31s/it] 35%|███▌ | 10624/30000 [6:48:16<12:21:43, 2.30s/it] 35%|███▌ | 10625/30000 [6:48:18<12:22:59, 2.30s/it] 35%|███▌ | 10626/30000 [6:48:21<12:24:22, 2.31s/it] 35%|███▌ | 10627/30000 [6:48:23<12:27:06, 2.31s/it] 35%|███▌ | 10628/30000 [6:48:25<12:27:29, 2.32s/it] 35%|███▌ | 10629/30000 [6:48:28<12:25:15, 2.31s/it] 35%|███▌ | 10630/30000 [6:48:30<12:24:49, 2.31s/it] 35%|███▌ | 10630/30000 [6:48:30<12:24:49, 2.31s/it] 35%|███▌ | 10631/30000 [6:48:32<12:25:09, 2.31s/it] 35%|███▌ | 10632/30000 [6:48:34<12:19:30, 2.29s/it] 35%|███▌ | 10633/30000 [6:48:37<12:19:19, 2.29s/it] 35%|███▌ | 10634/30000 [6:48:39<12:19:37, 2.29s/it] 35%|███▌ | 10635/30000 [6:48:41<12:18:54, 2.29s/it] 35%|███▌ | 10636/30000 [6:48:44<12:17:56, 2.29s/it] 35%|███▌ | 10637/30000 [6:48:46<12:18:10, 2.29s/it] 35%|███▌ | 10638/30000 [6:48:48<12:22:59, 2.30s/it] 35%|███▌ | 10639/30000 [6:48:50<12:15:18, 2.28s/it] 35%|███▌ | 10640/30000 [6:48:53<12:14:59, 2.28s/it] 35%|███▌ | 10640/30000 [6:48:53<12:14:59, 2.28s/it] 35%|███▌ | 10641/30000 [6:48:55<12:19:37, 2.29s/it] 35%|███▌ | 10642/30000 [6:48:57<12:18:21, 2.29s/it] 35%|███▌ | 10643/30000 [6:49:00<12:14:38, 2.28s/it] 35%|███▌ | 10644/30000 [6:49:02<12:15:14, 2.28s/it] 35%|███▌ | 10645/30000 [6:49:04<12:21:17, 2.30s/it] 35%|███▌ | 10646/30000 [6:49:07<12:26:20, 2.31s/it] 35%|███▌ | 10647/30000 [6:49:09<12:22:48, 2.30s/it] 35%|███▌ | 10648/30000 [6:49:11<12:23:23, 2.30s/it] 35%|███▌ | 10649/30000 [6:49:13<12:20:58, 2.30s/it] 36%|███▌ | 10650/30000 [6:49:16<12:18:17, 2.29s/it] 36%|███▌ | 10650/30000 [6:49:16<12:18:17, 2.29s/it] 36%|███▌ | 10651/30000 [6:49:18<12:17:59, 2.29s/it] 36%|███▌ | 10652/30000 [6:49:20<12:24:21, 2.31s/it] 36%|███▌ | 10653/30000 [6:49:23<12:23:45, 2.31s/it] 36%|███▌ | 10654/30000 [6:49:25<12:24:19, 2.31s/it] 36%|███▌ | 10655/30000 [6:49:27<12:28:44, 2.32s/it] 36%|███▌ | 10656/30000 [6:49:30<12:23:24, 2.31s/it] 36%|███▌ | 10657/30000 [6:49:32<12:18:02, 2.29s/it] 36%|███▌ | 10658/30000 [6:49:34<12:12:00, 2.27s/it] 36%|███▌ | 10659/30000 [6:49:36<12:16:05, 2.28s/it] 36%|███▌ | 10660/30000 [6:49:39<12:15:59, 2.28s/it] 36%|███▌ | 10660/30000 [6:49:39<12:15:59, 2.28s/it] 36%|███▌ | 10661/30000 [6:49:41<12:17:48, 2.29s/it] 36%|███▌ | 10662/30000 [6:49:43<12:26:09, 2.32s/it] 36%|███▌ | 10663/30000 [6:49:46<12:24:01, 2.31s/it] 36%|███▌ | 10664/30000 [6:49:48<12:20:05, 2.30s/it] 36%|███▌ | 10665/30000 [6:49:50<12:20:24, 2.30s/it] 36%|███▌ | 10666/30000 [6:49:53<12:24:35, 2.31s/it] 36%|███▌ | 10667/30000 [6:49:55<12:18:42, 2.29s/it] 36%|███▌ | 10668/30000 [6:49:57<12:16:22, 2.29s/it] 36%|███▌ | 10669/30000 [6:49:59<12:24:12, 2.31s/it] 36%|███▌ | 10670/30000 [6:50:02<12:25:02, 2.31s/it] 36%|███▌ | 10670/30000 [6:50:02<12:25:02, 2.31s/it] 36%|███▌ | 10671/30000 [6:50:04<12:22:48, 2.31s/it] 36%|███▌ | 10672/30000 [6:50:06<12:18:58, 2.29s/it] 36%|███▌ | 10673/30000 [6:50:09<12:23:12, 2.31s/it] 36%|███▌ | 10674/30000 [6:50:11<12:21:51, 2.30s/it] 36%|███▌ | 10675/30000 [6:50:13<12:20:08, 2.30s/it] 36%|███▌ | 10676/30000 [6:50:16<12:27:57, 2.32s/it] 36%|███▌ | 10677/30000 [6:50:18<12:26:46, 2.32s/it] 36%|███▌ | 10678/30000 [6:50:20<12:23:29, 2.31s/it] 36%|███▌ | 10679/30000 [6:50:22<12:20:02, 2.30s/it] 36%|███▌ | 10680/30000 [6:50:25<12:23:16, 2.31s/it] 36%|███▌ | 10680/30000 [6:50:25<12:23:16, 2.31s/it] 36%|███▌ | 10681/30000 [6:50:27<12:21:59, 2.30s/it] 36%|███▌ | 10682/30000 [6:50:29<12:17:01, 2.29s/it] 36%|███▌ | 10683/30000 [6:50:32<12:21:21, 2.30s/it] 36%|███▌ | 10684/30000 [6:50:34<12:26:18, 2.32s/it] 36%|███▌ | 10685/30000 [6:50:36<12:27:32, 2.32s/it] 36%|███▌ | 10686/30000 [6:50:39<12:28:19, 2.32s/it] 36%|███▌ | 10687/30000 [6:50:41<12:30:04, 2.33s/it] 36%|███▌ | 10688/30000 [6:50:43<12:22:18, 2.31s/it] 36%|███▌ | 10689/30000 [6:50:46<12:21:26, 2.30s/it] 36%|███▌ | 10690/30000 [6:50:48<12:25:27, 2.32s/it] 36%|███▌ | 10690/30000 [6:50:48<12:25:27, 2.32s/it] 36%|███▌ | 10691/30000 [6:50:50<12:26:27, 2.32s/it] 36%|███▌ | 10692/30000 [6:50:53<12:22:09, 2.31s/it] 36%|███▌ | 10693/30000 [6:50:55<12:26:07, 2.32s/it] 36%|███▌ | 10694/30000 [6:50:57<12:21:29, 2.30s/it] 36%|███▌ | 10695/30000 [6:50:59<12:19:25, 2.30s/it] 36%|███▌ | 10696/30000 [6:51:02<12:19:21, 2.30s/it] 36%|███▌ | 10697/30000 [6:51:04<12:19:56, 2.30s/it] 36%|███▌ | 10698/30000 [6:51:06<12:16:47, 2.29s/it] 36%|███▌ | 10699/30000 [6:51:09<12:17:03, 2.29s/it] 36%|███▌ | 10700/30000 [6:51:11<12:24:58, 2.32s/it] 36%|███▌ | 10700/30000 [6:51:11<12:24:58, 2.32s/it] 36%|███▌ | 10701/30000 [6:51:13<12:27:01, 2.32s/it] 36%|███▌ | 10702/30000 [6:51:16<12:20:38, 2.30s/it] 36%|███▌ | 10703/30000 [6:51:18<12:25:07, 2.32s/it] 36%|███▌ | 10704/30000 [6:51:20<12:23:56, 2.31s/it] 36%|███▌ | 10705/30000 [6:51:22<12:17:35, 2.29s/it] 36%|███▌ | 10706/30000 [6:51:25<12:23:54, 2.31s/it] 36%|███▌ | 10707/30000 [6:51:27<12:23:55, 2.31s/it] 36%|███▌ | 10708/30000 [6:51:29<12:20:51, 2.30s/it] 36%|███▌ | 10709/30000 [6:51:32<12:23:24, 2.31s/it] 36%|███▌ | 10710/30000 [6:51:34<12:26:28, 2.32s/it] 36%|███▌ | 10710/30000 [6:51:34<12:26:28, 2.32s/it] 36%|███▌ | 10711/30000 [6:51:36<12:21:20, 2.31s/it] 36%|███▌ | 10712/30000 [6:51:39<12:14:45, 2.29s/it] 36%|███▌ | 10713/30000 [6:51:41<12:20:26, 2.30s/it] 36%|███▌ | 10714/30000 [6:51:43<12:17:53, 2.30s/it] 36%|███▌ | 10715/30000 [6:51:45<12:13:11, 2.28s/it] 36%|███▌ | 10716/30000 [6:51:48<12:22:11, 2.31s/it] 36%|███▌ | 10717/30000 [6:51:50<12:23:30, 2.31s/it] 36%|███▌ | 10718/30000 [6:51:52<12:21:08, 2.31s/it] 36%|███▌ | 10719/30000 [6:51:55<12:19:14, 2.30s/it] 36%|███▌ | 10720/30000 [6:51:57<12:24:10, 2.32s/it] 36%|███▌ | 10720/30000 [6:51:57<12:24:10, 2.32s/it] 36%|███▌ | 10721/30000 [6:51:59<12:18:29, 2.30s/it] 36%|███▌ | 10722/30000 [6:52:02<12:12:27, 2.28s/it] 36%|███▌ | 10723/30000 [6:52:04<12:18:46, 2.30s/it] 36%|███▌ | 10724/30000 [6:52:06<12:13:38, 2.28s/it] 36%|███▌ | 10725/30000 [6:52:09<12:14:22, 2.29s/it] 36%|███▌ | 10726/30000 [6:52:11<12:20:30, 2.31s/it] 36%|███▌ | 10727/30000 [6:52:13<12:18:23, 2.30s/it] 36%|███▌ | 10728/30000 [6:52:15<12:10:53, 2.28s/it] 36%|███▌ | 10729/30000 [6:52:18<12:17:12, 2.30s/it] 36%|███▌ | 10730/30000 [6:52:20<12:12:42, 2.28s/it] 36%|███▌ | 10730/30000 [6:52:20<12:12:42, 2.28s/it] 36%|███▌ | 10731/30000 [6:52:22<12:09:31, 2.27s/it] 36%|███▌ | 10732/30000 [6:52:25<12:18:39, 2.30s/it] 36%|███▌ | 10733/30000 [6:52:27<12:19:46, 2.30s/it] 36%|███▌ | 10734/30000 [6:52:29<12:17:15, 2.30s/it] 36%|███▌ | 10735/30000 [6:52:31<12:21:37, 2.31s/it] 36%|███▌ | 10736/30000 [6:52:34<12:19:02, 2.30s/it] 36%|███▌ | 10737/30000 [6:52:36<12:15:18, 2.29s/it] 36%|███▌ | 10738/30000 [6:52:38<12:15:42, 2.29s/it] 36%|███▌ | 10739/30000 [6:52:41<12:24:14, 2.32s/it] 36%|███▌ | 10740/30000 [6:52:43<12:20:21, 2.31s/it] 36%|███▌ | 10740/30000 [6:52:43<12:20:21, 2.31s/it] 36%|███▌ | 10741/30000 [6:52:45<12:17:39, 2.30s/it] 36%|███▌ | 10742/30000 [6:52:48<12:22:50, 2.31s/it] 36%|███▌ | 10743/30000 [6:52:50<12:22:01, 2.31s/it] 36%|███▌ | 10744/30000 [6:52:52<12:19:31, 2.30s/it] 36%|███▌ | 10745/30000 [6:52:55<12:21:17, 2.31s/it] 36%|███▌ | 10746/30000 [6:52:57<12:21:40, 2.31s/it] 36%|███▌ | 10747/30000 [6:52:59<12:14:41, 2.29s/it] 36%|███▌ | 10748/30000 [6:53:01<12:16:35, 2.30s/it] 36%|███▌ | 10749/30000 [6:53:04<12:20:48, 2.31s/it] 36%|███▌ | 10750/30000 [6:53:06<12:25:26, 2.32s/it] 36%|███▌ | 10750/30000 [6:53:06<12:25:26, 2.32s/it] 36%|███▌ | 10751/30000 [6:53:08<12:17:56, 2.30s/it] 36%|███▌ | 10752/30000 [6:53:11<12:20:35, 2.31s/it] 36%|███▌ | 10753/30000 [6:53:13<12:21:55, 2.31s/it] 36%|███▌ | 10754/30000 [6:53:15<12:19:16, 2.30s/it] 36%|███▌ | 10755/30000 [6:53:18<12:16:23, 2.30s/it] 36%|███▌ | 10756/30000 [6:53:20<12:26:03, 2.33s/it] 36%|███▌ | 10757/30000 [6:53:22<12:25:45, 2.33s/it] 36%|███▌ | 10758/30000 [6:53:25<12:22:03, 2.31s/it] 36%|███▌ | 10759/30000 [6:53:27<12:18:04, 2.30s/it] 36%|███▌ | 10760/30000 [6:53:29<12:16:01, 2.30s/it] 36%|███▌ | 10760/30000 [6:53:29<12:16:01, 2.30s/it] 36%|███▌ | 10761/30000 [6:53:31<12:13:17, 2.29s/it] 36%|███▌ | 10762/30000 [6:53:34<12:12:10, 2.28s/it] 36%|███▌ | 10763/30000 [6:53:36<12:16:54, 2.30s/it] 36%|███▌ | 10764/30000 [6:53:38<12:18:22, 2.30s/it] 36%|███▌ | 10765/30000 [6:53:41<12:12:09, 2.28s/it] 36%|███▌ | 10766/30000 [6:53:43<12:16:34, 2.30s/it] 36%|███▌ | 10767/30000 [6:53:45<12:17:32, 2.30s/it] 36%|███▌ | 10768/30000 [6:53:47<12:15:09, 2.29s/it] 36%|███▌ | 10769/30000 [6:53:50<12:17:32, 2.30s/it] 36%|███▌ | 10770/30000 [6:53:52<12:14:39, 2.29s/it] 36%|███▌ | 10770/30000 [6:53:52<12:14:39, 2.29s/it] 36%|███▌ | 10771/30000 [6:53:54<12:12:43, 2.29s/it] 36%|███▌ | 10772/30000 [6:53:57<12:09:25, 2.28s/it] 36%|███▌ | 10773/30000 [6:53:59<12:20:40, 2.31s/it] 36%|███▌ | 10774/30000 [6:54:01<12:22:11, 2.32s/it] 36%|███▌ | 10775/30000 [6:54:04<12:15:01, 2.29s/it] 36%|███▌ | 10776/30000 [6:54:06<12:17:12, 2.30s/it] 36%|███▌ | 10777/30000 [6:54:08<12:18:04, 2.30s/it] 36%|███▌ | 10778/30000 [6:54:10<12:17:08, 2.30s/it] 36%|███▌ | 10779/30000 [6:54:13<12:18:16, 2.30s/it] 36%|███▌ | 10780/30000 [6:54:15<12:24:10, 2.32s/it] 36%|███▌ | 10780/30000 [6:54:15<12:24:10, 2.32s/it] 36%|███▌ | 10781/30000 [6:54:17<12:22:54, 2.32s/it] 36%|███▌ | 10782/30000 [6:54:20<12:18:14, 2.30s/it] 36%|███▌ | 10783/30000 [6:54:22<12:19:33, 2.31s/it] 36%|███▌ | 10784/30000 [6:54:24<12:26:12, 2.33s/it] 36%|███▌ | 10785/30000 [6:54:27<12:20:14, 2.31s/it] 36%|███▌ | 10786/30000 [6:54:29<12:18:22, 2.31s/it] 36%|███▌ | 10787/30000 [6:54:31<12:14:24, 2.29s/it] 36%|███▌ | 10788/30000 [6:54:34<12:18:52, 2.31s/it] 36%|███▌ | 10789/30000 [6:54:36<12:21:46, 2.32s/it] 36%|███▌ | 10790/30000 [6:54:38<12:13:13, 2.29s/it] 36%|███▌ | 10790/30000 [6:54:38<12:13:13, 2.29s/it] 36%|███▌ | 10791/30000 [6:54:40<12:17:15, 2.30s/it] 36%|███▌ | 10792/30000 [6:54:43<12:17:55, 2.31s/it] 36%|███▌ | 10793/30000 [6:54:45<12:11:10, 2.28s/it] 36%|███▌ | 10794/30000 [6:54:47<12:07:57, 2.27s/it] 36%|███▌ | 10795/30000 [6:54:50<12:13:53, 2.29s/it] 36%|███▌ | 10796/30000 [6:54:52<12:12:39, 2.29s/it] 36%|███▌ | 10797/30000 [6:54:54<12:09:38, 2.28s/it] 36%|███▌ | 10798/30000 [6:54:56<12:05:12, 2.27s/it] 36%|███▌ | 10799/30000 [6:54:59<12:11:42, 2.29s/it] 36%|███▌ | 10800/30000 [6:55:01<12:12:45, 2.29s/it] +{'loss': 0.014, 'grad_norm': 0.0789620578289032, 'learning_rate': 2.3343518204814407e-05, 'flow_matching_loss': 0.01399024147540331, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.03} +{'loss': 0.0152, 'grad_norm': 0.07901521027088165, 'learning_rate': 2.3329772385212798e-05, 'flow_matching_loss': 0.0152158847078681, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.03} +{'loss': 0.0151, 'grad_norm': 0.07819824665784836, 'learning_rate': 2.3316016444152296e-05, 'flow_matching_loss': 0.015124813746660947, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.03} +{'loss': 0.0157, 'grad_norm': 0.08373551815748215, 'learning_rate': 2.3302250398347668e-05, 'flow_matching_loss': 0.015654110908508302, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.03} +{'loss': 0.0153, 'grad_norm': 0.09592315554618835, 'learning_rate': 2.328847426452595e-05, 'flow_matching_loss': 0.015310317277908325, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.04} +{'loss': 0.0153, 'grad_norm': 0.08175301551818848, 'learning_rate': 2.3274688059426456e-05, 'flow_matching_loss': 0.015259873028844595, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.04} +{'loss': 0.0165, 'grad_norm': 0.07946216315031052, 'learning_rate': 2.3260891799800715e-05, 'flow_matching_loss': 0.01654182290658355, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.04} +{'loss': 0.0158, 'grad_norm': 0.08434632420539856, 'learning_rate': 2.3247085502412482e-05, 'flow_matching_loss': 0.015810263995081186, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.05} +{'loss': 0.0151, 'grad_norm': 0.07897178083658218, 'learning_rate': 2.3233269184037717e-05, 'flow_matching_loss': 0.015070579294115306, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.05} +{'loss': 0.0174, 'grad_norm': 0.08933214098215103, 'learning_rate': 2.3219442861464547e-05, 'flow_matching_loss': 0.01744923125952482, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.05} +{'loss': 0.0147, 'grad_norm': 0.07698569446802139, 'learning_rate': 2.3205606551493255e-05, 'flow_matching_loss': 0.014710999373346568, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.05} +{'loss': 0.016, 'grad_norm': 0.08459034562110901, 'learning_rate': 2.319176027093627e-05, 'flow_matching_loss': 0.016022367868572474, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.06} +{'loss': 0.0161, 'grad_norm': 0.06623293459415436, 'learning_rate': 2.3177904036618117e-05, 'flow_matching_loss': 0.016146218962967397, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.06} +{'loss': 0.0148, 'grad_norm': 0.07118456810712814, 'learning_rate': 2.316403786537544e-05, 'flow_matching_loss': 0.014846988394856453, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.06} +{'loss': 0.016, 'grad_norm': 0.09385216981172562, 'learning_rate': 2.3150161774056933e-05, 'flow_matching_loss': 0.016045124363154174, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.07} +{'loss': 0.0142, 'grad_norm': 0.08517618477344513, 'learning_rate': 2.313627577952336e-05, 'flow_matching_loss': 0.01423181276768446, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.07} +{'loss': 0.0146, 'grad_norm': 0.09296092391014099, 'learning_rate': 2.312237989864752e-05, 'flow_matching_loss': 0.014636414125561714, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.07} +{'loss': 0.0166, 'grad_norm': 0.09841103106737137, 'learning_rate': 2.3108474148314205e-05, 'flow_matching_loss': 0.01664766389876604, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.08} +{'loss': 0.0151, 'grad_norm': 0.08104795962572098, 'learning_rate': 2.309455854542023e-05, 'flow_matching_loss': 0.015077568311244249, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.08} +{'loss': 0.015, 'grad_norm': 0.08692153543233871, 'learning_rate': 2.308063310687435e-05, 'flow_matching_loss': 0.015044816303998232, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.08} +{'loss': 0.0171, 'grad_norm': 0.07784572243690491, 'learning_rate': 2.306669784959729e-05, 'flow_matching_loss': 0.017141340766102076, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.08} +{'loss': 0.0153, 'grad_norm': 0.09102693200111389, 'learning_rate': 2.3052752790521703e-05, 'flow_matching_loss': 0.015262208692729474, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.09} +{'loss': 0.0153, 'grad_norm': 0.0829348936676979, 'learning_rate': 2.3038797946592152e-05, 'flow_matching_loss': 0.015302827395498753, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.09} +{'loss': 0.0153, 'grad_norm': 0.08671735227108002, 'learning_rate': 2.3024833334765087e-05, 'flow_matching_loss': 0.015304843988269567, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.09} +{'loss': 0.0143, 'grad_norm': 0.06356758624315262, 'learning_rate': 2.301085897200883e-05, 'flow_matching_loss': 0.014335061516612768, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.1} +{'loss': 0.0152, 'grad_norm': 0.07158395648002625, 'learning_rate': 2.299687487530354e-05, 'flow_matching_loss': 0.01520595671609044, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.1} +{'loss': 0.0156, 'grad_norm': 0.07198874652385712, 'learning_rate': 2.2982881061641228e-05, 'flow_matching_loss': 0.015587270818650722, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.1} +{'loss': 0.016, 'grad_norm': 0.08251528441905975, 'learning_rate': 2.296887754802569e-05, 'flow_matching_loss': 0.01595425121486187, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.1} +{'loss': 0.0161, 'grad_norm': 0.06464014202356339, 'learning_rate': 2.295486435147251e-05, 'flow_matching_loss': 0.01611713245511055, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.11} +{'loss': 0.0152, 'grad_norm': 0.07106533646583557, 'learning_rate': 2.294084148900905e-05, 'flow_matching_loss': 0.015240859426558017, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.11} +{'loss': 0.0149, 'grad_norm': 0.09394925832748413, 'learning_rate': 2.292680897767441e-05, 'flow_matching_loss': 0.014856058545410633, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.11} +{'loss': 0.0162, 'grad_norm': 0.07793572545051575, 'learning_rate': 2.291276683451941e-05, 'flow_matching_loss': 0.016244260128587484, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.12} +{'loss': 0.0152, 'grad_norm': 0.09921017289161682, 'learning_rate': 2.2898715076606584e-05, 'flow_matching_loss': 0.015232953336089849, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.12} +{'loss': 0.0131, 'grad_norm': 0.06439618021249771, 'learning_rate': 2.2884653721010135e-05, 'flow_matching_loss': 0.013086782768368721, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.12} +{'loss': 0.0168, 'grad_norm': 0.10020428895950317, 'learning_rate': 2.2870582784815946e-05, 'flow_matching_loss': 0.016818947438150645, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.12} +{'loss': 0.0146, 'grad_norm': 0.06057731434702873, 'learning_rate': 2.2856502285121525e-05, 'flow_matching_loss': 0.014591674879193306, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.13} +{'loss': 0.0165, 'grad_norm': 0.10226329416036606, 'learning_rate': 2.2842412239036004e-05, 'flow_matching_loss': 0.01649843892082572, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.13} +{'loss': 0.0152, 'grad_norm': 0.08608047664165497, 'learning_rate': 2.2828312663680125e-05, 'flow_matching_loss': 0.015220282226800918, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.13} +{'loss': 0.0158, 'grad_norm': 0.08744465559720993, 'learning_rate': 2.2814203576186194e-05, 'flow_matching_loss': 0.01581614101305604, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.14} +{'loss': 0.0153, 'grad_norm': 0.07064255326986313, 'learning_rate': 2.2800084993698093e-05, 'flow_matching_loss': 0.01531647825613618, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.14} +{'loss': 0.0151, 'grad_norm': 0.09361318498849869, 'learning_rate': 2.2785956933371224e-05, 'flow_matching_loss': 0.015096652973443269, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.14} 36%|███▌ | 10800/30000 [6:55:01<12:12:45, 2.29s/it] 36%|███▌ | 10801/30000 [6:55:03<12:10:11, 2.28s/it] 36%|███▌ | 10802/30000 [6:55:06<12:10:59, 2.28s/it] 36%|███▌ | 10803/30000 [6:55:08<12:09:57, 2.28s/it] 36%|███▌ | 10804/30000 [6:55:10<12:09:27, 2.28s/it] 36%|███▌ | 10805/30000 [6:55:12<12:06:29, 2.27s/it] 36%|███▌ | 10806/30000 [6:55:15<12:11:50, 2.29s/it] 36%|███▌ | 10807/30000 [6:55:17<12:13:32, 2.29s/it] 36%|███▌ | 10808/30000 [6:55:19<12:10:37, 2.28s/it] 36%|███▌ | 10809/30000 [6:55:22<12:10:33, 2.28s/it] 36%|███▌ | 10810/30000 [6:55:24<12:15:45, 2.30s/it] 36%|███▌ | 10810/30000 [6:55:24<12:15:45, 2.30s/it] 36%|███▌ | 10811/30000 [6:55:26<12:15:28, 2.30s/it] 36%|███▌ | 10812/30000 [6:55:29<12:15:41, 2.30s/it] 36%|███▌ | 10813/30000 [6:55:31<12:18:03, 2.31s/it] 36%|███▌ | 10814/30000 [6:55:33<12:24:22, 2.33s/it] 36%|███▌ | 10815/30000 [6:55:35<12:20:56, 2.32s/it] 36%|███▌ | 10816/30000 [6:55:38<12:16:28, 2.30s/it] 36%|███▌ | 10817/30000 [6:55:40<12:14:49, 2.30s/it] 36%|███▌ | 10818/30000 [6:55:42<12:20:11, 2.32s/it] 36%|███▌ | 10819/30000 [6:55:45<12:20:18, 2.32s/it] 36%|███▌ | 10820/30000 [6:55:47<12:15:03, 2.30s/it] 36%|███▌ | 10820/30000 [6:55:47<12:15:03, 2.30s/it] 36%|███▌ | 10821/30000 [6:55:49<12:14:54, 2.30s/it] 36%|███▌ | 10822/30000 [6:55:52<12:21:38, 2.32s/it] 36%|███▌ | 10823/30000 [6:55:54<12:14:51, 2.30s/it] 36%|███▌ | 10824/30000 [6:55:56<12:13:44, 2.30s/it] 36%|███▌ | 10825/30000 [6:55:58<12:11:10, 2.29s/it] 36%|███▌ | 10826/30000 [6:56:01<12:14:23, 2.30s/it] 36%|███▌ | 10827/30000 [6:56:03<12:16:48, 2.31s/it] 36%|███▌ | 10828/30000 [6:56:05<12:11:33, 2.29s/it] 36%|███▌ | 10829/30000 [6:56:08<12:11:55, 2.29s/it] 36%|███▌ | 10830/30000 [6:56:10<12:20:51, 2.32s/it] 36%|███▌ | 10830/30000 [6:56:10<12:20:51, 2.32s/it] 36%|███▌ | 10831/30000 [6:56:12<12:21:53, 2.32s/it] 36%|███▌ | 10832/30000 [6:56:15<12:17:52, 2.31s/it] 36%|███▌ | 10833/30000 [6:56:17<12:17:31, 2.31s/it] 36%|███▌ | 10834/30000 [6:56:19<12:14:33, 2.30s/it] 36%|███▌ | 10835/30000 [6:56:21<12:11:33, 2.29s/it] 36%|███▌ | 10836/30000 [6:56:24<12:11:46, 2.29s/it] 36%|███▌ | 10837/30000 [6:56:26<12:16:20, 2.31s/it] 36%|███▌ | 10838/30000 [6:56:28<12:12:54, 2.29s/it] 36%|███▌ | 10839/30000 [6:56:31<12:06:56, 2.28s/it] 36%|███▌ | 10840/30000 [6:56:33<12:03:33, 2.27s/it] 36%|███▌ | 10840/30000 [6:56:33<12:03:33, 2.27s/it] 36%|███▌ | 10841/30000 [6:56:35<12:09:53, 2.29s/it] 36%|███▌ | 10842/30000 [6:56:38<12:14:27, 2.30s/it] 36%|███▌ | 10843/30000 [6:56:40<12:09:05, 2.28s/it] 36%|███▌ | 10844/30000 [6:56:42<12:09:50, 2.29s/it] 36%|███▌ | 10845/30000 [6:56:44<12:15:22, 2.30s/it] 36%|███▌ | 10846/30000 [6:56:47<12:15:15, 2.30s/it] 36%|███▌ | 10847/30000 [6:56:49<12:11:43, 2.29s/it] 36%|███▌ | 10848/30000 [6:56:51<12:11:37, 2.29s/it] 36%|███▌ | 10849/30000 [6:56:54<12:18:15, 2.31s/it] 36%|███▌ | 10850/30000 [6:56:56<12:20:08, 2.32s/it] 36%|███▌ | 10850/30000 [6:56:56<12:20:08, 2.32s/it] 36%|███▌ | 10851/30000 [6:56:58<12:15:34, 2.30s/it] 36%|███▌ | 10852/30000 [6:57:01<12:14:44, 2.30s/it] 36%|███▌ | 10853/30000 [6:57:03<12:20:02, 2.32s/it] 36%|███▌ | 10854/30000 [6:57:05<12:17:44, 2.31s/it] 36%|███▌ | 10855/30000 [6:57:07<12:13:18, 2.30s/it] 36%|███▌ | 10856/30000 [6:57:10<12:10:57, 2.29s/it] 36%|███▌ | 10857/30000 [6:57:12<12:11:22, 2.29s/it] 36%|███▌ | 10858/30000 [6:57:14<12:09:58, 2.29s/it] 36%|███▌ | 10859/30000 [6:57:17<12:08:05, 2.28s/it] 36%|███▌ | 10860/30000 [6:57:19<12:10:38, 2.29s/it] 36%|███▌ | 10860/30000 [6:57:19<12:10:38, 2.29s/it] 36%|███▌ | 10861/30000 [6:57:21<12:11:57, 2.29s/it] 36%|███▌ | 10862/30000 [6:57:23<12:08:12, 2.28s/it] 36%|███▌ | 10863/30000 [6:57:26<12:05:09, 2.27s/it] 36%|███▌ | 10864/30000 [6:57:28<12:01:37, 2.26s/it] 36%|███▌ | 10865/30000 [6:57:30<12:04:41, 2.27s/it] 36%|███▌ | 10866/30000 [6:57:32<12:01:04, 2.26s/it] 36%|███▌ | 10867/30000 [6:57:35<11:59:40, 2.26s/it] 36%|███▌ | 10868/30000 [6:57:37<12:11:46, 2.29s/it] 36%|███▌ | 10869/30000 [6:57:39<12:12:25, 2.30s/it] 36%|███▌ | 10870/30000 [6:57:42<12:11:32, 2.29s/it] 36%|███▌ | 10870/30000 [6:57:42<12:11:32, 2.29s/it] 36%|███▌ | 10871/30000 [6:57:44<12:09:14, 2.29s/it] 36%|███▌ | 10872/30000 [6:57:46<12:14:45, 2.30s/it] 36%|███▌ | 10873/30000 [6:57:49<12:14:19, 2.30s/it] 36%|███▌ | 10874/30000 [6:57:51<12:08:04, 2.28s/it] 36%|███▋ | 10875/30000 [6:57:53<12:07:16, 2.28s/it] 36%|███▋ | 10876/30000 [6:57:55<12:14:33, 2.30s/it] 36%|███▋ | 10877/30000 [6:57:58<12:16:30, 2.31s/it] 36%|███▋ | 10878/30000 [6:58:00<12:16:34, 2.31s/it] 36%|███▋ | 10879/30000 [6:58:02<12:14:24, 2.30s/it] 36%|███▋ | 10880/30000 [6:58:05<12:16:55, 2.31s/it] 36%|███▋ | 10880/30000 [6:58:05<12:16:55, 2.31s/it] 36%|███▋ | 10881/30000 [6:58:07<12:14:18, 2.30s/it] 36%|███▋ | 10882/30000 [6:58:09<12:13:06, 2.30s/it] 36%|███▋ | 10883/30000 [6:58:12<12:11:07, 2.29s/it] 36%|███▋ | 10884/30000 [6:58:14<12:13:12, 2.30s/it] 36%|███▋ | 10885/30000 [6:58:16<12:11:19, 2.30s/it] 36%|███▋ | 10886/30000 [6:58:18<12:06:47, 2.28s/it] 36%|███▋ | 10887/30000 [6:58:21<12:11:15, 2.30s/it] 36%|███▋ | 10888/30000 [6:58:23<12:11:36, 2.30s/it] 36%|███▋ | 10889/30000 [6:58:25<12:10:51, 2.29s/it] 36%|███▋ | 10890/30000 [6:58:28<12:08:08, 2.29s/it] 36%|███▋ | 10890/30000 [6:58:28<12:08:08, 2.29s/it] 36%|███▋ | 10891/30000 [6:58:30<12:11:51, 2.30s/it] 36%|███▋ | 10892/30000 [6:58:32<12:08:56, 2.29s/it] 36%|███▋ | 10893/30000 [6:58:35<12:08:34, 2.29s/it] 36%|███▋ | 10894/30000 [6:58:37<12:15:47, 2.31s/it] 36%|███▋ | 10895/30000 [6:58:39<12:12:13, 2.30s/it] 36%|███▋ | 10896/30000 [6:58:41<12:10:00, 2.29s/it] 36%|███▋ | 10897/30000 [6:58:44<12:09:05, 2.29s/it] 36%|███▋ | 10898/30000 [6:58:46<12:13:28, 2.30s/it] 36%|███▋ | 10899/30000 [6:58:48<12:13:28, 2.30s/it] 36%|███▋ | 10900/30000 [6:58:51<12:13:48, 2.31s/it] 36%|███▋ | 10900/30000 [6:58:51<12:13:48, 2.31s/it] 36%|███▋ | 10901/30000 [6:58:53<12:16:55, 2.32s/it] 36%|███▋ | 10902/30000 [6:58:55<12:21:08, 2.33s/it] 36%|███▋ | 10903/30000 [6:58:58<12:17:27, 2.32s/it] 36%|███▋ | 10904/30000 [6:59:00<12:11:56, 2.30s/it] 36%|███▋ | 10905/30000 [6:59:02<12:11:11, 2.30s/it] 36%|███▋ | 10906/30000 [6:59:04<12:11:58, 2.30s/it] 36%|███▋ | 10907/30000 [6:59:07<12:08:32, 2.29s/it] 36%|███▋ | 10908/30000 [6:59:09<12:10:36, 2.30s/it] 36%|███▋ | 10909/30000 [6:59:11<12:16:15, 2.31s/it] 36%|███▋ | 10910/30000 [6:59:14<12:20:58, 2.33s/it] 36%|███▋ | 10910/30000 [6:59:14<12:20:58, 2.33s/it] 36%|███▋ | 10911/30000 [6:59:16<12:18:31, 2.32s/it] 36%|███▋ | 10912/30000 [6:59:18<12:12:48, 2.30s/it] 36%|███▋ | 10913/30000 [6:59:21<12:14:27, 2.31s/it] 36%|███▋ | 10914/30000 [6:59:23<12:11:47, 2.30s/it] 36%|███▋ | 10915/30000 [6:59:25<12:05:53, 2.28s/it] 36%|███▋ | 10916/30000 [6:59:28<12:22:24, 2.33s/it] 36%|███▋ | 10917/30000 [6:59:30<12:21:26, 2.33s/it] 36%|███▋ | 10918/30000 [6:59:32<12:16:52, 2.32s/it] 36%|███▋ | 10919/30000 [6:59:35<12:13:01, 2.30s/it] 36%|███▋ | 10920/30000 [6:59:37<12:15:08, 2.31s/it] 36%|███▋ | 10920/30000 [6:59:37<12:15:08, 2.31s/it] 36%|███▋ | 10921/30000 [6:59:39<12:11:46, 2.30s/it] 36%|███▋ | 10922/30000 [6:59:41<12:07:33, 2.29s/it] 36%|███▋ | 10923/30000 [6:59:44<12:12:14, 2.30s/it] 36%|███▋ | 10924/30000 [6:59:46<12:15:12, 2.31s/it] 36%|███▋ | 10925/30000 [6:59:48<12:12:17, 2.30s/it] 36%|███▋ | 10926/30000 [6:59:51<12:09:22, 2.29s/it] 36%|███▋ | 10927/30000 [6:59:53<12:13:42, 2.31s/it] 36%|███▋ | 10928/30000 [6:59:55<12:16:32, 2.32s/it] 36%|███▋ | 10929/30000 [6:59:58<12:10:16, 2.30s/it] 36%|███▋ | 10930/30000 [7:00:00<12:13:56, 2.31s/it] 36%|███▋ | 10930/30000 [7:00:00<12:13:56, 2.31s/it] 36%|███▋ | 10931/30000 [7:00:02<12:14:32, 2.31s/it] 36%|███▋ | 10932/30000 [7:00:05<12:12:38, 2.31s/it] 36%|███▋ | 10933/30000 [7:00:07<12:17:57, 2.32s/it] 36%|███▋ | 10934/30000 [7:00:09<12:26:37, 2.35s/it] 36%|███▋ | 10935/30000 [7:00:12<12:24:30, 2.34s/it] 36%|███▋ | 10936/30000 [7:00:14<12:17:22, 2.32s/it] 36%|███▋ | 10937/30000 [7:00:16<12:13:33, 2.31s/it] 36%|███▋ | 10938/30000 [7:00:18<12:12:18, 2.31s/it] 36%|███▋ | 10939/30000 [7:00:21<12:07:33, 2.29s/it] 36%|███▋ | 10940/30000 [7:00:23<12:08:01, 2.29s/it] 36%|███▋ | 10940/30000 [7:00:23<12:08:01, 2.29s/it] 36%|███▋ | 10941/30000 [7:00:25<12:14:04, 2.31s/it] 36%|███▋ | 10942/30000 [7:00:28<12:12:35, 2.31s/it] 36%|███▋ | 10943/30000 [7:00:30<12:09:48, 2.30s/it] 36%|███▋ | 10944/30000 [7:00:32<12:10:14, 2.30s/it] 36%|███▋ | 10945/30000 [7:00:35<12:09:52, 2.30s/it] 36%|███▋ | 10946/30000 [7:00:37<12:05:03, 2.28s/it] 36%|███▋ | 10947/30000 [7:00:39<12:08:25, 2.29s/it] 36%|███▋ | 10948/30000 [7:00:41<12:11:07, 2.30s/it] 36%|███▋ | 10949/30000 [7:00:44<12:07:28, 2.29s/it] 36%|███▋ | 10950/30000 [7:00:46<12:02:52, 2.28s/it] 36%|███▋ | 10950/30000 [7:00:46<12:02:52, 2.28s/it] 37%|███▋ | 10951/30000 [7:00:48<12:10:45, 2.30s/it] 37%|███▋ | 10952/30000 [7:00:51<12:11:35, 2.30s/it] 37%|███▋ | 10953/30000 [7:00:53<12:08:51, 2.30s/it] 37%|███▋ | 10954/30000 [7:00:55<12:14:04, 2.31s/it] 37%|███▋ | 10955/30000 [7:00:58<12:13:10, 2.31s/it] 37%|███▋ | 10956/30000 [7:01:00<12:10:05, 2.30s/it] 37%|███▋ | 10957/30000 [7:01:02<12:17:47, 2.32s/it] 37%|███▋ | 10958/30000 [7:01:05<12:19:28, 2.33s/it] 37%|███▋ | 10959/30000 [7:01:07<12:13:54, 2.31s/it] 37%|███▋ | 10960/30000 [7:01:09<12:08:10, 2.29s/it] 37%|███▋ | 10960/30000 [7:01:09<12:08:10, 2.29s/it] 37%|███▋ | 10961/30000 [7:01:11<12:12:39, 2.31s/it] 37%|███▋ | 10962/30000 [7:01:14<12:06:11, 2.29s/it] 37%|███▋ | 10963/30000 [7:01:16<12:00:16, 2.27s/it] 37%|███▋ | 10964/30000 [7:01:18<12:09:20, 2.30s/it] 37%|███▋ | 10965/30000 [7:01:21<12:11:56, 2.31s/it] 37%|███▋ | 10966/30000 [7:01:23<12:09:16, 2.30s/it] 37%|███▋ | 10967/30000 [7:01:25<12:12:50, 2.31s/it] 37%|███▋ | 10968/30000 [7:01:27<12:12:26, 2.31s/it] 37%|███▋ | 10969/30000 [7:01:30<12:07:48, 2.29s/it] 37%|███▋ | 10970/30000 [7:01:32<12:06:08, 2.29s/it] 37%|███▋ | 10970/30000 [7:01:32<12:06:08, 2.29s/it] 37%|███▋ | 10971/30000 [7:01:34<12:10:27, 2.30s/it] 37%|███▋ | 10972/30000 [7:01:37<12:11:21, 2.31s/it] 37%|███▋ | 10973/30000 [7:01:39<12:07:16, 2.29s/it] 37%|███▋ | 10974/30000 [7:01:41<12:14:07, 2.32s/it] 37%|███▋ | 10975/30000 [7:01:44<12:13:22, 2.31s/it] 37%|███▋ | 10976/30000 [7:01:46<12:09:46, 2.30s/it] 37%|███▋ | 10977/30000 [7:01:48<12:11:41, 2.31s/it] 37%|███▋ | 10978/30000 [7:01:50<12:09:51, 2.30s/it] 37%|███▋ | 10979/30000 [7:01:53<12:07:57, 2.30s/it] 37%|███▋ | 10980/30000 [7:01:55<12:07:39, 2.30s/it] 37%|███▋ | 10980/30000 [7:01:55<12:07:39, 2.30s/it] 37%|███▋ | 10981/30000 [7:01:57<12:14:46, 2.32s/it] 37%|███▋ | 10982/30000 [7:02:00<12:19:16, 2.33s/it] 37%|███▋ | 10983/30000 [7:02:02<12:12:21, 2.31s/it] 37%|███▋ | 10984/30000 [7:02:04<12:13:48, 2.32s/it] 37%|███▋ | 10985/30000 [7:02:07<12:09:42, 2.30s/it] 37%|███▋ | 10986/30000 [7:02:09<12:06:31, 2.29s/it] 37%|███▋ | 10987/30000 [7:02:11<12:05:00, 2.29s/it] 37%|███▋ | 10988/30000 [7:02:14<12:10:39, 2.31s/it] 37%|███▋ | 10989/30000 [7:02:16<12:12:35, 2.31s/it] 37%|███▋ | 10990/30000 [7:02:18<12:13:21, 2.31s/it] 37%|███▋ | 10990/30000 [7:02:18<12:13:21, 2.31s/it] 37%|███▋ | 10991/30000 [7:02:21<12:16:28, 2.32s/it] 37%|███▋ | 10992/30000 [7:02:23<12:17:06, 2.33s/it] 37%|███▋ | 10993/30000 [7:02:25<12:11:45, 2.31s/it] 37%|███▋ | 10994/30000 [7:02:27<12:14:33, 2.32s/it] 37%|███▋ | 10995/30000 [7:02:30<12:11:13, 2.31s/it] 37%|███▋ | 10996/30000 [7:02:32<12:07:55, 2.30s/it] 37%|███▋ | 10997/30000 [7:02:34<12:10:57, 2.31s/it] 37%|███▋ | 10998/30000 [7:02:37<12:10:41, 2.31s/it] 37%|███▋ | 10999/30000 [7:02:39<12:06:26, 2.29s/it] 37%|███▋ | 11000/30000 [7:02:41<12:04:04, 2.29s/it] 37%|███▋ | 11000/30000 [7:02:41<12:04:04, 2.29s/it] 37%|███▋ | 11001/30000 [7:02:44<12:09:04, 2.30s/it] 37%|███▋ | 11002/30000 [7:02:46<12:01:32, 2.28s/it] 37%|███▋ | 11003/30000 [7:02:48<11:58:46, 2.27s/it] 37%|███▋ | 11004/30000 [7:02:50<12:06:00, 2.29s/it] 37%|███▋ | 11005/30000 [7:02:53<12:07:51, 2.30s/it] 37%|███▋ | 11006/30000 [7:02:55<12:06:15, 2.29s/it] 37%|███▋ | 11007/30000 [7:02:57<12:08:37, 2.30s/it] 37%|███▋ | 11008/30000 [7:03:00<12:07:10, 2.30s/it] 37%|███▋ | 11009/30000 [7:03:02<12:02:38, 2.28s/it] 37%|███▋ | 11010/30000 [7:03:04<12:04:04, 2.29s/it] 37%|███▋ | 11010/30000 [7:03:04<12:04:04, 2.29s/it] 37%|███▋ | 11011/30000 [7:03:06<12:03:34, 2.29s/it] 37%|███▋ | 11012/30000 [7:03:09<12:01:26, 2.28s/it] 37%|███▋ | 11013/30000 [7:03:11<12:02:01, 2.28s/it] 37%|███▋ | 11014/30000 [7:03:13<12:10:53, 2.31s/it] 37%|███▋ | 11015/30000 [7:03:16<12:08:49, 2.30s/it] 37%|███▋ | 11016/30000 [7:03:18<12:07:58, 2.30s/it] 37%|███▋ | 11017/30000 [7:03:20<12:05:07, 2.29s/it] 37%|███▋ | 11018/30000 [7:03:23<12:10:11, 2.31s/it] 37%|███▋ | 11019/30000 [7:03:25<12:04:03, 2.29s/it] 37%|███▋ | 11020/30000 [7:03:27<11:59:44, 2.28s/it] 37%|███▋ | 11020/30000 [7:03:27<11:59:44, 2.28s/it] 37%|███▋ | 11021/30000 [7:03:29<12:07:22, 2.30s/it] 37%|███▋ | 11022/30000 [7:03:32<12:08:22, 2.30s/it] 37%|███▋ | 11023/30000 [7:03:34<12:09:50, 2.31s/it] 37%|███▋ | 11024/30000 [7:03:36<12:05:52, 2.30s/it] 37%|███▋ | 11025/30000 [7:03:39<12:11:51, 2.31s/it] 37%|███▋ | 11026/30000 [7:03:41<12:08:01, 2.30s/it] 37%|███▋ | 11027/30000 [7:03:43<12:04:45, 2.29s/it] 37%|███▋ | 11028/30000 [7:03:45<12:04:47, 2.29s/it] 37%|███▋ | 11029/30000 [7:03:48<12:09:31, 2.31s/it] 37%|███▋ | 11030/30000 [7:03:50<12:12:40, 2.32s/it] 37%|███▋ | 11030/30000 [7:03:50<12:12:40, 2.32s/it] 37%|███▋ | 11031/30000 [7:03:52<12:07:48, 2.30s/it] 37%|███▋ | 11032/30000 [7:03:55<12:02:23, 2.29s/it] 37%|███▋ | 11033/30000 [7:03:57<12:07:15, 2.30s/it] 37%|███▋ | 11034/30000 [7:03:59<12:06:22, 2.30s/it] 37%|███▋ | 11035/30000 [7:04:02<12:06:23, 2.30s/it] 37%|███▋ | 11036/30000 [7:04:04<12:02:50, 2.29s/it] 37%|███▋ | 11037/30000 [7:04:06<12:09:14, 2.31s/it] 37%|███▋ | 11038/30000 [7:04:08<12:02:58, 2.29s/it] 37%|███▋ | 11039/30000 [7:04:11<11:59:45, 2.28s/it] 37%|███▋ | 11040/30000 [7:04:13<12:03:19, 2.29s/it] 37%|███▋ | 11040/30000 [7:04:13<12:03:19, 2.29s/it] 37%|███▋ | 11041/30000 [7:04:15<12:04:11, 2.29s/it] 37%|███▋ | 11042/30000 [7:04:18<12:00:01, 2.28s/it] 37%|███▋ | 11043/30000 [7:04:20<11:57:42, 2.27s/it] 37%|███▋ | 11044/30000 [7:04:22<12:03:00, 2.29s/it] 37%|███▋ | 11045/30000 [7:04:24<12:01:19, 2.28s/it] 37%|███▋ | 11046/30000 [7:04:27<12:00:32, 2.28s/it] 37%|███▋ | 11047/30000 [7:04:29<12:00:56, 2.28s/it] 37%|███▋ | 11048/30000 [7:04:31<12:04:29, 2.29s/it] 37%|███▋ | 11049/30000 [7:04:34<12:03:02, 2.29s/it] 37%|███▋ | 11050/30000 [7:04:36<12:00:36, 2.28s/it] 37%|███▋ | 11050/30000 [7:04:36<12:00:36, 2.28s/it] 37%|███▋ | 11051/30000 [7:04:38<11:58:08, 2.27s/it] 37%|███▋ | 11052/30000 [7:04:40<12:05:00, 2.30s/it] 37%|███▋ | 11053/30000 [7:04:43<12:06:12, 2.30s/it] 37%|███▋ | 11054/30000 [7:04:45<12:04:15, 2.29s/it] 37%|███▋ | 11055/30000 [7:04:47<12:01:50, 2.29s/it] 37%|███▋ | 11056/30000 [7:04:50<12:06:00, 2.30s/it] 37%|███▋ | 11057/30000 [7:04:52<12:07:36, 2.30s/it] 37%|███▋ | 11058/30000 [7:04:54<12:04:07, 2.29s/it] 37%|███▋ | 11059/30000 [7:04:57<12:03:14, 2.29s/it] 37%|███▋ | 11060/30000 [7:04:59<12:10:33, 2.31s/it] 37%|███▋ | 11060/30000 [7:04:59<12:10:33, 2.31s/it] 37%|███▋ | 11061/30000 [7:05:01<12:10:03, 2.31s/it] 37%|███▋ | 11062/30000 [7:05:03<12:09:34, 2.31s/it] 37%|███▋ | 11063/30000 [7:05:06<12:04:45, 2.30s/it] 37%|███▋ | 11064/30000 [7:05:08<12:10:14, 2.31s/it] 37%|███▋ | 11065/30000 [7:05:10<12:10:24, 2.31s/it] 37%|███▋ | 11066/30000 [7:05:13<12:06:47, 2.30s/it] 37%|███▋ | 11067/30000 [7:05:15<12:06:02, 2.30s/it] 37%|███▋ | 11068/30000 [7:05:17<12:10:04, 2.31s/it] 37%|███▋ | 11069/30000 [7:05:20<12:11:12, 2.32s/it] 37%|███▋ | 11070/30000 [7:05:22<12:02:48, 2.29s/it] 37%|███▋ | 11070/30000 [7:05:22<12:02:48, 2.29s/it] 37%|███▋ | 11071/30000 [7:05:24<12:05:27, 2.30s/it] 37%|███▋ | 11072/30000 [7:05:27<12:07:09, 2.31s/it] 37%|███▋ | 11073/30000 [7:05:29<12:02:52, 2.29s/it] 37%|███▋ | 11074/30000 [7:05:31<12:00:58, 2.29s/it] 37%|███▋ | 11075/30000 [7:05:33<12:03:26, 2.29s/it] 37%|███▋ | 11076/30000 [7:05:36<12:06:42, 2.30s/it] 37%|███▋ | 11077/30000 [7:05:38<12:04:18, 2.30s/it] 37%|███▋ | 11078/30000 [7:05:40<12:02:35, 2.29s/it] 37%|███▋ | 11079/30000 [7:05:43<12:05:50, 2.30s/it] 37%|███▋ | 11080/30000 [7:05:45<12:01:15, 2.29s/it] 37%|███▋ | 11080/30000 [7:05:45<12:01:15, 2.29s/it] 37%|███▋ | 11081/30000 [7:05:47<11:59:47, 2.28s/it] 37%|███▋ | 11082/30000 [7:05:49<12:04:55, 2.30s/it] 37%|███▋ | 11083/30000 [7:05:52<12:07:01, 2.31s/it] 37%|███▋ | 11084/30000 [7:05:54<12:03:41, 2.30s/it] 37%|███▋ | 11085/30000 [7:05:56<12:01:01, 2.29s/it] 37%|███▋ | 11086/30000 [7:05:59<12:08:18, 2.31s/it] 37%|███▋ | 11087/30000 [7:06:01<12:11:04, 2.32s/it] 37%|███▋ | 11088/30000 [7:06:03<12:12:19, 2.32s/it] 37%|███▋ | 11089/30000 [7:06:06<12:08:18, 2.31s/it] 37%|███▋ | 11090/30000 [7:06:08<12:11:11, 2.32s/it] 37%|███▋ | 11090/30000 [7:06:08<12:11:11, 2.32s/it] 37%|███▋ | 11091/30000 [7:06:10<12:11:35, 2.32s/it] 37%|███▋ | 11092/30000 [7:06:13<12:07:34, 2.31s/it] 37%|███▋ | 11093/30000 [7:06:15<12:03:12, 2.30s/it] 37%|███▋ | 11094/30000 [7:06:17<12:04:57, 2.30s/it] 37%|███▋ | 11095/30000 [7:06:19<11:59:14, 2.28s/it] 37%|███▋ | 11096/30000 [7:06:22<11:56:06, 2.27s/it] 37%|███▋ | 11097/30000 [7:06:24<12:03:08, 2.30s/it] 37%|███▋ | 11098/30000 [7:06:26<12:06:30, 2.31s/it] 37%|███▋ | 11099/30000 [7:06:29<12:03:51, 2.30s/it] 37%|███▋ | 11100/30000 [7:06:31<12:04:50, 2.30s/it] 37%|███▋ | 11100/30000 [7:06:31<12:04:50, 2.30s/it] 37%|███▋ | 11101/30000 [7:06:33<12:09:11, 2.32s/it] 37%|███▋ | 11102/30000 [7:06:36<12:13:38, 2.33s/it] 37%|███▋ | 11103/30000 [7:06:38<12:08:42, 2.31s/it] 37%|███▋ | 11104/30000 [7:06:40<12:09:27, 2.32s/it] 37%|███▋ | 11105/30000 [7:06:43<12:09:08, 2.32s/it] 37%|███▋ | 11106/30000 [7:06:45<12:08:51, 2.31s/it] 37%|███▋ | 11107/30000 [7:06:47<12:02:30, 2.29s/it] 37%|███▋ | 11108/30000 [7:06:49<12:02:20, 2.29s/it] 37%|███▋ | 11109/30000 [7:06:52<12:05:38, 2.30s/it] 37%|███▋ | 11110/30000 [7:06:54<12:10:26, 2.32s/it] 37%|███▋ | 11110/30000 [7:06:54<12:10:26, 2.32s/it] 37%|███▋ | 11111/30000 [7:06:56<12:07:26, 2.31s/it] 37%|███▋ | 11112/30000 [7:06:59<11:57:41, 2.28s/it] 37%|███▋ | 11113/30000 [7:07:01<11:59:55, 2.29s/it] 37%|███▋ | 11114/30000 [7:07:03<12:02:49, 2.30s/it] 37%|███▋ | 11115/30000 [7:07:05<11:58:24, 2.28s/it] 37%|███▋ | 11116/30000 [7:07:08<12:01:16, 2.29s/it] 37%|███▋ | 11117/30000 [7:07:10<12:04:51, 2.30s/it] 37%|███▋ | 11118/30000 [7:07:12<12:04:21, 2.30s/it] 37%|███▋ | 11119/30000 [7:07:15<12:00:00, 2.29s/it] 37%|███▋ | 11120/30000 [7:07:17<11:59:46, 2.29s/it] 37%|███▋ | 11120/30000 [7:07:17<11:59:46, 2.29s/it] 37%|███▋ | 11121/30000 [7:07:19<12:00:36, 2.29s/it] 37%|███▋ | 11122/30000 [7:07:22<12:02:00, 2.29s/it] 37%|███▋ | 11123/30000 [7:07:24<11:59:36, 2.29s/it] 37%|███▋ | 11124/30000 [7:07:26<11:59:11, 2.29s/it] 37%|███▋ | 11125/30000 [7:07:28<12:05:03, 2.30s/it] 37%|███▋ | 11126/30000 [7:07:31<12:04:02, 2.30s/it] 37%|███▋ | 11127/30000 [7:07:33<11:59:36, 2.29s/it] 37%|███▋ | 11128/30000 [7:07:35<11:55:53, 2.28s/it] 37%|███▋ | 11129/30000 [7:07:37<11:52:27, 2.27s/it] 37%|███▋ | 11130/30000 [7:07:40<11:48:35, 2.25s/it] 37%|███▋ | 11130/30000 [7:07:40<11:48:35, 2.25s/it] 37%|███▋ | 11131/30000 [7:07:42<11:52:48, 2.27s/it] 37%|███▋ | 11132/30000 [7:07:44<11:58:55, 2.29s/it] 37%|███▋ | 11133/30000 [7:07:47<11:57:30, 2.28s/it] 37%|███▋ | 11134/30000 [7:07:49<11:56:38, 2.28s/it] 37%|███▋ | 11135/30000 [7:07:51<12:03:51, 2.30s/it] 37%|███▋ | 11136/30000 [7:07:54<12:05:38, 2.31s/it] 37%|███▋ | 11137/30000 [7:07:56<12:02:20, 2.30s/it] 37%|███▋ | 11138/30000 [7:07:58<11:58:46, 2.29s/it] 37%|███▋ | 11139/30000 [7:08:00<12:01:39, 2.30s/it] 37%|███▋ | 11140/30000 [7:08:03<12:05:44, 2.31s/it] 37%|███▋ | 11140/30000 [7:08:03<12:05:44, 2.31s/it] 37%|███▋ | 11141/30000 [7:08:05<12:02:08, 2.30s/it] 37%|███▋ | 11142/30000 [7:08:07<12:08:33, 2.32s/it] 37%|███▋ | 11143/30000 [7:08:10<12:07:44, 2.32s/it] 37%|███▋ | 11144/30000 [7:08:12<12:01:43, 2.30s/it] 37%|███▋ | 11145/30000 [7:08:14<11:55:46, 2.28s/it] 37%|███▋ | 11146/30000 [7:08:16<11:55:01, 2.28s/it] 37%|███▋ | 11147/30000 [7:08:19<11:50:49, 2.26s/it] 37%|███▋ | 11148/30000 [7:08:21<11:52:36, 2.27s/it] 37%|███▋ | 11149/30000 [7:08:23<11:59:52, 2.29s/it] 37%|███▋ | 11150/30000 [7:08:26<11:57:31, 2.28s/it] 37%|███▋ | 11150/30000 [7:08:26<11:57:31, 2.28s/it] 37%|███▋ | 11151/30000 [7:08:28<11:57:26, 2.28s/it] 37%|███▋ | 11152/30000 [7:08:30<12:02:46, 2.30s/it] 37%|███▋ | 11153/30000 [7:08:32<12:02:43, 2.30s/it] 37%|███▋ | 11154/30000 [7:08:35<11:59:29, 2.29s/it] 37%|███▋ | 11155/30000 [7:08:37<11:57:53, 2.29s/it] 37%|███▋ | 11156/30000 [7:08:39<12:02:50, 2.30s/it] 37%|███▋ | 11157/30000 [7:08:42<12:02:57, 2.30s/it] 37%|███▋ | 11158/30000 [7:08:44<12:03:13, 2.30s/it] 37%|███▋ | 11159/30000 [7:08:46<12:07:03, 2.32s/it] 37%|███▋ | 11160/30000 [7:08:49<12:05:23, 2.31s/it] 37%|███▋ | 11160/30000 [7:08:49<12:05:23, 2.31s/it] 37%|███▋ | 11161/30000 [7:08:51<12:03:10, 2.30s/it] 37%|███▋ | 11162/30000 [7:08:53<12:06:13, 2.31s/it] 37%|███▋ | 11163/30000 [7:08:56<12:05:10, 2.31s/it] 37%|███▋ | 11164/30000 [7:08:58<12:03:27, 2.30s/it] 37%|███▋ | 11165/30000 [7:09:00<12:05:39, 2.31s/it] 37%|███▋ | 11166/30000 [7:09:02<12:05:41, 2.31s/it] 37%|███▋ | 11167/30000 [7:09:05<12:00:41, 2.30s/it] 37%|███▋ | 11168/30000 [7:09:07<11:58:52, 2.29s/it] 37%|███▋ | 11169/30000 [7:09:09<12:02:52, 2.30s/it] 37%|███▋ | 11170/30000 [7:09:12<12:06:05, 2.31s/it] 37%|███▋ | 11170/30000 [7:09:12<12:06:05, 2.31s/it] 37%|███▋ | 11171/30000 [7:09:14<12:03:15, 2.30s/it] 37%|███▋ | 11172/30000 [7:09:16<12:08:04, 2.32s/it] 37%|███▋ | 11173/30000 [7:09:19<12:07:58, 2.32s/it] 37%|███▋ | 11174/30000 [7:09:21<12:03:50, 2.31s/it] 37%|███▋ | 11175/30000 [7:09:23<12:00:49, 2.30s/it] 37%|███▋ | 11176/30000 [7:09:26<12:02:29, 2.30s/it] 37%|███▋ | 11177/30000 [7:09:28<11:56:47, 2.28s/it] 37%|███▋ | 11178/30000 [7:09:30<11:55:09, 2.28s/it] 37%|███▋ | 11179/30000 [7:09:32<12:00:05, 2.30s/it] 37%|███▋ | 11180/30000 [7:09:35<11:59:32, 2.29s/it] 37%|███▋ | 11180/30000 [7:09:35<11:59:32, 2.29s/it] 37%|███▋ | 11181/30000 [7:09:37<12:00:10, 2.30s/it] 37%|███▋ | 11182/30000 [7:09:39<12:05:24, 2.31s/it] 37%|███▋ | 11183/30000 [7:09:42<12:01:49, 2.30s/it] 37%|███▋ | 11184/30000 [7:09:44<11:53:43, 2.28s/it] 37%|███▋ | 11185/30000 [7:09:46<12:01:03, 2.30s/it] 37%|███▋ | 11186/30000 [7:09:48<12:02:06, 2.30s/it] 37%|███▋ | 11187/30000 [7:09:51<11:56:42, 2.29s/it] 37%|███▋ | 11188/30000 [7:09:53<12:02:37, 2.30s/it] 37%|███▋ | 11189/30000 [7:09:55<12:02:56, 2.31s/it] 37%|███▋ | 11190/30000 [7:09:58<12:01:36, 2.30s/it] 37%|███▋ | 11190/30000 [7:09:58<12:01:36, 2.30s/it] 37%|███▋ | 11191/30000 [7:10:00<11:57:40, 2.29s/it] 37%|███▋ | 11192/30000 [7:10:02<12:01:24, 2.30s/it] 37%|███▋ | 11193/30000 [7:10:04<11:55:05, 2.28s/it] 37%|███▋ | 11194/30000 [7:10:07<11:54:26, 2.28s/it] 37%|███▋ | 11195/30000 [7:10:09<12:01:37, 2.30s/it] 37%|███▋ | 11196/30000 [7:10:11<12:00:24, 2.30s/it] 37%|███▋ | 11197/30000 [7:10:14<11:58:57, 2.29s/it] 37%|███▋ | 11198/30000 [7:10:16<12:02:35, 2.31s/it] 37%|███▋ | 11199/30000 [7:10:18<11:58:45, 2.29s/it] 37%|███▋ | 11200/30000 [7:10:21<11:56:41, 2.29s/it] 37%|███▋ | 11200/30000 [7:10:21<11:56:41, 2.29s/it] 37%|███▋ | 11201/30000 [7:10:23<11:56:47, 2.29s/it] 37%|███▋ | 11202/30000 [7:10:25<11:58:42, 2.29s/it] 37%|███▋ | 11203/30000 [7:10:27<11:53:30, 2.28s/it] 37%|███▋ | 11204/30000 [7:10:30<11:55:13, 2.28s/it] 37%|███▋ | 11205/30000 [7:10:32<12:02:08, 2.31s/it] 37%|███▋ | 11206/30000 [7:10:34<11:59:57, 2.30s/it] 37%|███▋ | 11207/30000 [7:10:37<11:56:22, 2.29s/it] 37%|███▋ | 11208/30000 [7:10:39<12:01:52, 2.30s/it] 37%|███▋ | 11209/30000 [7:10:41<12:00:57, 2.30s/it] 37%|███▋ | 11210/30000 [7:10:44<11:59:47, 2.30s/it] 37%|███▋ | 11210/30000 [7:10:44<11:59:47, 2.30s/it] 37%|███▋ | 11211/30000 [7:10:46<12:00:13, 2.30s/it] 37%|███▋ | 11212/30000 [7:10:48<12:01:19, 2.30s/it] 37%|███▋ | 11213/30000 [7:10:50<12:00:16, 2.30s/it] 37%|███▋ | 11214/30000 [7:10:53<11:56:26, 2.29s/it] 37%|███▋ | 11215/30000 [7:10:55<12:01:13, 2.30s/it] 37%|███▋ | 11216/30000 [7:10:57<12:01:38, 2.31s/it] 37%|███▋ | 11217/30000 [7:11:00<11:56:54, 2.29s/it] 37%|███▋ | 11218/30000 [7:11:02<11:59:58, 2.30s/it] 37%|███▋ | 11219/30000 [7:11:04<11:58:58, 2.30s/it] 37%|███▋ | 11220/30000 [7:11:07<11:59:24, 2.30s/it] +{'loss': 0.0162, 'grad_norm': 0.07742153853178024, 'learning_rate': 2.2771819412372505e-05, 'flow_matching_loss': 0.01618814719840884, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.15} +{'loss': 0.0145, 'grad_norm': 0.07723217457532883, 'learning_rate': 2.2757672447880374e-05, 'flow_matching_loss': 0.01446524541825056, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.15} +{'loss': 0.0155, 'grad_norm': 0.09021779894828796, 'learning_rate': 2.2743516057084712e-05, 'flow_matching_loss': 0.01546453284099698, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.15} +{'loss': 0.0156, 'grad_norm': 0.08657471090555191, 'learning_rate': 2.2729350257186877e-05, 'flow_matching_loss': 0.015567675977945328, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.15} +{'loss': 0.0147, 'grad_norm': 0.07971818745136261, 'learning_rate': 2.2715175065399655e-05, 'flow_matching_loss': 0.0146649525500834, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.16} +{'loss': 0.0147, 'grad_norm': 0.10466691106557846, 'learning_rate': 2.270099049894723e-05, 'flow_matching_loss': 0.014686002396047115, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.16} +{'loss': 0.0154, 'grad_norm': 0.07063841074705124, 'learning_rate': 2.268679657506519e-05, 'flow_matching_loss': 0.015355703327804804, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.16} +{'loss': 0.0153, 'grad_norm': 0.0990372970700264, 'learning_rate': 2.2672593311000498e-05, 'flow_matching_loss': 0.015317789558321238, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.17} +{'loss': 0.0149, 'grad_norm': 0.0790305882692337, 'learning_rate': 2.2658380724011452e-05, 'flow_matching_loss': 0.01491024810820818, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.17} +{'loss': 0.0144, 'grad_norm': 0.07145271450281143, 'learning_rate': 2.264415883136769e-05, 'flow_matching_loss': 0.014408760797232389, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.17} +{'loss': 0.016, 'grad_norm': 0.08453219383955002, 'learning_rate': 2.262992765035015e-05, 'flow_matching_loss': 0.01596514331176877, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.17} +{'loss': 0.0149, 'grad_norm': 0.09978280961513519, 'learning_rate': 2.2615687198251062e-05, 'flow_matching_loss': 0.014927462488412858, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.18} +{'loss': 0.0163, 'grad_norm': 0.0764085054397583, 'learning_rate': 2.2601437492373914e-05, 'flow_matching_loss': 0.016310163866728546, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.18} +{'loss': 0.0167, 'grad_norm': 0.06889592111110687, 'learning_rate': 2.2587178550033444e-05, 'flow_matching_loss': 0.016740838531404734, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.18} +{'loss': 0.0151, 'grad_norm': 0.0708032101392746, 'learning_rate': 2.2572910388555615e-05, 'flow_matching_loss': 0.015051318891346454, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.19} +{'loss': 0.0148, 'grad_norm': 0.064992256462574, 'learning_rate': 2.255863302527758e-05, 'flow_matching_loss': 0.01482566650956869, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.19} +{'loss': 0.0171, 'grad_norm': 0.07450073957443237, 'learning_rate': 2.2544346477547685e-05, 'flow_matching_loss': 0.017122190352529286, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.19} +{'loss': 0.0133, 'grad_norm': 0.07609791308641434, 'learning_rate': 2.253005076272544e-05, 'flow_matching_loss': 0.013338679913431406, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.19} +{'loss': 0.0171, 'grad_norm': 0.09863721579313278, 'learning_rate': 2.2515745898181477e-05, 'flow_matching_loss': 0.017076717037707568, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.2} +{'loss': 0.0166, 'grad_norm': 0.08072847872972488, 'learning_rate': 2.250143190129756e-05, 'flow_matching_loss': 0.016562686953693628, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.2} +{'loss': 0.0154, 'grad_norm': 0.10498252511024475, 'learning_rate': 2.2487108789466547e-05, 'flow_matching_loss': 0.015399628784507513, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.2} +{'loss': 0.0156, 'grad_norm': 0.0737716555595398, 'learning_rate': 2.247277658009236e-05, 'flow_matching_loss': 0.015554165095090866, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.21} +{'loss': 0.0141, 'grad_norm': 0.07801555842161179, 'learning_rate': 2.245843529059e-05, 'flow_matching_loss': 0.014117324445396661, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.21} +{'loss': 0.0144, 'grad_norm': 0.0937659963965416, 'learning_rate': 2.244408493838547e-05, 'flow_matching_loss': 0.01441505951806903, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.21} +{'loss': 0.0146, 'grad_norm': 0.06170578673481941, 'learning_rate': 2.242972554091581e-05, 'flow_matching_loss': 0.014626496750861407, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.22} +{'loss': 0.0149, 'grad_norm': 0.07572762668132782, 'learning_rate': 2.2415357115629045e-05, 'flow_matching_loss': 0.014931832719594241, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.22} +{'loss': 0.0154, 'grad_norm': 0.07334236055612564, 'learning_rate': 2.2400979679984162e-05, 'flow_matching_loss': 0.015417482610791922, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.22} +{'loss': 0.018, 'grad_norm': 0.10359115898609161, 'learning_rate': 2.2386593251451096e-05, 'flow_matching_loss': 0.01797833852469921, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.22} +{'loss': 0.0156, 'grad_norm': 0.08786690980195999, 'learning_rate': 2.2372197847510722e-05, 'flow_matching_loss': 0.015619401074945926, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.23} +{'loss': 0.0145, 'grad_norm': 0.09775540977716446, 'learning_rate': 2.2357793485654813e-05, 'flow_matching_loss': 0.014547743089497089, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.23} +{'loss': 0.0159, 'grad_norm': 0.061824627220630646, 'learning_rate': 2.2343380183386017e-05, 'flow_matching_loss': 0.015903067588806153, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.23} +{'loss': 0.0146, 'grad_norm': 0.06892683357000351, 'learning_rate': 2.232895795821787e-05, 'flow_matching_loss': 0.014599829260259867, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.24} +{'loss': 0.0143, 'grad_norm': 0.08416149020195007, 'learning_rate': 2.2314526827674723e-05, 'flow_matching_loss': 0.014280756842345, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.24} +{'loss': 0.0156, 'grad_norm': 0.0943167433142662, 'learning_rate': 2.230008680929177e-05, 'flow_matching_loss': 0.0155974212102592, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.24} +{'loss': 0.0168, 'grad_norm': 0.0660262480378151, 'learning_rate': 2.2285637920614987e-05, 'flow_matching_loss': 0.016756225004792215, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.24} +{'loss': 0.0146, 'grad_norm': 0.07345934957265854, 'learning_rate': 2.2271180179201143e-05, 'flow_matching_loss': 0.014585093501955271, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.25} +{'loss': 0.017, 'grad_norm': 0.08071698993444443, 'learning_rate': 2.225671360261775e-05, 'flow_matching_loss': 0.01695232056081295, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.25} +{'loss': 0.0159, 'grad_norm': 0.06679379940032959, 'learning_rate': 2.2242238208443068e-05, 'flow_matching_loss': 0.015945157408714293, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.25} +{'loss': 0.0153, 'grad_norm': 0.07699058204889297, 'learning_rate': 2.2227754014266063e-05, 'flow_matching_loss': 0.015291140601038932, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.26} +{'loss': 0.0161, 'grad_norm': 0.07417980581521988, 'learning_rate': 2.22132610376864e-05, 'flow_matching_loss': 0.016107641626149416, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.26} +{'loss': 0.017, 'grad_norm': 0.08416034281253815, 'learning_rate': 2.219875929631441e-05, 'flow_matching_loss': 0.016982611920684575, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.26} +{'loss': 0.015, 'grad_norm': 0.08660227805376053, 'learning_rate': 2.2184248807771074e-05, 'flow_matching_loss': 0.01496037794277072, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.26} 37%|███▋ | 11220/30000 [7:11:07<11:59:24, 2.30s/it] 37%|███▋ | 11221/30000 [7:11:09<11:58:39, 2.30s/it] 37%|███▋ | 11222/30000 [7:11:11<12:04:26, 2.31s/it] 37%|███▋ | 11223/30000 [7:11:13<12:04:58, 2.32s/it] 37%|███▋ | 11224/30000 [7:11:16<12:00:30, 2.30s/it] 37%|███▋ | 11225/30000 [7:11:18<12:02:12, 2.31s/it] 37%|███▋ | 11226/30000 [7:11:20<11:57:38, 2.29s/it] 37%|███▋ | 11227/30000 [7:11:23<11:58:54, 2.30s/it] 37%|███▋ | 11228/30000 [7:11:25<11:55:33, 2.29s/it] 37%|███▋ | 11229/30000 [7:11:27<12:08:47, 2.33s/it] 37%|███▋ | 11230/30000 [7:11:30<12:07:17, 2.32s/it] 37%|███▋ | 11230/30000 [7:11:30<12:07:17, 2.32s/it] 37%|███▋ | 11231/30000 [7:11:32<11:58:04, 2.30s/it] 37%|███▋ | 11232/30000 [7:11:34<12:00:47, 2.30s/it] 37%|███▋ | 11233/30000 [7:11:36<11:59:39, 2.30s/it] 37%|███▋ | 11234/30000 [7:11:39<11:57:24, 2.29s/it] 37%|███▋ | 11235/30000 [7:11:41<11:57:39, 2.29s/it] 37%|███▋ | 11236/30000 [7:11:43<12:05:11, 2.32s/it] 37%|███▋ | 11237/30000 [7:11:46<12:08:54, 2.33s/it] 37%|███▋ | 11238/30000 [7:11:48<12:01:21, 2.31s/it] 37%|███▋ | 11239/30000 [7:11:50<11:59:51, 2.30s/it] 37%|███▋ | 11240/30000 [7:11:53<12:03:04, 2.31s/it] 37%|███▋ | 11240/30000 [7:11:53<12:03:04, 2.31s/it] 37%|███▋ | 11241/30000 [7:11:55<11:57:56, 2.30s/it] 37%|███▋ | 11242/30000 [7:11:57<11:54:16, 2.28s/it] 37%|███▋ | 11243/30000 [7:12:00<11:58:48, 2.30s/it] 37%|███▋ | 11244/30000 [7:12:02<12:02:02, 2.31s/it] 37%|███▋ | 11245/30000 [7:12:04<11:58:55, 2.30s/it] 37%|███▋ | 11246/30000 [7:12:06<11:52:07, 2.28s/it] 37%|███▋ | 11247/30000 [7:12:09<11:57:48, 2.30s/it] 37%|███▋ | 11248/30000 [7:12:11<11:57:32, 2.30s/it] 37%|███▋ | 11249/30000 [7:12:13<11:54:33, 2.29s/it] 38%|███▊ | 11250/30000 [7:12:16<11:55:28, 2.29s/it] 38%|███▊ | 11250/30000 [7:12:16<11:55:28, 2.29s/it] 38%|███▊ | 11251/30000 [7:12:18<11:54:24, 2.29s/it] 38%|███▊ | 11252/30000 [7:12:20<11:53:53, 2.28s/it] 38%|███▊ | 11253/30000 [7:12:22<11:54:38, 2.29s/it] 38%|███▊ | 11254/30000 [7:12:25<12:00:30, 2.31s/it] 38%|███▊ | 11255/30000 [7:12:27<11:58:13, 2.30s/it] 38%|███▊ | 11256/30000 [7:12:29<11:57:38, 2.30s/it] 38%|███▊ | 11257/30000 [7:12:32<11:55:39, 2.29s/it] 38%|███▊ | 11258/30000 [7:12:34<11:59:56, 2.30s/it] 38%|███▊ | 11259/30000 [7:12:36<11:54:05, 2.29s/it] 38%|███▊ | 11260/30000 [7:12:38<11:53:50, 2.29s/it] 38%|███▊ | 11260/30000 [7:12:39<11:53:50, 2.29s/it] 38%|███▊ | 11261/30000 [7:12:41<11:56:41, 2.29s/it] 38%|███▊ | 11262/30000 [7:12:43<12:02:28, 2.31s/it] 38%|███▊ | 11263/30000 [7:12:45<12:00:36, 2.31s/it] 38%|███▊ | 11264/30000 [7:12:48<11:54:07, 2.29s/it] 38%|███▊ | 11265/30000 [7:12:50<11:58:37, 2.30s/it] 38%|███▊ | 11266/30000 [7:12:52<11:57:21, 2.30s/it] 38%|███▊ | 11267/30000 [7:12:55<11:54:03, 2.29s/it] 38%|███▊ | 11268/30000 [7:12:57<11:55:27, 2.29s/it] 38%|███▊ | 11269/30000 [7:12:59<12:01:31, 2.31s/it] 38%|███▊ | 11270/30000 [7:13:02<12:00:59, 2.31s/it] 38%|███▊ | 11270/30000 [7:13:02<12:00:59, 2.31s/it] 38%|███▊ | 11271/30000 [7:13:04<11:53:48, 2.29s/it] 38%|███▊ | 11272/30000 [7:13:06<11:48:10, 2.27s/it] 38%|███▊ | 11273/30000 [7:13:08<11:54:28, 2.29s/it] 38%|███▊ | 11274/30000 [7:13:11<11:58:50, 2.30s/it] 38%|███▊ | 11275/30000 [7:13:13<11:55:05, 2.29s/it] 38%|███▊ | 11276/30000 [7:13:15<11:52:39, 2.28s/it] 38%|███▊ | 11277/30000 [7:13:18<11:58:31, 2.30s/it] 38%|███▊ | 11278/30000 [7:13:20<12:02:47, 2.32s/it] 38%|███▊ | 11279/30000 [7:13:22<11:58:25, 2.30s/it] 38%|███▊ | 11280/30000 [7:13:24<11:56:14, 2.30s/it] 38%|███▊ | 11280/30000 [7:13:24<11:56:14, 2.30s/it] 38%|███▊ | 11281/30000 [7:13:27<12:02:22, 2.32s/it] 38%|███▊ | 11282/30000 [7:13:29<11:57:44, 2.30s/it] 38%|███▊ | 11283/30000 [7:13:31<11:55:42, 2.29s/it] 38%|███▊ | 11284/30000 [7:13:34<11:52:19, 2.28s/it] 38%|███▊ | 11285/30000 [7:13:36<11:59:16, 2.31s/it] 38%|███▊ | 11286/30000 [7:13:38<12:01:44, 2.31s/it] 38%|███▊ | 11287/30000 [7:13:41<11:57:44, 2.30s/it] 38%|███▊ | 11288/30000 [7:13:43<11:53:39, 2.29s/it] 38%|███▊ | 11289/30000 [7:13:45<11:58:39, 2.30s/it] 38%|███▊ | 11290/30000 [7:13:47<11:58:35, 2.30s/it] 38%|███▊ | 11290/30000 [7:13:48<11:58:35, 2.30s/it] 38%|███▊ | 11291/30000 [7:13:50<11:54:12, 2.29s/it] 38%|███▊ | 11292/30000 [7:13:52<11:53:30, 2.29s/it] 38%|███▊ | 11293/30000 [7:13:54<12:00:57, 2.31s/it] 38%|███▊ | 11294/30000 [7:13:57<12:03:04, 2.32s/it] 38%|███▊ | 11295/30000 [7:13:59<11:58:28, 2.30s/it] 38%|███▊ | 11296/30000 [7:14:01<11:58:47, 2.31s/it] 38%|███▊ | 11297/30000 [7:14:04<12:00:29, 2.31s/it] 38%|███▊ | 11298/30000 [7:14:06<11:56:17, 2.30s/it] 38%|███▊ | 11299/30000 [7:14:08<11:52:47, 2.29s/it] 38%|███▊ | 11300/30000 [7:14:10<11:56:48, 2.30s/it] 38%|███▊ | 11300/30000 [7:14:10<11:56:48, 2.30s/it] 38%|███▊ | 11301/30000 [7:14:13<12:00:23, 2.31s/it] 38%|███▊ | 11302/30000 [7:14:15<11:56:47, 2.30s/it] 38%|███▊ | 11303/30000 [7:14:17<11:52:37, 2.29s/it] 38%|███▊ | 11304/30000 [7:14:20<11:57:45, 2.30s/it] 38%|███▊ | 11305/30000 [7:14:22<11:53:26, 2.29s/it] 38%|███▊ | 11306/30000 [7:14:24<11:51:42, 2.28s/it] 38%|███▊ | 11307/30000 [7:14:26<11:51:08, 2.28s/it] 38%|███▊ | 11308/30000 [7:14:29<11:55:46, 2.30s/it] 38%|███▊ | 11309/30000 [7:14:31<11:58:56, 2.31s/it] 38%|███▊ | 11310/30000 [7:14:33<11:58:10, 2.31s/it] 38%|███▊ | 11310/30000 [7:14:34<11:58:10, 2.31s/it] 38%|███▊ | 11311/30000 [7:14:36<11:56:45, 2.30s/it] 38%|███▊ | 11312/30000 [7:14:38<12:00:58, 2.31s/it] 38%|███▊ | 11313/30000 [7:14:40<11:59:29, 2.31s/it] 38%|███▊ | 11314/30000 [7:14:43<11:52:02, 2.29s/it] 38%|███▊ | 11315/30000 [7:14:45<11:53:17, 2.29s/it] 38%|███▊ | 11316/30000 [7:14:47<11:58:42, 2.31s/it] 38%|███▊ | 11317/30000 [7:14:50<12:04:34, 2.33s/it] 38%|███▊ | 11318/30000 [7:14:52<11:59:52, 2.31s/it] 38%|███▊ | 11319/30000 [7:14:54<11:53:15, 2.29s/it] 38%|███▊ | 11320/30000 [7:14:56<11:57:02, 2.30s/it] 38%|███▊ | 11320/30000 [7:14:57<11:57:02, 2.30s/it] 38%|███▊ | 11321/30000 [7:14:59<11:52:12, 2.29s/it] 38%|███▊ | 11322/30000 [7:15:01<11:50:49, 2.28s/it] 38%|███▊ | 11323/30000 [7:15:03<11:57:07, 2.30s/it] 38%|███▊ | 11324/30000 [7:15:06<11:58:07, 2.31s/it] 38%|███▊ | 11325/30000 [7:15:08<11:53:50, 2.29s/it] 38%|███▊ | 11326/30000 [7:15:10<11:52:17, 2.29s/it] 38%|███▊ | 11327/30000 [7:15:13<11:56:04, 2.30s/it] 38%|███▊ | 11328/30000 [7:15:15<11:50:08, 2.28s/it] 38%|███▊ | 11329/30000 [7:15:17<11:46:38, 2.27s/it] 38%|███▊ | 11330/30000 [7:15:19<11:50:58, 2.28s/it] 38%|███▊ | 11330/30000 [7:15:19<11:50:58, 2.28s/it] 38%|███▊ | 11331/30000 [7:15:22<11:51:08, 2.29s/it] 38%|███▊ | 11332/30000 [7:15:24<11:51:20, 2.29s/it] 38%|███▊ | 11333/30000 [7:15:26<11:54:55, 2.30s/it] 38%|███▊ | 11334/30000 [7:15:29<11:55:28, 2.30s/it] 38%|███▊ | 11335/30000 [7:15:31<11:52:32, 2.29s/it] 38%|███▊ | 11336/30000 [7:15:33<11:51:40, 2.29s/it] 38%|███▊ | 11337/30000 [7:15:35<11:58:44, 2.31s/it] 38%|███▊ | 11338/30000 [7:15:38<11:59:27, 2.31s/it] 38%|███▊ | 11339/30000 [7:15:40<11:56:01, 2.30s/it] 38%|███▊ | 11340/30000 [7:15:42<11:57:51, 2.31s/it] 38%|███▊ | 11340/30000 [7:15:42<11:57:51, 2.31s/it] 38%|███▊ | 11341/30000 [7:15:45<11:57:10, 2.31s/it] 38%|███▊ | 11342/30000 [7:15:47<11:50:31, 2.28s/it] 38%|███▊ | 11343/30000 [7:15:49<11:47:37, 2.28s/it] 38%|███▊ | 11344/30000 [7:15:52<11:53:23, 2.29s/it] 38%|███▊ | 11345/30000 [7:15:54<11:48:26, 2.28s/it] 38%|███▊ | 11346/30000 [7:15:56<11:45:44, 2.27s/it] 38%|███▊ | 11347/30000 [7:15:58<11:53:09, 2.29s/it] 38%|███▊ | 11348/30000 [7:16:01<11:54:49, 2.30s/it] 38%|███▊ | 11349/30000 [7:16:03<11:53:53, 2.30s/it] 38%|███▊ | 11350/30000 [7:16:05<11:59:07, 2.31s/it] 38%|███▊ | 11350/30000 [7:16:05<11:59:07, 2.31s/it] 38%|███▊ | 11351/30000 [7:16:08<11:57:09, 2.31s/it] 38%|███▊ | 11352/30000 [7:16:10<11:53:26, 2.30s/it] 38%|███▊ | 11353/30000 [7:16:12<11:52:07, 2.29s/it] 38%|███▊ | 11354/30000 [7:16:15<11:59:24, 2.31s/it] 38%|███▊ | 11355/30000 [7:16:17<11:55:05, 2.30s/it] 38%|███▊ | 11356/30000 [7:16:19<11:53:29, 2.30s/it] 38%|███▊ | 11357/30000 [7:16:21<11:58:51, 2.31s/it] 38%|███▊ | 11358/30000 [7:16:24<12:00:19, 2.32s/it] 38%|███▊ | 11359/30000 [7:16:26<11:54:24, 2.30s/it] 38%|███▊ | 11360/30000 [7:16:28<11:51:38, 2.29s/it] 38%|███▊ | 11360/30000 [7:16:28<11:51:38, 2.29s/it] 38%|███▊ | 11361/30000 [7:16:31<11:55:48, 2.30s/it] 38%|███▊ | 11362/30000 [7:16:33<11:56:20, 2.31s/it] 38%|███▊ | 11363/30000 [7:16:35<11:52:12, 2.29s/it] 38%|███▊ | 11364/30000 [7:16:38<11:58:12, 2.31s/it] 38%|███▊ | 11365/30000 [7:16:40<11:59:10, 2.32s/it] 38%|███▊ | 11366/30000 [7:16:42<11:53:53, 2.30s/it] 38%|███▊ | 11367/30000 [7:16:44<11:49:44, 2.29s/it] 38%|███▊ | 11368/30000 [7:16:47<11:56:44, 2.31s/it] 38%|███▊ | 11369/30000 [7:16:49<11:50:48, 2.29s/it] 38%|███▊ | 11370/30000 [7:16:51<11:51:47, 2.29s/it] 38%|███▊ | 11370/30000 [7:16:51<11:51:47, 2.29s/it] 38%|███▊ | 11371/30000 [7:16:54<11:57:08, 2.31s/it] 38%|███▊ | 11372/30000 [7:16:56<11:58:05, 2.31s/it] 38%|███▊ | 11373/30000 [7:16:58<11:55:50, 2.31s/it] 38%|███▊ | 11374/30000 [7:17:01<11:54:13, 2.30s/it] 38%|███▊ | 11375/30000 [7:17:03<11:56:28, 2.31s/it] 38%|███▊ | 11376/30000 [7:17:05<11:53:24, 2.30s/it] 38%|███▊ | 11377/30000 [7:17:07<11:51:31, 2.29s/it] 38%|███▊ | 11378/30000 [7:17:10<11:55:54, 2.31s/it] 38%|███▊ | 11379/30000 [7:17:12<11:58:54, 2.32s/it] 38%|███▊ | 11380/30000 [7:17:14<11:53:26, 2.30s/it] 38%|███▊ | 11380/30000 [7:17:14<11:53:26, 2.30s/it] 38%|███▊ | 11381/30000 [7:17:17<11:54:04, 2.30s/it] 38%|███▊ | 11382/30000 [7:17:19<11:59:14, 2.32s/it] 38%|███▊ | 11383/30000 [7:17:21<11:54:51, 2.30s/it] 38%|███▊ | 11384/30000 [7:17:24<11:51:40, 2.29s/it] 38%|███▊ | 11385/30000 [7:17:26<11:54:44, 2.30s/it] 38%|███▊ | 11386/30000 [7:17:28<11:54:26, 2.30s/it] 38%|███▊ | 11387/30000 [7:17:30<11:48:37, 2.28s/it] 38%|███▊ | 11388/30000 [7:17:33<11:54:44, 2.30s/it] 38%|███▊ | 11389/30000 [7:17:35<11:55:54, 2.31s/it] 38%|███▊ | 11390/30000 [7:17:37<11:53:45, 2.30s/it] 38%|███▊ | 11390/30000 [7:17:37<11:53:45, 2.30s/it] 38%|███▊ | 11391/30000 [7:17:40<11:53:08, 2.30s/it] 38%|███▊ | 11392/30000 [7:17:42<11:56:23, 2.31s/it] 38%|███▊ | 11393/30000 [7:17:44<11:51:10, 2.29s/it] 38%|███▊ | 11394/30000 [7:17:47<11:49:39, 2.29s/it] 38%|███▊ | 11395/30000 [7:17:49<11:53:02, 2.30s/it] 38%|███▊ | 11396/30000 [7:17:51<11:50:55, 2.29s/it] 38%|███▊ | 11397/30000 [7:17:53<11:49:48, 2.29s/it] 38%|███▊ | 11398/30000 [7:17:56<11:55:46, 2.31s/it] 38%|███▊ | 11399/30000 [7:17:58<11:50:41, 2.29s/it] 38%|███▊ | 11400/30000 [7:18:00<11:52:02, 2.30s/it] 38%|███▊ | 11400/30000 [7:18:00<11:52:02, 2.30s/it] 38%|███▊ | 11401/30000 [7:18:03<11:57:17, 2.31s/it] 38%|███▊ | 11402/30000 [7:18:05<12:00:24, 2.32s/it] 38%|███▊ | 11403/30000 [7:18:07<11:54:48, 2.31s/it] 38%|███▊ | 11404/30000 [7:18:10<11:53:53, 2.30s/it] 38%|███▊ | 11405/30000 [7:18:12<11:56:57, 2.31s/it] 38%|███▊ | 11406/30000 [7:18:14<11:52:06, 2.30s/it] 38%|███▊ | 11407/30000 [7:18:16<11:50:05, 2.29s/it] 38%|███▊ | 11408/30000 [7:18:19<11:56:15, 2.31s/it] 38%|███▊ | 11409/30000 [7:18:21<11:56:05, 2.31s/it] 38%|███▊ | 11410/30000 [7:18:23<11:50:47, 2.29s/it] 38%|███▊ | 11410/30000 [7:18:23<11:50:47, 2.29s/it] 38%|███▊ | 11411/30000 [7:18:26<11:52:19, 2.30s/it] 38%|███▊ | 11412/30000 [7:18:28<11:52:49, 2.30s/it] 38%|███▊ | 11413/30000 [7:18:30<11:49:41, 2.29s/it] 38%|███▊ | 11414/30000 [7:18:33<11:52:06, 2.30s/it] 38%|███▊ | 11415/30000 [7:18:35<11:49:26, 2.29s/it] 38%|███▊ | 11416/30000 [7:18:37<11:46:42, 2.28s/it] 38%|███▊ | 11417/30000 [7:18:39<11:43:46, 2.27s/it] 38%|███▊ | 11418/30000 [7:18:42<11:47:43, 2.29s/it] 38%|███▊ | 11419/30000 [7:18:44<11:44:28, 2.27s/it] 38%|███▊ | 11420/30000 [7:18:46<11:46:05, 2.28s/it] 38%|███▊ | 11420/30000 [7:18:46<11:46:05, 2.28s/it] 38%|███▊ | 11421/30000 [7:18:49<11:51:21, 2.30s/it] 38%|███▊ | 11422/30000 [7:18:51<11:54:41, 2.31s/it] 38%|███▊ | 11423/30000 [7:18:53<11:47:37, 2.29s/it] 38%|███▊ | 11424/30000 [7:18:56<11:54:05, 2.31s/it] 38%|███▊ | 11425/30000 [7:18:58<11:50:23, 2.29s/it] 38%|███▊ | 11426/30000 [7:19:00<11:48:57, 2.29s/it] 38%|███▊ | 11427/30000 [7:19:02<11:44:32, 2.28s/it] 38%|███▊ | 11428/30000 [7:19:05<11:52:57, 2.30s/it] 38%|███▊ | 11429/30000 [7:19:07<11:56:00, 2.31s/it] 38%|███▊ | 11430/30000 [7:19:09<11:53:56, 2.31s/it] 38%|███▊ | 11430/30000 [7:19:09<11:53:56, 2.31s/it] 38%|███▊ | 11431/30000 [7:19:12<11:54:33, 2.31s/it] 38%|███▊ | 11432/30000 [7:19:14<11:53:36, 2.31s/it] 38%|███▊ | 11433/30000 [7:19:16<11:51:45, 2.30s/it] 38%|███▊ | 11434/30000 [7:19:18<11:48:35, 2.29s/it] 38%|███▊ | 11435/30000 [7:19:21<11:56:11, 2.31s/it] 38%|███▊ | 11436/30000 [7:19:23<11:59:04, 2.32s/it] 38%|███▊ | 11437/30000 [7:19:25<11:54:50, 2.31s/it] 38%|███▊ | 11438/30000 [7:19:28<11:52:12, 2.30s/it] 38%|███▊ | 11439/30000 [7:19:30<11:55:00, 2.31s/it] 38%|███▊ | 11440/30000 [7:19:32<11:47:16, 2.29s/it] 38%|███▊ | 11440/30000 [7:19:32<11:47:16, 2.29s/it] 38%|███▊ | 11441/30000 [7:19:35<11:46:56, 2.29s/it] 38%|███▊ | 11442/30000 [7:19:37<11:51:37, 2.30s/it] 38%|███▊ | 11443/30000 [7:19:39<11:50:35, 2.30s/it] 38%|███▊ | 11444/30000 [7:19:42<11:49:02, 2.29s/it] 38%|███▊ | 11445/30000 [7:19:44<11:48:42, 2.29s/it] 38%|███▊ | 11446/30000 [7:19:46<11:55:50, 2.31s/it] 38%|███▊ | 11447/30000 [7:19:48<11:53:56, 2.31s/it] 38%|███▊ | 11448/30000 [7:19:51<11:50:58, 2.30s/it] 38%|███▊ | 11449/30000 [7:19:53<11:50:06, 2.30s/it] 38%|███▊ | 11450/30000 [7:19:55<11:48:31, 2.29s/it] 38%|███▊ | 11450/30000 [7:19:55<11:48:31, 2.29s/it] 38%|███▊ | 11451/30000 [7:19:58<11:46:27, 2.29s/it] 38%|███▊ | 11452/30000 [7:20:00<11:43:59, 2.28s/it] 38%|███▊ | 11453/30000 [7:20:02<11:51:52, 2.30s/it] 38%|███▊ | 11454/30000 [7:20:05<11:55:48, 2.32s/it] 38%|███▊ | 11455/30000 [7:20:07<11:52:25, 2.30s/it] 38%|███▊ | 11456/30000 [7:20:09<11:49:05, 2.29s/it] 38%|███▊ | 11457/30000 [7:20:11<11:52:43, 2.31s/it] 38%|███▊ | 11458/30000 [7:20:14<11:49:52, 2.30s/it] 38%|███▊ | 11459/30000 [7:20:16<11:49:07, 2.29s/it] 38%|███▊ | 11460/30000 [7:20:18<11:55:06, 2.31s/it] 38%|███▊ | 11460/30000 [7:20:18<11:55:06, 2.31s/it] 38%|███▊ | 11461/30000 [7:20:21<11:58:00, 2.32s/it] 38%|███▊ | 11462/30000 [7:20:23<11:52:25, 2.31s/it] 38%|███▊ | 11463/30000 [7:20:25<11:46:43, 2.29s/it] 38%|███▊ | 11464/30000 [7:20:28<11:50:26, 2.30s/it] 38%|███▊ | 11465/30000 [7:20:30<11:50:45, 2.30s/it] 38%|███▊ | 11466/30000 [7:20:32<11:46:19, 2.29s/it] 38%|███▊ | 11467/30000 [7:20:34<11:45:54, 2.29s/it] 38%|███▊ | 11468/30000 [7:20:37<11:54:28, 2.31s/it] 38%|███▊ | 11469/30000 [7:20:39<11:57:40, 2.32s/it] 38%|███▊ | 11470/30000 [7:20:41<11:52:39, 2.31s/it] 38%|███▊ | 11470/30000 [7:20:41<11:52:39, 2.31s/it] 38%|███▊ | 11471/30000 [7:20:44<11:52:11, 2.31s/it] 38%|███▊ | 11472/30000 [7:20:46<11:52:55, 2.31s/it] 38%|███▊ | 11473/30000 [7:20:48<11:48:30, 2.29s/it] 38%|███▊ | 11474/30000 [7:20:50<11:43:23, 2.28s/it] 38%|███▊ | 11475/30000 [7:20:53<11:50:19, 2.30s/it] 38%|███▊ | 11476/30000 [7:20:55<11:52:26, 2.31s/it] 38%|███▊ | 11477/30000 [7:20:57<11:50:50, 2.30s/it] 38%|███▊ | 11478/30000 [7:21:00<11:48:24, 2.29s/it] 38%|███▊ | 11479/30000 [7:21:02<11:53:52, 2.31s/it] 38%|███▊ | 11480/30000 [7:21:04<11:58:31, 2.33s/it] 38%|███▊ | 11480/30000 [7:21:04<11:58:31, 2.33s/it] 38%|███▊ | 11481/30000 [7:21:07<11:53:17, 2.31s/it] 38%|███▊ | 11482/30000 [7:21:09<11:51:04, 2.30s/it] 38%|███▊ | 11483/30000 [7:21:11<11:54:36, 2.32s/it] 38%|███▊ | 11484/30000 [7:21:14<11:54:01, 2.31s/it] 38%|███▊ | 11485/30000 [7:21:16<11:51:06, 2.30s/it] 38%|███▊ | 11486/30000 [7:21:18<11:47:25, 2.29s/it] 38%|███▊ | 11487/30000 [7:21:21<11:51:41, 2.31s/it] 38%|███▊ | 11488/30000 [7:21:23<11:50:58, 2.30s/it] 38%|███▊ | 11489/30000 [7:21:25<11:52:22, 2.31s/it] 38%|███▊ | 11490/30000 [7:21:27<11:48:39, 2.30s/it] 38%|███▊ | 11490/30000 [7:21:27<11:48:39, 2.30s/it] 38%|███▊ | 11491/30000 [7:21:30<11:44:09, 2.28s/it] 38%|███▊ | 11492/30000 [7:21:32<11:54:27, 2.32s/it] 38%|███▊ | 11493/30000 [7:21:34<11:56:45, 2.32s/it] 38%|███▊ | 11494/30000 [7:21:37<11:53:49, 2.31s/it] 38%|███▊ | 11495/30000 [7:21:39<11:48:33, 2.30s/it] 38%|███▊ | 11496/30000 [7:21:41<11:54:05, 2.32s/it] 38%|███▊ | 11497/30000 [7:21:44<11:47:51, 2.30s/it] 38%|███▊ | 11498/30000 [7:21:46<11:46:42, 2.29s/it] 38%|███▊ | 11499/30000 [7:21:48<11:53:51, 2.32s/it] 38%|███▊ | 11500/30000 [7:21:51<11:53:40, 2.31s/it] 38%|███▊ | 11500/30000 [7:21:51<11:53:40, 2.31s/it] 38%|███▊ | 11501/30000 [7:21:53<11:51:39, 2.31s/it] 38%|███▊ | 11502/30000 [7:21:55<11:52:15, 2.31s/it] 38%|███▊ | 11503/30000 [7:21:58<11:56:08, 2.32s/it] 38%|███▊ | 11504/30000 [7:22:00<11:51:15, 2.31s/it] 38%|███▊ | 11505/30000 [7:22:02<11:48:54, 2.30s/it] 38%|███▊ | 11506/30000 [7:22:04<11:51:18, 2.31s/it] 38%|███▊ | 11507/30000 [7:22:07<11:50:33, 2.31s/it] 38%|███▊ | 11508/30000 [7:22:09<11:49:51, 2.30s/it] 38%|███▊ | 11509/30000 [7:22:11<11:51:01, 2.31s/it] 38%|███▊ | 11510/30000 [7:22:14<11:55:16, 2.32s/it] 38%|███▊ | 11510/30000 [7:22:14<11:55:16, 2.32s/it] 38%|███▊ | 11511/30000 [7:22:16<11:52:10, 2.31s/it] 38%|███▊ | 11512/30000 [7:22:18<11:46:59, 2.29s/it] 38%|███▊ | 11513/30000 [7:22:20<11:46:03, 2.29s/it] 38%|███▊ | 11514/30000 [7:22:23<11:49:59, 2.30s/it] 38%|███▊ | 11515/30000 [7:22:25<11:50:02, 2.30s/it] 38%|███▊ | 11516/30000 [7:22:27<11:44:44, 2.29s/it] 38%|███▊ | 11517/30000 [7:22:30<11:43:14, 2.28s/it] 38%|███▊ | 11518/30000 [7:22:32<11:50:04, 2.31s/it] 38%|███▊ | 11519/30000 [7:22:34<11:52:42, 2.31s/it] 38%|███▊ | 11520/30000 [7:22:37<11:46:37, 2.29s/it] 38%|███▊ | 11520/30000 [7:22:37<11:46:37, 2.29s/it] 38%|███▊ | 11521/30000 [7:22:39<11:48:15, 2.30s/it] 38%|███▊ | 11522/30000 [7:22:41<11:48:00, 2.30s/it] 38%|███▊ | 11523/30000 [7:22:43<11:44:56, 2.29s/it] 38%|███▊ | 11524/30000 [7:22:46<11:43:59, 2.29s/it] 38%|███▊ | 11525/30000 [7:22:48<11:49:46, 2.31s/it] 38%|███▊ | 11526/30000 [7:22:50<11:51:48, 2.31s/it] 38%|███▊ | 11527/30000 [7:22:53<11:49:18, 2.30s/it] 38%|███▊ | 11528/30000 [7:22:55<11:45:08, 2.29s/it] 38%|███▊ | 11529/30000 [7:22:57<11:51:14, 2.31s/it] 38%|███▊ | 11530/30000 [7:23:00<11:52:09, 2.31s/it] 38%|███▊ | 11530/30000 [7:23:00<11:52:09, 2.31s/it] 38%|███▊ | 11531/30000 [7:23:02<11:50:01, 2.31s/it] 38%|███▊ | 11532/30000 [7:23:04<11:50:00, 2.31s/it] 38%|███▊ | 11533/30000 [7:23:07<11:56:38, 2.33s/it] 38%|███▊ | 11534/30000 [7:23:09<11:59:34, 2.34s/it] 38%|███▊ | 11535/30000 [7:23:11<11:50:53, 2.31s/it] 38%|███▊ | 11536/30000 [7:23:13<11:48:35, 2.30s/it] 38%|███▊ | 11537/30000 [7:23:16<11:52:26, 2.32s/it] 38%|███▊ | 11538/30000 [7:23:18<11:45:32, 2.29s/it] 38%|███▊ | 11539/30000 [7:23:20<11:42:53, 2.28s/it] 38%|███▊ | 11540/30000 [7:23:23<11:44:06, 2.29s/it] 38%|███▊ | 11540/30000 [7:23:23<11:44:06, 2.29s/it] 38%|███▊ | 11541/30000 [7:23:25<11:52:43, 2.32s/it] 38%|███▊ | 11542/30000 [7:23:27<11:54:58, 2.32s/it] 38%|███▊ | 11543/30000 [7:23:30<11:48:09, 2.30s/it] 38%|███▊ | 11544/30000 [7:23:32<11:45:06, 2.29s/it] 38%|███▊ | 11545/30000 [7:23:34<11:45:22, 2.29s/it] 38%|███▊ | 11546/30000 [7:23:36<11:44:05, 2.29s/it] 38%|███▊ | 11547/30000 [7:23:39<11:43:41, 2.29s/it] 38%|███▊ | 11548/30000 [7:23:41<11:49:37, 2.31s/it] 38%|███▊ | 11549/30000 [7:23:43<11:49:10, 2.31s/it] 38%|███▊ | 11550/30000 [7:23:46<11:47:02, 2.30s/it] 38%|███▊ | 11550/30000 [7:23:46<11:47:02, 2.30s/it] 39%|███▊ | 11551/30000 [7:23:48<11:44:27, 2.29s/it] 39%|███▊ | 11552/30000 [7:23:50<11:49:50, 2.31s/it] 39%|███▊ | 11553/30000 [7:23:53<11:47:34, 2.30s/it] 39%|███▊ | 11554/30000 [7:23:55<11:46:36, 2.30s/it] 39%|███▊ | 11555/30000 [7:23:57<11:40:07, 2.28s/it] 39%|███▊ | 11556/30000 [7:24:00<11:50:24, 2.31s/it] 39%|███▊ | 11557/30000 [7:24:02<11:51:11, 2.31s/it] 39%|███▊ | 11558/30000 [7:24:04<11:49:21, 2.31s/it] 39%|███▊ | 11559/30000 [7:24:06<11:42:35, 2.29s/it] 39%|███▊ | 11560/30000 [7:24:09<11:48:55, 2.31s/it] 39%|███▊ | 11560/30000 [7:24:09<11:48:55, 2.31s/it] 39%|███▊ | 11561/30000 [7:24:11<11:43:38, 2.29s/it] 39%|███▊ | 11562/30000 [7:24:13<11:42:01, 2.28s/it] 39%|███▊ | 11563/30000 [7:24:16<11:43:16, 2.29s/it] 39%|███▊ | 11564/30000 [7:24:18<11:46:09, 2.30s/it] 39%|███▊ | 11565/30000 [7:24:20<11:44:01, 2.29s/it] 39%|███▊ | 11566/30000 [7:24:22<11:41:32, 2.28s/it] 39%|███▊ | 11567/30000 [7:24:25<11:41:06, 2.28s/it] 39%|███▊ | 11568/30000 [7:24:27<11:41:10, 2.28s/it] 39%|███▊ | 11569/30000 [7:24:29<11:41:16, 2.28s/it] 39%|███▊ | 11570/30000 [7:24:32<11:41:58, 2.29s/it] 39%|███▊ | 11570/30000 [7:24:32<11:41:58, 2.29s/it] 39%|███▊ | 11571/30000 [7:24:34<11:48:32, 2.31s/it] 39%|███▊ | 11572/30000 [7:24:36<11:52:59, 2.32s/it] 39%|███▊ | 11573/30000 [7:24:39<11:53:50, 2.32s/it] 39%|███▊ | 11574/30000 [7:24:41<11:49:55, 2.31s/it] 39%|███▊ | 11575/30000 [7:24:43<11:46:10, 2.30s/it] 39%|███▊ | 11576/30000 [7:24:45<11:49:16, 2.31s/it] 39%|███▊ | 11577/30000 [7:24:48<11:50:02, 2.31s/it] 39%|███▊ | 11578/30000 [7:24:50<11:45:30, 2.30s/it] 39%|███▊ | 11579/30000 [7:24:52<11:45:26, 2.30s/it] 39%|███▊ | 11580/30000 [7:24:55<11:47:55, 2.31s/it] 39%|███▊ | 11580/30000 [7:24:55<11:47:55, 2.31s/it] 39%|███▊ | 11581/30000 [7:24:57<11:43:49, 2.29s/it] 39%|███▊ | 11582/30000 [7:24:59<11:42:28, 2.29s/it] 39%|███▊ | 11583/30000 [7:25:02<11:44:46, 2.30s/it] 39%|███▊ | 11584/30000 [7:25:04<11:43:36, 2.29s/it] 39%|███▊ | 11585/30000 [7:25:06<11:43:38, 2.29s/it] 39%|███▊ | 11586/30000 [7:25:08<11:45:49, 2.30s/it] 39%|███▊ | 11587/30000 [7:25:11<11:50:46, 2.32s/it] 39%|███▊ | 11588/30000 [7:25:13<11:52:22, 2.32s/it] 39%|███▊ | 11589/30000 [7:25:15<11:48:07, 2.31s/it] 39%|███▊ | 11590/30000 [7:25:18<11:46:55, 2.30s/it] 39%|███▊ | 11590/30000 [7:25:18<11:46:55, 2.30s/it] 39%|███▊ | 11591/30000 [7:25:20<11:52:13, 2.32s/it] 39%|███▊ | 11592/30000 [7:25:22<11:52:31, 2.32s/it] 39%|███▊ | 11593/30000 [7:25:25<11:45:58, 2.30s/it] 39%|███▊ | 11594/30000 [7:25:27<11:42:08, 2.29s/it] 39%|███▊ | 11595/30000 [7:25:29<11:44:44, 2.30s/it] 39%|███▊ | 11596/30000 [7:25:31<11:40:11, 2.28s/it] 39%|███▊ | 11597/30000 [7:25:34<11:41:56, 2.29s/it] 39%|███▊ | 11598/30000 [7:25:36<11:48:55, 2.31s/it] 39%|███▊ | 11599/30000 [7:25:38<11:49:19, 2.31s/it] 39%|███▊ | 11600/30000 [7:25:41<11:44:29, 2.30s/it] 39%|███▊ | 11600/30000 [7:25:41<11:44:29, 2.30s/it] 39%|███▊ | 11601/30000 [7:25:43<11:42:09, 2.29s/it] 39%|███▊ | 11602/30000 [7:25:45<11:46:13, 2.30s/it] 39%|███▊ | 11603/30000 [7:25:48<11:50:13, 2.32s/it] 39%|███▊ | 11604/30000 [7:25:50<11:48:29, 2.31s/it] 39%|███▊ | 11605/30000 [7:25:52<11:46:12, 2.30s/it] 39%|███▊ | 11606/30000 [7:25:55<11:48:39, 2.31s/it] 39%|███▊ | 11607/30000 [7:25:57<11:41:55, 2.29s/it] 39%|███▊ | 11608/30000 [7:25:59<11:37:05, 2.27s/it] 39%|███▊ | 11609/30000 [7:26:01<11:43:28, 2.30s/it] 39%|███▊ | 11610/30000 [7:26:04<11:46:30, 2.31s/it] 39%|███▊ | 11610/30000 [7:26:04<11:46:30, 2.31s/it] 39%|███▊ | 11611/30000 [7:26:06<11:43:21, 2.29s/it] 39%|███▊ | 11612/30000 [7:26:08<11:41:39, 2.29s/it] 39%|███▊ | 11613/30000 [7:26:11<11:46:00, 2.30s/it] 39%|███▊ | 11614/30000 [7:26:13<11:42:52, 2.29s/it] 39%|███▊ | 11615/30000 [7:26:15<11:41:20, 2.29s/it] 39%|███▊ | 11616/30000 [7:26:17<11:46:26, 2.31s/it] 39%|███▊ | 11617/30000 [7:26:20<11:46:25, 2.31s/it] 39%|███▊ | 11618/30000 [7:26:22<11:43:28, 2.30s/it] 39%|███▊ | 11619/30000 [7:26:24<11:40:10, 2.29s/it] 39%|███▊ | 11620/30000 [7:26:27<11:46:21, 2.31s/it] 39%|███▊ | 11620/30000 [7:26:27<11:46:21, 2.31s/it] 39%|███▊ | 11621/30000 [7:26:29<11:45:20, 2.30s/it] 39%|███▊ | 11622/30000 [7:26:31<11:42:39, 2.29s/it] 39%|███▊ | 11623/30000 [7:26:34<11:48:06, 2.31s/it] 39%|███▊ | 11624/30000 [7:26:36<11:47:16, 2.31s/it] 39%|███▉ | 11625/30000 [7:26:38<11:46:54, 2.31s/it] 39%|███▉ | 11626/30000 [7:26:40<11:44:23, 2.30s/it] 39%|███▉ | 11627/30000 [7:26:43<11:47:59, 2.31s/it] 39%|███▉ | 11628/30000 [7:26:45<11:46:45, 2.31s/it] 39%|███▉ | 11629/30000 [7:26:47<11:47:55, 2.31s/it] 39%|███▉ | 11630/30000 [7:26:50<11:45:55, 2.31s/it] +{'loss': 0.0152, 'grad_norm': 0.06262461096048355, 'learning_rate': 2.216972958968801e-05, 'flow_matching_loss': 0.015199510287493468, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.27} +{'loss': 0.0149, 'grad_norm': 0.07628375291824341, 'learning_rate': 2.215520165970743e-05, 'flow_matching_loss': 0.014879797771573066, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.27} +{'loss': 0.0137, 'grad_norm': 0.08823966234922409, 'learning_rate': 2.2140665035482143e-05, 'flow_matching_loss': 0.013696678634732961, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.27} +{'loss': 0.0153, 'grad_norm': 0.07645256817340851, 'learning_rate': 2.2126119734675518e-05, 'flow_matching_loss': 0.01529853567481041, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.28} +{'loss': 0.0146, 'grad_norm': 0.09255778044462204, 'learning_rate': 2.2111565774961463e-05, 'flow_matching_loss': 0.014607799891382455, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.28} +{'loss': 0.0152, 'grad_norm': 0.06922861188650131, 'learning_rate': 2.2097003174024417e-05, 'flow_matching_loss': 0.015212753508239984, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.28} +{'loss': 0.0153, 'grad_norm': 0.07110816985368729, 'learning_rate': 2.2082431949559315e-05, 'flow_matching_loss': 0.01532007846981287, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.28} +{'loss': 0.0154, 'grad_norm': 0.09647352993488312, 'learning_rate': 2.2067852119271558e-05, 'flow_matching_loss': 0.015359319094568492, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.29} +{'loss': 0.016, 'grad_norm': 0.06821310520172119, 'learning_rate': 2.2053263700877026e-05, 'flow_matching_loss': 0.016029846854507923, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.29} +{'loss': 0.016, 'grad_norm': 0.0653790608048439, 'learning_rate': 2.2038666712102015e-05, 'flow_matching_loss': 0.01601950218901038, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.29} +{'loss': 0.016, 'grad_norm': 0.07359831780195236, 'learning_rate': 2.2024061170683246e-05, 'flow_matching_loss': 0.01600588848814368, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.3} +{'loss': 0.0169, 'grad_norm': 0.08427239954471588, 'learning_rate': 2.200944709436783e-05, 'flow_matching_loss': 0.016894906386733057, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.3} +{'loss': 0.0162, 'grad_norm': 0.10443778336048126, 'learning_rate': 2.1994824500913246e-05, 'flow_matching_loss': 0.01615295372903347, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.3} +{'loss': 0.0171, 'grad_norm': 0.07021120935678482, 'learning_rate': 2.1980193408087322e-05, 'flow_matching_loss': 0.017092229519039392, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.31} +{'loss': 0.0159, 'grad_norm': 0.08641776442527771, 'learning_rate': 2.1965553833668224e-05, 'flow_matching_loss': 0.015939043182879687, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.31} +{'loss': 0.0155, 'grad_norm': 0.06871704757213593, 'learning_rate': 2.195090579544441e-05, 'flow_matching_loss': 0.015527061093598605, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.31} +{'loss': 0.0153, 'grad_norm': 0.09730838239192963, 'learning_rate': 2.193624931121462e-05, 'flow_matching_loss': 0.015334117412567138, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.31} +{'loss': 0.0149, 'grad_norm': 0.0696071907877922, 'learning_rate': 2.1921584398787872e-05, 'flow_matching_loss': 0.014886845368891954, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.32} +{'loss': 0.0148, 'grad_norm': 0.06843945384025574, 'learning_rate': 2.1906911075983418e-05, 'flow_matching_loss': 0.014826918579638004, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.32} +{'loss': 0.0132, 'grad_norm': 0.08574014157056808, 'learning_rate': 2.1892229360630728e-05, 'flow_matching_loss': 0.013193984609097242, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.32} +{'loss': 0.0149, 'grad_norm': 0.08976937085390091, 'learning_rate': 2.187753927056947e-05, 'flow_matching_loss': 0.01494209235534072, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.33} +{'loss': 0.0166, 'grad_norm': 0.11180431395769119, 'learning_rate': 2.186284082364949e-05, 'flow_matching_loss': 0.01659237714484334, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.33} +{'loss': 0.0157, 'grad_norm': 0.0947609543800354, 'learning_rate': 2.1848134037730786e-05, 'flow_matching_loss': 0.015749915968626738, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.33} +{'loss': 0.0154, 'grad_norm': 0.09999451041221619, 'learning_rate': 2.183341893068349e-05, 'flow_matching_loss': 0.015376142505556346, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.33} +{'loss': 0.0155, 'grad_norm': 0.05818157643079758, 'learning_rate': 2.1818695520387842e-05, 'flow_matching_loss': 0.015470939502120019, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.34} +{'loss': 0.0136, 'grad_norm': 0.09030024707317352, 'learning_rate': 2.1803963824734173e-05, 'flow_matching_loss': 0.013636673521250486, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.34} +{'loss': 0.015, 'grad_norm': 0.13053344190120697, 'learning_rate': 2.1789223861622887e-05, 'flow_matching_loss': 0.015031356085091829, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.34} +{'loss': 0.0157, 'grad_norm': 0.11050936579704285, 'learning_rate': 2.1774475648964428e-05, 'flow_matching_loss': 0.01568855345249176, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.35} +{'loss': 0.0152, 'grad_norm': 0.11162535846233368, 'learning_rate': 2.1759719204679267e-05, 'flow_matching_loss': 0.015165898017585278, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.35} +{'loss': 0.016, 'grad_norm': 0.06844819337129593, 'learning_rate': 2.174495454669787e-05, 'flow_matching_loss': 0.01603771522641182, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.35} +{'loss': 0.0175, 'grad_norm': 0.0929417759180069, 'learning_rate': 2.1730181692960692e-05, 'flow_matching_loss': 0.01752981897443533, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.35} +{'loss': 0.0144, 'grad_norm': 0.10741381347179413, 'learning_rate': 2.171540066141814e-05, 'flow_matching_loss': 0.014415863528847694, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.36} +{'loss': 0.0161, 'grad_norm': 0.10457614064216614, 'learning_rate': 2.170061147003056e-05, 'flow_matching_loss': 0.016136183124035596, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.36} +{'loss': 0.0174, 'grad_norm': 0.08770859241485596, 'learning_rate': 2.1685814136768225e-05, 'flow_matching_loss': 0.017379524558782576, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.36} +{'loss': 0.0143, 'grad_norm': 0.0676586702466011, 'learning_rate': 2.167100867961128e-05, 'flow_matching_loss': 0.014251954201608897, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.37} +{'loss': 0.0175, 'grad_norm': 0.07541871070861816, 'learning_rate': 2.1656195116549756e-05, 'flow_matching_loss': 0.01751046357676387, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.37} +{'loss': 0.0152, 'grad_norm': 0.07343531399965286, 'learning_rate': 2.1641373465583527e-05, 'flow_matching_loss': 0.015153719484806061, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.37} +{'loss': 0.0151, 'grad_norm': 0.07062983512878418, 'learning_rate': 2.1626543744722292e-05, 'flow_matching_loss': 0.015052888542413712, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.38} +{'loss': 0.0142, 'grad_norm': 0.07133804261684418, 'learning_rate': 2.161170597198557e-05, 'flow_matching_loss': 0.014208865258842707, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.38} +{'loss': 0.0147, 'grad_norm': 0.0738903284072876, 'learning_rate': 2.1596860165402645e-05, 'flow_matching_loss': 0.014730602409690619, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.38} +{'loss': 0.0161, 'grad_norm': 0.0890546441078186, 'learning_rate': 2.1582006343012576e-05, 'flow_matching_loss': 0.016088690049946307, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.38} 39%|███▉ | 11630/30000 [7:26:50<11:45:55, 2.31s/it] 39%|███▉ | 11631/30000 [7:26:52<11:48:14, 2.31s/it] 39%|███▉ | 11632/30000 [7:26:54<11:41:29, 2.29s/it] 39%|███▉ | 11633/30000 [7:26:57<11:38:25, 2.28s/it] 39%|███▉ | 11634/30000 [7:26:59<11:48:41, 2.32s/it] 39%|███▉ | 11635/30000 [7:27:01<11:48:46, 2.32s/it] 39%|███▉ | 11636/30000 [7:27:04<11:46:39, 2.31s/it] 39%|███▉ | 11637/30000 [7:27:06<11:45:44, 2.31s/it] 39%|███▉ | 11638/30000 [7:27:08<11:48:02, 2.31s/it] 39%|███▉ | 11639/30000 [7:27:10<11:42:30, 2.30s/it] 39%|███▉ | 11640/30000 [7:27:13<11:40:32, 2.29s/it] 39%|███▉ | 11640/30000 [7:27:13<11:40:32, 2.29s/it] 39%|███▉ | 11641/30000 [7:27:15<11:47:55, 2.31s/it] 39%|███▉ | 11642/30000 [7:27:17<11:51:20, 2.32s/it] 39%|███▉ | 11643/30000 [7:27:20<11:47:53, 2.31s/it] 39%|███▉ | 11644/30000 [7:27:22<11:46:08, 2.31s/it] 39%|███▉ | 11645/30000 [7:27:24<11:50:48, 2.32s/it] 39%|███▉ | 11646/30000 [7:27:27<11:52:34, 2.33s/it] 39%|███▉ | 11647/30000 [7:27:29<11:46:31, 2.31s/it] 39%|███▉ | 11648/30000 [7:27:31<11:43:36, 2.30s/it] 39%|███▉ | 11649/30000 [7:27:34<11:45:46, 2.31s/it] 39%|███▉ | 11650/30000 [7:27:36<11:40:02, 2.29s/it] 39%|███▉ | 11650/30000 [7:27:36<11:40:02, 2.29s/it] 39%|███▉ | 11651/30000 [7:27:38<11:38:44, 2.28s/it] 39%|███▉ | 11652/30000 [7:27:40<11:45:50, 2.31s/it] 39%|███▉ | 11653/30000 [7:27:43<11:48:14, 2.32s/it] 39%|███▉ | 11654/30000 [7:27:45<11:43:37, 2.30s/it] 39%|███▉ | 11655/30000 [7:27:47<11:46:18, 2.31s/it] 39%|███▉ | 11656/30000 [7:27:50<11:43:55, 2.30s/it] 39%|███▉ | 11657/30000 [7:27:52<11:40:17, 2.29s/it] 39%|███▉ | 11658/30000 [7:27:54<11:37:19, 2.28s/it] 39%|███▉ | 11659/30000 [7:27:57<11:40:51, 2.29s/it] 39%|███▉ | 11660/30000 [7:27:59<11:38:53, 2.29s/it] 39%|███▉ | 11660/30000 [7:27:59<11:38:53, 2.29s/it] 39%|███▉ | 11661/30000 [7:28:01<11:39:25, 2.29s/it] 39%|███▉ | 11662/30000 [7:28:03<11:47:37, 2.32s/it] 39%|███▉ | 11663/30000 [7:28:06<11:45:39, 2.31s/it] 39%|███▉ | 11664/30000 [7:28:08<11:42:46, 2.30s/it] 39%|███▉ | 11665/30000 [7:28:10<11:41:27, 2.30s/it] 39%|███▉ | 11666/30000 [7:28:13<11:39:58, 2.29s/it] 39%|███▉ | 11667/30000 [7:28:15<11:35:50, 2.28s/it] 39%|███▉ | 11668/30000 [7:28:17<11:38:55, 2.29s/it] 39%|███▉ | 11669/30000 [7:28:19<11:41:29, 2.30s/it] 39%|███▉ | 11670/30000 [7:28:22<11:38:47, 2.29s/it] 39%|███▉ | 11670/30000 [7:28:22<11:38:47, 2.29s/it] 39%|███▉ | 11671/30000 [7:28:24<11:33:13, 2.27s/it] 39%|███▉ | 11672/30000 [7:28:26<11:38:20, 2.29s/it] 39%|███▉ | 11673/30000 [7:28:29<11:35:27, 2.28s/it] 39%|███▉ | 11674/30000 [7:28:31<11:35:28, 2.28s/it] 39%|███▉ | 11675/30000 [7:28:33<11:42:59, 2.30s/it] 39%|███▉ | 11676/30000 [7:28:36<11:45:39, 2.31s/it] 39%|███▉ | 11677/30000 [7:28:38<11:43:27, 2.30s/it] 39%|███▉ | 11678/30000 [7:28:40<11:48:54, 2.32s/it] 39%|███▉ | 11679/30000 [7:28:42<11:47:25, 2.32s/it] 39%|███▉ | 11680/30000 [7:28:45<11:41:45, 2.30s/it] 39%|███▉ | 11680/30000 [7:28:45<11:41:45, 2.30s/it] 39%|███▉ | 11681/30000 [7:28:47<11:38:20, 2.29s/it] 39%|███▉ | 11682/30000 [7:28:49<11:42:12, 2.30s/it] 39%|███▉ | 11683/30000 [7:28:52<11:38:40, 2.29s/it] 39%|███▉ | 11684/30000 [7:28:54<11:37:26, 2.28s/it] 39%|███▉ | 11685/30000 [7:28:56<11:45:07, 2.31s/it] 39%|███▉ | 11686/30000 [7:28:59<11:51:09, 2.33s/it] 39%|███▉ | 11687/30000 [7:29:01<11:45:56, 2.31s/it] 39%|███▉ | 11688/30000 [7:29:03<11:46:53, 2.32s/it] 39%|███▉ | 11689/30000 [7:29:06<11:45:01, 2.31s/it] 39%|███▉ | 11690/30000 [7:29:08<11:44:50, 2.31s/it] 39%|███▉ | 11690/30000 [7:29:08<11:44:50, 2.31s/it] 39%|███▉ | 11691/30000 [7:29:10<11:43:02, 2.30s/it] 39%|███▉ | 11692/30000 [7:29:12<11:46:53, 2.32s/it] 39%|███▉ | 11693/30000 [7:29:15<11:49:15, 2.32s/it] 39%|███▉ | 11694/30000 [7:29:17<11:45:45, 2.31s/it] 39%|███▉ | 11695/30000 [7:29:19<11:48:10, 2.32s/it] 39%|███▉ | 11696/30000 [7:29:22<11:42:55, 2.30s/it] 39%|███▉ | 11697/30000 [7:29:24<11:39:19, 2.29s/it] 39%|███▉ | 11698/30000 [7:29:26<11:40:05, 2.30s/it] 39%|███▉ | 11699/30000 [7:29:29<11:39:36, 2.29s/it] 39%|███▉ | 11700/30000 [7:29:31<11:36:21, 2.28s/it] 39%|███▉ | 11700/30000 [7:29:31<11:36:21, 2.28s/it] 39%|███▉ | 11701/30000 [7:29:33<11:36:26, 2.28s/it] 39%|███▉ | 11702/30000 [7:29:35<11:44:12, 2.31s/it] 39%|███▉ | 11703/30000 [7:29:38<11:45:10, 2.31s/it] 39%|███▉ | 11704/30000 [7:29:40<11:37:54, 2.29s/it] 39%|███▉ | 11705/30000 [7:29:42<11:39:29, 2.29s/it] 39%|███▉ | 11706/30000 [7:29:45<11:41:31, 2.30s/it] 39%|███▉ | 11707/30000 [7:29:47<11:39:48, 2.30s/it] 39%|███▉ | 11708/30000 [7:29:49<11:45:05, 2.31s/it] 39%|███▉ | 11709/30000 [7:29:52<11:49:05, 2.33s/it] 39%|███▉ | 11710/30000 [7:29:54<11:41:55, 2.30s/it] 39%|███▉ | 11710/30000 [7:29:54<11:41:55, 2.30s/it] 39%|███▉ | 11711/30000 [7:29:56<11:35:08, 2.28s/it] 39%|███▉ | 11712/30000 [7:29:58<11:36:07, 2.28s/it] 39%|███▉ | 11713/30000 [7:30:01<11:34:20, 2.28s/it] 39%|███▉ | 11714/30000 [7:30:03<11:33:57, 2.28s/it] 39%|███▉ | 11715/30000 [7:30:05<11:41:23, 2.30s/it] 39%|███▉ | 11716/30000 [7:30:08<11:46:29, 2.32s/it] 39%|███▉ | 11717/30000 [7:30:10<11:44:21, 2.31s/it] 39%|███▉ | 11718/30000 [7:30:12<11:45:39, 2.32s/it] 39%|███▉ | 11719/30000 [7:30:15<11:44:26, 2.31s/it] 39%|███▉ | 11720/30000 [7:30:17<11:37:46, 2.29s/it] 39%|███▉ | 11720/30000 [7:30:17<11:37:46, 2.29s/it] 39%|███▉ | 11721/30000 [7:30:19<11:40:09, 2.30s/it] 39%|███▉ | 11722/30000 [7:30:21<11:39:09, 2.30s/it] 39%|███▉ | 11723/30000 [7:30:24<11:38:07, 2.29s/it] 39%|███▉ | 11724/30000 [7:30:26<11:36:50, 2.29s/it] 39%|███▉ | 11725/30000 [7:30:28<11:44:18, 2.31s/it] 39%|███▉ | 11726/30000 [7:30:31<11:42:24, 2.31s/it] 39%|███▉ | 11727/30000 [7:30:33<11:38:45, 2.29s/it] 39%|███▉ | 11728/30000 [7:30:35<11:40:57, 2.30s/it] 39%|███▉ | 11729/30000 [7:30:38<11:39:36, 2.30s/it] 39%|███▉ | 11730/30000 [7:30:40<11:35:18, 2.28s/it] 39%|███▉ | 11730/30000 [7:30:40<11:35:18, 2.28s/it] 39%|███▉ | 11731/30000 [7:30:42<11:32:40, 2.27s/it] 39%|███▉ | 11732/30000 [7:30:44<11:38:27, 2.29s/it] 39%|███▉ | 11733/30000 [7:30:47<11:41:42, 2.30s/it] 39%|███▉ | 11734/30000 [7:30:49<11:39:43, 2.30s/it] 39%|███▉ | 11735/30000 [7:30:51<11:38:20, 2.29s/it] 39%|███▉ | 11736/30000 [7:30:54<11:37:25, 2.29s/it] 39%|███▉ | 11737/30000 [7:30:56<11:35:16, 2.28s/it] 39%|███▉ | 11738/30000 [7:30:58<11:32:48, 2.28s/it] 39%|███▉ | 11739/30000 [7:31:00<11:35:48, 2.29s/it] 39%|███▉ | 11740/30000 [7:31:03<11:39:21, 2.30s/it] 39%|███▉ | 11740/30000 [7:31:03<11:39:21, 2.30s/it] 39%|███▉ | 11741/30000 [7:31:05<11:39:00, 2.30s/it] 39%|███▉ | 11742/30000 [7:31:07<11:41:01, 2.30s/it] 39%|███▉ | 11743/30000 [7:31:10<11:38:14, 2.29s/it] 39%|███▉ | 11744/30000 [7:31:12<11:33:57, 2.28s/it] 39%|███▉ | 11745/30000 [7:31:14<11:30:19, 2.27s/it] 39%|███▉ | 11746/30000 [7:31:16<11:37:51, 2.29s/it] 39%|███▉ | 11747/30000 [7:31:19<11:34:40, 2.28s/it] 39%|███▉ | 11748/30000 [7:31:21<11:33:36, 2.28s/it] 39%|███▉ | 11749/30000 [7:31:23<11:40:40, 2.30s/it] 39%|███▉ | 11750/30000 [7:31:26<11:38:33, 2.30s/it] 39%|███▉ | 11750/30000 [7:31:26<11:38:33, 2.30s/it] 39%|███▉ | 11751/30000 [7:31:28<11:37:53, 2.29s/it] 39%|███▉ | 11752/30000 [7:31:30<11:35:39, 2.29s/it] 39%|███▉ | 11753/30000 [7:31:32<11:38:09, 2.30s/it] 39%|███▉ | 11754/30000 [7:31:35<11:33:48, 2.28s/it] 39%|███▉ | 11755/30000 [7:31:37<11:31:43, 2.27s/it] 39%|███▉ | 11756/30000 [7:31:39<11:36:56, 2.29s/it] 39%|███▉ | 11757/30000 [7:31:42<11:37:31, 2.29s/it] 39%|███▉ | 11758/30000 [7:31:44<11:36:34, 2.29s/it] 39%|███▉ | 11759/30000 [7:31:46<11:34:55, 2.29s/it] 39%|███▉ | 11760/30000 [7:31:49<11:39:00, 2.30s/it] 39%|███▉ | 11760/30000 [7:31:49<11:39:00, 2.30s/it] 39%|███▉ | 11761/30000 [7:31:51<11:40:05, 2.30s/it] 39%|███▉ | 11762/30000 [7:31:53<11:37:41, 2.30s/it] 39%|███▉ | 11763/30000 [7:31:55<11:37:52, 2.30s/it] 39%|███▉ | 11764/30000 [7:31:58<11:39:44, 2.30s/it] 39%|███▉ | 11765/30000 [7:32:00<11:37:33, 2.30s/it] 39%|███▉ | 11766/30000 [7:32:02<11:37:19, 2.29s/it] 39%|███▉ | 11767/30000 [7:32:05<11:38:27, 2.30s/it] 39%|███▉ | 11768/30000 [7:32:07<11:37:49, 2.30s/it] 39%|███▉ | 11769/30000 [7:32:09<11:35:32, 2.29s/it] 39%|███▉ | 11770/30000 [7:32:11<11:33:15, 2.28s/it] 39%|███▉ | 11770/30000 [7:32:11<11:33:15, 2.28s/it] 39%|███▉ | 11771/30000 [7:32:14<11:40:39, 2.31s/it] 39%|███▉ | 11772/30000 [7:32:16<11:44:49, 2.32s/it] 39%|███▉ | 11773/30000 [7:32:18<11:40:04, 2.30s/it] 39%|███▉ | 11774/30000 [7:32:21<11:37:02, 2.29s/it] 39%|███▉ | 11775/30000 [7:32:23<11:39:44, 2.30s/it] 39%|███▉ | 11776/30000 [7:32:25<11:40:40, 2.31s/it] 39%|███▉ | 11777/30000 [7:32:28<11:38:01, 2.30s/it] 39%|███▉ | 11778/30000 [7:32:30<11:34:58, 2.29s/it] 39%|███▉ | 11779/30000 [7:32:32<11:40:26, 2.31s/it] 39%|███▉ | 11780/30000 [7:32:35<11:41:28, 2.31s/it] 39%|███▉ | 11780/30000 [7:32:35<11:41:28, 2.31s/it] 39%|███▉ | 11781/30000 [7:32:37<11:39:38, 2.30s/it] 39%|███▉ | 11782/30000 [7:32:39<11:36:18, 2.29s/it] 39%|███▉ | 11783/30000 [7:32:41<11:38:55, 2.30s/it] 39%|███▉ | 11784/30000 [7:32:44<11:40:49, 2.31s/it] 39%|███▉ | 11785/30000 [7:32:46<11:38:26, 2.30s/it] 39%|███▉ | 11786/30000 [7:32:48<11:33:01, 2.28s/it] 39%|███▉ | 11787/30000 [7:32:51<11:35:41, 2.29s/it] 39%|███▉ | 11788/30000 [7:32:53<11:37:12, 2.30s/it] 39%|███▉ | 11789/30000 [7:32:55<11:35:24, 2.29s/it] 39%|███▉ | 11790/30000 [7:32:57<11:36:31, 2.29s/it] 39%|███▉ | 11790/30000 [7:32:58<11:36:31, 2.29s/it] 39%|███▉ | 11791/30000 [7:33:00<11:40:17, 2.31s/it] 39%|███▉ | 11792/30000 [7:33:02<11:39:57, 2.31s/it] 39%|███▉ | 11793/30000 [7:33:04<11:33:26, 2.29s/it] 39%|███▉ | 11794/30000 [7:33:07<11:30:18, 2.27s/it] 39%|███▉ | 11795/30000 [7:33:09<11:34:29, 2.29s/it] 39%|███▉ | 11796/30000 [7:33:11<11:32:00, 2.28s/it] 39%|███▉ | 11797/30000 [7:33:13<11:31:25, 2.28s/it] 39%|███▉ | 11798/30000 [7:33:16<11:36:18, 2.30s/it] 39%|███▉ | 11799/30000 [7:33:18<11:34:16, 2.29s/it] 39%|███▉ | 11800/30000 [7:33:20<11:32:31, 2.28s/it] 39%|███▉ | 11800/30000 [7:33:20<11:32:31, 2.28s/it] 39%|███▉ | 11801/30000 [7:33:23<11:33:20, 2.29s/it] 39%|███▉ | 11802/30000 [7:33:25<11:38:23, 2.30s/it] 39%|███▉ | 11803/30000 [7:33:27<11:37:32, 2.30s/it] 39%|███▉ | 11804/30000 [7:33:30<11:39:06, 2.31s/it] 39%|███▉ | 11805/30000 [7:33:32<11:38:57, 2.30s/it] 39%|███▉ | 11806/30000 [7:33:34<11:43:50, 2.32s/it] 39%|███▉ | 11807/30000 [7:33:37<11:42:31, 2.32s/it] 39%|███▉ | 11808/30000 [7:33:39<11:34:27, 2.29s/it] 39%|███▉ | 11809/30000 [7:33:41<11:38:38, 2.30s/it] 39%|███▉ | 11810/30000 [7:33:43<11:39:40, 2.31s/it] 39%|███▉ | 11810/30000 [7:33:43<11:39:40, 2.31s/it] 39%|███▉ | 11811/30000 [7:33:46<11:34:52, 2.29s/it] 39%|███▉ | 11812/30000 [7:33:48<11:34:25, 2.29s/it] 39%|███▉ | 11813/30000 [7:33:50<11:41:12, 2.31s/it] 39%|███▉ | 11814/30000 [7:33:53<11:41:59, 2.32s/it] 39%|███▉ | 11815/30000 [7:33:55<11:37:09, 2.30s/it] 39%|███▉ | 11816/30000 [7:33:57<11:35:54, 2.30s/it] 39%|███▉ | 11817/30000 [7:34:00<11:36:13, 2.30s/it] 39%|███▉ | 11818/30000 [7:34:02<11:34:51, 2.29s/it] 39%|███▉ | 11819/30000 [7:34:04<11:33:01, 2.29s/it] 39%|███▉ | 11820/30000 [7:34:06<11:38:52, 2.31s/it] 39%|███▉ | 11820/30000 [7:34:06<11:38:52, 2.31s/it] 39%|███▉ | 11821/30000 [7:34:09<11:38:48, 2.31s/it] 39%|███▉ | 11822/30000 [7:34:11<11:36:26, 2.30s/it] 39%|███▉ | 11823/30000 [7:34:13<11:37:36, 2.30s/it] 39%|███▉ | 11824/30000 [7:34:16<11:36:37, 2.30s/it] 39%|███▉ | 11825/30000 [7:34:18<11:32:09, 2.28s/it] 39%|███▉ | 11826/30000 [7:34:20<11:32:22, 2.29s/it] 39%|███▉ | 11827/30000 [7:34:22<11:34:00, 2.29s/it] 39%|███▉ | 11828/30000 [7:34:25<11:35:05, 2.30s/it] 39%|███▉ | 11829/30000 [7:34:27<11:35:47, 2.30s/it] 39%|███▉ | 11830/30000 [7:34:29<11:41:30, 2.32s/it] 39%|███▉ | 11830/30000 [7:34:29<11:41:30, 2.32s/it] 39%|███▉ | 11831/30000 [7:34:32<11:39:44, 2.31s/it] 39%|███▉ | 11832/30000 [7:34:34<11:35:02, 2.30s/it] 39%|███▉ | 11833/30000 [7:34:36<11:36:41, 2.30s/it] 39%|███▉ | 11834/30000 [7:34:39<11:37:09, 2.30s/it] 39%|███▉ | 11835/30000 [7:34:41<11:35:06, 2.30s/it] 39%|███▉ | 11836/30000 [7:34:43<11:35:14, 2.30s/it] 39%|███▉ | 11837/30000 [7:34:46<11:40:24, 2.31s/it] 39%|███▉ | 11838/30000 [7:34:48<11:40:07, 2.31s/it] 39%|███▉ | 11839/30000 [7:34:50<11:35:58, 2.30s/it] 39%|███▉ | 11840/30000 [7:34:52<11:38:21, 2.31s/it] 39%|███▉ | 11840/30000 [7:34:52<11:38:21, 2.31s/it] 39%|███▉ | 11841/30000 [7:34:55<11:38:12, 2.31s/it] 39%|███▉ | 11842/30000 [7:34:57<11:35:36, 2.30s/it] 39%|███▉ | 11843/30000 [7:34:59<11:38:37, 2.31s/it] 39%|███▉ | 11844/30000 [7:35:02<11:39:57, 2.31s/it] 39%|███▉ | 11845/30000 [7:35:04<11:35:15, 2.30s/it] 39%|███▉ | 11846/30000 [7:35:06<11:32:48, 2.29s/it] 39%|███▉ | 11847/30000 [7:35:08<11:31:10, 2.28s/it] 39%|███▉ | 11848/30000 [7:35:11<11:29:35, 2.28s/it] 39%|███▉ | 11849/30000 [7:35:13<11:26:10, 2.27s/it] 40%|███▉ | 11850/30000 [7:35:15<11:32:33, 2.29s/it] 40%|███▉ | 11850/30000 [7:35:15<11:32:33, 2.29s/it] 40%|███▉ | 11851/30000 [7:35:18<11:32:04, 2.29s/it] 40%|███▉ | 11852/30000 [7:35:20<11:29:20, 2.28s/it] 40%|███▉ | 11853/30000 [7:35:22<11:36:33, 2.30s/it] 40%|███▉ | 11854/30000 [7:35:25<11:37:24, 2.31s/it] 40%|███▉ | 11855/30000 [7:35:27<11:33:38, 2.29s/it] 40%|███▉ | 11856/30000 [7:35:29<11:37:40, 2.31s/it] 40%|███▉ | 11857/30000 [7:35:31<11:38:03, 2.31s/it] 40%|███▉ | 11858/30000 [7:35:34<11:34:54, 2.30s/it] 40%|███▉ | 11859/30000 [7:35:36<11:32:56, 2.29s/it] 40%|███▉ | 11860/30000 [7:35:38<11:35:19, 2.30s/it] 40%|███▉ | 11860/30000 [7:35:38<11:35:19, 2.30s/it] 40%|███▉ | 11861/30000 [7:35:41<11:34:08, 2.30s/it] 40%|███▉ | 11862/30000 [7:35:43<11:33:56, 2.30s/it] 40%|███▉ | 11863/30000 [7:35:45<11:40:48, 2.32s/it] 40%|███▉ | 11864/30000 [7:35:48<11:36:23, 2.30s/it] 40%|███▉ | 11865/30000 [7:35:50<11:31:26, 2.29s/it] 40%|███▉ | 11866/30000 [7:35:52<11:36:03, 2.30s/it] 40%|███▉ | 11867/30000 [7:35:54<11:37:22, 2.31s/it] 40%|███▉ | 11868/30000 [7:35:57<11:34:03, 2.30s/it] 40%|███▉ | 11869/30000 [7:35:59<11:33:47, 2.30s/it] 40%|███▉ | 11870/30000 [7:36:01<11:35:22, 2.30s/it] 40%|███▉ | 11870/30000 [7:36:01<11:35:22, 2.30s/it] 40%|███▉ | 11871/30000 [7:36:04<11:32:05, 2.29s/it] 40%|███▉ | 11872/30000 [7:36:06<11:29:01, 2.28s/it] 40%|███▉ | 11873/30000 [7:36:08<11:38:38, 2.31s/it] 40%|███▉ | 11874/30000 [7:36:11<11:36:48, 2.31s/it] 40%|███▉ | 11875/30000 [7:36:13<11:33:32, 2.30s/it] 40%|███▉ | 11876/30000 [7:36:15<11:40:42, 2.32s/it] 40%|███▉ | 11877/30000 [7:36:17<11:38:08, 2.31s/it] 40%|███▉ | 11878/30000 [7:36:20<11:35:09, 2.30s/it] 40%|███▉ | 11879/30000 [7:36:22<11:30:31, 2.29s/it] 40%|███▉ | 11880/30000 [7:36:24<11:34:12, 2.30s/it] 40%|███▉ | 11880/30000 [7:36:24<11:34:12, 2.30s/it] 40%|███▉ | 11881/30000 [7:36:27<11:31:36, 2.29s/it] 40%|███▉ | 11882/30000 [7:36:29<11:27:41, 2.28s/it] 40%|███▉ | 11883/30000 [7:36:31<11:32:33, 2.29s/it] 40%|███▉ | 11884/30000 [7:36:33<11:34:09, 2.30s/it] 40%|███▉ | 11885/30000 [7:36:36<11:30:47, 2.29s/it] 40%|███▉ | 11886/30000 [7:36:38<11:30:06, 2.29s/it] 40%|███▉ | 11887/30000 [7:36:40<11:33:00, 2.30s/it] 40%|███▉ | 11888/30000 [7:36:43<11:33:41, 2.30s/it] 40%|███▉ | 11889/30000 [7:36:45<11:30:01, 2.29s/it] 40%|███▉ | 11890/30000 [7:36:47<11:28:23, 2.28s/it] 40%|███▉ | 11890/30000 [7:36:47<11:28:23, 2.28s/it] 40%|███▉ | 11891/30000 [7:36:50<11:32:05, 2.29s/it] 40%|███▉ | 11892/30000 [7:36:52<11:30:27, 2.29s/it] 40%|███▉ | 11893/30000 [7:36:54<11:28:26, 2.28s/it] 40%|███▉ | 11894/30000 [7:36:56<11:35:48, 2.31s/it] 40%|███▉ | 11895/30000 [7:36:59<11:32:36, 2.30s/it] 40%|███▉ | 11896/30000 [7:37:01<11:30:30, 2.29s/it] 40%|███▉ | 11897/30000 [7:37:03<11:26:05, 2.27s/it] 40%|███▉ | 11898/30000 [7:37:06<11:31:24, 2.29s/it] 40%|███▉ | 11899/30000 [7:37:08<11:35:06, 2.30s/it] 40%|███▉ | 11900/30000 [7:37:10<11:33:40, 2.30s/it] 40%|███▉ | 11900/30000 [7:37:10<11:33:40, 2.30s/it] 40%|███▉ | 11901/30000 [7:37:12<11:33:18, 2.30s/it] 40%|███▉ | 11902/30000 [7:37:15<11:36:49, 2.31s/it] 40%|███▉ | 11903/30000 [7:37:17<11:30:10, 2.29s/it] 40%|███▉ | 11904/30000 [7:37:19<11:26:26, 2.28s/it] 40%|███▉ | 11905/30000 [7:37:22<11:30:14, 2.29s/it] 40%|███▉ | 11906/30000 [7:37:24<11:33:09, 2.30s/it] 40%|███▉ | 11907/30000 [7:37:26<11:30:50, 2.29s/it] 40%|███▉ | 11908/30000 [7:37:28<11:29:20, 2.29s/it] 40%|███▉ | 11909/30000 [7:37:31<11:37:49, 2.31s/it] 40%|███▉ | 11910/30000 [7:37:33<11:36:29, 2.31s/it] 40%|███▉ | 11910/30000 [7:37:33<11:36:29, 2.31s/it] 40%|███▉ | 11911/30000 [7:37:35<11:33:45, 2.30s/it] 40%|███▉ | 11912/30000 [7:37:38<11:33:49, 2.30s/it] 40%|███▉ | 11913/30000 [7:37:40<11:36:12, 2.31s/it] 40%|███▉ | 11914/30000 [7:37:42<11:36:04, 2.31s/it] 40%|███▉ | 11915/30000 [7:37:45<11:31:45, 2.30s/it] 40%|███▉ | 11916/30000 [7:37:47<11:30:45, 2.29s/it] 40%|███▉ | 11917/30000 [7:37:49<11:34:27, 2.30s/it] 40%|███▉ | 11918/30000 [7:37:52<11:32:15, 2.30s/it] 40%|███▉ | 11919/30000 [7:37:54<11:29:54, 2.29s/it] 40%|███▉ | 11920/30000 [7:37:56<11:30:50, 2.29s/it] 40%|███▉ | 11920/30000 [7:37:56<11:30:50, 2.29s/it] 40%|███▉ | 11921/30000 [7:37:58<11:30:39, 2.29s/it] 40%|███▉ | 11922/30000 [7:38:01<11:28:57, 2.29s/it] 40%|███▉ | 11923/30000 [7:38:03<11:29:25, 2.29s/it] 40%|███▉ | 11924/30000 [7:38:05<11:37:32, 2.32s/it] 40%|███▉ | 11925/30000 [7:38:08<11:42:21, 2.33s/it] 40%|███▉ | 11926/30000 [7:38:10<11:38:20, 2.32s/it] 40%|███▉ | 11927/30000 [7:38:12<11:26:47, 2.28s/it] 40%|███▉ | 11928/30000 [7:38:14<11:29:58, 2.29s/it] 40%|███▉ | 11929/30000 [7:38:17<11:28:09, 2.28s/it] 40%|███▉ | 11930/30000 [7:38:19<11:25:43, 2.28s/it] 40%|███▉ | 11930/30000 [7:38:19<11:25:43, 2.28s/it] 40%|███▉ | 11931/30000 [7:38:21<11:24:47, 2.27s/it] 40%|███▉ | 11932/30000 [7:38:24<11:33:38, 2.30s/it] 40%|███▉ | 11933/30000 [7:38:26<11:38:33, 2.32s/it] 40%|███▉ | 11934/30000 [7:38:28<11:36:21, 2.31s/it] 40%|███▉ | 11935/30000 [7:38:31<11:38:17, 2.32s/it] 40%|███▉ | 11936/30000 [7:38:33<11:39:10, 2.32s/it] 40%|███▉ | 11937/30000 [7:38:35<11:33:00, 2.30s/it] 40%|███▉ | 11938/30000 [7:38:37<11:28:51, 2.29s/it] 40%|███▉ | 11939/30000 [7:38:40<11:32:32, 2.30s/it] 40%|███▉ | 11940/30000 [7:38:42<11:37:18, 2.32s/it] 40%|███▉ | 11940/30000 [7:38:42<11:37:18, 2.32s/it] 40%|███▉ | 11941/30000 [7:38:44<11:35:37, 2.31s/it] 40%|███▉ | 11942/30000 [7:38:47<11:34:00, 2.31s/it] 40%|███▉ | 11943/30000 [7:38:49<11:32:53, 2.30s/it] 40%|███▉ | 11944/30000 [7:38:51<11:27:41, 2.29s/it] 40%|███▉ | 11945/30000 [7:38:54<11:23:56, 2.27s/it] 40%|███▉ | 11946/30000 [7:38:56<11:29:08, 2.29s/it] 40%|███▉ | 11947/30000 [7:38:58<11:29:34, 2.29s/it] 40%|███▉ | 11948/30000 [7:39:00<11:28:52, 2.29s/it] 40%|███▉ | 11949/30000 [7:39:03<11:33:16, 2.30s/it] 40%|███▉ | 11950/30000 [7:39:05<11:35:07, 2.31s/it] 40%|███▉ | 11950/30000 [7:39:05<11:35:07, 2.31s/it] 40%|███▉ | 11951/30000 [7:39:07<11:30:20, 2.29s/it] 40%|███▉ | 11952/30000 [7:39:10<11:26:11, 2.28s/it] 40%|███▉ | 11953/30000 [7:39:12<11:31:12, 2.30s/it] 40%|███▉ | 11954/30000 [7:39:14<11:26:09, 2.28s/it] 40%|███▉ | 11955/30000 [7:39:16<11:26:55, 2.28s/it] 40%|███▉ | 11956/30000 [7:39:19<11:32:48, 2.30s/it] 40%|███▉ | 11957/30000 [7:39:21<11:35:50, 2.31s/it] 40%|███▉ | 11958/30000 [7:39:23<11:34:23, 2.31s/it] 40%|███▉ | 11959/30000 [7:39:26<11:31:08, 2.30s/it] 40%|███▉ | 11960/30000 [7:39:28<11:33:55, 2.31s/it] 40%|███▉ | 11960/30000 [7:39:28<11:33:55, 2.31s/it] 40%|███▉ | 11961/30000 [7:39:30<11:33:37, 2.31s/it] 40%|███▉ | 11962/30000 [7:39:33<11:27:34, 2.29s/it] 40%|███▉ | 11963/30000 [7:39:35<11:33:44, 2.31s/it] 40%|███▉ | 11964/30000 [7:39:37<11:40:43, 2.33s/it] 40%|███▉ | 11965/30000 [7:39:40<11:30:32, 2.30s/it] 40%|███▉ | 11966/30000 [7:39:42<11:30:14, 2.30s/it] 40%|███▉ | 11967/30000 [7:39:44<11:34:36, 2.31s/it] 40%|███▉ | 11968/30000 [7:39:47<11:31:35, 2.30s/it] 40%|███▉ | 11969/30000 [7:39:49<11:27:08, 2.29s/it] 40%|███▉ | 11970/30000 [7:39:51<11:31:58, 2.30s/it] 40%|███▉ | 11970/30000 [7:39:51<11:31:58, 2.30s/it] 40%|███▉ | 11971/30000 [7:39:53<11:34:21, 2.31s/it] 40%|███▉ | 11972/30000 [7:39:56<11:32:47, 2.31s/it] 40%|███▉ | 11973/30000 [7:39:58<11:38:47, 2.33s/it] 40%|███▉ | 11974/30000 [7:40:00<11:41:51, 2.34s/it] 40%|███▉ | 11975/30000 [7:40:03<11:30:07, 2.30s/it] 40%|███▉ | 11976/30000 [7:40:05<11:23:29, 2.28s/it] 40%|███▉ | 11977/30000 [7:40:07<11:29:16, 2.29s/it] 40%|███▉ | 11978/30000 [7:40:09<11:24:34, 2.28s/it] 40%|███▉ | 11979/30000 [7:40:12<11:27:14, 2.29s/it] 40%|███▉ | 11980/30000 [7:40:14<11:34:57, 2.31s/it] 40%|███▉ | 11980/30000 [7:40:14<11:34:57, 2.31s/it] 40%|███▉ | 11981/30000 [7:40:16<11:32:05, 2.30s/it] 40%|███▉ | 11982/30000 [7:40:19<11:32:55, 2.31s/it] 40%|███▉ | 11983/30000 [7:40:21<11:33:35, 2.31s/it] 40%|███▉ | 11984/30000 [7:40:23<11:33:13, 2.31s/it] 40%|███▉ | 11985/30000 [7:40:26<11:27:44, 2.29s/it] 40%|███▉ | 11986/30000 [7:40:28<11:25:36, 2.28s/it] 40%|███▉ | 11987/30000 [7:40:30<11:30:53, 2.30s/it] 40%|███▉ | 11988/30000 [7:40:33<11:33:41, 2.31s/it] 40%|███▉ | 11989/30000 [7:40:35<11:32:41, 2.31s/it] 40%|███▉ | 11990/30000 [7:40:37<11:36:31, 2.32s/it] 40%|███▉ | 11990/30000 [7:40:37<11:36:31, 2.32s/it] 40%|███▉ | 11991/30000 [7:40:40<11:37:06, 2.32s/it] 40%|███▉ | 11992/30000 [7:40:42<11:31:30, 2.30s/it] 40%|███▉ | 11993/30000 [7:40:44<11:30:14, 2.30s/it] 40%|███▉ | 11994/30000 [7:40:46<11:29:27, 2.30s/it] 40%|███▉ | 11995/30000 [7:40:49<11:25:22, 2.28s/it] 40%|███▉ | 11996/30000 [7:40:51<11:25:06, 2.28s/it] 40%|███▉ | 11997/30000 [7:40:53<11:31:33, 2.30s/it] 40%|███▉ | 11998/30000 [7:40:56<11:25:53, 2.29s/it] 40%|███▉ | 11999/30000 [7:40:58<11:23:06, 2.28s/it] 40%|████ | 12000/30000 [7:41:00<11:26:20, 2.29s/it] 40%|████ | 12000/30000 [7:41:00<11:26:20, 2.29s/it] 40%|████ | 12001/30000 [7:41:02<11:27:04, 2.29s/it] 40%|████ | 12002/30000 [7:41:05<11:29:16, 2.30s/it] 40%|████ | 12003/30000 [7:41:07<11:32:28, 2.31s/it] 40%|████ | 12004/30000 [7:41:09<11:33:04, 2.31s/it] 40%|████ | 12005/30000 [7:41:12<11:29:12, 2.30s/it] 40%|████ | 12006/30000 [7:41:14<11:29:08, 2.30s/it] 40%|████ | 12007/30000 [7:41:16<11:31:52, 2.31s/it] 40%|████ | 12008/30000 [7:41:18<11:25:27, 2.29s/it] 40%|████ | 12009/30000 [7:41:21<11:22:18, 2.28s/it] 40%|████ | 12010/30000 [7:41:23<11:24:17, 2.28s/it] 40%|████ | 12010/30000 [7:41:23<11:24:17, 2.28s/it] 40%|████ | 12011/30000 [7:41:25<11:21:40, 2.27s/it] 40%|████ | 12012/30000 [7:41:28<11:25:09, 2.29s/it] 40%|████ | 12013/30000 [7:41:30<11:36:45, 2.32s/it] 40%|████ | 12014/30000 [7:41:32<11:38:18, 2.33s/it] 40%|████ | 12015/30000 [7:41:35<11:35:05, 2.32s/it] 40%|████ | 12016/30000 [7:41:37<11:30:19, 2.30s/it] 40%|████ | 12017/30000 [7:41:39<11:32:40, 2.31s/it] 40%|████ | 12018/30000 [7:41:41<11:28:06, 2.30s/it] 40%|████ | 12019/30000 [7:41:44<11:26:39, 2.29s/it] 40%|████ | 12020/30000 [7:41:46<11:32:01, 2.31s/it] 40%|████ | 12020/30000 [7:41:46<11:32:01, 2.31s/it] 40%|████ | 12021/30000 [7:41:48<11:37:08, 2.33s/it] 40%|████ | 12022/30000 [7:41:51<11:34:34, 2.32s/it] 40%|████ | 12023/30000 [7:41:53<11:35:02, 2.32s/it] 40%|████ | 12024/30000 [7:41:55<11:32:29, 2.31s/it] 40%|████ | 12025/30000 [7:41:58<11:30:05, 2.30s/it] 40%|████ | 12026/30000 [7:42:00<11:28:32, 2.30s/it] 40%|████ | 12027/30000 [7:42:02<11:34:27, 2.32s/it] 40%|████ | 12028/30000 [7:42:05<11:36:18, 2.32s/it] 40%|████ | 12029/30000 [7:42:07<11:29:14, 2.30s/it] 40%|████ | 12030/30000 [7:42:09<11:29:11, 2.30s/it] 40%|████ | 12030/30000 [7:42:09<11:29:11, 2.30s/it] 40%|████ | 12031/30000 [7:42:12<11:30:52, 2.31s/it] 40%|████ | 12032/30000 [7:42:14<11:26:19, 2.29s/it] 40%|████ | 12033/30000 [7:42:16<11:23:30, 2.28s/it] 40%|████ | 12034/30000 [7:42:18<11:27:11, 2.29s/it] 40%|████ | 12035/30000 [7:42:21<11:22:42, 2.28s/it] 40%|████ | 12036/30000 [7:42:23<11:25:32, 2.29s/it] 40%|████ | 12037/30000 [7:42:25<11:30:15, 2.31s/it] 40%|████ | 12038/30000 [7:42:28<11:31:25, 2.31s/it] 40%|████ | 12039/30000 [7:42:30<11:25:30, 2.29s/it] 40%|████ | 12040/30000 [7:42:32<11:19:43, 2.27s/it] +{'loss': 0.0148, 'grad_norm': 0.06722988933324814, 'learning_rate': 2.156714452286416e-05, 'flow_matching_loss': 0.014838141109794378, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.39} +{'loss': 0.0159, 'grad_norm': 0.07635676115751266, 'learning_rate': 2.155227472301591e-05, 'flow_matching_loss': 0.015889868140220642, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.39} +{'loss': 0.0151, 'grad_norm': 0.10639965534210205, 'learning_rate': 2.1537396961536027e-05, 'flow_matching_loss': 0.015074862632900476, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.39} +{'loss': 0.0153, 'grad_norm': 0.0752687081694603, 'learning_rate': 2.1522511256502403e-05, 'flow_matching_loss': 0.015277268923819065, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.4} +{'loss': 0.0152, 'grad_norm': 0.08551640063524246, 'learning_rate': 2.1507617626002566e-05, 'flow_matching_loss': 0.015159731544554234, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.4} +{'loss': 0.0166, 'grad_norm': 0.07033184170722961, 'learning_rate': 2.1492716088133685e-05, 'flow_matching_loss': 0.016628528572618963, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.4} +{'loss': 0.0148, 'grad_norm': 0.07217340171337128, 'learning_rate': 2.1477806661002536e-05, 'flow_matching_loss': 0.014843813702464103, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.4} +{'loss': 0.0163, 'grad_norm': 0.07894200086593628, 'learning_rate': 2.146288936272548e-05, 'flow_matching_loss': 0.016312492452561854, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.41} +{'loss': 0.0155, 'grad_norm': 0.09515032172203064, 'learning_rate': 2.1447964211428433e-05, 'flow_matching_loss': 0.01553204944357276, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.41} +{'loss': 0.014, 'grad_norm': 0.08343199640512466, 'learning_rate': 2.1433031225246865e-05, 'flow_matching_loss': 0.013957070000469685, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.41} +{'loss': 0.0184, 'grad_norm': 0.07973859459161758, 'learning_rate': 2.1418090422325766e-05, 'flow_matching_loss': 0.018415308836847544, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.42} +{'loss': 0.0146, 'grad_norm': 0.0724276602268219, 'learning_rate': 2.1403141820819623e-05, 'flow_matching_loss': 0.0146258813329041, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.42} +{'loss': 0.0155, 'grad_norm': 0.0834338590502739, 'learning_rate': 2.1388185438892377e-05, 'flow_matching_loss': 0.015520735085010529, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.42} +{'loss': 0.0157, 'grad_norm': 0.07127588987350464, 'learning_rate': 2.1373221294717465e-05, 'flow_matching_loss': 0.015677267871797084, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.42} +{'loss': 0.0167, 'grad_norm': 0.07928606122732162, 'learning_rate': 2.1358249406477724e-05, 'flow_matching_loss': 0.01666880175471306, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.43} +{'loss': 0.0157, 'grad_norm': 0.06755295395851135, 'learning_rate': 2.1343269792365408e-05, 'flow_matching_loss': 0.015719879325479268, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.43} +{'loss': 0.0164, 'grad_norm': 0.09301071614027023, 'learning_rate': 2.132828247058217e-05, 'flow_matching_loss': 0.016388945654034613, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.43} +{'loss': 0.0157, 'grad_norm': 0.085683174431324, 'learning_rate': 2.1313287459339004e-05, 'flow_matching_loss': 0.01569692250341177, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.44} +{'loss': 0.0146, 'grad_norm': 0.08098357915878296, 'learning_rate': 2.1298284776856276e-05, 'flow_matching_loss': 0.01459726169705391, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.44} +{'loss': 0.0158, 'grad_norm': 0.08477999269962311, 'learning_rate': 2.128327444136365e-05, 'flow_matching_loss': 0.015756877791136503, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.44} +{'loss': 0.0156, 'grad_norm': 0.07628018409013748, 'learning_rate': 2.1268256471100105e-05, 'flow_matching_loss': 0.015641486924141645, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.44} +{'loss': 0.0145, 'grad_norm': 0.061489708721637726, 'learning_rate': 2.1253230884313893e-05, 'flow_matching_loss': 0.014458597265183925, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.45} +{'loss': 0.016, 'grad_norm': 0.08221996575593948, 'learning_rate': 2.123819769926251e-05, 'flow_matching_loss': 0.016034630592912436, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.45} +{'loss': 0.015, 'grad_norm': 0.08176316320896149, 'learning_rate': 2.1223156934212702e-05, 'flow_matching_loss': 0.01497499030083418, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.45} +{'loss': 0.0155, 'grad_norm': 0.07133590430021286, 'learning_rate': 2.1208108607440418e-05, 'flow_matching_loss': 0.0155030257999897, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.46} +{'loss': 0.0153, 'grad_norm': 0.06520935893058777, 'learning_rate': 2.119305273723078e-05, 'flow_matching_loss': 0.015317054651677608, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.46} +{'loss': 0.0143, 'grad_norm': 0.0804227665066719, 'learning_rate': 2.1177989341878106e-05, 'flow_matching_loss': 0.01432994194328785, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.46} +{'loss': 0.0158, 'grad_norm': 0.06417582929134369, 'learning_rate': 2.116291843968583e-05, 'flow_matching_loss': 0.01584387058392167, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.47} +{'loss': 0.016, 'grad_norm': 0.09675505012273788, 'learning_rate': 2.114784004896653e-05, 'flow_matching_loss': 0.01601812206208706, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.47} +{'loss': 0.0163, 'grad_norm': 0.06801251322031021, 'learning_rate': 2.113275418804187e-05, 'flow_matching_loss': 0.01629976453259587, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.47} +{'loss': 0.0144, 'grad_norm': 0.08308740705251694, 'learning_rate': 2.1117660875242583e-05, 'flow_matching_loss': 0.014379793219268322, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.47} +{'loss': 0.0156, 'grad_norm': 0.09500250220298767, 'learning_rate': 2.1102560128908478e-05, 'flow_matching_loss': 0.015599736105650663, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.48} +{'loss': 0.0141, 'grad_norm': 0.07440651953220367, 'learning_rate': 2.1087451967388384e-05, 'flow_matching_loss': 0.014109668787568808, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.48} +{'loss': 0.0149, 'grad_norm': 0.0827968567609787, 'learning_rate': 2.1072336409040135e-05, 'flow_matching_loss': 0.01485594203695655, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.48} +{'loss': 0.017, 'grad_norm': 0.07440180331468582, 'learning_rate': 2.1057213472230573e-05, 'flow_matching_loss': 0.016974958404898644, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.49} +{'loss': 0.0172, 'grad_norm': 0.09330865740776062, 'learning_rate': 2.1042083175335478e-05, 'flow_matching_loss': 0.017234695609658957, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.49} +{'loss': 0.0146, 'grad_norm': 0.08215612173080444, 'learning_rate': 2.1026945536739595e-05, 'flow_matching_loss': 0.014584231749176979, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.49} +{'loss': 0.0161, 'grad_norm': 0.07790908962488174, 'learning_rate': 2.1011800574836586e-05, 'flow_matching_loss': 0.01606825226917863, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.49} +{'loss': 0.0148, 'grad_norm': 0.05681914463639259, 'learning_rate': 2.0996648308028992e-05, 'flow_matching_loss': 0.014840306714177132, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.5} +{'loss': 0.0142, 'grad_norm': 0.11534086614847183, 'learning_rate': 2.0981488754728266e-05, 'flow_matching_loss': 0.014231746643781662, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.5} +{'loss': 0.0152, 'grad_norm': 0.07636135816574097, 'learning_rate': 2.0966321933354685e-05, 'flow_matching_loss': 0.015246598236262798, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.5} 40%|████ | 12040/30000 [7:42:32<11:19:43, 2.27s/it] 40%|████ | 12041/30000 [7:42:34<11:26:20, 2.29s/it] 40%|████ | 12042/30000 [7:42:37<11:22:57, 2.28s/it] 40%|████ | 12043/30000 [7:42:39<11:24:38, 2.29s/it] 40%|████ | 12044/30000 [7:42:41<11:30:33, 2.31s/it] 40%|████ | 12045/30000 [7:42:44<11:29:52, 2.31s/it] 40%|████ | 12046/30000 [7:42:46<11:28:03, 2.30s/it] 40%|████ | 12047/30000 [7:42:48<11:25:57, 2.29s/it] 40%|████ | 12048/30000 [7:42:51<11:30:07, 2.31s/it] 40%|████ | 12049/30000 [7:42:53<11:31:17, 2.31s/it] 40%|████ | 12050/30000 [7:42:55<11:26:14, 2.29s/it] 40%|████ | 12050/30000 [7:42:55<11:26:14, 2.29s/it] 40%|████ | 12051/30000 [7:42:57<11:25:19, 2.29s/it] 40%|████ | 12052/30000 [7:43:00<11:29:31, 2.31s/it] 40%|████ | 12053/30000 [7:43:02<11:30:15, 2.31s/it] 40%|████ | 12054/30000 [7:43:04<11:24:21, 2.29s/it] 40%|████ | 12055/30000 [7:43:07<11:21:37, 2.28s/it] 40%|████ | 12056/30000 [7:43:09<11:26:05, 2.29s/it] 40%|████ | 12057/30000 [7:43:11<11:27:49, 2.30s/it] 40%|████ | 12058/30000 [7:43:13<11:25:56, 2.29s/it] 40%|████ | 12059/30000 [7:43:16<11:20:49, 2.28s/it] 40%|████ | 12060/30000 [7:43:18<11:28:06, 2.30s/it] 40%|████ | 12060/30000 [7:43:18<11:28:06, 2.30s/it] 40%|████ | 12061/30000 [7:43:20<11:27:06, 2.30s/it] 40%|████ | 12062/30000 [7:43:23<11:27:20, 2.30s/it] 40%|████ | 12063/30000 [7:43:25<11:25:25, 2.29s/it] 40%|████ | 12064/30000 [7:43:27<11:28:02, 2.30s/it] 40%|████ | 12065/30000 [7:43:30<11:26:20, 2.30s/it] 40%|████ | 12066/30000 [7:43:32<11:25:42, 2.29s/it] 40%|████ | 12067/30000 [7:43:34<11:24:38, 2.29s/it] 40%|████ | 12068/30000 [7:43:36<11:24:30, 2.29s/it] 40%|████ | 12069/30000 [7:43:39<11:20:21, 2.28s/it] 40%|████ | 12070/30000 [7:43:41<11:20:08, 2.28s/it] 40%|████ | 12070/30000 [7:43:41<11:20:08, 2.28s/it] 40%|████ | 12071/30000 [7:43:43<11:25:46, 2.29s/it] 40%|████ | 12072/30000 [7:43:46<11:26:28, 2.30s/it] 40%|████ | 12073/30000 [7:43:48<11:22:49, 2.29s/it] 40%|████ | 12074/30000 [7:43:50<11:20:37, 2.28s/it] 40%|████ | 12075/30000 [7:43:52<11:25:46, 2.30s/it] 40%|████ | 12076/30000 [7:43:55<11:32:46, 2.32s/it] 40%|████ | 12077/30000 [7:43:57<11:25:32, 2.29s/it] 40%|████ | 12078/30000 [7:43:59<11:21:39, 2.28s/it] 40%|████ | 12079/30000 [7:44:02<11:24:46, 2.29s/it] 40%|████ | 12080/30000 [7:44:04<11:21:37, 2.28s/it] 40%|████ | 12080/30000 [7:44:04<11:21:37, 2.28s/it] 40%|████ | 12081/30000 [7:44:06<11:23:17, 2.29s/it] 40%|████ | 12082/30000 [7:44:08<11:22:52, 2.29s/it] 40%|████ | 12083/30000 [7:44:11<11:24:01, 2.29s/it] 40%|████ | 12084/30000 [7:44:13<11:20:41, 2.28s/it] 40%|████ | 12085/30000 [7:44:15<11:22:05, 2.28s/it] 40%|████ | 12086/30000 [7:44:17<10:24:24, 2.09s/it] 40%|████ | 12087/30000 [7:44:18<9:04:25, 1.82s/it] 40%|████ | 12088/30000 [7:44:19<8:07:38, 1.63s/it] 40%|████ | 12089/30000 [7:44:21<7:28:45, 1.50s/it] 40%|████ | 12090/30000 [7:44:22<6:59:35, 1.41s/it] 40%|████ | 12090/30000 [7:44:22<6:59:35, 1.41s/it] 40%|████ | 12091/30000 [7:44:23<6:43:24, 1.35s/it] 40%|████ | 12092/30000 [7:44:25<7:33:10, 1.52s/it] 40%|████ | 12093/30000 [7:44:27<8:44:23, 1.76s/it] 40%|████ | 12094/30000 [7:44:29<9:34:29, 1.93s/it] 40%|████ | 12095/30000 [7:44:32<10:06:17, 2.03s/it] 40%|████ | 12096/30000 [7:44:34<10:32:51, 2.12s/it] 40%|████ | 12097/30000 [7:44:36<10:50:01, 2.18s/it] 40%|████ | 12098/30000 [7:44:39<10:56:25, 2.20s/it] 40%|████ | 12099/30000 [7:44:41<11:08:12, 2.24s/it] 40%|████ | 12100/30000 [7:44:43<11:16:40, 2.27s/it] 40%|████ | 12100/30000 [7:44:43<11:16:40, 2.27s/it] 40%|████ | 12101/30000 [7:44:46<11:16:09, 2.27s/it] 40%|████ | 12102/30000 [7:44:48<11:18:25, 2.27s/it] 40%|████ | 12103/30000 [7:44:50<11:22:30, 2.29s/it] 40%|████ | 12104/30000 [7:44:52<11:23:08, 2.29s/it] 40%|████ | 12105/30000 [7:44:55<11:20:50, 2.28s/it] 40%|████ | 12106/30000 [7:44:57<11:28:32, 2.31s/it] 40%|████ | 12107/30000 [7:44:59<11:27:46, 2.31s/it] 40%|████ | 12108/30000 [7:45:02<11:25:34, 2.30s/it] 40%|████ | 12109/30000 [7:45:04<11:23:52, 2.29s/it] 40%|████ | 12110/30000 [7:45:06<11:27:47, 2.31s/it] 40%|████ | 12110/30000 [7:45:06<11:27:47, 2.31s/it] 40%|████ | 12111/30000 [7:45:09<11:24:26, 2.30s/it] 40%|████ | 12112/30000 [7:45:11<11:20:44, 2.28s/it] 40%|████ | 12113/30000 [7:45:13<11:26:59, 2.30s/it] 40%|████ | 12114/30000 [7:45:15<11:25:46, 2.30s/it] 40%|████ | 12115/30000 [7:45:18<11:23:53, 2.29s/it] 40%|████ | 12116/30000 [7:45:20<11:25:18, 2.30s/it] 40%|████ | 12117/30000 [7:45:22<11:24:37, 2.30s/it] 40%|████ | 12118/30000 [7:45:25<11:22:42, 2.29s/it] 40%|████ | 12119/30000 [7:45:27<11:22:09, 2.29s/it] 40%|████ | 12120/30000 [7:45:29<11:24:29, 2.30s/it] 40%|████ | 12120/30000 [7:45:29<11:24:29, 2.30s/it] 40%|████ | 12121/30000 [7:45:31<11:19:38, 2.28s/it] 40%|████ | 12122/30000 [7:45:34<11:18:33, 2.28s/it] 40%|████ | 12123/30000 [7:45:36<11:27:01, 2.31s/it] 40%|████ | 12124/30000 [7:45:38<11:30:05, 2.32s/it] 40%|████ | 12125/30000 [7:45:41<11:28:49, 2.31s/it] 40%|████ | 12126/30000 [7:45:43<11:24:25, 2.30s/it] 40%|████ | 12127/30000 [7:45:45<11:26:10, 2.30s/it] 40%|████ | 12128/30000 [7:45:48<11:21:39, 2.29s/it] 40%|████ | 12129/30000 [7:45:50<11:19:32, 2.28s/it] 40%|████ | 12130/30000 [7:45:52<11:25:12, 2.30s/it] 40%|████ | 12130/30000 [7:45:52<11:25:12, 2.30s/it] 40%|████ | 12131/30000 [7:45:55<11:28:42, 2.31s/it] 40%|████ | 12132/30000 [7:45:57<11:23:14, 2.29s/it] 40%|████ | 12133/30000 [7:45:59<11:26:04, 2.30s/it] 40%|████ | 12134/30000 [7:46:01<11:30:35, 2.32s/it] 40%|████ | 12135/30000 [7:46:04<11:27:09, 2.31s/it] 40%|████ | 12136/30000 [7:46:06<11:27:19, 2.31s/it] 40%|████ | 12137/30000 [7:46:08<11:28:29, 2.31s/it] 40%|████ | 12138/30000 [7:46:11<11:24:50, 2.30s/it] 40%|████ | 12139/30000 [7:46:13<11:18:05, 2.28s/it] 40%|████ | 12140/30000 [7:46:15<11:23:31, 2.30s/it] 40%|████ | 12140/30000 [7:46:15<11:23:31, 2.30s/it] 40%|████ | 12141/30000 [7:46:18<11:24:10, 2.30s/it] 40%|████ | 12142/30000 [7:46:20<11:19:56, 2.28s/it] 40%|████ | 12143/30000 [7:46:22<11:23:32, 2.30s/it] 40%|████ | 12144/30000 [7:46:24<11:24:28, 2.30s/it] 40%|████ | 12145/30000 [7:46:27<11:20:43, 2.29s/it] 40%|████ | 12146/30000 [7:46:29<11:22:04, 2.29s/it] 40%|████ | 12147/30000 [7:46:31<11:22:02, 2.29s/it] 40%|████ | 12148/30000 [7:46:34<11:20:09, 2.29s/it] 40%|████ | 12149/30000 [7:46:36<11:17:32, 2.28s/it] 40%|████ | 12150/30000 [7:46:38<11:23:02, 2.30s/it] 40%|████ | 12150/30000 [7:46:38<11:23:02, 2.30s/it] 41%|████ | 12151/30000 [7:46:40<11:20:17, 2.29s/it] 41%|████ | 12152/30000 [7:46:43<11:19:22, 2.28s/it] 41%|████ | 12153/30000 [7:46:45<11:25:56, 2.31s/it] 41%|████ | 12154/30000 [7:46:47<11:28:59, 2.32s/it] 41%|████ | 12155/30000 [7:46:50<11:25:51, 2.31s/it] 41%|████ | 12156/30000 [7:46:52<11:26:14, 2.31s/it] 41%|████ | 12157/30000 [7:46:54<11:27:59, 2.31s/it] 41%|████ | 12158/30000 [7:46:57<11:24:28, 2.30s/it] 41%|████ | 12159/30000 [7:46:59<11:21:57, 2.29s/it] 41%|████ | 12160/30000 [7:47:01<11:29:35, 2.32s/it] 41%|████ | 12160/30000 [7:47:01<11:29:35, 2.32s/it] 41%|████ | 12161/30000 [7:47:04<11:27:28, 2.31s/it] 41%|████ | 12162/30000 [7:47:06<11:20:13, 2.29s/it] 41%|████ | 12163/30000 [7:47:08<11:18:29, 2.28s/it] 41%|████ | 12164/30000 [7:47:10<11:22:39, 2.30s/it] 41%|████ | 12165/30000 [7:47:13<11:21:48, 2.29s/it] 41%|████ | 12166/30000 [7:47:15<11:22:22, 2.30s/it] 41%|████ | 12167/30000 [7:47:17<11:27:01, 2.31s/it] 41%|████ | 12168/30000 [7:47:20<11:25:30, 2.31s/it] 41%|████ | 12169/30000 [7:47:22<11:20:12, 2.29s/it] 41%|████ | 12170/30000 [7:47:24<11:17:56, 2.28s/it] 41%|████ | 12170/30000 [7:47:24<11:17:56, 2.28s/it] 41%|████ | 12171/30000 [7:47:26<11:24:52, 2.30s/it] 41%|████ | 12172/30000 [7:47:29<11:23:02, 2.30s/it] 41%|████ | 12173/30000 [7:47:31<11:17:18, 2.28s/it] 41%|████ | 12174/30000 [7:47:33<11:23:06, 2.30s/it] 41%|████ | 12175/30000 [7:47:36<11:23:19, 2.30s/it] 41%|████ | 12176/30000 [7:47:38<11:19:20, 2.29s/it] 41%|████ | 12177/30000 [7:47:40<11:21:16, 2.29s/it] 41%|████ | 12178/30000 [7:47:42<11:19:31, 2.29s/it] 41%|████ | 12179/30000 [7:47:45<11:16:14, 2.28s/it] 41%|████ | 12180/30000 [7:47:47<11:13:33, 2.27s/it] 41%|████ | 12180/30000 [7:47:47<11:13:33, 2.27s/it] 41%|████ | 12181/30000 [7:47:49<11:19:51, 2.29s/it] 41%|████ | 12182/30000 [7:47:52<11:19:03, 2.29s/it] 41%|████ | 12183/30000 [7:47:54<11:17:45, 2.28s/it] 41%|████ | 12184/30000 [7:47:56<11:20:47, 2.29s/it] 41%|████ | 12185/30000 [7:47:58<11:19:29, 2.29s/it] 41%|████ | 12186/30000 [7:48:01<11:18:17, 2.28s/it] 41%|████ | 12187/30000 [7:48:03<11:16:41, 2.28s/it] 41%|████ | 12188/30000 [7:48:05<11:23:22, 2.30s/it] 41%|████ | 12189/30000 [7:48:08<11:21:06, 2.29s/it] 41%|████ | 12190/30000 [7:48:10<11:19:37, 2.29s/it] 41%|████ | 12190/30000 [7:48:10<11:19:37, 2.29s/it] 41%|████ | 12191/30000 [7:48:12<11:22:45, 2.30s/it] 41%|████ | 12192/30000 [7:48:15<11:22:46, 2.30s/it] 41%|████ | 12193/30000 [7:48:17<11:16:47, 2.28s/it] 41%|████ | 12194/30000 [7:48:19<11:15:37, 2.28s/it] 41%|████ | 12195/30000 [7:48:21<11:21:15, 2.30s/it] 41%|████ | 12196/30000 [7:48:24<11:22:52, 2.30s/it] 41%|████ | 12197/30000 [7:48:26<11:19:07, 2.29s/it] 41%|████ | 12198/30000 [7:48:28<11:19:31, 2.29s/it] 41%|████ | 12199/30000 [7:48:31<11:23:28, 2.30s/it] 41%|████ | 12200/30000 [7:48:33<11:18:20, 2.29s/it] 41%|████ | 12200/30000 [7:48:33<11:18:20, 2.29s/it] 41%|████ | 12201/30000 [7:48:35<11:16:10, 2.28s/it] 41%|████ | 12202/30000 [7:48:37<11:12:11, 2.27s/it] 41%|████ | 12203/30000 [7:48:40<11:15:52, 2.28s/it] 41%|████ | 12204/30000 [7:48:42<11:18:37, 2.29s/it] 41%|████ | 12205/30000 [7:48:44<11:15:47, 2.28s/it] 41%|████ | 12206/30000 [7:48:47<11:19:11, 2.29s/it] 41%|████ | 12207/30000 [7:48:49<11:23:35, 2.31s/it] 41%|████ | 12208/30000 [7:48:51<11:21:27, 2.30s/it] 41%|████ | 12209/30000 [7:48:53<11:17:28, 2.28s/it] 41%|████ | 12210/30000 [7:48:56<11:23:33, 2.31s/it] 41%|████ | 12210/30000 [7:48:56<11:23:33, 2.31s/it] 41%|████ | 12211/30000 [7:48:58<11:25:03, 2.31s/it] 41%|████ | 12212/30000 [7:49:00<11:22:36, 2.30s/it] 41%|████ | 12213/30000 [7:49:03<11:18:34, 2.29s/it] 41%|████ | 12214/30000 [7:49:05<11:16:44, 2.28s/it] 41%|████ | 12215/30000 [7:49:07<11:13:05, 2.27s/it] 41%|████ | 12216/30000 [7:49:09<11:12:25, 2.27s/it] 41%|████ | 12217/30000 [7:49:12<11:18:04, 2.29s/it] 41%|████ | 12218/30000 [7:49:14<11:18:34, 2.29s/it] 41%|████ | 12219/30000 [7:49:16<11:19:01, 2.29s/it] 41%|████ | 12220/30000 [7:49:19<11:17:24, 2.29s/it] 41%|████ | 12220/30000 [7:49:19<11:17:24, 2.29s/it] 41%|████ | 12221/30000 [7:49:21<11:20:19, 2.30s/it] 41%|████ | 12222/30000 [7:49:23<11:20:59, 2.30s/it] 41%|████ | 12223/30000 [7:49:25<11:18:58, 2.29s/it] 41%|████ | 12224/30000 [7:49:28<11:18:21, 2.29s/it] 41%|████ | 12225/30000 [7:49:30<11:20:06, 2.30s/it] 41%|████ | 12226/30000 [7:49:32<11:15:37, 2.28s/it] 41%|████ | 12227/30000 [7:49:35<11:12:52, 2.27s/it] 41%|████ | 12228/30000 [7:49:37<11:19:06, 2.29s/it] 41%|████ | 12229/30000 [7:49:39<11:18:21, 2.29s/it] 41%|████ | 12230/30000 [7:49:41<11:17:43, 2.29s/it] 41%|████ | 12230/30000 [7:49:42<11:17:43, 2.29s/it] 41%|████ | 12231/30000 [7:49:44<11:22:04, 2.30s/it] 41%|████ | 12232/30000 [7:49:46<11:20:19, 2.30s/it] 41%|████ | 12233/30000 [7:49:48<11:19:09, 2.29s/it] 41%|████ | 12234/30000 [7:49:51<11:16:47, 2.29s/it] 41%|████ | 12235/30000 [7:49:53<11:18:05, 2.29s/it] 41%|████ | 12236/30000 [7:49:55<11:19:44, 2.30s/it] 41%|████ | 12237/30000 [7:49:58<11:18:30, 2.29s/it] 41%|████ | 12238/30000 [7:50:00<11:21:08, 2.30s/it] 41%|████ | 12239/30000 [7:50:02<11:20:23, 2.30s/it] 41%|████ | 12240/30000 [7:50:04<11:18:42, 2.29s/it] 41%|████ | 12240/30000 [7:50:04<11:18:42, 2.29s/it] 41%|████ | 12241/30000 [7:50:07<11:22:26, 2.31s/it] 41%|████ | 12242/30000 [7:50:09<11:19:55, 2.30s/it] 41%|████ | 12243/30000 [7:50:11<11:16:06, 2.28s/it] 41%|████ | 12244/30000 [7:50:14<11:19:23, 2.30s/it] 41%|████ | 12245/30000 [7:50:16<11:19:16, 2.30s/it] 41%|████ | 12246/30000 [7:50:18<11:16:02, 2.28s/it] 41%|████ | 12247/30000 [7:50:20<11:16:50, 2.29s/it] 41%|████ | 12248/30000 [7:50:23<11:21:02, 2.30s/it] 41%|████ | 12249/30000 [7:50:25<11:16:07, 2.29s/it] 41%|████ | 12250/30000 [7:50:27<11:11:43, 2.27s/it] 41%|████ | 12250/30000 [7:50:27<11:11:43, 2.27s/it] 41%|████ | 12251/30000 [7:50:30<11:20:07, 2.30s/it] 41%|████ | 12252/30000 [7:50:32<11:20:37, 2.30s/it] 41%|████ | 12253/30000 [7:50:34<11:17:44, 2.29s/it] 41%|████ | 12254/30000 [7:50:37<11:23:20, 2.31s/it] 41%|████ | 12255/30000 [7:50:39<11:20:11, 2.30s/it] 41%|████ | 12256/30000 [7:50:41<11:17:53, 2.29s/it] 41%|████ | 12257/30000 [7:50:43<11:21:33, 2.30s/it] 41%|████ | 12258/30000 [7:50:46<11:18:30, 2.29s/it] 41%|████ | 12259/30000 [7:50:48<11:16:47, 2.29s/it] 41%|████ | 12260/30000 [7:50:50<11:17:36, 2.29s/it] 41%|████ | 12260/30000 [7:50:50<11:17:36, 2.29s/it] 41%|████ | 12261/30000 [7:50:53<11:19:41, 2.30s/it] 41%|████ | 12262/30000 [7:50:55<11:15:55, 2.29s/it] 41%|████ | 12263/30000 [7:50:57<11:15:15, 2.28s/it] 41%|████ | 12264/30000 [7:51:00<11:20:02, 2.30s/it] 41%|████ | 12265/30000 [7:51:02<11:15:49, 2.29s/it] 41%|████ | 12266/30000 [7:51:04<11:14:40, 2.28s/it] 41%|████ | 12267/30000 [7:51:06<11:20:09, 2.30s/it] 41%|████ | 12268/30000 [7:51:09<11:23:00, 2.31s/it] 41%|████ | 12269/30000 [7:51:11<11:17:23, 2.29s/it] 41%|████ | 12270/30000 [7:51:13<11:17:36, 2.29s/it] 41%|████ | 12270/30000 [7:51:13<11:17:36, 2.29s/it] 41%|████ | 12271/30000 [7:51:16<11:18:17, 2.30s/it] 41%|████ | 12272/30000 [7:51:18<11:15:15, 2.29s/it] 41%|████ | 12273/30000 [7:51:20<11:12:32, 2.28s/it] 41%|████ | 12274/30000 [7:51:22<11:17:52, 2.29s/it] 41%|████ | 12275/30000 [7:51:25<11:15:53, 2.29s/it] 41%|████ | 12276/30000 [7:51:27<11:14:18, 2.28s/it] 41%|████ | 12277/30000 [7:51:29<11:17:45, 2.29s/it] 41%|████ | 12278/30000 [7:51:32<11:20:45, 2.30s/it] 41%|████ | 12279/30000 [7:51:34<11:15:43, 2.29s/it] 41%|████ | 12280/30000 [7:51:36<11:16:08, 2.29s/it] 41%|████ | 12280/30000 [7:51:36<11:16:08, 2.29s/it] 41%|████ | 12281/30000 [7:51:39<11:22:32, 2.31s/it] 41%|████ | 12282/30000 [7:51:41<11:23:11, 2.31s/it] 41%|████ | 12283/30000 [7:51:43<11:19:37, 2.30s/it] 41%|████ | 12284/30000 [7:51:45<11:13:47, 2.28s/it] 41%|████ | 12285/30000 [7:51:48<11:20:38, 2.31s/it] 41%|████ | 12286/30000 [7:51:50<11:22:36, 2.31s/it] 41%|████ | 12287/30000 [7:51:52<11:20:01, 2.30s/it] 41%|████ | 12288/30000 [7:51:55<11:16:16, 2.29s/it] 41%|████ | 12289/30000 [7:51:57<11:20:33, 2.31s/it] 41%|████ | 12290/30000 [7:51:59<11:17:26, 2.30s/it] 41%|████ | 12290/30000 [7:51:59<11:17:26, 2.30s/it] 41%|████ | 12291/30000 [7:52:01<11:16:30, 2.29s/it] 41%|████ | 12292/30000 [7:52:04<11:15:55, 2.29s/it] 41%|████ | 12293/30000 [7:52:06<11:17:07, 2.29s/it] 41%|████ | 12294/30000 [7:52:08<11:13:29, 2.28s/it] 41%|████ | 12295/30000 [7:52:11<11:08:45, 2.27s/it] 41%|████ | 12296/30000 [7:52:13<11:14:16, 2.29s/it] 41%|████ | 12297/30000 [7:52:15<11:16:47, 2.29s/it] 41%|████ | 12298/30000 [7:52:17<11:13:44, 2.28s/it] 41%|████ | 12299/30000 [7:52:20<11:13:33, 2.28s/it] 41%|████ | 12300/30000 [7:52:22<11:21:33, 2.31s/it] 41%|████ | 12300/30000 [7:52:22<11:21:33, 2.31s/it] 41%|████ | 12301/30000 [7:52:24<11:18:56, 2.30s/it] 41%|████ | 12302/30000 [7:52:27<11:14:45, 2.29s/it] 41%|████ | 12303/30000 [7:52:29<11:14:41, 2.29s/it] 41%|████ | 12304/30000 [7:52:31<11:19:59, 2.31s/it] 41%|████ | 12305/30000 [7:52:34<11:20:52, 2.31s/it] 41%|████ | 12306/30000 [7:52:36<11:15:55, 2.29s/it] 41%|████ | 12307/30000 [7:52:38<11:15:22, 2.29s/it] 41%|████ | 12308/30000 [7:52:40<11:18:05, 2.30s/it] 41%|████ | 12309/30000 [7:52:43<11:14:40, 2.29s/it] 41%|████ | 12310/30000 [7:52:45<11:13:26, 2.28s/it] 41%|████ | 12310/30000 [7:52:45<11:13:26, 2.28s/it] 41%|████ | 12311/30000 [7:52:47<11:15:46, 2.29s/it] 41%|████ | 12312/30000 [7:52:50<11:15:46, 2.29s/it] 41%|████ | 12313/30000 [7:52:52<11:14:50, 2.29s/it] 41%|████ | 12314/30000 [7:52:54<11:12:01, 2.28s/it] 41%|████ | 12315/30000 [7:52:56<11:17:59, 2.30s/it] 41%|████ | 12316/30000 [7:52:59<11:21:14, 2.31s/it] 41%|████ | 12317/30000 [7:53:01<11:17:50, 2.30s/it] 41%|████ | 12318/30000 [7:53:03<11:16:54, 2.30s/it] 41%|████ | 12319/30000 [7:53:06<11:22:07, 2.31s/it] 41%|████ | 12320/30000 [7:53:08<11:19:59, 2.31s/it] 41%|████ | 12320/30000 [7:53:08<11:19:59, 2.31s/it] 41%|████ | 12321/30000 [7:53:10<11:17:02, 2.30s/it] 41%|████ | 12322/30000 [7:53:13<11:09:56, 2.27s/it] 41%|████ | 12323/30000 [7:53:15<11:15:11, 2.29s/it] 41%|████ | 12324/30000 [7:53:17<11:17:47, 2.30s/it] 41%|████ | 12325/30000 [7:53:19<11:14:45, 2.29s/it] 41%|████ | 12326/30000 [7:53:22<11:16:06, 2.30s/it] 41%|████ | 12327/30000 [7:53:24<11:16:23, 2.30s/it] 41%|████ | 12328/30000 [7:53:26<11:12:31, 2.28s/it] 41%|████ | 12329/30000 [7:53:29<11:12:14, 2.28s/it] 41%|████ | 12330/30000 [7:53:31<11:15:05, 2.29s/it] 41%|████ | 12330/30000 [7:53:31<11:15:05, 2.29s/it] 41%|████ | 12331/30000 [7:53:33<11:14:28, 2.29s/it] 41%|████ | 12332/30000 [7:53:35<11:11:11, 2.28s/it] 41%|████ | 12333/30000 [7:53:38<11:10:56, 2.28s/it] 41%|████ | 12334/30000 [7:53:40<11:21:00, 2.31s/it] 41%|████ | 12335/30000 [7:53:42<11:22:41, 2.32s/it] 41%|████ | 12336/30000 [7:53:45<11:19:51, 2.31s/it] 41%|████ | 12337/30000 [7:53:47<11:11:23, 2.28s/it] 41%|████ | 12338/30000 [7:53:49<11:14:18, 2.29s/it] 41%|████ | 12339/30000 [7:53:51<11:09:39, 2.28s/it] 41%|████ | 12340/30000 [7:53:54<11:10:40, 2.28s/it] 41%|████ | 12340/30000 [7:53:54<11:10:40, 2.28s/it] 41%|████ | 12341/30000 [7:53:56<11:19:20, 2.31s/it] 41%|████ | 12342/30000 [7:53:58<11:19:13, 2.31s/it] 41%|████ | 12343/30000 [7:54:01<11:16:22, 2.30s/it] 41%|████ | 12344/30000 [7:54:03<11:11:10, 2.28s/it] 41%|████ | 12345/30000 [7:54:05<11:18:59, 2.31s/it] 41%|████ | 12346/30000 [7:54:08<11:20:31, 2.31s/it] 41%|████ | 12347/30000 [7:54:10<11:15:19, 2.30s/it] 41%|████ | 12348/30000 [7:54:12<11:16:29, 2.30s/it] 41%|████ | 12349/30000 [7:54:15<11:20:22, 2.31s/it] 41%|████ | 12350/30000 [7:54:17<11:21:00, 2.32s/it] 41%|████ | 12350/30000 [7:54:17<11:21:00, 2.32s/it] 41%|████ | 12351/30000 [7:54:19<11:18:39, 2.31s/it] 41%|████ | 12352/30000 [7:54:22<11:20:53, 2.31s/it] 41%|████ | 12353/30000 [7:54:24<11:20:05, 2.31s/it] 41%|████ | 12354/30000 [7:54:26<11:13:19, 2.29s/it] 41%|████ | 12355/30000 [7:54:28<11:12:52, 2.29s/it] 41%|████ | 12356/30000 [7:54:31<11:21:15, 2.32s/it] 41%|████ | 12357/30000 [7:54:33<11:22:08, 2.32s/it] 41%|████ | 12358/30000 [7:54:35<11:18:59, 2.31s/it] 41%|████ | 12359/30000 [7:54:38<11:09:56, 2.28s/it] 41%|████ | 12360/30000 [7:54:40<11:13:00, 2.29s/it] 41%|████ | 12360/30000 [7:54:40<11:13:00, 2.29s/it] 41%|████ | 12361/30000 [7:54:42<11:07:53, 2.27s/it] 41%|████ | 12362/30000 [7:54:44<11:09:11, 2.28s/it] 41%|████ | 12363/30000 [7:54:47<11:14:33, 2.29s/it] 41%|████ | 12364/30000 [7:54:49<11:17:51, 2.31s/it] 41%|████ | 12365/30000 [7:54:51<11:15:28, 2.30s/it] 41%|████ | 12366/30000 [7:54:54<11:17:27, 2.31s/it] 41%|████ | 12367/30000 [7:54:56<11:22:48, 2.32s/it] 41%|████ | 12368/30000 [7:54:58<11:18:57, 2.31s/it] 41%|████ | 12369/30000 [7:55:01<11:10:28, 2.28s/it] 41%|████ | 12370/30000 [7:55:03<11:16:24, 2.30s/it] 41%|████ | 12370/30000 [7:55:03<11:16:24, 2.30s/it] 41%|████ | 12371/30000 [7:55:05<11:18:06, 2.31s/it] 41%|████ | 12372/30000 [7:55:07<11:12:46, 2.29s/it] 41%|████ | 12373/30000 [7:55:10<11:18:25, 2.31s/it] 41%|████ | 12374/30000 [7:55:12<11:21:50, 2.32s/it] 41%|████▏ | 12375/30000 [7:55:14<11:17:02, 2.30s/it] 41%|████▏ | 12376/30000 [7:55:17<11:14:36, 2.30s/it] 41%|████▏ | 12377/30000 [7:55:19<11:18:21, 2.31s/it] 41%|████▏ | 12378/30000 [7:55:21<11:14:12, 2.30s/it] 41%|████▏ | 12379/30000 [7:55:24<11:11:21, 2.29s/it] 41%|████▏ | 12380/30000 [7:55:26<11:22:05, 2.32s/it] 41%|████▏ | 12380/30000 [7:55:26<11:22:05, 2.32s/it] 41%|████▏ | 12381/30000 [7:55:28<11:22:01, 2.32s/it] 41%|████▏ | 12382/30000 [7:55:31<11:16:14, 2.30s/it] 41%|████▏ | 12383/30000 [7:55:33<11:16:14, 2.30s/it] 41%|████▏ | 12384/30000 [7:55:35<11:19:00, 2.31s/it] 41%|████▏ | 12385/30000 [7:55:37<11:15:43, 2.30s/it] 41%|████▏ | 12386/30000 [7:55:40<11:10:44, 2.28s/it] 41%|████▏ | 12387/30000 [7:55:42<11:17:07, 2.31s/it] 41%|████▏ | 12388/30000 [7:55:44<11:19:51, 2.32s/it] 41%|████▏ | 12389/30000 [7:55:47<11:18:54, 2.31s/it] 41%|████▏ | 12390/30000 [7:55:49<11:20:28, 2.32s/it] 41%|████▏ | 12390/30000 [7:55:49<11:20:28, 2.32s/it] 41%|████▏ | 12391/30000 [7:55:51<11:17:31, 2.31s/it] 41%|████▏ | 12392/30000 [7:55:54<11:11:38, 2.29s/it] 41%|████▏ | 12393/30000 [7:55:56<11:12:02, 2.29s/it] 41%|████▏ | 12394/30000 [7:55:58<11:12:45, 2.29s/it] 41%|████▏ | 12395/30000 [7:56:00<11:09:21, 2.28s/it] 41%|████▏ | 12396/30000 [7:56:03<11:09:36, 2.28s/it] 41%|████▏ | 12397/30000 [7:56:05<11:16:27, 2.31s/it] 41%|████▏ | 12398/30000 [7:56:07<11:17:15, 2.31s/it] 41%|████▏ | 12399/30000 [7:56:10<11:15:51, 2.30s/it] 41%|████▏ | 12400/30000 [7:56:12<11:17:34, 2.31s/it] 41%|████▏ | 12400/30000 [7:56:12<11:17:34, 2.31s/it] 41%|████▏ | 12401/30000 [7:56:14<11:16:09, 2.31s/it] 41%|████▏ | 12402/30000 [7:56:17<11:14:12, 2.30s/it] 41%|████▏ | 12403/30000 [7:56:19<11:17:25, 2.31s/it] 41%|████▏ | 12404/30000 [7:56:21<11:17:49, 2.31s/it] 41%|████▏ | 12405/30000 [7:56:23<11:14:20, 2.30s/it] 41%|████▏ | 12406/30000 [7:56:26<11:12:02, 2.29s/it] 41%|████▏ | 12407/30000 [7:56:28<11:16:23, 2.31s/it] 41%|████▏ | 12408/30000 [7:56:30<11:12:39, 2.29s/it] 41%|████▏ | 12409/30000 [7:56:33<11:11:30, 2.29s/it] 41%|████▏ | 12410/30000 [7:56:35<11:17:03, 2.31s/it] 41%|████▏ | 12410/30000 [7:56:35<11:17:03, 2.31s/it] 41%|████▏ | 12411/30000 [7:56:37<11:16:08, 2.31s/it] 41%|████▏ | 12412/30000 [7:56:40<11:13:29, 2.30s/it] 41%|████▏ | 12413/30000 [7:56:42<11:19:19, 2.32s/it] 41%|████▏ | 12414/30000 [7:56:44<11:21:12, 2.32s/it] 41%|████▏ | 12415/30000 [7:56:47<11:16:48, 2.31s/it] 41%|████▏ | 12416/30000 [7:56:49<11:17:10, 2.31s/it] 41%|████▏ | 12417/30000 [7:56:51<11:19:12, 2.32s/it] 41%|████▏ | 12418/30000 [7:56:53<11:15:10, 2.30s/it] 41%|████▏ | 12419/30000 [7:56:56<11:12:07, 2.29s/it] 41%|████▏ | 12420/30000 [7:56:58<11:18:23, 2.32s/it] 41%|████▏ | 12420/30000 [7:56:58<11:18:23, 2.32s/it] 41%|████▏ | 12421/30000 [7:57:00<11:20:34, 2.32s/it] 41%|████▏ | 12422/30000 [7:57:03<11:16:39, 2.31s/it] 41%|████▏ | 12423/30000 [7:57:05<11:20:00, 2.32s/it] 41%|████▏ | 12424/30000 [7:57:07<11:17:23, 2.31s/it] 41%|████▏ | 12425/30000 [7:57:10<11:14:20, 2.30s/it] 41%|████▏ | 12426/30000 [7:57:12<11:11:49, 2.29s/it] 41%|████▏ | 12427/30000 [7:57:14<11:14:36, 2.30s/it] 41%|████▏ | 12428/30000 [7:57:17<11:14:24, 2.30s/it] 41%|████▏ | 12429/30000 [7:57:19<11:10:33, 2.29s/it] 41%|████▏ | 12430/30000 [7:57:21<11:17:49, 2.31s/it] 41%|████▏ | 12430/30000 [7:57:21<11:17:49, 2.31s/it] 41%|████▏ | 12431/30000 [7:57:23<11:17:58, 2.32s/it] 41%|████▏ | 12432/30000 [7:57:26<11:12:52, 2.30s/it] 41%|████▏ | 12433/30000 [7:57:28<11:18:00, 2.32s/it] 41%|████▏ | 12434/30000 [7:57:30<11:18:30, 2.32s/it] 41%|████▏ | 12435/30000 [7:57:33<11:13:39, 2.30s/it] 41%|████▏ | 12436/30000 [7:57:35<11:18:52, 2.32s/it] 41%|████▏ | 12437/30000 [7:57:37<11:17:36, 2.31s/it] 41%|████▏ | 12438/30000 [7:57:40<11:13:58, 2.30s/it] 41%|████▏ | 12439/30000 [7:57:42<11:15:19, 2.31s/it] 41%|████▏ | 12440/30000 [7:57:44<11:19:21, 2.32s/it] 41%|████▏ | 12440/30000 [7:57:44<11:19:21, 2.32s/it] 41%|████▏ | 12441/30000 [7:57:47<11:15:49, 2.31s/it] 41%|████▏ | 12442/30000 [7:57:49<11:12:47, 2.30s/it] 41%|████▏ | 12443/30000 [7:57:51<11:18:26, 2.32s/it] 41%|████▏ | 12444/30000 [7:57:54<11:16:54, 2.31s/it] 41%|████▏ | 12445/30000 [7:57:56<11:14:16, 2.30s/it] 41%|████▏ | 12446/30000 [7:57:58<11:12:58, 2.30s/it] 41%|████▏ | 12447/30000 [7:58:00<11:15:09, 2.31s/it] 41%|████▏ | 12448/30000 [7:58:03<11:11:06, 2.29s/it] 41%|████▏ | 12449/30000 [7:58:05<11:12:05, 2.30s/it] 42%|████▏ | 12450/30000 [7:58:07<11:18:19, 2.32s/it] +{'loss': 0.0148, 'grad_norm': 0.07598406821489334, 'learning_rate': 2.095114786233736e-05, 'flow_matching_loss': 0.014846167713403701, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.51} +{'loss': 0.0156, 'grad_norm': 0.07993654161691666, 'learning_rate': 2.0935966560114236e-05, 'flow_matching_loss': 0.015623101498931646, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.51} +{'loss': 0.0148, 'grad_norm': 0.09625408053398132, 'learning_rate': 2.092077804513201e-05, 'flow_matching_loss': 0.014811740815639496, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.51} +{'loss': 0.0154, 'grad_norm': 0.0879182294011116, 'learning_rate': 2.0905582335846167e-05, 'flow_matching_loss': 0.015382303670048714, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.51} +{'loss': 0.0157, 'grad_norm': 0.08180984854698181, 'learning_rate': 2.0890379450720927e-05, 'flow_matching_loss': 0.015659973304718734, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.52} +{'loss': 0.0155, 'grad_norm': 0.085471510887146, 'learning_rate': 2.087516940822923e-05, 'flow_matching_loss': 0.015476400032639503, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.52} +{'loss': 0.0157, 'grad_norm': 0.07904262095689774, 'learning_rate': 2.085995222685271e-05, 'flow_matching_loss': 0.015680381935089825, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.52} +{'loss': 0.0145, 'grad_norm': 0.06339608877897263, 'learning_rate': 2.084472792508167e-05, 'flow_matching_loss': 0.01445759292691946, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.53} +{'loss': 0.0149, 'grad_norm': 0.07699710875749588, 'learning_rate': 2.082949652141508e-05, 'flow_matching_loss': 0.014933250844478607, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.53} +{'loss': 0.0148, 'grad_norm': 0.07096560299396515, 'learning_rate': 2.081425803436054e-05, 'flow_matching_loss': 0.014755536336451769, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.53} +{'loss': 0.0159, 'grad_norm': 0.09391763061285019, 'learning_rate': 2.079901248243423e-05, 'flow_matching_loss': 0.015896176639944314, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.54} +{'loss': 0.0165, 'grad_norm': 0.06897689402103424, 'learning_rate': 2.078375988416095e-05, 'flow_matching_loss': 0.016461908631026746, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.54} +{'loss': 0.0143, 'grad_norm': 0.07040475308895111, 'learning_rate': 2.076850025807404e-05, 'flow_matching_loss': 0.01433317419141531, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.54} +{'loss': 0.0166, 'grad_norm': 0.06450559198856354, 'learning_rate': 2.0753233622715374e-05, 'flow_matching_loss': 0.016563307587057353, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.54} +{'loss': 0.0139, 'grad_norm': 0.05456933379173279, 'learning_rate': 2.073795999663536e-05, 'flow_matching_loss': 0.01387160662561655, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.55} +{'loss': 0.0151, 'grad_norm': 0.062114790081977844, 'learning_rate': 2.07226793983929e-05, 'flow_matching_loss': 0.015134937688708305, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.55} +{'loss': 0.0171, 'grad_norm': 0.07330848276615143, 'learning_rate': 2.070739184655536e-05, 'flow_matching_loss': 0.017116803489625455, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.55} +{'loss': 0.0163, 'grad_norm': 0.08307880908250809, 'learning_rate': 2.069209735969855e-05, 'flow_matching_loss': 0.016279212012887, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.56} +{'loss': 0.0167, 'grad_norm': 0.10090626776218414, 'learning_rate': 2.067679595640672e-05, 'flow_matching_loss': 0.016737996879965068, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.56} +{'loss': 0.0146, 'grad_norm': 0.0924069732427597, 'learning_rate': 2.066148765527252e-05, 'flow_matching_loss': 0.014578737970441579, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.56} +{'loss': 0.0149, 'grad_norm': 0.09856583178043365, 'learning_rate': 2.0646172474896976e-05, 'flow_matching_loss': 0.014851547870784998, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.56} +{'loss': 0.0147, 'grad_norm': 0.08253976702690125, 'learning_rate': 2.063085043388947e-05, 'flow_matching_loss': 0.014652405120432376, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.57} +{'loss': 0.0144, 'grad_norm': 0.08341598510742188, 'learning_rate': 2.0615521550867732e-05, 'flow_matching_loss': 0.01444775452837348, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.57} +{'loss': 0.0143, 'grad_norm': 0.07259522378444672, 'learning_rate': 2.0600185844457803e-05, 'flow_matching_loss': 0.014345444086939097, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.57} +{'loss': 0.015, 'grad_norm': 0.06408794969320297, 'learning_rate': 2.058484333329401e-05, 'flow_matching_loss': 0.014955624379217625, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.58} +{'loss': 0.0156, 'grad_norm': 0.07086075842380524, 'learning_rate': 2.0569494036018955e-05, 'flow_matching_loss': 0.015570421051234008, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.58} +{'loss': 0.0144, 'grad_norm': 0.09264744818210602, 'learning_rate': 2.055413797128347e-05, 'flow_matching_loss': 0.014438960142433644, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.58} +{'loss': 0.0141, 'grad_norm': 0.07077410817146301, 'learning_rate': 2.0538775157746634e-05, 'flow_matching_loss': 0.014104956574738026, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.58} +{'loss': 0.0137, 'grad_norm': 0.08152949810028076, 'learning_rate': 2.0523405614075706e-05, 'flow_matching_loss': 0.013657112792134284, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.59} +{'loss': 0.0147, 'grad_norm': 0.05459417402744293, 'learning_rate': 2.0508029358946132e-05, 'flow_matching_loss': 0.014723737724125385, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.59} +{'loss': 0.015, 'grad_norm': 0.06143946945667267, 'learning_rate': 2.0492646411041515e-05, 'flow_matching_loss': 0.014982997626066207, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.59} +{'loss': 0.016, 'grad_norm': 0.10025094449520111, 'learning_rate': 2.0477256789053587e-05, 'flow_matching_loss': 0.01595185389742255, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.6} +{'loss': 0.0143, 'grad_norm': 0.08835582435131073, 'learning_rate': 2.0461860511682194e-05, 'flow_matching_loss': 0.014266297779977321, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.6} +{'loss': 0.0149, 'grad_norm': 0.09285423159599304, 'learning_rate': 2.0446457597635254e-05, 'flow_matching_loss': 0.014878789894282818, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.6} +{'loss': 0.0151, 'grad_norm': 0.08217331022024155, 'learning_rate': 2.0431048065628763e-05, 'flow_matching_loss': 0.015094347018748521, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.6} +{'loss': 0.0157, 'grad_norm': 0.0708748996257782, 'learning_rate': 2.0415631934386764e-05, 'flow_matching_loss': 0.015729272831231357, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.61} +{'loss': 0.0158, 'grad_norm': 0.07710058987140656, 'learning_rate': 2.04002092226413e-05, 'flow_matching_loss': 0.01575240436941385, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.61} +{'loss': 0.0149, 'grad_norm': 0.07109581679105759, 'learning_rate': 2.038477994913242e-05, 'flow_matching_loss': 0.014913765713572502, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.61} +{'loss': 0.0164, 'grad_norm': 0.07884085178375244, 'learning_rate': 2.036934413260816e-05, 'flow_matching_loss': 0.016417440492659807, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.62} +{'loss': 0.016, 'grad_norm': 0.09533672779798508, 'learning_rate': 2.0353901791824473e-05, 'flow_matching_loss': 0.01597119215875864, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.62} +{'loss': 0.0148, 'grad_norm': 0.0970696434378624, 'learning_rate': 2.0338452945545275e-05, 'flow_matching_loss': 0.014754175208508968, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.62} 42%|████▏ | 12450/30000 [7:58:07<11:18:19, 2.32s/it] 42%|████▏ | 12451/30000 [7:58:10<11:19:43, 2.32s/it] 42%|████▏ | 12452/30000 [7:58:12<11:19:01, 2.32s/it] 42%|████▏ | 12453/30000 [7:58:14<11:16:09, 2.31s/it] 42%|████▏ | 12454/30000 [7:58:17<11:19:35, 2.32s/it] 42%|████▏ | 12455/30000 [7:58:19<11:13:46, 2.30s/it] 42%|████▏ | 12456/30000 [7:58:21<11:11:57, 2.30s/it] 42%|████▏ | 12457/30000 [7:58:24<11:12:41, 2.30s/it] 42%|████▏ | 12458/30000 [7:58:26<11:13:41, 2.30s/it] 42%|████▏ | 12459/30000 [7:58:28<11:08:47, 2.29s/it] 42%|████▏ | 12460/30000 [7:58:30<11:08:49, 2.29s/it] 42%|████▏ | 12460/30000 [7:58:30<11:08:49, 2.29s/it] 42%|████▏ | 12461/30000 [7:58:33<11:14:58, 2.31s/it] 42%|████▏ | 12462/30000 [7:58:35<11:13:33, 2.30s/it] 42%|████▏ | 12463/30000 [7:58:37<11:06:45, 2.28s/it] 42%|████▏ | 12464/30000 [7:58:40<11:05:46, 2.28s/it] 42%|████▏ | 12465/30000 [7:58:42<11:07:03, 2.28s/it] 42%|████▏ | 12466/30000 [7:58:44<11:06:44, 2.28s/it] 42%|████▏ | 12467/30000 [7:58:46<11:03:53, 2.27s/it] 42%|████▏ | 12468/30000 [7:58:49<11:11:28, 2.30s/it] 42%|████▏ | 12469/30000 [7:58:51<11:10:40, 2.30s/it] 42%|████▏ | 12470/30000 [7:58:53<11:09:49, 2.29s/it] 42%|████▏ | 12470/30000 [7:58:53<11:09:49, 2.29s/it] 42%|████▏ | 12471/30000 [7:58:56<11:11:48, 2.30s/it] 42%|████▏ | 12472/30000 [7:58:58<11:15:07, 2.31s/it] 42%|████▏ | 12473/30000 [7:59:00<11:14:28, 2.31s/it] 42%|████▏ | 12474/30000 [7:59:02<11:08:56, 2.29s/it] 42%|████▏ | 12475/30000 [7:59:05<11:07:35, 2.29s/it] 42%|████▏ | 12476/30000 [7:59:07<11:14:02, 2.31s/it] 42%|████▏ | 12477/30000 [7:59:09<11:15:42, 2.31s/it] 42%|████▏ | 12478/30000 [7:59:12<11:12:43, 2.30s/it] 42%|████▏ | 12479/30000 [7:59:14<11:11:51, 2.30s/it] 42%|████▏ | 12480/30000 [7:59:16<11:12:27, 2.30s/it] 42%|████▏ | 12480/30000 [7:59:16<11:12:27, 2.30s/it] 42%|████▏ | 12481/30000 [7:59:19<11:11:28, 2.30s/it] 42%|████▏ | 12482/30000 [7:59:21<11:07:33, 2.29s/it] 42%|████▏ | 12483/30000 [7:59:23<11:12:44, 2.30s/it] 42%|████▏ | 12484/30000 [7:59:26<11:15:55, 2.32s/it] 42%|████▏ | 12485/30000 [7:59:28<11:12:50, 2.30s/it] 42%|████▏ | 12486/30000 [7:59:30<11:11:37, 2.30s/it] 42%|████▏ | 12487/30000 [7:59:32<11:14:01, 2.31s/it] 42%|████▏ | 12488/30000 [7:59:35<11:06:55, 2.29s/it] 42%|████▏ | 12489/30000 [7:59:37<11:05:27, 2.28s/it] 42%|████▏ | 12490/30000 [7:59:39<11:08:55, 2.29s/it] 42%|████▏ | 12490/30000 [7:59:39<11:08:55, 2.29s/it] 42%|████▏ | 12491/30000 [7:59:42<11:08:29, 2.29s/it] 42%|████▏ | 12492/30000 [7:59:44<11:10:06, 2.30s/it] 42%|████▏ | 12493/30000 [7:59:46<11:09:15, 2.29s/it] 42%|████▏ | 12494/30000 [7:59:48<11:14:04, 2.31s/it] 42%|████▏ | 12495/30000 [7:59:51<11:14:35, 2.31s/it] 42%|████▏ | 12496/30000 [7:59:53<11:09:54, 2.30s/it] 42%|████▏ | 12497/30000 [7:59:55<11:10:24, 2.30s/it] 42%|████▏ | 12498/30000 [7:59:58<11:11:31, 2.30s/it] 42%|████▏ | 12499/30000 [8:00:00<11:06:27, 2.28s/it] 42%|████▏ | 12500/30000 [8:00:02<11:05:47, 2.28s/it] 42%|████▏ | 12500/30000 [8:00:02<11:05:47, 2.28s/it] 42%|████▏ | 12501/30000 [8:00:05<11:08:22, 2.29s/it] 42%|████▏ | 12502/30000 [8:00:07<11:13:48, 2.31s/it] 42%|████▏ | 12503/30000 [8:00:09<11:07:41, 2.29s/it] 42%|████▏ | 12504/30000 [8:00:11<11:04:06, 2.28s/it] 42%|████▏ | 12505/30000 [8:00:14<11:09:19, 2.30s/it] 42%|████▏ | 12506/30000 [8:00:16<11:10:31, 2.30s/it] 42%|████▏ | 12507/30000 [8:00:18<11:10:02, 2.30s/it] 42%|████▏ | 12508/30000 [8:00:21<11:08:23, 2.29s/it] 42%|████▏ | 12509/30000 [8:00:23<11:13:50, 2.31s/it] 42%|████▏ | 12510/30000 [8:00:25<11:17:16, 2.32s/it] 42%|████▏ | 12510/30000 [8:00:25<11:17:16, 2.32s/it] 42%|████▏ | 12511/30000 [8:00:28<11:12:24, 2.31s/it] 42%|████▏ | 12512/30000 [8:00:30<11:09:54, 2.30s/it] 42%|████▏ | 12513/30000 [8:00:32<11:13:35, 2.31s/it] 42%|████▏ | 12514/30000 [8:00:34<11:13:04, 2.31s/it] 42%|████▏ | 12515/30000 [8:00:37<11:10:01, 2.30s/it] 42%|████▏ | 12516/30000 [8:00:39<11:11:40, 2.31s/it] 42%|████▏ | 12517/30000 [8:00:41<11:15:02, 2.32s/it] 42%|████▏ | 12518/30000 [8:00:44<11:10:16, 2.30s/it] 42%|████▏ | 12519/30000 [8:00:46<11:08:16, 2.29s/it] 42%|████▏ | 12520/30000 [8:00:48<11:11:38, 2.31s/it] 42%|████▏ | 12520/30000 [8:00:48<11:11:38, 2.31s/it] 42%|████▏ | 12521/30000 [8:00:51<11:08:48, 2.30s/it] 42%|████▏ | 12522/30000 [8:00:53<11:03:53, 2.28s/it] 42%|████▏ | 12523/30000 [8:00:55<11:03:40, 2.28s/it] 42%|████▏ | 12524/30000 [8:00:57<11:08:29, 2.30s/it] 42%|████▏ | 12525/30000 [8:01:00<11:11:08, 2.30s/it] 42%|████▏ | 12526/30000 [8:01:02<11:06:45, 2.29s/it] 42%|████▏ | 12527/30000 [8:01:04<11:00:13, 2.27s/it] 42%|████▏ | 12528/30000 [8:01:07<11:04:24, 2.28s/it] 42%|████▏ | 12529/30000 [8:01:09<11:02:15, 2.27s/it] 42%|████▏ | 12530/30000 [8:01:11<10:59:54, 2.27s/it] 42%|████▏ | 12530/30000 [8:01:11<10:59:54, 2.27s/it] 42%|████▏ | 12531/30000 [8:01:13<11:07:18, 2.29s/it] 42%|████▏ | 12532/30000 [8:01:16<11:08:50, 2.30s/it] 42%|████▏ | 12533/30000 [8:01:18<11:05:06, 2.28s/it] 42%|████▏ | 12534/30000 [8:01:20<11:02:58, 2.28s/it] 42%|████▏ | 12535/30000 [8:01:23<11:07:57, 2.29s/it] 42%|████▏ | 12536/30000 [8:01:25<11:02:43, 2.28s/it] 42%|████▏ | 12537/30000 [8:01:27<10:59:54, 2.27s/it] 42%|████▏ | 12538/30000 [8:01:29<11:07:31, 2.29s/it] 42%|████▏ | 12539/30000 [8:01:32<11:08:33, 2.30s/it] 42%|████▏ | 12540/30000 [8:01:34<11:06:25, 2.29s/it] 42%|████▏ | 12540/30000 [8:01:34<11:06:25, 2.29s/it] 42%|████▏ | 12541/30000 [8:01:36<11:10:41, 2.30s/it] 42%|████▏ | 12542/30000 [8:01:39<11:10:42, 2.31s/it] 42%|████▏ | 12543/30000 [8:01:41<11:06:00, 2.29s/it] 42%|████▏ | 12544/30000 [8:01:43<11:04:01, 2.28s/it] 42%|████▏ | 12545/30000 [8:01:45<11:08:57, 2.30s/it] 42%|████▏ | 12546/30000 [8:01:48<11:06:51, 2.29s/it] 42%|████▏ | 12547/30000 [8:01:50<11:06:03, 2.29s/it] 42%|████▏ | 12548/30000 [8:01:52<11:12:32, 2.31s/it] 42%|████▏ | 12549/30000 [8:01:55<11:10:06, 2.30s/it] 42%|████▏ | 12550/30000 [8:01:57<11:07:00, 2.29s/it] 42%|████▏ | 12550/30000 [8:01:57<11:07:00, 2.29s/it] 42%|████▏ | 12551/30000 [8:01:59<11:04:09, 2.28s/it] 42%|████▏ | 12552/30000 [8:02:02<11:07:37, 2.30s/it] 42%|████▏ | 12553/30000 [8:02:04<11:12:18, 2.31s/it] 42%|████▏ | 12554/30000 [8:02:06<11:09:34, 2.30s/it] 42%|████▏ | 12555/30000 [8:02:08<11:11:05, 2.31s/it] 42%|████▏ | 12556/30000 [8:02:11<11:11:19, 2.31s/it] 42%|████▏ | 12557/30000 [8:02:13<11:05:48, 2.29s/it] 42%|████▏ | 12558/30000 [8:02:15<11:06:39, 2.29s/it] 42%|████▏ | 12559/30000 [8:02:18<11:11:54, 2.31s/it] 42%|████▏ | 12560/30000 [8:02:20<11:13:01, 2.32s/it] 42%|████▏ | 12560/30000 [8:02:20<11:13:01, 2.32s/it] 42%|████▏ | 12561/30000 [8:02:22<11:10:19, 2.31s/it] 42%|████▏ | 12562/30000 [8:02:25<11:11:35, 2.31s/it] 42%|████▏ | 12563/30000 [8:02:27<11:10:11, 2.31s/it] 42%|████▏ | 12564/30000 [8:02:29<11:06:52, 2.29s/it] 42%|████▏ | 12565/30000 [8:02:32<11:08:30, 2.30s/it] 42%|████▏ | 12566/30000 [8:02:34<11:12:36, 2.31s/it] 42%|████▏ | 12567/30000 [8:02:36<11:13:23, 2.32s/it] 42%|████▏ | 12568/30000 [8:02:38<11:07:47, 2.30s/it] 42%|████▏ | 12569/30000 [8:02:41<11:11:08, 2.31s/it] 42%|████▏ | 12570/30000 [8:02:43<11:07:59, 2.30s/it] 42%|████▏ | 12570/30000 [8:02:43<11:07:59, 2.30s/it] 42%|████▏ | 12571/30000 [8:02:45<11:07:12, 2.30s/it] 42%|████▏ | 12572/30000 [8:02:48<11:06:56, 2.30s/it] 42%|████▏ | 12573/30000 [8:02:50<11:09:50, 2.31s/it] 42%|████▏ | 12574/30000 [8:02:52<11:04:28, 2.29s/it] 42%|████▏ | 12575/30000 [8:02:54<11:00:48, 2.28s/it] 42%|████▏ | 12576/30000 [8:02:57<11:05:23, 2.29s/it] 42%|████▏ | 12577/30000 [8:02:59<11:02:48, 2.28s/it] 42%|████▏ | 12578/30000 [8:03:01<10:56:39, 2.26s/it] 42%|████▏ | 12579/30000 [8:03:04<11:03:49, 2.29s/it] 42%|████▏ | 12580/30000 [8:03:06<11:01:06, 2.28s/it] 42%|████▏ | 12580/30000 [8:03:06<11:01:06, 2.28s/it] 42%|████▏ | 12581/30000 [8:03:08<10:59:56, 2.27s/it] 42%|████▏ | 12582/30000 [8:03:10<11:06:59, 2.30s/it] 42%|████▏ | 12583/30000 [8:03:13<11:08:42, 2.30s/it] 42%|████▏ | 12584/30000 [8:03:15<11:07:49, 2.30s/it] 42%|████▏ | 12585/30000 [8:03:17<11:11:25, 2.31s/it] 42%|████▏ | 12586/30000 [8:03:20<11:10:02, 2.31s/it] 42%|████▏ | 12587/30000 [8:03:22<11:06:31, 2.30s/it] 42%|████▏ | 12588/30000 [8:03:24<11:14:11, 2.32s/it] 42%|████▏ | 12589/30000 [8:03:27<11:14:18, 2.32s/it] 42%|████▏ | 12590/30000 [8:03:29<11:11:29, 2.31s/it] 42%|████▏ | 12590/30000 [8:03:29<11:11:29, 2.31s/it] 42%|████▏ | 12591/30000 [8:03:31<11:09:17, 2.31s/it] 42%|████▏ | 12592/30000 [8:03:34<11:07:24, 2.30s/it] 42%|████▏ | 12593/30000 [8:03:36<11:06:03, 2.30s/it] 42%|████▏ | 12594/30000 [8:03:38<11:04:29, 2.29s/it] 42%|████▏ | 12595/30000 [8:03:40<11:08:10, 2.30s/it] 42%|████▏ | 12596/30000 [8:03:43<11:08:56, 2.31s/it] 42%|████▏ | 12597/30000 [8:03:45<11:08:10, 2.30s/it] 42%|████▏ | 12598/30000 [8:03:47<11:16:44, 2.33s/it] 42%|████▏ | 12599/30000 [8:03:50<11:07:29, 2.30s/it] 42%|████▏ | 12600/30000 [8:03:52<11:04:14, 2.29s/it] 42%|████▏ | 12600/30000 [8:03:52<11:04:14, 2.29s/it] 42%|████▏ | 12601/30000 [8:03:54<11:08:06, 2.30s/it] 42%|████▏ | 12602/30000 [8:03:57<11:05:33, 2.30s/it] 42%|████▏ | 12603/30000 [8:03:59<11:05:10, 2.29s/it] 42%|████▏ | 12604/30000 [8:04:01<11:11:45, 2.32s/it] 42%|████▏ | 12605/30000 [8:04:04<11:12:04, 2.32s/it] 42%|████▏ | 12606/30000 [8:04:06<11:08:27, 2.31s/it] 42%|████▏ | 12607/30000 [8:04:08<11:10:48, 2.31s/it] 42%|████▏ | 12608/30000 [8:04:10<11:08:46, 2.31s/it] 42%|████▏ | 12609/30000 [8:04:13<11:07:23, 2.30s/it] 42%|████▏ | 12610/30000 [8:04:15<11:07:48, 2.30s/it] 42%|████▏ | 12610/30000 [8:04:15<11:07:48, 2.30s/it] 42%|████▏ | 12611/30000 [8:04:17<11:05:06, 2.29s/it] 42%|████▏ | 12612/30000 [8:04:20<11:03:45, 2.29s/it] 42%|████▏ | 12613/30000 [8:04:22<11:08:20, 2.31s/it] 42%|████▏ | 12614/30000 [8:04:24<11:09:45, 2.31s/it] 42%|████▏ | 12615/30000 [8:04:27<11:05:46, 2.30s/it] 42%|████▏ | 12616/30000 [8:04:29<11:02:14, 2.29s/it] 42%|████▏ | 12617/30000 [8:04:31<11:06:39, 2.30s/it] 42%|████▏ | 12618/30000 [8:04:33<11:04:01, 2.29s/it] 42%|████▏ | 12619/30000 [8:04:36<11:02:00, 2.29s/it] 42%|████▏ | 12620/30000 [8:04:38<11:09:44, 2.31s/it] 42%|████▏ | 12620/30000 [8:04:38<11:09:44, 2.31s/it] 42%|████▏ | 12621/30000 [8:04:40<11:11:52, 2.32s/it] 42%|████▏ | 12622/30000 [8:04:43<11:07:35, 2.30s/it] 42%|████▏ | 12623/30000 [8:04:45<11:05:08, 2.30s/it] 42%|████▏ | 12624/30000 [8:04:47<11:06:06, 2.30s/it] 42%|████▏ | 12625/30000 [8:04:50<11:02:13, 2.29s/it] 42%|████▏ | 12626/30000 [8:04:52<11:00:13, 2.28s/it] 42%|████▏ | 12627/30000 [8:04:54<11:04:06, 2.29s/it] 42%|████▏ | 12628/30000 [8:04:56<11:05:00, 2.30s/it] 42%|████▏ | 12629/30000 [8:04:59<11:03:39, 2.29s/it] 42%|████▏ | 12630/30000 [8:05:01<11:07:53, 2.31s/it] 42%|████▏ | 12630/30000 [8:05:01<11:07:53, 2.31s/it] 42%|████▏ | 12631/30000 [8:05:03<11:08:28, 2.31s/it] 42%|████▏ | 12632/30000 [8:05:06<11:03:22, 2.29s/it] 42%|████▏ | 12633/30000 [8:05:08<11:01:45, 2.29s/it] 42%|████▏ | 12634/30000 [8:05:10<11:07:42, 2.31s/it] 42%|████▏ | 12635/30000 [8:05:12<11:00:52, 2.28s/it] 42%|████▏ | 12636/30000 [8:05:15<11:00:19, 2.28s/it] 42%|████▏ | 12637/30000 [8:05:17<11:07:56, 2.31s/it] 42%|████▏ | 12638/30000 [8:05:19<11:09:41, 2.31s/it] 42%|████▏ | 12639/30000 [8:05:22<11:03:04, 2.29s/it] 42%|████▏ | 12640/30000 [8:05:24<11:01:05, 2.28s/it] 42%|████▏ | 12640/30000 [8:05:24<11:01:05, 2.28s/it] 42%|████▏ | 12641/30000 [8:05:26<11:03:54, 2.29s/it] 42%|████▏ | 12642/30000 [8:05:29<11:01:19, 2.29s/it] 42%|████▏ | 12643/30000 [8:05:31<11:00:00, 2.28s/it] 42%|████▏ | 12644/30000 [8:05:33<11:09:18, 2.31s/it] 42%|████▏ | 12645/30000 [8:05:36<11:11:36, 2.32s/it] 42%|████▏ | 12646/30000 [8:05:38<11:09:12, 2.31s/it] 42%|████▏ | 12647/30000 [8:05:40<11:05:23, 2.30s/it] 42%|████▏ | 12648/30000 [8:05:42<11:02:42, 2.29s/it] 42%|████▏ | 12649/30000 [8:05:45<10:58:20, 2.28s/it] 42%|████▏ | 12650/30000 [8:05:47<10:52:48, 2.26s/it] 42%|████▏ | 12650/30000 [8:05:47<10:52:48, 2.26s/it] 42%|████▏ | 12651/30000 [8:05:49<10:57:30, 2.27s/it] 42%|████▏ | 12652/30000 [8:05:51<10:55:43, 2.27s/it] 42%|████▏ | 12653/30000 [8:05:54<10:58:13, 2.28s/it] 42%|████▏ | 12654/30000 [8:05:56<10:58:36, 2.28s/it] 42%|████▏ | 12655/30000 [8:05:58<10:57:47, 2.28s/it] 42%|████▏ | 12656/30000 [8:06:00<10:53:25, 2.26s/it] 42%|████▏ | 12657/30000 [8:06:03<10:51:56, 2.26s/it] 42%|████▏ | 12658/30000 [8:06:05<11:02:32, 2.29s/it] 42%|████▏ | 12659/30000 [8:06:07<11:08:24, 2.31s/it] 42%|████▏ | 12660/30000 [8:06:10<11:04:42, 2.30s/it] 42%|████▏ | 12660/30000 [8:06:10<11:04:42, 2.30s/it] 42%|████▏ | 12661/30000 [8:06:12<11:03:33, 2.30s/it] 42%|████▏ | 12662/30000 [8:06:14<11:09:16, 2.32s/it] 42%|████▏ | 12663/30000 [8:06:17<11:06:42, 2.31s/it] 42%|████▏ | 12664/30000 [8:06:19<11:01:01, 2.29s/it] 42%|████▏ | 12665/30000 [8:06:21<10:58:30, 2.28s/it] 42%|████▏ | 12666/30000 [8:06:23<11:04:29, 2.30s/it] 42%|████▏ | 12667/30000 [8:06:26<11:05:01, 2.30s/it] 42%|████▏ | 12668/30000 [8:06:28<11:01:19, 2.29s/it] 42%|████▏ | 12669/30000 [8:06:30<10:58:35, 2.28s/it] 42%|████▏ | 12670/30000 [8:06:33<11:05:09, 2.30s/it] 42%|████▏ | 12670/30000 [8:06:33<11:05:09, 2.30s/it] 42%|████▏ | 12671/30000 [8:06:35<11:08:21, 2.31s/it] 42%|████▏ | 12672/30000 [8:06:37<11:03:23, 2.30s/it] 42%|████▏ | 12673/30000 [8:06:40<11:07:21, 2.31s/it] 42%|████▏ | 12674/30000 [8:06:42<11:06:41, 2.31s/it] 42%|████▏ | 12675/30000 [8:06:44<11:02:50, 2.30s/it] 42%|████▏ | 12676/30000 [8:06:46<11:01:39, 2.29s/it] 42%|████▏ | 12677/30000 [8:06:49<11:05:35, 2.31s/it] 42%|████▏ | 12678/30000 [8:06:51<11:06:18, 2.31s/it] 42%|████▏ | 12679/30000 [8:06:53<11:00:56, 2.29s/it] 42%|████▏ | 12680/30000 [8:06:56<10:59:27, 2.28s/it] 42%|████▏ | 12680/30000 [8:06:56<10:59:27, 2.28s/it] 42%|████▏ | 12681/30000 [8:06:58<11:04:58, 2.30s/it] 42%|████▏ | 12682/30000 [8:07:00<11:05:11, 2.30s/it] 42%|████▏ | 12683/30000 [8:07:03<11:01:58, 2.29s/it] 42%|████▏ | 12684/30000 [8:07:05<11:00:50, 2.29s/it] 42%|████▏ | 12685/30000 [8:07:07<11:02:17, 2.29s/it] 42%|████▏ | 12686/30000 [8:07:09<10:58:30, 2.28s/it] 42%|████▏ | 12687/30000 [8:07:12<10:55:19, 2.27s/it] 42%|████▏ | 12688/30000 [8:07:14<11:00:29, 2.29s/it] 42%|████▏ | 12689/30000 [8:07:16<11:01:36, 2.29s/it] 42%|████▏ | 12690/30000 [8:07:19<10:57:22, 2.28s/it] 42%|████▏ | 12690/30000 [8:07:19<10:57:22, 2.28s/it] 42%|████▏ | 12691/30000 [8:07:21<10:58:21, 2.28s/it] 42%|████▏ | 12692/30000 [8:07:23<11:01:55, 2.29s/it] 42%|████▏ | 12693/30000 [8:07:25<10:59:20, 2.29s/it] 42%|████▏ | 12694/30000 [8:07:28<10:57:06, 2.28s/it] 42%|████▏ | 12695/30000 [8:07:30<11:01:22, 2.29s/it] 42%|████▏ | 12696/30000 [8:07:32<10:58:03, 2.28s/it] 42%|████▏ | 12697/30000 [8:07:35<10:58:06, 2.28s/it] 42%|████▏ | 12698/30000 [8:07:37<11:04:09, 2.30s/it] 42%|████▏ | 12699/30000 [8:07:39<11:03:28, 2.30s/it] 42%|████▏ | 12700/30000 [8:07:41<11:02:24, 2.30s/it] 42%|████▏ | 12700/30000 [8:07:41<11:02:24, 2.30s/it] 42%|████▏ | 12701/30000 [8:07:44<11:00:57, 2.29s/it] 42%|████▏ | 12702/30000 [8:07:46<11:06:38, 2.31s/it] 42%|████▏ | 12703/30000 [8:07:48<11:04:41, 2.31s/it] 42%|████▏ | 12704/30000 [8:07:51<10:59:27, 2.29s/it] 42%|████▏ | 12705/30000 [8:07:53<11:04:11, 2.30s/it] 42%|████▏ | 12706/30000 [8:07:55<11:05:45, 2.31s/it] 42%|████▏ | 12707/30000 [8:07:58<11:05:16, 2.31s/it] 42%|████▏ | 12708/30000 [8:08:00<11:02:30, 2.30s/it] 42%|████▏ | 12709/30000 [8:08:02<11:06:05, 2.31s/it] 42%|████▏ | 12710/30000 [8:08:04<11:01:28, 2.30s/it] 42%|████▏ | 12710/30000 [8:08:05<11:01:28, 2.30s/it] 42%|████▏ | 12711/30000 [8:08:07<10:59:58, 2.29s/it] 42%|████▏ | 12712/30000 [8:08:09<11:02:40, 2.30s/it] 42%|████▏ | 12713/30000 [8:08:11<11:02:45, 2.30s/it] 42%|████▏ | 12714/30000 [8:08:14<11:00:58, 2.29s/it] 42%|████▏ | 12715/30000 [8:08:16<11:01:25, 2.30s/it] 42%|████▏ | 12716/30000 [8:08:18<11:00:20, 2.29s/it] 42%|████▏ | 12717/30000 [8:08:21<10:58:42, 2.29s/it] 42%|████▏ | 12718/30000 [8:08:23<10:58:31, 2.29s/it] 42%|████▏ | 12719/30000 [8:08:25<11:04:02, 2.31s/it] 42%|████▏ | 12720/30000 [8:08:27<11:07:30, 2.32s/it] 42%|████▏ | 12720/30000 [8:08:28<11:07:30, 2.32s/it] 42%|████▏ | 12721/30000 [8:08:30<11:00:20, 2.29s/it] 42%|████▏ | 12722/30000 [8:08:32<11:03:33, 2.30s/it] 42%|████▏ | 12723/30000 [8:08:34<11:03:18, 2.30s/it] 42%|████▏ | 12724/30000 [8:08:37<11:01:43, 2.30s/it] 42%|████▏ | 12725/30000 [8:08:39<11:04:21, 2.31s/it] 42%|████▏ | 12726/30000 [8:08:41<11:07:31, 2.32s/it] 42%|████▏ | 12727/30000 [8:08:44<10:57:48, 2.29s/it] 42%|████▏ | 12728/30000 [8:08:46<10:55:56, 2.28s/it] 42%|████▏ | 12729/30000 [8:08:48<11:01:37, 2.30s/it] 42%|████▏ | 12730/30000 [8:08:50<10:56:35, 2.28s/it] 42%|████▏ | 12730/30000 [8:08:50<10:56:35, 2.28s/it] 42%|████▏ | 12731/30000 [8:08:53<10:51:23, 2.26s/it] 42%|████▏ | 12732/30000 [8:08:55<10:58:11, 2.29s/it] 42%|████▏ | 12733/30000 [8:08:57<10:57:33, 2.28s/it] 42%|████▏ | 12734/30000 [8:09:00<10:57:23, 2.28s/it] 42%|████▏ | 12735/30000 [8:09:02<11:01:01, 2.30s/it] 42%|████▏ | 12736/30000 [8:09:04<11:02:12, 2.30s/it] 42%|████▏ | 12737/30000 [8:09:06<10:59:20, 2.29s/it] 42%|████▏ | 12738/30000 [8:09:09<11:03:59, 2.31s/it] 42%|████▏ | 12739/30000 [8:09:11<11:03:49, 2.31s/it] 42%|████▏ | 12740/30000 [8:09:13<11:00:59, 2.30s/it] 42%|████▏ | 12740/30000 [8:09:13<11:00:59, 2.30s/it] 42%|████▏ | 12741/30000 [8:09:16<11:05:04, 2.31s/it] 42%|████▏ | 12742/30000 [8:09:18<11:03:50, 2.31s/it] 42%|████▏ | 12743/30000 [8:09:20<11:01:07, 2.30s/it] 42%|████▏ | 12744/30000 [8:09:23<11:00:31, 2.30s/it] 42%|████▏ | 12745/30000 [8:09:25<11:04:07, 2.31s/it] 42%|████▏ | 12746/30000 [8:09:27<11:01:10, 2.30s/it] 42%|████▏ | 12747/30000 [8:09:29<10:58:57, 2.29s/it] 42%|████▏ | 12748/30000 [8:09:32<11:05:16, 2.31s/it] 42%|████▏ | 12749/30000 [8:09:34<11:02:05, 2.30s/it] 42%|████▎ | 12750/30000 [8:09:36<10:59:39, 2.29s/it] 42%|████▎ | 12750/30000 [8:09:36<10:59:39, 2.29s/it] 43%|████▎ | 12751/30000 [8:09:39<11:04:24, 2.31s/it] 43%|████▎ | 12752/30000 [8:09:41<11:04:03, 2.31s/it] 43%|████▎ | 12753/30000 [8:09:43<11:00:13, 2.30s/it] 43%|████▎ | 12754/30000 [8:09:46<11:05:21, 2.31s/it] 43%|████▎ | 12755/30000 [8:09:48<11:04:43, 2.31s/it] 43%|████▎ | 12756/30000 [8:09:50<11:02:28, 2.31s/it] 43%|████▎ | 12757/30000 [8:09:53<11:01:47, 2.30s/it] 43%|████▎ | 12758/30000 [8:09:55<11:06:12, 2.32s/it] 43%|████▎ | 12759/30000 [8:09:57<11:01:51, 2.30s/it] 43%|████▎ | 12760/30000 [8:09:59<10:58:32, 2.29s/it] 43%|████▎ | 12760/30000 [8:09:59<10:58:32, 2.29s/it] 43%|████▎ | 12761/30000 [8:10:02<11:02:27, 2.31s/it] 43%|████▎ | 12762/30000 [8:10:04<10:58:47, 2.29s/it] 43%|████▎ | 12763/30000 [8:10:06<10:58:05, 2.29s/it] 43%|████▎ | 12764/30000 [8:10:09<11:04:41, 2.31s/it] 43%|████▎ | 12765/30000 [8:10:11<11:06:15, 2.32s/it] 43%|████▎ | 12766/30000 [8:10:13<11:04:18, 2.31s/it] 43%|████▎ | 12767/30000 [8:10:16<11:03:03, 2.31s/it] 43%|████▎ | 12768/30000 [8:10:18<11:00:50, 2.30s/it] 43%|████▎ | 12769/30000 [8:10:20<10:58:07, 2.29s/it] 43%|████▎ | 12770/30000 [8:10:22<10:56:05, 2.28s/it] 43%|████▎ | 12770/30000 [8:10:22<10:56:05, 2.28s/it] 43%|████▎ | 12771/30000 [8:10:25<11:03:10, 2.31s/it] 43%|████▎ | 12772/30000 [8:10:27<11:00:06, 2.30s/it] 43%|████▎ | 12773/30000 [8:10:29<10:59:01, 2.30s/it] 43%|████▎ | 12774/30000 [8:10:32<11:04:31, 2.31s/it] 43%|████▎ | 12775/30000 [8:10:34<10:58:35, 2.29s/it] 43%|████▎ | 12776/30000 [8:10:36<10:55:28, 2.28s/it] 43%|████▎ | 12777/30000 [8:10:39<11:00:10, 2.30s/it] 43%|████▎ | 12778/30000 [8:10:41<10:57:52, 2.29s/it] 43%|████▎ | 12779/30000 [8:10:43<10:56:39, 2.29s/it] 43%|████▎ | 12780/30000 [8:10:46<11:06:58, 2.32s/it] 43%|████▎ | 12780/30000 [8:10:46<11:06:58, 2.32s/it] 43%|████▎ | 12781/30000 [8:10:48<11:04:26, 2.32s/it] 43%|████▎ | 12782/30000 [8:10:50<11:00:37, 2.30s/it] 43%|████▎ | 12783/30000 [8:10:52<11:01:25, 2.31s/it] 43%|████▎ | 12784/30000 [8:10:55<10:58:39, 2.30s/it] 43%|████▎ | 12785/30000 [8:10:57<10:55:14, 2.28s/it] 43%|████▎ | 12786/30000 [8:10:59<10:53:01, 2.28s/it] 43%|████▎ | 12787/30000 [8:11:01<10:53:21, 2.28s/it] 43%|████▎ | 12788/30000 [8:11:04<10:55:29, 2.28s/it] 43%|████▎ | 12789/30000 [8:11:06<10:53:32, 2.28s/it] 43%|████▎ | 12790/30000 [8:11:08<11:04:16, 2.32s/it] 43%|████▎ | 12790/30000 [8:11:08<11:04:16, 2.32s/it] 43%|████▎ | 12791/30000 [8:11:11<11:03:23, 2.31s/it] 43%|████▎ | 12792/30000 [8:11:13<10:56:09, 2.29s/it] 43%|████▎ | 12793/30000 [8:11:15<11:00:17, 2.30s/it] 43%|████▎ | 12794/30000 [8:11:18<10:57:15, 2.29s/it] 43%|████▎ | 12795/30000 [8:11:20<10:54:40, 2.28s/it] 43%|████▎ | 12796/30000 [8:11:22<10:55:06, 2.28s/it] 43%|████▎ | 12797/30000 [8:11:25<11:03:03, 2.31s/it] 43%|████▎ | 12798/30000 [8:11:27<11:01:04, 2.31s/it] 43%|████▎ | 12799/30000 [8:11:29<10:57:56, 2.29s/it] 43%|████▎ | 12800/30000 [8:11:31<11:03:55, 2.32s/it] 43%|████▎ | 12800/30000 [8:11:31<11:03:55, 2.32s/it] 43%|████▎ | 12801/30000 [8:11:34<11:02:11, 2.31s/it] 43%|████▎ | 12802/30000 [8:11:36<10:58:24, 2.30s/it] 43%|████▎ | 12803/30000 [8:11:38<10:59:29, 2.30s/it] 43%|████▎ | 12804/30000 [8:11:41<11:00:14, 2.30s/it] 43%|████▎ | 12805/30000 [8:11:43<10:59:54, 2.30s/it] 43%|████▎ | 12806/30000 [8:11:45<10:59:38, 2.30s/it] 43%|████▎ | 12807/30000 [8:11:48<11:03:17, 2.31s/it] 43%|████▎ | 12808/30000 [8:11:50<11:03:15, 2.31s/it] 43%|████▎ | 12809/30000 [8:11:52<10:58:18, 2.30s/it] 43%|████▎ | 12810/30000 [8:11:54<10:56:13, 2.29s/it] 43%|████▎ | 12810/30000 [8:11:54<10:56:13, 2.29s/it] 43%|████▎ | 12811/30000 [8:11:57<11:01:17, 2.31s/it] 43%|████▎ | 12812/30000 [8:11:59<10:59:32, 2.30s/it] 43%|████▎ | 12813/30000 [8:12:01<10:58:58, 2.30s/it] 43%|████▎ | 12814/30000 [8:12:04<11:07:28, 2.33s/it] 43%|████▎ | 12815/30000 [8:12:06<11:04:15, 2.32s/it] 43%|████▎ | 12816/30000 [8:12:08<11:00:49, 2.31s/it] 43%|████▎ | 12817/30000 [8:12:11<10:58:10, 2.30s/it] 43%|████▎ | 12818/30000 [8:12:13<10:59:40, 2.30s/it] 43%|████▎ | 12819/30000 [8:12:15<10:54:48, 2.29s/it] 43%|████▎ | 12820/30000 [8:12:17<10:48:27, 2.26s/it] 43%|████▎ | 12820/30000 [8:12:17<10:48:27, 2.26s/it] 43%|████▎ | 12821/30000 [8:12:20<11:00:00, 2.31s/it] 43%|████▎ | 12822/30000 [8:12:22<10:58:26, 2.30s/it] 43%|████▎ | 12823/30000 [8:12:24<10:57:11, 2.30s/it] 43%|████▎ | 12824/30000 [8:12:27<10:59:00, 2.30s/it] 43%|████▎ | 12825/30000 [8:12:29<10:58:07, 2.30s/it] 43%|████▎ | 12826/30000 [8:12:31<10:52:56, 2.28s/it] 43%|████▎ | 12827/30000 [8:12:33<10:51:57, 2.28s/it] 43%|████▎ | 12828/30000 [8:12:36<10:59:36, 2.30s/it] 43%|████▎ | 12829/30000 [8:12:38<11:01:56, 2.31s/it] 43%|████▎ | 12830/30000 [8:12:40<10:57:16, 2.30s/it] 43%|████▎ | 12830/30000 [8:12:40<10:57:16, 2.30s/it] 43%|████▎ | 12831/30000 [8:12:43<10:56:37, 2.29s/it] 43%|████▎ | 12832/30000 [8:12:45<10:55:22, 2.29s/it] 43%|████▎ | 12833/30000 [8:12:47<10:51:40, 2.28s/it] 43%|████▎ | 12834/30000 [8:12:50<10:51:43, 2.28s/it] 43%|████▎ | 12835/30000 [8:12:52<10:57:42, 2.30s/it] 43%|████▎ | 12836/30000 [8:12:54<11:00:35, 2.31s/it] 43%|████▎ | 12837/30000 [8:12:56<10:58:15, 2.30s/it] 43%|████▎ | 12838/30000 [8:12:59<11:00:02, 2.31s/it] 43%|████▎ | 12839/30000 [8:13:01<10:56:06, 2.29s/it] 43%|████▎ | 12840/30000 [8:13:03<10:50:48, 2.28s/it] 43%|████▎ | 12840/30000 [8:13:03<10:50:48, 2.28s/it] 43%|████▎ | 12841/30000 [8:13:06<10:48:58, 2.27s/it] 43%|████▎ | 12842/30000 [8:13:08<10:52:41, 2.28s/it] 43%|████▎ | 12843/30000 [8:13:10<10:48:12, 2.27s/it] 43%|████▎ | 12844/30000 [8:13:12<10:47:28, 2.26s/it] 43%|████▎ | 12845/30000 [8:13:15<10:53:53, 2.29s/it] 43%|████▎ | 12846/30000 [8:13:17<10:58:14, 2.30s/it] 43%|████▎ | 12847/30000 [8:13:19<10:56:10, 2.30s/it] 43%|████▎ | 12848/30000 [8:13:22<10:51:55, 2.28s/it] 43%|████▎ | 12849/30000 [8:13:24<10:58:15, 2.30s/it] 43%|████▎ | 12850/30000 [8:13:26<10:59:21, 2.31s/it] 43%|████▎ | 12850/30000 [8:13:26<10:59:21, 2.31s/it] 43%|████▎ | 12851/30000 [8:13:28<10:55:36, 2.29s/it] 43%|████▎ | 12852/30000 [8:13:31<10:54:11, 2.29s/it] 43%|████▎ | 12853/30000 [8:13:33<11:02:01, 2.32s/it] 43%|████▎ | 12854/30000 [8:13:36<11:06:52, 2.33s/it] 43%|████▎ | 12855/30000 [8:13:38<10:56:38, 2.30s/it] 43%|████▎ | 12856/30000 [8:13:40<10:57:50, 2.30s/it] 43%|████▎ | 12857/30000 [8:13:42<10:55:15, 2.29s/it] 43%|████▎ | 12858/30000 [8:13:45<10:52:40, 2.28s/it] 43%|████▎ | 12859/30000 [8:13:47<10:51:03, 2.28s/it] 43%|████▎ | 12860/30000 [8:13:49<10:58:12, 2.30s/it] +{'loss': 0.016, 'grad_norm': 0.07719206809997559, 'learning_rate': 2.0322997612542366e-05, 'flow_matching_loss': 0.015970749128609896, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.63} +{'loss': 0.0158, 'grad_norm': 0.09329578280448914, 'learning_rate': 2.0307535811595427e-05, 'flow_matching_loss': 0.01575769213959575, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.63} +{'loss': 0.0158, 'grad_norm': 0.0748058333992958, 'learning_rate': 2.0292067561492013e-05, 'flow_matching_loss': 0.015780258271843196, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.63} +{'loss': 0.0162, 'grad_norm': 0.09164014458656311, 'learning_rate': 2.02765928810275e-05, 'flow_matching_loss': 0.016214762441813947, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.63} +{'loss': 0.0152, 'grad_norm': 0.07162585109472275, 'learning_rate': 2.0261111789005087e-05, 'flow_matching_loss': 0.015178795717656613, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.64} +{'loss': 0.0158, 'grad_norm': 0.08761367201805115, 'learning_rate': 2.024562430423576e-05, 'flow_matching_loss': 0.01577666401863098, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.64} +{'loss': 0.0151, 'grad_norm': 0.06986001878976822, 'learning_rate': 2.0230130445538273e-05, 'flow_matching_loss': 0.015112879686057568, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.64} +{'loss': 0.0153, 'grad_norm': 0.07255560159683228, 'learning_rate': 2.0214630231739127e-05, 'flow_matching_loss': 0.01534789465367794, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.65} +{'loss': 0.0155, 'grad_norm': 0.0966716930270195, 'learning_rate': 2.0199123681672543e-05, 'flow_matching_loss': 0.015469399094581605, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.65} +{'loss': 0.0143, 'grad_norm': 0.07361998409032822, 'learning_rate': 2.0183610814180437e-05, 'flow_matching_loss': 0.014341234602034092, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.65} +{'loss': 0.0167, 'grad_norm': 0.07613180577754974, 'learning_rate': 2.016809164811241e-05, 'flow_matching_loss': 0.016720857936888933, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.65} +{'loss': 0.0155, 'grad_norm': 0.07348378747701645, 'learning_rate': 2.015256620232571e-05, 'flow_matching_loss': 0.015492786001414061, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.66} +{'loss': 0.0156, 'grad_norm': 0.08892955631017685, 'learning_rate': 2.0137034495685222e-05, 'flow_matching_loss': 0.015598589554429054, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.66} +{'loss': 0.0145, 'grad_norm': 0.06663551181554794, 'learning_rate': 2.012149654706343e-05, 'flow_matching_loss': 0.01454709293320775, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.66} +{'loss': 0.016, 'grad_norm': 0.08209995925426483, 'learning_rate': 2.010595237534041e-05, 'flow_matching_loss': 0.016047302167862654, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.67} +{'loss': 0.0158, 'grad_norm': 0.08488032966852188, 'learning_rate': 2.0090401999403793e-05, 'flow_matching_loss': 0.015839790645986795, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.67} +{'loss': 0.0155, 'grad_norm': 0.08656375110149384, 'learning_rate': 2.007484543814875e-05, 'flow_matching_loss': 0.015500704944133758, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.67} +{'loss': 0.0156, 'grad_norm': 0.08931925147771835, 'learning_rate': 2.0059282710477972e-05, 'flow_matching_loss': 0.01557289930060506, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.67} +{'loss': 0.0144, 'grad_norm': 0.06347820907831192, 'learning_rate': 2.0043713835301647e-05, 'flow_matching_loss': 0.01441049836575985, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.68} +{'loss': 0.0149, 'grad_norm': 0.07370192557573318, 'learning_rate': 2.0028138831537417e-05, 'flow_matching_loss': 0.01488736756145954, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.68} +{'loss': 0.0143, 'grad_norm': 0.08057195693254471, 'learning_rate': 2.0012557718110387e-05, 'flow_matching_loss': 0.014337416645139456, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.68} +{'loss': 0.0153, 'grad_norm': 0.07211877405643463, 'learning_rate': 1.9996970513953077e-05, 'flow_matching_loss': 0.015252651926130057, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.69} +{'loss': 0.0149, 'grad_norm': 0.07587935030460358, 'learning_rate': 1.9981377238005405e-05, 'flow_matching_loss': 0.014883488696068526, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.69} +{'loss': 0.0144, 'grad_norm': 0.08962063491344452, 'learning_rate': 1.9965777909214674e-05, 'flow_matching_loss': 0.014429643470793962, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.69} +{'loss': 0.0134, 'grad_norm': 0.07572191208600998, 'learning_rate': 1.9950172546535545e-05, 'flow_matching_loss': 0.013441132102161645, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.7} +{'loss': 0.0158, 'grad_norm': 0.07448488473892212, 'learning_rate': 1.9934561168930002e-05, 'flow_matching_loss': 0.01582878902554512, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.7} +{'loss': 0.0145, 'grad_norm': 0.10022494941949844, 'learning_rate': 1.991894379536734e-05, 'flow_matching_loss': 0.014469576254487037, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.7} +{'loss': 0.015, 'grad_norm': 0.09445875883102417, 'learning_rate': 1.9903320444824137e-05, 'flow_matching_loss': 0.014997255895286799, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.7} +{'loss': 0.0156, 'grad_norm': 0.07282574474811554, 'learning_rate': 1.9887691136284245e-05, 'flow_matching_loss': 0.015600121766328811, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.71} +{'loss': 0.0168, 'grad_norm': 0.09825586527585983, 'learning_rate': 1.9872055888738745e-05, 'flow_matching_loss': 0.016763430461287497, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.71} +{'loss': 0.0149, 'grad_norm': 0.06771338731050491, 'learning_rate': 1.985641472118593e-05, 'flow_matching_loss': 0.014882063400000334, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.71} +{'loss': 0.0155, 'grad_norm': 0.0662817433476448, 'learning_rate': 1.9840767652631294e-05, 'flow_matching_loss': 0.015473625343292951, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.72} +{'loss': 0.0159, 'grad_norm': 0.061968859285116196, 'learning_rate': 1.9825114702087513e-05, 'flow_matching_loss': 0.015894069988280534, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.72} +{'loss': 0.0149, 'grad_norm': 0.07768125087022781, 'learning_rate': 1.9809455888574388e-05, 'flow_matching_loss': 0.014881713222712279, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.72} +{'loss': 0.0169, 'grad_norm': 0.09999608993530273, 'learning_rate': 1.9793791231118864e-05, 'flow_matching_loss': 0.016949024703353643, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.72} +{'loss': 0.0153, 'grad_norm': 0.08304872363805771, 'learning_rate': 1.9778120748754958e-05, 'flow_matching_loss': 0.015344189573079347, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.73} +{'loss': 0.0151, 'grad_norm': 0.10192036628723145, 'learning_rate': 1.97624444605238e-05, 'flow_matching_loss': 0.015131905023008586, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.73} +{'loss': 0.0145, 'grad_norm': 0.06647082418203354, 'learning_rate': 1.9746762385473552e-05, 'flow_matching_loss': 0.01452829958871007, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.73} +{'loss': 0.0154, 'grad_norm': 0.09458453208208084, 'learning_rate': 1.9731074542659414e-05, 'flow_matching_loss': 0.015444056503474713, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.74} +{'loss': 0.0163, 'grad_norm': 0.10168302804231644, 'learning_rate': 1.9715380951143603e-05, 'flow_matching_loss': 0.016327164974063633, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.74} +{'loss': 0.0155, 'grad_norm': 0.10346796363592148, 'learning_rate': 1.9699681629995295e-05, 'flow_matching_loss': 0.015480757225304842, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.74} 43%|████▎ | 12860/30000 [8:13:49<10:58:12, 2.30s/it] 43%|████▎ | 12861/30000 [8:13:52<11:01:59, 2.32s/it] 43%|████▎ | 12862/30000 [8:13:54<10:57:47, 2.30s/it] 43%|████▎ | 12863/30000 [8:13:56<10:54:16, 2.29s/it] 43%|████▎ | 12864/30000 [8:13:58<10:54:55, 2.29s/it] 43%|████▎ | 12865/30000 [8:14:01<10:51:12, 2.28s/it] 43%|████▎ | 12866/30000 [8:14:03<10:48:04, 2.27s/it] 43%|████▎ | 12867/30000 [8:14:05<10:53:00, 2.29s/it] 43%|████▎ | 12868/30000 [8:14:08<10:54:44, 2.29s/it] 43%|████▎ | 12869/30000 [8:14:10<10:48:47, 2.27s/it] 43%|████▎ | 12870/30000 [8:14:12<10:52:09, 2.28s/it] 43%|████▎ | 12870/30000 [8:14:12<10:52:09, 2.28s/it] 43%|████▎ | 12871/30000 [8:14:14<10:55:21, 2.30s/it] 43%|████▎ | 12872/30000 [8:14:17<10:52:59, 2.29s/it] 43%|████▎ | 12873/30000 [8:14:19<10:49:28, 2.28s/it] 43%|████▎ | 12874/30000 [8:14:21<10:46:43, 2.27s/it] 43%|████▎ | 12875/30000 [8:14:23<10:53:48, 2.29s/it] 43%|████▎ | 12876/30000 [8:14:26<10:57:24, 2.30s/it] 43%|████▎ | 12877/30000 [8:14:28<10:57:08, 2.30s/it] 43%|████▎ | 12878/30000 [8:14:30<10:53:24, 2.29s/it] 43%|████▎ | 12879/30000 [8:14:33<10:52:52, 2.29s/it] 43%|████▎ | 12880/30000 [8:14:35<10:47:40, 2.27s/it] 43%|████▎ | 12880/30000 [8:14:35<10:47:40, 2.27s/it] 43%|████▎ | 12881/30000 [8:14:37<10:45:15, 2.26s/it] 43%|████▎ | 12882/30000 [8:14:39<10:53:01, 2.29s/it] 43%|████▎ | 12883/30000 [8:14:42<10:57:45, 2.31s/it] 43%|████▎ | 12884/30000 [8:14:44<10:55:52, 2.30s/it] 43%|████▎ | 12885/30000 [8:14:46<10:54:07, 2.29s/it] 43%|████▎ | 12886/30000 [8:14:49<10:55:19, 2.30s/it] 43%|████▎ | 12887/30000 [8:14:51<10:51:00, 2.28s/it] 43%|████▎ | 12888/30000 [8:14:53<10:49:35, 2.28s/it] 43%|████▎ | 12889/30000 [8:14:56<10:56:05, 2.30s/it] 43%|████▎ | 12890/30000 [8:14:58<10:52:52, 2.29s/it] 43%|████▎ | 12890/30000 [8:14:58<10:52:52, 2.29s/it] 43%|████▎ | 12891/30000 [8:15:00<10:50:13, 2.28s/it] 43%|████▎ | 12892/30000 [8:15:02<10:51:09, 2.28s/it] 43%|████▎ | 12893/30000 [8:15:05<10:56:13, 2.30s/it] 43%|████▎ | 12894/30000 [8:15:07<10:53:55, 2.29s/it] 43%|████▎ | 12895/30000 [8:15:09<10:50:39, 2.28s/it] 43%|████▎ | 12896/30000 [8:15:12<10:53:05, 2.29s/it] 43%|████▎ | 12897/30000 [8:15:14<10:52:38, 2.29s/it] 43%|████▎ | 12898/30000 [8:15:16<10:53:23, 2.29s/it] 43%|████▎ | 12899/30000 [8:15:18<10:50:49, 2.28s/it] 43%|████▎ | 12900/30000 [8:15:21<10:54:35, 2.30s/it] 43%|████▎ | 12900/30000 [8:15:21<10:54:35, 2.30s/it] 43%|████▎ | 12901/30000 [8:15:23<10:54:02, 2.30s/it] 43%|████▎ | 12902/30000 [8:15:25<10:50:41, 2.28s/it] 43%|████▎ | 12903/30000 [8:15:28<10:48:49, 2.28s/it] 43%|████▎ | 12904/30000 [8:15:30<10:50:46, 2.28s/it] 43%|████▎ | 12905/30000 [8:15:32<10:48:44, 2.28s/it] 43%|████▎ | 12906/30000 [8:15:34<10:45:53, 2.27s/it] 43%|████▎ | 12907/30000 [8:15:37<10:50:39, 2.28s/it] 43%|████▎ | 12908/30000 [8:15:39<10:55:09, 2.30s/it] 43%|████▎ | 12909/30000 [8:15:41<10:52:08, 2.29s/it] 43%|████▎ | 12910/30000 [8:15:44<10:57:21, 2.31s/it] 43%|████▎ | 12910/30000 [8:15:44<10:57:21, 2.31s/it] 43%|████▎ | 12911/30000 [8:15:46<10:51:49, 2.29s/it] 43%|████▎ | 12912/30000 [8:15:48<10:49:31, 2.28s/it] 43%|████▎ | 12913/30000 [8:15:50<10:50:55, 2.29s/it] 43%|████▎ | 12914/30000 [8:15:53<10:57:09, 2.31s/it] 43%|████▎ | 12915/30000 [8:15:55<10:51:32, 2.29s/it] 43%|████▎ | 12916/30000 [8:15:57<10:52:47, 2.29s/it] 43%|████▎ | 12917/30000 [8:16:00<10:57:22, 2.31s/it] 43%|████▎ | 12918/30000 [8:16:02<10:59:40, 2.32s/it] 43%|████▎ | 12919/30000 [8:16:04<10:54:56, 2.30s/it] 43%|████▎ | 12920/30000 [8:16:07<10:55:17, 2.30s/it] 43%|████▎ | 12920/30000 [8:16:07<10:55:17, 2.30s/it] 43%|████▎ | 12921/30000 [8:16:09<10:56:16, 2.31s/it] 43%|████▎ | 12922/30000 [8:16:11<10:56:01, 2.30s/it] 43%|████▎ | 12923/30000 [8:16:13<10:52:18, 2.29s/it] 43%|████▎ | 12924/30000 [8:16:16<10:59:36, 2.32s/it] 43%|████▎ | 12925/30000 [8:16:18<11:03:47, 2.33s/it] 43%|████▎ | 12926/30000 [8:16:20<10:56:47, 2.31s/it] 43%|████▎ | 12927/30000 [8:16:23<10:51:40, 2.29s/it] 43%|████▎ | 12928/30000 [8:16:25<10:51:22, 2.29s/it] 43%|████▎ | 12929/30000 [8:16:27<10:47:55, 2.28s/it] 43%|████▎ | 12930/30000 [8:16:30<10:48:46, 2.28s/it] 43%|████▎ | 12930/30000 [8:16:30<10:48:46, 2.28s/it] 43%|████▎ | 12931/30000 [8:16:32<10:54:11, 2.30s/it] 43%|████▎ | 12932/30000 [8:16:34<10:56:24, 2.31s/it] 43%|████▎ | 12933/30000 [8:16:36<10:53:39, 2.30s/it] 43%|████▎ | 12934/30000 [8:16:39<10:57:10, 2.31s/it] 43%|████▎ | 12935/30000 [8:16:41<10:54:00, 2.30s/it] 43%|████▎ | 12936/30000 [8:16:43<10:50:09, 2.29s/it] 43%|████▎ | 12937/30000 [8:16:46<10:53:38, 2.30s/it] 43%|████▎ | 12938/30000 [8:16:48<10:52:09, 2.29s/it] 43%|████▎ | 12939/30000 [8:16:50<10:50:04, 2.29s/it] 43%|████▎ | 12940/30000 [8:16:53<10:48:46, 2.28s/it] 43%|████▎ | 12940/30000 [8:16:53<10:48:46, 2.28s/it] 43%|████▎ | 12941/30000 [8:16:55<10:56:19, 2.31s/it] 43%|████▎ | 12942/30000 [8:16:57<10:56:57, 2.31s/it] 43%|████▎ | 12943/30000 [8:16:59<10:51:51, 2.29s/it] 43%|████▎ | 12944/30000 [8:17:02<10:55:12, 2.30s/it] 43%|████▎ | 12945/30000 [8:17:04<10:56:51, 2.31s/it] 43%|████▎ | 12946/30000 [8:17:06<10:55:24, 2.31s/it] 43%|████▎ | 12947/30000 [8:17:09<10:59:24, 2.32s/it] 43%|████▎ | 12948/30000 [8:17:11<10:58:02, 2.32s/it] 43%|████▎ | 12949/30000 [8:17:13<10:55:55, 2.31s/it] 43%|████▎ | 12950/30000 [8:17:16<10:55:59, 2.31s/it] 43%|████▎ | 12950/30000 [8:17:16<10:55:59, 2.31s/it] 43%|████▎ | 12951/30000 [8:17:18<10:58:28, 2.32s/it] 43%|████▎ | 12952/30000 [8:17:20<10:58:47, 2.32s/it] 43%|████▎ | 12953/30000 [8:17:23<10:53:20, 2.30s/it] 43%|████▎ | 12954/30000 [8:17:25<11:00:31, 2.32s/it] 43%|████▎ | 12955/30000 [8:17:27<10:57:52, 2.32s/it] 43%|████▎ | 12956/30000 [8:17:30<10:54:14, 2.30s/it] 43%|████▎ | 12957/30000 [8:17:32<10:58:19, 2.32s/it] 43%|████▎ | 12958/30000 [8:17:34<11:01:01, 2.33s/it] 43%|████▎ | 12959/30000 [8:17:37<10:57:35, 2.32s/it] 43%|████▎ | 12960/30000 [8:17:39<10:48:36, 2.28s/it] 43%|████▎ | 12960/30000 [8:17:39<10:48:36, 2.28s/it] 43%|████▎ | 12961/30000 [8:17:41<10:53:07, 2.30s/it] 43%|████▎ | 12962/30000 [8:17:43<10:45:49, 2.27s/it] 43%|████▎ | 12963/30000 [8:17:46<10:45:43, 2.27s/it] 43%|████▎ | 12964/30000 [8:17:48<10:51:06, 2.29s/it] 43%|████▎ | 12965/30000 [8:17:50<10:56:38, 2.31s/it] 43%|████▎ | 12966/30000 [8:17:53<10:54:00, 2.30s/it] 43%|████▎ | 12967/30000 [8:17:55<10:57:30, 2.32s/it] 43%|████▎ | 12968/30000 [8:17:57<10:57:05, 2.31s/it] 43%|████▎ | 12969/30000 [8:17:59<10:55:18, 2.31s/it] 43%|████▎ | 12970/30000 [8:18:02<10:58:57, 2.32s/it] 43%|████▎ | 12970/30000 [8:18:02<10:58:57, 2.32s/it] 43%|████▎ | 12971/30000 [8:18:04<10:58:12, 2.32s/it] 43%|████▎ | 12972/30000 [8:18:06<10:52:17, 2.30s/it] 43%|████▎ | 12973/30000 [8:18:09<10:49:57, 2.29s/it] 43%|████▎ | 12974/30000 [8:18:11<10:57:21, 2.32s/it] 43%|████▎ | 12975/30000 [8:18:13<10:54:15, 2.31s/it] 43%|████▎ | 12976/30000 [8:18:16<10:49:13, 2.29s/it] 43%|████▎ | 12977/30000 [8:18:18<10:53:56, 2.30s/it] 43%|████▎ | 12978/30000 [8:18:20<10:53:02, 2.30s/it] 43%|████▎ | 12979/30000 [8:18:23<10:52:41, 2.30s/it] 43%|████▎ | 12980/30000 [8:18:25<10:57:10, 2.32s/it] 43%|████▎ | 12980/30000 [8:18:25<10:57:10, 2.32s/it] 43%|████▎ | 12981/30000 [8:18:27<10:57:16, 2.32s/it] 43%|████▎ | 12982/30000 [8:18:29<10:54:50, 2.31s/it] 43%|████▎ | 12983/30000 [8:18:32<10:53:34, 2.30s/it] 43%|████▎ | 12984/30000 [8:18:34<10:55:09, 2.31s/it] 43%|████▎ | 12985/30000 [8:18:36<10:51:09, 2.30s/it] 43%|████▎ | 12986/30000 [8:18:39<10:50:00, 2.29s/it] 43%|████▎ | 12987/30000 [8:18:41<10:51:52, 2.30s/it] 43%|████▎ | 12988/30000 [8:18:43<10:48:30, 2.29s/it] 43%|████▎ | 12989/30000 [8:18:45<10:45:59, 2.28s/it] 43%|████▎ | 12990/30000 [8:18:48<10:52:02, 2.30s/it] 43%|████▎ | 12990/30000 [8:18:48<10:52:02, 2.30s/it] 43%|████▎ | 12991/30000 [8:18:50<10:45:18, 2.28s/it] 43%|████▎ | 12992/30000 [8:18:52<10:46:17, 2.28s/it] 43%|████▎ | 12993/30000 [8:18:55<10:52:09, 2.30s/it] 43%|████▎ | 12994/30000 [8:18:57<10:53:29, 2.31s/it] 43%|████▎ | 12995/30000 [8:18:59<10:47:50, 2.29s/it] 43%|████▎ | 12996/30000 [8:19:02<10:51:38, 2.30s/it] 43%|████▎ | 12997/30000 [8:19:04<10:53:48, 2.31s/it] 43%|████▎ | 12998/30000 [8:19:06<10:51:55, 2.30s/it] 43%|████▎ | 12999/30000 [8:19:08<10:49:35, 2.29s/it] 43%|████▎ | 13000/30000 [8:19:11<10:47:57, 2.29s/it] 43%|████▎ | 13000/30000 [8:19:11<10:47:57, 2.29s/it] 43%|████▎ | 13001/30000 [8:19:13<10:45:25, 2.28s/it] 43%|████▎ | 13002/30000 [8:19:15<10:46:12, 2.28s/it] 43%|████▎ | 13003/30000 [8:19:18<10:50:50, 2.30s/it] 43%|████▎ | 13004/30000 [8:19:20<10:51:02, 2.30s/it] 43%|████▎ | 13005/30000 [8:19:22<10:40:21, 2.26s/it] 43%|████▎ | 13006/30000 [8:19:24<10:50:13, 2.30s/it] 43%|████▎ | 13007/30000 [8:19:27<10:51:46, 2.30s/it] 43%|████▎ | 13008/30000 [8:19:29<10:46:29, 2.28s/it] 43%|████▎ | 13009/30000 [8:19:31<10:48:22, 2.29s/it] 43%|████▎ | 13010/30000 [8:19:34<10:50:58, 2.30s/it] 43%|████▎ | 13010/30000 [8:19:34<10:50:58, 2.30s/it] 43%|████▎ | 13011/30000 [8:19:36<10:46:11, 2.28s/it] 43%|████▎ | 13012/30000 [8:19:38<10:45:44, 2.28s/it] 43%|████▎ | 13013/30000 [8:19:41<10:54:00, 2.31s/it] 43%|████▎ | 13014/30000 [8:19:43<10:55:31, 2.32s/it] 43%|████▎ | 13015/30000 [8:19:45<10:49:14, 2.29s/it] 43%|████▎ | 13016/30000 [8:19:47<10:46:15, 2.28s/it] 43%|████▎ | 13017/30000 [8:19:50<10:52:37, 2.31s/it] 43%|████▎ | 13018/30000 [8:19:52<10:49:05, 2.29s/it] 43%|████▎ | 13019/30000 [8:19:54<10:45:34, 2.28s/it] 43%|████▎ | 13020/30000 [8:19:57<10:46:03, 2.28s/it] 43%|████▎ | 13020/30000 [8:19:57<10:46:03, 2.28s/it] 43%|████▎ | 13021/30000 [8:19:59<10:47:49, 2.29s/it] 43%|████▎ | 13022/30000 [8:20:01<10:45:59, 2.28s/it] 43%|████▎ | 13023/30000 [8:20:03<10:41:36, 2.27s/it] 43%|████▎ | 13024/30000 [8:20:06<10:49:24, 2.30s/it] 43%|████▎ | 13025/30000 [8:20:08<10:49:29, 2.30s/it] 43%|████▎ | 13026/30000 [8:20:10<10:45:52, 2.28s/it] 43%|████▎ | 13027/30000 [8:20:12<10:44:13, 2.28s/it] 43%|████▎ | 13028/30000 [8:20:15<10:47:41, 2.29s/it] 43%|████▎ | 13029/30000 [8:20:17<10:48:49, 2.29s/it] 43%|████▎ | 13030/30000 [8:20:19<10:48:19, 2.29s/it] 43%|████▎ | 13030/30000 [8:20:19<10:48:19, 2.29s/it] 43%|████▎ | 13031/30000 [8:20:22<10:48:30, 2.29s/it] 43%|████▎ | 13032/30000 [8:20:24<10:48:52, 2.29s/it] 43%|████▎ | 13033/30000 [8:20:26<10:44:25, 2.28s/it] 43%|████▎ | 13034/30000 [8:20:29<10:46:15, 2.29s/it] 43%|████▎ | 13035/30000 [8:20:31<10:54:03, 2.31s/it] 43%|████▎ | 13036/30000 [8:20:33<10:59:02, 2.33s/it] 43%|████▎ | 13037/30000 [8:20:36<10:51:19, 2.30s/it] 43%|████▎ | 13038/30000 [8:20:38<10:49:12, 2.30s/it] 43%|████▎ | 13039/30000 [8:20:40<10:49:43, 2.30s/it] 43%|████▎ | 13040/30000 [8:20:42<10:49:37, 2.30s/it] 43%|████▎ | 13040/30000 [8:20:42<10:49:37, 2.30s/it] 43%|████▎ | 13041/30000 [8:20:45<10:46:53, 2.29s/it] 43%|████▎ | 13042/30000 [8:20:47<10:46:12, 2.29s/it] 43%|████▎ | 13043/30000 [8:20:49<10:46:34, 2.29s/it] 43%|████▎ | 13044/30000 [8:20:52<10:45:44, 2.29s/it] 43%|████▎ | 13045/30000 [8:20:54<10:40:58, 2.27s/it] 43%|████▎ | 13046/30000 [8:20:56<10:47:31, 2.29s/it] 43%|████▎ | 13047/30000 [8:20:58<10:47:40, 2.29s/it] 43%|████▎ | 13048/30000 [8:21:01<10:45:39, 2.29s/it] 43%|████▎ | 13049/30000 [8:21:03<10:41:35, 2.27s/it] 44%|████▎ | 13050/30000 [8:21:05<10:50:06, 2.30s/it] 44%|████▎ | 13050/30000 [8:21:05<10:50:06, 2.30s/it] 44%|████▎ | 13051/30000 [8:21:08<10:50:43, 2.30s/it] 44%|████▎ | 13052/30000 [8:21:10<10:46:55, 2.29s/it] 44%|████▎ | 13053/30000 [8:21:12<10:47:36, 2.29s/it] 44%|████▎ | 13054/30000 [8:21:15<10:53:27, 2.31s/it] 44%|████▎ | 13055/30000 [8:21:17<10:47:30, 2.29s/it] 44%|████▎ | 13056/30000 [8:21:19<10:43:46, 2.28s/it] 44%|████▎ | 13057/30000 [8:21:21<10:49:48, 2.30s/it] 44%|████▎ | 13058/30000 [8:21:24<10:50:57, 2.31s/it] 44%|████▎ | 13059/30000 [8:21:26<10:48:06, 2.30s/it] 44%|████▎ | 13060/30000 [8:21:28<10:43:20, 2.28s/it] 44%|████▎ | 13060/30000 [8:21:28<10:43:20, 2.28s/it] 44%|████▎ | 13061/30000 [8:21:31<10:48:29, 2.30s/it] 44%|████▎ | 13062/30000 [8:21:33<10:52:46, 2.31s/it] 44%|████▎ | 13063/30000 [8:21:35<10:48:32, 2.30s/it] 44%|████▎ | 13064/30000 [8:21:37<10:46:42, 2.29s/it] 44%|████▎ | 13065/30000 [8:21:40<10:48:31, 2.30s/it] 44%|████▎ | 13066/30000 [8:21:42<10:45:06, 2.29s/it] 44%|████▎ | 13067/30000 [8:21:44<10:43:02, 2.28s/it] 44%|████▎ | 13068/30000 [8:21:47<10:49:00, 2.30s/it] 44%|████▎ | 13069/30000 [8:21:49<10:53:47, 2.32s/it] 44%|████▎ | 13070/30000 [8:21:51<10:47:40, 2.30s/it] 44%|████▎ | 13070/30000 [8:21:51<10:47:40, 2.30s/it] 44%|████▎ | 13071/30000 [8:21:53<10:46:16, 2.29s/it] 44%|████▎ | 13072/30000 [8:21:56<10:48:07, 2.30s/it] 44%|████▎ | 13073/30000 [8:21:58<10:43:40, 2.28s/it] 44%|████▎ | 13074/30000 [8:22:00<10:41:17, 2.27s/it] 44%|████▎ | 13075/30000 [8:22:03<10:48:07, 2.30s/it] 44%|████▎ | 13076/30000 [8:22:05<10:48:59, 2.30s/it] 44%|████▎ | 13077/30000 [8:22:07<10:49:13, 2.30s/it] 44%|████▎ | 13078/30000 [8:22:10<10:47:41, 2.30s/it] 44%|████▎ | 13079/30000 [8:22:12<10:43:55, 2.28s/it] 44%|████▎ | 13080/30000 [8:22:14<10:42:57, 2.28s/it] 44%|████▎ | 13080/30000 [8:22:14<10:42:57, 2.28s/it] 44%|████▎ | 13081/30000 [8:22:16<10:43:11, 2.28s/it] 44%|████▎ | 13082/30000 [8:22:19<10:46:41, 2.29s/it] 44%|████▎ | 13083/30000 [8:22:21<10:50:07, 2.31s/it] 44%|████▎ | 13084/30000 [8:22:23<10:47:46, 2.30s/it] 44%|████▎ | 13085/30000 [8:22:26<10:47:25, 2.30s/it] 44%|████▎ | 13086/30000 [8:22:28<10:51:35, 2.31s/it] 44%|████▎ | 13087/30000 [8:22:30<10:50:15, 2.31s/it] 44%|████▎ | 13088/30000 [8:22:32<10:46:42, 2.29s/it] 44%|████▎ | 13089/30000 [8:22:35<10:49:01, 2.30s/it] 44%|████▎ | 13090/30000 [8:22:37<10:48:33, 2.30s/it] 44%|████▎ | 13090/30000 [8:22:37<10:48:33, 2.30s/it] 44%|████▎ | 13091/30000 [8:22:39<10:47:25, 2.30s/it] 44%|████▎ | 13092/30000 [8:22:42<10:49:01, 2.30s/it] 44%|████▎ | 13093/30000 [8:22:44<10:48:35, 2.30s/it] 44%|████▎ | 13094/30000 [8:22:46<10:46:13, 2.29s/it] 44%|████▎ | 13095/30000 [8:22:49<10:46:50, 2.30s/it] 44%|████▎ | 13096/30000 [8:22:51<10:49:18, 2.30s/it] 44%|████▎ | 13097/30000 [8:22:53<10:49:07, 2.30s/it] 44%|████▎ | 13098/30000 [8:22:55<10:47:34, 2.30s/it] 44%|████▎ | 13099/30000 [8:22:58<10:51:44, 2.31s/it] 44%|████▎ | 13100/30000 [8:23:00<10:52:37, 2.32s/it] 44%|████▎ | 13100/30000 [8:23:00<10:52:37, 2.32s/it] 44%|████▎ | 13101/30000 [8:23:02<10:46:28, 2.30s/it] 44%|████▎ | 13102/30000 [8:23:05<10:48:36, 2.30s/it] 44%|████▎ | 13103/30000 [8:23:07<10:53:12, 2.32s/it] 44%|████▎ | 13104/30000 [8:23:09<10:52:21, 2.32s/it] 44%|████▎ | 13105/30000 [8:23:12<10:49:51, 2.31s/it] 44%|████▎ | 13106/30000 [8:23:14<10:51:23, 2.31s/it] 44%|████▎ | 13107/30000 [8:23:16<10:48:16, 2.30s/it] 44%|████▎ | 13108/30000 [8:23:19<10:48:49, 2.30s/it] 44%|████▎ | 13109/30000 [8:23:21<10:48:12, 2.30s/it] 44%|████▎ | 13110/30000 [8:23:23<10:49:59, 2.31s/it] 44%|████▎ | 13110/30000 [8:23:23<10:49:59, 2.31s/it] 44%|████▎ | 13111/30000 [8:23:25<10:46:02, 2.30s/it] 44%|████▎ | 13112/30000 [8:23:28<10:44:45, 2.29s/it] 44%|████▎ | 13113/30000 [8:23:30<10:51:14, 2.31s/it] 44%|████▎ | 13114/30000 [8:23:32<10:51:27, 2.31s/it] 44%|████▎ | 13115/30000 [8:23:35<10:49:39, 2.31s/it] 44%|████▎ | 13116/30000 [8:23:37<10:52:14, 2.32s/it] 44%|████▎ | 13117/30000 [8:23:39<10:56:30, 2.33s/it] 44%|████▎ | 13118/30000 [8:23:42<10:50:28, 2.31s/it] 44%|████▎ | 13119/30000 [8:23:44<10:46:22, 2.30s/it] 44%|████▎ | 13120/30000 [8:23:46<10:53:34, 2.32s/it] 44%|████▎ | 13120/30000 [8:23:46<10:53:34, 2.32s/it] 44%|████▎ | 13121/30000 [8:23:49<10:50:45, 2.31s/it] 44%|████▎ | 13122/30000 [8:23:51<10:48:50, 2.31s/it] 44%|████▎ | 13123/30000 [8:23:53<10:51:05, 2.31s/it] 44%|████▎ | 13124/30000 [8:23:56<10:51:45, 2.32s/it] 44%|████▍ | 13125/30000 [8:23:58<10:47:32, 2.30s/it] 44%|████▍ | 13126/30000 [8:24:00<10:51:31, 2.32s/it] 44%|████▍ | 13127/30000 [8:24:03<10:50:46, 2.31s/it] 44%|████▍ | 13128/30000 [8:24:05<10:46:34, 2.30s/it] 44%|████▍ | 13129/30000 [8:24:07<10:48:57, 2.31s/it] 44%|████▍ | 13130/30000 [8:24:09<10:48:07, 2.31s/it] 44%|████▍ | 13130/30000 [8:24:09<10:48:07, 2.31s/it] 44%|████▍ | 13131/30000 [8:24:12<10:41:20, 2.28s/it] 44%|████▍ | 13132/30000 [8:24:14<10:42:34, 2.29s/it] 44%|████▍ | 13133/30000 [8:24:16<10:47:11, 2.30s/it] 44%|████▍ | 13134/30000 [8:24:19<10:47:00, 2.30s/it] 44%|████▍ | 13135/30000 [8:24:21<10:45:58, 2.30s/it] 44%|████▍ | 13136/30000 [8:24:23<10:52:24, 2.32s/it] 44%|████▍ | 13137/30000 [8:24:26<10:52:16, 2.32s/it] 44%|████▍ | 13138/30000 [8:24:28<10:47:59, 2.31s/it] 44%|████▍ | 13139/30000 [8:24:30<10:52:46, 2.32s/it] 44%|████▍ | 13140/30000 [8:24:33<10:52:43, 2.32s/it] 44%|████▍ | 13140/30000 [8:24:33<10:52:43, 2.32s/it] 44%|████▍ | 13141/30000 [8:24:35<10:52:19, 2.32s/it] 44%|████▍ | 13142/30000 [8:24:37<10:48:03, 2.31s/it] 44%|████▍ | 13143/30000 [8:24:39<10:49:48, 2.31s/it] 44%|████▍ | 13144/30000 [8:24:42<10:45:39, 2.30s/it] 44%|████▍ | 13145/30000 [8:24:44<10:42:20, 2.29s/it] 44%|████▍ | 13146/30000 [8:24:46<10:48:04, 2.31s/it] 44%|████▍ | 13147/30000 [8:24:49<10:51:11, 2.32s/it] 44%|████▍ | 13148/30000 [8:24:51<10:47:55, 2.31s/it] 44%|████▍ | 13149/30000 [8:24:53<10:50:44, 2.32s/it] 44%|████▍ | 13150/30000 [8:24:56<10:46:27, 2.30s/it] 44%|████▍ | 13150/30000 [8:24:56<10:46:27, 2.30s/it] 44%|████▍ | 13151/30000 [8:24:58<10:43:10, 2.29s/it] 44%|████▍ | 13152/30000 [8:25:00<10:47:21, 2.31s/it] 44%|████▍ | 13153/30000 [8:25:02<10:46:48, 2.30s/it] 44%|████▍ | 13154/30000 [8:25:05<10:41:15, 2.28s/it] 44%|████▍ | 13155/30000 [8:25:07<10:43:57, 2.29s/it] 44%|████▍ | 13156/30000 [8:25:09<10:50:27, 2.32s/it] 44%|████▍ | 13157/30000 [8:25:12<10:49:41, 2.31s/it] 44%|████▍ | 13158/30000 [8:25:14<10:44:46, 2.30s/it] 44%|████▍ | 13159/30000 [8:25:16<10:45:45, 2.30s/it] 44%|████▍ | 13160/30000 [8:25:18<10:40:51, 2.28s/it] 44%|████▍ | 13160/30000 [8:25:19<10:40:51, 2.28s/it] 44%|████▍ | 13161/30000 [8:25:21<10:37:18, 2.27s/it] 44%|████▍ | 13162/30000 [8:25:23<10:44:04, 2.30s/it] 44%|████▍ | 13163/30000 [8:25:25<10:46:21, 2.30s/it] 44%|████▍ | 13164/30000 [8:25:28<10:45:06, 2.30s/it] 44%|████▍ | 13165/30000 [8:25:30<10:49:41, 2.32s/it] 44%|████▍ | 13166/30000 [8:25:32<10:52:26, 2.33s/it] 44%|████▍ | 13167/30000 [8:25:35<10:49:12, 2.31s/it] 44%|████▍ | 13168/30000 [8:25:37<10:48:00, 2.31s/it] 44%|████▍ | 13169/30000 [8:25:39<10:48:39, 2.31s/it] 44%|████▍ | 13170/30000 [8:25:42<10:43:58, 2.30s/it] 44%|████▍ | 13170/30000 [8:25:42<10:43:58, 2.30s/it] 44%|████▍ | 13171/30000 [8:25:44<10:43:20, 2.29s/it] 44%|████▍ | 13172/30000 [8:25:46<10:47:45, 2.31s/it] 44%|████▍ | 13173/30000 [8:25:48<10:45:59, 2.30s/it] 44%|████▍ | 13174/30000 [8:25:51<10:45:32, 2.30s/it] 44%|████▍ | 13175/30000 [8:25:53<10:50:21, 2.32s/it] 44%|████▍ | 13176/30000 [8:25:55<10:48:27, 2.31s/it] 44%|████▍ | 13177/30000 [8:25:58<10:45:06, 2.30s/it] 44%|████▍ | 13178/30000 [8:26:00<10:49:42, 2.32s/it] 44%|████▍ | 13179/30000 [8:26:02<10:50:20, 2.32s/it] 44%|████▍ | 13180/30000 [8:26:05<10:50:27, 2.32s/it] 44%|████▍ | 13180/30000 [8:26:05<10:50:27, 2.32s/it] 44%|████▍ | 13181/30000 [8:26:07<10:48:49, 2.31s/it] 44%|████▍ | 13182/30000 [8:26:09<10:51:40, 2.32s/it] 44%|████▍ | 13183/30000 [8:26:12<10:44:36, 2.30s/it] 44%|████▍ | 13184/30000 [8:26:14<10:37:25, 2.27s/it] 44%|████▍ | 13185/30000 [8:26:16<10:43:43, 2.30s/it] 44%|████▍ | 13186/30000 [8:26:18<10:40:03, 2.28s/it] 44%|████▍ | 13187/30000 [8:26:21<10:37:05, 2.27s/it] 44%|████▍ | 13188/30000 [8:26:23<10:44:15, 2.30s/it] 44%|████▍ | 13189/30000 [8:26:25<10:46:53, 2.31s/it] 44%|████▍ | 13190/30000 [8:26:28<10:43:16, 2.30s/it] 44%|████▍ | 13190/30000 [8:26:28<10:43:16, 2.30s/it] 44%|████▍ | 13191/30000 [8:26:30<10:47:48, 2.31s/it] 44%|████▍ | 13192/30000 [8:26:32<10:42:52, 2.29s/it] 44%|████▍ | 13193/30000 [8:26:34<10:40:00, 2.28s/it] 44%|████▍ | 13194/30000 [8:26:37<10:42:41, 2.29s/it] 44%|████▍ | 13195/30000 [8:26:39<10:43:17, 2.30s/it] 44%|████▍ | 13196/30000 [8:26:41<10:41:11, 2.29s/it] 44%|████▍ | 13197/30000 [8:26:44<10:39:09, 2.28s/it] 44%|████▍ | 13198/30000 [8:26:46<10:48:34, 2.32s/it] 44%|████▍ | 13199/30000 [8:26:48<10:50:24, 2.32s/it] 44%|████▍ | 13200/30000 [8:26:51<10:44:51, 2.30s/it] 44%|████▍ | 13200/30000 [8:26:51<10:44:51, 2.30s/it] 44%|████▍ | 13201/30000 [8:26:53<10:47:05, 2.31s/it] 44%|████▍ | 13202/30000 [8:26:55<10:40:28, 2.29s/it] 44%|████▍ | 13203/30000 [8:26:57<10:34:23, 2.27s/it] 44%|████▍ | 13204/30000 [8:27:00<10:41:34, 2.29s/it] 44%|████▍ | 13205/30000 [8:27:02<10:45:13, 2.31s/it] 44%|████▍ | 13206/30000 [8:27:04<10:42:58, 2.30s/it] 44%|████▍ | 13207/30000 [8:27:07<10:42:18, 2.29s/it] 44%|████▍ | 13208/30000 [8:27:09<10:46:40, 2.31s/it] 44%|████▍ | 13209/30000 [8:27:11<10:42:43, 2.30s/it] 44%|████▍ | 13210/30000 [8:27:14<10:43:26, 2.30s/it] 44%|████▍ | 13210/30000 [8:27:14<10:43:26, 2.30s/it] 44%|████▍ | 13211/30000 [8:27:16<10:47:36, 2.31s/it] 44%|████▍ | 13212/30000 [8:27:18<10:49:01, 2.32s/it] 44%|████▍ | 13213/30000 [8:27:21<10:46:55, 2.31s/it] 44%|████▍ | 13214/30000 [8:27:23<10:48:46, 2.32s/it] 44%|████▍ | 13215/30000 [8:27:25<10:47:20, 2.31s/it] 44%|████▍ | 13216/30000 [8:27:27<10:43:06, 2.30s/it] 44%|████▍ | 13217/30000 [8:27:30<10:39:07, 2.28s/it] 44%|████▍ | 13218/30000 [8:27:32<10:43:23, 2.30s/it] 44%|████▍ | 13219/30000 [8:27:34<10:42:18, 2.30s/it] 44%|████▍ | 13220/30000 [8:27:37<10:38:59, 2.28s/it] 44%|████▍ | 13220/30000 [8:27:37<10:38:59, 2.28s/it] 44%|████▍ | 13221/30000 [8:27:39<10:43:53, 2.30s/it] 44%|████▍ | 13222/30000 [8:27:41<10:44:36, 2.31s/it] 44%|████▍ | 13223/30000 [8:27:44<10:39:56, 2.29s/it] 44%|████▍ | 13224/30000 [8:27:46<10:40:27, 2.29s/it] 44%|████▍ | 13225/30000 [8:27:48<10:43:06, 2.30s/it] 44%|████▍ | 13226/30000 [8:27:50<10:40:37, 2.29s/it] 44%|████▍ | 13227/30000 [8:27:53<10:39:16, 2.29s/it] 44%|████▍ | 13228/30000 [8:27:55<10:45:32, 2.31s/it] 44%|████▍ | 13229/30000 [8:27:57<10:44:44, 2.31s/it] 44%|████▍ | 13230/30000 [8:28:00<10:42:43, 2.30s/it] 44%|████▍ | 13230/30000 [8:28:00<10:42:43, 2.30s/it] 44%|████▍ | 13231/30000 [8:28:02<10:46:03, 2.31s/it] 44%|████▍ | 13232/30000 [8:28:04<10:44:19, 2.31s/it] 44%|████▍ | 13233/30000 [8:28:06<10:38:01, 2.28s/it] 44%|████▍ | 13234/30000 [8:28:09<10:36:18, 2.28s/it] 44%|████▍ | 13235/30000 [8:28:11<10:41:21, 2.30s/it] 44%|████▍ | 13236/30000 [8:28:13<10:40:45, 2.29s/it] 44%|████▍ | 13237/30000 [8:28:16<10:41:04, 2.29s/it] 44%|████▍ | 13238/30000 [8:28:18<10:44:44, 2.31s/it] 44%|████▍ | 13239/30000 [8:28:20<10:43:16, 2.30s/it] 44%|████▍ | 13240/30000 [8:28:23<10:38:57, 2.29s/it] 44%|████▍ | 13240/30000 [8:28:23<10:38:57, 2.29s/it] 44%|████▍ | 13241/30000 [8:28:25<10:38:40, 2.29s/it] 44%|████▍ | 13242/30000 [8:28:27<10:46:48, 2.32s/it] 44%|████▍ | 13243/30000 [8:28:30<10:46:42, 2.32s/it] 44%|████▍ | 13244/30000 [8:28:32<10:39:46, 2.29s/it] 44%|████▍ | 13245/30000 [8:28:34<10:40:45, 2.29s/it] 44%|████▍ | 13246/30000 [8:28:36<10:46:29, 2.32s/it] 44%|████▍ | 13247/30000 [8:28:39<10:44:27, 2.31s/it] 44%|████▍ | 13248/30000 [8:28:41<10:40:49, 2.30s/it] 44%|████▍ | 13249/30000 [8:28:43<10:38:23, 2.29s/it] 44%|████▍ | 13250/30000 [8:28:46<10:36:43, 2.28s/it] 44%|████▍ | 13250/30000 [8:28:46<10:36:43, 2.28s/it] 44%|████▍ | 13251/30000 [8:28:48<10:34:47, 2.27s/it] 44%|████▍ | 13252/30000 [8:28:50<10:32:37, 2.27s/it] 44%|████▍ | 13253/30000 [8:28:52<10:38:57, 2.29s/it] 44%|████▍ | 13254/30000 [8:28:55<10:41:12, 2.30s/it] 44%|████▍ | 13255/30000 [8:28:57<10:37:47, 2.29s/it] 44%|████▍ | 13256/30000 [8:28:59<10:36:55, 2.28s/it] 44%|████▍ | 13257/30000 [8:29:02<10:40:48, 2.30s/it] 44%|████▍ | 13258/30000 [8:29:04<10:43:28, 2.31s/it] 44%|████▍ | 13259/30000 [8:29:06<10:40:01, 2.29s/it] 44%|████▍ | 13260/30000 [8:29:08<10:39:26, 2.29s/it] 44%|████▍ | 13260/30000 [8:29:08<10:39:26, 2.29s/it] 44%|████▍ | 13261/30000 [8:29:11<10:44:21, 2.31s/it] 44%|████▍ | 13262/30000 [8:29:13<10:44:23, 2.31s/it] 44%|████▍ | 13263/30000 [8:29:15<10:39:19, 2.29s/it] 44%|████▍ | 13264/30000 [8:29:18<10:37:55, 2.29s/it] 44%|████▍ | 13265/30000 [8:29:20<10:41:00, 2.30s/it] 44%|████▍ | 13266/30000 [8:29:22<10:36:09, 2.28s/it] 44%|████▍ | 13267/30000 [8:29:24<10:37:15, 2.29s/it] 44%|████▍ | 13268/30000 [8:29:27<10:43:27, 2.31s/it] 44%|████▍ | 13269/30000 [8:29:29<10:37:36, 2.29s/it] 44%|████▍ | 13270/30000 [8:29:31<10:36:18, 2.28s/it] +{'loss': 0.0145, 'grad_norm': 0.08600758016109467, 'learning_rate': 1.968397659829066e-05, 'flow_matching_loss': 0.014520786050707103, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.74} +{'loss': 0.014, 'grad_norm': 0.0809231773018837, 'learning_rate': 1.966826587511279e-05, 'flow_matching_loss': 0.0139972859993577, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.75} +{'loss': 0.015, 'grad_norm': 0.07978958636522293, 'learning_rate': 1.965254947955169e-05, 'flow_matching_loss': 0.014974929951131343, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.75} +{'loss': 0.0168, 'grad_norm': 0.059731822460889816, 'learning_rate': 1.963682743070427e-05, 'flow_matching_loss': 0.016817395016551016, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.75} +{'loss': 0.0143, 'grad_norm': 0.0708240196108818, 'learning_rate': 1.962109974767429e-05, 'flow_matching_loss': 0.014251423068344593, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.76} +{'loss': 0.0149, 'grad_norm': 0.09659311175346375, 'learning_rate': 1.960536644957238e-05, 'flow_matching_loss': 0.014946888294070959, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.76} +{'loss': 0.0153, 'grad_norm': 0.08539491891860962, 'learning_rate': 1.9589627555515974e-05, 'flow_matching_loss': 0.01531663304194808, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.76} +{'loss': 0.0156, 'grad_norm': 0.07342422753572464, 'learning_rate': 1.9573883084629317e-05, 'flow_matching_loss': 0.015591784007847309, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.76} +{'loss': 0.0154, 'grad_norm': 0.06626555323600769, 'learning_rate': 1.9558133056043422e-05, 'flow_matching_loss': 0.015442391578108072, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.77} +{'loss': 0.014, 'grad_norm': 0.06999494880437851, 'learning_rate': 1.9542377488896062e-05, 'flow_matching_loss': 0.014002919103950262, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.77} +{'loss': 0.0149, 'grad_norm': 0.05268583074212074, 'learning_rate': 1.9526616402331733e-05, 'flow_matching_loss': 0.014945900067687035, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.77} +{'loss': 0.0149, 'grad_norm': 0.08314146101474762, 'learning_rate': 1.9510849815501638e-05, 'flow_matching_loss': 0.014918960630893707, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.78} +{'loss': 0.015, 'grad_norm': 0.09538402408361435, 'learning_rate': 1.949507774756368e-05, 'flow_matching_loss': 0.01502603879198432, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.78} +{'loss': 0.0159, 'grad_norm': 0.07004507631063461, 'learning_rate': 1.9479300217682394e-05, 'flow_matching_loss': 0.015883041266351938, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.78} +{'loss': 0.0153, 'grad_norm': 0.07908450067043304, 'learning_rate': 1.946351724502898e-05, 'flow_matching_loss': 0.015340164862573146, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.79} +{'loss': 0.0149, 'grad_norm': 0.06623215973377228, 'learning_rate': 1.9447728848781227e-05, 'flow_matching_loss': 0.014865301176905632, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.79} +{'loss': 0.0147, 'grad_norm': 0.08199667185544968, 'learning_rate': 1.943193504812353e-05, 'flow_matching_loss': 0.01474988404661417, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.79} +{'loss': 0.0146, 'grad_norm': 0.06327107548713684, 'learning_rate': 1.9416135862246846e-05, 'flow_matching_loss': 0.014598784409463406, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.79} +{'loss': 0.014, 'grad_norm': 0.07784723490476608, 'learning_rate': 1.9400331310348673e-05, 'flow_matching_loss': 0.013952832203358412, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.8} +{'loss': 0.0164, 'grad_norm': 0.11010678857564926, 'learning_rate': 1.9384521411633033e-05, 'flow_matching_loss': 0.01637130118906498, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.8} +{'loss': 0.0157, 'grad_norm': 0.08306363224983215, 'learning_rate': 1.936870618531045e-05, 'flow_matching_loss': 0.015689415950328113, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.8} +{'loss': 0.0152, 'grad_norm': 0.08190848678350449, 'learning_rate': 1.9352885650597904e-05, 'flow_matching_loss': 0.015155722852796316, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.81} +{'loss': 0.0148, 'grad_norm': 0.10564404726028442, 'learning_rate': 1.9337059826718847e-05, 'flow_matching_loss': 0.014823535922914743, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.81} +{'loss': 0.0136, 'grad_norm': 0.08159254491329193, 'learning_rate': 1.932122873290314e-05, 'flow_matching_loss': 0.013628110382705927, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.81} +{'loss': 0.0156, 'grad_norm': 0.06844203174114227, 'learning_rate': 1.9305392388387054e-05, 'flow_matching_loss': 0.01560172438621521, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.81} +{'loss': 0.0157, 'grad_norm': 0.0820152759552002, 'learning_rate': 1.928955081241325e-05, 'flow_matching_loss': 0.015737169701606036, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.82} +{'loss': 0.0149, 'grad_norm': 0.11021903902292252, 'learning_rate': 1.9273704024230725e-05, 'flow_matching_loss': 0.014866608940064908, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.82} +{'loss': 0.0157, 'grad_norm': 0.0738174244761467, 'learning_rate': 1.925785204309483e-05, 'flow_matching_loss': 0.01565711610019207, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.82} +{'loss': 0.0145, 'grad_norm': 0.07491621375083923, 'learning_rate': 1.9241994888267212e-05, 'flow_matching_loss': 0.014496132358908653, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.83} +{'loss': 0.0158, 'grad_norm': 0.0811389908194542, 'learning_rate': 1.9226132579015815e-05, 'flow_matching_loss': 0.01579592386260629, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.83} +{'loss': 0.0174, 'grad_norm': 0.07972889393568039, 'learning_rate': 1.9210265134614832e-05, 'flow_matching_loss': 0.017404117342084646, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.83} +{'loss': 0.0154, 'grad_norm': 0.07584629207849503, 'learning_rate': 1.9194392574344712e-05, 'flow_matching_loss': 0.015355046186596156, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.83} +{'loss': 0.0148, 'grad_norm': 0.06667227298021317, 'learning_rate': 1.9178514917492103e-05, 'flow_matching_loss': 0.014824478514492511, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.84} +{'loss': 0.0168, 'grad_norm': 0.06817618757486343, 'learning_rate': 1.9162632183349866e-05, 'flow_matching_loss': 0.016842051409184933, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.84} +{'loss': 0.0151, 'grad_norm': 0.07528634369373322, 'learning_rate': 1.914674439121701e-05, 'flow_matching_loss': 0.01513356128707528, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.84} +{'loss': 0.0153, 'grad_norm': 0.08590903878211975, 'learning_rate': 1.9130851560398715e-05, 'flow_matching_loss': 0.01531507158651948, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.85} +{'loss': 0.0138, 'grad_norm': 0.05860443413257599, 'learning_rate': 1.911495371020625e-05, 'flow_matching_loss': 0.013833930902183055, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.85} +{'loss': 0.0154, 'grad_norm': 0.1030806228518486, 'learning_rate': 1.909905085995702e-05, 'flow_matching_loss': 0.015372515469789506, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.85} +{'loss': 0.0157, 'grad_norm': 0.08523275703191757, 'learning_rate': 1.908314302897448e-05, 'flow_matching_loss': 0.015737086720764636, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.86} +{'loss': 0.0161, 'grad_norm': 0.0719156339764595, 'learning_rate': 1.9067230236588142e-05, 'flow_matching_loss': 0.016124567296355963, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.86} +{'loss': 0.014, 'grad_norm': 0.10529834032058716, 'learning_rate': 1.905131250213355e-05, 'flow_matching_loss': 0.014044593367725611, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.86} 44%|████▍ | 13270/30000 [8:29:31<10:36:18, 2.28s/it] 44%|████▍ | 13271/30000 [8:29:34<10:35:40, 2.28s/it] 44%|████▍ | 13272/30000 [8:29:36<10:43:57, 2.31s/it] 44%|████▍ | 13273/30000 [8:29:38<10:44:17, 2.31s/it] 44%|████▍ | 13274/30000 [8:29:41<10:42:25, 2.30s/it] 44%|████▍ | 13275/30000 [8:29:43<10:41:10, 2.30s/it] 44%|████▍ | 13276/30000 [8:29:45<10:42:01, 2.30s/it] 44%|████▍ | 13277/30000 [8:29:47<10:40:00, 2.30s/it] 44%|████▍ | 13278/30000 [8:29:50<10:37:22, 2.29s/it] 44%|████▍ | 13279/30000 [8:29:52<10:40:16, 2.30s/it] 44%|████▍ | 13280/30000 [8:29:54<10:42:32, 2.31s/it] 44%|████▍ | 13280/30000 [8:29:54<10:42:32, 2.31s/it] 44%|████▍ | 13281/30000 [8:29:57<10:39:35, 2.30s/it] 44%|████▍ | 13282/30000 [8:29:59<10:43:01, 2.31s/it] 44%|████▍ | 13283/30000 [8:30:01<10:40:41, 2.30s/it] 44%|████▍ | 13284/30000 [8:30:04<10:37:17, 2.29s/it] 44%|████▍ | 13285/30000 [8:30:06<10:32:47, 2.27s/it] 44%|████▍ | 13286/30000 [8:30:08<10:39:44, 2.30s/it] 44%|████▍ | 13287/30000 [8:30:10<10:41:11, 2.30s/it] 44%|████▍ | 13288/30000 [8:30:13<10:37:13, 2.29s/it] 44%|████▍ | 13289/30000 [8:30:15<10:40:50, 2.30s/it] 44%|████▍ | 13290/30000 [8:30:17<10:39:55, 2.30s/it] 44%|████▍ | 13290/30000 [8:30:17<10:39:55, 2.30s/it] 44%|████▍ | 13291/30000 [8:30:20<10:39:07, 2.30s/it] 44%|████▍ | 13292/30000 [8:30:22<10:35:35, 2.28s/it] 44%|████▍ | 13293/30000 [8:30:24<10:41:22, 2.30s/it] 44%|████▍ | 13294/30000 [8:30:27<10:38:57, 2.29s/it] 44%|████▍ | 13295/30000 [8:30:29<10:35:22, 2.28s/it] 44%|████▍ | 13296/30000 [8:30:31<10:40:56, 2.30s/it] 44%|████▍ | 13297/30000 [8:30:33<10:41:32, 2.30s/it] 44%|████▍ | 13298/30000 [8:30:36<10:37:34, 2.29s/it] 44%|████▍ | 13299/30000 [8:30:38<10:38:18, 2.29s/it] 44%|████▍ | 13300/30000 [8:30:40<10:40:28, 2.30s/it] 44%|████▍ | 13300/30000 [8:30:40<10:40:28, 2.30s/it] 44%|████▍ | 13301/30000 [8:30:43<10:39:22, 2.30s/it] 44%|████▍ | 13302/30000 [8:30:45<10:39:42, 2.30s/it] 44%|████▍ | 13303/30000 [8:30:47<10:43:51, 2.31s/it] 44%|████▍ | 13304/30000 [8:30:50<10:42:37, 2.31s/it] 44%|████▍ | 13305/30000 [8:30:52<10:40:05, 2.30s/it] 44%|████▍ | 13306/30000 [8:30:54<10:41:12, 2.30s/it] 44%|████▍ | 13307/30000 [8:30:56<10:40:48, 2.30s/it] 44%|████▍ | 13308/30000 [8:30:59<10:38:50, 2.30s/it] 44%|████▍ | 13309/30000 [8:31:01<10:39:04, 2.30s/it] 44%|████▍ | 13310/30000 [8:31:03<10:41:12, 2.31s/it] 44%|████▍ | 13310/30000 [8:31:03<10:41:12, 2.31s/it] 44%|████▍ | 13311/30000 [8:31:06<10:36:43, 2.29s/it] 44%|████▍ | 13312/30000 [8:31:08<10:31:36, 2.27s/it] 44%|████▍ | 13313/30000 [8:31:10<10:37:39, 2.29s/it] 44%|████▍ | 13314/30000 [8:31:12<10:36:10, 2.29s/it] 44%|████▍ | 13315/30000 [8:31:15<10:32:29, 2.27s/it] 44%|████▍ | 13316/30000 [8:31:17<10:40:20, 2.30s/it] 44%|████▍ | 13317/30000 [8:31:19<10:42:43, 2.31s/it] 44%|████▍ | 13318/30000 [8:31:22<10:38:44, 2.30s/it] 44%|████▍ | 13319/30000 [8:31:24<10:41:00, 2.31s/it] 44%|████▍ | 13320/30000 [8:31:26<10:39:32, 2.30s/it] 44%|████▍ | 13320/30000 [8:31:26<10:39:32, 2.30s/it] 44%|████▍ | 13321/30000 [8:31:29<10:45:19, 2.32s/it] 44%|████▍ | 13322/30000 [8:31:31<10:41:14, 2.31s/it] 44%|████▍ | 13323/30000 [8:31:33<10:40:59, 2.31s/it] 44%|████▍ | 13324/30000 [8:31:35<10:33:13, 2.28s/it] 44%|████▍ | 13325/30000 [8:31:38<10:34:32, 2.28s/it] 44%|████▍ | 13326/30000 [8:31:40<10:39:39, 2.30s/it] 44%|████▍ | 13327/30000 [8:31:42<10:40:28, 2.30s/it] 44%|████▍ | 13328/30000 [8:31:45<10:35:24, 2.29s/it] 44%|████▍ | 13329/30000 [8:31:47<10:37:03, 2.29s/it] 44%|████▍ | 13330/30000 [8:31:49<10:33:28, 2.28s/it] 44%|████▍ | 13330/30000 [8:31:49<10:33:28, 2.28s/it] 44%|████▍ | 13331/30000 [8:31:51<10:29:56, 2.27s/it] 44%|████▍ | 13332/30000 [8:31:54<10:36:33, 2.29s/it] 44%|████▍ | 13333/30000 [8:31:56<10:34:57, 2.29s/it] 44%|████▍ | 13334/30000 [8:31:58<10:32:48, 2.28s/it] 44%|████▍ | 13335/30000 [8:32:01<10:40:25, 2.31s/it] 44%|████▍ | 13336/30000 [8:32:03<10:37:27, 2.30s/it] 44%|████▍ | 13337/30000 [8:32:05<10:30:58, 2.27s/it] 44%|████▍ | 13338/30000 [8:32:07<10:31:14, 2.27s/it] 44%|████▍ | 13339/30000 [8:32:10<10:35:17, 2.29s/it] 44%|████▍ | 13340/30000 [8:32:12<10:33:47, 2.28s/it] 44%|████▍ | 13340/30000 [8:32:12<10:33:47, 2.28s/it] 44%|████▍ | 13341/30000 [8:32:14<10:40:59, 2.31s/it] 44%|████▍ | 13342/30000 [8:32:17<10:38:17, 2.30s/it] 44%|████▍ | 13343/30000 [8:32:19<10:28:31, 2.26s/it] 44%|████▍ | 13344/30000 [8:32:21<10:35:40, 2.29s/it] 44%|████▍ | 13345/30000 [8:32:23<10:32:33, 2.28s/it] 44%|████▍ | 13346/30000 [8:32:26<10:33:38, 2.28s/it] 44%|████▍ | 13347/30000 [8:32:28<10:35:36, 2.29s/it] 44%|████▍ | 13348/30000 [8:32:30<10:41:33, 2.31s/it] 44%|████▍ | 13349/30000 [8:32:33<10:41:05, 2.31s/it] 44%|████▍ | 13350/30000 [8:32:35<10:38:51, 2.30s/it] 44%|████▍ | 13350/30000 [8:32:35<10:38:51, 2.30s/it] 45%|████▍ | 13351/30000 [8:32:37<10:42:26, 2.32s/it] 45%|████▍ | 13352/30000 [8:32:40<10:41:27, 2.31s/it] 45%|████▍ | 13353/30000 [8:32:42<10:39:06, 2.30s/it] 45%|████▍ | 13354/30000 [8:32:44<10:41:16, 2.31s/it] 45%|████▍ | 13355/30000 [8:32:47<10:37:30, 2.30s/it] 45%|████▍ | 13356/30000 [8:32:49<10:40:02, 2.31s/it] 45%|████▍ | 13357/30000 [8:32:51<10:43:45, 2.32s/it] 45%|████▍ | 13358/30000 [8:32:54<10:47:03, 2.33s/it] 45%|████▍ | 13359/30000 [8:32:56<10:42:50, 2.32s/it] 45%|████▍ | 13360/30000 [8:32:58<10:45:53, 2.33s/it] 45%|████▍ | 13360/30000 [8:32:58<10:45:53, 2.33s/it] 45%|████▍ | 13361/30000 [8:33:00<10:41:33, 2.31s/it] 45%|████▍ | 13362/30000 [8:33:03<10:30:19, 2.27s/it] 45%|████▍ | 13363/30000 [8:33:05<10:35:00, 2.29s/it] 45%|████▍ | 13364/30000 [8:33:07<10:42:56, 2.32s/it] 45%|████▍ | 13365/30000 [8:33:10<10:48:31, 2.34s/it] 45%|████▍ | 13366/30000 [8:33:12<10:45:03, 2.33s/it] 45%|████▍ | 13367/30000 [8:33:14<10:43:22, 2.32s/it] 45%|████▍ | 13368/30000 [8:33:17<10:38:40, 2.30s/it] 45%|████▍ | 13369/30000 [8:33:19<10:36:15, 2.30s/it] 45%|████▍ | 13370/30000 [8:33:21<10:37:31, 2.30s/it] 45%|████▍ | 13370/30000 [8:33:21<10:37:31, 2.30s/it] 45%|████▍ | 13371/30000 [8:33:23<10:35:15, 2.29s/it] 45%|████▍ | 13372/30000 [8:33:26<10:38:50, 2.31s/it] 45%|████▍ | 13373/30000 [8:33:28<10:40:38, 2.31s/it] 45%|████▍ | 13374/30000 [8:33:30<10:40:07, 2.31s/it] 45%|████▍ | 13375/30000 [8:33:33<10:34:54, 2.29s/it] 45%|████▍ | 13376/30000 [8:33:35<10:37:26, 2.30s/it] 45%|████▍ | 13377/30000 [8:33:37<10:32:54, 2.28s/it] 45%|████▍ | 13378/30000 [8:33:40<10:33:37, 2.29s/it] 45%|████▍ | 13379/30000 [8:33:42<10:34:39, 2.29s/it] 45%|████▍ | 13380/30000 [8:33:44<10:28:35, 2.27s/it] 45%|████▍ | 13380/30000 [8:33:44<10:28:35, 2.27s/it] 45%|████▍ | 13381/30000 [8:33:46<10:34:09, 2.29s/it] 45%|████▍ | 13382/30000 [8:33:49<10:48:26, 2.34s/it] 45%|████▍ | 13383/30000 [8:33:51<10:43:01, 2.32s/it] 45%|████▍ | 13384/30000 [8:33:53<10:39:48, 2.31s/it] 45%|████▍ | 13385/30000 [8:33:56<10:42:26, 2.32s/it] 45%|████▍ | 13386/30000 [8:33:58<10:40:16, 2.31s/it] 45%|████▍ | 13387/30000 [8:34:00<10:35:15, 2.29s/it] 45%|████▍ | 13388/30000 [8:34:03<10:39:05, 2.31s/it] 45%|████▍ | 13389/30000 [8:34:05<10:43:06, 2.32s/it] 45%|████▍ | 13390/30000 [8:34:07<10:38:28, 2.31s/it] 45%|████▍ | 13390/30000 [8:34:07<10:38:28, 2.31s/it] 45%|████▍ | 13391/30000 [8:34:10<10:33:46, 2.29s/it] 45%|████▍ | 13392/30000 [8:34:12<10:37:28, 2.30s/it] 45%|████▍ | 13393/30000 [8:34:14<10:36:13, 2.30s/it] 45%|████▍ | 13394/30000 [8:34:16<10:36:03, 2.30s/it] 45%|████▍ | 13395/30000 [8:34:19<10:39:54, 2.31s/it] 45%|████▍ | 13396/30000 [8:34:21<10:40:21, 2.31s/it] 45%|████▍ | 13397/30000 [8:34:23<10:26:20, 2.26s/it] 45%|████▍ | 13398/30000 [8:34:26<10:32:13, 2.28s/it] 45%|████▍ | 13399/30000 [8:34:28<10:34:07, 2.29s/it] 45%|████▍ | 13400/30000 [8:34:30<10:25:24, 2.26s/it] 45%|████▍ | 13400/30000 [8:34:30<10:25:24, 2.26s/it] 45%|████▍ | 13401/30000 [8:34:32<10:29:36, 2.28s/it] 45%|████▍ | 13402/30000 [8:34:35<10:33:49, 2.29s/it] 45%|████▍ | 13403/30000 [8:34:37<10:32:11, 2.29s/it] 45%|████▍ | 13404/30000 [8:34:39<10:35:36, 2.30s/it] 45%|████▍ | 13405/30000 [8:34:42<10:37:10, 2.30s/it] 45%|████▍ | 13406/30000 [8:34:44<10:22:19, 2.25s/it] 45%|████▍ | 13407/30000 [8:34:46<10:33:36, 2.29s/it] 45%|████▍ | 13408/30000 [8:34:48<10:33:22, 2.29s/it] 45%|████▍ | 13409/30000 [8:34:51<10:29:36, 2.28s/it] 45%|████▍ | 13410/30000 [8:34:53<10:29:20, 2.28s/it] 45%|████▍ | 13410/30000 [8:34:53<10:29:20, 2.28s/it] 45%|████▍ | 13411/30000 [8:34:55<10:33:54, 2.29s/it] 45%|████▍ | 13412/30000 [8:34:58<10:32:50, 2.29s/it] 45%|████▍ | 13413/30000 [8:35:00<10:32:04, 2.29s/it] 45%|████▍ | 13414/30000 [8:35:02<10:36:01, 2.30s/it] 45%|████▍ | 13415/30000 [8:35:05<10:40:38, 2.32s/it] 45%|████▍ | 13416/30000 [8:35:07<10:40:10, 2.32s/it] 45%|████▍ | 13417/30000 [8:35:09<10:35:53, 2.30s/it] 45%|████▍ | 13418/30000 [8:35:11<10:36:24, 2.30s/it] 45%|████▍ | 13419/30000 [8:35:14<10:39:38, 2.31s/it] 45%|████▍ | 13420/30000 [8:35:16<10:35:44, 2.30s/it] 45%|████▍ | 13420/30000 [8:35:16<10:35:44, 2.30s/it] 45%|████▍ | 13421/30000 [8:35:18<10:37:41, 2.31s/it] 45%|████▍ | 13422/30000 [8:35:21<10:37:59, 2.31s/it] 45%|████▍ | 13423/30000 [8:35:23<10:34:48, 2.30s/it] 45%|████▍ | 13424/30000 [8:35:25<10:27:03, 2.27s/it] 45%|████▍ | 13425/30000 [8:35:27<10:26:47, 2.27s/it] 45%|████▍ | 13426/30000 [8:35:30<10:24:58, 2.26s/it] 45%|████▍ | 13427/30000 [8:35:32<10:23:05, 2.26s/it] 45%|████▍ | 13428/30000 [8:35:34<10:30:13, 2.28s/it] 45%|████▍ | 13429/30000 [8:35:37<10:34:01, 2.30s/it] 45%|████▍ | 13430/30000 [8:35:39<10:27:14, 2.27s/it] 45%|████▍ | 13430/30000 [8:35:39<10:27:14, 2.27s/it] 45%|████▍ | 13431/30000 [8:35:41<10:34:13, 2.30s/it] 45%|████▍ | 13432/30000 [8:35:43<10:27:05, 2.27s/it] 45%|████▍ | 13433/30000 [8:35:46<10:25:44, 2.27s/it] 45%|████▍ | 13434/30000 [8:35:48<10:29:04, 2.28s/it] 45%|████▍ | 13435/30000 [8:35:50<10:33:11, 2.29s/it] 45%|████▍ | 13436/30000 [8:35:53<10:31:02, 2.29s/it] 45%|████▍ | 13437/30000 [8:35:55<10:27:50, 2.27s/it] 45%|████▍ | 13438/30000 [8:35:57<10:35:36, 2.30s/it] 45%|████▍ | 13439/30000 [8:35:59<10:33:01, 2.29s/it] 45%|████▍ | 13440/30000 [8:36:02<10:23:27, 2.26s/it] 45%|████▍ | 13440/30000 [8:36:02<10:23:27, 2.26s/it] 45%|████▍ | 13441/30000 [8:36:04<10:34:07, 2.30s/it] 45%|████▍ | 13442/30000 [8:36:06<10:38:09, 2.31s/it] 45%|████▍ | 13443/30000 [8:36:09<10:37:37, 2.31s/it] 45%|████▍ | 13444/30000 [8:36:11<10:36:19, 2.31s/it] 45%|████▍ | 13445/30000 [8:36:13<10:38:05, 2.31s/it] 45%|████▍ | 13446/30000 [8:36:16<10:34:04, 2.30s/it] 45%|████▍ | 13447/30000 [8:36:18<10:31:03, 2.29s/it] 45%|████▍ | 13448/30000 [8:36:20<10:33:55, 2.30s/it] 45%|████▍ | 13449/30000 [8:36:22<10:30:57, 2.29s/it] 45%|████▍ | 13450/30000 [8:36:25<10:27:53, 2.28s/it] 45%|████▍ | 13450/30000 [8:36:25<10:27:53, 2.28s/it] 45%|████▍ | 13451/30000 [8:36:27<10:36:04, 2.31s/it] 45%|████▍ | 13452/30000 [8:36:29<10:35:30, 2.30s/it] 45%|████▍ | 13453/30000 [8:36:32<10:31:41, 2.29s/it] 45%|████▍ | 13454/30000 [8:36:34<10:31:18, 2.29s/it] 45%|████▍ | 13455/30000 [8:36:36<10:34:37, 2.30s/it] 45%|████▍ | 13456/30000 [8:36:38<10:34:40, 2.30s/it] 45%|████▍ | 13457/30000 [8:36:41<10:29:57, 2.28s/it] 45%|████▍ | 13458/30000 [8:36:43<10:35:20, 2.30s/it] 45%|████▍ | 13459/30000 [8:36:45<10:36:32, 2.31s/it] 45%|████▍ | 13460/30000 [8:36:48<10:34:36, 2.30s/it] 45%|████▍ | 13460/30000 [8:36:48<10:34:36, 2.30s/it] 45%|████▍ | 13461/30000 [8:36:50<10:39:46, 2.32s/it] 45%|████▍ | 13462/30000 [8:36:52<10:44:20, 2.34s/it] 45%|████▍ | 13463/30000 [8:36:55<10:38:34, 2.32s/it] 45%|████▍ | 13464/30000 [8:36:57<10:33:04, 2.30s/it] 45%|████▍ | 13465/30000 [8:36:59<10:37:58, 2.31s/it] 45%|████▍ | 13466/30000 [8:37:02<10:41:23, 2.33s/it] 45%|████▍ | 13467/30000 [8:37:04<10:49:13, 2.36s/it] 45%|████▍ | 13468/30000 [8:37:06<10:45:55, 2.34s/it] 45%|████▍ | 13469/30000 [8:37:09<10:55:16, 2.38s/it] 45%|████▍ | 13470/30000 [8:37:11<10:51:27, 2.36s/it] 45%|████▍ | 13470/30000 [8:37:11<10:51:27, 2.36s/it] 45%|████▍ | 13471/30000 [8:37:13<10:42:37, 2.33s/it] 45%|████▍ | 13472/30000 [8:37:16<10:36:48, 2.31s/it] 45%|████▍ | 13473/30000 [8:37:18<10:36:43, 2.31s/it] 45%|████▍ | 13474/30000 [8:37:20<10:35:06, 2.31s/it] 45%|████▍ | 13475/30000 [8:37:23<10:34:44, 2.30s/it] 45%|████▍ | 13476/30000 [8:37:25<10:37:13, 2.31s/it] 45%|████▍ | 13477/30000 [8:37:27<10:36:17, 2.31s/it] 45%|████▍ | 13478/30000 [8:37:30<10:35:26, 2.31s/it] 45%|████▍ | 13479/30000 [8:37:32<10:32:11, 2.30s/it] 45%|████▍ | 13480/30000 [8:37:34<10:31:24, 2.29s/it] 45%|████▍ | 13480/30000 [8:37:34<10:31:24, 2.29s/it] 45%|████▍ | 13481/30000 [8:37:36<10:36:22, 2.31s/it] 45%|████▍ | 13482/30000 [8:37:39<10:45:24, 2.34s/it] 45%|████▍ | 13483/30000 [8:37:41<10:39:10, 2.32s/it] 45%|████▍ | 13484/30000 [8:37:43<10:34:43, 2.31s/it] 45%|████▍ | 13485/30000 [8:37:46<10:36:05, 2.31s/it] 45%|████▍ | 13486/30000 [8:37:48<10:40:59, 2.33s/it] 45%|████▍ | 13487/30000 [8:37:50<10:43:03, 2.34s/it] 45%|████▍ | 13488/30000 [8:37:53<10:38:58, 2.32s/it] 45%|████▍ | 13489/30000 [8:37:55<10:32:28, 2.30s/it] 45%|████▍ | 13490/30000 [8:37:57<10:26:39, 2.28s/it] 45%|████▍ | 13490/30000 [8:37:57<10:26:39, 2.28s/it] 45%|████▍ | 13491/30000 [8:38:00<10:25:51, 2.27s/it] 45%|████▍ | 13492/30000 [8:38:02<10:26:43, 2.28s/it] 45%|████▍ | 13493/30000 [8:38:04<10:37:54, 2.32s/it] 45%|████▍ | 13494/30000 [8:38:06<10:35:01, 2.31s/it] 45%|████▍ | 13495/30000 [8:38:09<10:17:48, 2.25s/it] 45%|████▍ | 13496/30000 [8:38:11<10:17:21, 2.24s/it] 45%|████▍ | 13497/30000 [8:38:13<10:20:11, 2.25s/it] 45%|████▍ | 13498/30000 [8:38:15<10:21:58, 2.26s/it] 45%|████▍ | 13499/30000 [8:38:18<10:30:33, 2.29s/it] 45%|████▌ | 13500/30000 [8:38:20<10:30:48, 2.29s/it] 45%|████▌ | 13500/30000 [8:38:20<10:30:48, 2.29s/it] 45%|████▌ | 13501/30000 [8:38:22<10:29:58, 2.29s/it] 45%|████▌ | 13502/30000 [8:38:25<10:30:04, 2.29s/it] 45%|████▌ | 13503/30000 [8:38:27<10:29:24, 2.29s/it] 45%|████▌ | 13504/30000 [8:38:29<10:20:08, 2.26s/it] 45%|████▌ | 13505/30000 [8:38:31<10:26:24, 2.28s/it] 45%|████▌ | 13506/30000 [8:38:34<10:28:12, 2.29s/it] 45%|████▌ | 13507/30000 [8:38:36<10:26:46, 2.28s/it] 45%|████▌ | 13508/30000 [8:38:38<10:31:51, 2.30s/it] 45%|████▌ | 13509/30000 [8:38:41<10:38:37, 2.32s/it] 45%|████▌ | 13510/30000 [8:38:43<10:42:44, 2.34s/it] 45%|████▌ | 13510/30000 [8:38:43<10:42:44, 2.34s/it] 45%|████▌ | 13511/30000 [8:38:45<10:37:57, 2.32s/it] 45%|████▌ | 13512/30000 [8:38:48<10:31:38, 2.30s/it] 45%|████▌ | 13513/30000 [8:38:50<10:33:55, 2.31s/it] 45%|████▌ | 13514/30000 [8:38:52<10:28:56, 2.29s/it] 45%|████▌ | 13515/30000 [8:38:54<10:25:50, 2.28s/it] 45%|████▌ | 13516/30000 [8:38:57<10:33:19, 2.31s/it] 45%|████▌ | 13517/30000 [8:38:59<10:36:53, 2.32s/it] 45%|████▌ | 13518/30000 [8:39:01<10:32:10, 2.30s/it] 45%|████▌ | 13519/30000 [8:39:04<10:35:35, 2.31s/it] 45%|████▌ | 13520/30000 [8:39:06<10:28:31, 2.29s/it] 45%|████▌ | 13520/30000 [8:39:06<10:28:31, 2.29s/it] 45%|████▌ | 13521/30000 [8:39:08<10:21:22, 2.26s/it] 45%|████▌ | 13522/30000 [8:39:11<10:31:39, 2.30s/it] 45%|████▌ | 13523/30000 [8:39:13<10:39:06, 2.33s/it] 45%|████▌ | 13524/30000 [8:39:15<10:27:03, 2.28s/it] 45%|████▌ | 13525/30000 [8:39:17<10:32:23, 2.30s/it] 45%|████▌ | 13526/30000 [8:39:20<10:30:55, 2.30s/it] 45%|████▌ | 13527/30000 [8:39:22<10:28:56, 2.29s/it] 45%|████▌ | 13528/30000 [8:39:24<10:28:36, 2.29s/it] 45%|████▌ | 13529/30000 [8:39:27<10:29:56, 2.29s/it] 45%|████▌ | 13530/30000 [8:39:29<10:07:08, 2.21s/it] 45%|████▌ | 13530/30000 [8:39:29<10:07:08, 2.21s/it] 45%|████▌ | 13531/30000 [8:39:31<10:17:44, 2.25s/it] 45%|████▌ | 13532/30000 [8:39:33<10:19:11, 2.26s/it] 45%|████▌ | 13533/30000 [8:39:36<10:19:53, 2.26s/it] 45%|████▌ | 13534/30000 [8:39:38<10:29:48, 2.29s/it] 45%|████▌ | 13535/30000 [8:39:40<10:33:30, 2.31s/it] 45%|████▌ | 13536/30000 [8:39:43<10:28:21, 2.29s/it] 45%|████▌ | 13537/30000 [8:39:45<10:30:30, 2.30s/it] 45%|████▌ | 13538/30000 [8:39:47<10:32:43, 2.31s/it] 45%|████▌ | 13539/30000 [8:39:50<10:38:46, 2.33s/it] 45%|████▌ | 13540/30000 [8:39:52<10:32:12, 2.30s/it] 45%|████▌ | 13540/30000 [8:39:52<10:32:12, 2.30s/it] 45%|████▌ | 13541/30000 [8:39:54<10:42:44, 2.34s/it] 45%|████▌ | 13542/30000 [8:39:57<10:38:39, 2.33s/it] 45%|████▌ | 13543/30000 [8:39:59<10:41:34, 2.34s/it] 45%|████▌ | 13544/30000 [8:40:01<10:41:58, 2.34s/it] 45%|████▌ | 13545/30000 [8:40:04<10:40:19, 2.33s/it] 45%|████▌ | 13546/30000 [8:40:06<10:32:03, 2.30s/it] 45%|████▌ | 13547/30000 [8:40:08<10:34:15, 2.31s/it] 45%|████▌ | 13548/30000 [8:40:10<10:40:37, 2.34s/it] 45%|████▌ | 13549/30000 [8:40:13<10:32:17, 2.31s/it] 45%|████▌ | 13550/30000 [8:40:15<10:34:55, 2.32s/it] 45%|████▌ | 13550/30000 [8:40:15<10:34:55, 2.32s/it] 45%|████▌ | 13551/30000 [8:40:17<10:36:23, 2.32s/it] 45%|████▌ | 13552/30000 [8:40:20<10:29:52, 2.30s/it] 45%|████▌ | 13553/30000 [8:40:22<10:27:34, 2.29s/it] 45%|████▌ | 13554/30000 [8:40:24<10:41:07, 2.34s/it] 45%|████▌ | 13555/30000 [8:40:27<10:34:03, 2.31s/it] 45%|████▌ | 13556/30000 [8:40:29<10:30:40, 2.30s/it] 45%|████▌ | 13557/30000 [8:40:31<10:37:49, 2.33s/it] 45%|████▌ | 13558/30000 [8:40:34<10:36:13, 2.32s/it] 45%|████▌ | 13559/30000 [8:40:36<10:32:28, 2.31s/it] 45%|████▌ | 13560/30000 [8:40:38<10:36:33, 2.32s/it] 45%|████▌ | 13560/30000 [8:40:38<10:36:33, 2.32s/it] 45%|████▌ | 13561/30000 [8:40:41<10:42:11, 2.34s/it] 45%|████▌ | 13562/30000 [8:40:43<10:30:12, 2.30s/it] 45%|████▌ | 13563/30000 [8:40:45<10:35:34, 2.32s/it] 45%|████▌ | 13564/30000 [8:40:48<10:37:19, 2.33s/it] 45%|████▌ | 13565/30000 [8:40:50<10:33:57, 2.31s/it] 45%|████▌ | 13566/30000 [8:40:52<10:38:51, 2.33s/it] 45%|████▌ | 13567/30000 [8:40:54<10:32:32, 2.31s/it] 45%|████▌ | 13568/30000 [8:40:57<10:31:39, 2.31s/it] 45%|████▌ | 13569/30000 [8:40:59<10:32:32, 2.31s/it] 45%|████▌ | 13570/30000 [8:41:01<10:22:19, 2.27s/it] 45%|████▌ | 13570/30000 [8:41:01<10:22:19, 2.27s/it] 45%|████▌ | 13571/30000 [8:41:03<10:18:04, 2.26s/it] 45%|████▌ | 13572/30000 [8:41:06<10:26:21, 2.29s/it] 45%|████▌ | 13573/30000 [8:41:08<10:26:56, 2.29s/it] 45%|████▌ | 13574/30000 [8:41:10<10:24:08, 2.28s/it] 45%|████▌ | 13575/30000 [8:41:13<10:28:12, 2.29s/it] 45%|████▌ | 13576/30000 [8:41:15<10:33:30, 2.31s/it] 45%|████▌ | 13577/30000 [8:41:17<10:19:47, 2.26s/it] 45%|████▌ | 13578/30000 [8:41:20<10:28:40, 2.30s/it] 45%|████▌ | 13579/30000 [8:41:22<10:30:29, 2.30s/it] 45%|████▌ | 13580/30000 [8:41:24<10:29:04, 2.30s/it] 45%|████▌ | 13580/30000 [8:41:24<10:29:04, 2.30s/it] 45%|████▌ | 13581/30000 [8:41:27<10:32:45, 2.31s/it] 45%|████▌ | 13582/30000 [8:41:29<10:25:10, 2.28s/it] 45%|████▌ | 13583/30000 [8:41:31<10:14:51, 2.25s/it] 45%|████▌ | 13584/30000 [8:41:33<10:19:21, 2.26s/it] 45%|████▌ | 13585/30000 [8:41:35<10:17:48, 2.26s/it] 45%|████▌ | 13586/30000 [8:41:38<10:20:24, 2.27s/it] 45%|████▌ | 13587/30000 [8:41:40<10:24:33, 2.28s/it] 45%|████▌ | 13588/30000 [8:41:42<10:26:57, 2.29s/it] 45%|████▌ | 13589/30000 [8:41:45<10:26:21, 2.29s/it] 45%|████▌ | 13590/30000 [8:41:47<10:22:37, 2.28s/it] 45%|████▌ | 13590/30000 [8:41:47<10:22:37, 2.28s/it] 45%|████▌ | 13591/30000 [8:41:49<10:25:47, 2.29s/it] 45%|████▌ | 13592/30000 [8:41:52<10:23:57, 2.28s/it] 45%|████▌ | 13593/30000 [8:41:54<10:23:15, 2.28s/it] 45%|████▌ | 13594/30000 [8:41:56<10:38:29, 2.34s/it] 45%|████▌ | 13595/30000 [8:41:59<10:32:37, 2.31s/it] 45%|████▌ | 13596/30000 [8:42:01<10:31:13, 2.31s/it] 45%|████▌ | 13597/30000 [8:42:03<10:24:30, 2.28s/it] 45%|████▌ | 13598/30000 [8:42:05<10:31:35, 2.31s/it] 45%|████▌ | 13599/30000 [8:42:08<10:30:59, 2.31s/it] 45%|████▌ | 13600/30000 [8:42:10<10:29:59, 2.30s/it] 45%|████▌ | 13600/30000 [8:42:10<10:29:59, 2.30s/it] 45%|████▌ | 13601/30000 [8:42:12<10:32:05, 2.31s/it] 45%|████▌ | 13602/30000 [8:42:15<10:35:07, 2.32s/it] 45%|████▌ | 13603/30000 [8:42:17<10:30:54, 2.31s/it] 45%|████▌ | 13604/30000 [8:42:19<10:28:44, 2.30s/it] 45%|████▌ | 13605/30000 [8:42:22<10:28:12, 2.30s/it] 45%|████▌ | 13606/30000 [8:42:24<10:37:30, 2.33s/it] 45%|████▌ | 13607/30000 [8:42:26<10:45:14, 2.36s/it] 45%|████▌ | 13608/30000 [8:42:29<10:36:57, 2.33s/it] 45%|████▌ | 13609/30000 [8:42:31<10:36:24, 2.33s/it] 45%|████▌ | 13610/30000 [8:42:33<10:34:46, 2.32s/it] 45%|████▌ | 13610/30000 [8:42:33<10:34:46, 2.32s/it] 45%|████▌ | 13611/30000 [8:42:36<10:30:00, 2.31s/it] 45%|████▌ | 13612/30000 [8:42:38<10:29:45, 2.31s/it] 45%|████▌ | 13613/30000 [8:42:40<10:35:32, 2.33s/it] 45%|████▌ | 13614/30000 [8:42:43<10:36:51, 2.33s/it] 45%|████▌ | 13615/30000 [8:42:45<10:41:22, 2.35s/it] 45%|████▌ | 13616/30000 [8:42:47<10:35:45, 2.33s/it] 45%|████▌ | 13617/30000 [8:42:50<10:38:16, 2.34s/it] 45%|████▌ | 13618/30000 [8:42:52<10:37:18, 2.33s/it] 45%|████▌ | 13619/30000 [8:42:54<10:20:35, 2.27s/it] 45%|████▌ | 13620/30000 [8:42:56<10:28:42, 2.30s/it] 45%|████▌ | 13620/30000 [8:42:56<10:28:42, 2.30s/it] 45%|████▌ | 13621/30000 [8:42:59<10:32:59, 2.32s/it] 45%|████▌ | 13622/30000 [8:43:01<10:31:38, 2.31s/it] 45%|████▌ | 13623/30000 [8:43:03<10:39:26, 2.34s/it] 45%|████▌ | 13624/30000 [8:43:06<10:35:56, 2.33s/it] 45%|████▌ | 13625/30000 [8:43:08<10:34:28, 2.32s/it] 45%|████▌ | 13626/30000 [8:43:10<10:29:09, 2.31s/it] 45%|████▌ | 13627/30000 [8:43:13<10:31:57, 2.32s/it] 45%|████▌ | 13628/30000 [8:43:15<10:34:47, 2.33s/it] 45%|████▌ | 13629/30000 [8:43:17<10:29:45, 2.31s/it] 45%|████▌ | 13630/30000 [8:43:20<10:25:05, 2.29s/it] 45%|████▌ | 13630/30000 [8:43:20<10:25:05, 2.29s/it] 45%|████▌ | 13631/30000 [8:43:22<10:28:07, 2.30s/it] 45%|████▌ | 13632/30000 [8:43:24<10:29:36, 2.31s/it] 45%|████▌ | 13633/30000 [8:43:27<10:30:48, 2.31s/it] 45%|████▌ | 13634/30000 [8:43:29<10:25:15, 2.29s/it] 45%|████▌ | 13635/30000 [8:43:31<10:24:57, 2.29s/it] 45%|████▌ | 13636/30000 [8:43:33<10:18:22, 2.27s/it] 45%|████▌ | 13637/30000 [8:43:36<10:22:34, 2.28s/it] 45%|████▌ | 13638/30000 [8:43:38<10:29:15, 2.31s/it] 45%|████▌ | 13639/30000 [8:43:40<10:32:20, 2.32s/it] 45%|████▌ | 13640/30000 [8:43:43<10:23:39, 2.29s/it] 45%|████▌ | 13640/30000 [8:43:43<10:23:39, 2.29s/it] 45%|████▌ | 13641/30000 [8:43:45<10:25:49, 2.30s/it] 45%|████▌ | 13642/30000 [8:43:47<10:26:41, 2.30s/it] 45%|████▌ | 13643/30000 [8:43:49<10:22:47, 2.28s/it] 45%|████▌ | 13644/30000 [8:43:52<10:20:00, 2.27s/it] 45%|████▌ | 13645/30000 [8:43:54<10:27:59, 2.30s/it] 45%|████▌ | 13646/30000 [8:43:56<10:30:48, 2.31s/it] 45%|████▌ | 13647/30000 [8:43:59<10:37:10, 2.34s/it] 45%|████▌ | 13648/30000 [8:44:01<10:22:47, 2.29s/it] 45%|████▌ | 13649/30000 [8:44:03<10:27:23, 2.30s/it] 46%|████▌ | 13650/30000 [8:44:05<10:16:29, 2.26s/it] 46%|████▌ | 13650/30000 [8:44:05<10:16:29, 2.26s/it] 46%|████▌ | 13651/30000 [8:44:08<10:20:14, 2.28s/it] 46%|████▌ | 13652/30000 [8:44:10<10:33:22, 2.32s/it] 46%|████▌ | 13653/30000 [8:44:13<10:41:33, 2.35s/it] 46%|████▌ | 13654/30000 [8:44:15<10:31:32, 2.32s/it] 46%|████▌ | 13655/30000 [8:44:17<10:23:34, 2.29s/it] 46%|████▌ | 13656/30000 [8:44:19<10:36:05, 2.34s/it] 46%|████▌ | 13657/30000 [8:44:22<10:42:04, 2.36s/it] 46%|████▌ | 13658/30000 [8:44:24<10:34:40, 2.33s/it] 46%|████▌ | 13659/30000 [8:44:26<10:33:22, 2.33s/it] 46%|████▌ | 13660/30000 [8:44:29<10:31:36, 2.32s/it] 46%|████▌ | 13660/30000 [8:44:29<10:31:36, 2.32s/it] 46%|████▌ | 13661/30000 [8:44:31<10:29:11, 2.31s/it] 46%|████▌ | 13662/30000 [8:44:33<10:28:07, 2.31s/it] 46%|████▌ | 13663/30000 [8:44:36<10:30:18, 2.31s/it] 46%|████▌ | 13664/30000 [8:44:38<10:33:10, 2.33s/it] 46%|████▌ | 13665/30000 [8:44:40<10:25:07, 2.30s/it] 46%|████▌ | 13666/30000 [8:44:43<10:23:04, 2.29s/it] 46%|████▌ | 13667/30000 [8:44:45<10:26:16, 2.30s/it] 46%|████▌ | 13668/30000 [8:44:47<10:33:34, 2.33s/it] 46%|████▌ | 13669/30000 [8:44:50<10:31:21, 2.32s/it] 46%|████▌ | 13670/30000 [8:44:52<10:29:22, 2.31s/it] 46%|████▌ | 13670/30000 [8:44:52<10:29:22, 2.31s/it] 46%|████▌ | 13671/30000 [8:44:54<10:29:11, 2.31s/it] 46%|████▌ | 13672/30000 [8:44:56<10:17:23, 2.27s/it] 46%|████▌ | 13673/30000 [8:44:59<10:18:04, 2.27s/it] 46%|████▌ | 13674/30000 [8:45:01<10:21:51, 2.29s/it] 46%|████▌ | 13675/30000 [8:45:03<10:22:38, 2.29s/it] 46%|████▌ | 13676/30000 [8:45:06<10:21:40, 2.29s/it] 46%|████▌ | 13677/30000 [8:45:08<10:28:31, 2.31s/it] 46%|████▌ | 13678/30000 [8:45:10<10:29:24, 2.31s/it] 46%|████▌ | 13679/30000 [8:45:12<10:25:43, 2.30s/it] 46%|████▌ | 13680/30000 [8:45:15<10:26:53, 2.30s/it] +{'loss': 0.0146, 'grad_norm': 0.08640825003385544, 'learning_rate': 1.9035389844952256e-05, 'flow_matching_loss': 0.014580122940242291, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.86} +{'loss': 0.0179, 'grad_norm': 0.08943463861942291, 'learning_rate': 1.901946228439179e-05, 'flow_matching_loss': 0.01789060728624463, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.87} +{'loss': 0.0149, 'grad_norm': 0.07823865860700607, 'learning_rate': 1.9003529839805637e-05, 'flow_matching_loss': 0.014929715357720853, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.87} +{'loss': 0.0141, 'grad_norm': 0.05171496793627739, 'learning_rate': 1.8987592530553217e-05, 'flow_matching_loss': 0.01413062671199441, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.87} +{'loss': 0.0174, 'grad_norm': 0.09016017615795135, 'learning_rate': 1.897165037599987e-05, 'flow_matching_loss': 0.01737849349156022, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.88} +{'loss': 0.0145, 'grad_norm': 0.06757014989852905, 'learning_rate': 1.8955703395516814e-05, 'flow_matching_loss': 0.014546248503029347, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.88} +{'loss': 0.0159, 'grad_norm': 0.08513949066400528, 'learning_rate': 1.8939751608481134e-05, 'flow_matching_loss': 0.015926550049334763, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.88} +{'loss': 0.0152, 'grad_norm': 0.073252372443676, 'learning_rate': 1.892379503427575e-05, 'flow_matching_loss': 0.015233403258025646, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.88} +{'loss': 0.0152, 'grad_norm': 0.07889381051063538, 'learning_rate': 1.8907833692289412e-05, 'flow_matching_loss': 0.01518319258466363, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.89} +{'loss': 0.0153, 'grad_norm': 0.06803280115127563, 'learning_rate': 1.889186760191666e-05, 'flow_matching_loss': 0.01534176552668214, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.89} +{'loss': 0.0158, 'grad_norm': 0.0681178942322731, 'learning_rate': 1.887589678255779e-05, 'flow_matching_loss': 0.015776305738836528, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.89} +{'loss': 0.0152, 'grad_norm': 0.07514775544404984, 'learning_rate': 1.8859921253618852e-05, 'flow_matching_loss': 0.015209975559264422, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.9} +{'loss': 0.0161, 'grad_norm': 0.10330091416835785, 'learning_rate': 1.884394103451163e-05, 'flow_matching_loss': 0.01608736077323556, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.9} +{'loss': 0.0143, 'grad_norm': 0.06634590029716492, 'learning_rate': 1.8827956144653583e-05, 'flow_matching_loss': 0.014346883445978165, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.9} +{'loss': 0.0146, 'grad_norm': 0.07627879828214645, 'learning_rate': 1.8811966603467874e-05, 'flow_matching_loss': 0.014579633716493845, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.9} +{'loss': 0.0163, 'grad_norm': 0.10773918777704239, 'learning_rate': 1.8795972430383293e-05, 'flow_matching_loss': 0.016319816652685403, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.91} +{'loss': 0.0144, 'grad_norm': 0.09140139818191528, 'learning_rate': 1.8779973644834273e-05, 'flow_matching_loss': 0.014405209477990866, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.91} +{'loss': 0.0139, 'grad_norm': 0.08395938575267792, 'learning_rate': 1.876397026626085e-05, 'flow_matching_loss': 0.01390032097697258, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.91} +{'loss': 0.0145, 'grad_norm': 0.0703798308968544, 'learning_rate': 1.8747962314108633e-05, 'flow_matching_loss': 0.01453215116634965, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.92} +{'loss': 0.0149, 'grad_norm': 0.07989762723445892, 'learning_rate': 1.873194980782879e-05, 'flow_matching_loss': 0.014888463728129863, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.92} +{'loss': 0.0141, 'grad_norm': 0.06615065783262253, 'learning_rate': 1.871593276687804e-05, 'flow_matching_loss': 0.014136711228638887, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.92} +{'loss': 0.0142, 'grad_norm': 0.0751226618885994, 'learning_rate': 1.869991121071858e-05, 'flow_matching_loss': 0.014244797639548779, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.92} +{'loss': 0.0156, 'grad_norm': 0.07030288130044937, 'learning_rate': 1.8683885158818117e-05, 'flow_matching_loss': 0.01558488653972745, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.93} +{'loss': 0.0137, 'grad_norm': 0.07937522232532501, 'learning_rate': 1.8667854630649825e-05, 'flow_matching_loss': 0.013695767801254987, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.93} +{'loss': 0.0157, 'grad_norm': 0.0662912130355835, 'learning_rate': 1.86518196456923e-05, 'flow_matching_loss': 0.015679743234068157, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.93} +{'loss': 0.0154, 'grad_norm': 0.11082117259502411, 'learning_rate': 1.8635780223429565e-05, 'flow_matching_loss': 0.015433083847165107, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.94} +{'loss': 0.0161, 'grad_norm': 0.080501027405262, 'learning_rate': 1.8619736383351017e-05, 'flow_matching_loss': 0.016111167799681424, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.94} +{'loss': 0.0144, 'grad_norm': 0.055343423038721085, 'learning_rate': 1.8603688144951454e-05, 'flow_matching_loss': 0.014421588554978371, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.94} +{'loss': 0.0153, 'grad_norm': 0.08445049077272415, 'learning_rate': 1.8587635527730983e-05, 'flow_matching_loss': 0.015340539254248142, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.95} +{'loss': 0.0147, 'grad_norm': 0.08315341174602509, 'learning_rate': 1.8571578551195056e-05, 'flow_matching_loss': 0.014655168913304806, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.95} +{'loss': 0.0132, 'grad_norm': 0.07376855611801147, 'learning_rate': 1.855551723485441e-05, 'flow_matching_loss': 0.0132493382319808, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.95} +{'loss': 0.0148, 'grad_norm': 0.08220674842596054, 'learning_rate': 1.8539451598225066e-05, 'flow_matching_loss': 0.014801244996488095, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.95} +{'loss': 0.0155, 'grad_norm': 0.07899833470582962, 'learning_rate': 1.8523381660828276e-05, 'flow_matching_loss': 0.015522986184805632, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.96} +{'loss': 0.0154, 'grad_norm': 0.09519131481647491, 'learning_rate': 1.850730744219054e-05, 'flow_matching_loss': 0.015388775616884232, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.96} +{'loss': 0.0142, 'grad_norm': 0.07668602466583252, 'learning_rate': 1.849122896184354e-05, 'flow_matching_loss': 0.01417188048362732, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.96} +{'loss': 0.0164, 'grad_norm': 0.07650433480739594, 'learning_rate': 1.8475146239324146e-05, 'flow_matching_loss': 0.01639600405469537, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.97} +{'loss': 0.0135, 'grad_norm': 0.07764936983585358, 'learning_rate': 1.8459059294174396e-05, 'flow_matching_loss': 0.013482035044580698, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.97} +{'loss': 0.0156, 'grad_norm': 0.07656349241733551, 'learning_rate': 1.844296814594143e-05, 'flow_matching_loss': 0.015584578178822995, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.97} +{'loss': 0.0147, 'grad_norm': 0.10196517407894135, 'learning_rate': 1.8426872814177516e-05, 'flow_matching_loss': 0.01471948130056262, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.97} +{'loss': 0.0149, 'grad_norm': 0.07997869700193405, 'learning_rate': 1.8410773318440003e-05, 'flow_matching_loss': 0.014948063902556897, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.98} +{'loss': 0.014, 'grad_norm': 0.08206649124622345, 'learning_rate': 1.839466967829129e-05, 'flow_matching_loss': 0.01395276077091694, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.98} 46%|████▌ | 13680/30000 [8:45:15<10:26:53, 2.30s/it] 46%|████▌ | 13681/30000 [8:45:17<10:20:48, 2.28s/it] 46%|████▌ | 13682/30000 [8:45:19<10:18:35, 2.27s/it] 46%|████▌ | 13683/30000 [8:45:22<10:18:04, 2.27s/it] 46%|████▌ | 13684/30000 [8:45:24<10:22:48, 2.29s/it] 46%|████▌ | 13685/30000 [8:45:26<10:25:18, 2.30s/it] 46%|████▌ | 13686/30000 [8:45:29<10:25:08, 2.30s/it] 46%|████▌ | 13687/30000 [8:45:31<10:27:31, 2.31s/it] 46%|████▌ | 13688/30000 [8:45:33<10:23:15, 2.29s/it] 46%|████▌ | 13689/30000 [8:45:35<10:19:36, 2.28s/it] 46%|████▌ | 13690/30000 [8:45:38<10:18:24, 2.27s/it] 46%|████▌ | 13690/30000 [8:45:38<10:18:24, 2.27s/it] 46%|████▌ | 13691/30000 [8:45:40<10:31:59, 2.33s/it] 46%|████▌ | 13692/30000 [8:45:42<10:31:33, 2.32s/it] 46%|████▌ | 13693/30000 [8:45:45<10:27:27, 2.31s/it] 46%|████▌ | 13694/30000 [8:45:47<10:25:40, 2.30s/it] 46%|████▌ | 13695/30000 [8:45:49<10:31:19, 2.32s/it] 46%|████▌ | 13696/30000 [8:45:52<10:27:44, 2.31s/it] 46%|████▌ | 13697/30000 [8:45:54<10:25:48, 2.30s/it] 46%|████▌ | 13698/30000 [8:45:56<10:26:13, 2.30s/it] 46%|████▌ | 13699/30000 [8:45:58<10:24:42, 2.30s/it] 46%|████▌ | 13700/30000 [8:46:01<10:25:24, 2.30s/it] 46%|████▌ | 13700/30000 [8:46:01<10:25:24, 2.30s/it] 46%|████▌ | 13701/30000 [8:46:03<10:25:28, 2.30s/it] 46%|████▌ | 13702/30000 [8:46:05<10:20:23, 2.28s/it] 46%|████▌ | 13703/30000 [8:46:08<10:16:56, 2.27s/it] 46%|████▌ | 13704/30000 [8:46:10<10:19:20, 2.28s/it] 46%|████▌ | 13705/30000 [8:46:12<10:23:40, 2.30s/it] 46%|████▌ | 13706/30000 [8:46:14<10:18:58, 2.28s/it] 46%|████▌ | 13707/30000 [8:46:17<10:15:15, 2.27s/it] 46%|████▌ | 13708/30000 [8:46:19<10:26:47, 2.31s/it] 46%|████▌ | 13709/30000 [8:46:21<10:24:57, 2.30s/it] 46%|████▌ | 13710/30000 [8:46:24<10:29:03, 2.32s/it] 46%|████▌ | 13710/30000 [8:46:24<10:29:03, 2.32s/it] 46%|████▌ | 13711/30000 [8:46:26<10:23:12, 2.30s/it] 46%|████▌ | 13712/30000 [8:46:28<10:25:25, 2.30s/it] 46%|████▌ | 13713/30000 [8:46:31<10:23:21, 2.30s/it] 46%|████▌ | 13714/30000 [8:46:33<10:15:55, 2.27s/it] 46%|████▌ | 13715/30000 [8:46:35<10:22:32, 2.29s/it] 46%|████▌ | 13716/30000 [8:46:37<10:25:22, 2.30s/it] 46%|████▌ | 13717/30000 [8:46:40<10:22:17, 2.29s/it] 46%|████▌ | 13718/30000 [8:46:42<10:16:14, 2.27s/it] 46%|████▌ | 13719/30000 [8:46:44<10:13:49, 2.26s/it] 46%|████▌ | 13720/30000 [8:46:46<10:11:28, 2.25s/it] 46%|████▌ | 13720/30000 [8:46:46<10:11:28, 2.25s/it] 46%|████▌ | 13721/30000 [8:46:49<10:11:16, 2.25s/it] 46%|████▌ | 13722/30000 [8:46:51<10:16:52, 2.27s/it] 46%|████▌ | 13723/30000 [8:46:53<10:18:36, 2.28s/it] 46%|████▌ | 13724/30000 [8:46:56<10:17:52, 2.28s/it] 46%|████▌ | 13725/30000 [8:46:58<10:28:20, 2.32s/it] 46%|████▌ | 13726/30000 [8:47:00<10:26:18, 2.31s/it] 46%|████▌ | 13727/30000 [8:47:02<10:21:26, 2.29s/it] 46%|████▌ | 13728/30000 [8:47:05<10:17:24, 2.28s/it] 46%|████▌ | 13729/30000 [8:47:07<10:22:54, 2.30s/it] 46%|████▌ | 13730/30000 [8:47:09<10:19:18, 2.28s/it] 46%|████▌ | 13730/30000 [8:47:09<10:19:18, 2.28s/it] 46%|████▌ | 13731/30000 [8:47:12<10:27:46, 2.32s/it] 46%|████▌ | 13732/30000 [8:47:14<10:26:08, 2.31s/it] 46%|████▌ | 13733/30000 [8:47:16<10:26:33, 2.31s/it] 46%|████▌ | 13734/30000 [8:47:19<10:24:21, 2.30s/it] 46%|████▌ | 13735/30000 [8:47:21<10:21:36, 2.29s/it] 46%|████▌ | 13736/30000 [8:47:23<10:15:07, 2.27s/it] 46%|████▌ | 13737/30000 [8:47:25<10:21:04, 2.29s/it] 46%|████▌ | 13738/30000 [8:47:28<10:18:53, 2.28s/it] 46%|████▌ | 13739/30000 [8:47:30<10:24:11, 2.30s/it] 46%|████▌ | 13740/30000 [8:47:32<10:22:33, 2.30s/it] 46%|████▌ | 13740/30000 [8:47:32<10:22:33, 2.30s/it] 46%|████▌ | 13741/30000 [8:47:35<10:21:15, 2.29s/it] 46%|████▌ | 13742/30000 [8:47:37<10:20:33, 2.29s/it] 46%|████▌ | 13743/30000 [8:47:39<10:23:03, 2.30s/it] 46%|████▌ | 13744/30000 [8:47:42<10:25:08, 2.31s/it] 46%|████▌ | 13745/30000 [8:47:44<10:23:06, 2.30s/it] 46%|████▌ | 13746/30000 [8:47:46<10:18:32, 2.28s/it] 46%|████▌ | 13747/30000 [8:47:48<10:21:15, 2.29s/it] 46%|████▌ | 13748/30000 [8:47:49<8:27:44, 1.87s/it] 46%|████▌ | 13749/30000 [8:48:07<30:30:06, 6.76s/it] 46%|████▌ | 13750/30000 [8:48:10<24:22:01, 5.40s/it] 46%|████▌ | 13750/30000 [8:48:10<24:22:01, 5.40s/it] 46%|████▌ | 13751/30000 [8:48:12<20:15:57, 4.49s/it] 46%|████▌ | 13752/30000 [8:48:14<17:20:11, 3.84s/it] 46%|████▌ | 13753/30000 [8:48:17<15:19:02, 3.39s/it] 46%|████▌ | 13754/30000 [8:48:19<13:51:39, 3.07s/it] 46%|████▌ | 13755/30000 [8:48:21<12:45:23, 2.83s/it] 46%|████▌ | 13756/30000 [8:48:24<12:16:28, 2.72s/it] 46%|████▌ | 13757/30000 [8:48:26<11:38:32, 2.58s/it] 46%|████▌ | 13758/30000 [8:48:28<11:19:48, 2.51s/it] 46%|████▌ | 13759/30000 [8:48:31<11:01:58, 2.45s/it] 46%|████▌ | 13760/30000 [8:48:33<10:45:44, 2.39s/it] 46%|████▌ | 13760/30000 [8:48:33<10:45:44, 2.39s/it] 46%|████▌ | 13761/30000 [8:48:35<10:42:43, 2.37s/it] 46%|████▌ | 13762/30000 [8:48:38<10:38:17, 2.36s/it] 46%|████▌ | 13763/30000 [8:48:40<10:31:08, 2.33s/it] 46%|████▌ | 13764/30000 [8:48:42<10:30:54, 2.33s/it] 46%|████▌ | 13765/30000 [8:48:44<10:28:53, 2.32s/it] 46%|████▌ | 13766/30000 [8:48:47<10:31:05, 2.33s/it] 46%|████▌ | 13767/30000 [8:48:49<10:21:34, 2.30s/it] 46%|████▌ | 13768/30000 [8:48:51<10:21:39, 2.30s/it] 46%|████▌ | 13769/30000 [8:48:54<10:18:40, 2.29s/it] 46%|████▌ | 13770/30000 [8:48:56<10:21:59, 2.30s/it] 46%|████▌ | 13770/30000 [8:48:56<10:21:59, 2.30s/it] 46%|████▌ | 13771/30000 [8:48:58<10:26:13, 2.32s/it] 46%|████▌ | 13772/30000 [8:49:01<10:26:57, 2.32s/it] 46%|████▌ | 13773/30000 [8:49:03<10:20:44, 2.30s/it] 46%|████▌ | 13774/30000 [8:49:05<10:24:47, 2.31s/it] 46%|████▌ | 13775/30000 [8:49:08<10:27:51, 2.32s/it] 46%|████▌ | 13776/30000 [8:49:10<10:21:16, 2.30s/it] 46%|████▌ | 13777/30000 [8:49:12<10:12:18, 2.26s/it] 46%|████▌ | 13778/30000 [8:49:14<10:16:44, 2.28s/it] 46%|████▌ | 13779/30000 [8:49:17<10:19:31, 2.29s/it] 46%|████▌ | 13780/30000 [8:49:19<10:20:17, 2.29s/it] 46%|████▌ | 13780/30000 [8:49:19<10:20:17, 2.29s/it] 46%|████▌ | 13781/30000 [8:49:21<10:19:18, 2.29s/it] 46%|████▌ | 13782/30000 [8:49:24<10:21:55, 2.30s/it] 46%|████▌ | 13783/30000 [8:49:26<10:30:26, 2.33s/it] 46%|████▌ | 13784/30000 [8:49:28<10:25:22, 2.31s/it] 46%|████▌ | 13785/30000 [8:49:31<10:37:42, 2.36s/it] 46%|████▌ | 13786/30000 [8:49:33<10:34:11, 2.35s/it] 46%|████▌ | 13787/30000 [8:49:35<10:29:43, 2.33s/it] 46%|████▌ | 13788/30000 [8:49:38<10:26:48, 2.32s/it] 46%|████▌ | 13789/30000 [8:49:40<10:26:46, 2.32s/it] 46%|████▌ | 13790/30000 [8:49:42<10:23:28, 2.31s/it] 46%|████▌ | 13790/30000 [8:49:42<10:23:28, 2.31s/it] 46%|████▌ | 13791/30000 [8:49:44<10:21:45, 2.30s/it] 46%|████▌ | 13792/30000 [8:49:47<10:19:03, 2.29s/it] 46%|████▌ | 13793/30000 [8:49:49<10:35:01, 2.35s/it] 46%|████▌ | 13794/30000 [8:49:52<10:32:30, 2.34s/it] 46%|████▌ | 13795/30000 [8:49:54<10:35:37, 2.35s/it] 46%|████▌ | 13796/30000 [8:49:56<10:28:46, 2.33s/it] 46%|████▌ | 13797/30000 [8:49:59<10:30:19, 2.33s/it] 46%|████▌ | 13798/30000 [8:50:01<10:24:45, 2.31s/it] 46%|████▌ | 13799/30000 [8:50:03<10:21:25, 2.30s/it] 46%|████▌ | 13800/30000 [8:50:05<10:25:44, 2.32s/it] 46%|████▌ | 13800/30000 [8:50:05<10:25:44, 2.32s/it] 46%|████▌ | 13801/30000 [8:50:08<10:26:11, 2.32s/it] 46%|████▌ | 13802/30000 [8:50:10<10:33:21, 2.35s/it] 46%|████▌ | 13803/30000 [8:50:12<10:30:05, 2.33s/it] 46%|████▌ | 13804/30000 [8:50:15<10:31:28, 2.34s/it] 46%|████▌ | 13805/30000 [8:50:17<10:31:21, 2.34s/it] 46%|████▌ | 13806/30000 [8:50:19<10:27:10, 2.32s/it] 46%|████▌ | 13807/30000 [8:50:22<10:28:57, 2.33s/it] 46%|████▌ | 13808/30000 [8:50:24<10:23:02, 2.31s/it] 46%|████▌ | 13809/30000 [8:50:26<10:22:16, 2.31s/it] 46%|████▌ | 13810/30000 [8:50:29<10:22:31, 2.31s/it] 46%|████▌ | 13810/30000 [8:50:29<10:22:31, 2.31s/it] 46%|████▌ | 13811/30000 [8:50:31<10:27:05, 2.32s/it] 46%|████▌ | 13812/30000 [8:50:33<10:21:02, 2.30s/it] 46%|████▌ | 13813/30000 [8:50:36<10:15:35, 2.28s/it] 46%|████▌ | 13814/30000 [8:50:38<10:17:53, 2.29s/it] 46%|████▌ | 13815/30000 [8:50:40<10:14:21, 2.28s/it] 46%|████▌ | 13816/30000 [8:50:42<10:13:32, 2.27s/it] 46%|████▌ | 13817/30000 [8:50:45<10:21:26, 2.30s/it] 46%|████▌ | 13818/30000 [8:50:47<10:24:14, 2.31s/it] 46%|████▌ | 13819/30000 [8:50:49<10:30:47, 2.34s/it] 46%|████▌ | 13820/30000 [8:50:52<10:26:27, 2.32s/it] 46%|████▌ | 13820/30000 [8:50:52<10:26:27, 2.32s/it] 46%|████▌ | 13821/30000 [8:50:54<10:27:36, 2.33s/it] 46%|████▌ | 13822/30000 [8:50:56<10:29:58, 2.34s/it] 46%|████▌ | 13823/30000 [8:50:59<10:25:01, 2.32s/it] 46%|████▌ | 13824/30000 [8:51:01<10:22:26, 2.31s/it] 46%|████▌ | 13825/30000 [8:51:03<10:20:25, 2.30s/it] 46%|████▌ | 13826/30000 [8:51:06<10:16:45, 2.29s/it] 46%|████▌ | 13827/30000 [8:51:08<10:18:22, 2.29s/it] 46%|████▌ | 13828/30000 [8:51:10<10:15:38, 2.28s/it] 46%|████▌ | 13829/30000 [8:51:12<10:07:41, 2.25s/it] 46%|████▌ | 13830/30000 [8:51:15<10:09:18, 2.26s/it] 46%|████▌ | 13830/30000 [8:51:15<10:09:18, 2.26s/it] 46%|████▌ | 13831/30000 [8:51:17<10:12:38, 2.27s/it] 46%|████▌ | 13832/30000 [8:51:19<10:15:44, 2.29s/it] 46%|████▌ | 13833/30000 [8:51:21<10:12:05, 2.27s/it] 46%|████▌ | 13834/30000 [8:51:24<10:16:46, 2.29s/it] 46%|████▌ | 13835/30000 [8:51:26<10:13:32, 2.28s/it] 46%|████▌ | 13836/30000 [8:51:28<10:09:36, 2.26s/it] 46%|████▌ | 13837/30000 [8:51:30<10:07:45, 2.26s/it] 46%|████▌ | 13838/30000 [8:51:33<10:13:55, 2.28s/it] 46%|████▌ | 13839/30000 [8:51:35<10:13:25, 2.28s/it] 46%|████▌ | 13840/30000 [8:51:37<10:13:18, 2.28s/it] 46%|████▌ | 13840/30000 [8:51:37<10:13:18, 2.28s/it] 46%|████▌ | 13841/30000 [8:51:40<10:22:12, 2.31s/it] 46%|████▌ | 13842/30000 [8:51:42<10:29:43, 2.34s/it] 46%|████▌ | 13843/30000 [8:51:44<10:14:48, 2.28s/it] 46%|████▌ | 13844/30000 [8:51:47<10:18:39, 2.30s/it] 46%|████▌ | 13845/30000 [8:51:49<10:17:51, 2.29s/it] 46%|████▌ | 13846/30000 [8:51:51<10:16:18, 2.29s/it] 46%|████▌ | 13847/30000 [8:51:53<10:16:23, 2.29s/it] 46%|████▌ | 13848/30000 [8:51:56<10:16:21, 2.29s/it] 46%|████▌ | 13849/30000 [8:51:58<10:16:54, 2.29s/it] 46%|████▌ | 13850/30000 [8:52:00<10:11:00, 2.27s/it] 46%|████▌ | 13850/30000 [8:52:00<10:11:00, 2.27s/it] 46%|████▌ | 13851/30000 [8:52:03<10:13:15, 2.28s/it] 46%|████▌ | 13852/30000 [8:52:05<10:11:44, 2.27s/it] 46%|████▌ | 13853/30000 [8:52:07<10:09:35, 2.27s/it] 46%|████▌ | 13854/30000 [8:52:09<10:13:38, 2.28s/it] 46%|████▌ | 13855/30000 [8:52:12<10:15:02, 2.29s/it] 46%|████▌ | 13856/30000 [8:52:14<10:16:03, 2.29s/it] 46%|████▌ | 13857/30000 [8:52:16<10:16:50, 2.29s/it] 46%|████▌ | 13858/30000 [8:52:19<10:21:37, 2.31s/it] 46%|████▌ | 13859/30000 [8:52:21<10:21:17, 2.31s/it] 46%|████▌ | 13860/30000 [8:52:23<10:14:46, 2.29s/it] 46%|████▌ | 13860/30000 [8:52:23<10:14:46, 2.29s/it] 46%|████▌ | 13861/30000 [8:52:26<10:18:53, 2.30s/it] 46%|████▌ | 13862/30000 [8:52:28<10:21:59, 2.31s/it] 46%|████▌ | 13863/30000 [8:52:30<10:15:55, 2.29s/it] 46%|████▌ | 13864/30000 [8:52:32<10:15:00, 2.29s/it] 46%|████▌ | 13865/30000 [8:52:35<10:18:10, 2.30s/it] 46%|████▌ | 13866/30000 [8:52:37<10:17:31, 2.30s/it] 46%|████▌ | 13867/30000 [8:52:39<10:19:24, 2.30s/it] 46%|████▌ | 13868/30000 [8:52:42<10:22:51, 2.32s/it] 46%|████▌ | 13869/30000 [8:52:44<10:19:42, 2.31s/it] 46%|████▌ | 13870/30000 [8:52:46<10:22:08, 2.31s/it] 46%|████▌ | 13870/30000 [8:52:46<10:22:08, 2.31s/it] 46%|████▌ | 13871/30000 [8:52:49<10:22:51, 2.32s/it] 46%|████▌ | 13872/30000 [8:52:51<10:16:32, 2.29s/it] 46%|████▌ | 13873/30000 [8:52:53<10:09:11, 2.27s/it] 46%|████▌ | 13874/30000 [8:52:55<10:17:57, 2.30s/it] 46%|████▋ | 13875/30000 [8:52:58<10:17:43, 2.30s/it] 46%|████▋ | 13876/30000 [8:53:00<10:14:41, 2.29s/it] 46%|████▋ | 13877/30000 [8:53:02<10:13:52, 2.28s/it] 46%|████▋ | 13878/30000 [8:53:05<10:24:45, 2.33s/it] 46%|████▋ | 13879/30000 [8:53:07<10:22:34, 2.32s/it] 46%|████▋ | 13880/30000 [8:53:09<10:18:49, 2.30s/it] 46%|████▋ | 13880/30000 [8:53:09<10:18:49, 2.30s/it] 46%|████▋ | 13881/30000 [8:53:12<10:21:07, 2.31s/it] 46%|████▋ | 13882/30000 [8:53:14<10:23:17, 2.32s/it] 46%|████▋ | 13883/30000 [8:53:16<10:19:38, 2.31s/it] 46%|████▋ | 13884/30000 [8:53:18<10:18:32, 2.30s/it] 46%|████▋ | 13885/30000 [8:53:21<10:16:57, 2.30s/it] 46%|████▋ | 13886/30000 [8:53:23<10:10:14, 2.27s/it] 46%|████▋ | 13887/30000 [8:53:25<10:11:48, 2.28s/it] 46%|████▋ | 13888/30000 [8:53:28<10:08:29, 2.27s/it] 46%|████▋ | 13889/30000 [8:53:30<10:08:03, 2.26s/it] 46%|████▋ | 13890/30000 [8:53:32<10:12:00, 2.28s/it] 46%|████▋ | 13890/30000 [8:53:32<10:12:00, 2.28s/it] 46%|████▋ | 13891/30000 [8:53:34<10:08:09, 2.27s/it] 46%|████▋ | 13892/30000 [8:53:37<10:09:54, 2.27s/it] 46%|████▋ | 13893/30000 [8:53:39<10:08:56, 2.27s/it] 46%|████▋ | 13894/30000 [8:53:41<10:10:01, 2.27s/it] 46%|████▋ | 13895/30000 [8:53:43<10:11:12, 2.28s/it] 46%|████▋ | 13896/30000 [8:53:46<10:15:35, 2.29s/it] 46%|████▋ | 13897/30000 [8:53:48<10:12:37, 2.28s/it] 46%|████▋ | 13898/30000 [8:53:50<10:15:50, 2.29s/it] 46%|████▋ | 13899/30000 [8:53:53<10:16:27, 2.30s/it] 46%|████▋ | 13900/30000 [8:53:55<10:15:47, 2.29s/it] 46%|████▋ | 13900/30000 [8:53:55<10:15:47, 2.29s/it] 46%|████▋ | 13901/30000 [8:53:57<10:21:38, 2.32s/it] 46%|████▋ | 13902/30000 [8:54:00<10:11:51, 2.28s/it] 46%|████▋ | 13903/30000 [8:54:02<10:14:49, 2.29s/it] 46%|████▋ | 13904/30000 [8:54:04<10:12:10, 2.28s/it] 46%|████▋ | 13905/30000 [8:54:06<10:07:51, 2.27s/it] 46%|████▋ | 13906/30000 [8:54:09<10:14:07, 2.29s/it] 46%|████▋ | 13907/30000 [8:54:11<10:19:57, 2.31s/it] 46%|████▋ | 13908/30000 [8:54:13<10:14:14, 2.29s/it] 46%|████▋ | 13909/30000 [8:54:16<10:15:08, 2.29s/it] 46%|████▋ | 13910/30000 [8:54:18<10:13:05, 2.29s/it] 46%|████▋ | 13910/30000 [8:54:18<10:13:05, 2.29s/it] 46%|████▋ | 13911/30000 [8:54:20<10:10:20, 2.28s/it] 46%|████▋ | 13912/30000 [8:54:22<10:12:21, 2.28s/it] 46%|████▋ | 13913/30000 [8:54:25<10:18:23, 2.31s/it] 46%|████▋ | 13914/30000 [8:54:27<10:15:08, 2.29s/it] 46%|████▋ | 13915/30000 [8:54:29<10:14:05, 2.29s/it] 46%|████▋ | 13916/30000 [8:54:32<10:19:11, 2.31s/it] 46%|████▋ | 13917/30000 [8:54:34<10:10:11, 2.28s/it] 46%|████▋ | 13918/30000 [8:54:36<10:07:08, 2.27s/it] 46%|████▋ | 13919/30000 [8:54:38<10:11:41, 2.28s/it] 46%|████▋ | 13920/30000 [8:54:41<10:12:40, 2.29s/it] 46%|████▋ | 13920/30000 [8:54:41<10:12:40, 2.29s/it] 46%|████▋ | 13921/30000 [8:54:43<10:11:23, 2.28s/it] 46%|████▋ | 13922/30000 [8:54:45<10:07:28, 2.27s/it] 46%|████▋ | 13923/30000 [8:54:48<10:17:26, 2.30s/it] 46%|████▋ | 13924/30000 [8:54:50<10:11:10, 2.28s/it] 46%|████▋ | 13925/30000 [8:54:52<10:17:11, 2.30s/it] 46%|████▋ | 13926/30000 [8:54:55<10:18:14, 2.31s/it] 46%|████▋ | 13927/30000 [8:54:57<10:18:18, 2.31s/it] 46%|████▋ | 13928/30000 [8:54:59<9:50:26, 2.20s/it] 46%|████▋ | 13929/30000 [8:55:01<9:44:22, 2.18s/it] 46%|████▋ | 13930/30000 [8:55:03<9:53:23, 2.22s/it] 46%|████▋ | 13930/30000 [8:55:03<9:53:23, 2.22s/it] 46%|████▋ | 13931/30000 [8:55:05<10:00:00, 2.24s/it] 46%|████▋ | 13932/30000 [8:55:08<10:03:03, 2.25s/it] 46%|████▋ | 13933/30000 [8:55:10<10:01:14, 2.25s/it] 46%|████▋ | 13934/30000 [8:55:12<10:07:15, 2.27s/it] 46%|████▋ | 13935/30000 [8:55:15<10:11:05, 2.28s/it] 46%|████▋ | 13936/30000 [8:55:17<10:06:50, 2.27s/it] 46%|████▋ | 13937/30000 [8:55:19<10:07:12, 2.27s/it] 46%|████▋ | 13938/30000 [8:55:21<10:09:05, 2.28s/it] 46%|████▋ | 13939/30000 [8:55:24<10:07:41, 2.27s/it] 46%|████▋ | 13940/30000 [8:55:26<10:11:38, 2.29s/it] 46%|████▋ | 13940/30000 [8:55:26<10:11:38, 2.29s/it] 46%|████▋ | 13941/30000 [8:55:28<10:15:55, 2.30s/it] 46%|████▋ | 13942/30000 [8:55:31<10:17:24, 2.31s/it] 46%|████▋ | 13943/30000 [8:55:33<10:11:31, 2.29s/it] 46%|████▋ | 13944/30000 [8:55:35<10:15:43, 2.30s/it] 46%|████▋ | 13945/30000 [8:55:38<10:13:59, 2.29s/it] 46%|████▋ | 13946/30000 [8:55:40<10:21:05, 2.32s/it] 46%|████▋ | 13947/30000 [8:55:42<10:16:45, 2.31s/it] 46%|████▋ | 13948/30000 [8:55:45<10:19:15, 2.31s/it] 46%|████▋ | 13949/30000 [8:55:47<10:19:10, 2.31s/it] 46%|████▋ | 13950/30000 [8:55:49<10:23:18, 2.33s/it] 46%|████▋ | 13950/30000 [8:55:49<10:23:18, 2.33s/it] 47%|████▋ | 13951/30000 [8:55:51<10:20:56, 2.32s/it] 47%|████▋ | 13952/30000 [8:55:54<10:18:51, 2.31s/it] 47%|████▋ | 13953/30000 [8:55:56<10:16:56, 2.31s/it] 47%|████▋ | 13954/30000 [8:55:58<10:15:03, 2.30s/it] 47%|████▋ | 13955/30000 [8:56:01<10:15:06, 2.30s/it] 47%|████▋ | 13956/30000 [8:56:03<10:09:49, 2.28s/it] 47%|████▋ | 13957/30000 [8:56:05<10:09:44, 2.28s/it] 47%|████▋ | 13958/30000 [8:56:07<10:09:22, 2.28s/it] 47%|████▋ | 13959/30000 [8:56:10<10:02:10, 2.25s/it] 47%|████▋ | 13960/30000 [8:56:12<10:06:13, 2.27s/it] 47%|████▋ | 13960/30000 [8:56:12<10:06:13, 2.27s/it] 47%|████▋ | 13961/30000 [8:56:14<10:09:41, 2.28s/it] 47%|████▋ | 13962/30000 [8:56:17<10:07:07, 2.27s/it] 47%|████▋ | 13963/30000 [8:56:19<10:12:04, 2.29s/it] 47%|████▋ | 13964/30000 [8:56:21<10:08:36, 2.28s/it] 47%|████▋ | 13965/30000 [8:56:23<10:11:37, 2.29s/it] 47%|████▋ | 13966/30000 [8:56:26<10:08:51, 2.28s/it] 47%|████▋ | 13967/30000 [8:56:28<10:22:41, 2.33s/it] 47%|████▋ | 13968/30000 [8:56:30<10:23:31, 2.33s/it] 47%|████▋ | 13969/30000 [8:56:33<10:26:15, 2.34s/it] 47%|████▋ | 13970/30000 [8:56:35<10:24:31, 2.34s/it] 47%|████▋ | 13970/30000 [8:56:35<10:24:31, 2.34s/it] 47%|████▋ | 13971/30000 [8:56:38<10:29:11, 2.36s/it] 47%|████▋ | 13972/30000 [8:56:40<10:20:50, 2.32s/it] 47%|████▋ | 13973/30000 [8:56:42<10:08:39, 2.28s/it] 47%|████▋ | 13974/30000 [8:56:44<10:10:07, 2.28s/it] 47%|████▋ | 13975/30000 [8:56:47<10:09:28, 2.28s/it] 47%|████▋ | 13976/30000 [8:56:49<10:07:48, 2.28s/it] 47%|████▋ | 13977/30000 [8:56:51<10:16:44, 2.31s/it] 47%|████▋ | 13978/30000 [8:56:54<10:22:23, 2.33s/it] 47%|████▋ | 13979/30000 [8:56:56<10:18:38, 2.32s/it] 47%|████▋ | 13980/30000 [8:56:58<10:21:14, 2.33s/it] 47%|████▋ | 13980/30000 [8:56:58<10:21:14, 2.33s/it] 47%|████▋ | 13981/30000 [8:57:01<10:19:56, 2.32s/it] 47%|████▋ | 13982/30000 [8:57:03<10:14:09, 2.30s/it] 47%|████▋ | 13983/30000 [8:57:05<10:16:08, 2.31s/it] 47%|████▋ | 13984/30000 [8:57:07<10:15:14, 2.30s/it] 47%|████▋ | 13985/30000 [8:57:10<10:13:30, 2.30s/it] 47%|████▋ | 13986/30000 [8:57:12<10:15:49, 2.31s/it] 47%|████▋ | 13987/30000 [8:57:14<10:10:27, 2.29s/it] 47%|████▋ | 13988/30000 [8:57:17<10:08:52, 2.28s/it] 47%|████▋ | 13989/30000 [8:57:19<10:13:30, 2.30s/it] 47%|████▋ | 13990/30000 [8:57:21<10:16:23, 2.31s/it] 47%|████▋ | 13990/30000 [8:57:21<10:16:23, 2.31s/it] 47%|████▋ | 13991/30000 [8:57:23<10:10:08, 2.29s/it] 47%|████▋ | 13992/30000 [8:57:26<10:13:52, 2.30s/it] 47%|████▋ | 13993/30000 [8:57:28<10:16:52, 2.31s/it] 47%|████▋ | 13994/30000 [8:57:31<10:25:18, 2.34s/it] 47%|████▋ | 13995/30000 [8:57:33<10:24:59, 2.34s/it] 47%|████▋ | 13996/30000 [8:57:35<10:25:17, 2.34s/it] 47%|████▋ | 13997/30000 [8:57:37<10:18:38, 2.32s/it] 47%|████▋ | 13998/30000 [8:57:40<10:16:40, 2.31s/it] 47%|████▋ | 13999/30000 [8:57:42<10:19:47, 2.32s/it] 47%|████▋ | 14000/30000 [8:57:44<10:16:51, 2.31s/it] 47%|████▋ | 14000/30000 [8:57:44<10:16:51, 2.31s/it] 47%|████▋ | 14001/30000 [8:57:47<10:13:29, 2.30s/it] 47%|████▋ | 14002/30000 [8:57:49<10:20:05, 2.33s/it] 47%|████▋ | 14003/30000 [8:57:51<10:20:00, 2.33s/it] 47%|████▋ | 14004/30000 [8:57:54<10:15:14, 2.31s/it] 47%|████▋ | 14005/30000 [8:57:56<10:13:49, 2.30s/it] 47%|████▋ | 14006/30000 [8:57:58<10:11:40, 2.29s/it] 47%|████▋ | 14007/30000 [8:58:00<10:10:59, 2.29s/it] 47%|████▋ | 14008/30000 [8:58:03<10:17:22, 2.32s/it] 47%|████▋ | 14009/30000 [8:58:05<10:08:01, 2.28s/it] 47%|████▋ | 14010/30000 [8:58:07<9:54:41, 2.23s/it] 47%|████▋ | 14010/30000 [8:58:07<9:54:41, 2.23s/it] 47%|████▋ | 14011/30000 [8:58:09<9:55:30, 2.23s/it] 47%|████▋ | 14012/30000 [8:58:12<9:51:21, 2.22s/it] 47%|████▋ | 14013/30000 [8:58:14<10:00:56, 2.26s/it] 47%|████▋ | 14014/30000 [8:58:16<10:02:57, 2.26s/it] 47%|████▋ | 14015/30000 [8:58:18<9:54:55, 2.23s/it] 47%|████▋ | 14016/30000 [8:58:21<10:06:04, 2.28s/it] 47%|████▋ | 14017/30000 [8:58:23<10:16:20, 2.31s/it] 47%|████▋ | 14018/30000 [8:58:25<10:12:50, 2.30s/it] 47%|████▋ | 14019/30000 [8:58:28<10:19:37, 2.33s/it] 47%|████▋ | 14020/30000 [8:58:30<10:22:11, 2.34s/it] 47%|████▋ | 14020/30000 [8:58:30<10:22:11, 2.34s/it] 47%|████▋ | 14021/30000 [8:58:33<10:22:12, 2.34s/it] 47%|████▋ | 14022/30000 [8:58:35<10:17:12, 2.32s/it] 47%|████▋ | 14023/30000 [8:58:37<10:17:29, 2.32s/it] 47%|████▋ | 14024/30000 [8:58:39<10:18:43, 2.32s/it] 47%|████▋ | 14025/30000 [8:58:42<10:15:55, 2.31s/it] 47%|████▋ | 14026/30000 [8:58:44<10:07:32, 2.28s/it] 47%|████▋ | 14027/30000 [8:58:46<10:14:00, 2.31s/it] 47%|████▋ | 14028/30000 [8:58:49<10:09:30, 2.29s/it] 47%|████▋ | 14029/30000 [8:58:51<10:10:19, 2.29s/it] 47%|████▋ | 14030/30000 [8:58:53<10:25:48, 2.35s/it] 47%|████▋ | 14030/30000 [8:58:53<10:25:48, 2.35s/it] 47%|████▋ | 14031/30000 [8:58:56<10:19:48, 2.33s/it] 47%|████▋ | 14032/30000 [8:58:58<10:17:43, 2.32s/it] 47%|████▋ | 14033/30000 [8:59:00<10:18:34, 2.32s/it] 47%|████▋ | 14034/30000 [8:59:03<10:19:53, 2.33s/it] 47%|████▋ | 14035/30000 [8:59:05<10:14:30, 2.31s/it] 47%|████▋ | 14036/30000 [8:59:07<10:14:45, 2.31s/it] 47%|████▋ | 14037/30000 [8:59:10<10:17:08, 2.32s/it] 47%|████▋ | 14038/30000 [8:59:12<10:07:03, 2.28s/it] 47%|████▋ | 14039/30000 [8:59:14<10:08:48, 2.29s/it] 47%|████▋ | 14040/30000 [8:59:16<10:12:06, 2.30s/it] 47%|████▋ | 14040/30000 [8:59:16<10:12:06, 2.30s/it] 47%|████▋ | 14041/30000 [8:59:19<10:09:26, 2.29s/it] 47%|████▋ | 14042/30000 [8:59:21<10:09:35, 2.29s/it] 47%|████▋ | 14043/30000 [8:59:23<10:11:45, 2.30s/it] 47%|████▋ | 14044/30000 [8:59:25<9:55:37, 2.24s/it] 47%|████▋ | 14045/30000 [8:59:28<10:06:28, 2.28s/it] 47%|████▋ | 14046/30000 [8:59:30<10:06:10, 2.28s/it] 47%|████▋ | 14047/30000 [8:59:32<10:06:15, 2.28s/it] 47%|████▋ | 14048/30000 [8:59:35<10:06:42, 2.28s/it] 47%|████▋ | 14049/30000 [8:59:37<10:24:56, 2.35s/it] 47%|████▋ | 14050/30000 [8:59:39<10:21:45, 2.34s/it] 47%|████▋ | 14050/30000 [8:59:39<10:21:45, 2.34s/it] 47%|████▋ | 14051/30000 [8:59:42<10:16:08, 2.32s/it] 47%|████▋ | 14052/30000 [8:59:44<10:11:58, 2.30s/it] 47%|████▋ | 14053/30000 [8:59:46<10:12:46, 2.31s/it] 47%|████▋ | 14054/30000 [8:59:48<10:09:43, 2.29s/it] 47%|████▋ | 14055/30000 [8:59:51<10:09:23, 2.29s/it] 47%|████▋ | 14056/30000 [8:59:53<10:12:37, 2.31s/it] 47%|████▋ | 14057/30000 [8:59:55<10:13:49, 2.31s/it] 47%|████▋ | 14058/30000 [8:59:58<10:11:30, 2.30s/it] 47%|████▋ | 14059/30000 [9:00:00<10:14:52, 2.31s/it] 47%|████▋ | 14060/30000 [9:00:02<10:13:20, 2.31s/it] 47%|████▋ | 14060/30000 [9:00:02<10:13:20, 2.31s/it] 47%|████▋ | 14061/30000 [9:00:05<10:08:41, 2.29s/it] 47%|████▋ | 14062/30000 [9:00:07<10:09:33, 2.29s/it] 47%|████▋ | 14063/30000 [9:00:09<10:07:49, 2.29s/it] 47%|████▋ | 14064/30000 [9:00:11<10:07:42, 2.29s/it] 47%|████▋ | 14065/30000 [9:00:14<10:05:49, 2.28s/it] 47%|████▋ | 14066/30000 [9:00:16<10:11:31, 2.30s/it] 47%|████▋ | 14067/30000 [9:00:18<10:13:29, 2.31s/it] 47%|████▋ | 14068/30000 [9:00:21<10:15:45, 2.32s/it] 47%|████▋ | 14069/30000 [9:00:23<10:17:12, 2.32s/it] 47%|████▋ | 14070/30000 [9:00:25<10:13:45, 2.31s/it] 47%|████▋ | 14070/30000 [9:00:25<10:13:45, 2.31s/it] 47%|████▋ | 14071/30000 [9:00:28<10:11:55, 2.30s/it] 47%|████▋ | 14072/30000 [9:00:30<10:09:50, 2.30s/it] 47%|████▋ | 14073/30000 [9:00:32<10:14:48, 2.32s/it] 47%|████▋ | 14074/30000 [9:00:35<10:16:34, 2.32s/it] 47%|████▋ | 14075/30000 [9:00:37<10:11:03, 2.30s/it] 47%|████▋ | 14076/30000 [9:00:39<10:12:09, 2.31s/it] 47%|████▋ | 14077/30000 [9:00:42<10:13:44, 2.31s/it] 47%|████▋ | 14078/30000 [9:00:44<10:11:39, 2.30s/it] 47%|████▋ | 14079/30000 [9:00:46<10:06:39, 2.29s/it] 47%|████▋ | 14080/30000 [9:00:48<10:10:28, 2.30s/it] 47%|████▋ | 14080/30000 [9:00:48<10:10:28, 2.30s/it] 47%|████▋ | 14081/30000 [9:00:51<10:13:30, 2.31s/it] 47%|████▋ | 14082/30000 [9:00:53<10:09:28, 2.30s/it] 47%|████▋ | 14083/30000 [9:00:55<10:08:35, 2.29s/it] 47%|████▋ | 14084/30000 [9:00:58<10:09:38, 2.30s/it] 47%|████▋ | 14085/30000 [9:01:00<10:04:28, 2.28s/it] 47%|████▋ | 14086/30000 [9:01:02<10:02:00, 2.27s/it] 47%|████▋ | 14087/30000 [9:01:04<9:59:59, 2.26s/it] 47%|████▋ | 14088/30000 [9:01:07<10:01:02, 2.27s/it] 47%|████▋ | 14089/30000 [9:01:09<10:01:48, 2.27s/it] 47%|████▋ | 14090/30000 [9:01:11<10:10:55, 2.30s/it] +{'loss': 0.0145, 'grad_norm': 0.07001964747905731, 'learning_rate': 1.8378561913298822e-05, 'flow_matching_loss': 0.014522182196378708, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.98} +{'loss': 0.0134, 'grad_norm': 0.06686311215162277, 'learning_rate': 1.8362450043035053e-05, 'flow_matching_loss': 0.013379440829157829, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.99} +{'loss': 0.0149, 'grad_norm': 0.1170818954706192, 'learning_rate': 1.834633408707742e-05, 'flow_matching_loss': 0.014914518874138593, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.99} +{'loss': 0.0157, 'grad_norm': 0.10649386048316956, 'learning_rate': 1.8330214065008336e-05, 'flow_matching_loss': 0.015666478686034678, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.99} +{'loss': 0.0157, 'grad_norm': 0.07949764281511307, 'learning_rate': 1.831408999641514e-05, 'flow_matching_loss': 0.01572956470772624, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 3.99} +{'loss': 0.0151, 'grad_norm': 0.09257955104112625, 'learning_rate': 1.8297961900890107e-05, 'flow_matching_loss': 0.015082302689552306, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.0} +{'loss': 0.0136, 'grad_norm': 0.09366997331380844, 'learning_rate': 1.8281829798030385e-05, 'flow_matching_loss': 0.013616311084479093, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.0} +{'loss': 0.0149, 'grad_norm': 0.09710550308227539, 'learning_rate': 1.8265693707438002e-05, 'flow_matching_loss': 0.014855632465332747, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.0} +{'loss': 0.0148, 'grad_norm': 0.09384413808584213, 'learning_rate': 1.824955364871982e-05, 'flow_matching_loss': 0.014842930156737566, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.01} +{'loss': 0.014, 'grad_norm': 0.07173143327236176, 'learning_rate': 1.8233409641487553e-05, 'flow_matching_loss': 0.014005714654922485, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.01} +{'loss': 0.016, 'grad_norm': 0.07374656945466995, 'learning_rate': 1.8217261705357677e-05, 'flow_matching_loss': 0.015977469459176063, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.01} +{'loss': 0.0148, 'grad_norm': 0.11079880595207214, 'learning_rate': 1.8201109859951462e-05, 'flow_matching_loss': 0.01483305199071765, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.02} +{'loss': 0.0135, 'grad_norm': 0.1053965613245964, 'learning_rate': 1.818495412489492e-05, 'flow_matching_loss': 0.013477620761841535, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.02} +{'loss': 0.0161, 'grad_norm': 0.06615139544010162, 'learning_rate': 1.8168794519818802e-05, 'flow_matching_loss': 0.016141959093511106, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.02} +{'loss': 0.0153, 'grad_norm': 0.06525973230600357, 'learning_rate': 1.8152631064358537e-05, 'flow_matching_loss': 0.015326834749430418, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.02} +{'loss': 0.015, 'grad_norm': 0.06568615883588791, 'learning_rate': 1.8136463778154258e-05, 'flow_matching_loss': 0.015027104504406452, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.03} +{'loss': 0.0153, 'grad_norm': 0.07240273058414459, 'learning_rate': 1.812029268085074e-05, 'flow_matching_loss': 0.015257480647414922, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.03} +{'loss': 0.016, 'grad_norm': 0.08979135751724243, 'learning_rate': 1.8104117792097387e-05, 'flow_matching_loss': 0.016022813506424427, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.03} +{'loss': 0.0162, 'grad_norm': 0.0773392915725708, 'learning_rate': 1.808793913154822e-05, 'flow_matching_loss': 0.016239756904542445, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.04} +{'loss': 0.0155, 'grad_norm': 0.08641500771045685, 'learning_rate': 1.8071756718861838e-05, 'flow_matching_loss': 0.015460313204675913, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.04} +{'loss': 0.015, 'grad_norm': 0.09780822694301605, 'learning_rate': 1.805557057370139e-05, 'flow_matching_loss': 0.015027948655188084, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.04} +{'loss': 0.016, 'grad_norm': 0.06996931880712509, 'learning_rate': 1.8039380715734574e-05, 'flow_matching_loss': 0.015985894575715066, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.04} +{'loss': 0.0152, 'grad_norm': 0.09220394492149353, 'learning_rate': 1.802318716463359e-05, 'flow_matching_loss': 0.015214587468653918, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.05} +{'loss': 0.0152, 'grad_norm': 0.08730532974004745, 'learning_rate': 1.8006989940075138e-05, 'flow_matching_loss': 0.015239682048559189, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.05} +{'loss': 0.0139, 'grad_norm': 0.07310034334659576, 'learning_rate': 1.7990789061740368e-05, 'flow_matching_loss': 0.013948339875787497, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.05} +{'loss': 0.0155, 'grad_norm': 0.08298029750585556, 'learning_rate': 1.7974584549314875e-05, 'flow_matching_loss': 0.01551080159842968, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.06} +{'loss': 0.0164, 'grad_norm': 0.0760258287191391, 'learning_rate': 1.7958376422488672e-05, 'flow_matching_loss': 0.016402858030050992, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.06} +{'loss': 0.0166, 'grad_norm': 0.07214951515197754, 'learning_rate': 1.7942164700956156e-05, 'flow_matching_loss': 0.01656693695113063, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.06} +{'loss': 0.014, 'grad_norm': 0.08165285736322403, 'learning_rate': 1.79259494044161e-05, 'flow_matching_loss': 0.01397476801648736, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.06} +{'loss': 0.0146, 'grad_norm': 0.08159542083740234, 'learning_rate': 1.7909730552571617e-05, 'flow_matching_loss': 0.014646024908870459, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.07} +{'loss': 0.0158, 'grad_norm': 0.10157568007707596, 'learning_rate': 1.7893508165130143e-05, 'flow_matching_loss': 0.015780727937817575, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.07} +{'loss': 0.0159, 'grad_norm': 0.07735645025968552, 'learning_rate': 1.787728226180341e-05, 'flow_matching_loss': 0.015913321543484925, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.07} +{'loss': 0.0153, 'grad_norm': 0.07655656337738037, 'learning_rate': 1.786105286230742e-05, 'flow_matching_loss': 0.015310766361653804, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.08} +{'loss': 0.015, 'grad_norm': 0.08246251940727234, 'learning_rate': 1.784481998636242e-05, 'flow_matching_loss': 0.01502293711528182, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.08} +{'loss': 0.017, 'grad_norm': 0.10165663808584213, 'learning_rate': 1.782858365369289e-05, 'flow_matching_loss': 0.016965329740196468, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.08} +{'loss': 0.0151, 'grad_norm': 0.11116847395896912, 'learning_rate': 1.78123438840275e-05, 'flow_matching_loss': 0.01510065170004964, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.08} +{'loss': 0.0162, 'grad_norm': 0.059428513050079346, 'learning_rate': 1.779610069709911e-05, 'flow_matching_loss': 0.016220449563115837, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.09} +{'loss': 0.0162, 'grad_norm': 0.07638953626155853, 'learning_rate': 1.7779854112644713e-05, 'flow_matching_loss': 0.01618818137794733, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.09} +{'loss': 0.0159, 'grad_norm': 0.07663699984550476, 'learning_rate': 1.7763604150405446e-05, 'flow_matching_loss': 0.01594740841537714, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.09} +{'loss': 0.0156, 'grad_norm': 0.07886145263910294, 'learning_rate': 1.7747350830126548e-05, 'flow_matching_loss': 0.015631217695772647, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.1} +{'loss': 0.0159, 'grad_norm': 0.07668419927358627, 'learning_rate': 1.7731094171557332e-05, 'flow_matching_loss': 0.015890952292829753, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.1} 47%|████▋ | 14090/30000 [9:01:11<10:10:55, 2.30s/it] 47%|████▋ | 14091/30000 [9:01:14<10:12:36, 2.31s/it] 47%|████▋ | 14092/30000 [9:01:16<10:10:29, 2.30s/it] 47%|████▋ | 14093/30000 [9:01:18<10:08:38, 2.30s/it] 47%|████▋ | 14094/30000 [9:01:21<10:19:11, 2.34s/it] 47%|████▋ | 14095/30000 [9:01:23<10:14:44, 2.32s/it] 47%|████▋ | 14096/30000 [9:01:25<9:59:42, 2.26s/it] 47%|████▋ | 14097/30000 [9:01:26<8:36:49, 1.95s/it] 47%|████▋ | 14098/30000 [9:01:27<7:37:30, 1.73s/it] 47%|████▋ | 14099/30000 [9:01:29<6:55:23, 1.57s/it] 47%|████▋ | 14100/30000 [9:01:30<6:26:01, 1.46s/it] 47%|████▋ | 14100/30000 [9:01:30<6:26:01, 1.46s/it] 47%|████▋ | 14101/30000 [9:01:31<6:05:29, 1.38s/it] 47%|████▋ | 14102/30000 [9:01:32<5:51:12, 1.33s/it] 47%|████▋ | 14103/30000 [9:01:35<7:21:29, 1.67s/it] 47%|████▋ | 14104/30000 [9:01:37<8:22:24, 1.90s/it] 47%|████▋ | 14105/30000 [9:01:39<8:52:25, 2.01s/it] 47%|████▋ | 14106/30000 [9:01:42<9:19:04, 2.11s/it] 47%|████▋ | 14107/30000 [9:01:44<9:31:34, 2.16s/it] 47%|████▋ | 14108/30000 [9:01:46<9:44:16, 2.21s/it] 47%|████▋ | 14109/30000 [9:01:49<9:44:28, 2.21s/it] 47%|████▋ | 14110/30000 [9:01:51<9:57:01, 2.25s/it] 47%|████▋ | 14110/30000 [9:01:51<9:57:01, 2.25s/it] 47%|████▋ | 14111/30000 [9:01:53<10:00:57, 2.27s/it] 47%|████▋ | 14112/30000 [9:01:55<10:00:18, 2.27s/it] 47%|████▋ | 14113/30000 [9:01:58<9:56:14, 2.25s/it] 47%|████▋ | 14114/30000 [9:02:00<10:01:41, 2.27s/it] 47%|████▋ | 14115/30000 [9:02:02<10:02:11, 2.27s/it] 47%|████▋ | 14116/30000 [9:02:05<10:00:55, 2.27s/it] 47%|████▋ | 14117/30000 [9:02:07<10:00:26, 2.27s/it] 47%|████▋ | 14118/30000 [9:02:09<10:07:41, 2.30s/it] 47%|████▋ | 14119/30000 [9:02:11<10:04:56, 2.29s/it] 47%|████▋ | 14120/30000 [9:02:14<10:00:56, 2.27s/it] 47%|████▋ | 14120/30000 [9:02:14<10:00:56, 2.27s/it] 47%|████▋ | 14121/30000 [9:02:16<10:07:03, 2.29s/it] 47%|████▋ | 14122/30000 [9:02:18<10:04:53, 2.29s/it] 47%|████▋ | 14123/30000 [9:02:21<10:07:12, 2.29s/it] 47%|████▋ | 14124/30000 [9:02:23<10:07:26, 2.30s/it] 47%|████▋ | 14125/30000 [9:02:25<10:04:46, 2.29s/it] 47%|████▋ | 14126/30000 [9:02:27<9:59:55, 2.27s/it] 47%|████▋ | 14127/30000 [9:02:30<10:00:15, 2.27s/it] 47%|████▋ | 14128/30000 [9:02:32<10:12:18, 2.31s/it] 47%|████▋ | 14129/30000 [9:02:34<10:15:43, 2.33s/it] 47%|████▋ | 14130/30000 [9:02:37<10:13:02, 2.32s/it] 47%|████▋ | 14130/30000 [9:02:37<10:13:02, 2.32s/it] 47%|████▋ | 14131/30000 [9:02:39<10:15:09, 2.33s/it] 47%|████▋ | 14132/30000 [9:02:41<10:15:32, 2.33s/it] 47%|████▋ | 14133/30000 [9:02:44<10:09:39, 2.31s/it] 47%|████▋ | 14134/30000 [9:02:46<10:06:13, 2.29s/it] 47%|████▋ | 14135/30000 [9:02:48<10:06:54, 2.30s/it] 47%|████▋ | 14136/30000 [9:02:50<10:05:56, 2.29s/it] 47%|████▋ | 14137/30000 [9:02:53<10:05:14, 2.29s/it] 47%|████▋ | 14138/30000 [9:02:55<10:07:13, 2.30s/it] 47%|████▋ | 14139/30000 [9:02:57<10:03:30, 2.28s/it] 47%|████▋ | 14140/30000 [9:02:59<9:52:24, 2.24s/it] 47%|████▋ | 14140/30000 [9:03:00<9:52:24, 2.24s/it] 47%|████▋ | 14141/30000 [9:03:02<9:59:25, 2.27s/it] 47%|████▋ | 14142/30000 [9:03:04<10:03:44, 2.28s/it] 47%|████▋ | 14143/30000 [9:03:06<10:01:53, 2.28s/it] 47%|████▋ | 14144/30000 [9:03:09<10:00:12, 2.27s/it] 47%|████▋ | 14145/30000 [9:03:11<10:06:32, 2.30s/it] 47%|████▋ | 14146/30000 [9:03:13<10:08:03, 2.30s/it] 47%|████▋ | 14147/30000 [9:03:16<10:06:12, 2.29s/it] 47%|████▋ | 14148/30000 [9:03:18<10:16:02, 2.33s/it] 47%|████▋ | 14149/30000 [9:03:20<10:02:07, 2.28s/it] 47%|████▋ | 14150/30000 [9:03:22<9:53:15, 2.25s/it] 47%|████▋ | 14150/30000 [9:03:22<9:53:15, 2.25s/it] 47%|████▋ | 14151/30000 [9:03:25<10:02:02, 2.28s/it] 47%|████▋ | 14152/30000 [9:03:27<10:00:58, 2.28s/it] 47%|████▋ | 14153/30000 [9:03:29<10:05:54, 2.29s/it] 47%|████▋ | 14154/30000 [9:03:32<10:12:03, 2.32s/it] 47%|████▋ | 14155/30000 [9:03:34<10:09:47, 2.31s/it] 47%|████▋ | 14156/30000 [9:03:36<10:03:39, 2.29s/it] 47%|████▋ | 14157/30000 [9:03:39<10:10:08, 2.31s/it] 47%|████▋ | 14158/30000 [9:03:41<10:09:02, 2.31s/it] 47%|████▋ | 14159/30000 [9:03:43<10:13:42, 2.32s/it] 47%|████▋ | 14160/30000 [9:03:45<10:09:04, 2.31s/it] 47%|████▋ | 14160/30000 [9:03:46<10:09:04, 2.31s/it] 47%|████▋ | 14161/30000 [9:03:48<10:03:51, 2.29s/it] 47%|████▋ | 14162/30000 [9:03:50<10:11:24, 2.32s/it] 47%|████▋ | 14163/30000 [9:03:52<10:08:47, 2.31s/it] 47%|████▋ | 14164/30000 [9:03:55<10:11:49, 2.32s/it] 47%|████▋ | 14165/30000 [9:03:57<10:14:07, 2.33s/it] 47%|████▋ | 14166/30000 [9:03:59<10:03:16, 2.29s/it] 47%|████▋ | 14167/30000 [9:04:02<10:05:39, 2.30s/it] 47%|████▋ | 14168/30000 [9:04:04<10:09:04, 2.31s/it] 47%|████▋ | 14169/30000 [9:04:06<10:10:07, 2.31s/it] 47%|████▋ | 14170/30000 [9:04:09<10:16:14, 2.34s/it] 47%|████▋ | 14170/30000 [9:04:09<10:16:14, 2.34s/it] 47%|████▋ | 14171/30000 [9:04:11<10:16:18, 2.34s/it] 47%|████▋ | 14172/30000 [9:04:13<10:12:55, 2.32s/it] 47%|████▋ | 14173/30000 [9:04:16<10:10:33, 2.31s/it] 47%|████▋ | 14174/30000 [9:04:18<10:10:21, 2.31s/it] 47%|████▋ | 14175/30000 [9:04:20<10:10:42, 2.32s/it] 47%|████▋ | 14176/30000 [9:04:22<10:06:41, 2.30s/it] 47%|████▋ | 14177/30000 [9:04:25<10:14:59, 2.33s/it] 47%|████▋ | 14178/30000 [9:04:27<10:16:49, 2.34s/it] 47%|████▋ | 14179/30000 [9:04:30<10:20:51, 2.35s/it] 47%|████▋ | 14180/30000 [9:04:32<10:15:19, 2.33s/it] 47%|████▋ | 14180/30000 [9:04:32<10:15:19, 2.33s/it] 47%|████▋ | 14181/30000 [9:04:34<10:03:22, 2.29s/it] 47%|████▋ | 14182/30000 [9:04:36<9:53:05, 2.25s/it] 47%|████▋ | 14183/30000 [9:04:39<9:57:51, 2.27s/it] 47%|████▋ | 14184/30000 [9:04:41<9:58:19, 2.27s/it] 47%|████▋ | 14185/30000 [9:04:43<9:57:52, 2.27s/it] 47%|████▋ | 14186/30000 [9:04:45<10:05:18, 2.30s/it] 47%|████▋ | 14187/30000 [9:04:48<10:05:10, 2.30s/it] 47%|████▋ | 14188/30000 [9:04:50<10:06:39, 2.30s/it] 47%|████▋ | 14189/30000 [9:04:52<10:06:58, 2.30s/it] 47%|████▋ | 14190/30000 [9:04:55<10:06:59, 2.30s/it] 47%|████▋ | 14190/30000 [9:04:55<10:06:59, 2.30s/it] 47%|████▋ | 14191/30000 [9:04:57<10:03:51, 2.29s/it] 47%|████▋ | 14192/30000 [9:04:59<10:06:37, 2.30s/it] 47%|████▋ | 14193/30000 [9:05:02<10:09:36, 2.31s/it] 47%|████▋ | 14194/30000 [9:05:04<10:06:12, 2.30s/it] 47%|████▋ | 14195/30000 [9:05:06<10:05:50, 2.30s/it] 47%|████▋ | 14196/30000 [9:05:09<10:13:50, 2.33s/it] 47%|████▋ | 14197/30000 [9:05:11<10:01:52, 2.29s/it] 47%|████▋ | 14198/30000 [9:05:13<10:00:26, 2.28s/it] 47%|████▋ | 14199/30000 [9:05:15<10:08:47, 2.31s/it] 47%|████▋ | 14200/30000 [9:05:18<10:03:23, 2.29s/it] 47%|████▋ | 14200/30000 [9:05:18<10:03:23, 2.29s/it] 47%|████▋ | 14201/30000 [9:05:20<10:03:58, 2.29s/it] 47%|████▋ | 14202/30000 [9:05:22<10:11:37, 2.32s/it] 47%|████▋ | 14203/30000 [9:05:25<10:11:52, 2.32s/it] 47%|████▋ | 14204/30000 [9:05:27<10:09:42, 2.32s/it] 47%|████▋ | 14205/30000 [9:05:29<10:19:08, 2.35s/it] 47%|████▋ | 14206/30000 [9:05:32<10:13:12, 2.33s/it] 47%|████▋ | 14207/30000 [9:05:34<10:09:32, 2.32s/it] 47%|████▋ | 14208/30000 [9:05:36<10:08:20, 2.31s/it] 47%|████▋ | 14209/30000 [9:05:39<10:05:34, 2.30s/it] 47%|████▋ | 14210/30000 [9:05:41<10:04:16, 2.30s/it] 47%|████▋ | 14210/30000 [9:05:41<10:04:16, 2.30s/it] 47%|████▋ | 14211/30000 [9:05:43<10:05:44, 2.30s/it] 47%|████▋ | 14212/30000 [9:05:45<10:08:08, 2.31s/it] 47%|████▋ | 14213/30000 [9:05:48<10:09:45, 2.32s/it] 47%|████▋ | 14214/30000 [9:05:50<10:05:23, 2.30s/it] 47%|████▋ | 14215/30000 [9:05:52<10:05:40, 2.30s/it] 47%|████▋ | 14216/30000 [9:05:55<10:09:14, 2.32s/it] 47%|████▋ | 14217/30000 [9:05:57<10:09:44, 2.32s/it] 47%|████▋ | 14218/30000 [9:05:59<10:06:43, 2.31s/it] 47%|████▋ | 14219/30000 [9:06:02<10:10:03, 2.32s/it] 47%|████▋ | 14220/30000 [9:06:04<10:06:18, 2.31s/it] 47%|████▋ | 14220/30000 [9:06:04<10:06:18, 2.31s/it] 47%|████▋ | 14221/30000 [9:06:06<10:00:18, 2.28s/it] 47%|████▋ | 14222/30000 [9:06:09<10:06:10, 2.31s/it] 47%|████▋ | 14223/30000 [9:06:11<10:12:05, 2.33s/it] 47%|████▋ | 14224/30000 [9:06:13<10:14:39, 2.34s/it] 47%|████▋ | 14225/30000 [9:06:16<10:14:25, 2.34s/it] 47%|████▋ | 14226/30000 [9:06:18<10:11:04, 2.32s/it] 47%|████▋ | 14227/30000 [9:06:20<10:07:09, 2.31s/it] 47%|████▋ | 14228/30000 [9:06:23<10:07:40, 2.31s/it] 47%|████▋ | 14229/30000 [9:06:25<10:06:03, 2.31s/it] 47%|████▋ | 14230/30000 [9:06:27<10:03:47, 2.30s/it] 47%|████▋ | 14230/30000 [9:06:27<10:03:47, 2.30s/it] 47%|████▋ | 14231/30000 [9:06:29<10:09:43, 2.32s/it] 47%|████▋ | 14232/30000 [9:06:32<10:10:09, 2.32s/it] 47%|████▋ | 14233/30000 [9:06:34<10:05:24, 2.30s/it] 47%|████▋ | 14234/30000 [9:06:36<10:05:15, 2.30s/it] 47%|████▋ | 14235/30000 [9:06:39<10:09:45, 2.32s/it] 47%|████▋ | 14236/30000 [9:06:41<10:11:52, 2.33s/it] 47%|████▋ | 14237/30000 [9:06:43<10:06:24, 2.31s/it] 47%|████▋ | 14238/30000 [9:06:46<10:15:30, 2.34s/it] 47%|████▋ | 14239/30000 [9:06:48<10:10:21, 2.32s/it] 47%|████▋ | 14240/30000 [9:06:50<10:07:25, 2.31s/it] 47%|████▋ | 14240/30000 [9:06:50<10:07:25, 2.31s/it] 47%|████▋ | 14241/30000 [9:06:53<10:05:12, 2.30s/it] 47%|████▋ | 14242/30000 [9:06:55<10:16:27, 2.35s/it] 47%|████▋ | 14243/30000 [9:06:57<10:19:24, 2.36s/it] 47%|████▋ | 14244/30000 [9:07:00<10:10:52, 2.33s/it] 47%|████▋ | 14245/30000 [9:07:02<10:12:41, 2.33s/it] 47%|████▋ | 14246/30000 [9:07:04<10:11:31, 2.33s/it] 47%|████▋ | 14247/30000 [9:07:07<10:07:04, 2.31s/it] 47%|████▋ | 14248/30000 [9:07:09<10:06:32, 2.31s/it] 47%|████▋ | 14249/30000 [9:07:11<10:06:05, 2.31s/it] 48%|████▊ | 14250/30000 [9:07:13<10:02:40, 2.30s/it] 48%|████▊ | 14250/30000 [9:07:14<10:02:40, 2.30s/it] 48%|████▊ | 14251/30000 [9:07:16<10:03:03, 2.30s/it] 48%|████▊ | 14252/30000 [9:07:18<10:03:39, 2.30s/it] 48%|████▊ | 14253/30000 [9:07:20<10:07:36, 2.32s/it] 48%|████▊ | 14254/30000 [9:07:23<10:03:59, 2.30s/it] 48%|████▊ | 14255/30000 [9:07:25<10:03:24, 2.30s/it] 48%|████▊ | 14256/30000 [9:07:27<9:59:57, 2.29s/it] 48%|████▊ | 14257/30000 [9:07:30<10:02:01, 2.29s/it] 48%|████▊ | 14258/30000 [9:07:32<10:08:15, 2.32s/it] 48%|████▊ | 14259/30000 [9:07:34<10:03:56, 2.30s/it] 48%|████▊ | 14260/30000 [9:07:37<10:06:20, 2.31s/it] 48%|████▊ | 14260/30000 [9:07:37<10:06:20, 2.31s/it] 48%|████▊ | 14261/30000 [9:07:39<10:08:13, 2.32s/it] 48%|████▊ | 14262/30000 [9:07:41<10:04:54, 2.31s/it] 48%|████▊ | 14263/30000 [9:07:43<10:03:06, 2.30s/it] 48%|████▊ | 14264/30000 [9:07:46<10:06:29, 2.31s/it] 48%|████▊ | 14265/30000 [9:07:48<10:06:39, 2.31s/it] 48%|████▊ | 14266/30000 [9:07:50<9:54:52, 2.27s/it] 48%|████▊ | 14267/30000 [9:07:53<9:59:52, 2.29s/it] 48%|████▊ | 14268/30000 [9:07:55<10:06:43, 2.31s/it] 48%|████▊ | 14269/30000 [9:07:57<10:03:54, 2.30s/it] 48%|████▊ | 14270/30000 [9:08:00<10:07:42, 2.32s/it] 48%|████▊ | 14270/30000 [9:08:00<10:07:42, 2.32s/it] 48%|████▊ | 14271/30000 [9:08:02<10:09:34, 2.33s/it] 48%|████▊ | 14272/30000 [9:08:04<10:07:52, 2.32s/it] 48%|████▊ | 14273/30000 [9:08:07<10:08:25, 2.32s/it] 48%|████▊ | 14274/30000 [9:08:09<10:06:14, 2.31s/it] 48%|████▊ | 14275/30000 [9:08:11<10:05:18, 2.31s/it] 48%|████▊ | 14276/30000 [9:08:13<10:02:36, 2.30s/it] 48%|████▊ | 14277/30000 [9:08:16<10:07:52, 2.32s/it] 48%|████▊ | 14278/30000 [9:08:18<10:00:40, 2.29s/it] 48%|████▊ | 14279/30000 [9:08:20<10:01:38, 2.30s/it] 48%|████▊ | 14280/30000 [9:08:23<10:04:09, 2.31s/it] 48%|████▊ | 14280/30000 [9:08:23<10:04:09, 2.31s/it] 48%|████▊ | 14281/30000 [9:08:25<10:04:10, 2.31s/it] 48%|████▊ | 14282/30000 [9:08:27<10:01:40, 2.30s/it] 48%|████▊ | 14283/30000 [9:08:30<10:04:39, 2.31s/it] 48%|████▊ | 14284/30000 [9:08:32<10:05:19, 2.31s/it] 48%|████▊ | 14285/30000 [9:08:34<10:01:40, 2.30s/it] 48%|████▊ | 14286/30000 [9:08:37<10:04:30, 2.31s/it] 48%|████▊ | 14287/30000 [9:08:39<10:05:00, 2.31s/it] 48%|████▊ | 14288/30000 [9:08:41<10:01:26, 2.30s/it] 48%|████▊ | 14289/30000 [9:08:43<10:06:53, 2.32s/it] 48%|████▊ | 14290/30000 [9:08:46<10:02:28, 2.30s/it] 48%|████▊ | 14290/30000 [9:08:46<10:02:28, 2.30s/it] 48%|████▊ | 14291/30000 [9:08:48<10:03:04, 2.30s/it] 48%|████▊ | 14292/30000 [9:08:50<9:57:35, 2.28s/it] 48%|████▊ | 14293/30000 [9:08:53<10:00:15, 2.29s/it] 48%|████▊ | 14294/30000 [9:08:55<10:01:04, 2.30s/it] 48%|████▊ | 14295/30000 [9:08:57<9:58:24, 2.29s/it] 48%|████▊ | 14296/30000 [9:08:59<10:02:06, 2.30s/it] 48%|████▊ | 14297/30000 [9:09:02<10:05:18, 2.31s/it] 48%|████▊ | 14298/30000 [9:09:04<9:59:11, 2.29s/it] 48%|████▊ | 14299/30000 [9:09:06<10:03:43, 2.31s/it] 48%|████▊ | 14300/30000 [9:09:09<10:06:00, 2.32s/it] 48%|████▊ | 14300/30000 [9:09:09<10:06:00, 2.32s/it] 48%|████▊ | 14301/30000 [9:09:11<10:04:18, 2.31s/it] 48%|████▊ | 14302/30000 [9:09:13<10:01:25, 2.30s/it] 48%|████▊ | 14303/30000 [9:09:16<10:04:14, 2.31s/it] 48%|████▊ | 14304/30000 [9:09:18<9:59:41, 2.29s/it] 48%|████▊ | 14305/30000 [9:09:20<10:00:08, 2.29s/it] 48%|████▊ | 14306/30000 [9:09:23<10:06:13, 2.32s/it] 48%|████▊ | 14307/30000 [9:09:25<10:06:31, 2.32s/it] 48%|████▊ | 14308/30000 [9:09:27<9:59:38, 2.29s/it] 48%|████▊ | 14309/30000 [9:09:29<9:58:45, 2.29s/it] 48%|████▊ | 14310/30000 [9:09:32<9:55:53, 2.28s/it] 48%|████▊ | 14310/30000 [9:09:32<9:55:53, 2.28s/it] 48%|████▊ | 14311/30000 [9:09:34<9:54:32, 2.27s/it] 48%|████▊ | 14312/30000 [9:09:36<9:58:41, 2.29s/it] 48%|████▊ | 14313/30000 [9:09:39<10:01:21, 2.30s/it] 48%|████▊ | 14314/30000 [9:09:41<9:59:35, 2.29s/it] 48%|████▊ | 14315/30000 [9:09:43<10:01:46, 2.30s/it] 48%|████▊ | 14316/30000 [9:09:45<9:59:51, 2.29s/it] 48%|████▊ | 14317/30000 [9:09:48<10:07:26, 2.32s/it] 48%|████▊ | 14318/30000 [9:09:50<10:01:19, 2.30s/it] 48%|████▊ | 14319/30000 [9:09:52<10:06:00, 2.32s/it] 48%|████▊ | 14320/30000 [9:09:55<10:01:19, 2.30s/it] 48%|████▊ | 14320/30000 [9:09:55<10:01:19, 2.30s/it] 48%|████▊ | 14321/30000 [9:09:57<9:59:37, 2.29s/it] 48%|████▊ | 14322/30000 [9:09:59<9:57:55, 2.29s/it] 48%|████▊ | 14323/30000 [9:10:02<10:10:19, 2.34s/it] 48%|████▊ | 14324/30000 [9:10:04<10:00:33, 2.30s/it] 48%|████▊ | 14325/30000 [9:10:06<10:00:45, 2.30s/it] 48%|████▊ | 14326/30000 [9:10:09<10:00:28, 2.30s/it] 48%|████▊ | 14327/30000 [9:10:11<9:48:26, 2.25s/it] 48%|████▊ | 14328/30000 [9:10:13<9:53:12, 2.27s/it] 48%|████▊ | 14329/30000 [9:10:15<9:54:45, 2.28s/it] 48%|████▊ | 14330/30000 [9:10:18<9:53:48, 2.27s/it] 48%|████▊ | 14330/30000 [9:10:18<9:53:48, 2.27s/it] 48%|████▊ | 14331/30000 [9:10:20<9:56:15, 2.28s/it] 48%|████▊ | 14332/30000 [9:10:22<9:58:11, 2.29s/it] 48%|████▊ | 14333/30000 [9:10:24<9:57:14, 2.29s/it] 48%|████▊ | 14334/30000 [9:10:27<9:53:32, 2.27s/it] 48%|████▊ | 14335/30000 [9:10:29<9:54:51, 2.28s/it] 48%|████▊ | 14336/30000 [9:10:31<9:54:35, 2.28s/it] 48%|████▊ | 14337/30000 [9:10:34<9:56:54, 2.29s/it] 48%|████▊ | 14338/30000 [9:10:36<10:04:40, 2.32s/it] 48%|████▊ | 14339/30000 [9:10:38<9:57:59, 2.29s/it] 48%|████▊ | 14340/30000 [9:10:40<9:54:49, 2.28s/it] 48%|████▊ | 14340/30000 [9:10:40<9:54:49, 2.28s/it] 48%|████▊ | 14341/30000 [9:10:43<9:54:28, 2.28s/it] 48%|████▊ | 14342/30000 [9:10:45<9:53:10, 2.27s/it] 48%|████▊ | 14343/30000 [9:10:47<9:59:38, 2.30s/it] 48%|████▊ | 14344/30000 [9:10:50<10:00:50, 2.30s/it] 48%|████▊ | 14345/30000 [9:10:52<9:52:53, 2.27s/it] 48%|████▊ | 14346/30000 [9:10:54<10:02:00, 2.31s/it] 48%|████▊ | 14347/30000 [9:10:57<10:07:14, 2.33s/it] 48%|████▊ | 14348/30000 [9:10:59<10:00:54, 2.30s/it] 48%|████▊ | 14349/30000 [9:11:01<9:57:16, 2.29s/it] 48%|████▊ | 14350/30000 [9:11:03<9:59:20, 2.30s/it] 48%|████▊ | 14350/30000 [9:11:03<9:59:20, 2.30s/it] 48%|████▊ | 14351/30000 [9:11:06<9:57:44, 2.29s/it] 48%|████▊ | 14352/30000 [9:11:08<9:42:35, 2.23s/it] 48%|████▊ | 14353/30000 [9:11:10<9:50:46, 2.27s/it] 48%|████▊ | 14354/30000 [9:11:12<9:48:43, 2.26s/it] 48%|████▊ | 14355/30000 [9:11:15<9:41:48, 2.23s/it] 48%|████▊ | 14356/30000 [9:11:17<9:44:48, 2.24s/it] 48%|████▊ | 14357/30000 [9:11:19<9:37:07, 2.21s/it] 48%|████▊ | 14358/30000 [9:11:21<9:44:50, 2.24s/it] 48%|████▊ | 14359/30000 [9:11:24<9:51:49, 2.27s/it] 48%|████▊ | 14360/30000 [9:11:26<9:50:02, 2.26s/it] 48%|████▊ | 14360/30000 [9:11:26<9:50:02, 2.26s/it] 48%|████▊ | 14361/30000 [9:11:28<9:57:14, 2.29s/it] 48%|████▊ | 14362/30000 [9:11:31<9:59:05, 2.30s/it] 48%|████▊ | 14363/30000 [9:11:33<10:01:37, 2.31s/it] 48%|████▊ | 14364/30000 [9:11:35<9:53:36, 2.28s/it] 48%|████▊ | 14365/30000 [9:11:37<9:58:59, 2.30s/it] 48%|████▊ | 14366/30000 [9:11:40<9:57:33, 2.29s/it] 48%|████▊ | 14367/30000 [9:11:42<9:56:57, 2.29s/it] 48%|████▊ | 14368/30000 [9:11:44<10:00:29, 2.30s/it] 48%|████▊ | 14369/30000 [9:11:47<10:03:14, 2.32s/it] 48%|████▊ | 14370/30000 [9:11:49<9:59:29, 2.30s/it] 48%|████▊ | 14370/30000 [9:11:49<9:59:29, 2.30s/it] 48%|████▊ | 14371/30000 [9:11:51<9:59:34, 2.30s/it] 48%|████▊ | 14372/30000 [9:11:54<9:59:15, 2.30s/it] 48%|████▊ | 14373/30000 [9:11:56<9:56:27, 2.29s/it] 48%|████▊ | 14374/30000 [9:11:58<9:59:57, 2.30s/it] 48%|████▊ | 14375/30000 [9:12:01<10:11:47, 2.35s/it] 48%|████▊ | 14376/30000 [9:12:03<10:08:37, 2.34s/it] 48%|████▊ | 14377/30000 [9:12:05<10:03:00, 2.32s/it] 48%|████▊ | 14378/30000 [9:12:07<10:03:21, 2.32s/it] 48%|████▊ | 14379/30000 [9:12:10<10:08:27, 2.34s/it] 48%|████▊ | 14380/30000 [9:12:12<10:08:13, 2.34s/it] 48%|████▊ | 14380/30000 [9:12:12<10:08:13, 2.34s/it] 48%|████▊ | 14381/30000 [9:12:14<9:58:36, 2.30s/it] 48%|████▊ | 14382/30000 [9:12:17<9:57:37, 2.30s/it] 48%|████▊ | 14383/30000 [9:12:19<10:02:17, 2.31s/it] 48%|████▊ | 14384/30000 [9:12:21<9:57:14, 2.29s/it] 48%|████▊ | 14385/30000 [9:12:24<10:03:29, 2.32s/it] 48%|████▊ | 14386/30000 [9:12:26<9:59:46, 2.30s/it] 48%|████▊ | 14387/30000 [9:12:28<9:57:35, 2.30s/it] 48%|████▊ | 14388/30000 [9:12:31<9:59:59, 2.31s/it] 48%|████▊ | 14389/30000 [9:12:33<10:03:01, 2.32s/it] 48%|████▊ | 14390/30000 [9:12:35<10:03:13, 2.32s/it] 48%|████▊ | 14390/30000 [9:12:35<10:03:13, 2.32s/it] 48%|████▊ | 14391/30000 [9:12:37<9:58:53, 2.30s/it] 48%|████▊ | 14392/30000 [9:12:40<9:58:47, 2.30s/it] 48%|████▊ | 14393/30000 [9:12:42<9:56:14, 2.29s/it] 48%|████▊ | 14394/30000 [9:12:44<10:02:59, 2.32s/it] 48%|████▊ | 14395/30000 [9:12:47<10:08:09, 2.34s/it] 48%|████▊ | 14396/30000 [9:12:49<10:05:46, 2.33s/it] 48%|████▊ | 14397/30000 [9:12:51<10:01:25, 2.31s/it] 48%|████▊ | 14398/30000 [9:12:54<9:59:45, 2.31s/it] 48%|████▊ | 14399/30000 [9:12:56<10:08:52, 2.34s/it] 48%|████▊ | 14400/30000 [9:12:58<10:05:43, 2.33s/it] 48%|████▊ | 14400/30000 [9:12:58<10:05:43, 2.33s/it] 48%|████▊ | 14401/30000 [9:13:01<10:03:00, 2.32s/it] 48%|████▊ | 14402/30000 [9:13:03<10:02:24, 2.32s/it] 48%|████▊ | 14403/30000 [9:13:05<10:11:42, 2.35s/it] 48%|████▊ | 14404/30000 [9:13:08<10:06:54, 2.33s/it] 48%|████▊ | 14405/30000 [9:13:10<10:00:26, 2.31s/it] 48%|████▊ | 14406/30000 [9:13:12<10:01:24, 2.31s/it] 48%|████▊ | 14407/30000 [9:13:15<10:00:41, 2.31s/it] 48%|████▊ | 14408/30000 [9:13:17<9:56:48, 2.30s/it] 48%|████▊ | 14409/30000 [9:13:19<9:56:48, 2.30s/it] 48%|████▊ | 14410/30000 [9:13:22<10:02:13, 2.32s/it] 48%|████▊ | 14410/30000 [9:13:22<10:02:13, 2.32s/it] 48%|████▊ | 14411/30000 [9:13:24<10:03:18, 2.32s/it] 48%|████▊ | 14412/30000 [9:13:26<9:53:46, 2.29s/it] 48%|████▊ | 14413/30000 [9:13:28<9:57:38, 2.30s/it] 48%|████▊ | 14414/30000 [9:13:31<9:52:33, 2.28s/it] 48%|████▊ | 14415/30000 [9:13:33<9:49:21, 2.27s/it] 48%|████▊ | 14416/30000 [9:13:35<9:56:07, 2.30s/it] 48%|████▊ | 14417/30000 [9:13:38<9:58:20, 2.30s/it] 48%|████▊ | 14418/30000 [9:13:40<9:55:34, 2.29s/it] 48%|████▊ | 14419/30000 [9:13:42<9:55:43, 2.29s/it] 48%|████▊ | 14420/30000 [9:13:44<9:55:32, 2.29s/it] 48%|████▊ | 14420/30000 [9:13:44<9:55:32, 2.29s/it] 48%|████▊ | 14421/30000 [9:13:47<9:49:35, 2.27s/it] 48%|████▊ | 14422/30000 [9:13:49<9:50:16, 2.27s/it] 48%|████▊ | 14423/30000 [9:13:51<9:37:16, 2.22s/it] 48%|████▊ | 14424/30000 [9:13:53<9:45:16, 2.25s/it] 48%|████▊ | 14425/30000 [9:13:56<9:48:55, 2.27s/it] 48%|████▊ | 14426/30000 [9:13:58<9:48:15, 2.27s/it] 48%|████▊ | 14427/30000 [9:14:00<9:48:33, 2.27s/it] 48%|████▊ | 14428/30000 [9:14:03<9:53:18, 2.29s/it] 48%|████▊ | 14429/30000 [9:14:05<9:50:56, 2.28s/it] 48%|████▊ | 14430/30000 [9:14:07<9:51:42, 2.28s/it] 48%|████▊ | 14430/30000 [9:14:07<9:51:42, 2.28s/it] 48%|████▊ | 14431/30000 [9:14:09<9:53:59, 2.29s/it] 48%|████▊ | 14432/30000 [9:14:12<9:51:23, 2.28s/it] 48%|████▊ | 14433/30000 [9:14:14<9:58:18, 2.31s/it] 48%|████▊ | 14434/30000 [9:14:16<10:10:25, 2.35s/it] 48%|████▊ | 14435/30000 [9:14:19<9:59:28, 2.31s/it] 48%|████▊ | 14436/30000 [9:14:21<9:55:33, 2.30s/it] 48%|████▊ | 14437/30000 [9:14:23<9:58:59, 2.31s/it] 48%|████▊ | 14438/30000 [9:14:26<9:54:53, 2.29s/it] 48%|████▊ | 14439/30000 [9:14:28<9:51:39, 2.28s/it] 48%|████▊ | 14440/30000 [9:14:30<9:54:20, 2.29s/it] 48%|████▊ | 14440/30000 [9:14:30<9:54:20, 2.29s/it] 48%|████▊ | 14441/30000 [9:14:32<9:59:10, 2.31s/it] 48%|████▊ | 14442/30000 [9:14:35<10:05:42, 2.34s/it] 48%|████▊ | 14443/30000 [9:14:37<10:03:39, 2.33s/it] 48%|████▊ | 14444/30000 [9:14:40<10:05:26, 2.34s/it] 48%|████▊ | 14445/30000 [9:14:42<10:00:25, 2.32s/it] 48%|████▊ | 14446/30000 [9:14:44<9:54:49, 2.29s/it] 48%|████▊ | 14447/30000 [9:14:46<10:01:45, 2.32s/it] 48%|████▊ | 14448/30000 [9:14:49<10:01:58, 2.32s/it] 48%|████▊ | 14449/30000 [9:14:51<9:58:52, 2.31s/it] 48%|████▊ | 14450/30000 [9:14:53<9:59:05, 2.31s/it] 48%|████▊ | 14450/30000 [9:14:53<9:59:05, 2.31s/it] 48%|████▊ | 14451/30000 [9:14:56<9:59:54, 2.31s/it] 48%|████▊ | 14452/30000 [9:14:58<9:54:13, 2.29s/it] 48%|████▊ | 14453/30000 [9:15:00<9:50:34, 2.28s/it] 48%|████▊ | 14454/30000 [9:15:02<9:54:42, 2.30s/it] 48%|████▊ | 14455/30000 [9:15:05<9:49:43, 2.28s/it] 48%|████▊ | 14456/30000 [9:15:07<9:59:42, 2.31s/it] 48%|████▊ | 14457/30000 [9:15:09<9:58:27, 2.31s/it] 48%|████▊ | 14458/30000 [9:15:12<10:00:07, 2.32s/it] 48%|████▊ | 14459/30000 [9:15:14<9:56:43, 2.30s/it] 48%|████▊ | 14460/30000 [9:15:16<9:53:44, 2.29s/it] 48%|████▊ | 14460/30000 [9:15:16<9:53:44, 2.29s/it] 48%|████▊ | 14461/30000 [9:15:19<10:01:32, 2.32s/it] 48%|████▊ | 14462/30000 [9:15:21<9:57:27, 2.31s/it] 48%|████▊ | 14463/30000 [9:15:23<9:54:53, 2.30s/it] 48%|████▊ | 14464/30000 [9:15:26<10:01:27, 2.32s/it] 48%|████▊ | 14465/30000 [9:15:28<10:00:33, 2.32s/it] 48%|████▊ | 14466/30000 [9:15:30<9:46:36, 2.27s/it] 48%|████▊ | 14467/30000 [9:15:32<9:56:41, 2.30s/it] 48%|████▊ | 14468/30000 [9:15:35<9:59:17, 2.32s/it] 48%|████▊ | 14469/30000 [9:15:37<9:53:04, 2.29s/it] 48%|████▊ | 14470/30000 [9:15:39<9:56:14, 2.30s/it] 48%|████▊ | 14470/30000 [9:15:39<9:56:14, 2.30s/it] 48%|████▊ | 14471/30000 [9:15:42<9:59:45, 2.32s/it] 48%|████▊ | 14472/30000 [9:15:44<9:57:18, 2.31s/it] 48%|████▊ | 14473/30000 [9:15:46<9:54:27, 2.30s/it] 48%|████▊ | 14474/30000 [9:15:49<9:53:45, 2.29s/it] 48%|████▊ | 14475/30000 [9:15:51<9:51:41, 2.29s/it] 48%|████▊ | 14476/30000 [9:15:53<9:46:30, 2.27s/it] 48%|████▊ | 14477/30000 [9:15:55<9:50:19, 2.28s/it] 48%|████▊ | 14478/30000 [9:15:58<9:56:55, 2.31s/it] 48%|████▊ | 14479/30000 [9:16:00<9:54:33, 2.30s/it] 48%|████▊ | 14480/30000 [9:16:02<9:49:17, 2.28s/it] 48%|████▊ | 14480/30000 [9:16:02<9:49:17, 2.28s/it] 48%|████▊ | 14481/30000 [9:16:05<9:54:11, 2.30s/it] 48%|████▊ | 14482/30000 [9:16:07<9:57:44, 2.31s/it] 48%|████▊ | 14483/30000 [9:16:09<9:54:35, 2.30s/it] 48%|████▊ | 14484/30000 [9:16:11<9:49:12, 2.28s/it] 48%|████▊ | 14485/30000 [9:16:14<9:56:14, 2.31s/it] 48%|████▊ | 14486/30000 [9:16:16<9:54:03, 2.30s/it] 48%|████▊ | 14487/30000 [9:16:18<9:54:30, 2.30s/it] 48%|████▊ | 14488/30000 [9:16:21<9:54:57, 2.30s/it] 48%|████▊ | 14489/30000 [9:16:23<9:57:38, 2.31s/it] 48%|████▊ | 14490/30000 [9:16:25<9:56:06, 2.31s/it] 48%|████▊ | 14490/30000 [9:16:25<9:56:06, 2.31s/it] 48%|████▊ | 14491/30000 [9:16:28<9:53:28, 2.30s/it] 48%|████▊ | 14492/30000 [9:16:30<9:53:40, 2.30s/it] 48%|████▊ | 14493/30000 [9:16:32<9:53:18, 2.30s/it] 48%|████▊ | 14494/30000 [9:16:34<9:51:29, 2.29s/it] 48%|████▊ | 14495/30000 [9:16:37<9:45:06, 2.26s/it] 48%|████▊ | 14496/30000 [9:16:39<9:50:06, 2.28s/it] 48%|████▊ | 14497/30000 [9:16:41<9:55:37, 2.31s/it] 48%|████▊ | 14498/30000 [9:16:44<9:52:57, 2.30s/it] 48%|████▊ | 14499/30000 [9:16:46<9:53:02, 2.30s/it] 48%|████▊ | 14500/30000 [9:16:48<9:56:01, 2.31s/it] +{'loss': 0.0145, 'grad_norm': 0.06442062556743622, 'learning_rate': 1.771483419445117e-05, 'flow_matching_loss': 0.014469109661877155, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.1} +{'loss': 0.0129, 'grad_norm': 0.06407693773508072, 'learning_rate': 1.7698570918565474e-05, 'flow_matching_loss': 0.012871387787163258, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.11} +{'loss': 0.0131, 'grad_norm': 0.06688642501831055, 'learning_rate': 1.768230436366165e-05, 'flow_matching_loss': 0.013115144986659288, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.11} +{'loss': 0.0162, 'grad_norm': 0.07322639226913452, 'learning_rate': 1.7666034549505102e-05, 'flow_matching_loss': 0.016165948752313853, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.11} +{'loss': 0.0161, 'grad_norm': 0.06608857214450836, 'learning_rate': 1.7649761495865185e-05, 'flow_matching_loss': 0.016100504156202077, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.11} +{'loss': 0.0145, 'grad_norm': 0.08379761129617691, 'learning_rate': 1.7633485222515188e-05, 'flow_matching_loss': 0.01454313714057207, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.12} +{'loss': 0.0161, 'grad_norm': 0.07886158674955368, 'learning_rate': 1.761720574923233e-05, 'flow_matching_loss': 0.01610182672739029, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.12} +{'loss': 0.0159, 'grad_norm': 0.07624044269323349, 'learning_rate': 1.7600923095797704e-05, 'flow_matching_loss': 0.01586903017014265, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.12} +{'loss': 0.0146, 'grad_norm': 0.09627790749073029, 'learning_rate': 1.758463728199626e-05, 'flow_matching_loss': 0.014557879511266946, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.13} +{'loss': 0.0139, 'grad_norm': 0.11454810947179794, 'learning_rate': 1.7568348327616803e-05, 'flow_matching_loss': 0.013889676611870527, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.13} +{'loss': 0.0152, 'grad_norm': 0.07685293257236481, 'learning_rate': 1.755205625245195e-05, 'flow_matching_loss': 0.015228831861168145, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.13} +{'loss': 0.0137, 'grad_norm': 0.08030282706022263, 'learning_rate': 1.7535761076298104e-05, 'flow_matching_loss': 0.013746820762753487, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.13} +{'loss': 0.0135, 'grad_norm': 0.08963262289762497, 'learning_rate': 1.7519462818955448e-05, 'flow_matching_loss': 0.013496614899486304, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.14} +{'loss': 0.0149, 'grad_norm': 0.07776547968387604, 'learning_rate': 1.7503161500227896e-05, 'flow_matching_loss': 0.01493463758379221, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.14} +{'loss': 0.0147, 'grad_norm': 0.07679387181997299, 'learning_rate': 1.7486857139923093e-05, 'flow_matching_loss': 0.014659299422055483, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.14} +{'loss': 0.0149, 'grad_norm': 0.09313786029815674, 'learning_rate': 1.747054975785236e-05, 'flow_matching_loss': 0.014898673817515373, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.15} +{'loss': 0.0159, 'grad_norm': 0.07560387253761292, 'learning_rate': 1.745423937383072e-05, 'flow_matching_loss': 0.015903094317764044, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.15} +{'loss': 0.0132, 'grad_norm': 0.07156052440404892, 'learning_rate': 1.7437926007676825e-05, 'flow_matching_loss': 0.013232088647782803, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.15} +{'loss': 0.0147, 'grad_norm': 0.07507237046957016, 'learning_rate': 1.742160967921294e-05, 'flow_matching_loss': 0.014724692143499852, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.15} +{'loss': 0.0158, 'grad_norm': 0.07447381317615509, 'learning_rate': 1.7405290408264954e-05, 'flow_matching_loss': 0.015805251710116864, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.16} +{'loss': 0.0144, 'grad_norm': 0.055837150663137436, 'learning_rate': 1.738896821466232e-05, 'flow_matching_loss': 0.014357852935791015, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.16} +{'loss': 0.0142, 'grad_norm': 0.07191658765077591, 'learning_rate': 1.737264311823804e-05, 'flow_matching_loss': 0.014213691931217909, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.16} +{'loss': 0.0148, 'grad_norm': 0.07382476329803467, 'learning_rate': 1.7356315138828645e-05, 'flow_matching_loss': 0.014819152653217316, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.17} +{'loss': 0.0155, 'grad_norm': 0.07806475460529327, 'learning_rate': 1.7339984296274174e-05, 'flow_matching_loss': 0.01553381560370326, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.17} +{'loss': 0.0149, 'grad_norm': 0.09105653315782547, 'learning_rate': 1.7323650610418134e-05, 'flow_matching_loss': 0.01489891717210412, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.17} +{'loss': 0.0148, 'grad_norm': 0.09984341263771057, 'learning_rate': 1.730731410110749e-05, 'flow_matching_loss': 0.014844807796180248, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.18} +{'loss': 0.015, 'grad_norm': 0.07517286390066147, 'learning_rate': 1.7290974788192655e-05, 'flow_matching_loss': 0.014979499392211437, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.18} +{'loss': 0.0154, 'grad_norm': 0.08493130654096603, 'learning_rate': 1.727463269152743e-05, 'flow_matching_loss': 0.015357793681323529, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.18} +{'loss': 0.0157, 'grad_norm': 0.07967507839202881, 'learning_rate': 1.7258287830968996e-05, 'flow_matching_loss': 0.015724270697683097, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.18} +{'loss': 0.0152, 'grad_norm': 0.07800158858299255, 'learning_rate': 1.7241940226377908e-05, 'flow_matching_loss': 0.015233211684972048, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.19} +{'loss': 0.0147, 'grad_norm': 0.07308946549892426, 'learning_rate': 1.7225589897618043e-05, 'flow_matching_loss': 0.014747006166726351, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.19} +{'loss': 0.0158, 'grad_norm': 0.09524010866880417, 'learning_rate': 1.7209236864556588e-05, 'flow_matching_loss': 0.015788565017282964, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.19} +{'loss': 0.0145, 'grad_norm': 0.07999199628829956, 'learning_rate': 1.719288114706403e-05, 'flow_matching_loss': 0.014461743272840976, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.2} +{'loss': 0.0161, 'grad_norm': 0.1039251759648323, 'learning_rate': 1.7176522765014104e-05, 'flow_matching_loss': 0.016062608733773232, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.2} +{'loss': 0.0168, 'grad_norm': 0.07494623214006424, 'learning_rate': 1.7160161738283782e-05, 'flow_matching_loss': 0.01680302172899246, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.2} +{'loss': 0.0158, 'grad_norm': 0.06992339342832565, 'learning_rate': 1.714379808675326e-05, 'flow_matching_loss': 0.015759934671223164, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.2} +{'loss': 0.0169, 'grad_norm': 0.1049075722694397, 'learning_rate': 1.712743183030592e-05, 'flow_matching_loss': 0.01689430568367243, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.21} +{'loss': 0.0152, 'grad_norm': 0.10233863443136215, 'learning_rate': 1.7111062988828294e-05, 'flow_matching_loss': 0.01518243681639433, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.21} +{'loss': 0.0159, 'grad_norm': 0.08596283942461014, 'learning_rate': 1.709469158221008e-05, 'flow_matching_loss': 0.01587499277666211, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.21} +{'loss': 0.0154, 'grad_norm': 0.0785202756524086, 'learning_rate': 1.7078317630344083e-05, 'flow_matching_loss': 0.015418623015284539, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.22} +{'loss': 0.015, 'grad_norm': 0.08539605885744095, 'learning_rate': 1.706194115312618e-05, 'flow_matching_loss': 0.014982212614268065, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.22} 48%|████▊ | 14500/30000 [9:16:48<9:56:01, 2.31s/it] 48%|████▊ | 14501/30000 [9:16:51<9:58:59, 2.32s/it] 48%|████▊ | 14502/30000 [9:16:53<9:52:37, 2.29s/it] 48%|████▊ | 14503/30000 [9:16:55<9:52:33, 2.29s/it] 48%|████▊ | 14504/30000 [9:16:57<9:47:35, 2.28s/it] 48%|████▊ | 14505/30000 [9:17:00<9:48:00, 2.28s/it] 48%|████▊ | 14506/30000 [9:17:02<9:50:37, 2.29s/it] 48%|████▊ | 14507/30000 [9:17:04<9:54:06, 2.30s/it] 48%|████▊ | 14508/30000 [9:17:07<9:54:23, 2.30s/it] 48%|████▊ | 14509/30000 [9:17:09<9:52:30, 2.29s/it] 48%|████▊ | 14510/30000 [9:17:11<9:51:34, 2.29s/it] 48%|████▊ | 14510/30000 [9:17:11<9:51:34, 2.29s/it] 48%|████▊ | 14511/30000 [9:17:13<9:51:47, 2.29s/it] 48%|████▊ | 14512/30000 [9:17:16<9:54:10, 2.30s/it] 48%|████▊ | 14513/30000 [9:17:18<9:50:58, 2.29s/it] 48%|████▊ | 14514/30000 [9:17:20<9:52:19, 2.29s/it] 48%|████▊ | 14515/30000 [9:17:23<9:53:28, 2.30s/it] 48%|████▊ | 14516/30000 [9:17:25<9:55:01, 2.31s/it] 48%|████▊ | 14517/30000 [9:17:27<9:52:02, 2.29s/it] 48%|████▊ | 14518/30000 [9:17:29<9:47:19, 2.28s/it] 48%|████▊ | 14519/30000 [9:17:32<9:53:00, 2.30s/it] 48%|████▊ | 14520/30000 [9:17:34<9:53:17, 2.30s/it] 48%|████▊ | 14520/30000 [9:17:34<9:53:17, 2.30s/it] 48%|████▊ | 14521/30000 [9:17:36<9:54:14, 2.30s/it] 48%|████▊ | 14522/30000 [9:17:39<9:53:39, 2.30s/it] 48%|████▊ | 14523/30000 [9:17:41<9:58:23, 2.32s/it] 48%|████▊ | 14524/30000 [9:17:43<9:40:06, 2.25s/it] 48%|████▊ | 14525/30000 [9:17:45<9:34:05, 2.23s/it] 48%|████▊ | 14526/30000 [9:17:48<9:42:16, 2.26s/it] 48%|████▊ | 14527/30000 [9:17:50<9:43:26, 2.26s/it] 48%|████▊ | 14528/30000 [9:17:52<9:46:45, 2.28s/it] 48%|████▊ | 14529/30000 [9:17:55<9:50:07, 2.29s/it] 48%|████▊ | 14530/30000 [9:17:57<9:39:57, 2.25s/it] 48%|████▊ | 14530/30000 [9:17:57<9:39:57, 2.25s/it] 48%|████▊ | 14531/30000 [9:17:59<9:41:12, 2.25s/it] 48%|████▊ | 14532/30000 [9:18:01<9:43:32, 2.26s/it] 48%|████▊ | 14533/30000 [9:18:04<9:43:31, 2.26s/it] 48%|████▊ | 14534/30000 [9:18:06<9:43:55, 2.27s/it] 48%|████▊ | 14535/30000 [9:18:08<9:47:44, 2.28s/it] 48%|████▊ | 14536/30000 [9:18:10<9:47:15, 2.28s/it] 48%|████▊ | 14537/30000 [9:18:13<9:42:10, 2.26s/it] 48%|████▊ | 14538/30000 [9:18:15<9:55:53, 2.31s/it] 48%|████▊ | 14539/30000 [9:18:17<9:56:06, 2.31s/it] 48%|████▊ | 14540/30000 [9:18:20<9:48:15, 2.28s/it] 48%|████▊ | 14540/30000 [9:18:20<9:48:15, 2.28s/it] 48%|████▊ | 14541/30000 [9:18:22<9:54:11, 2.31s/it] 48%|████▊ | 14542/30000 [9:18:24<9:56:28, 2.32s/it] 48%|████▊ | 14543/30000 [9:18:27<9:50:26, 2.29s/it] 48%|████▊ | 14544/30000 [9:18:29<9:51:46, 2.30s/it] 48%|████▊ | 14545/30000 [9:18:31<9:54:33, 2.31s/it] 48%|████▊ | 14546/30000 [9:18:33<9:45:19, 2.27s/it] 48%|████▊ | 14547/30000 [9:18:36<9:51:24, 2.30s/it] 48%|████▊ | 14548/30000 [9:18:38<9:58:17, 2.32s/it] 48%|████▊ | 14549/30000 [9:18:40<9:53:21, 2.30s/it] 48%|████▊ | 14550/30000 [9:18:43<9:46:02, 2.28s/it] 48%|████▊ | 14550/30000 [9:18:43<9:46:02, 2.28s/it] 49%|████▊ | 14551/30000 [9:18:45<9:51:33, 2.30s/it] 49%|████▊ | 14552/30000 [9:18:47<9:55:59, 2.31s/it] 49%|████▊ | 14553/30000 [9:18:50<9:50:31, 2.29s/it] 49%|████▊ | 14554/30000 [9:18:52<9:50:56, 2.30s/it] 49%|████▊ | 14555/30000 [9:18:54<9:48:58, 2.29s/it] 49%|████▊ | 14556/30000 [9:18:56<9:46:19, 2.28s/it] 49%|████▊ | 14557/30000 [9:18:59<9:48:36, 2.29s/it] 49%|████▊ | 14558/30000 [9:19:01<9:40:51, 2.26s/it] 49%|████▊ | 14559/30000 [9:19:03<9:43:21, 2.27s/it] 49%|████▊ | 14560/30000 [9:19:05<9:45:02, 2.27s/it] 49%|████▊ | 14560/30000 [9:19:05<9:45:02, 2.27s/it] 49%|████▊ | 14561/30000 [9:19:08<9:45:20, 2.27s/it] 49%|████▊ | 14562/30000 [9:19:10<9:53:17, 2.31s/it] 49%|████▊ | 14563/30000 [9:19:12<9:54:12, 2.31s/it] 49%|████▊ | 14564/30000 [9:19:15<9:55:02, 2.31s/it] 49%|████▊ | 14565/30000 [9:19:17<9:50:22, 2.29s/it] 49%|████▊ | 14566/30000 [9:19:19<9:52:33, 2.30s/it] 49%|████▊ | 14567/30000 [9:19:22<9:55:05, 2.31s/it] 49%|████▊ | 14568/30000 [9:19:24<9:51:50, 2.30s/it] 49%|████▊ | 14569/30000 [9:19:26<9:54:50, 2.31s/it] 49%|████▊ | 14570/30000 [9:19:29<9:57:47, 2.32s/it] 49%|████▊ | 14570/30000 [9:19:29<9:57:47, 2.32s/it] 49%|████▊ | 14571/30000 [9:19:31<9:53:43, 2.31s/it] 49%|████▊ | 14572/30000 [9:19:33<9:56:46, 2.32s/it] 49%|████▊ | 14573/30000 [9:19:35<9:53:14, 2.31s/it] 49%|████▊ | 14574/30000 [9:19:38<9:49:24, 2.29s/it] 49%|████▊ | 14575/30000 [9:19:40<9:55:36, 2.32s/it] 49%|████▊ | 14576/30000 [9:19:42<9:54:47, 2.31s/it] 49%|████▊ | 14577/30000 [9:19:45<10:00:36, 2.34s/it] 49%|████▊ | 14578/30000 [9:19:47<9:57:24, 2.32s/it] 49%|████▊ | 14579/30000 [9:19:49<9:57:17, 2.32s/it] 49%|████▊ | 14580/30000 [9:19:52<9:50:30, 2.30s/it] 49%|████▊ | 14580/30000 [9:19:52<9:50:30, 2.30s/it] 49%|████▊ | 14581/30000 [9:19:54<9:47:11, 2.28s/it] 49%|████▊ | 14582/30000 [9:19:56<9:54:26, 2.31s/it] 49%|████▊ | 14583/30000 [9:19:59<9:55:10, 2.32s/it] 49%|████▊ | 14584/30000 [9:20:01<9:44:10, 2.27s/it] 49%|████▊ | 14585/30000 [9:20:03<9:45:27, 2.28s/it] 49%|████▊ | 14586/30000 [9:20:05<9:36:57, 2.25s/it] 49%|████▊ | 14587/30000 [9:20:08<9:45:04, 2.28s/it] 49%|████▊ | 14588/30000 [9:20:10<9:49:51, 2.30s/it] 49%|████▊ | 14589/30000 [9:20:12<9:43:21, 2.27s/it] 49%|████▊ | 14590/30000 [9:20:14<9:47:25, 2.29s/it] 49%|████▊ | 14590/30000 [9:20:14<9:47:25, 2.29s/it] 49%|████▊ | 14591/30000 [9:20:17<9:43:35, 2.27s/it] 49%|████▊ | 14592/30000 [9:20:19<9:29:16, 2.22s/it] 49%|████▊ | 14593/30000 [9:20:21<9:35:23, 2.24s/it] 49%|████▊ | 14594/30000 [9:20:24<9:49:06, 2.29s/it] 49%|████▊ | 14595/30000 [9:20:26<9:45:28, 2.28s/it] 49%|████▊ | 14596/30000 [9:20:28<9:44:03, 2.27s/it] 49%|████▊ | 14597/30000 [9:20:30<9:43:13, 2.27s/it] 49%|████▊ | 14598/30000 [9:20:33<9:40:15, 2.26s/it] 49%|████▊ | 14599/30000 [9:20:35<9:43:50, 2.27s/it] 49%|████▊ | 14600/30000 [9:20:37<9:47:18, 2.29s/it] 49%|████▊ | 14600/30000 [9:20:37<9:47:18, 2.29s/it] 49%|████▊ | 14601/30000 [9:20:39<9:44:11, 2.28s/it] 49%|████▊ | 14602/30000 [9:20:42<9:52:43, 2.31s/it] 49%|████▊ | 14603/30000 [9:20:44<9:51:06, 2.30s/it] 49%|████▊ | 14604/30000 [9:20:46<9:46:14, 2.28s/it] 49%|████▊ | 14605/30000 [9:20:49<9:42:09, 2.27s/it] 49%|████▊ | 14606/30000 [9:20:51<9:50:28, 2.30s/it] 49%|████▊ | 14607/30000 [9:20:53<9:50:27, 2.30s/it] 49%|████▊ | 14608/30000 [9:20:55<9:42:13, 2.27s/it] 49%|████▊ | 14609/30000 [9:20:58<9:51:35, 2.31s/it] 49%|████▊ | 14610/30000 [9:21:00<9:53:13, 2.31s/it] 49%|████▊ | 14610/30000 [9:21:00<9:53:13, 2.31s/it] 49%|████▊ | 14611/30000 [9:21:02<9:54:39, 2.32s/it] 49%|████▊ | 14612/30000 [9:21:05<9:48:36, 2.30s/it] 49%|████▊ | 14613/30000 [9:21:07<9:52:25, 2.31s/it] 49%|████▊ | 14614/30000 [9:21:09<9:53:30, 2.31s/it] 49%|████▊ | 14615/30000 [9:21:12<9:45:15, 2.28s/it] 49%|████▊ | 14616/30000 [9:21:14<9:51:54, 2.31s/it] 49%|████▊ | 14617/30000 [9:21:16<9:51:52, 2.31s/it] 49%|████▊ | 14618/30000 [9:21:19<9:49:47, 2.30s/it] 49%|████▊ | 14619/30000 [9:21:21<9:52:34, 2.31s/it] 49%|████▊ | 14620/30000 [9:21:23<9:52:14, 2.31s/it] 49%|████▊ | 14620/30000 [9:21:23<9:52:14, 2.31s/it] 49%|████▊ | 14621/30000 [9:21:26<9:51:42, 2.31s/it] 49%|████▊ | 14622/30000 [9:21:28<9:48:54, 2.30s/it] 49%|████▊ | 14623/30000 [9:21:30<9:54:44, 2.32s/it] 49%|████▊ | 14624/30000 [9:21:32<9:52:02, 2.31s/it] 49%|████▉ | 14625/30000 [9:21:35<9:38:03, 2.26s/it] 49%|████▉ | 14626/30000 [9:21:37<9:46:10, 2.29s/it] 49%|████▉ | 14627/30000 [9:21:39<9:46:19, 2.29s/it] 49%|████▉ | 14628/30000 [9:21:41<9:40:26, 2.27s/it] 49%|████▉ | 14629/30000 [9:21:44<9:45:36, 2.29s/it] 49%|████▉ | 14630/30000 [9:21:46<9:47:34, 2.29s/it] 49%|████▉ | 14630/30000 [9:21:46<9:47:34, 2.29s/it] 49%|████▉ | 14631/30000 [9:21:48<9:47:35, 2.29s/it] 49%|████▉ | 14632/30000 [9:21:51<9:39:05, 2.26s/it] 49%|████▉ | 14633/30000 [9:21:53<9:46:01, 2.29s/it] 49%|████▉ | 14634/30000 [9:21:55<9:42:50, 2.28s/it] 49%|████▉ | 14635/30000 [9:21:57<9:41:18, 2.27s/it] 49%|████▉ | 14636/30000 [9:22:00<9:44:19, 2.28s/it] 49%|████▉ | 14637/30000 [9:22:02<9:44:56, 2.28s/it] 49%|████▉ | 14638/30000 [9:22:04<9:45:05, 2.29s/it] 49%|████▉ | 14639/30000 [9:22:07<9:40:02, 2.27s/it] 49%|████▉ | 14640/30000 [9:22:09<9:44:22, 2.28s/it] 49%|████▉ | 14640/30000 [9:22:09<9:44:22, 2.28s/it] 49%|████▉ | 14641/30000 [9:22:11<9:49:49, 2.30s/it] 49%|████▉ | 14642/30000 [9:22:13<9:48:25, 2.30s/it] 49%|████▉ | 14643/30000 [9:22:16<9:49:04, 2.30s/it] 49%|████▉ | 14644/30000 [9:22:18<9:48:13, 2.30s/it] 49%|████▉ | 14645/30000 [9:22:20<9:43:17, 2.28s/it] 49%|████▉ | 14646/30000 [9:22:23<9:39:21, 2.26s/it] 49%|████▉ | 14647/30000 [9:22:25<9:52:57, 2.32s/it] 49%|████▉ | 14648/30000 [9:22:27<9:51:54, 2.31s/it] 49%|████▉ | 14649/30000 [9:22:30<9:48:26, 2.30s/it] 49%|████▉ | 14650/30000 [9:22:32<9:44:04, 2.28s/it] 49%|████▉ | 14650/30000 [9:22:32<9:44:04, 2.28s/it] 49%|████▉ | 14651/30000 [9:22:34<9:45:05, 2.29s/it] 49%|████▉ | 14652/30000 [9:22:36<9:41:36, 2.27s/it] 49%|████▉ | 14653/30000 [9:22:39<9:42:41, 2.28s/it] 49%|████▉ | 14654/30000 [9:22:41<9:48:15, 2.30s/it] 49%|████▉ | 14655/30000 [9:22:43<9:51:27, 2.31s/it] 49%|████▉ | 14656/30000 [9:22:46<9:48:45, 2.30s/it] 49%|████▉ | 14657/30000 [9:22:48<9:45:41, 2.29s/it] 49%|████▉ | 14658/30000 [9:22:50<9:50:30, 2.31s/it] 49%|████▉ | 14659/30000 [9:22:52<9:43:55, 2.28s/it] 49%|████▉ | 14660/30000 [9:22:55<9:42:10, 2.28s/it] 49%|████▉ | 14660/30000 [9:22:55<9:42:10, 2.28s/it] 49%|████▉ | 14661/30000 [9:22:57<9:47:29, 2.30s/it] 49%|████▉ | 14662/30000 [9:22:59<9:55:13, 2.33s/it] 49%|████▉ | 14663/30000 [9:23:02<9:49:26, 2.31s/it] 49%|████▉ | 14664/30000 [9:23:04<9:47:40, 2.30s/it] 49%|████▉ | 14665/30000 [9:23:06<9:51:21, 2.31s/it] 49%|████▉ | 14666/30000 [9:23:09<9:50:48, 2.31s/it] 49%|████▉ | 14667/30000 [9:23:11<9:47:09, 2.30s/it] 49%|████▉ | 14668/30000 [9:23:13<9:46:29, 2.30s/it] 49%|████▉ | 14669/30000 [9:23:16<9:50:34, 2.31s/it] 49%|████▉ | 14670/30000 [9:23:18<9:50:36, 2.31s/it] 49%|████▉ | 14670/30000 [9:23:18<9:50:36, 2.31s/it] 49%|████▉ | 14671/30000 [9:23:20<9:48:06, 2.30s/it] 49%|████▉ | 14672/30000 [9:23:22<9:49:00, 2.31s/it] 49%|████▉ | 14673/30000 [9:23:25<9:51:55, 2.32s/it] 49%|████▉ | 14674/30000 [9:23:27<9:49:12, 2.31s/it] 49%|████▉ | 14675/30000 [9:23:29<9:56:11, 2.33s/it] 49%|████▉ | 14676/30000 [9:23:32<9:51:29, 2.32s/it] 49%|████▉ | 14677/30000 [9:23:34<9:51:58, 2.32s/it] 49%|████▉ | 14678/30000 [9:23:36<9:39:53, 2.27s/it] 49%|████▉ | 14679/30000 [9:23:38<9:37:15, 2.26s/it] 49%|████▉ | 14680/30000 [9:23:41<9:45:00, 2.29s/it] 49%|████▉ | 14680/30000 [9:23:41<9:45:00, 2.29s/it] 49%|████▉ | 14681/30000 [9:23:43<9:48:02, 2.30s/it] 49%|████▉ | 14682/30000 [9:23:45<9:45:52, 2.29s/it] 49%|████▉ | 14683/30000 [9:23:48<9:45:49, 2.29s/it] 49%|████▉ | 14684/30000 [9:23:50<9:49:08, 2.31s/it] 49%|████▉ | 14685/30000 [9:23:52<9:46:55, 2.30s/it] 49%|████▉ | 14686/30000 [9:23:55<9:46:06, 2.30s/it] 49%|████▉ | 14687/30000 [9:23:57<9:44:58, 2.29s/it] 49%|████▉ | 14688/30000 [9:23:59<9:46:49, 2.30s/it] 49%|████▉ | 14689/30000 [9:24:02<9:47:02, 2.30s/it] 49%|████▉ | 14690/30000 [9:24:04<9:45:09, 2.29s/it] 49%|████▉ | 14690/30000 [9:24:04<9:45:09, 2.29s/it] 49%|████▉ | 14691/30000 [9:24:06<9:48:30, 2.31s/it] 49%|████▉ | 14692/30000 [9:24:09<9:57:14, 2.34s/it] 49%|████▉ | 14693/30000 [9:24:11<9:50:52, 2.32s/it] 49%|████▉ | 14694/30000 [9:24:13<9:53:36, 2.33s/it] 49%|████▉ | 14695/30000 [9:24:16<9:54:17, 2.33s/it] 49%|████▉ | 14696/30000 [9:24:18<9:47:36, 2.30s/it] 49%|████▉ | 14697/30000 [9:24:20<9:44:11, 2.29s/it] 49%|████▉ | 14698/30000 [9:24:22<9:47:26, 2.30s/it] 49%|████▉ | 14699/30000 [9:24:25<9:48:23, 2.31s/it] 49%|████▉ | 14700/30000 [9:24:27<9:46:38, 2.30s/it] 49%|████▉ | 14700/30000 [9:24:27<9:46:38, 2.30s/it] 49%|████▉ | 14701/30000 [9:24:29<9:50:10, 2.31s/it] 49%|████▉ | 14702/30000 [9:24:32<9:55:35, 2.34s/it] 49%|████▉ | 14703/30000 [9:24:34<9:53:15, 2.33s/it] 49%|████▉ | 14704/30000 [9:24:36<9:48:11, 2.31s/it] 49%|████▉ | 14705/30000 [9:24:39<9:43:47, 2.29s/it] 49%|████▉ | 14706/30000 [9:24:41<9:48:51, 2.31s/it] 49%|████▉ | 14707/30000 [9:24:43<9:51:16, 2.32s/it] 49%|████▉ | 14708/30000 [9:24:45<9:45:13, 2.30s/it] 49%|████▉ | 14709/30000 [9:24:48<9:45:24, 2.30s/it] 49%|████▉ | 14710/30000 [9:24:50<9:43:40, 2.29s/it] 49%|████▉ | 14710/30000 [9:24:50<9:43:40, 2.29s/it] 49%|████▉ | 14711/30000 [9:24:52<9:40:46, 2.28s/it] 49%|████▉ | 14712/30000 [9:24:55<9:41:03, 2.28s/it] 49%|████▉ | 14713/30000 [9:24:57<9:44:23, 2.29s/it] 49%|████▉ | 14714/30000 [9:24:59<9:42:45, 2.29s/it] 49%|████▉ | 14715/30000 [9:25:01<9:43:56, 2.29s/it] 49%|████▉ | 14716/30000 [9:25:04<9:47:12, 2.31s/it] 49%|████▉ | 14717/30000 [9:25:06<9:46:00, 2.30s/it] 49%|████▉ | 14718/30000 [9:25:08<9:51:53, 2.32s/it] 49%|████▉ | 14719/30000 [9:25:11<9:46:00, 2.30s/it] 49%|████▉ | 14720/30000 [9:25:13<9:57:21, 2.35s/it] 49%|████▉ | 14720/30000 [9:25:13<9:57:21, 2.35s/it] 49%|████▉ | 14721/30000 [9:25:15<9:53:08, 2.33s/it] 49%|████▉ | 14722/30000 [9:25:18<9:49:49, 2.32s/it] 49%|████▉ | 14723/30000 [9:25:20<9:47:47, 2.31s/it] 49%|████▉ | 14724/30000 [9:25:22<9:46:35, 2.30s/it] 49%|████▉ | 14725/30000 [9:25:25<9:43:23, 2.29s/it] 49%|████▉ | 14726/30000 [9:25:27<9:38:57, 2.27s/it] 49%|████▉ | 14727/30000 [9:25:29<9:44:09, 2.29s/it] 49%|████▉ | 14728/30000 [9:25:31<9:36:11, 2.26s/it] 49%|████▉ | 14729/30000 [9:25:34<9:40:27, 2.28s/it] 49%|████▉ | 14730/30000 [9:25:36<9:43:26, 2.29s/it] 49%|████▉ | 14730/30000 [9:25:36<9:43:26, 2.29s/it] 49%|████▉ | 14731/30000 [9:25:38<9:48:00, 2.31s/it] 49%|████▉ | 14732/30000 [9:25:41<9:43:53, 2.29s/it] 49%|████▉ | 14733/30000 [9:25:43<9:49:12, 2.32s/it] 49%|████▉ | 14734/30000 [9:25:45<9:48:23, 2.31s/it] 49%|████▉ | 14735/30000 [9:25:48<9:46:11, 2.30s/it] 49%|████▉ | 14736/30000 [9:25:50<9:48:33, 2.31s/it] 49%|████▉ | 14737/30000 [9:25:52<9:49:51, 2.32s/it] 49%|████▉ | 14738/30000 [9:25:54<9:44:20, 2.30s/it] 49%|████▉ | 14739/30000 [9:25:57<9:47:21, 2.31s/it] 49%|████▉ | 14740/30000 [9:25:59<9:51:03, 2.32s/it] 49%|████▉ | 14740/30000 [9:25:59<9:51:03, 2.32s/it] 49%|████▉ | 14741/30000 [9:26:01<9:48:10, 2.31s/it] 49%|████▉ | 14742/30000 [9:26:04<9:45:31, 2.30s/it] 49%|████▉ | 14743/30000 [9:26:06<9:47:48, 2.31s/it] 49%|████▉ | 14744/30000 [9:26:08<9:48:02, 2.31s/it] 49%|████▉ | 14745/30000 [9:26:11<9:56:01, 2.34s/it] 49%|████▉ | 14746/30000 [9:26:13<9:53:21, 2.33s/it] 49%|████▉ | 14747/30000 [9:26:15<9:55:05, 2.34s/it] 49%|████▉ | 14748/30000 [9:26:18<9:45:10, 2.30s/it] 49%|████▉ | 14749/30000 [9:26:20<9:46:48, 2.31s/it] 49%|████▉ | 14750/30000 [9:26:22<9:39:41, 2.28s/it] 49%|████▉ | 14750/30000 [9:26:22<9:39:41, 2.28s/it] 49%|████▉ | 14751/30000 [9:26:24<9:39:39, 2.28s/it] 49%|████▉ | 14752/30000 [9:26:27<9:43:02, 2.29s/it] 49%|████▉ | 14753/30000 [9:26:29<9:43:11, 2.29s/it] 49%|████▉ | 14754/30000 [9:26:31<9:43:24, 2.30s/it] 49%|████▉ | 14755/30000 [9:26:34<9:54:50, 2.34s/it] 49%|████▉ | 14756/30000 [9:26:36<9:53:46, 2.34s/it] 49%|████▉ | 14757/30000 [9:26:38<9:49:06, 2.32s/it] 49%|████▉ | 14758/30000 [9:26:41<9:46:24, 2.31s/it] 49%|████▉ | 14759/30000 [9:26:43<9:42:52, 2.29s/it] 49%|████▉ | 14760/30000 [9:26:45<9:42:28, 2.29s/it] 49%|████▉ | 14760/30000 [9:26:45<9:42:28, 2.29s/it] 49%|████▉ | 14761/30000 [9:26:48<9:42:12, 2.29s/it] 49%|████▉ | 14762/30000 [9:26:50<9:40:54, 2.29s/it] 49%|████▉ | 14763/30000 [9:26:52<9:44:24, 2.30s/it] 49%|████▉ | 14764/30000 [9:26:54<9:41:49, 2.29s/it] 49%|████▉ | 14765/30000 [9:26:57<9:40:12, 2.29s/it] 49%|████▉ | 14766/30000 [9:26:59<9:44:04, 2.30s/it] 49%|████▉ | 14767/30000 [9:27:01<9:46:05, 2.31s/it] 49%|████▉ | 14768/30000 [9:27:04<9:42:15, 2.29s/it] 49%|████▉ | 14769/30000 [9:27:06<9:58:48, 2.36s/it] 49%|████▉ | 14770/30000 [9:27:08<9:53:09, 2.34s/it] 49%|████▉ | 14770/30000 [9:27:09<9:53:09, 2.34s/it] 49%|████▉ | 14771/30000 [9:27:11<9:49:17, 2.32s/it] 49%|████▉ | 14772/30000 [9:27:13<9:50:57, 2.33s/it] 49%|████▉ | 14773/30000 [9:27:16<9:58:04, 2.36s/it] 49%|████▉ | 14774/30000 [9:27:18<9:52:56, 2.34s/it] 49%|████▉ | 14775/30000 [9:27:20<9:52:07, 2.33s/it] 49%|████▉ | 14776/30000 [9:27:22<9:51:23, 2.33s/it] 49%|████▉ | 14777/30000 [9:27:25<9:49:35, 2.32s/it] 49%|████▉ | 14778/30000 [9:27:27<9:36:43, 2.27s/it] 49%|████▉ | 14779/30000 [9:27:29<9:39:48, 2.29s/it] 49%|████▉ | 14780/30000 [9:27:32<9:39:18, 2.28s/it] 49%|████▉ | 14780/30000 [9:27:32<9:39:18, 2.28s/it] 49%|████▉ | 14781/30000 [9:27:34<9:39:42, 2.29s/it] 49%|████▉ | 14782/30000 [9:27:36<9:33:59, 2.26s/it] 49%|████▉ | 14783/30000 [9:27:38<9:35:58, 2.27s/it] 49%|████▉ | 14784/30000 [9:27:41<9:40:14, 2.29s/it] 49%|████▉ | 14785/30000 [9:27:43<9:44:15, 2.30s/it] 49%|████▉ | 14786/30000 [9:27:45<9:44:14, 2.30s/it] 49%|████▉ | 14787/30000 [9:27:48<9:39:46, 2.29s/it] 49%|████▉ | 14788/30000 [9:27:50<9:43:56, 2.30s/it] 49%|████▉ | 14789/30000 [9:27:52<9:46:42, 2.31s/it] 49%|████▉ | 14790/30000 [9:27:54<9:41:40, 2.29s/it] 49%|████▉ | 14790/30000 [9:27:54<9:41:40, 2.29s/it] 49%|████▉ | 14791/30000 [9:27:57<9:49:02, 2.32s/it] 49%|████▉ | 14792/30000 [9:27:59<9:50:30, 2.33s/it] 49%|████▉ | 14793/30000 [9:28:02<9:49:29, 2.33s/it] 49%|████▉ | 14794/30000 [9:28:04<9:45:46, 2.31s/it] 49%|████▉ | 14795/30000 [9:28:06<9:50:05, 2.33s/it] 49%|████▉ | 14796/30000 [9:28:08<9:49:51, 2.33s/it] 49%|████▉ | 14797/30000 [9:28:11<9:36:57, 2.28s/it] 49%|████▉ | 14798/30000 [9:28:13<9:41:48, 2.30s/it] 49%|████▉ | 14799/30000 [9:28:15<9:41:59, 2.30s/it] 49%|████▉ | 14800/30000 [9:28:18<9:41:39, 2.30s/it] 49%|████▉ | 14800/30000 [9:28:18<9:41:39, 2.30s/it] 49%|████▉ | 14801/30000 [9:28:20<9:48:29, 2.32s/it] 49%|████▉ | 14802/30000 [9:28:22<9:53:52, 2.34s/it] 49%|████▉ | 14803/30000 [9:28:25<9:49:21, 2.33s/it] 49%|████▉ | 14804/30000 [9:28:27<9:46:34, 2.32s/it] 49%|████▉ | 14805/30000 [9:28:29<9:48:49, 2.33s/it] 49%|████▉ | 14806/30000 [9:28:32<9:44:29, 2.31s/it] 49%|████▉ | 14807/30000 [9:28:34<9:51:02, 2.33s/it] 49%|████▉ | 14808/30000 [9:28:36<9:53:02, 2.34s/it] 49%|████▉ | 14809/30000 [9:28:39<9:50:32, 2.33s/it] 49%|████▉ | 14810/30000 [9:28:41<9:49:09, 2.33s/it] 49%|████▉ | 14810/30000 [9:28:41<9:49:09, 2.33s/it] 49%|████▉ | 14811/30000 [9:28:43<9:51:33, 2.34s/it] 49%|████▉ | 14812/30000 [9:28:46<9:50:18, 2.33s/it] 49%|████▉ | 14813/30000 [9:28:48<9:44:12, 2.31s/it] 49%|████▉ | 14814/30000 [9:28:50<9:39:49, 2.29s/it] 49%|████▉ | 14815/30000 [9:28:53<9:52:33, 2.34s/it] 49%|████▉ | 14816/30000 [9:28:55<9:43:18, 2.30s/it] 49%|████▉ | 14817/30000 [9:28:57<9:39:30, 2.29s/it] 49%|████▉ | 14818/30000 [9:28:59<9:42:16, 2.30s/it] 49%|████▉ | 14819/30000 [9:29:02<9:35:27, 2.27s/it] 49%|████▉ | 14820/30000 [9:29:04<9:33:24, 2.27s/it] 49%|████▉ | 14820/30000 [9:29:04<9:33:24, 2.27s/it] 49%|████▉ | 14821/30000 [9:29:06<9:35:45, 2.28s/it] 49%|████▉ | 14822/30000 [9:29:08<9:27:08, 2.24s/it] 49%|████▉ | 14823/30000 [9:29:11<9:33:12, 2.27s/it] 49%|████▉ | 14824/30000 [9:29:13<9:35:30, 2.28s/it] 49%|████▉ | 14825/30000 [9:29:15<9:38:08, 2.29s/it] 49%|████▉ | 14826/30000 [9:29:18<9:38:30, 2.29s/it] 49%|████▉ | 14827/30000 [9:29:20<9:38:00, 2.29s/it] 49%|████▉ | 14828/30000 [9:29:22<9:37:18, 2.28s/it] 49%|████▉ | 14829/30000 [9:29:24<9:36:06, 2.28s/it] 49%|████▉ | 14830/30000 [9:29:27<9:41:01, 2.30s/it] 49%|████▉ | 14830/30000 [9:29:27<9:41:01, 2.30s/it] 49%|████▉ | 14831/30000 [9:29:29<9:43:37, 2.31s/it] 49%|████▉ | 14832/30000 [9:29:31<9:42:37, 2.30s/it] 49%|████▉ | 14833/30000 [9:29:34<9:48:00, 2.33s/it] 49%|████▉ | 14834/30000 [9:29:36<9:50:05, 2.33s/it] 49%|████▉ | 14835/30000 [9:29:38<9:45:11, 2.32s/it] 49%|████▉ | 14836/30000 [9:29:41<9:43:54, 2.31s/it] 49%|████▉ | 14837/30000 [9:29:43<9:43:55, 2.31s/it] 49%|████▉ | 14838/30000 [9:29:45<9:43:43, 2.31s/it] 49%|████▉ | 14839/30000 [9:29:47<9:34:59, 2.28s/it] 49%|████▉ | 14840/30000 [9:29:50<9:38:25, 2.29s/it] 49%|████▉ | 14840/30000 [9:29:50<9:38:25, 2.29s/it] 49%|████▉ | 14841/30000 [9:29:52<9:40:16, 2.30s/it] 49%|████▉ | 14842/30000 [9:29:54<9:40:05, 2.30s/it] 49%|████▉ | 14843/30000 [9:29:57<9:39:00, 2.29s/it] 49%|████▉ | 14844/30000 [9:29:59<9:36:17, 2.28s/it] 49%|████▉ | 14845/30000 [9:30:01<9:32:09, 2.27s/it] 49%|████▉ | 14846/30000 [9:30:03<9:35:10, 2.28s/it] 49%|████▉ | 14847/30000 [9:30:06<9:29:37, 2.26s/it] 49%|████▉ | 14848/30000 [9:30:08<9:18:51, 2.21s/it] 49%|████▉ | 14849/30000 [9:30:10<9:29:20, 2.25s/it] 50%|████▉ | 14850/30000 [9:30:12<9:32:35, 2.27s/it] 50%|████▉ | 14850/30000 [9:30:12<9:32:35, 2.27s/it] 50%|████▉ | 14851/30000 [9:30:15<9:33:49, 2.27s/it] 50%|████▉ | 14852/30000 [9:30:17<9:34:54, 2.28s/it] 50%|████▉ | 14853/30000 [9:30:19<9:39:05, 2.29s/it] 50%|████▉ | 14854/30000 [9:30:22<9:35:17, 2.28s/it] 50%|████▉ | 14855/30000 [9:30:24<9:41:21, 2.30s/it] 50%|████▉ | 14856/30000 [9:30:26<9:42:29, 2.31s/it] 50%|████▉ | 14857/30000 [9:30:28<9:40:04, 2.30s/it] 50%|████▉ | 14858/30000 [9:30:31<9:38:25, 2.29s/it] 50%|████▉ | 14859/30000 [9:30:33<9:36:19, 2.28s/it] 50%|████▉ | 14860/30000 [9:30:35<9:35:53, 2.28s/it] 50%|████▉ | 14860/30000 [9:30:35<9:35:53, 2.28s/it] 50%|████▉ | 14861/30000 [9:30:38<9:33:33, 2.27s/it] 50%|████▉ | 14862/30000 [9:30:40<9:31:44, 2.27s/it] 50%|████▉ | 14863/30000 [9:30:42<9:46:40, 2.33s/it] 50%|████▉ | 14864/30000 [9:30:45<9:44:22, 2.32s/it] 50%|████▉ | 14865/30000 [9:30:47<9:43:24, 2.31s/it] 50%|████▉ | 14866/30000 [9:30:49<9:41:55, 2.31s/it] 50%|████▉ | 14867/30000 [9:30:52<9:47:44, 2.33s/it] 50%|████▉ | 14868/30000 [9:30:54<9:44:41, 2.32s/it] 50%|████▉ | 14869/30000 [9:30:56<9:47:49, 2.33s/it] 50%|████▉ | 14870/30000 [9:30:58<9:38:12, 2.29s/it] 50%|████▉ | 14870/30000 [9:30:58<9:38:12, 2.29s/it] 50%|████▉ | 14871/30000 [9:31:01<9:46:56, 2.33s/it] 50%|████▉ | 14872/30000 [9:31:03<9:45:07, 2.32s/it] 50%|████▉ | 14873/30000 [9:31:05<9:44:02, 2.32s/it] 50%|████▉ | 14874/30000 [9:31:08<9:40:42, 2.30s/it] 50%|████▉ | 14875/30000 [9:31:10<9:44:38, 2.32s/it] 50%|████▉ | 14876/30000 [9:31:12<9:43:49, 2.32s/it] 50%|████▉ | 14877/30000 [9:31:15<9:48:26, 2.33s/it] 50%|████▉ | 14878/30000 [9:31:17<9:36:11, 2.29s/it] 50%|████▉ | 14879/30000 [9:31:19<9:35:59, 2.29s/it] 50%|████▉ | 14880/30000 [9:31:21<9:33:32, 2.28s/it] 50%|████▉ | 14880/30000 [9:31:22<9:33:32, 2.28s/it] 50%|████▉ | 14881/30000 [9:31:24<9:36:35, 2.29s/it] 50%|████▉ | 14882/30000 [9:31:26<9:41:36, 2.31s/it] 50%|████▉ | 14883/30000 [9:31:28<9:41:23, 2.31s/it] 50%|████▉ | 14884/30000 [9:31:31<9:37:44, 2.29s/it] 50%|████▉ | 14885/30000 [9:31:33<9:34:27, 2.28s/it] 50%|████▉ | 14886/30000 [9:31:35<9:41:08, 2.31s/it] 50%|████▉ | 14887/30000 [9:31:38<9:40:03, 2.30s/it] 50%|████▉ | 14888/30000 [9:31:40<9:37:31, 2.29s/it] 50%|████▉ | 14889/30000 [9:31:42<9:39:28, 2.30s/it] 50%|████▉ | 14890/30000 [9:31:45<9:40:52, 2.31s/it] 50%|████▉ | 14890/30000 [9:31:45<9:40:52, 2.31s/it] 50%|████▉ | 14891/30000 [9:31:47<9:45:30, 2.33s/it] 50%|████▉ | 14892/30000 [9:31:49<9:38:58, 2.30s/it] 50%|████▉ | 14893/30000 [9:31:51<9:38:20, 2.30s/it] 50%|████▉ | 14894/30000 [9:31:54<9:31:46, 2.27s/it] 50%|████▉ | 14895/30000 [9:31:56<9:29:16, 2.26s/it] 50%|████▉ | 14896/30000 [9:31:58<9:32:00, 2.27s/it] 50%|████▉ | 14897/30000 [9:32:00<9:33:14, 2.28s/it] 50%|████▉ | 14898/30000 [9:32:03<9:29:29, 2.26s/it] 50%|████▉ | 14899/30000 [9:32:05<9:32:00, 2.27s/it] 50%|████▉ | 14900/30000 [9:32:07<9:37:40, 2.30s/it] 50%|████▉ | 14900/30000 [9:32:07<9:37:40, 2.30s/it] 50%|████▉ | 14901/30000 [9:32:10<9:47:13, 2.33s/it] 50%|████▉ | 14902/30000 [9:32:12<9:40:54, 2.31s/it] 50%|████▉ | 14903/30000 [9:32:14<9:37:19, 2.29s/it] 50%|████▉ | 14904/30000 [9:32:17<9:37:16, 2.29s/it] 50%|████▉ | 14905/30000 [9:32:19<9:35:33, 2.29s/it] 50%|████▉ | 14906/30000 [9:32:21<9:32:55, 2.28s/it] 50%|████▉ | 14907/30000 [9:32:23<9:36:06, 2.29s/it] 50%|████▉ | 14908/30000 [9:32:26<9:34:40, 2.28s/it] 50%|████▉ | 14909/30000 [9:32:28<9:25:52, 2.25s/it] 50%|████▉ | 14910/30000 [9:32:30<9:32:50, 2.28s/it] +{'loss': 0.0138, 'grad_norm': 0.07026781886816025, 'learning_rate': 1.704556217045536e-05, 'flow_matching_loss': 0.013810612075030804, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.22} +{'loss': 0.0153, 'grad_norm': 0.0818885937333107, 'learning_rate': 1.7029180702233612e-05, 'flow_matching_loss': 0.015287307277321815, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.22} +{'loss': 0.0155, 'grad_norm': 0.080250583589077, 'learning_rate': 1.7012796768365975e-05, 'flow_matching_loss': 0.01548102181404829, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.23} +{'loss': 0.0151, 'grad_norm': 0.07762870192527771, 'learning_rate': 1.6996410388760465e-05, 'flow_matching_loss': 0.015137349348515272, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.23} +{'loss': 0.0149, 'grad_norm': 0.08936086297035217, 'learning_rate': 1.698002158332809e-05, 'flow_matching_loss': 0.014925900567322969, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.23} +{'loss': 0.0155, 'grad_norm': 0.08747684955596924, 'learning_rate': 1.6963630371982783e-05, 'flow_matching_loss': 0.015466941148042678, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.24} +{'loss': 0.0156, 'grad_norm': 0.07859471440315247, 'learning_rate': 1.6947236774641416e-05, 'flow_matching_loss': 0.015566913038492202, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.24} +{'loss': 0.016, 'grad_norm': 0.09522400796413422, 'learning_rate': 1.6930840811223755e-05, 'flow_matching_loss': 0.01602203520014882, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.24} +{'loss': 0.0157, 'grad_norm': 0.09615881741046906, 'learning_rate': 1.6914442501652448e-05, 'flow_matching_loss': 0.015669668186455966, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.24} +{'loss': 0.0142, 'grad_norm': 0.07316907495260239, 'learning_rate': 1.6898041865852974e-05, 'flow_matching_loss': 0.014179198537021875, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.25} +{'loss': 0.0143, 'grad_norm': 0.06751125305891037, 'learning_rate': 1.6881638923753663e-05, 'flow_matching_loss': 0.014304464962333441, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.25} +{'loss': 0.0152, 'grad_norm': 0.07092851400375366, 'learning_rate': 1.6865233695285632e-05, 'flow_matching_loss': 0.015227616764605045, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.25} +{'loss': 0.015, 'grad_norm': 0.07415245473384857, 'learning_rate': 1.6848826200382782e-05, 'flow_matching_loss': 0.014966520294547082, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.26} +{'loss': 0.0145, 'grad_norm': 0.07499248534440994, 'learning_rate': 1.6832416458981762e-05, 'flow_matching_loss': 0.01451623197644949, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.26} +{'loss': 0.0148, 'grad_norm': 0.07175338268280029, 'learning_rate': 1.681600449102196e-05, 'flow_matching_loss': 0.014770927652716637, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.26} +{'loss': 0.0147, 'grad_norm': 0.07027586549520493, 'learning_rate': 1.6799590316445468e-05, 'flow_matching_loss': 0.014665864594280719, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.27} +{'loss': 0.0138, 'grad_norm': 0.07819302380084991, 'learning_rate': 1.678317395519705e-05, 'flow_matching_loss': 0.013845999538898469, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.27} +{'loss': 0.0162, 'grad_norm': 0.076007179915905, 'learning_rate': 1.676675542722413e-05, 'flow_matching_loss': 0.016228861827403306, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.27} +{'loss': 0.0143, 'grad_norm': 0.07354190945625305, 'learning_rate': 1.6750334752476776e-05, 'flow_matching_loss': 0.014277228061109782, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.27} +{'loss': 0.0143, 'grad_norm': 0.09136087447404861, 'learning_rate': 1.673391195090765e-05, 'flow_matching_loss': 0.014257513172924518, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.28} +{'loss': 0.0146, 'grad_norm': 0.09176922589540482, 'learning_rate': 1.6717487042472006e-05, 'flow_matching_loss': 0.014636557269841432, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.28} +{'loss': 0.0163, 'grad_norm': 0.08882725238800049, 'learning_rate': 1.6701060047127655e-05, 'flow_matching_loss': 0.016269149631261824, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.28} +{'loss': 0.014, 'grad_norm': 0.06758830696344376, 'learning_rate': 1.6684630984834947e-05, 'flow_matching_loss': 0.014009154122322797, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.29} +{'loss': 0.0158, 'grad_norm': 0.06822438538074493, 'learning_rate': 1.666819987555675e-05, 'flow_matching_loss': 0.01577377216890454, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.29} +{'loss': 0.0166, 'grad_norm': 0.0625971108675003, 'learning_rate': 1.6651766739258392e-05, 'flow_matching_loss': 0.016556976083666086, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.29} +{'loss': 0.0156, 'grad_norm': 0.08312332630157471, 'learning_rate': 1.6635331595907704e-05, 'flow_matching_loss': 0.015623989142477512, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.29} +{'loss': 0.0141, 'grad_norm': 0.0933290421962738, 'learning_rate': 1.6618894465474923e-05, 'flow_matching_loss': 0.014134983625262975, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.3} +{'loss': 0.0147, 'grad_norm': 0.09101437032222748, 'learning_rate': 1.660245536793271e-05, 'flow_matching_loss': 0.014747746102511883, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.3} +{'loss': 0.0151, 'grad_norm': 0.09943174570798874, 'learning_rate': 1.6586014323256126e-05, 'flow_matching_loss': 0.01505565932020545, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.3} +{'loss': 0.0149, 'grad_norm': 0.07880614697933197, 'learning_rate': 1.656957135142259e-05, 'flow_matching_loss': 0.014893659763038159, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.31} +{'loss': 0.0144, 'grad_norm': 0.08433464169502258, 'learning_rate': 1.655312647241186e-05, 'flow_matching_loss': 0.014372179564088584, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.31} +{'loss': 0.0161, 'grad_norm': 0.08304093778133392, 'learning_rate': 1.6536679706206015e-05, 'flow_matching_loss': 0.016091348510235547, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.31} +{'loss': 0.0147, 'grad_norm': 0.0659429207444191, 'learning_rate': 1.6520231072789427e-05, 'flow_matching_loss': 0.01472490355372429, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.31} +{'loss': 0.0134, 'grad_norm': 0.06995540112257004, 'learning_rate': 1.6503780592148737e-05, 'flow_matching_loss': 0.013412334769964219, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.32} +{'loss': 0.0155, 'grad_norm': 0.07858077436685562, 'learning_rate': 1.648732828427283e-05, 'flow_matching_loss': 0.015463532041758299, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.32} +{'loss': 0.014, 'grad_norm': 0.08520372211933136, 'learning_rate': 1.647087416915281e-05, 'flow_matching_loss': 0.013993626181036233, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.32} +{'loss': 0.0141, 'grad_norm': 0.08398837596178055, 'learning_rate': 1.6454418266781987e-05, 'flow_matching_loss': 0.014067402575165033, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.33} +{'loss': 0.0154, 'grad_norm': 0.09523718804121017, 'learning_rate': 1.6437960597155824e-05, 'flow_matching_loss': 0.015399997308850288, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.33} +{'loss': 0.0156, 'grad_norm': 0.0763525739312172, 'learning_rate': 1.642150118027194e-05, 'flow_matching_loss': 0.015567205101251601, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.33} +{'loss': 0.0158, 'grad_norm': 0.07996321469545364, 'learning_rate': 1.640504003613009e-05, 'flow_matching_loss': 0.015831305459141732, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.34} +{'loss': 0.0151, 'grad_norm': 0.07606387138366699, 'learning_rate': 1.6388577184732107e-05, 'flow_matching_loss': 0.015092579182237387, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.34} 50%|████▉ | 14910/30000 [9:32:30<9:32:50, 2.28s/it] 50%|████▉ | 14911/30000 [9:32:33<9:46:06, 2.33s/it] 50%|████▉ | 14912/30000 [9:32:35<9:43:34, 2.32s/it] 50%|████▉ | 14913/30000 [9:32:37<9:41:19, 2.31s/it] 50%|████▉ | 14914/30000 [9:32:40<9:45:35, 2.33s/it] 50%|████▉ | 14915/30000 [9:32:42<9:46:03, 2.33s/it] 50%|████▉ | 14916/30000 [9:32:44<9:50:19, 2.35s/it] 50%|████▉ | 14917/30000 [9:32:47<9:42:33, 2.32s/it] 50%|████▉ | 14918/30000 [9:32:49<9:51:32, 2.35s/it] 50%|████▉ | 14919/30000 [9:32:51<9:43:41, 2.32s/it] 50%|████▉ | 14920/30000 [9:32:54<9:37:52, 2.30s/it] 50%|████▉ | 14920/30000 [9:32:54<9:37:52, 2.30s/it] 50%|████▉ | 14921/30000 [9:32:56<9:38:26, 2.30s/it] 50%|████▉ | 14922/30000 [9:32:58<9:43:11, 2.32s/it] 50%|████▉ | 14923/30000 [9:33:01<9:44:38, 2.33s/it] 50%|████▉ | 14924/30000 [9:33:03<9:41:24, 2.31s/it] 50%|████▉ | 14925/30000 [9:33:05<9:38:07, 2.30s/it] 50%|████▉ | 14926/30000 [9:33:07<9:39:46, 2.31s/it] 50%|████▉ | 14927/30000 [9:33:10<9:34:24, 2.29s/it] 50%|████▉ | 14928/30000 [9:33:12<9:34:02, 2.29s/it] 50%|████▉ | 14929/30000 [9:33:14<9:41:01, 2.31s/it] 50%|████▉ | 14930/30000 [9:33:17<9:42:09, 2.32s/it] 50%|████▉ | 14930/30000 [9:33:17<9:42:09, 2.32s/it] 50%|████▉ | 14931/30000 [9:33:19<9:40:43, 2.31s/it] 50%|████▉ | 14932/30000 [9:33:21<9:32:44, 2.28s/it] 50%|████▉ | 14933/30000 [9:33:23<9:36:50, 2.30s/it] 50%|████▉ | 14934/30000 [9:33:26<9:33:51, 2.29s/it] 50%|████▉ | 14935/30000 [9:33:28<9:32:31, 2.28s/it] 50%|████▉ | 14936/30000 [9:33:30<9:34:35, 2.29s/it] 50%|████▉ | 14937/30000 [9:33:33<9:36:25, 2.30s/it] 50%|████▉ | 14938/30000 [9:33:35<9:35:39, 2.29s/it] 50%|████▉ | 14939/30000 [9:33:37<9:31:51, 2.28s/it] 50%|████▉ | 14940/30000 [9:33:39<9:33:58, 2.29s/it] 50%|████▉ | 14940/30000 [9:33:39<9:33:58, 2.29s/it] 50%|████▉ | 14941/30000 [9:33:42<9:33:16, 2.28s/it] 50%|████▉ | 14942/30000 [9:33:44<9:30:44, 2.27s/it] 50%|████▉ | 14943/30000 [9:33:46<9:36:23, 2.30s/it] 50%|████▉ | 14944/30000 [9:33:49<9:32:24, 2.28s/it] 50%|████▉ | 14945/30000 [9:33:51<9:29:32, 2.27s/it] 50%|████▉ | 14946/30000 [9:33:53<9:27:40, 2.26s/it] 50%|████▉ | 14947/30000 [9:33:55<9:39:19, 2.31s/it] 50%|████▉ | 14948/30000 [9:33:58<9:30:10, 2.27s/it] 50%|████▉ | 14949/30000 [9:34:00<9:33:08, 2.28s/it] 50%|████▉ | 14950/30000 [9:34:02<9:30:59, 2.28s/it] 50%|████▉ | 14950/30000 [9:34:02<9:30:59, 2.28s/it] 50%|████▉ | 14951/30000 [9:34:04<9:29:52, 2.27s/it] 50%|████▉ | 14952/30000 [9:34:07<9:32:34, 2.28s/it] 50%|████▉ | 14953/30000 [9:34:09<9:34:36, 2.29s/it] 50%|████▉ | 14954/30000 [9:34:11<9:29:47, 2.27s/it] 50%|████▉ | 14955/30000 [9:34:14<9:45:14, 2.33s/it] 50%|████▉ | 14956/30000 [9:34:16<9:40:41, 2.32s/it] 50%|████▉ | 14957/30000 [9:34:18<9:44:22, 2.33s/it] 50%|████▉ | 14958/30000 [9:34:21<9:45:04, 2.33s/it] 50%|████▉ | 14959/30000 [9:34:23<9:45:26, 2.34s/it] 50%|████▉ | 14960/30000 [9:34:25<9:40:08, 2.31s/it] 50%|████▉ | 14960/30000 [9:34:25<9:40:08, 2.31s/it] 50%|████▉ | 14961/30000 [9:34:28<9:37:32, 2.30s/it] 50%|████▉ | 14962/30000 [9:34:30<9:50:29, 2.36s/it] 50%|████▉ | 14963/30000 [9:34:32<9:46:08, 2.34s/it] 50%|████▉ | 14964/30000 [9:34:35<9:41:41, 2.32s/it] 50%|████▉ | 14965/30000 [9:34:37<9:41:13, 2.32s/it] 50%|████▉ | 14966/30000 [9:34:39<9:40:54, 2.32s/it] 50%|████▉ | 14967/30000 [9:34:42<9:38:17, 2.31s/it] 50%|████▉ | 14968/30000 [9:34:44<9:32:56, 2.29s/it] 50%|████▉ | 14969/30000 [9:34:46<9:34:43, 2.29s/it] 50%|████▉ | 14970/30000 [9:34:49<9:35:25, 2.30s/it] 50%|████▉ | 14970/30000 [9:34:49<9:35:25, 2.30s/it] 50%|████▉ | 14971/30000 [9:34:51<9:32:03, 2.28s/it] 50%|████▉ | 14972/30000 [9:34:53<9:35:50, 2.30s/it] 50%|████▉ | 14973/30000 [9:34:55<9:35:02, 2.30s/it] 50%|████▉ | 14974/30000 [9:34:58<9:32:34, 2.29s/it] 50%|████▉ | 14975/30000 [9:35:00<9:35:40, 2.30s/it] 50%|████▉ | 14976/30000 [9:35:02<9:46:43, 2.34s/it] 50%|████▉ | 14977/30000 [9:35:05<9:40:02, 2.32s/it] 50%|████▉ | 14978/30000 [9:35:07<9:34:02, 2.29s/it] 50%|████▉ | 14979/30000 [9:35:09<9:38:55, 2.31s/it] 50%|████▉ | 14980/30000 [9:35:12<9:35:51, 2.30s/it] 50%|████▉ | 14980/30000 [9:35:12<9:35:51, 2.30s/it] 50%|████▉ | 14981/30000 [9:35:14<9:35:29, 2.30s/it] 50%|████▉ | 14982/30000 [9:35:16<9:36:55, 2.30s/it] 50%|████▉ | 14983/30000 [9:35:18<9:34:39, 2.30s/it] 50%|████▉ | 14984/30000 [9:35:21<9:33:21, 2.29s/it] 50%|████▉ | 14985/30000 [9:35:23<9:35:28, 2.30s/it] 50%|████▉ | 14986/30000 [9:35:25<9:41:19, 2.32s/it] 50%|████▉ | 14987/30000 [9:35:28<9:41:37, 2.32s/it] 50%|████▉ | 14988/30000 [9:35:30<9:38:42, 2.31s/it] 50%|████▉ | 14989/30000 [9:35:32<9:41:22, 2.32s/it] 50%|████▉ | 14990/30000 [9:35:35<9:41:16, 2.32s/it] 50%|████▉ | 14990/30000 [9:35:35<9:41:16, 2.32s/it] 50%|████▉ | 14991/30000 [9:35:37<9:36:18, 2.30s/it] 50%|████▉ | 14992/30000 [9:35:39<9:35:54, 2.30s/it] 50%|████▉ | 14993/30000 [9:35:42<9:34:02, 2.30s/it] 50%|████▉ | 14994/30000 [9:35:44<9:29:34, 2.28s/it] 50%|████▉ | 14995/30000 [9:35:46<9:28:17, 2.27s/it] 50%|████▉ | 14996/30000 [9:35:48<9:33:12, 2.29s/it] 50%|████▉ | 14997/30000 [9:35:50<9:20:03, 2.24s/it] 50%|████▉ | 14998/30000 [9:35:53<9:24:57, 2.26s/it] 50%|████▉ | 14999/30000 [9:35:55<9:27:01, 2.27s/it] 50%|█████ | 15000/30000 [9:35:57<9:33:04, 2.29s/it] 50%|█████ | 15000/30000 [9:35:57<9:33:04, 2.29s/it] 50%|█████ | 15001/30000 [9:36:00<9:30:06, 2.28s/it] 50%|█████ | 15002/30000 [9:36:02<9:33:27, 2.29s/it] 50%|█████ | 15003/30000 [9:36:04<9:29:04, 2.28s/it] 50%|█████ | 15004/30000 [9:36:06<9:26:15, 2.27s/it] 50%|█████ | 15005/30000 [9:36:09<9:30:26, 2.28s/it] 50%|█████ | 15006/30000 [9:36:11<9:29:44, 2.28s/it] 50%|█████ | 15007/30000 [9:36:13<9:26:06, 2.27s/it] 50%|█████ | 15008/30000 [9:36:16<9:31:45, 2.29s/it] 50%|█████ | 15009/30000 [9:36:18<9:42:40, 2.33s/it] 50%|█████ | 15010/30000 [9:36:20<9:37:38, 2.31s/it] 50%|█████ | 15010/30000 [9:36:20<9:37:38, 2.31s/it] 50%|█████ | 15011/30000 [9:36:23<9:39:17, 2.32s/it] 50%|█████ | 15012/30000 [9:36:25<9:38:10, 2.31s/it] 50%|█████ | 15013/30000 [9:36:27<9:36:34, 2.31s/it] 50%|█████ | 15014/30000 [9:36:30<9:36:29, 2.31s/it] 50%|█████ | 15015/30000 [9:36:32<9:33:49, 2.30s/it] 50%|█████ | 15016/30000 [9:36:34<9:29:39, 2.28s/it] 50%|█████ | 15017/30000 [9:36:36<9:29:08, 2.28s/it] 50%|█████ | 15018/30000 [9:36:39<9:27:56, 2.27s/it] 50%|█████ | 15019/30000 [9:36:41<9:26:26, 2.27s/it] 50%|█████ | 15020/30000 [9:36:43<9:24:28, 2.26s/it] 50%|█████ | 15020/30000 [9:36:43<9:24:28, 2.26s/it] 50%|█████ | 15021/30000 [9:36:45<9:28:34, 2.28s/it] 50%|█████ | 15022/30000 [9:36:48<9:27:56, 2.28s/it] 50%|█████ | 15023/30000 [9:36:50<9:29:52, 2.28s/it] 50%|█████ | 15024/30000 [9:36:52<9:30:36, 2.29s/it] 50%|█████ | 15025/30000 [9:36:55<9:29:25, 2.28s/it] 50%|█████ | 15026/30000 [9:36:57<9:27:20, 2.27s/it] 50%|█████ | 15027/30000 [9:36:59<9:31:07, 2.29s/it] 50%|█████ | 15028/30000 [9:37:01<9:30:57, 2.29s/it] 50%|█████ | 15029/30000 [9:37:04<9:26:03, 2.27s/it] 50%|█████ | 15030/30000 [9:37:06<9:28:23, 2.28s/it] 50%|█████ | 15030/30000 [9:37:06<9:28:23, 2.28s/it] 50%|█████ | 15031/30000 [9:37:08<9:29:46, 2.28s/it] 50%|█████ | 15032/30000 [9:37:11<9:28:52, 2.28s/it] 50%|█████ | 15033/30000 [9:37:13<9:34:07, 2.30s/it] 50%|█████ | 15034/30000 [9:37:15<9:32:58, 2.30s/it] 50%|█████ | 15035/30000 [9:37:17<9:31:07, 2.29s/it] 50%|█████ | 15036/30000 [9:37:20<9:32:57, 2.30s/it] 50%|█████ | 15037/30000 [9:37:22<9:34:12, 2.30s/it] 50%|█████ | 15038/30000 [9:37:24<9:27:59, 2.28s/it] 50%|█████ | 15039/30000 [9:37:27<9:31:33, 2.29s/it] 50%|█████ | 15040/30000 [9:37:29<9:33:25, 2.30s/it] 50%|█████ | 15040/30000 [9:37:29<9:33:25, 2.30s/it] 50%|█████ | 15041/30000 [9:37:31<9:38:15, 2.32s/it] 50%|█████ | 15042/30000 [9:37:34<9:33:48, 2.30s/it] 50%|█████ | 15043/30000 [9:37:36<9:39:22, 2.32s/it] 50%|█████ | 15044/30000 [9:37:38<9:41:08, 2.33s/it] 50%|█████ | 15045/30000 [9:37:41<9:36:19, 2.31s/it] 50%|█████ | 15046/30000 [9:37:43<9:40:26, 2.33s/it] 50%|█████ | 15047/30000 [9:37:45<9:36:17, 2.31s/it] 50%|█████ | 15048/30000 [9:37:47<9:33:30, 2.30s/it] 50%|█████ | 15049/30000 [9:37:50<9:42:08, 2.34s/it] 50%|█████ | 15050/30000 [9:37:52<9:38:11, 2.32s/it] 50%|█████ | 15050/30000 [9:37:52<9:38:11, 2.32s/it] 50%|█████ | 15051/30000 [9:37:55<9:38:17, 2.32s/it] 50%|█████ | 15052/30000 [9:37:57<9:34:06, 2.30s/it] 50%|█████ | 15053/30000 [9:37:59<9:38:03, 2.32s/it] 50%|█████ | 15054/30000 [9:38:01<9:34:48, 2.31s/it] 50%|█████ | 15055/30000 [9:38:04<9:34:17, 2.31s/it] 50%|█████ | 15056/30000 [9:38:06<9:34:56, 2.31s/it] 50%|█████ | 15057/30000 [9:38:08<9:34:04, 2.31s/it] 50%|█████ | 15058/30000 [9:38:11<9:30:47, 2.29s/it] 50%|█████ | 15059/30000 [9:38:13<9:33:23, 2.30s/it] 50%|█████ | 15060/30000 [9:38:15<9:30:26, 2.29s/it] 50%|█████ | 15060/30000 [9:38:15<9:30:26, 2.29s/it] 50%|█████ | 15061/30000 [9:38:17<9:28:25, 2.28s/it] 50%|█████ | 15062/30000 [9:38:20<9:26:44, 2.28s/it] 50%|█████ | 15063/30000 [9:38:22<9:22:52, 2.26s/it] 50%|█████ | 15064/30000 [9:38:24<9:18:17, 2.24s/it] 50%|█████ | 15065/30000 [9:38:27<9:28:00, 2.28s/it] 50%|█████ | 15066/30000 [9:38:29<9:30:29, 2.29s/it] 50%|█████ | 15067/30000 [9:38:31<9:29:23, 2.29s/it] 50%|█████ | 15068/30000 [9:38:33<9:27:11, 2.28s/it] 50%|█████ | 15069/30000 [9:38:36<9:29:56, 2.29s/it] 50%|█████ | 15070/30000 [9:38:38<9:27:41, 2.28s/it] 50%|█████ | 15070/30000 [9:38:38<9:27:41, 2.28s/it] 50%|█████ | 15071/30000 [9:38:40<9:33:48, 2.31s/it] 50%|█████ | 15072/30000 [9:38:43<9:32:11, 2.30s/it] 50%|█████ | 15073/30000 [9:38:45<9:34:07, 2.31s/it] 50%|█████ | 15074/30000 [9:38:47<9:29:54, 2.29s/it] 50%|█████ | 15075/30000 [9:38:49<9:26:46, 2.28s/it] 50%|█████ | 15076/30000 [9:38:52<9:32:50, 2.30s/it] 50%|█████ | 15077/30000 [9:38:54<9:35:30, 2.31s/it] 50%|█████ | 15078/30000 [9:38:56<9:29:45, 2.29s/it] 50%|█████ | 15079/30000 [9:38:59<9:26:11, 2.28s/it] 50%|█████ | 15080/30000 [9:39:01<9:29:21, 2.29s/it] 50%|█████ | 15080/30000 [9:39:01<9:29:21, 2.29s/it] 50%|█████ | 15081/30000 [9:39:03<9:29:36, 2.29s/it] 50%|█████ | 15082/30000 [9:39:06<9:30:01, 2.29s/it] 50%|█████ | 15083/30000 [9:39:08<9:38:03, 2.33s/it] 50%|█████ | 15084/30000 [9:39:10<9:38:37, 2.33s/it] 50%|█████ | 15085/30000 [9:39:12<9:32:20, 2.30s/it] 50%|█████ | 15086/30000 [9:39:15<9:37:19, 2.32s/it] 50%|█████ | 15087/30000 [9:39:17<9:36:38, 2.32s/it] 50%|█████ | 15088/30000 [9:39:19<9:20:08, 2.25s/it] 50%|█████ | 15089/30000 [9:39:22<9:28:18, 2.29s/it] 50%|█████ | 15090/30000 [9:39:24<9:28:59, 2.29s/it] 50%|█████ | 15090/30000 [9:39:24<9:28:59, 2.29s/it] 50%|█████ | 15091/30000 [9:39:26<9:29:17, 2.29s/it] 50%|█████ | 15092/30000 [9:39:28<9:25:01, 2.27s/it] 50%|█████ | 15093/30000 [9:39:31<9:30:49, 2.30s/it] 50%|█████ | 15094/30000 [9:39:33<9:29:10, 2.29s/it] 50%|█████ | 15095/30000 [9:39:35<9:26:40, 2.28s/it] 50%|█████ | 15096/30000 [9:39:38<9:31:46, 2.30s/it] 50%|█████ | 15097/30000 [9:39:40<9:30:42, 2.30s/it] 50%|█████ | 15098/30000 [9:39:42<9:31:27, 2.30s/it] 50%|█████ | 15099/30000 [9:39:45<9:29:31, 2.29s/it] 50%|█████ | 15100/30000 [9:39:47<9:33:31, 2.31s/it] 50%|█████ | 15100/30000 [9:39:47<9:33:31, 2.31s/it] 50%|█████ | 15101/30000 [9:39:49<9:32:27, 2.31s/it] 50%|█████ | 15102/30000 [9:39:51<9:25:13, 2.28s/it] 50%|█████ | 15103/30000 [9:39:54<9:29:42, 2.29s/it] 50%|█████ | 15104/30000 [9:39:56<9:29:45, 2.29s/it] 50%|█████ | 15105/30000 [9:39:58<9:27:24, 2.29s/it] 50%|█████ | 15106/30000 [9:40:01<9:21:59, 2.26s/it] 50%|█████ | 15107/30000 [9:40:03<9:15:21, 2.24s/it] 50%|█████ | 15108/30000 [9:40:05<9:19:23, 2.25s/it] 50%|█████ | 15109/30000 [9:40:07<9:18:46, 2.25s/it] 50%|█████ | 15110/30000 [9:40:09<9:16:29, 2.24s/it] 50%|█████ | 15110/30000 [9:40:10<9:16:29, 2.24s/it] 50%|█████ | 15111/30000 [9:40:12<9:18:29, 2.25s/it] 50%|█████ | 15112/30000 [9:40:14<9:22:02, 2.27s/it] 50%|█████ | 15113/30000 [9:40:16<9:28:21, 2.29s/it] 50%|█████ | 15114/30000 [9:40:19<9:28:49, 2.29s/it] 50%|█████ | 15115/30000 [9:40:21<9:18:12, 2.25s/it] 50%|█████ | 15116/30000 [9:40:23<9:19:47, 2.26s/it] 50%|█████ | 15117/30000 [9:40:25<9:17:57, 2.25s/it] 50%|█████ | 15118/30000 [9:40:28<9:19:23, 2.26s/it] 50%|█████ | 15119/30000 [9:40:30<9:23:47, 2.27s/it] 50%|█████ | 15120/30000 [9:40:32<9:24:02, 2.27s/it] 50%|█████ | 15120/30000 [9:40:32<9:24:02, 2.27s/it] 50%|█████ | 15121/30000 [9:40:34<9:26:26, 2.28s/it] 50%|█████ | 15122/30000 [9:40:37<9:29:39, 2.30s/it] 50%|█████ | 15123/30000 [9:40:39<9:28:37, 2.29s/it] 50%|█████ | 15124/30000 [9:40:41<9:24:27, 2.28s/it] 50%|█████ | 15125/30000 [9:40:44<9:24:12, 2.28s/it] 50%|█████ | 15126/30000 [9:40:46<9:26:17, 2.28s/it] 50%|█████ | 15127/30000 [9:40:48<9:24:54, 2.28s/it] 50%|█████ | 15128/30000 [9:40:51<9:27:37, 2.29s/it] 50%|█████ | 15129/30000 [9:40:53<9:29:09, 2.30s/it] 50%|█████ | 15130/30000 [9:40:55<9:28:22, 2.29s/it] 50%|█████ | 15130/30000 [9:40:55<9:28:22, 2.29s/it] 50%|█████ | 15131/30000 [9:40:57<9:31:44, 2.31s/it] 50%|█████ | 15132/30000 [9:41:00<9:33:54, 2.32s/it] 50%|█████ | 15133/30000 [9:41:02<9:23:20, 2.27s/it] 50%|█████ | 15134/30000 [9:41:04<9:22:11, 2.27s/it] 50%|█████ | 15135/30000 [9:41:07<9:30:26, 2.30s/it] 50%|█████ | 15136/30000 [9:41:09<9:31:15, 2.31s/it] 50%|█████ | 15137/30000 [9:41:11<9:28:23, 2.29s/it] 50%|█████ | 15138/30000 [9:41:14<9:41:27, 2.35s/it] 50%|█████ | 15139/30000 [9:41:16<9:38:32, 2.34s/it] 50%|█████ | 15140/30000 [9:41:18<9:31:34, 2.31s/it] 50%|█████ | 15140/30000 [9:41:18<9:31:34, 2.31s/it] 50%|█████ | 15141/30000 [9:41:20<9:30:49, 2.30s/it] 50%|█████ | 15142/30000 [9:41:23<9:29:32, 2.30s/it] 50%|█████ | 15143/30000 [9:41:25<9:27:15, 2.29s/it] 50%|█████ | 15144/30000 [9:41:27<9:27:30, 2.29s/it] 50%|█████ | 15145/30000 [9:41:30<9:29:21, 2.30s/it] 50%|█████ | 15146/30000 [9:41:32<9:28:37, 2.30s/it] 50%|█████ | 15147/30000 [9:41:34<9:28:59, 2.30s/it] 50%|█████ | 15148/30000 [9:41:37<9:26:41, 2.29s/it] 50%|█████ | 15149/30000 [9:41:39<9:23:32, 2.28s/it] 50%|█████ | 15150/30000 [9:41:41<9:21:05, 2.27s/it] 50%|█████ | 15150/30000 [9:41:41<9:21:05, 2.27s/it] 51%|█████ | 15151/30000 [9:41:43<9:26:05, 2.29s/it] 51%|█████ | 15152/30000 [9:41:46<9:24:02, 2.28s/it] 51%|█████ | 15153/30000 [9:41:48<9:25:28, 2.29s/it] 51%|█████ | 15154/30000 [9:41:50<9:30:19, 2.30s/it] 51%|█████ | 15155/30000 [9:41:53<9:32:12, 2.31s/it] 51%|█████ | 15156/30000 [9:41:55<9:30:11, 2.30s/it] 51%|█████ | 15157/30000 [9:41:57<9:30:45, 2.31s/it] 51%|█████ | 15158/30000 [9:41:59<9:30:18, 2.31s/it] 51%|█████ | 15159/30000 [9:42:02<9:28:27, 2.30s/it] 51%|█████ | 15160/30000 [9:42:04<9:28:56, 2.30s/it] 51%|█████ | 15160/30000 [9:42:04<9:28:56, 2.30s/it] 51%|█████ | 15161/30000 [9:42:06<9:29:50, 2.30s/it] 51%|█████ | 15162/30000 [9:42:09<9:19:42, 2.26s/it] 51%|█████ | 15163/30000 [9:42:11<9:21:07, 2.27s/it] 51%|█████ | 15164/30000 [9:42:13<9:16:23, 2.25s/it] 51%|█████ | 15165/30000 [9:42:15<9:19:40, 2.26s/it] 51%|█████ | 15166/30000 [9:42:18<9:25:12, 2.29s/it] 51%|█████ | 15167/30000 [9:42:20<9:28:22, 2.30s/it] 51%|█████ | 15168/30000 [9:42:22<9:25:04, 2.29s/it] 51%|█████ | 15169/30000 [9:42:25<9:27:05, 2.29s/it] 51%|█████ | 15170/30000 [9:42:27<9:26:46, 2.29s/it] 51%|█████ | 15170/30000 [9:42:27<9:26:46, 2.29s/it] 51%|█████ | 15171/30000 [9:42:29<9:27:14, 2.30s/it] 51%|█████ | 15172/30000 [9:42:31<9:24:44, 2.29s/it] 51%|█████ | 15173/30000 [9:42:34<9:25:56, 2.29s/it] 51%|█████ | 15174/30000 [9:42:36<9:21:47, 2.27s/it] 51%|█████ | 15175/30000 [9:42:38<9:20:08, 2.27s/it] 51%|█████ | 15176/30000 [9:42:41<9:27:24, 2.30s/it] 51%|█████ | 15177/30000 [9:42:43<9:22:28, 2.28s/it] 51%|█████ | 15178/30000 [9:42:45<9:16:08, 2.25s/it] 51%|█████ | 15179/30000 [9:42:47<9:20:35, 2.27s/it] 51%|█████ | 15180/30000 [9:42:50<9:19:36, 2.27s/it] 51%|█████ | 15180/30000 [9:42:50<9:19:36, 2.27s/it] 51%|█████ | 15181/30000 [9:42:52<9:18:44, 2.26s/it] 51%|█████ | 15182/30000 [9:42:54<9:24:52, 2.29s/it] 51%|█████ | 15183/30000 [9:42:57<9:27:54, 2.30s/it] 51%|█████ | 15184/30000 [9:42:59<9:24:13, 2.28s/it] 51%|█████ | 15185/30000 [9:43:01<9:27:34, 2.30s/it] 51%|█████ | 15186/30000 [9:43:03<9:25:05, 2.29s/it] 51%|█████ | 15187/30000 [9:43:06<9:36:36, 2.34s/it] 51%|█████ | 15188/30000 [9:43:08<9:35:55, 2.33s/it] 51%|█████ | 15189/30000 [9:43:11<9:44:22, 2.37s/it] 51%|█████ | 15190/30000 [9:43:13<9:36:46, 2.34s/it] 51%|█████ | 15190/30000 [9:43:13<9:36:46, 2.34s/it] 51%|█████ | 15191/30000 [9:43:15<9:33:34, 2.32s/it] 51%|█████ | 15192/30000 [9:43:17<9:27:59, 2.30s/it] 51%|█████ | 15193/30000 [9:43:20<9:22:25, 2.28s/it] 51%|█████ | 15194/30000 [9:43:22<9:20:01, 2.27s/it] 51%|█████ | 15195/30000 [9:43:24<9:27:11, 2.30s/it] 51%|█████ | 15196/30000 [9:43:27<9:29:30, 2.31s/it] 51%|█████ | 15197/30000 [9:43:29<9:31:37, 2.32s/it] 51%|█████ | 15198/30000 [9:43:31<9:35:27, 2.33s/it] 51%|█████ | 15199/30000 [9:43:34<9:36:13, 2.34s/it] 51%|█████ | 15200/30000 [9:43:36<9:34:40, 2.33s/it] 51%|█████ | 15200/30000 [9:43:36<9:34:40, 2.33s/it] 51%|█████ | 15201/30000 [9:43:38<9:31:49, 2.32s/it] 51%|█████ | 15202/30000 [9:43:41<9:31:34, 2.32s/it] 51%|█████ | 15203/30000 [9:43:43<9:32:26, 2.32s/it] 51%|█████ | 15204/30000 [9:43:45<9:28:06, 2.30s/it] 51%|█████ | 15205/30000 [9:43:47<9:24:02, 2.29s/it] 51%|█████ | 15206/30000 [9:43:50<9:24:35, 2.29s/it] 51%|█████ | 15207/30000 [9:43:52<9:19:34, 2.27s/it] 51%|█████ | 15208/30000 [9:43:54<9:16:48, 2.26s/it] 51%|█████ | 15209/30000 [9:43:57<9:25:59, 2.30s/it] 51%|█████ | 15210/30000 [9:43:59<9:23:29, 2.29s/it] 51%|█████ | 15210/30000 [9:43:59<9:23:29, 2.29s/it] 51%|█████ | 15211/30000 [9:44:01<9:21:14, 2.28s/it] 51%|█████ | 15212/30000 [9:44:03<9:26:04, 2.30s/it] 51%|█████ | 15213/30000 [9:44:06<9:27:07, 2.30s/it] 51%|█████ | 15214/30000 [9:44:08<9:24:41, 2.29s/it] 51%|█████ | 15215/30000 [9:44:10<9:21:17, 2.28s/it] 51%|█████ | 15216/30000 [9:44:12<9:14:18, 2.25s/it] 51%|█████ | 15217/30000 [9:44:15<9:15:30, 2.25s/it] 51%|█████ | 15218/30000 [9:44:17<9:23:10, 2.29s/it] 51%|█████ | 15219/30000 [9:44:19<9:24:53, 2.29s/it] 51%|█████ | 15220/30000 [9:44:22<9:24:26, 2.29s/it] 51%|█████ | 15220/30000 [9:44:22<9:24:26, 2.29s/it] 51%|█████ | 15221/30000 [9:44:24<9:22:40, 2.28s/it] 51%|█████ | 15222/30000 [9:44:26<9:25:45, 2.30s/it] 51%|█████ | 15223/30000 [9:44:29<9:31:58, 2.32s/it] 51%|█████ | 15224/30000 [9:44:31<9:25:53, 2.30s/it] 51%|█████ | 15225/30000 [9:44:33<9:23:24, 2.29s/it] 51%|█████ | 15226/30000 [9:44:36<9:40:50, 2.36s/it] 51%|█████ | 15227/30000 [9:44:38<9:33:58, 2.33s/it] 51%|█████ | 15228/30000 [9:44:40<9:29:31, 2.31s/it] 51%|█████ | 15229/30000 [9:44:42<9:27:19, 2.30s/it] 51%|█████ | 15230/30000 [9:44:45<9:27:18, 2.30s/it] 51%|█████ | 15230/30000 [9:44:45<9:27:18, 2.30s/it] 51%|█████ | 15231/30000 [9:44:47<9:24:49, 2.29s/it] 51%|█████ | 15232/30000 [9:44:49<9:29:42, 2.31s/it] 51%|█████ | 15233/30000 [9:44:52<9:26:34, 2.30s/it] 51%|█████ | 15234/30000 [9:44:54<9:23:54, 2.29s/it] 51%|█████ | 15235/30000 [9:44:56<9:22:36, 2.29s/it] 51%|█████ | 15236/30000 [9:44:59<9:28:18, 2.31s/it] 51%|█████ | 15237/30000 [9:45:01<9:27:29, 2.31s/it] 51%|█████ | 15238/30000 [9:45:03<9:31:28, 2.32s/it] 51%|█████ | 15239/30000 [9:45:06<9:32:50, 2.33s/it] 51%|█████ | 15240/30000 [9:45:08<9:30:19, 2.32s/it] 51%|█████ | 15240/30000 [9:45:08<9:30:19, 2.32s/it] 51%|█████ | 15241/30000 [9:45:10<9:33:14, 2.33s/it] 51%|█████ | 15242/30000 [9:45:13<9:31:16, 2.32s/it] 51%|█████ | 15243/30000 [9:45:15<9:28:11, 2.31s/it] 51%|█████ | 15244/30000 [9:45:17<9:30:16, 2.32s/it] 51%|█████ | 15245/30000 [9:45:19<9:34:18, 2.34s/it] 51%|█████ | 15246/30000 [9:45:22<9:32:58, 2.33s/it] 51%|█████ | 15247/30000 [9:45:24<9:27:08, 2.31s/it] 51%|█████ | 15248/30000 [9:45:26<9:30:06, 2.32s/it] 51%|█████ | 15249/30000 [9:45:29<9:29:02, 2.31s/it] 51%|█████ | 15250/30000 [9:45:31<9:23:01, 2.29s/it] 51%|█████ | 15250/30000 [9:45:31<9:23:01, 2.29s/it] 51%|█████ | 15251/30000 [9:45:33<9:22:06, 2.29s/it] 51%|█████ | 15252/30000 [9:45:36<9:21:24, 2.28s/it] 51%|█████ | 15253/30000 [9:45:38<9:17:32, 2.27s/it] 51%|█████ | 15254/30000 [9:45:40<9:20:11, 2.28s/it] 51%|█████ | 15255/30000 [9:45:42<9:28:18, 2.31s/it] 51%|█████ | 15256/30000 [9:45:45<9:28:46, 2.31s/it] 51%|█████ | 15257/30000 [9:45:47<9:26:00, 2.30s/it] 51%|█████ | 15258/30000 [9:45:49<9:23:54, 2.30s/it] 51%|█████ | 15259/30000 [9:45:52<9:28:35, 2.31s/it] 51%|█████ | 15260/30000 [9:45:54<9:26:03, 2.30s/it] 51%|█████ | 15260/30000 [9:45:54<9:26:03, 2.30s/it] 51%|█████ | 15261/30000 [9:45:56<9:21:49, 2.29s/it] 51%|█████ | 15262/30000 [9:45:58<9:22:59, 2.29s/it] 51%|█████ | 15263/30000 [9:46:01<9:24:16, 2.30s/it] 51%|█████ | 15264/30000 [9:46:03<9:22:30, 2.29s/it] 51%|█████ | 15265/30000 [9:46:05<9:22:56, 2.29s/it] 51%|█████ | 15266/30000 [9:46:08<9:25:04, 2.30s/it] 51%|█████ | 15267/30000 [9:46:10<9:26:20, 2.31s/it] 51%|█████ | 15268/30000 [9:46:12<9:21:22, 2.29s/it] 51%|█████ | 15269/30000 [9:46:15<9:26:09, 2.31s/it] 51%|█████ | 15270/30000 [9:46:17<9:26:01, 2.31s/it] 51%|█████ | 15270/30000 [9:46:17<9:26:01, 2.31s/it] 51%|█████ | 15271/30000 [9:46:19<9:20:59, 2.29s/it] 51%|█████ | 15272/30000 [9:46:21<9:18:47, 2.28s/it] 51%|█████ | 15273/30000 [9:46:24<9:23:13, 2.29s/it] 51%|█████ | 15274/30000 [9:46:26<9:21:03, 2.29s/it] 51%|█████ | 15275/30000 [9:46:28<9:26:53, 2.31s/it] 51%|█████ | 15276/30000 [9:46:31<9:25:41, 2.31s/it] 51%|█████ | 15277/30000 [9:46:33<9:25:26, 2.30s/it] 51%|█████ | 15278/30000 [9:46:35<9:20:29, 2.28s/it] 51%|█████ | 15279/30000 [9:46:37<9:16:28, 2.27s/it] 51%|█████ | 15280/30000 [9:46:40<9:20:22, 2.28s/it] 51%|█████ | 15280/30000 [9:46:40<9:20:22, 2.28s/it] 51%|█████ | 15281/30000 [9:46:42<9:24:16, 2.30s/it] 51%|█████ | 15282/30000 [9:46:44<9:24:10, 2.30s/it] 51%|█████ | 15283/30000 [9:46:47<9:23:23, 2.30s/it] 51%|█████ | 15284/30000 [9:46:49<9:28:42, 2.32s/it] 51%|█████ | 15285/30000 [9:46:51<9:27:15, 2.31s/it] 51%|█████ | 15286/30000 [9:46:54<9:24:27, 2.30s/it] 51%|█████ | 15287/30000 [9:46:56<9:20:13, 2.28s/it] 51%|█████ | 15288/30000 [9:46:58<9:25:18, 2.31s/it] 51%|█████ | 15289/30000 [9:47:01<9:24:25, 2.30s/it] 51%|█████ | 15290/30000 [9:47:03<9:19:51, 2.28s/it] 51%|█████ | 15290/30000 [9:47:03<9:19:51, 2.28s/it] 51%|█████ | 15291/30000 [9:47:05<9:22:53, 2.30s/it] 51%|█████ | 15292/30000 [9:47:08<9:32:37, 2.34s/it] 51%|█████ | 15293/30000 [9:47:10<9:30:46, 2.33s/it] 51%|█████ | 15294/30000 [9:47:12<9:25:12, 2.31s/it] 51%|█████ | 15295/30000 [9:47:14<9:23:42, 2.30s/it] 51%|█████ | 15296/30000 [9:47:17<9:21:08, 2.29s/it] 51%|█████ | 15297/30000 [9:47:19<9:19:59, 2.29s/it] 51%|█████ | 15298/30000 [9:47:21<9:16:23, 2.27s/it] 51%|█████ | 15299/30000 [9:47:24<9:23:50, 2.30s/it] 51%|█████ | 15300/30000 [9:47:26<9:25:05, 2.31s/it] 51%|█████ | 15300/30000 [9:47:26<9:25:05, 2.31s/it] 51%|█████ | 15301/30000 [9:47:28<9:26:20, 2.31s/it] 51%|█████ | 15302/30000 [9:47:30<9:24:11, 2.30s/it] 51%|█████ | 15303/30000 [9:47:33<9:26:29, 2.31s/it] 51%|█████ | 15304/30000 [9:47:35<9:29:01, 2.32s/it] 51%|█████ | 15305/30000 [9:47:37<9:26:08, 2.31s/it] 51%|█████ | 15306/30000 [9:47:40<9:23:27, 2.30s/it] 51%|█████ | 15307/30000 [9:47:42<9:27:25, 2.32s/it] 51%|█████ | 15308/30000 [9:47:44<9:25:39, 2.31s/it] 51%|█████ | 15309/30000 [9:47:47<9:24:01, 2.30s/it] 51%|█████ | 15310/30000 [9:47:49<9:20:27, 2.29s/it] 51%|█████ | 15310/30000 [9:47:49<9:20:27, 2.29s/it] 51%|█████ | 15311/30000 [9:47:51<9:25:25, 2.31s/it] 51%|█████ | 15312/30000 [9:47:54<9:26:32, 2.31s/it] 51%|█████ | 15313/30000 [9:47:56<9:25:08, 2.31s/it] 51%|█████ | 15314/30000 [9:47:58<9:21:14, 2.29s/it] 51%|█████ | 15315/30000 [9:48:00<9:26:21, 2.31s/it] 51%|█████ | 15316/30000 [9:48:03<9:23:18, 2.30s/it] 51%|█████ | 15317/30000 [9:48:05<9:02:59, 2.22s/it] 51%|█████ | 15318/30000 [9:48:07<9:08:55, 2.24s/it] 51%|█████ | 15319/30000 [9:48:09<9:08:01, 2.24s/it] 51%|█████ | 15320/30000 [9:48:12<9:10:34, 2.25s/it] +{'loss': 0.015, 'grad_norm': 0.07518859952688217, 'learning_rate': 1.6372112646081907e-05, 'flow_matching_loss': 0.014973573014140129, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.34} +{'loss': 0.0156, 'grad_norm': 0.06764617562294006, 'learning_rate': 1.6355646440185464e-05, 'flow_matching_loss': 0.015584727562963963, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.34} +{'loss': 0.0151, 'grad_norm': 0.0923362523317337, 'learning_rate': 1.6339178587050764e-05, 'flow_matching_loss': 0.015131119545549155, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.35} +{'loss': 0.0149, 'grad_norm': 0.09131257236003876, 'learning_rate': 1.632270910668781e-05, 'flow_matching_loss': 0.014864864107221366, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.35} +{'loss': 0.0151, 'grad_norm': 0.07823561877012253, 'learning_rate': 1.6306238019108562e-05, 'flow_matching_loss': 0.015050220862030983, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.35} +{'loss': 0.0165, 'grad_norm': 0.10205482691526413, 'learning_rate': 1.6289765344326952e-05, 'flow_matching_loss': 0.01654695477336645, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.36} +{'loss': 0.0158, 'grad_norm': 0.06716150045394897, 'learning_rate': 1.6273291102358833e-05, 'flow_matching_loss': 0.015807953476905823, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.36} +{'loss': 0.0142, 'grad_norm': 0.08109015226364136, 'learning_rate': 1.6256815313221963e-05, 'flow_matching_loss': 0.01418626019731164, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.36} +{'loss': 0.0139, 'grad_norm': 0.06968723237514496, 'learning_rate': 1.6240337996935972e-05, 'flow_matching_loss': 0.013870825339108706, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.36} +{'loss': 0.015, 'grad_norm': 0.08531917631626129, 'learning_rate': 1.6223859173522367e-05, 'flow_matching_loss': 0.015030915476381778, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.37} +{'loss': 0.0152, 'grad_norm': 0.06523136794567108, 'learning_rate': 1.6207378863004455e-05, 'flow_matching_loss': 0.015157500188797713, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.37} +{'loss': 0.0137, 'grad_norm': 0.07243933528661728, 'learning_rate': 1.6190897085407382e-05, 'flow_matching_loss': 0.013722294755280019, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.37} +{'loss': 0.0163, 'grad_norm': 0.0734253078699112, 'learning_rate': 1.617441386075805e-05, 'flow_matching_loss': 0.01632539490237832, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.38} +{'loss': 0.016, 'grad_norm': 0.10683631896972656, 'learning_rate': 1.6157929209085142e-05, 'flow_matching_loss': 0.015991386491805315, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.38} +{'loss': 0.014, 'grad_norm': 0.06084362417459488, 'learning_rate': 1.614144315041906e-05, 'flow_matching_loss': 0.014014955004677176, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.38} +{'loss': 0.017, 'grad_norm': 0.08103403449058533, 'learning_rate': 1.612495570479192e-05, 'flow_matching_loss': 0.017047799471765757, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.38} +{'loss': 0.0161, 'grad_norm': 0.08495491743087769, 'learning_rate': 1.6108466892237517e-05, 'flow_matching_loss': 0.016107522509992124, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.39} +{'loss': 0.0141, 'grad_norm': 0.07165364176034927, 'learning_rate': 1.609197673279132e-05, 'flow_matching_loss': 0.014140655938535928, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.39} +{'loss': 0.014, 'grad_norm': 0.09411190450191498, 'learning_rate': 1.6075485246490428e-05, 'flow_matching_loss': 0.013967789895832538, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.39} +{'loss': 0.014, 'grad_norm': 0.08396762609481812, 'learning_rate': 1.605899245337355e-05, 'flow_matching_loss': 0.014032442681491375, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.4} +{'loss': 0.0148, 'grad_norm': 0.08314624428749084, 'learning_rate': 1.6042498373480983e-05, 'flow_matching_loss': 0.014782024826854468, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.4} +{'loss': 0.0149, 'grad_norm': 0.09240645915269852, 'learning_rate': 1.602600302685459e-05, 'flow_matching_loss': 0.014922874327749015, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.4} +{'loss': 0.0152, 'grad_norm': 0.0910346582531929, 'learning_rate': 1.6009506433537774e-05, 'flow_matching_loss': 0.015184484422206879, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.41} +{'loss': 0.0143, 'grad_norm': 0.09902123361825943, 'learning_rate': 1.599300861357545e-05, 'flow_matching_loss': 0.014332036394625903, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.41} +{'loss': 0.0141, 'grad_norm': 0.07101614028215408, 'learning_rate': 1.5976509587014032e-05, 'flow_matching_loss': 0.01413049902766943, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.41} +{'loss': 0.015, 'grad_norm': 0.04130292311310768, 'learning_rate': 1.5960009373901384e-05, 'flow_matching_loss': 0.015009857714176178, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.41} +{'loss': 0.0149, 'grad_norm': 0.08816580474376678, 'learning_rate': 1.5943507994286823e-05, 'flow_matching_loss': 0.014895289670675993, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.42} +{'loss': 0.0144, 'grad_norm': 0.11421776562929153, 'learning_rate': 1.5927005468221082e-05, 'flow_matching_loss': 0.01442231060937047, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.42} +{'loss': 0.0143, 'grad_norm': 0.08985734730958939, 'learning_rate': 1.5910501815756285e-05, 'flow_matching_loss': 0.014334757812321186, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.42} +{'loss': 0.0178, 'grad_norm': 0.10437354445457458, 'learning_rate': 1.589399705694593e-05, 'flow_matching_loss': 0.017768604215234518, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.43} +{'loss': 0.0135, 'grad_norm': 0.06458356976509094, 'learning_rate': 1.587749121184485e-05, 'flow_matching_loss': 0.013526966050267219, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.43} +{'loss': 0.0143, 'grad_norm': 0.07025489211082458, 'learning_rate': 1.5860984300509206e-05, 'flow_matching_loss': 0.014334492199122906, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.43} +{'loss': 0.013, 'grad_norm': 0.06986387073993683, 'learning_rate': 1.5844476342996453e-05, 'flow_matching_loss': 0.01299317916855216, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.43} +{'loss': 0.0129, 'grad_norm': 0.06992793083190918, 'learning_rate': 1.5827967359365306e-05, 'flow_matching_loss': 0.012946550734341145, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.44} +{'loss': 0.0138, 'grad_norm': 0.09060268849134445, 'learning_rate': 1.5811457369675758e-05, 'flow_matching_loss': 0.013781459257006645, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.44} +{'loss': 0.0146, 'grad_norm': 0.08867060393095016, 'learning_rate': 1.5794946393988982e-05, 'flow_matching_loss': 0.014552890788763762, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.44} +{'loss': 0.0146, 'grad_norm': 0.07358456403017044, 'learning_rate': 1.5778434452367382e-05, 'flow_matching_loss': 0.014612020272761584, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.45} +{'loss': 0.0145, 'grad_norm': 0.07012782990932465, 'learning_rate': 1.576192156487452e-05, 'flow_matching_loss': 0.014483146741986275, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.45} +{'loss': 0.0143, 'grad_norm': 0.07718883454799652, 'learning_rate': 1.574540775157511e-05, 'flow_matching_loss': 0.014289630856364966, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.45} +{'loss': 0.0154, 'grad_norm': 0.09731829166412354, 'learning_rate': 1.5728893032534998e-05, 'flow_matching_loss': 0.015394666884094476, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.45} +{'loss': 0.0128, 'grad_norm': 0.07412710040807724, 'learning_rate': 1.571237742782112e-05, 'flow_matching_loss': 0.012836019974201918, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.46} 51%|█████ | 15320/30000 [9:48:12<9:10:34, 2.25s/it] 51%|█████ | 15321/30000 [9:48:14<9:15:38, 2.27s/it] 51%|█████ | 15322/30000 [9:48:16<9:18:38, 2.28s/it] 51%|█████ | 15323/30000 [9:48:19<9:20:58, 2.29s/it] 51%|█████ | 15324/30000 [9:48:21<9:26:16, 2.32s/it] 51%|█████ | 15325/30000 [9:48:23<9:24:39, 2.31s/it] 51%|█████ | 15326/30000 [9:48:26<9:33:22, 2.34s/it] 51%|█████ | 15327/30000 [9:48:28<9:26:34, 2.32s/it] 51%|█████ | 15328/30000 [9:48:30<9:22:19, 2.30s/it] 51%|█████ | 15329/30000 [9:48:32<9:24:52, 2.31s/it] 51%|█████ | 15330/30000 [9:48:35<9:20:59, 2.29s/it] 51%|█████ | 15330/30000 [9:48:35<9:20:59, 2.29s/it] 51%|█████ | 15331/30000 [9:48:37<9:19:05, 2.29s/it] 51%|█████ | 15332/30000 [9:48:39<9:19:30, 2.29s/it] 51%|█████ | 15333/30000 [9:48:42<9:21:44, 2.30s/it] 51%|█████ | 15334/30000 [9:48:44<9:15:58, 2.27s/it] 51%|█████ | 15335/30000 [9:48:46<9:10:44, 2.25s/it] 51%|█████ | 15336/30000 [9:48:48<9:17:35, 2.28s/it] 51%|█████ | 15337/30000 [9:48:51<9:19:50, 2.29s/it] 51%|█████ | 15338/30000 [9:48:53<9:19:13, 2.29s/it] 51%|█████ | 15339/30000 [9:48:55<9:22:32, 2.30s/it] 51%|█████ | 15340/30000 [9:48:58<9:20:05, 2.29s/it] 51%|█████ | 15340/30000 [9:48:58<9:20:05, 2.29s/it] 51%|█████ | 15341/30000 [9:49:00<9:19:37, 2.29s/it] 51%|█████ | 15342/30000 [9:49:02<9:32:27, 2.34s/it] 51%|█████ | 15343/30000 [9:49:05<9:33:02, 2.35s/it] 51%|█████ | 15344/30000 [9:49:07<9:31:20, 2.34s/it] 51%|█████ | 15345/30000 [9:49:09<9:25:09, 2.31s/it] 51%|█████ | 15346/30000 [9:49:12<9:20:17, 2.29s/it] 51%|█████ | 15347/30000 [9:49:14<9:25:10, 2.31s/it] 51%|█████ | 15348/30000 [9:49:16<9:24:22, 2.31s/it] 51%|█████ | 15349/30000 [9:49:18<9:20:40, 2.30s/it] 51%|█████ | 15350/30000 [9:49:21<9:17:01, 2.28s/it] 51%|█████ | 15350/30000 [9:49:21<9:17:01, 2.28s/it] 51%|█████ | 15351/30000 [9:49:22<8:38:57, 2.13s/it] 51%|█████ | 15352/30000 [9:49:24<7:33:42, 1.86s/it] 51%|█████ | 15353/30000 [9:49:25<7:20:40, 1.81s/it] 51%|█████ | 15354/30000 [9:49:27<7:32:21, 1.85s/it] 51%|█████ | 15355/30000 [9:49:30<8:08:43, 2.00s/it] 51%|█████ | 15356/30000 [9:49:32<8:41:06, 2.14s/it] 51%|█████ | 15357/30000 [9:49:34<8:43:49, 2.15s/it] 51%|█████ | 15358/30000 [9:49:37<9:32:06, 2.34s/it] 51%|█████ | 15359/30000 [9:49:38<8:10:21, 2.01s/it] 51%|█████ | 15360/30000 [9:49:41<8:29:24, 2.09s/it] 51%|█████ | 15360/30000 [9:49:41<8:29:24, 2.09s/it] 51%|█████ | 15361/30000 [9:49:43<8:41:55, 2.14s/it] 51%|█████ | 15362/30000 [9:49:45<8:52:22, 2.18s/it] 51%|█████ | 15363/30000 [9:49:48<9:17:11, 2.28s/it] 51%|█████ | 15364/30000 [9:49:55<15:14:46, 3.75s/it] 51%|█████ | 15365/30000 [9:49:57<13:09:43, 3.24s/it] 51%|█████ | 15366/30000 [9:50:00<13:13:40, 3.25s/it] 51%|█████ | 15367/30000 [9:50:03<12:07:09, 2.98s/it] 51%|█████ | 15368/30000 [9:50:04<9:58:53, 2.46s/it] 51%|█████ | 15369/30000 [9:50:06<9:38:54, 2.37s/it] 51%|█████ | 15370/30000 [9:50:08<9:29:39, 2.34s/it] 51%|█████ | 15370/30000 [9:50:08<9:29:39, 2.34s/it] 51%|█████ | 15371/30000 [9:50:11<9:32:31, 2.35s/it] 51%|█████ | 15372/30000 [9:50:17<14:56:51, 3.68s/it] 51%|█████ | 15373/30000 [9:50:20<13:14:56, 3.26s/it] 51%|█████ | 15374/30000 [9:50:22<12:04:45, 2.97s/it] 51%|█████▏ | 15375/30000 [9:50:24<11:16:40, 2.78s/it] 51%|█████▏ | 15376/30000 [9:50:26<10:05:54, 2.49s/it] 51%|█████▏ | 15377/30000 [9:50:27<8:36:13, 2.12s/it] 51%|█████▏ | 15378/30000 [9:50:29<7:29:06, 1.84s/it] 51%|█████▏ | 15379/30000 [9:50:31<7:49:45, 1.93s/it] 51%|█████▏ | 15380/30000 [9:50:41<17:30:55, 4.31s/it] 51%|█████▏ | 15380/30000 [9:50:41<17:30:55, 4.31s/it] 51%|█████▏ | 15381/30000 [9:50:42<13:44:43, 3.38s/it] 51%|█████▏ | 15382/30000 [9:50:45<13:27:02, 3.31s/it] 51%|█████▏ | 15383/30000 [9:50:47<12:09:46, 3.00s/it] 51%|█████▏ | 15384/30000 [9:50:49<11:15:41, 2.77s/it] 51%|█████▏ | 15385/30000 [9:50:51<10:19:08, 2.54s/it] 51%|█████▏ | 15386/30000 [9:50:53<9:06:13, 2.24s/it] 51%|█████▏ | 15387/30000 [9:50:54<7:52:15, 1.94s/it] 51%|█████▏ | 15388/30000 [9:51:04<18:04:57, 4.46s/it] 51%|█████▏ | 15389/30000 [9:51:06<14:37:25, 3.60s/it] 51%|█████▏ | 15390/30000 [9:51:07<11:41:54, 2.88s/it] 51%|█████▏ | 15390/30000 [9:51:07<11:41:54, 2.88s/it] 51%|█████▏ | 15391/30000 [9:51:10<11:02:50, 2.72s/it] 51%|█████▏ | 15392/30000 [9:51:12<10:25:33, 2.57s/it] 51%|█████▏ | 15393/30000 [9:51:13<9:07:26, 2.25s/it] 51%|█████▏ | 15394/30000 [9:51:16<9:09:09, 2.26s/it] 51%|█████▏ | 15395/30000 [9:51:18<9:11:12, 2.26s/it] 51%|█████▏ | 15396/30000 [9:51:28<18:57:01, 4.67s/it] 51%|█████▏ | 15397/30000 [9:51:30<15:29:43, 3.82s/it] 51%|█████▏ | 15398/30000 [9:51:32<12:41:08, 3.13s/it] 51%|█████▏ | 15399/30000 [9:51:34<11:38:41, 2.87s/it] 51%|█████▏ | 15400/30000 [9:51:36<10:56:31, 2.70s/it] 51%|█████▏ | 15400/30000 [9:51:36<10:56:31, 2.70s/it] 51%|█████▏ | 15401/30000 [9:51:38<10:14:41, 2.53s/it] 51%|█████▏ | 15402/30000 [9:51:41<9:57:40, 2.46s/it] 51%|█████▏ | 15403/30000 [9:51:42<9:02:16, 2.23s/it] 51%|█████▏ | 15404/30000 [9:51:53<18:57:14, 4.67s/it] 51%|█████▏ | 15405/30000 [9:51:54<14:42:38, 3.63s/it] 51%|█████▏ | 15406/30000 [9:51:55<12:19:47, 3.04s/it] 51%|█████▏ | 15407/30000 [9:51:58<11:24:13, 2.81s/it] 51%|█████▏ | 15408/30000 [9:52:00<10:49:13, 2.67s/it] 51%|█████▏ | 15409/30000 [9:52:02<10:21:01, 2.55s/it] 51%|█████▏ | 15410/30000 [9:52:05<10:01:19, 2.47s/it] 51%|█████▏ | 15410/30000 [9:52:05<10:01:19, 2.47s/it] 51%|█████▏ | 15411/30000 [9:52:07<9:43:05, 2.40s/it] 51%|█████▏ | 15412/30000 [9:52:14<16:03:27, 3.96s/it] 51%|█████▏ | 15413/30000 [9:52:17<14:07:17, 3.49s/it] 51%|█████▏ | 15414/30000 [9:52:18<11:37:50, 2.87s/it] 51%|█████▏ | 15415/30000 [9:52:21<10:48:30, 2.67s/it] 51%|█████▏ | 15416/30000 [9:52:23<10:20:00, 2.55s/it] 51%|█████▏ | 15417/30000 [9:52:25<9:38:25, 2.38s/it] 51%|█████▏ | 15418/30000 [9:52:27<9:02:43, 2.23s/it] 51%|█████▏ | 15419/30000 [9:52:29<9:10:08, 2.26s/it] 51%|█████▏ | 15420/30000 [9:52:39<18:09:28, 4.48s/it] 51%|█████▏ | 15420/30000 [9:52:39<18:09:28, 4.48s/it] 51%|█████▏ | 15421/30000 [9:52:41<15:15:50, 3.77s/it] 51%|█████▏ | 15422/30000 [9:52:42<12:34:13, 3.10s/it] 51%|█████▏ | 15423/30000 [9:52:44<10:57:00, 2.70s/it] 51%|█████▏ | 15424/30000 [9:52:46<10:31:10, 2.60s/it] 51%|█████▏ | 15425/30000 [9:52:49<10:10:36, 2.51s/it] 51%|█████▏ | 15426/30000 [9:52:51<9:54:02, 2.45s/it] 51%|█████▏ | 15427/30000 [9:52:53<9:47:23, 2.42s/it] 51%|█████▏ | 15428/30000 [9:53:00<15:21:26, 3.79s/it] 51%|█████▏ | 15429/30000 [9:53:02<12:11:48, 3.01s/it] 51%|█████▏ | 15430/30000 [9:53:03<10:50:58, 2.68s/it] 51%|█████▏ | 15430/30000 [9:53:04<10:50:58, 2.68s/it] 51%|█████▏ | 15431/30000 [9:53:06<10:26:14, 2.58s/it] 51%|█████▏ | 15432/30000 [9:53:07<8:49:37, 2.18s/it] 51%|█████▏ | 15433/30000 [9:53:08<7:43:53, 1.91s/it] 51%|█████▏ | 15434/30000 [9:53:10<7:23:01, 1.82s/it] 51%|█████▏ | 15435/30000 [9:53:12<7:55:53, 1.96s/it] 51%|█████▏ | 15436/30000 [9:53:24<20:11:51, 4.99s/it] 51%|█████▏ | 15437/30000 [9:53:27<16:56:40, 4.19s/it] 51%|█████▏ | 15438/30000 [9:53:29<14:36:17, 3.61s/it] 51%|█████▏ | 15439/30000 [9:53:31<12:59:39, 3.21s/it] 51%|█████▏ | 15440/30000 [9:53:33<10:54:05, 2.70s/it] 51%|█████▏ | 15440/30000 [9:53:33<10:54:05, 2.70s/it] 51%|█████▏ | 15441/30000 [9:53:35<10:03:10, 2.49s/it] 51%|█████▏ | 15442/30000 [9:53:37<10:06:39, 2.50s/it] 51%|█████▏ | 15443/30000 [9:53:40<9:54:44, 2.45s/it] 51%|█████▏ | 15444/30000 [9:53:45<13:01:22, 3.22s/it] 51%|█████▏ | 15445/30000 [9:53:46<10:49:18, 2.68s/it] 51%|█████▏ | 15446/30000 [9:53:48<10:23:05, 2.57s/it] 51%|█████▏ | 15447/30000 [9:53:51<10:11:46, 2.52s/it] 51%|█████▏ | 15448/30000 [9:53:53<10:06:56, 2.50s/it] 51%|█████▏ | 15449/30000 [9:53:55<9:44:39, 2.41s/it] 52%|█████▏ | 15450/30000 [9:53:57<8:38:04, 2.14s/it] 52%|█████▏ | 15450/30000 [9:53:57<8:38:04, 2.14s/it] 52%|█████▏ | 15451/30000 [9:53:59<8:14:48, 2.04s/it] 52%|█████▏ | 15452/30000 [9:54:08<17:11:56, 4.26s/it] 52%|█████▏ | 15453/30000 [9:54:09<13:28:31, 3.33s/it] 52%|█████▏ | 15454/30000 [9:54:11<11:15:44, 2.79s/it] 52%|█████▏ | 15455/30000 [9:54:13<10:34:51, 2.62s/it] 52%|█████▏ | 15456/30000 [9:54:15<10:16:01, 2.54s/it] 52%|█████▏ | 15457/30000 [9:54:18<10:01:34, 2.48s/it] 52%|█████▏ | 15458/30000 [9:54:20<9:47:09, 2.42s/it] 52%|█████▏ | 15459/30000 [9:54:22<9:02:55, 2.24s/it] 52%|█████▏ | 15460/30000 [9:54:29<15:18:39, 3.79s/it] 52%|█████▏ | 15460/30000 [9:54:29<15:18:39, 3.79s/it] 52%|█████▏ | 15461/30000 [9:54:31<13:27:00, 3.33s/it] 52%|█████▏ | 15462/30000 [9:54:34<12:10:15, 3.01s/it] 52%|█████▏ | 15463/30000 [9:54:36<11:16:17, 2.79s/it] 52%|█████▏ | 15464/30000 [9:54:38<10:30:46, 2.60s/it] 52%|█████▏ | 15465/30000 [9:54:41<10:17:31, 2.55s/it] 52%|█████▏ | 15466/30000 [9:54:43<9:36:38, 2.38s/it] 52%|█████▏ | 15467/30000 [9:54:44<8:11:16, 2.03s/it] 52%|█████▏ | 15468/30000 [9:54:50<12:56:46, 3.21s/it] 52%|█████▏ | 15469/30000 [9:54:52<11:48:31, 2.93s/it] 52%|█████▏ | 15470/30000 [9:54:54<11:03:48, 2.74s/it] 52%|█████▏ | 15470/30000 [9:54:54<11:03:48, 2.74s/it] 52%|█████▏ | 15471/30000 [9:54:56<9:59:05, 2.47s/it] 52%|█████▏ | 15472/30000 [9:54:57<8:25:38, 2.09s/it] 52%|█████▏ | 15473/30000 [9:54:59<8:03:22, 2.00s/it] 52%|█████▏ | 15474/30000 [9:55:02<8:31:02, 2.11s/it] 52%|█████▏ | 15475/30000 [9:55:04<8:43:14, 2.16s/it] 52%|█████▏ | 15476/30000 [9:55:15<19:33:51, 4.85s/it] 52%|█████▏ | 15477/30000 [9:55:17<16:14:28, 4.03s/it] 52%|█████▏ | 15478/30000 [9:55:19<14:08:22, 3.51s/it] 52%|█████▏ | 15479/30000 [9:55:21<12:19:33, 3.06s/it] 52%|█████▏ | 15480/30000 [9:55:23<10:04:20, 2.50s/it] 52%|█████▏ | 15480/30000 [9:55:23<10:04:20, 2.50s/it] 52%|█████▏ | 15481/30000 [9:55:24<8:31:32, 2.11s/it] 52%|█████▏ | 15482/30000 [9:55:26<8:49:22, 2.19s/it] 52%|█████▏ | 15483/30000 [9:55:28<8:59:09, 2.23s/it] 52%|█████▏ | 15484/30000 [9:55:36<15:43:40, 3.90s/it] 52%|█████▏ | 15485/30000 [9:55:37<12:27:49, 3.09s/it] 52%|█████▏ | 15486/30000 [9:55:40<11:35:51, 2.88s/it] 52%|█████▏ | 15487/30000 [9:55:42<10:53:49, 2.70s/it] 52%|█████▏ | 15488/30000 [9:55:44<10:25:27, 2.59s/it] 52%|█████▏ | 15489/30000 [9:55:47<9:55:45, 2.46s/it] 52%|█████▏ | 15490/30000 [9:55:49<9:32:07, 2.37s/it] 52%|█████▏ | 15490/30000 [9:55:49<9:32:07, 2.37s/it] 52%|█████▏ | 15491/30000 [9:55:50<8:08:32, 2.02s/it] 52%|█████▏ | 15492/30000 [9:56:00<18:13:14, 4.52s/it] 52%|█████▏ | 15493/30000 [9:56:02<14:14:40, 3.53s/it] 52%|█████▏ | 15494/30000 [9:56:03<11:25:57, 2.84s/it] 52%|█████▏ | 15495/30000 [9:56:04<9:25:19, 2.34s/it] 52%|█████▏ | 15496/30000 [9:56:06<9:23:15, 2.33s/it] 52%|█████▏ | 15497/30000 [9:56:08<9:14:05, 2.29s/it] 52%|█████▏ | 15498/30000 [9:56:11<9:11:21, 2.28s/it] 52%|█████▏ | 15499/30000 [9:56:13<9:17:16, 2.31s/it] 52%|█████▏ | 15500/30000 [9:56:23<18:51:07, 4.68s/it] 52%|█████▏ | 15500/30000 [9:56:23<18:51:07, 4.68s/it] 52%|█████▏ | 15501/30000 [9:56:26<15:59:27, 3.97s/it] 52%|█████▏ | 15502/30000 [9:56:27<12:50:13, 3.19s/it] 52%|█████▏ | 15503/30000 [9:56:28<10:25:32, 2.59s/it] 52%|█████▏ | 15504/30000 [9:56:30<9:34:08, 2.38s/it] 52%|█████▏ | 15505/30000 [9:56:32<9:39:39, 2.40s/it] 52%|█████▏ | 15506/30000 [9:56:35<9:29:21, 2.36s/it] 52%|█████▏ | 15507/30000 [9:56:37<9:25:49, 2.34s/it] 52%|█████▏ | 15508/30000 [9:56:46<17:33:00, 4.36s/it] 52%|█████▏ | 15509/30000 [9:56:48<15:02:54, 3.74s/it] 52%|█████▏ | 15510/30000 [9:56:50<13:02:40, 3.24s/it] 52%|█████▏ | 15510/30000 [9:56:50<13:02:40, 3.24s/it] 52%|█████▏ | 15511/30000 [9:56:52<11:12:30, 2.78s/it] 52%|█████▏ | 15512/30000 [9:56:54<10:17:09, 2.56s/it] 52%|█████▏ | 15513/30000 [9:56:56<9:11:01, 2.28s/it] 52%|█████▏ | 15514/30000 [9:56:58<9:11:16, 2.28s/it] 52%|█████▏ | 15515/30000 [9:57:00<9:10:47, 2.28s/it] 52%|█████▏ | 15516/30000 [9:57:10<17:29:28, 4.35s/it] 52%|█████▏ | 15517/30000 [9:57:12<15:02:19, 3.74s/it] 52%|█████▏ | 15518/30000 [9:57:14<13:17:06, 3.30s/it] 52%|█████▏ | 15519/30000 [9:57:16<12:02:47, 2.99s/it] 52%|█████▏ | 15520/30000 [9:57:19<11:13:59, 2.79s/it] 52%|█████▏ | 15520/30000 [9:57:19<11:13:59, 2.79s/it] 52%|█████▏ | 15521/30000 [9:57:21<10:07:15, 2.52s/it] 52%|█████▏ | 15522/30000 [9:57:22<8:32:41, 2.12s/it] 52%|█████▏ | 15523/30000 [9:57:24<8:43:04, 2.17s/it] 52%|█████▏ | 15524/30000 [9:57:34<17:33:30, 4.37s/it] 52%|█████▏ | 15525/30000 [9:57:36<15:00:09, 3.73s/it] 52%|█████▏ | 15526/30000 [9:57:38<13:14:19, 3.29s/it] 52%|█████▏ | 15527/30000 [9:57:41<12:05:26, 3.01s/it] 52%|█████▏ | 15528/30000 [9:57:43<11:12:17, 2.79s/it] 52%|█████▏ | 15529/30000 [9:57:44<9:49:19, 2.44s/it] 52%|█████▏ | 15530/30000 [9:57:47<9:34:23, 2.38s/it] 52%|█████▏ | 15530/30000 [9:57:47<9:34:23, 2.38s/it] 52%|█████▏ | 15531/30000 [9:57:49<9:30:46, 2.37s/it] 52%|█████▏ | 15532/30000 [9:57:54<13:13:43, 3.29s/it] 52%|█████▏ | 15533/30000 [9:57:56<10:41:50, 2.66s/it] 52%|█████▏ | 15534/30000 [9:57:57<8:55:49, 2.22s/it] 52%|█████▏ | 15535/30000 [9:57:59<8:28:36, 2.11s/it] 52%|█████▏ | 15536/30000 [9:58:01<8:43:59, 2.17s/it] 52%|█████▏ | 15537/30000 [9:58:03<8:52:03, 2.21s/it] 52%|█████▏ | 15538/30000 [9:58:06<8:59:54, 2.24s/it] 52%|█████▏ | 15539/30000 [9:58:08<9:07:41, 2.27s/it] 52%|█████▏ | 15540/30000 [9:58:17<17:36:26, 4.38s/it] 52%|█████▏ | 15540/30000 [9:58:17<17:36:26, 4.38s/it] 52%|█████▏ | 15541/30000 [9:58:20<15:05:04, 3.76s/it] 52%|█████▏ | 15542/30000 [9:58:22<13:16:28, 3.31s/it] 52%|█████▏ | 15543/30000 [9:58:24<12:05:05, 3.01s/it] 52%|█████▏ | 15544/30000 [9:58:26<11:11:14, 2.79s/it] 52%|█████▏ | 15545/30000 [9:58:29<10:33:09, 2.63s/it] 52%|█████▏ | 15546/30000 [9:58:31<10:11:26, 2.54s/it] 52%|█████▏ | 15547/30000 [9:58:33<9:52:36, 2.46s/it] 52%|█████▏ | 15548/30000 [9:58:36<9:39:34, 2.41s/it] 52%|█████▏ | 15549/30000 [9:58:38<9:30:45, 2.37s/it] 52%|█████▏ | 15550/30000 [9:58:40<9:24:43, 2.34s/it] 52%|█████▏ | 15550/30000 [9:58:40<9:24:43, 2.34s/it] 52%|█████▏ | 15551/30000 [9:58:42<9:18:59, 2.32s/it] 52%|█████▏ | 15552/30000 [9:58:45<9:13:44, 2.30s/it] 52%|█████▏ | 15553/30000 [9:58:47<9:20:57, 2.33s/it] 52%|█████▏ | 15554/30000 [9:58:49<9:15:45, 2.31s/it] 52%|█████▏ | 15555/30000 [9:58:52<9:08:59, 2.28s/it] 52%|█████▏ | 15556/30000 [9:58:54<9:15:26, 2.31s/it] 52%|█████▏ | 15557/30000 [9:58:56<9:14:14, 2.30s/it] 52%|█████▏ | 15558/30000 [9:58:58<9:10:31, 2.29s/it] 52%|█████▏ | 15559/30000 [9:59:01<9:10:52, 2.29s/it] 52%|█████▏ | 15560/30000 [9:59:03<9:08:16, 2.28s/it] 52%|█████▏ | 15560/30000 [9:59:03<9:08:16, 2.28s/it] 52%|█████▏ | 15561/30000 [9:59:05<9:08:39, 2.28s/it] 52%|█████▏ | 15562/30000 [9:59:08<9:07:53, 2.28s/it] 52%|█████▏ | 15563/30000 [9:59:10<9:15:41, 2.31s/it] 52%|█████▏ | 15564/30000 [9:59:12<9:15:28, 2.31s/it] 52%|█████▏ | 15565/30000 [9:59:14<9:12:45, 2.30s/it] 52%|█████▏ | 15566/30000 [9:59:17<9:11:02, 2.29s/it] 52%|█████▏ | 15567/30000 [9:59:19<9:13:08, 2.30s/it] 52%|█████▏ | 15568/30000 [9:59:21<9:13:23, 2.30s/it] 52%|█████▏ | 15569/30000 [9:59:24<9:11:32, 2.29s/it] 52%|█████▏ | 15570/30000 [9:59:26<9:11:40, 2.29s/it] 52%|█████▏ | 15570/30000 [9:59:26<9:11:40, 2.29s/it] 52%|█████▏ | 15571/30000 [9:59:28<9:18:06, 2.32s/it] 52%|█████▏ | 15572/30000 [9:59:31<9:12:35, 2.30s/it] 52%|█████▏ | 15573/30000 [9:59:33<9:11:51, 2.30s/it] 52%|█████▏ | 15574/30000 [9:59:35<9:14:58, 2.31s/it] 52%|█████▏ | 15575/30000 [9:59:38<9:14:14, 2.31s/it] 52%|█████▏ | 15576/30000 [9:59:40<9:11:35, 2.29s/it] 52%|█████▏ | 15577/30000 [9:59:42<9:12:44, 2.30s/it] 52%|█████▏ | 15578/30000 [9:59:44<9:17:06, 2.32s/it] 52%|█████▏ | 15579/30000 [9:59:47<9:18:07, 2.32s/it] 52%|█████▏ | 15580/30000 [9:59:49<9:22:01, 2.34s/it] 52%|█████▏ | 15580/30000 [9:59:49<9:22:01, 2.34s/it] 52%|█████▏ | 15581/30000 [9:59:51<9:16:00, 2.31s/it] 52%|█████▏ | 15582/30000 [9:59:54<9:17:59, 2.32s/it] 52%|█████▏ | 15583/30000 [9:59:56<9:18:36, 2.32s/it] 52%|█████▏ | 15584/30000 [9:59:58<9:15:17, 2.31s/it] 52%|█████▏ | 15585/30000 [10:00:01<9:13:04, 2.30s/it] 52%|█████▏ | 15586/30000 [10:00:03<9:14:05, 2.31s/it] 52%|█████▏ | 15587/30000 [10:00:05<9:12:27, 2.30s/it] 52%|█████▏ | 15588/30000 [10:00:08<9:09:50, 2.29s/it] 52%|█████▏ | 15589/30000 [10:00:10<9:09:33, 2.29s/it] 52%|█████▏ | 15590/30000 [10:00:12<9:12:22, 2.30s/it] 52%|█████▏ | 15590/30000 [10:00:12<9:12:22, 2.30s/it] 52%|█████▏ | 15591/30000 [10:00:14<9:11:08, 2.29s/it] 52%|█████▏ | 15592/30000 [10:00:17<9:08:18, 2.28s/it] 52%|█████▏ | 15593/30000 [10:00:19<9:11:44, 2.30s/it] 52%|█████▏ | 15594/30000 [10:00:21<9:10:40, 2.29s/it] 52%|█████▏ | 15595/30000 [10:00:24<9:09:14, 2.29s/it] 52%|█████▏ | 15596/30000 [10:00:26<9:12:30, 2.30s/it] 52%|█████▏ | 15597/30000 [10:00:28<9:11:44, 2.30s/it] 52%|█████▏ | 15598/30000 [10:00:30<9:08:56, 2.29s/it] 52%|█████▏ | 15599/30000 [10:00:33<9:04:12, 2.27s/it] 52%|█████▏ | 15600/30000 [10:00:35<9:08:36, 2.29s/it] 52%|█████▏ | 15600/30000 [10:00:35<9:08:36, 2.29s/it] 52%|█████▏ | 15601/30000 [10:00:37<9:06:12, 2.28s/it] 52%|█████▏ | 15602/30000 [10:00:40<9:07:20, 2.28s/it] 52%|█████▏ | 15603/30000 [10:00:42<9:10:17, 2.29s/it] 52%|█████▏ | 15604/30000 [10:00:44<9:07:22, 2.28s/it] 52%|█████▏ | 15605/30000 [10:00:46<9:06:00, 2.28s/it] 52%|█████▏ | 15606/30000 [10:00:49<9:03:33, 2.27s/it] 52%|█████▏ | 15607/30000 [10:00:51<9:06:41, 2.28s/it] 52%|█████▏ | 15608/30000 [10:00:53<9:06:29, 2.28s/it] 52%|█████▏ | 15609/30000 [10:00:55<9:05:48, 2.28s/it] 52%|█████▏ | 15610/30000 [10:00:58<9:07:02, 2.28s/it] 52%|█████▏ | 15610/30000 [10:00:58<9:07:02, 2.28s/it] 52%|█████▏ | 15611/30000 [10:01:00<9:11:58, 2.30s/it] 52%|█████▏ | 15612/30000 [10:01:02<9:09:55, 2.29s/it] 52%|█████▏ | 15613/30000 [10:01:05<9:07:36, 2.28s/it] 52%|█████▏ | 15614/30000 [10:01:07<9:10:17, 2.30s/it] 52%|█████▏ | 15615/30000 [10:01:09<9:10:49, 2.30s/it] 52%|█████▏ | 15616/30000 [10:01:12<9:10:45, 2.30s/it] 52%|█████▏ | 15617/30000 [10:01:14<9:10:20, 2.30s/it] 52%|█████▏ | 15618/30000 [10:01:16<9:14:42, 2.31s/it] 52%|█████▏ | 15619/30000 [10:01:19<9:15:54, 2.32s/it] 52%|█████▏ | 15620/30000 [10:01:21<9:13:46, 2.31s/it] 52%|█████▏ | 15620/30000 [10:01:21<9:13:46, 2.31s/it] 52%|█████▏ | 15621/30000 [10:01:23<9:10:50, 2.30s/it] 52%|█████▏ | 15622/30000 [10:01:25<9:12:32, 2.31s/it] 52%|█████▏ | 15623/30000 [10:01:28<9:11:38, 2.30s/it] 52%|█████▏ | 15624/30000 [10:01:30<9:08:46, 2.29s/it] 52%|█████▏ | 15625/30000 [10:01:32<9:07:42, 2.29s/it] 52%|█████▏ | 15626/30000 [10:01:35<9:10:41, 2.30s/it] 52%|█████▏ | 15627/30000 [10:01:37<9:10:58, 2.30s/it] 52%|█████▏ | 15628/30000 [10:01:39<9:07:15, 2.28s/it] 52%|█████▏ | 15629/30000 [10:01:41<9:06:23, 2.28s/it] 52%|█████▏ | 15630/30000 [10:01:44<9:04:14, 2.27s/it] 52%|█████▏ | 15630/30000 [10:01:44<9:04:14, 2.27s/it] 52%|█████▏ | 15631/30000 [10:01:46<9:03:44, 2.27s/it] 52%|█████▏ | 15632/30000 [10:01:48<9:04:37, 2.27s/it] 52%|█████▏ | 15633/30000 [10:01:51<9:09:16, 2.29s/it] 52%|█████▏ | 15634/30000 [10:01:53<9:08:54, 2.29s/it] 52%|█████▏ | 15635/30000 [10:01:55<9:08:23, 2.29s/it] 52%|█████▏ | 15636/30000 [10:01:57<9:05:05, 2.28s/it] 52%|█████▏ | 15637/30000 [10:02:00<9:08:09, 2.29s/it] 52%|█████▏ | 15638/30000 [10:02:02<9:08:49, 2.29s/it] 52%|█████▏ | 15639/30000 [10:02:04<9:09:06, 2.29s/it] 52%|█████▏ | 15640/30000 [10:02:07<9:09:45, 2.30s/it] 52%|█████▏ | 15640/30000 [10:02:07<9:09:45, 2.30s/it] 52%|█████▏ | 15641/30000 [10:02:09<9:11:21, 2.30s/it] 52%|█████▏ | 15642/30000 [10:02:11<9:09:59, 2.30s/it] 52%|█████▏ | 15643/30000 [10:02:14<9:13:14, 2.31s/it] 52%|█████▏ | 15644/30000 [10:02:16<9:09:52, 2.30s/it] 52%|█████▏ | 15645/30000 [10:02:18<9:09:55, 2.30s/it] 52%|█████▏ | 15646/30000 [10:02:20<9:07:01, 2.29s/it] 52%|█████▏ | 15647/30000 [10:02:23<9:06:27, 2.28s/it] 52%|█████▏ | 15648/30000 [10:02:25<9:08:28, 2.29s/it] 52%|█████▏ | 15649/30000 [10:02:27<9:09:20, 2.30s/it] 52%|█████▏ | 15650/30000 [10:02:30<9:05:52, 2.28s/it] 52%|█████▏ | 15650/30000 [10:02:30<9:05:52, 2.28s/it] 52%|█████▏ | 15651/30000 [10:02:32<9:06:46, 2.29s/it] 52%|█████▏ | 15652/30000 [10:02:34<9:09:38, 2.30s/it] 52%|█████▏ | 15653/30000 [10:02:36<9:10:36, 2.30s/it] 52%|█████▏ | 15654/30000 [10:02:39<9:10:04, 2.30s/it] 52%|█████▏ | 15655/30000 [10:02:41<9:07:39, 2.29s/it] 52%|█████▏ | 15656/30000 [10:02:43<9:09:13, 2.30s/it] 52%|█████▏ | 15657/30000 [10:02:46<9:08:12, 2.29s/it] 52%|█████▏ | 15658/30000 [10:02:48<9:07:53, 2.29s/it] 52%|█████▏ | 15659/30000 [10:02:50<9:11:30, 2.31s/it] 52%|█████▏ | 15660/30000 [10:02:53<9:10:15, 2.30s/it] 52%|█████▏ | 15660/30000 [10:02:53<9:10:15, 2.30s/it] 52%|█████▏ | 15661/30000 [10:02:55<9:07:25, 2.29s/it] 52%|█████▏ | 15662/30000 [10:02:57<9:05:44, 2.28s/it] 52%|█████▏ | 15663/30000 [10:02:59<9:08:28, 2.30s/it] 52%|█████▏ | 15664/30000 [10:03:02<9:07:29, 2.29s/it] 52%|█████▏ | 15665/30000 [10:03:04<9:07:59, 2.29s/it] 52%|█████▏ | 15666/30000 [10:03:06<9:09:44, 2.30s/it] 52%|█████▏ | 15667/30000 [10:03:09<9:13:27, 2.32s/it] 52%|█████▏ | 15668/30000 [10:03:11<9:09:20, 2.30s/it] 52%|█████▏ | 15669/30000 [10:03:13<9:06:07, 2.29s/it] 52%|█████▏ | 15670/30000 [10:03:15<9:08:09, 2.30s/it] 52%|█████▏ | 15670/30000 [10:03:16<9:08:09, 2.30s/it] 52%|█████▏ | 15671/30000 [10:03:18<9:08:31, 2.30s/it] 52%|█████▏ | 15672/30000 [10:03:20<9:08:35, 2.30s/it] 52%|█████▏ | 15673/30000 [10:03:22<9:08:12, 2.30s/it] 52%|█████▏ | 15674/30000 [10:03:25<9:09:56, 2.30s/it] 52%|█████▏ | 15675/30000 [10:03:27<9:08:55, 2.30s/it] 52%|█████▏ | 15676/30000 [10:03:29<9:08:36, 2.30s/it] 52%|█████▏ | 15677/30000 [10:03:32<9:14:05, 2.32s/it] 52%|█████▏ | 15678/30000 [10:03:34<9:09:19, 2.30s/it] 52%|█████▏ | 15679/30000 [10:03:36<9:08:47, 2.30s/it] 52%|█████▏ | 15680/30000 [10:03:38<9:07:54, 2.30s/it] 52%|█████▏ | 15680/30000 [10:03:39<9:07:54, 2.30s/it] 52%|█████▏ | 15681/30000 [10:03:41<9:08:21, 2.30s/it] 52%|█████▏ | 15682/30000 [10:03:43<9:06:00, 2.29s/it] 52%|█████▏ | 15683/30000 [10:03:45<9:05:13, 2.28s/it] 52%|█████▏ | 15684/30000 [10:03:48<9:07:10, 2.29s/it] 52%|█████▏ | 15685/30000 [10:03:50<9:07:34, 2.30s/it] 52%|█████▏ | 15686/30000 [10:03:52<9:05:20, 2.29s/it] 52%|█████▏ | 15687/30000 [10:03:55<9:06:59, 2.29s/it] 52%|█████▏ | 15688/30000 [10:03:57<9:09:34, 2.30s/it] 52%|█████▏ | 15689/30000 [10:03:59<9:09:47, 2.31s/it] 52%|█████▏ | 15690/30000 [10:04:01<9:10:51, 2.31s/it] 52%|█████▏ | 15690/30000 [10:04:01<9:10:51, 2.31s/it] 52%|█████▏ | 15691/30000 [10:04:04<9:18:14, 2.34s/it] 52%|█████▏ | 15692/30000 [10:04:06<9:10:55, 2.31s/it] 52%|█████▏ | 15693/30000 [10:04:08<9:08:02, 2.30s/it] 52%|█████▏ | 15694/30000 [10:04:11<9:08:19, 2.30s/it] 52%|█████▏ | 15695/30000 [10:04:13<9:08:17, 2.30s/it] 52%|█████▏ | 15696/30000 [10:04:15<9:07:10, 2.30s/it] 52%|█████▏ | 15697/30000 [10:04:18<9:09:33, 2.31s/it] 52%|█████▏ | 15698/30000 [10:04:20<9:09:58, 2.31s/it] 52%|█████▏ | 15699/30000 [10:04:22<9:13:12, 2.32s/it] 52%|█████▏ | 15700/30000 [10:04:25<9:08:34, 2.30s/it] 52%|█████▏ | 15700/30000 [10:04:25<9:08:34, 2.30s/it] 52%|█████▏ | 15701/30000 [10:04:27<9:11:43, 2.32s/it] 52%|█████▏ | 15702/30000 [10:04:29<9:10:59, 2.31s/it] 52%|█████▏ | 15703/30000 [10:04:31<9:08:54, 2.30s/it] 52%|█████▏ | 15704/30000 [10:04:34<9:13:26, 2.32s/it] 52%|█████▏ | 15705/30000 [10:04:36<9:12:11, 2.32s/it] 52%|█████▏ | 15706/30000 [10:04:38<9:09:08, 2.31s/it] 52%|█████▏ | 15707/30000 [10:04:41<9:07:40, 2.30s/it] 52%|█████▏ | 15708/30000 [10:04:43<9:08:22, 2.30s/it] 52%|█████▏ | 15709/30000 [10:04:45<9:04:27, 2.29s/it] 52%|█████▏ | 15710/30000 [10:04:48<9:03:17, 2.28s/it] 52%|█████▏ | 15710/30000 [10:04:48<9:03:17, 2.28s/it] 52%|█████▏ | 15711/30000 [10:04:50<9:07:31, 2.30s/it] 52%|█████▏ | 15712/30000 [10:04:52<9:07:49, 2.30s/it] 52%|█████▏ | 15713/30000 [10:04:54<9:07:35, 2.30s/it] 52%|█████▏ | 15714/30000 [10:04:57<9:13:08, 2.32s/it] 52%|█████▏ | 15715/30000 [10:04:59<9:15:47, 2.33s/it] 52%|█████▏ | 15716/30000 [10:05:01<9:08:17, 2.30s/it] 52%|█████▏ | 15717/30000 [10:05:04<9:07:37, 2.30s/it] 52%|█████▏ | 15718/30000 [10:05:06<9:08:05, 2.30s/it] 52%|█████▏ | 15719/30000 [10:05:08<8:55:54, 2.25s/it] 52%|█████▏ | 15720/30000 [10:05:11<9:04:39, 2.29s/it] 52%|█████▏ | 15720/30000 [10:05:11<9:04:39, 2.29s/it] 52%|█████▏ | 15721/30000 [10:05:13<9:05:06, 2.29s/it] 52%|█████▏ | 15722/30000 [10:05:15<9:05:20, 2.29s/it] 52%|█████▏ | 15723/30000 [10:05:17<9:05:12, 2.29s/it] 52%|█████▏ | 15724/30000 [10:05:20<9:08:53, 2.31s/it] 52%|█████▏ | 15725/30000 [10:05:22<9:05:37, 2.29s/it] 52%|█████▏ | 15726/30000 [10:05:24<8:57:55, 2.26s/it] 52%|█████▏ | 15727/30000 [10:05:27<9:04:00, 2.29s/it] 52%|█████▏ | 15728/30000 [10:05:29<9:07:08, 2.30s/it] 52%|█████▏ | 15729/30000 [10:05:31<9:05:06, 2.29s/it] 52%|█████▏ | 15730/30000 [10:05:33<9:04:05, 2.29s/it] +{'loss': 0.0137, 'grad_norm': 0.06988117098808289, 'learning_rate': 1.569586095750149e-05, 'flow_matching_loss': 0.013701215013861656, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.46} +{'loss': 0.0152, 'grad_norm': 0.08507321774959564, 'learning_rate': 1.5679343641645193e-05, 'flow_matching_loss': 0.015245670918375254, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.46} +{'loss': 0.0148, 'grad_norm': 0.0634031891822815, 'learning_rate': 1.5662825500322304e-05, 'flow_matching_loss': 0.014805424027144909, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.47} +{'loss': 0.0165, 'grad_norm': 0.0807950347661972, 'learning_rate': 1.5646306553603932e-05, 'flow_matching_loss': 0.016491972841322422, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.47} +{'loss': 0.014, 'grad_norm': 0.08035554736852646, 'learning_rate': 1.5629786821562154e-05, 'flow_matching_loss': 0.013970221858471632, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.47} +{'loss': 0.0162, 'grad_norm': 0.09006202220916748, 'learning_rate': 1.561326632427001e-05, 'flow_matching_loss': 0.016221597231924532, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.47} +{'loss': 0.014, 'grad_norm': 0.07957533746957779, 'learning_rate': 1.5596745081801447e-05, 'flow_matching_loss': 0.013975668884813785, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.48} +{'loss': 0.0134, 'grad_norm': 0.08115212619304657, 'learning_rate': 1.5580223114231348e-05, 'flow_matching_loss': 0.013437022175639867, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.48} +{'loss': 0.0151, 'grad_norm': 0.09862697124481201, 'learning_rate': 1.5563700441635445e-05, 'flow_matching_loss': 0.015148298721760512, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.48} +{'loss': 0.0146, 'grad_norm': 0.07992416620254517, 'learning_rate': 1.5547177084090358e-05, 'flow_matching_loss': 0.014630807191133499, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.49} +{'loss': 0.013, 'grad_norm': 0.06291518360376358, 'learning_rate': 1.5530653061673514e-05, 'flow_matching_loss': 0.013019058294594288, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.49} +{'loss': 0.0132, 'grad_norm': 0.07245729863643646, 'learning_rate': 1.5514128394463167e-05, 'flow_matching_loss': 0.013219672534614802, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.49} +{'loss': 0.0163, 'grad_norm': 0.09696550667285919, 'learning_rate': 1.549760310253834e-05, 'flow_matching_loss': 0.016330114286392926, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.5} +{'loss': 0.0131, 'grad_norm': 0.08611389994621277, 'learning_rate': 1.5481077205978816e-05, 'flow_matching_loss': 0.013116530794650316, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.5} +{'loss': 0.0141, 'grad_norm': 0.09240510314702988, 'learning_rate': 1.5464550724865135e-05, 'flow_matching_loss': 0.014116436708718538, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.5} +{'loss': 0.0153, 'grad_norm': 0.09991602599620819, 'learning_rate': 1.5448023679278517e-05, 'flow_matching_loss': 0.015336194448173046, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.5} +{'loss': 0.0139, 'grad_norm': 0.09660439193248749, 'learning_rate': 1.543149608930088e-05, 'flow_matching_loss': 0.013883461616933346, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.51} +{'loss': 0.0151, 'grad_norm': 0.07507980614900589, 'learning_rate': 1.5414967975014805e-05, 'flow_matching_loss': 0.0151103631593287, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.51} +{'loss': 0.0151, 'grad_norm': 0.08128144592046738, 'learning_rate': 1.539843935650351e-05, 'flow_matching_loss': 0.015098787751048804, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.51} +{'loss': 0.0143, 'grad_norm': 0.08194848150014877, 'learning_rate': 1.5381910253850834e-05, 'flow_matching_loss': 0.01432015746831894, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.52} +{'loss': 0.0159, 'grad_norm': 0.08704596012830734, 'learning_rate': 1.5365380687141183e-05, 'flow_matching_loss': 0.015895967744290827, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.52} +{'loss': 0.015, 'grad_norm': 0.07523554563522339, 'learning_rate': 1.5348850676459544e-05, 'flow_matching_loss': 0.015021363086998463, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.52} +{'loss': 0.0139, 'grad_norm': 0.06428109109401703, 'learning_rate': 1.5332320241891438e-05, 'flow_matching_loss': 0.013903049007058143, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.52} +{'loss': 0.0161, 'grad_norm': 0.08954824507236481, 'learning_rate': 1.5315789403522908e-05, 'flow_matching_loss': 0.016111268568783997, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.53} +{'loss': 0.0149, 'grad_norm': 0.0713387206196785, 'learning_rate': 1.529925818144047e-05, 'flow_matching_loss': 0.014935821387916803, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.53} +{'loss': 0.0138, 'grad_norm': 0.06443995982408524, 'learning_rate': 1.5282726595731126e-05, 'flow_matching_loss': 0.013843467459082603, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.53} +{'loss': 0.0145, 'grad_norm': 0.10092808306217194, 'learning_rate': 1.5266194666482304e-05, 'flow_matching_loss': 0.014464623294770718, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.54} +{'loss': 0.0156, 'grad_norm': 0.07754936814308167, 'learning_rate': 1.5249662413781866e-05, 'flow_matching_loss': 0.015624248888343573, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.54} +{'loss': 0.0143, 'grad_norm': 0.10332988202571869, 'learning_rate': 1.5233129857718054e-05, 'flow_matching_loss': 0.014295940473675727, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.54} +{'loss': 0.0128, 'grad_norm': 0.06545239686965942, 'learning_rate': 1.521659701837948e-05, 'flow_matching_loss': 0.0127662293612957, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.54} +{'loss': 0.0146, 'grad_norm': 0.1132861077785492, 'learning_rate': 1.5200063915855107e-05, 'flow_matching_loss': 0.014642744418233633, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.55} +{'loss': 0.0147, 'grad_norm': 0.0766669437289238, 'learning_rate': 1.5183530570234212e-05, 'flow_matching_loss': 0.014675755146890878, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.55} +{'loss': 0.0145, 'grad_norm': 0.06509560346603394, 'learning_rate': 1.5166997001606361e-05, 'flow_matching_loss': 0.014484820887446404, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.55} +{'loss': 0.014, 'grad_norm': 0.078965924680233, 'learning_rate': 1.5150463230061417e-05, 'flow_matching_loss': 0.01397568229585886, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.56} +{'loss': 0.0147, 'grad_norm': 0.062429189682006836, 'learning_rate': 1.5133929275689455e-05, 'flow_matching_loss': 0.014735007099807263, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.56} +{'loss': 0.0141, 'grad_norm': 0.08984331786632538, 'learning_rate': 1.5117395158580795e-05, 'flow_matching_loss': 0.014139562658965588, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.56} +{'loss': 0.0145, 'grad_norm': 0.06667698919773102, 'learning_rate': 1.510086089882595e-05, 'flow_matching_loss': 0.014524329639971256, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.57} +{'loss': 0.0148, 'grad_norm': 0.09919349104166031, 'learning_rate': 1.5084326516515599e-05, 'flow_matching_loss': 0.014795872196555138, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.57} +{'loss': 0.0151, 'grad_norm': 0.08123046159744263, 'learning_rate': 1.5067792031740582e-05, 'flow_matching_loss': 0.01512451721355319, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.57} +{'loss': 0.0154, 'grad_norm': 0.1065877228975296, 'learning_rate': 1.505125746459185e-05, 'flow_matching_loss': 0.015394161827862263, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.57} +{'loss': 0.0146, 'grad_norm': 0.08976095914840698, 'learning_rate': 1.5034722835160466e-05, 'flow_matching_loss': 0.014578016474843025, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.58} 52%|█████▏ | 15730/30000 [10:05:33<9:04:05, 2.29s/it] 52%|█████▏ | 15731/30000 [10:05:36<9:06:47, 2.30s/it] 52%|█████▏ | 15732/30000 [10:05:38<9:04:22, 2.29s/it] 52%|█████▏ | 15733/30000 [10:05:40<9:02:27, 2.28s/it] 52%|█████▏ | 15734/30000 [10:05:43<9:06:13, 2.30s/it] 52%|█████▏ | 15735/30000 [10:05:45<9:08:35, 2.31s/it] 52%|█████▏ | 15736/30000 [10:05:47<9:05:06, 2.29s/it] 52%|█████▏ | 15737/30000 [10:05:50<9:04:13, 2.29s/it] 52%|█████▏ | 15738/30000 [10:05:52<9:08:33, 2.31s/it] 52%|█████▏ | 15739/30000 [10:05:54<9:05:19, 2.29s/it] 52%|█████▏ | 15740/30000 [10:05:56<9:03:58, 2.29s/it] 52%|█████▏ | 15740/30000 [10:05:56<9:03:58, 2.29s/it] 52%|█████▏ | 15741/30000 [10:05:59<9:07:40, 2.30s/it] 52%|█████▏ | 15742/30000 [10:06:01<9:07:33, 2.30s/it] 52%|█████▏ | 15743/30000 [10:06:03<8:58:14, 2.27s/it] 52%|█████▏ | 15744/30000 [10:06:05<8:55:54, 2.26s/it] 52%|█████▏ | 15745/30000 [10:06:08<9:00:20, 2.27s/it] 52%|█████▏ | 15746/30000 [10:06:10<8:59:35, 2.27s/it] 52%|█████▏ | 15747/30000 [10:06:12<9:00:44, 2.28s/it] 52%|█████▏ | 15748/30000 [10:06:15<9:05:23, 2.30s/it] 52%|█████▏ | 15749/30000 [10:06:17<8:59:41, 2.27s/it] 52%|█████▎ | 15750/30000 [10:06:19<8:58:17, 2.27s/it] 52%|█████▎ | 15750/30000 [10:06:19<8:58:17, 2.27s/it] 53%|█████▎ | 15751/30000 [10:06:21<9:02:08, 2.28s/it] 53%|█████▎ | 15752/30000 [10:06:24<9:05:00, 2.30s/it] 53%|█████▎ | 15753/30000 [10:06:26<9:04:40, 2.29s/it] 53%|█████▎ | 15754/30000 [10:06:28<9:03:40, 2.29s/it] 53%|█████▎ | 15755/30000 [10:06:31<9:06:49, 2.30s/it] 53%|█████▎ | 15756/30000 [10:06:33<9:05:54, 2.30s/it] 53%|█████▎ | 15757/30000 [10:06:35<9:04:07, 2.29s/it] 53%|█████▎ | 15758/30000 [10:06:38<9:04:00, 2.29s/it] 53%|█████▎ | 15759/30000 [10:06:40<9:01:09, 2.28s/it] 53%|█████▎ | 15760/30000 [10:06:42<9:00:34, 2.28s/it] 53%|█████▎ | 15760/30000 [10:06:42<9:00:34, 2.28s/it] 53%|█████▎ | 15761/30000 [10:06:44<9:00:24, 2.28s/it] 53%|█████▎ | 15762/30000 [10:06:47<9:13:31, 2.33s/it] 53%|█████▎ | 15763/30000 [10:06:49<9:10:25, 2.32s/it] 53%|█████▎ | 15764/30000 [10:06:51<9:06:13, 2.30s/it] 53%|█████▎ | 15765/30000 [10:06:54<9:03:30, 2.29s/it] 53%|█████▎ | 15766/30000 [10:06:56<9:06:58, 2.31s/it] 53%|█████▎ | 15767/30000 [10:06:58<9:04:08, 2.29s/it] 53%|█████▎ | 15768/30000 [10:07:01<9:04:09, 2.29s/it] 53%|█████▎ | 15769/30000 [10:07:03<9:03:32, 2.29s/it] 53%|█████▎ | 15770/30000 [10:07:05<9:02:34, 2.29s/it] 53%|█████▎ | 15770/30000 [10:07:05<9:02:34, 2.29s/it] 53%|█████▎ | 15771/30000 [10:07:07<9:03:08, 2.29s/it] 53%|█████▎ | 15772/30000 [10:07:10<9:01:17, 2.28s/it] 53%|█████▎ | 15773/30000 [10:07:12<9:03:49, 2.29s/it] 53%|█████▎ | 15774/30000 [10:07:14<9:05:06, 2.30s/it] 53%|█████▎ | 15775/30000 [10:07:17<9:03:52, 2.29s/it] 53%|█████▎ | 15776/30000 [10:07:19<9:03:19, 2.29s/it] 53%|█████▎ | 15777/30000 [10:07:21<9:05:40, 2.30s/it] 53%|█████▎ | 15778/30000 [10:07:23<9:05:34, 2.30s/it] 53%|█████▎ | 15779/30000 [10:07:26<9:02:33, 2.29s/it] 53%|█████▎ | 15780/30000 [10:07:28<8:59:25, 2.28s/it] 53%|█████▎ | 15780/30000 [10:07:28<8:59:25, 2.28s/it] 53%|█████▎ | 15781/30000 [10:07:30<9:03:44, 2.29s/it] 53%|█████▎ | 15782/30000 [10:07:33<8:59:38, 2.28s/it] 53%|█████▎ | 15783/30000 [10:07:35<9:01:48, 2.29s/it] 53%|█████▎ | 15784/30000 [10:07:37<9:02:48, 2.29s/it] 53%|█████▎ | 15785/30000 [10:07:39<9:04:23, 2.30s/it] 53%|█████▎ | 15786/30000 [10:07:42<9:00:53, 2.28s/it] 53%|█████▎ | 15787/30000 [10:07:44<9:02:27, 2.29s/it] 53%|█████▎ | 15788/30000 [10:07:46<9:05:32, 2.30s/it] 53%|█████▎ | 15789/30000 [10:07:49<9:03:03, 2.29s/it] 53%|█████▎ | 15790/30000 [10:07:51<8:59:26, 2.28s/it] 53%|█████▎ | 15790/30000 [10:07:51<8:59:26, 2.28s/it] 53%|█████▎ | 15791/30000 [10:07:53<8:58:37, 2.27s/it] 53%|█████▎ | 15792/30000 [10:07:55<9:02:50, 2.29s/it] 53%|█████▎ | 15793/30000 [10:07:58<9:04:45, 2.30s/it] 53%|█████▎ | 15794/30000 [10:08:00<9:04:20, 2.30s/it] 53%|█████▎ | 15795/30000 [10:08:02<9:00:08, 2.28s/it] 53%|█████▎ | 15796/30000 [10:08:05<9:09:52, 2.32s/it] 53%|█████▎ | 15797/30000 [10:08:07<9:06:22, 2.31s/it] 53%|█████▎ | 15798/30000 [10:08:09<9:04:13, 2.30s/it] 53%|█████▎ | 15799/30000 [10:08:12<9:02:53, 2.29s/it] 53%|█████▎ | 15800/30000 [10:08:14<9:06:28, 2.31s/it] 53%|█████▎ | 15800/30000 [10:08:14<9:06:28, 2.31s/it] 53%|█████▎ | 15801/30000 [10:08:16<9:06:20, 2.31s/it] 53%|█████▎ | 15802/30000 [10:08:19<9:06:01, 2.31s/it] 53%|█████▎ | 15803/30000 [10:08:21<9:00:41, 2.29s/it] 53%|█████▎ | 15804/30000 [10:08:23<8:58:34, 2.28s/it] 53%|█████▎ | 15805/30000 [10:08:25<8:55:51, 2.26s/it] 53%|█████▎ | 15806/30000 [10:08:28<8:56:35, 2.27s/it] 53%|█████▎ | 15807/30000 [10:08:30<9:01:04, 2.29s/it] 53%|█████▎ | 15808/30000 [10:08:32<9:02:35, 2.29s/it] 53%|█████▎ | 15809/30000 [10:08:34<8:58:38, 2.28s/it] 53%|█████▎ | 15810/30000 [10:08:37<8:59:45, 2.28s/it] 53%|█████▎ | 15810/30000 [10:08:37<8:59:45, 2.28s/it] 53%|█████▎ | 15811/30000 [10:08:39<9:05:38, 2.31s/it] 53%|█████▎ | 15812/30000 [10:08:41<8:55:39, 2.27s/it] 53%|█████▎ | 15813/30000 [10:08:43<8:49:29, 2.24s/it] 53%|█████▎ | 15814/30000 [10:08:46<8:53:41, 2.26s/it] 53%|█████▎ | 15815/30000 [10:08:48<8:47:56, 2.23s/it] 53%|█████▎ | 15816/30000 [10:08:50<8:49:48, 2.24s/it] 53%|█████▎ | 15817/30000 [10:08:52<8:54:27, 2.26s/it] 53%|█████▎ | 15818/30000 [10:08:55<8:57:31, 2.27s/it] 53%|█████▎ | 15819/30000 [10:08:57<8:59:15, 2.28s/it] 53%|█████▎ | 15820/30000 [10:08:59<9:03:32, 2.30s/it] 53%|█████▎ | 15820/30000 [10:08:59<9:03:32, 2.30s/it] 53%|█████▎ | 15821/30000 [10:09:02<9:03:57, 2.30s/it] 53%|█████▎ | 15822/30000 [10:09:04<9:07:20, 2.32s/it] 53%|█████▎ | 15823/30000 [10:09:06<9:06:07, 2.31s/it] 53%|█████▎ | 15824/30000 [10:09:09<9:07:05, 2.32s/it] 53%|█████▎ | 15825/30000 [10:09:11<9:05:50, 2.31s/it] 53%|█████▎ | 15826/30000 [10:09:13<9:04:34, 2.31s/it] 53%|█████▎ | 15827/30000 [10:09:16<9:07:13, 2.32s/it] 53%|█████▎ | 15828/30000 [10:09:18<9:02:03, 2.29s/it] 53%|█████▎ | 15829/30000 [10:09:20<8:59:33, 2.28s/it] 53%|█████▎ | 15830/30000 [10:09:22<9:02:33, 2.30s/it] 53%|█████▎ | 15830/30000 [10:09:23<9:02:33, 2.30s/it] 53%|█████▎ | 15831/30000 [10:09:25<9:03:39, 2.30s/it] 53%|█████▎ | 15832/30000 [10:09:27<8:59:26, 2.28s/it] 53%|█████▎ | 15833/30000 [10:09:29<9:03:28, 2.30s/it] 53%|█████▎ | 15834/30000 [10:09:32<9:02:07, 2.30s/it] 53%|█████▎ | 15835/30000 [10:09:34<9:00:28, 2.29s/it] 53%|█████▎ | 15836/30000 [10:09:36<8:58:30, 2.28s/it] 53%|█████▎ | 15837/30000 [10:09:39<9:01:17, 2.29s/it] 53%|█████▎ | 15838/30000 [10:09:41<9:00:10, 2.29s/it] 53%|█████▎ | 15839/30000 [10:09:43<9:00:04, 2.29s/it] 53%|█████▎ | 15840/30000 [10:09:45<9:03:31, 2.30s/it] 53%|█████▎ | 15840/30000 [10:09:45<9:03:31, 2.30s/it] 53%|█████▎ | 15841/30000 [10:09:48<9:03:44, 2.30s/it] 53%|█████▎ | 15842/30000 [10:09:50<8:59:53, 2.29s/it] 53%|█████▎ | 15843/30000 [10:09:52<9:05:08, 2.31s/it] 53%|█████▎ | 15844/30000 [10:09:55<9:03:49, 2.30s/it] 53%|█████▎ | 15845/30000 [10:09:57<8:58:56, 2.28s/it] 53%|█████▎ | 15846/30000 [10:09:59<9:04:21, 2.31s/it] 53%|█████▎ | 15847/30000 [10:10:01<8:58:54, 2.28s/it] 53%|█████▎ | 15848/30000 [10:10:04<8:58:26, 2.28s/it] 53%|█████▎ | 15849/30000 [10:10:06<9:03:28, 2.30s/it] 53%|█████▎ | 15850/30000 [10:10:08<9:04:44, 2.31s/it] 53%|█████▎ | 15850/30000 [10:10:08<9:04:44, 2.31s/it] 53%|█████▎ | 15851/30000 [10:10:11<9:01:37, 2.30s/it] 53%|█████▎ | 15852/30000 [10:10:13<8:59:46, 2.29s/it] 53%|█████▎ | 15853/30000 [10:10:15<9:01:24, 2.30s/it] 53%|█████▎ | 15854/30000 [10:10:18<8:58:33, 2.28s/it] 53%|█████▎ | 15855/30000 [10:10:20<8:59:58, 2.29s/it] 53%|█████▎ | 15856/30000 [10:10:22<9:05:41, 2.31s/it] 53%|█████▎ | 15857/30000 [10:10:25<9:05:43, 2.32s/it] 53%|█████▎ | 15858/30000 [10:10:27<9:01:29, 2.30s/it] 53%|█████▎ | 15859/30000 [10:10:29<9:06:30, 2.32s/it] 53%|█████▎ | 15860/30000 [10:10:31<9:04:08, 2.31s/it] 53%|█████▎ | 15860/30000 [10:10:31<9:04:08, 2.31s/it] 53%|█████▎ | 15861/30000 [10:10:34<9:00:12, 2.29s/it] 53%|█████▎ | 15862/30000 [10:10:36<8:59:31, 2.29s/it] 53%|█████▎ | 15863/30000 [10:10:38<8:59:34, 2.29s/it] 53%|█████▎ | 15864/30000 [10:10:41<8:58:03, 2.28s/it] 53%|█████▎ | 15865/30000 [10:10:43<8:56:11, 2.28s/it] 53%|█████▎ | 15866/30000 [10:10:45<8:58:53, 2.29s/it] 53%|█████▎ | 15867/30000 [10:10:47<9:00:30, 2.29s/it] 53%|█████▎ | 15868/30000 [10:10:50<8:59:40, 2.29s/it] 53%|█████▎ | 15869/30000 [10:10:52<9:00:39, 2.30s/it] 53%|█████▎ | 15870/30000 [10:10:54<8:59:15, 2.29s/it] 53%|█████▎ | 15870/30000 [10:10:54<8:59:15, 2.29s/it] 53%|█████▎ | 15871/30000 [10:10:57<8:57:49, 2.28s/it] 53%|█████▎ | 15872/30000 [10:10:59<8:56:21, 2.28s/it] 53%|█████▎ | 15873/30000 [10:11:01<8:56:49, 2.28s/it] 53%|█████▎ | 15874/30000 [10:11:03<8:55:02, 2.27s/it] 53%|█████▎ | 15875/30000 [10:11:06<8:54:16, 2.27s/it] 53%|█████▎ | 15876/30000 [10:11:08<8:56:41, 2.28s/it] 53%|█████▎ | 15877/30000 [10:11:10<8:54:56, 2.27s/it] 53%|█████▎ | 15878/30000 [10:11:12<8:54:40, 2.27s/it] 53%|█████▎ | 15879/30000 [10:11:15<8:58:08, 2.29s/it] 53%|█████▎ | 15880/30000 [10:11:17<8:57:39, 2.28s/it] 53%|█████▎ | 15880/30000 [10:11:17<8:57:39, 2.28s/it] 53%|█████▎ | 15881/30000 [10:11:19<9:04:54, 2.32s/it] 53%|█████▎ | 15882/30000 [10:11:22<8:58:29, 2.29s/it] 53%|█████▎ | 15883/30000 [10:11:24<9:00:48, 2.30s/it] 53%|█████▎ | 15884/30000 [10:11:26<9:02:32, 2.31s/it] 53%|█████▎ | 15885/30000 [10:11:29<8:59:52, 2.29s/it] 53%|█████▎ | 15886/30000 [10:11:31<9:00:35, 2.30s/it] 53%|█████▎ | 15887/30000 [10:11:33<8:59:41, 2.29s/it] 53%|█████▎ | 15888/30000 [10:11:35<8:56:31, 2.28s/it] 53%|█████▎ | 15889/30000 [10:11:38<8:50:09, 2.25s/it] 53%|█████▎ | 15890/30000 [10:11:40<8:58:37, 2.29s/it] 53%|█████▎ | 15890/30000 [10:11:40<8:58:37, 2.29s/it] 53%|█████▎ | 15891/30000 [10:11:42<9:00:12, 2.30s/it] 53%|█████▎ | 15892/30000 [10:11:45<8:55:23, 2.28s/it] 53%|█████▎ | 15893/30000 [10:11:47<9:01:42, 2.30s/it] 53%|█████▎ | 15894/30000 [10:11:49<9:00:06, 2.30s/it] 53%|█████▎ | 15895/30000 [10:11:51<8:53:52, 2.27s/it] 53%|█████▎ | 15896/30000 [10:11:54<8:53:09, 2.27s/it] 53%|█████▎ | 15897/30000 [10:11:56<8:58:16, 2.29s/it] 53%|█████▎ | 15898/30000 [10:11:58<9:01:05, 2.30s/it] 53%|█████▎ | 15899/30000 [10:12:01<9:01:23, 2.30s/it] 53%|█████▎ | 15900/30000 [10:12:03<8:57:54, 2.29s/it] 53%|█████▎ | 15900/30000 [10:12:03<8:57:54, 2.29s/it] 53%|█████▎ | 15901/30000 [10:12:05<9:02:09, 2.31s/it] 53%|█████▎ | 15902/30000 [10:12:07<8:56:23, 2.28s/it] 53%|█████▎ | 15903/30000 [10:12:10<8:52:12, 2.27s/it] 53%|█████▎ | 15904/30000 [10:12:12<9:01:52, 2.31s/it] 53%|█████▎ | 15905/30000 [10:12:14<9:04:53, 2.32s/it] 53%|█████▎ | 15906/30000 [10:12:17<8:58:20, 2.29s/it] 53%|█████▎ | 15907/30000 [10:12:19<8:57:43, 2.29s/it] 53%|█████▎ | 15908/30000 [10:12:21<9:01:02, 2.30s/it] 53%|█████▎ | 15909/30000 [10:12:23<8:37:05, 2.20s/it] 53%|█████▎ | 15910/30000 [10:12:25<8:41:03, 2.22s/it] 53%|█████▎ | 15910/30000 [10:12:26<8:41:03, 2.22s/it] 53%|█████▎ | 15911/30000 [10:12:28<8:43:00, 2.23s/it] 53%|█████▎ | 15912/30000 [10:12:30<8:48:13, 2.25s/it] 53%|█████▎ | 15913/30000 [10:12:32<8:51:30, 2.26s/it] 53%|█████▎ | 15914/30000 [10:12:35<9:01:19, 2.31s/it] 53%|█████▎ | 15915/30000 [10:12:37<8:59:10, 2.30s/it] 53%|█████▎ | 15916/30000 [10:12:39<8:59:26, 2.30s/it] 53%|█████▎ | 15917/30000 [10:12:42<8:57:53, 2.29s/it] 53%|█████▎ | 15918/30000 [10:12:44<8:49:27, 2.26s/it] 53%|█████▎ | 15919/30000 [10:12:46<8:55:24, 2.28s/it] 53%|█████▎ | 15920/30000 [10:12:48<8:58:47, 2.30s/it] 53%|█████▎ | 15920/30000 [10:12:48<8:58:47, 2.30s/it] 53%|█████▎ | 15921/30000 [10:12:51<9:00:18, 2.30s/it] 53%|█████▎ | 15922/30000 [10:12:53<8:48:24, 2.25s/it] 53%|█████▎ | 15923/30000 [10:12:55<8:34:42, 2.19s/it] 53%|█████▎ | 15924/30000 [10:12:57<7:59:46, 2.05s/it] 53%|█████▎ | 15925/30000 [10:12:59<7:48:14, 2.00s/it] 53%|█████▎ | 15926/30000 [10:13:01<8:12:25, 2.10s/it] 53%|█████▎ | 15927/30000 [10:13:03<8:24:19, 2.15s/it] 53%|█████▎ | 15928/30000 [10:13:05<8:31:19, 2.18s/it] 53%|█████▎ | 15929/30000 [10:13:08<8:45:09, 2.24s/it] 53%|█████▎ | 15930/30000 [10:13:09<7:34:06, 1.94s/it] 53%|█████▎ | 15930/30000 [10:13:09<7:34:06, 1.94s/it] 53%|█████▎ | 15931/30000 [10:13:13<10:27:31, 2.68s/it] 53%|█████▎ | 15932/30000 [10:13:17<11:29:57, 2.94s/it] 53%|█████▎ | 15933/30000 [10:13:20<11:45:12, 3.01s/it] 53%|█████▎ | 15934/30000 [10:13:22<9:54:25, 2.54s/it] 53%|█████▎ | 15935/30000 [10:13:24<9:32:13, 2.44s/it] 53%|█████▎ | 15936/30000 [10:13:26<9:22:09, 2.40s/it] 53%|█████▎ | 15937/30000 [10:13:28<9:13:19, 2.36s/it] 53%|█████▎ | 15938/30000 [10:13:31<9:07:48, 2.34s/it] 53%|█████▎ | 15939/30000 [10:13:35<11:02:06, 2.83s/it] 53%|█████▎ | 15940/30000 [10:13:37<10:37:45, 2.72s/it] 53%|█████▎ | 15940/30000 [10:13:37<10:37:45, 2.72s/it] 53%|█████▎ | 15941/30000 [10:13:42<12:52:32, 3.30s/it] 53%|█████▎ | 15942/30000 [10:13:43<10:30:08, 2.69s/it] 53%|█████▎ | 15943/30000 [10:13:45<10:06:09, 2.59s/it] 53%|█████▎ | 15944/30000 [10:13:48<9:38:29, 2.47s/it] 53%|█████▎ | 15945/30000 [10:13:50<9:23:48, 2.41s/it] 53%|█████▎ | 15946/30000 [10:13:52<9:12:13, 2.36s/it] 53%|█████▎ | 15947/30000 [10:13:55<10:27:20, 2.68s/it] 53%|█████▎ | 15948/30000 [10:14:01<14:17:01, 3.66s/it] 53%|█████▎ | 15949/30000 [10:14:07<16:12:02, 4.15s/it] 53%|█████▎ | 15950/30000 [10:14:08<13:03:09, 3.34s/it] 53%|█████▎ | 15950/30000 [10:14:08<13:03:09, 3.34s/it] 53%|█████▎ | 15951/30000 [10:14:09<10:32:07, 2.70s/it] 53%|█████▎ | 15952/30000 [10:14:12<10:05:01, 2.58s/it] 53%|█████▎ | 15953/30000 [10:14:14<9:41:58, 2.49s/it] 53%|█████▎ | 15954/30000 [10:14:16<9:32:07, 2.44s/it] 53%|█████▎ | 15955/30000 [10:14:19<9:26:41, 2.42s/it] 53%|█████▎ | 15956/30000 [10:14:23<11:17:09, 2.89s/it] 53%|█████▎ | 15957/30000 [10:14:28<14:19:42, 3.67s/it] 53%|█████▎ | 15958/30000 [10:14:30<12:42:56, 3.26s/it] 53%|█████▎ | 15959/30000 [10:14:33<11:31:09, 2.95s/it] 53%|█████▎ | 15960/30000 [10:14:34<9:53:20, 2.54s/it] 53%|█████▎ | 15960/30000 [10:14:34<9:53:20, 2.54s/it] 53%|█████▎ | 15961/30000 [10:14:36<9:07:39, 2.34s/it] 53%|█████▎ | 15962/30000 [10:14:38<9:06:48, 2.34s/it] 53%|█████▎ | 15963/30000 [10:14:41<9:02:38, 2.32s/it] 53%|█████▎ | 15964/30000 [10:14:45<11:46:03, 3.02s/it] 53%|█████▎ | 15965/30000 [10:14:50<13:11:36, 3.38s/it] 53%|█████▎ | 15966/30000 [10:14:52<11:43:55, 3.01s/it] 53%|█████▎ | 15967/30000 [10:14:54<10:54:12, 2.80s/it] 53%|█████▎ | 15968/30000 [10:14:56<10:22:41, 2.66s/it] 53%|█████▎ | 15969/30000 [10:14:58<9:36:48, 2.47s/it] 53%|█████▎ | 15970/30000 [10:15:00<9:00:48, 2.31s/it] 53%|█████▎ | 15970/30000 [10:15:00<9:00:48, 2.31s/it] 53%|█████▎ | 15971/30000 [10:15:03<9:01:36, 2.32s/it] 53%|█████▎ | 15972/30000 [10:15:07<11:01:02, 2.83s/it] 53%|█████▎ | 15973/30000 [10:15:11<12:54:59, 3.32s/it] 53%|█████▎ | 15974/30000 [10:15:12<10:32:56, 2.71s/it] 53%|█████▎ | 15975/30000 [10:15:15<10:05:11, 2.59s/it] 53%|█████▎ | 15976/30000 [10:15:17<9:51:58, 2.53s/it] 53%|█████▎ | 15977/30000 [10:15:19<9:36:12, 2.47s/it] 53%|█████▎ | 15978/30000 [10:15:22<9:25:24, 2.42s/it] 53%|█████▎ | 15979/30000 [10:15:24<9:18:43, 2.39s/it] 53%|█████▎ | 15980/30000 [10:15:28<10:57:25, 2.81s/it] 53%|█████▎ | 15980/30000 [10:15:28<10:57:25, 2.81s/it] 53%|█████▎ | 15981/30000 [10:15:33<13:18:27, 3.42s/it] 53%|█████▎ | 15982/30000 [10:15:34<11:17:32, 2.90s/it] 53%|█████▎ | 15983/30000 [10:15:36<9:43:42, 2.50s/it] 53%|█████▎ | 15984/30000 [10:15:38<9:30:04, 2.44s/it] 53%|█████▎ | 15985/30000 [10:15:41<9:17:58, 2.39s/it] 53%|█████▎ | 15986/30000 [10:15:43<9:12:12, 2.36s/it] 53%|█████▎ | 15987/30000 [10:15:45<9:05:09, 2.33s/it] 53%|█████▎ | 15988/30000 [10:15:53<15:00:39, 3.86s/it] 53%|█████▎ | 15989/30000 [10:15:56<14:57:09, 3.84s/it] 53%|█████▎ | 15990/30000 [10:15:59<13:12:01, 3.39s/it] 53%|█████▎ | 15990/30000 [10:15:59<13:12:01, 3.39s/it] 53%|█████▎ | 15991/30000 [10:16:01<11:30:48, 2.96s/it] 53%|█████▎ | 15992/30000 [10:16:03<10:18:02, 2.65s/it] 53%|█████▎ | 15993/30000 [10:16:05<9:52:37, 2.54s/it] 53%|█████▎ | 15994/30000 [10:16:07<9:37:07, 2.47s/it] 53%|█████▎ | 15995/30000 [10:16:10<9:30:28, 2.44s/it] 53%|█████▎ | 15996/30000 [10:16:13<10:11:54, 2.62s/it] 53%|█████▎ | 15997/30000 [10:16:16<10:55:21, 2.81s/it] 53%|█████▎ | 15998/30000 [10:16:18<10:18:15, 2.65s/it] 53%|█████▎ | 15999/30000 [10:16:20<9:55:24, 2.55s/it] 53%|█████▎ | 16000/30000 [10:16:23<9:29:10, 2.44s/it] 53%|█████▎ | 16000/30000 [10:16:23<9:29:10, 2.44s/it] 53%|█████▎ | 16001/30000 [10:16:25<9:15:34, 2.38s/it] 53%|█████▎ | 16002/30000 [10:16:26<8:01:37, 2.06s/it] 53%|█████▎ | 16003/30000 [10:16:28<8:19:34, 2.14s/it] 53%|█████▎ | 16004/30000 [10:16:36<14:28:21, 3.72s/it] 53%|█████▎ | 16005/30000 [10:16:37<11:45:29, 3.02s/it] 53%|█████▎ | 16006/30000 [10:16:38<9:36:35, 2.47s/it] 53%|█████▎ | 16007/30000 [10:16:40<8:48:21, 2.27s/it] 53%|█████▎ | 16008/30000 [10:16:43<8:52:38, 2.28s/it] 53%|█████▎ | 16009/30000 [10:16:45<8:49:46, 2.27s/it] 53%|█████▎ | 16010/30000 [10:16:47<8:54:38, 2.29s/it] 53%|█████▎ | 16010/30000 [10:16:47<8:54:38, 2.29s/it] 53%|█████▎ | 16011/30000 [10:16:50<8:58:04, 2.31s/it] 53%|█████▎ | 16012/30000 [10:16:58<16:03:03, 4.13s/it] 53%|█████▎ | 16013/30000 [10:17:00<13:47:14, 3.55s/it] 53%|█████▎ | 16014/30000 [10:17:02<11:18:25, 2.91s/it] 53%|█████▎ | 16015/30000 [10:17:03<9:18:52, 2.40s/it] 53%|█████▎ | 16016/30000 [10:17:05<8:59:28, 2.31s/it] 53%|█████▎ | 16017/30000 [10:17:07<8:58:08, 2.31s/it] 53%|█████▎ | 16018/30000 [10:17:09<8:58:19, 2.31s/it] 53%|█████▎ | 16019/30000 [10:17:12<8:48:21, 2.27s/it] 53%|█████▎ | 16020/30000 [10:17:21<17:36:16, 4.53s/it] 53%|█████▎ | 16020/30000 [10:17:21<17:36:16, 4.53s/it] 53%|█████▎ | 16021/30000 [10:17:24<14:58:08, 3.85s/it] 53%|█████▎ | 16022/30000 [10:17:26<12:53:41, 3.32s/it] 53%|█████▎ | 16023/30000 [10:17:27<10:23:43, 2.68s/it] 53%|█████▎ | 16024/30000 [10:17:28<8:40:52, 2.24s/it] 53%|█████▎ | 16025/30000 [10:17:30<8:38:55, 2.23s/it] 53%|█████▎ | 16026/30000 [10:17:33<8:42:28, 2.24s/it] 53%|█████▎ | 16027/30000 [10:17:35<8:50:02, 2.28s/it] 53%|█████▎ | 16028/30000 [10:17:41<13:17:44, 3.43s/it] 53%|█████▎ | 16029/30000 [10:17:43<11:52:54, 3.06s/it] 53%|█████▎ | 16030/30000 [10:17:46<11:02:45, 2.85s/it] 53%|█████▎ | 16030/30000 [10:17:46<11:02:45, 2.85s/it] 53%|█████▎ | 16031/30000 [10:17:48<10:25:23, 2.69s/it] 53%|█████▎ | 16032/30000 [10:17:50<9:56:24, 2.56s/it] 53%|█████▎ | 16033/30000 [10:17:53<9:39:28, 2.49s/it] 53%|█████▎ | 16034/30000 [10:17:55<9:24:47, 2.43s/it] 53%|█████▎ | 16035/30000 [10:17:57<9:15:46, 2.39s/it] 53%|█████▎ | 16036/30000 [10:18:02<12:04:52, 3.11s/it] 53%|█████▎ | 16037/30000 [10:18:04<11:06:10, 2.86s/it] 53%|█████▎ | 16038/30000 [10:18:06<10:22:11, 2.67s/it] 53%|█████▎ | 16039/30000 [10:18:09<10:00:49, 2.58s/it] 53%|█████▎ | 16040/30000 [10:18:11<9:45:17, 2.52s/it] 53%|█████▎ | 16040/30000 [10:18:11<9:45:17, 2.52s/it] 53%|█████▎ | 16041/30000 [10:18:13<8:29:58, 2.19s/it] 53%|█████▎ | 16042/30000 [10:18:14<7:19:24, 1.89s/it] 53%|█████▎ | 16043/30000 [10:18:15<6:33:52, 1.69s/it] 53%|█████▎ | 16044/30000 [10:18:25<16:25:11, 4.24s/it] 53%|█████▎ | 16045/30000 [10:18:27<13:26:10, 3.47s/it] 53%|█████▎ | 16046/30000 [10:18:28<11:10:46, 2.88s/it] 53%|█████▎ | 16047/30000 [10:18:31<10:29:42, 2.71s/it] 53%|█████▎ | 16048/30000 [10:18:33<10:00:03, 2.58s/it] 53%|█████▎ | 16049/30000 [10:18:35<9:42:19, 2.50s/it] 54%|█████▎ | 16050/30000 [10:18:38<9:24:22, 2.43s/it] 54%|█████▎ | 16050/30000 [10:18:38<9:24:22, 2.43s/it] 54%|█████▎ | 16051/30000 [10:18:39<8:23:34, 2.17s/it] 54%|█████▎ | 16052/30000 [10:18:47<15:08:54, 3.91s/it] 54%|█████▎ | 16053/30000 [10:18:49<13:08:14, 3.39s/it] 54%|█████▎ | 16054/30000 [10:18:50<10:35:39, 2.73s/it] 54%|█████▎ | 16055/30000 [10:18:52<8:48:00, 2.27s/it] 54%|█████▎ | 16056/30000 [10:18:53<8:07:52, 2.10s/it] 54%|█████▎ | 16057/30000 [10:18:56<8:17:26, 2.14s/it] 54%|█████▎ | 16058/30000 [10:18:58<8:28:17, 2.19s/it] 54%|█████▎ | 16059/30000 [10:19:00<8:40:45, 2.24s/it] 54%|█████▎ | 16060/30000 [10:19:10<16:48:00, 4.34s/it] 54%|█████▎ | 16060/30000 [10:19:10<16:48:00, 4.34s/it] 54%|█████▎ | 16061/30000 [10:19:12<14:24:21, 3.72s/it] 54%|█████▎ | 16062/30000 [10:19:14<12:47:40, 3.30s/it] 54%|█████▎ | 16063/30000 [10:19:16<11:37:35, 3.00s/it] 54%|█████▎ | 16064/30000 [10:19:19<10:45:58, 2.78s/it] 54%|█████▎ | 16065/30000 [10:19:21<10:13:49, 2.64s/it] 54%|█████▎ | 16066/30000 [10:19:23<9:47:11, 2.53s/it] 54%|█████▎ | 16067/30000 [10:19:26<9:30:07, 2.46s/it] 54%|█████▎ | 16068/30000 [10:19:30<12:19:01, 3.18s/it] 54%|█████▎ | 16069/30000 [10:19:33<11:18:54, 2.92s/it] 54%|█████▎ | 16070/30000 [10:19:35<10:35:18, 2.74s/it] 54%|█████▎ | 16070/30000 [10:19:35<10:35:18, 2.74s/it] 54%|█████▎ | 16071/30000 [10:19:37<9:55:45, 2.57s/it] 54%|█████▎ | 16072/30000 [10:19:40<9:36:00, 2.48s/it] 54%|█████▎ | 16073/30000 [10:19:41<8:06:21, 2.10s/it] 54%|█████▎ | 16074/30000 [10:19:43<8:17:19, 2.14s/it] 54%|█████▎ | 16075/30000 [10:19:45<8:28:41, 2.19s/it] 54%|█████▎ | 16076/30000 [10:19:53<14:46:16, 3.82s/it] 54%|█████▎ | 16077/30000 [10:19:55<13:02:37, 3.37s/it] 54%|█████▎ | 16078/30000 [10:19:58<11:47:55, 3.05s/it] 54%|█████▎ | 16079/30000 [10:20:00<10:49:28, 2.80s/it] 54%|█████▎ | 16080/30000 [10:20:02<10:12:50, 2.64s/it] 54%|█████▎ | 16080/30000 [10:20:02<10:12:50, 2.64s/it] 54%|█████▎ | 16081/30000 [10:20:04<9:04:48, 2.35s/it] 54%|█████▎ | 16082/30000 [10:20:05<7:45:13, 2.01s/it] 54%|█████▎ | 16083/30000 [10:20:06<7:19:23, 1.89s/it] 54%|█████▎ | 16084/30000 [10:20:16<15:42:58, 4.07s/it] 54%|█████▎ | 16085/30000 [10:20:17<12:29:21, 3.23s/it] 54%|█████▎ | 16086/30000 [10:20:19<11:27:12, 2.96s/it] 54%|█████▎ | 16087/30000 [10:20:22<10:43:55, 2.78s/it] 54%|█████▎ | 16088/30000 [10:20:24<10:09:25, 2.63s/it] 54%|█████▎ | 16089/30000 [10:20:26<9:45:33, 2.53s/it] 54%|█████▎ | 16090/30000 [10:20:28<8:58:18, 2.32s/it] 54%|█████▎ | 16090/30000 [10:20:28<8:58:18, 2.32s/it] 54%|█████▎ | 16091/30000 [10:20:31<9:12:16, 2.38s/it] 54%|█████▎ | 16092/30000 [10:20:36<13:16:49, 3.44s/it] 54%|█████▎ | 16093/30000 [10:20:39<12:00:46, 3.11s/it] 54%|█████▎ | 16094/30000 [10:20:41<10:29:13, 2.71s/it] 54%|█████▎ | 16095/30000 [10:20:43<10:04:36, 2.61s/it] 54%|█████▎ | 16096/30000 [10:20:45<9:43:06, 2.52s/it] 54%|█████▎ | 16097/30000 [10:20:48<9:27:17, 2.45s/it] 54%|█████▎ | 16098/30000 [10:20:50<9:17:20, 2.41s/it] 54%|█████▎ | 16099/30000 [10:20:52<8:46:33, 2.27s/it] 54%|█████▎ | 16100/30000 [10:20:53<7:42:44, 2.00s/it] 54%|█████▎ | 16100/30000 [10:20:53<7:42:44, 2.00s/it] 54%|█████▎ | 16101/30000 [10:20:54<6:48:06, 1.76s/it] 54%|█████▎ | 16102/30000 [10:20:56<6:06:47, 1.58s/it] 54%|█████▎ | 16103/30000 [10:20:57<5:38:27, 1.46s/it] 54%|█████▎ | 16104/30000 [10:20:58<5:21:45, 1.39s/it] 54%|█████▎ | 16105/30000 [10:20:59<5:25:34, 1.41s/it] 54%|█████▎ | 16106/30000 [10:21:02<6:29:55, 1.68s/it] 54%|█████▎ | 16107/30000 [10:21:04<7:18:08, 1.89s/it] 54%|█████▎ | 16108/30000 [10:21:06<7:55:02, 2.05s/it] 54%|█████▎ | 16109/30000 [10:21:09<8:09:18, 2.11s/it] 54%|█████▎ | 16110/30000 [10:21:11<8:23:55, 2.18s/it] 54%|█████▎ | 16110/30000 [10:21:11<8:23:55, 2.18s/it] 54%|█████▎ | 16111/30000 [10:21:13<8:30:32, 2.21s/it] 54%|█████▎ | 16112/30000 [10:21:16<8:36:22, 2.23s/it] 54%|█████▎ | 16113/30000 [10:21:18<8:45:07, 2.27s/it] 54%|█████▎ | 16114/30000 [10:21:20<8:44:14, 2.27s/it] 54%|█████▎ | 16115/30000 [10:21:23<8:42:59, 2.26s/it] 54%|█████▎ | 16116/30000 [10:21:25<8:45:53, 2.27s/it] 54%|█████▎ | 16117/30000 [10:21:27<8:44:33, 2.27s/it] 54%|█████▎ | 16118/30000 [10:21:29<8:44:05, 2.27s/it] 54%|█████▎ | 16119/30000 [10:21:32<8:43:15, 2.26s/it] 54%|█████▎ | 16120/30000 [10:21:34<8:47:41, 2.28s/it] 54%|█████▎ | 16120/30000 [10:21:34<8:47:41, 2.28s/it] 54%|█████▎ | 16121/30000 [10:21:36<8:49:48, 2.29s/it] 54%|█████▎ | 16122/30000 [10:21:38<8:49:07, 2.29s/it] 54%|█████▎ | 16123/30000 [10:21:41<8:53:17, 2.31s/it] 54%|█████▎ | 16124/30000 [10:21:43<8:51:39, 2.30s/it] 54%|█████▍ | 16125/30000 [10:21:45<8:49:26, 2.29s/it] 54%|█████▍ | 16126/30000 [10:21:48<8:47:30, 2.28s/it] 54%|█████▍ | 16127/30000 [10:21:50<8:53:25, 2.31s/it] 54%|█████▍ | 16128/30000 [10:21:52<8:51:58, 2.30s/it] 54%|█████▍ | 16129/30000 [10:21:55<8:48:59, 2.29s/it] 54%|█████▍ | 16130/30000 [10:21:57<8:52:06, 2.30s/it] 54%|█████▍ | 16130/30000 [10:21:57<8:52:06, 2.30s/it] 54%|█████▍ | 16131/30000 [10:21:59<8:52:16, 2.30s/it] 54%|█████▍ | 16132/30000 [10:22:01<8:49:39, 2.29s/it] 54%|█████▍ | 16133/30000 [10:22:04<8:50:22, 2.29s/it] 54%|█████▍ | 16134/30000 [10:22:06<8:48:09, 2.29s/it] 54%|█████▍ | 16135/30000 [10:22:08<8:45:14, 2.27s/it] 54%|█████▍ | 16136/30000 [10:22:11<8:45:10, 2.27s/it] 54%|█████▍ | 16137/30000 [10:22:13<8:48:33, 2.29s/it] 54%|█████▍ | 16138/30000 [10:22:15<8:45:50, 2.28s/it] 54%|█████▍ | 16139/30000 [10:22:17<8:41:55, 2.26s/it] 54%|█████▍ | 16140/30000 [10:22:20<8:47:34, 2.28s/it] +{'loss': 0.0133, 'grad_norm': 0.07386556267738342, 'learning_rate': 1.5018188163537566e-05, 'flow_matching_loss': 0.013344048336148263, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.58} +{'loss': 0.0141, 'grad_norm': 0.06501052528619766, 'learning_rate': 1.5001653469814331e-05, 'flow_matching_loss': 0.0140821504406631, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.58} +{'loss': 0.014, 'grad_norm': 0.07561297714710236, 'learning_rate': 1.4985118774081978e-05, 'flow_matching_loss': 0.01398624125868082, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.59} +{'loss': 0.0142, 'grad_norm': 0.08032281696796417, 'learning_rate': 1.4968584096431708e-05, 'flow_matching_loss': 0.014233425352722406, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.59} +{'loss': 0.0145, 'grad_norm': 0.0841061919927597, 'learning_rate': 1.4952049456954731e-05, 'flow_matching_loss': 0.014516256004571914, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.59} +{'loss': 0.0152, 'grad_norm': 0.08409544825553894, 'learning_rate': 1.4935514875742182e-05, 'flow_matching_loss': 0.015196537598967552, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.59} +{'loss': 0.0134, 'grad_norm': 0.08574692159891129, 'learning_rate': 1.4918980372885143e-05, 'flow_matching_loss': 0.013386325351893902, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.6} +{'loss': 0.0149, 'grad_norm': 0.08868012577295303, 'learning_rate': 1.4902445968474591e-05, 'flow_matching_loss': 0.01491455240175128, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.6} +{'loss': 0.0153, 'grad_norm': 0.08070006966590881, 'learning_rate': 1.4885911682601378e-05, 'flow_matching_loss': 0.015310697630047799, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.6} +{'loss': 0.0145, 'grad_norm': 0.07494336366653442, 'learning_rate': 1.4869377535356235e-05, 'flow_matching_loss': 0.014509843662381172, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.61} +{'loss': 0.0139, 'grad_norm': 0.07871320843696594, 'learning_rate': 1.4852843546829704e-05, 'flow_matching_loss': 0.013870447874069214, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.61} +{'loss': 0.0148, 'grad_norm': 0.08036062121391296, 'learning_rate': 1.483630973711214e-05, 'flow_matching_loss': 0.014847736433148384, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.61} +{'loss': 0.0145, 'grad_norm': 0.09614580124616623, 'learning_rate': 1.4819776126293677e-05, 'flow_matching_loss': 0.01454058000817895, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.61} +{'loss': 0.0141, 'grad_norm': 0.08636300265789032, 'learning_rate': 1.4803242734464214e-05, 'flow_matching_loss': 0.014092055894434453, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.62} +{'loss': 0.0143, 'grad_norm': 0.06117107719182968, 'learning_rate': 1.4786709581713383e-05, 'flow_matching_loss': 0.014255230128765107, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.62} +{'loss': 0.0154, 'grad_norm': 0.08179186284542084, 'learning_rate': 1.4770176688130524e-05, 'flow_matching_loss': 0.015391869191080331, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.62} +{'loss': 0.0167, 'grad_norm': 0.09183014184236526, 'learning_rate': 1.475364407380466e-05, 'flow_matching_loss': 0.016708184219896793, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.63} +{'loss': 0.0142, 'grad_norm': 0.08650922775268555, 'learning_rate': 1.4737111758824476e-05, 'flow_matching_loss': 0.014158191904425622, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.63} +{'loss': 0.0146, 'grad_norm': 0.08347480744123459, 'learning_rate': 1.4720579763278289e-05, 'flow_matching_loss': 0.014552545547485352, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.63} +{'loss': 0.016, 'grad_norm': 0.07432330399751663, 'learning_rate': 1.4704048107254042e-05, 'flow_matching_loss': 0.01603951258584857, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.63} +{'loss': 0.0155, 'grad_norm': 0.06635594367980957, 'learning_rate': 1.4687516810839246e-05, 'flow_matching_loss': 0.015504970960319043, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.64} +{'loss': 0.0159, 'grad_norm': 0.10431186854839325, 'learning_rate': 1.4670985894120997e-05, 'flow_matching_loss': 0.015857152361422776, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.64} +{'loss': 0.0146, 'grad_norm': 0.08237912505865097, 'learning_rate': 1.465445537718591e-05, 'flow_matching_loss': 0.01458518197759986, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.64} +{'loss': 0.0136, 'grad_norm': 0.0918801799416542, 'learning_rate': 1.463792528012012e-05, 'flow_matching_loss': 0.01357204644009471, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.65} +{'loss': 0.0175, 'grad_norm': 0.06362897902727127, 'learning_rate': 1.4621395623009252e-05, 'flow_matching_loss': 0.017459400463849306, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.65} +{'loss': 0.0154, 'grad_norm': 0.07856226712465286, 'learning_rate': 1.4604866425938407e-05, 'flow_matching_loss': 0.015372572839260102, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.65} +{'loss': 0.0142, 'grad_norm': 0.06872551143169403, 'learning_rate': 1.4588337708992115e-05, 'flow_matching_loss': 0.01423191586509347, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.66} +{'loss': 0.0144, 'grad_norm': 0.06902491301298141, 'learning_rate': 1.4571809492254318e-05, 'flow_matching_loss': 0.014382013771682977, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.66} +{'loss': 0.0167, 'grad_norm': 0.08575188368558884, 'learning_rate': 1.4555281795808371e-05, 'flow_matching_loss': 0.016723122913390398, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.66} +{'loss': 0.015, 'grad_norm': 0.09838937222957611, 'learning_rate': 1.4538754639736966e-05, 'flow_matching_loss': 0.01502839382737875, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.66} +{'loss': 0.013, 'grad_norm': 0.07068893313407898, 'learning_rate': 1.4522228044122174e-05, 'flow_matching_loss': 0.013034147769212722, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.67} +{'loss': 0.0156, 'grad_norm': 0.07665523886680603, 'learning_rate': 1.4505702029045359e-05, 'flow_matching_loss': 0.015559915732592345, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.67} +{'loss': 0.0152, 'grad_norm': 0.09281795471906662, 'learning_rate': 1.4489176614587189e-05, 'flow_matching_loss': 0.015184084419161081, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.67} +{'loss': 0.0143, 'grad_norm': 0.0919865220785141, 'learning_rate': 1.4472651820827595e-05, 'flow_matching_loss': 0.014278464205563068, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.68} +{'loss': 0.015, 'grad_norm': 0.07334495335817337, 'learning_rate': 1.4456127667845766e-05, 'flow_matching_loss': 0.014968716073781252, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.68} +{'loss': 0.0146, 'grad_norm': 0.08618137985467911, 'learning_rate': 1.4439604175720104e-05, 'flow_matching_loss': 0.014573677815496922, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.68} +{'loss': 0.0142, 'grad_norm': 0.067264124751091, 'learning_rate': 1.4423081364528212e-05, 'flow_matching_loss': 0.014227053429931403, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.68} +{'loss': 0.0147, 'grad_norm': 0.10153862088918686, 'learning_rate': 1.4406559254346863e-05, 'flow_matching_loss': 0.014692451898008585, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.69} +{'loss': 0.0148, 'grad_norm': 0.0748070776462555, 'learning_rate': 1.4390037865251978e-05, 'flow_matching_loss': 0.014821956120431422, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.69} +{'loss': 0.0146, 'grad_norm': 0.07195401936769485, 'learning_rate': 1.4373517217318597e-05, 'flow_matching_loss': 0.014645923487842083, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.69} +{'loss': 0.0137, 'grad_norm': 0.10976959764957428, 'learning_rate': 1.4356997330620875e-05, 'flow_matching_loss': 0.013721899781376123, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.7} 54%|█████▍ | 16140/30000 [10:22:20<8:47:34, 2.28s/it] 54%|█████▍ | 16141/30000 [10:22:22<8:49:25, 2.29s/it] 54%|█████▍ | 16142/30000 [10:22:24<8:48:36, 2.29s/it] 54%|█████▍ | 16143/30000 [10:22:27<8:45:09, 2.27s/it] 54%|█████▍ | 16144/30000 [10:22:29<8:51:32, 2.30s/it] 54%|█████▍ | 16145/30000 [10:22:31<8:51:19, 2.30s/it] 54%|█████▍ | 16146/30000 [10:22:33<8:49:49, 2.29s/it] 54%|█████▍ | 16147/30000 [10:22:36<8:48:01, 2.29s/it] 54%|█████▍ | 16148/30000 [10:22:38<8:47:54, 2.29s/it] 54%|█████▍ | 16149/30000 [10:22:40<8:42:27, 2.26s/it] 54%|█████▍ | 16150/30000 [10:22:42<8:42:35, 2.26s/it] 54%|█████▍ | 16150/30000 [10:22:43<8:42:35, 2.26s/it] 54%|█████▍ | 16151/30000 [10:22:45<8:47:42, 2.29s/it] 54%|█████▍ | 16152/30000 [10:22:47<8:50:39, 2.30s/it] 54%|█████▍ | 16153/30000 [10:22:49<8:47:26, 2.29s/it] 54%|█████▍ | 16154/30000 [10:22:52<8:46:57, 2.28s/it] 54%|█████▍ | 16155/30000 [10:22:54<8:53:00, 2.31s/it] 54%|█████▍ | 16156/30000 [10:22:56<8:48:54, 2.29s/it] 54%|█████▍ | 16157/30000 [10:22:59<8:46:08, 2.28s/it] 54%|█████▍ | 16158/30000 [10:23:01<8:48:22, 2.29s/it] 54%|█████▍ | 16159/30000 [10:23:03<8:48:16, 2.29s/it] 54%|█████▍ | 16160/30000 [10:23:05<8:47:57, 2.29s/it] 54%|█████▍ | 16160/30000 [10:23:06<8:47:57, 2.29s/it] 54%|█████▍ | 16161/30000 [10:23:08<8:45:22, 2.28s/it] 54%|█████▍ | 16162/30000 [10:23:10<8:46:08, 2.28s/it] 54%|█████▍ | 16163/30000 [10:23:12<8:49:06, 2.29s/it] 54%|█████▍ | 16164/30000 [10:23:15<8:44:49, 2.28s/it] 54%|█████▍ | 16165/30000 [10:23:17<8:48:53, 2.29s/it] 54%|█████▍ | 16166/30000 [10:23:19<8:49:42, 2.30s/it] 54%|█████▍ | 16167/30000 [10:23:21<8:45:29, 2.28s/it] 54%|█████▍ | 16168/30000 [10:23:24<8:46:02, 2.28s/it] 54%|█████▍ | 16169/30000 [10:23:26<8:57:25, 2.33s/it] 54%|█████▍ | 16170/30000 [10:23:28<8:56:34, 2.33s/it] 54%|█████▍ | 16170/30000 [10:23:29<8:56:34, 2.33s/it] 54%|█████▍ | 16171/30000 [10:23:31<8:53:02, 2.31s/it] 54%|█████▍ | 16172/30000 [10:23:33<8:51:07, 2.30s/it] 54%|█████▍ | 16173/30000 [10:23:35<8:49:09, 2.30s/it] 54%|█████▍ | 16174/30000 [10:23:38<8:45:21, 2.28s/it] 54%|█████▍ | 16175/30000 [10:23:40<8:46:19, 2.28s/it] 54%|█████▍ | 16176/30000 [10:23:42<8:48:19, 2.29s/it] 54%|█████▍ | 16177/30000 [10:23:44<8:49:22, 2.30s/it] 54%|█████▍ | 16178/30000 [10:23:47<8:47:13, 2.29s/it] 54%|█████▍ | 16179/30000 [10:23:49<8:47:23, 2.29s/it] 54%|█████▍ | 16180/30000 [10:23:51<8:50:24, 2.30s/it] 54%|█████▍ | 16180/30000 [10:23:51<8:50:24, 2.30s/it] 54%|█████▍ | 16181/30000 [10:23:54<8:51:10, 2.31s/it] 54%|█████▍ | 16182/30000 [10:23:56<8:48:28, 2.29s/it] 54%|█████▍ | 16183/30000 [10:23:58<8:46:04, 2.28s/it] 54%|█████▍ | 16184/30000 [10:24:01<8:50:20, 2.30s/it] 54%|█████▍ | 16185/30000 [10:24:03<8:51:20, 2.31s/it] 54%|█████▍ | 16186/30000 [10:24:05<8:47:06, 2.29s/it] 54%|█████▍ | 16187/30000 [10:24:07<8:47:41, 2.29s/it] 54%|█████▍ | 16188/30000 [10:24:10<8:47:48, 2.29s/it] 54%|█████▍ | 16189/30000 [10:24:12<8:45:36, 2.28s/it] 54%|█████▍ | 16190/30000 [10:24:14<8:42:57, 2.27s/it] 54%|█████▍ | 16190/30000 [10:24:14<8:42:57, 2.27s/it] 54%|█████▍ | 16191/30000 [10:24:17<8:48:16, 2.30s/it] 54%|█████▍ | 16192/30000 [10:24:19<8:53:02, 2.32s/it] 54%|█████▍ | 16193/30000 [10:24:21<8:51:01, 2.31s/it] 54%|█████▍ | 16194/30000 [10:24:24<8:49:11, 2.30s/it] 54%|█████▍ | 16195/30000 [10:24:26<8:52:23, 2.31s/it] 54%|█████▍ | 16196/30000 [10:24:28<8:53:22, 2.32s/it] 54%|█████▍ | 16197/30000 [10:24:30<8:50:55, 2.31s/it] 54%|█████▍ | 16198/30000 [10:24:33<8:48:13, 2.30s/it] 54%|█████▍ | 16199/30000 [10:24:35<8:46:26, 2.29s/it] 54%|█████▍ | 16200/30000 [10:24:37<8:46:32, 2.29s/it] 54%|█████▍ | 16200/30000 [10:24:37<8:46:32, 2.29s/it] 54%|█████▍ | 16201/30000 [10:24:40<8:46:35, 2.29s/it] 54%|█████▍ | 16202/30000 [10:24:42<8:50:38, 2.31s/it] 54%|█████▍ | 16203/30000 [10:24:44<8:52:29, 2.32s/it] 54%|█████▍ | 16204/30000 [10:24:47<8:47:06, 2.29s/it] 54%|█████▍ | 16205/30000 [10:24:49<8:44:14, 2.28s/it] 54%|█████▍ | 16206/30000 [10:24:51<8:48:42, 2.30s/it] 54%|█████▍ | 16207/30000 [10:24:53<8:50:53, 2.31s/it] 54%|█████▍ | 16208/30000 [10:24:56<8:47:55, 2.30s/it] 54%|█████▍ | 16209/30000 [10:24:58<8:44:58, 2.28s/it] 54%|█████▍ | 16210/30000 [10:25:00<8:50:09, 2.31s/it] 54%|█████▍ | 16210/30000 [10:25:00<8:50:09, 2.31s/it] 54%|█████▍ | 16211/30000 [10:25:03<8:53:18, 2.32s/it] 54%|█████▍ | 16212/30000 [10:25:05<8:47:43, 2.30s/it] 54%|█████▍ | 16213/30000 [10:25:07<8:49:34, 2.30s/it] 54%|█████▍ | 16214/30000 [10:25:09<8:45:31, 2.29s/it] 54%|█████▍ | 16215/30000 [10:25:12<8:42:24, 2.27s/it] 54%|█████▍ | 16216/30000 [10:25:14<8:44:27, 2.28s/it] 54%|█████▍ | 16217/30000 [10:25:16<8:46:08, 2.29s/it] 54%|█████▍ | 16218/30000 [10:25:19<8:43:00, 2.28s/it] 54%|█████▍ | 16219/30000 [10:25:21<8:41:43, 2.27s/it] 54%|█████▍ | 16220/30000 [10:25:23<8:44:26, 2.28s/it] 54%|█████▍ | 16220/30000 [10:25:23<8:44:26, 2.28s/it] 54%|█████▍ | 16221/30000 [10:25:25<8:42:29, 2.28s/it] 54%|█████▍ | 16222/30000 [10:25:28<8:42:56, 2.28s/it] 54%|█████▍ | 16223/30000 [10:25:30<8:46:21, 2.29s/it] 54%|█████▍ | 16224/30000 [10:25:32<8:47:14, 2.30s/it] 54%|█████▍ | 16225/30000 [10:25:35<8:46:55, 2.30s/it] 54%|█████▍ | 16226/30000 [10:25:37<8:46:40, 2.29s/it] 54%|█████▍ | 16227/30000 [10:25:39<8:46:34, 2.29s/it] 54%|█████▍ | 16228/30000 [10:25:41<8:42:09, 2.27s/it] 54%|█████▍ | 16229/30000 [10:25:44<8:45:38, 2.29s/it] 54%|█████▍ | 16230/30000 [10:25:46<8:44:08, 2.28s/it] 54%|█████▍ | 16230/30000 [10:25:46<8:44:08, 2.28s/it] 54%|█████▍ | 16231/30000 [10:25:48<8:43:45, 2.28s/it] 54%|█████▍ | 16232/30000 [10:25:51<8:43:40, 2.28s/it] 54%|█████▍ | 16233/30000 [10:25:53<8:49:25, 2.31s/it] 54%|█████▍ | 16234/30000 [10:25:55<8:45:50, 2.29s/it] 54%|█████▍ | 16235/30000 [10:25:58<8:46:23, 2.29s/it] 54%|█████▍ | 16236/30000 [10:26:00<8:48:52, 2.31s/it] 54%|█████▍ | 16237/30000 [10:26:02<8:50:36, 2.31s/it] 54%|█████▍ | 16238/30000 [10:26:04<8:47:44, 2.30s/it] 54%|█████▍ | 16239/30000 [10:26:07<8:50:12, 2.31s/it] 54%|█████▍ | 16240/30000 [10:26:09<8:52:31, 2.32s/it] 54%|█████▍ | 16240/30000 [10:26:09<8:52:31, 2.32s/it] 54%|█████▍ | 16241/30000 [10:26:11<8:48:57, 2.31s/it] 54%|█████▍ | 16242/30000 [10:26:14<8:45:35, 2.29s/it] 54%|█████▍ | 16243/30000 [10:26:16<8:49:40, 2.31s/it] 54%|█████▍ | 16244/30000 [10:26:18<8:46:21, 2.30s/it] 54%|█████▍ | 16245/30000 [10:26:21<8:46:01, 2.29s/it] 54%|█████▍ | 16246/30000 [10:26:23<8:51:00, 2.32s/it] 54%|█████▍ | 16247/30000 [10:26:25<8:49:43, 2.31s/it] 54%|█████▍ | 16248/30000 [10:26:28<8:49:02, 2.31s/it] 54%|█████▍ | 16249/30000 [10:26:30<8:52:44, 2.32s/it] 54%|█████▍ | 16250/30000 [10:26:32<8:51:25, 2.32s/it] 54%|█████▍ | 16250/30000 [10:26:32<8:51:25, 2.32s/it] 54%|█████▍ | 16251/30000 [10:26:34<8:46:30, 2.30s/it] 54%|█████▍ | 16252/30000 [10:26:37<8:47:23, 2.30s/it] 54%|█████▍ | 16253/30000 [10:26:39<8:45:02, 2.29s/it] 54%|█████▍ | 16254/30000 [10:26:41<8:43:49, 2.29s/it] 54%|█████▍ | 16255/30000 [10:26:44<8:44:39, 2.29s/it] 54%|█████▍ | 16256/30000 [10:26:46<8:44:22, 2.29s/it] 54%|█████▍ | 16257/30000 [10:26:48<8:44:16, 2.29s/it] 54%|█████▍ | 16258/30000 [10:26:50<8:43:09, 2.28s/it] 54%|█████▍ | 16259/30000 [10:26:53<8:47:37, 2.30s/it] 54%|█████▍ | 16260/30000 [10:26:55<8:43:07, 2.28s/it] 54%|█████▍ | 16260/30000 [10:26:55<8:43:07, 2.28s/it] 54%|█████▍ | 16261/30000 [10:26:57<8:41:23, 2.28s/it] 54%|█████▍ | 16262/30000 [10:27:00<8:45:37, 2.30s/it] 54%|█████▍ | 16263/30000 [10:27:02<8:46:08, 2.30s/it] 54%|█████▍ | 16264/30000 [10:27:04<8:44:56, 2.29s/it] 54%|█████▍ | 16265/30000 [10:27:07<8:47:42, 2.31s/it] 54%|█████▍ | 16266/30000 [10:27:09<8:48:55, 2.31s/it] 54%|█████▍ | 16267/30000 [10:27:11<8:47:30, 2.30s/it] 54%|█████▍ | 16268/30000 [10:27:13<8:44:36, 2.29s/it] 54%|█████▍ | 16269/30000 [10:27:16<8:45:26, 2.30s/it] 54%|█████▍ | 16270/30000 [10:27:18<8:42:27, 2.28s/it] 54%|█████▍ | 16270/30000 [10:27:18<8:42:27, 2.28s/it] 54%|█████▍ | 16271/30000 [10:27:20<8:40:59, 2.28s/it] 54%|█████▍ | 16272/30000 [10:27:23<8:47:14, 2.30s/it] 54%|█████▍ | 16273/30000 [10:27:25<8:48:11, 2.31s/it] 54%|█████▍ | 16274/30000 [10:27:27<8:45:49, 2.30s/it] 54%|█████▍ | 16275/30000 [10:27:29<8:43:28, 2.29s/it] 54%|█████▍ | 16276/30000 [10:27:32<8:41:57, 2.28s/it] 54%|█████▍ | 16277/30000 [10:27:34<8:39:29, 2.27s/it] 54%|█████▍ | 16278/30000 [10:27:36<8:39:55, 2.27s/it] 54%|█████▍ | 16279/30000 [10:27:39<8:42:51, 2.29s/it] 54%|█████▍ | 16280/30000 [10:27:41<8:41:21, 2.28s/it] 54%|█████▍ | 16280/30000 [10:27:41<8:41:21, 2.28s/it] 54%|█████▍ | 16281/30000 [10:27:43<8:41:19, 2.28s/it] 54%|█████▍ | 16282/30000 [10:27:45<8:45:18, 2.30s/it] 54%|█████▍ | 16283/30000 [10:27:48<8:44:39, 2.29s/it] 54%|█████▍ | 16284/30000 [10:27:50<8:41:13, 2.28s/it] 54%|█████▍ | 16285/30000 [10:27:52<8:40:16, 2.28s/it] 54%|█████▍ | 16286/30000 [10:27:55<8:45:12, 2.30s/it] 54%|█████▍ | 16287/30000 [10:27:57<8:47:55, 2.31s/it] 54%|█████▍ | 16288/30000 [10:27:59<8:43:04, 2.29s/it] 54%|█████▍ | 16289/30000 [10:28:01<8:42:22, 2.29s/it] 54%|█████▍ | 16290/30000 [10:28:04<8:48:09, 2.31s/it] 54%|█████▍ | 16290/30000 [10:28:04<8:48:09, 2.31s/it] 54%|█████▍ | 16291/30000 [10:28:06<8:47:41, 2.31s/it] 54%|█████▍ | 16292/30000 [10:28:08<8:42:08, 2.29s/it] 54%|█████▍ | 16293/30000 [10:28:11<8:44:20, 2.30s/it] 54%|█████▍ | 16294/30000 [10:28:13<8:45:47, 2.30s/it] 54%|█████▍ | 16295/30000 [10:28:15<8:41:23, 2.28s/it] 54%|█████▍ | 16296/30000 [10:28:17<8:37:47, 2.27s/it] 54%|█████▍ | 16297/30000 [10:28:20<8:43:26, 2.29s/it] 54%|█████▍ | 16298/30000 [10:28:22<8:45:42, 2.30s/it] 54%|█████▍ | 16299/30000 [10:28:24<8:44:04, 2.30s/it] 54%|█████▍ | 16300/30000 [10:28:27<8:41:22, 2.28s/it] 54%|█████▍ | 16300/30000 [10:28:27<8:41:22, 2.28s/it] 54%|█████▍ | 16301/30000 [10:28:29<8:46:25, 2.31s/it] 54%|█████▍ | 16302/30000 [10:28:31<8:49:46, 2.32s/it] 54%|█████▍ | 16303/30000 [10:28:34<8:45:55, 2.30s/it] 54%|█████▍ | 16304/30000 [10:28:36<8:46:02, 2.30s/it] 54%|█████▍ | 16305/30000 [10:28:38<8:50:50, 2.33s/it] 54%|█████▍ | 16306/30000 [10:28:41<8:50:41, 2.33s/it] 54%|█████▍ | 16307/30000 [10:28:43<8:44:46, 2.30s/it] 54%|█████▍ | 16308/30000 [10:28:45<8:42:42, 2.29s/it] 54%|█████▍ | 16309/30000 [10:28:48<8:47:41, 2.31s/it] 54%|█████▍ | 16310/30000 [10:28:50<8:47:19, 2.31s/it] 54%|█████▍ | 16310/30000 [10:28:50<8:47:19, 2.31s/it] 54%|█████▍ | 16311/30000 [10:28:52<8:42:02, 2.29s/it] 54%|█████▍ | 16312/30000 [10:28:54<8:42:01, 2.29s/it] 54%|█████▍ | 16313/30000 [10:28:57<8:46:45, 2.31s/it] 54%|█████▍ | 16314/30000 [10:28:59<8:47:22, 2.31s/it] 54%|█████▍ | 16315/30000 [10:29:01<8:44:40, 2.30s/it] 54%|█████▍ | 16316/30000 [10:29:04<8:44:20, 2.30s/it] 54%|█████▍ | 16317/30000 [10:29:06<8:45:05, 2.30s/it] 54%|█████▍ | 16318/30000 [10:29:08<8:41:40, 2.29s/it] 54%|█████▍ | 16319/30000 [10:29:10<8:37:40, 2.27s/it] 54%|█████▍ | 16320/30000 [10:29:13<8:43:59, 2.30s/it] 54%|█████▍ | 16320/30000 [10:29:13<8:43:59, 2.30s/it] 54%|█████▍ | 16321/30000 [10:29:15<8:42:56, 2.29s/it] 54%|█████▍ | 16322/30000 [10:29:17<8:40:30, 2.28s/it] 54%|█████▍ | 16323/30000 [10:29:20<8:43:19, 2.30s/it] 54%|█████▍ | 16324/30000 [10:29:22<8:42:45, 2.29s/it] 54%|█████▍ | 16325/30000 [10:29:24<8:42:03, 2.29s/it] 54%|█████▍ | 16326/30000 [10:29:27<8:40:17, 2.28s/it] 54%|█████▍ | 16327/30000 [10:29:29<8:44:34, 2.30s/it] 54%|█████▍ | 16328/30000 [10:29:31<8:47:33, 2.32s/it] 54%|█████▍ | 16329/30000 [10:29:33<8:46:12, 2.31s/it] 54%|█████▍ | 16330/30000 [10:29:36<8:43:26, 2.30s/it] 54%|█████▍ | 16330/30000 [10:29:36<8:43:26, 2.30s/it] 54%|█████▍ | 16331/30000 [10:29:38<8:48:04, 2.32s/it] 54%|█████▍ | 16332/30000 [10:29:40<8:47:30, 2.32s/it] 54%|█████▍ | 16333/30000 [10:29:43<8:46:01, 2.31s/it] 54%|█████▍ | 16334/30000 [10:29:45<8:46:33, 2.31s/it] 54%|█████▍ | 16335/30000 [10:29:47<8:48:29, 2.32s/it] 54%|█████▍ | 16336/30000 [10:29:50<8:48:34, 2.32s/it] 54%|█████▍ | 16337/30000 [10:29:52<8:50:05, 2.33s/it] 54%|█████▍ | 16338/30000 [10:29:54<8:46:31, 2.31s/it] 54%|█████▍ | 16339/30000 [10:29:57<8:40:44, 2.29s/it] 54%|█████▍ | 16340/30000 [10:29:59<8:45:18, 2.31s/it] 54%|█████▍ | 16340/30000 [10:29:59<8:45:18, 2.31s/it] 54%|█████▍ | 16341/30000 [10:30:01<8:48:05, 2.32s/it] 54%|█████▍ | 16342/30000 [10:30:04<8:44:24, 2.30s/it] 54%|█████▍ | 16343/30000 [10:30:06<8:43:58, 2.30s/it] 54%|█████▍ | 16344/30000 [10:30:08<8:42:59, 2.30s/it] 54%|█████▍ | 16345/30000 [10:30:10<8:40:03, 2.29s/it] 54%|█████▍ | 16346/30000 [10:30:13<8:38:23, 2.28s/it] 54%|█████▍ | 16347/30000 [10:30:15<8:41:17, 2.29s/it] 54%|█████▍ | 16348/30000 [10:30:17<8:38:50, 2.28s/it] 54%|█████▍ | 16349/30000 [10:30:19<8:38:01, 2.28s/it] 55%|█████▍ | 16350/30000 [10:30:22<8:40:07, 2.29s/it] 55%|█████▍ | 16350/30000 [10:30:22<8:40:07, 2.29s/it] 55%|█████▍ | 16351/30000 [10:30:24<8:42:23, 2.30s/it] 55%|█████▍ | 16352/30000 [10:30:26<8:43:32, 2.30s/it] 55%|█████▍ | 16353/30000 [10:30:29<8:38:29, 2.28s/it] 55%|█████▍ | 16354/30000 [10:30:31<8:37:35, 2.28s/it] 55%|█████▍ | 16355/30000 [10:30:33<8:41:59, 2.30s/it] 55%|█████▍ | 16356/30000 [10:30:36<8:41:43, 2.29s/it] 55%|█████▍ | 16357/30000 [10:30:38<8:39:37, 2.29s/it] 55%|█████▍ | 16358/30000 [10:30:40<8:40:37, 2.29s/it] 55%|█████▍ | 16359/30000 [10:30:42<8:41:01, 2.29s/it] 55%|█████▍ | 16360/30000 [10:30:45<8:40:20, 2.29s/it] 55%|█████▍ | 16360/30000 [10:30:45<8:40:20, 2.29s/it] 55%|█████▍ | 16361/30000 [10:30:47<8:39:33, 2.29s/it] 55%|█████▍ | 16362/30000 [10:30:49<8:42:54, 2.30s/it] 55%|█████▍ | 16363/30000 [10:30:52<8:42:44, 2.30s/it] 55%|█████▍ | 16364/30000 [10:30:54<8:38:56, 2.28s/it] 55%|█████▍ | 16365/30000 [10:30:56<8:41:48, 2.30s/it] 55%|█████▍ | 16366/30000 [10:30:58<8:41:44, 2.30s/it] 55%|█████▍ | 16367/30000 [10:31:01<8:39:18, 2.29s/it] 55%|█████▍ | 16368/30000 [10:31:03<8:34:12, 2.26s/it] 55%|█████▍ | 16369/30000 [10:31:05<8:36:47, 2.27s/it] 55%|█████▍ | 16370/30000 [10:31:08<8:36:34, 2.27s/it] 55%|█████▍ | 16370/30000 [10:31:08<8:36:34, 2.27s/it] 55%|█████▍ | 16371/30000 [10:31:10<8:34:54, 2.27s/it] 55%|█████▍ | 16372/30000 [10:31:12<8:38:09, 2.28s/it] 55%|█████▍ | 16373/30000 [10:31:14<8:39:59, 2.29s/it] 55%|█████▍ | 16374/30000 [10:31:17<8:36:34, 2.27s/it] 55%|█████▍ | 16375/30000 [10:31:19<8:42:33, 2.30s/it] 55%|█████▍ | 16376/30000 [10:31:21<8:42:30, 2.30s/it] 55%|█████▍ | 16377/30000 [10:31:24<8:39:37, 2.29s/it] 55%|█████▍ | 16378/30000 [10:31:26<8:40:16, 2.29s/it] 55%|█████▍ | 16379/30000 [10:31:28<8:43:07, 2.30s/it] 55%|█████▍ | 16380/30000 [10:31:30<8:39:48, 2.29s/it] 55%|█████▍ | 16380/30000 [10:31:30<8:39:48, 2.29s/it] 55%|█████▍ | 16381/30000 [10:31:33<8:35:29, 2.27s/it] 55%|█████▍ | 16382/30000 [10:31:35<8:38:25, 2.28s/it] 55%|█████▍ | 16383/30000 [10:31:37<8:38:17, 2.28s/it] 55%|█████▍ | 16384/30000 [10:31:40<8:36:43, 2.28s/it] 55%|█████▍ | 16385/30000 [10:31:42<8:42:39, 2.30s/it] 55%|█████▍ | 16386/30000 [10:31:44<8:44:12, 2.31s/it] 55%|█████▍ | 16387/30000 [10:31:47<8:42:43, 2.30s/it] 55%|█████▍ | 16388/30000 [10:31:49<8:44:10, 2.31s/it] 55%|█████▍ | 16389/30000 [10:31:51<8:43:54, 2.31s/it] 55%|█████▍ | 16390/30000 [10:31:53<8:41:58, 2.30s/it] 55%|█████▍ | 16390/30000 [10:31:53<8:41:58, 2.30s/it] 55%|█████▍ | 16391/30000 [10:31:56<8:41:54, 2.30s/it] 55%|█████▍ | 16392/30000 [10:31:58<8:42:08, 2.30s/it] 55%|█████▍ | 16393/30000 [10:32:00<8:40:31, 2.30s/it] 55%|█████▍ | 16394/30000 [10:32:03<8:37:58, 2.28s/it] 55%|█████▍ | 16395/30000 [10:32:05<8:42:08, 2.30s/it] 55%|█████▍ | 16396/30000 [10:32:07<8:37:38, 2.28s/it] 55%|█████▍ | 16397/30000 [10:32:09<8:35:24, 2.27s/it] 55%|█████▍ | 16398/30000 [10:32:12<8:43:09, 2.31s/it] 55%|█████▍ | 16399/30000 [10:32:14<8:39:44, 2.29s/it] 55%|█████▍ | 16400/30000 [10:32:16<8:35:53, 2.28s/it] 55%|█████▍ | 16400/30000 [10:32:16<8:35:53, 2.28s/it] 55%|█████▍ | 16401/30000 [10:32:19<8:41:45, 2.30s/it] 55%|█████▍ | 16402/30000 [10:32:21<8:41:54, 2.30s/it] 55%|█████▍ | 16403/30000 [10:32:23<8:41:40, 2.30s/it] 55%|█████▍ | 16404/30000 [10:32:26<8:41:47, 2.30s/it] 55%|█████▍ | 16405/30000 [10:32:28<8:41:51, 2.30s/it] 55%|█████▍ | 16406/30000 [10:32:30<8:40:24, 2.30s/it] 55%|█████▍ | 16407/30000 [10:32:32<8:44:22, 2.31s/it] 55%|█████▍ | 16408/30000 [10:32:35<8:47:24, 2.33s/it] 55%|█████▍ | 16409/30000 [10:32:37<8:44:52, 2.32s/it] 55%|█████▍ | 16410/30000 [10:32:39<8:43:59, 2.31s/it] 55%|█████▍ | 16410/30000 [10:32:39<8:43:59, 2.31s/it] 55%|█████▍ | 16411/30000 [10:32:42<8:47:54, 2.33s/it] 55%|█████▍ | 16412/30000 [10:32:44<8:44:29, 2.32s/it] 55%|█████▍ | 16413/30000 [10:32:46<8:39:48, 2.30s/it] 55%|█████▍ | 16414/30000 [10:32:49<8:43:46, 2.31s/it] 55%|█████▍ | 16415/30000 [10:32:51<8:41:02, 2.30s/it] 55%|█████▍ | 16416/30000 [10:32:53<8:40:47, 2.30s/it] 55%|█████▍ | 16417/30000 [10:32:56<8:44:20, 2.32s/it] 55%|█████▍ | 16418/30000 [10:32:58<8:47:38, 2.33s/it] 55%|█████▍ | 16419/30000 [10:33:00<8:43:09, 2.31s/it] 55%|█████▍ | 16420/30000 [10:33:03<8:42:09, 2.31s/it] 55%|█████▍ | 16420/30000 [10:33:03<8:42:09, 2.31s/it] 55%|█████▍ | 16421/30000 [10:33:05<8:42:08, 2.31s/it] 55%|█████▍ | 16422/30000 [10:33:07<8:37:37, 2.29s/it] 55%|█████▍ | 16423/30000 [10:33:09<8:37:36, 2.29s/it] 55%|█████▍ | 16424/30000 [10:33:12<8:38:55, 2.29s/it] 55%|█████▍ | 16425/30000 [10:33:14<8:36:55, 2.28s/it] 55%|█████▍ | 16426/30000 [10:33:16<8:38:47, 2.29s/it] 55%|█████▍ | 16427/30000 [10:33:19<8:43:50, 2.32s/it] 55%|█████▍ | 16428/30000 [10:33:21<8:43:35, 2.31s/it] 55%|█████▍ | 16429/30000 [10:33:23<8:38:52, 2.29s/it] 55%|█████▍ | 16430/30000 [10:33:26<8:43:26, 2.31s/it] 55%|█████▍ | 16430/30000 [10:33:26<8:43:26, 2.31s/it] 55%|█████▍ | 16431/30000 [10:33:28<8:41:17, 2.31s/it] 55%|█████▍ | 16432/30000 [10:33:30<8:40:29, 2.30s/it] 55%|█████▍ | 16433/30000 [10:33:32<8:38:00, 2.29s/it] 55%|█████▍ | 16434/30000 [10:33:35<8:39:04, 2.30s/it] 55%|█████▍ | 16435/30000 [10:33:37<8:38:40, 2.29s/it] 55%|█████▍ | 16436/30000 [10:33:39<8:36:19, 2.28s/it] 55%|█████▍ | 16437/30000 [10:33:42<8:42:23, 2.31s/it] 55%|█████▍ | 16438/30000 [10:33:44<8:42:56, 2.31s/it] 55%|█████▍ | 16439/30000 [10:33:46<8:39:27, 2.30s/it] 55%|█████▍ | 16440/30000 [10:33:49<8:44:14, 2.32s/it] 55%|█████▍ | 16440/30000 [10:33:49<8:44:14, 2.32s/it] 55%|█████▍ | 16441/30000 [10:33:51<8:44:35, 2.32s/it] 55%|█████▍ | 16442/30000 [10:33:53<8:42:18, 2.31s/it] 55%|█████▍ | 16443/30000 [10:33:56<8:42:26, 2.31s/it] 55%|█████▍ | 16444/30000 [10:33:58<8:43:07, 2.32s/it] 55%|█████▍ | 16445/30000 [10:34:00<8:39:37, 2.30s/it] 55%|█████▍ | 16446/30000 [10:34:02<8:37:08, 2.29s/it] 55%|█████▍ | 16447/30000 [10:34:05<8:40:27, 2.30s/it] 55%|█████▍ | 16448/30000 [10:34:07<8:42:54, 2.32s/it] 55%|█████▍ | 16449/30000 [10:34:09<8:39:03, 2.30s/it] 55%|█████▍ | 16450/30000 [10:34:12<8:42:58, 2.32s/it] 55%|█████▍ | 16450/30000 [10:34:12<8:42:58, 2.32s/it] 55%|█████▍ | 16451/30000 [10:34:14<8:41:03, 2.31s/it] 55%|█████▍ | 16452/30000 [10:34:16<8:38:04, 2.29s/it] 55%|█████▍ | 16453/30000 [10:34:19<8:38:58, 2.30s/it] 55%|█████▍ | 16454/30000 [10:34:21<8:40:33, 2.31s/it] 55%|█████▍ | 16455/30000 [10:34:23<8:37:11, 2.29s/it] 55%|█████▍ | 16456/30000 [10:34:25<8:37:18, 2.29s/it] 55%|█████▍ | 16457/30000 [10:34:28<8:40:15, 2.30s/it] 55%|█████▍ | 16458/30000 [10:34:30<8:36:12, 2.29s/it] 55%|█████▍ | 16459/30000 [10:34:32<8:36:47, 2.29s/it] 55%|█████▍ | 16460/30000 [10:34:35<8:41:10, 2.31s/it] 55%|█████▍ | 16460/30000 [10:34:35<8:41:10, 2.31s/it] 55%|█████▍ | 16461/30000 [10:34:37<8:42:59, 2.32s/it] 55%|█████▍ | 16462/30000 [10:34:39<8:37:21, 2.29s/it] 55%|█████▍ | 16463/30000 [10:34:42<8:40:22, 2.31s/it] 55%|█████▍ | 16464/30000 [10:34:44<8:39:13, 2.30s/it] 55%|█████▍ | 16465/30000 [10:34:46<8:39:29, 2.30s/it] 55%|█████▍ | 16466/30000 [10:34:48<8:36:52, 2.29s/it] 55%|█████▍ | 16467/30000 [10:34:51<8:42:46, 2.32s/it] 55%|█████▍ | 16468/30000 [10:34:53<8:38:08, 2.30s/it] 55%|█████▍ | 16469/30000 [10:34:55<8:35:08, 2.28s/it] 55%|█████▍ | 16470/30000 [10:34:58<8:37:51, 2.30s/it] 55%|█████▍ | 16470/30000 [10:34:58<8:37:51, 2.30s/it] 55%|█████▍ | 16471/30000 [10:35:00<8:37:12, 2.29s/it] 55%|█████▍ | 16472/30000 [10:35:02<8:36:39, 2.29s/it] 55%|█████▍ | 16473/30000 [10:35:04<8:35:49, 2.29s/it] 55%|█████▍ | 16474/30000 [10:35:07<8:40:44, 2.31s/it] 55%|█████▍ | 16475/30000 [10:35:09<8:42:13, 2.32s/it] 55%|█████▍ | 16476/30000 [10:35:11<8:36:36, 2.29s/it] 55%|█████▍ | 16477/30000 [10:35:14<8:39:44, 2.31s/it] 55%|█████▍ | 16478/30000 [10:35:16<8:40:30, 2.31s/it] 55%|█████▍ | 16479/30000 [10:35:18<8:38:09, 2.30s/it] 55%|█████▍ | 16480/30000 [10:35:21<8:36:47, 2.29s/it] 55%|█████▍ | 16480/30000 [10:35:21<8:36:47, 2.29s/it] 55%|█████▍ | 16481/30000 [10:35:23<8:35:56, 2.29s/it] 55%|█████▍ | 16482/30000 [10:35:25<8:34:48, 2.28s/it] 55%|█████▍ | 16483/30000 [10:35:27<8:34:28, 2.28s/it] 55%|█████▍ | 16484/30000 [10:35:30<8:39:24, 2.31s/it] 55%|█████▍ | 16485/30000 [10:35:32<8:39:33, 2.31s/it] 55%|█████▍ | 16486/30000 [10:35:34<8:38:25, 2.30s/it] 55%|█████▍ | 16487/30000 [10:35:37<8:34:29, 2.28s/it] 55%|█████▍ | 16488/30000 [10:35:39<8:38:09, 2.30s/it] 55%|█████▍ | 16489/30000 [10:35:41<8:35:37, 2.29s/it] 55%|█████▍ | 16490/30000 [10:35:44<8:33:56, 2.28s/it] 55%|█████▍ | 16490/30000 [10:35:44<8:33:56, 2.28s/it] 55%|█████▍ | 16491/30000 [10:35:46<8:34:52, 2.29s/it] 55%|█████▍ | 16492/30000 [10:35:48<8:38:11, 2.30s/it] 55%|█████▍ | 16493/30000 [10:35:50<8:36:45, 2.30s/it] 55%|█████▍ | 16494/30000 [10:35:53<8:31:47, 2.27s/it] 55%|█████▍ | 16495/30000 [10:35:55<8:35:51, 2.29s/it] 55%|█████▍ | 16496/30000 [10:35:57<8:36:33, 2.30s/it] 55%|█████▍ | 16497/30000 [10:36:00<8:37:27, 2.30s/it] 55%|█████▍ | 16498/30000 [10:36:02<8:33:09, 2.28s/it] 55%|█████▍ | 16499/30000 [10:36:04<8:35:35, 2.29s/it] 55%|█████▌ | 16500/30000 [10:36:06<8:31:53, 2.28s/it] 55%|█████▌ | 16500/30000 [10:36:06<8:31:53, 2.28s/it] 55%|█████▌ | 16501/30000 [10:36:09<8:32:43, 2.28s/it] 55%|█████▌ | 16502/30000 [10:36:11<8:36:30, 2.30s/it] 55%|█████▌ | 16503/30000 [10:36:13<8:36:58, 2.30s/it] 55%|█████▌ | 16504/30000 [10:36:16<8:35:41, 2.29s/it] 55%|█████▌ | 16505/30000 [10:36:18<8:37:49, 2.30s/it] 55%|█████▌ | 16506/30000 [10:36:20<8:43:02, 2.33s/it] 55%|█████▌ | 16507/30000 [10:36:23<8:43:15, 2.33s/it] 55%|█████▌ | 16508/30000 [10:36:25<8:36:32, 2.30s/it] 55%|█████▌ | 16509/30000 [10:36:27<8:34:36, 2.29s/it] 55%|█████▌ | 16510/30000 [10:36:29<8:39:01, 2.31s/it] 55%|█████▌ | 16510/30000 [10:36:30<8:39:01, 2.31s/it] 55%|█████▌ | 16511/30000 [10:36:32<8:34:00, 2.29s/it] 55%|█████▌ | 16512/30000 [10:36:34<8:33:05, 2.28s/it] 55%|█████▌ | 16513/30000 [10:36:36<8:35:21, 2.29s/it] 55%|█████▌ | 16514/30000 [10:36:39<8:36:16, 2.30s/it] 55%|█████▌ | 16515/30000 [10:36:41<8:33:35, 2.29s/it] 55%|█████▌ | 16516/30000 [10:36:43<8:32:06, 2.28s/it] 55%|█████▌ | 16517/30000 [10:36:45<8:34:49, 2.29s/it] 55%|█████▌ | 16518/30000 [10:36:48<8:35:29, 2.29s/it] 55%|█████▌ | 16519/30000 [10:36:50<8:30:21, 2.27s/it] 55%|█████▌ | 16520/30000 [10:36:52<8:35:32, 2.29s/it] 55%|█████▌ | 16520/30000 [10:36:52<8:35:32, 2.29s/it] 55%|█████▌ | 16521/30000 [10:36:55<8:38:24, 2.31s/it] 55%|█████▌ | 16522/30000 [10:36:57<8:36:57, 2.30s/it] 55%|█████▌ | 16523/30000 [10:36:59<8:38:24, 2.31s/it] 55%|█████▌ | 16524/30000 [10:37:02<8:37:39, 2.30s/it] 55%|█████▌ | 16525/30000 [10:37:04<8:39:01, 2.31s/it] 55%|█████▌ | 16526/30000 [10:37:06<8:38:59, 2.31s/it] 55%|█████▌ | 16527/30000 [10:37:08<8:34:11, 2.29s/it] 55%|█████▌ | 16528/30000 [10:37:11<8:34:25, 2.29s/it] 55%|█████▌ | 16529/30000 [10:37:13<8:35:05, 2.29s/it] 55%|█████▌ | 16530/30000 [10:37:15<8:36:46, 2.30s/it] 55%|█████▌ | 16530/30000 [10:37:15<8:36:46, 2.30s/it] 55%|█████▌ | 16531/30000 [10:37:18<8:34:51, 2.29s/it] 55%|█████▌ | 16532/30000 [10:37:20<8:37:07, 2.30s/it] 55%|█████▌ | 16533/30000 [10:37:22<8:36:28, 2.30s/it] 55%|█████▌ | 16534/30000 [10:37:25<8:34:21, 2.29s/it] 55%|█████▌ | 16535/30000 [10:37:27<8:32:37, 2.28s/it] 55%|█████▌ | 16536/30000 [10:37:29<8:36:36, 2.30s/it] 55%|█████▌ | 16537/30000 [10:37:31<8:39:46, 2.32s/it] 55%|█████▌ | 16538/30000 [10:37:34<8:38:18, 2.31s/it] 55%|█████▌ | 16539/30000 [10:37:36<8:38:04, 2.31s/it] 55%|█████▌ | 16540/30000 [10:37:38<8:35:58, 2.30s/it] 55%|█████▌ | 16540/30000 [10:37:38<8:35:58, 2.30s/it] 55%|█████▌ | 16541/30000 [10:37:41<8:34:26, 2.29s/it] 55%|█████▌ | 16542/30000 [10:37:43<8:31:52, 2.28s/it] 55%|█████▌ | 16543/30000 [10:37:45<8:31:50, 2.28s/it] 55%|█████▌ | 16544/30000 [10:37:47<8:34:11, 2.29s/it] 55%|█████▌ | 16545/30000 [10:37:50<8:30:54, 2.28s/it] 55%|█████▌ | 16546/30000 [10:37:52<8:29:49, 2.27s/it] 55%|█████▌ | 16547/30000 [10:37:54<8:35:31, 2.30s/it] 55%|█████▌ | 16548/30000 [10:37:57<8:35:56, 2.30s/it] 55%|█████▌ | 16549/30000 [10:37:59<8:33:41, 2.29s/it] 55%|█████▌ | 16550/30000 [10:38:01<8:32:28, 2.29s/it] +{'loss': 0.0137, 'grad_norm': 0.05918185040354729, 'learning_rate': 1.434047822523203e-05, 'flow_matching_loss': 0.013732035458087922, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.7} +{'loss': 0.0166, 'grad_norm': 0.09673801809549332, 'learning_rate': 1.4323959921224322e-05, 'flow_matching_loss': 0.016583086363971232, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.7} +{'loss': 0.0138, 'grad_norm': 0.0769217386841774, 'learning_rate': 1.4307442438669066e-05, 'flow_matching_loss': 0.013821309991180896, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.7} +{'loss': 0.0146, 'grad_norm': 0.06258105486631393, 'learning_rate': 1.4290925797636546e-05, 'flow_matching_loss': 0.014570187497884034, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.71} +{'loss': 0.0145, 'grad_norm': 0.07292283326387405, 'learning_rate': 1.4274410018196039e-05, 'flow_matching_loss': 0.014515000022947788, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.71} +{'loss': 0.0152, 'grad_norm': 0.07036242634057999, 'learning_rate': 1.4257895120415783e-05, 'flow_matching_loss': 0.015164814982563257, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.71} +{'loss': 0.0158, 'grad_norm': 0.1164296418428421, 'learning_rate': 1.4241381124362933e-05, 'flow_matching_loss': 0.015823520347476007, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.72} +{'loss': 0.0164, 'grad_norm': 0.09073644876480103, 'learning_rate': 1.4224868050103546e-05, 'flow_matching_loss': 0.016368670947849752, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.72} +{'loss': 0.0148, 'grad_norm': 0.07728391885757446, 'learning_rate': 1.4208355917702573e-05, 'flow_matching_loss': 0.014774301089346409, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.72} +{'loss': 0.0143, 'grad_norm': 0.07799604535102844, 'learning_rate': 1.41918447472238e-05, 'flow_matching_loss': 0.014340469986200333, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.73} +{'loss': 0.0148, 'grad_norm': 0.09126769751310349, 'learning_rate': 1.4175334558729869e-05, 'flow_matching_loss': 0.014757714699953794, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.73} +{'loss': 0.0148, 'grad_norm': 0.08212810009717941, 'learning_rate': 1.4158825372282211e-05, 'flow_matching_loss': 0.014838451333343983, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.73} +{'loss': 0.0143, 'grad_norm': 0.0762224942445755, 'learning_rate': 1.4142317207941042e-05, 'flow_matching_loss': 0.014319622702896596, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.73} +{'loss': 0.0147, 'grad_norm': 0.05543021485209465, 'learning_rate': 1.4125810085765339e-05, 'flow_matching_loss': 0.01468813568353653, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.74} +{'loss': 0.0165, 'grad_norm': 0.08326709270477295, 'learning_rate': 1.410930402581281e-05, 'flow_matching_loss': 0.016467265505343675, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.74} +{'loss': 0.0146, 'grad_norm': 0.10428942739963531, 'learning_rate': 1.4092799048139876e-05, 'flow_matching_loss': 0.014605036098510027, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.74} +{'loss': 0.0146, 'grad_norm': 0.07454412430524826, 'learning_rate': 1.4076295172801647e-05, 'flow_matching_loss': 0.014631668943911791, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.75} +{'loss': 0.0134, 'grad_norm': 0.08118512481451035, 'learning_rate': 1.4059792419851884e-05, 'flow_matching_loss': 0.013392662908881903, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.75} +{'loss': 0.0156, 'grad_norm': 0.06738359481096268, 'learning_rate': 1.4043290809342984e-05, 'flow_matching_loss': 0.015646727196872233, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.75} +{'loss': 0.0141, 'grad_norm': 0.06856400519609451, 'learning_rate': 1.4026790361325958e-05, 'flow_matching_loss': 0.014073270931839943, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.75} +{'loss': 0.0147, 'grad_norm': 0.08104920387268066, 'learning_rate': 1.4010291095850415e-05, 'flow_matching_loss': 0.014685491938143969, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.76} +{'loss': 0.0144, 'grad_norm': 0.06822150200605392, 'learning_rate': 1.3993793032964515e-05, 'flow_matching_loss': 0.014405467920005322, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.76} +{'loss': 0.0149, 'grad_norm': 0.10137245804071426, 'learning_rate': 1.3977296192714956e-05, 'flow_matching_loss': 0.014871872961521149, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.76} +{'loss': 0.014, 'grad_norm': 0.09212284535169601, 'learning_rate': 1.396080059514696e-05, 'flow_matching_loss': 0.013979556784033776, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.77} +{'loss': 0.0162, 'grad_norm': 0.09794077277183533, 'learning_rate': 1.3944306260304232e-05, 'flow_matching_loss': 0.0162409876473248, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.77} +{'loss': 0.0143, 'grad_norm': 0.0828828364610672, 'learning_rate': 1.3927813208228936e-05, 'flow_matching_loss': 0.014263976644724608, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.77} +{'loss': 0.0136, 'grad_norm': 0.0660286694765091, 'learning_rate': 1.39113214589617e-05, 'flow_matching_loss': 0.013555902522057295, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.77} +{'loss': 0.0133, 'grad_norm': 0.08267956227064133, 'learning_rate': 1.3894831032541545e-05, 'flow_matching_loss': 0.013314308505505324, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.78} +{'loss': 0.0139, 'grad_norm': 0.06778787076473236, 'learning_rate': 1.3878341949005895e-05, 'flow_matching_loss': 0.013946569617837667, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.78} +{'loss': 0.0149, 'grad_norm': 0.07828576117753983, 'learning_rate': 1.3861854228390546e-05, 'flow_matching_loss': 0.014925912115722894, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.78} +{'loss': 0.0147, 'grad_norm': 0.08249921351671219, 'learning_rate': 1.3845367890729621e-05, 'flow_matching_loss': 0.014729516115039587, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.79} +{'loss': 0.0153, 'grad_norm': 0.09129665791988373, 'learning_rate': 1.3828882956055591e-05, 'flow_matching_loss': 0.015345989447087049, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.79} +{'loss': 0.0136, 'grad_norm': 0.06301514059305191, 'learning_rate': 1.38123994443992e-05, 'flow_matching_loss': 0.013588733691722154, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.79} +{'loss': 0.0132, 'grad_norm': 0.08729828894138336, 'learning_rate': 1.3795917375789465e-05, 'flow_matching_loss': 0.013186279777437448, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.79} +{'loss': 0.0147, 'grad_norm': 0.07717011868953705, 'learning_rate': 1.3779436770253658e-05, 'flow_matching_loss': 0.014674048591405153, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.8} +{'loss': 0.0141, 'grad_norm': 0.07736579328775406, 'learning_rate': 1.3762957647817258e-05, 'flow_matching_loss': 0.014055029395967722, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.8} +{'loss': 0.0138, 'grad_norm': 0.0796428769826889, 'learning_rate': 1.374648002850397e-05, 'flow_matching_loss': 0.013815392181277275, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.8} +{'loss': 0.0132, 'grad_norm': 0.07976143062114716, 'learning_rate': 1.3730003932335642e-05, 'flow_matching_loss': 0.01315568359568715, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.81} +{'loss': 0.0139, 'grad_norm': 0.07132702320814133, 'learning_rate': 1.3713529379332292e-05, 'flow_matching_loss': 0.013944532535970212, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.81} +{'loss': 0.0138, 'grad_norm': 0.07411444932222366, 'learning_rate': 1.3697056389512054e-05, 'flow_matching_loss': 0.013842474948614836, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.81} +{'loss': 0.0143, 'grad_norm': 0.0819624662399292, 'learning_rate': 1.3680584982891162e-05, 'flow_matching_loss': 0.014342760108411312, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.82} 55%|█████▌ | 16550/30000 [10:38:01<8:32:28, 2.29s/it] 55%|█████▌ | 16551/30000 [10:38:03<8:31:58, 2.28s/it] 55%|█████▌ | 16552/30000 [10:38:06<8:32:38, 2.29s/it] 55%|█████▌ | 16553/30000 [10:38:08<8:32:27, 2.29s/it] 55%|█████▌ | 16554/30000 [10:38:10<8:36:54, 2.31s/it] 55%|█████▌ | 16555/30000 [10:38:13<8:38:30, 2.31s/it] 55%|█████▌ | 16556/30000 [10:38:15<8:33:59, 2.29s/it] 55%|█████▌ | 16557/30000 [10:38:17<8:31:18, 2.28s/it] 55%|█████▌ | 16558/30000 [10:38:20<8:33:25, 2.29s/it] 55%|█████▌ | 16559/30000 [10:38:22<8:29:18, 2.27s/it] 55%|█████▌ | 16560/30000 [10:38:24<8:31:56, 2.29s/it] 55%|█████▌ | 16560/30000 [10:38:24<8:31:56, 2.29s/it] 55%|█████▌ | 16561/30000 [10:38:26<8:37:52, 2.31s/it] 55%|█████▌ | 16562/30000 [10:38:29<8:36:48, 2.31s/it] 55%|█████▌ | 16563/30000 [10:38:31<8:33:52, 2.29s/it] 55%|█████▌ | 16564/30000 [10:38:33<8:31:57, 2.29s/it] 55%|█████▌ | 16565/30000 [10:38:36<8:35:21, 2.30s/it] 55%|█████▌ | 16566/30000 [10:38:38<8:37:54, 2.31s/it] 55%|█████▌ | 16567/30000 [10:38:40<8:36:55, 2.31s/it] 55%|█████▌ | 16568/30000 [10:38:43<8:38:27, 2.32s/it] 55%|█████▌ | 16569/30000 [10:38:45<8:38:27, 2.32s/it] 55%|█████▌ | 16570/30000 [10:38:47<8:34:29, 2.30s/it] 55%|█████▌ | 16570/30000 [10:38:47<8:34:29, 2.30s/it] 55%|█████▌ | 16571/30000 [10:38:49<8:33:22, 2.29s/it] 55%|█████▌ | 16572/30000 [10:38:52<8:34:45, 2.30s/it] 55%|█████▌ | 16573/30000 [10:38:54<8:29:34, 2.28s/it] 55%|█████▌ | 16574/30000 [10:38:56<8:28:57, 2.27s/it] 55%|█████▌ | 16575/30000 [10:38:59<8:33:04, 2.29s/it] 55%|█████▌ | 16576/30000 [10:39:01<8:36:51, 2.31s/it] 55%|█████▌ | 16577/30000 [10:39:03<8:34:39, 2.30s/it] 55%|█████▌ | 16578/30000 [10:39:06<8:34:46, 2.30s/it] 55%|█████▌ | 16579/30000 [10:39:08<8:36:18, 2.31s/it] 55%|█████▌ | 16580/30000 [10:39:10<8:31:22, 2.29s/it] 55%|█████▌ | 16580/30000 [10:39:10<8:31:22, 2.29s/it] 55%|█████▌ | 16581/30000 [10:39:12<8:31:46, 2.29s/it] 55%|█████▌ | 16582/30000 [10:39:15<8:33:31, 2.30s/it] 55%|█████▌ | 16583/30000 [10:39:17<8:34:03, 2.30s/it] 55%|█████▌ | 16584/30000 [10:39:19<8:33:31, 2.30s/it] 55%|█████▌ | 16585/30000 [10:39:22<8:35:54, 2.31s/it] 55%|█████▌ | 16586/30000 [10:39:24<8:36:51, 2.31s/it] 55%|█████▌ | 16587/30000 [10:39:26<8:35:37, 2.31s/it] 55%|█████▌ | 16588/30000 [10:39:29<8:33:41, 2.30s/it] 55%|█████▌ | 16589/30000 [10:39:31<8:35:59, 2.31s/it] 55%|█████▌ | 16590/30000 [10:39:33<8:35:02, 2.30s/it] 55%|█████▌ | 16590/30000 [10:39:33<8:35:02, 2.30s/it] 55%|█████▌ | 16591/30000 [10:39:35<8:31:32, 2.29s/it] 55%|█████▌ | 16592/30000 [10:39:38<8:36:43, 2.31s/it] 55%|█████▌ | 16593/30000 [10:39:40<8:38:12, 2.32s/it] 55%|█████▌ | 16594/30000 [10:39:42<8:35:56, 2.31s/it] 55%|█████▌ | 16595/30000 [10:39:45<8:36:21, 2.31s/it] 55%|█████▌ | 16596/30000 [10:39:47<8:35:27, 2.31s/it] 55%|█████▌ | 16597/30000 [10:39:49<8:32:30, 2.29s/it] 55%|█████▌ | 16598/30000 [10:39:52<8:31:33, 2.29s/it] 55%|█████▌ | 16599/30000 [10:39:54<8:36:26, 2.31s/it] 55%|█████▌ | 16600/30000 [10:39:56<8:39:59, 2.33s/it] 55%|█████▌ | 16600/30000 [10:39:56<8:39:59, 2.33s/it] 55%|█████▌ | 16601/30000 [10:39:59<8:37:50, 2.32s/it] 55%|█████▌ | 16602/30000 [10:40:01<8:38:20, 2.32s/it] 55%|█████▌ | 16603/30000 [10:40:03<8:42:14, 2.34s/it] 55%|█████▌ | 16604/30000 [10:40:06<8:42:41, 2.34s/it] 55%|█████▌ | 16605/30000 [10:40:08<8:37:29, 2.32s/it] 55%|█████▌ | 16606/30000 [10:40:10<8:33:03, 2.30s/it] 55%|█████▌ | 16607/30000 [10:40:13<8:36:08, 2.31s/it] 55%|█████▌ | 16608/30000 [10:40:15<8:34:00, 2.30s/it] 55%|█████▌ | 16609/30000 [10:40:17<8:30:06, 2.29s/it] 55%|█████▌ | 16610/30000 [10:40:19<8:33:47, 2.30s/it] 55%|█████▌ | 16610/30000 [10:40:19<8:33:47, 2.30s/it] 55%|█████▌ | 16611/30000 [10:40:22<8:31:03, 2.29s/it] 55%|█████▌ | 16612/30000 [10:40:24<8:29:37, 2.28s/it] 55%|█████▌ | 16613/30000 [10:40:26<8:34:06, 2.30s/it] 55%|█████▌ | 16614/30000 [10:40:29<8:34:21, 2.31s/it] 55%|█████▌ | 16615/30000 [10:40:31<8:31:17, 2.29s/it] 55%|█████▌ | 16616/30000 [10:40:33<8:35:07, 2.31s/it] 55%|█████▌ | 16617/30000 [10:40:36<8:36:22, 2.32s/it] 55%|█████▌ | 16618/30000 [10:40:38<8:32:05, 2.30s/it] 55%|█████▌ | 16619/30000 [10:40:40<8:32:19, 2.30s/it] 55%|█████▌ | 16620/30000 [10:40:42<8:33:33, 2.30s/it] 55%|█████▌ | 16620/30000 [10:40:42<8:33:33, 2.30s/it] 55%|█████▌ | 16621/30000 [10:40:45<8:34:33, 2.31s/it] 55%|█████▌ | 16622/30000 [10:40:47<8:31:37, 2.29s/it] 55%|█████▌ | 16623/30000 [10:40:49<8:33:18, 2.30s/it] 55%|█████▌ | 16624/30000 [10:40:52<8:33:50, 2.30s/it] 55%|█████▌ | 16625/30000 [10:40:54<8:30:41, 2.29s/it] 55%|█████▌ | 16626/30000 [10:40:56<8:28:49, 2.28s/it] 55%|█████▌ | 16627/30000 [10:40:58<8:32:50, 2.30s/it] 55%|█████▌ | 16628/30000 [10:41:01<8:29:34, 2.29s/it] 55%|█████▌ | 16629/30000 [10:41:03<8:24:48, 2.27s/it] 55%|█████▌ | 16630/30000 [10:41:05<8:27:54, 2.28s/it] 55%|█████▌ | 16630/30000 [10:41:05<8:27:54, 2.28s/it] 55%|█████▌ | 16631/30000 [10:41:07<8:26:19, 2.27s/it] 55%|█████▌ | 16632/30000 [10:41:10<8:26:48, 2.27s/it] 55%|█████▌ | 16633/30000 [10:41:12<8:32:50, 2.30s/it] 55%|█████▌ | 16634/30000 [10:41:14<8:31:20, 2.30s/it] 55%|█████▌ | 16635/30000 [10:41:17<8:32:10, 2.30s/it] 55%|█████▌ | 16636/30000 [10:41:19<8:34:45, 2.31s/it] 55%|█████▌ | 16637/30000 [10:41:21<8:34:04, 2.31s/it] 55%|█████▌ | 16638/30000 [10:41:24<8:31:21, 2.30s/it] 55%|█████▌ | 16639/30000 [10:41:26<8:34:15, 2.31s/it] 55%|█████▌ | 16640/30000 [10:41:28<8:34:43, 2.31s/it] 55%|█████▌ | 16640/30000 [10:41:28<8:34:43, 2.31s/it] 55%|█████▌ | 16641/30000 [10:41:31<8:34:49, 2.31s/it] 55%|█████▌ | 16642/30000 [10:41:33<8:32:09, 2.30s/it] 55%|█████▌ | 16643/30000 [10:41:35<8:33:52, 2.31s/it] 55%|█████▌ | 16644/30000 [10:41:37<8:28:54, 2.29s/it] 55%|█████▌ | 16645/30000 [10:41:40<8:27:56, 2.28s/it] 55%|█████▌ | 16646/30000 [10:41:42<8:32:43, 2.30s/it] 55%|█████▌ | 16647/30000 [10:41:44<8:33:45, 2.31s/it] 55%|█████▌ | 16648/30000 [10:41:47<8:29:22, 2.29s/it] 55%|█████▌ | 16649/30000 [10:41:49<8:34:43, 2.31s/it] 56%|█████▌ | 16650/30000 [10:41:51<8:35:29, 2.32s/it] 56%|█████▌ | 16650/30000 [10:41:51<8:35:29, 2.32s/it] 56%|█████▌ | 16651/30000 [10:41:54<8:34:00, 2.31s/it] 56%|█████▌ | 16652/30000 [10:41:56<8:35:19, 2.32s/it] 56%|█████▌ | 16653/30000 [10:41:58<8:31:17, 2.30s/it] 56%|█████▌ | 16654/30000 [10:42:00<8:30:26, 2.29s/it] 56%|█████▌ | 16655/30000 [10:42:03<8:30:27, 2.30s/it] 56%|█████▌ | 16656/30000 [10:42:05<8:35:03, 2.32s/it] 56%|█████▌ | 16657/30000 [10:42:07<8:30:17, 2.29s/it] 56%|█████▌ | 16658/30000 [10:42:10<8:31:05, 2.30s/it] 56%|█████▌ | 16659/30000 [10:42:12<8:35:19, 2.32s/it] 56%|█████▌ | 16660/30000 [10:42:14<8:34:00, 2.31s/it] 56%|█████▌ | 16660/30000 [10:42:14<8:34:00, 2.31s/it] 56%|█████▌ | 16661/30000 [10:42:17<8:32:05, 2.30s/it] 56%|█████▌ | 16662/30000 [10:42:19<8:34:39, 2.32s/it] 56%|█████▌ | 16663/30000 [10:42:21<8:32:02, 2.30s/it] 56%|█████▌ | 16664/30000 [10:42:24<8:29:38, 2.29s/it] 56%|█████▌ | 16665/30000 [10:42:26<8:35:44, 2.32s/it] 56%|█████▌ | 16666/30000 [10:42:28<8:36:01, 2.32s/it] 56%|█████▌ | 16667/30000 [10:42:30<8:30:52, 2.30s/it] 56%|█████▌ | 16668/30000 [10:42:33<8:30:29, 2.30s/it] 56%|█████▌ | 16669/30000 [10:42:35<8:31:20, 2.30s/it] 56%|█████▌ | 16670/30000 [10:42:37<8:29:09, 2.29s/it] 56%|█████▌ | 16670/30000 [10:42:37<8:29:09, 2.29s/it] 56%|█████▌ | 16671/30000 [10:42:40<8:29:33, 2.29s/it] 56%|█████▌ | 16672/30000 [10:42:42<8:32:38, 2.31s/it] 56%|█████▌ | 16673/30000 [10:42:44<8:30:11, 2.30s/it] 56%|█████▌ | 16674/30000 [10:42:47<8:27:29, 2.28s/it] 56%|█████▌ | 16675/30000 [10:42:49<8:33:59, 2.31s/it] 56%|█████▌ | 16676/30000 [10:42:51<8:31:23, 2.30s/it] 56%|█████▌ | 16677/30000 [10:42:53<8:29:17, 2.29s/it] 56%|█████▌ | 16678/30000 [10:42:56<8:29:23, 2.29s/it] 56%|█████▌ | 16679/30000 [10:42:58<8:26:28, 2.28s/it] 56%|█████▌ | 16680/30000 [10:43:00<8:25:58, 2.28s/it] 56%|█████▌ | 16680/30000 [10:43:00<8:25:58, 2.28s/it] 56%|█████▌ | 16681/30000 [10:43:03<8:28:37, 2.29s/it] 56%|█████▌ | 16682/30000 [10:43:05<8:29:27, 2.30s/it] 56%|█████▌ | 16683/30000 [10:43:07<8:26:55, 2.28s/it] 56%|█████▌ | 16684/30000 [10:43:09<8:23:29, 2.27s/it] 56%|█████▌ | 16685/30000 [10:43:12<8:27:18, 2.29s/it] 56%|█████▌ | 16686/30000 [10:43:14<8:25:08, 2.28s/it] 56%|█████▌ | 16687/30000 [10:43:16<8:23:12, 2.27s/it] 56%|█████▌ | 16688/30000 [10:43:19<8:29:25, 2.30s/it] 56%|█████▌ | 16689/30000 [10:43:21<8:28:36, 2.29s/it] 56%|█████▌ | 16690/30000 [10:43:23<8:26:21, 2.28s/it] 56%|█████▌ | 16690/30000 [10:43:23<8:26:21, 2.28s/it] 56%|█████▌ | 16691/30000 [10:43:25<8:25:05, 2.28s/it] 56%|█████▌ | 16692/30000 [10:43:28<8:28:31, 2.29s/it] 56%|█████▌ | 16693/30000 [10:43:30<8:27:45, 2.29s/it] 56%|█████▌ | 16694/30000 [10:43:32<8:25:42, 2.28s/it] 56%|█████▌ | 16695/30000 [10:43:35<8:26:31, 2.28s/it] 56%|█████▌ | 16696/30000 [10:43:37<8:29:16, 2.30s/it] 56%|█████▌ | 16697/30000 [10:43:39<8:28:27, 2.29s/it] 56%|█████▌ | 16698/30000 [10:43:41<8:28:00, 2.29s/it] 56%|█████▌ | 16699/30000 [10:43:44<8:31:12, 2.31s/it] 56%|█████▌ | 16700/30000 [10:43:46<8:30:49, 2.30s/it] 56%|█████▌ | 16700/30000 [10:43:46<8:30:49, 2.30s/it] 56%|█████▌ | 16701/30000 [10:43:48<8:28:15, 2.29s/it] 56%|█████▌ | 16702/30000 [10:43:51<8:27:12, 2.29s/it] 56%|█████▌ | 16703/30000 [10:43:53<8:31:26, 2.31s/it] 56%|█████▌ | 16704/30000 [10:43:55<8:25:41, 2.28s/it] 56%|█████▌ | 16705/30000 [10:43:57<8:24:17, 2.28s/it] 56%|█████▌ | 16706/30000 [10:44:00<8:26:46, 2.29s/it] 56%|█████▌ | 16707/30000 [10:44:02<8:30:19, 2.30s/it] 56%|█████▌ | 16708/30000 [10:44:04<8:25:10, 2.28s/it] 56%|█████▌ | 16709/30000 [10:44:07<8:23:46, 2.27s/it] 56%|█████▌ | 16710/30000 [10:44:09<8:27:39, 2.29s/it] 56%|█████▌ | 16710/30000 [10:44:09<8:27:39, 2.29s/it] 56%|█████▌ | 16711/30000 [10:44:11<8:26:20, 2.29s/it] 56%|█████▌ | 16712/30000 [10:44:14<8:27:18, 2.29s/it] 56%|█████▌ | 16713/30000 [10:44:16<8:27:33, 2.29s/it] 56%|█████▌ | 16714/30000 [10:44:18<8:30:26, 2.31s/it] 56%|█████▌ | 16715/30000 [10:44:20<8:31:32, 2.31s/it] 56%|█████▌ | 16716/30000 [10:44:23<8:25:31, 2.28s/it] 56%|█████▌ | 16717/30000 [10:44:25<8:23:11, 2.27s/it] 56%|█████▌ | 16718/30000 [10:44:27<8:27:52, 2.29s/it] 56%|█████▌ | 16719/30000 [10:44:30<8:25:27, 2.28s/it] 56%|█████▌ | 16720/30000 [10:44:32<8:24:12, 2.28s/it] 56%|█████▌ | 16720/30000 [10:44:32<8:24:12, 2.28s/it] 56%|█████▌ | 16721/30000 [10:44:34<8:29:07, 2.30s/it] 56%|█████▌ | 16722/30000 [10:44:37<8:30:46, 2.31s/it] 56%|█████▌ | 16723/30000 [10:44:39<8:28:32, 2.30s/it] 56%|█████▌ | 16724/30000 [10:44:41<8:25:08, 2.28s/it] 56%|█████▌ | 16725/30000 [10:44:43<8:29:19, 2.30s/it] 56%|█████▌ | 16726/30000 [10:44:46<8:30:07, 2.31s/it] 56%|█████▌ | 16727/30000 [10:44:48<8:26:27, 2.29s/it] 56%|█████▌ | 16728/30000 [10:44:50<8:27:13, 2.29s/it] 56%|█████▌ | 16729/30000 [10:44:53<8:27:03, 2.29s/it] 56%|█████▌ | 16730/30000 [10:44:55<8:25:57, 2.29s/it] 56%|█████▌ | 16730/30000 [10:44:55<8:25:57, 2.29s/it] 56%|█████▌ | 16731/30000 [10:44:57<8:29:12, 2.30s/it] 56%|█████▌ | 16732/30000 [10:44:59<8:30:00, 2.31s/it] 56%|█████▌ | 16733/30000 [10:45:02<8:30:49, 2.31s/it] 56%|█████▌ | 16734/30000 [10:45:04<8:29:03, 2.30s/it] 56%|█████▌ | 16735/30000 [10:45:06<8:29:14, 2.30s/it] 56%|█████▌ | 16736/30000 [10:45:09<8:35:01, 2.33s/it] 56%|█████▌ | 16737/30000 [10:45:11<8:32:52, 2.32s/it] 56%|█████▌ | 16738/30000 [10:45:13<8:29:02, 2.30s/it] 56%|█████▌ | 16739/30000 [10:45:16<8:29:21, 2.30s/it] 56%|█████▌ | 16740/30000 [10:45:18<8:29:31, 2.31s/it] 56%|█████▌ | 16740/30000 [10:45:18<8:29:31, 2.31s/it] 56%|█████▌ | 16741/30000 [10:45:20<8:26:20, 2.29s/it] 56%|█████▌ | 16742/30000 [10:45:22<8:21:28, 2.27s/it] 56%|█████▌ | 16743/30000 [10:45:25<8:21:55, 2.27s/it] 56%|█████▌ | 16744/30000 [10:45:27<8:24:50, 2.29s/it] 56%|█████▌ | 16745/30000 [10:45:29<8:22:30, 2.27s/it] 56%|█████▌ | 16746/30000 [10:45:32<8:22:00, 2.27s/it] 56%|█████▌ | 16747/30000 [10:45:34<8:26:01, 2.29s/it] 56%|█████▌ | 16748/30000 [10:45:36<8:27:50, 2.30s/it] 56%|█████▌ | 16749/30000 [10:45:38<8:25:12, 2.29s/it] 56%|█████▌ | 16750/30000 [10:45:41<8:21:14, 2.27s/it] 56%|█████▌ | 16750/30000 [10:45:41<8:21:14, 2.27s/it] 56%|█████▌ | 16751/30000 [10:45:43<8:25:05, 2.29s/it] 56%|█████▌ | 16752/30000 [10:45:45<8:25:06, 2.29s/it] 56%|█████▌ | 16753/30000 [10:45:48<8:23:25, 2.28s/it] 56%|█████▌ | 16754/30000 [10:45:50<8:28:20, 2.30s/it] 56%|█████▌ | 16755/30000 [10:45:52<8:29:52, 2.31s/it] 56%|█████▌ | 16756/30000 [10:45:55<8:27:40, 2.30s/it] 56%|█████▌ | 16757/30000 [10:45:57<8:26:48, 2.30s/it] 56%|█████▌ | 16758/30000 [10:45:59<8:28:01, 2.30s/it] 56%|█████▌ | 16759/30000 [10:46:01<8:28:55, 2.31s/it] 56%|█████▌ | 16760/30000 [10:46:04<8:25:41, 2.29s/it] 56%|█████▌ | 16760/30000 [10:46:04<8:25:41, 2.29s/it] 56%|█████▌ | 16761/30000 [10:46:06<8:27:33, 2.30s/it] 56%|█████▌ | 16762/30000 [10:46:08<8:30:14, 2.31s/it] 56%|█████▌ | 16763/30000 [10:46:11<8:38:14, 2.35s/it] 56%|█████▌ | 16764/30000 [10:46:13<8:31:57, 2.32s/it] 56%|█████▌ | 16765/30000 [10:46:15<8:26:42, 2.30s/it] 56%|█████▌ | 16766/30000 [10:46:18<8:27:03, 2.30s/it] 56%|█████▌ | 16767/30000 [10:46:20<8:23:08, 2.28s/it] 56%|█████▌ | 16768/30000 [10:46:22<8:24:55, 2.29s/it] 56%|█████▌ | 16769/30000 [10:46:24<8:28:55, 2.31s/it] 56%|█████▌ | 16770/30000 [10:46:27<8:30:45, 2.32s/it] 56%|█████▌ | 16770/30000 [10:46:27<8:30:45, 2.32s/it] 56%|█████▌ | 16771/30000 [10:46:29<8:32:30, 2.32s/it] 56%|█████▌ | 16772/30000 [10:46:31<8:27:38, 2.30s/it] 56%|█████▌ | 16773/30000 [10:46:34<8:29:48, 2.31s/it] 56%|█████▌ | 16774/30000 [10:46:36<8:31:48, 2.32s/it] 56%|█████▌ | 16775/30000 [10:46:38<8:27:46, 2.30s/it] 56%|█████▌ | 16776/30000 [10:46:41<8:25:28, 2.29s/it] 56%|█████▌ | 16777/30000 [10:46:43<8:29:21, 2.31s/it] 56%|█████▌ | 16778/30000 [10:46:45<8:23:18, 2.28s/it] 56%|█████▌ | 16779/30000 [10:46:47<8:20:18, 2.27s/it] 56%|█████▌ | 16780/30000 [10:46:50<8:25:32, 2.29s/it] 56%|█████▌ | 16780/30000 [10:46:50<8:25:32, 2.29s/it] 56%|█████▌ | 16781/30000 [10:46:52<8:26:39, 2.30s/it] 56%|█████▌ | 16782/30000 [10:46:54<8:25:17, 2.29s/it] 56%|█████▌ | 16783/30000 [10:46:57<8:21:26, 2.28s/it] 56%|█████▌ | 16784/30000 [10:46:59<8:25:54, 2.30s/it] 56%|█████▌ | 16785/30000 [10:47:01<8:21:52, 2.28s/it] 56%|█████▌ | 16786/30000 [10:47:03<8:22:24, 2.28s/it] 56%|█████▌ | 16787/30000 [10:47:06<8:25:37, 2.30s/it] 56%|█████▌ | 16788/30000 [10:47:08<8:25:57, 2.30s/it] 56%|█████▌ | 16789/30000 [10:47:10<8:22:34, 2.28s/it] 56%|█████▌ | 16790/30000 [10:47:13<8:27:22, 2.30s/it] 56%|█████▌ | 16790/30000 [10:47:13<8:27:22, 2.30s/it] 56%|█████▌ | 16791/30000 [10:47:15<8:23:42, 2.29s/it] 56%|█████▌ | 16792/30000 [10:47:17<8:23:16, 2.29s/it] 56%|█████▌ | 16793/30000 [10:47:20<8:23:12, 2.29s/it] 56%|█████▌ | 16794/30000 [10:47:22<8:28:09, 2.31s/it] 56%|█████▌ | 16795/30000 [10:47:24<8:31:42, 2.33s/it] 56%|█████▌ | 16796/30000 [10:47:27<8:28:36, 2.31s/it] 56%|█████▌ | 16797/30000 [10:47:29<8:30:24, 2.32s/it] 56%|█████▌ | 16798/30000 [10:47:31<8:27:51, 2.31s/it] 56%|█████▌ | 16799/30000 [10:47:33<8:25:12, 2.30s/it] 56%|█████▌ | 16800/30000 [10:47:36<8:24:22, 2.29s/it] 56%|█████▌ | 16800/30000 [10:47:36<8:24:22, 2.29s/it] 56%|█████▌ | 16801/30000 [10:47:38<8:30:13, 2.32s/it] 56%|█████▌ | 16802/30000 [10:47:40<8:33:27, 2.33s/it] 56%|█████▌ | 16803/30000 [10:47:43<8:30:15, 2.32s/it] 56%|█████▌ | 16804/30000 [10:47:45<8:29:00, 2.31s/it] 56%|█████▌ | 16805/30000 [10:47:47<8:28:58, 2.31s/it] 56%|█████▌ | 16806/30000 [10:47:50<8:24:57, 2.30s/it] 56%|█████▌ | 16807/30000 [10:47:52<8:21:34, 2.28s/it] 56%|█████▌ | 16808/30000 [10:47:54<8:27:07, 2.31s/it] 56%|█████▌ | 16809/30000 [10:47:57<8:26:54, 2.31s/it] 56%|█████▌ | 16810/30000 [10:47:59<8:25:52, 2.30s/it] 56%|█████▌ | 16810/30000 [10:47:59<8:25:52, 2.30s/it] 56%|█████▌ | 16811/30000 [10:48:01<8:28:49, 2.31s/it] 56%|█████▌ | 16812/30000 [10:48:03<8:26:32, 2.30s/it] 56%|█████▌ | 16813/30000 [10:48:06<8:25:32, 2.30s/it] 56%|█████▌ | 16814/30000 [10:48:08<8:23:43, 2.29s/it] 56%|█████▌ | 16815/30000 [10:48:10<8:25:15, 2.30s/it] 56%|█████▌ | 16816/30000 [10:48:13<8:22:15, 2.29s/it] 56%|█████▌ | 16817/30000 [10:48:15<8:19:13, 2.27s/it] 56%|█████▌ | 16818/30000 [10:48:17<8:25:33, 2.30s/it] 56%|█████▌ | 16819/30000 [10:48:20<8:28:22, 2.31s/it] 56%|█████▌ | 16820/30000 [10:48:22<8:19:03, 2.27s/it] 56%|█████▌ | 16820/30000 [10:48:22<8:19:03, 2.27s/it] 56%|█████▌ | 16821/30000 [10:48:24<8:26:25, 2.31s/it] 56%|█████▌ | 16822/30000 [10:48:26<8:27:15, 2.31s/it] 56%|█████▌ | 16823/30000 [10:48:29<8:26:17, 2.31s/it] 56%|█████▌ | 16824/30000 [10:48:31<8:28:56, 2.32s/it] 56%|█████▌ | 16825/30000 [10:48:33<8:27:59, 2.31s/it] 56%|█████▌ | 16826/30000 [10:48:36<8:23:49, 2.29s/it] 56%|█████▌ | 16827/30000 [10:48:38<8:23:18, 2.29s/it] 56%|█████▌ | 16828/30000 [10:48:40<8:24:30, 2.30s/it] 56%|█████▌ | 16829/30000 [10:48:43<8:25:17, 2.30s/it] 56%|█████▌ | 16830/30000 [10:48:45<8:19:32, 2.28s/it] 56%|█████▌ | 16830/30000 [10:48:45<8:19:32, 2.28s/it] 56%|█████▌ | 16831/30000 [10:48:47<8:26:50, 2.31s/it] 56%|█████▌ | 16832/30000 [10:48:49<8:25:15, 2.30s/it] 56%|█████▌ | 16833/30000 [10:48:52<8:23:36, 2.29s/it] 56%|█████▌ | 16834/30000 [10:48:54<8:29:05, 2.32s/it] 56%|█████▌ | 16835/30000 [10:48:56<8:29:32, 2.32s/it] 56%|█████▌ | 16836/30000 [10:48:59<8:25:02, 2.30s/it] 56%|█████▌ | 16837/30000 [10:49:01<8:27:29, 2.31s/it] 56%|█████▌ | 16838/30000 [10:49:03<8:24:56, 2.30s/it] 56%|█████▌ | 16839/30000 [10:49:06<8:22:03, 2.29s/it] 56%|█████▌ | 16840/30000 [10:49:08<8:24:24, 2.30s/it] 56%|█████▌ | 16840/30000 [10:49:08<8:24:24, 2.30s/it] 56%|█████▌ | 16841/30000 [10:49:10<8:25:07, 2.30s/it] 56%|█████▌ | 16842/30000 [10:49:12<8:22:52, 2.29s/it] 56%|█████▌ | 16843/30000 [10:49:15<8:21:38, 2.29s/it] 56%|█████▌ | 16844/30000 [10:49:17<8:24:26, 2.30s/it] 56%|█████▌ | 16845/30000 [10:49:19<8:22:02, 2.29s/it] 56%|█████▌ | 16846/30000 [10:49:22<8:20:47, 2.28s/it] 56%|█████▌ | 16847/30000 [10:49:24<8:27:32, 2.32s/it] 56%|█████▌ | 16848/30000 [10:49:26<8:27:43, 2.32s/it] 56%|█████▌ | 16849/30000 [10:49:29<8:25:10, 2.30s/it] 56%|█████▌ | 16850/30000 [10:49:31<8:25:35, 2.31s/it] 56%|█████▌ | 16850/30000 [10:49:31<8:25:35, 2.31s/it] 56%|█████▌ | 16851/30000 [10:49:33<8:27:19, 2.31s/it] 56%|█████▌ | 16852/30000 [10:49:35<8:23:39, 2.30s/it] 56%|█████▌ | 16853/30000 [10:49:38<8:22:52, 2.30s/it] 56%|█████▌ | 16854/30000 [10:49:40<8:26:26, 2.31s/it] 56%|█████▌ | 16855/30000 [10:49:42<8:25:33, 2.31s/it] 56%|█████▌ | 16856/30000 [10:49:45<8:23:00, 2.30s/it] 56%|█████▌ | 16857/30000 [10:49:47<8:24:49, 2.30s/it] 56%|█████▌ | 16858/30000 [10:49:49<8:25:39, 2.31s/it] 56%|█████▌ | 16859/30000 [10:49:52<8:20:16, 2.28s/it] 56%|█████▌ | 16860/30000 [10:49:54<8:18:45, 2.28s/it] 56%|█████▌ | 16860/30000 [10:49:54<8:18:45, 2.28s/it] 56%|█████▌ | 16861/30000 [10:49:56<8:22:58, 2.30s/it] 56%|█████▌ | 16862/30000 [10:49:58<8:19:58, 2.28s/it] 56%|█████▌ | 16863/30000 [10:50:01<8:19:24, 2.28s/it] 56%|█████▌ | 16864/30000 [10:50:03<8:24:48, 2.31s/it] 56%|█████▌ | 16865/30000 [10:50:05<8:26:26, 2.31s/it] 56%|█████▌ | 16866/30000 [10:50:08<8:23:09, 2.30s/it] 56%|█████▌ | 16867/30000 [10:50:10<8:22:59, 2.30s/it] 56%|█████▌ | 16868/30000 [10:50:12<8:22:28, 2.30s/it] 56%|█████▌ | 16869/30000 [10:50:14<8:20:02, 2.28s/it] 56%|█████▌ | 16870/30000 [10:50:17<8:17:14, 2.27s/it] 56%|█████▌ | 16870/30000 [10:50:17<8:17:14, 2.27s/it] 56%|█████▌ | 16871/30000 [10:50:19<8:21:48, 2.29s/it] 56%|█████▌ | 16872/30000 [10:50:21<8:24:13, 2.30s/it] 56%|█████▌ | 16873/30000 [10:50:24<8:22:37, 2.30s/it] 56%|█████▌ | 16874/30000 [10:50:26<8:22:14, 2.30s/it] 56%|█████▋ | 16875/30000 [10:50:28<8:24:03, 2.30s/it] 56%|█████▋ | 16876/30000 [10:50:31<8:22:32, 2.30s/it] 56%|█████▋ | 16877/30000 [10:50:33<8:19:44, 2.28s/it] 56%|█████▋ | 16878/30000 [10:50:35<8:19:11, 2.28s/it] 56%|█████▋ | 16879/30000 [10:50:37<8:21:24, 2.29s/it] 56%|█████▋ | 16880/30000 [10:50:40<8:21:01, 2.29s/it] 56%|█████▋ | 16880/30000 [10:50:40<8:21:01, 2.29s/it] 56%|█████▋ | 16881/30000 [10:50:42<8:21:11, 2.29s/it] 56%|█████▋ | 16882/30000 [10:50:44<8:23:43, 2.30s/it] 56%|█████▋ | 16883/30000 [10:50:47<8:21:12, 2.29s/it] 56%|█████▋ | 16884/30000 [10:50:49<8:17:21, 2.28s/it] 56%|█████▋ | 16885/30000 [10:50:51<8:15:50, 2.27s/it] 56%|█████▋ | 16886/30000 [10:50:53<8:21:25, 2.29s/it] 56%|█████▋ | 16887/30000 [10:50:56<8:18:08, 2.28s/it] 56%|█████▋ | 16888/30000 [10:50:58<8:19:28, 2.29s/it] 56%|█████▋ | 16889/30000 [10:51:00<8:20:21, 2.29s/it] 56%|█████▋ | 16890/30000 [10:51:03<8:24:48, 2.31s/it] 56%|█████▋ | 16890/30000 [10:51:03<8:24:48, 2.31s/it] 56%|█████▋ | 16891/30000 [10:51:05<8:27:27, 2.32s/it] 56%|█████▋ | 16892/30000 [10:51:07<8:24:09, 2.31s/it] 56%|█████▋ | 16893/30000 [10:51:10<8:20:41, 2.29s/it] 56%|█████▋ | 16894/30000 [10:51:12<8:25:17, 2.31s/it] 56%|█████▋ | 16895/30000 [10:51:14<8:24:00, 2.31s/it] 56%|█████▋ | 16896/30000 [10:51:16<8:22:20, 2.30s/it] 56%|█████▋ | 16897/30000 [10:51:19<8:23:43, 2.31s/it] 56%|█████▋ | 16898/30000 [10:51:21<8:28:31, 2.33s/it] 56%|█████▋ | 16899/30000 [10:51:23<8:28:05, 2.33s/it] 56%|█████▋ | 16900/30000 [10:51:26<8:21:42, 2.30s/it] 56%|█████▋ | 16900/30000 [10:51:26<8:21:42, 2.30s/it] 56%|█████▋ | 16901/30000 [10:51:28<8:21:59, 2.30s/it] 56%|█████▋ | 16902/30000 [10:51:30<8:21:22, 2.30s/it] 56%|█████▋ | 16903/30000 [10:51:33<8:19:26, 2.29s/it] 56%|█████▋ | 16904/30000 [10:51:35<8:17:58, 2.28s/it] 56%|█████▋ | 16905/30000 [10:51:37<8:23:00, 2.30s/it] 56%|█████▋ | 16906/30000 [10:51:40<8:24:07, 2.31s/it] 56%|█████▋ | 16907/30000 [10:51:42<8:23:33, 2.31s/it] 56%|█████▋ | 16908/30000 [10:51:44<8:20:51, 2.30s/it] 56%|█████▋ | 16909/30000 [10:51:46<8:23:52, 2.31s/it] 56%|█████▋ | 16910/30000 [10:51:49<8:21:37, 2.30s/it] 56%|█████▋ | 16910/30000 [10:51:49<8:21:37, 2.30s/it] 56%|█████▋ | 16911/30000 [10:51:51<8:19:51, 2.29s/it] 56%|█████▋ | 16912/30000 [10:51:53<8:17:40, 2.28s/it] 56%|█████▋ | 16913/30000 [10:51:56<8:21:29, 2.30s/it] 56%|█████▋ | 16914/30000 [10:51:58<8:23:50, 2.31s/it] 56%|█████▋ | 16915/30000 [10:52:00<8:23:35, 2.31s/it] 56%|█████▋ | 16916/30000 [10:52:02<8:21:31, 2.30s/it] 56%|█████▋ | 16917/30000 [10:52:05<8:20:19, 2.29s/it] 56%|█████▋ | 16918/30000 [10:52:07<8:18:53, 2.29s/it] 56%|█████▋ | 16919/30000 [10:52:09<8:15:14, 2.27s/it] 56%|█████▋ | 16920/30000 [10:52:12<8:16:06, 2.28s/it] 56%|█████▋ | 16920/30000 [10:52:12<8:16:06, 2.28s/it] 56%|█████▋ | 16921/30000 [10:52:14<8:21:05, 2.30s/it] 56%|█████▋ | 16922/30000 [10:52:16<8:22:53, 2.31s/it] 56%|█████▋ | 16923/30000 [10:52:19<8:20:06, 2.29s/it] 56%|█████▋ | 16924/30000 [10:52:21<8:19:51, 2.29s/it] 56%|█████▋ | 16925/30000 [10:52:23<8:16:28, 2.28s/it] 56%|█████▋ | 16926/30000 [10:52:25<8:16:22, 2.28s/it] 56%|█████▋ | 16927/30000 [10:52:28<8:13:29, 2.26s/it] 56%|█████▋ | 16928/30000 [10:52:30<8:17:19, 2.28s/it] 56%|█████▋ | 16929/30000 [10:52:32<8:19:25, 2.29s/it] 56%|█████▋ | 16930/30000 [10:52:34<8:17:35, 2.28s/it] 56%|█████▋ | 16930/30000 [10:52:35<8:17:35, 2.28s/it] 56%|█████▋ | 16931/30000 [10:52:37<8:22:57, 2.31s/it] 56%|█████▋ | 16932/30000 [10:52:39<8:20:11, 2.30s/it] 56%|█████▋ | 16933/30000 [10:52:41<8:17:43, 2.29s/it] 56%|█████▋ | 16934/30000 [10:52:44<8:17:21, 2.28s/it] 56%|█████▋ | 16935/30000 [10:52:46<8:17:58, 2.29s/it] 56%|█████▋ | 16936/30000 [10:52:48<8:16:53, 2.28s/it] 56%|█████▋ | 16937/30000 [10:52:50<8:15:18, 2.28s/it] 56%|█████▋ | 16938/30000 [10:52:53<8:20:34, 2.30s/it] 56%|█████▋ | 16939/30000 [10:52:55<8:23:13, 2.31s/it] 56%|█████▋ | 16940/30000 [10:52:57<8:17:58, 2.29s/it] 56%|█████▋ | 16940/30000 [10:52:57<8:17:58, 2.29s/it] 56%|█████▋ | 16941/30000 [10:53:00<8:21:04, 2.30s/it] 56%|█████▋ | 16942/30000 [10:53:02<8:18:54, 2.29s/it] 56%|█████▋ | 16943/30000 [10:53:04<8:17:59, 2.29s/it] 56%|█████▋ | 16944/30000 [10:53:07<8:22:06, 2.31s/it] 56%|█████▋ | 16945/30000 [10:53:09<8:21:02, 2.30s/it] 56%|█████▋ | 16946/30000 [10:53:11<8:17:30, 2.29s/it] 56%|█████▋ | 16947/30000 [10:53:13<8:17:20, 2.29s/it] 56%|█████▋ | 16948/30000 [10:53:16<8:17:05, 2.29s/it] 56%|█████▋ | 16949/30000 [10:53:18<8:16:30, 2.28s/it] 56%|█████▋ | 16950/30000 [10:53:20<8:13:33, 2.27s/it] 56%|█████▋ | 16950/30000 [10:53:20<8:13:33, 2.27s/it] 57%|█████▋ | 16951/30000 [10:53:23<8:16:43, 2.28s/it] 57%|█████▋ | 16952/30000 [10:53:25<8:15:38, 2.28s/it] 57%|█████▋ | 16953/30000 [10:53:27<8:16:24, 2.28s/it] 57%|█████▋ | 16954/30000 [10:53:29<8:20:52, 2.30s/it] 57%|█████▋ | 16955/30000 [10:53:32<8:23:35, 2.32s/it] 57%|█████▋ | 16956/30000 [10:53:34<8:17:03, 2.29s/it] 57%|█████▋ | 16957/30000 [10:53:36<8:21:59, 2.31s/it] 57%|█████▋ | 16958/30000 [10:53:39<8:24:03, 2.32s/it] 57%|█████▋ | 16959/30000 [10:53:41<8:22:55, 2.31s/it] 57%|█████▋ | 16960/30000 [10:53:43<8:25:07, 2.32s/it] +{'loss': 0.0128, 'grad_norm': 0.07260286062955856, 'learning_rate': 1.3664115179483921e-05, 'flow_matching_loss': 0.012819636892527343, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.82} +{'loss': 0.0137, 'grad_norm': 0.07077622413635254, 'learning_rate': 1.364764699930271e-05, 'flow_matching_loss': 0.013659362960606814, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.82} +{'loss': 0.0127, 'grad_norm': 0.06072103604674339, 'learning_rate': 1.363118046235791e-05, 'flow_matching_loss': 0.012741822376847267, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.82} +{'loss': 0.0136, 'grad_norm': 0.06319291144609451, 'learning_rate': 1.3614715588657921e-05, 'flow_matching_loss': 0.013578874990344048, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.83} +{'loss': 0.014, 'grad_norm': 0.0997951552271843, 'learning_rate': 1.3598252398209118e-05, 'flow_matching_loss': 0.014027332328259945, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.83} +{'loss': 0.0159, 'grad_norm': 0.06984248012304306, 'learning_rate': 1.358179091101582e-05, 'flow_matching_loss': 0.015883042383939026, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.83} +{'loss': 0.0146, 'grad_norm': 0.06538951396942139, 'learning_rate': 1.3565331147080299e-05, 'flow_matching_loss': 0.014560313243418933, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.84} +{'loss': 0.0164, 'grad_norm': 0.07660477608442307, 'learning_rate': 1.3548873126402716e-05, 'flow_matching_loss': 0.016424162313342094, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.84} +{'loss': 0.0123, 'grad_norm': 0.06121942773461342, 'learning_rate': 1.3532416868981117e-05, 'flow_matching_loss': 0.012269951775670052, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.84} +{'loss': 0.0151, 'grad_norm': 0.08641815185546875, 'learning_rate': 1.3515962394811401e-05, 'flow_matching_loss': 0.015092259366065263, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.84} +{'loss': 0.0152, 'grad_norm': 0.10017720609903336, 'learning_rate': 1.3499509723887313e-05, 'flow_matching_loss': 0.01519012525677681, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.85} +{'loss': 0.0142, 'grad_norm': 0.08840885013341904, 'learning_rate': 1.3483058876200396e-05, 'flow_matching_loss': 0.014194582030177117, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.85} +{'loss': 0.0142, 'grad_norm': 0.06737853586673737, 'learning_rate': 1.3466609871739987e-05, 'flow_matching_loss': 0.014235251024365425, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.85} +{'loss': 0.0143, 'grad_norm': 0.0949140414595604, 'learning_rate': 1.3450162730493172e-05, 'flow_matching_loss': 0.014257429353892802, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.86} +{'loss': 0.0126, 'grad_norm': 0.08374945819377899, 'learning_rate': 1.3433717472444781e-05, 'flow_matching_loss': 0.012631885055452586, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.86} +{'loss': 0.0139, 'grad_norm': 0.07398287951946259, 'learning_rate': 1.3417274117577343e-05, 'flow_matching_loss': 0.013917269837111234, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.86} +{'loss': 0.0149, 'grad_norm': 0.08838286995887756, 'learning_rate': 1.3400832685871102e-05, 'flow_matching_loss': 0.014922299329191447, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.86} +{'loss': 0.0154, 'grad_norm': 0.08320791274309158, 'learning_rate': 1.3384393197303936e-05, 'flow_matching_loss': 0.015351281687617301, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.87} +{'loss': 0.0144, 'grad_norm': 0.08866966515779495, 'learning_rate': 1.336795567185138e-05, 'flow_matching_loss': 0.014377421233803034, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.87} +{'loss': 0.0145, 'grad_norm': 0.0696142315864563, 'learning_rate': 1.3351520129486573e-05, 'flow_matching_loss': 0.014532910566776991, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.87} +{'loss': 0.015, 'grad_norm': 0.08698644489049911, 'learning_rate': 1.3335086590180252e-05, 'flow_matching_loss': 0.014980449620634318, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.88} +{'loss': 0.0151, 'grad_norm': 0.0822606235742569, 'learning_rate': 1.3318655073900708e-05, 'flow_matching_loss': 0.015058276988565922, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.88} +{'loss': 0.0139, 'grad_norm': 0.07582014799118042, 'learning_rate': 1.3302225600613796e-05, 'flow_matching_loss': 0.013919273298233748, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.88} +{'loss': 0.0134, 'grad_norm': 0.07408515363931656, 'learning_rate': 1.3285798190282869e-05, 'flow_matching_loss': 0.01336130676791072, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.89} +{'loss': 0.0137, 'grad_norm': 0.06974935531616211, 'learning_rate': 1.3269372862868776e-05, 'flow_matching_loss': 0.013689538650214671, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.89} +{'loss': 0.0142, 'grad_norm': 0.0813729390501976, 'learning_rate': 1.3252949638329845e-05, 'flow_matching_loss': 0.014184916391968727, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.89} +{'loss': 0.0155, 'grad_norm': 0.08019409328699112, 'learning_rate': 1.3236528536621834e-05, 'flow_matching_loss': 0.015482174325734377, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.89} +{'loss': 0.0146, 'grad_norm': 0.07172039151191711, 'learning_rate': 1.3220109577697943e-05, 'flow_matching_loss': 0.014624634850770235, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.9} +{'loss': 0.0146, 'grad_norm': 0.08011574298143387, 'learning_rate': 1.3203692781508745e-05, 'flow_matching_loss': 0.014635954238474368, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.9} +{'loss': 0.0147, 'grad_norm': 0.09822243452072144, 'learning_rate': 1.3187278168002198e-05, 'flow_matching_loss': 0.014664537832140922, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.9} +{'loss': 0.0145, 'grad_norm': 0.09535945951938629, 'learning_rate': 1.3170865757123606e-05, 'flow_matching_loss': 0.014527781028300523, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.91} +{'loss': 0.0137, 'grad_norm': 0.08278372138738632, 'learning_rate': 1.3154455568815592e-05, 'flow_matching_loss': 0.013727824855595826, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.91} +{'loss': 0.0146, 'grad_norm': 0.08199728280305862, 'learning_rate': 1.3138047623018087e-05, 'flow_matching_loss': 0.014565099030733109, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.91} +{'loss': 0.0143, 'grad_norm': 0.09505245089530945, 'learning_rate': 1.3121641939668293e-05, 'flow_matching_loss': 0.014319510199129581, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.91} +{'loss': 0.0144, 'grad_norm': 0.09958186745643616, 'learning_rate': 1.3105238538700659e-05, 'flow_matching_loss': 0.01437838664278388, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.92} +{'loss': 0.0135, 'grad_norm': 0.08008596301078796, 'learning_rate': 1.3088837440046866e-05, 'flow_matching_loss': 0.013476345408707856, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.92} +{'loss': 0.0143, 'grad_norm': 0.07631220668554306, 'learning_rate': 1.3072438663635787e-05, 'flow_matching_loss': 0.01429281458258629, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.92} +{'loss': 0.0146, 'grad_norm': 0.08522520959377289, 'learning_rate': 1.3056042229393497e-05, 'flow_matching_loss': 0.01456923158839345, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.93} +{'loss': 0.0147, 'grad_norm': 0.08634315431118011, 'learning_rate': 1.30396481572432e-05, 'flow_matching_loss': 0.014662631135433913, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.93} +{'loss': 0.0173, 'grad_norm': 0.07365034520626068, 'learning_rate': 1.3023256467105237e-05, 'flow_matching_loss': 0.017257215082645418, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.93} +{'loss': 0.0132, 'grad_norm': 0.05530531331896782, 'learning_rate': 1.3006867178897067e-05, 'flow_matching_loss': 0.013243357837200164, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.93} 57%|█████▋ | 16960/30000 [10:53:43<8:25:07, 2.32s/it] 57%|█████▋ | 16961/30000 [10:53:46<8:21:23, 2.31s/it] 57%|█████▋ | 16962/30000 [10:53:48<8:20:00, 2.30s/it] 57%|█████▋ | 16963/30000 [10:53:50<8:21:24, 2.31s/it] 57%|█████▋ | 16964/30000 [10:53:53<8:22:39, 2.31s/it] 57%|█████▋ | 16965/30000 [10:53:55<8:18:39, 2.30s/it] 57%|█████▋ | 16966/30000 [10:53:57<8:16:18, 2.28s/it] 57%|█████▋ | 16967/30000 [10:53:59<8:20:42, 2.31s/it] 57%|█████▋ | 16968/30000 [10:54:02<8:19:29, 2.30s/it] 57%|█████▋ | 16969/30000 [10:54:04<8:19:06, 2.30s/it] 57%|█████▋ | 16970/30000 [10:54:06<8:23:19, 2.32s/it] 57%|█████▋ | 16970/30000 [10:54:06<8:23:19, 2.32s/it] 57%|█████▋ | 16971/30000 [10:54:09<8:25:10, 2.33s/it] 57%|█████▋ | 16972/30000 [10:54:11<8:19:26, 2.30s/it] 57%|█████▋ | 16973/30000 [10:54:13<8:19:22, 2.30s/it] 57%|█████▋ | 16974/30000 [10:54:16<8:19:50, 2.30s/it] 57%|█████▋ | 16975/30000 [10:54:18<8:18:31, 2.30s/it] 57%|█████▋ | 16976/30000 [10:54:20<8:17:05, 2.29s/it] 57%|█████▋ | 16977/30000 [10:54:23<8:20:34, 2.31s/it] 57%|█████▋ | 16978/30000 [10:54:25<8:17:58, 2.29s/it] 57%|█████▋ | 16979/30000 [10:54:27<8:14:52, 2.28s/it] 57%|█████▋ | 16980/30000 [10:54:29<8:19:40, 2.30s/it] 57%|█████▋ | 16980/30000 [10:54:29<8:19:40, 2.30s/it] 57%|█████▋ | 16981/30000 [10:54:32<8:22:27, 2.32s/it] 57%|█████▋ | 16982/30000 [10:54:34<8:21:21, 2.31s/it] 57%|█████▋ | 16983/30000 [10:54:36<8:22:02, 2.31s/it] 57%|█████▋ | 16984/30000 [10:54:39<8:20:32, 2.31s/it] 57%|█████▋ | 16985/30000 [10:54:41<8:18:22, 2.30s/it] 57%|█████▋ | 16986/30000 [10:54:43<8:17:54, 2.30s/it] 57%|█████▋ | 16987/30000 [10:54:46<8:24:10, 2.32s/it] 57%|█████▋ | 16988/30000 [10:54:48<8:23:05, 2.32s/it] 57%|█████▋ | 16989/30000 [10:54:50<8:18:58, 2.30s/it] 57%|█████▋ | 16990/30000 [10:54:53<8:22:06, 2.32s/it] 57%|█████▋ | 16990/30000 [10:54:53<8:22:06, 2.32s/it] 57%|█████▋ | 16991/30000 [10:54:55<8:22:00, 2.32s/it] 57%|█████▋ | 16992/30000 [10:54:57<8:19:30, 2.30s/it] 57%|█████▋ | 16993/30000 [10:54:59<8:20:02, 2.31s/it] 57%|█████▋ | 16994/30000 [10:55:02<8:19:14, 2.30s/it] 57%|█████▋ | 16995/30000 [10:55:04<8:17:45, 2.30s/it] 57%|█████▋ | 16996/30000 [10:55:06<8:14:49, 2.28s/it] 57%|█████▋ | 16997/30000 [10:55:09<8:17:58, 2.30s/it] 57%|█████▋ | 16998/30000 [10:55:11<8:14:52, 2.28s/it] 57%|█████▋ | 16999/30000 [10:55:13<8:13:16, 2.28s/it] 57%|█████▋ | 17000/30000 [10:55:15<8:20:11, 2.31s/it] 57%|█████▋ | 17000/30000 [10:55:16<8:20:11, 2.31s/it] 57%|█████▋ | 17001/30000 [10:55:18<8:21:53, 2.32s/it] 57%|█████▋ | 17002/30000 [10:55:20<8:19:21, 2.31s/it] 57%|█████▋ | 17003/30000 [10:55:22<8:16:09, 2.29s/it] 57%|█████▋ | 17004/30000 [10:55:25<8:20:56, 2.31s/it] 57%|█████▋ | 17005/30000 [10:55:27<8:15:44, 2.29s/it] 57%|█████▋ | 17006/30000 [10:55:29<8:13:10, 2.28s/it] 57%|█████▋ | 17007/30000 [10:55:32<8:17:24, 2.30s/it] 57%|█████▋ | 17008/30000 [10:55:34<8:18:41, 2.30s/it] 57%|█████▋ | 17009/30000 [10:55:36<8:19:10, 2.31s/it] 57%|█████▋ | 17010/30000 [10:55:38<8:19:25, 2.31s/it] 57%|█████▋ | 17010/30000 [10:55:38<8:19:25, 2.31s/it] 57%|█████▋ | 17011/30000 [10:55:41<8:19:37, 2.31s/it] 57%|█████▋ | 17012/30000 [10:55:43<8:16:35, 2.29s/it] 57%|█████▋ | 17013/30000 [10:55:45<8:14:02, 2.28s/it] 57%|█████▋ | 17014/30000 [10:55:48<8:19:21, 2.31s/it] 57%|█████▋ | 17015/30000 [10:55:50<8:18:27, 2.30s/it] 57%|█████▋ | 17016/30000 [10:55:52<8:17:18, 2.30s/it] 57%|█████▋ | 17017/30000 [10:55:55<8:15:14, 2.29s/it] 57%|█████▋ | 17018/30000 [10:55:57<8:20:09, 2.31s/it] 57%|█████▋ | 17019/30000 [10:55:59<8:20:13, 2.31s/it] 57%|█████▋ | 17020/30000 [10:56:01<8:15:41, 2.29s/it] 57%|█████▋ | 17020/30000 [10:56:01<8:15:41, 2.29s/it] 57%|█████▋ | 17021/30000 [10:56:04<8:14:43, 2.29s/it] 57%|█████▋ | 17022/30000 [10:56:06<8:17:14, 2.30s/it] 57%|█████▋ | 17023/30000 [10:56:08<8:16:00, 2.29s/it] 57%|█████▋ | 17024/30000 [10:56:11<8:13:48, 2.28s/it] 57%|█████▋ | 17025/30000 [10:56:13<8:14:53, 2.29s/it] 57%|█████▋ | 17026/30000 [10:56:15<8:15:06, 2.29s/it] 57%|█████▋ | 17027/30000 [10:56:17<8:14:46, 2.29s/it] 57%|█████▋ | 17028/30000 [10:56:20<8:13:12, 2.28s/it] 57%|█████▋ | 17029/30000 [10:56:22<8:17:57, 2.30s/it] 57%|█████▋ | 17030/30000 [10:56:24<8:18:39, 2.31s/it] 57%|█████▋ | 17030/30000 [10:56:24<8:18:39, 2.31s/it] 57%|█████▋ | 17031/30000 [10:56:27<8:15:31, 2.29s/it] 57%|█████▋ | 17032/30000 [10:56:29<8:15:59, 2.29s/it] 57%|█████▋ | 17033/30000 [10:56:31<8:18:52, 2.31s/it] 57%|█████▋ | 17034/30000 [10:56:34<8:17:23, 2.30s/it] 57%|█████▋ | 17035/30000 [10:56:36<8:18:20, 2.31s/it] 57%|█████▋ | 17036/30000 [10:56:38<8:18:45, 2.31s/it] 57%|█████▋ | 17037/30000 [10:56:40<8:17:24, 2.30s/it] 57%|█████▋ | 17038/30000 [10:56:43<8:13:41, 2.29s/it] 57%|█████▋ | 17039/30000 [10:56:45<8:09:43, 2.27s/it] 57%|█████▋ | 17040/30000 [10:56:47<8:15:04, 2.29s/it] 57%|█████▋ | 17040/30000 [10:56:47<8:15:04, 2.29s/it] 57%|█████▋ | 17041/30000 [10:56:50<8:17:28, 2.30s/it] 57%|█████▋ | 17042/30000 [10:56:52<8:15:46, 2.30s/it] 57%|█████▋ | 17043/30000 [10:56:54<8:15:49, 2.30s/it] 57%|█████▋ | 17044/30000 [10:56:57<8:18:08, 2.31s/it] 57%|█████▋ | 17045/30000 [10:56:59<8:15:07, 2.29s/it] 57%|█████▋ | 17046/30000 [10:57:01<8:13:59, 2.29s/it] 57%|█████▋ | 17047/30000 [10:57:03<8:14:37, 2.29s/it] 57%|█████▋ | 17048/30000 [10:57:06<8:18:13, 2.31s/it] 57%|█████▋ | 17049/30000 [10:57:08<8:20:00, 2.32s/it] 57%|█████▋ | 17050/30000 [10:57:10<8:19:08, 2.31s/it] 57%|█████▋ | 17050/30000 [10:57:10<8:19:08, 2.31s/it] 57%|█████▋ | 17051/30000 [10:57:13<8:16:40, 2.30s/it] 57%|█████▋ | 17052/30000 [10:57:15<8:17:52, 2.31s/it] 57%|█████▋ | 17053/30000 [10:57:17<8:18:15, 2.31s/it] 57%|█████▋ | 17054/30000 [10:57:20<8:14:05, 2.29s/it] 57%|█████▋ | 17055/30000 [10:57:22<8:14:48, 2.29s/it] 57%|█████▋ | 17056/30000 [10:57:24<8:16:48, 2.30s/it] 57%|█████▋ | 17057/30000 [10:57:26<8:13:47, 2.29s/it] 57%|█████▋ | 17058/30000 [10:57:29<8:11:29, 2.28s/it] 57%|█████▋ | 17059/30000 [10:57:31<8:16:11, 2.30s/it] 57%|█████▋ | 17060/30000 [10:57:33<8:16:09, 2.30s/it] 57%|█████▋ | 17060/30000 [10:57:33<8:16:09, 2.30s/it] 57%|█████▋ | 17061/30000 [10:57:36<8:12:10, 2.28s/it] 57%|█████▋ | 17062/30000 [10:57:38<8:12:50, 2.29s/it] 57%|█████▋ | 17063/30000 [10:57:40<8:14:19, 2.29s/it] 57%|█████▋ | 17064/30000 [10:57:42<8:11:00, 2.28s/it] 57%|█████▋ | 17065/30000 [10:57:45<8:10:03, 2.27s/it] 57%|█████▋ | 17066/30000 [10:57:47<8:15:58, 2.30s/it] 57%|█████▋ | 17067/30000 [10:57:49<8:19:25, 2.32s/it] 57%|█████▋ | 17068/30000 [10:57:52<8:15:03, 2.30s/it] 57%|█████▋ | 17069/30000 [10:57:54<8:14:49, 2.30s/it] 57%|█████▋ | 17070/30000 [10:57:56<8:17:36, 2.31s/it] 57%|█████▋ | 17070/30000 [10:57:56<8:17:36, 2.31s/it] 57%|█████▋ | 17071/30000 [10:57:59<8:18:18, 2.31s/it] 57%|█████▋ | 17072/30000 [10:58:01<8:13:37, 2.29s/it] 57%|█████▋ | 17073/30000 [10:58:03<8:12:36, 2.29s/it] 57%|█████▋ | 17074/30000 [10:58:05<8:16:12, 2.30s/it] 57%|█████▋ | 17075/30000 [10:58:08<8:18:54, 2.32s/it] 57%|█████▋ | 17076/30000 [10:58:10<8:16:07, 2.30s/it] 57%|█████▋ | 17077/30000 [10:58:12<8:18:36, 2.31s/it] 57%|█████▋ | 17078/30000 [10:58:15<8:18:51, 2.32s/it] 57%|█████▋ | 17079/30000 [10:58:17<8:12:22, 2.29s/it] 57%|█████▋ | 17080/30000 [10:58:19<8:11:07, 2.28s/it] 57%|█████▋ | 17080/30000 [10:58:19<8:11:07, 2.28s/it] 57%|█████▋ | 17081/30000 [10:58:22<8:17:10, 2.31s/it] 57%|█████▋ | 17082/30000 [10:58:24<8:14:48, 2.30s/it] 57%|█████▋ | 17083/30000 [10:58:26<8:13:46, 2.29s/it] 57%|█████▋ | 17084/30000 [10:58:29<8:18:25, 2.32s/it] 57%|█████▋ | 17085/30000 [10:58:31<8:16:55, 2.31s/it] 57%|█████▋ | 17086/30000 [10:58:33<8:15:43, 2.30s/it] 57%|█████▋ | 17087/30000 [10:58:35<8:14:22, 2.30s/it] 57%|█████▋ | 17088/30000 [10:58:38<8:19:46, 2.32s/it] 57%|█████▋ | 17089/30000 [10:58:40<8:21:12, 2.33s/it] 57%|█████▋ | 17090/30000 [10:58:42<8:18:11, 2.32s/it] 57%|█████▋ | 17090/30000 [10:58:42<8:18:11, 2.32s/it] 57%|█████▋ | 17091/30000 [10:58:45<8:14:03, 2.30s/it] 57%|█████▋ | 17092/30000 [10:58:47<8:18:07, 2.32s/it] 57%|█████▋ | 17093/30000 [10:58:49<8:18:23, 2.32s/it] 57%|█████▋ | 17094/30000 [10:58:52<8:13:25, 2.29s/it] 57%|█████▋ | 17095/30000 [10:58:54<8:14:30, 2.30s/it] 57%|█████▋ | 17096/30000 [10:58:56<8:13:26, 2.29s/it] 57%|█████▋ | 17097/30000 [10:58:58<8:14:25, 2.30s/it] 57%|█████▋ | 17098/30000 [10:59:01<8:13:23, 2.29s/it] 57%|█████▋ | 17099/30000 [10:59:03<8:18:20, 2.32s/it] 57%|█████▋ | 17100/30000 [10:59:05<8:17:15, 2.31s/it] 57%|█████▋ | 17100/30000 [10:59:05<8:17:15, 2.31s/it] 57%|█████▋ | 17101/30000 [10:59:08<8:15:40, 2.31s/it] 57%|█████▋ | 17102/30000 [10:59:10<8:13:36, 2.30s/it] 57%|█████▋ | 17103/30000 [10:59:12<8:18:11, 2.32s/it] 57%|█████▋ | 17104/30000 [10:59:15<8:15:44, 2.31s/it] 57%|█████▋ | 17105/30000 [10:59:17<8:11:32, 2.29s/it] 57%|█████▋ | 17106/30000 [10:59:19<8:16:03, 2.31s/it] 57%|█████▋ | 17107/30000 [10:59:22<8:16:07, 2.31s/it] 57%|█████▋ | 17108/30000 [10:59:24<8:12:54, 2.29s/it] 57%|█████▋ | 17109/30000 [10:59:26<8:12:13, 2.29s/it] 57%|█████▋ | 17110/30000 [10:59:28<8:14:51, 2.30s/it] 57%|█████▋ | 17110/30000 [10:59:28<8:14:51, 2.30s/it] 57%|█████▋ | 17111/30000 [10:59:31<8:16:08, 2.31s/it] 57%|█████▋ | 17112/30000 [10:59:33<8:16:14, 2.31s/it] 57%|█████▋ | 17113/30000 [10:59:35<8:16:15, 2.31s/it] 57%|█████▋ | 17114/30000 [10:59:38<8:19:51, 2.33s/it] 57%|█████▋ | 17115/30000 [10:59:40<8:20:38, 2.33s/it] 57%|█████▋ | 17116/30000 [10:59:42<8:16:09, 2.31s/it] 57%|█████▋ | 17117/30000 [10:59:45<8:16:56, 2.31s/it] 57%|█████▋ | 17118/30000 [10:59:47<8:14:58, 2.31s/it] 57%|█████▋ | 17119/30000 [10:59:49<8:11:33, 2.29s/it] 57%|█████▋ | 17120/30000 [10:59:51<8:11:27, 2.29s/it] 57%|█████▋ | 17120/30000 [10:59:52<8:11:27, 2.29s/it] 57%|█████▋ | 17121/30000 [10:59:54<8:18:24, 2.32s/it] 57%|█████▋ | 17122/30000 [10:59:56<8:22:35, 2.34s/it] 57%|█████▋ | 17123/30000 [10:59:59<8:20:26, 2.33s/it] 57%|█████▋ | 17124/30000 [11:00:01<8:15:27, 2.31s/it] 57%|█████▋ | 17125/30000 [11:00:03<8:17:03, 2.32s/it] 57%|█████▋ | 17126/30000 [11:00:05<8:15:41, 2.31s/it] 57%|█████▋ | 17127/30000 [11:00:08<8:12:27, 2.30s/it] 57%|█████▋ | 17128/30000 [11:00:10<8:10:16, 2.29s/it] 57%|█████▋ | 17129/30000 [11:00:12<8:12:46, 2.30s/it] 57%|█████▋ | 17130/30000 [11:00:15<8:08:56, 2.28s/it] 57%|█████▋ | 17130/30000 [11:00:15<8:08:56, 2.28s/it] 57%|█████▋ | 17131/30000 [11:00:17<8:08:21, 2.28s/it] 57%|█████▋ | 17132/30000 [11:00:19<8:12:32, 2.30s/it] 57%|█████▋ | 17133/30000 [11:00:21<8:11:55, 2.29s/it] 57%|█████▋ | 17134/30000 [11:00:24<8:07:42, 2.27s/it] 57%|█████▋ | 17135/30000 [11:00:26<8:11:52, 2.29s/it] 57%|█████▋ | 17136/30000 [11:00:28<8:11:50, 2.29s/it] 57%|█████▋ | 17137/30000 [11:00:31<8:11:52, 2.29s/it] 57%|█████▋ | 17138/30000 [11:00:33<8:12:44, 2.30s/it] 57%|█████▋ | 17139/30000 [11:00:35<8:11:50, 2.29s/it] 57%|█████▋ | 17140/30000 [11:00:37<8:11:12, 2.29s/it] 57%|█████▋ | 17140/30000 [11:00:38<8:11:12, 2.29s/it] 57%|█████▋ | 17141/30000 [11:00:40<8:11:43, 2.29s/it] 57%|█████▋ | 17142/30000 [11:00:42<8:12:05, 2.30s/it] 57%|█████▋ | 17143/30000 [11:00:44<8:08:07, 2.28s/it] 57%|█████▋ | 17144/30000 [11:00:47<8:07:53, 2.28s/it] 57%|█████▋ | 17145/30000 [11:00:49<8:11:36, 2.29s/it] 57%|█████▋ | 17146/30000 [11:00:51<8:10:20, 2.29s/it] 57%|█████▋ | 17147/30000 [11:00:53<8:09:54, 2.29s/it] 57%|█████▋ | 17148/30000 [11:00:56<8:14:06, 2.31s/it] 57%|█████▋ | 17149/30000 [11:00:58<8:14:53, 2.31s/it] 57%|█████▋ | 17150/30000 [11:01:00<8:11:11, 2.29s/it] 57%|█████▋ | 17150/30000 [11:01:00<8:11:11, 2.29s/it] 57%|█████▋ | 17151/30000 [11:01:03<8:13:41, 2.31s/it] 57%|█████▋ | 17152/30000 [11:01:05<8:12:19, 2.30s/it] 57%|█████▋ | 17153/30000 [11:01:07<8:11:04, 2.29s/it] 57%|█████▋ | 17154/30000 [11:01:10<8:09:40, 2.29s/it] 57%|█████▋ | 17155/30000 [11:01:12<8:13:50, 2.31s/it] 57%|█████▋ | 17156/30000 [11:01:14<8:10:15, 2.29s/it] 57%|█████▋ | 17157/30000 [11:01:16<8:09:11, 2.29s/it] 57%|█████▋ | 17158/30000 [11:01:19<8:11:24, 2.30s/it] 57%|█████▋ | 17159/30000 [11:01:21<8:12:59, 2.30s/it] 57%|█████▋ | 17160/30000 [11:01:23<8:10:40, 2.29s/it] 57%|█████▋ | 17160/30000 [11:01:23<8:10:40, 2.29s/it] 57%|█████▋ | 17161/30000 [11:01:26<8:12:16, 2.30s/it] 57%|█████▋ | 17162/30000 [11:01:28<8:14:13, 2.31s/it] 57%|█████▋ | 17163/30000 [11:01:30<8:12:05, 2.30s/it] 57%|█████▋ | 17164/30000 [11:01:33<8:10:08, 2.29s/it] 57%|█████▋ | 17165/30000 [11:01:35<8:07:51, 2.28s/it] 57%|█████▋ | 17166/30000 [11:01:37<8:10:00, 2.29s/it] 57%|█████▋ | 17167/30000 [11:01:39<8:11:20, 2.30s/it] 57%|█████▋ | 17168/30000 [11:01:42<8:10:54, 2.30s/it] 57%|█████▋ | 17169/30000 [11:01:44<8:09:32, 2.29s/it] 57%|█████▋ | 17170/30000 [11:01:46<8:15:46, 2.32s/it] 57%|█████▋ | 17170/30000 [11:01:46<8:15:46, 2.32s/it] 57%|█████▋ | 17171/30000 [11:01:49<8:16:54, 2.32s/it] 57%|█████▋ | 17172/30000 [11:01:51<8:11:52, 2.30s/it] 57%|█████▋ | 17173/30000 [11:01:53<8:09:27, 2.29s/it] 57%|█████▋ | 17174/30000 [11:01:56<8:13:13, 2.31s/it] 57%|█████▋ | 17175/30000 [11:01:58<8:13:58, 2.31s/it] 57%|█████▋ | 17176/30000 [11:02:00<8:11:44, 2.30s/it] 57%|█████▋ | 17177/30000 [11:02:02<8:10:09, 2.29s/it] 57%|█████▋ | 17178/30000 [11:02:05<8:11:24, 2.30s/it] 57%|█████▋ | 17179/30000 [11:02:07<8:08:49, 2.29s/it] 57%|█████▋ | 17180/30000 [11:02:09<8:04:54, 2.27s/it] 57%|█████▋ | 17180/30000 [11:02:09<8:04:54, 2.27s/it] 57%|█████▋ | 17181/30000 [11:02:12<8:09:50, 2.29s/it] 57%|█████▋ | 17182/30000 [11:02:14<8:11:35, 2.30s/it] 57%|█████▋ | 17183/30000 [11:02:16<8:07:38, 2.28s/it] 57%|█████▋ | 17184/30000 [11:02:18<8:05:52, 2.27s/it] 57%|█████▋ | 17185/30000 [11:02:19<6:41:53, 1.88s/it] 57%|█████▋ | 17186/30000 [11:02:35<21:48:03, 6.12s/it] 57%|█████▋ | 17187/30000 [11:02:38<17:45:07, 4.99s/it] 57%|█████▋ | 17188/30000 [11:02:40<14:52:30, 4.18s/it] 57%|█████▋ | 17189/30000 [11:02:42<12:53:54, 3.62s/it] 57%|█████▋ | 17190/30000 [11:02:45<11:33:44, 3.25s/it] 57%|█████▋ | 17190/30000 [11:02:45<11:33:44, 3.25s/it] 57%|█████▋ | 17191/30000 [11:02:47<10:34:19, 2.97s/it] 57%|█████▋ | 17192/30000 [11:02:49<9:48:49, 2.76s/it] 57%|█████▋ | 17193/30000 [11:02:52<9:20:52, 2.63s/it] 57%|█████▋ | 17194/30000 [11:02:54<9:01:26, 2.54s/it] 57%|█████▋ | 17195/30000 [11:02:56<8:42:38, 2.45s/it] 57%|█████▋ | 17196/30000 [11:02:59<8:35:32, 2.42s/it] 57%|█████▋ | 17197/30000 [11:03:01<8:30:14, 2.39s/it] 57%|█████▋ | 17198/30000 [11:03:03<8:24:21, 2.36s/it] 57%|█████▋ | 17199/30000 [11:03:05<8:19:15, 2.34s/it] 57%|█████▋ | 17200/30000 [11:03:08<8:21:22, 2.35s/it] 57%|█████▋ | 17200/30000 [11:03:08<8:21:22, 2.35s/it] 57%|█████▋ | 17201/30000 [11:03:10<8:19:40, 2.34s/it] 57%|█████▋ | 17202/30000 [11:03:12<8:13:24, 2.31s/it] 57%|█████▋ | 17203/30000 [11:03:15<8:15:32, 2.32s/it] 57%|█████▋ | 17204/30000 [11:03:17<8:15:34, 2.32s/it] 57%|█████▋ | 17205/30000 [11:03:19<8:12:00, 2.31s/it] 57%|█████▋ | 17206/30000 [11:03:22<8:18:56, 2.34s/it] 57%|█████▋ | 17207/30000 [11:03:24<8:20:26, 2.35s/it] 57%|█████▋ | 17208/30000 [11:03:26<8:16:33, 2.33s/it] 57%|█████▋ | 17209/30000 [11:03:29<8:13:01, 2.31s/it] 57%|█████▋ | 17210/30000 [11:03:31<8:15:42, 2.33s/it] 57%|█████▋ | 17210/30000 [11:03:31<8:15:42, 2.33s/it] 57%|█████▋ | 17211/30000 [11:03:33<8:12:26, 2.31s/it] 57%|█████▋ | 17212/30000 [11:03:36<8:11:04, 2.30s/it] 57%|█████▋ | 17213/30000 [11:03:38<8:16:48, 2.33s/it] 57%|█████▋ | 17214/30000 [11:03:40<8:14:44, 2.32s/it] 57%|█████▋ | 17215/30000 [11:03:43<8:14:28, 2.32s/it] 57%|█████▋ | 17216/30000 [11:03:45<8:12:32, 2.31s/it] 57%|█████▋ | 17217/30000 [11:03:47<8:13:23, 2.32s/it] 57%|█████▋ | 17218/30000 [11:03:50<8:09:39, 2.30s/it] 57%|█████▋ | 17219/30000 [11:03:52<8:08:02, 2.29s/it] 57%|█████▋ | 17220/30000 [11:03:54<8:09:22, 2.30s/it] 57%|█████▋ | 17220/30000 [11:03:54<8:09:22, 2.30s/it] 57%|█████▋ | 17221/30000 [11:03:56<8:11:00, 2.31s/it] 57%|█████▋ | 17222/30000 [11:03:59<8:09:09, 2.30s/it] 57%|█████▋ | 17223/30000 [11:04:01<8:13:51, 2.32s/it] 57%|█████▋ | 17224/30000 [11:04:03<8:15:30, 2.33s/it] 57%|█████▋ | 17225/30000 [11:04:06<8:12:25, 2.31s/it] 57%|█████▋ | 17226/30000 [11:04:08<8:10:46, 2.31s/it] 57%|█████▋ | 17227/30000 [11:04:10<8:10:39, 2.30s/it] 57%|█████▋ | 17228/30000 [11:04:13<8:08:18, 2.29s/it] 57%|█████▋ | 17229/30000 [11:04:15<8:08:29, 2.29s/it] 57%|█████▋ | 17230/30000 [11:04:17<8:13:31, 2.32s/it] 57%|█████▋ | 17230/30000 [11:04:17<8:13:31, 2.32s/it] 57%|█████▋ | 17231/30000 [11:04:20<8:14:47, 2.32s/it] 57%|█████▋ | 17232/30000 [11:04:22<8:14:01, 2.32s/it] 57%|█████▋ | 17233/30000 [11:04:24<8:13:58, 2.32s/it] 57%|█████▋ | 17234/30000 [11:04:27<8:12:50, 2.32s/it] 57%|█████▋ | 17235/30000 [11:04:29<8:09:52, 2.30s/it] 57%|█████▋ | 17236/30000 [11:04:31<8:07:10, 2.29s/it] 57%|█████▋ | 17237/30000 [11:04:33<8:10:02, 2.30s/it] 57%|█████▋ | 17238/30000 [11:04:36<8:11:05, 2.31s/it] 57%|█████▋ | 17239/30000 [11:04:38<8:09:08, 2.30s/it] 57%|█████▋ | 17240/30000 [11:04:40<8:12:05, 2.31s/it] 57%|█████▋ | 17240/30000 [11:04:40<8:12:05, 2.31s/it] 57%|█████▋ | 17241/30000 [11:04:43<8:11:33, 2.31s/it] 57%|█████▋ | 17242/30000 [11:04:45<8:08:07, 2.30s/it] 57%|█████▋ | 17243/30000 [11:04:47<8:10:55, 2.31s/it] 57%|█████▋ | 17244/30000 [11:04:50<8:11:40, 2.31s/it] 57%|█████▋ | 17245/30000 [11:04:52<8:07:34, 2.29s/it] 57%|█████▋ | 17246/30000 [11:04:54<8:07:35, 2.29s/it] 57%|█████▋ | 17247/30000 [11:04:56<8:09:43, 2.30s/it] 57%|█████▋ | 17248/30000 [11:04:59<8:08:47, 2.30s/it] 57%|█████▋ | 17249/30000 [11:05:01<8:04:34, 2.28s/it] 57%|█████▊ | 17250/30000 [11:05:03<8:09:21, 2.30s/it] 57%|█████▊ | 17250/30000 [11:05:03<8:09:21, 2.30s/it] 58%|█████▊ | 17251/30000 [11:05:06<8:11:43, 2.31s/it] 58%|█████▊ | 17252/30000 [11:05:08<8:08:08, 2.30s/it] 58%|█████▊ | 17253/30000 [11:05:10<8:09:59, 2.31s/it] 58%|█████▊ | 17254/30000 [11:05:12<8:07:20, 2.29s/it] 58%|█████▊ | 17255/30000 [11:05:15<8:07:08, 2.29s/it] 58%|█████▊ | 17256/30000 [11:05:17<8:10:42, 2.31s/it] 58%|█████▊ | 17257/30000 [11:05:19<8:08:36, 2.30s/it] 58%|█████▊ | 17258/30000 [11:05:22<8:07:21, 2.29s/it] 58%|█████▊ | 17259/30000 [11:05:24<8:06:13, 2.29s/it] 58%|█████▊ | 17260/30000 [11:05:26<8:08:37, 2.30s/it] 58%|█████▊ | 17260/30000 [11:05:26<8:08:37, 2.30s/it] 58%|█████▊ | 17261/30000 [11:05:29<8:08:35, 2.30s/it] 58%|█████▊ | 17262/30000 [11:05:31<8:06:27, 2.29s/it] 58%|█████▊ | 17263/30000 [11:05:33<8:10:54, 2.31s/it] 58%|█████▊ | 17264/30000 [11:05:36<8:10:44, 2.31s/it] 58%|█████▊ | 17265/30000 [11:05:38<8:06:47, 2.29s/it] 58%|█████▊ | 17266/30000 [11:05:40<8:09:51, 2.31s/it] 58%|█████▊ | 17267/30000 [11:05:42<8:11:20, 2.32s/it] 58%|█████▊ | 17268/30000 [11:05:45<8:07:51, 2.30s/it] 58%|█████▊ | 17269/30000 [11:05:47<8:10:50, 2.31s/it] 58%|█████▊ | 17270/30000 [11:05:49<8:10:27, 2.31s/it] 58%|█████▊ | 17270/30000 [11:05:49<8:10:27, 2.31s/it] 58%|█████▊ | 17271/30000 [11:05:52<8:09:13, 2.31s/it] 58%|█████▊ | 17272/30000 [11:05:54<8:08:11, 2.30s/it] 58%|█████▊ | 17273/30000 [11:05:56<8:09:00, 2.31s/it] 58%|█████▊ | 17274/30000 [11:05:59<8:06:49, 2.30s/it] 58%|█████▊ | 17275/30000 [11:06:01<8:06:12, 2.29s/it] 58%|█████▊ | 17276/30000 [11:06:03<8:07:37, 2.30s/it] 58%|█████▊ | 17277/30000 [11:06:05<8:06:17, 2.29s/it] 58%|█████▊ | 17278/30000 [11:06:08<8:05:10, 2.29s/it] 58%|█████▊ | 17279/30000 [11:06:10<8:10:17, 2.31s/it] 58%|█████▊ | 17280/30000 [11:06:12<8:10:01, 2.31s/it] 58%|█████▊ | 17280/30000 [11:06:12<8:10:01, 2.31s/it] 58%|█████▊ | 17281/30000 [11:06:15<8:06:40, 2.30s/it] 58%|█████▊ | 17282/30000 [11:06:17<8:07:40, 2.30s/it] 58%|█████▊ | 17283/30000 [11:06:19<8:06:27, 2.30s/it] 58%|█████▊ | 17284/30000 [11:06:22<8:04:51, 2.29s/it] 58%|█████▊ | 17285/30000 [11:06:24<8:04:15, 2.29s/it] 58%|█████▊ | 17286/30000 [11:06:26<8:09:50, 2.31s/it] 58%|█████▊ | 17287/30000 [11:06:29<8:11:25, 2.32s/it] 58%|█████▊ | 17288/30000 [11:06:31<8:07:35, 2.30s/it] 58%|█████▊ | 17289/30000 [11:06:33<8:09:42, 2.31s/it] 58%|█████▊ | 17290/30000 [11:06:35<8:07:07, 2.30s/it] 58%|█████▊ | 17290/30000 [11:06:35<8:07:07, 2.30s/it] 58%|█████▊ | 17291/30000 [11:06:38<8:04:48, 2.29s/it] 58%|█████▊ | 17292/30000 [11:06:40<8:04:18, 2.29s/it] 58%|█████▊ | 17293/30000 [11:06:42<8:03:52, 2.28s/it] 58%|█████▊ | 17294/30000 [11:06:45<8:05:20, 2.29s/it] 58%|█████▊ | 17295/30000 [11:06:47<8:05:35, 2.29s/it] 58%|█████▊ | 17296/30000 [11:06:49<8:10:43, 2.32s/it] 58%|█████▊ | 17297/30000 [11:06:51<8:10:25, 2.32s/it] 58%|█████▊ | 17298/30000 [11:06:54<8:06:36, 2.30s/it] 58%|█████▊ | 17299/30000 [11:06:56<8:05:41, 2.29s/it] 58%|█████▊ | 17300/30000 [11:06:58<8:05:04, 2.29s/it] 58%|█████▊ | 17300/30000 [11:06:58<8:05:04, 2.29s/it] 58%|█████▊ | 17301/30000 [11:07:01<8:04:09, 2.29s/it] 58%|█████▊ | 17302/30000 [11:07:03<8:03:48, 2.29s/it] 58%|█████▊ | 17303/30000 [11:07:05<8:09:23, 2.31s/it] 58%|█████▊ | 17304/30000 [11:07:08<8:09:04, 2.31s/it] 58%|█████▊ | 17305/30000 [11:07:10<8:04:30, 2.29s/it] 58%|█████▊ | 17306/30000 [11:07:12<8:02:02, 2.28s/it] 58%|█████▊ | 17307/30000 [11:07:14<8:07:05, 2.30s/it] 58%|█████▊ | 17308/30000 [11:07:17<8:08:44, 2.31s/it] 58%|█████▊ | 17309/30000 [11:07:19<8:05:25, 2.29s/it] 58%|█████▊ | 17310/30000 [11:07:21<8:05:49, 2.30s/it] 58%|█████▊ | 17310/30000 [11:07:21<8:05:49, 2.30s/it] 58%|█████▊ | 17311/30000 [11:07:24<8:11:02, 2.32s/it] 58%|█████▊ | 17312/30000 [11:07:26<8:13:32, 2.33s/it] 58%|█████▊ | 17313/30000 [11:07:28<8:10:17, 2.32s/it] 58%|█████▊ | 17314/30000 [11:07:31<8:06:40, 2.30s/it] 58%|█████▊ | 17315/30000 [11:07:33<8:08:09, 2.31s/it] 58%|█████▊ | 17316/30000 [11:07:35<8:04:54, 2.29s/it] 58%|█████▊ | 17317/30000 [11:07:37<8:04:16, 2.29s/it] 58%|█████▊ | 17318/30000 [11:07:40<8:02:55, 2.28s/it] 58%|█████▊ | 17319/30000 [11:07:42<8:07:16, 2.31s/it] 58%|█████▊ | 17320/30000 [11:07:44<8:03:29, 2.29s/it] 58%|█████▊ | 17320/30000 [11:07:44<8:03:29, 2.29s/it] 58%|█████▊ | 17321/30000 [11:07:47<8:00:59, 2.28s/it] 58%|█████▊ | 17322/30000 [11:07:49<8:04:10, 2.29s/it] 58%|█████▊ | 17323/30000 [11:07:51<8:06:18, 2.30s/it] 58%|█████▊ | 17324/30000 [11:07:53<8:03:09, 2.29s/it] 58%|█████▊ | 17325/30000 [11:07:56<8:03:49, 2.29s/it] 58%|█████▊ | 17326/30000 [11:07:58<8:06:49, 2.30s/it] 58%|█████▊ | 17327/30000 [11:08:00<8:08:03, 2.31s/it] 58%|█████▊ | 17328/30000 [11:08:03<8:05:16, 2.30s/it] 58%|█████▊ | 17329/30000 [11:08:05<8:03:56, 2.29s/it] 58%|█████▊ | 17330/30000 [11:08:07<8:05:51, 2.30s/it] 58%|█████▊ | 17330/30000 [11:08:07<8:05:51, 2.30s/it] 58%|█████▊ | 17331/30000 [11:08:10<8:03:00, 2.29s/it] 58%|█████▊ | 17332/30000 [11:08:12<8:02:37, 2.29s/it] 58%|█████▊ | 17333/30000 [11:08:14<8:01:31, 2.28s/it] 58%|█████▊ | 17334/30000 [11:08:16<8:06:40, 2.31s/it] 58%|█████▊ | 17335/30000 [11:08:19<8:08:45, 2.32s/it] 58%|█████▊ | 17336/30000 [11:08:21<8:06:06, 2.30s/it] 58%|█████▊ | 17337/30000 [11:08:23<8:04:21, 2.29s/it] 58%|█████▊ | 17338/30000 [11:08:26<8:06:32, 2.31s/it] 58%|█████▊ | 17339/30000 [11:08:28<8:04:28, 2.30s/it] 58%|█████▊ | 17340/30000 [11:08:30<8:02:02, 2.28s/it] 58%|█████▊ | 17340/30000 [11:08:30<8:02:02, 2.28s/it] 58%|█████▊ | 17341/30000 [11:08:33<8:03:30, 2.29s/it] 58%|█████▊ | 17342/30000 [11:08:35<8:07:13, 2.31s/it] 58%|█████▊ | 17343/30000 [11:08:37<8:08:31, 2.32s/it] 58%|█████▊ | 17344/30000 [11:08:40<8:07:25, 2.31s/it] 58%|█████▊ | 17345/30000 [11:08:42<8:04:37, 2.30s/it] 58%|█████▊ | 17346/30000 [11:08:44<8:07:53, 2.31s/it] 58%|█████▊ | 17347/30000 [11:08:46<8:03:36, 2.29s/it] 58%|█████▊ | 17348/30000 [11:08:49<8:02:56, 2.29s/it] 58%|█████▊ | 17349/30000 [11:08:51<8:00:56, 2.28s/it] 58%|█████▊ | 17350/30000 [11:08:53<8:06:38, 2.31s/it] 58%|█████▊ | 17350/30000 [11:08:53<8:06:38, 2.31s/it] 58%|█████▊ | 17351/30000 [11:08:56<8:08:44, 2.32s/it] 58%|█████▊ | 17352/30000 [11:08:58<8:05:08, 2.30s/it] 58%|█████▊ | 17353/30000 [11:09:00<8:02:23, 2.29s/it] 58%|█████▊ | 17354/30000 [11:09:02<8:05:08, 2.30s/it] 58%|█████▊ | 17355/30000 [11:09:05<8:05:16, 2.30s/it] 58%|█████▊ | 17356/30000 [11:09:07<8:02:54, 2.29s/it] 58%|█████▊ | 17357/30000 [11:09:09<8:02:31, 2.29s/it] 58%|█████▊ | 17358/30000 [11:09:12<8:05:11, 2.30s/it] 58%|█████▊ | 17359/30000 [11:09:14<8:02:23, 2.29s/it] 58%|█████▊ | 17360/30000 [11:09:16<8:02:46, 2.29s/it] 58%|█████▊ | 17360/30000 [11:09:16<8:02:46, 2.29s/it] 58%|█████▊ | 17361/30000 [11:09:19<8:01:29, 2.29s/it] 58%|█████▊ | 17362/30000 [11:09:21<8:02:06, 2.29s/it] 58%|█████▊ | 17363/30000 [11:09:23<8:00:00, 2.28s/it] 58%|█████▊ | 17364/30000 [11:09:25<8:00:42, 2.28s/it] 58%|█████▊ | 17365/30000 [11:09:28<8:04:41, 2.30s/it] 58%|█████▊ | 17366/30000 [11:09:30<8:05:31, 2.31s/it] 58%|█████▊ | 17367/30000 [11:09:32<8:03:03, 2.29s/it] 58%|█████▊ | 17368/30000 [11:09:35<8:03:05, 2.29s/it] 58%|█████▊ | 17369/30000 [11:09:37<8:05:39, 2.31s/it] 58%|█████▊ | 17370/30000 [11:09:39<8:02:16, 2.29s/it] +{'loss': 0.0141, 'grad_norm': 0.07021339237689972, 'learning_rate': 1.2990480312533213e-05, 'flow_matching_loss': 0.014088227227330209, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.94} +{'loss': 0.0158, 'grad_norm': 0.06388810276985168, 'learning_rate': 1.2974095887925258e-05, 'flow_matching_loss': 0.015809244010597467, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.94} +{'loss': 0.0141, 'grad_norm': 0.064362071454525, 'learning_rate': 1.2957713924981837e-05, 'flow_matching_loss': 0.014120496436953544, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.94} +{'loss': 0.0133, 'grad_norm': 0.06198849529027939, 'learning_rate': 1.2941334443608572e-05, 'flow_matching_loss': 0.0133256901986897, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.95} +{'loss': 0.0141, 'grad_norm': 0.07580011337995529, 'learning_rate': 1.2924957463708073e-05, 'flow_matching_loss': 0.014105560164898634, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.95} +{'loss': 0.0143, 'grad_norm': 0.0650518462061882, 'learning_rate': 1.2908583005179924e-05, 'flow_matching_loss': 0.014293930120766163, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.95} +{'loss': 0.0162, 'grad_norm': 0.06986358761787415, 'learning_rate': 1.2892211087920624e-05, 'flow_matching_loss': 0.016169224679470063, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.95} +{'loss': 0.0122, 'grad_norm': 0.06352328509092331, 'learning_rate': 1.287584173182361e-05, 'flow_matching_loss': 0.012169232312589883, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.96} +{'loss': 0.0142, 'grad_norm': 0.06794797629117966, 'learning_rate': 1.2859474956779188e-05, 'flow_matching_loss': 0.014230998419225215, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.96} +{'loss': 0.0141, 'grad_norm': 0.09690054506063461, 'learning_rate': 1.2843110782674532e-05, 'flow_matching_loss': 0.014136901125311852, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.96} +{'loss': 0.014, 'grad_norm': 0.0981927216053009, 'learning_rate': 1.2826749229393655e-05, 'flow_matching_loss': 0.013950208015739918, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.97} +{'loss': 0.0156, 'grad_norm': 0.07120291143655777, 'learning_rate': 1.2810390316817381e-05, 'flow_matching_loss': 0.01558370068669319, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.97} +{'loss': 0.0139, 'grad_norm': 0.1037883460521698, 'learning_rate': 1.2794034064823346e-05, 'flow_matching_loss': 0.013906182814389467, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.97} +{'loss': 0.0141, 'grad_norm': 0.08093751221895218, 'learning_rate': 1.277768049328593e-05, 'flow_matching_loss': 0.014103555306792259, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.98} +{'loss': 0.0149, 'grad_norm': 0.09241246432065964, 'learning_rate': 1.2761329622076269e-05, 'flow_matching_loss': 0.014909974485635757, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.98} +{'loss': 0.0136, 'grad_norm': 0.07650624960660934, 'learning_rate': 1.274498147106221e-05, 'flow_matching_loss': 0.013594804238528014, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.98} +{'loss': 0.013, 'grad_norm': 0.069866843521595, 'learning_rate': 1.272863606010829e-05, 'flow_matching_loss': 0.013011347502470016, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.98} +{'loss': 0.0153, 'grad_norm': 0.07052939385175705, 'learning_rate': 1.2712293409075739e-05, 'flow_matching_loss': 0.015285687800496817, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.99} +{'loss': 0.0149, 'grad_norm': 0.08401324599981308, 'learning_rate': 1.2695953537822413e-05, 'flow_matching_loss': 0.014901444409042596, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.99} +{'loss': 0.015, 'grad_norm': 0.09070569276809692, 'learning_rate': 1.2679616466202792e-05, 'flow_matching_loss': 0.014991042017936707, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 4.99} +{'loss': 0.0172, 'grad_norm': 0.09058485925197601, 'learning_rate': 1.2663282214067967e-05, 'flow_matching_loss': 0.01724343542009592, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.0} +{'loss': 0.0161, 'grad_norm': 0.07499388605356216, 'learning_rate': 1.264695080126559e-05, 'flow_matching_loss': 0.016057119145989417, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.0} +{'loss': 0.0132, 'grad_norm': 0.08485870808362961, 'learning_rate': 1.263062224763986e-05, 'flow_matching_loss': 0.013174483273178338, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.0} +{'loss': 0.0148, 'grad_norm': 0.0808720663189888, 'learning_rate': 1.2614296573031522e-05, 'flow_matching_loss': 0.01475791959092021, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.0} +{'loss': 0.0149, 'grad_norm': 0.07600932568311691, 'learning_rate': 1.259797379727781e-05, 'flow_matching_loss': 0.01492586052045226, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.01} +{'loss': 0.0134, 'grad_norm': 0.08669573813676834, 'learning_rate': 1.2581653940212423e-05, 'flow_matching_loss': 0.013432506565004587, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.01} +{'loss': 0.0143, 'grad_norm': 0.08033598214387894, 'learning_rate': 1.2565337021665537e-05, 'flow_matching_loss': 0.014336793962866069, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.01} +{'loss': 0.0148, 'grad_norm': 0.09820553660392761, 'learning_rate': 1.2549023061463739e-05, 'flow_matching_loss': 0.014789143856614829, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.02} +{'loss': 0.0144, 'grad_norm': 0.06980694085359573, 'learning_rate': 1.2532712079430038e-05, 'flow_matching_loss': 0.014397968538105488, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.02} +{'loss': 0.0159, 'grad_norm': 0.08698073774576187, 'learning_rate': 1.2516404095383813e-05, 'flow_matching_loss': 0.015867799799889327, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.02} +{'loss': 0.0155, 'grad_norm': 0.1154361218214035, 'learning_rate': 1.2500099129140796e-05, 'flow_matching_loss': 0.015494402032345533, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.02} +{'loss': 0.0142, 'grad_norm': 0.08798468858003616, 'learning_rate': 1.2483797200513065e-05, 'flow_matching_loss': 0.014247441850602626, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.03} +{'loss': 0.0149, 'grad_norm': 0.10355237871408463, 'learning_rate': 1.2467498329308992e-05, 'flow_matching_loss': 0.014925187267363072, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.03} +{'loss': 0.015, 'grad_norm': 0.058941714465618134, 'learning_rate': 1.2451202535333253e-05, 'flow_matching_loss': 0.015004857070744037, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.03} +{'loss': 0.0156, 'grad_norm': 0.07094793021678925, 'learning_rate': 1.2434909838386768e-05, 'flow_matching_loss': 0.015615834947675466, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.04} +{'loss': 0.013, 'grad_norm': 0.07840409129858017, 'learning_rate': 1.2418620258266703e-05, 'flow_matching_loss': 0.012983580958098173, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.04} +{'loss': 0.0152, 'grad_norm': 0.07920239865779877, 'learning_rate': 1.2402333814766432e-05, 'flow_matching_loss': 0.015236461628228426, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.04} +{'loss': 0.0133, 'grad_norm': 0.08205486088991165, 'learning_rate': 1.2386050527675515e-05, 'flow_matching_loss': 0.013295307569205762, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.05} +{'loss': 0.013, 'grad_norm': 0.07248560339212418, 'learning_rate': 1.2369770416779693e-05, 'flow_matching_loss': 0.012957388907670975, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.05} +{'loss': 0.0136, 'grad_norm': 0.07421214133501053, 'learning_rate': 1.2353493501860828e-05, 'flow_matching_loss': 0.013592023402452469, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.05} +{'loss': 0.0153, 'grad_norm': 0.10148954391479492, 'learning_rate': 1.2337219802696905e-05, 'flow_matching_loss': 0.01527148773893714, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.05} 58%|█████▊ | 17370/30000 [11:09:39<8:02:16, 2.29s/it] 58%|█████▊ | 17371/30000 [11:09:41<8:01:03, 2.29s/it] 58%|█████▊ | 17372/30000 [11:09:44<8:01:34, 2.29s/it] 58%|█████▊ | 17373/30000 [11:09:46<8:02:29, 2.29s/it] 58%|█████▊ | 17374/30000 [11:09:48<8:00:41, 2.28s/it] 58%|█████▊ | 17375/30000 [11:09:51<7:57:22, 2.27s/it] 58%|█████▊ | 17376/30000 [11:09:53<8:03:45, 2.30s/it] 58%|█████▊ | 17377/30000 [11:09:55<8:00:19, 2.28s/it] 58%|█████▊ | 17378/30000 [11:09:57<7:59:24, 2.28s/it] 58%|█████▊ | 17379/30000 [11:10:00<8:01:34, 2.29s/it] 58%|█████▊ | 17380/30000 [11:10:02<8:04:23, 2.30s/it] 58%|█████▊ | 17380/30000 [11:10:02<8:04:23, 2.30s/it] 58%|█████▊ | 17381/30000 [11:10:04<8:02:38, 2.29s/it] 58%|█████▊ | 17382/30000 [11:10:07<8:01:48, 2.29s/it] 58%|█████▊ | 17383/30000 [11:10:09<8:06:41, 2.31s/it] 58%|█████▊ | 17384/30000 [11:10:11<8:05:34, 2.31s/it] 58%|█████▊ | 17385/30000 [11:10:14<8:02:14, 2.29s/it] 58%|█████▊ | 17386/30000 [11:10:16<8:00:22, 2.28s/it] 58%|█████▊ | 17387/30000 [11:10:18<8:04:28, 2.30s/it] 58%|█████▊ | 17388/30000 [11:10:20<8:04:22, 2.30s/it] 58%|█████▊ | 17389/30000 [11:10:23<8:01:12, 2.29s/it] 58%|█████▊ | 17390/30000 [11:10:25<8:03:30, 2.30s/it] 58%|█████▊ | 17390/30000 [11:10:25<8:03:30, 2.30s/it] 58%|█████▊ | 17391/30000 [11:10:27<8:06:19, 2.31s/it] 58%|█████▊ | 17392/30000 [11:10:30<8:03:56, 2.30s/it] 58%|█████▊ | 17393/30000 [11:10:32<8:00:14, 2.29s/it] 58%|█████▊ | 17394/30000 [11:10:34<8:06:22, 2.31s/it] 58%|█████▊ | 17395/30000 [11:10:37<8:04:42, 2.31s/it] 58%|█████▊ | 17396/30000 [11:10:39<8:01:38, 2.29s/it] 58%|█████▊ | 17397/30000 [11:10:41<8:03:41, 2.30s/it] 58%|█████▊ | 17398/30000 [11:10:43<8:01:57, 2.29s/it] 58%|█████▊ | 17399/30000 [11:10:46<8:02:07, 2.30s/it] 58%|█████▊ | 17400/30000 [11:10:48<8:00:37, 2.29s/it] 58%|█████▊ | 17400/30000 [11:10:48<8:00:37, 2.29s/it] 58%|█████▊ | 17401/30000 [11:10:50<8:04:15, 2.31s/it] 58%|█████▊ | 17402/30000 [11:10:53<8:03:19, 2.30s/it] 58%|█████▊ | 17403/30000 [11:10:55<8:00:38, 2.29s/it] 58%|█████▊ | 17404/30000 [11:10:57<7:59:51, 2.29s/it] 58%|█████▊ | 17405/30000 [11:11:00<8:02:02, 2.30s/it] 58%|█████▊ | 17406/30000 [11:11:02<7:58:08, 2.28s/it] 58%|█████▊ | 17407/30000 [11:11:04<7:57:34, 2.28s/it] 58%|█████▊ | 17408/30000 [11:11:06<8:02:09, 2.30s/it] 58%|█████▊ | 17409/30000 [11:11:09<8:02:57, 2.30s/it] 58%|█████▊ | 17410/30000 [11:11:11<8:01:41, 2.30s/it] 58%|█████▊ | 17410/30000 [11:11:11<8:01:41, 2.30s/it] 58%|█████▊ | 17411/30000 [11:11:13<8:01:38, 2.30s/it] 58%|█████▊ | 17412/30000 [11:11:16<8:02:50, 2.30s/it] 58%|█████▊ | 17413/30000 [11:11:18<7:59:34, 2.29s/it] 58%|█████▊ | 17414/30000 [11:11:20<7:58:30, 2.28s/it] 58%|█████▊ | 17415/30000 [11:11:22<8:04:13, 2.31s/it] 58%|█████▊ | 17416/30000 [11:11:25<8:05:05, 2.31s/it] 58%|█████▊ | 17417/30000 [11:11:27<8:03:01, 2.30s/it] 58%|█████▊ | 17418/30000 [11:11:29<8:02:20, 2.30s/it] 58%|█████▊ | 17419/30000 [11:11:32<8:04:32, 2.31s/it] 58%|█████▊ | 17420/30000 [11:11:34<7:58:46, 2.28s/it] 58%|█████▊ | 17420/30000 [11:11:34<7:58:46, 2.28s/it] 58%|█████▊ | 17421/30000 [11:11:36<7:57:47, 2.28s/it] 58%|█████▊ | 17422/30000 [11:11:39<8:04:23, 2.31s/it] 58%|█████▊ | 17423/30000 [11:11:41<8:05:37, 2.32s/it] 58%|█████▊ | 17424/30000 [11:11:43<8:04:12, 2.31s/it] 58%|█████▊ | 17425/30000 [11:11:46<8:05:52, 2.32s/it] 58%|█████▊ | 17426/30000 [11:11:48<8:05:49, 2.32s/it] 58%|█████▊ | 17427/30000 [11:11:50<7:59:07, 2.29s/it] 58%|█████▊ | 17428/30000 [11:11:52<7:59:48, 2.29s/it] 58%|█████▊ | 17429/30000 [11:11:55<8:02:10, 2.30s/it] 58%|█████▊ | 17430/30000 [11:11:57<8:01:42, 2.30s/it] 58%|█████▊ | 17430/30000 [11:11:57<8:01:42, 2.30s/it] 58%|█████▊ | 17431/30000 [11:11:59<8:02:06, 2.30s/it] 58%|█████▊ | 17432/30000 [11:12:02<8:04:08, 2.31s/it] 58%|█████▊ | 17433/30000 [11:12:04<8:04:32, 2.31s/it] 58%|█████▊ | 17434/30000 [11:12:06<8:01:37, 2.30s/it] 58%|█████▊ | 17435/30000 [11:12:09<8:03:29, 2.31s/it] 58%|█████▊ | 17436/30000 [11:12:11<8:03:15, 2.31s/it] 58%|█████▊ | 17437/30000 [11:12:13<8:01:59, 2.30s/it] 58%|█████▊ | 17438/30000 [11:12:15<8:01:55, 2.30s/it] 58%|█████▊ | 17439/30000 [11:12:18<8:05:21, 2.32s/it] 58%|█████▊ | 17440/30000 [11:12:20<7:59:12, 2.29s/it] 58%|█████▊ | 17440/30000 [11:12:20<7:59:12, 2.29s/it] 58%|█████▊ | 17441/30000 [11:12:22<7:59:44, 2.29s/it] 58%|█████▊ | 17442/30000 [11:12:25<8:02:53, 2.31s/it] 58%|█████▊ | 17443/30000 [11:12:27<8:03:02, 2.31s/it] 58%|█████▊ | 17444/30000 [11:12:29<8:01:22, 2.30s/it] 58%|█████▊ | 17445/30000 [11:12:32<8:04:04, 2.31s/it] 58%|█████▊ | 17446/30000 [11:12:34<8:04:26, 2.32s/it] 58%|█████▊ | 17447/30000 [11:12:36<8:01:53, 2.30s/it] 58%|█████▊ | 17448/30000 [11:12:38<8:00:17, 2.30s/it] 58%|█████▊ | 17449/30000 [11:12:41<8:06:14, 2.32s/it] 58%|█████▊ | 17450/30000 [11:12:43<8:00:56, 2.30s/it] 58%|█████▊ | 17450/30000 [11:12:43<8:00:56, 2.30s/it] 58%|█████▊ | 17451/30000 [11:12:45<7:58:41, 2.29s/it] 58%|█████▊ | 17452/30000 [11:12:48<8:02:19, 2.31s/it] 58%|█████▊ | 17453/30000 [11:12:50<8:05:06, 2.32s/it] 58%|█████▊ | 17454/30000 [11:12:52<8:00:57, 2.30s/it] 58%|█████▊ | 17455/30000 [11:12:55<8:04:23, 2.32s/it] 58%|█████▊ | 17456/30000 [11:12:57<8:04:12, 2.32s/it] 58%|█████▊ | 17457/30000 [11:12:59<7:58:55, 2.29s/it] 58%|█████▊ | 17458/30000 [11:13:02<8:02:21, 2.31s/it] 58%|█████▊ | 17459/30000 [11:13:04<8:02:24, 2.31s/it] 58%|█████▊ | 17460/30000 [11:13:06<7:59:49, 2.30s/it] 58%|█████▊ | 17460/30000 [11:13:06<7:59:49, 2.30s/it] 58%|█████▊ | 17461/30000 [11:13:08<8:03:23, 2.31s/it] 58%|█████▊ | 17462/30000 [11:13:11<8:03:29, 2.31s/it] 58%|█████▊ | 17463/30000 [11:13:13<8:01:57, 2.31s/it] 58%|█████▊ | 17464/30000 [11:13:15<7:59:23, 2.29s/it] 58%|█████▊ | 17465/30000 [11:13:18<8:02:26, 2.31s/it] 58%|█████▊ | 17466/30000 [11:13:20<8:00:59, 2.30s/it] 58%|█████▊ | 17467/30000 [11:13:22<7:58:41, 2.29s/it] 58%|█████▊ | 17468/30000 [11:13:25<8:03:07, 2.31s/it] 58%|█████▊ | 17469/30000 [11:13:27<8:04:43, 2.32s/it] 58%|█████▊ | 17470/30000 [11:13:29<8:01:38, 2.31s/it] 58%|█████▊ | 17470/30000 [11:13:29<8:01:38, 2.31s/it] 58%|█████▊ | 17471/30000 [11:13:32<8:05:23, 2.32s/it] 58%|█████▊ | 17472/30000 [11:13:34<8:03:39, 2.32s/it] 58%|█████▊ | 17473/30000 [11:13:36<8:01:38, 2.31s/it] 58%|█████▊ | 17474/30000 [11:13:39<8:02:50, 2.31s/it] 58%|█████▊ | 17475/30000 [11:13:41<8:04:44, 2.32s/it] 58%|█████▊ | 17476/30000 [11:13:43<8:01:46, 2.31s/it] 58%|█████▊ | 17477/30000 [11:13:45<7:59:51, 2.30s/it] 58%|█████▊ | 17478/30000 [11:13:48<7:58:21, 2.29s/it] 58%|█████▊ | 17479/30000 [11:13:50<7:57:33, 2.29s/it] 58%|█████▊ | 17480/30000 [11:13:52<7:56:33, 2.28s/it] 58%|█████▊ | 17480/30000 [11:13:52<7:56:33, 2.28s/it] 58%|█████▊ | 17481/30000 [11:13:55<8:01:00, 2.31s/it] 58%|█████▊ | 17482/30000 [11:13:57<7:57:09, 2.29s/it] 58%|█████▊ | 17483/30000 [11:13:59<7:57:29, 2.29s/it] 58%|█████▊ | 17484/30000 [11:14:01<8:02:02, 2.31s/it] 58%|█████▊ | 17485/30000 [11:14:04<8:02:45, 2.31s/it] 58%|█████▊ | 17486/30000 [11:14:06<8:00:04, 2.30s/it] 58%|█████▊ | 17487/30000 [11:14:08<8:03:15, 2.32s/it] 58%|█████▊ | 17488/30000 [11:14:11<8:02:28, 2.31s/it] 58%|█████▊ | 17489/30000 [11:14:13<8:01:04, 2.31s/it] 58%|█████▊ | 17490/30000 [11:14:15<8:01:21, 2.31s/it] 58%|█████▊ | 17490/30000 [11:14:15<8:01:21, 2.31s/it] 58%|█████▊ | 17491/30000 [11:14:18<8:02:57, 2.32s/it] 58%|█████▊ | 17492/30000 [11:14:20<7:58:57, 2.30s/it] 58%|█████▊ | 17493/30000 [11:14:22<7:58:30, 2.30s/it] 58%|█████▊ | 17494/30000 [11:14:25<8:03:49, 2.32s/it] 58%|█████▊ | 17495/30000 [11:14:27<8:04:22, 2.32s/it] 58%|█████▊ | 17496/30000 [11:14:29<8:03:46, 2.32s/it] 58%|█████▊ | 17497/30000 [11:14:32<8:04:58, 2.33s/it] 58%|█████▊ | 17498/30000 [11:14:34<8:04:13, 2.32s/it] 58%|█████▊ | 17499/30000 [11:14:36<8:02:34, 2.32s/it] 58%|█████▊ | 17500/30000 [11:14:38<7:59:13, 2.30s/it] 58%|█████▊ | 17500/30000 [11:14:39<7:59:13, 2.30s/it] 58%|█████▊ | 17501/30000 [11:14:41<8:04:58, 2.33s/it] 58%|█████▊ | 17502/30000 [11:14:43<8:05:13, 2.33s/it] 58%|█████▊ | 17503/30000 [11:14:45<8:00:38, 2.31s/it] 58%|█████▊ | 17504/30000 [11:14:48<8:04:20, 2.33s/it] 58%|█████▊ | 17505/30000 [11:14:50<8:01:56, 2.31s/it] 58%|█████▊ | 17506/30000 [11:14:52<7:58:11, 2.30s/it] 58%|█████▊ | 17507/30000 [11:14:55<7:59:40, 2.30s/it] 58%|█████▊ | 17508/30000 [11:14:57<7:57:04, 2.29s/it] 58%|█████▊ | 17509/30000 [11:14:59<7:56:00, 2.29s/it] 58%|█████▊ | 17510/30000 [11:15:02<7:56:35, 2.29s/it] 58%|█████▊ | 17510/30000 [11:15:02<7:56:35, 2.29s/it] 58%|█████▊ | 17511/30000 [11:15:04<8:00:07, 2.31s/it] 58%|█████▊ | 17512/30000 [11:15:06<7:57:34, 2.29s/it] 58%|█████▊ | 17513/30000 [11:15:08<7:54:52, 2.28s/it] 58%|█████▊ | 17514/30000 [11:15:11<7:58:11, 2.30s/it] 58%|█████▊ | 17515/30000 [11:15:13<8:01:30, 2.31s/it] 58%|█████▊ | 17516/30000 [11:15:15<7:57:31, 2.30s/it] 58%|█████▊ | 17517/30000 [11:15:18<7:59:18, 2.30s/it] 58%|█████▊ | 17518/30000 [11:15:20<8:02:19, 2.32s/it] 58%|█████▊ | 17519/30000 [11:15:22<7:58:35, 2.30s/it] 58%|█████▊ | 17520/30000 [11:15:25<7:55:15, 2.28s/it] 58%|█████▊ | 17520/30000 [11:15:25<7:55:15, 2.28s/it] 58%|█████▊ | 17521/30000 [11:15:27<7:58:34, 2.30s/it] 58%|█████▊ | 17522/30000 [11:15:29<7:54:21, 2.28s/it] 58%|█████▊ | 17523/30000 [11:15:31<7:53:30, 2.28s/it] 58%|█████▊ | 17524/30000 [11:15:34<7:55:51, 2.29s/it] 58%|█████▊ | 17525/30000 [11:15:36<7:59:14, 2.30s/it] 58%|█████▊ | 17526/30000 [11:15:38<7:55:45, 2.29s/it] 58%|█████▊ | 17527/30000 [11:15:41<7:57:57, 2.30s/it] 58%|█████▊ | 17528/30000 [11:15:43<7:59:54, 2.31s/it] 58%|█████▊ | 17529/30000 [11:15:45<7:55:00, 2.29s/it] 58%|█████▊ | 17530/30000 [11:15:47<7:53:51, 2.28s/it] 58%|█████▊ | 17530/30000 [11:15:47<7:53:51, 2.28s/it] 58%|█████▊ | 17531/30000 [11:15:50<8:01:04, 2.31s/it] 58%|█████▊ | 17532/30000 [11:15:52<8:00:02, 2.31s/it] 58%|█████▊ | 17533/30000 [11:15:54<7:56:29, 2.29s/it] 58%|█████▊ | 17534/30000 [11:15:57<7:56:59, 2.30s/it] 58%|█████▊ | 17535/30000 [11:15:59<8:01:11, 2.32s/it] 58%|█████▊ | 17536/30000 [11:16:01<7:59:35, 2.31s/it] 58%|█████▊ | 17537/30000 [11:16:04<7:55:06, 2.29s/it] 58%|█████▊ | 17538/30000 [11:16:06<7:57:53, 2.30s/it] 58%|█████▊ | 17539/30000 [11:16:08<7:55:35, 2.29s/it] 58%|█████▊ | 17540/30000 [11:16:10<7:53:36, 2.28s/it] 58%|█████▊ | 17540/30000 [11:16:10<7:53:36, 2.28s/it] 58%|█████▊ | 17541/30000 [11:16:13<7:55:36, 2.29s/it] 58%|█████▊ | 17542/30000 [11:16:15<7:57:06, 2.30s/it] 58%|█████▊ | 17543/30000 [11:16:17<7:55:00, 2.29s/it] 58%|█████▊ | 17544/30000 [11:16:20<7:49:23, 2.26s/it] 58%|█████▊ | 17545/30000 [11:16:22<7:54:04, 2.28s/it] 58%|█████▊ | 17546/30000 [11:16:24<7:56:14, 2.29s/it] 58%|█████▊ | 17547/30000 [11:16:26<7:55:04, 2.29s/it] 58%|█████▊ | 17548/30000 [11:16:29<7:54:21, 2.29s/it] 58%|█████▊ | 17549/30000 [11:16:31<7:57:06, 2.30s/it] 58%|█████▊ | 17550/30000 [11:16:33<7:55:10, 2.29s/it] 58%|█████▊ | 17550/30000 [11:16:33<7:55:10, 2.29s/it] 59%|█████▊ | 17551/30000 [11:16:36<7:56:22, 2.30s/it] 59%|█████▊ | 17552/30000 [11:16:38<7:58:51, 2.31s/it] 59%|█████▊ | 17553/30000 [11:16:40<7:57:59, 2.30s/it] 59%|█████▊ | 17554/30000 [11:16:43<7:56:23, 2.30s/it] 59%|█████▊ | 17555/30000 [11:16:45<7:54:43, 2.29s/it] 59%|█████▊ | 17556/30000 [11:16:47<7:56:51, 2.30s/it] 59%|█████▊ | 17557/30000 [11:16:49<7:56:29, 2.30s/it] 59%|█████▊ | 17558/30000 [11:16:52<7:52:53, 2.28s/it] 59%|█████▊ | 17559/30000 [11:16:54<7:52:05, 2.28s/it] 59%|█████▊ | 17560/30000 [11:16:56<7:57:35, 2.30s/it] 59%|█████▊ | 17560/30000 [11:16:56<7:57:35, 2.30s/it] 59%|█████▊ | 17561/30000 [11:16:59<8:03:34, 2.33s/it] 59%|█████▊ | 17562/30000 [11:17:01<8:10:56, 2.37s/it] 59%|█████▊ | 17563/30000 [11:17:04<8:16:27, 2.40s/it] 59%|█████▊ | 17564/30000 [11:17:06<8:19:24, 2.41s/it] 59%|█████▊ | 17565/30000 [11:17:09<8:23:09, 2.43s/it] 59%|█████▊ | 17566/30000 [11:17:11<8:27:13, 2.45s/it] 59%|█████▊ | 17567/30000 [11:17:13<8:28:39, 2.45s/it] 59%|█████▊ | 17568/30000 [11:17:16<8:30:15, 2.46s/it] 59%|█████▊ | 17569/30000 [11:17:18<8:31:29, 2.47s/it] 59%|█████▊ | 17570/30000 [11:17:21<8:31:00, 2.47s/it] 59%|█████▊ | 17570/30000 [11:17:21<8:31:00, 2.47s/it] 59%|█████▊ | 17571/30000 [11:17:23<8:31:15, 2.47s/it] 59%|█████▊ | 17572/30000 [11:17:26<8:29:39, 2.46s/it] 59%|█████▊ | 17573/30000 [11:17:28<8:27:41, 2.45s/it] 59%|█████▊ | 17574/30000 [11:17:30<8:08:00, 2.36s/it] 59%|█████▊ | 17575/30000 [11:17:32<6:57:30, 2.02s/it] 59%|█████▊ | 17576/30000 [11:17:33<6:08:02, 1.78s/it] 59%|█████▊ | 17577/30000 [11:17:34<5:32:44, 1.61s/it] 59%|█████▊ | 17578/30000 [11:17:35<5:10:34, 1.50s/it] 59%|█████▊ | 17579/30000 [11:17:37<4:54:52, 1.42s/it] 59%|█████▊ | 17580/30000 [11:17:38<4:39:59, 1.35s/it] 59%|█████▊ | 17580/30000 [11:17:38<4:39:59, 1.35s/it] 59%|█████▊ | 17581/30000 [11:17:39<4:30:19, 1.31s/it] 59%|█████▊ | 17582/30000 [11:17:40<4:22:59, 1.27s/it] 59%|█████▊ | 17583/30000 [11:17:41<4:18:01, 1.25s/it] 59%|█████▊ | 17584/30000 [11:17:42<4:14:52, 1.23s/it] 59%|█████▊ | 17585/30000 [11:17:44<4:11:28, 1.22s/it] 59%|█████▊ | 17586/30000 [11:17:45<4:10:19, 1.21s/it] 59%|█████▊ | 17587/30000 [11:17:46<4:10:03, 1.21s/it] 59%|█████▊ | 17588/30000 [11:17:47<4:09:32, 1.21s/it] 59%|█████▊ | 17589/30000 [11:17:48<4:09:20, 1.21s/it] 59%|█████▊ | 17590/30000 [11:17:50<4:08:45, 1.20s/it] 59%|█████▊ | 17590/30000 [11:17:50<4:08:45, 1.20s/it] 59%|█████▊ | 17591/30000 [11:17:51<4:08:50, 1.20s/it] 59%|█████▊ | 17592/30000 [11:17:52<4:08:46, 1.20s/it] 59%|█████▊ | 17593/30000 [11:17:53<4:08:10, 1.20s/it] 59%|█████▊ | 17594/30000 [11:17:54<4:06:37, 1.19s/it] 59%|█████▊ | 17595/30000 [11:17:56<4:09:26, 1.21s/it] 59%|█████▊ | 17596/30000 [11:17:57<4:10:00, 1.21s/it] 59%|█████▊ | 17597/30000 [11:17:58<4:09:36, 1.21s/it] 59%|█████▊ | 17598/30000 [11:17:59<4:08:49, 1.20s/it] 59%|█████▊ | 17599/30000 [11:18:01<4:08:17, 1.20s/it] 59%|█████▊ | 17600/30000 [11:18:03<5:01:06, 1.46s/it] 59%|█████▊ | 17600/30000 [11:18:03<5:01:06, 1.46s/it] 59%|█████▊ | 17601/30000 [11:18:04<4:44:45, 1.38s/it] 59%|█████▊ | 17602/30000 [11:18:05<4:32:07, 1.32s/it] 59%|█████▊ | 17603/30000 [11:18:06<4:27:19, 1.29s/it] 59%|█████▊ | 17604/30000 [11:18:07<4:21:40, 1.27s/it] 59%|█████▊ | 17605/30000 [11:18:09<4:18:37, 1.25s/it] 59%|█████▊ | 17606/30000 [11:18:10<4:16:10, 1.24s/it] 59%|█████▊ | 17607/30000 [11:18:11<4:14:19, 1.23s/it] 59%|█████▊ | 17608/30000 [11:18:17<8:47:52, 2.56s/it] 59%|█████▊ | 17609/30000 [11:18:18<7:22:18, 2.14s/it] 59%|█████▊ | 17610/30000 [11:18:19<6:23:40, 1.86s/it] 59%|█████▊ | 17610/30000 [11:18:19<6:23:40, 1.86s/it] 59%|█████▊ | 17611/30000 [11:18:20<5:43:02, 1.66s/it] 59%|█████▊ | 17612/30000 [11:18:21<5:12:53, 1.52s/it] 59%|█████▊ | 17613/30000 [11:18:23<4:53:05, 1.42s/it] 59%|█████▊ | 17614/30000 [11:18:24<4:39:48, 1.36s/it] 59%|█████▊ | 17615/30000 [11:18:25<4:30:24, 1.31s/it] 59%|█████▊ | 17616/30000 [11:18:31<9:19:29, 2.71s/it] 59%|█████▊ | 17617/30000 [11:18:32<7:43:56, 2.25s/it] 59%|█████▊ | 17618/30000 [11:18:33<6:38:15, 1.93s/it] 59%|█████▊ | 17619/30000 [11:18:35<5:52:43, 1.71s/it] 59%|█████▊ | 17620/30000 [11:18:36<5:18:49, 1.55s/it] 59%|█████▊ | 17620/30000 [11:18:36<5:18:49, 1.55s/it] 59%|█████▊ | 17621/30000 [11:18:37<4:57:53, 1.44s/it] 59%|█████▊ | 17622/30000 [11:18:38<4:41:50, 1.37s/it] 59%|█████▊ | 17623/30000 [11:18:39<4:30:41, 1.31s/it] 59%|█████▊ | 17624/30000 [11:18:46<10:00:42, 2.91s/it] 59%|█████▉ | 17625/30000 [11:18:47<8:13:31, 2.39s/it] 59%|█████▉ | 17626/30000 [11:18:48<6:57:56, 2.03s/it] 59%|█████▉ | 17627/30000 [11:18:49<6:07:07, 1.78s/it] 59%|█████▉ | 17628/30000 [11:18:51<5:31:24, 1.61s/it] 59%|█████▉ | 17629/30000 [11:18:52<5:07:46, 1.49s/it] 59%|█████▉ | 17630/30000 [11:18:53<4:51:56, 1.42s/it] 59%|█████▉ | 17630/30000 [11:18:53<4:51:56, 1.42s/it] 59%|█████▉ | 17631/30000 [11:18:54<4:41:25, 1.37s/it] 59%|█████▉ | 17632/30000 [11:19:01<10:28:47, 3.05s/it] 59%|█████▉ | 17633/30000 [11:19:03<8:33:06, 2.49s/it] 59%|█████▉ | 17634/30000 [11:19:04<7:11:04, 2.09s/it] 59%|█████▉ | 17635/30000 [11:19:05<6:15:50, 1.82s/it] 59%|█████▉ | 17636/30000 [11:19:06<5:34:43, 1.62s/it] 59%|█████▉ | 17637/30000 [11:19:07<5:07:49, 1.49s/it] 59%|█████▉ | 17638/30000 [11:19:08<4:49:20, 1.40s/it] 59%|█████▉ | 17639/30000 [11:19:10<4:35:29, 1.34s/it] 59%|█████▉ | 17640/30000 [11:19:17<10:43:20, 3.12s/it] 59%|█████▉ | 17640/30000 [11:19:17<10:43:20, 3.12s/it] 59%|█████▉ | 17641/30000 [11:19:18<8:42:22, 2.54s/it] 59%|█████▉ | 17642/30000 [11:19:19<7:18:08, 2.13s/it] 59%|█████▉ | 17643/30000 [11:19:20<6:19:51, 1.84s/it] 59%|█████▉ | 17644/30000 [11:19:22<5:38:01, 1.64s/it] 59%|█████▉ | 17645/30000 [11:19:23<5:09:14, 1.50s/it] 59%|█████▉ | 17646/30000 [11:19:24<4:49:09, 1.40s/it] 59%|█████▉ | 17647/30000 [11:19:25<4:36:01, 1.34s/it] 59%|█████▉ | 17648/30000 [11:19:33<11:10:02, 3.25s/it] 59%|█████▉ | 17649/30000 [11:19:34<9:01:06, 2.63s/it] 59%|█████▉ | 17650/30000 [11:19:35<7:30:47, 2.19s/it] 59%|█████▉ | 17650/30000 [11:19:35<7:30:47, 2.19s/it] 59%|█████▉ | 17651/30000 [11:19:36<6:30:09, 1.90s/it] 59%|█████▉ | 17652/30000 [11:19:38<5:45:25, 1.68s/it] 59%|█████▉ | 17653/30000 [11:19:39<5:15:42, 1.53s/it] 59%|█████▉ | 17654/30000 [11:19:40<4:53:47, 1.43s/it] 59%|█████▉ | 17655/30000 [11:19:41<4:38:05, 1.35s/it] 59%|█████▉ | 17656/30000 [11:19:49<10:55:49, 3.19s/it] 59%|█████▉ | 17657/30000 [11:19:50<8:52:29, 2.59s/it] 59%|█████▉ | 17658/30000 [11:19:51<7:26:26, 2.17s/it] 59%|█████▉ | 17659/30000 [11:19:52<6:26:35, 1.88s/it] 59%|█████▉ | 17660/30000 [11:19:53<5:42:32, 1.67s/it] 59%|█████▉ | 17660/30000 [11:19:53<5:42:32, 1.67s/it] 59%|█████▉ | 17661/30000 [11:19:55<5:13:04, 1.52s/it] 59%|█████▉ | 17662/30000 [11:19:56<4:52:21, 1.42s/it] 59%|█████▉ | 17663/30000 [11:19:57<4:37:44, 1.35s/it] 59%|█████▉ | 17664/30000 [11:20:04<10:21:10, 3.02s/it] 59%|█████▉ | 17665/30000 [11:20:05<8:26:32, 2.46s/it] 59%|█████▉ | 17666/30000 [11:20:06<7:06:32, 2.07s/it] 59%|█████▉ | 17667/30000 [11:20:07<6:11:41, 1.81s/it] 59%|█████▉ | 17668/30000 [11:20:09<5:32:21, 1.62s/it] 59%|█████▉ | 17669/30000 [11:20:10<5:05:51, 1.49s/it] 59%|█████▉ | 17670/30000 [11:20:11<4:47:20, 1.40s/it] 59%|█████▉ | 17670/30000 [11:20:11<4:47:20, 1.40s/it] 59%|█████▉ | 17671/30000 [11:20:12<4:35:07, 1.34s/it] 59%|█████▉ | 17672/30000 [11:20:20<10:55:44, 3.19s/it] 59%|█████▉ | 17673/30000 [11:20:21<9:28:49, 2.77s/it] 59%|█████▉ | 17674/30000 [11:20:23<7:50:13, 2.29s/it] 59%|█████▉ | 17675/30000 [11:20:24<6:41:42, 1.96s/it] 59%|█████▉ | 17676/30000 [11:20:25<5:53:40, 1.72s/it] 59%|█████▉ | 17677/30000 [11:20:26<5:20:30, 1.56s/it] 59%|█████▉ | 17678/30000 [11:20:27<4:58:16, 1.45s/it] 59%|█████▉ | 17679/30000 [11:20:29<4:42:05, 1.37s/it] 59%|█████▉ | 17680/30000 [11:20:36<11:15:34, 3.29s/it] 59%|█████▉ | 17680/30000 [11:20:36<11:15:34, 3.29s/it] 59%|█████▉ | 17681/30000 [11:20:38<9:17:21, 2.71s/it] 59%|█████▉ | 17682/30000 [11:20:39<7:42:41, 2.25s/it] 59%|█████▉ | 17683/30000 [11:20:40<6:37:05, 1.93s/it] 59%|█████▉ | 17684/30000 [11:20:41<5:50:20, 1.71s/it] 59%|█████▉ | 17685/30000 [11:20:42<5:19:40, 1.56s/it] 59%|█████▉ | 17686/30000 [11:20:44<4:56:27, 1.44s/it] 59%|█████▉ | 17687/30000 [11:20:45<4:39:53, 1.36s/it] 59%|█████▉ | 17688/30000 [11:20:52<10:53:49, 3.19s/it] 59%|█████▉ | 17689/30000 [11:20:54<9:16:47, 2.71s/it] 59%|█████▉ | 17690/30000 [11:20:55<7:41:45, 2.25s/it] 59%|█████▉ | 17690/30000 [11:20:55<7:41:45, 2.25s/it] 59%|█████▉ | 17691/30000 [11:20:56<6:37:01, 1.94s/it] 59%|█████▉ | 17692/30000 [11:20:57<5:49:29, 1.70s/it] 59%|█████▉ | 17693/30000 [11:20:59<5:16:40, 1.54s/it] 59%|█████▉ | 17694/30000 [11:21:00<4:54:33, 1.44s/it] 59%|█████▉ | 17695/30000 [11:21:01<4:38:01, 1.36s/it] 59%|█████▉ | 17696/30000 [11:21:09<11:10:15, 3.27s/it] 59%|█████▉ | 17697/30000 [11:21:10<9:01:04, 2.64s/it] 59%|█████▉ | 17698/30000 [11:21:11<7:30:39, 2.20s/it] 59%|█████▉ | 17699/30000 [11:21:12<6:28:35, 1.90s/it] 59%|█████▉ | 17700/30000 [11:21:13<5:44:14, 1.68s/it] 59%|█████▉ | 17700/30000 [11:21:13<5:44:14, 1.68s/it] 59%|█████▉ | 17701/30000 [11:21:14<5:13:56, 1.53s/it] 59%|█████▉ | 17702/30000 [11:21:16<4:52:04, 1.42s/it] 59%|█████▉ | 17703/30000 [11:21:17<4:36:08, 1.35s/it] 59%|█████▉ | 17704/30000 [11:21:24<11:01:56, 3.23s/it] 59%|█████▉ | 17705/30000 [11:21:26<9:09:54, 2.68s/it] 59%|█████▉ | 17706/30000 [11:21:27<7:36:59, 2.23s/it] 59%|█████▉ | 17707/30000 [11:21:28<6:32:40, 1.92s/it] 59%|█████▉ | 17708/30000 [11:21:29<5:46:01, 1.69s/it] 59%|█████▉ | 17709/30000 [11:21:31<5:15:36, 1.54s/it] 59%|█████▉ | 17710/30000 [11:21:32<4:53:59, 1.44s/it] 59%|█████▉ | 17710/30000 [11:21:32<4:53:59, 1.44s/it] 59%|█████▉ | 17711/30000 [11:21:33<4:38:12, 1.36s/it] 59%|█████▉ | 17712/30000 [11:21:41<11:49:23, 3.46s/it] 59%|█████▉ | 17713/30000 [11:21:43<9:33:53, 2.80s/it] 59%|█████▉ | 17714/30000 [11:21:44<7:53:02, 2.31s/it] 59%|█████▉ | 17715/30000 [11:21:45<6:44:41, 1.98s/it] 59%|█████▉ | 17716/30000 [11:21:46<5:54:31, 1.73s/it] 59%|█████▉ | 17717/30000 [11:21:47<5:21:24, 1.57s/it] 59%|█████▉ | 17718/30000 [11:21:48<4:58:20, 1.46s/it] 59%|█████▉ | 17719/30000 [11:21:50<4:41:02, 1.37s/it] 59%|█████▉ | 17720/30000 [11:21:57<10:40:14, 3.13s/it] 59%|█████▉ | 17720/30000 [11:21:57<10:40:14, 3.13s/it] 59%|█████▉ | 17721/30000 [11:21:58<9:04:45, 2.66s/it] 59%|█████▉ | 17722/30000 [11:22:00<7:32:44, 2.21s/it] 59%|█████▉ | 17723/30000 [11:22:01<6:29:12, 1.90s/it] 59%|█████▉ | 17724/30000 [11:22:02<5:44:28, 1.68s/it] 59%|█████▉ | 17725/30000 [11:22:03<5:14:06, 1.54s/it] 59%|█████▉ | 17726/30000 [11:22:04<4:52:43, 1.43s/it] 59%|█████▉ | 17727/30000 [11:22:06<4:37:10, 1.36s/it] 59%|█████▉ | 17728/30000 [11:22:13<11:08:14, 3.27s/it] 59%|█████▉ | 17729/30000 [11:22:15<9:22:16, 2.75s/it] 59%|█████▉ | 17730/30000 [11:22:16<7:45:00, 2.27s/it] 59%|█████▉ | 17730/30000 [11:22:16<7:45:00, 2.27s/it] 59%|█████▉ | 17731/30000 [11:22:17<6:38:55, 1.95s/it] 59%|█████▉ | 17732/30000 [11:22:18<5:50:38, 1.71s/it] 59%|█████▉ | 17733/30000 [11:22:20<5:18:13, 1.56s/it] 59%|█████▉ | 17734/30000 [11:22:21<4:56:37, 1.45s/it] 59%|█████▉ | 17735/30000 [11:22:22<4:44:11, 1.39s/it] 59%|█████▉ | 17736/30000 [11:22:30<11:36:02, 3.41s/it] 59%|█████▉ | 17737/30000 [11:22:32<9:45:40, 2.87s/it] 59%|█████▉ | 17738/30000 [11:22:33<8:01:09, 2.35s/it] 59%|█████▉ | 17739/30000 [11:22:34<6:50:04, 2.01s/it] 59%|█████▉ | 17740/30000 [11:22:35<5:58:45, 1.76s/it] 59%|█████▉ | 17740/30000 [11:22:35<5:58:45, 1.76s/it] 59%|█████▉ | 17741/30000 [11:22:36<5:24:35, 1.59s/it] 59%|█████▉ | 17742/30000 [11:22:38<4:59:43, 1.47s/it] 59%|█████▉ | 17743/30000 [11:22:39<4:40:43, 1.37s/it] 59%|█████▉ | 17744/30000 [11:22:46<10:39:17, 3.13s/it] 59%|█████▉ | 17745/30000 [11:22:48<9:07:55, 2.68s/it] 59%|█████▉ | 17746/30000 [11:22:49<7:36:56, 2.24s/it] 59%|█████▉ | 17747/30000 [11:22:50<6:33:06, 1.92s/it] 59%|█████▉ | 17748/30000 [11:22:51<5:46:27, 1.70s/it] 59%|█████▉ | 17749/30000 [11:22:52<5:15:01, 1.54s/it] 59%|█████▉ | 17750/30000 [11:22:54<4:53:46, 1.44s/it] 59%|█████▉ | 17750/30000 [11:22:54<4:53:46, 1.44s/it] 59%|█████▉ | 17751/30000 [11:22:55<4:38:52, 1.37s/it] 59%|█████▉ | 17752/30000 [11:23:02<11:01:55, 3.24s/it] 59%|█████▉ | 17753/30000 [11:23:05<10:14:07, 3.01s/it] 59%|█████▉ | 17754/30000 [11:23:06<8:23:15, 2.47s/it] 59%|█████▉ | 17755/30000 [11:23:07<7:05:03, 2.08s/it] 59%|█████▉ | 17756/30000 [11:23:08<6:08:56, 1.81s/it] 59%|█████▉ | 17757/30000 [11:23:10<5:30:28, 1.62s/it] 59%|█████▉ | 17758/30000 [11:23:11<5:04:16, 1.49s/it] 59%|█████▉ | 17759/30000 [11:23:12<4:45:43, 1.40s/it] 59%|█████▉ | 17760/30000 [11:23:19<11:00:42, 3.24s/it] 59%|█████▉ | 17760/30000 [11:23:19<11:00:42, 3.24s/it] 59%|█████▉ | 17761/30000 [11:23:22<10:36:52, 3.12s/it] 59%|█████▉ | 17762/30000 [11:23:23<8:36:11, 2.53s/it] 59%|█████▉ | 17763/30000 [11:23:25<7:13:53, 2.13s/it] 59%|█████▉ | 17764/30000 [11:23:26<6:15:43, 1.84s/it] 59%|█████▉ | 17765/30000 [11:23:27<5:35:44, 1.65s/it] 59%|█████▉ | 17766/30000 [11:23:28<5:08:35, 1.51s/it] 59%|█████▉ | 17767/30000 [11:23:29<4:49:14, 1.42s/it] 59%|█████▉ | 17768/30000 [11:23:36<9:43:48, 2.86s/it] 59%|█████▉ | 17769/30000 [11:23:39<10:04:36, 2.97s/it] 59%|█████▉ | 17770/30000 [11:23:40<8:14:12, 2.42s/it] 59%|█████▉ | 17770/30000 [11:23:40<8:14:12, 2.42s/it] 59%|█████▉ | 17771/30000 [11:23:41<6:58:37, 2.05s/it] 59%|█████▉ | 17772/30000 [11:23:42<6:04:54, 1.79s/it] 59%|█████▉ | 17773/30000 [11:23:44<5:28:33, 1.61s/it] 59%|█████▉ | 17774/30000 [11:23:45<5:04:33, 1.49s/it] 59%|█████▉ | 17775/30000 [11:23:46<4:45:41, 1.40s/it] 59%|█████▉ | 17776/30000 [11:23:52<9:07:31, 2.69s/it] 59%|█████▉ | 17777/30000 [11:23:55<9:39:33, 2.84s/it] 59%|█████▉ | 17778/30000 [11:23:56<7:57:10, 2.34s/it] 59%|█████▉ | 17779/30000 [11:23:57<6:47:08, 2.00s/it] 59%|█████▉ | 17780/30000 [11:23:58<5:56:40, 1.75s/it] +{'loss': 0.0151, 'grad_norm': 0.06719441711902618, 'learning_rate': 1.2320949339062008e-05, 'flow_matching_loss': 0.015120538789778948, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.06} +{'loss': 0.0152, 'grad_norm': 0.10341599583625793, 'learning_rate': 1.2304682130726281e-05, 'flow_matching_loss': 0.015150679368525743, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.06} +{'loss': 0.015, 'grad_norm': 0.07313315570354462, 'learning_rate': 1.2288418197455913e-05, 'flow_matching_loss': 0.014969553798437119, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.06} +{'loss': 0.0147, 'grad_norm': 0.09451326727867126, 'learning_rate': 1.2272157559013126e-05, 'flow_matching_loss': 0.01470245597884059, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.07} +{'loss': 0.0141, 'grad_norm': 0.07537028193473816, 'learning_rate': 1.2255900235156123e-05, 'flow_matching_loss': 0.014056878909468652, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.07} +{'loss': 0.0139, 'grad_norm': 0.09515228867530823, 'learning_rate': 1.2239646245639087e-05, 'flow_matching_loss': 0.013893928192555905, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.07} +{'loss': 0.0138, 'grad_norm': 0.09155604243278503, 'learning_rate': 1.2223395610212142e-05, 'flow_matching_loss': 0.013839660491794348, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.07} +{'loss': 0.0144, 'grad_norm': 0.060941897332668304, 'learning_rate': 1.2207148348621348e-05, 'flow_matching_loss': 0.014405519422143697, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.08} +{'loss': 0.014, 'grad_norm': 0.06991862505674362, 'learning_rate': 1.219090448060866e-05, 'flow_matching_loss': 0.014006106834858655, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.08} +{'loss': 0.0151, 'grad_norm': 0.07323029637336731, 'learning_rate': 1.217466402591191e-05, 'flow_matching_loss': 0.015146032441407442, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.08} +{'loss': 0.0138, 'grad_norm': 0.06799393147230148, 'learning_rate': 1.2158427004264785e-05, 'flow_matching_loss': 0.013844878878444433, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.09} +{'loss': 0.0145, 'grad_norm': 0.07743974775075912, 'learning_rate': 1.2142193435396788e-05, 'flow_matching_loss': 0.01452613091096282, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.09} +{'loss': 0.0137, 'grad_norm': 0.06681108474731445, 'learning_rate': 1.2125963339033234e-05, 'flow_matching_loss': 0.01370912017300725, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.09} +{'loss': 0.0139, 'grad_norm': 0.06763824075460434, 'learning_rate': 1.2109736734895237e-05, 'flow_matching_loss': 0.013859811332076788, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.09} +{'loss': 0.0145, 'grad_norm': 0.07905834168195724, 'learning_rate': 1.2093513642699634e-05, 'flow_matching_loss': 0.014519299659878016, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.1} +{'loss': 0.0146, 'grad_norm': 0.07283028960227966, 'learning_rate': 1.2077294082159024e-05, 'flow_matching_loss': 0.014625543542206288, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.1} +{'loss': 0.0147, 'grad_norm': 0.0842815563082695, 'learning_rate': 1.2061078072981697e-05, 'flow_matching_loss': 0.014715418498963118, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.1} +{'loss': 0.0138, 'grad_norm': 0.09055306762456894, 'learning_rate': 1.2044865634871626e-05, 'flow_matching_loss': 0.013832768145948648, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.11} +{'loss': 0.0145, 'grad_norm': 0.0697324350476265, 'learning_rate': 1.2028656787528468e-05, 'flow_matching_loss': 0.014508138597011565, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.11} +{'loss': 0.0128, 'grad_norm': 0.07877586781978607, 'learning_rate': 1.2012451550647491e-05, 'flow_matching_loss': 0.012834996730089188, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.11} +{'loss': 0.0127, 'grad_norm': 0.07297014445066452, 'learning_rate': 1.1996249943919589e-05, 'flow_matching_loss': 0.012734392285346985, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.11} +{'loss': 0.0127, 'grad_norm': 0.0695299282670021, 'learning_rate': 1.1980051987031237e-05, 'flow_matching_loss': 0.012701442185789346, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.12} +{'loss': 0.0128, 'grad_norm': 0.08383118361234665, 'learning_rate': 1.1963857699664483e-05, 'flow_matching_loss': 0.01277652671560645, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.12} +{'loss': 0.0152, 'grad_norm': 0.07866144180297852, 'learning_rate': 1.1947667101496911e-05, 'flow_matching_loss': 0.015230201464146376, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.12} +{'loss': 0.0145, 'grad_norm': 0.07512625306844711, 'learning_rate': 1.1931480212201633e-05, 'flow_matching_loss': 0.014541693776845933, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.13} +{'loss': 0.0128, 'grad_norm': 0.07307911664247513, 'learning_rate': 1.1915297051447237e-05, 'flow_matching_loss': 0.012751576118171215, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.13} +{'loss': 0.0133, 'grad_norm': 0.07951916009187698, 'learning_rate': 1.1899117638897793e-05, 'flow_matching_loss': 0.01326951738446951, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.13} +{'loss': 0.0145, 'grad_norm': 0.06883057206869125, 'learning_rate': 1.1882941994212811e-05, 'flow_matching_loss': 0.0145078394562006, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.14} +{'loss': 0.0144, 'grad_norm': 0.0774250403046608, 'learning_rate': 1.1866770137047221e-05, 'flow_matching_loss': 0.014405203424394131, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.14} +{'loss': 0.015, 'grad_norm': 0.0801144614815712, 'learning_rate': 1.1850602087051363e-05, 'flow_matching_loss': 0.014974320307374, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.14} +{'loss': 0.0137, 'grad_norm': 0.07801534235477448, 'learning_rate': 1.183443786387094e-05, 'flow_matching_loss': 0.013733332697302102, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.14} +{'loss': 0.0147, 'grad_norm': 0.0861586406826973, 'learning_rate': 1.1818277487147005e-05, 'flow_matching_loss': 0.014694625418633223, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.15} +{'loss': 0.0141, 'grad_norm': 0.06927123665809631, 'learning_rate': 1.180212097651594e-05, 'flow_matching_loss': 0.014064605440944433, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.15} +{'loss': 0.0153, 'grad_norm': 0.07302751392126083, 'learning_rate': 1.1785968351609422e-05, 'flow_matching_loss': 0.015349931176751852, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.15} +{'loss': 0.0138, 'grad_norm': 0.06895443797111511, 'learning_rate': 1.1769819632054427e-05, 'flow_matching_loss': 0.013842426054179669, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.16} +{'loss': 0.0143, 'grad_norm': 0.07778438925743103, 'learning_rate': 1.1753674837473165e-05, 'flow_matching_loss': 0.014253380242735147, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.16} +{'loss': 0.0131, 'grad_norm': 0.059446923434734344, 'learning_rate': 1.1737533987483081e-05, 'flow_matching_loss': 0.013100327830761671, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.16} +{'loss': 0.0138, 'grad_norm': 0.0643453598022461, 'learning_rate': 1.1721397101696835e-05, 'flow_matching_loss': 0.013806027546525, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.16} +{'loss': 0.0137, 'grad_norm': 0.07297095656394958, 'learning_rate': 1.1705264199722257e-05, 'flow_matching_loss': 0.013737361878156662, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.17} +{'loss': 0.0155, 'grad_norm': 0.07082313299179077, 'learning_rate': 1.1689135301162356e-05, 'flow_matching_loss': 0.015510430093854666, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.17} +{'loss': 0.0133, 'grad_norm': 0.07747276872396469, 'learning_rate': 1.1673010425615258e-05, 'flow_matching_loss': 0.013255203794687986, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.17} 59%|█████▉ | 17780/30000 [11:23:58<5:56:40, 1.75s/it] 59%|█████▉ | 17781/30000 [11:24:00<5:22:34, 1.58s/it] 59%|█████▉ | 17782/30000 [11:24:01<4:58:50, 1.47s/it] 59%|█████▉ | 17783/30000 [11:24:02<4:41:14, 1.38s/it] 59%|█████▉ | 17784/30000 [11:24:08<9:44:57, 2.87s/it] 59%|█████▉ | 17785/30000 [11:24:10<8:55:41, 2.63s/it] 59%|█████▉ | 17786/30000 [11:24:12<7:25:58, 2.19s/it] 59%|█████▉ | 17787/30000 [11:24:13<6:24:29, 1.89s/it] 59%|█████▉ | 17788/30000 [11:24:14<5:40:13, 1.67s/it] 59%|█████▉ | 17789/30000 [11:24:15<5:10:54, 1.53s/it] 59%|█████▉ | 17790/30000 [11:24:16<4:49:53, 1.42s/it] 59%|█████▉ | 17790/30000 [11:24:16<4:49:53, 1.42s/it] 59%|█████▉ | 17791/30000 [11:24:18<4:35:52, 1.36s/it] 59%|█████▉ | 17792/30000 [11:24:23<8:57:18, 2.64s/it] 59%|█████▉ | 17793/30000 [11:24:26<8:48:36, 2.60s/it] 59%|█████▉ | 17794/30000 [11:24:27<7:20:36, 2.17s/it] 59%|█████▉ | 17795/30000 [11:24:28<6:20:48, 1.87s/it] 59%|█████▉ | 17796/30000 [11:24:29<5:39:43, 1.67s/it] 59%|█████▉ | 17797/30000 [11:24:30<5:11:50, 1.53s/it] 59%|█████▉ | 17798/30000 [11:24:32<4:51:02, 1.43s/it] 59%|█████▉ | 17799/30000 [11:24:33<4:36:05, 1.36s/it] 59%|█████▉ | 17800/30000 [11:24:38<8:50:03, 2.61s/it] 59%|█████▉ | 17800/30000 [11:24:38<8:50:03, 2.61s/it] 59%|█████▉ | 17801/30000 [11:24:40<8:14:30, 2.43s/it] 59%|█████▉ | 17802/30000 [11:24:42<6:57:25, 2.05s/it] 59%|█████▉ | 17803/30000 [11:24:43<6:04:01, 1.79s/it] 59%|█████▉ | 17804/30000 [11:24:44<5:26:13, 1.60s/it] 59%|█████▉ | 17805/30000 [11:24:45<5:00:53, 1.48s/it] 59%|█████▉ | 17806/30000 [11:24:46<4:43:09, 1.39s/it] 59%|█████▉ | 17807/30000 [11:24:47<4:29:53, 1.33s/it] 59%|█████▉ | 17808/30000 [11:24:54<9:42:52, 2.87s/it] 59%|█████▉ | 17809/30000 [11:24:56<8:57:36, 2.65s/it] 59%|█████▉ | 17810/30000 [11:24:57<7:28:58, 2.21s/it] 59%|█████▉ | 17810/30000 [11:24:57<7:28:58, 2.21s/it] 59%|█████▉ | 17811/30000 [11:24:58<6:26:29, 1.90s/it] 59%|█████▉ | 17812/30000 [11:25:00<5:42:09, 1.68s/it] 59%|█████▉ | 17813/30000 [11:25:01<5:10:01, 1.53s/it] 59%|█████▉ | 17814/30000 [11:25:02<4:50:23, 1.43s/it] 59%|█████▉ | 17815/30000 [11:25:03<4:35:00, 1.35s/it] 59%|█████▉ | 17816/30000 [11:25:09<9:08:50, 2.70s/it] 59%|█████▉ | 17817/30000 [11:25:12<9:07:10, 2.69s/it] 59%|█████▉ | 17818/30000 [11:25:13<7:34:26, 2.24s/it] 59%|█████▉ | 17819/30000 [11:25:14<6:30:57, 1.93s/it] 59%|█████▉ | 17820/30000 [11:25:15<5:46:35, 1.71s/it] 59%|█████▉ | 17820/30000 [11:25:15<5:46:35, 1.71s/it] 59%|█████▉ | 17821/30000 [11:25:16<5:15:07, 1.55s/it] 59%|█████▉ | 17822/30000 [11:25:18<4:53:03, 1.44s/it] 59%|█████▉ | 17823/30000 [11:25:19<4:37:40, 1.37s/it] 59%|█████▉ | 17824/30000 [11:25:26<10:15:21, 3.03s/it] 59%|█████▉ | 17825/30000 [11:25:27<8:23:38, 2.48s/it] 59%|█████▉ | 17826/30000 [11:25:28<7:03:46, 2.09s/it] 59%|█████▉ | 17827/30000 [11:25:29<6:08:54, 1.82s/it] 59%|█████▉ | 17828/30000 [11:25:30<5:29:40, 1.63s/it] 59%|█████▉ | 17829/30000 [11:25:32<5:02:38, 1.49s/it] 59%|█████▉ | 17830/30000 [11:25:33<4:42:31, 1.39s/it] 59%|█████▉ | 17830/30000 [11:25:33<4:42:31, 1.39s/it] 59%|█████▉ | 17831/30000 [11:25:34<4:29:35, 1.33s/it] 59%|█████▉ | 17832/30000 [11:25:41<10:12:57, 3.02s/it] 59%|█████▉ | 17833/30000 [11:25:42<8:45:02, 2.59s/it] 59%|█████▉ | 17834/30000 [11:25:44<7:18:13, 2.16s/it] 59%|█████▉ | 17835/30000 [11:25:45<6:20:15, 1.88s/it] 59%|█████▉ | 17836/30000 [11:25:46<5:38:06, 1.67s/it] 59%|█████▉ | 17837/30000 [11:25:47<5:09:07, 1.52s/it] 59%|█████▉ | 17838/30000 [11:25:48<4:47:17, 1.42s/it] 59%|█████▉ | 17839/30000 [11:25:50<4:32:35, 1.34s/it] 59%|█████▉ | 17840/30000 [11:25:56<9:19:56, 2.76s/it] 59%|█████▉ | 17840/30000 [11:25:56<9:19:56, 2.76s/it] 59%|█████▉ | 17841/30000 [11:25:57<7:43:49, 2.29s/it] 59%|█████▉ | 17842/30000 [11:25:58<6:36:11, 1.96s/it] 59%|█████▉ | 17843/30000 [11:25:59<5:49:44, 1.73s/it] 59%|█████▉ | 17844/30000 [11:26:00<5:16:16, 1.56s/it] 59%|█████▉ | 17845/30000 [11:26:02<4:52:36, 1.44s/it] 59%|█████▉ | 17846/30000 [11:26:03<4:36:45, 1.37s/it] 59%|█████▉ | 17847/30000 [11:26:04<4:25:36, 1.31s/it] 59%|█████▉ | 17848/30000 [11:26:11<9:54:10, 2.93s/it] 59%|█████▉ | 17849/30000 [11:26:12<8:06:56, 2.40s/it] 60%|█████▉ | 17850/30000 [11:26:13<6:51:42, 2.03s/it] 60%|█████▉ | 17850/30000 [11:26:13<6:51:42, 2.03s/it] 60%|█████▉ | 17851/30000 [11:26:14<6:02:55, 1.79s/it] 60%|█████▉ | 17852/30000 [11:26:15<5:27:33, 1.62s/it] 60%|█████▉ | 17853/30000 [11:26:17<5:01:06, 1.49s/it] 60%|█████▉ | 17854/30000 [11:26:18<4:42:44, 1.40s/it] 60%|█████▉ | 17855/30000 [11:26:19<4:30:31, 1.34s/it] 60%|█████▉ | 17856/30000 [11:26:25<9:18:59, 2.76s/it] 60%|█████▉ | 17857/30000 [11:26:26<7:42:43, 2.29s/it] 60%|█████▉ | 17858/30000 [11:26:27<6:35:51, 1.96s/it] 60%|█████▉ | 17859/30000 [11:26:29<5:50:16, 1.73s/it] 60%|█████▉ | 17860/30000 [11:26:30<5:15:53, 1.56s/it] 60%|█████▉ | 17860/30000 [11:26:30<5:15:53, 1.56s/it] 60%|█████▉ | 17861/30000 [11:26:31<4:54:45, 1.46s/it] 60%|█████▉ | 17862/30000 [11:26:32<4:37:53, 1.37s/it] 60%|█████▉ | 17863/30000 [11:26:33<4:26:23, 1.32s/it] 60%|█████▉ | 17864/30000 [11:26:40<9:38:43, 2.86s/it] 60%|█████▉ | 17865/30000 [11:26:41<7:58:46, 2.37s/it] 60%|█████▉ | 17866/30000 [11:26:42<6:46:20, 2.01s/it] 60%|█████▉ | 17867/30000 [11:26:43<5:56:19, 1.76s/it] 60%|█████▉ | 17868/30000 [11:26:45<5:20:39, 1.59s/it] 60%|█████▉ | 17869/30000 [11:26:46<4:56:02, 1.46s/it] 60%|█████▉ | 17870/30000 [11:26:47<4:38:18, 1.38s/it] 60%|█████▉ | 17870/30000 [11:26:47<4:38:18, 1.38s/it] 60%|█████▉ | 17871/30000 [11:26:48<4:25:30, 1.31s/it] 60%|█████▉ | 17872/30000 [11:26:55<9:42:44, 2.88s/it] 60%|█████▉ | 17873/30000 [11:26:56<8:02:22, 2.39s/it] 60%|█████▉ | 17874/30000 [11:26:57<6:49:24, 2.03s/it] 60%|█████▉ | 17875/30000 [11:26:58<5:59:22, 1.78s/it] 60%|█████▉ | 17876/30000 [11:26:59<5:23:26, 1.60s/it] 60%|█████▉ | 17877/30000 [11:27:01<4:59:35, 1.48s/it] 60%|█████▉ | 17878/30000 [11:27:02<4:42:06, 1.40s/it] 60%|█████▉ | 17879/30000 [11:27:03<4:29:12, 1.33s/it] 60%|█████▉ | 17880/30000 [11:27:10<10:01:37, 2.98s/it] 60%|█████▉ | 17880/30000 [11:27:10<10:01:37, 2.98s/it] 60%|█████▉ | 17881/30000 [11:27:11<8:21:19, 2.48s/it] 60%|█████▉ | 17882/30000 [11:27:12<7:02:02, 2.09s/it] 60%|█████▉ | 17883/30000 [11:27:14<6:08:19, 1.82s/it] 60%|█████▉ | 17884/30000 [11:27:15<5:28:07, 1.62s/it] 60%|█████▉ | 17885/30000 [11:27:16<5:00:20, 1.49s/it] 60%|█████▉ | 17886/30000 [11:27:17<4:41:50, 1.40s/it] 60%|█████▉ | 17887/30000 [11:27:18<4:29:24, 1.33s/it] 60%|█████▉ | 17888/30000 [11:27:25<10:18:20, 3.06s/it] 60%|█████▉ | 17889/30000 [11:27:27<9:15:48, 2.75s/it] 60%|█████▉ | 17890/30000 [11:27:29<7:40:24, 2.28s/it] 60%|█████▉ | 17890/30000 [11:27:29<7:40:24, 2.28s/it] 60%|█████▉ | 17891/30000 [11:27:30<6:34:22, 1.95s/it] 60%|█████▉ | 17892/30000 [11:27:31<5:46:21, 1.72s/it] 60%|█████▉ | 17893/30000 [11:27:32<5:13:51, 1.56s/it] 60%|█████▉ | 17894/30000 [11:27:33<4:50:38, 1.44s/it] 60%|█████▉ | 17895/30000 [11:27:34<4:34:06, 1.36s/it] 60%|█████▉ | 17896/30000 [11:27:42<10:20:23, 3.08s/it] 60%|█████▉ | 17897/30000 [11:27:44<9:49:04, 2.92s/it] 60%|█████▉ | 17898/30000 [11:27:45<8:03:11, 2.40s/it] 60%|█████▉ | 17899/30000 [11:27:46<6:50:20, 2.03s/it] 60%|█████▉ | 17900/30000 [11:27:48<5:58:32, 1.78s/it] 60%|█████▉ | 17900/30000 [11:27:48<5:58:32, 1.78s/it] 60%|█████▉ | 17901/30000 [11:27:49<5:21:38, 1.60s/it] 60%|█████▉ | 17902/30000 [11:27:50<4:55:45, 1.47s/it] 60%|█████▉ | 17903/30000 [11:27:51<4:37:23, 1.38s/it] 60%|█████▉ | 17904/30000 [11:27:58<10:25:42, 3.10s/it] 60%|█████▉ | 17905/30000 [11:28:00<8:39:05, 2.58s/it] 60%|█████▉ | 17906/30000 [11:28:01<7:14:57, 2.16s/it] 60%|█████▉ | 17907/30000 [11:28:02<6:16:11, 1.87s/it] 60%|█████▉ | 17908/30000 [11:28:03<5:34:37, 1.66s/it] 60%|█████▉ | 17909/30000 [11:28:04<5:05:55, 1.52s/it] 60%|█████▉ | 17910/30000 [11:28:06<4:46:06, 1.42s/it] 60%|█████▉ | 17910/30000 [11:28:06<4:46:06, 1.42s/it] 60%|█████▉ | 17911/30000 [11:28:07<4:32:16, 1.35s/it] 60%|█████▉ | 17912/30000 [11:28:13<9:55:09, 2.95s/it] 60%|█████▉ | 17913/30000 [11:28:15<8:07:00, 2.42s/it] 60%|█████▉ | 17914/30000 [11:28:16<6:51:21, 2.04s/it] 60%|█████▉ | 17915/30000 [11:28:17<5:59:39, 1.79s/it] 60%|█████▉ | 17916/30000 [11:28:18<5:22:17, 1.60s/it] 60%|█████▉ | 17917/30000 [11:28:19<4:56:18, 1.47s/it] 60%|█████▉ | 17918/30000 [11:28:20<4:37:59, 1.38s/it] 60%|█████▉ | 17919/30000 [11:28:22<4:25:39, 1.32s/it] 60%|█████▉ | 17920/30000 [11:28:29<10:24:49, 3.10s/it] 60%|█████▉ | 17920/30000 [11:28:29<10:24:49, 3.10s/it] 60%|█████▉ | 17921/30000 [11:28:30<8:29:29, 2.53s/it] 60%|█████▉ | 17922/30000 [11:28:31<7:07:09, 2.12s/it] 60%|█████▉ | 17923/30000 [11:28:32<6:11:15, 1.84s/it] 60%|█████▉ | 17924/30000 [11:28:34<5:31:16, 1.65s/it] 60%|█████▉ | 17925/30000 [11:28:35<5:03:59, 1.51s/it] 60%|█████▉ | 17926/30000 [11:28:36<4:43:19, 1.41s/it] 60%|█████▉ | 17927/30000 [11:28:37<4:29:30, 1.34s/it] 60%|█████▉ | 17928/30000 [11:28:44<9:43:25, 2.90s/it] 60%|█████▉ | 17929/30000 [11:28:45<8:00:10, 2.39s/it] 60%|█████▉ | 17930/30000 [11:28:46<6:47:39, 2.03s/it] 60%|█████▉ | 17930/30000 [11:28:46<6:47:39, 2.03s/it] 60%|█████▉ | 17931/30000 [11:28:47<5:57:47, 1.78s/it] 60%|█████▉ | 17932/30000 [11:28:48<5:21:09, 1.60s/it] 60%|█████▉ | 17933/30000 [11:28:50<4:56:02, 1.47s/it] 60%|█████▉ | 17934/30000 [11:28:51<4:37:48, 1.38s/it] 60%|█████▉ | 17935/30000 [11:28:52<4:24:55, 1.32s/it] 60%|█████▉ | 17936/30000 [11:28:59<9:41:37, 2.89s/it] 60%|█████▉ | 17937/30000 [11:29:00<7:57:54, 2.38s/it] 60%|█████▉ | 17938/30000 [11:29:01<6:44:38, 2.01s/it] 60%|█████▉ | 17939/30000 [11:29:02<5:55:33, 1.77s/it] 60%|█████▉ | 17940/30000 [11:29:03<5:20:29, 1.59s/it] 60%|█████▉ | 17940/30000 [11:29:03<5:20:29, 1.59s/it] 60%|█████▉ | 17941/30000 [11:29:04<4:56:18, 1.47s/it] 60%|█████▉ | 17942/30000 [11:29:06<4:38:01, 1.38s/it] 60%|█████▉ | 17943/30000 [11:29:07<4:25:15, 1.32s/it] 60%|█████▉ | 17944/30000 [11:29:13<9:25:36, 2.81s/it] 60%|█████▉ | 17945/30000 [11:29:14<7:49:50, 2.34s/it] 60%|█████▉ | 17946/30000 [11:29:16<6:40:55, 2.00s/it] 60%|█████▉ | 17947/30000 [11:29:17<5:52:21, 1.75s/it] 60%|█████▉ | 17948/30000 [11:29:18<5:17:17, 1.58s/it] 60%|█████▉ | 17949/30000 [11:29:19<4:53:42, 1.46s/it] 60%|█████▉ | 17950/30000 [11:29:20<4:36:57, 1.38s/it] 60%|█████▉ | 17950/30000 [11:29:20<4:36:57, 1.38s/it] 60%|█████▉ | 17951/30000 [11:29:21<4:25:17, 1.32s/it] 60%|█████▉ | 17952/30000 [11:29:29<10:32:22, 3.15s/it] 60%|█████▉ | 17953/30000 [11:29:30<8:34:37, 2.56s/it] 60%|█████▉ | 17954/30000 [11:29:31<7:09:48, 2.14s/it] 60%|█████▉ | 17955/30000 [11:29:32<6:12:11, 1.85s/it] 60%|█████▉ | 17956/30000 [11:29:34<5:30:46, 1.65s/it] 60%|█████▉ | 17957/30000 [11:29:35<5:01:56, 1.50s/it] 60%|█████▉ | 17958/30000 [11:29:36<4:42:16, 1.41s/it] 60%|█████▉ | 17959/30000 [11:29:37<4:27:44, 1.33s/it] 60%|█████▉ | 17960/30000 [11:29:45<10:50:07, 3.24s/it] 60%|█████▉ | 17960/30000 [11:29:45<10:50:07, 3.24s/it] 60%|█████▉ | 17961/30000 [11:29:46<8:45:43, 2.62s/it] 60%|█████▉ | 17962/30000 [11:29:47<7:17:36, 2.18s/it] 60%|█████▉ | 17963/30000 [11:29:48<6:18:43, 1.89s/it] 60%|█████▉ | 17964/30000 [11:29:49<5:34:48, 1.67s/it] 60%|█████▉ | 17965/30000 [11:29:51<5:06:02, 1.53s/it] 60%|█████▉ | 17966/30000 [11:29:52<4:45:23, 1.42s/it] 60%|█████▉ | 17967/30000 [11:29:53<4:30:29, 1.35s/it] 60%|█████▉ | 17968/30000 [11:30:00<10:09:20, 3.04s/it] 60%|█████▉ | 17969/30000 [11:30:01<8:16:37, 2.48s/it] 60%|█████▉ | 17970/30000 [11:30:02<6:58:15, 2.09s/it] 60%|█████▉ | 17970/30000 [11:30:02<6:58:15, 2.09s/it] 60%|█████▉ | 17971/30000 [11:30:04<6:04:34, 1.82s/it] 60%|█████▉ | 17972/30000 [11:30:05<5:26:25, 1.63s/it] 60%|█████▉ | 17973/30000 [11:30:06<5:00:48, 1.50s/it] 60%|█████▉ | 17974/30000 [11:30:07<4:40:34, 1.40s/it] 60%|█████▉ | 17975/30000 [11:30:08<4:27:54, 1.34s/it] 60%|█████▉ | 17976/30000 [11:30:15<9:51:18, 2.95s/it] 60%|█████▉ | 17977/30000 [11:30:16<8:05:13, 2.42s/it] 60%|█████▉ | 17978/30000 [11:30:17<6:50:13, 2.05s/it] 60%|█████▉ | 17979/30000 [11:30:19<5:58:01, 1.79s/it] 60%|█████▉ | 17980/30000 [11:30:20<5:21:17, 1.60s/it] 60%|█████▉ | 17980/30000 [11:30:20<5:21:17, 1.60s/it] 60%|█████▉ | 17981/30000 [11:30:21<4:56:13, 1.48s/it] 60%|█████▉ | 17982/30000 [11:30:22<4:38:23, 1.39s/it] 60%|█████▉ | 17983/30000 [11:30:23<4:25:29, 1.33s/it] 60%|█████▉ | 17984/30000 [11:30:30<10:05:03, 3.02s/it] 60%|█████▉ | 17985/30000 [11:30:31<8:15:29, 2.47s/it] 60%|█████▉ | 17986/30000 [11:30:33<6:56:16, 2.08s/it] 60%|█████▉ | 17987/30000 [11:30:34<6:02:46, 1.81s/it] 60%|█████▉ | 17988/30000 [11:30:35<5:24:34, 1.62s/it] 60%|█████▉ | 17989/30000 [11:30:36<4:58:05, 1.49s/it] 60%|█████▉ | 17990/30000 [11:30:37<4:39:25, 1.40s/it] 60%|█████▉ | 17990/30000 [11:30:37<4:39:25, 1.40s/it] 60%|█████▉ | 17991/30000 [11:30:38<4:25:58, 1.33s/it] 60%|█████▉ | 17992/30000 [11:30:45<9:55:48, 2.98s/it] 60%|█████▉ | 17993/30000 [11:30:47<8:12:33, 2.46s/it] 60%|█████▉ | 17994/30000 [11:30:48<6:55:17, 2.08s/it] 60%|█████▉ | 17995/30000 [11:30:49<6:02:20, 1.81s/it] 60%|█████▉ | 17996/30000 [11:30:50<5:23:48, 1.62s/it] 60%|█████▉ | 17997/30000 [11:30:51<4:58:22, 1.49s/it] 60%|█████▉ | 17998/30000 [11:30:52<4:38:29, 1.39s/it] 60%|█████▉ | 17999/30000 [11:30:54<4:27:46, 1.34s/it] 60%|██████ | 18000/30000 [11:31:01<10:39:51, 3.20s/it] 60%|██████ | 18000/30000 [11:31:01<10:39:51, 3.20s/it] 60%|██████ | 18001/30000 [11:31:02<8:37:47, 2.59s/it] 60%|██████ | 18002/30000 [11:31:04<7:12:05, 2.16s/it] 60%|██████ | 18003/30000 [11:31:05<6:13:44, 1.87s/it] 60%|██████ | 18004/30000 [11:31:06<5:31:40, 1.66s/it] 60%|██████ | 18005/30000 [11:31:07<5:03:43, 1.52s/it] 60%|██████ | 18006/30000 [11:31:08<4:44:42, 1.42s/it] 60%|██████ | 18007/30000 [11:31:09<4:31:07, 1.36s/it] 60%|██████ | 18008/30000 [11:31:17<10:40:42, 3.21s/it] 60%|██████ | 18009/30000 [11:31:18<8:53:58, 2.67s/it] 60%|██████ | 18010/30000 [11:31:20<7:24:43, 2.23s/it] 60%|██████ | 18010/30000 [11:31:20<7:24:43, 2.23s/it] 60%|██████ | 18011/30000 [11:31:21<6:23:22, 1.92s/it] 60%|██████ | 18012/30000 [11:31:22<5:38:36, 1.69s/it] 60%|██████ | 18013/30000 [11:31:23<5:07:35, 1.54s/it] 60%|██████ | 18014/30000 [11:31:24<4:45:46, 1.43s/it] 60%|██████ | 18015/30000 [11:31:25<4:29:49, 1.35s/it] 60%|██████ | 18016/30000 [11:31:33<10:36:39, 3.19s/it] 60%|██████ | 18017/30000 [11:31:35<9:05:14, 2.73s/it] 60%|██████ | 18018/30000 [11:31:36<7:31:12, 2.26s/it] 60%|██████ | 18019/30000 [11:31:37<6:27:24, 1.94s/it] 60%|██████ | 18020/30000 [11:31:38<5:42:05, 1.71s/it] 60%|██████ | 18020/30000 [11:31:38<5:42:05, 1.71s/it] 60%|██████ | 18021/30000 [11:31:39<5:10:55, 1.56s/it] 60%|██████ | 18022/30000 [11:31:41<4:48:42, 1.45s/it] 60%|██████ | 18023/30000 [11:31:42<4:33:46, 1.37s/it] 60%|██████ | 18024/30000 [11:31:48<9:22:06, 2.82s/it] 60%|██████ | 18025/30000 [11:31:51<9:22:23, 2.82s/it] 60%|██████ | 18026/30000 [11:31:52<7:43:53, 2.32s/it] 60%|██████ | 18027/30000 [11:31:53<6:36:12, 1.99s/it] 60%|██████ | 18028/30000 [11:31:54<5:47:24, 1.74s/it] 60%|██████ | 18029/30000 [11:31:55<5:14:36, 1.58s/it] 60%|██████ | 18030/30000 [11:31:57<4:50:48, 1.46s/it] 60%|██████ | 18030/30000 [11:31:57<4:50:48, 1.46s/it] 60%|██████ | 18031/30000 [11:31:58<4:33:43, 1.37s/it] 60%|██████ | 18032/30000 [11:32:03<8:26:35, 2.54s/it] 60%|██████ | 18033/30000 [11:32:06<9:05:32, 2.74s/it] 60%|██████ | 18034/30000 [11:32:07<7:32:38, 2.27s/it] 60%|██████ | 18035/30000 [11:32:09<6:27:25, 1.94s/it] 60%|██████ | 18036/30000 [11:32:10<5:40:28, 1.71s/it] 60%|██████ | 18037/30000 [11:32:11<5:12:21, 1.57s/it] 60%|██████ | 18038/30000 [11:32:12<4:49:49, 1.45s/it] 60%|██████ | 18039/30000 [11:32:13<4:33:13, 1.37s/it] 60%|██████ | 18040/30000 [11:32:18<8:10:26, 2.46s/it] 60%|██████ | 18040/30000 [11:32:18<8:10:26, 2.46s/it] 60%|██████ | 18041/30000 [11:32:21<8:18:57, 2.50s/it] 60%|██████ | 18042/30000 [11:32:22<6:59:37, 2.11s/it] 60%|██████ | 18043/30000 [11:32:23<6:04:24, 1.83s/it] 60%|██████ | 18044/30000 [11:32:25<5:25:03, 1.63s/it] 60%|██████ | 18045/30000 [11:32:26<4:57:21, 1.49s/it] 60%|██████ | 18046/30000 [11:32:27<4:39:14, 1.40s/it] 60%|██████ | 18047/30000 [11:32:28<4:26:49, 1.34s/it] 60%|██████ | 18048/30000 [11:32:34<8:43:13, 2.63s/it] 60%|██████ | 18049/30000 [11:32:36<8:36:23, 2.59s/it] 60%|██████ | 18050/30000 [11:32:37<7:11:18, 2.17s/it] 60%|██████ | 18050/30000 [11:32:37<7:11:18, 2.17s/it] 60%|██████ | 18051/30000 [11:32:39<6:12:48, 1.87s/it] 60%|██████ | 18052/30000 [11:32:40<5:30:00, 1.66s/it] 60%|██████ | 18053/30000 [11:32:41<5:00:16, 1.51s/it] 60%|██████ | 18054/30000 [11:32:42<4:41:39, 1.41s/it] 60%|██████ | 18055/30000 [11:32:43<4:28:25, 1.35s/it] 60%|██████ | 18056/30000 [11:32:49<9:12:30, 2.78s/it] 60%|██████ | 18057/30000 [11:32:51<8:08:49, 2.46s/it] 60%|██████ | 18058/30000 [11:32:52<6:51:49, 2.07s/it] 60%|██████ | 18059/30000 [11:32:53<5:59:51, 1.81s/it] 60%|██████ | 18060/30000 [11:32:55<5:21:39, 1.62s/it] 60%|██████ | 18060/30000 [11:32:55<5:21:39, 1.62s/it] 60%|██████ | 18061/30000 [11:32:56<4:55:18, 1.48s/it] 60%|██████ | 18062/30000 [11:32:57<4:38:33, 1.40s/it] 60%|██████ | 18063/30000 [11:32:58<4:26:15, 1.34s/it] 60%|██████ | 18064/30000 [11:33:05<9:23:08, 2.83s/it] 60%|██████ | 18065/30000 [11:33:07<8:38:44, 2.61s/it] 60%|██████ | 18066/30000 [11:33:08<7:13:05, 2.18s/it] 60%|██████ | 18067/30000 [11:33:09<6:14:54, 1.89s/it] 60%|██████ | 18068/30000 [11:33:10<5:33:50, 1.68s/it] 60%|██████ | 18069/30000 [11:33:11<5:03:22, 1.53s/it] 60%|██████ | 18070/30000 [11:33:13<4:43:51, 1.43s/it] 60%|██████ | 18070/30000 [11:33:13<4:43:51, 1.43s/it] 60%|██████ | 18071/30000 [11:33:14<4:31:45, 1.37s/it] 60%|██████ | 18072/30000 [11:33:21<9:59:56, 3.02s/it] 60%|██████ | 18073/30000 [11:33:22<8:25:50, 2.54s/it] 60%|██████ | 18074/30000 [11:33:23<7:03:47, 2.13s/it] 60%|██████ | 18075/30000 [11:33:24<6:08:03, 1.85s/it] 60%|██████ | 18076/30000 [11:33:26<5:28:01, 1.65s/it] 60%|██████ | 18077/30000 [11:33:27<4:59:43, 1.51s/it] 60%|██████ | 18078/30000 [11:33:28<4:40:30, 1.41s/it] 60%|██████ | 18079/30000 [11:33:29<4:27:33, 1.35s/it] 60%|██████ | 18080/30000 [11:33:36<10:06:22, 3.05s/it] 60%|██████ | 18080/30000 [11:33:36<10:06:22, 3.05s/it] 60%|██████ | 18081/30000 [11:33:38<8:25:01, 2.54s/it] 60%|██████ | 18082/30000 [11:33:39<7:02:42, 2.13s/it] 60%|██████ | 18083/30000 [11:33:40<6:06:53, 1.85s/it] 60%|██████ | 18084/30000 [11:33:41<5:25:58, 1.64s/it] 60%|██████ | 18085/30000 [11:33:42<4:57:35, 1.50s/it] 60%|██████ | 18086/30000 [11:33:43<4:38:22, 1.40s/it] 60%|██████ | 18087/30000 [11:33:45<4:25:49, 1.34s/it] 60%|██████ | 18088/30000 [11:33:52<10:14:44, 3.10s/it] 60%|██████ | 18089/30000 [11:33:54<9:05:33, 2.75s/it] 60%|██████ | 18090/30000 [11:33:55<7:32:09, 2.28s/it] 60%|██████ | 18090/30000 [11:33:55<7:32:09, 2.28s/it] 60%|██████ | 18091/30000 [11:33:56<6:27:45, 1.95s/it] 60%|██████ | 18092/30000 [11:33:57<5:40:00, 1.71s/it] 60%|██████ | 18093/30000 [11:33:58<5:07:19, 1.55s/it] 60%|██████ | 18094/30000 [11:34:00<4:45:55, 1.44s/it] 60%|██████ | 18095/30000 [11:34:01<4:31:29, 1.37s/it] 60%|██████ | 18096/30000 [11:34:07<9:30:50, 2.88s/it] 60%|██████ | 18097/30000 [11:34:10<9:11:32, 2.78s/it] 60%|██████ | 18098/30000 [11:34:11<7:35:22, 2.30s/it] 60%|██████ | 18099/30000 [11:34:12<6:28:56, 1.96s/it] 60%|██████ | 18100/30000 [11:34:13<5:41:46, 1.72s/it] 60%|██████ | 18100/30000 [11:34:13<5:41:46, 1.72s/it] 60%|██████ | 18101/30000 [11:34:15<5:09:37, 1.56s/it] 60%|██████ | 18102/30000 [11:34:16<4:48:05, 1.45s/it] 60%|██████ | 18103/30000 [11:34:17<4:32:57, 1.38s/it] 60%|██████ | 18104/30000 [11:34:22<8:26:51, 2.56s/it] 60%|██████ | 18105/30000 [11:34:25<8:11:27, 2.48s/it] 60%|██████ | 18106/30000 [11:34:26<6:53:36, 2.09s/it] 60%|██████ | 18107/30000 [11:34:27<6:00:12, 1.82s/it] 60%|██████ | 18108/30000 [11:34:28<5:20:37, 1.62s/it] 60%|██████ | 18109/30000 [11:34:29<4:54:03, 1.48s/it] 60%|██████ | 18110/30000 [11:34:30<4:37:08, 1.40s/it] 60%|██████ | 18110/30000 [11:34:30<4:37:08, 1.40s/it] 60%|██████ | 18111/30000 [11:34:32<4:25:47, 1.34s/it] 60%|██████ | 18112/30000 [11:34:38<9:12:58, 2.79s/it] 60%|██████ | 18113/30000 [11:34:40<8:59:18, 2.72s/it] 60%|██████ | 18114/30000 [11:34:42<7:26:48, 2.26s/it] 60%|██████ | 18115/30000 [11:34:43<6:23:25, 1.94s/it] 60%|██████ | 18116/30000 [11:34:44<5:37:41, 1.70s/it] 60%|██████ | 18117/30000 [11:34:45<5:04:57, 1.54s/it] 60%|██████ | 18118/30000 [11:34:46<4:43:41, 1.43s/it] 60%|██████ | 18119/30000 [11:34:47<4:28:19, 1.36s/it] 60%|██████ | 18120/30000 [11:34:54<9:43:09, 2.95s/it] 60%|██████ | 18120/30000 [11:34:54<9:43:09, 2.95s/it] 60%|██████ | 18121/30000 [11:34:57<9:15:52, 2.81s/it] 60%|██████ | 18122/30000 [11:34:58<7:39:18, 2.32s/it] 60%|██████ | 18123/30000 [11:34:59<6:32:41, 1.98s/it] 60%|██████ | 18124/30000 [11:35:00<5:44:23, 1.74s/it] 60%|██████ | 18125/30000 [11:35:01<5:11:23, 1.57s/it] 60%|██████ | 18126/30000 [11:35:02<4:47:36, 1.45s/it] 60%|██████ | 18127/30000 [11:35:04<4:32:06, 1.38s/it] 60%|██████ | 18128/30000 [11:35:10<9:00:41, 2.73s/it] 60%|██████ | 18129/30000 [11:35:13<9:27:47, 2.87s/it] 60%|██████ | 18130/30000 [11:35:14<7:46:51, 2.36s/it] 60%|██████ | 18130/30000 [11:35:14<7:46:51, 2.36s/it] 60%|██████ | 18131/30000 [11:35:15<6:37:02, 2.01s/it] 60%|██████ | 18132/30000 [11:35:16<5:47:33, 1.76s/it] 60%|██████ | 18133/30000 [11:35:17<5:11:51, 1.58s/it] 60%|██████ | 18134/30000 [11:35:19<4:49:06, 1.46s/it] 60%|██████ | 18135/30000 [11:35:20<4:32:40, 1.38s/it] 60%|██████ | 18136/30000 [11:35:26<8:50:59, 2.69s/it] 60%|██████ | 18137/30000 [11:35:29<9:11:59, 2.79s/it] 60%|██████ | 18138/30000 [11:35:30<7:36:10, 2.31s/it] 60%|██████ | 18139/30000 [11:35:31<6:29:09, 1.97s/it] 60%|██████ | 18140/30000 [11:35:32<5:41:25, 1.73s/it] 60%|██████ | 18140/30000 [11:35:32<5:41:25, 1.73s/it] 60%|██████ | 18141/30000 [11:35:33<5:08:59, 1.56s/it] 60%|██████ | 18142/30000 [11:35:34<4:47:06, 1.45s/it] 60%|██████ | 18143/30000 [11:35:36<4:31:59, 1.38s/it] 60%|██████ | 18144/30000 [11:35:41<8:04:56, 2.45s/it] 60%|██████ | 18145/30000 [11:35:43<8:24:44, 2.55s/it] 60%|██████ | 18146/30000 [11:35:45<7:02:22, 2.14s/it] 60%|██████ | 18147/30000 [11:35:46<6:06:17, 1.85s/it] 60%|██████ | 18148/30000 [11:35:47<5:25:30, 1.65s/it] 60%|██████ | 18149/30000 [11:35:48<4:57:04, 1.50s/it] 60%|██████ | 18150/30000 [11:35:49<4:38:21, 1.41s/it] 60%|██████ | 18150/30000 [11:35:49<4:38:21, 1.41s/it] 61%|██████ | 18151/30000 [11:35:50<4:26:03, 1.35s/it] 61%|██████ | 18152/30000 [11:35:56<8:26:47, 2.57s/it] 61%|██████ | 18153/30000 [11:35:59<9:07:38, 2.77s/it] 61%|██████ | 18154/30000 [11:36:00<7:32:19, 2.29s/it] 61%|██████ | 18155/30000 [11:36:02<6:27:48, 1.96s/it] 61%|██████ | 18156/30000 [11:36:03<5:41:18, 1.73s/it] 61%|██████ | 18157/30000 [11:36:04<5:07:57, 1.56s/it] 61%|██████ | 18158/30000 [11:36:05<4:45:55, 1.45s/it] 61%|██████ | 18159/30000 [11:36:06<4:29:58, 1.37s/it] 61%|██████ | 18160/30000 [11:36:12<8:21:28, 2.54s/it] 61%|██████ | 18160/30000 [11:36:12<8:21:28, 2.54s/it] 61%|██████ | 18161/30000 [11:36:15<9:38:46, 2.93s/it] 61%|██████ | 18162/30000 [11:36:17<7:54:21, 2.40s/it] 61%|██████ | 18163/30000 [11:36:18<6:42:22, 2.04s/it] 61%|██████ | 18164/30000 [11:36:19<5:50:17, 1.78s/it] 61%|██████ | 18165/30000 [11:36:20<5:14:18, 1.59s/it] 61%|██████ | 18166/30000 [11:36:21<4:50:34, 1.47s/it] 61%|██████ | 18167/30000 [11:36:22<4:34:46, 1.39s/it] 61%|██████ | 18168/30000 [11:36:27<7:20:57, 2.24s/it] 61%|██████ | 18169/30000 [11:36:30<8:27:57, 2.58s/it] 61%|██████ | 18170/30000 [11:36:31<7:03:58, 2.15s/it] 61%|██████ | 18170/30000 [11:36:31<7:03:58, 2.15s/it] 61%|██████ | 18171/30000 [11:36:32<6:06:56, 1.86s/it] 61%|██████ | 18172/30000 [11:36:34<5:26:29, 1.66s/it] 61%|██████ | 18173/30000 [11:36:35<4:58:30, 1.51s/it] 61%|██████ | 18174/30000 [11:36:36<4:38:42, 1.41s/it] 61%|██████ | 18175/30000 [11:36:37<4:26:22, 1.35s/it] 61%|██████ | 18176/30000 [11:36:42<7:37:36, 2.32s/it] 61%|██████ | 18177/30000 [11:36:45<8:57:10, 2.73s/it] 61%|██████ | 18178/30000 [11:36:47<7:25:18, 2.26s/it] 61%|██████ | 18179/30000 [11:36:48<6:21:25, 1.94s/it] 61%|██████ | 18180/30000 [11:36:49<5:36:19, 1.71s/it] 61%|██████ | 18180/30000 [11:36:49<5:36:19, 1.71s/it] 61%|██████ | 18181/30000 [11:36:50<5:04:32, 1.55s/it] 61%|██████ | 18182/30000 [11:36:51<4:43:28, 1.44s/it] 61%|██████ | 18183/30000 [11:36:52<4:29:51, 1.37s/it] 61%|██████ | 18184/30000 [11:36:57<7:23:44, 2.25s/it] 61%|██████ | 18185/30000 [11:37:01<9:30:30, 2.90s/it] 61%|██████ | 18186/30000 [11:37:02<7:48:39, 2.38s/it] 61%|██████ | 18187/30000 [11:37:04<6:39:46, 2.03s/it] 61%|██████ | 18188/30000 [11:37:05<5:48:57, 1.77s/it] 61%|██████ | 18189/30000 [11:37:06<5:14:03, 1.60s/it] 61%|██████ | 18190/30000 [11:37:07<4:49:58, 1.47s/it] +{'loss': 0.0152, 'grad_norm': 0.08447699248790741, 'learning_rate': 1.1656889592674208e-05, 'flow_matching_loss': 0.0152359863743186, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.18} +{'loss': 0.0156, 'grad_norm': 0.1200331598520279, 'learning_rate': 1.1640772821927533e-05, 'flow_matching_loss': 0.015629077143967152, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.18} +{'loss': 0.0138, 'grad_norm': 0.08382770419120789, 'learning_rate': 1.162466013295864e-05, 'flow_matching_loss': 0.013760239910334348, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.18} +{'loss': 0.0135, 'grad_norm': 0.08971860259771347, 'learning_rate': 1.160855154534595e-05, 'flow_matching_loss': 0.013536443933844566, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.18} +{'loss': 0.0135, 'grad_norm': 0.0659145936369896, 'learning_rate': 1.1592447078662931e-05, 'flow_matching_loss': 0.013499094918370248, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.19} +{'loss': 0.016, 'grad_norm': 0.07897434383630753, 'learning_rate': 1.1576346752478025e-05, 'flow_matching_loss': 0.015971735958009957, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.19} +{'loss': 0.014, 'grad_norm': 0.08359299600124359, 'learning_rate': 1.1560250586354643e-05, 'flow_matching_loss': 0.013998060021549463, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.19} +{'loss': 0.013, 'grad_norm': 0.0731058418750763, 'learning_rate': 1.1544158599851146e-05, 'flow_matching_loss': 0.01304289735853672, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.2} +{'loss': 0.014, 'grad_norm': 0.0802319198846817, 'learning_rate': 1.1528070812520809e-05, 'flow_matching_loss': 0.01395136322826147, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.2} +{'loss': 0.0131, 'grad_norm': 0.07824064046144485, 'learning_rate': 1.1511987243911821e-05, 'flow_matching_loss': 0.013076271954923868, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.2} +{'loss': 0.0135, 'grad_norm': 0.06238299980759621, 'learning_rate': 1.1495907913567228e-05, 'flow_matching_loss': 0.013538339547812939, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.21} +{'loss': 0.0125, 'grad_norm': 0.07444217056035995, 'learning_rate': 1.1479832841024936e-05, 'flow_matching_loss': 0.012480934709310531, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.21} +{'loss': 0.0147, 'grad_norm': 0.08569744974374771, 'learning_rate': 1.1463762045817675e-05, 'flow_matching_loss': 0.014742120262235403, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.21} +{'loss': 0.0134, 'grad_norm': 0.07056215405464172, 'learning_rate': 1.1447695547472964e-05, 'flow_matching_loss': 0.01336787510663271, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.21} +{'loss': 0.0141, 'grad_norm': 0.06577488034963608, 'learning_rate': 1.1431633365513132e-05, 'flow_matching_loss': 0.014095176011323929, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.22} +{'loss': 0.0139, 'grad_norm': 0.07080544531345367, 'learning_rate': 1.1415575519455235e-05, 'flow_matching_loss': 0.013916104938834906, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.22} +{'loss': 0.013, 'grad_norm': 0.076821468770504, 'learning_rate': 1.1399522028811069e-05, 'flow_matching_loss': 0.012974327336996794, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.22} +{'loss': 0.0139, 'grad_norm': 0.07235971093177795, 'learning_rate': 1.1383472913087143e-05, 'flow_matching_loss': 0.013941044453531504, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.23} +{'loss': 0.0126, 'grad_norm': 0.06894263625144958, 'learning_rate': 1.1367428191784643e-05, 'flow_matching_loss': 0.012575377989560366, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.23} +{'loss': 0.0141, 'grad_norm': 0.07341308891773224, 'learning_rate': 1.1351387884399413e-05, 'flow_matching_loss': 0.014110368583351374, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.23} +{'loss': 0.0147, 'grad_norm': 0.08293548971414566, 'learning_rate': 1.1335352010421953e-05, 'flow_matching_loss': 0.014685844257473945, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.23} +{'loss': 0.0152, 'grad_norm': 0.11656209826469421, 'learning_rate': 1.1319320589337352e-05, 'flow_matching_loss': 0.01524152709171176, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.24} +{'loss': 0.014, 'grad_norm': 0.07882246375083923, 'learning_rate': 1.1303293640625299e-05, 'flow_matching_loss': 0.014020074717700481, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.24} +{'loss': 0.0139, 'grad_norm': 0.07227837294340134, 'learning_rate': 1.1287271183760052e-05, 'flow_matching_loss': 0.013929210789501666, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.24} +{'loss': 0.0141, 'grad_norm': 0.07670918107032776, 'learning_rate': 1.1271253238210398e-05, 'flow_matching_loss': 0.014053185191005468, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.25} +{'loss': 0.0138, 'grad_norm': 0.09517525136470795, 'learning_rate': 1.1255239823439663e-05, 'flow_matching_loss': 0.01381723051890731, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.25} +{'loss': 0.0141, 'grad_norm': 0.0841747596859932, 'learning_rate': 1.1239230958905652e-05, 'flow_matching_loss': 0.01412128945812583, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.25} +{'loss': 0.0143, 'grad_norm': 0.07077088207006454, 'learning_rate': 1.1223226664060641e-05, 'flow_matching_loss': 0.014334895368665456, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.25} +{'loss': 0.012, 'grad_norm': 0.07647416740655899, 'learning_rate': 1.1207226958351362e-05, 'flow_matching_loss': 0.011991481855511665, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.26} +{'loss': 0.0128, 'grad_norm': 0.05713275074958801, 'learning_rate': 1.1191231861218955e-05, 'flow_matching_loss': 0.012789621390402317, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.26} +{'loss': 0.013, 'grad_norm': 0.08305735886096954, 'learning_rate': 1.1175241392098988e-05, 'flow_matching_loss': 0.01304421704262495, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.26} +{'loss': 0.0136, 'grad_norm': 0.0990513414144516, 'learning_rate': 1.1159255570421376e-05, 'flow_matching_loss': 0.013603393826633692, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.27} +{'loss': 0.0143, 'grad_norm': 0.0787111148238182, 'learning_rate': 1.1143274415610408e-05, 'flow_matching_loss': 0.01425888789817691, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.27} +{'loss': 0.0148, 'grad_norm': 0.08488097786903381, 'learning_rate': 1.1127297947084692e-05, 'flow_matching_loss': 0.014824128057807685, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.27} +{'loss': 0.0145, 'grad_norm': 0.08265654742717743, 'learning_rate': 1.1111326184257135e-05, 'flow_matching_loss': 0.014517893455922604, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.27} +{'loss': 0.0148, 'grad_norm': 0.08194331079721451, 'learning_rate': 1.109535914653495e-05, 'flow_matching_loss': 0.014811432175338268, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.28} +{'loss': 0.0141, 'grad_norm': 0.07347078621387482, 'learning_rate': 1.1079396853319582e-05, 'flow_matching_loss': 0.014137065783143044, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.28} +{'loss': 0.0143, 'grad_norm': 0.09449905902147293, 'learning_rate': 1.1063439324006724e-05, 'flow_matching_loss': 0.01434409050270915, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.28} +{'loss': 0.0142, 'grad_norm': 0.07819166034460068, 'learning_rate': 1.1047486577986285e-05, 'flow_matching_loss': 0.014212740398943425, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.29} +{'loss': 0.0129, 'grad_norm': 0.06693124771118164, 'learning_rate': 1.1031538634642353e-05, 'flow_matching_loss': 0.012921795994043351, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.29} +{'loss': 0.0141, 'grad_norm': 0.0788087248802185, 'learning_rate': 1.1015595513353172e-05, 'flow_matching_loss': 0.014065996371209621, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.29} 61%|██████ | 18190/30000 [11:37:07<4:49:58, 1.47s/it] 61%|██████ | 18191/30000 [11:37:08<4:33:37, 1.39s/it] 61%|██████ | 18192/30000 [11:37:12<6:25:12, 1.96s/it] 61%|██████ | 18193/30000 [11:37:16<8:22:32, 2.55s/it] 61%|██████ | 18194/30000 [11:37:17<7:01:02, 2.14s/it] 61%|██████ | 18195/30000 [11:37:18<6:05:22, 1.86s/it] 61%|██████ | 18196/30000 [11:37:19<5:24:34, 1.65s/it] 61%|██████ | 18197/30000 [11:37:20<4:56:08, 1.51s/it] 61%|██████ | 18198/30000 [11:37:21<4:37:10, 1.41s/it] 61%|██████ | 18199/30000 [11:37:23<4:24:28, 1.34s/it] 61%|██████ | 18200/30000 [11:37:26<6:39:31, 2.03s/it] 61%|██████ | 18200/30000 [11:37:26<6:39:31, 2.03s/it] 61%|██████ | 18201/30000 [11:37:31<8:52:53, 2.71s/it] 61%|██████ | 18202/30000 [11:37:32<7:21:53, 2.25s/it] 61%|██████ | 18203/30000 [11:37:33<6:19:04, 1.93s/it] 61%|██████ | 18204/30000 [11:37:34<5:34:13, 1.70s/it] 61%|██████ | 18205/30000 [11:37:35<5:03:22, 1.54s/it] 61%|██████ | 18206/30000 [11:37:36<4:41:55, 1.43s/it] 61%|██████ | 18207/30000 [11:37:38<4:27:54, 1.36s/it] 61%|██████ | 18208/30000 [11:37:41<6:14:25, 1.91s/it] 61%|██████ | 18209/30000 [11:37:46<9:19:19, 2.85s/it] 61%|██████ | 18210/30000 [11:37:47<7:40:08, 2.34s/it] 61%|██████ | 18210/30000 [11:37:47<7:40:08, 2.34s/it] 61%|██████ | 18211/30000 [11:37:48<6:32:25, 2.00s/it] 61%|██████ | 18212/30000 [11:37:49<5:43:43, 1.75s/it] 61%|██████ | 18213/30000 [11:37:51<5:09:32, 1.58s/it] 61%|██████ | 18214/30000 [11:37:52<4:47:42, 1.46s/it] 61%|██████ | 18215/30000 [11:37:53<4:31:57, 1.38s/it] 61%|██████ | 18216/30000 [11:37:55<5:31:45, 1.69s/it] 61%|██████ | 18217/30000 [11:38:01<9:26:52, 2.89s/it] 61%|██████ | 18218/30000 [11:38:02<7:44:58, 2.37s/it] 61%|██████ | 18219/30000 [11:38:03<6:36:53, 2.02s/it] 61%|██████ | 18220/30000 [11:38:05<5:47:39, 1.77s/it] 61%|██████ | 18220/30000 [11:38:05<5:47:39, 1.77s/it] 61%|██████ | 18221/30000 [11:38:06<5:14:05, 1.60s/it] 61%|██████ | 18222/30000 [11:38:07<4:49:55, 1.48s/it] 61%|██████ | 18223/30000 [11:38:08<4:32:42, 1.39s/it] 61%|██████ | 18224/30000 [11:38:10<5:06:38, 1.56s/it] 61%|██████ | 18225/30000 [11:38:16<8:58:08, 2.74s/it] 61%|██████ | 18226/30000 [11:38:17<7:25:05, 2.27s/it] 61%|██████ | 18227/30000 [11:38:18<6:21:28, 1.94s/it] 61%|██████ | 18228/30000 [11:38:19<5:35:27, 1.71s/it] 61%|██████ | 18229/30000 [11:38:20<5:03:03, 1.54s/it] 61%|██████ | 18230/30000 [11:38:21<4:42:18, 1.44s/it] 61%|██████ | 18230/30000 [11:38:21<4:42:18, 1.44s/it] 61%|██████ | 18231/30000 [11:38:23<4:28:08, 1.37s/it] 61%|██████ | 18232/30000 [11:38:25<5:31:04, 1.69s/it] 61%|██████ | 18233/30000 [11:38:31<9:30:29, 2.91s/it] 61%|██████ | 18234/30000 [11:38:32<7:47:38, 2.38s/it] 61%|██████ | 18235/30000 [11:38:33<6:35:51, 2.02s/it] 61%|██████ | 18236/30000 [11:38:34<5:45:19, 1.76s/it] 61%|██████ | 18237/30000 [11:38:36<5:09:51, 1.58s/it] 61%|██████ | 18238/30000 [11:38:37<4:47:03, 1.46s/it] 61%|██████ | 18239/30000 [11:38:38<4:31:25, 1.38s/it] 61%|██████ | 18240/30000 [11:38:41<5:43:25, 1.75s/it] 61%|██████ | 18240/30000 [11:38:41<5:43:25, 1.75s/it] 61%|██████ | 18241/30000 [11:38:46<9:40:51, 2.96s/it] 61%|██████ | 18242/30000 [11:38:47<7:54:19, 2.42s/it] 61%|██████ | 18243/30000 [11:38:49<6:40:53, 2.05s/it] 61%|██████ | 18244/30000 [11:38:50<5:49:09, 1.78s/it] 61%|██████ | 18245/30000 [11:38:51<5:12:32, 1.60s/it] 61%|██████ | 18246/30000 [11:38:52<4:48:44, 1.47s/it] 61%|██████ | 18247/30000 [11:38:53<4:32:14, 1.39s/it] 61%|██████ | 18248/30000 [11:38:56<5:52:04, 1.80s/it] 61%|██████ | 18249/30000 [11:39:02<9:47:22, 3.00s/it] 61%|██████ | 18250/30000 [11:39:03<8:00:23, 2.45s/it] 61%|██████ | 18250/30000 [11:39:03<8:00:23, 2.45s/it] 61%|██████ | 18251/30000 [11:39:04<6:46:36, 2.08s/it] 61%|██████ | 18252/30000 [11:39:05<5:53:11, 1.80s/it] 61%|██████ | 18253/30000 [11:39:07<5:16:05, 1.61s/it] 61%|██████ | 18254/30000 [11:39:08<4:51:02, 1.49s/it] 61%|██████ | 18255/30000 [11:39:09<4:33:54, 1.40s/it] 61%|██████ | 18256/30000 [11:39:12<5:55:55, 1.82s/it] 61%|██████ | 18257/30000 [11:39:17<9:01:36, 2.77s/it] 61%|██████ | 18258/30000 [11:39:18<7:26:51, 2.28s/it] 61%|██████ | 18259/30000 [11:39:19<6:22:49, 1.96s/it] 61%|██████ | 18260/30000 [11:39:20<5:35:50, 1.72s/it] 61%|██████ | 18260/30000 [11:39:20<5:35:50, 1.72s/it] 61%|██████ | 18261/30000 [11:39:21<5:04:10, 1.55s/it] 61%|██████ | 18262/30000 [11:39:23<4:43:15, 1.45s/it] 61%|██████ | 18263/30000 [11:39:24<4:28:01, 1.37s/it] 61%|██████ | 18264/30000 [11:39:27<6:01:52, 1.85s/it] 61%|██████ | 18265/30000 [11:39:32<9:09:15, 2.81s/it] 61%|██████ | 18266/30000 [11:39:33<7:32:54, 2.32s/it] 61%|██████ | 18267/30000 [11:39:34<6:27:21, 1.98s/it] 61%|██████ | 18268/30000 [11:39:35<5:39:32, 1.74s/it] 61%|██████ | 18269/30000 [11:39:37<5:06:06, 1.57s/it] 61%|██████ | 18270/30000 [11:39:38<4:44:10, 1.45s/it] 61%|██████ | 18270/30000 [11:39:38<4:44:10, 1.45s/it] 61%|██████ | 18271/30000 [11:39:39<4:29:14, 1.38s/it] 61%|██████ | 18272/30000 [11:39:42<6:15:46, 1.92s/it] 61%|██████ | 18273/30000 [11:39:47<9:19:20, 2.86s/it] 61%|██████ | 18274/30000 [11:39:48<7:39:46, 2.35s/it] 61%|██████ | 18275/30000 [11:39:50<6:30:31, 2.00s/it] 61%|██████ | 18276/30000 [11:39:51<5:41:39, 1.75s/it] 61%|██████ | 18277/30000 [11:39:52<5:08:24, 1.58s/it] 61%|██████ | 18278/30000 [11:39:53<4:47:53, 1.47s/it] 61%|██████ | 18279/30000 [11:39:54<4:31:56, 1.39s/it] 61%|██████ | 18280/30000 [11:39:57<5:38:45, 1.73s/it] 61%|██████ | 18280/30000 [11:39:57<5:38:45, 1.73s/it] 61%|██████ | 18281/30000 [11:40:02<8:35:47, 2.64s/it] 61%|██████ | 18282/30000 [11:40:03<7:09:37, 2.20s/it] 61%|██████ | 18283/30000 [11:40:04<6:12:01, 1.91s/it] 61%|██████ | 18284/30000 [11:40:05<5:29:08, 1.69s/it] 61%|██████ | 18285/30000 [11:40:06<4:59:01, 1.53s/it] 61%|██████ | 18286/30000 [11:40:08<4:38:34, 1.43s/it] 61%|██████ | 18287/30000 [11:40:09<4:24:17, 1.35s/it] 61%|██████ | 18288/30000 [11:40:12<6:14:22, 1.92s/it] 61%|██████ | 18289/30000 [11:40:17<9:02:44, 2.78s/it] 61%|██████ | 18290/30000 [11:40:18<7:28:04, 2.30s/it] 61%|██████ | 18290/30000 [11:40:18<7:28:04, 2.30s/it] 61%|██████ | 18291/30000 [11:40:19<6:22:53, 1.96s/it] 61%|██████ | 18292/30000 [11:40:20<5:36:11, 1.72s/it] 61%|██████ | 18293/30000 [11:40:21<5:03:32, 1.56s/it] 61%|██████ | 18294/30000 [11:40:23<4:42:21, 1.45s/it] 61%|██████ | 18295/30000 [11:40:24<4:27:32, 1.37s/it] 61%|██████ | 18296/30000 [11:40:27<6:03:10, 1.86s/it] 61%|██████ | 18297/30000 [11:40:31<8:15:48, 2.54s/it] 61%|██████ | 18298/30000 [11:40:32<6:54:49, 2.13s/it] 61%|██████ | 18299/30000 [11:40:33<5:59:32, 1.84s/it] 61%|██████ | 18300/30000 [11:40:34<5:20:11, 1.64s/it] 61%|██████ | 18300/30000 [11:40:34<5:20:11, 1.64s/it] 61%|██████ | 18301/30000 [11:40:36<4:53:07, 1.50s/it] 61%|██████ | 18302/30000 [11:40:37<4:34:13, 1.41s/it] 61%|██████ | 18303/30000 [11:40:38<4:21:23, 1.34s/it] 61%|██████ | 18304/30000 [11:40:42<6:31:31, 2.01s/it] 61%|██████ | 18305/30000 [11:40:46<8:47:04, 2.70s/it] 61%|██████ | 18306/30000 [11:40:47<7:18:24, 2.25s/it] 61%|██████ | 18307/30000 [11:40:48<6:15:40, 1.93s/it] 61%|██████ | 18308/30000 [11:40:49<5:31:06, 1.70s/it] 61%|██████ | 18309/30000 [11:40:51<4:59:42, 1.54s/it] 61%|██████ | 18310/30000 [11:40:52<4:39:09, 1.43s/it] 61%|██████ | 18310/30000 [11:40:52<4:39:09, 1.43s/it] 61%|██████ | 18311/30000 [11:40:53<4:25:49, 1.36s/it] 61%|██████ | 18312/30000 [11:40:56<6:27:04, 1.99s/it] 61%|██████ | 18313/30000 [11:41:02<9:57:27, 3.07s/it] 61%|██████ | 18314/30000 [11:41:03<8:05:45, 2.49s/it] 61%|██████ | 18315/30000 [11:41:04<6:49:11, 2.10s/it] 61%|██████ | 18316/30000 [11:41:06<5:55:14, 1.82s/it] 61%|██████ | 18317/30000 [11:41:07<5:17:07, 1.63s/it] 61%|██████ | 18318/30000 [11:41:08<4:52:05, 1.50s/it] 61%|██████ | 18319/30000 [11:41:09<4:34:23, 1.41s/it] 61%|██████ | 18320/30000 [11:41:12<6:03:19, 1.87s/it] 61%|██████ | 18320/30000 [11:41:12<6:03:19, 1.87s/it] 61%|██████ | 18321/30000 [11:41:17<8:48:30, 2.72s/it] 61%|██████ | 18322/30000 [11:41:18<7:18:45, 2.25s/it] 61%|██████ | 18323/30000 [11:41:19<6:16:49, 1.94s/it] 61%|██████ | 18324/30000 [11:41:20<5:33:09, 1.71s/it] 61%|██████ | 18325/30000 [11:41:21<5:01:46, 1.55s/it] 61%|██████ | 18326/30000 [11:41:23<4:40:30, 1.44s/it] 61%|██████ | 18327/30000 [11:41:24<4:24:47, 1.36s/it] 61%|██████ | 18328/30000 [11:41:27<6:03:39, 1.87s/it] 61%|██████ | 18329/30000 [11:41:32<8:49:35, 2.72s/it] 61%|██████ | 18330/30000 [11:41:33<7:19:12, 2.26s/it] 61%|██████ | 18330/30000 [11:41:33<7:19:12, 2.26s/it] 61%|██████ | 18331/30000 [11:41:34<6:17:04, 1.94s/it] 61%|██████ | 18332/30000 [11:41:35<5:31:48, 1.71s/it] 61%|██████ | 18333/30000 [11:41:36<4:59:55, 1.54s/it] 61%|██████ | 18334/30000 [11:41:37<4:38:11, 1.43s/it] 61%|██████ | 18335/30000 [11:41:39<4:24:05, 1.36s/it] 61%|██████ | 18336/30000 [11:41:42<6:34:28, 2.03s/it] 61%|██████ | 18337/30000 [11:41:47<8:50:48, 2.73s/it] 61%|██████ | 18338/30000 [11:41:48<7:19:17, 2.26s/it] 61%|██████ | 18339/30000 [11:41:49<6:16:47, 1.94s/it] 61%|██████ | 18340/30000 [11:41:50<5:31:17, 1.70s/it] 61%|██████ | 18340/30000 [11:41:50<5:31:17, 1.70s/it] 61%|██████ | 18341/30000 [11:41:51<5:00:19, 1.55s/it] 61%|██████ | 18342/30000 [11:41:52<4:39:15, 1.44s/it] 61%|██████ | 18343/30000 [11:41:54<4:24:43, 1.36s/it] 61%|██████ | 18344/30000 [11:41:58<7:09:02, 2.21s/it] 61%|██████ | 18345/30000 [11:42:02<9:10:42, 2.84s/it] 61%|██████ | 18346/30000 [11:42:03<7:33:16, 2.33s/it] 61%|██████ | 18347/30000 [11:42:04<6:25:28, 1.98s/it] 61%|██████ | 18348/30000 [11:42:06<5:38:03, 1.74s/it] 61%|██████ | 18349/30000 [11:42:07<5:05:03, 1.57s/it] 61%|██████ | 18350/30000 [11:42:08<4:42:53, 1.46s/it] 61%|██████ | 18350/30000 [11:42:08<4:42:53, 1.46s/it] 61%|██████ | 18351/30000 [11:42:09<4:27:14, 1.38s/it] 61%|██████ | 18352/30000 [11:42:12<5:59:58, 1.85s/it] 61%|██████ | 18353/30000 [11:42:16<8:24:09, 2.60s/it] 61%|██████ | 18354/30000 [11:42:18<7:01:54, 2.17s/it] 61%|██████ | 18355/30000 [11:42:19<6:04:49, 1.88s/it] 61%|██████ | 18356/30000 [11:42:20<5:23:11, 1.67s/it] 61%|██████ | 18357/30000 [11:42:21<4:54:00, 1.52s/it] 61%|██████ | 18358/30000 [11:42:22<4:34:22, 1.41s/it] 61%|██████ | 18359/30000 [11:42:24<4:20:56, 1.34s/it] 61%|██████ | 18360/30000 [11:42:27<6:38:53, 2.06s/it] 61%|██████ | 18360/30000 [11:42:27<6:38:53, 2.06s/it] 61%|██████ | 18361/30000 [11:42:32<9:20:50, 2.89s/it] 61%|██████ | 18362/30000 [11:42:33<7:41:21, 2.38s/it] 61%|██████ | 18363/30000 [11:42:34<6:30:14, 2.01s/it] 61%|██████ | 18364/30000 [11:42:36<5:41:50, 1.76s/it] 61%|██████ | 18365/30000 [11:42:37<5:07:41, 1.59s/it] 61%|██████ | 18366/30000 [11:42:38<4:43:25, 1.46s/it] 61%|██████ | 18367/30000 [11:42:39<4:27:29, 1.38s/it] 61%|██████ | 18368/30000 [11:42:42<5:49:06, 1.80s/it] 61%|██████ | 18369/30000 [11:42:48<9:43:09, 3.01s/it] 61%|██████ | 18370/30000 [11:42:49<7:55:42, 2.45s/it] 61%|██████ | 18370/30000 [11:42:49<7:55:42, 2.45s/it] 61%|██████ | 18371/30000 [11:42:50<6:40:45, 2.07s/it] 61%|██████ | 18372/30000 [11:42:51<5:47:14, 1.79s/it] 61%|██████ | 18373/30000 [11:42:52<5:10:08, 1.60s/it] 61%|██████ | 18374/30000 [11:42:54<4:46:29, 1.48s/it] 61%|██████▏ | 18375/30000 [11:42:55<4:29:10, 1.39s/it] 61%|██████▏ | 18376/30000 [11:42:56<4:32:08, 1.40s/it] 61%|██████▏ | 18377/30000 [11:43:03<9:41:45, 3.00s/it] 61%|██████▏ | 18378/30000 [11:43:04<7:54:25, 2.45s/it] 61%|██████▏ | 18379/30000 [11:43:05<6:40:20, 2.07s/it] 61%|██████▏ | 18380/30000 [11:43:06<5:47:21, 1.79s/it] 61%|██████▏ | 18380/30000 [11:43:06<5:47:21, 1.79s/it] 61%|██████▏ | 18381/30000 [11:43:08<5:11:00, 1.61s/it] 61%|██████▏ | 18382/30000 [11:43:09<4:46:30, 1.48s/it] 61%|██████▏ | 18383/30000 [11:43:10<4:31:11, 1.40s/it] 61%|██████▏ | 18384/30000 [11:43:11<4:20:07, 1.34s/it] 61%|██████▏ | 18385/30000 [11:43:18<9:19:06, 2.89s/it] 61%|██████▏ | 18386/30000 [11:43:19<7:38:32, 2.37s/it] 61%|██████▏ | 18387/30000 [11:43:20<6:29:11, 2.01s/it] 61%|██████▏ | 18388/30000 [11:43:21<5:40:03, 1.76s/it] 61%|██████▏ | 18389/30000 [11:43:22<5:06:13, 1.58s/it] 61%|██████▏ | 18390/30000 [11:43:24<4:43:09, 1.46s/it] 61%|██████▏ | 18390/30000 [11:43:24<4:43:09, 1.46s/it] 61%|██████▏ | 18391/30000 [11:43:25<4:26:45, 1.38s/it] 61%|██████▏ | 18392/30000 [11:43:26<4:15:26, 1.32s/it] 61%|██████▏ | 18393/30000 [11:43:32<9:13:19, 2.86s/it] 61%|██████▏ | 18394/30000 [11:43:34<7:35:17, 2.35s/it] 61%|██████▏ | 18395/30000 [11:43:35<6:27:37, 2.00s/it] 61%|██████▏ | 18396/30000 [11:43:36<5:39:00, 1.75s/it] 61%|██████▏ | 18397/30000 [11:43:37<5:03:52, 1.57s/it] 61%|██████▏ | 18398/30000 [11:43:38<4:42:00, 1.46s/it] 61%|██████▏ | 18399/30000 [11:43:39<4:27:58, 1.39s/it] 61%|██████▏ | 18400/30000 [11:43:41<4:16:27, 1.33s/it] 61%|██████▏ | 18400/30000 [11:43:41<4:16:27, 1.33s/it] 61%|██████▏ | 18401/30000 [11:43:46<8:36:35, 2.67s/it] 61%|██████▏ | 18402/30000 [11:43:48<7:09:27, 2.22s/it] 61%|██████▏ | 18403/30000 [11:43:49<6:09:16, 1.91s/it] 61%|██████▏ | 18404/30000 [11:43:50<5:27:22, 1.69s/it] 61%|██████▏ | 18405/30000 [11:43:51<4:57:01, 1.54s/it] 61%|██████▏ | 18406/30000 [11:43:52<4:36:39, 1.43s/it] 61%|██████▏ | 18407/30000 [11:43:54<4:22:25, 1.36s/it] 61%|██████▏ | 18408/30000 [11:43:55<4:46:15, 1.48s/it] 61%|██████▏ | 18409/30000 [11:44:01<8:53:36, 2.76s/it] 61%|██████▏ | 18410/30000 [11:44:02<7:21:39, 2.29s/it] 61%|██████▏ | 18410/30000 [11:44:02<7:21:39, 2.29s/it] 61%|██████▏ | 18411/30000 [11:44:03<6:17:30, 1.95s/it] 61%|██████▏ | 18412/30000 [11:44:05<5:32:15, 1.72s/it] 61%|██████▏ | 18413/30000 [11:44:06<5:00:48, 1.56s/it] 61%|██████▏ | 18414/30000 [11:44:07<4:40:03, 1.45s/it] 61%|██████▏ | 18415/30000 [11:44:08<4:25:34, 1.38s/it] 61%|██████▏ | 18416/30000 [11:44:11<5:28:41, 1.70s/it] 61%|██████▏ | 18417/30000 [11:44:16<8:47:25, 2.73s/it] 61%|██████▏ | 18418/30000 [11:44:17<7:16:51, 2.26s/it] 61%|██████▏ | 18419/30000 [11:44:18<6:14:42, 1.94s/it] 61%|██████▏ | 18420/30000 [11:44:19<5:30:28, 1.71s/it] 61%|██████▏ | 18420/30000 [11:44:19<5:30:28, 1.71s/it] 61%|██████▏ | 18421/30000 [11:44:21<4:59:33, 1.55s/it] 61%|██████▏ | 18422/30000 [11:44:22<4:37:17, 1.44s/it] 61%|██████▏ | 18423/30000 [11:44:23<4:22:08, 1.36s/it] 61%|██████▏ | 18424/30000 [11:44:25<5:35:18, 1.74s/it] 61%|██████▏ | 18425/30000 [11:44:31<9:12:54, 2.87s/it] 61%|██████▏ | 18426/30000 [11:44:32<7:34:35, 2.36s/it] 61%|██████▏ | 18427/30000 [11:44:33<6:27:03, 2.01s/it] 61%|██████▏ | 18428/30000 [11:44:35<5:38:43, 1.76s/it] 61%|██████▏ | 18429/30000 [11:44:36<5:04:26, 1.58s/it] 61%|██████▏ | 18430/30000 [11:44:37<4:41:17, 1.46s/it] 61%|██████▏ | 18430/30000 [11:44:37<4:41:17, 1.46s/it] 61%|██████▏ | 18431/30000 [11:44:38<4:25:19, 1.38s/it] 61%|██████▏ | 18432/30000 [11:44:40<5:22:49, 1.67s/it] 61%|██████▏ | 18433/30000 [11:44:47<9:44:59, 3.03s/it] 61%|██████▏ | 18434/30000 [11:44:48<7:56:33, 2.47s/it] 61%|██████▏ | 18435/30000 [11:44:49<6:42:22, 2.09s/it] 61%|██████▏ | 18436/30000 [11:44:50<5:50:05, 1.82s/it] 61%|██████▏ | 18437/30000 [11:44:51<5:12:44, 1.62s/it] 61%|██████▏ | 18438/30000 [11:44:52<4:47:22, 1.49s/it] 61%|██████▏ | 18439/30000 [11:44:54<4:32:06, 1.41s/it] 61%|██████▏ | 18440/30000 [11:44:56<5:18:01, 1.65s/it] 61%|██████▏ | 18440/30000 [11:44:56<5:18:01, 1.65s/it] 61%|██████▏ | 18441/30000 [11:45:02<9:07:29, 2.84s/it] 61%|██████▏ | 18442/30000 [11:45:03<7:30:20, 2.34s/it] 61%|██████▏ | 18443/30000 [11:45:04<6:24:03, 1.99s/it] 61%|██████▏ | 18444/30000 [11:45:05<5:35:57, 1.74s/it] 61%|██████▏ | 18445/30000 [11:45:06<5:03:15, 1.57s/it] 61%|██████▏ | 18446/30000 [11:45:07<4:41:42, 1.46s/it] 61%|██████▏ | 18447/30000 [11:45:09<4:26:18, 1.38s/it] 61%|██████▏ | 18448/30000 [11:45:11<5:33:00, 1.73s/it] 61%|██████▏ | 18449/30000 [11:45:17<9:06:11, 2.84s/it] 62%|██████▏ | 18450/30000 [11:45:18<7:29:39, 2.34s/it] 62%|██████▏ | 18450/30000 [11:45:18<7:29:39, 2.34s/it] 62%|██████▏ | 18451/30000 [11:45:19<6:23:49, 1.99s/it] 62%|██████▏ | 18452/30000 [11:45:20<5:36:32, 1.75s/it] 62%|██████▏ | 18453/30000 [11:45:21<5:03:10, 1.58s/it] 62%|██████▏ | 18454/30000 [11:45:22<4:40:00, 1.46s/it] 62%|██████▏ | 18455/30000 [11:45:24<4:25:15, 1.38s/it] 62%|██████▏ | 18456/30000 [11:45:27<6:00:41, 1.87s/it] 62%|██████▏ | 18457/30000 [11:45:32<9:29:21, 2.96s/it] 62%|██████▏ | 18458/30000 [11:45:33<7:46:22, 2.42s/it] 62%|██████▏ | 18459/30000 [11:45:35<6:35:19, 2.06s/it] 62%|██████▏ | 18460/30000 [11:45:36<5:44:14, 1.79s/it] 62%|██████▏ | 18460/30000 [11:45:36<5:44:14, 1.79s/it] 62%|██████▏ | 18461/30000 [11:45:37<5:07:34, 1.60s/it] 62%|██████▏ | 18462/30000 [11:45:38<4:43:52, 1.48s/it] 62%|██████▏ | 18463/30000 [11:45:39<4:26:14, 1.38s/it] 62%|██████▏ | 18464/30000 [11:45:42<6:02:56, 1.89s/it] 62%|██████▏ | 18465/30000 [11:45:48<9:16:41, 2.90s/it] 62%|██████▏ | 18466/30000 [11:45:49<7:39:52, 2.39s/it] 62%|██████▏ | 18467/30000 [11:45:50<6:30:55, 2.03s/it] 62%|██████▏ | 18468/30000 [11:45:51<5:41:41, 1.78s/it] 62%|██████▏ | 18469/30000 [11:45:52<5:06:47, 1.60s/it] 62%|██████▏ | 18470/30000 [11:45:54<4:42:01, 1.47s/it] 62%|██████▏ | 18470/30000 [11:45:54<4:42:01, 1.47s/it] 62%|██████▏ | 18471/30000 [11:45:55<4:24:58, 1.38s/it] 62%|██████▏ | 18472/30000 [11:45:58<5:51:11, 1.83s/it] 62%|██████▏ | 18473/30000 [11:46:03<9:21:11, 2.92s/it] 62%|██████▏ | 18474/30000 [11:46:04<7:39:50, 2.39s/it] 62%|██████▏ | 18475/30000 [11:46:05<6:30:24, 2.03s/it] 62%|██████▏ | 18476/30000 [11:46:07<5:41:25, 1.78s/it] 62%|██████▏ | 18477/30000 [11:46:08<5:06:06, 1.59s/it] 62%|██████▏ | 18478/30000 [11:46:09<4:41:43, 1.47s/it] 62%|██████▏ | 18479/30000 [11:46:10<4:25:55, 1.38s/it] 62%|██████▏ | 18480/30000 [11:46:13<5:32:34, 1.73s/it] 62%|██████▏ | 18480/30000 [11:46:13<5:32:34, 1.73s/it] 62%|██████▏ | 18481/30000 [11:46:19<9:52:34, 3.09s/it] 62%|██████▏ | 18482/30000 [11:46:20<8:02:05, 2.51s/it] 62%|██████▏ | 18483/30000 [11:46:21<6:45:53, 2.11s/it] 62%|██████▏ | 18484/30000 [11:46:22<5:50:47, 1.83s/it] 62%|██████▏ | 18485/30000 [11:46:24<5:13:31, 1.63s/it] 62%|██████▏ | 18486/30000 [11:46:25<4:47:55, 1.50s/it] 62%|██████▏ | 18487/30000 [11:46:26<4:31:38, 1.42s/it] 62%|██████▏ | 18488/30000 [11:46:28<5:09:56, 1.62s/it] 62%|██████▏ | 18489/30000 [11:46:34<8:57:52, 2.80s/it] 62%|██████▏ | 18490/30000 [11:46:35<7:24:01, 2.31s/it] 62%|██████▏ | 18490/30000 [11:46:35<7:24:01, 2.31s/it] 62%|██████▏ | 18491/30000 [11:46:36<6:19:41, 1.98s/it] 62%|██████▏ | 18492/30000 [11:46:37<5:32:46, 1.73s/it] 62%|██████▏ | 18493/30000 [11:46:38<5:00:30, 1.57s/it] 62%|██████▏ | 18494/30000 [11:46:40<4:40:04, 1.46s/it] 62%|██████▏ | 18495/30000 [11:46:41<4:24:19, 1.38s/it] 62%|██████▏ | 18496/30000 [11:46:44<5:48:28, 1.82s/it] 62%|██████▏ | 18497/30000 [11:46:49<8:57:09, 2.80s/it] 62%|██████▏ | 18498/30000 [11:46:50<7:22:07, 2.31s/it] 62%|██████▏ | 18499/30000 [11:46:51<6:18:09, 1.97s/it] 62%|██████▏ | 18500/30000 [11:46:52<5:31:57, 1.73s/it] 62%|██████▏ | 18500/30000 [11:46:52<5:31:57, 1.73s/it] 62%|██████▏ | 18501/30000 [11:46:53<5:00:07, 1.57s/it] 62%|██████▏ | 18502/30000 [11:46:55<4:37:14, 1.45s/it] 62%|██████▏ | 18503/30000 [11:46:56<4:20:34, 1.36s/it] 62%|██████▏ | 18504/30000 [11:46:59<5:45:44, 1.80s/it] 62%|██████▏ | 18505/30000 [11:47:04<9:21:25, 2.93s/it] 62%|██████▏ | 18506/30000 [11:47:05<7:39:20, 2.40s/it] 62%|██████▏ | 18507/30000 [11:47:06<6:29:50, 2.04s/it] 62%|██████▏ | 18508/30000 [11:47:08<5:39:31, 1.77s/it] 62%|██████▏ | 18509/30000 [11:47:09<5:05:59, 1.60s/it] 62%|██████▏ | 18510/30000 [11:47:10<4:43:03, 1.48s/it] 62%|██████▏ | 18510/30000 [11:47:10<4:43:03, 1.48s/it] 62%|██████▏ | 18511/30000 [11:47:11<4:26:47, 1.39s/it] 62%|██████▏ | 18512/30000 [11:47:14<5:51:14, 1.83s/it] 62%|██████▏ | 18513/30000 [11:47:20<9:49:22, 3.08s/it] 62%|██████▏ | 18514/30000 [11:47:21<7:59:15, 2.50s/it] 62%|██████▏ | 18515/30000 [11:47:22<6:43:29, 2.11s/it] 62%|██████▏ | 18516/30000 [11:47:24<5:50:04, 1.83s/it] 62%|██████▏ | 18517/30000 [11:47:25<5:13:20, 1.64s/it] 62%|██████▏ | 18518/30000 [11:47:26<4:48:20, 1.51s/it] 62%|██████▏ | 18519/30000 [11:47:27<4:29:30, 1.41s/it] 62%|██████▏ | 18520/30000 [11:47:30<6:07:53, 1.92s/it] 62%|██████▏ | 18520/30000 [11:47:30<6:07:53, 1.92s/it] 62%|██████▏ | 18521/30000 [11:47:35<9:15:21, 2.90s/it] 62%|██████▏ | 18522/30000 [11:47:37<7:36:09, 2.38s/it] 62%|██████▏ | 18523/30000 [11:47:38<6:27:00, 2.02s/it] 62%|██████▏ | 18524/30000 [11:47:39<5:38:03, 1.77s/it] 62%|██████▏ | 18525/30000 [11:47:40<5:03:25, 1.59s/it] 62%|██████▏ | 18526/30000 [11:47:41<4:40:16, 1.47s/it] 62%|██████▏ | 18527/30000 [11:47:43<4:24:04, 1.38s/it] 62%|██████▏ | 18528/30000 [11:47:47<7:08:45, 2.24s/it] 62%|██████▏ | 18529/30000 [11:47:51<9:14:35, 2.90s/it] 62%|██████▏ | 18530/30000 [11:47:52<7:35:27, 2.38s/it] 62%|██████▏ | 18530/30000 [11:47:52<7:35:27, 2.38s/it] 62%|██████▏ | 18531/30000 [11:47:54<6:28:51, 2.03s/it] 62%|██████▏ | 18532/30000 [11:47:55<5:39:13, 1.77s/it] 62%|██████▏ | 18533/30000 [11:47:56<5:04:03, 1.59s/it] 62%|██████▏ | 18534/30000 [11:47:57<4:39:37, 1.46s/it] 62%|██████▏ | 18535/30000 [11:47:58<4:24:11, 1.38s/it] 62%|██████▏ | 18536/30000 [11:48:03<7:08:01, 2.24s/it] 62%|██████▏ | 18537/30000 [11:48:06<8:15:01, 2.59s/it] 62%|██████▏ | 18538/30000 [11:48:07<6:54:57, 2.17s/it] 62%|██████▏ | 18539/30000 [11:48:08<6:01:28, 1.89s/it] 62%|██████▏ | 18540/30000 [11:48:10<5:20:50, 1.68s/it] 62%|██████▏ | 18540/30000 [11:48:10<5:20:50, 1.68s/it] 62%|██████▏ | 18541/30000 [11:48:11<4:51:34, 1.53s/it] 62%|██████▏ | 18542/30000 [11:48:12<4:31:05, 1.42s/it] 62%|██████▏ | 18543/30000 [11:48:13<4:17:50, 1.35s/it] 62%|██████▏ | 18544/30000 [11:48:18<7:53:13, 2.48s/it] 62%|██████▏ | 18545/30000 [11:48:22<8:45:26, 2.75s/it] 62%|██████▏ | 18546/30000 [11:48:23<7:13:43, 2.27s/it] 62%|██████▏ | 18547/30000 [11:48:24<6:11:32, 1.95s/it] 62%|██████▏ | 18548/30000 [11:48:25<5:25:50, 1.71s/it] 62%|██████▏ | 18549/30000 [11:48:26<4:54:50, 1.54s/it] 62%|██████▏ | 18550/30000 [11:48:27<4:33:13, 1.43s/it] 62%|██████▏ | 18550/30000 [11:48:27<4:33:13, 1.43s/it] 62%|██████▏ | 18551/30000 [11:48:29<4:18:29, 1.35s/it] 62%|██████▏ | 18552/30000 [11:48:34<8:05:20, 2.54s/it] 62%|██████▏ | 18553/30000 [11:48:37<8:46:17, 2.76s/it] 62%|██████▏ | 18554/30000 [11:48:38<7:15:31, 2.28s/it] 62%|██████▏ | 18555/30000 [11:48:40<6:14:07, 1.96s/it] 62%|██████▏ | 18556/30000 [11:48:41<5:28:36, 1.72s/it] 62%|██████▏ | 18557/30000 [11:48:42<4:57:01, 1.56s/it] 62%|██████▏ | 18558/30000 [11:48:43<4:35:42, 1.45s/it] 62%|██████▏ | 18559/30000 [11:48:44<4:19:18, 1.36s/it] 62%|██████▏ | 18560/30000 [11:48:49<7:42:16, 2.42s/it] 62%|██████▏ | 18560/30000 [11:48:49<7:42:16, 2.42s/it] 62%|██████▏ | 18561/30000 [11:48:52<8:18:07, 2.61s/it] 62%|██████▏ | 18562/30000 [11:48:53<6:54:58, 2.18s/it] 62%|██████▏ | 18563/30000 [11:48:55<5:57:56, 1.88s/it] 62%|██████▏ | 18564/30000 [11:48:56<5:16:50, 1.66s/it] 62%|██████▏ | 18565/30000 [11:48:57<4:49:37, 1.52s/it] 62%|██████▏ | 18566/30000 [11:48:58<4:29:38, 1.41s/it] 62%|██████▏ | 18567/30000 [11:48:59<4:15:26, 1.34s/it] 62%|██████▏ | 18568/30000 [11:49:05<8:09:37, 2.57s/it] 62%|██████▏ | 18569/30000 [11:49:08<9:05:10, 2.86s/it] 62%|██████▏ | 18570/30000 [11:49:09<7:27:45, 2.35s/it] 62%|██████▏ | 18570/30000 [11:49:09<7:27:45, 2.35s/it] 62%|██████▏ | 18571/30000 [11:49:11<6:21:42, 2.00s/it] 62%|██████▏ | 18572/30000 [11:49:12<5:34:08, 1.75s/it] 62%|██████▏ | 18573/30000 [11:49:13<4:59:39, 1.57s/it] 62%|██████▏ | 18574/30000 [11:49:14<4:36:39, 1.45s/it] 62%|██████▏ | 18575/30000 [11:49:15<4:20:34, 1.37s/it] 62%|██████▏ | 18576/30000 [11:49:20<7:38:30, 2.41s/it] 62%|██████▏ | 18577/30000 [11:49:23<8:01:50, 2.53s/it] 62%|██████▏ | 18578/30000 [11:49:24<6:44:50, 2.13s/it] 62%|██████▏ | 18579/30000 [11:49:25<5:51:10, 1.84s/it] 62%|██████▏ | 18580/30000 [11:49:26<5:12:49, 1.64s/it] 62%|██████▏ | 18580/30000 [11:49:26<5:12:49, 1.64s/it] 62%|██████▏ | 18581/30000 [11:49:28<4:45:14, 1.50s/it] 62%|██████▏ | 18582/30000 [11:49:29<4:27:53, 1.41s/it] 62%|██████▏ | 18583/30000 [11:49:30<4:14:35, 1.34s/it] 62%|██████▏ | 18584/30000 [11:49:35<7:57:10, 2.51s/it] 62%|██████▏ | 18585/30000 [11:49:38<7:48:17, 2.46s/it] 62%|██████▏ | 18586/30000 [11:49:39<6:34:08, 2.07s/it] 62%|██████▏ | 18587/30000 [11:49:40<5:44:00, 1.81s/it] 62%|██████▏ | 18588/30000 [11:49:41<5:07:08, 1.61s/it] 62%|██████▏ | 18589/30000 [11:49:42<4:42:46, 1.49s/it] 62%|██████▏ | 18590/30000 [11:49:43<4:25:16, 1.39s/it] 62%|██████▏ | 18590/30000 [11:49:43<4:25:16, 1.39s/it] 62%|██████▏ | 18591/30000 [11:49:45<4:13:11, 1.33s/it] 62%|██████▏ | 18592/30000 [11:49:50<7:50:00, 2.47s/it] 62%|██████▏ | 18593/30000 [11:49:52<7:12:56, 2.28s/it] 62%|██████▏ | 18594/30000 [11:49:53<6:09:35, 1.94s/it] 62%|██████▏ | 18595/30000 [11:49:54<5:26:51, 1.72s/it] 62%|██████▏ | 18596/30000 [11:49:55<4:54:08, 1.55s/it] 62%|██████▏ | 18597/30000 [11:49:56<4:32:37, 1.43s/it] 62%|██████▏ | 18598/30000 [11:49:57<4:18:11, 1.36s/it] 62%|██████▏ | 18599/30000 [11:49:59<4:06:32, 1.30s/it] 62%|██████▏ | 18600/30000 [11:50:04<8:16:45, 2.61s/it] +{'loss': 0.0153, 'grad_norm': 0.08602231740951538, 'learning_rate': 1.0999657233491156e-05, 'flow_matching_loss': 0.015297961980104446, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.3} +{'loss': 0.0154, 'grad_norm': 0.11109869927167892, 'learning_rate': 1.0983723814422808e-05, 'flow_matching_loss': 0.015396787598729134, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.3} +{'loss': 0.0138, 'grad_norm': 0.08248860388994217, 'learning_rate': 1.0967795275508735e-05, 'flow_matching_loss': 0.013839747942984103, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.3} +{'loss': 0.015, 'grad_norm': 0.118488609790802, 'learning_rate': 1.095187163610361e-05, 'flow_matching_loss': 0.014963547699153423, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.3} +{'loss': 0.0133, 'grad_norm': 0.07503626495599747, 'learning_rate': 1.0935952915556163e-05, 'flow_matching_loss': 0.01330208983272314, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.31} +{'loss': 0.0151, 'grad_norm': 0.08472942560911179, 'learning_rate': 1.0920039133209137e-05, 'flow_matching_loss': 0.015089655946940183, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.31} +{'loss': 0.0142, 'grad_norm': 0.07356730103492737, 'learning_rate': 1.0904130308399283e-05, 'flow_matching_loss': 0.014178075548261403, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.31} +{'loss': 0.0148, 'grad_norm': 0.06548425555229187, 'learning_rate': 1.0888226460457318e-05, 'flow_matching_loss': 0.014837050437927246, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.32} +{'loss': 0.014, 'grad_norm': 0.07497602701187134, 'learning_rate': 1.087232760870792e-05, 'flow_matching_loss': 0.013993861619383097, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.32} +{'loss': 0.0136, 'grad_norm': 0.08522956073284149, 'learning_rate': 1.0856433772469686e-05, 'flow_matching_loss': 0.01361953942105174, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.32} +{'loss': 0.0143, 'grad_norm': 0.0748744085431099, 'learning_rate': 1.0840544971055135e-05, 'flow_matching_loss': 0.014280647691339254, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.32} +{'loss': 0.0123, 'grad_norm': 0.07844077050685883, 'learning_rate': 1.0824661223770655e-05, 'flow_matching_loss': 0.012320141308009624, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.33} +{'loss': 0.0167, 'grad_norm': 0.12603890895843506, 'learning_rate': 1.0808782549916498e-05, 'flow_matching_loss': 0.0167470739223063, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.33} +{'loss': 0.0142, 'grad_norm': 0.06117792800068855, 'learning_rate': 1.079290896878675e-05, 'flow_matching_loss': 0.014190875925123692, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.33} +{'loss': 0.0146, 'grad_norm': 0.08312884718179703, 'learning_rate': 1.07770404996693e-05, 'flow_matching_loss': 0.01464774226769805, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.34} +{'loss': 0.0123, 'grad_norm': 0.10944189876317978, 'learning_rate': 1.0761177161845848e-05, 'flow_matching_loss': 0.012319139763712883, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.34} +{'loss': 0.015, 'grad_norm': 0.06116833910346031, 'learning_rate': 1.0745318974591841e-05, 'flow_matching_loss': 0.015030022151768208, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.34} +{'loss': 0.0145, 'grad_norm': 0.08094902336597443, 'learning_rate': 1.0729465957176469e-05, 'flow_matching_loss': 0.014492313284426928, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.34} +{'loss': 0.0133, 'grad_norm': 0.0730765238404274, 'learning_rate': 1.0713618128862642e-05, 'flow_matching_loss': 0.01333428779616952, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.35} +{'loss': 0.0139, 'grad_norm': 0.05362412706017494, 'learning_rate': 1.0697775508906969e-05, 'flow_matching_loss': 0.0139237679541111, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.35} +{'loss': 0.0138, 'grad_norm': 0.060967400670051575, 'learning_rate': 1.068193811655972e-05, 'flow_matching_loss': 0.013760816212743521, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.35} +{'loss': 0.0147, 'grad_norm': 0.0787210762500763, 'learning_rate': 1.0666105971064835e-05, 'flow_matching_loss': 0.014660908747464419, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.36} +{'loss': 0.0137, 'grad_norm': 0.09715977311134338, 'learning_rate': 1.0650279091659852e-05, 'flow_matching_loss': 0.013746879808604718, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.36} +{'loss': 0.0131, 'grad_norm': 0.07471799105405807, 'learning_rate': 1.0634457497575926e-05, 'flow_matching_loss': 0.013131294772028923, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.36} +{'loss': 0.0146, 'grad_norm': 0.0822833850979805, 'learning_rate': 1.061864120803778e-05, 'flow_matching_loss': 0.014621380716562271, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.37} +{'loss': 0.0139, 'grad_norm': 0.06635168194770813, 'learning_rate': 1.0602830242263701e-05, 'flow_matching_loss': 0.013868337869644165, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.37} +{'loss': 0.0136, 'grad_norm': 0.07220041751861572, 'learning_rate': 1.0587024619465506e-05, 'flow_matching_loss': 0.013565827300772071, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.37} +{'loss': 0.0136, 'grad_norm': 0.05307985842227936, 'learning_rate': 1.0571224358848517e-05, 'flow_matching_loss': 0.01360450629144907, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.37} +{'loss': 0.0143, 'grad_norm': 0.07804977148771286, 'learning_rate': 1.0555429479611537e-05, 'flow_matching_loss': 0.014258207194507122, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.38} +{'loss': 0.0138, 'grad_norm': 0.07200499624013901, 'learning_rate': 1.0539640000946835e-05, 'flow_matching_loss': 0.013797235395759343, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.38} +{'loss': 0.0125, 'grad_norm': 0.06504882127046585, 'learning_rate': 1.0523855942040111e-05, 'flow_matching_loss': 0.012477932963520288, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.38} +{'loss': 0.0133, 'grad_norm': 0.05906660482287407, 'learning_rate': 1.0508077322070496e-05, 'flow_matching_loss': 0.013339301571249962, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.39} +{'loss': 0.0139, 'grad_norm': 0.08986791223287582, 'learning_rate': 1.0492304160210495e-05, 'flow_matching_loss': 0.013868662621825933, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.39} +{'loss': 0.0152, 'grad_norm': 0.08087968826293945, 'learning_rate': 1.0476536475625985e-05, 'flow_matching_loss': 0.015176227409392595, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.39} +{'loss': 0.016, 'grad_norm': 0.0920604020357132, 'learning_rate': 1.0460774287476195e-05, 'flow_matching_loss': 0.01600948264822364, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.39} +{'loss': 0.0132, 'grad_norm': 0.06203969195485115, 'learning_rate': 1.044501761491366e-05, 'flow_matching_loss': 0.013185323029756547, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.4} +{'loss': 0.013, 'grad_norm': 0.08109024167060852, 'learning_rate': 1.0429266477084235e-05, 'flow_matching_loss': 0.012988746911287308, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.4} +{'loss': 0.015, 'grad_norm': 0.08345464617013931, 'learning_rate': 1.0413520893127034e-05, 'flow_matching_loss': 0.014999587927013636, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.4} +{'loss': 0.0131, 'grad_norm': 0.0789119228720665, 'learning_rate': 1.0397780882174424e-05, 'flow_matching_loss': 0.013098410982638597, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.41} +{'loss': 0.0139, 'grad_norm': 0.07319469749927521, 'learning_rate': 1.0382046463351998e-05, 'flow_matching_loss': 0.013907326105982065, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.41} +{'loss': 0.0149, 'grad_norm': 0.0759134367108345, 'learning_rate': 1.0366317655778568e-05, 'flow_matching_loss': 0.01491400832310319, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.41} 62%|██████▏ | 18600/30000 [11:50:04<8:16:45, 2.61s/it] 62%|██████▏ | 18601/30000 [11:50:06<7:33:09, 2.39s/it] 62%|██████▏ | 18602/30000 [11:50:07<6:23:45, 2.02s/it] 62%|██████▏ | 18603/30000 [11:50:08<5:36:35, 1.77s/it] 62%|██████▏ | 18604/30000 [11:50:10<5:02:34, 1.59s/it] 62%|██████▏ | 18605/30000 [11:50:11<4:38:07, 1.46s/it] 62%|██████▏ | 18606/30000 [11:50:12<4:22:14, 1.38s/it] 62%|██████▏ | 18607/30000 [11:50:13<4:12:40, 1.33s/it] 62%|██████▏ | 18608/30000 [11:50:19<8:34:02, 2.71s/it] 62%|██████▏ | 18609/30000 [11:50:21<7:21:57, 2.33s/it] 62%|██████▏ | 18610/30000 [11:50:22<6:16:04, 1.98s/it] 62%|██████▏ | 18610/30000 [11:50:22<6:16:04, 1.98s/it] 62%|██████▏ | 18611/30000 [11:50:23<5:30:38, 1.74s/it] 62%|██████▏ | 18612/30000 [11:50:24<4:57:21, 1.57s/it] 62%|██████▏ | 18613/30000 [11:50:25<4:34:39, 1.45s/it] 62%|██████▏ | 18614/30000 [11:50:26<4:20:13, 1.37s/it] 62%|██████▏ | 18615/30000 [11:50:28<4:08:40, 1.31s/it] 62%|██████▏ | 18616/30000 [11:50:34<9:18:14, 2.94s/it] 62%|██████▏ | 18617/30000 [11:50:36<7:36:55, 2.41s/it] 62%|██████▏ | 18618/30000 [11:50:37<6:27:12, 2.04s/it] 62%|██████▏ | 18619/30000 [11:50:38<5:38:38, 1.79s/it] 62%|██████▏ | 18620/30000 [11:50:39<5:03:25, 1.60s/it] 62%|██████▏ | 18620/30000 [11:50:39<5:03:25, 1.60s/it] 62%|██████▏ | 18621/30000 [11:50:40<4:40:37, 1.48s/it] 62%|██████▏ | 18622/30000 [11:50:41<4:23:43, 1.39s/it] 62%|██████▏ | 18623/30000 [11:50:43<4:11:12, 1.32s/it] 62%|██████▏ | 18624/30000 [11:50:49<8:40:24, 2.74s/it] 62%|██████▏ | 18625/30000 [11:50:50<7:18:31, 2.31s/it] 62%|██████▏ | 18626/30000 [11:50:51<6:13:47, 1.97s/it] 62%|██████▏ | 18627/30000 [11:50:52<5:28:46, 1.73s/it] 62%|██████▏ | 18628/30000 [11:50:54<4:56:48, 1.57s/it] 62%|██████▏ | 18629/30000 [11:50:55<4:34:01, 1.45s/it] 62%|██████▏ | 18630/30000 [11:50:56<4:19:32, 1.37s/it] 62%|██████▏ | 18630/30000 [11:50:56<4:19:32, 1.37s/it] 62%|██████▏ | 18631/30000 [11:50:57<4:09:05, 1.31s/it] 62%|██████▏ | 18632/30000 [11:51:03<8:25:46, 2.67s/it] 62%|██████▏ | 18633/30000 [11:51:04<7:00:25, 2.22s/it] 62%|██████▏ | 18634/30000 [11:51:05<6:00:25, 1.90s/it] 62%|██████▏ | 18635/30000 [11:51:06<5:21:04, 1.70s/it] 62%|██████▏ | 18636/30000 [11:51:08<4:51:16, 1.54s/it] 62%|██████▏ | 18637/30000 [11:51:09<4:31:23, 1.43s/it] 62%|██████▏ | 18638/30000 [11:51:10<4:17:31, 1.36s/it] 62%|██████▏ | 18639/30000 [11:51:11<4:06:41, 1.30s/it] 62%|██████▏ | 18640/30000 [11:51:17<8:22:28, 2.65s/it] 62%|██████▏ | 18640/30000 [11:51:17<8:22:28, 2.65s/it] 62%|██████▏ | 18641/30000 [11:51:18<6:58:41, 2.21s/it] 62%|██████▏ | 18642/30000 [11:51:19<5:59:59, 1.90s/it] 62%|██████▏ | 18643/30000 [11:51:21<5:21:17, 1.70s/it] 62%|██████▏ | 18644/30000 [11:51:22<4:52:16, 1.54s/it] 62%|██████▏ | 18645/30000 [11:51:23<4:31:13, 1.43s/it] 62%|██████▏ | 18646/30000 [11:51:24<4:17:11, 1.36s/it] 62%|██████▏ | 18647/30000 [11:51:25<4:07:36, 1.31s/it] 62%|██████▏ | 18648/30000 [11:51:31<7:56:14, 2.52s/it] 62%|██████▏ | 18649/30000 [11:51:33<7:21:38, 2.33s/it] 62%|██████▏ | 18650/30000 [11:51:34<6:15:36, 1.99s/it] 62%|██████▏ | 18650/30000 [11:51:34<6:15:36, 1.99s/it] 62%|██████▏ | 18651/30000 [11:51:35<5:30:38, 1.75s/it] 62%|██████▏ | 18652/30000 [11:51:36<4:57:54, 1.58s/it] 62%|██████▏ | 18653/30000 [11:51:37<4:35:13, 1.46s/it] 62%|██████▏ | 18654/30000 [11:51:38<4:18:24, 1.37s/it] 62%|██████▏ | 18655/30000 [11:51:40<4:07:20, 1.31s/it] 62%|██████▏ | 18656/30000 [11:51:45<8:06:26, 2.57s/it] 62%|██████▏ | 18657/30000 [11:51:47<7:41:17, 2.44s/it] 62%|██████▏ | 18658/30000 [11:51:48<6:28:43, 2.06s/it] 62%|██████▏ | 18659/30000 [11:51:50<5:40:09, 1.80s/it] 62%|██████▏ | 18660/30000 [11:51:51<5:03:59, 1.61s/it] 62%|██████▏ | 18660/30000 [11:51:51<5:03:59, 1.61s/it] 62%|██████▏ | 18661/30000 [11:51:52<4:41:05, 1.49s/it] 62%|██████▏ | 18662/30000 [11:51:53<4:24:06, 1.40s/it] 62%|██████▏ | 18663/30000 [11:51:54<4:10:25, 1.33s/it] 62%|██████▏ | 18664/30000 [11:52:00<7:59:28, 2.54s/it] 62%|██████▏ | 18665/30000 [11:52:01<7:04:20, 2.25s/it] 62%|██████▏ | 18666/30000 [11:52:02<6:03:24, 1.92s/it] 62%|██████▏ | 18667/30000 [11:52:04<5:22:03, 1.71s/it] 62%|██████▏ | 18668/30000 [11:52:05<4:51:37, 1.54s/it] 62%|██████▏ | 18669/30000 [11:52:06<4:31:50, 1.44s/it] 62%|██████▏ | 18670/30000 [11:52:07<4:17:53, 1.37s/it] 62%|██████▏ | 18670/30000 [11:52:07<4:17:53, 1.37s/it] 62%|██████▏ | 18671/30000 [11:52:08<4:07:40, 1.31s/it] 62%|██████▏ | 18672/30000 [11:52:14<8:25:55, 2.68s/it] 62%|██████▏ | 18673/30000 [11:52:16<7:30:42, 2.39s/it] 62%|██████▏ | 18674/30000 [11:52:17<6:21:57, 2.02s/it] 62%|██████▏ | 18675/30000 [11:52:18<5:34:49, 1.77s/it] 62%|██████▏ | 18676/30000 [11:52:19<5:00:09, 1.59s/it] 62%|██████▏ | 18677/30000 [11:52:21<4:35:45, 1.46s/it] 62%|██████▏ | 18678/30000 [11:52:22<4:19:56, 1.38s/it] 62%|██████▏ | 18679/30000 [11:52:23<4:09:04, 1.32s/it] 62%|██████▏ | 18680/30000 [11:52:29<8:53:25, 2.83s/it] 62%|██████▏ | 18680/30000 [11:52:29<8:53:25, 2.83s/it] 62%|██████▏ | 18681/30000 [11:52:31<8:06:09, 2.58s/it] 62%|██████▏ | 18682/30000 [11:52:32<6:46:30, 2.16s/it] 62%|██████▏ | 18683/30000 [11:52:34<5:52:10, 1.87s/it] 62%|██████▏ | 18684/30000 [11:52:35<5:12:58, 1.66s/it] 62%|██████▏ | 18685/30000 [11:52:36<4:45:16, 1.51s/it] 62%|██████▏ | 18686/30000 [11:52:37<4:26:58, 1.42s/it] 62%|██████▏ | 18687/30000 [11:52:38<4:13:55, 1.35s/it] 62%|██████▏ | 18688/30000 [11:52:44<8:02:37, 2.56s/it] 62%|██████▏ | 18689/30000 [11:52:46<7:59:59, 2.55s/it] 62%|██████▏ | 18690/30000 [11:52:47<6:42:14, 2.13s/it] 62%|██████▏ | 18690/30000 [11:52:47<6:42:14, 2.13s/it] 62%|██████▏ | 18691/30000 [11:52:49<5:51:54, 1.87s/it] 62%|██████▏ | 18692/30000 [11:52:50<5:15:03, 1.67s/it] 62%|██████▏ | 18693/30000 [11:52:51<4:47:48, 1.53s/it] 62%|██████▏ | 18694/30000 [11:52:52<4:28:30, 1.42s/it] 62%|██████▏ | 18695/30000 [11:52:53<4:14:11, 1.35s/it] 62%|██████▏ | 18696/30000 [11:52:59<7:53:51, 2.52s/it] 62%|██████▏ | 18697/30000 [11:53:01<7:43:15, 2.46s/it] 62%|██████▏ | 18698/30000 [11:53:02<6:31:56, 2.08s/it] 62%|██████▏ | 18699/30000 [11:53:03<5:41:56, 1.82s/it] 62%|██████▏ | 18700/30000 [11:53:05<5:05:44, 1.62s/it] 62%|██████▏ | 18700/30000 [11:53:05<5:05:44, 1.62s/it] 62%|██████▏ | 18701/30000 [11:53:06<4:40:18, 1.49s/it] 62%|██████▏ | 18702/30000 [11:53:07<4:23:05, 1.40s/it] 62%|██████▏ | 18703/30000 [11:53:08<4:10:54, 1.33s/it] 62%|██████▏ | 18704/30000 [11:53:13<7:35:11, 2.42s/it] 62%|██████▏ | 18705/30000 [11:53:16<7:48:49, 2.49s/it] 62%|██████▏ | 18706/30000 [11:53:17<6:34:30, 2.10s/it] 62%|██████▏ | 18707/30000 [11:53:18<5:43:37, 1.83s/it] 62%|██████▏ | 18708/30000 [11:53:19<5:06:31, 1.63s/it] 62%|██████▏ | 18709/30000 [11:53:20<4:39:22, 1.48s/it] 62%|██████▏ | 18710/30000 [11:53:22<4:22:29, 1.39s/it] 62%|██████▏ | 18710/30000 [11:53:22<4:22:29, 1.39s/it] 62%|██████▏ | 18711/30000 [11:53:23<4:10:31, 1.33s/it] 62%|██████▏ | 18712/30000 [11:53:28<7:56:21, 2.53s/it] 62%|██████▏ | 18713/30000 [11:53:31<8:11:45, 2.61s/it] 62%|██████▏ | 18714/30000 [11:53:32<6:50:36, 2.18s/it] 62%|██████▏ | 18715/30000 [11:53:33<5:54:18, 1.88s/it] 62%|██████▏ | 18716/30000 [11:53:34<5:13:45, 1.67s/it] 62%|██████▏ | 18717/30000 [11:53:36<4:46:08, 1.52s/it] 62%|██████▏ | 18718/30000 [11:53:37<4:26:12, 1.42s/it] 62%|██████▏ | 18719/30000 [11:53:38<4:12:03, 1.34s/it] 62%|██████▏ | 18720/30000 [11:53:44<8:52:42, 2.83s/it] 62%|██████▏ | 18720/30000 [11:53:44<8:52:42, 2.83s/it] 62%|██████▏ | 18721/30000 [11:53:47<8:28:37, 2.71s/it] 62%|██████▏ | 18722/30000 [11:53:48<7:01:36, 2.24s/it] 62%|██████▏ | 18723/30000 [11:53:49<6:01:51, 1.93s/it] 62%|██████▏ | 18724/30000 [11:53:50<5:19:19, 1.70s/it] 62%|██████▏ | 18725/30000 [11:53:51<4:49:08, 1.54s/it] 62%|██████▏ | 18726/30000 [11:53:53<4:29:44, 1.44s/it] 62%|██████▏ | 18727/30000 [11:53:54<4:15:26, 1.36s/it] 62%|██████▏ | 18728/30000 [11:53:59<8:13:07, 2.62s/it] 62%|██████▏ | 18729/30000 [11:54:02<8:22:35, 2.68s/it] 62%|██████▏ | 18730/30000 [11:54:03<6:58:36, 2.23s/it] 62%|██████▏ | 18730/30000 [11:54:03<6:58:36, 2.23s/it] 62%|██████▏ | 18731/30000 [11:54:05<6:00:28, 1.92s/it] 62%|██████▏ | 18732/30000 [11:54:06<5:17:51, 1.69s/it] 62%|██████▏ | 18733/30000 [11:54:07<4:47:17, 1.53s/it] 62%|██████▏ | 18734/30000 [11:54:08<4:28:00, 1.43s/it] 62%|██████▏ | 18735/30000 [11:54:09<4:14:56, 1.36s/it] 62%|██████▏ | 18736/30000 [11:54:15<8:21:11, 2.67s/it] 62%|██████▏ | 18737/30000 [11:54:17<7:53:34, 2.52s/it] 62%|██████▏ | 18738/30000 [11:54:18<6:37:01, 2.12s/it] 62%|██████▏ | 18739/30000 [11:54:20<5:44:33, 1.84s/it] 62%|██████▏ | 18740/30000 [11:54:21<5:07:11, 1.64s/it] 62%|██████▏ | 18740/30000 [11:54:21<5:07:11, 1.64s/it] 62%|██████▏ | 18741/30000 [11:54:22<4:40:57, 1.50s/it] 62%|██████▏ | 18742/30000 [11:54:23<4:23:17, 1.40s/it] 62%|██████▏ | 18743/30000 [11:54:24<4:10:46, 1.34s/it] 62%|██████▏ | 18744/30000 [11:54:31<9:01:41, 2.89s/it] 62%|██████▏ | 18745/30000 [11:54:33<8:04:27, 2.58s/it] 62%|██████▏ | 18746/30000 [11:54:34<6:44:31, 2.16s/it] 62%|██████▏ | 18747/30000 [11:54:35<5:50:03, 1.87s/it] 62%|██████▏ | 18748/30000 [11:54:36<5:10:43, 1.66s/it] 62%|██████▏ | 18749/30000 [11:54:37<4:43:21, 1.51s/it] 62%|██████▎ | 18750/30000 [11:54:38<4:24:59, 1.41s/it] 62%|██████▎ | 18750/30000 [11:54:38<4:24:59, 1.41s/it] 63%|██████▎ | 18751/30000 [11:54:40<4:12:47, 1.35s/it] 63%|██████▎ | 18752/30000 [11:54:46<9:18:17, 2.98s/it] 63%|██████▎ | 18753/30000 [11:54:48<8:02:54, 2.58s/it] 63%|██████▎ | 18754/30000 [11:54:49<6:45:12, 2.16s/it] 63%|██████▎ | 18755/30000 [11:54:50<5:51:01, 1.87s/it] 63%|██████▎ | 18756/30000 [11:54:52<5:11:33, 1.66s/it] 63%|██████▎ | 18757/30000 [11:54:53<4:43:19, 1.51s/it] 63%|██████▎ | 18758/30000 [11:54:54<4:25:27, 1.42s/it] 63%|██████▎ | 18759/30000 [11:54:55<4:13:18, 1.35s/it] 63%|██████▎ | 18760/30000 [11:55:02<9:42:52, 3.11s/it] 63%|██████▎ | 18760/30000 [11:55:02<9:42:52, 3.11s/it] 63%|██████▎ | 18761/30000 [11:55:05<8:46:55, 2.81s/it] 63%|██████▎ | 18762/30000 [11:55:06<7:14:03, 2.32s/it] 63%|██████▎ | 18763/30000 [11:55:07<6:11:26, 1.98s/it] 63%|██████▎ | 18764/30000 [11:55:08<5:26:43, 1.74s/it] 63%|██████▎ | 18765/30000 [11:55:09<4:53:50, 1.57s/it] 63%|██████▎ | 18766/30000 [11:55:10<4:32:38, 1.46s/it] 63%|██████▎ | 18767/30000 [11:55:12<4:17:25, 1.38s/it] 63%|██████▎ | 18768/30000 [11:55:18<9:11:20, 2.95s/it] 63%|██████▎ | 18769/30000 [11:55:20<8:30:31, 2.73s/it] 63%|██████▎ | 18770/30000 [11:55:22<7:02:42, 2.26s/it] 63%|██████▎ | 18770/30000 [11:55:22<7:02:42, 2.26s/it] 63%|██████▎ | 18771/30000 [11:55:23<6:03:08, 1.94s/it] 63%|██████▎ | 18772/30000 [11:55:24<5:19:21, 1.71s/it] 63%|██████▎ | 18773/30000 [11:55:25<4:49:32, 1.55s/it] 63%|██████▎ | 18774/30000 [11:55:26<4:28:57, 1.44s/it] 63%|██████▎ | 18775/30000 [11:55:28<4:14:40, 1.36s/it] 63%|██████▎ | 18776/30000 [11:55:34<9:01:56, 2.90s/it] 63%|██████▎ | 18777/30000 [11:55:36<8:18:24, 2.66s/it] 63%|██████▎ | 18778/30000 [11:55:37<6:55:56, 2.22s/it] 63%|██████▎ | 18779/30000 [11:55:38<5:57:17, 1.91s/it] 63%|██████▎ | 18780/30000 [11:55:40<5:15:33, 1.69s/it] 63%|██████▎ | 18780/30000 [11:55:40<5:15:33, 1.69s/it] 63%|██████▎ | 18781/30000 [11:55:41<4:46:18, 1.53s/it] 63%|██████▎ | 18782/30000 [11:55:42<4:26:50, 1.43s/it] 63%|██████▎ | 18783/30000 [11:55:43<4:13:03, 1.35s/it] 63%|██████▎ | 18784/30000 [11:55:50<9:28:14, 3.04s/it] 63%|██████▎ | 18785/30000 [11:55:52<8:15:59, 2.65s/it] 63%|██████▎ | 18786/30000 [11:55:53<6:52:08, 2.21s/it] 63%|██████▎ | 18787/30000 [11:55:54<5:54:54, 1.90s/it] 63%|██████▎ | 18788/30000 [11:55:55<5:13:02, 1.68s/it] 63%|██████▎ | 18789/30000 [11:55:57<4:45:48, 1.53s/it] 63%|██████▎ | 18790/30000 [11:55:58<4:25:08, 1.42s/it] 63%|██████▎ | 18790/30000 [11:55:58<4:25:08, 1.42s/it] 63%|██████▎ | 18791/30000 [11:55:59<4:11:01, 1.34s/it] 63%|██████▎ | 18792/30000 [11:56:06<9:52:38, 3.17s/it] 63%|██████▎ | 18793/30000 [11:56:08<8:33:26, 2.75s/it] 63%|██████▎ | 18794/30000 [11:56:09<7:04:37, 2.27s/it] 63%|██████▎ | 18795/30000 [11:56:10<6:03:32, 1.95s/it] 63%|██████▎ | 18796/30000 [11:56:12<5:19:56, 1.71s/it] 63%|██████▎ | 18797/30000 [11:56:13<4:50:20, 1.56s/it] 63%|██████▎ | 18798/30000 [11:56:14<4:30:03, 1.45s/it] 63%|██████▎ | 18799/30000 [11:56:15<4:15:39, 1.37s/it] 63%|██████▎ | 18800/30000 [11:56:23<10:15:26, 3.30s/it] 63%|██████▎ | 18800/30000 [11:56:23<10:15:26, 3.30s/it] 63%|██████▎ | 18801/30000 [11:56:24<8:31:01, 2.74s/it] 63%|██████▎ | 18802/30000 [11:56:26<7:02:40, 2.26s/it] 63%|██████▎ | 18803/30000 [11:56:27<6:02:02, 1.94s/it] 63%|██████▎ | 18804/30000 [11:56:28<5:18:24, 1.71s/it] 63%|██████▎ | 18805/30000 [11:56:29<4:48:34, 1.55s/it] 63%|██████▎ | 18806/30000 [11:56:30<4:28:24, 1.44s/it] 63%|██████▎ | 18807/30000 [11:56:32<4:14:37, 1.36s/it] 63%|██████▎ | 18808/30000 [11:56:38<9:27:57, 3.04s/it] 63%|██████▎ | 18809/30000 [11:56:40<7:57:57, 2.56s/it] 63%|██████▎ | 18810/30000 [11:56:41<6:39:50, 2.14s/it] 63%|██████▎ | 18810/30000 [11:56:41<6:39:50, 2.14s/it] 63%|██████▎ | 18811/30000 [11:56:42<5:45:49, 1.85s/it] 63%|██████▎ | 18812/30000 [11:56:43<5:08:05, 1.65s/it] 63%|██████▎ | 18813/30000 [11:56:45<4:41:40, 1.51s/it] 63%|██████▎ | 18814/30000 [11:56:46<4:23:08, 1.41s/it] 63%|██████▎ | 18815/30000 [11:56:47<4:10:02, 1.34s/it] 63%|██████▎ | 18816/30000 [11:56:54<9:50:52, 3.17s/it] 63%|██████▎ | 18817/30000 [11:56:56<7:59:58, 2.58s/it] 63%|██████▎ | 18818/30000 [11:56:57<6:43:35, 2.17s/it] 63%|██████▎ | 18819/30000 [11:56:58<5:48:55, 1.87s/it] 63%|██████▎ | 18820/30000 [11:56:59<5:09:51, 1.66s/it] 63%|██████▎ | 18820/30000 [11:56:59<5:09:51, 1.66s/it] 63%|██████▎ | 18821/30000 [11:57:00<4:44:34, 1.53s/it] 63%|██████▎ | 18822/30000 [11:57:02<4:25:22, 1.42s/it] 63%|██████▎ | 18823/30000 [11:57:03<4:11:48, 1.35s/it] 63%|██████▎ | 18824/30000 [11:57:09<8:38:37, 2.78s/it] 63%|██████▎ | 18825/30000 [11:57:10<7:09:53, 2.31s/it] 63%|██████▎ | 18826/30000 [11:57:11<6:06:18, 1.97s/it] 63%|██████▎ | 18827/30000 [11:57:12<5:22:49, 1.73s/it] 63%|██████▎ | 18828/30000 [11:57:14<4:52:46, 1.57s/it] 63%|██████▎ | 18829/30000 [11:57:15<4:33:24, 1.47s/it] 63%|██████▎ | 18830/30000 [11:57:16<4:17:19, 1.38s/it] 63%|██████▎ | 18830/30000 [11:57:16<4:17:19, 1.38s/it] 63%|██████▎ | 18831/30000 [11:57:17<4:06:16, 1.32s/it] 63%|██████▎ | 18832/30000 [11:57:24<8:47:42, 2.84s/it] 63%|██████▎ | 18833/30000 [11:57:25<7:25:42, 2.39s/it] 63%|██████▎ | 18834/30000 [11:57:26<6:18:10, 2.03s/it] 63%|██████▎ | 18835/30000 [11:57:27<5:31:53, 1.78s/it] 63%|██████▎ | 18836/30000 [11:57:29<4:57:51, 1.60s/it] 63%|██████▎ | 18837/30000 [11:57:30<4:33:55, 1.47s/it] 63%|██████▎ | 18838/30000 [11:57:31<4:17:37, 1.38s/it] 63%|██████▎ | 18839/30000 [11:57:32<4:06:10, 1.32s/it] 63%|██████▎ | 18840/30000 [11:57:39<9:41:50, 3.13s/it] 63%|██████▎ | 18840/30000 [11:57:39<9:41:50, 3.13s/it] 63%|██████▎ | 18841/30000 [11:57:41<8:21:47, 2.70s/it] 63%|██████▎ | 18842/30000 [11:57:42<6:56:55, 2.24s/it] 63%|██████▎ | 18843/30000 [11:57:43<5:59:43, 1.93s/it] 63%|██████▎ | 18844/30000 [11:57:45<5:16:31, 1.70s/it] 63%|██████▎ | 18845/30000 [11:57:46<4:47:31, 1.55s/it] 63%|██████▎ | 18846/30000 [11:57:47<4:27:35, 1.44s/it] 63%|██████▎ | 18847/30000 [11:57:48<4:14:58, 1.37s/it] 63%|██████▎ | 18848/30000 [11:57:55<9:17:52, 3.00s/it] 63%|██████▎ | 18849/30000 [11:57:56<7:35:24, 2.45s/it] 63%|██████▎ | 18850/30000 [11:57:57<6:24:20, 2.07s/it] 63%|██████▎ | 18850/30000 [11:57:57<6:24:20, 2.07s/it] 63%|██████▎ | 18851/30000 [11:57:59<5:36:46, 1.81s/it] 63%|██████▎ | 18852/30000 [11:58:00<5:00:41, 1.62s/it] 63%|██████▎ | 18853/30000 [11:58:01<4:36:02, 1.49s/it] 63%|██████▎ | 18854/30000 [11:58:02<4:19:38, 1.40s/it] 63%|██████▎ | 18855/30000 [11:58:03<4:08:24, 1.34s/it] 63%|██████▎ | 18856/30000 [11:58:10<9:15:35, 2.99s/it] 63%|██████▎ | 18857/30000 [11:58:11<7:34:41, 2.45s/it] 63%|██████▎ | 18858/30000 [11:58:13<6:25:25, 2.08s/it] 63%|██████▎ | 18859/30000 [11:58:14<5:36:31, 1.81s/it] 63%|██████▎ | 18860/30000 [11:58:15<5:00:32, 1.62s/it] 63%|██████▎ | 18860/30000 [11:58:15<5:00:32, 1.62s/it] 63%|██████▎ | 18861/30000 [11:58:16<4:36:44, 1.49s/it] 63%|██████▎ | 18862/30000 [11:58:17<4:19:30, 1.40s/it] 63%|██████▎ | 18863/30000 [11:58:18<4:08:10, 1.34s/it] 63%|██████▎ | 18864/30000 [11:58:25<9:00:26, 2.91s/it] 63%|██████▎ | 18865/30000 [11:58:26<7:24:15, 2.39s/it] 63%|██████▎ | 18866/30000 [11:58:27<6:16:15, 2.03s/it] 63%|██████▎ | 18867/30000 [11:58:29<5:30:10, 1.78s/it] 63%|██████▎ | 18868/30000 [11:58:30<4:55:53, 1.59s/it] 63%|██████▎ | 18869/30000 [11:58:31<4:32:42, 1.47s/it] 63%|██████▎ | 18870/30000 [11:58:32<4:16:20, 1.38s/it] 63%|██████▎ | 18870/30000 [11:58:32<4:16:20, 1.38s/it] 63%|██████▎ | 18871/30000 [11:58:33<4:04:29, 1.32s/it] 63%|██████▎ | 18872/30000 [11:58:42<10:55:17, 3.53s/it] 63%|██████▎ | 18873/30000 [11:58:43<8:43:56, 2.83s/it] 63%|██████▎ | 18874/30000 [11:58:44<7:11:52, 2.33s/it] 63%|██████▎ | 18875/30000 [11:58:46<6:08:22, 1.99s/it] 63%|██████▎ | 18876/30000 [11:58:47<5:23:14, 1.74s/it] 63%|██████▎ | 18877/30000 [11:58:48<4:52:45, 1.58s/it] 63%|██████▎ | 18878/30000 [11:58:49<4:30:26, 1.46s/it] 63%|██████▎ | 18879/30000 [11:58:50<4:14:31, 1.37s/it] 63%|██████▎ | 18880/30000 [11:58:59<10:38:56, 3.45s/it] 63%|██████▎ | 18880/30000 [11:58:59<10:38:56, 3.45s/it] 63%|██████▎ | 18881/30000 [11:59:00<8:34:42, 2.78s/it] 63%|██████▎ | 18882/30000 [11:59:01<7:07:10, 2.31s/it] 63%|██████▎ | 18883/30000 [11:59:02<6:05:13, 1.97s/it] 63%|██████▎ | 18884/30000 [11:59:03<5:19:48, 1.73s/it] 63%|██████▎ | 18885/30000 [11:59:05<4:49:06, 1.56s/it] 63%|██████▎ | 18886/30000 [11:59:06<4:27:14, 1.44s/it] 63%|██████▎ | 18887/30000 [11:59:07<4:12:37, 1.36s/it] 63%|██████▎ | 18888/30000 [11:59:14<9:17:38, 3.01s/it] 63%|██████▎ | 18889/30000 [11:59:15<7:36:19, 2.46s/it] 63%|██████▎ | 18890/30000 [11:59:16<6:24:04, 2.07s/it] 63%|██████▎ | 18890/30000 [11:59:16<6:24:04, 2.07s/it] 63%|██████▎ | 18891/30000 [11:59:17<5:35:50, 1.81s/it] 63%|██████▎ | 18892/30000 [11:59:18<5:00:52, 1.63s/it] 63%|██████▎ | 18893/30000 [11:59:20<4:37:06, 1.50s/it] 63%|██████▎ | 18894/30000 [11:59:21<4:19:19, 1.40s/it] 63%|██████▎ | 18895/30000 [11:59:22<4:06:58, 1.33s/it] 63%|██████▎ | 18896/30000 [11:59:30<9:52:11, 3.20s/it] 63%|██████▎ | 18897/30000 [11:59:31<7:59:53, 2.59s/it] 63%|██████▎ | 18898/30000 [11:59:32<6:40:37, 2.17s/it] 63%|██████▎ | 18899/30000 [11:59:33<5:46:28, 1.87s/it] 63%|██████▎ | 18900/30000 [11:59:34<5:08:03, 1.67s/it] 63%|██████▎ | 18900/30000 [11:59:34<5:08:03, 1.67s/it] 63%|██████▎ | 18901/30000 [11:59:35<4:41:54, 1.52s/it] 63%|██████▎ | 18902/30000 [11:59:37<4:23:02, 1.42s/it] 63%|██████▎ | 18903/30000 [11:59:38<4:09:03, 1.35s/it] 63%|██████▎ | 18904/30000 [11:59:46<10:09:08, 3.29s/it] 63%|██████▎ | 18905/30000 [11:59:47<8:11:32, 2.66s/it] 63%|██████▎ | 18906/30000 [11:59:48<6:48:23, 2.21s/it] 63%|██████▎ | 18907/30000 [11:59:49<5:52:36, 1.91s/it] 63%|██████▎ | 18908/30000 [11:59:50<5:11:53, 1.69s/it] 63%|██████▎ | 18909/30000 [11:59:52<4:44:05, 1.54s/it] 63%|██████▎ | 18910/30000 [11:59:53<4:23:57, 1.43s/it] 63%|██████▎ | 18910/30000 [11:59:53<4:23:57, 1.43s/it] 63%|██████▎ | 18911/30000 [11:59:54<4:09:39, 1.35s/it] 63%|██████▎ | 18912/30000 [12:00:01<9:54:22, 3.22s/it] 63%|██████▎ | 18913/30000 [12:00:03<8:00:56, 2.60s/it] 63%|██████▎ | 18914/30000 [12:00:04<6:40:23, 2.17s/it] 63%|██████▎ | 18915/30000 [12:00:05<5:46:42, 1.88s/it] 63%|██████▎ | 18916/30000 [12:00:06<5:07:05, 1.66s/it] 63%|██████▎ | 18917/30000 [12:00:07<4:41:01, 1.52s/it] 63%|██████▎ | 18918/30000 [12:00:09<4:22:34, 1.42s/it] 63%|██████▎ | 18919/30000 [12:00:10<4:08:38, 1.35s/it] 63%|██████▎ | 18920/30000 [12:00:18<10:25:30, 3.39s/it] 63%|██████▎ | 18920/30000 [12:00:18<10:25:30, 3.39s/it] 63%|██████▎ | 18921/30000 [12:00:19<8:22:10, 2.72s/it] 63%|██████▎ | 18922/30000 [12:00:20<6:55:15, 2.25s/it] 63%|██████▎ | 18923/30000 [12:00:21<5:56:16, 1.93s/it] 63%|██████▎ | 18924/30000 [12:00:23<5:13:59, 1.70s/it] 63%|██████▎ | 18925/30000 [12:00:24<4:46:54, 1.55s/it] 63%|██████▎ | 18926/30000 [12:00:25<4:26:33, 1.44s/it] 63%|██████▎ | 18927/30000 [12:00:26<4:12:40, 1.37s/it] 63%|██████▎ | 18928/30000 [12:00:33<9:37:45, 3.13s/it] 63%|██████▎ | 18929/30000 [12:00:35<7:48:53, 2.54s/it] 63%|██████▎ | 18930/30000 [12:00:36<6:32:25, 2.13s/it] 63%|██████▎ | 18930/30000 [12:00:36<6:32:25, 2.13s/it] 63%|██████▎ | 18931/30000 [12:00:37<5:39:58, 1.84s/it] 63%|██████▎ | 18932/30000 [12:00:38<5:02:32, 1.64s/it] 63%|██████▎ | 18933/30000 [12:00:39<4:37:23, 1.50s/it] 63%|██████▎ | 18934/30000 [12:00:40<4:19:46, 1.41s/it] 63%|██████▎ | 18935/30000 [12:00:42<4:07:03, 1.34s/it] 63%|██████▎ | 18936/30000 [12:00:50<10:30:54, 3.42s/it] 63%|██████▎ | 18937/30000 [12:00:51<8:26:21, 2.75s/it] 63%|██████▎ | 18938/30000 [12:00:52<6:59:17, 2.27s/it] 63%|██████▎ | 18939/30000 [12:00:53<5:59:22, 1.95s/it] 63%|██████▎ | 18940/30000 [12:00:55<5:16:19, 1.72s/it] 63%|██████▎ | 18940/30000 [12:00:55<5:16:19, 1.72s/it] 63%|██████▎ | 18941/30000 [12:00:56<4:46:53, 1.56s/it] 63%|██████▎ | 18942/30000 [12:00:57<4:26:06, 1.44s/it] 63%|██████▎ | 18943/30000 [12:00:58<4:11:05, 1.36s/it] 63%|██████▎ | 18944/30000 [12:01:06<10:15:33, 3.34s/it] 63%|██████▎ | 18945/30000 [12:01:08<8:45:11, 2.85s/it] 63%|██████▎ | 18946/30000 [12:01:09<7:11:40, 2.34s/it] 63%|██████▎ | 18947/30000 [12:01:10<6:07:52, 2.00s/it] 63%|██████▎ | 18948/30000 [12:01:11<5:22:20, 1.75s/it] 63%|██████▎ | 18949/30000 [12:01:12<4:51:21, 1.58s/it] 63%|██████▎ | 18950/30000 [12:01:14<4:29:44, 1.46s/it] 63%|██████▎ | 18950/30000 [12:01:14<4:29:44, 1.46s/it] 63%|██████▎ | 18951/30000 [12:01:15<4:14:13, 1.38s/it] 63%|██████▎ | 18952/30000 [12:01:22<9:23:14, 3.06s/it] 63%|██████▎ | 18953/30000 [12:01:25<9:12:30, 3.00s/it] 63%|██████▎ | 18954/30000 [12:01:26<7:31:12, 2.45s/it] 63%|██████▎ | 18955/30000 [12:01:27<6:23:36, 2.08s/it] 63%|██████▎ | 18956/30000 [12:01:28<5:33:44, 1.81s/it] 63%|██████▎ | 18957/30000 [12:01:29<4:59:07, 1.63s/it] 63%|██████▎ | 18958/30000 [12:01:31<4:35:29, 1.50s/it] 63%|██████▎ | 18959/30000 [12:01:32<4:18:04, 1.40s/it] 63%|██████▎ | 18960/30000 [12:01:39<9:10:13, 2.99s/it] 63%|██████▎ | 18960/30000 [12:01:39<9:10:13, 2.99s/it] 63%|██████▎ | 18961/30000 [12:01:42<9:27:21, 3.08s/it] 63%|██████▎ | 18962/30000 [12:01:43<7:41:06, 2.51s/it] 63%|██████▎ | 18963/30000 [12:01:44<6:27:58, 2.11s/it] 63%|██████▎ | 18964/30000 [12:01:45<5:36:34, 1.83s/it] 63%|██████▎ | 18965/30000 [12:01:47<5:01:12, 1.64s/it] 63%|██████▎ | 18966/30000 [12:01:48<4:38:18, 1.51s/it] 63%|██████▎ | 18967/30000 [12:01:49<4:21:21, 1.42s/it] 63%|██████▎ | 18968/30000 [12:01:55<8:26:19, 2.75s/it] 63%|██████▎ | 18969/30000 [12:01:58<8:55:19, 2.91s/it] 63%|██████▎ | 18970/30000 [12:01:59<7:19:40, 2.39s/it] 63%|██████▎ | 18970/30000 [12:01:59<7:19:40, 2.39s/it] 63%|██████▎ | 18971/30000 [12:02:00<6:13:40, 2.03s/it] 63%|██████▎ | 18972/30000 [12:02:02<5:26:02, 1.77s/it] 63%|██████▎ | 18973/30000 [12:02:03<4:53:54, 1.60s/it] 63%|██████▎ | 18974/30000 [12:02:04<4:29:37, 1.47s/it] 63%|██████▎ | 18975/30000 [12:02:05<4:14:10, 1.38s/it] 63%|██████▎ | 18976/30000 [12:02:11<8:18:14, 2.71s/it] 63%|██████▎ | 18977/30000 [12:02:14<8:33:31, 2.80s/it] 63%|██████▎ | 18978/30000 [12:02:15<7:05:16, 2.32s/it] 63%|██████▎ | 18979/30000 [12:02:16<6:03:51, 1.98s/it] 63%|██████▎ | 18980/30000 [12:02:18<5:18:39, 1.74s/it] 63%|██████▎ | 18980/30000 [12:02:18<5:18:39, 1.74s/it] 63%|██████▎ | 18981/30000 [12:02:19<4:49:02, 1.57s/it] 63%|██████▎ | 18982/30000 [12:02:20<4:27:50, 1.46s/it] 63%|██████▎ | 18983/30000 [12:02:21<4:11:40, 1.37s/it] 63%|██████▎ | 18984/30000 [12:02:27<8:24:14, 2.75s/it] 63%|██████▎ | 18985/30000 [12:02:30<8:59:35, 2.94s/it] 63%|██████▎ | 18986/30000 [12:02:32<7:21:42, 2.41s/it] 63%|██████▎ | 18987/30000 [12:02:33<6:13:44, 2.04s/it] 63%|██████▎ | 18988/30000 [12:02:34<5:25:58, 1.78s/it] 63%|██████▎ | 18989/30000 [12:02:35<4:53:50, 1.60s/it] 63%|██████▎ | 18990/30000 [12:02:36<4:30:48, 1.48s/it] 63%|██████▎ | 18990/30000 [12:02:36<4:30:48, 1.48s/it] 63%|██████▎ | 18991/30000 [12:02:38<4:14:35, 1.39s/it] 63%|██████▎ | 18992/30000 [12:02:43<8:02:47, 2.63s/it] 63%|██████▎ | 18993/30000 [12:02:46<8:45:52, 2.87s/it] 63%|██████▎ | 18994/30000 [12:02:48<7:12:27, 2.36s/it] 63%|██████▎ | 18995/30000 [12:02:49<6:08:19, 2.01s/it] 63%|██████▎ | 18996/30000 [12:02:50<5:22:29, 1.76s/it] 63%|██████▎ | 18997/30000 [12:02:51<4:52:33, 1.60s/it] 63%|██████▎ | 18998/30000 [12:02:52<4:29:18, 1.47s/it] 63%|██████▎ | 18999/30000 [12:02:54<4:12:59, 1.38s/it] 63%|██████▎ | 19000/30000 [12:02:59<8:01:07, 2.62s/it] 63%|██████▎ | 19000/30000 [12:02:59<8:01:07, 2.62s/it] 63%|██████▎ | 19001/30000 [12:03:03<8:55:04, 2.92s/it] 63%|██████▎ | 19002/30000 [12:03:04<7:20:14, 2.40s/it] 63%|██████▎ | 19003/30000 [12:03:05<6:13:26, 2.04s/it] 63%|██████▎ | 19004/30000 [12:03:06<5:25:36, 1.78s/it] 63%|██████▎ | 19005/30000 [12:03:07<4:52:53, 1.60s/it] 63%|██████▎ | 19006/30000 [12:03:09<4:31:54, 1.48s/it] 63%|██████▎ | 19007/30000 [12:03:10<4:15:12, 1.39s/it] 63%|██████▎ | 19008/30000 [12:03:15<7:48:26, 2.56s/it] 63%|██████▎ | 19009/30000 [12:03:19<9:00:16, 2.95s/it] 63%|██████▎ | 19010/30000 [12:03:20<7:21:37, 2.41s/it] +{'loss': 0.013, 'grad_norm': 0.09008701890707016, 'learning_rate': 1.035059447856611e-05, 'flow_matching_loss': 0.01303564989939332, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.41} +{'loss': 0.0137, 'grad_norm': 0.06842056661844254, 'learning_rate': 1.0334876950819777e-05, 'flow_matching_loss': 0.013703648746013642, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.42} +{'loss': 0.0142, 'grad_norm': 0.08440735936164856, 'learning_rate': 1.0319165091637837e-05, 'flow_matching_loss': 0.014206151198595762, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.42} +{'loss': 0.0142, 'grad_norm': 0.06344017386436462, 'learning_rate': 1.0303458920111686e-05, 'flow_matching_loss': 0.014173902198672294, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.42} +{'loss': 0.015, 'grad_norm': 0.06240127235651016, 'learning_rate': 1.0287758455325803e-05, 'flow_matching_loss': 0.014974792581051588, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.43} +{'loss': 0.014, 'grad_norm': 0.06940951943397522, 'learning_rate': 1.0272063716357729e-05, 'flow_matching_loss': 0.014034784585237502, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.43} +{'loss': 0.0131, 'grad_norm': 0.060240402817726135, 'learning_rate': 1.0256374722278057e-05, 'flow_matching_loss': 0.013105412293225528, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.43} +{'loss': 0.0144, 'grad_norm': 0.06963799148797989, 'learning_rate': 1.0240691492150388e-05, 'flow_matching_loss': 0.0144074821844697, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.43} +{'loss': 0.0129, 'grad_norm': 0.07454155385494232, 'learning_rate': 1.0225014045031334e-05, 'flow_matching_loss': 0.012875640019774436, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.44} +{'loss': 0.0132, 'grad_norm': 0.07150642573833466, 'learning_rate': 1.0209342399970468e-05, 'flow_matching_loss': 0.013245443347841501, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.44} +{'loss': 0.0135, 'grad_norm': 0.0615660659968853, 'learning_rate': 1.019367657601031e-05, 'flow_matching_loss': 0.013530796952545643, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.44} +{'loss': 0.0137, 'grad_norm': 0.0727241039276123, 'learning_rate': 1.0178016592186323e-05, 'flow_matching_loss': 0.013718448393046857, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.45} +{'loss': 0.0161, 'grad_norm': 0.08528797328472137, 'learning_rate': 1.0162362467526861e-05, 'flow_matching_loss': 0.016086266189813615, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.45} +{'loss': 0.0144, 'grad_norm': 0.05863480642437935, 'learning_rate': 1.0146714221053159e-05, 'flow_matching_loss': 0.014361578039824963, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.45} +{'loss': 0.0148, 'grad_norm': 0.10753107070922852, 'learning_rate': 1.0131071871779314e-05, 'flow_matching_loss': 0.014821816515177489, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.46} +{'loss': 0.0158, 'grad_norm': 0.06376317143440247, 'learning_rate': 1.0115435438712249e-05, 'flow_matching_loss': 0.01580682899802923, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.46} +{'loss': 0.0158, 'grad_norm': 0.10683897882699966, 'learning_rate': 1.0099804940851716e-05, 'flow_matching_loss': 0.015841268002986908, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.46} +{'loss': 0.0134, 'grad_norm': 0.07325812429189682, 'learning_rate': 1.0084180397190238e-05, 'flow_matching_loss': 0.013363798428326846, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.46} +{'loss': 0.0151, 'grad_norm': 0.07019786536693573, 'learning_rate': 1.006856182671311e-05, 'flow_matching_loss': 0.01513790600001812, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.47} +{'loss': 0.0146, 'grad_norm': 0.09300339221954346, 'learning_rate': 1.0052949248398367e-05, 'flow_matching_loss': 0.014613651018589736, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.47} +{'loss': 0.0147, 'grad_norm': 0.07203436642885208, 'learning_rate': 1.003734268121676e-05, 'flow_matching_loss': 0.014719314686954022, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.47} +{'loss': 0.0141, 'grad_norm': 0.08211177587509155, 'learning_rate': 1.0021742144131744e-05, 'flow_matching_loss': 0.014135550428181886, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.48} +{'loss': 0.0143, 'grad_norm': 0.08199828863143921, 'learning_rate': 1.000614765609944e-05, 'flow_matching_loss': 0.014328669290989637, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.48} +{'loss': 0.0131, 'grad_norm': 0.08433878421783447, 'learning_rate': 9.990559236068628e-06, 'flow_matching_loss': 0.013131102174520492, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.48} +{'loss': 0.0135, 'grad_norm': 0.06485473364591599, 'learning_rate': 9.974976902980703e-06, 'flow_matching_loss': 0.013476364128291608, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.48} +{'loss': 0.0149, 'grad_norm': 0.07683218270540237, 'learning_rate': 9.959400675769668e-06, 'flow_matching_loss': 0.014918736554682255, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.49} +{'loss': 0.0147, 'grad_norm': 0.075995534658432, 'learning_rate': 9.943830573362106e-06, 'flow_matching_loss': 0.014722555316984654, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.49} +{'loss': 0.0156, 'grad_norm': 0.08984941989183426, 'learning_rate': 9.928266614677168e-06, 'flow_matching_loss': 0.015584193542599678, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.49} +{'loss': 0.0144, 'grad_norm': 0.08136141300201416, 'learning_rate': 9.912708818626524e-06, 'flow_matching_loss': 0.014371374063193798, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.5} +{'loss': 0.0136, 'grad_norm': 0.09291591495275497, 'learning_rate': 9.897157204114369e-06, 'flow_matching_loss': 0.01358921229839325, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.5} +{'loss': 0.0132, 'grad_norm': 0.06537451595067978, 'learning_rate': 9.881611790037375e-06, 'flow_matching_loss': 0.01319147488102317, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.5} +{'loss': 0.0142, 'grad_norm': 0.07002439349889755, 'learning_rate': 9.866072595284684e-06, 'flow_matching_loss': 0.014242850244045258, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.5} +{'loss': 0.0146, 'grad_norm': 0.0660470575094223, 'learning_rate': 9.850539638737894e-06, 'flow_matching_loss': 0.014595110714435578, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.51} +{'loss': 0.0153, 'grad_norm': 0.08294948935508728, 'learning_rate': 9.835012939271006e-06, 'flow_matching_loss': 0.01525071607902646, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.51} +{'loss': 0.0137, 'grad_norm': 0.08121013641357422, 'learning_rate': 9.819492515750418e-06, 'flow_matching_loss': 0.01372428284958005, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.51} +{'loss': 0.0133, 'grad_norm': 0.06462250649929047, 'learning_rate': 9.803978387034918e-06, 'flow_matching_loss': 0.013317237049341202, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.52} +{'loss': 0.0169, 'grad_norm': 0.07488692551851273, 'learning_rate': 9.788470571975624e-06, 'flow_matching_loss': 0.016907600592821835, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.52} +{'loss': 0.0134, 'grad_norm': 0.06947974860668182, 'learning_rate': 9.772969089416008e-06, 'flow_matching_loss': 0.013380698207765818, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.52} +{'loss': 0.0137, 'grad_norm': 0.06764917075634003, 'learning_rate': 9.757473958191822e-06, 'flow_matching_loss': 0.013738674949854613, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.53} +{'loss': 0.0132, 'grad_norm': 0.06537255644798279, 'learning_rate': 9.741985197131118e-06, 'flow_matching_loss': 0.013209135830402374, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.53} +{'loss': 0.0124, 'grad_norm': 0.06341597437858582, 'learning_rate': 9.726502825054195e-06, 'flow_matching_loss': 0.012434760574251413, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.53} 63%|██████▎ | 19010/30000 [12:03:20<7:21:37, 2.41s/it] 63%|██████▎ | 19011/30000 [12:03:21<6:15:01, 2.05s/it] 63%|██████▎ | 19012/30000 [12:03:22<5:26:47, 1.78s/it] 63%|██████▎ | 19013/30000 [12:03:24<4:54:13, 1.61s/it] 63%|██████▎ | 19014/30000 [12:03:25<4:31:20, 1.48s/it] 63%|██████▎ | 19015/30000 [12:03:26<4:15:08, 1.39s/it] 63%|██████▎ | 19016/30000 [12:03:31<7:56:12, 2.60s/it] 63%|██████▎ | 19017/30000 [12:03:34<8:16:30, 2.71s/it] 63%|██████▎ | 19018/30000 [12:03:36<6:51:19, 2.25s/it] 63%|██████▎ | 19019/30000 [12:03:37<5:53:24, 1.93s/it] 63%|██████▎ | 19020/30000 [12:03:38<5:11:25, 1.70s/it] 63%|██████▎ | 19020/30000 [12:03:38<5:11:25, 1.70s/it] 63%|██████▎ | 19021/30000 [12:03:39<4:43:17, 1.55s/it] 63%|██████▎ | 19022/30000 [12:03:40<4:22:31, 1.43s/it] 63%|██████▎ | 19023/30000 [12:03:41<4:07:05, 1.35s/it] 63%|██████▎ | 19024/30000 [12:03:47<7:44:58, 2.54s/it] 63%|██████▎ | 19025/30000 [12:03:50<8:16:16, 2.71s/it] 63%|██████▎ | 19026/30000 [12:03:51<6:51:29, 2.25s/it] 63%|██████▎ | 19027/30000 [12:03:52<5:52:52, 1.93s/it] 63%|██████▎ | 19028/30000 [12:03:53<5:11:21, 1.70s/it] 63%|██████▎ | 19029/30000 [12:03:55<4:43:09, 1.55s/it] 63%|██████▎ | 19030/30000 [12:03:56<4:23:15, 1.44s/it] 63%|██████▎ | 19030/30000 [12:03:56<4:23:15, 1.44s/it] 63%|██████▎ | 19031/30000 [12:03:57<4:09:11, 1.36s/it] 63%|██████▎ | 19032/30000 [12:04:03<8:10:51, 2.69s/it] 63%|██████▎ | 19033/30000 [12:04:06<8:37:13, 2.83s/it] 63%|██████▎ | 19034/30000 [12:04:07<7:06:31, 2.33s/it] 63%|██████▎ | 19035/30000 [12:04:08<6:05:10, 2.00s/it] 63%|██████▎ | 19036/30000 [12:04:09<5:19:17, 1.75s/it] 63%|██████▎ | 19037/30000 [12:04:11<4:49:05, 1.58s/it] 63%|██████▎ | 19038/30000 [12:04:12<4:27:13, 1.46s/it] 63%|██████▎ | 19039/30000 [12:04:13<4:13:06, 1.39s/it] 63%|██████▎ | 19040/30000 [12:04:18<7:31:51, 2.47s/it] 63%|██████▎ | 19040/30000 [12:04:18<7:31:51, 2.47s/it] 63%|██████▎ | 19041/30000 [12:04:21<7:55:14, 2.60s/it] 63%|██████▎ | 19042/30000 [12:04:22<6:37:05, 2.17s/it] 63%|██████▎ | 19043/30000 [12:04:23<5:45:39, 1.89s/it] 63%|██████▎ | 19044/30000 [12:04:25<5:05:54, 1.68s/it] 63%|██████▎ | 19045/30000 [12:04:26<4:39:17, 1.53s/it] 63%|██████▎ | 19046/30000 [12:04:27<4:19:42, 1.42s/it] 63%|██████▎ | 19047/30000 [12:04:28<4:07:18, 1.35s/it] 63%|██████▎ | 19048/30000 [12:04:33<7:36:41, 2.50s/it] 63%|██████▎ | 19049/30000 [12:04:36<8:01:48, 2.64s/it] 64%|██████▎ | 19050/30000 [12:04:37<6:43:47, 2.21s/it] 64%|██████▎ | 19050/30000 [12:04:38<6:43:47, 2.21s/it] 64%|██████▎ | 19051/30000 [12:04:39<5:49:43, 1.92s/it] 64%|██████▎ | 19052/30000 [12:04:40<5:09:06, 1.69s/it] 64%|██████▎ | 19053/30000 [12:04:41<4:40:58, 1.54s/it] 64%|██████▎ | 19054/30000 [12:04:42<4:21:26, 1.43s/it] 64%|██████▎ | 19055/30000 [12:04:43<4:07:54, 1.36s/it] 64%|██████▎ | 19056/30000 [12:04:49<8:11:03, 2.69s/it] 64%|██████▎ | 19057/30000 [12:04:52<8:32:02, 2.81s/it] 64%|██████▎ | 19058/30000 [12:04:53<7:02:34, 2.32s/it] 64%|██████▎ | 19059/30000 [12:04:55<6:00:23, 1.98s/it] 64%|██████▎ | 19060/30000 [12:04:56<5:17:02, 1.74s/it] 64%|██████▎ | 19060/30000 [12:04:56<5:17:02, 1.74s/it] 64%|██████▎ | 19061/30000 [12:04:57<4:47:01, 1.57s/it] 64%|██████▎ | 19062/30000 [12:04:58<4:25:36, 1.46s/it] 64%|██████▎ | 19063/30000 [12:04:59<4:10:34, 1.37s/it] 64%|██████▎ | 19064/30000 [12:05:05<8:16:53, 2.73s/it] 64%|██████▎ | 19065/30000 [12:05:09<9:00:45, 2.97s/it] 64%|██████▎ | 19066/30000 [12:05:10<7:22:27, 2.43s/it] 64%|██████▎ | 19067/30000 [12:05:11<6:14:42, 2.06s/it] 64%|██████▎ | 19068/30000 [12:05:12<5:26:28, 1.79s/it] 64%|██████▎ | 19069/30000 [12:05:14<4:54:16, 1.62s/it] 64%|██████▎ | 19070/30000 [12:05:15<4:31:37, 1.49s/it] 64%|██████▎ | 19070/30000 [12:05:15<4:31:37, 1.49s/it] 64%|██████▎ | 19071/30000 [12:05:16<4:15:20, 1.40s/it] 64%|██████▎ | 19072/30000 [12:05:21<7:42:32, 2.54s/it] 64%|██████▎ | 19073/30000 [12:05:25<8:39:11, 2.85s/it] 64%|██████▎ | 19074/30000 [12:05:26<7:07:50, 2.35s/it] 64%|██████▎ | 19075/30000 [12:05:27<6:05:21, 2.01s/it] 64%|██████▎ | 19076/30000 [12:05:28<5:19:06, 1.75s/it] 64%|██████▎ | 19077/30000 [12:05:29<4:48:38, 1.59s/it] 64%|██████▎ | 19078/30000 [12:05:31<4:26:24, 1.46s/it] 64%|██████▎ | 19079/30000 [12:05:32<4:13:31, 1.39s/it] 64%|██████▎ | 19080/30000 [12:05:37<7:36:22, 2.51s/it] 64%|██████▎ | 19080/30000 [12:05:37<7:36:22, 2.51s/it] 64%|██████▎ | 19081/30000 [12:05:42<9:32:49, 3.15s/it] 64%|██████▎ | 19082/30000 [12:05:43<7:44:45, 2.55s/it] 64%|██████▎ | 19083/30000 [12:05:44<6:29:53, 2.14s/it] 64%|██████▎ | 19084/30000 [12:05:45<5:36:36, 1.85s/it] 64%|██████▎ | 19085/30000 [12:05:46<5:00:40, 1.65s/it] 64%|██████▎ | 19086/30000 [12:05:48<4:35:40, 1.52s/it] 64%|██████▎ | 19087/30000 [12:05:49<4:17:51, 1.42s/it] 64%|██████▎ | 19088/30000 [12:05:54<7:59:59, 2.64s/it] 64%|██████▎ | 19089/30000 [12:05:59<10:19:18, 3.41s/it] 64%|██████▎ | 19090/30000 [12:06:01<8:17:29, 2.74s/it] 64%|██████▎ | 19090/30000 [12:06:01<8:17:29, 2.74s/it] 64%|██████▎ | 19091/30000 [12:06:02<6:53:28, 2.27s/it] 64%|██████▎ | 19092/30000 [12:06:03<5:53:39, 1.95s/it] 64%|██████▎ | 19093/30000 [12:06:04<5:12:36, 1.72s/it] 64%|██████▎ | 19094/30000 [12:06:05<4:44:39, 1.57s/it] 64%|██████▎ | 19095/30000 [12:06:07<4:23:28, 1.45s/it] 64%|██████▎ | 19096/30000 [12:06:11<6:49:27, 2.25s/it] 64%|██████▎ | 19097/30000 [12:06:16<9:25:33, 3.11s/it] 64%|██████▎ | 19098/30000 [12:06:17<7:40:10, 2.53s/it] 64%|██████▎ | 19099/30000 [12:06:18<6:27:09, 2.13s/it] 64%|██████▎ | 19100/30000 [12:06:19<5:35:15, 1.85s/it] 64%|██████▎ | 19100/30000 [12:06:19<5:35:15, 1.85s/it] 64%|██████▎ | 19101/30000 [12:06:21<5:00:32, 1.65s/it] 64%|██████▎ | 19102/30000 [12:06:22<4:35:33, 1.52s/it] 64%|██████▎ | 19103/30000 [12:06:23<4:16:48, 1.41s/it] 64%|██████▎ | 19104/30000 [12:06:26<6:08:52, 2.03s/it] 64%|██████▎ | 19105/30000 [12:06:32<9:02:19, 2.99s/it] 64%|██████▎ | 19106/30000 [12:06:33<7:23:09, 2.44s/it] 64%|██████▎ | 19107/30000 [12:06:34<6:14:22, 2.06s/it] 64%|██████▎ | 19108/30000 [12:06:35<5:25:04, 1.79s/it] 64%|██████▎ | 19109/30000 [12:06:36<4:51:42, 1.61s/it] 64%|██████▎ | 19110/30000 [12:06:37<4:28:43, 1.48s/it] 64%|██████▎ | 19110/30000 [12:06:37<4:28:43, 1.48s/it] 64%|██████▎ | 19111/30000 [12:06:39<4:13:02, 1.39s/it] 64%|██████▎ | 19112/30000 [12:06:42<5:33:25, 1.84s/it] 64%|██████▎ | 19113/30000 [12:06:48<9:41:18, 3.20s/it] 64%|██████▎ | 19114/30000 [12:06:49<7:51:07, 2.60s/it] 64%|██████▎ | 19115/30000 [12:06:50<6:34:47, 2.18s/it] 64%|██████▎ | 19116/30000 [12:06:51<5:40:11, 1.88s/it] 64%|██████▎ | 19117/30000 [12:06:53<5:02:55, 1.67s/it] 64%|██████▎ | 19118/30000 [12:06:54<4:37:00, 1.53s/it] 64%|██████▎ | 19119/30000 [12:06:55<4:23:18, 1.45s/it] 64%|██████▎ | 19120/30000 [12:06:57<5:03:06, 1.67s/it] 64%|██████▎ | 19120/30000 [12:06:57<5:03:06, 1.67s/it] 64%|██████▎ | 19121/30000 [12:07:05<10:03:34, 3.33s/it] 64%|██████▎ | 19122/30000 [12:07:06<8:06:16, 2.68s/it] 64%|██████▎ | 19123/30000 [12:07:07<6:45:19, 2.24s/it] 64%|██████▎ | 19124/30000 [12:07:08<5:47:25, 1.92s/it] 64%|██████▍ | 19125/30000 [12:07:09<5:08:23, 1.70s/it] 64%|██████▍ | 19126/30000 [12:07:10<4:41:06, 1.55s/it] 64%|██████▍ | 19127/30000 [12:07:12<4:21:13, 1.44s/it] 64%|██████▍ | 19128/30000 [12:07:13<4:08:23, 1.37s/it] 64%|██████▍ | 19129/30000 [12:07:20<9:07:31, 3.02s/it] 64%|██████▍ | 19130/30000 [12:07:21<7:26:04, 2.46s/it] 64%|██████▍ | 19130/30000 [12:07:21<7:26:04, 2.46s/it] 64%|██████▍ | 19131/30000 [12:07:22<6:17:10, 2.08s/it] 64%|██████▍ | 19132/30000 [12:07:23<5:28:06, 1.81s/it] 64%|██████▍ | 19133/30000 [12:07:24<4:53:55, 1.62s/it] 64%|██████▍ | 19134/30000 [12:07:26<4:31:00, 1.50s/it] 64%|██████▍ | 19135/30000 [12:07:27<4:14:15, 1.40s/it] 64%|██████▍ | 19136/30000 [12:07:29<4:33:07, 1.51s/it] 64%|██████▍ | 19137/30000 [12:07:36<9:42:58, 3.22s/it] 64%|██████▍ | 19138/30000 [12:07:37<7:51:17, 2.60s/it] 64%|██████▍ | 19139/30000 [12:07:38<6:34:41, 2.18s/it] 64%|██████▍ | 19140/30000 [12:07:39<5:40:03, 1.88s/it] 64%|██████▍ | 19140/30000 [12:07:39<5:40:03, 1.88s/it] 64%|██████▍ | 19141/30000 [12:07:41<5:03:26, 1.68s/it] 64%|██████▍ | 19142/30000 [12:07:42<4:37:04, 1.53s/it] 64%|██████▍ | 19143/30000 [12:07:43<4:19:49, 1.44s/it] 64%|██████▍ | 19144/30000 [12:07:45<5:12:13, 1.73s/it] 64%|██████▍ | 19145/30000 [12:07:53<10:16:24, 3.41s/it] 64%|██████▍ | 19146/30000 [12:07:54<8:15:22, 2.74s/it] 64%|██████▍ | 19147/30000 [12:07:55<6:51:02, 2.27s/it] 64%|██████▍ | 19148/30000 [12:07:56<5:51:43, 1.94s/it] 64%|██████▍ | 19149/30000 [12:07:57<5:09:39, 1.71s/it] 64%|██████▍ | 19150/30000 [12:07:59<4:40:36, 1.55s/it] 64%|██████▍ | 19150/30000 [12:07:59<4:40:36, 1.55s/it] 64%|██████▍ | 19151/30000 [12:08:00<4:21:01, 1.44s/it] 64%|██████▍ | 19152/30000 [12:08:01<4:20:42, 1.44s/it] 64%|██████▍ | 19153/30000 [12:08:08<9:34:12, 3.18s/it] 64%|██████▍ | 19154/30000 [12:08:10<7:44:51, 2.57s/it] 64%|██████▍ | 19155/30000 [12:08:11<6:28:35, 2.15s/it] 64%|██████▍ | 19156/30000 [12:08:12<5:35:22, 1.86s/it] 64%|██████▍ | 19157/30000 [12:08:13<4:57:18, 1.65s/it] 64%|██████▍ | 19158/30000 [12:08:14<4:31:56, 1.50s/it] 64%|██████▍ | 19159/30000 [12:08:15<4:15:05, 1.41s/it] 64%|██████▍ | 19160/30000 [12:08:17<4:07:22, 1.37s/it] 64%|██████▍ | 19160/30000 [12:08:17<4:07:22, 1.37s/it] 64%|██████▍ | 19161/30000 [12:08:24<9:46:42, 3.25s/it] 64%|██████▍ | 19162/30000 [12:08:26<7:54:15, 2.63s/it] 64%|██████▍ | 19163/30000 [12:08:27<6:35:14, 2.19s/it] 64%|██████▍ | 19164/30000 [12:08:28<5:39:57, 1.88s/it] 64%|██████▍ | 19165/30000 [12:08:29<5:02:37, 1.68s/it] 64%|██████▍ | 19166/30000 [12:08:30<4:38:53, 1.54s/it] 64%|██████▍ | 19167/30000 [12:08:31<4:19:34, 1.44s/it] 64%|██████▍ | 19168/30000 [12:08:33<4:11:55, 1.40s/it] 64%|██████▍ | 19169/30000 [12:08:41<10:21:02, 3.44s/it] 64%|██████▍ | 19170/30000 [12:08:42<8:18:09, 2.76s/it] 64%|██████▍ | 19170/30000 [12:08:42<8:18:09, 2.76s/it] 64%|██████▍ | 19171/30000 [12:08:43<6:52:47, 2.29s/it] 64%|██████▍ | 19172/30000 [12:08:45<5:52:40, 1.95s/it] 64%|██████▍ | 19173/30000 [12:08:46<5:12:41, 1.73s/it] 64%|██████▍ | 19174/30000 [12:08:47<4:45:07, 1.58s/it] 64%|██████▍ | 19175/30000 [12:08:48<4:24:41, 1.47s/it] 64%|██████▍ | 19176/30000 [12:08:50<4:20:36, 1.44s/it] 64%|██████▍ | 19177/30000 [12:08:57<9:54:35, 3.30s/it] 64%|██████▍ | 19178/30000 [12:08:58<7:58:52, 2.66s/it] 64%|██████▍ | 19179/30000 [12:08:59<6:39:14, 2.21s/it] 64%|██████▍ | 19180/30000 [12:09:01<5:42:15, 1.90s/it] 64%|██████▍ | 19180/30000 [12:09:01<5:42:15, 1.90s/it] 64%|██████▍ | 19181/30000 [12:09:02<5:03:38, 1.68s/it] 64%|██████▍ | 19182/30000 [12:09:03<4:37:06, 1.54s/it] 64%|██████▍ | 19183/30000 [12:09:04<4:18:15, 1.43s/it] 64%|██████▍ | 19184/30000 [12:09:06<4:45:26, 1.58s/it] 64%|██████▍ | 19185/30000 [12:09:13<9:02:37, 3.01s/it] 64%|██████▍ | 19186/30000 [12:09:14<7:22:44, 2.46s/it] 64%|██████▍ | 19187/30000 [12:09:15<6:14:42, 2.08s/it] 64%|██████▍ | 19188/30000 [12:09:16<5:25:41, 1.81s/it] 64%|██████▍ | 19189/30000 [12:09:17<4:52:06, 1.62s/it] 64%|██████▍ | 19190/30000 [12:09:18<4:28:30, 1.49s/it] 64%|██████▍ | 19190/30000 [12:09:18<4:28:30, 1.49s/it] 64%|██████▍ | 19191/30000 [12:09:20<4:10:52, 1.39s/it] 64%|██████▍ | 19192/30000 [12:09:23<5:34:46, 1.86s/it] 64%|██████▍ | 19193/30000 [12:09:28<8:57:50, 2.99s/it] 64%|██████▍ | 19194/30000 [12:09:29<7:19:49, 2.44s/it] 64%|██████▍ | 19195/30000 [12:09:30<6:10:54, 2.06s/it] 64%|██████▍ | 19196/30000 [12:09:32<5:23:07, 1.79s/it] 64%|██████▍ | 19197/30000 [12:09:33<4:52:16, 1.62s/it] 64%|██████▍ | 19198/30000 [12:09:34<4:28:14, 1.49s/it] 64%|██████▍ | 19199/30000 [12:09:35<4:11:05, 1.39s/it] 64%|██████▍ | 19200/30000 [12:09:38<5:27:15, 1.82s/it] 64%|██████▍ | 19200/30000 [12:09:38<5:27:15, 1.82s/it] 64%|██████▍ | 19201/30000 [12:09:44<9:08:46, 3.05s/it] 64%|██████▍ | 19202/30000 [12:09:45<7:27:22, 2.49s/it] 64%|██████▍ | 19203/30000 [12:09:46<6:17:10, 2.10s/it] 64%|██████▍ | 19204/30000 [12:09:47<5:27:31, 1.82s/it] 64%|██████▍ | 19205/30000 [12:09:49<4:53:59, 1.63s/it] 64%|██████▍ | 19206/30000 [12:09:50<4:29:50, 1.50s/it] 64%|██████▍ | 19207/30000 [12:09:51<4:12:39, 1.40s/it] 64%|██████▍ | 19208/30000 [12:09:53<5:00:33, 1.67s/it] 64%|██████▍ | 19209/30000 [12:09:59<8:44:39, 2.92s/it] 64%|██████▍ | 19210/30000 [12:10:00<7:09:01, 2.39s/it] 64%|██████▍ | 19210/30000 [12:10:00<7:09:01, 2.39s/it] 64%|██████▍ | 19211/30000 [12:10:02<6:04:16, 2.03s/it] 64%|██████▍ | 19212/30000 [12:10:03<5:18:02, 1.77s/it] 64%|██████▍ | 19213/30000 [12:10:04<4:46:23, 1.59s/it] 64%|██████▍ | 19214/30000 [12:10:05<4:27:09, 1.49s/it] 64%|██████▍ | 19215/30000 [12:10:06<4:10:26, 1.39s/it] 64%|██████▍ | 19216/30000 [12:10:09<5:26:26, 1.82s/it] 64%|██████▍ | 19217/30000 [12:10:14<8:37:04, 2.88s/it] 64%|██████▍ | 19218/30000 [12:10:16<7:04:32, 2.36s/it] 64%|██████▍ | 19219/30000 [12:10:17<6:00:43, 2.01s/it] 64%|██████▍ | 19220/30000 [12:10:18<5:15:39, 1.76s/it] 64%|██████▍ | 19220/30000 [12:10:18<5:15:39, 1.76s/it] 64%|██████▍ | 19221/30000 [12:10:19<4:44:17, 1.58s/it] 64%|██████▍ | 19222/30000 [12:10:20<4:21:22, 1.46s/it] 64%|██████▍ | 19223/30000 [12:10:21<4:06:31, 1.37s/it] 64%|██████▍ | 19224/30000 [12:10:25<5:42:28, 1.91s/it] 64%|██████▍ | 19225/30000 [12:10:30<8:47:17, 2.94s/it] 64%|██████▍ | 19226/30000 [12:10:31<7:12:14, 2.41s/it] 64%|██████▍ | 19227/30000 [12:10:32<6:05:37, 2.04s/it] 64%|██████▍ | 19228/30000 [12:10:33<5:18:47, 1.78s/it] 64%|██████▍ | 19229/30000 [12:10:35<4:46:48, 1.60s/it] 64%|██████▍ | 19230/30000 [12:10:36<4:25:22, 1.48s/it] 64%|██████▍ | 19230/30000 [12:10:36<4:25:22, 1.48s/it] 64%|██████▍ | 19231/30000 [12:10:37<4:09:23, 1.39s/it] 64%|██████▍ | 19232/30000 [12:10:40<5:18:21, 1.77s/it] 64%|██████▍ | 19233/30000 [12:10:45<8:25:59, 2.82s/it] 64%|██████▍ | 19234/30000 [12:10:46<6:56:45, 2.32s/it] 64%|██████▍ | 19235/30000 [12:10:47<5:54:25, 1.98s/it] 64%|██████▍ | 19236/30000 [12:10:48<5:10:54, 1.73s/it] 64%|██████▍ | 19237/30000 [12:10:50<4:40:24, 1.56s/it] 64%|██████▍ | 19238/30000 [12:10:51<4:19:37, 1.45s/it] 64%|██████▍ | 19239/30000 [12:10:52<4:07:56, 1.38s/it] 64%|██████▍ | 19240/30000 [12:10:54<4:24:22, 1.47s/it] 64%|██████▍ | 19240/30000 [12:10:54<4:24:22, 1.47s/it] 64%|██████▍ | 19241/30000 [12:10:59<7:58:53, 2.67s/it] 64%|██████▍ | 19242/30000 [12:11:00<6:38:39, 2.22s/it] 64%|██████▍ | 19243/30000 [12:11:02<5:42:23, 1.91s/it] 64%|██████▍ | 19244/30000 [12:11:03<5:02:48, 1.69s/it] 64%|██████▍ | 19245/30000 [12:11:04<4:36:05, 1.54s/it] 64%|██████▍ | 19246/30000 [12:11:05<4:17:18, 1.44s/it] 64%|██████▍ | 19247/30000 [12:11:06<4:04:36, 1.36s/it] 64%|██████▍ | 19248/30000 [12:11:09<4:58:16, 1.66s/it] 64%|██████▍ | 19249/30000 [12:11:14<8:21:21, 2.80s/it] 64%|██████▍ | 19250/30000 [12:11:15<6:53:33, 2.31s/it] 64%|██████▍ | 19250/30000 [12:11:15<6:53:33, 2.31s/it] 64%|██████▍ | 19251/30000 [12:11:16<5:53:33, 1.97s/it] 64%|██████▍ | 19252/30000 [12:11:18<5:09:58, 1.73s/it] 64%|██████▍ | 19253/30000 [12:11:19<4:42:31, 1.58s/it] 64%|██████▍ | 19254/30000 [12:11:20<4:22:48, 1.47s/it] 64%|██████▍ | 19255/30000 [12:11:21<4:08:35, 1.39s/it] 64%|██████▍ | 19256/30000 [12:11:24<5:21:01, 1.79s/it] 64%|██████▍ | 19257/30000 [12:11:29<8:25:18, 2.82s/it] 64%|██████▍ | 19258/30000 [12:11:30<6:55:57, 2.32s/it] 64%|██████▍ | 19259/30000 [12:11:32<5:54:18, 1.98s/it] 64%|██████▍ | 19260/30000 [12:11:33<5:10:44, 1.74s/it] 64%|██████▍ | 19260/30000 [12:11:33<5:10:44, 1.74s/it] 64%|██████▍ | 19261/30000 [12:11:34<4:40:56, 1.57s/it] 64%|██████▍ | 19262/30000 [12:11:35<4:20:39, 1.46s/it] 64%|██████▍ | 19263/30000 [12:11:36<4:06:14, 1.38s/it] 64%|██████▍ | 19264/30000 [12:11:39<5:33:10, 1.86s/it] 64%|██████▍ | 19265/30000 [12:11:44<8:03:31, 2.70s/it] 64%|██████▍ | 19266/30000 [12:11:45<6:41:21, 2.24s/it] 64%|██████▍ | 19267/30000 [12:11:46<5:44:40, 1.93s/it] 64%|██████▍ | 19268/30000 [12:11:47<5:04:10, 1.70s/it] 64%|██████▍ | 19269/30000 [12:11:49<4:36:08, 1.54s/it] 64%|██████▍ | 19270/30000 [12:11:50<4:17:16, 1.44s/it] 64%|██████▍ | 19270/30000 [12:11:50<4:17:16, 1.44s/it] 64%|██████▍ | 19271/30000 [12:11:51<4:04:33, 1.37s/it] 64%|██████▍ | 19272/30000 [12:11:55<6:01:03, 2.02s/it] 64%|██████▍ | 19273/30000 [12:11:59<8:07:02, 2.72s/it] 64%|██████▍ | 19274/30000 [12:12:00<6:43:25, 2.26s/it] 64%|██████▍ | 19275/30000 [12:12:01<5:45:39, 1.93s/it] 64%|██████▍ | 19276/30000 [12:12:02<5:04:41, 1.70s/it] 64%|██████▍ | 19277/30000 [12:12:04<4:36:14, 1.55s/it] 64%|██████▍ | 19278/30000 [12:12:05<4:17:44, 1.44s/it] 64%|██████▍ | 19279/30000 [12:12:06<4:03:54, 1.37s/it] 64%|██████▍ | 19280/30000 [12:12:10<6:25:36, 2.16s/it] 64%|██████▍ | 19280/30000 [12:12:10<6:25:36, 2.16s/it] 64%|██████▍ | 19281/30000 [12:12:14<7:51:14, 2.64s/it] 64%|██████▍ | 19282/30000 [12:12:15<6:32:57, 2.20s/it] 64%|██████▍ | 19283/30000 [12:12:16<5:38:29, 1.90s/it] 64%|██████▍ | 19284/30000 [12:12:17<4:59:50, 1.68s/it] 64%|██████▍ | 19285/30000 [12:12:19<4:33:01, 1.53s/it] 64%|██████▍ | 19286/30000 [12:12:20<4:14:43, 1.43s/it] 64%|██████▍ | 19287/30000 [12:12:21<4:02:04, 1.36s/it] 64%|██████▍ | 19288/30000 [12:12:25<6:11:18, 2.08s/it] 64%|██████▍ | 19289/30000 [12:12:29<7:47:01, 2.62s/it] 64%|██████▍ | 19290/30000 [12:12:30<6:30:09, 2.19s/it] 64%|██████▍ | 19290/30000 [12:12:30<6:30:09, 2.19s/it] 64%|██████▍ | 19291/30000 [12:12:31<5:36:37, 1.89s/it] 64%|██████▍ | 19292/30000 [12:12:32<4:58:19, 1.67s/it] 64%|██████▍ | 19293/30000 [12:12:33<4:34:16, 1.54s/it] 64%|██████▍ | 19294/30000 [12:12:34<4:14:30, 1.43s/it] 64%|██████▍ | 19295/30000 [12:12:36<4:02:17, 1.36s/it] 64%|██████▍ | 19296/30000 [12:12:39<5:37:22, 1.89s/it] 64%|██████▍ | 19297/30000 [12:12:43<7:53:40, 2.66s/it] 64%|██████▍ | 19298/30000 [12:12:44<6:34:23, 2.21s/it] 64%|██████▍ | 19299/30000 [12:12:46<5:39:55, 1.91s/it] 64%|██████▍ | 19300/30000 [12:12:47<5:00:48, 1.69s/it] 64%|██████▍ | 19300/30000 [12:12:47<5:00:48, 1.69s/it] 64%|██████▍ | 19301/30000 [12:12:48<4:33:55, 1.54s/it] 64%|██████▍ | 19302/30000 [12:12:49<4:15:42, 1.43s/it] 64%|██████▍ | 19303/30000 [12:12:50<4:03:01, 1.36s/it] 64%|██████▍ | 19304/30000 [12:12:53<5:18:06, 1.78s/it] 64%|██████▍ | 19305/30000 [12:12:57<7:34:31, 2.55s/it] 64%|██████▍ | 19306/30000 [12:12:59<6:20:42, 2.14s/it] 64%|██████▍ | 19307/30000 [12:13:00<5:29:22, 1.85s/it] 64%|██████▍ | 19308/30000 [12:13:01<4:53:45, 1.65s/it] 64%|██████▍ | 19309/30000 [12:13:02<4:29:24, 1.51s/it] 64%|██████▍ | 19310/30000 [12:13:03<4:12:14, 1.42s/it] 64%|██████▍ | 19310/30000 [12:13:03<4:12:14, 1.42s/it] 64%|██████▍ | 19311/30000 [12:13:05<4:00:59, 1.35s/it] 64%|██████▍ | 19312/30000 [12:13:08<5:57:52, 2.01s/it] 64%|██████▍ | 19313/30000 [12:13:13<8:29:36, 2.86s/it] 64%|██████▍ | 19314/30000 [12:13:14<6:58:59, 2.35s/it] 64%|██████▍ | 19315/30000 [12:13:15<5:56:42, 2.00s/it] 64%|██████▍ | 19316/30000 [12:13:16<5:11:46, 1.75s/it] 64%|██████▍ | 19317/30000 [12:13:18<4:40:52, 1.58s/it] 64%|██████▍ | 19318/30000 [12:13:19<4:20:05, 1.46s/it] 64%|██████▍ | 19319/30000 [12:13:20<4:05:28, 1.38s/it] 64%|██████▍ | 19320/30000 [12:13:23<5:26:35, 1.83s/it] 64%|██████▍ | 19320/30000 [12:13:23<5:26:35, 1.83s/it] 64%|██████▍ | 19321/30000 [12:13:28<8:27:47, 2.85s/it] 64%|██████▍ | 19322/30000 [12:13:29<6:57:09, 2.34s/it] 64%|██████▍ | 19323/30000 [12:13:31<5:55:38, 2.00s/it] 64%|██████▍ | 19324/30000 [12:13:32<5:12:12, 1.75s/it] 64%|██████▍ | 19325/30000 [12:13:33<4:40:49, 1.58s/it] 64%|██████▍ | 19326/30000 [12:13:34<4:19:52, 1.46s/it] 64%|██████▍ | 19327/30000 [12:13:35<4:05:39, 1.38s/it] 64%|██████▍ | 19328/30000 [12:13:39<6:24:32, 2.16s/it] 64%|██████▍ | 19329/30000 [12:13:44<8:53:17, 3.00s/it] 64%|██████▍ | 19330/30000 [12:13:45<7:15:10, 2.45s/it] 64%|██████▍ | 19330/30000 [12:13:45<7:15:10, 2.45s/it] 64%|██████▍ | 19331/30000 [12:13:47<6:08:13, 2.07s/it] 64%|██████▍ | 19332/30000 [12:13:48<5:20:04, 1.80s/it] 64%|██████▍ | 19333/30000 [12:13:49<4:46:12, 1.61s/it] 64%|██████▍ | 19334/30000 [12:13:50<4:23:01, 1.48s/it] 64%|██████▍ | 19335/30000 [12:13:51<4:07:20, 1.39s/it] 64%|██████▍ | 19336/30000 [12:13:54<4:55:04, 1.66s/it] 64%|██████▍ | 19337/30000 [12:13:59<8:06:07, 2.74s/it] 64%|██████▍ | 19338/30000 [12:14:00<6:42:58, 2.27s/it] 64%|██████▍ | 19339/30000 [12:14:01<5:45:44, 1.95s/it] 64%|██████▍ | 19340/30000 [12:14:02<5:04:38, 1.71s/it] 64%|██████▍ | 19340/30000 [12:14:02<5:04:38, 1.71s/it] 64%|██████▍ | 19341/30000 [12:14:03<4:36:46, 1.56s/it] 64%|██████▍ | 19342/30000 [12:14:05<4:16:58, 1.45s/it] 64%|██████▍ | 19343/30000 [12:14:06<4:04:35, 1.38s/it] 64%|██████▍ | 19344/30000 [12:14:08<4:49:38, 1.63s/it] 64%|██████▍ | 19345/30000 [12:14:13<8:00:19, 2.70s/it] 64%|██████▍ | 19346/30000 [12:14:15<6:38:32, 2.24s/it] 64%|██████▍ | 19347/30000 [12:14:16<5:41:47, 1.93s/it] 64%|██████▍ | 19348/30000 [12:14:17<5:02:03, 1.70s/it] 64%|██████▍ | 19349/30000 [12:14:18<4:34:08, 1.54s/it] 64%|██████▍ | 19350/30000 [12:14:19<4:13:51, 1.43s/it] 64%|██████▍ | 19350/30000 [12:14:19<4:13:51, 1.43s/it] 65%|██████▍ | 19351/30000 [12:14:20<4:00:34, 1.36s/it] 65%|██████▍ | 19352/30000 [12:14:23<4:42:42, 1.59s/it] 65%|██████▍ | 19353/30000 [12:14:28<8:28:12, 2.86s/it] 65%|██████▍ | 19354/30000 [12:14:30<6:58:22, 2.36s/it] 65%|██████▍ | 19355/30000 [12:14:31<5:56:05, 2.01s/it] 65%|██████▍ | 19356/30000 [12:14:32<5:11:37, 1.76s/it] 65%|██████▍ | 19357/30000 [12:14:33<4:41:39, 1.59s/it] 65%|██████▍ | 19358/30000 [12:14:34<4:20:15, 1.47s/it] 65%|██████▍ | 19359/30000 [12:14:35<4:04:54, 1.38s/it] 65%|██████▍ | 19360/30000 [12:14:37<4:38:33, 1.57s/it] 65%|██████▍ | 19360/30000 [12:14:37<4:38:33, 1.57s/it] 65%|██████▍ | 19361/30000 [12:14:44<8:37:39, 2.92s/it] 65%|██████▍ | 19362/30000 [12:14:45<7:03:52, 2.39s/it] 65%|██████▍ | 19363/30000 [12:14:46<5:59:24, 2.03s/it] 65%|██████▍ | 19364/30000 [12:14:47<5:14:25, 1.77s/it] 65%|██████▍ | 19365/30000 [12:14:48<4:42:58, 1.60s/it] 65%|██████▍ | 19366/30000 [12:14:49<4:21:23, 1.47s/it] 65%|██████▍ | 19367/30000 [12:14:51<4:08:17, 1.40s/it] 65%|██████▍ | 19368/30000 [12:14:53<4:42:01, 1.59s/it] 65%|██████▍ | 19369/30000 [12:14:58<8:06:50, 2.75s/it] 65%|██████▍ | 19370/30000 [12:14:59<6:43:03, 2.28s/it] 65%|██████▍ | 19370/30000 [12:14:59<6:43:03, 2.28s/it] 65%|██████▍ | 19371/30000 [12:15:01<5:45:17, 1.95s/it] 65%|██████▍ | 19372/30000 [12:15:02<5:04:00, 1.72s/it] 65%|██████▍ | 19373/30000 [12:15:03<4:36:02, 1.56s/it] 65%|██████▍ | 19374/30000 [12:15:04<4:16:41, 1.45s/it] 65%|██████▍ | 19375/30000 [12:15:05<4:03:36, 1.38s/it] 65%|██████▍ | 19376/30000 [12:15:08<5:06:57, 1.73s/it] 65%|██████▍ | 19377/30000 [12:15:13<8:34:39, 2.91s/it] 65%|██████▍ | 19378/30000 [12:15:15<7:01:39, 2.38s/it] 65%|██████▍ | 19379/30000 [12:15:16<5:57:07, 2.02s/it] 65%|██████▍ | 19380/30000 [12:15:17<5:11:56, 1.76s/it] 65%|██████▍ | 19380/30000 [12:15:17<5:11:56, 1.76s/it] 65%|██████▍ | 19381/30000 [12:15:18<4:39:45, 1.58s/it] 65%|██████▍ | 19382/30000 [12:15:19<4:17:44, 1.46s/it] 65%|██████▍ | 19383/30000 [12:15:20<4:03:55, 1.38s/it] 65%|██████▍ | 19384/30000 [12:15:23<5:29:22, 1.86s/it] 65%|██████▍ | 19385/30000 [12:15:28<8:13:42, 2.79s/it] 65%|██████▍ | 19386/30000 [12:15:30<6:47:35, 2.30s/it] 65%|██████▍ | 19387/30000 [12:15:31<5:48:35, 1.97s/it] 65%|██████▍ | 19388/30000 [12:15:32<5:06:20, 1.73s/it] 65%|██████▍ | 19389/30000 [12:15:33<4:35:50, 1.56s/it] 65%|██████▍ | 19390/30000 [12:15:34<4:15:41, 1.45s/it] 65%|██████▍ | 19390/30000 [12:15:34<4:15:41, 1.45s/it] 65%|██████▍ | 19391/30000 [12:15:35<4:01:37, 1.37s/it] 65%|██████▍ | 19392/30000 [12:15:39<5:37:08, 1.91s/it] 65%|██████▍ | 19393/30000 [12:15:43<7:41:22, 2.61s/it] 65%|██████▍ | 19394/30000 [12:15:44<6:24:40, 2.18s/it] 65%|██████▍ | 19395/30000 [12:15:45<5:32:19, 1.88s/it] 65%|██████▍ | 19396/30000 [12:15:46<4:54:18, 1.67s/it] 65%|██████▍ | 19397/30000 [12:15:48<4:29:16, 1.52s/it] 65%|██████▍ | 19398/30000 [12:15:49<4:11:08, 1.42s/it] 65%|██████▍ | 19399/30000 [12:15:50<3:57:46, 1.35s/it] 65%|██████▍ | 19400/30000 [12:15:53<5:29:00, 1.86s/it] 65%|██████▍ | 19400/30000 [12:15:53<5:29:00, 1.86s/it] 65%|██████▍ | 19401/30000 [12:15:57<7:06:09, 2.41s/it] 65%|██████▍ | 19402/30000 [12:15:58<6:00:00, 2.04s/it] 65%|██████▍ | 19403/30000 [12:15:59<5:15:30, 1.79s/it] 65%|██████▍ | 19404/30000 [12:16:00<4:45:07, 1.61s/it] 65%|██████▍ | 19405/30000 [12:16:02<4:22:29, 1.49s/it] 65%|██████▍ | 19406/30000 [12:16:03<4:06:52, 1.40s/it] 65%|██████▍ | 19407/30000 [12:16:04<3:56:16, 1.34s/it] 65%|██████▍ | 19408/30000 [12:16:08<6:37:17, 2.25s/it] 65%|██████▍ | 19409/30000 [12:16:12<7:42:08, 2.62s/it] 65%|██████▍ | 19410/30000 [12:16:13<6:25:33, 2.18s/it] 65%|██████▍ | 19410/30000 [12:16:13<6:25:33, 2.18s/it] 65%|██████▍ | 19411/30000 [12:16:14<5:33:18, 1.89s/it] 65%|██████▍ | 19412/30000 [12:16:15<4:54:59, 1.67s/it] 65%|██████▍ | 19413/30000 [12:16:16<4:29:45, 1.53s/it] 65%|██████▍ | 19414/30000 [12:16:18<4:13:11, 1.44s/it] 65%|██████▍ | 19415/30000 [12:16:19<3:58:52, 1.35s/it] 65%|██████▍ | 19416/30000 [12:16:23<6:49:26, 2.32s/it] 65%|██████▍ | 19417/30000 [12:16:27<7:44:07, 2.63s/it] 65%|██████▍ | 19418/30000 [12:16:28<6:27:22, 2.20s/it] 65%|██████▍ | 19419/30000 [12:16:29<5:37:14, 1.91s/it] 65%|██████▍ | 19420/30000 [12:16:30<4:59:35, 1.70s/it] 65%|██████▍ | 19420/30000 [12:16:30<4:59:35, 1.70s/it] 65%|██████▍ | 19421/30000 [12:16:32<4:30:41, 1.54s/it] 65%|██████▍ | 19422/30000 [12:16:33<4:11:43, 1.43s/it] 65%|██████▍ | 19423/30000 [12:16:34<3:59:24, 1.36s/it] 65%|██████▍ | 19424/30000 [12:16:38<6:17:32, 2.14s/it] 65%|██████▍ | 19425/30000 [12:16:40<6:34:53, 2.24s/it] 65%|██████▍ | 19426/30000 [12:16:42<5:38:36, 1.92s/it] 65%|██████▍ | 19427/30000 [12:16:43<5:00:58, 1.71s/it] 65%|██████▍ | 19428/30000 [12:16:44<4:32:50, 1.55s/it] 65%|██████▍ | 19429/30000 [12:16:45<4:13:12, 1.44s/it] 65%|██████▍ | 19430/30000 [12:16:46<3:59:55, 1.36s/it] +{'loss': 0.0129, 'grad_norm': 0.06695058941841125, 'learning_rate': 9.711026860773595e-06, 'flow_matching_loss': 0.01286294530145824, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.53} +{'loss': 0.0133, 'grad_norm': 0.07664590328931808, 'learning_rate': 9.69555732309408e-06, 'flow_matching_loss': 0.013298728410154581, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.54} +{'loss': 0.0153, 'grad_norm': 0.09982796758413315, 'learning_rate': 9.68009423081259e-06, 'flow_matching_loss': 0.01533393943682313, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.54} +{'loss': 0.0134, 'grad_norm': 0.08752105385065079, 'learning_rate': 9.664637602718249e-06, 'flow_matching_loss': 0.013436759077012538, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.54} +{'loss': 0.0134, 'grad_norm': 0.07009952515363693, 'learning_rate': 9.649187457592313e-06, 'flow_matching_loss': 0.013410498481243848, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.55} +{'loss': 0.0135, 'grad_norm': 0.06905335187911987, 'learning_rate': 9.63374381420816e-06, 'flow_matching_loss': 0.013525201100856065, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.55} +{'loss': 0.0145, 'grad_norm': 0.08070108294487, 'learning_rate': 9.618306691331279e-06, 'flow_matching_loss': 0.014533462934195995, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.55} +{'loss': 0.0151, 'grad_norm': 0.08117672055959702, 'learning_rate': 9.602876107719228e-06, 'flow_matching_loss': 0.01508248345926404, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.55} +{'loss': 0.0138, 'grad_norm': 0.09232920408248901, 'learning_rate': 9.587452082121621e-06, 'flow_matching_loss': 0.013818306941539049, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.56} +{'loss': 0.014, 'grad_norm': 0.06534752994775772, 'learning_rate': 9.57203463328011e-06, 'flow_matching_loss': 0.01401618393138051, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.56} +{'loss': 0.0132, 'grad_norm': 0.0608723908662796, 'learning_rate': 9.556623779928341e-06, 'flow_matching_loss': 0.013245942629873752, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.56} +{'loss': 0.0129, 'grad_norm': 0.08211256563663483, 'learning_rate': 9.541219540791949e-06, 'flow_matching_loss': 0.012943431548774243, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.57} +{'loss': 0.0147, 'grad_norm': 0.06107352301478386, 'learning_rate': 9.525821934588551e-06, 'flow_matching_loss': 0.014719156827777625, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.57} +{'loss': 0.0139, 'grad_norm': 0.0753227025270462, 'learning_rate': 9.510430980027684e-06, 'flow_matching_loss': 0.013904716819524765, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.57} +{'loss': 0.0135, 'grad_norm': 0.056349169462919235, 'learning_rate': 9.495046695810809e-06, 'flow_matching_loss': 0.013522779569029808, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.57} +{'loss': 0.0137, 'grad_norm': 0.09230038523674011, 'learning_rate': 9.47966910063128e-06, 'flow_matching_loss': 0.013702734280377627, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.58} +{'loss': 0.0141, 'grad_norm': 0.07966207712888718, 'learning_rate': 9.464298213174325e-06, 'flow_matching_loss': 0.014062075223773719, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.58} +{'loss': 0.0125, 'grad_norm': 0.0641256645321846, 'learning_rate': 9.448934052117024e-06, 'flow_matching_loss': 0.012519476935267448, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.58} +{'loss': 0.0141, 'grad_norm': 0.0813443511724472, 'learning_rate': 9.433576636128284e-06, 'flow_matching_loss': 0.01410133419558406, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.59} +{'loss': 0.0128, 'grad_norm': 0.07929657399654388, 'learning_rate': 9.41822598386881e-06, 'flow_matching_loss': 0.012764712329953909, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.59} +{'loss': 0.0146, 'grad_norm': 0.05423362925648689, 'learning_rate': 9.402882113991094e-06, 'flow_matching_loss': 0.01461963662877679, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.59} +{'loss': 0.0153, 'grad_norm': 0.08455118536949158, 'learning_rate': 9.387545045139381e-06, 'flow_matching_loss': 0.015321825724095106, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.59} +{'loss': 0.0132, 'grad_norm': 0.06820191442966461, 'learning_rate': 9.372214795949655e-06, 'flow_matching_loss': 0.013162200059741735, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.6} +{'loss': 0.0147, 'grad_norm': 0.07933379709720612, 'learning_rate': 9.356891385049618e-06, 'flow_matching_loss': 0.014720608107745647, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.6} +{'loss': 0.014, 'grad_norm': 0.06991490721702576, 'learning_rate': 9.341574831058664e-06, 'flow_matching_loss': 0.014043298736214637, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.6} +{'loss': 0.0138, 'grad_norm': 0.06425200402736664, 'learning_rate': 9.326265152587844e-06, 'flow_matching_loss': 0.01379737388342619, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.61} +{'loss': 0.0139, 'grad_norm': 0.0696304515004158, 'learning_rate': 9.310962368239861e-06, 'flow_matching_loss': 0.013860025443136692, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.61} +{'loss': 0.0146, 'grad_norm': 0.07177449762821198, 'learning_rate': 9.295666496609036e-06, 'flow_matching_loss': 0.014556194189935923, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.61} +{'loss': 0.0138, 'grad_norm': 0.06869437545537949, 'learning_rate': 9.280377556281306e-06, 'flow_matching_loss': 0.013794043008238078, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.62} +{'loss': 0.0163, 'grad_norm': 0.06978055089712143, 'learning_rate': 9.265095565834172e-06, 'flow_matching_loss': 0.01628536283969879, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.62} +{'loss': 0.0124, 'grad_norm': 0.09864863008260727, 'learning_rate': 9.249820543836688e-06, 'flow_matching_loss': 0.012428372167050838, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.62} +{'loss': 0.0146, 'grad_norm': 0.08128707855939865, 'learning_rate': 9.234552508849454e-06, 'flow_matching_loss': 0.014627764001488686, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.62} +{'loss': 0.0132, 'grad_norm': 0.07800736278295517, 'learning_rate': 9.21929147942456e-06, 'flow_matching_loss': 0.013230200204998255, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.63} +{'loss': 0.0126, 'grad_norm': 0.07449322938919067, 'learning_rate': 9.204037474105611e-06, 'flow_matching_loss': 0.01255423789843917, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.63} +{'loss': 0.0139, 'grad_norm': 0.0706578940153122, 'learning_rate': 9.188790511427653e-06, 'flow_matching_loss': 0.013917815499007702, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.63} +{'loss': 0.0138, 'grad_norm': 0.07333850860595703, 'learning_rate': 9.173550609917186e-06, 'flow_matching_loss': 0.01381744146347046, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.64} +{'loss': 0.0149, 'grad_norm': 0.099968321621418, 'learning_rate': 9.158317788092124e-06, 'flow_matching_loss': 0.014906936138868333, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.64} +{'loss': 0.0138, 'grad_norm': 0.05746352672576904, 'learning_rate': 9.143092064461783e-06, 'flow_matching_loss': 0.013764949236065149, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.64} +{'loss': 0.0125, 'grad_norm': 0.047256603837013245, 'learning_rate': 9.127873457526855e-06, 'flow_matching_loss': 0.012484820792451501, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.64} +{'loss': 0.0139, 'grad_norm': 0.07210560142993927, 'learning_rate': 9.112661985779382e-06, 'flow_matching_loss': 0.013861607294529676, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.65} +{'loss': 0.0148, 'grad_norm': 0.08028285950422287, 'learning_rate': 9.09745766770274e-06, 'flow_matching_loss': 0.014828621596097945, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.65} +{'loss': 0.014, 'grad_norm': 0.08074948191642761, 'learning_rate': 9.082260521771601e-06, 'flow_matching_loss': 0.014000285137444734, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.65} 65%|██████▍ | 19430/30000 [12:16:46<3:59:55, 1.36s/it] 65%|██████▍ | 19431/30000 [12:16:48<3:51:04, 1.31s/it] 65%|██████▍ | 19432/30000 [12:16:52<6:15:56, 2.13s/it] 65%|██████▍ | 19433/30000 [12:16:54<6:28:59, 2.21s/it] 65%|██████▍ | 19434/30000 [12:16:55<5:33:55, 1.90s/it] 65%|██████▍ | 19435/30000 [12:16:56<4:56:24, 1.68s/it] 65%|██████▍ | 19436/30000 [12:16:57<4:29:28, 1.53s/it] 65%|██████▍ | 19437/30000 [12:16:59<4:10:26, 1.42s/it] 65%|██████▍ | 19438/30000 [12:17:00<3:57:33, 1.35s/it] 65%|██████▍ | 19439/30000 [12:17:01<3:49:11, 1.30s/it] 65%|██████▍ | 19440/30000 [12:17:05<6:35:40, 2.25s/it] 65%|██████▍ | 19440/30000 [12:17:05<6:35:40, 2.25s/it] 65%|██████▍ | 19441/30000 [12:17:08<7:13:59, 2.47s/it] 65%|██████▍ | 19442/30000 [12:17:10<6:05:27, 2.08s/it] 65%|██████▍ | 19443/30000 [12:17:11<5:18:27, 1.81s/it] 65%|██████▍ | 19444/30000 [12:17:12<4:46:17, 1.63s/it] 65%|██████▍ | 19445/30000 [12:17:13<4:22:26, 1.49s/it] 65%|██████▍ | 19446/30000 [12:17:14<4:06:16, 1.40s/it] 65%|██████▍ | 19447/30000 [12:17:16<3:54:21, 1.33s/it] 65%|██████▍ | 19448/30000 [12:17:19<5:51:43, 2.00s/it] 65%|██████▍ | 19449/30000 [12:17:22<6:51:12, 2.34s/it] 65%|██████▍ | 19450/30000 [12:17:23<5:48:46, 1.98s/it] 65%|██████▍ | 19450/30000 [12:17:23<5:48:46, 1.98s/it] 65%|██████▍ | 19451/30000 [12:17:25<5:08:27, 1.75s/it] 65%|██████▍ | 19452/30000 [12:17:26<4:38:47, 1.59s/it] 65%|██████▍ | 19453/30000 [12:17:27<4:16:48, 1.46s/it] 65%|██████▍ | 19454/30000 [12:17:28<4:02:44, 1.38s/it] 65%|██████▍ | 19455/30000 [12:17:29<3:52:18, 1.32s/it] 65%|██████▍ | 19456/30000 [12:17:33<6:04:42, 2.08s/it] 65%|██████▍ | 19457/30000 [12:17:37<7:12:55, 2.46s/it] 65%|██████▍ | 19458/30000 [12:17:38<6:04:16, 2.07s/it] 65%|██████▍ | 19459/30000 [12:17:39<5:17:58, 1.81s/it] 65%|██████▍ | 19460/30000 [12:17:40<4:44:39, 1.62s/it] 65%|██████▍ | 19460/30000 [12:17:40<4:44:39, 1.62s/it] 65%|██████▍ | 19461/30000 [12:17:41<4:21:14, 1.49s/it] 65%|██████▍ | 19462/30000 [12:17:42<4:04:49, 1.39s/it] 65%|██████▍ | 19463/30000 [12:17:44<3:55:04, 1.34s/it] 65%|██████▍ | 19464/30000 [12:17:48<6:19:23, 2.16s/it] 65%|██████▍ | 19465/30000 [12:17:51<7:28:39, 2.56s/it] 65%|██████▍ | 19466/30000 [12:17:52<6:15:53, 2.14s/it] 65%|██████▍ | 19467/30000 [12:17:54<5:26:06, 1.86s/it] 65%|██████▍ | 19468/30000 [12:17:55<4:50:42, 1.66s/it] 65%|██████▍ | 19469/30000 [12:17:56<4:25:38, 1.51s/it] 65%|██████▍ | 19470/30000 [12:17:57<4:08:45, 1.42s/it] 65%|██████▍ | 19470/30000 [12:17:57<4:08:45, 1.42s/it] 65%|██████▍ | 19471/30000 [12:17:58<3:55:52, 1.34s/it] 65%|██████▍ | 19472/30000 [12:18:02<6:00:42, 2.06s/it] 65%|██████▍ | 19473/30000 [12:18:06<7:16:44, 2.49s/it] 65%|██████▍ | 19474/30000 [12:18:07<6:07:32, 2.10s/it] 65%|██████▍ | 19475/30000 [12:18:08<5:20:35, 1.83s/it] 65%|██████▍ | 19476/30000 [12:18:09<4:45:48, 1.63s/it] 65%|██████▍ | 19477/30000 [12:18:10<4:21:51, 1.49s/it] 65%|██████▍ | 19478/30000 [12:18:11<4:05:41, 1.40s/it] 65%|██████▍ | 19479/30000 [12:18:13<3:53:46, 1.33s/it] 65%|██████▍ | 19480/30000 [12:18:16<5:26:22, 1.86s/it] 65%|██████▍ | 19480/30000 [12:18:16<5:26:22, 1.86s/it] 65%|██████▍ | 19481/30000 [12:18:20<7:35:43, 2.60s/it] 65%|██████▍ | 19482/30000 [12:18:21<6:20:33, 2.17s/it] 65%|██████▍ | 19483/30000 [12:18:22<5:29:19, 1.88s/it] 65%|██████▍ | 19484/30000 [12:18:24<4:51:56, 1.67s/it] 65%|██████▍ | 19485/30000 [12:18:25<4:26:44, 1.52s/it] 65%|██████▍ | 19486/30000 [12:18:26<4:09:24, 1.42s/it] 65%|██████▍ | 19487/30000 [12:18:27<3:56:45, 1.35s/it] 65%|██████▍ | 19488/30000 [12:18:31<6:15:00, 2.14s/it] 65%|██████▍ | 19489/30000 [12:18:35<7:52:41, 2.70s/it] 65%|██████▍ | 19490/30000 [12:18:36<6:32:30, 2.24s/it] 65%|██████▍ | 19490/30000 [12:18:36<6:32:30, 2.24s/it] 65%|██████▍ | 19491/30000 [12:18:37<5:37:13, 1.93s/it] 65%|██████▍ | 19492/30000 [12:18:39<4:58:25, 1.70s/it] 65%|██████▍ | 19493/30000 [12:18:40<4:31:39, 1.55s/it] 65%|██████▍ | 19494/30000 [12:18:41<4:13:04, 1.45s/it] 65%|██████▍ | 19495/30000 [12:18:42<3:58:21, 1.36s/it] 65%|██████▍ | 19496/30000 [12:18:46<5:49:40, 2.00s/it] 65%|██████▍ | 19497/30000 [12:18:49<7:19:09, 2.51s/it] 65%|██████▍ | 19498/30000 [12:18:51<6:09:01, 2.11s/it] 65%|██████▍ | 19499/30000 [12:18:52<5:20:54, 1.83s/it] 65%|██████▌ | 19500/30000 [12:18:53<4:47:36, 1.64s/it] 65%|██████▌ | 19500/30000 [12:18:53<4:47:36, 1.64s/it] 65%|██████▌ | 19501/30000 [12:18:54<4:23:23, 1.51s/it] 65%|██████▌ | 19502/30000 [12:18:55<4:06:47, 1.41s/it] 65%|██████▌ | 19503/30000 [12:18:57<3:54:20, 1.34s/it] 65%|██████▌ | 19504/30000 [12:19:00<5:54:57, 2.03s/it] 65%|██████▌ | 19505/30000 [12:19:04<7:31:16, 2.58s/it] 65%|██████▌ | 19506/30000 [12:19:05<6:17:04, 2.16s/it] 65%|██████▌ | 19507/30000 [12:19:06<5:26:06, 1.86s/it] 65%|██████▌ | 19508/30000 [12:19:08<4:49:43, 1.66s/it] 65%|██████▌ | 19509/30000 [12:19:09<4:24:31, 1.51s/it] 65%|██████▌ | 19510/30000 [12:19:10<4:06:50, 1.41s/it] 65%|██████▌ | 19510/30000 [12:19:10<4:06:50, 1.41s/it] 65%|██████▌ | 19511/30000 [12:19:11<3:55:29, 1.35s/it] 65%|██████▌ | 19512/30000 [12:19:15<6:03:15, 2.08s/it] 65%|██████▌ | 19513/30000 [12:19:18<7:07:15, 2.44s/it] 65%|██████▌ | 19514/30000 [12:19:19<6:00:38, 2.06s/it] 65%|██████▌ | 19515/30000 [12:19:21<5:15:20, 1.80s/it] 65%|██████▌ | 19516/30000 [12:19:22<4:43:35, 1.62s/it] 65%|██████▌ | 19517/30000 [12:19:23<4:20:09, 1.49s/it] 65%|██████▌ | 19518/30000 [12:19:24<4:04:16, 1.40s/it] 65%|██████▌ | 19519/30000 [12:19:25<3:52:47, 1.33s/it] 65%|██████▌ | 19520/30000 [12:19:29<6:12:57, 2.14s/it] 65%|██████▌ | 19520/30000 [12:19:29<6:12:57, 2.14s/it] 65%|██████▌ | 19521/30000 [12:19:33<7:16:25, 2.50s/it] 65%|██████▌ | 19522/30000 [12:19:34<6:06:51, 2.10s/it] 65%|██████▌ | 19523/30000 [12:19:35<5:20:26, 1.84s/it] 65%|██████▌ | 19524/30000 [12:19:36<4:45:33, 1.64s/it] 65%|██████▌ | 19525/30000 [12:19:37<4:20:58, 1.49s/it] 65%|██████▌ | 19526/30000 [12:19:39<4:05:09, 1.40s/it] 65%|██████▌ | 19527/30000 [12:19:40<3:53:52, 1.34s/it] 65%|██████▌ | 19528/30000 [12:19:44<6:05:04, 2.09s/it] 65%|██████▌ | 19529/30000 [12:19:47<6:55:14, 2.38s/it] 65%|██████▌ | 19530/30000 [12:19:48<5:52:30, 2.02s/it] 65%|██████▌ | 19530/30000 [12:19:48<5:52:30, 2.02s/it] 65%|██████▌ | 19531/30000 [12:19:49<5:13:17, 1.80s/it] 65%|██████▌ | 19532/30000 [12:19:50<4:40:49, 1.61s/it] 65%|██████▌ | 19533/30000 [12:19:51<4:17:26, 1.48s/it] 65%|██████▌ | 19534/30000 [12:19:53<4:01:40, 1.39s/it] 65%|██████▌ | 19535/30000 [12:19:54<3:50:27, 1.32s/it] 65%|██████▌ | 19536/30000 [12:19:58<6:36:10, 2.27s/it] 65%|██████▌ | 19537/30000 [12:20:01<7:25:36, 2.56s/it] 65%|██████▌ | 19538/30000 [12:20:03<6:13:22, 2.14s/it] 65%|██████▌ | 19539/30000 [12:20:04<5:23:50, 1.86s/it] 65%|██████▌ | 19540/30000 [12:20:05<4:47:02, 1.65s/it] 65%|██████▌ | 19540/30000 [12:20:05<4:47:02, 1.65s/it] 65%|██████▌ | 19541/30000 [12:20:06<4:23:27, 1.51s/it] 65%|██████▌ | 19542/30000 [12:20:07<4:07:55, 1.42s/it] 65%|██████▌ | 19543/30000 [12:20:09<3:55:51, 1.35s/it] 65%|██████▌ | 19544/30000 [12:20:13<6:24:48, 2.21s/it] 65%|██████▌ | 19545/30000 [12:20:16<7:29:31, 2.58s/it] 65%|██████▌ | 19546/30000 [12:20:17<6:16:35, 2.16s/it] 65%|██████▌ | 19547/30000 [12:20:19<5:26:54, 1.88s/it] 65%|██████▌ | 19548/30000 [12:20:20<4:49:37, 1.66s/it] 65%|██████▌ | 19549/30000 [12:20:21<4:23:51, 1.51s/it] 65%|██████▌ | 19550/30000 [12:20:22<4:07:00, 1.42s/it] 65%|██████▌ | 19550/30000 [12:20:22<4:07:00, 1.42s/it] 65%|██████▌ | 19551/30000 [12:20:23<3:55:18, 1.35s/it] 65%|██████▌ | 19552/30000 [12:20:27<5:49:44, 2.01s/it] 65%|██████▌ | 19553/30000 [12:20:30<6:38:53, 2.29s/it] 65%|██████▌ | 19554/30000 [12:20:31<5:40:09, 1.95s/it] 65%|██████▌ | 19555/30000 [12:20:32<5:00:23, 1.73s/it] 65%|██████▌ | 19556/30000 [12:20:33<4:31:29, 1.56s/it] 65%|██████▌ | 19557/30000 [12:20:35<4:11:34, 1.45s/it] 65%|██████▌ | 19558/30000 [12:20:36<3:58:09, 1.37s/it] 65%|██████▌ | 19559/30000 [12:20:37<3:49:07, 1.32s/it] 65%|██████▌ | 19560/30000 [12:20:42<6:57:46, 2.40s/it] 65%|██████▌ | 19560/30000 [12:20:42<6:57:46, 2.40s/it] 65%|██████▌ | 19561/30000 [12:20:45<7:19:47, 2.53s/it] 65%|██████▌ | 19562/30000 [12:20:46<6:09:06, 2.12s/it] 65%|██████▌ | 19563/30000 [12:20:47<5:21:02, 1.85s/it] 65%|██████▌ | 19564/30000 [12:20:48<4:46:16, 1.65s/it] 65%|██████▌ | 19565/30000 [12:20:49<4:24:07, 1.52s/it] 65%|██████▌ | 19566/30000 [12:20:51<4:07:20, 1.42s/it] 65%|██████▌ | 19567/30000 [12:20:52<3:54:39, 1.35s/it] 65%|██████▌ | 19568/30000 [12:20:57<7:26:16, 2.57s/it] 65%|██████▌ | 19569/30000 [12:21:00<7:51:38, 2.71s/it] 65%|██████▌ | 19570/30000 [12:21:02<6:32:04, 2.26s/it] 65%|██████▌ | 19570/30000 [12:21:02<6:32:04, 2.26s/it] 65%|██████▌ | 19571/30000 [12:21:03<5:38:33, 1.95s/it] 65%|██████▌ | 19572/30000 [12:21:04<4:58:30, 1.72s/it] 65%|██████▌ | 19573/30000 [12:21:05<4:30:08, 1.55s/it] 65%|██████▌ | 19574/30000 [12:21:06<4:11:13, 1.45s/it] 65%|██████▌ | 19575/30000 [12:21:07<3:57:28, 1.37s/it] 65%|██████▌ | 19576/30000 [12:21:11<6:09:26, 2.13s/it] 65%|██████▌ | 19577/30000 [12:21:14<6:39:07, 2.30s/it] 65%|██████▌ | 19578/30000 [12:21:15<5:39:57, 1.96s/it] 65%|██████▌ | 19579/30000 [12:21:16<5:00:12, 1.73s/it] 65%|██████▌ | 19580/30000 [12:21:18<4:31:19, 1.56s/it] 65%|██████▌ | 19580/30000 [12:21:18<4:31:19, 1.56s/it] 65%|██████▌ | 19581/30000 [12:21:19<4:12:39, 1.45s/it] 65%|██████▌ | 19582/30000 [12:21:20<3:59:01, 1.38s/it] 65%|██████▌ | 19583/30000 [12:21:21<3:48:47, 1.32s/it] 65%|██████▌ | 19584/30000 [12:21:26<6:43:48, 2.33s/it] 65%|██████▌ | 19585/30000 [12:21:29<7:07:03, 2.46s/it] 65%|██████▌ | 19586/30000 [12:21:30<6:00:04, 2.07s/it] 65%|██████▌ | 19587/30000 [12:21:31<5:14:29, 1.81s/it] 65%|██████▌ | 19588/30000 [12:21:32<4:41:53, 1.62s/it] 65%|██████▌ | 19589/30000 [12:21:33<4:18:49, 1.49s/it] 65%|██████▌ | 19590/30000 [12:21:35<4:03:27, 1.40s/it] 65%|██████▌ | 19590/30000 [12:21:35<4:03:27, 1.40s/it] 65%|██████▌ | 19591/30000 [12:21:36<3:52:17, 1.34s/it] 65%|██████▌ | 19592/30000 [12:21:40<6:15:21, 2.16s/it] 65%|██████▌ | 19593/30000 [12:21:43<7:28:25, 2.59s/it] 65%|██████▌ | 19594/30000 [12:21:45<6:15:14, 2.16s/it] 65%|██████▌ | 19595/30000 [12:21:46<5:25:32, 1.88s/it] 65%|██████▌ | 19596/30000 [12:21:47<4:51:23, 1.68s/it] 65%|██████▌ | 19597/30000 [12:21:48<4:25:08, 1.53s/it] 65%|██████▌ | 19598/30000 [12:21:49<4:07:29, 1.43s/it] 65%|██████▌ | 19599/30000 [12:21:51<3:55:00, 1.36s/it] 65%|██████▌ | 19600/30000 [12:21:54<5:28:48, 1.90s/it] 65%|██████▌ | 19600/30000 [12:21:54<5:28:48, 1.90s/it] 65%|██████▌ | 19601/30000 [12:21:57<7:03:20, 2.44s/it] 65%|██████▌ | 19602/30000 [12:21:59<5:57:23, 2.06s/it] 65%|██████▌ | 19603/30000 [12:22:00<5:12:46, 1.80s/it] 65%|██████▌ | 19604/30000 [12:22:01<4:39:57, 1.62s/it] 65%|██████▌ | 19605/30000 [12:22:02<4:16:13, 1.48s/it] 65%|██████▌ | 19606/30000 [12:22:03<4:01:23, 1.39s/it] 65%|██████▌ | 19607/30000 [12:22:05<3:51:17, 1.34s/it] 65%|██████▌ | 19608/30000 [12:22:08<5:28:25, 1.90s/it] 65%|██████▌ | 19609/30000 [12:22:12<7:38:41, 2.65s/it] 65%|██████▌ | 19610/30000 [12:22:13<6:22:46, 2.21s/it] 65%|██████▌ | 19610/30000 [12:22:13<6:22:46, 2.21s/it] 65%|██████▌ | 19611/30000 [12:22:15<5:31:03, 1.91s/it] 65%|██████▌ | 19612/30000 [12:22:16<4:52:55, 1.69s/it] 65%|██████▌ | 19613/30000 [12:22:17<4:27:27, 1.54s/it] 65%|██████▌ | 19614/30000 [12:22:18<4:09:54, 1.44s/it] 65%|██████▌ | 19615/30000 [12:22:19<3:57:25, 1.37s/it] 65%|██████▌ | 19616/30000 [12:22:22<5:12:23, 1.81s/it] 65%|██████▌ | 19617/30000 [12:22:27<7:37:15, 2.64s/it] 65%|██████▌ | 19618/30000 [12:22:28<6:20:37, 2.20s/it] 65%|██████▌ | 19619/30000 [12:22:29<5:28:25, 1.90s/it] 65%|██████▌ | 19620/30000 [12:22:30<4:50:25, 1.68s/it] 65%|██████▌ | 19620/30000 [12:22:30<4:50:25, 1.68s/it] 65%|██████▌ | 19621/30000 [12:22:32<4:24:46, 1.53s/it] 65%|██████▌ | 19622/30000 [12:22:33<4:07:12, 1.43s/it] 65%|██████▌ | 19623/30000 [12:22:34<3:54:07, 1.35s/it] 65%|██████▌ | 19624/30000 [12:22:37<5:04:44, 1.76s/it] 65%|██████▌ | 19625/30000 [12:22:41<7:18:56, 2.54s/it] 65%|██████▌ | 19626/30000 [12:22:42<6:08:08, 2.13s/it] 65%|██████▌ | 19627/30000 [12:22:43<5:19:23, 1.85s/it] 65%|██████▌ | 19628/30000 [12:22:44<4:44:25, 1.65s/it] 65%|██████▌ | 19629/30000 [12:22:46<4:20:54, 1.51s/it] 65%|██████▌ | 19630/30000 [12:22:47<4:04:12, 1.41s/it] 65%|██████▌ | 19630/30000 [12:22:47<4:04:12, 1.41s/it] 65%|██████▌ | 19631/30000 [12:22:48<3:52:27, 1.35s/it] 65%|██████▌ | 19632/30000 [12:22:51<5:38:46, 1.96s/it] 65%|██████▌ | 19633/30000 [12:22:57<8:19:30, 2.89s/it] 65%|██████▌ | 19634/30000 [12:22:58<6:51:19, 2.38s/it] 65%|██████▌ | 19635/30000 [12:22:59<5:49:02, 2.02s/it] 65%|██████▌ | 19636/30000 [12:23:00<5:04:24, 1.76s/it] 65%|██████▌ | 19637/30000 [12:23:01<4:33:38, 1.58s/it] 65%|██████▌ | 19638/30000 [12:23:02<4:13:41, 1.47s/it] 65%|██████▌ | 19639/30000 [12:23:04<3:59:45, 1.39s/it] 65%|██████▌ | 19640/30000 [12:23:06<4:53:50, 1.70s/it] 65%|██████▌ | 19640/30000 [12:23:06<4:53:50, 1.70s/it] 65%|██████▌ | 19641/30000 [12:23:12<8:13:02, 2.86s/it] 65%|██████▌ | 19642/30000 [12:23:13<6:45:32, 2.35s/it] 65%|██████▌ | 19643/30000 [12:23:14<5:46:28, 2.01s/it] 65%|██████▌ | 19644/30000 [12:23:15<5:03:15, 1.76s/it] 65%|██████▌ | 19645/30000 [12:23:16<4:33:04, 1.58s/it] 65%|██████▌ | 19646/30000 [12:23:17<4:12:08, 1.46s/it] 65%|██████▌ | 19647/30000 [12:23:19<3:58:05, 1.38s/it] 65%|██████▌ | 19648/30000 [12:23:20<4:19:28, 1.50s/it] 65%|██████▌ | 19649/30000 [12:23:26<7:39:20, 2.66s/it] 66%|██████▌ | 19650/30000 [12:23:27<6:21:47, 2.21s/it] 66%|██████▌ | 19650/30000 [12:23:27<6:21:47, 2.21s/it] 66%|██████▌ | 19651/30000 [12:23:28<5:29:59, 1.91s/it] 66%|██████▌ | 19652/30000 [12:23:29<4:51:28, 1.69s/it] 66%|██████▌ | 19653/30000 [12:23:31<4:25:07, 1.54s/it] 66%|██████▌ | 19654/30000 [12:23:32<4:07:13, 1.43s/it] 66%|██████▌ | 19655/30000 [12:23:33<3:54:55, 1.36s/it] 66%|██████▌ | 19656/30000 [12:23:35<4:47:59, 1.67s/it] 66%|██████▌ | 19657/30000 [12:23:41<7:48:45, 2.72s/it] 66%|██████▌ | 19658/30000 [12:23:42<6:28:33, 2.25s/it] 66%|██████▌ | 19659/30000 [12:23:43<5:33:43, 1.94s/it] 66%|██████▌ | 19660/30000 [12:23:44<4:54:32, 1.71s/it] 66%|██████▌ | 19660/30000 [12:23:44<4:54:32, 1.71s/it] 66%|██████▌ | 19661/30000 [12:23:45<4:27:19, 1.55s/it] 66%|██████▌ | 19662/30000 [12:23:46<4:09:10, 1.45s/it] 66%|██████▌ | 19663/30000 [12:23:48<3:55:25, 1.37s/it] 66%|██████▌ | 19664/30000 [12:23:51<5:15:31, 1.83s/it] 66%|██████▌ | 19665/30000 [12:23:56<8:10:09, 2.85s/it] 66%|██████▌ | 19666/30000 [12:23:57<6:43:23, 2.34s/it] 66%|██████▌ | 19667/30000 [12:23:58<5:43:44, 2.00s/it] 66%|██████▌ | 19668/30000 [12:23:59<5:01:36, 1.75s/it] 66%|██████▌ | 19669/30000 [12:24:00<4:31:27, 1.58s/it] 66%|██████▌ | 19670/30000 [12:24:02<4:11:56, 1.46s/it] 66%|██████▌ | 19670/30000 [12:24:02<4:11:56, 1.46s/it] 66%|██████▌ | 19671/30000 [12:24:03<3:58:33, 1.39s/it] 66%|██████▌ | 19672/30000 [12:24:04<4:11:34, 1.46s/it] 66%|██████▌ | 19673/30000 [12:24:09<7:11:14, 2.51s/it] 66%|██████▌ | 19674/30000 [12:24:11<6:02:34, 2.11s/it] 66%|██████▌ | 19675/30000 [12:24:12<5:15:19, 1.83s/it] 66%|██████▌ | 19676/30000 [12:24:13<4:41:40, 1.64s/it] 66%|██████▌ | 19677/30000 [12:24:14<4:17:33, 1.50s/it] 66%|██████▌ | 19678/30000 [12:24:15<4:02:03, 1.41s/it] 66%|██████▌ | 19679/30000 [12:24:17<3:50:51, 1.34s/it] 66%|██████▌ | 19680/30000 [12:24:18<4:17:21, 1.50s/it] 66%|██████▌ | 19680/30000 [12:24:18<4:17:21, 1.50s/it] 66%|██████▌ | 19681/30000 [12:24:23<7:06:34, 2.48s/it] 66%|██████▌ | 19682/30000 [12:24:24<5:59:32, 2.09s/it] 66%|██████▌ | 19683/30000 [12:24:26<5:13:24, 1.82s/it] 66%|██████▌ | 19684/30000 [12:24:27<4:41:50, 1.64s/it] 66%|██████▌ | 19685/30000 [12:24:28<4:19:29, 1.51s/it] 66%|██████▌ | 19686/30000 [12:24:29<4:04:51, 1.42s/it] 66%|██████▌ | 19687/30000 [12:24:30<3:52:20, 1.35s/it] 66%|██████▌ | 19688/30000 [12:24:33<4:57:47, 1.73s/it] 66%|██████▌ | 19689/30000 [12:24:38<7:45:26, 2.71s/it] 66%|██████▌ | 19690/30000 [12:24:39<6:27:22, 2.25s/it] 66%|██████▌ | 19690/30000 [12:24:39<6:27:22, 2.25s/it] 66%|██████▌ | 19691/30000 [12:24:40<5:31:36, 1.93s/it] 66%|██████▌ | 19692/30000 [12:24:42<4:52:17, 1.70s/it] 66%|██████▌ | 19693/30000 [12:24:43<4:24:17, 1.54s/it] 66%|██████▌ | 19694/30000 [12:24:44<4:05:16, 1.43s/it] 66%|██████▌ | 19695/30000 [12:24:45<3:53:54, 1.36s/it] 66%|██████▌ | 19696/30000 [12:24:47<4:40:48, 1.64s/it] 66%|██████▌ | 19697/30000 [12:24:51<6:44:29, 2.36s/it] 66%|██████▌ | 19698/30000 [12:24:53<5:43:58, 2.00s/it] 66%|██████▌ | 19699/30000 [12:24:54<5:02:19, 1.76s/it] 66%|██████▌ | 19700/30000 [12:24:55<4:32:27, 1.59s/it] 66%|██████▌ | 19700/30000 [12:24:55<4:32:27, 1.59s/it] 66%|██████▌ | 19701/30000 [12:24:56<4:11:07, 1.46s/it] 66%|██████▌ | 19702/30000 [12:24:57<3:57:40, 1.38s/it] 66%|██████▌ | 19703/30000 [12:24:59<3:48:30, 1.33s/it] 66%|██████▌ | 19704/30000 [12:25:00<4:13:41, 1.48s/it] 66%|██████▌ | 19705/30000 [12:25:05<7:04:19, 2.47s/it] 66%|██████▌ | 19706/30000 [12:25:06<5:57:00, 2.08s/it] 66%|██████▌ | 19707/30000 [12:25:07<5:10:55, 1.81s/it] 66%|██████▌ | 19708/30000 [12:25:09<4:38:41, 1.62s/it] 66%|██████▌ | 19709/30000 [12:25:10<4:16:15, 1.49s/it] 66%|██████▌ | 19710/30000 [12:25:11<4:01:39, 1.41s/it] 66%|██████▌ | 19710/30000 [12:25:11<4:01:39, 1.41s/it] 66%|██████▌ | 19711/30000 [12:25:12<3:50:48, 1.35s/it] 66%|██████▌ | 19712/30000 [12:25:14<4:24:43, 1.54s/it] 66%|██████▌ | 19713/30000 [12:25:20<7:43:01, 2.70s/it] 66%|██████▌ | 19714/30000 [12:25:21<6:23:56, 2.24s/it] 66%|██████▌ | 19715/30000 [12:25:22<5:29:59, 1.93s/it] 66%|██████▌ | 19716/30000 [12:25:23<4:50:58, 1.70s/it] 66%|██████▌ | 19717/30000 [12:25:24<4:24:28, 1.54s/it] 66%|██████▌ | 19718/30000 [12:25:26<4:06:31, 1.44s/it] 66%|██████▌ | 19719/30000 [12:25:27<3:54:00, 1.37s/it] 66%|██████▌ | 19720/30000 [12:25:29<4:15:20, 1.49s/it] 66%|██████▌ | 19720/30000 [12:25:29<4:15:20, 1.49s/it] 66%|██████▌ | 19721/30000 [12:25:34<7:28:08, 2.62s/it] 66%|██████▌ | 19722/30000 [12:25:35<6:14:14, 2.18s/it] 66%|██████▌ | 19723/30000 [12:25:36<5:22:21, 1.88s/it] 66%|██████▌ | 19724/30000 [12:25:37<4:46:22, 1.67s/it] 66%|██████▌ | 19725/30000 [12:25:39<4:21:07, 1.52s/it] 66%|██████▌ | 19726/30000 [12:25:40<4:03:33, 1.42s/it] 66%|██████▌ | 19727/30000 [12:25:41<3:51:41, 1.35s/it] 66%|██████▌ | 19728/30000 [12:25:43<4:48:45, 1.69s/it] 66%|██████▌ | 19729/30000 [12:25:49<7:57:44, 2.79s/it] 66%|██████▌ | 19730/30000 [12:25:50<6:34:34, 2.31s/it] 66%|██████▌ | 19730/30000 [12:25:50<6:34:34, 2.31s/it] 66%|██████▌ | 19731/30000 [12:25:51<5:37:57, 1.97s/it] 66%|██████▌ | 19732/30000 [12:25:52<4:56:29, 1.73s/it] 66%|██████▌ | 19733/30000 [12:25:53<4:28:00, 1.57s/it] 66%|██████▌ | 19734/30000 [12:25:55<4:09:30, 1.46s/it] 66%|██████▌ | 19735/30000 [12:25:56<3:56:11, 1.38s/it] 66%|██████▌ | 19736/30000 [12:25:58<4:32:48, 1.59s/it] 66%|██████▌ | 19737/30000 [12:26:03<7:41:17, 2.70s/it] 66%|██████▌ | 19738/30000 [12:26:04<6:23:16, 2.24s/it] 66%|██████▌ | 19739/30000 [12:26:06<5:29:06, 1.92s/it] 66%|██████▌ | 19740/30000 [12:26:07<4:50:15, 1.70s/it] 66%|██████▌ | 19740/30000 [12:26:07<4:50:15, 1.70s/it] 66%|██████▌ | 19741/30000 [12:26:08<4:23:16, 1.54s/it] 66%|██████▌ | 19742/30000 [12:26:09<4:04:49, 1.43s/it] 66%|██████▌ | 19743/30000 [12:26:10<3:52:35, 1.36s/it] 66%|██████▌ | 19744/30000 [12:26:12<4:16:10, 1.50s/it] 66%|██████▌ | 19745/30000 [12:26:18<7:55:29, 2.78s/it] 66%|██████▌ | 19746/30000 [12:26:19<6:33:03, 2.30s/it] 66%|██████▌ | 19747/30000 [12:26:20<5:35:55, 1.97s/it] 66%|██████▌ | 19748/30000 [12:26:21<4:55:38, 1.73s/it] 66%|██████▌ | 19749/30000 [12:26:23<4:27:40, 1.57s/it] 66%|██████▌ | 19750/30000 [12:26:24<4:08:18, 1.45s/it] 66%|██████▌ | 19750/30000 [12:26:24<4:08:18, 1.45s/it] 66%|██████▌ | 19751/30000 [12:26:25<3:55:17, 1.38s/it] 66%|██████▌ | 19752/30000 [12:26:27<4:22:06, 1.53s/it] 66%|██████▌ | 19753/30000 [12:26:32<7:34:59, 2.66s/it] 66%|██████▌ | 19754/30000 [12:26:33<6:18:21, 2.22s/it] 66%|██████▌ | 19755/30000 [12:26:35<5:25:21, 1.91s/it] 66%|██████▌ | 19756/30000 [12:26:36<4:47:31, 1.68s/it] 66%|██████▌ | 19757/30000 [12:26:37<4:21:13, 1.53s/it] 66%|██████▌ | 19758/30000 [12:26:38<4:03:32, 1.43s/it] 66%|██████▌ | 19759/30000 [12:26:39<3:53:30, 1.37s/it] 66%|██████▌ | 19760/30000 [12:26:42<4:56:53, 1.74s/it] 66%|██████▌ | 19760/30000 [12:26:42<4:56:53, 1.74s/it] 66%|██████▌ | 19761/30000 [12:26:46<7:17:20, 2.56s/it] 66%|██████▌ | 19762/30000 [12:26:48<6:05:45, 2.14s/it] 66%|██████▌ | 19763/30000 [12:26:49<5:17:15, 1.86s/it] 66%|██████▌ | 19764/30000 [12:26:50<4:42:51, 1.66s/it] 66%|██████▌ | 19765/30000 [12:26:51<4:17:40, 1.51s/it] 66%|██████▌ | 19766/30000 [12:26:52<4:02:00, 1.42s/it] 66%|██████▌ | 19767/30000 [12:26:54<3:50:26, 1.35s/it] 66%|██████▌ | 19768/30000 [12:26:56<4:33:55, 1.61s/it] 66%|██████▌ | 19769/30000 [12:27:01<7:35:32, 2.67s/it] 66%|██████▌ | 19770/30000 [12:27:02<6:19:07, 2.22s/it] 66%|██████▌ | 19770/30000 [12:27:02<6:19:07, 2.22s/it] 66%|██████▌ | 19771/30000 [12:27:03<5:27:43, 1.92s/it] 66%|██████▌ | 19772/30000 [12:27:04<4:50:25, 1.70s/it] 66%|██████▌ | 19773/30000 [12:27:06<4:23:43, 1.55s/it] 66%|██████▌ | 19774/30000 [12:27:07<4:05:18, 1.44s/it] 66%|██████▌ | 19775/30000 [12:27:08<3:53:29, 1.37s/it] 66%|██████▌ | 19776/30000 [12:27:10<4:20:26, 1.53s/it] 66%|██████▌ | 19777/30000 [12:27:16<8:14:45, 2.90s/it] 66%|██████▌ | 19778/30000 [12:27:17<6:46:03, 2.38s/it] 66%|██████▌ | 19779/30000 [12:27:18<5:44:58, 2.03s/it] 66%|██████▌ | 19780/30000 [12:27:20<5:00:56, 1.77s/it] 66%|██████▌ | 19780/30000 [12:27:20<5:00:56, 1.77s/it] 66%|██████▌ | 19781/30000 [12:27:21<4:30:30, 1.59s/it] 66%|██████▌ | 19782/30000 [12:27:22<4:09:54, 1.47s/it] 66%|██████▌ | 19783/30000 [12:27:23<3:55:57, 1.39s/it] 66%|██████▌ | 19784/30000 [12:27:25<4:18:46, 1.52s/it] 66%|██████▌ | 19785/30000 [12:27:31<7:59:59, 2.82s/it] 66%|██████▌ | 19786/30000 [12:27:32<6:35:38, 2.32s/it] 66%|██████▌ | 19787/30000 [12:27:33<5:37:14, 1.98s/it] 66%|██████▌ | 19788/30000 [12:27:34<4:55:41, 1.74s/it] 66%|██████▌ | 19789/30000 [12:27:35<4:26:52, 1.57s/it] 66%|██████▌ | 19790/30000 [12:27:37<4:07:48, 1.46s/it] 66%|██████▌ | 19790/30000 [12:27:37<4:07:48, 1.46s/it] 66%|██████▌ | 19791/30000 [12:27:38<3:54:44, 1.38s/it] 66%|██████▌ | 19792/30000 [12:27:40<4:26:54, 1.57s/it] 66%|██████▌ | 19793/30000 [12:27:46<7:53:26, 2.78s/it] 66%|██████▌ | 19794/30000 [12:27:47<6:31:12, 2.30s/it] 66%|██████▌ | 19795/30000 [12:27:48<5:35:02, 1.97s/it] 66%|██████▌ | 19796/30000 [12:27:49<4:53:51, 1.73s/it] 66%|██████▌ | 19797/30000 [12:27:50<4:24:43, 1.56s/it] 66%|██████▌ | 19798/30000 [12:27:51<4:07:14, 1.45s/it] 66%|██████▌ | 19799/30000 [12:27:53<3:53:37, 1.37s/it] 66%|██████▌ | 19800/30000 [12:27:54<4:03:33, 1.43s/it] 66%|██████▌ | 19800/30000 [12:27:54<4:03:33, 1.43s/it] 66%|██████▌ | 19801/30000 [12:28:00<7:31:01, 2.65s/it] 66%|██████▌ | 19802/30000 [12:28:01<6:15:10, 2.21s/it] 66%|██████▌ | 19803/30000 [12:28:02<5:23:37, 1.90s/it] 66%|██████▌ | 19804/30000 [12:28:03<4:46:51, 1.69s/it] 66%|██████▌ | 19805/30000 [12:28:04<4:20:15, 1.53s/it] 66%|██████▌ | 19806/30000 [12:28:06<4:03:03, 1.43s/it] 66%|██████▌ | 19807/30000 [12:28:07<3:50:35, 1.36s/it] 66%|██████▌ | 19808/30000 [12:28:08<3:42:45, 1.31s/it] 66%|██████▌ | 19809/30000 [12:28:14<7:40:03, 2.71s/it] 66%|██████▌ | 19810/30000 [12:28:15<6:21:23, 2.25s/it] 66%|██████▌ | 19810/30000 [12:28:15<6:21:23, 2.25s/it] 66%|██████▌ | 19811/30000 [12:28:16<5:28:37, 1.94s/it] 66%|██████▌ | 19812/30000 [12:28:18<4:50:06, 1.71s/it] 66%|██████▌ | 19813/30000 [12:28:19<4:24:10, 1.56s/it] 66%|██████▌ | 19814/30000 [12:28:20<4:05:17, 1.44s/it] 66%|██████▌ | 19815/30000 [12:28:21<3:52:18, 1.37s/it] 66%|██████▌ | 19816/30000 [12:28:22<3:44:10, 1.32s/it] 66%|██████▌ | 19817/30000 [12:28:28<7:37:45, 2.70s/it] 66%|██████▌ | 19818/30000 [12:28:29<6:20:08, 2.24s/it] 66%|██████▌ | 19819/30000 [12:28:31<5:26:32, 1.92s/it] 66%|██████▌ | 19820/30000 [12:28:32<4:48:12, 1.70s/it] 66%|██████▌ | 19820/30000 [12:28:32<4:48:12, 1.70s/it] 66%|██████▌ | 19821/30000 [12:28:33<4:20:38, 1.54s/it] 66%|██████▌ | 19822/30000 [12:28:34<4:04:55, 1.44s/it] 66%|██████▌ | 19823/30000 [12:28:35<3:54:01, 1.38s/it] 66%|██████▌ | 19824/30000 [12:28:37<3:48:42, 1.35s/it] 66%|██████▌ | 19825/30000 [12:28:42<7:32:55, 2.67s/it] 66%|██████▌ | 19826/30000 [12:28:44<6:16:21, 2.22s/it] 66%|██████▌ | 19827/30000 [12:28:45<5:23:59, 1.91s/it] 66%|██████▌ | 19828/30000 [12:28:46<4:46:25, 1.69s/it] 66%|██████▌ | 19829/30000 [12:28:47<4:20:30, 1.54s/it] 66%|██████▌ | 19830/30000 [12:28:48<4:04:49, 1.44s/it] 66%|██████▌ | 19830/30000 [12:28:48<4:04:49, 1.44s/it] 66%|██████▌ | 19831/30000 [12:28:50<3:52:14, 1.37s/it] 66%|██████▌ | 19832/30000 [12:28:51<4:13:12, 1.49s/it] 66%|██████▌ | 19833/30000 [12:28:57<7:29:04, 2.65s/it] 66%|██████▌ | 19834/30000 [12:28:58<6:14:07, 2.21s/it] 66%|██████▌ | 19835/30000 [12:28:59<5:22:19, 1.90s/it] 66%|██████▌ | 19836/30000 [12:29:00<4:45:18, 1.68s/it] 66%|██████▌ | 19837/30000 [12:29:01<4:20:14, 1.54s/it] 66%|██████▌ | 19838/30000 [12:29:03<4:02:28, 1.43s/it] 66%|██████▌ | 19839/30000 [12:29:04<3:50:26, 1.36s/it] 66%|██████▌ | 19840/30000 [12:29:06<4:22:09, 1.55s/it] 66%|██████▌ | 19840/30000 [12:29:06<4:22:09, 1.55s/it] 66%|██████▌ | 19841/30000 [12:29:11<7:46:29, 2.76s/it] 66%|██████▌ | 19842/30000 [12:29:12<6:26:20, 2.28s/it] 66%|██████▌ | 19843/30000 [12:29:14<5:31:30, 1.96s/it] 66%|██████▌ | 19844/30000 [12:29:15<4:52:13, 1.73s/it] 66%|██████▌ | 19845/30000 [12:29:16<4:23:23, 1.56s/it] 66%|██████▌ | 19846/30000 [12:29:17<4:06:33, 1.46s/it] 66%|██████▌ | 19847/30000 [12:29:18<3:54:05, 1.38s/it] 66%|██████▌ | 19848/30000 [12:29:21<4:36:59, 1.64s/it] 66%|██████▌ | 19849/30000 [12:29:26<7:35:11, 2.69s/it] 66%|██████▌ | 19850/30000 [12:29:27<6:18:06, 2.24s/it] +{'loss': 0.0138, 'grad_norm': 0.08663112670183182, 'learning_rate': 9.067070566451936e-06, 'flow_matching_loss': 0.013784564472734928, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.66} +{'loss': 0.0148, 'grad_norm': 0.08776795864105225, 'learning_rate': 9.051887820200966e-06, 'flow_matching_loss': 0.014770177751779556, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.66} +{'loss': 0.013, 'grad_norm': 0.06020805612206459, 'learning_rate': 9.03671230146717e-06, 'flow_matching_loss': 0.013020809832960367, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.66} +{'loss': 0.014, 'grad_norm': 0.06758236140012741, 'learning_rate': 9.021544028690228e-06, 'flow_matching_loss': 0.013973519951105118, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.66} +{'loss': 0.0134, 'grad_norm': 0.08342023193836212, 'learning_rate': 9.006383020301024e-06, 'flow_matching_loss': 0.013431855663657188, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.67} +{'loss': 0.0128, 'grad_norm': 0.07798338681459427, 'learning_rate': 8.991229294721613e-06, 'flow_matching_loss': 0.012797566130757332, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.67} +{'loss': 0.0152, 'grad_norm': 0.07702511548995972, 'learning_rate': 8.976082870365191e-06, 'flow_matching_loss': 0.015186396986246109, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.67} +{'loss': 0.013, 'grad_norm': 0.06734813749790192, 'learning_rate': 8.960943765636105e-06, 'flow_matching_loss': 0.013011136092245579, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.68} +{'loss': 0.0133, 'grad_norm': 0.07860258966684341, 'learning_rate': 8.945811998929792e-06, 'flow_matching_loss': 0.013332582451403141, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.68} +{'loss': 0.0127, 'grad_norm': 0.06208709254860878, 'learning_rate': 8.93068758863277e-06, 'flow_matching_loss': 0.012676289584487677, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.68} +{'loss': 0.013, 'grad_norm': 0.06711776554584503, 'learning_rate': 8.915570553122628e-06, 'flow_matching_loss': 0.013011361099779606, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.69} +{'loss': 0.0132, 'grad_norm': 0.06706066429615021, 'learning_rate': 8.900460910767982e-06, 'flow_matching_loss': 0.013220662064850331, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.69} +{'loss': 0.0132, 'grad_norm': 0.07574845850467682, 'learning_rate': 8.885358679928486e-06, 'flow_matching_loss': 0.01316726803779602, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.69} +{'loss': 0.0124, 'grad_norm': 0.06340748816728592, 'learning_rate': 8.870263878954767e-06, 'flow_matching_loss': 0.012383767776191235, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.69} +{'loss': 0.014, 'grad_norm': 0.07075715810060501, 'learning_rate': 8.855176526188428e-06, 'flow_matching_loss': 0.013981906231492758, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.7} +{'loss': 0.014, 'grad_norm': 0.07737661153078079, 'learning_rate': 8.840096639962033e-06, 'flow_matching_loss': 0.014002531953155994, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.7} +{'loss': 0.0135, 'grad_norm': 0.06957145780324936, 'learning_rate': 8.825024238599056e-06, 'flow_matching_loss': 0.01345181129872799, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.7} +{'loss': 0.0144, 'grad_norm': 0.06828322261571884, 'learning_rate': 8.809959340413891e-06, 'flow_matching_loss': 0.014417755044996738, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.71} +{'loss': 0.0136, 'grad_norm': 0.052430395036935806, 'learning_rate': 8.79490196371181e-06, 'flow_matching_loss': 0.013552786596119405, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.71} +{'loss': 0.0139, 'grad_norm': 0.09208868443965912, 'learning_rate': 8.779852126788949e-06, 'flow_matching_loss': 0.013859143014997244, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.71} +{'loss': 0.0147, 'grad_norm': 0.07321837544441223, 'learning_rate': 8.764809847932277e-06, 'flow_matching_loss': 0.014707072917371989, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.71} +{'loss': 0.0132, 'grad_norm': 0.0803893432021141, 'learning_rate': 8.749775145419577e-06, 'flow_matching_loss': 0.013189580477774143, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.72} +{'loss': 0.0161, 'grad_norm': 0.08422481268644333, 'learning_rate': 8.734748037519427e-06, 'flow_matching_loss': 0.01605983395129442, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.72} +{'loss': 0.014, 'grad_norm': 0.07347878813743591, 'learning_rate': 8.719728542491195e-06, 'flow_matching_loss': 0.013996883947402238, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.72} +{'loss': 0.0147, 'grad_norm': 0.07745840400457382, 'learning_rate': 8.704716678584976e-06, 'flow_matching_loss': 0.014734817482531071, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.73} +{'loss': 0.0146, 'grad_norm': 0.08333516865968704, 'learning_rate': 8.689712464041596e-06, 'flow_matching_loss': 0.01462929304689169, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.73} +{'loss': 0.0142, 'grad_norm': 0.07650727778673172, 'learning_rate': 8.674715917092592e-06, 'flow_matching_loss': 0.014246050082147122, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.73} +{'loss': 0.014, 'grad_norm': 0.09963200241327286, 'learning_rate': 8.65972705596018e-06, 'flow_matching_loss': 0.013957061059772968, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.73} +{'loss': 0.0136, 'grad_norm': 0.06418734043836594, 'learning_rate': 8.64474589885725e-06, 'flow_matching_loss': 0.013646051660180092, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.74} +{'loss': 0.0144, 'grad_norm': 0.07088744640350342, 'learning_rate': 8.629772463987315e-06, 'flow_matching_loss': 0.01441462356597185, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.74} +{'loss': 0.0136, 'grad_norm': 0.07893220335245132, 'learning_rate': 8.614806769544507e-06, 'flow_matching_loss': 0.013560963235795498, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.74} +{'loss': 0.0116, 'grad_norm': 0.07044023275375366, 'learning_rate': 8.59984883371356e-06, 'flow_matching_loss': 0.011633584834635257, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.75} +{'loss': 0.0137, 'grad_norm': 0.059705816209316254, 'learning_rate': 8.584898674669772e-06, 'flow_matching_loss': 0.01374082574620843, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.75} +{'loss': 0.0142, 'grad_norm': 0.09106852859258652, 'learning_rate': 8.569956310579001e-06, 'flow_matching_loss': 0.014201650209724904, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.75} +{'loss': 0.0134, 'grad_norm': 0.09440051019191742, 'learning_rate': 8.555021759597627e-06, 'flow_matching_loss': 0.013414052408188581, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.76} +{'loss': 0.0155, 'grad_norm': 0.08630137145519257, 'learning_rate': 8.54009503987253e-06, 'flow_matching_loss': 0.015476438403129577, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.76} +{'loss': 0.0128, 'grad_norm': 0.06602420657873154, 'learning_rate': 8.525176169541095e-06, 'flow_matching_loss': 0.012779364548623561, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.76} +{'loss': 0.0143, 'grad_norm': 0.06605277210474014, 'learning_rate': 8.510265166731151e-06, 'flow_matching_loss': 0.014344696793705225, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.76} +{'loss': 0.0139, 'grad_norm': 0.07499410957098007, 'learning_rate': 8.495362049560964e-06, 'flow_matching_loss': 0.013928807619959117, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.77} +{'loss': 0.0136, 'grad_norm': 0.08595845848321915, 'learning_rate': 8.480466836139241e-06, 'flow_matching_loss': 0.013621231261640786, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.77} +{'loss': 0.0135, 'grad_norm': 0.061808791011571884, 'learning_rate': 8.465579544565066e-06, 'flow_matching_loss': 0.013450589217245578, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.77} +{'loss': 0.0153, 'grad_norm': 0.10419429838657379, 'learning_rate': 8.450700192927897e-06, 'flow_matching_loss': 0.015342164039611816, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.78} 66%|██████▌ | 19850/30000 [12:29:27<6:18:06, 2.24s/it] 66%|██████▌ | 19851/30000 [12:29:28<5:24:59, 1.92s/it] 66%|██████▌ | 19852/30000 [12:29:29<4:46:43, 1.70s/it] 66%|██████▌ | 19853/30000 [12:29:31<4:20:22, 1.54s/it] 66%|██████▌ | 19854/30000 [12:29:32<4:03:11, 1.44s/it] 66%|██████▌ | 19855/30000 [12:29:33<3:50:40, 1.36s/it] 66%|██████▌ | 19856/30000 [12:29:36<4:52:21, 1.73s/it] 66%|██████▌ | 19857/30000 [12:29:40<7:29:23, 2.66s/it] 66%|██████▌ | 19858/30000 [12:29:42<6:13:29, 2.21s/it] 66%|██████▌ | 19859/30000 [12:29:43<5:21:58, 1.91s/it] 66%|██████▌ | 19860/30000 [12:29:44<4:44:58, 1.69s/it] 66%|██████▌ | 19860/30000 [12:29:44<4:44:58, 1.69s/it] 66%|██████▌ | 19861/30000 [12:29:45<4:18:42, 1.53s/it] 66%|██████▌ | 19862/30000 [12:29:46<4:02:15, 1.43s/it] 66%|██████▌ | 19863/30000 [12:29:47<3:50:14, 1.36s/it] 66%|██████▌ | 19864/30000 [12:29:49<4:19:37, 1.54s/it] 66%|██████▌ | 19865/30000 [12:29:54<7:05:22, 2.52s/it] 66%|██████▌ | 19866/30000 [12:29:55<5:56:29, 2.11s/it] 66%|██████▌ | 19867/30000 [12:29:57<5:12:39, 1.85s/it] 66%|██████▌ | 19868/30000 [12:29:58<4:37:50, 1.65s/it] 66%|██████▌ | 19869/30000 [12:29:59<4:14:09, 1.51s/it] 66%|██████▌ | 19870/30000 [12:30:00<3:58:32, 1.41s/it] 66%|██████▌ | 19870/30000 [12:30:00<3:58:32, 1.41s/it] 66%|██████▌ | 19871/30000 [12:30:01<3:47:18, 1.35s/it] 66%|██████▌ | 19872/30000 [12:30:03<4:20:09, 1.54s/it] 66%|██████▌ | 19873/30000 [12:30:08<7:04:15, 2.51s/it] 66%|██████▌ | 19874/30000 [12:30:09<5:55:52, 2.11s/it] 66%|██████▋ | 19875/30000 [12:30:10<5:08:47, 1.83s/it] 66%|██████▋ | 19876/30000 [12:30:12<4:34:41, 1.63s/it] 66%|██████▋ | 19877/30000 [12:30:13<4:11:26, 1.49s/it] 66%|██████▋ | 19878/30000 [12:30:14<3:55:43, 1.40s/it] 66%|██████▋ | 19879/30000 [12:30:15<3:44:48, 1.33s/it] 66%|██████▋ | 19880/30000 [12:30:18<5:01:22, 1.79s/it] 66%|██████▋ | 19880/30000 [12:30:18<5:01:22, 1.79s/it] 66%|██████▋ | 19881/30000 [12:30:22<7:12:31, 2.56s/it] 66%|██████▋ | 19882/30000 [12:30:24<6:01:11, 2.14s/it] 66%|██████▋ | 19883/30000 [12:30:25<5:12:57, 1.86s/it] 66%|██████▋ | 19884/30000 [12:30:26<4:37:54, 1.65s/it] 66%|██████▋ | 19885/30000 [12:30:27<4:13:45, 1.51s/it] 66%|██████▋ | 19886/30000 [12:30:28<3:57:49, 1.41s/it] 66%|██████▋ | 19887/30000 [12:30:29<3:46:28, 1.34s/it] 66%|██████▋ | 19888/30000 [12:30:32<5:04:29, 1.81s/it] 66%|██████▋ | 19889/30000 [12:30:37<7:25:06, 2.64s/it] 66%|██████▋ | 19890/30000 [12:30:38<6:12:47, 2.21s/it] 66%|██████▋ | 19890/30000 [12:30:38<6:12:47, 2.21s/it] 66%|██████▋ | 19891/30000 [12:30:39<5:21:32, 1.91s/it] 66%|██████▋ | 19892/30000 [12:30:41<4:44:15, 1.69s/it] 66%|██████▋ | 19893/30000 [12:30:42<4:18:13, 1.53s/it] 66%|██████▋ | 19894/30000 [12:30:43<4:00:57, 1.43s/it] 66%|██████▋ | 19895/30000 [12:30:44<3:48:22, 1.36s/it] 66%|██████▋ | 19896/30000 [12:30:46<4:40:19, 1.66s/it] 66%|██████▋ | 19897/30000 [12:30:51<6:55:11, 2.47s/it] 66%|██████▋ | 19898/30000 [12:30:52<5:49:15, 2.07s/it] 66%|██████▋ | 19899/30000 [12:30:53<5:05:08, 1.81s/it] 66%|██████▋ | 19900/30000 [12:30:54<4:32:59, 1.62s/it] 66%|██████▋ | 19900/30000 [12:30:54<4:32:59, 1.62s/it] 66%|██████▋ | 19901/30000 [12:30:55<4:10:10, 1.49s/it] 66%|██████▋ | 19902/30000 [12:30:57<3:54:48, 1.40s/it] 66%|██████▋ | 19903/30000 [12:30:58<3:44:44, 1.34s/it] 66%|██████▋ | 19904/30000 [12:31:00<4:47:33, 1.71s/it] 66%|██████▋ | 19905/30000 [12:31:05<7:03:21, 2.52s/it] 66%|██████▋ | 19906/30000 [12:31:06<5:55:50, 2.12s/it] 66%|██████▋ | 19907/30000 [12:31:07<5:10:49, 1.85s/it] 66%|██████▋ | 19908/30000 [12:31:08<4:37:24, 1.65s/it] 66%|██████▋ | 19909/30000 [12:31:10<4:13:16, 1.51s/it] 66%|██████▋ | 19910/30000 [12:31:11<3:56:36, 1.41s/it] 66%|██████▋ | 19910/30000 [12:31:11<3:56:36, 1.41s/it] 66%|██████▋ | 19911/30000 [12:31:12<3:45:58, 1.34s/it] 66%|██████▋ | 19912/30000 [12:31:16<5:43:42, 2.04s/it] 66%|██████▋ | 19913/30000 [12:31:19<7:10:12, 2.56s/it] 66%|██████▋ | 19914/30000 [12:31:21<6:00:38, 2.15s/it] 66%|██████▋ | 19915/30000 [12:31:22<5:13:39, 1.87s/it] 66%|██████▋ | 19916/30000 [12:31:23<4:38:43, 1.66s/it] 66%|██████▋ | 19917/30000 [12:31:24<4:13:24, 1.51s/it] 66%|██████▋ | 19918/30000 [12:31:25<3:57:41, 1.41s/it] 66%|██████▋ | 19919/30000 [12:31:27<3:45:49, 1.34s/it] 66%|██████▋ | 19920/30000 [12:31:31<6:09:59, 2.20s/it] 66%|██████▋ | 19920/30000 [12:31:31<6:09:59, 2.20s/it] 66%|██████▋ | 19921/30000 [12:31:33<6:38:44, 2.37s/it] 66%|██████▋ | 19922/30000 [12:31:35<5:38:16, 2.01s/it] 66%|██████▋ | 19923/30000 [12:31:36<4:56:55, 1.77s/it] 66%|██████▋ | 19924/30000 [12:31:37<4:26:40, 1.59s/it] 66%|██████▋ | 19925/30000 [12:31:38<4:05:58, 1.46s/it] 66%|██████▋ | 19926/30000 [12:31:39<3:51:07, 1.38s/it] 66%|██████▋ | 19927/30000 [12:31:41<3:42:10, 1.32s/it] 66%|██████▋ | 19928/30000 [12:31:45<6:18:45, 2.26s/it] 66%|██████▋ | 19929/30000 [12:31:47<6:28:41, 2.32s/it] 66%|██████▋ | 19930/30000 [12:31:49<5:31:02, 1.97s/it] 66%|██████▋ | 19930/30000 [12:31:49<5:31:02, 1.97s/it] 66%|██████▋ | 19931/30000 [12:31:50<4:51:45, 1.74s/it] 66%|██████▋ | 19932/30000 [12:31:51<4:23:04, 1.57s/it] 66%|██████▋ | 19933/30000 [12:31:52<4:03:23, 1.45s/it] 66%|██████▋ | 19934/30000 [12:31:53<3:50:19, 1.37s/it] 66%|██████▋ | 19935/30000 [12:31:55<3:40:51, 1.32s/it] 66%|██████▋ | 19936/30000 [12:31:59<6:31:26, 2.33s/it] 66%|██████▋ | 19937/30000 [12:32:02<6:38:25, 2.38s/it] 66%|██████▋ | 19938/30000 [12:32:03<5:37:38, 2.01s/it] 66%|██████▋ | 19939/30000 [12:32:04<4:56:30, 1.77s/it] 66%|██████▋ | 19940/30000 [12:32:05<4:26:13, 1.59s/it] 66%|██████▋ | 19940/30000 [12:32:05<4:26:13, 1.59s/it] 66%|██████▋ | 19941/30000 [12:32:06<4:04:30, 1.46s/it] 66%|██████▋ | 19942/30000 [12:32:08<3:50:41, 1.38s/it] 66%|██████▋ | 19943/30000 [12:32:09<3:40:52, 1.32s/it] 66%|██████▋ | 19944/30000 [12:32:13<6:02:55, 2.17s/it] 66%|██████▋ | 19945/30000 [12:32:16<6:32:30, 2.34s/it] 66%|██████▋ | 19946/30000 [12:32:17<5:33:48, 1.99s/it] 66%|██████▋ | 19947/30000 [12:32:18<4:54:04, 1.76s/it] 66%|██████▋ | 19948/30000 [12:32:19<4:25:06, 1.58s/it] 66%|██████▋ | 19949/30000 [12:32:20<4:04:32, 1.46s/it] 66%|██████▋ | 19950/30000 [12:32:22<3:50:31, 1.38s/it] 66%|██████▋ | 19950/30000 [12:32:22<3:50:31, 1.38s/it] 67%|██████▋ | 19951/30000 [12:32:23<3:41:32, 1.32s/it] 67%|██████▋ | 19952/30000 [12:32:27<6:11:32, 2.22s/it] 67%|██████▋ | 19953/30000 [12:32:29<6:13:10, 2.23s/it] 67%|██████▋ | 19954/30000 [12:32:31<5:20:33, 1.91s/it] 67%|██████▋ | 19955/30000 [12:32:32<4:44:31, 1.70s/it] 67%|██████▋ | 19956/30000 [12:32:33<4:18:17, 1.54s/it] 67%|██████▋ | 19957/30000 [12:32:34<3:59:21, 1.43s/it] 67%|██████▋ | 19958/30000 [12:32:35<3:46:10, 1.35s/it] 67%|██████▋ | 19959/30000 [12:32:36<3:37:40, 1.30s/it] 67%|██████▋ | 19960/30000 [12:32:40<5:53:10, 2.11s/it] 67%|██████▋ | 19960/30000 [12:32:40<5:53:10, 2.11s/it] 67%|██████▋ | 19961/30000 [12:32:43<6:26:59, 2.31s/it] 67%|██████▋ | 19962/30000 [12:32:44<5:30:18, 1.97s/it] 67%|██████▋ | 19963/30000 [12:32:46<4:51:58, 1.75s/it] 67%|██████▋ | 19964/30000 [12:32:47<4:22:59, 1.57s/it] 67%|██████▋ | 19965/30000 [12:32:48<4:03:21, 1.46s/it] 67%|██████▋ | 19966/30000 [12:32:49<3:49:52, 1.37s/it] 67%|██████▋ | 19967/30000 [12:32:50<3:41:17, 1.32s/it] 67%|██████▋ | 19968/30000 [12:32:55<6:18:13, 2.26s/it] 67%|██████▋ | 19969/30000 [12:32:57<6:07:59, 2.20s/it] 67%|██████▋ | 19970/30000 [12:32:58<5:16:47, 1.90s/it] 67%|██████▋ | 19970/30000 [12:32:58<5:16:47, 1.90s/it] 67%|██████▋ | 19971/30000 [12:32:59<4:41:59, 1.69s/it] 67%|██████▋ | 19972/30000 [12:33:00<4:16:24, 1.53s/it] 67%|██████▋ | 19973/30000 [12:33:02<3:58:26, 1.43s/it] 67%|██████▋ | 19974/30000 [12:33:03<3:45:04, 1.35s/it] 67%|██████▋ | 19975/30000 [12:33:04<3:37:10, 1.30s/it] 67%|██████▋ | 19976/30000 [12:33:09<7:02:34, 2.53s/it] 67%|██████▋ | 19977/30000 [12:33:11<6:24:24, 2.30s/it] 67%|██████▋ | 19978/30000 [12:33:12<5:27:12, 1.96s/it] 67%|██████▋ | 19979/30000 [12:33:13<4:49:09, 1.73s/it] 67%|██████▋ | 19980/30000 [12:33:15<4:21:42, 1.57s/it] 67%|██████▋ | 19980/30000 [12:33:15<4:21:42, 1.57s/it] 67%|██████▋ | 19981/30000 [12:33:16<4:01:52, 1.45s/it] 67%|██████▋ | 19982/30000 [12:33:17<3:47:44, 1.36s/it] 67%|██████▋ | 19983/30000 [12:33:18<3:37:40, 1.30s/it] 67%|██████▋ | 19984/30000 [12:33:23<6:54:50, 2.49s/it] 67%|██████▋ | 19985/30000 [12:33:25<6:32:19, 2.35s/it] 67%|██████▋ | 19986/30000 [12:33:27<5:33:38, 2.00s/it] 67%|██████▋ | 19987/30000 [12:33:28<4:52:58, 1.76s/it] 67%|██████▋ | 19988/30000 [12:33:29<4:24:06, 1.58s/it] 67%|██████▋ | 19989/30000 [12:33:30<4:04:14, 1.46s/it] 67%|██████▋ | 19990/30000 [12:33:31<3:50:05, 1.38s/it] 67%|██████▋ | 19990/30000 [12:33:31<3:50:05, 1.38s/it] 67%|██████▋ | 19991/30000 [12:33:33<3:40:13, 1.32s/it] 67%|██████▋ | 19992/30000 [12:33:38<7:20:30, 2.64s/it] 67%|██████▋ | 19993/30000 [12:33:40<6:47:28, 2.44s/it] 67%|██████▋ | 19994/30000 [12:33:41<5:44:12, 2.06s/it] 67%|██████▋ | 19995/30000 [12:33:43<5:00:50, 1.80s/it] 67%|██████▋ | 19996/30000 [12:33:44<4:28:41, 1.61s/it] 67%|██████▋ | 19997/30000 [12:33:45<4:06:07, 1.48s/it] 67%|██████▋ | 19998/30000 [12:33:46<3:50:33, 1.38s/it] 67%|██████▋ | 19999/30000 [12:33:47<3:40:42, 1.32s/it] 67%|██████▋ | 20000/30000 [12:33:53<7:31:01, 2.71s/it] 67%|██████▋ | 20000/30000 [12:33:53<7:31:01, 2.71s/it] 67%|██████▋ | 20001/30000 [12:33:55<6:19:26, 2.28s/it] 67%|██████▋ | 20002/30000 [12:33:56<5:23:55, 1.94s/it] 67%|██████▋ | 20003/30000 [12:33:57<4:46:27, 1.72s/it] 67%|██████▋ | 20004/30000 [12:33:58<4:18:31, 1.55s/it] 67%|██████▋ | 20005/30000 [12:33:59<3:58:50, 1.43s/it] 67%|██████▋ | 20006/30000 [12:34:00<3:46:29, 1.36s/it] 67%|██████▋ | 20007/30000 [12:34:02<3:37:37, 1.31s/it] 67%|██████▋ | 20008/30000 [12:34:08<7:31:21, 2.71s/it] 67%|██████▋ | 20009/30000 [12:34:09<6:22:53, 2.30s/it] 67%|██████▋ | 20010/30000 [12:34:10<5:27:49, 1.97s/it] 67%|██████▋ | 20010/30000 [12:34:10<5:27:49, 1.97s/it] 67%|██████▋ | 20011/30000 [12:34:11<4:48:57, 1.74s/it] 67%|██████▋ | 20012/30000 [12:34:12<4:20:01, 1.56s/it] 67%|██████▋ | 20013/30000 [12:34:14<4:00:48, 1.45s/it] 67%|██████▋ | 20014/30000 [12:34:15<3:48:02, 1.37s/it] 67%|██████▋ | 20015/30000 [12:34:16<3:38:51, 1.32s/it] 67%|██████▋ | 20016/30000 [12:34:21<6:33:59, 2.37s/it] 67%|██████▋ | 20017/30000 [12:34:23<6:15:46, 2.26s/it] 67%|██████▋ | 20018/30000 [12:34:24<5:21:15, 1.93s/it] 67%|██████▋ | 20019/30000 [12:34:25<4:44:18, 1.71s/it] 67%|██████▋ | 20020/30000 [12:34:26<4:17:23, 1.55s/it] 67%|██████▋ | 20020/30000 [12:34:26<4:17:23, 1.55s/it] 67%|██████▋ | 20021/30000 [12:34:28<3:58:40, 1.44s/it] 67%|██████▋ | 20022/30000 [12:34:29<3:46:43, 1.36s/it] 67%|██████▋ | 20023/30000 [12:34:30<3:37:48, 1.31s/it] 67%|██████▋ | 20024/30000 [12:34:36<7:18:55, 2.64s/it] 67%|██████▋ | 20025/30000 [12:34:37<6:32:21, 2.36s/it] 67%|██████▋ | 20026/30000 [12:34:39<5:33:12, 2.00s/it] 67%|██████▋ | 20027/30000 [12:34:40<4:53:27, 1.77s/it] 67%|██████▋ | 20028/30000 [12:34:41<4:23:32, 1.59s/it] 67%|██████▋ | 20029/30000 [12:34:42<4:03:06, 1.46s/it] 67%|██████▋ | 20030/30000 [12:34:43<3:48:24, 1.37s/it] 67%|██████▋ | 20030/30000 [12:34:43<3:48:24, 1.37s/it] 67%|██████▋ | 20031/30000 [12:34:44<3:39:05, 1.32s/it] 67%|██████▋ | 20032/30000 [12:34:52<8:39:11, 3.13s/it] 67%|██████▋ | 20033/30000 [12:34:53<7:02:04, 2.54s/it] 67%|██████▋ | 20034/30000 [12:34:54<5:53:58, 2.13s/it] 67%|██████▋ | 20035/30000 [12:34:55<5:07:24, 1.85s/it] 67%|██████▋ | 20036/30000 [12:34:56<4:33:38, 1.65s/it] 67%|██████▋ | 20037/30000 [12:34:58<4:09:50, 1.50s/it] 67%|██████▋ | 20038/30000 [12:34:59<3:54:58, 1.42s/it] 67%|██████▋ | 20039/30000 [12:35:00<3:43:27, 1.35s/it] 67%|██████▋ | 20040/30000 [12:35:06<7:29:50, 2.71s/it] 67%|██████▋ | 20040/30000 [12:35:06<7:29:50, 2.71s/it] 67%|██████▋ | 20041/30000 [12:35:07<6:25:25, 2.32s/it] 67%|██████▋ | 20042/30000 [12:35:09<5:28:19, 1.98s/it] 67%|██████▋ | 20043/30000 [12:35:10<4:49:01, 1.74s/it] 67%|██████▋ | 20044/30000 [12:35:11<4:20:11, 1.57s/it] 67%|██████▋ | 20045/30000 [12:35:12<4:00:35, 1.45s/it] 67%|██████▋ | 20046/30000 [12:35:13<3:47:42, 1.37s/it] 67%|██████▋ | 20047/30000 [12:35:14<3:37:25, 1.31s/it] 67%|██████▋ | 20048/30000 [12:35:20<7:32:46, 2.73s/it] 67%|██████▋ | 20049/30000 [12:35:22<6:15:36, 2.26s/it] 67%|██████▋ | 20050/30000 [12:35:23<5:20:58, 1.94s/it] 67%|██████▋ | 20050/30000 [12:35:23<5:20:58, 1.94s/it] 67%|██████▋ | 20051/30000 [12:35:24<4:44:33, 1.72s/it] 67%|██████▋ | 20052/30000 [12:35:25<4:18:15, 1.56s/it] 67%|██████▋ | 20053/30000 [12:35:26<3:58:31, 1.44s/it] 67%|██████▋ | 20054/30000 [12:35:28<3:45:57, 1.36s/it] 67%|██████▋ | 20055/30000 [12:35:29<3:36:14, 1.30s/it] 67%|██████▋ | 20056/30000 [12:35:35<7:34:24, 2.74s/it] 67%|██████▋ | 20057/30000 [12:35:36<6:17:01, 2.28s/it] 67%|██████▋ | 20058/30000 [12:35:37<5:21:51, 1.94s/it] 67%|██████▋ | 20059/30000 [12:35:38<4:44:31, 1.72s/it] 67%|██████▋ | 20060/30000 [12:35:40<4:17:03, 1.55s/it] 67%|██████▋ | 20060/30000 [12:35:40<4:17:03, 1.55s/it] 67%|██████▋ | 20061/30000 [12:35:41<3:57:43, 1.44s/it] 67%|██████▋ | 20062/30000 [12:35:42<3:45:31, 1.36s/it] 67%|██████▋ | 20063/30000 [12:35:43<3:36:22, 1.31s/it] 67%|██████▋ | 20064/30000 [12:35:48<6:58:39, 2.53s/it] 67%|██████▋ | 20065/30000 [12:35:50<6:05:09, 2.21s/it] 67%|██████▋ | 20066/30000 [12:35:51<5:13:15, 1.89s/it] 67%|██████▋ | 20067/30000 [12:35:52<4:38:33, 1.68s/it] 67%|██████▋ | 20068/30000 [12:35:53<4:13:22, 1.53s/it] 67%|██████▋ | 20069/30000 [12:35:55<3:56:13, 1.43s/it] 67%|██████▋ | 20070/30000 [12:35:56<3:44:30, 1.36s/it] 67%|██████▋ | 20070/30000 [12:35:56<3:44:30, 1.36s/it] 67%|██████▋ | 20071/30000 [12:35:57<3:36:35, 1.31s/it] 67%|██████▋ | 20072/30000 [12:36:02<6:28:39, 2.35s/it] 67%|██████▋ | 20073/30000 [12:36:04<6:07:12, 2.22s/it] 67%|██████▋ | 20074/30000 [12:36:05<5:15:14, 1.91s/it] 67%|██████▋ | 20075/30000 [12:36:06<4:40:35, 1.70s/it] 67%|██████▋ | 20076/30000 [12:36:07<4:14:34, 1.54s/it] 67%|██████▋ | 20077/30000 [12:36:08<3:56:21, 1.43s/it] 67%|██████▋ | 20078/30000 [12:36:10<3:44:36, 1.36s/it] 67%|██████▋ | 20079/30000 [12:36:11<3:35:32, 1.30s/it] 67%|██████▋ | 20080/30000 [12:36:16<6:51:41, 2.49s/it] 67%|██████▋ | 20080/30000 [12:36:16<6:51:41, 2.49s/it] 67%|██████▋ | 20081/30000 [12:36:18<6:12:44, 2.25s/it] 67%|██████▋ | 20082/30000 [12:36:19<5:20:29, 1.94s/it] 67%|██████▋ | 20083/30000 [12:36:20<4:43:14, 1.71s/it] 67%|██████▋ | 20084/30000 [12:36:21<4:16:11, 1.55s/it] 67%|██████▋ | 20085/30000 [12:36:22<3:57:17, 1.44s/it] 67%|██████▋ | 20086/30000 [12:36:24<3:45:50, 1.37s/it] 67%|██████▋ | 20087/30000 [12:36:25<3:36:59, 1.31s/it] 67%|██████▋ | 20088/30000 [12:36:31<7:19:22, 2.66s/it] 67%|██████▋ | 20089/30000 [12:36:32<6:06:05, 2.22s/it] 67%|██████▋ | 20090/30000 [12:36:33<5:14:23, 1.90s/it] 67%|██████▋ | 20090/30000 [12:36:33<5:14:23, 1.90s/it] 67%|██████▋ | 20091/30000 [12:36:34<4:38:59, 1.69s/it] 67%|██████▋ | 20092/30000 [12:36:35<4:12:34, 1.53s/it] 67%|██████▋ | 20093/30000 [12:36:37<3:55:26, 1.43s/it] 67%|██████▋ | 20094/30000 [12:36:38<3:43:43, 1.36s/it] 67%|██████▋ | 20095/30000 [12:36:39<3:35:20, 1.30s/it] 67%|██████▋ | 20096/30000 [12:36:44<6:36:46, 2.40s/it] 67%|██████▋ | 20097/30000 [12:36:46<6:00:40, 2.19s/it] 67%|██████▋ | 20098/30000 [12:36:47<5:10:47, 1.88s/it] 67%|██████▋ | 20099/30000 [12:36:48<4:37:16, 1.68s/it] 67%|██████▋ | 20100/30000 [12:36:49<4:13:11, 1.53s/it] 67%|██████▋ | 20100/30000 [12:36:49<4:13:11, 1.53s/it] 67%|██████▋ | 20101/30000 [12:36:50<3:55:21, 1.43s/it] 67%|██████▋ | 20102/30000 [12:36:52<3:43:47, 1.36s/it] 67%|██████▋ | 20103/30000 [12:36:53<3:35:08, 1.30s/it] 67%|██████▋ | 20104/30000 [12:36:58<7:17:25, 2.65s/it] 67%|██████▋ | 20105/30000 [12:37:00<6:29:30, 2.36s/it] 67%|██████▋ | 20106/30000 [12:37:01<5:30:34, 2.00s/it] 67%|██████▋ | 20107/30000 [12:37:03<4:50:39, 1.76s/it] 67%|██████▋ | 20108/30000 [12:37:04<4:20:52, 1.58s/it] 67%|██████▋ | 20109/30000 [12:37:05<4:00:09, 1.46s/it] 67%|██████▋ | 20110/30000 [12:37:06<3:46:42, 1.38s/it] 67%|██████▋ | 20110/30000 [12:37:06<3:46:42, 1.38s/it] 67%|██████▋ | 20111/30000 [12:37:07<3:40:02, 1.34s/it] 67%|██████▋ | 20112/30000 [12:37:13<6:58:53, 2.54s/it] 67%|██████▋ | 20113/30000 [12:37:15<6:29:38, 2.36s/it] 67%|██████▋ | 20114/30000 [12:37:16<5:30:30, 2.01s/it] 67%|██████▋ | 20115/30000 [12:37:17<4:50:33, 1.76s/it] 67%|██████▋ | 20116/30000 [12:37:18<4:21:01, 1.58s/it] 67%|██████▋ | 20117/30000 [12:37:19<4:02:49, 1.47s/it] 67%|██████▋ | 20118/30000 [12:37:21<3:48:59, 1.39s/it] 67%|██████▋ | 20119/30000 [12:37:22<3:39:01, 1.33s/it] 67%|██████▋ | 20120/30000 [12:37:26<5:48:55, 2.12s/it] 67%|██████▋ | 20120/30000 [12:37:26<5:48:55, 2.12s/it] 67%|██████▋ | 20121/30000 [12:37:28<5:47:58, 2.11s/it] 67%|██████▋ | 20122/30000 [12:37:29<5:01:24, 1.83s/it] 67%|██████▋ | 20123/30000 [12:37:30<4:29:52, 1.64s/it] 67%|██████▋ | 20124/30000 [12:37:31<4:06:52, 1.50s/it] 67%|██████▋ | 20125/30000 [12:37:33<3:50:44, 1.40s/it] 67%|██████▋ | 20126/30000 [12:37:34<3:40:18, 1.34s/it] 67%|██████▋ | 20127/30000 [12:37:35<3:33:01, 1.29s/it] 67%|██████▋ | 20128/30000 [12:37:39<5:55:44, 2.16s/it] 67%|██████▋ | 20129/30000 [12:37:41<6:06:31, 2.23s/it] 67%|██████▋ | 20130/30000 [12:37:43<5:15:01, 1.92s/it] 67%|██████▋ | 20130/30000 [12:37:43<5:15:01, 1.92s/it] 67%|██████▋ | 20131/30000 [12:37:44<4:40:05, 1.70s/it] 67%|██████▋ | 20132/30000 [12:37:45<4:14:07, 1.55s/it] 67%|██████▋ | 20133/30000 [12:37:46<3:55:03, 1.43s/it] 67%|██████▋ | 20134/30000 [12:37:47<3:43:42, 1.36s/it] 67%|██████▋ | 20135/30000 [12:37:49<3:34:59, 1.31s/it] 67%|██████▋ | 20136/30000 [12:37:53<6:23:30, 2.33s/it] 67%|██████▋ | 20137/30000 [12:37:55<6:06:44, 2.23s/it] 67%|██████▋ | 20138/30000 [12:37:56<5:14:41, 1.91s/it] 67%|██████▋ | 20139/30000 [12:37:58<4:39:15, 1.70s/it] 67%|██████▋ | 20140/30000 [12:37:59<4:13:08, 1.54s/it] 67%|██████▋ | 20140/30000 [12:37:59<4:13:08, 1.54s/it] 67%|██████▋ | 20141/30000 [12:38:00<3:56:47, 1.44s/it] 67%|██████▋ | 20142/30000 [12:38:01<3:44:42, 1.37s/it] 67%|██████▋ | 20143/30000 [12:38:02<3:35:19, 1.31s/it] 67%|██████▋ | 20144/30000 [12:38:07<6:29:45, 2.37s/it] 67%|██████▋ | 20145/30000 [12:38:09<5:59:54, 2.19s/it] 67%|██████▋ | 20146/30000 [12:38:10<5:09:35, 1.89s/it] 67%|██████▋ | 20147/30000 [12:38:11<4:35:18, 1.68s/it] 67%|██████▋ | 20148/30000 [12:38:13<4:09:56, 1.52s/it] 67%|██████▋ | 20149/30000 [12:38:14<3:52:30, 1.42s/it] 67%|██████▋ | 20150/30000 [12:38:15<3:41:04, 1.35s/it] 67%|██████▋ | 20150/30000 [12:38:15<3:41:04, 1.35s/it] 67%|██████▋ | 20151/30000 [12:38:16<3:32:44, 1.30s/it] 67%|██████▋ | 20152/30000 [12:38:20<6:04:41, 2.22s/it] 67%|██████▋ | 20153/30000 [12:38:23<5:57:23, 2.18s/it] 67%|██████▋ | 20154/30000 [12:38:24<5:07:00, 1.87s/it] 67%|██████▋ | 20155/30000 [12:38:25<4:33:30, 1.67s/it] 67%|██████▋ | 20156/30000 [12:38:26<4:09:27, 1.52s/it] 67%|██████▋ | 20157/30000 [12:38:27<3:52:59, 1.42s/it] 67%|██████▋ | 20158/30000 [12:38:28<3:41:58, 1.35s/it] 67%|██████▋ | 20159/30000 [12:38:30<3:33:29, 1.30s/it] 67%|██████▋ | 20160/30000 [12:38:34<5:58:01, 2.18s/it] 67%|██████▋ | 20160/30000 [12:38:34<5:58:01, 2.18s/it] 67%|██████▋ | 20161/30000 [12:38:36<6:09:25, 2.25s/it] 67%|██████▋ | 20162/30000 [12:38:37<5:15:36, 1.92s/it] 67%|██████▋ | 20163/30000 [12:38:39<4:38:56, 1.70s/it] 67%|██████▋ | 20164/30000 [12:38:40<4:12:53, 1.54s/it] 67%|██████▋ | 20165/30000 [12:38:41<3:55:09, 1.43s/it] 67%|██████▋ | 20166/30000 [12:38:42<3:44:07, 1.37s/it] 67%|██████▋ | 20167/30000 [12:38:43<3:35:53, 1.32s/it] 67%|██████▋ | 20168/30000 [12:38:48<6:00:45, 2.20s/it] 67%|██████▋ | 20169/30000 [12:38:51<6:41:38, 2.45s/it] 67%|██████▋ | 20170/30000 [12:38:52<5:38:48, 2.07s/it] 67%|██████▋ | 20170/30000 [12:38:52<5:38:48, 2.07s/it] 67%|██████▋ | 20171/30000 [12:38:53<4:55:30, 1.80s/it] 67%|██████▋ | 20172/30000 [12:38:54<4:24:28, 1.61s/it] 67%|██████▋ | 20173/30000 [12:38:55<4:02:49, 1.48s/it] 67%|██████▋ | 20174/30000 [12:38:57<3:49:05, 1.40s/it] 67%|██████▋ | 20175/30000 [12:38:58<3:38:49, 1.34s/it] 67%|██████▋ | 20176/30000 [12:39:02<6:19:23, 2.32s/it] 67%|██████▋ | 20177/30000 [12:39:05<6:49:28, 2.50s/it] 67%|██████▋ | 20178/30000 [12:39:07<5:43:31, 2.10s/it] 67%|██████▋ | 20179/30000 [12:39:08<4:58:42, 1.82s/it] 67%|██████▋ | 20180/30000 [12:39:09<4:27:47, 1.64s/it] 67%|██████▋ | 20180/30000 [12:39:09<4:27:47, 1.64s/it] 67%|██████▋ | 20181/30000 [12:39:10<4:04:58, 1.50s/it] 67%|██████▋ | 20182/30000 [12:39:11<3:49:43, 1.40s/it] 67%|██████▋ | 20183/30000 [12:39:12<3:39:32, 1.34s/it] 67%|██████▋ | 20184/30000 [12:39:17<6:01:37, 2.21s/it] 67%|██████▋ | 20185/30000 [12:39:19<5:59:20, 2.20s/it] 67%|██████▋ | 20186/30000 [12:39:20<5:08:38, 1.89s/it] 67%|██████▋ | 20187/30000 [12:39:21<4:36:13, 1.69s/it] 67%|██████▋ | 20188/30000 [12:39:22<4:11:30, 1.54s/it] 67%|██████▋ | 20189/30000 [12:39:24<3:53:19, 1.43s/it] 67%|██████▋ | 20190/30000 [12:39:25<3:43:20, 1.37s/it] 67%|██████▋ | 20190/30000 [12:39:25<3:43:20, 1.37s/it] 67%|██████▋ | 20191/30000 [12:39:26<3:34:22, 1.31s/it] 67%|██████▋ | 20192/30000 [12:39:30<5:56:11, 2.18s/it] 67%|██████▋ | 20193/30000 [12:39:33<6:06:22, 2.24s/it] 67%|██████▋ | 20194/30000 [12:39:34<5:14:13, 1.92s/it] 67%|██████▋ | 20195/30000 [12:39:35<4:38:42, 1.71s/it] 67%|██████▋ | 20196/30000 [12:39:36<4:12:32, 1.55s/it] 67%|██████▋ | 20197/30000 [12:39:37<3:54:00, 1.43s/it] 67%|██████▋ | 20198/30000 [12:39:39<3:42:37, 1.36s/it] 67%|██████▋ | 20199/30000 [12:39:40<3:34:19, 1.31s/it] 67%|██████▋ | 20200/30000 [12:39:44<6:16:44, 2.31s/it] 67%|██████▋ | 20200/30000 [12:39:44<6:16:44, 2.31s/it] 67%|██████▋ | 20201/30000 [12:39:46<6:01:41, 2.21s/it] 67%|██████▋ | 20202/30000 [12:39:48<5:10:51, 1.90s/it] 67%|██████▋ | 20203/30000 [12:39:49<4:35:24, 1.69s/it] 67%|██████▋ | 20204/30000 [12:39:50<4:10:51, 1.54s/it] 67%|██████▋ | 20205/30000 [12:39:51<3:53:21, 1.43s/it] 67%|██████▋ | 20206/30000 [12:39:52<3:41:53, 1.36s/it] 67%|██████▋ | 20207/30000 [12:39:53<3:34:02, 1.31s/it] 67%|██████▋ | 20208/30000 [12:39:59<6:41:39, 2.46s/it] 67%|██████▋ | 20209/30000 [12:40:01<6:30:13, 2.39s/it] 67%|██████▋ | 20210/30000 [12:40:02<5:30:07, 2.02s/it] 67%|██████▋ | 20210/30000 [12:40:02<5:30:07, 2.02s/it] 67%|██████▋ | 20211/30000 [12:40:03<4:49:04, 1.77s/it] 67%|██████▋ | 20212/30000 [12:40:04<4:19:45, 1.59s/it] 67%|██████▋ | 20213/30000 [12:40:05<3:58:24, 1.46s/it] 67%|██████▋ | 20214/30000 [12:40:07<3:44:55, 1.38s/it] 67%|██████▋ | 20215/30000 [12:40:08<3:35:50, 1.32s/it] 67%|██████▋ | 20216/30000 [12:40:13<6:54:02, 2.54s/it] 67%|██████▋ | 20217/30000 [12:40:15<6:22:50, 2.35s/it] 67%|██████▋ | 20218/30000 [12:40:16<5:25:32, 2.00s/it] 67%|██████▋ | 20219/30000 [12:40:18<4:45:58, 1.75s/it] 67%|██████▋ | 20220/30000 [12:40:19<4:17:41, 1.58s/it] 67%|██████▋ | 20220/30000 [12:40:19<4:17:41, 1.58s/it] 67%|██████▋ | 20221/30000 [12:40:20<3:57:38, 1.46s/it] 67%|██████▋ | 20222/30000 [12:40:21<3:44:05, 1.38s/it] 67%|██████▋ | 20223/30000 [12:40:22<3:35:15, 1.32s/it] 67%|██████▋ | 20224/30000 [12:40:28<6:58:55, 2.57s/it] 67%|██████▋ | 20225/30000 [12:40:30<6:24:24, 2.36s/it] 67%|██████▋ | 20226/30000 [12:40:31<5:26:12, 2.00s/it] 67%|██████▋ | 20227/30000 [12:40:32<4:46:27, 1.76s/it] 67%|██████▋ | 20228/30000 [12:40:33<4:17:46, 1.58s/it] 67%|██████▋ | 20229/30000 [12:40:34<3:56:57, 1.46s/it] 67%|██████▋ | 20230/30000 [12:40:35<3:43:25, 1.37s/it] 67%|██████▋ | 20230/30000 [12:40:35<3:43:25, 1.37s/it] 67%|██████▋ | 20231/30000 [12:40:37<3:34:39, 1.32s/it] 67%|██████▋ | 20232/30000 [12:40:42<6:55:25, 2.55s/it] 67%|██████▋ | 20233/30000 [12:40:45<6:52:37, 2.53s/it] 67%|██████▋ | 20234/30000 [12:40:46<5:45:40, 2.12s/it] 67%|██████▋ | 20235/30000 [12:40:47<4:59:13, 1.84s/it] 67%|██████▋ | 20236/30000 [12:40:48<4:26:35, 1.64s/it] 67%|██████▋ | 20237/30000 [12:40:49<4:03:41, 1.50s/it] 67%|██████▋ | 20238/30000 [12:40:50<3:48:17, 1.40s/it] 67%|██████▋ | 20239/30000 [12:40:52<3:37:57, 1.34s/it] 67%|██████▋ | 20240/30000 [12:40:56<6:17:13, 2.32s/it] 67%|██████▋ | 20240/30000 [12:40:56<6:17:13, 2.32s/it] 67%|██████▋ | 20241/30000 [12:40:58<6:08:08, 2.26s/it] 67%|██████▋ | 20242/30000 [12:41:00<5:16:38, 1.95s/it] 67%|██████▋ | 20243/30000 [12:41:01<4:39:58, 1.72s/it] 67%|██████▋ | 20244/30000 [12:41:02<4:12:01, 1.55s/it] 67%|██████▋ | 20245/30000 [12:41:03<3:53:17, 1.43s/it] 67%|██████▋ | 20246/30000 [12:41:04<3:41:00, 1.36s/it] 67%|██████▋ | 20247/30000 [12:41:05<3:32:41, 1.31s/it] 67%|██████▋ | 20248/30000 [12:41:11<6:34:51, 2.43s/it] 67%|██████▋ | 20249/30000 [12:41:13<6:42:08, 2.47s/it] 68%|██████▊ | 20250/30000 [12:41:14<5:40:27, 2.10s/it] 68%|██████▊ | 20250/30000 [12:41:14<5:40:27, 2.10s/it] 68%|██████▊ | 20251/30000 [12:41:15<4:56:20, 1.82s/it] 68%|██████▊ | 20252/30000 [12:41:17<4:24:13, 1.63s/it] 68%|██████▊ | 20253/30000 [12:41:18<4:01:39, 1.49s/it] 68%|██████▊ | 20254/30000 [12:41:19<3:46:52, 1.40s/it] 68%|██████▊ | 20255/30000 [12:41:20<3:36:02, 1.33s/it] 68%|██████▊ | 20256/30000 [12:41:25<6:13:51, 2.30s/it] 68%|██████▊ | 20257/30000 [12:41:27<5:55:45, 2.19s/it] 68%|██████▊ | 20258/30000 [12:41:28<5:07:13, 1.89s/it] 68%|██████▊ | 20259/30000 [12:41:29<4:33:11, 1.68s/it] 68%|██████▊ | 20260/30000 [12:41:30<4:08:11, 1.53s/it] 68%|██████▊ | 20260/30000 [12:41:30<4:08:11, 1.53s/it] 68%|██████▊ | 20261/30000 [12:41:31<3:50:36, 1.42s/it] 68%|██████▊ | 20262/30000 [12:41:33<3:39:54, 1.35s/it] 68%|██████▊ | 20263/30000 [12:41:34<3:32:05, 1.31s/it] 68%|██████▊ | 20264/30000 [12:41:40<7:27:05, 2.76s/it] 68%|██████▊ | 20265/30000 [12:41:41<6:16:08, 2.32s/it] 68%|██████▊ | 20266/30000 [12:41:42<5:20:19, 1.97s/it] 68%|██████▊ | 20267/30000 [12:41:44<4:42:37, 1.74s/it] 68%|██████▊ | 20268/30000 [12:41:45<4:14:52, 1.57s/it] 68%|██████▊ | 20269/30000 [12:41:46<3:56:11, 1.46s/it] 68%|██████▊ | 20270/30000 [12:41:47<3:43:46, 1.38s/it] +{'loss': 0.0135, 'grad_norm': 0.061743445694446564, 'learning_rate': 8.435828799307551e-06, 'flow_matching_loss': 0.01352948397397995, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.78} +{'loss': 0.0133, 'grad_norm': 0.07601521909236908, 'learning_rate': 8.420965381774172e-06, 'flow_matching_loss': 0.013265693560242653, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.78} +{'loss': 0.0136, 'grad_norm': 0.09289446473121643, 'learning_rate': 8.406109958388222e-06, 'flow_matching_loss': 0.013604666013270617, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.78} +{'loss': 0.0134, 'grad_norm': 0.06888096034526825, 'learning_rate': 8.391262547200434e-06, 'flow_matching_loss': 0.013449338171631098, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.79} +{'loss': 0.0147, 'grad_norm': 0.08243697136640549, 'learning_rate': 8.376423166251816e-06, 'flow_matching_loss': 0.014733711257576942, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.79} +{'loss': 0.0142, 'grad_norm': 0.07165886461734772, 'learning_rate': 8.361591833573612e-06, 'flow_matching_loss': 0.014213358238339423, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.79} +{'loss': 0.0145, 'grad_norm': 0.06593390554189682, 'learning_rate': 8.346768567187287e-06, 'flow_matching_loss': 0.014538467675447465, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.8} +{'loss': 0.0135, 'grad_norm': 0.06419916450977325, 'learning_rate': 8.331953385104516e-06, 'flow_matching_loss': 0.0134547533467412, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.8} +{'loss': 0.0134, 'grad_norm': 0.06922859698534012, 'learning_rate': 8.317146305327134e-06, 'flow_matching_loss': 0.013423800468444824, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.8} +{'loss': 0.0141, 'grad_norm': 0.07326412200927734, 'learning_rate': 8.302347345847146e-06, 'flow_matching_loss': 0.014095025975257159, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.8} +{'loss': 0.0145, 'grad_norm': 0.06709704548120499, 'learning_rate': 8.287556524646683e-06, 'flow_matching_loss': 0.01449103830382228, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.81} +{'loss': 0.013, 'grad_norm': 0.056809037923812866, 'learning_rate': 8.272773859697978e-06, 'flow_matching_loss': 0.012989426264539361, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.81} +{'loss': 0.0132, 'grad_norm': 0.09189996868371964, 'learning_rate': 8.257999368963374e-06, 'flow_matching_loss': 0.013232431933283806, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.81} +{'loss': 0.0135, 'grad_norm': 0.07713286578655243, 'learning_rate': 8.243233070395266e-06, 'flow_matching_loss': 0.0134742833673954, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.82} +{'loss': 0.0139, 'grad_norm': 0.07694512605667114, 'learning_rate': 8.2284749819361e-06, 'flow_matching_loss': 0.013946933019906283, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.82} +{'loss': 0.0148, 'grad_norm': 0.07786765694618225, 'learning_rate': 8.213725121518346e-06, 'flow_matching_loss': 0.014757164008915424, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.82} +{'loss': 0.015, 'grad_norm': 0.08803462982177734, 'learning_rate': 8.198983507064469e-06, 'flow_matching_loss': 0.014996055513620377, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.82} +{'loss': 0.0146, 'grad_norm': 0.08339009433984756, 'learning_rate': 8.18425015648692e-06, 'flow_matching_loss': 0.014623521734029055, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.83} +{'loss': 0.0139, 'grad_norm': 0.07796301692724228, 'learning_rate': 8.169525087688122e-06, 'flow_matching_loss': 0.0139027233235538, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.83} +{'loss': 0.0149, 'grad_norm': 0.092775858938694, 'learning_rate': 8.154808318560413e-06, 'flow_matching_loss': 0.014880090206861495, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.83} +{'loss': 0.0133, 'grad_norm': 0.0720883309841156, 'learning_rate': 8.140099866986058e-06, 'flow_matching_loss': 0.013261178508400917, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.84} +{'loss': 0.0129, 'grad_norm': 0.07374638319015503, 'learning_rate': 8.125399750837202e-06, 'flow_matching_loss': 0.012893917225301265, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.84} +{'loss': 0.0139, 'grad_norm': 0.07515527307987213, 'learning_rate': 8.110707987975888e-06, 'flow_matching_loss': 0.01385287856683135, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.84} +{'loss': 0.0147, 'grad_norm': 0.07491724193096161, 'learning_rate': 8.096024596253982e-06, 'flow_matching_loss': 0.014749199524521828, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.85} +{'loss': 0.0132, 'grad_norm': 0.0765443816781044, 'learning_rate': 8.081349593513197e-06, 'flow_matching_loss': 0.013229366485029459, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.85} +{'loss': 0.0132, 'grad_norm': 0.0906120091676712, 'learning_rate': 8.066682997585045e-06, 'flow_matching_loss': 0.013214761950075626, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.85} +{'loss': 0.0131, 'grad_norm': 0.10625690221786499, 'learning_rate': 8.052024826290819e-06, 'flow_matching_loss': 0.013111216295510531, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.85} +{'loss': 0.0158, 'grad_norm': 0.09113506972789764, 'learning_rate': 8.037375097441574e-06, 'flow_matching_loss': 0.015779935475438833, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.86} +{'loss': 0.0134, 'grad_norm': 0.07908552139997482, 'learning_rate': 8.022733828838124e-06, 'flow_matching_loss': 0.013436411134898662, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.86} +{'loss': 0.0147, 'grad_norm': 0.09008011966943741, 'learning_rate': 8.008101038270988e-06, 'flow_matching_loss': 0.014688399713486433, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.86} +{'loss': 0.0144, 'grad_norm': 0.07192651182413101, 'learning_rate': 7.99347674352038e-06, 'flow_matching_loss': 0.014431864488869905, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.87} +{'loss': 0.0135, 'grad_norm': 0.059935007244348526, 'learning_rate': 7.978860962356202e-06, 'flow_matching_loss': 0.013540288247168065, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.87} +{'loss': 0.014, 'grad_norm': 0.08810340613126755, 'learning_rate': 7.964253712537997e-06, 'flow_matching_loss': 0.013977660052478313, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.87} +{'loss': 0.0138, 'grad_norm': 0.06040380895137787, 'learning_rate': 7.949655011814968e-06, 'flow_matching_loss': 0.013802854204550385, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.87} +{'loss': 0.0131, 'grad_norm': 0.06618435680866241, 'learning_rate': 7.9350648779259e-06, 'flow_matching_loss': 0.013088070787489414, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.88} +{'loss': 0.0135, 'grad_norm': 0.06056733801960945, 'learning_rate': 7.92048332859919e-06, 'flow_matching_loss': 0.013462936785072089, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.88} +{'loss': 0.0148, 'grad_norm': 0.07528995722532272, 'learning_rate': 7.90591038155278e-06, 'flow_matching_loss': 0.014848919026553631, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.88} +{'loss': 0.0128, 'grad_norm': 0.07406371086835861, 'learning_rate': 7.891346054494194e-06, 'flow_matching_loss': 0.012801005970686674, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.89} +{'loss': 0.0138, 'grad_norm': 0.0773933008313179, 'learning_rate': 7.876790365120447e-06, 'flow_matching_loss': 0.013847187533974648, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.89} +{'loss': 0.0136, 'grad_norm': 0.06827635318040848, 'learning_rate': 7.862243331118088e-06, 'flow_matching_loss': 0.013645104505121708, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.89} +{'loss': 0.0136, 'grad_norm': 0.06843166053295135, 'learning_rate': 7.847704970163127e-06, 'flow_matching_loss': 0.013626991864293813, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.89} +{'loss': 0.0136, 'grad_norm': 0.07399114966392517, 'learning_rate': 7.833175299921046e-06, 'flow_matching_loss': 0.013644114509224892, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.9} 68%|██████▊ | 20270/30000 [12:41:47<3:43:46, 1.38s/it] 68%|██████▊ | 20271/30000 [12:41:48<3:34:36, 1.32s/it] 68%|██████▊ | 20272/30000 [12:41:54<7:03:26, 2.61s/it] 68%|██████▊ | 20273/30000 [12:41:55<5:58:42, 2.21s/it] 68%|██████▊ | 20274/30000 [12:41:56<5:08:45, 1.90s/it] 68%|██████▊ | 20275/30000 [12:41:58<4:34:06, 1.69s/it] 68%|██████▊ | 20276/30000 [12:41:59<4:09:27, 1.54s/it] 68%|██████▊ | 20277/30000 [12:42:00<3:52:14, 1.43s/it] 68%|██████▊ | 20278/30000 [12:42:01<3:40:35, 1.36s/it] 68%|██████▊ | 20279/30000 [12:42:02<3:32:20, 1.31s/it] 68%|██████▊ | 20280/30000 [12:42:07<6:22:58, 2.36s/it] 68%|██████▊ | 20280/30000 [12:42:07<6:22:58, 2.36s/it] 68%|██████▊ | 20281/30000 [12:42:09<5:32:30, 2.05s/it] 68%|██████▊ | 20282/30000 [12:42:10<4:51:17, 1.80s/it] 68%|██████▊ | 20283/30000 [12:42:11<4:21:50, 1.62s/it] 68%|██████▊ | 20284/30000 [12:42:12<4:00:11, 1.48s/it] 68%|██████▊ | 20285/30000 [12:42:13<3:45:19, 1.39s/it] 68%|██████▊ | 20286/30000 [12:42:14<3:35:55, 1.33s/it] 68%|██████▊ | 20287/30000 [12:42:16<3:29:26, 1.29s/it] 68%|██████▊ | 20288/30000 [12:42:21<6:44:12, 2.50s/it] 68%|██████▊ | 20289/30000 [12:42:23<6:04:24, 2.25s/it] 68%|██████▊ | 20290/30000 [12:42:24<5:13:01, 1.93s/it] 68%|██████▊ | 20290/30000 [12:42:24<5:13:01, 1.93s/it] 68%|██████▊ | 20291/30000 [12:42:25<4:35:52, 1.70s/it] 68%|██████▊ | 20292/30000 [12:42:26<4:10:15, 1.55s/it] 68%|██████▊ | 20293/30000 [12:42:27<3:52:20, 1.44s/it] 68%|██████▊ | 20294/30000 [12:42:29<3:40:18, 1.36s/it] 68%|██████▊ | 20295/30000 [12:42:30<3:32:56, 1.32s/it] 68%|██████▊ | 20296/30000 [12:42:35<6:40:18, 2.48s/it] 68%|██████▊ | 20297/30000 [12:42:37<6:10:12, 2.29s/it] 68%|██████▊ | 20298/30000 [12:42:38<5:15:08, 1.95s/it] 68%|██████▊ | 20299/30000 [12:42:39<4:38:32, 1.72s/it] 68%|██████▊ | 20300/30000 [12:42:40<4:11:40, 1.56s/it] 68%|██████▊ | 20300/30000 [12:42:40<4:11:40, 1.56s/it] 68%|██████▊ | 20301/30000 [12:42:42<3:53:41, 1.45s/it] 68%|██████▊ | 20302/30000 [12:42:43<3:41:10, 1.37s/it] 68%|██████▊ | 20303/30000 [12:42:44<3:32:48, 1.32s/it] 68%|██████▊ | 20304/30000 [12:42:49<6:45:35, 2.51s/it] 68%|██████▊ | 20305/30000 [12:42:51<5:54:03, 2.19s/it] 68%|██████▊ | 20306/30000 [12:42:52<5:04:45, 1.89s/it] 68%|██████▊ | 20307/30000 [12:42:53<4:30:06, 1.67s/it] 68%|██████▊ | 20308/30000 [12:42:54<4:06:00, 1.52s/it] 68%|██████▊ | 20309/30000 [12:42:55<3:48:23, 1.41s/it] 68%|██████▊ | 20310/30000 [12:42:57<3:37:51, 1.35s/it] 68%|██████▊ | 20310/30000 [12:42:57<3:37:51, 1.35s/it] 68%|██████▊ | 20311/30000 [12:42:58<3:30:30, 1.30s/it] 68%|██████▊ | 20312/30000 [12:43:04<7:07:39, 2.65s/it] 68%|██████▊ | 20313/30000 [12:43:05<6:16:56, 2.33s/it] 68%|██████▊ | 20314/30000 [12:43:06<5:21:12, 1.99s/it] 68%|██████▊ | 20315/30000 [12:43:07<4:42:08, 1.75s/it] 68%|██████▊ | 20316/30000 [12:43:09<4:14:19, 1.58s/it] 68%|██████▊ | 20317/30000 [12:43:10<3:55:08, 1.46s/it] 68%|██████▊ | 20318/30000 [12:43:11<3:42:08, 1.38s/it] 68%|██████▊ | 20319/30000 [12:43:12<3:31:56, 1.31s/it] 68%|██████▊ | 20320/30000 [12:43:17<6:33:03, 2.44s/it] 68%|██████▊ | 20320/30000 [12:43:17<6:33:03, 2.44s/it] 68%|██████▊ | 20321/30000 [12:43:20<6:24:49, 2.39s/it] 68%|██████▊ | 20322/30000 [12:43:21<5:26:12, 2.02s/it] 68%|██████▊ | 20323/30000 [12:43:22<4:46:12, 1.77s/it] 68%|██████▊ | 20324/30000 [12:43:23<4:17:24, 1.60s/it] 68%|██████▊ | 20325/30000 [12:43:24<3:57:09, 1.47s/it] 68%|██████▊ | 20326/30000 [12:43:25<3:44:26, 1.39s/it] 68%|██████▊ | 20327/30000 [12:43:27<3:34:08, 1.33s/it] 68%|██████▊ | 20328/30000 [12:43:31<6:10:33, 2.30s/it] 68%|██████▊ | 20329/30000 [12:43:33<6:06:55, 2.28s/it] 68%|██████▊ | 20330/30000 [12:43:35<5:13:37, 1.95s/it] 68%|██████▊ | 20330/30000 [12:43:35<5:13:37, 1.95s/it] 68%|██████▊ | 20331/30000 [12:43:36<4:36:10, 1.71s/it] 68%|██████▊ | 20332/30000 [12:43:37<4:09:43, 1.55s/it] 68%|██████▊ | 20333/30000 [12:43:38<3:50:43, 1.43s/it] 68%|██████▊ | 20334/30000 [12:43:39<3:41:06, 1.37s/it] 68%|██████▊ | 20335/30000 [12:43:41<3:32:35, 1.32s/it] 68%|██████▊ | 20336/30000 [12:43:46<6:33:29, 2.44s/it] 68%|██████▊ | 20337/30000 [12:43:48<6:12:20, 2.31s/it] 68%|██████▊ | 20338/30000 [12:43:49<5:16:52, 1.97s/it] 68%|██████▊ | 20339/30000 [12:43:50<4:38:25, 1.73s/it] 68%|██████▊ | 20340/30000 [12:43:51<4:12:08, 1.57s/it] 68%|██████▊ | 20340/30000 [12:43:51<4:12:08, 1.57s/it] 68%|██████▊ | 20341/30000 [12:43:52<3:53:04, 1.45s/it] 68%|██████▊ | 20342/30000 [12:43:53<3:40:16, 1.37s/it] 68%|██████▊ | 20343/30000 [12:43:55<3:31:52, 1.32s/it] 68%|██████▊ | 20344/30000 [12:44:00<6:53:51, 2.57s/it] 68%|██████▊ | 20345/30000 [12:44:02<6:08:52, 2.29s/it] 68%|██████▊ | 20346/30000 [12:44:03<5:14:44, 1.96s/it] 68%|██████▊ | 20347/30000 [12:44:04<4:38:12, 1.73s/it] 68%|██████▊ | 20348/30000 [12:44:05<4:10:43, 1.56s/it] 68%|██████▊ | 20349/30000 [12:44:07<3:52:34, 1.45s/it] 68%|██████▊ | 20350/30000 [12:44:08<3:40:15, 1.37s/it] 68%|██████▊ | 20350/30000 [12:44:08<3:40:15, 1.37s/it] 68%|██████▊ | 20351/30000 [12:44:09<3:32:08, 1.32s/it] 68%|██████▊ | 20352/30000 [12:44:15<7:33:20, 2.82s/it] 68%|██████▊ | 20353/30000 [12:44:16<6:15:25, 2.33s/it] 68%|██████▊ | 20354/30000 [12:44:18<5:18:49, 1.98s/it] 68%|██████▊ | 20355/30000 [12:44:19<4:40:01, 1.74s/it] 68%|██████▊ | 20356/30000 [12:44:20<4:12:10, 1.57s/it] 68%|██████▊ | 20357/30000 [12:44:21<3:52:20, 1.45s/it] 68%|██████▊ | 20358/30000 [12:44:22<3:40:22, 1.37s/it] 68%|██████▊ | 20359/30000 [12:44:24<3:31:53, 1.32s/it] 68%|██████▊ | 20360/30000 [12:44:29<7:14:11, 2.70s/it] 68%|██████▊ | 20360/30000 [12:44:29<7:14:11, 2.70s/it] 68%|██████▊ | 20361/30000 [12:44:31<6:03:05, 2.26s/it] 68%|██████▊ | 20362/30000 [12:44:32<5:10:22, 1.93s/it] 68%|██████▊ | 20363/30000 [12:44:33<4:33:42, 1.70s/it] 68%|██████▊ | 20364/30000 [12:44:34<4:08:00, 1.54s/it] 68%|██████▊ | 20365/30000 [12:44:35<3:50:19, 1.43s/it] 68%|██████▊ | 20366/30000 [12:44:37<3:38:27, 1.36s/it] 68%|██████▊ | 20367/30000 [12:44:38<3:29:28, 1.30s/it] 68%|██████▊ | 20368/30000 [12:44:44<7:18:34, 2.73s/it] 68%|██████▊ | 20369/30000 [12:44:46<6:39:28, 2.49s/it] 68%|██████▊ | 20370/30000 [12:44:47<5:35:42, 2.09s/it] 68%|██████▊ | 20370/30000 [12:44:47<5:35:42, 2.09s/it] 68%|██████▊ | 20371/30000 [12:44:48<4:52:00, 1.82s/it] 68%|██████▊ | 20372/30000 [12:44:49<4:20:34, 1.62s/it] 68%|██████▊ | 20373/30000 [12:44:50<3:59:00, 1.49s/it] 68%|██████▊ | 20374/30000 [12:44:52<3:44:38, 1.40s/it] 68%|██████▊ | 20375/30000 [12:44:53<3:34:05, 1.33s/it] 68%|██████▊ | 20376/30000 [12:44:59<7:18:29, 2.73s/it] 68%|██████▊ | 20377/30000 [12:45:01<6:43:27, 2.52s/it] 68%|██████▊ | 20378/30000 [12:45:02<5:38:41, 2.11s/it] 68%|██████▊ | 20379/30000 [12:45:03<4:54:12, 1.83s/it] 68%|██████▊ | 20380/30000 [12:45:04<4:22:11, 1.64s/it] 68%|██████▊ | 20380/30000 [12:45:04<4:22:11, 1.64s/it] 68%|██████▊ | 20381/30000 [12:45:05<4:00:24, 1.50s/it] 68%|██████▊ | 20382/30000 [12:45:07<3:45:50, 1.41s/it] 68%|██████▊ | 20383/30000 [12:45:08<3:34:33, 1.34s/it] 68%|██████▊ | 20384/30000 [12:45:13<6:48:57, 2.55s/it] 68%|██████▊ | 20385/30000 [12:45:15<6:14:01, 2.33s/it] 68%|██████▊ | 20386/30000 [12:45:16<5:18:50, 1.99s/it] 68%|██████▊ | 20387/30000 [12:45:17<4:40:01, 1.75s/it] 68%|██████▊ | 20388/30000 [12:45:19<4:11:41, 1.57s/it] 68%|██████▊ | 20389/30000 [12:45:20<3:52:24, 1.45s/it] 68%|██████▊ | 20390/30000 [12:45:21<3:38:45, 1.37s/it] 68%|██████▊ | 20390/30000 [12:45:21<3:38:45, 1.37s/it] 68%|██████▊ | 20391/30000 [12:45:22<3:30:11, 1.31s/it] 68%|██████▊ | 20392/30000 [12:45:27<6:43:00, 2.52s/it] 68%|██████▊ | 20393/30000 [12:45:29<6:13:15, 2.33s/it] 68%|██████▊ | 20394/30000 [12:45:31<5:17:28, 1.98s/it] 68%|██████▊ | 20395/30000 [12:45:32<4:38:56, 1.74s/it] 68%|██████▊ | 20396/30000 [12:45:33<4:11:35, 1.57s/it] 68%|██████▊ | 20397/30000 [12:45:34<3:51:54, 1.45s/it] 68%|██████▊ | 20398/30000 [12:45:35<3:38:21, 1.36s/it] 68%|██████▊ | 20399/30000 [12:45:36<3:29:34, 1.31s/it] 68%|██████▊ | 20400/30000 [12:45:42<7:03:57, 2.65s/it] 68%|██████▊ | 20400/30000 [12:45:42<7:03:57, 2.65s/it] 68%|██████▊ | 20401/30000 [12:45:44<6:17:35, 2.36s/it] 68%|██████▊ | 20402/30000 [12:45:45<5:20:58, 2.01s/it] 68%|██████▊ | 20403/30000 [12:45:46<4:40:55, 1.76s/it] 68%|██████▊ | 20404/30000 [12:45:47<4:12:48, 1.58s/it] 68%|██████▊ | 20405/30000 [12:45:49<3:54:07, 1.46s/it] 68%|██████▊ | 20406/30000 [12:45:50<3:40:36, 1.38s/it] 68%|██████▊ | 20407/30000 [12:45:51<3:31:31, 1.32s/it] 68%|██████▊ | 20408/30000 [12:45:57<7:32:53, 2.83s/it] 68%|██████▊ | 20409/30000 [12:45:58<6:12:57, 2.33s/it] 68%|██████▊ | 20410/30000 [12:46:00<5:17:19, 1.99s/it] 68%|██████▊ | 20410/30000 [12:46:00<5:17:19, 1.99s/it] 68%|██████▊ | 20411/30000 [12:46:01<4:38:24, 1.74s/it] 68%|██████▊ | 20412/30000 [12:46:02<4:11:21, 1.57s/it] 68%|██████▊ | 20413/30000 [12:46:03<3:52:21, 1.45s/it] 68%|██████▊ | 20414/30000 [12:46:04<3:38:26, 1.37s/it] 68%|██████▊ | 20415/30000 [12:46:06<3:29:33, 1.31s/it] 68%|██████▊ | 20416/30000 [12:46:12<7:36:50, 2.86s/it] 68%|██████▊ | 20417/30000 [12:46:13<6:16:38, 2.36s/it] 68%|██████▊ | 20418/30000 [12:46:14<5:19:35, 2.00s/it] 68%|██████▊ | 20419/30000 [12:46:16<4:40:27, 1.76s/it] 68%|██████▊ | 20420/30000 [12:46:17<4:12:07, 1.58s/it] 68%|██████▊ | 20420/30000 [12:46:17<4:12:07, 1.58s/it] 68%|██████▊ | 20421/30000 [12:46:18<3:52:37, 1.46s/it] 68%|██████▊ | 20422/30000 [12:46:19<3:39:26, 1.37s/it] 68%|██████▊ | 20423/30000 [12:46:20<3:30:16, 1.32s/it] 68%|██████▊ | 20424/30000 [12:46:27<7:52:06, 2.96s/it] 68%|██████▊ | 20425/30000 [12:46:28<6:26:46, 2.42s/it] 68%|██████▊ | 20426/30000 [12:46:29<5:26:15, 2.04s/it] 68%|██████▊ | 20427/30000 [12:46:31<4:44:40, 1.78s/it] 68%|██████▊ | 20428/30000 [12:46:32<4:14:54, 1.60s/it] 68%|██████▊ | 20429/30000 [12:46:33<3:54:21, 1.47s/it] 68%|██████▊ | 20430/30000 [12:46:34<3:40:50, 1.38s/it] 68%|██████▊ | 20430/30000 [12:46:34<3:40:50, 1.38s/it] 68%|██████▊ | 20431/30000 [12:46:35<3:31:20, 1.33s/it] 68%|██████▊ | 20432/30000 [12:46:41<7:26:58, 2.80s/it] 68%|██████▊ | 20433/30000 [12:46:43<6:09:09, 2.32s/it] 68%|██████▊ | 20434/30000 [12:46:44<5:14:19, 1.97s/it] 68%|██████▊ | 20435/30000 [12:46:45<4:38:15, 1.75s/it] 68%|██████▊ | 20436/30000 [12:46:46<4:10:51, 1.57s/it] 68%|██████▊ | 20437/30000 [12:46:47<3:51:39, 1.45s/it] 68%|██████▊ | 20438/30000 [12:46:49<3:38:53, 1.37s/it] 68%|██████▊ | 20439/30000 [12:46:50<3:29:52, 1.32s/it] 68%|██████▊ | 20440/30000 [12:46:56<7:26:32, 2.80s/it] 68%|██████▊ | 20440/30000 [12:46:56<7:26:32, 2.80s/it] 68%|██████▊ | 20441/30000 [12:46:57<6:08:32, 2.31s/it] 68%|██████▊ | 20442/30000 [12:46:58<5:13:59, 1.97s/it] 68%|██████▊ | 20443/30000 [12:47:00<4:36:30, 1.74s/it] 68%|██████▊ | 20444/30000 [12:47:01<4:09:27, 1.57s/it] 68%|██████▊ | 20445/30000 [12:47:02<3:51:05, 1.45s/it] 68%|██████▊ | 20446/30000 [12:47:03<3:38:31, 1.37s/it] 68%|██████▊ | 20447/30000 [12:47:04<3:29:06, 1.31s/it] 68%|██████▊ | 20448/30000 [12:47:11<7:24:51, 2.79s/it] 68%|██████▊ | 20449/30000 [12:47:12<6:07:12, 2.31s/it] 68%|██████▊ | 20450/30000 [12:47:13<5:12:45, 1.96s/it] 68%|██████▊ | 20450/30000 [12:47:13<5:12:45, 1.96s/it] 68%|██████▊ | 20451/30000 [12:47:14<4:34:51, 1.73s/it] 68%|██████▊ | 20452/30000 [12:47:15<4:08:13, 1.56s/it] 68%|██████▊ | 20453/30000 [12:47:16<3:49:26, 1.44s/it] 68%|██████▊ | 20454/30000 [12:47:18<3:37:18, 1.37s/it] 68%|██████▊ | 20455/30000 [12:47:19<3:28:17, 1.31s/it] 68%|██████▊ | 20456/30000 [12:47:25<7:39:01, 2.89s/it] 68%|██████▊ | 20457/30000 [12:47:27<6:45:02, 2.55s/it] 68%|██████▊ | 20458/30000 [12:47:28<5:39:24, 2.13s/it] 68%|██████▊ | 20459/30000 [12:47:29<4:54:07, 1.85s/it] 68%|██████▊ | 20460/30000 [12:47:31<4:23:31, 1.66s/it] 68%|██████▊ | 20460/30000 [12:47:31<4:23:31, 1.66s/it] 68%|██████▊ | 20461/30000 [12:47:32<4:00:19, 1.51s/it] 68%|██████▊ | 20462/30000 [12:47:33<3:44:58, 1.42s/it] 68%|██████▊ | 20463/30000 [12:47:34<3:33:59, 1.35s/it] 68%|██████▊ | 20464/30000 [12:47:40<7:27:20, 2.81s/it] 68%|██████▊ | 20465/30000 [12:47:42<6:21:52, 2.40s/it] 68%|██████▊ | 20466/30000 [12:47:43<5:22:36, 2.03s/it] 68%|██████▊ | 20467/30000 [12:47:44<4:41:46, 1.77s/it] 68%|██████▊ | 20468/30000 [12:47:45<4:13:07, 1.59s/it] 68%|██████▊ | 20469/30000 [12:47:47<3:52:16, 1.46s/it] 68%|██████▊ | 20470/30000 [12:47:48<3:38:47, 1.38s/it] 68%|██████▊ | 20470/30000 [12:47:48<3:38:47, 1.38s/it] 68%|██████▊ | 20471/30000 [12:47:49<3:29:34, 1.32s/it] 68%|██████▊ | 20472/30000 [12:47:55<7:21:11, 2.78s/it] 68%|██████▊ | 20473/30000 [12:47:56<6:08:21, 2.32s/it] 68%|██████▊ | 20474/30000 [12:47:57<5:13:52, 1.98s/it] 68%|██████▊ | 20475/30000 [12:47:59<4:36:35, 1.74s/it] 68%|██████▊ | 20476/30000 [12:48:00<4:09:46, 1.57s/it] 68%|██████▊ | 20477/30000 [12:48:01<3:50:32, 1.45s/it] 68%|██████▊ | 20478/30000 [12:48:02<3:36:54, 1.37s/it] 68%|██████▊ | 20479/30000 [12:48:03<3:28:27, 1.31s/it] 68%|██████▊ | 20480/30000 [12:48:09<7:13:58, 2.74s/it] 68%|██████▊ | 20480/30000 [12:48:09<7:13:58, 2.74s/it] 68%|██████▊ | 20481/30000 [12:48:11<6:02:21, 2.28s/it] 68%|██████▊ | 20482/30000 [12:48:12<5:09:02, 1.95s/it] 68%|██████▊ | 20483/30000 [12:48:13<4:32:52, 1.72s/it] 68%|██████▊ | 20484/30000 [12:48:14<4:06:22, 1.55s/it] 68%|██████▊ | 20485/30000 [12:48:15<3:48:27, 1.44s/it] 68%|██████▊ | 20486/30000 [12:48:17<3:37:03, 1.37s/it] 68%|██████▊ | 20487/30000 [12:48:18<3:28:16, 1.31s/it] 68%|██████▊ | 20488/30000 [12:48:25<8:05:42, 3.06s/it] 68%|██████▊ | 20489/30000 [12:48:26<6:35:51, 2.50s/it] 68%|██████▊ | 20490/30000 [12:48:27<5:32:47, 2.10s/it] 68%|██████▊ | 20490/30000 [12:48:27<5:32:47, 2.10s/it] 68%|██████▊ | 20491/30000 [12:48:28<4:49:52, 1.83s/it] 68%|██████▊ | 20492/30000 [12:48:30<4:18:29, 1.63s/it] 68%|██████▊ | 20493/30000 [12:48:31<3:56:44, 1.49s/it] 68%|██████▊ | 20494/30000 [12:48:32<3:41:24, 1.40s/it] 68%|██████▊ | 20495/30000 [12:48:33<3:31:08, 1.33s/it] 68%|██████▊ | 20496/30000 [12:48:41<8:29:10, 3.21s/it] 68%|██████▊ | 20497/30000 [12:48:42<6:51:35, 2.60s/it] 68%|██████▊ | 20498/30000 [12:48:43<5:43:03, 2.17s/it] 68%|██████▊ | 20499/30000 [12:48:44<4:56:02, 1.87s/it] 68%|██████▊ | 20500/30000 [12:48:45<4:22:12, 1.66s/it] 68%|██████▊ | 20500/30000 [12:48:45<4:22:12, 1.66s/it] 68%|██████▊ | 20501/30000 [12:48:47<4:00:05, 1.52s/it] 68%|██████▊ | 20502/30000 [12:48:48<3:44:45, 1.42s/it] 68%|██████▊ | 20503/30000 [12:48:49<3:33:53, 1.35s/it] 68%|██████▊ | 20504/30000 [12:48:55<7:12:59, 2.74s/it] 68%|██████▊ | 20505/30000 [12:48:56<5:58:52, 2.27s/it] 68%|██████▊ | 20506/30000 [12:48:57<5:06:53, 1.94s/it] 68%|██████▊ | 20507/30000 [12:48:58<4:31:04, 1.71s/it] 68%|██████▊ | 20508/30000 [12:49:00<4:05:34, 1.55s/it] 68%|██████▊ | 20509/30000 [12:49:01<3:47:02, 1.44s/it] 68%|██████▊ | 20510/30000 [12:49:02<3:35:37, 1.36s/it] 68%|██████▊ | 20510/30000 [12:49:02<3:35:37, 1.36s/it] 68%|██████▊ | 20511/30000 [12:49:03<3:27:33, 1.31s/it] 68%|██████▊ | 20512/30000 [12:49:09<7:04:02, 2.68s/it] 68%|██████▊ | 20513/30000 [12:49:10<5:52:13, 2.23s/it] 68%|██████▊ | 20514/30000 [12:49:11<5:02:09, 1.91s/it] 68%|██████▊ | 20515/30000 [12:49:13<4:27:54, 1.69s/it] 68%|██████▊ | 20516/30000 [12:49:14<4:03:59, 1.54s/it] 68%|██████▊ | 20517/30000 [12:49:15<3:46:29, 1.43s/it] 68%|██████▊ | 20518/30000 [12:49:16<3:34:19, 1.36s/it] 68%|██████▊ | 20519/30000 [12:49:17<3:26:24, 1.31s/it] 68%|██████▊ | 20520/30000 [12:49:23<7:09:47, 2.72s/it] 68%|██████▊ | 20520/30000 [12:49:23<7:09:47, 2.72s/it] 68%|██████▊ | 20521/30000 [12:49:25<6:11:25, 2.35s/it] 68%|██████▊ | 20522/30000 [12:49:26<5:15:35, 2.00s/it] 68%|██████▊ | 20523/30000 [12:49:27<4:37:34, 1.76s/it] 68%|██████▊ | 20524/30000 [12:49:28<4:09:01, 1.58s/it] 68%|██████▊ | 20525/30000 [12:49:30<3:49:59, 1.46s/it] 68%|██████▊ | 20526/30000 [12:49:31<3:37:46, 1.38s/it] 68%|██████▊ | 20527/30000 [12:49:32<3:27:52, 1.32s/it] 68%|██████▊ | 20528/30000 [12:49:38<7:05:13, 2.69s/it] 68%|██████▊ | 20529/30000 [12:49:39<6:04:30, 2.31s/it] 68%|██████▊ | 20530/30000 [12:49:40<5:10:46, 1.97s/it] 68%|██████▊ | 20530/30000 [12:49:40<5:10:46, 1.97s/it] 68%|██████▊ | 20531/30000 [12:49:42<4:34:01, 1.74s/it] 68%|██████▊ | 20532/30000 [12:49:43<4:06:55, 1.56s/it] 68%|██████▊ | 20533/30000 [12:49:44<3:47:53, 1.44s/it] 68%|██████▊ | 20534/30000 [12:49:45<3:34:37, 1.36s/it] 68%|██████▊ | 20535/30000 [12:49:46<3:25:32, 1.30s/it] 68%|██████▊ | 20536/30000 [12:49:52<6:51:40, 2.61s/it] 68%|██████▊ | 20537/30000 [12:49:54<6:12:20, 2.36s/it] 68%|██████▊ | 20538/30000 [12:49:55<5:16:14, 2.01s/it] 68%|██████▊ | 20539/30000 [12:49:56<4:37:19, 1.76s/it] 68%|██████▊ | 20540/30000 [12:49:57<4:09:53, 1.58s/it] 68%|██████▊ | 20540/30000 [12:49:57<4:09:53, 1.58s/it] 68%|██████▊ | 20541/30000 [12:49:58<3:51:00, 1.47s/it] 68%|██████▊ | 20542/30000 [12:50:00<3:36:48, 1.38s/it] 68%|██████▊ | 20543/30000 [12:50:01<3:26:56, 1.31s/it] 68%|██████▊ | 20544/30000 [12:50:07<7:02:10, 2.68s/it] 68%|██████▊ | 20545/30000 [12:50:08<6:08:53, 2.34s/it] 68%|██████▊ | 20546/30000 [12:50:09<5:13:36, 1.99s/it] 68%|██████▊ | 20547/30000 [12:50:11<4:35:49, 1.75s/it] 68%|██████▊ | 20548/30000 [12:50:12<4:08:04, 1.57s/it] 68%|██████▊ | 20549/30000 [12:50:13<3:48:45, 1.45s/it] 68%|██████▊ | 20550/30000 [12:50:14<3:35:08, 1.37s/it] 68%|██████▊ | 20550/30000 [12:50:14<3:35:08, 1.37s/it] 69%|██████▊ | 20551/30000 [12:50:15<3:26:16, 1.31s/it] 69%|██████▊ | 20552/30000 [12:50:22<7:54:26, 3.01s/it] 69%|██████▊ | 20553/30000 [12:50:23<6:27:00, 2.46s/it] 69%|██████▊ | 20554/30000 [12:50:25<5:26:13, 2.07s/it] 69%|██████▊ | 20555/30000 [12:50:26<4:44:07, 1.80s/it] 69%|██████▊ | 20556/30000 [12:50:27<4:13:50, 1.61s/it] 69%|██████▊ | 20557/30000 [12:50:28<3:52:15, 1.48s/it] 69%|██████▊ | 20558/30000 [12:50:29<3:38:53, 1.39s/it] 69%|██████▊ | 20559/30000 [12:50:30<3:29:21, 1.33s/it] 69%|██████▊ | 20560/30000 [12:50:37<7:55:20, 3.02s/it] 69%|██████▊ | 20560/30000 [12:50:37<7:55:20, 3.02s/it] 69%|██████▊ | 20561/30000 [12:50:39<6:28:22, 2.47s/it] 69%|██████▊ | 20562/30000 [12:50:40<5:26:39, 2.08s/it] 69%|██████▊ | 20563/30000 [12:50:41<4:44:36, 1.81s/it] 69%|██████▊ | 20564/30000 [12:50:42<4:14:22, 1.62s/it] 69%|██████▊ | 20565/30000 [12:50:43<3:53:02, 1.48s/it] 69%|██████▊ | 20566/30000 [12:50:44<3:39:01, 1.39s/it] 69%|██████▊ | 20567/30000 [12:50:46<3:29:08, 1.33s/it] 69%|██████▊ | 20568/30000 [12:50:52<7:31:35, 2.87s/it] 69%|██████▊ | 20569/30000 [12:50:53<6:11:16, 2.36s/it] 69%|██████▊ | 20570/30000 [12:50:54<5:14:41, 2.00s/it] 69%|██████▊ | 20570/30000 [12:50:54<5:14:41, 2.00s/it] 69%|██████▊ | 20571/30000 [12:50:56<4:36:12, 1.76s/it] 69%|██████▊ | 20572/30000 [12:50:57<4:07:55, 1.58s/it] 69%|██████▊ | 20573/30000 [12:50:58<3:48:43, 1.46s/it] 69%|██████▊ | 20574/30000 [12:50:59<3:35:43, 1.37s/it] 69%|██████▊ | 20575/30000 [12:51:00<3:25:38, 1.31s/it] 69%|██████▊ | 20576/30000 [12:51:07<7:30:53, 2.87s/it] 69%|██████▊ | 20577/30000 [12:51:08<6:11:01, 2.36s/it] 69%|██████▊ | 20578/30000 [12:51:09<5:14:45, 2.00s/it] 69%|██████▊ | 20579/30000 [12:51:10<4:36:11, 1.76s/it] 69%|██████▊ | 20580/30000 [12:51:12<4:08:35, 1.58s/it] 69%|██████▊ | 20580/30000 [12:51:12<4:08:35, 1.58s/it] 69%|██████▊ | 20581/30000 [12:51:13<3:48:50, 1.46s/it] 69%|██████▊ | 20582/30000 [12:51:14<3:36:12, 1.38s/it] 69%|██████▊ | 20583/30000 [12:51:15<3:26:55, 1.32s/it] 69%|██████▊ | 20584/30000 [12:51:22<7:37:13, 2.91s/it] 69%|██████▊ | 20585/30000 [12:51:23<6:15:12, 2.39s/it] 69%|██████▊ | 20586/30000 [12:51:24<5:17:46, 2.03s/it] 69%|██████▊ | 20587/30000 [12:51:25<4:38:14, 1.77s/it] 69%|██████▊ | 20588/30000 [12:51:26<4:10:12, 1.60s/it] 69%|██████▊ | 20589/30000 [12:51:28<3:49:50, 1.47s/it] 69%|██████▊ | 20590/30000 [12:51:29<3:36:45, 1.38s/it] 69%|██████▊ | 20590/30000 [12:51:29<3:36:45, 1.38s/it] 69%|██████▊ | 20591/30000 [12:51:30<3:26:48, 1.32s/it] 69%|██████▊ | 20592/30000 [12:51:36<6:58:04, 2.67s/it] 69%|██████▊ | 20593/30000 [12:51:38<6:45:35, 2.59s/it] 69%|██████▊ | 20594/30000 [12:51:39<5:39:21, 2.16s/it] 69%|██████▊ | 20595/30000 [12:51:41<4:53:59, 1.88s/it] 69%|██████▊ | 20596/30000 [12:51:42<4:20:45, 1.66s/it] 69%|██████▊ | 20597/30000 [12:51:43<3:56:34, 1.51s/it] 69%|██████▊ | 20598/30000 [12:51:44<3:41:33, 1.41s/it] 69%|██████▊ | 20599/30000 [12:51:45<3:30:20, 1.34s/it] 69%|██████▊ | 20600/30000 [12:51:52<7:23:18, 2.83s/it] 69%|██████▊ | 20600/30000 [12:51:52<7:23:18, 2.83s/it] 69%|██████▊ | 20601/30000 [12:51:53<6:33:31, 2.51s/it] 69%|██████▊ | 20602/30000 [12:51:54<5:30:17, 2.11s/it] 69%|██████▊ | 20603/30000 [12:51:56<4:47:00, 1.83s/it] 69%|██████▊ | 20604/30000 [12:51:57<4:14:48, 1.63s/it] 69%|██████▊ | 20605/30000 [12:51:58<3:53:11, 1.49s/it] 69%|██████▊ | 20606/30000 [12:51:59<3:39:04, 1.40s/it] 69%|██████▊ | 20607/30000 [12:52:00<3:28:38, 1.33s/it] 69%|██████▊ | 20608/30000 [12:52:07<7:32:21, 2.89s/it] 69%|██████▊ | 20609/30000 [12:52:09<7:20:05, 2.81s/it] 69%|██████▊ | 20610/30000 [12:52:11<6:02:59, 2.32s/it] 69%|██████▊ | 20610/30000 [12:52:11<6:02:59, 2.32s/it] 69%|██████▊ | 20611/30000 [12:52:12<5:09:56, 1.98s/it] 69%|██████▊ | 20612/30000 [12:52:13<4:32:04, 1.74s/it] 69%|██████▊ | 20613/30000 [12:52:14<4:05:39, 1.57s/it] 69%|██████▊ | 20614/30000 [12:52:15<3:47:35, 1.45s/it] 69%|██████▊ | 20615/30000 [12:52:17<3:34:49, 1.37s/it] 69%|██████▊ | 20616/30000 [12:52:24<7:57:48, 3.06s/it] 69%|██████▊ | 20617/30000 [12:52:26<7:51:30, 3.02s/it] 69%|██████▊ | 20618/30000 [12:52:28<6:24:24, 2.46s/it] 69%|██████▊ | 20619/30000 [12:52:29<5:24:20, 2.07s/it] 69%|██████▊ | 20620/30000 [12:52:30<4:40:55, 1.80s/it] 69%|██████▊ | 20620/30000 [12:52:30<4:40:55, 1.80s/it] 69%|██████▊ | 20621/30000 [12:52:31<4:11:08, 1.61s/it] 69%|██████▊ | 20622/30000 [12:52:32<3:16:18, 1.26s/it] 69%|██████▊ | 20623/30000 [12:52:50<16:32:58, 6.35s/it] 69%|██████▊ | 20624/30000 [12:52:51<12:30:51, 4.81s/it] 69%|██████▉ | 20625/30000 [12:52:52<9:40:18, 3.71s/it] 69%|██████▉ | 20626/30000 [12:52:53<7:40:41, 2.95s/it] 69%|██████▉ | 20627/30000 [12:52:54<6:17:25, 2.42s/it] 69%|██████▉ | 20628/30000 [12:52:56<5:18:56, 2.04s/it] 69%|██████▉ | 20629/30000 [12:52:57<4:38:23, 1.78s/it] 69%|██████▉ | 20630/30000 [12:53:07<11:25:25, 4.39s/it] 69%|██████▉ | 20630/30000 [12:53:07<11:25:25, 4.39s/it] 69%|██████▉ | 20631/30000 [12:53:08<8:54:53, 3.43s/it] 69%|██████▉ | 20632/30000 [12:53:10<7:09:37, 2.75s/it] 69%|██████▉ | 20633/30000 [12:53:11<5:54:55, 2.27s/it] 69%|██████▉ | 20634/30000 [12:53:12<5:03:09, 1.94s/it] 69%|██████▉ | 20635/30000 [12:53:13<4:28:22, 1.72s/it] 69%|██████▉ | 20636/30000 [12:53:14<4:03:27, 1.56s/it] 69%|██████▉ | 20637/30000 [12:53:16<3:45:43, 1.45s/it] 69%|██████▉ | 20638/30000 [12:53:22<7:37:19, 2.93s/it] 69%|██████▉ | 20639/30000 [12:53:23<6:14:52, 2.40s/it] 69%|██████▉ | 20640/30000 [12:53:24<5:17:47, 2.04s/it] 69%|██████▉ | 20640/30000 [12:53:24<5:17:47, 2.04s/it] 69%|██████▉ | 20641/30000 [12:53:25<4:37:14, 1.78s/it] 69%|██████▉ | 20642/30000 [12:53:27<4:08:51, 1.60s/it] 69%|██████▉ | 20643/30000 [12:53:28<3:50:01, 1.48s/it] 69%|██████▉ | 20644/30000 [12:53:29<3:36:30, 1.39s/it] 69%|██████▉ | 20645/30000 [12:53:30<3:27:13, 1.33s/it] 69%|██████▉ | 20646/30000 [12:53:37<7:33:31, 2.91s/it] 69%|██████▉ | 20647/30000 [12:53:38<6:12:42, 2.39s/it] 69%|██████▉ | 20648/30000 [12:53:39<5:16:30, 2.03s/it] 69%|██████▉ | 20649/30000 [12:53:40<4:36:43, 1.78s/it] 69%|██████▉ | 20650/30000 [12:53:42<4:08:17, 1.59s/it] 69%|██████▉ | 20650/30000 [12:53:42<4:08:17, 1.59s/it] 69%|██████▉ | 20651/30000 [12:53:43<3:49:24, 1.47s/it] 69%|██████▉ | 20652/30000 [12:53:44<3:35:42, 1.38s/it] 69%|██████▉ | 20653/30000 [12:53:45<3:25:48, 1.32s/it] 69%|██████▉ | 20654/30000 [12:53:52<8:04:11, 3.11s/it] 69%|██████▉ | 20655/30000 [12:53:54<6:34:11, 2.53s/it] 69%|██████▉ | 20656/30000 [12:53:55<5:31:05, 2.13s/it] 69%|██████▉ | 20657/30000 [12:53:56<4:46:34, 1.84s/it] 69%|██████▉ | 20658/30000 [12:53:57<4:15:31, 1.64s/it] 69%|██████▉ | 20659/30000 [12:53:58<3:54:16, 1.50s/it] 69%|██████▉ | 20660/30000 [12:53:59<3:39:38, 1.41s/it] 69%|██████▉ | 20660/30000 [12:53:59<3:39:38, 1.41s/it] 69%|██████▉ | 20661/30000 [12:54:01<3:28:54, 1.34s/it] 69%|██████▉ | 20662/30000 [12:54:08<8:24:54, 3.24s/it] 69%|██████▉ | 20663/30000 [12:54:09<6:47:39, 2.62s/it] 69%|██████▉ | 20664/30000 [12:54:11<5:41:05, 2.19s/it] 69%|██████▉ | 20665/30000 [12:54:12<4:53:20, 1.89s/it] 69%|██████▉ | 20666/30000 [12:54:13<4:20:11, 1.67s/it] 69%|██████▉ | 20667/30000 [12:54:14<3:57:19, 1.53s/it] 69%|██████▉ | 20668/30000 [12:54:15<3:41:33, 1.42s/it] 69%|██████▉ | 20669/30000 [12:54:17<3:30:22, 1.35s/it] 69%|██████▉ | 20670/30000 [12:54:23<7:46:57, 3.00s/it] 69%|██████▉ | 20670/30000 [12:54:23<7:46:57, 3.00s/it] 69%|██████▉ | 20671/30000 [12:54:25<6:21:18, 2.45s/it] 69%|██████▉ | 20672/30000 [12:54:26<5:22:20, 2.07s/it] 69%|██████▉ | 20673/30000 [12:54:27<4:40:05, 1.80s/it] 69%|██████▉ | 20674/30000 [12:54:28<4:10:17, 1.61s/it] 69%|██████▉ | 20675/30000 [12:54:29<3:50:59, 1.49s/it] 69%|██████▉ | 20676/30000 [12:54:30<3:36:46, 1.39s/it] 69%|██████▉ | 20677/30000 [12:54:32<3:26:55, 1.33s/it] 69%|██████▉ | 20678/30000 [12:54:38<6:57:53, 2.69s/it] 69%|██████▉ | 20679/30000 [12:54:39<5:46:17, 2.23s/it] 69%|██████▉ | 20680/30000 [12:54:40<4:58:02, 1.92s/it] 69%|██████▉ | 20680/30000 [12:54:40<4:58:02, 1.92s/it] 69%|██████▉ | 20681/30000 [12:54:41<4:23:04, 1.69s/it] 69%|██████▉ | 20682/30000 [12:54:42<3:58:54, 1.54s/it] 69%|██████▉ | 20683/30000 [12:54:43<3:42:46, 1.43s/it] 69%|██████▉ | 20684/30000 [12:54:45<3:32:00, 1.37s/it] 69%|██████▉ | 20685/30000 [12:54:46<3:23:29, 1.31s/it] 69%|██████▉ | 20686/30000 [12:54:54<8:21:39, 3.23s/it] 69%|██████▉ | 20687/30000 [12:54:55<6:45:43, 2.61s/it] 69%|██████▉ | 20688/30000 [12:54:56<5:39:39, 2.19s/it] 69%|██████▉ | 20689/30000 [12:54:57<4:52:01, 1.88s/it] 69%|██████▉ | 20690/30000 [12:54:58<4:21:14, 1.68s/it] +{'loss': 0.0133, 'grad_norm': 0.07332953810691833, 'learning_rate': 7.81865433804676e-06, 'flow_matching_loss': 0.013299732701852918, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.9} +{'loss': 0.0141, 'grad_norm': 0.08378478139638901, 'learning_rate': 7.80414210218461e-06, 'flow_matching_loss': 0.014063518401235342, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.9} +{'loss': 0.0135, 'grad_norm': 0.07136157900094986, 'learning_rate': 7.789638609968332e-06, 'flow_matching_loss': 0.013520581554621458, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.91} +{'loss': 0.0145, 'grad_norm': 0.07059333473443985, 'learning_rate': 7.775143879021037e-06, 'flow_matching_loss': 0.014487337227910757, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.91} +{'loss': 0.014, 'grad_norm': 0.08462020009756088, 'learning_rate': 7.760657926955187e-06, 'flow_matching_loss': 0.014042341988533736, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.91} +{'loss': 0.0133, 'grad_norm': 0.0887245386838913, 'learning_rate': 7.746180771372584e-06, 'flow_matching_loss': 0.01333738248795271, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.92} +{'loss': 0.0137, 'grad_norm': 0.0699739158153534, 'learning_rate': 7.731712429864325e-06, 'flow_matching_loss': 0.013730249833315611, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.92} +{'loss': 0.0138, 'grad_norm': 0.08325962722301483, 'learning_rate': 7.717252920010828e-06, 'flow_matching_loss': 0.013774011377245187, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.92} +{'loss': 0.0153, 'grad_norm': 0.08618428558111191, 'learning_rate': 7.702802259381742e-06, 'flow_matching_loss': 0.015258386079221965, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.92} +{'loss': 0.0148, 'grad_norm': 0.09495894610881805, 'learning_rate': 7.688360465535999e-06, 'flow_matching_loss': 0.014751589111983775, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.93} +{'loss': 0.0139, 'grad_norm': 0.08616695553064346, 'learning_rate': 7.673927556021734e-06, 'flow_matching_loss': 0.01393575118854642, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.93} +{'loss': 0.015, 'grad_norm': 0.08938861638307571, 'learning_rate': 7.659503548376284e-06, 'flow_matching_loss': 0.01504011070355773, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.93} +{'loss': 0.0132, 'grad_norm': 0.09314657002687454, 'learning_rate': 7.645088460126193e-06, 'flow_matching_loss': 0.01316290646791458, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.94} +{'loss': 0.014, 'grad_norm': 0.05843929573893547, 'learning_rate': 7.630682308787145e-06, 'flow_matching_loss': 0.014024066366255283, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.94} +{'loss': 0.0141, 'grad_norm': 0.06680796295404434, 'learning_rate': 7.6162851118639754e-06, 'flow_matching_loss': 0.014066519681364298, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.94} +{'loss': 0.013, 'grad_norm': 0.060835324227809906, 'learning_rate': 7.601896886850632e-06, 'flow_matching_loss': 0.01297402698546648, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.94} +{'loss': 0.015, 'grad_norm': 0.15657220780849457, 'learning_rate': 7.587517651230168e-06, 'flow_matching_loss': 0.014955730922520161, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.95} +{'loss': 0.0144, 'grad_norm': 0.08897427469491959, 'learning_rate': 7.573147422474702e-06, 'flow_matching_loss': 0.014421947579830886, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.95} +{'loss': 0.013, 'grad_norm': 0.059640828520059586, 'learning_rate': 7.55878621804543e-06, 'flow_matching_loss': 0.01303082974627614, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.95} +{'loss': 0.0139, 'grad_norm': 0.06590395420789719, 'learning_rate': 7.544434055392564e-06, 'flow_matching_loss': 0.013911883253604174, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.96} +{'loss': 0.0139, 'grad_norm': 0.06536798179149628, 'learning_rate': 7.530090951955334e-06, 'flow_matching_loss': 0.013857920933514834, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.96} +{'loss': 0.014, 'grad_norm': 0.06690935045480728, 'learning_rate': 7.515756925161959e-06, 'flow_matching_loss': 0.014002930279821158, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.96} +{'loss': 0.0134, 'grad_norm': 0.10565229505300522, 'learning_rate': 7.501431992429643e-06, 'flow_matching_loss': 0.013378847390413284, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.96} +{'loss': 0.0137, 'grad_norm': 0.06711839884519577, 'learning_rate': 7.487116171164519e-06, 'flow_matching_loss': 0.013727994449436665, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.97} +{'loss': 0.0143, 'grad_norm': 0.07583244144916534, 'learning_rate': 7.472809478761672e-06, 'flow_matching_loss': 0.014269156102091073, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.97} +{'loss': 0.0134, 'grad_norm': 0.0664479061961174, 'learning_rate': 7.4585119326050775e-06, 'flow_matching_loss': 0.013395992852747441, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.97} +{'loss': 0.0131, 'grad_norm': 0.06291623413562775, 'learning_rate': 7.444223550067599e-06, 'flow_matching_loss': 0.013059158436954021, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.98} +{'loss': 0.0128, 'grad_norm': 0.06246177479624748, 'learning_rate': 7.429944348510964e-06, 'flow_matching_loss': 0.012821360025554895, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.98} +{'loss': 0.0149, 'grad_norm': 0.10775981843471527, 'learning_rate': 7.415674345285763e-06, 'flow_matching_loss': 0.014922027755528689, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.98} +{'loss': 0.0135, 'grad_norm': 0.06878874450922012, 'learning_rate': 7.401413557731387e-06, 'flow_matching_loss': 0.013516869209706784, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.98} +{'loss': 0.0127, 'grad_norm': 0.07063031941652298, 'learning_rate': 7.387162003176037e-06, 'flow_matching_loss': 0.012658418715000152, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.99} +{'loss': 0.0135, 'grad_norm': 0.06531249731779099, 'learning_rate': 7.372919698936697e-06, 'flow_matching_loss': 0.013461919967085123, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.99} +{'loss': 0.0152, 'grad_norm': 0.08464805036783218, 'learning_rate': 7.358686662319104e-06, 'flow_matching_loss': 0.015170445758849382, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 5.99} +{'loss': 0.0128, 'grad_norm': 0.06139141321182251, 'learning_rate': 7.344462910617752e-06, 'flow_matching_loss': 0.012818939611315727, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.0} +{'loss': 0.013, 'grad_norm': 0.06412608921527863, 'learning_rate': 7.330248461115832e-06, 'flow_matching_loss': 0.013043529726564885, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.0} +{'loss': 0.0134, 'grad_norm': 0.06606021523475647, 'learning_rate': 7.316043331085242e-06, 'flow_matching_loss': 0.01338115893304348, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.0} +{'loss': 0.0146, 'grad_norm': 0.08576945215463638, 'learning_rate': 7.30184753778655e-06, 'flow_matching_loss': 0.014607445616275073, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.01} +{'loss': 0.0147, 'grad_norm': 0.0753631666302681, 'learning_rate': 7.2876610984689935e-06, 'flow_matching_loss': 0.014650371577590703, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.01} +{'loss': 0.0135, 'grad_norm': 0.06642290949821472, 'learning_rate': 7.273484030370423e-06, 'flow_matching_loss': 0.013512080255895852, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.01} +{'loss': 0.0143, 'grad_norm': 0.08033954352140427, 'learning_rate': 7.259316350717324e-06, 'flow_matching_loss': 0.014321845024824142, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.01} +{'loss': 0.0129, 'grad_norm': 0.06816756725311279, 'learning_rate': 7.245158076724757e-06, 'flow_matching_loss': 0.01292652329429984, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.02} +{'loss': 0.0134, 'grad_norm': 0.05639943107962608, 'learning_rate': 7.23100922559636e-06, 'flow_matching_loss': 0.013356755953282118, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.02} 69%|██████▉ | 20690/30000 [12:54:58<4:21:14, 1.68s/it] 69%|██████▉ | 20691/30000 [12:54:59<3:59:56, 1.55s/it] 69%|██████▉ | 20692/30000 [12:55:01<3:43:49, 1.44s/it] 69%|██████▉ | 20693/30000 [12:55:02<3:31:59, 1.37s/it] 69%|██████▉ | 20694/30000 [12:55:10<8:38:32, 3.34s/it] 69%|██████▉ | 20695/30000 [12:55:11<6:56:49, 2.69s/it] 69%|██████▉ | 20696/30000 [12:55:12<5:47:10, 2.24s/it] 69%|██████▉ | 20697/30000 [12:55:13<4:56:47, 1.91s/it] 69%|██████▉ | 20698/30000 [12:55:15<4:21:22, 1.69s/it] 69%|██████▉ | 20699/30000 [12:55:16<3:58:18, 1.54s/it] 69%|██████▉ | 20700/30000 [12:55:17<3:42:22, 1.43s/it] 69%|██████▉ | 20700/30000 [12:55:17<3:42:22, 1.43s/it] 69%|██████▉ | 20701/30000 [12:55:18<3:31:08, 1.36s/it] 69%|██████▉ | 20702/30000 [12:55:25<8:02:59, 3.12s/it] 69%|██████▉ | 20703/30000 [12:55:26<6:32:23, 2.53s/it] 69%|██████▉ | 20704/30000 [12:55:28<5:29:37, 2.13s/it] 69%|██████▉ | 20705/30000 [12:55:29<4:44:52, 1.84s/it] 69%|██████▉ | 20706/30000 [12:55:30<4:13:44, 1.64s/it] 69%|██████▉ | 20707/30000 [12:55:31<3:52:55, 1.50s/it] 69%|██████▉ | 20708/30000 [12:55:32<3:38:20, 1.41s/it] 69%|██████▉ | 20709/30000 [12:55:34<3:28:04, 1.34s/it] 69%|██████▉ | 20710/30000 [12:55:40<7:30:27, 2.91s/it] 69%|██████▉ | 20710/30000 [12:55:40<7:30:27, 2.91s/it] 69%|██████▉ | 20711/30000 [12:55:41<6:09:58, 2.39s/it] 69%|██████▉ | 20712/30000 [12:55:42<5:14:34, 2.03s/it] 69%|██████▉ | 20713/30000 [12:55:44<4:33:36, 1.77s/it] 69%|██████▉ | 20714/30000 [12:55:45<4:05:44, 1.59s/it] 69%|██████▉ | 20715/30000 [12:55:46<3:47:33, 1.47s/it] 69%|██████▉ | 20716/30000 [12:55:47<3:34:34, 1.39s/it] 69%|██████▉ | 20717/30000 [12:55:48<3:25:06, 1.33s/it] 69%|██████▉ | 20718/30000 [12:55:56<8:13:46, 3.19s/it] 69%|██████▉ | 20719/30000 [12:55:57<6:41:13, 2.59s/it] 69%|██████▉ | 20720/30000 [12:55:58<5:36:09, 2.17s/it] 69%|██████▉ | 20720/30000 [12:55:58<5:36:09, 2.17s/it] 69%|██████▉ | 20721/30000 [12:55:59<4:49:08, 1.87s/it] 69%|██████▉ | 20722/30000 [12:56:01<4:17:24, 1.66s/it] 69%|██████▉ | 20723/30000 [12:56:02<3:54:56, 1.52s/it] 69%|██████▉ | 20724/30000 [12:56:03<3:39:59, 1.42s/it] 69%|██████▉ | 20725/30000 [12:56:04<3:28:52, 1.35s/it] 69%|██████▉ | 20726/30000 [12:56:12<8:24:32, 3.26s/it] 69%|██████▉ | 20727/30000 [12:56:13<6:47:16, 2.64s/it] 69%|██████▉ | 20728/30000 [12:56:14<5:39:59, 2.20s/it] 69%|██████▉ | 20729/30000 [12:56:15<4:51:34, 1.89s/it] 69%|██████▉ | 20730/30000 [12:56:17<4:17:41, 1.67s/it] 69%|██████▉ | 20730/30000 [12:56:17<4:17:41, 1.67s/it] 69%|██████▉ | 20731/30000 [12:56:18<3:55:33, 1.52s/it] 69%|██████▉ | 20732/30000 [12:56:19<3:39:20, 1.42s/it] 69%|██████▉ | 20733/30000 [12:56:20<3:27:48, 1.35s/it] 69%|██████▉ | 20734/30000 [12:56:28<8:39:46, 3.37s/it] 69%|██████▉ | 20735/30000 [12:56:29<6:58:00, 2.71s/it] 69%|██████▉ | 20736/30000 [12:56:31<5:48:35, 2.26s/it] 69%|██████▉ | 20737/30000 [12:56:32<4:57:54, 1.93s/it] 69%|██████▉ | 20738/30000 [12:56:33<4:21:41, 1.70s/it] 69%|██████▉ | 20739/30000 [12:56:34<3:58:08, 1.54s/it] 69%|██████▉ | 20740/30000 [12:56:35<3:41:22, 1.43s/it] 69%|██████▉ | 20740/30000 [12:56:35<3:41:22, 1.43s/it] 69%|██████▉ | 20741/30000 [12:56:36<3:30:26, 1.36s/it] 69%|██████▉ | 20742/30000 [12:56:44<8:07:02, 3.16s/it] 69%|██████▉ | 20743/30000 [12:56:45<6:34:13, 2.56s/it] 69%|██████▉ | 20744/30000 [12:56:46<5:31:18, 2.15s/it] 69%|██████▉ | 20745/30000 [12:56:47<4:45:26, 1.85s/it] 69%|██████▉ | 20746/30000 [12:56:49<4:13:50, 1.65s/it] 69%|██████▉ | 20747/30000 [12:56:50<3:53:26, 1.51s/it] 69%|██████▉ | 20748/30000 [12:56:51<3:38:40, 1.42s/it] 69%|██████▉ | 20749/30000 [12:56:52<3:28:13, 1.35s/it] 69%|██████▉ | 20750/30000 [12:56:59<7:38:46, 2.98s/it] 69%|██████▉ | 20750/30000 [12:56:59<7:38:46, 2.98s/it] 69%|██████▉ | 20751/30000 [12:57:00<6:16:52, 2.44s/it] 69%|██████▉ | 20752/30000 [12:57:01<5:20:17, 2.08s/it] 69%|██████▉ | 20753/30000 [12:57:02<4:38:07, 1.80s/it] 69%|██████▉ | 20754/30000 [12:57:04<4:08:54, 1.62s/it] 69%|██████▉ | 20755/30000 [12:57:05<3:48:38, 1.48s/it] 69%|██████▉ | 20756/30000 [12:57:06<3:36:13, 1.40s/it] 69%|██████▉ | 20757/30000 [12:57:07<3:26:15, 1.34s/it] 69%|██████▉ | 20758/30000 [12:57:14<7:48:29, 3.04s/it] 69%|██████▉ | 20759/30000 [12:57:15<6:21:57, 2.48s/it] 69%|██████▉ | 20760/30000 [12:57:17<5:22:13, 2.09s/it] 69%|██████▉ | 20760/30000 [12:57:17<5:22:13, 2.09s/it] 69%|██████▉ | 20761/30000 [12:57:18<4:39:50, 1.82s/it] 69%|██████▉ | 20762/30000 [12:57:19<4:09:43, 1.62s/it] 69%|██████▉ | 20763/30000 [12:57:20<3:49:04, 1.49s/it] 69%|██████▉ | 20764/30000 [12:57:21<3:35:05, 1.40s/it] 69%|██████▉ | 20765/30000 [12:57:22<3:25:12, 1.33s/it] 69%|██████▉ | 20766/30000 [12:57:30<7:53:29, 3.08s/it] 69%|██████▉ | 20767/30000 [12:57:31<6:24:58, 2.50s/it] 69%|██████▉ | 20768/30000 [12:57:32<5:25:00, 2.11s/it] 69%|██████▉ | 20769/30000 [12:57:33<4:41:05, 1.83s/it] 69%|██████▉ | 20770/30000 [12:57:34<4:10:42, 1.63s/it] 69%|██████▉ | 20770/30000 [12:57:34<4:10:42, 1.63s/it] 69%|██████▉ | 20771/30000 [12:57:36<3:50:43, 1.50s/it] 69%|██████▉ | 20772/30000 [12:57:37<3:36:35, 1.41s/it] 69%|██████▉ | 20773/30000 [12:57:38<3:26:50, 1.34s/it] 69%|██████▉ | 20774/30000 [12:57:45<7:46:11, 3.03s/it] 69%|██████▉ | 20775/30000 [12:57:46<6:20:15, 2.47s/it] 69%|██████▉ | 20776/30000 [12:57:47<5:20:51, 2.09s/it] 69%|██████▉ | 20777/30000 [12:57:48<4:38:41, 1.81s/it] 69%|██████▉ | 20778/30000 [12:57:50<4:08:44, 1.62s/it] 69%|██████▉ | 20779/30000 [12:57:51<3:48:48, 1.49s/it] 69%|██████▉ | 20780/30000 [12:57:52<3:36:08, 1.41s/it] 69%|██████▉ | 20780/30000 [12:57:52<3:36:08, 1.41s/it] 69%|██████▉ | 20781/30000 [12:57:53<3:26:05, 1.34s/it] 69%|██████▉ | 20782/30000 [12:58:00<7:51:10, 3.07s/it] 69%|██████▉ | 20783/30000 [12:58:01<6:23:50, 2.50s/it] 69%|██████▉ | 20784/30000 [12:58:03<5:22:42, 2.10s/it] 69%|██████▉ | 20785/30000 [12:58:04<4:40:24, 1.83s/it] 69%|██████▉ | 20786/30000 [12:58:05<4:10:15, 1.63s/it] 69%|██████▉ | 20787/30000 [12:58:06<3:50:19, 1.50s/it] 69%|██████▉ | 20788/30000 [12:58:07<3:36:09, 1.41s/it] 69%|██████▉ | 20789/30000 [12:58:09<3:26:12, 1.34s/it] 69%|██████▉ | 20790/30000 [12:58:16<7:53:59, 3.09s/it] 69%|██████▉ | 20790/30000 [12:58:16<7:53:59, 3.09s/it] 69%|██████▉ | 20791/30000 [12:58:17<6:25:18, 2.51s/it] 69%|██████▉ | 20792/30000 [12:58:18<5:24:44, 2.12s/it] 69%|██████▉ | 20793/30000 [12:58:19<4:40:26, 1.83s/it] 69%|██████▉ | 20794/30000 [12:58:20<4:09:45, 1.63s/it] 69%|██████▉ | 20795/30000 [12:58:22<3:49:34, 1.50s/it] 69%|██████▉ | 20796/30000 [12:58:23<3:35:29, 1.40s/it] 69%|██████▉ | 20797/30000 [12:58:24<3:25:24, 1.34s/it] 69%|██████▉ | 20798/30000 [12:58:30<7:18:52, 2.86s/it] 69%|██████▉ | 20799/30000 [12:58:32<6:00:21, 2.35s/it] 69%|██████▉ | 20800/30000 [12:58:33<5:07:12, 2.00s/it] 69%|██████▉ | 20800/30000 [12:58:33<5:07:12, 2.00s/it] 69%|██████▉ | 20801/30000 [12:58:34<4:29:13, 1.76s/it] 69%|██████▉ | 20802/30000 [12:58:35<4:02:25, 1.58s/it] 69%|██████▉ | 20803/30000 [12:58:36<3:44:27, 1.46s/it] 69%|██████▉ | 20804/30000 [12:58:37<3:31:59, 1.38s/it] 69%|██████▉ | 20805/30000 [12:58:39<3:22:58, 1.32s/it] 69%|██████▉ | 20806/30000 [12:58:45<7:33:58, 2.96s/it] 69%|██████▉ | 20807/30000 [12:58:47<6:11:29, 2.42s/it] 69%|██████▉ | 20808/30000 [12:58:48<5:14:45, 2.05s/it] 69%|██████▉ | 20809/30000 [12:58:49<4:33:48, 1.79s/it] 69%|██████▉ | 20810/30000 [12:58:50<4:05:01, 1.60s/it] 69%|██████▉ | 20810/30000 [12:58:50<4:05:01, 1.60s/it] 69%|██████▉ | 20811/30000 [12:58:51<3:46:32, 1.48s/it] 69%|██████▉ | 20812/30000 [12:58:52<3:33:25, 1.39s/it] 69%|██████▉ | 20813/30000 [12:58:54<3:24:02, 1.33s/it] 69%|██████▉ | 20814/30000 [12:59:01<7:37:47, 2.99s/it] 69%|██████▉ | 20815/30000 [12:59:02<6:13:38, 2.44s/it] 69%|██████▉ | 20816/30000 [12:59:03<5:16:11, 2.07s/it] 69%|██████▉ | 20817/30000 [12:59:04<4:34:47, 1.80s/it] 69%|██████▉ | 20818/30000 [12:59:05<4:05:27, 1.60s/it] 69%|██████▉ | 20819/30000 [12:59:06<3:46:34, 1.48s/it] 69%|██████▉ | 20820/30000 [12:59:08<3:33:44, 1.40s/it] 69%|██████▉ | 20820/30000 [12:59:08<3:33:44, 1.40s/it] 69%|██████▉ | 20821/30000 [12:59:09<3:24:34, 1.34s/it] 69%|██████▉ | 20822/30000 [12:59:17<8:41:17, 3.41s/it] 69%|██████▉ | 20823/30000 [12:59:18<6:58:00, 2.73s/it] 69%|██████▉ | 20824/30000 [12:59:19<5:47:06, 2.27s/it] 69%|██████▉ | 20825/30000 [12:59:21<4:59:54, 1.96s/it] 69%|██████▉ | 20826/30000 [12:59:22<4:22:25, 1.72s/it] 69%|██████▉ | 20827/30000 [12:59:23<3:58:30, 1.56s/it] 69%|██████▉ | 20828/30000 [12:59:24<3:41:20, 1.45s/it] 69%|██████▉ | 20829/30000 [12:59:25<3:29:45, 1.37s/it] 69%|██████▉ | 20830/30000 [12:59:33<7:55:47, 3.11s/it] 69%|██████▉ | 20830/30000 [12:59:33<7:55:47, 3.11s/it] 69%|██████▉ | 20831/30000 [12:59:34<6:27:01, 2.53s/it] 69%|██████▉ | 20832/30000 [12:59:35<5:25:22, 2.13s/it] 69%|██████▉ | 20833/30000 [12:59:36<4:41:14, 1.84s/it] 69%|██████▉ | 20834/30000 [12:59:37<4:11:33, 1.65s/it] 69%|██████▉ | 20835/30000 [12:59:38<3:50:00, 1.51s/it] 69%|██████▉ | 20836/30000 [12:59:40<3:35:13, 1.41s/it] 69%|██████▉ | 20837/30000 [12:59:41<3:25:03, 1.34s/it] 69%|██████▉ | 20838/30000 [12:59:48<8:14:25, 3.24s/it] 69%|██████▉ | 20839/30000 [12:59:50<6:39:32, 2.62s/it] 69%|██████▉ | 20840/30000 [12:59:51<5:34:22, 2.19s/it] 69%|██████▉ | 20840/30000 [12:59:51<5:34:22, 2.19s/it] 69%|██████▉ | 20841/30000 [12:59:52<4:47:27, 1.88s/it] 69%|██████▉ | 20842/30000 [12:59:53<4:14:54, 1.67s/it] 69%|██████▉ | 20843/30000 [12:59:54<3:52:47, 1.53s/it] 69%|██████▉ | 20844/30000 [12:59:56<3:37:40, 1.43s/it] 69%|██████▉ | 20845/30000 [12:59:57<3:27:52, 1.36s/it] 69%|██████▉ | 20846/30000 [13:00:04<8:14:02, 3.24s/it] 69%|██████▉ | 20847/30000 [13:00:06<6:40:02, 2.62s/it] 69%|██████▉ | 20848/30000 [13:00:07<5:34:20, 2.19s/it] 69%|██████▉ | 20849/30000 [13:00:08<4:46:17, 1.88s/it] 70%|██████▉ | 20850/30000 [13:00:09<4:14:03, 1.67s/it] 70%|██████▉ | 20850/30000 [13:00:09<4:14:03, 1.67s/it] 70%|██████▉ | 20851/30000 [13:00:10<3:52:14, 1.52s/it] 70%|██████▉ | 20852/30000 [13:00:11<3:37:08, 1.42s/it] 70%|██████▉ | 20853/30000 [13:00:13<3:26:24, 1.35s/it] 70%|██████▉ | 20854/30000 [13:00:21<8:39:55, 3.41s/it] 70%|██████▉ | 20855/30000 [13:00:22<6:57:47, 2.74s/it] 70%|██████▉ | 20856/30000 [13:00:23<5:46:33, 2.27s/it] 70%|██████▉ | 20857/30000 [13:00:24<4:55:41, 1.94s/it] 70%|██████▉ | 20858/30000 [13:00:26<4:20:29, 1.71s/it] 70%|██████▉ | 20859/30000 [13:00:27<3:56:34, 1.55s/it] 70%|██████▉ | 20860/30000 [13:00:28<3:39:55, 1.44s/it] 70%|██████▉ | 20860/30000 [13:00:28<3:39:55, 1.44s/it] 70%|██████▉ | 20861/30000 [13:00:29<3:27:55, 1.37s/it] 70%|██████▉ | 20862/30000 [13:00:36<8:00:07, 3.15s/it] 70%|██████▉ | 20863/30000 [13:00:38<6:28:21, 2.55s/it] 70%|██████▉ | 20864/30000 [13:00:39<5:26:18, 2.14s/it] 70%|██████▉ | 20865/30000 [13:00:40<4:41:43, 1.85s/it] 70%|██████▉ | 20866/30000 [13:00:41<4:10:24, 1.64s/it] 70%|██████▉ | 20867/30000 [13:00:42<3:49:14, 1.51s/it] 70%|██████▉ | 20868/30000 [13:00:43<3:34:29, 1.41s/it] 70%|██████▉ | 20869/30000 [13:00:45<3:25:48, 1.35s/it] 70%|██████▉ | 20870/30000 [13:00:53<8:29:04, 3.35s/it] 70%|██████▉ | 20870/30000 [13:00:53<8:29:04, 3.35s/it] 70%|██████▉ | 20871/30000 [13:00:54<6:49:46, 2.69s/it] 70%|██████▉ | 20872/30000 [13:00:55<5:41:21, 2.24s/it] 70%|██████▉ | 20873/30000 [13:00:56<4:51:17, 1.91s/it] 70%|██████▉ | 20874/30000 [13:00:57<4:16:50, 1.69s/it] 70%|██████▉ | 20875/30000 [13:00:59<3:54:17, 1.54s/it] 70%|██████▉ | 20876/30000 [13:01:00<3:38:18, 1.44s/it] 70%|██████▉ | 20877/30000 [13:01:01<3:27:13, 1.36s/it] 70%|██████▉ | 20878/30000 [13:01:08<8:08:34, 3.21s/it] 70%|██████▉ | 20879/30000 [13:01:10<6:35:41, 2.60s/it] 70%|██████▉ | 20880/30000 [13:01:11<5:31:09, 2.18s/it] 70%|██████▉ | 20880/30000 [13:01:11<5:31:09, 2.18s/it] 70%|██████▉ | 20881/30000 [13:01:12<4:45:29, 1.88s/it] 70%|██████▉ | 20882/30000 [13:01:13<4:13:57, 1.67s/it] 70%|██████▉ | 20883/30000 [13:01:14<3:52:13, 1.53s/it] 70%|██████▉ | 20884/30000 [13:01:16<3:36:53, 1.43s/it] 70%|██████▉ | 20885/30000 [13:01:17<3:25:53, 1.36s/it] 70%|██████▉ | 20886/30000 [13:01:24<7:47:17, 3.08s/it] 70%|██████▉ | 20887/30000 [13:01:25<6:20:47, 2.51s/it] 70%|██████▉ | 20888/30000 [13:01:26<5:20:40, 2.11s/it] 70%|██████▉ | 20889/30000 [13:01:27<4:37:34, 1.83s/it] 70%|██████▉ | 20890/30000 [13:01:29<4:07:52, 1.63s/it] 70%|██████▉ | 20890/30000 [13:01:29<4:07:52, 1.63s/it] 70%|██████▉ | 20891/30000 [13:01:30<3:48:07, 1.50s/it] 70%|██████▉ | 20892/30000 [13:01:31<3:34:10, 1.41s/it] 70%|██████▉ | 20893/30000 [13:01:32<3:23:50, 1.34s/it] 70%|██████▉ | 20894/30000 [13:01:39<7:32:00, 2.98s/it] 70%|██████▉ | 20895/30000 [13:01:40<6:09:14, 2.43s/it] 70%|██████▉ | 20896/30000 [13:01:41<5:12:13, 2.06s/it] 70%|██████▉ | 20897/30000 [13:01:42<4:31:49, 1.79s/it] 70%|██████▉ | 20898/30000 [13:01:44<4:02:44, 1.60s/it] 70%|██████▉ | 20899/30000 [13:01:45<3:43:30, 1.47s/it] 70%|██████▉ | 20900/30000 [13:01:46<3:30:04, 1.39s/it] 70%|██████▉ | 20900/30000 [13:01:46<3:30:04, 1.39s/it] 70%|██████▉ | 20901/30000 [13:01:47<3:21:25, 1.33s/it] 70%|██████▉ | 20902/30000 [13:01:55<8:36:13, 3.40s/it] 70%|██████▉ | 20903/30000 [13:01:57<6:54:28, 2.73s/it] 70%|██████▉ | 20904/30000 [13:01:58<5:44:54, 2.28s/it] 70%|██████▉ | 20905/30000 [13:01:59<4:54:15, 1.94s/it] 70%|██████▉ | 20906/30000 [13:02:00<4:18:35, 1.71s/it] 70%|██████▉ | 20907/30000 [13:02:01<3:54:54, 1.55s/it] 70%|██████▉ | 20908/30000 [13:02:02<3:38:23, 1.44s/it] 70%|██████▉ | 20909/30000 [13:02:04<3:26:45, 1.36s/it] 70%|██████▉ | 20910/30000 [13:02:11<7:55:11, 3.14s/it] 70%|██████▉ | 20910/30000 [13:02:11<7:55:11, 3.14s/it] 70%|██████▉ | 20911/30000 [13:02:12<6:24:43, 2.54s/it] 70%|██████▉ | 20912/30000 [13:02:13<5:23:13, 2.13s/it] 70%|██████▉ | 20913/30000 [13:02:14<4:39:21, 1.84s/it] 70%|██████▉ | 20914/30000 [13:02:16<4:08:10, 1.64s/it] 70%|██████▉ | 20915/30000 [13:02:17<3:47:30, 1.50s/it] 70%|██████▉ | 20916/30000 [13:02:18<3:33:31, 1.41s/it] 70%|██████▉ | 20917/30000 [13:02:19<3:24:14, 1.35s/it] 70%|██████▉ | 20918/30000 [13:02:27<8:02:36, 3.19s/it] 70%|██████▉ | 20919/30000 [13:02:28<6:30:10, 2.58s/it] 70%|██████▉ | 20920/30000 [13:02:29<5:27:08, 2.16s/it] 70%|██████▉ | 20920/30000 [13:02:29<5:27:08, 2.16s/it] 70%|██████▉ | 20921/30000 [13:02:30<4:42:13, 1.87s/it] 70%|██████▉ | 20922/30000 [13:02:31<4:10:14, 1.65s/it] 70%|██████▉ | 20923/30000 [13:02:33<3:48:55, 1.51s/it] 70%|██████▉ | 20924/30000 [13:02:34<3:33:43, 1.41s/it] 70%|██████▉ | 20925/30000 [13:02:35<3:23:52, 1.35s/it] 70%|██████▉ | 20926/30000 [13:02:43<8:07:53, 3.23s/it] 70%|██████▉ | 20927/30000 [13:02:44<6:33:58, 2.61s/it] 70%|██████▉ | 20928/30000 [13:02:45<5:29:58, 2.18s/it] 70%|██████▉ | 20929/30000 [13:02:46<4:43:43, 1.88s/it] 70%|██████▉ | 20930/30000 [13:02:47<4:11:26, 1.66s/it] 70%|██████▉ | 20930/30000 [13:02:47<4:11:26, 1.66s/it] 70%|██████▉ | 20931/30000 [13:02:48<3:49:59, 1.52s/it] 70%|██████▉ | 20932/30000 [13:02:50<3:34:43, 1.42s/it] 70%|██████▉ | 20933/30000 [13:02:51<3:23:58, 1.35s/it] 70%|██████▉ | 20934/30000 [13:02:58<7:46:53, 3.09s/it] 70%|██████▉ | 20935/30000 [13:02:59<6:19:24, 2.51s/it] 70%|██████▉ | 20936/30000 [13:03:00<5:19:04, 2.11s/it] 70%|██████▉ | 20937/30000 [13:03:01<4:35:24, 1.82s/it] 70%|██████▉ | 20938/30000 [13:03:03<4:05:47, 1.63s/it] 70%|██████▉ | 20939/30000 [13:03:04<3:46:11, 1.50s/it] 70%|██████▉ | 20940/30000 [13:03:05<3:32:02, 1.40s/it] 70%|██████▉ | 20940/30000 [13:03:05<3:32:02, 1.40s/it] 70%|██████▉ | 20941/30000 [13:03:06<3:22:53, 1.34s/it] 70%|██████▉ | 20942/30000 [13:03:15<8:40:47, 3.45s/it] 70%|██████▉ | 20943/30000 [13:03:16<6:57:32, 2.77s/it] 70%|██████▉ | 20944/30000 [13:03:17<5:46:06, 2.29s/it] 70%|██████▉ | 20945/30000 [13:03:18<4:53:55, 1.95s/it] 70%|██████▉ | 20946/30000 [13:03:19<4:18:07, 1.71s/it] 70%|██████▉ | 20947/30000 [13:03:20<3:54:53, 1.56s/it] 70%|██████▉ | 20948/30000 [13:03:22<3:38:46, 1.45s/it] 70%|██████▉ | 20949/30000 [13:03:23<3:28:19, 1.38s/it] 70%|██████▉ | 20950/30000 [13:03:30<7:34:01, 3.01s/it] 70%|██████▉ | 20950/30000 [13:03:30<7:34:01, 3.01s/it] 70%|██████▉ | 20951/30000 [13:03:31<6:12:40, 2.47s/it] 70%|██████▉ | 20952/30000 [13:03:32<5:16:38, 2.10s/it] 70%|██████▉ | 20953/30000 [13:03:33<4:36:40, 1.83s/it] 70%|██████▉ | 20954/30000 [13:03:34<4:07:33, 1.64s/it] 70%|██████▉ | 20955/30000 [13:03:36<3:47:53, 1.51s/it] 70%|██████▉ | 20956/30000 [13:03:37<3:32:47, 1.41s/it] 70%|██████▉ | 20957/30000 [13:03:38<3:22:53, 1.35s/it] 70%|██████▉ | 20958/30000 [13:03:45<7:54:17, 3.15s/it] 70%|██████▉ | 20959/30000 [13:03:47<6:24:43, 2.55s/it] 70%|██████▉ | 20960/30000 [13:03:48<5:22:51, 2.14s/it] 70%|██████▉ | 20960/30000 [13:03:48<5:22:51, 2.14s/it] 70%|██████▉ | 20961/30000 [13:03:49<4:38:41, 1.85s/it] 70%|██████▉ | 20962/30000 [13:03:50<4:07:27, 1.64s/it] 70%|██████▉ | 20963/30000 [13:03:51<3:47:09, 1.51s/it] 70%|██████▉ | 20964/30000 [13:03:52<3:32:57, 1.41s/it] 70%|██████▉ | 20965/30000 [13:03:54<3:22:43, 1.35s/it] 70%|██████▉ | 20966/30000 [13:04:01<8:16:23, 3.30s/it] 70%|██████▉ | 20967/30000 [13:04:03<6:40:02, 2.66s/it] 70%|██████▉ | 20968/30000 [13:04:04<5:33:58, 2.22s/it] 70%|██████▉ | 20969/30000 [13:04:05<4:46:54, 1.91s/it] 70%|██████▉ | 20970/30000 [13:04:06<4:13:20, 1.68s/it] 70%|██████▉ | 20970/30000 [13:04:06<4:13:20, 1.68s/it] 70%|██████▉ | 20971/30000 [13:04:07<3:51:05, 1.54s/it] 70%|██████▉ | 20972/30000 [13:04:09<3:35:36, 1.43s/it] 70%|██████▉ | 20973/30000 [13:04:10<3:26:04, 1.37s/it] 70%|██████▉ | 20974/30000 [13:04:17<8:11:14, 3.27s/it] 70%|██████▉ | 20975/30000 [13:04:19<6:36:29, 2.64s/it] 70%|██████▉ | 20976/30000 [13:04:20<5:31:21, 2.20s/it] 70%|██████▉ | 20977/30000 [13:04:21<4:44:31, 1.89s/it] 70%|██████▉ | 20978/30000 [13:04:22<4:11:31, 1.67s/it] 70%|██████▉ | 20979/30000 [13:04:23<3:50:13, 1.53s/it] 70%|██████▉ | 20980/30000 [13:04:25<3:34:48, 1.43s/it] 70%|██████▉ | 20980/30000 [13:04:25<3:34:48, 1.43s/it] 70%|██████▉ | 20981/30000 [13:04:26<3:24:11, 1.36s/it] 70%|██████▉ | 20982/30000 [13:04:34<8:26:18, 3.37s/it] 70%|██████▉ | 20983/30000 [13:04:35<6:47:06, 2.71s/it] 70%|██████▉ | 20984/30000 [13:04:36<5:38:22, 2.25s/it] 70%|██████▉ | 20985/30000 [13:04:37<4:49:03, 1.92s/it] 70%|██████▉ | 20986/30000 [13:04:38<4:14:48, 1.70s/it] 70%|██████▉ | 20987/30000 [13:04:40<3:52:18, 1.55s/it] 70%|██████▉ | 20988/30000 [13:04:41<3:35:53, 1.44s/it] 70%|██████▉ | 20989/30000 [13:04:42<3:24:33, 1.36s/it] 70%|██████▉ | 20990/30000 [13:04:50<8:17:02, 3.31s/it] 70%|██████▉ | 20990/30000 [13:04:50<8:17:02, 3.31s/it] 70%|██████▉ | 20991/30000 [13:04:51<6:40:49, 2.67s/it] 70%|██████▉ | 20992/30000 [13:04:52<5:34:19, 2.23s/it] 70%|██████▉ | 20993/30000 [13:04:53<4:46:23, 1.91s/it] 70%|██████▉ | 20994/30000 [13:04:55<4:12:52, 1.68s/it] 70%|██████▉ | 20995/30000 [13:04:56<3:50:31, 1.54s/it] 70%|██████▉ | 20996/30000 [13:04:57<3:34:43, 1.43s/it] 70%|██████▉ | 20997/30000 [13:04:58<3:23:25, 1.36s/it] 70%|██████▉ | 20998/30000 [13:05:06<7:54:38, 3.16s/it] 70%|██████▉ | 20999/30000 [13:05:07<6:24:32, 2.56s/it] 70%|███████ | 21000/30000 [13:05:08<5:23:51, 2.16s/it] 70%|███████ | 21000/30000 [13:05:08<5:23:51, 2.16s/it] 70%|███████ | 21001/30000 [13:05:09<4:41:13, 1.87s/it] 70%|███████ | 21002/30000 [13:05:10<4:09:30, 1.66s/it] 70%|███████ | 21003/30000 [13:05:12<3:48:36, 1.52s/it] 70%|███████ | 21004/30000 [13:05:13<3:33:32, 1.42s/it] 70%|███████ | 21005/30000 [13:05:14<3:22:48, 1.35s/it] 70%|███████ | 21006/30000 [13:05:21<7:58:04, 3.19s/it] 70%|███████ | 21007/30000 [13:05:23<6:26:40, 2.58s/it] 70%|███████ | 21008/30000 [13:05:24<5:22:49, 2.15s/it] 70%|███████ | 21009/30000 [13:05:25<4:38:19, 1.86s/it] 70%|███████ | 21010/30000 [13:05:26<4:07:05, 1.65s/it] 70%|███████ | 21010/30000 [13:05:26<4:07:05, 1.65s/it] 70%|███████ | 21011/30000 [13:05:27<3:46:45, 1.51s/it] 70%|███████ | 21012/30000 [13:05:28<3:31:49, 1.41s/it] 70%|███████ | 21013/30000 [13:05:30<3:22:11, 1.35s/it] 70%|███████ | 21014/30000 [13:05:38<8:26:16, 3.38s/it] 70%|███████ | 21015/30000 [13:05:39<6:47:45, 2.72s/it] 70%|███████ | 21016/30000 [13:05:40<5:39:18, 2.27s/it] 70%|███████ | 21017/30000 [13:05:41<4:50:01, 1.94s/it] 70%|███████ | 21018/30000 [13:05:42<4:15:08, 1.70s/it] 70%|███████ | 21019/30000 [13:05:44<3:53:08, 1.56s/it] 70%|███████ | 21020/30000 [13:05:45<3:36:34, 1.45s/it] 70%|███████ | 21020/30000 [13:05:45<3:36:34, 1.45s/it] 70%|███████ | 21021/30000 [13:05:46<3:24:36, 1.37s/it] 70%|███████ | 21022/30000 [13:05:54<8:14:02, 3.30s/it] 70%|███████ | 21023/30000 [13:05:55<6:38:38, 2.66s/it] 70%|███████ | 21024/30000 [13:05:56<5:32:19, 2.22s/it] 70%|███████ | 21025/30000 [13:05:57<4:44:57, 1.90s/it] 70%|███████ | 21026/30000 [13:05:59<4:11:41, 1.68s/it] 70%|███████ | 21027/30000 [13:06:00<3:49:08, 1.53s/it] 70%|███████ | 21028/30000 [13:06:01<3:33:13, 1.43s/it] 70%|███████ | 21029/30000 [13:06:02<3:22:29, 1.35s/it] 70%|███████ | 21030/30000 [13:06:10<8:32:39, 3.43s/it] 70%|███████ | 21030/30000 [13:06:10<8:32:39, 3.43s/it] 70%|███████ | 21031/30000 [13:06:12<6:51:12, 2.75s/it] 70%|███████ | 21032/30000 [13:06:13<5:40:53, 2.28s/it] 70%|███████ | 21033/30000 [13:06:14<4:50:42, 1.95s/it] 70%|███████ | 21034/30000 [13:06:15<4:16:00, 1.71s/it] 70%|███████ | 21035/30000 [13:06:16<3:52:44, 1.56s/it] 70%|███████ | 21036/30000 [13:06:17<3:36:01, 1.45s/it] 70%|███████ | 21037/30000 [13:06:19<3:24:17, 1.37s/it] 70%|███████ | 21038/30000 [13:06:27<8:33:58, 3.44s/it] 70%|███████ | 21039/30000 [13:06:28<6:50:44, 2.75s/it] 70%|███████ | 21040/30000 [13:06:29<5:40:50, 2.28s/it] 70%|███████ | 21040/30000 [13:06:29<5:40:50, 2.28s/it] 70%|███████ | 21041/30000 [13:06:30<4:51:11, 1.95s/it] 70%|███████ | 21042/30000 [13:06:32<4:16:17, 1.72s/it] 70%|███████ | 21043/30000 [13:06:33<3:52:59, 1.56s/it] 70%|███████ | 21044/30000 [13:06:34<3:36:10, 1.45s/it] 70%|███████ | 21045/30000 [13:06:35<3:24:33, 1.37s/it] 70%|███████ | 21046/30000 [13:06:42<7:30:52, 3.02s/it] 70%|███████ | 21047/30000 [13:06:43<6:07:34, 2.46s/it] 70%|███████ | 21048/30000 [13:06:44<5:11:26, 2.09s/it] 70%|███████ | 21049/30000 [13:06:46<4:30:26, 1.81s/it] 70%|███████ | 21050/30000 [13:06:47<4:00:58, 1.62s/it] 70%|███████ | 21050/30000 [13:06:47<4:00:58, 1.62s/it] 70%|███████ | 21051/30000 [13:06:48<3:41:34, 1.49s/it] 70%|███████ | 21052/30000 [13:06:49<3:28:23, 1.40s/it] 70%|███████ | 21053/30000 [13:06:50<3:19:27, 1.34s/it] 70%|███████ | 21054/30000 [13:06:58<7:46:29, 3.13s/it] 70%|███████ | 21055/30000 [13:06:59<6:18:16, 2.54s/it] 70%|███████ | 21056/30000 [13:07:00<5:17:49, 2.13s/it] 70%|███████ | 21057/30000 [13:07:01<4:34:22, 1.84s/it] 70%|███████ | 21058/30000 [13:07:02<4:03:31, 1.63s/it] 70%|███████ | 21059/30000 [13:07:03<3:43:35, 1.50s/it] 70%|███████ | 21060/30000 [13:07:05<3:29:47, 1.41s/it] 70%|███████ | 21060/30000 [13:07:05<3:29:47, 1.41s/it] 70%|███████ | 21061/30000 [13:07:06<3:19:30, 1.34s/it] 70%|███████ | 21062/30000 [13:07:13<7:51:33, 3.17s/it] 70%|███████ | 21063/30000 [13:07:14<6:21:50, 2.56s/it] 70%|███████ | 21064/30000 [13:07:16<5:20:44, 2.15s/it] 70%|███████ | 21065/30000 [13:07:17<4:36:40, 1.86s/it] 70%|███████ | 21066/30000 [13:07:18<4:05:41, 1.65s/it] 70%|███████ | 21067/30000 [13:07:19<3:44:51, 1.51s/it] 70%|███████ | 21068/30000 [13:07:20<3:30:23, 1.41s/it] 70%|███████ | 21069/30000 [13:07:21<3:20:30, 1.35s/it] 70%|███████ | 21070/30000 [13:07:29<7:47:39, 3.14s/it] 70%|███████ | 21070/30000 [13:07:29<7:47:39, 3.14s/it] 70%|███████ | 21071/30000 [13:07:30<6:19:15, 2.55s/it] 70%|███████ | 21072/30000 [13:07:31<5:18:16, 2.14s/it] 70%|███████ | 21073/30000 [13:07:32<4:33:59, 1.84s/it] 70%|███████ | 21074/30000 [13:07:33<4:03:21, 1.64s/it] 70%|███████ | 21075/30000 [13:07:35<3:43:34, 1.50s/it] 70%|███████ | 21076/30000 [13:07:36<3:29:43, 1.41s/it] 70%|███████ | 21077/30000 [13:07:37<3:19:24, 1.34s/it] 70%|███████ | 21078/30000 [13:07:45<8:05:41, 3.27s/it] 70%|███████ | 21079/30000 [13:07:46<6:31:11, 2.63s/it] 70%|███████ | 21080/30000 [13:07:47<5:26:52, 2.20s/it] 70%|███████ | 21080/30000 [13:07:47<5:26:52, 2.20s/it] 70%|███████ | 21081/30000 [13:07:48<4:40:49, 1.89s/it] 70%|███████ | 21082/30000 [13:07:49<4:08:31, 1.67s/it] 70%|███████ | 21083/30000 [13:07:51<3:46:56, 1.53s/it] 70%|███████ | 21084/30000 [13:07:52<3:31:51, 1.43s/it] 70%|███████ | 21085/30000 [13:07:53<3:21:08, 1.35s/it] 70%|███████ | 21086/30000 [13:08:01<8:02:55, 3.25s/it] 70%|███████ | 21087/30000 [13:08:02<6:29:13, 2.62s/it] 70%|███████ | 21088/30000 [13:08:03<5:25:27, 2.19s/it] 70%|███████ | 21089/30000 [13:08:04<4:39:22, 1.88s/it] 70%|███████ | 21090/30000 [13:08:05<4:07:26, 1.67s/it] 70%|███████ | 21090/30000 [13:08:05<4:07:26, 1.67s/it] 70%|███████ | 21091/30000 [13:08:07<3:46:16, 1.52s/it] 70%|███████ | 21092/30000 [13:08:08<3:31:09, 1.42s/it] 70%|███████ | 21093/30000 [13:08:09<3:20:35, 1.35s/it] 70%|███████ | 21094/30000 [13:08:16<7:57:35, 3.22s/it] 70%|███████ | 21095/30000 [13:08:18<6:27:34, 2.61s/it] 70%|███████ | 21096/30000 [13:08:19<5:24:07, 2.18s/it] 70%|███████ | 21097/30000 [13:08:20<4:38:41, 1.88s/it] 70%|███████ | 21098/30000 [13:08:21<4:06:54, 1.66s/it] 70%|███████ | 21099/30000 [13:08:22<3:45:16, 1.52s/it] 70%|███████ | 21100/30000 [13:08:24<3:30:03, 1.42s/it] 70%|███████ | 21100/30000 [13:08:24<3:30:03, 1.42s/it] 70%|███████ | 21101/30000 [13:08:25<3:20:16, 1.35s/it] 70%|███████ | 21102/30000 [13:08:33<8:36:41, 3.48s/it] 70%|███████ | 21103/30000 [13:08:34<6:53:08, 2.79s/it] 70%|███████ | 21104/30000 [13:08:36<5:41:53, 2.31s/it] 70%|███████ | 21105/30000 [13:08:37<4:51:18, 1.96s/it] 70%|███████ | 21106/30000 [13:08:38<4:15:58, 1.73s/it] 70%|███████ | 21107/30000 [13:08:39<3:52:10, 1.57s/it] 70%|███████ | 21108/30000 [13:08:40<3:35:29, 1.45s/it] 70%|███████ | 21109/30000 [13:08:41<3:23:45, 1.38s/it] 70%|███████ | 21110/30000 [13:08:50<8:20:27, 3.38s/it] +{'loss': 0.0145, 'grad_norm': 0.08841076493263245, 'learning_rate': 7.216869814524323e-06, 'flow_matching_loss': 0.014520961977541446, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.02} +{'loss': 0.0138, 'grad_norm': 0.0628279596567154, 'learning_rate': 7.202739860689357e-06, 'flow_matching_loss': 0.013761247135698795, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.03} +{'loss': 0.0129, 'grad_norm': 0.07391525059938431, 'learning_rate': 7.188619381260698e-06, 'flow_matching_loss': 0.012860330194234848, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.03} +{'loss': 0.0142, 'grad_norm': 0.05804761126637459, 'learning_rate': 7.174508393396057e-06, 'flow_matching_loss': 0.014246062375605106, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.03} +{'loss': 0.014, 'grad_norm': 0.08102737367153168, 'learning_rate': 7.160406914241615e-06, 'flow_matching_loss': 0.014035649597644806, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.03} +{'loss': 0.013, 'grad_norm': 0.07982810586690903, 'learning_rate': 7.146314960931996e-06, 'flow_matching_loss': 0.013018817361444234, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.04} +{'loss': 0.0136, 'grad_norm': 0.10457555204629898, 'learning_rate': 7.132232550590247e-06, 'flow_matching_loss': 0.013623427972197533, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.04} +{'loss': 0.014, 'grad_norm': 0.06905245780944824, 'learning_rate': 7.118159700327839e-06, 'flow_matching_loss': 0.013978890888392926, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.04} +{'loss': 0.0131, 'grad_norm': 0.06040935963392258, 'learning_rate': 7.104096427244603e-06, 'flow_matching_loss': 0.013053791597485542, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.05} +{'loss': 0.0135, 'grad_norm': 0.07075101137161255, 'learning_rate': 7.090042748428737e-06, 'flow_matching_loss': 0.013516444247215986, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.05} +{'loss': 0.0128, 'grad_norm': 0.07539436221122742, 'learning_rate': 7.075998680956797e-06, 'flow_matching_loss': 0.01284086937084794, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.05} +{'loss': 0.0123, 'grad_norm': 0.06344272196292877, 'learning_rate': 7.061964241893639e-06, 'flow_matching_loss': 0.01232607001438737, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.05} +{'loss': 0.0149, 'grad_norm': 0.07526150345802307, 'learning_rate': 7.047939448292442e-06, 'flow_matching_loss': 0.014897357299923896, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.06} +{'loss': 0.0128, 'grad_norm': 0.06881296634674072, 'learning_rate': 7.033924317194644e-06, 'flow_matching_loss': 0.012764256540685892, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.06} +{'loss': 0.0134, 'grad_norm': 0.06596582382917404, 'learning_rate': 7.0199188656299524e-06, 'flow_matching_loss': 0.013440481945872306, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.06} +{'loss': 0.0127, 'grad_norm': 0.07813560962677002, 'learning_rate': 7.005923110616313e-06, 'flow_matching_loss': 0.012699814140796661, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.07} +{'loss': 0.0138, 'grad_norm': 0.08119695633649826, 'learning_rate': 6.9919370691598855e-06, 'flow_matching_loss': 0.013788906764239073, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.07} +{'loss': 0.0129, 'grad_norm': 0.062298137694597244, 'learning_rate': 6.977960758255024e-06, 'flow_matching_loss': 0.012904537934809923, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.07} +{'loss': 0.0133, 'grad_norm': 0.07072030752897263, 'learning_rate': 6.963994194884276e-06, 'flow_matching_loss': 0.013342053163796663, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.08} +{'loss': 0.0144, 'grad_norm': 0.08403238654136658, 'learning_rate': 6.950037396018328e-06, 'flow_matching_loss': 0.014421308133751153, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.08} +{'loss': 0.0141, 'grad_norm': 0.05878273397684097, 'learning_rate': 6.9360903786160056e-06, 'flow_matching_loss': 0.01406254032626748, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.08} +{'loss': 0.0128, 'grad_norm': 0.05908052623271942, 'learning_rate': 6.9221531596242485e-06, 'flow_matching_loss': 0.01279848925769329, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.08} +{'loss': 0.014, 'grad_norm': 0.07026775181293488, 'learning_rate': 6.908225755978089e-06, 'flow_matching_loss': 0.013967563072219491, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.09} +{'loss': 0.0134, 'grad_norm': 0.07203994691371918, 'learning_rate': 6.8943081846006466e-06, 'flow_matching_loss': 0.01340379510074854, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.09} +{'loss': 0.0142, 'grad_norm': 0.09151970595121384, 'learning_rate': 6.88040046240307e-06, 'flow_matching_loss': 0.014152648579329252, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.09} +{'loss': 0.0134, 'grad_norm': 0.07707943767309189, 'learning_rate': 6.866502606284566e-06, 'flow_matching_loss': 0.013432913646101951, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.1} +{'loss': 0.0135, 'grad_norm': 0.06905467808246613, 'learning_rate': 6.852614633132335e-06, 'flow_matching_loss': 0.013512122631072997, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.1} +{'loss': 0.0141, 'grad_norm': 0.075456902384758, 'learning_rate': 6.8387365598215665e-06, 'flow_matching_loss': 0.014145964104682208, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.1} +{'loss': 0.0144, 'grad_norm': 0.07432951778173447, 'learning_rate': 6.82486840321544e-06, 'flow_matching_loss': 0.014378399308770894, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.1} +{'loss': 0.015, 'grad_norm': 0.07796572893857956, 'learning_rate': 6.811010180165069e-06, 'flow_matching_loss': 0.015034446585923433, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.11} +{'loss': 0.0125, 'grad_norm': 0.08578699082136154, 'learning_rate': 6.797161907509499e-06, 'flow_matching_loss': 0.012520063109695912, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.11} +{'loss': 0.0126, 'grad_norm': 0.07925533503293991, 'learning_rate': 6.783323602075688e-06, 'flow_matching_loss': 0.012584083434194326, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.11} +{'loss': 0.0123, 'grad_norm': 0.08869442343711853, 'learning_rate': 6.7694952806784784e-06, 'flow_matching_loss': 0.012253145501017571, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.12} +{'loss': 0.0139, 'grad_norm': 0.0829630047082901, 'learning_rate': 6.7556769601205944e-06, 'flow_matching_loss': 0.013936540205031633, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.12} +{'loss': 0.0156, 'grad_norm': 0.06480823457241058, 'learning_rate': 6.741868657192594e-06, 'flow_matching_loss': 0.01557383881881833, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.12} +{'loss': 0.014, 'grad_norm': 0.08003394305706024, 'learning_rate': 6.728070388672864e-06, 'flow_matching_loss': 0.013954813312739133, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.12} +{'loss': 0.0131, 'grad_norm': 0.06920632719993591, 'learning_rate': 6.714282171327604e-06, 'flow_matching_loss': 0.013074188027530909, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.13} +{'loss': 0.0132, 'grad_norm': 0.07120053470134735, 'learning_rate': 6.700504021910791e-06, 'flow_matching_loss': 0.013159275893121958, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.13} +{'loss': 0.0128, 'grad_norm': 0.04894782975316048, 'learning_rate': 6.68673595716419e-06, 'flow_matching_loss': 0.012831538729369641, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.13} +{'loss': 0.0141, 'grad_norm': 0.06352882832288742, 'learning_rate': 6.6729779938172856e-06, 'flow_matching_loss': 0.014128577802330256, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.14} +{'loss': 0.0131, 'grad_norm': 0.089161217212677, 'learning_rate': 6.659230148587311e-06, 'flow_matching_loss': 0.013059527147561312, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.14} +{'loss': 0.0128, 'grad_norm': 0.08149219304323196, 'learning_rate': 6.645492438179188e-06, 'flow_matching_loss': 0.012751314416527748, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.14} 70%|███████ | 21110/30000 [13:08:50<8:20:27, 3.38s/it] 70%|███████ | 21111/30000 [13:08:51<6:42:38, 2.72s/it] 70%|███████ | 21112/30000 [13:08:52<5:34:58, 2.26s/it] 70%|███████ | 21113/30000 [13:08:53<4:46:35, 1.93s/it] 70%|███████ | 21114/30000 [13:08:54<4:12:17, 1.70s/it] 70%|███████ | 21115/30000 [13:08:55<3:49:17, 1.55s/it] 70%|███████ | 21116/30000 [13:08:57<3:33:18, 1.44s/it] 70%|███████ | 21117/30000 [13:08:58<3:21:53, 1.36s/it] 70%|███████ | 21118/30000 [13:09:06<8:26:59, 3.42s/it] 70%|███████ | 21119/30000 [13:09:07<6:45:52, 2.74s/it] 70%|███████ | 21120/30000 [13:09:08<5:37:03, 2.28s/it] 70%|███████ | 21120/30000 [13:09:08<5:37:03, 2.28s/it] 70%|███████ | 21121/30000 [13:09:10<4:47:23, 1.94s/it] 70%|███████ | 21122/30000 [13:09:11<4:12:23, 1.71s/it] 70%|███████ | 21123/30000 [13:09:12<3:49:38, 1.55s/it] 70%|███████ | 21124/30000 [13:09:13<3:33:14, 1.44s/it] 70%|███████ | 21125/30000 [13:09:14<3:22:23, 1.37s/it] 70%|███████ | 21126/30000 [13:09:21<7:33:06, 3.06s/it] 70%|███████ | 21127/30000 [13:09:22<6:09:18, 2.50s/it] 70%|███████ | 21128/30000 [13:09:24<5:10:46, 2.10s/it] 70%|███████ | 21129/30000 [13:09:25<4:29:00, 1.82s/it] 70%|███████ | 21130/30000 [13:09:26<4:00:22, 1.63s/it] 70%|███████ | 21130/30000 [13:09:26<4:00:22, 1.63s/it] 70%|███████ | 21131/30000 [13:09:27<3:40:51, 1.49s/it] 70%|███████ | 21132/30000 [13:09:28<3:27:23, 1.40s/it] 70%|███████ | 21133/30000 [13:09:30<3:17:50, 1.34s/it] 70%|███████ | 21134/30000 [13:09:36<6:51:44, 2.79s/it] 70%|███████ | 21135/30000 [13:09:37<5:40:08, 2.30s/it] 70%|███████ | 21136/30000 [13:09:38<4:50:54, 1.97s/it] 70%|███████ | 21137/30000 [13:09:39<4:16:04, 1.73s/it] 70%|███████ | 21138/30000 [13:09:40<3:51:04, 1.56s/it] 70%|███████ | 21139/30000 [13:09:42<3:34:22, 1.45s/it] 70%|███████ | 21140/30000 [13:09:43<3:21:27, 1.36s/it] 70%|███████ | 21140/30000 [13:09:43<3:21:27, 1.36s/it] 70%|███████ | 21141/30000 [13:09:44<3:13:21, 1.31s/it] 70%|███████ | 21142/30000 [13:09:51<7:25:36, 3.02s/it] 70%|███████ | 21143/30000 [13:09:52<6:03:34, 2.46s/it] 70%|███████ | 21144/30000 [13:09:53<5:06:58, 2.08s/it] 70%|███████ | 21145/30000 [13:09:54<4:25:42, 1.80s/it] 70%|███████ | 21146/30000 [13:09:56<3:57:13, 1.61s/it] 70%|███████ | 21147/30000 [13:09:57<3:38:24, 1.48s/it] 70%|███████ | 21148/30000 [13:09:58<3:25:27, 1.39s/it] 70%|███████ | 21149/30000 [13:09:59<3:16:28, 1.33s/it] 70%|███████ | 21150/30000 [13:10:07<7:43:03, 3.14s/it] 70%|███████ | 21150/30000 [13:10:07<7:43:03, 3.14s/it] 71%|███████ | 21151/30000 [13:10:08<6:16:34, 2.55s/it] 71%|███████ | 21152/30000 [13:10:09<5:16:36, 2.15s/it] 71%|███████ | 21153/30000 [13:10:10<4:32:52, 1.85s/it] 71%|███████ | 21154/30000 [13:10:11<4:02:29, 1.64s/it] 71%|███████ | 21155/30000 [13:10:12<3:42:07, 1.51s/it] 71%|███████ | 21156/30000 [13:10:14<3:27:50, 1.41s/it] 71%|███████ | 21157/30000 [13:10:15<3:18:02, 1.34s/it] 71%|███████ | 21158/30000 [13:10:23<8:17:55, 3.38s/it] 71%|███████ | 21159/30000 [13:10:24<6:40:28, 2.72s/it] 71%|███████ | 21160/30000 [13:10:25<5:33:01, 2.26s/it] 71%|███████ | 21160/30000 [13:10:25<5:33:01, 2.26s/it] 71%|███████ | 21161/30000 [13:10:26<4:44:56, 1.93s/it] 71%|███████ | 21162/30000 [13:10:28<4:11:12, 1.71s/it] 71%|███████ | 21163/30000 [13:10:29<3:47:51, 1.55s/it] 71%|███████ | 21164/30000 [13:10:30<3:31:46, 1.44s/it] 71%|███████ | 21165/30000 [13:10:31<3:20:51, 1.36s/it] 71%|███████ | 21166/30000 [13:10:39<8:26:04, 3.44s/it] 71%|███████ | 21167/30000 [13:10:41<6:45:39, 2.76s/it] 71%|███████ | 21168/30000 [13:10:42<5:36:42, 2.29s/it] 71%|███████ | 21169/30000 [13:10:43<4:47:28, 1.95s/it] 71%|███████ | 21170/30000 [13:10:44<4:12:59, 1.72s/it] 71%|███████ | 21170/30000 [13:10:44<4:12:59, 1.72s/it] 71%|███████ | 21171/30000 [13:10:45<3:49:36, 1.56s/it] 71%|███████ | 21172/30000 [13:10:47<3:33:12, 1.45s/it] 71%|███████ | 21173/30000 [13:10:48<3:21:44, 1.37s/it] 71%|███████ | 21174/30000 [13:10:56<8:20:43, 3.40s/it] 71%|███████ | 21175/30000 [13:10:57<6:42:16, 2.73s/it] 71%|███████ | 21176/30000 [13:10:58<5:34:11, 2.27s/it] 71%|███████ | 21177/30000 [13:10:59<4:45:17, 1.94s/it] 71%|███████ | 21178/30000 [13:11:01<4:10:42, 1.71s/it] 71%|███████ | 21179/30000 [13:11:02<3:48:23, 1.55s/it] 71%|███████ | 21180/30000 [13:11:03<3:32:51, 1.45s/it] 71%|███████ | 21180/30000 [13:11:03<3:32:51, 1.45s/it] 71%|███████ | 21181/30000 [13:11:04<3:21:06, 1.37s/it] 71%|███████ | 21182/30000 [13:11:12<8:05:10, 3.30s/it] 71%|███████ | 21183/30000 [13:11:13<6:30:41, 2.66s/it] 71%|███████ | 21184/30000 [13:11:14<5:25:30, 2.22s/it] 71%|███████ | 21185/30000 [13:11:15<4:39:14, 1.90s/it] 71%|███████ | 21186/30000 [13:11:17<4:07:29, 1.68s/it] 71%|███████ | 21187/30000 [13:11:18<3:46:44, 1.54s/it] 71%|███████ | 21188/30000 [13:11:19<3:31:17, 1.44s/it] 71%|███████ | 21189/30000 [13:11:20<3:20:16, 1.36s/it] 71%|███████ | 21190/30000 [13:11:28<7:53:01, 3.22s/it] 71%|███████ | 21190/30000 [13:11:28<7:53:01, 3.22s/it] 71%|███████ | 21191/30000 [13:11:29<6:22:42, 2.61s/it] 71%|███████ | 21192/30000 [13:11:30<5:20:23, 2.18s/it] 71%|███████ | 21193/30000 [13:11:31<4:35:14, 1.88s/it] 71%|███████ | 21194/30000 [13:11:32<4:03:52, 1.66s/it] 71%|███████ | 21195/30000 [13:11:34<3:43:14, 1.52s/it] 71%|███████ | 21196/30000 [13:11:35<3:28:44, 1.42s/it] 71%|███████ | 21197/30000 [13:11:36<3:18:04, 1.35s/it] 71%|███████ | 21198/30000 [13:11:44<7:59:53, 3.27s/it] 71%|███████ | 21199/30000 [13:11:45<6:27:43, 2.64s/it] 71%|███████ | 21200/30000 [13:11:46<5:23:45, 2.21s/it] 71%|███████ | 21200/30000 [13:11:46<5:23:45, 2.21s/it] 71%|███████ | 21201/30000 [13:11:47<4:38:07, 1.90s/it] 71%|███████ | 21202/30000 [13:11:49<4:06:01, 1.68s/it] 71%|███████ | 21203/30000 [13:11:50<3:44:03, 1.53s/it] 71%|███████ | 21204/30000 [13:11:51<3:29:09, 1.43s/it] 71%|███████ | 21205/30000 [13:11:52<3:20:33, 1.37s/it] 71%|███████ | 21206/30000 [13:12:01<8:31:33, 3.49s/it] 71%|███████ | 21207/30000 [13:12:02<6:49:09, 2.79s/it] 71%|███████ | 21208/30000 [13:12:03<5:37:26, 2.30s/it] 71%|███████ | 21209/30000 [13:12:04<4:48:04, 1.97s/it] 71%|███████ | 21210/30000 [13:12:05<4:13:07, 1.73s/it] 71%|███████ | 21210/30000 [13:12:05<4:13:07, 1.73s/it] 71%|███████ | 21211/30000 [13:12:06<3:50:02, 1.57s/it] 71%|███████ | 21212/30000 [13:12:08<3:33:36, 1.46s/it] 71%|███████ | 21213/30000 [13:12:09<3:21:45, 1.38s/it] 71%|███████ | 21214/30000 [13:12:16<7:57:19, 3.26s/it] 71%|███████ | 21215/30000 [13:12:18<6:25:39, 2.63s/it] 71%|███████ | 21216/30000 [13:12:19<5:22:16, 2.20s/it] 71%|███████ | 21217/30000 [13:12:20<4:36:36, 1.89s/it] 71%|███████ | 21218/30000 [13:12:21<4:04:29, 1.67s/it] 71%|███████ | 21219/30000 [13:12:22<3:43:11, 1.53s/it] 71%|███████ | 21220/30000 [13:12:24<3:27:55, 1.42s/it] 71%|███████ | 21220/30000 [13:12:24<3:27:55, 1.42s/it] 71%|███████ | 21221/30000 [13:12:25<3:17:16, 1.35s/it] 71%|███████ | 21222/30000 [13:12:32<7:40:59, 3.15s/it] 71%|███████ | 21223/30000 [13:12:33<6:14:17, 2.56s/it] 71%|███████ | 21224/30000 [13:12:34<5:14:10, 2.15s/it] 71%|███████ | 21225/30000 [13:12:36<4:31:39, 1.86s/it] 71%|███████ | 21226/30000 [13:12:37<4:01:37, 1.65s/it] 71%|███████ | 21227/30000 [13:12:38<3:40:26, 1.51s/it] 71%|███████ | 21228/30000 [13:12:39<3:26:14, 1.41s/it] 71%|███████ | 21229/30000 [13:12:40<3:16:11, 1.34s/it] 71%|███████ | 21230/30000 [13:12:48<7:56:29, 3.26s/it] 71%|███████ | 21230/30000 [13:12:48<7:56:29, 3.26s/it] 71%|███████ | 21231/30000 [13:12:49<6:24:33, 2.63s/it] 71%|███████ | 21232/30000 [13:12:50<5:21:12, 2.20s/it] 71%|███████ | 21233/30000 [13:12:52<4:35:54, 1.89s/it] 71%|███████ | 21234/30000 [13:12:53<4:04:38, 1.67s/it] 71%|███████ | 21235/30000 [13:12:54<3:43:22, 1.53s/it] 71%|███████ | 21236/30000 [13:12:55<3:28:33, 1.43s/it] 71%|███████ | 21237/30000 [13:12:56<3:18:14, 1.36s/it] 71%|███████ | 21238/30000 [13:13:03<7:23:26, 3.04s/it] 71%|███████ | 21239/30000 [13:13:04<6:01:19, 2.47s/it] 71%|███████ | 21240/30000 [13:13:06<5:04:53, 2.09s/it] 71%|███████ | 21240/30000 [13:13:06<5:04:53, 2.09s/it] 71%|███████ | 21241/30000 [13:13:07<4:25:06, 1.82s/it] 71%|███████ | 21242/30000 [13:13:08<3:56:22, 1.62s/it] 71%|███████ | 21243/30000 [13:13:09<3:38:08, 1.49s/it] 71%|███████ | 21244/30000 [13:13:10<3:25:31, 1.41s/it] 71%|███████ | 21245/30000 [13:13:12<3:15:10, 1.34s/it] 71%|███████ | 21246/30000 [13:13:18<7:14:42, 2.98s/it] 71%|███████ | 21247/30000 [13:13:20<5:55:07, 2.43s/it] 71%|███████ | 21248/30000 [13:13:21<5:00:17, 2.06s/it] 71%|███████ | 21249/30000 [13:13:22<4:21:29, 1.79s/it] 71%|███████ | 21250/30000 [13:13:23<3:53:58, 1.60s/it] 71%|███████ | 21250/30000 [13:13:23<3:53:58, 1.60s/it] 71%|███████ | 21251/30000 [13:13:24<3:36:23, 1.48s/it] 71%|███████ | 21252/30000 [13:13:25<3:25:14, 1.41s/it] 71%|███████ | 21253/30000 [13:13:27<3:17:48, 1.36s/it] 71%|███████ | 21254/30000 [13:13:34<7:52:13, 3.24s/it] 71%|███████ | 21255/30000 [13:13:36<6:21:33, 2.62s/it] 71%|███████ | 21256/30000 [13:13:37<5:19:03, 2.19s/it] 71%|███████ | 21257/30000 [13:13:38<4:33:50, 1.88s/it] 71%|███████ | 21258/30000 [13:13:39<4:02:53, 1.67s/it] 71%|███████ | 21259/30000 [13:13:40<3:42:18, 1.53s/it] 71%|███████ | 21260/30000 [13:13:41<3:28:03, 1.43s/it] 71%|███████ | 21260/30000 [13:13:41<3:28:03, 1.43s/it] 71%|███████ | 21261/30000 [13:13:43<3:17:26, 1.36s/it] 71%|███████ | 21262/30000 [13:13:50<8:02:21, 3.31s/it] 71%|███████ | 21263/30000 [13:13:52<6:29:20, 2.67s/it] 71%|███████ | 21264/30000 [13:13:53<5:24:27, 2.23s/it] 71%|███████ | 21265/30000 [13:13:54<4:38:17, 1.91s/it] 71%|███████ | 21266/30000 [13:13:55<4:05:25, 1.69s/it] 71%|███████ | 21267/30000 [13:13:56<3:43:46, 1.54s/it] 71%|███████ | 21268/30000 [13:13:58<3:28:46, 1.43s/it] 71%|███████ | 21269/30000 [13:13:59<3:18:14, 1.36s/it] 71%|███████ | 21270/30000 [13:14:07<8:02:54, 3.32s/it] 71%|███████ | 21270/30000 [13:14:07<8:02:54, 3.32s/it] 71%|███████ | 21271/30000 [13:14:08<6:30:41, 2.69s/it] 71%|███████ | 21272/30000 [13:14:09<5:26:26, 2.24s/it] 71%|███████ | 21273/30000 [13:14:10<4:39:27, 1.92s/it] 71%|███████ | 21274/30000 [13:14:11<4:06:23, 1.69s/it] 71%|███████ | 21275/30000 [13:14:13<3:44:23, 1.54s/it] 71%|███████ | 21276/30000 [13:14:14<3:28:38, 1.43s/it] 71%|███████ | 21277/30000 [13:14:15<3:17:43, 1.36s/it] 71%|███████ | 21278/30000 [13:14:22<7:04:33, 2.92s/it] 71%|███████ | 21279/30000 [13:14:23<5:49:11, 2.40s/it] 71%|███████ | 21280/30000 [13:14:24<4:56:25, 2.04s/it] 71%|███████ | 21280/30000 [13:14:24<4:56:25, 2.04s/it] 71%|███████ | 21281/30000 [13:14:25<4:19:45, 1.79s/it] 71%|███████ | 21282/30000 [13:14:26<3:52:26, 1.60s/it] 71%|███████ | 21283/30000 [13:14:27<3:34:30, 1.48s/it] 71%|███████ | 21284/30000 [13:14:29<3:21:31, 1.39s/it] 71%|███████ | 21285/30000 [13:14:30<3:12:39, 1.33s/it] 71%|███████ | 21286/30000 [13:14:38<7:50:53, 3.24s/it] 71%|███████ | 21287/30000 [13:14:39<6:20:23, 2.62s/it] 71%|███████ | 21288/30000 [13:14:40<5:18:24, 2.19s/it] 71%|███████ | 21289/30000 [13:14:41<4:33:53, 1.89s/it] 71%|███████ | 21290/30000 [13:14:42<4:02:17, 1.67s/it] 71%|███████ | 21290/30000 [13:14:42<4:02:17, 1.67s/it] 71%|███████ | 21291/30000 [13:14:43<3:41:33, 1.53s/it] 71%|███████ | 21292/30000 [13:14:45<3:26:40, 1.42s/it] 71%|███████ | 21293/30000 [13:14:46<3:16:49, 1.36s/it] 71%|███████ | 21294/30000 [13:14:53<7:48:12, 3.23s/it] 71%|███████ | 21295/30000 [13:14:55<6:18:35, 2.61s/it] 71%|███████ | 21296/30000 [13:14:56<5:17:04, 2.19s/it] 71%|███████ | 21297/30000 [13:14:57<4:32:52, 1.88s/it] 71%|███████ | 21298/30000 [13:14:58<4:01:36, 1.67s/it] 71%|███████ | 21299/30000 [13:14:59<3:40:14, 1.52s/it] 71%|███████ | 21300/30000 [13:15:00<3:25:18, 1.42s/it] 71%|███████ | 21300/30000 [13:15:00<3:25:18, 1.42s/it] 71%|███████ | 21301/30000 [13:15:02<3:16:11, 1.35s/it] 71%|███████ | 21302/30000 [13:15:09<7:53:00, 3.26s/it] 71%|███████ | 21303/30000 [13:15:11<6:21:26, 2.63s/it] 71%|███████ | 21304/30000 [13:15:12<5:19:02, 2.20s/it] 71%|███████ | 21305/30000 [13:15:13<4:34:37, 1.90s/it] 71%|███████ | 21306/30000 [13:15:14<4:04:41, 1.69s/it] 71%|███████ | 21307/30000 [13:15:15<3:43:12, 1.54s/it] 71%|███████ | 21308/30000 [13:15:17<3:27:39, 1.43s/it] 71%|███████ | 21309/30000 [13:15:18<3:18:44, 1.37s/it] 71%|███████ | 21310/30000 [13:15:26<8:04:32, 3.35s/it] 71%|███████ | 21310/30000 [13:15:26<8:04:32, 3.35s/it] 71%|███████ | 21311/30000 [13:15:27<6:30:10, 2.69s/it] 71%|███████ | 21312/30000 [13:15:28<5:24:43, 2.24s/it] 71%|███████ | 21313/30000 [13:15:29<4:38:25, 1.92s/it] 71%|███████ | 21314/30000 [13:15:30<4:05:57, 1.70s/it] 71%|███████ | 21315/30000 [13:15:32<3:44:06, 1.55s/it] 71%|███████ | 21316/30000 [13:15:33<3:28:59, 1.44s/it] 71%|███████ | 21317/30000 [13:15:34<3:17:23, 1.36s/it] 71%|███████ | 21318/30000 [13:15:42<8:20:22, 3.46s/it] 71%|███████ | 21319/30000 [13:15:44<6:41:18, 2.77s/it] 71%|███████ | 21320/30000 [13:15:45<5:34:07, 2.31s/it] 71%|███████ | 21320/30000 [13:15:45<5:34:07, 2.31s/it] 71%|███████ | 21321/30000 [13:15:46<4:44:32, 1.97s/it] 71%|███████ | 21322/30000 [13:15:47<4:09:55, 1.73s/it] 71%|███████ | 21323/30000 [13:15:48<3:46:53, 1.57s/it] 71%|███████ | 21324/30000 [13:15:49<3:30:24, 1.46s/it] 71%|███████ | 21325/30000 [13:15:51<3:18:46, 1.37s/it] 71%|███████ | 21326/30000 [13:15:59<8:11:39, 3.40s/it] 71%|███████ | 21327/30000 [13:16:00<6:34:42, 2.73s/it] 71%|███████ | 21328/30000 [13:16:01<5:28:21, 2.27s/it] 71%|███████ | 21329/30000 [13:16:02<4:40:13, 1.94s/it] 71%|███████ | 21330/30000 [13:16:03<4:06:19, 1.70s/it] 71%|███████ | 21330/30000 [13:16:03<4:06:19, 1.70s/it] 71%|███████ | 21331/30000 [13:16:05<3:43:54, 1.55s/it] 71%|███████ | 21332/30000 [13:16:06<3:28:10, 1.44s/it] 71%|███████ | 21333/30000 [13:16:07<3:17:30, 1.37s/it] 71%|███████ | 21334/30000 [13:16:15<7:43:10, 3.21s/it] 71%|███████ | 21335/30000 [13:16:16<6:14:52, 2.60s/it] 71%|███████ | 21336/30000 [13:16:17<5:13:50, 2.17s/it] 71%|███████ | 21337/30000 [13:16:18<4:30:36, 1.87s/it] 71%|███████ | 21338/30000 [13:16:19<3:59:49, 1.66s/it] 71%|███████ | 21339/30000 [13:16:20<3:38:38, 1.51s/it] 71%|███████ | 21340/30000 [13:16:22<3:24:16, 1.42s/it] 71%|███████ | 21340/30000 [13:16:22<3:24:16, 1.42s/it] 71%|███████ | 21341/30000 [13:16:23<3:14:02, 1.34s/it] 71%|███████ | 21342/30000 [13:16:31<7:52:24, 3.27s/it] 71%|███████ | 21343/30000 [13:16:32<6:21:05, 2.64s/it] 71%|███████ | 21344/30000 [13:16:33<5:17:57, 2.20s/it] 71%|███████ | 21345/30000 [13:16:34<4:32:45, 1.89s/it] 71%|███████ | 21346/30000 [13:16:35<4:01:08, 1.67s/it] 71%|███████ | 21347/30000 [13:16:36<3:40:25, 1.53s/it] 71%|███████ | 21348/30000 [13:16:38<3:25:48, 1.43s/it] 71%|███████ | 21349/30000 [13:16:39<3:14:52, 1.35s/it] 71%|███████ | 21350/30000 [13:16:46<7:32:29, 3.14s/it] 71%|███████ | 21350/30000 [13:16:46<7:32:29, 3.14s/it] 71%|███████ | 21351/30000 [13:16:47<6:07:29, 2.55s/it] 71%|███████ | 21352/30000 [13:16:48<5:09:45, 2.15s/it] 71%|███████ | 21353/30000 [13:16:50<4:28:09, 1.86s/it] 71%|███████ | 21354/30000 [13:16:51<3:57:36, 1.65s/it] 71%|███████ | 21355/30000 [13:16:52<3:37:09, 1.51s/it] 71%|███████ | 21356/30000 [13:16:53<3:22:48, 1.41s/it] 71%|███████ | 21357/30000 [13:16:54<3:13:06, 1.34s/it] 71%|███████ | 21358/30000 [13:17:03<8:13:18, 3.42s/it] 71%|███████ | 21359/30000 [13:17:04<6:35:28, 2.75s/it] 71%|███████ | 21360/30000 [13:17:05<5:28:19, 2.28s/it] 71%|███████ | 21360/30000 [13:17:05<5:28:19, 2.28s/it] 71%|███████ | 21361/30000 [13:17:06<4:40:01, 1.94s/it] 71%|███████ | 21362/30000 [13:17:07<4:06:38, 1.71s/it] 71%|███████ | 21363/30000 [13:17:08<3:42:23, 1.54s/it] 71%|███████ | 21364/30000 [13:17:10<3:26:46, 1.44s/it] 71%|███████ | 21365/30000 [13:17:11<3:15:52, 1.36s/it] 71%|███████ | 21366/30000 [13:17:18<7:36:10, 3.17s/it] 71%|███████ | 21367/30000 [13:17:19<6:09:36, 2.57s/it] 71%|███████ | 21368/30000 [13:17:21<5:09:52, 2.15s/it] 71%|███████ | 21369/30000 [13:17:22<4:28:40, 1.87s/it] 71%|███████ | 21370/30000 [13:17:23<3:59:27, 1.66s/it] 71%|███████ | 21370/30000 [13:17:23<3:59:27, 1.66s/it] 71%|███████ | 21371/30000 [13:17:24<3:39:00, 1.52s/it] 71%|███████ | 21372/30000 [13:17:25<3:24:07, 1.42s/it] 71%|███████ | 21373/30000 [13:17:27<3:14:25, 1.35s/it] 71%|███████ | 21374/30000 [13:17:34<7:32:17, 3.15s/it] 71%|███████▏ | 21375/30000 [13:17:35<6:07:03, 2.55s/it] 71%|███████▏ | 21376/30000 [13:17:36<5:07:26, 2.14s/it] 71%|███████▏ | 21377/30000 [13:17:37<4:26:10, 1.85s/it] 71%|███████▏ | 21378/30000 [13:17:39<3:57:18, 1.65s/it] 71%|███████▏ | 21379/30000 [13:17:40<3:37:34, 1.51s/it] 71%|███████▏ | 21380/30000 [13:17:41<3:23:33, 1.42s/it] 71%|███████▏ | 21380/30000 [13:17:41<3:23:33, 1.42s/it] 71%|███████▏ | 21381/30000 [13:17:42<3:14:02, 1.35s/it] 71%|███████▏ | 21382/30000 [13:17:50<7:52:17, 3.29s/it] 71%|███████▏ | 21383/30000 [13:17:51<6:21:00, 2.65s/it] 71%|███████▏ | 21384/30000 [13:17:52<5:17:51, 2.21s/it] 71%|███████▏ | 21385/30000 [13:17:54<4:32:55, 1.90s/it] 71%|███████▏ | 21386/30000 [13:17:55<4:02:11, 1.69s/it] 71%|███████▏ | 21387/30000 [13:17:56<3:40:04, 1.53s/it] 71%|███████▏ | 21388/30000 [13:17:57<3:25:00, 1.43s/it] 71%|███████▏ | 21389/30000 [13:17:58<3:14:23, 1.35s/it] 71%|███████▏ | 21390/30000 [13:18:05<7:25:20, 3.10s/it] 71%|███████▏ | 21390/30000 [13:18:05<7:25:20, 3.10s/it] 71%|███████▏ | 21391/30000 [13:18:07<6:01:43, 2.52s/it] 71%|███████▏ | 21392/30000 [13:18:08<5:03:34, 2.12s/it] 71%|███████▏ | 21393/30000 [13:18:09<4:22:19, 1.83s/it] 71%|███████▏ | 21394/30000 [13:18:10<3:54:04, 1.63s/it] 71%|███████▏ | 21395/30000 [13:18:11<3:34:34, 1.50s/it] 71%|███████▏ | 21396/30000 [13:18:12<3:21:15, 1.40s/it] 71%|███████▏ | 21397/30000 [13:18:14<3:11:15, 1.33s/it] 71%|███████▏ | 21398/30000 [13:18:21<7:44:54, 3.24s/it] 71%|███████▏ | 21399/30000 [13:18:22<6:15:20, 2.62s/it] 71%|███████▏ | 21400/30000 [13:18:24<5:13:33, 2.19s/it] 71%|███████▏ | 21400/30000 [13:18:24<5:13:33, 2.19s/it] 71%|███████▏ | 21401/30000 [13:18:25<4:30:36, 1.89s/it] 71%|███████▏ | 21402/30000 [13:18:26<3:59:40, 1.67s/it] 71%|███████▏ | 21403/30000 [13:18:27<3:38:58, 1.53s/it] 71%|███████▏ | 21404/30000 [13:18:28<3:23:51, 1.42s/it] 71%|███████▏ | 21405/30000 [13:18:30<3:15:06, 1.36s/it] 71%|███████▏ | 21406/30000 [13:18:37<7:50:09, 3.28s/it] 71%|███████▏ | 21407/30000 [13:18:39<6:19:47, 2.65s/it] 71%|███████▏ | 21408/30000 [13:18:40<5:16:43, 2.21s/it] 71%|███████▏ | 21409/30000 [13:18:41<4:31:29, 1.90s/it] 71%|███████▏ | 21410/30000 [13:18:42<4:01:05, 1.68s/it] 71%|███████▏ | 21410/30000 [13:18:42<4:01:05, 1.68s/it] 71%|███████▏ | 21411/30000 [13:18:43<3:39:42, 1.53s/it] 71%|███████▏ | 21412/30000 [13:18:44<3:25:26, 1.44s/it] 71%|███████▏ | 21413/30000 [13:18:46<3:15:13, 1.36s/it] 71%|███████▏ | 21414/30000 [13:18:53<7:12:58, 3.03s/it] 71%|███████▏ | 21415/30000 [13:18:54<5:51:45, 2.46s/it] 71%|███████▏ | 21416/30000 [13:18:55<4:56:46, 2.07s/it] 71%|███████▏ | 21417/30000 [13:18:56<4:17:35, 1.80s/it] 71%|███████▏ | 21418/30000 [13:18:57<3:50:31, 1.61s/it] 71%|███████▏ | 21419/30000 [13:18:58<3:31:30, 1.48s/it] 71%|███████▏ | 21420/30000 [13:19:00<3:19:05, 1.39s/it] 71%|███████▏ | 21420/30000 [13:19:00<3:19:05, 1.39s/it] 71%|███████▏ | 21421/30000 [13:19:01<3:10:33, 1.33s/it] 71%|███████▏ | 21422/30000 [13:19:08<7:10:39, 3.01s/it] 71%|███████▏ | 21423/30000 [13:19:09<5:51:25, 2.46s/it] 71%|███████▏ | 21424/30000 [13:19:10<4:56:36, 2.08s/it] 71%|███████▏ | 21425/30000 [13:19:11<4:17:19, 1.80s/it] 71%|███████▏ | 21426/30000 [13:19:12<3:49:48, 1.61s/it] 71%|███████▏ | 21427/30000 [13:19:14<3:31:35, 1.48s/it] 71%|███████▏ | 21428/30000 [13:19:15<3:19:38, 1.40s/it] 71%|███████▏ | 21429/30000 [13:19:16<3:10:39, 1.33s/it] 71%|███████▏ | 21430/30000 [13:19:23<7:28:02, 3.14s/it] 71%|███████▏ | 21430/30000 [13:19:23<7:28:02, 3.14s/it] 71%|███████▏ | 21431/30000 [13:19:24<6:03:41, 2.55s/it] 71%|███████▏ | 21432/30000 [13:19:26<5:04:45, 2.13s/it] 71%|███████▏ | 21433/30000 [13:19:27<4:23:23, 1.84s/it] 71%|███████▏ | 21434/30000 [13:19:28<3:54:38, 1.64s/it] 71%|███████▏ | 21435/30000 [13:19:29<3:35:01, 1.51s/it] 71%|███████▏ | 21436/30000 [13:19:30<3:21:25, 1.41s/it] 71%|███████▏ | 21437/30000 [13:19:32<3:12:06, 1.35s/it] 71%|███████▏ | 21438/30000 [13:19:40<7:57:32, 3.35s/it] 71%|███████▏ | 21439/30000 [13:19:41<6:24:10, 2.69s/it] 71%|███████▏ | 21440/30000 [13:19:42<5:19:57, 2.24s/it] 71%|███████▏ | 21440/30000 [13:19:42<5:19:57, 2.24s/it] 71%|███████▏ | 21441/30000 [13:19:43<4:34:09, 1.92s/it] 71%|███████▏ | 21442/30000 [13:19:44<4:01:44, 1.69s/it] 71%|███████▏ | 21443/30000 [13:19:45<3:40:07, 1.54s/it] 71%|███████▏ | 21444/30000 [13:19:47<3:24:34, 1.43s/it] 71%|███████▏ | 21445/30000 [13:19:48<3:14:15, 1.36s/it] 71%|███████▏ | 21446/30000 [13:19:55<7:39:44, 3.22s/it] 71%|███████▏ | 21447/30000 [13:19:57<6:11:56, 2.61s/it] 71%|███████▏ | 21448/30000 [13:19:58<5:11:22, 2.18s/it] 71%|███████▏ | 21449/30000 [13:19:59<4:27:48, 1.88s/it] 72%|███████▏ | 21450/30000 [13:20:00<3:57:17, 1.67s/it] 72%|███████▏ | 21450/30000 [13:20:00<3:57:17, 1.67s/it] 72%|███████▏ | 21451/30000 [13:20:01<3:38:18, 1.53s/it] 72%|███████▏ | 21452/30000 [13:20:03<3:23:57, 1.43s/it] 72%|███████▏ | 21453/30000 [13:20:04<3:13:52, 1.36s/it] 72%|███████▏ | 21454/30000 [13:20:11<7:39:31, 3.23s/it] 72%|███████▏ | 21455/30000 [13:20:12<6:11:24, 2.61s/it] 72%|███████▏ | 21456/30000 [13:20:14<5:10:48, 2.18s/it] 72%|███████▏ | 21457/30000 [13:20:15<4:27:29, 1.88s/it] 72%|███████▏ | 21458/30000 [13:20:16<3:56:55, 1.66s/it] 72%|███████▏ | 21459/30000 [13:20:17<3:36:05, 1.52s/it] 72%|███████▏ | 21460/30000 [13:20:18<3:22:16, 1.42s/it] 72%|███████▏ | 21460/30000 [13:20:18<3:22:16, 1.42s/it] 72%|███████▏ | 21461/30000 [13:20:20<3:12:12, 1.35s/it] 72%|███████▏ | 21462/30000 [13:20:27<7:17:28, 3.07s/it] 72%|███████▏ | 21463/30000 [13:20:28<5:56:07, 2.50s/it] 72%|███████▏ | 21464/30000 [13:20:29<5:00:19, 2.11s/it] 72%|███████▏ | 21465/30000 [13:20:30<4:20:08, 1.83s/it] 72%|███████▏ | 21466/30000 [13:20:31<3:52:36, 1.64s/it] 72%|███████▏ | 21467/30000 [13:20:33<3:33:25, 1.50s/it] 72%|███████▏ | 21468/30000 [13:20:34<3:20:38, 1.41s/it] 72%|███████▏ | 21469/30000 [13:20:35<3:11:19, 1.35s/it] 72%|███████▏ | 21470/30000 [13:20:42<7:07:12, 3.00s/it] 72%|███████▏ | 21470/30000 [13:20:42<7:07:12, 3.00s/it] 72%|███████▏ | 21471/30000 [13:20:43<5:48:46, 2.45s/it] 72%|███████▏ | 21472/30000 [13:20:44<4:55:06, 2.08s/it] 72%|███████▏ | 21473/30000 [13:20:45<4:16:12, 1.80s/it] 72%|███████▏ | 21474/30000 [13:20:46<3:48:33, 1.61s/it] 72%|███████▏ | 21475/30000 [13:20:48<3:30:35, 1.48s/it] 72%|███████▏ | 21476/30000 [13:20:49<3:18:25, 1.40s/it] 72%|███████▏ | 21477/30000 [13:20:50<3:09:52, 1.34s/it] 72%|███████▏ | 21478/30000 [13:20:57<7:16:30, 3.07s/it] 72%|███████▏ | 21479/30000 [13:20:58<5:55:14, 2.50s/it] 72%|███████▏ | 21480/30000 [13:21:00<4:59:12, 2.11s/it] 72%|███████▏ | 21480/30000 [13:21:00<4:59:12, 2.11s/it] 72%|███████▏ | 21481/30000 [13:21:01<4:19:53, 1.83s/it] 72%|███████▏ | 21482/30000 [13:21:02<3:51:21, 1.63s/it] 72%|███████▏ | 21483/30000 [13:21:03<3:32:09, 1.49s/it] 72%|███████▏ | 21484/30000 [13:21:04<3:19:40, 1.41s/it] 72%|███████▏ | 21485/30000 [13:21:05<3:10:04, 1.34s/it] 72%|███████▏ | 21486/30000 [13:21:13<7:16:12, 3.07s/it] 72%|███████▏ | 21487/30000 [13:21:14<5:55:12, 2.50s/it] 72%|███████▏ | 21488/30000 [13:21:15<4:57:46, 2.10s/it] 72%|███████▏ | 21489/30000 [13:21:16<4:17:55, 1.82s/it] 72%|███████▏ | 21490/30000 [13:21:17<3:49:48, 1.62s/it] 72%|███████▏ | 21490/30000 [13:21:17<3:49:48, 1.62s/it] 72%|███████▏ | 21491/30000 [13:21:18<3:31:24, 1.49s/it] 72%|███████▏ | 21492/30000 [13:21:20<3:18:44, 1.40s/it] 72%|███████▏ | 21493/30000 [13:21:21<3:09:51, 1.34s/it] 72%|███████▏ | 21494/30000 [13:21:28<7:28:25, 3.16s/it] 72%|███████▏ | 21495/30000 [13:21:29<6:03:31, 2.56s/it] 72%|███████▏ | 21496/30000 [13:21:31<5:04:45, 2.15s/it] 72%|███████▏ | 21497/30000 [13:21:32<4:22:40, 1.85s/it] 72%|███████▏ | 21498/30000 [13:21:33<3:52:36, 1.64s/it] 72%|███████▏ | 21499/30000 [13:21:34<3:33:11, 1.50s/it] 72%|███████▏ | 21500/30000 [13:21:35<3:19:59, 1.41s/it] 72%|███████▏ | 21500/30000 [13:21:35<3:19:59, 1.41s/it] 72%|███████▏ | 21501/30000 [13:21:36<3:10:57, 1.35s/it] 72%|███████▏ | 21502/30000 [13:21:45<7:55:55, 3.36s/it] 72%|███████▏ | 21503/30000 [13:21:46<6:23:18, 2.71s/it] 72%|███████▏ | 21504/30000 [13:21:47<5:18:36, 2.25s/it] 72%|███████▏ | 21505/30000 [13:21:48<4:32:17, 1.92s/it] 72%|███████▏ | 21506/30000 [13:21:49<3:59:36, 1.69s/it] 72%|███████▏ | 21507/30000 [13:21:50<3:37:56, 1.54s/it] 72%|███████▏ | 21508/30000 [13:21:52<3:23:08, 1.44s/it] 72%|███████▏ | 21509/30000 [13:21:53<3:12:54, 1.36s/it] 72%|███████▏ | 21510/30000 [13:22:00<7:31:08, 3.19s/it] 72%|███████▏ | 21510/30000 [13:22:00<7:31:08, 3.19s/it] 72%|███████▏ | 21511/30000 [13:22:01<6:05:06, 2.58s/it] 72%|███████▏ | 21512/30000 [13:22:03<5:06:07, 2.16s/it] 72%|███████▏ | 21513/30000 [13:22:04<4:23:58, 1.87s/it] 72%|███████▏ | 21514/30000 [13:22:05<3:54:47, 1.66s/it] 72%|███████▏ | 21515/30000 [13:22:06<3:34:55, 1.52s/it] 72%|███████▏ | 21516/30000 [13:22:07<3:21:23, 1.42s/it] 72%|███████▏ | 21517/30000 [13:22:09<3:11:40, 1.36s/it] 72%|███████▏ | 21518/30000 [13:22:16<7:14:00, 3.07s/it] 72%|███████▏ | 21519/30000 [13:22:17<5:53:20, 2.50s/it] 72%|███████▏ | 21520/30000 [13:22:18<4:58:12, 2.11s/it] 72%|███████▏ | 21520/30000 [13:22:18<4:58:12, 2.11s/it] 72%|███████▏ | 21521/30000 [13:22:19<4:17:25, 1.82s/it] 72%|███████▏ | 21522/30000 [13:22:20<3:49:59, 1.63s/it] 72%|███████▏ | 21523/30000 [13:22:21<3:31:38, 1.50s/it] 72%|███████▏ | 21524/30000 [13:22:23<3:18:59, 1.41s/it] 72%|███████▏ | 21525/30000 [13:22:24<3:09:31, 1.34s/it] 72%|███████▏ | 21526/30000 [13:22:31<7:09:21, 3.04s/it] 72%|███████▏ | 21527/30000 [13:22:32<5:50:20, 2.48s/it] 72%|███████▏ | 21528/30000 [13:22:33<4:55:21, 2.09s/it] 72%|███████▏ | 21529/30000 [13:22:34<4:16:48, 1.82s/it] 72%|███████▏ | 21530/30000 [13:22:36<3:49:29, 1.63s/it] +{'loss': 0.0128, 'grad_norm': 0.055885475128889084, 'learning_rate': 6.631764879285531e-06, 'flow_matching_loss': 0.012835640367120505, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.14} +{'loss': 0.0132, 'grad_norm': 0.09167298674583435, 'learning_rate': 6.618047488586614e-06, 'flow_matching_loss': 0.013199290819466114, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.15} +{'loss': 0.0145, 'grad_norm': 0.06155386567115784, 'learning_rate': 6.604340282750371e-06, 'flow_matching_loss': 0.014526690077036618, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.15} +{'loss': 0.0134, 'grad_norm': 0.0793628916144371, 'learning_rate': 6.590643278432342e-06, 'flow_matching_loss': 0.01342006828635931, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.15} +{'loss': 0.0135, 'grad_norm': 0.06371542811393738, 'learning_rate': 6.576956492275682e-06, 'flow_matching_loss': 0.013493011053651572, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.16} +{'loss': 0.0146, 'grad_norm': 0.06251689046621323, 'learning_rate': 6.563279940911124e-06, 'flow_matching_loss': 0.014578517433255911, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.16} +{'loss': 0.012, 'grad_norm': 0.06362323462963104, 'learning_rate': 6.549613640956963e-06, 'flow_matching_loss': 0.011952918861061335, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.16} +{'loss': 0.0145, 'grad_norm': 0.08182073384523392, 'learning_rate': 6.535957609019053e-06, 'flow_matching_loss': 0.014548214431852103, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.17} +{'loss': 0.0136, 'grad_norm': 0.05403905361890793, 'learning_rate': 6.5223118616907545e-06, 'flow_matching_loss': 0.01357681406661868, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.17} +{'loss': 0.014, 'grad_norm': 0.07819663733243942, 'learning_rate': 6.508676415552939e-06, 'flow_matching_loss': 0.01395200053229928, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.17} +{'loss': 0.0147, 'grad_norm': 0.07384762912988663, 'learning_rate': 6.495051287173952e-06, 'flow_matching_loss': 0.014665459468960761, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.17} +{'loss': 0.0134, 'grad_norm': 0.06337407231330872, 'learning_rate': 6.481436493109621e-06, 'flow_matching_loss': 0.013386011682450771, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.18} +{'loss': 0.0144, 'grad_norm': 0.06983184069395065, 'learning_rate': 6.467832049903194e-06, 'flow_matching_loss': 0.014396184496581555, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.18} +{'loss': 0.0152, 'grad_norm': 0.08792704343795776, 'learning_rate': 6.454237974085361e-06, 'flow_matching_loss': 0.01518079824745655, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.18} +{'loss': 0.0125, 'grad_norm': 0.0681145116686821, 'learning_rate': 6.440654282174203e-06, 'flow_matching_loss': 0.012457642331719398, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.19} +{'loss': 0.0141, 'grad_norm': 0.08124446868896484, 'learning_rate': 6.427080990675187e-06, 'flow_matching_loss': 0.014071778766810894, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.19} +{'loss': 0.014, 'grad_norm': 0.08365116268396378, 'learning_rate': 6.41351811608114e-06, 'flow_matching_loss': 0.01398466075770557, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.19} +{'loss': 0.013, 'grad_norm': 0.06582597643136978, 'learning_rate': 6.399965674872228e-06, 'flow_matching_loss': 0.013002427946776151, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.19} +{'loss': 0.0145, 'grad_norm': 0.07184753566980362, 'learning_rate': 6.386423683515957e-06, 'flow_matching_loss': 0.014462905190885066, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.2} +{'loss': 0.0141, 'grad_norm': 0.09338876605033875, 'learning_rate': 6.372892158467119e-06, 'flow_matching_loss': 0.01407859562896192, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.2} +{'loss': 0.0142, 'grad_norm': 0.0934266522526741, 'learning_rate': 6.359371116167791e-06, 'flow_matching_loss': 0.014183247555047273, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.2} +{'loss': 0.0126, 'grad_norm': 0.0759960487484932, 'learning_rate': 6.345860573047314e-06, 'flow_matching_loss': 0.012562664225697517, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.21} +{'loss': 0.0144, 'grad_norm': 0.07659218460321426, 'learning_rate': 6.332360545522268e-06, 'flow_matching_loss': 0.01439686557278037, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.21} +{'loss': 0.0141, 'grad_norm': 0.09161235392093658, 'learning_rate': 6.31887104999647e-06, 'flow_matching_loss': 0.014106225594878197, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.21} +{'loss': 0.0129, 'grad_norm': 0.08210441470146179, 'learning_rate': 6.305392102860924e-06, 'flow_matching_loss': 0.012935131508857011, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.21} +{'loss': 0.0129, 'grad_norm': 0.06348702311515808, 'learning_rate': 6.291923720493814e-06, 'flow_matching_loss': 0.012861873675137758, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.22} +{'loss': 0.014, 'grad_norm': 0.07163286209106445, 'learning_rate': 6.278465919260507e-06, 'flow_matching_loss': 0.01403709566220641, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.22} +{'loss': 0.0126, 'grad_norm': 0.10804980248212814, 'learning_rate': 6.26501871551349e-06, 'flow_matching_loss': 0.012616245355457068, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.22} +{'loss': 0.0129, 'grad_norm': 0.08127621561288834, 'learning_rate': 6.251582125592394e-06, 'flow_matching_loss': 0.012905366905033588, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.23} +{'loss': 0.0134, 'grad_norm': 0.07373716682195663, 'learning_rate': 6.2381561658239386e-06, 'flow_matching_loss': 0.01336753936484456, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.23} +{'loss': 0.0127, 'grad_norm': 0.10665709525346756, 'learning_rate': 6.2247408525219276e-06, 'flow_matching_loss': 0.012651353422552348, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.23} +{'loss': 0.0133, 'grad_norm': 0.06875603646039963, 'learning_rate': 6.2113362019872335e-06, 'flow_matching_loss': 0.01331365630030632, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.24} +{'loss': 0.0128, 'grad_norm': 0.06800201535224915, 'learning_rate': 6.197942230507772e-06, 'flow_matching_loss': 0.012810017354786396, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.24} +{'loss': 0.0139, 'grad_norm': 0.07994898408651352, 'learning_rate': 6.184558954358475e-06, 'flow_matching_loss': 0.013947608321905136, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.24} +{'loss': 0.014, 'grad_norm': 0.08925843983888626, 'learning_rate': 6.1711863898012965e-06, 'flow_matching_loss': 0.01400615433230996, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.24} +{'loss': 0.0132, 'grad_norm': 0.0774691253900528, 'learning_rate': 6.157824553085155e-06, 'flow_matching_loss': 0.013187952153384686, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.25} +{'loss': 0.0132, 'grad_norm': 0.06112616881728172, 'learning_rate': 6.144473460445945e-06, 'flow_matching_loss': 0.013222523499280214, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.25} +{'loss': 0.0134, 'grad_norm': 0.09023316949605942, 'learning_rate': 6.131133128106503e-06, 'flow_matching_loss': 0.013364616129547358, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.25} +{'loss': 0.0133, 'grad_norm': 0.07730896770954132, 'learning_rate': 6.117803572276584e-06, 'flow_matching_loss': 0.013325887639075518, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.26} +{'loss': 0.0135, 'grad_norm': 0.08541591465473175, 'learning_rate': 6.104484809152866e-06, 'flow_matching_loss': 0.013505143113434315, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.26} +{'loss': 0.0148, 'grad_norm': 0.08976373076438904, 'learning_rate': 6.091176854918892e-06, 'flow_matching_loss': 0.014753915183246136, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.26} +{'loss': 0.0143, 'grad_norm': 0.06672367453575134, 'learning_rate': 6.077879725745092e-06, 'flow_matching_loss': 0.014273176621645688, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.26} 72%|███████▏ | 21530/30000 [13:22:36<3:49:29, 1.63s/it] 72%|███████▏ | 21531/30000 [13:22:37<3:30:51, 1.49s/it] 72%|███████▏ | 21532/30000 [13:22:38<3:17:42, 1.40s/it] 72%|███████▏ | 21533/30000 [13:22:39<3:07:59, 1.33s/it] 72%|███████▏ | 21534/30000 [13:22:46<7:02:33, 2.99s/it] 72%|███████▏ | 21535/30000 [13:22:47<5:45:29, 2.45s/it] 72%|███████▏ | 21536/30000 [13:22:48<4:52:08, 2.07s/it] 72%|███████▏ | 21537/30000 [13:22:50<4:13:56, 1.80s/it] 72%|███████▏ | 21538/30000 [13:22:51<3:47:27, 1.61s/it] 72%|███████▏ | 21539/30000 [13:22:52<3:29:42, 1.49s/it] 72%|███████▏ | 21540/30000 [13:22:53<3:17:13, 1.40s/it] 72%|███████▏ | 21540/30000 [13:22:53<3:17:13, 1.40s/it] 72%|███████▏ | 21541/30000 [13:22:54<3:08:47, 1.34s/it] 72%|███████▏ | 21542/30000 [13:23:01<7:08:36, 3.04s/it] 72%|███████▏ | 21543/30000 [13:23:02<5:50:21, 2.49s/it] 72%|███████▏ | 21544/30000 [13:23:04<4:55:22, 2.10s/it] 72%|███████▏ | 21545/30000 [13:23:05<4:16:01, 1.82s/it] 72%|███████▏ | 21546/30000 [13:23:06<3:48:21, 1.62s/it] 72%|███████▏ | 21547/30000 [13:23:07<3:30:16, 1.49s/it] 72%|███████▏ | 21548/30000 [13:23:08<3:17:01, 1.40s/it] 72%|███████▏ | 21549/30000 [13:23:10<3:08:21, 1.34s/it] 72%|███████▏ | 21550/30000 [13:23:18<8:01:39, 3.42s/it] 72%|███████▏ | 21550/30000 [13:23:18<8:01:39, 3.42s/it] 72%|███████▏ | 21551/30000 [13:23:19<6:26:26, 2.74s/it] 72%|███████▏ | 21552/30000 [13:23:20<5:21:46, 2.29s/it] 72%|███████▏ | 21553/30000 [13:23:21<4:34:21, 1.95s/it] 72%|███████▏ | 21554/30000 [13:23:23<4:01:26, 1.72s/it] 72%|███████▏ | 21555/30000 [13:23:24<3:39:07, 1.56s/it] 72%|███████▏ | 21556/30000 [13:23:25<3:23:44, 1.45s/it] 72%|███████▏ | 21557/30000 [13:23:26<3:12:45, 1.37s/it] 72%|███████▏ | 21558/30000 [13:23:34<7:33:12, 3.22s/it] 72%|███████▏ | 21559/30000 [13:23:35<6:06:16, 2.60s/it] 72%|███████▏ | 21560/30000 [13:23:36<5:06:28, 2.18s/it] 72%|███████▏ | 21560/30000 [13:23:36<5:06:28, 2.18s/it] 72%|███████▏ | 21561/30000 [13:23:37<4:24:04, 1.88s/it] 72%|███████▏ | 21562/30000 [13:23:38<3:53:30, 1.66s/it] 72%|███████▏ | 21563/30000 [13:23:40<3:33:46, 1.52s/it] 72%|███████▏ | 21564/30000 [13:23:41<3:19:37, 1.42s/it] 72%|███████▏ | 21565/30000 [13:23:42<3:09:33, 1.35s/it] 72%|███████▏ | 21566/30000 [13:23:49<6:55:10, 2.95s/it] 72%|███████▏ | 21567/30000 [13:23:50<5:39:58, 2.42s/it] 72%|███████▏ | 21568/30000 [13:23:51<4:48:34, 2.05s/it] 72%|███████▏ | 21569/30000 [13:23:52<4:11:03, 1.79s/it] 72%|███████▏ | 21570/30000 [13:23:53<3:44:51, 1.60s/it] 72%|███████▏ | 21570/30000 [13:23:53<3:44:51, 1.60s/it] 72%|███████▏ | 21571/30000 [13:23:55<3:27:52, 1.48s/it] 72%|███████▏ | 21572/30000 [13:23:56<3:15:48, 1.39s/it] 72%|███████▏ | 21573/30000 [13:23:57<3:07:31, 1.34s/it] 72%|███████▏ | 21574/30000 [13:24:05<7:58:56, 3.41s/it] 72%|███████▏ | 21575/30000 [13:24:06<6:24:15, 2.74s/it] 72%|███████▏ | 21576/30000 [13:24:08<5:19:29, 2.28s/it] 72%|███████▏ | 21577/30000 [13:24:09<4:32:54, 1.94s/it] 72%|███████▏ | 21578/30000 [13:24:10<4:00:52, 1.72s/it] 72%|███████▏ | 21579/30000 [13:24:11<3:38:46, 1.56s/it] 72%|███████▏ | 21580/30000 [13:24:12<3:23:06, 1.45s/it] 72%|███████▏ | 21580/30000 [13:24:12<3:23:06, 1.45s/it] 72%|███████▏ | 21581/30000 [13:24:13<3:12:04, 1.37s/it] 72%|███████▏ | 21582/30000 [13:24:21<7:38:59, 3.27s/it] 72%|███████▏ | 21583/30000 [13:24:22<6:09:52, 2.64s/it] 72%|███████▏ | 21584/30000 [13:24:23<5:08:46, 2.20s/it] 72%|███████▏ | 21585/30000 [13:24:25<4:25:19, 1.89s/it] 72%|███████▏ | 21586/30000 [13:24:26<3:54:46, 1.67s/it] 72%|███████▏ | 21587/30000 [13:24:27<3:34:11, 1.53s/it] 72%|███████▏ | 21588/30000 [13:24:28<3:19:57, 1.43s/it] 72%|███████▏ | 21589/30000 [13:24:29<3:10:50, 1.36s/it] 72%|███████▏ | 21590/30000 [13:24:37<7:43:44, 3.31s/it] 72%|███████▏ | 21590/30000 [13:24:37<7:43:44, 3.31s/it] 72%|███████▏ | 21591/30000 [13:24:38<6:13:31, 2.67s/it] 72%|███████▏ | 21592/30000 [13:24:40<5:11:08, 2.22s/it] 72%|███████▏ | 21593/30000 [13:24:41<4:27:24, 1.91s/it] 72%|███████▏ | 21594/30000 [13:24:42<3:55:45, 1.68s/it] 72%|███████▏ | 21595/30000 [13:24:43<3:35:43, 1.54s/it] 72%|███████▏ | 21596/30000 [13:24:44<3:20:55, 1.43s/it] 72%|███████▏ | 21597/30000 [13:24:46<3:10:13, 1.36s/it] 72%|███████▏ | 21598/30000 [13:24:52<6:56:37, 2.98s/it] 72%|███████▏ | 21599/30000 [13:24:53<5:41:05, 2.44s/it] 72%|███████▏ | 21600/30000 [13:24:55<4:48:45, 2.06s/it] 72%|███████▏ | 21600/30000 [13:24:55<4:48:45, 2.06s/it] 72%|███████▏ | 21601/30000 [13:24:56<4:11:23, 1.80s/it] 72%|███████▏ | 21602/30000 [13:24:57<3:45:52, 1.61s/it] 72%|███████▏ | 21603/30000 [13:24:58<3:27:53, 1.49s/it] 72%|███████▏ | 21604/30000 [13:24:59<3:15:17, 1.40s/it] 72%|███████▏ | 21605/30000 [13:25:01<3:06:23, 1.33s/it] 72%|███████▏ | 21606/30000 [13:25:08<7:13:39, 3.10s/it] 72%|███████▏ | 21607/30000 [13:25:09<5:52:27, 2.52s/it] 72%|███████▏ | 21608/30000 [13:25:10<4:56:33, 2.12s/it] 72%|███████▏ | 21609/30000 [13:25:11<4:15:43, 1.83s/it] 72%|███████▏ | 21610/30000 [13:25:12<3:47:07, 1.62s/it] 72%|███████▏ | 21610/30000 [13:25:12<3:47:07, 1.62s/it] 72%|███████▏ | 21611/30000 [13:25:14<3:30:38, 1.51s/it] 72%|███████▏ | 21612/30000 [13:25:15<3:18:33, 1.42s/it] 72%|███████▏ | 21613/30000 [13:25:16<3:08:47, 1.35s/it] 72%|███████▏ | 21614/30000 [13:25:24<7:26:16, 3.19s/it] 72%|███████▏ | 21615/30000 [13:25:25<6:01:24, 2.59s/it] 72%|███████▏ | 21616/30000 [13:25:26<5:04:50, 2.18s/it] 72%|███████▏ | 21617/30000 [13:25:27<4:22:27, 1.88s/it] 72%|███████▏ | 21618/30000 [13:25:28<3:52:11, 1.66s/it] 72%|███████▏ | 21619/30000 [13:25:29<3:31:58, 1.52s/it] 72%|███████▏ | 21620/30000 [13:25:31<3:18:12, 1.42s/it] 72%|███████▏ | 21620/30000 [13:25:31<3:18:12, 1.42s/it] 72%|███████▏ | 21621/30000 [13:25:32<3:10:56, 1.37s/it] 72%|███████▏ | 21622/30000 [13:25:38<6:43:14, 2.89s/it] 72%|███████▏ | 21623/30000 [13:25:40<5:30:40, 2.37s/it] 72%|███████▏ | 21624/30000 [13:25:41<4:41:04, 2.01s/it] 72%|███████▏ | 21625/30000 [13:25:42<4:05:37, 1.76s/it] 72%|███████▏ | 21626/30000 [13:25:43<3:40:35, 1.58s/it] 72%|███████▏ | 21627/30000 [13:25:44<3:25:15, 1.47s/it] 72%|███████▏ | 21628/30000 [13:25:45<3:13:21, 1.39s/it] 72%|███████▏ | 21629/30000 [13:25:47<3:05:13, 1.33s/it] 72%|███████▏ | 21630/30000 [13:25:54<7:09:39, 3.08s/it] 72%|███████▏ | 21630/30000 [13:25:54<7:09:39, 3.08s/it] 72%|███████▏ | 21631/30000 [13:25:55<5:49:34, 2.51s/it] 72%|███████▏ | 21632/30000 [13:25:56<4:54:56, 2.11s/it] 72%|███████▏ | 21633/30000 [13:25:57<4:16:11, 1.84s/it] 72%|███████▏ | 21634/30000 [13:25:59<3:48:56, 1.64s/it] 72%|███████▏ | 21635/30000 [13:26:00<3:30:10, 1.51s/it] 72%|███████▏ | 21636/30000 [13:26:01<3:16:38, 1.41s/it] 72%|███████▏ | 21637/30000 [13:26:02<3:07:36, 1.35s/it] 72%|███████▏ | 21638/30000 [13:26:10<7:26:02, 3.20s/it] 72%|███████▏ | 21639/30000 [13:26:11<6:00:59, 2.59s/it] 72%|███████▏ | 21640/30000 [13:26:12<5:02:25, 2.17s/it] 72%|███████▏ | 21640/30000 [13:26:12<5:02:25, 2.17s/it] 72%|███████▏ | 21641/30000 [13:26:13<4:20:12, 1.87s/it] 72%|███████▏ | 21642/30000 [13:26:14<3:50:42, 1.66s/it] 72%|███████▏ | 21643/30000 [13:26:15<3:30:47, 1.51s/it] 72%|███████▏ | 21644/30000 [13:26:17<3:17:08, 1.42s/it] 72%|███████▏ | 21645/30000 [13:26:18<3:07:20, 1.35s/it] 72%|███████▏ | 21646/30000 [13:26:25<7:25:04, 3.20s/it] 72%|███████▏ | 21647/30000 [13:26:27<6:00:11, 2.59s/it] 72%|███████▏ | 21648/30000 [13:26:28<5:01:36, 2.17s/it] 72%|███████▏ | 21649/30000 [13:26:29<4:20:20, 1.87s/it] 72%|███████▏ | 21650/30000 [13:26:30<3:50:45, 1.66s/it] 72%|███████▏ | 21650/30000 [13:26:30<3:50:45, 1.66s/it] 72%|███████▏ | 21651/30000 [13:26:31<3:31:45, 1.52s/it] 72%|███████▏ | 21652/30000 [13:26:32<3:17:59, 1.42s/it] 72%|███████▏ | 21653/30000 [13:26:34<3:08:14, 1.35s/it] 72%|███████▏ | 21654/30000 [13:26:40<6:43:47, 2.90s/it] 72%|███████▏ | 21655/30000 [13:26:41<5:31:53, 2.39s/it] 72%|███████▏ | 21656/30000 [13:26:43<4:42:57, 2.03s/it] 72%|███████▏ | 21657/30000 [13:26:44<4:06:42, 1.77s/it] 72%|███████▏ | 21658/30000 [13:26:45<3:40:56, 1.59s/it] 72%|███████▏ | 21659/30000 [13:26:46<3:24:16, 1.47s/it] 72%|███████▏ | 21660/30000 [13:26:47<3:12:21, 1.38s/it] 72%|███████▏ | 21660/30000 [13:26:47<3:12:21, 1.38s/it] 72%|███████▏ | 21661/30000 [13:26:48<3:04:29, 1.33s/it] 72%|███████▏ | 21662/30000 [13:26:57<7:51:17, 3.39s/it] 72%|███████▏ | 21663/30000 [13:26:58<6:18:47, 2.73s/it] 72%|███████▏ | 21664/30000 [13:26:59<5:15:11, 2.27s/it] 72%|███████▏ | 21665/30000 [13:27:00<4:29:59, 1.94s/it] 72%|███████▏ | 21666/30000 [13:27:01<3:57:42, 1.71s/it] 72%|███████▏ | 21667/30000 [13:27:03<3:35:52, 1.55s/it] 72%|███████▏ | 21668/30000 [13:27:04<3:20:52, 1.45s/it] 72%|███████▏ | 21669/30000 [13:27:05<3:09:38, 1.37s/it] 72%|███████▏ | 21670/30000 [13:27:12<7:19:42, 3.17s/it] 72%|███████▏ | 21670/30000 [13:27:12<7:19:42, 3.17s/it] 72%|███████▏ | 21671/30000 [13:27:13<5:55:55, 2.56s/it] 72%|███████▏ | 21672/30000 [13:27:15<4:58:38, 2.15s/it] 72%|███████▏ | 21673/30000 [13:27:16<4:17:51, 1.86s/it] 72%|███████▏ | 21674/30000 [13:27:17<3:48:13, 1.64s/it] 72%|███████▏ | 21675/30000 [13:27:18<3:29:20, 1.51s/it] 72%|███████▏ | 21676/30000 [13:27:19<3:16:00, 1.41s/it] 72%|███████▏ | 21677/30000 [13:27:21<3:06:00, 1.34s/it] 72%|███████▏ | 21678/30000 [13:27:29<7:48:04, 3.37s/it] 72%|███████▏ | 21679/30000 [13:27:30<6:16:33, 2.72s/it] 72%|███████▏ | 21680/30000 [13:27:31<5:13:38, 2.26s/it] 72%|███████▏ | 21680/30000 [13:27:31<5:13:38, 2.26s/it] 72%|███████▏ | 21681/30000 [13:27:32<4:28:18, 1.94s/it] 72%|███████▏ | 21682/30000 [13:27:33<3:55:57, 1.70s/it] 72%|███████▏ | 21683/30000 [13:27:35<3:34:14, 1.55s/it] 72%|███████▏ | 21684/30000 [13:27:36<3:18:49, 1.43s/it] 72%|███████▏ | 21685/30000 [13:27:37<3:08:32, 1.36s/it] 72%|███████▏ | 21686/30000 [13:27:44<7:11:35, 3.11s/it] 72%|███████▏ | 21687/30000 [13:27:45<5:51:34, 2.54s/it] 72%|███████▏ | 21688/30000 [13:27:47<4:56:22, 2.14s/it] 72%|███████▏ | 21689/30000 [13:27:48<4:16:53, 1.85s/it] 72%|███████▏ | 21690/30000 [13:27:49<3:48:08, 1.65s/it] 72%|███████▏ | 21690/30000 [13:27:49<3:48:08, 1.65s/it] 72%|███████▏ | 21691/30000 [13:27:50<3:29:27, 1.51s/it] 72%|███████▏ | 21692/30000 [13:27:51<3:16:24, 1.42s/it] 72%|███████▏ | 21693/30000 [13:27:52<3:06:55, 1.35s/it] 72%|███████▏ | 21694/30000 [13:28:01<8:03:43, 3.49s/it] 72%|███████▏ | 21695/30000 [13:28:02<6:26:53, 2.80s/it] 72%|███████▏ | 21696/30000 [13:28:03<5:20:11, 2.31s/it] 72%|███████▏ | 21697/30000 [13:28:05<4:33:25, 1.98s/it] 72%|███████▏ | 21698/30000 [13:28:06<4:01:37, 1.75s/it] 72%|███████▏ | 21699/30000 [13:28:07<3:38:50, 1.58s/it] 72%|███████▏ | 21700/30000 [13:28:08<3:22:05, 1.46s/it] 72%|███████▏ | 21700/30000 [13:28:08<3:22:05, 1.46s/it] 72%|███████▏ | 21701/30000 [13:28:09<3:10:37, 1.38s/it] 72%|███████▏ | 21702/30000 [13:28:17<7:34:21, 3.29s/it] 72%|███████▏ | 21703/30000 [13:28:18<6:06:00, 2.65s/it] 72%|███████▏ | 21704/30000 [13:28:19<5:05:34, 2.21s/it] 72%|███████▏ | 21705/30000 [13:28:21<4:21:56, 1.89s/it] 72%|███████▏ | 21706/30000 [13:28:22<3:52:10, 1.68s/it] 72%|███████▏ | 21707/30000 [13:28:23<3:31:10, 1.53s/it] 72%|███████▏ | 21708/30000 [13:28:24<3:17:15, 1.43s/it] 72%|███████▏ | 21709/30000 [13:28:25<3:07:31, 1.36s/it] 72%|███████▏ | 21710/30000 [13:28:33<7:49:03, 3.39s/it] 72%|███████▏ | 21710/30000 [13:28:33<7:49:03, 3.39s/it] 72%|███████▏ | 21711/30000 [13:28:35<6:17:35, 2.73s/it] 72%|███████▏ | 21712/30000 [13:28:36<5:14:51, 2.28s/it] 72%|███████▏ | 21713/30000 [13:28:37<4:29:22, 1.95s/it] 72%|███████▏ | 21714/30000 [13:28:38<3:56:54, 1.72s/it] 72%|███████▏ | 21715/30000 [13:28:39<3:34:56, 1.56s/it] 72%|███████▏ | 21716/30000 [13:28:41<3:19:26, 1.44s/it] 72%|███████▏ | 21717/30000 [13:28:42<3:08:31, 1.37s/it] 72%|███████▏ | 21718/30000 [13:28:49<7:28:21, 3.25s/it] 72%|███████▏ | 21719/30000 [13:28:51<6:01:58, 2.62s/it] 72%|███████▏ | 21720/30000 [13:28:52<5:03:01, 2.20s/it] 72%|███████▏ | 21720/30000 [13:28:52<5:03:01, 2.20s/it] 72%|███████▏ | 21721/30000 [13:28:53<4:22:23, 1.90s/it] 72%|███████▏ | 21722/30000 [13:28:54<3:51:49, 1.68s/it] 72%|███████▏ | 21723/30000 [13:28:55<3:31:31, 1.53s/it] 72%|███████▏ | 21724/30000 [13:28:56<3:17:09, 1.43s/it] 72%|███████▏ | 21725/30000 [13:28:58<3:07:09, 1.36s/it] 72%|███████▏ | 21726/30000 [13:29:05<7:08:09, 3.10s/it] 72%|███████▏ | 21727/30000 [13:29:06<5:47:52, 2.52s/it] 72%|███████▏ | 21728/30000 [13:29:07<4:53:01, 2.13s/it] 72%|███████▏ | 21729/30000 [13:29:08<4:13:44, 1.84s/it] 72%|███████▏ | 21730/30000 [13:29:10<3:46:12, 1.64s/it] 72%|███████▏ | 21730/30000 [13:29:10<3:46:12, 1.64s/it] 72%|███████▏ | 21731/30000 [13:29:11<3:28:21, 1.51s/it] 72%|███████▏ | 21732/30000 [13:29:12<3:15:35, 1.42s/it] 72%|███████▏ | 21733/30000 [13:29:13<3:07:51, 1.36s/it] 72%|███████▏ | 21734/30000 [13:29:20<6:49:38, 2.97s/it] 72%|███████▏ | 21735/30000 [13:29:21<5:34:50, 2.43s/it] 72%|███████▏ | 21736/30000 [13:29:22<4:44:07, 2.06s/it] 72%|███████▏ | 21737/30000 [13:29:23<4:06:46, 1.79s/it] 72%|███████▏ | 21738/30000 [13:29:25<3:41:12, 1.61s/it] 72%|███████▏ | 21739/30000 [13:29:26<3:24:02, 1.48s/it] 72%|███████▏ | 21740/30000 [13:29:27<3:11:47, 1.39s/it] 72%|███████▏ | 21740/30000 [13:29:27<3:11:47, 1.39s/it] 72%|███████▏ | 21741/30000 [13:29:28<3:02:55, 1.33s/it] 72%|███████▏ | 21742/30000 [13:29:36<7:22:28, 3.21s/it] 72%|███████▏ | 21743/30000 [13:29:37<5:57:47, 2.60s/it] 72%|███████▏ | 21744/30000 [13:29:38<4:59:54, 2.18s/it] 72%|███████▏ | 21745/30000 [13:29:39<4:18:55, 1.88s/it] 72%|███████▏ | 21746/30000 [13:29:41<3:49:53, 1.67s/it] 72%|███████▏ | 21747/30000 [13:29:42<3:30:12, 1.53s/it] 72%|███████▏ | 21748/30000 [13:29:43<3:16:36, 1.43s/it] 72%|███████▏ | 21749/30000 [13:29:44<3:06:43, 1.36s/it] 72%|███████▎ | 21750/30000 [13:29:51<7:06:18, 3.10s/it] 72%|███████▎ | 21750/30000 [13:29:51<7:06:18, 3.10s/it] 73%|███████▎ | 21751/30000 [13:29:52<5:46:05, 2.52s/it] 73%|███████▎ | 21752/30000 [13:29:54<4:52:19, 2.13s/it] 73%|███████▎ | 21753/30000 [13:29:55<4:13:09, 1.84s/it] 73%|███████▎ | 21754/30000 [13:29:56<3:44:51, 1.64s/it] 73%|███████▎ | 21755/30000 [13:29:57<3:26:11, 1.50s/it] 73%|███████▎ | 21756/30000 [13:29:58<3:13:28, 1.41s/it] 73%|███████▎ | 21757/30000 [13:30:00<3:04:40, 1.34s/it] 73%|███████▎ | 21758/30000 [13:30:08<7:36:19, 3.32s/it] 73%|███████▎ | 21759/30000 [13:30:09<6:08:24, 2.68s/it] 73%|███████▎ | 21760/30000 [13:30:10<5:06:59, 2.24s/it] 73%|███████▎ | 21760/30000 [13:30:10<5:06:59, 2.24s/it] 73%|███████▎ | 21761/30000 [13:30:11<4:23:45, 1.92s/it] 73%|███████▎ | 21762/30000 [13:30:12<3:53:15, 1.70s/it] 73%|███████▎ | 21763/30000 [13:30:13<3:32:14, 1.55s/it] 73%|███████▎ | 21764/30000 [13:30:15<3:17:02, 1.44s/it] 73%|███████▎ | 21765/30000 [13:30:16<3:06:21, 1.36s/it] 73%|███████▎ | 21766/30000 [13:30:23<7:25:53, 3.25s/it] 73%|███████▎ | 21767/30000 [13:30:25<6:00:07, 2.62s/it] 73%|███████▎ | 21768/30000 [13:30:26<5:01:04, 2.19s/it] 73%|███████▎ | 21769/30000 [13:30:27<4:18:54, 1.89s/it] 73%|███████▎ | 21770/30000 [13:30:28<3:48:41, 1.67s/it] 73%|███████▎ | 21770/30000 [13:30:28<3:48:41, 1.67s/it] 73%|███████▎ | 21771/30000 [13:30:29<3:29:04, 1.52s/it] 73%|███████▎ | 21772/30000 [13:30:31<3:15:19, 1.42s/it] 73%|███████▎ | 21773/30000 [13:30:32<3:05:32, 1.35s/it] 73%|███████▎ | 21774/30000 [13:30:39<7:24:21, 3.24s/it] 73%|███████▎ | 21775/30000 [13:30:41<5:58:52, 2.62s/it] 73%|███████▎ | 21776/30000 [13:30:42<5:01:31, 2.20s/it] 73%|███████▎ | 21777/30000 [13:30:43<4:19:08, 1.89s/it] 73%|███████▎ | 21778/30000 [13:30:44<3:51:09, 1.69s/it] 73%|███████▎ | 21779/30000 [13:30:45<3:31:55, 1.55s/it] 73%|███████▎ | 21780/30000 [13:30:47<3:16:25, 1.43s/it] 73%|███████▎ | 21780/30000 [13:30:47<3:16:25, 1.43s/it] 73%|███████▎ | 21781/30000 [13:30:48<3:06:41, 1.36s/it] 73%|███████▎ | 21782/30000 [13:30:55<7:23:50, 3.24s/it] 73%|███████▎ | 21783/30000 [13:30:57<5:58:19, 2.62s/it] 73%|███████▎ | 21784/30000 [13:30:58<5:06:46, 2.24s/it] 73%|███████▎ | 21785/30000 [13:30:59<4:22:07, 1.91s/it] 73%|███████▎ | 21786/30000 [13:31:00<3:51:04, 1.69s/it] 73%|███████▎ | 21787/30000 [13:31:01<3:31:07, 1.54s/it] 73%|███████▎ | 21788/30000 [13:31:03<3:17:09, 1.44s/it] 73%|███████▎ | 21789/30000 [13:31:04<3:06:44, 1.36s/it] 73%|███████▎ | 21790/30000 [13:31:11<6:54:00, 3.03s/it] 73%|███████▎ | 21790/30000 [13:31:11<6:54:00, 3.03s/it] 73%|███████▎ | 21791/30000 [13:31:12<5:37:26, 2.47s/it] 73%|███████▎ | 21792/30000 [13:31:13<4:55:37, 2.16s/it] 73%|███████▎ | 21793/30000 [13:31:14<4:14:15, 1.86s/it] 73%|███████▎ | 21794/30000 [13:31:16<3:45:59, 1.65s/it] 73%|███████▎ | 21795/30000 [13:31:17<3:27:08, 1.51s/it] 73%|███████▎ | 21796/30000 [13:31:18<3:13:49, 1.42s/it] 73%|███████▎ | 21797/30000 [13:31:19<3:04:23, 1.35s/it] 73%|███████▎ | 21798/30000 [13:31:26<7:02:10, 3.09s/it] 73%|███████▎ | 21799/30000 [13:31:28<5:43:12, 2.51s/it] 73%|███████▎ | 21800/30000 [13:31:30<5:30:41, 2.42s/it] 73%|███████▎ | 21800/30000 [13:31:30<5:30:41, 2.42s/it] 73%|███████▎ | 21801/30000 [13:31:31<4:39:27, 2.05s/it] 73%|███████▎ | 21802/30000 [13:31:32<4:03:15, 1.78s/it] 73%|███████▎ | 21803/30000 [13:31:33<3:38:44, 1.60s/it] 73%|███████▎ | 21804/30000 [13:31:34<3:21:41, 1.48s/it] 73%|███████▎ | 21805/30000 [13:31:36<3:10:05, 1.39s/it] 73%|███████▎ | 21806/30000 [13:31:42<6:50:47, 3.01s/it] 73%|███████▎ | 21807/30000 [13:31:44<5:35:23, 2.46s/it] 73%|███████▎ | 21808/30000 [13:31:46<5:26:40, 2.39s/it] 73%|███████▎ | 21809/30000 [13:31:47<4:36:48, 2.03s/it] 73%|███████▎ | 21810/30000 [13:31:48<4:01:48, 1.77s/it] 73%|███████▎ | 21810/30000 [13:31:48<4:01:48, 1.77s/it] 73%|███████▎ | 21811/30000 [13:31:49<3:37:52, 1.60s/it] 73%|███████▎ | 21812/30000 [13:31:51<3:20:32, 1.47s/it] 73%|███████▎ | 21813/30000 [13:31:52<3:08:30, 1.38s/it] 73%|███████▎ | 21814/30000 [13:31:59<6:56:02, 3.05s/it] 73%|███████▎ | 21815/30000 [13:32:00<5:38:57, 2.48s/it] 73%|███████▎ | 21816/30000 [13:32:02<5:42:57, 2.51s/it] 73%|███████▎ | 21817/30000 [13:32:04<4:47:35, 2.11s/it] 73%|███████▎ | 21818/30000 [13:32:05<4:08:43, 1.82s/it] 73%|███████▎ | 21819/30000 [13:32:06<3:43:01, 1.64s/it] 73%|███████▎ | 21820/30000 [13:32:07<3:24:35, 1.50s/it] 73%|███████▎ | 21820/30000 [13:32:07<3:24:35, 1.50s/it] 73%|███████▎ | 21821/30000 [13:32:08<3:12:11, 1.41s/it] 73%|███████▎ | 21822/30000 [13:32:15<6:38:49, 2.93s/it] 73%|███████▎ | 21823/30000 [13:32:16<5:26:36, 2.40s/it] 73%|███████▎ | 21824/30000 [13:32:19<5:55:45, 2.61s/it] 73%|███████▎ | 21825/30000 [13:32:20<4:56:42, 2.18s/it] 73%|███████▎ | 21826/30000 [13:32:21<4:15:21, 1.87s/it] 73%|███████▎ | 21827/30000 [13:32:23<3:47:44, 1.67s/it] 73%|███████▎ | 21828/30000 [13:32:24<3:27:39, 1.52s/it] 73%|███████▎ | 21829/30000 [13:32:25<3:13:29, 1.42s/it] 73%|███████▎ | 21830/30000 [13:32:31<6:22:27, 2.81s/it] 73%|███████▎ | 21830/30000 [13:32:31<6:22:27, 2.81s/it] 73%|███████▎ | 21831/30000 [13:32:32<5:15:44, 2.32s/it] 73%|███████▎ | 21832/30000 [13:32:35<5:46:25, 2.54s/it] 73%|███████▎ | 21833/30000 [13:32:36<4:50:22, 2.13s/it] 73%|███████▎ | 21834/30000 [13:32:38<4:10:31, 1.84s/it] 73%|███████▎ | 21835/30000 [13:32:39<3:43:57, 1.65s/it] 73%|███████▎ | 21836/30000 [13:32:40<3:25:04, 1.51s/it] 73%|███████▎ | 21837/30000 [13:32:41<3:12:15, 1.41s/it] 73%|███████▎ | 21838/30000 [13:32:47<5:55:07, 2.61s/it] 73%|███████▎ | 21839/30000 [13:32:48<4:55:39, 2.17s/it] 73%|███████▎ | 21840/30000 [13:32:52<6:09:55, 2.72s/it] 73%|███████▎ | 21840/30000 [13:32:52<6:09:55, 2.72s/it] 73%|███████▎ | 21841/30000 [13:32:53<5:06:31, 2.25s/it] 73%|███████▎ | 21842/30000 [13:32:54<4:21:37, 1.92s/it] 73%|███████▎ | 21843/30000 [13:32:55<3:50:55, 1.70s/it] 73%|███████▎ | 21844/30000 [13:32:56<3:29:46, 1.54s/it] 73%|███████▎ | 21845/30000 [13:32:58<3:15:26, 1.44s/it] 73%|███████▎ | 21846/30000 [13:33:02<5:09:05, 2.27s/it] 73%|███████▎ | 21847/30000 [13:33:03<4:23:47, 1.94s/it] 73%|███████▎ | 21848/30000 [13:33:08<6:20:24, 2.80s/it] 73%|███████▎ | 21849/30000 [13:33:09<5:13:46, 2.31s/it] 73%|███████▎ | 21850/30000 [13:33:10<4:27:07, 1.97s/it] 73%|███████▎ | 21850/30000 [13:33:10<4:27:07, 1.97s/it] 73%|███████▎ | 21851/30000 [13:33:11<3:55:12, 1.73s/it] 73%|███████▎ | 21852/30000 [13:33:12<3:32:56, 1.57s/it] 73%|███████▎ | 21853/30000 [13:33:14<3:17:29, 1.45s/it] 73%|███████▎ | 21854/30000 [13:33:17<4:53:28, 2.16s/it] 73%|███████▎ | 21855/30000 [13:33:19<4:13:02, 1.86s/it] 73%|███████▎ | 21856/30000 [13:33:24<6:46:44, 3.00s/it] 73%|███████▎ | 21857/30000 [13:33:25<5:32:01, 2.45s/it] 73%|███████▎ | 21858/30000 [13:33:27<4:39:48, 2.06s/it] 73%|███████▎ | 21859/30000 [13:33:28<4:03:37, 1.80s/it] 73%|███████▎ | 21860/30000 [13:33:29<3:38:55, 1.61s/it] 73%|███████▎ | 21860/30000 [13:33:29<3:38:55, 1.61s/it] 73%|███████▎ | 21861/30000 [13:33:30<3:21:37, 1.49s/it] 73%|███████▎ | 21862/30000 [13:33:33<4:08:26, 1.83s/it] 73%|███████▎ | 21863/30000 [13:33:34<3:43:11, 1.65s/it] 73%|███████▎ | 21864/30000 [13:33:40<6:27:41, 2.86s/it] 73%|███████▎ | 21865/30000 [13:33:41<5:18:58, 2.35s/it] 73%|███████▎ | 21866/30000 [13:33:42<4:30:51, 2.00s/it] 73%|███████▎ | 21867/30000 [13:33:43<3:58:14, 1.76s/it] 73%|███████▎ | 21868/30000 [13:33:44<3:34:58, 1.59s/it] 73%|███████▎ | 21869/30000 [13:33:46<3:18:32, 1.47s/it] 73%|███████▎ | 21870/30000 [13:33:49<4:41:01, 2.07s/it] 73%|███████▎ | 21870/30000 [13:33:49<4:41:01, 2.07s/it] 73%|███████▎ | 21871/30000 [13:33:50<4:04:01, 1.80s/it] 73%|███████▎ | 21872/30000 [13:33:56<6:26:02, 2.85s/it] 73%|███████▎ | 21873/30000 [13:33:57<5:17:20, 2.34s/it] 73%|███████▎ | 21874/30000 [13:33:58<4:29:16, 1.99s/it] 73%|███████▎ | 21875/30000 [13:33:59<3:56:23, 1.75s/it] 73%|███████▎ | 21876/30000 [13:34:00<3:32:55, 1.57s/it] 73%|███████▎ | 21877/30000 [13:34:01<3:17:01, 1.46s/it] 73%|███████▎ | 21878/30000 [13:34:06<5:23:34, 2.39s/it] 73%|███████▎ | 21879/30000 [13:34:07<4:34:02, 2.02s/it] 73%|███████▎ | 21880/30000 [13:34:11<5:53:02, 2.61s/it] 73%|███████▎ | 21880/30000 [13:34:11<5:53:02, 2.61s/it] 73%|███████▎ | 21881/30000 [13:34:12<4:54:46, 2.18s/it] 73%|███████▎ | 21882/30000 [13:34:13<4:13:35, 1.87s/it] 73%|███████▎ | 21883/30000 [13:34:15<3:45:30, 1.67s/it] 73%|███████▎ | 21884/30000 [13:34:16<3:26:05, 1.52s/it] 73%|███████▎ | 21885/30000 [13:34:17<3:12:11, 1.42s/it] 73%|███████▎ | 21886/30000 [13:34:22<5:44:38, 2.55s/it] 73%|███████▎ | 21887/30000 [13:34:23<4:48:29, 2.13s/it] 73%|███████▎ | 21888/30000 [13:34:27<5:52:05, 2.60s/it] 73%|███████▎ | 21889/30000 [13:34:28<4:53:23, 2.17s/it] 73%|███████▎ | 21890/30000 [13:34:29<4:12:17, 1.87s/it] 73%|███████▎ | 21890/30000 [13:34:29<4:12:17, 1.87s/it] 73%|███████▎ | 21891/30000 [13:34:31<3:44:41, 1.66s/it] 73%|███████▎ | 21892/30000 [13:34:32<3:25:10, 1.52s/it] 73%|███████▎ | 21893/30000 [13:34:33<3:11:22, 1.42s/it] 73%|███████▎ | 21894/30000 [13:34:38<5:52:37, 2.61s/it] 73%|███████▎ | 21895/30000 [13:34:39<4:54:27, 2.18s/it] 73%|███████▎ | 21896/30000 [13:34:43<5:31:29, 2.45s/it] 73%|███████▎ | 21897/30000 [13:34:44<4:39:21, 2.07s/it] 73%|███████▎ | 21898/30000 [13:34:45<4:02:32, 1.80s/it] 73%|███████▎ | 21899/30000 [13:34:46<3:38:31, 1.62s/it] 73%|███████▎ | 21900/30000 [13:34:47<3:23:17, 1.51s/it] 73%|███████▎ | 21900/30000 [13:34:47<3:23:17, 1.51s/it] 73%|███████▎ | 21901/30000 [13:34:49<3:11:10, 1.42s/it] 73%|███████▎ | 21902/30000 [13:34:54<6:00:25, 2.67s/it] 73%|███████▎ | 21903/30000 [13:34:55<4:59:21, 2.22s/it] 73%|███████▎ | 21904/30000 [13:34:57<4:36:21, 2.05s/it] 73%|███████▎ | 21905/30000 [13:34:59<4:22:51, 1.95s/it] 73%|███████▎ | 21906/30000 [13:35:00<3:52:03, 1.72s/it] 73%|███████▎ | 21907/30000 [13:35:01<3:30:29, 1.56s/it] 73%|███████▎ | 21908/30000 [13:35:02<3:15:35, 1.45s/it] 73%|███████▎ | 21909/30000 [13:35:03<3:05:28, 1.38s/it] 73%|███████▎ | 21910/30000 [13:35:09<5:56:53, 2.65s/it] 73%|███████▎ | 21910/30000 [13:35:09<5:56:53, 2.65s/it] 73%|███████▎ | 21911/30000 [13:35:10<4:57:02, 2.20s/it] 73%|███████▎ | 21912/30000 [13:35:12<4:49:24, 2.15s/it] 73%|███████▎ | 21913/30000 [13:35:15<4:58:45, 2.22s/it] 73%|███████▎ | 21914/30000 [13:35:16<4:16:35, 1.90s/it] 73%|███████▎ | 21915/30000 [13:35:17<3:47:28, 1.69s/it] 73%|███████▎ | 21916/30000 [13:35:18<3:27:19, 1.54s/it] 73%|███████▎ | 21917/30000 [13:35:19<3:12:07, 1.43s/it] 73%|███████▎ | 21918/30000 [13:35:25<5:57:51, 2.66s/it] 73%|███████▎ | 21919/30000 [13:35:26<4:57:25, 2.21s/it] 73%|███████▎ | 21920/30000 [13:35:28<4:49:54, 2.15s/it] 73%|███████▎ | 21920/30000 [13:35:28<4:49:54, 2.15s/it] 73%|███████▎ | 21921/30000 [13:35:31<5:23:56, 2.41s/it] 73%|███████▎ | 21922/30000 [13:35:32<4:34:17, 2.04s/it] 73%|███████▎ | 21923/30000 [13:35:33<4:00:10, 1.78s/it] 73%|███████▎ | 21924/30000 [13:35:35<3:35:45, 1.60s/it] 73%|███████▎ | 21925/30000 [13:35:36<3:18:48, 1.48s/it] 73%|███████▎ | 21926/30000 [13:35:41<5:51:41, 2.61s/it] 73%|███████▎ | 21927/30000 [13:35:42<4:53:16, 2.18s/it] 73%|███████▎ | 21928/30000 [13:35:44<4:57:35, 2.21s/it] 73%|███████▎ | 21929/30000 [13:35:47<5:09:35, 2.30s/it] 73%|███████▎ | 21930/30000 [13:35:48<4:23:30, 1.96s/it] 73%|███████▎ | 21930/30000 [13:35:48<4:23:30, 1.96s/it] 73%|███████▎ | 21931/30000 [13:35:49<3:52:22, 1.73s/it] 73%|███████▎ | 21932/30000 [13:35:51<3:30:34, 1.57s/it] 73%|███████▎ | 21933/30000 [13:35:52<3:15:11, 1.45s/it] 73%|███████▎ | 21934/30000 [13:35:57<5:32:58, 2.48s/it] 73%|███████▎ | 21935/30000 [13:35:58<4:40:28, 2.09s/it] 73%|███████▎ | 21936/30000 [13:36:00<4:58:17, 2.22s/it] 73%|███████▎ | 21937/30000 [13:36:04<5:44:52, 2.57s/it] 73%|███████▎ | 21938/30000 [13:36:05<4:47:35, 2.14s/it] 73%|███████▎ | 21939/30000 [13:36:06<4:08:50, 1.85s/it] 73%|███████▎ | 21940/30000 [13:36:07<3:42:02, 1.65s/it] 73%|███████▎ | 21940/30000 [13:36:07<3:42:02, 1.65s/it] 73%|███████▎ | 21941/30000 [13:36:08<3:22:51, 1.51s/it] 73%|███████▎ | 21942/30000 [13:36:12<4:48:37, 2.15s/it] 73%|███████▎ | 21943/30000 [13:36:13<4:09:31, 1.86s/it] 73%|███████▎ | 21944/30000 [13:36:16<4:43:16, 2.11s/it] 73%|███████▎ | 21945/30000 [13:36:20<6:01:57, 2.70s/it] 73%|███████▎ | 21946/30000 [13:36:21<5:00:45, 2.24s/it] 73%|███████▎ | 21947/30000 [13:36:22<4:19:14, 1.93s/it] 73%|███████▎ | 21948/30000 [13:36:24<3:49:13, 1.71s/it] 73%|███████▎ | 21949/30000 [13:36:25<3:28:28, 1.55s/it] 73%|███████▎ | 21950/30000 [13:36:27<4:02:14, 1.81s/it] +{'loss': 0.0141, 'grad_norm': 0.06204155087471008, 'learning_rate': 6.064593437788727e-06, 'flow_matching_loss': 0.014067056216299533, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.27} +{'loss': 0.013, 'grad_norm': 0.07347016781568527, 'learning_rate': 6.051318007193881e-06, 'flow_matching_loss': 0.012958420999348163, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.27} +{'loss': 0.0133, 'grad_norm': 0.08531838655471802, 'learning_rate': 6.038053450091469e-06, 'flow_matching_loss': 0.013255035411566496, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.27} +{'loss': 0.0138, 'grad_norm': 0.07470180839300156, 'learning_rate': 6.0247997825991705e-06, 'flow_matching_loss': 0.013811799231916666, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.28} +{'loss': 0.0136, 'grad_norm': 0.07131385803222656, 'learning_rate': 6.011557020821444e-06, 'flow_matching_loss': 0.013600754830986262, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.28} +{'loss': 0.0133, 'grad_norm': 0.05311321094632149, 'learning_rate': 5.99832518084949e-06, 'flow_matching_loss': 0.01325705014169216, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.28} +{'loss': 0.0145, 'grad_norm': 0.0680558905005455, 'learning_rate': 5.98510427876124e-06, 'flow_matching_loss': 0.014485093299299479, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.28} +{'loss': 0.0137, 'grad_norm': 0.07896234840154648, 'learning_rate': 5.971894330621344e-06, 'flow_matching_loss': 0.013706157077103853, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.29} +{'loss': 0.0146, 'grad_norm': 0.062067266553640366, 'learning_rate': 5.958695352481134e-06, 'flow_matching_loss': 0.01459311619400978, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.29} +{'loss': 0.0127, 'grad_norm': 0.0762583315372467, 'learning_rate': 5.945507360378608e-06, 'flow_matching_loss': 0.01266593225300312, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.29} +{'loss': 0.015, 'grad_norm': 0.0864642858505249, 'learning_rate': 5.9323303703384225e-06, 'flow_matching_loss': 0.014999859035015106, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.3} +{'loss': 0.0135, 'grad_norm': 0.10498881340026855, 'learning_rate': 5.919164398371857e-06, 'flow_matching_loss': 0.013467316515743732, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.3} +{'loss': 0.0132, 'grad_norm': 0.08467771112918854, 'learning_rate': 5.906009460476818e-06, 'flow_matching_loss': 0.01315827462822199, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.3} +{'loss': 0.0143, 'grad_norm': 0.07239816337823868, 'learning_rate': 5.892865572637789e-06, 'flow_matching_loss': 0.014299181196838618, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.3} +{'loss': 0.0134, 'grad_norm': 0.06359751522541046, 'learning_rate': 5.8797327508258396e-06, 'flow_matching_loss': 0.013373817130923272, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.31} +{'loss': 0.0136, 'grad_norm': 0.05801743268966675, 'learning_rate': 5.866611010998584e-06, 'flow_matching_loss': 0.013608177565038205, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.31} +{'loss': 0.0138, 'grad_norm': 0.06648078560829163, 'learning_rate': 5.853500369100172e-06, 'flow_matching_loss': 0.013848559930920601, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.31} +{'loss': 0.0129, 'grad_norm': 0.06789704412221909, 'learning_rate': 5.840400841061265e-06, 'flow_matching_loss': 0.01289151906967163, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.32} +{'loss': 0.0125, 'grad_norm': 0.07673487067222595, 'learning_rate': 5.827312442799034e-06, 'flow_matching_loss': 0.012534800916910172, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.32} +{'loss': 0.0119, 'grad_norm': 0.0708501935005188, 'learning_rate': 5.814235190217114e-06, 'flow_matching_loss': 0.011932060308754443, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.32} +{'loss': 0.0129, 'grad_norm': 0.09044141322374344, 'learning_rate': 5.801169099205599e-06, 'flow_matching_loss': 0.012880152091383934, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.33} +{'loss': 0.0131, 'grad_norm': 0.08360414206981659, 'learning_rate': 5.788114185641022e-06, 'flow_matching_loss': 0.01305748987942934, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.33} +{'loss': 0.0134, 'grad_norm': 0.06152678653597832, 'learning_rate': 5.775070465386328e-06, 'flow_matching_loss': 0.01337619349360466, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.33} +{'loss': 0.0142, 'grad_norm': 0.10867657512426376, 'learning_rate': 5.7620379542908765e-06, 'flow_matching_loss': 0.01417469447478652, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.33} +{'loss': 0.0139, 'grad_norm': 0.06831344962120056, 'learning_rate': 5.749016668190395e-06, 'flow_matching_loss': 0.013875481858849525, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.34} +{'loss': 0.0146, 'grad_norm': 0.0568140484392643, 'learning_rate': 5.736006622906973e-06, 'flow_matching_loss': 0.014637913275510072, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.34} +{'loss': 0.0146, 'grad_norm': 0.05983157828450203, 'learning_rate': 5.723007834249034e-06, 'flow_matching_loss': 0.014577757753431796, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.34} +{'loss': 0.014, 'grad_norm': 0.05902672931551933, 'learning_rate': 5.710020318011346e-06, 'flow_matching_loss': 0.013965468388050795, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.35} +{'loss': 0.0144, 'grad_norm': 0.07469973713159561, 'learning_rate': 5.697044089974955e-06, 'flow_matching_loss': 0.01444997675716877, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.35} +{'loss': 0.0144, 'grad_norm': 0.058961521834135056, 'learning_rate': 5.68407916590721e-06, 'flow_matching_loss': 0.014425041899085046, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.35} +{'loss': 0.0125, 'grad_norm': 0.06402431428432465, 'learning_rate': 5.671125561561713e-06, 'flow_matching_loss': 0.012464318796992302, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.35} +{'loss': 0.0147, 'grad_norm': 0.06572264432907104, 'learning_rate': 5.658183292678315e-06, 'flow_matching_loss': 0.014658653642982244, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.36} +{'loss': 0.0125, 'grad_norm': 0.054180342704057693, 'learning_rate': 5.645252374983091e-06, 'flow_matching_loss': 0.012549732904881239, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.36} +{'loss': 0.0144, 'grad_norm': 0.05498868227005005, 'learning_rate': 5.6323328241883224e-06, 'flow_matching_loss': 0.014381828717887402, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.36} +{'loss': 0.0139, 'grad_norm': 0.09776416420936584, 'learning_rate': 5.61942465599249e-06, 'flow_matching_loss': 0.013947288691997527, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.37} +{'loss': 0.014, 'grad_norm': 0.08617467433214188, 'learning_rate': 5.60652788608023e-06, 'flow_matching_loss': 0.013974087685346604, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.37} +{'loss': 0.0127, 'grad_norm': 0.06929370760917664, 'learning_rate': 5.593642530122336e-06, 'flow_matching_loss': 0.012676784163340926, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.37} +{'loss': 0.0157, 'grad_norm': 0.0818684920668602, 'learning_rate': 5.58076860377573e-06, 'flow_matching_loss': 0.015684251580387353, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.37} +{'loss': 0.0133, 'grad_norm': 0.08244957029819489, 'learning_rate': 5.567906122683439e-06, 'flow_matching_loss': 0.013344424497336149, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.38} +{'loss': 0.0148, 'grad_norm': 0.08186900615692139, 'learning_rate': 5.555055102474604e-06, 'flow_matching_loss': 0.014805641956627369, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.38} +{'loss': 0.0135, 'grad_norm': 0.06675300747156143, 'learning_rate': 5.542215558764421e-06, 'flow_matching_loss': 0.013509374391287566, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.38} +{'loss': 0.0137, 'grad_norm': 0.05708472430706024, 'learning_rate': 5.529387507154138e-06, 'flow_matching_loss': 0.013714950624853372, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.39} 73%|███████▎ | 21950/30000 [13:36:27<4:02:14, 1.81s/it] 73%|███████▎ | 21951/30000 [13:36:28<3:36:37, 1.61s/it] 73%|███████▎ | 21952/30000 [13:36:32<4:43:58, 2.12s/it] 73%|███████▎ | 21953/30000 [13:36:35<5:35:32, 2.50s/it] 73%|███████▎ | 21954/30000 [13:36:36<4:41:47, 2.10s/it] 73%|███████▎ | 21955/30000 [13:36:37<4:04:31, 1.82s/it] 73%|███████▎ | 21956/30000 [13:36:38<3:39:16, 1.64s/it] 73%|███████▎ | 21957/30000 [13:36:40<3:20:55, 1.50s/it] 73%|███████▎ | 21958/30000 [13:36:42<4:13:20, 1.89s/it] 73%|███████▎ | 21959/30000 [13:36:44<3:44:35, 1.68s/it] 73%|███████▎ | 21960/30000 [13:36:46<4:14:29, 1.90s/it] 73%|███████▎ | 21960/30000 [13:36:46<4:14:29, 1.90s/it] 73%|███████▎ | 21961/30000 [13:36:51<6:13:41, 2.79s/it] 73%|███████▎ | 21962/30000 [13:36:52<5:08:19, 2.30s/it] 73%|███████▎ | 21963/30000 [13:36:53<4:23:06, 1.96s/it] 73%|███████▎ | 21964/30000 [13:36:54<3:51:39, 1.73s/it] 73%|███████▎ | 21965/30000 [13:36:56<3:29:24, 1.56s/it] 73%|███████▎ | 21966/30000 [13:36:59<4:44:26, 2.12s/it] 73%|███████▎ | 21967/30000 [13:37:00<4:05:48, 1.84s/it] 73%|███████▎ | 21968/30000 [13:37:02<4:14:14, 1.90s/it] 73%|███████▎ | 21969/30000 [13:37:07<6:22:36, 2.86s/it] 73%|███████▎ | 21970/30000 [13:37:09<5:14:26, 2.35s/it] 73%|███████▎ | 21970/30000 [13:37:09<5:14:26, 2.35s/it] 73%|███████▎ | 21971/30000 [13:37:10<4:27:38, 2.00s/it] 73%|███████▎ | 21972/30000 [13:37:11<3:56:56, 1.77s/it] 73%|███████▎ | 21973/30000 [13:37:12<3:34:57, 1.61s/it] 73%|███████▎ | 21974/30000 [13:37:16<4:46:17, 2.14s/it] 73%|███████▎ | 21975/30000 [13:37:17<4:07:16, 1.85s/it] 73%|███████▎ | 21976/30000 [13:37:18<3:56:01, 1.76s/it] 73%|███████▎ | 21977/30000 [13:37:23<5:36:22, 2.52s/it] 73%|███████▎ | 21978/30000 [13:37:24<4:42:02, 2.11s/it] 73%|███████▎ | 21979/30000 [13:37:25<4:04:52, 1.83s/it] 73%|███████▎ | 21980/30000 [13:37:26<3:38:08, 1.63s/it] 73%|███████▎ | 21980/30000 [13:37:26<3:38:08, 1.63s/it] 73%|███████▎ | 21981/30000 [13:37:27<3:20:15, 1.50s/it] 73%|███████▎ | 21982/30000 [13:37:32<5:10:38, 2.32s/it] 73%|███████▎ | 21983/30000 [13:37:33<4:24:42, 1.98s/it] 73%|███████▎ | 21984/30000 [13:37:35<4:25:31, 1.99s/it] 73%|███████▎ | 21985/30000 [13:37:38<5:36:23, 2.52s/it] 73%|███████▎ | 21986/30000 [13:37:40<4:42:14, 2.11s/it] 73%|███████▎ | 21987/30000 [13:37:41<4:04:56, 1.83s/it] 73%|███████▎ | 21988/30000 [13:37:42<3:39:08, 1.64s/it] 73%|███████▎ | 21989/30000 [13:37:43<3:21:39, 1.51s/it] 73%|███████▎ | 21990/30000 [13:37:47<5:12:44, 2.34s/it] 73%|███████▎ | 21990/30000 [13:37:48<5:12:44, 2.34s/it] 73%|███████▎ | 21991/30000 [13:37:49<4:25:27, 1.99s/it] 73%|███████▎ | 21992/30000 [13:37:51<4:41:04, 2.11s/it] 73%|███████▎ | 21993/30000 [13:37:56<6:26:55, 2.90s/it] 73%|███████▎ | 21994/30000 [13:37:57<5:17:19, 2.38s/it] 73%|███████▎ | 21995/30000 [13:37:58<4:30:03, 2.02s/it] 73%|███████▎ | 21996/30000 [13:37:59<3:56:22, 1.77s/it] 73%|███████▎ | 21997/30000 [13:38:01<3:32:53, 1.60s/it] 73%|███████▎ | 21998/30000 [13:38:04<4:51:52, 2.19s/it] 73%|███████▎ | 21999/30000 [13:38:05<4:11:18, 1.88s/it] 73%|███████▎ | 22000/30000 [13:38:07<4:00:59, 1.81s/it] 73%|███████▎ | 22000/30000 [13:38:07<4:00:59, 1.81s/it] 73%|███████▎ | 22001/30000 [13:38:11<5:51:46, 2.64s/it] 73%|███████▎ | 22002/30000 [13:38:13<4:52:59, 2.20s/it] 73%|███████▎ | 22003/30000 [13:38:14<4:12:17, 1.89s/it] 73%|███████▎ | 22004/30000 [13:38:15<3:44:13, 1.68s/it] 73%|███████▎ | 22005/30000 [13:38:16<3:24:04, 1.53s/it] 73%|███████▎ | 22006/30000 [13:38:20<4:54:21, 2.21s/it] 73%|███████▎ | 22007/30000 [13:38:21<4:12:20, 1.89s/it] 73%|███████▎ | 22008/30000 [13:38:23<4:04:41, 1.84s/it] 73%|███████▎ | 22009/30000 [13:38:27<5:33:38, 2.51s/it] 73%|███████▎ | 22010/30000 [13:38:28<4:40:45, 2.11s/it] 73%|███████▎ | 22010/30000 [13:38:28<4:40:45, 2.11s/it] 73%|███████▎ | 22011/30000 [13:38:29<4:03:42, 1.83s/it] 73%|███████▎ | 22012/30000 [13:38:30<3:37:52, 1.64s/it] 73%|███████▎ | 22013/30000 [13:38:32<3:19:12, 1.50s/it] 73%|███████▎ | 22014/30000 [13:38:35<4:21:28, 1.96s/it] 73%|███████▎ | 22015/30000 [13:38:36<3:49:59, 1.73s/it] 73%|███████▎ | 22016/30000 [13:38:38<4:24:06, 1.98s/it] 73%|███████▎ | 22017/30000 [13:38:43<5:46:56, 2.61s/it] 73%|███████▎ | 22018/30000 [13:38:44<4:51:15, 2.19s/it] 73%|███████▎ | 22019/30000 [13:38:45<4:11:05, 1.89s/it] 73%|███████▎ | 22020/30000 [13:38:46<3:43:17, 1.68s/it] 73%|███████▎ | 22020/30000 [13:38:46<3:43:17, 1.68s/it] 73%|███████▎ | 22021/30000 [13:38:47<3:23:51, 1.53s/it] 73%|███████▎ | 22022/30000 [13:38:50<4:12:10, 1.90s/it] 73%|███████▎ | 22023/30000 [13:38:51<3:43:03, 1.68s/it] 73%|███████▎ | 22024/30000 [13:38:54<4:13:56, 1.91s/it] 73%|███████▎ | 22025/30000 [13:38:58<5:36:59, 2.54s/it] 73%|███████▎ | 22026/30000 [13:38:59<4:42:13, 2.12s/it] 73%|███████▎ | 22027/30000 [13:39:00<4:04:58, 1.84s/it] 73%|███████▎ | 22028/30000 [13:39:01<3:40:08, 1.66s/it] 73%|███████▎ | 22029/30000 [13:39:02<3:21:00, 1.51s/it] 73%|███████▎ | 22030/30000 [13:39:06<4:37:14, 2.09s/it] 73%|███████▎ | 22030/30000 [13:39:06<4:37:14, 2.09s/it] 73%|███████▎ | 22031/30000 [13:39:07<4:01:28, 1.82s/it] 73%|███████▎ | 22032/30000 [13:39:10<4:29:06, 2.03s/it] 73%|███████▎ | 22033/30000 [13:39:13<5:36:45, 2.54s/it] 73%|███████▎ | 22034/30000 [13:39:14<4:42:04, 2.12s/it] 73%|███████▎ | 22035/30000 [13:39:16<4:04:53, 1.84s/it] 73%|███████▎ | 22036/30000 [13:39:17<3:39:15, 1.65s/it] 73%|███████▎ | 22037/30000 [13:39:18<3:20:09, 1.51s/it] 73%|███████▎ | 22038/30000 [13:39:21<4:22:52, 1.98s/it] 73%|███████▎ | 22039/30000 [13:39:22<3:50:05, 1.73s/it] 73%|███████▎ | 22040/30000 [13:39:25<4:27:52, 2.02s/it] 73%|███████▎ | 22040/30000 [13:39:25<4:27:52, 2.02s/it] 73%|███████▎ | 22041/30000 [13:39:29<5:36:22, 2.54s/it] 73%|███████▎ | 22042/30000 [13:39:30<4:41:33, 2.12s/it] 73%|███████▎ | 22043/30000 [13:39:31<4:03:52, 1.84s/it] 73%|███████▎ | 22044/30000 [13:39:32<3:37:06, 1.64s/it] 73%|███████▎ | 22045/30000 [13:39:33<3:19:21, 1.50s/it] 73%|███████▎ | 22046/30000 [13:39:36<4:06:22, 1.86s/it] 73%|███████▎ | 22047/30000 [13:39:37<3:38:59, 1.65s/it] 73%|███████▎ | 22048/30000 [13:39:39<4:02:20, 1.83s/it] 73%|███████▎ | 22049/30000 [13:39:43<5:12:57, 2.36s/it] 74%|███████▎ | 22050/30000 [13:39:44<4:25:16, 2.00s/it] 74%|███████▎ | 22050/30000 [13:39:44<4:25:16, 2.00s/it] 74%|███████▎ | 22051/30000 [13:39:45<3:52:51, 1.76s/it] 74%|███████▎ | 22052/30000 [13:39:47<3:29:21, 1.58s/it] 74%|███████▎ | 22053/30000 [13:39:48<3:13:04, 1.46s/it] 74%|███████▎ | 22054/30000 [13:39:51<4:17:19, 1.94s/it] 74%|███████▎ | 22055/30000 [13:39:52<3:46:39, 1.71s/it] 74%|███████▎ | 22056/30000 [13:39:55<4:24:22, 2.00s/it] 74%|███████▎ | 22057/30000 [13:39:58<5:25:46, 2.46s/it] 74%|███████▎ | 22058/30000 [13:39:59<4:34:19, 2.07s/it] 74%|███████▎ | 22059/30000 [13:40:01<3:58:11, 1.80s/it] 74%|███████▎ | 22060/30000 [13:40:02<3:33:25, 1.61s/it] 74%|███████▎ | 22060/30000 [13:40:02<3:33:25, 1.61s/it] 74%|███████▎ | 22061/30000 [13:40:03<3:16:29, 1.48s/it] 74%|███████▎ | 22062/30000 [13:40:05<3:54:42, 1.77s/it] 74%|███████▎ | 22063/30000 [13:40:06<3:30:30, 1.59s/it] 74%|███████▎ | 22064/30000 [13:40:10<4:41:58, 2.13s/it] 74%|███████▎ | 22065/30000 [13:40:13<5:25:03, 2.46s/it] 74%|███████▎ | 22066/30000 [13:40:14<4:33:38, 2.07s/it] 74%|███████▎ | 22067/30000 [13:40:15<3:58:49, 1.81s/it] 74%|███████▎ | 22068/30000 [13:40:17<3:32:59, 1.61s/it] 74%|███████▎ | 22069/30000 [13:40:18<3:16:17, 1.49s/it] 74%|███████▎ | 22070/30000 [13:40:20<4:02:45, 1.84s/it] 74%|███████▎ | 22070/30000 [13:40:20<4:02:45, 1.84s/it] 74%|███████▎ | 22071/30000 [13:40:22<3:36:27, 1.64s/it] 74%|███████▎ | 22072/30000 [13:40:25<4:45:49, 2.16s/it] 74%|███████▎ | 22073/30000 [13:40:28<5:06:38, 2.32s/it] 74%|███████▎ | 22074/30000 [13:40:29<4:21:03, 1.98s/it] 74%|███████▎ | 22075/30000 [13:40:30<3:50:11, 1.74s/it] 74%|███████▎ | 22076/30000 [13:40:31<3:28:52, 1.58s/it] 74%|███████▎ | 22077/30000 [13:40:32<3:12:44, 1.46s/it] 74%|███████▎ | 22078/30000 [13:40:35<3:56:00, 1.79s/it] 74%|███████▎ | 22079/30000 [13:40:36<3:31:23, 1.60s/it] 74%|███████▎ | 22080/30000 [13:40:39<4:08:34, 1.88s/it] 74%|███████▎ | 22080/30000 [13:40:39<4:08:34, 1.88s/it] 74%|███████▎ | 22081/30000 [13:40:42<5:11:20, 2.36s/it] 74%|███████▎ | 22082/30000 [13:40:43<4:24:29, 2.00s/it] 74%|███████▎ | 22083/30000 [13:40:45<3:51:00, 1.75s/it] 74%|███████▎ | 22084/30000 [13:40:46<3:28:13, 1.58s/it] 74%|███████▎ | 22085/30000 [13:40:47<3:12:13, 1.46s/it] 74%|███████▎ | 22086/30000 [13:40:50<4:04:37, 1.85s/it] 74%|███████▎ | 22087/30000 [13:40:51<3:36:58, 1.65s/it] 74%|███████▎ | 22088/30000 [13:40:54<4:22:49, 1.99s/it] 74%|███████▎ | 22089/30000 [13:40:57<5:09:53, 2.35s/it] 74%|███████▎ | 22090/30000 [13:40:58<4:23:42, 2.00s/it] 74%|███████▎ | 22090/30000 [13:40:58<4:23:42, 2.00s/it] 74%|███████▎ | 22091/30000 [13:40:59<3:50:58, 1.75s/it] 74%|███████▎ | 22092/30000 [13:41:00<3:28:48, 1.58s/it] 74%|███████▎ | 22093/30000 [13:41:02<3:13:01, 1.46s/it] 74%|███████▎ | 22094/30000 [13:41:05<4:17:58, 1.96s/it] 74%|███████▎ | 22095/30000 [13:41:06<3:46:31, 1.72s/it] 74%|███████▎ | 22096/30000 [13:41:08<4:06:50, 1.87s/it] 74%|███████▎ | 22097/30000 [13:41:11<4:49:15, 2.20s/it] 74%|███████▎ | 22098/30000 [13:41:12<4:08:40, 1.89s/it] 74%|███████▎ | 22099/30000 [13:41:13<3:41:02, 1.68s/it] 74%|███████▎ | 22100/30000 [13:41:15<3:21:39, 1.53s/it] 74%|███████▎ | 22100/30000 [13:41:15<3:21:39, 1.53s/it] 74%|███████▎ | 22101/30000 [13:41:16<3:08:40, 1.43s/it] 74%|███████▎ | 22102/30000 [13:41:19<4:23:52, 2.00s/it] 74%|███████▎ | 22103/30000 [13:41:20<3:50:32, 1.75s/it] 74%|███████▎ | 22104/30000 [13:41:22<3:52:48, 1.77s/it] 74%|███████▎ | 22105/30000 [13:41:26<5:13:53, 2.39s/it] 74%|███████▎ | 22106/30000 [13:41:27<4:25:30, 2.02s/it] 74%|███████▎ | 22107/30000 [13:41:28<3:52:53, 1.77s/it] 74%|███████▎ | 22108/30000 [13:41:29<3:29:32, 1.59s/it] 74%|███████▎ | 22109/30000 [13:41:31<3:13:02, 1.47s/it] 74%|███████▎ | 22110/30000 [13:41:34<4:15:19, 1.94s/it] 74%|███████▎ | 22110/30000 [13:41:34<4:15:19, 1.94s/it] 74%|███████▎ | 22111/30000 [13:41:35<3:44:08, 1.70s/it] 74%|███████▎ | 22112/30000 [13:41:36<3:41:05, 1.68s/it] 74%|███████▎ | 22113/30000 [13:41:40<5:13:41, 2.39s/it] 74%|███████▎ | 22114/30000 [13:41:42<4:25:04, 2.02s/it] 74%|███████▎ | 22115/30000 [13:41:43<3:52:34, 1.77s/it] 74%|███████▎ | 22116/30000 [13:41:44<3:29:25, 1.59s/it] 74%|███████▎ | 22117/30000 [13:41:45<3:13:34, 1.47s/it] 74%|███████▎ | 22118/30000 [13:41:48<4:08:56, 1.89s/it] 74%|███████▎ | 22119/30000 [13:41:49<3:39:54, 1.67s/it] 74%|███████▎ | 22120/30000 [13:41:50<3:22:38, 1.54s/it] 74%|███████▎ | 22120/30000 [13:41:50<3:22:38, 1.54s/it] 74%|███████▎ | 22121/30000 [13:41:54<4:51:49, 2.22s/it] 74%|███████▎ | 22122/30000 [13:41:55<4:10:28, 1.91s/it] 74%|███████▎ | 22123/30000 [13:41:57<3:42:01, 1.69s/it] 74%|███████▎ | 22124/30000 [13:41:58<3:21:11, 1.53s/it] 74%|███████▍ | 22125/30000 [13:41:59<3:07:28, 1.43s/it] 74%|███████▍ | 22126/30000 [13:42:02<4:29:37, 2.05s/it] 74%|███████▍ | 22127/30000 [13:42:04<3:55:09, 1.79s/it] 74%|███████▍ | 22128/30000 [13:42:05<3:31:15, 1.61s/it] 74%|███████▍ | 22129/30000 [13:42:09<5:17:19, 2.42s/it] 74%|███████▍ | 22130/30000 [13:42:10<4:27:41, 2.04s/it] 74%|███████▍ | 22130/30000 [13:42:10<4:27:41, 2.04s/it] 74%|███████▍ | 22131/30000 [13:42:12<3:54:27, 1.79s/it] 74%|███████▍ | 22132/30000 [13:42:13<3:30:19, 1.60s/it] 74%|███████▍ | 22133/30000 [13:42:14<3:13:18, 1.47s/it] 74%|███████▍ | 22134/30000 [13:42:17<4:28:33, 2.05s/it] 74%|███████▍ | 22135/30000 [13:42:18<3:54:06, 1.79s/it] 74%|███████▍ | 22136/30000 [13:42:20<3:30:19, 1.60s/it] 74%|███████▍ | 22137/30000 [13:42:23<4:39:30, 2.13s/it] 74%|███████▍ | 22138/30000 [13:42:24<4:02:13, 1.85s/it] 74%|███████▍ | 22139/30000 [13:42:25<3:35:54, 1.65s/it] 74%|███████▍ | 22140/30000 [13:42:27<3:18:01, 1.51s/it] 74%|███████▍ | 22140/30000 [13:42:27<3:18:01, 1.51s/it] 74%|███████▍ | 22141/30000 [13:42:28<3:05:33, 1.42s/it] 74%|███████▍ | 22142/30000 [13:42:31<4:21:14, 1.99s/it] 74%|███████▍ | 22143/30000 [13:42:32<3:48:45, 1.75s/it] 74%|███████▍ | 22144/30000 [13:42:33<3:27:24, 1.58s/it] 74%|███████▍ | 22145/30000 [13:42:38<5:35:34, 2.56s/it] 74%|███████▍ | 22146/30000 [13:42:40<4:42:45, 2.16s/it] 74%|███████▍ | 22147/30000 [13:42:41<4:04:57, 1.87s/it] 74%|███████▍ | 22148/30000 [13:42:42<3:37:46, 1.66s/it] 74%|███████▍ | 22149/30000 [13:42:43<3:18:47, 1.52s/it] 74%|███████▍ | 22150/30000 [13:42:46<4:14:17, 1.94s/it] 74%|███████▍ | 22150/30000 [13:42:46<4:14:17, 1.94s/it] 74%|███████▍ | 22151/30000 [13:42:47<3:43:42, 1.71s/it] 74%|███████▍ | 22152/30000 [13:42:48<3:22:51, 1.55s/it] 74%|███████▍ | 22153/30000 [13:42:53<5:15:01, 2.41s/it] 74%|███████▍ | 22154/30000 [13:42:54<4:26:46, 2.04s/it] 74%|███████▍ | 22155/30000 [13:42:55<3:53:01, 1.78s/it] 74%|███████▍ | 22156/30000 [13:42:56<3:29:16, 1.60s/it] 74%|███████▍ | 22157/30000 [13:42:57<3:12:28, 1.47s/it] 74%|███████▍ | 22158/30000 [13:43:01<4:21:13, 2.00s/it] 74%|███████▍ | 22159/30000 [13:43:02<3:49:05, 1.75s/it] 74%|███████▍ | 22160/30000 [13:43:03<3:26:39, 1.58s/it] 74%|███████▍ | 22160/30000 [13:43:03<3:26:39, 1.58s/it] 74%|███████▍ | 22161/30000 [13:43:07<5:07:31, 2.35s/it] 74%|███████▍ | 22162/30000 [13:43:08<4:21:41, 2.00s/it] 74%|███████▍ | 22163/30000 [13:43:10<3:49:17, 1.76s/it] 74%|███████▍ | 22164/30000 [13:43:11<3:26:32, 1.58s/it] 74%|███████▍ | 22165/30000 [13:43:12<3:10:58, 1.46s/it] 74%|███████▍ | 22166/30000 [13:43:15<4:02:05, 1.85s/it] 74%|███████▍ | 22167/30000 [13:43:16<3:36:27, 1.66s/it] 74%|███████▍ | 22168/30000 [13:43:17<3:18:15, 1.52s/it] 74%|███████▍ | 22169/30000 [13:43:22<5:14:53, 2.41s/it] 74%|███████▍ | 22170/30000 [13:43:23<4:26:27, 2.04s/it] 74%|███████▍ | 22170/30000 [13:43:23<4:26:27, 2.04s/it] 74%|███████▍ | 22171/30000 [13:43:24<3:52:43, 1.78s/it] 74%|███████▍ | 22172/30000 [13:43:25<3:28:12, 1.60s/it] 74%|███████▍ | 22173/30000 [13:43:26<3:12:01, 1.47s/it] 74%|███████▍ | 22174/30000 [13:43:29<3:43:40, 1.71s/it] 74%|███████▍ | 22175/30000 [13:43:30<3:22:19, 1.55s/it] 74%|███████▍ | 22176/30000 [13:43:31<3:08:36, 1.45s/it] 74%|███████▍ | 22177/30000 [13:43:36<5:30:14, 2.53s/it] 74%|███████▍ | 22178/30000 [13:43:37<4:36:23, 2.12s/it] 74%|███████▍ | 22179/30000 [13:43:38<3:59:52, 1.84s/it] 74%|███████▍ | 22180/30000 [13:43:40<3:33:48, 1.64s/it] 74%|███████▍ | 22180/30000 [13:43:40<3:33:48, 1.64s/it] 74%|███████▍ | 22181/30000 [13:43:41<3:16:01, 1.50s/it] 74%|███████▍ | 22182/30000 [13:43:43<3:42:34, 1.71s/it] 74%|███████▍ | 22183/30000 [13:43:44<3:20:54, 1.54s/it] 74%|███████▍ | 22184/30000 [13:43:46<3:21:31, 1.55s/it] 74%|███████▍ | 22185/30000 [13:43:51<5:53:47, 2.72s/it] 74%|███████▍ | 22186/30000 [13:43:52<4:52:22, 2.24s/it] 74%|███████▍ | 22187/30000 [13:43:53<4:11:27, 1.93s/it] 74%|███████▍ | 22188/30000 [13:43:55<3:42:47, 1.71s/it] 74%|███████▍ | 22189/30000 [13:43:56<3:21:59, 1.55s/it] 74%|███████▍ | 22190/30000 [13:43:58<3:30:36, 1.62s/it] 74%|███████▍ | 22190/30000 [13:43:58<3:30:36, 1.62s/it] 74%|███████▍ | 22191/30000 [13:43:59<3:13:19, 1.49s/it] 74%|███████▍ | 22192/30000 [13:44:02<4:04:41, 1.88s/it] 74%|███████▍ | 22193/30000 [13:44:06<5:44:18, 2.65s/it] 74%|███████▍ | 22194/30000 [13:44:07<4:46:06, 2.20s/it] 74%|███████▍ | 22195/30000 [13:44:08<4:06:19, 1.89s/it] 74%|███████▍ | 22196/30000 [13:44:09<3:38:34, 1.68s/it] 74%|███████▍ | 22197/30000 [13:44:11<3:19:27, 1.53s/it] 74%|███████▍ | 22198/30000 [13:44:13<3:35:36, 1.66s/it] 74%|███████▍ | 22199/30000 [13:44:14<3:16:38, 1.51s/it] 74%|███████▍ | 22200/30000 [13:44:16<3:31:06, 1.62s/it] 74%|███████▍ | 22200/30000 [13:44:16<3:31:06, 1.62s/it] 74%|███████▍ | 22201/30000 [13:44:20<5:09:19, 2.38s/it] 74%|███████▍ | 22202/30000 [13:44:21<4:22:32, 2.02s/it] 74%|███████▍ | 22203/30000 [13:44:22<3:49:44, 1.77s/it] 74%|███████▍ | 22204/30000 [13:44:23<3:26:56, 1.59s/it] 74%|███████▍ | 22205/30000 [13:44:25<3:12:03, 1.48s/it] 74%|███████▍ | 22206/30000 [13:44:26<3:26:40, 1.59s/it] 74%|███████▍ | 22207/30000 [13:44:28<3:10:26, 1.47s/it] 74%|███████▍ | 22208/30000 [13:44:30<3:30:54, 1.62s/it] 74%|███████▍ | 22209/30000 [13:44:34<5:10:02, 2.39s/it] 74%|███████▍ | 22210/30000 [13:44:35<4:22:15, 2.02s/it] 74%|███████▍ | 22210/30000 [13:44:35<4:22:15, 2.02s/it] 74%|███████▍ | 22211/30000 [13:44:36<3:48:56, 1.76s/it] 74%|███████▍ | 22212/30000 [13:44:37<3:26:32, 1.59s/it] 74%|███████▍ | 22213/30000 [13:44:38<3:10:30, 1.47s/it] 74%|███████▍ | 22214/30000 [13:44:40<3:23:55, 1.57s/it] 74%|███████▍ | 22215/30000 [13:44:41<3:08:53, 1.46s/it] 74%|███████▍ | 22216/30000 [13:44:44<3:52:41, 1.79s/it] 74%|███████▍ | 22217/30000 [13:44:48<5:20:36, 2.47s/it] 74%|███████▍ | 22218/30000 [13:44:49<4:30:31, 2.09s/it] 74%|███████▍ | 22219/30000 [13:44:50<3:56:01, 1.82s/it] 74%|███████▍ | 22220/30000 [13:44:52<3:30:25, 1.62s/it] 74%|███████▍ | 22220/30000 [13:44:52<3:30:25, 1.62s/it] 74%|███████▍ | 22221/30000 [13:44:53<3:13:29, 1.49s/it] 74%|███████▍ | 22222/30000 [13:44:55<3:42:41, 1.72s/it] 74%|███████▍ | 22223/30000 [13:44:56<3:22:11, 1.56s/it] 74%|███████▍ | 22224/30000 [13:44:59<3:57:53, 1.84s/it] 74%|███████▍ | 22225/30000 [13:45:03<5:35:00, 2.59s/it] 74%|███████▍ | 22226/30000 [13:45:04<4:40:54, 2.17s/it] 74%|███████▍ | 22227/30000 [13:45:05<4:02:45, 1.87s/it] 74%|███████▍ | 22228/30000 [13:45:07<3:35:35, 1.66s/it] 74%|███████▍ | 22229/30000 [13:45:08<3:16:55, 1.52s/it] 74%|███████▍ | 22230/30000 [13:45:10<3:54:35, 1.81s/it] 74%|███████▍ | 22230/30000 [13:45:10<3:54:35, 1.81s/it] 74%|███████▍ | 22231/30000 [13:45:12<3:30:38, 1.63s/it] 74%|███████▍ | 22232/30000 [13:45:13<3:37:50, 1.68s/it] 74%|███████▍ | 22233/30000 [13:45:18<5:43:34, 2.65s/it] 74%|███████▍ | 22234/30000 [13:45:19<4:45:45, 2.21s/it] 74%|███████▍ | 22235/30000 [13:45:21<4:05:46, 1.90s/it] 74%|███████▍ | 22236/30000 [13:45:22<3:37:48, 1.68s/it] 74%|███████▍ | 22237/30000 [13:45:23<3:17:50, 1.53s/it] 74%|███████▍ | 22238/30000 [13:45:25<3:25:11, 1.59s/it] 74%|███████▍ | 22239/30000 [13:45:26<3:08:46, 1.46s/it] 74%|███████▍ | 22240/30000 [13:45:28<3:43:49, 1.73s/it] 74%|███████▍ | 22240/30000 [13:45:28<3:43:49, 1.73s/it] 74%|███████▍ | 22241/30000 [13:45:34<6:07:54, 2.85s/it] 74%|███████▍ | 22242/30000 [13:45:35<5:02:58, 2.34s/it] 74%|███████▍ | 22243/30000 [13:45:36<4:17:54, 1.99s/it] 74%|███████▍ | 22244/30000 [13:45:37<3:45:52, 1.75s/it] 74%|███████▍ | 22245/30000 [13:45:38<3:24:45, 1.58s/it] 74%|███████▍ | 22246/30000 [13:45:40<3:09:13, 1.46s/it] 74%|███████▍ | 22247/30000 [13:45:41<2:57:17, 1.37s/it] 74%|███████▍ | 22248/30000 [13:45:43<3:20:21, 1.55s/it] 74%|███████▍ | 22249/30000 [13:45:48<6:04:09, 2.82s/it] 74%|███████▍ | 22250/30000 [13:45:50<4:59:46, 2.32s/it] 74%|███████▍ | 22250/30000 [13:45:50<4:59:46, 2.32s/it] 74%|███████▍ | 22251/30000 [13:45:51<4:15:16, 1.98s/it] 74%|███████▍ | 22252/30000 [13:45:52<3:44:25, 1.74s/it] 74%|███████▍ | 22253/30000 [13:45:53<3:24:00, 1.58s/it] 74%|███████▍ | 22254/30000 [13:45:54<3:09:38, 1.47s/it] 74%|███████▍ | 22255/30000 [13:45:56<2:58:27, 1.38s/it] 74%|███████▍ | 22256/30000 [13:45:57<3:10:34, 1.48s/it] 74%|███████▍ | 22257/30000 [13:46:03<6:00:48, 2.80s/it] 74%|███████▍ | 22258/30000 [13:46:04<4:59:22, 2.32s/it] 74%|███████▍ | 22259/30000 [13:46:06<4:16:34, 1.99s/it] 74%|███████▍ | 22260/30000 [13:46:07<3:45:35, 1.75s/it] 74%|███████▍ | 22260/30000 [13:46:07<3:45:35, 1.75s/it] 74%|███████▍ | 22261/30000 [13:46:08<3:23:48, 1.58s/it] 74%|███████▍ | 22262/30000 [13:46:09<3:08:15, 1.46s/it] 74%|███████▍ | 22263/30000 [13:46:10<2:57:44, 1.38s/it] 74%|███████▍ | 22264/30000 [13:46:12<2:56:24, 1.37s/it] 74%|███████▍ | 22265/30000 [13:46:18<5:56:20, 2.76s/it] 74%|███████▍ | 22266/30000 [13:46:19<4:55:36, 2.29s/it] 74%|███████▍ | 22267/30000 [13:46:20<4:14:44, 1.98s/it] 74%|███████▍ | 22268/30000 [13:46:21<3:44:26, 1.74s/it] 74%|███████▍ | 22269/30000 [13:46:22<3:22:49, 1.57s/it] 74%|███████▍ | 22270/30000 [13:46:24<3:08:02, 1.46s/it] 74%|███████▍ | 22270/30000 [13:46:24<3:08:02, 1.46s/it] 74%|███████▍ | 22271/30000 [13:46:25<2:56:21, 1.37s/it] 74%|███████▍ | 22272/30000 [13:46:26<2:52:04, 1.34s/it] 74%|███████▍ | 22273/30000 [13:46:32<6:00:50, 2.80s/it] 74%|███████▍ | 22274/30000 [13:46:33<4:58:29, 2.32s/it] 74%|███████▍ | 22275/30000 [13:46:35<4:15:49, 1.99s/it] 74%|███████▍ | 22276/30000 [13:46:36<3:45:35, 1.75s/it] 74%|███████▍ | 22277/30000 [13:46:37<3:24:36, 1.59s/it] 74%|███████▍ | 22278/30000 [13:46:38<3:08:41, 1.47s/it] 74%|███████▍ | 22279/30000 [13:46:39<2:57:17, 1.38s/it] 74%|███████▍ | 22280/30000 [13:46:41<2:50:58, 1.33s/it] 74%|███████▍ | 22280/30000 [13:46:41<2:50:58, 1.33s/it] 74%|███████▍ | 22281/30000 [13:46:47<5:47:37, 2.70s/it] 74%|███████▍ | 22282/30000 [13:46:48<4:49:18, 2.25s/it] 74%|███████▍ | 22283/30000 [13:46:49<4:08:19, 1.93s/it] 74%|███████▍ | 22284/30000 [13:46:50<3:39:58, 1.71s/it] 74%|███████▍ | 22285/30000 [13:46:51<3:19:14, 1.55s/it] 74%|███████▍ | 22286/30000 [13:46:53<3:04:52, 1.44s/it] 74%|███████▍ | 22287/30000 [13:46:54<2:55:04, 1.36s/it] 74%|███████▍ | 22288/30000 [13:46:55<2:49:42, 1.32s/it] 74%|███████▍ | 22289/30000 [13:47:01<6:00:43, 2.81s/it] 74%|███████▍ | 22290/30000 [13:47:02<4:57:44, 2.32s/it] 74%|███████▍ | 22290/30000 [13:47:02<4:57:44, 2.32s/it] 74%|███████▍ | 22291/30000 [13:47:04<4:14:23, 1.98s/it] 74%|███████▍ | 22292/30000 [13:47:05<3:44:17, 1.75s/it] 74%|███████▍ | 22293/30000 [13:47:06<3:22:37, 1.58s/it] 74%|███████▍ | 22294/30000 [13:47:07<3:07:37, 1.46s/it] 74%|███████▍ | 22295/30000 [13:47:08<2:56:28, 1.37s/it] 74%|███████▍ | 22296/30000 [13:47:10<2:49:57, 1.32s/it] 74%|███████▍ | 22297/30000 [13:47:16<5:51:34, 2.74s/it] 74%|███████▍ | 22298/30000 [13:47:17<4:51:01, 2.27s/it] 74%|███████▍ | 22299/30000 [13:47:18<4:09:25, 1.94s/it] 74%|███████▍ | 22300/30000 [13:47:19<3:39:12, 1.71s/it] 74%|███████▍ | 22300/30000 [13:47:19<3:39:12, 1.71s/it] 74%|███████▍ | 22301/30000 [13:47:20<3:18:25, 1.55s/it] 74%|███████▍ | 22302/30000 [13:47:21<3:04:15, 1.44s/it] 74%|███████▍ | 22303/30000 [13:47:23<2:54:07, 1.36s/it] 74%|███████▍ | 22304/30000 [13:47:24<2:48:14, 1.31s/it] 74%|███████▍ | 22305/30000 [13:47:29<5:25:21, 2.54s/it] 74%|███████▍ | 22306/30000 [13:47:30<4:33:31, 2.13s/it] 74%|███████▍ | 22307/30000 [13:47:32<3:58:00, 1.86s/it] 74%|███████▍ | 22308/30000 [13:47:33<3:32:26, 1.66s/it] 74%|███████▍ | 22309/30000 [13:47:34<3:14:45, 1.52s/it] 74%|███████▍ | 22310/30000 [13:47:35<3:01:50, 1.42s/it] 74%|███████▍ | 22310/30000 [13:47:35<3:01:50, 1.42s/it] 74%|███████▍ | 22311/30000 [13:47:36<2:52:32, 1.35s/it] 74%|███████▍ | 22312/30000 [13:47:38<2:46:27, 1.30s/it] 74%|███████▍ | 22313/30000 [13:47:43<5:37:47, 2.64s/it] 74%|███████▍ | 22314/30000 [13:47:44<4:41:18, 2.20s/it] 74%|███████▍ | 22315/30000 [13:47:46<4:02:10, 1.89s/it] 74%|███████▍ | 22316/30000 [13:47:47<3:35:11, 1.68s/it] 74%|███████▍ | 22317/30000 [13:47:48<3:16:14, 1.53s/it] 74%|███████▍ | 22318/30000 [13:47:49<3:02:55, 1.43s/it] 74%|███████▍ | 22319/30000 [13:47:50<2:54:29, 1.36s/it] 74%|███████▍ | 22320/30000 [13:47:52<2:48:41, 1.32s/it] 74%|███████▍ | 22320/30000 [13:47:52<2:48:41, 1.32s/it] 74%|███████▍ | 22321/30000 [13:47:58<6:09:19, 2.89s/it] 74%|███████▍ | 22322/30000 [13:47:59<5:03:35, 2.37s/it] 74%|███████▍ | 22323/30000 [13:48:01<4:17:24, 2.01s/it] 74%|███████▍ | 22324/30000 [13:48:02<3:45:30, 1.76s/it] 74%|███████▍ | 22325/30000 [13:48:03<3:23:51, 1.59s/it] 74%|███████▍ | 22326/30000 [13:48:04<3:07:58, 1.47s/it] 74%|███████▍ | 22327/30000 [13:48:05<2:57:00, 1.38s/it] 74%|███████▍ | 22328/30000 [13:48:06<2:50:15, 1.33s/it] 74%|███████▍ | 22329/30000 [13:48:13<5:53:23, 2.76s/it] 74%|███████▍ | 22330/30000 [13:48:14<4:52:20, 2.29s/it] 74%|███████▍ | 22330/30000 [13:48:14<4:52:20, 2.29s/it] 74%|███████▍ | 22331/30000 [13:48:15<4:10:00, 1.96s/it] 74%|███████▍ | 22332/30000 [13:48:16<3:40:19, 1.72s/it] 74%|███████▍ | 22333/30000 [13:48:17<3:19:51, 1.56s/it] 74%|███████▍ | 22334/30000 [13:48:18<3:04:47, 1.45s/it] 74%|███████▍ | 22335/30000 [13:48:20<2:54:26, 1.37s/it] 74%|███████▍ | 22336/30000 [13:48:21<2:47:45, 1.31s/it] 74%|███████▍ | 22337/30000 [13:48:27<5:48:27, 2.73s/it] 74%|███████▍ | 22338/30000 [13:48:28<4:48:54, 2.26s/it] 74%|███████▍ | 22339/30000 [13:48:29<4:08:02, 1.94s/it] 74%|███████▍ | 22340/30000 [13:48:30<3:38:30, 1.71s/it] 74%|███████▍ | 22340/30000 [13:48:30<3:38:30, 1.71s/it] 74%|███████▍ | 22341/30000 [13:48:32<3:18:02, 1.55s/it] 74%|███████▍ | 22342/30000 [13:48:33<3:04:11, 1.44s/it] 74%|███████▍ | 22343/30000 [13:48:34<2:53:47, 1.36s/it] 74%|███████▍ | 22344/30000 [13:48:35<2:47:58, 1.32s/it] 74%|███████▍ | 22345/30000 [13:48:41<5:58:47, 2.81s/it] 74%|███████▍ | 22346/30000 [13:48:43<4:55:39, 2.32s/it] 74%|███████▍ | 22347/30000 [13:48:44<4:12:55, 1.98s/it] 74%|███████▍ | 22348/30000 [13:48:45<3:42:18, 1.74s/it] 74%|███████▍ | 22349/30000 [13:48:46<3:21:27, 1.58s/it] 74%|███████▍ | 22350/30000 [13:48:47<3:06:42, 1.46s/it] 74%|███████▍ | 22350/30000 [13:48:47<3:06:42, 1.46s/it] 75%|███████▍ | 22351/30000 [13:48:49<2:55:51, 1.38s/it] 75%|███████▍ | 22352/30000 [13:48:50<2:49:21, 1.33s/it] 75%|███████▍ | 22353/30000 [13:48:56<5:58:44, 2.81s/it] 75%|███████▍ | 22354/30000 [13:48:57<4:56:24, 2.33s/it] 75%|███████▍ | 22355/30000 [13:48:58<4:13:04, 1.99s/it] 75%|███████▍ | 22356/30000 [13:49:00<3:44:01, 1.76s/it] 75%|███████▍ | 22357/30000 [13:49:01<3:21:50, 1.58s/it] 75%|███████▍ | 22358/30000 [13:49:02<3:07:24, 1.47s/it] 75%|███████▍ | 22359/30000 [13:49:03<2:56:32, 1.39s/it] 75%|███████▍ | 22360/30000 [13:49:04<2:49:18, 1.33s/it] 75%|███████▍ | 22360/30000 [13:49:04<2:49:18, 1.33s/it] 75%|███████▍ | 22361/30000 [13:49:10<5:38:29, 2.66s/it] 75%|███████▍ | 22362/30000 [13:49:11<4:41:51, 2.21s/it] 75%|███████▍ | 22363/30000 [13:49:13<4:02:36, 1.91s/it] 75%|███████▍ | 22364/30000 [13:49:14<3:35:47, 1.70s/it] 75%|███████▍ | 22365/30000 [13:49:15<3:16:11, 1.54s/it] 75%|███████▍ | 22366/30000 [13:49:16<3:03:15, 1.44s/it] 75%|███████▍ | 22367/30000 [13:49:17<2:52:53, 1.36s/it] 75%|███████▍ | 22368/30000 [13:49:19<2:47:29, 1.32s/it] 75%|███████▍ | 22369/30000 [13:49:24<5:34:55, 2.63s/it] 75%|███████▍ | 22370/30000 [13:49:25<4:39:45, 2.20s/it] +{'loss': 0.0148, 'grad_norm': 0.0721307247877121, 'learning_rate': 5.516570963231058e-06, 'flow_matching_loss': 0.014757471345365047, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.39} +{'loss': 0.0133, 'grad_norm': 0.07982407510280609, 'learning_rate': 5.503765942568484e-06, 'flow_matching_loss': 0.0133464558981359, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.39} +{'loss': 0.0133, 'grad_norm': 0.07761149108409882, 'learning_rate': 5.49097246072573e-06, 'flow_matching_loss': 0.013324925862252712, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.4} +{'loss': 0.0136, 'grad_norm': 0.08745390176773071, 'learning_rate': 5.4781905332480805e-06, 'flow_matching_loss': 0.013628732413053513, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.4} +{'loss': 0.0134, 'grad_norm': 0.0676802396774292, 'learning_rate': 5.4654201756667795e-06, 'flow_matching_loss': 0.013379075285047292, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.4} +{'loss': 0.0124, 'grad_norm': 0.06918632984161377, 'learning_rate': 5.452661403499017e-06, 'flow_matching_loss': 0.012372338585555554, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.4} +{'loss': 0.0129, 'grad_norm': 0.08009369671344757, 'learning_rate': 5.4399142322479005e-06, 'flow_matching_loss': 0.012910364009439946, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.41} +{'loss': 0.0137, 'grad_norm': 0.0697103887796402, 'learning_rate': 5.427178677402452e-06, 'flow_matching_loss': 0.013682690355926752, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.41} +{'loss': 0.0135, 'grad_norm': 0.09547895938158035, 'learning_rate': 5.41445475443757e-06, 'flow_matching_loss': 0.013462392427027226, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.41} +{'loss': 0.0135, 'grad_norm': 0.07074116915464401, 'learning_rate': 5.4017424788140194e-06, 'flow_matching_loss': 0.013490174245089293, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.42} +{'loss': 0.0144, 'grad_norm': 0.05957217514514923, 'learning_rate': 5.389041865978414e-06, 'flow_matching_loss': 0.014390055648982525, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.42} +{'loss': 0.0138, 'grad_norm': 0.06613312661647797, 'learning_rate': 5.37635293136319e-06, 'flow_matching_loss': 0.013776670210063457, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.42} +{'loss': 0.0148, 'grad_norm': 0.10950074344873428, 'learning_rate': 5.363675690386609e-06, 'flow_matching_loss': 0.014812473487108947, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.42} +{'loss': 0.0144, 'grad_norm': 0.059068094938993454, 'learning_rate': 5.351010158452708e-06, 'flow_matching_loss': 0.014358712639659644, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.43} +{'loss': 0.0134, 'grad_norm': 0.08689611405134201, 'learning_rate': 5.338356350951308e-06, 'flow_matching_loss': 0.013437368627637624, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.43} +{'loss': 0.0127, 'grad_norm': 0.08115945756435394, 'learning_rate': 5.325714283257974e-06, 'flow_matching_loss': 0.012738493084907532, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.43} +{'loss': 0.0145, 'grad_norm': 0.06740902364253998, 'learning_rate': 5.313083970734013e-06, 'flow_matching_loss': 0.014465564768761397, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.44} +{'loss': 0.0164, 'grad_norm': 0.08645253628492355, 'learning_rate': 5.300465428726439e-06, 'flow_matching_loss': 0.0163935667835176, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.44} +{'loss': 0.0136, 'grad_norm': 0.06534223258495331, 'learning_rate': 5.287858672567979e-06, 'flow_matching_loss': 0.013616079930216074, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.44} +{'loss': 0.0138, 'grad_norm': 0.0572090670466423, 'learning_rate': 5.275263717577028e-06, 'flow_matching_loss': 0.013818234950304032, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.44} +{'loss': 0.0125, 'grad_norm': 0.07186184823513031, 'learning_rate': 5.2626805790576435e-06, 'flow_matching_loss': 0.012545681186020374, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.45} +{'loss': 0.0138, 'grad_norm': 0.06590911746025085, 'learning_rate': 5.250109272299523e-06, 'flow_matching_loss': 0.013784188590943814, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.45} +{'loss': 0.0126, 'grad_norm': 0.09146437048912048, 'learning_rate': 5.237549812577987e-06, 'flow_matching_loss': 0.012599368393421174, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.45} +{'loss': 0.0146, 'grad_norm': 0.08692826330661774, 'learning_rate': 5.225002215153973e-06, 'flow_matching_loss': 0.014612434897571802, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.46} +{'loss': 0.013, 'grad_norm': 0.06001284718513489, 'learning_rate': 5.21246649527399e-06, 'flow_matching_loss': 0.013037483114749194, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.46} +{'loss': 0.0124, 'grad_norm': 0.06237281858921051, 'learning_rate': 5.199942668170118e-06, 'flow_matching_loss': 0.012439053039997816, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.46} +{'loss': 0.0133, 'grad_norm': 0.06958480924367905, 'learning_rate': 5.187430749059986e-06, 'flow_matching_loss': 0.013310950435698032, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.46} +{'loss': 0.0144, 'grad_norm': 0.07408503443002701, 'learning_rate': 5.174930753146763e-06, 'flow_matching_loss': 0.01437306934967637, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.47} +{'loss': 0.0124, 'grad_norm': 0.06287696212530136, 'learning_rate': 5.162442695619114e-06, 'flow_matching_loss': 0.012378826271742582, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.47} +{'loss': 0.0129, 'grad_norm': 0.08866557478904724, 'learning_rate': 5.149966591651215e-06, 'flow_matching_loss': 0.012902287859469652, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.47} +{'loss': 0.0141, 'grad_norm': 0.07543319463729858, 'learning_rate': 5.137502456402705e-06, 'flow_matching_loss': 0.014118932094424963, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.48} +{'loss': 0.0144, 'grad_norm': 0.07786840200424194, 'learning_rate': 5.125050305018683e-06, 'flow_matching_loss': 0.014378662779927253, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.48} +{'loss': 0.0139, 'grad_norm': 0.06162134185433388, 'learning_rate': 5.112610152629687e-06, 'flow_matching_loss': 0.013949399814009666, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.48} +{'loss': 0.0146, 'grad_norm': 0.06187082827091217, 'learning_rate': 5.10018201435167e-06, 'flow_matching_loss': 0.014580365642905236, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.49} +{'loss': 0.012, 'grad_norm': 0.0684242695569992, 'learning_rate': 5.087765905286004e-06, 'flow_matching_loss': 0.011973320227116346, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.49} +{'loss': 0.013, 'grad_norm': 0.08950722962617874, 'learning_rate': 5.0753618405194245e-06, 'flow_matching_loss': 0.012989166751503944, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.49} +{'loss': 0.013, 'grad_norm': 0.05878158658742905, 'learning_rate': 5.062969835124042e-06, 'flow_matching_loss': 0.013042198866605759, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.49} +{'loss': 0.0134, 'grad_norm': 0.0706595927476883, 'learning_rate': 5.05058990415731e-06, 'flow_matching_loss': 0.013409404084086418, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.5} +{'loss': 0.0133, 'grad_norm': 0.07735515385866165, 'learning_rate': 5.038222062662007e-06, 'flow_matching_loss': 0.013332903943955898, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.5} +{'loss': 0.0156, 'grad_norm': 0.08529850840568542, 'learning_rate': 5.025866325666237e-06, 'flow_matching_loss': 0.015594257693737745, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.5} +{'loss': 0.0146, 'grad_norm': 0.07035086303949356, 'learning_rate': 5.013522708183383e-06, 'flow_matching_loss': 0.014552695769816637, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.51} +{'loss': 0.0149, 'grad_norm': 0.07600998133420944, 'learning_rate': 5.001191225212096e-06, 'flow_matching_loss': 0.014857903495430947, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.51} 75%|███████▍ | 22370/30000 [13:49:25<4:39:45, 2.20s/it] 75%|███████▍ | 22371/30000 [13:49:27<4:01:39, 1.90s/it] 75%|███████▍ | 22372/30000 [13:49:28<3:34:07, 1.68s/it] 75%|███████▍ | 22373/30000 [13:49:29<3:14:52, 1.53s/it] 75%|███████▍ | 22374/30000 [13:49:30<3:01:47, 1.43s/it] 75%|███████▍ | 22375/30000 [13:49:31<2:52:19, 1.36s/it] 75%|███████▍ | 22376/30000 [13:49:33<2:46:08, 1.31s/it] 75%|███████▍ | 22377/30000 [13:49:39<6:14:22, 2.95s/it] 75%|███████▍ | 22378/30000 [13:49:41<5:06:31, 2.41s/it] 75%|███████▍ | 22379/30000 [13:49:42<4:19:19, 2.04s/it] 75%|███████▍ | 22380/30000 [13:49:43<3:46:53, 1.79s/it] 75%|███████▍ | 22380/30000 [13:49:43<3:46:53, 1.79s/it] 75%|███████▍ | 22381/30000 [13:49:44<3:24:52, 1.61s/it] 75%|███████▍ | 22382/30000 [13:49:45<3:08:48, 1.49s/it] 75%|███████▍ | 22383/30000 [13:49:46<2:56:23, 1.39s/it] 75%|███████▍ | 22384/30000 [13:49:48<2:49:56, 1.34s/it] 75%|███████▍ | 22385/30000 [13:49:54<6:05:19, 2.88s/it] 75%|███████▍ | 22386/30000 [13:49:55<5:00:38, 2.37s/it] 75%|███████▍ | 22387/30000 [13:49:57<4:15:15, 2.01s/it] 75%|███████▍ | 22388/30000 [13:49:58<3:43:02, 1.76s/it] 75%|███████▍ | 22389/30000 [13:49:59<3:21:58, 1.59s/it] 75%|███████▍ | 22390/30000 [13:50:00<3:06:09, 1.47s/it] 75%|███████▍ | 22390/30000 [13:50:00<3:06:09, 1.47s/it] 75%|███████▍ | 22391/30000 [13:50:01<2:55:06, 1.38s/it] 75%|███████▍ | 22392/30000 [13:50:02<2:48:23, 1.33s/it] 75%|███████▍ | 22393/30000 [13:50:09<6:07:20, 2.90s/it] 75%|███████▍ | 22394/30000 [13:50:10<5:01:05, 2.38s/it] 75%|███████▍ | 22395/30000 [13:50:11<4:15:36, 2.02s/it] 75%|███████▍ | 22396/30000 [13:50:13<3:43:31, 1.76s/it] 75%|███████▍ | 22397/30000 [13:50:14<3:21:52, 1.59s/it] 75%|███████▍ | 22398/30000 [13:50:15<3:05:39, 1.47s/it] 75%|███████▍ | 22399/30000 [13:50:16<2:54:18, 1.38s/it] 75%|███████▍ | 22400/30000 [13:50:17<2:46:56, 1.32s/it] 75%|███████▍ | 22400/30000 [13:50:17<2:46:56, 1.32s/it] 75%|███████▍ | 22401/30000 [13:50:23<5:50:02, 2.76s/it] 75%|███████▍ | 22402/30000 [13:50:25<4:49:19, 2.28s/it] 75%|███████▍ | 22403/30000 [13:50:26<4:08:16, 1.96s/it] 75%|███████▍ | 22404/30000 [13:50:27<3:38:01, 1.72s/it] 75%|███████▍ | 22405/30000 [13:50:28<3:17:46, 1.56s/it] 75%|███████▍ | 22406/30000 [13:50:29<3:03:53, 1.45s/it] 75%|███████▍ | 22407/30000 [13:50:30<2:53:39, 1.37s/it] 75%|███████▍ | 22408/30000 [13:50:32<2:47:08, 1.32s/it] 75%|███████▍ | 22409/30000 [13:50:37<5:37:22, 2.67s/it] 75%|███████▍ | 22410/30000 [13:50:39<4:40:32, 2.22s/it] 75%|███████▍ | 22410/30000 [13:50:39<4:40:32, 2.22s/it] 75%|███████▍ | 22411/30000 [13:50:40<4:01:17, 1.91s/it] 75%|███████▍ | 22412/30000 [13:50:41<3:33:07, 1.69s/it] 75%|███████▍ | 22413/30000 [13:50:42<3:14:13, 1.54s/it] 75%|███████▍ | 22414/30000 [13:50:43<3:00:57, 1.43s/it] 75%|███████▍ | 22415/30000 [13:50:45<2:51:24, 1.36s/it] 75%|███████▍ | 22416/30000 [13:50:46<2:45:04, 1.31s/it] 75%|███████▍ | 22417/30000 [13:50:52<5:53:08, 2.79s/it] 75%|███████▍ | 22418/30000 [13:50:53<4:51:54, 2.31s/it] 75%|███████▍ | 22419/30000 [13:50:54<4:10:48, 1.98s/it] 75%|███████▍ | 22420/30000 [13:50:56<3:41:08, 1.75s/it] 75%|███████▍ | 22420/30000 [13:50:56<3:41:08, 1.75s/it] 75%|███████▍ | 22421/30000 [13:50:57<3:19:38, 1.58s/it] 75%|███████▍ | 22422/30000 [13:50:58<3:04:36, 1.46s/it] 75%|███████▍ | 22423/30000 [13:50:59<2:53:11, 1.37s/it] 75%|███████▍ | 22424/30000 [13:51:00<2:46:51, 1.32s/it] 75%|███████▍ | 22425/30000 [13:51:07<6:10:58, 2.94s/it] 75%|███████▍ | 22426/30000 [13:51:08<5:03:45, 2.41s/it] 75%|███████▍ | 22427/30000 [13:51:09<4:17:52, 2.04s/it] 75%|███████▍ | 22428/30000 [13:51:11<3:45:05, 1.78s/it] 75%|███████▍ | 22429/30000 [13:51:12<3:22:36, 1.61s/it] 75%|███████▍ | 22430/30000 [13:51:13<3:06:49, 1.48s/it] 75%|███████▍ | 22430/30000 [13:51:13<3:06:49, 1.48s/it] 75%|███████▍ | 22431/30000 [13:51:14<2:55:19, 1.39s/it] 75%|███████▍ | 22432/30000 [13:51:15<2:47:23, 1.33s/it] 75%|███████▍ | 22433/30000 [13:51:22<5:57:55, 2.84s/it] 75%|███████▍ | 22434/30000 [13:51:23<4:56:30, 2.35s/it] 75%|███████▍ | 22435/30000 [13:51:24<4:12:39, 2.00s/it] 75%|███████▍ | 22436/30000 [13:51:25<3:40:43, 1.75s/it] 75%|███████▍ | 22437/30000 [13:51:26<3:19:05, 1.58s/it] 75%|███████▍ | 22438/30000 [13:51:28<3:03:59, 1.46s/it] 75%|███████▍ | 22439/30000 [13:51:29<2:53:30, 1.38s/it] 75%|███████▍ | 22440/30000 [13:51:30<2:45:35, 1.31s/it] 75%|███████▍ | 22440/30000 [13:51:30<2:45:35, 1.31s/it] 75%|███████▍ | 22441/30000 [13:51:36<6:00:02, 2.86s/it] 75%|███████▍ | 22442/30000 [13:51:38<4:56:13, 2.35s/it] 75%|███████▍ | 22443/30000 [13:51:39<4:11:05, 1.99s/it] 75%|███████▍ | 22444/30000 [13:51:40<3:41:19, 1.76s/it] 75%|███████▍ | 22445/30000 [13:51:41<3:20:07, 1.59s/it] 75%|███████▍ | 22446/30000 [13:51:42<3:04:46, 1.47s/it] 75%|███████▍ | 22447/30000 [13:51:44<2:53:58, 1.38s/it] 75%|███████▍ | 22448/30000 [13:51:45<2:47:19, 1.33s/it] 75%|███████▍ | 22449/30000 [13:51:51<6:04:25, 2.90s/it] 75%|███████▍ | 22450/30000 [13:51:52<4:59:01, 2.38s/it] 75%|███████▍ | 22450/30000 [13:51:52<4:59:01, 2.38s/it] 75%|███████▍ | 22451/30000 [13:51:54<4:14:19, 2.02s/it] 75%|███████▍ | 22452/30000 [13:51:55<3:42:43, 1.77s/it] 75%|███████▍ | 22453/30000 [13:51:56<3:20:09, 1.59s/it] 75%|███████▍ | 22454/30000 [13:51:57<3:05:04, 1.47s/it] 75%|███████▍ | 22455/30000 [13:51:58<2:53:54, 1.38s/it] 75%|███████▍ | 22456/30000 [13:52:00<2:46:40, 1.33s/it] 75%|███████▍ | 22457/30000 [13:52:06<6:06:52, 2.92s/it] 75%|███████▍ | 22458/30000 [13:52:07<5:00:36, 2.39s/it] 75%|███████▍ | 22459/30000 [13:52:09<4:14:13, 2.02s/it] 75%|███████▍ | 22460/30000 [13:52:10<3:42:27, 1.77s/it] 75%|███████▍ | 22460/30000 [13:52:10<3:42:27, 1.77s/it] 75%|███████▍ | 22461/30000 [13:52:11<3:19:28, 1.59s/it] 75%|███████▍ | 22462/30000 [13:52:12<3:03:45, 1.46s/it] 75%|███████▍ | 22463/30000 [13:52:13<2:53:18, 1.38s/it] 75%|███████▍ | 22464/30000 [13:52:14<2:46:48, 1.33s/it] 75%|███████▍ | 22465/30000 [13:52:20<5:34:15, 2.66s/it] 75%|███████▍ | 22466/30000 [13:52:21<4:37:19, 2.21s/it] 75%|███████▍ | 22467/30000 [13:52:23<3:58:08, 1.90s/it] 75%|███████▍ | 22468/30000 [13:52:24<3:31:48, 1.69s/it] 75%|███████▍ | 22469/30000 [13:52:25<3:12:58, 1.54s/it] 75%|███████▍ | 22470/30000 [13:52:26<3:00:08, 1.44s/it] 75%|███████▍ | 22470/30000 [13:52:26<3:00:08, 1.44s/it] 75%|███████▍ | 22471/30000 [13:52:27<2:50:45, 1.36s/it] 75%|███████▍ | 22472/30000 [13:52:29<2:44:28, 1.31s/it] 75%|███████▍ | 22473/30000 [13:52:35<5:52:48, 2.81s/it] 75%|███████▍ | 22474/30000 [13:52:36<4:51:29, 2.32s/it] 75%|███████▍ | 22475/30000 [13:52:37<4:08:39, 1.98s/it] 75%|███████▍ | 22476/30000 [13:52:38<3:38:37, 1.74s/it] 75%|███████▍ | 22477/30000 [13:52:40<3:17:41, 1.58s/it] 75%|███████▍ | 22478/30000 [13:52:41<3:03:34, 1.46s/it] 75%|███████▍ | 22479/30000 [13:52:42<2:52:30, 1.38s/it] 75%|███████▍ | 22480/30000 [13:52:43<2:45:18, 1.32s/it] 75%|███████▍ | 22480/30000 [13:52:43<2:45:18, 1.32s/it] 75%|███████▍ | 22481/30000 [13:52:50<6:05:51, 2.92s/it] 75%|███████▍ | 22482/30000 [13:52:51<5:00:01, 2.39s/it] 75%|███████▍ | 22483/30000 [13:52:52<4:14:32, 2.03s/it] 75%|███████▍ | 22484/30000 [13:52:53<3:42:54, 1.78s/it] 75%|███████▍ | 22485/30000 [13:52:55<3:20:52, 1.60s/it] 75%|███████▍ | 22486/30000 [13:52:56<3:04:35, 1.47s/it] 75%|███████▍ | 22487/30000 [13:52:57<2:53:36, 1.39s/it] 75%|███████▍ | 22488/30000 [13:52:58<2:46:22, 1.33s/it] 75%|███████▍ | 22489/30000 [13:53:05<6:08:07, 2.94s/it] 75%|███████▍ | 22490/30000 [13:53:06<5:01:12, 2.41s/it] 75%|███████▍ | 22490/30000 [13:53:06<5:01:12, 2.41s/it] 75%|███████▍ | 22491/30000 [13:53:07<4:15:41, 2.04s/it] 75%|███████▍ | 22492/30000 [13:53:08<3:43:10, 1.78s/it] 75%|███████▍ | 22493/30000 [13:53:09<3:20:36, 1.60s/it] 75%|███████▍ | 22494/30000 [13:53:11<3:04:51, 1.48s/it] 75%|███████▍ | 22495/30000 [13:53:12<2:53:01, 1.38s/it] 75%|███████▍ | 22496/30000 [13:53:13<2:46:26, 1.33s/it] 75%|███████▍ | 22497/30000 [13:53:20<6:18:44, 3.03s/it] 75%|███████▍ | 22498/30000 [13:53:21<5:09:01, 2.47s/it] 75%|███████▍ | 22499/30000 [13:53:22<4:21:05, 2.09s/it] 75%|███████▌ | 22500/30000 [13:53:24<3:48:12, 1.83s/it] 75%|███████▌ | 22500/30000 [13:53:24<3:48:12, 1.83s/it] 75%|███████▌ | 22501/30000 [13:53:25<3:25:36, 1.65s/it] 75%|███████▌ | 22502/30000 [13:53:26<3:08:09, 1.51s/it] 75%|███████▌ | 22503/30000 [13:53:27<2:55:25, 1.40s/it] 75%|███████▌ | 22504/30000 [13:53:28<2:46:45, 1.33s/it] 75%|███████▌ | 22505/30000 [13:53:35<6:04:11, 2.92s/it] 75%|███████▌ | 22506/30000 [13:53:36<4:58:52, 2.39s/it] 75%|███████▌ | 22507/30000 [13:53:37<4:14:34, 2.04s/it] 75%|███████▌ | 22508/30000 [13:53:39<3:41:58, 1.78s/it] 75%|███████▌ | 22509/30000 [13:53:40<3:19:50, 1.60s/it] 75%|███████▌ | 22510/30000 [13:53:41<3:04:07, 1.48s/it] 75%|███████▌ | 22510/30000 [13:53:41<3:04:07, 1.48s/it] 75%|███████▌ | 22511/30000 [13:53:42<2:52:41, 1.38s/it] 75%|███████▌ | 22512/30000 [13:53:43<2:45:24, 1.33s/it] 75%|███████▌ | 22513/30000 [13:53:49<5:46:24, 2.78s/it] 75%|███████▌ | 22514/30000 [13:53:51<4:47:08, 2.30s/it] 75%|███████▌ | 22515/30000 [13:53:52<4:05:27, 1.97s/it] 75%|███████▌ | 22516/30000 [13:53:53<3:36:33, 1.74s/it] 75%|███████▌ | 22517/30000 [13:53:54<3:15:42, 1.57s/it] 75%|███████▌ | 22518/30000 [13:53:55<3:01:43, 1.46s/it] 75%|███████▌ | 22519/30000 [13:53:57<2:51:02, 1.37s/it] 75%|███████▌ | 22520/30000 [13:53:58<2:44:00, 1.32s/it] 75%|███████▌ | 22520/30000 [13:53:58<2:44:00, 1.32s/it] 75%|███████▌ | 22521/30000 [13:54:05<6:14:23, 3.00s/it] 75%|███████▌ | 22522/30000 [13:54:06<5:05:53, 2.45s/it] 75%|███████▌ | 22523/30000 [13:54:07<4:18:02, 2.07s/it] 75%|███████▌ | 22524/30000 [13:54:08<3:45:07, 1.81s/it] 75%|███████▌ | 22525/30000 [13:54:09<3:21:18, 1.62s/it] 75%|███████▌ | 22526/30000 [13:54:11<3:04:56, 1.48s/it] 75%|███████▌ | 22527/30000 [13:54:12<2:52:50, 1.39s/it] 75%|███████▌ | 22528/30000 [13:54:13<2:45:50, 1.33s/it] 75%|███████▌ | 22529/30000 [13:54:20<6:22:27, 3.07s/it] 75%|███████▌ | 22530/30000 [13:54:21<5:11:28, 2.50s/it] 75%|███████▌ | 22530/30000 [13:54:21<5:11:28, 2.50s/it] 75%|███████▌ | 22531/30000 [13:54:22<4:22:15, 2.11s/it] 75%|███████▌ | 22532/30000 [13:54:24<3:47:06, 1.82s/it] 75%|███████▌ | 22533/30000 [13:54:25<3:23:34, 1.64s/it] 75%|███████▌ | 22534/30000 [13:54:26<3:07:00, 1.50s/it] 75%|███████▌ | 22535/30000 [13:54:27<2:55:22, 1.41s/it] 75%|███████▌ | 22536/30000 [13:54:28<2:48:34, 1.36s/it] 75%|███████▌ | 22537/30000 [13:54:35<6:21:03, 3.06s/it] 75%|███████▌ | 22538/30000 [13:54:37<5:10:43, 2.50s/it] 75%|███████▌ | 22539/30000 [13:54:38<4:21:49, 2.11s/it] 75%|███████▌ | 22540/30000 [13:54:39<3:48:02, 1.83s/it] 75%|███████▌ | 22540/30000 [13:54:39<3:48:02, 1.83s/it] 75%|███████▌ | 22541/30000 [13:54:40<3:24:29, 1.64s/it] 75%|███████▌ | 22542/30000 [13:54:41<3:07:41, 1.51s/it] 75%|███████▌ | 22543/30000 [13:54:43<2:54:55, 1.41s/it] 75%|███████▌ | 22544/30000 [13:54:44<2:47:04, 1.34s/it] 75%|███████▌ | 22545/30000 [13:54:50<5:57:16, 2.88s/it] 75%|███████▌ | 22546/30000 [13:54:51<4:54:42, 2.37s/it] 75%|███████▌ | 22547/30000 [13:54:53<4:10:27, 2.02s/it] 75%|███████▌ | 22548/30000 [13:54:54<3:39:23, 1.77s/it] 75%|███████▌ | 22549/30000 [13:54:55<3:18:44, 1.60s/it] 75%|███████▌ | 22550/30000 [13:54:56<3:03:07, 1.47s/it] 75%|███████▌ | 22550/30000 [13:54:56<3:03:07, 1.47s/it] 75%|███████▌ | 22551/30000 [13:54:57<2:51:46, 1.38s/it] 75%|███████▌ | 22552/30000 [13:54:59<2:44:38, 1.33s/it] 75%|███████▌ | 22553/30000 [13:55:04<5:33:27, 2.69s/it] 75%|███████▌ | 22554/30000 [13:55:06<4:37:16, 2.23s/it] 75%|███████▌ | 22555/30000 [13:55:07<3:58:34, 1.92s/it] 75%|███████▌ | 22556/30000 [13:55:08<3:30:53, 1.70s/it] 75%|███████▌ | 22557/30000 [13:55:09<3:10:53, 1.54s/it] 75%|███████▌ | 22558/30000 [13:55:10<2:58:08, 1.44s/it] 75%|███████▌ | 22559/30000 [13:55:11<2:48:21, 1.36s/it] 75%|███████▌ | 22560/30000 [13:55:13<2:42:06, 1.31s/it] 75%|███████▌ | 22560/30000 [13:55:13<2:42:06, 1.31s/it] 75%|███████▌ | 22561/30000 [13:55:18<5:26:53, 2.64s/it] 75%|███████▌ | 22562/30000 [13:55:20<4:32:28, 2.20s/it] 75%|███████▌ | 22563/30000 [13:55:21<3:55:17, 1.90s/it] 75%|███████▌ | 22564/30000 [13:55:22<3:29:05, 1.69s/it] 75%|███████▌ | 22565/30000 [13:55:23<3:10:32, 1.54s/it] 75%|███████▌ | 22566/30000 [13:55:24<2:57:45, 1.43s/it] 75%|███████▌ | 22567/30000 [13:55:26<2:47:59, 1.36s/it] 75%|███████▌ | 22568/30000 [13:55:27<2:42:07, 1.31s/it] 75%|███████▌ | 22569/30000 [13:55:33<5:40:06, 2.75s/it] 75%|███████▌ | 22570/30000 [13:55:34<4:41:40, 2.27s/it] 75%|███████▌ | 22570/30000 [13:55:34<4:41:40, 2.27s/it] 75%|███████▌ | 22571/30000 [13:55:35<4:01:42, 1.95s/it] 75%|███████▌ | 22572/30000 [13:55:36<3:33:30, 1.72s/it] 75%|███████▌ | 22573/30000 [13:55:38<3:13:36, 1.56s/it] 75%|███████▌ | 22574/30000 [13:55:39<2:59:29, 1.45s/it] 75%|███████▌ | 22575/30000 [13:55:40<2:49:05, 1.37s/it] 75%|███████▌ | 22576/30000 [13:55:41<2:43:31, 1.32s/it] 75%|███████▌ | 22577/30000 [13:55:47<5:22:53, 2.61s/it] 75%|███████▌ | 22578/30000 [13:55:48<4:29:22, 2.18s/it] 75%|███████▌ | 22579/30000 [13:55:49<3:53:14, 1.89s/it] 75%|███████▌ | 22580/30000 [13:55:50<3:27:00, 1.67s/it] 75%|███████▌ | 22580/30000 [13:55:50<3:27:00, 1.67s/it] 75%|███████▌ | 22581/30000 [13:55:51<3:08:21, 1.52s/it] 75%|███████▌ | 22582/30000 [13:55:53<2:56:48, 1.43s/it] 75%|███████▌ | 22583/30000 [13:55:54<2:48:08, 1.36s/it] 75%|███████▌ | 22584/30000 [13:55:55<2:42:13, 1.31s/it] 75%|███████▌ | 22585/30000 [13:56:01<5:39:12, 2.74s/it] 75%|███████▌ | 22586/30000 [13:56:02<4:40:40, 2.27s/it] 75%|███████▌ | 22587/30000 [13:56:04<4:00:36, 1.95s/it] 75%|███████▌ | 22588/30000 [13:56:05<3:32:27, 1.72s/it] 75%|███████▌ | 22589/30000 [13:56:06<3:12:31, 1.56s/it] 75%|███████▌ | 22590/30000 [13:56:07<2:59:09, 1.45s/it] 75%|███████▌ | 22590/30000 [13:56:07<2:59:09, 1.45s/it] 75%|███████▌ | 22591/30000 [13:56:08<2:50:07, 1.38s/it] 75%|███████▌ | 22592/30000 [13:56:10<2:42:55, 1.32s/it] 75%|███████▌ | 22593/30000 [13:56:15<5:27:39, 2.65s/it] 75%|███████▌ | 22594/30000 [13:56:16<4:32:29, 2.21s/it] 75%|███████▌ | 22595/30000 [13:56:18<3:54:46, 1.90s/it] 75%|███████▌ | 22596/30000 [13:56:19<3:28:42, 1.69s/it] 75%|███████▌ | 22597/30000 [13:56:20<3:09:34, 1.54s/it] 75%|███████▌ | 22598/30000 [13:56:21<2:56:35, 1.43s/it] 75%|███████▌ | 22599/30000 [13:56:22<2:47:14, 1.36s/it] 75%|███████▌ | 22600/30000 [13:56:24<2:41:16, 1.31s/it] 75%|███████▌ | 22600/30000 [13:56:24<2:41:16, 1.31s/it] 75%|███████▌ | 22601/30000 [13:56:29<5:29:11, 2.67s/it] 75%|███████▌ | 22602/30000 [13:56:31<4:34:54, 2.23s/it] 75%|███████▌ | 22603/30000 [13:56:32<3:56:36, 1.92s/it] 75%|███████▌ | 22604/30000 [13:56:33<3:29:41, 1.70s/it] 75%|███████▌ | 22605/30000 [13:56:34<3:11:08, 1.55s/it] 75%|███████▌ | 22606/30000 [13:56:35<2:57:56, 1.44s/it] 75%|███████▌ | 22607/30000 [13:56:37<2:48:18, 1.37s/it] 75%|███████▌ | 22608/30000 [13:56:38<2:42:11, 1.32s/it] 75%|███████▌ | 22609/30000 [13:56:44<5:31:51, 2.69s/it] 75%|███████▌ | 22610/30000 [13:56:45<4:35:09, 2.23s/it] 75%|███████▌ | 22610/30000 [13:56:45<4:35:09, 2.23s/it] 75%|███████▌ | 22611/30000 [13:56:46<3:56:16, 1.92s/it] 75%|███████▌ | 22612/30000 [13:56:47<3:29:04, 1.70s/it] 75%|███████▌ | 22613/30000 [13:56:48<3:09:40, 1.54s/it] 75%|███████▌ | 22614/30000 [13:56:50<2:56:45, 1.44s/it] 75%|███████▌ | 22615/30000 [13:56:51<2:47:07, 1.36s/it] 75%|███████▌ | 22616/30000 [13:56:52<2:40:56, 1.31s/it] 75%|███████▌ | 22617/30000 [13:56:58<5:30:35, 2.69s/it] 75%|███████▌ | 22618/30000 [13:56:59<4:34:14, 2.23s/it] 75%|███████▌ | 22619/30000 [13:57:00<3:55:36, 1.92s/it] 75%|███████▌ | 22620/30000 [13:57:01<3:28:50, 1.70s/it] 75%|███████▌ | 22620/30000 [13:57:01<3:28:50, 1.70s/it] 75%|███████▌ | 22621/30000 [13:57:03<3:11:17, 1.56s/it] 75%|███████▌ | 22622/30000 [13:57:04<2:58:17, 1.45s/it] 75%|███████▌ | 22623/30000 [13:57:05<2:48:14, 1.37s/it] 75%|███████▌ | 22624/30000 [13:57:06<2:42:08, 1.32s/it] 75%|███████▌ | 22625/30000 [13:57:12<5:27:30, 2.66s/it] 75%|███████▌ | 22626/30000 [13:57:13<4:32:04, 2.21s/it] 75%|███████▌ | 22627/30000 [13:57:14<3:53:42, 1.90s/it] 75%|███████▌ | 22628/30000 [13:57:16<3:26:40, 1.68s/it] 75%|███████▌ | 22629/30000 [13:57:17<3:07:33, 1.53s/it] 75%|███████▌ | 22630/30000 [13:57:18<2:55:13, 1.43s/it] 75%|███████▌ | 22630/30000 [13:57:18<2:55:13, 1.43s/it] 75%|███████▌ | 22631/30000 [13:57:19<2:46:27, 1.36s/it] 75%|███████▌ | 22632/30000 [13:57:20<2:40:44, 1.31s/it] 75%|███████▌ | 22633/30000 [13:57:26<5:32:49, 2.71s/it] 75%|███████▌ | 22634/30000 [13:57:27<4:36:18, 2.25s/it] 75%|███████▌ | 22635/30000 [13:57:29<3:56:55, 1.93s/it] 75%|███████▌ | 22636/30000 [13:57:30<3:29:37, 1.71s/it] 75%|███████▌ | 22637/30000 [13:57:31<3:10:33, 1.55s/it] 75%|███████▌ | 22638/30000 [13:57:32<2:57:09, 1.44s/it] 75%|███████▌ | 22639/30000 [13:57:33<2:46:57, 1.36s/it] 75%|███████▌ | 22640/30000 [13:57:35<2:41:28, 1.32s/it] 75%|███████▌ | 22640/30000 [13:57:35<2:41:28, 1.32s/it] 75%|███████▌ | 22641/30000 [13:57:41<5:47:57, 2.84s/it] 75%|███████▌ | 22642/30000 [13:57:42<4:48:20, 2.35s/it] 75%|███████▌ | 22643/30000 [13:57:43<4:05:18, 2.00s/it] 75%|███████▌ | 22644/30000 [13:57:45<3:34:45, 1.75s/it] 75%|███████▌ | 22645/30000 [13:57:46<3:13:18, 1.58s/it] 75%|███████▌ | 22646/30000 [13:57:47<3:02:41, 1.49s/it] 75%|███████▌ | 22647/30000 [13:57:48<2:51:12, 1.40s/it] 75%|███████▌ | 22648/30000 [13:57:49<2:43:50, 1.34s/it] 75%|███████▌ | 22649/30000 [13:57:56<5:55:46, 2.90s/it] 76%|███████▌ | 22650/30000 [13:57:57<4:53:02, 2.39s/it] 76%|███████▌ | 22650/30000 [13:57:57<4:53:02, 2.39s/it] 76%|███████▌ | 22651/30000 [13:57:58<4:08:37, 2.03s/it] 76%|███████▌ | 22652/30000 [13:57:59<3:37:06, 1.77s/it] 76%|███████▌ | 22653/30000 [13:58:01<3:15:24, 1.60s/it] 76%|███████▌ | 22654/30000 [13:58:02<2:59:35, 1.47s/it] 76%|███████▌ | 22655/30000 [13:58:03<2:49:02, 1.38s/it] 76%|███████▌ | 22656/30000 [13:58:04<2:42:23, 1.33s/it] 76%|███████▌ | 22657/30000 [13:58:10<5:41:44, 2.79s/it] 76%|███████▌ | 22658/30000 [13:58:12<4:42:04, 2.31s/it] 76%|███████▌ | 22659/30000 [13:58:13<4:01:26, 1.97s/it] 76%|███████▌ | 22660/30000 [13:58:14<3:32:33, 1.74s/it] 76%|███████▌ | 22660/30000 [13:58:14<3:32:33, 1.74s/it] 76%|███████▌ | 22661/30000 [13:58:15<3:12:38, 1.57s/it] 76%|███████▌ | 22662/30000 [13:58:16<2:57:54, 1.45s/it] 76%|███████▌ | 22663/30000 [13:58:17<2:47:03, 1.37s/it] 76%|███████▌ | 22664/30000 [13:58:19<2:40:30, 1.31s/it] 76%|███████▌ | 22665/30000 [13:58:25<5:35:39, 2.75s/it] 76%|███████▌ | 22666/30000 [13:58:26<4:38:23, 2.28s/it] 76%|███████▌ | 22667/30000 [13:58:27<3:58:21, 1.95s/it] 76%|███████▌ | 22668/30000 [13:58:28<3:29:40, 1.72s/it] 76%|███████▌ | 22669/30000 [13:58:29<3:10:10, 1.56s/it] 76%|███████▌ | 22670/30000 [13:58:31<2:57:13, 1.45s/it] 76%|███████▌ | 22670/30000 [13:58:31<2:57:13, 1.45s/it] 76%|███████▌ | 22671/30000 [13:58:32<2:47:05, 1.37s/it] 76%|███████▌ | 22672/30000 [13:58:33<2:40:54, 1.32s/it] 76%|███████▌ | 22673/30000 [13:58:39<5:47:09, 2.84s/it] 76%|███████▌ | 22674/30000 [13:58:41<4:46:01, 2.34s/it] 76%|███████▌ | 22675/30000 [13:58:42<4:03:54, 2.00s/it] 76%|███████▌ | 22676/30000 [13:58:43<3:34:33, 1.76s/it] 76%|███████▌ | 22677/30000 [13:58:44<3:13:38, 1.59s/it] 76%|███████▌ | 22678/30000 [13:58:45<2:59:06, 1.47s/it] 76%|███████▌ | 22679/30000 [13:58:47<2:48:21, 1.38s/it] 76%|███████▌ | 22680/30000 [13:58:48<2:41:35, 1.32s/it] 76%|███████▌ | 22680/30000 [13:58:48<2:41:35, 1.32s/it] 76%|███████▌ | 22681/30000 [13:58:54<5:51:27, 2.88s/it] 76%|███████▌ | 22682/30000 [13:58:55<4:48:40, 2.37s/it] 76%|███████▌ | 22683/30000 [13:58:57<4:05:34, 2.01s/it] 76%|███████▌ | 22684/30000 [13:58:58<3:34:28, 1.76s/it] 76%|███████▌ | 22685/30000 [13:58:59<3:13:16, 1.59s/it] 76%|███████▌ | 22686/30000 [13:59:00<2:58:17, 1.46s/it] 76%|███████▌ | 22687/30000 [13:59:01<2:47:30, 1.37s/it] 76%|███████▌ | 22688/30000 [13:59:03<2:40:41, 1.32s/it] 76%|███████▌ | 22689/30000 [13:59:09<6:04:32, 2.99s/it] 76%|███████▌ | 22690/30000 [13:59:11<4:57:49, 2.44s/it] 76%|███████▌ | 22690/30000 [13:59:11<4:57:49, 2.44s/it] 76%|███████▌ | 22691/30000 [13:59:12<4:11:11, 2.06s/it] 76%|███████▌ | 22692/30000 [13:59:13<3:39:20, 1.80s/it] 76%|███████▌ | 22693/30000 [13:59:14<3:17:26, 1.62s/it] 76%|███████▌ | 22694/30000 [13:59:15<3:01:46, 1.49s/it] 76%|███████▌ | 22695/30000 [13:59:17<2:50:21, 1.40s/it] 76%|███████▌ | 22696/30000 [13:59:18<2:42:59, 1.34s/it] 76%|███████▌ | 22697/30000 [13:59:25<6:05:05, 3.00s/it] 76%|███████▌ | 22698/30000 [13:59:26<4:58:33, 2.45s/it] 76%|███████▌ | 22699/30000 [13:59:27<4:12:33, 2.08s/it] 76%|███████▌ | 22700/30000 [13:59:28<3:39:33, 1.80s/it] 76%|███████▌ | 22700/30000 [13:59:28<3:39:33, 1.80s/it] 76%|███████▌ | 22701/30000 [13:59:29<3:17:33, 1.62s/it] 76%|███████▌ | 22702/30000 [13:59:31<3:01:12, 1.49s/it] 76%|███████▌ | 22703/30000 [13:59:32<2:49:54, 1.40s/it] 76%|███████▌ | 22704/30000 [13:59:33<2:41:57, 1.33s/it] 76%|███████▌ | 22705/30000 [13:59:39<5:38:52, 2.79s/it] 76%|███████▌ | 22706/30000 [13:59:40<4:39:14, 2.30s/it] 76%|███████▌ | 22707/30000 [13:59:41<3:59:01, 1.97s/it] 76%|███████▌ | 22708/30000 [13:59:43<3:30:32, 1.73s/it] 76%|███████▌ | 22709/30000 [13:59:44<3:10:26, 1.57s/it] 76%|███████▌ | 22710/30000 [13:59:45<2:57:01, 1.46s/it] 76%|███████▌ | 22710/30000 [13:59:45<2:57:01, 1.46s/it] 76%|███████▌ | 22711/30000 [13:59:46<2:46:18, 1.37s/it] 76%|███████▌ | 22712/30000 [13:59:47<2:38:57, 1.31s/it] 76%|███████▌ | 22713/30000 [13:59:53<5:26:48, 2.69s/it] 76%|███████▌ | 22714/30000 [13:59:54<4:31:25, 2.24s/it] 76%|███████▌ | 22715/30000 [13:59:56<3:53:09, 1.92s/it] 76%|███████▌ | 22716/30000 [13:59:57<3:25:49, 1.70s/it] 76%|███████▌ | 22717/30000 [13:59:58<3:07:02, 1.54s/it] 76%|███████▌ | 22718/30000 [13:59:59<2:53:57, 1.43s/it] 76%|███████▌ | 22719/30000 [14:00:00<2:44:36, 1.36s/it] 76%|███████▌ | 22720/30000 [14:00:01<2:38:11, 1.30s/it] 76%|███████▌ | 22720/30000 [14:00:02<2:38:11, 1.30s/it] 76%|███████▌ | 22721/30000 [14:00:08<6:04:49, 3.01s/it] 76%|███████▌ | 22722/30000 [14:00:10<4:57:41, 2.45s/it] 76%|███████▌ | 22723/30000 [14:00:11<4:11:24, 2.07s/it] 76%|███████▌ | 22724/30000 [14:00:12<3:39:08, 1.81s/it] 76%|███████▌ | 22725/30000 [14:00:13<3:16:29, 1.62s/it] 76%|███████▌ | 22726/30000 [14:00:14<3:00:59, 1.49s/it] 76%|███████▌ | 22727/30000 [14:00:16<2:49:36, 1.40s/it] 76%|███████▌ | 22728/30000 [14:00:17<2:42:31, 1.34s/it] 76%|███████▌ | 22729/30000 [14:00:23<5:57:05, 2.95s/it] 76%|███████▌ | 22730/30000 [14:00:25<4:52:02, 2.41s/it] 76%|███████▌ | 22730/30000 [14:00:25<4:52:02, 2.41s/it] 76%|███████▌ | 22731/30000 [14:00:26<4:07:32, 2.04s/it] 76%|███████▌ | 22732/30000 [14:00:27<3:36:08, 1.78s/it] 76%|███████▌ | 22733/30000 [14:00:28<3:14:30, 1.61s/it] 76%|███████▌ | 22734/30000 [14:00:29<2:59:23, 1.48s/it] 76%|███████▌ | 22735/30000 [14:00:31<2:48:56, 1.40s/it] 76%|███████▌ | 22736/30000 [14:00:32<2:41:54, 1.34s/it] 76%|███████▌ | 22737/30000 [14:00:38<5:51:33, 2.90s/it] 76%|███████▌ | 22738/30000 [14:00:39<4:48:25, 2.38s/it] 76%|███████▌ | 22739/30000 [14:00:41<4:04:16, 2.02s/it] 76%|███████▌ | 22740/30000 [14:00:42<3:33:17, 1.76s/it] 76%|███████▌ | 22740/30000 [14:00:42<3:33:17, 1.76s/it] 76%|███████▌ | 22741/30000 [14:00:43<3:12:02, 1.59s/it] 76%|███████▌ | 22742/30000 [14:00:44<2:56:52, 1.46s/it] 76%|███████▌ | 22743/30000 [14:00:45<2:47:02, 1.38s/it] 76%|███████▌ | 22744/30000 [14:00:47<2:39:45, 1.32s/it] 76%|███████▌ | 22745/30000 [14:00:53<5:44:06, 2.85s/it] 76%|███████▌ | 22746/30000 [14:00:54<4:42:49, 2.34s/it] 76%|███████▌ | 22747/30000 [14:00:55<4:01:08, 1.99s/it] 76%|███████▌ | 22748/30000 [14:00:56<3:31:49, 1.75s/it] 76%|███████▌ | 22749/30000 [14:00:58<3:11:02, 1.58s/it] 76%|███████▌ | 22750/30000 [14:00:59<2:56:33, 1.46s/it] 76%|███████▌ | 22750/30000 [14:00:59<2:56:33, 1.46s/it] 76%|███████▌ | 22751/30000 [14:01:00<2:46:18, 1.38s/it] 76%|███████▌ | 22752/30000 [14:01:01<2:39:48, 1.32s/it] 76%|███████▌ | 22753/30000 [14:01:07<5:27:12, 2.71s/it] 76%|███████▌ | 22754/30000 [14:01:08<4:31:17, 2.25s/it] 76%|███████▌ | 22755/30000 [14:01:10<3:52:54, 1.93s/it] 76%|███████▌ | 22756/30000 [14:01:11<3:25:48, 1.70s/it] 76%|███████▌ | 22757/30000 [14:01:12<3:06:43, 1.55s/it] 76%|███████▌ | 22758/30000 [14:01:13<2:53:35, 1.44s/it] 76%|███████▌ | 22759/30000 [14:01:14<2:43:37, 1.36s/it] 76%|███████▌ | 22760/30000 [14:01:15<2:36:51, 1.30s/it] 76%|███████▌ | 22760/30000 [14:01:15<2:36:51, 1.30s/it] 76%|███████▌ | 22761/30000 [14:01:22<5:45:13, 2.86s/it] 76%|███████▌ | 22762/30000 [14:01:23<4:44:31, 2.36s/it] 76%|███████▌ | 22763/30000 [14:01:24<4:01:07, 2.00s/it] 76%|███████▌ | 22764/30000 [14:01:25<3:31:59, 1.76s/it] 76%|███████▌ | 22765/30000 [14:01:27<3:10:37, 1.58s/it] 76%|███████▌ | 22766/30000 [14:01:28<2:56:16, 1.46s/it] 76%|███████▌ | 22767/30000 [14:01:29<2:46:20, 1.38s/it] 76%|███████▌ | 22768/30000 [14:01:30<2:39:46, 1.33s/it] 76%|███████▌ | 22769/30000 [14:01:36<5:29:53, 2.74s/it] 76%|███████▌ | 22770/30000 [14:01:37<4:33:08, 2.27s/it] 76%|███████▌ | 22770/30000 [14:01:37<4:33:08, 2.27s/it] 76%|███████▌ | 22771/30000 [14:01:39<3:54:38, 1.95s/it] 76%|███████▌ | 22772/30000 [14:01:40<3:27:07, 1.72s/it] 76%|███████▌ | 22773/30000 [14:01:41<3:08:30, 1.57s/it] 76%|███████▌ | 22774/30000 [14:01:42<2:54:41, 1.45s/it] 76%|███████▌ | 22775/30000 [14:01:43<2:45:04, 1.37s/it] 76%|███████▌ | 22776/30000 [14:01:45<2:38:43, 1.32s/it] 76%|███████▌ | 22777/30000 [14:01:51<5:27:45, 2.72s/it] 76%|███████▌ | 22778/30000 [14:01:52<4:31:55, 2.26s/it] 76%|███████▌ | 22779/30000 [14:01:53<3:53:23, 1.94s/it] 76%|███████▌ | 22780/30000 [14:01:54<3:26:44, 1.72s/it] 76%|███████▌ | 22780/30000 [14:01:54<3:26:44, 1.72s/it] 76%|███████▌ | 22781/30000 [14:01:55<3:07:33, 1.56s/it] 76%|███████▌ | 22782/30000 [14:01:57<2:54:49, 1.45s/it] 76%|███████▌ | 22783/30000 [14:01:58<2:44:22, 1.37s/it] 76%|███████▌ | 22784/30000 [14:01:59<2:38:33, 1.32s/it] 76%|███████▌ | 22785/30000 [14:02:04<5:13:23, 2.61s/it] 76%|███████▌ | 22786/30000 [14:02:06<4:21:05, 2.17s/it] 76%|███████▌ | 22787/30000 [14:02:07<3:45:54, 1.88s/it] 76%|███████▌ | 22788/30000 [14:02:08<3:20:53, 1.67s/it] 76%|███████▌ | 22789/30000 [14:02:09<3:02:23, 1.52s/it] 76%|███████▌ | 22790/30000 [14:02:10<2:51:17, 1.43s/it] +{'loss': 0.0129, 'grad_norm': 0.08999856561422348, 'learning_rate': 4.9888718917363e-06, 'flow_matching_loss': 0.012877288274466991, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.51} +{'loss': 0.0125, 'grad_norm': 0.07064348459243774, 'learning_rate': 4.9765647227251395e-06, 'flow_matching_loss': 0.012458019005134702, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.51} +{'loss': 0.0123, 'grad_norm': 0.06244400888681412, 'learning_rate': 4.964269733132992e-06, 'flow_matching_loss': 0.012317993957549334, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.52} +{'loss': 0.0137, 'grad_norm': 0.06266000121831894, 'learning_rate': 4.951986937899428e-06, 'flow_matching_loss': 0.013677875138819217, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.52} +{'loss': 0.0132, 'grad_norm': 0.07671651244163513, 'learning_rate': 4.9397163519492e-06, 'flow_matching_loss': 0.01320616789162159, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.52} +{'loss': 0.0131, 'grad_norm': 0.06471577286720276, 'learning_rate': 4.9274579901922275e-06, 'flow_matching_loss': 0.013066184241324663, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.53} +{'loss': 0.012, 'grad_norm': 0.07541292160749435, 'learning_rate': 4.9152118675235705e-06, 'flow_matching_loss': 0.012017757818102836, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.53} +{'loss': 0.0138, 'grad_norm': 0.08353924751281738, 'learning_rate': 4.902977998823432e-06, 'flow_matching_loss': 0.013801049906760454, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.53} +{'loss': 0.014, 'grad_norm': 0.060540370643138885, 'learning_rate': 4.890756398957114e-06, 'flow_matching_loss': 0.013974407128989697, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.53} +{'loss': 0.0134, 'grad_norm': 0.07175809890031815, 'learning_rate': 4.8785470827750075e-06, 'flow_matching_loss': 0.013436923641711474, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.54} +{'loss': 0.0135, 'grad_norm': 0.06997458636760712, 'learning_rate': 4.866350065112588e-06, 'flow_matching_loss': 0.013483384158462286, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.54} +{'loss': 0.0117, 'grad_norm': 0.08805360645055771, 'learning_rate': 4.854165360790381e-06, 'flow_matching_loss': 0.011747473757714033, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.54} +{'loss': 0.0123, 'grad_norm': 0.07783672213554382, 'learning_rate': 4.8419929846139425e-06, 'flow_matching_loss': 0.012278868863359094, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.55} +{'loss': 0.0146, 'grad_norm': 0.08615543693304062, 'learning_rate': 4.829832951373872e-06, 'flow_matching_loss': 0.014619407430291175, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.55} +{'loss': 0.013, 'grad_norm': 0.07616326212882996, 'learning_rate': 4.817685275845746e-06, 'flow_matching_loss': 0.013019076734781265, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.55} +{'loss': 0.0143, 'grad_norm': 0.07834882289171219, 'learning_rate': 4.805549972790143e-06, 'flow_matching_loss': 0.014324491005390882, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.56} +{'loss': 0.0132, 'grad_norm': 0.07489871978759766, 'learning_rate': 4.7934270569525966e-06, 'flow_matching_loss': 0.013236889429390431, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.56} +{'loss': 0.0127, 'grad_norm': 0.08917434513568878, 'learning_rate': 4.7813165430635874e-06, 'flow_matching_loss': 0.01274797460064292, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.56} +{'loss': 0.0139, 'grad_norm': 0.07994019985198975, 'learning_rate': 4.769218445838544e-06, 'flow_matching_loss': 0.013948941975831986, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.56} +{'loss': 0.0134, 'grad_norm': 0.07774246484041214, 'learning_rate': 4.757132779977787e-06, 'flow_matching_loss': 0.013383117970079184, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.57} +{'loss': 0.0127, 'grad_norm': 0.06253936886787415, 'learning_rate': 4.745059560166542e-06, 'flow_matching_loss': 0.012692431826144457, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.57} +{'loss': 0.0127, 'grad_norm': 0.0599326528608799, 'learning_rate': 4.732998801074911e-06, 'flow_matching_loss': 0.012691887281835078, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.57} +{'loss': 0.0128, 'grad_norm': 0.06588095426559448, 'learning_rate': 4.7209505173578445e-06, 'flow_matching_loss': 0.01280516739934683, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.58} +{'loss': 0.0148, 'grad_norm': 0.108277827501297, 'learning_rate': 4.7089147236551555e-06, 'flow_matching_loss': 0.014797658193856478, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.58} +{'loss': 0.0127, 'grad_norm': 0.05974648892879486, 'learning_rate': 4.696891434591464e-06, 'flow_matching_loss': 0.012664176430553198, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.58} +{'loss': 0.0143, 'grad_norm': 0.08135362714529037, 'learning_rate': 4.684880664776199e-06, 'flow_matching_loss': 0.014306690264493226, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.58} +{'loss': 0.0143, 'grad_norm': 0.07090885937213898, 'learning_rate': 4.6728824288035784e-06, 'flow_matching_loss': 0.014345387928187847, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.59} +{'loss': 0.0142, 'grad_norm': 0.07416519522666931, 'learning_rate': 4.660896741252586e-06, 'flow_matching_loss': 0.014230594597756862, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.59} +{'loss': 0.0141, 'grad_norm': 0.07003627717494965, 'learning_rate': 4.6489236166869714e-06, 'flow_matching_loss': 0.014115734305232764, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.59} +{'loss': 0.0117, 'grad_norm': 0.07473664730787277, 'learning_rate': 4.6369630696552e-06, 'flow_matching_loss': 0.01167108230292797, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.6} +{'loss': 0.0133, 'grad_norm': 0.06458549946546555, 'learning_rate': 4.625015114690474e-06, 'flow_matching_loss': 0.013322941493242979, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.6} +{'loss': 0.0125, 'grad_norm': 0.06708625704050064, 'learning_rate': 4.613079766310682e-06, 'flow_matching_loss': 0.012455717101693154, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.6} +{'loss': 0.0128, 'grad_norm': 0.06664470583200455, 'learning_rate': 4.601157039018396e-06, 'flow_matching_loss': 0.012809778098016977, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.6} +{'loss': 0.0119, 'grad_norm': 0.079779714345932, 'learning_rate': 4.589246947300849e-06, 'flow_matching_loss': 0.011927387677133083, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.61} +{'loss': 0.0126, 'grad_norm': 0.06724656373262405, 'learning_rate': 4.577349505629936e-06, 'flow_matching_loss': 0.012552298791706561, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.61} +{'loss': 0.0138, 'grad_norm': 0.08309617638587952, 'learning_rate': 4.565464728462165e-06, 'flow_matching_loss': 0.013801949005573988, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.61} +{'loss': 0.0142, 'grad_norm': 0.07396794855594635, 'learning_rate': 4.553592630238663e-06, 'flow_matching_loss': 0.014249459188431502, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.62} +{'loss': 0.0128, 'grad_norm': 0.08411047607660294, 'learning_rate': 4.541733225385146e-06, 'flow_matching_loss': 0.012820951733738184, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.62} +{'loss': 0.0126, 'grad_norm': 0.09513940662145615, 'learning_rate': 4.529886528311905e-06, 'flow_matching_loss': 0.012647601962089538, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.62} +{'loss': 0.0139, 'grad_norm': 0.08614011108875275, 'learning_rate': 4.5180525534138045e-06, 'flow_matching_loss': 0.013887646794319152, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.62} +{'loss': 0.0144, 'grad_norm': 0.07358040660619736, 'learning_rate': 4.506231315070235e-06, 'flow_matching_loss': 0.0144280762411654, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.63} +{'loss': 0.0132, 'grad_norm': 0.07864868640899658, 'learning_rate': 4.494422827645115e-06, 'flow_matching_loss': 0.013163793366402388, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.63} 76%|███████▌ | 22790/30000 [14:02:10<2:51:17, 1.43s/it] 76%|███████▌ | 22791/30000 [14:02:12<2:41:53, 1.35s/it] 76%|███████▌ | 22792/30000 [14:02:13<2:36:11, 1.30s/it] 76%|███████▌ | 22793/30000 [14:02:19<5:29:28, 2.74s/it] 76%|███████▌ | 22794/30000 [14:02:20<4:33:20, 2.28s/it] 76%|███████▌ | 22795/30000 [14:02:21<3:54:04, 1.95s/it] 76%|███████▌ | 22796/30000 [14:02:22<3:28:00, 1.73s/it] 76%|███████▌ | 22797/30000 [14:02:24<3:09:20, 1.58s/it] 76%|███████▌ | 22798/30000 [14:02:25<2:54:54, 1.46s/it] 76%|███████▌ | 22799/30000 [14:02:26<2:45:02, 1.38s/it] 76%|███████▌ | 22800/30000 [14:02:27<2:38:06, 1.32s/it] 76%|███████▌ | 22800/30000 [14:02:27<2:38:06, 1.32s/it] 76%|███████▌ | 22801/30000 [14:02:33<5:12:25, 2.60s/it] 76%|███████▌ | 22802/30000 [14:02:34<4:21:07, 2.18s/it] 76%|███████▌ | 22803/30000 [14:02:35<3:45:37, 1.88s/it] 76%|███████▌ | 22804/30000 [14:02:36<3:21:02, 1.68s/it] 76%|███████▌ | 22805/30000 [14:02:38<3:03:19, 1.53s/it] 76%|███████▌ | 22806/30000 [14:02:39<2:50:10, 1.42s/it] 76%|███████▌ | 22807/30000 [14:02:40<2:40:55, 1.34s/it] 76%|███████▌ | 22808/30000 [14:02:41<2:34:53, 1.29s/it] 76%|███████▌ | 22809/30000 [14:02:47<5:18:10, 2.65s/it] 76%|███████▌ | 22810/30000 [14:02:48<4:26:09, 2.22s/it] 76%|███████▌ | 22810/30000 [14:02:48<4:26:09, 2.22s/it] 76%|███████▌ | 22811/30000 [14:02:49<3:50:25, 1.92s/it] 76%|███████▌ | 22812/30000 [14:02:51<3:24:24, 1.71s/it] 76%|███████▌ | 22813/30000 [14:02:52<3:05:23, 1.55s/it] 76%|███████▌ | 22814/30000 [14:02:53<2:52:02, 1.44s/it] 76%|███████▌ | 22815/30000 [14:02:54<2:42:35, 1.36s/it] 76%|███████▌ | 22816/30000 [14:02:55<2:36:21, 1.31s/it] 76%|███████▌ | 22817/30000 [14:03:01<5:19:15, 2.67s/it] 76%|███████▌ | 22818/30000 [14:03:02<4:25:11, 2.22s/it] 76%|███████▌ | 22819/30000 [14:03:03<3:48:28, 1.91s/it] 76%|███████▌ | 22820/30000 [14:03:05<3:22:29, 1.69s/it] 76%|███████▌ | 22820/30000 [14:03:05<3:22:29, 1.69s/it] 76%|███████▌ | 22821/30000 [14:03:06<3:04:03, 1.54s/it] 76%|███████▌ | 22822/30000 [14:03:07<2:51:48, 1.44s/it] 76%|███████▌ | 22823/30000 [14:03:08<2:42:32, 1.36s/it] 76%|███████▌ | 22824/30000 [14:03:09<2:36:40, 1.31s/it] 76%|███████▌ | 22825/30000 [14:03:15<5:23:33, 2.71s/it] 76%|███████▌ | 22826/30000 [14:03:17<4:29:08, 2.25s/it] 76%|███████▌ | 22827/30000 [14:03:18<3:50:50, 1.93s/it] 76%|███████▌ | 22828/30000 [14:03:19<3:23:48, 1.71s/it] 76%|███████▌ | 22829/30000 [14:03:20<3:05:01, 1.55s/it] 76%|███████▌ | 22830/30000 [14:03:21<2:51:32, 1.44s/it] 76%|███████▌ | 22830/30000 [14:03:21<2:51:32, 1.44s/it] 76%|███████▌ | 22831/30000 [14:03:22<2:41:56, 1.36s/it] 76%|███████▌ | 22832/30000 [14:03:24<2:35:55, 1.31s/it] 76%|███████▌ | 22833/30000 [14:03:30<5:36:12, 2.81s/it] 76%|███████▌ | 22834/30000 [14:03:31<4:37:32, 2.32s/it] 76%|███████▌ | 22835/30000 [14:03:32<3:56:35, 1.98s/it] 76%|███████▌ | 22836/30000 [14:03:34<3:28:06, 1.74s/it] 76%|███████▌ | 22837/30000 [14:03:35<3:07:50, 1.57s/it] 76%|███████▌ | 22838/30000 [14:03:36<2:54:06, 1.46s/it] 76%|███████▌ | 22839/30000 [14:03:37<2:44:08, 1.38s/it] 76%|███████▌ | 22840/30000 [14:03:38<2:37:18, 1.32s/it] 76%|███████▌ | 22840/30000 [14:03:38<2:37:18, 1.32s/it] 76%|███████▌ | 22841/30000 [14:03:44<5:16:38, 2.65s/it] 76%|███████▌ | 22842/30000 [14:03:45<4:23:10, 2.21s/it] 76%|███████▌ | 22843/30000 [14:03:46<3:46:02, 1.90s/it] 76%|███████▌ | 22844/30000 [14:03:48<3:21:39, 1.69s/it] 76%|███████▌ | 22845/30000 [14:03:49<3:03:01, 1.53s/it] 76%|███████▌ | 22846/30000 [14:03:50<2:50:24, 1.43s/it] 76%|███████▌ | 22847/30000 [14:03:51<2:41:00, 1.35s/it] 76%|███████▌ | 22848/30000 [14:03:52<2:35:37, 1.31s/it] 76%|███████▌ | 22849/30000 [14:03:59<5:34:16, 2.80s/it] 76%|███████▌ | 22850/30000 [14:04:00<4:36:12, 2.32s/it] 76%|███████▌ | 22850/30000 [14:04:00<4:36:12, 2.32s/it] 76%|███████▌ | 22851/30000 [14:04:01<3:56:40, 1.99s/it] 76%|███████▌ | 22852/30000 [14:04:02<3:27:27, 1.74s/it] 76%|███████▌ | 22853/30000 [14:04:03<3:07:03, 1.57s/it] 76%|███████▌ | 22854/30000 [14:04:05<2:53:44, 1.46s/it] 76%|███████▌ | 22855/30000 [14:04:06<2:43:19, 1.37s/it] 76%|███████▌ | 22856/30000 [14:04:07<2:36:49, 1.32s/it] 76%|███████▌ | 22857/30000 [14:04:13<5:36:55, 2.83s/it] 76%|███████▌ | 22858/30000 [14:04:14<4:37:32, 2.33s/it] 76%|███████▌ | 22859/30000 [14:04:16<3:56:57, 1.99s/it] 76%|███████▌ | 22860/30000 [14:04:17<3:28:33, 1.75s/it] 76%|███████▌ | 22860/30000 [14:04:17<3:28:33, 1.75s/it] 76%|███████▌ | 22861/30000 [14:04:18<3:08:00, 1.58s/it] 76%|███████▌ | 22862/30000 [14:04:19<2:54:00, 1.46s/it] 76%|███████▌ | 22863/30000 [14:04:20<2:44:03, 1.38s/it] 76%|███████▌ | 22864/30000 [14:04:22<2:38:49, 1.34s/it] 76%|███████▌ | 22865/30000 [14:04:28<5:28:55, 2.77s/it] 76%|███████▌ | 22866/30000 [14:04:29<4:32:10, 2.29s/it] 76%|███████▌ | 22867/30000 [14:04:30<3:53:01, 1.96s/it] 76%|███████▌ | 22868/30000 [14:04:31<3:24:31, 1.72s/it] 76%|███████▌ | 22869/30000 [14:04:32<3:05:03, 1.56s/it] 76%|███████▌ | 22870/30000 [14:04:34<2:52:25, 1.45s/it] 76%|███████▌ | 22870/30000 [14:04:34<2:52:25, 1.45s/it] 76%|███████▌ | 22871/30000 [14:04:35<2:43:40, 1.38s/it] 76%|███████▌ | 22872/30000 [14:04:36<2:37:17, 1.32s/it] 76%|███████▌ | 22873/30000 [14:04:42<5:35:11, 2.82s/it] 76%|███████▌ | 22874/30000 [14:04:43<4:36:18, 2.33s/it] 76%|███████▋ | 22875/30000 [14:04:45<3:54:57, 1.98s/it] 76%|███████▋ | 22876/30000 [14:04:46<3:26:44, 1.74s/it] 76%|███████▋ | 22877/30000 [14:04:47<3:06:42, 1.57s/it] 76%|███████▋ | 22878/30000 [14:04:48<2:53:16, 1.46s/it] 76%|███████▋ | 22879/30000 [14:04:49<2:43:26, 1.38s/it] 76%|███████▋ | 22880/30000 [14:04:51<2:38:26, 1.34s/it] 76%|███████▋ | 22880/30000 [14:04:51<2:38:26, 1.34s/it] 76%|███████▋ | 22881/30000 [14:04:57<5:34:42, 2.82s/it] 76%|███████▋ | 22882/30000 [14:04:58<4:35:56, 2.33s/it] 76%|███████▋ | 22883/30000 [14:04:59<3:55:50, 1.99s/it] 76%|███████▋ | 22884/30000 [14:05:00<3:26:54, 1.74s/it] 76%|███████▋ | 22885/30000 [14:05:02<3:07:36, 1.58s/it] 76%|███████▋ | 22886/30000 [14:05:03<2:53:35, 1.46s/it] 76%|███████▋ | 22887/30000 [14:05:04<2:43:29, 1.38s/it] 76%|███████▋ | 22888/30000 [14:05:05<2:37:36, 1.33s/it] 76%|███████▋ | 22889/30000 [14:05:11<5:14:46, 2.66s/it] 76%|███████▋ | 22890/30000 [14:05:12<4:21:48, 2.21s/it] 76%|███████▋ | 22890/30000 [14:05:12<4:21:48, 2.21s/it] 76%|███████▋ | 22891/30000 [14:05:13<3:46:07, 1.91s/it] 76%|███████▋ | 22892/30000 [14:05:15<3:20:17, 1.69s/it] 76%|███████▋ | 22893/30000 [14:05:16<3:02:27, 1.54s/it] 76%|███████▋ | 22894/30000 [14:05:17<2:49:22, 1.43s/it] 76%|███████▋ | 22895/30000 [14:05:18<2:40:00, 1.35s/it] 76%|███████▋ | 22896/30000 [14:05:19<2:34:50, 1.31s/it] 76%|███████▋ | 22897/30000 [14:05:26<5:35:38, 2.84s/it] 76%|███████▋ | 22898/30000 [14:05:27<4:36:39, 2.34s/it] 76%|███████▋ | 22899/30000 [14:05:28<3:56:30, 2.00s/it] 76%|███████▋ | 22900/30000 [14:05:29<3:28:16, 1.76s/it] 76%|███████▋ | 22900/30000 [14:05:29<3:28:16, 1.76s/it] 76%|███████▋ | 22901/30000 [14:05:30<3:07:19, 1.58s/it] 76%|███████▋ | 22902/30000 [14:05:32<2:53:29, 1.47s/it] 76%|███████▋ | 22903/30000 [14:05:33<2:42:33, 1.37s/it] 76%|███████▋ | 22904/30000 [14:05:34<2:36:06, 1.32s/it] 76%|███████▋ | 22905/30000 [14:05:40<5:18:15, 2.69s/it] 76%|███████▋ | 22906/30000 [14:05:41<4:24:40, 2.24s/it] 76%|███████▋ | 22907/30000 [14:05:42<3:47:15, 1.92s/it] 76%|███████▋ | 22908/30000 [14:05:43<3:21:24, 1.70s/it] 76%|███████▋ | 22909/30000 [14:05:45<3:03:08, 1.55s/it] 76%|███████▋ | 22910/30000 [14:05:46<2:51:21, 1.45s/it] 76%|███████▋ | 22910/30000 [14:05:46<2:51:21, 1.45s/it] 76%|███████▋ | 22911/30000 [14:05:47<2:43:07, 1.38s/it] 76%|███████▋ | 22912/30000 [14:05:48<2:37:10, 1.33s/it] 76%|███████▋ | 22913/30000 [14:05:54<5:22:07, 2.73s/it] 76%|███████▋ | 22914/30000 [14:05:55<4:27:53, 2.27s/it] 76%|███████▋ | 22915/30000 [14:05:57<3:49:21, 1.94s/it] 76%|███████▋ | 22916/30000 [14:05:58<3:22:15, 1.71s/it] 76%|███████▋ | 22917/30000 [14:05:59<3:04:50, 1.57s/it] 76%|███████▋ | 22918/30000 [14:06:00<2:52:09, 1.46s/it] 76%|███████▋ | 22919/30000 [14:06:01<2:41:49, 1.37s/it] 76%|███████▋ | 22920/30000 [14:06:03<2:35:29, 1.32s/it] 76%|███████▋ | 22920/30000 [14:06:03<2:35:29, 1.32s/it] 76%|███████▋ | 22921/30000 [14:06:09<5:47:27, 2.94s/it] 76%|███████▋ | 22922/30000 [14:06:11<4:44:44, 2.41s/it] 76%|███████▋ | 22923/30000 [14:06:12<4:01:30, 2.05s/it] 76%|███████▋ | 22924/30000 [14:06:13<3:31:36, 1.79s/it] 76%|███████▋ | 22925/30000 [14:06:14<3:10:07, 1.61s/it] 76%|███████▋ | 22926/30000 [14:06:15<2:54:27, 1.48s/it] 76%|███████▋ | 22927/30000 [14:06:16<2:43:46, 1.39s/it] 76%|███████▋ | 22928/30000 [14:06:18<2:36:49, 1.33s/it] 76%|███████▋ | 22929/30000 [14:06:24<5:23:59, 2.75s/it] 76%|███████▋ | 22930/30000 [14:06:25<4:28:28, 2.28s/it] 76%|███████▋ | 22930/30000 [14:06:25<4:28:28, 2.28s/it] 76%|███████▋ | 22931/30000 [14:06:26<3:50:23, 1.96s/it] 76%|███████▋ | 22932/30000 [14:06:27<3:23:46, 1.73s/it] 76%|███████▋ | 22933/30000 [14:06:29<3:04:40, 1.57s/it] 76%|███████▋ | 22934/30000 [14:06:30<2:51:07, 1.45s/it] 76%|███████▋ | 22935/30000 [14:06:31<2:41:08, 1.37s/it] 76%|███████▋ | 22936/30000 [14:06:32<2:34:55, 1.32s/it] 76%|███████▋ | 22937/30000 [14:06:39<5:49:35, 2.97s/it] 76%|███████▋ | 22938/30000 [14:06:40<4:45:57, 2.43s/it] 76%|███████▋ | 22939/30000 [14:06:41<4:02:16, 2.06s/it] 76%|███████▋ | 22940/30000 [14:06:42<3:31:19, 1.80s/it] 76%|███████▋ | 22940/30000 [14:06:42<3:31:19, 1.80s/it] 76%|███████▋ | 22941/30000 [14:06:44<3:10:01, 1.62s/it] 76%|███████▋ | 22942/30000 [14:06:45<2:56:15, 1.50s/it] 76%|███████▋ | 22943/30000 [14:06:46<2:45:14, 1.40s/it] 76%|███████▋ | 22944/30000 [14:06:47<2:38:26, 1.35s/it] 76%|███████▋ | 22945/30000 [14:06:54<5:51:39, 2.99s/it] 76%|███████▋ | 22946/30000 [14:06:55<4:47:34, 2.45s/it] 76%|███████▋ | 22947/30000 [14:06:56<4:03:08, 2.07s/it] 76%|███████▋ | 22948/30000 [14:06:58<3:32:10, 1.81s/it] 76%|███████▋ | 22949/30000 [14:06:59<3:10:25, 1.62s/it] 76%|███████▋ | 22950/30000 [14:07:00<2:55:16, 1.49s/it] 76%|███████▋ | 22950/30000 [14:07:00<2:55:16, 1.49s/it] 77%|███████▋ | 22951/30000 [14:07:01<2:44:33, 1.40s/it] 77%|███████▋ | 22952/30000 [14:07:02<2:37:06, 1.34s/it] 77%|███████▋ | 22953/30000 [14:07:09<5:26:15, 2.78s/it] 77%|███████▋ | 22954/30000 [14:07:10<4:29:50, 2.30s/it] 77%|███████▋ | 22955/30000 [14:07:11<3:51:20, 1.97s/it] 77%|███████▋ | 22956/30000 [14:07:12<3:23:34, 1.73s/it] 77%|███████▋ | 22957/30000 [14:07:13<3:04:48, 1.57s/it] 77%|███████▋ | 22958/30000 [14:07:14<2:51:14, 1.46s/it] 77%|███████▋ | 22959/30000 [14:07:16<2:41:01, 1.37s/it] 77%|███████▋ | 22960/30000 [14:07:17<2:34:30, 1.32s/it] 77%|███████▋ | 22960/30000 [14:07:17<2:34:30, 1.32s/it] 77%|███████▋ | 22961/30000 [14:07:23<5:23:24, 2.76s/it] 77%|███████▋ | 22962/30000 [14:07:24<4:27:12, 2.28s/it] 77%|███████▋ | 22963/30000 [14:07:25<3:48:38, 1.95s/it] 77%|███████▋ | 22964/30000 [14:07:26<3:21:50, 1.72s/it] 77%|███████▋ | 22965/30000 [14:07:28<3:02:57, 1.56s/it] 77%|███████▋ | 22966/30000 [14:07:29<2:49:55, 1.45s/it] 77%|███████▋ | 22967/30000 [14:07:30<2:40:05, 1.37s/it] 77%|███████▋ | 22968/30000 [14:07:31<2:34:23, 1.32s/it] 77%|███████▋ | 22969/30000 [14:07:38<5:59:02, 3.06s/it] 77%|███████▋ | 22970/30000 [14:07:40<4:52:29, 2.50s/it] 77%|███████▋ | 22970/30000 [14:07:40<4:52:29, 2.50s/it] 77%|███████▋ | 22971/30000 [14:07:41<4:06:25, 2.10s/it] 77%|███████▋ | 22972/30000 [14:07:42<3:34:02, 1.83s/it] 77%|███████▋ | 22973/30000 [14:07:43<3:11:50, 1.64s/it] 77%|███████▋ | 22974/30000 [14:07:44<2:55:00, 1.49s/it] 77%|███████▋ | 22975/30000 [14:07:45<2:43:21, 1.40s/it] 77%|███████▋ | 22976/30000 [14:07:47<2:36:32, 1.34s/it] 77%|███████▋ | 22977/30000 [14:07:54<5:59:50, 3.07s/it] 77%|███████▋ | 22978/30000 [14:07:55<4:52:44, 2.50s/it] 77%|███████▋ | 22979/30000 [14:07:56<4:06:28, 2.11s/it] 77%|███████▋ | 22980/30000 [14:07:57<3:33:57, 1.83s/it] 77%|███████▋ | 22980/30000 [14:07:57<3:33:57, 1.83s/it] 77%|███████▋ | 22981/30000 [14:07:59<3:12:06, 1.64s/it] 77%|███████▋ | 22982/30000 [14:08:00<2:55:23, 1.50s/it] 77%|███████▋ | 22983/30000 [14:08:01<2:43:56, 1.40s/it] 77%|███████▋ | 22984/30000 [14:08:02<2:35:55, 1.33s/it] 77%|███████▋ | 22985/30000 [14:08:08<5:23:59, 2.77s/it] 77%|███████▋ | 22986/30000 [14:08:09<4:27:59, 2.29s/it] 77%|███████▋ | 22987/30000 [14:08:11<3:49:24, 1.96s/it] 77%|███████▋ | 22988/30000 [14:08:12<3:22:06, 1.73s/it] 77%|███████▋ | 22989/30000 [14:08:13<3:02:16, 1.56s/it] 77%|███████▋ | 22990/30000 [14:08:14<2:48:30, 1.44s/it] 77%|███████▋ | 22990/30000 [14:08:14<2:48:30, 1.44s/it] 77%|███████▋ | 22991/30000 [14:08:15<2:39:11, 1.36s/it] 77%|███████▋ | 22992/30000 [14:08:16<2:33:36, 1.32s/it] 77%|███████▋ | 22993/30000 [14:08:23<5:36:05, 2.88s/it] 77%|███████▋ | 22994/30000 [14:08:24<4:36:01, 2.36s/it] 77%|███████▋ | 22995/30000 [14:08:25<3:54:23, 2.01s/it] 77%|███████▋ | 22996/30000 [14:08:26<3:25:33, 1.76s/it] 77%|███████▋ | 22997/30000 [14:08:28<3:05:14, 1.59s/it] 77%|███████▋ | 22998/30000 [14:08:29<2:51:12, 1.47s/it] 77%|███████▋ | 22999/30000 [14:08:30<2:41:00, 1.38s/it] 77%|███████▋ | 23000/30000 [14:08:31<2:34:21, 1.32s/it] 77%|███████▋ | 23000/30000 [14:08:31<2:34:21, 1.32s/it] 77%|███████▋ | 23001/30000 [14:08:38<5:41:24, 2.93s/it] 77%|███████▋ | 23002/30000 [14:08:39<4:40:00, 2.40s/it] 77%|███████▋ | 23003/30000 [14:08:40<3:57:55, 2.04s/it] 77%|███████▋ | 23004/30000 [14:08:41<3:28:26, 1.79s/it] 77%|███████▋ | 23005/30000 [14:08:43<3:08:18, 1.62s/it] 77%|███████▋ | 23006/30000 [14:08:44<2:52:35, 1.48s/it] 77%|███████▋ | 23007/30000 [14:08:45<2:42:07, 1.39s/it] 77%|███████▋ | 23008/30000 [14:08:46<2:35:14, 1.33s/it] 77%|███████▋ | 23009/30000 [14:08:52<5:12:07, 2.68s/it] 77%|███████▋ | 23010/30000 [14:08:53<4:19:08, 2.22s/it] 77%|███████▋ | 23010/30000 [14:08:53<4:19:08, 2.22s/it] 77%|███████▋ | 23011/30000 [14:08:54<3:43:08, 1.92s/it] 77%|███████▋ | 23012/30000 [14:08:56<3:17:53, 1.70s/it] 77%|███████▋ | 23013/30000 [14:08:57<3:00:19, 1.55s/it] 77%|███████▋ | 23014/30000 [14:08:58<2:48:24, 1.45s/it] 77%|███████▋ | 23015/30000 [14:08:59<2:39:48, 1.37s/it] 77%|███████▋ | 23016/30000 [14:09:00<2:33:13, 1.32s/it] 77%|███████▋ | 23017/30000 [14:09:06<5:05:48, 2.63s/it] 77%|███████▋ | 23018/30000 [14:09:07<4:14:08, 2.18s/it] 77%|███████▋ | 23019/30000 [14:09:08<3:39:35, 1.89s/it] 77%|███████▋ | 23020/30000 [14:09:10<3:16:03, 1.69s/it] 77%|███████▋ | 23020/30000 [14:09:10<3:16:03, 1.69s/it] 77%|███████▋ | 23021/30000 [14:09:11<2:58:29, 1.53s/it] 77%|███████▋ | 23022/30000 [14:09:12<2:46:51, 1.43s/it] 77%|███████▋ | 23023/30000 [14:09:13<2:38:57, 1.37s/it] 77%|███████▋ | 23024/30000 [14:09:14<2:32:47, 1.31s/it] 77%|███████▋ | 23025/30000 [14:09:20<5:13:47, 2.70s/it] 77%|███████▋ | 23026/30000 [14:09:22<4:20:29, 2.24s/it] 77%|███████▋ | 23027/30000 [14:09:23<3:43:25, 1.92s/it] 77%|███████▋ | 23028/30000 [14:09:24<3:17:06, 1.70s/it] 77%|███████▋ | 23029/30000 [14:09:25<2:58:18, 1.53s/it] 77%|███████▋ | 23030/30000 [14:09:26<2:46:55, 1.44s/it] 77%|███████▋ | 23030/30000 [14:09:26<2:46:55, 1.44s/it] 77%|███████▋ | 23031/30000 [14:09:27<2:39:22, 1.37s/it] 77%|███████▋ | 23032/30000 [14:09:29<2:32:46, 1.32s/it] 77%|███████▋ | 23033/30000 [14:09:35<5:27:21, 2.82s/it] 77%|███████▋ | 23034/30000 [14:09:36<4:29:57, 2.33s/it] 77%|███████▋ | 23035/30000 [14:09:37<3:50:29, 1.99s/it] 77%|███████▋ | 23036/30000 [14:09:39<3:22:46, 1.75s/it] 77%|███████▋ | 23037/30000 [14:09:40<3:03:26, 1.58s/it] 77%|███████▋ | 23038/30000 [14:09:41<2:49:54, 1.46s/it] 77%|███████▋ | 23039/30000 [14:09:42<2:39:18, 1.37s/it] 77%|███████▋ | 23040/30000 [14:09:43<2:32:21, 1.31s/it] 77%|███████▋ | 23040/30000 [14:09:43<2:32:21, 1.31s/it] 77%|███████▋ | 23041/30000 [14:09:49<5:01:20, 2.60s/it] 77%|███████▋ | 23042/30000 [14:09:50<4:11:04, 2.17s/it] 77%|███████▋ | 23043/30000 [14:09:51<3:37:10, 1.87s/it] 77%|███████▋ | 23044/30000 [14:09:52<3:13:19, 1.67s/it] 77%|███████▋ | 23045/30000 [14:09:54<2:55:26, 1.51s/it] 77%|███████▋ | 23046/30000 [14:09:55<2:43:41, 1.41s/it] 77%|███████▋ | 23047/30000 [14:09:56<2:35:15, 1.34s/it] 77%|███████▋ | 23048/30000 [14:09:57<2:29:27, 1.29s/it] 77%|███████▋ | 23049/30000 [14:10:03<5:02:53, 2.61s/it] 77%|███████▋ | 23050/30000 [14:10:04<4:12:44, 2.18s/it] 77%|███████▋ | 23050/30000 [14:10:04<4:12:44, 2.18s/it] 77%|███████▋ | 23051/30000 [14:10:05<3:38:03, 1.88s/it] 77%|███████▋ | 23052/30000 [14:10:06<3:13:25, 1.67s/it] 77%|███████▋ | 23053/30000 [14:10:07<2:56:32, 1.52s/it] 77%|███████▋ | 23054/30000 [14:10:09<2:45:06, 1.43s/it] 77%|███████▋ | 23055/30000 [14:10:10<2:36:18, 1.35s/it] 77%|███████▋ | 23056/30000 [14:10:11<2:31:52, 1.31s/it] 77%|███████▋ | 23057/30000 [14:10:16<4:54:15, 2.54s/it] 77%|███████▋ | 23058/30000 [14:10:18<4:06:44, 2.13s/it] 77%|███████▋ | 23059/30000 [14:10:19<3:34:00, 1.85s/it] 77%|███████▋ | 23060/30000 [14:10:20<3:10:41, 1.65s/it] 77%|███████▋ | 23060/30000 [14:10:20<3:10:41, 1.65s/it] 77%|███████▋ | 23061/30000 [14:10:21<2:55:06, 1.51s/it] 77%|███████▋ | 23062/30000 [14:10:22<2:43:22, 1.41s/it] 77%|███████▋ | 23063/30000 [14:10:24<2:35:20, 1.34s/it] 77%|███████▋ | 23064/30000 [14:10:25<2:29:49, 1.30s/it] 77%|███████▋ | 23065/30000 [14:10:31<5:10:10, 2.68s/it] 77%|███████▋ | 23066/30000 [14:10:32<4:18:30, 2.24s/it] 77%|███████▋ | 23067/30000 [14:10:33<3:41:34, 1.92s/it] 77%|███████▋ | 23068/30000 [14:10:34<3:16:12, 1.70s/it] 77%|███████▋ | 23069/30000 [14:10:35<2:58:30, 1.55s/it] 77%|███████▋ | 23070/30000 [14:10:37<2:46:00, 1.44s/it] 77%|███████▋ | 23070/30000 [14:10:37<2:46:00, 1.44s/it] 77%|███████▋ | 23071/30000 [14:10:38<2:36:56, 1.36s/it] 77%|███████▋ | 23072/30000 [14:10:39<2:29:58, 1.30s/it] 77%|███████▋ | 23073/30000 [14:10:45<5:15:53, 2.74s/it] 77%|███████▋ | 23074/30000 [14:10:46<4:21:51, 2.27s/it] 77%|███████▋ | 23075/30000 [14:10:47<3:44:27, 1.94s/it] 77%|███████▋ | 23076/30000 [14:10:49<3:17:55, 1.72s/it] 77%|███████▋ | 23077/30000 [14:10:50<2:59:49, 1.56s/it] 77%|███████▋ | 23078/30000 [14:10:51<2:47:04, 1.45s/it] 77%|███████▋ | 23079/30000 [14:10:52<2:37:26, 1.36s/it] 77%|███████▋ | 23080/30000 [14:10:53<2:31:14, 1.31s/it] 77%|███████▋ | 23080/30000 [14:10:53<2:31:14, 1.31s/it] 77%|███████▋ | 23081/30000 [14:10:59<5:16:48, 2.75s/it] 77%|███████▋ | 23082/30000 [14:11:01<4:22:14, 2.27s/it] 77%|███████▋ | 23083/30000 [14:11:02<3:44:02, 1.94s/it] 77%|███████▋ | 23084/30000 [14:11:03<3:17:48, 1.72s/it] 77%|███████▋ | 23085/30000 [14:11:04<2:59:39, 1.56s/it] 77%|███████▋ | 23086/30000 [14:11:05<2:46:47, 1.45s/it] 77%|███████▋ | 23087/30000 [14:11:06<2:37:09, 1.36s/it] 77%|███████▋ | 23088/30000 [14:11:08<2:31:18, 1.31s/it] 77%|███████▋ | 23089/30000 [14:11:14<5:24:13, 2.81s/it] 77%|███████▋ | 23090/30000 [14:11:15<4:27:44, 2.32s/it] 77%|███████▋ | 23090/30000 [14:11:15<4:27:44, 2.32s/it] 77%|███████▋ | 23091/30000 [14:11:16<3:48:02, 1.98s/it] 77%|███████▋ | 23092/30000 [14:11:18<3:20:23, 1.74s/it] 77%|███████▋ | 23093/30000 [14:11:19<3:00:32, 1.57s/it] 77%|███████▋ | 23094/30000 [14:11:20<2:47:11, 1.45s/it] 77%|███████▋ | 23095/30000 [14:11:21<2:37:15, 1.37s/it] 77%|███████▋ | 23096/30000 [14:11:22<2:31:30, 1.32s/it] 77%|███████▋ | 23097/30000 [14:11:28<5:04:40, 2.65s/it] 77%|███████▋ | 23098/30000 [14:11:29<4:14:21, 2.21s/it] 77%|███████▋ | 23099/30000 [14:11:30<3:39:03, 1.90s/it] 77%|███████▋ | 23100/30000 [14:11:32<3:14:16, 1.69s/it] 77%|███████▋ | 23100/30000 [14:11:32<3:14:16, 1.69s/it] 77%|███████▋ | 23101/30000 [14:11:33<2:57:08, 1.54s/it] 77%|███████▋ | 23102/30000 [14:11:34<2:45:05, 1.44s/it] 77%|███████▋ | 23103/30000 [14:11:35<2:36:11, 1.36s/it] 77%|███████▋ | 23104/30000 [14:11:36<2:30:01, 1.31s/it] 77%|███████▋ | 23105/30000 [14:11:42<5:12:36, 2.72s/it] 77%|███████▋ | 23106/30000 [14:11:44<4:19:34, 2.26s/it] 77%|███████▋ | 23107/30000 [14:11:45<3:42:24, 1.94s/it] 77%|███████▋ | 23108/30000 [14:11:46<3:16:26, 1.71s/it] 77%|███████▋ | 23109/30000 [14:11:47<2:58:57, 1.56s/it] 77%|███████▋ | 23110/30000 [14:11:48<2:46:04, 1.45s/it] 77%|███████▋ | 23110/30000 [14:11:48<2:46:04, 1.45s/it] 77%|███████▋ | 23111/30000 [14:11:49<2:36:08, 1.36s/it] 77%|███████▋ | 23112/30000 [14:11:51<2:30:27, 1.31s/it] 77%|███████▋ | 23113/30000 [14:11:57<5:16:37, 2.76s/it] 77%|███████▋ | 23114/30000 [14:11:58<4:21:51, 2.28s/it] 77%|███████▋ | 23115/30000 [14:11:59<3:43:38, 1.95s/it] 77%|███████▋ | 23116/30000 [14:12:00<3:17:22, 1.72s/it] 77%|███████▋ | 23117/30000 [14:12:01<2:59:19, 1.56s/it] 77%|███████▋ | 23118/30000 [14:12:03<2:47:11, 1.46s/it] 77%|███████▋ | 23119/30000 [14:12:04<2:37:32, 1.37s/it] 77%|███████▋ | 23120/30000 [14:12:05<2:31:06, 1.32s/it] 77%|███████▋ | 23120/30000 [14:12:05<2:31:06, 1.32s/it] 77%|███████▋ | 23121/30000 [14:12:11<5:25:47, 2.84s/it] 77%|███████▋ | 23122/30000 [14:12:13<4:28:23, 2.34s/it] 77%|███████▋ | 23123/30000 [14:12:14<3:48:49, 2.00s/it] 77%|███████▋ | 23124/30000 [14:12:15<3:20:02, 1.75s/it] 77%|███████▋ | 23125/30000 [14:12:16<3:00:58, 1.58s/it] 77%|███████▋ | 23126/30000 [14:12:17<2:48:25, 1.47s/it] 77%|███████▋ | 23127/30000 [14:12:19<2:37:56, 1.38s/it] 77%|███████▋ | 23128/30000 [14:12:20<2:31:43, 1.32s/it] 77%|███████▋ | 23129/30000 [14:12:26<5:29:33, 2.88s/it] 77%|███████▋ | 23130/30000 [14:12:27<4:31:22, 2.37s/it] 77%|███████▋ | 23130/30000 [14:12:27<4:31:22, 2.37s/it] 77%|███████▋ | 23131/30000 [14:12:29<3:51:33, 2.02s/it] 77%|███████▋ | 23132/30000 [14:12:30<3:22:26, 1.77s/it] 77%|███████▋ | 23133/30000 [14:12:31<3:01:49, 1.59s/it] 77%|███████▋ | 23134/30000 [14:12:32<2:48:03, 1.47s/it] 77%|███████▋ | 23135/30000 [14:12:33<2:37:59, 1.38s/it] 77%|███████▋ | 23136/30000 [14:12:35<2:31:48, 1.33s/it] 77%|███████▋ | 23137/30000 [14:12:41<5:23:06, 2.82s/it] 77%|███████▋ | 23138/30000 [14:12:42<4:27:39, 2.34s/it] 77%|███████▋ | 23139/30000 [14:12:43<3:47:21, 1.99s/it] 77%|███████▋ | 23140/30000 [14:12:44<3:19:49, 1.75s/it] 77%|███████▋ | 23140/30000 [14:12:44<3:19:49, 1.75s/it] 77%|███████▋ | 23141/30000 [14:12:46<3:01:23, 1.59s/it] 77%|███████▋ | 23142/30000 [14:12:47<2:47:44, 1.47s/it] 77%|███████▋ | 23143/30000 [14:12:48<2:38:45, 1.39s/it] 77%|███████▋ | 23144/30000 [14:12:49<2:31:54, 1.33s/it] 77%|███████▋ | 23145/30000 [14:12:55<5:03:14, 2.65s/it] 77%|███████▋ | 23146/30000 [14:12:56<4:12:26, 2.21s/it] 77%|███████▋ | 23147/30000 [14:12:57<3:37:34, 1.90s/it] 77%|███████▋ | 23148/30000 [14:12:59<3:13:01, 1.69s/it] 77%|███████▋ | 23149/30000 [14:13:00<2:55:46, 1.54s/it] 77%|███████▋ | 23150/30000 [14:13:01<2:44:06, 1.44s/it] 77%|███████▋ | 23150/30000 [14:13:01<2:44:06, 1.44s/it] 77%|███████▋ | 23151/30000 [14:13:02<2:35:40, 1.36s/it] 77%|███████▋ | 23152/30000 [14:13:03<2:32:01, 1.33s/it] 77%|███████▋ | 23153/30000 [14:13:09<5:04:35, 2.67s/it] 77%|███████▋ | 23154/30000 [14:13:10<4:13:25, 2.22s/it] 77%|███████▋ | 23155/30000 [14:13:12<3:37:37, 1.91s/it] 77%|███████▋ | 23156/30000 [14:13:13<3:13:15, 1.69s/it] 77%|███████▋ | 23157/30000 [14:13:14<2:56:35, 1.55s/it] 77%|███████▋ | 23158/30000 [14:13:15<2:45:10, 1.45s/it] 77%|███████▋ | 23159/30000 [14:13:16<2:36:01, 1.37s/it] 77%|███████▋ | 23160/30000 [14:13:18<2:29:24, 1.31s/it] 77%|███████▋ | 23160/30000 [14:13:18<2:29:24, 1.31s/it] 77%|███████▋ | 23161/30000 [14:13:24<5:17:47, 2.79s/it] 77%|███████▋ | 23162/30000 [14:13:25<4:22:57, 2.31s/it] 77%|███████▋ | 23163/30000 [14:13:26<3:45:20, 1.98s/it] 77%|███████▋ | 23164/30000 [14:13:27<3:18:33, 1.74s/it] 77%|███████▋ | 23165/30000 [14:13:29<2:59:57, 1.58s/it] 77%|███████▋ | 23166/30000 [14:13:30<2:46:21, 1.46s/it] 77%|███████▋ | 23167/30000 [14:13:31<2:37:26, 1.38s/it] 77%|███████▋ | 23168/30000 [14:13:32<2:32:00, 1.33s/it] 77%|███████▋ | 23169/30000 [14:13:39<5:39:24, 2.98s/it] 77%|███████▋ | 23170/30000 [14:13:40<4:37:56, 2.44s/it] 77%|███████▋ | 23170/30000 [14:13:40<4:37:56, 2.44s/it] 77%|███████▋ | 23171/30000 [14:13:41<3:54:56, 2.06s/it] 77%|███████▋ | 23172/30000 [14:13:43<3:25:34, 1.81s/it] 77%|███████▋ | 23173/30000 [14:13:44<3:04:05, 1.62s/it] 77%|███████▋ | 23174/30000 [14:13:45<2:49:10, 1.49s/it] 77%|███████▋ | 23175/30000 [14:13:46<2:38:27, 1.39s/it] 77%|███████▋ | 23176/30000 [14:13:47<2:32:04, 1.34s/it] 77%|███████▋ | 23177/30000 [14:13:54<5:23:43, 2.85s/it] 77%|███████▋ | 23178/30000 [14:13:55<4:26:05, 2.34s/it] 77%|███████▋ | 23179/30000 [14:13:56<3:47:31, 2.00s/it] 77%|███████▋ | 23180/30000 [14:13:57<3:21:05, 1.77s/it] 77%|███████▋ | 23180/30000 [14:13:57<3:21:05, 1.77s/it] 77%|███████▋ | 23181/30000 [14:13:58<3:01:12, 1.59s/it] 77%|███████▋ | 23182/30000 [14:14:00<2:46:40, 1.47s/it] 77%|███████▋ | 23183/30000 [14:14:01<2:36:40, 1.38s/it] 77%|███████▋ | 23184/30000 [14:14:02<2:30:45, 1.33s/it] 77%|███████▋ | 23185/30000 [14:14:08<5:16:22, 2.79s/it] 77%|███████▋ | 23186/30000 [14:14:09<4:21:06, 2.30s/it] 77%|███████▋ | 23187/30000 [14:14:11<3:43:13, 1.97s/it] 77%|███████▋ | 23188/30000 [14:14:12<3:16:33, 1.73s/it] 77%|███████▋ | 23189/30000 [14:14:13<2:57:07, 1.56s/it] 77%|███████▋ | 23190/30000 [14:14:14<2:44:14, 1.45s/it] 77%|███████▋ | 23190/30000 [14:14:14<2:44:14, 1.45s/it] 77%|███████▋ | 23191/30000 [14:14:15<2:34:50, 1.36s/it] 77%|███████▋ | 23192/30000 [14:14:16<2:29:08, 1.31s/it] 77%|███████▋ | 23193/30000 [14:14:22<5:09:57, 2.73s/it] 77%|███████▋ | 23194/30000 [14:14:24<4:16:47, 2.26s/it] 77%|███████▋ | 23195/30000 [14:14:25<3:39:58, 1.94s/it] 77%|███████▋ | 23196/30000 [14:14:26<3:14:45, 1.72s/it] 77%|███████▋ | 23197/30000 [14:14:27<2:56:51, 1.56s/it] 77%|███████▋ | 23198/30000 [14:14:28<2:44:17, 1.45s/it] 77%|███████▋ | 23199/30000 [14:14:30<2:34:51, 1.37s/it] 77%|███████▋ | 23200/30000 [14:14:31<2:28:37, 1.31s/it] 77%|███████▋ | 23200/30000 [14:14:31<2:28:37, 1.31s/it] 77%|███████▋ | 23201/30000 [14:14:37<5:16:27, 2.79s/it] 77%|███████▋ | 23202/30000 [14:14:38<4:21:07, 2.30s/it] 77%|███████▋ | 23203/30000 [14:14:39<3:43:29, 1.97s/it] 77%|███████▋ | 23204/30000 [14:14:41<3:16:49, 1.74s/it] 77%|███████▋ | 23205/30000 [14:14:42<2:58:11, 1.57s/it] 77%|███████▋ | 23206/30000 [14:14:43<2:44:57, 1.46s/it] 77%|███████▋ | 23207/30000 [14:14:44<2:35:06, 1.37s/it] 77%|███████▋ | 23208/30000 [14:14:45<2:28:59, 1.32s/it] 77%|███████▋ | 23209/30000 [14:14:52<5:16:30, 2.80s/it] 77%|███████▋ | 23210/30000 [14:14:53<4:21:19, 2.31s/it] +{'loss': 0.0129, 'grad_norm': 0.0875745415687561, 'learning_rate': 4.482627105486865e-06, 'flow_matching_loss': 0.012885838840156793, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.63} +{'loss': 0.0122, 'grad_norm': 0.05601033940911293, 'learning_rate': 4.470844162928412e-06, 'flow_matching_loss': 0.012179369479417801, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.64} +{'loss': 0.013, 'grad_norm': 0.07324472069740295, 'learning_rate': 4.459074014287129e-06, 'flow_matching_loss': 0.013040865957736968, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.64} +{'loss': 0.0138, 'grad_norm': 0.061665039509534836, 'learning_rate': 4.447316673864869e-06, 'flow_matching_loss': 0.01375096058472991, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.64} +{'loss': 0.0126, 'grad_norm': 0.06747863441705704, 'learning_rate': 4.435572155947903e-06, 'flow_matching_loss': 0.012608497217297555, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.65} +{'loss': 0.0134, 'grad_norm': 0.07583832740783691, 'learning_rate': 4.423840474806929e-06, 'flow_matching_loss': 0.013383533526211976, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.65} +{'loss': 0.0138, 'grad_norm': 0.06424327939748764, 'learning_rate': 4.412121644697043e-06, 'flow_matching_loss': 0.013770818710327148, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.65} +{'loss': 0.0123, 'grad_norm': 0.07237479835748672, 'learning_rate': 4.400415679857727e-06, 'flow_matching_loss': 0.012279549334198236, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.65} +{'loss': 0.0138, 'grad_norm': 0.10594811290502548, 'learning_rate': 4.388722594512838e-06, 'flow_matching_loss': 0.013779634702950716, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.66} +{'loss': 0.0134, 'grad_norm': 0.07461182028055191, 'learning_rate': 4.377042402870577e-06, 'flow_matching_loss': 0.013413378689438104, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.66} +{'loss': 0.0122, 'grad_norm': 0.07149967551231384, 'learning_rate': 4.365375119123476e-06, 'flow_matching_loss': 0.012193301040679217, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.66} +{'loss': 0.0124, 'grad_norm': 0.06420587748289108, 'learning_rate': 4.353720757448383e-06, 'flow_matching_loss': 0.012400860339403153, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.67} +{'loss': 0.0128, 'grad_norm': 0.08046424388885498, 'learning_rate': 4.342079332006445e-06, 'flow_matching_loss': 0.0127859428524971, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.67} +{'loss': 0.0132, 'grad_norm': 0.07746373862028122, 'learning_rate': 4.330450856943101e-06, 'flow_matching_loss': 0.013156667817384005, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.67} +{'loss': 0.0131, 'grad_norm': 0.07540087401866913, 'learning_rate': 4.318835346388036e-06, 'flow_matching_loss': 0.013050335831940174, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.67} +{'loss': 0.0126, 'grad_norm': 0.0715625062584877, 'learning_rate': 4.307232814455189e-06, 'flow_matching_loss': 0.01263108616694808, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.68} +{'loss': 0.0132, 'grad_norm': 0.06669336557388306, 'learning_rate': 4.29564327524274e-06, 'flow_matching_loss': 0.013175473175942897, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.68} +{'loss': 0.0132, 'grad_norm': 0.0659121423959732, 'learning_rate': 4.284066742833061e-06, 'flow_matching_loss': 0.01316832723096013, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.68} +{'loss': 0.0129, 'grad_norm': 0.08752994984388351, 'learning_rate': 4.272503231292741e-06, 'flow_matching_loss': 0.012881690915673972, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.69} +{'loss': 0.0146, 'grad_norm': 0.09754519909620285, 'learning_rate': 4.260952754672532e-06, 'flow_matching_loss': 0.014578353054821492, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.69} +{'loss': 0.0148, 'grad_norm': 0.08505729585886002, 'learning_rate': 4.249415327007353e-06, 'flow_matching_loss': 0.014796303771436214, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.69} +{'loss': 0.0136, 'grad_norm': 0.09604617953300476, 'learning_rate': 4.237890962316264e-06, 'flow_matching_loss': 0.013581185881048442, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.69} +{'loss': 0.0131, 'grad_norm': 0.08292097598314285, 'learning_rate': 4.226379674602452e-06, 'flow_matching_loss': 0.013079790864139795, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.7} +{'loss': 0.0135, 'grad_norm': 0.08257924020290375, 'learning_rate': 4.214881477853227e-06, 'flow_matching_loss': 0.013464473187923431, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.7} +{'loss': 0.0134, 'grad_norm': 0.07382367551326752, 'learning_rate': 4.2033963860399745e-06, 'flow_matching_loss': 0.013366041425615548, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.7} +{'loss': 0.0119, 'grad_norm': 0.055135373026132584, 'learning_rate': 4.1919244131181696e-06, 'flow_matching_loss': 0.01186885442584753, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.71} +{'loss': 0.0134, 'grad_norm': 0.06911183893680573, 'learning_rate': 4.180465573027335e-06, 'flow_matching_loss': 0.013424212764948606, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.71} +{'loss': 0.012, 'grad_norm': 0.07754072546958923, 'learning_rate': 4.169019879691046e-06, 'flow_matching_loss': 0.011955693410709501, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.71} +{'loss': 0.0132, 'grad_norm': 0.08831287175416946, 'learning_rate': 4.157587347016891e-06, 'flow_matching_loss': 0.01323012774810195, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.72} +{'loss': 0.0131, 'grad_norm': 0.07804005593061447, 'learning_rate': 4.14616798889649e-06, 'flow_matching_loss': 0.013057221658527851, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.72} +{'loss': 0.0118, 'grad_norm': 0.06287261843681335, 'learning_rate': 4.134761819205425e-06, 'flow_matching_loss': 0.011813301593065262, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.72} +{'loss': 0.0125, 'grad_norm': 0.08171433955430984, 'learning_rate': 4.1233688518032794e-06, 'flow_matching_loss': 0.012514743395149708, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.72} +{'loss': 0.0121, 'grad_norm': 0.0632585734128952, 'learning_rate': 4.1119891005335795e-06, 'flow_matching_loss': 0.01213830653578043, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.73} +{'loss': 0.0122, 'grad_norm': 0.06418593972921371, 'learning_rate': 4.1006225792237895e-06, 'flow_matching_loss': 0.012153969053179026, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.73} +{'loss': 0.0127, 'grad_norm': 0.06217648833990097, 'learning_rate': 4.089269301685314e-06, 'flow_matching_loss': 0.012737945653498173, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.73} +{'loss': 0.012, 'grad_norm': 0.08387526124715805, 'learning_rate': 4.077929281713453e-06, 'flow_matching_loss': 0.011958282627165317, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.74} +{'loss': 0.0134, 'grad_norm': 0.08157632499933243, 'learning_rate': 4.066602533087398e-06, 'flow_matching_loss': 0.01344501655548811, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.74} +{'loss': 0.0134, 'grad_norm': 0.08237400650978088, 'learning_rate': 4.0552890695702175e-06, 'flow_matching_loss': 0.013352810218930245, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.74} +{'loss': 0.0133, 'grad_norm': 0.081477589905262, 'learning_rate': 4.04398890490883e-06, 'flow_matching_loss': 0.013337617088109256, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.74} +{'loss': 0.0134, 'grad_norm': 0.07833698391914368, 'learning_rate': 4.032702052834013e-06, 'flow_matching_loss': 0.013405650667846204, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.75} +{'loss': 0.0134, 'grad_norm': 0.06393641233444214, 'learning_rate': 4.021428527060352e-06, 'flow_matching_loss': 0.013357904460281133, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.75} +{'loss': 0.0138, 'grad_norm': 0.05421995371580124, 'learning_rate': 4.010168341286241e-06, 'flow_matching_loss': 0.013832518458366394, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.75} 77%|███████▋ | 23210/30000 [14:14:53<4:21:19, 2.31s/it] 77%|███████▋ | 23211/30000 [14:14:54<3:43:45, 1.98s/it] 77%|███████▋ | 23212/30000 [14:14:55<3:17:01, 1.74s/it] 77%|███████▋ | 23213/30000 [14:14:56<2:59:13, 1.58s/it] 77%|███████▋ | 23214/30000 [14:14:57<2:45:25, 1.46s/it] 77%|███████▋ | 23215/30000 [14:14:59<2:35:34, 1.38s/it] 77%|███████▋ | 23216/30000 [14:15:00<2:29:00, 1.32s/it] 77%|███████▋ | 23217/30000 [14:15:06<5:17:54, 2.81s/it] 77%|███████▋ | 23218/30000 [14:15:07<4:22:29, 2.32s/it] 77%|███████▋ | 23219/30000 [14:15:09<3:43:54, 1.98s/it] 77%|███████▋ | 23220/30000 [14:15:10<3:16:28, 1.74s/it] 77%|███████▋ | 23220/30000 [14:15:10<3:16:28, 1.74s/it] 77%|███████▋ | 23221/30000 [14:15:11<2:58:13, 1.58s/it] 77%|███████▋ | 23222/30000 [14:15:12<2:44:28, 1.46s/it] 77%|███████▋ | 23223/30000 [14:15:13<2:35:18, 1.37s/it] 77%|███████▋ | 23224/30000 [14:15:14<2:29:06, 1.32s/it] 77%|███████▋ | 23225/30000 [14:15:21<5:35:25, 2.97s/it] 77%|███████▋ | 23226/30000 [14:15:22<4:34:25, 2.43s/it] 77%|███████▋ | 23227/30000 [14:15:24<3:52:04, 2.06s/it] 77%|███████▋ | 23228/30000 [14:15:25<3:22:33, 1.79s/it] 77%|███████▋ | 23229/30000 [14:15:26<3:02:00, 1.61s/it] 77%|███████▋ | 23230/30000 [14:15:27<2:47:40, 1.49s/it] 77%|███████▋ | 23230/30000 [14:15:27<2:47:40, 1.49s/it] 77%|███████▋ | 23231/30000 [14:15:28<2:36:56, 1.39s/it] 77%|███████▋ | 23232/30000 [14:15:30<2:30:27, 1.33s/it] 77%|███████▋ | 23233/30000 [14:15:36<5:35:54, 2.98s/it] 77%|███████▋ | 23234/30000 [14:15:38<4:35:11, 2.44s/it] 77%|███████▋ | 23235/30000 [14:15:39<3:54:18, 2.08s/it] 77%|███████▋ | 23236/30000 [14:15:40<3:25:29, 1.82s/it] 77%|███████▋ | 23237/30000 [14:15:41<3:05:12, 1.64s/it] 77%|███████▋ | 23238/30000 [14:15:42<2:49:37, 1.51s/it] 77%|███████▋ | 23239/30000 [14:15:44<2:38:10, 1.40s/it] 77%|███████▋ | 23240/30000 [14:15:45<2:31:12, 1.34s/it] 77%|███████▋ | 23240/30000 [14:15:45<2:31:12, 1.34s/it] 77%|███████▋ | 23241/30000 [14:15:51<5:19:14, 2.83s/it] 77%|███████▋ | 23242/30000 [14:15:52<4:22:28, 2.33s/it] 77%|███████▋ | 23243/30000 [14:15:53<3:44:26, 1.99s/it] 77%|███████▋ | 23244/30000 [14:15:55<3:16:45, 1.75s/it] 77%|███████▋ | 23245/30000 [14:15:56<2:57:50, 1.58s/it] 77%|███████▋ | 23246/30000 [14:15:57<2:44:04, 1.46s/it] 77%|███████▋ | 23247/30000 [14:15:58<2:34:21, 1.37s/it] 77%|███████▋ | 23248/30000 [14:15:59<2:28:26, 1.32s/it] 77%|███████▋ | 23249/30000 [14:16:05<5:09:28, 2.75s/it] 78%|███████▊ | 23250/30000 [14:16:07<4:15:39, 2.27s/it] 78%|███████▊ | 23250/30000 [14:16:07<4:15:39, 2.27s/it] 78%|███████▊ | 23251/30000 [14:16:08<3:38:58, 1.95s/it] 78%|███████▊ | 23252/30000 [14:16:09<3:14:16, 1.73s/it] 78%|███████▊ | 23253/30000 [14:16:10<2:55:48, 1.56s/it] 78%|███████▊ | 23254/30000 [14:16:11<2:42:56, 1.45s/it] 78%|███████▊ | 23255/30000 [14:16:13<2:33:37, 1.37s/it] 78%|███████▊ | 23256/30000 [14:16:14<2:27:43, 1.31s/it] 78%|███████▊ | 23257/30000 [14:16:20<5:02:02, 2.69s/it] 78%|███████▊ | 23258/30000 [14:16:21<4:10:59, 2.23s/it] 78%|███████▊ | 23259/30000 [14:16:22<3:35:45, 1.92s/it] 78%|███████▊ | 23260/30000 [14:16:23<3:11:00, 1.70s/it] 78%|███████▊ | 23260/30000 [14:16:23<3:11:00, 1.70s/it] 78%|███████▊ | 23261/30000 [14:16:24<2:54:18, 1.55s/it] 78%|███████▊ | 23262/30000 [14:16:26<2:41:56, 1.44s/it] 78%|███████▊ | 23263/30000 [14:16:27<2:33:42, 1.37s/it] 78%|███████▊ | 23264/30000 [14:16:28<2:27:56, 1.32s/it] 78%|███████▊ | 23265/30000 [14:16:34<5:11:36, 2.78s/it] 78%|███████▊ | 23266/30000 [14:16:35<4:18:07, 2.30s/it] 78%|███████▊ | 23267/30000 [14:16:37<3:41:43, 1.98s/it] 78%|███████▊ | 23268/30000 [14:16:38<3:15:15, 1.74s/it] 78%|███████▊ | 23269/30000 [14:16:39<2:57:07, 1.58s/it] 78%|███████▊ | 23270/30000 [14:16:40<2:43:27, 1.46s/it] 78%|███████▊ | 23270/30000 [14:16:40<2:43:27, 1.46s/it] 78%|███████▊ | 23271/30000 [14:16:41<2:34:13, 1.38s/it] 78%|███████▊ | 23272/30000 [14:16:43<2:28:25, 1.32s/it] 78%|███████▊ | 23273/30000 [14:16:49<5:06:34, 2.73s/it] 78%|███████▊ | 23274/30000 [14:16:50<4:14:03, 2.27s/it] 78%|███████▊ | 23275/30000 [14:16:51<3:37:41, 1.94s/it] 78%|███████▊ | 23276/30000 [14:16:52<3:11:56, 1.71s/it] 78%|███████▊ | 23277/30000 [14:16:53<2:54:28, 1.56s/it] 78%|███████▊ | 23278/30000 [14:16:54<2:42:12, 1.45s/it] 78%|███████▊ | 23279/30000 [14:16:56<2:33:00, 1.37s/it] 78%|███████▊ | 23280/30000 [14:16:57<2:27:24, 1.32s/it] 78%|███████▊ | 23280/30000 [14:16:57<2:27:24, 1.32s/it] 78%|███████▊ | 23281/30000 [14:17:04<5:29:20, 2.94s/it] 78%|███████▊ | 23282/30000 [14:17:05<4:29:40, 2.41s/it] 78%|███████▊ | 23283/30000 [14:17:06<3:48:59, 2.05s/it] 78%|███████▊ | 23284/30000 [14:17:07<3:20:27, 1.79s/it] 78%|███████▊ | 23285/30000 [14:17:08<3:00:12, 1.61s/it] 78%|███████▊ | 23286/30000 [14:17:09<2:45:34, 1.48s/it] 78%|███████▊ | 23287/30000 [14:17:11<2:35:19, 1.39s/it] 78%|███████▊ | 23288/30000 [14:17:12<2:28:09, 1.32s/it] 78%|███████▊ | 23289/30000 [14:17:18<5:22:00, 2.88s/it] 78%|███████▊ | 23290/30000 [14:17:20<4:24:31, 2.37s/it] 78%|███████▊ | 23290/30000 [14:17:20<4:24:31, 2.37s/it] 78%|███████▊ | 23291/30000 [14:17:21<3:45:24, 2.02s/it] 78%|███████▊ | 23292/30000 [14:17:22<3:17:53, 1.77s/it] 78%|███████▊ | 23293/30000 [14:17:23<2:58:44, 1.60s/it] 78%|███████▊ | 23294/30000 [14:17:24<2:44:43, 1.47s/it] 78%|███████▊ | 23295/30000 [14:17:25<2:34:58, 1.39s/it] 78%|███████▊ | 23296/30000 [14:17:27<2:28:38, 1.33s/it] 78%|███████▊ | 23297/30000 [14:17:33<5:04:43, 2.73s/it] 78%|███████▊ | 23298/30000 [14:17:34<4:12:18, 2.26s/it] 78%|███████▊ | 23299/30000 [14:17:35<3:36:52, 1.94s/it] 78%|███████▊ | 23300/30000 [14:17:36<3:11:33, 1.72s/it] 78%|███████▊ | 23300/30000 [14:17:36<3:11:33, 1.72s/it] 78%|███████▊ | 23301/30000 [14:17:37<2:54:40, 1.56s/it] 78%|███████▊ | 23302/30000 [14:17:39<2:41:45, 1.45s/it] 78%|███████▊ | 23303/30000 [14:17:40<2:32:07, 1.36s/it] 78%|███████▊ | 23304/30000 [14:17:41<2:26:59, 1.32s/it] 78%|███████▊ | 23305/30000 [14:17:47<5:08:48, 2.77s/it] 78%|███████▊ | 23306/30000 [14:17:48<4:15:34, 2.29s/it] 78%|███████▊ | 23307/30000 [14:17:50<3:38:50, 1.96s/it] 78%|███████▊ | 23308/30000 [14:17:51<3:12:54, 1.73s/it] 78%|███████▊ | 23309/30000 [14:17:52<2:55:58, 1.58s/it] 78%|███████▊ | 23310/30000 [14:17:53<2:42:14, 1.46s/it] 78%|███████▊ | 23310/30000 [14:17:53<2:42:14, 1.46s/it] 78%|███████▊ | 23311/30000 [14:17:54<2:33:01, 1.37s/it] 78%|███████▊ | 23312/30000 [14:17:55<2:26:45, 1.32s/it] 78%|███████▊ | 23313/30000 [14:18:02<5:16:47, 2.84s/it] 78%|███████▊ | 23314/30000 [14:18:03<4:20:54, 2.34s/it] 78%|███████▊ | 23315/30000 [14:18:04<3:42:09, 1.99s/it] 78%|███████▊ | 23316/30000 [14:18:05<3:15:34, 1.76s/it] 78%|███████▊ | 23317/30000 [14:18:07<2:56:18, 1.58s/it] 78%|███████▊ | 23318/30000 [14:18:08<2:42:51, 1.46s/it] 78%|███████▊ | 23319/30000 [14:18:09<2:34:27, 1.39s/it] 78%|███████▊ | 23320/30000 [14:18:10<2:28:34, 1.33s/it] 78%|███████▊ | 23320/30000 [14:18:10<2:28:34, 1.33s/it] 78%|███████▊ | 23321/30000 [14:18:16<5:03:29, 2.73s/it] 78%|███████▊ | 23322/30000 [14:18:17<4:11:15, 2.26s/it] 78%|███████▊ | 23323/30000 [14:18:19<3:35:46, 1.94s/it] 78%|███████▊ | 23324/30000 [14:18:20<3:10:00, 1.71s/it] 78%|███████▊ | 23325/30000 [14:18:21<2:53:08, 1.56s/it] 78%|███████▊ | 23326/30000 [14:18:22<2:40:43, 1.44s/it] 78%|███████▊ | 23327/30000 [14:18:23<2:31:18, 1.36s/it] 78%|███████▊ | 23328/30000 [14:18:24<2:25:14, 1.31s/it] 78%|███████▊ | 23329/30000 [14:18:32<5:37:46, 3.04s/it] 78%|███████▊ | 23330/30000 [14:18:33<4:34:33, 2.47s/it] 78%|███████▊ | 23330/30000 [14:18:33<4:34:33, 2.47s/it] 78%|███████▊ | 23331/30000 [14:18:34<3:51:55, 2.09s/it] 78%|███████▊ | 23332/30000 [14:18:35<3:21:31, 1.81s/it] 78%|███████▊ | 23333/30000 [14:18:36<3:00:04, 1.62s/it] 78%|███████▊ | 23334/30000 [14:18:37<2:45:32, 1.49s/it] 78%|███████▊ | 23335/30000 [14:18:39<2:35:15, 1.40s/it] 78%|███████▊ | 23336/30000 [14:18:40<2:27:45, 1.33s/it] 78%|███████▊ | 23337/30000 [14:18:47<5:47:43, 3.13s/it] 78%|███████▊ | 23338/30000 [14:18:48<4:43:14, 2.55s/it] 78%|███████▊ | 23339/30000 [14:18:49<3:57:45, 2.14s/it] 78%|███████▊ | 23340/30000 [14:18:51<3:26:40, 1.86s/it] 78%|███████▊ | 23340/30000 [14:18:51<3:26:40, 1.86s/it] 78%|███████▊ | 23341/30000 [14:18:52<3:04:15, 1.66s/it] 78%|███████▊ | 23342/30000 [14:18:53<2:48:06, 1.51s/it] 78%|███████▊ | 23343/30000 [14:18:54<2:36:32, 1.41s/it] 78%|███████▊ | 23344/30000 [14:18:55<2:29:23, 1.35s/it] 78%|███████▊ | 23345/30000 [14:19:02<5:14:17, 2.83s/it] 78%|███████▊ | 23346/30000 [14:19:03<4:18:58, 2.34s/it] 78%|███████▊ | 23347/30000 [14:19:04<3:40:41, 1.99s/it] 78%|███████▊ | 23348/30000 [14:19:05<3:14:14, 1.75s/it] 78%|███████▊ | 23349/30000 [14:19:06<2:54:52, 1.58s/it] 78%|███████▊ | 23350/30000 [14:19:08<2:42:06, 1.46s/it] 78%|███████▊ | 23350/30000 [14:19:08<2:42:06, 1.46s/it] 78%|███████▊ | 23351/30000 [14:19:09<2:32:32, 1.38s/it] 78%|███████▊ | 23352/30000 [14:19:10<2:27:39, 1.33s/it] 78%|███████▊ | 23353/30000 [14:19:16<5:01:27, 2.72s/it] 78%|███████▊ | 23354/30000 [14:19:17<4:09:36, 2.25s/it] 78%|███████▊ | 23355/30000 [14:19:18<3:33:39, 1.93s/it] 78%|███████▊ | 23356/30000 [14:19:19<3:08:58, 1.71s/it] 78%|███████▊ | 23357/30000 [14:19:21<2:51:19, 1.55s/it] 78%|███████▊ | 23358/30000 [14:19:22<2:39:34, 1.44s/it] 78%|███████▊ | 23359/30000 [14:19:23<2:30:27, 1.36s/it] 78%|███████▊ | 23360/30000 [14:19:24<2:25:32, 1.32s/it] 78%|███████▊ | 23360/30000 [14:19:24<2:25:32, 1.32s/it] 78%|███████▊ | 23361/30000 [14:19:30<4:58:11, 2.69s/it] 78%|███████▊ | 23362/30000 [14:19:31<4:07:33, 2.24s/it] 78%|███████▊ | 23363/30000 [14:19:33<3:32:46, 1.92s/it] 78%|███████▊ | 23364/30000 [14:19:34<3:07:42, 1.70s/it] 78%|███████▊ | 23365/30000 [14:19:35<2:50:21, 1.54s/it] 78%|███████▊ | 23366/30000 [14:19:36<2:38:46, 1.44s/it] 78%|███████▊ | 23367/30000 [14:19:37<2:29:39, 1.35s/it] 78%|███████▊ | 23368/30000 [14:19:38<2:24:40, 1.31s/it] 78%|███████▊ | 23369/30000 [14:19:45<5:12:19, 2.83s/it] 78%|███████▊ | 23370/30000 [14:19:46<4:17:03, 2.33s/it] 78%|███████▊ | 23370/30000 [14:19:46<4:17:03, 2.33s/it] 78%|███████▊ | 23371/30000 [14:19:47<3:39:51, 1.99s/it] 78%|███████▊ | 23372/30000 [14:19:48<3:13:08, 1.75s/it] 78%|███████▊ | 23373/30000 [14:19:50<2:54:37, 1.58s/it] 78%|███████▊ | 23374/30000 [14:19:51<2:41:44, 1.46s/it] 78%|███████▊ | 23375/30000 [14:19:52<2:32:02, 1.38s/it] 78%|███████▊ | 23376/30000 [14:19:53<2:25:41, 1.32s/it] 78%|███████▊ | 23377/30000 [14:19:59<4:59:44, 2.72s/it] 78%|███████▊ | 23378/30000 [14:20:00<4:08:35, 2.25s/it] 78%|███████▊ | 23379/30000 [14:20:01<3:34:03, 1.94s/it] 78%|███████▊ | 23380/30000 [14:20:03<3:08:32, 1.71s/it] 78%|███████▊ | 23380/30000 [14:20:03<3:08:32, 1.71s/it] 78%|███████▊ | 23381/30000 [14:20:04<2:53:00, 1.57s/it] 78%|███████▊ | 23382/30000 [14:20:05<2:39:59, 1.45s/it] 78%|███████▊ | 23383/30000 [14:20:06<2:30:29, 1.36s/it] 78%|███████▊ | 23384/30000 [14:20:07<2:24:30, 1.31s/it] 78%|███████▊ | 23385/30000 [14:20:13<4:49:31, 2.63s/it] 78%|███████▊ | 23386/30000 [14:20:14<4:01:25, 2.19s/it] 78%|███████▊ | 23387/30000 [14:20:15<3:28:25, 1.89s/it] 78%|███████▊ | 23388/30000 [14:20:17<3:05:03, 1.68s/it] 78%|███████▊ | 23389/30000 [14:20:18<2:48:56, 1.53s/it] 78%|███████▊ | 23390/30000 [14:20:19<2:37:49, 1.43s/it] 78%|███████▊ | 23390/30000 [14:20:19<2:37:49, 1.43s/it] 78%|███████▊ | 23391/30000 [14:20:20<2:29:58, 1.36s/it] 78%|███████▊ | 23392/30000 [14:20:21<2:24:22, 1.31s/it] 78%|███████▊ | 23393/30000 [14:20:28<5:04:39, 2.77s/it] 78%|███████▊ | 23394/30000 [14:20:29<4:12:02, 2.29s/it] 78%|███████▊ | 23395/30000 [14:20:30<3:35:55, 1.96s/it] 78%|███████▊ | 23396/30000 [14:20:31<3:09:36, 1.72s/it] 78%|███████▊ | 23397/30000 [14:20:32<2:51:30, 1.56s/it] 78%|███████▊ | 23398/30000 [14:20:33<2:38:43, 1.44s/it] 78%|███████▊ | 23399/30000 [14:20:35<2:29:40, 1.36s/it] 78%|███████▊ | 23400/30000 [14:20:36<2:23:53, 1.31s/it] 78%|███████▊ | 23400/30000 [14:20:36<2:23:53, 1.31s/it] 78%|███████▊ | 23401/30000 [14:20:43<5:30:16, 3.00s/it] 78%|███████▊ | 23402/30000 [14:20:44<4:30:51, 2.46s/it] 78%|███████▊ | 23403/30000 [14:20:45<3:49:16, 2.09s/it] 78%|███████▊ | 23404/30000 [14:20:46<3:19:16, 1.81s/it] 78%|███████▊ | 23405/30000 [14:20:48<2:57:49, 1.62s/it] 78%|███████▊ | 23406/30000 [14:20:49<2:43:39, 1.49s/it] 78%|███████▊ | 23407/30000 [14:20:50<2:34:07, 1.40s/it] 78%|███████▊ | 23408/30000 [14:20:51<2:26:55, 1.34s/it] 78%|███████▊ | 23409/30000 [14:20:57<5:00:59, 2.74s/it] 78%|███████▊ | 23410/30000 [14:20:58<4:09:20, 2.27s/it] 78%|███████▊ | 23410/30000 [14:20:58<4:09:20, 2.27s/it] 78%|███████▊ | 23411/30000 [14:20:59<3:32:51, 1.94s/it] 78%|███████▊ | 23412/30000 [14:21:01<3:08:33, 1.72s/it] 78%|███████▊ | 23413/30000 [14:21:02<2:51:32, 1.56s/it] 78%|███████▊ | 23414/30000 [14:21:03<2:39:21, 1.45s/it] 78%|███████▊ | 23415/30000 [14:21:04<2:30:00, 1.37s/it] 78%|███████▊ | 23416/30000 [14:21:05<2:24:01, 1.31s/it] 78%|███████▊ | 23417/30000 [14:21:12<5:08:36, 2.81s/it] 78%|███████▊ | 23418/30000 [14:21:13<4:14:27, 2.32s/it] 78%|███████▊ | 23419/30000 [14:21:14<3:37:35, 1.98s/it] 78%|███████▊ | 23420/30000 [14:21:15<3:11:18, 1.74s/it] 78%|███████▊ | 23420/30000 [14:21:15<3:11:18, 1.74s/it] 78%|███████▊ | 23421/30000 [14:21:16<2:53:04, 1.58s/it] 78%|███████▊ | 23422/30000 [14:21:18<2:40:12, 1.46s/it] 78%|███████▊ | 23423/30000 [14:21:19<2:30:58, 1.38s/it] 78%|███████▊ | 23424/30000 [14:21:20<2:24:53, 1.32s/it] 78%|███████▊ | 23425/30000 [14:21:27<5:39:21, 3.10s/it] 78%|███████▊ | 23426/30000 [14:21:28<4:36:20, 2.52s/it] 78%|███████▊ | 23427/30000 [14:21:30<3:53:00, 2.13s/it] 78%|███████▊ | 23428/30000 [14:21:31<3:22:23, 1.85s/it] 78%|███████▊ | 23429/30000 [14:21:32<3:01:36, 1.66s/it] 78%|███████▊ | 23430/30000 [14:21:33<2:45:50, 1.51s/it] 78%|███████▊ | 23430/30000 [14:21:33<2:45:50, 1.51s/it] 78%|███████▊ | 23431/30000 [14:21:34<2:34:47, 1.41s/it] 78%|███████▊ | 23432/30000 [14:21:36<2:27:01, 1.34s/it] 78%|███████▊ | 23433/30000 [14:21:42<5:13:31, 2.86s/it] 78%|███████▊ | 23434/30000 [14:21:43<4:17:48, 2.36s/it] 78%|███████▊ | 23435/30000 [14:21:44<3:39:23, 2.01s/it] 78%|███████▊ | 23436/30000 [14:21:46<3:12:33, 1.76s/it] 78%|███████▊ | 23437/30000 [14:21:47<2:53:54, 1.59s/it] 78%|███████▊ | 23438/30000 [14:21:48<2:40:39, 1.47s/it] 78%|███████▊ | 23439/30000 [14:21:49<2:30:59, 1.38s/it] 78%|███████▊ | 23440/30000 [14:21:50<2:25:20, 1.33s/it] 78%|███████▊ | 23440/30000 [14:21:50<2:25:20, 1.33s/it] 78%|███████▊ | 23441/30000 [14:21:56<5:00:43, 2.75s/it] 78%|███████▊ | 23442/30000 [14:21:58<4:08:48, 2.28s/it] 78%|███████▊ | 23443/30000 [14:21:59<3:32:33, 1.94s/it] 78%|███████▊ | 23444/30000 [14:22:00<3:07:13, 1.71s/it] 78%|███████▊ | 23445/30000 [14:22:01<2:50:22, 1.56s/it] 78%|███████▊ | 23446/30000 [14:22:02<2:38:06, 1.45s/it] 78%|███████▊ | 23447/30000 [14:22:03<2:28:49, 1.36s/it] 78%|███████▊ | 23448/30000 [14:22:05<2:23:33, 1.31s/it] 78%|███████▊ | 23449/30000 [14:22:11<4:58:10, 2.73s/it] 78%|███████▊ | 23450/30000 [14:22:12<4:06:49, 2.26s/it] 78%|███████▊ | 23450/30000 [14:22:12<4:06:49, 2.26s/it] 78%|███████▊ | 23451/30000 [14:22:13<3:31:15, 1.94s/it] 78%|███████▊ | 23452/30000 [14:22:14<3:07:48, 1.72s/it] 78%|███████▊ | 23453/30000 [14:22:15<2:50:41, 1.56s/it] 78%|███████▊ | 23454/30000 [14:22:17<2:39:46, 1.46s/it] 78%|███████▊ | 23455/30000 [14:22:18<2:29:59, 1.37s/it] 78%|███████▊ | 23456/30000 [14:22:19<2:23:41, 1.32s/it] 78%|███████▊ | 23457/30000 [14:22:25<5:00:08, 2.75s/it] 78%|███████▊ | 23458/30000 [14:22:26<4:08:23, 2.28s/it] 78%|███████▊ | 23459/30000 [14:22:27<3:32:32, 1.95s/it] 78%|███████▊ | 23460/30000 [14:22:29<3:08:09, 1.73s/it] 78%|███████▊ | 23460/30000 [14:22:29<3:08:09, 1.73s/it] 78%|███████▊ | 23461/30000 [14:22:30<2:50:23, 1.56s/it] 78%|███████▊ | 23462/30000 [14:22:31<2:38:25, 1.45s/it] 78%|███████▊ | 23463/30000 [14:22:32<2:30:30, 1.38s/it] 78%|███████▊ | 23464/30000 [14:22:33<2:24:36, 1.33s/it] 78%|███████▊ | 23465/30000 [14:22:39<4:44:53, 2.62s/it] 78%|███████▊ | 23466/30000 [14:22:40<3:57:36, 2.18s/it] 78%|███████▊ | 23467/30000 [14:22:41<3:25:11, 1.88s/it] 78%|███████▊ | 23468/30000 [14:22:43<3:02:28, 1.68s/it] 78%|███████▊ | 23469/30000 [14:22:44<2:46:20, 1.53s/it] 78%|███████▊ | 23470/30000 [14:22:45<2:35:18, 1.43s/it] 78%|███████▊ | 23470/30000 [14:22:45<2:35:18, 1.43s/it] 78%|███████▊ | 23471/30000 [14:22:46<2:26:44, 1.35s/it] 78%|███████▊ | 23472/30000 [14:22:47<2:20:40, 1.29s/it] 78%|███████▊ | 23473/30000 [14:22:54<5:11:09, 2.86s/it] 78%|███████▊ | 23474/30000 [14:22:55<4:16:11, 2.36s/it] 78%|███████▊ | 23475/30000 [14:22:56<3:38:24, 2.01s/it] 78%|███████▊ | 23476/30000 [14:22:57<3:11:39, 1.76s/it] 78%|███████▊ | 23477/30000 [14:22:59<2:52:08, 1.58s/it] 78%|███████▊ | 23478/30000 [14:23:00<2:38:40, 1.46s/it] 78%|███████▊ | 23479/30000 [14:23:01<2:29:04, 1.37s/it] 78%|███████▊ | 23480/30000 [14:23:02<2:23:35, 1.32s/it] 78%|███████▊ | 23480/30000 [14:23:02<2:23:35, 1.32s/it] 78%|███████▊ | 23481/30000 [14:23:09<5:09:08, 2.85s/it] 78%|███████▊ | 23482/30000 [14:23:10<4:15:16, 2.35s/it] 78%|███████▊ | 23483/30000 [14:23:11<3:36:38, 1.99s/it] 78%|███████▊ | 23484/30000 [14:23:12<3:09:35, 1.75s/it] 78%|███████▊ | 23485/30000 [14:23:13<2:50:59, 1.57s/it] 78%|███████▊ | 23486/30000 [14:23:14<2:37:58, 1.46s/it] 78%|███████▊ | 23487/30000 [14:23:16<2:28:42, 1.37s/it] 78%|███████▊ | 23488/30000 [14:23:17<2:22:52, 1.32s/it] 78%|███████▊ | 23489/30000 [14:23:22<4:39:21, 2.57s/it] 78%|███████▊ | 23490/30000 [14:23:23<3:53:30, 2.15s/it] 78%|███████▊ | 23490/30000 [14:23:23<3:53:30, 2.15s/it] 78%|███████▊ | 23491/30000 [14:23:25<3:22:37, 1.87s/it] 78%|███████▊ | 23492/30000 [14:23:26<3:00:51, 1.67s/it] 78%|███████▊ | 23493/30000 [14:23:27<2:45:34, 1.53s/it] 78%|███████▊ | 23494/30000 [14:23:28<2:34:44, 1.43s/it] 78%|███████▊ | 23495/30000 [14:23:29<2:26:10, 1.35s/it] 78%|███████▊ | 23496/30000 [14:23:31<2:21:38, 1.31s/it] 78%|███████▊ | 23497/30000 [14:23:36<4:45:25, 2.63s/it] 78%|███████▊ | 23498/30000 [14:23:38<3:58:00, 2.20s/it] 78%|███████▊ | 23499/30000 [14:23:39<3:24:28, 1.89s/it] 78%|███████▊ | 23500/30000 [14:23:40<3:01:00, 1.67s/it] 78%|███████▊ | 23500/30000 [14:23:40<3:01:00, 1.67s/it] 78%|███████▊ | 23501/30000 [14:23:41<2:45:37, 1.53s/it] 78%|███████▊ | 23502/30000 [14:23:42<2:34:52, 1.43s/it] 78%|███████▊ | 23503/30000 [14:23:43<2:26:36, 1.35s/it] 78%|███████▊ | 23504/30000 [14:23:45<2:21:13, 1.30s/it] 78%|███████▊ | 23505/30000 [14:23:51<5:08:18, 2.85s/it] 78%|███████▊ | 23506/30000 [14:23:52<4:13:32, 2.34s/it] 78%|███████▊ | 23507/30000 [14:23:53<3:36:26, 2.00s/it] 78%|███████▊ | 23508/30000 [14:23:55<3:10:09, 1.76s/it] 78%|███████▊ | 23509/30000 [14:23:56<2:51:50, 1.59s/it] 78%|███████▊ | 23510/30000 [14:23:57<2:38:32, 1.47s/it] 78%|███████▊ | 23510/30000 [14:23:57<2:38:32, 1.47s/it] 78%|███████▊ | 23511/30000 [14:23:58<2:29:20, 1.38s/it] 78%|███████▊ | 23512/30000 [14:23:59<2:22:58, 1.32s/it] 78%|███████▊ | 23513/30000 [14:24:05<4:43:48, 2.63s/it] 78%|███████▊ | 23514/30000 [14:24:06<3:56:32, 2.19s/it] 78%|███████▊ | 23515/30000 [14:24:07<3:24:15, 1.89s/it] 78%|███████▊ | 23516/30000 [14:24:09<3:01:39, 1.68s/it] 78%|███████▊ | 23517/30000 [14:24:10<2:45:05, 1.53s/it] 78%|███████▊ | 23518/30000 [14:24:11<2:33:38, 1.42s/it] 78%|███████▊ | 23519/30000 [14:24:12<2:25:35, 1.35s/it] 78%|███████▊ | 23520/30000 [14:24:13<2:20:46, 1.30s/it] 78%|███████▊ | 23520/30000 [14:24:13<2:20:46, 1.30s/it] 78%|███████▊ | 23521/30000 [14:24:20<5:06:10, 2.84s/it] 78%|███████▊ | 23522/30000 [14:24:21<4:12:00, 2.33s/it] 78%|███████▊ | 23523/30000 [14:24:22<3:35:45, 2.00s/it] 78%|███████▊ | 23524/30000 [14:24:23<3:09:29, 1.76s/it] 78%|███████▊ | 23525/30000 [14:24:24<2:50:25, 1.58s/it] 78%|███████▊ | 23526/30000 [14:24:26<2:37:57, 1.46s/it] 78%|███████▊ | 23527/30000 [14:24:27<2:28:41, 1.38s/it] 78%|███████▊ | 23528/30000 [14:24:28<2:23:57, 1.33s/it] 78%|███████▊ | 23529/30000 [14:24:34<5:04:46, 2.83s/it] 78%|███████▊ | 23530/30000 [14:24:36<4:11:20, 2.33s/it] 78%|███████▊ | 23530/30000 [14:24:36<4:11:20, 2.33s/it] 78%|███████▊ | 23531/30000 [14:24:37<3:34:51, 1.99s/it] 78%|███████▊ | 23532/30000 [14:24:38<3:09:08, 1.75s/it] 78%|███████▊ | 23533/30000 [14:24:39<2:51:12, 1.59s/it] 78%|███████▊ | 23534/30000 [14:24:40<2:38:02, 1.47s/it] 78%|███████▊ | 23535/30000 [14:24:41<2:28:19, 1.38s/it] 78%|███████▊ | 23536/30000 [14:24:43<2:22:27, 1.32s/it] 78%|███████▊ | 23537/30000 [14:24:49<4:48:34, 2.68s/it] 78%|███████▊ | 23538/30000 [14:24:50<4:00:34, 2.23s/it] 78%|███████▊ | 23539/30000 [14:24:51<3:25:46, 1.91s/it] 78%|███████▊ | 23540/30000 [14:24:52<3:02:46, 1.70s/it] 78%|███████▊ | 23540/30000 [14:24:52<3:02:46, 1.70s/it] 78%|███████▊ | 23541/30000 [14:24:53<2:45:46, 1.54s/it] 78%|███████▊ | 23542/30000 [14:24:54<2:34:01, 1.43s/it] 78%|███████▊ | 23543/30000 [14:24:56<2:25:23, 1.35s/it] 78%|███████▊ | 23544/30000 [14:24:57<2:20:04, 1.30s/it] 78%|███████▊ | 23545/30000 [14:25:03<4:55:08, 2.74s/it] 78%|███████▊ | 23546/30000 [14:25:04<4:04:46, 2.28s/it] 78%|███████▊ | 23547/30000 [14:25:05<3:29:01, 1.94s/it] 78%|███████▊ | 23548/30000 [14:25:06<3:04:41, 1.72s/it] 78%|███████▊ | 23549/30000 [14:25:08<2:47:56, 1.56s/it] 78%|███████▊ | 23550/30000 [14:25:09<2:35:35, 1.45s/it] 78%|███████▊ | 23550/30000 [14:25:09<2:35:35, 1.45s/it] 79%|███████▊ | 23551/30000 [14:25:10<2:26:28, 1.36s/it] 79%|███████▊ | 23552/30000 [14:25:11<2:21:29, 1.32s/it] 79%|███████▊ | 23553/30000 [14:25:17<5:01:14, 2.80s/it] 79%|███████▊ | 23554/30000 [14:25:19<4:09:10, 2.32s/it] 79%|███████▊ | 23555/30000 [14:25:20<3:33:51, 1.99s/it] 79%|███████▊ | 23556/30000 [14:25:21<3:09:37, 1.77s/it] 79%|███████▊ | 23557/30000 [14:25:22<2:51:08, 1.59s/it] 79%|███████▊ | 23558/30000 [14:25:23<2:38:03, 1.47s/it] 79%|███████▊ | 23559/30000 [14:25:25<2:28:32, 1.38s/it] 79%|███████▊ | 23560/30000 [14:25:26<2:21:49, 1.32s/it] 79%|███████▊ | 23560/30000 [14:25:26<2:21:49, 1.32s/it] 79%|███████▊ | 23561/30000 [14:25:32<5:05:56, 2.85s/it] 79%|███████▊ | 23562/30000 [14:25:33<4:12:51, 2.36s/it] 79%|███████▊ | 23563/30000 [14:25:35<3:35:19, 2.01s/it] 79%|███████▊ | 23564/30000 [14:25:36<3:08:56, 1.76s/it] 79%|███████▊ | 23565/30000 [14:25:37<2:51:10, 1.60s/it] 79%|███████▊ | 23566/30000 [14:25:38<2:37:55, 1.47s/it] 79%|███████▊ | 23567/30000 [14:25:39<2:28:36, 1.39s/it] 79%|███████▊ | 23568/30000 [14:25:41<2:23:15, 1.34s/it] 79%|███████▊ | 23569/30000 [14:25:47<5:15:06, 2.94s/it] 79%|███████▊ | 23570/30000 [14:25:49<4:18:33, 2.41s/it] 79%|███████▊ | 23570/30000 [14:25:49<4:18:33, 2.41s/it] 79%|███████▊ | 23571/30000 [14:25:50<3:39:10, 2.05s/it] 79%|███████▊ | 23572/30000 [14:25:51<3:11:23, 1.79s/it] 79%|███████▊ | 23573/30000 [14:25:52<2:51:47, 1.60s/it] 79%|███████▊ | 23574/30000 [14:25:53<2:37:58, 1.47s/it] 79%|███████▊ | 23575/30000 [14:25:54<2:28:47, 1.39s/it] 79%|███████▊ | 23576/30000 [14:25:56<2:22:55, 1.33s/it] 79%|███████▊ | 23577/30000 [14:26:02<5:19:21, 2.98s/it] 79%|███████▊ | 23578/30000 [14:26:04<4:20:48, 2.44s/it] 79%|███████▊ | 23579/30000 [14:26:05<3:40:23, 2.06s/it] 79%|███████▊ | 23580/30000 [14:26:06<3:12:08, 1.80s/it] 79%|███████▊ | 23580/30000 [14:26:06<3:12:08, 1.80s/it] 79%|███████▊ | 23581/30000 [14:26:07<2:51:49, 1.61s/it] 79%|███████▊ | 23582/30000 [14:26:08<2:38:20, 1.48s/it] 79%|███████▊ | 23583/30000 [14:26:10<2:28:29, 1.39s/it] 79%|███████▊ | 23584/30000 [14:26:11<2:21:56, 1.33s/it] 79%|███████▊ | 23585/30000 [14:26:17<4:58:12, 2.79s/it] 79%|███████▊ | 23586/30000 [14:26:18<4:05:52, 2.30s/it] 79%|███████▊ | 23587/30000 [14:26:19<3:30:33, 1.97s/it] 79%|███████▊ | 23588/30000 [14:26:20<3:05:13, 1.73s/it] 79%|███████▊ | 23589/30000 [14:26:22<2:47:44, 1.57s/it] 79%|███████▊ | 23590/30000 [14:26:23<2:35:28, 1.46s/it] 79%|███████▊ | 23590/30000 [14:26:23<2:35:28, 1.46s/it] 79%|███████▊ | 23591/30000 [14:26:24<2:26:30, 1.37s/it] 79%|███████▊ | 23592/30000 [14:26:25<2:20:55, 1.32s/it] 79%|███████▊ | 23593/30000 [14:26:31<4:59:26, 2.80s/it] 79%|███████▊ | 23594/30000 [14:26:33<4:07:07, 2.31s/it] 79%|███████▊ | 23595/30000 [14:26:34<3:31:29, 1.98s/it] 79%|███████▊ | 23596/30000 [14:26:35<3:07:17, 1.75s/it] 79%|███████▊ | 23597/30000 [14:26:36<2:50:26, 1.60s/it] 79%|███████▊ | 23598/30000 [14:26:37<2:37:13, 1.47s/it] 79%|███████▊ | 23599/30000 [14:26:39<2:27:26, 1.38s/it] 79%|███████▊ | 23600/30000 [14:26:40<2:21:23, 1.33s/it] 79%|███████▊ | 23600/30000 [14:26:40<2:21:23, 1.33s/it] 79%|███████▊ | 23601/30000 [14:26:46<4:57:10, 2.79s/it] 79%|███████▊ | 23602/30000 [14:26:47<4:05:07, 2.30s/it] 79%|███████▊ | 23603/30000 [14:26:48<3:30:03, 1.97s/it] 79%|███████▊ | 23604/30000 [14:26:50<3:04:09, 1.73s/it] 79%|███████▊ | 23605/30000 [14:26:51<2:46:41, 1.56s/it] 79%|███████▊ | 23606/30000 [14:26:52<2:34:13, 1.45s/it] 79%|███████▊ | 23607/30000 [14:26:53<2:25:38, 1.37s/it] 79%|███████▊ | 23608/30000 [14:26:54<2:20:08, 1.32s/it] 79%|███████▊ | 23609/30000 [14:27:01<5:11:27, 2.92s/it] 79%|███████▊ | 23610/30000 [14:27:02<4:17:02, 2.41s/it] 79%|███████▊ | 23610/30000 [14:27:02<4:17:02, 2.41s/it] 79%|███████▊ | 23611/30000 [14:27:03<3:37:59, 2.05s/it] 79%|███████▊ | 23612/30000 [14:27:05<3:10:12, 1.79s/it] 79%|███████▊ | 23613/30000 [14:27:06<2:49:59, 1.60s/it] 79%|███████▊ | 23614/30000 [14:27:07<2:36:35, 1.47s/it] 79%|███████▊ | 23615/30000 [14:27:08<2:27:27, 1.39s/it] 79%|███████▊ | 23616/30000 [14:27:09<2:21:27, 1.33s/it] 79%|███████▊ | 23617/30000 [14:27:16<5:02:48, 2.85s/it] 79%|███████▊ | 23618/30000 [14:27:17<4:09:06, 2.34s/it] 79%|███████▊ | 23619/30000 [14:27:18<3:32:42, 2.00s/it] 79%|███████▊ | 23620/30000 [14:27:19<3:06:08, 1.75s/it] +{'loss': 0.0136, 'grad_norm': 0.07249770313501358, 'learning_rate': 3.998921509193868e-06, 'flow_matching_loss': 0.01359674073755741, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.76} +{'loss': 0.0133, 'grad_norm': 0.07241462171077728, 'learning_rate': 3.987688044449193e-06, 'flow_matching_loss': 0.01328016333281994, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.76} +{'loss': 0.0151, 'grad_norm': 0.08209222555160522, 'learning_rate': 3.9764679607019385e-06, 'flow_matching_loss': 0.015134180057793855, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.76} +{'loss': 0.013, 'grad_norm': 0.06588281691074371, 'learning_rate': 3.965261271585561e-06, 'flow_matching_loss': 0.012993327621370555, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.76} +{'loss': 0.0128, 'grad_norm': 0.07698290050029755, 'learning_rate': 3.954067990717251e-06, 'flow_matching_loss': 0.01275297049432993, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.77} +{'loss': 0.0124, 'grad_norm': 0.07327555119991302, 'learning_rate': 3.9428881316978935e-06, 'flow_matching_loss': 0.012368461955338717, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.77} +{'loss': 0.0145, 'grad_norm': 0.06738276779651642, 'learning_rate': 3.931721708112075e-06, 'flow_matching_loss': 0.014484945405274629, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.77} +{'loss': 0.0126, 'grad_norm': 0.05898813530802727, 'learning_rate': 3.920568733528047e-06, 'flow_matching_loss': 0.012577784899622201, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.78} +{'loss': 0.0132, 'grad_norm': 0.06367369741201401, 'learning_rate': 3.909429221497734e-06, 'flow_matching_loss': 0.013249537535011768, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.78} +{'loss': 0.0131, 'grad_norm': 0.08027568459510803, 'learning_rate': 3.898303185556689e-06, 'flow_matching_loss': 0.013080786634236573, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.78} +{'loss': 0.0138, 'grad_norm': 0.09910979121923447, 'learning_rate': 3.887190639224097e-06, 'flow_matching_loss': 0.013767514564096927, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.78} +{'loss': 0.0121, 'grad_norm': 0.060933541506528854, 'learning_rate': 3.876091596002746e-06, 'flow_matching_loss': 0.01205474054440856, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.79} +{'loss': 0.0141, 'grad_norm': 0.07226121425628662, 'learning_rate': 3.865006069379019e-06, 'flow_matching_loss': 0.01407859642058611, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.79} +{'loss': 0.0124, 'grad_norm': 0.06077126786112785, 'learning_rate': 3.853934072822881e-06, 'flow_matching_loss': 0.012378246989101171, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.79} +{'loss': 0.0128, 'grad_norm': 0.11632164567708969, 'learning_rate': 3.8428756197878514e-06, 'flow_matching_loss': 0.01279826508834958, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.8} +{'loss': 0.0139, 'grad_norm': 0.08741648495197296, 'learning_rate': 3.831830723710992e-06, 'flow_matching_loss': 0.013852845970541238, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.8} +{'loss': 0.0123, 'grad_norm': 0.07019244134426117, 'learning_rate': 3.820799398012889e-06, 'flow_matching_loss': 0.012309214193373918, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.8} +{'loss': 0.0137, 'grad_norm': 0.10938937216997147, 'learning_rate': 3.80978165609765e-06, 'flow_matching_loss': 0.013724331324920059, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.81} +{'loss': 0.0143, 'grad_norm': 0.09971470385789871, 'learning_rate': 3.798777511352864e-06, 'flow_matching_loss': 0.014336697943508624, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.81} +{'loss': 0.0136, 'grad_norm': 0.08182620257139206, 'learning_rate': 3.787786977149613e-06, 'flow_matching_loss': 0.013575168885290622, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.81} +{'loss': 0.0131, 'grad_norm': 0.0722547248005867, 'learning_rate': 3.776810066842427e-06, 'flow_matching_loss': 0.01306431544944644, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.81} +{'loss': 0.0128, 'grad_norm': 0.06513359397649765, 'learning_rate': 3.7658467937692866e-06, 'flow_matching_loss': 0.012784101441502571, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.82} +{'loss': 0.0123, 'grad_norm': 0.05964573472738266, 'learning_rate': 3.7548971712516e-06, 'flow_matching_loss': 0.012299287132918835, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.82} +{'loss': 0.0135, 'grad_norm': 0.09415420889854431, 'learning_rate': 3.743961212594197e-06, 'flow_matching_loss': 0.013486617524176837, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.82} +{'loss': 0.013, 'grad_norm': 0.05950843542814255, 'learning_rate': 3.7330389310852975e-06, 'flow_matching_loss': 0.013028914667665958, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.83} +{'loss': 0.0148, 'grad_norm': 0.07325046509504318, 'learning_rate': 3.7221303399965008e-06, 'flow_matching_loss': 0.014827377535402775, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.83} +{'loss': 0.0127, 'grad_norm': 0.09654416888952255, 'learning_rate': 3.711235452582775e-06, 'flow_matching_loss': 0.01272427411749959, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.83} +{'loss': 0.0126, 'grad_norm': 0.06549131870269775, 'learning_rate': 3.7003542820824347e-06, 'flow_matching_loss': 0.012641285918653012, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.83} +{'loss': 0.0143, 'grad_norm': 0.07705140113830566, 'learning_rate': 3.6894868417171252e-06, 'flow_matching_loss': 0.01434050127863884, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.84} +{'loss': 0.0129, 'grad_norm': 0.06565580517053604, 'learning_rate': 3.678633144691818e-06, 'flow_matching_loss': 0.012918719463050366, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.84} +{'loss': 0.0132, 'grad_norm': 0.06412441283464432, 'learning_rate': 3.667793204194776e-06, 'flow_matching_loss': 0.01315924758091569, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.84} +{'loss': 0.0108, 'grad_norm': 0.06377682834863663, 'learning_rate': 3.656967033397544e-06, 'flow_matching_loss': 0.010768975596874952, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.85} +{'loss': 0.0142, 'grad_norm': 0.11339374631643295, 'learning_rate': 3.646154645454948e-06, 'flow_matching_loss': 0.014235550072044134, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.85} +{'loss': 0.0126, 'grad_norm': 0.06091293692588806, 'learning_rate': 3.6353560535050524e-06, 'flow_matching_loss': 0.012568653002381325, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.85} +{'loss': 0.0123, 'grad_norm': 0.06215246394276619, 'learning_rate': 3.624571270669172e-06, 'flow_matching_loss': 0.012308130413293839, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.85} +{'loss': 0.0131, 'grad_norm': 0.07203428447246552, 'learning_rate': 3.6138003100518296e-06, 'flow_matching_loss': 0.013055917713791132, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.86} +{'loss': 0.0131, 'grad_norm': 0.054761312901973724, 'learning_rate': 3.6030431847407595e-06, 'flow_matching_loss': 0.013073442317545415, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.86} +{'loss': 0.0143, 'grad_norm': 0.10924729704856873, 'learning_rate': 3.592299907806881e-06, 'flow_matching_loss': 0.01434020409360528, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.86} +{'loss': 0.0124, 'grad_norm': 0.07767850160598755, 'learning_rate': 3.5815704923042892e-06, 'flow_matching_loss': 0.012352512683719397, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.87} +{'loss': 0.0135, 'grad_norm': 0.07494913041591644, 'learning_rate': 3.5708549512702392e-06, 'flow_matching_loss': 0.013451638445258141, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.87} +{'loss': 0.0139, 'grad_norm': 0.08847980201244354, 'learning_rate': 3.5601532977251217e-06, 'flow_matching_loss': 0.013869556691497564, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.87} 79%|███████▊ | 23620/30000 [14:27:19<3:06:08, 1.75s/it] 79%|███████▊ | 23621/30000 [14:27:20<2:48:13, 1.58s/it] 79%|███████▊ | 23622/30000 [14:27:22<2:35:40, 1.46s/it] 79%|███████▊ | 23623/30000 [14:27:23<2:26:33, 1.38s/it] 79%|███████▊ | 23624/30000 [14:27:24<2:21:05, 1.33s/it] 79%|███████▉ | 23625/30000 [14:27:30<4:44:37, 2.68s/it] 79%|███████▉ | 23626/30000 [14:27:31<3:56:50, 2.23s/it] 79%|███████▉ | 23627/30000 [14:27:32<3:23:50, 1.92s/it] 79%|███████▉ | 23628/30000 [14:27:33<3:00:24, 1.70s/it] 79%|███████▉ | 23629/30000 [14:27:35<2:44:33, 1.55s/it] 79%|███████▉ | 23630/30000 [14:27:36<2:33:10, 1.44s/it] 79%|███████▉ | 23630/30000 [14:27:36<2:33:10, 1.44s/it] 79%|███████▉ | 23631/30000 [14:27:37<2:25:22, 1.37s/it] 79%|███████▉ | 23632/30000 [14:27:38<2:20:11, 1.32s/it] 79%|███████▉ | 23633/30000 [14:27:45<5:14:59, 2.97s/it] 79%|███████▉ | 23634/30000 [14:27:46<4:18:19, 2.43s/it] 79%|███████▉ | 23635/30000 [14:27:47<3:38:56, 2.06s/it] 79%|███████▉ | 23636/30000 [14:27:49<3:11:23, 1.80s/it] 79%|███████▉ | 23637/30000 [14:27:50<2:52:18, 1.62s/it] 79%|███████▉ | 23638/30000 [14:27:51<2:39:09, 1.50s/it] 79%|███████▉ | 23639/30000 [14:27:52<2:28:40, 1.40s/it] 79%|███████▉ | 23640/30000 [14:27:53<2:22:47, 1.35s/it] 79%|███████▉ | 23640/30000 [14:27:53<2:22:47, 1.35s/it] 79%|███████▉ | 23641/30000 [14:28:01<5:32:47, 3.14s/it] 79%|███████▉ | 23642/30000 [14:28:02<4:30:37, 2.55s/it] 79%|███████▉ | 23643/30000 [14:28:03<3:47:28, 2.15s/it] 79%|███████▉ | 23644/30000 [14:28:04<3:17:15, 1.86s/it] 79%|███████▉ | 23645/30000 [14:28:05<2:55:38, 1.66s/it] 79%|███████▉ | 23646/30000 [14:28:07<2:40:30, 1.52s/it] 79%|███████▉ | 23647/30000 [14:28:08<2:30:33, 1.42s/it] 79%|███████▉ | 23648/30000 [14:28:09<2:24:06, 1.36s/it] 79%|███████▉ | 23649/30000 [14:28:15<4:55:42, 2.79s/it] 79%|███████▉ | 23650/30000 [14:28:16<4:03:47, 2.30s/it] 79%|███████▉ | 23650/30000 [14:28:16<4:03:47, 2.30s/it] 79%|███████▉ | 23651/30000 [14:28:18<3:29:42, 1.98s/it] 79%|███████▉ | 23652/30000 [14:28:19<3:04:45, 1.75s/it] 79%|███████▉ | 23653/30000 [14:28:20<2:48:21, 1.59s/it] 79%|███████▉ | 23654/30000 [14:28:21<2:35:42, 1.47s/it] 79%|███████▉ | 23655/30000 [14:28:22<2:26:57, 1.39s/it] 79%|███████▉ | 23656/30000 [14:28:24<2:20:59, 1.33s/it] 79%|███████▉ | 23657/30000 [14:28:30<4:47:41, 2.72s/it] 79%|███████▉ | 23658/30000 [14:28:31<3:58:40, 2.26s/it] 79%|███████▉ | 23659/30000 [14:28:32<3:24:28, 1.93s/it] 79%|███████▉ | 23660/30000 [14:28:33<3:01:18, 1.72s/it] 79%|███████▉ | 23660/30000 [14:28:33<3:01:18, 1.72s/it] 79%|███████▉ | 23661/30000 [14:28:34<2:45:27, 1.57s/it] 79%|███████▉ | 23662/30000 [14:28:36<2:33:18, 1.45s/it] 79%|███████▉ | 23663/30000 [14:28:37<2:24:38, 1.37s/it] 79%|███████▉ | 23664/30000 [14:28:38<2:19:05, 1.32s/it] 79%|███████▉ | 23665/30000 [14:28:45<5:14:43, 2.98s/it] 79%|███████▉ | 23666/30000 [14:28:46<4:17:22, 2.44s/it] 79%|███████▉ | 23667/30000 [14:28:47<3:38:08, 2.07s/it] 79%|███████▉ | 23668/30000 [14:28:48<3:10:31, 1.81s/it] 79%|███████▉ | 23669/30000 [14:28:50<2:51:28, 1.63s/it] 79%|███████▉ | 23670/30000 [14:28:51<2:38:02, 1.50s/it] 79%|███████▉ | 23670/30000 [14:28:51<2:38:02, 1.50s/it] 79%|███████▉ | 23671/30000 [14:28:52<2:27:40, 1.40s/it] 79%|███████▉ | 23672/30000 [14:28:53<2:20:50, 1.34s/it] 79%|███████▉ | 23673/30000 [14:29:00<5:13:04, 2.97s/it] 79%|███████▉ | 23674/30000 [14:29:01<4:16:23, 2.43s/it] 79%|███████▉ | 23675/30000 [14:29:02<3:37:05, 2.06s/it] 79%|███████▉ | 23676/30000 [14:29:03<3:09:50, 1.80s/it] 79%|███████▉ | 23677/30000 [14:29:05<2:50:21, 1.62s/it] 79%|███████▉ | 23678/30000 [14:29:06<2:36:26, 1.48s/it] 79%|███████▉ | 23679/30000 [14:29:07<2:27:20, 1.40s/it] 79%|███████▉ | 23680/30000 [14:29:08<2:20:35, 1.33s/it] 79%|███████▉ | 23680/30000 [14:29:08<2:20:35, 1.33s/it] 79%|███████▉ | 23681/30000 [14:29:15<5:12:28, 2.97s/it] 79%|███████▉ | 23682/30000 [14:29:16<4:15:30, 2.43s/it] 79%|███████▉ | 23683/30000 [14:29:17<3:36:57, 2.06s/it] 79%|███████▉ | 23684/30000 [14:29:19<3:10:49, 1.81s/it] 79%|███████▉ | 23685/30000 [14:29:20<2:51:45, 1.63s/it] 79%|███████▉ | 23686/30000 [14:29:21<2:37:27, 1.50s/it] 79%|███████▉ | 23687/30000 [14:29:22<2:26:41, 1.39s/it] 79%|███████▉ | 23688/30000 [14:29:23<2:20:02, 1.33s/it] 79%|███████▉ | 23689/30000 [14:29:30<5:13:18, 2.98s/it] 79%|███████▉ | 23690/30000 [14:29:31<4:15:57, 2.43s/it] 79%|███████▉ | 23690/30000 [14:29:31<4:15:57, 2.43s/it] 79%|███████▉ | 23691/30000 [14:29:32<3:37:02, 2.06s/it] 79%|███████▉ | 23692/30000 [14:29:34<3:10:19, 1.81s/it] 79%|███████▉ | 23693/30000 [14:29:35<2:50:56, 1.63s/it] 79%|███████▉ | 23694/30000 [14:29:36<2:37:11, 1.50s/it] 79%|███████▉ | 23695/30000 [14:29:37<2:27:12, 1.40s/it] 79%|███████▉ | 23696/30000 [14:29:39<2:22:11, 1.35s/it] 79%|███████▉ | 23697/30000 [14:29:45<5:08:06, 2.93s/it] 79%|███████▉ | 23698/30000 [14:29:46<4:12:40, 2.41s/it] 79%|███████▉ | 23699/30000 [14:29:48<3:34:46, 2.05s/it] 79%|███████▉ | 23700/30000 [14:29:49<3:08:17, 1.79s/it] 79%|███████▉ | 23700/30000 [14:29:49<3:08:17, 1.79s/it] 79%|███████▉ | 23701/30000 [14:29:50<2:49:51, 1.62s/it] 79%|███████▉ | 23702/30000 [14:29:51<2:37:02, 1.50s/it] 79%|███████▉ | 23703/30000 [14:29:52<2:26:48, 1.40s/it] 79%|███████▉ | 23704/30000 [14:29:54<2:20:32, 1.34s/it] 79%|███████▉ | 23705/30000 [14:30:00<5:05:16, 2.91s/it] 79%|███████▉ | 23706/30000 [14:30:01<4:10:02, 2.38s/it] 79%|███████▉ | 23707/30000 [14:30:02<3:32:43, 2.03s/it] 79%|███████▉ | 23708/30000 [14:30:04<3:06:22, 1.78s/it] 79%|███████▉ | 23709/30000 [14:30:05<2:48:59, 1.61s/it] 79%|███████▉ | 23710/30000 [14:30:06<2:35:30, 1.48s/it] 79%|███████▉ | 23710/30000 [14:30:06<2:35:30, 1.48s/it] 79%|███████▉ | 23711/30000 [14:30:07<2:25:41, 1.39s/it] 79%|███████▉ | 23712/30000 [14:30:08<2:19:35, 1.33s/it] 79%|███████▉ | 23713/30000 [14:30:15<5:03:49, 2.90s/it] 79%|███████▉ | 23714/30000 [14:30:16<4:09:17, 2.38s/it] 79%|███████▉ | 23715/30000 [14:30:17<3:31:40, 2.02s/it] 79%|███████▉ | 23716/30000 [14:30:19<3:06:04, 1.78s/it] 79%|███████▉ | 23717/30000 [14:30:20<2:47:57, 1.60s/it] 79%|███████▉ | 23718/30000 [14:30:21<2:35:26, 1.48s/it] 79%|███████▉ | 23719/30000 [14:30:22<2:26:05, 1.40s/it] 79%|███████▉ | 23720/30000 [14:30:23<2:19:44, 1.34s/it] 79%|███████▉ | 23720/30000 [14:30:23<2:19:44, 1.34s/it] 79%|███████▉ | 23721/30000 [14:30:30<4:58:10, 2.85s/it] 79%|███████▉ | 23722/30000 [14:30:31<4:06:19, 2.35s/it] 79%|███████▉ | 23723/30000 [14:30:32<3:31:46, 2.02s/it] 79%|███████▉ | 23724/30000 [14:30:33<3:05:24, 1.77s/it] 79%|███████▉ | 23725/30000 [14:30:35<2:47:20, 1.60s/it] 79%|███████▉ | 23726/30000 [14:30:36<2:34:28, 1.48s/it] 79%|███████▉ | 23727/30000 [14:30:37<2:24:58, 1.39s/it] 79%|███████▉ | 23728/30000 [14:30:38<2:19:20, 1.33s/it] 79%|███████▉ | 23729/30000 [14:30:44<4:45:32, 2.73s/it] 79%|███████▉ | 23730/30000 [14:30:45<3:56:53, 2.27s/it] 79%|███████▉ | 23730/30000 [14:30:45<3:56:53, 2.27s/it] 79%|███████▉ | 23731/30000 [14:30:47<3:24:06, 1.95s/it] 79%|███████▉ | 23732/30000 [14:30:48<3:00:10, 1.72s/it] 79%|███████▉ | 23733/30000 [14:30:49<2:43:16, 1.56s/it] 79%|███████▉ | 23734/30000 [14:30:50<2:31:33, 1.45s/it] 79%|███████▉ | 23735/30000 [14:30:51<2:22:38, 1.37s/it] 79%|███████▉ | 23736/30000 [14:30:52<2:17:11, 1.31s/it] 79%|███████▉ | 23737/30000 [14:31:00<5:20:20, 3.07s/it] 79%|███████▉ | 23738/30000 [14:31:01<4:21:01, 2.50s/it] 79%|███████▉ | 23739/30000 [14:31:02<3:40:06, 2.11s/it] 79%|███████▉ | 23740/30000 [14:31:03<3:11:34, 1.84s/it] 79%|███████▉ | 23740/30000 [14:31:03<3:11:34, 1.84s/it] 79%|███████▉ | 23741/30000 [14:31:04<2:50:56, 1.64s/it] 79%|███████▉ | 23742/30000 [14:31:06<2:36:39, 1.50s/it] 79%|███████▉ | 23743/30000 [14:31:07<2:26:43, 1.41s/it] 79%|███████▉ | 23744/30000 [14:31:08<2:20:24, 1.35s/it] 79%|███████▉ | 23745/30000 [14:31:15<5:19:02, 3.06s/it] 79%|███████▉ | 23746/30000 [14:31:16<4:20:22, 2.50s/it] 79%|███████▉ | 23747/30000 [14:31:17<3:39:22, 2.10s/it] 79%|███████▉ | 23748/30000 [14:31:19<3:10:38, 1.83s/it] 79%|███████▉ | 23749/30000 [14:31:20<2:50:48, 1.64s/it] 79%|███████▉ | 23750/30000 [14:31:21<2:36:29, 1.50s/it] 79%|███████▉ | 23750/30000 [14:31:21<2:36:29, 1.50s/it] 79%|███████▉ | 23751/30000 [14:31:22<2:26:18, 1.40s/it] 79%|███████▉ | 23752/30000 [14:31:23<2:19:33, 1.34s/it] 79%|███████▉ | 23753/30000 [14:31:30<4:59:34, 2.88s/it] 79%|███████▉ | 23754/30000 [14:31:31<4:06:07, 2.36s/it] 79%|███████▉ | 23755/30000 [14:31:32<3:29:49, 2.02s/it] 79%|███████▉ | 23756/30000 [14:31:33<3:04:13, 1.77s/it] 79%|███████▉ | 23757/30000 [14:31:35<2:46:22, 1.60s/it] 79%|███████▉ | 23758/30000 [14:31:36<2:33:26, 1.47s/it] 79%|███████▉ | 23759/30000 [14:31:37<2:24:15, 1.39s/it] 79%|███████▉ | 23760/30000 [14:31:38<2:18:19, 1.33s/it] 79%|███████▉ | 23760/30000 [14:31:38<2:18:19, 1.33s/it] 79%|███████▉ | 23761/30000 [14:31:45<5:15:45, 3.04s/it] 79%|███████▉ | 23762/30000 [14:31:46<4:17:41, 2.48s/it] 79%|███████▉ | 23763/30000 [14:31:47<3:38:18, 2.10s/it] 79%|███████▉ | 23764/30000 [14:31:49<3:10:34, 1.83s/it] 79%|███████▉ | 23765/30000 [14:31:50<2:51:01, 1.65s/it] 79%|███████▉ | 23766/30000 [14:31:51<2:36:31, 1.51s/it] 79%|███████▉ | 23767/30000 [14:31:52<2:26:23, 1.41s/it] 79%|███████▉ | 23768/30000 [14:31:53<2:20:19, 1.35s/it] 79%|███████▉ | 23769/30000 [14:32:01<5:20:52, 3.09s/it] 79%|███████▉ | 23770/30000 [14:32:02<4:21:32, 2.52s/it] 79%|███████▉ | 23770/30000 [14:32:02<4:21:32, 2.52s/it] 79%|███████▉ | 23771/30000 [14:32:03<3:40:41, 2.13s/it] 79%|███████▉ | 23772/30000 [14:32:04<3:12:04, 1.85s/it] 79%|███████▉ | 23773/30000 [14:32:05<2:51:45, 1.65s/it] 79%|███████▉ | 23774/30000 [14:32:07<2:37:42, 1.52s/it] 79%|███████▉ | 23775/30000 [14:32:08<2:27:07, 1.42s/it] 79%|███████▉ | 23776/30000 [14:32:09<2:20:04, 1.35s/it] 79%|███████▉ | 23777/30000 [14:32:16<5:16:00, 3.05s/it] 79%|███████▉ | 23778/30000 [14:32:17<4:17:58, 2.49s/it] 79%|███████▉ | 23779/30000 [14:32:18<3:37:50, 2.10s/it] 79%|███████▉ | 23780/30000 [14:32:20<3:09:00, 1.82s/it] 79%|███████▉ | 23780/30000 [14:32:20<3:09:00, 1.82s/it] 79%|███████▉ | 23781/30000 [14:32:21<2:49:32, 1.64s/it] 79%|███████▉ | 23782/30000 [14:32:22<2:35:45, 1.50s/it] 79%|███████▉ | 23783/30000 [14:32:23<2:26:26, 1.41s/it] 79%|███████▉ | 23784/30000 [14:32:24<2:19:05, 1.34s/it] 79%|███████▉ | 23785/30000 [14:32:31<5:10:45, 3.00s/it] 79%|███████▉ | 23786/30000 [14:32:32<4:14:02, 2.45s/it] 79%|███████▉ | 23787/30000 [14:32:34<3:34:51, 2.07s/it] 79%|███████▉ | 23788/30000 [14:32:35<3:07:46, 1.81s/it] 79%|███████▉ | 23789/30000 [14:32:36<2:48:29, 1.63s/it] 79%|███████▉ | 23790/30000 [14:32:37<2:35:19, 1.50s/it] 79%|███████▉ | 23790/30000 [14:32:37<2:35:19, 1.50s/it] 79%|███████▉ | 23791/30000 [14:32:38<2:26:11, 1.41s/it] 79%|███████▉ | 23792/30000 [14:32:40<2:19:46, 1.35s/it] 79%|███████▉ | 23793/30000 [14:32:47<5:21:10, 3.10s/it] 79%|███████▉ | 23794/30000 [14:32:48<4:21:01, 2.52s/it] 79%|███████▉ | 23795/30000 [14:32:49<3:39:46, 2.13s/it] 79%|███████▉ | 23796/30000 [14:32:50<3:10:43, 1.84s/it] 79%|███████▉ | 23797/30000 [14:32:52<2:51:00, 1.65s/it] 79%|███████▉ | 23798/30000 [14:32:53<2:36:10, 1.51s/it] 79%|███████▉ | 23799/30000 [14:32:54<2:26:19, 1.42s/it] 79%|███████▉ | 23800/30000 [14:32:55<2:20:07, 1.36s/it] 79%|███████▉ | 23800/30000 [14:32:55<2:20:07, 1.36s/it] 79%|███████▉ | 23801/30000 [14:33:03<5:26:56, 3.16s/it] 79%|███████▉ | 23802/30000 [14:33:04<4:25:08, 2.57s/it] 79%|███████▉ | 23803/30000 [14:33:05<3:43:05, 2.16s/it] 79%|███████▉ | 23804/30000 [14:33:06<3:14:04, 1.88s/it] 79%|███████▉ | 23805/30000 [14:33:07<2:52:51, 1.67s/it] 79%|███████▉ | 23806/30000 [14:33:09<2:37:41, 1.53s/it] 79%|███████▉ | 23807/30000 [14:33:10<2:27:32, 1.43s/it] 79%|███████▉ | 23808/30000 [14:33:11<2:21:50, 1.37s/it] 79%|███████▉ | 23809/30000 [14:33:18<5:15:29, 3.06s/it] 79%|███████▉ | 23810/30000 [14:33:19<4:16:43, 2.49s/it] 79%|███████▉ | 23810/30000 [14:33:19<4:16:43, 2.49s/it] 79%|███████▉ | 23811/30000 [14:33:20<3:36:27, 2.10s/it] 79%|███████▉ | 23812/30000 [14:33:22<3:09:37, 1.84s/it] 79%|███████▉ | 23813/30000 [14:33:23<2:49:45, 1.65s/it] 79%|███████▉ | 23814/30000 [14:33:24<2:35:36, 1.51s/it] 79%|███████▉ | 23815/30000 [14:33:25<2:26:18, 1.42s/it] 79%|███████▉ | 23816/30000 [14:33:26<2:19:44, 1.36s/it] 79%|███████▉ | 23817/30000 [14:33:33<5:18:42, 3.09s/it] 79%|███████▉ | 23818/30000 [14:33:35<4:19:37, 2.52s/it] 79%|███████▉ | 23819/30000 [14:33:36<3:39:03, 2.13s/it] 79%|███████▉ | 23820/30000 [14:33:37<3:10:30, 1.85s/it] 79%|███████▉ | 23820/30000 [14:33:37<3:10:30, 1.85s/it] 79%|███████▉ | 23821/30000 [14:33:38<2:51:21, 1.66s/it] 79%|███████▉ | 23822/30000 [14:33:39<2:36:24, 1.52s/it] 79%|███████▉ | 23823/30000 [14:33:41<2:26:07, 1.42s/it] 79%|███████▉ | 23824/30000 [14:33:42<2:19:02, 1.35s/it] 79%|███████▉ | 23825/30000 [14:33:48<4:53:58, 2.86s/it] 79%|███████▉ | 23826/30000 [14:33:49<4:01:36, 2.35s/it] 79%|███████▉ | 23827/30000 [14:33:51<3:25:50, 2.00s/it] 79%|███████▉ | 23828/30000 [14:33:52<3:00:25, 1.75s/it] 79%|███████▉ | 23829/30000 [14:33:53<2:43:38, 1.59s/it] 79%|███████▉ | 23830/30000 [14:33:54<2:31:29, 1.47s/it] 79%|███████▉ | 23830/30000 [14:33:54<2:31:29, 1.47s/it] 79%|███████▉ | 23831/30000 [14:33:55<2:22:21, 1.38s/it] 79%|███████▉ | 23832/30000 [14:33:57<2:16:11, 1.32s/it] 79%|███████▉ | 23833/30000 [14:34:03<4:39:46, 2.72s/it] 79%|███████▉ | 23834/30000 [14:34:04<3:51:57, 2.26s/it] 79%|███████▉ | 23835/30000 [14:34:05<3:19:40, 1.94s/it] 79%|███████▉ | 23836/30000 [14:34:06<2:56:04, 1.71s/it] 79%|███████▉ | 23837/30000 [14:34:07<2:39:56, 1.56s/it] 79%|███████▉ | 23838/30000 [14:34:08<2:28:06, 1.44s/it] 79%|███████▉ | 23839/30000 [14:34:10<2:25:30, 1.42s/it] 79%|███████▉ | 23840/30000 [14:34:11<2:18:57, 1.35s/it] 79%|███████▉ | 23840/30000 [14:34:11<2:18:57, 1.35s/it] 79%|███████▉ | 23841/30000 [14:34:17<4:31:35, 2.65s/it] 79%|███████▉ | 23842/30000 [14:34:18<3:46:39, 2.21s/it] 79%|███████▉ | 23843/30000 [14:34:19<3:15:34, 1.91s/it] 79%|███████▉ | 23844/30000 [14:34:20<2:53:40, 1.69s/it] 79%|███████▉ | 23845/30000 [14:34:21<2:38:19, 1.54s/it] 79%|███████▉ | 23846/30000 [14:34:23<2:27:37, 1.44s/it] 79%|███████▉ | 23847/30000 [14:34:24<2:19:53, 1.36s/it] 79%|███████▉ | 23848/30000 [14:34:25<2:14:59, 1.32s/it] 79%|███████▉ | 23849/30000 [14:34:31<4:44:14, 2.77s/it] 80%|███████▉ | 23850/30000 [14:34:32<3:55:08, 2.29s/it] 80%|███████▉ | 23850/30000 [14:34:32<3:55:08, 2.29s/it] 80%|███████▉ | 23851/30000 [14:34:34<3:20:56, 1.96s/it] 80%|███████▉ | 23852/30000 [14:34:35<2:57:18, 1.73s/it] 80%|███████▉ | 23853/30000 [14:34:36<2:41:29, 1.58s/it] 80%|███████▉ | 23854/30000 [14:34:37<2:29:22, 1.46s/it] 80%|███████▉ | 23855/30000 [14:34:38<2:21:29, 1.38s/it] 80%|███████▉ | 23856/30000 [14:34:40<2:16:00, 1.33s/it] 80%|███████▉ | 23857/30000 [14:34:47<5:15:21, 3.08s/it] 80%|███████▉ | 23858/30000 [14:34:48<4:16:45, 2.51s/it] 80%|███████▉ | 23859/30000 [14:34:49<3:36:26, 2.11s/it] 80%|███████▉ | 23860/30000 [14:34:50<3:07:58, 1.84s/it] 80%|███████▉ | 23860/30000 [14:34:50<3:07:58, 1.84s/it] 80%|███████▉ | 23861/30000 [14:34:52<2:48:28, 1.65s/it] 80%|███████▉ | 23862/30000 [14:34:53<2:33:49, 1.50s/it] 80%|███████▉ | 23863/30000 [14:34:54<2:24:09, 1.41s/it] 80%|███████▉ | 23864/30000 [14:34:55<2:17:45, 1.35s/it] 80%|███████▉ | 23865/30000 [14:35:02<5:22:04, 3.15s/it] 80%|███████▉ | 23866/30000 [14:35:04<4:21:37, 2.56s/it] 80%|███████▉ | 23867/30000 [14:35:05<3:39:46, 2.15s/it] 80%|███████▉ | 23868/30000 [14:35:06<3:09:54, 1.86s/it] 80%|███████▉ | 23869/30000 [14:35:07<2:49:26, 1.66s/it] 80%|███████▉ | 23870/30000 [14:35:08<2:34:37, 1.51s/it] 80%|███████▉ | 23870/30000 [14:35:08<2:34:37, 1.51s/it] 80%|███████▉ | 23871/30000 [14:35:10<2:24:11, 1.41s/it] 80%|███████▉ | 23872/30000 [14:35:11<2:17:37, 1.35s/it] 80%|███████▉ | 23873/30000 [14:35:18<5:07:18, 3.01s/it] 80%|███████▉ | 23874/30000 [14:35:19<4:11:15, 2.46s/it] 80%|███████▉ | 23875/30000 [14:35:20<3:32:46, 2.08s/it] 80%|███████▉ | 23876/30000 [14:35:21<3:05:27, 1.82s/it] 80%|███████▉ | 23877/30000 [14:35:22<2:46:30, 1.63s/it] 80%|███████▉ | 23878/30000 [14:35:24<2:33:22, 1.50s/it] 80%|███████▉ | 23879/30000 [14:35:25<2:22:58, 1.40s/it] 80%|███████▉ | 23880/30000 [14:35:26<2:17:19, 1.35s/it] 80%|███████▉ | 23880/30000 [14:35:26<2:17:19, 1.35s/it] 80%|███████▉ | 23881/30000 [14:35:33<4:58:58, 2.93s/it] 80%|███████▉ | 23882/30000 [14:35:34<4:05:01, 2.40s/it] 80%|███████▉ | 23883/30000 [14:35:35<3:27:22, 2.03s/it] 80%|███████▉ | 23884/30000 [14:35:36<3:01:31, 1.78s/it] 80%|███████▉ | 23885/30000 [14:35:37<2:43:14, 1.60s/it] 80%|███████▉ | 23886/30000 [14:35:39<2:30:57, 1.48s/it] 80%|███████▉ | 23887/30000 [14:35:40<2:22:00, 1.39s/it] 80%|███████▉ | 23888/30000 [14:35:41<2:15:54, 1.33s/it] 80%|███████▉ | 23889/30000 [14:35:48<5:00:40, 2.95s/it] 80%|███████▉ | 23890/30000 [14:35:49<4:06:07, 2.42s/it] 80%|███████▉ | 23890/30000 [14:35:49<4:06:07, 2.42s/it] 80%|███████▉ | 23891/30000 [14:35:50<3:28:58, 2.05s/it] 80%|███████▉ | 23892/30000 [14:35:51<3:02:07, 1.79s/it] 80%|███████▉ | 23893/30000 [14:35:52<2:44:13, 1.61s/it] 80%|███████▉ | 23894/30000 [14:35:54<2:31:25, 1.49s/it] 80%|███████▉ | 23895/30000 [14:35:55<2:20:56, 1.39s/it] 80%|███████▉ | 23896/30000 [14:35:56<2:15:07, 1.33s/it] 80%|███████▉ | 23897/30000 [14:36:03<5:01:53, 2.97s/it] 80%|███████▉ | 23898/30000 [14:36:04<4:06:42, 2.43s/it] 80%|███████▉ | 23899/30000 [14:36:05<3:29:02, 2.06s/it] 80%|███████▉ | 23900/30000 [14:36:06<3:02:47, 1.80s/it] 80%|███████▉ | 23900/30000 [14:36:06<3:02:47, 1.80s/it] 80%|███████▉ | 23901/30000 [14:36:07<2:44:18, 1.62s/it] 80%|███████▉ | 23902/30000 [14:36:09<2:31:52, 1.49s/it] 80%|███████▉ | 23903/30000 [14:36:10<2:22:12, 1.40s/it] 80%|███████▉ | 23904/30000 [14:36:11<2:15:31, 1.33s/it] 80%|███████▉ | 23905/30000 [14:36:18<5:09:24, 3.05s/it] 80%|███████▉ | 23906/30000 [14:36:19<4:12:22, 2.48s/it] 80%|███████▉ | 23907/30000 [14:36:20<3:33:04, 2.10s/it] 80%|███████▉ | 23908/30000 [14:36:22<3:06:03, 1.83s/it] 80%|███████▉ | 23909/30000 [14:36:23<2:46:42, 1.64s/it] 80%|███████▉ | 23910/30000 [14:36:24<2:33:09, 1.51s/it] 80%|███████▉ | 23910/30000 [14:36:24<2:33:09, 1.51s/it] 80%|███████▉ | 23911/30000 [14:36:25<2:24:20, 1.42s/it] 80%|███████▉ | 23912/30000 [14:36:26<2:17:40, 1.36s/it] 80%|███████▉ | 23913/30000 [14:36:33<5:05:04, 3.01s/it] 80%|███████▉ | 23914/30000 [14:36:34<4:08:56, 2.45s/it] 80%|███████▉ | 23915/30000 [14:36:36<3:29:49, 2.07s/it] 80%|███████▉ | 23916/30000 [14:36:37<3:02:55, 1.80s/it] 80%|███████▉ | 23917/30000 [14:36:38<2:43:51, 1.62s/it] 80%|███████▉ | 23918/30000 [14:36:39<2:30:37, 1.49s/it] 80%|███████▉ | 23919/30000 [14:36:40<2:21:36, 1.40s/it] 80%|███████▉ | 23920/30000 [14:36:42<2:15:55, 1.34s/it] 80%|███████▉ | 23920/30000 [14:36:42<2:15:55, 1.34s/it] 80%|███████▉ | 23921/30000 [14:36:49<5:09:34, 3.06s/it] 80%|███████▉ | 23922/30000 [14:36:50<4:13:06, 2.50s/it] 80%|███████▉ | 23923/30000 [14:36:51<3:33:24, 2.11s/it] 80%|███████▉ | 23924/30000 [14:36:52<3:05:48, 1.83s/it] 80%|███████▉ | 23925/30000 [14:36:53<2:46:44, 1.65s/it] 80%|███████▉ | 23926/30000 [14:36:55<2:32:45, 1.51s/it] 80%|███████▉ | 23927/30000 [14:36:56<2:22:30, 1.41s/it] 80%|███████▉ | 23928/30000 [14:36:57<2:16:06, 1.34s/it] 80%|███████▉ | 23929/30000 [14:37:04<5:05:11, 3.02s/it] 80%|███████▉ | 23930/30000 [14:37:05<4:10:00, 2.47s/it] 80%|███████▉ | 23930/30000 [14:37:05<4:10:00, 2.47s/it] 80%|███████▉ | 23931/30000 [14:37:06<3:31:24, 2.09s/it] 80%|███████▉ | 23932/30000 [14:37:08<3:03:57, 1.82s/it] 80%|███████▉ | 23933/30000 [14:37:09<2:45:15, 1.63s/it] 80%|███████▉ | 23934/30000 [14:37:10<2:32:30, 1.51s/it] 80%|███████▉ | 23935/30000 [14:37:11<2:22:51, 1.41s/it] 80%|███████▉ | 23936/30000 [14:37:12<2:17:33, 1.36s/it] 80%|███████▉ | 23937/30000 [14:37:19<4:57:13, 2.94s/it] 80%|███████▉ | 23938/30000 [14:37:20<4:04:20, 2.42s/it] 80%|███████▉ | 23939/30000 [14:37:21<3:27:05, 2.05s/it] 80%|███████▉ | 23940/30000 [14:37:23<3:01:50, 1.80s/it] 80%|███████▉ | 23940/30000 [14:37:23<3:01:50, 1.80s/it] 80%|███████▉ | 23941/30000 [14:37:24<2:43:44, 1.62s/it] 80%|███████▉ | 23942/30000 [14:37:25<2:30:44, 1.49s/it] 80%|███████▉ | 23943/30000 [14:37:26<2:21:11, 1.40s/it] 80%|███████▉ | 23944/30000 [14:37:27<2:15:07, 1.34s/it] 80%|███████▉ | 23945/30000 [14:37:34<4:56:32, 2.94s/it] 80%|███████▉ | 23946/30000 [14:37:35<4:04:03, 2.42s/it] 80%|███████▉ | 23947/30000 [14:37:36<3:27:34, 2.06s/it] 80%|███████▉ | 23948/30000 [14:37:38<3:02:43, 1.81s/it] 80%|███████▉ | 23949/30000 [14:37:39<2:43:32, 1.62s/it] 80%|███████▉ | 23950/30000 [14:37:40<2:30:31, 1.49s/it] 80%|███████▉ | 23950/30000 [14:37:40<2:30:31, 1.49s/it] 80%|███████▉ | 23951/30000 [14:37:41<2:21:15, 1.40s/it] 80%|███████▉ | 23952/30000 [14:37:42<2:15:27, 1.34s/it] 80%|███████▉ | 23953/30000 [14:37:50<5:12:25, 3.10s/it] 80%|███████▉ | 23954/30000 [14:37:51<4:13:57, 2.52s/it] 80%|███████▉ | 23955/30000 [14:37:52<3:33:26, 2.12s/it] 80%|███████▉ | 23956/30000 [14:37:53<3:05:08, 1.84s/it] 80%|███████▉ | 23957/30000 [14:37:54<2:46:55, 1.66s/it] 80%|███████▉ | 23958/30000 [14:37:56<2:33:20, 1.52s/it] 80%|███████▉ | 23959/30000 [14:37:57<2:22:30, 1.42s/it] 80%|███████▉ | 23960/30000 [14:37:58<2:15:30, 1.35s/it] 80%|███████▉ | 23960/30000 [14:37:58<2:15:30, 1.35s/it] 80%|███████▉ | 23961/30000 [14:38:04<4:46:12, 2.84s/it] 80%|███████▉ | 23962/30000 [14:38:06<3:55:37, 2.34s/it] 80%|███████▉ | 23963/30000 [14:38:07<3:20:38, 1.99s/it] 80%|███████▉ | 23964/30000 [14:38:08<2:56:07, 1.75s/it] 80%|███████▉ | 23965/30000 [14:38:09<2:39:34, 1.59s/it] 80%|███████▉ | 23966/30000 [14:38:10<2:27:20, 1.47s/it] 80%|███████▉ | 23967/30000 [14:38:11<2:18:38, 1.38s/it] 80%|███████▉ | 23968/30000 [14:38:13<2:13:32, 1.33s/it] 80%|███████▉ | 23969/30000 [14:38:19<4:51:41, 2.90s/it] 80%|███████▉ | 23970/30000 [14:38:20<3:59:57, 2.39s/it] 80%|███████▉ | 23970/30000 [14:38:20<3:59:57, 2.39s/it] 80%|███████▉ | 23971/30000 [14:38:22<3:23:58, 2.03s/it] 80%|███████▉ | 23972/30000 [14:38:23<2:59:43, 1.79s/it] 80%|███████▉ | 23973/30000 [14:38:24<2:43:06, 1.62s/it] 80%|███████▉ | 23974/30000 [14:38:25<2:29:36, 1.49s/it] 80%|███████▉ | 23975/30000 [14:38:26<2:21:13, 1.41s/it] 80%|███████▉ | 23976/30000 [14:38:28<2:15:58, 1.35s/it] 80%|███████▉ | 23977/30000 [14:38:36<5:30:37, 3.29s/it] 80%|███████▉ | 23978/30000 [14:38:37<4:26:38, 2.66s/it] 80%|███████▉ | 23979/30000 [14:38:38<3:42:30, 2.22s/it] 80%|███████▉ | 23980/30000 [14:38:39<3:11:33, 1.91s/it] 80%|███████▉ | 23980/30000 [14:38:39<3:11:33, 1.91s/it] 80%|███████▉ | 23981/30000 [14:38:40<2:50:28, 1.70s/it] 80%|███████▉ | 23982/30000 [14:38:41<2:35:06, 1.55s/it] 80%|███████▉ | 23983/30000 [14:38:43<2:24:04, 1.44s/it] 80%|███████▉ | 23984/30000 [14:38:44<2:18:29, 1.38s/it] 80%|███████▉ | 23985/30000 [14:38:51<5:09:21, 3.09s/it] 80%|███████▉ | 23986/30000 [14:38:52<4:11:37, 2.51s/it] 80%|███████▉ | 23987/30000 [14:38:53<3:31:29, 2.11s/it] 80%|███████▉ | 23988/30000 [14:38:54<3:03:15, 1.83s/it] 80%|███████▉ | 23989/30000 [14:38:56<2:43:25, 1.63s/it] 80%|███████▉ | 23990/30000 [14:38:57<2:30:18, 1.50s/it] 80%|███████▉ | 23990/30000 [14:38:57<2:30:18, 1.50s/it] 80%|███████▉ | 23991/30000 [14:38:58<2:20:12, 1.40s/it] 80%|███████▉ | 23992/30000 [14:38:59<2:13:24, 1.33s/it] 80%|███████▉ | 23993/30000 [14:39:06<4:52:24, 2.92s/it] 80%|███████▉ | 23994/30000 [14:39:07<4:00:06, 2.40s/it] 80%|███████▉ | 23995/30000 [14:39:08<3:25:09, 2.05s/it] 80%|███████▉ | 23996/30000 [14:39:09<2:59:58, 1.80s/it] 80%|███████▉ | 23997/30000 [14:39:11<2:41:49, 1.62s/it] 80%|███████▉ | 23998/30000 [14:39:12<2:29:16, 1.49s/it] 80%|███████▉ | 23999/30000 [14:39:13<2:19:34, 1.40s/it] 80%|████████ | 24000/30000 [14:39:14<2:12:58, 1.33s/it] 80%|████████ | 24000/30000 [14:39:14<2:12:58, 1.33s/it] 80%|████████ | 24001/30000 [14:39:21<4:55:45, 2.96s/it] 80%|████████ | 24002/30000 [14:39:22<4:02:35, 2.43s/it] 80%|████████ | 24003/30000 [14:39:23<3:25:30, 2.06s/it] 80%|████████ | 24004/30000 [14:39:25<3:00:11, 1.80s/it] 80%|████████ | 24005/30000 [14:39:26<2:42:50, 1.63s/it] 80%|████████ | 24006/30000 [14:39:27<2:30:10, 1.50s/it] 80%|████████ | 24007/30000 [14:39:28<2:20:59, 1.41s/it] 80%|████████ | 24008/30000 [14:39:29<2:14:22, 1.35s/it] 80%|████████ | 24009/30000 [14:39:36<4:56:03, 2.96s/it] 80%|████████ | 24010/30000 [14:39:37<4:02:38, 2.43s/it] 80%|████████ | 24010/30000 [14:39:37<4:02:38, 2.43s/it] 80%|████████ | 24011/30000 [14:39:38<3:25:59, 2.06s/it] 80%|████████ | 24012/30000 [14:39:40<2:59:30, 1.80s/it] 80%|████████ | 24013/30000 [14:39:41<2:41:43, 1.62s/it] 80%|████████ | 24014/30000 [14:39:43<2:50:34, 1.71s/it] 80%|████████ | 24015/30000 [14:39:44<2:34:36, 1.55s/it] 80%|████████ | 24016/30000 [14:39:45<2:24:55, 1.45s/it] 80%|████████ | 24017/30000 [14:39:52<4:57:25, 2.98s/it] 80%|████████ | 24018/30000 [14:39:53<4:03:08, 2.44s/it] 80%|████████ | 24019/30000 [14:39:54<3:25:55, 2.07s/it] 80%|████████ | 24020/30000 [14:39:55<3:00:44, 1.81s/it] 80%|████████ | 24020/30000 [14:39:55<3:00:44, 1.81s/it] 80%|████████ | 24021/30000 [14:39:57<2:42:31, 1.63s/it] 80%|████████ | 24022/30000 [14:39:59<3:09:20, 1.90s/it] 80%|████████ | 24023/30000 [14:40:00<2:47:42, 1.68s/it] 80%|████████ | 24024/30000 [14:40:01<2:33:07, 1.54s/it] 80%|████████ | 24025/30000 [14:40:07<4:21:52, 2.63s/it] 80%|████████ | 24026/30000 [14:40:08<3:38:17, 2.19s/it] 80%|████████ | 24027/30000 [14:40:09<3:08:52, 1.90s/it] 80%|████████ | 24028/30000 [14:40:10<2:47:43, 1.69s/it] 80%|████████ | 24029/30000 [14:40:11<2:32:33, 1.53s/it] 80%|████████ | 24030/30000 [14:40:13<2:29:48, 1.51s/it] +{'loss': 0.0123, 'grad_norm': 0.06703509390354156, 'learning_rate': 3.5494655446724544e-06, 'flow_matching_loss': 0.012346963956952095, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.88} +{'loss': 0.0134, 'grad_norm': 0.07025184482336044, 'learning_rate': 3.538791705098866e-06, 'flow_matching_loss': 0.013439282868057489, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.88} +{'loss': 0.0136, 'grad_norm': 0.08834854513406754, 'learning_rate': 3.5281317919740718e-06, 'flow_matching_loss': 0.013571306783705949, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.88} +{'loss': 0.0115, 'grad_norm': 0.07961033284664154, 'learning_rate': 3.5174858182508825e-06, 'flow_matching_loss': 0.011514554172754288, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.88} +{'loss': 0.0143, 'grad_norm': 0.06444607675075531, 'learning_rate': 3.506853796865153e-06, 'flow_matching_loss': 0.01433885945007205, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.89} +{'loss': 0.0127, 'grad_norm': 0.07557566463947296, 'learning_rate': 3.496235740735789e-06, 'flow_matching_loss': 0.01267557665705681, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.89} +{'loss': 0.0132, 'grad_norm': 0.08679758012294769, 'learning_rate': 3.485631662764738e-06, 'flow_matching_loss': 0.01323378849774599, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.89} +{'loss': 0.0135, 'grad_norm': 0.06883301585912704, 'learning_rate': 3.475041575836952e-06, 'flow_matching_loss': 0.013517966307699681, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.9} +{'loss': 0.0127, 'grad_norm': 0.06750419735908508, 'learning_rate': 3.4644654928203796e-06, 'flow_matching_loss': 0.012650252226740122, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.9} +{'loss': 0.0121, 'grad_norm': 0.06011442467570305, 'learning_rate': 3.4539034265659663e-06, 'flow_matching_loss': 0.012091747112572193, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.9} +{'loss': 0.013, 'grad_norm': 0.07578044384717941, 'learning_rate': 3.4433553899076163e-06, 'flow_matching_loss': 0.012967042252421379, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.9} +{'loss': 0.0133, 'grad_norm': 0.05966169387102127, 'learning_rate': 3.432821395662189e-06, 'flow_matching_loss': 0.013312952546402811, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.91} +{'loss': 0.0139, 'grad_norm': 0.07922252267599106, 'learning_rate': 3.4223014566294785e-06, 'flow_matching_loss': 0.01387314973399043, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.91} +{'loss': 0.0149, 'grad_norm': 0.08457576483488083, 'learning_rate': 3.4117955855922e-06, 'flow_matching_loss': 0.014905789587646722, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.91} +{'loss': 0.013, 'grad_norm': 0.05490583926439285, 'learning_rate': 3.4013037953159855e-06, 'flow_matching_loss': 0.012990378495305776, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.92} +{'loss': 0.0143, 'grad_norm': 0.07260622084140778, 'learning_rate': 3.390826098549345e-06, 'flow_matching_loss': 0.014326185639947652, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.92} +{'loss': 0.012, 'grad_norm': 0.0552930124104023, 'learning_rate': 3.3803625080236684e-06, 'flow_matching_loss': 0.012019568216055631, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.92} +{'loss': 0.0155, 'grad_norm': 0.11994647234678268, 'learning_rate': 3.369913036453205e-06, 'flow_matching_loss': 0.01547076292335987, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.92} +{'loss': 0.0132, 'grad_norm': 0.06702899187803268, 'learning_rate': 3.3594776965350434e-06, 'flow_matching_loss': 0.013233691826462746, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.93} +{'loss': 0.014, 'grad_norm': 0.06085817888379097, 'learning_rate': 3.34905650094911e-06, 'flow_matching_loss': 0.013987453561276198, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.93} +{'loss': 0.0124, 'grad_norm': 0.06913836300373077, 'learning_rate': 3.338649462358145e-06, 'flow_matching_loss': 0.01239976929500699, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.93} +{'loss': 0.0128, 'grad_norm': 0.06143622472882271, 'learning_rate': 3.3282565934076765e-06, 'flow_matching_loss': 0.012844567652791739, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.94} +{'loss': 0.0133, 'grad_norm': 0.07414962351322174, 'learning_rate': 3.31787790672602e-06, 'flow_matching_loss': 0.013333524577319622, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.94} +{'loss': 0.0122, 'grad_norm': 0.0696955993771553, 'learning_rate': 3.307513414924255e-06, 'flow_matching_loss': 0.012160107120871544, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.94} +{'loss': 0.0141, 'grad_norm': 0.06021328270435333, 'learning_rate': 3.2971631305962246e-06, 'flow_matching_loss': 0.014105782844126225, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.95} +{'loss': 0.0126, 'grad_norm': 0.05701705440878868, 'learning_rate': 3.2868270663184964e-06, 'flow_matching_loss': 0.012647306313738227, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.95} +{'loss': 0.0142, 'grad_norm': 0.08307893574237823, 'learning_rate': 3.276505234650362e-06, 'flow_matching_loss': 0.014173644315451383, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.95} +{'loss': 0.0133, 'grad_norm': 0.07117629796266556, 'learning_rate': 3.266197648133821e-06, 'flow_matching_loss': 0.0133087407797575, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.95} +{'loss': 0.0135, 'grad_norm': 0.09680023044347763, 'learning_rate': 3.255904319293562e-06, 'flow_matching_loss': 0.013530096691101789, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.96} +{'loss': 0.0127, 'grad_norm': 0.06917626410722733, 'learning_rate': 3.2456252606369485e-06, 'flow_matching_loss': 0.012747881188988685, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.96} +{'loss': 0.0121, 'grad_norm': 0.06282377243041992, 'learning_rate': 3.235360484654011e-06, 'flow_matching_loss': 0.012068322021514177, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.96} +{'loss': 0.0131, 'grad_norm': 0.07004904747009277, 'learning_rate': 3.225110003817418e-06, 'flow_matching_loss': 0.01312935994938016, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.97} +{'loss': 0.0142, 'grad_norm': 0.11056961119174957, 'learning_rate': 3.214873830582467e-06, 'flow_matching_loss': 0.014184669684618711, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.97} +{'loss': 0.0131, 'grad_norm': 0.06396222859621048, 'learning_rate': 3.2046519773870803e-06, 'flow_matching_loss': 0.013077223859727382, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.97} +{'loss': 0.0138, 'grad_norm': 0.06966537982225418, 'learning_rate': 3.1944444566517654e-06, 'flow_matching_loss': 0.01376980347558856, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.97} +{'loss': 0.0135, 'grad_norm': 0.0663403794169426, 'learning_rate': 3.18425128077963e-06, 'flow_matching_loss': 0.013524494040757418, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.98} +{'loss': 0.0135, 'grad_norm': 0.08207444101572037, 'learning_rate': 3.174072462156342e-06, 'flow_matching_loss': 0.013540730904787779, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.98} +{'loss': 0.0116, 'grad_norm': 0.061608705669641495, 'learning_rate': 3.163908013150124e-06, 'flow_matching_loss': 0.011592795746400953, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.98} +{'loss': 0.0124, 'grad_norm': 0.07592878490686417, 'learning_rate': 3.153757946111741e-06, 'flow_matching_loss': 0.012418582849204541, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.99} +{'loss': 0.0124, 'grad_norm': 0.07717892527580261, 'learning_rate': 3.1436222733744776e-06, 'flow_matching_loss': 0.012359482422471046, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.99} +{'loss': 0.012, 'grad_norm': 0.06441289931535721, 'learning_rate': 3.133501007254138e-06, 'flow_matching_loss': 0.012033763714134693, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.99} + 80%|████████ | 24030/30000 [14:40:13<2:29:48, 1.51s/it] 80%|████████ | 24031/30000 [14:40:14<2:20:18, 1.41s/it] 80%|████████ | 24032/30000 [14:40:15<2:14:10, 1.35s/it] 80%|████████ | 24033/30000 [14:40:20<4:03:32, 2.45s/it] 80%|████████ | 24034/30000 [14:40:21<3:25:20, 2.07s/it] 80%|████████ | 24035/30000 [14:40:23<2:59:12, 1.80s/it] 80%|████████ | 24036/30000 [14:40:24<2:40:48, 1.62s/it] 80%|████████ | 24037/30000 [14:40:25<2:28:20, 1.49s/it] 80%|████████ | 24038/30000 [14:40:26<2:20:25, 1.41s/it] 80%|████████ | 24039/30000 [14:40:27<2:15:16, 1.36s/it] 80%|████████ | 24040/30000 [14:40:29<2:10:39, 1.32s/it] 80%|████████ | 24040/30000 [14:40:29<2:10:39, 1.32s/it] 80%|████████ | 24041/30000 [14:40:35<4:46:59, 2.89s/it] 80%|████████ | 24042/30000 [14:40:36<3:57:07, 2.39s/it] 80%|████████ | 24043/30000 [14:40:38<3:21:34, 2.03s/it] 80%|████████ | 24044/30000 [14:40:39<2:56:48, 1.78s/it] 80%|████████ | 24045/30000 [14:40:40<2:39:22, 1.61s/it] 80%|████████ | 24046/30000 [14:40:41<2:27:11, 1.48s/it] 80%|████████ | 24047/30000 [14:40:42<2:18:15, 1.39s/it] 80%|████████ | 24048/30000 [14:40:44<2:12:27, 1.34s/it] 80%|████████ | 24049/30000 [14:40:51<5:07:22, 3.10s/it] 80%|████████ | 24050/30000 [14:40:52<4:09:55, 2.52s/it] 80%|████████ | 24050/30000 [14:40:52<4:09:55, 2.52s/it] 80%|████████ | 24051/30000 [14:40:53<3:30:25, 2.12s/it] 80%|████████ | 24052/30000 [14:40:54<3:02:59, 1.85s/it] 80%|████████ | 24053/30000 [14:40:56<2:43:29, 1.65s/it] 80%|████████ | 24054/30000 [14:40:58<3:09:26, 1.91s/it] 80%|████████ | 24055/30000 [14:40:59<2:47:13, 1.69s/it] 80%|████████ | 24056/30000 [14:41:00<2:32:06, 1.54s/it] 80%|████████ | 24057/30000 [14:41:10<6:18:58, 3.83s/it] 80%|████████ | 24058/30000 [14:41:11<4:59:29, 3.02s/it] 80%|████████ | 24059/30000 [14:41:11<3:42:43, 2.25s/it] 80%|████████ | 24060/30000 [14:41:28<10:51:55, 6.59s/it] 80%|████████ | 24060/30000 [14:41:28<10:51:55, 6.59s/it] 80%|████████ | 24061/30000 [14:41:29<8:11:55, 4.97s/it] 80%|████████ | 24062/30000 [14:41:30<6:18:23, 3.82s/it] 80%|████████ | 24063/30000 [14:41:31<4:59:15, 3.02s/it] 80%|████████ | 24064/30000 [14:41:33<4:04:45, 2.47s/it] 80%|████████ | 24065/30000 [14:41:34<3:26:50, 2.09s/it] 80%|████████ | 24066/30000 [14:41:35<3:00:40, 1.83s/it] 80%|████████ | 24067/30000 [14:41:43<6:09:12, 3.73s/it] 80%|████████ | 24068/30000 [14:41:44<4:52:55, 2.96s/it] 80%|████████ | 24069/30000 [14:41:46<4:00:20, 2.43s/it] 80%|████████ | 24070/30000 [14:41:47<3:22:51, 2.05s/it] 80%|████████ | 24070/30000 [14:41:47<3:22:51, 2.05s/it] 80%|████████ | 24071/30000 [14:41:48<2:56:23, 1.78s/it] 80%|████████ | 24072/30000 [14:41:49<2:38:51, 1.61s/it] 80%|████████ | 24073/30000 [14:41:50<2:27:01, 1.49s/it] 80%|████████ | 24074/30000 [14:41:51<2:18:48, 1.41s/it] 80%|████████ | 24075/30000 [14:41:59<5:29:46, 3.34s/it] 80%|████████ | 24076/30000 [14:42:01<4:25:16, 2.69s/it] 80%|████████ | 24077/30000 [14:42:02<3:40:57, 2.24s/it] 80%|████████ | 24078/30000 [14:42:03<3:09:48, 1.92s/it] 80%|████████ | 24079/30000 [14:42:04<2:48:07, 1.70s/it] 80%|████████ | 24080/30000 [14:42:05<2:36:23, 1.59s/it] 80%|████████ | 24080/30000 [14:42:05<2:36:23, 1.59s/it] 80%|████████ | 24081/30000 [14:42:07<2:24:39, 1.47s/it] 80%|████████ | 24082/30000 [14:42:08<2:17:05, 1.39s/it] 80%|████████ | 24083/30000 [14:42:15<5:00:52, 3.05s/it] 80%|████████ | 24084/30000 [14:42:16<4:05:39, 2.49s/it] 80%|████████ | 24085/30000 [14:42:17<3:26:42, 2.10s/it] 80%|████████ | 24086/30000 [14:42:18<2:58:58, 1.82s/it] 80%|████████ | 24087/30000 [14:42:19<2:40:30, 1.63s/it] 80%|████████ | 24088/30000 [14:42:21<2:27:08, 1.49s/it] 80%|████████ | 24089/30000 [14:42:22<2:18:16, 1.40s/it] 80%|████████ | 24090/30000 [14:42:23<2:11:37, 1.34s/it] 80%|████████ | 24090/30000 [14:42:23<2:11:37, 1.34s/it] 80%|████████ | 24091/30000 [14:42:30<5:13:44, 3.19s/it] 80%|████████ | 24092/30000 [14:42:32<4:13:56, 2.58s/it] 80%|████████ | 24093/30000 [14:42:33<3:34:43, 2.18s/it] 80%|████████ | 24094/30000 [14:42:34<3:05:38, 1.89s/it] 80%|████████ | 24095/30000 [14:42:35<2:44:15, 1.67s/it] 80%|████████ | 24096/30000 [14:42:36<2:30:06, 1.53s/it] 80%|████████ | 24097/30000 [14:42:38<2:20:06, 1.42s/it] 80%|████████ | 24098/30000 [14:42:39<2:13:34, 1.36s/it] 80%|████████ | 24099/30000 [14:42:46<4:55:30, 3.00s/it] 80%|████████ | 24100/30000 [14:42:47<4:01:09, 2.45s/it] 80%|████████ | 24100/30000 [14:42:47<4:01:09, 2.45s/it] 80%|████████ | 24101/30000 [14:42:48<3:24:05, 2.08s/it] 80%|████████ | 24102/30000 [14:42:49<2:57:33, 1.81s/it] 80%|████████ | 24103/30000 [14:42:50<2:39:02, 1.62s/it] 80%|████████ | 24104/30000 [14:42:52<2:26:41, 1.49s/it] 80%|████████ | 24105/30000 [14:42:53<2:17:45, 1.40s/it] 80%|████████ | 24106/30000 [14:42:54<2:11:19, 1.34s/it] 80%|████████ | 24107/30000 [14:43:01<4:47:39, 2.93s/it] 80%|████████ | 24108/30000 [14:43:02<3:55:53, 2.40s/it] 80%|████████ | 24109/30000 [14:43:03<3:21:26, 2.05s/it] 80%|████████ | 24110/30000 [14:43:04<2:56:16, 1.80s/it] 80%|████████ | 24110/30000 [14:43:04<2:56:16, 1.80s/it] 80%|████████ | 24111/30000 [14:43:05<2:37:49, 1.61s/it] 80%|████████ | 24112/30000 [14:43:07<2:25:20, 1.48s/it] 80%|████████ | 24113/30000 [14:43:08<2:17:02, 1.40s/it] 80%|████████ | 24114/30000 [14:43:09<2:11:27, 1.34s/it] 80%|████████ | 24115/30000 [14:43:17<5:29:27, 3.36s/it] 80%|████████ | 24116/30000 [14:43:18<4:24:33, 2.70s/it] 80%|████████ | 24117/30000 [14:43:19<3:40:27, 2.25s/it] 80%|████████ | 24118/30000 [14:43:21<3:08:43, 1.93s/it] 80%|████████ | 24119/30000 [14:43:22<2:46:44, 1.70s/it] 80%|████████ | 24120/30000 [14:43:23<2:32:05, 1.55s/it] 80%|████████ | 24120/30000 [14:43:23<2:32:05, 1.55s/it] 80%|████████ | 24121/30000 [14:43:24<2:21:53, 1.45s/it] 80%|████████ | 24122/30000 [14:43:26<2:24:31, 1.48s/it] 80%|████████ | 24123/30000 [14:43:33<5:23:32, 3.30s/it] 80%|████████ | 24124/30000 [14:43:34<4:20:44, 2.66s/it] 80%|████████ | 24125/30000 [14:43:36<3:37:37, 2.22s/it] 80%|████████ | 24126/30000 [14:43:37<3:06:49, 1.91s/it] 80%|████████ | 24127/30000 [14:43:38<2:45:04, 1.69s/it] 80%|████████ | 24128/30000 [14:43:39<2:30:59, 1.54s/it] 80%|████████ | 24129/30000 [14:43:40<2:21:48, 1.45s/it] 80%|████████ | 24130/30000 [14:43:42<2:14:47, 1.38s/it] 80%|████████ | 24130/30000 [14:43:42<2:14:47, 1.38s/it] 80%|████████ | 24131/30000 [14:43:49<4:56:18, 3.03s/it] 80%|████████ | 24132/30000 [14:43:50<4:02:58, 2.48s/it] 80%|████████ | 24133/30000 [14:43:51<3:25:38, 2.10s/it] 80%|████████ | 24134/30000 [14:43:52<2:58:33, 1.83s/it] 80%|████████ | 24135/30000 [14:43:53<2:39:20, 1.63s/it] 80%|████████ | 24136/30000 [14:43:55<2:27:00, 1.50s/it] 80%|████████ | 24137/30000 [14:43:56<2:18:06, 1.41s/it] 80%|████████ | 24138/30000 [14:43:57<2:11:51, 1.35s/it] 80%|████████ | 24139/30000 [14:44:04<4:54:14, 3.01s/it] 80%|████████ | 24140/30000 [14:44:05<4:00:52, 2.47s/it] 80%|████████ | 24140/30000 [14:44:05<4:00:52, 2.47s/it] 80%|████████ | 24141/30000 [14:44:06<3:24:03, 2.09s/it] 80%|████████ | 24142/30000 [14:44:07<2:57:26, 1.82s/it] 80%|████████ | 24143/30000 [14:44:09<2:38:49, 1.63s/it] 80%|████████ | 24144/30000 [14:44:10<2:26:33, 1.50s/it] 80%|████████ | 24145/30000 [14:44:11<2:17:24, 1.41s/it] 80%|████████ | 24146/30000 [14:44:12<2:12:11, 1.35s/it] 80%|████████ | 24147/30000 [14:44:19<4:56:09, 3.04s/it] 80%|████████ | 24148/30000 [14:44:20<4:01:55, 2.48s/it] 80%|████████ | 24149/30000 [14:44:22<3:24:20, 2.10s/it] 80%|████████ | 24150/30000 [14:44:23<2:57:45, 1.82s/it] 80%|████████ | 24150/30000 [14:44:23<2:57:45, 1.82s/it] 81%|████████ | 24151/30000 [14:44:24<2:38:27, 1.63s/it] 81%|████████ | 24152/30000 [14:44:25<2:26:50, 1.51s/it] 81%|████████ | 24153/30000 [14:44:26<2:20:56, 1.45s/it] 81%|████████ | 24154/30000 [14:44:28<2:13:38, 1.37s/it] 81%|████████ | 24155/30000 [14:44:34<4:43:23, 2.91s/it] 81%|████████ | 24156/30000 [14:44:35<3:52:52, 2.39s/it] 81%|████████ | 24157/30000 [14:44:36<3:17:36, 2.03s/it] 81%|████████ | 24158/30000 [14:44:38<2:52:57, 1.78s/it] 81%|████████ | 24159/30000 [14:44:39<2:35:42, 1.60s/it] 81%|████████ | 24160/30000 [14:44:40<2:23:43, 1.48s/it] 81%|████████ | 24160/30000 [14:44:40<2:23:43, 1.48s/it] 81%|████████ | 24161/30000 [14:44:41<2:15:41, 1.39s/it] 81%|████████ | 24162/30000 [14:44:42<2:10:18, 1.34s/it] 81%|████████ | 24163/30000 [14:44:48<4:26:40, 2.74s/it] 81%|████████ | 24164/30000 [14:44:50<3:40:39, 2.27s/it] 81%|████████ | 24165/30000 [14:44:51<3:09:03, 1.94s/it] 81%|████████ | 24166/30000 [14:44:52<2:48:29, 1.73s/it] 81%|████████ | 24167/30000 [14:44:53<2:32:45, 1.57s/it] 81%|████████ | 24168/30000 [14:44:55<2:23:26, 1.48s/it] 81%|████████ | 24169/30000 [14:44:56<2:15:51, 1.40s/it] 81%|████████ | 24170/30000 [14:44:57<2:10:31, 1.34s/it] 81%|████████ | 24170/30000 [14:44:57<2:10:31, 1.34s/it] 81%|████████ | 24171/30000 [14:45:03<4:39:23, 2.88s/it] 81%|████████ | 24172/30000 [14:45:05<3:49:52, 2.37s/it] 81%|████████ | 24173/30000 [14:45:06<3:16:04, 2.02s/it] 81%|████████ | 24174/30000 [14:45:07<2:51:17, 1.76s/it] 81%|████████ | 24175/30000 [14:45:08<2:33:39, 1.58s/it] 81%|████████ | 24176/30000 [14:45:09<2:22:24, 1.47s/it] 81%|████████ | 24177/30000 [14:45:10<2:14:17, 1.38s/it] 81%|████████ | 24178/30000 [14:45:12<2:08:52, 1.33s/it] 81%|████████ | 24179/30000 [14:45:19<4:48:28, 2.97s/it] 81%|████████ | 24180/30000 [14:45:20<3:56:13, 2.44s/it] 81%|████████ | 24180/30000 [14:45:20<3:56:13, 2.44s/it] 81%|████████ | 24181/30000 [14:45:21<3:21:02, 2.07s/it] 81%|████████ | 24182/30000 [14:45:22<2:55:20, 1.81s/it] 81%|████████ | 24183/30000 [14:45:23<2:36:37, 1.62s/it] 81%|████████ | 24184/30000 [14:45:24<2:23:26, 1.48s/it] 81%|████████ | 24185/30000 [14:45:26<2:14:47, 1.39s/it] 81%|████████ | 24186/30000 [14:45:27<2:08:31, 1.33s/it] 81%|████████ | 24187/30000 [14:45:34<4:59:54, 3.10s/it] 81%|████████ | 24188/30000 [14:45:35<4:03:54, 2.52s/it] 81%|████████ | 24189/30000 [14:45:36<3:25:18, 2.12s/it] 81%|████████ | 24190/30000 [14:45:38<2:58:00, 1.84s/it] 81%|████████ | 24190/30000 [14:45:38<2:58:00, 1.84s/it] 81%|████████ | 24191/30000 [14:45:39<2:38:44, 1.64s/it] 81%|████████ | 24192/30000 [14:45:40<2:25:07, 1.50s/it] 81%|████████ | 24193/30000 [14:45:41<2:16:15, 1.41s/it] 81%|████████ | 24194/30000 [14:45:42<2:10:21, 1.35s/it] 81%|████████ | 24195/30000 [14:45:50<5:13:18, 3.24s/it] 81%|████████ | 24196/30000 [14:45:51<4:13:38, 2.62s/it] 81%|████████ | 24197/30000 [14:45:52<3:32:27, 2.20s/it] 81%|████████ | 24198/30000 [14:45:54<3:02:45, 1.89s/it] 81%|████████ | 24199/30000 [14:45:55<2:41:47, 1.67s/it] 81%|████████ | 24200/30000 [14:45:56<2:27:47, 1.53s/it] 81%|████████ | 24200/30000 [14:45:56<2:27:47, 1.53s/it] 81%|████████ | 24201/30000 [14:45:57<2:18:32, 1.43s/it] 81%|████████ | 24202/30000 [14:45:58<2:11:02, 1.36s/it] 81%|████████ | 24203/30000 [14:46:06<5:04:17, 3.15s/it] 81%|████████ | 24204/30000 [14:46:07<4:07:00, 2.56s/it] 81%|████████ | 24205/30000 [14:46:08<3:27:34, 2.15s/it] 81%|████████ | 24206/30000 [14:46:09<2:59:29, 1.86s/it] 81%|████████ | 24207/30000 [14:46:10<2:39:22, 1.65s/it] 81%|████████ | 24208/30000 [14:46:12<2:26:46, 1.52s/it] 81%|████████ | 24209/30000 [14:46:13<2:17:35, 1.43s/it] 81%|████████ | 24210/30000 [14:46:14<2:11:43, 1.37s/it] 81%|████████ | 24210/30000 [14:46:14<2:11:43, 1.37s/it] 81%|████████ | 24211/30000 [14:46:21<5:04:59, 3.16s/it] 81%|████████ | 24212/30000 [14:46:23<4:07:41, 2.57s/it] 81%|████████ | 24213/30000 [14:46:24<3:28:03, 2.16s/it] 81%|████████ | 24214/30000 [14:46:25<2:59:31, 1.86s/it] 81%|████████ | 24215/30000 [14:46:26<2:40:16, 1.66s/it] 81%|████████ | 24216/30000 [14:46:27<2:27:06, 1.53s/it] 81%|████████ | 24217/30000 [14:46:28<2:18:16, 1.43s/it] 81%|████████ | 24218/30000 [14:46:30<2:11:00, 1.36s/it] 81%|████████ | 24219/30000 [14:46:37<5:07:01, 3.19s/it] 81%|████████ | 24220/30000 [14:46:38<4:09:26, 2.59s/it] 81%|████████ | 24220/30000 [14:46:38<4:09:26, 2.59s/it] 81%|████████ | 24221/30000 [14:46:40<3:29:31, 2.18s/it] 81%|████████ | 24222/30000 [14:46:41<3:00:58, 1.88s/it] 81%|████████ | 24223/30000 [14:46:42<2:40:49, 1.67s/it] 81%|████████ | 24224/30000 [14:46:43<2:26:54, 1.53s/it] 81%|████████ | 24225/30000 [14:46:44<2:17:22, 1.43s/it] 81%|████████ | 24226/30000 [14:46:45<2:10:18, 1.35s/it] 81%|████████ | 24227/30000 [14:46:52<4:46:55, 2.98s/it] 81%|████████ | 24228/30000 [14:46:53<3:55:32, 2.45s/it] 81%|████████ | 24229/30000 [14:46:55<3:19:17, 2.07s/it] 81%|████████ | 24230/30000 [14:46:56<2:53:18, 1.80s/it] 81%|████████ | 24230/30000 [14:46:56<2:53:18, 1.80s/it] 81%|████████ | 24231/30000 [14:46:57<2:35:31, 1.62s/it] 81%|████████ | 24232/30000 [14:46:58<2:23:18, 1.49s/it] 81%|████████ | 24233/30000 [14:46:59<2:14:30, 1.40s/it] 81%|████████ | 24234/30000 [14:47:01<2:08:33, 1.34s/it] 81%|████████ | 24235/30000 [14:47:08<4:56:47, 3.09s/it] 81%|████████ | 24236/30000 [14:47:09<4:01:43, 2.52s/it] 81%|████████ | 24237/30000 [14:47:10<3:24:00, 2.12s/it] 81%|████████ | 24238/30000 [14:47:11<2:57:36, 1.85s/it] 81%|████████ | 24239/30000 [14:47:13<2:37:34, 1.64s/it] 81%|████████ | 24240/30000 [14:47:14<2:24:31, 1.51s/it] 81%|████████ | 24240/30000 [14:47:14<2:24:31, 1.51s/it] 81%|████████ | 24241/30000 [14:47:15<2:15:55, 1.42s/it] 81%|████████ | 24242/30000 [14:47:16<2:09:57, 1.35s/it] 81%|████████ | 24243/30000 [14:47:23<4:44:41, 2.97s/it] 81%|████████ | 24244/30000 [14:47:24<3:53:16, 2.43s/it] 81%|████████ | 24245/30000 [14:47:25<3:18:17, 2.07s/it] 81%|████████ | 24246/30000 [14:47:26<2:53:44, 1.81s/it] 81%|████████ | 24247/30000 [14:47:28<2:36:07, 1.63s/it] 81%|████████ | 24248/30000 [14:47:29<2:23:35, 1.50s/it] 81%|████████ | 24249/30000 [14:47:30<2:17:21, 1.43s/it] 81%|████████ | 24250/30000 [14:47:31<2:10:39, 1.36s/it] 81%|████████ | 24250/30000 [14:47:31<2:10:39, 1.36s/it] 81%|████████ | 24251/30000 [14:47:39<5:15:23, 3.29s/it] 81%|████████ | 24252/30000 [14:47:40<4:14:17, 2.65s/it] 81%|████████ | 24253/30000 [14:47:42<3:33:11, 2.23s/it] 81%|████████ | 24254/30000 [14:47:43<3:02:53, 1.91s/it] 81%|████████ | 24255/30000 [14:47:44<2:41:57, 1.69s/it] 81%|████████ | 24256/30000 [14:47:45<2:27:28, 1.54s/it] 81%|████████ | 24257/30000 [14:47:46<2:17:42, 1.44s/it] 81%|████████ | 24258/30000 [14:47:47<2:10:53, 1.37s/it] 81%|████████ | 24259/30000 [14:47:54<4:29:21, 2.82s/it] 81%|████████ | 24260/30000 [14:47:55<3:42:12, 2.32s/it] 81%|████████ | 24260/30000 [14:47:55<3:42:12, 2.32s/it] 81%|████████ | 24261/30000 [14:47:56<3:10:55, 2.00s/it] 81%|████████ | 24262/30000 [14:47:57<2:46:54, 1.75s/it] 81%|████████ | 24263/30000 [14:47:58<2:30:42, 1.58s/it] 81%|████████ | 24264/30000 [14:48:00<2:19:19, 1.46s/it] 81%|████████ | 24265/30000 [14:48:01<2:13:30, 1.40s/it] 81%|████████ | 24266/30000 [14:48:02<2:10:16, 1.36s/it] 81%|████████ | 24267/30000 [14:48:09<4:53:36, 3.07s/it] 81%|████████ | 24268/30000 [14:48:10<3:59:31, 2.51s/it] 81%|████████ | 24269/30000 [14:48:12<3:23:03, 2.13s/it] 81%|████████ | 24270/30000 [14:48:13<2:55:25, 1.84s/it] 81%|████████ | 24270/30000 [14:48:13<2:55:25, 1.84s/it] 81%|████████ | 24271/30000 [14:48:14<2:37:02, 1.64s/it] 81%|████████ | 24272/30000 [14:48:15<2:24:08, 1.51s/it] 81%|████████ | 24273/30000 [14:48:16<2:15:12, 1.42s/it] 81%|████████ | 24274/30000 [14:48:18<2:09:37, 1.36s/it] 81%|████████ | 24275/30000 [14:48:25<4:58:32, 3.13s/it] 81%|████████ | 24276/30000 [14:48:26<4:02:45, 2.54s/it] 81%|████████ | 24277/30000 [14:48:27<3:24:30, 2.14s/it] 81%|████████ | 24278/30000 [14:48:28<2:57:21, 1.86s/it] 81%|████████ | 24279/30000 [14:48:30<2:38:38, 1.66s/it] 81%|████████ | 24280/30000 [14:48:31<2:25:26, 1.53s/it] 81%|████████ | 24280/30000 [14:48:31<2:25:26, 1.53s/it] 81%|████████ | 24281/30000 [14:48:32<2:15:56, 1.43s/it] 81%|████████ | 24282/30000 [14:48:33<2:09:41, 1.36s/it] 81%|████████ | 24283/30000 [14:48:40<4:36:29, 2.90s/it] 81%|████████ | 24284/30000 [14:48:41<3:47:19, 2.39s/it] 81%|████████ | 24285/30000 [14:48:42<3:13:12, 2.03s/it] 81%|████████ | 24286/30000 [14:48:43<2:49:22, 1.78s/it] 81%|████████ | 24287/30000 [14:48:45<2:33:06, 1.61s/it] 81%|████████ | 24288/30000 [14:48:46<2:22:30, 1.50s/it] 81%|████████ | 24289/30000 [14:48:47<2:14:59, 1.42s/it] 81%|████████ | 24290/30000 [14:48:48<2:09:00, 1.36s/it] 81%|████████ | 24290/30000 [14:48:48<2:09:00, 1.36s/it] 81%|████████ | 24291/30000 [14:48:55<4:56:00, 3.11s/it] 81%|████████ | 24292/30000 [14:48:57<4:01:44, 2.54s/it] 81%|████████ | 24293/30000 [14:48:58<3:23:09, 2.14s/it] 81%|████████ | 24294/30000 [14:48:59<2:56:16, 1.85s/it] 81%|████████ | 24295/30000 [14:49:00<2:37:22, 1.66s/it] 81%|████████ | 24296/30000 [14:49:01<2:24:07, 1.52s/it] 81%|████████ | 24297/30000 [14:49:03<2:14:48, 1.42s/it] 81%|████████ | 24298/30000 [14:49:04<2:09:21, 1.36s/it] 81%|████████ | 24299/30000 [14:49:11<5:06:20, 3.22s/it] 81%|████████ | 24300/30000 [14:49:13<4:08:23, 2.61s/it] 81%|████████ | 24300/30000 [14:49:13<4:08:23, 2.61s/it] 81%|████████ | 24301/30000 [14:49:14<3:28:21, 2.19s/it] 81%|████████ | 24302/30000 [14:49:15<3:00:30, 1.90s/it] 81%|████████ | 24303/30000 [14:49:16<2:40:19, 1.69s/it] 81%|████████ | 24304/30000 [14:49:17<2:26:18, 1.54s/it] 81%|████████ | 24305/30000 [14:49:19<2:16:23, 1.44s/it] 81%|████████ | 24306/30000 [14:49:20<2:10:41, 1.38s/it] 81%|████████ | 24307/30000 [14:49:27<5:08:26, 3.25s/it] 81%|████████ | 24308/30000 [14:49:29<4:09:05, 2.63s/it] 81%|████████ | 24309/30000 [14:49:30<3:28:44, 2.20s/it] 81%|████████ | 24310/30000 [14:49:31<2:59:47, 1.90s/it] 81%|████████ | 24310/30000 [14:49:31<2:59:47, 1.90s/it] 81%|████████ | 24311/30000 [14:49:32<2:39:07, 1.68s/it] 81%|████████ | 24312/30000 [14:49:33<2:25:20, 1.53s/it] 81%|████████ | 24313/30000 [14:49:35<2:15:44, 1.43s/it] 81%|████████ | 24314/30000 [14:49:36<2:11:14, 1.38s/it] 81%|████████ | 24315/30000 [14:49:42<4:33:53, 2.89s/it] 81%|████████ | 24316/30000 [14:49:43<3:45:23, 2.38s/it] 81%|████████ | 24317/30000 [14:49:45<3:11:53, 2.03s/it] 81%|████████ | 24318/30000 [14:49:46<2:48:57, 1.78s/it] 81%|████████ | 24319/30000 [14:49:47<2:31:49, 1.60s/it] 81%|████████ | 24320/30000 [14:49:48<2:20:00, 1.48s/it] 81%|████████ | 24320/30000 [14:49:48<2:20:00, 1.48s/it] 81%|████████ | 24321/30000 [14:49:49<2:12:20, 1.40s/it] 81%|████████ | 24322/30000 [14:49:51<2:06:33, 1.34s/it] 81%|████████ | 24323/30000 [14:49:57<4:41:36, 2.98s/it] 81%|████████ | 24324/30000 [14:49:59<3:50:38, 2.44s/it] 81%|████████ | 24325/30000 [14:50:00<3:15:23, 2.07s/it] 81%|████████ | 24326/30000 [14:50:01<2:50:02, 1.80s/it] 81%|████████ | 24327/30000 [14:50:02<2:31:57, 1.61s/it] 81%|████████ | 24328/30000 [14:50:03<2:20:21, 1.48s/it] 81%|████████ | 24329/30000 [14:50:05<2:12:28, 1.40s/it] 81%|████████ | 24330/30000 [14:50:06<2:07:11, 1.35s/it] 81%|████████ | 24330/30000 [14:50:06<2:07:11, 1.35s/it] 81%|████████ | 24331/30000 [14:50:14<5:08:40, 3.27s/it] 81%|████████ | 24332/30000 [14:50:15<4:09:25, 2.64s/it] 81%|████████ | 24333/30000 [14:50:16<3:28:41, 2.21s/it] 81%|████████ | 24334/30000 [14:50:17<2:59:34, 1.90s/it] 81%|████████ | 24335/30000 [14:50:18<2:39:13, 1.69s/it] 81%|████████ | 24336/30000 [14:50:19<2:25:43, 1.54s/it] 81%|████████ | 24337/30000 [14:50:21<2:15:55, 1.44s/it] 81%|████████ | 24338/30000 [14:50:22<2:09:28, 1.37s/it] 81%|████████ | 24339/30000 [14:50:29<4:43:56, 3.01s/it] 81%|████████ | 24340/30000 [14:50:30<3:52:07, 2.46s/it] 81%|████████ | 24340/30000 [14:50:30<3:52:07, 2.46s/it] 81%|████████ | 24341/30000 [14:50:31<3:16:55, 2.09s/it] 81%|████████ | 24342/30000 [14:50:32<2:51:15, 1.82s/it] 81%|████████ | 24343/30000 [14:50:33<2:33:08, 1.62s/it] 81%|████████ | 24344/30000 [14:50:35<2:21:10, 1.50s/it] 81%|████████ | 24345/30000 [14:50:36<2:12:31, 1.41s/it] 81%|████████ | 24346/30000 [14:50:37<2:11:47, 1.40s/it] 81%|████████ | 24347/30000 [14:50:45<5:05:37, 3.24s/it] 81%|████████ | 24348/30000 [14:50:46<4:07:06, 2.62s/it] 81%|████████ | 24349/30000 [14:50:47<3:26:26, 2.19s/it] 81%|████████ | 24350/30000 [14:50:48<2:58:34, 1.90s/it] 81%|████████ | 24350/30000 [14:50:48<2:58:34, 1.90s/it] 81%|████████ | 24351/30000 [14:50:50<2:38:34, 1.68s/it] 81%|████████ | 24352/30000 [14:50:51<2:24:54, 1.54s/it] 81%|████████ | 24353/30000 [14:50:52<2:15:32, 1.44s/it] 81%|████████ | 24354/30000 [14:50:53<2:08:55, 1.37s/it] 81%|████████ | 24355/30000 [14:51:00<4:42:35, 3.00s/it] 81%|████████ | 24356/30000 [14:51:01<3:52:27, 2.47s/it] 81%|████████ | 24357/30000 [14:51:02<3:16:36, 2.09s/it] 81%|████████ | 24358/30000 [14:51:04<2:50:31, 1.81s/it] 81%|████████ | 24359/30000 [14:51:05<2:33:08, 1.63s/it] 81%|████████ | 24360/30000 [14:51:06<2:21:49, 1.51s/it] 81%|████████ | 24360/30000 [14:51:06<2:21:49, 1.51s/it] 81%|████████ | 24361/30000 [14:51:07<2:13:49, 1.42s/it] 81%|████████ | 24362/30000 [14:51:08<2:07:05, 1.35s/it] 81%|████████ | 24363/30000 [14:51:16<5:00:28, 3.20s/it] 81%|████████ | 24364/30000 [14:51:17<4:03:24, 2.59s/it] 81%|████████ | 24365/30000 [14:51:18<3:24:01, 2.17s/it] 81%|████████ | 24366/30000 [14:51:20<2:56:23, 1.88s/it] 81%|████████ | 24367/30000 [14:51:21<2:37:00, 1.67s/it] 81%|████████ | 24368/30000 [14:51:22<2:24:02, 1.53s/it] 81%|████████ | 24369/30000 [14:51:23<2:14:38, 1.43s/it] 81%|████████ | 24370/30000 [14:51:24<2:08:30, 1.37s/it] 81%|████████ | 24370/30000 [14:51:24<2:08:30, 1.37s/it] 81%|████████ | 24371/30000 [14:51:31<4:34:47, 2.93s/it] 81%|████████ | 24372/30000 [14:51:32<3:46:02, 2.41s/it] 81%|████████ | 24373/30000 [14:51:33<3:12:21, 2.05s/it] 81%|████████ | 24374/30000 [14:51:35<2:48:30, 1.80s/it] 81%|████████▏ | 24375/30000 [14:51:36<2:31:39, 1.62s/it] 81%|████████▏ | 24376/30000 [14:51:37<2:19:42, 1.49s/it] 81%|████████▏ | 24377/30000 [14:51:38<2:11:12, 1.40s/it] 81%|████████▏ | 24378/30000 [14:51:39<2:05:02, 1.33s/it] 81%|████████▏ | 24379/30000 [14:51:46<4:36:29, 2.95s/it] 81%|████████▏ | 24380/30000 [14:51:47<3:47:53, 2.43s/it] 81%|████████▏ | 24380/30000 [14:51:47<3:47:53, 2.43s/it] 81%|████████▏ | 24381/30000 [14:51:48<3:13:35, 2.07s/it] 81%|████████▏ | 24382/30000 [14:51:50<2:48:36, 1.80s/it] 81%|████████▏ | 24383/30000 [14:51:51<2:31:18, 1.62s/it] 81%|████████▏ | 24384/30000 [14:51:52<2:19:52, 1.49s/it] 81%|████████▏ | 24385/30000 [14:51:53<2:11:33, 1.41s/it] 81%|████████▏ | 24386/30000 [14:51:54<2:06:16, 1.35s/it] 81%|████████▏ | 24387/30000 [14:52:00<4:09:59, 2.67s/it] 81%|████████▏ | 24388/30000 [14:52:01<3:27:56, 2.22s/it] 81%|████████▏ | 24389/30000 [14:52:03<2:58:19, 1.91s/it] 81%|████████▏ | 24390/30000 [14:52:04<2:38:13, 1.69s/it] 81%|████████▏ | 24390/30000 [14:52:04<2:38:13, 1.69s/it] 81%|████████▏ | 24391/30000 [14:52:05<2:23:37, 1.54s/it] 81%|████████▏ | 24392/30000 [14:52:06<2:13:50, 1.43s/it] 81%|████████▏ | 24393/30000 [14:52:07<2:07:31, 1.36s/it] 81%|████████▏ | 24394/30000 [14:52:09<2:03:38, 1.32s/it] 81%|████████▏ | 24395/30000 [14:52:15<4:22:52, 2.81s/it] 81%|████████▏ | 24396/30000 [14:52:16<3:37:57, 2.33s/it] 81%|████████▏ | 24397/30000 [14:52:17<3:06:08, 1.99s/it] 81%|████████▏ | 24398/30000 [14:52:18<2:43:42, 1.75s/it] 81%|████████▏ | 24399/30000 [14:52:20<2:28:24, 1.59s/it] 81%|████████▏ | 24400/30000 [14:52:21<2:17:43, 1.48s/it] 81%|████████▏ | 24400/30000 [14:52:21<2:17:43, 1.48s/it] 81%|████████▏ | 24401/30000 [14:52:22<2:10:22, 1.40s/it] 81%|████████▏ | 24402/30000 [14:52:23<2:05:35, 1.35s/it] 81%|████████▏ | 24403/30000 [14:52:30<4:26:46, 2.86s/it] 81%|████████▏ | 24404/30000 [14:52:31<3:39:18, 2.35s/it] 81%|████████▏ | 24405/30000 [14:52:32<3:07:01, 2.01s/it] 81%|████████▏ | 24406/30000 [14:52:33<2:44:02, 1.76s/it] 81%|████████▏ | 24407/30000 [14:52:34<2:27:38, 1.58s/it] 81%|████████▏ | 24408/30000 [14:52:36<2:16:12, 1.46s/it] 81%|████████▏ | 24409/30000 [14:52:37<2:08:54, 1.38s/it] 81%|████████▏ | 24410/30000 [14:52:38<2:03:22, 1.32s/it] 81%|████████▏ | 24410/30000 [14:52:38<2:03:22, 1.32s/it] 81%|████████▏ | 24411/30000 [14:52:44<4:25:09, 2.85s/it] 81%|████████▏ | 24412/30000 [14:52:46<3:39:07, 2.35s/it] 81%|████████▏ | 24413/30000 [14:52:47<3:06:58, 2.01s/it] 81%|████████▏ | 24414/30000 [14:52:48<2:43:51, 1.76s/it] 81%|████████▏ | 24415/30000 [14:52:49<2:27:12, 1.58s/it] 81%|████████▏ | 24416/30000 [14:52:50<2:16:18, 1.46s/it] 81%|████████▏ | 24417/30000 [14:52:51<2:08:38, 1.38s/it] 81%|████████▏ | 24418/30000 [14:52:53<2:03:37, 1.33s/it] 81%|████████▏ | 24419/30000 [14:52:59<4:33:37, 2.94s/it] 81%|████████▏ | 24420/30000 [14:53:01<3:43:50, 2.41s/it] 81%|████████▏ | 24420/30000 [14:53:01<3:43:50, 2.41s/it] 81%|████████▏ | 24421/30000 [14:53:02<3:10:05, 2.04s/it] 81%|████████▏ | 24422/30000 [14:53:03<2:45:41, 1.78s/it] 81%|████████▏ | 24423/30000 [14:53:04<2:28:36, 1.60s/it] 81%|████████▏ | 24424/30000 [14:53:05<2:17:17, 1.48s/it] 81%|████████▏ | 24425/30000 [14:53:06<2:09:23, 1.39s/it] 81%|████████▏ | 24426/30000 [14:53:08<2:03:09, 1.33s/it] 81%|████████▏ | 24427/30000 [14:53:15<4:48:31, 3.11s/it] 81%|████████▏ | 24428/30000 [14:53:16<3:54:31, 2.53s/it] 81%|████████▏ | 24429/30000 [14:53:17<3:17:37, 2.13s/it] 81%|████████▏ | 24430/30000 [14:53:18<2:51:11, 1.84s/it] 81%|████████▏ | 24430/30000 [14:53:18<2:51:11, 1.84s/it] 81%|████████▏ | 24431/30000 [14:53:20<2:31:45, 1.64s/it] 81%|████████▏ | 24432/30000 [14:53:21<2:19:11, 1.50s/it] 81%|████████▏ | 24433/30000 [14:53:22<2:10:06, 1.40s/it] 81%|████████▏ | 24434/30000 [14:53:23<2:04:05, 1.34s/it] 81%|████████▏ | 24435/30000 [14:53:30<4:41:49, 3.04s/it] 81%|████████▏ | 24436/30000 [14:53:31<3:49:58, 2.48s/it] 81%|████████▏ | 24437/30000 [14:53:33<3:14:03, 2.09s/it] 81%|████████▏ | 24438/30000 [14:53:34<2:48:39, 1.82s/it] 81%|████████▏ | 24439/30000 [14:53:35<2:31:16, 1.63s/it] 81%|████████▏ | 24440/30000 [14:53:36<2:18:45, 1.50s/it] {'loss': 0.0137, 'grad_norm': 0.08312865346670151, 'learning_rate': 3.1233941600490123e-06, 'flow_matching_loss': 0.013655327167361975, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 6.99} +{'loss': 0.0133, 'grad_norm': 0.08324886858463287, 'learning_rate': 3.113301744039875e-06, 'flow_matching_loss': 0.013295511901378631, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.0} +{'loss': 0.0139, 'grad_norm': 0.07051778584718704, 'learning_rate': 3.103223771489959e-06, 'flow_matching_loss': 0.013923694007098674, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.0} +{'loss': 0.0122, 'grad_norm': 0.0754796490073204, 'learning_rate': 3.09316025464495e-06, 'flow_matching_loss': 0.012239718856289983, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.0} +{'loss': 0.0138, 'grad_norm': 0.08114205300807953, 'learning_rate': 3.0831112057329775e-06, 'flow_matching_loss': 0.013776348996907472, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.01} +{'loss': 0.0134, 'grad_norm': 0.06838233768939972, 'learning_rate': 3.073076636964579e-06, 'flow_matching_loss': 0.013368572387844325, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.01} +{'loss': 0.0121, 'grad_norm': 0.05184592679142952, 'learning_rate': 3.063056560532698e-06, 'flow_matching_loss': 0.012081048637628555, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.01} +{'loss': 0.0122, 'grad_norm': 0.08445174992084503, 'learning_rate': 3.0530509886126807e-06, 'flow_matching_loss': 0.012163841351866721, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.01} +{'loss': 0.0121, 'grad_norm': 0.0796457901597023, 'learning_rate': 3.043059933362236e-06, 'flow_matching_loss': 0.012078788690268993, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.02} +{'loss': 0.0138, 'grad_norm': 0.08410722017288208, 'learning_rate': 3.0330834069214337e-06, 'flow_matching_loss': 0.013791468925774097, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.02} +{'loss': 0.0133, 'grad_norm': 0.09685815870761871, 'learning_rate': 3.023121421412703e-06, 'flow_matching_loss': 0.013339803740382194, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.02} +{'loss': 0.0138, 'grad_norm': 0.07762318849563599, 'learning_rate': 3.0131739889407896e-06, 'flow_matching_loss': 0.013773251511156558, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.03} +{'loss': 0.0127, 'grad_norm': 0.07465093582868576, 'learning_rate': 3.0032411215927645e-06, 'flow_matching_loss': 0.0127343506552279, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.03} +{'loss': 0.0132, 'grad_norm': 0.07047083973884583, 'learning_rate': 2.9933228314379975e-06, 'flow_matching_loss': 0.013217219337821006, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.03} +{'loss': 0.0122, 'grad_norm': 0.06911014765501022, 'learning_rate': 2.9834191305281405e-06, 'flow_matching_loss': 0.012204410322010518, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.04} +{'loss': 0.013, 'grad_norm': 0.07424776256084442, 'learning_rate': 2.973530030897135e-06, 'flow_matching_loss': 0.01300343768671155, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.04} +{'loss': 0.012, 'grad_norm': 0.07900994271039963, 'learning_rate': 2.9636555445611646e-06, 'flow_matching_loss': 0.012008262239396572, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.04} +{'loss': 0.0132, 'grad_norm': 0.08802741020917892, 'learning_rate': 2.953795683518659e-06, 'flow_matching_loss': 0.01315589053556323, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.04} +{'loss': 0.013, 'grad_norm': 0.09260044246912003, 'learning_rate': 2.9439504597502815e-06, 'flow_matching_loss': 0.012994121666997671, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.05} +{'loss': 0.0138, 'grad_norm': 0.0671023279428482, 'learning_rate': 2.934119885218902e-06, 'flow_matching_loss': 0.013760613091289998, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.05} +{'loss': 0.0117, 'grad_norm': 0.07491301745176315, 'learning_rate': 2.9243039718696045e-06, 'flow_matching_loss': 0.011709636403247714, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.05} +{'loss': 0.0118, 'grad_norm': 0.054729197174310684, 'learning_rate': 2.914502731629641e-06, 'flow_matching_loss': 0.011811607237905263, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.06} +{'loss': 0.0145, 'grad_norm': 0.07196380943059921, 'learning_rate': 2.904716176408451e-06, 'flow_matching_loss': 0.014450524374842644, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.06} +{'loss': 0.0136, 'grad_norm': 0.10089542716741562, 'learning_rate': 2.8949443180976172e-06, 'flow_matching_loss': 0.013633861392736434, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.06} +{'loss': 0.0133, 'grad_norm': 0.0828099399805069, 'learning_rate': 2.885187168570864e-06, 'flow_matching_loss': 0.013279086165130138, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.06} +{'loss': 0.0144, 'grad_norm': 0.07810799777507782, 'learning_rate': 2.875444739684058e-06, 'flow_matching_loss': 0.0144026062451303, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.07} +{'loss': 0.0123, 'grad_norm': 0.07579539716243744, 'learning_rate': 2.865717043275165e-06, 'flow_matching_loss': 0.012314677145332098, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.07} +{'loss': 0.0143, 'grad_norm': 0.06534378230571747, 'learning_rate': 2.8560040911642528e-06, 'flow_matching_loss': 0.014299785066395999, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.07} +{'loss': 0.014, 'grad_norm': 0.06601939350366592, 'learning_rate': 2.846305895153472e-06, 'flow_matching_loss': 0.013996868208050729, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.08} +{'loss': 0.0143, 'grad_norm': 0.08378820866346359, 'learning_rate': 2.8366224670270487e-06, 'flow_matching_loss': 0.014250252489000559, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.08} +{'loss': 0.0114, 'grad_norm': 0.05324731767177582, 'learning_rate': 2.8269538185512535e-06, 'flow_matching_loss': 0.011393609549850225, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.08} +{'loss': 0.0126, 'grad_norm': 0.07163026183843613, 'learning_rate': 2.817299961474417e-06, 'flow_matching_loss': 0.012587910983711481, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.08} +{'loss': 0.0128, 'grad_norm': 0.0645531490445137, 'learning_rate': 2.8076609075268816e-06, 'flow_matching_loss': 0.012789037358015776, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.09} +{'loss': 0.0139, 'grad_norm': 0.07644204050302505, 'learning_rate': 2.798036668421006e-06, 'flow_matching_loss': 0.013863526284694672, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.09} +{'loss': 0.0128, 'grad_norm': 0.09393570572137833, 'learning_rate': 2.7884272558511433e-06, 'flow_matching_loss': 0.012814339715987444, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.09} +{'loss': 0.0134, 'grad_norm': 0.05040053278207779, 'learning_rate': 2.7788326814936456e-06, 'flow_matching_loss': 0.013429556041955948, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.1} +{'loss': 0.0145, 'grad_norm': 0.06803322583436966, 'learning_rate': 2.7692529570068176e-06, 'flow_matching_loss': 0.014519502688199282, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.1} +{'loss': 0.015, 'grad_norm': 0.08708856999874115, 'learning_rate': 2.759688094030933e-06, 'flow_matching_loss': 0.01496996358036995, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.1} +{'loss': 0.0132, 'grad_norm': 0.07699897140264511, 'learning_rate': 2.750138104188201e-06, 'flow_matching_loss': 0.013233906030654908, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.11} +{'loss': 0.0128, 'grad_norm': 0.0747479796409607, 'learning_rate': 2.740602999082759e-06, 'flow_matching_loss': 0.012809615302830934, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.11} +{'loss': 0.0124, 'grad_norm': 0.05632628872990608, 'learning_rate': 2.731082790300654e-06, 'flow_matching_loss': 0.01238290909677744, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.11} 81%|████████▏ | 24440/30000 [14:53:36<2:18:45, 1.50s/it] 81%|████████▏ | 24441/30000 [14:53:37<2:10:05, 1.40s/it] 81%|████████▏ | 24442/30000 [14:53:38<2:03:44, 1.34s/it] 81%|████████▏ | 24443/30000 [14:53:46<4:57:37, 3.21s/it] 81%|████████▏ | 24444/30000 [14:53:47<4:00:30, 2.60s/it] 81%|████████▏ | 24445/30000 [14:53:48<3:22:43, 2.19s/it] 81%|████████▏ | 24446/30000 [14:53:50<2:54:28, 1.88s/it] 81%|████████▏ | 24447/30000 [14:53:51<2:34:38, 1.67s/it] 81%|████████▏ | 24448/30000 [14:53:52<2:20:57, 1.52s/it] 81%|████████▏ | 24449/30000 [14:53:53<2:12:04, 1.43s/it] 82%|████████▏ | 24450/30000 [14:53:54<2:04:33, 1.35s/it] 82%|████████▏ | 24450/30000 [14:53:54<2:04:33, 1.35s/it] 82%|████████▏ | 24451/30000 [14:54:02<4:50:59, 3.15s/it] 82%|████████▏ | 24452/30000 [14:54:03<3:56:20, 2.56s/it] 82%|████████▏ | 24453/30000 [14:54:04<3:18:20, 2.15s/it] 82%|████████▏ | 24454/30000 [14:54:05<2:51:19, 1.85s/it] 82%|████████▏ | 24455/30000 [14:54:06<2:32:02, 1.65s/it] 82%|████████▏ | 24456/30000 [14:54:08<2:18:58, 1.50s/it] 82%|████████▏ | 24457/30000 [14:54:09<2:10:24, 1.41s/it] 82%|████████▏ | 24458/30000 [14:54:10<2:04:37, 1.35s/it] 82%|████████▏ | 24459/30000 [14:54:17<4:54:23, 3.19s/it] 82%|████████▏ | 24460/30000 [14:54:19<3:58:56, 2.59s/it] 82%|████████▏ | 24460/30000 [14:54:19<3:58:56, 2.59s/it] 82%|████████▏ | 24461/30000 [14:54:20<3:21:06, 2.18s/it] 82%|████████▏ | 24462/30000 [14:54:21<2:54:09, 1.89s/it] 82%|████████▏ | 24463/30000 [14:54:22<2:34:16, 1.67s/it] 82%|████████▏ | 24464/30000 [14:54:23<2:21:25, 1.53s/it] 82%|████████▏ | 24465/30000 [14:54:25<2:11:47, 1.43s/it] 82%|████████▏ | 24466/30000 [14:54:26<2:05:16, 1.36s/it] 82%|████████▏ | 24467/30000 [14:54:33<4:55:05, 3.20s/it] 82%|████████▏ | 24468/30000 [14:54:35<3:59:36, 2.60s/it] 82%|████████▏ | 24469/30000 [14:54:36<3:20:23, 2.17s/it] 82%|████████▏ | 24470/30000 [14:54:37<2:52:51, 1.88s/it] 82%|████████▏ | 24470/30000 [14:54:37<2:52:51, 1.88s/it] 82%|████████▏ | 24471/30000 [14:54:38<2:33:57, 1.67s/it] 82%|████████▏ | 24472/30000 [14:54:39<2:20:44, 1.53s/it] 82%|████████▏ | 24473/30000 [14:54:40<2:11:22, 1.43s/it] 82%|████████▏ | 24474/30000 [14:54:42<2:05:10, 1.36s/it] 82%|████████▏ | 24475/30000 [14:54:50<5:06:27, 3.33s/it] 82%|████████▏ | 24476/30000 [14:54:51<4:07:11, 2.68s/it] 82%|████████▏ | 24477/30000 [14:54:52<3:25:37, 2.23s/it] 82%|████████▏ | 24478/30000 [14:54:53<2:56:04, 1.91s/it] 82%|████████▏ | 24479/30000 [14:54:54<2:35:27, 1.69s/it] 82%|████████▏ | 24480/30000 [14:54:55<2:21:17, 1.54s/it] 82%|████████▏ | 24480/30000 [14:54:55<2:21:17, 1.54s/it] 82%|████████▏ | 24481/30000 [14:54:57<2:11:35, 1.43s/it] 82%|████████▏ | 24482/30000 [14:54:58<2:05:04, 1.36s/it] 82%|████████▏ | 24483/30000 [14:55:04<4:23:26, 2.87s/it] 82%|████████▏ | 24484/30000 [14:55:05<3:36:50, 2.36s/it] 82%|████████▏ | 24485/30000 [14:55:07<3:04:56, 2.01s/it] 82%|████████▏ | 24486/30000 [14:55:08<2:42:01, 1.76s/it] 82%|████████▏ | 24487/30000 [14:55:09<2:26:15, 1.59s/it] 82%|████████▏ | 24488/30000 [14:55:10<2:14:48, 1.47s/it] 82%|████████▏ | 24489/30000 [14:55:11<2:07:15, 1.39s/it] 82%|████████▏ | 24490/30000 [14:55:13<2:03:15, 1.34s/it] 82%|████████▏ | 24490/30000 [14:55:13<2:03:15, 1.34s/it] 82%|████████▏ | 24491/30000 [14:55:20<5:02:53, 3.30s/it] 82%|████████▏ | 24492/30000 [14:55:22<4:04:04, 2.66s/it] 82%|████████▏ | 24493/30000 [14:55:23<3:24:01, 2.22s/it] 82%|████████▏ | 24494/30000 [14:55:24<2:54:43, 1.90s/it] 82%|████████▏ | 24495/30000 [14:55:25<2:34:35, 1.68s/it] 82%|████████▏ | 24496/30000 [14:55:26<2:21:48, 1.55s/it] 82%|████████▏ | 24497/30000 [14:55:28<2:12:33, 1.45s/it] 82%|████████▏ | 24498/30000 [14:55:29<2:06:39, 1.38s/it] 82%|████████▏ | 24499/30000 [14:55:36<4:44:33, 3.10s/it] 82%|████████▏ | 24500/30000 [14:55:37<3:52:07, 2.53s/it] 82%|████████▏ | 24500/30000 [14:55:37<3:52:07, 2.53s/it] 82%|████████▏ | 24501/30000 [14:55:38<3:15:31, 2.13s/it] 82%|████████▏ | 24502/30000 [14:55:40<2:49:18, 1.85s/it] 82%|████████▏ | 24503/30000 [14:55:41<2:30:38, 1.64s/it] 82%|████████▏ | 24504/30000 [14:55:42<2:18:35, 1.51s/it] 82%|████████▏ | 24505/30000 [14:55:43<2:09:27, 1.41s/it] 82%|████████▏ | 24506/30000 [14:55:44<2:03:15, 1.35s/it] 82%|████████▏ | 24507/30000 [14:55:52<4:46:28, 3.13s/it] 82%|████████▏ | 24508/30000 [14:55:53<3:53:48, 2.55s/it] 82%|████████▏ | 24509/30000 [14:55:54<3:16:48, 2.15s/it] 82%|████████▏ | 24510/30000 [14:55:55<2:50:34, 1.86s/it] 82%|████████▏ | 24510/30000 [14:55:55<2:50:34, 1.86s/it] 82%|████████▏ | 24511/30000 [14:55:56<2:31:13, 1.65s/it] 82%|████████▏ | 24512/30000 [14:55:57<2:18:04, 1.51s/it] 82%|████████▏ | 24513/30000 [14:55:59<2:10:04, 1.42s/it] 82%|████████▏ | 24514/30000 [14:56:00<2:04:39, 1.36s/it] 82%|████████▏ | 24515/30000 [14:56:07<4:48:04, 3.15s/it] 82%|████████▏ | 24516/30000 [14:56:08<3:53:34, 2.56s/it] 82%|████████▏ | 24517/30000 [14:56:10<3:16:39, 2.15s/it] 82%|████████▏ | 24518/30000 [14:56:11<2:50:06, 1.86s/it] 82%|████████▏ | 24519/30000 [14:56:12<2:31:16, 1.66s/it] 82%|████████▏ | 24520/30000 [14:56:13<2:18:07, 1.51s/it] 82%|████████▏ | 24520/30000 [14:56:13<2:18:07, 1.51s/it] 82%|████████▏ | 24521/30000 [14:56:14<2:11:59, 1.45s/it] 82%|████████▏ | 24522/30000 [14:56:16<2:05:25, 1.37s/it] 82%|████████▏ | 24523/30000 [14:56:23<4:48:35, 3.16s/it] 82%|████████▏ | 24524/30000 [14:56:24<3:54:02, 2.56s/it] 82%|████████▏ | 24525/30000 [14:56:25<3:17:04, 2.16s/it] 82%|████████▏ | 24526/30000 [14:56:27<2:50:10, 1.87s/it] 82%|████████▏ | 24527/30000 [14:56:28<2:31:41, 1.66s/it] 82%|████████▏ | 24528/30000 [14:56:29<2:18:53, 1.52s/it] 82%|████████▏ | 24529/30000 [14:56:30<2:11:19, 1.44s/it] 82%|████████▏ | 24530/30000 [14:56:31<2:05:10, 1.37s/it] 82%|████████▏ | 24530/30000 [14:56:31<2:05:10, 1.37s/it] 82%|████████▏ | 24531/30000 [14:56:39<4:45:11, 3.13s/it] 82%|████████▏ | 24532/30000 [14:56:40<3:51:42, 2.54s/it] 82%|████████▏ | 24533/30000 [14:56:41<3:14:27, 2.13s/it] 82%|████████▏ | 24534/30000 [14:56:42<2:48:34, 1.85s/it] 82%|████████▏ | 24535/30000 [14:56:43<2:30:15, 1.65s/it] 82%|████████▏ | 24536/30000 [14:56:45<2:17:37, 1.51s/it] 82%|████████▏ | 24537/30000 [14:56:46<2:09:32, 1.42s/it] 82%|████████▏ | 24538/30000 [14:56:47<2:02:55, 1.35s/it] 82%|████████▏ | 24539/30000 [14:56:54<4:39:03, 3.07s/it] 82%|████████▏ | 24540/30000 [14:56:55<3:47:15, 2.50s/it] 82%|████████▏ | 24540/30000 [14:56:55<3:47:15, 2.50s/it] 82%|████████▏ | 24541/30000 [14:56:56<3:11:26, 2.10s/it] 82%|████████▏ | 24542/30000 [14:56:58<2:46:21, 1.83s/it] 82%|████████▏ | 24543/30000 [14:56:59<2:28:06, 1.63s/it] 82%|████████▏ | 24544/30000 [14:57:00<2:15:34, 1.49s/it] 82%|████████▏ | 24545/30000 [14:57:01<2:08:29, 1.41s/it] 82%|████████▏ | 24546/30000 [14:57:02<2:02:17, 1.35s/it] 82%|████████▏ | 24547/30000 [14:57:10<4:57:40, 3.28s/it] 82%|████████▏ | 24548/30000 [14:57:11<4:01:15, 2.66s/it] 82%|████████▏ | 24549/30000 [14:57:13<3:22:24, 2.23s/it] 82%|████████▏ | 24550/30000 [14:57:14<2:53:11, 1.91s/it] 82%|████████▏ | 24550/30000 [14:57:14<2:53:11, 1.91s/it] 82%|████████▏ | 24551/30000 [14:57:15<2:33:10, 1.69s/it] 82%|████████▏ | 24552/30000 [14:57:16<2:19:46, 1.54s/it] 82%|████████▏ | 24553/30000 [14:57:17<2:11:20, 1.45s/it] 82%|████████▏ | 24554/30000 [14:57:18<2:04:31, 1.37s/it] 82%|████████▏ | 24555/30000 [14:57:26<4:53:08, 3.23s/it] 82%|████████▏ | 24556/30000 [14:57:27<3:56:49, 2.61s/it] 82%|████████▏ | 24557/30000 [14:57:28<3:18:12, 2.18s/it] 82%|████████▏ | 24558/30000 [14:57:30<2:50:53, 1.88s/it] 82%|████████▏ | 24559/30000 [14:57:31<2:31:31, 1.67s/it] 82%|████████▏ | 24560/30000 [14:57:32<2:18:31, 1.53s/it] 82%|████████▏ | 24560/30000 [14:57:32<2:18:31, 1.53s/it] 82%|████████▏ | 24561/30000 [14:57:33<2:09:33, 1.43s/it] 82%|████████▏ | 24562/30000 [14:57:34<2:03:07, 1.36s/it] 82%|████████▏ | 24563/30000 [14:57:40<4:08:35, 2.74s/it] 82%|████████▏ | 24564/30000 [14:57:42<3:26:33, 2.28s/it] 82%|████████▏ | 24565/30000 [14:57:43<2:56:59, 1.95s/it] 82%|████████▏ | 24566/30000 [14:57:44<2:35:52, 1.72s/it] 82%|████████▏ | 24567/30000 [14:57:45<2:21:01, 1.56s/it] 82%|████████▏ | 24568/30000 [14:57:46<2:10:39, 1.44s/it] 82%|████████▏ | 24569/30000 [14:57:47<2:04:30, 1.38s/it] 82%|████████▏ | 24570/30000 [14:57:49<1:59:57, 1.33s/it] 82%|████████▏ | 24570/30000 [14:57:49<1:59:57, 1.33s/it] 82%|████████▏ | 24571/30000 [14:57:55<4:25:54, 2.94s/it] 82%|████████▏ | 24572/30000 [14:57:57<3:37:50, 2.41s/it] 82%|████████▏ | 24573/30000 [14:57:58<3:04:56, 2.04s/it] 82%|████████▏ | 24574/30000 [14:57:59<2:41:28, 1.79s/it] 82%|████████▏ | 24575/30000 [14:58:00<2:24:50, 1.60s/it] 82%|████████▏ | 24576/30000 [14:58:01<2:14:09, 1.48s/it] 82%|████████▏ | 24577/30000 [14:58:03<2:06:51, 1.40s/it] 82%|████████▏ | 24578/30000 [14:58:04<2:01:12, 1.34s/it] 82%|████████▏ | 24579/30000 [14:58:11<4:34:54, 3.04s/it] 82%|████████▏ | 24580/30000 [14:58:12<3:44:11, 2.48s/it] 82%|████████▏ | 24580/30000 [14:58:12<3:44:11, 2.48s/it] 82%|████████▏ | 24581/30000 [14:58:13<3:09:32, 2.10s/it] 82%|████████▏ | 24582/30000 [14:58:14<2:44:38, 1.82s/it] 82%|████████▏ | 24583/30000 [14:58:16<2:27:52, 1.64s/it] 82%|████████▏ | 24584/30000 [14:58:17<2:16:16, 1.51s/it] 82%|████████▏ | 24585/30000 [14:58:18<2:07:59, 1.42s/it] 82%|████████▏ | 24586/30000 [14:58:19<2:01:59, 1.35s/it] 82%|████████▏ | 24587/30000 [14:58:25<4:08:43, 2.76s/it] 82%|████████▏ | 24588/30000 [14:58:26<3:25:39, 2.28s/it] 82%|████████▏ | 24589/30000 [14:58:28<2:56:07, 1.95s/it] 82%|████████▏ | 24590/30000 [14:58:29<2:34:28, 1.71s/it] 82%|████████▏ | 24590/30000 [14:58:29<2:34:28, 1.71s/it] 82%|████████▏ | 24591/30000 [14:58:30<2:19:50, 1.55s/it] 82%|████████▏ | 24592/30000 [14:58:31<2:09:56, 1.44s/it] 82%|████████▏ | 24593/30000 [14:58:32<2:03:19, 1.37s/it] 82%|████████▏ | 24594/30000 [14:58:33<1:59:20, 1.32s/it] 82%|████████▏ | 24595/30000 [14:58:39<4:03:51, 2.71s/it] 82%|████████▏ | 24596/30000 [14:58:41<3:22:40, 2.25s/it] 82%|████████▏ | 24597/30000 [14:58:42<2:55:06, 1.94s/it] 82%|████████▏ | 24598/30000 [14:58:43<2:34:08, 1.71s/it] 82%|████████▏ | 24599/30000 [14:58:44<2:20:01, 1.56s/it] 82%|████████▏ | 24600/30000 [14:58:45<2:10:17, 1.45s/it] 82%|████████▏ | 24600/30000 [14:58:45<2:10:17, 1.45s/it] 82%|████████▏ | 24601/30000 [14:58:47<2:04:06, 1.38s/it] 82%|████████▏ | 24602/30000 [14:58:48<1:59:22, 1.33s/it] 82%|████████▏ | 24603/30000 [14:58:55<4:32:31, 3.03s/it] 82%|████████▏ | 24604/30000 [14:58:56<3:43:17, 2.48s/it] 82%|████████▏ | 24605/30000 [14:58:57<3:10:42, 2.12s/it] 82%|████████▏ | 24606/30000 [14:58:58<2:45:07, 1.84s/it] 82%|████████▏ | 24607/30000 [14:59:00<2:27:46, 1.64s/it] 82%|████████▏ | 24608/30000 [14:59:01<2:15:38, 1.51s/it] 82%|████████▏ | 24609/30000 [14:59:02<2:07:10, 1.42s/it] 82%|████████▏ | 24610/30000 [14:59:03<2:01:55, 1.36s/it] 82%|████████▏ | 24610/30000 [14:59:03<2:01:55, 1.36s/it] 82%|████████▏ | 24611/30000 [14:59:10<4:32:58, 3.04s/it] 82%|████████▏ | 24612/30000 [14:59:11<3:42:38, 2.48s/it] 82%|████████▏ | 24613/30000 [14:59:13<3:08:17, 2.10s/it] 82%|████████▏ | 24614/30000 [14:59:14<2:43:30, 1.82s/it] 82%|████████▏ | 24615/30000 [14:59:15<2:26:45, 1.64s/it] 82%|████████▏ | 24616/30000 [14:59:16<2:15:00, 1.50s/it] 82%|████████▏ | 24617/30000 [14:59:17<2:06:32, 1.41s/it] 82%|████████▏ | 24618/30000 [14:59:19<2:00:47, 1.35s/it] 82%|████████▏ | 24619/30000 [14:59:25<4:20:27, 2.90s/it] 82%|████████▏ | 24620/30000 [14:59:26<3:34:23, 2.39s/it] 82%|████████▏ | 24620/30000 [14:59:26<3:34:23, 2.39s/it] 82%|████████▏ | 24621/30000 [14:59:28<3:03:28, 2.05s/it] 82%|████████▏ | 24622/30000 [14:59:29<2:40:03, 1.79s/it] 82%|████████▏ | 24623/30000 [14:59:30<2:24:07, 1.61s/it] 82%|████████▏ | 24624/30000 [14:59:31<2:13:00, 1.48s/it] 82%|████████▏ | 24625/30000 [14:59:32<2:05:04, 1.40s/it] 82%|████████▏ | 24626/30000 [14:59:33<1:59:13, 1.33s/it] 82%|████████▏ | 24627/30000 [14:59:40<4:14:39, 2.84s/it] 82%|████████▏ | 24628/30000 [14:59:41<3:30:40, 2.35s/it] 82%|████████▏ | 24629/30000 [14:59:42<3:00:12, 2.01s/it] 82%|████████▏ | 24630/30000 [14:59:43<2:38:40, 1.77s/it] 82%|████████▏ | 24630/30000 [14:59:43<2:38:40, 1.77s/it] 82%|████████▏ | 24631/30000 [14:59:45<2:22:32, 1.59s/it] 82%|████████▏ | 24632/30000 [14:59:46<2:11:58, 1.48s/it] 82%|████████▏ | 24633/30000 [14:59:47<2:04:29, 1.39s/it] 82%|████████▏ | 24634/30000 [14:59:48<1:59:12, 1.33s/it] 82%|████████▏ | 24635/30000 [14:59:56<4:44:51, 3.19s/it] 82%|████████▏ | 24636/30000 [14:59:57<3:51:08, 2.59s/it] 82%|████████▏ | 24637/30000 [14:59:58<3:13:40, 2.17s/it] 82%|████████▏ | 24638/30000 [14:59:59<2:47:15, 1.87s/it] 82%|████████▏ | 24639/30000 [15:00:00<2:28:37, 1.66s/it] 82%|████████▏ | 24640/30000 [15:00:02<2:16:34, 1.53s/it] 82%|████████▏ | 24640/30000 [15:00:02<2:16:34, 1.53s/it] 82%|████████▏ | 24641/30000 [15:00:03<2:08:20, 1.44s/it] 82%|████████▏ | 24642/30000 [15:00:04<2:01:42, 1.36s/it] 82%|████████▏ | 24643/30000 [15:00:10<4:13:16, 2.84s/it] 82%|████████▏ | 24644/30000 [15:00:12<3:29:44, 2.35s/it] 82%|████████▏ | 24645/30000 [15:00:13<2:59:10, 2.01s/it] 82%|████████▏ | 24646/30000 [15:00:14<2:36:25, 1.75s/it] 82%|████████▏ | 24647/30000 [15:00:15<2:20:23, 1.57s/it] 82%|████████▏ | 24648/30000 [15:00:16<2:10:40, 1.47s/it] 82%|████████▏ | 24649/30000 [15:00:18<2:02:58, 1.38s/it] 82%|████████▏ | 24650/30000 [15:00:19<1:58:21, 1.33s/it] 82%|████████▏ | 24650/30000 [15:00:19<1:58:21, 1.33s/it] 82%|████████▏ | 24651/30000 [15:00:25<4:14:05, 2.85s/it] 82%|████████▏ | 24652/30000 [15:00:26<3:29:20, 2.35s/it] 82%|████████▏ | 24653/30000 [15:00:27<2:58:35, 2.00s/it] 82%|████████▏ | 24654/30000 [15:00:29<2:37:08, 1.76s/it] 82%|████████▏ | 24655/30000 [15:00:30<2:21:32, 1.59s/it] 82%|████████▏ | 24656/30000 [15:00:31<2:11:17, 1.47s/it] 82%|████████▏ | 24657/30000 [15:00:32<2:03:34, 1.39s/it] 82%|████████▏ | 24658/30000 [15:00:33<1:58:30, 1.33s/it] 82%|████████▏ | 24659/30000 [15:00:40<4:08:15, 2.79s/it] 82%|████████▏ | 24660/30000 [15:00:41<3:25:50, 2.31s/it] 82%|████████▏ | 24660/30000 [15:00:41<3:25:50, 2.31s/it] 82%|████████▏ | 24661/30000 [15:00:42<2:57:46, 2.00s/it] 82%|████████▏ | 24662/30000 [15:00:43<2:35:51, 1.75s/it] 82%|████████▏ | 24663/30000 [15:00:44<2:20:05, 1.57s/it] 82%|████████▏ | 24664/30000 [15:00:46<2:11:13, 1.48s/it] 82%|████████▏ | 24665/30000 [15:00:47<2:04:55, 1.40s/it] 82%|████████▏ | 24666/30000 [15:00:48<2:00:32, 1.36s/it] 82%|████████▏ | 24667/30000 [15:00:55<4:24:58, 2.98s/it] 82%|████████▏ | 24668/30000 [15:00:56<3:37:24, 2.45s/it] 82%|████████▏ | 24669/30000 [15:00:57<3:03:29, 2.07s/it] 82%|████████▏ | 24670/30000 [15:00:59<2:39:40, 1.80s/it] 82%|████████▏ | 24670/30000 [15:00:59<2:39:40, 1.80s/it] 82%|████████▏ | 24671/30000 [15:01:00<2:23:12, 1.61s/it] 82%|████████▏ | 24672/30000 [15:01:01<2:12:21, 1.49s/it] 82%|████████▏ | 24673/30000 [15:01:02<2:04:56, 1.41s/it] 82%|████████▏ | 24674/30000 [15:01:03<1:59:21, 1.34s/it] 82%|████████▏ | 24675/30000 [15:01:11<4:43:47, 3.20s/it] 82%|████████▏ | 24676/30000 [15:01:12<3:50:21, 2.60s/it] 82%|████████▏ | 24677/30000 [15:01:13<3:13:47, 2.18s/it] 82%|████████▏ | 24678/30000 [15:01:14<2:46:36, 1.88s/it] 82%|████████▏ | 24679/30000 [15:01:16<2:27:54, 1.67s/it] 82%|████████▏ | 24680/30000 [15:01:17<2:15:08, 1.52s/it] 82%|████████▏ | 24680/30000 [15:01:17<2:15:08, 1.52s/it] 82%|████████▏ | 24681/30000 [15:01:18<2:06:23, 1.43s/it] 82%|████████▏ | 24682/30000 [15:01:19<2:01:11, 1.37s/it] 82%|████████▏ | 24683/30000 [15:01:26<4:22:00, 2.96s/it] 82%|████████▏ | 24684/30000 [15:01:27<3:34:29, 2.42s/it] 82%|████████▏ | 24685/30000 [15:01:28<3:02:28, 2.06s/it] 82%|████████▏ | 24686/30000 [15:01:29<2:38:42, 1.79s/it] 82%|████████▏ | 24687/30000 [15:01:31<2:22:44, 1.61s/it] 82%|████████▏ | 24688/30000 [15:01:32<2:10:50, 1.48s/it] 82%|████████▏ | 24689/30000 [15:01:33<2:03:02, 1.39s/it] 82%|████████▏ | 24690/30000 [15:01:34<1:57:03, 1.32s/it] 82%|████████▏ | 24690/30000 [15:01:34<1:57:03, 1.32s/it] 82%|████████▏ | 24691/30000 [15:01:40<3:58:15, 2.69s/it] 82%|████████▏ | 24692/30000 [15:01:41<3:17:36, 2.23s/it] 82%|████████▏ | 24693/30000 [15:01:42<2:49:47, 1.92s/it] 82%|████████▏ | 24694/30000 [15:01:44<2:29:42, 1.69s/it] 82%|████████▏ | 24695/30000 [15:01:45<2:15:30, 1.53s/it] 82%|████████▏ | 24696/30000 [15:01:46<2:06:21, 1.43s/it] 82%|████████▏ | 24697/30000 [15:01:47<1:59:30, 1.35s/it] 82%|████████▏ | 24698/30000 [15:01:48<1:54:51, 1.30s/it] 82%|████████▏ | 24699/30000 [15:01:54<3:53:30, 2.64s/it] 82%|████████▏ | 24700/30000 [15:01:55<3:14:31, 2.20s/it] 82%|████████▏ | 24700/30000 [15:01:55<3:14:31, 2.20s/it] 82%|████████▏ | 24701/30000 [15:01:56<2:48:39, 1.91s/it] 82%|████████▏ | 24702/30000 [15:01:58<2:29:32, 1.69s/it] 82%|████████▏ | 24703/30000 [15:01:59<2:15:53, 1.54s/it] 82%|████████▏ | 24704/30000 [15:02:00<2:07:04, 1.44s/it] 82%|████████▏ | 24705/30000 [15:02:01<2:00:37, 1.37s/it] 82%|████████▏ | 24706/30000 [15:02:02<1:56:03, 1.32s/it] 82%|████████▏ | 24707/30000 [15:02:10<4:48:01, 3.27s/it] 82%|████████▏ | 24708/30000 [15:02:11<3:53:30, 2.65s/it] 82%|████████▏ | 24709/30000 [15:02:13<3:15:32, 2.22s/it] 82%|████████▏ | 24710/30000 [15:02:14<2:48:07, 1.91s/it] 82%|████████▏ | 24710/30000 [15:02:14<2:48:07, 1.91s/it] 82%|████████▏ | 24711/30000 [15:02:15<2:28:57, 1.69s/it] 82%|████████▏ | 24712/30000 [15:02:16<2:20:17, 1.59s/it] 82%|████████▏ | 24713/30000 [15:02:18<2:09:20, 1.47s/it] 82%|████████▏ | 24714/30000 [15:02:19<2:02:12, 1.39s/it] 82%|████████▏ | 24715/30000 [15:02:27<5:03:46, 3.45s/it] 82%|████████▏ | 24716/30000 [15:02:28<4:03:48, 2.77s/it] 82%|████████▏ | 24717/30000 [15:02:29<3:23:10, 2.31s/it] 82%|████████▏ | 24718/30000 [15:02:31<2:53:34, 1.97s/it] 82%|████████▏ | 24719/30000 [15:02:32<2:32:38, 1.73s/it] 82%|████████▏ | 24720/30000 [15:02:33<2:22:47, 1.62s/it] 82%|████████▏ | 24720/30000 [15:02:33<2:22:47, 1.62s/it] 82%|████████▏ | 24721/30000 [15:02:34<2:13:31, 1.52s/it] 82%|████████▏ | 24722/30000 [15:02:36<2:05:47, 1.43s/it] 82%|████████▏ | 24723/30000 [15:02:44<5:14:32, 3.58s/it] 82%|████████▏ | 24724/30000 [15:02:45<4:11:23, 2.86s/it] 82%|████████▏ | 24725/30000 [15:02:47<3:27:09, 2.36s/it] 82%|████████▏ | 24726/30000 [15:02:48<2:55:23, 2.00s/it] 82%|████████▏ | 24727/30000 [15:02:49<2:34:10, 1.75s/it] 82%|████████▏ | 24728/30000 [15:02:50<2:20:57, 1.60s/it] 82%|████████▏ | 24729/30000 [15:02:51<2:09:53, 1.48s/it] 82%|████████▏ | 24730/30000 [15:02:53<2:04:08, 1.41s/it] 82%|████████▏ | 24730/30000 [15:02:53<2:04:08, 1.41s/it] 82%|████████▏ | 24731/30000 [15:03:01<4:58:32, 3.40s/it] 82%|████████▏ | 24732/30000 [15:03:02<3:59:46, 2.73s/it] 82%|████████▏ | 24733/30000 [15:03:03<3:18:46, 2.26s/it] 82%|████████▏ | 24734/30000 [15:03:04<2:49:58, 1.94s/it] 82%|████████▏ | 24735/30000 [15:03:05<2:30:15, 1.71s/it] 82%|████████▏ | 24736/30000 [15:03:07<2:16:54, 1.56s/it] 82%|████████▏ | 24737/30000 [15:03:08<2:07:05, 1.45s/it] 82%|████████▏ | 24738/30000 [15:03:09<2:00:13, 1.37s/it] 82%|████████▏ | 24739/30000 [15:03:18<5:26:57, 3.73s/it] 82%|████████▏ | 24740/30000 [15:03:19<4:20:12, 2.97s/it] 82%|████████▏ | 24740/30000 [15:03:19<4:20:12, 2.97s/it] 82%|████████▏ | 24741/30000 [15:03:21<3:34:31, 2.45s/it] 82%|████████▏ | 24742/30000 [15:03:22<3:00:58, 2.07s/it] 82%|████████▏ | 24743/30000 [15:03:23<2:37:08, 1.79s/it] 82%|████████▏ | 24744/30000 [15:03:24<2:21:18, 1.61s/it] 82%|████████▏ | 24745/30000 [15:03:25<2:10:09, 1.49s/it] 82%|████████▏ | 24746/30000 [15:03:27<2:02:43, 1.40s/it] 82%|████████▏ | 24747/30000 [15:03:36<5:26:13, 3.73s/it] 82%|████████▏ | 24748/30000 [15:03:37<4:19:11, 2.96s/it] 82%|████████▏ | 24749/30000 [15:03:38<3:32:00, 2.42s/it] 82%|████████▎ | 24750/30000 [15:03:39<2:59:03, 2.05s/it] 82%|████████▎ | 24750/30000 [15:03:39<2:59:03, 2.05s/it] 83%|████████▎ | 24751/30000 [15:03:40<2:36:21, 1.79s/it] 83%|████████▎ | 24752/30000 [15:03:42<2:20:23, 1.61s/it] 83%|████████▎ | 24753/30000 [15:03:43<2:09:20, 1.48s/it] 83%|████████▎ | 24754/30000 [15:03:44<2:01:48, 1.39s/it] 83%|████████▎ | 24755/30000 [15:03:53<5:31:29, 3.79s/it] 83%|████████▎ | 24756/30000 [15:03:54<4:22:28, 3.00s/it] 83%|████████▎ | 24757/30000 [15:03:56<3:35:32, 2.47s/it] 83%|████████▎ | 24758/30000 [15:03:57<3:02:20, 2.09s/it] 83%|████████▎ | 24759/30000 [15:03:58<2:38:04, 1.81s/it] 83%|████████▎ | 24760/30000 [15:03:59<2:22:19, 1.63s/it] 83%|████████▎ | 24760/30000 [15:03:59<2:22:19, 1.63s/it] 83%|████████▎ | 24761/30000 [15:04:00<2:10:46, 1.50s/it] 83%|████████▎ | 24762/30000 [15:04:02<2:02:25, 1.40s/it] 83%|████████▎ | 24763/30000 [15:04:11<5:30:19, 3.78s/it] 83%|████████▎ | 24764/30000 [15:04:12<4:22:10, 3.00s/it] 83%|████████▎ | 24765/30000 [15:04:13<3:34:24, 2.46s/it] 83%|████████▎ | 24766/30000 [15:04:15<3:00:53, 2.07s/it] 83%|████████▎ | 24767/30000 [15:04:16<2:37:25, 1.80s/it] 83%|████████▎ | 24768/30000 [15:04:17<2:21:50, 1.63s/it] 83%|████████▎ | 24769/30000 [15:04:18<2:11:02, 1.50s/it] 83%|████████▎ | 24770/30000 [15:04:19<2:03:14, 1.41s/it] 83%|████████▎ | 24770/30000 [15:04:19<2:03:14, 1.41s/it] 83%|████████▎ | 24771/30000 [15:04:28<5:03:33, 3.48s/it] 83%|████████▎ | 24772/30000 [15:04:29<4:03:04, 2.79s/it] 83%|████████▎ | 24773/30000 [15:04:30<3:24:13, 2.34s/it] 83%|████████▎ | 24774/30000 [15:04:31<2:53:57, 2.00s/it] 83%|████████▎ | 24775/30000 [15:04:32<2:32:30, 1.75s/it] 83%|████████▎ | 24776/30000 [15:04:34<2:17:04, 1.57s/it] 83%|████████▎ | 24777/30000 [15:04:35<2:06:45, 1.46s/it] 83%|████████▎ | 24778/30000 [15:04:36<2:00:02, 1.38s/it] 83%|████████▎ | 24779/30000 [15:04:45<5:13:10, 3.60s/it] 83%|████████▎ | 24780/30000 [15:04:46<4:09:23, 2.87s/it] 83%|████████▎ | 24780/30000 [15:04:46<4:09:23, 2.87s/it] 83%|████████▎ | 24781/30000 [15:04:47<3:27:19, 2.38s/it] 83%|████████▎ | 24782/30000 [15:04:48<2:56:02, 2.02s/it] 83%|████████▎ | 24783/30000 [15:04:50<2:34:02, 1.77s/it] 83%|████████▎ | 24784/30000 [15:04:51<2:19:15, 1.60s/it] 83%|████████▎ | 24785/30000 [15:04:52<2:08:37, 1.48s/it] 83%|████████▎ | 24786/30000 [15:04:53<2:01:47, 1.40s/it] 83%|████████▎ | 24787/30000 [15:05:01<4:35:58, 3.18s/it] 83%|████████▎ | 24788/30000 [15:05:02<3:43:32, 2.57s/it] 83%|████████▎ | 24789/30000 [15:05:03<3:07:43, 2.16s/it] 83%|████████▎ | 24790/30000 [15:05:04<2:41:52, 1.86s/it] 83%|████████▎ | 24790/30000 [15:05:04<2:41:52, 1.86s/it] 83%|████████▎ | 24791/30000 [15:05:05<2:23:30, 1.65s/it] 83%|████████▎ | 24792/30000 [15:05:06<2:11:44, 1.52s/it] 83%|████████▎ | 24793/30000 [15:05:08<2:03:45, 1.43s/it] 83%|████████▎ | 24794/30000 [15:05:09<1:57:22, 1.35s/it] 83%|████████▎ | 24795/30000 [15:05:16<4:29:59, 3.11s/it] 83%|████████▎ | 24796/30000 [15:05:17<3:39:38, 2.53s/it] 83%|████████▎ | 24797/30000 [15:05:18<3:04:41, 2.13s/it] 83%|████████▎ | 24798/30000 [15:05:20<2:39:39, 1.84s/it] 83%|████████▎ | 24799/30000 [15:05:21<2:22:20, 1.64s/it] 83%|████████▎ | 24800/30000 [15:05:22<2:10:34, 1.51s/it] 83%|████████▎ | 24800/30000 [15:05:22<2:10:34, 1.51s/it] 83%|████████▎ | 24801/30000 [15:05:23<2:02:27, 1.41s/it] 83%|████████▎ | 24802/30000 [15:05:24<1:56:34, 1.35s/it] 83%|████████▎ | 24803/30000 [15:05:32<4:43:00, 3.27s/it] 83%|████████▎ | 24804/30000 [15:05:33<3:48:25, 2.64s/it] 83%|████████▎ | 24805/30000 [15:05:35<3:13:11, 2.23s/it] 83%|████████▎ | 24806/30000 [15:05:36<2:45:20, 1.91s/it] 83%|████████▎ | 24807/30000 [15:05:37<2:26:50, 1.70s/it] 83%|████████▎ | 24808/30000 [15:05:38<2:16:14, 1.57s/it] 83%|████████▎ | 24809/30000 [15:05:40<2:10:11, 1.50s/it] 83%|████████▎ | 24810/30000 [15:05:41<2:02:07, 1.41s/it] 83%|████████▎ | 24810/30000 [15:05:41<2:02:07, 1.41s/it] 83%|████████▎ | 24811/30000 [15:05:47<4:19:06, 3.00s/it] 83%|████████▎ | 24812/30000 [15:05:49<3:31:53, 2.45s/it] 83%|████████▎ | 24813/30000 [15:05:50<2:59:39, 2.08s/it] 83%|████████▎ | 24814/30000 [15:05:51<2:36:10, 1.81s/it] 83%|████████▎ | 24815/30000 [15:05:52<2:19:51, 1.62s/it] 83%|████████▎ | 24816/30000 [15:05:53<2:08:57, 1.49s/it] 83%|████████▎ | 24817/30000 [15:05:55<2:01:48, 1.41s/it] 83%|████████▎ | 24818/30000 [15:05:56<1:56:20, 1.35s/it] 83%|████████▎ | 24819/30000 [15:06:04<4:42:36, 3.27s/it] 83%|████████▎ | 24820/30000 [15:06:05<3:48:37, 2.65s/it] 83%|████████▎ | 24820/30000 [15:06:05<3:48:37, 2.65s/it] 83%|████████▎ | 24821/30000 [15:06:06<3:11:19, 2.22s/it] 83%|████████▎ | 24822/30000 [15:06:07<2:44:29, 1.91s/it] 83%|████████▎ | 24823/30000 [15:06:08<2:25:57, 1.69s/it] 83%|████████▎ | 24824/30000 [15:06:10<2:15:11, 1.57s/it] 83%|████████▎ | 24825/30000 [15:06:11<2:09:22, 1.50s/it] 83%|████████▎ | 24826/30000 [15:06:12<2:01:10, 1.41s/it] 83%|████████▎ | 24827/30000 [15:06:19<4:32:17, 3.16s/it] 83%|████████▎ | 24828/30000 [15:06:21<3:40:29, 2.56s/it] 83%|████████▎ | 24829/30000 [15:06:22<3:05:33, 2.15s/it] 83%|████████▎ | 24830/30000 [15:06:23<2:40:23, 1.86s/it] 83%|████████▎ | 24830/30000 [15:06:23<2:40:23, 1.86s/it] 83%|████████▎ | 24831/30000 [15:06:24<2:22:53, 1.66s/it] 83%|████████▎ | 24832/30000 [15:06:25<2:11:21, 1.53s/it] 83%|████████▎ | 24833/30000 [15:06:26<2:02:37, 1.42s/it] 83%|████████▎ | 24834/30000 [15:06:28<1:57:20, 1.36s/it] 83%|████████▎ | 24835/30000 [15:06:36<4:47:00, 3.33s/it] 83%|████████▎ | 24836/30000 [15:06:37<3:51:28, 2.69s/it] 83%|████████▎ | 24837/30000 [15:06:38<3:14:16, 2.26s/it] 83%|████████▎ | 24838/30000 [15:06:39<2:46:52, 1.94s/it] 83%|████████▎ | 24839/30000 [15:06:41<2:28:55, 1.73s/it] 83%|████████▎ | 24840/30000 [15:06:42<2:15:04, 1.57s/it] 83%|████████▎ | 24840/30000 [15:06:42<2:15:04, 1.57s/it] 83%|████████▎ | 24841/30000 [15:06:43<2:04:26, 1.45s/it] 83%|████████▎ | 24842/30000 [15:06:44<1:57:27, 1.37s/it] 83%|████████▎ | 24843/30000 [15:06:53<5:21:26, 3.74s/it] 83%|████████▎ | 24844/30000 [15:06:54<4:14:52, 2.97s/it] 83%|████████▎ | 24845/30000 [15:06:56<3:28:36, 2.43s/it] 83%|████████▎ | 24846/30000 [15:06:57<2:59:31, 2.09s/it] 83%|████████▎ | 24847/30000 [15:06:58<2:38:05, 1.84s/it] 83%|████████▎ | 24848/30000 [15:06:59<2:21:32, 1.65s/it] 83%|████████▎ | 24849/30000 [15:07:01<2:11:23, 1.53s/it] 83%|████████▎ | 24850/30000 [15:07:02<2:03:12, 1.44s/it] +{'loss': 0.0133, 'grad_norm': 0.051959872245788574, 'learning_rate': 2.7215774894098474e-06, 'flow_matching_loss': 0.01327802212908864, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.11} +{'loss': 0.0113, 'grad_norm': 0.07458533346652985, 'learning_rate': 2.7120871079601696e-06, 'flow_matching_loss': 0.01134128523990512, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.12} +{'loss': 0.0133, 'grad_norm': 0.06368128955364227, 'learning_rate': 2.7026116574833314e-06, 'flow_matching_loss': 0.013349977415055036, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.12} +{'loss': 0.0126, 'grad_norm': 0.0845566913485527, 'learning_rate': 2.6931511494928953e-06, 'flow_matching_loss': 0.012562236189842224, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.12} +{'loss': 0.0126, 'grad_norm': 0.08125726133584976, 'learning_rate': 2.683705595484275e-06, 'flow_matching_loss': 0.012606816738843918, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.13} +{'loss': 0.0137, 'grad_norm': 0.07409604638814926, 'learning_rate': 2.6742750069347014e-06, 'flow_matching_loss': 0.013678825367242098, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.13} +{'loss': 0.0137, 'grad_norm': 0.0936979278922081, 'learning_rate': 2.664859395303239e-06, 'flow_matching_loss': 0.013686820492148399, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.13} +{'loss': 0.0151, 'grad_norm': 0.09818891435861588, 'learning_rate': 2.655458772030739e-06, 'flow_matching_loss': 0.015125338081270456, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.13} +{'loss': 0.0121, 'grad_norm': 0.11127855628728867, 'learning_rate': 2.6460731485398403e-06, 'flow_matching_loss': 0.01205681897699833, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.14} +{'loss': 0.0142, 'grad_norm': 0.0842173844575882, 'learning_rate': 2.6367025362349695e-06, 'flow_matching_loss': 0.014237461239099502, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.14} +{'loss': 0.0136, 'grad_norm': 0.07710163295269012, 'learning_rate': 2.6273469465022965e-06, 'flow_matching_loss': 0.013627373520284891, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.14} +{'loss': 0.013, 'grad_norm': 0.06570237129926682, 'learning_rate': 2.618006390709753e-06, 'flow_matching_loss': 0.013048865739256144, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.15} +{'loss': 0.0126, 'grad_norm': 0.058185964822769165, 'learning_rate': 2.60868088020699e-06, 'flow_matching_loss': 0.012613368593156338, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.15} +{'loss': 0.0128, 'grad_norm': 0.06023018807172775, 'learning_rate': 2.599370426325382e-06, 'flow_matching_loss': 0.012759647518396377, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.15} +{'loss': 0.0143, 'grad_norm': 0.06926777213811874, 'learning_rate': 2.5900750403780093e-06, 'flow_matching_loss': 0.014306988939642907, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.15} +{'loss': 0.0126, 'grad_norm': 0.0714886337518692, 'learning_rate': 2.580794733659639e-06, 'flow_matching_loss': 0.012614351883530616, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.16} +{'loss': 0.0143, 'grad_norm': 0.06699349731206894, 'learning_rate': 2.5715295174467236e-06, 'flow_matching_loss': 0.014344602916389703, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.16} +{'loss': 0.0125, 'grad_norm': 0.060588542371988297, 'learning_rate': 2.5622794029973746e-06, 'flow_matching_loss': 0.012466399976983667, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.16} +{'loss': 0.0136, 'grad_norm': 0.06389699131250381, 'learning_rate': 2.553044401551351e-06, 'flow_matching_loss': 0.0136436621658504, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.17} +{'loss': 0.0144, 'grad_norm': 0.08247794955968857, 'learning_rate': 2.54382452433005e-06, 'flow_matching_loss': 0.01439961064606905, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.17} +{'loss': 0.0131, 'grad_norm': 0.10571398586034775, 'learning_rate': 2.53461978253649e-06, 'flow_matching_loss': 0.013059384655207396, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.17} +{'loss': 0.0123, 'grad_norm': 0.07228204607963562, 'learning_rate': 2.5254301873553065e-06, 'flow_matching_loss': 0.012258846126496793, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.17} +{'loss': 0.0132, 'grad_norm': 0.060404930263757706, 'learning_rate': 2.5162557499527184e-06, 'flow_matching_loss': 0.013164240680634976, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.18} +{'loss': 0.0127, 'grad_norm': 0.061132289469242096, 'learning_rate': 2.5070964814765335e-06, 'flow_matching_loss': 0.012725449353456496, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.18} +{'loss': 0.0137, 'grad_norm': 0.08273901045322418, 'learning_rate': 2.4979523930561295e-06, 'flow_matching_loss': 0.013678031507879496, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.18} +{'loss': 0.0119, 'grad_norm': 0.07530416548252106, 'learning_rate': 2.4888234958024333e-06, 'flow_matching_loss': 0.01191875832155347, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.19} +{'loss': 0.0142, 'grad_norm': 0.08369842916727066, 'learning_rate': 2.479709800807915e-06, 'flow_matching_loss': 0.014158102590590716, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.19} +{'loss': 0.0131, 'grad_norm': 0.09264060854911804, 'learning_rate': 2.4706113191465767e-06, 'flow_matching_loss': 0.013087652344256639, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.19} +{'loss': 0.013, 'grad_norm': 0.0836668461561203, 'learning_rate': 2.46152806187393e-06, 'flow_matching_loss': 0.012982225371524691, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.2} +{'loss': 0.0121, 'grad_norm': 0.07441049069166183, 'learning_rate': 2.452460040026989e-06, 'flow_matching_loss': 0.012134405691176653, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.2} +{'loss': 0.0139, 'grad_norm': 0.07621421664953232, 'learning_rate': 2.4434072646242555e-06, 'flow_matching_loss': 0.013944985345005989, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.2} +{'loss': 0.0124, 'grad_norm': 0.0761420950293541, 'learning_rate': 2.4343697466657015e-06, 'flow_matching_loss': 0.012367180921137333, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.2} +{'loss': 0.0117, 'grad_norm': 0.0716809332370758, 'learning_rate': 2.42534749713277e-06, 'flow_matching_loss': 0.01172742024064064, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.21} +{'loss': 0.0136, 'grad_norm': 0.05760239437222481, 'learning_rate': 2.4163405269883425e-06, 'flow_matching_loss': 0.013584075309336186, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.21} +{'loss': 0.0138, 'grad_norm': 0.06101386994123459, 'learning_rate': 2.407348847176735e-06, 'flow_matching_loss': 0.013777763582766056, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.21} +{'loss': 0.0134, 'grad_norm': 0.07487001270055771, 'learning_rate': 2.398372468623688e-06, 'flow_matching_loss': 0.013429543375968933, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.22} +{'loss': 0.0122, 'grad_norm': 0.06392122060060501, 'learning_rate': 2.3894114022363446e-06, 'flow_matching_loss': 0.012228234857320785, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.22} +{'loss': 0.0123, 'grad_norm': 0.07510365545749664, 'learning_rate': 2.3804656589032514e-06, 'flow_matching_loss': 0.012323313299566508, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.22} +{'loss': 0.0126, 'grad_norm': 0.06851806491613388, 'learning_rate': 2.3715352494943237e-06, 'flow_matching_loss': 0.012571858149021864, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.22} +{'loss': 0.013, 'grad_norm': 0.0656377524137497, 'learning_rate': 2.3626201848608582e-06, 'flow_matching_loss': 0.01299173878505826, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.23} +{'loss': 0.0132, 'grad_norm': 0.08375021815299988, 'learning_rate': 2.353720475835494e-06, 'flow_matching_loss': 0.013223565369844436, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.23} 83%|████████▎ | 24850/30000 [15:07:02<2:03:12, 1.44s/it] 83%|████████▎ | 24851/30000 [15:07:11<5:16:04, 3.68s/it] 83%|████████▎ | 24852/30000 [15:07:12<4:11:08, 2.93s/it] 83%|████████▎ | 24853/30000 [15:07:13<3:26:24, 2.41s/it] 83%|████████▎ | 24854/30000 [15:07:14<2:54:34, 2.04s/it] 83%|████████▎ | 24855/30000 [15:07:16<2:33:02, 1.78s/it] 83%|████████▎ | 24856/30000 [15:07:17<2:17:41, 1.61s/it] 83%|████████▎ | 24857/30000 [15:07:18<2:06:46, 1.48s/it] 83%|████████▎ | 24858/30000 [15:07:19<2:00:23, 1.40s/it] 83%|████████▎ | 24859/30000 [15:07:27<4:37:41, 3.24s/it] 83%|████████▎ | 24860/30000 [15:07:28<3:44:56, 2.63s/it] 83%|████████▎ | 24860/30000 [15:07:28<3:44:56, 2.63s/it] 83%|████████▎ | 24861/30000 [15:07:29<3:09:19, 2.21s/it] 83%|████████▎ | 24862/30000 [15:07:30<2:42:49, 1.90s/it] 83%|████████▎ | 24863/30000 [15:07:31<2:23:59, 1.68s/it] 83%|████████▎ | 24864/30000 [15:07:33<2:11:54, 1.54s/it] 83%|████████▎ | 24865/30000 [15:07:34<2:02:42, 1.43s/it] 83%|████████▎ | 24866/30000 [15:07:35<1:57:14, 1.37s/it] 83%|████████▎ | 24867/30000 [15:07:44<5:05:07, 3.57s/it] 83%|████████▎ | 24868/30000 [15:07:45<4:04:02, 2.85s/it] 83%|████████▎ | 24869/30000 [15:07:46<3:21:33, 2.36s/it] 83%|████████▎ | 24870/30000 [15:07:47<2:51:23, 2.00s/it] 83%|████████▎ | 24870/30000 [15:07:47<2:51:23, 2.00s/it] 83%|████████▎ | 24871/30000 [15:07:49<2:30:24, 1.76s/it] 83%|████████▎ | 24872/30000 [15:07:50<2:16:01, 1.59s/it] 83%|████████▎ | 24873/30000 [15:07:51<2:06:48, 1.48s/it] 83%|████████▎ | 24874/30000 [15:07:52<1:59:11, 1.40s/it] 83%|████████▎ | 24875/30000 [15:08:01<5:14:01, 3.68s/it] 83%|████████▎ | 24876/30000 [15:08:02<4:09:35, 2.92s/it] 83%|████████▎ | 24877/30000 [15:08:03<3:25:18, 2.40s/it] 83%|████████▎ | 24878/30000 [15:08:05<2:54:00, 2.04s/it] 83%|████████▎ | 24879/30000 [15:08:06<2:31:38, 1.78s/it] 83%|████████▎ | 24880/30000 [15:08:07<2:16:37, 1.60s/it] 83%|████████▎ | 24880/30000 [15:08:07<2:16:37, 1.60s/it] 83%|████████▎ | 24881/30000 [15:08:08<2:06:32, 1.48s/it] 83%|████████▎ | 24882/30000 [15:08:09<1:58:59, 1.40s/it] 83%|████████▎ | 24883/30000 [15:08:19<5:21:05, 3.77s/it] 83%|████████▎ | 24884/30000 [15:08:20<4:14:41, 2.99s/it] 83%|████████▎ | 24885/30000 [15:08:21<3:28:55, 2.45s/it] 83%|████████▎ | 24886/30000 [15:08:22<2:56:04, 2.07s/it] 83%|████████▎ | 24887/30000 [15:08:23<2:33:21, 1.80s/it] 83%|████████▎ | 24888/30000 [15:08:25<2:17:18, 1.61s/it] 83%|████████▎ | 24889/30000 [15:08:26<2:06:21, 1.48s/it] 83%|████████▎ | 24890/30000 [15:08:27<2:02:08, 1.43s/it] 83%|████████▎ | 24890/30000 [15:08:27<2:02:08, 1.43s/it] 83%|████████▎ | 24891/30000 [15:08:35<4:45:52, 3.36s/it] 83%|████████▎ | 24892/30000 [15:08:36<3:49:42, 2.70s/it] 83%|████████▎ | 24893/30000 [15:08:37<3:11:18, 2.25s/it] 83%|████████▎ | 24894/30000 [15:08:39<2:43:55, 1.93s/it] 83%|████████▎ | 24895/30000 [15:08:40<2:25:07, 1.71s/it] 83%|████████▎ | 24896/30000 [15:08:41<2:12:20, 1.56s/it] 83%|████████▎ | 24897/30000 [15:08:42<2:04:02, 1.46s/it] 83%|████████▎ | 24898/30000 [15:08:43<1:57:42, 1.38s/it] 83%|████████▎ | 24899/30000 [15:08:52<5:02:17, 3.56s/it] 83%|████████▎ | 24900/30000 [15:08:53<4:01:33, 2.84s/it] 83%|████████▎ | 24900/30000 [15:08:53<4:01:33, 2.84s/it] 83%|████████▎ | 24901/30000 [15:08:54<3:19:55, 2.35s/it] 83%|████████▎ | 24902/30000 [15:08:56<2:50:48, 2.01s/it] 83%|████████▎ | 24903/30000 [15:08:57<2:29:54, 1.76s/it] 83%|████████▎ | 24904/30000 [15:08:58<2:15:29, 1.60s/it] 83%|████████▎ | 24905/30000 [15:08:59<2:05:34, 1.48s/it] 83%|████████▎ | 24906/30000 [15:09:00<2:00:25, 1.42s/it] 83%|████████▎ | 24907/30000 [15:09:08<4:35:33, 3.25s/it] 83%|████████▎ | 24908/30000 [15:09:09<3:42:43, 2.62s/it] 83%|████████▎ | 24909/30000 [15:09:10<3:06:22, 2.20s/it] 83%|████████▎ | 24910/30000 [15:09:12<2:43:01, 1.92s/it] 83%|████████▎ | 24910/30000 [15:09:12<2:43:01, 1.92s/it] 83%|████████▎ | 24911/30000 [15:09:13<2:24:23, 1.70s/it] 83%|████████▎ | 24912/30000 [15:09:14<2:11:43, 1.55s/it] 83%|████████▎ | 24913/30000 [15:09:15<2:03:26, 1.46s/it] 83%|████████▎ | 24914/30000 [15:09:16<1:57:01, 1.38s/it] 83%|████████▎ | 24915/30000 [15:09:25<4:57:45, 3.51s/it] 83%|████████▎ | 24916/30000 [15:09:26<3:57:43, 2.81s/it] 83%|████████▎ | 24917/30000 [15:09:27<3:17:17, 2.33s/it] 83%|████████▎ | 24918/30000 [15:09:28<2:48:47, 1.99s/it] 83%|████████▎ | 24919/30000 [15:09:30<2:29:46, 1.77s/it] 83%|████████▎ | 24920/30000 [15:09:31<2:16:11, 1.61s/it] 83%|████████▎ | 24920/30000 [15:09:31<2:16:11, 1.61s/it] 83%|████████▎ | 24921/30000 [15:09:32<2:06:07, 1.49s/it] 83%|████████▎ | 24922/30000 [15:09:33<1:58:56, 1.41s/it] 83%|████████▎ | 24923/30000 [15:09:41<4:45:01, 3.37s/it] 83%|████████▎ | 24924/30000 [15:09:43<3:49:03, 2.71s/it] 83%|████████▎ | 24925/30000 [15:09:44<3:11:03, 2.26s/it] 83%|████████▎ | 24926/30000 [15:09:45<2:43:29, 1.93s/it] 83%|████████▎ | 24927/30000 [15:09:46<2:23:36, 1.70s/it] 83%|████████▎ | 24928/30000 [15:09:47<2:10:31, 1.54s/it] 83%|████████▎ | 24929/30000 [15:09:48<2:01:57, 1.44s/it] 83%|████████▎ | 24930/30000 [15:09:50<1:57:30, 1.39s/it] 83%|████████▎ | 24930/30000 [15:09:50<1:57:30, 1.39s/it] 83%|████████▎ | 24931/30000 [15:09:59<5:06:51, 3.63s/it] 83%|████████▎ | 24932/30000 [15:10:00<4:04:28, 2.89s/it] 83%|████████▎ | 24933/30000 [15:10:01<3:21:25, 2.39s/it] 83%|████████▎ | 24934/30000 [15:10:02<2:50:57, 2.02s/it] 83%|████████▎ | 24935/30000 [15:10:03<2:29:30, 1.77s/it] 83%|████████▎ | 24936/30000 [15:10:05<2:14:57, 1.60s/it] 83%|████████▎ | 24937/30000 [15:10:06<2:07:42, 1.51s/it] 83%|████████▎ | 24938/30000 [15:10:07<2:02:07, 1.45s/it] 83%|████████▎ | 24939/30000 [15:10:16<5:13:12, 3.71s/it] 83%|████████▎ | 24940/30000 [15:10:17<4:09:48, 2.96s/it] 83%|████████▎ | 24940/30000 [15:10:17<4:09:48, 2.96s/it] 83%|████████▎ | 24941/30000 [15:10:19<3:28:32, 2.47s/it] 83%|████████▎ | 24942/30000 [15:10:20<2:55:28, 2.08s/it] 83%|████████▎ | 24943/30000 [15:10:21<2:32:30, 1.81s/it] 83%|████████▎ | 24944/30000 [15:10:22<2:17:05, 1.63s/it] 83%|████████▎ | 24945/30000 [15:10:23<2:05:25, 1.49s/it] 83%|████████▎ | 24946/30000 [15:10:25<1:58:30, 1.41s/it] 83%|████████▎ | 24947/30000 [15:10:33<4:43:14, 3.36s/it] 83%|████████▎ | 24948/30000 [15:10:34<3:48:46, 2.72s/it] 83%|████████▎ | 24949/30000 [15:10:35<3:11:11, 2.27s/it] 83%|████████▎ | 24950/30000 [15:10:36<2:43:26, 1.94s/it] 83%|████████▎ | 24950/30000 [15:10:36<2:43:26, 1.94s/it] 83%|████████▎ | 24951/30000 [15:10:37<2:24:17, 1.71s/it] 83%|████████▎ | 24952/30000 [15:10:39<2:11:14, 1.56s/it] 83%|████████▎ | 24953/30000 [15:10:40<2:01:29, 1.44s/it] 83%|████████▎ | 24954/30000 [15:10:41<1:58:00, 1.40s/it] 83%|████████▎ | 24955/30000 [15:10:49<4:51:10, 3.46s/it] 83%|████████▎ | 24956/30000 [15:10:50<3:53:24, 2.78s/it] 83%|████████▎ | 24957/30000 [15:10:52<3:13:36, 2.30s/it] 83%|████████▎ | 24958/30000 [15:10:53<2:45:19, 1.97s/it] 83%|████████▎ | 24959/30000 [15:10:54<2:25:50, 1.74s/it] 83%|████████▎ | 24960/30000 [15:10:55<2:12:52, 1.58s/it] 83%|████████▎ | 24960/30000 [15:10:55<2:12:52, 1.58s/it] 83%|████████▎ | 24961/30000 [15:10:56<2:03:22, 1.47s/it] 83%|████████▎ | 24962/30000 [15:10:58<1:57:37, 1.40s/it] 83%|████████▎ | 24963/30000 [15:11:07<5:07:23, 3.66s/it] 83%|████████▎ | 24964/30000 [15:11:08<4:07:07, 2.94s/it] 83%|████████▎ | 24965/30000 [15:11:09<3:23:53, 2.43s/it] 83%|████████▎ | 24966/30000 [15:11:11<2:58:09, 2.12s/it] 83%|████████▎ | 24967/30000 [15:11:12<2:37:51, 1.88s/it] 83%|████████▎ | 24968/30000 [15:11:13<2:21:50, 1.69s/it] 83%|████████▎ | 24969/30000 [15:11:14<2:09:28, 1.54s/it] 83%|████████▎ | 24970/30000 [15:11:16<2:01:31, 1.45s/it] 83%|████████▎ | 24970/30000 [15:11:16<2:01:31, 1.45s/it] 83%|████████▎ | 24971/30000 [15:11:23<4:32:59, 3.26s/it] 83%|████████▎ | 24972/30000 [15:11:24<3:40:52, 2.64s/it] 83%|████████▎ | 24973/30000 [15:11:25<3:04:44, 2.21s/it] 83%|████████▎ | 24974/30000 [15:11:27<2:38:51, 1.90s/it] 83%|████████▎ | 24975/30000 [15:11:28<2:20:49, 1.68s/it] 83%|████████▎ | 24976/30000 [15:11:29<2:08:50, 1.54s/it] 83%|████████▎ | 24977/30000 [15:11:30<1:59:59, 1.43s/it] 83%|████████▎ | 24978/30000 [15:11:31<1:53:53, 1.36s/it] 83%|████████▎ | 24979/30000 [15:11:41<5:23:58, 3.87s/it] 83%|████████▎ | 24980/30000 [15:11:42<4:16:33, 3.07s/it] 83%|████████▎ | 24980/30000 [15:11:42<4:16:33, 3.07s/it] 83%|████████▎ | 24981/30000 [15:11:43<3:29:36, 2.51s/it] 83%|████████▎ | 24982/30000 [15:11:45<2:56:02, 2.10s/it] 83%|████████▎ | 24983/30000 [15:11:46<2:32:48, 1.83s/it] 83%|████████▎ | 24984/30000 [15:11:47<2:16:37, 1.63s/it] 83%|████████▎ | 24985/30000 [15:11:48<2:06:04, 1.51s/it] 83%|████████▎ | 24986/30000 [15:11:49<1:58:33, 1.42s/it] 83%|████████▎ | 24987/30000 [15:11:57<4:21:32, 3.13s/it] 83%|████████▎ | 24988/30000 [15:11:58<3:33:48, 2.56s/it] 83%|████████▎ | 24989/30000 [15:11:59<3:00:48, 2.16s/it] 83%|████████▎ | 24990/30000 [15:12:00<2:38:28, 1.90s/it] 83%|████████▎ | 24990/30000 [15:12:00<2:38:28, 1.90s/it] 83%|████████▎ | 24991/30000 [15:12:01<2:20:49, 1.69s/it] 83%|████████▎ | 24992/30000 [15:12:03<2:08:55, 1.54s/it] 83%|████████▎ | 24993/30000 [15:12:04<1:59:42, 1.43s/it] 83%|████████▎ | 24994/30000 [15:12:05<1:54:00, 1.37s/it] 83%|████████▎ | 24995/30000 [15:12:12<4:08:38, 2.98s/it] 83%|████████▎ | 24996/30000 [15:12:13<3:24:26, 2.45s/it] 83%|████████▎ | 24997/30000 [15:12:14<2:54:23, 2.09s/it] 83%|████████▎ | 24998/30000 [15:12:15<2:32:28, 1.83s/it] 83%|████████▎ | 24999/30000 [15:12:17<2:17:45, 1.65s/it] 83%|████████▎ | 25000/30000 [15:12:18<2:07:22, 1.53s/it] 83%|████████▎ | 25000/30000 [15:12:18<2:07:22, 1.53s/it] 83%|████████▎ | 25001/30000 [15:12:19<1:59:04, 1.43s/it] 83%|████████▎ | 25002/30000 [15:12:20<1:52:49, 1.35s/it] 83%|████████▎ | 25003/30000 [15:12:28<4:36:10, 3.32s/it] 83%|████████▎ | 25004/30000 [15:12:29<3:43:05, 2.68s/it] 83%|████████▎ | 25005/30000 [15:12:31<3:06:19, 2.24s/it] 83%|████████▎ | 25006/30000 [15:12:32<2:39:48, 1.92s/it] 83%|████████▎ | 25007/30000 [15:12:33<2:21:55, 1.71s/it] 83%|████████▎ | 25008/30000 [15:12:34<2:09:10, 1.55s/it] 83%|████████▎ | 25009/30000 [15:12:35<1:59:30, 1.44s/it] 83%|████████▎ | 25010/30000 [15:12:37<1:53:14, 1.36s/it] 83%|████████▎ | 25010/30000 [15:12:37<1:53:14, 1.36s/it] 83%|████████▎ | 25011/30000 [15:12:44<4:12:43, 3.04s/it] 83%|████████▎ | 25012/30000 [15:12:45<3:26:00, 2.48s/it] 83%|████████▎ | 25013/30000 [15:12:46<2:53:24, 2.09s/it] 83%|████████▎ | 25014/30000 [15:12:47<2:30:32, 1.81s/it] 83%|████████▎ | 25015/30000 [15:12:48<2:14:55, 1.62s/it] 83%|████████▎ | 25016/30000 [15:12:50<2:06:58, 1.53s/it] 83%|████████▎ | 25017/30000 [15:12:51<2:02:17, 1.47s/it] 83%|████████▎ | 25018/30000 [15:12:52<1:54:29, 1.38s/it] 83%|████████▎ | 25019/30000 [15:13:00<4:40:06, 3.37s/it] 83%|████████▎ | 25020/30000 [15:13:01<3:44:50, 2.71s/it] 83%|████████▎ | 25020/30000 [15:13:01<3:44:50, 2.71s/it] 83%|████████▎ | 25021/30000 [15:13:02<3:06:45, 2.25s/it] 83%|████████▎ | 25022/30000 [15:13:04<2:39:50, 1.93s/it] 83%|████████▎ | 25023/30000 [15:13:05<2:21:27, 1.71s/it] 83%|████████▎ | 25024/30000 [15:13:06<2:08:22, 1.55s/it] 83%|████████▎ | 25025/30000 [15:13:07<2:00:07, 1.45s/it] 83%|████████▎ | 25026/30000 [15:13:08<1:54:01, 1.38s/it] 83%|████████▎ | 25027/30000 [15:13:17<5:03:57, 3.67s/it] 83%|████████▎ | 25028/30000 [15:13:19<4:01:33, 2.92s/it] 83%|████████▎ | 25029/30000 [15:13:20<3:19:37, 2.41s/it] 83%|████████▎ | 25030/30000 [15:13:21<2:50:32, 2.06s/it] 83%|████████▎ | 25030/30000 [15:13:21<2:50:32, 2.06s/it] 83%|████████▎ | 25031/30000 [15:13:22<2:29:44, 1.81s/it] 83%|████████▎ | 25032/30000 [15:13:23<2:14:47, 1.63s/it] 83%|████████▎ | 25033/30000 [15:13:25<2:07:21, 1.54s/it] 83%|████████▎ | 25034/30000 [15:13:26<1:58:56, 1.44s/it] 83%|████████▎ | 25035/30000 [15:13:34<4:48:21, 3.48s/it] 83%|████████▎ | 25036/30000 [15:13:35<3:51:40, 2.80s/it] 83%|████████▎ | 25037/30000 [15:13:37<3:11:51, 2.32s/it] 83%|████████▎ | 25038/30000 [15:13:38<2:44:30, 1.99s/it] 83%|████████▎ | 25039/30000 [15:13:39<2:24:14, 1.74s/it] 83%|████████▎ | 25040/30000 [15:13:40<2:10:55, 1.58s/it] 83%|████████▎ | 25040/30000 [15:13:40<2:10:55, 1.58s/it] 83%|████████▎ | 25041/30000 [15:13:41<2:01:38, 1.47s/it] 83%|████████▎ | 25042/30000 [15:13:43<1:55:03, 1.39s/it] 83%|████████▎ | 25043/30000 [15:13:51<4:43:39, 3.43s/it] 83%|████████▎ | 25044/30000 [15:13:52<3:47:36, 2.76s/it] 83%|████████▎ | 25045/30000 [15:13:53<3:09:46, 2.30s/it] 83%|████████▎ | 25046/30000 [15:13:54<2:42:32, 1.97s/it] 83%|████████▎ | 25047/30000 [15:13:56<2:24:17, 1.75s/it] 83%|████████▎ | 25048/30000 [15:13:57<2:10:14, 1.58s/it] 83%|████████▎ | 25049/30000 [15:13:58<2:00:38, 1.46s/it] 84%|████████▎ | 25050/30000 [15:13:59<1:53:22, 1.37s/it] 84%|████████▎ | 25050/30000 [15:13:59<1:53:22, 1.37s/it] 84%|████████▎ | 25051/30000 [15:14:08<5:05:14, 3.70s/it] 84%|████████▎ | 25052/30000 [15:14:10<4:03:29, 2.95s/it] 84%|████████▎ | 25053/30000 [15:14:11<3:19:21, 2.42s/it] 84%|████████▎ | 25054/30000 [15:14:12<2:51:14, 2.08s/it] 84%|████████▎ | 25055/30000 [15:14:13<2:29:51, 1.82s/it] 84%|████████▎ | 25056/30000 [15:14:14<2:14:15, 1.63s/it] 84%|████████▎ | 25057/30000 [15:14:16<2:03:34, 1.50s/it] 84%|████████▎ | 25058/30000 [15:14:17<1:56:03, 1.41s/it] 84%|████████▎ | 25059/30000 [15:14:25<4:38:26, 3.38s/it] 84%|████████▎ | 25060/30000 [15:14:26<3:44:17, 2.72s/it] 84%|████████▎ | 25060/30000 [15:14:26<3:44:17, 2.72s/it] 84%|████████▎ | 25061/30000 [15:14:27<3:06:35, 2.27s/it] 84%|████████▎ | 25062/30000 [15:14:28<2:39:40, 1.94s/it] 84%|████████▎ | 25063/30000 [15:14:30<2:20:58, 1.71s/it] 84%|████████▎ | 25064/30000 [15:14:31<2:10:02, 1.58s/it] 84%|████████▎ | 25065/30000 [15:14:32<2:02:27, 1.49s/it] 84%|████████▎ | 25066/30000 [15:14:33<1:55:25, 1.40s/it] 84%|████████▎ | 25067/30000 [15:14:42<4:52:43, 3.56s/it] 84%|████████▎ | 25068/30000 [15:14:43<3:53:41, 2.84s/it] 84%|████████▎ | 25069/30000 [15:14:44<3:12:54, 2.35s/it] 84%|████████▎ | 25070/30000 [15:14:45<2:44:02, 2.00s/it] 84%|████████▎ | 25070/30000 [15:14:45<2:44:02, 2.00s/it] 84%|████████▎ | 25071/30000 [15:14:47<2:23:43, 1.75s/it] 84%|████████▎ | 25072/30000 [15:14:48<2:10:23, 1.59s/it] 84%|████████▎ | 25073/30000 [15:14:49<2:00:27, 1.47s/it] 84%|████████▎ | 25074/30000 [15:14:50<1:53:30, 1.38s/it] 84%|████████▎ | 25075/30000 [15:14:59<4:46:37, 3.49s/it] 84%|████████▎ | 25076/30000 [15:15:00<3:49:00, 2.79s/it] 84%|████████▎ | 25077/30000 [15:15:01<3:09:44, 2.31s/it] 84%|████████▎ | 25078/30000 [15:15:02<2:41:24, 1.97s/it] 84%|████████▎ | 25079/30000 [15:15:03<2:21:57, 1.73s/it] 84%|████████▎ | 25080/30000 [15:15:05<2:10:23, 1.59s/it] 84%|████████▎ | 25080/30000 [15:15:05<2:10:23, 1.59s/it] 84%|████████▎ | 25081/30000 [15:15:06<2:01:40, 1.48s/it] 84%|████████▎ | 25082/30000 [15:15:07<1:56:00, 1.42s/it] 84%|████████▎ | 25083/30000 [15:15:15<4:35:47, 3.37s/it] 84%|████████▎ | 25084/30000 [15:15:16<3:43:07, 2.72s/it] 84%|████████▎ | 25085/30000 [15:15:17<3:06:37, 2.28s/it] 84%|████████▎ | 25086/30000 [15:15:19<2:39:27, 1.95s/it] 84%|████████▎ | 25087/30000 [15:15:20<2:21:37, 1.73s/it] 84%|████████▎ | 25088/30000 [15:15:21<2:09:26, 1.58s/it] 84%|████████▎ | 25089/30000 [15:15:22<2:00:52, 1.48s/it] 84%|████████▎ | 25090/30000 [15:15:23<1:53:50, 1.39s/it] 84%|████████▎ | 25090/30000 [15:15:23<1:53:50, 1.39s/it] 84%|████████▎ | 25091/30000 [15:15:31<4:32:24, 3.33s/it] 84%|████████▎ | 25092/30000 [15:15:33<3:39:57, 2.69s/it] 84%|████████▎ | 25093/30000 [15:15:34<3:03:04, 2.24s/it] 84%|████████▎ | 25094/30000 [15:15:35<2:37:13, 1.92s/it] 84%|████████▎ | 25095/30000 [15:15:36<2:18:46, 1.70s/it] 84%|████████▎ | 25096/30000 [15:15:37<2:06:28, 1.55s/it] 84%|████████▎ | 25097/30000 [15:15:38<1:58:02, 1.44s/it] 84%|████████▎ | 25098/30000 [15:15:40<1:51:38, 1.37s/it] 84%|████████▎ | 25099/30000 [15:15:47<4:22:16, 3.21s/it] 84%|████████▎ | 25100/30000 [15:15:48<3:32:49, 2.61s/it] 84%|████████▎ | 25100/30000 [15:15:48<3:32:49, 2.61s/it] 84%|████████▎ | 25101/30000 [15:15:50<2:58:16, 2.18s/it] 84%|████████▎ | 25102/30000 [15:15:51<2:33:26, 1.88s/it] 84%|████████▎ | 25103/30000 [15:15:52<2:15:36, 1.66s/it] 84%|████████▎ | 25104/30000 [15:15:53<2:03:46, 1.52s/it] 84%|████████▎ | 25105/30000 [15:15:54<1:55:35, 1.42s/it] 84%|████████▎ | 25106/30000 [15:15:55<1:50:17, 1.35s/it] 84%|████████▎ | 25107/30000 [15:16:04<4:38:11, 3.41s/it] 84%|████████▎ | 25108/30000 [15:16:05<3:43:32, 2.74s/it] 84%|████████▎ | 25109/30000 [15:16:06<3:05:56, 2.28s/it] 84%|████████▎ | 25110/30000 [15:16:07<2:38:45, 1.95s/it] 84%|████████▎ | 25110/30000 [15:16:07<2:38:45, 1.95s/it] 84%|████████▎ | 25111/30000 [15:16:08<2:20:06, 1.72s/it] 84%|████████▎ | 25112/30000 [15:16:10<2:09:18, 1.59s/it] 84%|████████▎ | 25113/30000 [15:16:11<2:00:54, 1.48s/it] 84%|████████▎ | 25114/30000 [15:16:12<1:55:01, 1.41s/it] 84%|████████▎ | 25115/30000 [15:16:20<4:36:42, 3.40s/it] 84%|████████▎ | 25116/30000 [15:16:21<3:42:42, 2.74s/it] 84%|████████▎ | 25117/30000 [15:16:23<3:05:23, 2.28s/it] 84%|████████▎ | 25118/30000 [15:16:24<2:38:57, 1.95s/it] 84%|████████▎ | 25119/30000 [15:16:25<2:19:49, 1.72s/it] 84%|████████▎ | 25120/30000 [15:16:26<2:11:27, 1.62s/it] 84%|████████▎ | 25120/30000 [15:16:26<2:11:27, 1.62s/it] 84%|████████▎ | 25121/30000 [15:16:28<2:03:29, 1.52s/it] 84%|████████▎ | 25122/30000 [15:16:29<1:57:05, 1.44s/it] 84%|████████▎ | 25123/30000 [15:16:37<4:32:55, 3.36s/it] 84%|████████▎ | 25124/30000 [15:16:38<3:39:26, 2.70s/it] 84%|████████▍ | 25125/30000 [15:16:39<3:02:34, 2.25s/it] 84%|████████▍ | 25126/30000 [15:16:40<2:39:19, 1.96s/it] 84%|████████▍ | 25127/30000 [15:16:42<2:21:20, 1.74s/it] 84%|████████▍ | 25128/30000 [15:16:43<2:08:33, 1.58s/it] 84%|████████▍ | 25129/30000 [15:16:44<1:59:21, 1.47s/it] 84%|████████▍ | 25130/30000 [15:16:45<1:52:58, 1.39s/it] 84%|████████▍ | 25130/30000 [15:16:45<1:52:58, 1.39s/it] 84%|████████▍ | 25131/30000 [15:16:54<4:54:42, 3.63s/it] 84%|████████▍ | 25132/30000 [15:16:55<3:54:17, 2.89s/it] 84%|████████▍ | 25133/30000 [15:16:56<3:12:22, 2.37s/it] 84%|████████▍ | 25134/30000 [15:16:58<2:43:37, 2.02s/it] 84%|████████▍ | 25135/30000 [15:16:59<2:22:57, 1.76s/it] 84%|████████▍ | 25136/30000 [15:17:00<2:09:25, 1.60s/it] 84%|████████▍ | 25137/30000 [15:17:01<2:00:09, 1.48s/it] 84%|████████▍ | 25138/30000 [15:17:02<1:53:23, 1.40s/it] 84%|████████▍ | 25139/30000 [15:17:11<4:40:16, 3.46s/it] 84%|████████▍ | 25140/30000 [15:17:12<3:44:57, 2.78s/it] 84%|████████▍ | 25140/30000 [15:17:12<3:44:57, 2.78s/it] 84%|████████▍ | 25141/30000 [15:17:13<3:07:01, 2.31s/it] 84%|████████▍ | 25142/30000 [15:17:14<2:40:14, 1.98s/it] 84%|████████▍ | 25143/30000 [15:17:15<2:21:21, 1.75s/it] 84%|████████▍ | 25144/30000 [15:17:17<2:08:12, 1.58s/it] 84%|████████▍ | 25145/30000 [15:17:18<1:58:40, 1.47s/it] 84%|████████▍ | 25146/30000 [15:17:19<1:52:15, 1.39s/it] 84%|████████▍ | 25147/30000 [15:17:28<4:49:05, 3.57s/it] 84%|████████▍ | 25148/30000 [15:17:29<3:50:51, 2.85s/it] 84%|████████▍ | 25149/30000 [15:17:30<3:11:16, 2.37s/it] 84%|████████▍ | 25150/30000 [15:17:31<2:42:34, 2.01s/it] 84%|████████▍ | 25150/30000 [15:17:31<2:42:34, 2.01s/it] 84%|████████▍ | 25151/30000 [15:17:33<2:22:23, 1.76s/it] 84%|████████▍ | 25152/30000 [15:17:34<2:07:55, 1.58s/it] 84%|████████▍ | 25153/30000 [15:17:35<1:58:17, 1.46s/it] 84%|████████▍ | 25154/30000 [15:17:36<1:51:52, 1.39s/it] 84%|████████▍ | 25155/30000 [15:17:43<4:13:38, 3.14s/it] 84%|████████▍ | 25156/30000 [15:17:45<3:27:46, 2.57s/it] 84%|████████▍ | 25157/30000 [15:17:46<2:54:26, 2.16s/it] 84%|████████▍ | 25158/30000 [15:17:47<2:30:28, 1.86s/it] 84%|████████▍ | 25159/30000 [15:17:48<2:14:20, 1.66s/it] 84%|████████▍ | 25160/30000 [15:17:49<2:02:40, 1.52s/it] 84%|████████▍ | 25160/30000 [15:17:49<2:02:40, 1.52s/it] 84%|████████▍ | 25161/30000 [15:17:51<1:55:57, 1.44s/it] 84%|████████▍ | 25162/30000 [15:17:52<1:50:21, 1.37s/it] 84%|████████▍ | 25163/30000 [15:18:01<4:49:48, 3.59s/it] 84%|████████▍ | 25164/30000 [15:18:02<3:50:53, 2.86s/it] 84%|████████▍ | 25165/30000 [15:18:03<3:11:13, 2.37s/it] 84%|████████▍ | 25166/30000 [15:18:04<2:42:18, 2.01s/it] 84%|████████▍ | 25167/30000 [15:18:05<2:22:05, 1.76s/it] 84%|████████▍ | 25168/30000 [15:18:07<2:09:48, 1.61s/it] 84%|████████▍ | 25169/30000 [15:18:08<2:00:32, 1.50s/it] 84%|████████▍ | 25170/30000 [15:18:09<1:53:15, 1.41s/it] 84%|████████▍ | 25170/30000 [15:18:09<1:53:15, 1.41s/it] 84%|████████▍ | 25171/30000 [15:18:17<4:36:23, 3.43s/it] 84%|████████▍ | 25172/30000 [15:18:18<3:42:15, 2.76s/it] 84%|████████▍ | 25173/30000 [15:18:20<3:04:48, 2.30s/it] 84%|████████▍ | 25174/30000 [15:18:21<2:38:36, 1.97s/it] 84%|████████▍ | 25175/30000 [15:18:22<2:20:45, 1.75s/it] 84%|████████▍ | 25176/30000 [15:18:23<2:07:27, 1.59s/it] 84%|████████▍ | 25177/30000 [15:18:24<1:57:56, 1.47s/it] 84%|████████▍ | 25178/30000 [15:18:26<1:51:48, 1.39s/it] 84%|████████▍ | 25179/30000 [15:18:34<4:34:31, 3.42s/it] 84%|████████▍ | 25180/30000 [15:18:35<3:40:12, 2.74s/it] 84%|████████▍ | 25180/30000 [15:18:35<3:40:12, 2.74s/it] 84%|████████▍ | 25181/30000 [15:18:36<3:03:26, 2.28s/it] 84%|████████▍ | 25182/30000 [15:18:37<2:37:52, 1.97s/it] 84%|████████▍ | 25183/30000 [15:18:39<2:19:19, 1.74s/it] 84%|████████▍ | 25184/30000 [15:18:40<2:06:43, 1.58s/it] 84%|████████▍ | 25185/30000 [15:18:41<1:59:09, 1.48s/it] 84%|████████▍ | 25186/30000 [15:18:42<1:52:26, 1.40s/it] 84%|████████▍ | 25187/30000 [15:18:50<4:26:50, 3.33s/it] 84%|████████▍ | 25188/30000 [15:18:51<3:36:52, 2.70s/it] 84%|████████▍ | 25189/30000 [15:18:53<3:00:43, 2.25s/it] 84%|████████▍ | 25190/30000 [15:18:54<2:34:43, 1.93s/it] 84%|████████▍ | 25190/30000 [15:18:54<2:34:43, 1.93s/it] 84%|████████▍ | 25191/30000 [15:18:55<2:16:55, 1.71s/it] 84%|████████▍ | 25192/30000 [15:18:56<2:05:46, 1.57s/it] 84%|████████▍ | 25193/30000 [15:18:57<1:58:42, 1.48s/it] 84%|████████▍ | 25194/30000 [15:18:59<1:52:49, 1.41s/it] 84%|████████▍ | 25195/30000 [15:19:07<4:29:50, 3.37s/it] 84%|████████▍ | 25196/30000 [15:19:08<3:37:33, 2.72s/it] 84%|████████▍ | 25197/30000 [15:19:09<3:00:55, 2.26s/it] 84%|████████▍ | 25198/30000 [15:19:10<2:37:16, 1.97s/it] 84%|████████▍ | 25199/30000 [15:19:11<2:18:30, 1.73s/it] 84%|████████▍ | 25200/30000 [15:19:13<2:08:01, 1.60s/it] 84%|████████▍ | 25200/30000 [15:19:13<2:08:01, 1.60s/it] 84%|████████▍ | 25201/30000 [15:19:14<1:58:29, 1.48s/it] 84%|████████▍ | 25202/30000 [15:19:15<1:51:46, 1.40s/it] 84%|████████▍ | 25203/30000 [15:19:23<4:38:17, 3.48s/it] 84%|████████▍ | 25204/30000 [15:19:25<3:44:00, 2.80s/it] 84%|████████▍ | 25205/30000 [15:19:26<3:06:09, 2.33s/it] 84%|████████▍ | 25206/30000 [15:19:27<2:39:00, 1.99s/it] 84%|████████▍ | 25207/30000 [15:19:28<2:19:54, 1.75s/it] 84%|████████▍ | 25208/30000 [15:19:30<2:07:27, 1.60s/it] 84%|████████▍ | 25209/30000 [15:19:31<1:58:31, 1.48s/it] 84%|████████▍ | 25210/30000 [15:19:32<1:51:49, 1.40s/it] 84%|████████▍ | 25210/30000 [15:19:32<1:51:49, 1.40s/it] 84%|████████▍ | 25211/30000 [15:19:42<5:13:50, 3.93s/it] 84%|████████▍ | 25212/30000 [15:19:43<4:08:39, 3.12s/it] 84%|████████▍ | 25213/30000 [15:19:44<3:22:30, 2.54s/it] 84%|████████▍ | 25214/30000 [15:19:45<2:49:39, 2.13s/it] 84%|████████▍ | 25215/30000 [15:19:47<2:26:48, 1.84s/it] 84%|████████▍ | 25216/30000 [15:19:48<2:11:47, 1.65s/it] 84%|████████▍ | 25217/30000 [15:19:49<2:00:55, 1.52s/it] 84%|████████▍ | 25218/30000 [15:19:50<1:54:03, 1.43s/it] 84%|████████▍ | 25219/30000 [15:19:59<4:53:33, 3.68s/it] 84%|████████▍ | 25220/30000 [15:20:00<3:53:30, 2.93s/it] 84%|████████▍ | 25220/30000 [15:20:00<3:53:30, 2.93s/it] 84%|████████▍ | 25221/30000 [15:20:02<3:12:25, 2.42s/it] 84%|████████▍ | 25222/30000 [15:20:03<2:46:29, 2.09s/it] 84%|████████▍ | 25223/30000 [15:20:04<2:24:02, 1.81s/it] 84%|████████▍ | 25224/30000 [15:20:05<2:11:52, 1.66s/it] 84%|████████▍ | 25225/30000 [15:20:07<2:01:42, 1.53s/it] 84%|████████▍ | 25226/30000 [15:20:08<1:55:52, 1.46s/it] 84%|████████▍ | 25227/30000 [15:20:15<4:19:06, 3.26s/it] 84%|████████▍ | 25228/30000 [15:20:16<3:29:41, 2.64s/it] 84%|████████▍ | 25229/30000 [15:20:18<2:54:35, 2.20s/it] 84%|████████▍ | 25230/30000 [15:20:19<2:30:27, 1.89s/it] 84%|████████▍ | 25230/30000 [15:20:19<2:30:27, 1.89s/it] 84%|████████▍ | 25231/30000 [15:20:20<2:13:27, 1.68s/it] 84%|████████▍ | 25232/30000 [15:20:21<2:01:42, 1.53s/it] 84%|████████▍ | 25233/30000 [15:20:22<1:54:33, 1.44s/it] 84%|████████▍ | 25234/30000 [15:20:24<1:48:50, 1.37s/it] 84%|████████▍ | 25235/30000 [15:20:31<4:09:49, 3.15s/it] 84%|████████▍ | 25236/30000 [15:20:32<3:22:47, 2.55s/it] 84%|████████▍ | 25237/30000 [15:20:33<2:50:28, 2.15s/it] 84%|████████▍ | 25238/30000 [15:20:34<2:27:34, 1.86s/it] 84%|████████▍ | 25239/30000 [15:20:36<2:11:22, 1.66s/it] 84%|████████▍ | 25240/30000 [15:20:37<1:59:59, 1.51s/it] 84%|████████▍ | 25240/30000 [15:20:37<1:59:59, 1.51s/it] 84%|████████▍ | 25241/30000 [15:20:38<1:52:02, 1.41s/it] 84%|████████▍ | 25242/30000 [15:20:39<1:50:36, 1.39s/it] 84%|████████▍ | 25243/30000 [15:20:47<4:14:55, 3.22s/it] 84%|████████▍ | 25244/30000 [15:20:48<3:26:26, 2.60s/it] 84%|████████▍ | 25245/30000 [15:20:49<2:52:55, 2.18s/it] 84%|████████▍ | 25246/30000 [15:20:50<2:29:14, 1.88s/it] 84%|████████▍ | 25247/30000 [15:20:52<2:12:38, 1.67s/it] 84%|████████▍ | 25248/30000 [15:20:53<2:01:29, 1.53s/it] 84%|████████▍ | 25249/30000 [15:20:54<1:53:41, 1.44s/it] 84%|████████▍ | 25250/30000 [15:20:55<1:48:26, 1.37s/it] 84%|████████▍ | 25250/30000 [15:20:55<1:48:26, 1.37s/it] 84%|████████▍ | 25251/30000 [15:21:03<4:10:54, 3.17s/it] 84%|████████▍ | 25252/30000 [15:21:04<3:23:31, 2.57s/it] 84%|████████▍ | 25253/30000 [15:21:05<2:50:32, 2.16s/it] 84%|████████▍ | 25254/30000 [15:21:06<2:27:24, 1.86s/it] 84%|████████▍ | 25255/30000 [15:21:07<2:10:56, 1.66s/it] 84%|████████▍ | 25256/30000 [15:21:09<1:59:50, 1.52s/it] 84%|████████▍ | 25257/30000 [15:21:10<1:51:59, 1.42s/it] 84%|████████▍ | 25258/30000 [15:21:11<1:50:55, 1.40s/it] 84%|████████▍ | 25259/30000 [15:21:19<4:32:17, 3.45s/it] 84%|████████▍ | 25260/30000 [15:21:20<3:38:25, 2.76s/it] +{'loss': 0.0119, 'grad_norm': 0.06097510829567909, 'learning_rate': 2.344836133232215e-06, 'flow_matching_loss': 0.011923977453261615, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.23} +{'loss': 0.0138, 'grad_norm': 0.06924326717853546, 'learning_rate': 2.335967167846342e-06, 'flow_matching_loss': 0.013761421758681535, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.24} +{'loss': 0.0127, 'grad_norm': 0.07396054267883301, 'learning_rate': 2.3271135904544993e-06, 'flow_matching_loss': 0.012687346246093512, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.24} +{'loss': 0.012, 'grad_norm': 0.07609918713569641, 'learning_rate': 2.3182754118146206e-06, 'flow_matching_loss': 0.012041572760790587, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.24} +{'loss': 0.0129, 'grad_norm': 0.066389299929142, 'learning_rate': 2.309452642665923e-06, 'flow_matching_loss': 0.012917874939739704, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.24} +{'loss': 0.0135, 'grad_norm': 0.09029030054807663, 'learning_rate': 2.3006452937289065e-06, 'flow_matching_loss': 0.013467625342309475, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.25} +{'loss': 0.0132, 'grad_norm': 0.06256437301635742, 'learning_rate': 2.291853375705327e-06, 'flow_matching_loss': 0.013243778143078089, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.25} +{'loss': 0.0133, 'grad_norm': 0.06800330430269241, 'learning_rate': 2.283076899278197e-06, 'flow_matching_loss': 0.013294597156345844, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.25} +{'loss': 0.0133, 'grad_norm': 0.06033209338784218, 'learning_rate': 2.274315875111763e-06, 'flow_matching_loss': 0.013315598294138909, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.26} +{'loss': 0.013, 'grad_norm': 0.06581635028123856, 'learning_rate': 2.265570313851493e-06, 'flow_matching_loss': 0.013028513081371785, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.26} +{'loss': 0.0131, 'grad_norm': 0.07426039129495621, 'learning_rate': 2.2568402261240685e-06, 'flow_matching_loss': 0.013104200642555953, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.26} +{'loss': 0.0131, 'grad_norm': 0.0683746412396431, 'learning_rate': 2.248125622537373e-06, 'flow_matching_loss': 0.013127768971025943, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.27} +{'loss': 0.0157, 'grad_norm': 0.09765830636024475, 'learning_rate': 2.239426513680467e-06, 'flow_matching_loss': 0.015658169891685246, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.27} +{'loss': 0.0114, 'grad_norm': 0.07160168886184692, 'learning_rate': 2.2307429101235915e-06, 'flow_matching_loss': 0.0113798966165632, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.27} +{'loss': 0.0138, 'grad_norm': 0.08232659846544266, 'learning_rate': 2.222074822418145e-06, 'flow_matching_loss': 0.01377755831927061, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.27} +{'loss': 0.0129, 'grad_norm': 0.06931172311306, 'learning_rate': 2.213422261096668e-06, 'flow_matching_loss': 0.012920684739947319, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.28} +{'loss': 0.0125, 'grad_norm': 0.07889042049646378, 'learning_rate': 2.204785236672835e-06, 'flow_matching_loss': 0.01247064326889813, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.28} +{'loss': 0.0129, 'grad_norm': 0.07014771550893784, 'learning_rate': 2.1961637596414536e-06, 'flow_matching_loss': 0.012855493929237127, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.28} +{'loss': 0.0145, 'grad_norm': 0.08160828799009323, 'learning_rate': 2.1875578404784275e-06, 'flow_matching_loss': 0.014467110205441714, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.29} +{'loss': 0.0127, 'grad_norm': 0.058091193437576294, 'learning_rate': 2.1789674896407583e-06, 'flow_matching_loss': 0.012672700826078654, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.29} +{'loss': 0.0128, 'grad_norm': 0.06610129773616791, 'learning_rate': 2.170392717566534e-06, 'flow_matching_loss': 0.012809596210718154, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.29} +{'loss': 0.0134, 'grad_norm': 0.054400645196437836, 'learning_rate': 2.161833534674909e-06, 'flow_matching_loss': 0.0133501716889441, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.29} +{'loss': 0.0137, 'grad_norm': 0.053762346506118774, 'learning_rate': 2.1532899513661015e-06, 'flow_matching_loss': 0.013675120938569308, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.3} +{'loss': 0.013, 'grad_norm': 0.08040399849414825, 'learning_rate': 2.1447619780213682e-06, 'flow_matching_loss': 0.012978781666606664, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.3} +{'loss': 0.0125, 'grad_norm': 0.055258795619010925, 'learning_rate': 2.136249625003001e-06, 'flow_matching_loss': 0.012538180407136678, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.3} +{'loss': 0.0131, 'grad_norm': 0.057001613080501556, 'learning_rate': 2.1277529026543107e-06, 'flow_matching_loss': 0.013060044962912798, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.31} +{'loss': 0.0137, 'grad_norm': 0.058778002858161926, 'learning_rate': 2.1192718212996188e-06, 'flow_matching_loss': 0.013655280694365501, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.31} +{'loss': 0.0105, 'grad_norm': 0.05762752890586853, 'learning_rate': 2.110806391244235e-06, 'flow_matching_loss': 0.010538684530183673, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.31} +{'loss': 0.0136, 'grad_norm': 0.07295507192611694, 'learning_rate': 2.1023566227744596e-06, 'flow_matching_loss': 0.013550690282136202, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.31} +{'loss': 0.0138, 'grad_norm': 0.07096055150032043, 'learning_rate': 2.0939225261575563e-06, 'flow_matching_loss': 0.013786217477172614, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.32} +{'loss': 0.0124, 'grad_norm': 0.095338836312294, 'learning_rate': 2.0855041116417473e-06, 'flow_matching_loss': 0.012375852558761835, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.32} +{'loss': 0.0123, 'grad_norm': 0.09436263144016266, 'learning_rate': 2.0771013894562003e-06, 'flow_matching_loss': 0.012291854154318571, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.32} +{'loss': 0.0127, 'grad_norm': 0.06214461475610733, 'learning_rate': 2.0687143698110113e-06, 'flow_matching_loss': 0.0126788261346519, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.33} +{'loss': 0.0133, 'grad_norm': 0.1134079322218895, 'learning_rate': 2.060343062897208e-06, 'flow_matching_loss': 0.013309497106820344, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.33} +{'loss': 0.0127, 'grad_norm': 0.06954015791416168, 'learning_rate': 2.0519874788867127e-06, 'flow_matching_loss': 0.012689704913645983, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.33} +{'loss': 0.0141, 'grad_norm': 0.0855197012424469, 'learning_rate': 2.043647627932347e-06, 'flow_matching_loss': 0.014092088025063277, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.33} +{'loss': 0.0131, 'grad_norm': 0.06703419238328934, 'learning_rate': 2.0353235201678193e-06, 'flow_matching_loss': 0.013058912847191096, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.34} +{'loss': 0.0123, 'grad_norm': 0.07556241005659103, 'learning_rate': 2.0270151657076995e-06, 'flow_matching_loss': 0.012271210271865129, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.34} +{'loss': 0.0146, 'grad_norm': 0.0758795291185379, 'learning_rate': 2.0187225746474286e-06, 'flow_matching_loss': 0.014575333055108785, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.34} +{'loss': 0.0126, 'grad_norm': 0.06054588779807091, 'learning_rate': 2.0104457570632845e-06, 'flow_matching_loss': 0.012605851236730813, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.35} +{'loss': 0.0125, 'grad_norm': 0.0611952468752861, 'learning_rate': 2.0021847230123775e-06, 'flow_matching_loss': 0.012461605109274387, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.35} 84%|████████▍ | 25260/30000 [15:21:20<3:38:25, 2.76s/it] 84%|████████▍ | 25261/30000 [15:21:22<3:00:52, 2.29s/it] 84%|████████▍ | 25262/30000 [15:21:23<2:34:43, 1.96s/it] 84%|████████▍ | 25263/30000 [15:21:24<2:15:59, 1.72s/it] 84%|████████▍ | 25264/30000 [15:21:25<2:03:43, 1.57s/it] 84%|████████▍ | 25265/30000 [15:21:26<1:55:03, 1.46s/it] 84%|████████▍ | 25266/30000 [15:21:28<1:48:47, 1.38s/it] 84%|████████▍ | 25267/30000 [15:21:36<4:41:02, 3.56s/it] 84%|████████▍ | 25268/30000 [15:21:37<3:44:12, 2.84s/it] 84%|████████▍ | 25269/30000 [15:21:39<3:04:35, 2.34s/it] 84%|████████▍ | 25270/30000 [15:21:40<2:37:14, 1.99s/it] 84%|████████▍ | 25270/30000 [15:21:40<2:37:14, 1.99s/it] 84%|████████▍ | 25271/30000 [15:21:41<2:18:38, 1.76s/it] 84%|████████▍ | 25272/30000 [15:21:42<2:07:43, 1.62s/it] 84%|████████▍ | 25273/30000 [15:21:43<1:58:08, 1.50s/it] 84%|████████▍ | 25274/30000 [15:21:45<1:52:43, 1.43s/it] 84%|████████▍ | 25275/30000 [15:21:52<4:14:12, 3.23s/it] 84%|████████▍ | 25276/30000 [15:21:53<3:25:51, 2.61s/it] 84%|████████▍ | 25277/30000 [15:21:55<2:51:54, 2.18s/it] 84%|████████▍ | 25278/30000 [15:21:56<2:28:11, 1.88s/it] 84%|████████▍ | 25279/30000 [15:21:57<2:11:05, 1.67s/it] 84%|████████▍ | 25280/30000 [15:21:58<2:00:34, 1.53s/it] 84%|████████▍ | 25280/30000 [15:21:58<2:00:34, 1.53s/it] 84%|████████▍ | 25281/30000 [15:21:59<1:53:01, 1.44s/it] 84%|████████▍ | 25282/30000 [15:22:01<1:47:27, 1.37s/it] 84%|████████▍ | 25283/30000 [15:22:08<4:18:18, 3.29s/it] 84%|████████▍ | 25284/30000 [15:22:10<3:30:00, 2.67s/it] 84%|████████▍ | 25285/30000 [15:22:11<2:56:03, 2.24s/it] 84%|████████▍ | 25286/30000 [15:22:12<2:30:43, 1.92s/it] 84%|████████▍ | 25287/30000 [15:22:13<2:13:06, 1.69s/it] 84%|████████▍ | 25288/30000 [15:22:14<2:01:15, 1.54s/it] 84%|████████▍ | 25289/30000 [15:22:16<1:53:44, 1.45s/it] 84%|████████▍ | 25290/30000 [15:22:17<1:47:52, 1.37s/it] 84%|████████▍ | 25290/30000 [15:22:17<1:47:52, 1.37s/it] 84%|████████▍ | 25291/30000 [15:22:24<4:11:21, 3.20s/it] 84%|████████▍ | 25292/30000 [15:22:25<3:23:05, 2.59s/it] 84%|████████▍ | 25293/30000 [15:22:27<2:50:00, 2.17s/it] 84%|████████▍ | 25294/30000 [15:22:28<2:26:49, 1.87s/it] 84%|████████▍ | 25295/30000 [15:22:29<2:10:47, 1.67s/it] 84%|████████▍ | 25296/30000 [15:22:30<2:00:42, 1.54s/it] 84%|████████▍ | 25297/30000 [15:22:31<1:53:14, 1.44s/it] 84%|████████▍ | 25298/30000 [15:22:33<1:47:17, 1.37s/it] 84%|████████▍ | 25299/30000 [15:22:41<4:21:48, 3.34s/it] 84%|████████▍ | 25300/30000 [15:22:42<3:31:12, 2.70s/it] 84%|████████▍ | 25300/30000 [15:22:42<3:31:12, 2.70s/it] 84%|████████▍ | 25301/30000 [15:22:43<2:54:57, 2.23s/it] 84%|████████▍ | 25302/30000 [15:22:44<2:29:46, 1.91s/it] 84%|████████▍ | 25303/30000 [15:22:45<2:12:56, 1.70s/it] 84%|████████▍ | 25304/30000 [15:22:46<2:00:44, 1.54s/it] 84%|████████▍ | 25305/30000 [15:22:48<1:53:07, 1.45s/it] 84%|████████▍ | 25306/30000 [15:22:49<1:47:10, 1.37s/it] 84%|████████▍ | 25307/30000 [15:22:57<4:34:31, 3.51s/it] 84%|████████▍ | 25308/30000 [15:22:59<3:40:26, 2.82s/it] 84%|████████▍ | 25309/30000 [15:23:00<3:02:31, 2.33s/it] 84%|████████▍ | 25310/30000 [15:23:01<2:36:04, 2.00s/it] 84%|████████▍ | 25310/30000 [15:23:01<2:36:04, 2.00s/it] 84%|████████▍ | 25311/30000 [15:23:02<2:16:46, 1.75s/it] 84%|████████▍ | 25312/30000 [15:23:03<2:03:23, 1.58s/it] 84%|████████▍ | 25313/30000 [15:23:05<1:57:58, 1.51s/it] 84%|████████▍ | 25314/30000 [15:23:06<1:49:51, 1.41s/it] 84%|████████▍ | 25315/30000 [15:23:14<4:27:17, 3.42s/it] 84%|████████▍ | 25316/30000 [15:23:15<3:35:11, 2.76s/it] 84%|████████▍ | 25317/30000 [15:23:16<2:58:06, 2.28s/it] 84%|████████▍ | 25318/30000 [15:23:17<2:31:58, 1.95s/it] 84%|████████▍ | 25319/30000 [15:23:19<2:14:24, 1.72s/it] 84%|████████▍ | 25320/30000 [15:23:20<2:02:53, 1.58s/it] 84%|████████▍ | 25320/30000 [15:23:20<2:02:53, 1.58s/it] 84%|████████▍ | 25321/30000 [15:23:21<1:54:20, 1.47s/it] 84%|████████▍ | 25322/30000 [15:23:22<1:49:47, 1.41s/it] 84%|████████▍ | 25323/30000 [15:23:31<4:49:54, 3.72s/it] 84%|████████▍ | 25324/30000 [15:23:33<3:52:52, 2.99s/it] 84%|████████▍ | 25325/30000 [15:23:34<3:13:53, 2.49s/it] 84%|████████▍ | 25326/30000 [15:23:35<2:43:10, 2.09s/it] 84%|████████▍ | 25327/30000 [15:23:36<2:21:42, 1.82s/it] 84%|████████▍ | 25328/30000 [15:23:38<2:07:21, 1.64s/it] 84%|████████▍ | 25329/30000 [15:23:39<1:57:06, 1.50s/it] 84%|████████▍ | 25330/30000 [15:23:40<1:52:42, 1.45s/it] 84%|████████▍ | 25330/30000 [15:23:40<1:52:42, 1.45s/it] 84%|████████▍ | 25331/30000 [15:23:49<4:50:34, 3.73s/it] 84%|████████▍ | 25332/30000 [15:23:50<3:50:55, 2.97s/it] 84%|████████▍ | 25333/30000 [15:23:52<3:10:09, 2.44s/it] 84%|████████▍ | 25334/30000 [15:23:53<2:40:50, 2.07s/it] 84%|████████▍ | 25335/30000 [15:23:54<2:21:16, 1.82s/it] 84%|████████▍ | 25336/30000 [15:23:55<2:06:15, 1.62s/it] 84%|████████▍ | 25337/30000 [15:23:56<1:55:58, 1.49s/it] 84%|████████▍ | 25338/30000 [15:23:58<1:49:34, 1.41s/it] 84%|████████▍ | 25339/30000 [15:24:06<4:28:17, 3.45s/it] 84%|████████▍ | 25340/30000 [15:24:07<3:35:18, 2.77s/it] 84%|████████▍ | 25340/30000 [15:24:07<3:35:18, 2.77s/it] 84%|████████▍ | 25341/30000 [15:24:08<2:59:01, 2.31s/it] 84%|████████▍ | 25342/30000 [15:24:09<2:32:40, 1.97s/it] 84%|████████▍ | 25343/30000 [15:24:11<2:15:07, 1.74s/it] 84%|████████▍ | 25344/30000 [15:24:12<2:02:52, 1.58s/it] 84%|████████▍ | 25345/30000 [15:24:13<1:54:34, 1.48s/it] 84%|████████▍ | 25346/30000 [15:24:14<1:48:16, 1.40s/it] 84%|████████▍ | 25347/30000 [15:24:23<4:28:25, 3.46s/it] 84%|████████▍ | 25348/30000 [15:24:24<3:37:42, 2.81s/it] 84%|████████▍ | 25349/30000 [15:24:25<3:04:42, 2.38s/it] 84%|████████▍ | 25350/30000 [15:24:26<2:37:08, 2.03s/it] 84%|████████▍ | 25350/30000 [15:24:26<2:37:08, 2.03s/it] 85%|████████▍ | 25351/30000 [15:24:28<2:17:14, 1.77s/it] 85%|████████▍ | 25352/30000 [15:24:29<2:03:56, 1.60s/it] 85%|████████▍ | 25353/30000 [15:24:30<1:54:31, 1.48s/it] 85%|████████▍ | 25354/30000 [15:24:31<1:48:41, 1.40s/it] 85%|████████▍ | 25355/30000 [15:24:39<4:11:50, 3.25s/it] 85%|████████▍ | 25356/30000 [15:24:40<3:24:53, 2.65s/it] 85%|████████▍ | 25357/30000 [15:24:41<2:51:24, 2.21s/it] 85%|████████▍ | 25358/30000 [15:24:43<2:29:52, 1.94s/it] 85%|████████▍ | 25359/30000 [15:24:44<2:12:02, 1.71s/it] 85%|████████▍ | 25360/30000 [15:24:45<1:59:29, 1.55s/it] 85%|████████▍ | 25360/30000 [15:24:45<1:59:29, 1.55s/it] 85%|████████▍ | 25361/30000 [15:24:46<1:52:02, 1.45s/it] 85%|████████▍ | 25362/30000 [15:24:47<1:45:52, 1.37s/it] 85%|████████▍ | 25363/30000 [15:24:56<4:30:40, 3.50s/it] 85%|████████▍ | 25364/30000 [15:24:57<3:36:46, 2.81s/it] 85%|████████▍ | 25365/30000 [15:24:58<2:59:13, 2.32s/it] 85%|████████▍ | 25366/30000 [15:24:59<2:33:06, 1.98s/it] 85%|████████▍ | 25367/30000 [15:25:01<2:14:33, 1.74s/it] 85%|████████▍ | 25368/30000 [15:25:02<2:02:06, 1.58s/it] 85%|████████▍ | 25369/30000 [15:25:03<1:53:23, 1.47s/it] 85%|████████▍ | 25370/30000 [15:25:04<1:46:48, 1.38s/it] 85%|████████▍ | 25370/30000 [15:25:04<1:46:48, 1.38s/it] 85%|████████▍ | 25371/30000 [15:25:13<4:43:35, 3.68s/it] 85%|████████▍ | 25372/30000 [15:25:14<3:47:22, 2.95s/it] 85%|████████▍ | 25373/30000 [15:25:16<3:06:39, 2.42s/it] 85%|████████▍ | 25374/30000 [15:25:17<2:40:20, 2.08s/it] 85%|████████▍ | 25375/30000 [15:25:18<2:20:53, 1.83s/it] 85%|████████▍ | 25376/30000 [15:25:19<2:06:04, 1.64s/it] 85%|████████▍ | 25377/30000 [15:25:20<1:55:40, 1.50s/it] 85%|████████▍ | 25378/30000 [15:25:22<1:48:31, 1.41s/it] 85%|████████▍ | 25379/30000 [15:25:31<4:45:36, 3.71s/it] 85%|████████▍ | 25380/30000 [15:25:32<3:52:23, 3.02s/it] 85%|████████▍ | 25380/30000 [15:25:32<3:52:23, 3.02s/it] 85%|████████▍ | 25381/30000 [15:25:34<3:13:36, 2.51s/it] 85%|████████▍ | 25382/30000 [15:25:35<2:42:12, 2.11s/it] 85%|████████▍ | 25383/30000 [15:25:36<2:21:06, 1.83s/it] 85%|████████▍ | 25384/30000 [15:25:37<2:06:35, 1.65s/it] 85%|████████▍ | 25385/30000 [15:25:38<1:56:39, 1.52s/it] 85%|████████▍ | 25386/30000 [15:25:39<1:49:13, 1.42s/it] 85%|████████▍ | 25387/30000 [15:25:45<3:34:08, 2.79s/it] 85%|████████▍ | 25388/30000 [15:25:47<3:00:36, 2.35s/it] 85%|████████▍ | 25389/30000 [15:25:48<2:35:02, 2.02s/it] 85%|████████▍ | 25390/30000 [15:25:49<2:19:23, 1.81s/it] 85%|████████▍ | 25390/30000 [15:25:49<2:19:23, 1.81s/it] 85%|████████▍ | 25391/30000 [15:25:51<2:04:24, 1.62s/it] 85%|████████▍ | 25392/30000 [15:25:52<1:54:30, 1.49s/it] 85%|████████▍ | 25393/30000 [15:25:53<1:47:23, 1.40s/it] 85%|████████▍ | 25394/30000 [15:25:54<1:42:34, 1.34s/it] 85%|████████▍ | 25395/30000 [15:26:03<4:28:12, 3.49s/it] 85%|████████▍ | 25396/30000 [15:26:04<3:34:28, 2.80s/it] 85%|████████▍ | 25397/30000 [15:26:05<2:57:02, 2.31s/it] 85%|████████▍ | 25398/30000 [15:26:06<2:31:02, 1.97s/it] 85%|████████▍ | 25399/30000 [15:26:07<2:12:50, 1.73s/it] 85%|████████▍ | 25400/30000 [15:26:09<2:01:45, 1.59s/it] 85%|████████▍ | 25400/30000 [15:26:09<2:01:45, 1.59s/it] 85%|████████▍ | 25401/30000 [15:26:10<1:52:56, 1.47s/it] 85%|████████▍ | 25402/30000 [15:26:11<1:46:06, 1.38s/it] 85%|████████▍ | 25403/30000 [15:26:19<4:15:54, 3.34s/it] 85%|████████▍ | 25404/30000 [15:26:20<3:27:15, 2.71s/it] 85%|████████▍ | 25405/30000 [15:26:21<2:52:53, 2.26s/it] 85%|████████▍ | 25406/30000 [15:26:22<2:27:22, 1.92s/it] 85%|████████▍ | 25407/30000 [15:26:24<2:10:23, 1.70s/it] 85%|████████▍ | 25408/30000 [15:26:25<1:58:17, 1.55s/it] 85%|████████▍ | 25409/30000 [15:26:26<1:50:35, 1.45s/it] 85%|████████▍ | 25410/30000 [15:26:27<1:44:20, 1.36s/it] 85%|████████▍ | 25410/30000 [15:26:27<1:44:20, 1.36s/it] 85%|████████▍ | 25411/30000 [15:26:36<4:24:51, 3.46s/it] 85%|████████▍ | 25412/30000 [15:26:37<3:32:17, 2.78s/it] 85%|████████▍ | 25413/30000 [15:26:38<2:56:00, 2.30s/it] 85%|████████▍ | 25414/30000 [15:26:39<2:30:17, 1.97s/it] 85%|████████▍ | 25415/30000 [15:26:40<2:12:03, 1.73s/it] 85%|████████▍ | 25416/30000 [15:26:41<1:59:51, 1.57s/it] 85%|████████▍ | 25417/30000 [15:26:43<1:50:43, 1.45s/it] 85%|████████▍ | 25418/30000 [15:26:44<1:44:38, 1.37s/it] 85%|████████▍ | 25419/30000 [15:26:52<4:15:19, 3.34s/it] 85%|████████▍ | 25420/30000 [15:26:53<3:25:39, 2.69s/it] 85%|████████▍ | 25420/30000 [15:26:53<3:25:39, 2.69s/it] 85%|████████▍ | 25421/30000 [15:26:54<2:51:58, 2.25s/it] 85%|████████▍ | 25422/30000 [15:26:55<2:27:17, 1.93s/it] 85%|████████▍ | 25423/30000 [15:26:57<2:09:58, 1.70s/it] 85%|████████▍ | 25424/30000 [15:26:58<1:58:03, 1.55s/it] 85%|████████▍ | 25425/30000 [15:26:59<1:50:08, 1.44s/it] 85%|████████▍ | 25426/30000 [15:27:00<1:44:30, 1.37s/it] 85%|████████▍ | 25427/30000 [15:27:07<3:49:11, 3.01s/it] 85%|████████▍ | 25428/30000 [15:27:08<3:08:11, 2.47s/it] 85%|████████▍ | 25429/30000 [15:27:09<2:39:05, 2.09s/it] 85%|████████▍ | 25430/30000 [15:27:11<2:18:10, 1.81s/it] 85%|████████▍ | 25430/30000 [15:27:11<2:18:10, 1.81s/it] 85%|████████▍ | 25431/30000 [15:27:12<2:04:15, 1.63s/it] 85%|████████▍ | 25432/30000 [15:27:13<1:54:15, 1.50s/it] 85%|████████▍ | 25433/30000 [15:27:14<1:47:23, 1.41s/it] 85%|████████▍ | 25434/30000 [15:27:15<1:42:58, 1.35s/it] 85%|████████▍ | 25435/30000 [15:27:22<3:52:40, 3.06s/it] 85%|████████▍ | 25436/30000 [15:27:24<3:09:30, 2.49s/it] 85%|████████▍ | 25437/30000 [15:27:25<2:39:18, 2.09s/it] 85%|████████▍ | 25438/30000 [15:27:26<2:18:15, 1.82s/it] 85%|████████▍ | 25439/30000 [15:27:27<2:03:54, 1.63s/it] 85%|████████▍ | 25440/30000 [15:27:28<1:53:50, 1.50s/it] 85%|████████▍ | 25440/30000 [15:27:28<1:53:50, 1.50s/it] 85%|████████▍ | 25441/30000 [15:27:29<1:46:53, 1.41s/it] 85%|████████▍ | 25442/30000 [15:27:31<1:42:07, 1.34s/it] 85%|████████▍ | 25443/30000 [15:27:37<3:44:40, 2.96s/it] 85%|████████▍ | 25444/30000 [15:27:39<3:04:07, 2.42s/it] 85%|████████▍ | 25445/30000 [15:27:40<2:36:53, 2.07s/it] 85%|████████▍ | 25446/30000 [15:27:41<2:16:32, 1.80s/it] 85%|████████▍ | 25447/30000 [15:27:42<2:02:13, 1.61s/it] 85%|████████▍ | 25448/30000 [15:27:43<1:52:34, 1.48s/it] 85%|████████▍ | 25449/30000 [15:27:45<1:45:41, 1.39s/it] 85%|████████▍ | 25450/30000 [15:27:46<1:40:30, 1.33s/it] 85%|████████▍ | 25450/30000 [15:27:46<1:40:30, 1.33s/it] 85%|████████▍ | 25451/30000 [15:27:52<3:35:05, 2.84s/it] 85%|████████▍ | 25452/30000 [15:27:53<2:57:07, 2.34s/it] 85%|████████▍ | 25453/30000 [15:27:54<2:31:52, 2.00s/it] 85%|████████▍ | 25454/30000 [15:27:56<2:13:03, 1.76s/it] 85%|████████▍ | 25455/30000 [15:27:57<1:59:45, 1.58s/it] 85%|████████▍ | 25456/30000 [15:27:58<1:50:42, 1.46s/it] 85%|████████▍ | 25457/30000 [15:27:59<1:44:30, 1.38s/it] 85%|████████▍ | 25458/30000 [15:28:00<1:39:29, 1.31s/it] 85%|████████▍ | 25459/30000 [15:28:09<4:15:56, 3.38s/it] 85%|████████▍ | 25460/30000 [15:28:10<3:27:01, 2.74s/it] 85%|████████▍ | 25460/30000 [15:28:10<3:27:01, 2.74s/it] 85%|████████▍ | 25461/30000 [15:28:11<2:52:27, 2.28s/it] 85%|████████▍ | 25462/30000 [15:28:12<2:27:18, 1.95s/it] 85%|████████▍ | 25463/30000 [15:28:13<2:09:46, 1.72s/it] 85%|████████▍ | 25464/30000 [15:28:15<1:58:16, 1.56s/it] 85%|████████▍ | 25465/30000 [15:28:16<1:49:43, 1.45s/it] 85%|████████▍ | 25466/30000 [15:28:17<1:43:35, 1.37s/it] 85%|████████▍ | 25467/30000 [15:28:27<4:58:12, 3.95s/it] 85%|████████▍ | 25468/30000 [15:28:28<3:54:58, 3.11s/it] 85%|████████▍ | 25469/30000 [15:28:29<3:11:12, 2.53s/it] 85%|████████▍ | 25470/30000 [15:28:30<2:40:12, 2.12s/it] 85%|████████▍ | 25470/30000 [15:28:30<2:40:12, 2.12s/it] 85%|████████▍ | 25471/30000 [15:28:32<2:19:01, 1.84s/it] 85%|████████▍ | 25472/30000 [15:28:33<2:04:50, 1.65s/it] 85%|████████▍ | 25473/30000 [15:28:34<1:54:14, 1.51s/it] 85%|████████▍ | 25474/30000 [15:28:35<1:47:29, 1.43s/it] 85%|████████▍ | 25475/30000 [15:28:43<4:15:56, 3.39s/it] 85%|████████▍ | 25476/30000 [15:28:44<3:25:21, 2.72s/it] 85%|████████▍ | 25477/30000 [15:28:46<2:50:28, 2.26s/it] 85%|████████▍ | 25478/30000 [15:28:47<2:25:48, 1.93s/it] 85%|████████▍ | 25479/30000 [15:28:48<2:09:25, 1.72s/it] 85%|████████▍ | 25480/30000 [15:28:49<1:58:27, 1.57s/it] 85%|████████▍ | 25480/30000 [15:28:49<1:58:27, 1.57s/it] 85%|████████▍ | 25481/30000 [15:28:50<1:50:00, 1.46s/it] 85%|████████▍ | 25482/30000 [15:28:52<1:44:00, 1.38s/it] 85%|████████▍ | 25483/30000 [15:29:01<4:42:19, 3.75s/it] 85%|████████▍ | 25484/30000 [15:29:02<3:43:53, 2.97s/it] 85%|████████▍ | 25485/30000 [15:29:03<3:03:10, 2.43s/it] 85%|████████▍ | 25486/30000 [15:29:04<2:34:50, 2.06s/it] 85%|████████▍ | 25487/30000 [15:29:06<2:17:28, 1.83s/it] 85%|████████▍ | 25488/30000 [15:29:07<2:03:43, 1.65s/it] 85%|████████▍ | 25489/30000 [15:29:08<1:55:11, 1.53s/it] 85%|████████▍ | 25490/30000 [15:29:09<1:49:16, 1.45s/it] 85%|████████▍ | 25490/30000 [15:29:09<1:49:16, 1.45s/it] 85%|████████▍ | 25491/30000 [15:29:17<4:06:46, 3.28s/it] 85%|████████▍ | 25492/30000 [15:29:18<3:19:19, 2.65s/it] 85%|████████▍ | 25493/30000 [15:29:19<2:46:23, 2.22s/it] 85%|████████▍ | 25494/30000 [15:29:21<2:24:13, 1.92s/it] 85%|████████▍ | 25495/30000 [15:29:22<2:07:55, 1.70s/it] 85%|████████▍ | 25496/30000 [15:29:23<1:58:43, 1.58s/it] 85%|████████▍ | 25497/30000 [15:29:24<1:51:12, 1.48s/it] 85%|████████▍ | 25498/30000 [15:29:26<1:46:10, 1.42s/it] 85%|████████▍ | 25499/30000 [15:29:34<4:20:44, 3.48s/it] 85%|████████▌ | 25500/30000 [15:29:35<3:29:06, 2.79s/it] 85%|████████▌ | 25500/30000 [15:29:35<3:29:06, 2.79s/it] 85%|████████▌ | 25501/30000 [15:29:36<2:53:10, 2.31s/it] 85%|████████▌ | 25502/30000 [15:29:37<2:27:29, 1.97s/it] 85%|████████▌ | 25503/30000 [15:29:39<2:09:38, 1.73s/it] 85%|████████▌ | 25504/30000 [15:29:40<1:58:31, 1.58s/it] 85%|████████▌ | 25505/30000 [15:29:41<1:49:55, 1.47s/it] 85%|████████▌ | 25506/30000 [15:29:42<1:43:35, 1.38s/it] 85%|████████▌ | 25507/30000 [15:29:51<4:32:19, 3.64s/it] 85%|████████▌ | 25508/30000 [15:29:52<3:38:20, 2.92s/it] 85%|████████▌ | 25509/30000 [15:29:53<2:59:29, 2.40s/it] 85%|████████▌ | 25510/30000 [15:29:55<2:31:54, 2.03s/it] 85%|████████▌ | 25510/30000 [15:29:55<2:31:54, 2.03s/it] 85%|████████▌ | 25511/30000 [15:29:56<2:13:00, 1.78s/it] 85%|████████▌ | 25512/30000 [15:29:57<1:59:54, 1.60s/it] 85%|████████▌ | 25513/30000 [15:29:58<1:50:46, 1.48s/it] 85%|████████▌ | 25514/30000 [15:29:59<1:45:20, 1.41s/it] 85%|████████▌ | 25515/30000 [15:30:08<4:29:45, 3.61s/it] 85%|████████▌ | 25516/30000 [15:30:09<3:35:11, 2.88s/it] 85%|████████▌ | 25517/30000 [15:30:11<3:00:20, 2.41s/it] 85%|████████▌ | 25518/30000 [15:30:12<2:32:29, 2.04s/it] 85%|████████▌ | 25519/30000 [15:30:13<2:13:02, 1.78s/it] 85%|████████▌ | 25520/30000 [15:30:14<1:59:56, 1.61s/it] 85%|████████▌ | 25520/30000 [15:30:14<1:59:56, 1.61s/it] 85%|████████▌ | 25521/30000 [15:30:15<1:50:44, 1.48s/it] 85%|████████▌ | 25522/30000 [15:30:17<1:43:56, 1.39s/it] 85%|████████▌ | 25523/30000 [15:30:26<4:36:31, 3.71s/it] 85%|████████▌ | 25524/30000 [15:30:27<3:39:42, 2.95s/it] 85%|████████▌ | 25525/30000 [15:30:28<3:00:16, 2.42s/it] 85%|████████▌ | 25526/30000 [15:30:29<2:32:44, 2.05s/it] 85%|████████▌ | 25527/30000 [15:30:30<2:13:46, 1.79s/it] 85%|████████▌ | 25528/30000 [15:30:32<2:03:09, 1.65s/it] 85%|████████▌ | 25529/30000 [15:30:33<1:52:51, 1.51s/it] 85%|████████▌ | 25530/30000 [15:30:34<1:45:50, 1.42s/it] 85%|████████▌ | 25530/30000 [15:30:34<1:45:50, 1.42s/it] 85%|████████▌ | 25531/30000 [15:30:42<4:04:09, 3.28s/it] 85%|████████▌ | 25532/30000 [15:30:43<3:17:03, 2.65s/it] 85%|████████▌ | 25533/30000 [15:30:44<2:45:03, 2.22s/it] 85%|████████▌ | 25534/30000 [15:30:45<2:22:07, 1.91s/it] 85%|████████▌ | 25535/30000 [15:30:47<2:05:23, 1.69s/it] 85%|████████▌ | 25536/30000 [15:30:48<1:54:38, 1.54s/it] 85%|████████▌ | 25537/30000 [15:30:49<1:47:14, 1.44s/it] 85%|████████▌ | 25538/30000 [15:30:50<1:41:22, 1.36s/it] 85%|████████▌ | 25539/30000 [15:30:59<4:17:29, 3.46s/it] 85%|████████▌ | 25540/30000 [15:31:00<3:26:31, 2.78s/it] 85%|████████▌ | 25540/30000 [15:31:00<3:26:31, 2.78s/it] 85%|████████▌ | 25541/30000 [15:31:01<2:51:52, 2.31s/it] 85%|████████▌ | 25542/30000 [15:31:02<2:26:32, 1.97s/it] 85%|████████▌ | 25543/30000 [15:31:03<2:08:41, 1.73s/it] 85%|████████▌ | 25544/30000 [15:31:04<1:56:45, 1.57s/it] 85%|████████▌ | 25545/30000 [15:31:06<1:48:13, 1.46s/it] 85%|████████▌ | 25546/30000 [15:31:07<1:42:19, 1.38s/it] 85%|████████▌ | 25547/30000 [15:31:14<3:57:47, 3.20s/it] 85%|████████▌ | 25548/30000 [15:31:16<3:13:07, 2.60s/it] 85%|████████▌ | 25549/30000 [15:31:17<2:42:26, 2.19s/it] 85%|████████▌ | 25550/30000 [15:31:18<2:20:11, 1.89s/it] 85%|████████▌ | 25550/30000 [15:31:18<2:20:11, 1.89s/it] 85%|████████▌ | 25551/30000 [15:31:19<2:06:18, 1.70s/it] 85%|████████▌ | 25552/30000 [15:31:20<1:55:32, 1.56s/it] 85%|████████▌ | 25553/30000 [15:31:22<1:47:39, 1.45s/it] 85%|████████▌ | 25554/30000 [15:31:23<1:42:12, 1.38s/it] 85%|████████▌ | 25555/30000 [15:31:31<4:09:01, 3.36s/it] 85%|████████▌ | 25556/30000 [15:31:32<3:20:26, 2.71s/it] 85%|████████▌ | 25557/30000 [15:31:33<2:48:34, 2.28s/it] 85%|████████▌ | 25558/30000 [15:31:34<2:24:37, 1.95s/it] 85%|████████▌ | 25559/30000 [15:31:36<2:07:22, 1.72s/it] 85%|████████▌ | 25560/30000 [15:31:37<1:56:12, 1.57s/it] 85%|████████▌ | 25560/30000 [15:31:37<1:56:12, 1.57s/it] 85%|████████▌ | 25561/30000 [15:31:38<1:47:33, 1.45s/it] 85%|████████▌ | 25562/30000 [15:31:39<1:41:31, 1.37s/it] 85%|████████▌ | 25563/30000 [15:31:48<4:28:48, 3.63s/it] 85%|████████▌ | 25564/30000 [15:31:49<3:33:57, 2.89s/it] 85%|████████▌ | 25565/30000 [15:31:51<2:56:14, 2.38s/it] 85%|████████▌ | 25566/30000 [15:31:52<2:29:13, 2.02s/it] 85%|████████▌ | 25567/30000 [15:31:53<2:10:42, 1.77s/it] 85%|████████▌ | 25568/30000 [15:31:54<1:57:37, 1.59s/it] 85%|████████▌ | 25569/30000 [15:31:55<1:48:56, 1.48s/it] 85%|████████▌ | 25570/30000 [15:31:56<1:42:10, 1.38s/it] 85%|████████▌ | 25570/30000 [15:31:56<1:42:10, 1.38s/it] 85%|████████▌ | 25571/30000 [15:32:05<4:26:18, 3.61s/it] 85%|████████▌ | 25572/30000 [15:32:06<3:32:34, 2.88s/it] 85%|████████▌ | 25573/30000 [15:32:08<2:54:37, 2.37s/it] 85%|████████▌ | 25574/30000 [15:32:09<2:28:31, 2.01s/it] 85%|████████▌ | 25575/30000 [15:32:10<2:09:47, 1.76s/it] 85%|████████▌ | 25576/30000 [15:32:11<1:57:06, 1.59s/it] 85%|████████▌ | 25577/30000 [15:32:12<1:48:43, 1.47s/it] 85%|████████▌ | 25578/30000 [15:32:14<1:42:59, 1.40s/it] 85%|████████▌ | 25579/30000 [15:32:23<4:31:32, 3.69s/it] 85%|████████▌ | 25580/30000 [15:32:24<3:35:26, 2.92s/it] 85%|████████▌ | 25580/30000 [15:32:24<3:35:26, 2.92s/it] 85%|████████▌ | 25581/30000 [15:32:25<2:57:01, 2.40s/it] 85%|████████▌ | 25582/30000 [15:32:26<2:29:46, 2.03s/it] 85%|████████▌ | 25583/30000 [15:32:27<2:11:08, 1.78s/it] 85%|████████▌ | 25584/30000 [15:32:28<1:58:24, 1.61s/it] 85%|████████▌ | 25585/30000 [15:32:30<1:49:08, 1.48s/it] 85%|████████▌ | 25586/30000 [15:32:31<1:42:41, 1.40s/it] 85%|████████▌ | 25587/30000 [15:32:40<4:24:18, 3.59s/it] 85%|████████▌ | 25588/30000 [15:32:41<3:31:10, 2.87s/it] 85%|████████▌ | 25589/30000 [15:32:42<2:54:18, 2.37s/it] 85%|████████▌ | 25590/30000 [15:32:43<2:27:36, 2.01s/it] 85%|████████▌ | 25590/30000 [15:32:43<2:27:36, 2.01s/it] 85%|████████▌ | 25591/30000 [15:32:44<2:09:36, 1.76s/it] 85%|████████▌ | 25592/30000 [15:32:46<1:57:37, 1.60s/it] 85%|████████▌ | 25593/30000 [15:32:47<1:49:47, 1.49s/it] 85%|████████▌ | 25594/30000 [15:32:48<1:44:17, 1.42s/it] 85%|████████▌ | 25595/30000 [15:32:57<4:33:40, 3.73s/it] 85%|████████▌ | 25596/30000 [15:32:58<3:37:34, 2.96s/it] 85%|████████▌ | 25597/30000 [15:33:00<2:58:29, 2.43s/it] 85%|████████▌ | 25598/30000 [15:33:01<2:30:48, 2.06s/it] 85%|████████▌ | 25599/30000 [15:33:02<2:11:37, 1.79s/it] 85%|████████▌ | 25600/30000 [15:33:03<1:58:48, 1.62s/it] 85%|████████▌ | 25600/30000 [15:33:03<1:58:48, 1.62s/it] 85%|████████▌ | 25601/30000 [15:33:04<1:50:51, 1.51s/it] 85%|████████▌ | 25602/30000 [15:33:06<1:43:58, 1.42s/it] 85%|████████▌ | 25603/30000 [15:33:14<4:12:31, 3.45s/it] 85%|████████▌ | 25604/30000 [15:33:15<3:22:29, 2.76s/it] 85%|████████▌ | 25605/30000 [15:33:16<2:48:21, 2.30s/it] 85%|████████▌ | 25606/30000 [15:33:17<2:23:24, 1.96s/it] 85%|████████▌ | 25607/30000 [15:33:18<2:06:26, 1.73s/it] 85%|████████▌ | 25608/30000 [15:33:20<1:54:44, 1.57s/it] 85%|████████▌ | 25609/30000 [15:33:21<1:46:33, 1.46s/it] 85%|████████▌ | 25610/30000 [15:33:22<1:40:41, 1.38s/it] 85%|████████▌ | 25610/30000 [15:33:22<1:40:41, 1.38s/it] 85%|████████▌ | 25611/30000 [15:33:29<3:51:01, 3.16s/it] 85%|████████▌ | 25612/30000 [15:33:31<3:08:04, 2.57s/it] 85%|████████▌ | 25613/30000 [15:33:32<2:39:42, 2.18s/it] 85%|████████▌ | 25614/30000 [15:33:33<2:21:53, 1.94s/it] 85%|████████▌ | 25615/30000 [15:33:34<2:05:30, 1.72s/it] 85%|████████▌ | 25616/30000 [15:33:36<1:54:05, 1.56s/it] 85%|████████▌ | 25617/30000 [15:33:37<1:46:16, 1.45s/it] 85%|████████▌ | 25618/30000 [15:33:38<1:41:02, 1.38s/it] 85%|████████▌ | 25619/30000 [15:33:46<4:15:27, 3.50s/it] 85%|████████▌ | 25620/30000 [15:33:48<3:24:33, 2.80s/it] 85%|████████▌ | 25620/30000 [15:33:48<3:24:33, 2.80s/it] 85%|████████▌ | 25621/30000 [15:33:49<2:49:04, 2.32s/it] 85%|████████▌ | 25622/30000 [15:33:50<2:24:13, 1.98s/it] 85%|████████▌ | 25623/30000 [15:33:51<2:07:35, 1.75s/it] 85%|████████▌ | 25624/30000 [15:33:52<1:56:29, 1.60s/it] 85%|████████▌ | 25625/30000 [15:33:54<1:47:51, 1.48s/it] 85%|████████▌ | 25626/30000 [15:33:55<1:42:57, 1.41s/it] 85%|████████▌ | 25627/30000 [15:34:03<4:16:39, 3.52s/it] 85%|████████▌ | 25628/30000 [15:34:05<3:25:22, 2.82s/it] 85%|████████▌ | 25629/30000 [15:34:06<2:50:03, 2.33s/it] 85%|████████▌ | 25630/30000 [15:34:07<2:24:43, 1.99s/it] 85%|████████▌ | 25630/30000 [15:34:07<2:24:43, 1.99s/it] 85%|████████▌ | 25631/30000 [15:34:08<2:07:09, 1.75s/it] 85%|████████▌ | 25632/30000 [15:34:09<1:55:06, 1.58s/it] 85%|████████▌ | 25633/30000 [15:34:11<1:46:23, 1.46s/it] 85%|████████▌ | 25634/30000 [15:34:12<1:40:35, 1.38s/it] 85%|████████▌ | 25635/30000 [15:34:20<4:14:46, 3.50s/it] 85%|████████▌ | 25636/30000 [15:34:21<3:23:54, 2.80s/it] 85%|████████▌ | 25637/30000 [15:34:23<2:49:29, 2.33s/it] 85%|████████▌ | 25638/30000 [15:34:24<2:25:32, 2.00s/it] 85%|████████▌ | 25639/30000 [15:34:25<2:09:29, 1.78s/it] 85%|████████▌ | 25640/30000 [15:34:26<1:57:16, 1.61s/it] 85%|████████▌ | 25640/30000 [15:34:26<1:57:16, 1.61s/it] 85%|████████▌ | 25641/30000 [15:34:27<1:48:20, 1.49s/it] 85%|████████▌ | 25642/30000 [15:34:29<1:42:00, 1.40s/it] 85%|████████▌ | 25643/30000 [15:34:37<4:08:19, 3.42s/it] 85%|████████▌ | 25644/30000 [15:34:38<3:19:44, 2.75s/it] 85%|████████▌ | 25645/30000 [15:34:39<2:45:47, 2.28s/it] 85%|████████▌ | 25646/30000 [15:34:40<2:21:36, 1.95s/it] 85%|████████▌ | 25647/30000 [15:34:42<2:04:45, 1.72s/it] 85%|████████▌ | 25648/30000 [15:34:43<1:53:25, 1.56s/it] 85%|████████▌ | 25649/30000 [15:34:44<1:46:15, 1.47s/it] 86%|████████▌ | 25650/30000 [15:34:45<1:41:40, 1.40s/it] 86%|████████▌ | 25650/30000 [15:34:45<1:41:40, 1.40s/it] 86%|████████▌ | 25651/30000 [15:34:53<4:10:42, 3.46s/it] 86%|████████▌ | 25652/30000 [15:34:55<3:23:32, 2.81s/it] 86%|████████▌ | 25653/30000 [15:34:56<2:48:29, 2.33s/it] 86%|████████▌ | 25654/30000 [15:34:57<2:23:49, 1.99s/it] 86%|████████▌ | 25655/30000 [15:34:58<2:06:09, 1.74s/it] 86%|████████▌ | 25656/30000 [15:35:00<1:54:33, 1.58s/it] 86%|████████▌ | 25657/30000 [15:35:01<1:46:28, 1.47s/it] 86%|████████▌ | 25658/30000 [15:35:02<1:41:55, 1.41s/it] 86%|████████▌ | 25659/30000 [15:35:10<4:04:31, 3.38s/it] 86%|████████▌ | 25660/30000 [15:35:11<3:16:34, 2.72s/it] 86%|████████▌ | 25660/30000 [15:35:11<3:16:34, 2.72s/it] 86%|████████▌ | 25661/30000 [15:35:12<2:43:11, 2.26s/it] 86%|████████▌ | 25662/30000 [15:35:14<2:19:56, 1.94s/it] 86%|████████▌ | 25663/30000 [15:35:15<2:03:11, 1.70s/it] 86%|████████▌ | 25664/30000 [15:35:16<1:52:19, 1.55s/it] 86%|████████▌ | 25665/30000 [15:35:17<1:44:36, 1.45s/it] 86%|████████▌ | 25666/30000 [15:35:18<1:39:38, 1.38s/it] 86%|████████▌ | 25667/30000 [15:35:27<4:26:59, 3.70s/it] 86%|████████▌ | 25668/30000 [15:35:29<3:31:57, 2.94s/it] 86%|████████▌ | 25669/30000 [15:35:30<2:54:25, 2.42s/it] 86%|████████▌ | 25670/30000 [15:35:31<2:27:56, 2.05s/it] +{'loss': 0.0115, 'grad_norm': 0.05084116384387016, 'learning_rate': 1.9939394825326494e-06, 'flow_matching_loss': 0.011523432936519385, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.35} +{'loss': 0.0121, 'grad_norm': 0.05574171245098114, 'learning_rate': 1.9857100456428385e-06, 'flow_matching_loss': 0.012106937356293201, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.36} +{'loss': 0.0131, 'grad_norm': 0.07533165067434311, 'learning_rate': 1.977496422342492e-06, 'flow_matching_loss': 0.01310834102332592, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.36} +{'loss': 0.013, 'grad_norm': 0.07708322256803513, 'learning_rate': 1.9692986226119345e-06, 'flow_matching_loss': 0.01304339999333024, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.36} +{'loss': 0.0143, 'grad_norm': 0.08604677021503448, 'learning_rate': 1.9611166564122658e-06, 'flow_matching_loss': 0.014299921691417694, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.36} +{'loss': 0.0129, 'grad_norm': 0.07884234935045242, 'learning_rate': 1.9529505336853456e-06, 'flow_matching_loss': 0.012930291332304477, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.37} +{'loss': 0.0144, 'grad_norm': 0.07091029733419418, 'learning_rate': 1.9448002643537833e-06, 'flow_matching_loss': 0.014421731792390346, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.37} +{'loss': 0.0117, 'grad_norm': 0.06459653377532959, 'learning_rate': 1.9366658583209205e-06, 'flow_matching_loss': 0.011736518330872059, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.37} +{'loss': 0.0122, 'grad_norm': 0.07143402099609375, 'learning_rate': 1.9285473254708363e-06, 'flow_matching_loss': 0.012224908731877804, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.38} +{'loss': 0.0137, 'grad_norm': 0.09013223648071289, 'learning_rate': 1.9204446756683092e-06, 'flow_matching_loss': 0.013674024678766727, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.38} +{'loss': 0.0132, 'grad_norm': 0.07861611992120743, 'learning_rate': 1.912357918758822e-06, 'flow_matching_loss': 0.013178820349276065, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.38} +{'loss': 0.0123, 'grad_norm': 0.07647135108709335, 'learning_rate': 1.9042870645685478e-06, 'flow_matching_loss': 0.012334919348359109, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.38} +{'loss': 0.0134, 'grad_norm': 0.05840656906366348, 'learning_rate': 1.8962321229043322e-06, 'flow_matching_loss': 0.01341190179809928, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.39} +{'loss': 0.014, 'grad_norm': 0.06756342202425003, 'learning_rate': 1.8881931035536943e-06, 'flow_matching_loss': 0.014032341074198484, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.39} +{'loss': 0.0129, 'grad_norm': 0.06421398371458054, 'learning_rate': 1.8801700162847952e-06, 'flow_matching_loss': 0.012930939253419638, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.39} +{'loss': 0.0133, 'grad_norm': 0.07883691787719727, 'learning_rate': 1.87216287084645e-06, 'flow_matching_loss': 0.01330068688839674, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.4} +{'loss': 0.0133, 'grad_norm': 0.09124581515789032, 'learning_rate': 1.8641716769680895e-06, 'flow_matching_loss': 0.013317269645631314, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.4} +{'loss': 0.0122, 'grad_norm': 0.062327321618795395, 'learning_rate': 1.8561964443597662e-06, 'flow_matching_loss': 0.012218899372965098, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.4} +{'loss': 0.013, 'grad_norm': 0.07481712102890015, 'learning_rate': 1.848237182712147e-06, 'flow_matching_loss': 0.013039053231477738, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.4} +{'loss': 0.0127, 'grad_norm': 0.056531988084316254, 'learning_rate': 1.8402939016964788e-06, 'flow_matching_loss': 0.01274887048639357, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.41} +{'loss': 0.0133, 'grad_norm': 0.047009702771902084, 'learning_rate': 1.832366610964601e-06, 'flow_matching_loss': 0.013257852755486965, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.41} +{'loss': 0.0126, 'grad_norm': 0.05993375554680824, 'learning_rate': 1.8244553201489166e-06, 'flow_matching_loss': 0.01259139897301793, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.41} +{'loss': 0.0123, 'grad_norm': 0.0734214186668396, 'learning_rate': 1.8165600388623882e-06, 'flow_matching_loss': 0.012274976540356874, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.42} +{'loss': 0.0134, 'grad_norm': 0.07600793242454529, 'learning_rate': 1.8086807766985331e-06, 'flow_matching_loss': 0.013438132591545582, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.42} +{'loss': 0.0121, 'grad_norm': 0.05948229506611824, 'learning_rate': 1.8008175432313946e-06, 'flow_matching_loss': 0.012121975794434548, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.42} +{'loss': 0.0119, 'grad_norm': 0.08612322062253952, 'learning_rate': 1.7929703480155429e-06, 'flow_matching_loss': 0.01191346487030387, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.43} +{'loss': 0.0139, 'grad_norm': 0.07376115769147873, 'learning_rate': 1.7851392005860595e-06, 'flow_matching_loss': 0.013862775871530175, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.43} +{'loss': 0.013, 'grad_norm': 0.05657510459423065, 'learning_rate': 1.7773241104585243e-06, 'flow_matching_loss': 0.012960921786725522, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.43} +{'loss': 0.0121, 'grad_norm': 0.07718539983034134, 'learning_rate': 1.7695250871290108e-06, 'flow_matching_loss': 0.012099429173395038, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.43} +{'loss': 0.0137, 'grad_norm': 0.0667220875620842, 'learning_rate': 1.761742140074074e-06, 'flow_matching_loss': 0.013707152009010315, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.44} +{'loss': 0.0127, 'grad_norm': 0.08278442174196243, 'learning_rate': 1.7539752787507208e-06, 'flow_matching_loss': 0.012684958055615425, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.44} +{'loss': 0.0137, 'grad_norm': 0.07153280079364777, 'learning_rate': 1.7462245125964233e-06, 'flow_matching_loss': 0.01366911232471466, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.44} +{'loss': 0.0148, 'grad_norm': 0.0994393453001976, 'learning_rate': 1.7384898510290931e-06, 'flow_matching_loss': 0.014841235801577568, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.45} +{'loss': 0.0134, 'grad_norm': 0.07436629384756088, 'learning_rate': 1.730771303447069e-06, 'flow_matching_loss': 0.01342151453718543, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.45} +{'loss': 0.0129, 'grad_norm': 0.0859152302145958, 'learning_rate': 1.7230688792291204e-06, 'flow_matching_loss': 0.0129134276881814, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.45} +{'loss': 0.0145, 'grad_norm': 0.08573093265295029, 'learning_rate': 1.7153825877344175e-06, 'flow_matching_loss': 0.014512934908270835, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.45} +{'loss': 0.0128, 'grad_norm': 0.05831589549779892, 'learning_rate': 1.7077124383025278e-06, 'flow_matching_loss': 0.012800470925867558, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.46} +{'loss': 0.0123, 'grad_norm': 0.06885861605405807, 'learning_rate': 1.7000584402534058e-06, 'flow_matching_loss': 0.012330307718366384, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.46} +{'loss': 0.0129, 'grad_norm': 0.050423312932252884, 'learning_rate': 1.6924206028873774e-06, 'flow_matching_loss': 0.01293637678027153, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.46} +{'loss': 0.0136, 'grad_norm': 0.06327241659164429, 'learning_rate': 1.6847989354851434e-06, 'flow_matching_loss': 0.013649784866720438, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.47} +{'loss': 0.0115, 'grad_norm': 0.06406784802675247, 'learning_rate': 1.6771934473077443e-06, 'flow_matching_loss': 0.011483726277947425, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.47} 86%|████████▌ | 25670/30000 [15:35:31<2:27:56, 2.05s/it] 86%|████████▌ | 25671/30000 [15:35:32<2:10:50, 1.81s/it] 86%|████████▌ | 25672/30000 [15:35:33<1:58:01, 1.64s/it] 86%|████████▌ | 25673/30000 [15:35:35<1:48:33, 1.51s/it] 86%|████████▌ | 25674/30000 [15:35:36<1:41:38, 1.41s/it] 86%|████████▌ | 25675/30000 [15:35:44<4:11:46, 3.49s/it] 86%|████████▌ | 25676/30000 [15:35:45<3:22:10, 2.81s/it] 86%|████████▌ | 25677/30000 [15:35:47<2:47:51, 2.33s/it] 86%|████████▌ | 25678/30000 [15:35:48<2:23:26, 1.99s/it] 86%|████████▌ | 25679/30000 [15:35:49<2:08:05, 1.78s/it] 86%|████████▌ | 25680/30000 [15:35:50<1:55:21, 1.60s/it] 86%|████████▌ | 25680/30000 [15:35:50<1:55:21, 1.60s/it] 86%|████████▌ | 25681/30000 [15:35:51<1:45:53, 1.47s/it] 86%|████████▌ | 25682/30000 [15:35:53<1:40:23, 1.40s/it] 86%|████████▌ | 25683/30000 [15:35:59<3:34:57, 2.99s/it] 86%|████████▌ | 25684/30000 [15:36:01<2:56:15, 2.45s/it] 86%|████████▌ | 25685/30000 [15:36:02<2:29:31, 2.08s/it] 86%|████████▌ | 25686/30000 [15:36:03<2:10:14, 1.81s/it] 86%|████████▌ | 25687/30000 [15:36:04<1:56:38, 1.62s/it] 86%|████████▌ | 25688/30000 [15:36:06<1:49:54, 1.53s/it] 86%|████████▌ | 25689/30000 [15:36:07<1:42:52, 1.43s/it] 86%|████████▌ | 25690/30000 [15:36:08<1:37:44, 1.36s/it] 86%|████████▌ | 25690/30000 [15:36:08<1:37:44, 1.36s/it] 86%|████████▌ | 25691/30000 [15:36:16<3:55:56, 3.29s/it] 86%|████████▌ | 25692/30000 [15:36:17<3:10:32, 2.65s/it] 86%|████████▌ | 25693/30000 [15:36:18<2:40:01, 2.23s/it] 86%|████████▌ | 25694/30000 [15:36:19<2:18:12, 1.93s/it] 86%|████████▌ | 25695/30000 [15:36:20<2:01:56, 1.70s/it] 86%|████████▌ | 25696/30000 [15:36:22<1:53:29, 1.58s/it] 86%|████████▌ | 25697/30000 [15:36:23<1:45:05, 1.47s/it] 86%|████████▌ | 25698/30000 [15:36:24<1:38:48, 1.38s/it] 86%|████████▌ | 25699/30000 [15:36:33<4:20:07, 3.63s/it] 86%|████████▌ | 25700/30000 [15:36:34<3:27:07, 2.89s/it] 86%|████████▌ | 25700/30000 [15:36:34<3:27:07, 2.89s/it] 86%|████████▌ | 25701/30000 [15:36:35<2:50:34, 2.38s/it] 86%|████████▌ | 25702/30000 [15:36:37<2:24:28, 2.02s/it] 86%|████████▌ | 25703/30000 [15:36:38<2:05:57, 1.76s/it] 86%|████████▌ | 25704/30000 [15:36:39<1:53:48, 1.59s/it] 86%|████████▌ | 25705/30000 [15:36:40<1:45:10, 1.47s/it] 86%|████████▌ | 25706/30000 [15:36:41<1:39:37, 1.39s/it] 86%|████████▌ | 25707/30000 [15:36:50<4:19:18, 3.62s/it] 86%|████████▌ | 25708/30000 [15:36:51<3:26:24, 2.89s/it] 86%|████████▌ | 25709/30000 [15:36:53<2:49:57, 2.38s/it] 86%|████████▌ | 25710/30000 [15:36:54<2:24:03, 2.01s/it] 86%|████████▌ | 25710/30000 [15:36:54<2:24:03, 2.01s/it] 86%|████████▌ | 25711/30000 [15:36:55<2:05:58, 1.76s/it] 86%|████████▌ | 25712/30000 [15:36:56<1:53:10, 1.58s/it] 86%|████████▌ | 25713/30000 [15:36:57<1:44:23, 1.46s/it] 86%|████████▌ | 25714/30000 [15:36:58<1:38:56, 1.39s/it] 86%|████████▌ | 25715/30000 [15:37:07<4:08:07, 3.47s/it] 86%|████████▌ | 25716/30000 [15:37:08<3:21:05, 2.82s/it] 86%|████████▌ | 25717/30000 [15:37:09<2:46:25, 2.33s/it] 86%|████████▌ | 25718/30000 [15:37:10<2:23:17, 2.01s/it] 86%|████████▌ | 25719/30000 [15:37:12<2:05:08, 1.75s/it] 86%|████████▌ | 25720/30000 [15:37:13<1:52:33, 1.58s/it] 86%|████████▌ | 25720/30000 [15:37:13<1:52:33, 1.58s/it] 86%|████████▌ | 25721/30000 [15:37:14<1:43:54, 1.46s/it] 86%|████████▌ | 25722/30000 [15:37:15<1:38:18, 1.38s/it] 86%|████████▌ | 25723/30000 [15:37:24<4:14:54, 3.58s/it] 86%|████████▌ | 25724/30000 [15:37:25<3:23:19, 2.85s/it] 86%|████████▌ | 25725/30000 [15:37:26<2:47:47, 2.35s/it] 86%|████████▌ | 25726/30000 [15:37:27<2:22:09, 2.00s/it] 86%|████████▌ | 25727/30000 [15:37:29<2:04:20, 1.75s/it] 86%|████████▌ | 25728/30000 [15:37:30<1:52:21, 1.58s/it] 86%|████████▌ | 25729/30000 [15:37:31<1:43:53, 1.46s/it] 86%|████████▌ | 25730/30000 [15:37:32<1:38:19, 1.38s/it] 86%|████████▌ | 25730/30000 [15:37:32<1:38:19, 1.38s/it] 86%|████████▌ | 25731/30000 [15:37:41<4:17:10, 3.61s/it] 86%|████████▌ | 25732/30000 [15:37:42<3:25:20, 2.89s/it] 86%|████████▌ | 25733/30000 [15:37:43<2:49:20, 2.38s/it] 86%|████████▌ | 25734/30000 [15:37:45<2:23:51, 2.02s/it] 86%|████████▌ | 25735/30000 [15:37:46<2:05:26, 1.76s/it] 86%|████████▌ | 25736/30000 [15:37:47<1:53:32, 1.60s/it] 86%|████████▌ | 25737/30000 [15:37:48<1:44:51, 1.48s/it] 86%|████████▌ | 25738/30000 [15:37:49<1:38:47, 1.39s/it] 86%|████████▌ | 25739/30000 [15:37:59<4:27:00, 3.76s/it] 86%|████████▌ | 25740/30000 [15:38:00<3:31:47, 2.98s/it] 86%|████████▌ | 25740/30000 [15:38:00<3:31:47, 2.98s/it] 86%|████████▌ | 25741/30000 [15:38:01<2:54:00, 2.45s/it] 86%|████████▌ | 25742/30000 [15:38:02<2:26:56, 2.07s/it] 86%|████████▌ | 25743/30000 [15:38:03<2:07:48, 1.80s/it] 86%|████████▌ | 25744/30000 [15:38:05<1:54:51, 1.62s/it] 86%|████████▌ | 25745/30000 [15:38:06<1:45:13, 1.48s/it] 86%|████████▌ | 25746/30000 [15:38:07<1:39:04, 1.40s/it] 86%|████████▌ | 25747/30000 [15:38:15<3:56:23, 3.33s/it] 86%|████████▌ | 25748/30000 [15:38:16<3:10:20, 2.69s/it] 86%|████████▌ | 25749/30000 [15:38:17<2:38:31, 2.24s/it] 86%|████████▌ | 25750/30000 [15:38:18<2:15:59, 1.92s/it] 86%|████████▌ | 25750/30000 [15:38:18<2:15:59, 1.92s/it] 86%|████████▌ | 25751/30000 [15:38:19<2:00:22, 1.70s/it] 86%|████████▌ | 25752/30000 [15:38:21<1:50:39, 1.56s/it] 86%|████████▌ | 25753/30000 [15:38:22<1:43:41, 1.46s/it] 86%|████████▌ | 25754/30000 [15:38:23<1:38:09, 1.39s/it] 86%|████████▌ | 25755/30000 [15:38:31<3:48:29, 3.23s/it] 86%|████████▌ | 25756/30000 [15:38:32<3:05:22, 2.62s/it] 86%|████████▌ | 25757/30000 [15:38:33<2:35:08, 2.19s/it] 86%|████████▌ | 25758/30000 [15:38:34<2:13:35, 1.89s/it] 86%|████████▌ | 25759/30000 [15:38:35<1:58:17, 1.67s/it] 86%|████████▌ | 25760/30000 [15:38:37<1:48:04, 1.53s/it] 86%|████████▌ | 25760/30000 [15:38:37<1:48:04, 1.53s/it] 86%|████████▌ | 25761/30000 [15:38:38<1:40:42, 1.43s/it] 86%|████████▌ | 25762/30000 [15:38:39<1:36:44, 1.37s/it] 86%|████████▌ | 25763/30000 [15:38:48<4:13:01, 3.58s/it] 86%|████████▌ | 25764/30000 [15:38:49<3:22:00, 2.86s/it] 86%|████████▌ | 25765/30000 [15:38:50<2:46:49, 2.36s/it] 86%|████████▌ | 25766/30000 [15:38:51<2:22:04, 2.01s/it] 86%|████████▌ | 25767/30000 [15:38:53<2:05:02, 1.77s/it] 86%|████████▌ | 25768/30000 [15:38:54<1:52:28, 1.59s/it] 86%|████████▌ | 25769/30000 [15:38:55<1:44:35, 1.48s/it] 86%|████████▌ | 25770/30000 [15:38:56<1:39:49, 1.42s/it] 86%|████████▌ | 25770/30000 [15:38:56<1:39:49, 1.42s/it] 86%|████████▌ | 25771/30000 [15:39:05<4:08:01, 3.52s/it] 86%|████████▌ | 25772/30000 [15:39:06<3:18:15, 2.81s/it] 86%|████████▌ | 25773/30000 [15:39:07<2:45:37, 2.35s/it] 86%|████████▌ | 25774/30000 [15:39:08<2:21:09, 2.00s/it] 86%|████████▌ | 25775/30000 [15:39:10<2:04:27, 1.77s/it] 86%|████████▌ | 25776/30000 [15:39:11<1:52:32, 1.60s/it] 86%|████████▌ | 25777/30000 [15:39:12<1:44:33, 1.49s/it] 86%|████████▌ | 25778/30000 [15:39:13<1:40:17, 1.43s/it] 86%|████████▌ | 25779/30000 [15:39:21<4:03:36, 3.46s/it] 86%|████████▌ | 25780/30000 [15:39:23<3:15:04, 2.77s/it] 86%|████████▌ | 25780/30000 [15:39:23<3:15:04, 2.77s/it] 86%|████████▌ | 25781/30000 [15:39:24<2:42:05, 2.31s/it] 86%|████████▌ | 25782/30000 [15:39:25<2:18:20, 1.97s/it] 86%|████████▌ | 25783/30000 [15:39:26<2:02:17, 1.74s/it] 86%|████████▌ | 25784/30000 [15:39:27<1:51:07, 1.58s/it] 86%|████████▌ | 25785/30000 [15:39:29<1:42:59, 1.47s/it] 86%|████████▌ | 25786/30000 [15:39:30<1:37:36, 1.39s/it] 86%|████████▌ | 25787/30000 [15:39:39<4:16:25, 3.65s/it] 86%|████████▌ | 25788/30000 [15:39:40<3:25:32, 2.93s/it] 86%|████████▌ | 25789/30000 [15:39:41<2:49:35, 2.42s/it] 86%|████████▌ | 25790/30000 [15:39:42<2:25:09, 2.07s/it] 86%|████████▌ | 25790/30000 [15:39:43<2:25:09, 2.07s/it] 86%|████████▌ | 25791/30000 [15:39:44<2:08:54, 1.84s/it] 86%|████████▌ | 25792/30000 [15:39:45<1:57:03, 1.67s/it] 86%|████████▌ | 25793/30000 [15:39:46<1:47:25, 1.53s/it] 86%|████████▌ | 25794/30000 [15:39:47<1:40:37, 1.44s/it] 86%|████████▌ | 25795/30000 [15:39:56<4:00:23, 3.43s/it] 86%|████████▌ | 25796/30000 [15:39:57<3:12:50, 2.75s/it] 86%|████████▌ | 25797/30000 [15:39:58<2:40:17, 2.29s/it] 86%|████████▌ | 25798/30000 [15:39:59<2:17:20, 1.96s/it] 86%|████████▌ | 25799/30000 [15:40:00<2:00:44, 1.72s/it] 86%|████████▌ | 25800/30000 [15:40:01<1:49:39, 1.57s/it] 86%|████████▌ | 25800/30000 [15:40:02<1:49:39, 1.57s/it] 86%|████████▌ | 25801/30000 [15:40:03<1:41:38, 1.45s/it] 86%|████████▌ | 25802/30000 [15:40:04<1:36:23, 1.38s/it] 86%|████████▌ | 25803/30000 [15:40:12<3:54:48, 3.36s/it] 86%|████████▌ | 25804/30000 [15:40:13<3:10:17, 2.72s/it] 86%|████████▌ | 25805/30000 [15:40:14<2:41:00, 2.30s/it] 86%|████████▌ | 25806/30000 [15:40:16<2:17:26, 1.97s/it] 86%|████████▌ | 25807/30000 [15:40:17<2:00:42, 1.73s/it] 86%|████████▌ | 25808/30000 [15:40:18<1:49:09, 1.56s/it] 86%|████████▌ | 25809/30000 [15:40:19<1:40:44, 1.44s/it] 86%|████████▌ | 25810/30000 [15:40:20<1:35:29, 1.37s/it] 86%|████████▌ | 25810/30000 [15:40:20<1:35:29, 1.37s/it] 86%|████████▌ | 25811/30000 [15:40:28<3:56:49, 3.39s/it] 86%|████████▌ | 25812/30000 [15:40:30<3:11:15, 2.74s/it] 86%|████████▌ | 25813/30000 [15:40:31<2:40:08, 2.29s/it] 86%|████████▌ | 25814/30000 [15:40:32<2:19:24, 2.00s/it] 86%|████████▌ | 25815/30000 [15:40:33<2:04:01, 1.78s/it] 86%|████████▌ | 25816/30000 [15:40:35<1:53:22, 1.63s/it] 86%|████████▌ | 25817/30000 [15:40:36<1:44:24, 1.50s/it] 86%|████████▌ | 25818/30000 [15:40:37<1:38:00, 1.41s/it] 86%|████████▌ | 25819/30000 [15:40:45<3:53:21, 3.35s/it] 86%|████████▌ | 25820/30000 [15:40:46<3:08:07, 2.70s/it] 86%|████████▌ | 25820/30000 [15:40:46<3:08:07, 2.70s/it] 86%|████████▌ | 25821/30000 [15:40:47<2:36:35, 2.25s/it] 86%|████████▌ | 25822/30000 [15:40:49<2:14:17, 1.93s/it] 86%|████████▌ | 25823/30000 [15:40:50<1:59:56, 1.72s/it] 86%|████████▌ | 25824/30000 [15:40:51<1:49:25, 1.57s/it] 86%|████████▌ | 25825/30000 [15:40:52<1:43:13, 1.48s/it] 86%|████████▌ | 25826/30000 [15:40:54<1:39:59, 1.44s/it] 86%|████████▌ | 25827/30000 [15:41:00<3:26:17, 2.97s/it] 86%|████████▌ | 25828/30000 [15:41:01<2:49:09, 2.43s/it] 86%|████████▌ | 25829/30000 [15:41:03<2:23:47, 2.07s/it] 86%|████████▌ | 25830/30000 [15:41:04<2:05:48, 1.81s/it] 86%|████████▌ | 25830/30000 [15:41:04<2:05:48, 1.81s/it] 86%|████████▌ | 25831/30000 [15:41:05<1:53:06, 1.63s/it] 86%|████████▌ | 25832/30000 [15:41:06<1:44:05, 1.50s/it] 86%|████████▌ | 25833/30000 [15:41:07<1:37:30, 1.40s/it] 86%|████████▌ | 25834/30000 [15:41:09<1:32:34, 1.33s/it] 86%|████████▌ | 25835/30000 [15:41:16<3:42:05, 3.20s/it] 86%|████████▌ | 25836/30000 [15:41:17<3:00:14, 2.60s/it] 86%|████████▌ | 25837/30000 [15:41:18<2:31:21, 2.18s/it] 86%|████████▌ | 25838/30000 [15:41:20<2:10:57, 1.89s/it] 86%|████████▌ | 25839/30000 [15:41:21<1:57:31, 1.69s/it] 86%|████████▌ | 25840/30000 [15:41:22<1:47:26, 1.55s/it] 86%|████████▌ | 25840/30000 [15:41:22<1:47:26, 1.55s/it] 86%|████████▌ | 25841/30000 [15:41:23<1:40:57, 1.46s/it] 86%|████████▌ | 25842/30000 [15:41:25<1:35:22, 1.38s/it] 86%|████████▌ | 25843/30000 [15:41:32<3:45:15, 3.25s/it] 86%|████████▌ | 25844/30000 [15:41:33<3:02:17, 2.63s/it] 86%|████████▌ | 25845/30000 [15:41:35<2:32:36, 2.20s/it] 86%|████████▌ | 25846/30000 [15:41:36<2:11:22, 1.90s/it] 86%|████████▌ | 25847/30000 [15:41:37<1:58:54, 1.72s/it] 86%|████████▌ | 25848/30000 [15:41:38<1:48:18, 1.57s/it] 86%|████████▌ | 25849/30000 [15:41:39<1:40:50, 1.46s/it] 86%|████████▌ | 25850/30000 [15:41:41<1:35:53, 1.39s/it] 86%|████████▌ | 25850/30000 [15:41:41<1:35:53, 1.39s/it] 86%|████████▌ | 25851/30000 [15:41:47<3:23:07, 2.94s/it] 86%|████████▌ | 25852/30000 [15:41:48<2:46:37, 2.41s/it] 86%|████████▌ | 25853/30000 [15:41:50<2:20:38, 2.03s/it] 86%|████████▌ | 25854/30000 [15:41:51<2:02:47, 1.78s/it] 86%|████████▌ | 25855/30000 [15:41:52<1:51:02, 1.61s/it] 86%|████████▌ | 25856/30000 [15:41:53<1:42:59, 1.49s/it] 86%|████████▌ | 25857/30000 [15:41:54<1:37:35, 1.41s/it] 86%|████████▌ | 25858/30000 [15:41:56<1:33:31, 1.35s/it] 86%|████████▌ | 25859/30000 [15:42:02<3:24:58, 2.97s/it] 86%|████████▌ | 25860/30000 [15:42:04<2:47:48, 2.43s/it] 86%|████████▌ | 25860/30000 [15:42:04<2:47:48, 2.43s/it] 86%|████████▌ | 25861/30000 [15:42:05<2:21:57, 2.06s/it] 86%|████████▌ | 25862/30000 [15:42:06<2:03:24, 1.79s/it] 86%|████████▌ | 25863/30000 [15:42:07<1:50:43, 1.61s/it] 86%|████████▌ | 25864/30000 [15:42:08<1:42:23, 1.49s/it] 86%|████████▌ | 25865/30000 [15:42:09<1:35:37, 1.39s/it] 86%|████████▌ | 25866/30000 [15:42:11<1:34:15, 1.37s/it] 86%|████████▌ | 25867/30000 [15:42:19<4:04:09, 3.54s/it] 86%|████████▌ | 25868/30000 [15:42:21<3:15:16, 2.84s/it] 86%|████████▌ | 25869/30000 [15:42:22<2:41:00, 2.34s/it] 86%|████████▌ | 25870/30000 [15:42:23<2:17:07, 1.99s/it] 86%|████████▌ | 25870/30000 [15:42:23<2:17:07, 1.99s/it] 86%|████████▌ | 25871/30000 [15:42:24<2:00:20, 1.75s/it] 86%|████████▌ | 25872/30000 [15:42:25<1:49:15, 1.59s/it] 86%|████████▌ | 25873/30000 [15:42:27<1:41:10, 1.47s/it] 86%|████████▌ | 25874/30000 [15:42:28<1:36:31, 1.40s/it] 86%|████████▋ | 25875/30000 [15:42:36<3:55:10, 3.42s/it] 86%|████████▋ | 25876/30000 [15:42:37<3:08:59, 2.75s/it] 86%|████████▋ | 25877/30000 [15:42:38<2:37:16, 2.29s/it] 86%|████████▋ | 25878/30000 [15:42:39<2:14:03, 1.95s/it] 86%|████████▋ | 25879/30000 [15:42:41<1:58:30, 1.73s/it] 86%|████████▋ | 25880/30000 [15:42:42<1:48:56, 1.59s/it] 86%|████████▋ | 25880/30000 [15:42:42<1:48:56, 1.59s/it] 86%|████████▋ | 25881/30000 [15:42:43<1:41:27, 1.48s/it] 86%|████████▋ | 25882/30000 [15:42:44<1:35:22, 1.39s/it] 86%|████████▋ | 25883/30000 [15:42:52<3:53:34, 3.40s/it] 86%|████████▋ | 25884/30000 [15:42:54<3:08:26, 2.75s/it] 86%|████████▋ | 25885/30000 [15:42:55<2:36:38, 2.28s/it] 86%|████████▋ | 25886/30000 [15:42:56<2:14:58, 1.97s/it] 86%|████████▋ | 25887/30000 [15:42:57<1:59:28, 1.74s/it] 86%|████████▋ | 25888/30000 [15:42:59<1:48:10, 1.58s/it] 86%|████████▋ | 25889/30000 [15:43:00<1:40:04, 1.46s/it] 86%|████████▋ | 25890/30000 [15:43:01<1:34:36, 1.38s/it] 86%|████████▋ | 25890/30000 [15:43:01<1:34:36, 1.38s/it] 86%|████████▋ | 25891/30000 [15:43:09<3:45:55, 3.30s/it] 86%|████████▋ | 25892/30000 [15:43:10<3:02:19, 2.66s/it] 86%|████████▋ | 25893/30000 [15:43:11<2:31:43, 2.22s/it] 86%|████████▋ | 25894/30000 [15:43:12<2:10:13, 1.90s/it] 86%|████████▋ | 25895/30000 [15:43:14<1:57:52, 1.72s/it] 86%|████████▋ | 25896/30000 [15:43:15<1:46:56, 1.56s/it] 86%|████████▋ | 25897/30000 [15:43:16<1:39:58, 1.46s/it] 86%|████████▋ | 25898/30000 [15:43:17<1:35:42, 1.40s/it] 86%|████████▋ | 25899/30000 [15:43:25<3:45:12, 3.29s/it] 86%|████████▋ | 25900/30000 [15:43:26<3:01:22, 2.65s/it] 86%|████████▋ | 25900/30000 [15:43:26<3:01:22, 2.65s/it] 86%|████████▋ | 25901/30000 [15:43:27<2:31:21, 2.22s/it] 86%|████████▋ | 25902/30000 [15:43:28<2:10:55, 1.92s/it] 86%|████████▋ | 25903/30000 [15:43:30<1:56:08, 1.70s/it] 86%|████████▋ | 25904/30000 [15:43:31<1:47:39, 1.58s/it] 86%|████████▋ | 25905/30000 [15:43:32<1:39:15, 1.45s/it] 86%|████████▋ | 25906/30000 [15:43:33<1:34:04, 1.38s/it] 86%|████████▋ | 25907/30000 [15:43:42<4:04:11, 3.58s/it] 86%|████████▋ | 25908/30000 [15:43:43<3:15:00, 2.86s/it] 86%|████████▋ | 25909/30000 [15:43:44<2:41:40, 2.37s/it] 86%|████████▋ | 25910/30000 [15:43:46<2:16:59, 2.01s/it] 86%|████████▋ | 25910/30000 [15:43:46<2:16:59, 2.01s/it] 86%|████████▋ | 25911/30000 [15:43:47<2:00:38, 1.77s/it] 86%|████████▋ | 25912/30000 [15:43:48<1:48:14, 1.59s/it] 86%|████████▋ | 25913/30000 [15:43:49<1:39:37, 1.46s/it] 86%|████████▋ | 25914/30000 [15:43:50<1:33:39, 1.38s/it] 86%|████████▋ | 25915/30000 [15:43:58<3:49:14, 3.37s/it] 86%|████████▋ | 25916/30000 [15:44:00<3:04:17, 2.71s/it] 86%|████████▋ | 25917/30000 [15:44:01<2:33:14, 2.25s/it] 86%|████████▋ | 25918/30000 [15:44:02<2:10:58, 1.93s/it] 86%|████████▋ | 25919/30000 [15:44:03<1:55:42, 1.70s/it] 86%|████████▋ | 25920/30000 [15:44:04<1:45:30, 1.55s/it] 86%|████████▋ | 25920/30000 [15:44:04<1:45:30, 1.55s/it] 86%|████████▋ | 25921/30000 [15:44:05<1:38:29, 1.45s/it] 86%|████████▋ | 25922/30000 [15:44:07<1:33:14, 1.37s/it] 86%|████████▋ | 25923/30000 [15:44:14<3:44:45, 3.31s/it] 86%|████████▋ | 25924/30000 [15:44:16<3:00:54, 2.66s/it] 86%|████████▋ | 25925/30000 [15:44:17<2:31:12, 2.23s/it] 86%|████████▋ | 25926/30000 [15:44:18<2:09:15, 1.90s/it] 86%|████████▋ | 25927/30000 [15:44:19<1:54:14, 1.68s/it] 86%|████████▋ | 25928/30000 [15:44:20<1:45:16, 1.55s/it] 86%|████████▋ | 25929/30000 [15:44:22<1:37:39, 1.44s/it] 86%|████████▋ | 25930/30000 [15:44:23<1:32:36, 1.37s/it] 86%|████████▋ | 25930/30000 [15:44:23<1:32:36, 1.37s/it] 86%|████████▋ | 25931/30000 [15:44:31<3:42:28, 3.28s/it] 86%|████████▋ | 25932/30000 [15:44:32<2:59:57, 2.65s/it] 86%|████████▋ | 25933/30000 [15:44:33<2:30:02, 2.21s/it] 86%|████████▋ | 25934/30000 [15:44:34<2:08:51, 1.90s/it] 86%|████████▋ | 25935/30000 [15:44:35<1:54:14, 1.69s/it] 86%|████████▋ | 25936/30000 [15:44:36<1:44:24, 1.54s/it] 86%|████████▋ | 25937/30000 [15:44:38<1:37:40, 1.44s/it] 86%|████████▋ | 25938/30000 [15:44:39<1:32:23, 1.36s/it] 86%|████████▋ | 25939/30000 [15:44:47<3:45:16, 3.33s/it] 86%|████████▋ | 25940/30000 [15:44:48<3:02:20, 2.69s/it] 86%|████████▋ | 25940/30000 [15:44:48<3:02:20, 2.69s/it] 86%|████████▋ | 25941/30000 [15:44:49<2:32:29, 2.25s/it] 86%|████████▋ | 25942/30000 [15:44:50<2:10:33, 1.93s/it] 86%|████████▋ | 25943/30000 [15:44:52<1:56:26, 1.72s/it] 86%|████████▋ | 25944/30000 [15:44:53<1:46:18, 1.57s/it] 86%|████████▋ | 25945/30000 [15:44:54<1:40:16, 1.48s/it] 86%|████████▋ | 25946/30000 [15:44:55<1:37:13, 1.44s/it] 86%|████████▋ | 25947/30000 [15:45:03<3:49:15, 3.39s/it] 86%|████████▋ | 25948/30000 [15:45:05<3:04:17, 2.73s/it] 86%|████████▋ | 25949/30000 [15:45:06<2:32:42, 2.26s/it] 86%|████████▋ | 25950/30000 [15:45:07<2:10:54, 1.94s/it] 86%|████████▋ | 25950/30000 [15:45:07<2:10:54, 1.94s/it] 87%|████████▋ | 25951/30000 [15:45:08<1:56:04, 1.72s/it] 87%|████████▋ | 25952/30000 [15:45:09<1:46:24, 1.58s/it] 87%|████████▋ | 25953/30000 [15:45:11<1:39:53, 1.48s/it] 87%|████████▋ | 25954/30000 [15:45:12<1:34:34, 1.40s/it] 87%|████████▋ | 25955/30000 [15:45:20<3:51:57, 3.44s/it] 87%|████████▋ | 25956/30000 [15:45:21<3:05:46, 2.76s/it] 87%|████████▋ | 25957/30000 [15:45:22<2:34:16, 2.29s/it] 87%|████████▋ | 25958/30000 [15:45:24<2:11:57, 1.96s/it] 87%|████████▋ | 25959/30000 [15:45:25<1:55:57, 1.72s/it] 87%|████████▋ | 25960/30000 [15:45:26<1:46:13, 1.58s/it] 87%|████████▋ | 25960/30000 [15:45:26<1:46:13, 1.58s/it] 87%|████████▋ | 25961/30000 [15:45:27<1:39:00, 1.47s/it] 87%|████████▋ | 25962/30000 [15:45:28<1:33:10, 1.38s/it] 87%|████████▋ | 25963/30000 [15:45:37<4:03:18, 3.62s/it] 87%|████████▋ | 25964/30000 [15:45:38<3:13:44, 2.88s/it] 87%|████████▋ | 25965/30000 [15:45:40<2:39:35, 2.37s/it] 87%|████████▋ | 25966/30000 [15:45:41<2:15:46, 2.02s/it] 87%|████████▋ | 25967/30000 [15:45:42<1:59:58, 1.78s/it] 87%|████████▋ | 25968/30000 [15:45:43<1:48:50, 1.62s/it] 87%|████████▋ | 25969/30000 [15:45:45<1:40:58, 1.50s/it] 87%|████████▋ | 25970/30000 [15:45:46<1:34:49, 1.41s/it] 87%|████████▋ | 25970/30000 [15:45:46<1:34:49, 1.41s/it] 87%|████████▋ | 25971/30000 [15:45:53<3:27:42, 3.09s/it] 87%|████████▋ | 25972/30000 [15:45:54<2:49:59, 2.53s/it] 87%|████████▋ | 25973/30000 [15:45:55<2:23:07, 2.13s/it] 87%|████████▋ | 25974/30000 [15:45:56<2:03:47, 1.84s/it] 87%|████████▋ | 25975/30000 [15:45:57<1:50:05, 1.64s/it] 87%|████████▋ | 25976/30000 [15:45:59<1:40:58, 1.51s/it] 87%|████████▋ | 25977/30000 [15:46:00<1:34:47, 1.41s/it] 87%|████████▋ | 25978/30000 [15:46:01<1:30:08, 1.34s/it] 87%|████████▋ | 25979/30000 [15:46:08<3:25:25, 3.07s/it] 87%|████████▋ | 25980/30000 [15:46:09<2:47:48, 2.50s/it] 87%|████████▋ | 25980/30000 [15:46:09<2:47:48, 2.50s/it] 87%|████████▋ | 25981/30000 [15:46:11<2:21:07, 2.11s/it] 87%|████████▋ | 25982/30000 [15:46:12<2:04:31, 1.86s/it] 87%|████████▋ | 25983/30000 [15:46:13<1:52:42, 1.68s/it] 87%|████████▋ | 25984/30000 [15:46:14<1:43:29, 1.55s/it] 87%|████████▋ | 25985/30000 [15:46:15<1:36:36, 1.44s/it] 87%|████████▋ | 25986/30000 [15:46:17<1:31:36, 1.37s/it] 87%|████████▋ | 25987/30000 [15:46:24<3:40:05, 3.29s/it] 87%|████████▋ | 25988/30000 [15:46:26<2:57:29, 2.65s/it] 87%|████████▋ | 25989/30000 [15:46:27<2:28:26, 2.22s/it] 87%|████████▋ | 25990/30000 [15:46:28<2:07:52, 1.91s/it] 87%|████████▋ | 25990/30000 [15:46:28<2:07:52, 1.91s/it] 87%|████████▋ | 25991/30000 [15:46:29<1:54:05, 1.71s/it] 87%|████████▋ | 25992/30000 [15:46:31<1:44:36, 1.57s/it] 87%|████████▋ | 25993/30000 [15:46:32<1:37:10, 1.46s/it] 87%|████████▋ | 25994/30000 [15:46:33<1:34:17, 1.41s/it] 87%|████████▋ | 25995/30000 [15:46:41<3:43:10, 3.34s/it] 87%|████████▋ | 25996/30000 [15:46:42<2:59:57, 2.70s/it] 87%|████████▋ | 25997/30000 [15:46:43<2:29:44, 2.24s/it] 87%|████████▋ | 25998/30000 [15:46:44<2:08:16, 1.92s/it] 87%|████████▋ | 25999/30000 [15:46:46<1:52:55, 1.69s/it] 87%|████████▋ | 26000/30000 [15:46:47<1:42:34, 1.54s/it] 87%|████████▋ | 26000/30000 [15:46:47<1:42:34, 1.54s/it] 87%|████████▋ | 26001/30000 [15:46:48<1:35:43, 1.44s/it] 87%|████████▋ | 26002/30000 [15:46:49<1:31:04, 1.37s/it] 87%|████████▋ | 26003/30000 [15:46:58<3:59:08, 3.59s/it] 87%|████████▋ | 26004/30000 [15:46:59<3:10:32, 2.86s/it] 87%|████████▋ | 26005/30000 [15:47:00<2:37:27, 2.36s/it] 87%|████████▋ | 26006/30000 [15:47:01<2:13:18, 2.00s/it] 87%|████████▋ | 26007/30000 [15:47:03<1:57:10, 1.76s/it] 87%|████████▋ | 26008/30000 [15:47:04<1:45:32, 1.59s/it] 87%|████████▋ | 26009/30000 [15:47:05<1:37:28, 1.47s/it] 87%|████████▋ | 26010/30000 [15:47:06<1:32:10, 1.39s/it] 87%|████████▋ | 26010/30000 [15:47:06<1:32:10, 1.39s/it] 87%|████████▋ | 26011/30000 [15:47:14<3:43:24, 3.36s/it] 87%|████████▋ | 26012/30000 [15:47:15<2:59:46, 2.70s/it] 87%|████████▋ | 26013/30000 [15:47:17<2:29:31, 2.25s/it] 87%|████████▋ | 26014/30000 [15:47:18<2:07:45, 1.92s/it] 87%|████████▋ | 26015/30000 [15:47:19<1:52:45, 1.70s/it] 87%|████████▋ | 26016/30000 [15:47:20<1:43:27, 1.56s/it] 87%|████████▋ | 26017/30000 [15:47:21<1:36:32, 1.45s/it] 87%|████████▋ | 26018/30000 [15:47:23<1:34:00, 1.42s/it] 87%|████████▋ | 26019/30000 [15:47:30<3:38:27, 3.29s/it] 87%|████████▋ | 26020/30000 [15:47:32<2:56:38, 2.66s/it] 87%|████████▋ | 26020/30000 [15:47:32<2:56:38, 2.66s/it] 87%|████████▋ | 26021/30000 [15:47:33<2:27:24, 2.22s/it] 87%|████████▋ | 26022/30000 [15:47:34<2:06:45, 1.91s/it] 87%|████████▋ | 26023/30000 [15:47:35<1:54:14, 1.72s/it] 87%|████████▋ | 26024/30000 [15:47:36<1:44:21, 1.57s/it] 87%|████████▋ | 26025/30000 [15:47:38<1:37:01, 1.46s/it] 87%|████████▋ | 26026/30000 [15:47:39<1:31:46, 1.39s/it] 87%|████████▋ | 26027/30000 [15:47:46<3:31:11, 3.19s/it] 87%|████████▋ | 26028/30000 [15:47:47<2:51:14, 2.59s/it] 87%|████████▋ | 26029/30000 [15:47:49<2:23:45, 2.17s/it] 87%|████████▋ | 26030/30000 [15:47:50<2:04:45, 1.89s/it] 87%|████████▋ | 26030/30000 [15:47:50<2:04:45, 1.89s/it] 87%|████████▋ | 26031/30000 [15:47:51<1:51:32, 1.69s/it] 87%|████████▋ | 26032/30000 [15:47:52<1:42:22, 1.55s/it] 87%|████████▋ | 26033/30000 [15:47:54<1:36:31, 1.46s/it] 87%|████████▋ | 26034/30000 [15:47:55<1:33:25, 1.41s/it] 87%|████████▋ | 26035/30000 [15:48:02<3:19:39, 3.02s/it] 87%|████████▋ | 26036/30000 [15:48:03<2:42:38, 2.46s/it] 87%|████████▋ | 26037/30000 [15:48:04<2:17:19, 2.08s/it] 87%|████████▋ | 26038/30000 [15:48:05<1:59:36, 1.81s/it] 87%|████████▋ | 26039/30000 [15:48:06<1:46:43, 1.62s/it] 87%|████████▋ | 26040/30000 [15:48:08<1:38:53, 1.50s/it] 87%|████████▋ | 26040/30000 [15:48:08<1:38:53, 1.50s/it] 87%|████████▋ | 26041/30000 [15:48:09<1:32:59, 1.41s/it] 87%|████████▋ | 26042/30000 [15:48:10<1:31:26, 1.39s/it] 87%|████████▋ | 26043/30000 [15:48:19<3:53:02, 3.53s/it] 87%|████████▋ | 26044/30000 [15:48:20<3:06:24, 2.83s/it] 87%|████████▋ | 26045/30000 [15:48:21<2:33:47, 2.33s/it] 87%|████████▋ | 26046/30000 [15:48:22<2:10:24, 1.98s/it] 87%|████████▋ | 26047/30000 [15:48:23<1:54:24, 1.74s/it] 87%|████████▋ | 26048/30000 [15:48:24<1:43:39, 1.57s/it] 87%|████████▋ | 26049/30000 [15:48:26<1:36:06, 1.46s/it] 87%|████████▋ | 26050/30000 [15:48:27<1:30:35, 1.38s/it] 87%|████████▋ | 26050/30000 [15:48:27<1:30:35, 1.38s/it] 87%|████████▋ | 26051/30000 [15:48:36<3:57:17, 3.61s/it] 87%|████████▋ | 26052/30000 [15:48:37<3:09:17, 2.88s/it] 87%|████████▋ | 26053/30000 [15:48:38<2:36:37, 2.38s/it] 87%|████████▋ | 26054/30000 [15:48:39<2:12:40, 2.02s/it] 87%|████████▋ | 26055/30000 [15:48:40<1:55:59, 1.76s/it] 87%|████████▋ | 26056/30000 [15:48:42<1:44:21, 1.59s/it] 87%|████████▋ | 26057/30000 [15:48:43<1:36:49, 1.47s/it] 87%|████████▋ | 26058/30000 [15:48:44<1:31:22, 1.39s/it] 87%|████████▋ | 26059/30000 [15:48:52<3:40:24, 3.36s/it] 87%|████████▋ | 26060/30000 [15:48:53<2:57:20, 2.70s/it] 87%|████████▋ | 26060/30000 [15:48:53<2:57:20, 2.70s/it] 87%|████████▋ | 26061/30000 [15:48:54<2:26:57, 2.24s/it] 87%|████████▋ | 26062/30000 [15:48:55<2:06:00, 1.92s/it] 87%|████████▋ | 26063/30000 [15:48:57<1:51:04, 1.69s/it] 87%|████████▋ | 26064/30000 [15:48:58<1:41:03, 1.54s/it] 87%|████████▋ | 26065/30000 [15:48:59<1:34:21, 1.44s/it] 87%|████████▋ | 26066/30000 [15:49:00<1:29:16, 1.36s/it] 87%|████████▋ | 26067/30000 [15:49:09<3:47:00, 3.46s/it] 87%|████████▋ | 26068/30000 [15:49:10<3:01:47, 2.77s/it] 87%|████████▋ | 26069/30000 [15:49:11<2:30:54, 2.30s/it] 87%|████████▋ | 26070/30000 [15:49:12<2:08:36, 1.96s/it] 87%|████████▋ | 26070/30000 [15:49:12<2:08:36, 1.96s/it] 87%|████████▋ | 26071/30000 [15:49:13<1:53:12, 1.73s/it] 87%|████████▋ | 26072/30000 [15:49:14<1:42:32, 1.57s/it] 87%|████████▋ | 26073/30000 [15:49:16<1:35:32, 1.46s/it] 87%|████████▋ | 26074/30000 [15:49:17<1:30:27, 1.38s/it] 87%|████████▋ | 26075/30000 [15:49:25<3:46:39, 3.46s/it] 87%|████████▋ | 26076/30000 [15:49:26<3:01:38, 2.78s/it] 87%|████████▋ | 26077/30000 [15:49:28<2:30:53, 2.31s/it] 87%|████████▋ | 26078/30000 [15:49:29<2:08:43, 1.97s/it] 87%|████████▋ | 26079/30000 [15:49:30<1:53:13, 1.73s/it] 87%|████████▋ | 26080/30000 [15:49:31<1:42:05, 1.56s/it] +{'loss': 0.0131, 'grad_norm': 0.08338239789009094, 'learning_rate': 1.669604147596563e-06, 'flow_matching_loss': 0.013094308972358703, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.47} +{'loss': 0.0127, 'grad_norm': 0.07019822299480438, 'learning_rate': 1.662031045573319e-06, 'flow_matching_loss': 0.012733108550310134, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.47} +{'loss': 0.0128, 'grad_norm': 0.06902701407670975, 'learning_rate': 1.6544741504400402e-06, 'flow_matching_loss': 0.012831385992467404, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.48} +{'loss': 0.0117, 'grad_norm': 0.07583564519882202, 'learning_rate': 1.6469334713790712e-06, 'flow_matching_loss': 0.011672129109501838, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.48} +{'loss': 0.0135, 'grad_norm': 0.08908869326114655, 'learning_rate': 1.639409017553048e-06, 'flow_matching_loss': 0.013492080755531788, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.48} +{'loss': 0.0126, 'grad_norm': 0.05291089415550232, 'learning_rate': 1.6319007981048884e-06, 'flow_matching_loss': 0.012594279460608959, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.49} +{'loss': 0.0122, 'grad_norm': 0.06040407344698906, 'learning_rate': 1.6244088221577896e-06, 'flow_matching_loss': 0.012221523933112622, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.49} +{'loss': 0.0138, 'grad_norm': 0.08165033161640167, 'learning_rate': 1.6169330988152081e-06, 'flow_matching_loss': 0.013818352203816175, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.49} +{'loss': 0.0128, 'grad_norm': 0.07168484479188919, 'learning_rate': 1.6094736371608475e-06, 'flow_matching_loss': 0.012833732645958662, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.49} +{'loss': 0.0128, 'grad_norm': 0.10635149478912354, 'learning_rate': 1.602030446258666e-06, 'flow_matching_loss': 0.012817358877509833, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.5} +{'loss': 0.0128, 'grad_norm': 0.07729887217283249, 'learning_rate': 1.5946035351528388e-06, 'flow_matching_loss': 0.012820358015596866, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.5} +{'loss': 0.0118, 'grad_norm': 0.07085059583187103, 'learning_rate': 1.5871929128677621e-06, 'flow_matching_loss': 0.011788236815482378, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.5} +{'loss': 0.0132, 'grad_norm': 0.08461789041757584, 'learning_rate': 1.5797985884080412e-06, 'flow_matching_loss': 0.013153692241758108, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.51} +{'loss': 0.0131, 'grad_norm': 0.08148098737001419, 'learning_rate': 1.572420570758476e-06, 'flow_matching_loss': 0.013119586743414402, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.51} +{'loss': 0.0148, 'grad_norm': 0.07059703767299652, 'learning_rate': 1.5650588688840567e-06, 'flow_matching_loss': 0.014783224556595087, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.51} +{'loss': 0.0143, 'grad_norm': 0.07809881120920181, 'learning_rate': 1.5577134917299413e-06, 'flow_matching_loss': 0.014253124594688416, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.52} +{'loss': 0.0126, 'grad_norm': 0.08103885501623154, 'learning_rate': 1.5503844482214618e-06, 'flow_matching_loss': 0.012570601049810648, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.52} +{'loss': 0.0121, 'grad_norm': 0.05658887326717377, 'learning_rate': 1.5430717472640932e-06, 'flow_matching_loss': 0.012107066623866559, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.52} +{'loss': 0.0135, 'grad_norm': 0.07372228056192398, 'learning_rate': 1.5357753977434541e-06, 'flow_matching_loss': 0.013507523015141487, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.52} +{'loss': 0.0132, 'grad_norm': 0.0726289302110672, 'learning_rate': 1.5284954085253043e-06, 'flow_matching_loss': 0.013177156448364258, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.53} +{'loss': 0.0129, 'grad_norm': 0.06559240072965622, 'learning_rate': 1.5212317884555115e-06, 'flow_matching_loss': 0.012900643981993198, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.53} +{'loss': 0.0125, 'grad_norm': 0.07608447223901749, 'learning_rate': 1.5139845463600605e-06, 'flow_matching_loss': 0.012548348773270846, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.53} +{'loss': 0.0122, 'grad_norm': 0.07705956697463989, 'learning_rate': 1.5067536910450352e-06, 'flow_matching_loss': 0.012182949110865593, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.54} +{'loss': 0.0105, 'grad_norm': 0.07828722149133682, 'learning_rate': 1.4995392312966044e-06, 'flow_matching_loss': 0.010484562814235687, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.54} +{'loss': 0.0126, 'grad_norm': 0.07037383317947388, 'learning_rate': 1.492341175881014e-06, 'flow_matching_loss': 0.012635072227567434, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.54} +{'loss': 0.0133, 'grad_norm': 0.06791326403617859, 'learning_rate': 1.4851595335445895e-06, 'flow_matching_loss': 0.013254914432764053, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.54} +{'loss': 0.0136, 'grad_norm': 0.06689749658107758, 'learning_rate': 1.4779943130136953e-06, 'flow_matching_loss': 0.013586146710440516, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.55} +{'loss': 0.0118, 'grad_norm': 0.07148636132478714, 'learning_rate': 1.4708455229947537e-06, 'flow_matching_loss': 0.01177665013819933, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.55} +{'loss': 0.0122, 'grad_norm': 0.07021035999059677, 'learning_rate': 1.4637131721742148e-06, 'flow_matching_loss': 0.012249214015901089, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.55} +{'loss': 0.0129, 'grad_norm': 0.05520514398813248, 'learning_rate': 1.4565972692185626e-06, 'flow_matching_loss': 0.012855331040918827, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.56} +{'loss': 0.0124, 'grad_norm': 0.06433721631765366, 'learning_rate': 1.4494978227742855e-06, 'flow_matching_loss': 0.012447944236919284, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.56} +{'loss': 0.0138, 'grad_norm': 0.05606786161661148, 'learning_rate': 1.442414841467885e-06, 'flow_matching_loss': 0.013791683688759803, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.56} +{'loss': 0.0133, 'grad_norm': 0.07977280765771866, 'learning_rate': 1.4353483339058498e-06, 'flow_matching_loss': 0.013292414788156747, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.56} +{'loss': 0.0122, 'grad_norm': 0.0483664870262146, 'learning_rate': 1.4282983086746498e-06, 'flow_matching_loss': 0.012237032130360603, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.57} +{'loss': 0.0111, 'grad_norm': 0.061599250882864, 'learning_rate': 1.4212647743407325e-06, 'flow_matching_loss': 0.011142915533855557, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.57} +{'loss': 0.0131, 'grad_norm': 0.06774485856294632, 'learning_rate': 1.4142477394505065e-06, 'flow_matching_loss': 0.013078386243432761, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.57} +{'loss': 0.0128, 'grad_norm': 0.09033896028995514, 'learning_rate': 1.4072472125303299e-06, 'flow_matching_loss': 0.01283123278990388, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.58} +{'loss': 0.0126, 'grad_norm': 0.06380046904087067, 'learning_rate': 1.4002632020865031e-06, 'flow_matching_loss': 0.01255258172750473, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.58} +{'loss': 0.0128, 'grad_norm': 0.07161778956651688, 'learning_rate': 1.393295716605258e-06, 'flow_matching_loss': 0.012843643315136433, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.58} +{'loss': 0.0131, 'grad_norm': 0.05444419011473656, 'learning_rate': 1.386344764552742e-06, 'flow_matching_loss': 0.013148422073572873, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.59} +{'loss': 0.012, 'grad_norm': 0.07700629532337189, 'learning_rate': 1.3794103543750246e-06, 'flow_matching_loss': 0.012031706795096398, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.59} 87%|████████▋ | 26080/30000 [15:49:31<1:42:05, 1.56s/it] 87%|████████▋ | 26081/30000 [15:49:32<1:34:45, 1.45s/it] 87%|████████▋ | 26082/30000 [15:49:34<1:31:11, 1.40s/it] 87%|████████▋ | 26083/30000 [15:49:40<3:18:16, 3.04s/it] 87%|████████▋ | 26084/30000 [15:49:42<2:41:49, 2.48s/it] 87%|████████▋ | 26085/30000 [15:49:43<2:16:41, 2.09s/it] 87%|████████▋ | 26086/30000 [15:49:44<1:58:58, 1.82s/it] 87%|████████▋ | 26087/30000 [15:49:45<1:46:13, 1.63s/it] 87%|████████▋ | 26088/30000 [15:49:46<1:37:39, 1.50s/it] 87%|████████▋ | 26089/30000 [15:49:48<1:31:29, 1.40s/it] 87%|████████▋ | 26090/30000 [15:49:49<1:27:38, 1.34s/it] 87%|████████▋ | 26090/30000 [15:49:49<1:27:38, 1.34s/it] 87%|████████▋ | 26091/30000 [15:49:56<3:28:32, 3.20s/it] 87%|████████▋ | 26092/30000 [15:49:58<2:51:41, 2.64s/it] 87%|████████▋ | 26093/30000 [15:49:59<2:22:57, 2.20s/it] 87%|████████▋ | 26094/30000 [15:50:00<2:04:53, 1.92s/it] 87%|████████▋ | 26095/30000 [15:50:01<1:50:24, 1.70s/it] 87%|████████▋ | 26096/30000 [15:50:02<1:40:38, 1.55s/it] 87%|████████▋ | 26097/30000 [15:50:04<1:33:29, 1.44s/it] 87%|████████▋ | 26098/30000 [15:50:05<1:28:36, 1.36s/it] 87%|████████▋ | 26099/30000 [15:50:12<3:22:25, 3.11s/it] 87%|████████▋ | 26100/30000 [15:50:13<2:44:58, 2.54s/it] 87%|████████▋ | 26100/30000 [15:50:13<2:44:58, 2.54s/it] 87%|████████▋ | 26101/30000 [15:50:14<2:18:29, 2.13s/it] 87%|████████▋ | 26102/30000 [15:50:16<1:59:52, 1.85s/it] 87%|████████▋ | 26103/30000 [15:50:17<1:46:37, 1.64s/it] 87%|████████▋ | 26104/30000 [15:50:18<1:38:13, 1.51s/it] 87%|████████▋ | 26105/30000 [15:50:19<1:32:13, 1.42s/it] 87%|████████▋ | 26106/30000 [15:50:20<1:27:58, 1.36s/it] 87%|████████▋ | 26107/30000 [15:50:28<3:31:25, 3.26s/it] 87%|████████▋ | 26108/30000 [15:50:29<2:52:06, 2.65s/it] 87%|████████▋ | 26109/30000 [15:50:30<2:24:34, 2.23s/it] 87%|████████▋ | 26110/30000 [15:50:32<2:04:01, 1.91s/it] 87%|████████▋ | 26110/30000 [15:50:32<2:04:01, 1.91s/it] 87%|████████▋ | 26111/30000 [15:50:33<1:49:30, 1.69s/it] 87%|████████▋ | 26112/30000 [15:50:34<1:40:55, 1.56s/it] 87%|████████▋ | 26113/30000 [15:50:35<1:33:46, 1.45s/it] 87%|████████▋ | 26114/30000 [15:50:37<1:31:28, 1.41s/it] 87%|████████▋ | 26115/30000 [15:50:45<3:39:55, 3.40s/it] 87%|████████▋ | 26116/30000 [15:50:46<2:58:03, 2.75s/it] 87%|████████▋ | 26117/30000 [15:50:47<2:28:52, 2.30s/it] 87%|████████▋ | 26118/30000 [15:50:48<2:07:09, 1.97s/it] 87%|████████▋ | 26119/30000 [15:50:50<1:53:51, 1.76s/it] 87%|████████▋ | 26120/30000 [15:50:51<1:42:36, 1.59s/it] 87%|████████▋ | 26120/30000 [15:50:51<1:42:36, 1.59s/it] 87%|████████▋ | 26121/30000 [15:50:52<1:34:43, 1.47s/it] 87%|████████▋ | 26122/30000 [15:50:53<1:29:24, 1.38s/it] 87%|████████▋ | 26123/30000 [15:51:02<4:00:26, 3.72s/it] 87%|████████▋ | 26124/30000 [15:51:03<3:11:19, 2.96s/it] 87%|████████▋ | 26125/30000 [15:51:05<2:36:53, 2.43s/it] 87%|████████▋ | 26126/30000 [15:51:06<2:12:24, 2.05s/it] 87%|████████▋ | 26127/30000 [15:51:07<1:55:30, 1.79s/it] 87%|████████▋ | 26128/30000 [15:51:08<1:44:06, 1.61s/it] 87%|████████▋ | 26129/30000 [15:51:09<1:35:45, 1.48s/it] 87%|████████▋ | 26130/30000 [15:51:11<1:29:28, 1.39s/it] 87%|████████▋ | 26130/30000 [15:51:11<1:29:28, 1.39s/it] 87%|████████▋ | 26131/30000 [15:51:20<3:58:15, 3.69s/it] 87%|████████▋ | 26132/30000 [15:51:21<3:09:33, 2.94s/it] 87%|████████▋ | 26133/30000 [15:51:22<2:35:41, 2.42s/it] 87%|████████▋ | 26134/30000 [15:51:23<2:12:40, 2.06s/it] 87%|████████▋ | 26135/30000 [15:51:24<1:55:21, 1.79s/it] 87%|████████▋ | 26136/30000 [15:51:26<1:44:01, 1.62s/it] 87%|████████▋ | 26137/30000 [15:51:27<1:36:01, 1.49s/it] 87%|████████▋ | 26138/30000 [15:51:28<1:30:39, 1.41s/it] 87%|████████▋ | 26139/30000 [15:51:36<3:36:39, 3.37s/it] 87%|████████▋ | 26140/30000 [15:51:37<2:54:40, 2.72s/it] 87%|████████▋ | 26140/30000 [15:51:37<2:54:40, 2.72s/it] 87%|████████▋ | 26141/30000 [15:51:38<2:25:23, 2.26s/it] 87%|████████▋ | 26142/30000 [15:51:40<2:04:19, 1.93s/it] 87%|████████▋ | 26143/30000 [15:51:41<1:50:32, 1.72s/it] 87%|████████▋ | 26144/30000 [15:51:42<1:39:54, 1.55s/it] 87%|████████▋ | 26145/30000 [15:51:43<1:32:46, 1.44s/it] 87%|████████▋ | 26146/30000 [15:51:44<1:28:33, 1.38s/it] 87%|████████▋ | 26147/30000 [15:51:52<3:26:25, 3.21s/it] 87%|████████▋ | 26148/30000 [15:51:53<2:47:19, 2.61s/it] 87%|████████▋ | 26149/30000 [15:51:54<2:19:52, 2.18s/it] 87%|████████▋ | 26150/30000 [15:51:55<2:00:12, 1.87s/it] 87%|████████▋ | 26150/30000 [15:51:55<2:00:12, 1.87s/it] 87%|████████▋ | 26151/30000 [15:51:57<1:46:58, 1.67s/it] 87%|████████▋ | 26152/30000 [15:51:58<1:37:55, 1.53s/it] 87%|████████▋ | 26153/30000 [15:51:59<1:31:29, 1.43s/it] 87%|████████▋ | 26154/30000 [15:52:00<1:26:57, 1.36s/it] 87%|████████▋ | 26155/30000 [15:52:08<3:25:14, 3.20s/it] 87%|████████▋ | 26156/30000 [15:52:09<2:48:42, 2.63s/it] 87%|████████▋ | 26157/30000 [15:52:10<2:20:47, 2.20s/it] 87%|████████▋ | 26158/30000 [15:52:11<2:01:08, 1.89s/it] 87%|████████▋ | 26159/30000 [15:52:12<1:47:18, 1.68s/it] 87%|████████▋ | 26160/30000 [15:52:14<1:38:06, 1.53s/it] 87%|████████▋ | 26160/30000 [15:52:14<1:38:06, 1.53s/it] 87%|████████▋ | 26161/30000 [15:52:15<1:31:22, 1.43s/it] 87%|████████▋ | 26162/30000 [15:52:16<1:26:46, 1.36s/it] 87%|████████▋ | 26163/30000 [15:52:23<3:04:28, 2.88s/it] 87%|████████▋ | 26164/30000 [15:52:24<2:31:41, 2.37s/it] 87%|████████▋ | 26165/30000 [15:52:25<2:09:03, 2.02s/it] 87%|████████▋ | 26166/30000 [15:52:26<1:53:21, 1.77s/it] 87%|████████▋ | 26167/30000 [15:52:27<1:41:52, 1.59s/it] 87%|████████▋ | 26168/30000 [15:52:28<1:33:41, 1.47s/it] 87%|████████▋ | 26169/30000 [15:52:30<1:28:15, 1.38s/it] 87%|████████▋ | 26170/30000 [15:52:31<1:24:31, 1.32s/it] 87%|████████▋ | 26170/30000 [15:52:31<1:24:31, 1.32s/it] 87%|████████▋ | 26171/30000 [15:52:37<3:04:11, 2.89s/it] 87%|████████▋ | 26172/30000 [15:52:39<2:31:20, 2.37s/it] 87%|████████▋ | 26173/30000 [15:52:40<2:08:54, 2.02s/it] 87%|████████▋ | 26174/30000 [15:52:41<1:52:58, 1.77s/it] 87%|████████▋ | 26175/30000 [15:52:42<1:42:56, 1.61s/it] 87%|████████▋ | 26176/30000 [15:52:43<1:36:52, 1.52s/it] 87%|████████▋ | 26177/30000 [15:52:45<1:30:44, 1.42s/it] 87%|████████▋ | 26178/30000 [15:52:46<1:25:49, 1.35s/it] 87%|████████▋ | 26179/30000 [15:52:52<3:00:28, 2.83s/it] 87%|████████▋ | 26180/30000 [15:52:53<2:28:38, 2.33s/it] 87%|████████▋ | 26180/30000 [15:52:53<2:28:38, 2.33s/it] 87%|████████▋ | 26181/30000 [15:52:55<2:07:12, 2.00s/it] 87%|████████▋ | 26182/30000 [15:52:56<1:52:03, 1.76s/it] 87%|████████▋ | 26183/30000 [15:52:57<1:40:44, 1.58s/it] 87%|████████▋ | 26184/30000 [15:52:58<1:33:30, 1.47s/it] 87%|████████▋ | 26185/30000 [15:52:59<1:29:04, 1.40s/it] 87%|████████▋ | 26186/30000 [15:53:01<1:26:06, 1.35s/it] 87%|████████▋ | 26187/30000 [15:53:08<3:20:45, 3.16s/it] 87%|████████▋ | 26188/30000 [15:53:09<2:42:56, 2.56s/it] 87%|████████▋ | 26189/30000 [15:53:10<2:16:37, 2.15s/it] 87%|████████▋ | 26190/30000 [15:53:11<1:58:04, 1.86s/it] 87%|████████▋ | 26190/30000 [15:53:12<1:58:04, 1.86s/it] 87%|████████▋ | 26191/30000 [15:53:13<1:45:00, 1.65s/it] 87%|████████▋ | 26192/30000 [15:53:14<1:35:50, 1.51s/it] 87%|████████▋ | 26193/30000 [15:53:15<1:30:00, 1.42s/it] 87%|████████▋ | 26194/30000 [15:53:16<1:27:38, 1.38s/it] 87%|████████▋ | 26195/30000 [15:53:25<3:49:41, 3.62s/it] 87%|████████▋ | 26196/30000 [15:53:26<3:03:09, 2.89s/it] 87%|████████▋ | 26197/30000 [15:53:28<2:31:23, 2.39s/it] 87%|████████▋ | 26198/30000 [15:53:29<2:08:14, 2.02s/it] 87%|████████▋ | 26199/30000 [15:53:30<1:51:57, 1.77s/it] 87%|████████▋ | 26200/30000 [15:53:31<1:42:07, 1.61s/it] 87%|████████▋ | 26200/30000 [15:53:31<1:42:07, 1.61s/it] 87%|████████▋ | 26201/30000 [15:53:33<1:37:09, 1.53s/it] 87%|████████▋ | 26202/30000 [15:53:34<1:31:58, 1.45s/it] 87%|████████▋ | 26203/30000 [15:53:42<3:43:00, 3.52s/it] 87%|████████▋ | 26204/30000 [15:53:43<2:58:16, 2.82s/it] 87%|████████▋ | 26205/30000 [15:53:45<2:27:50, 2.34s/it] 87%|████████▋ | 26206/30000 [15:53:46<2:07:04, 2.01s/it] 87%|████████▋ | 26207/30000 [15:53:47<1:51:08, 1.76s/it] 87%|████████▋ | 26208/30000 [15:53:48<1:40:03, 1.58s/it] 87%|████████▋ | 26209/30000 [15:53:49<1:32:02, 1.46s/it] 87%|████████▋ | 26210/30000 [15:53:50<1:27:20, 1.38s/it] 87%|████████▋ | 26210/30000 [15:53:51<1:27:20, 1.38s/it] 87%|████████▋ | 26211/30000 [15:53:58<3:27:28, 3.29s/it] 87%|████████▋ | 26212/30000 [15:53:59<2:47:23, 2.65s/it] 87%|████████▋ | 26213/30000 [15:54:01<2:19:21, 2.21s/it] 87%|████████▋ | 26214/30000 [15:54:02<1:59:41, 1.90s/it] 87%|████████▋ | 26215/30000 [15:54:03<1:45:58, 1.68s/it] 87%|████████▋ | 26216/30000 [15:54:04<1:36:30, 1.53s/it] 87%|████████▋ | 26217/30000 [15:54:05<1:30:20, 1.43s/it] 87%|████████▋ | 26218/30000 [15:54:07<1:27:47, 1.39s/it] 87%|████████▋ | 26219/30000 [15:54:15<3:36:00, 3.43s/it] 87%|████████▋ | 26220/30000 [15:54:16<2:54:01, 2.76s/it] 87%|████████▋ | 26220/30000 [15:54:16<2:54:01, 2.76s/it] 87%|████████▋ | 26221/30000 [15:54:17<2:24:37, 2.30s/it] 87%|████████▋ | 26222/30000 [15:54:18<2:05:19, 1.99s/it] 87%|████████▋ | 26223/30000 [15:54:20<1:50:10, 1.75s/it] 87%|████████▋ | 26224/30000 [15:54:21<1:39:59, 1.59s/it] 87%|████████▋ | 26225/30000 [15:54:22<1:32:38, 1.47s/it] 87%|████████▋ | 26226/30000 [15:54:23<1:27:04, 1.38s/it] 87%|████████▋ | 26227/30000 [15:54:31<3:35:08, 3.42s/it] 87%|████████▋ | 26228/30000 [15:54:33<2:52:33, 2.74s/it] 87%|████████▋ | 26229/30000 [15:54:34<2:23:28, 2.28s/it] 87%|████████▋ | 26230/30000 [15:54:35<2:02:58, 1.96s/it] 87%|████████▋ | 26230/30000 [15:54:35<2:02:58, 1.96s/it] 87%|████████▋ | 26231/30000 [15:54:36<1:48:30, 1.73s/it] 87%|████████▋ | 26232/30000 [15:54:37<1:38:46, 1.57s/it] 87%|████████▋ | 26233/30000 [15:54:39<1:31:59, 1.47s/it] 87%|████████▋ | 26234/30000 [15:54:40<1:27:22, 1.39s/it] 87%|████████▋ | 26235/30000 [15:54:47<3:25:08, 3.27s/it] 87%|████████▋ | 26236/30000 [15:54:49<2:45:33, 2.64s/it] 87%|████████▋ | 26237/30000 [15:54:50<2:19:05, 2.22s/it] 87%|████████▋ | 26238/30000 [15:54:51<1:59:29, 1.91s/it] 87%|████████▋ | 26239/30000 [15:54:52<1:46:02, 1.69s/it] 87%|████████▋ | 26240/30000 [15:54:53<1:36:28, 1.54s/it] 87%|████████▋ | 26240/30000 [15:54:53<1:36:28, 1.54s/it] 87%|████████▋ | 26241/30000 [15:54:55<1:29:39, 1.43s/it] 87%|████████▋ | 26242/30000 [15:54:56<1:24:57, 1.36s/it] 87%|████████▋ | 26243/30000 [15:55:05<3:53:31, 3.73s/it] 87%|████████▋ | 26244/30000 [15:55:06<3:05:15, 2.96s/it] 87%|████████▋ | 26245/30000 [15:55:07<2:31:58, 2.43s/it] 87%|████████▋ | 26246/30000 [15:55:09<2:09:07, 2.06s/it] 87%|████████▋ | 26247/30000 [15:55:10<1:52:37, 1.80s/it] 87%|████████▋ | 26248/30000 [15:55:11<1:41:12, 1.62s/it] 87%|████████▋ | 26249/30000 [15:55:12<1:33:17, 1.49s/it] 88%|████████▊ | 26250/30000 [15:55:13<1:27:56, 1.41s/it] 88%|████████▊ | 26250/30000 [15:55:13<1:27:56, 1.41s/it] 88%|████████▊ | 26251/30000 [15:55:21<3:19:07, 3.19s/it] 88%|████████▊ | 26252/30000 [15:55:22<2:41:27, 2.58s/it] 88%|████████▊ | 26253/30000 [15:55:23<2:17:33, 2.20s/it] 88%|████████▊ | 26254/30000 [15:55:24<1:57:47, 1.89s/it] 88%|████████▊ | 26255/30000 [15:55:26<1:44:16, 1.67s/it] 88%|████████▊ | 26256/30000 [15:55:27<1:35:17, 1.53s/it] 88%|████████▊ | 26257/30000 [15:55:28<1:29:01, 1.43s/it] 88%|████████▊ | 26258/30000 [15:55:29<1:24:36, 1.36s/it] 88%|████████▊ | 26259/30000 [15:55:36<3:12:39, 3.09s/it] 88%|████████▊ | 26260/30000 [15:55:37<2:36:41, 2.51s/it] 88%|████████▊ | 26260/30000 [15:55:37<2:36:41, 2.51s/it] 88%|████████▊ | 26261/30000 [15:55:39<2:12:35, 2.13s/it] 88%|████████▊ | 26262/30000 [15:55:40<1:55:04, 1.85s/it] 88%|████████▊ | 26263/30000 [15:55:41<1:42:36, 1.65s/it] 88%|████████▊ | 26264/30000 [15:55:42<1:34:34, 1.52s/it] 88%|████████▊ | 26265/30000 [15:55:43<1:29:01, 1.43s/it] 88%|████████▊ | 26266/30000 [15:55:45<1:24:38, 1.36s/it] 88%|████████▊ | 26267/30000 [15:55:52<3:22:57, 3.26s/it] 88%|████████▊ | 26268/30000 [15:55:54<2:44:40, 2.65s/it] 88%|████████▊ | 26269/30000 [15:55:55<2:18:24, 2.23s/it] 88%|████████▊ | 26270/30000 [15:55:56<1:58:43, 1.91s/it] 88%|████████▊ | 26270/30000 [15:55:56<1:58:43, 1.91s/it] 88%|████████▊ | 26271/30000 [15:55:57<1:45:20, 1.69s/it] 88%|████████▊ | 26272/30000 [15:55:58<1:35:48, 1.54s/it] 88%|████████▊ | 26273/30000 [15:56:00<1:29:34, 1.44s/it] 88%|████████▊ | 26274/30000 [15:56:01<1:24:45, 1.36s/it] 88%|████████▊ | 26275/30000 [15:56:09<3:23:22, 3.28s/it] 88%|████████▊ | 26276/30000 [15:56:10<2:44:25, 2.65s/it] 88%|████████▊ | 26277/30000 [15:56:11<2:18:33, 2.23s/it] 88%|████████▊ | 26278/30000 [15:56:12<1:58:38, 1.91s/it] 88%|████████▊ | 26279/30000 [15:56:13<1:44:51, 1.69s/it] 88%|████████▊ | 26280/30000 [15:56:15<1:35:33, 1.54s/it] 88%|████████▊ | 26280/30000 [15:56:15<1:35:33, 1.54s/it] 88%|████████▊ | 26281/30000 [15:56:16<1:28:56, 1.44s/it] 88%|████████▊ | 26282/30000 [15:56:17<1:24:08, 1.36s/it] 88%|████████▊ | 26283/30000 [15:56:24<3:16:20, 3.17s/it] 88%|████████▊ | 26284/30000 [15:56:25<2:39:04, 2.57s/it] 88%|████████▊ | 26285/30000 [15:56:27<2:14:20, 2.17s/it] 88%|████████▊ | 26286/30000 [15:56:28<1:56:56, 1.89s/it] 88%|████████▊ | 26287/30000 [15:56:29<1:43:33, 1.67s/it] 88%|████████▊ | 26288/30000 [15:56:30<1:35:07, 1.54s/it] 88%|████████▊ | 26289/30000 [15:56:31<1:28:45, 1.44s/it] 88%|████████▊ | 26290/30000 [15:56:33<1:24:18, 1.36s/it] 88%|████████▊ | 26290/30000 [15:56:33<1:24:18, 1.36s/it] 88%|████████▊ | 26291/30000 [15:56:41<3:35:36, 3.49s/it] 88%|████████▊ | 26292/30000 [15:56:42<2:52:35, 2.79s/it] 88%|████████▊ | 26293/30000 [15:56:44<2:23:31, 2.32s/it] 88%|████████▊ | 26294/30000 [15:56:45<2:02:50, 1.99s/it] 88%|████████▊ | 26295/30000 [15:56:46<1:48:31, 1.76s/it] 88%|████████▊ | 26296/30000 [15:56:47<1:38:43, 1.60s/it] 88%|████████▊ | 26297/30000 [15:56:48<1:31:45, 1.49s/it] 88%|████████▊ | 26298/30000 [15:56:50<1:26:19, 1.40s/it] 88%|████████▊ | 26299/30000 [15:56:57<3:13:31, 3.14s/it] 88%|████████▊ | 26300/30000 [15:56:58<2:37:47, 2.56s/it] 88%|████████▊ | 26300/30000 [15:56:58<2:37:47, 2.56s/it] 88%|████████▊ | 26301/30000 [15:56:59<2:13:30, 2.17s/it] 88%|████████▊ | 26302/30000 [15:57:00<1:55:19, 1.87s/it] 88%|████████▊ | 26303/30000 [15:57:02<1:42:35, 1.66s/it] 88%|████████▊ | 26304/30000 [15:57:03<1:33:27, 1.52s/it] 88%|████████▊ | 26305/30000 [15:57:04<1:27:07, 1.41s/it] 88%|████████▊ | 26306/30000 [15:57:05<1:23:21, 1.35s/it] 88%|████████▊ | 26307/30000 [15:57:13<3:27:54, 3.38s/it] 88%|████████▊ | 26308/30000 [15:57:14<2:47:10, 2.72s/it] 88%|████████▊ | 26309/30000 [15:57:16<2:18:21, 2.25s/it] 88%|████████▊ | 26310/30000 [15:57:17<1:58:48, 1.93s/it] 88%|████████▊ | 26310/30000 [15:57:17<1:58:48, 1.93s/it] 88%|████████▊ | 26311/30000 [15:57:18<1:44:53, 1.71s/it] 88%|████████▊ | 26312/30000 [15:57:19<1:35:30, 1.55s/it] 88%|████████▊ | 26313/30000 [15:57:20<1:29:11, 1.45s/it] 88%|████████▊ | 26314/30000 [15:57:22<1:24:09, 1.37s/it] 88%|████████▊ | 26315/30000 [15:57:30<3:28:13, 3.39s/it] 88%|████████▊ | 26316/30000 [15:57:31<2:47:50, 2.73s/it] 88%|████████▊ | 26317/30000 [15:57:32<2:20:23, 2.29s/it] 88%|████████▊ | 26318/30000 [15:57:33<2:00:34, 1.96s/it] 88%|████████▊ | 26319/30000 [15:57:35<1:46:17, 1.73s/it] 88%|████████▊ | 26320/30000 [15:57:36<1:36:24, 1.57s/it] 88%|████████▊ | 26320/30000 [15:57:36<1:36:24, 1.57s/it] 88%|████████▊ | 26321/30000 [15:57:37<1:30:12, 1.47s/it] 88%|████████▊ | 26322/30000 [15:57:38<1:25:28, 1.39s/it] 88%|████████▊ | 26323/30000 [15:57:45<3:00:47, 2.95s/it] 88%|████████▊ | 26324/30000 [15:57:46<2:29:13, 2.44s/it] 88%|████████▊ | 26325/30000 [15:57:47<2:06:29, 2.07s/it] 88%|████████▊ | 26326/30000 [15:57:48<1:50:21, 1.80s/it] 88%|████████▊ | 26327/30000 [15:57:50<1:38:51, 1.61s/it] 88%|████████▊ | 26328/30000 [15:57:51<1:30:48, 1.48s/it] 88%|████████▊ | 26329/30000 [15:57:52<1:25:42, 1.40s/it] 88%|████████▊ | 26330/30000 [15:57:53<1:21:48, 1.34s/it] 88%|████████▊ | 26330/30000 [15:57:53<1:21:48, 1.34s/it] 88%|████████▊ | 26331/30000 [15:58:00<3:10:29, 3.12s/it] 88%|████████▊ | 26332/30000 [15:58:02<2:35:28, 2.54s/it] 88%|████████▊ | 26333/30000 [15:58:03<2:12:07, 2.16s/it] 88%|████████▊ | 26334/30000 [15:58:04<1:54:24, 1.87s/it] 88%|████████▊ | 26335/30000 [15:58:05<1:41:27, 1.66s/it] 88%|████████▊ | 26336/30000 [15:58:06<1:32:41, 1.52s/it] 88%|████████▊ | 26337/30000 [15:58:08<1:26:48, 1.42s/it] 88%|████████▊ | 26338/30000 [15:58:09<1:23:03, 1.36s/it] 88%|████████▊ | 26339/30000 [15:58:18<3:53:11, 3.82s/it] 88%|████████▊ | 26340/30000 [15:58:20<3:04:42, 3.03s/it] 88%|████████▊ | 26340/30000 [15:58:20<3:04:42, 3.03s/it] 88%|████████▊ | 26341/30000 [15:58:21<2:31:35, 2.49s/it] 88%|████████▊ | 26342/30000 [15:58:22<2:07:17, 2.09s/it] 88%|████████▊ | 26343/30000 [15:58:23<1:50:32, 1.81s/it] 88%|████████▊ | 26344/30000 [15:58:24<1:39:17, 1.63s/it] 88%|████████▊ | 26345/30000 [15:58:26<1:32:15, 1.51s/it] 88%|████████▊ | 26346/30000 [15:58:27<1:26:16, 1.42s/it] 88%|████████▊ | 26347/30000 [15:58:34<3:19:12, 3.27s/it] 88%|████████▊ | 26348/30000 [15:58:36<2:40:59, 2.64s/it] 88%|████████▊ | 26349/30000 [15:58:37<2:15:49, 2.23s/it] 88%|████████▊ | 26350/30000 [15:58:38<1:56:30, 1.92s/it] 88%|████████▊ | 26350/30000 [15:58:38<1:56:30, 1.92s/it] 88%|████████▊ | 26351/30000 [15:58:39<1:44:04, 1.71s/it] 88%|████████▊ | 26352/30000 [15:58:40<1:35:11, 1.57s/it] 88%|████████▊ | 26353/30000 [15:58:42<1:30:29, 1.49s/it] 88%|████████▊ | 26354/30000 [15:58:43<1:25:25, 1.41s/it] 88%|████████▊ | 26355/30000 [15:58:51<3:25:45, 3.39s/it] 88%|████████▊ | 26356/30000 [15:58:52<2:45:11, 2.72s/it] 88%|████████▊ | 26357/30000 [15:58:53<2:17:57, 2.27s/it] 88%|████████▊ | 26358/30000 [15:58:55<1:57:58, 1.94s/it] 88%|████████▊ | 26359/30000 [15:58:56<1:44:01, 1.71s/it] 88%|████████▊ | 26360/30000 [15:58:57<1:34:28, 1.56s/it] 88%|████████▊ | 26360/30000 [15:58:57<1:34:28, 1.56s/it] 88%|████████▊ | 26361/30000 [15:58:58<1:28:16, 1.46s/it] 88%|████████▊ | 26362/30000 [15:58:59<1:24:24, 1.39s/it] 88%|████████▊ | 26363/30000 [15:59:07<3:20:56, 3.32s/it] 88%|████████▊ | 26364/30000 [15:59:08<2:42:02, 2.67s/it] 88%|████████▊ | 26365/30000 [15:59:10<2:14:34, 2.22s/it] 88%|████████▊ | 26366/30000 [15:59:11<1:55:31, 1.91s/it] 88%|████████▊ | 26367/30000 [15:59:12<1:42:08, 1.69s/it] 88%|████████▊ | 26368/30000 [15:59:13<1:33:07, 1.54s/it] 88%|████████▊ | 26369/30000 [15:59:14<1:27:32, 1.45s/it] 88%|████████▊ | 26370/30000 [15:59:16<1:23:14, 1.38s/it] 88%|████████▊ | 26370/30000 [15:59:16<1:23:14, 1.38s/it] 88%|████████▊ | 26371/30000 [15:59:23<3:18:35, 3.28s/it] 88%|████████▊ | 26372/30000 [15:59:24<2:40:39, 2.66s/it] 88%|████████▊ | 26373/30000 [15:59:26<2:14:13, 2.22s/it] 88%|████████▊ | 26374/30000 [15:59:27<1:56:05, 1.92s/it] 88%|████████▊ | 26375/30000 [15:59:28<1:42:51, 1.70s/it] 88%|████████▊ | 26376/30000 [15:59:29<1:33:40, 1.55s/it] 88%|████████▊ | 26377/30000 [15:59:30<1:27:01, 1.44s/it] 88%|████████▊ | 26378/30000 [15:59:32<1:22:08, 1.36s/it] 88%|████████▊ | 26379/30000 [15:59:40<3:24:04, 3.38s/it] 88%|████████▊ | 26380/30000 [15:59:41<2:43:43, 2.71s/it] 88%|████████▊ | 26380/30000 [15:59:41<2:43:43, 2.71s/it] 88%|████████▊ | 26381/30000 [15:59:42<2:16:28, 2.26s/it] 88%|████████▊ | 26382/30000 [15:59:43<1:56:46, 1.94s/it] 88%|████████▊ | 26383/30000 [15:59:44<1:42:57, 1.71s/it] 88%|████████▊ | 26384/30000 [15:59:46<1:33:32, 1.55s/it] 88%|████████▊ | 26385/30000 [15:59:47<1:27:14, 1.45s/it] 88%|████████▊ | 26386/30000 [15:59:48<1:22:20, 1.37s/it] 88%|████████▊ | 26387/30000 [15:59:57<3:40:53, 3.67s/it] 88%|████████▊ | 26388/30000 [15:59:58<2:56:00, 2.92s/it] 88%|████████▊ | 26389/30000 [15:59:59<2:24:27, 2.40s/it] 88%|████████▊ | 26390/30000 [16:00:01<2:02:07, 2.03s/it] 88%|████████▊ | 26390/30000 [16:00:01<2:02:07, 2.03s/it] 88%|████████▊ | 26391/30000 [16:00:02<1:46:29, 1.77s/it] 88%|████████▊ | 26392/30000 [16:00:03<1:35:45, 1.59s/it] 88%|████████▊ | 26393/30000 [16:00:04<1:28:47, 1.48s/it] 88%|████████▊ | 26394/30000 [16:00:05<1:23:50, 1.40s/it] 88%|████████▊ | 26395/30000 [16:00:13<3:16:11, 3.27s/it] 88%|████████▊ | 26396/30000 [16:00:14<2:38:28, 2.64s/it] 88%|████████▊ | 26397/30000 [16:00:15<2:12:08, 2.20s/it] 88%|████████▊ | 26398/30000 [16:00:16<1:53:34, 1.89s/it] 88%|████████▊ | 26399/30000 [16:00:18<1:40:22, 1.67s/it] 88%|████████▊ | 26400/30000 [16:00:19<1:31:25, 1.52s/it] 88%|████████▊ | 26400/30000 [16:00:19<1:31:25, 1.52s/it] 88%|████████▊ | 26401/30000 [16:00:20<1:25:45, 1.43s/it] 88%|████████▊ | 26402/30000 [16:00:21<1:21:14, 1.35s/it] 88%|████████▊ | 26403/30000 [16:00:28<3:01:51, 3.03s/it] 88%|████████▊ | 26404/30000 [16:00:29<2:28:47, 2.48s/it] 88%|████████▊ | 26405/30000 [16:00:31<2:05:46, 2.10s/it] 88%|████████▊ | 26406/30000 [16:00:32<1:50:02, 1.84s/it] 88%|████████▊ | 26407/30000 [16:00:33<1:38:07, 1.64s/it] 88%|████████▊ | 26408/30000 [16:00:34<1:29:52, 1.50s/it] 88%|████████▊ | 26409/30000 [16:00:35<1:24:25, 1.41s/it] 88%|████████▊ | 26410/30000 [16:00:37<1:19:59, 1.34s/it] 88%|████████▊ | 26410/30000 [16:00:37<1:19:59, 1.34s/it] 88%|████████▊ | 26411/30000 [16:00:45<3:27:42, 3.47s/it] 88%|████████▊ | 26412/30000 [16:00:46<2:45:59, 2.78s/it] 88%|████████▊ | 26413/30000 [16:00:47<2:18:05, 2.31s/it] 88%|████████▊ | 26414/30000 [16:00:49<1:58:06, 1.98s/it] 88%|████████▊ | 26415/30000 [16:00:50<1:43:53, 1.74s/it] 88%|████████▊ | 26416/30000 [16:00:51<1:34:19, 1.58s/it] 88%|████████▊ | 26417/30000 [16:00:52<1:27:24, 1.46s/it] 88%|████████▊ | 26418/30000 [16:00:53<1:24:57, 1.42s/it] 88%|████████▊ | 26419/30000 [16:01:01<3:07:20, 3.14s/it] 88%|████████▊ | 26420/30000 [16:01:02<2:33:09, 2.57s/it] 88%|████████▊ | 26420/30000 [16:01:02<2:33:09, 2.57s/it] 88%|████████▊ | 26421/30000 [16:01:03<2:09:32, 2.17s/it] 88%|████████▊ | 26422/30000 [16:01:04<1:51:23, 1.87s/it] 88%|████████▊ | 26423/30000 [16:01:05<1:38:52, 1.66s/it] 88%|████████▊ | 26424/30000 [16:01:07<1:30:55, 1.53s/it] 88%|████████▊ | 26425/30000 [16:01:08<1:25:30, 1.43s/it] 88%|████████▊ | 26426/30000 [16:01:09<1:21:28, 1.37s/it] 88%|████████▊ | 26427/30000 [16:01:17<3:22:19, 3.40s/it] 88%|████████▊ | 26428/30000 [16:01:18<2:43:27, 2.75s/it] 88%|████████▊ | 26429/30000 [16:01:20<2:15:56, 2.28s/it] 88%|████████▊ | 26430/30000 [16:01:21<1:55:59, 1.95s/it] 88%|████████▊ | 26430/30000 [16:01:21<1:55:59, 1.95s/it] 88%|████████▊ | 26431/30000 [16:01:22<1:41:52, 1.71s/it] 88%|████████▊ | 26432/30000 [16:01:23<1:32:29, 1.56s/it] 88%|████████▊ | 26433/30000 [16:01:24<1:25:45, 1.44s/it] 88%|████████▊ | 26434/30000 [16:01:26<1:21:34, 1.37s/it] 88%|████████▊ | 26435/30000 [16:01:33<3:09:32, 3.19s/it] 88%|████████▊ | 26436/30000 [16:01:34<2:34:46, 2.61s/it] 88%|████████▊ | 26437/30000 [16:01:35<2:09:41, 2.18s/it] 88%|████████▊ | 26438/30000 [16:01:37<1:51:27, 1.88s/it] 88%|████████▊ | 26439/30000 [16:01:38<1:39:11, 1.67s/it] 88%|████████▊ | 26440/30000 [16:01:39<1:30:45, 1.53s/it] 88%|████████▊ | 26440/30000 [16:01:39<1:30:45, 1.53s/it] 88%|████████▊ | 26441/30000 [16:01:40<1:24:59, 1.43s/it] 88%|████████▊ | 26442/30000 [16:01:41<1:20:40, 1.36s/it] 88%|████████▊ | 26443/30000 [16:01:49<3:11:41, 3.23s/it] 88%|████████▊ | 26444/30000 [16:01:50<2:36:49, 2.65s/it] 88%|████████▊ | 26445/30000 [16:01:51<2:10:26, 2.20s/it] 88%|████████▊ | 26446/30000 [16:01:53<1:52:09, 1.89s/it] 88%|████████▊ | 26447/30000 [16:01:54<1:39:14, 1.68s/it] 88%|████████▊ | 26448/30000 [16:01:55<1:30:23, 1.53s/it] 88%|████████▊ | 26449/30000 [16:01:56<1:23:57, 1.42s/it] 88%|████████▊ | 26450/30000 [16:01:57<1:20:09, 1.35s/it] 88%|████████▊ | 26450/30000 [16:01:57<1:20:09, 1.35s/it] 88%|████████▊ | 26451/30000 [16:02:05<3:12:11, 3.25s/it] 88%|████████▊ | 26452/30000 [16:02:06<2:36:26, 2.65s/it] 88%|████████▊ | 26453/30000 [16:02:07<2:10:38, 2.21s/it] 88%|████████▊ | 26454/30000 [16:02:09<1:53:05, 1.91s/it] 88%|████████▊ | 26455/30000 [16:02:10<1:39:53, 1.69s/it] 88%|████████▊ | 26456/30000 [16:02:11<1:30:43, 1.54s/it] 88%|████████▊ | 26457/30000 [16:02:12<1:24:25, 1.43s/it] 88%|████████▊ | 26458/30000 [16:02:13<1:20:11, 1.36s/it] 88%|████████▊ | 26459/30000 [16:02:21<3:18:46, 3.37s/it] 88%|████████▊ | 26460/30000 [16:02:23<2:39:50, 2.71s/it] 88%|████████▊ | 26460/30000 [16:02:23<2:39:50, 2.71s/it] 88%|████████▊ | 26461/30000 [16:02:24<2:14:45, 2.28s/it] 88%|████████▊ | 26462/30000 [16:02:25<1:55:01, 1.95s/it] 88%|████████▊ | 26463/30000 [16:02:26<1:41:14, 1.72s/it] 88%|████████▊ | 26464/30000 [16:02:27<1:31:21, 1.55s/it] 88%|████████▊ | 26465/30000 [16:02:29<1:24:33, 1.44s/it] 88%|████████▊ | 26466/30000 [16:02:30<1:20:08, 1.36s/it] 88%|████████▊ | 26467/30000 [16:02:37<3:12:10, 3.26s/it] 88%|████████▊ | 26468/30000 [16:02:39<2:37:38, 2.68s/it] 88%|████████▊ | 26469/30000 [16:02:40<2:13:14, 2.26s/it] 88%|████████▊ | 26470/30000 [16:02:41<1:54:05, 1.94s/it] 88%|████████▊ | 26470/30000 [16:02:41<1:54:05, 1.94s/it] 88%|████████▊ | 26471/30000 [16:02:42<1:40:52, 1.72s/it] 88%|████████▊ | 26472/30000 [16:02:44<1:31:25, 1.55s/it] 88%|████████▊ | 26473/30000 [16:02:45<1:25:14, 1.45s/it] 88%|████████▊ | 26474/30000 [16:02:46<1:23:38, 1.42s/it] 88%|████████▊ | 26475/30000 [16:02:54<3:17:41, 3.37s/it] 88%|████████▊ | 26476/30000 [16:02:55<2:39:34, 2.72s/it] 88%|████████▊ | 26477/30000 [16:02:56<2:13:40, 2.28s/it] 88%|████████▊ | 26478/30000 [16:02:58<1:54:25, 1.95s/it] 88%|████████▊ | 26479/30000 [16:02:59<1:42:36, 1.75s/it] 88%|████████▊ | 26480/30000 [16:03:00<1:33:13, 1.59s/it] 88%|████████▊ | 26480/30000 [16:03:00<1:33:13, 1.59s/it] 88%|████████▊ | 26481/30000 [16:03:01<1:26:01, 1.47s/it] 88%|████████▊ | 26482/30000 [16:03:03<1:21:09, 1.38s/it] 88%|████████▊ | 26483/30000 [16:03:11<3:21:08, 3.43s/it] 88%|████████▊ | 26484/30000 [16:03:12<2:41:55, 2.76s/it] 88%|████████▊ | 26485/30000 [16:03:13<2:15:57, 2.32s/it] 88%|████████▊ | 26486/30000 [16:03:14<1:56:18, 1.99s/it] 88%|████████▊ | 26487/30000 [16:03:16<1:42:06, 1.74s/it] 88%|████████▊ | 26488/30000 [16:03:17<1:32:19, 1.58s/it] 88%|████████▊ | 26489/30000 [16:03:18<1:25:33, 1.46s/it] 88%|████████▊ | 26490/30000 [16:03:19<1:20:28, 1.38s/it] +{'loss': 0.012, 'grad_norm': 0.06295322626829147, 'learning_rate': 1.3724924944980637e-06, 'flow_matching_loss': 0.012030580546706915, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.59} +{'loss': 0.0118, 'grad_norm': 0.049774814397096634, 'learning_rate': 1.365591193327712e-06, 'flow_matching_loss': 0.01177627956494689, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.59} +{'loss': 0.0139, 'grad_norm': 0.06342910975217819, 'learning_rate': 1.3587064592496984e-06, 'flow_matching_loss': 0.01392419608309865, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.6} +{'loss': 0.0143, 'grad_norm': 0.09233888983726501, 'learning_rate': 1.3518383006296292e-06, 'flow_matching_loss': 0.014274237211793661, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.6} +{'loss': 0.0116, 'grad_norm': 0.0687977597117424, 'learning_rate': 1.344986725812959e-06, 'flow_matching_loss': 0.011621550004929305, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.6} +{'loss': 0.0119, 'grad_norm': 0.14373205602169037, 'learning_rate': 1.3381517431250046e-06, 'flow_matching_loss': 0.01190667850896716, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.61} +{'loss': 0.0128, 'grad_norm': 0.055491507053375244, 'learning_rate': 1.3313333608709095e-06, 'flow_matching_loss': 0.012780462950468063, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.61} +{'loss': 0.0132, 'grad_norm': 0.08118662983179092, 'learning_rate': 1.3245315873356528e-06, 'flow_matching_loss': 0.013227985426783561, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.61} +{'loss': 0.0124, 'grad_norm': 0.07563182711601257, 'learning_rate': 1.3177464307840304e-06, 'flow_matching_loss': 0.01240280456840992, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.61} +{'loss': 0.0121, 'grad_norm': 0.05798719823360443, 'learning_rate': 1.310977899460642e-06, 'flow_matching_loss': 0.012079085130244494, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.62} +{'loss': 0.0131, 'grad_norm': 0.07265697419643402, 'learning_rate': 1.3042260015899006e-06, 'flow_matching_loss': 0.01311165578663349, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.62} +{'loss': 0.0139, 'grad_norm': 0.08371012657880783, 'learning_rate': 1.2974907453759948e-06, 'flow_matching_loss': 0.01389027489349246, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.62} +{'loss': 0.0139, 'grad_norm': 0.07087790220975876, 'learning_rate': 1.2907721390028937e-06, 'flow_matching_loss': 0.013912614714354277, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.63} +{'loss': 0.0135, 'grad_norm': 0.070008784532547, 'learning_rate': 1.2840701906343417e-06, 'flow_matching_loss': 0.013536378741264343, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.63} +{'loss': 0.0137, 'grad_norm': 0.0680910050868988, 'learning_rate': 1.2773849084138312e-06, 'flow_matching_loss': 0.013747624401003123, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.63} +{'loss': 0.0115, 'grad_norm': 0.07855898886919022, 'learning_rate': 1.2707163004646177e-06, 'flow_matching_loss': 0.011536851711571216, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.63} +{'loss': 0.0122, 'grad_norm': 0.0681988000869751, 'learning_rate': 1.2640643748896879e-06, 'flow_matching_loss': 0.012234514066949486, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.64} +{'loss': 0.0125, 'grad_norm': 0.05895388498902321, 'learning_rate': 1.257429139771753e-06, 'flow_matching_loss': 0.012548262905329465, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.64} +{'loss': 0.013, 'grad_norm': 0.07444070279598236, 'learning_rate': 1.2508106031732553e-06, 'flow_matching_loss': 0.012987198401242494, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.64} +{'loss': 0.0124, 'grad_norm': 0.061005447059869766, 'learning_rate': 1.2442087731363366e-06, 'flow_matching_loss': 0.012397184036672115, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.65} +{'loss': 0.0133, 'grad_norm': 0.08798597007989883, 'learning_rate': 1.237623657682847e-06, 'flow_matching_loss': 0.013311865646392107, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.65} +{'loss': 0.0128, 'grad_norm': 0.09124507009983063, 'learning_rate': 1.2310552648143208e-06, 'flow_matching_loss': 0.012807992938905954, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.65} +{'loss': 0.0123, 'grad_norm': 0.0875726267695427, 'learning_rate': 1.2245036025119738e-06, 'flow_matching_loss': 0.012266082037240267, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.65} +{'loss': 0.0126, 'grad_norm': 0.09540621936321259, 'learning_rate': 1.2179686787366962e-06, 'flow_matching_loss': 0.012627287488430738, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.66} +{'loss': 0.0142, 'grad_norm': 0.08137194812297821, 'learning_rate': 1.2114505014290329e-06, 'flow_matching_loss': 0.014216389041393995, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.66} +{'loss': 0.0144, 'grad_norm': 0.06338533759117126, 'learning_rate': 1.2049490785091833e-06, 'flow_matching_loss': 0.014367375895380974, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.66} +{'loss': 0.0142, 'grad_norm': 0.06757400929927826, 'learning_rate': 1.1984644178769949e-06, 'flow_matching_loss': 0.014211819786578418, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.67} +{'loss': 0.0128, 'grad_norm': 0.06758058071136475, 'learning_rate': 1.1919965274119381e-06, 'flow_matching_loss': 0.012798390164971352, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.67} +{'loss': 0.0127, 'grad_norm': 0.07073196023702621, 'learning_rate': 1.18554541497311e-06, 'flow_matching_loss': 0.012690815702080726, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.67} +{'loss': 0.0126, 'grad_norm': 0.06943254917860031, 'learning_rate': 1.1791110883992213e-06, 'flow_matching_loss': 0.012623388040810823, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.68} +{'loss': 0.0123, 'grad_norm': 0.06511859595775604, 'learning_rate': 1.1726935555085817e-06, 'flow_matching_loss': 0.012302027270197868, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.68} +{'loss': 0.013, 'grad_norm': 0.06441745162010193, 'learning_rate': 1.166292824099105e-06, 'flow_matching_loss': 0.013017509318888187, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.68} +{'loss': 0.0121, 'grad_norm': 0.06387417018413544, 'learning_rate': 1.1599089019482768e-06, 'flow_matching_loss': 0.012132040224969387, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.68} +{'loss': 0.0136, 'grad_norm': 0.0675065740942955, 'learning_rate': 1.1535417968131718e-06, 'flow_matching_loss': 0.013589953817427159, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.69} +{'loss': 0.0128, 'grad_norm': 0.06698855757713318, 'learning_rate': 1.147191516430418e-06, 'flow_matching_loss': 0.012828375585377217, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.69} +{'loss': 0.0123, 'grad_norm': 0.06951673328876495, 'learning_rate': 1.140858068516205e-06, 'flow_matching_loss': 0.01225200816988945, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.69} +{'loss': 0.013, 'grad_norm': 0.07945740967988968, 'learning_rate': 1.1345414607662736e-06, 'flow_matching_loss': 0.012968241423368453, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.7} +{'loss': 0.0139, 'grad_norm': 0.06808175146579742, 'learning_rate': 1.1282417008558937e-06, 'flow_matching_loss': 0.013866188656538724, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.7} +{'loss': 0.0111, 'grad_norm': 0.05359599366784096, 'learning_rate': 1.1219587964398674e-06, 'flow_matching_loss': 0.01109476275742054, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.7} +{'loss': 0.0143, 'grad_norm': 0.07413151860237122, 'learning_rate': 1.1156927551525198e-06, 'flow_matching_loss': 0.014301272202283144, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.7} +{'loss': 0.0129, 'grad_norm': 0.07639714330434799, 'learning_rate': 1.109443584607675e-06, 'flow_matching_loss': 0.012859277240931988, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.71} 88%|████████▊ | 26490/30000 [16:03:19<1:20:28, 1.38s/it] 88%|████████▊ | 26491/30000 [16:03:27<3:14:24, 3.32s/it] 88%|████████▊ | 26492/30000 [16:03:28<2:36:47, 2.68s/it] 88%|████████▊ | 26493/30000 [16:03:29<2:11:45, 2.25s/it] 88%|████████▊ | 26494/30000 [16:03:31<1:52:36, 1.93s/it] 88%|████████▊ | 26495/30000 [16:03:32<1:39:49, 1.71s/it] 88%|████████▊ | 26496/30000 [16:03:33<1:30:44, 1.55s/it] 88%|████████▊ | 26497/30000 [16:03:34<1:23:59, 1.44s/it] 88%|████████▊ | 26498/30000 [16:03:35<1:19:53, 1.37s/it] 88%|████████▊ | 26499/30000 [16:03:42<2:44:33, 2.82s/it] 88%|████████▊ | 26500/30000 [16:03:43<2:15:50, 2.33s/it] 88%|████████▊ | 26500/30000 [16:03:43<2:15:50, 2.33s/it] 88%|████████▊ | 26501/30000 [16:03:44<1:56:37, 2.00s/it] 88%|████████▊ | 26502/30000 [16:03:45<1:41:59, 1.75s/it] 88%|████████▊ | 26503/30000 [16:03:46<1:32:40, 1.59s/it] 88%|████████▊ | 26504/30000 [16:03:48<1:28:56, 1.53s/it] 88%|████████▊ | 26505/30000 [16:03:49<1:23:30, 1.43s/it] 88%|████████▊ | 26506/30000 [16:03:50<1:20:13, 1.38s/it] 88%|████████▊ | 26507/30000 [16:03:58<3:04:37, 3.17s/it] 88%|████████▊ | 26508/30000 [16:03:59<2:29:41, 2.57s/it] 88%|████████▊ | 26509/30000 [16:04:00<2:05:38, 2.16s/it] 88%|████████▊ | 26510/30000 [16:04:01<1:48:37, 1.87s/it] 88%|████████▊ | 26510/30000 [16:04:01<1:48:37, 1.87s/it] 88%|████████▊ | 26511/30000 [16:04:02<1:36:36, 1.66s/it] 88%|████████▊ | 26512/30000 [16:04:04<1:28:25, 1.52s/it] 88%|████████▊ | 26513/30000 [16:04:05<1:22:44, 1.42s/it] 88%|████████▊ | 26514/30000 [16:04:06<1:18:32, 1.35s/it] 88%|████████▊ | 26515/30000 [16:04:13<2:54:56, 3.01s/it] 88%|████████▊ | 26516/30000 [16:04:14<2:23:21, 2.47s/it] 88%|████████▊ | 26517/30000 [16:04:15<2:01:20, 2.09s/it] 88%|████████▊ | 26518/30000 [16:04:16<1:45:21, 1.82s/it] 88%|████████▊ | 26519/30000 [16:04:18<1:34:11, 1.62s/it] 88%|████████▊ | 26520/30000 [16:04:19<1:27:20, 1.51s/it] 88%|████████▊ | 26520/30000 [16:04:19<1:27:20, 1.51s/it] 88%|████████▊ | 26521/30000 [16:04:20<1:22:17, 1.42s/it] 88%|████████▊ | 26522/30000 [16:04:21<1:18:25, 1.35s/it] 88%|████████▊ | 26523/30000 [16:04:28<2:57:25, 3.06s/it] 88%|████████▊ | 26524/30000 [16:04:29<2:24:50, 2.50s/it] 88%|████████▊ | 26525/30000 [16:04:31<2:02:00, 2.11s/it] 88%|████████▊ | 26526/30000 [16:04:32<1:46:21, 1.84s/it] 88%|████████▊ | 26527/30000 [16:04:33<1:34:47, 1.64s/it] 88%|████████▊ | 26528/30000 [16:04:34<1:28:59, 1.54s/it] 88%|████████▊ | 26529/30000 [16:04:36<1:22:55, 1.43s/it] 88%|████████▊ | 26530/30000 [16:04:37<1:18:12, 1.35s/it] 88%|████████▊ | 26530/30000 [16:04:37<1:18:12, 1.35s/it] 88%|████████▊ | 26531/30000 [16:04:44<2:55:44, 3.04s/it] 88%|████████▊ | 26532/30000 [16:04:45<2:23:21, 2.48s/it] 88%|████████▊ | 26533/30000 [16:04:46<2:00:47, 2.09s/it] 88%|████████▊ | 26534/30000 [16:04:47<1:44:42, 1.81s/it] 88%|████████▊ | 26535/30000 [16:04:48<1:33:57, 1.63s/it] 88%|████████▊ | 26536/30000 [16:04:50<1:28:20, 1.53s/it] 88%|████████▊ | 26537/30000 [16:04:51<1:22:22, 1.43s/it] 88%|████████▊ | 26538/30000 [16:04:52<1:18:22, 1.36s/it] 88%|████████▊ | 26539/30000 [16:04:59<3:00:00, 3.12s/it] 88%|████████▊ | 26540/30000 [16:05:00<2:26:28, 2.54s/it] 88%|████████▊ | 26540/30000 [16:05:01<2:26:28, 2.54s/it] 88%|████████▊ | 26541/30000 [16:05:02<2:03:18, 2.14s/it] 88%|████████▊ | 26542/30000 [16:05:03<1:46:34, 1.85s/it] 88%|████████▊ | 26543/30000 [16:05:04<1:34:46, 1.64s/it] 88%|████████▊ | 26544/30000 [16:05:05<1:26:56, 1.51s/it] 88%|████████▊ | 26545/30000 [16:05:06<1:21:15, 1.41s/it] 88%|████████▊ | 26546/30000 [16:05:08<1:17:06, 1.34s/it] 88%|████████▊ | 26547/30000 [16:05:16<3:13:59, 3.37s/it] 88%|████████▊ | 26548/30000 [16:05:17<2:36:19, 2.72s/it] 88%|████████▊ | 26549/30000 [16:05:18<2:09:52, 2.26s/it] 88%|████████▊ | 26550/30000 [16:05:19<1:51:01, 1.93s/it] 88%|████████▊ | 26550/30000 [16:05:19<1:51:01, 1.93s/it] 89%|████████▊ | 26551/30000 [16:05:20<1:37:59, 1.70s/it] 89%|████████▊ | 26552/30000 [16:05:22<1:29:09, 1.55s/it] 89%|████████▊ | 26553/30000 [16:05:23<1:22:36, 1.44s/it] 89%|████████▊ | 26554/30000 [16:05:24<1:18:08, 1.36s/it] 89%|████████▊ | 26555/30000 [16:05:31<3:00:25, 3.14s/it] 89%|████████▊ | 26556/30000 [16:05:32<2:26:19, 2.55s/it] 89%|████████▊ | 26557/30000 [16:05:34<2:02:42, 2.14s/it] 89%|████████▊ | 26558/30000 [16:05:35<1:46:08, 1.85s/it] 89%|████████▊ | 26559/30000 [16:05:36<1:34:41, 1.65s/it] 89%|████████▊ | 26560/30000 [16:05:37<1:26:48, 1.51s/it] 89%|████████▊ | 26560/30000 [16:05:37<1:26:48, 1.51s/it] 89%|████████▊ | 26561/30000 [16:05:38<1:22:27, 1.44s/it] 89%|████████▊ | 26562/30000 [16:05:40<1:17:58, 1.36s/it] 89%|████████▊ | 26563/30000 [16:05:48<3:24:54, 3.58s/it] 89%|████████▊ | 26564/30000 [16:05:50<2:43:37, 2.86s/it] 89%|████████▊ | 26565/30000 [16:05:51<2:14:50, 2.36s/it] 89%|████████▊ | 26566/30000 [16:05:52<1:54:56, 2.01s/it] 89%|████████▊ | 26567/30000 [16:05:53<1:40:54, 1.76s/it] 89%|████████▊ | 26568/30000 [16:05:54<1:31:45, 1.60s/it] 89%|████████▊ | 26569/30000 [16:05:56<1:24:34, 1.48s/it] 89%|████████▊ | 26570/30000 [16:05:57<1:19:33, 1.39s/it] 89%|████████▊ | 26570/30000 [16:05:57<1:19:33, 1.39s/it] 89%|████████▊ | 26571/30000 [16:06:06<3:30:18, 3.68s/it] 89%|████████▊ | 26572/30000 [16:06:07<2:47:39, 2.93s/it] 89%|████████▊ | 26573/30000 [16:06:08<2:17:57, 2.42s/it] 89%|████████▊ | 26574/30000 [16:06:09<1:56:29, 2.04s/it] 89%|████████▊ | 26575/30000 [16:06:10<1:41:23, 1.78s/it] 89%|████████▊ | 26576/30000 [16:06:12<1:31:37, 1.61s/it] 89%|████████▊ | 26577/30000 [16:06:13<1:24:20, 1.48s/it] 89%|████████▊ | 26578/30000 [16:06:14<1:19:48, 1.40s/it] 89%|████████▊ | 26579/30000 [16:06:22<3:03:55, 3.23s/it] 89%|████████▊ | 26580/30000 [16:06:23<2:28:43, 2.61s/it] 89%|████████▊ | 26580/30000 [16:06:23<2:28:43, 2.61s/it] 89%|████████▊ | 26581/30000 [16:06:24<2:06:13, 2.22s/it] 89%|████████▊ | 26582/30000 [16:06:25<1:48:11, 1.90s/it] 89%|████████▊ | 26583/30000 [16:06:26<1:35:45, 1.68s/it] 89%|████████▊ | 26584/30000 [16:06:28<1:27:29, 1.54s/it] 89%|████████▊ | 26585/30000 [16:06:29<1:21:29, 1.43s/it] 89%|████████▊ | 26586/30000 [16:06:30<1:17:21, 1.36s/it] 89%|████████▊ | 26587/30000 [16:06:39<3:22:31, 3.56s/it] 89%|████████▊ | 26588/30000 [16:06:40<2:41:51, 2.85s/it] 89%|████████▊ | 26589/30000 [16:06:41<2:13:33, 2.35s/it] 89%|████████▊ | 26590/30000 [16:06:42<1:53:31, 2.00s/it] 89%|████████▊ | 26590/30000 [16:06:42<1:53:31, 2.00s/it] 89%|████████▊ | 26591/30000 [16:06:43<1:39:05, 1.74s/it] 89%|████████▊ | 26592/30000 [16:06:45<1:31:45, 1.62s/it] 89%|████████▊ | 26593/30000 [16:06:46<1:25:05, 1.50s/it] 89%|████████▊ | 26594/30000 [16:06:47<1:20:03, 1.41s/it] 89%|████████▊ | 26595/30000 [16:06:54<2:51:56, 3.03s/it] 89%|████████▊ | 26596/30000 [16:06:55<2:21:02, 2.49s/it] 89%|████████▊ | 26597/30000 [16:06:56<1:58:54, 2.10s/it] 89%|████████▊ | 26598/30000 [16:06:57<1:43:17, 1.82s/it] 89%|████████▊ | 26599/30000 [16:06:59<1:32:14, 1.63s/it] 89%|████████▊ | 26600/30000 [16:07:00<1:24:40, 1.49s/it] 89%|████████▊ | 26600/30000 [16:07:00<1:24:40, 1.49s/it] 89%|████████▊ | 26601/30000 [16:07:01<1:20:00, 1.41s/it] 89%|████████▊ | 26602/30000 [16:07:02<1:17:02, 1.36s/it] 89%|████████▊ | 26603/30000 [16:07:11<3:17:55, 3.50s/it] 89%|████████▊ | 26604/30000 [16:07:12<2:38:25, 2.80s/it] 89%|████████▊ | 26605/30000 [16:07:13<2:11:51, 2.33s/it] 89%|████████▊ | 26606/30000 [16:07:14<1:52:00, 1.98s/it] 89%|████████▊ | 26607/30000 [16:07:16<1:38:19, 1.74s/it] 89%|████████▊ | 26608/30000 [16:07:17<1:29:04, 1.58s/it] 89%|████████▊ | 26609/30000 [16:07:18<1:22:32, 1.46s/it] 89%|████████▊ | 26610/30000 [16:07:19<1:17:44, 1.38s/it] 89%|████████▊ | 26610/30000 [16:07:19<1:17:44, 1.38s/it] 89%|████████▊ | 26611/30000 [16:07:27<3:01:13, 3.21s/it] 89%|████████▊ | 26612/30000 [16:07:28<2:29:20, 2.64s/it] 89%|████████▊ | 26613/30000 [16:07:29<2:04:35, 2.21s/it] 89%|████████▊ | 26614/30000 [16:07:30<1:49:08, 1.93s/it] 89%|████████▊ | 26615/30000 [16:07:32<1:35:53, 1.70s/it] 89%|████████▊ | 26616/30000 [16:07:33<1:27:19, 1.55s/it] 89%|████████▊ | 26617/30000 [16:07:34<1:21:28, 1.45s/it] 89%|████████▊ | 26618/30000 [16:07:35<1:16:49, 1.36s/it] 89%|████████▊ | 26619/30000 [16:07:43<3:08:03, 3.34s/it] 89%|████████▊ | 26620/30000 [16:07:44<2:31:09, 2.68s/it] 89%|████████▊ | 26620/30000 [16:07:44<2:31:09, 2.68s/it] 89%|████████▊ | 26621/30000 [16:07:45<2:06:00, 2.24s/it] 89%|████████▊ | 26622/30000 [16:07:47<1:49:06, 1.94s/it] 89%|████████▊ | 26623/30000 [16:07:48<1:37:50, 1.74s/it] 89%|████████▊ | 26624/30000 [16:07:49<1:28:45, 1.58s/it] 89%|████████▉ | 26625/30000 [16:07:50<1:24:08, 1.50s/it] 89%|████████▉ | 26626/30000 [16:07:52<1:18:46, 1.40s/it] 89%|████████▉ | 26627/30000 [16:07:59<3:06:41, 3.32s/it] 89%|████████▉ | 26628/30000 [16:08:01<2:30:52, 2.68s/it] 89%|████████▉ | 26629/30000 [16:08:02<2:05:43, 2.24s/it] 89%|████████▉ | 26630/30000 [16:08:03<1:47:38, 1.92s/it] 89%|████████▉ | 26630/30000 [16:08:03<1:47:38, 1.92s/it] 89%|████████▉ | 26631/30000 [16:08:04<1:35:10, 1.69s/it] 89%|████████▉ | 26632/30000 [16:08:05<1:26:25, 1.54s/it] 89%|████████▉ | 26633/30000 [16:08:07<1:20:34, 1.44s/it] 89%|████████▉ | 26634/30000 [16:08:08<1:16:38, 1.37s/it] 89%|████████▉ | 26635/30000 [16:08:15<2:57:34, 3.17s/it] 89%|████████▉ | 26636/30000 [16:08:16<2:24:15, 2.57s/it] 89%|████████▉ | 26637/30000 [16:08:17<2:00:50, 2.16s/it] 89%|████████▉ | 26638/30000 [16:08:19<1:44:35, 1.87s/it] 89%|████████▉ | 26639/30000 [16:08:20<1:32:46, 1.66s/it] 89%|████████▉ | 26640/30000 [16:08:21<1:24:54, 1.52s/it] 89%|████████▉ | 26640/30000 [16:08:21<1:24:54, 1.52s/it] 89%|████████▉ | 26641/30000 [16:08:22<1:19:17, 1.42s/it] 89%|████████▉ | 26642/30000 [16:08:23<1:15:09, 1.34s/it] 89%|████████▉ | 26643/30000 [16:08:29<2:34:28, 2.76s/it] 89%|████████▉ | 26644/30000 [16:08:31<2:08:28, 2.30s/it] 89%|████████▉ | 26645/30000 [16:08:32<1:49:53, 1.97s/it] 89%|████████▉ | 26646/30000 [16:08:33<1:36:44, 1.73s/it] 89%|████████▉ | 26647/30000 [16:08:34<1:28:12, 1.58s/it] 89%|████████▉ | 26648/30000 [16:08:35<1:21:58, 1.47s/it] 89%|████████▉ | 26649/30000 [16:08:37<1:17:37, 1.39s/it] 89%|████████▉ | 26650/30000 [16:08:38<1:14:05, 1.33s/it] 89%|████████▉ | 26650/30000 [16:08:38<1:14:05, 1.33s/it] 89%|████████▉ | 26651/30000 [16:08:46<3:05:32, 3.32s/it] 89%|████████▉ | 26652/30000 [16:08:47<2:29:32, 2.68s/it] 89%|████████▉ | 26653/30000 [16:08:48<2:04:34, 2.23s/it] 89%|████████▉ | 26654/30000 [16:08:49<1:47:02, 1.92s/it] 89%|████████▉ | 26655/30000 [16:08:51<1:35:04, 1.71s/it] 89%|████████▉ | 26656/30000 [16:08:52<1:26:29, 1.55s/it] 89%|████████▉ | 26657/30000 [16:08:53<1:20:33, 1.45s/it] 89%|████████▉ | 26658/30000 [16:08:54<1:16:13, 1.37s/it] 89%|████████▉ | 26659/30000 [16:09:02<3:03:52, 3.30s/it] 89%|████████▉ | 26660/30000 [16:09:03<2:28:37, 2.67s/it] 89%|████████▉ | 26660/30000 [16:09:03<2:28:37, 2.67s/it] 89%|████████▉ | 26661/30000 [16:09:04<2:03:51, 2.23s/it] 89%|████████▉ | 26662/30000 [16:09:06<1:45:59, 1.91s/it] 89%|████████▉ | 26663/30000 [16:09:07<1:33:56, 1.69s/it] 89%|████████▉ | 26664/30000 [16:09:08<1:26:11, 1.55s/it] 89%|████████▉ | 26665/30000 [16:09:09<1:20:46, 1.45s/it] 89%|████████▉ | 26666/30000 [16:09:10<1:17:35, 1.40s/it] 89%|████████▉ | 26667/30000 [16:09:18<3:03:34, 3.30s/it] 89%|████████▉ | 26668/30000 [16:09:19<2:28:01, 2.67s/it] 89%|████████▉ | 26669/30000 [16:09:21<2:03:25, 2.22s/it] 89%|████████▉ | 26670/30000 [16:09:22<1:45:42, 1.90s/it] 89%|████████▉ | 26670/30000 [16:09:22<1:45:42, 1.90s/it] 89%|████████▉ | 26671/30000 [16:09:23<1:33:30, 1.69s/it] 89%|████████▉ | 26672/30000 [16:09:24<1:25:05, 1.53s/it] 89%|████████▉ | 26673/30000 [16:09:25<1:19:11, 1.43s/it] 89%|████████▉ | 26674/30000 [16:09:26<1:15:28, 1.36s/it] 89%|████████▉ | 26675/30000 [16:09:34<3:01:10, 3.27s/it] 89%|████████▉ | 26676/30000 [16:09:35<2:26:18, 2.64s/it] 89%|████████▉ | 26677/30000 [16:09:37<2:02:34, 2.21s/it] 89%|████████▉ | 26678/30000 [16:09:38<1:46:14, 1.92s/it] 89%|████████▉ | 26679/30000 [16:09:39<1:34:04, 1.70s/it] 89%|████████▉ | 26680/30000 [16:09:40<1:27:38, 1.58s/it] 89%|████████▉ | 26680/30000 [16:09:40<1:27:38, 1.58s/it] 89%|████████▉ | 26681/30000 [16:09:42<1:21:31, 1.47s/it] 89%|████████▉ | 26682/30000 [16:09:43<1:16:25, 1.38s/it] 89%|████████▉ | 26683/30000 [16:09:50<2:58:22, 3.23s/it] 89%|████████▉ | 26684/30000 [16:09:51<2:24:27, 2.61s/it] 89%|████████▉ | 26685/30000 [16:09:53<2:00:17, 2.18s/it] 89%|████████▉ | 26686/30000 [16:09:54<1:43:33, 1.87s/it] 89%|████████▉ | 26687/30000 [16:09:55<1:31:38, 1.66s/it] 89%|████████▉ | 26688/30000 [16:09:56<1:25:03, 1.54s/it] 89%|████████▉ | 26689/30000 [16:09:57<1:19:04, 1.43s/it] 89%|████████▉ | 26690/30000 [16:09:59<1:15:53, 1.38s/it] 89%|████████▉ | 26690/30000 [16:09:59<1:15:53, 1.38s/it] 89%|████████▉ | 26691/30000 [16:10:07<3:09:49, 3.44s/it] 89%|████████▉ | 26692/30000 [16:10:08<2:32:16, 2.76s/it] 89%|████████▉ | 26693/30000 [16:10:09<2:06:17, 2.29s/it] 89%|████████▉ | 26694/30000 [16:10:10<1:47:39, 1.95s/it] 89%|████████▉ | 26695/30000 [16:10:12<1:34:49, 1.72s/it] 89%|████████▉ | 26696/30000 [16:10:13<1:26:02, 1.56s/it] 89%|████████▉ | 26697/30000 [16:10:14<1:19:51, 1.45s/it] 89%|████████▉ | 26698/30000 [16:10:15<1:16:11, 1.38s/it] 89%|████████▉ | 26699/30000 [16:10:23<3:04:20, 3.35s/it] 89%|████████▉ | 26700/30000 [16:10:24<2:28:26, 2.70s/it] 89%|████████▉ | 26700/30000 [16:10:24<2:28:26, 2.70s/it] 89%|████████▉ | 26701/30000 [16:10:25<2:03:48, 2.25s/it] 89%|████████▉ | 26702/30000 [16:10:27<1:45:42, 1.92s/it] 89%|████████▉ | 26703/30000 [16:10:28<1:32:58, 1.69s/it] 89%|████████▉ | 26704/30000 [16:10:29<1:24:54, 1.55s/it] 89%|████████▉ | 26705/30000 [16:10:30<1:19:08, 1.44s/it] 89%|████████▉ | 26706/30000 [16:10:31<1:14:42, 1.36s/it] 89%|████████▉ | 26707/30000 [16:10:39<3:00:49, 3.29s/it] 89%|████████▉ | 26708/30000 [16:10:40<2:26:02, 2.66s/it] 89%|████████▉ | 26709/30000 [16:10:42<2:03:31, 2.25s/it] 89%|████████▉ | 26710/30000 [16:10:43<1:45:39, 1.93s/it] 89%|████████▉ | 26710/30000 [16:10:43<1:45:39, 1.93s/it] 89%|████████▉ | 26711/30000 [16:10:44<1:34:43, 1.73s/it] 89%|████████▉ | 26712/30000 [16:10:45<1:26:59, 1.59s/it] 89%|████████▉ | 26713/30000 [16:10:47<1:20:17, 1.47s/it] 89%|████████▉ | 26714/30000 [16:10:48<1:15:24, 1.38s/it] 89%|████████▉ | 26715/30000 [16:10:55<3:00:54, 3.30s/it] 89%|████████▉ | 26716/30000 [16:10:57<2:25:40, 2.66s/it] 89%|████████▉ | 26717/30000 [16:10:58<2:01:30, 2.22s/it] 89%|████████▉ | 26718/30000 [16:10:59<1:44:07, 1.90s/it] 89%|████████▉ | 26719/30000 [16:11:00<1:31:55, 1.68s/it] 89%|████████▉ | 26720/30000 [16:11:01<1:24:43, 1.55s/it] 89%|████████▉ | 26720/30000 [16:11:01<1:24:43, 1.55s/it] 89%|████████▉ | 26721/30000 [16:11:03<1:18:29, 1.44s/it] 89%|████████▉ | 26722/30000 [16:11:04<1:13:58, 1.35s/it] 89%|████████▉ | 26723/30000 [16:11:11<2:46:52, 3.06s/it] 89%|████████▉ | 26724/30000 [16:11:12<2:16:27, 2.50s/it] 89%|████████▉ | 26725/30000 [16:11:13<1:54:47, 2.10s/it] 89%|████████▉ | 26726/30000 [16:11:14<1:39:13, 1.82s/it] 89%|████████▉ | 26727/30000 [16:11:15<1:28:39, 1.63s/it] 89%|████████▉ | 26728/30000 [16:11:17<1:21:44, 1.50s/it] 89%|████████▉ | 26729/30000 [16:11:18<1:16:29, 1.40s/it] 89%|████████▉ | 26730/30000 [16:11:19<1:12:53, 1.34s/it] 89%|████████▉ | 26730/30000 [16:11:19<1:12:53, 1.34s/it] 89%|████████▉ | 26731/30000 [16:11:27<2:53:28, 3.18s/it] 89%|████████▉ | 26732/30000 [16:11:28<2:20:54, 2.59s/it] 89%|████████▉ | 26733/30000 [16:11:29<2:00:04, 2.21s/it] 89%|████████▉ | 26734/30000 [16:11:30<1:43:08, 1.89s/it] 89%|████████▉ | 26735/30000 [16:11:31<1:31:17, 1.68s/it] 89%|████████▉ | 26736/30000 [16:11:33<1:23:02, 1.53s/it] 89%|████████▉ | 26737/30000 [16:11:34<1:17:18, 1.42s/it] 89%|████████▉ | 26738/30000 [16:11:35<1:13:04, 1.34s/it] 89%|████████▉ | 26739/30000 [16:11:43<3:03:07, 3.37s/it] 89%|████████▉ | 26740/30000 [16:11:44<2:27:17, 2.71s/it] 89%|████████▉ | 26740/30000 [16:11:44<2:27:17, 2.71s/it] 89%|████████▉ | 26741/30000 [16:11:45<2:02:40, 2.26s/it] 89%|████████▉ | 26742/30000 [16:11:47<1:44:45, 1.93s/it] 89%|████████▉ | 26743/30000 [16:11:48<1:32:18, 1.70s/it] 89%|████████▉ | 26744/30000 [16:11:49<1:23:39, 1.54s/it] 89%|████████▉ | 26745/30000 [16:11:50<1:18:16, 1.44s/it] 89%|████████▉ | 26746/30000 [16:11:51<1:14:43, 1.38s/it] 89%|████████▉ | 26747/30000 [16:12:00<3:07:20, 3.46s/it] 89%|████████▉ | 26748/30000 [16:12:01<2:30:13, 2.77s/it] 89%|████████▉ | 26749/30000 [16:12:02<2:05:02, 2.31s/it] 89%|████████▉ | 26750/30000 [16:12:03<1:46:19, 1.96s/it] 89%|████████▉ | 26750/30000 [16:12:03<1:46:19, 1.96s/it] 89%|████████▉ | 26751/30000 [16:12:04<1:33:16, 1.72s/it] 89%|████████▉ | 26752/30000 [16:12:06<1:26:23, 1.60s/it] 89%|████████▉ | 26753/30000 [16:12:07<1:20:07, 1.48s/it] 89%|████████▉ | 26754/30000 [16:12:08<1:15:36, 1.40s/it] 89%|████████▉ | 26755/30000 [16:12:17<3:11:30, 3.54s/it] 89%|████████▉ | 26756/30000 [16:12:18<2:33:13, 2.83s/it] 89%|████████▉ | 26757/30000 [16:12:19<2:06:19, 2.34s/it] 89%|████████▉ | 26758/30000 [16:12:20<1:47:04, 1.98s/it] 89%|████████▉ | 26759/30000 [16:12:21<1:33:56, 1.74s/it] 89%|████████▉ | 26760/30000 [16:12:22<1:25:08, 1.58s/it] 89%|████████▉ | 26760/30000 [16:12:22<1:25:08, 1.58s/it] 89%|████████▉ | 26761/30000 [16:12:24<1:19:12, 1.47s/it] 89%|████████▉ | 26762/30000 [16:12:25<1:14:14, 1.38s/it] 89%|████████▉ | 26763/30000 [16:12:33<3:03:55, 3.41s/it] 89%|████████▉ | 26764/30000 [16:12:34<2:27:46, 2.74s/it] 89%|████████▉ | 26765/30000 [16:12:35<2:02:27, 2.27s/it] 89%|████████▉ | 26766/30000 [16:12:37<1:44:39, 1.94s/it] 89%|████████▉ | 26767/30000 [16:12:38<1:31:56, 1.71s/it] 89%|████████▉ | 26768/30000 [16:12:39<1:23:07, 1.54s/it] 89%|████████▉ | 26769/30000 [16:12:40<1:17:15, 1.43s/it] 89%|████████▉ | 26770/30000 [16:12:41<1:13:12, 1.36s/it] 89%|████████▉ | 26770/30000 [16:12:41<1:13:12, 1.36s/it] 89%|████████▉ | 26771/30000 [16:12:49<2:53:27, 3.22s/it] 89%|████████▉ | 26772/30000 [16:12:50<2:20:15, 2.61s/it] 89%|████████▉ | 26773/30000 [16:12:51<1:56:47, 2.17s/it] 89%|████████▉ | 26774/30000 [16:12:52<1:40:41, 1.87s/it] 89%|████████▉ | 26775/30000 [16:12:54<1:30:01, 1.67s/it] 89%|████████▉ | 26776/30000 [16:12:55<1:22:33, 1.54s/it] 89%|████████▉ | 26777/30000 [16:12:56<1:17:14, 1.44s/it] 89%|████████▉ | 26778/30000 [16:12:57<1:13:09, 1.36s/it] 89%|████████▉ | 26779/30000 [16:13:05<3:02:45, 3.40s/it] 89%|████████▉ | 26780/30000 [16:13:06<2:26:51, 2.74s/it] 89%|████████▉ | 26780/30000 [16:13:06<2:26:51, 2.74s/it] 89%|████████▉ | 26781/30000 [16:13:08<2:02:11, 2.28s/it] 89%|████████▉ | 26782/30000 [16:13:09<1:44:44, 1.95s/it] 89%|████████▉ | 26783/30000 [16:13:10<1:32:07, 1.72s/it] 89%|████████▉ | 26784/30000 [16:13:11<1:23:14, 1.55s/it] 89%|████████▉ | 26785/30000 [16:13:12<1:17:03, 1.44s/it] 89%|████████▉ | 26786/30000 [16:13:14<1:12:58, 1.36s/it] 89%|████████▉ | 26787/30000 [16:13:21<2:52:15, 3.22s/it] 89%|████████▉ | 26788/30000 [16:13:22<2:19:04, 2.60s/it] 89%|████████▉ | 26789/30000 [16:13:23<1:56:18, 2.17s/it] 89%|████████▉ | 26790/30000 [16:13:25<1:40:15, 1.87s/it] 89%|████████▉ | 26790/30000 [16:13:25<1:40:15, 1.87s/it] 89%|████████▉ | 26791/30000 [16:13:26<1:29:12, 1.67s/it] 89%|████████▉ | 26792/30000 [16:13:27<1:21:11, 1.52s/it] 89%|████████▉ | 26793/30000 [16:13:28<1:16:05, 1.42s/it] 89%|████████▉ | 26794/30000 [16:13:29<1:12:28, 1.36s/it] 89%|████████▉ | 26795/30000 [16:13:37<2:51:05, 3.20s/it] 89%|████████▉ | 26796/30000 [16:13:38<2:18:46, 2.60s/it] 89%|████████▉ | 26797/30000 [16:13:39<1:58:09, 2.21s/it] 89%|████████▉ | 26798/30000 [16:13:41<1:43:11, 1.93s/it] 89%|████████▉ | 26799/30000 [16:13:42<1:32:55, 1.74s/it] 89%|████████▉ | 26800/30000 [16:13:43<1:24:17, 1.58s/it] 89%|████████▉ | 26800/30000 [16:13:43<1:24:17, 1.58s/it] 89%|████████▉ | 26801/30000 [16:13:44<1:17:35, 1.46s/it] 89%|████████▉ | 26802/30000 [16:13:46<1:13:20, 1.38s/it] 89%|████████▉ | 26803/30000 [16:13:52<2:37:03, 2.95s/it] 89%|████████▉ | 26804/30000 [16:13:53<2:08:48, 2.42s/it] 89%|████████▉ | 26805/30000 [16:13:55<1:49:14, 2.05s/it] 89%|████████▉ | 26806/30000 [16:13:56<1:35:25, 1.79s/it] 89%|████████▉ | 26807/30000 [16:13:57<1:25:41, 1.61s/it] 89%|████████▉ | 26808/30000 [16:13:58<1:18:41, 1.48s/it] 89%|████████▉ | 26809/30000 [16:13:59<1:15:04, 1.41s/it] 89%|████████▉ | 26810/30000 [16:14:01<1:11:42, 1.35s/it] 89%|████████▉ | 26810/30000 [16:14:01<1:11:42, 1.35s/it] 89%|████████▉ | 26811/30000 [16:14:07<2:34:15, 2.90s/it] 89%|████████▉ | 26812/30000 [16:14:08<2:07:48, 2.41s/it] 89%|████████▉ | 26813/30000 [16:14:10<1:48:55, 2.05s/it] 89%|████████▉ | 26814/30000 [16:14:11<1:37:12, 1.83s/it] 89%|████████▉ | 26815/30000 [16:14:12<1:26:46, 1.63s/it] 89%|████████▉ | 26816/30000 [16:14:13<1:19:22, 1.50s/it] 89%|████████▉ | 26817/30000 [16:14:14<1:14:31, 1.40s/it] 89%|████████▉ | 26818/30000 [16:14:16<1:11:15, 1.34s/it] 89%|████████▉ | 26819/30000 [16:14:23<2:44:37, 3.11s/it] 89%|████████▉ | 26820/30000 [16:14:24<2:13:46, 2.52s/it] 89%|████████▉ | 26820/30000 [16:14:24<2:13:46, 2.52s/it] 89%|████████▉ | 26821/30000 [16:14:25<1:54:23, 2.16s/it] 89%|████████▉ | 26822/30000 [16:14:26<1:38:42, 1.86s/it] 89%|████████▉ | 26823/30000 [16:14:28<1:27:58, 1.66s/it] 89%|████████▉ | 26824/30000 [16:14:29<1:20:25, 1.52s/it] 89%|████████▉ | 26825/30000 [16:14:30<1:14:57, 1.42s/it] 89%|████████▉ | 26826/30000 [16:14:31<1:10:56, 1.34s/it] 89%|████████▉ | 26827/30000 [16:14:38<2:46:07, 3.14s/it] 89%|████████▉ | 26828/30000 [16:14:40<2:14:34, 2.55s/it] 89%|████████▉ | 26829/30000 [16:14:41<1:54:24, 2.16s/it] 89%|████████▉ | 26830/30000 [16:14:42<1:38:52, 1.87s/it] 89%|████████▉ | 26830/30000 [16:14:42<1:38:52, 1.87s/it] 89%|████████▉ | 26831/30000 [16:14:43<1:28:02, 1.67s/it] 89%|████████▉ | 26832/30000 [16:14:45<1:20:33, 1.53s/it] 89%|████████▉ | 26833/30000 [16:14:46<1:14:58, 1.42s/it] 89%|████████▉ | 26834/30000 [16:14:47<1:11:31, 1.36s/it] 89%|████████▉ | 26835/30000 [16:14:54<2:47:08, 3.17s/it] 89%|████████▉ | 26836/30000 [16:14:55<2:15:03, 2.56s/it] 89%|████████▉ | 26837/30000 [16:14:57<1:55:13, 2.19s/it] 89%|████████▉ | 26838/30000 [16:14:58<1:39:17, 1.88s/it] 89%|████████▉ | 26839/30000 [16:14:59<1:28:06, 1.67s/it] 89%|████████▉ | 26840/30000 [16:15:00<1:20:23, 1.53s/it] 89%|████████▉ | 26840/30000 [16:15:00<1:20:23, 1.53s/it] 89%|████████▉ | 26841/30000 [16:15:01<1:15:16, 1.43s/it] 89%|████████▉ | 26842/30000 [16:15:03<1:11:59, 1.37s/it] 89%|████████▉ | 26843/30000 [16:15:10<2:39:48, 3.04s/it] 89%|████████▉ | 26844/30000 [16:15:11<2:10:56, 2.49s/it] 89%|████████▉ | 26845/30000 [16:15:12<1:50:33, 2.10s/it] 89%|████████▉ | 26846/30000 [16:15:13<1:35:42, 1.82s/it] 89%|████████▉ | 26847/30000 [16:15:15<1:27:19, 1.66s/it] 89%|████████▉ | 26848/30000 [16:15:16<1:20:31, 1.53s/it] 89%|████████▉ | 26849/30000 [16:15:17<1:15:12, 1.43s/it] 90%|████████▉ | 26850/30000 [16:15:18<1:11:10, 1.36s/it] 90%|████████▉ | 26850/30000 [16:15:18<1:11:10, 1.36s/it] 90%|████████▉ | 26851/30000 [16:15:27<3:03:58, 3.51s/it] 90%|████████▉ | 26852/30000 [16:15:28<2:27:12, 2.81s/it] 90%|████████▉ | 26853/30000 [16:15:29<2:02:05, 2.33s/it] 90%|████████▉ | 26854/30000 [16:15:30<1:44:01, 1.98s/it] 90%|████████▉ | 26855/30000 [16:15:31<1:31:13, 1.74s/it] 90%|████████▉ | 26856/30000 [16:15:33<1:22:38, 1.58s/it] 90%|████████▉ | 26857/30000 [16:15:34<1:16:18, 1.46s/it] 90%|████████▉ | 26858/30000 [16:15:35<1:12:16, 1.38s/it] 90%|████████▉ | 26859/30000 [16:15:44<3:20:00, 3.82s/it] 90%|████████▉ | 26860/30000 [16:15:46<2:38:16, 3.02s/it] 90%|████████▉ | 26860/30000 [16:15:46<2:38:16, 3.02s/it] 90%|████████▉ | 26861/30000 [16:15:47<2:09:27, 2.47s/it] 90%|████████▉ | 26862/30000 [16:15:48<1:48:57, 2.08s/it] 90%|████████▉ | 26863/30000 [16:15:49<1:34:33, 1.81s/it] 90%|████████▉ | 26864/30000 [16:15:50<1:24:42, 1.62s/it] 90%|████████▉ | 26865/30000 [16:15:52<1:17:43, 1.49s/it] 90%|████████▉ | 26866/30000 [16:15:53<1:13:46, 1.41s/it] 90%|████████▉ | 26867/30000 [16:16:00<2:37:24, 3.01s/it] 90%|████████▉ | 26868/30000 [16:16:01<2:08:26, 2.46s/it] 90%|████████▉ | 26869/30000 [16:16:02<1:48:22, 2.08s/it] 90%|████████▉ | 26870/30000 [16:16:03<1:34:03, 1.80s/it] 90%|████████▉ | 26870/30000 [16:16:03<1:34:03, 1.80s/it] 90%|████████▉ | 26871/30000 [16:16:04<1:24:12, 1.61s/it] 90%|████████▉ | 26872/30000 [16:16:05<1:17:21, 1.48s/it] 90%|████████▉ | 26873/30000 [16:16:07<1:12:59, 1.40s/it] 90%|████████▉ | 26874/30000 [16:16:08<1:09:30, 1.33s/it] 90%|████████▉ | 26875/30000 [16:16:19<3:46:10, 4.34s/it] 90%|████████▉ | 26876/30000 [16:16:20<2:56:28, 3.39s/it] 90%|████████▉ | 26877/30000 [16:16:21<2:21:55, 2.73s/it] 90%|████████▉ | 26878/30000 [16:16:23<1:59:16, 2.29s/it] 90%|████████▉ | 26879/30000 [16:16:24<1:42:07, 1.96s/it] 90%|████████▉ | 26880/30000 [16:16:25<1:30:06, 1.73s/it] 90%|████████▉ | 26880/30000 [16:16:25<1:30:06, 1.73s/it] 90%|████████▉ | 26881/30000 [16:16:26<1:21:36, 1.57s/it] 90%|████████▉ | 26882/30000 [16:16:28<1:15:47, 1.46s/it] 90%|████████▉ | 26883/30000 [16:16:44<5:02:34, 5.82s/it] 90%|████████▉ | 26884/30000 [16:16:45<3:50:06, 4.43s/it] 90%|████████▉ | 26885/30000 [16:16:46<2:59:34, 3.46s/it] 90%|████████▉ | 26886/30000 [16:16:47<2:24:00, 2.77s/it] 90%|████████▉ | 26887/30000 [16:16:48<1:59:31, 2.30s/it] 90%|████████▉ | 26888/30000 [16:16:49<1:42:14, 1.97s/it] 90%|████████▉ | 26889/30000 [16:16:51<1:30:08, 1.74s/it] 90%|████████▉ | 26890/30000 [16:16:52<1:21:54, 1.58s/it] 90%|████████▉ | 26890/30000 [16:16:52<1:21:54, 1.58s/it] 90%|████████▉ | 26891/30000 [16:17:08<5:10:15, 5.99s/it] 90%|████████▉ | 26892/30000 [16:17:09<3:56:18, 4.56s/it] 90%|████████▉ | 26893/30000 [16:17:11<3:03:57, 3.55s/it] 90%|████████▉ | 26894/30000 [16:17:12<2:27:12, 2.84s/it] 90%|████████▉ | 26895/30000 [16:17:13<2:01:36, 2.35s/it] 90%|████████▉ | 26896/30000 [16:17:14<1:45:50, 2.05s/it] 90%|████████▉ | 26897/30000 [16:17:16<1:32:34, 1.79s/it] 90%|████████▉ | 26898/30000 [16:17:17<1:23:03, 1.61s/it] 90%|████████▉ | 26899/30000 [16:17:33<5:06:22, 5.93s/it] 90%|████████▉ | 26900/30000 [16:17:34<3:52:25, 4.50s/it] +{'loss': 0.0128, 'grad_norm': 0.07460569590330124, 'learning_rate': 1.1032112923986708e-06, 'flow_matching_loss': 0.012814912293106318, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.71} +{'loss': 0.011, 'grad_norm': 0.06333214789628983, 'learning_rate': 1.0969958860983299e-06, 'flow_matching_loss': 0.010979409143328667, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.71} +{'loss': 0.0131, 'grad_norm': 0.07062237709760666, 'learning_rate': 1.0907973732589537e-06, 'flow_matching_loss': 0.013121406454592944, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.72} +{'loss': 0.0147, 'grad_norm': 0.08114725351333618, 'learning_rate': 1.0846157614123215e-06, 'flow_matching_loss': 0.014736282359808683, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.72} +{'loss': 0.0129, 'grad_norm': 0.08635713160037994, 'learning_rate': 1.078451058069671e-06, 'flow_matching_loss': 0.012898176675662398, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.72} +{'loss': 0.0117, 'grad_norm': 0.076515793800354, 'learning_rate': 1.0723032707217056e-06, 'flow_matching_loss': 0.011656746733933687, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.72} +{'loss': 0.0127, 'grad_norm': 0.08632136136293411, 'learning_rate': 1.066172406838558e-06, 'flow_matching_loss': 0.012701408099383115, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.73} +{'loss': 0.0137, 'grad_norm': 0.08342651277780533, 'learning_rate': 1.0600584738698127e-06, 'flow_matching_loss': 0.01372673511505127, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.73} +{'loss': 0.012, 'grad_norm': 0.060280416160821915, 'learning_rate': 1.0539614792444735e-06, 'flow_matching_loss': 0.012027610000222922, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.73} +{'loss': 0.0129, 'grad_norm': 0.06948684900999069, 'learning_rate': 1.0478814303709627e-06, 'flow_matching_loss': 0.012871930655092002, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.74} +{'loss': 0.0126, 'grad_norm': 0.07391028106212616, 'learning_rate': 1.041818334637109e-06, 'flow_matching_loss': 0.012626357562839984, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.74} +{'loss': 0.0132, 'grad_norm': 0.06805191189050674, 'learning_rate': 1.0357721994101526e-06, 'flow_matching_loss': 0.013215960375964642, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.74} +{'loss': 0.0118, 'grad_norm': 0.06617675721645355, 'learning_rate': 1.0297430320367135e-06, 'flow_matching_loss': 0.011777885910123586, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.75} +{'loss': 0.0144, 'grad_norm': 0.06166217848658562, 'learning_rate': 1.0237308398428014e-06, 'flow_matching_loss': 0.014371013455092906, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.75} +{'loss': 0.0127, 'grad_norm': 0.08461878448724747, 'learning_rate': 1.0177356301337925e-06, 'flow_matching_loss': 0.012715864833444357, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.75} +{'loss': 0.0127, 'grad_norm': 0.07615501433610916, 'learning_rate': 1.0117574101944348e-06, 'flow_matching_loss': 0.012742830999195575, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.75} +{'loss': 0.0143, 'grad_norm': 0.08782289177179337, 'learning_rate': 1.0057961872888293e-06, 'flow_matching_loss': 0.0142555829603225, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.76} +{'loss': 0.0143, 'grad_norm': 0.07511448115110397, 'learning_rate': 9.998519686604235e-07, 'flow_matching_loss': 0.014334477111697197, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.76} +{'loss': 0.0131, 'grad_norm': 0.07824164628982544, 'learning_rate': 9.939247615320047e-07, 'flow_matching_loss': 0.013113170489668846, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.76} +{'loss': 0.0136, 'grad_norm': 0.08294370025396347, 'learning_rate': 9.880145731056834e-07, 'flow_matching_loss': 0.013630472868680955, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.77} +{'loss': 0.0119, 'grad_norm': 0.07212171703577042, 'learning_rate': 9.821214105629034e-07, 'flow_matching_loss': 0.011904085148125887, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.77} +{'loss': 0.0133, 'grad_norm': 0.06830432265996933, 'learning_rate': 9.762452810644068e-07, 'flow_matching_loss': 0.013267873041331767, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.77} +{'loss': 0.0128, 'grad_norm': 0.07221584022045135, 'learning_rate': 9.703861917502487e-07, 'flow_matching_loss': 0.012770872190594673, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.77} +{'loss': 0.0126, 'grad_norm': 0.05683606490492821, 'learning_rate': 9.645441497397744e-07, 'flow_matching_loss': 0.012591711012646556, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.78} +{'loss': 0.0128, 'grad_norm': 0.05289946123957634, 'learning_rate': 9.587191621316154e-07, 'flow_matching_loss': 0.012788553535938264, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.78} +{'loss': 0.0122, 'grad_norm': 0.07604807615280151, 'learning_rate': 9.529112360036802e-07, 'flow_matching_loss': 0.01221606694161892, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.78} +{'loss': 0.0127, 'grad_norm': 0.07585832476615906, 'learning_rate': 9.471203784131449e-07, 'flow_matching_loss': 0.01271046306937933, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.79} +{'loss': 0.0124, 'grad_norm': 0.06289848685264587, 'learning_rate': 9.41346596396453e-07, 'flow_matching_loss': 0.012375692836940289, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.79} +{'loss': 0.0137, 'grad_norm': 0.07865127921104431, 'learning_rate': 9.355898969692889e-07, 'flow_matching_loss': 0.013712527696043253, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.79} +{'loss': 0.0133, 'grad_norm': 0.05315452069044113, 'learning_rate': 9.29850287126589e-07, 'flow_matching_loss': 0.013340232986956834, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.79} +{'loss': 0.0135, 'grad_norm': 0.08113444596529007, 'learning_rate': 9.24127773842518e-07, 'flow_matching_loss': 0.013532031606882811, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.8} +{'loss': 0.0127, 'grad_norm': 0.06716790795326233, 'learning_rate': 9.18422364070467e-07, 'flow_matching_loss': 0.012681691348552704, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.8} +{'loss': 0.0133, 'grad_norm': 0.06646155565977097, 'learning_rate': 9.127340647430538e-07, 'flow_matching_loss': 0.013341965340077876, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.8} +{'loss': 0.012, 'grad_norm': 0.08402863889932632, 'learning_rate': 9.070628827720929e-07, 'flow_matching_loss': 0.011977728363126517, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.81} +{'loss': 0.0133, 'grad_norm': 0.06793270260095596, 'learning_rate': 9.014088250486052e-07, 'flow_matching_loss': 0.013303213939070701, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.81} +{'loss': 0.0121, 'grad_norm': 0.06043362617492676, 'learning_rate': 8.95771898442807e-07, 'flow_matching_loss': 0.012101596593856812, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.81} +{'loss': 0.0124, 'grad_norm': 0.07232126593589783, 'learning_rate': 8.90152109804091e-07, 'flow_matching_loss': 0.01237841546535492, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.81} +{'loss': 0.0126, 'grad_norm': 0.0723879411816597, 'learning_rate': 8.845494659610365e-07, 'flow_matching_loss': 0.01260833302512765, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.82} +{'loss': 0.012, 'grad_norm': 0.055900465697050095, 'learning_rate': 8.789639737213812e-07, 'flow_matching_loss': 0.012015944626182318, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.82} +{'loss': 0.0141, 'grad_norm': 0.10025962442159653, 'learning_rate': 8.733956398720244e-07, 'flow_matching_loss': 0.0141375245526433, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.82} +{'loss': 0.0117, 'grad_norm': 0.05434897169470787, 'learning_rate': 8.678444711790157e-07, 'flow_matching_loss': 0.011661839764565229, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.83} + 90%|████████▉ | 26900/30000 [16:17:34<3:52:25, 4.50s/it] 90%|████████▉ | 26901/30000 [16:17:35<3:01:30, 3.51s/it] 90%|████████▉ | 26902/30000 [16:17:36<2:24:58, 2.81s/it] 90%|████████▉ | 26903/30000 [16:17:37<1:59:39, 2.32s/it] 90%|████████▉ | 26904/30000 [16:17:39<1:42:25, 1.99s/it] 90%|████████▉ | 26905/30000 [16:17:40<1:30:10, 1.75s/it] 90%|████████▉ | 26906/30000 [16:17:41<1:22:17, 1.60s/it] 90%|████████▉ | 26907/30000 [16:17:57<5:08:45, 5.99s/it] 90%|████████▉ | 26908/30000 [16:17:59<3:55:48, 4.58s/it] 90%|████████▉ | 26909/30000 [16:18:00<3:03:39, 3.57s/it] 90%|████████▉ | 26910/30000 [16:18:01<2:26:38, 2.85s/it] 90%|████████▉ | 26910/30000 [16:18:01<2:26:38, 2.85s/it] 90%|████████▉ | 26911/30000 [16:18:02<2:00:47, 2.35s/it] 90%|████████▉ | 26912/30000 [16:18:03<1:43:11, 2.01s/it] 90%|████████▉ | 26913/30000 [16:18:05<1:30:46, 1.76s/it] 90%|████████▉ | 26914/30000 [16:18:06<1:21:52, 1.59s/it] 90%|████████▉ | 26915/30000 [16:18:17<3:48:33, 4.45s/it] 90%|████████▉ | 26916/30000 [16:18:18<2:58:14, 3.47s/it] 90%|████████▉ | 26917/30000 [16:18:19<2:22:51, 2.78s/it] 90%|████████▉ | 26918/30000 [16:18:20<1:58:19, 2.30s/it] 90%|████████▉ | 26919/30000 [16:18:22<1:40:59, 1.97s/it] 90%|████████▉ | 26920/30000 [16:18:23<1:28:52, 1.73s/it] 90%|████████▉ | 26920/30000 [16:18:23<1:28:52, 1.73s/it] 90%|████████▉ | 26921/30000 [16:18:24<1:20:53, 1.58s/it] 90%|████████▉ | 26922/30000 [16:18:25<1:15:25, 1.47s/it] 90%|████████▉ | 26923/30000 [16:18:35<3:29:50, 4.09s/it] 90%|████████▉ | 26924/30000 [16:18:37<2:45:27, 3.23s/it] 90%|████████▉ | 26925/30000 [16:18:38<2:17:16, 2.68s/it] 90%|████████▉ | 26926/30000 [16:18:39<1:54:16, 2.23s/it] 90%|████████▉ | 26927/30000 [16:18:40<1:39:03, 1.93s/it] 90%|████████▉ | 26928/30000 [16:18:42<1:27:52, 1.72s/it] 90%|████████▉ | 26929/30000 [16:18:43<1:19:41, 1.56s/it] 90%|████████▉ | 26930/30000 [16:18:44<1:14:13, 1.45s/it] 90%|████████▉ | 26930/30000 [16:18:44<1:14:13, 1.45s/it] 90%|████████▉ | 26931/30000 [16:18:54<3:17:17, 3.86s/it] 90%|████████▉ | 26932/30000 [16:18:55<2:40:25, 3.14s/it] 90%|████████▉ | 26933/30000 [16:18:56<2:10:44, 2.56s/it] 90%|████████▉ | 26934/30000 [16:18:57<1:49:50, 2.15s/it] 90%|████████▉ | 26935/30000 [16:18:59<1:35:09, 1.86s/it] 90%|████████▉ | 26936/30000 [16:19:00<1:26:29, 1.69s/it] 90%|████████▉ | 26937/30000 [16:19:01<1:18:53, 1.55s/it] 90%|████████▉ | 26938/30000 [16:19:02<1:13:56, 1.45s/it] 90%|████████▉ | 26939/30000 [16:19:12<3:24:20, 4.01s/it] 90%|████████▉ | 26940/30000 [16:19:13<2:41:25, 3.17s/it] 90%|████████▉ | 26940/30000 [16:19:13<2:41:25, 3.17s/it] 90%|████████▉ | 26941/30000 [16:19:15<2:11:32, 2.58s/it] 90%|████████▉ | 26942/30000 [16:19:16<1:50:05, 2.16s/it] 90%|████████▉ | 26943/30000 [16:19:17<1:35:07, 1.87s/it] 90%|████████▉ | 26944/30000 [16:19:18<1:25:18, 1.67s/it] 90%|████████▉ | 26945/30000 [16:19:19<1:18:00, 1.53s/it] 90%|████████▉ | 26946/30000 [16:19:21<1:15:03, 1.47s/it] 90%|████████▉ | 26947/30000 [16:19:31<3:31:52, 4.16s/it] 90%|████████▉ | 26948/30000 [16:19:33<2:48:24, 3.31s/it] 90%|████████▉ | 26949/30000 [16:19:34<2:17:10, 2.70s/it] 90%|████████▉ | 26950/30000 [16:19:35<1:54:08, 2.25s/it] 90%|████████▉ | 26950/30000 [16:19:35<1:54:08, 2.25s/it] 90%|████████▉ | 26951/30000 [16:19:36<1:38:05, 1.93s/it] 90%|████████▉ | 26952/30000 [16:19:37<1:26:59, 1.71s/it] 90%|████████▉ | 26953/30000 [16:19:39<1:19:58, 1.57s/it] 90%|████████▉ | 26954/30000 [16:19:43<2:02:05, 2.40s/it] 90%|████████▉ | 26955/30000 [16:19:48<2:36:19, 3.08s/it] 90%|████████▉ | 26956/30000 [16:19:49<2:11:08, 2.59s/it] 90%|████████▉ | 26957/30000 [16:19:50<1:51:10, 2.19s/it] 90%|████████▉ | 26958/30000 [16:19:52<1:35:43, 1.89s/it] 90%|████████▉ | 26959/30000 [16:19:53<1:24:48, 1.67s/it] 90%|████████▉ | 26960/30000 [16:19:54<1:17:51, 1.54s/it] 90%|████████▉ | 26960/30000 [16:19:54<1:17:51, 1.54s/it] 90%|████████▉ | 26961/30000 [16:19:55<1:13:10, 1.44s/it] 90%|████████▉ | 26962/30000 [16:20:07<3:46:15, 4.47s/it] 90%|████████▉ | 26963/30000 [16:20:08<2:56:32, 3.49s/it] 90%|████████▉ | 26964/30000 [16:20:09<2:21:30, 2.80s/it] 90%|████████▉ | 26965/30000 [16:20:10<1:57:19, 2.32s/it] 90%|████████▉ | 26966/30000 [16:20:12<1:41:05, 2.00s/it] 90%|████████▉ | 26967/30000 [16:20:13<1:28:32, 1.75s/it] 90%|████████▉ | 26968/30000 [16:20:14<1:20:13, 1.59s/it] 90%|████████▉ | 26969/30000 [16:20:15<1:14:39, 1.48s/it] 90%|████████▉ | 26970/30000 [16:20:30<4:39:01, 5.53s/it] 90%|████████▉ | 26970/30000 [16:20:30<4:39:01, 5.53s/it] 90%|████████▉ | 26971/30000 [16:20:31<3:33:10, 4.22s/it] 90%|████████▉ | 26972/30000 [16:20:32<2:47:20, 3.32s/it] 90%|████████▉ | 26973/30000 [16:20:34<2:15:24, 2.68s/it] 90%|████████▉ | 26974/30000 [16:20:35<1:52:29, 2.23s/it] 90%|████████▉ | 26975/30000 [16:20:36<1:38:45, 1.96s/it] 90%|████████▉ | 26976/30000 [16:20:37<1:27:48, 1.74s/it] 90%|████████▉ | 26977/30000 [16:20:39<1:19:35, 1.58s/it] 90%|████████▉ | 26978/30000 [16:20:54<4:43:14, 5.62s/it] 90%|████████▉ | 26979/30000 [16:20:55<3:36:27, 4.30s/it] 90%|████████▉ | 26980/30000 [16:20:56<2:50:51, 3.39s/it] 90%|████████▉ | 26980/30000 [16:20:56<2:50:51, 3.39s/it] 90%|████████▉ | 26981/30000 [16:20:57<2:17:44, 2.74s/it] 90%|████████▉ | 26982/30000 [16:20:59<1:56:48, 2.32s/it] 90%|████████▉ | 26983/30000 [16:21:00<1:39:32, 1.98s/it] 90%|████████▉ | 26984/30000 [16:21:01<1:27:49, 1.75s/it] 90%|████████▉ | 26985/30000 [16:21:02<1:20:10, 1.60s/it] 90%|████████▉ | 26986/30000 [16:21:13<3:29:21, 4.17s/it] 90%|████████▉ | 26987/30000 [16:21:14<2:44:01, 3.27s/it] 90%|████████▉ | 26988/30000 [16:21:15<2:13:00, 2.65s/it] 90%|████████▉ | 26989/30000 [16:21:16<1:50:45, 2.21s/it] 90%|████████▉ | 26990/30000 [16:21:17<1:35:16, 1.90s/it] 90%|████████▉ | 26990/30000 [16:21:17<1:35:16, 1.90s/it] 90%|████████▉ | 26991/30000 [16:21:18<1:24:35, 1.69s/it] 90%|████████▉ | 26992/30000 [16:21:20<1:18:06, 1.56s/it] 90%|████████▉ | 26993/30000 [16:21:21<1:14:57, 1.50s/it] 90%|████████▉ | 26994/30000 [16:21:29<2:51:17, 3.42s/it] 90%|████████▉ | 26995/30000 [16:21:30<2:18:06, 2.76s/it] 90%|████████▉ | 26996/30000 [16:21:31<1:55:34, 2.31s/it] 90%|████████▉ | 26997/30000 [16:21:33<1:40:21, 2.01s/it] 90%|████████▉ | 26998/30000 [16:21:34<1:27:54, 1.76s/it] 90%|████████▉ | 26999/30000 [16:21:35<1:19:45, 1.59s/it] 90%|█████████ | 27000/30000 [16:21:36<1:15:05, 1.50s/it] 90%|█████████ | 27000/30000 [16:21:36<1:15:05, 1.50s/it] 90%|█████████ | 27001/30000 [16:21:38<1:11:10, 1.42s/it] 90%|█████████ | 27002/30000 [16:21:43<2:03:04, 2.46s/it] 90%|█████████ | 27003/30000 [16:21:44<1:44:27, 2.09s/it] 90%|█████████ | 27004/30000 [16:21:45<1:31:32, 1.83s/it] 90%|█████████ | 27005/30000 [16:21:46<1:22:04, 1.64s/it] 90%|█████████ | 27006/30000 [16:21:47<1:14:43, 1.50s/it] 90%|█████████ | 27007/30000 [16:21:49<1:11:01, 1.42s/it] 90%|█████████ | 27008/30000 [16:21:50<1:08:20, 1.37s/it] 90%|█████████ | 27009/30000 [16:21:51<1:06:44, 1.34s/it] 90%|█████████ | 27010/30000 [16:22:01<3:14:02, 3.89s/it] 90%|█████████ | 27010/30000 [16:22:01<3:14:02, 3.89s/it] 90%|█████████ | 27011/30000 [16:22:02<2:33:54, 3.09s/it] 90%|█████████ | 27012/30000 [16:22:03<2:05:45, 2.53s/it] 90%|█████████ | 27013/30000 [16:22:05<1:45:44, 2.12s/it] 90%|█████████ | 27014/30000 [16:22:06<1:32:10, 1.85s/it] 90%|█████████ | 27015/30000 [16:22:07<1:22:48, 1.66s/it] 90%|█████████ | 27016/30000 [16:22:08<1:16:52, 1.55s/it] 90%|█████████ | 27017/30000 [16:22:10<1:11:54, 1.45s/it] 90%|█████████ | 27018/30000 [16:22:18<2:52:13, 3.47s/it] 90%|█████████ | 27019/30000 [16:22:19<2:18:53, 2.80s/it] 90%|█████████ | 27020/30000 [16:22:20<1:55:55, 2.33s/it] 90%|█████████ | 27020/30000 [16:22:20<1:55:55, 2.33s/it] 90%|█████████ | 27021/30000 [16:22:22<1:40:46, 2.03s/it] 90%|█████████ | 27022/30000 [16:22:23<1:28:48, 1.79s/it] 90%|█████████ | 27023/30000 [16:22:24<1:19:56, 1.61s/it] 90%|█████████ | 27024/30000 [16:22:25<1:13:57, 1.49s/it] 90%|█████████ | 27025/30000 [16:22:26<1:11:48, 1.45s/it] 90%|█████████ | 27026/30000 [16:22:32<2:09:20, 2.61s/it] 90%|█████████ | 27027/30000 [16:22:33<1:48:42, 2.19s/it] 90%|█████████ | 27028/30000 [16:22:34<1:34:03, 1.90s/it] 90%|█████████ | 27029/30000 [16:22:36<1:26:49, 1.75s/it] 90%|█████████ | 27030/30000 [16:22:37<1:20:23, 1.62s/it] 90%|█████████ | 27030/30000 [16:22:37<1:20:23, 1.62s/it] 90%|█████████ | 27031/30000 [16:22:38<1:15:44, 1.53s/it] 90%|█████████ | 27032/30000 [16:22:40<1:10:50, 1.43s/it] 90%|█████████ | 27033/30000 [16:22:41<1:07:56, 1.37s/it] 90%|█████████ | 27034/30000 [16:22:46<2:10:47, 2.65s/it] 90%|█████████ | 27035/30000 [16:22:48<1:49:33, 2.22s/it] 90%|█████████ | 27036/30000 [16:22:49<1:34:25, 1.91s/it] 90%|█████████ | 27037/30000 [16:22:50<1:24:32, 1.71s/it] 90%|█████████ | 27038/30000 [16:22:51<1:20:32, 1.63s/it] 90%|█████████ | 27039/30000 [16:22:53<1:14:08, 1.50s/it] 90%|█████████ | 27040/30000 [16:22:54<1:10:51, 1.44s/it] 90%|█████████ | 27040/30000 [16:22:54<1:10:51, 1.44s/it] 90%|█████████ | 27041/30000 [16:22:55<1:07:59, 1.38s/it] 90%|█████████ | 27042/30000 [16:23:00<2:01:28, 2.46s/it] 90%|█████████ | 27043/30000 [16:23:01<1:43:10, 2.09s/it] 90%|█████████ | 27044/30000 [16:23:03<1:30:38, 1.84s/it] 90%|█████████ | 27045/30000 [16:23:04<1:21:43, 1.66s/it] 90%|█████████ | 27046/30000 [16:23:05<1:15:48, 1.54s/it] 90%|█████████ | 27047/30000 [16:23:06<1:12:30, 1.47s/it] 90%|█████████ | 27048/30000 [16:23:08<1:08:38, 1.40s/it] 90%|█████████ | 27049/30000 [16:23:09<1:05:50, 1.34s/it] 90%|█████████ | 27050/30000 [16:23:14<2:02:26, 2.49s/it] 90%|█████████ | 27050/30000 [16:23:14<2:02:26, 2.49s/it] 90%|█████████ | 27051/30000 [16:23:16<1:47:46, 2.19s/it] 90%|█████████ | 27052/30000 [16:23:17<1:33:50, 1.91s/it] 90%|█████████ | 27053/30000 [16:23:18<1:24:26, 1.72s/it] 90%|█████████ | 27054/30000 [16:23:19<1:18:02, 1.59s/it] 90%|█████████ | 27055/30000 [16:23:21<1:12:41, 1.48s/it] 90%|█████████ | 27056/30000 [16:23:22<1:09:21, 1.41s/it] 90%|█████████ | 27057/30000 [16:23:23<1:07:11, 1.37s/it] 90%|█████████ | 27058/30000 [16:23:28<1:56:03, 2.37s/it] 90%|█████████ | 27059/30000 [16:23:29<1:41:17, 2.07s/it] 90%|█████████ | 27060/30000 [16:23:30<1:28:14, 1.80s/it] 90%|█████████ | 27060/30000 [16:23:30<1:28:14, 1.80s/it] 90%|█████████ | 27061/30000 [16:23:32<1:21:50, 1.67s/it] 90%|█████████ | 27062/30000 [16:23:33<1:14:51, 1.53s/it] 90%|█████████ | 27063/30000 [16:23:34<1:10:44, 1.45s/it] 90%|█████████ | 27064/30000 [16:23:35<1:07:12, 1.37s/it] 90%|█████████ | 27065/30000 [16:23:37<1:06:09, 1.35s/it] 90%|█████████ | 27066/30000 [16:23:41<1:56:38, 2.39s/it] 90%|█████████ | 27067/30000 [16:23:43<1:39:03, 2.03s/it] 90%|█████████ | 27068/30000 [16:23:44<1:26:37, 1.77s/it] 90%|█████████ | 27069/30000 [16:23:45<1:19:06, 1.62s/it] 90%|█████████ | 27070/30000 [16:23:47<1:16:13, 1.56s/it] 90%|█████████ | 27070/30000 [16:23:47<1:16:13, 1.56s/it] 90%|█████████ | 27071/30000 [16:23:48<1:12:07, 1.48s/it] 90%|█████████ | 27072/30000 [16:23:49<1:08:07, 1.40s/it] 90%|█████████ | 27073/30000 [16:23:50<1:05:38, 1.35s/it] 90%|█████████ | 27074/30000 [16:23:56<2:03:25, 2.53s/it] 90%|█████████ | 27075/30000 [16:23:57<1:45:57, 2.17s/it] 90%|█████████ | 27076/30000 [16:23:58<1:31:53, 1.89s/it] 90%|█████████ | 27077/30000 [16:23:59<1:22:48, 1.70s/it] 90%|█████████ | 27078/30000 [16:24:01<1:15:38, 1.55s/it] 90%|█████████ | 27079/30000 [16:24:02<1:16:46, 1.58s/it] 90%|█████████ | 27080/30000 [16:24:04<1:14:35, 1.53s/it] 90%|█████████ | 27080/30000 [16:24:04<1:14:35, 1.53s/it] 90%|█████████ | 27081/30000 [16:24:05<1:15:28, 1.55s/it] 90%|█████████ | 27082/30000 [16:24:09<1:45:12, 2.16s/it] 90%|█████████ | 27083/30000 [16:24:10<1:31:06, 1.87s/it] 90%|█████████ | 27084/30000 [16:24:11<1:21:18, 1.67s/it] 90%|█████████ | 27085/30000 [16:24:12<1:14:50, 1.54s/it] 90%|█████████ | 27086/30000 [16:24:14<1:09:53, 1.44s/it] 90%|█████████ | 27087/30000 [16:24:15<1:06:22, 1.37s/it] 90%|█████████ | 27088/30000 [16:24:16<1:05:27, 1.35s/it] 90%|█████████ | 27089/30000 [16:24:18<1:06:35, 1.37s/it] 90%|█████████ | 27090/30000 [16:24:24<2:20:35, 2.90s/it] 90%|█████████ | 27090/30000 [16:24:24<2:20:35, 2.90s/it] 90%|█████████ | 27091/30000 [16:24:25<1:55:53, 2.39s/it] 90%|█████████ | 27092/30000 [16:24:27<1:39:20, 2.05s/it] 90%|█████████ | 27093/30000 [16:24:28<1:26:48, 1.79s/it] 90%|█████████ | 27094/30000 [16:24:29<1:17:55, 1.61s/it] 90%|█████████ | 27095/30000 [16:24:30<1:14:06, 1.53s/it] 90%|█████████ | 27096/30000 [16:24:31<1:09:20, 1.43s/it] 90%|█████████ | 27097/30000 [16:24:33<1:05:59, 1.36s/it] 90%|█████████ | 27098/30000 [16:24:42<3:03:39, 3.80s/it] 90%|█████████ | 27099/30000 [16:24:43<2:25:55, 3.02s/it] 90%|█████████ | 27100/30000 [16:24:45<1:59:45, 2.48s/it] 90%|█████████ | 27100/30000 [16:24:45<1:59:45, 2.48s/it] 90%|█████████ | 27101/30000 [16:24:46<1:41:06, 2.09s/it] 90%|█████████ | 27102/30000 [16:24:47<1:29:21, 1.85s/it] 90%|█████████ | 27103/30000 [16:24:48<1:19:37, 1.65s/it] 90%|█████████ | 27104/30000 [16:24:49<1:13:33, 1.52s/it] 90%|█████████ | 27105/30000 [16:24:51<1:10:13, 1.46s/it] 90%|█████████ | 27106/30000 [16:24:57<2:20:22, 2.91s/it] 90%|█████████ | 27107/30000 [16:24:58<1:56:02, 2.41s/it] 90%|█████████ | 27108/30000 [16:24:59<1:38:51, 2.05s/it] 90%|█████████ | 27109/30000 [16:25:01<1:26:34, 1.80s/it] 90%|█████████ | 27110/30000 [16:25:02<1:17:47, 1.62s/it] 90%|█████████ | 27110/30000 [16:25:02<1:17:47, 1.62s/it] 90%|█████████ | 27111/30000 [16:25:03<1:14:04, 1.54s/it] 90%|█████████ | 27112/30000 [16:25:05<1:14:00, 1.54s/it] 90%|█████████ | 27113/30000 [16:25:06<1:09:03, 1.44s/it] 90%|█████████ | 27114/30000 [16:25:12<2:14:24, 2.79s/it] 90%|█████████ | 27115/30000 [16:25:13<1:54:05, 2.37s/it] 90%|█████████ | 27116/30000 [16:25:15<1:37:39, 2.03s/it] 90%|█████████ | 27117/30000 [16:25:16<1:25:49, 1.79s/it] 90%|█████████ | 27118/30000 [16:25:17<1:17:33, 1.61s/it] 90%|█████████ | 27119/30000 [16:25:18<1:15:10, 1.57s/it] 90%|█████████ | 27120/30000 [16:25:20<1:09:49, 1.45s/it] 90%|█████████ | 27120/30000 [16:25:20<1:09:49, 1.45s/it] 90%|█████████ | 27121/30000 [16:25:21<1:06:31, 1.39s/it] 90%|█████████ | 27122/30000 [16:25:28<2:33:08, 3.19s/it] 90%|█████████ | 27123/30000 [16:25:30<2:05:05, 2.61s/it] 90%|█████████ | 27124/30000 [16:25:31<1:45:11, 2.19s/it] 90%|█████████ | 27125/30000 [16:25:32<1:36:32, 2.01s/it] 90%|█████████ | 27126/30000 [16:25:34<1:28:42, 1.85s/it] 90%|█████████ | 27127/30000 [16:25:35<1:19:42, 1.66s/it] 90%|█████████ | 27128/30000 [16:25:36<1:13:05, 1.53s/it] 90%|█████████ | 27129/30000 [16:25:37<1:08:50, 1.44s/it] 90%|█████████ | 27130/30000 [16:25:42<1:52:09, 2.34s/it] 90%|█████████ | 27130/30000 [16:25:42<1:52:09, 2.34s/it] 90%|█████████ | 27131/30000 [16:25:43<1:35:56, 2.01s/it] 90%|█████████ | 27132/30000 [16:25:44<1:24:19, 1.76s/it] 90%|█████████ | 27133/30000 [16:25:46<1:16:24, 1.60s/it] 90%|█████████ | 27134/30000 [16:25:47<1:10:07, 1.47s/it] 90%|█████████ | 27135/30000 [16:25:48<1:06:23, 1.39s/it] 90%|█████████ | 27136/30000 [16:25:49<1:03:40, 1.33s/it] 90%|█████████ | 27137/30000 [16:25:50<1:01:52, 1.30s/it] 90%|█████████ | 27138/30000 [16:25:59<2:43:49, 3.43s/it] 90%|█████████ | 27139/30000 [16:26:00<2:11:54, 2.77s/it] 90%|█████████ | 27140/30000 [16:26:01<1:49:32, 2.30s/it] 90%|█████████ | 27140/30000 [16:26:01<1:49:32, 2.30s/it] 90%|█████████ | 27141/30000 [16:26:02<1:34:07, 1.98s/it] 90%|█████████ | 27142/30000 [16:26:04<1:22:59, 1.74s/it] 90%|█████████ | 27143/30000 [16:26:05<1:15:20, 1.58s/it] 90%|█████████ | 27144/30000 [16:26:06<1:12:01, 1.51s/it] 90%|█████████ | 27145/30000 [16:26:07<1:07:31, 1.42s/it] 90%|█████████ | 27146/30000 [16:26:13<2:06:53, 2.67s/it] 90%|█████████ | 27147/30000 [16:26:14<1:45:57, 2.23s/it] 90%|█████████ | 27148/30000 [16:26:15<1:30:55, 1.91s/it] 90%|█████████ | 27149/30000 [16:26:17<1:26:16, 1.82s/it] 90%|█████████ | 27150/30000 [16:26:18<1:21:25, 1.71s/it] 90%|█████████ | 27150/30000 [16:26:18<1:21:25, 1.71s/it] 91%|█████████ | 27151/30000 [16:26:20<1:14:45, 1.57s/it] 91%|█████████ | 27152/30000 [16:26:21<1:09:34, 1.47s/it] 91%|█████████ | 27153/30000 [16:26:22<1:05:36, 1.38s/it] 91%|█████████ | 27154/30000 [16:26:27<1:57:57, 2.49s/it] 91%|█████████ | 27155/30000 [16:26:29<1:42:16, 2.16s/it] 91%|█████████ | 27156/30000 [16:26:30<1:29:56, 1.90s/it] 91%|█████████ | 27157/30000 [16:26:31<1:21:29, 1.72s/it] 91%|█████████ | 27158/30000 [16:26:32<1:14:18, 1.57s/it] 91%|█████████ | 27159/30000 [16:26:34<1:10:56, 1.50s/it] 91%|█████████ | 27160/30000 [16:26:35<1:06:47, 1.41s/it] 91%|█████████ | 27160/30000 [16:26:35<1:06:47, 1.41s/it] 91%|█████████ | 27161/30000 [16:26:36<1:05:59, 1.39s/it] 91%|█████████ | 27162/30000 [16:26:42<2:08:08, 2.71s/it] 91%|█████████ | 27163/30000 [16:26:51<3:39:38, 4.65s/it] 91%|█████████ | 27164/30000 [16:26:52<2:50:34, 3.61s/it] 91%|█████████ | 27165/30000 [16:26:54<2:16:42, 2.89s/it] 91%|█████████ | 27166/30000 [16:26:55<1:52:17, 2.38s/it] 91%|█████████ | 27167/30000 [16:26:56<1:35:39, 2.03s/it] 91%|█████████ | 27168/30000 [16:26:57<1:25:44, 1.82s/it] 91%|█████████ | 27169/30000 [16:26:58<1:16:57, 1.63s/it] 91%|█████████ | 27170/30000 [16:27:00<1:11:47, 1.52s/it] 91%|█████████ | 27170/30000 [16:27:00<1:11:47, 1.52s/it] 91%|█████████ | 27171/30000 [16:27:12<3:43:28, 4.74s/it] 91%|█████████ | 27172/30000 [16:27:13<2:52:54, 3.67s/it] 91%|█████████ | 27173/30000 [16:27:14<2:17:40, 2.92s/it] 91%|█████████ | 27174/30000 [16:27:16<1:53:05, 2.40s/it] 91%|█████████ | 27175/30000 [16:27:17<1:35:56, 2.04s/it] 91%|█████████ | 27176/30000 [16:27:18<1:24:19, 1.79s/it] 91%|█████████ | 27177/30000 [16:27:19<1:15:29, 1.60s/it] 91%|█████████ | 27178/30000 [16:27:20<1:09:24, 1.48s/it] 91%|█████████ | 27179/30000 [16:27:32<3:37:22, 4.62s/it] 91%|█████████ | 27180/30000 [16:27:33<2:48:55, 3.59s/it] 91%|█████████ | 27180/30000 [16:27:33<2:48:55, 3.59s/it] 91%|█████████ | 27181/30000 [16:27:35<2:15:08, 2.88s/it] 91%|█████████ | 27182/30000 [16:27:36<1:51:17, 2.37s/it] 91%|█████████ | 27183/30000 [16:27:37<1:34:24, 2.01s/it] 91%|█████████ | 27184/30000 [16:27:38<1:24:08, 1.79s/it] 91%|█████████ | 27185/30000 [16:27:40<1:17:03, 1.64s/it] 91%|█████████ | 27186/30000 [16:27:41<1:10:20, 1.50s/it] 91%|█████████ | 27187/30000 [16:27:56<4:18:01, 5.50s/it] 91%|█████████ | 27188/30000 [16:27:57<3:17:00, 4.20s/it] 91%|█████████ | 27189/30000 [16:27:58<2:34:59, 3.31s/it] 91%|█████████ | 27190/30000 [16:27:59<2:04:54, 2.67s/it] 91%|█████████ | 27190/30000 [16:27:59<2:04:54, 2.67s/it] 91%|█████████ | 27191/30000 [16:28:00<1:43:56, 2.22s/it] 91%|█████████ | 27192/30000 [16:28:02<1:29:57, 1.92s/it] 91%|█████████ | 27193/30000 [16:28:03<1:20:45, 1.73s/it] 91%|█████████ | 27194/30000 [16:28:04<1:12:59, 1.56s/it] 91%|█████████ | 27195/30000 [16:28:20<4:33:36, 5.85s/it] 91%|█████████ | 27196/30000 [16:28:21<3:27:59, 4.45s/it] 91%|█████████ | 27197/30000 [16:28:22<2:42:10, 3.47s/it] 91%|█████████ | 27198/30000 [16:28:23<2:10:22, 2.79s/it] 91%|█████████ | 27199/30000 [16:28:25<1:48:02, 2.31s/it] 91%|█████████ | 27200/30000 [16:28:26<1:32:11, 1.98s/it] 91%|█████████ | 27200/30000 [16:28:26<1:32:11, 1.98s/it] 91%|█████████ | 27201/30000 [16:28:27<1:21:11, 1.74s/it] 91%|█████████ | 27202/30000 [16:28:28<1:13:37, 1.58s/it] 91%|█████████ | 27203/30000 [16:28:43<4:16:56, 5.51s/it] 91%|█████████ | 27204/30000 [16:28:44<3:16:19, 4.21s/it] 91%|█████████ | 27205/30000 [16:28:45<2:34:09, 3.31s/it] 91%|█████████ | 27206/30000 [16:28:46<2:04:22, 2.67s/it] 91%|█████████ | 27207/30000 [16:28:48<1:44:04, 2.24s/it] 91%|█████████ | 27208/30000 [16:28:49<1:32:16, 1.98s/it] 91%|█████████ | 27209/30000 [16:28:50<1:23:29, 1.80s/it] 91%|█████████ | 27210/30000 [16:28:52<1:16:01, 1.64s/it] 91%|█████████ | 27210/30000 [16:28:52<1:16:01, 1.64s/it] 91%|█████████ | 27211/30000 [16:29:04<3:43:45, 4.81s/it] 91%|█████████ | 27212/30000 [16:29:05<2:52:58, 3.72s/it] 91%|█████████ | 27213/30000 [16:29:06<2:18:01, 2.97s/it] 91%|█████████ | 27214/30000 [16:29:08<1:53:22, 2.44s/it] 91%|█████████ | 27215/30000 [16:29:09<1:36:11, 2.07s/it] 91%|█████████ | 27216/30000 [16:29:10<1:23:55, 1.81s/it] 91%|█████████ | 27217/30000 [16:29:11<1:15:26, 1.63s/it] 91%|█████████ | 27218/30000 [16:29:12<1:09:02, 1.49s/it] 91%|█████████ | 27219/30000 [16:29:25<3:48:54, 4.94s/it] 91%|█████████ | 27220/30000 [16:29:26<2:56:46, 3.82s/it] 91%|█████████ | 27220/30000 [16:29:27<2:56:46, 3.82s/it] 91%|█████████ | 27221/30000 [16:29:28<2:20:35, 3.04s/it] 91%|█████████ | 27222/30000 [16:29:29<1:54:46, 2.48s/it] 91%|█████████ | 27223/30000 [16:29:30<1:37:04, 2.10s/it] 91%|█████████ | 27224/30000 [16:29:31<1:24:44, 1.83s/it] 91%|█████████ | 27225/30000 [16:29:32<1:15:54, 1.64s/it] 91%|█████████ | 27226/30000 [16:29:34<1:09:46, 1.51s/it] 91%|█████████ | 27227/30000 [16:29:49<4:19:04, 5.61s/it] 91%|█████████ | 27228/30000 [16:29:50<3:17:35, 4.28s/it] 91%|█████████ | 27229/30000 [16:29:51<2:34:59, 3.36s/it] 91%|█████████ | 27230/30000 [16:29:52<2:05:16, 2.71s/it] 91%|█████████ | 27230/30000 [16:29:52<2:05:16, 2.71s/it] 91%|█████████ | 27231/30000 [16:29:54<1:44:06, 2.26s/it] 91%|█████████ | 27232/30000 [16:29:55<1:29:27, 1.94s/it] 91%|█████████ | 27233/30000 [16:29:56<1:18:59, 1.71s/it] 91%|█████████ | 27234/30000 [16:29:57<1:12:04, 1.56s/it] 91%|█████████ | 27235/30000 [16:30:12<4:11:30, 5.46s/it] 91%|█████████ | 27236/30000 [16:30:13<3:12:18, 4.17s/it] 91%|█████████ | 27237/30000 [16:30:14<2:30:59, 3.28s/it] 91%|█████████ | 27238/30000 [16:30:15<2:02:03, 2.65s/it] 91%|█████████ | 27239/30000 [16:30:17<1:41:34, 2.21s/it] 91%|█████████ | 27240/30000 [16:30:18<1:29:07, 1.94s/it] 91%|█████████ | 27240/30000 [16:30:18<1:29:07, 1.94s/it] 91%|█████████ | 27241/30000 [16:30:19<1:19:11, 1.72s/it] 91%|█████████ | 27242/30000 [16:30:20<1:11:49, 1.56s/it] 91%|█████████ | 27243/30000 [16:30:36<4:25:42, 5.78s/it] 91%|█████████ | 27244/30000 [16:30:37<3:22:22, 4.41s/it] 91%|█████████ | 27245/30000 [16:30:38<2:40:03, 3.49s/it] 91%|█████████ | 27246/30000 [16:30:40<2:08:38, 2.80s/it] 91%|█████████ | 27247/30000 [16:30:41<1:48:37, 2.37s/it] 91%|█████████ | 27248/30000 [16:30:42<1:32:13, 2.01s/it] 91%|█████████ | 27249/30000 [16:30:43<1:21:52, 1.79s/it] 91%|█████████ | 27250/30000 [16:30:45<1:15:32, 1.65s/it] 91%|█████████ | 27250/30000 [16:30:45<1:15:32, 1.65s/it] 91%|█████████ | 27251/30000 [16:30:57<3:38:45, 4.77s/it] 91%|█████████ | 27252/30000 [16:30:58<2:49:21, 3.70s/it] 91%|█████████ | 27253/30000 [16:30:59<2:15:40, 2.96s/it] 91%|█████████ | 27254/30000 [16:31:00<1:51:32, 2.44s/it] 91%|█████████ | 27255/30000 [16:31:02<1:34:25, 2.06s/it] 91%|█████████ | 27256/30000 [16:31:03<1:23:36, 1.83s/it] 91%|█████████ | 27257/30000 [16:31:04<1:16:41, 1.68s/it] 91%|█████████ | 27258/30000 [16:31:06<1:13:24, 1.61s/it] 91%|█████████ | 27259/30000 [16:31:19<3:57:15, 5.19s/it] 91%|█████████ | 27260/30000 [16:31:20<3:02:26, 3.99s/it] 91%|█████████ | 27260/30000 [16:31:20<3:02:26, 3.99s/it] 91%|█████████ | 27261/30000 [16:31:22<2:24:05, 3.16s/it] 91%|█████████ | 27262/30000 [16:31:23<1:57:03, 2.57s/it] 91%|█████████ | 27263/30000 [16:31:24<1:38:37, 2.16s/it] 91%|█████████ | 27264/30000 [16:31:25<1:25:41, 1.88s/it] 91%|█████████ | 27265/30000 [16:31:26<1:16:29, 1.68s/it] 91%|█████████ | 27266/30000 [16:31:28<1:10:14, 1.54s/it] 91%|█████████ | 27267/30000 [16:31:44<4:26:18, 5.85s/it] 91%|█████████ | 27268/30000 [16:31:45<3:22:20, 4.44s/it] 91%|█████████ | 27269/30000 [16:31:46<2:37:53, 3.47s/it] 91%|█████████ | 27270/30000 [16:31:47<2:06:41, 2.78s/it] 91%|█████████ | 27270/30000 [16:31:47<2:06:41, 2.78s/it] 91%|█████████ | 27271/30000 [16:31:48<1:46:30, 2.34s/it] 91%|█████████ | 27272/30000 [16:31:50<1:31:55, 2.02s/it] 91%|█████████ | 27273/30000 [16:31:51<1:20:47, 1.78s/it] 91%|█████████ | 27274/30000 [16:31:52<1:13:00, 1.61s/it] 91%|█████████ | 27275/30000 [16:32:04<3:33:46, 4.71s/it] 91%|█████████ | 27276/30000 [16:32:05<2:45:53, 3.65s/it] 91%|█████████ | 27277/30000 [16:32:07<2:13:05, 2.93s/it] 91%|█████████ | 27278/30000 [16:32:08<1:49:15, 2.41s/it] 91%|█████████ | 27279/30000 [16:32:09<1:32:44, 2.04s/it] 91%|█████████ | 27280/30000 [16:32:10<1:21:31, 1.80s/it] 91%|█████████ | 27280/30000 [16:32:10<1:21:31, 1.80s/it] 91%|█████████ | 27281/30000 [16:32:11<1:13:22, 1.62s/it] 91%|█████████ | 27282/30000 [16:32:13<1:07:36, 1.49s/it] 91%|█████████ | 27283/30000 [16:32:24<3:24:15, 4.51s/it] 91%|█████████ | 27284/30000 [16:32:25<2:38:41, 3.51s/it] 91%|█████████ | 27285/30000 [16:32:26<2:06:48, 2.80s/it] 91%|█████████ | 27286/30000 [16:32:28<1:44:51, 2.32s/it] 91%|█████████ | 27287/30000 [16:32:29<1:30:50, 2.01s/it] 91%|█████████ | 27288/30000 [16:32:30<1:21:33, 1.80s/it] 91%|█████████ | 27289/30000 [16:32:31<1:12:55, 1.61s/it] 91%|█████████ | 27290/30000 [16:32:33<1:07:00, 1.48s/it] 91%|█████████ | 27290/30000 [16:32:33<1:07:00, 1.48s/it] 91%|█████████ | 27291/30000 [16:32:40<2:29:04, 3.30s/it] 91%|█████████ | 27292/30000 [16:32:41<1:59:56, 2.66s/it] 91%|█████████ | 27293/30000 [16:32:42<1:39:49, 2.21s/it] 91%|█████████ | 27294/30000 [16:32:44<1:25:21, 1.89s/it] 91%|█████████ | 27295/30000 [16:32:45<1:15:59, 1.69s/it] 91%|█████████ | 27296/30000 [16:32:46<1:09:07, 1.53s/it] 91%|█████████ | 27297/30000 [16:32:47<1:04:49, 1.44s/it] 91%|█████████ | 27298/30000 [16:32:48<1:02:13, 1.38s/it] 91%|█████████ | 27299/30000 [16:32:55<2:14:53, 3.00s/it] 91%|█████████ | 27300/30000 [16:32:56<1:49:58, 2.44s/it] 91%|█████████ | 27300/30000 [16:32:56<1:49:58, 2.44s/it] 91%|█████████ | 27301/30000 [16:32:58<1:32:54, 2.07s/it] 91%|█████████ | 27302/30000 [16:32:59<1:20:53, 1.80s/it] 91%|█████████ | 27303/30000 [16:33:00<1:12:37, 1.62s/it] 91%|█████████ | 27304/30000 [16:33:01<1:06:32, 1.48s/it] 91%|█████████ | 27305/30000 [16:33:02<1:02:21, 1.39s/it] 91%|█████████ | 27306/30000 [16:33:03<59:34, 1.33s/it] 91%|█████████ | 27307/30000 [16:33:10<2:05:04, 2.79s/it] 91%|█████████ | 27308/30000 [16:33:11<1:43:19, 2.30s/it] 91%|█████████ | 27309/30000 [16:33:12<1:28:25, 1.97s/it] 91%|█████████ | 27310/30000 [16:33:13<1:17:21, 1.73s/it] 91%|█████████ | 27310/30000 [16:33:13<1:17:21, 1.73s/it] 91%|█████████ | 27311/30000 [16:33:14<1:11:33, 1.60s/it] 91%|█████████ | 27312/30000 [16:33:16<1:06:08, 1.48s/it] 91%|█████████ | 27313/30000 [16:33:17<1:02:41, 1.40s/it] 91%|█████████ | 27314/30000 [16:33:18<1:00:04, 1.34s/it] 91%|█████████ | 27315/30000 [16:33:25<2:12:54, 2.97s/it] 91%|█████████ | 27316/30000 [16:33:26<1:48:33, 2.43s/it] 91%|█████████ | 27317/30000 [16:33:27<1:31:52, 2.05s/it] 91%|█████████ | 27318/30000 [16:33:28<1:20:01, 1.79s/it] 91%|█████████ | 27319/30000 [16:33:30<1:11:57, 1.61s/it] 91%|█████████ | 27320/30000 [16:33:31<1:06:40, 1.49s/it] {'loss': 0.0133, 'grad_norm': 0.07360409945249557, 'learning_rate': 8.623104743875493e-07, 'flow_matching_loss': 0.01329895220696926, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.83} +{'loss': 0.0117, 'grad_norm': 0.06889282166957855, 'learning_rate': 8.567936562219547e-07, 'flow_matching_loss': 0.011700825206935406, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.83} +{'loss': 0.0124, 'grad_norm': 0.05514807626605034, 'learning_rate': 8.512940233856847e-07, 'flow_matching_loss': 0.012411681888625025, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.84} +{'loss': 0.0136, 'grad_norm': 0.07261635363101959, 'learning_rate': 8.458115825613122e-07, 'flow_matching_loss': 0.013622082956135272, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.84} +{'loss': 0.0134, 'grad_norm': 0.06534697115421295, 'learning_rate': 8.403463404105199e-07, 'flow_matching_loss': 0.013382064178586006, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.84} +{'loss': 0.0139, 'grad_norm': 0.0829702615737915, 'learning_rate': 8.348983035740926e-07, 'flow_matching_loss': 0.013899722788482904, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.84} +{'loss': 0.0139, 'grad_norm': 0.06289325654506683, 'learning_rate': 8.294674786719064e-07, 'flow_matching_loss': 0.01394154028967023, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.85} +{'loss': 0.0122, 'grad_norm': 0.06706035882234573, 'learning_rate': 8.240538723029295e-07, 'flow_matching_loss': 0.012227454036474229, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.85} +{'loss': 0.014, 'grad_norm': 0.09092912077903748, 'learning_rate': 8.186574910451983e-07, 'flow_matching_loss': 0.014013869501650334, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.85} +{'loss': 0.0136, 'grad_norm': 0.07205277681350708, 'learning_rate': 8.132783414558326e-07, 'flow_matching_loss': 0.013561711274087429, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.86} +{'loss': 0.0127, 'grad_norm': 0.05128224194049835, 'learning_rate': 8.079164300710023e-07, 'flow_matching_loss': 0.012738823611289262, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.86} +{'loss': 0.0129, 'grad_norm': 0.07306075096130371, 'learning_rate': 8.025717634059326e-07, 'flow_matching_loss': 0.012938710395246744, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.86} +{'loss': 0.0127, 'grad_norm': 0.050664253532886505, 'learning_rate': 7.972443479549035e-07, 'flow_matching_loss': 0.012652888894081116, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.86} +{'loss': 0.0133, 'grad_norm': 0.06696998327970505, 'learning_rate': 7.919341901912236e-07, 'flow_matching_loss': 0.013250813633203507, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.87} +{'loss': 0.013, 'grad_norm': 0.07620339840650558, 'learning_rate': 7.866412965672365e-07, 'flow_matching_loss': 0.01301035787910223, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.87} +{'loss': 0.0131, 'grad_norm': 0.05864829197525978, 'learning_rate': 7.813656735143044e-07, 'flow_matching_loss': 0.013091527577489614, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.87} +{'loss': 0.0154, 'grad_norm': 0.07775416225194931, 'learning_rate': 7.761073274428076e-07, 'flow_matching_loss': 0.015439269319176673, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.88} +{'loss': 0.0121, 'grad_norm': 0.07371184974908829, 'learning_rate': 7.708662647421355e-07, 'flow_matching_loss': 0.01208119154907763, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.88} +{'loss': 0.0134, 'grad_norm': 0.07182905077934265, 'learning_rate': 7.65642491780672e-07, 'flow_matching_loss': 0.013441320974379778, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.88} +{'loss': 0.0118, 'grad_norm': 0.07131637632846832, 'learning_rate': 7.604360149057916e-07, 'flow_matching_loss': 0.011770930513739586, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.88} +{'loss': 0.0127, 'grad_norm': 0.06346842646598816, 'learning_rate': 7.552468404438568e-07, 'flow_matching_loss': 0.012733687087893486, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.89} +{'loss': 0.0123, 'grad_norm': 0.08467470109462738, 'learning_rate': 7.500749747002022e-07, 'flow_matching_loss': 0.012267770152539015, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.89} +{'loss': 0.0142, 'grad_norm': 0.05807049199938774, 'learning_rate': 7.449204239591345e-07, 'flow_matching_loss': 0.014162540528923273, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.89} +{'loss': 0.0122, 'grad_norm': 0.06340792775154114, 'learning_rate': 7.397831944839167e-07, 'flow_matching_loss': 0.012174832914024592, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.9} +{'loss': 0.0123, 'grad_norm': 0.0591316893696785, 'learning_rate': 7.346632925167706e-07, 'flow_matching_loss': 0.012255798745900392, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.9} +{'loss': 0.0125, 'grad_norm': 0.07291005551815033, 'learning_rate': 7.295607242788566e-07, 'flow_matching_loss': 0.012497935723513365, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.9} +{'loss': 0.0137, 'grad_norm': 0.07010766863822937, 'learning_rate': 7.244754959702765e-07, 'flow_matching_loss': 0.01368441954255104, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.91} +{'loss': 0.0136, 'grad_norm': 0.09366802871227264, 'learning_rate': 7.194076137700628e-07, 'flow_matching_loss': 0.013609485048800708, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.91} +{'loss': 0.0122, 'grad_norm': 0.07857147604227066, 'learning_rate': 7.143570838361696e-07, 'flow_matching_loss': 0.012198489531874657, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.91} +{'loss': 0.0117, 'grad_norm': 0.07387354224920273, 'learning_rate': 7.093239123054696e-07, 'flow_matching_loss': 0.011665806733071804, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.91} +{'loss': 0.0127, 'grad_norm': 0.06353069096803665, 'learning_rate': 7.043081052937389e-07, 'flow_matching_loss': 0.012706049717962742, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.92} +{'loss': 0.0124, 'grad_norm': 0.06387545168399811, 'learning_rate': 6.993096688956558e-07, 'flow_matching_loss': 0.01239713579416275, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.92} +{'loss': 0.0139, 'grad_norm': 0.06471024453639984, 'learning_rate': 6.943286091847884e-07, 'flow_matching_loss': 0.01392208132892847, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.92} +{'loss': 0.013, 'grad_norm': 0.08426330238580704, 'learning_rate': 6.89364932213602e-07, 'flow_matching_loss': 0.013023110665380954, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.93} +{'loss': 0.013, 'grad_norm': 0.06377990543842316, 'learning_rate': 6.844186440134253e-07, 'flow_matching_loss': 0.01304463716223836, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.93} +{'loss': 0.0118, 'grad_norm': 0.05850611999630928, 'learning_rate': 6.794897505944692e-07, 'flow_matching_loss': 0.011838026624172927, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.93} +{'loss': 0.0116, 'grad_norm': 0.06821798533201218, 'learning_rate': 6.745782579457993e-07, 'flow_matching_loss': 0.011643405631184578, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.93} +{'loss': 0.0135, 'grad_norm': 0.09847799688577652, 'learning_rate': 6.696841720353453e-07, 'flow_matching_loss': 0.013462896458804607, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.94} +{'loss': 0.0124, 'grad_norm': 0.0605456717312336, 'learning_rate': 6.648074988098834e-07, 'flow_matching_loss': 0.012438499554991721, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.94} +{'loss': 0.0123, 'grad_norm': 0.08166225999593735, 'learning_rate': 6.599482441950322e-07, 'flow_matching_loss': 0.012334281299263238, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.94} +{'loss': 0.0137, 'grad_norm': 0.08016535639762878, 'learning_rate': 6.551064140952417e-07, 'flow_matching_loss': 0.013688144274055958, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.95} +{'loss': 0.0128, 'grad_norm': 0.07792257517576218, 'learning_rate': 6.502820143937943e-07, 'flow_matching_loss': 0.012790690921247006, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.95} 91%|█████████ | 27320/30000 [16:33:31<1:06:40, 1.49s/it] 91%|█████████ | 27321/30000 [16:33:32<1:02:58, 1.41s/it] 91%|█████████ | 27322/30000 [16:33:33<1:00:34, 1.36s/it] 91%|█████████ | 27323/30000 [16:33:40<2:15:47, 3.04s/it] 91%|█████████ | 27324/30000 [16:33:41<1:50:36, 2.48s/it] 91%|█████████ | 27325/30000 [16:33:43<1:33:27, 2.10s/it] 91%|█████████ | 27326/30000 [16:33:44<1:21:00, 1.82s/it] 91%|█████████ | 27327/30000 [16:33:45<1:12:21, 1.62s/it] 91%|█████████ | 27328/30000 [16:33:46<1:06:31, 1.49s/it] 91%|█████████ | 27329/30000 [16:33:47<1:02:16, 1.40s/it] 91%|█████████ | 27330/30000 [16:33:48<59:22, 1.33s/it] 91%|█████████ | 27330/30000 [16:33:48<59:22, 1.33s/it] 91%|█████████ | 27331/30000 [16:33:55<2:09:32, 2.91s/it] 91%|█████████ | 27332/30000 [16:33:56<1:46:33, 2.40s/it] 91%|█████████ | 27333/30000 [16:33:57<1:30:48, 2.04s/it] 91%|█████████ | 27334/30000 [16:33:59<1:18:54, 1.78s/it] 91%|█████████ | 27335/30000 [16:34:00<1:11:05, 1.60s/it] 91%|█████████ | 27336/30000 [16:34:01<1:05:25, 1.47s/it] 91%|█████████ | 27337/30000 [16:34:02<1:01:44, 1.39s/it] 91%|█████████ | 27338/30000 [16:34:03<58:54, 1.33s/it] 91%|█████████ | 27339/30000 [16:34:10<2:15:38, 3.06s/it] 91%|█████████ | 27340/30000 [16:34:12<1:50:24, 2.49s/it] 91%|█████████ | 27340/30000 [16:34:12<1:50:24, 2.49s/it] 91%|█████████ | 27341/30000 [16:34:13<1:32:43, 2.09s/it] 91%|█████████ | 27342/30000 [16:34:14<1:20:16, 1.81s/it] 91%|█████████ | 27343/30000 [16:34:15<1:11:57, 1.62s/it] 91%|█████████ | 27344/30000 [16:34:16<1:06:07, 1.49s/it] 91%|█████████ | 27345/30000 [16:34:18<1:02:19, 1.41s/it] 91%|█████████ | 27346/30000 [16:34:19<59:13, 1.34s/it] 91%|█████████ | 27347/30000 [16:34:26<2:14:57, 3.05s/it] 91%|█████████ | 27348/30000 [16:34:27<1:49:55, 2.49s/it] 91%|█████████ | 27349/30000 [16:34:28<1:32:44, 2.10s/it] 91%|█████████ | 27350/30000 [16:34:29<1:20:29, 1.82s/it] 91%|█████████ | 27350/30000 [16:34:29<1:20:29, 1.82s/it] 91%|█████████ | 27351/30000 [16:34:30<1:11:48, 1.63s/it] 91%|█████████ | 27352/30000 [16:34:32<1:06:36, 1.51s/it] 91%|█████████ | 27353/30000 [16:34:33<1:02:13, 1.41s/it] 91%|█████████ | 27354/30000 [16:34:34<58:48, 1.33s/it] 91%|█████████ | 27355/30000 [16:34:41<2:07:44, 2.90s/it] 91%|█████████ | 27356/30000 [16:34:42<1:45:30, 2.39s/it] 91%|█████████ | 27357/30000 [16:34:43<1:29:37, 2.03s/it] 91%|█████████ | 27358/30000 [16:34:44<1:17:57, 1.77s/it] 91%|█████████ | 27359/30000 [16:34:45<1:10:23, 1.60s/it] 91%|█████████ | 27360/30000 [16:34:46<1:04:39, 1.47s/it] 91%|█████████ | 27360/30000 [16:34:47<1:04:39, 1.47s/it] 91%|█████████ | 27361/30000 [16:34:48<1:00:52, 1.38s/it] 91%|█████████ | 27362/30000 [16:34:49<57:56, 1.32s/it] 91%|█████████ | 27363/30000 [16:34:55<2:01:08, 2.76s/it] 91%|█████████ | 27364/30000 [16:34:56<1:40:13, 2.28s/it] 91%|█████████ | 27365/30000 [16:34:57<1:25:45, 1.95s/it] 91%|█████████ | 27366/30000 [16:34:58<1:15:17, 1.72s/it] 91%|█████████ | 27367/30000 [16:35:00<1:08:34, 1.56s/it] 91%|█████████ | 27368/30000 [16:35:01<1:04:37, 1.47s/it] 91%|█████████ | 27369/30000 [16:35:02<1:01:36, 1.41s/it] 91%|█████████ | 27370/30000 [16:35:03<58:53, 1.34s/it] 91%|█████████ | 27370/30000 [16:35:03<58:53, 1.34s/it] 91%|█████████ | 27371/30000 [16:35:10<2:07:35, 2.91s/it] 91%|█████████ | 27372/30000 [16:35:11<1:44:28, 2.39s/it] 91%|█████████ | 27373/30000 [16:35:12<1:28:42, 2.03s/it] 91%|█████████ | 27374/30000 [16:35:13<1:17:12, 1.76s/it] 91%|█████████▏| 27375/30000 [16:35:15<1:09:23, 1.59s/it] 91%|█████████▏| 27376/30000 [16:35:16<1:04:53, 1.48s/it] 91%|█████████▏| 27377/30000 [16:35:17<1:00:51, 1.39s/it] 91%|█████████▏| 27378/30000 [16:35:18<58:10, 1.33s/it] 91%|█████████▏| 27379/30000 [16:35:25<2:11:25, 3.01s/it] 91%|█████████▏| 27380/30000 [16:35:26<1:47:11, 2.45s/it] 91%|█████████▏| 27380/30000 [16:35:26<1:47:11, 2.45s/it] 91%|█████████▏| 27381/30000 [16:35:28<1:30:59, 2.08s/it] 91%|█████████▏| 27382/30000 [16:35:29<1:19:02, 1.81s/it] 91%|█████████▏| 27383/30000 [16:35:30<1:10:36, 1.62s/it] 91%|█████████▏| 27384/30000 [16:35:31<1:05:13, 1.50s/it] 91%|█████████▏| 27385/30000 [16:35:32<1:01:02, 1.40s/it] 91%|█████████▏| 27386/30000 [16:35:33<58:06, 1.33s/it] 91%|█████████▏| 27387/30000 [16:35:40<2:08:27, 2.95s/it] 91%|█████████▏| 27388/30000 [16:35:41<1:45:07, 2.41s/it] 91%|█████████▏| 27389/30000 [16:35:43<1:29:03, 2.05s/it] 91%|█████████▏| 27390/30000 [16:35:44<1:18:12, 1.80s/it] 91%|█████████▏| 27390/30000 [16:35:44<1:18:12, 1.80s/it] 91%|█████████▏| 27391/30000 [16:35:45<1:10:31, 1.62s/it] 91%|█████████▏| 27392/30000 [16:35:46<1:04:51, 1.49s/it] 91%|█████████▏| 27393/30000 [16:35:47<1:00:42, 1.40s/it] 91%|█████████▏| 27394/30000 [16:35:49<57:44, 1.33s/it] 91%|█████████▏| 27395/30000 [16:35:56<2:12:21, 3.05s/it] 91%|█████████▏| 27396/30000 [16:35:57<1:47:40, 2.48s/it] 91%|█████████▏| 27397/30000 [16:35:58<1:30:28, 2.09s/it] 91%|█████████▏| 27398/30000 [16:35:59<1:18:42, 1.82s/it] 91%|█████████▏| 27399/30000 [16:36:00<1:10:02, 1.62s/it] 91%|█████████▏| 27400/30000 [16:36:01<1:04:20, 1.48s/it] 91%|█████████▏| 27400/30000 [16:36:01<1:04:20, 1.48s/it] 91%|█████████▏| 27401/30000 [16:36:03<1:00:00, 1.39s/it] 91%|█████████▏| 27402/30000 [16:36:04<57:01, 1.32s/it] 91%|█████████▏| 27403/30000 [16:36:10<2:07:50, 2.95s/it] 91%|█████████▏| 27404/30000 [16:36:12<1:44:36, 2.42s/it] 91%|█████████▏| 27405/30000 [16:36:13<1:28:48, 2.05s/it] 91%|█████████▏| 27406/30000 [16:36:14<1:17:25, 1.79s/it] 91%|█████████▏| 27407/30000 [16:36:15<1:09:34, 1.61s/it] 91%|█████████▏| 27408/30000 [16:36:16<1:04:02, 1.48s/it] 91%|█████████▏| 27409/30000 [16:36:18<59:53, 1.39s/it] 91%|█████████▏| 27410/30000 [16:36:19<57:06, 1.32s/it] 91%|█████████▏| 27410/30000 [16:36:19<57:06, 1.32s/it] 91%|█████████▏| 27411/30000 [16:36:24<1:50:40, 2.56s/it] 91%|█████████▏| 27412/30000 [16:36:25<1:32:27, 2.14s/it] 91%|█████████▏| 27413/30000 [16:36:27<1:20:08, 1.86s/it] 91%|█████████▏| 27414/30000 [16:36:28<1:11:10, 1.65s/it] 91%|█████████▏| 27415/30000 [16:36:29<1:05:16, 1.52s/it] 91%|█████████▏| 27416/30000 [16:36:30<1:01:01, 1.42s/it] 91%|█████████▏| 27417/30000 [16:36:31<59:31, 1.38s/it] 91%|█████████▏| 27418/30000 [16:36:33<56:45, 1.32s/it] 91%|█████████▏| 27419/30000 [16:36:38<1:49:00, 2.53s/it] 91%|█████████▏| 27420/30000 [16:36:39<1:31:22, 2.12s/it] 91%|█████████▏| 27420/30000 [16:36:39<1:31:22, 2.12s/it] 91%|█████████▏| 27421/30000 [16:36:40<1:19:13, 1.84s/it] 91%|█████████▏| 27422/30000 [16:36:41<1:10:32, 1.64s/it] 91%|█████████▏| 27423/30000 [16:36:43<1:05:15, 1.52s/it] 91%|█████████▏| 27424/30000 [16:36:44<1:01:21, 1.43s/it] 91%|█████████▏| 27425/30000 [16:36:45<58:15, 1.36s/it] 91%|█████████▏| 27426/30000 [16:36:46<57:20, 1.34s/it] 91%|█████████▏| 27427/30000 [16:36:52<1:57:58, 2.75s/it] 91%|█████████▏| 27428/30000 [16:36:54<1:37:29, 2.27s/it] 91%|█████████▏| 27429/30000 [16:36:55<1:23:34, 1.95s/it] 91%|█████████▏| 27430/30000 [16:36:56<1:13:20, 1.71s/it] 91%|█████████▏| 27430/30000 [16:36:56<1:13:20, 1.71s/it] 91%|█████████▏| 27431/30000 [16:36:57<1:06:24, 1.55s/it] 91%|█████████▏| 27432/30000 [16:36:58<1:01:44, 1.44s/it] 91%|█████████▏| 27433/30000 [16:36:59<58:08, 1.36s/it] 91%|█████████▏| 27434/30000 [16:37:01<55:39, 1.30s/it] 91%|█████████▏| 27435/30000 [16:37:07<1:54:52, 2.69s/it] 91%|█████████▏| 27436/30000 [16:37:08<1:35:04, 2.22s/it] 91%|█████████▏| 27437/30000 [16:37:09<1:21:43, 1.91s/it] 91%|█████████▏| 27438/30000 [16:37:10<1:12:09, 1.69s/it] 91%|█████████▏| 27439/30000 [16:37:11<1:05:43, 1.54s/it] 91%|█████████▏| 27440/30000 [16:37:12<1:01:00, 1.43s/it] 91%|█████████▏| 27440/30000 [16:37:12<1:01:00, 1.43s/it] 91%|█████████▏| 27441/30000 [16:37:14<58:04, 1.36s/it] 91%|█████████▏| 27442/30000 [16:37:15<55:41, 1.31s/it] 91%|█████████▏| 27443/30000 [16:37:21<1:54:03, 2.68s/it] 91%|█████████▏| 27444/30000 [16:37:22<1:35:41, 2.25s/it] 91%|█████████▏| 27445/30000 [16:37:23<1:22:46, 1.94s/it] 91%|█████████▏| 27446/30000 [16:37:24<1:13:19, 1.72s/it] 91%|█████████▏| 27447/30000 [16:37:26<1:06:27, 1.56s/it] 91%|█████████▏| 27448/30000 [16:37:27<1:01:34, 1.45s/it] 91%|█████████▏| 27449/30000 [16:37:28<58:12, 1.37s/it] 92%|█████████▏| 27450/30000 [16:37:29<55:36, 1.31s/it] 92%|█████████▏| 27450/30000 [16:37:29<55:36, 1.31s/it] 92%|█████████▏| 27451/30000 [16:37:37<2:15:46, 3.20s/it] 92%|█████████▏| 27452/30000 [16:37:38<1:49:33, 2.58s/it] 92%|█████████▏| 27453/30000 [16:37:39<1:31:46, 2.16s/it] 92%|█████████▏| 27454/30000 [16:37:40<1:19:10, 1.87s/it] 92%|█████████▏| 27455/30000 [16:37:41<1:10:33, 1.66s/it] 92%|█████████▏| 27456/30000 [16:37:43<1:04:21, 1.52s/it] 92%|█████████▏| 27457/30000 [16:37:44<1:00:00, 1.42s/it] 92%|█████████▏| 27458/30000 [16:37:45<57:00, 1.35s/it] 92%|█████████▏| 27459/30000 [16:37:53<2:25:02, 3.42s/it] 92%|█████████▏| 27460/30000 [16:37:54<1:55:57, 2.74s/it] 92%|█████████▏| 27460/30000 [16:37:54<1:55:57, 2.74s/it] 92%|█████████▏| 27461/30000 [16:37:56<1:35:50, 2.26s/it] 92%|█████████▏| 27462/30000 [16:37:57<1:21:42, 1.93s/it] 92%|█████████▏| 27463/30000 [16:37:58<1:12:35, 1.72s/it] 92%|█████████▏| 27464/30000 [16:37:59<1:05:50, 1.56s/it] 92%|█████████▏| 27465/30000 [16:38:00<1:00:53, 1.44s/it] 92%|█████████▏| 27466/30000 [16:38:01<57:30, 1.36s/it] 92%|█████████▏| 27467/30000 [16:38:10<2:34:55, 3.67s/it] 92%|█████████▏| 27468/30000 [16:38:12<2:03:01, 2.92s/it] 92%|█████████▏| 27469/30000 [16:38:13<1:40:55, 2.39s/it] 92%|█████████▏| 27470/30000 [16:38:14<1:25:19, 2.02s/it] 92%|█████████▏| 27470/30000 [16:38:14<1:25:19, 2.02s/it] 92%|█████████▏| 27471/30000 [16:38:15<1:14:20, 1.76s/it] 92%|█████████▏| 27472/30000 [16:38:16<1:07:06, 1.59s/it] 92%|█████████▏| 27473/30000 [16:38:18<1:01:58, 1.47s/it] 92%|█████████▏| 27474/30000 [16:38:19<58:27, 1.39s/it] 92%|█████████▏| 27475/30000 [16:38:29<2:48:14, 4.00s/it] 92%|█████████▏| 27476/30000 [16:38:30<2:12:02, 3.14s/it] 92%|█████████▏| 27477/30000 [16:38:31<1:46:59, 2.54s/it] 92%|█████████▏| 27478/30000 [16:38:32<1:29:21, 2.13s/it] 92%|█████████▏| 27479/30000 [16:38:33<1:17:22, 1.84s/it] 92%|█████████▏| 27480/30000 [16:38:35<1:09:10, 1.65s/it] 92%|█████████▏| 27480/30000 [16:38:35<1:09:10, 1.65s/it] 92%|█████████▏| 27481/30000 [16:38:36<1:04:05, 1.53s/it] 92%|█████████▏| 27482/30000 [16:38:37<59:42, 1.42s/it] 92%|█████████▏| 27483/30000 [16:38:47<2:45:28, 3.94s/it] 92%|█████████▏| 27484/30000 [16:38:48<2:10:10, 3.10s/it] 92%|█████████▏| 27485/30000 [16:38:49<1:45:40, 2.52s/it] 92%|█████████▏| 27486/30000 [16:38:50<1:28:34, 2.11s/it] 92%|█████████▏| 27487/30000 [16:38:52<1:16:47, 1.83s/it] 92%|█████████▏| 27488/30000 [16:38:53<1:08:15, 1.63s/it] 92%|█████████▏| 27489/30000 [16:38:54<1:02:27, 1.49s/it] 92%|█████████▏| 27490/30000 [16:38:55<58:20, 1.39s/it] 92%|█████████▏| 27490/30000 [16:38:55<58:20, 1.39s/it] 92%|█████████▏| 27491/30000 [16:39:07<3:06:20, 4.46s/it] 92%|█████████▏| 27492/30000 [16:39:08<2:24:32, 3.46s/it] 92%|█████████▏| 27493/30000 [16:39:09<1:55:46, 2.77s/it] 92%|█████████▏| 27494/30000 [16:39:10<1:35:19, 2.28s/it] 92%|█████████▏| 27495/30000 [16:39:11<1:21:04, 1.94s/it] 92%|█████████▏| 27496/30000 [16:39:12<1:02:25, 1.50s/it] 92%|█████████▏| 27497/30000 [16:39:28<4:07:08, 5.92s/it] 92%|█████████▏| 27498/30000 [16:39:29<3:07:47, 4.50s/it] 92%|█████████▏| 27499/30000 [16:39:30<2:25:43, 3.50s/it] 92%|█████████▏| 27500/30000 [16:39:31<1:56:23, 2.79s/it] 92%|█████████▏| 27500/30000 [16:39:31<1:56:23, 2.79s/it] 92%|█████████▏| 27501/30000 [16:39:33<1:36:02, 2.31s/it] 92%|█████████▏| 27502/30000 [16:39:34<1:21:48, 1.97s/it] 92%|█████████▏| 27503/30000 [16:39:35<1:11:45, 1.72s/it] 92%|█████████▏| 27504/30000 [16:39:43<2:32:16, 3.66s/it] 92%|█████████▏| 27505/30000 [16:39:44<2:00:57, 2.91s/it] 92%|█████████▏| 27506/30000 [16:39:45<1:39:14, 2.39s/it] 92%|█████████▏| 27507/30000 [16:39:47<1:24:13, 2.03s/it] 92%|█████████▏| 27508/30000 [16:39:48<1:13:30, 1.77s/it] 92%|█████████▏| 27509/30000 [16:39:49<1:06:07, 1.59s/it] 92%|█████████▏| 27510/30000 [16:39:50<1:00:35, 1.46s/it] 92%|█████████▏| 27510/30000 [16:39:50<1:00:35, 1.46s/it] 92%|█████████▏| 27511/30000 [16:39:51<57:03, 1.38s/it] 92%|█████████▏| 27512/30000 [16:39:58<2:00:02, 2.89s/it] 92%|█████████▏| 27513/30000 [16:39:59<1:38:27, 2.38s/it] 92%|█████████▏| 27514/30000 [16:40:00<1:23:38, 2.02s/it] 92%|█████████▏| 27515/30000 [16:40:01<1:14:16, 1.79s/it] 92%|█████████▏| 27516/30000 [16:40:03<1:07:55, 1.64s/it] 92%|█████████▏| 27517/30000 [16:40:04<1:03:45, 1.54s/it] 92%|█████████▏| 27518/30000 [16:40:05<58:48, 1.42s/it] 92%|█████████▏| 27519/30000 [16:40:06<55:27, 1.34s/it] 92%|█████████▏| 27520/30000 [16:40:13<2:08:08, 3.10s/it] 92%|█████████▏| 27520/30000 [16:40:13<2:08:08, 3.10s/it] 92%|█████████▏| 27521/30000 [16:40:15<1:44:02, 2.52s/it] 92%|█████████▏| 27522/30000 [16:40:16<1:27:28, 2.12s/it] 92%|█████████▏| 27523/30000 [16:40:17<1:15:33, 1.83s/it] 92%|█████████▏| 27524/30000 [16:40:18<1:07:26, 1.63s/it] 92%|█████████▏| 27525/30000 [16:40:19<1:01:45, 1.50s/it] 92%|█████████▏| 27526/30000 [16:40:20<57:27, 1.39s/it] 92%|█████████▏| 27527/30000 [16:40:22<55:24, 1.34s/it] 92%|█████████▏| 27528/30000 [16:40:31<2:36:28, 3.80s/it] 92%|█████████▏| 27529/30000 [16:40:32<2:04:18, 3.02s/it] 92%|█████████▏| 27530/30000 [16:40:34<1:41:39, 2.47s/it] 92%|█████████▏| 27530/30000 [16:40:34<1:41:39, 2.47s/it] 92%|█████████▏| 27531/30000 [16:40:35<1:25:58, 2.09s/it] 92%|█████████▏| 27532/30000 [16:40:36<1:15:26, 1.83s/it] 92%|█████████▏| 27533/30000 [16:40:37<1:07:34, 1.64s/it] 92%|█████████▏| 27534/30000 [16:40:38<1:02:16, 1.52s/it] 92%|█████████▏| 27535/30000 [16:40:40<58:38, 1.43s/it] 92%|█████████▏| 27536/30000 [16:40:51<2:57:18, 4.32s/it] 92%|█████████▏| 27537/30000 [16:40:52<2:21:45, 3.45s/it] 92%|█████████▏| 27538/30000 [16:40:54<1:58:26, 2.89s/it] 92%|█████████▏| 27539/30000 [16:40:55<1:37:39, 2.38s/it] 92%|█████████▏| 27540/30000 [16:40:56<1:23:25, 2.03s/it] 92%|█████████▏| 27540/30000 [16:40:56<1:23:25, 2.03s/it] 92%|█████████▏| 27541/30000 [16:40:57<1:12:58, 1.78s/it] 92%|█████████▏| 27542/30000 [16:40:58<1:05:17, 1.59s/it] 92%|█████████▏| 27543/30000 [16:41:00<1:00:59, 1.49s/it] 92%|█████████▏| 27544/30000 [16:41:12<3:13:37, 4.73s/it] 92%|█████████▏| 27545/30000 [16:41:13<2:29:43, 3.66s/it] 92%|█████████▏| 27546/30000 [16:41:14<1:59:35, 2.92s/it] 92%|█████████▏| 27547/30000 [16:41:16<1:38:19, 2.41s/it] 92%|█████████▏| 27548/30000 [16:41:17<1:23:24, 2.04s/it] 92%|█████████▏| 27549/30000 [16:41:18<1:13:35, 1.80s/it] 92%|█████████▏| 27550/30000 [16:41:19<1:06:51, 1.64s/it] 92%|█████████▏| 27550/30000 [16:41:19<1:06:51, 1.64s/it] 92%|█████████▏| 27551/30000 [16:41:20<1:01:23, 1.50s/it] 92%|█████████▏| 27552/30000 [16:41:34<3:32:30, 5.21s/it] 92%|█████████▏| 27553/30000 [16:41:36<2:43:58, 4.02s/it] 92%|█████████▏| 27554/30000 [16:41:37<2:09:24, 3.17s/it] 92%|█████████▏| 27555/30000 [16:41:38<1:44:52, 2.57s/it] 92%|█████████▏| 27556/30000 [16:41:39<1:27:57, 2.16s/it] 92%|█████████▏| 27557/30000 [16:41:40<1:16:08, 1.87s/it] 92%|█████████▏| 27558/30000 [16:41:42<1:07:52, 1.67s/it] 92%|█████████▏| 27559/30000 [16:41:43<1:02:54, 1.55s/it] 92%|█████████▏| 27560/30000 [16:41:56<3:22:32, 4.98s/it] 92%|█████████▏| 27560/30000 [16:41:56<3:22:32, 4.98s/it] 92%|█████████▏| 27561/30000 [16:41:57<2:36:28, 3.85s/it] 92%|█████████▏| 27562/30000 [16:41:58<2:03:55, 3.05s/it] 92%|█████████▏| 27563/30000 [16:41:59<1:40:54, 2.48s/it] 92%|█████████▏| 27564/30000 [16:42:01<1:25:10, 2.10s/it] 92%|█████████▏| 27565/30000 [16:42:02<1:14:12, 1.83s/it] 92%|█████████▏| 27566/30000 [16:42:03<1:06:31, 1.64s/it] 92%|█████████▏| 27567/30000 [16:42:04<1:00:47, 1.50s/it] 92%|█████████▏| 27568/30000 [16:42:18<3:26:43, 5.10s/it] 92%|█████████▏| 27569/30000 [16:42:19<2:38:40, 3.92s/it] 92%|█████████▏| 27570/30000 [16:42:20<2:06:08, 3.11s/it] 92%|█████████▏| 27570/30000 [16:42:20<2:06:08, 3.11s/it] 92%|█████████▏| 27571/30000 [16:42:21<1:42:29, 2.53s/it] 92%|█████████▏| 27572/30000 [16:42:22<1:27:17, 2.16s/it] 92%|█████████▏| 27573/30000 [16:42:24<1:15:42, 1.87s/it] 92%|█████████▏| 27574/30000 [16:42:25<1:07:21, 1.67s/it] 92%|█████████▏| 27575/30000 [16:42:26<1:01:34, 1.52s/it] 92%|█████████▏| 27576/30000 [16:42:38<3:07:36, 4.64s/it] 92%|█████████▏| 27577/30000 [16:42:39<2:25:32, 3.60s/it] 92%|█████████▏| 27578/30000 [16:42:40<1:56:26, 2.88s/it] 92%|█████████▏| 27579/30000 [16:42:41<1:35:35, 2.37s/it] 92%|█████████▏| 27580/30000 [16:42:43<1:21:28, 2.02s/it] 92%|█████████▏| 27580/30000 [16:42:43<1:21:28, 2.02s/it] 92%|█████████▏| 27581/30000 [16:42:44<1:11:13, 1.77s/it] 92%|█████████▏| 27582/30000 [16:42:45<1:03:58, 1.59s/it] 92%|█████████▏| 27583/30000 [16:42:46<59:06, 1.47s/it] 92%|█████████▏| 27584/30000 [16:42:54<2:19:22, 3.46s/it] 92%|█████████▏| 27585/30000 [16:42:56<1:51:36, 2.77s/it] 92%|█████████▏| 27586/30000 [16:42:57<1:32:34, 2.30s/it] 92%|█████████▏| 27587/30000 [16:42:58<1:20:02, 1.99s/it] 92%|█████████▏| 27588/30000 [16:42:59<1:10:07, 1.74s/it] 92%|█████████▏| 27589/30000 [16:43:00<1:03:14, 1.57s/it] 92%|█████████▏| 27590/30000 [16:43:02<58:54, 1.47s/it] 92%|█████████▏| 27590/30000 [16:43:02<58:54, 1.47s/it] 92%|█████████▏| 27591/30000 [16:43:03<55:40, 1.39s/it] 92%|█████████▏| 27592/30000 [16:43:16<3:22:22, 5.04s/it] 92%|█████████▏| 27593/30000 [16:43:17<2:35:26, 3.87s/it] 92%|█████████▏| 27594/30000 [16:43:19<2:03:15, 3.07s/it] 92%|█████████▏| 27595/30000 [16:43:20<1:40:07, 2.50s/it] 92%|█████████▏| 27596/30000 [16:43:21<1:24:12, 2.10s/it] 92%|█████████▏| 27597/30000 [16:43:22<1:14:17, 1.86s/it] 92%|█████████▏| 27598/30000 [16:43:24<1:07:57, 1.70s/it] 92%|█████████▏| 27599/30000 [16:43:25<1:01:51, 1.55s/it] 92%|█████████▏| 27600/30000 [16:43:38<3:25:21, 5.13s/it] 92%|█████████▏| 27600/30000 [16:43:38<3:25:21, 5.13s/it] 92%|█████████▏| 27601/30000 [16:43:39<2:37:43, 3.94s/it] 92%|█████████▏| 27602/30000 [16:43:41<2:04:50, 3.12s/it] 92%|█████████▏| 27603/30000 [16:43:42<1:41:33, 2.54s/it] 92%|█████████▏| 27604/30000 [16:43:43<1:25:33, 2.14s/it] 92%|█████████▏| 27605/30000 [16:43:44<1:14:07, 1.86s/it] 92%|█████████▏| 27606/30000 [16:43:45<1:06:10, 1.66s/it] 92%|█████████▏| 27607/30000 [16:43:47<1:01:47, 1.55s/it] 92%|█████████▏| 27608/30000 [16:43:58<3:02:34, 4.58s/it] 92%|█████████▏| 27609/30000 [16:44:00<2:21:46, 3.56s/it] 92%|█████████▏| 27610/30000 [16:44:01<1:53:39, 2.85s/it] 92%|█████████▏| 27610/30000 [16:44:01<1:53:39, 2.85s/it] 92%|█████████▏| 27611/30000 [16:44:02<1:33:16, 2.34s/it] 92%|█████████▏| 27612/30000 [16:44:03<1:19:45, 2.00s/it] 92%|█████████▏| 27613/30000 [16:44:04<1:10:07, 1.76s/it] 92%|█████████▏| 27614/30000 [16:44:06<1:03:23, 1.59s/it] 92%|█████████▏| 27615/30000 [16:44:07<57:58, 1.46s/it] 92%|█████████▏| 27616/30000 [16:44:18<2:51:31, 4.32s/it] 92%|█████████▏| 27617/30000 [16:44:19<2:13:58, 3.37s/it] 92%|█████████▏| 27618/30000 [16:44:20<1:47:44, 2.71s/it] 92%|█████████▏| 27619/30000 [16:44:21<1:29:56, 2.27s/it] 92%|█████████▏| 27620/30000 [16:44:22<1:17:04, 1.94s/it] 92%|█████████▏| 27620/30000 [16:44:22<1:17:04, 1.94s/it] 92%|█████████▏| 27621/30000 [16:44:24<1:08:09, 1.72s/it] 92%|█████████▏| 27622/30000 [16:44:25<1:02:01, 1.57s/it] 92%|█████████▏| 27623/30000 [16:44:26<57:23, 1.45s/it] 92%|█████████▏| 27624/30000 [16:44:39<3:16:41, 4.97s/it] 92%|█████████▏| 27625/30000 [16:44:40<2:31:34, 3.83s/it] 92%|█████████▏| 27626/30000 [16:44:42<2:00:19, 3.04s/it] 92%|█████████▏| 27627/30000 [16:44:43<1:38:27, 2.49s/it] 92%|█████████▏| 27628/30000 [16:44:44<1:22:58, 2.10s/it] 92%|█████████▏| 27629/30000 [16:44:45<1:12:46, 1.84s/it] 92%|█████████▏| 27630/30000 [16:44:46<1:05:46, 1.67s/it] 92%|█████████▏| 27630/30000 [16:44:46<1:05:46, 1.67s/it] 92%|█████████▏| 27631/30000 [16:44:48<1:00:42, 1.54s/it] 92%|█████████▏| 27632/30000 [16:45:01<3:17:38, 5.01s/it] 92%|█████████▏| 27633/30000 [16:45:02<2:32:30, 3.87s/it] 92%|█████████▏| 27634/30000 [16:45:03<2:00:39, 3.06s/it] 92%|█████████▏| 27635/30000 [16:45:04<1:38:40, 2.50s/it] 92%|█████████▏| 27636/30000 [16:45:06<1:22:48, 2.10s/it] 92%|█████████▏| 27637/30000 [16:45:07<1:11:50, 1.82s/it] 92%|█████████▏| 27638/30000 [16:45:08<1:04:14, 1.63s/it] 92%|█████████▏| 27639/30000 [16:45:09<59:05, 1.50s/it] 92%|█████████▏| 27640/30000 [16:45:20<2:51:10, 4.35s/it] 92%|█████████▏| 27640/30000 [16:45:20<2:51:10, 4.35s/it] 92%|█████████▏| 27641/30000 [16:45:21<2:14:13, 3.41s/it] 92%|█████████▏| 27642/30000 [16:45:23<1:47:48, 2.74s/it] 92%|█████████▏| 27643/30000 [16:45:24<1:29:20, 2.27s/it] 92%|█████████▏| 27644/30000 [16:45:25<1:16:45, 1.95s/it] 92%|█████████▏| 27645/30000 [16:45:26<1:07:55, 1.73s/it] 92%|█████████▏| 27646/30000 [16:45:27<1:01:34, 1.57s/it] 92%|█████████▏| 27647/30000 [16:45:28<57:02, 1.45s/it] 92%|█████████▏| 27648/30000 [16:45:43<3:30:27, 5.37s/it] 92%|█████████▏| 27649/30000 [16:45:44<2:41:17, 4.12s/it] 92%|█████████▏| 27650/30000 [16:45:45<2:06:46, 3.24s/it] 92%|█████████▏| 27650/30000 [16:45:45<2:06:46, 3.24s/it] 92%|█████████▏| 27651/30000 [16:45:47<1:42:28, 2.62s/it] 92%|█████████▏| 27652/30000 [16:45:48<1:25:28, 2.18s/it] 92%|█████████▏| 27653/30000 [16:45:49<1:14:10, 1.90s/it] 92%|█████████▏| 27654/30000 [16:45:50<1:07:20, 1.72s/it] 92%|█████████▏| 27655/30000 [16:45:51<1:00:52, 1.56s/it] 92%|█████████▏| 27656/30000 [16:46:06<3:32:01, 5.43s/it] 92%|█████████▏| 27657/30000 [16:46:07<2:42:10, 4.15s/it] 92%|█████████▏| 27658/30000 [16:46:08<2:07:21, 3.26s/it] 92%|█████████▏| 27659/30000 [16:46:09<1:42:56, 2.64s/it] 92%|█████████▏| 27660/30000 [16:46:11<1:25:44, 2.20s/it] 92%|█████████▏| 27660/30000 [16:46:11<1:25:44, 2.20s/it] 92%|█████████▏| 27661/30000 [16:46:12<1:13:59, 1.90s/it] 92%|█████████▏| 27662/30000 [16:46:13<1:05:39, 1.68s/it] 92%|█████████▏| 27663/30000 [16:46:14<59:59, 1.54s/it] 92%|█████████▏| 27664/30000 [16:46:24<2:38:47, 4.08s/it] 92%|█████████▏| 27665/30000 [16:46:25<2:04:56, 3.21s/it] 92%|█████████▏| 27666/30000 [16:46:27<1:41:24, 2.61s/it] 92%|█████████▏| 27667/30000 [16:46:28<1:24:52, 2.18s/it] 92%|█████████▏| 27668/30000 [16:46:29<1:13:02, 1.88s/it] 92%|█████████▏| 27669/30000 [16:46:30<1:05:09, 1.68s/it] 92%|█████████▏| 27670/30000 [16:46:31<59:26, 1.53s/it] 92%|█████████▏| 27670/30000 [16:46:31<59:26, 1.53s/it] 92%|█████████▏| 27671/30000 [16:46:33<55:45, 1.44s/it] 92%|█████████▏| 27672/30000 [16:46:46<3:17:23, 5.09s/it] 92%|█████████▏| 27673/30000 [16:46:47<2:31:46, 3.91s/it] 92%|█████████▏| 27674/30000 [16:46:49<2:00:01, 3.10s/it] 92%|█████████▏| 27675/30000 [16:46:50<1:37:52, 2.53s/it] 92%|█████████▏| 27676/30000 [16:46:51<1:23:03, 2.14s/it] 92%|█████████▏| 27677/30000 [16:46:52<1:12:53, 1.88s/it] 92%|█████████▏| 27678/30000 [16:46:53<1:05:06, 1.68s/it] 92%|█████████▏| 27679/30000 [16:46:55<59:22, 1.53s/it] 92%|█████████▏| 27680/30000 [16:47:06<2:53:00, 4.47s/it] 92%|█████████▏| 27680/30000 [16:47:06<2:53:00, 4.47s/it] 92%|█████████▏| 27681/30000 [16:47:07<2:14:49, 3.49s/it] 92%|█████████▏| 27682/30000 [16:47:08<1:48:48, 2.82s/it] 92%|█████████▏| 27683/30000 [16:47:10<1:30:17, 2.34s/it] 92%|█████████▏| 27684/30000 [16:47:11<1:16:49, 1.99s/it] 92%|█████████▏| 27685/30000 [16:47:12<1:07:27, 1.75s/it] 92%|█████████▏| 27686/30000 [16:47:13<1:01:11, 1.59s/it] 92%|█████████▏| 27687/30000 [16:47:14<56:38, 1.47s/it] 92%|█████████▏| 27688/30000 [16:47:27<3:00:15, 4.68s/it] 92%|█████████▏| 27689/30000 [16:47:28<2:19:47, 3.63s/it] 92%|█████████▏| 27690/30000 [16:47:29<1:51:35, 2.90s/it] 92%|█████████▏| 27690/30000 [16:47:29<1:51:35, 2.90s/it] 92%|█████████▏| 27691/30000 [16:47:30<1:31:49, 2.39s/it] 92%|█████████▏| 27692/30000 [16:47:31<1:18:32, 2.04s/it] 92%|█████████▏| 27693/30000 [16:47:33<1:08:53, 1.79s/it] 92%|█████████▏| 27694/30000 [16:47:34<1:01:56, 1.61s/it] 92%|█████████▏| 27695/30000 [16:47:35<57:41, 1.50s/it] 92%|█████████▏| 27696/30000 [16:47:47<3:02:42, 4.76s/it] 92%|█████████▏| 27697/30000 [16:47:49<2:21:41, 3.69s/it] 92%|█████████▏| 27698/30000 [16:47:50<1:53:00, 2.95s/it] 92%|█████████▏| 27699/30000 [16:47:51<1:33:19, 2.43s/it] 92%|█████████▏| 27700/30000 [16:47:52<1:19:07, 2.06s/it] 92%|█████████▏| 27700/30000 [16:47:52<1:19:07, 2.06s/it] 92%|█████████▏| 27701/30000 [16:47:53<1:09:05, 1.80s/it] 92%|█████████▏| 27702/30000 [16:47:55<1:02:14, 1.63s/it] 92%|█████████▏| 27703/30000 [16:47:56<57:30, 1.50s/it] 92%|█████████▏| 27704/30000 [16:48:07<2:44:51, 4.31s/it] 92%|█████████▏| 27705/30000 [16:48:08<2:09:30, 3.39s/it] 92%|█████████▏| 27706/30000 [16:48:09<1:44:09, 2.72s/it] 92%|█████████▏| 27707/30000 [16:48:10<1:26:37, 2.27s/it] 92%|█████████▏| 27708/30000 [16:48:12<1:14:30, 1.95s/it] 92%|█████████▏| 27709/30000 [16:48:13<1:05:49, 1.72s/it] 92%|█████████▏| 27710/30000 [16:48:14<59:38, 1.56s/it] 92%|█████████▏| 27710/30000 [16:48:14<59:38, 1.56s/it] 92%|█████████▏| 27711/30000 [16:48:15<55:22, 1.45s/it] 92%|█████████▏| 27712/30000 [16:48:28<3:09:27, 4.97s/it] 92%|█████████▏| 27713/30000 [16:48:29<2:26:13, 3.84s/it] 92%|█████████▏| 27714/30000 [16:48:31<1:55:58, 3.04s/it] 92%|█████████▏| 27715/30000 [16:48:32<1:35:10, 2.50s/it] 92%|█████████▏| 27716/30000 [16:48:33<1:20:31, 2.12s/it] 92%|█████████▏| 27717/30000 [16:48:34<1:09:38, 1.83s/it] 92%|█████████▏| 27718/30000 [16:48:35<1:02:23, 1.64s/it] 92%|█████████▏| 27719/30000 [16:48:37<57:42, 1.52s/it] 92%|█████████▏| 27720/30000 [16:48:50<3:07:29, 4.93s/it] 92%|█████████▏| 27720/30000 [16:48:50<3:07:29, 4.93s/it] 92%|█████████▏| 27721/30000 [16:48:51<2:24:44, 3.81s/it] 92%|█████████▏| 27722/30000 [16:48:52<1:55:04, 3.03s/it] 92%|█████████▏| 27723/30000 [16:48:53<1:34:02, 2.48s/it] 92%|█████████▏| 27724/30000 [16:48:54<1:19:18, 2.09s/it] 92%|█████████▏| 27725/30000 [16:48:56<1:09:06, 1.82s/it] 92%|█████████▏| 27726/30000 [16:48:57<1:02:00, 1.64s/it] 92%|█████████▏| 27727/30000 [16:48:58<57:02, 1.51s/it] 92%|█████████▏| 27728/30000 [16:49:10<2:57:23, 4.68s/it] 92%|█████████▏| 27729/30000 [16:49:11<2:17:45, 3.64s/it] 92%|█████████▏| 27730/30000 [16:49:12<1:49:42, 2.90s/it] 92%|█████████▏| 27730/30000 [16:49:12<1:49:42, 2.90s/it] 92%|█████████▏| 27731/30000 [16:49:14<1:30:37, 2.40s/it] 92%|█████████▏| 27732/30000 [16:49:15<1:16:49, 2.03s/it] 92%|█████████▏| 27733/30000 [16:49:16<1:07:09, 1.78s/it] 92%|█████████▏| 27734/30000 [16:49:17<1:00:38, 1.61s/it] 92%|█████████▏| 27735/30000 [16:49:18<55:48, 1.48s/it] 92%|█████████▏| 27736/30000 [16:49:29<2:36:54, 4.16s/it] 92%|█████████▏| 27737/30000 [16:49:30<2:03:09, 3.27s/it] 92%|█████████▏| 27738/30000 [16:49:31<1:39:38, 2.64s/it] 92%|█████████▏| 27739/30000 [16:49:32<1:23:40, 2.22s/it] 92%|█████████▏| 27740/30000 [16:49:34<1:12:16, 1.92s/it] +{'loss': 0.0134, 'grad_norm': 0.08440513908863068, 'learning_rate': 6.454750509527874e-07, 'flow_matching_loss': 0.01342535400763154, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.95} +{'loss': 0.0124, 'grad_norm': 0.08133941888809204, 'learning_rate': 6.406855296131387e-07, 'flow_matching_loss': 0.01235794834792614, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.95} +{'loss': 0.0132, 'grad_norm': 0.06799145042896271, 'learning_rate': 6.359134561945662e-07, 'flow_matching_loss': 0.01319186231121421, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.96} +{'loss': 0.012, 'grad_norm': 0.060073595494031906, 'learning_rate': 6.311588364955883e-07, 'flow_matching_loss': 0.01200198447331786, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.96} +{'loss': 0.0126, 'grad_norm': 0.10197519510984421, 'learning_rate': 6.264216762935187e-07, 'flow_matching_loss': 0.012581582553684712, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.96} +{'loss': 0.013, 'grad_norm': 0.06462494283914566, 'learning_rate': 6.21701981344453e-07, 'flow_matching_loss': 0.012986408732831477, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.97} +{'loss': 0.0141, 'grad_norm': 0.058957260102033615, 'learning_rate': 6.16999757383262e-07, 'flow_matching_loss': 0.014058680273592472, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.97} +{'loss': 0.0115, 'grad_norm': 0.056843101978302, 'learning_rate': 6.123150101235952e-07, 'flow_matching_loss': 0.011486744321882725, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.97} +{'loss': 0.0132, 'grad_norm': 0.09613320976495743, 'learning_rate': 6.076477452578594e-07, 'flow_matching_loss': 0.013243414368480444, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.97} +{'loss': 0.0138, 'grad_norm': 0.11645399034023285, 'learning_rate': 6.029979684572262e-07, 'flow_matching_loss': 0.013841983955353498, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.98} +{'loss': 0.0119, 'grad_norm': 0.06327846646308899, 'learning_rate': 5.983656853716097e-07, 'flow_matching_loss': 0.011886368552222849, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.98} +{'loss': 0.0133, 'grad_norm': 0.07263582199811935, 'learning_rate': 5.937509016296689e-07, 'flow_matching_loss': 0.013328510150313378, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.98} +{'loss': 0.0135, 'grad_norm': 0.07780887931585312, 'learning_rate': 5.89153622838805e-07, 'flow_matching_loss': 0.01349923647940159, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.99} +{'loss': 0.0121, 'grad_norm': 0.045072682201862335, 'learning_rate': 5.845738545851431e-07, 'flow_matching_loss': 0.012073251558467746, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.99} +{'loss': 0.0136, 'grad_norm': 0.10172402113676071, 'learning_rate': 5.800116024335334e-07, 'flow_matching_loss': 0.013600567076355218, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 7.99} +{'loss': 0.0132, 'grad_norm': 0.15186269581317902, 'learning_rate': 5.75466871927543e-07, 'flow_matching_loss': 0.013160080462694169, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.0} +{'loss': 0.0121, 'grad_norm': 0.08182662725448608, 'learning_rate': 5.709396685894463e-07, 'flow_matching_loss': 0.012103529646992683, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.0} +{'loss': 0.011, 'grad_norm': 0.06037335470318794, 'learning_rate': 5.664299979202264e-07, 'flow_matching_loss': 0.010976522881537675, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.0} +{'loss': 0.0138, 'grad_norm': 0.08376156538724899, 'learning_rate': 5.61937865399555e-07, 'flow_matching_loss': 0.01377509068697691, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.0} +{'loss': 0.0131, 'grad_norm': 0.058316901326179504, 'learning_rate': 5.574632764857974e-07, 'flow_matching_loss': 0.0131288249976933, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.01} +{'loss': 0.0133, 'grad_norm': 0.060219451785087585, 'learning_rate': 5.53006236616001e-07, 'flow_matching_loss': 0.013312409073114395, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.01} +{'loss': 0.013, 'grad_norm': 0.07643022388219833, 'learning_rate': 5.485667512058873e-07, 'flow_matching_loss': 0.012958084791898727, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.01} +{'loss': 0.0136, 'grad_norm': 0.06345991790294647, 'learning_rate': 5.441448256498544e-07, 'flow_matching_loss': 0.013588641304522753, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.02} +{'loss': 0.012, 'grad_norm': 0.07162240892648697, 'learning_rate': 5.397404653209559e-07, 'flow_matching_loss': 0.011964343767613173, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.02} +{'loss': 0.014, 'grad_norm': 0.0637969970703125, 'learning_rate': 5.35353675570906e-07, 'flow_matching_loss': 0.013982978090643883, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.02} +{'loss': 0.013, 'grad_norm': 0.07577679306268692, 'learning_rate': 5.309844617300674e-07, 'flow_matching_loss': 0.012996282801032067, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.02} +{'loss': 0.0125, 'grad_norm': 0.06356759369373322, 'learning_rate': 5.266328291074468e-07, 'flow_matching_loss': 0.012534740380942822, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.03} +{'loss': 0.014, 'grad_norm': 0.06428677588701248, 'learning_rate': 5.222987829906861e-07, 'flow_matching_loss': 0.014026763383299112, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.03} +{'loss': 0.0126, 'grad_norm': 0.08340828865766525, 'learning_rate': 5.179823286460644e-07, 'flow_matching_loss': 0.012560094147920609, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.03} +{'loss': 0.0133, 'grad_norm': 0.056733693927526474, 'learning_rate': 5.13683471318478e-07, 'flow_matching_loss': 0.013333381339907645, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.04} +{'loss': 0.0121, 'grad_norm': 0.06672048568725586, 'learning_rate': 5.094022162314449e-07, 'flow_matching_loss': 0.012102200090885163, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.04} +{'loss': 0.0112, 'grad_norm': 0.07147487252950668, 'learning_rate': 5.051385685870924e-07, 'flow_matching_loss': 0.01123367426916957, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.04} +{'loss': 0.0133, 'grad_norm': 0.11078888922929764, 'learning_rate': 5.008925335661546e-07, 'flow_matching_loss': 0.013323393370956182, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.04} +{'loss': 0.0142, 'grad_norm': 0.07782618701457977, 'learning_rate': 4.966641163279662e-07, 'flow_matching_loss': 0.014161836355924606, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.05} +{'loss': 0.0123, 'grad_norm': 0.06411916762590408, 'learning_rate': 4.924533220104521e-07, 'flow_matching_loss': 0.012251785770058633, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.05} +{'loss': 0.0133, 'grad_norm': 0.10096710920333862, 'learning_rate': 4.882601557301231e-07, 'flow_matching_loss': 0.01325656920671463, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.05} +{'loss': 0.0128, 'grad_norm': 0.07032284140586853, 'learning_rate': 4.840846225820717e-07, 'flow_matching_loss': 0.01282265093177557, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.06} +{'loss': 0.0138, 'grad_norm': 0.06450319290161133, 'learning_rate': 4.79926727639966e-07, 'flow_matching_loss': 0.013832969404757024, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.06} +{'loss': 0.0147, 'grad_norm': 0.06648196280002594, 'learning_rate': 4.7578647595603817e-07, 'flow_matching_loss': 0.014668611530214547, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.06} +{'loss': 0.0122, 'grad_norm': 0.06535237282514572, 'learning_rate': 4.716638725610889e-07, 'flow_matching_loss': 0.012159868609160185, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.07} +{'loss': 0.0116, 'grad_norm': 0.08388612419366837, 'learning_rate': 4.6755892246446606e-07, 'flow_matching_loss': 0.011569353844970464, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.07} +{'loss': 0.013, 'grad_norm': 0.06443262100219727, 'learning_rate': 4.634716306540715e-07, 'flow_matching_loss': 0.012968243844807149, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.07} 92%|█████████▏| 27740/30000 [16:49:34<1:12:16, 1.92s/it] 92%|█████████▏| 27741/30000 [16:49:35<1:03:58, 1.70s/it] 92%|█████████▏| 27742/30000 [16:49:36<58:26, 1.55s/it] 92%|█████████▏| 27743/30000 [16:49:37<54:18, 1.44s/it] 92%|█████████▏| 27744/30000 [16:49:44<1:57:52, 3.13s/it] 92%|█████████▏| 27745/30000 [16:49:46<1:35:57, 2.55s/it] 92%|█████████▏| 27746/30000 [16:49:47<1:21:12, 2.16s/it] 92%|█████████▏| 27747/30000 [16:49:48<1:10:13, 1.87s/it] 92%|█████████▏| 27748/30000 [16:49:49<1:02:42, 1.67s/it] 92%|█████████▏| 27749/30000 [16:49:50<57:21, 1.53s/it] 92%|█████████▎| 27750/30000 [16:49:52<53:28, 1.43s/it] 92%|█████████▎| 27750/30000 [16:49:52<53:28, 1.43s/it] 93%|█████████▎| 27751/30000 [16:49:53<50:37, 1.35s/it] 93%|█████████▎| 27752/30000 [16:49:59<1:50:47, 2.96s/it] 93%|█████████▎| 27753/30000 [16:50:01<1:32:05, 2.46s/it] 93%|█████████▎| 27754/30000 [16:50:02<1:18:09, 2.09s/it] 93%|█████████▎| 27755/30000 [16:50:03<1:09:06, 1.85s/it] 93%|█████████▎| 27756/30000 [16:50:04<1:01:46, 1.65s/it] 93%|█████████▎| 27757/30000 [16:50:06<56:36, 1.51s/it] 93%|█████████▎| 27758/30000 [16:50:07<53:01, 1.42s/it] 93%|█████████▎| 27759/30000 [16:50:08<50:49, 1.36s/it] 93%|█████████▎| 27760/30000 [16:50:15<1:56:07, 3.11s/it] 93%|█████████▎| 27760/30000 [16:50:15<1:56:07, 3.11s/it] 93%|█████████▎| 27761/30000 [16:50:16<1:34:39, 2.54s/it] 93%|█████████▎| 27762/30000 [16:50:18<1:20:05, 2.15s/it] 93%|█████████▎| 27763/30000 [16:50:19<1:11:42, 1.92s/it] 93%|█████████▎| 27764/30000 [16:50:20<1:03:32, 1.70s/it] 93%|█████████▎| 27765/30000 [16:50:22<58:06, 1.56s/it] 93%|█████████▎| 27766/30000 [16:50:23<54:08, 1.45s/it] 93%|█████████▎| 27767/30000 [16:50:24<51:04, 1.37s/it] 93%|█████████▎| 27768/30000 [16:50:32<2:02:37, 3.30s/it] 93%|█████████▎| 27769/30000 [16:50:33<1:39:06, 2.67s/it] 93%|█████████▎| 27770/30000 [16:50:34<1:22:33, 2.22s/it] 93%|█████████▎| 27770/30000 [16:50:34<1:22:33, 2.22s/it] 93%|█████████▎| 27771/30000 [16:50:35<1:11:04, 1.91s/it] 93%|█████████▎| 27772/30000 [16:50:36<1:03:11, 1.70s/it] 93%|█████████▎| 27773/30000 [16:50:38<57:38, 1.55s/it] 93%|█████████▎| 27774/30000 [16:50:39<53:34, 1.44s/it] 93%|█████████▎| 27775/30000 [16:50:40<50:55, 1.37s/it] 93%|█████████▎| 27776/30000 [16:50:47<1:53:09, 3.05s/it] 93%|█████████▎| 27777/30000 [16:50:48<1:32:22, 2.49s/it] 93%|█████████▎| 27778/30000 [16:50:49<1:18:01, 2.11s/it] 93%|█████████▎| 27779/30000 [16:50:51<1:07:53, 1.83s/it] 93%|█████████▎| 27780/30000 [16:50:52<1:00:45, 1.64s/it] 93%|█████████▎| 27780/30000 [16:50:52<1:00:45, 1.64s/it] 93%|█████████▎| 27781/30000 [16:50:53<55:51, 1.51s/it] 93%|█████████▎| 27782/30000 [16:50:54<52:14, 1.41s/it] 93%|█████████▎| 27783/30000 [16:50:55<49:56, 1.35s/it] 93%|█████████▎| 27784/30000 [16:51:02<1:48:28, 2.94s/it] 93%|█████████▎| 27785/30000 [16:51:03<1:29:05, 2.41s/it] 93%|█████████▎| 27786/30000 [16:51:04<1:15:32, 2.05s/it] 93%|█████████▎| 27787/30000 [16:51:06<1:06:07, 1.79s/it] 93%|█████████▎| 27788/30000 [16:51:07<59:24, 1.61s/it] 93%|█████████▎| 27789/30000 [16:51:08<54:38, 1.48s/it] 93%|█████████▎| 27790/30000 [16:51:09<51:12, 1.39s/it] 93%|█████████▎| 27790/30000 [16:51:09<51:12, 1.39s/it] 93%|█████████▎| 27791/30000 [16:51:10<48:51, 1.33s/it] 93%|█████████▎| 27792/30000 [16:51:17<1:43:05, 2.80s/it] 93%|█████████▎| 27793/30000 [16:51:18<1:25:20, 2.32s/it] 93%|█████████▎| 27794/30000 [16:51:19<1:13:01, 1.99s/it] 93%|█████████▎| 27795/30000 [16:51:20<1:04:23, 1.75s/it] 93%|█████████▎| 27796/30000 [16:51:21<58:38, 1.60s/it] 93%|█████████▎| 27797/30000 [16:51:23<54:38, 1.49s/it] 93%|█████████▎| 27798/30000 [16:51:24<51:41, 1.41s/it] 93%|█████████▎| 27799/30000 [16:51:25<49:37, 1.35s/it] 93%|█████████▎| 27800/30000 [16:51:31<1:37:12, 2.65s/it] 93%|█████████▎| 27800/30000 [16:51:31<1:37:12, 2.65s/it] 93%|█████████▎| 27801/30000 [16:51:32<1:20:39, 2.20s/it] 93%|█████████▎| 27802/30000 [16:51:33<1:09:18, 1.89s/it] 93%|█████████▎| 27803/30000 [16:51:34<1:01:26, 1.68s/it] 93%|█████████▎| 27804/30000 [16:51:36<56:04, 1.53s/it] 93%|█████████▎| 27805/30000 [16:51:37<52:11, 1.43s/it] 93%|█████████▎| 27806/30000 [16:51:38<49:33, 1.36s/it] 93%|█████████▎| 27807/30000 [16:51:39<47:36, 1.30s/it] 93%|█████████▎| 27808/30000 [16:51:43<1:16:42, 2.10s/it] 93%|█████████▎| 27809/30000 [16:51:44<1:06:26, 1.82s/it] 93%|█████████▎| 27810/30000 [16:51:45<59:29, 1.63s/it] 93%|█████████▎| 27810/30000 [16:51:45<59:29, 1.63s/it] 93%|█████████▎| 27811/30000 [16:51:47<54:21, 1.49s/it] 93%|█████████▎| 27812/30000 [16:51:48<51:10, 1.40s/it] 93%|█████████▎| 27813/30000 [16:51:49<48:45, 1.34s/it] 93%|█████████▎| 27814/30000 [16:51:50<47:00, 1.29s/it] 93%|█████████▎| 27815/30000 [16:51:51<45:42, 1.26s/it] 93%|█████████▎| 27816/30000 [16:51:55<1:12:55, 2.00s/it] 93%|█████████▎| 27817/30000 [16:51:56<1:03:37, 1.75s/it] 93%|█████████▎| 27818/30000 [16:51:57<57:23, 1.58s/it] 93%|█████████▎| 27819/30000 [16:51:59<52:41, 1.45s/it] 93%|█████████▎| 27820/30000 [16:52:00<49:51, 1.37s/it] 93%|█████████▎| 27820/30000 [16:52:00<49:51, 1.37s/it] 93%|█████████▎| 27821/30000 [16:52:01<47:56, 1.32s/it] 93%|█████████▎| 27822/30000 [16:52:02<46:18, 1.28s/it] 93%|█████████▎| 27823/30000 [16:52:03<45:09, 1.24s/it] 93%|█████████▎| 27824/30000 [16:52:08<1:23:19, 2.30s/it] 93%|█████████▎| 27825/30000 [16:52:09<1:10:46, 1.95s/it] 93%|█████████▎| 27826/30000 [16:52:10<1:02:24, 1.72s/it] 93%|█████████▎| 27827/30000 [16:52:12<56:51, 1.57s/it] 93%|█████████▎| 27828/30000 [16:52:13<52:52, 1.46s/it] 93%|█████████▎| 27829/30000 [16:52:14<49:45, 1.38s/it] 93%|█████████▎| 27830/30000 [16:52:15<47:41, 1.32s/it] 93%|█████████▎| 27830/30000 [16:52:15<47:41, 1.32s/it] 93%|█████████▎| 27831/30000 [16:52:16<46:25, 1.28s/it] 93%|█████████▎| 27832/30000 [16:52:21<1:20:46, 2.24s/it] 93%|█████████▎| 27833/30000 [16:52:22<1:09:45, 1.93s/it] 93%|█████████▎| 27834/30000 [16:52:23<1:01:49, 1.71s/it] 93%|█████████▎| 27835/30000 [16:52:24<56:05, 1.55s/it] 93%|█████████▎| 27836/30000 [16:52:26<52:05, 1.44s/it] 93%|█████████▎| 27837/30000 [16:52:27<49:16, 1.37s/it] 93%|█████████▎| 27838/30000 [16:52:28<47:19, 1.31s/it] 93%|█████████▎| 27839/30000 [16:52:29<46:11, 1.28s/it] 93%|█████████▎| 27840/30000 [16:52:34<1:21:32, 2.26s/it] 93%|█████████▎| 27840/30000 [16:52:34<1:21:32, 2.26s/it] 93%|█████████▎| 27841/30000 [16:52:35<1:10:13, 1.95s/it] 93%|█████████▎| 27842/30000 [16:52:36<1:02:21, 1.73s/it] 93%|█████████▎| 27843/30000 [16:52:37<56:38, 1.58s/it] 93%|█████████▎| 27844/30000 [16:52:39<52:34, 1.46s/it] 93%|█████████▎| 27845/30000 [16:52:40<49:48, 1.39s/it] 93%|█████████▎| 27846/30000 [16:52:41<47:59, 1.34s/it] 93%|█████████▎| 27847/30000 [16:52:42<46:24, 1.29s/it] 93%|█████████▎| 27848/30000 [16:52:46<1:15:04, 2.09s/it] 93%|█████████▎| 27849/30000 [16:52:47<1:05:27, 1.83s/it] 93%|█████████▎| 27850/30000 [16:52:49<58:47, 1.64s/it] 93%|█████████▎| 27850/30000 [16:52:49<58:47, 1.64s/it] 93%|█████████▎| 27851/30000 [16:52:50<54:36, 1.52s/it] 93%|█████████▎| 27852/30000 [16:52:51<51:12, 1.43s/it] 93%|█████████▎| 27853/30000 [16:52:52<48:53, 1.37s/it] 93%|█████████▎| 27854/30000 [16:52:53<47:15, 1.32s/it] 93%|█████████▎| 27855/30000 [16:52:55<45:53, 1.28s/it] 93%|█████████▎| 27856/30000 [16:52:59<1:19:12, 2.22s/it] 93%|█████████▎| 27857/30000 [16:53:00<1:08:25, 1.92s/it] 93%|█████████▎| 27858/30000 [16:53:01<1:01:04, 1.71s/it] 93%|█████████▎| 27859/30000 [16:53:03<55:30, 1.56s/it] 93%|█████████▎| 27860/30000 [16:53:04<52:04, 1.46s/it] 93%|█████████▎| 27860/30000 [16:53:04<52:04, 1.46s/it] 93%|█████████▎| 27861/30000 [16:53:05<49:25, 1.39s/it] 93%|█████████▎| 27862/30000 [16:53:06<47:36, 1.34s/it] 93%|█████████▎| 27863/30000 [16:53:08<46:29, 1.31s/it] 93%|█████████▎| 27864/30000 [16:53:12<1:17:24, 2.17s/it] 93%|█████████▎| 27865/30000 [16:53:13<1:07:08, 1.89s/it] 93%|█████████▎| 27866/30000 [16:53:14<1:00:08, 1.69s/it] 93%|█████████▎| 27867/30000 [16:53:15<54:56, 1.55s/it] 93%|█████████▎| 27868/30000 [16:53:17<51:24, 1.45s/it] 93%|█████████▎| 27869/30000 [16:53:18<48:54, 1.38s/it] 93%|█████████▎| 27870/30000 [16:53:19<47:16, 1.33s/it] 93%|█████████▎| 27870/30000 [16:53:19<47:16, 1.33s/it] 93%|█████████▎| 27871/30000 [16:53:20<46:03, 1.30s/it] 93%|█████████▎| 27872/30000 [16:53:25<1:19:46, 2.25s/it] 93%|█████████▎| 27873/30000 [16:53:26<1:08:48, 1.94s/it] 93%|█████████▎| 27874/30000 [16:53:27<1:01:15, 1.73s/it] 93%|█████████▎| 27875/30000 [16:53:28<55:18, 1.56s/it] 93%|█████████▎| 27876/30000 [16:53:30<51:31, 1.46s/it] 93%|█████████▎| 27877/30000 [16:53:31<49:11, 1.39s/it] 93%|█████████▎| 27878/30000 [16:53:32<47:16, 1.34s/it] 93%|█████████▎| 27879/30000 [16:53:33<45:54, 1.30s/it] 93%|█████████▎| 27880/30000 [16:53:38<1:20:46, 2.29s/it] 93%|█████████▎| 27880/30000 [16:53:38<1:20:46, 2.29s/it] 93%|█████████▎| 27881/30000 [16:53:39<1:09:04, 1.96s/it] 93%|█████████▎| 27882/30000 [16:53:40<1:01:17, 1.74s/it] 93%|█████████▎| 27883/30000 [16:53:41<55:28, 1.57s/it] 93%|█████████▎| 27884/30000 [16:53:43<51:42, 1.47s/it] 93%|█████████▎| 27885/30000 [16:53:44<48:55, 1.39s/it] 93%|█████████▎| 27886/30000 [16:53:45<46:56, 1.33s/it] 93%|█████████▎| 27887/30000 [16:53:46<45:31, 1.29s/it] 93%|█████████▎| 27888/30000 [16:53:51<1:18:34, 2.23s/it] 93%|█████████▎| 27889/30000 [16:53:52<1:07:49, 1.93s/it] 93%|█████████▎| 27890/30000 [16:53:53<1:00:11, 1.71s/it] 93%|█████████▎| 27890/30000 [16:53:53<1:00:11, 1.71s/it] 93%|█████████▎| 27891/30000 [16:53:54<54:40, 1.56s/it] 93%|█████████▎| 27892/30000 [16:53:56<50:57, 1.45s/it] 93%|█████████▎| 27893/30000 [16:53:57<48:59, 1.40s/it] 93%|█████████▎| 27894/30000 [16:53:58<47:08, 1.34s/it] 93%|█████████▎| 27895/30000 [16:53:59<45:57, 1.31s/it] 93%|█████████▎| 27896/30000 [16:54:04<1:21:21, 2.32s/it] 93%|█████████▎| 27897/30000 [16:54:05<1:09:22, 1.98s/it] 93%|█████████▎| 27898/30000 [16:54:06<1:01:09, 1.75s/it] 93%|█████████▎| 27899/30000 [16:54:08<55:16, 1.58s/it] 93%|█████████▎| 27900/30000 [16:54:09<51:55, 1.48s/it] 93%|█████████▎| 27900/30000 [16:54:09<51:55, 1.48s/it] 93%|█████████▎| 27901/30000 [16:54:10<49:11, 1.41s/it] 93%|█████████▎| 27902/30000 [16:54:11<47:03, 1.35s/it] 93%|█████████▎| 27903/30000 [16:54:12<46:11, 1.32s/it] 93%|█████████▎| 27904/30000 [16:54:18<1:26:14, 2.47s/it] 93%|█████████▎| 27905/30000 [16:54:19<1:12:54, 2.09s/it] 93%|█████████▎| 27906/30000 [16:54:20<1:03:32, 1.82s/it] 93%|█████████▎| 27907/30000 [16:54:21<56:54, 1.63s/it] 93%|█████████▎| 27908/30000 [16:54:22<52:36, 1.51s/it] 93%|█████████▎| 27909/30000 [16:54:24<49:34, 1.42s/it] 93%|█████████▎| 27910/30000 [16:54:25<47:50, 1.37s/it] 93%|█████████▎| 27910/30000 [16:54:25<47:50, 1.37s/it] 93%|█████████▎| 27911/30000 [16:54:26<46:14, 1.33s/it] 93%|█████████▎| 27912/30000 [16:54:32<1:28:25, 2.54s/it] 93%|█████████▎| 27913/30000 [16:54:33<1:14:05, 2.13s/it] 93%|█████████▎| 27914/30000 [16:54:34<1:04:48, 1.86s/it] 93%|█████████▎| 27915/30000 [16:54:35<57:32, 1.66s/it] 93%|█████████▎| 27916/30000 [16:54:36<52:52, 1.52s/it] 93%|█████████▎| 27917/30000 [16:54:38<49:39, 1.43s/it] 93%|█████████▎| 27918/30000 [16:54:39<47:15, 1.36s/it] 93%|█████████▎| 27919/30000 [16:54:40<45:51, 1.32s/it] 93%|█████████▎| 27920/30000 [16:54:44<1:18:37, 2.27s/it] 93%|█████████▎| 27920/30000 [16:54:44<1:18:37, 2.27s/it] 93%|█████████▎| 27921/30000 [16:54:46<1:07:26, 1.95s/it] 93%|█████████▎| 27922/30000 [16:54:47<59:44, 1.72s/it] 93%|█████████▎| 27923/30000 [16:54:48<54:00, 1.56s/it] 93%|█████████▎| 27924/30000 [16:54:49<50:17, 1.45s/it] 93%|█████████▎| 27925/30000 [16:54:50<47:41, 1.38s/it] 93%|█████████▎| 27926/30000 [16:54:52<46:07, 1.33s/it] 93%|█████████▎| 27927/30000 [16:54:53<44:59, 1.30s/it] 93%|█████████▎| 27928/30000 [16:54:58<1:20:03, 2.32s/it] 93%|█████████▎| 27929/30000 [16:54:59<1:08:24, 1.98s/it] 93%|█████████▎| 27930/30000 [16:55:00<1:00:31, 1.75s/it] 93%|█████████▎| 27930/30000 [16:55:00<1:00:31, 1.75s/it] 93%|█████████▎| 27931/30000 [16:55:01<55:13, 1.60s/it] 93%|█████████▎| 27932/30000 [16:55:02<51:27, 1.49s/it] 93%|█████████▎| 27933/30000 [16:55:04<48:49, 1.42s/it] 93%|█████████▎| 27934/30000 [16:55:05<46:47, 1.36s/it] 93%|█████████▎| 27935/30000 [16:55:06<45:22, 1.32s/it] 93%|█████████▎| 27936/30000 [16:55:11<1:20:43, 2.35s/it] 93%|█████████▎| 27937/30000 [16:55:12<1:08:38, 2.00s/it] 93%|█████████▎| 27938/30000 [16:55:13<1:00:36, 1.76s/it] 93%|█████████▎| 27939/30000 [16:55:14<54:33, 1.59s/it] 93%|█████████▎| 27940/30000 [16:55:16<50:35, 1.47s/it] 93%|█████████▎| 27940/30000 [16:55:16<50:35, 1.47s/it] 93%|█████████▎| 27941/30000 [16:55:17<47:48, 1.39s/it] 93%|█████████▎| 27942/30000 [16:55:18<45:35, 1.33s/it] 93%|█████████▎| 27943/30000 [16:55:19<44:11, 1.29s/it] 93%|█████████▎| 27944/30000 [16:55:24<1:20:27, 2.35s/it] 93%|█████████▎| 27945/30000 [16:55:25<1:08:41, 2.01s/it] 93%|█████████▎| 27946/30000 [16:55:27<1:01:02, 1.78s/it] 93%|█████████▎| 27947/30000 [16:55:28<55:09, 1.61s/it] 93%|█████████▎| 27948/30000 [16:55:29<50:50, 1.49s/it] 93%|█████████▎| 27949/30000 [16:55:30<47:49, 1.40s/it] 93%|█████████▎| 27950/30000 [16:55:31<46:20, 1.36s/it] 93%|█████████▎| 27950/30000 [16:55:31<46:20, 1.36s/it] 93%|█████████▎| 27951/30000 [16:55:33<44:54, 1.31s/it] 93%|█████████▎| 27952/30000 [16:55:37<1:15:51, 2.22s/it] 93%|█████████▎| 27953/30000 [16:55:38<1:05:33, 1.92s/it] 93%|█████████▎| 27954/30000 [16:55:39<58:28, 1.71s/it] 93%|█████████▎| 27955/30000 [16:55:41<52:59, 1.55s/it] 93%|█████████▎| 27956/30000 [16:55:42<49:37, 1.46s/it] 93%|█████████▎| 27957/30000 [16:55:43<47:21, 1.39s/it] 93%|█████████▎| 27958/30000 [16:55:44<45:22, 1.33s/it] 93%|█████████▎| 27959/30000 [16:55:45<44:12, 1.30s/it] 93%|█████████▎| 27960/30000 [16:55:50<1:17:28, 2.28s/it] 93%|█████████▎| 27960/30000 [16:55:50<1:17:28, 2.28s/it] 93%|█████████▎| 27961/30000 [16:55:51<1:06:51, 1.97s/it] 93%|█████████▎| 27962/30000 [16:55:53<59:05, 1.74s/it] 93%|█████████▎| 27963/30000 [16:55:54<53:15, 1.57s/it] 93%|█████████▎| 27964/30000 [16:55:55<49:24, 1.46s/it] 93%|█████████▎| 27965/30000 [16:55:56<47:08, 1.39s/it] 93%|█████████▎| 27966/30000 [16:55:57<45:26, 1.34s/it] 93%|█████████▎| 27967/30000 [16:55:59<44:30, 1.31s/it] 93%|█████████▎| 27968/30000 [16:56:03<1:18:39, 2.32s/it] 93%|█████████▎| 27969/30000 [16:56:04<1:07:07, 1.98s/it] 93%|█████████▎| 27970/30000 [16:56:06<59:05, 1.75s/it] 93%|█████████▎| 27970/30000 [16:56:06<59:05, 1.75s/it] 93%|█████████▎| 27971/30000 [16:56:07<53:25, 1.58s/it] 93%|█████████▎| 27972/30000 [16:56:08<49:30, 1.46s/it] 93%|█████████▎| 27973/30000 [16:56:09<46:50, 1.39s/it] 93%|█████████▎| 27974/30000 [16:56:10<44:54, 1.33s/it] 93%|█████████▎| 27975/30000 [16:56:12<43:31, 1.29s/it] 93%|█████████▎| 27976/30000 [16:56:17<1:20:53, 2.40s/it] 93%|█████████▎| 27977/30000 [16:56:18<1:08:41, 2.04s/it] 93%|█████████▎| 27978/30000 [16:56:19<1:00:15, 1.79s/it] 93%|█████████▎| 27979/30000 [16:56:20<53:54, 1.60s/it] 93%|█████████▎| 27980/30000 [16:56:21<49:58, 1.48s/it] 93%|█████████▎| 27980/30000 [16:56:21<49:58, 1.48s/it] 93%|█████████▎| 27981/30000 [16:56:23<47:07, 1.40s/it] 93%|█████████▎| 27982/30000 [16:56:24<45:21, 1.35s/it] 93%|█████████▎| 27983/30000 [16:56:25<43:43, 1.30s/it] 93%|█████████▎| 27984/30000 [16:56:30<1:17:58, 2.32s/it] 93%|█████████▎| 27985/30000 [16:56:31<1:07:17, 2.00s/it] 93%|█████████▎| 27986/30000 [16:56:32<59:56, 1.79s/it] 93%|█████████▎| 27987/30000 [16:56:33<54:02, 1.61s/it] 93%|█████████▎| 27988/30000 [16:56:35<50:04, 1.49s/it] 93%|█████████▎| 27989/30000 [16:56:36<47:24, 1.41s/it] 93%|█████████▎| 27990/30000 [16:56:37<45:13, 1.35s/it] 93%|█████████▎| 27990/30000 [16:56:37<45:13, 1.35s/it] 93%|█████████▎| 27991/30000 [16:56:38<44:06, 1.32s/it] 93%|█████████▎| 27992/30000 [16:56:43<1:17:20, 2.31s/it] 93%|█████████▎| 27993/30000 [16:56:44<1:06:33, 1.99s/it] 93%|█████████▎| 27994/30000 [16:56:45<58:47, 1.76s/it] 93%|█████████▎| 27995/30000 [16:56:47<52:52, 1.58s/it] 93%|█████████▎| 27996/30000 [16:56:48<49:26, 1.48s/it] 93%|█████████▎| 27997/30000 [16:56:49<46:51, 1.40s/it] 93%|█████████▎| 27998/30000 [16:56:50<44:39, 1.34s/it] 93%|█████████▎| 27999/30000 [16:56:51<43:17, 1.30s/it] 93%|█████████▎| 28000/30000 [16:56:56<1:16:03, 2.28s/it] 93%|█████████▎| 28000/30000 [16:56:56<1:16:03, 2.28s/it] 93%|█████████▎| 28001/30000 [16:56:57<1:05:06, 1.95s/it] 93%|█████████▎| 28002/30000 [16:56:58<57:40, 1.73s/it] 93%|█████████▎| 28003/30000 [16:57:00<52:07, 1.57s/it] 93%|█████████▎| 28004/30000 [16:57:01<48:49, 1.47s/it] 93%|█████████▎| 28005/30000 [16:57:02<46:40, 1.40s/it] 93%|█████████▎| 28006/30000 [16:57:03<44:44, 1.35s/it] 93%|█████████▎| 28007/30000 [16:57:05<43:35, 1.31s/it] 93%|█████████▎| 28008/30000 [16:57:09<1:17:37, 2.34s/it] 93%|█████████▎| 28009/30000 [16:57:10<1:06:22, 2.00s/it] 93%|█████████▎| 28010/30000 [16:57:12<58:40, 1.77s/it] 93%|█████████▎| 28010/30000 [16:57:12<58:40, 1.77s/it] 93%|█████████▎| 28011/30000 [16:57:13<54:16, 1.64s/it] 93%|█████████▎| 28012/30000 [16:57:14<49:54, 1.51s/it] 93%|█████████▎| 28013/30000 [16:57:15<46:42, 1.41s/it] 93%|█████████▎| 28014/30000 [16:57:17<44:45, 1.35s/it] 93%|█████████▎| 28015/30000 [16:57:18<43:28, 1.31s/it] 93%|█████████▎| 28016/30000 [16:57:23<1:16:38, 2.32s/it] 93%|█████████▎| 28017/30000 [16:57:24<1:05:29, 1.98s/it] 93%|█████████▎| 28018/30000 [16:57:25<57:40, 1.75s/it] 93%|█████████▎| 28019/30000 [16:57:26<52:03, 1.58s/it] 93%|█████████▎| 28020/30000 [16:57:27<48:13, 1.46s/it] 93%|█████████▎| 28020/30000 [16:57:27<48:13, 1.46s/it] 93%|█████████▎| 28021/30000 [16:57:29<45:36, 1.38s/it] 93%|█████████▎| 28022/30000 [16:57:30<43:56, 1.33s/it] 93%|█████████▎| 28023/30000 [16:57:31<42:48, 1.30s/it] 93%|█████████▎| 28024/30000 [16:57:36<1:20:49, 2.45s/it] 93%|█████████▎| 28025/30000 [16:57:37<1:08:16, 2.07s/it] 93%|█████████▎| 28026/30000 [16:57:38<59:39, 1.81s/it] 93%|█████████▎| 28027/30000 [16:57:40<53:35, 1.63s/it] 93%|█████████▎| 28028/30000 [16:57:41<49:41, 1.51s/it] 93%|█████████▎| 28029/30000 [16:57:42<46:37, 1.42s/it] 93%|█████████▎| 28030/30000 [16:57:43<44:37, 1.36s/it] 93%|█████████▎| 28030/30000 [16:57:43<44:37, 1.36s/it] 93%|█████████▎| 28031/30000 [16:57:45<43:15, 1.32s/it] 93%|█████████▎| 28032/30000 [16:57:49<1:14:40, 2.28s/it] 93%|█████████▎| 28033/30000 [16:57:50<1:04:17, 1.96s/it] 93%|█████████▎| 28034/30000 [16:57:52<56:39, 1.73s/it] 93%|█████████▎| 28035/30000 [16:57:53<51:12, 1.56s/it] 93%|█████████▎| 28036/30000 [16:57:54<47:39, 1.46s/it] 93%|█████████▎| 28037/30000 [16:57:55<45:24, 1.39s/it] 93%|█████████▎| 28038/30000 [16:57:56<43:35, 1.33s/it] 93%|█████████▎| 28039/30000 [16:57:58<42:18, 1.29s/it] 93%|█████████▎| 28040/30000 [16:58:02<1:14:51, 2.29s/it] 93%|█████████▎| 28040/30000 [16:58:02<1:14:51, 2.29s/it] 93%|█████████▎| 28041/30000 [16:58:03<1:04:05, 1.96s/it] 93%|█████████▎| 28042/30000 [16:58:05<56:53, 1.74s/it] 93%|█████████▎| 28043/30000 [16:58:06<51:17, 1.57s/it] 93%|█████████▎| 28044/30000 [16:58:07<47:29, 1.46s/it] 93%|█████████▎| 28045/30000 [16:58:08<44:54, 1.38s/it] 93%|█████████▎| 28046/30000 [16:58:09<43:13, 1.33s/it] 93%|█████████▎| 28047/30000 [16:58:11<42:06, 1.29s/it] 93%|█████████▎| 28048/30000 [16:58:15<1:11:27, 2.20s/it] 93%|█████████▎| 28049/30000 [16:58:16<1:01:22, 1.89s/it] 94%|█████████▎| 28050/30000 [16:58:17<54:43, 1.68s/it] 94%|█████████▎| 28050/30000 [16:58:17<54:43, 1.68s/it] 94%|█████████▎| 28051/30000 [16:58:18<49:37, 1.53s/it] 94%|█████████▎| 28052/30000 [16:58:20<47:11, 1.45s/it] 94%|█████████▎| 28053/30000 [16:58:21<44:37, 1.38s/it] 94%|█████████▎| 28054/30000 [16:58:22<43:13, 1.33s/it] 94%|█████████▎| 28055/30000 [16:58:23<41:58, 1.29s/it] 94%|█████████▎| 28056/30000 [16:58:28<1:15:06, 2.32s/it] 94%|█████████▎| 28057/30000 [16:58:29<1:04:20, 1.99s/it] 94%|█████████▎| 28058/30000 [16:58:30<56:36, 1.75s/it] 94%|█████████▎| 28059/30000 [16:58:32<51:10, 1.58s/it] 94%|█████████▎| 28060/30000 [16:58:33<47:41, 1.47s/it] 94%|█████████▎| 28060/30000 [16:58:33<47:41, 1.47s/it] 94%|█████████▎| 28061/30000 [16:58:34<44:55, 1.39s/it] 94%|█████████▎| 28062/30000 [16:58:35<43:09, 1.34s/it] 94%|█████████▎| 28063/30000 [16:58:36<41:50, 1.30s/it] 94%|█████████▎| 28064/30000 [16:58:41<1:13:43, 2.29s/it] 94%|█████████▎| 28065/30000 [16:58:42<1:03:34, 1.97s/it] 94%|█████████▎| 28066/30000 [16:58:43<56:16, 1.75s/it] 94%|█████████▎| 28067/30000 [16:58:45<50:35, 1.57s/it] 94%|█████████▎| 28068/30000 [16:58:46<46:56, 1.46s/it] 94%|█████████▎| 28069/30000 [16:58:47<44:23, 1.38s/it] 94%|█████████▎| 28070/30000 [16:58:48<42:39, 1.33s/it] 94%|█████████▎| 28070/30000 [16:58:48<42:39, 1.33s/it] 94%|█████████▎| 28071/30000 [16:58:49<41:34, 1.29s/it] 94%|█████████▎| 28072/30000 [16:58:54<1:11:13, 2.22s/it] 94%|█████████▎| 28073/30000 [16:58:55<1:01:23, 1.91s/it] 94%|█████████▎| 28074/30000 [16:58:57<57:40, 1.80s/it] 94%|█████████▎| 28075/30000 [16:58:58<52:08, 1.63s/it] 94%|█████████▎| 28076/30000 [16:58:59<48:20, 1.51s/it] 94%|█████████▎| 28077/30000 [16:59:00<45:48, 1.43s/it] 94%|█████████▎| 28078/30000 [16:59:01<43:35, 1.36s/it] 94%|█████████▎| 28079/30000 [16:59:03<42:01, 1.31s/it] 94%|█████████▎| 28080/30000 [16:59:07<1:10:00, 2.19s/it] 94%|█████████▎| 28080/30000 [16:59:07<1:10:00, 2.19s/it] 94%|█████████▎| 28081/30000 [16:59:08<1:00:30, 1.89s/it] 94%|█████████▎| 28082/30000 [16:59:10<1:04:24, 2.01s/it] 94%|█████████▎| 28083/30000 [16:59:12<56:19, 1.76s/it] 94%|█████████▎| 28084/30000 [16:59:13<50:46, 1.59s/it] 94%|█████████▎| 28085/30000 [16:59:14<47:33, 1.49s/it] 94%|█████████▎| 28086/30000 [16:59:15<44:46, 1.40s/it] 94%|█████████▎| 28087/30000 [16:59:16<42:54, 1.35s/it] 94%|█████████▎| 28088/30000 [16:59:20<1:04:44, 2.03s/it] 94%|█████████▎| 28089/30000 [16:59:21<56:42, 1.78s/it] 94%|█████████▎| 28090/30000 [16:59:24<1:04:19, 2.02s/it] 94%|█████████▎| 28090/30000 [16:59:24<1:04:19, 2.02s/it] 94%|█████████▎| 28091/30000 [16:59:25<57:04, 1.79s/it] 94%|█████████▎| 28092/30000 [16:59:26<51:26, 1.62s/it] 94%|█████████▎| 28093/30000 [16:59:28<47:47, 1.50s/it] 94%|█████████▎| 28094/30000 [16:59:29<44:58, 1.42s/it] 94%|█████████▎| 28095/30000 [16:59:30<42:50, 1.35s/it] 94%|█████████▎| 28096/30000 [16:59:33<57:02, 1.80s/it] 94%|█████████▎| 28097/30000 [16:59:34<51:23, 1.62s/it] 94%|█████████▎| 28098/30000 [16:59:38<1:11:03, 2.24s/it] 94%|█████████▎| 28099/30000 [16:59:39<1:00:49, 1.92s/it] 94%|█████████▎| 28100/30000 [16:59:40<53:58, 1.70s/it] 94%|█████████▎| 28100/30000 [16:59:40<53:58, 1.70s/it] 94%|█████████▎| 28101/30000 [16:59:41<49:13, 1.56s/it] 94%|█████████▎| 28102/30000 [16:59:42<45:51, 1.45s/it] 94%|█████████▎| 28103/30000 [16:59:44<43:29, 1.38s/it] 94%|█████████▎| 28104/30000 [16:59:46<50:06, 1.59s/it] 94%|█████████▎| 28105/30000 [16:59:47<46:22, 1.47s/it] 94%|█████████▎| 28106/30000 [16:59:51<1:10:45, 2.24s/it] 94%|█████████▎| 28107/30000 [16:59:52<1:00:31, 1.92s/it] 94%|█████████▎| 28108/30000 [16:59:53<53:32, 1.70s/it] 94%|█████████▎| 28109/30000 [16:59:55<48:53, 1.55s/it] 94%|█████████▎| 28110/30000 [16:59:56<45:34, 1.45s/it] 94%|█████████▎| 28110/30000 [16:59:56<45:34, 1.45s/it] 94%|█████████▎| 28111/30000 [16:59:57<42:58, 1.36s/it] 94%|█████████▎| 28112/30000 [16:59:59<46:14, 1.47s/it] 94%|█████████▎| 28113/30000 [17:00:00<43:43, 1.39s/it] 94%|█████████▎| 28114/30000 [17:00:05<1:15:01, 2.39s/it] 94%|█████████▎| 28115/30000 [17:00:06<1:03:50, 2.03s/it] 94%|█████████▎| 28116/30000 [17:00:07<56:04, 1.79s/it] 94%|█████████▎| 28117/30000 [17:00:08<50:46, 1.62s/it] 94%|█████████▎| 28118/30000 [17:00:09<46:57, 1.50s/it] 94%|█████████▎| 28119/30000 [17:00:11<44:26, 1.42s/it] 94%|█████████▎| 28120/30000 [17:00:12<42:54, 1.37s/it] 94%|█████████▎| 28120/30000 [17:00:12<42:54, 1.37s/it] 94%|█████████▎| 28121/30000 [17:00:13<42:04, 1.34s/it] 94%|█████████▎| 28122/30000 [17:00:18<1:12:05, 2.30s/it] 94%|█████████▎| 28123/30000 [17:00:19<1:01:28, 1.97s/it] 94%|█████████▎| 28124/30000 [17:00:20<54:12, 1.73s/it] 94%|█████████▍| 28125/30000 [17:00:21<49:18, 1.58s/it] 94%|█████████▍| 28126/30000 [17:00:23<45:55, 1.47s/it] 94%|█████████▍| 28127/30000 [17:00:24<43:29, 1.39s/it] 94%|█████████▍| 28128/30000 [17:00:25<43:03, 1.38s/it] 94%|█████████▍| 28129/30000 [17:00:26<41:29, 1.33s/it] 94%|█████████▍| 28130/30000 [17:00:31<1:15:22, 2.42s/it] 94%|█████████▍| 28130/30000 [17:00:31<1:15:22, 2.42s/it] 94%|█████████▍| 28131/30000 [17:00:32<1:03:43, 2.05s/it] 94%|█████████▍| 28132/30000 [17:00:34<56:13, 1.81s/it] 94%|█████████▍| 28133/30000 [17:00:35<50:39, 1.63s/it] 94%|█████████▍| 28134/30000 [17:00:36<46:38, 1.50s/it] 94%|█████████▍| 28135/30000 [17:00:37<43:45, 1.41s/it] 94%|█████████▍| 28136/30000 [17:00:39<42:03, 1.35s/it] 94%|█████████▍| 28137/30000 [17:00:40<40:42, 1.31s/it] 94%|█████████▍| 28138/30000 [17:00:44<1:07:42, 2.18s/it] 94%|█████████▍| 28139/30000 [17:00:45<58:20, 1.88s/it] 94%|█████████▍| 28140/30000 [17:00:46<51:53, 1.67s/it] 94%|█████████▍| 28140/30000 [17:00:46<51:53, 1.67s/it] 94%|█████████▍| 28141/30000 [17:00:48<47:29, 1.53s/it] 94%|█████████▍| 28142/30000 [17:00:49<44:37, 1.44s/it] 94%|█████████▍| 28143/30000 [17:00:50<42:28, 1.37s/it] 94%|█████████▍| 28144/30000 [17:00:52<47:09, 1.52s/it] 94%|█████████▍| 28145/30000 [17:00:53<44:06, 1.43s/it] 94%|█████████▍| 28146/30000 [17:00:57<1:09:45, 2.26s/it] 94%|█████████▍| 28147/30000 [17:00:58<59:37, 1.93s/it] 94%|█████████▍| 28148/30000 [17:01:00<52:45, 1.71s/it] 94%|█████████▍| 28149/30000 [17:01:01<48:04, 1.56s/it] 94%|█████████▍| 28150/30000 [17:01:02<44:58, 1.46s/it] +{'loss': 0.0131, 'grad_norm': 0.07338638603687286, 'learning_rate': 4.5940200209635244e-07, 'flow_matching_loss': 0.013094702549278737, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.07} +{'loss': 0.0138, 'grad_norm': 0.077033132314682, 'learning_rate': 4.5535004173628657e-07, 'flow_matching_loss': 0.013768563978374004, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.08} +{'loss': 0.013, 'grad_norm': 0.058653004467487335, 'learning_rate': 4.5131575449739225e-07, 'flow_matching_loss': 0.012963672261685133, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.08} +{'loss': 0.0125, 'grad_norm': 0.07911240309476852, 'learning_rate': 4.472991452817099e-07, 'flow_matching_loss': 0.012479779217392207, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.08} +{'loss': 0.0134, 'grad_norm': 0.0876830667257309, 'learning_rate': 4.433002189697938e-07, 'flow_matching_loss': 0.013357452303171157, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.09} +{'loss': 0.0129, 'grad_norm': 0.08071833103895187, 'learning_rate': 4.393189804207187e-07, 'flow_matching_loss': 0.01290185023099184, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.09} +{'loss': 0.014, 'grad_norm': 0.071235790848732, 'learning_rate': 4.353554344720651e-07, 'flow_matching_loss': 0.013959024753421544, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.09} +{'loss': 0.0144, 'grad_norm': 0.0810093954205513, 'learning_rate': 4.314095859399153e-07, 'flow_matching_loss': 0.014391971472650766, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.09} +{'loss': 0.0127, 'grad_norm': 0.08817046880722046, 'learning_rate': 4.274814396188459e-07, 'flow_matching_loss': 0.012666654586791993, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.1} +{'loss': 0.0131, 'grad_norm': 0.06504296511411667, 'learning_rate': 4.235710002819271e-07, 'flow_matching_loss': 0.013068395014852286, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.1} +{'loss': 0.013, 'grad_norm': 0.06966384500265121, 'learning_rate': 4.196782726807097e-07, 'flow_matching_loss': 0.012955585308372975, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.1} +{'loss': 0.0129, 'grad_norm': 0.08155141025781631, 'learning_rate': 4.158032615452251e-07, 'flow_matching_loss': 0.012858011573553086, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.11} +{'loss': 0.014, 'grad_norm': 0.0736677423119545, 'learning_rate': 4.1194597158397853e-07, 'flow_matching_loss': 0.013975854031741619, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.11} +{'loss': 0.0128, 'grad_norm': 0.05996353551745415, 'learning_rate': 4.081064074839408e-07, 'flow_matching_loss': 0.012838307488709687, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.11} +{'loss': 0.0128, 'grad_norm': 0.08311943709850311, 'learning_rate': 4.042845739105433e-07, 'flow_matching_loss': 0.012813475821167231, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.11} +{'loss': 0.0128, 'grad_norm': 0.10000535100698471, 'learning_rate': 4.004804755076746e-07, 'flow_matching_loss': 0.01282686684280634, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.12} +{'loss': 0.012, 'grad_norm': 0.060644447803497314, 'learning_rate': 3.9669411689767063e-07, 'flow_matching_loss': 0.01199403554201126, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.12} +{'loss': 0.0126, 'grad_norm': 0.0671173706650734, 'learning_rate': 3.929255026813194e-07, 'flow_matching_loss': 0.012610387476161122, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.12} +{'loss': 0.0124, 'grad_norm': 0.06751329451799393, 'learning_rate': 3.8917463743783797e-07, 'flow_matching_loss': 0.012355474382638931, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.13} +{'loss': 0.0134, 'grad_norm': 0.06770852208137512, 'learning_rate': 3.854415257248839e-07, 'flow_matching_loss': 0.013383136224001646, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.13} +{'loss': 0.0116, 'grad_norm': 0.06095369905233383, 'learning_rate': 3.8172617207853864e-07, 'flow_matching_loss': 0.01156202405691147, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.13} +{'loss': 0.0128, 'grad_norm': 0.08495932817459106, 'learning_rate': 3.7802858101330773e-07, 'flow_matching_loss': 0.012757162936031819, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.14} +{'loss': 0.0131, 'grad_norm': 0.07662546634674072, 'learning_rate': 3.743487570221121e-07, 'flow_matching_loss': 0.013105278555303811, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.14} +{'loss': 0.013, 'grad_norm': 0.0981823280453682, 'learning_rate': 3.7068670457628517e-07, 'flow_matching_loss': 0.013023317977786063, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.14} +{'loss': 0.0131, 'grad_norm': 0.06958580017089844, 'learning_rate': 3.6704242812556576e-07, 'flow_matching_loss': 0.013122931867837907, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.14} +{'loss': 0.0132, 'grad_norm': 0.09407681971788406, 'learning_rate': 3.634159320980951e-07, 'flow_matching_loss': 0.013188391644507646, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.15} +{'loss': 0.0129, 'grad_norm': 0.06586829572916031, 'learning_rate': 3.5980722090040663e-07, 'flow_matching_loss': 0.012895617354661226, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.15} +{'loss': 0.0123, 'grad_norm': 0.07035885751247406, 'learning_rate': 3.5621629891742447e-07, 'flow_matching_loss': 0.012326311226934195, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.15} +{'loss': 0.0146, 'grad_norm': 0.06775034219026566, 'learning_rate': 3.526431705124583e-07, 'flow_matching_loss': 0.014644635748118163, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.16} +{'loss': 0.015, 'grad_norm': 0.08339616656303406, 'learning_rate': 3.4908784002719517e-07, 'flow_matching_loss': 0.014984537940472364, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.16} +{'loss': 0.0133, 'grad_norm': 0.09428761899471283, 'learning_rate': 3.4555031178169773e-07, 'flow_matching_loss': 0.013268484454602002, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.16} +{'loss': 0.0139, 'grad_norm': 0.0954204574227333, 'learning_rate': 3.420305900743942e-07, 'flow_matching_loss': 0.013878103997558355, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.16} +{'loss': 0.0132, 'grad_norm': 0.06296652555465698, 'learning_rate': 3.385286791820802e-07, 'flow_matching_loss': 0.013244531862437726, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.17} +{'loss': 0.0139, 'grad_norm': 0.084579236805439, 'learning_rate': 3.350445833599086e-07, 'flow_matching_loss': 0.013858631812036037, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.17} +{'loss': 0.0114, 'grad_norm': 0.06510187685489655, 'learning_rate': 3.3157830684138303e-07, 'flow_matching_loss': 0.011442597769200803, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.17} +{'loss': 0.0133, 'grad_norm': 0.06868989020586014, 'learning_rate': 3.281298538383559e-07, 'flow_matching_loss': 0.013250317052006721, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.18} +{'loss': 0.0132, 'grad_norm': 0.06951950490474701, 'learning_rate': 3.2469922854102387e-07, 'flow_matching_loss': 0.013215628173202276, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.18} +{'loss': 0.0145, 'grad_norm': 0.07312360405921936, 'learning_rate': 3.21286435117919e-07, 'flow_matching_loss': 0.014493253082036972, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.18} +{'loss': 0.0125, 'grad_norm': 0.07457323372364044, 'learning_rate': 3.1789147771590754e-07, 'flow_matching_loss': 0.012522770930081606, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.18} +{'loss': 0.0136, 'grad_norm': 0.04700348153710365, 'learning_rate': 3.14514360460183e-07, 'flow_matching_loss': 0.01359171224758029, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.19} +{'loss': 0.0125, 'grad_norm': 0.06821933388710022, 'learning_rate': 3.1115508745426293e-07, 'flow_matching_loss': 0.012532502226531506, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.19} + 94%|█████████▍| 28150/30000 [17:01:02<44:58, 1.46s/it] 94%|█████████▍| 28151/30000 [17:01:03<42:42, 1.39s/it] 94%|█████████▍| 28152/30000 [17:01:05<47:42, 1.55s/it] 94%|█████████▍| 28153/30000 [17:01:06<44:25, 1.44s/it] 94%|█████████▍| 28154/30000 [17:01:11<1:12:13, 2.35s/it] 94%|█████████▍| 28155/30000 [17:01:12<1:01:45, 2.01s/it] 94%|█████████▍| 28156/30000 [17:01:13<54:40, 1.78s/it] 94%|█████████▍| 28157/30000 [17:01:15<49:38, 1.62s/it] 94%|█████████▍| 28158/30000 [17:01:16<46:05, 1.50s/it] 94%|█████████▍| 28159/30000 [17:01:17<43:26, 1.42s/it] 94%|█████████▍| 28160/30000 [17:01:18<42:30, 1.39s/it] 94%|█████████▍| 28160/30000 [17:01:18<42:30, 1.39s/it] 94%|█████████▍| 28161/30000 [17:01:19<40:37, 1.33s/it] 94%|█████████▍| 28162/30000 [17:01:24<1:11:20, 2.33s/it] 94%|█████████▍| 28163/30000 [17:01:25<1:00:54, 1.99s/it] 94%|█████████▍| 28164/30000 [17:01:27<53:34, 1.75s/it] 94%|█████████▍| 28165/30000 [17:01:28<48:33, 1.59s/it] 94%|█████████▍| 28166/30000 [17:01:29<45:36, 1.49s/it] 94%|█████████▍| 28167/30000 [17:01:30<43:11, 1.41s/it] 94%|█████████▍| 28168/30000 [17:01:31<41:28, 1.36s/it] 94%|█████████▍| 28169/30000 [17:01:33<40:00, 1.31s/it] 94%|█████████▍| 28170/30000 [17:01:38<1:15:51, 2.49s/it] 94%|█████████▍| 28170/30000 [17:01:38<1:15:51, 2.49s/it] 94%|█████████▍| 28171/30000 [17:01:39<1:04:53, 2.13s/it] 94%|█████████▍| 28172/30000 [17:01:40<56:24, 1.85s/it] 94%|█████████▍| 28173/30000 [17:01:42<50:29, 1.66s/it] 94%|█████████▍| 28174/30000 [17:01:43<46:47, 1.54s/it] 94%|█████████▍| 28175/30000 [17:01:44<43:40, 1.44s/it] 94%|█████████▍| 28176/30000 [17:01:45<42:25, 1.40s/it] 94%|█████████▍| 28177/30000 [17:01:47<40:29, 1.33s/it] 94%|█████████▍| 28178/30000 [17:01:52<1:13:32, 2.42s/it] 94%|█████████▍| 28179/30000 [17:01:53<1:02:06, 2.05s/it] 94%|█████████▍| 28180/30000 [17:01:54<54:15, 1.79s/it] 94%|█████████▍| 28180/30000 [17:01:54<54:15, 1.79s/it] 94%|█████████▍| 28181/30000 [17:01:55<49:10, 1.62s/it] 94%|█████████▍| 28182/30000 [17:01:56<45:32, 1.50s/it] 94%|█████████▍| 28183/30000 [17:01:58<42:52, 1.42s/it] 94%|█████████▍| 28184/30000 [17:01:59<41:27, 1.37s/it] 94%|█████████▍| 28185/30000 [17:02:00<39:57, 1.32s/it] 94%|█████████▍| 28186/30000 [17:02:05<1:15:56, 2.51s/it] 94%|█████████▍| 28187/30000 [17:02:06<1:03:53, 2.11s/it] 94%|█████████▍| 28188/30000 [17:02:08<55:18, 1.83s/it] 94%|█████████▍| 28189/30000 [17:02:09<49:42, 1.65s/it] 94%|█████████▍| 28190/30000 [17:02:10<45:32, 1.51s/it] 94%|█████████▍| 28190/30000 [17:02:10<45:32, 1.51s/it] 94%|█████████▍| 28191/30000 [17:02:11<43:05, 1.43s/it] 94%|█████████▍| 28192/30000 [17:02:13<41:19, 1.37s/it] 94%|█████████▍| 28193/30000 [17:02:14<39:44, 1.32s/it] 94%|█████████▍| 28194/30000 [17:02:19<1:15:12, 2.50s/it] 94%|█████████▍| 28195/30000 [17:02:20<1:03:13, 2.10s/it] 94%|█████████▍| 28196/30000 [17:02:21<55:05, 1.83s/it] 94%|█████████▍| 28197/30000 [17:02:23<49:24, 1.64s/it] 94%|█████████▍| 28198/30000 [17:02:24<45:23, 1.51s/it] 94%|█████████▍| 28199/30000 [17:02:25<42:53, 1.43s/it] 94%|█████████▍| 28200/30000 [17:02:26<41:04, 1.37s/it] 94%|█████████▍| 28200/30000 [17:02:26<41:04, 1.37s/it] 94%|█████████▍| 28201/30000 [17:02:27<39:49, 1.33s/it] 94%|█████████▍| 28202/30000 [17:02:33<1:17:33, 2.59s/it] 94%|█████████▍| 28203/30000 [17:02:34<1:05:08, 2.17s/it] 94%|█████████▍| 28204/30000 [17:02:35<56:37, 1.89s/it] 94%|█████████▍| 28205/30000 [17:02:37<50:53, 1.70s/it] 94%|█████████▍| 28206/30000 [17:02:38<46:40, 1.56s/it] 94%|█████████▍| 28207/30000 [17:02:39<43:56, 1.47s/it] 94%|█████████▍| 28208/30000 [17:02:40<42:06, 1.41s/it] 94%|█████████▍| 28209/30000 [17:02:42<40:09, 1.35s/it] 94%|█████████▍| 28210/30000 [17:02:47<1:13:07, 2.45s/it] 94%|█████████▍| 28210/30000 [17:02:47<1:13:07, 2.45s/it] 94%|█████████▍| 28211/30000 [17:02:48<1:01:54, 2.08s/it] 94%|█████████▍| 28212/30000 [17:02:49<54:00, 1.81s/it] 94%|█████████▍| 28213/30000 [17:02:50<48:38, 1.63s/it] 94%|█████████▍| 28214/30000 [17:02:51<44:37, 1.50s/it] 94%|█████████▍| 28215/30000 [17:02:53<42:06, 1.42s/it] 94%|█████████▍| 28216/30000 [17:02:54<40:16, 1.35s/it] 94%|█████████▍| 28217/30000 [17:02:55<39:51, 1.34s/it] 94%|█████████▍| 28218/30000 [17:03:00<1:13:26, 2.47s/it] 94%|█████████▍| 28219/30000 [17:03:02<1:01:49, 2.08s/it] 94%|█████████▍| 28220/30000 [17:03:03<53:49, 1.81s/it] 94%|█████████▍| 28220/30000 [17:03:03<53:49, 1.81s/it] 94%|█████████▍| 28221/30000 [17:03:04<48:26, 1.63s/it] 94%|█████████▍| 28222/30000 [17:03:05<44:35, 1.50s/it] 94%|█████████▍| 28223/30000 [17:03:06<42:03, 1.42s/it] 94%|█████████▍| 28224/30000 [17:03:08<40:03, 1.35s/it] 94%|█████████▍| 28225/30000 [17:03:09<38:48, 1.31s/it] 94%|█████████▍| 28226/30000 [17:03:14<1:12:52, 2.46s/it] 94%|█████████▍| 28227/30000 [17:03:15<1:01:36, 2.08s/it] 94%|█████████▍| 28228/30000 [17:03:16<53:55, 1.83s/it] 94%|█████████▍| 28229/30000 [17:03:18<48:39, 1.65s/it] 94%|█████████▍| 28230/30000 [17:03:19<44:58, 1.52s/it] 94%|█████████▍| 28230/30000 [17:03:19<44:58, 1.52s/it] 94%|█████████▍| 28231/30000 [17:03:20<41:55, 1.42s/it] 94%|█████████▍| 28232/30000 [17:03:21<40:20, 1.37s/it] 94%|█████████▍| 28233/30000 [17:03:22<38:38, 1.31s/it] 94%|█████████▍| 28234/30000 [17:03:28<1:12:43, 2.47s/it] 94%|█████████▍| 28235/30000 [17:03:29<1:01:22, 2.09s/it] 94%|█████████▍| 28236/30000 [17:03:30<53:29, 1.82s/it] 94%|█████████▍| 28237/30000 [17:03:31<48:11, 1.64s/it] 94%|█████████▍| 28238/30000 [17:03:32<44:12, 1.51s/it] 94%|█████████▍| 28239/30000 [17:03:34<41:38, 1.42s/it] 94%|█████████▍| 28240/30000 [17:03:35<39:33, 1.35s/it] 94%|█████████▍| 28240/30000 [17:03:35<39:33, 1.35s/it] 94%|█████████▍| 28241/30000 [17:03:36<38:05, 1.30s/it] 94%|█████████▍| 28242/30000 [17:03:41<1:06:55, 2.28s/it] 94%|█████████▍| 28243/30000 [17:03:42<57:06, 1.95s/it] 94%|█████████▍| 28244/30000 [17:03:43<50:15, 1.72s/it] 94%|█████████▍| 28245/30000 [17:03:44<45:48, 1.57s/it] 94%|█████████▍| 28246/30000 [17:03:45<42:22, 1.45s/it] 94%|█████████▍| 28247/30000 [17:03:47<40:29, 1.39s/it] 94%|█████████▍| 28248/30000 [17:03:48<38:48, 1.33s/it] 94%|█████████▍| 28249/30000 [17:03:49<37:38, 1.29s/it] 94%|█████████▍| 28250/30000 [17:03:54<1:09:53, 2.40s/it] 94%|█████████▍| 28250/30000 [17:03:54<1:09:53, 2.40s/it] 94%|█████████▍| 28251/30000 [17:03:55<59:12, 2.03s/it] 94%|█████████▍| 28252/30000 [17:03:56<51:50, 1.78s/it] 94%|█████████▍| 28253/30000 [17:03:57<46:40, 1.60s/it] 94%|█████████▍| 28254/30000 [17:03:59<43:02, 1.48s/it] 94%|█████████▍| 28255/30000 [17:04:00<41:13, 1.42s/it] 94%|█████████▍| 28256/30000 [17:04:01<39:19, 1.35s/it] 94%|█████████▍| 28257/30000 [17:04:02<37:58, 1.31s/it] 94%|█████████▍| 28258/30000 [17:04:07<1:09:35, 2.40s/it] 94%|█████████▍| 28259/30000 [17:04:08<59:01, 2.03s/it] 94%|█████████▍| 28260/30000 [17:04:10<51:40, 1.78s/it] 94%|█████████▍| 28260/30000 [17:04:10<51:40, 1.78s/it] 94%|█████████▍| 28261/30000 [17:04:11<46:25, 1.60s/it] 94%|█████████▍| 28262/30000 [17:04:12<42:54, 1.48s/it] 94%|█████████▍| 28263/30000 [17:04:13<40:36, 1.40s/it] 94%|█████████▍| 28264/30000 [17:04:15<39:21, 1.36s/it] 94%|█████████▍| 28265/30000 [17:04:16<38:00, 1.31s/it] 94%|█████████▍| 28266/30000 [17:04:20<1:06:23, 2.30s/it] 94%|█████████▍| 28267/30000 [17:04:21<56:37, 1.96s/it] 94%|█████████▍| 28268/30000 [17:04:23<50:04, 1.73s/it] 94%|█████████▍| 28269/30000 [17:04:24<45:21, 1.57s/it] 94%|█████████▍| 28270/30000 [17:04:25<42:11, 1.46s/it] 94%|█████████▍| 28270/30000 [17:04:25<42:11, 1.46s/it] 94%|█████████▍| 28271/30000 [17:04:26<39:56, 1.39s/it] 94%|█████████▍| 28272/30000 [17:04:27<38:09, 1.32s/it] 94%|█████████▍| 28273/30000 [17:04:29<37:06, 1.29s/it] 94%|█████████▍| 28274/30000 [17:04:34<1:11:00, 2.47s/it] 94%|█████████▍| 28275/30000 [17:04:35<59:44, 2.08s/it] 94%|█████████▍| 28276/30000 [17:04:36<52:13, 1.82s/it] 94%|█████████▍| 28277/30000 [17:04:37<46:48, 1.63s/it] 94%|█████████▍| 28278/30000 [17:04:39<43:06, 1.50s/it] 94%|█████████▍| 28279/30000 [17:04:40<40:34, 1.41s/it] 94%|█████████▍| 28280/30000 [17:04:41<38:30, 1.34s/it] 94%|█████████▍| 28280/30000 [17:04:41<38:30, 1.34s/it] 94%|█████████▍| 28281/30000 [17:04:42<37:06, 1.30s/it] 94%|█████████▍| 28282/30000 [17:04:47<1:10:47, 2.47s/it] 94%|█████████▍| 28283/30000 [17:04:49<59:42, 2.09s/it] 94%|█████████▍| 28284/30000 [17:04:50<51:47, 1.81s/it] 94%|█████████▍| 28285/30000 [17:04:51<46:24, 1.62s/it] 94%|█████████▍| 28286/30000 [17:04:52<43:03, 1.51s/it] 94%|█████████▍| 28287/30000 [17:04:53<40:25, 1.42s/it] 94%|█████████▍| 28288/30000 [17:04:55<38:15, 1.34s/it] 94%|█████████▍| 28289/30000 [17:04:56<37:04, 1.30s/it] 94%|█████████▍| 28290/30000 [17:05:00<1:05:45, 2.31s/it] 94%|█████████▍| 28290/30000 [17:05:01<1:05:45, 2.31s/it] 94%|█████████▍| 28291/30000 [17:05:02<55:57, 1.96s/it] 94%|█████████▍| 28292/30000 [17:05:03<49:16, 1.73s/it] 94%|█████████▍| 28293/30000 [17:05:04<44:24, 1.56s/it] 94%|█████████▍| 28294/30000 [17:05:05<40:54, 1.44s/it] 94%|█████████▍| 28295/30000 [17:05:06<38:38, 1.36s/it] 94%|█████████▍| 28296/30000 [17:05:07<36:49, 1.30s/it] 94%|█████████▍| 28297/30000 [17:05:09<35:46, 1.26s/it] 94%|█████████▍| 28298/30000 [17:05:13<59:06, 2.08s/it] 94%|█████████▍| 28299/30000 [17:05:14<51:09, 1.80s/it] 94%|█████████▍| 28300/30000 [17:05:15<45:43, 1.61s/it] 94%|█████████▍| 28300/30000 [17:05:15<45:43, 1.61s/it] 94%|█████████▍| 28301/30000 [17:05:16<42:11, 1.49s/it] 94%|█████████▍| 28302/30000 [17:05:17<39:25, 1.39s/it] 94%|█████████▍| 28303/30000 [17:05:19<37:41, 1.33s/it] 94%|█████████▍| 28304/30000 [17:05:20<36:27, 1.29s/it] 94%|█████████▍| 28305/30000 [17:05:21<35:23, 1.25s/it] 94%|█████████▍| 28306/30000 [17:05:26<1:04:11, 2.27s/it] 94%|█████████▍| 28307/30000 [17:05:27<54:46, 1.94s/it] 94%|█████████▍| 28308/30000 [17:05:28<48:24, 1.72s/it] 94%|█████████▍| 28309/30000 [17:05:29<44:04, 1.56s/it] 94%|█████████▍| 28310/30000 [17:05:30<41:06, 1.46s/it] 94%|█████████▍| 28310/30000 [17:05:30<41:06, 1.46s/it] 94%|█████████▍| 28311/30000 [17:05:32<38:53, 1.38s/it] 94%|█████████▍| 28312/30000 [17:05:33<37:10, 1.32s/it] 94%|█████████▍| 28313/30000 [17:05:34<36:02, 1.28s/it] 94%|█████████▍| 28314/30000 [17:05:38<1:00:28, 2.15s/it] 94%|█████████▍| 28315/30000 [17:05:39<52:16, 1.86s/it] 94%|█████████▍| 28316/30000 [17:05:40<46:37, 1.66s/it] 94%|█████████▍| 28317/30000 [17:05:42<42:37, 1.52s/it] 94%|█████████▍| 28318/30000 [17:05:43<39:48, 1.42s/it] 94%|█████████▍| 28319/30000 [17:05:44<38:11, 1.36s/it] 94%|█████████▍| 28320/30000 [17:05:45<36:56, 1.32s/it] 94%|█████████▍| 28320/30000 [17:05:45<36:56, 1.32s/it] 94%|█████████▍| 28321/30000 [17:05:46<35:51, 1.28s/it] 94%|█████████▍| 28322/30000 [17:05:51<59:59, 2.15s/it] 94%|█████████▍| 28323/30000 [17:05:52<51:49, 1.85s/it] 94%|█████████▍| 28324/30000 [17:05:53<46:21, 1.66s/it] 94%|█████████▍| 28325/30000 [17:05:54<42:21, 1.52s/it] 94%|█████████▍| 28326/30000 [17:05:55<39:38, 1.42s/it] 94%|█████████▍| 28327/30000 [17:05:57<37:46, 1.35s/it] 94%|█████████▍| 28328/30000 [17:05:58<36:12, 1.30s/it] 94%|█████████▍| 28329/30000 [17:05:59<35:18, 1.27s/it] 94%|█████████▍| 28330/30000 [17:06:04<1:02:40, 2.25s/it] 94%|█████████▍| 28330/30000 [17:06:04<1:02:40, 2.25s/it] 94%|█████████▍| 28331/30000 [17:06:05<53:31, 1.92s/it] 94%|█████████▍| 28332/30000 [17:06:06<47:21, 1.70s/it] 94%|█████████▍| 28333/30000 [17:06:07<43:20, 1.56s/it] 94%|█████████▍| 28334/30000 [17:06:08<40:16, 1.45s/it] 94%|█████████▍| 28335/30000 [17:06:09<38:10, 1.38s/it] 94%|█████████▍| 28336/30000 [17:06:11<36:37, 1.32s/it] 94%|█████████▍| 28337/30000 [17:06:12<35:29, 1.28s/it] 94%|█████████▍| 28338/30000 [17:06:16<1:00:14, 2.17s/it] 94%|█████████▍| 28339/30000 [17:06:17<51:56, 1.88s/it] 94%|█████████▍| 28340/30000 [17:06:19<46:16, 1.67s/it] 94%|█████████▍| 28340/30000 [17:06:19<46:16, 1.67s/it] 94%|█████████▍| 28341/30000 [17:06:20<42:14, 1.53s/it] 94%|█████████▍| 28342/30000 [17:06:21<39:26, 1.43s/it] 94%|█████████▍| 28343/30000 [17:06:22<37:30, 1.36s/it] 94%|█████████▍| 28344/30000 [17:06:23<36:19, 1.32s/it] 94%|█████████▍| 28345/30000 [17:06:25<35:21, 1.28s/it] 94%|█████████▍| 28346/30000 [17:06:29<59:07, 2.14s/it] 94%|█████████▍| 28347/30000 [17:06:30<51:05, 1.85s/it] 94%|█████████▍| 28348/30000 [17:06:31<45:39, 1.66s/it] 94%|█████████▍| 28349/30000 [17:06:32<41:49, 1.52s/it] 94%|█████████▍| 28350/30000 [17:06:33<39:10, 1.42s/it] 94%|█████████▍| 28350/30000 [17:06:33<39:10, 1.42s/it] 95%|█████████▍| 28351/30000 [17:06:35<37:26, 1.36s/it] 95%|█████████▍| 28352/30000 [17:06:36<36:06, 1.31s/it] 95%|█████████▍| 28353/30000 [17:06:37<35:07, 1.28s/it] 95%|█████████▍| 28354/30000 [17:06:41<57:20, 2.09s/it] 95%|█████████▍| 28355/30000 [17:06:42<49:56, 1.82s/it] 95%|█████████▍| 28356/30000 [17:06:43<44:46, 1.63s/it] 95%|█████████▍| 28357/30000 [17:06:45<40:54, 1.49s/it] 95%|█████████▍| 28358/30000 [17:06:46<38:29, 1.41s/it] 95%|█████████▍| 28359/30000 [17:06:47<36:52, 1.35s/it] 95%|█████████▍| 28360/30000 [17:06:48<35:41, 1.31s/it] 95%|█████████▍| 28360/30000 [17:06:48<35:41, 1.31s/it] 95%|█████████▍| 28361/30000 [17:06:49<34:42, 1.27s/it] 95%|█████████▍| 28362/30000 [17:06:53<56:28, 2.07s/it] 95%|█████████▍| 28363/30000 [17:06:54<48:59, 1.80s/it] 95%|█████████▍| 28364/30000 [17:06:56<43:56, 1.61s/it] 95%|█████████▍| 28365/30000 [17:06:57<40:26, 1.48s/it] 95%|█████████▍| 28366/30000 [17:06:58<38:02, 1.40s/it] 95%|█████████▍| 28367/30000 [17:06:59<36:24, 1.34s/it] 95%|█████████▍| 28368/30000 [17:07:00<35:16, 1.30s/it] 95%|█████████▍| 28369/30000 [17:07:02<34:22, 1.26s/it] 95%|█████████▍| 28370/30000 [17:07:06<56:15, 2.07s/it] 95%|█████████▍| 28370/30000 [17:07:06<56:15, 2.07s/it] 95%|█████████▍| 28371/30000 [17:07:07<48:59, 1.80s/it] 95%|█████████▍| 28372/30000 [17:07:08<43:49, 1.62s/it] 95%|█████████▍| 28373/30000 [17:07:09<40:19, 1.49s/it] 95%|█████████▍| 28374/30000 [17:07:10<37:50, 1.40s/it] 95%|█████████▍| 28375/30000 [17:07:12<36:18, 1.34s/it] 95%|█████████▍| 28376/30000 [17:07:13<35:04, 1.30s/it] 95%|█████████▍| 28377/30000 [17:07:14<34:12, 1.26s/it] 95%|█████████▍| 28378/30000 [17:07:18<55:57, 2.07s/it] 95%|█████████▍| 28379/30000 [17:07:19<48:40, 1.80s/it] 95%|█████████▍| 28380/30000 [17:07:20<43:47, 1.62s/it] 95%|█████████▍| 28380/30000 [17:07:20<43:47, 1.62s/it] 95%|█████████▍| 28381/30000 [17:07:21<40:22, 1.50s/it] 95%|█████████▍| 28382/30000 [17:07:23<37:56, 1.41s/it] 95%|█████████▍| 28383/30000 [17:07:24<36:24, 1.35s/it] 95%|█████████▍| 28384/30000 [17:07:25<35:14, 1.31s/it] 95%|█████████▍| 28385/30000 [17:07:26<34:43, 1.29s/it] 95%|█████████▍| 28386/30000 [17:07:30<56:31, 2.10s/it] 95%|█████████▍| 28387/30000 [17:07:31<48:58, 1.82s/it] 95%|█████████▍| 28388/30000 [17:07:33<43:57, 1.64s/it] 95%|█████████▍| 28389/30000 [17:07:34<40:33, 1.51s/it] 95%|█████████▍| 28390/30000 [17:07:35<37:56, 1.41s/it] 95%|█████████▍| 28390/30000 [17:07:35<37:56, 1.41s/it] 95%|█████████▍| 28391/30000 [17:07:36<36:21, 1.36s/it] 95%|█████████▍| 28392/30000 [17:07:38<35:15, 1.32s/it] 95%|█████████▍| 28393/30000 [17:07:39<34:14, 1.28s/it] 95%|█████████▍| 28394/30000 [17:07:43<56:52, 2.12s/it] 95%|█████████▍| 28395/30000 [17:07:44<49:10, 1.84s/it] 95%|█████████▍| 28396/30000 [17:07:45<43:56, 1.64s/it] 95%|█████████▍| 28397/30000 [17:07:46<40:20, 1.51s/it] 95%|█████████▍| 28398/30000 [17:07:48<37:57, 1.42s/it] 95%|█████████▍| 28399/30000 [17:07:49<36:20, 1.36s/it] 95%|█████████▍| 28400/30000 [17:07:50<34:55, 1.31s/it] 95%|█████████▍| 28400/30000 [17:07:50<34:55, 1.31s/it] 95%|█████████▍| 28401/30000 [17:07:51<34:01, 1.28s/it] 95%|█████████▍| 28402/30000 [17:07:55<57:55, 2.17s/it] 95%|█████████▍| 28403/30000 [17:07:57<49:42, 1.87s/it] 95%|█████████▍| 28404/30000 [17:07:58<44:18, 1.67s/it] 95%|█████████▍| 28405/30000 [17:07:59<40:30, 1.52s/it] 95%|█████████▍| 28406/30000 [17:08:00<37:52, 1.43s/it] 95%|█████████▍| 28407/30000 [17:08:01<36:17, 1.37s/it] 95%|█████████▍| 28408/30000 [17:08:03<34:57, 1.32s/it] 95%|█████████▍| 28409/30000 [17:08:04<33:52, 1.28s/it] 95%|█████████▍| 28410/30000 [17:08:08<53:15, 2.01s/it] 95%|█████████▍| 28410/30000 [17:08:08<53:15, 2.01s/it] 95%|█████████▍| 28411/30000 [17:08:09<46:38, 1.76s/it] 95%|█████████▍| 28412/30000 [17:08:10<42:13, 1.60s/it] 95%|█████████▍| 28413/30000 [17:08:11<39:01, 1.48s/it] 95%|█████████▍| 28414/30000 [17:08:12<36:43, 1.39s/it] 95%|█████████▍| 28415/30000 [17:08:14<35:26, 1.34s/it] 95%|█████████▍| 28416/30000 [17:08:15<34:12, 1.30s/it] 95%|█████████▍| 28417/30000 [17:08:16<33:19, 1.26s/it] 95%|█████████▍| 28418/30000 [17:08:20<57:56, 2.20s/it] 95%|█████████▍| 28419/30000 [17:08:21<49:46, 1.89s/it] 95%|█████████▍| 28420/30000 [17:08:23<44:08, 1.68s/it] 95%|█████████▍| 28420/30000 [17:08:23<44:08, 1.68s/it] 95%|█████████▍| 28421/30000 [17:08:24<40:13, 1.53s/it] 95%|█████████▍| 28422/30000 [17:08:25<37:33, 1.43s/it] 95%|█████████▍| 28423/30000 [17:08:26<35:47, 1.36s/it] 95%|█████████▍| 28424/30000 [17:08:27<34:26, 1.31s/it] 95%|█████████▍| 28425/30000 [17:08:29<33:28, 1.28s/it] 95%|█████████▍| 28426/30000 [17:08:32<50:18, 1.92s/it] 95%|█████████▍| 28427/30000 [17:08:33<44:27, 1.70s/it] 95%|█████████▍| 28428/30000 [17:08:34<40:24, 1.54s/it] 95%|█████████▍| 28429/30000 [17:08:36<37:36, 1.44s/it] 95%|█████████▍| 28430/30000 [17:08:37<35:31, 1.36s/it] 95%|█████████▍| 28430/30000 [17:08:37<35:31, 1.36s/it] 95%|█████████▍| 28431/30000 [17:08:38<34:26, 1.32s/it] 95%|█████████▍| 28432/30000 [17:08:39<33:27, 1.28s/it] 95%|█████████▍| 28433/30000 [17:08:40<32:46, 1.25s/it] 95%|█████████▍| 28434/30000 [17:08:45<56:47, 2.18s/it] 95%|█████████▍| 28435/30000 [17:08:46<48:48, 1.87s/it] 95%|█████████▍| 28436/30000 [17:08:47<43:32, 1.67s/it] 95%|█████████▍| 28437/30000 [17:08:48<39:46, 1.53s/it] 95%|█████████▍| 28438/30000 [17:08:49<37:10, 1.43s/it] 95%|█████████▍| 28439/30000 [17:08:51<35:34, 1.37s/it] 95%|█████████▍| 28440/30000 [17:08:52<34:18, 1.32s/it] 95%|█████████▍| 28440/30000 [17:08:52<34:18, 1.32s/it] 95%|█████████▍| 28441/30000 [17:08:53<33:27, 1.29s/it] 95%|█████████▍| 28442/30000 [17:08:57<53:33, 2.06s/it] 95%|█████████▍| 28443/30000 [17:08:58<46:42, 1.80s/it] 95%|█████████▍| 28444/30000 [17:08:59<42:01, 1.62s/it] 95%|█████████▍| 28445/30000 [17:09:01<38:33, 1.49s/it] 95%|█████████▍| 28446/30000 [17:09:02<36:14, 1.40s/it] 95%|█████████▍| 28447/30000 [17:09:03<34:41, 1.34s/it] 95%|█████████▍| 28448/30000 [17:09:04<33:32, 1.30s/it] 95%|█████████▍| 28449/30000 [17:09:05<32:50, 1.27s/it] 95%|█████████▍| 28450/30000 [17:09:09<52:03, 2.01s/it] 95%|█████████▍| 28450/30000 [17:09:09<52:03, 2.01s/it] 95%|█████████▍| 28451/30000 [17:09:10<45:30, 1.76s/it] 95%|█████████▍| 28452/30000 [17:09:11<41:02, 1.59s/it] 95%|█████████▍| 28453/30000 [17:09:13<38:04, 1.48s/it] 95%|█████████▍| 28454/30000 [17:09:14<35:50, 1.39s/it] 95%|█████████▍| 28455/30000 [17:09:15<34:32, 1.34s/it] 95%|█████████▍| 28456/30000 [17:09:16<33:25, 1.30s/it] 95%|█████████▍| 28457/30000 [17:09:17<32:35, 1.27s/it] 95%|█████████▍| 28458/30000 [17:09:21<51:57, 2.02s/it] 95%|█████████▍| 28459/30000 [17:09:22<45:18, 1.76s/it] 95%|█████████▍| 28460/30000 [17:09:24<41:00, 1.60s/it] 95%|█████████▍| 28460/30000 [17:09:24<41:00, 1.60s/it] 95%|█████████▍| 28461/30000 [17:09:25<37:54, 1.48s/it] 95%|█████████▍| 28462/30000 [17:09:26<35:58, 1.40s/it] 95%|█████████▍| 28463/30000 [17:09:27<35:13, 1.38s/it] 95%|█████████▍| 28464/30000 [17:09:29<33:45, 1.32s/it] 95%|█████████▍| 28465/30000 [17:09:30<32:52, 1.28s/it] 95%|█████████▍| 28466/30000 [17:09:34<54:32, 2.13s/it] 95%|█████████▍| 28467/30000 [17:09:35<47:09, 1.85s/it] 95%|█████████▍| 28468/30000 [17:09:36<42:07, 1.65s/it] 95%|█████████▍| 28469/30000 [17:09:37<38:34, 1.51s/it] 95%|█████████▍| 28470/30000 [17:09:39<36:05, 1.42s/it] 95%|█████████▍| 28470/30000 [17:09:39<36:05, 1.42s/it] 95%|█████████▍| 28471/30000 [17:09:40<34:26, 1.35s/it] 95%|█████████▍| 28472/30000 [17:09:41<33:15, 1.31s/it] 95%|█████████▍| 28473/30000 [17:09:42<32:44, 1.29s/it] 95%|█████████▍| 28474/30000 [17:09:46<54:25, 2.14s/it] 95%|█████████▍| 28475/30000 [17:09:48<46:47, 1.84s/it] 95%|█████████▍| 28476/30000 [17:09:49<41:46, 1.64s/it] 95%|█████████▍| 28477/30000 [17:09:50<38:18, 1.51s/it] 95%|█████████▍| 28478/30000 [17:09:51<35:56, 1.42s/it] 95%|█████████▍| 28479/30000 [17:09:52<34:19, 1.35s/it] 95%|█████████▍| 28480/30000 [17:09:54<33:10, 1.31s/it] 95%|█████████▍| 28480/30000 [17:09:54<33:10, 1.31s/it] 95%|█████████▍| 28481/30000 [17:09:55<32:14, 1.27s/it] 95%|█████████▍| 28482/30000 [17:09:59<55:33, 2.20s/it] 95%|█████████▍| 28483/30000 [17:10:00<47:42, 1.89s/it] 95%|█████████▍| 28484/30000 [17:10:01<42:25, 1.68s/it] 95%|█████████▍| 28485/30000 [17:10:03<38:43, 1.53s/it] 95%|█████████▍| 28486/30000 [17:10:04<36:09, 1.43s/it] 95%|█████████▍| 28487/30000 [17:10:05<34:25, 1.36s/it] 95%|█████████▍| 28488/30000 [17:10:06<33:12, 1.32s/it] 95%|█████████▍| 28489/30000 [17:10:07<32:09, 1.28s/it] 95%|█████████▍| 28490/30000 [17:10:12<54:54, 2.18s/it] 95%|█████████▍| 28490/30000 [17:10:12<54:54, 2.18s/it] 95%|█████████▍| 28491/30000 [17:10:13<47:09, 1.88s/it] 95%|█████████▍| 28492/30000 [17:10:14<41:58, 1.67s/it] 95%|█████████▍| 28493/30000 [17:10:15<38:20, 1.53s/it] 95%|█████████▍| 28494/30000 [17:10:16<35:46, 1.43s/it] 95%|█████████▍| 28495/30000 [17:10:18<34:58, 1.39s/it] 95%|█████████▍| 28496/30000 [17:10:19<33:27, 1.34s/it] 95%|█████████▍| 28497/30000 [17:10:20<32:19, 1.29s/it] 95%|█████████▍| 28498/30000 [17:10:24<50:51, 2.03s/it] 95%|█████████▍| 28499/30000 [17:10:25<44:20, 1.77s/it] 95%|█████████▌| 28500/30000 [17:10:26<39:59, 1.60s/it] 95%|█████████▌| 28500/30000 [17:10:26<39:59, 1.60s/it] 95%|█████████▌| 28501/30000 [17:10:27<37:01, 1.48s/it] 95%|█████████▌| 28502/30000 [17:10:29<34:57, 1.40s/it] 95%|█████████▌| 28503/30000 [17:10:30<34:36, 1.39s/it] 95%|█████████▌| 28504/30000 [17:10:31<33:03, 1.33s/it] 95%|█████████▌| 28505/30000 [17:10:32<31:58, 1.28s/it] 95%|█████████▌| 28506/30000 [17:10:36<50:57, 2.05s/it] 95%|█████████▌| 28507/30000 [17:10:37<44:27, 1.79s/it] 95%|█████████▌| 28508/30000 [17:10:39<40:05, 1.61s/it] 95%|█████████▌| 28509/30000 [17:10:40<36:54, 1.49s/it] 95%|█████████▌| 28510/30000 [17:10:41<34:52, 1.40s/it] 95%|█████████▌| 28510/30000 [17:10:41<34:52, 1.40s/it] 95%|█████████▌| 28511/30000 [17:10:42<33:18, 1.34s/it] 95%|█████████▌| 28512/30000 [17:10:43<32:08, 1.30s/it] 95%|█████████▌| 28513/30000 [17:10:45<31:11, 1.26s/it] 95%|█████████▌| 28514/30000 [17:10:49<51:51, 2.09s/it] 95%|█████████▌| 28515/30000 [17:10:50<44:58, 1.82s/it] 95%|█████████▌| 28516/30000 [17:10:51<40:18, 1.63s/it] 95%|█████████▌| 28517/30000 [17:10:52<37:04, 1.50s/it] 95%|█████████▌| 28518/30000 [17:10:53<34:51, 1.41s/it] 95%|█████████▌| 28519/30000 [17:10:55<33:22, 1.35s/it] 95%|█████████▌| 28520/30000 [17:10:56<32:07, 1.30s/it] 95%|█████████▌| 28520/30000 [17:10:56<32:07, 1.30s/it] 95%|█████████▌| 28521/30000 [17:10:57<31:15, 1.27s/it] 95%|█████████▌| 28522/30000 [17:11:01<51:17, 2.08s/it] 95%|█████████▌| 28523/30000 [17:11:02<44:32, 1.81s/it] 95%|█████████▌| 28524/30000 [17:11:03<40:08, 1.63s/it] 95%|█████████▌| 28525/30000 [17:11:05<36:49, 1.50s/it] 95%|█████████▌| 28526/30000 [17:11:06<34:28, 1.40s/it] 95%|█████████▌| 28527/30000 [17:11:07<32:59, 1.34s/it] 95%|█████████▌| 28528/30000 [17:11:08<31:44, 1.29s/it] 95%|█████████▌| 28529/30000 [17:11:09<30:55, 1.26s/it] 95%|█████████▌| 28530/30000 [17:11:13<51:37, 2.11s/it] 95%|█████████▌| 28530/30000 [17:11:13<51:37, 2.11s/it] 95%|█████████▌| 28531/30000 [17:11:15<44:39, 1.82s/it] 95%|█████████▌| 28532/30000 [17:11:16<40:03, 1.64s/it] 95%|█████████▌| 28533/30000 [17:11:17<36:57, 1.51s/it] 95%|█████████▌| 28534/30000 [17:11:18<34:37, 1.42s/it] 95%|█████████▌| 28535/30000 [17:11:19<33:05, 1.36s/it] 95%|█████████▌| 28536/30000 [17:11:21<31:39, 1.30s/it] 95%|█████████▌| 28537/30000 [17:11:22<30:46, 1.26s/it] 95%|█████████▌| 28538/30000 [17:11:26<55:17, 2.27s/it] 95%|█████████▌| 28539/30000 [17:11:28<47:34, 1.95s/it] 95%|█████████▌| 28540/30000 [17:11:29<42:13, 1.74s/it] 95%|█████████▌| 28540/30000 [17:11:29<42:13, 1.74s/it] 95%|█████████▌| 28541/30000 [17:11:30<38:15, 1.57s/it] 95%|█████████▌| 28542/30000 [17:11:31<35:25, 1.46s/it] 95%|█████████▌| 28543/30000 [17:11:32<33:46, 1.39s/it] 95%|█████████▌| 28544/30000 [17:11:34<32:18, 1.33s/it] 95%|█████████▌| 28545/30000 [17:11:35<31:18, 1.29s/it] 95%|█████████▌| 28546/30000 [17:11:39<55:55, 2.31s/it] 95%|█████████▌| 28547/30000 [17:11:41<47:30, 1.96s/it] 95%|█████████▌| 28548/30000 [17:11:42<41:59, 1.74s/it] 95%|█████████▌| 28549/30000 [17:11:43<38:12, 1.58s/it] 95%|█████████▌| 28550/30000 [17:11:44<35:21, 1.46s/it] 95%|█████████▌| 28550/30000 [17:11:44<35:21, 1.46s/it] 95%|█████████▌| 28551/30000 [17:11:45<33:34, 1.39s/it] 95%|█████████▌| 28552/30000 [17:11:47<32:07, 1.33s/it] 95%|█████████▌| 28553/30000 [17:11:48<31:02, 1.29s/it] 95%|█████████▌| 28554/30000 [17:11:53<56:28, 2.34s/it] 95%|█████████▌| 28555/30000 [17:11:54<47:59, 1.99s/it] 95%|█████████▌| 28556/30000 [17:11:55<42:04, 1.75s/it] 95%|█████████▌| 28557/30000 [17:11:56<38:01, 1.58s/it] 95%|█████████▌| 28558/30000 [17:11:57<35:16, 1.47s/it] 95%|█████████▌| 28559/30000 [17:11:59<33:18, 1.39s/it] 95%|█████████▌| 28560/30000 [17:12:00<31:48, 1.33s/it] {'loss': 0.0137, 'grad_norm': 0.08348139375448227, 'learning_rate': 3.0781366277997716e-07, 'flow_matching_loss': 0.013746024947613478, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.19} +{'loss': 0.0143, 'grad_norm': 0.05789078772068024, 'learning_rate': 3.044900904974762e-07, 'flow_matching_loss': 0.01425830963999033, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.2} +{'loss': 0.0127, 'grad_norm': 0.06216936558485031, 'learning_rate': 3.011843746452081e-07, 'flow_matching_loss': 0.01273475270718336, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.2} +{'loss': 0.0152, 'grad_norm': 0.06656448543071747, 'learning_rate': 2.9789651923993464e-07, 'flow_matching_loss': 0.015218209475278854, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.2} +{'loss': 0.0118, 'grad_norm': 0.05962615832686424, 'learning_rate': 2.9462652827670855e-07, 'flow_matching_loss': 0.01184342335909605, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.2} +{'loss': 0.0132, 'grad_norm': 0.08668399602174759, 'learning_rate': 2.913744057288764e-07, 'flow_matching_loss': 0.013213720172643661, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.21} +{'loss': 0.0146, 'grad_norm': 0.1074739322066307, 'learning_rate': 2.8814015554807396e-07, 'flow_matching_loss': 0.014583468157798051, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.21} +{'loss': 0.0131, 'grad_norm': 0.055379968136548996, 'learning_rate': 2.849237816642192e-07, 'flow_matching_loss': 0.013093191105872393, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.21} +{'loss': 0.0126, 'grad_norm': 0.06330244988203049, 'learning_rate': 2.817252879855109e-07, 'flow_matching_loss': 0.012606904841959476, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.22} +{'loss': 0.0121, 'grad_norm': 0.07152049988508224, 'learning_rate': 2.785446783984219e-07, 'flow_matching_loss': 0.012128066271543503, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.22} +{'loss': 0.011, 'grad_norm': 0.06426088511943817, 'learning_rate': 2.7538195676769076e-07, 'flow_matching_loss': 0.011026451829820871, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.22} +{'loss': 0.0136, 'grad_norm': 0.06925376504659653, 'learning_rate': 2.722371269363216e-07, 'flow_matching_loss': 0.013621341623365878, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.23} +{'loss': 0.0135, 'grad_norm': 0.06542079150676727, 'learning_rate': 2.6911019272558125e-07, 'flow_matching_loss': 0.013530756160616875, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.23} +{'loss': 0.0116, 'grad_norm': 0.07104208320379257, 'learning_rate': 2.6600115793498857e-07, 'flow_matching_loss': 0.011573642492294312, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.23} +{'loss': 0.0135, 'grad_norm': 0.09723030030727386, 'learning_rate': 2.629100263423151e-07, 'flow_matching_loss': 0.013519202172756196, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.23} +{'loss': 0.0117, 'grad_norm': 0.06337190419435501, 'learning_rate': 2.598368017035763e-07, 'flow_matching_loss': 0.011686358274891973, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.24} +{'loss': 0.0136, 'grad_norm': 0.07275745272636414, 'learning_rate': 2.5678148775303e-07, 'flow_matching_loss': 0.01355264289304614, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.24} +{'loss': 0.0125, 'grad_norm': 0.07644246518611908, 'learning_rate': 2.537440882031683e-07, 'flow_matching_loss': 0.012549559026956559, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.24} +{'loss': 0.0136, 'grad_norm': 0.06513189524412155, 'learning_rate': 2.5072460674472385e-07, 'flow_matching_loss': 0.0135580369271338, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.25} +{'loss': 0.0116, 'grad_norm': 0.07351565361022949, 'learning_rate': 2.4772304704664683e-07, 'flow_matching_loss': 0.011599574983119965, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.25} +{'loss': 0.0126, 'grad_norm': 0.08159194886684418, 'learning_rate': 2.447394127561181e-07, 'flow_matching_loss': 0.012605345621705056, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.25} +{'loss': 0.0136, 'grad_norm': 0.07246667891740799, 'learning_rate': 2.41773707498531e-07, 'flow_matching_loss': 0.013568165712058544, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.25} +{'loss': 0.0126, 'grad_norm': 0.07399924099445343, 'learning_rate': 2.388259348774996e-07, 'flow_matching_loss': 0.012622840888798236, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.26} +{'loss': 0.0129, 'grad_norm': 0.07642663270235062, 'learning_rate': 2.358960984748454e-07, 'flow_matching_loss': 0.01289844149723649, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.26} +{'loss': 0.0116, 'grad_norm': 0.050575993955135345, 'learning_rate': 2.3298420185059565e-07, 'flow_matching_loss': 0.011633377242833375, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.26} +{'loss': 0.0134, 'grad_norm': 0.0717485323548317, 'learning_rate': 2.3009024854297844e-07, 'flow_matching_loss': 0.013370336405932904, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.27} +{'loss': 0.012, 'grad_norm': 0.07514847815036774, 'learning_rate': 2.272142420684209e-07, 'flow_matching_loss': 0.012039550207555294, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.27} +{'loss': 0.0139, 'grad_norm': 0.08123744279146194, 'learning_rate': 2.243561859215426e-07, 'flow_matching_loss': 0.013867509271949529, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.27} +{'loss': 0.0128, 'grad_norm': 0.06769362092018127, 'learning_rate': 2.2151608357514897e-07, 'flow_matching_loss': 0.012806138303130866, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.27} +{'loss': 0.0123, 'grad_norm': 0.06167908385396004, 'learning_rate': 2.186939384802361e-07, 'flow_matching_loss': 0.012290109973400831, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.28} +{'loss': 0.0128, 'grad_norm': 0.0783676877617836, 'learning_rate': 2.1588975406597266e-07, 'flow_matching_loss': 0.012751576770097018, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.28} +{'loss': 0.0126, 'grad_norm': 0.06741730868816376, 'learning_rate': 2.131035337397097e-07, 'flow_matching_loss': 0.012558369617909192, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.28} +{'loss': 0.0138, 'grad_norm': 0.07224731892347336, 'learning_rate': 2.1033528088696575e-07, 'flow_matching_loss': 0.013786687701940536, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.29} +{'loss': 0.0125, 'grad_norm': 0.08348555862903595, 'learning_rate': 2.0758499887143012e-07, 'flow_matching_loss': 0.012525017373263836, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.29} +{'loss': 0.0127, 'grad_norm': 0.06260176002979279, 'learning_rate': 2.0485269103495462e-07, 'flow_matching_loss': 0.01271587861701846, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.29} +{'loss': 0.0126, 'grad_norm': 0.07085982710123062, 'learning_rate': 2.021383606975502e-07, 'flow_matching_loss': 0.012587063387036324, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.3} +{'loss': 0.0122, 'grad_norm': 0.06609652191400528, 'learning_rate': 1.9944201115738524e-07, 'flow_matching_loss': 0.012246784474700689, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.3} +{'loss': 0.0127, 'grad_norm': 0.06083147972822189, 'learning_rate': 1.9676364569077732e-07, 'flow_matching_loss': 0.012722001224756242, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.3} +{'loss': 0.0129, 'grad_norm': 0.0665624588727951, 'learning_rate': 1.9410326755219477e-07, 'flow_matching_loss': 0.012918277457356454, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.3} +{'loss': 0.0118, 'grad_norm': 0.0912712886929512, 'learning_rate': 1.9146087997424677e-07, 'flow_matching_loss': 0.01184877110645175, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.31} +{'loss': 0.0136, 'grad_norm': 0.07156643271446228, 'learning_rate': 1.8883648616768335e-07, 'flow_matching_loss': 0.013595526944845915, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.31} 95%|█████████▌| 28560/30000 [17:12:00<31:48, 1.33s/it] 95%|█████████▌| 28561/30000 [17:12:01<30:44, 1.28s/it] 95%|█████████▌| 28562/30000 [17:12:06<54:41, 2.28s/it] 95%|█████████▌| 28563/30000 [17:12:07<46:52, 1.96s/it] 95%|█████████▌| 28564/30000 [17:12:08<41:19, 1.73s/it] 95%|█████████▌| 28565/30000 [17:12:09<37:24, 1.56s/it] 95%|█████████▌| 28566/30000 [17:12:10<34:53, 1.46s/it] 95%|█████████▌| 28567/30000 [17:12:12<33:12, 1.39s/it] 95%|█████████▌| 28568/30000 [17:12:13<31:42, 1.33s/it] 95%|█████████▌| 28569/30000 [17:12:14<30:44, 1.29s/it] 95%|█████████▌| 28570/30000 [17:12:18<52:53, 2.22s/it] 95%|█████████▌| 28570/30000 [17:12:18<52:53, 2.22s/it] 95%|█████████▌| 28571/30000 [17:12:20<45:29, 1.91s/it] 95%|█████████▌| 28572/30000 [17:12:21<40:18, 1.69s/it] 95%|█████████▌| 28573/30000 [17:12:22<36:42, 1.54s/it] 95%|█████████▌| 28574/30000 [17:12:23<34:07, 1.44s/it] 95%|█████████▌| 28575/30000 [17:12:24<32:21, 1.36s/it] 95%|█████████▌| 28576/30000 [17:12:25<31:10, 1.31s/it] 95%|█████████▌| 28577/30000 [17:12:27<30:16, 1.28s/it] 95%|█████████▌| 28578/30000 [17:12:32<55:30, 2.34s/it] 95%|█████████▌| 28579/30000 [17:12:33<47:16, 2.00s/it] 95%|█████████▌| 28580/30000 [17:12:34<41:30, 1.75s/it] 95%|█████████▌| 28580/30000 [17:12:34<41:30, 1.75s/it] 95%|█████████▌| 28581/30000 [17:12:35<37:35, 1.59s/it] 95%|█████████▌| 28582/30000 [17:12:36<34:47, 1.47s/it] 95%|█████████▌| 28583/30000 [17:12:38<32:58, 1.40s/it] 95%|█████████▌| 28584/30000 [17:12:39<31:39, 1.34s/it] 95%|█████████▌| 28585/30000 [17:12:40<30:45, 1.30s/it] 95%|█████████▌| 28586/30000 [17:12:44<50:25, 2.14s/it] 95%|█████████▌| 28587/30000 [17:12:45<43:28, 1.85s/it] 95%|█████████▌| 28588/30000 [17:12:46<38:42, 1.64s/it] 95%|█████████▌| 28589/30000 [17:12:48<35:32, 1.51s/it] 95%|█████████▌| 28590/30000 [17:12:49<33:11, 1.41s/it] 95%|█████████▌| 28590/30000 [17:12:49<33:11, 1.41s/it] 95%|█████████▌| 28591/30000 [17:12:50<31:45, 1.35s/it] 95%|█████████▌| 28592/30000 [17:12:51<30:33, 1.30s/it] 95%|█████████▌| 28593/30000 [17:12:52<29:43, 1.27s/it] 95%|█████████▌| 28594/30000 [17:12:57<51:23, 2.19s/it] 95%|█████████▌| 28595/30000 [17:12:58<44:02, 1.88s/it] 95%|█████████▌| 28596/30000 [17:12:59<39:10, 1.67s/it] 95%|█████████▌| 28597/30000 [17:13:00<35:44, 1.53s/it] 95%|█████████▌| 28598/30000 [17:13:01<33:27, 1.43s/it] 95%|█████████▌| 28599/30000 [17:13:03<31:50, 1.36s/it] 95%|█████████▌| 28600/30000 [17:13:04<30:34, 1.31s/it] 95%|█████████▌| 28600/30000 [17:13:04<30:34, 1.31s/it] 95%|█████████▌| 28601/30000 [17:13:05<29:40, 1.27s/it] 95%|█████████▌| 28602/30000 [17:13:09<49:12, 2.11s/it] 95%|█████████▌| 28603/30000 [17:13:10<42:32, 1.83s/it] 95%|█████████▌| 28604/30000 [17:13:11<38:04, 1.64s/it] 95%|█████████▌| 28605/30000 [17:13:13<35:08, 1.51s/it] 95%|█████████▌| 28606/30000 [17:13:14<32:48, 1.41s/it] 95%|█████████▌| 28607/30000 [17:13:15<31:22, 1.35s/it] 95%|█████████▌| 28608/30000 [17:13:16<30:19, 1.31s/it] 95%|█████████▌| 28609/30000 [17:13:17<29:26, 1.27s/it] 95%|█████████▌| 28610/30000 [17:13:21<47:01, 2.03s/it] 95%|█████████▌| 28610/30000 [17:13:21<47:01, 2.03s/it] 95%|█████████▌| 28611/30000 [17:13:22<40:58, 1.77s/it] 95%|█████████▌| 28612/30000 [17:13:24<36:52, 1.59s/it] 95%|█████████▌| 28613/30000 [17:13:25<34:01, 1.47s/it] 95%|█████████▌| 28614/30000 [17:13:26<32:17, 1.40s/it] 95%|█████████▌| 28615/30000 [17:13:27<30:58, 1.34s/it] 95%|█████████▌| 28616/30000 [17:13:28<29:48, 1.29s/it] 95%|█████████▌| 28617/30000 [17:13:30<29:03, 1.26s/it] 95%|█████████▌| 28618/30000 [17:13:34<50:56, 2.21s/it] 95%|█████████▌| 28619/30000 [17:13:35<43:45, 1.90s/it] 95%|█████████▌| 28620/30000 [17:13:36<38:50, 1.69s/it] 95%|█████████▌| 28620/30000 [17:13:36<38:50, 1.69s/it] 95%|█████████▌| 28621/30000 [17:13:38<35:20, 1.54s/it] 95%|█████████▌| 28622/30000 [17:13:39<32:59, 1.44s/it] 95%|█████████▌| 28623/30000 [17:13:40<31:26, 1.37s/it] 95%|█████████▌| 28624/30000 [17:13:41<30:33, 1.33s/it] 95%|█████████▌| 28625/30000 [17:13:42<29:57, 1.31s/it] 95%|█████████▌| 28626/30000 [17:13:46<47:52, 2.09s/it] 95%|█████████▌| 28627/30000 [17:13:48<41:30, 1.81s/it] 95%|█████████▌| 28628/30000 [17:13:49<37:12, 1.63s/it] 95%|█████████▌| 28629/30000 [17:13:50<34:10, 1.50s/it] 95%|█████████▌| 28630/30000 [17:13:51<32:00, 1.40s/it] 95%|█████████▌| 28630/30000 [17:13:51<32:00, 1.40s/it] 95%|█████████▌| 28631/30000 [17:13:52<30:41, 1.35s/it] 95%|█████████▌| 28632/30000 [17:13:54<29:48, 1.31s/it] 95%|█████████▌| 28633/30000 [17:13:55<29:06, 1.28s/it] 95%|█████████▌| 28634/30000 [17:13:59<47:25, 2.08s/it] 95%|█████████▌| 28635/30000 [17:14:00<41:11, 1.81s/it] 95%|█████████▌| 28636/30000 [17:14:01<36:54, 1.62s/it] 95%|█████████▌| 28637/30000 [17:14:02<33:59, 1.50s/it] 95%|█████████▌| 28638/30000 [17:14:03<31:51, 1.40s/it] 95%|█████████▌| 28639/30000 [17:14:05<30:23, 1.34s/it] 95%|█████████▌| 28640/30000 [17:14:06<29:07, 1.28s/it] 95%|█████████▌| 28640/30000 [17:14:06<29:07, 1.28s/it] 95%|█████████▌| 28641/30000 [17:14:07<28:27, 1.26s/it] 95%|█████████▌| 28642/30000 [17:14:12<52:22, 2.31s/it] 95%|█████████▌| 28643/30000 [17:14:13<44:39, 1.97s/it] 95%|█████████▌| 28644/30000 [17:14:14<39:19, 1.74s/it] 95%|█████████▌| 28645/30000 [17:14:15<35:32, 1.57s/it] 95%|█████████▌| 28646/30000 [17:14:17<32:46, 1.45s/it] 95%|█████████▌| 28647/30000 [17:14:18<31:02, 1.38s/it] 95%|█████████▌| 28648/30000 [17:14:19<29:43, 1.32s/it] 95%|█████████▌| 28649/30000 [17:14:20<28:50, 1.28s/it] 96%|█████████▌| 28650/30000 [17:14:25<50:36, 2.25s/it] 96%|█████████▌| 28650/30000 [17:14:25<50:36, 2.25s/it] 96%|█████████▌| 28651/30000 [17:14:26<43:19, 1.93s/it] 96%|█████████▌| 28652/30000 [17:14:27<38:18, 1.71s/it] 96%|█████████▌| 28653/30000 [17:14:28<34:52, 1.55s/it] 96%|█████████▌| 28654/30000 [17:14:29<32:22, 1.44s/it] 96%|█████████▌| 28655/30000 [17:14:31<30:40, 1.37s/it] 96%|█████████▌| 28656/30000 [17:14:32<29:25, 1.31s/it] 96%|█████████▌| 28657/30000 [17:14:33<28:41, 1.28s/it] 96%|█████████▌| 28658/30000 [17:14:38<53:39, 2.40s/it] 96%|█████████▌| 28659/30000 [17:14:39<45:16, 2.03s/it] 96%|█████████▌| 28660/30000 [17:14:40<39:39, 1.78s/it] 96%|█████████▌| 28660/30000 [17:14:40<39:39, 1.78s/it] 96%|█████████▌| 28661/30000 [17:14:41<35:41, 1.60s/it] 96%|█████████▌| 28662/30000 [17:14:43<32:54, 1.48s/it] 96%|█████████▌| 28663/30000 [17:14:44<30:56, 1.39s/it] 96%|█████████▌| 28664/30000 [17:14:45<29:40, 1.33s/it] 96%|█████████▌| 28665/30000 [17:14:46<28:42, 1.29s/it] 96%|█████████▌| 28666/30000 [17:14:50<48:30, 2.18s/it] 96%|█████████▌| 28667/30000 [17:14:52<41:45, 1.88s/it] 96%|█████████▌| 28668/30000 [17:14:53<37:05, 1.67s/it] 96%|█████████▌| 28669/30000 [17:14:54<33:51, 1.53s/it] 96%|█████████▌| 28670/30000 [17:14:55<31:34, 1.42s/it] 96%|█████████▌| 28670/30000 [17:14:55<31:34, 1.42s/it] 96%|█████████▌| 28671/30000 [17:14:56<30:03, 1.36s/it] 96%|█████████▌| 28672/30000 [17:14:58<28:57, 1.31s/it] 96%|█████████▌| 28673/30000 [17:14:59<28:20, 1.28s/it] 96%|█████████▌| 28674/30000 [17:15:03<45:35, 2.06s/it] 96%|█████████▌| 28675/30000 [17:15:04<39:37, 1.79s/it] 96%|█████████▌| 28676/30000 [17:15:05<35:33, 1.61s/it] 96%|█████████▌| 28677/30000 [17:15:06<32:49, 1.49s/it] 96%|█████████▌| 28678/30000 [17:15:07<30:50, 1.40s/it] 96%|█████████▌| 28679/30000 [17:15:09<29:42, 1.35s/it] 96%|█████████▌| 28680/30000 [17:15:10<28:35, 1.30s/it] 96%|█████████▌| 28680/30000 [17:15:10<28:35, 1.30s/it] 96%|█████████▌| 28681/30000 [17:15:11<27:49, 1.27s/it] 96%|█████████▌| 28682/30000 [17:15:15<46:06, 2.10s/it] 96%|█████████▌| 28683/30000 [17:15:16<40:06, 1.83s/it] 96%|█████████▌| 28684/30000 [17:15:18<35:52, 1.64s/it] 96%|█████████▌| 28685/30000 [17:15:19<32:42, 1.49s/it] 96%|█████████▌| 28686/30000 [17:15:20<31:00, 1.42s/it] 96%|█████████▌| 28687/30000 [17:15:21<29:30, 1.35s/it] 96%|█████████▌| 28688/30000 [17:15:22<28:28, 1.30s/it] 96%|█████████▌| 28689/30000 [17:15:23<27:40, 1.27s/it] 96%|█████████▌| 28690/30000 [17:15:27<45:23, 2.08s/it] 96%|█████████▌| 28690/30000 [17:15:27<45:23, 2.08s/it] 96%|█████████▌| 28691/30000 [17:15:29<39:19, 1.80s/it] 96%|█████████▌| 28692/30000 [17:15:30<35:16, 1.62s/it] 96%|█████████▌| 28693/30000 [17:15:31<32:24, 1.49s/it] 96%|█████████▌| 28694/30000 [17:15:32<30:15, 1.39s/it] 96%|█████████▌| 28695/30000 [17:15:33<28:57, 1.33s/it] 96%|█████████▌| 28696/30000 [17:15:35<27:57, 1.29s/it] 96%|█████████▌| 28697/30000 [17:15:36<27:12, 1.25s/it] 96%|█████████▌| 28698/30000 [17:15:40<44:06, 2.03s/it] 96%|█████████▌| 28699/30000 [17:15:41<38:35, 1.78s/it] 96%|█████████▌| 28700/30000 [17:15:42<34:52, 1.61s/it] 96%|█████████▌| 28700/30000 [17:15:42<34:52, 1.61s/it] 96%|█████████▌| 28701/30000 [17:15:43<32:10, 1.49s/it] 96%|█████████▌| 28702/30000 [17:15:44<30:05, 1.39s/it] 96%|█████████▌| 28703/30000 [17:15:46<28:50, 1.33s/it] 96%|█████████▌| 28704/30000 [17:15:47<27:56, 1.29s/it] 96%|█████████▌| 28705/30000 [17:15:48<27:25, 1.27s/it] 96%|█████████▌| 28706/30000 [17:15:52<47:52, 2.22s/it] 96%|█████████▌| 28707/30000 [17:15:54<41:00, 1.90s/it] 96%|█████████▌| 28708/30000 [17:15:55<36:20, 1.69s/it] 96%|█████████▌| 28709/30000 [17:15:56<33:10, 1.54s/it] 96%|█████████▌| 28710/30000 [17:15:57<30:54, 1.44s/it] 96%|█████████▌| 28710/30000 [17:15:57<30:54, 1.44s/it] 96%|█████████▌| 28711/30000 [17:15:58<29:19, 1.36s/it] 96%|█████████▌| 28712/30000 [17:15:59<28:13, 1.32s/it] 96%|█████████▌| 28713/30000 [17:16:01<27:26, 1.28s/it] 96%|█████████▌| 28714/30000 [17:16:05<46:56, 2.19s/it] 96%|█████████▌| 28715/30000 [17:16:06<40:16, 1.88s/it] 96%|█████████▌| 28716/30000 [17:16:07<35:40, 1.67s/it] 96%|█████████▌| 28717/30000 [17:16:09<32:31, 1.52s/it] 96%|█████████▌| 28718/30000 [17:16:10<30:23, 1.42s/it] 96%|█████████▌| 28719/30000 [17:16:11<28:52, 1.35s/it] 96%|█████████▌| 28720/30000 [17:16:12<27:44, 1.30s/it] 96%|█████████▌| 28720/30000 [17:16:12<27:44, 1.30s/it] 96%|█████████▌| 28721/30000 [17:16:13<27:03, 1.27s/it] 96%|█████████▌| 28722/30000 [17:16:18<46:43, 2.19s/it] 96%|█████████▌| 28723/30000 [17:16:19<40:03, 1.88s/it] 96%|█████████▌| 28724/30000 [17:16:20<35:26, 1.67s/it] 96%|█████████▌| 28725/30000 [17:16:21<32:24, 1.53s/it] 96%|█████████▌| 28726/30000 [17:16:22<30:23, 1.43s/it] 96%|█████████▌| 28727/30000 [17:16:24<28:46, 1.36s/it] 96%|█████████▌| 28728/30000 [17:16:25<27:35, 1.30s/it] 96%|█████████▌| 28729/30000 [17:16:26<26:55, 1.27s/it] 96%|█████████▌| 28730/30000 [17:16:30<46:52, 2.21s/it] 96%|█████████▌| 28730/30000 [17:16:30<46:52, 2.21s/it] 96%|█████████▌| 28731/30000 [17:16:31<40:12, 1.90s/it] 96%|█████████▌| 28732/30000 [17:16:33<35:49, 1.69s/it] 96%|█████████▌| 28733/30000 [17:16:34<32:29, 1.54s/it] 96%|█████████▌| 28734/30000 [17:16:35<30:15, 1.43s/it] 96%|█████████▌| 28735/30000 [17:16:36<28:43, 1.36s/it] 96%|█████████▌| 28736/30000 [17:16:37<27:34, 1.31s/it] 96%|█████████▌| 28737/30000 [17:16:39<26:47, 1.27s/it] 96%|█████████▌| 28738/30000 [17:16:43<45:34, 2.17s/it] 96%|█████████▌| 28739/30000 [17:16:44<39:07, 1.86s/it] 96%|█████████▌| 28740/30000 [17:16:45<34:41, 1.65s/it] 96%|█████████▌| 28740/30000 [17:16:45<34:41, 1.65s/it] 96%|█████████▌| 28741/30000 [17:16:46<31:41, 1.51s/it] 96%|█████████▌| 28742/30000 [17:16:48<29:36, 1.41s/it] 96%|█████████▌| 28743/30000 [17:16:49<28:11, 1.35s/it] 96%|█████████▌| 28744/30000 [17:16:50<27:02, 1.29s/it] 96%|█████████▌| 28745/30000 [17:16:51<26:16, 1.26s/it] 96%|█████████▌| 28746/30000 [17:16:56<47:31, 2.27s/it] 96%|█████████▌| 28747/30000 [17:16:57<40:36, 1.94s/it] 96%|█████████▌| 28748/30000 [17:16:58<36:09, 1.73s/it] 96%|█████████▌| 28749/30000 [17:16:59<32:51, 1.58s/it] 96%|█████████▌| 28750/30000 [17:17:01<30:23, 1.46s/it] 96%|█████████▌| 28750/30000 [17:17:01<30:23, 1.46s/it] 96%|█████████▌| 28751/30000 [17:17:02<28:42, 1.38s/it] 96%|█████████▌| 28752/30000 [17:17:03<27:27, 1.32s/it] 96%|█████████▌| 28753/30000 [17:17:04<26:35, 1.28s/it] 96%|█████████▌| 28754/30000 [17:17:08<44:20, 2.13s/it] 96%|█████████▌| 28755/30000 [17:17:09<38:20, 1.85s/it] 96%|█████████▌| 28756/30000 [17:17:11<34:14, 1.65s/it] 96%|█████████▌| 28757/30000 [17:17:12<31:07, 1.50s/it] 96%|█████████▌| 28758/30000 [17:17:13<29:12, 1.41s/it] 96%|█████████▌| 28759/30000 [17:17:14<27:42, 1.34s/it] 96%|█████████▌| 28760/30000 [17:17:15<26:37, 1.29s/it] 96%|█████████▌| 28760/30000 [17:17:15<26:37, 1.29s/it] 96%|█████████▌| 28761/30000 [17:17:17<26:00, 1.26s/it] 96%|█████████▌| 28762/30000 [17:17:21<44:56, 2.18s/it] 96%|█████████▌| 28763/30000 [17:17:22<38:37, 1.87s/it] 96%|█████████▌| 28764/30000 [17:17:23<34:13, 1.66s/it] 96%|█████████▌| 28765/30000 [17:17:24<31:15, 1.52s/it] 96%|█████████▌| 28766/30000 [17:17:26<29:05, 1.41s/it] 96%|█████████▌| 28767/30000 [17:17:27<27:42, 1.35s/it] 96%|█████████▌| 28768/30000 [17:17:28<26:35, 1.29s/it] 96%|█████████▌| 28769/30000 [17:17:29<25:51, 1.26s/it] 96%|█████████▌| 28770/30000 [17:17:33<44:59, 2.19s/it] 96%|█████████▌| 28770/30000 [17:17:33<44:59, 2.19s/it] 96%|█████████▌| 28771/30000 [17:17:35<38:43, 1.89s/it] 96%|█████████▌| 28772/30000 [17:17:36<34:28, 1.68s/it] 96%|█████████▌| 28773/30000 [17:17:37<31:25, 1.54s/it] 96%|█████████▌| 28774/30000 [17:17:38<29:15, 1.43s/it] 96%|█████████▌| 28775/30000 [17:17:39<27:41, 1.36s/it] 96%|█████████▌| 28776/30000 [17:17:41<26:33, 1.30s/it] 96%|█████████▌| 28777/30000 [17:17:42<25:48, 1.27s/it] 96%|█████████▌| 28778/30000 [17:17:46<45:56, 2.26s/it] 96%|█████████▌| 28779/30000 [17:17:47<39:17, 1.93s/it] 96%|█████████▌| 28780/30000 [17:17:49<34:48, 1.71s/it] 96%|█████████▌| 28780/30000 [17:17:49<34:48, 1.71s/it] 96%|█████████▌| 28781/30000 [17:17:50<31:42, 1.56s/it] 96%|█████████▌| 28782/30000 [17:17:51<29:36, 1.46s/it] 96%|█████████▌| 28783/30000 [17:17:52<27:55, 1.38s/it] 96%|█████████▌| 28784/30000 [17:17:53<26:41, 1.32s/it] 96%|█████████▌| 28785/30000 [17:17:55<25:49, 1.28s/it] 96%|█████████▌| 28786/30000 [17:17:59<45:09, 2.23s/it] 96%|█████████▌| 28787/30000 [17:18:00<38:36, 1.91s/it] 96%|█████████▌| 28788/30000 [17:18:01<34:11, 1.69s/it] 96%|█████████▌| 28789/30000 [17:18:03<31:03, 1.54s/it] 96%|█████████▌| 28790/30000 [17:18:04<28:56, 1.44s/it] 96%|█████████▌| 28790/30000 [17:18:04<28:56, 1.44s/it] 96%|█████████▌| 28791/30000 [17:18:05<27:35, 1.37s/it] 96%|█████████▌| 28792/30000 [17:18:06<26:29, 1.32s/it] 96%|█████████▌| 28793/30000 [17:18:07<25:37, 1.27s/it] 96%|█████████▌| 28794/30000 [17:18:12<45:47, 2.28s/it] 96%|█████████▌| 28795/30000 [17:18:13<38:56, 1.94s/it] 96%|█████████▌| 28796/30000 [17:18:14<34:21, 1.71s/it] 96%|█████████▌| 28797/30000 [17:18:16<31:12, 1.56s/it] 96%|█████████▌| 28798/30000 [17:18:17<28:55, 1.44s/it] 96%|█████████▌| 28799/30000 [17:18:18<29:27, 1.47s/it] 96%|█████████▌| 28800/30000 [17:18:19<27:40, 1.38s/it] 96%|█████████▌| 28800/30000 [17:18:19<27:40, 1.38s/it] 96%|█████████▌| 28801/30000 [17:18:21<26:25, 1.32s/it] 96%|█████████▌| 28802/30000 [17:18:26<48:06, 2.41s/it] 96%|█████████▌| 28803/30000 [17:18:27<40:41, 2.04s/it] 96%|█████████▌| 28804/30000 [17:18:28<35:34, 1.79s/it] 96%|█████████▌| 28805/30000 [17:18:29<31:52, 1.60s/it] 96%|█████████▌| 28806/30000 [17:18:30<29:22, 1.48s/it] 96%|█████████▌| 28807/30000 [17:18:31<27:42, 1.39s/it] 96%|█████████▌| 28808/30000 [17:18:33<26:21, 1.33s/it] 96%|█████████▌| 28809/30000 [17:18:34<25:34, 1.29s/it] 96%|█████████▌| 28810/30000 [17:18:38<42:10, 2.13s/it] 96%|█████████▌| 28810/30000 [17:18:38<42:10, 2.13s/it] 96%|█████████▌| 28811/30000 [17:18:39<36:27, 1.84s/it] 96%|█████████▌| 28812/30000 [17:18:40<32:24, 1.64s/it] 96%|█████████▌| 28813/30000 [17:18:41<29:45, 1.50s/it] 96%|█████████▌| 28814/30000 [17:18:43<27:45, 1.40s/it] 96%|█████████▌| 28815/30000 [17:18:44<26:42, 1.35s/it] 96%|█████████▌| 28816/30000 [17:18:45<25:32, 1.29s/it] 96%|█████████▌| 28817/30000 [17:18:46<24:55, 1.26s/it] 96%|█████████▌| 28818/30000 [17:18:51<44:49, 2.28s/it] 96%|█████████▌| 28819/30000 [17:18:52<38:16, 1.94s/it] 96%|█████████▌| 28820/30000 [17:18:53<33:45, 1.72s/it] 96%|█████████▌| 28820/30000 [17:18:53<33:45, 1.72s/it] 96%|█████████▌| 28821/30000 [17:18:54<30:36, 1.56s/it] 96%|█████████▌| 28822/30000 [17:18:56<28:21, 1.44s/it] 96%|█████████▌| 28823/30000 [17:18:57<26:54, 1.37s/it] 96%|█████████▌| 28824/30000 [17:18:58<25:47, 1.32s/it] 96%|█████████▌| 28825/30000 [17:18:59<24:57, 1.27s/it] 96%|█████████▌| 28826/30000 [17:19:03<41:40, 2.13s/it] 96%|█████████▌| 28827/30000 [17:19:04<36:00, 1.84s/it] 96%|█████████▌| 28828/30000 [17:19:06<32:07, 1.64s/it] 96%|█████████▌| 28829/30000 [17:19:07<29:24, 1.51s/it] 96%|█████████▌| 28830/30000 [17:19:08<27:32, 1.41s/it] 96%|█████████▌| 28830/30000 [17:19:08<27:32, 1.41s/it] 96%|█████████▌| 28831/30000 [17:19:09<26:23, 1.35s/it] 96%|█████████▌| 28832/30000 [17:19:10<25:26, 1.31s/it] 96%|█████████▌| 28833/30000 [17:19:12<24:46, 1.27s/it] 96%|█████████▌| 28834/30000 [17:19:16<43:07, 2.22s/it] 96%|█████████▌| 28835/30000 [17:19:17<37:03, 1.91s/it] 96%|█████████▌| 28836/30000 [17:19:18<32:43, 1.69s/it] 96%|█████████▌| 28837/30000 [17:19:20<29:45, 1.54s/it] 96%|█████████▌| 28838/30000 [17:19:21<27:49, 1.44s/it] 96%|█████████▌| 28839/30000 [17:19:22<26:36, 1.37s/it] 96%|█████████▌| 28840/30000 [17:19:23<25:40, 1.33s/it] 96%|█████████▌| 28840/30000 [17:19:23<25:40, 1.33s/it] 96%|█████████▌| 28841/30000 [17:19:24<25:02, 1.30s/it] 96%|█████████▌| 28842/30000 [17:19:29<43:19, 2.24s/it] 96%|█████████▌| 28843/30000 [17:19:30<36:59, 1.92s/it] 96%|█████████▌| 28844/30000 [17:19:31<32:53, 1.71s/it] 96%|█████████▌| 28845/30000 [17:19:32<29:46, 1.55s/it] 96%|█████████▌| 28846/30000 [17:19:34<27:43, 1.44s/it] 96%|█████████▌| 28847/30000 [17:19:35<26:17, 1.37s/it] 96%|█████████▌| 28848/30000 [17:19:36<25:14, 1.31s/it] 96%|█████████▌| 28849/30000 [17:19:37<24:25, 1.27s/it] 96%|█████████▌| 28850/30000 [17:19:42<45:25, 2.37s/it] 96%|█████████▌| 28850/30000 [17:19:42<45:25, 2.37s/it] 96%|█████████▌| 28851/30000 [17:19:43<38:32, 2.01s/it] 96%|█████████▌| 28852/30000 [17:19:45<33:45, 1.76s/it] 96%|█████████▌| 28853/30000 [17:19:46<30:22, 1.59s/it] 96%|█████████▌| 28854/30000 [17:19:47<27:53, 1.46s/it] 96%|█████████▌| 28855/30000 [17:19:48<26:19, 1.38s/it] 96%|█████████▌| 28856/30000 [17:19:49<25:11, 1.32s/it] 96%|█████████▌| 28857/30000 [17:19:50<24:20, 1.28s/it] 96%|█████████▌| 28858/30000 [17:19:55<41:30, 2.18s/it] 96%|█████████▌| 28859/30000 [17:19:56<35:36, 1.87s/it] 96%|█████████▌| 28860/30000 [17:19:57<31:40, 1.67s/it] 96%|█████████▌| 28860/30000 [17:19:57<31:40, 1.67s/it] 96%|█████████▌| 28861/30000 [17:19:58<28:53, 1.52s/it] 96%|█████████▌| 28862/30000 [17:19:59<26:56, 1.42s/it] 96%|█████████▌| 28863/30000 [17:20:01<25:55, 1.37s/it] 96%|█████████▌| 28864/30000 [17:20:02<24:43, 1.31s/it] 96%|█████████▌| 28865/30000 [17:20:03<23:56, 1.27s/it] 96%|█████████▌| 28866/30000 [17:20:07<38:22, 2.03s/it] 96%|█████████▌| 28867/30000 [17:20:08<33:25, 1.77s/it] 96%|█████████▌| 28868/30000 [17:20:09<30:07, 1.60s/it] 96%|█████████▌| 28869/30000 [17:20:10<27:41, 1.47s/it] 96%|█████████▌| 28870/30000 [17:20:12<26:05, 1.39s/it] 96%|█████████▌| 28870/30000 [17:20:12<26:05, 1.39s/it] 96%|█████████▌| 28871/30000 [17:20:13<24:54, 1.32s/it] 96%|█████████▌| 28872/30000 [17:20:14<24:02, 1.28s/it] 96%|█████████▌| 28873/30000 [17:20:15<23:29, 1.25s/it] 96%|█████████▌| 28874/30000 [17:20:19<40:27, 2.16s/it] 96%|█████████▋| 28875/30000 [17:20:20<34:49, 1.86s/it] 96%|█████████▋| 28876/30000 [17:20:22<30:56, 1.65s/it] 96%|█████████▋| 28877/30000 [17:20:23<28:21, 1.52s/it] 96%|█████████▋| 28878/30000 [17:20:24<26:34, 1.42s/it] 96%|█████████▋| 28879/30000 [17:20:25<25:13, 1.35s/it] 96%|█████████▋| 28880/30000 [17:20:26<24:16, 1.30s/it] 96%|█████████▋| 28880/30000 [17:20:26<24:16, 1.30s/it] 96%|█████████▋| 28881/30000 [17:20:28<23:37, 1.27s/it] 96%|█████████▋| 28882/30000 [17:20:32<39:22, 2.11s/it] 96%|█████████▋| 28883/30000 [17:20:33<34:07, 1.83s/it] 96%|█████████▋| 28884/30000 [17:20:34<30:25, 1.64s/it] 96%|█████████▋| 28885/30000 [17:20:35<27:52, 1.50s/it] 96%|█████████▋| 28886/30000 [17:20:36<26:01, 1.40s/it] 96%|█████████▋| 28887/30000 [17:20:38<24:46, 1.34s/it] 96%|█████████▋| 28888/30000 [17:20:39<23:47, 1.28s/it] 96%|█████████▋| 28889/30000 [17:20:40<23:15, 1.26s/it] 96%|█████████▋| 28890/30000 [17:20:44<38:23, 2.08s/it] 96%|█████████▋| 28890/30000 [17:20:44<38:23, 2.08s/it] 96%|█████████▋| 28891/30000 [17:20:45<33:24, 1.81s/it] 96%|█████████▋| 28892/30000 [17:20:46<30:02, 1.63s/it] 96%|█████████▋| 28893/30000 [17:20:47<27:31, 1.49s/it] 96%|█████████▋| 28894/30000 [17:20:49<25:52, 1.40s/it] 96%|█████████▋| 28895/30000 [17:20:50<24:39, 1.34s/it] 96%|█████████▋| 28896/30000 [17:20:51<23:41, 1.29s/it] 96%|█████████▋| 28897/30000 [17:20:52<23:09, 1.26s/it] 96%|█████████▋| 28898/30000 [17:20:56<37:49, 2.06s/it] 96%|█████████▋| 28899/30000 [17:20:57<32:58, 1.80s/it] 96%|█████████▋| 28900/30000 [17:20:59<29:38, 1.62s/it] 96%|█████████▋| 28900/30000 [17:20:59<29:38, 1.62s/it] 96%|█████████▋| 28901/30000 [17:21:00<27:14, 1.49s/it] 96%|█████████▋| 28902/30000 [17:21:01<25:35, 1.40s/it] 96%|█████████▋| 28903/30000 [17:21:02<24:59, 1.37s/it] 96%|█████████▋| 28904/30000 [17:21:03<23:51, 1.31s/it] 96%|█████████▋| 28905/30000 [17:21:05<23:11, 1.27s/it] 96%|█████████▋| 28906/30000 [17:21:09<40:12, 2.21s/it] 96%|█████████▋| 28907/30000 [17:21:10<34:33, 1.90s/it] 96%|█████████▋| 28908/30000 [17:21:11<30:38, 1.68s/it] 96%|█████████▋| 28909/30000 [17:21:12<27:45, 1.53s/it] 96%|█████████▋| 28910/30000 [17:21:14<25:47, 1.42s/it] 96%|█████████▋| 28910/30000 [17:21:14<25:47, 1.42s/it] 96%|█████████▋| 28911/30000 [17:21:15<24:33, 1.35s/it] 96%|█████████▋| 28912/30000 [17:21:16<23:36, 1.30s/it] 96%|█████████▋| 28913/30000 [17:21:17<22:55, 1.27s/it] 96%|█████████▋| 28914/30000 [17:21:21<38:22, 2.12s/it] 96%|█████████▋| 28915/30000 [17:21:23<33:21, 1.84s/it] 96%|█████████▋| 28916/30000 [17:21:24<29:57, 1.66s/it] 96%|█████████▋| 28917/30000 [17:21:25<27:28, 1.52s/it] 96%|█████████▋| 28918/30000 [17:21:26<25:40, 1.42s/it] 96%|█████████▋| 28919/30000 [17:21:28<27:36, 1.53s/it] 96%|█████████▋| 28920/30000 [17:21:29<25:45, 1.43s/it] 96%|█████████▋| 28920/30000 [17:21:29<25:45, 1.43s/it] 96%|█████████▋| 28921/30000 [17:21:30<24:30, 1.36s/it] 96%|█████████▋| 28922/30000 [17:21:34<37:11, 2.07s/it] 96%|█████████▋| 28923/30000 [17:21:35<32:26, 1.81s/it] 96%|█████████▋| 28924/30000 [17:21:36<29:03, 1.62s/it] 96%|█████████▋| 28925/30000 [17:21:38<26:54, 1.50s/it] 96%|█████████▋| 28926/30000 [17:21:39<25:22, 1.42s/it] 96%|█████████▋| 28927/30000 [17:21:41<28:37, 1.60s/it] 96%|█████████▋| 28928/30000 [17:21:42<26:37, 1.49s/it] 96%|█████████▋| 28929/30000 [17:21:43<25:05, 1.41s/it] 96%|█████████▋| 28930/30000 [17:21:46<33:03, 1.85s/it] 96%|█████████▋| 28930/30000 [17:21:46<33:03, 1.85s/it] 96%|█████████▋| 28931/30000 [17:21:47<29:25, 1.65s/it] 96%|█████████▋| 28932/30000 [17:21:49<26:57, 1.51s/it] 96%|█████████▋| 28933/30000 [17:21:50<25:17, 1.42s/it] 96%|█████████▋| 28934/30000 [17:21:51<24:02, 1.35s/it] 96%|█████████▋| 28935/30000 [17:21:54<33:33, 1.89s/it] 96%|█████████▋| 28936/30000 [17:21:55<29:46, 1.68s/it] 96%|█████████▋| 28937/30000 [17:21:57<27:06, 1.53s/it] 96%|█████████▋| 28938/30000 [17:21:59<32:53, 1.86s/it] 96%|█████████▋| 28939/30000 [17:22:00<29:13, 1.65s/it] 96%|█████████▋| 28940/30000 [17:22:02<26:49, 1.52s/it] 96%|█████████▋| 28940/30000 [17:22:02<26:49, 1.52s/it] 96%|█████████▋| 28941/30000 [17:22:03<25:04, 1.42s/it] 96%|█████████▋| 28942/30000 [17:22:04<23:49, 1.35s/it] 96%|█████████▋| 28943/30000 [17:22:07<31:31, 1.79s/it] 96%|█████████▋| 28944/30000 [17:22:08<28:22, 1.61s/it] 96%|█████████▋| 28945/30000 [17:22:09<26:10, 1.49s/it] 96%|█████████▋| 28946/30000 [17:22:11<30:14, 1.72s/it] 96%|█████████▋| 28947/30000 [17:22:13<27:27, 1.56s/it] 96%|█████████▋| 28948/30000 [17:22:14<25:28, 1.45s/it] 96%|█████████▋| 28949/30000 [17:22:15<23:58, 1.37s/it] 96%|█████████▋| 28950/30000 [17:22:16<23:00, 1.31s/it] 96%|█████████▋| 28950/30000 [17:22:16<23:00, 1.31s/it] 97%|█████████▋| 28951/30000 [17:22:19<32:14, 1.84s/it] 97%|█████████▋| 28952/30000 [17:22:20<28:46, 1.65s/it] 97%|█████████▋| 28953/30000 [17:22:22<26:22, 1.51s/it] 97%|█████████▋| 28954/30000 [17:22:24<32:33, 1.87s/it] 97%|█████████▋| 28955/30000 [17:22:25<28:52, 1.66s/it] 97%|█████████▋| 28956/30000 [17:22:27<26:25, 1.52s/it] 97%|█████████▋| 28957/30000 [17:22:28<24:45, 1.42s/it] 97%|█████████▋| 28958/30000 [17:22:29<23:46, 1.37s/it] 97%|█████████▋| 28959/30000 [17:22:33<36:41, 2.12s/it] 97%|█████████▋| 28960/30000 [17:22:34<31:44, 1.83s/it] 97%|█████████▋| 28960/30000 [17:22:34<31:44, 1.83s/it] 97%|█████████▋| 28961/30000 [17:22:35<28:27, 1.64s/it] 97%|█████████▋| 28962/30000 [17:22:37<29:34, 1.71s/it] 97%|█████████▋| 28963/30000 [17:22:38<26:44, 1.55s/it] 97%|█████████▋| 28964/30000 [17:22:40<24:53, 1.44s/it] 97%|█████████▋| 28965/30000 [17:22:41<23:33, 1.37s/it] 97%|█████████▋| 28966/30000 [17:22:42<22:42, 1.32s/it] 97%|█████████▋| 28967/30000 [17:22:46<34:39, 2.01s/it] 97%|█████████▋| 28968/30000 [17:22:47<30:18, 1.76s/it] 97%|█████████▋| 28969/30000 [17:22:48<27:23, 1.59s/it] 97%|█████████▋| 28970/30000 [17:22:50<29:23, 1.71s/it] +{'loss': 0.0129, 'grad_norm': 0.05490626394748688, 'learning_rate': 1.862300893213903e-07, 'flow_matching_loss': 0.012883279100060463, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.31} +{'loss': 0.012, 'grad_norm': 0.09323376417160034, 'learning_rate': 1.8364169260238595e-07, 'flow_matching_loss': 0.011970263067632913, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.32} +{'loss': 0.0136, 'grad_norm': 0.06173180788755417, 'learning_rate': 1.8107129915581777e-07, 'flow_matching_loss': 0.01355331065133214, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.32} +{'loss': 0.0122, 'grad_norm': 0.07750650495290756, 'learning_rate': 1.785189121049541e-07, 'flow_matching_loss': 0.012189502408728004, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.32} +{'loss': 0.0127, 'grad_norm': 0.07237403839826584, 'learning_rate': 1.7598453455118567e-07, 'flow_matching_loss': 0.01266820365563035, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.32} +{'loss': 0.0143, 'grad_norm': 0.06966681778430939, 'learning_rate': 1.7346816957402256e-07, 'flow_matching_loss': 0.01433944869786501, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.33} +{'loss': 0.0122, 'grad_norm': 0.07230470329523087, 'learning_rate': 1.7096982023108389e-07, 'flow_matching_loss': 0.012249280232936144, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.33} +{'loss': 0.0127, 'grad_norm': 0.057123810052871704, 'learning_rate': 1.6848948955809963e-07, 'flow_matching_loss': 0.012726344633847475, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.33} +{'loss': 0.0132, 'grad_norm': 0.07276846468448639, 'learning_rate': 1.66027180568904e-07, 'flow_matching_loss': 0.013213436212390661, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.34} +{'loss': 0.0132, 'grad_norm': 0.0581718347966671, 'learning_rate': 1.6358289625543865e-07, 'flow_matching_loss': 0.013235205598175526, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.34} +{'loss': 0.0127, 'grad_norm': 0.05975045636296272, 'learning_rate': 1.6115663958773618e-07, 'flow_matching_loss': 0.01274622669443488, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.34} +{'loss': 0.0135, 'grad_norm': 0.07526625692844391, 'learning_rate': 1.5874841351392832e-07, 'flow_matching_loss': 0.013512833695858716, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.34} +{'loss': 0.0155, 'grad_norm': 0.09914402663707733, 'learning_rate': 1.56358220960236e-07, 'flow_matching_loss': 0.015495777316391467, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.35} +{'loss': 0.0124, 'grad_norm': 0.07004306465387344, 'learning_rate': 1.5398606483096932e-07, 'flow_matching_loss': 0.012382709234952927, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.35} +{'loss': 0.0144, 'grad_norm': 0.11182279139757156, 'learning_rate': 1.5163194800852265e-07, 'flow_matching_loss': 0.01441841321066022, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.35} +{'loss': 0.0146, 'grad_norm': 0.09309599548578262, 'learning_rate': 1.4929587335336614e-07, 'flow_matching_loss': 0.014612728357315063, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.36} +{'loss': 0.0128, 'grad_norm': 0.07489442080259323, 'learning_rate': 1.4697784370405587e-07, 'flow_matching_loss': 0.012807410582900048, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.36} +{'loss': 0.0127, 'grad_norm': 0.07617061585187912, 'learning_rate': 1.4467786187721376e-07, 'flow_matching_loss': 0.012662817630916834, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.36} +{'loss': 0.0126, 'grad_norm': 0.06837426871061325, 'learning_rate': 1.4239593066753265e-07, 'flow_matching_loss': 0.012615935876965523, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.36} +{'loss': 0.0121, 'grad_norm': 0.06263464689254761, 'learning_rate': 1.4013205284777953e-07, 'flow_matching_loss': 0.012145032966509461, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.37} +{'loss': 0.0129, 'grad_norm': 0.07096876949071884, 'learning_rate': 1.3788623116877397e-07, 'flow_matching_loss': 0.012858329527080058, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.37} +{'loss': 0.0127, 'grad_norm': 0.06653342396020889, 'learning_rate': 1.3565846835940477e-07, 'flow_matching_loss': 0.012683265656232835, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.37} +{'loss': 0.013, 'grad_norm': 0.09980694204568863, 'learning_rate': 1.334487671266116e-07, 'flow_matching_loss': 0.013042287994176149, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.38} +{'loss': 0.0121, 'grad_norm': 0.07713396847248077, 'learning_rate': 1.3125713015538998e-07, 'flow_matching_loss': 0.012066153204068541, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.38} +{'loss': 0.0128, 'grad_norm': 0.07240042835474014, 'learning_rate': 1.290835601087864e-07, 'flow_matching_loss': 0.0128202011808753, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.38} +{'loss': 0.0122, 'grad_norm': 0.06319954991340637, 'learning_rate': 1.2692805962789323e-07, 'flow_matching_loss': 0.012177428510040045, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.39} +{'loss': 0.0124, 'grad_norm': 0.05570422112941742, 'learning_rate': 1.2479063133184698e-07, 'flow_matching_loss': 0.012377765867859125, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.39} +{'loss': 0.0121, 'grad_norm': 0.07644127309322357, 'learning_rate': 1.2267127781782516e-07, 'flow_matching_loss': 0.012061446346342564, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.39} +{'loss': 0.013, 'grad_norm': 0.08634291589260101, 'learning_rate': 1.205700016610395e-07, 'flow_matching_loss': 0.013047231175005437, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.39} +{'loss': 0.0126, 'grad_norm': 0.06040705367922783, 'learning_rate': 1.1848680541474088e-07, 'flow_matching_loss': 0.012642847094684839, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.4} +{'loss': 0.0126, 'grad_norm': 0.06766446679830551, 'learning_rate': 1.1642169161020955e-07, 'flow_matching_loss': 0.012642869073897601, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.4} +{'loss': 0.0125, 'grad_norm': 0.08120467513799667, 'learning_rate': 1.1437466275675156e-07, 'flow_matching_loss': 0.012465938832610846, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.4} +{'loss': 0.0135, 'grad_norm': 0.05903855711221695, 'learning_rate': 1.1234572134170229e-07, 'flow_matching_loss': 0.013501220662146807, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.41} +{'loss': 0.0121, 'grad_norm': 0.06431137025356293, 'learning_rate': 1.1033486983041463e-07, 'flow_matching_loss': 0.012055949214845896, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.41} +{'loss': 0.0128, 'grad_norm': 0.06793515384197235, 'learning_rate': 1.0834211066626409e-07, 'flow_matching_loss': 0.012819670606404543, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.41} +{'loss': 0.0118, 'grad_norm': 0.06529086083173752, 'learning_rate': 1.0636744627064044e-07, 'flow_matching_loss': 0.011801788024604321, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.41} +{'loss': 0.014, 'grad_norm': 0.05944078043103218, 'learning_rate': 1.0441087904294766e-07, 'flow_matching_loss': 0.013998137786984444, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.42} +{'loss': 0.0114, 'grad_norm': 0.05693169683218002, 'learning_rate': 1.0247241136059904e-07, 'flow_matching_loss': 0.011432184465229511, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.42} +{'loss': 0.0129, 'grad_norm': 0.07386791706085205, 'learning_rate': 1.0055204557901709e-07, 'flow_matching_loss': 0.012852784991264344, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.42} +{'loss': 0.0127, 'grad_norm': 0.071060411632061, 'learning_rate': 9.864978403162361e-08, 'flow_matching_loss': 0.012738364934921264, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.43} +{'loss': 0.0131, 'grad_norm': 0.06994062662124634, 'learning_rate': 9.676562902984965e-08, 'flow_matching_loss': 0.013066381588578225, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.43} 97%|█████████▋| 28970/30000 [17:22:50<29:23, 1.71s/it] 97%|█████████▋| 28971/30000 [17:22:51<26:31, 1.55s/it] 97%|█████████▋| 28972/30000 [17:22:52<24:37, 1.44s/it] 97%|█████████▋| 28973/30000 [17:22:53<23:13, 1.36s/it] 97%|█████████▋| 28974/30000 [17:22:55<22:21, 1.31s/it] 97%|█████████▋| 28975/30000 [17:22:58<33:27, 1.96s/it] 97%|█████████▋| 28976/30000 [17:22:59<29:24, 1.72s/it] 97%|█████████▋| 28977/30000 [17:23:01<26:45, 1.57s/it] 97%|█████████▋| 28978/30000 [17:23:03<28:47, 1.69s/it] 97%|█████████▋| 28979/30000 [17:23:04<26:21, 1.55s/it] 97%|█████████▋| 28980/30000 [17:23:05<24:35, 1.45s/it] 97%|█████████▋| 28980/30000 [17:23:05<24:35, 1.45s/it] 97%|█████████▋| 28981/30000 [17:23:06<23:14, 1.37s/it] 97%|█████████▋| 28982/30000 [17:23:07<22:13, 1.31s/it] 97%|█████████▋| 28983/30000 [17:23:10<31:43, 1.87s/it] 97%|█████████▋| 28984/30000 [17:23:12<28:05, 1.66s/it] 97%|█████████▋| 28985/30000 [17:23:13<25:39, 1.52s/it] 97%|█████████▋| 28986/30000 [17:23:15<28:19, 1.68s/it] 97%|█████████▋| 28987/30000 [17:23:16<25:42, 1.52s/it] 97%|█████████▋| 28988/30000 [17:23:17<23:55, 1.42s/it] 97%|█████████▋| 28989/30000 [17:23:18<23:01, 1.37s/it] 97%|█████████▋| 28990/30000 [17:23:20<22:03, 1.31s/it] 97%|█████████▋| 28990/30000 [17:23:20<22:03, 1.31s/it] 97%|█████████▋| 28991/30000 [17:23:26<48:05, 2.86s/it] 97%|█████████▋| 28992/30000 [17:23:27<39:30, 2.35s/it] 97%|█████████▋| 28993/30000 [17:23:28<33:31, 2.00s/it] 97%|█████████▋| 28994/30000 [17:23:30<29:29, 1.76s/it] 97%|█████████▋| 28995/30000 [17:23:31<26:26, 1.58s/it] 97%|█████████▋| 28996/30000 [17:23:32<24:22, 1.46s/it] 97%|█████████▋| 28997/30000 [17:23:33<22:56, 1.37s/it] 97%|█████████▋| 28998/30000 [17:23:34<21:49, 1.31s/it] 97%|█████████▋| 28999/30000 [17:23:39<38:44, 2.32s/it] 97%|█████████▋| 29000/30000 [17:23:40<33:02, 1.98s/it] 97%|█████████▋| 29000/30000 [17:23:40<33:02, 1.98s/it] 97%|█████████▋| 29001/30000 [17:23:41<29:05, 1.75s/it] 97%|█████████▋| 29002/30000 [17:23:43<26:12, 1.58s/it] 97%|█████████▋| 29003/30000 [17:23:44<24:12, 1.46s/it] 97%|█████████▋| 29004/30000 [17:23:45<22:49, 1.38s/it] 97%|█████████▋| 29005/30000 [17:23:46<22:00, 1.33s/it] 97%|█████████▋| 29006/30000 [17:23:47<21:22, 1.29s/it] 97%|█████████▋| 29007/30000 [17:23:52<36:01, 2.18s/it] 97%|█████████▋| 29008/30000 [17:23:53<31:06, 1.88s/it] 97%|█████████▋| 29009/30000 [17:23:54<27:43, 1.68s/it] 97%|█████████▋| 29010/30000 [17:23:55<25:30, 1.55s/it] 97%|█████████▋| 29010/30000 [17:23:55<25:30, 1.55s/it] 97%|█████████▋| 29011/30000 [17:23:56<24:05, 1.46s/it] 97%|█████████▋| 29012/30000 [17:23:58<22:42, 1.38s/it] 97%|█████████▋| 29013/30000 [17:23:59<21:57, 1.34s/it] 97%|█████████▋| 29014/30000 [17:24:00<21:59, 1.34s/it] 97%|█████████▋| 29015/30000 [17:24:04<32:17, 1.97s/it] 97%|█████████▋| 29016/30000 [17:24:05<28:23, 1.73s/it] 97%|█████████▋| 29017/30000 [17:24:06<25:40, 1.57s/it] 97%|█████████▋| 29018/30000 [17:24:07<23:52, 1.46s/it] 97%|█████████▋| 29019/30000 [17:24:08<22:38, 1.38s/it] 97%|█████████▋| 29020/30000 [17:24:10<21:46, 1.33s/it] 97%|█████████▋| 29020/30000 [17:24:10<21:46, 1.33s/it] 97%|█████████▋| 29021/30000 [17:24:11<21:01, 1.29s/it] 97%|█████████▋| 29022/30000 [17:24:12<20:28, 1.26s/it] 97%|█████████▋| 29023/30000 [17:24:17<36:37, 2.25s/it] 97%|█████████▋| 29024/30000 [17:24:18<31:16, 1.92s/it] 97%|█████████▋| 29025/30000 [17:24:19<27:42, 1.71s/it] 97%|█████████▋| 29026/30000 [17:24:20<25:06, 1.55s/it] 97%|█████████▋| 29027/30000 [17:24:21<23:15, 1.43s/it] 97%|█████████▋| 29028/30000 [17:24:23<21:58, 1.36s/it] 97%|█████████▋| 29029/30000 [17:24:24<21:06, 1.30s/it] 97%|█████████▋| 29030/30000 [17:24:25<20:25, 1.26s/it] 97%|█████████▋| 29030/30000 [17:24:25<20:25, 1.26s/it] 97%|█████████▋| 29031/30000 [17:24:30<38:31, 2.38s/it] 97%|█████████▋| 29032/30000 [17:24:31<32:34, 2.02s/it] 97%|█████████▋| 29033/30000 [17:24:32<28:26, 1.76s/it] 97%|█████████▋| 29034/30000 [17:24:33<25:35, 1.59s/it] 97%|█████████▋| 29035/30000 [17:24:35<23:29, 1.46s/it] 97%|█████████▋| 29036/30000 [17:24:36<22:07, 1.38s/it] 97%|█████████▋| 29037/30000 [17:24:37<21:02, 1.31s/it] 97%|█████████▋| 29038/30000 [17:24:38<20:20, 1.27s/it] 97%|█████████▋| 29039/30000 [17:24:43<36:27, 2.28s/it] 97%|█████████▋| 29040/30000 [17:24:44<31:19, 1.96s/it] 97%|█████████▋| 29040/30000 [17:24:44<31:19, 1.96s/it] 97%|█████████▋| 29041/30000 [17:24:45<27:52, 1.74s/it] 97%|█████████▋| 29042/30000 [17:24:46<25:06, 1.57s/it] 97%|█████████▋| 29043/30000 [17:24:47<23:15, 1.46s/it] 97%|█████████▋| 29044/30000 [17:24:49<22:26, 1.41s/it] 97%|█████████▋| 29045/30000 [17:24:50<21:23, 1.34s/it] 97%|█████████▋| 29046/30000 [17:24:51<20:35, 1.29s/it] 97%|█████████▋| 29047/30000 [17:24:55<33:32, 2.11s/it] 97%|█████████▋| 29048/30000 [17:24:56<29:05, 1.83s/it] 97%|█████████▋| 29049/30000 [17:24:58<25:54, 1.63s/it] 97%|█████████▋| 29050/30000 [17:24:59<23:39, 1.49s/it] 97%|█████████▋| 29050/30000 [17:24:59<23:39, 1.49s/it] 97%|█████████▋| 29051/30000 [17:25:00<22:06, 1.40s/it] 97%|█████████▋| 29052/30000 [17:25:01<21:04, 1.33s/it] 97%|█████████▋| 29053/30000 [17:25:02<20:18, 1.29s/it] 97%|█████████▋| 29054/30000 [17:25:03<19:51, 1.26s/it] 97%|█████████▋| 29055/30000 [17:25:08<34:30, 2.19s/it] 97%|█████████▋| 29056/30000 [17:25:09<29:37, 1.88s/it] 97%|█████████▋| 29057/30000 [17:25:10<26:23, 1.68s/it] 97%|█████████▋| 29058/30000 [17:25:11<23:58, 1.53s/it] 97%|█████████▋| 29059/30000 [17:25:13<22:16, 1.42s/it] 97%|█████████▋| 29060/30000 [17:25:14<21:11, 1.35s/it] 97%|█████████▋| 29060/30000 [17:25:14<21:11, 1.35s/it] 97%|█████████▋| 29061/30000 [17:25:15<20:26, 1.31s/it] 97%|█████████▋| 29062/30000 [17:25:16<19:51, 1.27s/it] 97%|█████████▋| 29063/30000 [17:25:21<36:35, 2.34s/it] 97%|█████████▋| 29064/30000 [17:25:22<31:09, 2.00s/it] 97%|█████████▋| 29065/30000 [17:25:23<27:12, 1.75s/it] 97%|█████████▋| 29066/30000 [17:25:24<24:32, 1.58s/it] 97%|█████████▋| 29067/30000 [17:25:26<22:37, 1.45s/it] 97%|█████████▋| 29068/30000 [17:25:27<21:18, 1.37s/it] 97%|█████████▋| 29069/30000 [17:25:28<20:22, 1.31s/it] 97%|█████████▋| 29070/30000 [17:25:29<19:56, 1.29s/it] 97%|█████████▋| 29070/30000 [17:25:29<19:56, 1.29s/it] 97%|█████████▋| 29071/30000 [17:25:34<35:53, 2.32s/it] 97%|█████████▋| 29072/30000 [17:25:35<30:34, 1.98s/it] 97%|█████████▋| 29073/30000 [17:25:36<26:53, 1.74s/it] 97%|█████████▋| 29074/30000 [17:25:37<24:15, 1.57s/it] 97%|█████████▋| 29075/30000 [17:25:39<22:30, 1.46s/it] 97%|█████████▋| 29076/30000 [17:25:40<21:10, 1.38s/it] 97%|█████████▋| 29077/30000 [17:25:41<20:18, 1.32s/it] 97%|█████████▋| 29078/30000 [17:25:42<19:50, 1.29s/it] 97%|█████████▋| 29079/30000 [17:25:47<34:57, 2.28s/it] 97%|█████████▋| 29080/30000 [17:25:48<29:49, 1.94s/it] 97%|█████████▋| 29080/30000 [17:25:48<29:49, 1.94s/it] 97%|█████████▋| 29081/30000 [17:25:49<26:19, 1.72s/it] 97%|█████████▋| 29082/30000 [17:25:50<23:47, 1.56s/it] 97%|█████████▋| 29083/30000 [17:25:52<21:59, 1.44s/it] 97%|█████████▋| 29084/30000 [17:25:53<20:54, 1.37s/it] 97%|█████████▋| 29085/30000 [17:25:54<19:59, 1.31s/it] 97%|█████████▋| 29086/30000 [17:25:55<19:21, 1.27s/it] 97%|█████████▋| 29087/30000 [17:25:59<32:34, 2.14s/it] 97%|█████████▋| 29088/30000 [17:26:00<28:16, 1.86s/it] 97%|█████████▋| 29089/30000 [17:26:02<25:15, 1.66s/it] 97%|█████████▋| 29090/30000 [17:26:03<23:08, 1.53s/it] 97%|█████████▋| 29090/30000 [17:26:03<23:08, 1.53s/it] 97%|█████████▋| 29091/30000 [17:26:04<21:24, 1.41s/it] 97%|█████████▋| 29092/30000 [17:26:05<20:31, 1.36s/it] 97%|█████████▋| 29093/30000 [17:26:06<19:43, 1.31s/it] 97%|█████████▋| 29094/30000 [17:26:08<19:03, 1.26s/it] 97%|█████████▋| 29095/30000 [17:26:12<32:01, 2.12s/it] 97%|█████████▋| 29096/30000 [17:26:13<27:39, 1.84s/it] 97%|█████████▋| 29097/30000 [17:26:14<24:43, 1.64s/it] 97%|█████████▋| 29098/30000 [17:26:15<22:37, 1.51s/it] 97%|█████████▋| 29099/30000 [17:26:16<21:03, 1.40s/it] 97%|█████████▋| 29100/30000 [17:26:18<20:09, 1.34s/it] 97%|█████████▋| 29100/30000 [17:26:18<20:09, 1.34s/it] 97%|█████████▋| 29101/30000 [17:26:19<19:48, 1.32s/it] 97%|█████████▋| 29102/30000 [17:26:20<19:11, 1.28s/it] 97%|█████████▋| 29103/30000 [17:26:25<34:20, 2.30s/it] 97%|█████████▋| 29104/30000 [17:26:26<29:19, 1.96s/it] 97%|█████████▋| 29105/30000 [17:26:27<25:43, 1.72s/it] 97%|█████████▋| 29106/30000 [17:26:28<23:20, 1.57s/it] 97%|█████████▋| 29107/30000 [17:26:30<21:36, 1.45s/it] 97%|█████████▋| 29108/30000 [17:26:31<20:18, 1.37s/it] 97%|█████████▋| 29109/30000 [17:26:32<19:36, 1.32s/it] 97%|█████████▋| 29110/30000 [17:26:33<18:58, 1.28s/it] 97%|█████████▋| 29110/30000 [17:26:33<18:58, 1.28s/it] 97%|█████████▋| 29111/30000 [17:26:38<35:31, 2.40s/it] 97%|█████████▋| 29112/30000 [17:26:39<30:05, 2.03s/it] 97%|█████████▋| 29113/30000 [17:26:40<26:10, 1.77s/it] 97%|█████████▋| 29114/30000 [17:26:42<23:34, 1.60s/it] 97%|█████████▋| 29115/30000 [17:26:43<21:34, 1.46s/it] 97%|█████████▋| 29116/30000 [17:26:44<20:13, 1.37s/it] 97%|█████████▋| 29117/30000 [17:26:45<19:17, 1.31s/it] 97%|█████████▋| 29118/30000 [17:26:46<18:42, 1.27s/it] 97%|█████████▋| 29119/30000 [17:26:51<34:50, 2.37s/it] 97%|█████████▋| 29120/30000 [17:26:52<29:38, 2.02s/it] 97%|█████████▋| 29120/30000 [17:26:52<29:38, 2.02s/it] 97%|█████████▋| 29121/30000 [17:26:54<25:53, 1.77s/it] 97%|█████████▋| 29122/30000 [17:26:55<23:15, 1.59s/it] 97%|█████████▋| 29123/30000 [17:26:56<21:24, 1.46s/it] 97%|█████████▋| 29124/30000 [17:26:57<20:08, 1.38s/it] 97%|█████████▋| 29125/30000 [17:26:58<19:10, 1.31s/it] 97%|█████████▋| 29126/30000 [17:26:59<18:30, 1.27s/it] 97%|█████████▋| 29127/30000 [17:27:05<35:02, 2.41s/it] 97%|█████████▋| 29128/30000 [17:27:06<29:39, 2.04s/it] 97%|█████████▋| 29129/30000 [17:27:07<25:49, 1.78s/it] 97%|█████████▋| 29130/30000 [17:27:08<23:10, 1.60s/it] 97%|█████████▋| 29130/30000 [17:27:08<23:10, 1.60s/it] 97%|█████████▋| 29131/30000 [17:27:09<21:12, 1.46s/it] 97%|█████████▋| 29132/30000 [17:27:10<19:58, 1.38s/it] 97%|█████████▋| 29133/30000 [17:27:12<19:02, 1.32s/it] 97%|█████████▋| 29134/30000 [17:27:13<18:21, 1.27s/it] 97%|█████████▋| 29135/30000 [17:27:18<37:35, 2.61s/it] 97%|█████████▋| 29136/30000 [17:27:20<31:13, 2.17s/it] 97%|█████████▋| 29137/30000 [17:27:21<26:52, 1.87s/it] 97%|█████████▋| 29138/30000 [17:27:22<23:55, 1.66s/it] 97%|█████████▋| 29139/30000 [17:27:23<21:47, 1.52s/it] 97%|█████████▋| 29140/30000 [17:27:24<20:10, 1.41s/it] 97%|█████████▋| 29140/30000 [17:27:24<20:10, 1.41s/it] 97%|█████████▋| 29141/30000 [17:27:25<19:08, 1.34s/it] 97%|█████████▋| 29142/30000 [17:27:27<18:29, 1.29s/it] 97%|█████████▋| 29143/30000 [17:27:32<33:56, 2.38s/it] 97%|█████████▋| 29144/30000 [17:27:33<28:51, 2.02s/it] 97%|█████████▋| 29145/30000 [17:27:34<25:15, 1.77s/it] 97%|█████████▋| 29146/30000 [17:27:35<22:56, 1.61s/it] 97%|█████████▋| 29147/30000 [17:27:36<21:00, 1.48s/it] 97%|█████████▋| 29148/30000 [17:27:38<19:49, 1.40s/it] 97%|█████████▋| 29149/30000 [17:27:39<19:08, 1.35s/it] 97%|█████████▋| 29150/30000 [17:27:40<18:31, 1.31s/it] 97%|█████████▋| 29150/30000 [17:27:40<18:31, 1.31s/it] 97%|█████████▋| 29151/30000 [17:27:44<30:17, 2.14s/it] 97%|█████████▋| 29152/30000 [17:27:45<26:09, 1.85s/it] 97%|█████████▋| 29153/30000 [17:27:46<23:14, 1.65s/it] 97%|█████████▋| 29154/30000 [17:27:48<21:22, 1.52s/it] 97%|█████████▋| 29155/30000 [17:27:49<19:50, 1.41s/it] 97%|█████████▋| 29156/30000 [17:27:50<18:56, 1.35s/it] 97%|█████████▋| 29157/30000 [17:27:51<18:09, 1.29s/it] 97%|█████████▋| 29158/30000 [17:27:52<17:51, 1.27s/it] 97%|█████████▋| 29159/30000 [17:27:57<30:26, 2.17s/it] 97%|█████████▋| 29160/30000 [17:27:58<26:15, 1.88s/it] 97%|█████████▋| 29160/30000 [17:27:58<26:15, 1.88s/it] 97%|█████████▋| 29161/30000 [17:27:59<23:16, 1.66s/it] 97%|█████████▋| 29162/30000 [17:28:00<21:10, 1.52s/it] 97%|█████████▋| 29163/30000 [17:28:01<19:42, 1.41s/it] 97%|█████████▋| 29164/30000 [17:28:03<18:46, 1.35s/it] 97%|█████████▋| 29165/30000 [17:28:04<18:06, 1.30s/it] 97%|█████████▋| 29166/30000 [17:28:05<17:36, 1.27s/it] 97%|█████████▋| 29167/30000 [17:28:09<30:55, 2.23s/it] 97%|█████████▋| 29168/30000 [17:28:11<26:35, 1.92s/it] 97%|█████████▋| 29169/30000 [17:28:12<23:34, 1.70s/it] 97%|█████████▋| 29170/30000 [17:28:13<21:24, 1.55s/it] 97%|█████████▋| 29170/30000 [17:28:13<21:24, 1.55s/it] 97%|█████████▋| 29171/30000 [17:28:14<19:49, 1.44s/it] 97%|█████████▋| 29172/30000 [17:28:15<18:43, 1.36s/it] 97%|█████████▋| 29173/30000 [17:28:17<18:01, 1.31s/it] 97%|█████████▋| 29174/30000 [17:28:18<17:34, 1.28s/it] 97%|█████████▋| 29175/30000 [17:28:22<30:02, 2.19s/it] 97%|█████████▋| 29176/30000 [17:28:23<25:53, 1.89s/it] 97%|█████████▋| 29177/30000 [17:28:24<22:57, 1.67s/it] 97%|█████████▋| 29178/30000 [17:28:26<20:54, 1.53s/it] 97%|█████████▋| 29179/30000 [17:28:27<19:29, 1.42s/it] 97%|█████████▋| 29180/30000 [17:28:28<18:32, 1.36s/it] 97%|█████████▋| 29180/30000 [17:28:28<18:32, 1.36s/it] 97%|█████████▋| 29181/30000 [17:28:29<17:52, 1.31s/it] 97%|█████████▋| 29182/30000 [17:28:30<17:19, 1.27s/it] 97%|█████████▋| 29183/30000 [17:28:35<30:20, 2.23s/it] 97%|█████████▋| 29184/30000 [17:28:36<26:05, 1.92s/it] 97%|█████████▋| 29185/30000 [17:28:37<23:01, 1.69s/it] 97%|█████████▋| 29186/30000 [17:28:38<20:59, 1.55s/it] 97%|█████████▋| 29187/30000 [17:28:40<19:19, 1.43s/it] 97%|█████████▋| 29188/30000 [17:28:41<18:32, 1.37s/it] 97%|█████████▋| 29189/30000 [17:28:42<17:47, 1.32s/it] 97%|█████████▋| 29190/30000 [17:28:43<17:16, 1.28s/it] 97%|█████████▋| 29190/30000 [17:28:43<17:16, 1.28s/it] 97%|█████████▋| 29191/30000 [17:28:48<30:21, 2.25s/it] 97%|█████████▋| 29192/30000 [17:28:49<26:03, 1.94s/it] 97%|█████████▋| 29193/30000 [17:28:50<23:05, 1.72s/it] 97%|█████████▋| 29194/30000 [17:28:51<20:57, 1.56s/it] 97%|█████████▋| 29195/30000 [17:28:52<19:20, 1.44s/it] 97%|█████████▋| 29196/30000 [17:28:54<18:22, 1.37s/it] 97%|█████████▋| 29197/30000 [17:28:55<17:36, 1.32s/it] 97%|█████████▋| 29198/30000 [17:28:56<17:03, 1.28s/it] 97%|█████████▋| 29199/30000 [17:29:00<29:43, 2.23s/it] 97%|█████████▋| 29200/30000 [17:29:02<25:31, 1.91s/it] 97%|█████████▋| 29200/30000 [17:29:02<25:31, 1.91s/it] 97%|█████████▋| 29201/30000 [17:29:03<22:30, 1.69s/it] 97%|█████████▋| 29202/30000 [17:29:04<20:27, 1.54s/it] 97%|█████████▋| 29203/30000 [17:29:05<18:54, 1.42s/it] 97%|█████████▋| 29204/30000 [17:29:06<17:59, 1.36s/it] 97%|█████████▋| 29205/30000 [17:29:08<17:13, 1.30s/it] 97%|█████████▋| 29206/30000 [17:29:09<16:41, 1.26s/it] 97%|█████████▋| 29207/30000 [17:29:14<32:05, 2.43s/it] 97%|█████████▋| 29208/30000 [17:29:15<27:03, 2.05s/it] 97%|█████████▋| 29209/30000 [17:29:16<23:37, 1.79s/it] 97%|█████████▋| 29210/30000 [17:29:17<21:10, 1.61s/it] 97%|█████████▋| 29210/30000 [17:29:17<21:10, 1.61s/it] 97%|█████████▋| 29211/30000 [17:29:19<19:27, 1.48s/it] 97%|█████████▋| 29212/30000 [17:29:20<18:15, 1.39s/it] 97%|█████████▋| 29213/30000 [17:29:21<17:18, 1.32s/it] 97%|█████████▋| 29214/30000 [17:29:22<16:38, 1.27s/it] 97%|█████████▋| 29215/30000 [17:29:27<29:54, 2.29s/it] 97%|█████████▋| 29216/30000 [17:29:28<25:31, 1.95s/it] 97%|█████████▋| 29217/30000 [17:29:29<22:32, 1.73s/it] 97%|█████████▋| 29218/30000 [17:29:30<20:28, 1.57s/it] 97%|█████████▋| 29219/30000 [17:29:31<19:05, 1.47s/it] 97%|█████████▋| 29220/30000 [17:29:33<18:01, 1.39s/it] 97%|█████████▋| 29220/30000 [17:29:33<18:01, 1.39s/it] 97%|█████████▋| 29221/30000 [17:29:34<17:14, 1.33s/it] 97%|█████████▋| 29222/30000 [17:29:35<16:38, 1.28s/it] 97%|█████████▋| 29223/30000 [17:29:39<28:02, 2.17s/it] 97%|█████████▋| 29224/30000 [17:29:40<24:09, 1.87s/it] 97%|█████████▋| 29225/30000 [17:29:42<21:25, 1.66s/it] 97%|█████████▋| 29226/30000 [17:29:43<19:46, 1.53s/it] 97%|█████████▋| 29227/30000 [17:29:44<18:23, 1.43s/it] 97%|█████████▋| 29228/30000 [17:29:45<17:28, 1.36s/it] 97%|█████████▋| 29229/30000 [17:29:46<16:43, 1.30s/it] 97%|█████████▋| 29230/30000 [17:29:48<16:10, 1.26s/it] 97%|█████████▋| 29230/30000 [17:29:48<16:10, 1.26s/it] 97%|█████████▋| 29231/30000 [17:29:52<26:49, 2.09s/it] 97%|█████████▋| 29232/30000 [17:29:53<23:14, 1.82s/it] 97%|█████████▋| 29233/30000 [17:29:54<20:46, 1.63s/it] 97%|█████████▋| 29234/30000 [17:29:55<19:01, 1.49s/it] 97%|█████████▋| 29235/30000 [17:29:56<17:42, 1.39s/it] 97%|█████████▋| 29236/30000 [17:29:57<16:51, 1.32s/it] 97%|█████████▋| 29237/30000 [17:29:59<16:15, 1.28s/it] 97%|█████████▋| 29238/30000 [17:30:00<15:47, 1.24s/it] 97%|█████████▋| 29239/30000 [17:30:04<28:35, 2.25s/it] 97%|█████████▋| 29240/30000 [17:30:06<24:24, 1.93s/it] 97%|█████████▋| 29240/30000 [17:30:06<24:24, 1.93s/it] 97%|█████████▋| 29241/30000 [17:30:07<21:34, 1.71s/it] 97%|█████████▋| 29242/30000 [17:30:08<19:54, 1.58s/it] 97%|█████████▋| 29243/30000 [17:30:09<18:16, 1.45s/it] 97%|█████████▋| 29244/30000 [17:30:10<17:11, 1.36s/it] 97%|█████████▋| 29245/30000 [17:30:12<16:26, 1.31s/it] 97%|█████████▋| 29246/30000 [17:30:13<15:56, 1.27s/it] 97%|█████████▋| 29247/30000 [17:30:17<27:53, 2.22s/it] 97%|█████████▋| 29248/30000 [17:30:18<23:50, 1.90s/it] 97%|█████████▋| 29249/30000 [17:30:20<21:06, 1.69s/it] 98%|█████████▊| 29250/30000 [17:30:21<19:10, 1.53s/it] 98%|█████████▊| 29250/30000 [17:30:21<19:10, 1.53s/it] 98%|█████████▊| 29251/30000 [17:30:22<17:41, 1.42s/it] 98%|█████████▊| 29252/30000 [17:30:23<16:43, 1.34s/it] 98%|█████████▊| 29253/30000 [17:30:24<15:57, 1.28s/it] 98%|█████████▊| 29254/30000 [17:30:25<15:42, 1.26s/it] 98%|█████████▊| 29255/30000 [17:30:29<25:55, 2.09s/it] 98%|█████████▊| 29256/30000 [17:30:31<22:31, 1.82s/it] 98%|█████████▊| 29257/30000 [17:30:32<20:05, 1.62s/it] 98%|█████████▊| 29258/30000 [17:30:33<18:35, 1.50s/it] 98%|█████████▊| 29259/30000 [17:30:34<17:21, 1.41s/it] 98%|█████████▊| 29260/30000 [17:30:35<16:34, 1.34s/it] 98%|█████████▊| 29260/30000 [17:30:35<16:34, 1.34s/it] 98%|█████████▊| 29261/30000 [17:30:36<15:54, 1.29s/it] 98%|█████████▊| 29262/30000 [17:30:38<15:32, 1.26s/it] 98%|█████████▊| 29263/30000 [17:30:42<27:37, 2.25s/it] 98%|█████████▊| 29264/30000 [17:30:44<24:11, 1.97s/it] 98%|█████████▊| 29265/30000 [17:30:45<21:48, 1.78s/it] 98%|█████████▊| 29266/30000 [17:30:46<19:38, 1.61s/it] 98%|█████████▊| 29267/30000 [17:30:47<18:07, 1.48s/it] 98%|█████████▊| 29268/30000 [17:30:48<17:00, 1.39s/it] 98%|█████████▊| 29269/30000 [17:30:50<16:16, 1.34s/it] 98%|█████████▊| 29270/30000 [17:30:51<15:47, 1.30s/it] 98%|█████████▊| 29270/30000 [17:30:51<15:47, 1.30s/it] 98%|█████████▊| 29271/30000 [17:30:55<25:44, 2.12s/it] 98%|█████████▊| 29272/30000 [17:30:56<22:17, 1.84s/it] 98%|█████████▊| 29273/30000 [17:30:57<19:48, 1.64s/it] 98%|█████████▊| 29274/30000 [17:30:58<18:05, 1.50s/it] 98%|█████████▊| 29275/30000 [17:31:00<17:01, 1.41s/it] 98%|█████████▊| 29276/30000 [17:31:01<16:11, 1.34s/it] 98%|█████████▊| 29277/30000 [17:31:02<15:31, 1.29s/it] 98%|█████████▊| 29278/30000 [17:31:03<15:04, 1.25s/it] 98%|█████████▊| 29279/30000 [17:31:08<27:52, 2.32s/it] 98%|█████████▊| 29280/30000 [17:31:09<23:40, 1.97s/it] 98%|█████████▊| 29280/30000 [17:31:09<23:40, 1.97s/it] 98%|█████████▊| 29281/30000 [17:31:10<20:46, 1.73s/it] 98%|█████████▊| 29282/30000 [17:31:11<18:43, 1.56s/it] 98%|█████████▊| 29283/30000 [17:31:13<17:14, 1.44s/it] 98%|█████████▊| 29284/30000 [17:31:14<16:12, 1.36s/it] 98%|█████████▊| 29285/30000 [17:31:15<15:34, 1.31s/it] 98%|█████████▊| 29286/30000 [17:31:16<15:04, 1.27s/it] 98%|█████████▊| 29287/30000 [17:31:21<27:11, 2.29s/it] 98%|█████████▊| 29288/30000 [17:31:22<23:09, 1.95s/it] 98%|█████████▊| 29289/30000 [17:31:23<20:20, 1.72s/it] 98%|█████████▊| 29290/30000 [17:31:24<18:22, 1.55s/it] 98%|█████████▊| 29290/30000 [17:31:24<18:22, 1.55s/it] 98%|█████████▊| 29291/30000 [17:31:26<17:05, 1.45s/it] 98%|█████████▊| 29292/30000 [17:31:27<16:09, 1.37s/it] 98%|█████████▊| 29293/30000 [17:31:28<15:34, 1.32s/it] 98%|█████████▊| 29294/30000 [17:31:29<15:02, 1.28s/it] 98%|█████████▊| 29295/30000 [17:31:34<27:22, 2.33s/it] 98%|█████████▊| 29296/30000 [17:31:35<23:15, 1.98s/it] 98%|█████████▊| 29297/30000 [17:31:36<20:30, 1.75s/it] 98%|█████████▊| 29298/30000 [17:31:37<18:26, 1.58s/it] 98%|█████████▊| 29299/30000 [17:31:39<17:00, 1.46s/it] 98%|█████████▊| 29300/30000 [17:31:40<16:02, 1.37s/it] 98%|█████████▊| 29300/30000 [17:31:40<16:02, 1.37s/it] 98%|█████████▊| 29301/30000 [17:31:41<15:22, 1.32s/it] 98%|█████████▊| 29302/30000 [17:31:42<14:49, 1.27s/it] 98%|█████████▊| 29303/30000 [17:31:47<26:58, 2.32s/it] 98%|█████████▊| 29304/30000 [17:31:48<22:58, 1.98s/it] 98%|█████████▊| 29305/30000 [17:31:49<20:07, 1.74s/it] 98%|█████████▊| 29306/30000 [17:31:50<18:08, 1.57s/it] 98%|█████████▊| 29307/30000 [17:31:52<16:44, 1.45s/it] 98%|█████████▊| 29308/30000 [17:31:53<15:48, 1.37s/it] 98%|█████████▊| 29309/30000 [17:31:54<15:04, 1.31s/it] 98%|█████████▊| 29310/30000 [17:31:55<14:34, 1.27s/it] 98%|█████████▊| 29310/30000 [17:31:55<14:34, 1.27s/it] 98%|█████████▊| 29311/30000 [17:32:00<28:00, 2.44s/it] 98%|█████████▊| 29312/30000 [17:32:01<23:34, 2.06s/it] 98%|█████████▊| 29313/30000 [17:32:03<20:31, 1.79s/it] 98%|█████████▊| 29314/30000 [17:32:04<18:29, 1.62s/it] 98%|█████████▊| 29315/30000 [17:32:05<16:51, 1.48s/it] 98%|█████████▊| 29316/30000 [17:32:06<15:47, 1.39s/it] 98%|█████████▊| 29317/30000 [17:32:07<15:00, 1.32s/it] 98%|█████████▊| 29318/30000 [17:32:09<14:29, 1.27s/it] 98%|█████████▊| 29319/30000 [17:32:14<27:50, 2.45s/it] 98%|█████████▊| 29320/30000 [17:32:15<23:25, 2.07s/it] 98%|█████████▊| 29320/30000 [17:32:15<23:25, 2.07s/it] 98%|█████████▊| 29321/30000 [17:32:16<20:17, 1.79s/it] 98%|█████████▊| 29322/30000 [17:32:17<18:13, 1.61s/it] 98%|█████████▊| 29323/30000 [17:32:18<16:42, 1.48s/it] 98%|█████████▊| 29324/30000 [17:32:20<15:36, 1.39s/it] 98%|█████████▊| 29325/30000 [17:32:21<14:51, 1.32s/it] 98%|█████████▊| 29326/30000 [17:32:22<14:17, 1.27s/it] 98%|█████████▊| 29327/30000 [17:32:27<27:22, 2.44s/it] 98%|█████████▊| 29328/30000 [17:32:28<23:02, 2.06s/it] 98%|█████████▊| 29329/30000 [17:32:29<19:57, 1.78s/it] 98%|█████████▊| 29330/30000 [17:32:31<17:54, 1.60s/it] 98%|█████████▊| 29330/30000 [17:32:31<17:54, 1.60s/it] 98%|█████████▊| 29331/30000 [17:32:32<16:26, 1.47s/it] 98%|█████████▊| 29332/30000 [17:32:33<15:26, 1.39s/it] 98%|█████████▊| 29333/30000 [17:32:34<14:42, 1.32s/it] 98%|█████████▊| 29334/30000 [17:32:35<14:07, 1.27s/it] 98%|█████████▊| 29335/30000 [17:32:40<27:01, 2.44s/it] 98%|█████████▊| 29336/30000 [17:32:42<22:44, 2.06s/it] 98%|█████████▊| 29337/30000 [17:32:43<19:45, 1.79s/it] 98%|█████████▊| 29338/30000 [17:32:44<17:40, 1.60s/it] 98%|█████████▊| 29339/30000 [17:32:45<16:10, 1.47s/it] 98%|█████████▊| 29340/30000 [17:32:46<15:12, 1.38s/it] 98%|█████████▊| 29340/30000 [17:32:46<15:12, 1.38s/it] 98%|█████████▊| 29341/30000 [17:32:47<14:31, 1.32s/it] 98%|█████████▊| 29342/30000 [17:32:49<14:01, 1.28s/it] 98%|█████████▊| 29343/30000 [17:32:53<24:26, 2.23s/it] 98%|█████████▊| 29344/30000 [17:32:54<20:52, 1.91s/it] 98%|█████████▊| 29345/30000 [17:32:55<18:26, 1.69s/it] 98%|█████████▊| 29346/30000 [17:32:57<16:47, 1.54s/it] 98%|█████████▊| 29347/30000 [17:32:58<15:41, 1.44s/it] 98%|█████████▊| 29348/30000 [17:32:59<14:59, 1.38s/it] 98%|█████████▊| 29349/30000 [17:33:00<14:18, 1.32s/it] 98%|█████████▊| 29350/30000 [17:33:01<14:06, 1.30s/it] 98%|█████████▊| 29350/30000 [17:33:01<14:06, 1.30s/it] 98%|█████████▊| 29351/30000 [17:33:06<24:37, 2.28s/it] 98%|█████████▊| 29352/30000 [17:33:07<21:01, 1.95s/it] 98%|█████████▊| 29353/30000 [17:33:08<18:31, 1.72s/it] 98%|█████████▊| 29354/30000 [17:33:10<16:45, 1.56s/it] 98%|█████████▊| 29355/30000 [17:33:11<15:30, 1.44s/it] 98%|█████████▊| 29356/30000 [17:33:12<14:41, 1.37s/it] 98%|█████████▊| 29357/30000 [17:33:13<14:08, 1.32s/it] 98%|█████████▊| 29358/30000 [17:33:14<13:36, 1.27s/it] 98%|█████████▊| 29359/30000 [17:33:19<24:07, 2.26s/it] 98%|█████████▊| 29360/30000 [17:33:20<20:34, 1.93s/it] 98%|█████████▊| 29360/30000 [17:33:20<20:34, 1.93s/it] 98%|█████████▊| 29361/30000 [17:33:21<18:10, 1.71s/it] 98%|█████████▊| 29362/30000 [17:33:22<16:30, 1.55s/it] 98%|█████████▊| 29363/30000 [17:33:24<15:29, 1.46s/it] 98%|█████████▊| 29364/30000 [17:33:25<15:05, 1.42s/it] 98%|█████████▊| 29365/30000 [17:33:26<14:20, 1.35s/it] 98%|█████████▊| 29366/30000 [17:33:27<13:48, 1.31s/it] 98%|█████████▊| 29367/30000 [17:33:32<23:44, 2.25s/it] 98%|█████████▊| 29368/30000 [17:33:33<20:17, 1.93s/it] 98%|█████████▊| 29369/30000 [17:33:34<17:51, 1.70s/it] 98%|█████████▊| 29370/30000 [17:33:35<16:10, 1.54s/it] 98%|█████████▊| 29370/30000 [17:33:35<16:10, 1.54s/it] 98%|█████████▊| 29371/30000 [17:33:36<14:56, 1.43s/it] 98%|█████████▊| 29372/30000 [17:33:38<14:08, 1.35s/it] 98%|█████████▊| 29373/30000 [17:33:39<13:33, 1.30s/it] 98%|█████████▊| 29374/30000 [17:33:40<13:12, 1.27s/it] 98%|█████████▊| 29375/30000 [17:33:45<23:37, 2.27s/it] 98%|█████████▊| 29376/30000 [17:33:46<20:12, 1.94s/it] 98%|█████████▊| 29377/30000 [17:33:47<17:43, 1.71s/it] 98%|█████████▊| 29378/30000 [17:33:48<16:01, 1.55s/it] 98%|█████████▊| 29379/30000 [17:33:49<14:50, 1.43s/it] 98%|█████████▊| 29380/30000 [17:33:50<13:58, 1.35s/it] 98%|█████████▊| 29380/30000 [17:33:50<13:58, 1.35s/it] 98%|█████████▊| 29381/30000 [17:33:52<13:27, 1.30s/it] 98%|█████████▊| 29382/30000 [17:33:53<13:04, 1.27s/it] 98%|█████████▊| 29383/30000 [17:33:57<22:35, 2.20s/it] 98%|█████████▊| 29384/30000 [17:33:58<19:27, 1.90s/it] 98%|█████████▊| 29385/30000 [17:34:00<17:11, 1.68s/it] 98%|█████████▊| 29386/30000 [17:34:01<15:33, 1.52s/it] 98%|█████████▊| 29387/30000 [17:34:02<14:29, 1.42s/it] 98%|█████████▊| 29388/30000 [17:34:03<13:44, 1.35s/it] 98%|█████████▊| 29389/30000 [17:34:04<13:11, 1.29s/it] 98%|█████████▊| 29390/30000 [17:34:05<12:49, 1.26s/it] +{'loss': 0.0119, 'grad_norm': 0.07181645184755325, 'learning_rate': 9.489958286311883e-08, 'flow_matching_loss': 0.01191741987131536, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.43} +{'loss': 0.0121, 'grad_norm': 0.05911954492330551, 'learning_rate': 9.30516477988541e-08, 'flow_matching_loss': 0.012062098365277052, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.43} +{'loss': 0.0126, 'grad_norm': 0.07594984769821167, 'learning_rate': 9.122182608247098e-08, 'flow_matching_loss': 0.01264209058135748, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.44} +{'loss': 0.0124, 'grad_norm': 0.05738458409905434, 'learning_rate': 8.94101199373759e-08, 'flow_matching_loss': 0.012350694090127946, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.44} +{'loss': 0.012, 'grad_norm': 0.056578561663627625, 'learning_rate': 8.761653156495964e-08, 'flow_matching_loss': 0.011987593956291676, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.44} +{'loss': 0.013, 'grad_norm': 0.07732746005058289, 'learning_rate': 8.584106314460549e-08, 'flow_matching_loss': 0.013049287442117929, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.45} +{'loss': 0.0137, 'grad_norm': 0.07627490162849426, 'learning_rate': 8.408371683367111e-08, 'flow_matching_loss': 0.013672502245754004, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.45} +{'loss': 0.0121, 'grad_norm': 0.06678113341331482, 'learning_rate': 8.234449476750172e-08, 'flow_matching_loss': 0.012105375621467828, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.45} +{'loss': 0.0129, 'grad_norm': 0.06250522285699844, 'learning_rate': 8.062339905941351e-08, 'flow_matching_loss': 0.012913932371884585, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.46} +{'loss': 0.0131, 'grad_norm': 0.06952090561389923, 'learning_rate': 7.89204318007003e-08, 'flow_matching_loss': 0.013064093980938196, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.46} +{'loss': 0.0119, 'grad_norm': 0.06045028194785118, 'learning_rate': 7.723559506062849e-08, 'flow_matching_loss': 0.011925913207232953, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.46} +{'loss': 0.0131, 'grad_norm': 0.054545897990465164, 'learning_rate': 7.556889088643381e-08, 'flow_matching_loss': 0.01306575033813715, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.46} +{'loss': 0.0137, 'grad_norm': 0.09854549914598465, 'learning_rate': 7.392032130331793e-08, 'flow_matching_loss': 0.013650075066834688, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.47} +{'loss': 0.0155, 'grad_norm': 0.08914481103420258, 'learning_rate': 7.228988831445015e-08, 'flow_matching_loss': 0.015546680241823197, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.47} +{'loss': 0.013, 'grad_norm': 0.07046489417552948, 'learning_rate': 7.06775939009574e-08, 'flow_matching_loss': 0.012990179564803838, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.47} +{'loss': 0.0131, 'grad_norm': 0.048914603888988495, 'learning_rate': 6.908344002193256e-08, 'flow_matching_loss': 0.013071298552677035, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.48} +{'loss': 0.0117, 'grad_norm': 0.0493750236928463, 'learning_rate': 6.750742861442116e-08, 'flow_matching_loss': 0.011695770639926195, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.48} +{'loss': 0.0121, 'grad_norm': 0.05984507501125336, 'learning_rate': 6.594956159342636e-08, 'flow_matching_loss': 0.012090576719492674, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.48} +{'loss': 0.0138, 'grad_norm': 0.05642218887805939, 'learning_rate': 6.440984085190394e-08, 'flow_matching_loss': 0.01382071441039443, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.48} +{'loss': 0.0124, 'grad_norm': 0.058013807982206345, 'learning_rate': 6.288826826075733e-08, 'flow_matching_loss': 0.012403417937457561, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.49} +{'loss': 0.0131, 'grad_norm': 0.07396427541971207, 'learning_rate': 6.138484566884261e-08, 'flow_matching_loss': 0.013050282746553421, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.49} +{'loss': 0.0122, 'grad_norm': 0.06211281195282936, 'learning_rate': 5.989957490296183e-08, 'flow_matching_loss': 0.012217097543179989, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.49} +{'loss': 0.0119, 'grad_norm': 0.06047975271940231, 'learning_rate': 5.843245776785466e-08, 'flow_matching_loss': 0.011936973966658115, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.5} +{'loss': 0.0124, 'grad_norm': 0.06770820170640945, 'learning_rate': 5.6983496046210115e-08, 'flow_matching_loss': 0.012397396750748158, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.5} +{'loss': 0.0127, 'grad_norm': 0.06579840928316116, 'learning_rate': 5.5552691498653165e-08, 'flow_matching_loss': 0.012652566749602556, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.5} +{'loss': 0.0122, 'grad_norm': 0.05626003444194794, 'learning_rate': 5.4140045863743106e-08, 'flow_matching_loss': 0.012185069266706705, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.5} +{'loss': 0.0138, 'grad_norm': 0.06347176432609558, 'learning_rate': 5.274556085798188e-08, 'flow_matching_loss': 0.013780882116407156, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.51} +{'loss': 0.0122, 'grad_norm': 0.055616140365600586, 'learning_rate': 5.136923817579742e-08, 'flow_matching_loss': 0.012242414522916079, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.51} +{'loss': 0.0128, 'grad_norm': 0.07311689108610153, 'learning_rate': 5.001107948955197e-08, 'flow_matching_loss': 0.012812222260981799, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.51} +{'loss': 0.0126, 'grad_norm': 0.06727994978427887, 'learning_rate': 4.867108644953711e-08, 'flow_matching_loss': 0.012554540857672691, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.52} +{'loss': 0.0147, 'grad_norm': 0.07841529697179794, 'learning_rate': 4.734926068397038e-08, 'flow_matching_loss': 0.014720562659204007, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.52} +{'loss': 0.0141, 'grad_norm': 0.06437932699918747, 'learning_rate': 4.604560379899536e-08, 'flow_matching_loss': 0.014147276151925325, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.52} +{'loss': 0.0131, 'grad_norm': 0.06689436733722687, 'learning_rate': 4.476011737867658e-08, 'flow_matching_loss': 0.013111739978194236, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.52} +{'loss': 0.0132, 'grad_norm': 0.05986233428120613, 'learning_rate': 4.349280298500291e-08, 'flow_matching_loss': 0.013202043808996677, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.53} +{'loss': 0.0129, 'grad_norm': 0.09957590699195862, 'learning_rate': 4.224366215787923e-08, 'flow_matching_loss': 0.012906148005276919, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.53} +{'loss': 0.0112, 'grad_norm': 0.07120413333177567, 'learning_rate': 4.101269641513139e-08, 'flow_matching_loss': 0.011154662817716599, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.53} +{'loss': 0.0144, 'grad_norm': 0.07022010535001755, 'learning_rate': 3.979990725249627e-08, 'flow_matching_loss': 0.014442384522408248, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.54} +{'loss': 0.014, 'grad_norm': 0.07072775810956955, 'learning_rate': 3.860529614363006e-08, 'flow_matching_loss': 0.014020628295838833, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.54} +{'loss': 0.013, 'grad_norm': 0.06555414944887161, 'learning_rate': 3.742886454009664e-08, 'flow_matching_loss': 0.013012716360390187, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.54} +{'loss': 0.0114, 'grad_norm': 0.06440197676420212, 'learning_rate': 3.627061387137087e-08, 'flow_matching_loss': 0.011351094581186771, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.55} +{'loss': 0.0137, 'grad_norm': 0.0823453813791275, 'learning_rate': 3.5130545544835305e-08, 'flow_matching_loss': 0.013685052189975977, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.55} +{'loss': 0.0139, 'grad_norm': 0.07791199535131454, 'learning_rate': 3.400866094578348e-08, 'flow_matching_loss': 0.013926566205918789, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.55} 98%|█████████▊| 29390/30000 [17:34:05<12:49, 1.26s/it] 98%|█████████▊| 29391/30000 [17:34:10<23:31, 2.32s/it] 98%|█████████▊| 29392/30000 [17:34:11<20:01, 1.98s/it] 98%|█████████▊| 29393/30000 [17:34:13<17:34, 1.74s/it] 98%|█████████▊| 29394/30000 [17:34:14<15:54, 1.57s/it] 98%|█████████▊| 29395/30000 [17:34:15<14:37, 1.45s/it] 98%|█████████▊| 29396/30000 [17:34:16<13:49, 1.37s/it] 98%|█████████▊| 29397/30000 [17:34:17<13:15, 1.32s/it] 98%|█████████▊| 29398/30000 [17:34:19<12:51, 1.28s/it] 98%|█████████▊| 29399/30000 [17:34:23<22:00, 2.20s/it] 98%|█████████▊| 29400/30000 [17:34:24<18:57, 1.90s/it] 98%|█████████▊| 29400/30000 [17:34:24<18:57, 1.90s/it] 98%|█████████▊| 29401/30000 [17:34:25<16:47, 1.68s/it] 98%|█████████▊| 29402/30000 [17:34:26<15:17, 1.53s/it] 98%|█████████▊| 29403/30000 [17:34:28<14:09, 1.42s/it] 98%|█████████▊| 29404/30000 [17:34:29<13:27, 1.35s/it] 98%|█████████▊| 29405/30000 [17:34:30<12:57, 1.31s/it] 98%|█████████▊| 29406/30000 [17:34:31<12:32, 1.27s/it] 98%|█████████▊| 29407/30000 [17:34:35<21:15, 2.15s/it] 98%|█████████▊| 29408/30000 [17:34:37<18:15, 1.85s/it] 98%|█████████▊| 29409/30000 [17:34:38<16:12, 1.64s/it] 98%|█████████▊| 29410/30000 [17:34:39<14:50, 1.51s/it] 98%|█████████▊| 29410/30000 [17:34:39<14:50, 1.51s/it] 98%|█████████▊| 29411/30000 [17:34:40<13:51, 1.41s/it] 98%|█████████▊| 29412/30000 [17:34:41<13:05, 1.34s/it] 98%|█████████▊| 29413/30000 [17:34:42<12:39, 1.29s/it] 98%|█████████▊| 29414/30000 [17:34:44<12:21, 1.27s/it] 98%|█████████▊| 29415/30000 [17:34:48<22:06, 2.27s/it] 98%|█████████▊| 29416/30000 [17:34:49<18:51, 1.94s/it] 98%|█████████▊| 29417/30000 [17:34:51<16:33, 1.70s/it] 98%|█████████▊| 29418/30000 [17:34:52<15:04, 1.55s/it] 98%|█████████▊| 29419/30000 [17:34:53<13:57, 1.44s/it] 98%|█████████▊| 29420/30000 [17:34:54<13:13, 1.37s/it] 98%|█████████▊| 29420/30000 [17:34:54<13:13, 1.37s/it] 98%|█████████▊| 29421/30000 [17:34:55<12:35, 1.30s/it] 98%|█████████▊| 29422/30000 [17:34:56<12:10, 1.26s/it] 98%|█████████▊| 29423/30000 [17:35:02<23:06, 2.40s/it] 98%|█████████▊| 29424/30000 [17:35:03<19:31, 2.03s/it] 98%|█████████▊| 29425/30000 [17:35:04<17:02, 1.78s/it] 98%|█████████▊| 29426/30000 [17:35:05<15:17, 1.60s/it] 98%|█████████▊| 29427/30000 [17:35:06<14:01, 1.47s/it] 98%|█████████▊| 29428/30000 [17:35:07<13:11, 1.38s/it] 98%|█████████▊| 29429/30000 [17:35:09<12:38, 1.33s/it] 98%|█████████▊| 29430/30000 [17:35:10<12:08, 1.28s/it] 98%|█████████▊| 29430/30000 [17:35:10<12:08, 1.28s/it] 98%|█████████▊| 29431/30000 [17:35:15<23:51, 2.52s/it] 98%|█████████▊| 29432/30000 [17:35:16<19:59, 2.11s/it] 98%|█████████▊| 29433/30000 [17:35:18<17:19, 1.83s/it] 98%|█████████▊| 29434/30000 [17:35:19<15:23, 1.63s/it] 98%|█████████▊| 29435/30000 [17:35:20<14:01, 1.49s/it] 98%|█████████▊| 29436/30000 [17:35:21<13:09, 1.40s/it] 98%|█████████▊| 29437/30000 [17:35:22<12:28, 1.33s/it] 98%|█████████▊| 29438/30000 [17:35:23<12:06, 1.29s/it] 98%|█████████▊| 29439/30000 [17:35:29<22:52, 2.45s/it] 98%|█████████▊| 29440/30000 [17:35:30<19:18, 2.07s/it] 98%|█████████▊| 29440/30000 [17:35:30<19:18, 2.07s/it] 98%|█████████▊| 29441/30000 [17:35:31<16:49, 1.81s/it] 98%|█████████▊| 29442/30000 [17:35:32<15:01, 1.62s/it] 98%|█████████▊| 29443/30000 [17:35:33<13:50, 1.49s/it] 98%|█████████▊| 29444/30000 [17:35:35<13:09, 1.42s/it] 98%|█████████▊| 29445/30000 [17:35:36<12:32, 1.36s/it] 98%|█████████▊| 29446/30000 [17:35:37<12:02, 1.30s/it] 98%|█████████▊| 29447/30000 [17:35:41<20:52, 2.26s/it] 98%|█████████▊| 29448/30000 [17:35:43<17:57, 1.95s/it] 98%|█████████▊| 29449/30000 [17:35:44<15:46, 1.72s/it] 98%|█████████▊| 29450/30000 [17:35:45<14:14, 1.55s/it] 98%|█████████▊| 29450/30000 [17:35:45<14:14, 1.55s/it] 98%|█████████▊| 29451/30000 [17:35:46<13:07, 1.43s/it] 98%|█████████▊| 29452/30000 [17:35:47<12:27, 1.36s/it] 98%|█████████▊| 29453/30000 [17:35:49<11:55, 1.31s/it] 98%|█████████▊| 29454/30000 [17:35:50<11:33, 1.27s/it] 98%|█████████▊| 29455/30000 [17:35:56<25:56, 2.86s/it] 98%|█████████▊| 29456/30000 [17:35:58<21:30, 2.37s/it] 98%|█████████▊| 29457/30000 [17:35:59<18:18, 2.02s/it] 98%|█████████▊| 29458/30000 [17:36:00<16:00, 1.77s/it] 98%|█████████▊| 29459/30000 [17:36:01<14:24, 1.60s/it] 98%|█████████▊| 29460/30000 [17:36:02<13:19, 1.48s/it] 98%|█████████▊| 29460/30000 [17:36:02<13:19, 1.48s/it] 98%|█████████▊| 29461/30000 [17:36:04<12:35, 1.40s/it] 98%|█████████▊| 29462/30000 [17:36:05<12:00, 1.34s/it] 98%|█████████▊| 29463/30000 [17:36:09<21:07, 2.36s/it] 98%|█████████▊| 29464/30000 [17:36:11<17:58, 2.01s/it] 98%|█████████▊| 29465/30000 [17:36:12<15:41, 1.76s/it] 98%|█████████▊| 29466/30000 [17:36:13<14:04, 1.58s/it] 98%|█████████▊| 29467/30000 [17:36:14<12:57, 1.46s/it] 98%|█████████▊| 29468/30000 [17:36:15<12:13, 1.38s/it] 98%|█████████▊| 29469/30000 [17:36:17<11:43, 1.32s/it] 98%|█████████▊| 29470/30000 [17:36:18<11:17, 1.28s/it] 98%|█████████▊| 29470/30000 [17:36:18<11:17, 1.28s/it] 98%|█████████▊| 29471/30000 [17:36:22<18:53, 2.14s/it] 98%|█████████▊| 29472/30000 [17:36:23<16:17, 1.85s/it] 98%|█████████▊| 29473/30000 [17:36:24<14:33, 1.66s/it] 98%|█████████▊| 29474/30000 [17:36:25<13:17, 1.52s/it] 98%|█████████▊| 29475/30000 [17:36:27<12:19, 1.41s/it] 98%|█████████▊| 29476/30000 [17:36:28<11:42, 1.34s/it] 98%|█████████▊| 29477/30000 [17:36:29<11:13, 1.29s/it] 98%|█████████▊| 29478/30000 [17:36:30<11:05, 1.27s/it] 98%|█████████▊| 29479/30000 [17:36:35<19:05, 2.20s/it] 98%|█████████▊| 29480/30000 [17:36:36<16:22, 1.89s/it] 98%|█████████▊| 29480/30000 [17:36:36<16:22, 1.89s/it] 98%|█████████▊| 29481/30000 [17:36:37<14:31, 1.68s/it] 98%|█████████▊| 29482/30000 [17:36:38<13:11, 1.53s/it] 98%|█████████▊| 29483/30000 [17:36:39<12:13, 1.42s/it] 98%|█████████▊| 29484/30000 [17:36:40<11:32, 1.34s/it] 98%|█████████▊| 29485/30000 [17:36:42<11:04, 1.29s/it] 98%|█████████▊| 29486/30000 [17:36:43<10:44, 1.25s/it] 98%|█████████▊| 29487/30000 [17:36:47<18:47, 2.20s/it] 98%|█████████▊| 29488/30000 [17:36:48<16:07, 1.89s/it] 98%|█████████▊| 29489/30000 [17:36:49<14:14, 1.67s/it] 98%|█████████▊| 29490/30000 [17:36:51<12:57, 1.52s/it] 98%|█████████▊| 29490/30000 [17:36:51<12:57, 1.52s/it] 98%|█████████▊| 29491/30000 [17:36:52<12:00, 1.42s/it] 98%|█████████▊| 29492/30000 [17:36:53<11:23, 1.35s/it] 98%|█████████▊| 29493/30000 [17:36:54<10:58, 1.30s/it] 98%|█████████▊| 29494/30000 [17:36:55<10:36, 1.26s/it] 98%|█████████▊| 29495/30000 [17:37:00<19:42, 2.34s/it] 98%|█████████▊| 29496/30000 [17:37:01<16:42, 1.99s/it] 98%|█████████▊| 29497/30000 [17:37:03<14:32, 1.74s/it] 98%|█████████▊| 29498/30000 [17:37:04<13:08, 1.57s/it] 98%|█████████▊| 29499/30000 [17:37:05<12:05, 1.45s/it] 98%|█████████▊| 29500/30000 [17:37:06<11:23, 1.37s/it] 98%|█████████▊| 29500/30000 [17:37:06<11:23, 1.37s/it] 98%|█████████▊| 29501/30000 [17:37:07<10:52, 1.31s/it] 98%|█████████▊| 29502/30000 [17:37:08<10:29, 1.26s/it] 98%|█████████▊| 29503/30000 [17:37:13<19:32, 2.36s/it] 98%|█████████▊| 29504/30000 [17:37:15<16:41, 2.02s/it] 98%|█████████▊| 29505/30000 [17:37:16<14:35, 1.77s/it] 98%|█████████▊| 29506/30000 [17:37:17<13:05, 1.59s/it] 98%|█████████▊| 29507/30000 [17:37:18<12:01, 1.46s/it] 98%|█████████▊| 29508/30000 [17:37:19<11:16, 1.38s/it] 98%|█████████▊| 29509/30000 [17:37:20<10:47, 1.32s/it] 98%|█████████▊| 29510/30000 [17:37:22<10:26, 1.28s/it] 98%|█████████▊| 29510/30000 [17:37:22<10:26, 1.28s/it] 98%|█████████▊| 29511/30000 [17:37:26<17:53, 2.19s/it] 98%|█████████▊| 29512/30000 [17:37:27<15:21, 1.89s/it] 98%|█████████▊| 29513/30000 [17:37:28<13:35, 1.67s/it] 98%|█████████▊| 29514/30000 [17:37:29<12:22, 1.53s/it] 98%|█████████▊| 29515/30000 [17:37:31<11:27, 1.42s/it] 98%|█████████▊| 29516/30000 [17:37:32<10:59, 1.36s/it] 98%|█████████▊| 29517/30000 [17:37:33<10:27, 1.30s/it] 98%|█████████▊| 29518/30000 [17:37:34<10:10, 1.27s/it] 98%|█████████▊| 29519/30000 [17:37:39<19:05, 2.38s/it] 98%|█████████▊| 29520/30000 [17:37:40<16:10, 2.02s/it] 98%|█████████▊| 29520/30000 [17:37:40<16:10, 2.02s/it] 98%|█████████▊| 29521/30000 [17:37:42<14:09, 1.77s/it] 98%|█████████▊| 29522/30000 [17:37:43<12:44, 1.60s/it] 98%|█████████▊| 29523/30000 [17:37:44<11:41, 1.47s/it] 98%|█████████▊| 29524/30000 [17:37:45<10:55, 1.38s/it] 98%|█████████▊| 29525/30000 [17:37:46<10:25, 1.32s/it] 98%|█████████▊| 29526/30000 [17:37:47<10:03, 1.27s/it] 98%|█████████▊| 29527/30000 [17:37:52<18:10, 2.31s/it] 98%|█████████▊| 29528/30000 [17:37:53<15:26, 1.96s/it] 98%|█████████▊| 29529/30000 [17:37:55<13:37, 1.74s/it] 98%|█████████▊| 29530/30000 [17:37:56<12:16, 1.57s/it] 98%|█████████▊| 29530/30000 [17:37:56<12:16, 1.57s/it] 98%|█████████▊| 29531/30000 [17:37:57<11:19, 1.45s/it] 98%|█████████▊| 29532/30000 [17:37:58<10:38, 1.37s/it] 98%|█████████▊| 29533/30000 [17:37:59<10:10, 1.31s/it] 98%|█████████▊| 29534/30000 [17:38:00<09:53, 1.27s/it] 98%|█████████▊| 29535/30000 [17:38:05<17:25, 2.25s/it] 98%|█████████▊| 29536/30000 [17:38:06<14:54, 1.93s/it] 98%|█████████▊| 29537/30000 [17:38:07<13:04, 1.70s/it] 98%|█████████▊| 29538/30000 [17:38:08<11:50, 1.54s/it] 98%|█████████▊| 29539/30000 [17:38:10<10:58, 1.43s/it] 98%|█████████▊| 29540/30000 [17:38:11<10:21, 1.35s/it] 98%|█████████▊| 29540/30000 [17:38:11<10:21, 1.35s/it] 98%|█████████▊| 29541/30000 [17:38:12<09:58, 1.30s/it] 98%|█████████▊| 29542/30000 [17:38:13<09:39, 1.26s/it] 98%|█████████▊| 29543/30000 [17:38:18<16:45, 2.20s/it] 98%|█████████▊| 29544/30000 [17:38:19<14:22, 1.89s/it] 98%|█████████▊| 29545/30000 [17:38:20<12:42, 1.68s/it] 98%|█████████▊| 29546/30000 [17:38:21<11:33, 1.53s/it] 98%|█████████▊| 29547/30000 [17:38:22<10:43, 1.42s/it] 98%|█████████▊| 29548/30000 [17:38:23<10:07, 1.34s/it] 98%|█████████▊| 29549/30000 [17:38:25<09:41, 1.29s/it] 98%|█████████▊| 29550/30000 [17:38:26<09:23, 1.25s/it] 98%|█████████▊| 29550/30000 [17:38:26<09:23, 1.25s/it] 99%|█████████▊| 29551/30000 [17:38:30<16:35, 2.22s/it] 99%|█████████▊| 29552/30000 [17:38:31<14:14, 1.91s/it] 99%|█████████▊| 29553/30000 [17:38:33<12:33, 1.69s/it] 99%|█████████▊| 29554/30000 [17:38:34<11:24, 1.54s/it] 99%|█████████▊| 29555/30000 [17:38:35<10:36, 1.43s/it] 99%|█████████▊| 29556/30000 [17:38:36<10:04, 1.36s/it] 99%|█████████▊| 29557/30000 [17:38:37<09:36, 1.30s/it] 99%|█████████▊| 29558/30000 [17:38:38<09:18, 1.26s/it] 99%|█████████▊| 29559/30000 [17:38:44<18:30, 2.52s/it] 99%|█████████▊| 29560/30000 [17:38:45<15:30, 2.12s/it] 99%|█████████▊| 29560/30000 [17:38:45<15:30, 2.12s/it] 99%|█████████▊| 29561/30000 [17:38:46<13:24, 1.83s/it] 99%|█████████▊| 29562/30000 [17:38:47<11:55, 1.63s/it] 99%|█████████▊| 29563/30000 [17:38:49<10:51, 1.49s/it] 99%|█████████▊| 29564/30000 [17:38:50<10:11, 1.40s/it] 99%|█████████▊| 29565/30000 [17:38:51<09:38, 1.33s/it] 99%|█████████▊| 29566/30000 [17:38:52<09:18, 1.29s/it] 99%|█████████▊| 29567/30000 [17:38:56<16:00, 2.22s/it] 99%|█████████▊| 29568/30000 [17:38:58<14:01, 1.95s/it] 99%|█████████▊| 29569/30000 [17:38:59<12:22, 1.72s/it] 99%|█████████▊| 29570/30000 [17:39:00<11:17, 1.57s/it] 99%|█████████▊| 29570/30000 [17:39:00<11:17, 1.57s/it] 99%|█████████▊| 29571/30000 [17:39:01<10:25, 1.46s/it] 99%|█████████▊| 29572/30000 [17:39:03<09:46, 1.37s/it] 99%|█████████▊| 29573/30000 [17:39:04<09:21, 1.31s/it] 99%|█████████▊| 29574/30000 [17:39:05<09:04, 1.28s/it] 99%|█████████▊| 29575/30000 [17:39:09<15:17, 2.16s/it] 99%|█████████▊| 29576/30000 [17:39:10<13:10, 1.86s/it] 99%|█████████▊| 29577/30000 [17:39:12<11:39, 1.65s/it] 99%|█████████▊| 29578/30000 [17:39:13<10:35, 1.51s/it] 99%|█████████▊| 29579/30000 [17:39:14<09:50, 1.40s/it] 99%|█████████▊| 29580/30000 [17:39:15<09:20, 1.34s/it] 99%|█████████▊| 29580/30000 [17:39:15<09:20, 1.34s/it] 99%|█████████▊| 29581/30000 [17:39:16<08:59, 1.29s/it] 99%|█████████▊| 29582/30000 [17:39:17<08:47, 1.26s/it] 99%|█████████▊| 29583/30000 [17:39:22<16:36, 2.39s/it] 99%|█████████▊| 29584/30000 [17:39:24<14:02, 2.02s/it] 99%|█████████▊| 29585/30000 [17:39:25<12:14, 1.77s/it] 99%|█████████▊| 29586/30000 [17:39:26<10:58, 1.59s/it] 99%|█████████▊| 29587/30000 [17:39:27<10:05, 1.47s/it] 99%|█████████▊| 29588/30000 [17:39:28<09:30, 1.38s/it] 99%|█████████▊| 29589/30000 [17:39:29<09:01, 1.32s/it] 99%|█████████▊| 29590/30000 [17:39:31<08:43, 1.28s/it] 99%|█████████▊| 29590/30000 [17:39:31<08:43, 1.28s/it] 99%|█████████▊| 29591/30000 [17:39:35<15:12, 2.23s/it] 99%|█████████▊| 29592/30000 [17:39:36<13:01, 1.92s/it] 99%|█████████▊| 29593/30000 [17:39:37<11:31, 1.70s/it] 99%|█████████▊| 29594/30000 [17:39:39<10:26, 1.54s/it] 99%|█████████▊| 29595/30000 [17:39:40<09:37, 1.43s/it] 99%|█████████▊| 29596/30000 [17:39:41<09:06, 1.35s/it] 99%|█████████▊| 29597/30000 [17:39:42<08:41, 1.29s/it] 99%|█████████▊| 29598/30000 [17:39:43<08:27, 1.26s/it] 99%|█████████▊| 29599/30000 [17:39:48<16:05, 2.41s/it] 99%|█████████▊| 29600/30000 [17:39:50<13:37, 2.04s/it] 99%|█████████▊| 29600/30000 [17:39:50<13:37, 2.04s/it] 99%|█████████▊| 29601/30000 [17:39:51<11:53, 1.79s/it] 99%|█████████▊| 29602/30000 [17:39:52<10:38, 1.60s/it] 99%|█████████▊| 29603/30000 [17:39:53<09:43, 1.47s/it] 99%|█████████▊| 29604/30000 [17:39:54<09:09, 1.39s/it] 99%|█████████▊| 29605/30000 [17:39:56<08:44, 1.33s/it] 99%|█████████▊| 29606/30000 [17:39:57<08:24, 1.28s/it] 99%|█████████▊| 29607/30000 [17:40:02<15:53, 2.43s/it] 99%|█████████▊| 29608/30000 [17:40:03<13:25, 2.06s/it] 99%|█████████▊| 29609/30000 [17:40:04<11:48, 1.81s/it] 99%|█████████▊| 29610/30000 [17:40:05<10:34, 1.63s/it] 99%|█████████▊| 29610/30000 [17:40:05<10:34, 1.63s/it] 99%|█████████▊| 29611/30000 [17:40:07<09:43, 1.50s/it] 99%|█████████▊| 29612/30000 [17:40:08<09:05, 1.41s/it] 99%|█████████▊| 29613/30000 [17:40:09<08:35, 1.33s/it] 99%|█████████▊| 29614/30000 [17:40:10<08:17, 1.29s/it] 99%|█████████▊| 29615/30000 [17:40:14<14:02, 2.19s/it] 99%|█████████▊| 29616/30000 [17:40:16<12:07, 1.89s/it] 99%|█████████▊| 29617/30000 [17:40:17<10:48, 1.69s/it] 99%|█████████▊| 29618/30000 [17:40:18<09:51, 1.55s/it] 99%|█████████▊| 29619/30000 [17:40:19<09:07, 1.44s/it] 99%|█████████▊| 29620/30000 [17:40:21<08:44, 1.38s/it] 99%|█████████▊| 29620/30000 [17:40:21<08:44, 1.38s/it] 99%|█████████▊| 29621/30000 [17:40:22<08:27, 1.34s/it] 99%|█████████▊| 29622/30000 [17:40:23<08:16, 1.31s/it] 99%|█████████▊| 29623/30000 [17:40:27<12:33, 2.00s/it] 99%|█████████▊| 29624/30000 [17:40:28<11:02, 1.76s/it] 99%|█████████▉| 29625/30000 [17:40:29<09:56, 1.59s/it] 99%|█████████▉| 29626/30000 [17:40:30<09:08, 1.47s/it] 99%|█████████▉| 29627/30000 [17:40:31<08:33, 1.38s/it] 99%|█████████▉| 29628/30000 [17:40:33<08:09, 1.32s/it] 99%|█████████▉| 29629/30000 [17:40:34<07:51, 1.27s/it] 99%|█████████▉| 29630/30000 [17:40:35<07:37, 1.24s/it] 99%|█████████▉| 29630/30000 [17:40:35<07:37, 1.24s/it] 99%|█████████▉| 29631/30000 [17:40:40<14:22, 2.34s/it] 99%|█████████▉| 29632/30000 [17:40:41<12:10, 1.99s/it] 99%|█████████▉| 29633/30000 [17:40:42<10:39, 1.74s/it] 99%|█████████▉| 29634/30000 [17:40:43<09:37, 1.58s/it] 99%|█████████▉| 29635/30000 [17:40:44<08:50, 1.45s/it] 99%|█████████▉| 29636/30000 [17:40:46<08:21, 1.38s/it] 99%|█████████▉| 29637/30000 [17:40:47<08:01, 1.33s/it] 99%|█████████▉| 29638/30000 [17:40:48<07:49, 1.30s/it] 99%|█████████▉| 29639/30000 [17:40:53<13:43, 2.28s/it] 99%|█████████▉| 29640/30000 [17:40:54<11:42, 1.95s/it] 99%|█████████▉| 29640/30000 [17:40:54<11:42, 1.95s/it] 99%|█████████▉| 29641/30000 [17:40:55<10:21, 1.73s/it] 99%|█████████▉| 29642/30000 [17:40:56<09:19, 1.56s/it] 99%|█████████▉| 29643/30000 [17:40:57<08:35, 1.44s/it] 99%|█████████▉| 29644/30000 [17:40:59<08:06, 1.37s/it] 99%|█████████▉| 29645/30000 [17:41:00<07:44, 1.31s/it] 99%|█████████▉| 29646/30000 [17:41:01<07:29, 1.27s/it] 99%|█████████▉| 29647/30000 [17:41:06<13:39, 2.32s/it] 99%|█████████▉| 29648/30000 [17:41:07<11:38, 1.98s/it] 99%|█████████▉| 29649/30000 [17:41:08<10:11, 1.74s/it] 99%|█████████▉| 29650/30000 [17:41:09<09:10, 1.57s/it] 99%|█████████▉| 29650/30000 [17:41:09<09:10, 1.57s/it] 99%|█████████▉| 29651/30000 [17:41:10<08:27, 1.45s/it] 99%|█████████▉| 29652/30000 [17:41:12<08:05, 1.39s/it] 99%|█████████▉| 29653/30000 [17:41:13<08:06, 1.40s/it] 99%|█████████▉| 29654/30000 [17:41:14<07:40, 1.33s/it] 99%|█████████▉| 29655/30000 [17:41:19<12:59, 2.26s/it] 99%|█████████▉| 29656/30000 [17:41:20<11:06, 1.94s/it] 99%|█████████▉| 29657/30000 [17:41:21<09:49, 1.72s/it] 99%|█████████▉| 29658/30000 [17:41:22<08:50, 1.55s/it] 99%|█████████▉| 29659/30000 [17:41:23<08:10, 1.44s/it] 99%|█████████▉| 29660/30000 [17:41:25<07:44, 1.37s/it] 99%|█████████▉| 29660/30000 [17:41:25<07:44, 1.37s/it] 99%|█████████▉| 29661/30000 [17:41:26<07:22, 1.30s/it] 99%|█████████▉| 29662/30000 [17:41:27<07:08, 1.27s/it] 99%|█████████▉| 29663/30000 [17:41:32<13:14, 2.36s/it] 99%|█████████▉| 29664/30000 [17:41:33<11:15, 2.01s/it] 99%|█████████▉| 29665/30000 [17:41:34<09:50, 1.76s/it] 99%|█████████▉| 29666/30000 [17:41:35<08:50, 1.59s/it] 99%|█████████▉| 29667/30000 [17:41:37<08:06, 1.46s/it] 99%|█████████▉| 29668/30000 [17:41:38<07:36, 1.37s/it] 99%|█████████▉| 29669/30000 [17:41:39<07:15, 1.32s/it] 99%|█████████▉| 29670/30000 [17:41:40<07:00, 1.27s/it] 99%|█████████▉| 29670/30000 [17:41:40<07:00, 1.27s/it] 99%|█████████▉| 29671/30000 [17:41:45<13:15, 2.42s/it] 99%|█████████▉| 29672/30000 [17:41:46<11:12, 2.05s/it] 99%|█████████▉| 29673/30000 [17:41:48<09:44, 1.79s/it] 99%|█████████▉| 29674/30000 [17:41:49<08:43, 1.61s/it] 99%|█████████▉| 29675/30000 [17:41:50<08:00, 1.48s/it] 99%|█████████▉| 29676/30000 [17:41:51<07:28, 1.39s/it] 99%|█████████▉| 29677/30000 [17:41:52<07:07, 1.32s/it] 99%|█████████▉| 29678/30000 [17:41:53<06:51, 1.28s/it] 99%|█████████▉| 29679/30000 [17:41:59<13:16, 2.48s/it] 99%|█████████▉| 29680/30000 [17:42:00<11:09, 2.09s/it] 99%|█████████▉| 29680/30000 [17:42:00<11:09, 2.09s/it] 99%|█████████▉| 29681/30000 [17:42:01<09:39, 1.82s/it] 99%|█████████▉| 29682/30000 [17:42:02<08:35, 1.62s/it] 99%|█████████▉| 29683/30000 [17:42:03<07:50, 1.48s/it] 99%|█████████▉| 29684/30000 [17:42:05<07:19, 1.39s/it] 99%|█████████▉| 29685/30000 [17:42:06<06:55, 1.32s/it] 99%|█████████▉| 29686/30000 [17:42:07<06:40, 1.27s/it] 99%|█████████▉| 29687/30000 [17:42:11<11:17, 2.16s/it] 99%|█████████▉| 29688/30000 [17:42:12<09:42, 1.87s/it] 99%|█████████▉| 29689/30000 [17:42:14<08:37, 1.66s/it] 99%|█████████▉| 29690/30000 [17:42:15<07:50, 1.52s/it] 99%|█████████▉| 29690/30000 [17:42:15<07:50, 1.52s/it] 99%|█████████▉| 29691/30000 [17:42:16<07:16, 1.41s/it] 99%|█████████▉| 29692/30000 [17:42:17<06:51, 1.34s/it] 99%|█████████▉| 29693/30000 [17:42:18<06:35, 1.29s/it] 99%|█████████▉| 29694/30000 [17:42:19<06:23, 1.25s/it] 99%|█████████▉| 29695/30000 [17:42:24<11:28, 2.26s/it] 99%|█████████▉| 29696/30000 [17:42:25<09:46, 1.93s/it] 99%|█████████▉| 29697/30000 [17:42:26<08:35, 1.70s/it] 99%|█████████▉| 29698/30000 [17:42:28<07:46, 1.55s/it] 99%|█████████▉| 29699/30000 [17:42:29<07:13, 1.44s/it] 99%|█████████▉| 29700/30000 [17:42:30<06:51, 1.37s/it] 99%|█████████▉| 29700/30000 [17:42:30<06:51, 1.37s/it] 99%|█████████▉| 29701/30000 [17:42:31<06:33, 1.32s/it] 99%|█████████▉| 29702/30000 [17:42:32<06:19, 1.27s/it] 99%|█████████▉| 29703/30000 [17:42:37<11:37, 2.35s/it] 99%|█████████▉| 29704/30000 [17:42:38<09:49, 1.99s/it] 99%|█████████▉| 29705/30000 [17:42:39<08:35, 1.75s/it] 99%|█████████▉| 29706/30000 [17:42:41<07:41, 1.57s/it] 99%|█████████▉| 29707/30000 [17:42:42<07:05, 1.45s/it] 99%|█████████▉| 29708/30000 [17:42:43<06:39, 1.37s/it] 99%|█████████▉| 29709/30000 [17:42:44<06:26, 1.33s/it] 99%|█████████▉| 29710/30000 [17:42:45<06:16, 1.30s/it] 99%|█████████▉| 29710/30000 [17:42:45<06:16, 1.30s/it] 99%|█████████▉| 29711/30000 [17:42:50<10:48, 2.24s/it] 99%|█████████▉| 29712/30000 [17:42:51<09:12, 1.92s/it] 99%|█████████▉| 29713/30000 [17:42:52<08:08, 1.70s/it] 99%|█████████▉| 29714/30000 [17:42:53<07:21, 1.54s/it] 99%|█████████▉| 29715/30000 [17:42:55<06:47, 1.43s/it] 99%|█████████▉| 29716/30000 [17:42:56<06:24, 1.35s/it] 99%|█████████▉| 29717/30000 [17:42:57<06:06, 1.30s/it] 99%|█████████▉| 29718/30000 [17:42:58<05:53, 1.25s/it] 99%|█████████▉| 29719/30000 [17:43:03<10:23, 2.22s/it] 99%|█████████▉| 29720/30000 [17:43:04<08:54, 1.91s/it] 99%|█████████▉| 29720/30000 [17:43:04<08:54, 1.91s/it] 99%|█████████▉| 29721/30000 [17:43:05<07:51, 1.69s/it] 99%|█████████▉| 29722/30000 [17:43:06<07:08, 1.54s/it] 99%|█████████▉| 29723/30000 [17:43:07<06:36, 1.43s/it] 99%|█████████▉| 29724/30000 [17:43:08<06:12, 1.35s/it] 99%|█████████▉| 29725/30000 [17:43:10<05:55, 1.29s/it] 99%|█████████▉| 29726/30000 [17:43:11<05:43, 1.25s/it] 99%|█████████▉| 29727/30000 [17:43:15<10:06, 2.22s/it] 99%|█████████▉| 29728/30000 [17:43:16<08:38, 1.91s/it] 99%|█████████▉| 29729/30000 [17:43:18<07:36, 1.69s/it] 99%|█████████▉| 29730/30000 [17:43:19<06:54, 1.53s/it] 99%|█████████▉| 29730/30000 [17:43:19<06:54, 1.53s/it] 99%|█████████▉| 29731/30000 [17:43:20<06:23, 1.43s/it] 99%|█████████▉| 29732/30000 [17:43:21<06:03, 1.35s/it] 99%|█████████▉| 29733/30000 [17:43:22<05:47, 1.30s/it] 99%|█████████▉| 29734/30000 [17:43:23<05:33, 1.26s/it] 99%|█████████▉| 29735/30000 [17:43:28<09:40, 2.19s/it] 99%|█████████▉| 29736/30000 [17:43:29<08:19, 1.89s/it] 99%|█████████▉| 29737/30000 [17:43:30<07:20, 1.68s/it] 99%|█████████▉| 29738/30000 [17:43:31<06:39, 1.52s/it] 99%|█████████▉| 29739/30000 [17:43:33<06:09, 1.42s/it] 99%|█████████▉| 29740/30000 [17:43:34<05:48, 1.34s/it] 99%|█████████▉| 29740/30000 [17:43:34<05:48, 1.34s/it] 99%|█████████▉| 29741/30000 [17:43:35<05:35, 1.29s/it] 99%|█████████▉| 29742/30000 [17:43:36<05:25, 1.26s/it] 99%|█████████▉| 29743/30000 [17:43:41<09:35, 2.24s/it] 99%|█████████▉| 29744/30000 [17:43:42<08:14, 1.93s/it] 99%|█████████▉| 29745/30000 [17:43:43<07:16, 1.71s/it] 99%|█████████▉| 29746/30000 [17:43:44<06:36, 1.56s/it] 99%|█████████▉| 29747/30000 [17:43:45<06:05, 1.44s/it] 99%|█████████▉| 29748/30000 [17:43:47<05:47, 1.38s/it] 99%|█████████▉| 29749/30000 [17:43:48<05:32, 1.32s/it] 99%|█████████▉| 29750/30000 [17:43:49<05:21, 1.29s/it] 99%|█████████▉| 29750/30000 [17:43:49<05:21, 1.29s/it] 99%|█████████▉| 29751/30000 [17:43:53<08:32, 2.06s/it] 99%|█████████▉| 29752/30000 [17:43:54<07:22, 1.79s/it] 99%|█████████▉| 29753/30000 [17:43:55<06:37, 1.61s/it] 99%|█████████▉| 29754/30000 [17:43:56<06:04, 1.48s/it] 99%|█████████▉| 29755/30000 [17:43:58<05:39, 1.39s/it] 99%|█████████▉| 29756/30000 [17:43:59<05:23, 1.33s/it] 99%|█████████▉| 29757/30000 [17:44:00<05:10, 1.28s/it] 99%|█████████▉| 29758/30000 [17:44:01<05:01, 1.24s/it] 99%|█████████▉| 29759/30000 [17:44:05<08:31, 2.12s/it] 99%|█████████▉| 29760/30000 [17:44:06<07:20, 1.84s/it] 99%|█████████▉| 29760/30000 [17:44:06<07:20, 1.84s/it] 99%|█████████▉| 29761/30000 [17:44:08<06:30, 1.64s/it] 99%|█████████▉| 29762/30000 [17:44:09<05:57, 1.50s/it] 99%|█████████▉| 29763/30000 [17:44:10<05:31, 1.40s/it] 99%|█████████▉| 29764/30000 [17:44:11<05:13, 1.33s/it] 99%|█████████▉| 29765/30000 [17:44:12<05:02, 1.29s/it] 99%|█████████▉| 29766/30000 [17:44:13<04:51, 1.25s/it] 99%|█████████▉| 29767/30000 [17:44:18<08:44, 2.25s/it] 99%|█████████▉| 29768/30000 [17:44:19<07:26, 1.92s/it] 99%|█████████▉| 29769/30000 [17:44:20<06:35, 1.71s/it] 99%|█████████▉| 29770/30000 [17:44:22<06:04, 1.58s/it] 99%|█████████▉| 29770/30000 [17:44:22<06:04, 1.58s/it] 99%|█████████▉| 29771/30000 [17:44:23<05:34, 1.46s/it] 99%|█████████▉| 29772/30000 [17:44:24<05:13, 1.37s/it] 99%|█████████▉| 29773/30000 [17:44:25<05:03, 1.34s/it] 99%|█████████▉| 29774/30000 [17:44:26<04:51, 1.29s/it] 99%|█████████▉| 29775/30000 [17:44:31<08:07, 2.17s/it] 99%|█████████▉| 29776/30000 [17:44:32<06:58, 1.87s/it] 99%|█████████▉| 29777/30000 [17:44:33<06:09, 1.66s/it] 99%|█████████▉| 29778/30000 [17:44:34<05:36, 1.52s/it] 99%|█████████▉| 29779/30000 [17:44:35<05:11, 1.41s/it] 99%|█████████▉| 29780/30000 [17:44:37<04:54, 1.34s/it] 99%|█████████▉| 29780/30000 [17:44:37<04:54, 1.34s/it] 99%|█████████▉| 29781/30000 [17:44:38<04:41, 1.29s/it] 99%|█████████▉| 29782/30000 [17:44:39<04:32, 1.25s/it] 99%|█████████▉| 29783/30000 [17:44:43<07:54, 2.19s/it] 99%|█████████▉| 29784/30000 [17:44:44<06:46, 1.88s/it] 99%|█████████▉| 29785/30000 [17:44:46<06:00, 1.68s/it] 99%|█████████▉| 29786/30000 [17:44:47<05:28, 1.54s/it] 99%|█████████▉| 29787/30000 [17:44:48<05:04, 1.43s/it] 99%|█████████▉| 29788/30000 [17:44:49<04:47, 1.35s/it] 99%|█████████▉| 29789/30000 [17:44:50<04:33, 1.30s/it] 99%|█████████▉| 29790/30000 [17:44:52<04:23, 1.26s/it] 99%|█████████▉| 29790/30000 [17:44:52<04:23, 1.26s/it] 99%|█████████▉| 29791/30000 [17:44:56<07:31, 2.16s/it] 99%|█████████▉| 29792/30000 [17:44:57<06:27, 1.86s/it] 99%|█████████▉| 29793/30000 [17:44:58<05:42, 1.65s/it] 99%|█████████▉| 29794/30000 [17:44:59<05:10, 1.51s/it] 99%|█████████▉| 29795/30000 [17:45:01<04:51, 1.42s/it] 99%|█████████▉| 29796/30000 [17:45:02<04:35, 1.35s/it] 99%|█████████▉| 29797/30000 [17:45:03<04:24, 1.30s/it] 99%|█████████▉| 29798/30000 [17:45:04<04:14, 1.26s/it] 99%|█████████▉| 29799/30000 [17:45:09<07:32, 2.25s/it] 99%|█████████▉| 29800/30000 [17:45:10<06:24, 1.92s/it] +{'loss': 0.0129, 'grad_norm': 0.05855482444167137, 'learning_rate': 3.290496143740829e-08, 'flow_matching_loss': 0.012910888250917196, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.55} +{'loss': 0.0157, 'grad_norm': 0.06846615672111511, 'learning_rate': 3.18194483608103e-08, 'flow_matching_loss': 0.01568738454952836, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.56} +{'loss': 0.0122, 'grad_norm': 0.06365924328565598, 'learning_rate': 3.075212303498942e-08, 'flow_matching_loss': 0.012224457133561373, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.56} +{'loss': 0.013, 'grad_norm': 0.06471985578536987, 'learning_rate': 2.9702986756846574e-08, 'flow_matching_loss': 0.012991876155138016, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.56} +{'loss': 0.0104, 'grad_norm': 0.07286480069160461, 'learning_rate': 2.867204080118202e-08, 'flow_matching_loss': 0.010391318844631315, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.57} +{'loss': 0.0138, 'grad_norm': 0.09886623173952103, 'learning_rate': 2.7659286420693708e-08, 'flow_matching_loss': 0.013837077841162681, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.57} +{'loss': 0.013, 'grad_norm': 0.07786650210618973, 'learning_rate': 2.6664724845968934e-08, 'flow_matching_loss': 0.012983535882085562, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.57} +{'loss': 0.0132, 'grad_norm': 0.0669604241847992, 'learning_rate': 2.5688357285499343e-08, 'flow_matching_loss': 0.013203028403222562, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.57} +{'loss': 0.0138, 'grad_norm': 0.07313407957553864, 'learning_rate': 2.473018492565926e-08, 'flow_matching_loss': 0.01383969197049737, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.58} +{'loss': 0.0114, 'grad_norm': 0.07329302281141281, 'learning_rate': 2.3790208930720702e-08, 'flow_matching_loss': 0.011396007612347603, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.58} +{'loss': 0.0122, 'grad_norm': 0.07752939313650131, 'learning_rate': 2.2868430442843368e-08, 'flow_matching_loss': 0.012226414494216442, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.58} +{'loss': 0.0118, 'grad_norm': 0.06800425797700882, 'learning_rate': 2.196485058207298e-08, 'flow_matching_loss': 0.011755072697997093, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.59} +{'loss': 0.0122, 'grad_norm': 0.07006639242172241, 'learning_rate': 2.1079470446346282e-08, 'flow_matching_loss': 0.012176356464624404, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.59} +{'loss': 0.0139, 'grad_norm': 0.058815620839595795, 'learning_rate': 2.0212291111482707e-08, 'flow_matching_loss': 0.013872453290969134, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.59} +{'loss': 0.0128, 'grad_norm': 0.0732850432395935, 'learning_rate': 1.9363313631187707e-08, 'flow_matching_loss': 0.012820127420127392, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.59} +{'loss': 0.0146, 'grad_norm': 0.07404571771621704, 'learning_rate': 1.853253903704777e-08, 'flow_matching_loss': 0.01455988809466362, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.6} +{'loss': 0.0116, 'grad_norm': 0.06814903765916824, 'learning_rate': 1.7719968338533733e-08, 'flow_matching_loss': 0.011584219802170993, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.6} +{'loss': 0.0121, 'grad_norm': 0.07298994064331055, 'learning_rate': 1.6925602522992467e-08, 'flow_matching_loss': 0.012126040551811457, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.6} +{'loss': 0.0122, 'grad_norm': 0.06856200844049454, 'learning_rate': 1.6149442555658533e-08, 'flow_matching_loss': 0.01219925144687295, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.61} +{'loss': 0.0139, 'grad_norm': 0.07330061495304108, 'learning_rate': 1.5391489379635858e-08, 'flow_matching_loss': 0.013906961306929588, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.61} +{'loss': 0.0128, 'grad_norm': 0.05756188929080963, 'learning_rate': 1.4651743915909399e-08, 'flow_matching_loss': 0.012819942738860845, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.61} +{'loss': 0.0151, 'grad_norm': 0.07487346231937408, 'learning_rate': 1.3930207063340139e-08, 'flow_matching_loss': 0.015142134949564933, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.62} +{'loss': 0.0123, 'grad_norm': 0.05687498673796654, 'learning_rate': 1.322687969866343e-08, 'flow_matching_loss': 0.012325819721445442, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.62} +{'loss': 0.0115, 'grad_norm': 0.07665029913187027, 'learning_rate': 1.2541762676488989e-08, 'flow_matching_loss': 0.011502622067928315, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.62} +{'loss': 0.0115, 'grad_norm': 0.07028832286596298, 'learning_rate': 1.1874856829297564e-08, 'flow_matching_loss': 0.011482204403728246, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.62} +{'loss': 0.0136, 'grad_norm': 0.0728665292263031, 'learning_rate': 1.1226162967442611e-08, 'flow_matching_loss': 0.013585858047008514, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.63} +{'loss': 0.0117, 'grad_norm': 0.08169899880886078, 'learning_rate': 1.059568187914861e-08, 'flow_matching_loss': 0.011659033037722111, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.63} +{'loss': 0.0124, 'grad_norm': 0.06996914744377136, 'learning_rate': 9.983414330511087e-09, 'flow_matching_loss': 0.012372460588812828, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.63} +{'loss': 0.0137, 'grad_norm': 0.08532674610614777, 'learning_rate': 9.389361065489932e-09, 'flow_matching_loss': 0.013657233864068984, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.64} +{'loss': 0.013, 'grad_norm': 0.08552336692810059, 'learning_rate': 8.81352280591774e-09, 'flow_matching_loss': 0.012985961511731148, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.64} +{'loss': 0.0111, 'grad_norm': 0.06328040361404419, 'learning_rate': 8.255900251491478e-09, 'flow_matching_loss': 0.011058480618521572, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.64} +{'loss': 0.0148, 'grad_norm': 0.0809445008635521, 'learning_rate': 7.716494079775816e-09, 'flow_matching_loss': 0.014783472754061222, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.64} +{'loss': 0.0125, 'grad_norm': 0.06661822646856308, 'learning_rate': 7.195304946199799e-09, 'flow_matching_loss': 0.012453899532556535, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.65} +{'loss': 0.0127, 'grad_norm': 0.06765200942754745, 'learning_rate': 6.692333484056845e-09, 'flow_matching_loss': 0.012731821648776532, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.65} +{'loss': 0.0137, 'grad_norm': 0.08420855551958084, 'learning_rate': 6.207580304506411e-09, 'flow_matching_loss': 0.013671270478516817, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.65} +{'loss': 0.013, 'grad_norm': 0.09778223931789398, 'learning_rate': 5.741045996567329e-09, 'flow_matching_loss': 0.013048023451119661, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.66} +{'loss': 0.0126, 'grad_norm': 0.08303812146186829, 'learning_rate': 5.2927311271244725e-09, 'flow_matching_loss': 0.012617102544754743, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.66} +{'loss': 0.0141, 'grad_norm': 0.0789141058921814, 'learning_rate': 4.8626362409204265e-09, 'flow_matching_loss': 0.014149058610200882, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.66} +{'loss': 0.0138, 'grad_norm': 0.08443633466959, 'learning_rate': 4.4507618605638126e-09, 'flow_matching_loss': 0.013815926481038332, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.66} +{'loss': 0.0143, 'grad_norm': 0.08866216987371445, 'learning_rate': 4.057108486519301e-09, 'flow_matching_loss': 0.014270150940865279, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.67} +{'loss': 0.0132, 'grad_norm': 0.07516106963157654, 'learning_rate': 3.6816765971126043e-09, 'flow_matching_loss': 0.01315668374300003, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.67} 99%|█████████▉| 29800/30000 [17:45:10<06:24, 1.92s/it] 99%|█████████▉| 29801/30000 [17:45:11<05:38, 1.70s/it] 99%|█████████▉| 29802/30000 [17:45:12<05:08, 1.56s/it] 99%|█████████▉| 29803/30000 [17:45:13<04:45, 1.45s/it] 99%|█████████▉| 29804/30000 [17:45:15<04:26, 1.36s/it] 99%|█████████▉| 29805/30000 [17:45:16<04:12, 1.30s/it] 99%|█████████▉| 29806/30000 [17:45:17<04:03, 1.26s/it] 99%|█████████▉| 29807/30000 [17:45:21<07:03, 2.20s/it] 99%|█████████▉| 29808/30000 [17:45:22<06:03, 1.89s/it] 99%|█████████▉| 29809/30000 [17:45:24<05:20, 1.68s/it] 99%|█████████▉| 29810/30000 [17:45:25<04:50, 1.53s/it] 99%|█████████▉| 29810/30000 [17:45:25<04:50, 1.53s/it] 99%|█████████▉| 29811/30000 [17:45:26<04:28, 1.42s/it] 99%|█████████▉| 29812/30000 [17:45:27<04:12, 1.34s/it] 99%|█████████▉| 29813/30000 [17:45:28<04:01, 1.29s/it] 99%|█████████▉| 29814/30000 [17:45:29<03:52, 1.25s/it] 99%|█████████▉| 29815/30000 [17:45:34<07:09, 2.32s/it] 99%|█████████▉| 29816/30000 [17:45:35<06:04, 1.98s/it] 99%|█████████▉| 29817/30000 [17:45:37<05:20, 1.75s/it] 99%|█████████▉| 29818/30000 [17:45:38<04:49, 1.59s/it] 99%|█████████▉| 29819/30000 [17:45:39<04:24, 1.46s/it] 99%|█████████▉| 29820/30000 [17:45:40<04:06, 1.37s/it] 99%|█████████▉| 29820/30000 [17:45:40<04:06, 1.37s/it] 99%|█████████▉| 29821/30000 [17:45:41<03:54, 1.31s/it] 99%|█████████▉| 29822/30000 [17:45:42<03:44, 1.26s/it] 99%|█████████▉| 29823/30000 [17:45:48<07:18, 2.48s/it] 99%|█████████▉| 29824/30000 [17:45:49<06:07, 2.09s/it] 99%|█████████▉| 29825/30000 [17:45:50<05:17, 1.81s/it] 99%|█████████▉| 29826/30000 [17:45:51<04:42, 1.62s/it] 99%|█████████▉| 29827/30000 [17:45:52<04:16, 1.49s/it] 99%|█████████▉| 29828/30000 [17:45:54<03:59, 1.39s/it] 99%|█████████▉| 29829/30000 [17:45:55<03:47, 1.33s/it] 99%|█████████▉| 29830/30000 [17:45:56<03:37, 1.28s/it] 99%|█████████▉| 29830/30000 [17:45:56<03:37, 1.28s/it] 99%|█████████▉| 29831/30000 [17:46:01<06:55, 2.46s/it] 99%|█████████▉| 29832/30000 [17:46:02<05:48, 2.07s/it] 99%|█████████▉| 29833/30000 [17:46:04<05:00, 1.80s/it] 99%|█████████▉| 29834/30000 [17:46:05<04:27, 1.61s/it] 99%|█████████▉| 29835/30000 [17:46:06<04:03, 1.48s/it] 99%|█████████▉| 29836/30000 [17:46:07<03:47, 1.38s/it] 99%|█████████▉| 29837/30000 [17:46:08<03:34, 1.32s/it] 99%|█████████▉| 29838/30000 [17:46:09<03:26, 1.27s/it] 99%|█████████▉| 29839/30000 [17:46:15<06:32, 2.44s/it] 99%|█████████▉| 29840/30000 [17:46:16<05:31, 2.07s/it] 99%|█████████▉| 29840/30000 [17:46:16<05:31, 2.07s/it] 99%|█████████▉| 29841/30000 [17:46:17<04:48, 1.82s/it] 99%|█████████▉| 29842/30000 [17:46:18<04:19, 1.64s/it] 99%|█████████▉| 29843/30000 [17:46:19<03:57, 1.51s/it] 99%|█████████▉| 29844/30000 [17:46:21<03:41, 1.42s/it] 99%|█████████▉| 29845/30000 [17:46:22<03:29, 1.35s/it] 99%|█████████▉| 29846/30000 [17:46:23<03:20, 1.30s/it] 99%|█████████▉| 29847/30000 [17:46:28<05:46, 2.26s/it] 99%|█████████▉| 29848/30000 [17:46:29<04:54, 1.93s/it] 99%|█████████▉| 29849/30000 [17:46:30<04:19, 1.72s/it] 100%|█████████▉| 29850/30000 [17:46:31<03:53, 1.56s/it] 100%|█████████▉| 29850/30000 [17:46:31<03:53, 1.56s/it] 100%|█████████▉| 29851/30000 [17:46:32<03:35, 1.45s/it] 100%|█████████▉| 29852/30000 [17:46:33<03:22, 1.37s/it] 100%|█████████▉| 29853/30000 [17:46:35<03:15, 1.33s/it] 100%|█████████▉| 29854/30000 [17:46:36<03:09, 1.30s/it] 100%|█████████▉| 29855/30000 [17:46:42<06:18, 2.61s/it] 100%|█████████▉| 29856/30000 [17:46:43<05:13, 2.18s/it] 100%|█████████▉| 29857/30000 [17:46:44<04:27, 1.87s/it] 100%|█████████▉| 29858/30000 [17:46:45<03:55, 1.66s/it] 100%|█████████▉| 29859/30000 [17:46:46<03:32, 1.51s/it] 100%|█████████▉| 29860/30000 [17:46:47<03:16, 1.41s/it] 100%|█████████▉| 29860/30000 [17:46:47<03:16, 1.41s/it] 100%|█████████▉| 29861/30000 [17:46:49<03:05, 1.33s/it] 100%|█████████▉| 29862/30000 [17:46:50<02:58, 1.29s/it] 100%|█████████▉| 29863/30000 [17:46:55<05:50, 2.56s/it] 100%|█████████▉| 29864/30000 [17:46:56<04:50, 2.14s/it] 100%|█████████▉| 29865/30000 [17:46:58<04:09, 1.85s/it] 100%|█████████▉| 29866/30000 [17:46:59<03:39, 1.64s/it] 100%|█████████▉| 29867/30000 [17:47:00<03:19, 1.50s/it] 100%|█████████▉| 29868/30000 [17:47:01<03:05, 1.40s/it] 100%|█████████▉| 29869/30000 [17:47:02<02:55, 1.34s/it] 100%|█████████▉| 29870/30000 [17:47:03<02:46, 1.28s/it] 100%|█████████▉| 29870/30000 [17:47:03<02:46, 1.28s/it] 100%|█████████▉| 29871/30000 [17:47:08<04:56, 2.30s/it] 100%|█████████▉| 29872/30000 [17:47:09<04:10, 1.96s/it] 100%|█████████▉| 29873/30000 [17:47:10<03:39, 1.73s/it] 100%|█████████▉| 29874/30000 [17:47:12<03:16, 1.56s/it] 100%|█████████▉| 29875/30000 [17:47:13<03:00, 1.44s/it] 100%|█████████▉| 29876/30000 [17:47:14<02:48, 1.36s/it] 100%|█████████▉| 29877/30000 [17:47:15<02:40, 1.31s/it] 100%|█████████▉| 29878/30000 [17:47:16<02:35, 1.27s/it] 100%|█████████▉| 29879/30000 [17:47:21<04:27, 2.21s/it] 100%|█████████▉| 29880/30000 [17:47:22<03:48, 1.90s/it] 100%|█████████▉| 29880/30000 [17:47:22<03:48, 1.90s/it] 100%|█████████▉| 29881/30000 [17:47:23<03:20, 1.69s/it] 100%|█████████▉| 29882/30000 [17:47:24<03:00, 1.53s/it] 100%|█████████▉| 29883/30000 [17:47:25<02:46, 1.43s/it] 100%|█████████▉| 29884/30000 [17:47:27<02:37, 1.36s/it] 100%|█████████▉| 29885/30000 [17:47:28<02:30, 1.31s/it] 100%|█████████▉| 29886/30000 [17:47:29<02:24, 1.27s/it] 100%|█████████▉| 29887/30000 [17:47:33<04:00, 2.13s/it] 100%|█████████▉| 29888/30000 [17:47:34<03:26, 1.85s/it] 100%|█████████▉| 29889/30000 [17:47:36<03:04, 1.66s/it] 100%|█████████▉| 29890/30000 [17:47:37<02:47, 1.53s/it] 100%|█████████▉| 29890/30000 [17:47:37<02:47, 1.53s/it] 100%|█████████▉| 29891/30000 [17:47:38<02:36, 1.43s/it] 100%|█████████▉| 29892/30000 [17:47:39<02:28, 1.37s/it] 100%|█████████▉| 29893/30000 [17:47:40<02:21, 1.33s/it] 100%|█████████▉| 29894/30000 [17:47:42<02:15, 1.28s/it] 100%|█████████▉| 29895/30000 [17:47:46<03:49, 2.19s/it] 100%|█████████▉| 29896/30000 [17:47:47<03:16, 1.89s/it] 100%|█████████▉| 29897/30000 [17:47:48<02:52, 1.67s/it] 100%|█████████▉| 29898/30000 [17:47:49<02:35, 1.53s/it] 100%|█████████▉| 29899/30000 [17:47:51<02:23, 1.42s/it] 100%|█████████▉| 29900/30000 [17:47:52<02:14, 1.35s/it] 100%|█████████▉| 29900/30000 [17:47:52<02:14, 1.35s/it] 100%|█████████▉| 29901/30000 [17:47:53<02:08, 1.29s/it] 100%|█████████▉| 29902/30000 [17:47:54<02:03, 1.26s/it] 100%|█████████▉| 29903/30000 [17:47:58<03:25, 2.12s/it] 100%|█████████▉| 29904/30000 [17:47:59<02:56, 1.84s/it] 100%|█████████▉| 29905/30000 [17:48:01<02:36, 1.64s/it] 100%|█████████▉| 29906/30000 [17:48:02<02:21, 1.50s/it] 100%|█████████▉| 29907/30000 [17:48:03<02:10, 1.40s/it] 100%|█████████▉| 29908/30000 [17:48:04<02:03, 1.34s/it] 100%|█████████▉| 29909/30000 [17:48:05<01:57, 1.29s/it] 100%|█████████▉| 29910/30000 [17:48:07<01:52, 1.26s/it] 100%|█████████▉| 29910/30000 [17:48:07<01:52, 1.26s/it] 100%|█████████▉| 29911/30000 [17:48:11<03:27, 2.33s/it] 100%|█████████▉| 29912/30000 [17:48:13<02:54, 1.99s/it] 100%|█████████▉| 29913/30000 [17:48:14<02:32, 1.75s/it] 100%|█████████▉| 29914/30000 [17:48:15<02:15, 1.57s/it] 100%|█████████▉| 29915/30000 [17:48:16<02:03, 1.45s/it] 100%|█████████▉| 29916/30000 [17:48:17<01:54, 1.37s/it] 100%|█████████▉| 29917/30000 [17:48:18<01:48, 1.30s/it] 100%|█████████▉| 29918/30000 [17:48:20<01:43, 1.26s/it] 100%|█████████▉| 29919/30000 [17:48:24<03:05, 2.29s/it] 100%|█████████▉| 29920/30000 [17:48:25<02:36, 1.96s/it] 100%|█████████▉| 29920/30000 [17:48:25<02:36, 1.96s/it] 100%|█████████▉| 29921/30000 [17:48:27<02:15, 1.72s/it] 100%|█████████▉| 29922/30000 [17:48:28<02:01, 1.56s/it] 100%|█████████▉| 29923/30000 [17:48:29<01:50, 1.44s/it] 100%|█████████▉| 29924/30000 [17:48:30<01:43, 1.36s/it] 100%|█████████▉| 29925/30000 [17:48:31<01:38, 1.31s/it] 100%|█████████▉| 29926/30000 [17:48:32<01:33, 1.26s/it] 100%|█████████▉| 29927/30000 [17:48:37<02:44, 2.26s/it] 100%|█████████▉| 29928/30000 [17:48:38<02:19, 1.94s/it] 100%|█████████▉| 29929/30000 [17:48:39<02:01, 1.71s/it] 100%|█████████▉| 29930/30000 [17:48:41<01:47, 1.54s/it] 100%|█████████▉| 29930/30000 [17:48:41<01:47, 1.54s/it] 100%|█████████▉| 29931/30000 [17:48:42<01:39, 1.44s/it] 100%|█████████▉| 29932/30000 [17:48:43<01:32, 1.36s/it] 100%|█████████▉| 29933/30000 [17:48:44<01:29, 1.33s/it] 100%|█████████▉| 29934/30000 [17:48:45<01:24, 1.28s/it] 100%|█████████▉| 29935/30000 [17:48:50<02:31, 2.33s/it] 100%|█████████▉| 29936/30000 [17:48:51<02:06, 1.98s/it] 100%|█████████▉| 29937/30000 [17:48:53<01:49, 1.74s/it] 100%|█████████▉| 29938/30000 [17:48:54<01:38, 1.58s/it] 100%|█████████▉| 29939/30000 [17:48:55<01:28, 1.46s/it] 100%|█████████▉| 29940/30000 [17:48:56<01:22, 1.38s/it] 100%|█████████▉| 29940/30000 [17:48:56<01:22, 1.38s/it] 100%|█████████▉| 29941/30000 [17:48:57<01:18, 1.32s/it] 100%|█████████▉| 29942/30000 [17:48:58<01:14, 1.28s/it] 100%|█████████▉| 29943/30000 [17:49:03<02:01, 2.14s/it] 100%|█████████▉| 29944/30000 [17:49:04<01:43, 1.85s/it] 100%|█████████▉| 29945/30000 [17:49:05<01:31, 1.66s/it] 100%|█████████▉| 29946/30000 [17:49:06<01:21, 1.51s/it] 100%|█████████▉| 29947/30000 [17:49:07<01:15, 1.42s/it] 100%|█████████▉| 29948/30000 [17:49:09<01:10, 1.35s/it] 100%|█████████▉| 29949/30000 [17:49:10<01:06, 1.30s/it] 100%|█████████▉| 29950/30000 [17:49:11<01:02, 1.26s/it] 100%|█████████▉| 29950/30000 [17:49:11<01:02, 1.26s/it] 100%|█████████▉| 29951/30000 [17:49:15<01:47, 2.20s/it] 100%|█████████▉| 29952/30000 [17:49:16<01:30, 1.89s/it] 100%|█████████▉| 29953/30000 [17:49:18<01:18, 1.67s/it] 100%|█████████▉| 29954/30000 [17:49:19<01:10, 1.53s/it] 100%|█████████▉| 29955/30000 [17:49:20<01:04, 1.43s/it] 100%|█████████▉| 29956/30000 [17:49:21<00:59, 1.36s/it] 100%|█████████▉| 29957/30000 [17:49:22<00:55, 1.30s/it] 100%|█████████▉| 29958/30000 [17:49:24<00:53, 1.26s/it] 100%|█████████▉| 29959/30000 [17:49:27<01:22, 2.02s/it] 100%|█████████▉| 29960/30000 [17:49:29<01:10, 1.77s/it] 100%|█████████▉| 29960/30000 [17:49:29<01:10, 1.77s/it] 100%|█████████▉| 29961/30000 [17:49:30<01:02, 1.60s/it] 100%|█████████▉| 29962/30000 [17:49:31<00:56, 1.48s/it] 100%|█████████▉| 29963/30000 [17:49:32<00:51, 1.40s/it] 100%|█████████▉| 29964/30000 [17:49:33<00:48, 1.33s/it] 100%|█████████▉| 29965/30000 [17:49:34<00:44, 1.29s/it] 100%|█████████▉| 29966/30000 [17:49:36<00:42, 1.25s/it] 100%|█████████▉| 29967/30000 [17:49:40<01:15, 2.29s/it] 100%|█████████▉| 29968/30000 [17:49:42<01:02, 1.95s/it] 100%|█████████▉| 29969/30000 [17:49:43<00:52, 1.71s/it] 100%|█████████▉| 29970/30000 [17:49:44<00:46, 1.56s/it] 100%|█████████▉| 29970/30000 [17:49:44<00:46, 1.56s/it] 100%|█████████▉| 29971/30000 [17:49:45<00:42, 1.45s/it] 100%|█████████▉| 29972/30000 [17:49:46<00:38, 1.37s/it] 100%|█████████▉| 29973/30000 [17:49:47<00:35, 1.31s/it] 100%|█████████▉| 29974/30000 [17:49:49<00:32, 1.27s/it] 100%|█████████▉| 29975/30000 [17:49:53<00:53, 2.14s/it] 100%|█████████▉| 29976/30000 [17:49:54<00:44, 1.85s/it] 100%|█████████▉| 29977/30000 [17:49:55<00:38, 1.66s/it] 100%|█████████▉| 29978/30000 [17:49:56<00:33, 1.52s/it] 100%|█████████▉| 29979/30000 [17:49:58<00:29, 1.41s/it] 100%|█████████▉| 29980/30000 [17:49:59<00:26, 1.34s/it] 100%|█████████▉| 29980/30000 [17:49:59<00:26, 1.34s/it] 100%|█████████▉| 29981/30000 [17:50:00<00:24, 1.29s/it] 100%|█████████▉| 29982/30000 [17:50:01<00:22, 1.25s/it] 100%|█████████▉| 29983/30000 [17:50:05<00:34, 2.06s/it] 100%|█████████▉| 29984/30000 [17:50:06<00:28, 1.79s/it] 100%|█████████▉| 29985/30000 [17:50:07<00:24, 1.60s/it] 100%|█████████▉| 29986/30000 [17:50:08<00:20, 1.47s/it] 100%|█████████▉| 29987/30000 [17:50:10<00:17, 1.38s/it] 100%|█████████▉| 29988/30000 [17:50:11<00:15, 1.31s/it] 100%|█████████▉| 29989/30000 [17:50:12<00:13, 1.27s/it] 100%|█████████▉| 29990/30000 [17:50:13<00:12, 1.24s/it] 100%|█████████▉| 29990/30000 [17:50:13<00:12, 1.24s/it] 100%|█████████▉| 29991/30000 [17:50:18<00:20, 2.32s/it] 100%|█████████▉| 29992/30000 [17:50:19<00:15, 1.98s/it] 100%|█████████▉| 29993/30000 [17:50:20<00:12, 1.74s/it] 100%|█████████▉| 29994/30000 [17:50:21<00:09, 1.57s/it] 100%|█████████▉| 29995/30000 [17:50:23<00:07, 1.45s/it] 100%|█████████▉| 29996/30000 [17:50:24<00:05, 1.36s/it] 100%|█████████▉| 29997/30000 [17:50:25<00:03, 1.30s/it] 100%|█████████▉| 29998/30000 [17:50:26<00:02, 1.26s/it] 100%|█████████▉| 29999/30000 [17:50:31<00:02, 2.36s/it] 100%|██████████| 30000/30000 [17:50:32<00:00, 2.00s/it] 100%|██████████| 30000/30000 [17:50:32<00:00, 2.00s/it] 100%|██████████| 30000/30000 [17:50:32<00:00, 2.00s/it] 100%|██████████| 30000/30000 [17:50:32<00:00, 2.14s/it] +wandb: updating run metadata +wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml +wandb: uploading output.log +wandb: uploading summary, console lines 3003-3003 +wandb: +wandb: Run history: +wandb: train/epoch ▁▁▁▂▂▂▂▂▂▂▃▃▃▃▃▃▃▃▃▃▄▄▄▅▅▆▆▆▆▇▇▇▇▇▇█████ +wandb: train/flow_matching_loss █▃▃▃▃▂▃▂▂▂▂▂▂▂▂▂▂▂▂▂▁▁▂▂▂▂▂▁▂▂▁▂▁▂▁▁▁▁▂▁ +wandb: train/global_step ▁▁▁▂▂▂▂▃▃▃▄▄▄▄▄▅▅▅▅▅▅▅▆▆▆▆▆▆▆▆▇▇▇▇▇▇▇███ +wandb: train/grad_norm █▅▆█▆▅▂▄▆▃▂▆█▃▂▅▃▃▂▃▃▄▂▃▃▂▁▄▄▄▁▃▃▂▃▃▄▁▂▃ +wandb: train/learning_rate ▂▄▄████████▇▇▇▇▇▇▆▅▅▄▄▄▃▃▃▃▃▃▃▂▂▂▂▂▁▁▁▁▁ +wandb: train/loss █▄▄▃▃▃▃▂▃▂▂▃▃▃▂▃▂▂▂▂▃▂▂▂▂▂▂▂▂▂▂▂▂▁▂▂▂▂▁▂ +wandb: train/mgd_loss ▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁ +wandb: train/mgd_token_mask_ratio_cfg ▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁ +wandb: +wandb: Run summary: +wandb: total_flos 0 +wandb: train/epoch 8.72854 +wandb: train/flow_matching_loss 0.0127 +wandb: train/global_step 30000 +wandb: train/grad_norm 0.06092 +wandb: train/learning_rate 0.0 +wandb: train/loss 0.0127 +wandb: train/mgd_loss 0 +wandb: train/mgd_token_mask_ratio_cfg 0.5 +wandb: train_loss 0.01563 +wandb: +3 ... +wandb: +wandb: 🚀 View run mgd_token_mask_ratio_0.5_phase3 at: https://wandb.ai/minje227_hyu-hanyang-university/huggingface/runs/0m3k76xv +wandb: ⭐️ View project at: https://wandb.ai/minje227_hyu-hanyang-university/huggingface +wandb: Synced 5 W&B file(s), 0 media file(s), 0 artifact file(s) and 0 other file(s) +wandb: Find logs at: ./wandb/run-20260603_014853-0m3k76xv/logs + +{'loss': 0.0126, 'grad_norm': 0.0706789493560791, 'learning_rate': 3.324466648528812e-09, 'flow_matching_loss': 0.012624203227460384, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.67} +{'loss': 0.0137, 'grad_norm': 0.06629828363656998, 'learning_rate': 2.9854790748123917e-09, 'flow_matching_loss': 0.013683277182281017, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.68} +{'loss': 0.0124, 'grad_norm': 0.06581724435091019, 'learning_rate': 2.664714287865522e-09, 'flow_matching_loss': 0.012416338268667459, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.68} +{'loss': 0.0126, 'grad_norm': 0.05887773260474205, 'learning_rate': 2.362172677446428e-09, 'flow_matching_loss': 0.012554429471492767, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.68} +{'loss': 0.0118, 'grad_norm': 0.0729188546538353, 'learning_rate': 2.0778546111727137e-09, 'flow_matching_loss': 0.011771690659224987, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.68} +{'loss': 0.0124, 'grad_norm': 0.08576650172472, 'learning_rate': 1.811760434516363e-09, 'flow_matching_loss': 0.012417888920754194, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.69} +{'loss': 0.0121, 'grad_norm': 0.06380762159824371, 'learning_rate': 1.5638904708087377e-09, 'flow_matching_loss': 0.012143871001899242, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.69} +{'loss': 0.0122, 'grad_norm': 0.054130036383867264, 'learning_rate': 1.3342450212355805e-09, 'flow_matching_loss': 0.01224596444517374, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.69} +{'loss': 0.0136, 'grad_norm': 0.09417291730642319, 'learning_rate': 1.122824364835351e-09, 'flow_matching_loss': 0.013569681253284217, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.7} +{'loss': 0.0122, 'grad_norm': 0.06119155511260033, 'learning_rate': 9.296287585042196e-10, 'flow_matching_loss': 0.012200099136680364, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.7} +{'loss': 0.012, 'grad_norm': 0.061947040259838104, 'learning_rate': 7.546584369960697e-10, 'flow_matching_loss': 0.012029681727290154, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.7} +{'loss': 0.0135, 'grad_norm': 0.07544105499982834, 'learning_rate': 5.97913612912504e-10, 'flow_matching_loss': 0.013541720062494277, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.71} +{'loss': 0.0118, 'grad_norm': 0.05005600303411484, 'learning_rate': 4.593944767161684e-10, 'flow_matching_loss': 0.011795561853796243, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.71} +{'loss': 0.0138, 'grad_norm': 0.08500850200653076, 'learning_rate': 3.391011967207591e-10, 'flow_matching_loss': 0.013798499293625355, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.71} +{'loss': 0.0118, 'grad_norm': 0.07519569247961044, 'learning_rate': 2.370339190910231e-10, 'flow_matching_loss': 0.011824683379381895, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.71} +{'loss': 0.0126, 'grad_norm': 0.062107235193252563, 'learning_rate': 1.531927678510847e-10, 'flow_matching_loss': 0.012577164452522993, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.72} +{'loss': 0.0135, 'grad_norm': 0.055698636919260025, 'learning_rate': 8.757784487611886e-11, 'flow_matching_loss': 0.013511533383280039, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.72} +{'loss': 0.0131, 'grad_norm': 0.0704827532172203, 'learning_rate': 4.018922989235119e-11, 'flow_matching_loss': 0.013075003866106272, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.72} +{'loss': 0.0128, 'grad_norm': 0.06343361735343933, 'learning_rate': 1.1026980482053972e-11, 'flow_matching_loss': 0.012844633590430022, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.73} +{'loss': 0.0127, 'grad_norm': 0.0609227679669857, 'learning_rate': 9.113208188082922e-14, 'flow_matching_loss': 0.012701811827719211, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.5, 'epoch': 8.73} +{'train_runtime': 64240.4931, 'train_samples_per_second': 59.775, 'train_steps_per_second': 0.467, 'train_loss': 0.015626614911605913, 'epoch': 8.73} +Saved final phase model to: /home/ext_minje/groot_robocasa/robocasa_v2/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.5/phase3 +███████▉| 29674/30000 [11:08:13<07:12, 1.33s/it] 99%|█████████▉| 29675/30000 [11:08:14<07:03, 1.30s/it] 99%|█████████▉| 29676/30000 [11:08:15<07:08, 1.32s/it] 99%|█████████▉| 29677/30000 [11:08:16<07:00, 1.30s/it] 99%|█████████▉| 29678/30000 [11:08:18<07:08, 1.33s/it] 99%|█████████▉| 29679/30000 [11:08:19<06:57, 1.30s/it] 99%|█████████▉| 29680/30000 [11:08:20<07:06, 1.33s/it] {'loss': 0.0149, 'grad_norm': 0.0972728505730629, 'learning_rate': 9.389361065489932e-09, 'flow_matching_loss': 0.014948657248169184, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.64} + 99%|█████████▉| 29680/30000 [11:08:20<07:06, 1.33s/it] 99%|█████████▉| 29681/30000 [11:08:22<07:08, 1.34s/it] 99%|█████████▉| 29682/30000 [11:08:23<07:08, 1.35s/it] 99%|█████████▉| 29683/30000 [11:08:25<07:11, 1.36s/it] 99%|█████████▉| 29684/30000 [11:08:26<06:59, 1.33s/it] 99%|█████████▉| 29685/30000 [11:08:27<07:03, 1.35s/it] 99%|█████████▉| 29686/30000 [11:08:29<07:02, 1.34s/it] 99%|█████████▉| 29687/30000 [11:08:30<06:52, 1.32s/it] 99%|█████████▉| 29688/30000 [11:08:31<06:54, 1.33s/it] 99%|█████████▉| 29689/30000 [11:08:32<06:47, 1.31s/it] 99%|█████████▉| 29690/30000 [11:08:34<06:53, 1.33s/it] {'loss': 0.0151, 'grad_norm': 0.07631339132785797, 'learning_rate': 8.81352280591774e-09, 'flow_matching_loss': 0.015118073113262654, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.64} + 99%|█████████▉| 29690/30000 [11:08:34<06:53, 1.33s/it] 99%|█████████▉| 29691/30000 [11:08:35<06:44, 1.31s/it] 99%|█████████▉| 29692/30000 [11:08:36<06:50, 1.33s/it] 99%|█████████▉| 29693/30000 [11:08:38<06:56, 1.36s/it] 99%|█████████▉| 29694/30000 [11:08:39<06:45, 1.33s/it] 99%|█████████▉| 29695/30000 [11:08:40<06:48, 1.34s/it] 99%|█████████▉| 29696/30000 [11:08:42<06:37, 1.31s/it] 99%|█████████▉| 29697/30000 [11:08:43<06:42, 1.33s/it] 99%|█████████▉| 29698/30000 [11:08:44<06:35, 1.31s/it] 99%|█████████▉| 29699/30000 [11:08:46<06:40, 1.33s/it] 99%|█████████▉| 29700/30000 [11:08:47<06:42, 1.34s/it] {'loss': 0.0162, 'grad_norm': 0.09784317761659622, 'learning_rate': 8.255900251491478e-09, 'flow_matching_loss': 0.016150859044864775, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.64} + 99%|█████████▉| 29700/30000 [11:08:47<06:42, 1.34s/it] 99%|█████████▉| 29701/30000 [11:08:48<06:35, 1.32s/it] 99%|█████████▉| 29702/30000 [11:08:50<06:38, 1.34s/it] 99%|█████████▉| 29703/30000 [11:08:51<06:27, 1.30s/it] 99%|█████████▉| 29704/30000 [11:08:52<06:33, 1.33s/it] 99%|█████████▉| 29705/30000 [11:08:54<06:34, 1.34s/it] 99%|█████████▉| 29706/30000 [11:08:55<06:34, 1.34s/it] 99%|█████████▉| 29707/30000 [11:08:56<06:35, 1.35s/it] 99%|█████████▉| 29708/30000 [11:08:58<06:29, 1.33s/it] 99%|█████████▉| 29709/30000 [11:08:59<06:31, 1.35s/it] 99%|█████████▉| 29710/30000 [11:09:00<06:18, 1.31s/it] {'loss': 0.0149, 'grad_norm': 0.09401462227106094, 'learning_rate': 7.716494079775816e-09, 'flow_matching_loss': 0.014934702683240176, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.64} + 99%|█████████▉| 29710/30000 [11:09:00<06:18, 1.31s/it] 99%|█████████▉| 29711/30000 [11:09:02<06:25, 1.33s/it] 99%|█████████▉| 29712/30000 [11:09:03<06:28, 1.35s/it] 99%|█████████▉| 29713/30000 [11:09:04<06:19, 1.32s/it] 99%|█████████▉| 29714/30000 [11:09:06<06:25, 1.35s/it] 99%|█████████▉| 29715/30000 [11:09:07<06:13, 1.31s/it] 99%|█████████▉| 29716/30000 [11:09:08<06:22, 1.35s/it] 99%|█████████▉| 29717/30000 [11:09:10<06:19, 1.34s/it] 99%|█████████▉| 29718/30000 [11:09:11<06:23, 1.36s/it] 99%|█████████▉| 29719/30000 [11:09:13<06:21, 1.36s/it] 99%|█████████▉| 29720/30000 [11:09:14<06:15, 1.34s/it] {'loss': 0.0132, 'grad_norm': 0.08865852653980255, 'learning_rate': 7.195304946199799e-09, 'flow_matching_loss': 0.013166244048625231, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.65} + 99%|█████████▉| 29720/30000 [11:09:14<06:15, 1.34s/it] 99%|█████████▉| 29721/30000 [11:09:15<06:19, 1.36s/it] 99%|█████████▉| 29722/30000 [11:09:16<06:07, 1.32s/it] 99%|█████████▉| 29723/30000 [11:09:18<06:11, 1.34s/it] 99%|█████████▉| 29724/30000 [11:09:19<06:13, 1.35s/it] 99%|█████████▉| 29725/30000 [11:09:20<06:05, 1.33s/it] 99%|█████████▉| 29726/30000 [11:09:22<06:08, 1.35s/it] 99%|█████████▉| 29727/30000 [11:09:23<05:59, 1.32s/it] 99%|█████████▉| 29728/30000 [11:09:25<06:03, 1.34s/it] 99%|█████████▉| 29729/30000 [11:09:26<06:02, 1.34s/it] 99%|█████████▉| 29730/30000 [11:09:27<06:03, 1.35s/it] {'loss': 0.0138, 'grad_norm': 0.08763492107391357, 'learning_rate': 6.692333484056845e-09, 'flow_matching_loss': 0.013800548017024994, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.65} + 99%|█████████▉| 29730/30000 [11:09:27<06:03, 1.35s/it] 99%|█████████▉| 29731/30000 [11:09:29<06:04, 1.36s/it] 99%|█████████▉| 29732/30000 [11:09:30<05:56, 1.33s/it] 99%|█████████▉| 29733/30000 [11:09:31<05:58, 1.34s/it] 99%|█████████▉| 29734/30000 [11:09:32<05:48, 1.31s/it] 99%|█████████▉| 29735/30000 [11:09:34<05:54, 1.34s/it] 99%|█████████▉| 29736/30000 [11:09:35<05:54, 1.34s/it] 99%|█████████▉| 29737/30000 [11:09:37<05:50, 1.33s/it] 99%|█████████▉| 29738/30000 [11:09:38<05:55, 1.36s/it] 99%|█████████▉| 29739/30000 [11:09:39<05:44, 1.32s/it] 99%|█████████▉| 29740/30000 [11:09:41<05:48, 1.34s/it] {'loss': 0.013, 'grad_norm': 0.09982739388942719, 'learning_rate': 6.207580304506411e-09, 'flow_matching_loss': 0.013015625718981027, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.65} + 99%|█████████▉| 29740/30000 [11:09:41<05:48, 1.34s/it] 99%|█████████▉| 29741/30000 [11:09:42<05:48, 1.34s/it] 99%|█████████▉| 29742/30000 [11:09:43<05:50, 1.36s/it] 99%|█████████▉| 29743/30000 [11:09:45<05:48, 1.36s/it] 99%|█████████▉| 29744/30000 [11:09:46<05:40, 1.33s/it] 99%|█████████▉| 29745/30000 [11:09:47<05:41, 1.34s/it] 99%|█████████▉| 29746/30000 [11:09:49<05:32, 1.31s/it] 99%|█████████▉| 29747/30000 [11:09:50<05:34, 1.32s/it] 99%|█████████▉| 29748/30000 [11:09:51<05:35, 1.33s/it] 99%|█████████▉| 29749/30000 [11:09:53<05:38, 1.35s/it] 99%|█████████▉| 29750/30000 [11:09:54<05:37, 1.35s/it] {'loss': 0.0158, 'grad_norm': 0.11751998960971832, 'learning_rate': 5.741045996567329e-09, 'flow_matching_loss': 0.01583666019141674, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.66} + 99%|█████████▉| 29750/30000 [11:09:54<05:37, 1.35s/it] 99%|█████████▉| 29751/30000 [11:09:55<05:31, 1.33s/it] 99%|█████████▉| 29752/30000 [11:09:57<05:33, 1.35s/it] 99%|█████████▉| 29753/30000 [11:09:58<05:35, 1.36s/it] 99%|█████████▉| 29754/30000 [11:09:59<05:37, 1.37s/it] 99%|█████████▉| 29755/30000 [11:10:01<05:35, 1.37s/it] 99%|█████████▉| 29756/30000 [11:10:02<05:23, 1.33s/it] 99%|█████████▉| 29757/30000 [11:10:03<05:25, 1.34s/it] 99%|█████████▉| 29758/30000 [11:10:05<05:17, 1.31s/it] 99%|█████████▉| 29759/30000 [11:10:06<05:19, 1.33s/it] 99%|█████████▉| 29760/30000 [11:10:07<05:20, 1.34s/it] {'loss': 0.0146, 'grad_norm': 0.08662352710962296, 'learning_rate': 5.2927311271244725e-09, 'flow_matching_loss': 0.014649682585150004, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.66} + 99%|█████████▉| 29760/30000 [11:10:07<05:20, 1.34s/it] 99%|█████████▉| 29761/30000 [11:10:09<05:26, 1.37s/it] 99%|█████████▉| 29762/30000 [11:10:10<05:25, 1.37s/it] 99%|█████████▉| 29763/30000 [11:10:11<05:17, 1.34s/it] 99%|█████████▉| 29764/30000 [11:10:13<05:18, 1.35s/it] 99%|█████████▉| 29765/30000 [11:10:14<05:17, 1.35s/it] 99%|█████████▉| 29766/30000 [11:10:16<05:19, 1.36s/it] 99%|█████████▉| 29767/30000 [11:10:17<05:19, 1.37s/it] 99%|█████████▉| 29768/30000 [11:10:18<05:09, 1.34s/it] 99%|█████████▉| 29769/30000 [11:10:20<05:12, 1.35s/it] 99%|█████████▉| 29770/30000 [11:10:21<05:05, 1.33s/it] {'loss': 0.014, 'grad_norm': 0.08532943576574326, 'learning_rate': 4.8626362409204265e-09, 'flow_matching_loss': 0.014005054719746113, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.66} + 99%|█████████▉| 29770/30000 [11:10:21<05:05, 1.33s/it] 99%|█████████▉| 29771/30000 [11:10:22<05:08, 1.35s/it] 99%|█████████▉| 29772/30000 [11:10:24<05:11, 1.37s/it] 99%|█████████▉| 29773/30000 [11:10:25<05:09, 1.36s/it] 99%|█████████▉| 29774/30000 [11:10:26<05:08, 1.37s/it] 99%|█████████▉| 29775/30000 [11:10:28<05:01, 1.34s/it] 99%|█████████▉| 29776/30000 [11:10:29<05:03, 1.35s/it] 99%|█████████▉| 29777/30000 [11:10:30<05:01, 1.35s/it] 99%|█████████▉| 29778/30000 [11:10:32<05:02, 1.36s/it] 99%|█████████▉| 29779/30000 [11:10:33<05:02, 1.37s/it] 99%|█████████▉| 29780/30000 [11:10:34<04:54, 1.34s/it] {'loss': 0.0151, 'grad_norm': 0.07754822820425034, 'learning_rate': 4.4507618605638126e-09, 'flow_matching_loss': 0.015095721511170269, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.66} + 99%|█████████▉| 29780/30000 [11:10:34<04:54, 1.34s/it] 99%|█████████▉| 29781/30000 [11:10:36<04:55, 1.35s/it] 99%|█████████▉| 29782/30000 [11:10:37<04:54, 1.35s/it] 99%|█████████▉| 29783/30000 [11:10:39<04:56, 1.37s/it] 99%|█████████▉| 29784/30000 [11:10:40<04:56, 1.37s/it] 99%|█████████▉| 29785/30000 [11:10:41<04:48, 1.34s/it] 99%|█████████▉| 29786/30000 [11:10:43<04:47, 1.35s/it] 99%|█████████▉| 29787/30000 [11:10:44<04:40, 1.32s/it] 99%|█████████▉| 29788/30000 [11:10:45<04:41, 1.33s/it] 99%|█████████▉| 29789/30000 [11:10:47<04:42, 1.34s/it] 99%|█████████▉| 29790/30000 [11:10:48<04:48, 1.37s/it] {'loss': 0.0141, 'grad_norm': 0.0658322274684906, 'learning_rate': 4.057108486519301e-09, 'flow_matching_loss': 0.014074154663830996, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.67} + 99%|█████████▉| 29790/30000 [11:10:48<04:48, 1.37s/it] 99%|█████████▉| 29791/30000 [11:10:49<04:47, 1.37s/it] 99%|█████████▉| 29792/30000 [11:10:51<04:36, 1.33s/it] 99%|█████████▉| 29793/30000 [11:10:52<04:38, 1.34s/it] 99%|█████████▉| 29794/30000 [11:10:53<04:44, 1.38s/it] 99%|█████████▉| 29795/30000 [11:10:55<04:42, 1.38s/it] 99%|█████████▉| 29796/30000 [11:10:56<04:41, 1.38s/it] 99%|█████████▉| 29797/30000 [11:10:57<04:31, 1.34s/it] 99%|█████████▉| 29798/30000 [11:10:59<04:31, 1.34s/it] 99%|█████████▉| 29799/30000 [11:11:00<04:30, 1.34s/it] 99%|█████████▉| 29800/30000 [11:11:02<04:34, 1.37s/it] {'loss': 0.0147, 'grad_norm': 0.08310340344905853, 'learning_rate': 3.6816765971126043e-09, 'flow_matching_loss': 0.014719462487846613, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.67} + 99%|█████████▉| 29800/30000 [11:11:02<04:34, 1.37s/it] 99%|█████████▉| 29801/30000 [11:11:03<04:32, 1.37s/it] 99%|█████████▉| 29802/30000 [11:11:04<04:30, 1.37s/it] 99%|█████████▉| 29803/30000 [11:11:06<04:29, 1.37s/it] 99%|█████████▉| 29804/30000 [11:11:07<04:25, 1.36s/it] 99%|█████████▉| 29805/30000 [11:11:08<04:24, 1.36s/it] 99%|█████████▉| 29806/30000 [11:11:10<04:22, 1.35s/it] 99%|█████████▉| 29807/30000 [11:11:11<04:22, 1.36s/it] 99%|█████████▉| 29808/30000 [11:11:12<04:21, 1.36s/it] 99%|█████████▉| 29809/30000 [11:11:14<04:14, 1.33s/it] 99%|█████████▉| 29810/30000 [11:11:15<04:14, 1.34s/it] {'loss': 0.0153, 'grad_norm': 0.10329257696866989, 'learning_rate': 3.324466648528812e-09, 'flow_matching_loss': 0.015255775302648544, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.67} + 99%|█████████▉| 29810/30000 [11:11:15<04:14, 1.34s/it] 99%|█████████▉| 29811/30000 [11:11:16<04:15, 1.35s/it] 99%|█████████▉| 29812/30000 [11:11:18<04:15, 1.36s/it] 99%|█████████▉| 29813/30000 [11:11:19<04:17, 1.37s/it] 99%|█████████▉| 29814/30000 [11:11:21<04:14, 1.37s/it] 99%|█████████▉| 29815/30000 [11:11:22<04:11, 1.36s/it] 99%|█████████▉| 29816/30000 [11:11:23<04:10, 1.36s/it] 99%|█████████▉| 29817/30000 [11:11:25<04:12, 1.38s/it] 99%|█████████▉| 29818/30000 [11:11:26<04:12, 1.39s/it] 99%|█████████▉| 29819/30000 [11:11:28<04:10, 1.39s/it] 99%|█████████▉| 29820/30000 [11:11:29<04:08, 1.38s/it] {'loss': 0.0138, 'grad_norm': 0.11584015190601349, 'learning_rate': 2.9854790748123917e-09, 'flow_matching_loss': 0.013790425937622785, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.68} + 99%|█████████▉| 29820/30000 [11:11:29<04:08, 1.38s/it] 99%|█████████▉| 29821/30000 [11:11:30<04:06, 1.37s/it] 99%|█████████▉| 29822/30000 [11:11:32<04:04, 1.37s/it] 99%|█████████▉| 29823/30000 [11:11:33<04:02, 1.37s/it] 99%|█████████▉| 29824/30000 [11:11:34<04:01, 1.37s/it] 99%|█████████▉| 29825/30000 [11:11:36<04:00, 1.38s/it] 99%|█████████▉| 29826/30000 [11:11:37<03:57, 1.36s/it] 99%|█████████▉| 29827/30000 [11:11:38<03:55, 1.36s/it] 99%|█████████▉| 29828/30000 [11:11:40<03:52, 1.35s/it] 99%|█████████▉| 29829/30000 [11:11:41<03:53, 1.37s/it] 99%|█████████▉| 29830/30000 [11:11:43<03:52, 1.37s/it] {'loss': 0.0141, 'grad_norm': 0.08774961531162262, 'learning_rate': 2.664714287865522e-09, 'flow_matching_loss': 0.014121217839419841, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.68} + 99%|█████████▉| 29830/30000 [11:11:43<03:52, 1.37s/it] 99%|█████████▉| 29831/30000 [11:11:44<03:46, 1.34s/it] 99%|█████████▉| 29832/30000 [11:11:45<03:45, 1.34s/it] 99%|█████████▉| 29833/30000 [11:11:46<03:44, 1.34s/it] 99%|█████████▉| 29834/30000 [11:11:48<03:47, 1.37s/it] 99%|█████████▉| 29835/30000 [11:11:49<03:45, 1.37s/it] 99%|█████████▉| 29836/30000 [11:11:51<03:44, 1.37s/it] 99%|█████████▉| 29837/30000 [11:11:52<03:41, 1.36s/it] 99%|█████████▉| 29838/30000 [11:11:53<03:38, 1.35s/it] 99%|█████████▉| 29839/30000 [11:11:55<03:39, 1.36s/it] 99%|█████████▉| 29840/30000 [11:11:56<03:39, 1.37s/it] {'loss': 0.0143, 'grad_norm': 0.08351629972457886, 'learning_rate': 2.362172677446428e-09, 'flow_matching_loss': 0.014284798875451088, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.68} + 99%|█████████▉| 29840/30000 [11:11:56<03:39, 1.37s/it] 99%|█████████▉| 29841/30000 [11:11:57<03:39, 1.38s/it] 99%|█████████▉| 29842/30000 [11:11:59<03:35, 1.37s/it] 99%|█████████▉| 29843/30000 [11:12:00<03:34, 1.37s/it] 99%|█████████▉| 29844/30000 [11:12:02<03:35, 1.38s/it] 99%|█████████▉| 29845/30000 [11:12:03<03:32, 1.37s/it] 99%|█████████▉| 29846/30000 [11:12:04<03:32, 1.38s/it] 99%|█████████▉| 29847/30000 [11:12:06<03:30, 1.38s/it] 99%|█████████▉| 29848/30000 [11:12:07<03:27, 1.36s/it] 99%|█████████▉| 29849/30000 [11:12:08<03:27, 1.37s/it] 100%|█████████▉| 29850/30000 [11:12:10<03:27, 1.38s/it] {'loss': 0.0159, 'grad_norm': 0.09413779526948929, 'learning_rate': 2.0778546111727137e-09, 'flow_matching_loss': 0.015877093095332383, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.68} + 100%|█████████▉| 29850/30000 [11:12:10<03:27, 1.38s/it] 100%|█████████▉| 29851/30000 [11:12:11<03:27, 1.39s/it] 100%|█████████▉| 29852/30000 [11:12:13<03:25, 1.39s/it] 100%|█████████▉| 29853/30000 [11:12:14<03:20, 1.36s/it] 100%|█████████▉| 29854/30000 [11:12:15<03:18, 1.36s/it] 100%|█████████▉| 29855/30000 [11:12:17<03:16, 1.36s/it] 100%|█████████▉| 29856/30000 [11:12:18<03:16, 1.36s/it] 100%|█████████▉| 29857/30000 [11:12:19<03:15, 1.37s/it] 100%|█████████▉| 29858/30000 [11:12:21<03:11, 1.35s/it] 100%|█████████▉| 29859/30000 [11:12:22<03:09, 1.34s/it] 100%|█████████▉| 29860/30000 [11:12:23<03:07, 1.34s/it] {'loss': 0.0157, 'grad_norm': 0.06286724656820297, 'learning_rate': 1.811760434516363e-09, 'flow_matching_loss': 0.015656255558133124, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.69} + 100%|█████████▉| 29860/30000 [11:12:23<03:07, 1.34s/it] 100%|█████████▉| 29861/30000 [11:12:25<03:08, 1.36s/it] 100%|█████████▉| 29862/30000 [11:12:26<03:07, 1.36s/it] 100%|█████████▉| 29863/30000 [11:12:28<03:07, 1.37s/it] 100%|█████████▉| 29864/30000 [11:12:29<03:03, 1.35s/it] 100%|█████████▉| 29865/30000 [11:12:30<03:01, 1.34s/it] 100%|█████████▉| 29866/30000 [11:12:32<03:03, 1.37s/it] 100%|█████████▉| 29867/30000 [11:12:33<03:01, 1.36s/it] 100%|█████████▉| 29868/30000 [11:12:34<03:01, 1.38s/it] 100%|█████████▉| 29869/30000 [11:12:36<02:59, 1.37s/it] 100%|█████████▉| 29870/30000 [11:12:37<02:56, 1.36s/it] {'loss': 0.0139, 'grad_norm': 0.07024824619293213, 'learning_rate': 1.5638904708087377e-09, 'flow_matching_loss': 0.013904739916324616, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.69} + 100%|█████████▉| 29870/30000 [11:12:37<02:56, 1.36s/it] 100%|█████████▉| 29871/30000 [11:12:38<02:55, 1.36s/it] 100%|█████████▉| 29872/30000 [11:12:40<02:53, 1.36s/it] 100%|█████████▉| 29873/30000 [11:12:41<02:53, 1.37s/it] 100%|█████████▉| 29874/30000 [11:12:43<02:52, 1.37s/it] 100%|█████████▉| 29875/30000 [11:12:44<02:49, 1.36s/it] 100%|█████████▉| 29876/30000 [11:12:45<02:47, 1.35s/it] 100%|█████████▉| 29877/30000 [11:12:47<02:46, 1.35s/it] 100%|█████████▉| 29878/30000 [11:12:48<02:47, 1.37s/it] 100%|█████████▉| 29879/30000 [11:12:49<02:45, 1.37s/it] 100%|█████████▉| 29880/30000 [11:12:51<02:44, 1.37s/it] {'loss': 0.015, 'grad_norm': 0.09886088967323303, 'learning_rate': 1.3342450212355805e-09, 'flow_matching_loss': 0.014962677471339702, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.69} + 100%|█████████▉| 29880/30000 [11:12:51<02:44, 1.37s/it] 100%|█████████▉| 29881/30000 [11:12:52<02:40, 1.35s/it] 100%|█████████▉| 29882/30000 [11:12:53<02:38, 1.34s/it] 100%|█████████▉| 29883/30000 [11:12:55<02:38, 1.35s/it] 100%|█████████▉| 29884/30000 [11:12:56<02:37, 1.36s/it] 100%|█████████▉| 29885/30000 [11:12:57<02:37, 1.37s/it] 100%|█████████▉| 29886/30000 [11:12:59<02:35, 1.37s/it] 100%|█████████▉| 29887/30000 [11:13:00<02:34, 1.36s/it] 100%|█████████▉| 29888/30000 [11:13:02<02:33, 1.37s/it] 100%|█████████▉| 29889/30000 [11:13:03<02:32, 1.38s/it] 100%|█████████▉| 29890/30000 [11:13:04<02:31, 1.38s/it] {'loss': 0.0151, 'grad_norm': 0.09117791801691055, 'learning_rate': 1.122824364835351e-09, 'flow_matching_loss': 0.015051961597055196, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.7} + 100%|█████████▉| 29890/30000 [11:13:04<02:31, 1.38s/it] 100%|█████████▉| 29891/30000 [11:13:06<02:30, 1.38s/it] 100%|█████████▉| 29892/30000 [11:13:07<02:26, 1.35s/it] 100%|█████████▉| 29893/30000 [11:13:08<02:24, 1.35s/it] 100%|█████████▉| 29894/30000 [11:13:10<02:23, 1.35s/it] 100%|█████████▉| 29895/30000 [11:13:11<02:21, 1.35s/it] 100%|█████████▉| 29896/30000 [11:13:12<02:21, 1.36s/it] 100%|█████████▉| 29897/30000 [11:13:14<02:20, 1.36s/it] 100%|█████████▉| 29898/30000 [11:13:15<02:17, 1.34s/it] 100%|█████████▉| 29899/30000 [11:13:16<02:15, 1.34s/it] 100%|█████████▉| 29900/30000 [11:13:18<02:15, 1.35s/it] {'loss': 0.0144, 'grad_norm': 0.06521416455507278, 'learning_rate': 9.296287585042196e-10, 'flow_matching_loss': 0.014378390740603209, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.7} + 100%|█████████▉| 29900/30000 [11:13:18<02:15, 1.35s/it] 100%|█████████▉| 29901/30000 [11:13:19<02:14, 1.36s/it] 100%|█████████▉| 29902/30000 [11:13:21<02:13, 1.37s/it] 100%|█████████▉| 29903/30000 [11:13:22<02:12, 1.36s/it] 100%|█████████▉| 29904/30000 [11:13:23<02:10, 1.36s/it] 100%|█████████▉| 29905/30000 [11:13:25<02:09, 1.36s/it] 100%|█████████▉| 29906/30000 [11:13:26<02:07, 1.36s/it] 100%|█████████▉| 29907/30000 [11:13:27<02:07, 1.37s/it] 100%|█████████▉| 29908/30000 [11:13:29<02:05, 1.36s/it] 100%|█████████▉| 29909/30000 [11:13:30<02:03, 1.35s/it] 100%|█████████▉| 29910/30000 [11:13:31<02:01, 1.35s/it] {'loss': 0.0137, 'grad_norm': 0.07205648720264435, 'learning_rate': 7.546584369960697e-10, 'flow_matching_loss': 0.01367904506623745, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.7} + 100%|█████████▉| 29910/30000 [11:13:31<02:01, 1.35s/it] 100%|█████████▉| 29911/30000 [11:13:33<02:00, 1.35s/it] 100%|█████████▉| 29912/30000 [11:13:34<01:59, 1.36s/it] 100%|█████████▉| 29913/30000 [11:13:36<01:58, 1.36s/it] 100%|█████████▉| 29914/30000 [11:13:37<01:57, 1.37s/it] 100%|█████████▉| 29915/30000 [11:13:38<01:53, 1.33s/it] 100%|█████████▉| 29916/30000 [11:13:40<01:52, 1.34s/it] 100%|█████████▉| 29917/30000 [11:13:41<01:51, 1.34s/it] 100%|█████████▉| 29918/30000 [11:13:42<01:50, 1.35s/it] 100%|█████████▉| 29919/30000 [11:13:44<01:49, 1.35s/it] 100%|█████████▉| 29920/30000 [11:13:45<01:48, 1.35s/it] {'loss': 0.0148, 'grad_norm': 0.06399117410182953, 'learning_rate': 5.97913612912504e-10, 'flow_matching_loss': 0.014763976959511638, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.71} + 100%|█████████▉| 29920/30000 [11:13:45<01:48, 1.35s/it] 100%|█████████▉| 29921/30000 [11:13:46<01:47, 1.36s/it] 100%|█████████▉| 29922/30000 [11:13:48<01:44, 1.34s/it] 100%|█████████▉| 29923/30000 [11:13:49<01:43, 1.35s/it] 100%|█████████▉| 29924/30000 [11:13:50<01:42, 1.35s/it] 100%|█████████▉| 29925/30000 [11:13:52<01:41, 1.35s/it] 100%|█████████▉| 29926/30000 [11:13:53<01:40, 1.36s/it] 100%|█████████▉| 29927/30000 [11:13:54<01:37, 1.33s/it] 100%|█████████▉| 29928/30000 [11:13:56<01:35, 1.33s/it] 100%|█████████▉| 29929/30000 [11:13:57<01:34, 1.33s/it] 100%|█████████▉| 29930/30000 [11:13:58<01:35, 1.37s/it] {'loss': 0.017, 'grad_norm': 0.09678608924150467, 'learning_rate': 4.593944767161684e-10, 'flow_matching_loss': 0.017035243194550277, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.71} + 100%|█████████▉| 29930/30000 [11:13:58<01:35, 1.37s/it] 100%|█████████▉| 29931/30000 [11:14:00<01:34, 1.37s/it] 100%|█████████▉| 29932/30000 [11:14:01<01:31, 1.35s/it] 100%|█████████▉| 29933/30000 [11:14:02<01:30, 1.36s/it] 100%|█████████▉| 29934/30000 [11:14:04<01:29, 1.35s/it] 100%|█████████▉| 29935/30000 [11:14:05<01:27, 1.35s/it] 100%|█████████▉| 29936/30000 [11:14:07<01:26, 1.36s/it] 100%|█████████▉| 29937/30000 [11:14:08<01:25, 1.35s/it] 100%|█████████▉| 29938/30000 [11:14:09<01:24, 1.36s/it] 100%|█████████▉| 29939/30000 [11:14:11<01:21, 1.33s/it] 100%|█████████▉| 29940/30000 [11:14:12<01:19, 1.33s/it] {'loss': 0.0146, 'grad_norm': 0.06919198483228683, 'learning_rate': 3.391011967207591e-10, 'flow_matching_loss': 0.014553564134985209, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.71} + 100%|█████████▉| 29940/30000 [11:14:12<01:19, 1.33s/it] 100%|█████████▉| 29941/30000 [11:14:13<01:19, 1.35s/it] 100%|█████████▉| 29942/30000 [11:14:15<01:18, 1.35s/it] 100%|█████████▉| 29943/30000 [11:14:16<01:17, 1.36s/it] 100%|█████████▉| 29944/30000 [11:14:17<01:14, 1.32s/it] 100%|█████████▉| 29945/30000 [11:14:19<01:13, 1.34s/it] 100%|█████████▉| 29946/30000 [11:14:20<01:10, 1.31s/it] 100%|█████████▉| 29947/30000 [11:14:21<01:10, 1.33s/it] 100%|█████████▉| 29948/30000 [11:14:23<01:09, 1.34s/it] 100%|█████████▉| 29949/30000 [11:14:24<01:08, 1.35s/it] 100%|█████████▉| 29950/30000 [11:14:25<01:08, 1.36s/it] {'loss': 0.0154, 'grad_norm': 0.10789285600185394, 'learning_rate': 2.370339190910231e-10, 'flow_matching_loss': 0.015397820062935353, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.71} + 100%|█████████▉| 29950/30000 [11:14:25<01:08, 1.36s/it] 100%|█████████▉| 29951/30000 [11:14:27<01:05, 1.34s/it] 100%|█████████▉| 29952/30000 [11:14:28<01:04, 1.34s/it] 100%|█████████▉| 29953/30000 [11:14:29<01:02, 1.34s/it] 100%|█████████▉| 29954/30000 [11:14:31<01:02, 1.35s/it] 100%|█████████▉| 29955/30000 [11:14:32<01:00, 1.35s/it] 100%|█████████▉| 29956/30000 [11:14:33<00:58, 1.33s/it] 100%|█████████▉| 29957/30000 [11:14:35<00:57, 1.35s/it] 100%|█████████▉| 29958/30000 [11:14:36<00:55, 1.32s/it] 100%|█████████▉| 29959/30000 [11:14:37<00:54, 1.34s/it] 100%|█████████▉| 29960/30000 [11:14:39<00:53, 1.34s/it] {'loss': 0.0131, 'grad_norm': 0.10595181584358215, 'learning_rate': 1.531927678510847e-10, 'flow_matching_loss': 0.013118467479944228, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.72} + 100%|█████████▉| 29960/30000 [11:14:39<00:53, 1.34s/it] 100%|█████████▉| 29961/30000 [11:14:40<00:52, 1.35s/it] 100%|█████████▉| 29962/30000 [11:14:41<00:51, 1.36s/it] 100%|█████████▉| 29963/30000 [11:14:43<00:49, 1.33s/it] 100%|█████████▉| 29964/30000 [11:14:44<00:48, 1.35s/it] 100%|█████████▉| 29965/30000 [11:14:45<00:46, 1.33s/it] 100%|█████████▉| 29966/30000 [11:14:47<00:45, 1.35s/it] 100%|█████████▉| 29967/30000 [11:14:48<00:44, 1.34s/it] 100%|█████████▉| 29968/30000 [11:14:49<00:42, 1.32s/it] 100%|█████████▉| 29969/30000 [11:14:51<00:41, 1.34s/it] 100%|█████████▉| 29970/30000 [11:14:52<00:39, 1.31s/it] {'loss': 0.014, 'grad_norm': 0.10274787992238998, 'learning_rate': 8.757784487611886e-11, 'flow_matching_loss': 0.014010875765234233, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.72} + 100%|█████████▉| 29970/30000 [11:14:52<00:39, 1.31s/it] 100%|█████████▉| 29971/30000 [11:14:53<00:39, 1.35s/it] 100%|█████████▉| 29972/30000 [11:14:55<00:37, 1.34s/it] 100%|█████████▉| 29973/30000 [11:14:56<00:36, 1.34s/it] 100%|█████████▉| 29974/30000 [11:14:57<00:34, 1.34s/it] 100%|█████████▉| 29975/30000 [11:14:59<00:32, 1.31s/it] 100%|█████████▉| 29976/30000 [11:15:00<00:32, 1.34s/it] 100%|█████████▉| 29977/30000 [11:15:01<00:30, 1.32s/it] 100%|█████████▉| 29978/30000 [11:15:03<00:29, 1.33s/it] 100%|█████████▉| 29979/30000 [11:15:04<00:28, 1.34s/it] 100%|█████████▉| 29980/30000 [11:15:05<00:26, 1.32s/it] {'loss': 0.0136, 'grad_norm': 0.0954645425081253, 'learning_rate': 4.018922989235119e-11, 'flow_matching_loss': 0.013644915632903576, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.72} + 100%|█████████▉| 29980/30000 [11:15:05<00:26, 1.32s/it] 100%|█████████▉| 29981/30000 [11:15:07<00:25, 1.34s/it] 100%|█████████▉| 29982/30000 [11:15:08<00:23, 1.31s/it] 100%|█████████▉| 29983/30000 [11:15:09<00:22, 1.34s/it] 100%|█████████▉| 29984/30000 [11:15:11<00:21, 1.31s/it] 100%|█████████▉| 29985/30000 [11:15:12<00:19, 1.33s/it] 100%|█████████▉| 29986/30000 [11:15:13<00:18, 1.35s/it] 100%|█████████▉| 29987/30000 [11:15:15<00:17, 1.34s/it] 100%|█████████▉| 29988/30000 [11:15:16<00:16, 1.35s/it] 100%|█████████▉| 29989/30000 [11:15:17<00:14, 1.30s/it] 100%|█████████▉| 29990/30000 [11:15:19<00:13, 1.33s/it] {'loss': 0.0132, 'grad_norm': 0.08894939720630646, 'learning_rate': 1.1026980482053972e-11, 'flow_matching_loss': 0.013218238297849894, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.73} + 100%|█████████▉| 29990/30000 [11:15:19<00:13, 1.33s/it] 100%|█████████▉| 29991/30000 [11:15:20<00:11, 1.31s/it] 100%|█████████▉| 29992/30000 [11:15:21<00:10, 1.33s/it] 100%|█████████▉| 29993/30000 [11:15:23<00:09, 1.34s/it] 100%|█████████▉| 29994/30000 [11:15:24<00:07, 1.32s/it] 100%|█████████▉| 29995/30000 [11:15:25<00:06, 1.33s/it] 100%|█████████▉| 29996/30000 [11:15:27<00:05, 1.30s/it] 100%|█████████▉| 29997/30000 [11:15:28<00:03, 1.32s/it] 100%|█████████▉| 29998/30000 [11:15:29<00:02, 1.29s/it] 100%|█████████▉| 29999/30000 [11:15:30<00:01, 1.31s/it] 100%|██████████| 30000/30000 [11:15:32<00:00, 1.30s/it] {'loss': 0.0125, 'grad_norm': 0.07938306778669357, 'learning_rate': 9.113208188082922e-14, 'flow_matching_loss': 0.012512949667870999, 'mgd_loss': 0.0, 'mgd_token_mask_ratio_cfg': 0.30000001192092896, 'epoch': 8.73} + 100%|██████████| 30000/30000 [11:15:32<00:00, 1.30s/it] {'train_runtime': 40539.3311, 'train_samples_per_second': 94.723, 'train_steps_per_second': 0.74, 'train_loss': 0.017189961644013724, 'epoch': 8.73} + 100%|██████████| 30000/30000 [11:15:32<00:00, 1.30s/it] 100%|██████████| 30000/30000 [11:15:32<00:00, 1.35s/it] +/home/ext_minje/miniforge3/envs/robocasa/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +Saved final phase model to: /home/ext_minje/groot_robocasa/robocasa_v2/MGD_Processing_line_Only/dit_core_only/mgd_token_mask_ratio_0.3/phase3 +/home/ext_minje/miniforge3/envs/robocasa/lib/python3.11/site-packages/torch/distributed/distributed_c10d.py:4631: UserWarning: No device id is provided via `init_process_group` or `barrier `. Using the current device set by the user. + warnings.warn( # warn only once +wandb: updating run metadata +wandb: uploading output.log; uploading wandb-summary.json; uploading config.yaml +wandb: uploading output.log; uploading config.yaml +wandb: uploading output.log +wandb: uploading summary, console lines 3003-3005 +wandb: +wandb: Run history: +wandb: train/epoch ▁▁▁▁▂▂▂▂▂▂▂▃▃▃▃▃▃▄▄▄▄▄▄▄▄▅▅▅▅▅▆▆▆▆▆▇▇▇██ +wandb: train/flow_matching_loss ██▇▄▆▄▄▄▄▄▆▃▅▅▄▄▃▃▄▃▄▂▄▂▄▁▂▁▃▁▁▂▂▁▂▁▂▃▂▂ +wandb: train/global_step ▁▁▁▁▁▂▂▂▂▂▃▃▃▃▄▄▄▄▅▅▅▅▅▅▅▆▆▆▆▆▆▆▇▇▇▇▇▇██ +wandb: train/grad_norm █▃▃▄▂▂▂▁▃▂▂▂▂▂▁▂▁▁▂▁▃▂▂▂▂▂▂▂▂▁▂▁▁▂▂▁▂▁▂▂ +wandb: train/learning_rate ▂▄████████▇▇▇▇▇▆▆▆▆▅▅▅▄▄▄▄▄▃▃▃▂▂▂▂▂▂▁▁▁▁ +wandb: train/loss █▂▂▂▁▁▁▁▁▁▂▁▁▁▂▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁ +wandb: train/mgd_loss ▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁ +wandb: train/mgd_token_mask_ratio_cfg ▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁ +wandb: +wandb: Run summary: +wandb: total_flos 0 +wandb: train/epoch 8.72854 +wandb: train/flow_matching_loss 0.01251 +wandb: train/global_step 30000 +wandb: train/grad_norm 0.07938 +wandb: train/learning_rate 0.0 +wandb: train/loss 0.0125 +wandb: train/mgd_loss 0 +wandb: train/mgd_token_mask_ratio_cfg 0.3 +wandb: train_loss 0.01719 +wandb: +3 ... +wandb: +wandb: 🚀 View run mgd_token_mask_ratio_0.3_phase3 at: https://wandb.ai/minje227_hyu-hanyang-university/huggingface/runs/jr27m8yt +wandb: ⭐️ View project at: https://wandb.ai/minje227_hyu-hanyang-university/huggingface +wandb: Synced 5 W&B file(s), 0 media file(s), 0 artifact file(s) and 0 other file(s) +wandb: Find logs at: ./wandb/run-20260603_015019-jr27m8yt/logs +[rank0]:[W603 13:06:36.080451837 ProcessGroupNCCL.cpp:1479] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator()) +WARNING: mimicgen environments not imported since mimicgen is not installed! + +================================================== +GR00T FINE-TUNING CONFIGURATION: +================================================== +config: /home/ext_minje/clvla/benchmarks/robocasa_v2/experiment_cfg/processing_line_only/mgd_dit_core_only3.yaml +dataset_soup: None +output_dir: /tmp/gr00t +output_root: None +data_config: panda_omron +batch_size: 64 +max_steps: 300000 +num_gpus: 2 +save_steps: 20000 +run_name: None +save_total_limit: 100 +seed: 42 +base_model_path: nvidia/GR00T-N1.5-3B +tune_llm: False +tune_visual: False +tune_projector: True +tune_diffusion_model: True +resume: False +learning_rate: 3e-05 +weight_decay: 1e-05 +warmup_ratio: 0.05 +lora_rank: 0 +lora_alpha: 16 +lora_dropout: 0.1 +lora_full_model: False +dataloader_num_workers: 8 +report_to: wandb +embodiment_tag: new_embodiment +video_backend: opencv +balance_dataset_weights: True +balance_trajectory_weights: True +ds_weights_alpha: 0.4 +================================================== + +Using 2 GPUs +Running torchrun command: ['/home/ext_minje/miniforge3/envs/robocasa/bin/python', '-m', 'torch.distributed.run', '--standalone', '--nproc_per_node=2', '--nnodes=1', '/home/ext_minje/clvla/benchmarks/robocasa_v2/my_scripts/gr00t_finetune.py', '--config', '/home/ext_minje/clvla/benchmarks/robocasa_v2/experiment_cfg/processing_line_only/mgd_dit_core_only3.yaml', '--batch-size', '64', '--num-gpus', '2'] diff --git a/RKD_Processing_line_Only/dit_core_only/rkd_temp_0.1/phase3/experiment_cfg/metadata.json b/RKD_Processing_line_Only/dit_core_only/rkd_temp_0.1/phase3/experiment_cfg/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..300eb070963337016e16465f50ba8351ea2eb25c --- /dev/null +++ b/RKD_Processing_line_Only/dit_core_only/rkd_temp_0.1/phase3/experiment_cfg/metadata.json @@ -0,0 +1,431 @@ +{ + "new_embodiment": { + "statistics": { + "state": { + "base_position": { + "max": [ + 7.3139495849609375, + 0.4876587688922882, + 0.7196521759033203 + ], + "min": [ + -4.94293737411499, + -6.890198230743408, + 0.6996827125549316 + ], + "mean": [ + 2.749288365724937, + -2.0455216239909983, + 0.700532551020533 + ], + "std": [ + 1.6786295897046262, + 1.312897969434244, + 0.00133751758906258 + ], + "q01": [ + -1.4241907881148037, + -5.1716547935950885, + 0.6999670898768614 + ], + "q99": [ + 6.198111545831555, + -0.5873531334104489, + 0.7038833014242587 + ] + }, + "base_rotation": { + "max": [ + 0.0, + 0.0, + 1.0, + 1.0 + ], + "min": [ + 0.0, + 0.0, + -1.0, + 0.0 + ], + "mean": [ + 0.0, + 0.0, + 0.23084999587450247, + 0.6238431088581847 + ], + "std": [ + 0.0, + 0.0, + 0.6556404371644047, + 0.3572252105732042 + ], + "q01": [ + 0.0, + 0.0, + -0.9999999999999999, + 1.7482354593273349e-06 + ], + "q99": [ + 0.0, + 0.0, + 0.9999999999999999, + 0.9999999999999999 + ] + }, + "end_effector_position_relative": { + "max": [ + 0.9014528393745422, + 0.8003877401351929, + 0.9829942584037781 + ], + "min": [ + -0.37471598386764526, + -0.8472502827644348, + -0.25070279836654663 + ], + "mean": [ + 0.28667332601863704, + -0.038315473170876704, + 0.45974658906777444 + ], + "std": [ + 0.17067058828305154, + 0.23569952037784195, + 0.21950702354181487 + ], + "q01": [ + 0.015172948963784228, + -0.44450502063220615, + 0.23314444103329338 + ], + "q99": [ + 0.5609069689572412, + 0.38454179128009547, + 0.7028102736473852 + ] + }, + "end_effector_rotation_relative": { + "max": [ + 0.9999998807907104, + 0.9984455108642578, + 0.9434727430343628, + 0.9062229990959167 + ], + "min": [ + -0.9999930262565613, + -0.9988337755203247, + -0.9618149995803833, + 2.1872274658107926e-07 + ], + "mean": [ + -0.25672297401336, + 0.024425335806687216, + -0.08740977164483847, + 0.16608739644018397 + ], + "std": [ + 0.7932585208392383, + 0.3102260195580416, + 0.3772472576315148, + 0.17451959300158773 + ], + "q01": [ + -0.99379006987882, + -0.5478102050038828, + -0.6101777952971636, + 0.002274085014134748 + ], + "q99": [ + 0.8804856038896288, + 0.5910611092873037, + 0.5216058042993562, + 0.5231965974012255 + ] + }, + "gripper_qpos": { + "max": [ + 0.055869169533252716, + 0.010916369967162609 + ], + "min": [ + -0.011436971835792065, + -0.05664053186774254 + ], + "mean": [ + 0.031651551516052555, + -0.03161481387920421 + ], + "std": [ + 0.013119769222217526, + 0.01306078225192402 + ], + "q01": [ + 0.0060999246446777336, + -0.04062601653964198 + ], + "q99": [ + 0.04054891621517283, + -0.0059163567113456345 + ] + } + }, + "action": { + "base_motion": { + "max": [ + 1.0, + 1.0, + 1.0, + 0.0 + ], + "min": [ + -1.0, + -1.0, + -1.0, + 0.0 + ], + "mean": [ + 0.008144896753718373, + -0.00018893135146078263, + -0.0008739062727221845, + 0.0 + ], + "std": [ + 0.11030045424364411, + 0.10148082570313594, + 0.08975698373467506, + 0.0 + ], + "q01": [ + -0.07287373067581518, + -0.09446994199569948, + -0.07702818259249572, + 0.0 + ], + "q99": [ + 0.04485485414407898, + 0.09626941605965177, + 0.07610665906122742, + 0.0 + ] + }, + "control_mode": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.9221974313425939 + ], + "std": [ + 0.38670194342612674 + ], + "q01": [ + -1.0 + ], + "q99": [ + -0.384693946883099 + ] + }, + "end_effector_position": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + 0.01976129923017491, + -0.020870631344490034, + -0.05993476517349181 + ], + "std": [ + 0.4347024990804053, + 0.4221662976569997, + 0.3845506843044066 + ], + "q01": [ + -0.8835792939767807, + -0.9280919251713778, + -0.783361071470956 + ], + "q99": [ + 0.7622084438449307, + 0.8625125252609077, + 0.7470974753250706 + ] + }, + "end_effector_rotation": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + 0.008089673068544335, + -0.026498254627927348, + 0.0029083551743059005 + ], + "std": [ + 0.11216734503733773, + 0.13206002613798629, + 0.12615870412692864 + ], + "q01": [ + -0.2814627972846672, + -0.4342386516115439, + -0.34489755628340574 + ], + "q99": [ + 0.31879353266939386, + 0.28411797683220563, + 0.3597718671669894 + ] + }, + "gripper_close": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.3824228874769045 + ], + "std": [ + 0.9240015096469786 + ], + "q01": [ + -1.0 + ], + "q99": [ + 0.5597272305813592 + ] + } + } + }, + "modalities": { + "video": { + "robot0_eye_in_hand": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + }, + "robot0_agentview_left": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + }, + "robot0_agentview_right": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + } + }, + "state": { + "base_position": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "base_rotation": { + "absolute": true, + "rotation_type": "quaternion", + "shape": [ + 4 + ], + "continuous": true + }, + "end_effector_position_relative": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "end_effector_rotation_relative": { + "absolute": true, + "rotation_type": "quaternion", + "shape": [ + 4 + ], + "continuous": true + }, + "gripper_qpos": { + "absolute": true, + "rotation_type": null, + "shape": [ + 2 + ], + "continuous": true + } + }, + "action": { + "base_motion": { + "absolute": true, + "rotation_type": null, + "shape": [ + 4 + ], + "continuous": true + }, + "control_mode": { + "absolute": true, + "rotation_type": null, + "shape": [ + 1 + ], + "continuous": true + }, + "end_effector_position": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "end_effector_rotation": { + "absolute": true, + "rotation_type": "axis_angle", + "shape": [ + 3 + ], + "continuous": true + }, + "gripper_close": { + "absolute": true, + "rotation_type": null, + "shape": [ + 1 + ], + "continuous": true + } + } + }, + "embodiment_tag": "new_embodiment" + } +} \ No newline at end of file diff --git a/RKD_Processing_line_Only/dit_core_only/rkd_temp_0.1/resolved_config.yaml b/RKD_Processing_line_Only/dit_core_only/rkd_temp_0.1/resolved_config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..10a2bb3f6d664ba707e80055319b75b5d9467a2f --- /dev/null +++ b/RKD_Processing_line_Only/dit_core_only/rkd_temp_0.1/resolved_config.yaml @@ -0,0 +1,33 @@ +experiment_name: rkd_temp_ablation_from_trained_base +policy_type: groot_rkd +base_model_path: ${HOME}/groot_robocasa/robocasa_v2/RKD_Processing_line_Only/rkd_temp_0.1/phase2 +output_root: ${HOME}/groot_robocasa/robocasa_v2/RKD_Processing_line_Only/dit_core_only +sweep: + rkd_temp: + - 0.1 + - 0.2 +dataset: + dataset_soup: my_atomic26_human +training: + num_gpus: 2 + batch_size: 64 +model: null +phases: +- name: phase3_rkd_fm + max_steps: 30000 + save_steps: 0 + trainable: + preset: freeze_processing_line + tune_llm: false + tune_visual: false + tune_projector: true + tune_diffusion_model: true + losses: + rkd_enabled: false + rkd_fm_loss_weight: 1.0 + rkd_loss_weight: 0.0 + rkd_action_temp: 0.1 + rkd_vlm_temp: 0.1 + rkd_relation_mode: token_pair_mean +resolved_sweep: + rkd_temp: 0.1 diff --git a/RKD_Processing_line_Only/dit_core_only/rkd_temp_0.1/rkd_dit_core_only2.yaml b/RKD_Processing_line_Only/dit_core_only/rkd_temp_0.1/rkd_dit_core_only2.yaml new file mode 100644 index 0000000000000000000000000000000000000000..bd8cafb77c75f95be0c97b9a77e18b188cea71c2 --- /dev/null +++ b/RKD_Processing_line_Only/dit_core_only/rkd_temp_0.1/rkd_dit_core_only2.yaml @@ -0,0 +1,31 @@ +experiment_name: rkd_temp_ablation_from_trained_base +policy_type: groot_rkd +base_model_path: ${HOME}/groot_robocasa/robocasa_v2/RKD_Processing_line_Only/rkd_temp_${sweep.rkd_temp}/phase2 +output_root: ${HOME}/groot_robocasa/robocasa_v2/RKD_Processing_line_Only/dit_core_only +sweep: + rkd_temp: + - 0.1 + - 0.2 +dataset: + dataset_soup: my_atomic26_human +training: + num_gpus: 2 + batch_size: 64 +model: null +phases: +- name: phase3_rkd_fm + max_steps: 30000 + save_steps: 0 + trainable: + preset: freeze_processing_line + tune_llm: false + tune_visual: false + tune_projector: true + tune_diffusion_model: true + losses: + rkd_enabled: false + rkd_fm_loss_weight: 1.0 + rkd_loss_weight: 0.0 + rkd_action_temp: ${sweep.rkd_temp} + rkd_vlm_temp: ${sweep.rkd_temp} + rkd_relation_mode: token_pair_mean diff --git a/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/mgd_phase3_aux_linear_decay.yaml b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/mgd_phase3_aux_linear_decay.yaml new file mode 100644 index 0000000000000000000000000000000000000000..5dfa65e6c8e1ae7a5158157f8202c466854e80a2 --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/mgd_phase3_aux_linear_decay.yaml @@ -0,0 +1,33 @@ +experiment_name: mgd_phase3_aux_linear_decay +policy_type: groot_mgd +base_model_path: ${HOME}/groot_robocasa/robocasa_v2/MGD_Processing_line_Only/mgd_token_mask_ratio_${sweep.mgd_token_mask_ratio}/phase2 +output_root: ${HOME}/groot_robocasa/robocasa_v2/processing_line_only/loss_strategy_ablation/mgd +sweep: + loss_strategy: + - aux_linear_decay + mgd_token_mask_ratio: + - 0.3 +dataset: + dataset_soup: my_atomic26_human +training: + num_gpus: 1 + batch_size: 128 +model: null +phases: +- name: phase3_fm_mgd_linear_decay + max_steps: 30000 + save_steps: 0 + trainable: + tune_llm: false + tune_visual: false + tune_projector: true + tune_diffusion_model: true + losses: + mgd_enabled: true + mgd_fm_loss_weight: 1.0 + mgd_loss_weight_schedule: linear_decay + mgd_loss_weight_start: 1.0 + mgd_loss_weight_end: 0.0 + mgd_token_mask_ratio: ${sweep.mgd_token_mask_ratio} + mgd_use_cosine_loss: true + mgd_use_mse_loss: false diff --git a/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/phase3/config.json b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/phase3/config.json new file mode 100644 index 0000000000000000000000000000000000000000..a83db7d6adefeb42268cbb8bd3856d3798d6d04f --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/phase3/config.json @@ -0,0 +1,78 @@ +{ + "action_dim": 32, + "action_head_cfg": { + "action_dim": 32, + "action_horizon": 16, + "add_pos_embed": true, + "backbone_embedding_dim": 2048, + "diffusion_model_cfg": { + "attention_head_dim": 48, + "cross_attention_dim": 2048, + "dropout": 0.2, + "final_dropout": true, + "interleave_self_attention": true, + "norm_type": "ada_norm", + "num_attention_heads": 32, + "num_layers": 16, + "output_dim": 1024, + "positional_embeddings": null + }, + "hidden_size": 1024, + "input_embedding_dim": 1536, + "max_action_dim": 32, + "max_state_dim": 64, + "model_dtype": "float32", + "noise_beta_alpha": 1.5, + "noise_beta_beta": 1.0, + "noise_s": 0.999, + "num_inference_timesteps": 4, + "num_target_vision_tokens": 32, + "num_timestep_buckets": 1000, + "tune_diffusion_model": true, + "tune_projector": true, + "use_vlln": true, + "vl_self_attention_cfg": { + "attention_head_dim": 64, + "dropout": 0.2, + "final_dropout": true, + "num_attention_heads": 32, + "num_layers": 4, + "positional_embeddings": null + } + }, + "action_horizon": 16, + "architectures": [ + "GR00T_N1_5_MGD" + ], + "attn_implementation": null, + "backbone_cfg": { + "eagle_path": "NVEagle/eagle_er-qwen3_1_7B-Siglip2_400M_stage1_5_128gpu_er_v7_1mlp_nops", + "load_bf16": false, + "project_to_dim": null, + "reproject_vision": false, + "select_layer": 12, + "tune_llm": false, + "tune_visual": true, + "use_flash_attention": true + }, + "compute_dtype": "bfloat16", + "hidden_size": 2048, + "mgd_enabled": true, + "mgd_fm_loss_weight": 1.0, + "mgd_loss_weight": 0.0, + "mgd_loss_weight_end": 0.0, + "mgd_loss_weight_schedule": "linear_decay", + "mgd_loss_weight_start": 1.0, + "mgd_pretrained_projector_path": null, + "mgd_sequence_hidden_dim": 512, + "mgd_target_dim": 512, + "mgd_target_pooling": "flatten", + "mgd_target_projection": "frozen_random", + "mgd_token_mask_ratio": 0.3, + "mgd_use_cosine_loss": true, + "mgd_use_mse_loss": false, + "model_dtype": "float32", + "model_type": "gr00t_n1_5", + "torch_dtype": "bfloat16", + "transformers_version": "4.51.3" +} diff --git a/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/phase3/experiment_cfg/metadata.json b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/phase3/experiment_cfg/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..300eb070963337016e16465f50ba8351ea2eb25c --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/phase3/experiment_cfg/metadata.json @@ -0,0 +1,431 @@ +{ + "new_embodiment": { + "statistics": { + "state": { + "base_position": { + "max": [ + 7.3139495849609375, + 0.4876587688922882, + 0.7196521759033203 + ], + "min": [ + -4.94293737411499, + -6.890198230743408, + 0.6996827125549316 + ], + "mean": [ + 2.749288365724937, + -2.0455216239909983, + 0.700532551020533 + ], + "std": [ + 1.6786295897046262, + 1.312897969434244, + 0.00133751758906258 + ], + "q01": [ + -1.4241907881148037, + -5.1716547935950885, + 0.6999670898768614 + ], + "q99": [ + 6.198111545831555, + -0.5873531334104489, + 0.7038833014242587 + ] + }, + "base_rotation": { + "max": [ + 0.0, + 0.0, + 1.0, + 1.0 + ], + "min": [ + 0.0, + 0.0, + -1.0, + 0.0 + ], + "mean": [ + 0.0, + 0.0, + 0.23084999587450247, + 0.6238431088581847 + ], + "std": [ + 0.0, + 0.0, + 0.6556404371644047, + 0.3572252105732042 + ], + "q01": [ + 0.0, + 0.0, + -0.9999999999999999, + 1.7482354593273349e-06 + ], + "q99": [ + 0.0, + 0.0, + 0.9999999999999999, + 0.9999999999999999 + ] + }, + "end_effector_position_relative": { + "max": [ + 0.9014528393745422, + 0.8003877401351929, + 0.9829942584037781 + ], + "min": [ + -0.37471598386764526, + -0.8472502827644348, + -0.25070279836654663 + ], + "mean": [ + 0.28667332601863704, + -0.038315473170876704, + 0.45974658906777444 + ], + "std": [ + 0.17067058828305154, + 0.23569952037784195, + 0.21950702354181487 + ], + "q01": [ + 0.015172948963784228, + -0.44450502063220615, + 0.23314444103329338 + ], + "q99": [ + 0.5609069689572412, + 0.38454179128009547, + 0.7028102736473852 + ] + }, + "end_effector_rotation_relative": { + "max": [ + 0.9999998807907104, + 0.9984455108642578, + 0.9434727430343628, + 0.9062229990959167 + ], + "min": [ + -0.9999930262565613, + -0.9988337755203247, + -0.9618149995803833, + 2.1872274658107926e-07 + ], + "mean": [ + -0.25672297401336, + 0.024425335806687216, + -0.08740977164483847, + 0.16608739644018397 + ], + "std": [ + 0.7932585208392383, + 0.3102260195580416, + 0.3772472576315148, + 0.17451959300158773 + ], + "q01": [ + -0.99379006987882, + -0.5478102050038828, + -0.6101777952971636, + 0.002274085014134748 + ], + "q99": [ + 0.8804856038896288, + 0.5910611092873037, + 0.5216058042993562, + 0.5231965974012255 + ] + }, + "gripper_qpos": { + "max": [ + 0.055869169533252716, + 0.010916369967162609 + ], + "min": [ + -0.011436971835792065, + -0.05664053186774254 + ], + "mean": [ + 0.031651551516052555, + -0.03161481387920421 + ], + "std": [ + 0.013119769222217526, + 0.01306078225192402 + ], + "q01": [ + 0.0060999246446777336, + -0.04062601653964198 + ], + "q99": [ + 0.04054891621517283, + -0.0059163567113456345 + ] + } + }, + "action": { + "base_motion": { + "max": [ + 1.0, + 1.0, + 1.0, + 0.0 + ], + "min": [ + -1.0, + -1.0, + -1.0, + 0.0 + ], + "mean": [ + 0.008144896753718373, + -0.00018893135146078263, + -0.0008739062727221845, + 0.0 + ], + "std": [ + 0.11030045424364411, + 0.10148082570313594, + 0.08975698373467506, + 0.0 + ], + "q01": [ + -0.07287373067581518, + -0.09446994199569948, + -0.07702818259249572, + 0.0 + ], + "q99": [ + 0.04485485414407898, + 0.09626941605965177, + 0.07610665906122742, + 0.0 + ] + }, + "control_mode": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.9221974313425939 + ], + "std": [ + 0.38670194342612674 + ], + "q01": [ + -1.0 + ], + "q99": [ + -0.384693946883099 + ] + }, + "end_effector_position": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + 0.01976129923017491, + -0.020870631344490034, + -0.05993476517349181 + ], + "std": [ + 0.4347024990804053, + 0.4221662976569997, + 0.3845506843044066 + ], + "q01": [ + -0.8835792939767807, + -0.9280919251713778, + -0.783361071470956 + ], + "q99": [ + 0.7622084438449307, + 0.8625125252609077, + 0.7470974753250706 + ] + }, + "end_effector_rotation": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + 0.008089673068544335, + -0.026498254627927348, + 0.0029083551743059005 + ], + "std": [ + 0.11216734503733773, + 0.13206002613798629, + 0.12615870412692864 + ], + "q01": [ + -0.2814627972846672, + -0.4342386516115439, + -0.34489755628340574 + ], + "q99": [ + 0.31879353266939386, + 0.28411797683220563, + 0.3597718671669894 + ] + }, + "gripper_close": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.3824228874769045 + ], + "std": [ + 0.9240015096469786 + ], + "q01": [ + -1.0 + ], + "q99": [ + 0.5597272305813592 + ] + } + } + }, + "modalities": { + "video": { + "robot0_eye_in_hand": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + }, + "robot0_agentview_left": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + }, + "robot0_agentview_right": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + } + }, + "state": { + "base_position": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "base_rotation": { + "absolute": true, + "rotation_type": "quaternion", + "shape": [ + 4 + ], + "continuous": true + }, + "end_effector_position_relative": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "end_effector_rotation_relative": { + "absolute": true, + "rotation_type": "quaternion", + "shape": [ + 4 + ], + "continuous": true + }, + "gripper_qpos": { + "absolute": true, + "rotation_type": null, + "shape": [ + 2 + ], + "continuous": true + } + }, + "action": { + "base_motion": { + "absolute": true, + "rotation_type": null, + "shape": [ + 4 + ], + "continuous": true + }, + "control_mode": { + "absolute": true, + "rotation_type": null, + "shape": [ + 1 + ], + "continuous": true + }, + "end_effector_position": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "end_effector_rotation": { + "absolute": true, + "rotation_type": "axis_angle", + "shape": [ + 3 + ], + "continuous": true + }, + "gripper_close": { + "absolute": true, + "rotation_type": null, + "shape": [ + 1 + ], + "continuous": true + } + } + }, + "embodiment_tag": "new_embodiment" + } +} \ No newline at end of file diff --git a/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/phase3/model.safetensors.index.json b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/phase3/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..579ccf56b744c24c73f571cce697c12343122f56 --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/phase3/model.safetensors.index.json @@ -0,0 +1,916 @@ +{ + "metadata": { + "total_size": 7642572676 + }, + "weight_map": { + "action_head.action_decoder.layer1.W": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer1.b": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer2.W": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer2.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W1.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W1.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W2.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W2.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W3.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W3.b": "model-00002-of-00002.safetensors", + "action_head.future_tokens.weight": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_1.bias": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_1.weight": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_2.bias": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_2.weight": "model-00002-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_1.bias": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_1.weight": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_2.bias": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.position_embedding.weight": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer1.W": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer1.b": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer2.W": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer2.b": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vlln.bias": "model-00002-of-00002.safetensors", + "action_head.vlln.weight": "model-00002-of-00002.safetensors", + "action_target_projector.projector.bias": "model-00002-of-00002.safetensors", + "action_target_projector.projector.weight": "model-00002-of-00002.safetensors", + "backbone.eagle_model.language_model.lm_head.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.embed_tokens.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.mlp1.0.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.mlp1.0.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.patch_embedding.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.patch_embedding.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.position_embedding.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.in_proj_bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.in_proj_weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.layernorm.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.probe": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.post_layernorm.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.post_layernorm.weight": "model-00001-of-00002.safetensors", + "sequence_mgd_head.attn_score.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.attn_score.weight": "model-00002-of-00002.safetensors", + "sequence_mgd_head.out_proj.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.out_proj.weight": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.0.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.0.weight": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.2.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.2.weight": "model-00002-of-00002.safetensors" + } +} diff --git a/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/phase3/trainer_state.json b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/phase3/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..b98fcd6322e984c769d0a2c424aae718a0ac84cf --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/phase3/trainer_state.json @@ -0,0 +1,60043 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 8.728542333430317, + "eval_steps": 500, + "global_step": 30000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "actual_token_mask_ratio": 0.30040227472782133, + "epoch": 0.002909514111143439, + "flow_matching_loss": 0.12298423051834106, + "grad_norm": 0.7868468165397644, + "kept_token_count": 572.53671875, + "learning_rate": 1.8e-07, + "loss": 0.1306, + "mgd_cos_sim": 0.9923962116241455, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076037780847400425, + "mgd_loss_weight": 0.9998499929904938, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.5814636230469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.3129852294922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10, + "valid_token_count": 818.38046875 + }, + { + "actual_token_mask_ratio": 0.2999395698308945, + "epoch": 0.005819028222286878, + "flow_matching_loss": 0.10989849641919136, + "grad_norm": 0.6629934310913086, + "kept_token_count": 573.0015625, + "learning_rate": 3.8e-07, + "loss": 0.1173, + "mgd_cos_sim": 0.9925511121749878, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074489060789346695, + "mgd_loss_weight": 0.9995166480541229, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.9244079589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.9990539550781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20, + "valid_token_count": 818.50078125 + }, + { + "actual_token_mask_ratio": 0.29944669306278227, + "epoch": 0.008728542333430317, + "flow_matching_loss": 0.10452512353658676, + "grad_norm": 0.6846851706504822, + "kept_token_count": 573.16953125, + "learning_rate": 5.800000000000001e-07, + "loss": 0.1121, + "mgd_cos_sim": 0.9924566149711609, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075433942489326, + "mgd_loss_weight": 0.9991833090782165, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.3287017822266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.9890960693359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 30, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.29940836429595946, + "epoch": 0.011638056444573755, + "flow_matching_loss": 0.08869744390249253, + "grad_norm": 0.45964956283569336, + "kept_token_count": 573.34140625, + "learning_rate": 7.799999999999999e-07, + "loss": 0.0962, + "mgd_cos_sim": 0.9924743115901947, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007525690132752061, + "mgd_loss_weight": 0.9988499581813812, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.19675598144534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.11659851074216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 40, + "valid_token_count": 818.37109375 + }, + { + "actual_token_mask_ratio": 0.29983765482902525, + "epoch": 0.014547570555717195, + "flow_matching_loss": 0.07611456364393235, + "grad_norm": 0.41570207476615906, + "kept_token_count": 572.93359375, + "learning_rate": 9.8e-07, + "loss": 0.084, + "mgd_cos_sim": 0.9921221792697906, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007877815049141645, + "mgd_loss_weight": 0.9985166192054749, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.30791625976565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.59783325195315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 50, + "valid_token_count": 818.29140625 + }, + { + "actual_token_mask_ratio": 0.2996689766645432, + "epoch": 0.017457084666860634, + "flow_matching_loss": 0.06639967262744903, + "grad_norm": 0.29384759068489075, + "kept_token_count": 572.8390625, + "learning_rate": 1.18e-06, + "loss": 0.0743, + "mgd_cos_sim": 0.9921103119850159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007889698818325996, + "mgd_loss_weight": 0.998183274269104, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.7526062011719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.31878967285155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 60, + "valid_token_count": 817.9578125 + }, + { + "actual_token_mask_ratio": 0.2999273419380188, + "epoch": 0.020366598778004074, + "flow_matching_loss": 0.05376637689769268, + "grad_norm": 0.280385285615921, + "kept_token_count": 572.81640625, + "learning_rate": 1.38e-06, + "loss": 0.0615, + "mgd_cos_sim": 0.9922336041927338, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007766398601233959, + "mgd_loss_weight": 0.9978499233722686, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.1280883789062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.2303955078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 70, + "valid_token_count": 818.22578125 + }, + { + "actual_token_mask_ratio": 0.2998680680990219, + "epoch": 0.02327611288914751, + "flow_matching_loss": 0.0524392981082201, + "grad_norm": 0.23638828098773956, + "kept_token_count": 572.9078125, + "learning_rate": 1.5800000000000001e-06, + "loss": 0.0604, + "mgd_cos_sim": 0.9920338749885559, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007966135116294026, + "mgd_loss_weight": 0.9975165903568268, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 334.17542724609376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.5535522460938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 80, + "valid_token_count": 818.28984375 + }, + { + "actual_token_mask_ratio": 0.29977086186408997, + "epoch": 0.02618562700029095, + "flow_matching_loss": 0.0412452545017004, + "grad_norm": 0.24359464645385742, + "kept_token_count": 572.93046875, + "learning_rate": 1.7800000000000001e-06, + "loss": 0.0491, + "mgd_cos_sim": 0.9921652913093567, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00783471530303359, + "mgd_loss_weight": 0.997183233499527, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.5385009765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.02231750488284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 90, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.2996587365865707, + "epoch": 0.02909514111143439, + "flow_matching_loss": 0.038543707318603994, + "grad_norm": 0.2080250382423401, + "kept_token_count": 573.2578125, + "learning_rate": 1.98e-06, + "loss": 0.0465, + "mgd_cos_sim": 0.9920610547065735, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007938943384215236, + "mgd_loss_weight": 0.9968498945236206, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 299.551513671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.0261932373047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 100, + "valid_token_count": 818.53671875 + }, + { + "actual_token_mask_ratio": 0.3000616401433945, + "epoch": 0.03200465522257783, + "flow_matching_loss": 0.03592816945165396, + "grad_norm": 0.20328080654144287, + "kept_token_count": 572.63828125, + "learning_rate": 2.1800000000000003e-06, + "loss": 0.0446, + "mgd_cos_sim": 0.9912972211837768, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008702791295945644, + "mgd_loss_weight": 0.9965165555477142, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 282.7716979980469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.2222534179688, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 110, + "valid_token_count": 818.12265625 + }, + { + "actual_token_mask_ratio": 0.2996388703584671, + "epoch": 0.03491416933372127, + "flow_matching_loss": 0.03582309428602457, + "grad_norm": 0.19341519474983215, + "kept_token_count": 572.7828125, + "learning_rate": 2.38e-06, + "loss": 0.0442, + "mgd_cos_sim": 0.9915530443191528, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008446947811171413, + "mgd_loss_weight": 0.9961831986904144, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 264.2916717529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.83973083496096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 120, + "valid_token_count": 817.8390625 + }, + { + "actual_token_mask_ratio": 0.300084662437439, + "epoch": 0.037823683444864704, + "flow_matching_loss": 0.029323955811560155, + "grad_norm": 0.18294887244701385, + "kept_token_count": 572.80703125, + "learning_rate": 2.58e-06, + "loss": 0.0381, + "mgd_cos_sim": 0.9911601841449738, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008839825633913279, + "mgd_loss_weight": 0.9958498656749726, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 241.89120330810547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.34095764160156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 130, + "valid_token_count": 818.39765625 + }, + { + "actual_token_mask_ratio": 0.29994952082633974, + "epoch": 0.04073319755600815, + "flow_matching_loss": 0.02703567687422037, + "grad_norm": 0.1437058299779892, + "kept_token_count": 572.69609375, + "learning_rate": 2.78e-06, + "loss": 0.036, + "mgd_cos_sim": 0.990979117155075, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009020880702883005, + "mgd_loss_weight": 0.9955165147781372, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 225.15384674072266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1117462158203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 140, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.29976836442947385, + "epoch": 0.043642711667151585, + "flow_matching_loss": 0.02648906372487545, + "grad_norm": 0.14391793310642242, + "kept_token_count": 572.89453125, + "learning_rate": 2.98e-06, + "loss": 0.0354, + "mgd_cos_sim": 0.9910950541496277, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008904949203133583, + "mgd_loss_weight": 0.9951831698417664, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 211.41138458251953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.76280212402344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 150, + "valid_token_count": 818.15078125 + }, + { + "actual_token_mask_ratio": 0.2996849089860916, + "epoch": 0.04655222577829502, + "flow_matching_loss": 0.024466136004775763, + "grad_norm": 0.12862570583820343, + "kept_token_count": 573.025, + "learning_rate": 3.18e-06, + "loss": 0.0334, + "mgd_cos_sim": 0.9910314440727234, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008968566730618476, + "mgd_loss_weight": 0.99484983086586, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 200.9302490234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5594207763672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 160, + "valid_token_count": 818.24140625 + }, + { + "actual_token_mask_ratio": 0.299516886472702, + "epoch": 0.049461739889438465, + "flow_matching_loss": 0.025922176241874696, + "grad_norm": 0.12622195482254028, + "kept_token_count": 573.14375, + "learning_rate": 3.38e-06, + "loss": 0.0355, + "mgd_cos_sim": 0.990370637178421, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009629361517727374, + "mgd_loss_weight": 0.9945164799690247, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 193.56197509765624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.59046325683596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 170, + "valid_token_count": 818.21171875 + }, + { + "actual_token_mask_ratio": 0.3002052277326584, + "epoch": 0.0523712540005819, + "flow_matching_loss": 0.023681787215173243, + "grad_norm": 0.1431080847978592, + "kept_token_count": 572.65078125, + "learning_rate": 3.58e-06, + "loss": 0.0328, + "mgd_cos_sim": 0.9907890379428863, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009210966620594263, + "mgd_loss_weight": 0.9941831409931183, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 188.46351165771483, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3180816650391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 180, + "valid_token_count": 818.31328125 + }, + { + "actual_token_mask_ratio": 0.2996139794588089, + "epoch": 0.055280768111725345, + "flow_matching_loss": 0.021583708748221398, + "grad_norm": 0.10964047908782959, + "kept_token_count": 573.146875, + "learning_rate": 3.7800000000000002e-06, + "loss": 0.0313, + "mgd_cos_sim": 0.9902590930461883, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009740916453301907, + "mgd_loss_weight": 0.9938497960567474, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 184.01599884033203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.87060546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 190, + "valid_token_count": 818.32734375 + }, + { + "actual_token_mask_ratio": 0.2999971568584442, + "epoch": 0.05819028222286878, + "flow_matching_loss": 0.021982060559093953, + "grad_norm": 0.12837542593479156, + "kept_token_count": 572.66484375, + "learning_rate": 3.98e-06, + "loss": 0.0318, + "mgd_cos_sim": 0.9901628196239471, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009837180189788342, + "mgd_loss_weight": 0.9935164451599121, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 183.34566345214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.5789520263672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 200, + "valid_token_count": 818.0859375 + }, + { + "actual_token_mask_ratio": 0.3002703130245209, + "epoch": 0.06109979633401222, + "flow_matching_loss": 0.022379160672426224, + "grad_norm": 0.10005210340023041, + "kept_token_count": 572.4453125, + "learning_rate": 4.18e-06, + "loss": 0.0319, + "mgd_cos_sim": 0.9904317021369934, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009568288829177618, + "mgd_loss_weight": 0.9931831061840057, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 180.20514678955078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2686340332031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 210, + "valid_token_count": 818.09921875 + }, + { + "actual_token_mask_ratio": 0.2995095461606979, + "epoch": 0.06400931044515566, + "flow_matching_loss": 0.023055230639874934, + "grad_norm": 0.09969936311244965, + "kept_token_count": 573.24921875, + "learning_rate": 4.3799999999999996e-06, + "loss": 0.033, + "mgd_cos_sim": 0.9900249600410461, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009975043032318354, + "mgd_loss_weight": 0.9928497612476349, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 179.6704299926758, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.27672424316404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 220, + "valid_token_count": 818.35234375 + }, + { + "actual_token_mask_ratio": 0.2998043894767761, + "epoch": 0.06691882455629909, + "flow_matching_loss": 0.02093523759394884, + "grad_norm": 0.11013385653495789, + "kept_token_count": 572.9453125, + "learning_rate": 4.58e-06, + "loss": 0.0302, + "mgd_cos_sim": 0.9906863868236542, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009313627053052187, + "mgd_loss_weight": 0.9925164163112641, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 181.35598754882812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.87352294921874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 230, + "valid_token_count": 818.26640625 + }, + { + "actual_token_mask_ratio": 0.30012513399124147, + "epoch": 0.06982833866744254, + "flow_matching_loss": 0.020363811682909728, + "grad_norm": 0.09588447958230972, + "kept_token_count": 572.77890625, + "learning_rate": 4.78e-06, + "loss": 0.0299, + "mgd_cos_sim": 0.9903521776199341, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00964781390503049, + "mgd_loss_weight": 0.9921830713748931, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 179.36323089599608, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.4397918701172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 240, + "valid_token_count": 818.39921875 + }, + { + "actual_token_mask_ratio": 0.2998598426580429, + "epoch": 0.07273785277858598, + "flow_matching_loss": 0.020915719121694563, + "grad_norm": 0.09199198335409164, + "kept_token_count": 572.8109375, + "learning_rate": 4.980000000000001e-06, + "loss": 0.03, + "mgd_cos_sim": 0.9907924890518188, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009207518398761749, + "mgd_loss_weight": 0.9918497323989868, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 181.0469223022461, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.56861572265626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 250, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.2989384591579437, + "epoch": 0.07564736688972941, + "flow_matching_loss": 0.019501696340739727, + "grad_norm": 0.10036198794841766, + "kept_token_count": 573.6796875, + "learning_rate": 5.18e-06, + "loss": 0.0288, + "mgd_cos_sim": 0.9906427323818207, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0093572529964149, + "mgd_loss_weight": 0.9915163815021515, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 178.3244613647461, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.36179809570314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 260, + "valid_token_count": 818.30078125 + }, + { + "actual_token_mask_ratio": 0.2996117889881134, + "epoch": 0.07855688100087285, + "flow_matching_loss": 0.021747706085443498, + "grad_norm": 0.11372004449367523, + "kept_token_count": 573.0640625, + "learning_rate": 5.38e-06, + "loss": 0.031, + "mgd_cos_sim": 0.99063760638237, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009362381137907504, + "mgd_loss_weight": 0.9911830425262451, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 181.32997436523436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.9364868164063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 270, + "valid_token_count": 818.2125 + }, + { + "actual_token_mask_ratio": 0.2992363542318344, + "epoch": 0.0814663951120163, + "flow_matching_loss": 0.01882209712639451, + "grad_norm": 0.08919690549373627, + "kept_token_count": 573.44921875, + "learning_rate": 5.58e-06, + "loss": 0.0282, + "mgd_cos_sim": 0.990554416179657, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009445580840110778, + "mgd_loss_weight": 0.9908496975898743, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 183.04608612060548, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.2097503662109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 280, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.3008067637681961, + "epoch": 0.08437590922315973, + "flow_matching_loss": 0.020709837228059767, + "grad_norm": 0.09664466232061386, + "kept_token_count": 572.04140625, + "learning_rate": 5.7800000000000006e-06, + "loss": 0.0293, + "mgd_cos_sim": 0.9912923097610473, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008707691077142954, + "mgd_loss_weight": 0.9905163466930389, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 183.9917007446289, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.91551513671874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 290, + "valid_token_count": 818.13984375 + }, + { + "actual_token_mask_ratio": 0.3000073403120041, + "epoch": 0.08728542333430317, + "flow_matching_loss": 0.01994443479925394, + "grad_norm": 0.10031113773584366, + "kept_token_count": 572.80859375, + "learning_rate": 5.98e-06, + "loss": 0.0291, + "mgd_cos_sim": 0.9907127141952514, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009287300053983927, + "mgd_loss_weight": 0.9901830136775971, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 182.1752166748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.65740966796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 300, + "valid_token_count": 818.30546875 + }, + { + "actual_token_mask_ratio": 0.30038678646087646, + "epoch": 0.09019493744544661, + "flow_matching_loss": 0.019343936070799827, + "grad_norm": 0.10550954937934875, + "kept_token_count": 572.36484375, + "learning_rate": 6.18e-06, + "loss": 0.0283, + "mgd_cos_sim": 0.9909050405025482, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009094956796616316, + "mgd_loss_weight": 0.9898496568202972, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 180.51683502197267, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7112060546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 310, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.29997038543224336, + "epoch": 0.09310445155659004, + "flow_matching_loss": 0.02015405995771289, + "grad_norm": 0.09889587759971619, + "kept_token_count": 572.790625, + "learning_rate": 6.38e-06, + "loss": 0.0291, + "mgd_cos_sim": 0.9910034120082856, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008996599027886987, + "mgd_loss_weight": 0.9895163178443909, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 181.1805221557617, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.6641845703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 320, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.30036543011665345, + "epoch": 0.09601396566773349, + "flow_matching_loss": 0.019876261055469514, + "grad_norm": 0.11172794550657272, + "kept_token_count": 572.4421875, + "learning_rate": 6.58e-06, + "loss": 0.0281, + "mgd_cos_sim": 0.9916632652282715, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008336748695001006, + "mgd_loss_weight": 0.9891829788684845, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 188.64177856445312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.3099029541016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 330, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.29982112944126127, + "epoch": 0.09892347977887693, + "flow_matching_loss": 0.01847883276641369, + "grad_norm": 0.10038074105978012, + "kept_token_count": 572.88046875, + "learning_rate": 6.78e-06, + "loss": 0.027, + "mgd_cos_sim": 0.9914081215858459, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008591876830905676, + "mgd_loss_weight": 0.9888496220111846, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 188.22547607421876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.3670623779297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 340, + "valid_token_count": 818.19140625 + }, + { + "actual_token_mask_ratio": 0.3003722369670868, + "epoch": 0.10183299389002037, + "flow_matching_loss": 0.017275003343820573, + "grad_norm": 0.0811794102191925, + "kept_token_count": 572.60546875, + "learning_rate": 6.98e-06, + "loss": 0.0257, + "mgd_cos_sim": 0.9915135085582734, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008486500475555658, + "mgd_loss_weight": 0.9885162889957428, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 186.38394470214843, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.7128845214844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 350, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.2991302669048309, + "epoch": 0.1047425080011638, + "flow_matching_loss": 0.019699203502386808, + "grad_norm": 0.08435431122779846, + "kept_token_count": 573.57109375, + "learning_rate": 7.180000000000001e-06, + "loss": 0.0286, + "mgd_cos_sim": 0.9910355865955353, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008964413497596978, + "mgd_loss_weight": 0.9881829380989074, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 187.36163177490235, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.09589233398435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 360, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.29916212856769564, + "epoch": 0.10765202211230725, + "flow_matching_loss": 0.018787751626223325, + "grad_norm": 0.08229030668735504, + "kept_token_count": 573.403125, + "learning_rate": 7.3800000000000005e-06, + "loss": 0.0272, + "mgd_cos_sim": 0.9914618551731109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008538151811808347, + "mgd_loss_weight": 0.9878495931625366, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 193.64216003417968, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.99224548339845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 370, + "valid_token_count": 818.16796875 + }, + { + "actual_token_mask_ratio": 0.30039132237434385, + "epoch": 0.11056153622345069, + "flow_matching_loss": 0.020100552961230277, + "grad_norm": 0.09932008385658264, + "kept_token_count": 572.534375, + "learning_rate": 7.5799999999999994e-06, + "loss": 0.0289, + "mgd_cos_sim": 0.9911343336105347, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008865676261484623, + "mgd_loss_weight": 0.9875162541866302, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 194.6553192138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.0729034423828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 380, + "valid_token_count": 818.36484375 + }, + { + "actual_token_mask_ratio": 0.3003829330205917, + "epoch": 0.11347105033459412, + "flow_matching_loss": 0.019941031746566296, + "grad_norm": 0.07665983587503433, + "kept_token_count": 572.28046875, + "learning_rate": 7.780000000000002e-06, + "loss": 0.0282, + "mgd_cos_sim": 0.9915983319282532, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008401671843603253, + "mgd_loss_weight": 0.987182903289795, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 194.1083541870117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.8733215332031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 390, + "valid_token_count": 817.99140625 + }, + { + "actual_token_mask_ratio": 0.29981414079666135, + "epoch": 0.11638056444573756, + "flow_matching_loss": 0.017976003885269164, + "grad_norm": 0.09829358756542206, + "kept_token_count": 572.72734375, + "learning_rate": 7.98e-06, + "loss": 0.0264, + "mgd_cos_sim": 0.9914295256137848, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008570479508489371, + "mgd_loss_weight": 0.9868495643138886, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 195.4292465209961, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.09437561035156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 400, + "valid_token_count": 817.96796875 + }, + { + "actual_token_mask_ratio": 0.2999721109867096, + "epoch": 0.11929007855688101, + "flow_matching_loss": 0.01691245539113879, + "grad_norm": 0.09057634323835373, + "kept_token_count": 572.665625, + "learning_rate": 8.18e-06, + "loss": 0.0256, + "mgd_cos_sim": 0.9912079989910125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008792001008987426, + "mgd_loss_weight": 0.9865162193775177, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 192.82005615234374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.604736328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 410, + "valid_token_count": 818.059375 + }, + { + "actual_token_mask_ratio": 0.3000555783510208, + "epoch": 0.12219959266802444, + "flow_matching_loss": 0.017917842883616687, + "grad_norm": 0.0918676108121872, + "kept_token_count": 572.69296875, + "learning_rate": 8.38e-06, + "loss": 0.0265, + "mgd_cos_sim": 0.9913364529609681, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008663552161306144, + "mgd_loss_weight": 0.9861828684806824, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 195.02977142333984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.38136901855466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 420, + "valid_token_count": 818.19765625 + }, + { + "actual_token_mask_ratio": 0.29959396123886106, + "epoch": 0.12510910677916787, + "flow_matching_loss": 0.019650856871157883, + "grad_norm": 0.0912281721830368, + "kept_token_count": 573.13671875, + "learning_rate": 8.58e-06, + "loss": 0.0285, + "mgd_cos_sim": 0.9910504639148712, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008949543116614223, + "mgd_loss_weight": 0.985849529504776, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 197.24274749755858, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.0780090332031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 430, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.30010594725608825, + "epoch": 0.12801862089031132, + "flow_matching_loss": 0.019737045653164385, + "grad_norm": 0.09940160065889359, + "kept_token_count": 572.72734375, + "learning_rate": 8.78e-06, + "loss": 0.0287, + "mgd_cos_sim": 0.9909236669540405, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009076334489509463, + "mgd_loss_weight": 0.9855161845684052, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 197.30914459228515, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.01458740234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 440, + "valid_token_count": 818.31171875 + }, + { + "actual_token_mask_ratio": 0.30083577930927274, + "epoch": 0.13092813500145475, + "flow_matching_loss": 0.018973677791655062, + "grad_norm": 0.08461462706327438, + "kept_token_count": 572.021875, + "learning_rate": 8.98e-06, + "loss": 0.0273, + "mgd_cos_sim": 0.9915483117103576, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008451682422310114, + "mgd_loss_weight": 0.9851828396320343, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 200.7408004760742, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.5640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 450, + "valid_token_count": 818.15078125 + }, + { + "actual_token_mask_ratio": 0.30025129318237304, + "epoch": 0.13383764911259818, + "flow_matching_loss": 0.0174494163133204, + "grad_norm": 0.09840268641710281, + "kept_token_count": 572.52890625, + "learning_rate": 9.18e-06, + "loss": 0.0259, + "mgd_cos_sim": 0.9914588570594788, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008541137352585793, + "mgd_loss_weight": 0.9848494946956634, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 194.28792877197264, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.5023712158203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 460, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.2999340772628784, + "epoch": 0.13674716322374164, + "flow_matching_loss": 0.018720953539013863, + "grad_norm": 0.08969737589359283, + "kept_token_count": 572.84296875, + "learning_rate": 9.38e-06, + "loss": 0.0274, + "mgd_cos_sim": 0.9911572217941285, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008842788077890873, + "mgd_loss_weight": 0.9845161497592926, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 202.04287567138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.48980407714845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 470, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.2998578369617462, + "epoch": 0.13965667733488507, + "flow_matching_loss": 0.01783325495198369, + "grad_norm": 0.08967273682355881, + "kept_token_count": 573.18203125, + "learning_rate": 9.580000000000001e-06, + "loss": 0.0262, + "mgd_cos_sim": 0.9915382623672485, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008461718773469329, + "mgd_loss_weight": 0.9841828048229218, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 205.34719543457032, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1835418701172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 480, + "valid_token_count": 818.66171875 + }, + { + "actual_token_mask_ratio": 0.30071536600589754, + "epoch": 0.1425661914460285, + "flow_matching_loss": 0.017506768554449083, + "grad_norm": 0.09768211096525192, + "kept_token_count": 572.18046875, + "learning_rate": 9.780000000000001e-06, + "loss": 0.0259, + "mgd_cos_sim": 0.9914737820625306, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008526235911995172, + "mgd_loss_weight": 0.9838494658470154, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 202.3220245361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.7251678466797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 490, + "valid_token_count": 818.23046875 + }, + { + "actual_token_mask_ratio": 0.30004405975341797, + "epoch": 0.14547570555717196, + "flow_matching_loss": 0.018386270571500062, + "grad_norm": 0.09573929756879807, + "kept_token_count": 572.81875, + "learning_rate": 9.980000000000001e-06, + "loss": 0.0268, + "mgd_cos_sim": 0.9914490640163421, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008550929836928844, + "mgd_loss_weight": 0.9835161149501801, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 203.83204193115233, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.112841796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 500, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.3011317282915115, + "epoch": 0.1483852196683154, + "flow_matching_loss": 0.018217719998210667, + "grad_norm": 0.09798221290111542, + "kept_token_count": 571.94375, + "learning_rate": 1.0179999999999999e-05, + "loss": 0.0269, + "mgd_cos_sim": 0.9911682069301605, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008831794466823339, + "mgd_loss_weight": 0.9831827700138092, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 203.1528549194336, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.87177734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 510, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.2999028950929642, + "epoch": 0.15129473377945882, + "flow_matching_loss": 0.016227405052632093, + "grad_norm": 0.09487675875425339, + "kept_token_count": 572.92734375, + "learning_rate": 1.0379999999999999e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.9913887441158294, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008611258026212454, + "mgd_loss_weight": 0.9828494310379028, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 202.78286590576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1462371826172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 520, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.2991794228553772, + "epoch": 0.15420424789060228, + "flow_matching_loss": 0.015192249324172735, + "grad_norm": 0.08367598056793213, + "kept_token_count": 573.515625, + "learning_rate": 1.058e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9917013943195343, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008298613503575325, + "mgd_loss_weight": 0.9825160801410675, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 203.0484634399414, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.3684020996094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 530, + "valid_token_count": 818.34609375 + }, + { + "actual_token_mask_ratio": 0.2999247878789902, + "epoch": 0.1571137620017457, + "flow_matching_loss": 0.018571688327938318, + "grad_norm": 0.08903180807828903, + "kept_token_count": 572.9921875, + "learning_rate": 1.078e-05, + "loss": 0.0269, + "mgd_cos_sim": 0.9914849519729614, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008515045046806335, + "mgd_loss_weight": 0.9821827411651611, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 207.81558837890626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4987426757813, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 540, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.3000282019376755, + "epoch": 0.16002327611288913, + "flow_matching_loss": 0.01685401229187846, + "grad_norm": 0.09864240884780884, + "kept_token_count": 572.80703125, + "learning_rate": 1.098e-05, + "loss": 0.0252, + "mgd_cos_sim": 0.9914991319179535, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008500887453556061, + "mgd_loss_weight": 0.9818493962287903, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 207.11953430175782, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.162744140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 550, + "valid_token_count": 818.32890625 + }, + { + "actual_token_mask_ratio": 0.3004197984933853, + "epoch": 0.1629327902240326, + "flow_matching_loss": 0.015302309393882751, + "grad_norm": 0.0822741761803627, + "kept_token_count": 572.43828125, + "learning_rate": 1.118e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9917700886726379, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008229924971237779, + "mgd_loss_weight": 0.9815160453319549, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 214.74059600830077, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.8814971923828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 560, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.29989812672138216, + "epoch": 0.16584230433517602, + "flow_matching_loss": 0.01839094525203109, + "grad_norm": 0.09893855452537537, + "kept_token_count": 572.95234375, + "learning_rate": 1.1380000000000001e-05, + "loss": 0.027, + "mgd_cos_sim": 0.991274356842041, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00872565764002502, + "mgd_loss_weight": 0.9811827123165131, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 207.02478485107423, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.2782348632812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 570, + "valid_token_count": 818.3828125 + }, + { + "actual_token_mask_ratio": 0.3000322014093399, + "epoch": 0.16875181844631945, + "flow_matching_loss": 0.016570845060050486, + "grad_norm": 0.09326931089162827, + "kept_token_count": 572.70078125, + "learning_rate": 1.1580000000000001e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.9916750133037567, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008324987115338444, + "mgd_loss_weight": 0.9808493554592133, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 208.2086624145508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.3328460693359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 580, + "valid_token_count": 818.18046875 + }, + { + "actual_token_mask_ratio": 0.29976330399513246, + "epoch": 0.1716613325574629, + "flow_matching_loss": 0.018586072605103254, + "grad_norm": 0.08105432242155075, + "kept_token_count": 573.0828125, + "learning_rate": 1.178e-05, + "loss": 0.027, + "mgd_cos_sim": 0.9914127469062806, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008587264455854893, + "mgd_loss_weight": 0.9805160164833069, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 211.53658447265624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.67095947265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 590, + "valid_token_count": 818.41484375 + }, + { + "actual_token_mask_ratio": 0.2996436059474945, + "epoch": 0.17457084666860634, + "flow_matching_loss": 0.01713618664070964, + "grad_norm": 0.11036527901887894, + "kept_token_count": 573.09375, + "learning_rate": 1.198e-05, + "loss": 0.0256, + "mgd_cos_sim": 0.9913201212882996, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008679875498637557, + "mgd_loss_weight": 0.9801826775074005, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 214.15314636230468, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.49097595214846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 600, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.3002097815275192, + "epoch": 0.17748036077974977, + "flow_matching_loss": 0.01618697540834546, + "grad_norm": 0.09376242011785507, + "kept_token_count": 572.41875, + "learning_rate": 1.2180000000000002e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9917263746261596, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008273617224767804, + "mgd_loss_weight": 0.9798493206501007, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 213.97198638916015, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.2684051513672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 610, + "valid_token_count": 817.9875 + }, + { + "actual_token_mask_ratio": 0.3007974594831467, + "epoch": 0.18038987489089323, + "flow_matching_loss": 0.016868201177567245, + "grad_norm": 0.10207559913396835, + "kept_token_count": 572.0375, + "learning_rate": 1.238e-05, + "loss": 0.0253, + "mgd_cos_sim": 0.9913590312004089, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008640959160402417, + "mgd_loss_weight": 0.9795159876346589, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 222.7941131591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.0796630859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 620, + "valid_token_count": 818.12265625 + }, + { + "actual_token_mask_ratio": 0.3004536211490631, + "epoch": 0.18329938900203666, + "flow_matching_loss": 0.016281219851225614, + "grad_norm": 0.09185953438282013, + "kept_token_count": 572.35625, + "learning_rate": 1.258e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9918486535549164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008151340670883656, + "mgd_loss_weight": 0.9791826367378235, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 217.37183227539063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.219775390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 630, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.29965119659900663, + "epoch": 0.18620890311318009, + "flow_matching_loss": 0.01674176435917616, + "grad_norm": 0.09244415163993835, + "kept_token_count": 573.12265625, + "learning_rate": 1.278e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.9919209539890289, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008079044753685594, + "mgd_loss_weight": 0.9788492918014526, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 215.68863372802736, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.6341583251953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 640, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.2996652126312256, + "epoch": 0.18911841722432354, + "flow_matching_loss": 0.01658902959898114, + "grad_norm": 0.09609563648700714, + "kept_token_count": 572.8859375, + "learning_rate": 1.298e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9913700520992279, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008629959495738148, + "mgd_loss_weight": 0.9785159528255463, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 218.30524291992188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.4223327636719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 650, + "valid_token_count": 818.015625 + }, + { + "actual_token_mask_ratio": 0.30029759705066683, + "epoch": 0.19202793133546697, + "flow_matching_loss": 0.014477152936160564, + "grad_norm": 0.08827081322669983, + "kept_token_count": 572.58984375, + "learning_rate": 1.3180000000000001e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9916854083538056, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00831458386965096, + "mgd_loss_weight": 0.978182601928711, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 220.8851287841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7304748535156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 660, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.30059835612773894, + "epoch": 0.1949374454466104, + "flow_matching_loss": 0.019673359580338, + "grad_norm": 0.08023352921009064, + "kept_token_count": 572.4484375, + "learning_rate": 1.338e-05, + "loss": 0.0282, + "mgd_cos_sim": 0.9912321209907532, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008767887204885482, + "mgd_loss_weight": 0.9778492629528046, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 226.97225341796874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.91941833496094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 670, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.30001629889011383, + "epoch": 0.19784695955775386, + "flow_matching_loss": 0.016981207206845282, + "grad_norm": 0.09237589687108994, + "kept_token_count": 572.75, + "learning_rate": 1.358e-05, + "loss": 0.025, + "mgd_cos_sim": 0.991813188791275, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008186796167865395, + "mgd_loss_weight": 0.9775159180164337, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 221.64757232666017, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.4463806152344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 680, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.29996550977230074, + "epoch": 0.2007564736688973, + "flow_matching_loss": 0.015475280303508043, + "grad_norm": 0.09632504731416702, + "kept_token_count": 572.93046875, + "learning_rate": 1.378e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9917451024055481, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008254903834313154, + "mgd_loss_weight": 0.9771825730800628, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 230.26202697753905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.79906005859374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 690, + "valid_token_count": 818.4296875 + }, + { + "actual_token_mask_ratio": 0.30063059329986574, + "epoch": 0.20366598778004075, + "flow_matching_loss": 0.015489729400724173, + "grad_norm": 0.10739781707525253, + "kept_token_count": 572.3953125, + "learning_rate": 1.3980000000000002e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9918319523334503, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008168047340586782, + "mgd_loss_weight": 0.976849228143692, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 219.6351119995117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.2512176513672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 700, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.29977903962135316, + "epoch": 0.20657550189118418, + "flow_matching_loss": 0.016447003092616798, + "grad_norm": 0.08710160106420517, + "kept_token_count": 572.93515625, + "learning_rate": 1.4180000000000001e-05, + "loss": 0.0251, + "mgd_cos_sim": 0.9911557734012604, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008844242803752423, + "mgd_loss_weight": 0.9765158891677856, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 221.6120864868164, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.80185852050784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 710, + "valid_token_count": 818.22421875 + }, + { + "actual_token_mask_ratio": 0.29991355240345, + "epoch": 0.2094850160023276, + "flow_matching_loss": 0.014723115414381028, + "grad_norm": 0.07994402945041656, + "kept_token_count": 572.7375, + "learning_rate": 1.438e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9920600116252899, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007939981715753674, + "mgd_loss_weight": 0.9761825382709504, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 224.2689178466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.10191345214844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 720, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.30004335641860963, + "epoch": 0.21239453011347106, + "flow_matching_loss": 0.017335518170148133, + "grad_norm": 0.0910758227109909, + "kept_token_count": 572.775, + "learning_rate": 1.458e-05, + "loss": 0.0254, + "mgd_cos_sim": 0.9917224943637848, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008277516812086105, + "mgd_loss_weight": 0.9758491933345794, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 231.74719848632813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.8020538330078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 730, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.3004371613264084, + "epoch": 0.2153040442246145, + "flow_matching_loss": 0.01621797727420926, + "grad_norm": 0.08162742853164673, + "kept_token_count": 572.44453125, + "learning_rate": 1.4779999999999999e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.991619324684143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008380660461261869, + "mgd_loss_weight": 0.975515854358673, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 229.11941223144532, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.39312438964845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 740, + "valid_token_count": 818.28984375 + }, + { + "actual_token_mask_ratio": 0.29967820942401885, + "epoch": 0.21821355833575792, + "flow_matching_loss": 0.015410409215837717, + "grad_norm": 0.09302666038274765, + "kept_token_count": 573.04609375, + "learning_rate": 1.498e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9917472004890442, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008252796391025186, + "mgd_loss_weight": 0.9751825034618378, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 229.10663604736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.12347412109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 750, + "valid_token_count": 818.25859375 + }, + { + "actual_token_mask_ratio": 0.2999436676502228, + "epoch": 0.22112307244690138, + "flow_matching_loss": 0.016365043260157107, + "grad_norm": 0.1190246120095253, + "kept_token_count": 572.71328125, + "learning_rate": 1.518e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9913747072219848, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008625287655740976, + "mgd_loss_weight": 0.9748491644859314, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 221.11355590820312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.49461364746094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 760, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.3001909613609314, + "epoch": 0.2240325865580448, + "flow_matching_loss": 0.017073845025151968, + "grad_norm": 0.10121878981590271, + "kept_token_count": 572.8078125, + "learning_rate": 1.5380000000000002e-05, + "loss": 0.025, + "mgd_cos_sim": 0.99189293384552, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00810706871561706, + "mgd_loss_weight": 0.9745158195495606, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 226.98277435302734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.3592559814453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 770, + "valid_token_count": 818.525 + }, + { + "actual_token_mask_ratio": 0.29948660731315613, + "epoch": 0.22694210066918824, + "flow_matching_loss": 0.016618749219924213, + "grad_norm": 0.09632641077041626, + "kept_token_count": 573.17421875, + "learning_rate": 1.558e-05, + "loss": 0.0249, + "mgd_cos_sim": 0.9915501296520233, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008449881756678223, + "mgd_loss_weight": 0.9741824686527252, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 222.60532836914064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.57249755859374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 780, + "valid_token_count": 818.21484375 + }, + { + "actual_token_mask_ratio": 0.29962012767791746, + "epoch": 0.2298516147803317, + "flow_matching_loss": 0.016817523166537285, + "grad_norm": 0.10152728855609894, + "kept_token_count": 573.07421875, + "learning_rate": 1.578e-05, + "loss": 0.0254, + "mgd_cos_sim": 0.9912164807319641, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008783517684787513, + "mgd_loss_weight": 0.9738491356372834, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 239.76136627197266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.1192352294922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 790, + "valid_token_count": 818.23671875 + }, + { + "actual_token_mask_ratio": 0.29980939328670503, + "epoch": 0.23276112889147513, + "flow_matching_loss": 0.01706353947520256, + "grad_norm": 0.0890803337097168, + "kept_token_count": 573.084375, + "learning_rate": 1.598e-05, + "loss": 0.0255, + "mgd_cos_sim": 0.9913849592208862, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008615039708092808, + "mgd_loss_weight": 0.9735157787799835, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 224.4856414794922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.5772369384766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 800, + "valid_token_count": 818.46640625 + }, + { + "actual_token_mask_ratio": 0.3007220447063446, + "epoch": 0.23567064300261856, + "flow_matching_loss": 0.015349625423550606, + "grad_norm": 0.08811840415000916, + "kept_token_count": 572.09921875, + "learning_rate": 1.618e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9914573609828949, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008542637247592211, + "mgd_loss_weight": 0.9731824398040771, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 237.7928237915039, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.64844665527346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 810, + "valid_token_count": 818.13046875 + }, + { + "actual_token_mask_ratio": 0.2998536854982376, + "epoch": 0.23858015711376201, + "flow_matching_loss": 0.01626686183735728, + "grad_norm": 0.08258960396051407, + "kept_token_count": 572.703125, + "learning_rate": 1.6380000000000002e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.9914618253707885, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008538175839930773, + "mgd_loss_weight": 0.9728491008281708, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 225.24411315917968, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.4713500976562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 820, + "valid_token_count": 817.975 + }, + { + "actual_token_mask_ratio": 0.2991182178258896, + "epoch": 0.24148967122490544, + "flow_matching_loss": 0.01433918746188283, + "grad_norm": 0.07856664806604385, + "kept_token_count": 573.621875, + "learning_rate": 1.658e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9915581047534943, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008441891614347696, + "mgd_loss_weight": 0.9725157439708709, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 236.17435302734376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.6305633544922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 830, + "valid_token_count": 818.43203125 + }, + { + "actual_token_mask_ratio": 0.29964096546173097, + "epoch": 0.24439918533604887, + "flow_matching_loss": 0.017513538338243963, + "grad_norm": 0.0886005088686943, + "kept_token_count": 573.1921875, + "learning_rate": 1.6780000000000002e-05, + "loss": 0.0258, + "mgd_cos_sim": 0.9914851307868957, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008514855429530144, + "mgd_loss_weight": 0.9721824109554291, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 237.0398742675781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.91527099609374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 840, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.30016421973705293, + "epoch": 0.24730869944719233, + "flow_matching_loss": 0.019179216399788857, + "grad_norm": 0.09070885926485062, + "kept_token_count": 572.71796875, + "learning_rate": 1.698e-05, + "loss": 0.0272, + "mgd_cos_sim": 0.9917918860912323, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008208109205588699, + "mgd_loss_weight": 0.9718490600585937, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 237.07872924804687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.6592651367188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 850, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.30011303126811983, + "epoch": 0.25021821355833573, + "flow_matching_loss": 0.01600086521357298, + "grad_norm": 0.09706658869981766, + "kept_token_count": 572.62265625, + "learning_rate": 1.718e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9916443049907684, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00835569780319929, + "mgd_loss_weight": 0.9715157151222229, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 239.43451843261718, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.10809631347655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 860, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.2997750163078308, + "epoch": 0.2531277276694792, + "flow_matching_loss": 0.015939547680318356, + "grad_norm": 0.09356820583343506, + "kept_token_count": 572.88359375, + "learning_rate": 1.7380000000000003e-05, + "loss": 0.0242, + "mgd_cos_sim": 0.9914584875106811, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008541503362357617, + "mgd_loss_weight": 0.9711823761463165, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 236.30876617431642, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.48174743652345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 870, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.299838051199913, + "epoch": 0.25603724178062265, + "flow_matching_loss": 0.016426939703524112, + "grad_norm": 0.10883860290050507, + "kept_token_count": 572.959375, + "learning_rate": 1.758e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.991392171382904, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008607842726632953, + "mgd_loss_weight": 0.9708490252494812, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 242.38714904785155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.1162414550781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 880, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.29984149932861326, + "epoch": 0.2589467558917661, + "flow_matching_loss": 0.015894751995801926, + "grad_norm": 0.08904118090867996, + "kept_token_count": 572.84140625, + "learning_rate": 1.778e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9918092846870422, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0081907257437706, + "mgd_loss_weight": 0.9705156862735749, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 242.94433135986327, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.93841857910155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 890, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.2998840302228928, + "epoch": 0.2618562700029095, + "flow_matching_loss": 0.017295110039412975, + "grad_norm": 0.1094084084033966, + "kept_token_count": 572.80859375, + "learning_rate": 1.798e-05, + "loss": 0.0251, + "mgd_cos_sim": 0.9919311940670014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008068816410377622, + "mgd_loss_weight": 0.9701823413372039, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 245.41105194091796, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.9841705322266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 900, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.29989901185035706, + "epoch": 0.26476578411405294, + "flow_matching_loss": 0.015911236591637134, + "grad_norm": 0.08320693671703339, + "kept_token_count": 572.84140625, + "learning_rate": 1.818e-05, + "loss": 0.0242, + "mgd_cos_sim": 0.9914840161800385, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008515974087640644, + "mgd_loss_weight": 0.9698489904403687, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 239.88990936279296, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.16172180175784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 910, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.30013653039932253, + "epoch": 0.26767529822519637, + "flow_matching_loss": 0.01685321656987071, + "grad_norm": 0.10530459135770798, + "kept_token_count": 572.86171875, + "learning_rate": 1.838e-05, + "loss": 0.0256, + "mgd_cos_sim": 0.9909927546977997, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009007251542061567, + "mgd_loss_weight": 0.9695156514644623, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 234.07848205566407, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.9468933105469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 920, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.2988359421491623, + "epoch": 0.27058481233633985, + "flow_matching_loss": 0.017401762958616018, + "grad_norm": 0.10431922972202301, + "kept_token_count": 573.6, + "learning_rate": 1.8579999999999998e-05, + "loss": 0.0256, + "mgd_cos_sim": 0.9915704607963562, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008429552707821131, + "mgd_loss_weight": 0.9691823065280915, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 235.0296188354492, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.42955627441404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 930, + "valid_token_count": 818.0609375 + }, + { + "actual_token_mask_ratio": 0.3001677244901657, + "epoch": 0.2734943264474833, + "flow_matching_loss": 0.016706756222993134, + "grad_norm": 0.09133479744195938, + "kept_token_count": 572.71953125, + "learning_rate": 1.878e-05, + "loss": 0.0249, + "mgd_cos_sim": 0.9915729880332946, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008426994318142533, + "mgd_loss_weight": 0.9688489615917206, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 234.972509765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.32665100097654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 940, + "valid_token_count": 818.36796875 + }, + { + "actual_token_mask_ratio": 0.3000658303499222, + "epoch": 0.2764038405586267, + "flow_matching_loss": 0.016911908611655237, + "grad_norm": 0.13532161712646484, + "kept_token_count": 572.8453125, + "learning_rate": 1.898e-05, + "loss": 0.0251, + "mgd_cos_sim": 0.9915937125682831, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008406282775104046, + "mgd_loss_weight": 0.9685156166553497, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 243.9560989379883, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.3746002197266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 950, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.299737349152565, + "epoch": 0.27931335466977014, + "flow_matching_loss": 0.017438561469316483, + "grad_norm": 0.09825092554092407, + "kept_token_count": 572.95078125, + "learning_rate": 1.918e-05, + "loss": 0.0261, + "mgd_cos_sim": 0.9910545587539673, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008945426717400552, + "mgd_loss_weight": 0.9681822717189789, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 235.01829376220704, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.9637786865234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 960, + "valid_token_count": 818.19453125 + }, + { + "actual_token_mask_ratio": 0.2996260404586792, + "epoch": 0.28222286878091357, + "flow_matching_loss": 0.016254172008484603, + "grad_norm": 0.0872480571269989, + "kept_token_count": 573.0171875, + "learning_rate": 1.938e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.9916175186634064, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008382482081651687, + "mgd_loss_weight": 0.967848926782608, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 246.97052001953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.6551239013672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 970, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.30005151629447935, + "epoch": 0.285132382892057, + "flow_matching_loss": 0.015980870090425016, + "grad_norm": 0.0848078727722168, + "kept_token_count": 572.65859375, + "learning_rate": 1.958e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9918352544307709, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008164745243266224, + "mgd_loss_weight": 0.9675155878067017, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 238.6302734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.74078369140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 980, + "valid_token_count": 818.14296875 + }, + { + "actual_token_mask_ratio": 0.2998352438211441, + "epoch": 0.2880418970032005, + "flow_matching_loss": 0.01724534258246422, + "grad_norm": 0.08781762421131134, + "kept_token_count": 573.10546875, + "learning_rate": 1.978e-05, + "loss": 0.0253, + "mgd_cos_sim": 0.9916488707065583, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008351111505180597, + "mgd_loss_weight": 0.9671822369098664, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 253.2133987426758, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.3353698730469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 990, + "valid_token_count": 818.525 + }, + { + "actual_token_mask_ratio": 0.2998127996921539, + "epoch": 0.2909514111143439, + "flow_matching_loss": 0.017421597614884377, + "grad_norm": 0.08923622965812683, + "kept_token_count": 572.946875, + "learning_rate": 1.9980000000000002e-05, + "loss": 0.0259, + "mgd_cos_sim": 0.9912436187267304, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008756376057863235, + "mgd_loss_weight": 0.9668488919734954, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 240.4791030883789, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.8916046142578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1000, + "valid_token_count": 818.28203125 + }, + { + "actual_token_mask_ratio": 0.3005711942911148, + "epoch": 0.29386092522548735, + "flow_matching_loss": 0.017317445762455464, + "grad_norm": 0.10735127329826355, + "kept_token_count": 572.2421875, + "learning_rate": 2.018e-05, + "loss": 0.0251, + "mgd_cos_sim": 0.9919224739074707, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008077523158863186, + "mgd_loss_weight": 0.9665155529975891, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 245.5456085205078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.2952056884766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1010, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.2996128022670746, + "epoch": 0.2967704393366308, + "flow_matching_loss": 0.016334470082074403, + "grad_norm": 0.09667499363422394, + "kept_token_count": 573.01640625, + "learning_rate": 2.038e-05, + "loss": 0.0245, + "mgd_cos_sim": 0.9915908992290496, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008409106777980924, + "mgd_loss_weight": 0.9661822021007538, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 245.04581146240236, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.32514038085935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1020, + "valid_token_count": 818.14140625 + }, + { + "actual_token_mask_ratio": 0.299856972694397, + "epoch": 0.2996799534477742, + "flow_matching_loss": 0.016614512261003255, + "grad_norm": 0.12056141346693039, + "kept_token_count": 572.7703125, + "learning_rate": 2.0580000000000003e-05, + "loss": 0.0245, + "mgd_cos_sim": 0.9917988717556, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008201130526140332, + "mgd_loss_weight": 0.9658488631248474, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 248.24874877929688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.70949096679686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1030, + "valid_token_count": 818.078125 + }, + { + "actual_token_mask_ratio": 0.3007429242134094, + "epoch": 0.30258946755891764, + "flow_matching_loss": 0.015666386857628823, + "grad_norm": 0.09555691480636597, + "kept_token_count": 572.2453125, + "learning_rate": 2.078e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9910978496074676, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008902159612625837, + "mgd_loss_weight": 0.9655155181884766, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 248.50113983154296, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.9170654296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1040, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.299749031662941, + "epoch": 0.3054989816700611, + "flow_matching_loss": 0.016908693127334118, + "grad_norm": 0.0912489965558052, + "kept_token_count": 572.88046875, + "learning_rate": 2.0980000000000002e-05, + "loss": 0.0249, + "mgd_cos_sim": 0.9916797161102295, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008320273784920573, + "mgd_loss_weight": 0.9651821672916412, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 240.91158905029297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.36024475097656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1050, + "valid_token_count": 818.1078125 + }, + { + "actual_token_mask_ratio": 0.29898176789283754, + "epoch": 0.30840849578120455, + "flow_matching_loss": 0.016217463836073875, + "grad_norm": 0.10271599143743515, + "kept_token_count": 573.72421875, + "learning_rate": 2.118e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9910977721214295, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008902232814580201, + "mgd_loss_weight": 0.9648488342761994, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 244.49718322753907, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.8421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1060, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.30031162202358247, + "epoch": 0.311318009892348, + "flow_matching_loss": 0.016739480011165143, + "grad_norm": 0.09574141353368759, + "kept_token_count": 572.6203125, + "learning_rate": 2.1380000000000002e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.991796338558197, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008203669218346477, + "mgd_loss_weight": 0.9645154774188995, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 242.8595230102539, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.09644775390626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1070, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.29955527782440183, + "epoch": 0.3142275240034914, + "flow_matching_loss": 0.015465401764959096, + "grad_norm": 0.10974612087011337, + "kept_token_count": 573.2015625, + "learning_rate": 2.1580000000000003e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9916583180427552, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008341695088893175, + "mgd_loss_weight": 0.9641821384429932, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 240.17723083496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.1473327636719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1080, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.3002567857503891, + "epoch": 0.31713703811463484, + "flow_matching_loss": 0.014447444956749678, + "grad_norm": 0.10045810788869858, + "kept_token_count": 572.47265625, + "learning_rate": 2.178e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9919190824031829, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008080914849415421, + "mgd_loss_weight": 0.9638487994670868, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 249.49347839355468, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.4277679443359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1090, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.3005100667476654, + "epoch": 0.32004655222577827, + "flow_matching_loss": 0.016776365134865044, + "grad_norm": 0.11825886368751526, + "kept_token_count": 572.51328125, + "learning_rate": 2.198e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.9917898118495941, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00821017692796886, + "mgd_loss_weight": 0.9635154426097869, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 245.07149963378907, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.1248046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1100, + "valid_token_count": 818.475 + }, + { + "actual_token_mask_ratio": 0.29985194802284243, + "epoch": 0.32295606633692175, + "flow_matching_loss": 0.01814859118312597, + "grad_norm": 0.14141945540905, + "kept_token_count": 573.059375, + "learning_rate": 2.2179999999999998e-05, + "loss": 0.0259, + "mgd_cos_sim": 0.9919218719005585, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008078131079673766, + "mgd_loss_weight": 0.9631821095943451, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 240.96858673095704, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.6336975097656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1110, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.3002129107713699, + "epoch": 0.3258655804480652, + "flow_matching_loss": 0.0157778263092041, + "grad_norm": 0.112735316157341, + "kept_token_count": 572.76015625, + "learning_rate": 2.238e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9915358185768127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008464182680472731, + "mgd_loss_weight": 0.9628487646579742, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 259.69500122070315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.0933898925781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1120, + "valid_token_count": 818.47734375 + }, + { + "actual_token_mask_ratio": 0.3000136137008667, + "epoch": 0.3287750945592086, + "flow_matching_loss": 0.017135406471788885, + "grad_norm": 0.10875969380140305, + "kept_token_count": 572.76875, + "learning_rate": 2.258e-05, + "loss": 0.0254, + "mgd_cos_sim": 0.9914296746253968, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00857033752836287, + "mgd_loss_weight": 0.9625154137611389, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 255.9958938598633, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.23882446289065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1130, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.2998995780944824, + "epoch": 0.33168460867035204, + "flow_matching_loss": 0.015068458393216133, + "grad_norm": 0.10872481018304825, + "kept_token_count": 572.78515625, + "learning_rate": 2.278e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9916971981525421, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008302789181470871, + "mgd_loss_weight": 0.9621820747852325, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 251.88438110351564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.6466491699219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1140, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.30018497109413145, + "epoch": 0.3345941227814955, + "flow_matching_loss": 0.01593236979097128, + "grad_norm": 0.09926165640354156, + "kept_token_count": 572.58203125, + "learning_rate": 2.298e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9913303196430207, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008669673930853605, + "mgd_loss_weight": 0.9618487298488617, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 252.2896530151367, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.9613525390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1150, + "valid_token_count": 818.18984375 + }, + { + "actual_token_mask_ratio": 0.299845364689827, + "epoch": 0.3375036368926389, + "flow_matching_loss": 0.01687541464343667, + "grad_norm": 0.10823974013328552, + "kept_token_count": 572.94921875, + "learning_rate": 2.3179999999999998e-05, + "loss": 0.0254, + "mgd_cos_sim": 0.9910965442657471, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008903450518846511, + "mgd_loss_weight": 0.9615153849124909, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 247.14851837158204, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.7187103271484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1160, + "valid_token_count": 818.31484375 + }, + { + "actual_token_mask_ratio": 0.29979583621025085, + "epoch": 0.3404131510037824, + "flow_matching_loss": 0.0169470326974988, + "grad_norm": 0.07995817065238953, + "kept_token_count": 572.98359375, + "learning_rate": 2.338e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9915959656238555, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008404020080342888, + "mgd_loss_weight": 0.96118203997612, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 247.62306518554686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.91697692871094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1170, + "valid_token_count": 818.31171875 + }, + { + "actual_token_mask_ratio": 0.2999196946620941, + "epoch": 0.3433226651149258, + "flow_matching_loss": 0.01779107069596648, + "grad_norm": 0.09602982550859451, + "kept_token_count": 572.7296875, + "learning_rate": 2.358e-05, + "loss": 0.0263, + "mgd_cos_sim": 0.9911300063133239, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008869984839111566, + "mgd_loss_weight": 0.9608486950397491, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 249.76672973632813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.5853637695312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1180, + "valid_token_count": 818.09140625 + }, + { + "actual_token_mask_ratio": 0.2998834580183029, + "epoch": 0.34623217922606925, + "flow_matching_loss": 0.015548764634877444, + "grad_norm": 0.08347360789775848, + "kept_token_count": 573.05703125, + "learning_rate": 2.378e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9917567789554596, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008243213780224324, + "mgd_loss_weight": 0.9605153501033783, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 252.8892028808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.5089965820313, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1190, + "valid_token_count": 818.51796875 + }, + { + "actual_token_mask_ratio": 0.2996847331523895, + "epoch": 0.3491416933372127, + "flow_matching_loss": 0.016864996682852507, + "grad_norm": 0.10157416015863419, + "kept_token_count": 573.23046875, + "learning_rate": 2.398e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9915300130844116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008469992270693182, + "mgd_loss_weight": 0.9601820111274719, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 252.98135681152343, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.9714416503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1200, + "valid_token_count": 818.52734375 + }, + { + "actual_token_mask_ratio": 0.29998424649238586, + "epoch": 0.3520512074483561, + "flow_matching_loss": 0.017066076304763554, + "grad_norm": 0.11750932037830353, + "kept_token_count": 572.6890625, + "learning_rate": 2.4180000000000002e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9917093873023987, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008290620613843202, + "mgd_loss_weight": 0.9598486602306366, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 255.82512664794922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.5738159179688, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1210, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.3002630233764648, + "epoch": 0.35496072155949954, + "flow_matching_loss": 0.01672248849645257, + "grad_norm": 0.10529099404811859, + "kept_token_count": 572.6046875, + "learning_rate": 2.438e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9915318787097931, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008468125155195594, + "mgd_loss_weight": 0.9595153152942657, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 262.08411865234376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.9258209228516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1220, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.29979958236217497, + "epoch": 0.357870235670643, + "flow_matching_loss": 0.01628937441855669, + "grad_norm": 0.11002469062805176, + "kept_token_count": 573.25, + "learning_rate": 2.4580000000000002e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.99157794713974, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008422056958079339, + "mgd_loss_weight": 0.9591819763183593, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 263.00146484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.83900146484376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1230, + "valid_token_count": 818.68984375 + }, + { + "actual_token_mask_ratio": 0.30032953917980193, + "epoch": 0.36077974978178645, + "flow_matching_loss": 0.01637560622766614, + "grad_norm": 0.09177234768867493, + "kept_token_count": 572.7015625, + "learning_rate": 2.478e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.9916822373867035, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008317763078957796, + "mgd_loss_weight": 0.958848625421524, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 252.6610076904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.16657104492185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1240, + "valid_token_count": 818.5296875 + }, + { + "actual_token_mask_ratio": 0.3006243646144867, + "epoch": 0.3636892638929299, + "flow_matching_loss": 0.017113596852868795, + "grad_norm": 0.10541198402643204, + "kept_token_count": 572.30859375, + "learning_rate": 2.498e-05, + "loss": 0.0254, + "mgd_cos_sim": 0.9913849413394928, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00861506862565875, + "mgd_loss_weight": 0.9585152864456177, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 261.8005004882813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.88588256835936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1250, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.2993743747472763, + "epoch": 0.3665987780040733, + "flow_matching_loss": 0.01686019208282232, + "grad_norm": 0.09049248695373535, + "kept_token_count": 573.37734375, + "learning_rate": 2.5180000000000003e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9914702951908112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00852970709092915, + "mgd_loss_weight": 0.9581819415092468, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 257.7563934326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.53468933105466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1260, + "valid_token_count": 818.38203125 + }, + { + "actual_token_mask_ratio": 0.2990357369184494, + "epoch": 0.36950829211521674, + "flow_matching_loss": 0.015614565927535295, + "grad_norm": 0.1130000501871109, + "kept_token_count": 573.50625, + "learning_rate": 2.538e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9915231823921203, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008476814674213528, + "mgd_loss_weight": 0.9578485906124115, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 262.8044921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.96936950683596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1270, + "valid_token_count": 818.17109375 + }, + { + "actual_token_mask_ratio": 0.2997406780719757, + "epoch": 0.37241780622636017, + "flow_matching_loss": 0.018588507641106845, + "grad_norm": 0.09754151105880737, + "kept_token_count": 572.9578125, + "learning_rate": 2.5580000000000002e-05, + "loss": 0.027, + "mgd_cos_sim": 0.9912062764167786, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008793718786910177, + "mgd_loss_weight": 0.9575152575969696, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 257.1968933105469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.9339324951172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1280, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.2997212171554565, + "epoch": 0.37532732033750366, + "flow_matching_loss": 0.01540938476100564, + "grad_norm": 0.0935387834906578, + "kept_token_count": 573.16875, + "learning_rate": 2.578e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.991220909357071, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00877910372801125, + "mgd_loss_weight": 0.9571819007396698, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 265.3660858154297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.86549987792966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1290, + "valid_token_count": 818.4890625 + }, + { + "actual_token_mask_ratio": 0.300834122300148, + "epoch": 0.3782368344486471, + "flow_matching_loss": 0.01648285398259759, + "grad_norm": 0.1026204526424408, + "kept_token_count": 572.2484375, + "learning_rate": 2.5980000000000002e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.9917608678340912, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008239133283495903, + "mgd_loss_weight": 0.9568485617637634, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 267.1064117431641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.9766906738281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1300, + "valid_token_count": 818.4765625 + }, + { + "actual_token_mask_ratio": 0.3000852733850479, + "epoch": 0.3811463485597905, + "flow_matching_loss": 0.016653112322092056, + "grad_norm": 0.0894680917263031, + "kept_token_count": 572.49453125, + "learning_rate": 2.618e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9914366543293, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008563353307545185, + "mgd_loss_weight": 0.9565152227878571, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 262.28124847412107, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.6527984619141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1310, + "valid_token_count": 817.94375 + }, + { + "actual_token_mask_ratio": 0.29980874955654147, + "epoch": 0.38405586267093395, + "flow_matching_loss": 0.01621163608506322, + "grad_norm": 0.09096764028072357, + "kept_token_count": 572.99921875, + "learning_rate": 2.6379999999999998e-05, + "loss": 0.0245, + "mgd_cos_sim": 0.9913500964641571, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008649907354265452, + "mgd_loss_weight": 0.9561818659305572, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 264.9293960571289, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.9725067138672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1320, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.30034993290901185, + "epoch": 0.3869653767820774, + "flow_matching_loss": 0.01693888073787093, + "grad_norm": 0.0834357813000679, + "kept_token_count": 572.43671875, + "learning_rate": 2.658e-05, + "loss": 0.0252, + "mgd_cos_sim": 0.9913167655467987, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008683240925893187, + "mgd_loss_weight": 0.9558485329151154, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 253.60105895996094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.7498046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1330, + "valid_token_count": 818.17421875 + }, + { + "actual_token_mask_ratio": 0.300553959608078, + "epoch": 0.3898748908932208, + "flow_matching_loss": 0.016253928188234568, + "grad_norm": 0.10236327350139618, + "kept_token_count": 572.3203125, + "learning_rate": 2.678e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.9912682473659515, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008731749653816224, + "mgd_loss_weight": 0.95551518201828, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 256.856689453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.8489959716797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1340, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.3005328714847565, + "epoch": 0.3927844050043643, + "flow_matching_loss": 0.01757618971168995, + "grad_norm": 0.11572469025850296, + "kept_token_count": 572.3078125, + "learning_rate": 2.698e-05, + "loss": 0.0263, + "mgd_cos_sim": 0.9908832669258117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009116711700335145, + "mgd_loss_weight": 0.9551818370819092, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 255.8961151123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.84443359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1350, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.2990217238664627, + "epoch": 0.3956939191155077, + "flow_matching_loss": 0.01650331951677799, + "grad_norm": 0.10158885270357132, + "kept_token_count": 573.49375, + "learning_rate": 2.718e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9910700500011445, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0089299529325217, + "mgd_loss_weight": 0.9548484981060028, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 273.41005859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.128125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1360, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.30003386437892915, + "epoch": 0.39860343322665115, + "flow_matching_loss": 0.016873825900256635, + "grad_norm": 0.1102931797504425, + "kept_token_count": 572.62734375, + "learning_rate": 2.738e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9914389669895172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008561043441295624, + "mgd_loss_weight": 0.9545151472091675, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 264.4633041381836, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.17076416015624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1370, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.29993888437747956, + "epoch": 0.4015129473377946, + "flow_matching_loss": 0.01450425200164318, + "grad_norm": 0.14097380638122559, + "kept_token_count": 572.74375, + "learning_rate": 2.758e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.990959084033966, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009040919691324234, + "mgd_loss_weight": 0.9541818082332612, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 263.5163055419922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.83428344726565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1380, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.300038093328476, + "epoch": 0.404422461448938, + "flow_matching_loss": 0.01658584186807275, + "grad_norm": 0.07886609435081482, + "kept_token_count": 572.59609375, + "learning_rate": 2.778e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9911937952041626, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008806208567693829, + "mgd_loss_weight": 0.9538484632968902, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 268.7933151245117, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.58929748535155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1390, + "valid_token_count": 818.04375 + }, + { + "actual_token_mask_ratio": 0.3003676444292068, + "epoch": 0.4073319755600815, + "flow_matching_loss": 0.016015144530683756, + "grad_norm": 0.11031931638717651, + "kept_token_count": 572.2890625, + "learning_rate": 2.798e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9916059255599976, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008394054416567088, + "mgd_loss_weight": 0.9535151124000549, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 262.0269577026367, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.10606994628904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1400, + "valid_token_count": 817.98203125 + }, + { + "actual_token_mask_ratio": 0.3003917157649994, + "epoch": 0.4102414896712249, + "flow_matching_loss": 0.018419509567320348, + "grad_norm": 0.1119493767619133, + "kept_token_count": 572.625, + "learning_rate": 2.818e-05, + "loss": 0.0272, + "mgd_cos_sim": 0.9907939791679382, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009206027165055275, + "mgd_loss_weight": 0.9531817734241486, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 256.54892578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.7403869628906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1410, + "valid_token_count": 818.49609375 + }, + { + "actual_token_mask_ratio": 0.29981389343738557, + "epoch": 0.41315100378236835, + "flow_matching_loss": 0.015301148500293493, + "grad_norm": 0.0834006741642952, + "kept_token_count": 572.9859375, + "learning_rate": 2.838e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.991595321893692, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00840468592941761, + "mgd_loss_weight": 0.9528484284877777, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 262.39034576416014, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0747131347656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1420, + "valid_token_count": 818.33203125 + }, + { + "actual_token_mask_ratio": 0.3001969248056412, + "epoch": 0.4160605178935118, + "flow_matching_loss": 0.01698862873017788, + "grad_norm": 0.10105960071086884, + "kept_token_count": 572.78125, + "learning_rate": 2.858e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9918334245681762, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008166572824120522, + "mgd_loss_weight": 0.9525150835514069, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 276.01963500976564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.1546234130859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1430, + "valid_token_count": 818.49140625 + }, + { + "actual_token_mask_ratio": 0.3004276782274246, + "epoch": 0.4189700320046552, + "flow_matching_loss": 0.01582336025312543, + "grad_norm": 0.08618830889463425, + "kept_token_count": 572.26484375, + "learning_rate": 2.8780000000000002e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9915618240833283, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00843818224966526, + "mgd_loss_weight": 0.952181738615036, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 269.66192321777345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.41058044433595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1440, + "valid_token_count": 818.02109375 + }, + { + "actual_token_mask_ratio": 0.3006235271692276, + "epoch": 0.42187954611579864, + "flow_matching_loss": 0.016862942557781934, + "grad_norm": 0.09708704054355621, + "kept_token_count": 572.29921875, + "learning_rate": 2.898e-05, + "loss": 0.0256, + "mgd_cos_sim": 0.9908486783504487, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00915132910013199, + "mgd_loss_weight": 0.9518483936786651, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 262.73133087158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.9962097167969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1450, + "valid_token_count": 818.30078125 + }, + { + "actual_token_mask_ratio": 0.3005703330039978, + "epoch": 0.42478906022694213, + "flow_matching_loss": 0.01731217885389924, + "grad_norm": 0.09158945828676224, + "kept_token_count": 572.315625, + "learning_rate": 2.9180000000000002e-05, + "loss": 0.0255, + "mgd_cos_sim": 0.9914141654968261, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008585830591619015, + "mgd_loss_weight": 0.9515150487422943, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 280.98362121582034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.1955627441406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1460, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.29950188398361205, + "epoch": 0.42769857433808556, + "flow_matching_loss": 0.0189359393902123, + "grad_norm": 0.1083645150065422, + "kept_token_count": 573.278125, + "learning_rate": 2.938e-05, + "loss": 0.0273, + "mgd_cos_sim": 0.9912164330482482, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008783542551100254, + "mgd_loss_weight": 0.9511817097663879, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 268.27451171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.939990234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1470, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.2997397243976593, + "epoch": 0.430608088449229, + "flow_matching_loss": 0.016364551428705455, + "grad_norm": 0.11315958946943283, + "kept_token_count": 573.1546875, + "learning_rate": 2.958e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.9915380954742432, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008461909415200353, + "mgd_loss_weight": 0.9508483588695527, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 269.1293579101563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.9511260986328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1480, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.2992875397205353, + "epoch": 0.4335176025603724, + "flow_matching_loss": 0.015896924026310443, + "grad_norm": 0.10753830522298813, + "kept_token_count": 573.3984375, + "learning_rate": 2.9780000000000003e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.9910841405391693, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008915858622640372, + "mgd_loss_weight": 0.9505150139331817, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 277.1376434326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.3438385009766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1490, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.29963871240615847, + "epoch": 0.43642711667151585, + "flow_matching_loss": 0.015857338532805444, + "grad_norm": 0.11383417993783951, + "kept_token_count": 573.11171875, + "learning_rate": 2.998e-05, + "loss": 0.0242, + "mgd_cos_sim": 0.9911748170852661, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00882519637234509, + "mgd_loss_weight": 0.9501816749572753, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 266.8564422607422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.0644256591797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1500, + "valid_token_count": 818.30703125 + }, + { + "actual_token_mask_ratio": 0.299945142865181, + "epoch": 0.4393366307826593, + "flow_matching_loss": 0.015942322369664908, + "grad_norm": 0.11037416756153107, + "kept_token_count": 573.0515625, + "learning_rate": 2.9999992618302022e-05, + "loss": 0.0249, + "mgd_cos_sim": 0.990544593334198, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00945540820248425, + "mgd_loss_weight": 0.9498483240604401, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 255.9006546020508, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.06630859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1510, + "valid_token_count": 818.57890625 + }, + { + "actual_token_mask_ratio": 0.2994301915168762, + "epoch": 0.44224614489380276, + "flow_matching_loss": 0.016553532518446445, + "grad_norm": 0.10635771602392197, + "kept_token_count": 573.46484375, + "learning_rate": 2.999996710133069e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9911218881607056, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008878100942820311, + "mgd_loss_weight": 0.9495149850845337, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 272.07749328613284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.8804901123047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1520, + "valid_token_count": 818.56953125 + }, + { + "actual_token_mask_ratio": 0.2998373955488205, + "epoch": 0.4451556590049462, + "flow_matching_loss": 0.015699005126953124, + "grad_norm": 0.10143670439720154, + "kept_token_count": 572.74375, + "learning_rate": 2.9999923357984923e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9913027584552765, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008697231952100992, + "mgd_loss_weight": 0.9491816401481629, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 261.1240524291992, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.11554260253905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1530, + "valid_token_count": 818.015625 + }, + { + "actual_token_mask_ratio": 0.30023249685764314, + "epoch": 0.4480651731160896, + "flow_matching_loss": 0.01589994328096509, + "grad_norm": 0.09209704399108887, + "kept_token_count": 572.43671875, + "learning_rate": 2.9999861388317874e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9913433194160461, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008656670153141022, + "mgd_loss_weight": 0.9488482892513275, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 266.47070770263673, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.9807983398438, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1540, + "valid_token_count": 818.03671875 + }, + { + "actual_token_mask_ratio": 0.29980834424495695, + "epoch": 0.45097468722723305, + "flow_matching_loss": 0.017883027065545322, + "grad_norm": 0.1298646181821823, + "kept_token_count": 572.8125, + "learning_rate": 2.9999781192404847e-05, + "loss": 0.0262, + "mgd_cos_sim": 0.9911964654922485, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00880352407693863, + "mgd_loss_weight": 0.9485149562358857, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 284.49156188964844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.2976837158203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1550, + "valid_token_count": 818.078125 + }, + { + "actual_token_mask_ratio": 0.2995987743139267, + "epoch": 0.4538842013383765, + "flow_matching_loss": 0.01595157105475664, + "grad_norm": 0.0993209257721901, + "kept_token_count": 573.19296875, + "learning_rate": 2.9999682770343283e-05, + "loss": 0.0242, + "mgd_cos_sim": 0.991316843032837, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008683155570179225, + "mgd_loss_weight": 0.9481816053390503, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 260.92444763183596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.3062774658203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1560, + "valid_token_count": 818.37578125 + }, + { + "actual_token_mask_ratio": 0.29968576431274413, + "epoch": 0.4567937154495199, + "flow_matching_loss": 0.0179245064035058, + "grad_norm": 0.1121869832277298, + "kept_token_count": 573.08515625, + "learning_rate": 2.9999566122252777e-05, + "loss": 0.026, + "mgd_cos_sim": 0.9914933025836945, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008506686193868518, + "mgd_loss_weight": 0.9478482604026794, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 291.1995330810547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.7280029296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1570, + "valid_token_count": 818.31953125 + }, + { + "actual_token_mask_ratio": 0.29989719986915586, + "epoch": 0.4597032295606634, + "flow_matching_loss": 0.01588129149749875, + "grad_norm": 0.10451764613389969, + "kept_token_count": 572.9109375, + "learning_rate": 2.9999431248275065e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9911440849304199, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008855918236076833, + "mgd_loss_weight": 0.9475149214267731, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 285.34999694824216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.68788452148436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1580, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.29978262484073637, + "epoch": 0.4626127436718068, + "flow_matching_loss": 0.015669688116759062, + "grad_norm": 0.08788777887821198, + "kept_token_count": 572.94453125, + "learning_rate": 2.9999278148574027e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9912653207778931, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00873467680066824, + "mgd_loss_weight": 0.9471815705299378, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 276.90956726074216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.5198699951172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1590, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.30038401782512664, + "epoch": 0.46552225778295026, + "flow_matching_loss": 0.016861907485872508, + "grad_norm": 0.09720059484243393, + "kept_token_count": 572.540625, + "learning_rate": 2.9999106823335705e-05, + "loss": 0.0259, + "mgd_cos_sim": 0.9904808759689331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009519124310463667, + "mgd_loss_weight": 0.9468482315540314, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 274.1320007324219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.86810302734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1600, + "valid_token_count": 818.36171875 + }, + { + "actual_token_mask_ratio": 0.3000356048345566, + "epoch": 0.4684317718940937, + "flow_matching_loss": 0.016321185044944286, + "grad_norm": 0.08649581670761108, + "kept_token_count": 572.83203125, + "learning_rate": 2.999891727276827e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9910890638828278, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00891093909740448, + "mgd_loss_weight": 0.9465148866176605, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 269.62953186035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.32025146484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1610, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.29988362491130827, + "epoch": 0.4713412860052371, + "flow_matching_loss": 0.014260740950703621, + "grad_norm": 0.09386331588029861, + "kept_token_count": 572.66875, + "learning_rate": 2.9998709497102037e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9915980935096741, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008401908120140434, + "mgd_loss_weight": 0.9461815357208252, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 276.60479736328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.92530822753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1620, + "valid_token_count": 817.9609375 + }, + { + "actual_token_mask_ratio": 0.2993158042430878, + "epoch": 0.47425080011638054, + "flow_matching_loss": 0.017450817674398423, + "grad_norm": 0.1133531779050827, + "kept_token_count": 573.3515625, + "learning_rate": 2.9998483496589485e-05, + "loss": 0.0259, + "mgd_cos_sim": 0.9910925209522248, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00890746172517538, + "mgd_loss_weight": 0.9458481967449188, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 273.13157653808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.720458984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1630, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.300342270731926, + "epoch": 0.47716031422752403, + "flow_matching_loss": 0.016791606973856687, + "grad_norm": 0.10638977587223053, + "kept_token_count": 572.55546875, + "learning_rate": 2.9998239271505217e-05, + "loss": 0.0254, + "mgd_cos_sim": 0.9908868551254273, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009113143105059862, + "mgd_loss_weight": 0.945514851808548, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 287.777685546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.22792053222656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1640, + "valid_token_count": 818.33359375 + }, + { + "actual_token_mask_ratio": 0.2999347299337387, + "epoch": 0.48006982833866746, + "flow_matching_loss": 0.018248208891600372, + "grad_norm": 0.09098754078149796, + "kept_token_count": 572.88203125, + "learning_rate": 2.9997976822145995e-05, + "loss": 0.0268, + "mgd_cos_sim": 0.990961354970932, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009038650803267956, + "mgd_loss_weight": 0.9451815068721772, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 272.6251922607422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.4751831054688, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1650, + "valid_token_count": 818.32265625 + }, + { + "actual_token_mask_ratio": 0.29947408735752107, + "epoch": 0.4829793424498109, + "flow_matching_loss": 0.01611850894987583, + "grad_norm": 0.10068126022815704, + "kept_token_count": 573.24140625, + "learning_rate": 2.9997696148830714e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9918635189533234, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008136487752199172, + "mgd_loss_weight": 0.9448481619358062, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 288.96494140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 359.76527709960936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1660, + "valid_token_count": 818.30390625 + }, + { + "actual_token_mask_ratio": 0.30007095634937286, + "epoch": 0.4858888565609543, + "flow_matching_loss": 0.0161186414770782, + "grad_norm": 0.10918652266263962, + "kept_token_count": 572.68046875, + "learning_rate": 2.9997397251900423e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.9911914944648743, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00880851484835148, + "mgd_loss_weight": 0.9445148169994354, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 288.56190185546876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.0189178466797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1670, + "valid_token_count": 818.19453125 + }, + { + "actual_token_mask_ratio": 0.30105651915073395, + "epoch": 0.48879837067209775, + "flow_matching_loss": 0.015358551871031523, + "grad_norm": 0.10107959806919098, + "kept_token_count": 571.71015625, + "learning_rate": 2.999708013171831e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9911769807338715, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008823003992438317, + "mgd_loss_weight": 0.9441814720630646, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 278.8373687744141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.437158203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1680, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.2997738063335419, + "epoch": 0.4917078847832412, + "flow_matching_loss": 0.016525575052946807, + "grad_norm": 0.09849560260772705, + "kept_token_count": 572.98359375, + "learning_rate": 2.9996744788669705e-05, + "loss": 0.0251, + "mgd_cos_sim": 0.990903663635254, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009096337435767054, + "mgd_loss_weight": 0.9438481330871582, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 275.74794311523436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.0695037841797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1690, + "valid_token_count": 818.28359375 + }, + { + "actual_token_mask_ratio": 0.30030744075775145, + "epoch": 0.49461739889438466, + "flow_matching_loss": 0.01709047295153141, + "grad_norm": 0.10977118462324142, + "kept_token_count": 572.41953125, + "learning_rate": 2.999639122316208e-05, + "loss": 0.0255, + "mgd_cos_sim": 0.9911114394664764, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008888558112084865, + "mgd_loss_weight": 0.9435147821903229, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 283.1634246826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.8344970703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1700, + "valid_token_count": 818.09765625 + }, + { + "actual_token_mask_ratio": 0.29882158637046813, + "epoch": 0.4975269130055281, + "flow_matching_loss": 0.014269448723644019, + "grad_norm": 0.09297585487365723, + "kept_token_count": 573.57421875, + "learning_rate": 2.999601943562505e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.991375082731247, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008624917594715953, + "mgd_loss_weight": 0.943181437253952, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 293.0205596923828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.15318908691404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1710, + "valid_token_count": 818.01171875 + }, + { + "actual_token_mask_ratio": 0.299538779258728, + "epoch": 0.5004364271166715, + "flow_matching_loss": 0.01887317020446062, + "grad_norm": 0.09422967582941055, + "kept_token_count": 573.1640625, + "learning_rate": 2.9995629426510377e-05, + "loss": 0.0281, + "mgd_cos_sim": 0.9902364909648895, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009763507731258869, + "mgd_loss_weight": 0.9428480982780456, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 287.7588287353516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.8352935791016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1720, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.299034121632576, + "epoch": 0.503345941227815, + "flow_matching_loss": 0.01750663910061121, + "grad_norm": 0.09421559423208237, + "kept_token_count": 573.60234375, + "learning_rate": 2.9995221196291952e-05, + "loss": 0.0258, + "mgd_cos_sim": 0.9911899268627167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008810066524893046, + "mgd_loss_weight": 0.9425147473812103, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 284.5504852294922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.2807586669922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1730, + "valid_token_count": 818.30546875 + }, + { + "actual_token_mask_ratio": 0.29977802038192747, + "epoch": 0.5062554553389584, + "flow_matching_loss": 0.015795608051121235, + "grad_norm": 0.11465110629796982, + "kept_token_count": 573.015625, + "learning_rate": 2.999479474546582e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.991818630695343, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008181373123079538, + "mgd_loss_weight": 0.942181408405304, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 292.8600341796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.66364135742185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1740, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.30036631524562835, + "epoch": 0.5091649694501018, + "flow_matching_loss": 0.016278713662177324, + "grad_norm": 0.13286279141902924, + "kept_token_count": 572.48515625, + "learning_rate": 2.9994350074550153e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.9913281083106995, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008671908406540751, + "mgd_loss_weight": 0.9418480634689331, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 286.5017425537109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.52638549804686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1750, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.300401446223259, + "epoch": 0.5120744835612453, + "flow_matching_loss": 0.016652113292366266, + "grad_norm": 0.11417844891548157, + "kept_token_count": 572.525, + "learning_rate": 2.9993887184085273e-05, + "loss": 0.0249, + "mgd_cos_sim": 0.991223031282425, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00877696881070733, + "mgd_loss_weight": 0.9415147125720977, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 284.16773681640626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.22314453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1760, + "valid_token_count": 818.35703125 + }, + { + "actual_token_mask_ratio": 0.30027985870838164, + "epoch": 0.5149839976723887, + "flow_matching_loss": 0.016862994991242886, + "grad_norm": 0.14014676213264465, + "kept_token_count": 572.67890625, + "learning_rate": 2.9993406074633628e-05, + "loss": 0.0254, + "mgd_cos_sim": 0.9909481883049012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009051804430782795, + "mgd_loss_weight": 0.9411813795566559, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 283.5854034423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.7378326416016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1770, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.3003873318433762, + "epoch": 0.5178935117835322, + "flow_matching_loss": 0.015342715289443731, + "grad_norm": 0.11446550488471985, + "kept_token_count": 572.5046875, + "learning_rate": 2.9992906746779824e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.991107314825058, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008892679819837213, + "mgd_loss_weight": 0.9408480226993561, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 273.48632049560547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.39496154785155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1780, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.301190173625946, + "epoch": 0.5208030258946756, + "flow_matching_loss": 0.016482719220221043, + "grad_norm": 0.11831924319267273, + "kept_token_count": 571.84375, + "learning_rate": 2.9992389201130576e-05, + "loss": 0.0249, + "mgd_cos_sim": 0.991096144914627, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008903855970129371, + "mgd_loss_weight": 0.9405146837234497, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 282.42283630371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.2094360351563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1790, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.29919967949390414, + "epoch": 0.523712540005819, + "flow_matching_loss": 0.017476927302777768, + "grad_norm": 0.11746416985988617, + "kept_token_count": 573.5328125, + "learning_rate": 2.999185343831476e-05, + "loss": 0.0258, + "mgd_cos_sim": 0.9911684632301331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008831539191305637, + "mgd_loss_weight": 0.9401813447475433, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 291.2647338867188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.8009063720703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1800, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.29992467164993286, + "epoch": 0.5266220541169625, + "flow_matching_loss": 0.016690770722925664, + "grad_norm": 0.10300502926111221, + "kept_token_count": 572.659375, + "learning_rate": 2.999129945898338e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9913536429405212, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008646352868527173, + "mgd_loss_weight": 0.9398479878902435, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 278.9255798339844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.2970825195313, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1810, + "valid_token_count": 817.99453125 + }, + { + "actual_token_mask_ratio": 0.3006744682788849, + "epoch": 0.5295315682281059, + "flow_matching_loss": 0.015911793150007725, + "grad_norm": 0.10621501505374908, + "kept_token_count": 572.19296875, + "learning_rate": 2.999072726380957e-05, + "loss": 0.0242, + "mgd_cos_sim": 0.9912313461303711, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008768643345683813, + "mgd_loss_weight": 0.9395146548748017, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 294.68272399902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.4227691650391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1820, + "valid_token_count": 818.20390625 + }, + { + "actual_token_mask_ratio": 0.29960173964500425, + "epoch": 0.5324410823392494, + "flow_matching_loss": 0.017728911153972148, + "grad_norm": 0.09386592358350754, + "kept_token_count": 573.10390625, + "learning_rate": 2.9990136853488603e-05, + "loss": 0.0261, + "mgd_cos_sim": 0.9911215007305145, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008878506626933813, + "mgd_loss_weight": 0.9391813039779663, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 290.09539489746095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.7117858886719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1830, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.29990739822387696, + "epoch": 0.5353505964503927, + "flow_matching_loss": 0.01600538045167923, + "grad_norm": 0.09199009835720062, + "kept_token_count": 572.95078125, + "learning_rate": 2.9989528228737878e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9913849174976349, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0086150873452425, + "mgd_loss_weight": 0.9388479590415955, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 288.09434204101564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.2956909179687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1840, + "valid_token_count": 818.39453125 + }, + { + "actual_token_mask_ratio": 0.3003960371017456, + "epoch": 0.5382601105615362, + "flow_matching_loss": 0.016204075980931522, + "grad_norm": 0.07716123014688492, + "kept_token_count": 572.559375, + "learning_rate": 2.9988901390296935e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9908770084381103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009122998267412186, + "mgd_loss_weight": 0.9385146200656891, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 279.9786865234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.5038543701172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1850, + "valid_token_count": 818.40546875 + }, + { + "actual_token_mask_ratio": 0.3001470178365707, + "epoch": 0.5411696246726797, + "flow_matching_loss": 0.01694620531052351, + "grad_norm": 0.11365952342748642, + "kept_token_count": 572.83515625, + "learning_rate": 2.9988256338927442e-05, + "loss": 0.0257, + "mgd_cos_sim": 0.990661108493805, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009338885918259621, + "mgd_loss_weight": 0.9381812691688538, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 291.7527740478516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.0297119140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1860, + "valid_token_count": 818.50234375 + }, + { + "actual_token_mask_ratio": 0.3000558793544769, + "epoch": 0.5440791387838231, + "flow_matching_loss": 0.015779570769518615, + "grad_norm": 0.11622485518455505, + "kept_token_count": 572.68671875, + "learning_rate": 2.9987593075413198e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.9906123399734497, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00938765862956643, + "mgd_loss_weight": 0.9378479301929474, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 291.3665283203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.0595703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1870, + "valid_token_count": 818.18828125 + }, + { + "actual_token_mask_ratio": 0.30080385208129884, + "epoch": 0.5469886528949666, + "flow_matching_loss": 0.016658682469278575, + "grad_norm": 0.10315025597810745, + "kept_token_count": 571.96875, + "learning_rate": 2.9986911600560128e-05, + "loss": 0.0249, + "mgd_cos_sim": 0.9912574112415313, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008742597931995989, + "mgd_loss_weight": 0.9375145852565765, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 279.2365966796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.1782928466797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1880, + "valid_token_count": 818.03984375 + }, + { + "actual_token_mask_ratio": 0.2999159634113312, + "epoch": 0.5498981670061099, + "flow_matching_loss": 0.01579405488446355, + "grad_norm": 0.12018252164125443, + "kept_token_count": 572.77890625, + "learning_rate": 2.998621191519629e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9912989199161529, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00870108068920672, + "mgd_loss_weight": 0.9371812343597412, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 297.2743194580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.8840301513672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1890, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.2995219349861145, + "epoch": 0.5528076811172534, + "flow_matching_loss": 0.017354588862508534, + "grad_norm": 0.11253601312637329, + "kept_token_count": 573.16015625, + "learning_rate": 2.9985494020171873e-05, + "loss": 0.0257, + "mgd_cos_sim": 0.9910627007484436, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008937293663620949, + "mgd_loss_weight": 0.9368478953838348, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 294.47783813476565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.47169494628906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1900, + "valid_token_count": 818.24296875 + }, + { + "actual_token_mask_ratio": 0.3000096768140793, + "epoch": 0.5557171952283969, + "flow_matching_loss": 0.015598054602742195, + "grad_norm": 0.10533475875854492, + "kept_token_count": 572.73828125, + "learning_rate": 2.9984757916359176e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9910089254379273, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008991073071956634, + "mgd_loss_weight": 0.936514550447464, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 288.11277160644534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.9655059814453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1910, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.29922102093696595, + "epoch": 0.5586267093395403, + "flow_matching_loss": 0.015247715730220079, + "grad_norm": 0.09062882512807846, + "kept_token_count": 573.38125, + "learning_rate": 2.998400360465264e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.991613119840622, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008386886306107044, + "mgd_loss_weight": 0.9361812055110932, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 277.7789367675781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.56968994140624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1920, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.3006599873304367, + "epoch": 0.5615362234506838, + "flow_matching_loss": 0.015981862042099237, + "grad_norm": 0.10029038786888123, + "kept_token_count": 572.18125, + "learning_rate": 2.998323108596883e-05, + "loss": 0.0245, + "mgd_cos_sim": 0.9909161448478698, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009083852730691433, + "mgd_loss_weight": 0.9358478605747222, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 283.7682739257813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.28094787597655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1930, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.29983504116535187, + "epoch": 0.5644457375618271, + "flow_matching_loss": 0.01657632291316986, + "grad_norm": 0.11773277819156647, + "kept_token_count": 572.925, + "learning_rate": 2.9982440361246422e-05, + "loss": 0.0249, + "mgd_cos_sim": 0.9911508858203888, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008849116880446673, + "mgd_loss_weight": 0.9355145156383514, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 291.8760131835937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.34739685058594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1940, + "valid_token_count": 818.27421875 + }, + { + "actual_token_mask_ratio": 0.2998333305120468, + "epoch": 0.5673552516729706, + "flow_matching_loss": 0.015001047123223542, + "grad_norm": 0.10616596043109894, + "kept_token_count": 573.05390625, + "learning_rate": 2.998163143144622e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9909034669399261, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009096546657383442, + "mgd_loss_weight": 0.9351811707019806, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 288.200439453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.90293273925784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1950, + "valid_token_count": 818.44765625 + }, + { + "actual_token_mask_ratio": 0.29906857907772066, + "epoch": 0.570264765784114, + "flow_matching_loss": 0.0192411832511425, + "grad_norm": 0.12244895100593567, + "kept_token_count": 573.4328125, + "learning_rate": 2.9980804297551155e-05, + "loss": 0.0276, + "mgd_cos_sim": 0.9910623967647553, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008937609475106, + "mgd_loss_weight": 0.9348478317260742, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 284.5409790039063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.64649963378906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1960, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.2999012768268585, + "epoch": 0.5731742798952575, + "flow_matching_loss": 0.0152212037704885, + "grad_norm": 0.1194990947842598, + "kept_token_count": 572.7765625, + "learning_rate": 2.997995896056627e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9911855697631836, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008814410911872983, + "mgd_loss_weight": 0.9345144808292389, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 287.852880859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.07716674804686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1970, + "valid_token_count": 818.13984375 + }, + { + "actual_token_mask_ratio": 0.3006465911865234, + "epoch": 0.576083794006401, + "flow_matching_loss": 0.01742344256490469, + "grad_norm": 0.10622438788414001, + "kept_token_count": 572.19609375, + "learning_rate": 2.997909542151873e-05, + "loss": 0.0256, + "mgd_cos_sim": 0.9912162721157074, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008783734822645783, + "mgd_loss_weight": 0.934181135892868, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.9073211669922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.9661926269531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1980, + "valid_token_count": 818.17734375 + }, + { + "actual_token_mask_ratio": 0.299935844540596, + "epoch": 0.5789933081175443, + "flow_matching_loss": 0.019160817191004754, + "grad_norm": 0.11905006319284439, + "kept_token_count": 572.83984375, + "learning_rate": 2.9978213681457823e-05, + "loss": 0.0274, + "mgd_cos_sim": 0.9911649525165558, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008835038635879755, + "mgd_loss_weight": 0.9338478028774262, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 289.865380859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.03855590820314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 1990, + "valid_token_count": 818.26484375 + }, + { + "actual_token_mask_ratio": 0.3000954508781433, + "epoch": 0.5819028222286878, + "flow_matching_loss": 0.017880126647651196, + "grad_norm": 0.1092580184340477, + "kept_token_count": 572.64609375, + "learning_rate": 2.9977313741454933e-05, + "loss": 0.0258, + "mgd_cos_sim": 0.9915230333805084, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008476965269073844, + "mgd_loss_weight": 0.9335144460201263, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 297.0670654296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.7104431152344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2000, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.3001412034034729, + "epoch": 0.5848123363398312, + "flow_matching_loss": 0.016569117177277805, + "grad_norm": 0.08351293951272964, + "kept_token_count": 572.65703125, + "learning_rate": 2.997639560260358e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9912154078483582, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008784587122499942, + "mgd_loss_weight": 0.93318110704422, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 288.79482421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.1472930908203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2010, + "valid_token_count": 818.24296875 + }, + { + "actual_token_mask_ratio": 0.30010000467300413, + "epoch": 0.5877218504509747, + "flow_matching_loss": 0.015828347485512494, + "grad_norm": 0.10866475850343704, + "kept_token_count": 572.7421875, + "learning_rate": 2.9975459266019387e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9913874268531799, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008612563787028193, + "mgd_loss_weight": 0.9328477680683136, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 283.7320526123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.6103546142578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2020, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.3003083199262619, + "epoch": 0.5906313645621182, + "flow_matching_loss": 0.015781966969370842, + "grad_norm": 0.1045861467719078, + "kept_token_count": 572.48046875, + "learning_rate": 2.9974504732840092e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9911203742027282, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008879636088386178, + "mgd_loss_weight": 0.9325144112110137, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 310.23157653808596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.86224670410155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2030, + "valid_token_count": 818.18828125 + }, + { + "actual_token_mask_ratio": 0.29941198229789734, + "epoch": 0.5935408786732616, + "flow_matching_loss": 0.01606724616140127, + "grad_norm": 0.0820777490735054, + "kept_token_count": 573.2546875, + "learning_rate": 2.997353200422555e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.9911016821861267, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008898299094289542, + "mgd_loss_weight": 0.9321810781955719, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 290.77195739746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.8222381591797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2040, + "valid_token_count": 818.24609375 + }, + { + "actual_token_mask_ratio": 0.3005182445049286, + "epoch": 0.596450392784405, + "flow_matching_loss": 0.016383586172014474, + "grad_norm": 0.13024497032165527, + "kept_token_count": 572.253125, + "learning_rate": 2.99725410813577e-05, + "loss": 0.0249, + "mgd_cos_sim": 0.99086292386055, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009137081634253263, + "mgd_loss_weight": 0.9318477272987366, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 300.8634460449219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.0911560058594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2050, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.2998429208993912, + "epoch": 0.5993599068955484, + "flow_matching_loss": 0.015838522370904683, + "grad_norm": 0.09058203548192978, + "kept_token_count": 573.00859375, + "learning_rate": 2.997153196544063e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9913025617599487, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00869743200019002, + "mgd_loss_weight": 0.9315143823623657, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.61455078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.61155090332034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2060, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.3005795985460281, + "epoch": 0.6022694210066919, + "flow_matching_loss": 0.01647542854771018, + "grad_norm": 0.09054163098335266, + "kept_token_count": 572.3875, + "learning_rate": 2.997050465770049e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9916437447071076, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008356264419853687, + "mgd_loss_weight": 0.9311810433864594, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 287.07981262207034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.32313537597656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2070, + "valid_token_count": 818.36953125 + }, + { + "actual_token_mask_ratio": 0.299971279501915, + "epoch": 0.6051789351178353, + "flow_matching_loss": 0.01692401636391878, + "grad_norm": 0.09926074743270874, + "kept_token_count": 572.76171875, + "learning_rate": 2.9969459159385562e-05, + "loss": 0.0254, + "mgd_cos_sim": 0.9908433735370636, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009156634751707315, + "mgd_loss_weight": 0.9308476924896241, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 292.33579711914064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.71222534179685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2080, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.300013467669487, + "epoch": 0.6080884492289788, + "flow_matching_loss": 0.016427267994731664, + "grad_norm": 0.12157589197158813, + "kept_token_count": 572.68984375, + "learning_rate": 2.9968395471766237e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9910329043865204, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008967106370255351, + "mgd_loss_weight": 0.9305143535137177, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 296.1399688720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 360.0567962646484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2090, + "valid_token_count": 818.13984375 + }, + { + "actual_token_mask_ratio": 0.30032976269721984, + "epoch": 0.6109979633401222, + "flow_matching_loss": 0.015328653622418642, + "grad_norm": 0.09601689130067825, + "kept_token_count": 572.5765625, + "learning_rate": 2.996731359613498e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9908371329307556, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009162884019315243, + "mgd_loss_weight": 0.9301810085773468, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 298.23948974609374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.80199279785154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2100, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.30011182725429536, + "epoch": 0.6139074774512656, + "flow_matching_loss": 0.01563164805993438, + "grad_norm": 0.12340479344129562, + "kept_token_count": 572.72578125, + "learning_rate": 2.996621353380638e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9908992767333984, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009100727550685406, + "mgd_loss_weight": 0.9298476576805115, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.8981475830078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.5440704345703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2110, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.3006245791912079, + "epoch": 0.6168169915624091, + "flow_matching_loss": 0.01701854532584548, + "grad_norm": 0.12835879623889923, + "kept_token_count": 572.43984375, + "learning_rate": 2.9965095286117113e-05, + "loss": 0.0253, + "mgd_cos_sim": 0.9910795867443085, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008920410880818963, + "mgd_loss_weight": 0.9295143187046051, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 296.3236083984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.8580657958984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2120, + "valid_token_count": 818.50234375 + }, + { + "actual_token_mask_ratio": 0.3000122159719467, + "epoch": 0.6197265056735525, + "flow_matching_loss": 0.015226689074188471, + "grad_norm": 0.08542129397392273, + "kept_token_count": 572.7828125, + "learning_rate": 2.996395885442596e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9916114032268524, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00838859798386693, + "mgd_loss_weight": 0.9291809737682343, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 300.665576171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.9301422119141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2130, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.30018088817596433, + "epoch": 0.622636019784696, + "flow_matching_loss": 0.015133844129741192, + "grad_norm": 0.09441673010587692, + "kept_token_count": 572.571875, + "learning_rate": 2.996280424011379e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9908931970596313, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00910679721273482, + "mgd_loss_weight": 0.9288476288318634, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 297.18287658691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.0545684814453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2140, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.2997792512178421, + "epoch": 0.6255455338958394, + "flow_matching_loss": 0.016927802842110394, + "grad_norm": 0.10683377087116241, + "kept_token_count": 573.0296875, + "learning_rate": 2.9961631444583565e-05, + "loss": 0.0251, + "mgd_cos_sim": 0.9911559462547302, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00884407078847289, + "mgd_loss_weight": 0.9285142838954925, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.51962890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.24741516113284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2150, + "valid_token_count": 818.35703125 + }, + { + "actual_token_mask_ratio": 0.30046020448207855, + "epoch": 0.6284550480069828, + "flow_matching_loss": 0.017002758383750916, + "grad_norm": 0.11009805649518967, + "kept_token_count": 572.3984375, + "learning_rate": 2.9960440469260347e-05, + "loss": 0.0253, + "mgd_cos_sim": 0.9910876274108886, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008912381250411272, + "mgd_loss_weight": 0.9281809389591217, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 293.16245422363284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.8276306152344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2160, + "valid_token_count": 818.25390625 + }, + { + "actual_token_mask_ratio": 0.2998696446418762, + "epoch": 0.6313645621181263, + "flow_matching_loss": 0.016755920182913542, + "grad_norm": 0.0921381413936615, + "kept_token_count": 572.72890625, + "learning_rate": 2.995923131559128e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9910994052886963, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008900595922023058, + "mgd_loss_weight": 0.9278475940227509, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 291.1195953369141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.8530731201172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2170, + "valid_token_count": 818.028125 + }, + { + "actual_token_mask_ratio": 0.2992514669895172, + "epoch": 0.6342740762292697, + "flow_matching_loss": 0.016131039895117282, + "grad_norm": 0.09835054725408554, + "kept_token_count": 573.34375, + "learning_rate": 2.99580039850456e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9906442344188691, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009355759620666504, + "mgd_loss_weight": 0.9275142550468445, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 296.3332794189453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.44300231933596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2180, + "valid_token_count": 818.18671875 + }, + { + "actual_token_mask_ratio": 0.2996241509914398, + "epoch": 0.6371835903404132, + "flow_matching_loss": 0.017236554343253375, + "grad_norm": 0.0987553745508194, + "kept_token_count": 573.07890625, + "learning_rate": 2.9956758479114635e-05, + "loss": 0.0254, + "mgd_cos_sim": 0.9912479817867279, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008752020308747888, + "mgd_loss_weight": 0.9271809041500092, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 290.23854064941406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.2327026367187, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2190, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.29992842078208926, + "epoch": 0.6400931044515565, + "flow_matching_loss": 0.017085126508027314, + "grad_norm": 0.0942176952958107, + "kept_token_count": 573.06328125, + "learning_rate": 2.9955494799311787e-05, + "loss": 0.0253, + "mgd_cos_sim": 0.9911875069141388, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008812489453703165, + "mgd_loss_weight": 0.9268475592136383, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 290.96949768066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.69617614746096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2200, + "valid_token_count": 818.5796875 + }, + { + "actual_token_mask_ratio": 0.2998939096927643, + "epoch": 0.6430026185627, + "flow_matching_loss": 0.01664211554452777, + "grad_norm": 0.11592062562704086, + "kept_token_count": 573.05703125, + "learning_rate": 2.995421294717254e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9911549925804138, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008845007559284569, + "mgd_loss_weight": 0.9265142202377319, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 310.5494842529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.419482421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2210, + "valid_token_count": 818.528125 + }, + { + "actual_token_mask_ratio": 0.2999308228492737, + "epoch": 0.6459121326738435, + "flow_matching_loss": 0.016412284597754477, + "grad_norm": 0.10899300128221512, + "kept_token_count": 572.98984375, + "learning_rate": 2.995291292425448e-05, + "loss": 0.0249, + "mgd_cos_sim": 0.9908016502857209, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009198353858664632, + "mgd_loss_weight": 0.9261808693408966, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 314.2169616699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.7066192626953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2220, + "valid_token_count": 818.4734375 + }, + { + "actual_token_mask_ratio": 0.2996350020170212, + "epoch": 0.6488216467849869, + "flow_matching_loss": 0.0145262292586267, + "grad_norm": 0.09522155672311783, + "kept_token_count": 572.859375, + "learning_rate": 2.9951594732137248e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9914160907268524, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008583897072821856, + "mgd_loss_weight": 0.9258475303649902, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.65216674804685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.36960754394534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2230, + "valid_token_count": 817.9390625 + }, + { + "actual_token_mask_ratio": 0.30042846500873566, + "epoch": 0.6517311608961304, + "flow_matching_loss": 0.01591274244710803, + "grad_norm": 0.10916910320520401, + "kept_token_count": 572.45546875, + "learning_rate": 2.995025837242257e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9912794351577758, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008720548218116164, + "mgd_loss_weight": 0.9255141854286194, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.68056030273436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.344921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2240, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.2998037219047546, + "epoch": 0.6546406750072737, + "flow_matching_loss": 0.016597098112106322, + "grad_norm": 0.12537166476249695, + "kept_token_count": 572.89765625, + "learning_rate": 2.994890384673426e-05, + "loss": 0.0249, + "mgd_cos_sim": 0.9909984648227692, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009001534665003419, + "mgd_loss_weight": 0.925180834531784, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 300.76605529785155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.3668975830078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2250, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.3000678837299347, + "epoch": 0.6575501891184172, + "flow_matching_loss": 0.014811728801578283, + "grad_norm": 0.09993059188127518, + "kept_token_count": 572.65234375, + "learning_rate": 2.9947531156718174e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9912904441356659, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008709568530321121, + "mgd_loss_weight": 0.9248475015163422, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 299.82163696289064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.9988677978516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2260, + "valid_token_count": 818.14765625 + }, + { + "actual_token_mask_ratio": 0.29951644241809844, + "epoch": 0.6604597032295607, + "flow_matching_loss": 0.015934500750154255, + "grad_norm": 0.13045841455459595, + "kept_token_count": 573.0578125, + "learning_rate": 2.994614030404228e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.990892106294632, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009107890957966446, + "mgd_loss_weight": 0.9245141446590424, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 324.3432281494141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.51161499023436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2270, + "valid_token_count": 818.08984375 + }, + { + "actual_token_mask_ratio": 0.29959786832332613, + "epoch": 0.6633692173407041, + "flow_matching_loss": 0.016562553960829972, + "grad_norm": 0.07949265837669373, + "kept_token_count": 573.10625, + "learning_rate": 2.994473129039659e-05, + "loss": 0.0245, + "mgd_cos_sim": 0.9914487421512603, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008551248954609037, + "mgd_loss_weight": 0.924180805683136, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.35863342285154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.39842529296874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2280, + "valid_token_count": 818.25703125 + }, + { + "actual_token_mask_ratio": 0.30156668126583097, + "epoch": 0.6662787314518476, + "flow_matching_loss": 0.01572197787463665, + "grad_norm": 0.08569908887147903, + "kept_token_count": 571.41328125, + "learning_rate": 2.9943304117493183e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9915093183517456, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008490680158138275, + "mgd_loss_weight": 0.9238474667072296, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 296.6512725830078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.24977722167966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2290, + "valid_token_count": 818.13828125 + }, + { + "actual_token_mask_ratio": 0.300111722946167, + "epoch": 0.669188245562991, + "flow_matching_loss": 0.017189539689570665, + "grad_norm": 0.08286447077989578, + "kept_token_count": 572.59765625, + "learning_rate": 2.9941858787066207e-05, + "loss": 0.0253, + "mgd_cos_sim": 0.9912308812141418, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008769114920869469, + "mgd_loss_weight": 0.9235141098499298, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 305.3335418701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.3500122070312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2300, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.3002012550830841, + "epoch": 0.6720977596741344, + "flow_matching_loss": 0.017378046363592147, + "grad_norm": 0.10216525942087173, + "kept_token_count": 572.5203125, + "learning_rate": 2.9940395300871888e-05, + "loss": 0.0256, + "mgd_cos_sim": 0.991050410270691, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008949590567499398, + "mgd_loss_weight": 0.923180776834488, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 305.9210235595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.3486297607422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2310, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.29975509345531465, + "epoch": 0.6750072737852778, + "flow_matching_loss": 0.015957340877503157, + "grad_norm": 0.11306405812501907, + "kept_token_count": 573.01640625, + "learning_rate": 2.993891366068849e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9913913786411286, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008608620055019856, + "mgd_loss_weight": 0.9228474259376526, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 300.4538177490234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.9503570556641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2320, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.2999982535839081, + "epoch": 0.6779167878964213, + "flow_matching_loss": 0.015820462070405484, + "grad_norm": 0.08069677650928497, + "kept_token_count": 572.82734375, + "learning_rate": 2.993741386831634e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.991799509525299, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008200492337346077, + "mgd_loss_weight": 0.9225140810012817, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.63085021972654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.0407196044922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2330, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.29968046844005586, + "epoch": 0.6808263020075648, + "flow_matching_loss": 0.015943012200295925, + "grad_norm": 0.12042155861854553, + "kept_token_count": 573.1265625, + "learning_rate": 2.9935895925577837e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9917004704475403, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008299526805058122, + "mgd_loss_weight": 0.9221807420253754, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.36109924316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.26917419433596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2340, + "valid_token_count": 818.37265625 + }, + { + "actual_token_mask_ratio": 0.300750607252121, + "epoch": 0.6837358161187082, + "flow_matching_loss": 0.01828477019444108, + "grad_norm": 0.09403672069311142, + "kept_token_count": 572.13828125, + "learning_rate": 2.9934359834317425e-05, + "loss": 0.0265, + "mgd_cos_sim": 0.9910647094249725, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00893527725711465, + "mgd_loss_weight": 0.9218473911285401, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 304.8444793701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.6765960693359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2350, + "valid_token_count": 818.22265625 + }, + { + "actual_token_mask_ratio": 0.2998662769794464, + "epoch": 0.6866453302298516, + "flow_matching_loss": 0.016871007066220044, + "grad_norm": 0.11662973463535309, + "kept_token_count": 572.9890625, + "learning_rate": 2.993280559640159e-05, + "loss": 0.0256, + "mgd_cos_sim": 0.9905671715736389, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009432833082973957, + "mgd_loss_weight": 0.9215140521526337, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 297.7117218017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.766796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2360, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.30051503479480746, + "epoch": 0.689554844340995, + "flow_matching_loss": 0.015439129434525967, + "grad_norm": 0.11548007279634476, + "kept_token_count": 572.40625, + "learning_rate": 2.9931233213718887e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9916354477405548, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008364534564316272, + "mgd_loss_weight": 0.9211807072162628, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.70118408203126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.41884765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2370, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.3003795802593231, + "epoch": 0.6924643584521385, + "flow_matching_loss": 0.017894089687615633, + "grad_norm": 0.11717240512371063, + "kept_token_count": 572.4859375, + "learning_rate": 2.992964268817991e-05, + "loss": 0.026, + "mgd_cos_sim": 0.9911788046360016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008821203559637069, + "mgd_loss_weight": 0.9208473563194275, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 309.6571990966797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.87431640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2380, + "valid_token_count": 818.27890625 + }, + { + "actual_token_mask_ratio": 0.3010792165994644, + "epoch": 0.695373872563282, + "flow_matching_loss": 0.017910534422844647, + "grad_norm": 0.11578381806612015, + "kept_token_count": 571.8578125, + "learning_rate": 2.9928034021717286e-05, + "loss": 0.0265, + "mgd_cos_sim": 0.9906891703605651, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009310828102752567, + "mgd_loss_weight": 0.9205140173435211, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 317.0167205810547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.42449951171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2390, + "valid_token_count": 818.19765625 + }, + { + "actual_token_mask_ratio": 0.2995228976011276, + "epoch": 0.6982833866744254, + "flow_matching_loss": 0.01574670374393463, + "grad_norm": 0.13125328719615936, + "kept_token_count": 573.3109375, + "learning_rate": 2.9926407216285707e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9911987662315369, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008801241964101791, + "mgd_loss_weight": 0.9201806724071503, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 295.3940002441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.4708618164062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2400, + "valid_token_count": 818.45859375 + }, + { + "actual_token_mask_ratio": 0.30010253489017485, + "epoch": 0.7011929007855688, + "flow_matching_loss": 0.018492572382092475, + "grad_norm": 0.11701690405607224, + "kept_token_count": 572.80234375, + "learning_rate": 2.9924762273861894e-05, + "loss": 0.0264, + "mgd_cos_sim": 0.991387528181076, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00861247037537396, + "mgd_loss_weight": 0.9198473274707795, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 300.9837951660156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.7424377441406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2410, + "valid_token_count": 818.41484375 + }, + { + "actual_token_mask_ratio": 0.30000682175159454, + "epoch": 0.7041024148967122, + "flow_matching_loss": 0.017326224967837332, + "grad_norm": 0.11171963065862656, + "kept_token_count": 572.8125, + "learning_rate": 2.992309919644461e-05, + "loss": 0.0255, + "mgd_cos_sim": 0.991159176826477, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008840811904519796, + "mgd_loss_weight": 0.9195139825344085, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 311.230908203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.82056884765626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2420, + "valid_token_count": 818.30859375 + }, + { + "actual_token_mask_ratio": 0.30053953528404237, + "epoch": 0.7070119290078557, + "flow_matching_loss": 0.016004685126245023, + "grad_norm": 0.09612458944320679, + "kept_token_count": 572.47890625, + "learning_rate": 2.9921417986054633e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9914004266262054, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008599555306136607, + "mgd_loss_weight": 0.9191806375980377, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 310.5407257080078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.18658752441405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2430, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.29998840391635895, + "epoch": 0.7099214431189991, + "flow_matching_loss": 0.01694033406674862, + "grad_norm": 0.09535031020641327, + "kept_token_count": 572.7625, + "learning_rate": 2.9919718644734815e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9914538741111756, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0085461197886616, + "mgd_loss_weight": 0.9188472926616669, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 297.9635925292969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.0824340820312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2440, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.30003073811531067, + "epoch": 0.7128309572301426, + "flow_matching_loss": 0.015842247847467662, + "grad_norm": 0.08537528663873672, + "kept_token_count": 572.76484375, + "learning_rate": 2.991800117455e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9909985184669494, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00900147333741188, + "mgd_loss_weight": 0.9185139536857605, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 311.55048217773435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.4152313232422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2450, + "valid_token_count": 818.27109375 + }, + { + "actual_token_mask_ratio": 0.3007550537586212, + "epoch": 0.715740471341286, + "flow_matching_loss": 0.017675547301769255, + "grad_norm": 0.08749031275510788, + "kept_token_count": 572.02265625, + "learning_rate": 2.9916265577587085e-05, + "loss": 0.0259, + "mgd_cos_sim": 0.9910852074623108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008914794633165002, + "mgd_loss_weight": 0.9181806087493897, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 321.1687896728516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.52460327148435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2460, + "valid_token_count": 818.06328125 + }, + { + "actual_token_mask_ratio": 0.29939988255500793, + "epoch": 0.7186499854524294, + "flow_matching_loss": 0.016484821680933237, + "grad_norm": 0.10574624687433243, + "kept_token_count": 573.31796875, + "learning_rate": 2.9914511855954986e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9914932012557983, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008506791945546866, + "mgd_loss_weight": 0.9178472578525543, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.4517303466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.07196655273435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2470, + "valid_token_count": 818.32578125 + }, + { + "actual_token_mask_ratio": 0.2997666418552399, + "epoch": 0.7215594995635729, + "flow_matching_loss": 0.0176145407371223, + "grad_norm": 0.106066033244133, + "kept_token_count": 573.0453125, + "learning_rate": 2.9912740011784626e-05, + "loss": 0.0257, + "mgd_cos_sim": 0.9911709666252136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008829039335250855, + "mgd_loss_weight": 0.9175139248371125, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.6409484863281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.7248016357422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2480, + "valid_token_count": 818.36171875 + }, + { + "actual_token_mask_ratio": 0.299508735537529, + "epoch": 0.7244690136747163, + "flow_matching_loss": 0.016945476830005645, + "grad_norm": 0.10571756213903427, + "kept_token_count": 573.1234375, + "learning_rate": 2.991095004722898e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.9917030334472656, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008296975865960121, + "mgd_loss_weight": 0.9171805679798126, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 303.4135040283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.47454223632815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2490, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.29961844682693484, + "epoch": 0.7273785277858598, + "flow_matching_loss": 0.017051800433546305, + "grad_norm": 0.11411608010530472, + "kept_token_count": 573.196875, + "learning_rate": 2.9909141964463017e-05, + "loss": 0.0249, + "mgd_cos_sim": 0.9914531409740448, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00854685897938907, + "mgd_loss_weight": 0.9168472290039062, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 302.89088134765626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.2644287109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2500, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.3003113687038422, + "epoch": 0.7302880418970032, + "flow_matching_loss": 0.01718676732853055, + "grad_norm": 0.11004273593425751, + "kept_token_count": 572.621875, + "learning_rate": 2.990731576568372e-05, + "loss": 0.0249, + "mgd_cos_sim": 0.9916127920150757, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008387211104854941, + "mgd_loss_weight": 0.9165138900279999, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 303.70302734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.7111541748047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2510, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.2997000217437744, + "epoch": 0.7331975560081466, + "flow_matching_loss": 0.016822850238531827, + "grad_norm": 0.11106458306312561, + "kept_token_count": 573.15625, + "learning_rate": 2.99054714531101e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.9913502216339112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008649768214672804, + "mgd_loss_weight": 0.9161805331707, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 315.00055541992185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.7756378173828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2520, + "valid_token_count": 818.44375 + }, + { + "actual_token_mask_ratio": 0.300206258893013, + "epoch": 0.7361070701192901, + "flow_matching_loss": 0.015433620288968086, + "grad_norm": 0.09599324315786362, + "kept_token_count": 572.76640625, + "learning_rate": 2.9903609028983178e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9916365265846252, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008363483380526305, + "mgd_loss_weight": 0.9158472001552582, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 302.7052368164062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.9238739013672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2530, + "valid_token_count": 818.47265625 + }, + { + "actual_token_mask_ratio": 0.30040937960147857, + "epoch": 0.7390165842304335, + "flow_matching_loss": 0.01558627439662814, + "grad_norm": 0.10777294635772705, + "kept_token_count": 572.48828125, + "learning_rate": 2.990172849556596e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9912165939807892, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008783401921391487, + "mgd_loss_weight": 0.9155138492584228, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.8734191894531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.77581481933595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2540, + "valid_token_count": 818.32265625 + }, + { + "actual_token_mask_ratio": 0.29963130354881284, + "epoch": 0.741926098341577, + "flow_matching_loss": 0.016808789502829314, + "grad_norm": 0.10372602939605713, + "kept_token_count": 572.99296875, + "learning_rate": 2.989982985514348e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9912547171115875, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008745278092101216, + "mgd_loss_weight": 0.915180504322052, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 298.8344757080078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.0236450195313, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2550, + "valid_token_count": 818.125 + }, + { + "actual_token_mask_ratio": 0.29972265660762787, + "epoch": 0.7448356124527203, + "flow_matching_loss": 0.015599124412983657, + "grad_norm": 0.10700434446334839, + "kept_token_count": 573.02265625, + "learning_rate": 2.9897913110022766e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9909747004508972, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009025305230170488, + "mgd_loss_weight": 0.9148471653461456, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 314.9010711669922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.003515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2560, + "valid_token_count": 818.27734375 + }, + { + "actual_token_mask_ratio": 0.3001212030649185, + "epoch": 0.7477451265638638, + "flow_matching_loss": 0.016706061363220216, + "grad_norm": 0.1337888091802597, + "kept_token_count": 572.68203125, + "learning_rate": 2.989597826253284e-05, + "loss": 0.0245, + "mgd_cos_sim": 0.9914986371994019, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008501369692385196, + "mgd_loss_weight": 0.9145138144493103, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 313.37952575683596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.0198944091797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2570, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.29945798218250275, + "epoch": 0.7506546406750073, + "flow_matching_loss": 0.017307483498007058, + "grad_norm": 0.10114797949790955, + "kept_token_count": 573.3328125, + "learning_rate": 2.989402531502473e-05, + "loss": 0.0258, + "mgd_cos_sim": 0.990687620639801, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009312374610453844, + "mgd_loss_weight": 0.914180475473404, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 310.7976440429687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.52003173828126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2580, + "valid_token_count": 818.41015625 + }, + { + "actual_token_mask_ratio": 0.29933314621448515, + "epoch": 0.7535641547861507, + "flow_matching_loss": 0.014930473640561104, + "grad_norm": 0.09511971473693848, + "kept_token_count": 573.27734375, + "learning_rate": 2.989205426987144e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9913340508937836, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008665952924638986, + "mgd_loss_weight": 0.913847130537033, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 315.47846069335935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.93421630859376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2590, + "valid_token_count": 818.19140625 + }, + { + "actual_token_mask_ratio": 0.30044027268886564, + "epoch": 0.7564736688972942, + "flow_matching_loss": 0.0159614116884768, + "grad_norm": 0.08938737958669662, + "kept_token_count": 572.43359375, + "learning_rate": 2.9890065129467988e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9911952078342438, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008804794633761048, + "mgd_loss_weight": 0.9135137796401978, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 321.319873046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.74623718261716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2600, + "valid_token_count": 818.27734375 + }, + { + "actual_token_mask_ratio": 0.3000040590763092, + "epoch": 0.7593831830084375, + "flow_matching_loss": 0.015530952904373407, + "grad_norm": 0.0920567661523819, + "kept_token_count": 572.83984375, + "learning_rate": 2.9888057896231364e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9919045507907868, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008095441805198788, + "mgd_loss_weight": 0.9131804406642914, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 317.54973754882815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.6953094482422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2610, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.3005356967449188, + "epoch": 0.762292697119581, + "flow_matching_loss": 0.01665625097230077, + "grad_norm": 0.10199524462223053, + "kept_token_count": 572.27578125, + "learning_rate": 2.988603257260054e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9911093890666962, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008890610001981259, + "mgd_loss_weight": 0.9128470957279206, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 318.8062316894531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.89837646484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2620, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.3001893639564514, + "epoch": 0.7652022112307245, + "flow_matching_loss": 0.017970907408744095, + "grad_norm": 0.13365234434604645, + "kept_token_count": 572.7671875, + "learning_rate": 2.988398916103648e-05, + "loss": 0.0262, + "mgd_cos_sim": 0.99096177816391, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009038212802261114, + "mgd_loss_weight": 0.9125137507915497, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 309.24973449707034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.3125915527344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2630, + "valid_token_count": 818.45859375 + }, + { + "actual_token_mask_ratio": 0.3009838402271271, + "epoch": 0.7681117253418679, + "flow_matching_loss": 0.016250984277576208, + "grad_norm": 0.12328746914863586, + "kept_token_count": 572.059375, + "learning_rate": 2.988192766402212e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9917751371860504, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008224862860515714, + "mgd_loss_weight": 0.9121804058551788, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 326.0777587890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.33435974121096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2640, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.3004907935857773, + "epoch": 0.7710212394530114, + "flow_matching_loss": 0.01563477609306574, + "grad_norm": 0.09321015328168869, + "kept_token_count": 572.29140625, + "learning_rate": 2.9879848084062378e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9915656924247742, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008434317493811249, + "mgd_loss_weight": 0.911847060918808, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.41489868164064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.68358459472654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2650, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.29963178634643556, + "epoch": 0.7739307535641547, + "flow_matching_loss": 0.015990360639989376, + "grad_norm": 0.10352909564971924, + "kept_token_count": 573.2375, + "learning_rate": 2.9877750423684132e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9912363231182099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008763661468401551, + "mgd_loss_weight": 0.9115137159824371, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 314.3992523193359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.41249694824216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2660, + "valid_token_count": 818.48515625 + }, + { + "actual_token_mask_ratio": 0.3007822513580322, + "epoch": 0.7768402676752982, + "flow_matching_loss": 0.015154664684087039, + "grad_norm": 0.10242529958486557, + "kept_token_count": 572.24140625, + "learning_rate": 2.987563468543624e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9914959192276, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008504091016948222, + "mgd_loss_weight": 0.9111803770065308, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 321.49843139648436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.08257446289065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2670, + "valid_token_count": 818.40390625 + }, + { + "actual_token_mask_ratio": 0.29965923726558685, + "epoch": 0.7797497817864416, + "flow_matching_loss": 0.015107930172234774, + "grad_norm": 0.10165613144636154, + "kept_token_count": 573.04609375, + "learning_rate": 2.9873500871889533e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9912699282169342, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008730065682902933, + "mgd_loss_weight": 0.9108470261096955, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 305.1924133300781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.92908935546876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2680, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.30014115273952485, + "epoch": 0.7826592958975851, + "flow_matching_loss": 0.015135679487138986, + "grad_norm": 0.09604977071285248, + "kept_token_count": 572.5296875, + "learning_rate": 2.9871348985636778e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9915471553802491, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008452840941026806, + "mgd_loss_weight": 0.9105136811733245, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 314.9140899658203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.903515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2690, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.30101052224636077, + "epoch": 0.7855688100087286, + "flow_matching_loss": 0.017175150476396082, + "grad_norm": 0.0824110209941864, + "kept_token_count": 572.071875, + "learning_rate": 2.986917902929273e-05, + "loss": 0.0251, + "mgd_cos_sim": 0.9912646234035491, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008735372545197606, + "mgd_loss_weight": 0.9101803421974182, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 309.1729248046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.59380493164065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2700, + "valid_token_count": 818.42421875 + }, + { + "actual_token_mask_ratio": 0.3000775039196014, + "epoch": 0.788478324119872, + "flow_matching_loss": 0.01725882915779948, + "grad_norm": 0.09898744523525238, + "kept_token_count": 572.69375, + "learning_rate": 2.9866991005494095e-05, + "loss": 0.0254, + "mgd_cos_sim": 0.9910325944423676, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0089674212038517, + "mgd_loss_weight": 0.9098469913005829, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.3402069091797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.9866912841797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2710, + "valid_token_count": 818.22109375 + }, + { + "actual_token_mask_ratio": 0.30063657760620116, + "epoch": 0.7913878382310154, + "flow_matching_loss": 0.013932534400373698, + "grad_norm": 0.11250139772891998, + "kept_token_count": 572.246875, + "learning_rate": 2.986478491689952e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9918376386165619, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008162374421954156, + "mgd_loss_weight": 0.9095136523246765, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.6166687011719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.475537109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2720, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.29992890954017637, + "epoch": 0.7942973523421588, + "flow_matching_loss": 0.014991690963506698, + "grad_norm": 0.11337137222290039, + "kept_token_count": 572.8375, + "learning_rate": 2.986256076618961e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9909594237804413, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009040579758584499, + "mgd_loss_weight": 0.9091803073883057, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 296.30613403320314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.6528289794922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2730, + "valid_token_count": 818.25703125 + }, + { + "actual_token_mask_ratio": 0.30035469233989714, + "epoch": 0.7972068664533023, + "flow_matching_loss": 0.015599800739437342, + "grad_norm": 0.11193813383579254, + "kept_token_count": 572.4765625, + "learning_rate": 2.986031855606692e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.991378927230835, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008621058240532876, + "mgd_loss_weight": 0.9088469564914703, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.6975067138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.4412506103516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2740, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.30062404870986936, + "epoch": 0.8001163805644458, + "flow_matching_loss": 0.017678539268672466, + "grad_norm": 0.10450281202793121, + "kept_token_count": 572.4421875, + "learning_rate": 2.9858058289255954e-05, + "loss": 0.0261, + "mgd_cos_sim": 0.9907728254795074, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009227180574089288, + "mgd_loss_weight": 0.9085136234760285, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 329.02467041015626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.8369995117188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2750, + "valid_token_count": 818.50546875 + }, + { + "actual_token_mask_ratio": 0.3001898914575577, + "epoch": 0.8030258946755892, + "flow_matching_loss": 0.016174093820154667, + "grad_norm": 0.0850616991519928, + "kept_token_count": 572.4859375, + "learning_rate": 2.985577996850314e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9918170034885406, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008182995952665806, + "mgd_loss_weight": 0.9081802666187286, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 312.1159393310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.4735382080078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2760, + "valid_token_count": 818.05390625 + }, + { + "actual_token_mask_ratio": 0.2996708571910858, + "epoch": 0.8059354087867326, + "flow_matching_loss": 0.015670901257544757, + "grad_norm": 0.08813679218292236, + "kept_token_count": 573.1484375, + "learning_rate": 2.9853483596576856e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.991492635011673, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008507357444614172, + "mgd_loss_weight": 0.9078469276428223, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.8911651611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.3951782226562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2770, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.30021598041057584, + "epoch": 0.808844922897876, + "flow_matching_loss": 0.017448237910866738, + "grad_norm": 0.09448104351758957, + "kept_token_count": 572.66875, + "learning_rate": 2.9851169176267412e-05, + "loss": 0.0256, + "mgd_cos_sim": 0.9910700798034668, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008929920801892876, + "mgd_loss_weight": 0.9075135886669159, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 317.5924865722656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.7792541503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2780, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.3001310497522354, + "epoch": 0.8117544370090195, + "flow_matching_loss": 0.017875050473958255, + "grad_norm": 0.1046818420290947, + "kept_token_count": 572.8984375, + "learning_rate": 2.9848836710387044e-05, + "loss": 0.0257, + "mgd_cos_sim": 0.9914255857467651, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008574413135647774, + "mgd_loss_weight": 0.907180231809616, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 319.51063232421876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.43842163085935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2790, + "valid_token_count": 818.58125 + }, + { + "actual_token_mask_ratio": 0.299153333902359, + "epoch": 0.814663951120163, + "flow_matching_loss": 0.01812336575239897, + "grad_norm": 0.0987393707036972, + "kept_token_count": 573.60078125, + "learning_rate": 2.9846486201769917e-05, + "loss": 0.0261, + "mgd_cos_sim": 0.9912157714366913, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008784226700663566, + "mgd_loss_weight": 0.9068468987941742, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 300.0199829101563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.6694732666016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2800, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.2997509628534317, + "epoch": 0.8175734652313064, + "flow_matching_loss": 0.017468917090445756, + "grad_norm": 0.10466306656599045, + "kept_token_count": 573.02890625, + "learning_rate": 2.9844117653272122e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9916847288608551, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008315265318378806, + "mgd_loss_weight": 0.9065135478973388, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 315.8778411865234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.12958068847655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2810, + "valid_token_count": 818.32578125 + }, + { + "actual_token_mask_ratio": 0.29991876184940336, + "epoch": 0.8204829793424498, + "flow_matching_loss": 0.016662492789328097, + "grad_norm": 0.11713580042123795, + "kept_token_count": 572.78984375, + "learning_rate": 2.9841731067771676e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9910653054714202, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00893469862639904, + "mgd_loss_weight": 0.906180202960968, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 323.3450866699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.7468994140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2820, + "valid_token_count": 818.17890625 + }, + { + "actual_token_mask_ratio": 0.30013447999954224, + "epoch": 0.8233924934535932, + "flow_matching_loss": 0.016546476259827615, + "grad_norm": 0.10657943040132523, + "kept_token_count": 572.725, + "learning_rate": 2.9839326448168492e-05, + "loss": 0.0249, + "mgd_cos_sim": 0.9907708287239074, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009229174070060252, + "mgd_loss_weight": 0.9058468639850616, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 329.5822387695313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.9345397949219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2830, + "valid_token_count": 818.33359375 + }, + { + "actual_token_mask_ratio": 0.29965759813785553, + "epoch": 0.8263020075647367, + "flow_matching_loss": 0.01630048230290413, + "grad_norm": 0.0890696719288826, + "kept_token_count": 572.9296875, + "learning_rate": 2.983690379738442e-05, + "loss": 0.0245, + "mgd_cos_sim": 0.990907621383667, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009092384623363614, + "mgd_loss_weight": 0.9055135130882264, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 314.12493591308595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.3870361328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2840, + "valid_token_count": 818.07265625 + }, + { + "actual_token_mask_ratio": 0.2996024668216705, + "epoch": 0.8292115216758801, + "flow_matching_loss": 0.01575464243069291, + "grad_norm": 0.09101409465074539, + "kept_token_count": 573.22578125, + "learning_rate": 2.9834463118363204e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9912400424480439, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0087599687743932, + "mgd_loss_weight": 0.90518017411232, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 323.67149353027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.69773254394534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2850, + "valid_token_count": 818.42734375 + }, + { + "actual_token_mask_ratio": 0.30027708113193513, + "epoch": 0.8321210357870236, + "flow_matching_loss": 0.016609236132353544, + "grad_norm": 0.10804763436317444, + "kept_token_count": 572.62265625, + "learning_rate": 2.98320044140705e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.9914055109024048, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008594497060403227, + "mgd_loss_weight": 0.904846829175949, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 303.4700134277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.56097106933595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2860, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.300858798623085, + "epoch": 0.835030549898167, + "flow_matching_loss": 0.018130486272275446, + "grad_norm": 0.10602765530347824, + "kept_token_count": 571.94375, + "learning_rate": 2.9829527687493868e-05, + "loss": 0.026, + "mgd_cos_sim": 0.9912862777709961, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00871373019181192, + "mgd_loss_weight": 0.9045134782791138, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 319.2729858398437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.5414184570312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2870, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.29946301579475404, + "epoch": 0.8379400640093104, + "flow_matching_loss": 0.015063322521746159, + "grad_norm": 0.08578293025493622, + "kept_token_count": 573.234375, + "learning_rate": 2.9827032941642765e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9916955769062042, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008304430358111858, + "mgd_loss_weight": 0.9041801393032074, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 326.9252899169922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.7379974365234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2880, + "valid_token_count": 818.27890625 + }, + { + "actual_token_mask_ratio": 0.29960193336009977, + "epoch": 0.8408495781204539, + "flow_matching_loss": 0.01626570140942931, + "grad_norm": 0.1027412936091423, + "kept_token_count": 573.03046875, + "learning_rate": 2.9824520179548535e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9914998769760132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008500131824985147, + "mgd_loss_weight": 0.903846800327301, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.7425933837891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.2248016357422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2890, + "valid_token_count": 818.14765625 + }, + { + "actual_token_mask_ratio": 0.2993305712938309, + "epoch": 0.8437590922315973, + "flow_matching_loss": 0.015368407033383847, + "grad_norm": 0.11800882965326309, + "kept_token_count": 573.3, + "learning_rate": 2.9821989404264428e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9914625823497772, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008537413971498608, + "mgd_loss_weight": 0.9035134494304657, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 321.3714233398438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.28711853027346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2900, + "valid_token_count": 818.21796875 + }, + { + "actual_token_mask_ratio": 0.2998376339673996, + "epoch": 0.8466686063427408, + "flow_matching_loss": 0.01643909029662609, + "grad_norm": 0.08141608536243439, + "kept_token_count": 572.82265625, + "learning_rate": 2.9819440618865572e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9915519535541535, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00844804341904819, + "mgd_loss_weight": 0.9031801044940948, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.31761779785154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.64470825195315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2910, + "valid_token_count": 818.12890625 + }, + { + "actual_token_mask_ratio": 0.29940766990184786, + "epoch": 0.8495781204538843, + "flow_matching_loss": 0.015004873182624579, + "grad_norm": 0.1010722741484642, + "kept_token_count": 573.33828125, + "learning_rate": 2.981687382644898e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9916248679161072, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008375129895284772, + "mgd_loss_weight": 0.9028467655181884, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 302.7353790283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.96226501464844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2920, + "valid_token_count": 818.35859375 + }, + { + "actual_token_mask_ratio": 0.29925849437713625, + "epoch": 0.8524876345650276, + "flow_matching_loss": 0.01675394307821989, + "grad_norm": 0.09728002548217773, + "kept_token_count": 573.18125, + "learning_rate": 2.9814289030133556e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.9913397967815399, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008660205779597164, + "mgd_loss_weight": 0.9025134146213531, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 334.5072448730469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.7180572509766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2930, + "valid_token_count": 817.96484375 + }, + { + "actual_token_mask_ratio": 0.3000120222568512, + "epoch": 0.8553971486761711, + "flow_matching_loss": 0.016031799465417863, + "grad_norm": 0.10816118866205215, + "kept_token_count": 572.696875, + "learning_rate": 2.9811686233060058e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9911820709705352, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008817917760461568, + "mgd_loss_weight": 0.9021800756454468, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.6007568359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.0787628173828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2940, + "valid_token_count": 818.14921875 + }, + { + "actual_token_mask_ratio": 0.30033867359161376, + "epoch": 0.8583066627873145, + "flow_matching_loss": 0.0143449273891747, + "grad_norm": 0.08930982649326324, + "kept_token_count": 572.54453125, + "learning_rate": 2.9809065438391138e-05, + "loss": 0.022, + "mgd_cos_sim": 0.991501373052597, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008498634910210967, + "mgd_loss_weight": 0.901846730709076, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.53582153320315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.7540618896484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2950, + "valid_token_count": 818.31640625 + }, + { + "actual_token_mask_ratio": 0.2998427152633667, + "epoch": 0.861216176898458, + "flow_matching_loss": 0.016570510063320398, + "grad_norm": 0.11145107448101044, + "kept_token_count": 572.80078125, + "learning_rate": 2.9806426649311304e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.9909679114818573, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009032083954662085, + "mgd_loss_weight": 0.9015133798122406, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.5286834716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.31341552734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2960, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.3000577390193939, + "epoch": 0.8641256910096013, + "flow_matching_loss": 0.01755892401561141, + "grad_norm": 0.11919663846492767, + "kept_token_count": 572.84296875, + "learning_rate": 2.980376986902694e-05, + "loss": 0.0257, + "mgd_cos_sim": 0.9909918427467346, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009008152317255736, + "mgd_loss_weight": 0.9011800467967988, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.6390686035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.55132751464845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2970, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.29971631765365603, + "epoch": 0.8670352051207448, + "flow_matching_loss": 0.016720920521765947, + "grad_norm": 0.1075282096862793, + "kept_token_count": 573.19609375, + "learning_rate": 2.9801095100766276e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.9913015604019165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008698445279151201, + "mgd_loss_weight": 0.9008466899394989, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 314.6243591308594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.9261810302734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2980, + "valid_token_count": 818.521875 + }, + { + "actual_token_mask_ratio": 0.3000102072954178, + "epoch": 0.8699447192318883, + "flow_matching_loss": 0.015913626085966824, + "grad_norm": 0.1124255433678627, + "kept_token_count": 572.9140625, + "learning_rate": 2.979840234777942e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9920407712459565, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007959237042814492, + "mgd_loss_weight": 0.9005133509635925, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 327.44028625488284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.1090789794922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 2990, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.2996072918176651, + "epoch": 0.8728542333430317, + "flow_matching_loss": 0.01612160699442029, + "grad_norm": 0.09837185591459274, + "kept_token_count": 573.246875, + "learning_rate": 2.979569161333831e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9912131607532502, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008786838594824075, + "mgd_loss_weight": 0.9001800119876862, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 304.27967529296876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.85862426757814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3000, + "valid_token_count": 818.46953125 + }, + { + "actual_token_mask_ratio": 0.30036507844924926, + "epoch": 0.8757637474541752, + "flow_matching_loss": 0.016722697671502827, + "grad_norm": 0.10300649702548981, + "kept_token_count": 572.3359375, + "learning_rate": 2.9792962900736744e-05, + "loss": 0.0245, + "mgd_cos_sim": 0.9913960456848144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008603951800614596, + "mgd_loss_weight": 0.8998466551303863, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.0493469238281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.8030517578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3010, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.3003066599369049, + "epoch": 0.8786732615653186, + "flow_matching_loss": 0.01627296321094036, + "grad_norm": 0.10614306479692459, + "kept_token_count": 572.42265625, + "learning_rate": 2.9790216213290372e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9915316760540008, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008468323666602374, + "mgd_loss_weight": 0.8995133221149445, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 308.4333068847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 361.25892028808596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3020, + "valid_token_count": 818.11015625 + }, + { + "actual_token_mask_ratio": 0.29947439432144163, + "epoch": 0.881582775676462, + "flow_matching_loss": 0.015596289280802011, + "grad_norm": 0.10107269883155823, + "kept_token_count": 573.275, + "learning_rate": 2.9787451554336672e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.991337651014328, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008662352384999395, + "mgd_loss_weight": 0.8991799712181091, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.1078369140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.44850463867186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3030, + "valid_token_count": 818.34453125 + }, + { + "actual_token_mask_ratio": 0.3003026157617569, + "epoch": 0.8844922897876055, + "flow_matching_loss": 0.01749651078134775, + "grad_norm": 0.11460654437541962, + "kept_token_count": 572.5328125, + "learning_rate": 2.9784668927234972e-05, + "loss": 0.0256, + "mgd_cos_sim": 0.9910105764865875, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008989428170025349, + "mgd_loss_weight": 0.8988466262817383, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 320.3437744140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.3163360595703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3040, + "valid_token_count": 818.25859375 + }, + { + "actual_token_mask_ratio": 0.30004723072052003, + "epoch": 0.8874018038987489, + "flow_matching_loss": 0.015540785528719426, + "grad_norm": 0.12499184906482697, + "kept_token_count": 572.85546875, + "learning_rate": 2.9781868335366418e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9917585909366607, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00824140957556665, + "mgd_loss_weight": 0.8985132873058319, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.54286499023436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.37232971191406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3050, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.29999827444553373, + "epoch": 0.8903113180098924, + "flow_matching_loss": 0.018142173066735268, + "grad_norm": 0.10998249053955078, + "kept_token_count": 572.9546875, + "learning_rate": 2.9779049782134e-05, + "loss": 0.0261, + "mgd_cos_sim": 0.9911708235740662, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00882917819544673, + "mgd_loss_weight": 0.8981799364089966, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 301.8980682373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.20667419433596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3060, + "valid_token_count": 818.50390625 + }, + { + "actual_token_mask_ratio": 0.3004422217607498, + "epoch": 0.8932208321210358, + "flow_matching_loss": 0.01663860259577632, + "grad_norm": 0.09482365846633911, + "kept_token_count": 572.24453125, + "learning_rate": 2.9776213270962523e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9917053341865539, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008294663857668638, + "mgd_loss_weight": 0.8978465974330903, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 334.6912475585938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.25368347167966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3070, + "valid_token_count": 818.00703125 + }, + { + "actual_token_mask_ratio": 0.2998256951570511, + "epoch": 0.8961303462321792, + "flow_matching_loss": 0.01490835975855589, + "grad_norm": 0.09958137571811676, + "kept_token_count": 572.9375, + "learning_rate": 2.977335880529861e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9913081169128418, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008691867114976048, + "mgd_loss_weight": 0.8975132524967193, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.0934814453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.64556274414065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3080, + "valid_token_count": 818.27578125 + }, + { + "actual_token_mask_ratio": 0.29986329674720763, + "epoch": 0.8990398603433226, + "flow_matching_loss": 0.013720017112791538, + "grad_norm": 0.09270965307950974, + "kept_token_count": 572.8578125, + "learning_rate": 2.977048638861072e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9912071049213409, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008792890096083283, + "mgd_loss_weight": 0.897179901599884, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.7660858154297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.30103454589846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3090, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.299615278840065, + "epoch": 0.9019493744544661, + "flow_matching_loss": 0.016323797032237054, + "grad_norm": 0.09857749193906784, + "kept_token_count": 573.134375, + "learning_rate": 2.97675960243891e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9915250301361084, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008474972238764167, + "mgd_loss_weight": 0.8968465626239777, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 313.12986145019534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.5242065429687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3100, + "valid_token_count": 818.31328125 + }, + { + "actual_token_mask_ratio": 0.3000666558742523, + "epoch": 0.9048588885656096, + "flow_matching_loss": 0.01654495419934392, + "grad_norm": 0.11395306885242462, + "kept_token_count": 572.65859375, + "learning_rate": 2.976468771614582e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9913138508796692, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008686145488172769, + "mgd_loss_weight": 0.8965132176876068, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 329.8677215576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.06004333496094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3110, + "valid_token_count": 818.15703125 + }, + { + "actual_token_mask_ratio": 0.30010400116443636, + "epoch": 0.907768402676753, + "flow_matching_loss": 0.015940205194056034, + "grad_norm": 0.10777396708726883, + "kept_token_count": 572.7078125, + "learning_rate": 2.9761761467414736e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9916197955608368, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008380206022411584, + "mgd_loss_weight": 0.896179872751236, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.0993133544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.1065185546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3120, + "valid_token_count": 818.28671875 + }, + { + "actual_token_mask_ratio": 0.2995474606752396, + "epoch": 0.9106779167878964, + "flow_matching_loss": 0.015034732781350613, + "grad_norm": 0.09150128066539764, + "kept_token_count": 573.13203125, + "learning_rate": 2.9758817281751538e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9919507801532745, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008049226459115743, + "mgd_loss_weight": 0.8958465278148651, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 323.5798278808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.50030822753905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3130, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.2994169682264328, + "epoch": 0.9135874308990398, + "flow_matching_loss": 0.014542723167687654, + "grad_norm": 0.10147879272699356, + "kept_token_count": 573.20078125, + "learning_rate": 2.975585516273367e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9915965855121612, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008403409970924258, + "mgd_loss_weight": 0.8955131828784942, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.62583618164064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.2854827880859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3140, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.30018151700496676, + "epoch": 0.9164969450101833, + "flow_matching_loss": 0.016136753279715775, + "grad_norm": 0.1063334047794342, + "kept_token_count": 572.73203125, + "learning_rate": 2.975287511396039e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9915419757366181, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008458019094541669, + "mgd_loss_weight": 0.8951798379421234, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 326.2118865966797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.17193603515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3150, + "valid_token_count": 818.40078125 + }, + { + "actual_token_mask_ratio": 0.29935707449913024, + "epoch": 0.9194064591213268, + "flow_matching_loss": 0.01567433774471283, + "grad_norm": 0.09993688762187958, + "kept_token_count": 573.315625, + "learning_rate": 2.9749877139052745e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.99144749045372, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008552519837394357, + "mgd_loss_weight": 0.894846498966217, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.36704711914064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.277099609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3160, + "valid_token_count": 818.26953125 + }, + { + "actual_token_mask_ratio": 0.29991941452026366, + "epoch": 0.9223159732324702, + "flow_matching_loss": 0.01572161531075835, + "grad_norm": 0.11593227833509445, + "kept_token_count": 572.88984375, + "learning_rate": 2.9746861241653555e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9916135191917419, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008386484626680613, + "mgd_loss_weight": 0.8945131480693818, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 315.6481048583984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.3576690673828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3170, + "valid_token_count": 818.31171875 + }, + { + "actual_token_mask_ratio": 0.3005938082933426, + "epoch": 0.9252254873436137, + "flow_matching_loss": 0.016972817480564117, + "grad_norm": 0.11819477379322052, + "kept_token_count": 572.25625, + "learning_rate": 2.9743827425427416e-05, + "loss": 0.0245, + "mgd_cos_sim": 0.991575300693512, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008424690412357449, + "mgd_loss_weight": 0.8941798031330108, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.3057556152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.9690795898438, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3180, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.30034820139408114, + "epoch": 0.928135001454757, + "flow_matching_loss": 0.015204628184437751, + "grad_norm": 0.09595781564712524, + "kept_token_count": 572.67734375, + "learning_rate": 2.9740775694060704e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9913689374923706, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008631078759208322, + "mgd_loss_weight": 0.8938464641571044, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 317.7878845214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.0454498291016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3190, + "valid_token_count": 818.51328125 + }, + { + "actual_token_mask_ratio": 0.30006442666053773, + "epoch": 0.9310445155659005, + "flow_matching_loss": 0.01805764203891158, + "grad_norm": 0.11216616630554199, + "kept_token_count": 572.71328125, + "learning_rate": 2.973770605126156e-05, + "loss": 0.0259, + "mgd_cos_sim": 0.9912493824958801, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008750618807971478, + "mgd_loss_weight": 0.8935131132602692, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.7398956298828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.8529357910156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3200, + "valid_token_count": 818.23828125 + }, + { + "actual_token_mask_ratio": 0.3008326917886734, + "epoch": 0.9339540296770439, + "flow_matching_loss": 0.017559347953647376, + "grad_norm": 0.0946555882692337, + "kept_token_count": 572.07734375, + "learning_rate": 2.9734618500759886e-05, + "loss": 0.0253, + "mgd_cos_sim": 0.9913813769817352, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008618636196479202, + "mgd_loss_weight": 0.8931797742843628, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.327392578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.5885833740234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3210, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.29985817074775695, + "epoch": 0.9368635437881874, + "flow_matching_loss": 0.01608071718364954, + "grad_norm": 0.09728346019983292, + "kept_token_count": 572.7625, + "learning_rate": 2.9731513046307356e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9910426676273346, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008957344945520163, + "mgd_loss_weight": 0.892846429347992, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.6466522216797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.1281677246094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3220, + "valid_token_count": 818.06640625 + }, + { + "actual_token_mask_ratio": 0.29966971278190613, + "epoch": 0.9397730578993309, + "flow_matching_loss": 0.015651555359363557, + "grad_norm": 0.09718504548072815, + "kept_token_count": 573.09296875, + "learning_rate": 2.972838969167738e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9915145218372345, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008485466660931706, + "mgd_loss_weight": 0.8925130784511566, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 324.451171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.81407165527344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3230, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.2999723881483078, + "epoch": 0.9426825720104742, + "flow_matching_loss": 0.014828610513359309, + "grad_norm": 0.10554530471563339, + "kept_token_count": 572.86875, + "learning_rate": 2.972524844066513e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9916798114776612, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008320189453661442, + "mgd_loss_weight": 0.8921797454357148, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 340.30931091308594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.01813049316405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3240, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.29985865354537966, + "epoch": 0.9455920861216177, + "flow_matching_loss": 0.0170354881323874, + "grad_norm": 0.11333117634057999, + "kept_token_count": 572.88203125, + "learning_rate": 2.9722089297087524e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9913411319255829, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008658877667039632, + "mgd_loss_weight": 0.8918463885784149, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 323.2137481689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.3202789306641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3250, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.2999477595090866, + "epoch": 0.9485016002327611, + "flow_matching_loss": 0.01662492789328098, + "grad_norm": 0.09204617887735367, + "kept_token_count": 572.8234375, + "learning_rate": 2.971891226478322e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.9909887373447418, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00901125743985176, + "mgd_loss_weight": 0.8915130496025085, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.9978363037109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.0592468261719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3260, + "valid_token_count": 818.26015625 + }, + { + "actual_token_mask_ratio": 0.2996887445449829, + "epoch": 0.9514111143439046, + "flow_matching_loss": 0.014668772742152215, + "grad_norm": 0.10226868838071823, + "kept_token_count": 572.88515625, + "learning_rate": 2.9715717347612615e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9917020320892334, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008297963952645659, + "mgd_loss_weight": 0.8911797106266022, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.8736297607422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.5028472900391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3270, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.30044215321540835, + "epoch": 0.9543206284550481, + "flow_matching_loss": 0.015892614889889956, + "grad_norm": 0.10826101899147034, + "kept_token_count": 572.2875, + "learning_rate": 2.971250454945783e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9912371695041656, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008762856433168053, + "mgd_loss_weight": 0.8908463537693023, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.82664184570314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.6042785644531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3280, + "valid_token_count": 818.06875 + }, + { + "actual_token_mask_ratio": 0.29939620196819305, + "epoch": 0.9572301425661914, + "flow_matching_loss": 0.01595190018415451, + "grad_norm": 0.10123433917760849, + "kept_token_count": 573.37421875, + "learning_rate": 2.970927387422272e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9911155045032501, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008884503366425633, + "mgd_loss_weight": 0.8905130207538605, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.5683959960937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.5964721679687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3290, + "valid_token_count": 818.40390625 + }, + { + "actual_token_mask_ratio": 0.3000210404396057, + "epoch": 0.9601396566773349, + "flow_matching_loss": 0.016884199157357217, + "grad_norm": 0.105926513671875, + "kept_token_count": 572.61640625, + "learning_rate": 2.9706025325832858e-05, + "loss": 0.0245, + "mgd_cos_sim": 0.9914812028408051, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008518778765574098, + "mgd_loss_weight": 0.8901796698570251, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.31488647460935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.4355865478516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3300, + "valid_token_count": 818.04765625 + }, + { + "actual_token_mask_ratio": 0.30016463696956636, + "epoch": 0.9630491707884783, + "flow_matching_loss": 0.017261251620948315, + "grad_norm": 0.09786217659711838, + "kept_token_count": 572.59140625, + "learning_rate": 2.9702758908235534e-05, + "loss": 0.025, + "mgd_cos_sim": 0.99128338098526, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008716627350077033, + "mgd_loss_weight": 0.8898463249206543, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.45896606445314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.68658447265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3310, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.300124517083168, + "epoch": 0.9659586848996218, + "flow_matching_loss": 0.018299470841884612, + "grad_norm": 0.10622949153184891, + "kept_token_count": 572.79375, + "learning_rate": 2.9699474625399765e-05, + "loss": 0.0259, + "mgd_cos_sim": 0.9915027022361755, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008497304515913129, + "mgd_loss_weight": 0.8895129859447479, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.1433410644531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.9290435791016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3320, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.299992573261261, + "epoch": 0.9688681990107652, + "flow_matching_loss": 0.017113642953336238, + "grad_norm": 0.10371142625808716, + "kept_token_count": 572.87109375, + "learning_rate": 2.9696172481316246e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.9914906740188598, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008509316528216005, + "mgd_loss_weight": 0.8891796410083771, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.82594299316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.11732788085936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3330, + "valid_token_count": 818.37890625 + }, + { + "actual_token_mask_ratio": 0.30021817088127134, + "epoch": 0.9717777131219086, + "flow_matching_loss": 0.015604731533676386, + "grad_norm": 0.09466497600078583, + "kept_token_count": 572.53984375, + "learning_rate": 2.9692852479997407e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.991469395160675, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008530606189742685, + "mgd_loss_weight": 0.8888462960720063, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.7404510498047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.11664123535155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3340, + "valid_token_count": 818.16484375 + }, + { + "actual_token_mask_ratio": 0.3008160382509232, + "epoch": 0.9746872272330521, + "flow_matching_loss": 0.01623265566304326, + "grad_norm": 0.10372858494520187, + "kept_token_count": 572.1484375, + "learning_rate": 2.9689514625477364e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.991459995508194, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008540007146075368, + "mgd_loss_weight": 0.8885129511356353, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.967041015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.3112518310547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3350, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.3002621978521347, + "epoch": 0.9775967413441955, + "flow_matching_loss": 0.0183151762932539, + "grad_norm": 0.10798584669828415, + "kept_token_count": 572.68671875, + "learning_rate": 2.968615892181192e-05, + "loss": 0.0262, + "mgd_cos_sim": 0.9910874485969543, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008912554662674665, + "mgd_loss_weight": 0.8881796061992645, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 326.27097778320314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.8919647216797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3360, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.29953921139240264, + "epoch": 0.980506255455339, + "flow_matching_loss": 0.01630024444311857, + "grad_norm": 0.1049310490489006, + "kept_token_count": 573.1515625, + "learning_rate": 2.9682785373078567e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9916280388832093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008371967077255248, + "mgd_loss_weight": 0.8878462612628937, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 324.7729766845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.860498046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3370, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.30050920844078066, + "epoch": 0.9834157695664824, + "flow_matching_loss": 0.017950238101184367, + "grad_norm": 0.10248858481645584, + "kept_token_count": 572.3609375, + "learning_rate": 2.9679393983376493e-05, + "loss": 0.0257, + "mgd_cos_sim": 0.9913129091262818, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008687102422118186, + "mgd_loss_weight": 0.8875129222869873, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.65209045410154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.24775390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3380, + "valid_token_count": 818.25703125 + }, + { + "actual_token_mask_ratio": 0.30038895905017854, + "epoch": 0.9863252836776258, + "flow_matching_loss": 0.016535849869251253, + "grad_norm": 0.12777520716190338, + "kept_token_count": 572.3640625, + "learning_rate": 2.9675984756826545e-05, + "loss": 0.0242, + "mgd_cos_sim": 0.9913978338241577, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008602154860273003, + "mgd_loss_weight": 0.887179571390152, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.392431640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.16742248535155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3390, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.30050723254680634, + "epoch": 0.9892347977887693, + "flow_matching_loss": 0.015996272955089808, + "grad_norm": 0.09938716143369675, + "kept_token_count": 572.43359375, + "learning_rate": 2.9672557697571273e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9913395643234253, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008660433068871497, + "mgd_loss_weight": 0.8868462264537811, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.8291931152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.4778778076172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3400, + "valid_token_count": 818.35078125 + }, + { + "actual_token_mask_ratio": 0.2996692568063736, + "epoch": 0.9921443118999127, + "flow_matching_loss": 0.018415915407240392, + "grad_norm": 0.10072813183069229, + "kept_token_count": 573.034375, + "learning_rate": 2.966911280977486e-05, + "loss": 0.0263, + "mgd_cos_sim": 0.9911477208137512, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008852278906852007, + "mgd_loss_weight": 0.8865128874778747, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.18585510253905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.1966827392578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3410, + "valid_token_count": 818.23515625 + }, + { + "actual_token_mask_ratio": 0.30002795457839965, + "epoch": 0.9950538260110562, + "flow_matching_loss": 0.015329874400049449, + "grad_norm": 0.0955052301287651, + "kept_token_count": 572.72421875, + "learning_rate": 2.966565009762318e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9913738548755646, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008626159373670816, + "mgd_loss_weight": 0.8861795365810394, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 316.6629699707031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.5087097167969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3420, + "valid_token_count": 818.21171875 + }, + { + "actual_token_mask_ratio": 0.30007174909114837, + "epoch": 0.9979633401221996, + "flow_matching_loss": 0.016216025967150927, + "grad_norm": 0.10649619251489639, + "kept_token_count": 572.678125, + "learning_rate": 2.966216956532375e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9910869657993316, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00891302078962326, + "mgd_loss_weight": 0.8858461976051331, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.8990997314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.3701202392578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3430, + "valid_token_count": 818.19453125 + }, + { + "actual_token_mask_ratio": 0.30067131817340853, + "epoch": 1.000872854233343, + "flow_matching_loss": 0.01613221298903227, + "grad_norm": 0.11945908516645432, + "kept_token_count": 572.343310546875, + "learning_rate": 2.9658671217105744e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9915541112422943, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00844588689506054, + "mgd_loss_weight": 0.8855128526687622, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.7915557861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.2622802734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3440, + "valid_token_count": 818.4120910644531 + }, + { + "actual_token_mask_ratio": 0.299643275141716, + "epoch": 1.0037823683444864, + "flow_matching_loss": 0.016206534951925276, + "grad_norm": 0.11205686628818512, + "kept_token_count": 573.01328125, + "learning_rate": 2.965515505721999e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9915840446949005, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008415952231734991, + "mgd_loss_weight": 0.8851795017719268, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 331.45009765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.4548828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3450, + "valid_token_count": 818.17109375 + }, + { + "actual_token_mask_ratio": 0.3002898871898651, + "epoch": 1.00669188245563, + "flow_matching_loss": 0.017049240227788687, + "grad_norm": 0.08595520257949829, + "kept_token_count": 572.4875, + "learning_rate": 2.9651621089938948e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.991509348154068, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008490654826164245, + "mgd_loss_weight": 0.884846168756485, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.8526245117188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.50475158691404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3460, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.2997012585401535, + "epoch": 1.0096013965667734, + "flow_matching_loss": 0.015341866947710513, + "grad_norm": 0.0971260741353035, + "kept_token_count": 573.05859375, + "learning_rate": 2.9648069319556725e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9914613425731659, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008538666786625982, + "mgd_loss_weight": 0.8845128118991852, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 327.79893188476564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.17145385742185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3470, + "valid_token_count": 818.30078125 + }, + { + "actual_token_mask_ratio": 0.30039735436439513, + "epoch": 1.0125109106779169, + "flow_matching_loss": 0.015983034670352936, + "grad_norm": 0.08815360069274902, + "kept_token_count": 572.52109375, + "learning_rate": 2.9644499750389057e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.991894143819809, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00810584737919271, + "mgd_loss_weight": 0.8841794729232788, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 329.2691680908203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.1774078369141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3480, + "valid_token_count": 818.34765625 + }, + { + "actual_token_mask_ratio": 0.30037271678447724, + "epoch": 1.0154204247890601, + "flow_matching_loss": 0.016532670799642803, + "grad_norm": 0.12584702670574188, + "kept_token_count": 572.58046875, + "learning_rate": 2.9640912386773305e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9917067587375641, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008293223893269896, + "mgd_loss_weight": 0.8838461339473724, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.3727294921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.837939453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3490, + "valid_token_count": 818.40390625 + }, + { + "actual_token_mask_ratio": 0.2997011750936508, + "epoch": 1.0183299389002036, + "flow_matching_loss": 0.014766738377511501, + "grad_norm": 0.10598361492156982, + "kept_token_count": 573.15, + "learning_rate": 2.9637307233068455e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.991709154844284, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00829084445722401, + "mgd_loss_weight": 0.8835127770900726, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.2541107177734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.99956665039065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3500, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.3000326186418533, + "epoch": 1.021239453011347, + "flow_matching_loss": 0.016389086563140153, + "grad_norm": 0.09244756400585175, + "kept_token_count": 572.74140625, + "learning_rate": 2.9633684293655108e-05, + "loss": 0.0242, + "mgd_cos_sim": 0.9911364555358887, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008863551681861282, + "mgd_loss_weight": 0.8831794440746308, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.37567443847655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.2911743164062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3510, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.2995055019855499, + "epoch": 1.0241489671224906, + "flow_matching_loss": 0.015148774534463883, + "grad_norm": 0.08759444206953049, + "kept_token_count": 573.22109375, + "learning_rate": 2.963004357293548e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9913232743740081, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008676740108057857, + "mgd_loss_weight": 0.8828460931777954, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 328.88104248046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.91099243164064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3520, + "valid_token_count": 818.30703125 + }, + { + "actual_token_mask_ratio": 0.2996843338012695, + "epoch": 1.027058481233634, + "flow_matching_loss": 0.015176471974700689, + "grad_norm": 0.11506817489862442, + "kept_token_count": 572.90234375, + "learning_rate": 2.962638507533339e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9912132978439331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008786693960428239, + "mgd_loss_weight": 0.8825127482414246, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.92258605957034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.1697113037109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3530, + "valid_token_count": 818.0671875 + }, + { + "actual_token_mask_ratio": 0.2996171176433563, + "epoch": 1.0299679953447773, + "flow_matching_loss": 0.016107574664056303, + "grad_norm": 0.09055285900831223, + "kept_token_count": 573.06875, + "learning_rate": 2.962270880529425e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9917018592357636, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008298143604770302, + "mgd_loss_weight": 0.8821794092655182, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 323.2476531982422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.9152099609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3540, + "valid_token_count": 818.21953125 + }, + { + "actual_token_mask_ratio": 0.29950434267520903, + "epoch": 1.0328775094559208, + "flow_matching_loss": 0.01588887022808194, + "grad_norm": 0.09153272211551666, + "kept_token_count": 573.153125, + "learning_rate": 2.9619014767285084e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.991683965921402, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008316033892333508, + "mgd_loss_weight": 0.8818460583686829, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 326.0485321044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.6700805664062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3550, + "valid_token_count": 818.21015625 + }, + { + "actual_token_mask_ratio": 0.3000816136598587, + "epoch": 1.0357870235670643, + "flow_matching_loss": 0.015822464041411875, + "grad_norm": 0.08968257158994675, + "kept_token_count": 572.6015625, + "learning_rate": 2.9615302965794488e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9916886806488037, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008311321679502725, + "mgd_loss_weight": 0.8815127193927765, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.7345306396484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.0001159667969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3560, + "valid_token_count": 818.0984375 + }, + { + "actual_token_mask_ratio": 0.30024919509887693, + "epoch": 1.0386965376782078, + "flow_matching_loss": 0.015831143688410522, + "grad_norm": 0.09011688083410263, + "kept_token_count": 572.5421875, + "learning_rate": 2.961157340533266e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9918830811977386, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008116922480985523, + "mgd_loss_weight": 0.8811793744564056, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 327.89027099609376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.2411376953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3570, + "valid_token_count": 818.21015625 + }, + { + "actual_token_mask_ratio": 0.3008967757225037, + "epoch": 1.0416060517893513, + "flow_matching_loss": 0.015093412436544894, + "grad_norm": 0.07883888483047485, + "kept_token_count": 572.03515625, + "learning_rate": 2.9607826090431364e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9913886606693267, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008611325174570084, + "mgd_loss_weight": 0.8808460235595703, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 340.89075927734376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.4861358642578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3580, + "valid_token_count": 818.23984375 + }, + { + "actual_token_mask_ratio": 0.29899742007255553, + "epoch": 1.0445155659004945, + "flow_matching_loss": 0.016687440406531095, + "grad_norm": 0.08365984261035919, + "kept_token_count": 573.684375, + "learning_rate": 2.9604061025643936e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9919132947921753, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00808669999241829, + "mgd_loss_weight": 0.8805126845836639, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.71962280273436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 362.4578887939453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3590, + "valid_token_count": 818.37734375 + }, + { + "actual_token_mask_ratio": 0.29972185492515563, + "epoch": 1.047425080011638, + "flow_matching_loss": 0.01620092811062932, + "grad_norm": 0.09988983720541, + "kept_token_count": 573.2125, + "learning_rate": 2.9600278215545292e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9913430690765381, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008656926220282912, + "mgd_loss_weight": 0.8801793396472931, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.9087799072266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.4629211425781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3600, + "valid_token_count": 818.553125 + }, + { + "actual_token_mask_ratio": 0.2999051004648209, + "epoch": 1.0503345941227815, + "flow_matching_loss": 0.01480531096458435, + "grad_norm": 0.10653846710920334, + "kept_token_count": 572.86484375, + "learning_rate": 2.95964776647319e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9915803611278534, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00841962993144989, + "mgd_loss_weight": 0.8798459947109223, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.4679382324219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.47691040039064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3610, + "valid_token_count": 818.27265625 + }, + { + "actual_token_mask_ratio": 0.29993079602718353, + "epoch": 1.053244108233925, + "flow_matching_loss": 0.017008900176733732, + "grad_norm": 0.1010105162858963, + "kept_token_count": 572.89375, + "learning_rate": 2.9592659377821787e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.9915791928768158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008420814853161573, + "mgd_loss_weight": 0.8795126497745513, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.24390869140626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.46929321289065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3620, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.29988665878772736, + "epoch": 1.0561536223450685, + "flow_matching_loss": 0.015575371496379376, + "grad_norm": 0.08217502385377884, + "kept_token_count": 572.89453125, + "learning_rate": 2.9588823359454528e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9921465039253234, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007853502873331308, + "mgd_loss_weight": 0.8791793048381805, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.9622283935547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.9423095703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3630, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.29991415739059446, + "epoch": 1.0590631364562118, + "flow_matching_loss": 0.0162224605679512, + "grad_norm": 0.11593392491340637, + "kept_token_count": 572.77109375, + "learning_rate": 2.958496961429126e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9914362251758575, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00856376914307475, + "mgd_loss_weight": 0.8788459599018097, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.4194000244141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.3225402832031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3640, + "valid_token_count": 818.14140625 + }, + { + "actual_token_mask_ratio": 0.29998801052570345, + "epoch": 1.0619726505673552, + "flow_matching_loss": 0.016468519531190395, + "grad_norm": 0.1253383606672287, + "kept_token_count": 572.8625, + "learning_rate": 2.9581098147014628e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9914249360561371, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008575069205835461, + "mgd_loss_weight": 0.8785126209259033, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 331.9081604003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.87510681152344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3650, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.2995897799730301, + "epoch": 1.0648821646784987, + "flow_matching_loss": 0.017465340811759232, + "grad_norm": 0.08254294842481613, + "kept_token_count": 573.12578125, + "learning_rate": 2.9577208962328843e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.9917433798313141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00825661988928914, + "mgd_loss_weight": 0.878179270029068, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 327.2647766113281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.52903442382814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3660, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.3000211328268051, + "epoch": 1.0677916787896422, + "flow_matching_loss": 0.015515084192156792, + "grad_norm": 0.11526375263929367, + "kept_token_count": 572.7265625, + "learning_rate": 2.9573302064959633e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9914206385612487, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008579374151304364, + "mgd_loss_weight": 0.8778459250926971, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 331.8292572021484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.6671447753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3670, + "valid_token_count": 818.20703125 + }, + { + "actual_token_mask_ratio": 0.30023294389247895, + "epoch": 1.0707011929007855, + "flow_matching_loss": 0.018078410439193247, + "grad_norm": 0.10023140907287598, + "kept_token_count": 572.5484375, + "learning_rate": 2.9569377459654235e-05, + "loss": 0.0259, + "mgd_cos_sim": 0.9910438418388366, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008956169011071324, + "mgd_loss_weight": 0.8775125861167907, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.47100219726565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.78891906738284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3680, + "valid_token_count": 818.19765625 + }, + { + "actual_token_mask_ratio": 0.30069389939308167, + "epoch": 1.073610707011929, + "flow_matching_loss": 0.01750310156494379, + "grad_norm": 0.11743244528770447, + "kept_token_count": 572.3078125, + "learning_rate": 2.9565435151181424e-05, + "loss": 0.0253, + "mgd_cos_sim": 0.9911361575126648, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008863842347636819, + "mgd_loss_weight": 0.8771792352199554, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 326.6688537597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.0455841064453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3690, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.3002540051937103, + "epoch": 1.0765202211230724, + "flow_matching_loss": 0.017213956732302905, + "grad_norm": 0.100335992872715, + "kept_token_count": 572.7078125, + "learning_rate": 2.9561475144331473e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.9914898693561554, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008510118629783393, + "mgd_loss_weight": 0.8768458962440491, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.06255493164065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.35044250488284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3700, + "valid_token_count": 818.453125 + }, + { + "actual_token_mask_ratio": 0.30017897486686707, + "epoch": 1.079429735234216, + "flow_matching_loss": 0.014212829526513815, + "grad_norm": 0.08429590612649918, + "kept_token_count": 572.5859375, + "learning_rate": 2.9557497443916162e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.992160415649414, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007839568192139268, + "mgd_loss_weight": 0.8765125513076782, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 331.61167907714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.4310638427734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3710, + "valid_token_count": 818.18828125 + }, + { + "actual_token_mask_ratio": 0.301262754201889, + "epoch": 1.0823392493453594, + "flow_matching_loss": 0.015276023279875517, + "grad_norm": 0.09593626856803894, + "kept_token_count": 571.7953125, + "learning_rate": 2.9553502054768774e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9914815127849579, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008518492290750145, + "mgd_loss_weight": 0.8761792004108429, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.3886993408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.0513671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3720, + "valid_token_count": 818.32890625 + }, + { + "actual_token_mask_ratio": 0.2992389440536499, + "epoch": 1.0852487634565027, + "flow_matching_loss": 0.015821308363229036, + "grad_norm": 0.09324038773775101, + "kept_token_count": 573.4, + "learning_rate": 2.954948898174409e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9915808379650116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008419164177030324, + "mgd_loss_weight": 0.875845867395401, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.80751953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.88714904785155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3730, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.29953827857971194, + "epoch": 1.0881582775676462, + "flow_matching_loss": 0.016192798037081957, + "grad_norm": 0.09476304054260254, + "kept_token_count": 573.15703125, + "learning_rate": 2.954545822971836e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9916169285774231, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008383085811510682, + "mgd_loss_weight": 0.8755125105381012, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.3885528564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.8198638916016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3740, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.30014040470123293, + "epoch": 1.0910677916787896, + "flow_matching_loss": 0.01538548655807972, + "grad_norm": 0.1006806343793869, + "kept_token_count": 572.6921875, + "learning_rate": 2.9541409803589335e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.99156374335289, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008436256367713214, + "mgd_loss_weight": 0.8751791715621948, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.40772399902346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.6017608642578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3750, + "valid_token_count": 818.29453125 + }, + { + "actual_token_mask_ratio": 0.30052005052566527, + "epoch": 1.0939773057899331, + "flow_matching_loss": 0.016106243152171372, + "grad_norm": 0.1008625328540802, + "kept_token_count": 572.2515625, + "learning_rate": 2.9537343708276235e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9910245180130005, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008975483244284987, + "mgd_loss_weight": 0.8748458325862885, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.3805450439453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.09507446289064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3760, + "valid_token_count": 818.11328125 + }, + { + "actual_token_mask_ratio": 0.3000652015209198, + "epoch": 1.0968868199010766, + "flow_matching_loss": 0.015948275569826365, + "grad_norm": 0.0771438404917717, + "kept_token_count": 572.99140625, + "learning_rate": 2.953325994871975e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.991604870557785, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008395136008039117, + "mgd_loss_weight": 0.8745124816894532, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 342.30628356933596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.29567565917966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3770, + "valid_token_count": 818.63359375 + }, + { + "actual_token_mask_ratio": 0.30001664757728574, + "epoch": 1.0997963340122199, + "flow_matching_loss": 0.017117160465568305, + "grad_norm": 0.09896475076675415, + "kept_token_count": 572.7515625, + "learning_rate": 2.9529158529882023e-05, + "loss": 0.0245, + "mgd_cos_sim": 0.9915198564529419, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008480144571512938, + "mgd_loss_weight": 0.8741791427135468, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.0113830566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.02093200683595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3780, + "valid_token_count": 818.23515625 + }, + { + "actual_token_mask_ratio": 0.299490362405777, + "epoch": 1.1027058481233634, + "flow_matching_loss": 0.014455750491470099, + "grad_norm": 0.10097717493772507, + "kept_token_count": 573.165625, + "learning_rate": 2.9525039456746677e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9918712079524994, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0081288012675941, + "mgd_loss_weight": 0.8738457977771759, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.4335021972656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.5858551025391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3790, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.30021512508392334, + "epoch": 1.1056153622345068, + "flow_matching_loss": 0.015044331178069114, + "grad_norm": 0.09554420411586761, + "kept_token_count": 572.52421875, + "learning_rate": 2.952090273431877e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9919256508350373, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008074348140507937, + "mgd_loss_weight": 0.8735124468803406, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.0945709228516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.7217987060547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3800, + "valid_token_count": 818.14609375 + }, + { + "actual_token_mask_ratio": 0.30046447217464445, + "epoch": 1.1085248763456503, + "flow_matching_loss": 0.015676317550241948, + "grad_norm": 0.10915607959032059, + "kept_token_count": 572.578125, + "learning_rate": 2.9516748367624808e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9914665222167969, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008533471543341876, + "mgd_loss_weight": 0.8731791079044342, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 340.5106964111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.10169982910156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3810, + "valid_token_count": 818.50859375 + }, + { + "actual_token_mask_ratio": 0.3005409687757492, + "epoch": 1.1114343904567938, + "flow_matching_loss": 0.016423432901501656, + "grad_norm": 0.11170583218336105, + "kept_token_count": 572.40546875, + "learning_rate": 2.951257636171274e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9918294310569763, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0081705657299608, + "mgd_loss_weight": 0.8728457629680634, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.4821746826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.2388458251953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3820, + "valid_token_count": 818.35703125 + }, + { + "actual_token_mask_ratio": 0.3004607170820236, + "epoch": 1.114343904567937, + "flow_matching_loss": 0.018140799645334483, + "grad_norm": 0.09374012798070908, + "kept_token_count": 572.44921875, + "learning_rate": 2.9508386721651944e-05, + "loss": 0.0256, + "mgd_cos_sim": 0.9914674758911133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008532528160139918, + "mgd_loss_weight": 0.8725124180316925, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 340.76905212402346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.6040283203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3830, + "valid_token_count": 818.32265625 + }, + { + "actual_token_mask_ratio": 0.2997961163520813, + "epoch": 1.1172534186790806, + "flow_matching_loss": 0.016860159020870923, + "grad_norm": 0.10922709852457047, + "kept_token_count": 572.9, + "learning_rate": 2.9504179452533228e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9918728590011596, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008127125492319465, + "mgd_loss_weight": 0.8721790730953216, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 338.1315643310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.9332214355469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3840, + "valid_token_count": 818.19296875 + }, + { + "actual_token_mask_ratio": 0.29985460340976716, + "epoch": 1.120162932790224, + "flow_matching_loss": 0.01715021673589945, + "grad_norm": 0.10939914733171463, + "kept_token_count": 573.0328125, + "learning_rate": 2.9499954559468826e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9912151336669922, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008784845890477299, + "mgd_loss_weight": 0.8718457281589508, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.5765655517578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.6315643310547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3850, + "valid_token_count": 818.45234375 + }, + { + "actual_token_mask_ratio": 0.3000206261873245, + "epoch": 1.1230724469013675, + "flow_matching_loss": 0.016144500020891427, + "grad_norm": 0.09586500376462936, + "kept_token_count": 572.72265625, + "learning_rate": 2.9495712047592378e-05, + "loss": 0.0242, + "mgd_cos_sim": 0.9907783985137939, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009221598785370589, + "mgd_loss_weight": 0.87151238322258, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.96143493652346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.778857421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3860, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.30031912624835966, + "epoch": 1.1259819610125108, + "flow_matching_loss": 0.016144629381597043, + "grad_norm": 0.0893401950597763, + "kept_token_count": 572.63203125, + "learning_rate": 2.949145192205893e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9918154597282409, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008184531005099416, + "mgd_loss_weight": 0.8711790442466736, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.3277191162109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.73399353027344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3870, + "valid_token_count": 818.41484375 + }, + { + "actual_token_mask_ratio": 0.2996318757534027, + "epoch": 1.1288914751236543, + "flow_matching_loss": 0.016773836966603995, + "grad_norm": 0.0927872285246849, + "kept_token_count": 573.13515625, + "learning_rate": 2.9487174188044953e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9913568437099457, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008643148466944695, + "mgd_loss_weight": 0.8708456933498383, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 329.0850372314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.6462890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3880, + "valid_token_count": 818.33203125 + }, + { + "actual_token_mask_ratio": 0.3010913759469986, + "epoch": 1.1318009892347978, + "flow_matching_loss": 0.01750456253066659, + "grad_norm": 0.09305281937122345, + "kept_token_count": 571.978125, + "learning_rate": 2.9482878850748276e-05, + "loss": 0.0251, + "mgd_cos_sim": 0.9912992358207703, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008700785273686052, + "mgd_loss_weight": 0.8705123484134674, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.2309143066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.49705810546874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3890, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.3005736380815506, + "epoch": 1.1347105033459413, + "flow_matching_loss": 0.015968074183911087, + "grad_norm": 0.09089519083499908, + "kept_token_count": 572.35390625, + "learning_rate": 2.9478565915388157e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.991491287946701, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008508719690144061, + "mgd_loss_weight": 0.870179009437561, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 340.1070098876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.717333984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3900, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.2995152175426483, + "epoch": 1.1376200174570847, + "flow_matching_loss": 0.014118577353656292, + "grad_norm": 0.08245997875928879, + "kept_token_count": 573.25390625, + "learning_rate": 2.9474235387205207e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9909887254238129, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009011289663612842, + "mgd_loss_weight": 0.8698456585407257, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.9958465576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.4172424316406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3910, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.3003595799207687, + "epoch": 1.140529531568228, + "flow_matching_loss": 0.016201662737876175, + "grad_norm": 0.10328376293182373, + "kept_token_count": 572.54453125, + "learning_rate": 2.946988727146143e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.991975212097168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008024779334664344, + "mgd_loss_weight": 0.8695123195648193, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.06566162109374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.1064788818359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3920, + "valid_token_count": 818.33828125 + }, + { + "actual_token_mask_ratio": 0.2999252825975418, + "epoch": 1.1434390456793715, + "flow_matching_loss": 0.016131071094423532, + "grad_norm": 0.10017510503530502, + "kept_token_count": 572.9125, + "learning_rate": 2.9465521573440196e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9912885308265686, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008711470384150743, + "mgd_loss_weight": 0.8691789746284485, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.99879760742186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.10980834960935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3930, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.3000411629676819, + "epoch": 1.146348559790515, + "flow_matching_loss": 0.015458988770842553, + "grad_norm": 0.08850337564945221, + "kept_token_count": 572.73046875, + "learning_rate": 2.946113829844624e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9915635883808136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008436409989371896, + "mgd_loss_weight": 0.8688456237316131, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.4764770507812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.3939971923828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3940, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.3009754031896591, + "epoch": 1.1492580739016585, + "flow_matching_loss": 0.016127416957169773, + "grad_norm": 0.11346711963415146, + "kept_token_count": 571.87578125, + "learning_rate": 2.945673745180566e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9916681766510009, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008331817481666804, + "mgd_loss_weight": 0.8685122907161713, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.6518188476563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.36252746582034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3950, + "valid_token_count": 818.10703125 + }, + { + "actual_token_mask_ratio": 0.30025024712085724, + "epoch": 1.152167588012802, + "flow_matching_loss": 0.01619767351076007, + "grad_norm": 0.09279756247997284, + "kept_token_count": 572.51328125, + "learning_rate": 2.945231903886589e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9919367134571075, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008063286170363426, + "mgd_loss_weight": 0.8681789338588715, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.7689178466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.308984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3960, + "valid_token_count": 818.16953125 + }, + { + "actual_token_mask_ratio": 0.29977173507213595, + "epoch": 1.1550771021239452, + "flow_matching_loss": 0.01655569113790989, + "grad_norm": 0.09649324417114258, + "kept_token_count": 572.87578125, + "learning_rate": 2.9447883064995728e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9920044243335724, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00799557943828404, + "mgd_loss_weight": 0.8678455948829651, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.10637817382815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.8343811035156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3970, + "valid_token_count": 818.12578125 + }, + { + "actual_token_mask_ratio": 0.300562372803688, + "epoch": 1.1579866162350887, + "flow_matching_loss": 0.017109496518969536, + "grad_norm": 0.10788794606924057, + "kept_token_count": 572.35, + "learning_rate": 2.9443429535585288e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.99160937666893, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008390620537102222, + "mgd_loss_weight": 0.8675122559070587, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.3841888427734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.9302185058594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3980, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.3003679871559143, + "epoch": 1.1608961303462322, + "flow_matching_loss": 0.018787152227014305, + "grad_norm": 0.1000562459230423, + "kept_token_count": 572.384375, + "learning_rate": 2.9438958456046044e-05, + "loss": 0.026, + "mgd_cos_sim": 0.9917117297649384, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008288272190839052, + "mgd_loss_weight": 0.8671788990497589, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 335.94330139160155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.57118835449216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 3990, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.3002925395965576, + "epoch": 1.1638056444573757, + "flow_matching_loss": 0.01567877335473895, + "grad_norm": 0.11450810730457306, + "kept_token_count": 572.559375, + "learning_rate": 2.9434469831810765e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9911365568637848, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008863462135195732, + "mgd_loss_weight": 0.8668455660343171, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.0989501953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 363.715966796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4000, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.2999194473028183, + "epoch": 1.1667151585685192, + "flow_matching_loss": 0.014687039237469435, + "grad_norm": 0.09609968215227127, + "kept_token_count": 572.971875, + "learning_rate": 2.9429963668333566e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9912188172340393, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008781194081529975, + "mgd_loss_weight": 0.8665122151374817, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 337.8234039306641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.3277160644531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4010, + "valid_token_count": 818.43359375 + }, + { + "actual_token_mask_ratio": 0.3002409428358078, + "epoch": 1.1696246726796624, + "flow_matching_loss": 0.01632083421573043, + "grad_norm": 0.10466449707746506, + "kept_token_count": 572.64765625, + "learning_rate": 2.9425439971089845e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.991016286611557, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00898371464572847, + "mgd_loss_weight": 0.8661788702011108, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.0601440429688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.0005126953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4020, + "valid_token_count": 818.35703125 + }, + { + "actual_token_mask_ratio": 0.2999497503042221, + "epoch": 1.172534186790806, + "flow_matching_loss": 0.01519060367718339, + "grad_norm": 0.1288101226091385, + "kept_token_count": 572.95234375, + "learning_rate": 2.9420898745576336e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9914577782154084, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00854223002679646, + "mgd_loss_weight": 0.8658455312252045, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 333.147802734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.34258422851565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4030, + "valid_token_count": 818.44453125 + }, + { + "actual_token_mask_ratio": 0.3009901523590088, + "epoch": 1.1754437009019494, + "flow_matching_loss": 0.015434423927217722, + "grad_norm": 0.10471536964178085, + "kept_token_count": 571.97890625, + "learning_rate": 2.9416339997311044e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9916665136814118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008333489391952753, + "mgd_loss_weight": 0.8655121803283692, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.48109741210936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.86842346191406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4040, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.3009176403284073, + "epoch": 1.1783532150130929, + "flow_matching_loss": 0.017299570608884096, + "grad_norm": 0.12206843495368958, + "kept_token_count": 572.003125, + "learning_rate": 2.941176373183329e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9913140177726746, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00868598222732544, + "mgd_loss_weight": 0.8651788413524628, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.8091766357422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.9708282470703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4050, + "valid_token_count": 818.21953125 + }, + { + "actual_token_mask_ratio": 0.2992436677217484, + "epoch": 1.1812627291242364, + "flow_matching_loss": 0.01625403817743063, + "grad_norm": 0.1062650978565216, + "kept_token_count": 573.3640625, + "learning_rate": 2.9407169954703653e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9909265339374542, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009073476959019899, + "mgd_loss_weight": 0.8648454964160919, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.8017150878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.9989837646484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4060, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.30018181800842286, + "epoch": 1.1841722432353796, + "flow_matching_loss": 0.015550210326910018, + "grad_norm": 0.10061972588300705, + "kept_token_count": 572.45859375, + "learning_rate": 2.9402558671504018e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9918184220790863, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008181576058268547, + "mgd_loss_weight": 0.8645121455192566, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.5769439697266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.0832000732422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4070, + "valid_token_count": 818.0125 + }, + { + "actual_token_mask_ratio": 0.2997844398021698, + "epoch": 1.187081757346523, + "flow_matching_loss": 0.017330511845648287, + "grad_norm": 0.1105508804321289, + "kept_token_count": 572.921875, + "learning_rate": 2.9397929887837524e-05, + "loss": 0.0245, + "mgd_cos_sim": 0.9917211949825286, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008278806041926146, + "mgd_loss_weight": 0.8641788065433502, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.0175262451172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.72072143554686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4080, + "valid_token_count": 818.20859375 + }, + { + "actual_token_mask_ratio": 0.3000743806362152, + "epoch": 1.1899912714576666, + "flow_matching_loss": 0.015697537641972303, + "grad_norm": 0.11016659438610077, + "kept_token_count": 572.77734375, + "learning_rate": 2.939328360932858e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9916795670986176, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008320426708087325, + "mgd_loss_weight": 0.8638454616069794, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.6468475341797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.81553649902344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4090, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.30005282163619995, + "epoch": 1.19290078556881, + "flow_matching_loss": 0.016928350552916527, + "grad_norm": 0.09456924349069595, + "kept_token_count": 572.73515625, + "learning_rate": 2.938861984162285e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9915195226669311, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008480464620515705, + "mgd_loss_weight": 0.8635121166706086, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.21477661132815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.34470825195314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4100, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.29975962042808535, + "epoch": 1.1958102996799536, + "flow_matching_loss": 0.014185144286602735, + "grad_norm": 0.1063275933265686, + "kept_token_count": 572.91953125, + "learning_rate": 2.938393859038726e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9913484632968903, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008651537913829088, + "mgd_loss_weight": 0.8631787717342376, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 330.5323883056641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.19447937011716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4110, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.29933212995529174, + "epoch": 1.1987198137910968, + "flow_matching_loss": 0.01505141481757164, + "grad_norm": 0.08682915568351746, + "kept_token_count": 573.478125, + "learning_rate": 2.9379239861309964e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9919614136219025, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008038577297702432, + "mgd_loss_weight": 0.8628454267978668, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.92303771972655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.9638366699219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4120, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.30043748021125793, + "epoch": 1.2016293279022403, + "flow_matching_loss": 0.0158346620388329, + "grad_norm": 0.0933358445763588, + "kept_token_count": 572.35, + "learning_rate": 2.937452366010036e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9912763774394989, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008723622281104326, + "mgd_loss_weight": 0.862512081861496, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.3268157958984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.0586273193359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4130, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.3008385688066483, + "epoch": 1.2045388420133838, + "flow_matching_loss": 0.015529195498675107, + "grad_norm": 0.09100587666034698, + "kept_token_count": 572.16171875, + "learning_rate": 2.936978999248908e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9918406128883361, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008159380825236439, + "mgd_loss_weight": 0.8621787428855896, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.84470825195314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.77528076171876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4140, + "valid_token_count": 818.35546875 + }, + { + "actual_token_mask_ratio": 0.2995974689722061, + "epoch": 1.2074483561245273, + "flow_matching_loss": 0.01575672412291169, + "grad_norm": 0.09280712157487869, + "kept_token_count": 573.253125, + "learning_rate": 2.936503886422798e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9916739165782928, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008326065260916948, + "mgd_loss_weight": 0.8618453919887543, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.2950958251953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.25389404296874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4150, + "valid_token_count": 818.4640625 + }, + { + "actual_token_mask_ratio": 0.29976860582828524, + "epoch": 1.2103578702356708, + "flow_matching_loss": 0.015375111438333987, + "grad_norm": 0.0805748701095581, + "kept_token_count": 572.915625, + "learning_rate": 2.9360270281090122e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9913482427597046, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008651758171617984, + "mgd_loss_weight": 0.8615120470523834, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 325.840478515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.4006072998047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4160, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.3001132011413574, + "epoch": 1.213267384346814, + "flow_matching_loss": 0.016238475497812033, + "grad_norm": 0.09456039220094681, + "kept_token_count": 572.60546875, + "learning_rate": 2.935548424886979e-05, + "loss": 0.024, + "mgd_cos_sim": 0.991044408082962, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008955588098615408, + "mgd_loss_weight": 0.861178708076477, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.4379821777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.4162231445313, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4170, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.2996809661388397, + "epoch": 1.2161768984579575, + "flow_matching_loss": 0.016077405028045177, + "grad_norm": 0.10518195480108261, + "kept_token_count": 572.971875, + "learning_rate": 2.935068077338246e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9915543258190155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008445673529058695, + "mgd_loss_weight": 0.8608453571796417, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.57545776367186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.97121276855466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4180, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.30007838606834414, + "epoch": 1.219086412569101, + "flow_matching_loss": 0.014605535939335824, + "grad_norm": 0.1076461523771286, + "kept_token_count": 572.6671875, + "learning_rate": 2.934585986046481e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9913813591003418, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00861863405443728, + "mgd_loss_weight": 0.8605120182037354, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.2916748046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.2861389160156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4190, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.299549800157547, + "epoch": 1.2219959266802445, + "flow_matching_loss": 0.015617991145700217, + "grad_norm": 0.10080476105213165, + "kept_token_count": 573.0640625, + "learning_rate": 2.93410215159747e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9919960200786591, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008003969909623265, + "mgd_loss_weight": 0.860178679227829, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.4053924560547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.29107360839845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4200, + "valid_token_count": 818.13515625 + }, + { + "actual_token_mask_ratio": 0.3005148500204086, + "epoch": 1.2249054407913877, + "flow_matching_loss": 0.015792144928127527, + "grad_norm": 0.11921565234661102, + "kept_token_count": 572.34375, + "learning_rate": 2.933616574579118e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9918802738189697, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00811973325908184, + "mgd_loss_weight": 0.8598453223705291, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.8248718261719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.2401397705078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4210, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.29979556798934937, + "epoch": 1.2278149549025312, + "flow_matching_loss": 0.01530947107821703, + "grad_norm": 0.12038428336381912, + "kept_token_count": 573.02109375, + "learning_rate": 2.933129255581447e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9918426632881164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008157334290444851, + "mgd_loss_weight": 0.8595119893550873, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.89853515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.0246215820313, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4220, + "valid_token_count": 818.36640625 + }, + { + "actual_token_mask_ratio": 0.3000039398670197, + "epoch": 1.2307244690136747, + "flow_matching_loss": 0.01695067957043648, + "grad_norm": 0.08729539066553116, + "kept_token_count": 572.8671875, + "learning_rate": 2.9326401951965954e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9917899787425994, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008210008079186083, + "mgd_loss_weight": 0.8591786384582519, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.87119140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.307177734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4230, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.2997822642326355, + "epoch": 1.2336339831248182, + "flow_matching_loss": 0.015072569344192743, + "grad_norm": 0.10388662666082382, + "kept_token_count": 572.8328125, + "learning_rate": 2.9321493940188165e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9918682515621186, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008131748298183084, + "mgd_loss_weight": 0.8588452935218811, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.4598358154297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.2265869140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4240, + "valid_token_count": 818.07578125 + }, + { + "actual_token_mask_ratio": 0.3002101719379425, + "epoch": 1.2365434972359617, + "flow_matching_loss": 0.01614850517362356, + "grad_norm": 0.09195312112569809, + "kept_token_count": 572.5265625, + "learning_rate": 2.931656852644482e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9918184995651245, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008181521203368902, + "mgd_loss_weight": 0.8585119545459747, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.3034393310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.3653228759766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4250, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.29994304180145265, + "epoch": 1.239453011347105, + "flow_matching_loss": 0.01569889662787318, + "grad_norm": 0.14198732376098633, + "kept_token_count": 572.87265625, + "learning_rate": 2.9311625716720748e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.991375309228897, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008624691376462578, + "mgd_loss_weight": 0.8581786036491394, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.02796630859376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.6270477294922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4260, + "valid_token_count": 818.32421875 + }, + { + "actual_token_mask_ratio": 0.29960988461971283, + "epoch": 1.2423625254582484, + "flow_matching_loss": 0.016056778468191625, + "grad_norm": 0.09831881523132324, + "kept_token_count": 573.03515625, + "learning_rate": 2.9306665517021927e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9920307219028472, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00796928172931075, + "mgd_loss_weight": 0.8578452646732331, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.0809600830078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.32895812988284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4270, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.3008878409862518, + "epoch": 1.245272039569392, + "flow_matching_loss": 0.015137470886111259, + "grad_norm": 0.11120934039354324, + "kept_token_count": 572.071875, + "learning_rate": 2.9301687933375483e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9917725145816803, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008227484999224544, + "mgd_loss_weight": 0.8575119197368621, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 332.5106903076172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.78049621582034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4280, + "valid_token_count": 818.28359375 + }, + { + "actual_token_mask_ratio": 0.2994273900985718, + "epoch": 1.2481815536805354, + "flow_matching_loss": 0.017478565406054258, + "grad_norm": 0.09561589360237122, + "kept_token_count": 573.2875, + "learning_rate": 2.929669297182963e-05, + "loss": 0.0249, + "mgd_cos_sim": 0.9913068294525147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008693176787346601, + "mgd_loss_weight": 0.8571785688400269, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.5455688476562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.1757110595703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4290, + "valid_token_count": 818.31171875 + }, + { + "actual_token_mask_ratio": 0.30033840537071227, + "epoch": 1.2510910677916787, + "flow_matching_loss": 0.01646695574745536, + "grad_norm": 0.11047369241714478, + "kept_token_count": 572.70859375, + "learning_rate": 2.929168063845373e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9917670786380768, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008232931792736053, + "mgd_loss_weight": 0.8568452298641205, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.82605895996096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.089794921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4300, + "valid_token_count": 818.56015625 + }, + { + "actual_token_mask_ratio": 0.30049885213375094, + "epoch": 1.2540005819028222, + "flow_matching_loss": 0.01617908300831914, + "grad_norm": 0.09339707344770432, + "kept_token_count": 572.45, + "learning_rate": 2.9286650939338236e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9912605583667755, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008739449735730886, + "mgd_loss_weight": 0.8565118849277497, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.8432586669922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.90916137695314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4310, + "valid_token_count": 818.36640625 + }, + { + "actual_token_mask_ratio": 0.30060647130012513, + "epoch": 1.2569100960139656, + "flow_matching_loss": 0.016552020516246557, + "grad_norm": 0.08038387447595596, + "kept_token_count": 572.2625, + "learning_rate": 2.9281603880594706e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.991796612739563, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008203397179022432, + "mgd_loss_weight": 0.8561785399913788, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.2710845947266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.8538024902344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4320, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.30021700263023376, + "epoch": 1.2598196101251091, + "flow_matching_loss": 0.016697438526898624, + "grad_norm": 0.11441706866025925, + "kept_token_count": 572.6578125, + "learning_rate": 2.9276539468355796e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9921050727367401, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007894918415695429, + "mgd_loss_weight": 0.8558451950550079, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.59190979003904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.21734924316405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4330, + "valid_token_count": 818.33984375 + }, + { + "actual_token_mask_ratio": 0.30043439865112304, + "epoch": 1.2627291242362526, + "flow_matching_loss": 0.017440257593989372, + "grad_norm": 0.10992281883955002, + "kept_token_count": 572.3265625, + "learning_rate": 2.9271457708775244e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9920293509960174, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007970663532614709, + "mgd_loss_weight": 0.8555118501186371, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.481396484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.6011566162109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4340, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.30046466588973997, + "epoch": 1.2656386383473959, + "flow_matching_loss": 0.014162879157811403, + "grad_norm": 0.09910757839679718, + "kept_token_count": 572.3375, + "learning_rate": 2.9266358608027863e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9913965463638306, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008603461831808091, + "mgd_loss_weight": 0.8551785051822662, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.3719757080078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.3266662597656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4350, + "valid_token_count": 818.16640625 + }, + { + "actual_token_mask_ratio": 0.29932370185852053, + "epoch": 1.2685481524585394, + "flow_matching_loss": 0.016926167253404855, + "grad_norm": 0.10515027493238449, + "kept_token_count": 573.34375, + "learning_rate": 2.9261242172309543e-05, + "loss": 0.0242, + "mgd_cos_sim": 0.991445130109787, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008554867142811418, + "mgd_loss_weight": 0.8548451662063599, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 336.9550323486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.40843505859374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4360, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.30040108859539033, + "epoch": 1.2714576665696828, + "flow_matching_loss": 0.015366548765450716, + "grad_norm": 0.10529349744319916, + "kept_token_count": 572.67109375, + "learning_rate": 2.925610840783723e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9919495403766632, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008050464931875468, + "mgd_loss_weight": 0.8545118153095246, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.54127502441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.6502227783203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4370, + "valid_token_count": 818.571875 + }, + { + "actual_token_mask_ratio": 0.3000537008047104, + "epoch": 1.2743671806808263, + "flow_matching_loss": 0.015041692461818457, + "grad_norm": 0.09614706039428711, + "kept_token_count": 572.72890625, + "learning_rate": 2.9250957320848946e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9914946854114532, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0085053033195436, + "mgd_loss_weight": 0.8541784703731536, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.6778564453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 364.9630157470703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4380, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.30045188665390016, + "epoch": 1.2772766947919698, + "flow_matching_loss": 0.01621135203167796, + "grad_norm": 0.09784404933452606, + "kept_token_count": 572.39296875, + "learning_rate": 2.924578891760374e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9921029925346374, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007897001458331943, + "mgd_loss_weight": 0.8538451313972473, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.70777893066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.140380859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4390, + "valid_token_count": 818.23046875 + }, + { + "actual_token_mask_ratio": 0.29926826655864713, + "epoch": 1.280186208903113, + "flow_matching_loss": 0.0156733145006001, + "grad_norm": 0.11638092994689941, + "kept_token_count": 573.3625, + "learning_rate": 2.924060320438171e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9915704309940339, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00842957478016615, + "mgd_loss_weight": 0.853511780500412, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.02166442871095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.14252319335935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4400, + "valid_token_count": 818.23671875 + }, + { + "actual_token_mask_ratio": 0.3003320127725601, + "epoch": 1.2830957230142566, + "flow_matching_loss": 0.01607404127717018, + "grad_norm": 0.10662135481834412, + "kept_token_count": 572.75, + "learning_rate": 2.9235400187483978e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.99159677028656, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008403225243091584, + "mgd_loss_weight": 0.8531784415245056, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.8336120605469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.7958038330078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4410, + "valid_token_count": 818.60625 + }, + { + "actual_token_mask_ratio": 0.29981698393821715, + "epoch": 1.2860052371254, + "flow_matching_loss": 0.017287857178598643, + "grad_norm": 0.13293875753879547, + "kept_token_count": 572.94453125, + "learning_rate": 2.9230179873232716e-05, + "loss": 0.0242, + "mgd_cos_sim": 0.9918423771858216, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008157612709328532, + "mgd_loss_weight": 0.8528450965881348, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.92528076171874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.37395629882815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4420, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.29997414648532866, + "epoch": 1.2889147512365435, + "flow_matching_loss": 0.016064136289060116, + "grad_norm": 0.09640010446310043, + "kept_token_count": 572.76015625, + "learning_rate": 2.922494226797108e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9918218493461609, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008178146556019783, + "mgd_loss_weight": 0.8525117456912994, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.4945465087891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.78212890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4430, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.3001607686281204, + "epoch": 1.291824265347687, + "flow_matching_loss": 0.017456461768597366, + "grad_norm": 0.10926943272352219, + "kept_token_count": 572.79453125, + "learning_rate": 2.9219687378063275e-05, + "loss": 0.0246, + "mgd_cos_sim": 0.9916116416454315, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008388348342850804, + "mgd_loss_weight": 0.8521784126758576, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 339.8931060791016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.4381622314453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4440, + "valid_token_count": 818.46484375 + }, + { + "actual_token_mask_ratio": 0.3003416657447815, + "epoch": 1.2947337794588303, + "flow_matching_loss": 0.016756721306592226, + "grad_norm": 0.104018934071064, + "kept_token_count": 572.4703125, + "learning_rate": 2.921441520989447e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9915290713310242, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008470924431458115, + "mgd_loss_weight": 0.8518450558185577, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.6049560546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.05816955566405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4450, + "valid_token_count": 818.2125 + }, + { + "actual_token_mask_ratio": 0.2999386817216873, + "epoch": 1.2976432935699738, + "flow_matching_loss": 0.01692116502672434, + "grad_norm": 0.09008263796567917, + "kept_token_count": 572.82734375, + "learning_rate": 2.9209125769870854e-05, + "loss": 0.0242, + "mgd_cos_sim": 0.9914271056652069, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008572909142822027, + "mgd_loss_weight": 0.8515117168426514, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.9764801025391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.87562255859376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4460, + "valid_token_count": 818.25078125 + }, + { + "actual_token_mask_ratio": 0.30019311606884, + "epoch": 1.3005528076811173, + "flow_matching_loss": 0.017489960044622423, + "grad_norm": 0.11078231036663055, + "kept_token_count": 572.63984375, + "learning_rate": 2.92038190644196e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9914241075515747, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008575893845409155, + "mgd_loss_weight": 0.851178377866745, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.24082336425784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.7608947753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4470, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.3007231891155243, + "epoch": 1.3034623217922607, + "flow_matching_loss": 0.017479053232818842, + "grad_norm": 0.10438009351491928, + "kept_token_count": 572.28203125, + "learning_rate": 2.9198495099988843e-05, + "loss": 0.0248, + "mgd_cos_sim": 0.9913480937480926, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008651898242533207, + "mgd_loss_weight": 0.8508450210094451, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 346.8363098144531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.3985107421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4480, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.29939168989658355, + "epoch": 1.3063718359034042, + "flow_matching_loss": 0.01648947997018695, + "grad_norm": 0.08264026045799255, + "kept_token_count": 573.22109375, + "learning_rate": 2.9193153883047712e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9918709576129914, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008129051327705384, + "mgd_loss_weight": 0.8505116879940033, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.55986328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.9945037841797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4490, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.3006296157836914, + "epoch": 1.3092813500145475, + "flow_matching_loss": 0.01666986122727394, + "grad_norm": 0.09068331867456436, + "kept_token_count": 572.3171875, + "learning_rate": 2.9187795420086287e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9919551849365235, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008044808823615312, + "mgd_loss_weight": 0.850178337097168, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.77313537597655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.3815185546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4500, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.3003106385469437, + "epoch": 1.312190864125691, + "flow_matching_loss": 0.018458660505712033, + "grad_norm": 0.11457667499780655, + "kept_token_count": 572.54765625, + "learning_rate": 2.9182419717615602e-05, + "loss": 0.0257, + "mgd_cos_sim": 0.9915231943130494, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008476798888295889, + "mgd_loss_weight": 0.8498449921607971, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.5430419921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.29893493652344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4510, + "valid_token_count": 818.28984375 + }, + { + "actual_token_mask_ratio": 0.29951503574848176, + "epoch": 1.3151003782368345, + "flow_matching_loss": 0.016521667316555976, + "grad_norm": 0.11976704001426697, + "kept_token_count": 573.26484375, + "learning_rate": 2.917702678216765e-05, + "loss": 0.024, + "mgd_cos_sim": 0.991176289319992, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008823734009638429, + "mgd_loss_weight": 0.8495116531848907, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.38042602539065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.49217529296874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4520, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.29993986785411836, + "epoch": 1.318009892347978, + "flow_matching_loss": 0.015677786618471145, + "grad_norm": 0.09594015777111053, + "kept_token_count": 573.01328125, + "learning_rate": 2.9171616620295347e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9914791345596313, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00852087801322341, + "mgd_loss_weight": 0.8491783022880555, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.2467498779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.5141265869141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4530, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.30006893277168273, + "epoch": 1.3209194064591214, + "flow_matching_loss": 0.0166276010684669, + "grad_norm": 0.09404192119836807, + "kept_token_count": 572.8515625, + "learning_rate": 2.9166189238572554e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9915809631347656, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008419026946648955, + "mgd_loss_weight": 0.8488449633121491, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.0642547607422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.3149108886719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4540, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.30004703104496, + "epoch": 1.3238289205702647, + "flow_matching_loss": 0.015931222308427095, + "grad_norm": 0.09215375036001205, + "kept_token_count": 572.82734375, + "learning_rate": 2.9160744643594052e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.991402930021286, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008597056847065687, + "mgd_loss_weight": 0.8485116183757782, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.405712890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.1066436767578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4550, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.2999539762735367, + "epoch": 1.3267384346814082, + "flow_matching_loss": 0.015401484444737434, + "grad_norm": 0.11217658221721649, + "kept_token_count": 572.9015625, + "learning_rate": 2.915528284197554e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9914677679538727, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008532238891348243, + "mgd_loss_weight": 0.8481782674789429, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.2930938720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.7658294677734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4560, + "valid_token_count": 818.37578125 + }, + { + "actual_token_mask_ratio": 0.30039499402046205, + "epoch": 1.3296479487925517, + "flow_matching_loss": 0.015908195823431014, + "grad_norm": 0.14405860006809235, + "kept_token_count": 572.48515625, + "learning_rate": 2.914980384035362e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.991530841588974, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00846917568705976, + "mgd_loss_weight": 0.8478449285030365, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.6498687744141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.06243896484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4570, + "valid_token_count": 818.29453125 + }, + { + "actual_token_mask_ratio": 0.30025544166564944, + "epoch": 1.3325574629036951, + "flow_matching_loss": 0.016312846448272467, + "grad_norm": 0.08844190090894699, + "kept_token_count": 572.584375, + "learning_rate": 2.9144307645385787e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9909551203250885, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00904486682265997, + "mgd_loss_weight": 0.8475115835666657, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.22731018066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.3367492675781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4580, + "valid_token_count": 818.27734375 + }, + { + "actual_token_mask_ratio": 0.29907236993312836, + "epoch": 1.3354669770148386, + "flow_matching_loss": 0.015952169615775346, + "grad_norm": 0.10554333031177521, + "kept_token_count": 573.5171875, + "learning_rate": 2.9138794263750454e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9913746476173401, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008625361043959856, + "mgd_loss_weight": 0.8471782386302948, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.7332000732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.29804077148435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4590, + "valid_token_count": 818.22265625 + }, + { + "actual_token_mask_ratio": 0.29982903599739075, + "epoch": 1.338376491125982, + "flow_matching_loss": 0.016592518985271455, + "grad_norm": 0.0922757014632225, + "kept_token_count": 572.8984375, + "learning_rate": 2.913326370214688e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9914650678634643, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008534939726814628, + "mgd_loss_weight": 0.8468448936939239, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.52161560058596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.4252532958984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4600, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.29954656064510343, + "epoch": 1.3412860052371254, + "flow_matching_loss": 0.015634861309081316, + "grad_norm": 0.09823169559240341, + "kept_token_count": 573.0765625, + "learning_rate": 2.9127715967295234e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9915580034255982, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008441989915445447, + "mgd_loss_weight": 0.8465115487575531, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.21544494628904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.37233581542966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4610, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.3001377731561661, + "epoch": 1.3441955193482689, + "flow_matching_loss": 0.01581954499706626, + "grad_norm": 0.10484209656715393, + "kept_token_count": 572.74140625, + "learning_rate": 2.912215106593653e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9911292254924774, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00887078233063221, + "mgd_loss_weight": 0.8461782038211823, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.55260314941404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.4111663818359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4620, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.3001251667737961, + "epoch": 1.3471050334594123, + "flow_matching_loss": 0.016360970679670572, + "grad_norm": 0.09163440763950348, + "kept_token_count": 572.72578125, + "learning_rate": 2.911656900483265e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9919578671455384, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008042134018614889, + "mgd_loss_weight": 0.8458448648452759, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.4822723388672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.3158386230469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4630, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.3004435390233994, + "epoch": 1.3500145475705558, + "flow_matching_loss": 0.017205691523849966, + "grad_norm": 0.12039762735366821, + "kept_token_count": 572.515625, + "learning_rate": 2.9110969790766322e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9917996346950531, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00820036414079368, + "mgd_loss_weight": 0.8455115139484406, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.6780578613281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.61614685058595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4640, + "valid_token_count": 818.40234375 + }, + { + "actual_token_mask_ratio": 0.30032513439655306, + "epoch": 1.352924061681699, + "flow_matching_loss": 0.016330127883702517, + "grad_norm": 0.07495306432247162, + "kept_token_count": 572.60546875, + "learning_rate": 2.910535343054112e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.992286640405655, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007713365508243442, + "mgd_loss_weight": 0.8451781690120697, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.1857971191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.1128753662109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4650, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.30007446706295016, + "epoch": 1.3558335757928426, + "flow_matching_loss": 0.015036561246961355, + "grad_norm": 0.10057288408279419, + "kept_token_count": 572.709375, + "learning_rate": 2.909971993098145e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9915713012218476, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008428691141307353, + "mgd_loss_weight": 0.8448448359966279, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.1216552734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.1125213623047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4660, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.299365359544754, + "epoch": 1.358743089903986, + "flow_matching_loss": 0.017852793913334608, + "grad_norm": 0.09750516712665558, + "kept_token_count": 573.28046875, + "learning_rate": 2.9094069298932552e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.991937792301178, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008062197593972087, + "mgd_loss_weight": 0.844511479139328, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.6773712158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.78404846191404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4670, + "valid_token_count": 818.23203125 + }, + { + "actual_token_mask_ratio": 0.30028619766235354, + "epoch": 1.3616526040151296, + "flow_matching_loss": 0.01627071527764201, + "grad_norm": 0.10650438070297241, + "kept_token_count": 572.6265625, + "learning_rate": 2.908840154126047e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9918422698974609, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008157754177227616, + "mgd_loss_weight": 0.8441781401634216, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.3018035888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.38560180664064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4680, + "valid_token_count": 818.37734375 + }, + { + "actual_token_mask_ratio": 0.29999426901340487, + "epoch": 1.364562118126273, + "flow_matching_loss": 0.014001255854964256, + "grad_norm": 0.09794788807630539, + "kept_token_count": 572.85625, + "learning_rate": 2.9082716664852063e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9918824255466461, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008117573941126466, + "mgd_loss_weight": 0.8438448011875153, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.14552307128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.3421569824219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4690, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.30051682591438295, + "epoch": 1.3674716322374163, + "flow_matching_loss": 0.016184072475880384, + "grad_norm": 0.0951603576540947, + "kept_token_count": 572.41796875, + "learning_rate": 2.9077014676615e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.991272383928299, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008727617049589752, + "mgd_loss_weight": 0.8435114443302154, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.72880859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.800146484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4700, + "valid_token_count": 818.34296875 + }, + { + "actual_token_mask_ratio": 0.2999627351760864, + "epoch": 1.3703811463485598, + "flow_matching_loss": 0.015422472637146712, + "grad_norm": 0.09481345862150192, + "kept_token_count": 572.9328125, + "learning_rate": 2.9071295583477724e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9919776082038879, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008022386254742742, + "mgd_loss_weight": 0.8431781113147736, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.04038696289064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.36158752441406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4710, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.30038041472434995, + "epoch": 1.3732906604597033, + "flow_matching_loss": 0.01485472172498703, + "grad_norm": 0.09098589420318604, + "kept_token_count": 572.575, + "learning_rate": 2.906555939238948e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9918159365653991, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008184067811816931, + "mgd_loss_weight": 0.8428447604179382, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 345.3492858886719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.7216552734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4720, + "valid_token_count": 818.41328125 + }, + { + "actual_token_mask_ratio": 0.29979179203510287, + "epoch": 1.3762001745708465, + "flow_matching_loss": 0.016517067700624465, + "grad_norm": 0.08654454350471497, + "kept_token_count": 572.9109375, + "learning_rate": 2.905980611032028e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9918359398841858, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008164048055186867, + "mgd_loss_weight": 0.8425114154815674, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.00087890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.29709777832034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4730, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.3000951141119003, + "epoch": 1.3791096886819902, + "flow_matching_loss": 0.01584893148392439, + "grad_norm": 0.09965462982654572, + "kept_token_count": 572.77265625, + "learning_rate": 2.9054035744260904e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9914739549160003, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008526041870936751, + "mgd_loss_weight": 0.842178076505661, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.8905548095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.7402008056641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4740, + "valid_token_count": 818.35625 + }, + { + "actual_token_mask_ratio": 0.3003861516714096, + "epoch": 1.3820192027931335, + "flow_matching_loss": 0.014459299668669701, + "grad_norm": 0.10209567844867706, + "kept_token_count": 572.3328125, + "learning_rate": 2.9048248301222885e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9918903231620788, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008109676418825984, + "mgd_loss_weight": 0.8418447256088257, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.9880828857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.4157745361328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4750, + "valid_token_count": 818.07421875 + }, + { + "actual_token_mask_ratio": 0.3006195515394211, + "epoch": 1.384928716904277, + "flow_matching_loss": 0.016246674861758948, + "grad_norm": 0.07948989421129227, + "kept_token_count": 572.23359375, + "learning_rate": 2.904244378823852e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9916397750377655, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008360223798081278, + "mgd_loss_weight": 0.8415113866329194, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.1761932373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.9263854980469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4760, + "valid_token_count": 818.20234375 + }, + { + "actual_token_mask_ratio": 0.300270214676857, + "epoch": 1.3878382310154205, + "flow_matching_loss": 0.014480166416615248, + "grad_norm": 0.11460284888744354, + "kept_token_count": 572.5546875, + "learning_rate": 2.9036622212360836e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9912632405757904, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008736758306622506, + "mgd_loss_weight": 0.8411780416965484, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 341.6786224365234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.61131286621094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4770, + "valid_token_count": 818.24765625 + }, + { + "actual_token_mask_ratio": 0.29942674934864044, + "epoch": 1.3907477451265637, + "flow_matching_loss": 0.015011664293706418, + "grad_norm": 0.10136102885007858, + "kept_token_count": 573.36484375, + "learning_rate": 2.9030783580663596e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9923033833503723, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076966154854744675, + "mgd_loss_weight": 0.8408446907997131, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.2680908203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.56568908691406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4780, + "valid_token_count": 818.42265625 + }, + { + "actual_token_mask_ratio": 0.3007355123758316, + "epoch": 1.3936572592377072, + "flow_matching_loss": 0.015623770747333765, + "grad_norm": 0.10162326693534851, + "kept_token_count": 572.17265625, + "learning_rate": 2.902492790024129e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9914912819862366, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008508724020794034, + "mgd_loss_weight": 0.8405113518238068, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 353.38526306152346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.7185913085938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4790, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.29958942234516145, + "epoch": 1.3965667733488507, + "flow_matching_loss": 0.016695829574018717, + "grad_norm": 0.10403508692979813, + "kept_token_count": 573.175, + "learning_rate": 2.901905517820912e-05, + "loss": 0.0239, + "mgd_cos_sim": 0.9914340078830719, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008566005621105433, + "mgd_loss_weight": 0.8401780068874359, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.7865936279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.74598388671876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4800, + "valid_token_count": 818.33671875 + }, + { + "actual_token_mask_ratio": 0.2993862360715866, + "epoch": 1.3994762874599942, + "flow_matching_loss": 0.01472566369920969, + "grad_norm": 0.09188200533390045, + "kept_token_count": 573.18125, + "learning_rate": 2.9013165421703e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.991711312532425, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008288669679313899, + "mgd_loss_weight": 0.8398446619510651, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.6488830566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.5058654785156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4810, + "valid_token_count": 818.1125 + }, + { + "actual_token_mask_ratio": 0.2998003035783768, + "epoch": 1.4023858015711377, + "flow_matching_loss": 0.015956289414316416, + "grad_norm": 0.08694033324718475, + "kept_token_count": 572.9765625, + "learning_rate": 2.9007258637879534e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9918092906475067, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008190700225532055, + "mgd_loss_weight": 0.8395113170146942, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.32212219238284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.6137359619141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4820, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.3007574647665024, + "epoch": 1.405295315682281, + "flow_matching_loss": 0.015654955897480248, + "grad_norm": 0.08658101409673691, + "kept_token_count": 572.28046875, + "learning_rate": 2.9001334833916025e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9917894423007965, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008210557978600264, + "mgd_loss_weight": 0.8391779720783233, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 329.03184509277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.2879974365234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4830, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.3009746581315994, + "epoch": 1.4082048297934244, + "flow_matching_loss": 0.014883141685277223, + "grad_norm": 0.08786710351705551, + "kept_token_count": 572.03046875, + "learning_rate": 2.899539401701046e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9920153558254242, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007984639704227447, + "mgd_loss_weight": 0.8388446271419525, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 367.0112060546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.22267150878906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4840, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.3001712322235107, + "epoch": 1.411114343904568, + "flow_matching_loss": 0.016068021953105926, + "grad_norm": 0.12252823263406754, + "kept_token_count": 572.603125, + "learning_rate": 2.8989436194381474e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9915201485157012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008479850366711617, + "mgd_loss_weight": 0.8385112881660461, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.4171630859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.4524291992187, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4850, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.2990299344062805, + "epoch": 1.4140238580157114, + "flow_matching_loss": 0.01571578122675419, + "grad_norm": 0.09150637686252594, + "kept_token_count": 573.58984375, + "learning_rate": 2.8983461373268404e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.991788113117218, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008211905090138315, + "mgd_loss_weight": 0.8381779372692109, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.34226989746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.1628112792969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4860, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.3000601053237915, + "epoch": 1.4169333721268549, + "flow_matching_loss": 0.016482723969966175, + "grad_norm": 0.10318803042173386, + "kept_token_count": 572.72890625, + "learning_rate": 2.8977469560931205e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9918784677982331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008121543005108834, + "mgd_loss_weight": 0.8378445923328399, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.85923461914064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.72507934570314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4870, + "valid_token_count": 818.25546875 + }, + { + "actual_token_mask_ratio": 0.30025874376296996, + "epoch": 1.4198428862379981, + "flow_matching_loss": 0.015074291825294494, + "grad_norm": 0.0911121815443039, + "kept_token_count": 572.53828125, + "learning_rate": 2.8971460764650502e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9914445042610168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008555505564436317, + "mgd_loss_weight": 0.8375112533569335, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.5512329101563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.18272094726564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4880, + "valid_token_count": 818.21640625 + }, + { + "actual_token_mask_ratio": 0.2996043711900711, + "epoch": 1.4227524003491416, + "flow_matching_loss": 0.015838383138179778, + "grad_norm": 0.10524741560220718, + "kept_token_count": 573.13359375, + "learning_rate": 2.8965434991727545e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9919386208057404, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008061389951035381, + "mgd_loss_weight": 0.8371779024600983, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.3305725097656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.10633544921876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4890, + "valid_token_count": 818.30078125 + }, + { + "actual_token_mask_ratio": 0.3002219468355179, + "epoch": 1.4256619144602851, + "flow_matching_loss": 0.014864584803581238, + "grad_norm": 0.11682091653347015, + "kept_token_count": 572.60234375, + "learning_rate": 2.895939224948422e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9914724349975585, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008527545584365725, + "mgd_loss_weight": 0.8368445634841919, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.1597869873047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.7167724609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4900, + "valid_token_count": 818.26484375 + }, + { + "actual_token_mask_ratio": 0.29980299770832064, + "epoch": 1.4285714285714286, + "flow_matching_loss": 0.013853665441274643, + "grad_norm": 0.0760420486330986, + "kept_token_count": 572.98828125, + "learning_rate": 2.895333254526303e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9919245839118958, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008075432386249304, + "mgd_loss_weight": 0.8365112185478211, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.835498046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.8830902099609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4910, + "valid_token_count": 818.32578125 + }, + { + "actual_token_mask_ratio": 0.29994221925735476, + "epoch": 1.431480942682572, + "flow_matching_loss": 0.01614925069734454, + "grad_norm": 0.09693972766399384, + "kept_token_count": 572.70390625, + "learning_rate": 2.8947255886427084e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9917764067649841, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008223596075549722, + "mgd_loss_weight": 0.8361778676509857, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 344.6468841552734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.9166534423828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4920, + "valid_token_count": 818.08203125 + }, + { + "actual_token_mask_ratio": 0.30073699057102204, + "epoch": 1.4343904567937154, + "flow_matching_loss": 0.015889839828014375, + "grad_norm": 0.0755230113863945, + "kept_token_count": 572.00078125, + "learning_rate": 2.8941162280360095e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9922084510326385, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007791556511074304, + "mgd_loss_weight": 0.8358445346355439, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.020703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.9036468505859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4930, + "valid_token_count": 818.00390625 + }, + { + "actual_token_mask_ratio": 0.3003161132335663, + "epoch": 1.4372999709048588, + "flow_matching_loss": 0.01664124121889472, + "grad_norm": 0.08450327813625336, + "kept_token_count": 572.58984375, + "learning_rate": 2.8935051734466368e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.991804015636444, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008195980731397868, + "mgd_loss_weight": 0.835511177778244, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.1747741699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.76499938964844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4940, + "valid_token_count": 818.35703125 + }, + { + "actual_token_mask_ratio": 0.2999446511268616, + "epoch": 1.4402094850160023, + "flow_matching_loss": 0.016584957018494606, + "grad_norm": 0.08786418288946152, + "kept_token_count": 572.903125, + "learning_rate": 2.8928924256170792e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9918080806732178, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00819190931506455, + "mgd_loss_weight": 0.8351778388023376, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 349.33656005859376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.12925415039064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4950, + "valid_token_count": 818.36640625 + }, + { + "actual_token_mask_ratio": 0.30021188855171205, + "epoch": 1.4431189991271458, + "flow_matching_loss": 0.01638560416176915, + "grad_norm": 0.09536168724298477, + "kept_token_count": 572.59375, + "learning_rate": 2.8922779852918836e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9919970810413361, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008002929156646132, + "mgd_loss_weight": 0.8348444998264313, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 343.60179443359374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.8952117919922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4960, + "valid_token_count": 818.23984375 + }, + { + "actual_token_mask_ratio": 0.30052370429039, + "epoch": 1.4460285132382893, + "flow_matching_loss": 0.017743773013353347, + "grad_norm": 0.09624242037534714, + "kept_token_count": 572.38515625, + "learning_rate": 2.8916618532176526e-05, + "loss": 0.0245, + "mgd_cos_sim": 0.9919179975986481, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008082003565505147, + "mgd_loss_weight": 0.8345111429691314, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.87083435058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.536474609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4970, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.30041842758655546, + "epoch": 1.4489380273494326, + "flow_matching_loss": 0.015390446875244379, + "grad_norm": 0.09616567939519882, + "kept_token_count": 572.43828125, + "learning_rate": 2.891044030143045e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9919466316699982, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008053361251950265, + "mgd_loss_weight": 0.8341778099536896, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.9164215087891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.99464721679686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4980, + "valid_token_count": 818.25390625 + }, + { + "actual_token_mask_ratio": 0.2993677705526352, + "epoch": 1.451847541460576, + "flow_matching_loss": 0.01455824263393879, + "grad_norm": 0.09827245026826859, + "kept_token_count": 573.48125, + "learning_rate": 2.890424516818774e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.991442596912384, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008557404391467571, + "mgd_loss_weight": 0.8338444590568542, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.3681976318359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.98675231933595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 4990, + "valid_token_count": 818.5171875 + }, + { + "actual_token_mask_ratio": 0.3006045579910278, + "epoch": 1.4547570555717195, + "flow_matching_loss": 0.01483602011576295, + "grad_norm": 0.09443396329879761, + "kept_token_count": 572.3140625, + "learning_rate": 2.8898033139976057e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9919171690940857, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008082834770902991, + "mgd_loss_weight": 0.8335111141204834, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.2668792724609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.7820739746094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5000, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.3001321077346802, + "epoch": 1.457666569682863, + "flow_matching_loss": 0.015475438442081213, + "grad_norm": 0.09716179221868515, + "kept_token_count": 572.75546875, + "learning_rate": 2.889180422434362e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9915027022361755, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00849730516783893, + "mgd_loss_weight": 0.833177775144577, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 334.4241027832031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.5675048828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5010, + "valid_token_count": 818.37421875 + }, + { + "actual_token_mask_ratio": 0.30027263462543485, + "epoch": 1.4605760837940065, + "flow_matching_loss": 0.014301463030278683, + "grad_norm": 0.08835688978433609, + "kept_token_count": 572.61015625, + "learning_rate": 2.8885558428859136e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.992099666595459, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007900340342894196, + "mgd_loss_weight": 0.8328444242477417, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.82402954101565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.06728515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5020, + "valid_token_count": 818.33203125 + }, + { + "actual_token_mask_ratio": 0.3002149909734726, + "epoch": 1.4634855979051498, + "flow_matching_loss": 0.014730309415608645, + "grad_norm": 0.0908481776714325, + "kept_token_count": 572.62421875, + "learning_rate": 2.8879295761111835e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9919884383678437, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008011570945382119, + "mgd_loss_weight": 0.8325110852718354, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.6530731201172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.4641418457031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5030, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.3002063363790512, + "epoch": 1.4663951120162932, + "flow_matching_loss": 0.01587672485038638, + "grad_norm": 0.10975483804941177, + "kept_token_count": 572.58984375, + "learning_rate": 2.8873016228711452e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9915912389755249, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008408758835867048, + "mgd_loss_weight": 0.8321777403354644, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.2515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.1302215576172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5040, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.2995319485664368, + "epoch": 1.4693046261274367, + "flow_matching_loss": 0.01666742879897356, + "grad_norm": 0.09912902116775513, + "kept_token_count": 573.03984375, + "learning_rate": 2.8866719839288215e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9918668210506439, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008133172988891602, + "mgd_loss_weight": 0.8318443894386292, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.91717834472655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.5715393066406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5050, + "valid_token_count": 818.078125 + }, + { + "actual_token_mask_ratio": 0.29996860325336455, + "epoch": 1.4722141402385802, + "flow_matching_loss": 0.015737443510442973, + "grad_norm": 0.13166101276874542, + "kept_token_count": 572.890625, + "learning_rate": 2.886040660049283e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9918400168418884, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00815997626632452, + "mgd_loss_weight": 0.8315110504627228, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.9231323242187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.1886932373047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5060, + "valid_token_count": 818.37890625 + }, + { + "actual_token_mask_ratio": 0.3000418245792389, + "epoch": 1.4751236543497237, + "flow_matching_loss": 0.013766432367265224, + "grad_norm": 0.09342740476131439, + "kept_token_count": 572.74140625, + "learning_rate": 2.8854076519996467e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9920095562934875, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00799044333398342, + "mgd_loss_weight": 0.831177705526352, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.69407653808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.19069519042966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5070, + "valid_token_count": 818.24765625 + }, + { + "actual_token_mask_ratio": 0.30016147494316103, + "epoch": 1.478033168460867, + "flow_matching_loss": 0.01325841285288334, + "grad_norm": 0.09341515600681305, + "kept_token_count": 572.77734375, + "learning_rate": 2.884772960549078e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9920229375362396, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007977061299607157, + "mgd_loss_weight": 0.8308443605899811, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.4153594970703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.69806213378905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5080, + "valid_token_count": 818.43984375 + }, + { + "actual_token_mask_ratio": 0.2999963670969009, + "epoch": 1.4809426825720104, + "flow_matching_loss": 0.01735846446827054, + "grad_norm": 0.11593066900968552, + "kept_token_count": 572.87734375, + "learning_rate": 2.8841365864687867e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.991637772321701, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008362226514145732, + "mgd_loss_weight": 0.8305110156536102, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.4247650146484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.12644958496094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5090, + "valid_token_count": 818.39296875 + }, + { + "actual_token_mask_ratio": 0.29977557957172396, + "epoch": 1.483852196683154, + "flow_matching_loss": 0.014959682244807482, + "grad_norm": 0.10555325448513031, + "kept_token_count": 572.88046875, + "learning_rate": 2.8834985305320276e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9916288554668427, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008371125860139727, + "mgd_loss_weight": 0.8301776766777038, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.46263427734374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.27955322265626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5100, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.29939577281475066, + "epoch": 1.4867617107942974, + "flow_matching_loss": 0.015061493590474128, + "grad_norm": 0.10022059828042984, + "kept_token_count": 573.415625, + "learning_rate": 2.882858793514098e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9921895384788513, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0078104648273438215, + "mgd_loss_weight": 0.8298443257808685, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.434033203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.63287353515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5110, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.300753453373909, + "epoch": 1.489671224905441, + "flow_matching_loss": 0.015196450892835856, + "grad_norm": 0.0993444174528122, + "kept_token_count": 572.1390625, + "learning_rate": 2.8822173761923398e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9916120827198028, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008387915464118123, + "mgd_loss_weight": 0.8295109868049622, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.6493835449219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.3381408691406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5120, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.2999865859746933, + "epoch": 1.4925807390165842, + "flow_matching_loss": 0.014810065645724535, + "grad_norm": 0.10539745539426804, + "kept_token_count": 572.79140625, + "learning_rate": 2.8815742793461343e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9918217360973358, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008178254636004568, + "mgd_loss_weight": 0.8291776418685913, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.3280364990234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.79664306640626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5130, + "valid_token_count": 818.25703125 + }, + { + "actual_token_mask_ratio": 0.29950074255466463, + "epoch": 1.4954902531277277, + "flow_matching_loss": 0.014841240830719471, + "grad_norm": 0.09318584203720093, + "kept_token_count": 572.97109375, + "learning_rate": 2.880929503756906e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9919988572597503, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008001143299043178, + "mgd_loss_weight": 0.8288442909717559, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.39490966796876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.78045654296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5140, + "valid_token_count": 817.94140625 + }, + { + "actual_token_mask_ratio": 0.2998477965593338, + "epoch": 1.4983997672388711, + "flow_matching_loss": 0.016051299590617418, + "grad_norm": 0.10882902890443802, + "kept_token_count": 572.8796875, + "learning_rate": 2.880283050208117e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9918446958065033, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008155314438045024, + "mgd_loss_weight": 0.8285109579563141, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.99639587402345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.7539001464844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5150, + "valid_token_count": 818.22109375 + }, + { + "actual_token_mask_ratio": 0.3000660210847855, + "epoch": 1.5013092813500144, + "flow_matching_loss": 0.014855440892279147, + "grad_norm": 0.07975371927022934, + "kept_token_count": 572.63046875, + "learning_rate": 2.8796349194852704e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9916917860507966, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008308218419551849, + "mgd_loss_weight": 0.8281776010990143, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.8173431396484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.4453857421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5160, + "valid_token_count": 818.12109375 + }, + { + "actual_token_mask_ratio": 0.3001578152179718, + "epoch": 1.504218795461158, + "flow_matching_loss": 0.015867028012871744, + "grad_norm": 0.08787348121404648, + "kept_token_count": 572.5796875, + "learning_rate": 2.8789851123759052e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9916791617870331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008320845384150743, + "mgd_loss_weight": 0.8278442621231079, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.0986694335937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.4196075439453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5170, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.30004648566246034, + "epoch": 1.5071283095723014, + "flow_matching_loss": 0.016567638516426085, + "grad_norm": 0.09675855934619904, + "kept_token_count": 572.77734375, + "learning_rate": 2.8783336296695985e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9913330137729645, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008666979288682342, + "mgd_loss_weight": 0.8275109231472015, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.8558807373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.7579315185547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5180, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.30028020441532133, + "epoch": 1.5100378236834449, + "flow_matching_loss": 0.014653340261429549, + "grad_norm": 0.11790139973163605, + "kept_token_count": 572.721875, + "learning_rate": 2.877680472157964e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9916657626628875, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008334236964583398, + "mgd_loss_weight": 0.8271775662899017, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.97398681640624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.2773712158203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5190, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.30040809214115144, + "epoch": 1.5129473377945883, + "flow_matching_loss": 0.017173528764396907, + "grad_norm": 0.08621352165937424, + "kept_token_count": 572.625, + "learning_rate": 2.8770256406346483e-05, + "loss": 0.0242, + "mgd_cos_sim": 0.9914688110351563, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008531184541061521, + "mgd_loss_weight": 0.8268442332744599, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.8022827148437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.4678100585937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5200, + "valid_token_count": 818.5125 + }, + { + "actual_token_mask_ratio": 0.2998228967189789, + "epoch": 1.5158568519057316, + "flow_matching_loss": 0.01498257154598832, + "grad_norm": 0.08009476214647293, + "kept_token_count": 572.88046875, + "learning_rate": 2.8763691358953348e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9916891396045685, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00831084162928164, + "mgd_loss_weight": 0.8265108823776245, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.1996276855469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.27329406738284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5210, + "valid_token_count": 818.19609375 + }, + { + "actual_token_mask_ratio": 0.30060078501701354, + "epoch": 1.5187663660168753, + "flow_matching_loss": 0.014312395825982094, + "grad_norm": 0.09300066530704498, + "kept_token_count": 572.33671875, + "learning_rate": 2.8757109587377377e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9919905960559845, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008009405992925166, + "mgd_loss_weight": 0.8261775374412537, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.8031005859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.04596557617185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5220, + "valid_token_count": 818.33515625 + }, + { + "actual_token_mask_ratio": 0.300412717461586, + "epoch": 1.5216758801280186, + "flow_matching_loss": 0.014619526825845242, + "grad_norm": 0.09216314554214478, + "kept_token_count": 572.29921875, + "learning_rate": 2.8750511099616047e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9921617984771729, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007838195888325572, + "mgd_loss_weight": 0.8258441984653473, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 351.71539001464845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.3212097167969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5230, + "valid_token_count": 818.05 + }, + { + "actual_token_mask_ratio": 0.30007392168045044, + "epoch": 1.524585394239162, + "flow_matching_loss": 0.014889163244515658, + "grad_norm": 0.0915384590625763, + "kept_token_count": 572.88984375, + "learning_rate": 2.8743895903687143e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.992018848657608, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007981147663667799, + "mgd_loss_weight": 0.825510847568512, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.13433532714845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 365.4133361816406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5240, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.2995899736881256, + "epoch": 1.5274949083503055, + "flow_matching_loss": 0.016760460380464793, + "grad_norm": 0.1037880927324295, + "kept_token_count": 573.04453125, + "learning_rate": 2.8737264007628753e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9915427684783935, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008457232499495149, + "mgd_loss_weight": 0.8251775085926056, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.3643798828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.59288635253904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5250, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.3005741715431213, + "epoch": 1.5304044224614488, + "flow_matching_loss": 0.014176778495311737, + "grad_norm": 0.10770857334136963, + "kept_token_count": 572.315625, + "learning_rate": 2.8730615419499247e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9920811712741852, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007918832963332534, + "mgd_loss_weight": 0.8248441636562347, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.4516326904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.06048278808595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5260, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.30007432103157045, + "epoch": 1.5333139365725925, + "flow_matching_loss": 0.017511685192584992, + "grad_norm": 0.10507527738809586, + "kept_token_count": 572.68515625, + "learning_rate": 2.8723950147377303e-05, + "loss": 0.0244, + "mgd_cos_sim": 0.9916463911533355, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0083536006975919, + "mgd_loss_weight": 0.8245108127593994, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.36302795410154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.48614501953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5270, + "valid_token_count": 818.21171875 + }, + { + "actual_token_mask_ratio": 0.2998892843723297, + "epoch": 1.5362234506837358, + "flow_matching_loss": 0.016390319634228944, + "grad_norm": 0.11360436677932739, + "kept_token_count": 572.69296875, + "learning_rate": 2.871726819936184e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.9915604829788208, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008439528010785579, + "mgd_loss_weight": 0.824177473783493, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.30006713867186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 366.4000183105469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5280, + "valid_token_count": 818.00078125 + }, + { + "actual_token_mask_ratio": 0.30040355026721954, + "epoch": 1.5391329647948793, + "flow_matching_loss": 0.015152146108448505, + "grad_norm": 0.11178037524223328, + "kept_token_count": 572.40234375, + "learning_rate": 2.8710569583572068e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9915054917335511, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008494511293247343, + "mgd_loss_weight": 0.8238441288471222, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.3193603515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.5888275146484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5290, + "valid_token_count": 818.18828125 + }, + { + "actual_token_mask_ratio": 0.3001609772443771, + "epoch": 1.5420424789060228, + "flow_matching_loss": 0.014252798911184073, + "grad_norm": 0.07738452404737473, + "kept_token_count": 572.76328125, + "learning_rate": 2.870385430814742e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9913364231586457, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008663578890264034, + "mgd_loss_weight": 0.8235107839107514, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.4028656005859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.2500274658203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5300, + "valid_token_count": 818.42265625 + }, + { + "actual_token_mask_ratio": 0.29957720041275027, + "epoch": 1.544951993017166, + "flow_matching_loss": 0.015759804751724005, + "grad_norm": 0.08879406750202179, + "kept_token_count": 573.13515625, + "learning_rate": 2.8697122381247605e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9919396877288819, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008060309244319797, + "mgd_loss_weight": 0.8231774389743804, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.05653991699216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.4473937988281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5310, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.30010317265987396, + "epoch": 1.5478615071283097, + "flow_matching_loss": 0.01735334387049079, + "grad_norm": 0.10240397602319717, + "kept_token_count": 572.8296875, + "learning_rate": 2.8690373811052545e-05, + "loss": 0.0243, + "mgd_cos_sim": 0.9915269136428833, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008473093621432781, + "mgd_loss_weight": 0.8228440940380096, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 347.0782135009766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.6465270996094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5320, + "valid_token_count": 818.44609375 + }, + { + "actual_token_mask_ratio": 0.3004722982645035, + "epoch": 1.550771021239453, + "flow_matching_loss": 0.014492020756006242, + "grad_norm": 0.09347990900278091, + "kept_token_count": 572.36015625, + "learning_rate": 2.8683608605762378e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9920488774776459, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00795112019404769, + "mgd_loss_weight": 0.8225107491016388, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.6115264892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.0578948974609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5330, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.3006587505340576, + "epoch": 1.5536805353505965, + "flow_matching_loss": 0.016435702610760928, + "grad_norm": 0.10973507165908813, + "kept_token_count": 572.19296875, + "learning_rate": 2.8676826773597485e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9917375206947326, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008262482192367315, + "mgd_loss_weight": 0.8221774101257324, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 352.1247192382813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.3892059326172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5340, + "valid_token_count": 818.19296875 + }, + { + "actual_token_mask_ratio": 0.30032438337802886, + "epoch": 1.55659004946174, + "flow_matching_loss": 0.016680600680410863, + "grad_norm": 0.1023983582854271, + "kept_token_count": 572.6203125, + "learning_rate": 2.8670028322798413e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9915718197822571, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008428179239854217, + "mgd_loss_weight": 0.8218440592288971, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.4485565185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.9681671142578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5350, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.29952513575553896, + "epoch": 1.5594995635728832, + "flow_matching_loss": 0.015487141534686088, + "grad_norm": 0.09868326783180237, + "kept_token_count": 572.99296875, + "learning_rate": 2.8663213261625932e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9919221043586731, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008077887585386634, + "mgd_loss_weight": 0.8215107142925262, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.7756713867187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.22552795410155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5360, + "valid_token_count": 818.00546875 + }, + { + "actual_token_mask_ratio": 0.2994173467159271, + "epoch": 1.562409077684027, + "flow_matching_loss": 0.015894448012113573, + "grad_norm": 0.08682239800691605, + "kept_token_count": 573.23046875, + "learning_rate": 2.865638159836098e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9918626129627228, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008137381030246615, + "mgd_loss_weight": 0.8211773753166198, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.8085113525391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.5534210205078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5370, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.29936218857765196, + "epoch": 1.5653185917951702, + "flow_matching_loss": 0.01572083467617631, + "grad_norm": 0.10423661768436432, + "kept_token_count": 573.25546875, + "learning_rate": 2.8649533341304677e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9919354796409607, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00806452021934092, + "mgd_loss_weight": 0.8208440244197845, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.4423461914063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.1642547607422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5380, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.29989027976989746, + "epoch": 1.5682281059063137, + "flow_matching_loss": 0.013421181682497263, + "grad_norm": 0.09513126313686371, + "kept_token_count": 572.71953125, + "learning_rate": 2.8642668498778297e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9918536961078643, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008146293135359884, + "mgd_loss_weight": 0.8205106854438782, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.2528594970703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.75999145507814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5390, + "valid_token_count": 818.040625 + }, + { + "actual_token_mask_ratio": 0.3001730561256409, + "epoch": 1.5711376200174572, + "flow_matching_loss": 0.015604668203741312, + "grad_norm": 0.10118267685174942, + "kept_token_count": 572.7375, + "learning_rate": 2.8635787079123268e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9917238414287567, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00827614520676434, + "mgd_loss_weight": 0.8201773405075073, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.79791870117185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.76109008789064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5400, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.3001114547252655, + "epoch": 1.5740471341286004, + "flow_matching_loss": 0.01580067314207554, + "grad_norm": 0.0997631624341011, + "kept_token_count": 572.7140625, + "learning_rate": 2.8628889090701175e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.991856449842453, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008143556490540504, + "mgd_loss_weight": 0.819843989610672, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 356.6606750488281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.68459167480466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5410, + "valid_token_count": 818.29453125 + }, + { + "actual_token_mask_ratio": 0.299663308262825, + "epoch": 1.576956648239744, + "flow_matching_loss": 0.014232462830841541, + "grad_norm": 0.08811504393815994, + "kept_token_count": 573.34140625, + "learning_rate": 2.862197454189371e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9919845640659333, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008015426760539412, + "mgd_loss_weight": 0.8195106565952301, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.150927734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.0422027587891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5420, + "valid_token_count": 818.67109375 + }, + { + "actual_token_mask_ratio": 0.29963340163230895, + "epoch": 1.5798661623508874, + "flow_matching_loss": 0.01537688821554184, + "grad_norm": 0.09776468575000763, + "kept_token_count": 573.1015625, + "learning_rate": 2.8615043441102708e-05, + "loss": 0.022, + "mgd_cos_sim": 0.99190913438797, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008090862911194563, + "mgd_loss_weight": 0.8191772997379303, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.50010375976564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.2451202392578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5430, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.2997938543558121, + "epoch": 1.5827756764620309, + "flow_matching_loss": 0.015667849406599998, + "grad_norm": 0.10011804848909378, + "kept_token_count": 573.0515625, + "learning_rate": 2.860809579675011e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.992062509059906, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007937497412785887, + "mgd_loss_weight": 0.8188439607620239, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.1868377685547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.6966949462891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5440, + "valid_token_count": 818.40234375 + }, + { + "actual_token_mask_ratio": 0.30056901276111603, + "epoch": 1.5856851905731744, + "flow_matching_loss": 0.015668241027742624, + "grad_norm": 0.11236163228750229, + "kept_token_count": 572.44296875, + "learning_rate": 2.860113161727796e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9914777219295502, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008522264799103141, + "mgd_loss_weight": 0.8185106217861176, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.6786712646484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.55934448242186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5450, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.30050802826881406, + "epoch": 1.5885947046843176, + "flow_matching_loss": 0.015725666470825673, + "grad_norm": 0.08977817744016647, + "kept_token_count": 572.30625, + "learning_rate": 2.8594150911148393e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.991741544008255, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00825844369828701, + "mgd_loss_weight": 0.8181772649288177, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.9494964599609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.5705108642578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5460, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.29987705647945406, + "epoch": 1.5915042187954611, + "flow_matching_loss": 0.016394039802253246, + "grad_norm": 0.0939643532037735, + "kept_token_count": 573.06953125, + "learning_rate": 2.8587153686843626e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9913555264472962, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008644476067274809, + "mgd_loss_weight": 0.8178439319133759, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.38063354492186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.6247100830078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5470, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.30052105486392977, + "epoch": 1.5944137329066046, + "flow_matching_loss": 0.014780057407915593, + "grad_norm": 0.07706665992736816, + "kept_token_count": 572.54765625, + "learning_rate": 2.858013995286594e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9916614115238189, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00833857525140047, + "mgd_loss_weight": 0.8175105810165405, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.24584045410154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.44866333007815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5480, + "valid_token_count": 818.5375 + }, + { + "actual_token_mask_ratio": 0.299152672290802, + "epoch": 1.597323247017748, + "flow_matching_loss": 0.01616267105564475, + "grad_norm": 0.10850497335195541, + "kept_token_count": 573.49609375, + "learning_rate": 2.857310971773769e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9918712735176086, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008128729183226824, + "mgd_loss_weight": 0.8171772360801697, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.9596649169922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.3237609863281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5490, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.29990128576755526, + "epoch": 1.6002327611288916, + "flow_matching_loss": 0.016396551672369242, + "grad_norm": 0.08500552177429199, + "kept_token_count": 572.915625, + "learning_rate": 2.8566062990001274e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9918661236763, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008133870689198375, + "mgd_loss_weight": 0.8168438971042633, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.8439453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.7497833251953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5500, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.30010671615600587, + "epoch": 1.6031422752400348, + "flow_matching_loss": 0.01591946044936776, + "grad_norm": 0.10329795628786087, + "kept_token_count": 572.6984375, + "learning_rate": 2.8558999778219126e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9917364180088043, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008263580081984401, + "mgd_loss_weight": 0.816510546207428, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 354.25132751464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.77430725097656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5510, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.29998208284378053, + "epoch": 1.6060517893511783, + "flow_matching_loss": 0.014798290934413672, + "grad_norm": 0.09275329113006592, + "kept_token_count": 572.84453125, + "learning_rate": 2.8551920090973722e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9914303481578827, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008569652773439884, + "mgd_loss_weight": 0.8161772072315217, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.3179473876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.16566162109376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5520, + "valid_token_count": 818.32265625 + }, + { + "actual_token_mask_ratio": 0.2995723694562912, + "epoch": 1.6089613034623218, + "flow_matching_loss": 0.016877881623804568, + "grad_norm": 0.07910080254077911, + "kept_token_count": 573.26171875, + "learning_rate": 2.854482393686754e-05, + "loss": 0.0238, + "mgd_cos_sim": 0.9915485739707947, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008451434690505267, + "mgd_loss_weight": 0.8158438622951507, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.84092407226564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.4100372314453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5530, + "valid_token_count": 818.44453125 + }, + { + "actual_token_mask_ratio": 0.2994828224182129, + "epoch": 1.6118708175734653, + "flow_matching_loss": 0.01582608949393034, + "grad_norm": 0.10597900301218033, + "kept_token_count": 573.14921875, + "learning_rate": 2.8537711324523086e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9918020486831665, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008197955321520567, + "mgd_loss_weight": 0.8155105173587799, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 350.68288269042966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.1831298828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5540, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.29981047511100767, + "epoch": 1.6147803316846088, + "flow_matching_loss": 0.014994781371206045, + "grad_norm": 0.08316110819578171, + "kept_token_count": 572.83203125, + "learning_rate": 2.8530582262582845e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9920480012893677, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007951989257708191, + "mgd_loss_weight": 0.8151771724224091, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.51710205078126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.7841003417969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5550, + "valid_token_count": 818.10859375 + }, + { + "actual_token_mask_ratio": 0.30053374767303465, + "epoch": 1.617689845795752, + "flow_matching_loss": 0.01548551833257079, + "grad_norm": 0.09055157750844955, + "kept_token_count": 572.24140625, + "learning_rate": 2.8523436759709306e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.99215646982193, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007843525521457196, + "mgd_loss_weight": 0.8148438334465027, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.3123474121094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.7371063232422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5560, + "valid_token_count": 818.11171875 + }, + { + "actual_token_mask_ratio": 0.2996632605791092, + "epoch": 1.6205993599068955, + "flow_matching_loss": 0.01604244615882635, + "grad_norm": 0.10093460977077484, + "kept_token_count": 572.8546875, + "learning_rate": 2.8516274824584926e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9917385876178741, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008261406095698476, + "mgd_loss_weight": 0.8145104825496674, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.72022399902346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.9674560546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5570, + "valid_token_count": 817.96953125 + }, + { + "actual_token_mask_ratio": 0.30010906755924227, + "epoch": 1.623508874018039, + "flow_matching_loss": 0.01367339538410306, + "grad_norm": 0.08845505863428116, + "kept_token_count": 572.61015625, + "learning_rate": 2.8509096465912136e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9923694133758545, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007630590675398707, + "mgd_loss_weight": 0.8141771376132965, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.1913116455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.6985290527344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5580, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.29946726858615874, + "epoch": 1.6264183881291823, + "flow_matching_loss": 0.01651562675833702, + "grad_norm": 0.10117488354444504, + "kept_token_count": 573.12578125, + "learning_rate": 2.8501901692413315e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9920376062393188, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007962393574416637, + "mgd_loss_weight": 0.8138437986373901, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.14866943359374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.8542938232422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5590, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.2999470353126526, + "epoch": 1.629327902240326, + "flow_matching_loss": 0.016886898688971995, + "grad_norm": 0.12485750019550323, + "kept_token_count": 572.7859375, + "learning_rate": 2.8494690512830788e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9917109847068787, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00828902036882937, + "mgd_loss_weight": 0.8135104477405548, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.84400329589846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.22864990234376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5600, + "valid_token_count": 818.20234375 + }, + { + "actual_token_mask_ratio": 0.299938702583313, + "epoch": 1.6322374163514692, + "flow_matching_loss": 0.015075522009283304, + "grad_norm": 0.09843269735574722, + "kept_token_count": 572.82734375, + "learning_rate": 2.8487462935926823e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9916908144950867, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008309190720319748, + "mgd_loss_weight": 0.8131771087646484, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.4302093505859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.2501953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5610, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.29986667931079863, + "epoch": 1.6351469304626127, + "flow_matching_loss": 0.016684328205883503, + "grad_norm": 0.10051257908344269, + "kept_token_count": 572.6484375, + "learning_rate": 2.8480218970483606e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9909990012645722, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.009000990819185973, + "mgd_loss_weight": 0.8128437638282776, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.76921997070315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.5664459228516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5620, + "valid_token_count": 817.915625 + }, + { + "actual_token_mask_ratio": 0.3001366913318634, + "epoch": 1.6380564445737562, + "flow_matching_loss": 0.01570164179429412, + "grad_norm": 0.09096468240022659, + "kept_token_count": 572.596875, + "learning_rate": 2.847295862530324e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9915750086307525, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008424990950152278, + "mgd_loss_weight": 0.8125104129314422, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.8686126708984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.33518676757814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5630, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.2993058443069458, + "epoch": 1.6409659586848995, + "flow_matching_loss": 0.014161956217139959, + "grad_norm": 0.09524857997894287, + "kept_token_count": 573.41796875, + "learning_rate": 2.846568190920773e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9918220937252045, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00817790455184877, + "mgd_loss_weight": 0.8121770799160004, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.5447052001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.48877868652346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5640, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.29978143274784086, + "epoch": 1.6438754727960432, + "flow_matching_loss": 0.01338663911446929, + "grad_norm": 0.08460738509893417, + "kept_token_count": 573.05234375, + "learning_rate": 2.8458388831038965e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9919114768505096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008088536281138658, + "mgd_loss_weight": 0.8118437230587006, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 363.3203369140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.79400329589845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5650, + "valid_token_count": 818.39296875 + }, + { + "actual_token_mask_ratio": 0.3005087524652481, + "epoch": 1.6467849869071864, + "flow_matching_loss": 0.01450835857540369, + "grad_norm": 0.09360882639884949, + "kept_token_count": 572.51328125, + "learning_rate": 2.8451079399658725e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9915217518806457, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008478259108960629, + "mgd_loss_weight": 0.8115103840827942, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.80302734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.5518371582031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5660, + "valid_token_count": 818.47265625 + }, + { + "actual_token_mask_ratio": 0.29987368285655974, + "epoch": 1.64969450101833, + "flow_matching_loss": 0.014752175193279982, + "grad_norm": 0.08428478240966797, + "kept_token_count": 573.18203125, + "learning_rate": 2.8443753623948665e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9920101881027221, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007989813946187497, + "mgd_loss_weight": 0.8111770451068878, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 357.84423828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.1956787109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5670, + "valid_token_count": 818.68046875 + }, + { + "actual_token_mask_ratio": 0.29993982911109923, + "epoch": 1.6526040151294734, + "flow_matching_loss": 0.017946243286132812, + "grad_norm": 0.0899510458111763, + "kept_token_count": 572.77890625, + "learning_rate": 2.8436411512810288e-05, + "loss": 0.025, + "mgd_cos_sim": 0.9912631571292877, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00873684803955257, + "mgd_loss_weight": 0.810843688249588, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.3601593017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.23609313964846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5680, + "valid_token_count": 818.18671875 + }, + { + "actual_token_mask_ratio": 0.299860343337059, + "epoch": 1.6555135292406167, + "flow_matching_loss": 0.016719349194318056, + "grad_norm": 0.1147741973400116, + "kept_token_count": 573.11328125, + "learning_rate": 2.842905307516495e-05, + "loss": 0.0236, + "mgd_cos_sim": 0.9915590107440948, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008440971933305263, + "mgd_loss_weight": 0.8105103552341462, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.76903076171874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.21832580566405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5690, + "valid_token_count": 818.56875 + }, + { + "actual_token_mask_ratio": 0.30001903176307676, + "epoch": 1.6584230433517604, + "flow_matching_loss": 0.01599601740017533, + "grad_norm": 0.10684935003519058, + "kept_token_count": 572.8296875, + "learning_rate": 2.8421678319953854e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9915661931037902, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00843380349688232, + "mgd_loss_weight": 0.8101770043373108, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.27288818359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.1606964111328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5700, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.29974507689476015, + "epoch": 1.6613325574629036, + "flow_matching_loss": 0.014617434050887824, + "grad_norm": 0.08873577415943146, + "kept_token_count": 573.11875, + "learning_rate": 2.841428725613801e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9919393360614777, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008060670737177134, + "mgd_loss_weight": 0.8098436594009399, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.5782440185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.7246978759766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5710, + "valid_token_count": 818.43984375 + }, + { + "actual_token_mask_ratio": 0.30037385523319243, + "epoch": 1.6642420715740471, + "flow_matching_loss": 0.01435395572334528, + "grad_norm": 0.10346662998199463, + "kept_token_count": 572.35625, + "learning_rate": 2.8406879892698272e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9922443747520446, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007755622128024697, + "mgd_loss_weight": 0.8095103204250336, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.8386260986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.5360137939453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5720, + "valid_token_count": 818.09140625 + }, + { + "actual_token_mask_ratio": 0.30029494166374204, + "epoch": 1.6671515856851906, + "flow_matching_loss": 0.015614073909819125, + "grad_norm": 0.10997097939252853, + "kept_token_count": 572.5640625, + "learning_rate": 2.8399456238635274e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9920762121677399, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007923786202445627, + "mgd_loss_weight": 0.8091769695281983, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.8167755126953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.3795654296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5730, + "valid_token_count": 818.29921875 + }, + { + "actual_token_mask_ratio": 0.30002403259277344, + "epoch": 1.6700610997963339, + "flow_matching_loss": 0.014712621551007032, + "grad_norm": 0.10122082382440567, + "kept_token_count": 572.76875, + "learning_rate": 2.8392016302969457e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9920053958892823, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007994604110717774, + "mgd_loss_weight": 0.8088436305522919, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.70680847167966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.6578887939453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5740, + "valid_token_count": 818.26640625 + }, + { + "actual_token_mask_ratio": 0.3000562459230423, + "epoch": 1.6729706139074776, + "flow_matching_loss": 0.014312946796417236, + "grad_norm": 0.0841047540307045, + "kept_token_count": 572.759375, + "learning_rate": 2.8384560094741055e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9921643733978271, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00783562259748578, + "mgd_loss_weight": 0.808510285615921, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.8976684570313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.0197692871094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5750, + "valid_token_count": 818.29453125 + }, + { + "actual_token_mask_ratio": 0.3005290269851685, + "epoch": 1.6758801280186209, + "flow_matching_loss": 0.01653529005125165, + "grad_norm": 0.11780483275651932, + "kept_token_count": 572.31015625, + "learning_rate": 2.8377087623010054e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9915637850761414, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008436213852837682, + "mgd_loss_weight": 0.8081769347190857, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.9140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.9767669677734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5760, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.29981495141983033, + "epoch": 1.6787896421297643, + "flow_matching_loss": 0.01665301201865077, + "grad_norm": 0.09478476643562317, + "kept_token_count": 573.04140625, + "learning_rate": 2.8369598896856213e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9920403122901916, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00795967928133905, + "mgd_loss_weight": 0.8078435957431793, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.1554229736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.74512939453126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5770, + "valid_token_count": 818.42109375 + }, + { + "actual_token_mask_ratio": 0.2995516210794449, + "epoch": 1.6816991562409078, + "flow_matching_loss": 0.017000545654445885, + "grad_norm": 0.10484646260738373, + "kept_token_count": 573.31796875, + "learning_rate": 2.836209392537904e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9920325577259064, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007967452285811305, + "mgd_loss_weight": 0.8075102508068085, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.26073303222654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.18017578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5780, + "valid_token_count": 818.50078125 + }, + { + "actual_token_mask_ratio": 0.2995096266269684, + "epoch": 1.684608670352051, + "flow_matching_loss": 0.015877889283001424, + "grad_norm": 0.1123742014169693, + "kept_token_count": 573.18046875, + "learning_rate": 2.8354572717697787e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9917838156223298, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008216195506975055, + "mgd_loss_weight": 0.8071769058704377, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.3415832519531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.06458740234376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5790, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.3003000855445862, + "epoch": 1.6875181844631948, + "flow_matching_loss": 0.015813223831355572, + "grad_norm": 0.09304574877023697, + "kept_token_count": 572.53046875, + "learning_rate": 2.8347035282951428e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.992178213596344, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00782178738154471, + "mgd_loss_weight": 0.8068435609340667, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.4581726074219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.15448303222655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5800, + "valid_token_count": 818.24609375 + }, + { + "actual_token_mask_ratio": 0.300340536236763, + "epoch": 1.690427698574338, + "flow_matching_loss": 0.01838668845593929, + "grad_norm": 0.0956462025642395, + "kept_token_count": 572.628125, + "learning_rate": 2.8339481630298653e-05, + "loss": 0.0255, + "mgd_cos_sim": 0.9911865830421448, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008813419286161661, + "mgd_loss_weight": 0.8065102159976959, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.81788940429686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.1932678222656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5810, + "valid_token_count": 818.43359375 + }, + { + "actual_token_mask_ratio": 0.30034204125404357, + "epoch": 1.6933372126854815, + "flow_matching_loss": 0.017486178316175938, + "grad_norm": 0.1019105538725853, + "kept_token_count": 572.4640625, + "learning_rate": 2.833191176891787e-05, + "loss": 0.024, + "mgd_cos_sim": 0.9919507563114166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008049224549904466, + "mgd_loss_weight": 0.8061768710613251, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.54076232910154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.1699249267578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5820, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.29996145963668824, + "epoch": 1.696246726796625, + "flow_matching_loss": 0.015903625264763833, + "grad_norm": 0.0861266553401947, + "kept_token_count": 572.971875, + "learning_rate": 2.832432570800717e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.992052960395813, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007947039557620883, + "mgd_loss_weight": 0.8058435320854187, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.2938598632812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.658251953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5830, + "valid_token_count": 818.48671875 + }, + { + "actual_token_mask_ratio": 0.3000057995319366, + "epoch": 1.6991562409077683, + "flow_matching_loss": 0.014589050877839327, + "grad_norm": 0.09197624772787094, + "kept_token_count": 572.8578125, + "learning_rate": 2.8316723456784332e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9920759677886963, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007924033934250474, + "mgd_loss_weight": 0.8055101811885834, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.9722106933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.2383331298828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5840, + "valid_token_count": 818.37265625 + }, + { + "actual_token_mask_ratio": 0.2993131399154663, + "epoch": 1.702065755018912, + "flow_matching_loss": 0.015168067440390587, + "grad_norm": 0.10943000763654709, + "kept_token_count": 573.225, + "learning_rate": 2.8309105024486808e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9919887602329254, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008011238742619753, + "mgd_loss_weight": 0.8051768362522125, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.5921966552734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.0032043457031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5850, + "valid_token_count": 818.09140625 + }, + { + "actual_token_mask_ratio": 0.2993341624736786, + "epoch": 1.7049752691300553, + "flow_matching_loss": 0.015662357304245235, + "grad_norm": 0.0984647199511528, + "kept_token_count": 573.1984375, + "learning_rate": 2.830147042037172e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9916537523269653, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00834624250419438, + "mgd_loss_weight": 0.8048434972763061, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.1529968261719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.4782653808594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5860, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.3002721220254898, + "epoch": 1.7078847832411987, + "flow_matching_loss": 0.016429779957979916, + "grad_norm": 0.10079433768987656, + "kept_token_count": 572.671875, + "learning_rate": 2.8293819653715816e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9920233726501465, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007976612774655223, + "mgd_loss_weight": 0.8045101463794708, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.0014923095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.89182434082034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5870, + "valid_token_count": 818.42265625 + }, + { + "actual_token_mask_ratio": 0.3000650376081467, + "epoch": 1.7107942973523422, + "flow_matching_loss": 0.01621749857440591, + "grad_norm": 0.10349713265895844, + "kept_token_count": 572.6390625, + "learning_rate": 2.828615273381551e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9919662535190582, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008033736608922482, + "mgd_loss_weight": 0.8041768074035645, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.1865539550781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.0734344482422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5880, + "valid_token_count": 818.13203125 + }, + { + "actual_token_mask_ratio": 0.30040210485458374, + "epoch": 1.7137038114634855, + "flow_matching_loss": 0.015645627956837417, + "grad_norm": 0.11179500073194504, + "kept_token_count": 572.59609375, + "learning_rate": 2.827846966998683e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9919024229049682, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008097583847120404, + "mgd_loss_weight": 0.8038434624671936, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.00868530273436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.00487060546874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5890, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.30014524459838865, + "epoch": 1.7166133255746292, + "flow_matching_loss": 0.01644788607954979, + "grad_norm": 0.10084218531847, + "kept_token_count": 572.71796875, + "learning_rate": 2.8270770471565422e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9917442739009857, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008255715668201446, + "mgd_loss_weight": 0.8035101115703582, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.3301635742188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.93690185546876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5900, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.29981683790683744, + "epoch": 1.7195228396857725, + "flow_matching_loss": 0.015827553067356347, + "grad_norm": 0.13535350561141968, + "kept_token_count": 573.003125, + "learning_rate": 2.826305514790654e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9915621876716614, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008437820943072438, + "mgd_loss_weight": 0.8031767785549164, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.03261108398436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.85374755859374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5910, + "valid_token_count": 818.36015625 + }, + { + "actual_token_mask_ratio": 0.2995951622724533, + "epoch": 1.722432353796916, + "flow_matching_loss": 0.01734087122604251, + "grad_norm": 0.12097389996051788, + "kept_token_count": 573.2046875, + "learning_rate": 2.825532370838503e-05, + "loss": 0.0241, + "mgd_cos_sim": 0.9915518403053284, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00844817035831511, + "mgd_loss_weight": 0.8028434216976166, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.2419891357422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.2783142089844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5920, + "valid_token_count": 818.38828125 + }, + { + "actual_token_mask_ratio": 0.2998621791601181, + "epoch": 1.7253418679080594, + "flow_matching_loss": 0.01587672084569931, + "grad_norm": 0.10381873697042465, + "kept_token_count": 572.76328125, + "learning_rate": 2.824757616239532e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9917331516742707, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008266831003129482, + "mgd_loss_weight": 0.8025100827217102, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.51575622558596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.6139678955078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5930, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.3006776958703995, + "epoch": 1.7282513820192027, + "flow_matching_loss": 0.01565287932753563, + "grad_norm": 0.08635018765926361, + "kept_token_count": 572.2796875, + "learning_rate": 2.8239812519351404e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9917762339115143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008223756402730941, + "mgd_loss_weight": 0.8021767437458038, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.53093872070315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.47373962402344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5940, + "valid_token_count": 818.33671875 + }, + { + "actual_token_mask_ratio": 0.300568425655365, + "epoch": 1.7311608961303462, + "flow_matching_loss": 0.01633796850219369, + "grad_norm": 0.11330366134643555, + "kept_token_count": 572.4171875, + "learning_rate": 2.823203278868685e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9917626559734345, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00823736097663641, + "mgd_loss_weight": 0.801843386888504, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.0840728759766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.9576934814453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5950, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.29971452057361603, + "epoch": 1.7340704102414897, + "flow_matching_loss": 0.016103491745889188, + "grad_norm": 0.08636799454689026, + "kept_token_count": 573.08359375, + "learning_rate": 2.8224236979854752e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9917300879955292, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008269910933449864, + "mgd_loss_weight": 0.8015100538730622, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.87694091796874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.2297332763672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5960, + "valid_token_count": 818.35390625 + }, + { + "actual_token_mask_ratio": 0.29985687136650085, + "epoch": 1.7369799243526332, + "flow_matching_loss": 0.018781817704439163, + "grad_norm": 0.10507874935865402, + "kept_token_count": 572.796875, + "learning_rate": 2.821642510232776e-05, + "loss": 0.0256, + "mgd_cos_sim": 0.9914949774742127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00850502741523087, + "mgd_loss_weight": 0.8011767089366912, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.4799072265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.07652282714844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5970, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.29982547760009765, + "epoch": 1.7398894384637766, + "flow_matching_loss": 0.016828620992600917, + "grad_norm": 0.11870038509368896, + "kept_token_count": 572.95234375, + "learning_rate": 2.8208597165598046e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.991933000087738, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008066996978595854, + "mgd_loss_weight": 0.800843358039856, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 355.94959716796876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.57178955078126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5980, + "valid_token_count": 818.29609375 + }, + { + "actual_token_mask_ratio": 0.3002637505531311, + "epoch": 1.74279895257492, + "flow_matching_loss": 0.015104265511035919, + "grad_norm": 0.10474660992622375, + "kept_token_count": 572.5515625, + "learning_rate": 2.8200753179177285e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.991901969909668, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00809802315197885, + "mgd_loss_weight": 0.8005100190639496, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.9378631591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.821923828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 5990, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.3004086554050446, + "epoch": 1.7457084666860634, + "flow_matching_loss": 0.015394558291882276, + "grad_norm": 0.1008305549621582, + "kept_token_count": 572.671875, + "learning_rate": 2.819289315259666e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9921614944934845, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007838498288765549, + "mgd_loss_weight": 0.8001766741275788, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.7312744140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.5059326171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6000, + "valid_token_count": 818.58359375 + }, + { + "actual_token_mask_ratio": 0.29965567886829375, + "epoch": 1.7486179807972069, + "flow_matching_loss": 0.018242911621928214, + "grad_norm": 0.08542145043611526, + "kept_token_count": 572.97109375, + "learning_rate": 2.8185017095406845e-05, + "loss": 0.0247, + "mgd_cos_sim": 0.9919168770313262, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0080831294413656, + "mgd_loss_weight": 0.7998433291912079, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.4073822021484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.33453674316405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6010, + "valid_token_count": 818.11953125 + }, + { + "actual_token_mask_ratio": 0.3001460313796997, + "epoch": 1.7515274949083504, + "flow_matching_loss": 0.01478852666914463, + "grad_norm": 0.08411138504743576, + "kept_token_count": 572.51796875, + "learning_rate": 2.8177125017177998e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9920287013053894, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007971297204494476, + "mgd_loss_weight": 0.799509984254837, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 348.5089965820313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.19578247070314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6020, + "valid_token_count": 818.05703125 + }, + { + "actual_token_mask_ratio": 0.3004928410053253, + "epoch": 1.7544370090194938, + "flow_matching_loss": 0.01526795569807291, + "grad_norm": 0.09230884164571762, + "kept_token_count": 572.40625, + "learning_rate": 2.8169216927499727e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9922973096370697, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007702685939148068, + "mgd_loss_weight": 0.7991766393184662, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.030712890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.38202514648435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6030, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.30018885135650636, + "epoch": 1.757346523130637, + "flow_matching_loss": 0.0146614296361804, + "grad_norm": 0.08449451625347137, + "kept_token_count": 572.7109375, + "learning_rate": 2.816129283598112e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9915202558040619, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008479743171483279, + "mgd_loss_weight": 0.7988432943820953, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.73572998046876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.00710144042966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6040, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.300286602973938, + "epoch": 1.7602560372417806, + "flow_matching_loss": 0.015200873650610448, + "grad_norm": 0.07156120985746384, + "kept_token_count": 572.64140625, + "learning_rate": 2.815335275225068e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9917167842388153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008283226191997528, + "mgd_loss_weight": 0.798509955406189, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.99896240234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.9593017578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6050, + "valid_token_count": 818.39140625 + }, + { + "actual_token_mask_ratio": 0.3001508295536041, + "epoch": 1.763165551352924, + "flow_matching_loss": 0.014925251342356205, + "grad_norm": 0.08584708720445633, + "kept_token_count": 572.53515625, + "learning_rate": 2.8145396685956374e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9919926047325134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008007381623610855, + "mgd_loss_weight": 0.7981766045093537, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.45177307128904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.9807891845703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6060, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.3004976034164429, + "epoch": 1.7660750654640673, + "flow_matching_loss": 0.015619026031345129, + "grad_norm": 0.08350906521081924, + "kept_token_count": 572.53125, + "learning_rate": 2.813742464676556e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9912543654441833, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008745627291500569, + "mgd_loss_weight": 0.7978432595729827, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.1066131591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.32196044921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6070, + "valid_token_count": 818.48671875 + }, + { + "actual_token_mask_ratio": 0.29857954382896423, + "epoch": 1.768984579575211, + "flow_matching_loss": 0.015275627188384534, + "grad_norm": 0.1157461628317833, + "kept_token_count": 573.84296875, + "learning_rate": 2.8129436644365025e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.992351233959198, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00764877200126648, + "mgd_loss_weight": 0.7975099205970764, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.58338317871096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.5882110595703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6080, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.29948853254318236, + "epoch": 1.7718940936863543, + "flow_matching_loss": 0.015092466678470372, + "grad_norm": 0.08802545815706253, + "kept_token_count": 573.1671875, + "learning_rate": 2.8121432688460938e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9921318829059601, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00786810927093029, + "mgd_loss_weight": 0.7971765697002411, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.5354278564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.59430541992185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6090, + "valid_token_count": 818.21328125 + }, + { + "actual_token_mask_ratio": 0.3002365976572037, + "epoch": 1.7748036077974978, + "flow_matching_loss": 0.01670172456651926, + "grad_norm": 0.09501034766435623, + "kept_token_count": 572.56484375, + "learning_rate": 2.8113412788778867e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.9918062686920166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00819373936392367, + "mgd_loss_weight": 0.7968432307243347, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.92877502441405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.19060668945315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6100, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.29968746304512023, + "epoch": 1.7777131219086413, + "flow_matching_loss": 0.014878320693969726, + "grad_norm": 0.08527588844299316, + "kept_token_count": 573.1140625, + "learning_rate": 2.8105376955063744e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.992196387052536, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007803615601733327, + "mgd_loss_weight": 0.7965098857879639, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 358.7955627441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.0223052978516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6110, + "valid_token_count": 818.36328125 + }, + { + "actual_token_mask_ratio": 0.2997651159763336, + "epoch": 1.7806226360197845, + "flow_matching_loss": 0.016332939080893994, + "grad_norm": 0.0919342041015625, + "kept_token_count": 572.921875, + "learning_rate": 2.8097325197079862e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9918552577495575, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008144733076915146, + "mgd_loss_weight": 0.7961765348911285, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.7886108398437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.83497619628906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6120, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.30060975849628446, + "epoch": 1.7835321501309283, + "flow_matching_loss": 0.013924339786171913, + "grad_norm": 0.0834452211856842, + "kept_token_count": 572.28984375, + "learning_rate": 2.808925752461087e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9917385935783386, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008261399855837227, + "mgd_loss_weight": 0.7958432018756867, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.18260192871094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.5315185546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6130, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.30005930066108705, + "epoch": 1.7864416642420715, + "flow_matching_loss": 0.016195862088352442, + "grad_norm": 0.09008028358221054, + "kept_token_count": 573.07109375, + "learning_rate": 2.808117394745975e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9918533980846405, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008146605594083667, + "mgd_loss_weight": 0.7955098450183868, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.4170227050781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.6451049804688, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6140, + "valid_token_count": 818.740625 + }, + { + "actual_token_mask_ratio": 0.3004649579524994, + "epoch": 1.789351178353215, + "flow_matching_loss": 0.01603146493434906, + "grad_norm": 0.11027734726667404, + "kept_token_count": 572.51953125, + "learning_rate": 2.807307447544881e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9919494330883026, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008050578273832798, + "mgd_loss_weight": 0.7951765060424805, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.6441986083984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.5465118408203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6150, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.29915752410888674, + "epoch": 1.7922606924643585, + "flow_matching_loss": 0.016362899076193572, + "grad_norm": 0.09646408259868622, + "kept_token_count": 573.44765625, + "learning_rate": 2.806495911841967e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9919436156749726, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008056386932730674, + "mgd_loss_weight": 0.7948431670665741, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.4414855957031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.3742218017578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6160, + "valid_token_count": 818.22890625 + }, + { + "actual_token_mask_ratio": 0.29979760348796847, + "epoch": 1.7951702065755017, + "flow_matching_loss": 0.016198218520730733, + "grad_norm": 0.0982213094830513, + "kept_token_count": 572.9859375, + "learning_rate": 2.8056827886233255e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9917253196239472, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008274673204869031, + "mgd_loss_weight": 0.7945098102092742, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.18938903808595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.38109130859374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6170, + "valid_token_count": 818.31953125 + }, + { + "actual_token_mask_ratio": 0.30046831965446474, + "epoch": 1.7980797206866455, + "flow_matching_loss": 0.016595635656267406, + "grad_norm": 0.10168983042240143, + "kept_token_count": 572.4828125, + "learning_rate": 2.8048680788769783e-05, + "loss": 0.023, + "mgd_cos_sim": 0.9919734179973603, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008026572363451123, + "mgd_loss_weight": 0.7941764771938324, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.8082305908203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.99054260253905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6180, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.300372326374054, + "epoch": 1.8009892347977887, + "flow_matching_loss": 0.014769930206239224, + "grad_norm": 0.09826585650444031, + "kept_token_count": 572.58125, + "learning_rate": 2.8040517835928742e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.991560423374176, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008439586963504553, + "mgd_loss_weight": 0.793843126296997, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.539501953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.08758544921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6190, + "valid_token_count": 818.40859375 + }, + { + "actual_token_mask_ratio": 0.30037304759025574, + "epoch": 1.8038987489089322, + "flow_matching_loss": 0.014507305901497602, + "grad_norm": 0.10828498750925064, + "kept_token_count": 572.72890625, + "learning_rate": 2.803233903762889e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9920971930027008, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00790280820801854, + "mgd_loss_weight": 0.7935097813606262, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.1916259765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.3275421142578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6200, + "valid_token_count": 818.615625 + }, + { + "actual_token_mask_ratio": 0.2995444357395172, + "epoch": 1.8068082630200757, + "flow_matching_loss": 0.013676232006400824, + "grad_norm": 0.08633208274841309, + "kept_token_count": 573.27578125, + "learning_rate": 2.8024144403808235e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9922111630439758, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00778885786421597, + "mgd_loss_weight": 0.7931764423847198, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.0597778320313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.964794921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6210, + "valid_token_count": 818.43046875 + }, + { + "actual_token_mask_ratio": 0.30039556324481964, + "epoch": 1.809717777131219, + "flow_matching_loss": 0.015444149263203144, + "grad_norm": 0.09592227637767792, + "kept_token_count": 572.45625, + "learning_rate": 2.8015933944424037e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9918652057647706, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008134792512282728, + "mgd_loss_weight": 0.7928430914878846, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.35243530273436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.52198791503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6220, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.29968003630638124, + "epoch": 1.8126272912423627, + "flow_matching_loss": 0.013941686600446701, + "grad_norm": 0.0860346257686615, + "kept_token_count": 573.29375, + "learning_rate": 2.800770766945277e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9919162631034851, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008083747373893857, + "mgd_loss_weight": 0.7925097525119782, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.1705657958984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.2541564941406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6230, + "valid_token_count": 818.6140625 + }, + { + "actual_token_mask_ratio": 0.299495792388916, + "epoch": 1.815536805353506, + "flow_matching_loss": 0.015271100588142872, + "grad_norm": 0.09245933592319489, + "kept_token_count": 573.06015625, + "learning_rate": 2.7999465588890135e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9919163465499878, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008083656337112189, + "mgd_loss_weight": 0.7921764075756073, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.9990203857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.1697998046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6240, + "valid_token_count": 818.07265625 + }, + { + "actual_token_mask_ratio": 0.3008892863988876, + "epoch": 1.8184463194646494, + "flow_matching_loss": 0.015898845065385102, + "grad_norm": 0.09493198245763779, + "kept_token_count": 572.10625, + "learning_rate": 2.799120771275105e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9921834349632264, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007816548738628627, + "mgd_loss_weight": 0.791843056678772, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.8804504394531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.5349395751953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6250, + "valid_token_count": 818.33359375 + }, + { + "actual_token_mask_ratio": 0.2995272547006607, + "epoch": 1.821355833575793, + "flow_matching_loss": 0.015724950656294822, + "grad_norm": 0.09834413230419159, + "kept_token_count": 573.2625, + "learning_rate": 2.7982934051069597e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9919336438179016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00806635431945324, + "mgd_loss_weight": 0.7915097177028656, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.07024536132815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.4096649169922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6260, + "valid_token_count": 818.39140625 + }, + { + "actual_token_mask_ratio": 0.3003502428531647, + "epoch": 1.8242653476869362, + "flow_matching_loss": 0.01485950816422701, + "grad_norm": 0.10990852862596512, + "kept_token_count": 572.5046875, + "learning_rate": 2.7974644613899063e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9921453773975373, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007854632753878831, + "mgd_loss_weight": 0.7911763727664948, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.3313507080078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.0051605224609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6270, + "valid_token_count": 818.27421875 + }, + { + "actual_token_mask_ratio": 0.30012906789779664, + "epoch": 1.8271748617980799, + "flow_matching_loss": 0.015581966657191516, + "grad_norm": 0.10148973762989044, + "kept_token_count": 572.6296875, + "learning_rate": 2.796633941131189e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9917914032936096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008208594471216201, + "mgd_loss_weight": 0.790843027830124, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.7628936767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.551904296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6280, + "valid_token_count": 818.19296875 + }, + { + "actual_token_mask_ratio": 0.29984164237976074, + "epoch": 1.8300843759092231, + "flow_matching_loss": 0.015936459880322218, + "grad_norm": 0.09998315572738647, + "kept_token_count": 572.8578125, + "learning_rate": 2.795801845339969e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9925055086612702, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074944784399122, + "mgd_loss_weight": 0.790509682893753, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.3491455078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.03680725097655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6290, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.2992583900690079, + "epoch": 1.8329938900203666, + "flow_matching_loss": 0.014231806807219981, + "grad_norm": 0.08148979395627975, + "kept_token_count": 573.415625, + "learning_rate": 2.7949681750273208e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9922938644886017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007706116279587149, + "mgd_loss_weight": 0.7901763379573822, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.40859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.6449310302734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6300, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.2997848242521286, + "epoch": 1.83590340413151, + "flow_matching_loss": 0.013721953053027392, + "grad_norm": 0.07902102172374725, + "kept_token_count": 572.875, + "learning_rate": 2.7941329312062322e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9920088648796082, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007991128601133823, + "mgd_loss_weight": 0.7898429930210114, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.5971221923828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.5828002929687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6310, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.2995757609605789, + "epoch": 1.8388129182426534, + "flow_matching_loss": 0.01599084595218301, + "grad_norm": 0.09768800437450409, + "kept_token_count": 573.11796875, + "learning_rate": 2.7932961148916037e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9919314563274384, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008068531891331077, + "mgd_loss_weight": 0.789509654045105, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.02179565429685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.69074096679685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6320, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.30052387416362764, + "epoch": 1.841722432353797, + "flow_matching_loss": 0.01433853544294834, + "grad_norm": 0.09393031895160675, + "kept_token_count": 572.2375, + "learning_rate": 2.7924577271002456e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.99231715798378, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076828282326459885, + "mgd_loss_weight": 0.7891763031482697, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.0519714355469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.24986572265624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6330, + "valid_token_count": 818.1 + }, + { + "actual_token_mask_ratio": 0.2993997871875763, + "epoch": 1.8446319464649403, + "flow_matching_loss": 0.01537980167195201, + "grad_norm": 0.08628673106431961, + "kept_token_count": 573.353125, + "learning_rate": 2.7916177688508782e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9920722186565399, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00792777370661497, + "mgd_loss_weight": 0.7888429582118988, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.2937957763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.4073760986328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6340, + "valid_token_count": 818.37421875 + }, + { + "actual_token_mask_ratio": 0.29986049234867096, + "epoch": 1.8475414605760838, + "flow_matching_loss": 0.015398361347615718, + "grad_norm": 0.07913151383399963, + "kept_token_count": 572.9796875, + "learning_rate": 2.7907762411641302e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9924088954925537, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007591098779812455, + "mgd_loss_weight": 0.7885096192359924, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.2772613525391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.6710693359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6350, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.2994632601737976, + "epoch": 1.8504509746872273, + "flow_matching_loss": 0.01653935294598341, + "grad_norm": 0.10048086196184158, + "kept_token_count": 573.19453125, + "learning_rate": 2.7899331450625367e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9920158743858337, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007984131202101707, + "mgd_loss_weight": 0.7881762683391571, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 366.4565032958984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.2945556640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6360, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.29938399195671084, + "epoch": 1.8533604887983706, + "flow_matching_loss": 0.015481125563383102, + "grad_norm": 0.08235948532819748, + "kept_token_count": 573.24140625, + "learning_rate": 2.7890884815705397e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9922586798667907, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007741322554647922, + "mgd_loss_weight": 0.7878429293632507, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 369.84007873535154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.7698455810547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6370, + "valid_token_count": 818.19921875 + }, + { + "actual_token_mask_ratio": 0.3004003703594208, + "epoch": 1.8562700029095143, + "flow_matching_loss": 0.017289834562689066, + "grad_norm": 0.20104695856571198, + "kept_token_count": 572.4796875, + "learning_rate": 2.788242251714484e-05, + "loss": 0.0237, + "mgd_cos_sim": 0.9919061481952667, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008093865867704154, + "mgd_loss_weight": 0.7875095844268799, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.23716735839844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.6001220703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6380, + "valid_token_count": 818.30078125 + }, + { + "actual_token_mask_ratio": 0.2998005300760269, + "epoch": 1.8591795170206575, + "flow_matching_loss": 0.01562052546069026, + "grad_norm": 0.1076345294713974, + "kept_token_count": 572.9796875, + "learning_rate": 2.787394456522619e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9923553049564362, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00764468926936388, + "mgd_loss_weight": 0.7871762335300445, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.9170623779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.1385223388672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6390, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.3001043856143951, + "epoch": 1.862089031131801, + "flow_matching_loss": 0.017145533580332994, + "grad_norm": 0.10305782407522202, + "kept_token_count": 572.48203125, + "learning_rate": 2.7865450970250965e-05, + "loss": 0.0235, + "mgd_cos_sim": 0.9918656289577484, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008134375512599944, + "mgd_loss_weight": 0.7868429005146027, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.8628265380859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.80457458496096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6400, + "valid_token_count": 817.9546875 + }, + { + "actual_token_mask_ratio": 0.3000065565109253, + "epoch": 1.8649985452429445, + "flow_matching_loss": 0.01527741439640522, + "grad_norm": 0.08588039875030518, + "kept_token_count": 572.84296875, + "learning_rate": 2.7856941742539676e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9920076251029968, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007992369681596756, + "mgd_loss_weight": 0.7865095496177673, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.87127075195315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.3127868652344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6410, + "valid_token_count": 818.35546875 + }, + { + "actual_token_mask_ratio": 0.3002581000328064, + "epoch": 1.8679080593540878, + "flow_matching_loss": 0.014942377712577582, + "grad_norm": 0.0955953449010849, + "kept_token_count": 572.82578125, + "learning_rate": 2.784841689243184e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9913811564445496, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008618855755776167, + "mgd_loss_weight": 0.7861762046813965, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.08063354492185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.2858947753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6420, + "valid_token_count": 818.62109375 + }, + { + "actual_token_mask_ratio": 0.30015375912189485, + "epoch": 1.8708175734652313, + "flow_matching_loss": 0.016201116051524876, + "grad_norm": 0.09455527365207672, + "kept_token_count": 572.63515625, + "learning_rate": 2.783987643028595e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9916639685630798, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008336037071421743, + "mgd_loss_weight": 0.7858428657054901, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.8599914550781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.0574554443359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6430, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.2997658997774124, + "epoch": 1.8737270875763747, + "flow_matching_loss": 0.015021471492946147, + "grad_norm": 0.11248814314603806, + "kept_token_count": 572.8640625, + "learning_rate": 2.783132036647948e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9921563506126404, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007843658700585366, + "mgd_loss_weight": 0.7855095148086548, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.7204162597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.152978515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6440, + "valid_token_count": 818.1 + }, + { + "actual_token_mask_ratio": 0.3002237617969513, + "epoch": 1.8766366016875182, + "flow_matching_loss": 0.01526015130802989, + "grad_norm": 0.0747879147529602, + "kept_token_count": 572.63046875, + "learning_rate": 2.7822748711408852e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9922023415565491, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007797651458531618, + "mgd_loss_weight": 0.7851761758327485, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.4186157226562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3662902832031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6450, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.3004394769668579, + "epoch": 1.8795461157986617, + "flow_matching_loss": 0.013828113209456205, + "grad_norm": 0.0942983329296112, + "kept_token_count": 572.5328125, + "learning_rate": 2.7814161475489428e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9920737504959106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007926246570423245, + "mgd_loss_weight": 0.7848428308963775, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.6723907470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.5173370361328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6460, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.3002811461687088, + "epoch": 1.882455629909805, + "flow_matching_loss": 0.015171274915337563, + "grad_norm": 0.08584915101528168, + "kept_token_count": 572.528125, + "learning_rate": 2.780555866915552e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9924227714538574, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007577217509970069, + "mgd_loss_weight": 0.7845094799995422, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 370.42945556640626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.67583618164065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6470, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.3000072568655014, + "epoch": 1.8853651440209485, + "flow_matching_loss": 0.016072924062609673, + "grad_norm": 0.08296915143728256, + "kept_token_count": 572.85546875, + "learning_rate": 2.7796940302860338e-05, + "loss": 0.0227, + "mgd_cos_sim": 0.9915778934955597, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008422115352004766, + "mgd_loss_weight": 0.7841761410236359, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.24066162109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.2582672119141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6480, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.3001152008771896, + "epoch": 1.888274658132092, + "flow_matching_loss": 0.014966982696205378, + "grad_norm": 0.0864119753241539, + "kept_token_count": 572.67109375, + "learning_rate": 2.7788306387076018e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9921613931655884, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007838604878634215, + "mgd_loss_weight": 0.783842796087265, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.23939514160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.3546539306641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6490, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.3002200722694397, + "epoch": 1.8911841722432354, + "flow_matching_loss": 0.01557296235114336, + "grad_norm": 0.1017560288310051, + "kept_token_count": 572.675, + "learning_rate": 2.7779656932293584e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9920069754123688, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007993023050948978, + "mgd_loss_weight": 0.7835094511508942, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.4764129638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.28130798339845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6500, + "valid_token_count": 818.36640625 + }, + { + "actual_token_mask_ratio": 0.3002092748880386, + "epoch": 1.894093686354379, + "flow_matching_loss": 0.01600720761343837, + "grad_norm": 0.10310179740190506, + "kept_token_count": 572.615625, + "learning_rate": 2.777099194902293e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.991891348361969, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008108646422624589, + "mgd_loss_weight": 0.7831761062145233, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.7382781982422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.15143432617185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6510, + "valid_token_count": 818.26640625 + }, + { + "actual_token_mask_ratio": 0.2996422290802002, + "epoch": 1.8970032004655222, + "flow_matching_loss": 0.014832999929785729, + "grad_norm": 0.11022034287452698, + "kept_token_count": 573.06328125, + "learning_rate": 2.776231144779283e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9920801460742951, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007919873716309667, + "mgd_loss_weight": 0.7828427612781524, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.81700134277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.2751434326172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6520, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.3005241990089417, + "epoch": 1.8999127145766657, + "flow_matching_loss": 0.014734744653105735, + "grad_norm": 0.0788157656788826, + "kept_token_count": 572.34375, + "learning_rate": 2.775361543915091e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9924875199794769, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007512494316324591, + "mgd_loss_weight": 0.7825094163417816, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.69476928710935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.2035705566406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6530, + "valid_token_count": 818.24609375 + }, + { + "actual_token_mask_ratio": 0.300268879532814, + "epoch": 1.9028222286878091, + "flow_matching_loss": 0.015150727611035109, + "grad_norm": 0.09396063536405563, + "kept_token_count": 572.7078125, + "learning_rate": 2.774490393366365e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9922486782073975, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007751321140676737, + "mgd_loss_weight": 0.7821760773658752, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.6740661621094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.22812805175784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6540, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.29911001324653624, + "epoch": 1.9057317427989524, + "flow_matching_loss": 0.014982727356255054, + "grad_norm": 0.08144277334213257, + "kept_token_count": 573.62578125, + "learning_rate": 2.7736176941916346e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9923450112342834, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007654982525855303, + "mgd_loss_weight": 0.78184272646904, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.0146759033203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.54933471679686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6550, + "valid_token_count": 818.42734375 + }, + { + "actual_token_mask_ratio": 0.2999176800251007, + "epoch": 1.9086412569100961, + "flow_matching_loss": 0.016096761357039212, + "grad_norm": 0.1026734933257103, + "kept_token_count": 572.8390625, + "learning_rate": 2.772743447451311e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9921566367149353, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007843363238498569, + "mgd_loss_weight": 0.781509381532669, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.32301025390626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4231872558594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6560, + "valid_token_count": 818.24453125 + }, + { + "actual_token_mask_ratio": 0.30010112524032595, + "epoch": 1.9115507710212394, + "flow_matching_loss": 0.013850516360253095, + "grad_norm": 0.08699935674667358, + "kept_token_count": 572.85, + "learning_rate": 2.7718676542076872e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.992246824502945, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00775317195802927, + "mgd_loss_weight": 0.7811760425567627, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.61192016601564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.23255920410156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6570, + "valid_token_count": 818.471875 + }, + { + "actual_token_mask_ratio": 0.2996668636798859, + "epoch": 1.9144602851323829, + "flow_matching_loss": 0.014995712507516145, + "grad_norm": 0.08515582233667374, + "kept_token_count": 573.1015625, + "learning_rate": 2.770990315524934e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9920992910861969, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007900703884661197, + "mgd_loss_weight": 0.7808426916599274, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.6542266845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.3186889648438, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6580, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.29978411793708803, + "epoch": 1.9173697992435264, + "flow_matching_loss": 0.015630471613258125, + "grad_norm": 0.10780853033065796, + "kept_token_count": 572.9453125, + "learning_rate": 2.7701114324691015e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9917787134647369, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008221289282664657, + "mgd_loss_weight": 0.780509352684021, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.587548828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.5045989990234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6590, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.2997666209936142, + "epoch": 1.9202793133546696, + "flow_matching_loss": 0.01454253140836954, + "grad_norm": 0.09032048285007477, + "kept_token_count": 572.8921875, + "learning_rate": 2.7692310061081145e-05, + "loss": 0.021, + "mgd_cos_sim": 0.99178626537323, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008213740447536111, + "mgd_loss_weight": 0.7801760077476502, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 362.8233306884766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6155731201172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6600, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.29921576380729675, + "epoch": 1.9231888274658133, + "flow_matching_loss": 0.014990570396184922, + "grad_norm": 0.09207337349653244, + "kept_token_count": 573.2609375, + "learning_rate": 2.7683490375117742e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.992053097486496, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007946915086358785, + "mgd_loss_weight": 0.7798426568508148, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.05189208984376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.5498107910156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6610, + "valid_token_count": 818.0265625 + }, + { + "actual_token_mask_ratio": 0.2993545472621918, + "epoch": 1.9260983415769566, + "flow_matching_loss": 0.014574695844203234, + "grad_norm": 0.09173353761434555, + "kept_token_count": 573.23515625, + "learning_rate": 2.7674655277517556e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9924269139766693, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007573080761358142, + "mgd_loss_weight": 0.779509323835373, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.9056823730469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9614562988281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6620, + "valid_token_count": 818.15078125 + }, + { + "actual_token_mask_ratio": 0.2998159497976303, + "epoch": 1.9290078556881, + "flow_matching_loss": 0.015647316724061965, + "grad_norm": 0.09523744136095047, + "kept_token_count": 572.9375, + "learning_rate": 2.7665804779016072e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9922246754169464, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007775320205837488, + "mgd_loss_weight": 0.7791759669780731, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.8861602783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3817504882812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6630, + "valid_token_count": 818.26796875 + }, + { + "actual_token_mask_ratio": 0.3002897709608078, + "epoch": 1.9319173697992436, + "flow_matching_loss": 0.014568277262151241, + "grad_norm": 0.09743563830852509, + "kept_token_count": 572.5640625, + "learning_rate": 2.765693889036747e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9924336433410644, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075663560535758735, + "mgd_loss_weight": 0.7788426280021667, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.9820098876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.7199951171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6640, + "valid_token_count": 818.28515625 + }, + { + "actual_token_mask_ratio": 0.3001023143529892, + "epoch": 1.9348268839103868, + "flow_matching_loss": 0.015495580155402422, + "grad_norm": 0.09777134656906128, + "kept_token_count": 572.55390625, + "learning_rate": 2.764805762234465e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9921312272548676, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007868777867406607, + "mgd_loss_weight": 0.7785092890262604, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.80959167480466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.52611083984374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6650, + "valid_token_count": 818.05390625 + }, + { + "actual_token_mask_ratio": 0.299761563539505, + "epoch": 1.9377363980215305, + "flow_matching_loss": 0.016075152438133955, + "grad_norm": 0.09087062627077103, + "kept_token_count": 573.04140625, + "learning_rate": 2.7639160985739182e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.9917422950267791, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008257695427164435, + "mgd_loss_weight": 0.7781759321689605, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.1920196533203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.29710693359374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6660, + "valid_token_count": 818.35625 + }, + { + "actual_token_mask_ratio": 0.29975866973400117, + "epoch": 1.9406459121326738, + "flow_matching_loss": 0.015123766753822565, + "grad_norm": 0.08332344144582748, + "kept_token_count": 572.884375, + "learning_rate": 2.7630248991361317e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9923829019069672, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007617097115144134, + "mgd_loss_weight": 0.7778425991535187, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.63816528320314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7264831542969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6670, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.3004212111234665, + "epoch": 1.9435554262438173, + "flow_matching_loss": 0.016255698539316653, + "grad_norm": 0.10932395607233047, + "kept_token_count": 572.41171875, + "learning_rate": 2.762132165003998e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9924656450748444, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007534350547939539, + "mgd_loss_weight": 0.7775092482566833, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.03720703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.8695922851563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6680, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.3001103937625885, + "epoch": 1.9464649403549608, + "flow_matching_loss": 0.014008199609816075, + "grad_norm": 0.11622331291437149, + "kept_token_count": 572.80546875, + "learning_rate": 2.7612378972622722e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9927103459835053, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007289660861715674, + "mgd_loss_weight": 0.7771759033203125, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.5987915039062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.52550048828124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6690, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.30021015405654905, + "epoch": 1.949374454466104, + "flow_matching_loss": 0.015124044939875602, + "grad_norm": 0.10434050858020782, + "kept_token_count": 572.8, + "learning_rate": 2.7603420969975743e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9916066348552703, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008393356716260314, + "mgd_loss_weight": 0.7768425643444061, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.9928466796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.9097930908203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6700, + "valid_token_count": 818.53125 + }, + { + "actual_token_mask_ratio": 0.299857497215271, + "epoch": 1.9522839685772477, + "flow_matching_loss": 0.01678608525544405, + "grad_norm": 0.09868375211954117, + "kept_token_count": 572.9453125, + "learning_rate": 2.7594447652983862e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.992491352558136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00750864869914949, + "mgd_loss_weight": 0.7765092134475708, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.52178649902345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.2958953857422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6710, + "valid_token_count": 818.32265625 + }, + { + "actual_token_mask_ratio": 0.3003624051809311, + "epoch": 1.955193482688391, + "flow_matching_loss": 0.016605083737522362, + "grad_norm": 0.10253126174211502, + "kept_token_count": 572.603125, + "learning_rate": 2.7585459032550497e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9923230290412903, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007676967838779092, + "mgd_loss_weight": 0.7761758744716645, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.9268798828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.1961669921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6720, + "valid_token_count": 818.42890625 + }, + { + "actual_token_mask_ratio": 0.29994978904724123, + "epoch": 1.9581029967995345, + "flow_matching_loss": 0.014603335130959749, + "grad_norm": 0.0876510962843895, + "kept_token_count": 572.8140625, + "learning_rate": 2.7576455119597684e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9926647782325745, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007335228566080332, + "mgd_loss_weight": 0.7758425295352935, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.8012664794922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.6692657470703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6730, + "valid_token_count": 818.25390625 + }, + { + "actual_token_mask_ratio": 0.30019004046916964, + "epoch": 1.961012510910678, + "flow_matching_loss": 0.015761607326567173, + "grad_norm": 0.09283314645290375, + "kept_token_count": 572.59140625, + "learning_rate": 2.7567435925066012e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9921544253826141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007845591427758336, + "mgd_loss_weight": 0.7755091786384583, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.65473022460935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4642028808594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6740, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.29968840777873995, + "epoch": 1.9639220250218212, + "flow_matching_loss": 0.01610939111560583, + "grad_norm": 0.1090560331940651, + "kept_token_count": 572.940625, + "learning_rate": 2.7558401459914665e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9920869886875152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00791301648132503, + "mgd_loss_weight": 0.7751758396625519, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.47596130371096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.59166259765624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6750, + "valid_token_count": 818.12265625 + }, + { + "actual_token_mask_ratio": 0.30004760921001433, + "epoch": 1.966831539132965, + "flow_matching_loss": 0.014828332420438527, + "grad_norm": 0.0914950892329216, + "kept_token_count": 572.690625, + "learning_rate": 2.7549351735121362e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9922057807445526, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007794211572036147, + "mgd_loss_weight": 0.774842494726181, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.2581329345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.73292846679686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6760, + "valid_token_count": 818.1875 + }, + { + "actual_token_mask_ratio": 0.3005615919828415, + "epoch": 1.9697410532441082, + "flow_matching_loss": 0.015592668391764163, + "grad_norm": 0.11296352744102478, + "kept_token_count": 572.2203125, + "learning_rate": 2.7540286761682377e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9919142901897431, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008085712091997266, + "mgd_loss_weight": 0.7745091497898102, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.1910034179688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.6826538085937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6770, + "valid_token_count": 818.11484375 + }, + { + "actual_token_mask_ratio": 0.3000668227672577, + "epoch": 1.9726505673552517, + "flow_matching_loss": 0.01684422604739666, + "grad_norm": 0.11080926656723022, + "kept_token_count": 572.77578125, + "learning_rate": 2.75312065506125e-05, + "loss": 0.0229, + "mgd_cos_sim": 0.9921132981777191, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007886683428660035, + "mgd_loss_weight": 0.7741758048534393, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.23227844238284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.7703887939453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6780, + "valid_token_count": 818.32734375 + }, + { + "actual_token_mask_ratio": 0.2999602109193802, + "epoch": 1.9755600814663952, + "flow_matching_loss": 0.016332439612597227, + "grad_norm": 0.10051589459180832, + "kept_token_count": 572.79140625, + "learning_rate": 2.7522111112945052e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9919520795345307, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00804792707785964, + "mgd_loss_weight": 0.7738424599170685, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.19398803710936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.2185882568359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6790, + "valid_token_count": 818.23046875 + }, + { + "actual_token_mask_ratio": 0.29986332058906556, + "epoch": 1.9784695955775384, + "flow_matching_loss": 0.01581426812335849, + "grad_norm": 0.10207096487283707, + "kept_token_count": 572.93671875, + "learning_rate": 2.751300045973184e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9919222116470336, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008077786443755031, + "mgd_loss_weight": 0.7735091149806976, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.51304016113284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.6497802734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6800, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.29960032999515535, + "epoch": 1.9813791096886821, + "flow_matching_loss": 0.015133300796151161, + "grad_norm": 0.1084214448928833, + "kept_token_count": 573.02265625, + "learning_rate": 2.750387460204317e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9918053030967713, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008194697787985206, + "mgd_loss_weight": 0.7731757760047913, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.6796173095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.45485229492186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6810, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.30103684961795807, + "epoch": 1.9842886237998254, + "flow_matching_loss": 0.014201348647475243, + "grad_norm": 0.07354968041181564, + "kept_token_count": 572.02890625, + "learning_rate": 2.7494733550967827e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.992407888174057, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007592122675850988, + "mgd_loss_weight": 0.772842425107956, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.56743774414065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.9269653320313, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6820, + "valid_token_count": 818.39296875 + }, + { + "actual_token_mask_ratio": 0.3014198273420334, + "epoch": 1.9871981379109689, + "flow_matching_loss": 0.015064736269414425, + "grad_norm": 0.08853761851787567, + "kept_token_count": 571.8625, + "learning_rate": 2.748557731761304e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9923635244369506, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007636488927528262, + "mgd_loss_weight": 0.772509080171585, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.2338073730469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.2624755859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6830, + "valid_token_count": 818.60234375 + }, + { + "actual_token_mask_ratio": 0.3003799945116043, + "epoch": 1.9901076520221124, + "flow_matching_loss": 0.016041153389960528, + "grad_norm": 0.10062578320503235, + "kept_token_count": 572.44765625, + "learning_rate": 2.74764059131045e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9920849859714508, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00791500904597342, + "mgd_loss_weight": 0.7721757471561432, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 360.8034942626953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.95485229492186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6840, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.2999114602804184, + "epoch": 1.9930171661332556, + "flow_matching_loss": 0.01689590625464916, + "grad_norm": 0.09903041273355484, + "kept_token_count": 572.83515625, + "learning_rate": 2.7467219348586322e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9919469594955445, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008053042320534587, + "mgd_loss_weight": 0.7718423902988434, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.8304504394531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.2384948730469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6850, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.29980082511901857, + "epoch": 1.9959266802443993, + "flow_matching_loss": 0.015110437106341124, + "grad_norm": 0.06945187598466873, + "kept_token_count": 572.87109375, + "learning_rate": 2.7458017635221056e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9921286880970002, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007871320331469178, + "mgd_loss_weight": 0.771509051322937, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.33042907714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.64989318847654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6860, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.30121963322162626, + "epoch": 1.9988361943555426, + "flow_matching_loss": 0.016901601385325194, + "grad_norm": 0.10442190617322922, + "kept_token_count": 571.83125, + "learning_rate": 2.7448800784189652e-05, + "loss": 0.0232, + "mgd_cos_sim": 0.991864675283432, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008135325508192182, + "mgd_loss_weight": 0.7711757123470306, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.5410614013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.662060546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6870, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.2998514711856842, + "epoch": 2.001745708466686, + "flow_matching_loss": 0.016116877552121878, + "grad_norm": 0.12130974978208542, + "kept_token_count": 573.1248962402344, + "learning_rate": 2.7439568806691448e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9921195149421692, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007880501030012966, + "mgd_loss_weight": 0.7708423554897308, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.26336059570315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.1365631103516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6880, + "valid_token_count": 818.5762268066406 + }, + { + "actual_token_mask_ratio": 0.2994454920291901, + "epoch": 2.0046552225778296, + "flow_matching_loss": 0.014020807668566703, + "grad_norm": 0.11265693604946136, + "kept_token_count": 573.33203125, + "learning_rate": 2.7430321713944167e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9921358466148377, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007864155154675246, + "mgd_loss_weight": 0.770509022474289, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.0554229736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.6970520019531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6890, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.29971586763858793, + "epoch": 2.007564736688973, + "flow_matching_loss": 0.014431642554700374, + "grad_norm": 0.11922623217105865, + "kept_token_count": 572.88671875, + "learning_rate": 2.7421059517183906e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9921436727046966, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007856329530477523, + "mgd_loss_weight": 0.7701756715774536, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.2861053466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.84760131835935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6900, + "valid_token_count": 818.07421875 + }, + { + "actual_token_mask_ratio": 0.30003839135169985, + "epoch": 2.0104742508001165, + "flow_matching_loss": 0.015816697850823402, + "grad_norm": 0.14033353328704834, + "kept_token_count": 572.85703125, + "learning_rate": 2.7411782227665097e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9919213593006134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008078644797205926, + "mgd_loss_weight": 0.7698423266410828, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.26149291992186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.15813598632815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6910, + "valid_token_count": 818.40859375 + }, + { + "actual_token_mask_ratio": 0.2996066898107529, + "epoch": 2.01338376491126, + "flow_matching_loss": 0.01652241563424468, + "grad_norm": 0.08561937510967255, + "kept_token_count": 573.13203125, + "learning_rate": 2.7402489856660528e-05, + "loss": 0.0228, + "mgd_cos_sim": 0.9918581724166871, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008141817152500152, + "mgd_loss_weight": 0.7695089876651764, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 359.82024536132815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9000579833984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6920, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.29956311285495757, + "epoch": 2.016293279022403, + "flow_matching_loss": 0.014094860292971134, + "grad_norm": 0.08951818197965622, + "kept_token_count": 573.17890625, + "learning_rate": 2.7393182415461307e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9920546352863312, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007945359777659177, + "mgd_loss_weight": 0.7691756367683411, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.5703430175781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.4798065185547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6930, + "valid_token_count": 818.31796875 + }, + { + "actual_token_mask_ratio": 0.29968092143535613, + "epoch": 2.019202793133547, + "flow_matching_loss": 0.01568352961912751, + "grad_norm": 0.08904808759689331, + "kept_token_count": 573.16328125, + "learning_rate": 2.7383859915376853e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9921842336654663, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007815761072561146, + "mgd_loss_weight": 0.7688422977924347, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 365.1634033203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.2108459472656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6940, + "valid_token_count": 818.4296875 + }, + { + "actual_token_mask_ratio": 0.29982973337173463, + "epoch": 2.02211230724469, + "flow_matching_loss": 0.016026033367961646, + "grad_norm": 0.11494824290275574, + "kept_token_count": 572.8578125, + "learning_rate": 2.7374522367734875e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9919278085231781, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008072195993736386, + "mgd_loss_weight": 0.7685089528560638, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.42723388671874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.33595275878906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6950, + "valid_token_count": 818.17109375 + }, + { + "actual_token_mask_ratio": 0.2995250165462494, + "epoch": 2.0250218213558338, + "flow_matching_loss": 0.015080900397151709, + "grad_norm": 0.08514682948589325, + "kept_token_count": 573.01171875, + "learning_rate": 2.7365169783881378e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9920557200908661, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007944291550666093, + "mgd_loss_weight": 0.7681756019592285, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.18291931152345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.0053192138672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6960, + "valid_token_count": 818.02734375 + }, + { + "actual_token_mask_ratio": 0.30017911195755004, + "epoch": 2.027931335466977, + "flow_matching_loss": 0.013839408196508885, + "grad_norm": 0.10357452183961868, + "kept_token_count": 572.6859375, + "learning_rate": 2.735580217518064e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9924825847148895, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007517413794994354, + "mgd_loss_weight": 0.7678422629833221, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.9557250976562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.9848236083984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6970, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.2996084809303284, + "epoch": 2.0308408495781203, + "flow_matching_loss": 0.016515038069337608, + "grad_norm": 0.10612986236810684, + "kept_token_count": 572.965625, + "learning_rate": 2.7346419553015175e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9924172282218933, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007582784863188863, + "mgd_loss_weight": 0.7675089180469513, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.8123260498047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4649658203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6980, + "valid_token_count": 818.05859375 + }, + { + "actual_token_mask_ratio": 0.30041839480400084, + "epoch": 2.033750363689264, + "flow_matching_loss": 0.016569180972874166, + "grad_norm": 0.10300998389720917, + "kept_token_count": 572.35703125, + "learning_rate": 2.733702192878577e-05, + "loss": 0.0224, + "mgd_cos_sim": 0.9923596858978272, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007640303391963243, + "mgd_loss_weight": 0.7671755731105805, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.83489685058595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.66539306640624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 6990, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.2995859682559967, + "epoch": 2.0366598778004072, + "flow_matching_loss": 0.014173614885658025, + "grad_norm": 0.10377665609121323, + "kept_token_count": 572.984375, + "learning_rate": 2.732760931391141e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.99209566116333, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007904339162632824, + "mgd_loss_weight": 0.7668422281742096, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.77215270996095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.957177734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7000, + "valid_token_count": 818.06328125 + }, + { + "actual_token_mask_ratio": 0.300256210565567, + "epoch": 2.039569391911551, + "flow_matching_loss": 0.015514685865491628, + "grad_norm": 0.09043513238430023, + "kept_token_count": 572.46796875, + "learning_rate": 2.731818171982932e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9920433104038239, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007956691598519684, + "mgd_loss_weight": 0.7665088832378387, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.74415283203126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.9019836425781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7010, + "valid_token_count": 818.11484375 + }, + { + "actual_token_mask_ratio": 0.300428694486618, + "epoch": 2.042478906022694, + "flow_matching_loss": 0.014571808092296124, + "grad_norm": 0.11063719540834427, + "kept_token_count": 572.41640625, + "learning_rate": 2.730873915799491e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.992031705379486, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007968279905617237, + "mgd_loss_weight": 0.7661755383014679, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.62982177734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.2886474609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7020, + "valid_token_count": 818.23984375 + }, + { + "actual_token_mask_ratio": 0.2997751235961914, + "epoch": 2.0453884201338375, + "flow_matching_loss": 0.015826235897839068, + "grad_norm": 0.09663024544715881, + "kept_token_count": 572.90078125, + "learning_rate": 2.7299281639881783e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9919332444667817, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008066750690340995, + "mgd_loss_weight": 0.7658421993255615, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.5549377441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.1066589355469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7030, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.3001063525676727, + "epoch": 2.048297934244981, + "flow_matching_loss": 0.013267173152416944, + "grad_norm": 0.09649913758039474, + "kept_token_count": 572.94453125, + "learning_rate": 2.728980917698172e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.991849809885025, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008150189183652401, + "mgd_loss_weight": 0.7655088484287262, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.19703369140626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.70693359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7040, + "valid_token_count": 818.61484375 + }, + { + "actual_token_mask_ratio": 0.29998749792575835, + "epoch": 2.0512074483561245, + "flow_matching_loss": 0.014452913589775563, + "grad_norm": 0.11399827897548676, + "kept_token_count": 572.6515625, + "learning_rate": 2.7280321780804655e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9920019388198853, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007998082973062991, + "mgd_loss_weight": 0.7651755034923553, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.44239501953126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.7446746826172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7050, + "valid_token_count": 818.05859375 + }, + { + "actual_token_mask_ratio": 0.2999880850315094, + "epoch": 2.054116962467268, + "flow_matching_loss": 0.015673591196537016, + "grad_norm": 0.0859922245144844, + "kept_token_count": 572.603125, + "learning_rate": 2.7270819462878666e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9920639455318451, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0079360646661371, + "mgd_loss_weight": 0.7648421645164489, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.5213134765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.0998596191406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7060, + "valid_token_count": 817.990625 + }, + { + "actual_token_mask_ratio": 0.2997849524021149, + "epoch": 2.0570264765784114, + "flow_matching_loss": 0.014723752345889806, + "grad_norm": 0.0953977033495903, + "kept_token_count": 573.15, + "learning_rate": 2.7261302234749965e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9924174249172211, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007582575269043446, + "mgd_loss_weight": 0.7645088136196136, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.32798461914064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.92608337402345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7070, + "valid_token_count": 818.5359375 + }, + { + "actual_token_mask_ratio": 0.2994776487350464, + "epoch": 2.0599359906895547, + "flow_matching_loss": 0.01418690076097846, + "grad_norm": 0.10104382783174515, + "kept_token_count": 573.14453125, + "learning_rate": 2.7251770107982888e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9920768439769745, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007923146849498153, + "mgd_loss_weight": 0.7641754746437073, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.4032806396484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.1638488769531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7080, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.29952473044395445, + "epoch": 2.0628455048006984, + "flow_matching_loss": 0.015784689877182245, + "grad_norm": 0.07584217190742493, + "kept_token_count": 573.23125, + "learning_rate": 2.7242223094159856e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9919903099536895, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008009680127725006, + "mgd_loss_weight": 0.7638421297073364, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.9824981689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.9019256591797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7090, + "valid_token_count": 818.34609375 + }, + { + "actual_token_mask_ratio": 0.3007632613182068, + "epoch": 2.0657550189118417, + "flow_matching_loss": 0.015257808286696673, + "grad_norm": 0.0935099869966507, + "kept_token_count": 572.18671875, + "learning_rate": 2.7232661204881408e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9916947901248931, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008305211691185832, + "mgd_loss_weight": 0.763508778810501, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 373.7426818847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.8578674316406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7100, + "valid_token_count": 818.30078125 + }, + { + "actual_token_mask_ratio": 0.29869556427001953, + "epoch": 2.0686645330229854, + "flow_matching_loss": 0.013824912626296282, + "grad_norm": 0.08391663432121277, + "kept_token_count": 573.9234375, + "learning_rate": 2.7223084451766123e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9923645198345185, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007635468477383256, + "mgd_loss_weight": 0.7631754457950592, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.8027648925781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.9691558837891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7110, + "valid_token_count": 818.36796875 + }, + { + "actual_token_mask_ratio": 0.30002849996089936, + "epoch": 2.0715740471341286, + "flow_matching_loss": 0.015258599352091552, + "grad_norm": 0.0892895758152008, + "kept_token_count": 572.6, + "learning_rate": 2.721349284645067e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9921673238277435, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007832679618149995, + "mgd_loss_weight": 0.7628420889377594, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.15785522460936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.84364624023436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7120, + "valid_token_count": 818.034375 + }, + { + "actual_token_mask_ratio": 0.3008792042732239, + "epoch": 2.074483561245272, + "flow_matching_loss": 0.015829792153090237, + "grad_norm": 0.1059207171201706, + "kept_token_count": 572.01875, + "learning_rate": 2.7203886400589752e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9918218135833741, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00817819251678884, + "mgd_loss_weight": 0.762508749961853, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 372.2345886230469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.4953247070313, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7130, + "valid_token_count": 818.20078125 + }, + { + "actual_token_mask_ratio": 0.3002607673406601, + "epoch": 2.0773930753564156, + "flow_matching_loss": 0.015373618900775909, + "grad_norm": 0.11196639388799667, + "kept_token_count": 572.5328125, + "learning_rate": 2.7194265125856105e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9922443449497222, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077556479722261425, + "mgd_loss_weight": 0.7621754109859467, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.68670959472655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.099169921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7140, + "valid_token_count": 818.20859375 + }, + { + "actual_token_mask_ratio": 0.3005712449550629, + "epoch": 2.080302589467559, + "flow_matching_loss": 0.014974101260304452, + "grad_norm": 0.1275627613067627, + "kept_token_count": 572.3203125, + "learning_rate": 2.718462903394049e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9919681489467621, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008031851286068559, + "mgd_loss_weight": 0.7618420541286468, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.71055908203124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.88946228027345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7150, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.300291645526886, + "epoch": 2.0832121035787026, + "flow_matching_loss": 0.01588266370818019, + "grad_norm": 0.08664874732494354, + "kept_token_count": 572.646875, + "learning_rate": 2.7174978136551662e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9921958744525909, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007804127177223563, + "mgd_loss_weight": 0.761508721113205, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.93776245117186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 368.3463073730469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7160, + "valid_token_count": 818.40703125 + }, + { + "actual_token_mask_ratio": 0.29951109290122985, + "epoch": 2.086121617689846, + "flow_matching_loss": 0.01597888208925724, + "grad_norm": 0.14355845749378204, + "kept_token_count": 573.07890625, + "learning_rate": 2.716531244541637e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.991790109872818, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00820988779887557, + "mgd_loss_weight": 0.7611753702163696, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.68922119140626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.19134521484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7170, + "valid_token_count": 818.11015625 + }, + { + "actual_token_mask_ratio": 0.29963467717170716, + "epoch": 2.089031131800989, + "flow_matching_loss": 0.015045672748237848, + "grad_norm": 0.10700108855962753, + "kept_token_count": 573.05546875, + "learning_rate": 2.715563197227934e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9922665894031525, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00773341478779912, + "mgd_loss_weight": 0.7608420252799988, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.7764129638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.79990844726564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7180, + "valid_token_count": 818.21953125 + }, + { + "actual_token_mask_ratio": 0.29925971031188964, + "epoch": 2.091940645912133, + "flow_matching_loss": 0.014776889327913522, + "grad_norm": 0.11182619631290436, + "kept_token_count": 573.51171875, + "learning_rate": 2.714593672890327e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.992309308052063, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007690692786127329, + "mgd_loss_weight": 0.7605086863040924, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.2145935058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.0517211914063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7190, + "valid_token_count": 818.434375 + }, + { + "actual_token_mask_ratio": 0.3001337945461273, + "epoch": 2.094850160023276, + "flow_matching_loss": 0.0156501323916018, + "grad_norm": 0.09229038655757904, + "kept_token_count": 572.67109375, + "learning_rate": 2.7136226727068783e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9920534014701843, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007946601416915656, + "mgd_loss_weight": 0.7601753354072571, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.60284729003905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.3390686035156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7200, + "valid_token_count": 818.25859375 + }, + { + "actual_token_mask_ratio": 0.29975616931915283, + "epoch": 2.0977596741344193, + "flow_matching_loss": 0.0138353293761611, + "grad_norm": 0.09992379695177078, + "kept_token_count": 572.91953125, + "learning_rate": 2.712650197857445e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.992338502407074, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007661495357751846, + "mgd_loss_weight": 0.7598419964313508, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.6046478271484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.98675231933595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7210, + "valid_token_count": 818.16953125 + }, + { + "actual_token_mask_ratio": 0.30021314024925233, + "epoch": 2.100669188245563, + "flow_matching_loss": 0.015565774030983448, + "grad_norm": 0.0941743403673172, + "kept_token_count": 572.58359375, + "learning_rate": 2.7116762495236763e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9921563267707825, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007843679748475552, + "mgd_loss_weight": 0.7595086514949798, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.4248809814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.768212890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7220, + "valid_token_count": 818.23203125 + }, + { + "actual_token_mask_ratio": 0.30096371471881866, + "epoch": 2.1035787023567063, + "flow_matching_loss": 0.014097403641790152, + "grad_norm": 0.08607631921768188, + "kept_token_count": 571.93046875, + "learning_rate": 2.7107008288890106e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9923444569110871, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007655541552230716, + "mgd_loss_weight": 0.7591753005981445, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.18512573242185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.5863433837891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7230, + "valid_token_count": 818.17265625 + }, + { + "actual_token_mask_ratio": 0.3002577364444733, + "epoch": 2.10648821646785, + "flow_matching_loss": 0.014395893644541502, + "grad_norm": 0.11736869812011719, + "kept_token_count": 572.6265625, + "learning_rate": 2.709723937138676e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9920769691467285, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007923036022111774, + "mgd_loss_weight": 0.7588419616222382, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.75214233398435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.14609680175784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7240, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.2997592866420746, + "epoch": 2.1093977305789933, + "flow_matching_loss": 0.014884146861732006, + "grad_norm": 0.07281100749969482, + "kept_token_count": 573.05546875, + "learning_rate": 2.7087455754596883e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9919352114200592, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008064784668385982, + "mgd_loss_weight": 0.7585086166858673, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.08477478027345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.3196685791016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7250, + "valid_token_count": 818.365625 + }, + { + "actual_token_mask_ratio": 0.30006169676780703, + "epoch": 2.112307244690137, + "flow_matching_loss": 0.014502557553350925, + "grad_norm": 0.09221065044403076, + "kept_token_count": 572.80703125, + "learning_rate": 2.7077657450408498e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9922467947006226, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007753198593854904, + "mgd_loss_weight": 0.7581752717494965, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.68675537109374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.561181640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7260, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.3000188499689102, + "epoch": 2.1152167588012802, + "flow_matching_loss": 0.014796335995197297, + "grad_norm": 0.0893847718834877, + "kept_token_count": 572.71875, + "learning_rate": 2.7067844470727456e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9922089874744415, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007791022351011634, + "mgd_loss_weight": 0.7578419268131256, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 375.2156616210938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.1730010986328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7270, + "valid_token_count": 818.19140625 + }, + { + "actual_token_mask_ratio": 0.2997644513845444, + "epoch": 2.1181262729124235, + "flow_matching_loss": 0.017001976259052753, + "grad_norm": 0.0951329916715622, + "kept_token_count": 572.80546875, + "learning_rate": 2.7058016827477455e-05, + "loss": 0.0231, + "mgd_cos_sim": 0.9919393002986908, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008060709526762366, + "mgd_loss_weight": 0.7575085818767547, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.38242797851564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.25640563964845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7280, + "valid_token_count": 818.01796875 + }, + { + "actual_token_mask_ratio": 0.2995888352394104, + "epoch": 2.121035787023567, + "flow_matching_loss": 0.014332551043480635, + "grad_norm": 0.10149599611759186, + "kept_token_count": 573.22109375, + "learning_rate": 2.7048174532600018e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9920996248722076, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007900379318743943, + "mgd_loss_weight": 0.7571752369403839, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.7835296630859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.2182281494141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7290, + "valid_token_count": 818.40859375 + }, + { + "actual_token_mask_ratio": 0.29888308942317965, + "epoch": 2.1239453011347105, + "flow_matching_loss": 0.01466513415798545, + "grad_norm": 0.09145834296941757, + "kept_token_count": 573.56953125, + "learning_rate": 2.7038317598054455e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9923948824405671, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007605130970478058, + "mgd_loss_weight": 0.7568418979644775, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.13921813964845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.2312286376953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7300, + "valid_token_count": 818.0796875 + }, + { + "actual_token_mask_ratio": 0.30037994384765626, + "epoch": 2.1268548152458537, + "flow_matching_loss": 0.014534266758710146, + "grad_norm": 0.09214992076158524, + "kept_token_count": 572.43515625, + "learning_rate": 2.702844603581787e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9923745334148407, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007625466864556074, + "mgd_loss_weight": 0.7565085530281067, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.895849609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 367.93433837890626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7310, + "valid_token_count": 818.21015625 + }, + { + "actual_token_mask_ratio": 0.3000977963209152, + "epoch": 2.1297643293569974, + "flow_matching_loss": 0.014919448643922806, + "grad_norm": 0.10692156851291656, + "kept_token_count": 572.634375, + "learning_rate": 2.701855985788514e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9923184096813202, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007681597955524921, + "mgd_loss_weight": 0.7561752021312713, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.0967254638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.66590576171876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7320, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.2997122436761856, + "epoch": 2.1326738434681407, + "flow_matching_loss": 0.01489013684913516, + "grad_norm": 0.11968663334846497, + "kept_token_count": 573.0875, + "learning_rate": 2.7008659076268912e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9918714582920074, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008128535747528077, + "mgd_loss_weight": 0.7558418691158295, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.7671417236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.2068725585938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7330, + "valid_token_count": 818.36328125 + }, + { + "actual_token_mask_ratio": 0.29990184009075166, + "epoch": 2.1355833575792844, + "flow_matching_loss": 0.015520717948675156, + "grad_norm": 0.08122032880783081, + "kept_token_count": 572.96015625, + "learning_rate": 2.6998743702999562e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.991987431049347, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008012562012299895, + "mgd_loss_weight": 0.7555085122585297, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.72853088378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.4343231201172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7340, + "valid_token_count": 818.39765625 + }, + { + "actual_token_mask_ratio": 0.29935296475887296, + "epoch": 2.1384928716904277, + "flow_matching_loss": 0.014649507589638233, + "grad_norm": 0.09189269691705704, + "kept_token_count": 573.35859375, + "learning_rate": 2.6988813750125202e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9922461092472077, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007753901509568095, + "mgd_loss_weight": 0.7551751732826233, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.8681915283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.52435607910155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7350, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.30032113194465637, + "epoch": 2.141402385801571, + "flow_matching_loss": 0.015194765850901603, + "grad_norm": 0.09816080331802368, + "kept_token_count": 572.76015625, + "learning_rate": 2.6978869229711668e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9922127068042755, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007787293754518032, + "mgd_loss_weight": 0.7548418343067169, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.6366027832031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.37372131347655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7360, + "valid_token_count": 818.60546875 + }, + { + "actual_token_mask_ratio": 0.299836078286171, + "epoch": 2.1443118999127146, + "flow_matching_loss": 0.015267386008054018, + "grad_norm": 0.10321148484945297, + "kept_token_count": 572.92734375, + "learning_rate": 2.6968910153842484e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.9921765685081482, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007823421061038971, + "mgd_loss_weight": 0.7545084774494171, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 368.98138732910155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.7373901367188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7370, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.2998746633529663, + "epoch": 2.147221414023858, + "flow_matching_loss": 0.014671179745346308, + "grad_norm": 0.07682063430547714, + "kept_token_count": 572.9859375, + "learning_rate": 2.6958936534618874e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9927421092987061, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00725789749994874, + "mgd_loss_weight": 0.7541751444339753, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.9897003173828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.6607604980469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7380, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.29987613558769227, + "epoch": 2.1501309281350016, + "flow_matching_loss": 0.015098744072020054, + "grad_norm": 0.1385166496038437, + "kept_token_count": 572.99140625, + "learning_rate": 2.694894838415971e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9920845448970794, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00791545188985765, + "mgd_loss_weight": 0.7538417935371399, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 364.56220092773435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.54099426269534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7390, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.300244602560997, + "epoch": 2.153040442246145, + "flow_matching_loss": 0.014759261347353458, + "grad_norm": 0.08870886266231537, + "kept_token_count": 572.56171875, + "learning_rate": 2.6938945714601555e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9924822926521302, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007517704972997308, + "mgd_loss_weight": 0.753508448600769, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.7652954101562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.3205841064453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7400, + "valid_token_count": 818.22578125 + }, + { + "actual_token_mask_ratio": 0.30016150772571565, + "epoch": 2.155949956357288, + "flow_matching_loss": 0.014145920984447002, + "grad_norm": 0.08621267229318619, + "kept_token_count": 572.80859375, + "learning_rate": 2.6928928538098587e-05, + "loss": 0.02, + "mgd_cos_sim": 0.992290872335434, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007709127385169268, + "mgd_loss_weight": 0.7531751096248627, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.8912658691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.0345886230469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7410, + "valid_token_count": 818.4875 + }, + { + "actual_token_mask_ratio": 0.2994467556476593, + "epoch": 2.158859470468432, + "flow_matching_loss": 0.016036611516028643, + "grad_norm": 0.0857425257563591, + "kept_token_count": 573.25859375, + "learning_rate": 2.6918896866822615e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.99174884557724, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008251160476356744, + "mgd_loss_weight": 0.7528417587280274, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.1620635986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.37981872558595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7420, + "valid_token_count": 818.29765625 + }, + { + "actual_token_mask_ratio": 0.29985071122646334, + "epoch": 2.161768984579575, + "flow_matching_loss": 0.014178232662379742, + "grad_norm": 0.08155979216098785, + "kept_token_count": 572.946875, + "learning_rate": 2.690885071296308e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9922961056232452, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007703904574736953, + "mgd_loss_weight": 0.752508419752121, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.67650146484374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.7318634033203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7430, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.30026977956295015, + "epoch": 2.164678498690719, + "flow_matching_loss": 0.015746762044727802, + "grad_norm": 0.09993217885494232, + "kept_token_count": 572.5875, + "learning_rate": 2.6898790088726993e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9920604348182678, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007939554331824184, + "mgd_loss_weight": 0.7521750748157501, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.86977233886716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.7257446289062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7440, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.3003627359867096, + "epoch": 2.167588012801862, + "flow_matching_loss": 0.015199992153793574, + "grad_norm": 0.09930087625980377, + "kept_token_count": 572.58984375, + "learning_rate": 2.6888715006338968e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9921203255653381, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007879668287932873, + "mgd_loss_weight": 0.7518417239189148, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.3622161865234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.50396118164065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7450, + "valid_token_count": 818.40546875 + }, + { + "actual_token_mask_ratio": 0.30031628012657163, + "epoch": 2.1704975269130053, + "flow_matching_loss": 0.015185787808150053, + "grad_norm": 0.09044661372900009, + "kept_token_count": 572.54375, + "learning_rate": 2.6878625478041172e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9923155784606934, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076844336465001105, + "mgd_loss_weight": 0.7515083849430084, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.62830810546876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.91054382324216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7460, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.2998490512371063, + "epoch": 2.173407041024149, + "flow_matching_loss": 0.017373015452176333, + "grad_norm": 0.11877970397472382, + "kept_token_count": 572.83828125, + "learning_rate": 2.6868521516093342e-05, + "loss": 0.0233, + "mgd_cos_sim": 0.992114931344986, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007885054545477032, + "mgd_loss_weight": 0.7511750400066376, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.0157897949219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.489208984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7470, + "valid_token_count": 818.16171875 + }, + { + "actual_token_mask_ratio": 0.30013464093208314, + "epoch": 2.1763165551352923, + "flow_matching_loss": 0.013230347353965045, + "grad_norm": 0.09889830648899078, + "kept_token_count": 572.846875, + "learning_rate": 2.6858403132772745e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9922838032245636, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077161898370832205, + "mgd_loss_weight": 0.7508416950702668, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.4028289794922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.35675048828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7480, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.30059888362884524, + "epoch": 2.179226069246436, + "flow_matching_loss": 0.015305565670132637, + "grad_norm": 0.09179942309856415, + "kept_token_count": 572.41171875, + "learning_rate": 2.6848270340374162e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9924624562263489, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007537549315020442, + "mgd_loss_weight": 0.7505083501338958, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.0492858886719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.44495849609376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7490, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.29962055683135985, + "epoch": 2.1821355833575793, + "flow_matching_loss": 0.013508548866957426, + "grad_norm": 0.075155109167099, + "kept_token_count": 573.24609375, + "learning_rate": 2.68381231512099e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9923264920711518, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007673499453812837, + "mgd_loss_weight": 0.750175005197525, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.48186950683595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.5893951416016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7500, + "valid_token_count": 818.47734375 + }, + { + "actual_token_mask_ratio": 0.3003526240587234, + "epoch": 2.1850450974687226, + "flow_matching_loss": 0.015463802590966224, + "grad_norm": 0.09067162126302719, + "kept_token_count": 572.44140625, + "learning_rate": 2.6827961577609745e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9925006031990051, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007499391119927168, + "mgd_loss_weight": 0.7498416602611542, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.3732482910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.45941467285155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7510, + "valid_token_count": 818.18515625 + }, + { + "actual_token_mask_ratio": 0.2989313304424286, + "epoch": 2.1879546115798663, + "flow_matching_loss": 0.014781611878424883, + "grad_norm": 0.08098921179771423, + "kept_token_count": 573.803125, + "learning_rate": 2.681778563192097e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9922054290771485, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007794568873941898, + "mgd_loss_weight": 0.7495083212852478, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.79462280273435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.14488525390624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7520, + "valid_token_count": 818.46640625 + }, + { + "actual_token_mask_ratio": 0.30001010894775393, + "epoch": 2.1908641256910095, + "flow_matching_loss": 0.014992263820022344, + "grad_norm": 0.09960509836673737, + "kept_token_count": 572.86171875, + "learning_rate": 2.6807595326508308e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9929725587368011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007027440378442407, + "mgd_loss_weight": 0.7491749703884125, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.3010284423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.1800567626953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7530, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.2999895215034485, + "epoch": 2.1937736398021532, + "flow_matching_loss": 0.01597170261666179, + "grad_norm": 0.08180137723684311, + "kept_token_count": 572.7390625, + "learning_rate": 2.679739067375394e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9923788726329803, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076211303472518924, + "mgd_loss_weight": 0.7488416254520416, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.42137756347654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.8879333496094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7540, + "valid_token_count": 818.18671875 + }, + { + "actual_token_mask_ratio": 0.29935274720191957, + "epoch": 2.1966831539132965, + "flow_matching_loss": 0.016304312273859978, + "grad_norm": 0.08216790109872818, + "kept_token_count": 573.3625, + "learning_rate": 2.6787171686057487e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9926069855690003, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073930316139012575, + "mgd_loss_weight": 0.7485082864761352, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.5288879394531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.70740661621096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7550, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.30072730779647827, + "epoch": 2.1995926680244398, + "flow_matching_loss": 0.01565131600946188, + "grad_norm": 0.0852532759308815, + "kept_token_count": 572.01328125, + "learning_rate": 2.677693837583597e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9918961524963379, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008103837352246046, + "mgd_loss_weight": 0.7481749355792999, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.1709350585937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.7455230712891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7560, + "valid_token_count": 818.00625 + }, + { + "actual_token_mask_ratio": 0.2999852210283279, + "epoch": 2.2025021821355835, + "flow_matching_loss": 0.01398888062685728, + "grad_norm": 0.08492271602153778, + "kept_token_count": 572.99453125, + "learning_rate": 2.6766690755523836e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9921111226081848, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007888872781768442, + "mgd_loss_weight": 0.7478415966033936, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.6814727783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.59807739257815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7570, + "valid_token_count": 818.54921875 + }, + { + "actual_token_mask_ratio": 0.30042511522769927, + "epoch": 2.2054116962467267, + "flow_matching_loss": 0.014160724729299546, + "grad_norm": 0.10186995565891266, + "kept_token_count": 572.31796875, + "learning_rate": 2.675642883757291e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9921041846275329, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007895820122212172, + "mgd_loss_weight": 0.7475082516670227, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.7769500732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.34473571777346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7580, + "valid_token_count": 818.0984375 + }, + { + "actual_token_mask_ratio": 0.2994176834821701, + "epoch": 2.2083212103578704, + "flow_matching_loss": 0.015471730846911669, + "grad_norm": 0.11531498283147812, + "kept_token_count": 573.215625, + "learning_rate": 2.674615263445239e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.991622656583786, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008377345791086555, + "mgd_loss_weight": 0.7471749007701873, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.6171844482422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3502563476562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7590, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.2992372095584869, + "epoch": 2.2112307244690137, + "flow_matching_loss": 0.016394510213285685, + "grad_norm": 0.08775721490383148, + "kept_token_count": 573.39921875, + "learning_rate": 2.673586215864882e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9923191487789154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007680837344378233, + "mgd_loss_weight": 0.7468415677547455, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 371.05161743164064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.2506896972656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7600, + "valid_token_count": 818.25234375 + }, + { + "actual_token_mask_ratio": 0.2999697685241699, + "epoch": 2.214140238580157, + "flow_matching_loss": 0.014564447477459908, + "grad_norm": 0.1149292066693306, + "kept_token_count": 572.69921875, + "learning_rate": 2.6725557422666113e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9923742592334748, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007625748356804252, + "mgd_loss_weight": 0.7465082108974457, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.3771453857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.91847229003906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7610, + "valid_token_count": 818.11171875 + }, + { + "actual_token_mask_ratio": 0.3009737253189087, + "epoch": 2.2170497526913007, + "flow_matching_loss": 0.014687269367277623, + "grad_norm": 0.09562471508979797, + "kept_token_count": 572.15625, + "learning_rate": 2.6715238439025493e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.992316049337387, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007683956529945135, + "mgd_loss_weight": 0.7461748719215393, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 376.3498809814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.5189666748047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7620, + "valid_token_count": 818.5 + }, + { + "actual_token_mask_ratio": 0.3000687718391418, + "epoch": 2.219959266802444, + "flow_matching_loss": 0.015517340321093798, + "grad_norm": 0.10713137686252594, + "kept_token_count": 572.55, + "learning_rate": 2.6704905220265493e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.992025226354599, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007974781980738043, + "mgd_loss_weight": 0.7458415329456329, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.79901733398435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.48362121582034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7630, + "valid_token_count": 818.00625 + }, + { + "actual_token_mask_ratio": 0.3000097513198853, + "epoch": 2.2228687809135876, + "flow_matching_loss": 0.015614028554409742, + "grad_norm": 0.09004797786474228, + "kept_token_count": 572.7546875, + "learning_rate": 2.6694557778941947e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9923963367938995, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007603664556518197, + "mgd_loss_weight": 0.7455081760883331, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.8014709472656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.8131896972656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7640, + "valid_token_count": 818.23359375 + }, + { + "actual_token_mask_ratio": 0.2997804582118988, + "epoch": 2.225778295024731, + "flow_matching_loss": 0.01563910795375705, + "grad_norm": 0.10098987817764282, + "kept_token_count": 573.015625, + "learning_rate": 2.6684196127627977e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9924581229686738, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007541865669190883, + "mgd_loss_weight": 0.7451748430728913, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.9503509521484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.6036682128906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7650, + "valid_token_count": 818.33828125 + }, + { + "actual_token_mask_ratio": 0.30045506954193113, + "epoch": 2.228687809135874, + "flow_matching_loss": 0.014152772258967161, + "grad_norm": 0.08152799308300018, + "kept_token_count": 572.425, + "learning_rate": 2.6673820278913963e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9914375364780426, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0085624685510993, + "mgd_loss_weight": 0.7448414921760559, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.1185791015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.79388122558595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7660, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.30001037418842313, + "epoch": 2.231597323247018, + "flow_matching_loss": 0.01431937413290143, + "grad_norm": 0.0834319144487381, + "kept_token_count": 572.6921875, + "learning_rate": 2.6663430245407544e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9921963691711426, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007803627522662282, + "mgd_loss_weight": 0.7445081472396851, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.8082580566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.3697570800781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7670, + "valid_token_count": 818.13828125 + }, + { + "actual_token_mask_ratio": 0.30017189085483553, + "epoch": 2.234506837358161, + "flow_matching_loss": 0.016108863148838283, + "grad_norm": 0.10205643624067307, + "kept_token_count": 572.67109375, + "learning_rate": 2.6653026039733585e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9922488451004028, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007751144468784332, + "mgd_loss_weight": 0.7441748082637787, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.99085998535156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.5439910888672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7680, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.30001142621040344, + "epoch": 2.237416351469305, + "flow_matching_loss": 0.01541302390396595, + "grad_norm": 0.10887663066387177, + "kept_token_count": 572.80234375, + "learning_rate": 2.6642607674534176e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9921069860458374, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007893006689846516, + "mgd_loss_weight": 0.7438414573669434, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.97841186523436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.70672912597655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7690, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.30015115439891815, + "epoch": 2.240325865580448, + "flow_matching_loss": 0.01648842403665185, + "grad_norm": 0.08165071904659271, + "kept_token_count": 572.74765625, + "learning_rate": 2.663217516246862e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9921326518058777, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007867331197485328, + "mgd_loss_weight": 0.743508118391037, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.5792663574219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 369.9552917480469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7700, + "valid_token_count": 818.38984375 + }, + { + "actual_token_mask_ratio": 0.30007520616054534, + "epoch": 2.2432353796915914, + "flow_matching_loss": 0.014627627935260535, + "grad_norm": 0.10469039529561996, + "kept_token_count": 572.81328125, + "learning_rate": 2.66217285162134e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9923535287380219, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007646481459960342, + "mgd_loss_weight": 0.7431747734546661, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.09300842285154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.69305114746095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7710, + "valid_token_count": 818.38671875 + }, + { + "actual_token_mask_ratio": 0.3000294864177704, + "epoch": 2.246144893802735, + "flow_matching_loss": 0.015885769668966532, + "grad_norm": 0.09730734676122665, + "kept_token_count": 572.815625, + "learning_rate": 2.6611267748462175e-05, + "loss": 0.0219, + "mgd_cos_sim": 0.9919040203094482, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008095977641642094, + "mgd_loss_weight": 0.7428414225578308, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.6574432373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.88593139648435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7720, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.299596506357193, + "epoch": 2.2490544079138783, + "flow_matching_loss": 0.01447759298607707, + "grad_norm": 0.10076918452978134, + "kept_token_count": 573.09453125, + "learning_rate": 2.6600792871925764e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9919143080711365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008085698774084448, + "mgd_loss_weight": 0.7425080835819244, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.8908966064453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.46519775390624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7730, + "valid_token_count": 818.24140625 + }, + { + "actual_token_mask_ratio": 0.2996433824300766, + "epoch": 2.2519639220250216, + "flow_matching_loss": 0.014223660714924336, + "grad_norm": 0.10532067716121674, + "kept_token_count": 572.91875, + "learning_rate": 2.659030389933213e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9921317219734191, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007868261076509953, + "mgd_loss_weight": 0.7421747446060181, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.3074127197266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.7117126464844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7740, + "valid_token_count": 818.03828125 + }, + { + "actual_token_mask_ratio": 0.3001820296049118, + "epoch": 2.2548734361361653, + "flow_matching_loss": 0.016231299564242362, + "grad_norm": 0.11149327456951141, + "kept_token_count": 572.56015625, + "learning_rate": 2.6579800843426358e-05, + "loss": 0.022, + "mgd_cos_sim": 0.9922596454620362, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007740362919867039, + "mgd_loss_weight": 0.7418413937091828, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.86199951171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.6975860595703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7750, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.299512654542923, + "epoch": 2.2577829502473086, + "flow_matching_loss": 0.015043954458087682, + "grad_norm": 0.09755193442106247, + "kept_token_count": 573.21796875, + "learning_rate": 2.656928371697067e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9924224317073822, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075775633566081526, + "mgd_loss_weight": 0.7415080487728118, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.5204345703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.17987060546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7760, + "valid_token_count": 818.31796875 + }, + { + "actual_token_mask_ratio": 0.3007637977600098, + "epoch": 2.2606924643584523, + "flow_matching_loss": 0.015215475670993328, + "grad_norm": 0.11041295528411865, + "kept_token_count": 571.90625, + "learning_rate": 2.6558752532744346e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9922586679458618, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007741330843418837, + "mgd_loss_weight": 0.7411747097969055, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.3538421630859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.1736175537109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7770, + "valid_token_count": 817.9046875 + }, + { + "actual_token_mask_ratio": 0.3003259301185608, + "epoch": 2.2636019784695955, + "flow_matching_loss": 0.01707569379359484, + "grad_norm": 0.09691999107599258, + "kept_token_count": 572.69375, + "learning_rate": 2.654820730354377e-05, + "loss": 0.0226, + "mgd_cos_sim": 0.9924947917461395, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007505219057202339, + "mgd_loss_weight": 0.7408413589000702, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.9660339355469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.23048400878906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7780, + "valid_token_count": 818.5109375 + }, + { + "actual_token_mask_ratio": 0.29971868097782134, + "epoch": 2.2665114925807393, + "flow_matching_loss": 0.016299881879240273, + "grad_norm": 0.09220912307500839, + "kept_token_count": 573.08046875, + "learning_rate": 2.6537648042182406e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9919543325901031, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00804567662999034, + "mgd_loss_weight": 0.7405080199241638, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.43426208496095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9043701171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7790, + "valid_token_count": 818.35859375 + }, + { + "actual_token_mask_ratio": 0.29938663840293883, + "epoch": 2.2694210066918825, + "flow_matching_loss": 0.014551330730319023, + "grad_norm": 0.09142190963029861, + "kept_token_count": 573.5296875, + "learning_rate": 2.6527074761490735e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9920404434204102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007959564263001085, + "mgd_loss_weight": 0.740174674987793, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.0219451904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 370.8763763427734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7800, + "valid_token_count": 818.615625 + }, + { + "actual_token_mask_ratio": 0.3006227910518646, + "epoch": 2.2723305208030258, + "flow_matching_loss": 0.01487332321703434, + "grad_norm": 0.076044462621212, + "kept_token_count": 572.32265625, + "learning_rate": 2.6516487474316303e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9919369459152222, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00806306367740035, + "mgd_loss_weight": 0.7398413240909576, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.3821624755859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.4863739013672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7810, + "valid_token_count": 818.32890625 + }, + { + "actual_token_mask_ratio": 0.29980830252170565, + "epoch": 2.2752400349141695, + "flow_matching_loss": 0.01428774856030941, + "grad_norm": 0.07704808562994003, + "kept_token_count": 573.021875, + "learning_rate": 2.650588619352365e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9922645509243011, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007735444651916623, + "mgd_loss_weight": 0.7395079910755158, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.1820526123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4337463378906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7820, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.29986820220947263, + "epoch": 2.2781495490253127, + "flow_matching_loss": 0.014517014101147651, + "grad_norm": 0.09700863063335419, + "kept_token_count": 573.0453125, + "learning_rate": 2.649527093199434e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9919634878635406, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008036507479846477, + "mgd_loss_weight": 0.7391746342182159, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.0277465820312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.9340576171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7830, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.3004374414682388, + "epoch": 2.281059063136456, + "flow_matching_loss": 0.013265125919133425, + "grad_norm": 0.0816044807434082, + "kept_token_count": 572.5171875, + "learning_rate": 2.6484641702626914e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9922527194023132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077472816221415995, + "mgd_loss_weight": 0.7388412952423096, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.97621154785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.9786712646484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7840, + "valid_token_count": 818.39765625 + }, + { + "actual_token_mask_ratio": 0.3001223742961884, + "epoch": 2.2839685772475997, + "flow_matching_loss": 0.014307865593582391, + "grad_norm": 0.09769066423177719, + "kept_token_count": 572.665625, + "learning_rate": 2.647399851833689e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9919603824615478, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008039611391723156, + "mgd_loss_weight": 0.7385079562664032, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.0723602294922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.2907470703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7850, + "valid_token_count": 818.23515625 + }, + { + "actual_token_mask_ratio": 0.2995551288127899, + "epoch": 2.286878091358743, + "flow_matching_loss": 0.015347109222784639, + "grad_norm": 0.10658468306064606, + "kept_token_count": 573.0, + "learning_rate": 2.6463341392056735e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9923121571540833, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007687847828492522, + "mgd_loss_weight": 0.7381745994091033, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.16143798828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.62880859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7860, + "valid_token_count": 818.05078125 + }, + { + "actual_token_mask_ratio": 0.3003811061382294, + "epoch": 2.2897876054698867, + "flow_matching_loss": 0.01389145916327834, + "grad_norm": 0.08654171973466873, + "kept_token_count": 572.33359375, + "learning_rate": 2.6452670336735867e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9924005925655365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007599404314532876, + "mgd_loss_weight": 0.7378412663936615, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.1007476806641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3325256347656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7870, + "valid_token_count": 818.06875 + }, + { + "actual_token_mask_ratio": 0.29985224902629853, + "epoch": 2.29269711958103, + "flow_matching_loss": 0.015525411441922187, + "grad_norm": 0.07885678112506866, + "kept_token_count": 572.9578125, + "learning_rate": 2.644198536534063e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9919189453125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008081052824854851, + "mgd_loss_weight": 0.7375079154968261, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.89231872558594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.6378692626953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7880, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.3001717984676361, + "epoch": 2.295606633692173, + "flow_matching_loss": 0.01574622355401516, + "grad_norm": 0.10215640813112259, + "kept_token_count": 572.465625, + "learning_rate": 2.6431286490854255e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.992205160856247, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007794832577928901, + "mgd_loss_weight": 0.7371745705604553, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 382.7238403320313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.5952850341797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7890, + "valid_token_count": 818.01171875 + }, + { + "actual_token_mask_ratio": 0.29986810088157656, + "epoch": 2.298516147803317, + "flow_matching_loss": 0.015933564864099026, + "grad_norm": 0.10182633250951767, + "kept_token_count": 572.86328125, + "learning_rate": 2.6420573726276903e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.992598158121109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007401833916082979, + "mgd_loss_weight": 0.736841231584549, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.8782684326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.3033813476562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7900, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.3001580148935318, + "epoch": 2.30142566191446, + "flow_matching_loss": 0.015877858828753234, + "grad_norm": 0.09821925312280655, + "kept_token_count": 572.6390625, + "learning_rate": 2.6409847084625575e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9924065887928009, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007593408320099115, + "mgd_loss_weight": 0.7365078806877137, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.2339752197266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1244750976563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7910, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.30007116198539735, + "epoch": 2.304335176025604, + "flow_matching_loss": 0.01492748586460948, + "grad_norm": 0.09989415109157562, + "kept_token_count": 572.83046875, + "learning_rate": 2.6399106578934167e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9921902894973755, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007809712691232562, + "mgd_loss_weight": 0.7361745417118073, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.31366577148435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.74281921386716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7920, + "valid_token_count": 818.40703125 + }, + { + "actual_token_mask_ratio": 0.30037887394428253, + "epoch": 2.307244690136747, + "flow_matching_loss": 0.016146164387464523, + "grad_norm": 0.10129109025001526, + "kept_token_count": 572.41328125, + "learning_rate": 2.63883522222534e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9923819839954376, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007618006831035018, + "mgd_loss_weight": 0.7358411967754364, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.4221893310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.10479125976565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7930, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.29983105361461637, + "epoch": 2.3101542042478904, + "flow_matching_loss": 0.014800666458904743, + "grad_norm": 0.09966105222702026, + "kept_token_count": 572.84296875, + "learning_rate": 2.6377584027650827e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9928201794624328, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007179810153320432, + "mgd_loss_weight": 0.7355078458786011, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.5616027832031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.4297271728516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7940, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.29968624413013456, + "epoch": 2.313063718359034, + "flow_matching_loss": 0.01574337938800454, + "grad_norm": 0.10390450805425644, + "kept_token_count": 573.02421875, + "learning_rate": 2.636680200821083e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9923179566860199, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007682044198736549, + "mgd_loss_weight": 0.7351745069026947, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.14667053222655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.96788635253904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7950, + "valid_token_count": 818.23671875 + }, + { + "actual_token_mask_ratio": 0.30086520314216614, + "epoch": 2.3159732324701774, + "flow_matching_loss": 0.01440771846100688, + "grad_norm": 0.10182882100343704, + "kept_token_count": 572.2484375, + "learning_rate": 2.635600617703457e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.992210853099823, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077891458291560415, + "mgd_loss_weight": 0.7348411619663239, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.4284393310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.7219970703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7960, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.30028477013111116, + "epoch": 2.318882746581321, + "flow_matching_loss": 0.014778417162597179, + "grad_norm": 0.07615428417921066, + "kept_token_count": 572.6359375, + "learning_rate": 2.6345196547240007e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9919544875621795, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008045518072322011, + "mgd_loss_weight": 0.734507817029953, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.3355316162109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.40386962890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7970, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.2999808520078659, + "epoch": 2.3217922606924644, + "flow_matching_loss": 0.015535592194646597, + "grad_norm": 0.11090660840272903, + "kept_token_count": 572.971875, + "learning_rate": 2.6334373131961854e-05, + "loss": 0.0212, + "mgd_cos_sim": 0.992267769575119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007732229679822922, + "mgd_loss_weight": 0.7341744720935821, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.1521850585938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8539245605469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7980, + "valid_token_count": 818.5046875 + }, + { + "actual_token_mask_ratio": 0.29976586997509, + "epoch": 2.3247017748036076, + "flow_matching_loss": 0.015003578085452319, + "grad_norm": 0.09629596024751663, + "kept_token_count": 573.1609375, + "learning_rate": 2.6323535944351583e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.992646050453186, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007353953830897808, + "mgd_loss_weight": 0.7338411271572113, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.255859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.33660278320315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 7990, + "valid_token_count": 818.52890625 + }, + { + "actual_token_mask_ratio": 0.3000535100698471, + "epoch": 2.3276112889147513, + "flow_matching_loss": 0.014203223027288913, + "grad_norm": 0.07831919193267822, + "kept_token_count": 572.7578125, + "learning_rate": 2.63126849975774e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9921181440353394, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00788186090067029, + "mgd_loss_weight": 0.7335077822208405, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.56866455078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.46759338378905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8000, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.3001464486122131, + "epoch": 2.3305208030258946, + "flow_matching_loss": 0.01489449255168438, + "grad_norm": 0.10983912646770477, + "kept_token_count": 572.59453125, + "learning_rate": 2.6301820304824234e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9918688595294952, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008131154533475637, + "mgd_loss_weight": 0.7331744432449341, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 380.0821594238281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.37624206542966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8010, + "valid_token_count": 818.15859375 + }, + { + "actual_token_mask_ratio": 0.2995018094778061, + "epoch": 2.3334303171370383, + "flow_matching_loss": 0.014364413172006606, + "grad_norm": 0.11185959726572037, + "kept_token_count": 573.33203125, + "learning_rate": 2.6290941879293703e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9921269357204437, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00787308462895453, + "mgd_loss_weight": 0.7328410923480988, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.45562133789065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.4445007324219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8020, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.2999140977859497, + "epoch": 2.3363398312481816, + "flow_matching_loss": 0.015634812787175177, + "grad_norm": 0.10471195727586746, + "kept_token_count": 572.671875, + "learning_rate": 2.628004973420412e-05, + "loss": 0.0213, + "mgd_cos_sim": 0.9922749161720276, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007725082943215967, + "mgd_loss_weight": 0.7325077474117279, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 361.07813415527346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.33876342773436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8030, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.29954820275306704, + "epoch": 2.339249345359325, + "flow_matching_loss": 0.01620014114305377, + "grad_norm": 0.08938848227262497, + "kept_token_count": 573.12578125, + "learning_rate": 2.626914388279047e-05, + "loss": 0.022, + "mgd_cos_sim": 0.992026150226593, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00797385722398758, + "mgd_loss_weight": 0.7321744084358215, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.54197692871094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.2896301269531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8040, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.2996304899454117, + "epoch": 2.3421588594704685, + "flow_matching_loss": 0.013805184047669172, + "grad_norm": 0.09921657294034958, + "kept_token_count": 572.99921875, + "learning_rate": 2.6258224338304394e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9922299563884736, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007770044403150678, + "mgd_loss_weight": 0.7318410575389862, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.1785369873047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.1756225585938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8050, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.3003892421722412, + "epoch": 2.345068373581612, + "flow_matching_loss": 0.01422469075769186, + "grad_norm": 0.11316291242837906, + "kept_token_count": 572.41875, + "learning_rate": 2.6247291114014166e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9920045614242554, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007995435688644647, + "mgd_loss_weight": 0.7315077185630798, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.9375946044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.14170227050784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8060, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.30077705085277556, + "epoch": 2.347977887692755, + "flow_matching_loss": 0.014558926783502102, + "grad_norm": 0.06572288274765015, + "kept_token_count": 572.003125, + "learning_rate": 2.6236344223204683e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9919381022453309, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008061907021328807, + "mgd_loss_weight": 0.731174373626709, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.83070068359376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9437286376953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8070, + "valid_token_count": 818.05390625 + }, + { + "actual_token_mask_ratio": 0.29998869001865386, + "epoch": 2.3508874018038988, + "flow_matching_loss": 0.015169471129775047, + "grad_norm": 0.08684328198432922, + "kept_token_count": 572.63359375, + "learning_rate": 2.6225383679177453e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9923684477806092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007631557248532772, + "mgd_loss_weight": 0.7308410227298736, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 374.1812042236328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.00914001464844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8080, + "valid_token_count": 818.03515625 + }, + { + "actual_token_mask_ratio": 0.30025573670864103, + "epoch": 2.353796915915042, + "flow_matching_loss": 0.014637199882417917, + "grad_norm": 0.11076239496469498, + "kept_token_count": 572.57265625, + "learning_rate": 2.6214409495250568e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9922067105770112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007793288445100188, + "mgd_loss_weight": 0.7305076897144318, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.8106292724609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.82097778320315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8090, + "valid_token_count": 818.26015625 + }, + { + "actual_token_mask_ratio": 0.2998734682798386, + "epoch": 2.3567064300261857, + "flow_matching_loss": 0.017843298241496088, + "grad_norm": 0.07978860288858414, + "kept_token_count": 572.99375, + "learning_rate": 2.620342168475869e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9923232674598694, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007676725601777434, + "mgd_loss_weight": 0.730174332857132, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.4613067626953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.24465026855466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8100, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.29954490065574646, + "epoch": 2.359615944137329, + "flow_matching_loss": 0.01624447200447321, + "grad_norm": 0.10385172069072723, + "kept_token_count": 573.01328125, + "learning_rate": 2.619242026105306e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9924381375312805, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007561855996027589, + "mgd_loss_weight": 0.7298409938812256, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.2184143066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1424865722656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8110, + "valid_token_count": 818.05234375 + }, + { + "actual_token_mask_ratio": 0.3002264738082886, + "epoch": 2.3625254582484727, + "flow_matching_loss": 0.014566758926957846, + "grad_norm": 0.09088371694087982, + "kept_token_count": 572.65859375, + "learning_rate": 2.618140523750143e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.992437607049942, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007562400866299867, + "mgd_loss_weight": 0.7295076549053192, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.4740905761719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3219268798828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8120, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.2992329806089401, + "epoch": 2.365434972359616, + "flow_matching_loss": 0.014980645850300789, + "grad_norm": 0.0806315466761589, + "kept_token_count": 573.44453125, + "learning_rate": 2.6170376627488103e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9924831390380859, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007516865571960807, + "mgd_loss_weight": 0.7291742980480194, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.5168090820313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.77012023925784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8130, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.299597230553627, + "epoch": 2.3683444864707592, + "flow_matching_loss": 0.015905466768890618, + "grad_norm": 0.10927560180425644, + "kept_token_count": 573.190625, + "learning_rate": 2.6159334444413873e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9920969843864441, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007903006672859193, + "mgd_loss_weight": 0.7288409650325776, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.9774505615234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.14698791503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8140, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.2991625517606735, + "epoch": 2.371254000581903, + "flow_matching_loss": 0.017892997339367868, + "grad_norm": 0.08804447948932648, + "kept_token_count": 573.57890625, + "learning_rate": 2.614827870169604e-05, + "loss": 0.0234, + "mgd_cos_sim": 0.9924614608287812, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007538538053631782, + "mgd_loss_weight": 0.7285076141357422, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.3373565673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.3765441894531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8150, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.3003173768520355, + "epoch": 2.374163514693046, + "flow_matching_loss": 0.013722914271056652, + "grad_norm": 0.0849420502781868, + "kept_token_count": 572.51796875, + "learning_rate": 2.6137209412768367e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9925297975540162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007470216788351536, + "mgd_loss_weight": 0.7281742691993713, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.2838592529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0717041015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8160, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.2999539911746979, + "epoch": 2.3770730288041895, + "flow_matching_loss": 0.016308766696602105, + "grad_norm": 0.1013740748167038, + "kept_token_count": 572.8015625, + "learning_rate": 2.612612659108109e-05, + "loss": 0.0218, + "mgd_cos_sim": 0.9923969209194183, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007603084901347756, + "mgd_loss_weight": 0.727840930223465, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.7407928466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.5471984863281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8170, + "valid_token_count": 818.23359375 + }, + { + "actual_token_mask_ratio": 0.3001424789428711, + "epoch": 2.379982542915333, + "flow_matching_loss": 0.0145704148337245, + "grad_norm": 0.07788285613059998, + "kept_token_count": 572.675, + "learning_rate": 2.611503025010088e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9924728691577911, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007527139922603965, + "mgd_loss_weight": 0.7275075852870941, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.98737182617185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.56015930175784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8180, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.29998964071273804, + "epoch": 2.3828920570264764, + "flow_matching_loss": 0.014455233700573444, + "grad_norm": 0.08939981460571289, + "kept_token_count": 572.73515625, + "learning_rate": 2.6103920403310832e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9922970414161683, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007702962541952729, + "mgd_loss_weight": 0.7271742403507233, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.37561950683596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.30230712890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8190, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.29938889741897584, + "epoch": 2.38580157113762, + "flow_matching_loss": 0.014723739493638277, + "grad_norm": 0.10429827868938446, + "kept_token_count": 573.2484375, + "learning_rate": 2.6092797064210465e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.992558765411377, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007441239198669791, + "mgd_loss_weight": 0.7268408954143524, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.9300811767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.5018829345703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8200, + "valid_token_count": 818.21796875 + }, + { + "actual_token_mask_ratio": 0.2993373781442642, + "epoch": 2.3887110852487634, + "flow_matching_loss": 0.015364088956266642, + "grad_norm": 0.09059911221265793, + "kept_token_count": 573.39296875, + "learning_rate": 2.6081660246315685e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9922141373157501, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00778585453517735, + "mgd_loss_weight": 0.7265075504779815, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.1739440917969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.7099975585937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8210, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.29963319897651675, + "epoch": 2.391620599359907, + "flow_matching_loss": 0.013582217413932085, + "grad_norm": 0.08408204466104507, + "kept_token_count": 573.053125, + "learning_rate": 2.6070509963158774e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9928442776203156, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00715570910833776, + "mgd_loss_weight": 0.7261742055416107, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.839208984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.50464782714846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8220, + "valid_token_count": 818.21484375 + }, + { + "actual_token_mask_ratio": 0.29989196062088014, + "epoch": 2.3945301134710504, + "flow_matching_loss": 0.014150504115968942, + "grad_norm": 0.09444590657949448, + "kept_token_count": 573.18671875, + "learning_rate": 2.6059346228288375e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9923465311527252, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007653460837900639, + "mgd_loss_weight": 0.7258408665657043, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.94248046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.95225830078124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8230, + "valid_token_count": 818.7171875 + }, + { + "actual_token_mask_ratio": 0.29966611266136167, + "epoch": 2.3974396275821936, + "flow_matching_loss": 0.014465534593909979, + "grad_norm": 0.10829909145832062, + "kept_token_count": 573.00078125, + "learning_rate": 2.604816905526948e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9922042548656463, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007795739406719804, + "mgd_loss_weight": 0.7255075156688691, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.82570190429686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.2076782226562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8240, + "valid_token_count": 818.18203125 + }, + { + "actual_token_mask_ratio": 0.3003557801246643, + "epoch": 2.4003491416933374, + "flow_matching_loss": 0.014119719620794058, + "grad_norm": 0.08530028164386749, + "kept_token_count": 572.63203125, + "learning_rate": 2.603697845768341e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9927207589149475, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072792414110153915, + "mgd_loss_weight": 0.7251741707324981, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.3579437255859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.83681335449216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8250, + "valid_token_count": 818.46171875 + }, + { + "actual_token_mask_ratio": 0.3000981241464615, + "epoch": 2.4032586558044806, + "flow_matching_loss": 0.014669765811413526, + "grad_norm": 0.08399280905723572, + "kept_token_count": 572.6890625, + "learning_rate": 2.6025774449127792e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9920489966869355, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007951013604179024, + "mgd_loss_weight": 0.7248408317565918, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.7635528564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.2423400878906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8260, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.3001750290393829, + "epoch": 2.406168169915624, + "flow_matching_loss": 0.013942151982337236, + "grad_norm": 0.09068997204303741, + "kept_token_count": 572.70625, + "learning_rate": 2.6014557043216562e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.992271339893341, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007728658989071846, + "mgd_loss_weight": 0.7245074808597565, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.0766571044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.08127746582034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8270, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.2995481461286545, + "epoch": 2.4090776840267676, + "flow_matching_loss": 0.01402661595493555, + "grad_norm": 0.10473766177892685, + "kept_token_count": 573.296875, + "learning_rate": 2.6003326253579913e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9923831880092621, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007616810267791152, + "mgd_loss_weight": 0.7241741418838501, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.9553192138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.2099975585937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8280, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.2991004168987274, + "epoch": 2.411987198137911, + "flow_matching_loss": 0.013944676611572504, + "grad_norm": 0.08029063045978546, + "kept_token_count": 573.75703125, + "learning_rate": 2.599208209386432e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.992052572965622, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007947420049458743, + "mgd_loss_weight": 0.7238407969474793, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 378.78465576171874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.03648681640624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8290, + "valid_token_count": 818.60234375 + }, + { + "actual_token_mask_ratio": 0.30008548200130464, + "epoch": 2.4148967122490546, + "flow_matching_loss": 0.016439757961779832, + "grad_norm": 0.11214832216501236, + "kept_token_count": 572.75703125, + "learning_rate": 2.5980824577732502e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9921566784381867, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007843312667682767, + "mgd_loss_weight": 0.7235074460506439, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.24708557128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.09788818359374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8300, + "valid_token_count": 818.33203125 + }, + { + "actual_token_mask_ratio": 0.3004282534122467, + "epoch": 2.417806226360198, + "flow_matching_loss": 0.014275787211954594, + "grad_norm": 0.09154179692268372, + "kept_token_count": 572.43359375, + "learning_rate": 2.5969553718863395e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9925597369670868, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007440262893214822, + "mgd_loss_weight": 0.7231741130352021, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.71018371582034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.51221923828126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8310, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.3000850170850754, + "epoch": 2.4207157404713415, + "flow_matching_loss": 0.015132080297917128, + "grad_norm": 0.08612929284572601, + "kept_token_count": 572.8546875, + "learning_rate": 2.595826953095216e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9922360360622406, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077639642171561715, + "mgd_loss_weight": 0.7228407561779022, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.70513610839845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.11098022460936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8320, + "valid_token_count": 818.465625 + }, + { + "actual_token_mask_ratio": 0.29950282275676726, + "epoch": 2.423625254582485, + "flow_matching_loss": 0.014492952730506659, + "grad_norm": 0.09392719715833664, + "kept_token_count": 573.225, + "learning_rate": 2.594697202771014e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9923299849033356, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00767002166248858, + "mgd_loss_weight": 0.7225074172019958, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.10138549804685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7039489746094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8330, + "valid_token_count": 818.31015625 + }, + { + "actual_token_mask_ratio": 0.29963702261447905, + "epoch": 2.426534768693628, + "flow_matching_loss": 0.015308167692273856, + "grad_norm": 0.06727878749370575, + "kept_token_count": 573.17109375, + "learning_rate": 2.5935661222864888e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9921837151050568, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007816283032298088, + "mgd_loss_weight": 0.7221740782260895, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.3921295166016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4136566162109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8340, + "valid_token_count": 818.38984375 + }, + { + "actual_token_mask_ratio": 0.299784055352211, + "epoch": 2.4294442828047718, + "flow_matching_loss": 0.016000039968639614, + "grad_norm": 0.08498529344797134, + "kept_token_count": 573.10703125, + "learning_rate": 2.592433713016008e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9923670947551727, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007632900727912784, + "mgd_loss_weight": 0.7218407213687896, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.3563262939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.14593200683595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8350, + "valid_token_count": 818.471875 + }, + { + "actual_token_mask_ratio": 0.2999949544668198, + "epoch": 2.432353796915915, + "flow_matching_loss": 0.01301511498168111, + "grad_norm": 0.09479008615016937, + "kept_token_count": 572.91640625, + "learning_rate": 2.5912999763355553e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9922697842121124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007730199908837676, + "mgd_loss_weight": 0.7215073883533478, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.5731719970703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.1334747314453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8360, + "valid_token_count": 818.44375 + }, + { + "actual_token_mask_ratio": 0.30010477006435393, + "epoch": 2.4352633110270583, + "flow_matching_loss": 0.014103740826249123, + "grad_norm": 0.08024419844150543, + "kept_token_count": 572.80546875, + "learning_rate": 2.590164913622729e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9923209011554718, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076790846884250644, + "mgd_loss_weight": 0.7211740374565124, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.71886291503904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.60544738769534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8370, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.29988047778606414, + "epoch": 2.438172825138202, + "flow_matching_loss": 0.015342954359948636, + "grad_norm": 0.09353189170360565, + "kept_token_count": 573.03125, + "learning_rate": 2.589028526256737e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9926823139190674, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007317674579098821, + "mgd_loss_weight": 0.7208406925201416, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.0981079101563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.7609344482422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8380, + "valid_token_count": 818.4765625 + }, + { + "actual_token_mask_ratio": 0.300597208738327, + "epoch": 2.4410823392493453, + "flow_matching_loss": 0.015204806067049503, + "grad_norm": 0.07193484157323837, + "kept_token_count": 572.55703125, + "learning_rate": 2.587890815618397e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9924563348293305, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075436573009938, + "mgd_loss_weight": 0.7205073535442352, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.313525390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1339508056641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8390, + "valid_token_count": 818.634375 + }, + { + "actual_token_mask_ratio": 0.3005319178104401, + "epoch": 2.443991853360489, + "flow_matching_loss": 0.01586719509214163, + "grad_norm": 0.09759300947189331, + "kept_token_count": 572.31875, + "learning_rate": 2.5867517830901343e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9927421391010285, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007257860479876399, + "mgd_loss_weight": 0.7201740026474, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.86524963378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.69321899414064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8400, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.2992886990308762, + "epoch": 2.4469013674716322, + "flow_matching_loss": 0.01471934150904417, + "grad_norm": 0.09634926915168762, + "kept_token_count": 573.45546875, + "learning_rate": 2.5856114300559812e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9926129758358002, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007387033896520734, + "mgd_loss_weight": 0.7198406636714936, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.98645935058596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.20689086914064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8410, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.3000608831644058, + "epoch": 2.4498108815827755, + "flow_matching_loss": 0.01532782707363367, + "grad_norm": 0.08552063256502151, + "kept_token_count": 572.7578125, + "learning_rate": 2.584469757901574e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9923852443695068, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00761475176550448, + "mgd_loss_weight": 0.7195073187351226, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.6743896484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.51186218261716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8420, + "valid_token_count": 818.29453125 + }, + { + "actual_token_mask_ratio": 0.2995427638292313, + "epoch": 2.452720395693919, + "flow_matching_loss": 0.013030633796006441, + "grad_norm": 0.08082013577222824, + "kept_token_count": 573.26875, + "learning_rate": 2.5833267680141526e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9922055423259735, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077944390010088686, + "mgd_loss_weight": 0.7191739678382874, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.4666809082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.229296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8430, + "valid_token_count": 818.42109375 + }, + { + "actual_token_mask_ratio": 0.3003364115953445, + "epoch": 2.4556299098050625, + "flow_matching_loss": 0.014842324890196323, + "grad_norm": 0.08362279087305069, + "kept_token_count": 572.61953125, + "learning_rate": 2.5821824617825564e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9920339822769165, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007966003380715847, + "mgd_loss_weight": 0.718840628862381, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.5140380859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.2337585449219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8440, + "valid_token_count": 818.41796875 + }, + { + "actual_token_mask_ratio": 0.30041658878326416, + "epoch": 2.458539423916206, + "flow_matching_loss": 0.014487131778150798, + "grad_norm": 0.09733878076076508, + "kept_token_count": 572.3890625, + "learning_rate": 2.5810368405972265e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9922563076019287, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00774368648417294, + "mgd_loss_weight": 0.7185072839260102, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.1595153808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 371.3998779296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8450, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.2997583746910095, + "epoch": 2.4614489380273494, + "flow_matching_loss": 0.014817821513861417, + "grad_norm": 0.09156084805727005, + "kept_token_count": 573.09296875, + "learning_rate": 2.5798899058501996e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9926300346851349, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007369964197278023, + "mgd_loss_weight": 0.7181739389896393, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.6736785888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.36541442871095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8460, + "valid_token_count": 818.42578125 + }, + { + "actual_token_mask_ratio": 0.299477806687355, + "epoch": 2.4643584521384927, + "flow_matching_loss": 0.013287033606320619, + "grad_norm": 0.08419202268123627, + "kept_token_count": 573.2828125, + "learning_rate": 2.578741658935111e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9920684099197388, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00793158747255802, + "mgd_loss_weight": 0.7178405940532684, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.80736389160154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.65818786621094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8470, + "valid_token_count": 818.365625 + }, + { + "actual_token_mask_ratio": 0.300000935792923, + "epoch": 2.4672679662496364, + "flow_matching_loss": 0.01421230286359787, + "grad_norm": 0.10646574199199677, + "kept_token_count": 572.92421875, + "learning_rate": 2.5775921012471875e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9922276556491851, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007772338390350342, + "mgd_loss_weight": 0.7175072491168976, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.5773986816406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.15577392578126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8480, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.29998467564582826, + "epoch": 2.4701774803607797, + "flow_matching_loss": 0.013547304179519415, + "grad_norm": 0.09343200922012329, + "kept_token_count": 572.83359375, + "learning_rate": 2.5764412341832505e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9922662436962127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007733753696084022, + "mgd_loss_weight": 0.7171739041805267, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.82481384277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.35784606933595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8490, + "valid_token_count": 818.31328125 + }, + { + "actual_token_mask_ratio": 0.2996069848537445, + "epoch": 2.4730869944719234, + "flow_matching_loss": 0.016159896831959487, + "grad_norm": 0.10051531344652176, + "kept_token_count": 573.084375, + "learning_rate": 2.5752890591417125e-05, + "loss": 0.0216, + "mgd_cos_sim": 0.9924795687198639, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007520440733060241, + "mgd_loss_weight": 0.7168405652046204, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 377.41566467285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.17665710449216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8500, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.2997722804546356, + "epoch": 2.4759965085830666, + "flow_matching_loss": 0.016856106370687483, + "grad_norm": 0.08081287890672684, + "kept_token_count": 572.8953125, + "learning_rate": 2.5741355775225744e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.99217689037323, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007823104131966829, + "mgd_loss_weight": 0.7165072143077851, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.7864105224609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8349151611328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8510, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.29974042177200316, + "epoch": 2.47890602269421, + "flow_matching_loss": 0.014774023555219173, + "grad_norm": 0.09466765075922012, + "kept_token_count": 572.99296875, + "learning_rate": 2.5729807907274254e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9922961056232452, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077038930729031565, + "mgd_loss_weight": 0.7161738693714141, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.974462890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9224151611328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8520, + "valid_token_count": 818.25546875 + }, + { + "actual_token_mask_ratio": 0.29975879192352295, + "epoch": 2.4818155368053536, + "flow_matching_loss": 0.014856158848851919, + "grad_norm": 0.10780717432498932, + "kept_token_count": 572.87109375, + "learning_rate": 2.5718247001594403e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9922559201717377, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007744078245013952, + "mgd_loss_weight": 0.7158405303955078, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.81935424804686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.5836151123047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8530, + "valid_token_count": 818.10234375 + }, + { + "actual_token_mask_ratio": 0.29946255683898926, + "epoch": 2.484725050916497, + "flow_matching_loss": 0.014561573509126902, + "grad_norm": 0.09309301525354385, + "kept_token_count": 573.2015625, + "learning_rate": 2.5706673072233776e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9923262655735016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007673741551116109, + "mgd_loss_weight": 0.7155071794986725, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.28013000488284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.7653472900391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8540, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.29974597692489624, + "epoch": 2.4876345650276406, + "flow_matching_loss": 0.014886464551091194, + "grad_norm": 0.08356621861457825, + "kept_token_count": 573.00546875, + "learning_rate": 2.5695086133255795e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9926065981388092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073933946434408425, + "mgd_loss_weight": 0.7151738405227661, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.4646026611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9112518310547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8550, + "valid_token_count": 818.28359375 + }, + { + "actual_token_mask_ratio": 0.3000637710094452, + "epoch": 2.490544079138784, + "flow_matching_loss": 0.015232678782194854, + "grad_norm": 0.09149016439914703, + "kept_token_count": 572.63984375, + "learning_rate": 2.5683486198739687e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9923528850078582, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076471161097288135, + "mgd_loss_weight": 0.7148404955863953, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.6932312011719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.62921752929685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8560, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.30050258338451385, + "epoch": 2.493453593249927, + "flow_matching_loss": 0.014838566444814206, + "grad_norm": 0.09884189069271088, + "kept_token_count": 572.21875, + "learning_rate": 2.5671873282780452e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9926129758358002, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007387028401717543, + "mgd_loss_weight": 0.7145071446895599, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 385.8701568603516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9208251953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8570, + "valid_token_count": 818.04140625 + }, + { + "actual_token_mask_ratio": 0.29969754815101624, + "epoch": 2.496363107361071, + "flow_matching_loss": 0.015180754940956831, + "grad_norm": 0.07834276556968689, + "kept_token_count": 572.91171875, + "learning_rate": 2.5660247399488887e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9924377501010895, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00756225255317986, + "mgd_loss_weight": 0.7141738116741181, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.84248046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.24883728027345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8580, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.2998173117637634, + "epoch": 2.499272621472214, + "flow_matching_loss": 0.014986086450517178, + "grad_norm": 0.0967559888958931, + "kept_token_count": 572.9984375, + "learning_rate": 2.564860856299154e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9925747573375702, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007425246667116881, + "mgd_loss_weight": 0.7138404548168182, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.10662536621095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.10564575195315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8590, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.29980901181697844, + "epoch": 2.5021821355833573, + "flow_matching_loss": 0.014567784871906042, + "grad_norm": 0.10975093394517899, + "kept_token_count": 572.88203125, + "learning_rate": 2.5636956787430693e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9927018821239472, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007298114756122231, + "mgd_loss_weight": 0.7135071158409119, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.22752685546874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.56878662109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8600, + "valid_token_count": 818.17734375 + }, + { + "actual_token_mask_ratio": 0.29981582164764403, + "epoch": 2.505091649694501, + "flow_matching_loss": 0.015030965954065324, + "grad_norm": 0.085194431245327, + "kept_token_count": 572.89921875, + "learning_rate": 2.5625292086964346e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9924318015575408, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007568190153688192, + "mgd_loss_weight": 0.7131737768650055, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.93520202636716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9801055908203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8610, + "valid_token_count": 818.20859375 + }, + { + "actual_token_mask_ratio": 0.29897748231887816, + "epoch": 2.5080011638056443, + "flow_matching_loss": 0.01589489635080099, + "grad_norm": 0.09344340115785599, + "kept_token_count": 573.80859375, + "learning_rate": 2.5613614475766212e-05, + "loss": 0.0214, + "mgd_cos_sim": 0.9923161685466766, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007683829497545957, + "mgd_loss_weight": 0.7128404259681702, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.6260284423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.3934356689453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8620, + "valid_token_count": 818.52890625 + }, + { + "actual_token_mask_ratio": 0.30019004344940187, + "epoch": 2.510910677916788, + "flow_matching_loss": 0.015061532519757748, + "grad_norm": 0.10831978917121887, + "kept_token_count": 572.54765625, + "learning_rate": 2.5601923968025694e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9923517167568207, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076482782140374185, + "mgd_loss_weight": 0.7125070869922638, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.3468078613281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.65492248535156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8630, + "valid_token_count": 818.14296875 + }, + { + "actual_token_mask_ratio": 0.30015707314014434, + "epoch": 2.5138201920279313, + "flow_matching_loss": 0.014098320994526148, + "grad_norm": 0.08847790956497192, + "kept_token_count": 572.5984375, + "learning_rate": 2.5590220577947862e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9926408529281616, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007359137991443276, + "mgd_loss_weight": 0.7121737420558929, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.8221008300781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.0000030517578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8640, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.30049126744270327, + "epoch": 2.516729706139075, + "flow_matching_loss": 0.014302333351224661, + "grad_norm": 0.0935431569814682, + "kept_token_count": 572.28359375, + "learning_rate": 2.5578504319753427e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9925943970680237, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074055954348295925, + "mgd_loss_weight": 0.7118403911590576, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.48789978027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9649230957031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8650, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.30012997388839724, + "epoch": 2.5196392202502182, + "flow_matching_loss": 0.01519374419003725, + "grad_norm": 0.07869967073202133, + "kept_token_count": 572.6, + "learning_rate": 2.556677520767876e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9924103796482087, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007589618302881717, + "mgd_loss_weight": 0.7115070521831512, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.0702178955078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 372.4778259277344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8660, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.29997886121273043, + "epoch": 2.5225487343613615, + "flow_matching_loss": 0.013818387687206269, + "grad_norm": 0.08293259888887405, + "kept_token_count": 572.85390625, + "learning_rate": 2.5555033255975835e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9920179426670075, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00798206846229732, + "mgd_loss_weight": 0.7111737072467804, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.7339569091797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.2914703369141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8670, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.29999957978725433, + "epoch": 2.525458248472505, + "flow_matching_loss": 0.013678819313645363, + "grad_norm": 0.0719650536775589, + "kept_token_count": 572.78125, + "learning_rate": 2.5543278478912228e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9927015423774719, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007298461860045791, + "mgd_loss_weight": 0.7108403623104096, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.2384338378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2889007568359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8680, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.3004955381155014, + "epoch": 2.5283677625836485, + "flow_matching_loss": 0.01480087051168084, + "grad_norm": 0.07384617626667023, + "kept_token_count": 572.61015625, + "learning_rate": 2.55315108907711e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9922552108764648, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007744806911796331, + "mgd_loss_weight": 0.7105070173740387, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.7120361328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1330596923828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8690, + "valid_token_count": 818.58984375 + }, + { + "actual_token_mask_ratio": 0.30015007853507997, + "epoch": 2.5312772766947917, + "flow_matching_loss": 0.01613095561042428, + "grad_norm": 0.08463851362466812, + "kept_token_count": 572.38828125, + "learning_rate": 2.551973050585119e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9924668431282043, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007533175032585859, + "mgd_loss_weight": 0.7101736724376678, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.1398223876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.8350433349609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8700, + "valid_token_count": 817.87265625 + }, + { + "actual_token_mask_ratio": 0.29937017560005186, + "epoch": 2.5341867908059355, + "flow_matching_loss": 0.014533717930316926, + "grad_norm": 0.09877162426710129, + "kept_token_count": 573.21484375, + "learning_rate": 2.5507937338466768e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9920875072479248, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007912498945370317, + "mgd_loss_weight": 0.709840327501297, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 379.7068115234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.2759552001953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8710, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.3000451326370239, + "epoch": 2.5370963049170787, + "flow_matching_loss": 0.013167052157223224, + "grad_norm": 0.08067155629396439, + "kept_token_count": 572.83671875, + "learning_rate": 2.549613140294764e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.992772388458252, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007227609539404512, + "mgd_loss_weight": 0.7095069885253906, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.2994903564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.3482635498047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8720, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.2999663412570953, + "epoch": 2.5400058190282224, + "flow_matching_loss": 0.013543590810149908, + "grad_norm": 0.08465619385242462, + "kept_token_count": 572.7265625, + "learning_rate": 2.548431271363914e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9928130626678466, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071869449689984325, + "mgd_loss_weight": 0.7091736376285553, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.9923889160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.01243896484374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8730, + "valid_token_count": 818.14609375 + }, + { + "actual_token_mask_ratio": 0.30062039792537687, + "epoch": 2.5429153331393657, + "flow_matching_loss": 0.015014170482754708, + "grad_norm": 0.08105818927288055, + "kept_token_count": 572.3828125, + "learning_rate": 2.5472481284902092e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9922063529491425, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00779364425688982, + "mgd_loss_weight": 0.7088402926921844, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 383.8024597167969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.90196228027344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8740, + "valid_token_count": 818.41171875 + }, + { + "actual_token_mask_ratio": 0.30012138187885284, + "epoch": 2.5458248472505094, + "flow_matching_loss": 0.014271708857268094, + "grad_norm": 0.07990943640470505, + "kept_token_count": 572.57578125, + "learning_rate": 2.5460637131112788e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9928646743297577, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007135323015972972, + "mgd_loss_weight": 0.708506953716278, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.7729248046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.39211120605466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8750, + "valid_token_count": 818.10390625 + }, + { + "actual_token_mask_ratio": 0.2998449236154556, + "epoch": 2.5487343613616527, + "flow_matching_loss": 0.01538711627945304, + "grad_norm": 0.07857827842235565, + "kept_token_count": 572.84140625, + "learning_rate": 2.5448780266662995e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9922860205173493, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077139927539974455, + "mgd_loss_weight": 0.7081736028194427, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.6636474609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8684509277344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8760, + "valid_token_count": 818.16015625 + }, + { + "actual_token_mask_ratio": 0.30009148418903353, + "epoch": 2.551643875472796, + "flow_matching_loss": 0.016339020524173974, + "grad_norm": 0.08151364326477051, + "kept_token_count": 572.59765625, + "learning_rate": 2.5436910705959926e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9923822343349457, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007617776095867157, + "mgd_loss_weight": 0.7078402638435364, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.52537536621094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.2278686523438, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8770, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.29955790340900423, + "epoch": 2.5545533895839396, + "flow_matching_loss": 0.015064245648682117, + "grad_norm": 0.08350734412670135, + "kept_token_count": 573.21015625, + "learning_rate": 2.542502846342621e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9923704564571381, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007629557000473142, + "mgd_loss_weight": 0.7075069189071655, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.82086791992185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9558135986328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8780, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.30017286241054536, + "epoch": 2.557462903695083, + "flow_matching_loss": 0.014911275263875723, + "grad_norm": 0.10075365006923676, + "kept_token_count": 572.49921875, + "learning_rate": 2.5413133553499892e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9925655782222748, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007434413442388177, + "mgd_loss_weight": 0.7071735680103302, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.2222839355469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1069793701172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8790, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.29966399669647215, + "epoch": 2.560372417806226, + "flow_matching_loss": 0.015707289054989814, + "grad_norm": 0.08608246594667435, + "kept_token_count": 573.01171875, + "learning_rate": 2.540122599063442e-05, + "loss": 0.0211, + "mgd_cos_sim": 0.9923436105251312, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00765638672746718, + "mgd_loss_weight": 0.7068402349948884, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.305712890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6131256103516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8800, + "valid_token_count": 818.19765625 + }, + { + "actual_token_mask_ratio": 0.3003094166517258, + "epoch": 2.56328193191737, + "flow_matching_loss": 0.014339689537882805, + "grad_norm": 0.09662666916847229, + "kept_token_count": 572.615625, + "learning_rate": 2.538930578929859e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9923228979110718, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007677104230970145, + "mgd_loss_weight": 0.7065068781375885, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.8052490234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.81658325195315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8810, + "valid_token_count": 818.38671875 + }, + { + "actual_token_mask_ratio": 0.29964015185832976, + "epoch": 2.566191446028513, + "flow_matching_loss": 0.016795053146779537, + "grad_norm": 0.18345797061920166, + "kept_token_count": 573.140625, + "learning_rate": 2.537737296397658e-05, + "loss": 0.0225, + "mgd_cos_sim": 0.991968160867691, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008031843090429902, + "mgd_loss_weight": 0.7061735391616821, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.7617950439453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.12713317871095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8820, + "valid_token_count": 818.35625 + }, + { + "actual_token_mask_ratio": 0.3005715787410736, + "epoch": 2.569100960139657, + "flow_matching_loss": 0.01651366800069809, + "grad_norm": 0.09479749947786331, + "kept_token_count": 572.43125, + "learning_rate": 2.5365427529167903e-05, + "loss": 0.022, + "mgd_cos_sim": 0.99225412607193, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007745888363569975, + "mgd_loss_weight": 0.7058402001857758, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.4355712890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6241394042969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8830, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.30008854269981383, + "epoch": 2.5720104742508, + "flow_matching_loss": 0.013247242011129857, + "grad_norm": 0.09282638877630234, + "kept_token_count": 572.61796875, + "learning_rate": 2.5353469499387375e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9922708392143249, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007729159714654088, + "mgd_loss_weight": 0.7055068433284759, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.96038513183595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.64996337890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8840, + "valid_token_count": 818.12734375 + }, + { + "actual_token_mask_ratio": 0.3002706915140152, + "epoch": 2.574919988361944, + "flow_matching_loss": 0.01498776813969016, + "grad_norm": 0.09805722534656525, + "kept_token_count": 572.57265625, + "learning_rate": 2.534149888916514e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9923460781574249, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007653925009071827, + "mgd_loss_weight": 0.7051735103130341, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.289892578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.89623107910154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8850, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.29993373453617095, + "epoch": 2.577829502473087, + "flow_matching_loss": 0.013480293191969394, + "grad_norm": 0.0852678120136261, + "kept_token_count": 572.884375, + "learning_rate": 2.5329515713046616e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9923084735870361, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007691532326862216, + "mgd_loss_weight": 0.7048401594161987, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.02065124511716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3578186035156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8860, + "valid_token_count": 818.3328125 + }, + { + "actual_token_mask_ratio": 0.2998544633388519, + "epoch": 2.5807390165842303, + "flow_matching_loss": 0.014092228841036558, + "grad_norm": 0.08396327495574951, + "kept_token_count": 573.059375, + "learning_rate": 2.531751998559249e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9925584852695465, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007441507233306765, + "mgd_loss_weight": 0.7045068144798279, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.2715148925781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1014038085938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8870, + "valid_token_count": 818.48359375 + }, + { + "actual_token_mask_ratio": 0.2997783362865448, + "epoch": 2.583648530695374, + "flow_matching_loss": 0.013519908580929042, + "grad_norm": 0.08533726632595062, + "kept_token_count": 573.11328125, + "learning_rate": 2.5305511721378703e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9925451040267944, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00745490905828774, + "mgd_loss_weight": 0.7041734755039215, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.47730712890626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.91407165527346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8880, + "valid_token_count": 818.47890625 + }, + { + "actual_token_mask_ratio": 0.3004255831241608, + "epoch": 2.5865580448065173, + "flow_matching_loss": 0.013739699590951204, + "grad_norm": 0.0750124379992485, + "kept_token_count": 572.31484375, + "learning_rate": 2.529349093499642e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9924026668071747, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007597348466515541, + "mgd_loss_weight": 0.7038401246070862, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.37847595214845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.19090576171874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8890, + "valid_token_count": 818.09140625 + }, + { + "actual_token_mask_ratio": 0.3000729620456696, + "epoch": 2.5894675589176606, + "flow_matching_loss": 0.014127221796661616, + "grad_norm": 0.10307270288467407, + "kept_token_count": 572.7140625, + "learning_rate": 2.528145764105205e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9923879683017731, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007612030673772097, + "mgd_loss_weight": 0.7035067856311799, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.0730712890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.0295349121094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8900, + "valid_token_count": 818.24609375 + }, + { + "actual_token_mask_ratio": 0.2999687224626541, + "epoch": 2.5923770730288043, + "flow_matching_loss": 0.014596236683428288, + "grad_norm": 0.09920785576105118, + "kept_token_count": 572.86640625, + "learning_rate": 2.5269411854167166e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9923486053943634, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007651401031762361, + "mgd_loss_weight": 0.7031734406948089, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.9828643798828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.36123046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8910, + "valid_token_count": 818.34765625 + }, + { + "actual_token_mask_ratio": 0.2994078457355499, + "epoch": 2.5952865871399475, + "flow_matching_loss": 0.01670250380411744, + "grad_norm": 0.09544530510902405, + "kept_token_count": 573.27421875, + "learning_rate": 2.5257353588978534e-05, + "loss": 0.0223, + "mgd_cos_sim": 0.9920636355876923, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007936366740614176, + "mgd_loss_weight": 0.7028400897979736, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.4118316650391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.08665771484374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8920, + "valid_token_count": 818.26953125 + }, + { + "actual_token_mask_ratio": 0.3002221524715424, + "epoch": 2.598196101251091, + "flow_matching_loss": 0.014545450266450644, + "grad_norm": 0.10296172648668289, + "kept_token_count": 572.5671875, + "learning_rate": 2.5245282860138083e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9923132479190826, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007686760975047946, + "mgd_loss_weight": 0.7025067508220673, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.8339813232422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.19346008300784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8930, + "valid_token_count": 818.21171875 + }, + { + "actual_token_mask_ratio": 0.3007997632026672, + "epoch": 2.6011056153622345, + "flow_matching_loss": 0.013844679482281208, + "grad_norm": 0.06768085807561874, + "kept_token_count": 572.0921875, + "learning_rate": 2.5233199682312894e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9926942884922028, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007305709924548864, + "mgd_loss_weight": 0.7021734058856964, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.1562072753906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.91277770996095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8940, + "valid_token_count": 818.20703125 + }, + { + "actual_token_mask_ratio": 0.29982872903347013, + "epoch": 2.604015129473378, + "flow_matching_loss": 0.01368642207235098, + "grad_norm": 0.08574308454990387, + "kept_token_count": 572.94453125, + "learning_rate": 2.5221104070185163e-05, + "loss": 0.019, + "mgd_cos_sim": 0.992370855808258, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007629126869142055, + "mgd_loss_weight": 0.7018400609493256, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.83021545410156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.20079345703124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8950, + "valid_token_count": 818.29765625 + }, + { + "actual_token_mask_ratio": 0.2991062134504318, + "epoch": 2.6069246435845215, + "flow_matching_loss": 0.014691543020308017, + "grad_norm": 0.06988488882780075, + "kept_token_count": 573.42421875, + "learning_rate": 2.5208996038452204e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9923865854740143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007613416016101837, + "mgd_loss_weight": 0.7015067160129547, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.7383697509766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7837890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8960, + "valid_token_count": 818.13203125 + }, + { + "actual_token_mask_ratio": 0.29980034828186036, + "epoch": 2.6098341576956647, + "flow_matching_loss": 0.013613775186240674, + "grad_norm": 0.07286258786916733, + "kept_token_count": 572.959375, + "learning_rate": 2.5196875601826403e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9922600746154785, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007739907735958696, + "mgd_loss_weight": 0.7011733710765838, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.34079895019534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.9179321289063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8970, + "valid_token_count": 818.27890625 + }, + { + "actual_token_mask_ratio": 0.30041683912277223, + "epoch": 2.6127436718068084, + "flow_matching_loss": 0.014034902770072221, + "grad_norm": 0.09934661537408829, + "kept_token_count": 572.29921875, + "learning_rate": 2.518474277503525e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9917437016963959, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008256288710981607, + "mgd_loss_weight": 0.700840026140213, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.64808044433596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.63963012695314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8980, + "valid_token_count": 818.06484375 + }, + { + "actual_token_mask_ratio": 0.2993658006191254, + "epoch": 2.6156531859179517, + "flow_matching_loss": 0.014084414206445217, + "grad_norm": 0.05953103303909302, + "kept_token_count": 573.32265625, + "learning_rate": 2.5172597572821266e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9922805488109588, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007719452446326614, + "mgd_loss_weight": 0.7005066871643066, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.1154449462891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2251342773437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 8990, + "valid_token_count": 818.29453125 + }, + { + "actual_token_mask_ratio": 0.29909224808216095, + "epoch": 2.618562700029095, + "flow_matching_loss": 0.013811266422271729, + "grad_norm": 0.0809660330414772, + "kept_token_count": 573.728125, + "learning_rate": 2.5160440009942015e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9924897909164428, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007510196557268501, + "mgd_loss_weight": 0.7001733362674714, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.0749481201172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6588104248047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9000, + "valid_token_count": 818.54921875 + }, + { + "actual_token_mask_ratio": 0.299935394525528, + "epoch": 2.6214722141402387, + "flow_matching_loss": 0.014223633240908385, + "grad_norm": 0.08239413797855377, + "kept_token_count": 572.9265625, + "learning_rate": 2.514827010117009e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.992354017496109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007645989628508687, + "mgd_loss_weight": 0.6998399913311004, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.81894836425784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.7626678466797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9010, + "valid_token_count": 818.38828125 + }, + { + "actual_token_mask_ratio": 0.3003134191036224, + "epoch": 2.624381728251382, + "flow_matching_loss": 0.014232487697154284, + "grad_norm": 0.07782375812530518, + "kept_token_count": 572.4859375, + "learning_rate": 2.5136087861293068e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9928710401058197, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007128963572904467, + "mgd_loss_weight": 0.699506652355194, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.9619140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.62298278808595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9020, + "valid_token_count": 818.20234375 + }, + { + "actual_token_mask_ratio": 0.29991153478622434, + "epoch": 2.627291242362525, + "flow_matching_loss": 0.014972795452922582, + "grad_norm": 0.07503053545951843, + "kept_token_count": 572.940625, + "learning_rate": 2.5123893305113525e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9923054099082946, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076945849228650335, + "mgd_loss_weight": 0.6991733014583588, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.2366058349609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.0627868652344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9030, + "valid_token_count": 818.38515625 + }, + { + "actual_token_mask_ratio": 0.2999882310628891, + "epoch": 2.630200756473669, + "flow_matching_loss": 0.014360361360013485, + "grad_norm": 0.07659457623958588, + "kept_token_count": 572.8328125, + "learning_rate": 2.5111686447448997e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9924047589302063, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007595219369977713, + "mgd_loss_weight": 0.6988399624824524, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.6344512939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.1387664794922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9040, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.3009793937206268, + "epoch": 2.633110270584812, + "flow_matching_loss": 0.01488440139219165, + "grad_norm": 0.09587208926677704, + "kept_token_count": 571.97890625, + "learning_rate": 2.5099467303131966e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9922744810581208, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007725499104708433, + "mgd_loss_weight": 0.698506623506546, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.6221893310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.46942749023435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9050, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.30001677572727203, + "epoch": 2.636019784695956, + "flow_matching_loss": 0.014161236491054297, + "grad_norm": 0.09469021111726761, + "kept_token_count": 572.7671875, + "learning_rate": 2.508723588700984e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9922487318515778, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077512651216238735, + "mgd_loss_weight": 0.6981732666492462, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.0219207763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7336486816406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9060, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.2997903048992157, + "epoch": 2.638929298807099, + "flow_matching_loss": 0.015490016527473926, + "grad_norm": 0.08069504052400589, + "kept_token_count": 573.0, + "learning_rate": 2.5074992213944952e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9925092816352844, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007490708399564028, + "mgd_loss_weight": 0.6978399336338044, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.3380584716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.92999267578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9070, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.29997349083423613, + "epoch": 2.641838812918243, + "flow_matching_loss": 0.01415353612974286, + "grad_norm": 0.09197385609149933, + "kept_token_count": 572.71875, + "learning_rate": 2.506273629881451e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9925875782966613, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007412425242364407, + "mgd_loss_weight": 0.697506582736969, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.10907287597655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.62108154296874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9080, + "valid_token_count": 818.14453125 + }, + { + "actual_token_mask_ratio": 0.3005366325378418, + "epoch": 2.644748327029386, + "flow_matching_loss": 0.01655249102041125, + "grad_norm": 0.10232706367969513, + "kept_token_count": 572.28984375, + "learning_rate": 2.5050468156510617e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9920213222503662, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007978693023324013, + "mgd_loss_weight": 0.6971732378005981, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.49757385253906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.82582397460936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9090, + "valid_token_count": 818.18515625 + }, + { + "actual_token_mask_ratio": 0.3002394139766693, + "epoch": 2.6476578411405294, + "flow_matching_loss": 0.013569427374750375, + "grad_norm": 0.07234499603509903, + "kept_token_count": 572.52734375, + "learning_rate": 2.5038187801940215e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9926812589168549, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007318747695535421, + "mgd_loss_weight": 0.6968398988246918, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.08063354492185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.54148864746094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9100, + "valid_token_count": 818.17734375 + }, + { + "actual_token_mask_ratio": 0.30030926764011384, + "epoch": 2.650567355251673, + "flow_matching_loss": 0.014935857057571411, + "grad_norm": 0.08712096512317657, + "kept_token_count": 572.71953125, + "learning_rate": 2.5025895250025088e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9924154698848724, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075845434330403805, + "mgd_loss_weight": 0.6965065479278565, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.21433715820314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9338317871094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9110, + "valid_token_count": 818.53203125 + }, + { + "actual_token_mask_ratio": 0.2999247401952744, + "epoch": 2.6534768693628163, + "flow_matching_loss": 0.015572716761380435, + "grad_norm": 0.08929014950990677, + "kept_token_count": 572.56640625, + "learning_rate": 2.501359051570186e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9922104060649872, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007789600407704711, + "mgd_loss_weight": 0.6961732089519501, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.55039978027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.5764678955078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9120, + "valid_token_count": 817.86171875 + }, + { + "actual_token_mask_ratio": 0.3003899395465851, + "epoch": 2.6563863834739596, + "flow_matching_loss": 0.01520875170826912, + "grad_norm": 0.07898793369531631, + "kept_token_count": 572.35625, + "learning_rate": 2.500127361392193e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9925616204738616, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007438379526138306, + "mgd_loss_weight": 0.6958398640155792, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.3337829589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.41668395996095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9130, + "valid_token_count": 818.10703125 + }, + { + "actual_token_mask_ratio": 0.29985744059085845, + "epoch": 2.6592958975851033, + "flow_matching_loss": 0.015346341766417027, + "grad_norm": 0.08635390549898148, + "kept_token_count": 572.88125, + "learning_rate": 2.49889445596515e-05, + "loss": 0.0208, + "mgd_cos_sim": 0.9921052157878876, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007894771499559283, + "mgd_loss_weight": 0.6955065131187439, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.6849639892578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.2562561035156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9140, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.2998348087072372, + "epoch": 2.6622054116962466, + "flow_matching_loss": 0.014514266140758991, + "grad_norm": 0.0836712047457695, + "kept_token_count": 572.80234375, + "learning_rate": 2.497660336787154e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9923267185688018, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007673274725675583, + "mgd_loss_weight": 0.6951731741428375, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.7950439453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.2622985839844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9150, + "valid_token_count": 818.09609375 + }, + { + "actual_token_mask_ratio": 0.2999740928411484, + "epoch": 2.6651149258073903, + "flow_matching_loss": 0.014217594265937805, + "grad_norm": 0.09412088245153427, + "kept_token_count": 572.76953125, + "learning_rate": 2.496425005357775e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9922425627708436, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007757449056953192, + "mgd_loss_weight": 0.6948398292064667, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.67237854003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.538134765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9160, + "valid_token_count": 818.21640625 + }, + { + "actual_token_mask_ratio": 0.2996217906475067, + "epoch": 2.6680244399185336, + "flow_matching_loss": 0.016089400742202997, + "grad_norm": 0.0974871963262558, + "kept_token_count": 573.17734375, + "learning_rate": 2.4951884631780578e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9921429574489593, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007857056939974428, + "mgd_loss_weight": 0.6945064842700959, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.1809326171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.91253356933595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9170, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.29967118203639986, + "epoch": 2.6709339540296773, + "flow_matching_loss": 0.013681190367788077, + "grad_norm": 0.07148488610982895, + "kept_token_count": 573.0953125, + "learning_rate": 2.493950711750518e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9922451436519623, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007754847826436162, + "mgd_loss_weight": 0.6941731393337249, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.75734558105466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1275238037109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9180, + "valid_token_count": 818.31953125 + }, + { + "actual_token_mask_ratio": 0.30028572380542756, + "epoch": 2.6738434681408205, + "flow_matching_loss": 0.01657603643834591, + "grad_norm": 0.0922766625881195, + "kept_token_count": 572.5578125, + "learning_rate": 2.49271175257914e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9925861060619354, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007413889560848474, + "mgd_loss_weight": 0.6938397943973541, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.7476043701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7973266601563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9190, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.2988994926214218, + "epoch": 2.676752982251964, + "flow_matching_loss": 0.015292311646044254, + "grad_norm": 0.07596750557422638, + "kept_token_count": 573.82421875, + "learning_rate": 2.4914715871693757e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9927234947681427, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007276521669700742, + "mgd_loss_weight": 0.6935064494609833, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.8689666748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9284942626953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9200, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.29905794858932494, + "epoch": 2.6796624963631075, + "flow_matching_loss": 0.01610009642317891, + "grad_norm": 0.08956771343946457, + "kept_token_count": 573.65625, + "learning_rate": 2.4902302170281438e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9921651124954224, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007834879774600267, + "mgd_loss_weight": 0.6931731104850769, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.9339263916016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.73145751953126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9210, + "valid_token_count": 818.40546875 + }, + { + "actual_token_mask_ratio": 0.3001516729593277, + "epoch": 2.6825720104742508, + "flow_matching_loss": 0.014664085675030946, + "grad_norm": 0.09002513438463211, + "kept_token_count": 572.778125, + "learning_rate": 2.4889876436638252e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9922263026237488, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007773707341402769, + "mgd_loss_weight": 0.6928397595882416, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.2615234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6478271484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9220, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.299919593334198, + "epoch": 2.685481524585394, + "flow_matching_loss": 0.0145580411888659, + "grad_norm": 0.09022394567728043, + "kept_token_count": 572.94296875, + "learning_rate": 2.4877438685862643e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9923294663429261, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076705299783498045, + "mgd_loss_weight": 0.6925064146518707, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.9584197998047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4561492919922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9230, + "valid_token_count": 818.39921875 + }, + { + "actual_token_mask_ratio": 0.2999327689409256, + "epoch": 2.6883910386965377, + "flow_matching_loss": 0.013912556041032077, + "grad_norm": 0.09512188285589218, + "kept_token_count": 572.925, + "learning_rate": 2.486498893306765e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9923900187015533, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007609984744340182, + "mgd_loss_weight": 0.6921730756759643, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.33937683105466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.92722473144534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9240, + "valid_token_count": 818.38515625 + }, + { + "actual_token_mask_ratio": 0.299774032831192, + "epoch": 2.691300552807681, + "flow_matching_loss": 0.014486795850098133, + "grad_norm": 0.11073649674654007, + "kept_token_count": 573.03984375, + "learning_rate": 2.485252719338089e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9924026727676392, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007597321039065718, + "mgd_loss_weight": 0.691839724779129, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.9159240722656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8687347412109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9250, + "valid_token_count": 818.36171875 + }, + { + "actual_token_mask_ratio": 0.3000771403312683, + "epoch": 2.6942100669188247, + "flow_matching_loss": 0.01497088810428977, + "grad_norm": 0.12125200033187866, + "kept_token_count": 572.90625, + "learning_rate": 2.4840053481944563e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9926551938056946, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007344813132658601, + "mgd_loss_weight": 0.6915063858032227, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.58143615722656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.8343841552734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9260, + "valid_token_count": 818.5234375 + }, + { + "actual_token_mask_ratio": 0.2993348717689514, + "epoch": 2.697119581029968, + "flow_matching_loss": 0.01459685992449522, + "grad_norm": 0.09639924764633179, + "kept_token_count": 573.4171875, + "learning_rate": 2.4827567813915398e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9926784574985504, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007321545947343111, + "mgd_loss_weight": 0.6911730408668518, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.7523254394531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.65403137207034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9270, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.3003928303718567, + "epoch": 2.7000290951411117, + "flow_matching_loss": 0.014422027487307787, + "grad_norm": 0.08513341844081879, + "kept_token_count": 572.43125, + "learning_rate": 2.481507020446466e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9927179455757141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007282060058787465, + "mgd_loss_weight": 0.6908396899700164, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.2017364501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.0538024902344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9280, + "valid_token_count": 818.21484375 + }, + { + "actual_token_mask_ratio": 0.2994506299495697, + "epoch": 2.702938609252255, + "flow_matching_loss": 0.01461621318012476, + "grad_norm": 0.08684160560369492, + "kept_token_count": 573.1171875, + "learning_rate": 2.4802560668778122e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9920462369918823, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007953751925379038, + "mgd_loss_weight": 0.6905063569545746, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.94165954589846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.98623046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9290, + "valid_token_count": 818.09921875 + }, + { + "actual_token_mask_ratio": 0.30039680898189547, + "epoch": 2.705848123363398, + "flow_matching_loss": 0.013519135117530823, + "grad_norm": 0.09475449472665787, + "kept_token_count": 572.321875, + "learning_rate": 2.479003922205605e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9926491260528565, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073508883360773325, + "mgd_loss_weight": 0.6901730000972748, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.9713073730469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.28985900878905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9300, + "valid_token_count": 818.06953125 + }, + { + "actual_token_mask_ratio": 0.30036795437335967, + "epoch": 2.708757637474542, + "flow_matching_loss": 0.014745824784040452, + "grad_norm": 0.091225765645504, + "kept_token_count": 572.575, + "learning_rate": 2.4777505879513183e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9924744009971619, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007525599654763937, + "mgd_loss_weight": 0.6898396611213684, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 387.9230987548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.2260345458984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9310, + "valid_token_count": 818.39140625 + }, + { + "actual_token_mask_ratio": 0.2994004637002945, + "epoch": 2.711667151585685, + "flow_matching_loss": 0.013249027822166682, + "grad_norm": 0.09052973240613937, + "kept_token_count": 573.45859375, + "learning_rate": 2.4764960656378714e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9920889914035798, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007911004777997731, + "mgd_loss_weight": 0.689506322145462, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.50559692382814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.6469329833984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9320, + "valid_token_count": 818.5234375 + }, + { + "actual_token_mask_ratio": 0.2998292773962021, + "epoch": 2.7145766656968284, + "flow_matching_loss": 0.012454483006149531, + "grad_norm": 0.09097950160503387, + "kept_token_count": 573.04375, + "learning_rate": 2.4752403567896274e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9928274869918823, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007172522461041808, + "mgd_loss_weight": 0.6891729652881622, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.56163635253904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.76381225585936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9330, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.30020858347415924, + "epoch": 2.717486179807972, + "flow_matching_loss": 0.013270444655790924, + "grad_norm": 0.09719116985797882, + "kept_token_count": 572.64375, + "learning_rate": 2.47398346293239e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9926110982894898, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007388897472992539, + "mgd_loss_weight": 0.6888396322727204, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.03272399902346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9192657470703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9340, + "valid_token_count": 818.30546875 + }, + { + "actual_token_mask_ratio": 0.3001975029706955, + "epoch": 2.7203956939191154, + "flow_matching_loss": 0.014035016298294067, + "grad_norm": 0.07986672222614288, + "kept_token_count": 572.60078125, + "learning_rate": 2.4727253855934055e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9924459159374237, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00755407982505858, + "mgd_loss_weight": 0.688506281375885, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.66766357421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.4726867675781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9350, + "valid_token_count": 818.23046875 + }, + { + "actual_token_mask_ratio": 0.29958640336990355, + "epoch": 2.723305208030259, + "flow_matching_loss": 0.014631418138742447, + "grad_norm": 0.09782984852790833, + "kept_token_count": 573.03828125, + "learning_rate": 2.4714661263013548e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9922488868236542, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007751118578016758, + "mgd_loss_weight": 0.6881729364395142, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.6654815673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0513854980469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9360, + "valid_token_count": 818.14453125 + }, + { + "actual_token_mask_ratio": 0.2999070525169373, + "epoch": 2.7262147221414024, + "flow_matching_loss": 0.014654483553022147, + "grad_norm": 0.08359870314598083, + "kept_token_count": 573.02109375, + "learning_rate": 2.4702056865863575e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9922006964683533, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00779929175041616, + "mgd_loss_weight": 0.6878395974636078, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.92564392089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.9377471923828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9370, + "valid_token_count": 818.4953125 + }, + { + "actual_token_mask_ratio": 0.2997988611459732, + "epoch": 2.729124236252546, + "flow_matching_loss": 0.014973325282335281, + "grad_norm": 0.08171258121728897, + "kept_token_count": 573.05625, + "learning_rate": 2.4689440679799672e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9922612130641937, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00773879112675786, + "mgd_loss_weight": 0.6875062465667725, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.05877075195315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.7343688964844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9380, + "valid_token_count": 818.41484375 + }, + { + "actual_token_mask_ratio": 0.3004830539226532, + "epoch": 2.7320337503636893, + "flow_matching_loss": 0.013681671302765607, + "grad_norm": 0.08325918018817902, + "kept_token_count": 572.42265625, + "learning_rate": 2.4676812720151678e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9928621709346771, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071378401946276425, + "mgd_loss_weight": 0.6871729075908661, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.3369140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.85621948242186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9390, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.29998004138469697, + "epoch": 2.7349432644748326, + "flow_matching_loss": 0.014346909988671542, + "grad_norm": 0.09148653596639633, + "kept_token_count": 572.971875, + "learning_rate": 2.466417300226377e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9920533478260041, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007946645049378276, + "mgd_loss_weight": 0.6868395626544952, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.8263641357422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.5065002441406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9400, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.299912890791893, + "epoch": 2.7378527785859763, + "flow_matching_loss": 0.014663801901042462, + "grad_norm": 0.07820779830217361, + "kept_token_count": 572.96875, + "learning_rate": 2.4651521541494394e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.992700845003128, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007299147499725223, + "mgd_loss_weight": 0.6865062117576599, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.5075347900391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.4159637451172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9410, + "valid_token_count": 818.4296875 + }, + { + "actual_token_mask_ratio": 0.3007064193487167, + "epoch": 2.7407622926971196, + "flow_matching_loss": 0.015394814312458038, + "grad_norm": 0.09076496958732605, + "kept_token_count": 572.13671875, + "learning_rate": 2.4638858353216267e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9927003979682922, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007299601333215833, + "mgd_loss_weight": 0.6861728727817535, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.2434906005859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.91456909179686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9420, + "valid_token_count": 818.15703125 + }, + { + "actual_token_mask_ratio": 0.2996224522590637, + "epoch": 2.743671806808263, + "flow_matching_loss": 0.015302196517586708, + "grad_norm": 0.10704518854618073, + "kept_token_count": 573.08828125, + "learning_rate": 2.4626183452816355e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.992220664024353, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007779340539127588, + "mgd_loss_weight": 0.6858395278453827, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.75917358398436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.66628112792966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9430, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.3001516073942184, + "epoch": 2.7465813209194065, + "flow_matching_loss": 0.012895661126822233, + "grad_norm": 0.09118098765611649, + "kept_token_count": 572.80546875, + "learning_rate": 2.4613496855695855e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9927908957004548, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007209117012098432, + "mgd_loss_weight": 0.6855061829090119, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.809130859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.98969116210935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9440, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.30040567815303804, + "epoch": 2.74949083503055, + "flow_matching_loss": 0.01338829742744565, + "grad_norm": 0.07965385913848877, + "kept_token_count": 572.33046875, + "learning_rate": 2.4600798577270183e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9927001118659973, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072998854797333475, + "mgd_loss_weight": 0.685172837972641, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.0045196533203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9807098388672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9450, + "valid_token_count": 818.08984375 + }, + { + "actual_token_mask_ratio": 0.29957341253757475, + "epoch": 2.752400349141693, + "flow_matching_loss": 0.015022718254476786, + "grad_norm": 0.11738407611846924, + "kept_token_count": 573.21171875, + "learning_rate": 2.4588088632968944e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9923932433128357, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007606772100552916, + "mgd_loss_weight": 0.6848394930362701, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 392.1273895263672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.4241180419922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9460, + "valid_token_count": 818.36953125 + }, + { + "actual_token_mask_ratio": 0.3001339495182037, + "epoch": 2.7553098632528368, + "flow_matching_loss": 0.013980103749781847, + "grad_norm": 0.09742606431245804, + "kept_token_count": 572.684375, + "learning_rate": 2.4575367038235916e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9921000063419342, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00790000893175602, + "mgd_loss_weight": 0.6845061480998993, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.4027099609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.35719604492186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9470, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.30075083374977113, + "epoch": 2.7582193773639805, + "flow_matching_loss": 0.013861521705985069, + "grad_norm": 0.10854766517877579, + "kept_token_count": 572.17109375, + "learning_rate": 2.4562633808529043e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.992176228761673, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007823761412873864, + "mgd_loss_weight": 0.6841728091239929, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.1507904052734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0974945068359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9480, + "valid_token_count": 818.26484375 + }, + { + "actual_token_mask_ratio": 0.30004550516605377, + "epoch": 2.7611288914751237, + "flow_matching_loss": 0.015431680623441935, + "grad_norm": 0.10088932514190674, + "kept_token_count": 572.60546875, + "learning_rate": 2.4549888959320396e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9920534014701843, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007946603093296289, + "mgd_loss_weight": 0.6838394582271576, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.76380615234376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1476623535156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9490, + "valid_token_count": 818.06015625 + }, + { + "actual_token_mask_ratio": 0.2996169775724411, + "epoch": 2.764038405586267, + "flow_matching_loss": 0.014624470844864845, + "grad_norm": 0.09706294536590576, + "kept_token_count": 573.25078125, + "learning_rate": 2.4537132506096162e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9920689821243286, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00793101186864078, + "mgd_loss_weight": 0.6835061132907867, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.27739868164065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.0047576904297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9500, + "valid_token_count": 818.48203125 + }, + { + "actual_token_mask_ratio": 0.3003657102584839, + "epoch": 2.7669479196974107, + "flow_matching_loss": 0.012759143207222223, + "grad_norm": 0.0974821224808693, + "kept_token_count": 572.41640625, + "learning_rate": 2.452436446435664e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.992464131116867, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075358724221587185, + "mgd_loss_weight": 0.6831727802753449, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.2978179931641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.56285095214844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9510, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.2995397120714188, + "epoch": 2.769857433808554, + "flow_matching_loss": 0.015930604841560125, + "grad_norm": 0.09346914291381836, + "kept_token_count": 573.20625, + "learning_rate": 2.451158484961621e-05, + "loss": 0.021, + "mgd_cos_sim": 0.9925181627273559, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074818401597440245, + "mgd_loss_weight": 0.682839423418045, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.16029663085936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.23793029785156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9520, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.2999413698911667, + "epoch": 2.7727669479196972, + "flow_matching_loss": 0.01644965326413512, + "grad_norm": 0.1289210021495819, + "kept_token_count": 572.87734375, + "learning_rate": 2.4498793677403304e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.992324835062027, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007675162749364972, + "mgd_loss_weight": 0.6825060844421387, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.4143524169922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.40857849121096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9530, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.3003875195980072, + "epoch": 2.775676462030841, + "flow_matching_loss": 0.015385892428457736, + "grad_norm": 0.0960303470492363, + "kept_token_count": 572.31171875, + "learning_rate": 2.4485990963260398e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9919767439365387, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008023256994783878, + "mgd_loss_weight": 0.6821727454662323, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.34796142578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.43468322753904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9540, + "valid_token_count": 818.04453125 + }, + { + "actual_token_mask_ratio": 0.2996337443590164, + "epoch": 2.778585976141984, + "flow_matching_loss": 0.014493020251393318, + "grad_norm": 0.09224079549312592, + "kept_token_count": 573.17109375, + "learning_rate": 2.4473176722744e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9926379501819611, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007362047769129276, + "mgd_loss_weight": 0.6818393886089325, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.7185089111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.1543273925781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9550, + "valid_token_count": 818.38671875 + }, + { + "actual_token_mask_ratio": 0.3000101149082184, + "epoch": 2.7814954902531275, + "flow_matching_loss": 0.01686596153303981, + "grad_norm": 0.09946519136428833, + "kept_token_count": 572.81875, + "learning_rate": 2.446035097142463e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9928379774093627, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00716202175244689, + "mgd_loss_weight": 0.6815060555934906, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.65028991699216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.6373626708984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9560, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.299469792842865, + "epoch": 2.784405004364271, + "flow_matching_loss": 0.015180468931794167, + "grad_norm": 0.09364635497331619, + "kept_token_count": 573.2015625, + "learning_rate": 2.444751372488677e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9923172175884247, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076827698852866885, + "mgd_loss_weight": 0.6811727046966553, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.7324676513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.85887451171874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9570, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.3001846134662628, + "epoch": 2.7873145184754144, + "flow_matching_loss": 0.016762829199433327, + "grad_norm": 0.07873963564634323, + "kept_token_count": 572.5953125, + "learning_rate": 2.443466499872889e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9921787977218628, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007821211405098439, + "mgd_loss_weight": 0.6808393597602844, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.0938323974609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.91419982910156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9580, + "valid_token_count": 818.2125 + }, + { + "actual_token_mask_ratio": 0.29956451058387756, + "epoch": 2.790224032586558, + "flow_matching_loss": 0.01335758352652192, + "grad_norm": 0.08217805624008179, + "kept_token_count": 573.359375, + "learning_rate": 2.4421804808563413e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9926167309284211, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007383279921486974, + "mgd_loss_weight": 0.680506020784378, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.32395935058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.81707153320315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9590, + "valid_token_count": 818.57578125 + }, + { + "actual_token_mask_ratio": 0.2997247129678726, + "epoch": 2.7931335466977014, + "flow_matching_loss": 0.017082657385617494, + "grad_norm": 0.09942258894443512, + "kept_token_count": 572.93203125, + "learning_rate": 2.440893317001667e-05, + "loss": 0.0221, + "mgd_cos_sim": 0.9926533281803132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007346672890707851, + "mgd_loss_weight": 0.6801726698875428, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.1164031982422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.83745727539065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9600, + "valid_token_count": 818.15078125 + }, + { + "actual_token_mask_ratio": 0.29999361336231234, + "epoch": 2.796043060808845, + "flow_matching_loss": 0.01312703751027584, + "grad_norm": 0.08495210856199265, + "kept_token_count": 572.74375, + "learning_rate": 2.4396050098728925e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9925291776657105, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007470844266936183, + "mgd_loss_weight": 0.6798393309116364, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.58636779785155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.82593994140626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9610, + "valid_token_count": 818.20078125 + }, + { + "actual_token_mask_ratio": 0.2996868550777435, + "epoch": 2.7989525749199884, + "flow_matching_loss": 0.0142715435475111, + "grad_norm": 0.08741825819015503, + "kept_token_count": 572.915625, + "learning_rate": 2.4383155610354314e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9924680173397065, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007531969761475921, + "mgd_loss_weight": 0.6795059859752655, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.45692443847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.7046630859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9620, + "valid_token_count": 818.08515625 + }, + { + "actual_token_mask_ratio": 0.3004615753889084, + "epoch": 2.8018620890311317, + "flow_matching_loss": 0.016708861757069825, + "grad_norm": 0.10138360410928726, + "kept_token_count": 572.4453125, + "learning_rate": 2.437024972056087e-05, + "loss": 0.0217, + "mgd_cos_sim": 0.9926142871379853, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007385722734034062, + "mgd_loss_weight": 0.6791726350784302, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.08897094726564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.0251770019531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9630, + "valid_token_count": 818.31328125 + }, + { + "actual_token_mask_ratio": 0.30051504373550414, + "epoch": 2.8047716031422754, + "flow_matching_loss": 0.014424900244921445, + "grad_norm": 0.1043820008635521, + "kept_token_count": 572.3140625, + "learning_rate": 2.4357332445030452e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9923861682415008, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007613817136734724, + "mgd_loss_weight": 0.6788392961025238, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.3454833984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.48248596191405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9640, + "valid_token_count": 818.19296875 + }, + { + "actual_token_mask_ratio": 0.3001765638589859, + "epoch": 2.8076811172534186, + "flow_matching_loss": 0.01520807547494769, + "grad_norm": 0.09288576245307922, + "kept_token_count": 572.5515625, + "learning_rate": 2.434440379945878e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9923838376998901, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007616170542314648, + "mgd_loss_weight": 0.678505951166153, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.66802673339845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9797637939453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9650, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.3004616409540176, + "epoch": 2.810590631364562, + "flow_matching_loss": 0.015106125641614199, + "grad_norm": 0.09558194130659103, + "kept_token_count": 572.48125, + "learning_rate": 2.4331463799555366e-05, + "loss": 0.0203, + "mgd_cos_sim": 0.9923796653747559, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076203443575650455, + "mgd_loss_weight": 0.6781726062297821, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.1709442138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.8335662841797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9660, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.3004336655139923, + "epoch": 2.8135001454757056, + "flow_matching_loss": 0.01308009196072817, + "grad_norm": 0.1028544008731842, + "kept_token_count": 572.4328125, + "learning_rate": 2.4318512461043544e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9927532553672791, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007246747985482216, + "mgd_loss_weight": 0.6778392612934112, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.9335906982422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.8845520019531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9670, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.29955304265022276, + "epoch": 2.816409659586849, + "flow_matching_loss": 0.013614736776798964, + "grad_norm": 0.10650968551635742, + "kept_token_count": 573.1453125, + "learning_rate": 2.4305549799660397e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9925031781196594, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00749681331217289, + "mgd_loss_weight": 0.6775059163570404, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.43992614746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.45946350097654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9680, + "valid_token_count": 818.25703125 + }, + { + "actual_token_mask_ratio": 0.3000490844249725, + "epoch": 2.8193191736979926, + "flow_matching_loss": 0.014742215815931558, + "grad_norm": 0.10331819951534271, + "kept_token_count": 572.796875, + "learning_rate": 2.429257583115679e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9927075505256653, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007292440021410584, + "mgd_loss_weight": 0.6771725714206696, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.9837188720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.29473266601565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9690, + "valid_token_count": 818.33828125 + }, + { + "actual_token_mask_ratio": 0.30118057131767273, + "epoch": 2.822228687809136, + "flow_matching_loss": 0.015281521342694759, + "grad_norm": 0.0807587057352066, + "kept_token_count": 571.871875, + "learning_rate": 2.4279590571297316e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9927583396434784, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007241661241278052, + "mgd_loss_weight": 0.6768392324447632, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.7602996826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.09999694824216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9700, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.3002555727958679, + "epoch": 2.8251382019202795, + "flow_matching_loss": 0.01467835595831275, + "grad_norm": 0.10091163963079453, + "kept_token_count": 572.43515625, + "learning_rate": 2.4266594035860288e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9924001038074494, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007599894283339381, + "mgd_loss_weight": 0.6765058815479279, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.32261047363284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.0805145263672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9710, + "valid_token_count": 818.0671875 + }, + { + "actual_token_mask_ratio": 0.2988651245832443, + "epoch": 2.828047716031423, + "flow_matching_loss": 0.014119541272521019, + "grad_norm": 0.09599827229976654, + "kept_token_count": 573.43515625, + "learning_rate": 2.425358624063773e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9925941824913025, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007405830454081297, + "mgd_loss_weight": 0.676172536611557, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.7558654785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9958740234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9720, + "valid_token_count": 817.8703125 + }, + { + "actual_token_mask_ratio": 0.3001326709985733, + "epoch": 2.830957230142566, + "flow_matching_loss": 0.012531755492091179, + "grad_norm": 0.08218852430582047, + "kept_token_count": 572.70703125, + "learning_rate": 2.424056720143532e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9927312135696411, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007268780143931508, + "mgd_loss_weight": 0.6758391976356506, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.5550079345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.0441497802734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9730, + "valid_token_count": 818.30859375 + }, + { + "actual_token_mask_ratio": 0.29979332685470583, + "epoch": 2.8338667442537098, + "flow_matching_loss": 0.013579271920025348, + "grad_norm": 0.07624441385269165, + "kept_token_count": 572.91796875, + "learning_rate": 2.422753693407244e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9925843417644501, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007415653625503183, + "mgd_loss_weight": 0.6755058467388153, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.94074096679685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.25975341796874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9740, + "valid_token_count": 818.22109375 + }, + { + "actual_token_mask_ratio": 0.2991510361433029, + "epoch": 2.836776258364853, + "flow_matching_loss": 0.01331524234265089, + "grad_norm": 0.08551632612943649, + "kept_token_count": 573.5234375, + "learning_rate": 2.421449545438209e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9924964487552643, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007503553200513125, + "mgd_loss_weight": 0.6751725077629089, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.4403564453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.5559997558594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9750, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.2992901861667633, + "epoch": 2.8396857724759963, + "flow_matching_loss": 0.014848471712321043, + "grad_norm": 0.10145364701747894, + "kept_token_count": 573.23046875, + "learning_rate": 2.4201442778210887e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9927476942539215, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007252322416752577, + "mgd_loss_weight": 0.6748391628265381, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.7923828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9672424316406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9760, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.30027785897254944, + "epoch": 2.84259528658714, + "flow_matching_loss": 0.015348713751882315, + "grad_norm": 0.11213743686676025, + "kept_token_count": 572.5546875, + "learning_rate": 2.418837892141907e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9920122563838959, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007987742824479938, + "mgd_loss_weight": 0.6745058119297027, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.0779296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.47662048339845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9770, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.29966396987438204, + "epoch": 2.8455048006982833, + "flow_matching_loss": 0.014631831552833319, + "grad_norm": 0.07426643371582031, + "kept_token_count": 573.21171875, + "learning_rate": 2.4175303899880458e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9923550069332123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007644998142495751, + "mgd_loss_weight": 0.6741724789142609, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.6031005859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.88441772460936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9780, + "valid_token_count": 818.48046875 + }, + { + "actual_token_mask_ratio": 0.29984102249145506, + "epoch": 2.848414314809427, + "flow_matching_loss": 0.01457291329279542, + "grad_norm": 0.09116030484437943, + "kept_token_count": 572.97578125, + "learning_rate": 2.4162217729482427e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.992418622970581, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00758135668002069, + "mgd_loss_weight": 0.6738391220569611, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.9348083496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.4472900390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9790, + "valid_token_count": 818.35390625 + }, + { + "actual_token_mask_ratio": 0.300574055314064, + "epoch": 2.8513238289205702, + "flow_matching_loss": 0.013722518365830183, + "grad_norm": 0.08229830861091614, + "kept_token_count": 572.29765625, + "learning_rate": 2.4149120426125918e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9926161944866181, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007383780088275671, + "mgd_loss_weight": 0.6735057830810547, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.6148284912109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9219085693359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9800, + "valid_token_count": 818.23671875 + }, + { + "actual_token_mask_ratio": 0.2999611884355545, + "epoch": 2.854233343031714, + "flow_matching_loss": 0.013625249359756708, + "grad_norm": 0.08687172830104828, + "kept_token_count": 572.8359375, + "learning_rate": 2.413601200572538e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9927258312702179, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007274172501638532, + "mgd_loss_weight": 0.6731724441051483, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.92363891601565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.2819122314453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9810, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.29971919059753416, + "epoch": 2.857142857142857, + "flow_matching_loss": 0.01432179007679224, + "grad_norm": 0.10056998580694199, + "kept_token_count": 573.15234375, + "learning_rate": 2.4122892484208794e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9922816395759583, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007718360284343362, + "mgd_loss_weight": 0.6728390872478485, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.9174072265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.5147277832031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9820, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.2997287780046463, + "epoch": 2.8600523712540005, + "flow_matching_loss": 0.015463404916226864, + "grad_norm": 0.10292281955480576, + "kept_token_count": 572.96484375, + "learning_rate": 2.4109761877517603e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.9923683345317841, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007631664490327239, + "mgd_loss_weight": 0.6725057542324067, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.37734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8795532226562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9830, + "valid_token_count": 818.19921875 + }, + { + "actual_token_mask_ratio": 0.29963105320930483, + "epoch": 2.862961885365144, + "flow_matching_loss": 0.015087911300361157, + "grad_norm": 0.07753323018550873, + "kept_token_count": 573.06875, + "learning_rate": 2.4096620201606737e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9925209999084472, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007479009637609124, + "mgd_loss_weight": 0.6721724033355713, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.4164978027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.6898712158203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9840, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.29918248355388644, + "epoch": 2.8658713994762874, + "flow_matching_loss": 0.012807784974575043, + "grad_norm": 0.06997063755989075, + "kept_token_count": 573.5484375, + "learning_rate": 2.4083467472444573e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9923839390277862, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007616070238873363, + "mgd_loss_weight": 0.6718390583992004, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.25526733398436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2816467285156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9850, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.30008082687854765, + "epoch": 2.8687809135874307, + "flow_matching_loss": 0.012785546574741602, + "grad_norm": 0.0678953155875206, + "kept_token_count": 572.6859375, + "learning_rate": 2.4070303706012912e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9926706910133362, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007329298136755824, + "mgd_loss_weight": 0.6715057194232941, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.7929718017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9983215332031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9860, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.3000256150960922, + "epoch": 2.8716904276985744, + "flow_matching_loss": 0.014742257352918387, + "grad_norm": 0.12823477387428284, + "kept_token_count": 572.5734375, + "learning_rate": 2.4057128918306975e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9929839074611664, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007016100827604533, + "mgd_loss_weight": 0.6711723685264588, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.82709350585935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.79041442871096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9870, + "valid_token_count": 817.98671875 + }, + { + "actual_token_mask_ratio": 0.29948435723781586, + "epoch": 2.8745999418097177, + "flow_matching_loss": 0.013432067260146141, + "grad_norm": 0.09201595932245255, + "kept_token_count": 573.18125, + "learning_rate": 2.4043943125335373e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9924973249435425, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075026748701930044, + "mgd_loss_weight": 0.6708390295505524, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.90138549804686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.35028076171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9880, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.29973597824573517, + "epoch": 2.8775094559208614, + "flow_matching_loss": 0.014779825974255801, + "grad_norm": 0.0789300799369812, + "kept_token_count": 572.98359375, + "learning_rate": 2.403074634312008e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9925956845283508, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00740432096645236, + "mgd_loss_weight": 0.6705056846141815, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.5579772949219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.90903930664064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9890, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.29971790611743926, + "epoch": 2.8804189700320046, + "flow_matching_loss": 0.015144731942564249, + "grad_norm": 0.08250878751277924, + "kept_token_count": 573.11484375, + "learning_rate": 2.4017538587696442e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9928994357585907, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007100569549947977, + "mgd_loss_weight": 0.6701723337173462, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.1918701171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.94012145996095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9900, + "valid_token_count": 818.40546875 + }, + { + "actual_token_mask_ratio": 0.2998272716999054, + "epoch": 2.8833284841431484, + "flow_matching_loss": 0.014062123745679856, + "grad_norm": 0.092986561357975, + "kept_token_count": 572.859375, + "learning_rate": 2.4004319875113118e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9926363289356231, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007363657653331757, + "mgd_loss_weight": 0.6698389947414398, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.9254913330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.96513671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9910, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.29953827559947965, + "epoch": 2.8862379982542916, + "flow_matching_loss": 0.014864942897111178, + "grad_norm": 0.09440125524997711, + "kept_token_count": 573.0296875, + "learning_rate": 2.3991090221432096e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9927007019519806, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007299305871129036, + "mgd_loss_weight": 0.669505649805069, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.4074676513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.95146484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9920, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.29984703063964846, + "epoch": 2.889147512365435, + "flow_matching_loss": 0.014495969284325838, + "grad_norm": 0.07676712423563004, + "kept_token_count": 572.9796875, + "learning_rate": 2.397784964272865e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9925795376300812, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007420469354838133, + "mgd_loss_weight": 0.6691723048686982, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.4178009033203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.7660705566406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9930, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.30027099549770353, + "epoch": 2.8920570264765786, + "flow_matching_loss": 0.01487259790301323, + "grad_norm": 0.0860225185751915, + "kept_token_count": 572.52265625, + "learning_rate": 2.3964598155091335e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9922133922576905, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007786618405953049, + "mgd_loss_weight": 0.6688389599323272, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.06908264160154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.8387817382812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9940, + "valid_token_count": 818.20859375 + }, + { + "actual_token_mask_ratio": 0.30022976994514466, + "epoch": 2.894966540587722, + "flow_matching_loss": 0.014640658721327781, + "grad_norm": 0.08891236782073975, + "kept_token_count": 572.53203125, + "learning_rate": 2.3951335774621952e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.992496234178543, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007503776345402002, + "mgd_loss_weight": 0.6685056209564209, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.7998413085937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.8110595703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9950, + "valid_token_count": 818.16796875 + }, + { + "actual_token_mask_ratio": 0.29978279769420624, + "epoch": 2.897876054698865, + "flow_matching_loss": 0.01536538852378726, + "grad_norm": 0.07421672344207764, + "kept_token_count": 572.828125, + "learning_rate": 2.3938062517435548e-05, + "loss": 0.0206, + "mgd_cos_sim": 0.992107880115509, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007892120582982897, + "mgd_loss_weight": 0.6681722700595856, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.8916961669922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.6120208740234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9960, + "valid_token_count": 818.065625 + }, + { + "actual_token_mask_ratio": 0.30039932727813723, + "epoch": 2.900785568810009, + "flow_matching_loss": 0.016947244014590978, + "grad_norm": 0.08705736696720123, + "kept_token_count": 572.42734375, + "learning_rate": 2.392477839966038e-05, + "loss": 0.0222, + "mgd_cos_sim": 0.9921132385730743, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007886754628270865, + "mgd_loss_weight": 0.6678389310836792, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.29212951660156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.8317901611328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9970, + "valid_token_count": 818.22109375 + }, + { + "actual_token_mask_ratio": 0.3002988398075104, + "epoch": 2.903695082921152, + "flow_matching_loss": 0.01469766665250063, + "grad_norm": 0.10550591349601746, + "kept_token_count": 572.5703125, + "learning_rate": 2.3911483437437904e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9926806330680847, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007319370983168483, + "mgd_loss_weight": 0.6675055861473084, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 447.4534912109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.35875549316404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9980, + "valid_token_count": 818.30546875 + }, + { + "actual_token_mask_ratio": 0.3002459406852722, + "epoch": 2.9066045970322953, + "flow_matching_loss": 0.013872987125068903, + "grad_norm": 0.11310060322284698, + "kept_token_count": 572.42109375, + "learning_rate": 2.3898177646922757e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9927460193634033, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007253991952165961, + "mgd_loss_weight": 0.667172235250473, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.37672729492186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.883154296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 9990, + "valid_token_count": 818.03125 + }, + { + "actual_token_mask_ratio": 0.2996715486049652, + "epoch": 2.909514111143439, + "flow_matching_loss": 0.01307884780690074, + "grad_norm": 0.08034206181764603, + "kept_token_count": 572.90859375, + "learning_rate": 2.3884861044282722e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9925005674362183, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007499439874663949, + "mgd_loss_weight": 0.6668389022350312, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.3367858886719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.6476654052734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10000, + "valid_token_count": 818.05390625 + }, + { + "actual_token_mask_ratio": 0.30011298060417174, + "epoch": 2.9124236252545828, + "flow_matching_loss": 0.013244828954339028, + "grad_norm": 0.09016148746013641, + "kept_token_count": 572.690625, + "learning_rate": 2.3871533645698732e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9923816025257111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007618386764079332, + "mgd_loss_weight": 0.6665055453777313, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.7208312988281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.35804138183596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10010, + "valid_token_count": 818.26171875 + }, + { + "actual_token_mask_ratio": 0.3003761261701584, + "epoch": 2.915333139365726, + "flow_matching_loss": 0.013939594943076372, + "grad_norm": 0.07051566243171692, + "kept_token_count": 572.3859375, + "learning_rate": 2.3858195467364833e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9930479049682617, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00695209251716733, + "mgd_loss_weight": 0.666172206401825, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.0189147949219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6347930908203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10020, + "valid_token_count": 818.13046875 + }, + { + "actual_token_mask_ratio": 0.2995611399412155, + "epoch": 2.9182426534768693, + "flow_matching_loss": 0.012372125592082738, + "grad_norm": 0.08238504081964493, + "kept_token_count": 573.17265625, + "learning_rate": 2.3844846525488166e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9927168488502502, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007283157482743263, + "mgd_loss_weight": 0.6658388674259186, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.7476409912109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.62195739746096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10030, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.3004473179578781, + "epoch": 2.921152167588013, + "flow_matching_loss": 0.013224831596016885, + "grad_norm": 0.07622499763965607, + "kept_token_count": 572.31796875, + "learning_rate": 2.3831486836288957e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9926802575588226, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007319738389924169, + "mgd_loss_weight": 0.6655055105686187, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.31199951171874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.21686096191405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10040, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.30045984089374544, + "epoch": 2.9240616816991563, + "flow_matching_loss": 0.013675618171691894, + "grad_norm": 0.09361939132213593, + "kept_token_count": 572.4015625, + "learning_rate": 2.381811641600048e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9927583754062652, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007241607038304209, + "mgd_loss_weight": 0.6651721775531769, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.49583740234374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.69647216796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10050, + "valid_token_count": 818.25234375 + }, + { + "actual_token_mask_ratio": 0.2998369991779327, + "epoch": 2.9269711958102995, + "flow_matching_loss": 0.014471614826470613, + "grad_norm": 0.09127058833837509, + "kept_token_count": 572.98125, + "learning_rate": 2.3804735280869073e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.99232057929039, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076794341206550595, + "mgd_loss_weight": 0.6648388266563415, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.14422302246095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.2771026611328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10060, + "valid_token_count": 818.35546875 + }, + { + "actual_token_mask_ratio": 0.29986870288848877, + "epoch": 2.9298807099214432, + "flow_matching_loss": 0.014152996707707644, + "grad_norm": 0.0990813598036766, + "kept_token_count": 572.934375, + "learning_rate": 2.3791343447154064e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9927040100097656, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007295988919213414, + "mgd_loss_weight": 0.6645054817199707, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.16455993652346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9961364746094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10070, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.3000716656446457, + "epoch": 2.9327902240325865, + "flow_matching_loss": 0.014763817004859448, + "grad_norm": 0.08413947373628616, + "kept_token_count": 572.5609375, + "learning_rate": 2.37779409311278e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9923911690711975, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007608821149915457, + "mgd_loss_weight": 0.6641721427440643, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.3996520996094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.0378936767578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10080, + "valid_token_count": 818.03046875 + }, + { + "actual_token_mask_ratio": 0.2995650678873062, + "epoch": 2.9356997381437298, + "flow_matching_loss": 0.014547484833747149, + "grad_norm": 0.10563594102859497, + "kept_token_count": 573.25234375, + "learning_rate": 2.3764527749075596e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.992702168226242, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072978366632014515, + "mgd_loss_weight": 0.663838791847229, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.4688446044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.14422607421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10090, + "valid_token_count": 818.42265625 + }, + { + "actual_token_mask_ratio": 0.3008598357439041, + "epoch": 2.9386092522548735, + "flow_matching_loss": 0.01438395194709301, + "grad_norm": 0.1036364808678627, + "kept_token_count": 572.06796875, + "learning_rate": 2.3751103917295735e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9925569593906403, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007443044986575842, + "mgd_loss_weight": 0.6635054528713227, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.70188293457034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.2075897216797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10100, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.299754798412323, + "epoch": 2.9415187663660167, + "flow_matching_loss": 0.016346349753439425, + "grad_norm": 0.09255241602659225, + "kept_token_count": 572.98203125, + "learning_rate": 2.373766945209944e-05, + "loss": 0.0215, + "mgd_cos_sim": 0.9922865152359008, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007713491329923272, + "mgd_loss_weight": 0.6631721079349517, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.41295166015624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.2712066650391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10110, + "valid_token_count": 818.26015625 + }, + { + "actual_token_mask_ratio": 0.2995396077632904, + "epoch": 2.9444282804771604, + "flow_matching_loss": 0.015312703885138034, + "grad_norm": 0.10326223820447922, + "kept_token_count": 573.18125, + "learning_rate": 2.3724224369810856e-05, + "loss": 0.0204, + "mgd_cos_sim": 0.9923316538333893, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007668355759233236, + "mgd_loss_weight": 0.6628387570381165, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.2170684814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.7607147216797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10120, + "valid_token_count": 818.29140625 + }, + { + "actual_token_mask_ratio": 0.299446302652359, + "epoch": 2.9473377945883037, + "flow_matching_loss": 0.014989270363003016, + "grad_norm": 0.09479750692844391, + "kept_token_count": 573.009375, + "learning_rate": 2.3710768686767015e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9927587032318115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072412941139191386, + "mgd_loss_weight": 0.6625054180622101, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.53477172851564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.4276123046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10130, + "valid_token_count": 817.93203125 + }, + { + "actual_token_mask_ratio": 0.2997070014476776, + "epoch": 2.9502473086994474, + "flow_matching_loss": 0.013920467905700206, + "grad_norm": 0.10222141444683075, + "kept_token_count": 572.9578125, + "learning_rate": 2.3697302419317856e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9925853610038757, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007414635736495256, + "mgd_loss_weight": 0.6621720731258393, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.97510681152346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.91518249511716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10140, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.29950572848320006, + "epoch": 2.9531568228105907, + "flow_matching_loss": 0.015030952077358961, + "grad_norm": 0.09890314191579819, + "kept_token_count": 573.20859375, + "learning_rate": 2.3683825583826146e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.992381626367569, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007618365669623017, + "mgd_loss_weight": 0.6618387281894684, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.07021484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.749365234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10150, + "valid_token_count": 818.28984375 + }, + { + "actual_token_mask_ratio": 0.3010158002376556, + "epoch": 2.956066336921734, + "flow_matching_loss": 0.013358835875988007, + "grad_norm": 0.09921837598085403, + "kept_token_count": 572.05859375, + "learning_rate": 2.3670338196667528e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9925496816635132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007450318709015846, + "mgd_loss_weight": 0.6615053832530975, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 386.9118621826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.7751770019531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10160, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.2998739153146744, + "epoch": 2.9589758510328776, + "flow_matching_loss": 0.013094387017190456, + "grad_norm": 0.08787307888269424, + "kept_token_count": 573.0328125, + "learning_rate": 2.3656840274230443e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9928649485111236, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007135059963911772, + "mgd_loss_weight": 0.6611720383167267, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.166650390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0802734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10170, + "valid_token_count": 818.47421875 + }, + { + "actual_token_mask_ratio": 0.30013264119625094, + "epoch": 2.961885365144021, + "flow_matching_loss": 0.014240407664328814, + "grad_norm": 0.11340807378292084, + "kept_token_count": 572.6515625, + "learning_rate": 2.3643331832916133e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9927050113677979, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00729499333538115, + "mgd_loss_weight": 0.6608386933803558, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.99046325683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.6860717773437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10180, + "valid_token_count": 818.22890625 + }, + { + "actual_token_mask_ratio": 0.29936653971672056, + "epoch": 2.964794879255164, + "flow_matching_loss": 0.015690984204411507, + "grad_norm": 0.09193477779626846, + "kept_token_count": 573.22890625, + "learning_rate": 2.362981288913863e-05, + "loss": 0.0207, + "mgd_cos_sim": 0.9924667537212372, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075332446489483115, + "mgd_loss_weight": 0.6605053544044495, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.7768127441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.20977783203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10190, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.2998678356409073, + "epoch": 2.967704393366308, + "flow_matching_loss": 0.01424955641850829, + "grad_norm": 0.08493702858686447, + "kept_token_count": 572.8578125, + "learning_rate": 2.3616283459324737e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9927754878997803, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007224493473768234, + "mgd_loss_weight": 0.6601720035076142, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.3780517578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.7703155517578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10200, + "valid_token_count": 818.21328125 + }, + { + "actual_token_mask_ratio": 0.29917084574699404, + "epoch": 2.970613907477451, + "flow_matching_loss": 0.013034122716635465, + "grad_norm": 0.07472527027130127, + "kept_token_count": 573.40390625, + "learning_rate": 2.3602743559913973e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9926676094532013, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007332390174269676, + "mgd_loss_weight": 0.6598386585712432, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.5506195068359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.28139038085936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10210, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.3000923812389374, + "epoch": 2.973523421588595, + "flow_matching_loss": 0.013449400290846824, + "grad_norm": 0.10025077313184738, + "kept_token_count": 572.79140625, + "learning_rate": 2.35891932073586e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.992666882276535, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007333124289289117, + "mgd_loss_weight": 0.6595053195953369, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.8745147705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.979248046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10220, + "valid_token_count": 818.37734375 + }, + { + "actual_token_mask_ratio": 0.30023455917835234, + "epoch": 2.976432935699738, + "flow_matching_loss": 0.014201260916888715, + "grad_norm": 0.12791065871715546, + "kept_token_count": 572.36484375, + "learning_rate": 2.357563241812357e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9927009046077728, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007299100514501333, + "mgd_loss_weight": 0.6591719686985016, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.02648010253904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.335009765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10230, + "valid_token_count": 817.93671875 + }, + { + "actual_token_mask_ratio": 0.29983610212802886, + "epoch": 2.979342449810882, + "flow_matching_loss": 0.014176464732736349, + "grad_norm": 0.08073386549949646, + "kept_token_count": 572.89375, + "learning_rate": 2.3562061208686522e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9924751102924347, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007524889567866922, + "mgd_loss_weight": 0.6588386297225952, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.1772491455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.357177734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10240, + "valid_token_count": 818.23515625 + }, + { + "actual_token_mask_ratio": 0.2995633751153946, + "epoch": 2.982251963922025, + "flow_matching_loss": 0.01478773234412074, + "grad_norm": 0.10393054038286209, + "kept_token_count": 573.0359375, + "learning_rate": 2.354847959553776e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9924778640270233, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007522155623883009, + "mgd_loss_weight": 0.6585052847862244, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.65032348632815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.12067260742185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10250, + "valid_token_count": 818.11796875 + }, + { + "actual_token_mask_ratio": 0.30033787786960603, + "epoch": 2.9851614780331683, + "flow_matching_loss": 0.01371011408045888, + "grad_norm": 0.09874382615089417, + "kept_token_count": 572.41796875, + "learning_rate": 2.353488759518022e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9923192679882049, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007680734107270837, + "mgd_loss_weight": 0.658171933889389, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.0013031005859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.8108795166016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10260, + "valid_token_count": 818.13203125 + }, + { + "actual_token_mask_ratio": 0.30014767944812776, + "epoch": 2.988070992144312, + "flow_matching_loss": 0.013981499150395394, + "grad_norm": 0.08190526813268661, + "kept_token_count": 572.69453125, + "learning_rate": 2.352128522412946e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.992698585987091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007301413826644421, + "mgd_loss_weight": 0.6578386008739472, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.5403350830078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.92462768554685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10270, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.3000343024730682, + "epoch": 2.9909805062554553, + "flow_matching_loss": 0.014019045140594244, + "grad_norm": 0.09337488561868668, + "kept_token_count": 572.8578125, + "learning_rate": 2.3507672498913653e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9921452164649963, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00785477226600051, + "mgd_loss_weight": 0.6575052440166473, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.5808990478516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.8861511230469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10280, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.29913780391216277, + "epoch": 2.9938900203665986, + "flow_matching_loss": 0.014503398165106773, + "grad_norm": 0.10610944032669067, + "kept_token_count": 573.459375, + "learning_rate": 2.3494049436073537e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9923453807830811, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007654630113393068, + "mgd_loss_weight": 0.657171905040741, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.8202789306641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.85887145996094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10290, + "valid_token_count": 818.21796875 + }, + { + "actual_token_mask_ratio": 0.30012576282024384, + "epoch": 2.9967995344777423, + "flow_matching_loss": 0.01422774400562048, + "grad_norm": 0.0840468481183052, + "kept_token_count": 572.665625, + "learning_rate": 2.348041605216242e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9931634962558746, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00683652120642364, + "mgd_loss_weight": 0.6568385660648346, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.16656188964845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.73675537109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10300, + "valid_token_count": 818.24765625 + }, + { + "actual_token_mask_ratio": 0.3002043545246124, + "epoch": 2.9997090485888855, + "flow_matching_loss": 0.014543017279356718, + "grad_norm": 0.0833943709731102, + "kept_token_count": 572.5578125, + "learning_rate": 2.3466772363746146e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.992294180393219, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00770583339035511, + "mgd_loss_weight": 0.6565052092075347, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.9305450439453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.00556030273435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10310, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.29947659373283386, + "epoch": 3.0026185627000292, + "flow_matching_loss": 0.01350017311051488, + "grad_norm": 0.08344680070877075, + "kept_token_count": 573.1373657226562, + "learning_rate": 2.345311838740309e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9924420952796936, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007557905465364456, + "mgd_loss_weight": 0.6561718761920929, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.76767578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9821044921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10320, + "valid_token_count": 818.1562866210937 + }, + { + "actual_token_mask_ratio": 0.29999403059482577, + "epoch": 3.0055280768111725, + "flow_matching_loss": 0.013296797312796116, + "grad_norm": 0.10601203143596649, + "kept_token_count": 572.91171875, + "learning_rate": 2.343945413972411e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9925156950950622, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007484316686168313, + "mgd_loss_weight": 0.6558385252952575, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.544482421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.4750610351563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10330, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.3003135472536087, + "epoch": 3.0084375909223158, + "flow_matching_loss": 0.013787064421921968, + "grad_norm": 0.07286805659532547, + "kept_token_count": 572.52734375, + "learning_rate": 2.3425779637312568e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.991960734128952, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008039271272718907, + "mgd_loss_weight": 0.6555051803588867, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.1736297607422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.2419799804687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10340, + "valid_token_count": 818.26640625 + }, + { + "actual_token_mask_ratio": 0.2992364376783371, + "epoch": 3.0113471050334595, + "flow_matching_loss": 0.014865655731409789, + "grad_norm": 0.09574443846940994, + "kept_token_count": 573.49609375, + "learning_rate": 2.3412094896784264e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9920647203922272, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007935291295871139, + "mgd_loss_weight": 0.6551718413829803, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.95102844238284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.25140686035155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10350, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.2996372848749161, + "epoch": 3.0142566191446027, + "flow_matching_loss": 0.016056319419294595, + "grad_norm": 0.0948672890663147, + "kept_token_count": 573.128125, + "learning_rate": 2.339839993476745e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.992599219083786, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007400773698464036, + "mgd_loss_weight": 0.6548384904861451, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.3980682373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.85405883789065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10360, + "valid_token_count": 818.32265625 + }, + { + "actual_token_mask_ratio": 0.30024070739746095, + "epoch": 3.0171661332557465, + "flow_matching_loss": 0.014735890552401543, + "grad_norm": 0.09417589753866196, + "kept_token_count": 572.68984375, + "learning_rate": 2.3384694767902804e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9930783212184906, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006921677757054568, + "mgd_loss_weight": 0.6545051515102387, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.6942962646484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.5203796386719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10370, + "valid_token_count": 818.40703125 + }, + { + "actual_token_mask_ratio": 0.29915540516376493, + "epoch": 3.0200756473668897, + "flow_matching_loss": 0.012192024290561676, + "grad_norm": 0.08652709424495697, + "kept_token_count": 573.5234375, + "learning_rate": 2.3370979412843384e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9923769414424897, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007623065076768398, + "mgd_loss_weight": 0.6541718065738678, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.41595458984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.74366455078126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10380, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.30023974776268003, + "epoch": 3.022985161478033, + "flow_matching_loss": 0.013270250894129277, + "grad_norm": 0.08895290642976761, + "kept_token_count": 572.778125, + "learning_rate": 2.3357253886254648e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9925762832164764, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007423720369115472, + "mgd_loss_weight": 0.6538384616374969, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.53004455566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.83909606933594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10390, + "valid_token_count": 818.53359375 + }, + { + "actual_token_mask_ratio": 0.29985537230968473, + "epoch": 3.0258946755891767, + "flow_matching_loss": 0.01313121048733592, + "grad_norm": 0.1462784856557846, + "kept_token_count": 572.8296875, + "learning_rate": 2.3343518204814407e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9925825893878937, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007417411031201482, + "mgd_loss_weight": 0.6535051167011261, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.83959045410154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6551879882812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10400, + "valid_token_count": 818.16171875 + }, + { + "actual_token_mask_ratio": 0.29988453090190886, + "epoch": 3.02880418970032, + "flow_matching_loss": 0.014450683444738387, + "grad_norm": 0.11100669205188751, + "kept_token_count": 572.85390625, + "learning_rate": 2.3329772385212798e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9921426951885224, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007857298105955124, + "mgd_loss_weight": 0.6531717777252197, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.96266174316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.2758422851563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10410, + "valid_token_count": 818.23359375 + }, + { + "actual_token_mask_ratio": 0.29986990690231324, + "epoch": 3.0317137038114637, + "flow_matching_loss": 0.01400980157777667, + "grad_norm": 0.09193316847085953, + "kept_token_count": 572.83125, + "learning_rate": 2.3316016444152296e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9926074802875519, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007392506720498205, + "mgd_loss_weight": 0.6528384268283844, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.3208435058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1029571533203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10420, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.29970845878124236, + "epoch": 3.034623217922607, + "flow_matching_loss": 0.014248610194772481, + "grad_norm": 0.1379977911710739, + "kept_token_count": 573.12734375, + "learning_rate": 2.3302250398347668e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9924717366695404, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007528247032314539, + "mgd_loss_weight": 0.6525050818920135, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.8463470458984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.0512329101563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10430, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.3002659499645233, + "epoch": 3.03753273203375, + "flow_matching_loss": 0.014102501515299081, + "grad_norm": 0.09334636479616165, + "kept_token_count": 572.5171875, + "learning_rate": 2.328847426452595e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9929506957530976, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070493079256266356, + "mgd_loss_weight": 0.6521717429161071, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.74754638671874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.0291442871094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10440, + "valid_token_count": 818.19140625 + }, + { + "actual_token_mask_ratio": 0.3007392376661301, + "epoch": 3.040442246144894, + "flow_matching_loss": 0.013952773064374924, + "grad_norm": 0.09671969711780548, + "kept_token_count": 572.3421875, + "learning_rate": 2.3274688059426456e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9926465749740601, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007353442814201116, + "mgd_loss_weight": 0.6518383920192719, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.26885986328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.3418792724609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10450, + "valid_token_count": 818.49140625 + }, + { + "actual_token_mask_ratio": 0.29975551664829253, + "epoch": 3.043351760256037, + "flow_matching_loss": 0.015311010647565126, + "grad_norm": 0.10629015415906906, + "kept_token_count": 572.89296875, + "learning_rate": 2.3260891799800715e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9924656689167023, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007534336159005761, + "mgd_loss_weight": 0.6515050530433655, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.1286285400391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.40370178222656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10460, + "valid_token_count": 818.13359375 + }, + { + "actual_token_mask_ratio": 0.29958280026912687, + "epoch": 3.046261274367181, + "flow_matching_loss": 0.014453156385570765, + "grad_norm": 0.09792477637529373, + "kept_token_count": 573.13125, + "learning_rate": 2.3247085502412482e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9926958560943604, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073041412979364395, + "mgd_loss_weight": 0.6511717081069947, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.56633911132815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6779266357422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10470, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.30000234842300416, + "epoch": 3.049170788478324, + "flow_matching_loss": 0.013696963526308536, + "grad_norm": 0.09245012700557709, + "kept_token_count": 572.9046875, + "learning_rate": 2.3233269184037717e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9926923155784607, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007307678367942572, + "mgd_loss_weight": 0.6508383572101593, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.88880310058596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.90332946777346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10480, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.2995109349489212, + "epoch": 3.0520803025894674, + "flow_matching_loss": 0.015800748020410538, + "grad_norm": 0.08945681899785995, + "kept_token_count": 573.2, + "learning_rate": 2.3219442861464547e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9921547472476959, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007845254242420196, + "mgd_loss_weight": 0.6505050241947175, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 394.5916046142578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.42243041992185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10490, + "valid_token_count": 818.28828125 + }, + { + "actual_token_mask_ratio": 0.29964030385017393, + "epoch": 3.054989816700611, + "flow_matching_loss": 0.013262417167425156, + "grad_norm": 0.07411216199398041, + "kept_token_count": 573.003125, + "learning_rate": 2.3205606551493255e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9928264498710633, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007173559954389929, + "mgd_loss_weight": 0.6501716673374176, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.0893890380859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3507537841797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10500, + "valid_token_count": 818.15234375 + }, + { + "actual_token_mask_ratio": 0.30148531794548034, + "epoch": 3.0578993308117544, + "flow_matching_loss": 0.014646316226571799, + "grad_norm": 0.09263354539871216, + "kept_token_count": 571.55, + "learning_rate": 2.319176027093627e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9923246145248413, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007675385242328048, + "mgd_loss_weight": 0.6498383283615112, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.781103515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1261871337891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10510, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.30064420104026796, + "epoch": 3.060808844922898, + "flow_matching_loss": 0.014995876979082823, + "grad_norm": 0.10968364775180817, + "kept_token_count": 572.37578125, + "learning_rate": 2.3177904036618117e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9922960102558136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007703994633629918, + "mgd_loss_weight": 0.6495049893856049, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.2292205810547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9551696777344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10520, + "valid_token_count": 818.43203125 + }, + { + "actual_token_mask_ratio": 0.3001121193170547, + "epoch": 3.0637183590340413, + "flow_matching_loss": 0.013233903702348471, + "grad_norm": 0.08777463436126709, + "kept_token_count": 572.521875, + "learning_rate": 2.316403786537544e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9924678206443787, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007532175071537495, + "mgd_loss_weight": 0.649171632528305, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.36745300292966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.1875335693359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10530, + "valid_token_count": 818.01796875 + }, + { + "actual_token_mask_ratio": 0.3010972738265991, + "epoch": 3.0666278731451846, + "flow_matching_loss": 0.014722701627761126, + "grad_norm": 0.09921426326036453, + "kept_token_count": 571.98984375, + "learning_rate": 2.3150161774056933e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9927271783351899, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007272830931469798, + "mgd_loss_weight": 0.6488382995128632, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.61631164550784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.64683532714844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10540, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.30026734471321104, + "epoch": 3.0695373872563283, + "flow_matching_loss": 0.013177568092942238, + "grad_norm": 0.11275660991668701, + "kept_token_count": 572.6703125, + "learning_rate": 2.313627577952336e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9925136625766754, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007486351253464818, + "mgd_loss_weight": 0.6485049486160278, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.81443481445314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.2786895751953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10550, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.30006004869937897, + "epoch": 3.0724469013674716, + "flow_matching_loss": 0.01340457210317254, + "grad_norm": 0.1047690212726593, + "kept_token_count": 572.73828125, + "learning_rate": 2.312237989864752e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9921745777130127, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007825426943600177, + "mgd_loss_weight": 0.648171603679657, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.97413940429686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.0785247802734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10560, + "valid_token_count": 818.26640625 + }, + { + "actual_token_mask_ratio": 0.30036008954048155, + "epoch": 3.0753564154786153, + "flow_matching_loss": 0.015249627828598022, + "grad_norm": 0.07955963909626007, + "kept_token_count": 572.5390625, + "learning_rate": 2.3108474148314205e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9923211753368377, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0076788478065282105, + "mgd_loss_weight": 0.6478382647037506, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.9514862060547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.6173095703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10570, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.300571209192276, + "epoch": 3.0782659295897585, + "flow_matching_loss": 0.013502646330744028, + "grad_norm": 0.07360154390335083, + "kept_token_count": 572.08984375, + "learning_rate": 2.309455854542023e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9928210973739624, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071789045352488754, + "mgd_loss_weight": 0.6475049138069153, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.08461303710936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.73642578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10580, + "valid_token_count": 817.9375 + }, + { + "actual_token_mask_ratio": 0.2995726644992828, + "epoch": 3.081175443700902, + "flow_matching_loss": 0.013734134379774332, + "grad_norm": 0.09842627495527267, + "kept_token_count": 573.328125, + "learning_rate": 2.308063310687435e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9922905027866363, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007709496142342687, + "mgd_loss_weight": 0.647171574831009, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.0883392333984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.88646240234374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10590, + "valid_token_count": 818.54375 + }, + { + "actual_token_mask_ratio": 0.30036247670650484, + "epoch": 3.0840849578120455, + "flow_matching_loss": 0.014346874691545964, + "grad_norm": 0.09978694468736649, + "kept_token_count": 572.53203125, + "learning_rate": 2.306669784959729e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.992311418056488, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007688576728105545, + "mgd_loss_weight": 0.646838229894638, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.24834899902345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.2466735839844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10600, + "valid_token_count": 818.32890625 + }, + { + "actual_token_mask_ratio": 0.2997140437364578, + "epoch": 3.0869944719231888, + "flow_matching_loss": 0.013177049811929464, + "grad_norm": 0.10173480957746506, + "kept_token_count": 572.89296875, + "learning_rate": 2.3052752790521703e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9921164274215698, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00788357206620276, + "mgd_loss_weight": 0.6465048789978027, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.41937866210935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.16014709472654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10610, + "valid_token_count": 818.0828125 + }, + { + "actual_token_mask_ratio": 0.3001676172018051, + "epoch": 3.0899039860343325, + "flow_matching_loss": 0.014375576563179494, + "grad_norm": 0.08836157619953156, + "kept_token_count": 572.67578125, + "learning_rate": 2.3038797946592152e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9925365209579468, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00746347508393228, + "mgd_loss_weight": 0.6461715400218964, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.7074737548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.68833618164064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10620, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.29985010623931885, + "epoch": 3.0928135001454757, + "flow_matching_loss": 0.013771765027195215, + "grad_norm": 0.07515139877796173, + "kept_token_count": 572.80625, + "learning_rate": 2.3024833334765087e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9926405131816864, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007359486958011985, + "mgd_loss_weight": 0.6458381950855255, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.1230895996094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.41603698730466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10630, + "valid_token_count": 818.11875 + }, + { + "actual_token_mask_ratio": 0.30052319169044495, + "epoch": 3.095723014256619, + "flow_matching_loss": 0.012842011731117963, + "grad_norm": 0.08149448037147522, + "kept_token_count": 572.24296875, + "learning_rate": 2.301085897200883e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9929118990898133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007088098581880331, + "mgd_loss_weight": 0.6455048501491547, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.49660339355466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.2608581542969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10640, + "valid_token_count": 818.1 + }, + { + "actual_token_mask_ratio": 0.3003105401992798, + "epoch": 3.0986325283677627, + "flow_matching_loss": 0.014529284927994013, + "grad_norm": 0.09140302240848541, + "kept_token_count": 572.6078125, + "learning_rate": 2.299687487530354e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9927845954895019, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007215397618710995, + "mgd_loss_weight": 0.6451715052127838, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.98276672363284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9132568359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10650, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.30020969212055204, + "epoch": 3.101542042478906, + "flow_matching_loss": 0.014254594594240189, + "grad_norm": 0.09535177052021027, + "kept_token_count": 572.6328125, + "learning_rate": 2.2982881061641228e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9926364302635193, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007363575091585517, + "mgd_loss_weight": 0.6448381602764129, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.17238159179686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.0560333251953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10660, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.30024178624153136, + "epoch": 3.1044515565900497, + "flow_matching_loss": 0.014294761512428523, + "grad_norm": 0.08346285670995712, + "kept_token_count": 572.48671875, + "learning_rate": 2.296887754802569e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9924774944782258, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007522498536854982, + "mgd_loss_weight": 0.6445048153400421, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.3614135742188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8047760009766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10670, + "valid_token_count": 818.11640625 + }, + { + "actual_token_mask_ratio": 0.29999049603939054, + "epoch": 3.107361070701193, + "flow_matching_loss": 0.014517890103161336, + "grad_norm": 0.0747387483716011, + "kept_token_count": 572.73515625, + "learning_rate": 2.295486435147251e-05, + "loss": 0.0194, + "mgd_cos_sim": 0.9924019396305084, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007598058553412557, + "mgd_loss_weight": 0.6441714763641357, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.2492645263672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.56988525390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10680, + "valid_token_count": 818.18671875 + }, + { + "actual_token_mask_ratio": 0.30027989447116854, + "epoch": 3.110270584812336, + "flow_matching_loss": 0.01424032635986805, + "grad_norm": 0.09008654206991196, + "kept_token_count": 572.46484375, + "learning_rate": 2.294084148900905e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9924809038639069, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007519099535420537, + "mgd_loss_weight": 0.6438381254673005, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.27626037597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0350769042969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10690, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.300221660733223, + "epoch": 3.11318009892348, + "flow_matching_loss": 0.013714074902236462, + "grad_norm": 0.10029441863298416, + "kept_token_count": 572.596875, + "learning_rate": 2.292680897767441e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9919589459896088, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.008041043439880014, + "mgd_loss_weight": 0.6435047805309295, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.21185302734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.77910766601565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10700, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.29940813183784487, + "epoch": 3.116089613034623, + "flow_matching_loss": 0.015070677828043699, + "grad_norm": 0.09920850396156311, + "kept_token_count": 573.1234375, + "learning_rate": 2.291276683451941e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9922585904598236, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077413962688297035, + "mgd_loss_weight": 0.6431714415550231, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.3935882568359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.43646850585935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10710, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.30023899078369143, + "epoch": 3.1189991271457664, + "flow_matching_loss": 0.013313102908432484, + "grad_norm": 0.09430774301290512, + "kept_token_count": 572.61640625, + "learning_rate": 2.2898715076606584e-05, + "loss": 0.018, + "mgd_cos_sim": 0.992675918340683, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00732408445328474, + "mgd_loss_weight": 0.6428380906581879, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 381.7628204345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.3088806152344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10720, + "valid_token_count": 818.29921875 + }, + { + "actual_token_mask_ratio": 0.29950576424598696, + "epoch": 3.12190864125691, + "flow_matching_loss": 0.011972033418715, + "grad_norm": 0.11038397252559662, + "kept_token_count": 573.21640625, + "learning_rate": 2.2884653721010135e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9929321110248566, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007067890930920839, + "mgd_loss_weight": 0.6425047516822815, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.23285217285155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.39122314453124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10730, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.30062692165374755, + "epoch": 3.1248181553680534, + "flow_matching_loss": 0.014779041148722171, + "grad_norm": 0.08161196112632751, + "kept_token_count": 572.159375, + "learning_rate": 2.2870582784815946e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9922927141189575, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0077072915621101854, + "mgd_loss_weight": 0.6421714067459107, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 388.3638397216797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.7784423828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10740, + "valid_token_count": 818.10078125 + }, + { + "actual_token_mask_ratio": 0.29901981055736543, + "epoch": 3.127727669479197, + "flow_matching_loss": 0.012821173667907715, + "grad_norm": 0.09846723824739456, + "kept_token_count": 573.625, + "learning_rate": 2.2856502285121525e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.99287930727005, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007120703905820846, + "mgd_loss_weight": 0.6418380558490753, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.505224609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.6042175292969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10750, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.29973557591438293, + "epoch": 3.1306371835903404, + "flow_matching_loss": 0.014397005829960108, + "grad_norm": 0.09292584657669067, + "kept_token_count": 573.0234375, + "learning_rate": 2.2842412239036004e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9926103472709655, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007389649422839284, + "mgd_loss_weight": 0.6415047228336335, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.3763732910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.45763244628904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10760, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.3002865850925446, + "epoch": 3.133546697701484, + "flow_matching_loss": 0.013555786665529012, + "grad_norm": 0.07599956542253494, + "kept_token_count": 572.62578125, + "learning_rate": 2.2828312663680125e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9927435159683228, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072564879897981886, + "mgd_loss_weight": 0.6411713659763336, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.3775360107422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9974609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10770, + "valid_token_count": 818.37578125 + }, + { + "actual_token_mask_ratio": 0.30066874623298645, + "epoch": 3.1364562118126273, + "flow_matching_loss": 0.01442967401817441, + "grad_norm": 0.08951536566019058, + "kept_token_count": 572.27578125, + "learning_rate": 2.2814203576186194e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9925018787384033, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007498111948370934, + "mgd_loss_weight": 0.6408380270004272, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.88231506347654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.47096862792966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10780, + "valid_token_count": 818.31484375 + }, + { + "actual_token_mask_ratio": 0.3004618793725967, + "epoch": 3.1393657259237706, + "flow_matching_loss": 0.013838099222630263, + "grad_norm": 0.10370893776416779, + "kept_token_count": 572.4875, + "learning_rate": 2.2800084993698093e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9926345586776734, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007365445559844375, + "mgd_loss_weight": 0.6405046880245209, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 401.4519287109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0189636230469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10790, + "valid_token_count": 818.38203125 + }, + { + "actual_token_mask_ratio": 0.2997018814086914, + "epoch": 3.1422752400349143, + "flow_matching_loss": 0.013531199283897876, + "grad_norm": 0.08630011230707169, + "kept_token_count": 573.1625, + "learning_rate": 2.2785956933371224e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.992624944448471, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007375044701620937, + "mgd_loss_weight": 0.640171331167221, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.76829223632814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.1102722167969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10800, + "valid_token_count": 818.46171875 + }, + { + "actual_token_mask_ratio": 0.3012795984745026, + "epoch": 3.1451847541460576, + "flow_matching_loss": 0.014618448540568351, + "grad_norm": 0.06901674717664719, + "kept_token_count": 571.721875, + "learning_rate": 2.2771819412372505e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9926603555679321, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007339661940932274, + "mgd_loss_weight": 0.6398379981517792, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.51377258300784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.6504150390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10810, + "valid_token_count": 818.23828125 + }, + { + "actual_token_mask_ratio": 0.3000893950462341, + "epoch": 3.148094268257201, + "flow_matching_loss": 0.013108388893306256, + "grad_norm": 0.10290700942277908, + "kept_token_count": 572.6734375, + "learning_rate": 2.2757672447880374e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9927500367164612, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00724996356293559, + "mgd_loss_weight": 0.6395046532154083, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.4057922363281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.08787536621094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10820, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.3002027541399002, + "epoch": 3.1510037823683446, + "flow_matching_loss": 0.01324644424021244, + "grad_norm": 0.08295915275812149, + "kept_token_count": 572.7109375, + "learning_rate": 2.2743516057084712e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9928577184677124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007142287911847234, + "mgd_loss_weight": 0.639171302318573, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.4974395751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.2153747558594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10830, + "valid_token_count": 818.39140625 + }, + { + "actual_token_mask_ratio": 0.29963878989219667, + "epoch": 3.153913296479488, + "flow_matching_loss": 0.013566224649548531, + "grad_norm": 0.09116274118423462, + "kept_token_count": 572.9671875, + "learning_rate": 2.2729350257186877e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9924843788146973, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075156140141189095, + "mgd_loss_weight": 0.6388379633426666, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.3388336181641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.88624267578126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10840, + "valid_token_count": 818.10234375 + }, + { + "actual_token_mask_ratio": 0.2997864425182343, + "epoch": 3.1568228105906315, + "flow_matching_loss": 0.01293633207678795, + "grad_norm": 0.10666243731975555, + "kept_token_count": 573.09921875, + "learning_rate": 2.2715175065399655e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9925800919532776, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007419898454099893, + "mgd_loss_weight": 0.6385046184062958, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.61705322265624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.32790222167966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10850, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.2994682192802429, + "epoch": 3.159732324701775, + "flow_matching_loss": 0.01300894059240818, + "grad_norm": 0.09227107465267181, + "kept_token_count": 573.24140625, + "learning_rate": 2.270099049894723e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9926408231258392, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007359182415530086, + "mgd_loss_weight": 0.638171273469925, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.0513671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.4840393066406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10860, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.29993650019168855, + "epoch": 3.162641838812918, + "flow_matching_loss": 0.013041118811815976, + "grad_norm": 0.08312058448791504, + "kept_token_count": 572.6859375, + "learning_rate": 2.268679657506519e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9928041279315949, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007195877796038985, + "mgd_loss_weight": 0.637837928533554, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.8374389648437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.4110992431641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10870, + "valid_token_count": 818.05234375 + }, + { + "actual_token_mask_ratio": 0.2997645169496536, + "epoch": 3.1655513529240618, + "flow_matching_loss": 0.01350225079804659, + "grad_norm": 0.08263193070888519, + "kept_token_count": 572.9046875, + "learning_rate": 2.2672593311000498e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9926885962486267, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007311407523229718, + "mgd_loss_weight": 0.6375045835971832, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.2337158203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.6215454101563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10880, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.3006370812654495, + "epoch": 3.168460867035205, + "flow_matching_loss": 0.013082838896661998, + "grad_norm": 0.07791130989789963, + "kept_token_count": 572.2953125, + "learning_rate": 2.2658380724011452e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9925345659255982, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007465422432869673, + "mgd_loss_weight": 0.6371712386608124, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.78538818359374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.3990783691406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10890, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.2997697412967682, + "epoch": 3.1713703811463487, + "flow_matching_loss": 0.013548286817967891, + "grad_norm": 0.10125747323036194, + "kept_token_count": 572.93203125, + "learning_rate": 2.264415883136769e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9925604820251465, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007439529988914728, + "mgd_loss_weight": 0.636837899684906, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.2677917480469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.0139495849609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10900, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.2999206006526947, + "epoch": 3.174279895257492, + "flow_matching_loss": 0.01364857517182827, + "grad_norm": 0.08837603032588959, + "kept_token_count": 572.9375, + "learning_rate": 2.262992765035015e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9928037285804748, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071962727699428795, + "mgd_loss_weight": 0.6365045487880707, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.6276916503906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.1455505371094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10910, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.30055356919765475, + "epoch": 3.1771894093686353, + "flow_matching_loss": 0.013491971790790558, + "grad_norm": 0.07927948236465454, + "kept_token_count": 572.20546875, + "learning_rate": 2.2615687198251062e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9928620874881744, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007137927925214171, + "mgd_loss_weight": 0.6361712038516998, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.3236877441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.4237396240234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10920, + "valid_token_count": 818.08046875 + }, + { + "actual_token_mask_ratio": 0.30004894733428955, + "epoch": 3.180098923479779, + "flow_matching_loss": 0.014227588009089232, + "grad_norm": 0.08430008590221405, + "kept_token_count": 572.73828125, + "learning_rate": 2.2601437492373914e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9922745823860168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007725412026047707, + "mgd_loss_weight": 0.6358378648757934, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 393.1200866699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.51491394042966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10930, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.29987830519676206, + "epoch": 3.1830084375909222, + "flow_matching_loss": 0.01474956665188074, + "grad_norm": 0.08159322291612625, + "kept_token_count": 573.06328125, + "learning_rate": 2.2587178550033444e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9925442576408386, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00745573933236301, + "mgd_loss_weight": 0.6355045139789581, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.8914001464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.7518615722656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10940, + "valid_token_count": 818.51953125 + }, + { + "actual_token_mask_ratio": 0.30035158395767214, + "epoch": 3.185917951702066, + "flow_matching_loss": 0.013418739289045334, + "grad_norm": 0.08144757151603699, + "kept_token_count": 572.48984375, + "learning_rate": 2.2572910388555615e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9924465954303742, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007553419470787049, + "mgd_loss_weight": 0.6351711750030518, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 390.2261535644531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.48671264648436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10950, + "valid_token_count": 818.253125 + }, + { + "actual_token_mask_ratio": 0.2999988108873367, + "epoch": 3.188827465813209, + "flow_matching_loss": 0.01289477376267314, + "grad_norm": 0.09927434474229813, + "kept_token_count": 572.81640625, + "learning_rate": 2.255863302527758e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9927311182022095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00726887141354382, + "mgd_loss_weight": 0.6348378300666809, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.5271759033203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 373.26544189453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10960, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.2999974310398102, + "epoch": 3.1917369799243525, + "flow_matching_loss": 0.01549711897969246, + "grad_norm": 0.11013124883174896, + "kept_token_count": 572.9140625, + "learning_rate": 2.2544346477547685e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9927384495735169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007261549215763807, + "mgd_loss_weight": 0.6345044791698455, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.8879150390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.65146484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10970, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.30004628002643585, + "epoch": 3.194646494035496, + "flow_matching_loss": 0.011720982287079096, + "grad_norm": 0.08497531712055206, + "kept_token_count": 572.82734375, + "learning_rate": 2.253005076272544e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9926333427429199, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007366665592417121, + "mgd_loss_weight": 0.6341711461544037, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.94590759277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.6183380126953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10980, + "valid_token_count": 818.37734375 + }, + { + "actual_token_mask_ratio": 0.29949443638324735, + "epoch": 3.1975560081466394, + "flow_matching_loss": 0.015750588569790124, + "grad_norm": 0.09386029839515686, + "kept_token_count": 573.24296875, + "learning_rate": 2.2515745898181477e-05, + "loss": 0.0205, + "mgd_cos_sim": 0.9925059676170349, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007494028424844146, + "mgd_loss_weight": 0.6338377892971039, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.5328674316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.76270751953126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 10990, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.3005539894104004, + "epoch": 3.200465522257783, + "flow_matching_loss": 0.01476603951305151, + "grad_norm": 0.07753652334213257, + "kept_token_count": 572.221875, + "learning_rate": 2.250143190129756e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9925233244895935, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074766643811017275, + "mgd_loss_weight": 0.6335044503211975, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.2287933349609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.3076873779297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11000, + "valid_token_count": 818.10078125 + }, + { + "actual_token_mask_ratio": 0.3001136392354965, + "epoch": 3.2033750363689264, + "flow_matching_loss": 0.013377304840832948, + "grad_norm": 0.08791203796863556, + "kept_token_count": 572.62890625, + "learning_rate": 2.2487108789466547e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9926020681858063, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007397931953892112, + "mgd_loss_weight": 0.6331711113452911, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.6604797363281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.75714111328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11010, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.29933592975139617, + "epoch": 3.2062845504800697, + "flow_matching_loss": 0.013923869468271732, + "grad_norm": 0.09764537960290909, + "kept_token_count": 573.26953125, + "learning_rate": 2.247277658009236e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9928802967071533, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007119707670062781, + "mgd_loss_weight": 0.6328377544879913, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.80304260253905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.2400390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11020, + "valid_token_count": 818.17734375 + }, + { + "actual_token_mask_ratio": 0.29999782145023346, + "epoch": 3.2091940645912134, + "flow_matching_loss": 0.012396353017538785, + "grad_norm": 0.08099430054426193, + "kept_token_count": 572.7421875, + "learning_rate": 2.245843529059e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9928443551063537, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007155636418610811, + "mgd_loss_weight": 0.6325044214725495, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.7681549072266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.37489318847656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11030, + "valid_token_count": 818.19765625 + }, + { + "actual_token_mask_ratio": 0.29984417259693147, + "epoch": 3.2121035787023566, + "flow_matching_loss": 0.01309740161523223, + "grad_norm": 0.1310371309518814, + "kept_token_count": 572.86796875, + "learning_rate": 2.244408493838547e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9925191581249238, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007480844017118215, + "mgd_loss_weight": 0.6321710705757141, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.281884765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.9284240722656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11040, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.29989284873008726, + "epoch": 3.2150130928135003, + "flow_matching_loss": 0.013949818350374699, + "grad_norm": 0.07890202105045319, + "kept_token_count": 572.9640625, + "learning_rate": 2.242972554091581e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9923623144626618, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007637687074020505, + "mgd_loss_weight": 0.6318377256393433, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.60191650390624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.42103271484376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11050, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.29946519434452057, + "epoch": 3.2179226069246436, + "flow_matching_loss": 0.01278103943914175, + "grad_norm": 0.0810532420873642, + "kept_token_count": 573.365625, + "learning_rate": 2.2415357115629045e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9929529666900635, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007047034380957484, + "mgd_loss_weight": 0.6315043866634369, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.96483459472654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.85915222167966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11060, + "valid_token_count": 818.46796875 + }, + { + "actual_token_mask_ratio": 0.3007821500301361, + "epoch": 3.220832121035787, + "flow_matching_loss": 0.013550741877406836, + "grad_norm": 0.09571616351604462, + "kept_token_count": 572.2703125, + "learning_rate": 2.2400979679984162e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9923008620738983, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007699145283550024, + "mgd_loss_weight": 0.6311710357666016, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.2751800537109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.55711975097654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11070, + "valid_token_count": 818.4453125 + }, + { + "actual_token_mask_ratio": 0.30051960349082946, + "epoch": 3.2237416351469306, + "flow_matching_loss": 0.01619565598666668, + "grad_norm": 0.09665276110172272, + "kept_token_count": 572.34765625, + "learning_rate": 2.2386593251451096e-05, + "loss": 0.0209, + "mgd_cos_sim": 0.9924689650535583, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007531054131686687, + "mgd_loss_weight": 0.6308376967906952, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.13136901855466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.12531127929685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11080, + "valid_token_count": 818.24609375 + }, + { + "actual_token_mask_ratio": 0.30002913773059847, + "epoch": 3.226651149258074, + "flow_matching_loss": 0.013755269441753626, + "grad_norm": 0.0901833325624466, + "kept_token_count": 572.72578125, + "learning_rate": 2.2372197847510722e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9928502023220063, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007149794837459922, + "mgd_loss_weight": 0.6305043518543243, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.40684204101564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.18010864257815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11090, + "valid_token_count": 818.2125 + }, + { + "actual_token_mask_ratio": 0.3005692481994629, + "epoch": 3.2295606633692175, + "flow_matching_loss": 0.01295668501406908, + "grad_norm": 0.07762297987937927, + "kept_token_count": 572.215625, + "learning_rate": 2.2357793485654813e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9923953294754029, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007604687800630927, + "mgd_loss_weight": 0.630171000957489, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.2830413818359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.02002868652346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11100, + "valid_token_count": 818.11796875 + }, + { + "actual_token_mask_ratio": 0.30072204768657684, + "epoch": 3.232470177480361, + "flow_matching_loss": 0.01430748188868165, + "grad_norm": 0.08182284235954285, + "kept_token_count": 572.2234375, + "learning_rate": 2.2343380183386017e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.992772763967514, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00722723277285695, + "mgd_loss_weight": 0.6298376619815826, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.2891021728516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.3521667480469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11110, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.30024288296699525, + "epoch": 3.235379691591504, + "flow_matching_loss": 0.013489841483533382, + "grad_norm": 0.08231237530708313, + "kept_token_count": 572.61328125, + "learning_rate": 2.232895795821787e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9927500486373901, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007249962911009789, + "mgd_loss_weight": 0.6295043170452118, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.44664001464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 374.66265258789065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11120, + "valid_token_count": 818.30546875 + }, + { + "actual_token_mask_ratio": 0.3002544343471527, + "epoch": 3.2382892057026478, + "flow_matching_loss": 0.012469272408634425, + "grad_norm": 0.09814906120300293, + "kept_token_count": 572.48125, + "learning_rate": 2.2314526827674723e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9930014193058014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00699857478030026, + "mgd_loss_weight": 0.629170972108841, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.41025695800784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.4143890380859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11130, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.29972060322761535, + "epoch": 3.241198719813791, + "flow_matching_loss": 0.013742129690945149, + "grad_norm": 0.09654634445905685, + "kept_token_count": 572.8625, + "learning_rate": 2.230008680929177e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9923527777194977, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007647227216511965, + "mgd_loss_weight": 0.62883762717247, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.23992309570315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.1683654785156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11140, + "valid_token_count": 818.0515625 + }, + { + "actual_token_mask_ratio": 0.29999718964099886, + "epoch": 3.2441082339249343, + "flow_matching_loss": 0.014956329297274352, + "grad_norm": 0.0783080905675888, + "kept_token_count": 572.83359375, + "learning_rate": 2.2285637920614987e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9922864854335784, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007713509863242507, + "mgd_loss_weight": 0.6285042822360992, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.2932861328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.08136291503905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11150, + "valid_token_count": 818.33125 + }, + { + "actual_token_mask_ratio": 0.29982338547706605, + "epoch": 3.247017748036078, + "flow_matching_loss": 0.012728347815573216, + "grad_norm": 0.08670330792665482, + "kept_token_count": 572.93203125, + "learning_rate": 2.2271180179201143e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9926230132579803, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007376988092437387, + "mgd_loss_weight": 0.6281709372997284, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.74456481933595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.13026123046876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11160, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.2993483364582062, + "epoch": 3.2499272621472213, + "flow_matching_loss": 0.014553721901029349, + "grad_norm": 0.07784765213727951, + "kept_token_count": 573.3171875, + "learning_rate": 2.225671360261775e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9924399733543396, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007560028741136193, + "mgd_loss_weight": 0.627837598323822, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 384.48008728027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.0694641113281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11170, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.3004697024822235, + "epoch": 3.252836776258365, + "flow_matching_loss": 0.014389035198837519, + "grad_norm": 0.08701024204492569, + "kept_token_count": 572.33203125, + "learning_rate": 2.2242238208443068e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9923943877220154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007605603244155645, + "mgd_loss_weight": 0.6275042474269867, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.1667419433594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.05564880371094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11180, + "valid_token_count": 818.16640625 + }, + { + "actual_token_mask_ratio": 0.30062718093395235, + "epoch": 3.2557462903695082, + "flow_matching_loss": 0.013837473187595605, + "grad_norm": 0.08118856698274612, + "kept_token_count": 572.065625, + "learning_rate": 2.2227754014266063e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9926801025867462, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007319891313090921, + "mgd_loss_weight": 0.6271709024906158, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.8986846923828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.1645568847656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11190, + "valid_token_count": 817.96953125 + }, + { + "actual_token_mask_ratio": 0.3002656728029251, + "epoch": 3.258655804480652, + "flow_matching_loss": 0.01519599687308073, + "grad_norm": 0.09777119755744934, + "kept_token_count": 572.70546875, + "learning_rate": 2.22132610376864e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9925690054893493, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007430993253365159, + "mgd_loss_weight": 0.6268375635147094, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.6838806152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.34861450195314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11200, + "valid_token_count": 818.46640625 + }, + { + "actual_token_mask_ratio": 0.2999243438243866, + "epoch": 3.261565318591795, + "flow_matching_loss": 0.015183372795581818, + "grad_norm": 0.08944815397262573, + "kept_token_count": 572.80703125, + "learning_rate": 2.219875929631441e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9926221907138825, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007377815758809447, + "mgd_loss_weight": 0.6265042126178741, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 391.39190063476565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.5109344482422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11210, + "valid_token_count": 818.20390625 + }, + { + "actual_token_mask_ratio": 0.3007773578166962, + "epoch": 3.2644748327029385, + "flow_matching_loss": 0.01365465708076954, + "grad_norm": 0.09795057028532028, + "kept_token_count": 572.41640625, + "learning_rate": 2.2184248807771074e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9926179230213166, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007382077444344759, + "mgd_loss_weight": 0.6261708736419678, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.85211791992185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.8592926025391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11220, + "valid_token_count": 818.64921875 + }, + { + "actual_token_mask_ratio": 0.29966012835502626, + "epoch": 3.267384346814082, + "flow_matching_loss": 0.013155909674242138, + "grad_norm": 0.06885725259780884, + "kept_token_count": 572.946875, + "learning_rate": 2.216972958968801e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9930054187774658, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006994566647335887, + "mgd_loss_weight": 0.625837528705597, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.766064453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.82867736816405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11230, + "valid_token_count": 818.10078125 + }, + { + "actual_token_mask_ratio": 0.2995861828327179, + "epoch": 3.2702938609252254, + "flow_matching_loss": 0.013588914088904858, + "grad_norm": 0.08476833999156952, + "kept_token_count": 573.11328125, + "learning_rate": 2.215520165970743e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9928866624832153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007113333558663726, + "mgd_loss_weight": 0.6255041778087616, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.0650146484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.79210510253904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11240, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.29925709366798403, + "epoch": 3.2732033750363687, + "flow_matching_loss": 0.012696932442486287, + "grad_norm": 0.07779379189014435, + "kept_token_count": 573.3421875, + "learning_rate": 2.2140665035482143e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9930359899997712, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006964013259857893, + "mgd_loss_weight": 0.6251708447933197, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.80156860351565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.5668487548828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11250, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.3004328519105911, + "epoch": 3.2761128891475124, + "flow_matching_loss": 0.014011797215789556, + "grad_norm": 0.08485876023769379, + "kept_token_count": 572.39609375, + "learning_rate": 2.2126119734675518e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9924646615982056, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075353369116783146, + "mgd_loss_weight": 0.6248374938964844, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.40032958984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.3877655029297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11260, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.2992900818586349, + "epoch": 3.2790224032586557, + "flow_matching_loss": 0.013483605068176984, + "grad_norm": 0.10035394877195358, + "kept_token_count": 573.253125, + "learning_rate": 2.2111565774961463e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9929231405258179, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007076859660446644, + "mgd_loss_weight": 0.6245041489601135, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.98041076660155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.6110321044922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11270, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.300496107339859, + "epoch": 3.2819319173697994, + "flow_matching_loss": 0.013652675505727529, + "grad_norm": 0.09580440074205399, + "kept_token_count": 572.19609375, + "learning_rate": 2.2097003174024417e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9925149559974671, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007485049450770021, + "mgd_loss_weight": 0.6241708099842072, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.8608093261719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.018408203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11280, + "valid_token_count": 818.00625 + }, + { + "actual_token_mask_ratio": 0.2998580545186996, + "epoch": 3.2848414314809427, + "flow_matching_loss": 0.01333632878959179, + "grad_norm": 0.07524150609970093, + "kept_token_count": 572.88671875, + "learning_rate": 2.2082431949559315e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.992666357755661, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007333645084872842, + "mgd_loss_weight": 0.6238374590873719, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.18085021972655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.2076141357422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11290, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.30015689730644224, + "epoch": 3.2877509455920864, + "flow_matching_loss": 0.014616982638835907, + "grad_norm": 0.11169575899839401, + "kept_token_count": 572.6015625, + "learning_rate": 2.2067852119271558e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9921290516853333, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007870937883853912, + "mgd_loss_weight": 0.6235041201114655, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.10694885253906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.6569091796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11300, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.3002223074436188, + "epoch": 3.2906604597032296, + "flow_matching_loss": 0.014184743631631136, + "grad_norm": 0.095701202750206, + "kept_token_count": 572.72109375, + "learning_rate": 2.2053263700877026e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9924395203590393, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007560472143813968, + "mgd_loss_weight": 0.6231707751750946, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.4229278564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.22587890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11310, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.30031796991825105, + "epoch": 3.293569973814373, + "flow_matching_loss": 0.01508287200704217, + "grad_norm": 0.11382956802845001, + "kept_token_count": 572.503125, + "learning_rate": 2.2038666712102015e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9922360360622406, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007763973996043205, + "mgd_loss_weight": 0.6228374242782593, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.44309997558594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.0010986328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11320, + "valid_token_count": 818.23203125 + }, + { + "actual_token_mask_ratio": 0.30025366842746737, + "epoch": 3.2964794879255166, + "flow_matching_loss": 0.014135812502354383, + "grad_norm": 0.0770670548081398, + "kept_token_count": 572.48828125, + "learning_rate": 2.2024061170683246e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9924394905567169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007560520945116877, + "mgd_loss_weight": 0.6225040853023529, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.7787384033203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.94191589355466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11330, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.29938005805015566, + "epoch": 3.29938900203666, + "flow_matching_loss": 0.014994903095066547, + "grad_norm": 0.10903401672840118, + "kept_token_count": 573.21953125, + "learning_rate": 2.200944709436783e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.992171174287796, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00782882971689105, + "mgd_loss_weight": 0.6221707403659821, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.8857757568359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1583526611328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11340, + "valid_token_count": 818.16171875 + }, + { + "actual_token_mask_ratio": 0.3002024173736572, + "epoch": 3.302298516147803, + "flow_matching_loss": 0.013473285920917987, + "grad_norm": 0.07325282692909241, + "kept_token_count": 572.6609375, + "learning_rate": 2.1994824500913246e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9928622364997863, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007137756142765283, + "mgd_loss_weight": 0.6218373954296113, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.7536163330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0138336181641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11350, + "valid_token_count": 818.32265625 + }, + { + "actual_token_mask_ratio": 0.29964116811752317, + "epoch": 3.305208030258947, + "flow_matching_loss": 0.015160701051354409, + "grad_norm": 0.0685858279466629, + "kept_token_count": 573.0671875, + "learning_rate": 2.1980193408087322e-05, + "loss": 0.0199, + "mgd_cos_sim": 0.9923953533172607, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007604638580232859, + "mgd_loss_weight": 0.6215040504932403, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.53908386230466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.12720947265626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11360, + "valid_token_count": 818.24609375 + }, + { + "actual_token_mask_ratio": 0.30039407014846803, + "epoch": 3.30811754437009, + "flow_matching_loss": 0.014080655295401812, + "grad_norm": 0.08754947781562805, + "kept_token_count": 572.53125, + "learning_rate": 2.1965553833668224e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9924794375896454, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007520568929612637, + "mgd_loss_weight": 0.6211707055568695, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.92991333007814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.54065856933596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11370, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.3001939833164215, + "epoch": 3.311027058481234, + "flow_matching_loss": 0.014031160064041615, + "grad_norm": 0.08107047528028488, + "kept_token_count": 572.44921875, + "learning_rate": 2.195090579544441e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9927186965942383, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007281304895877838, + "mgd_loss_weight": 0.6208373606204987, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.05979919433594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.94952697753905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11380, + "valid_token_count": 818.0140625 + }, + { + "actual_token_mask_ratio": 0.300497031211853, + "epoch": 3.313936572592377, + "flow_matching_loss": 0.013761811517179012, + "grad_norm": 0.09663104265928268, + "kept_token_count": 572.36015625, + "learning_rate": 2.193624931121462e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9925129294395447, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007487067999318242, + "mgd_loss_weight": 0.6205040216445923, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.9961456298828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.2227386474609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11390, + "valid_token_count": 818.23828125 + }, + { + "actual_token_mask_ratio": 0.30023670196533203, + "epoch": 3.3168460867035203, + "flow_matching_loss": 0.013414809480309486, + "grad_norm": 0.0797429159283638, + "kept_token_count": 572.57265625, + "learning_rate": 2.1921584398787872e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9929971933364868, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007002808945253491, + "mgd_loss_weight": 0.620170670747757, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.4134765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.66351928710935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11400, + "valid_token_count": 818.23828125 + }, + { + "actual_token_mask_ratio": 0.29956549108028413, + "epoch": 3.319755600814664, + "flow_matching_loss": 0.01319929389283061, + "grad_norm": 0.07734564691781998, + "kept_token_count": 573.23984375, + "learning_rate": 2.1906911075983418e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9922573268413544, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007742659142240882, + "mgd_loss_weight": 0.6198373258113861, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.499365234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.6789093017578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11410, + "valid_token_count": 818.40234375 + }, + { + "actual_token_mask_ratio": 0.2996195018291473, + "epoch": 3.3226651149258073, + "flow_matching_loss": 0.011740174330770969, + "grad_norm": 0.06261502206325531, + "kept_token_count": 573.10546875, + "learning_rate": 2.1892229360630728e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9929106831550598, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00708931633271277, + "mgd_loss_weight": 0.6195039868354797, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.0619689941406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.1360168457031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11420, + "valid_token_count": 818.27890625 + }, + { + "actual_token_mask_ratio": 0.3001582473516464, + "epoch": 3.325574629036951, + "flow_matching_loss": 0.013056488335132599, + "grad_norm": 0.07962830364704132, + "kept_token_count": 572.62109375, + "learning_rate": 2.187753927056947e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9924507439136505, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007549252267926932, + "mgd_loss_weight": 0.6191706359386444, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.64539489746096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.03118591308595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11430, + "valid_token_count": 818.21328125 + }, + { + "actual_token_mask_ratio": 0.2997511446475983, + "epoch": 3.3284841431480943, + "flow_matching_loss": 0.014249631017446519, + "grad_norm": 0.09113895893096924, + "kept_token_count": 572.8703125, + "learning_rate": 2.186284082364949e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9927307963371277, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007269213674589992, + "mgd_loss_weight": 0.618837296962738, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.59282836914065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.06492309570314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11440, + "valid_token_count": 818.09765625 + }, + { + "actual_token_mask_ratio": 0.30015069544315337, + "epoch": 3.3313936572592375, + "flow_matching_loss": 0.013380111195147038, + "grad_norm": 0.07572457939386368, + "kept_token_count": 572.69765625, + "learning_rate": 2.1848134037730786e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9925069272518158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007493075355887413, + "mgd_loss_weight": 0.6185039520263672, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.51531982421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.8675048828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11450, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.2990876048803329, + "epoch": 3.3343031713703812, + "flow_matching_loss": 0.01327688256278634, + "grad_norm": 0.08903202414512634, + "kept_token_count": 573.42421875, + "learning_rate": 2.183341893068349e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9928719341754914, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007128076907247305, + "mgd_loss_weight": 0.6181706011295318, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 389.49489135742186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.92608337402345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11460, + "valid_token_count": 818.1125 + }, + { + "actual_token_mask_ratio": 0.2995491802692413, + "epoch": 3.3372126854815245, + "flow_matching_loss": 0.01417238675057888, + "grad_norm": 0.08708401769399643, + "kept_token_count": 573.125, + "learning_rate": 2.1818695520387842e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9926284670829773, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007371528958901763, + "mgd_loss_weight": 0.61783726811409, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.2222564697266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.8274169921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11470, + "valid_token_count": 818.21640625 + }, + { + "actual_token_mask_ratio": 0.29946060180664064, + "epoch": 3.340122199592668, + "flow_matching_loss": 0.012397256121039391, + "grad_norm": 0.07383870333433151, + "kept_token_count": 573.18671875, + "learning_rate": 2.1803963824734173e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9926855444908143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073144679423421625, + "mgd_loss_weight": 0.6175039112567902, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.7742004394531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.18426818847655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11480, + "valid_token_count": 818.20390625 + }, + { + "actual_token_mask_ratio": 0.29915125370025636, + "epoch": 3.3430317137038115, + "flow_matching_loss": 0.013096197973936796, + "grad_norm": 0.07835721969604492, + "kept_token_count": 573.584375, + "learning_rate": 2.1789223861622887e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9926777482032776, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007322252774611115, + "mgd_loss_weight": 0.6171705722808838, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.4675598144531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.6968109130859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11490, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.29981873333454134, + "epoch": 3.3459412278149547, + "flow_matching_loss": 0.014060623478144407, + "grad_norm": 0.0939183384180069, + "kept_token_count": 573.0078125, + "learning_rate": 2.1774475648964428e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9927579939365387, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007242005970329046, + "mgd_loss_weight": 0.6168372333049774, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.9776092529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7188201904297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11500, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.29933044910430906, + "epoch": 3.3488507419260984, + "flow_matching_loss": 0.01331264590844512, + "grad_norm": 0.08719372749328613, + "kept_token_count": 573.325, + "learning_rate": 2.1759719204679267e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9924358189105987, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075641815084964035, + "mgd_loss_weight": 0.6165038764476776, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.39650573730466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.0508209228516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11510, + "valid_token_count": 818.25390625 + }, + { + "actual_token_mask_ratio": 0.3003409743309021, + "epoch": 3.3517602560372417, + "flow_matching_loss": 0.013920250069350004, + "grad_norm": 0.06925718486309052, + "kept_token_count": 572.56640625, + "learning_rate": 2.174495454669787e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9925123155117035, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074876826256513596, + "mgd_loss_weight": 0.6161705434322358, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.3132263183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.06036071777345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11520, + "valid_token_count": 818.34921875 + }, + { + "actual_token_mask_ratio": 0.30024227797985076, + "epoch": 3.3546697701483854, + "flow_matching_loss": 0.015402411762624978, + "grad_norm": 0.0829729288816452, + "kept_token_count": 572.7015625, + "learning_rate": 2.1730181692960692e-05, + "loss": 0.0201, + "mgd_cos_sim": 0.9924154162406922, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075845721177756785, + "mgd_loss_weight": 0.6158371925354004, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.9550445556641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7469879150391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11530, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.2996090441942215, + "epoch": 3.3575792842595287, + "flow_matching_loss": 0.012394499126821756, + "grad_norm": 0.08524807542562485, + "kept_token_count": 573.12734375, + "learning_rate": 2.171540066141814e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9923463046550751, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007653692876920104, + "mgd_loss_weight": 0.6155038475990295, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.30152893066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.5341766357422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11540, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.2999750763177872, + "epoch": 3.360488798370672, + "flow_matching_loss": 0.014978826139122247, + "grad_norm": 0.10029282420873642, + "kept_token_count": 572.8109375, + "learning_rate": 2.170061147003056e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9926445126533509, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073554928880184885, + "mgd_loss_weight": 0.6151705086231232, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.78534545898435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.063037109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11550, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.3002685785293579, + "epoch": 3.3633983124818156, + "flow_matching_loss": 0.015365920308977366, + "grad_norm": 0.10593941062688828, + "kept_token_count": 572.6140625, + "learning_rate": 2.1685814136768225e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.992861932516098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071380688343197106, + "mgd_loss_weight": 0.6148371577262879, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.99799194335935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.6066955566406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11560, + "valid_token_count": 818.33984375 + }, + { + "actual_token_mask_ratio": 0.29978508949279786, + "epoch": 3.366307826592959, + "flow_matching_loss": 0.012130060885101556, + "grad_norm": 0.08531372249126434, + "kept_token_count": 573.17421875, + "learning_rate": 2.167100867961128e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9927865624427795, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007213447522372007, + "mgd_loss_weight": 0.6145038187503815, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.66962890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.25042724609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11570, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.30011957585811616, + "epoch": 3.369217340704102, + "flow_matching_loss": 0.015593042969703675, + "grad_norm": 0.08219954371452332, + "kept_token_count": 572.8078125, + "learning_rate": 2.1656195116549756e-05, + "loss": 0.0202, + "mgd_cos_sim": 0.9924528777599335, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007547117210924626, + "mgd_loss_weight": 0.6141704738140106, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.88519592285155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.53578796386716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11580, + "valid_token_count": 818.43046875 + }, + { + "actual_token_mask_ratio": 0.3001025289297104, + "epoch": 3.372126854815246, + "flow_matching_loss": 0.013391223270446062, + "grad_norm": 0.07431723922491074, + "kept_token_count": 572.73359375, + "learning_rate": 2.1641373465583527e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9927286982536316, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007271308405324817, + "mgd_loss_weight": 0.6138371229171753, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.71253967285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.83518981933594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11590, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.3000739276409149, + "epoch": 3.375036368926389, + "flow_matching_loss": 0.01370416497811675, + "grad_norm": 0.07147545367479324, + "kept_token_count": 572.73125, + "learning_rate": 2.1626543744722292e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9924113392829895, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007588656712323427, + "mgd_loss_weight": 0.6135037839412689, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.2919128417969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.74951171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11600, + "valid_token_count": 818.27890625 + }, + { + "actual_token_mask_ratio": 0.2995032459497452, + "epoch": 3.377945883037533, + "flow_matching_loss": 0.012884540995582939, + "grad_norm": 0.09038183838129044, + "kept_token_count": 573.30078125, + "learning_rate": 2.161170597198557e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9928563356399536, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007143654394894838, + "mgd_loss_weight": 0.6131704390048981, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.01063537597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.02927551269534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11610, + "valid_token_count": 818.41484375 + }, + { + "actual_token_mask_ratio": 0.30014478266239164, + "epoch": 3.380855397148676, + "flow_matching_loss": 0.012794577237218618, + "grad_norm": 0.0731600821018219, + "kept_token_count": 572.83828125, + "learning_rate": 2.1596860165402645e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9927835047245026, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00721648964099586, + "mgd_loss_weight": 0.6128370940685273, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.5968383789062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.49123229980466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11620, + "valid_token_count": 818.50703125 + }, + { + "actual_token_mask_ratio": 0.2998071700334549, + "epoch": 3.38376491125982, + "flow_matching_loss": 0.014215131755918264, + "grad_norm": 0.08911897987127304, + "kept_token_count": 572.821875, + "learning_rate": 2.1582006343012576e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9926871836185456, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073128209449350836, + "mgd_loss_weight": 0.6125037491321563, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.53895263671876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.878857421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11630, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.30021998584270476, + "epoch": 3.386674425370963, + "flow_matching_loss": 0.013691172376275063, + "grad_norm": 0.10310805588960648, + "kept_token_count": 572.56171875, + "learning_rate": 2.156714452286416e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9919110536575317, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00808894275687635, + "mgd_loss_weight": 0.6121704041957855, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.6884826660156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2962188720703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11640, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.29980695247650146, + "epoch": 3.3895839394821063, + "flow_matching_loss": 0.013922380842268467, + "grad_norm": 0.09117773920297623, + "kept_token_count": 573.12890625, + "learning_rate": 2.155227472301591e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9930641412734985, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006935857003554702, + "mgd_loss_weight": 0.6118370592594147, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.0621398925781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.71324462890624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11650, + "valid_token_count": 818.528125 + }, + { + "actual_token_mask_ratio": 0.3004311680793762, + "epoch": 3.39249345359325, + "flow_matching_loss": 0.013178609590977431, + "grad_norm": 0.07489749789237976, + "kept_token_count": 572.5171875, + "learning_rate": 2.1537396961536027e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9927367806434632, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007263219216838479, + "mgd_loss_weight": 0.6115037202835083, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.4812072753906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.3669097900391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11660, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.3003204733133316, + "epoch": 3.3954029677043933, + "flow_matching_loss": 0.013620028086006642, + "grad_norm": 0.08643105626106262, + "kept_token_count": 572.41015625, + "learning_rate": 2.1522511256502403e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9927742004394531, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007225802727043628, + "mgd_loss_weight": 0.611170369386673, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.1339385986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0255554199219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11670, + "valid_token_count": 818.09921875 + }, + { + "actual_token_mask_ratio": 0.3004846066236496, + "epoch": 3.3983124818155366, + "flow_matching_loss": 0.01320688920095563, + "grad_norm": 0.09681936353445053, + "kept_token_count": 572.35, + "learning_rate": 2.1507617626002566e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9926768660545349, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00732314856722951, + "mgd_loss_weight": 0.6108370244503021, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.35362548828124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.78436584472655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11680, + "valid_token_count": 818.21484375 + }, + { + "actual_token_mask_ratio": 0.2996259808540344, + "epoch": 3.4012219959266803, + "flow_matching_loss": 0.014042918756604194, + "grad_norm": 0.08454965800046921, + "kept_token_count": 573.028125, + "learning_rate": 2.1492716088133685e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9923811078071594, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007618911797180772, + "mgd_loss_weight": 0.6105036854743957, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.837939453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.90020446777345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11690, + "valid_token_count": 818.17265625 + }, + { + "actual_token_mask_ratio": 0.29994277358055116, + "epoch": 3.4041315100378235, + "flow_matching_loss": 0.013160377275198698, + "grad_norm": 0.0810311883687973, + "kept_token_count": 572.84453125, + "learning_rate": 2.1477806661002536e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.992701655626297, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007298354571685195, + "mgd_loss_weight": 0.6101703345775604, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.3331756591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.8754913330078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11700, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.2989896535873413, + "epoch": 3.4070410241489673, + "flow_matching_loss": 0.014809564873576164, + "grad_norm": 0.11449752748012543, + "kept_token_count": 573.52265625, + "learning_rate": 2.146288936272548e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9931629717350006, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068370292894542216, + "mgd_loss_weight": 0.609836995601654, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.87481079101565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.3656372070312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11710, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.2996516525745392, + "epoch": 3.4099505382601105, + "flow_matching_loss": 0.014362975303083659, + "grad_norm": 0.07533589005470276, + "kept_token_count": 572.9953125, + "learning_rate": 2.1447964211428433e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9928307592868805, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007169248117133975, + "mgd_loss_weight": 0.6095036566257477, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.34764099121094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.1770324707031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11720, + "valid_token_count": 818.15703125 + }, + { + "actual_token_mask_ratio": 0.30101526975631715, + "epoch": 3.4128600523712542, + "flow_matching_loss": 0.012067047692835332, + "grad_norm": 0.09143334627151489, + "kept_token_count": 571.96796875, + "learning_rate": 2.1433031225246865e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9927557170391083, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007244298839941621, + "mgd_loss_weight": 0.6091702997684478, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.07000122070315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0711273193359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11730, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.30016280710697174, + "epoch": 3.4157695664823975, + "flow_matching_loss": 0.016453123930841685, + "grad_norm": 0.08138467371463776, + "kept_token_count": 572.56953125, + "learning_rate": 2.1418090422325766e-05, + "loss": 0.021, + "mgd_cos_sim": 0.99251788854599, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007482120487838984, + "mgd_loss_weight": 0.608836966753006, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.9836669921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1646697998047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11740, + "valid_token_count": 818.14765625 + }, + { + "actual_token_mask_ratio": 0.30034895837306974, + "epoch": 3.4186790805935408, + "flow_matching_loss": 0.013289045542478561, + "grad_norm": 0.09611117839813232, + "kept_token_count": 572.51875, + "learning_rate": 2.1403141820819623e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9926470696926117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007352943904697895, + "mgd_loss_weight": 0.6085036158561706, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.9878753662109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.5197723388672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11750, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.3002471148967743, + "epoch": 3.4215885947046845, + "flow_matching_loss": 0.013912733178585768, + "grad_norm": 0.09418144077062607, + "kept_token_count": 572.5390625, + "learning_rate": 2.1388185438892377e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9926037549972534, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073962494730949405, + "mgd_loss_weight": 0.6081702709197998, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.10870361328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.7947113037109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11760, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.29924386739730835, + "epoch": 3.4244981088158277, + "flow_matching_loss": 0.013809166382998228, + "grad_norm": 0.09072466194629669, + "kept_token_count": 573.59609375, + "learning_rate": 2.1373221294717465e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9929904997348785, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007009502733126283, + "mgd_loss_weight": 0.6078369319438934, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.4965545654297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.07688293457034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11770, + "valid_token_count": 818.5390625 + }, + { + "actual_token_mask_ratio": 0.3002625286579132, + "epoch": 3.427407622926971, + "flow_matching_loss": 0.015270532667636871, + "grad_norm": 0.09621861577033997, + "kept_token_count": 572.52421875, + "learning_rate": 2.1358249406477724e-05, + "loss": 0.0195, + "mgd_cos_sim": 0.9930785477161408, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006921463087201119, + "mgd_loss_weight": 0.6075035810470581, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.40001220703124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.86240844726564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11780, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.2999009430408478, + "epoch": 3.4303171370381147, + "flow_matching_loss": 0.014200087916105986, + "grad_norm": 0.08420558273792267, + "kept_token_count": 573.02578125, + "learning_rate": 2.1343269792365408e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9928129613399506, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007187033630907535, + "mgd_loss_weight": 0.6071702420711518, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.1943054199219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2894744873047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11790, + "valid_token_count": 818.4953125 + }, + { + "actual_token_mask_ratio": 0.29939426481723785, + "epoch": 3.433226651149258, + "flow_matching_loss": 0.014816062897443772, + "grad_norm": 0.09643980860710144, + "kept_token_count": 573.3953125, + "learning_rate": 2.132828247058217e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9926992952823639, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007300692237913608, + "mgd_loss_weight": 0.6068368971347808, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.98236694335935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.51291809082034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11800, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.2995800763368607, + "epoch": 3.4361361652604017, + "flow_matching_loss": 0.01429982976987958, + "grad_norm": 0.085020050406456, + "kept_token_count": 573.01328125, + "learning_rate": 2.1313287459339004e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9927056133747101, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072943820152431725, + "mgd_loss_weight": 0.6065035462379456, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.1501800537109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.3410919189453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11810, + "valid_token_count": 818.1 + }, + { + "actual_token_mask_ratio": 0.2993100434541702, + "epoch": 3.439045679371545, + "flow_matching_loss": 0.012871631234884263, + "grad_norm": 0.07647202908992767, + "kept_token_count": 573.37421875, + "learning_rate": 2.1298284776856276e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9926582872867584, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007341711176559329, + "mgd_loss_weight": 0.6061702072620392, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.4053039550781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9374664306641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11820, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.30005664825439454, + "epoch": 3.4419551934826886, + "flow_matching_loss": 0.014120385516434908, + "grad_norm": 0.09707281738519669, + "kept_token_count": 572.77734375, + "learning_rate": 2.128327444136365e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9930656850337982, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006934302672743797, + "mgd_loss_weight": 0.6058368623256684, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.63230590820314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.6617950439453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11830, + "valid_token_count": 818.31796875 + }, + { + "actual_token_mask_ratio": 0.30008515119552615, + "epoch": 3.444864707593832, + "flow_matching_loss": 0.014427826832979917, + "grad_norm": 0.10469984263181686, + "kept_token_count": 572.7609375, + "learning_rate": 2.1268256471100105e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9927954316139221, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007204560190439224, + "mgd_loss_weight": 0.6055035173892975, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.43529052734374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.7969451904297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11840, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.299380761384964, + "epoch": 3.447774221704975, + "flow_matching_loss": 0.012732047401368619, + "grad_norm": 0.09672552347183228, + "kept_token_count": 573.22421875, + "learning_rate": 2.1253230884313893e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9928455293178559, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071544615551829335, + "mgd_loss_weight": 0.6051701724529266, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.07898864746096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.2808502197266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11850, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.2995049923658371, + "epoch": 3.450683735816119, + "flow_matching_loss": 0.014098025299608708, + "grad_norm": 0.08723045140504837, + "kept_token_count": 573.0828125, + "learning_rate": 2.123819769926251e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9927114248275757, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007288583787158132, + "mgd_loss_weight": 0.6048368275165558, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.2296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9443664550781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11860, + "valid_token_count": 818.11328125 + }, + { + "actual_token_mask_ratio": 0.2998677581548691, + "epoch": 3.453593249927262, + "flow_matching_loss": 0.01372210904955864, + "grad_norm": 0.07293815910816193, + "kept_token_count": 572.9453125, + "learning_rate": 2.1223156934212702e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9927384495735169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007261544512584805, + "mgd_loss_weight": 0.6045034825801849, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 444.64147033691404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.6561004638672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11870, + "valid_token_count": 818.33828125 + }, + { + "actual_token_mask_ratio": 0.3000408589839935, + "epoch": 3.4565027640384054, + "flow_matching_loss": 0.013774482998996973, + "grad_norm": 0.08613631874322891, + "kept_token_count": 572.82578125, + "learning_rate": 2.1208108607440418e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9928199470043182, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007180050481110811, + "mgd_loss_weight": 0.6041701436042786, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.41969604492186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.7975128173828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11880, + "valid_token_count": 818.36640625 + }, + { + "actual_token_mask_ratio": 0.3004164814949036, + "epoch": 3.459412278149549, + "flow_matching_loss": 0.013691414892673493, + "grad_norm": 0.11008282005786896, + "kept_token_count": 572.4421875, + "learning_rate": 2.119305273723078e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9926872670650482, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00731273707933724, + "mgd_loss_weight": 0.6038367927074433, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.45457458496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.0364471435547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11890, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.2994158923625946, + "epoch": 3.4623217922606924, + "flow_matching_loss": 0.013494494184851647, + "grad_norm": 0.0804547443985939, + "kept_token_count": 573.1765625, + "learning_rate": 2.1177989341878106e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9926890432834625, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073109498713165525, + "mgd_loss_weight": 0.6035034477710723, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.9849060058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.53319091796874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11900, + "valid_token_count": 818.14140625 + }, + { + "actual_token_mask_ratio": 0.3000752627849579, + "epoch": 3.465231306371836, + "flow_matching_loss": 0.01424845764413476, + "grad_norm": 0.08437197655439377, + "kept_token_count": 572.70234375, + "learning_rate": 2.116291843968583e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9925819873809815, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007417997112497687, + "mgd_loss_weight": 0.603170108795166, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.1598693847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.60874328613284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11910, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.2998059391975403, + "epoch": 3.4681408204829793, + "flow_matching_loss": 0.014213340450078249, + "grad_norm": 0.08071445673704147, + "kept_token_count": 573.0375, + "learning_rate": 2.114784004896653e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9925536513328552, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007446357002481818, + "mgd_loss_weight": 0.6028367578983307, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.3860656738281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9066192626953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11920, + "valid_token_count": 818.40546875 + }, + { + "actual_token_mask_ratio": 0.3001620054244995, + "epoch": 3.4710503345941226, + "flow_matching_loss": 0.0148765804246068, + "grad_norm": 0.07405009865760803, + "kept_token_count": 572.67890625, + "learning_rate": 2.113275418804187e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9927042603492737, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007295733271166682, + "mgd_loss_weight": 0.6025034189224243, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.63950805664064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.11281127929686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11930, + "valid_token_count": 818.30859375 + }, + { + "actual_token_mask_ratio": 0.2996144235134125, + "epoch": 3.4739598487052663, + "flow_matching_loss": 0.012616855092346668, + "grad_norm": 0.08962972462177277, + "kept_token_count": 573.09609375, + "learning_rate": 2.1117660875242583e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9926912844181061, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00730870570987463, + "mgd_loss_weight": 0.6021700739860535, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.86155090332034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.35606079101564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11940, + "valid_token_count": 818.25546875 + }, + { + "actual_token_mask_ratio": 0.3000262796878815, + "epoch": 3.4768693628164096, + "flow_matching_loss": 0.013828911166638137, + "grad_norm": 0.09126489609479904, + "kept_token_count": 572.76484375, + "learning_rate": 2.1102560128908478e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9928272306919098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007172783836722374, + "mgd_loss_weight": 0.6018367230892181, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.54200439453126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8231262207031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11950, + "valid_token_count": 818.26484375 + }, + { + "actual_token_mask_ratio": 0.30024614930152893, + "epoch": 3.4797788769275533, + "flow_matching_loss": 0.012976881116628647, + "grad_norm": 0.06773883104324341, + "kept_token_count": 572.790625, + "learning_rate": 2.1087451967388384e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9931026756763458, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006897329119965434, + "mgd_loss_weight": 0.6015033900737763, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.992724609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.06712646484374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11960, + "valid_token_count": 818.559375 + }, + { + "actual_token_mask_ratio": 0.2993422657251358, + "epoch": 3.4826883910386965, + "flow_matching_loss": 0.013478652201592922, + "grad_norm": 0.07523626834154129, + "kept_token_count": 573.59921875, + "learning_rate": 2.1072336409040135e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9926092863082886, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007390711456537247, + "mgd_loss_weight": 0.6011700332164764, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.8695922851563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6649139404297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11970, + "valid_token_count": 818.6625 + }, + { + "actual_token_mask_ratio": 0.29976358711719514, + "epoch": 3.48559790514984, + "flow_matching_loss": 0.014763386081904173, + "grad_norm": 0.08922363072633743, + "kept_token_count": 573.15234375, + "learning_rate": 2.1057213472230573e-05, + "loss": 0.0192, + "mgd_cos_sim": 0.9926132619380951, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007386741926893592, + "mgd_loss_weight": 0.6008366942405701, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.42539978027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.7981750488281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11980, + "valid_token_count": 818.5109375 + }, + { + "actual_token_mask_ratio": 0.30040850341320036, + "epoch": 3.4885074192609835, + "flow_matching_loss": 0.015367316640913486, + "grad_norm": 0.10694566369056702, + "kept_token_count": 572.43515625, + "learning_rate": 2.1042083175335478e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9928233683109283, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00717663518153131, + "mgd_loss_weight": 0.6005033552646637, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.473583984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9683044433594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 11990, + "valid_token_count": 818.24296875 + }, + { + "actual_token_mask_ratio": 0.29955871105194093, + "epoch": 3.4914169333721268, + "flow_matching_loss": 0.013313559908419847, + "grad_norm": 0.08166109025478363, + "kept_token_count": 573.24296875, + "learning_rate": 2.1026945536739595e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9922061681747436, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007793827867135406, + "mgd_loss_weight": 0.6001699984073638, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.0640014648437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.0748229980469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12000, + "valid_token_count": 818.40078125 + }, + { + "actual_token_mask_ratio": 0.30059005320072174, + "epoch": 3.4943264474832705, + "flow_matching_loss": 0.014210054744035005, + "grad_norm": 0.08319613337516785, + "kept_token_count": 572.15, + "learning_rate": 2.1011800574836586e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9927642285823822, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007235774258151651, + "mgd_loss_weight": 0.599836665391922, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.76954345703126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.86322021484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12010, + "valid_token_count": 818.04921875 + }, + { + "actual_token_mask_ratio": 0.30008602440357207, + "epoch": 3.4972359615944137, + "flow_matching_loss": 0.013427525013685226, + "grad_norm": 0.0725327581167221, + "kept_token_count": 572.7328125, + "learning_rate": 2.0996648308028992e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9925050675868988, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007494925474748016, + "mgd_loss_weight": 0.5995033144950866, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.46288146972654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9039276123047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12020, + "valid_token_count": 818.28515625 + }, + { + "actual_token_mask_ratio": 0.3007152885198593, + "epoch": 3.500145475705557, + "flow_matching_loss": 0.012815701216459275, + "grad_norm": 0.102002814412117, + "kept_token_count": 572.11171875, + "learning_rate": 2.0981488754728266e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9927526593208313, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007247341703623534, + "mgd_loss_weight": 0.5991699695587158, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 445.5511016845703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.7320037841797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12030, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.30030712485313416, + "epoch": 3.5030549898167007, + "flow_matching_loss": 0.013376620225608348, + "grad_norm": 0.07757594436407089, + "kept_token_count": 572.62265625, + "learning_rate": 2.0966321933354685e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.992441076040268, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007558915670961142, + "mgd_loss_weight": 0.5988366305828094, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.8683685302734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.8601898193359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12040, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.3002002567052841, + "epoch": 3.505964503927844, + "flow_matching_loss": 0.013234683033078908, + "grad_norm": 0.06575194746255875, + "kept_token_count": 572.52578125, + "learning_rate": 2.095114786233736e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9930094540119171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006990534905344248, + "mgd_loss_weight": 0.5985032796859742, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.6448272705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.3131896972656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12050, + "valid_token_count": 818.13046875 + }, + { + "actual_token_mask_ratio": 0.2997380346059799, + "epoch": 3.5088740180389877, + "flow_matching_loss": 0.013300121203064919, + "grad_norm": 0.07929152995347977, + "kept_token_count": 572.88203125, + "learning_rate": 2.0935966560114236e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9927635610103607, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007236422039568424, + "mgd_loss_weight": 0.5981699407100678, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.9985626220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.18979797363284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12060, + "valid_token_count": 818.09296875 + }, + { + "actual_token_mask_ratio": 0.3001700282096863, + "epoch": 3.511783532150131, + "flow_matching_loss": 0.01348910378292203, + "grad_norm": 0.08796204626560211, + "kept_token_count": 572.5734375, + "learning_rate": 2.092077804513201e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9930522561073303, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006947749666869641, + "mgd_loss_weight": 0.5978365957736969, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.2151702880859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6091369628906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12070, + "valid_token_count": 818.16015625 + }, + { + "actual_token_mask_ratio": 0.2996275365352631, + "epoch": 3.514693046261274, + "flow_matching_loss": 0.013154718466103077, + "grad_norm": 0.07916850596666336, + "kept_token_count": 573.01953125, + "learning_rate": 2.0905582335846167e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.992551577091217, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007448426121845842, + "mgd_loss_weight": 0.5975032448768616, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 443.4133331298828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3873291015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12080, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.300322824716568, + "epoch": 3.517602560372418, + "flow_matching_loss": 0.014071758557111024, + "grad_norm": 0.08142280578613281, + "kept_token_count": 572.4265625, + "learning_rate": 2.0890379450720927e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9931297779083252, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006870207656174898, + "mgd_loss_weight": 0.5971699059009552, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.7269805908203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.3494567871094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12090, + "valid_token_count": 818.13359375 + }, + { + "actual_token_mask_ratio": 0.3003003627061844, + "epoch": 3.520512074483561, + "flow_matching_loss": 0.013699123170226812, + "grad_norm": 0.07756689190864563, + "kept_token_count": 572.57890625, + "learning_rate": 2.087516940822923e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9922171413898468, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007782866666093469, + "mgd_loss_weight": 0.5968365609645844, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 398.5078521728516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2844696044922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12100, + "valid_token_count": 818.32265625 + }, + { + "actual_token_mask_ratio": 0.30004613399505614, + "epoch": 3.5234215885947044, + "flow_matching_loss": 0.01416903045028448, + "grad_norm": 0.1123790442943573, + "kept_token_count": 572.87109375, + "learning_rate": 2.085995222685271e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.99266876578331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00733125600963831, + "mgd_loss_weight": 0.5965032160282135, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.4637023925781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.43104248046876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12110, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.2997413039207458, + "epoch": 3.526331102705848, + "flow_matching_loss": 0.013054258562624455, + "grad_norm": 0.08773031830787659, + "kept_token_count": 572.95859375, + "learning_rate": 2.084472792508167e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9931053996086121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068945879582315685, + "mgd_loss_weight": 0.5961698710918426, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.1702117919922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9624328613281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12120, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.30014452934265134, + "epoch": 3.5292406168169914, + "flow_matching_loss": 0.013115342613309622, + "grad_norm": 0.08191067725419998, + "kept_token_count": 572.63984375, + "learning_rate": 2.082949652141508e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.99260373711586, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073962591122835875, + "mgd_loss_weight": 0.5958365261554718, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.2362457275391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0044738769531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12130, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.2995569109916687, + "epoch": 3.532150130928135, + "flow_matching_loss": 0.013424838986247778, + "grad_norm": 0.130541130900383, + "kept_token_count": 573.3203125, + "learning_rate": 2.081425803436054e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9931015074253082, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006898485030978918, + "mgd_loss_weight": 0.595503181219101, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.29798278808596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.35670166015626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12140, + "valid_token_count": 818.5109375 + }, + { + "actual_token_mask_ratio": 0.2996987849473953, + "epoch": 3.5350596450392784, + "flow_matching_loss": 0.014613597933202983, + "grad_norm": 0.10302939265966415, + "kept_token_count": 573.1296875, + "learning_rate": 2.079901248243423e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9926357328891754, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007364284014329314, + "mgd_loss_weight": 0.5951698422431946, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.7372741699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.5821990966797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12150, + "valid_token_count": 818.41328125 + }, + { + "actual_token_mask_ratio": 0.2998537331819534, + "epoch": 3.537969159150422, + "flow_matching_loss": 0.014985609240829945, + "grad_norm": 0.0952880010008812, + "kept_token_count": 572.77109375, + "learning_rate": 2.078375988416095e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9930082976818084, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006991702504456044, + "mgd_loss_weight": 0.5948364973068238, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.9974853515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.66306762695314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12160, + "valid_token_count": 818.071875 + }, + { + "actual_token_mask_ratio": 0.29984511733055114, + "epoch": 3.5408786732615654, + "flow_matching_loss": 0.013134021684527398, + "grad_norm": 0.0768166184425354, + "kept_token_count": 572.7421875, + "learning_rate": 2.076850025807404e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9931234955787659, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068765037227422, + "mgd_loss_weight": 0.5945031464099884, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.5504638671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2794219970703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12170, + "valid_token_count": 818.01640625 + }, + { + "actual_token_mask_ratio": 0.3002360761165619, + "epoch": 3.5437881873727086, + "flow_matching_loss": 0.01535091707482934, + "grad_norm": 0.0711369663476944, + "kept_token_count": 572.603125, + "learning_rate": 2.0753233622715374e-05, + "loss": 0.0198, + "mgd_cos_sim": 0.9925662934780121, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074337093159556385, + "mgd_loss_weight": 0.5941698133945466, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.25348510742185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.44548950195315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12180, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.3001244693994522, + "epoch": 3.5466977014838523, + "flow_matching_loss": 0.012692635785788298, + "grad_norm": 0.07202795892953873, + "kept_token_count": 572.5546875, + "learning_rate": 2.073795999663536e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.993068253993988, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006931745819747448, + "mgd_loss_weight": 0.5938364565372467, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.5818603515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.94683532714845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12190, + "valid_token_count": 818.08515625 + }, + { + "actual_token_mask_ratio": 0.2998766005039215, + "epoch": 3.5496072155949956, + "flow_matching_loss": 0.013543640356510877, + "grad_norm": 0.07290488481521606, + "kept_token_count": 572.80234375, + "learning_rate": 2.07226793983929e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9925548315048218, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007445176411420107, + "mgd_loss_weight": 0.5935031175613403, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.6211395263672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6811553955078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12200, + "valid_token_count": 818.14765625 + }, + { + "actual_token_mask_ratio": 0.29989235997200014, + "epoch": 3.552516729706139, + "flow_matching_loss": 0.014685018546879291, + "grad_norm": 0.08612149953842163, + "kept_token_count": 572.93125, + "learning_rate": 2.070739184655536e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9929488897323608, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00705110989511013, + "mgd_loss_weight": 0.593169778585434, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.96348876953124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.2379180908203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12210, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.29995751678943633, + "epoch": 3.5554262438172826, + "flow_matching_loss": 0.014638245105743408, + "grad_norm": 0.09670612961053848, + "kept_token_count": 572.7015625, + "learning_rate": 2.069209735969855e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.992874538898468, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071254611480981115, + "mgd_loss_weight": 0.5928364217281341, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.53264770507815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5647338867187, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12220, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.3004982113838196, + "epoch": 3.558335757928426, + "flow_matching_loss": 0.014865185040980578, + "grad_norm": 0.08793709427118301, + "kept_token_count": 572.29765625, + "learning_rate": 2.067679595640672e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9927958130836487, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072041715029627085, + "mgd_loss_weight": 0.5925030887126923, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.1895385742188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.54339904785155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12230, + "valid_token_count": 818.15078125 + }, + { + "actual_token_mask_ratio": 0.30021762251853945, + "epoch": 3.5612452720395695, + "flow_matching_loss": 0.012803017627447844, + "grad_norm": 0.09908401221036911, + "kept_token_count": 572.51640625, + "learning_rate": 2.066148765527252e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9930298864841461, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006970103597268462, + "mgd_loss_weight": 0.5921697378158569, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.94263916015626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1085968017578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12240, + "valid_token_count": 818.13046875 + }, + { + "actual_token_mask_ratio": 0.300207594037056, + "epoch": 3.564154786150713, + "flow_matching_loss": 0.013164943642914295, + "grad_norm": 0.10389655828475952, + "kept_token_count": 572.65234375, + "learning_rate": 2.0646172474896976e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9930683970451355, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006931605655699969, + "mgd_loss_weight": 0.5918363928794861, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.063427734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3788146972656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12250, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.30014338791370393, + "epoch": 3.5670643002618565, + "flow_matching_loss": 0.012735148891806602, + "grad_norm": 0.09887872636318207, + "kept_token_count": 572.7796875, + "learning_rate": 2.063085043388947e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9927897691726685, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007210235018283129, + "mgd_loss_weight": 0.5915030539035797, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.2720428466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.2320098876953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12260, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.2995804876089096, + "epoch": 3.5699738143729998, + "flow_matching_loss": 0.013007084839046001, + "grad_norm": 0.0965321809053421, + "kept_token_count": 573.1328125, + "learning_rate": 2.0615521550867732e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9931436240673065, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006856372207403183, + "mgd_loss_weight": 0.5911697030067444, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.1297332763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.10174865722655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12270, + "valid_token_count": 818.27578125 + }, + { + "actual_token_mask_ratio": 0.2995190918445587, + "epoch": 3.572883328484143, + "flow_matching_loss": 0.012969118356704713, + "grad_norm": 0.08014195412397385, + "kept_token_count": 573.25234375, + "learning_rate": 2.0600185844457803e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9930099189281464, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00699006742797792, + "mgd_loss_weight": 0.5908363640308381, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.95692443847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.2298980712891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12280, + "valid_token_count": 818.36953125 + }, + { + "actual_token_mask_ratio": 0.3000866621732712, + "epoch": 3.5757928425952867, + "flow_matching_loss": 0.013177542388439179, + "grad_norm": 0.08229553699493408, + "kept_token_count": 572.7, + "learning_rate": 2.058484333329401e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9928095877170563, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007190427230671048, + "mgd_loss_weight": 0.5905030190944671, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.5834075927734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.484423828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12290, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.30062041580677035, + "epoch": 3.57870235670643, + "flow_matching_loss": 0.013643211405724286, + "grad_norm": 0.08982778340578079, + "kept_token_count": 572.34296875, + "learning_rate": 2.0569494036018955e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9928678154945374, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007132168440148234, + "mgd_loss_weight": 0.5901696681976318, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.6565856933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.38506774902345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12300, + "valid_token_count": 818.35703125 + }, + { + "actual_token_mask_ratio": 0.3000286132097244, + "epoch": 3.5816118708175733, + "flow_matching_loss": 0.012655851803719998, + "grad_norm": 0.11514369398355484, + "kept_token_count": 572.728125, + "learning_rate": 2.055413797128347e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9931654155254364, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006834593880921602, + "mgd_loss_weight": 0.5898363292217255, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.1740478515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1085601806641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12310, + "valid_token_count": 818.21484375 + }, + { + "actual_token_mask_ratio": 0.29977318942546843, + "epoch": 3.584521384928717, + "flow_matching_loss": 0.012591171357780696, + "grad_norm": 0.10184889286756516, + "kept_token_count": 572.9671875, + "learning_rate": 2.0538775157746634e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.993384712934494, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066152873914688826, + "mgd_loss_weight": 0.5895029842853546, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.3128326416016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.3296722412109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12320, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.29978579878807066, + "epoch": 3.5874308990398602, + "flow_matching_loss": 0.011747552454471588, + "grad_norm": 0.08857867121696472, + "kept_token_count": 572.86015625, + "learning_rate": 2.0523405614075706e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9930676877498626, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006932304380461574, + "mgd_loss_weight": 0.5891696393489838, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.48500061035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.33526306152345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12330, + "valid_token_count": 818.12109375 + }, + { + "actual_token_mask_ratio": 0.29967484772205355, + "epoch": 3.590340413151004, + "flow_matching_loss": 0.012829097360372544, + "grad_norm": 0.06444185972213745, + "kept_token_count": 572.9875, + "learning_rate": 2.0508029358946132e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9929128527641297, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070871416013687846, + "mgd_loss_weight": 0.5888362944126129, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.28504638671876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.2164764404297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12340, + "valid_token_count": 818.17421875 + }, + { + "actual_token_mask_ratio": 0.29867531061172486, + "epoch": 3.593249927262147, + "flow_matching_loss": 0.013478431198745967, + "grad_norm": 0.0714905709028244, + "kept_token_count": 573.90546875, + "learning_rate": 2.0492646411041515e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9927884638309479, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007211545202881098, + "mgd_loss_weight": 0.588502949476242, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.5406799316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.46677551269534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12350, + "valid_token_count": 818.31484375 + }, + { + "actual_token_mask_ratio": 0.3002872735261917, + "epoch": 3.596159441373291, + "flow_matching_loss": 0.014616075903177261, + "grad_norm": 0.10714909434318542, + "kept_token_count": 572.4640625, + "learning_rate": 2.0477256789053587e-05, + "loss": 0.019, + "mgd_cos_sim": 0.9926314949989319, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007368504442274571, + "mgd_loss_weight": 0.5881696045398712, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.5651580810547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2795043945313, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12360, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.300922492146492, + "epoch": 3.599068955484434, + "flow_matching_loss": 0.012614841759204864, + "grad_norm": 0.10970495641231537, + "kept_token_count": 572.20234375, + "learning_rate": 2.0461860511682194e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9926219761371613, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007378010731190443, + "mgd_loss_weight": 0.5878362655639648, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.7301727294922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6107147216797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12370, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.29969132542610166, + "epoch": 3.6019784695955774, + "flow_matching_loss": 0.012742339726537466, + "grad_norm": 0.12391204386949539, + "kept_token_count": 573.01171875, + "learning_rate": 2.0446457597635254e-05, + "loss": 0.017, + "mgd_cos_sim": 0.992737638950348, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072623422835022215, + "mgd_loss_weight": 0.5875029146671296, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.8585479736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.57373046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12380, + "valid_token_count": 818.23203125 + }, + { + "actual_token_mask_ratio": 0.29989464282989503, + "epoch": 3.604887983706721, + "flow_matching_loss": 0.013517068605870009, + "grad_norm": 0.08595266193151474, + "kept_token_count": 573.05546875, + "learning_rate": 2.0431048065628763e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9932209610939026, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006779053620994091, + "mgd_loss_weight": 0.5871695697307586, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.2842254638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3862365722656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12390, + "valid_token_count": 818.5265625 + }, + { + "actual_token_mask_ratio": 0.29974240362644194, + "epoch": 3.6077974978178644, + "flow_matching_loss": 0.014018672425299883, + "grad_norm": 0.07828351855278015, + "kept_token_count": 573.09609375, + "learning_rate": 2.0415631934386764e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.992672997713089, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007327018445357681, + "mgd_loss_weight": 0.5868362307548523, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.4457275390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.21485290527346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12400, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.3003156274557114, + "epoch": 3.6107070119290077, + "flow_matching_loss": 0.013771590031683444, + "grad_norm": 0.09014294296503067, + "kept_token_count": 572.54765625, + "learning_rate": 2.04002092226413e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.99295414686203, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007045850437134504, + "mgd_loss_weight": 0.586502879858017, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.03595275878905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9067413330078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12410, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.3001700520515442, + "epoch": 3.6136165260401514, + "flow_matching_loss": 0.013331948034465312, + "grad_norm": 0.10800334811210632, + "kept_token_count": 572.6484375, + "learning_rate": 2.038477994913242e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9929920434951782, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00700794979929924, + "mgd_loss_weight": 0.5861695408821106, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.9379150390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.78314208984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12420, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.30026001632213595, + "epoch": 3.6165260401512946, + "flow_matching_loss": 0.015016323421150446, + "grad_norm": 0.11067179590463638, + "kept_token_count": 572.64453125, + "learning_rate": 2.036934413260816e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9926989734172821, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007301034033298492, + "mgd_loss_weight": 0.5858361959457398, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.28871154785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.71811828613284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12430, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.2999830037355423, + "epoch": 3.619435554262438, + "flow_matching_loss": 0.014044744335114956, + "grad_norm": 0.08700945228338242, + "kept_token_count": 572.81484375, + "learning_rate": 2.0353901791824473e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9926968038082122, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007303199591115117, + "mgd_loss_weight": 0.5855028450489044, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.7411254882812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5894836425781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12440, + "valid_token_count": 818.2828125 + }, + { + "actual_token_mask_ratio": 0.2997090667486191, + "epoch": 3.6223450683735816, + "flow_matching_loss": 0.0126840197481215, + "grad_norm": 0.06868420541286469, + "kept_token_count": 573.20859375, + "learning_rate": 2.0338452945545275e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9928223669528962, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007177623314782977, + "mgd_loss_weight": 0.5851695120334626, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.3135498046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4264617919922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12450, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.3000311553478241, + "epoch": 3.6252545824847253, + "flow_matching_loss": 0.01387228276580572, + "grad_norm": 0.09330564737319946, + "kept_token_count": 572.84921875, + "learning_rate": 2.0322997612542366e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9931904196739196, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006809589313343167, + "mgd_loss_weight": 0.5848361551761627, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.11780700683596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.96778869628906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12460, + "valid_token_count": 818.39609375 + }, + { + "actual_token_mask_ratio": 0.30042007863521575, + "epoch": 3.6281640965958686, + "flow_matching_loss": 0.013891194202005862, + "grad_norm": 0.09259363263845444, + "kept_token_count": 572.41171875, + "learning_rate": 2.0307535811595427e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9927005767822266, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007299418561160564, + "mgd_loss_weight": 0.5845028162002563, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.45310363769534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.51430053710936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12470, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.30044586062431333, + "epoch": 3.631073610707012, + "flow_matching_loss": 0.014109705947339535, + "grad_norm": 0.06978919357061386, + "kept_token_count": 572.5078125, + "learning_rate": 2.0292067561492013e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9927393019199371, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007260696636512876, + "mgd_loss_weight": 0.58416947722435, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.5800537109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1455444335937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12480, + "valid_token_count": 818.38984375 + }, + { + "actual_token_mask_ratio": 0.3002105623483658, + "epoch": 3.6339831248181556, + "flow_matching_loss": 0.013980330247431994, + "grad_norm": 0.07954397797584534, + "kept_token_count": 572.57265625, + "learning_rate": 2.02765928810275e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9925565481185913, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074434512294828895, + "mgd_loss_weight": 0.5838361203670501, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.08014221191405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6759521484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12490, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.300105020403862, + "epoch": 3.636892638929299, + "flow_matching_loss": 0.013764329813420772, + "grad_norm": 0.07431183010339737, + "kept_token_count": 572.53828125, + "learning_rate": 2.0261111789005087e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9935027301311493, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006497269868850708, + "mgd_loss_weight": 0.5835027873516083, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.42713317871096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.76219787597654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12500, + "valid_token_count": 818.03359375 + }, + { + "actual_token_mask_ratio": 0.300581094622612, + "epoch": 3.639802153040442, + "flow_matching_loss": 0.013353298790752888, + "grad_norm": 0.08133693039417267, + "kept_token_count": 572.1921875, + "learning_rate": 2.024562430423576e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9929383158683777, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007061682781204582, + "mgd_loss_weight": 0.5831694364547729, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.7229248046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.7845764160156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12510, + "valid_token_count": 818.09140625 + }, + { + "actual_token_mask_ratio": 0.30020698010921476, + "epoch": 3.642711667151586, + "flow_matching_loss": 0.013689312059432268, + "grad_norm": 0.08927404880523682, + "kept_token_count": 572.5265625, + "learning_rate": 2.0230130445538273e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9926323115825653, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007367691257968545, + "mgd_loss_weight": 0.5828360915184021, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.75914611816404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.92474060058595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12520, + "valid_token_count": 818.13515625 + }, + { + "actual_token_mask_ratio": 0.30012896060943606, + "epoch": 3.645621181262729, + "flow_matching_loss": 0.013877553585916757, + "grad_norm": 0.07965933531522751, + "kept_token_count": 572.60546875, + "learning_rate": 2.0214630231739127e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9930685520172119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006931448820978403, + "mgd_loss_weight": 0.5825027525424957, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.0707275390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.5672882080078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12530, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.30018578469753265, + "epoch": 3.6485306953738723, + "flow_matching_loss": 0.013301426265388726, + "grad_norm": 0.06469118595123291, + "kept_token_count": 572.471875, + "learning_rate": 2.0199123681672543e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9927905321121215, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007209461648017168, + "mgd_loss_weight": 0.5821694016456604, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.2762908935547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0296630859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12540, + "valid_token_count": 818.03515625 + }, + { + "actual_token_mask_ratio": 0.30065953731536865, + "epoch": 3.651440209485016, + "flow_matching_loss": 0.013240918144583702, + "grad_norm": 0.08583569526672363, + "kept_token_count": 572.3, + "learning_rate": 2.0183610814180437e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9929848790168763, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007015120889991522, + "mgd_loss_weight": 0.5818360626697541, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.81348876953126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.28862915039065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12550, + "valid_token_count": 818.34453125 + }, + { + "actual_token_mask_ratio": 0.30017380714416503, + "epoch": 3.6543497235961593, + "flow_matching_loss": 0.014544273726642132, + "grad_norm": 0.08088529855012894, + "kept_token_count": 572.56328125, + "learning_rate": 2.016809164811241e-05, + "loss": 0.0189, + "mgd_cos_sim": 0.9925430715084076, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007456927001476288, + "mgd_loss_weight": 0.5815027177333831, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.263818359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.3731719970703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12560, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.29957334101200106, + "epoch": 3.657259237707303, + "flow_matching_loss": 0.013209454715251923, + "grad_norm": 0.07968824356794357, + "kept_token_count": 573.1390625, + "learning_rate": 2.015256620232571e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9928092360496521, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071907664183527235, + "mgd_loss_weight": 0.5811693668365479, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.0721496582031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.4812286376953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12570, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.30032235383987427, + "epoch": 3.6601687518184463, + "flow_matching_loss": 0.013482409995049238, + "grad_norm": 0.08273568749427795, + "kept_token_count": 572.55703125, + "learning_rate": 2.0137034495685222e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9927425146102905, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007257487997412682, + "mgd_loss_weight": 0.5808360278606415, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.0032165527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.7167236328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12580, + "valid_token_count": 818.31640625 + }, + { + "actual_token_mask_ratio": 0.2992252230644226, + "epoch": 3.66307826592959, + "flow_matching_loss": 0.012692655716091395, + "grad_norm": 0.09395083785057068, + "kept_token_count": 573.47265625, + "learning_rate": 2.012149654706343e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9927677750587464, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007232214882969856, + "mgd_loss_weight": 0.5805026888847351, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.2718841552734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.6321044921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12590, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.3001973181962967, + "epoch": 3.6659877800407332, + "flow_matching_loss": 0.014003513101488351, + "grad_norm": 0.07389902323484421, + "kept_token_count": 572.71484375, + "learning_rate": 2.010595237534041e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9930494427680969, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069505583960562944, + "mgd_loss_weight": 0.5801693379878998, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.5149169921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9682586669922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12600, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.30009744465351107, + "epoch": 3.6688972941518765, + "flow_matching_loss": 0.013988312985748052, + "grad_norm": 0.09136838465929031, + "kept_token_count": 572.55390625, + "learning_rate": 2.0090401999403793e-05, + "loss": 0.018, + "mgd_cos_sim": 0.993098795413971, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006901200767606497, + "mgd_loss_weight": 0.5798359930515289, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.29229431152345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.347998046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12610, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.3003249317407608, + "epoch": 3.67180680826302, + "flow_matching_loss": 0.013095596618950368, + "grad_norm": 0.08020293712615967, + "kept_token_count": 572.32578125, + "learning_rate": 2.007484543814875e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9931138455867767, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00688616274856031, + "mgd_loss_weight": 0.5795026540756225, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.3325500488281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.3616119384766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12620, + "valid_token_count": 817.9953125 + }, + { + "actual_token_mask_ratio": 0.30038564205169677, + "epoch": 3.6747163223741635, + "flow_matching_loss": 0.013587937410920858, + "grad_norm": 0.09208013862371445, + "kept_token_count": 572.39375, + "learning_rate": 2.0059282710477972e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9926818668842315, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007318139774724841, + "mgd_loss_weight": 0.5791693031787872, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.61754150390624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.6171539306641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12630, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.2998562902212143, + "epoch": 3.6776258364853067, + "flow_matching_loss": 0.012439023982733489, + "grad_norm": 0.06392264366149902, + "kept_token_count": 572.88828125, + "learning_rate": 2.0043713835301647e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9927674114704133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007232595281675458, + "mgd_loss_weight": 0.5788359642028809, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.1505828857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9427490234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12640, + "valid_token_count": 818.24453125 + }, + { + "actual_token_mask_ratio": 0.29992533326148985, + "epoch": 3.6805353505964504, + "flow_matching_loss": 0.013517768029123545, + "grad_norm": 0.07753030210733414, + "kept_token_count": 572.7984375, + "learning_rate": 2.0028138831537417e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.993059641122818, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006940367259085178, + "mgd_loss_weight": 0.57850261926651, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.12138671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.88462524414064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12650, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.29986642897129057, + "epoch": 3.6834448647075937, + "flow_matching_loss": 0.013150317780673503, + "grad_norm": 0.0939095988869667, + "kept_token_count": 572.9578125, + "learning_rate": 2.0012557718110387e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9931253671646119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006874641450121999, + "mgd_loss_weight": 0.5781692683696746, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.56768798828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.1200836181641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12660, + "valid_token_count": 818.35546875 + }, + { + "actual_token_mask_ratio": 0.29982950091362, + "epoch": 3.6863543788187374, + "flow_matching_loss": 0.01308548366650939, + "grad_norm": 0.0819488912820816, + "kept_token_count": 572.95078125, + "learning_rate": 1.9996970513953077e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9930975675582886, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006902413116768003, + "mgd_loss_weight": 0.5778359353542328, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.43138427734374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1593780517578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12670, + "valid_token_count": 818.30234375 + }, + { + "actual_token_mask_ratio": 0.3005042105913162, + "epoch": 3.6892638929298807, + "flow_matching_loss": 0.013624375499784947, + "grad_norm": 0.08501702547073364, + "kept_token_count": 572.2671875, + "learning_rate": 1.9981377238005405e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9928986072540283, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007101390231400728, + "mgd_loss_weight": 0.577502578496933, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.4971405029297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.55736083984374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12680, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.29977399706840513, + "epoch": 3.6921734070410244, + "flow_matching_loss": 0.01346116904169321, + "grad_norm": 0.07670200616121292, + "kept_token_count": 572.9703125, + "learning_rate": 1.9965777909214674e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9926934719085694, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007306535309180617, + "mgd_loss_weight": 0.5771692395210266, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.3321075439453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1088165283203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12690, + "valid_token_count": 818.26328125 + }, + { + "actual_token_mask_ratio": 0.2998010694980621, + "epoch": 3.6950829211521676, + "flow_matching_loss": 0.012042441591620445, + "grad_norm": 0.11413890868425369, + "kept_token_count": 573.028125, + "learning_rate": 1.9950172546535545e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9924244403839111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007575566554442048, + "mgd_loss_weight": 0.5768359005451202, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.15271911621096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1036010742188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12700, + "valid_token_count": 818.37578125 + }, + { + "actual_token_mask_ratio": 0.2998765349388123, + "epoch": 3.697992435263311, + "flow_matching_loss": 0.01375530455261469, + "grad_norm": 0.09021077305078506, + "kept_token_count": 572.9546875, + "learning_rate": 1.9934561168930002e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9927752912044525, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007224714569747448, + "mgd_loss_weight": 0.5765025436878204, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.4299041748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8622802734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12710, + "valid_token_count": 818.36328125 + }, + { + "actual_token_mask_ratio": 0.29970057904720304, + "epoch": 3.7009019493744546, + "flow_matching_loss": 0.013135042320936919, + "grad_norm": 0.0894394963979721, + "kept_token_count": 573.025, + "learning_rate": 1.991894379536734e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9928722321987152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007127768918871879, + "mgd_loss_weight": 0.5761692106723786, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.68418273925784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.66371765136716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12720, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.30028838515281675, + "epoch": 3.703811463485598, + "flow_matching_loss": 0.013493156339973212, + "grad_norm": 0.07860314846038818, + "kept_token_count": 572.69375, + "learning_rate": 1.9903320444824137e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9924694895744324, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007530514523386956, + "mgd_loss_weight": 0.5758358597755432, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.17906494140624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.86211853027345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12730, + "valid_token_count": 818.471875 + }, + { + "actual_token_mask_ratio": 0.29956718981266023, + "epoch": 3.706720977596741, + "flow_matching_loss": 0.013788098935037851, + "grad_norm": 0.08372165262699127, + "kept_token_count": 573.165625, + "learning_rate": 1.9887691136284245e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9927804410457611, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072195596061646935, + "mgd_loss_weight": 0.5755025148391724, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.98272399902345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.64942321777346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12740, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.2997570335865021, + "epoch": 3.709630491707885, + "flow_matching_loss": 0.01488687340170145, + "grad_norm": 0.08068577200174332, + "kept_token_count": 572.93828125, + "learning_rate": 1.9872055888738745e-05, + "loss": 0.0191, + "mgd_cos_sim": 0.9927019238471985, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007298081601038575, + "mgd_loss_weight": 0.575169175863266, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.5105712890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.93982849121096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12750, + "valid_token_count": 818.19296875 + }, + { + "actual_token_mask_ratio": 0.30034769475460055, + "epoch": 3.712540005819028, + "flow_matching_loss": 0.012538076844066381, + "grad_norm": 0.07843858003616333, + "kept_token_count": 572.54375, + "learning_rate": 1.985641472118593e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9928675711154937, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007132422085851431, + "mgd_loss_weight": 0.5748358249664307, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.1960510253906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.40513305664064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12760, + "valid_token_count": 818.32265625 + }, + { + "actual_token_mask_ratio": 0.2995254397392273, + "epoch": 3.715449519930172, + "flow_matching_loss": 0.013525017257779836, + "grad_norm": 0.06968022882938385, + "kept_token_count": 573.2640625, + "learning_rate": 1.9840767652631294e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9927132427692413, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072867573704570535, + "mgd_loss_weight": 0.5745024859905243, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.0528167724609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.94336853027346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12770, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.3001042515039444, + "epoch": 3.718359034041315, + "flow_matching_loss": 0.01416536644101143, + "grad_norm": 0.07594653964042664, + "kept_token_count": 572.85703125, + "learning_rate": 1.9825114702087513e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9928323864936829, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007167614623904228, + "mgd_loss_weight": 0.5741691410541534, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.6695922851562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.2928527832031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12780, + "valid_token_count": 818.4921875 + }, + { + "actual_token_mask_ratio": 0.2999675750732422, + "epoch": 3.7212685481524588, + "flow_matching_loss": 0.01244168058037758, + "grad_norm": 0.06692767888307571, + "kept_token_count": 572.80078125, + "learning_rate": 1.9809455888574388e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.992943388223648, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007056619087234139, + "mgd_loss_weight": 0.5738357901573181, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.9989440917969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0649017333984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12790, + "valid_token_count": 818.25078125 + }, + { + "actual_token_mask_ratio": 0.29986250400543213, + "epoch": 3.724178062263602, + "flow_matching_loss": 0.015089027117937803, + "grad_norm": 0.08787406980991364, + "kept_token_count": 572.9609375, + "learning_rate": 1.9793791231118864e-05, + "loss": 0.0193, + "mgd_cos_sim": 0.9925897121429443, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007410291815176606, + "mgd_loss_weight": 0.5735024511814117, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.9117156982422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.3636993408203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12800, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.2998515546321869, + "epoch": 3.7270875763747453, + "flow_matching_loss": 0.013388063199818135, + "grad_norm": 0.08464612811803818, + "kept_token_count": 573.0171875, + "learning_rate": 1.9778120748754958e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9926611006259918, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073388791643083096, + "mgd_loss_weight": 0.5731691062450409, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.78123779296874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.82721252441405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12810, + "valid_token_count": 818.42109375 + }, + { + "actual_token_mask_ratio": 0.3003911405801773, + "epoch": 3.729997090485889, + "flow_matching_loss": 0.012878082506358624, + "grad_norm": 0.09648917615413666, + "kept_token_count": 572.42265625, + "learning_rate": 1.97624444605238e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9928530693054199, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071469177957624195, + "mgd_loss_weight": 0.5728357613086701, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.9836120605469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9325469970703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12820, + "valid_token_count": 818.21015625 + }, + { + "actual_token_mask_ratio": 0.3009676694869995, + "epoch": 3.7329066045970323, + "flow_matching_loss": 0.012678525689989329, + "grad_norm": 0.0809156596660614, + "kept_token_count": 572.04140625, + "learning_rate": 1.9746762385473552e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9926477372646332, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007352250861003995, + "mgd_loss_weight": 0.5725024163722991, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.9537719726562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8733642578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12830, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.3002229332923889, + "epoch": 3.7358161187081755, + "flow_matching_loss": 0.013504156097769737, + "grad_norm": 0.08903378993272781, + "kept_token_count": 572.528125, + "learning_rate": 1.9731074542659414e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.992907440662384, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007092554681003093, + "mgd_loss_weight": 0.5721690714359283, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.24925842285154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.4763671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12840, + "valid_token_count": 818.16171875 + }, + { + "actual_token_mask_ratio": 0.30032232105731965, + "epoch": 3.7387256328193192, + "flow_matching_loss": 0.014202219340950251, + "grad_norm": 0.09283551573753357, + "kept_token_count": 572.52890625, + "learning_rate": 1.9715380951143603e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9926110684871674, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00738893230445683, + "mgd_loss_weight": 0.5718357264995575, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.3453643798828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.2113342285156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12850, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.2999359041452408, + "epoch": 3.7416351469304625, + "flow_matching_loss": 0.013783866912126541, + "grad_norm": 0.11907405406236649, + "kept_token_count": 572.78828125, + "learning_rate": 1.9699681629995295e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9928962588310242, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007103728828951716, + "mgd_loss_weight": 0.5715023875236511, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.81556701660156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4476379394531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12860, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.29970013797283174, + "epoch": 3.744544661041606, + "flow_matching_loss": 0.012422513775527477, + "grad_norm": 0.08916979283094406, + "kept_token_count": 572.975, + "learning_rate": 1.968397659829066e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9929145336151123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007085460424423218, + "mgd_loss_weight": 0.5711690366268158, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.66978759765624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2736053466797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12870, + "valid_token_count": 818.18359375 + }, + { + "actual_token_mask_ratio": 0.29971342980861665, + "epoch": 3.7474541751527495, + "flow_matching_loss": 0.012488322705030442, + "grad_norm": 0.0788990780711174, + "kept_token_count": 573.0, + "learning_rate": 1.966826587511279e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9929107546806335, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00708924587816, + "mgd_loss_weight": 0.5708356916904449, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.49561767578126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.951708984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12880, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.2998378038406372, + "epoch": 3.750363689263893, + "flow_matching_loss": 0.01302214413881302, + "grad_norm": 0.08745697140693665, + "kept_token_count": 572.796875, + "learning_rate": 1.965254947955169e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9926958084106445, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073041889350861314, + "mgd_loss_weight": 0.5705023527145385, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.0069213867188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.3709442138672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12890, + "valid_token_count": 818.09921875 + }, + { + "actual_token_mask_ratio": 0.2997554033994675, + "epoch": 3.7532732033750364, + "flow_matching_loss": 0.014475893136113882, + "grad_norm": 0.09190583974123001, + "kept_token_count": 572.925, + "learning_rate": 1.963682743070427e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9924908995628356, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007509101321920753, + "mgd_loss_weight": 0.5701690018177032, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.85584411621096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.2367797851563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12900, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.3000571697950363, + "epoch": 3.7561827174861797, + "flow_matching_loss": 0.012225091736763716, + "grad_norm": 0.10488617420196533, + "kept_token_count": 572.73203125, + "learning_rate": 1.962109974767429e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9928357124328613, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007164289196953178, + "mgd_loss_weight": 0.5698356628417969, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.2789703369141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.5603790283203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12910, + "valid_token_count": 818.26171875 + }, + { + "actual_token_mask_ratio": 0.29949933886528013, + "epoch": 3.7590922315973234, + "flow_matching_loss": 0.0129100420512259, + "grad_norm": 0.07541273534297943, + "kept_token_count": 573.28515625, + "learning_rate": 1.960536644957238e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9926315307617187, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007368475617840886, + "mgd_loss_weight": 0.569502317905426, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.9075103759766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.85777893066404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12920, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.2994459480047226, + "epoch": 3.7620017457084667, + "flow_matching_loss": 0.013470904529094696, + "grad_norm": 0.08099522441625595, + "kept_token_count": 573.334375, + "learning_rate": 1.9589627555515974e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9927845895290375, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072154067922383545, + "mgd_loss_weight": 0.5691689670085907, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.7579010009766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.89210815429686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12930, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.29961156845092773, + "epoch": 3.76491125981961, + "flow_matching_loss": 0.013793728034943342, + "grad_norm": 0.09210773557424545, + "kept_token_count": 573.2046875, + "learning_rate": 1.9573883084629317e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9925531804561615, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007446820056065917, + "mgd_loss_weight": 0.5688356339931488, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.9130432128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.5763336181641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12940, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.3005833774805069, + "epoch": 3.7678207739307537, + "flow_matching_loss": 0.014071017690002918, + "grad_norm": 0.08039843291044235, + "kept_token_count": 572.14375, + "learning_rate": 1.9558133056043422e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9927895247936249, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007210477208718658, + "mgd_loss_weight": 0.568502277135849, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.7401123046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.7683563232422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12950, + "valid_token_count": 818.03125 + }, + { + "actual_token_mask_ratio": 0.30026068687438967, + "epoch": 3.770730288041897, + "flow_matching_loss": 0.012165889889001847, + "grad_norm": 0.09499561041593552, + "kept_token_count": 572.4953125, + "learning_rate": 1.9542377488896062e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9925420045852661, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007457996672019363, + "mgd_loss_weight": 0.5681689381599426, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.64234313964846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2716552734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12960, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.29976398348808286, + "epoch": 3.77363980215304, + "flow_matching_loss": 0.013506453391164542, + "grad_norm": 0.06609472632408142, + "kept_token_count": 572.8640625, + "learning_rate": 1.9526616402331733e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.993274736404419, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006725276727229357, + "mgd_loss_weight": 0.5678355991840363, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.6255615234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.1307861328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12970, + "valid_token_count": 818.11171875 + }, + { + "actual_token_mask_ratio": 0.29968405365943906, + "epoch": 3.776549316264184, + "flow_matching_loss": 0.01299670897424221, + "grad_norm": 0.09858689457178116, + "kept_token_count": 573.146875, + "learning_rate": 1.9510849815501638e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9927793502807617, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007220639195293188, + "mgd_loss_weight": 0.5675022423267364, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.4654510498047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.35224609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12980, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.30070008635520934, + "epoch": 3.7794588303753276, + "flow_matching_loss": 0.013349883537739516, + "grad_norm": 0.0892663449048996, + "kept_token_count": 572.37890625, + "learning_rate": 1.949507774756368e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.993043738603592, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006956269731745124, + "mgd_loss_weight": 0.5671689093112946, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.73570861816404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5151611328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 12990, + "valid_token_count": 818.50390625 + }, + { + "actual_token_mask_ratio": 0.30008667409420015, + "epoch": 3.782368344486471, + "flow_matching_loss": 0.014351926650851965, + "grad_norm": 0.08086558431386948, + "kept_token_count": 572.74140625, + "learning_rate": 1.9479300217682394e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9923684537410736, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00763155035674572, + "mgd_loss_weight": 0.5668355584144592, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.662109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.63108825683594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13000, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.300022691488266, + "epoch": 3.785277858597614, + "flow_matching_loss": 0.01372713390737772, + "grad_norm": 0.09562984108924866, + "kept_token_count": 572.89140625, + "learning_rate": 1.946351724502898e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9924410998821258, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007558906823396683, + "mgd_loss_weight": 0.5665022134780884, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.4464385986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.59864807128906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13010, + "valid_token_count": 818.44140625 + }, + { + "actual_token_mask_ratio": 0.3004036843776703, + "epoch": 3.788187372708758, + "flow_matching_loss": 0.013396899122744799, + "grad_norm": 0.08631058037281036, + "kept_token_count": 572.36796875, + "learning_rate": 1.9447728848781227e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.992511647939682, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0074883468449115755, + "mgd_loss_weight": 0.566168874502182, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.7421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4722564697266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13020, + "valid_token_count": 818.1359375 + }, + { + "actual_token_mask_ratio": 0.30039999186992644, + "epoch": 3.791096886819901, + "flow_matching_loss": 0.013595364056527615, + "grad_norm": 0.07983507215976715, + "kept_token_count": 572.38671875, + "learning_rate": 1.943193504812353e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9929482042789459, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007051791809499264, + "mgd_loss_weight": 0.5658355295658112, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.1984436035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1146514892578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13030, + "valid_token_count": 818.159375 + }, + { + "actual_token_mask_ratio": 0.30009472370147705, + "epoch": 3.7940064009310444, + "flow_matching_loss": 0.01268404396250844, + "grad_norm": 0.07559817284345627, + "kept_token_count": 572.70859375, + "learning_rate": 1.9416135862246846e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9930695056915283, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006930497288703918, + "mgd_loss_weight": 0.5655021846294404, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.0088287353516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6220336914063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13040, + "valid_token_count": 818.26640625 + }, + { + "actual_token_mask_ratio": 0.2995174765586853, + "epoch": 3.796915915042188, + "flow_matching_loss": 0.011826609168201685, + "grad_norm": 0.07771098613739014, + "kept_token_count": 573.24453125, + "learning_rate": 1.9400331310348673e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9929089665412902, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007091041095554829, + "mgd_loss_weight": 0.5651688396930694, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.4439758300781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9818511962891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13050, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.3001975744962692, + "epoch": 3.7998254291533313, + "flow_matching_loss": 0.014104918576776981, + "grad_norm": 0.08554104715585709, + "kept_token_count": 572.6671875, + "learning_rate": 1.9384521411633033e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9928311049938202, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007168886624276638, + "mgd_loss_weight": 0.5648354947566986, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.8618957519531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.13442077636716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13060, + "valid_token_count": 818.32578125 + }, + { + "actual_token_mask_ratio": 0.30032541453838346, + "epoch": 3.8027349432644746, + "flow_matching_loss": 0.014108637627214194, + "grad_norm": 0.09063996374607086, + "kept_token_count": 572.4640625, + "learning_rate": 1.936870618531045e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9925958752632141, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007404116308316588, + "mgd_loss_weight": 0.5645021498203278, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.1018371582031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.50022583007814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13070, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.29988002181053164, + "epoch": 3.8056444573756183, + "flow_matching_loss": 0.013141191843897105, + "grad_norm": 0.07782526314258575, + "kept_token_count": 572.8734375, + "learning_rate": 1.9352885650597904e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9926835775375367, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007316423999145627, + "mgd_loss_weight": 0.5641688108444214, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.5340301513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6635040283203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13080, + "valid_token_count": 818.25 + }, + { + "actual_token_mask_ratio": 0.29938305020332334, + "epoch": 3.8085539714867616, + "flow_matching_loss": 0.013018697127699852, + "grad_norm": 0.10873347520828247, + "kept_token_count": 573.2171875, + "learning_rate": 1.9337059826718847e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9927586913108826, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007241309387609363, + "mgd_loss_weight": 0.5638354599475861, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.0688781738281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.166455078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13090, + "valid_token_count": 818.16171875 + }, + { + "actual_token_mask_ratio": 0.3003876268863678, + "epoch": 3.8114634855979053, + "flow_matching_loss": 0.012048173788934946, + "grad_norm": 0.09120252728462219, + "kept_token_count": 572.26875, + "learning_rate": 1.932122873290314e-05, + "loss": 0.016, + "mgd_cos_sim": 0.992919796705246, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007080208137631416, + "mgd_loss_weight": 0.5635021150112152, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.41466369628904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6675598144531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13100, + "valid_token_count": 817.97265625 + }, + { + "actual_token_mask_ratio": 0.30103825628757475, + "epoch": 3.8143729997090485, + "flow_matching_loss": 0.013262744061648846, + "grad_norm": 0.0909193605184555, + "kept_token_count": 571.92734375, + "learning_rate": 1.9305392388387054e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9925682246685028, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007431771932169795, + "mgd_loss_weight": 0.5631687760353088, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.2071197509766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.01478576660156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13110, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.299659264087677, + "epoch": 3.8172825138201922, + "flow_matching_loss": 0.013763402495533228, + "grad_norm": 0.06460416316986084, + "kept_token_count": 573.14296875, + "learning_rate": 1.928955081241325e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9925999224185944, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007400079490616917, + "mgd_loss_weight": 0.5628354251384735, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.1549102783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5300354003906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13120, + "valid_token_count": 818.37734375 + }, + { + "actual_token_mask_ratio": 0.3002649128437042, + "epoch": 3.8201920279313355, + "flow_matching_loss": 0.013211948983371257, + "grad_norm": 0.06770589202642441, + "kept_token_count": 572.55234375, + "learning_rate": 1.9273704024230725e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.993055808544159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006944194389507175, + "mgd_loss_weight": 0.5625020861625671, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.9095428466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.90309143066406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13130, + "valid_token_count": 818.24609375 + }, + { + "actual_token_mask_ratio": 0.30044332444667815, + "epoch": 3.8231015420424788, + "flow_matching_loss": 0.013890847098082304, + "grad_norm": 0.07534603029489517, + "kept_token_count": 572.50234375, + "learning_rate": 1.925785204309483e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9928127288818359, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007187271257862449, + "mgd_loss_weight": 0.5621687412261963, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.56127014160154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 375.6942657470703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13140, + "valid_token_count": 818.38203125 + }, + { + "actual_token_mask_ratio": 0.29976014196872713, + "epoch": 3.8260110561536225, + "flow_matching_loss": 0.012730932142585517, + "grad_norm": 0.09475608915090561, + "kept_token_count": 572.9390625, + "learning_rate": 1.9241994888267212e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9929333209991456, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007066694786772132, + "mgd_loss_weight": 0.5618353903293609, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.14207458496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5699768066406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13150, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.30001467764377593, + "epoch": 3.8289205702647657, + "flow_matching_loss": 0.013812564872205257, + "grad_norm": 0.09374354034662247, + "kept_token_count": 572.96640625, + "learning_rate": 1.9226132579015815e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9927501976490021, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007249804679304361, + "mgd_loss_weight": 0.5615020573139191, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.4001068115234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.0104248046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13160, + "valid_token_count": 818.540625 + }, + { + "actual_token_mask_ratio": 0.29943384528160094, + "epoch": 3.831830084375909, + "flow_matching_loss": 0.015670662466436624, + "grad_norm": 0.07787129282951355, + "kept_token_count": 573.31015625, + "learning_rate": 1.9210265134614832e-05, + "loss": 0.0197, + "mgd_cos_sim": 0.9928914427757263, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00710854884237051, + "mgd_loss_weight": 0.5611687004566193, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.51299438476565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9135375976563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13170, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.30021013617515563, + "epoch": 3.8347395984870527, + "flow_matching_loss": 0.013354067225009203, + "grad_norm": 0.0785263404250145, + "kept_token_count": 572.61484375, + "learning_rate": 1.9194392574344712e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9927215695381164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00727843539789319, + "mgd_loss_weight": 0.5608353614807129, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.80885009765626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.36891174316406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13180, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.30021595060825346, + "epoch": 3.837649112598196, + "flow_matching_loss": 0.01264815921895206, + "grad_norm": 0.08193079382181168, + "kept_token_count": 572.61796875, + "learning_rate": 1.9178514917492103e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9927537441253662, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072462542448192835, + "mgd_loss_weight": 0.5605020225048065, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.0500457763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8964538574219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13190, + "valid_token_count": 818.27421875 + }, + { + "actual_token_mask_ratio": 0.29964247941970823, + "epoch": 3.8405586267093397, + "flow_matching_loss": 0.014867044985294342, + "grad_norm": 0.08927516639232635, + "kept_token_count": 573.021875, + "learning_rate": 1.9162632183349866e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.993002200126648, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006997789489105344, + "mgd_loss_weight": 0.5601686656475067, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.7218780517578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2225799560547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13200, + "valid_token_count": 818.18203125 + }, + { + "actual_token_mask_ratio": 0.2994567215442657, + "epoch": 3.843468140820483, + "flow_matching_loss": 0.012997108977288008, + "grad_norm": 0.08093845099210739, + "kept_token_count": 573.37890625, + "learning_rate": 1.914674439121701e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9928082644939422, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00719174426048994, + "mgd_loss_weight": 0.5598353326320649, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.25336303710935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.50079040527345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13210, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.2991252958774567, + "epoch": 3.8463776549316266, + "flow_matching_loss": 0.013567680586129427, + "grad_norm": 0.08336493372917175, + "kept_token_count": 573.54921875, + "learning_rate": 1.9130851560398715e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9928121507167816, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007187858410179615, + "mgd_loss_weight": 0.5595019817352295, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.31415405273435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0484924316406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13220, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.3001475155353546, + "epoch": 3.84928716904277, + "flow_matching_loss": 0.011995017994195223, + "grad_norm": 0.07225118577480316, + "kept_token_count": 572.6703125, + "learning_rate": 1.911495371020625e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9930899024009705, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006910086097195745, + "mgd_loss_weight": 0.5591686367988586, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.9977966308594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0152191162109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13230, + "valid_token_count": 818.26796875 + }, + { + "actual_token_mask_ratio": 0.300403019785881, + "epoch": 3.852196683153913, + "flow_matching_loss": 0.012879230827093125, + "grad_norm": 0.0857790932059288, + "kept_token_count": 572.434375, + "learning_rate": 1.909905085995702e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9929734706878662, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007026539975777268, + "mgd_loss_weight": 0.5588352978229523, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.4506805419922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.95384216308594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13240, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.30011842846870423, + "epoch": 3.855106197265057, + "flow_matching_loss": 0.013363015558570623, + "grad_norm": 0.0790695995092392, + "kept_token_count": 572.7546875, + "learning_rate": 1.908314302897448e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9929618299007416, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007038168516010046, + "mgd_loss_weight": 0.558501946926117, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.32566833496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.26190185546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13250, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.29990178942680357, + "epoch": 3.8580157113762, + "flow_matching_loss": 0.014334041718393564, + "grad_norm": 0.09340524673461914, + "kept_token_count": 572.875, + "learning_rate": 1.9067230236588142e-05, + "loss": 0.0183, + "mgd_cos_sim": 0.9928539812564849, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007146021304652095, + "mgd_loss_weight": 0.5581686079502106, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.1221221923828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9417694091797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13260, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.30043112933635713, + "epoch": 3.8609252254873434, + "flow_matching_loss": 0.01246027983725071, + "grad_norm": 0.08451543748378754, + "kept_token_count": 572.35234375, + "learning_rate": 1.905131250213355e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9927991449832916, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007200853293761611, + "mgd_loss_weight": 0.5578352630138397, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 395.6665985107422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.7409240722656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13270, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.3004176586866379, + "epoch": 3.863834739598487, + "flow_matching_loss": 0.013201102754101158, + "grad_norm": 0.08040694147348404, + "kept_token_count": 572.48515625, + "learning_rate": 1.9035389844952256e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9933773219585419, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006622678134590387, + "mgd_loss_weight": 0.5575019121170044, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.89695739746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9285614013672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13280, + "valid_token_count": 818.32421875 + }, + { + "actual_token_mask_ratio": 0.30041552782058717, + "epoch": 3.8667442537096304, + "flow_matching_loss": 0.015593390725553036, + "grad_norm": 0.08663611859083176, + "kept_token_count": 572.42109375, + "learning_rate": 1.901946228439179e-05, + "loss": 0.0196, + "mgd_cos_sim": 0.9928829848766327, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007117023039609194, + "mgd_loss_weight": 0.557168573141098, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.2666900634766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.53150634765626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13290, + "valid_token_count": 818.22734375 + }, + { + "actual_token_mask_ratio": 0.2996831893920898, + "epoch": 3.869653767820774, + "flow_matching_loss": 0.012930064648389816, + "grad_norm": 0.07324035465717316, + "kept_token_count": 572.94140625, + "learning_rate": 1.9003529839805637e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9930033266544342, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006996675813570619, + "mgd_loss_weight": 0.5568352282047272, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.0618621826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.73253479003904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13300, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.2995472550392151, + "epoch": 3.8725632819319173, + "flow_matching_loss": 0.012026573345065118, + "grad_norm": 0.06773779541254044, + "kept_token_count": 573.109375, + "learning_rate": 1.8987592530553217e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9930911242961884, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006908871373161674, + "mgd_loss_weight": 0.5565018832683564, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.6761199951172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.15526123046874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13310, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.30004665851593015, + "epoch": 3.875472796043061, + "flow_matching_loss": 0.015964536182582378, + "grad_norm": 0.10026875138282776, + "kept_token_count": 572.84765625, + "learning_rate": 1.897165037599987e-05, + "loss": 0.02, + "mgd_cos_sim": 0.9927896916866302, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007210315996780991, + "mgd_loss_weight": 0.5561685383319854, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.6790283203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3618103027344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13320, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.2996764570474625, + "epoch": 3.8783823101542043, + "flow_matching_loss": 0.01248485380783677, + "grad_norm": 0.07465841621160507, + "kept_token_count": 572.94375, + "learning_rate": 1.8955703395516814e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9930643737316132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006935620261356234, + "mgd_loss_weight": 0.5558351933956146, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.4584533691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.604638671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13330, + "valid_token_count": 818.11171875 + }, + { + "actual_token_mask_ratio": 0.30049918591976166, + "epoch": 3.8812918242653476, + "flow_matching_loss": 0.01365891396999359, + "grad_norm": 0.10837821662425995, + "kept_token_count": 572.4890625, + "learning_rate": 1.8939751608481134e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.993002712726593, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006997273582965135, + "mgd_loss_weight": 0.5555018484592438, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.99094543457034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7236694335937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13340, + "valid_token_count": 818.42578125 + }, + { + "actual_token_mask_ratio": 0.30015774965286257, + "epoch": 3.8842013383764913, + "flow_matching_loss": 0.013758638501167297, + "grad_norm": 0.07039976865053177, + "kept_token_count": 572.71328125, + "learning_rate": 1.892379503427575e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9928247690200805, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007175223669037223, + "mgd_loss_weight": 0.5551685094833374, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.09581604003904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.62453308105466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13350, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.29952595829963685, + "epoch": 3.8871108524876345, + "flow_matching_loss": 0.013275930006057025, + "grad_norm": 0.08031964302062988, + "kept_token_count": 573.2984375, + "learning_rate": 1.8907833692289412e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9929859876632691, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007014010427519679, + "mgd_loss_weight": 0.5548351585865021, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.13275146484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.52525024414064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13360, + "valid_token_count": 818.44140625 + }, + { + "actual_token_mask_ratio": 0.2998964011669159, + "epoch": 3.890020366598778, + "flow_matching_loss": 0.013959917332977056, + "grad_norm": 0.07518412172794342, + "kept_token_count": 572.72578125, + "learning_rate": 1.889186760191666e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.992465740442276, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00753425401635468, + "mgd_loss_weight": 0.5545018136501312, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.3376708984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0252319335938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13370, + "valid_token_count": 818.05859375 + }, + { + "actual_token_mask_ratio": 0.3003080040216446, + "epoch": 3.8929298807099215, + "flow_matching_loss": 0.014160181675106288, + "grad_norm": 0.10257413983345032, + "kept_token_count": 572.73046875, + "learning_rate": 1.887589678255779e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9928235828876495, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007176408218219877, + "mgd_loss_weight": 0.5541684746742248, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.4282592773437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6296356201172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13380, + "valid_token_count": 818.540625 + }, + { + "actual_token_mask_ratio": 0.29917094111442566, + "epoch": 3.895839394821065, + "flow_matching_loss": 0.013568804133683443, + "grad_norm": 0.07803784310817719, + "kept_token_count": 573.49765625, + "learning_rate": 1.8859921253618852e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9930424630641937, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006957550533115864, + "mgd_loss_weight": 0.5538351237773895, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.39064636230466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.66892700195314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13390, + "valid_token_count": 818.31171875 + }, + { + "actual_token_mask_ratio": 0.3002522587776184, + "epoch": 3.898748908932208, + "flow_matching_loss": 0.014291948080062867, + "grad_norm": 0.10156882554292679, + "kept_token_count": 572.5265625, + "learning_rate": 1.884394103451163e-05, + "loss": 0.018, + "mgd_cos_sim": 0.993215137720108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006784852733835578, + "mgd_loss_weight": 0.5535017848014832, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.0257080078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.37034912109374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13400, + "valid_token_count": 818.19140625 + }, + { + "actual_token_mask_ratio": 0.2998934119939804, + "epoch": 3.9016584230433518, + "flow_matching_loss": 0.012358658947050571, + "grad_norm": 0.07826471328735352, + "kept_token_count": 572.87578125, + "learning_rate": 1.8827956144653583e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9929819166660309, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007018092973157764, + "mgd_loss_weight": 0.5531684398651123, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 444.4864562988281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1420135498047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13410, + "valid_token_count": 818.26796875 + }, + { + "actual_token_mask_ratio": 0.3001232147216797, + "epoch": 3.9045679371544955, + "flow_matching_loss": 0.012344640493392945, + "grad_norm": 0.0833456814289093, + "kept_token_count": 572.76875, + "learning_rate": 1.8811966603467874e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9929986596107483, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007001343136653304, + "mgd_loss_weight": 0.5528350889682769, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.44130859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.93128967285156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13420, + "valid_token_count": 818.38984375 + }, + { + "actual_token_mask_ratio": 0.30038134157657626, + "epoch": 3.9074774512656387, + "flow_matching_loss": 0.013137765228748322, + "grad_norm": 0.0888843759894371, + "kept_token_count": 572.4328125, + "learning_rate": 1.8795972430383293e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.992716646194458, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007283338811248541, + "mgd_loss_weight": 0.5525017559528351, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.44249267578124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4142364501953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13430, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.30001347362995145, + "epoch": 3.910386965376782, + "flow_matching_loss": 0.012636930774897337, + "grad_norm": 0.10331905633211136, + "kept_token_count": 572.7109375, + "learning_rate": 1.8779973644834273e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9929009854793549, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007099037664011121, + "mgd_loss_weight": 0.5521683990955353, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.2375549316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5487091064453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13440, + "valid_token_count": 818.175 + }, + { + "actual_token_mask_ratio": 0.29934770464897154, + "epoch": 3.9132964794879257, + "flow_matching_loss": 0.012483481038361787, + "grad_norm": 0.08119948208332062, + "kept_token_count": 573.36328125, + "learning_rate": 1.876397026626085e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9932787120342255, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006721288664266467, + "mgd_loss_weight": 0.5518350601196289, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.6994995117187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9171417236328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13450, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.2999705642461777, + "epoch": 3.916205993599069, + "flow_matching_loss": 0.012610227055847644, + "grad_norm": 0.09228958934545517, + "kept_token_count": 572.8359375, + "learning_rate": 1.8747962314108633e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9929147362709045, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007085273275151849, + "mgd_loss_weight": 0.5515017211437225, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.8697265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.6855865478516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13460, + "valid_token_count": 818.30078125 + }, + { + "actual_token_mask_ratio": 0.29944824874401094, + "epoch": 3.919115507710212, + "flow_matching_loss": 0.012761078402400016, + "grad_norm": 0.11645031720399857, + "kept_token_count": 573.18203125, + "learning_rate": 1.873194980782879e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9930851459503174, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006914844457060099, + "mgd_loss_weight": 0.5511683702468873, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.13511962890624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.12633361816404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13470, + "valid_token_count": 818.18671875 + }, + { + "actual_token_mask_ratio": 0.2997817426919937, + "epoch": 3.922025021821356, + "flow_matching_loss": 0.012286537606269122, + "grad_norm": 0.06915650516748428, + "kept_token_count": 573.11484375, + "learning_rate": 1.871593276687804e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.993412309885025, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006587682338431477, + "mgd_loss_weight": 0.5508350312709809, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.5321807861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2284210205078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13480, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.301207160949707, + "epoch": 3.924934535932499, + "flow_matching_loss": 0.012560846749693155, + "grad_norm": 0.0900006964802742, + "kept_token_count": 571.60078125, + "learning_rate": 1.869991121071858e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9929865956306457, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007013407209888101, + "mgd_loss_weight": 0.5505016863346099, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.7924530029297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2085235595703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13490, + "valid_token_count": 817.98359375 + }, + { + "actual_token_mask_ratio": 0.3003430277109146, + "epoch": 3.9278440500436425, + "flow_matching_loss": 0.013834279403090477, + "grad_norm": 0.0985836312174797, + "kept_token_count": 572.51796875, + "learning_rate": 1.8683885158818117e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9927240967750549, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007275894656777382, + "mgd_loss_weight": 0.5501683354377747, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.7339233398437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.30848083496096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13500, + "valid_token_count": 818.28359375 + }, + { + "actual_token_mask_ratio": 0.3001902103424072, + "epoch": 3.930753564154786, + "flow_matching_loss": 0.01167153064161539, + "grad_norm": 0.08314226567745209, + "kept_token_count": 572.5859375, + "learning_rate": 1.8667854630649825e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9934645831584931, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006535412184894085, + "mgd_loss_weight": 0.5498349964618683, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.6063293457031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.75828247070314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13510, + "valid_token_count": 818.20078125 + }, + { + "actual_token_mask_ratio": 0.2998602271080017, + "epoch": 3.9336630782659294, + "flow_matching_loss": 0.01378493132069707, + "grad_norm": 0.07157520949840546, + "kept_token_count": 572.903125, + "learning_rate": 1.86518196456923e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9932285487651825, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006771452538669109, + "mgd_loss_weight": 0.5495016515254975, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.36406860351565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8030517578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13520, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.3004458427429199, + "epoch": 3.936572592377073, + "flow_matching_loss": 0.013296845648437739, + "grad_norm": 0.09567684680223465, + "kept_token_count": 572.65, + "learning_rate": 1.8635780223429565e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9930145561695098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006985444668680429, + "mgd_loss_weight": 0.5491683065891266, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.1981231689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2460571289063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13530, + "valid_token_count": 818.5890625 + }, + { + "actual_token_mask_ratio": 0.29989193975925443, + "epoch": 3.9394821064882164, + "flow_matching_loss": 0.014256148971617222, + "grad_norm": 0.09025643020868301, + "kept_token_count": 572.93984375, + "learning_rate": 1.8619736383351017e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9929192304611206, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007080763019621373, + "mgd_loss_weight": 0.5488349616527557, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4690399169922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5387878417969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13540, + "valid_token_count": 818.35625 + }, + { + "actual_token_mask_ratio": 0.3004791349172592, + "epoch": 3.94239162059936, + "flow_matching_loss": 0.012781693506985903, + "grad_norm": 0.08445514738559723, + "kept_token_count": 572.34453125, + "learning_rate": 1.8603688144951454e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.993007504940033, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00699249655008316, + "mgd_loss_weight": 0.5485016167163849, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.3990905761719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.20862731933596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13550, + "valid_token_count": 818.19765625 + }, + { + "actual_token_mask_ratio": 0.29924718737602235, + "epoch": 3.9453011347105034, + "flow_matching_loss": 0.013415289670228958, + "grad_norm": 0.06984812766313553, + "kept_token_count": 573.57109375, + "learning_rate": 1.8587635527730983e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.993153190612793, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00684681748971343, + "mgd_loss_weight": 0.548168271780014, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.44899291992186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3465942382812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13560, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.2999746412038803, + "epoch": 3.9482106488216466, + "flow_matching_loss": 0.012977279908955097, + "grad_norm": 0.09403432905673981, + "kept_token_count": 572.7609375, + "learning_rate": 1.8571578551195056e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9933020949363709, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006697906507179141, + "mgd_loss_weight": 0.5478349328041077, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.7192077636719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6869262695312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13570, + "valid_token_count": 818.20234375 + }, + { + "actual_token_mask_ratio": 0.2992587834596634, + "epoch": 3.9511201629327903, + "flow_matching_loss": 0.01171500375494361, + "grad_norm": 0.07544045895338058, + "kept_token_count": 573.49140625, + "learning_rate": 1.855551723485441e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9930699467658997, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006930059241130948, + "mgd_loss_weight": 0.5475015819072724, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.0670166015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.14281311035154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13580, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.3003327906131744, + "epoch": 3.9540296770439336, + "flow_matching_loss": 0.012454320024698973, + "grad_norm": 0.12189741432666779, + "kept_token_count": 572.5140625, + "learning_rate": 1.8539451598225066e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9929939270019531, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007006079936400056, + "mgd_loss_weight": 0.5471682369709014, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.70374755859376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.90122375488284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13590, + "valid_token_count": 818.26328125 + }, + { + "actual_token_mask_ratio": 0.30060832798480985, + "epoch": 3.956939191155077, + "flow_matching_loss": 0.013522443268448115, + "grad_norm": 0.07167249172925949, + "kept_token_count": 572.38359375, + "learning_rate": 1.8523381660828276e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9927789151668549, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007221077149733901, + "mgd_loss_weight": 0.5468348979949951, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.1382232666016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.13457946777345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13600, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.3004745036363602, + "epoch": 3.9598487052662206, + "flow_matching_loss": 0.013050775695592164, + "grad_norm": 0.09397062659263611, + "kept_token_count": 572.35234375, + "learning_rate": 1.850730744219054e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9930831849575043, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006916806241497398, + "mgd_loss_weight": 0.5465015470981598, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.31953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8994934082031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13610, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.30014160871505735, + "epoch": 3.962758219377364, + "flow_matching_loss": 0.012329160142689944, + "grad_norm": 0.0816994458436966, + "kept_token_count": 572.80390625, + "learning_rate": 1.849122896184354e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9928397238254547, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007160262949764728, + "mgd_loss_weight": 0.5461682081222534, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.0441070556641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2403869628906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13620, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.2997151017189026, + "epoch": 3.9656677334885075, + "flow_matching_loss": 0.014026379957795144, + "grad_norm": 0.09945325553417206, + "kept_token_count": 573.0015625, + "learning_rate": 1.8475146239324146e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9930771648883819, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006922832503914833, + "mgd_loss_weight": 0.5458348631858826, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.87241821289064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.11585998535156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13630, + "valid_token_count": 818.24453125 + }, + { + "actual_token_mask_ratio": 0.29968827962875366, + "epoch": 3.968577247599651, + "flow_matching_loss": 0.011590918712317943, + "grad_norm": 0.0917133018374443, + "kept_token_count": 572.90703125, + "learning_rate": 1.8459059294174396e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9930863499641418, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069136468693614, + "mgd_loss_weight": 0.5455015122890472, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.947216796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.14385681152345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13640, + "valid_token_count": 818.07421875 + }, + { + "actual_token_mask_ratio": 0.2997032552957535, + "epoch": 3.9714867617107945, + "flow_matching_loss": 0.013407861720770598, + "grad_norm": 0.08201001584529877, + "kept_token_count": 572.98984375, + "learning_rate": 1.844296814594143e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9926479518413543, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0073520466685295105, + "mgd_loss_weight": 0.5451681792736054, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.98829650878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7830017089844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13650, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.3000932663679123, + "epoch": 3.9743962758219378, + "flow_matching_loss": 0.012928808480501175, + "grad_norm": 0.09124261885881424, + "kept_token_count": 572.77890625, + "learning_rate": 1.8426872814177516e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9931320130825043, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006867986544966698, + "mgd_loss_weight": 0.5448348224163055, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.76734619140626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7774719238281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13660, + "valid_token_count": 818.36015625 + }, + { + "actual_token_mask_ratio": 0.2999667853116989, + "epoch": 3.977305789933081, + "flow_matching_loss": 0.012596099358052015, + "grad_norm": 0.09691161662340164, + "kept_token_count": 572.765625, + "learning_rate": 1.8410773318440003e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.993083655834198, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006916346913203597, + "mgd_loss_weight": 0.5445014834403992, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.61614379882815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3761260986328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13670, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.29969724714756013, + "epoch": 3.9802153040442247, + "flow_matching_loss": 0.011908297799527645, + "grad_norm": 0.0950722023844719, + "kept_token_count": 573.17109375, + "learning_rate": 1.839466967829129e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9930085122585297, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006991493655368686, + "mgd_loss_weight": 0.5441681444644928, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.91436462402345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7947052001953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13680, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.2997446984052658, + "epoch": 3.983124818155368, + "flow_matching_loss": 0.012608662247657776, + "grad_norm": 0.07360988110303879, + "kept_token_count": 572.94140625, + "learning_rate": 1.8378561913298822e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9928513467311859, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007148652570322156, + "mgd_loss_weight": 0.543834787607193, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.9903869628906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.5623352050781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13690, + "valid_token_count": 818.18828125 + }, + { + "actual_token_mask_ratio": 0.3000980973243713, + "epoch": 3.9860343322665113, + "flow_matching_loss": 0.01137613020837307, + "grad_norm": 0.07328726351261139, + "kept_token_count": 572.6921875, + "learning_rate": 1.8362450043035053e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9927680194377899, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007231990667060017, + "mgd_loss_weight": 0.5435014545917511, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.54852600097655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4732727050781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13700, + "valid_token_count": 818.24765625 + }, + { + "actual_token_mask_ratio": 0.29990409314632416, + "epoch": 3.988943846377655, + "flow_matching_loss": 0.012874872423708439, + "grad_norm": 0.10787449032068253, + "kept_token_count": 572.63125, + "learning_rate": 1.834633408707742e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9928987085819244, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007101287040859461, + "mgd_loss_weight": 0.5431681036949157, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.8548248291016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.29216918945315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13710, + "valid_token_count": 817.93515625 + }, + { + "actual_token_mask_ratio": 0.3001679927110672, + "epoch": 3.9918533604887982, + "flow_matching_loss": 0.014470364339649678, + "grad_norm": 0.08467640727758408, + "kept_token_count": 572.628125, + "learning_rate": 1.8330214065008336e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9932331681251526, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006766839139163494, + "mgd_loss_weight": 0.5428347587585449, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.38315734863284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.5478546142578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13720, + "valid_token_count": 818.23515625 + }, + { + "actual_token_mask_ratio": 0.30053806602954863, + "epoch": 3.994762874599942, + "flow_matching_loss": 0.013759559486061335, + "grad_norm": 0.08471760153770447, + "kept_token_count": 572.39140625, + "learning_rate": 1.831408999641514e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9929202795028687, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007079731067642569, + "mgd_loss_weight": 0.5425014197826385, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.5212432861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.6685302734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13730, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.3001293659210205, + "epoch": 3.997672388711085, + "flow_matching_loss": 0.013382783439010382, + "grad_norm": 0.09442804008722305, + "kept_token_count": 572.64921875, + "learning_rate": 1.8297961900890107e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9930720210075379, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006927978573366999, + "mgd_loss_weight": 0.5421680688858033, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.6757110595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4796569824219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13740, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.29969740808010104, + "epoch": 4.000581902822229, + "flow_matching_loss": 0.01214041612111032, + "grad_norm": 0.06841745972633362, + "kept_token_count": 572.9377380371094, + "learning_rate": 1.8281829798030385e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.993622887134552, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006377116031944752, + "mgd_loss_weight": 0.5418347299098969, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.493408203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.3376770019531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13750, + "valid_token_count": 818.1338989257813 + }, + { + "actual_token_mask_ratio": 0.2997681349515915, + "epoch": 4.003491416933372, + "flow_matching_loss": 0.013420515134930611, + "grad_norm": 0.0887831449508667, + "kept_token_count": 572.97265625, + "learning_rate": 1.8265693707438002e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9928460776805877, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007153924647718668, + "mgd_loss_weight": 0.541501384973526, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 444.129736328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.4276947021484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13760, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.29962743520736695, + "epoch": 4.0064009310445154, + "flow_matching_loss": 0.012829962186515331, + "grad_norm": 0.08371182531118393, + "kept_token_count": 573.39609375, + "learning_rate": 1.824955364871982e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9929265856742859, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007073425035923719, + "mgd_loss_weight": 0.5411680340766907, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.3261688232422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8805358886719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13770, + "valid_token_count": 818.69921875 + }, + { + "actual_token_mask_ratio": 0.3001936346292496, + "epoch": 4.009310445155659, + "flow_matching_loss": 0.012425323016941547, + "grad_norm": 0.08211951702833176, + "kept_token_count": 572.51796875, + "learning_rate": 1.8233409641487553e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9930814445018769, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006918546929955483, + "mgd_loss_weight": 0.5408346951007843, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.8609558105469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8254913330078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13780, + "valid_token_count": 818.10859375 + }, + { + "actual_token_mask_ratio": 0.30022944808006286, + "epoch": 4.012219959266803, + "flow_matching_loss": 0.013975819759070873, + "grad_norm": 0.07974836230278015, + "kept_token_count": 572.66953125, + "learning_rate": 1.8217261705357677e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9926853120326996, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007314677257090807, + "mgd_loss_weight": 0.5405013501644135, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.57904663085935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9623077392578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13790, + "valid_token_count": 818.365625 + }, + { + "actual_token_mask_ratio": 0.30070329904556276, + "epoch": 4.015129473377946, + "flow_matching_loss": 0.01252801800146699, + "grad_norm": 0.09019407629966736, + "kept_token_count": 572.3265625, + "learning_rate": 1.8201109859951462e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9929136216640473, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007086389977484941, + "mgd_loss_weight": 0.5401680052280426, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.8609130859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6552429199219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13800, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.2993964284658432, + "epoch": 4.018038987489089, + "flow_matching_loss": 0.011616413155570627, + "grad_norm": 0.08449162542819977, + "kept_token_count": 573.40546875, + "learning_rate": 1.818495412489492e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9933380603790283, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006661935243755579, + "mgd_loss_weight": 0.5398346602916717, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.47222900390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.6878387451172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13810, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.3007306307554245, + "epoch": 4.020948501600233, + "flow_matching_loss": 0.014418786484748125, + "grad_norm": 0.07624999433755875, + "kept_token_count": 572.2515625, + "learning_rate": 1.8168794519818802e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9931001484394073, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006899844342842698, + "mgd_loss_weight": 0.5395013153553009, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.9688262939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.48984680175784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13820, + "valid_token_count": 818.36015625 + }, + { + "actual_token_mask_ratio": 0.2998745977878571, + "epoch": 4.023858015711376, + "flow_matching_loss": 0.013669950887560844, + "grad_norm": 0.08258149772882462, + "kept_token_count": 572.8796875, + "learning_rate": 1.8152631064358537e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9932060778141022, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006793929263949394, + "mgd_loss_weight": 0.53916797041893, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.5580596923828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1705596923828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13830, + "valid_token_count": 818.24453125 + }, + { + "actual_token_mask_ratio": 0.30062257647514345, + "epoch": 4.02676752982252, + "flow_matching_loss": 0.013542583398520946, + "grad_norm": 0.08320450782775879, + "kept_token_count": 572.3421875, + "learning_rate": 1.8136463778154258e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9929510116577148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007049002824351192, + "mgd_loss_weight": 0.5388346314430237, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.7842254638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6006317138672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13840, + "valid_token_count": 818.36015625 + }, + { + "actual_token_mask_ratio": 0.2999870181083679, + "epoch": 4.029677043933663, + "flow_matching_loss": 0.013602028135210276, + "grad_norm": 0.09018530696630478, + "kept_token_count": 572.80234375, + "learning_rate": 1.812029268085074e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9927008628845215, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072991409804672005, + "mgd_loss_weight": 0.5385012805461884, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.21042785644534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6127990722656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13850, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.3000682592391968, + "epoch": 4.032586558044806, + "flow_matching_loss": 0.01458757957443595, + "grad_norm": 0.09914688020944595, + "kept_token_count": 572.68828125, + "learning_rate": 1.8104117792097387e-05, + "loss": 0.0187, + "mgd_cos_sim": 0.9924413383007049, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007558664167299866, + "mgd_loss_weight": 0.5381679356098175, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.13190307617185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.69996032714846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13860, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.3003455072641373, + "epoch": 4.03549607215595, + "flow_matching_loss": 0.014324623625725508, + "grad_norm": 0.09712331742048264, + "kept_token_count": 572.36953125, + "learning_rate": 1.808793913154822e-05, + "loss": 0.0182, + "mgd_cos_sim": 0.9927778959274292, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007222097320482135, + "mgd_loss_weight": 0.5378345966339111, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.17173156738284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6491180419922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13870, + "valid_token_count": 818.07421875 + }, + { + "actual_token_mask_ratio": 0.29985481798648833, + "epoch": 4.038405586267094, + "flow_matching_loss": 0.013490796368569136, + "grad_norm": 0.08844935894012451, + "kept_token_count": 572.89921875, + "learning_rate": 1.8071756718861838e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9926724851131439, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007327513350173831, + "mgd_loss_weight": 0.5375012457370758, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.2578369140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3940002441406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13880, + "valid_token_count": 818.25390625 + }, + { + "actual_token_mask_ratio": 0.2998215168714523, + "epoch": 4.041315100378237, + "flow_matching_loss": 0.012829440739005804, + "grad_norm": 0.09103775769472122, + "kept_token_count": 572.8203125, + "learning_rate": 1.805557057370139e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9928750634193421, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071249368600547315, + "mgd_loss_weight": 0.5371679067611694, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.1765991210938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.63218383789064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13890, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.2998647004365921, + "epoch": 4.04422461448938, + "flow_matching_loss": 0.013768151868134737, + "grad_norm": 0.07029183954000473, + "kept_token_count": 572.89375, + "learning_rate": 1.8039380715734574e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9933213233947754, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006678680749610066, + "mgd_loss_weight": 0.5368345618247986, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.7609130859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.6186981201172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13900, + "valid_token_count": 818.26328125 + }, + { + "actual_token_mask_ratio": 0.30067841708660126, + "epoch": 4.047134128600524, + "flow_matching_loss": 0.012831824086606502, + "grad_norm": 0.08123976737260818, + "kept_token_count": 572.25390625, + "learning_rate": 1.802318716463359e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9927580893039704, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007241910835728049, + "mgd_loss_weight": 0.5365012109279632, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.1410766601563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.46197204589845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13910, + "valid_token_count": 818.29765625 + }, + { + "actual_token_mask_ratio": 0.30057633221149443, + "epoch": 4.0500436427116675, + "flow_matching_loss": 0.013581863231956958, + "grad_norm": 0.074997678399086, + "kept_token_count": 572.3171875, + "learning_rate": 1.8006989940075138e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9924875140190125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0075124924071133135, + "mgd_loss_weight": 0.5361678779125214, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.65631103515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.65919799804686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13920, + "valid_token_count": 818.26796875 + }, + { + "actual_token_mask_ratio": 0.3000499427318573, + "epoch": 4.05295315682281, + "flow_matching_loss": 0.012233192194253206, + "grad_norm": 0.08594168722629547, + "kept_token_count": 572.6734375, + "learning_rate": 1.7990789061740368e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.992443197965622, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007556814793497324, + "mgd_loss_weight": 0.535834527015686, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.6688629150391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.81093139648436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13930, + "valid_token_count": 818.16640625 + }, + { + "actual_token_mask_ratio": 0.3001074194908142, + "epoch": 4.055862670933954, + "flow_matching_loss": 0.013239951897412538, + "grad_norm": 0.09004746377468109, + "kept_token_count": 572.65703125, + "learning_rate": 1.7974584549314875e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9928347945213318, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007165194163098932, + "mgd_loss_weight": 0.5355011820793152, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.0667419433594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6503967285156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13940, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.30014869570732117, + "epoch": 4.058772185045098, + "flow_matching_loss": 0.014750509336590768, + "grad_norm": 0.11863074451684952, + "kept_token_count": 572.54453125, + "learning_rate": 1.7958376422488672e-05, + "loss": 0.0185, + "mgd_cos_sim": 0.9929156064987182, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007084386283531785, + "mgd_loss_weight": 0.5351678431034088, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.60308532714845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0553863525391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13950, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.29919918179512023, + "epoch": 4.0616816991562406, + "flow_matching_loss": 0.014786481391638517, + "grad_norm": 0.07377006858587265, + "kept_token_count": 573.3703125, + "learning_rate": 1.7942164700956156e-05, + "loss": 0.0186, + "mgd_cos_sim": 0.9928701281547546, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007129868073388934, + "mgd_loss_weight": 0.5348344922065735, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.6639801025391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1710174560547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13960, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.29964178800582886, + "epoch": 4.064591213267384, + "flow_matching_loss": 0.0125875661149621, + "grad_norm": 0.07670895755290985, + "kept_token_count": 572.98046875, + "learning_rate": 1.79259494044161e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9930357277393341, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006964277755469084, + "mgd_loss_weight": 0.5345011532306672, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.64177856445315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.41990356445314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13970, + "valid_token_count": 818.12890625 + }, + { + "actual_token_mask_ratio": 0.29946832954883573, + "epoch": 4.067500727378528, + "flow_matching_loss": 0.01325934063643217, + "grad_norm": 0.08279887586832047, + "kept_token_count": 573.19921875, + "learning_rate": 1.7909730552571617e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9931887865066529, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006811213446781039, + "mgd_loss_weight": 0.5341678082942962, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.65785827636716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1542114257812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13980, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.3001375049352646, + "epoch": 4.070410241489671, + "flow_matching_loss": 0.01348821297287941, + "grad_norm": 0.09173640608787537, + "kept_token_count": 572.75546875, + "learning_rate": 1.7893508165130143e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9926252365112305, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007374774990603327, + "mgd_loss_weight": 0.5338344573974609, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.48052978515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.86846923828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 13990, + "valid_token_count": 818.37734375 + }, + { + "actual_token_mask_ratio": 0.299175962805748, + "epoch": 4.0733197556008145, + "flow_matching_loss": 0.013509767223149538, + "grad_norm": 0.09290444105863571, + "kept_token_count": 573.32890625, + "learning_rate": 1.787728226180341e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9931062400341034, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006893756007775665, + "mgd_loss_weight": 0.5335011184215546, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.6106689453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.50583801269534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14000, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.30013281404972075, + "epoch": 4.076229269711958, + "flow_matching_loss": 0.013164425548166037, + "grad_norm": 0.08959043771028519, + "kept_token_count": 572.60546875, + "learning_rate": 1.786105286230742e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9931476056575775, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006852404726669192, + "mgd_loss_weight": 0.5331677734851837, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.4767761230469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5693664550781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14010, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.2992940485477448, + "epoch": 4.079138783823102, + "flow_matching_loss": 0.013294093776494265, + "grad_norm": 0.09854555130004883, + "kept_token_count": 573.38984375, + "learning_rate": 1.784481998636242e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9931367039680481, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006863292632624507, + "mgd_loss_weight": 0.5328344285488129, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.8597839355469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0957305908203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14020, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.29994866251945496, + "epoch": 4.082048297934245, + "flow_matching_loss": 0.014854592271149158, + "grad_norm": 0.08899784833192825, + "kept_token_count": 572.90859375, + "learning_rate": 1.782858365369289e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9932956576347352, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006704323645681143, + "mgd_loss_weight": 0.532501083612442, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 400.6966888427734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0410095214844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14030, + "valid_token_count": 818.38046875 + }, + { + "actual_token_mask_ratio": 0.2991182118654251, + "epoch": 4.084957812045388, + "flow_matching_loss": 0.013031400553882123, + "grad_norm": 0.08698966354131699, + "kept_token_count": 573.4625, + "learning_rate": 1.78123438840275e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9928120911121369, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007187907723709941, + "mgd_loss_weight": 0.5321677386760711, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.56702575683596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.14194946289064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14040, + "valid_token_count": 818.20078125 + }, + { + "actual_token_mask_ratio": 0.30002483129501345, + "epoch": 4.087867326156532, + "flow_matching_loss": 0.01432166900485754, + "grad_norm": 0.07867399603128433, + "kept_token_count": 572.68828125, + "learning_rate": 1.779610069709911e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9930449485778808, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006955052120611071, + "mgd_loss_weight": 0.5318343937397003, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.40919189453126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7403228759766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14050, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.29999602735042574, + "epoch": 4.090776840267675, + "flow_matching_loss": 0.014143274445086718, + "grad_norm": 0.073458231985569, + "kept_token_count": 572.7984375, + "learning_rate": 1.7779854112644713e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9929728925228118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007027107989415526, + "mgd_loss_weight": 0.531501054763794, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.1800231933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0844421386719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14060, + "valid_token_count": 818.27578125 + }, + { + "actual_token_mask_ratio": 0.3007743626832962, + "epoch": 4.093686354378819, + "flow_matching_loss": 0.013795361388474703, + "grad_norm": 0.08380843698978424, + "kept_token_count": 572.2203125, + "learning_rate": 1.7763604150405446e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9931933403015136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00680665373802185, + "mgd_loss_weight": 0.5311677038669587, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.43299255371096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8191253662109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14070, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.30049056112766265, + "epoch": 4.096595868489962, + "flow_matching_loss": 0.01375432349741459, + "grad_norm": 0.08756070584058762, + "kept_token_count": 572.49453125, + "learning_rate": 1.7747350830126548e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9926398575305939, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007360150245949626, + "mgd_loss_weight": 0.5308343589305877, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.89829711914064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.31443786621094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14080, + "valid_token_count": 818.42265625 + }, + { + "actual_token_mask_ratio": 0.30049343705177306, + "epoch": 4.099505382601105, + "flow_matching_loss": 0.013782591372728348, + "grad_norm": 0.07457011193037033, + "kept_token_count": 572.49375, + "learning_rate": 1.7731094171557332e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9930469632148743, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006953046284615993, + "mgd_loss_weight": 0.5305010199546814, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.3463439941406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.95180053710936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14090, + "valid_token_count": 818.42265625 + }, + { + "actual_token_mask_ratio": 0.30069148540496826, + "epoch": 4.102414896712249, + "flow_matching_loss": 0.012498394399881364, + "grad_norm": 0.07876817137002945, + "kept_token_count": 572.009375, + "learning_rate": 1.771483419445117e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9932553827762604, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006744626211002469, + "mgd_loss_weight": 0.5301676690578461, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4547515869141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.212646484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14100, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.3007997334003448, + "epoch": 4.105324410823393, + "flow_matching_loss": 0.011035160161554813, + "grad_norm": 0.06892801076173782, + "kept_token_count": 572.21796875, + "learning_rate": 1.7698570918565474e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9929871797561646, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007012827368453145, + "mgd_loss_weight": 0.5298343300819397, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.13926696777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3128967285156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14110, + "valid_token_count": 818.38515625 + }, + { + "actual_token_mask_ratio": 0.29974437952041627, + "epoch": 4.108233924934536, + "flow_matching_loss": 0.011216573137789964, + "grad_norm": 0.07413718104362488, + "kept_token_count": 572.99921875, + "learning_rate": 1.768230436366165e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9933683514595032, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006631656643003225, + "mgd_loss_weight": 0.5295009851455689, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.5051635742187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.44603576660154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14120, + "valid_token_count": 818.27265625 + }, + { + "actual_token_mask_ratio": 0.3001589149236679, + "epoch": 4.111143439045679, + "flow_matching_loss": 0.014556896314024926, + "grad_norm": 0.06556213647127151, + "kept_token_count": 572.64921875, + "learning_rate": 1.7666034549505102e-05, + "loss": 0.0184, + "mgd_cos_sim": 0.9926701784133911, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007329829363152384, + "mgd_loss_weight": 0.5291676342487335, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.4276428222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.55531005859376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14130, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.29984385669231417, + "epoch": 4.114052953156823, + "flow_matching_loss": 0.01365409716963768, + "grad_norm": 0.0792730301618576, + "kept_token_count": 572.91953125, + "learning_rate": 1.7649761495865185e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9933044135570526, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006695599760860205, + "mgd_loss_weight": 0.5288343012332917, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.87095031738284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.4438781738281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14140, + "valid_token_count": 818.27421875 + }, + { + "actual_token_mask_ratio": 0.2994080066680908, + "epoch": 4.1169624672679666, + "flow_matching_loss": 0.01313690422102809, + "grad_norm": 0.10326573252677917, + "kept_token_count": 573.26640625, + "learning_rate": 1.7633485222515188e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9929926574230195, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007007340714335441, + "mgd_loss_weight": 0.5285009443759918, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.6467559814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.42176208496096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14150, + "valid_token_count": 818.25859375 + }, + { + "actual_token_mask_ratio": 0.2991624891757965, + "epoch": 4.119871981379109, + "flow_matching_loss": 0.01378747159615159, + "grad_norm": 0.07352311164140701, + "kept_token_count": 573.39296875, + "learning_rate": 1.761720574923233e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9930030882358551, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069969234522432085, + "mgd_loss_weight": 0.5281676054000854, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.4527221679688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5788177490234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14160, + "valid_token_count": 818.15703125 + }, + { + "actual_token_mask_ratio": 0.2997377455234528, + "epoch": 4.122781495490253, + "flow_matching_loss": 0.01357361301779747, + "grad_norm": 0.06784863770008087, + "kept_token_count": 572.94765625, + "learning_rate": 1.7600923095797704e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9924054086208344, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007594592962414026, + "mgd_loss_weight": 0.5278342664241791, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.7705352783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.2821807861328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14170, + "valid_token_count": 818.18984375 + }, + { + "actual_token_mask_ratio": 0.29996476471424105, + "epoch": 4.125691009601397, + "flow_matching_loss": 0.01251072222366929, + "grad_norm": 0.09461230039596558, + "kept_token_count": 572.80390625, + "learning_rate": 1.758463728199626e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9929448544979096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007055147038772702, + "mgd_loss_weight": 0.5275009095668792, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.99523010253904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.08733520507815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14180, + "valid_token_count": 818.25078125 + }, + { + "actual_token_mask_ratio": 0.29924381971359254, + "epoch": 4.12860052371254, + "flow_matching_loss": 0.011978998035192489, + "grad_norm": 0.09849312156438828, + "kept_token_count": 573.31796875, + "learning_rate": 1.7568348327616803e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9929281771183014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070718100760132074, + "mgd_loss_weight": 0.5271675765514374, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.079736328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.07310791015624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14190, + "valid_token_count": 818.14296875 + }, + { + "actual_token_mask_ratio": 0.3001069962978363, + "epoch": 4.131510037823683, + "flow_matching_loss": 0.013035558629781008, + "grad_norm": 0.08500038832426071, + "kept_token_count": 572.7328125, + "learning_rate": 1.755205625245195e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9932310879230499, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006768915662541985, + "mgd_loss_weight": 0.526834225654602, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.9907196044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.7644470214844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14200, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.29969362914562225, + "epoch": 4.134419551934827, + "flow_matching_loss": 0.01174226813018322, + "grad_norm": 0.07709460705518723, + "kept_token_count": 573.10234375, + "learning_rate": 1.7535761076298104e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9929808676242828, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007019129116088152, + "mgd_loss_weight": 0.5265008807182312, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.2134613037109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.60382690429685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14210, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.2995947927236557, + "epoch": 4.137329066045971, + "flow_matching_loss": 0.01160811367444694, + "grad_norm": 0.0780733972787857, + "kept_token_count": 573.221875, + "learning_rate": 1.7519462818955448e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9931646168231965, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00683538201265037, + "mgd_loss_weight": 0.5261675417423248, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.7393341064453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5997650146484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14220, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.3002795696258545, + "epoch": 4.1402385801571135, + "flow_matching_loss": 0.012150706443935632, + "grad_norm": 0.07334549725055695, + "kept_token_count": 572.66015625, + "learning_rate": 1.7503161500227896e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9931008219718933, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006899191904813052, + "mgd_loss_weight": 0.5258341908454895, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.7894287109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6098876953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14230, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.2995279490947723, + "epoch": 4.143148094268257, + "flow_matching_loss": 0.012385637778788806, + "grad_norm": 0.08122090250253677, + "kept_token_count": 573.06640625, + "learning_rate": 1.7486857139923093e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9929073989391327, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007092606136575341, + "mgd_loss_weight": 0.5255008518695832, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.4480316162109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.80552673339844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14240, + "valid_token_count": 818.11171875 + }, + { + "actual_token_mask_ratio": 0.29945998191833495, + "epoch": 4.146057608379401, + "flow_matching_loss": 0.012747703585773706, + "grad_norm": 0.09138250350952148, + "kept_token_count": 573.3625, + "learning_rate": 1.747054975785236e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9922797799110412, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007720220368355513, + "mgd_loss_weight": 0.5251675069332122, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4092956542969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.3579437255859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14250, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.3005937188863754, + "epoch": 4.148967122490544, + "flow_matching_loss": 0.013248185627162457, + "grad_norm": 0.08267106860876083, + "kept_token_count": 572.32734375, + "learning_rate": 1.745423937383072e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9930711627006531, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069288301281630995, + "mgd_loss_weight": 0.524834156036377, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.0124450683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7182861328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14260, + "valid_token_count": 818.30078125 + }, + { + "actual_token_mask_ratio": 0.299258092045784, + "epoch": 4.1518766366016875, + "flow_matching_loss": 0.011485593300312758, + "grad_norm": 0.08178642392158508, + "kept_token_count": 573.44453125, + "learning_rate": 1.7437926007676825e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9934217751026153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00657821362838149, + "mgd_loss_weight": 0.5245008170604706, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 451.99683837890626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.65228576660155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14270, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.3004780501127243, + "epoch": 4.154786150712831, + "flow_matching_loss": 0.0128958560526371, + "grad_norm": 0.07176897674798965, + "kept_token_count": 572.38203125, + "learning_rate": 1.742160967921294e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9929368257522583, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007063168566673994, + "mgd_loss_weight": 0.5241674721240998, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.4555267333984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.39910583496095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14280, + "valid_token_count": 818.24921875 + }, + { + "actual_token_mask_ratio": 0.30028334856033323, + "epoch": 4.157695664823974, + "flow_matching_loss": 0.01401794645935297, + "grad_norm": 0.08925404399633408, + "kept_token_count": 572.61484375, + "learning_rate": 1.7405290408264954e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9929910838603974, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007008915627375245, + "mgd_loss_weight": 0.5238341271877289, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.6451721191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.14914855957034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14290, + "valid_token_count": 818.34921875 + }, + { + "actual_token_mask_ratio": 0.2998488485813141, + "epoch": 4.160605178935118, + "flow_matching_loss": 0.012762543093413115, + "grad_norm": 0.07387078553438187, + "kept_token_count": 572.95703125, + "learning_rate": 1.738896821466232e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.992959725856781, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007040277076885104, + "mgd_loss_weight": 0.523500782251358, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.78456420898436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7860046386719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14300, + "valid_token_count": 818.33515625 + }, + { + "actual_token_mask_ratio": 0.3003620535135269, + "epoch": 4.163514693046261, + "flow_matching_loss": 0.011974289268255233, + "grad_norm": 0.07661724835634232, + "kept_token_count": 572.6625, + "learning_rate": 1.737264311823804e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9930603384971619, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006939655914902687, + "mgd_loss_weight": 0.5231674373149872, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.7833953857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7445404052734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14310, + "valid_token_count": 818.5109375 + }, + { + "actual_token_mask_ratio": 0.30080288648605347, + "epoch": 4.166424207157405, + "flow_matching_loss": 0.012811396364122629, + "grad_norm": 0.07094120234251022, + "kept_token_count": 572.165625, + "learning_rate": 1.7356315138828645e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9927714884281158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007228508498519659, + "mgd_loss_weight": 0.5228340923786163, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.38901672363284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.01015319824216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14320, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.30050948858261106, + "epoch": 4.169333721268548, + "flow_matching_loss": 0.012777499202638865, + "grad_norm": 0.0766824409365654, + "kept_token_count": 572.178125, + "learning_rate": 1.7339984296274174e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9930716574192047, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006928326329216361, + "mgd_loss_weight": 0.52250075340271, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.5521270751953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0673492431641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14330, + "valid_token_count": 817.99375 + }, + { + "actual_token_mask_ratio": 0.30053054094314574, + "epoch": 4.172243235379692, + "flow_matching_loss": 0.012572300899773837, + "grad_norm": 0.08319103717803955, + "kept_token_count": 572.328125, + "learning_rate": 1.7323650610418134e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9934479534626007, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006552049890160561, + "mgd_loss_weight": 0.5221674025058747, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.13727416992185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.44887390136716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14340, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.3007813334465027, + "epoch": 4.175152749490835, + "flow_matching_loss": 0.012299367599189282, + "grad_norm": 0.06745085120201111, + "kept_token_count": 572.1078125, + "learning_rate": 1.730731410110749e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9931579053401947, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00684208027087152, + "mgd_loss_weight": 0.5218340575695037, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.9077606201172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.09490356445315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14350, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.29901416301727296, + "epoch": 4.178062263601978, + "flow_matching_loss": 0.012939773686230182, + "grad_norm": 0.08256794512271881, + "kept_token_count": 573.7296875, + "learning_rate": 1.7290974788192655e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9929303765296936, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007069628406316042, + "mgd_loss_weight": 0.5215007245540619, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.3169738769531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.97364807128906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14360, + "valid_token_count": 818.4640625 + }, + { + "actual_token_mask_ratio": 0.2999838888645172, + "epoch": 4.180971777713122, + "flow_matching_loss": 0.013079569861292838, + "grad_norm": 0.07989223301410675, + "kept_token_count": 572.95703125, + "learning_rate": 1.727463269152743e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9929645597934723, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007035453571006656, + "mgd_loss_weight": 0.5211673676967621, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.990673828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.17221984863284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14370, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.2996583044528961, + "epoch": 4.183881291824266, + "flow_matching_loss": 0.013627079874277115, + "grad_norm": 0.07666434347629547, + "kept_token_count": 572.9796875, + "learning_rate": 1.7258287830968996e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9928602516651154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007139733526855707, + "mgd_loss_weight": 0.5208340287208557, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.8192840576172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.04010009765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14380, + "valid_token_count": 818.14140625 + }, + { + "actual_token_mask_ratio": 0.3000903069972992, + "epoch": 4.186790805935408, + "flow_matching_loss": 0.013286911975592374, + "grad_norm": 0.06551344692707062, + "kept_token_count": 572.85, + "learning_rate": 1.7241940226377908e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9929945170879364, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007005474530160427, + "mgd_loss_weight": 0.5205006897449493, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.0860900878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2314910888672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14390, + "valid_token_count": 818.46640625 + }, + { + "actual_token_mask_ratio": 0.30013485848903654, + "epoch": 4.189700320046552, + "flow_matching_loss": 0.013031222764402628, + "grad_norm": 0.08024027943611145, + "kept_token_count": 572.61171875, + "learning_rate": 1.7225589897618043e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9928804695606231, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007119532208889723, + "mgd_loss_weight": 0.5201673328876495, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.4565948486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.385400390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14400, + "valid_token_count": 818.17265625 + }, + { + "actual_token_mask_ratio": 0.2998274862766266, + "epoch": 4.192609834157696, + "flow_matching_loss": 0.013200327940285205, + "grad_norm": 0.07145766913890839, + "kept_token_count": 572.9203125, + "learning_rate": 1.7209236864556588e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9932684957981109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006731501268222928, + "mgd_loss_weight": 0.5198339998722077, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.52236022949216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1421752929688, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14410, + "valid_token_count": 818.26015625 + }, + { + "actual_token_mask_ratio": 0.29954011738300323, + "epoch": 4.195519348268839, + "flow_matching_loss": 0.01263768393546343, + "grad_norm": 0.07675053924322128, + "kept_token_count": 573.2234375, + "learning_rate": 1.719288114706403e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9929855585098266, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007014440326020122, + "mgd_loss_weight": 0.5195006489753723, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.76527099609376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.04017333984376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14420, + "valid_token_count": 818.35390625 + }, + { + "actual_token_mask_ratio": 0.300380265712738, + "epoch": 4.198428862379982, + "flow_matching_loss": 0.01374188307672739, + "grad_norm": 0.09625092893838882, + "kept_token_count": 572.44609375, + "learning_rate": 1.7176522765014104e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9931881189346313, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006811887212097645, + "mgd_loss_weight": 0.5191673040390015, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 446.2824279785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.08074645996095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14430, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.2994574010372162, + "epoch": 4.201338376491126, + "flow_matching_loss": 0.014410076569765807, + "grad_norm": 0.08334723860025406, + "kept_token_count": 573.409375, + "learning_rate": 1.7160161738283782e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9931412696838379, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068587358109653, + "mgd_loss_weight": 0.5188339650630951, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.2813354492188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.9542755126953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14440, + "valid_token_count": 818.5234375 + }, + { + "actual_token_mask_ratio": 0.3001926392316818, + "epoch": 4.20424789060227, + "flow_matching_loss": 0.012984929978847504, + "grad_norm": 0.06510523706674576, + "kept_token_count": 572.65234375, + "learning_rate": 1.714379808675326e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9925581276416778, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007441850239410996, + "mgd_loss_weight": 0.5185006141662598, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.51446228027345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.80731811523435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14450, + "valid_token_count": 818.30078125 + }, + { + "actual_token_mask_ratio": 0.3003547877073288, + "epoch": 4.207157404713413, + "flow_matching_loss": 0.014397131837904453, + "grad_norm": 0.0748356357216835, + "kept_token_count": 572.5140625, + "learning_rate": 1.712743183030592e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9933684527873993, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006631564209237694, + "mgd_loss_weight": 0.5181672751903534, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.99049987792966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.33746337890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14460, + "valid_token_count": 818.28984375 + }, + { + "actual_token_mask_ratio": 0.29990609884262087, + "epoch": 4.210066918824556, + "flow_matching_loss": 0.013032055366784335, + "grad_norm": 0.08843587338924408, + "kept_token_count": 572.8453125, + "learning_rate": 1.7111062988828294e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9929681241512298, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00703187077306211, + "mgd_loss_weight": 0.5178339302539825, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.32716979980466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.260546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14470, + "valid_token_count": 818.24453125 + }, + { + "actual_token_mask_ratio": 0.30027718245983126, + "epoch": 4.2129764329357, + "flow_matching_loss": 0.013897678349167109, + "grad_norm": 0.07256796211004257, + "kept_token_count": 572.58359375, + "learning_rate": 1.709469158221008e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9931606590747833, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006839337898418308, + "mgd_loss_weight": 0.5175005793571472, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.2077606201172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5292266845703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14480, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.30003770291805265, + "epoch": 4.215885947046843, + "flow_matching_loss": 0.013921290356665849, + "grad_norm": 0.08721663057804108, + "kept_token_count": 572.646875, + "learning_rate": 1.7078317630344083e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9929230272769928, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007076980452984572, + "mgd_loss_weight": 0.5171672403812408, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.6731231689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.38262634277345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14490, + "valid_token_count": 818.11484375 + }, + { + "actual_token_mask_ratio": 0.30020090341567995, + "epoch": 4.2187954611579865, + "flow_matching_loss": 0.012462233286350965, + "grad_norm": 0.0719216912984848, + "kept_token_count": 572.4265625, + "learning_rate": 1.706194115312618e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9933326423168183, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006667347997426987, + "mgd_loss_weight": 0.51683389544487, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.7546691894531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.53619079589845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14500, + "valid_token_count": 817.98515625 + }, + { + "actual_token_mask_ratio": 0.2998176008462906, + "epoch": 4.22170497526913, + "flow_matching_loss": 0.012103119539096952, + "grad_norm": 0.07642707973718643, + "kept_token_count": 572.87890625, + "learning_rate": 1.704556217045536e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.993038535118103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069614676758646965, + "mgd_loss_weight": 0.5165005505084992, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.1868408203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.05983276367186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14510, + "valid_token_count": 818.18203125 + }, + { + "actual_token_mask_ratio": 0.2993746966123581, + "epoch": 4.224614489380274, + "flow_matching_loss": 0.013474936131387949, + "grad_norm": 0.09893157333135605, + "kept_token_count": 573.190625, + "learning_rate": 1.7029180702233612e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9926165819168091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007383412634953857, + "mgd_loss_weight": 0.5161672055721283, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.1161346435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.21507873535154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14520, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.29952124655246737, + "epoch": 4.227524003491417, + "flow_matching_loss": 0.013295622635632753, + "grad_norm": 0.09820456802845001, + "kept_token_count": 573.096875, + "learning_rate": 1.7012796768365975e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.993057119846344, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006942888768389821, + "mgd_loss_weight": 0.5158338606357574, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.71412048339846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8857818603516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14530, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.30004933178424836, + "epoch": 4.2304335176025605, + "flow_matching_loss": 0.012813306413590908, + "grad_norm": 0.08883559703826904, + "kept_token_count": 572.74453125, + "learning_rate": 1.6996410388760465e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9929738402366638, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007026163348928094, + "mgd_loss_weight": 0.5155005156993866, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.2823211669922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0137023925781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14540, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.3001229792833328, + "epoch": 4.233343031713704, + "flow_matching_loss": 0.012774431332945823, + "grad_norm": 0.10992547869682312, + "kept_token_count": 572.66484375, + "learning_rate": 1.698002158332809e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.99350825548172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006491744192317128, + "mgd_loss_weight": 0.5151671767234802, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.27435302734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.38303833007814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14550, + "valid_token_count": 818.23515625 + }, + { + "actual_token_mask_ratio": 0.2996074289083481, + "epoch": 4.236252545824847, + "flow_matching_loss": 0.013415213953703643, + "grad_norm": 0.09755221754312515, + "kept_token_count": 572.996875, + "learning_rate": 1.6963630371982783e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9926497280597687, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00735027170740068, + "mgd_loss_weight": 0.5148338258266449, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.0266967773438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2750518798828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14560, + "valid_token_count": 818.10390625 + }, + { + "actual_token_mask_ratio": 0.2999624997377396, + "epoch": 4.239162059935991, + "flow_matching_loss": 0.013294212613254785, + "grad_norm": 0.08329858630895615, + "kept_token_count": 572.90859375, + "learning_rate": 1.6947236774641416e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9930895209312439, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006910489406436682, + "mgd_loss_weight": 0.514500480890274, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.5298614501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.84429321289065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14570, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.3005940169095993, + "epoch": 4.242071574047134, + "flow_matching_loss": 0.014137656334787608, + "grad_norm": 0.09042302519083023, + "kept_token_count": 572.18125, + "learning_rate": 1.6930840811223755e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9931038618087769, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006896148901432753, + "mgd_loss_weight": 0.5141671419143676, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.16599731445314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9159271240234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14580, + "valid_token_count": 818.10234375 + }, + { + "actual_token_mask_ratio": 0.2990807116031647, + "epoch": 4.244981088158277, + "flow_matching_loss": 0.013133939355611801, + "grad_norm": 0.0775349885225296, + "kept_token_count": 573.46875, + "learning_rate": 1.6914442501652448e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9930263340473175, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069736478384584185, + "mgd_loss_weight": 0.5138337910175323, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.9321014404297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1255187988281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14590, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.2996480494737625, + "epoch": 4.247890602269421, + "flow_matching_loss": 0.012472143024206161, + "grad_norm": 0.0802975669503212, + "kept_token_count": 573.1734375, + "learning_rate": 1.6898041865852974e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9932045876979828, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006795421056449413, + "mgd_loss_weight": 0.513500452041626, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.17893371582034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9343780517578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14600, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.29961816072463987, + "epoch": 4.250800116380565, + "flow_matching_loss": 0.01210037199780345, + "grad_norm": 0.07359752804040909, + "kept_token_count": 573.15234375, + "learning_rate": 1.6881638923753663e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9933578848838807, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006642119167372585, + "mgd_loss_weight": 0.5131671071052551, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.9302276611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.75255432128904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14610, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.3000738322734833, + "epoch": 4.2537096304917075, + "flow_matching_loss": 0.01279488867148757, + "grad_norm": 0.07863727957010269, + "kept_token_count": 572.903125, + "learning_rate": 1.6865233695285632e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9928323984146118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007167608058080077, + "mgd_loss_weight": 0.5128337562084198, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.97288208007814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.451123046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14620, + "valid_token_count": 818.52265625 + }, + { + "actual_token_mask_ratio": 0.2994861364364624, + "epoch": 4.256619144602851, + "flow_matching_loss": 0.01299857757985592, + "grad_norm": 0.07542131096124649, + "kept_token_count": 573.12265625, + "learning_rate": 1.6848826200382782e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9932228982448578, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006777091789990664, + "mgd_loss_weight": 0.512500423192978, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.5976135253906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.34531860351564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14630, + "valid_token_count": 818.14921875 + }, + { + "actual_token_mask_ratio": 0.30050398111343385, + "epoch": 4.259528658713995, + "flow_matching_loss": 0.012190768867731095, + "grad_norm": 0.08040349185466766, + "kept_token_count": 572.41015625, + "learning_rate": 1.6832416458981762e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9930066704750061, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006993330223485828, + "mgd_loss_weight": 0.5121670663356781, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.2915283203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6310577392578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14640, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.29996545016765597, + "epoch": 4.262438172825139, + "flow_matching_loss": 0.01312020355835557, + "grad_norm": 0.08065086603164673, + "kept_token_count": 572.71875, + "learning_rate": 1.681600449102196e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9928660809993743, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007133921468630433, + "mgd_loss_weight": 0.5118337273597717, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.7101593017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.37163696289065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14650, + "valid_token_count": 818.12421875 + }, + { + "actual_token_mask_ratio": 0.3008996665477753, + "epoch": 4.265347686936281, + "flow_matching_loss": 0.012562898825854063, + "grad_norm": 0.08812884986400604, + "kept_token_count": 572.16796875, + "learning_rate": 1.6799590316445468e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9931106805801392, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006889312295243144, + "mgd_loss_weight": 0.5115003883838654, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.07158203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.97104797363284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14660, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.30002752542495725, + "epoch": 4.268257201047425, + "flow_matching_loss": 0.011952795460820198, + "grad_norm": 0.08768542855978012, + "kept_token_count": 572.86015625, + "learning_rate": 1.678317395519705e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9931947946548462, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006805208744481206, + "mgd_loss_weight": 0.5111670315265655, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.56598510742185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.45546264648436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14670, + "valid_token_count": 818.4 + }, + { + "actual_token_mask_ratio": 0.30056907832622526, + "epoch": 4.271166715158569, + "flow_matching_loss": 0.014107178151607513, + "grad_norm": 0.09587374329566956, + "kept_token_count": 572.421875, + "learning_rate": 1.676675542722413e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.9929076552391052, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007092349743470549, + "mgd_loss_weight": 0.5108336985111237, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.9851837158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.13628540039065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14680, + "valid_token_count": 818.41640625 + }, + { + "actual_token_mask_ratio": 0.2998658001422882, + "epoch": 4.274076229269712, + "flow_matching_loss": 0.012192516773939132, + "grad_norm": 0.07911574095487595, + "kept_token_count": 572.94921875, + "learning_rate": 1.6750334752476776e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9929924070835113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070075957570225, + "mgd_loss_weight": 0.5105003476142883, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.4645629882813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2806396484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14690, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.30086217522621156, + "epoch": 4.276985743380855, + "flow_matching_loss": 0.012185504660010338, + "grad_norm": 0.08608243614435196, + "kept_token_count": 571.99609375, + "learning_rate": 1.673391195090765e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9930758714675904, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00692411633208394, + "mgd_loss_weight": 0.5101670026779175, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.2971221923828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5889556884766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14700, + "valid_token_count": 818.14921875 + }, + { + "actual_token_mask_ratio": 0.30001129806041715, + "epoch": 4.279895257491999, + "flow_matching_loss": 0.012764027202501893, + "grad_norm": 0.08386742323637009, + "kept_token_count": 572.790625, + "learning_rate": 1.6717487042472006e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9929735124111175, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007026480929926038, + "mgd_loss_weight": 0.5098336637020111, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4461700439453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.3643310546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14710, + "valid_token_count": 818.28515625 + }, + { + "actual_token_mask_ratio": 0.3000759333372116, + "epoch": 4.282804771603142, + "flow_matching_loss": 0.01419128868728876, + "grad_norm": 0.07051364332437515, + "kept_token_count": 572.80234375, + "learning_rate": 1.6701060047127655e-05, + "loss": 0.0177, + "mgd_cos_sim": 0.993101704120636, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006898286659270525, + "mgd_loss_weight": 0.5095003128051758, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.12330627441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.24456481933595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14720, + "valid_token_count": 818.38203125 + }, + { + "actual_token_mask_ratio": 0.29996098577976227, + "epoch": 4.285714285714286, + "flow_matching_loss": 0.011631798185408115, + "grad_norm": 0.08065053075551987, + "kept_token_count": 572.74765625, + "learning_rate": 1.6684630984834947e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9930217206478119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006978259701281786, + "mgd_loss_weight": 0.5091669738292695, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.0704742431641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3656402587891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14730, + "valid_token_count": 818.16796875 + }, + { + "actual_token_mask_ratio": 0.30024783611297606, + "epoch": 4.288623799825429, + "flow_matching_loss": 0.013887082785367965, + "grad_norm": 0.07866601645946503, + "kept_token_count": 572.65625, + "learning_rate": 1.666819987555675e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9931015729904175, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006898437114432454, + "mgd_loss_weight": 0.5088336288928985, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.8313690185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.05979309082034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14740, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.30007962584495546, + "epoch": 4.291533313936573, + "flow_matching_loss": 0.013546064868569374, + "grad_norm": 0.07463032007217407, + "kept_token_count": 572.609375, + "learning_rate": 1.6651766739258392e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9929681420326233, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007031860807910561, + "mgd_loss_weight": 0.5085002779960632, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.43731994628905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.53717651367185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14750, + "valid_token_count": 818.1078125 + }, + { + "actual_token_mask_ratio": 0.2999055773019791, + "epoch": 4.294442828047716, + "flow_matching_loss": 0.013872588612139225, + "grad_norm": 0.09412600100040436, + "kept_token_count": 572.8296875, + "learning_rate": 1.6635331595907704e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9931195318698883, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006880465755239129, + "mgd_loss_weight": 0.5081669390201569, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.8137634277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3421356201172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14760, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.2990898758172989, + "epoch": 4.2973523421588595, + "flow_matching_loss": 0.011884310003370046, + "grad_norm": 0.07552257180213928, + "kept_token_count": 573.4640625, + "learning_rate": 1.6618894465474923e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9930525720119476, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006947435438632965, + "mgd_loss_weight": 0.507833594083786, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.85437927246096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6599517822266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14770, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.30041449069976806, + "epoch": 4.300261856270003, + "flow_matching_loss": 0.012603654433041811, + "grad_norm": 0.10179132223129272, + "kept_token_count": 572.51875, + "learning_rate": 1.660245536793271e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9927516639232635, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007248336542397737, + "mgd_loss_weight": 0.5075002491474152, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.35841979980466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8822937011719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14780, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.3008368104696274, + "epoch": 4.303171370381146, + "flow_matching_loss": 0.012757605966180563, + "grad_norm": 0.08293239027261734, + "kept_token_count": 572.13046875, + "learning_rate": 1.6586014323256126e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9934036672115326, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006596326641738415, + "mgd_loss_weight": 0.5071669042110443, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.95298156738284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.72987670898436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14790, + "valid_token_count": 818.30546875 + }, + { + "actual_token_mask_ratio": 0.30010792315006257, + "epoch": 4.30608088449229, + "flow_matching_loss": 0.012986967992037534, + "grad_norm": 0.0745205208659172, + "kept_token_count": 572.68828125, + "learning_rate": 1.656957135142259e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9932488918304443, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006751103699207306, + "mgd_loss_weight": 0.5068335652351379, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.3986846923828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.79746704101564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14800, + "valid_token_count": 818.24921875 + }, + { + "actual_token_mask_ratio": 0.29963698983192444, + "epoch": 4.3089903986034335, + "flow_matching_loss": 0.012423314433544875, + "grad_norm": 0.08515193313360214, + "kept_token_count": 573.1875, + "learning_rate": 1.655312647241186e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9925775170326233, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007422478683292866, + "mgd_loss_weight": 0.5065002143383026, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.00563659667966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1328826904297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14810, + "valid_token_count": 818.41484375 + }, + { + "actual_token_mask_ratio": 0.2997140496969223, + "epoch": 4.311899912714576, + "flow_matching_loss": 0.013658144790679217, + "grad_norm": 0.09056068956851959, + "kept_token_count": 572.90234375, + "learning_rate": 1.6536679706206015e-05, + "loss": 0.0172, + "mgd_cos_sim": 0.9929651618003845, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007034847047179937, + "mgd_loss_weight": 0.5061668753623962, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.8351287841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6247222900391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14820, + "valid_token_count": 818.096875 + }, + { + "actual_token_mask_ratio": 0.30011771321296693, + "epoch": 4.31480942682572, + "flow_matching_loss": 0.012687364453449845, + "grad_norm": 0.0779329463839531, + "kept_token_count": 572.8015625, + "learning_rate": 1.6520231072789427e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9931964099407196, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006803602213039994, + "mgd_loss_weight": 0.5058335304260254, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 445.3640930175781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.237890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14830, + "valid_token_count": 818.42265625 + }, + { + "actual_token_mask_ratio": 0.30010469257831573, + "epoch": 4.317718940936864, + "flow_matching_loss": 0.012128205504268408, + "grad_norm": 0.08942503482103348, + "kept_token_count": 572.675, + "learning_rate": 1.6503780592148737e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9929518103599548, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0070481913164258, + "mgd_loss_weight": 0.50550017952919, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.06922912597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.97560119628906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14840, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.2997816622257233, + "epoch": 4.320628455048007, + "flow_matching_loss": 0.013593540154397487, + "grad_norm": 0.0678936019539833, + "kept_token_count": 572.98125, + "learning_rate": 1.648732828427283e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9930160164833068, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006983988638967276, + "mgd_loss_weight": 0.5051668465137482, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.36961364746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1572967529297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14850, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.3000901430845261, + "epoch": 4.32353796915915, + "flow_matching_loss": 0.011780211888253688, + "grad_norm": 0.07762016355991364, + "kept_token_count": 572.7421875, + "learning_rate": 1.647087416915281e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9932978510856628, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006702142627909779, + "mgd_loss_weight": 0.5048334896564484, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.12281799316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0642456054687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14860, + "valid_token_count": 818.30859375 + }, + { + "actual_token_mask_ratio": 0.300268343091011, + "epoch": 4.326447483270294, + "flow_matching_loss": 0.011885860189795493, + "grad_norm": 0.0667591243982315, + "kept_token_count": 572.68125, + "learning_rate": 1.6454418266781987e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9933170318603516, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066829736344516276, + "mgd_loss_weight": 0.504500150680542, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.1880584716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.7190216064453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14870, + "valid_token_count": 818.43046875 + }, + { + "actual_token_mask_ratio": 0.29960655570030215, + "epoch": 4.329356997381438, + "flow_matching_loss": 0.013374095782637596, + "grad_norm": 0.09903121739625931, + "kept_token_count": 573.3296875, + "learning_rate": 1.6437960597155824e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9929027497768402, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007097240956500173, + "mgd_loss_weight": 0.5041668117046356, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.6994293212891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.4018615722656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14880, + "valid_token_count": 818.58671875 + }, + { + "actual_token_mask_ratio": 0.29988706707954405, + "epoch": 4.3322665114925805, + "flow_matching_loss": 0.012808823119848966, + "grad_norm": 0.07654448598623276, + "kept_token_count": 572.8703125, + "learning_rate": 1.642150118027194e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9930992186069488, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006900779018178582, + "mgd_loss_weight": 0.5038334548473358, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.4699768066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.7283477783203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14890, + "valid_token_count": 818.25078125 + }, + { + "actual_token_mask_ratio": 0.29981482923030855, + "epoch": 4.335176025603724, + "flow_matching_loss": 0.013337665889412164, + "grad_norm": 0.07909411191940308, + "kept_token_count": 572.73046875, + "learning_rate": 1.640504003613009e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9932147622108459, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006785237789154052, + "mgd_loss_weight": 0.503500121831894, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.86566772460935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.2699462890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14900, + "valid_token_count": 817.97265625 + }, + { + "actual_token_mask_ratio": 0.3005913257598877, + "epoch": 4.338085539714868, + "flow_matching_loss": 0.013398527726531028, + "grad_norm": 0.07880674302577972, + "kept_token_count": 572.3015625, + "learning_rate": 1.6388577184732107e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9930403709411622, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006959624635055661, + "mgd_loss_weight": 0.5031667709350586, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.07989501953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1893585205078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14910, + "valid_token_count": 818.26171875 + }, + { + "actual_token_mask_ratio": 0.2995902091264725, + "epoch": 4.340995053826011, + "flow_matching_loss": 0.013056343607604503, + "grad_norm": 0.08847059309482574, + "kept_token_count": 573.14140625, + "learning_rate": 1.6372112646081907e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9933010816574097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006698921602219343, + "mgd_loss_weight": 0.5028334259986877, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.96234436035155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0002197265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14920, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.2993497133255005, + "epoch": 4.343904567937154, + "flow_matching_loss": 0.014224905893206597, + "grad_norm": 0.0776701346039772, + "kept_token_count": 573.340625, + "learning_rate": 1.6355646440185464e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9935812294483185, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006418763333931566, + "mgd_loss_weight": 0.5025000870227814, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.744921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.68910522460936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14930, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.29983888268470765, + "epoch": 4.346814082048298, + "flow_matching_loss": 0.012873435486108064, + "grad_norm": 0.08153735101222992, + "kept_token_count": 572.8296875, + "learning_rate": 1.6339178587050764e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9930832922458649, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006916703144088388, + "mgd_loss_weight": 0.5021667361259461, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.5527740478516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.6075103759766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14940, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.3005895882844925, + "epoch": 4.349723596159441, + "flow_matching_loss": 0.013292293157428503, + "grad_norm": 0.10435212403535843, + "kept_token_count": 572.3078125, + "learning_rate": 1.632270910668781e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9927133858203888, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0072865938302129505, + "mgd_loss_weight": 0.5018333971500397, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.1479217529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3764129638672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14950, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.2995288848876953, + "epoch": 4.352633110270585, + "flow_matching_loss": 0.013424551114439964, + "grad_norm": 0.09598628431558609, + "kept_token_count": 573.2296875, + "learning_rate": 1.6306238019108562e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9932810306549072, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006718954350799322, + "mgd_loss_weight": 0.5015000522136688, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.90556640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2519073486328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14960, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.2998792290687561, + "epoch": 4.355542624381728, + "flow_matching_loss": 0.014183442760258912, + "grad_norm": 0.07800886034965515, + "kept_token_count": 572.79375, + "learning_rate": 1.6289765344326952e-05, + "loss": 0.0178, + "mgd_cos_sim": 0.9928199887275696, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007180004566907883, + "mgd_loss_weight": 0.5011667013168335, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.5659149169922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6781433105469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14970, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.2999813437461853, + "epoch": 4.358452138492872, + "flow_matching_loss": 0.013812905736267566, + "grad_norm": 0.1180361658334732, + "kept_token_count": 572.7625, + "learning_rate": 1.6273291102358833e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9928102672100068, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007189721567556262, + "mgd_loss_weight": 0.5008333623409271, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.90509643554685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8298400878906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14980, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.29993854463100433, + "epoch": 4.361361652604015, + "flow_matching_loss": 0.012486141268163919, + "grad_norm": 0.10311220586299896, + "kept_token_count": 572.7546875, + "learning_rate": 1.6256815313221963e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9929961323738098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007003865111619234, + "mgd_loss_weight": 0.5005000174045563, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.416455078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1333923339844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 14990, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.30054253339767456, + "epoch": 4.364271166715159, + "flow_matching_loss": 0.011994313914328814, + "grad_norm": 0.0980309396982193, + "kept_token_count": 572.40234375, + "learning_rate": 1.6240337996935972e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9929618537425995, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007038131635636091, + "mgd_loss_weight": 0.5001666724681855, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.1108123779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5181945800781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15000, + "valid_token_count": 818.34765625 + }, + { + "actual_token_mask_ratio": 0.30036489069461825, + "epoch": 4.367180680826302, + "flow_matching_loss": 0.013024977128952741, + "grad_norm": 0.07805965840816498, + "kept_token_count": 572.51328125, + "learning_rate": 1.6223859173522367e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9933816254138946, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006618364062160253, + "mgd_loss_weight": 0.4998333305120468, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.0310485839844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.33092346191404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15010, + "valid_token_count": 818.30546875 + }, + { + "actual_token_mask_ratio": 0.3002969563007355, + "epoch": 4.370090194937445, + "flow_matching_loss": 0.014055432565510274, + "grad_norm": 0.0862772986292839, + "kept_token_count": 572.696875, + "learning_rate": 1.6207378863004455e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9932431817054749, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067568161524832245, + "mgd_loss_weight": 0.4994999825954437, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.36463928222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.56127014160154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15020, + "valid_token_count": 818.4859375 + }, + { + "actual_token_mask_ratio": 0.3004354566335678, + "epoch": 4.372999709048589, + "flow_matching_loss": 0.011906365677714349, + "grad_norm": 0.10167891532182693, + "kept_token_count": 572.28359375, + "learning_rate": 1.6190897085407382e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9932899475097656, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006710057053714991, + "mgd_loss_weight": 0.4991666376590729, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.40846252441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2013732910156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15030, + "valid_token_count": 818.05234375 + }, + { + "actual_token_mask_ratio": 0.3004930704832077, + "epoch": 4.3759092231597325, + "flow_matching_loss": 0.014463968109339476, + "grad_norm": 0.08780977129936218, + "kept_token_count": 572.39453125, + "learning_rate": 1.617441386075805e-05, + "loss": 0.0181, + "mgd_cos_sim": 0.9927804410457611, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00721955937333405, + "mgd_loss_weight": 0.4988332986831665, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.1159393310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.34254760742186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15040, + "valid_token_count": 818.28671875 + }, + { + "actual_token_mask_ratio": 0.2988845020532608, + "epoch": 4.378818737270876, + "flow_matching_loss": 0.014130098931491374, + "grad_norm": 0.09434499591588974, + "kept_token_count": 573.59921875, + "learning_rate": 1.6157929209085142e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9931823492050171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006817642832174897, + "mgd_loss_weight": 0.49849994480609894, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.0493560791016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8189331054688, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15050, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.29950831830501556, + "epoch": 4.381728251382019, + "flow_matching_loss": 0.012501041544601322, + "grad_norm": 0.07621661573648453, + "kept_token_count": 573.19296875, + "learning_rate": 1.614144315041906e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9928896963596344, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007110302429646253, + "mgd_loss_weight": 0.49816660583019257, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.85337524414064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.66923828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15060, + "valid_token_count": 818.26796875 + }, + { + "actual_token_mask_ratio": 0.30003806948661804, + "epoch": 4.384637765493163, + "flow_matching_loss": 0.015161134116351605, + "grad_norm": 0.08377570658922195, + "kept_token_count": 572.790625, + "learning_rate": 1.612495570479192e-05, + "loss": 0.0188, + "mgd_cos_sim": 0.9926355898380279, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007364405272528529, + "mgd_loss_weight": 0.49783326387405397, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.4898223876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0253631591797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15070, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.2995904713869095, + "epoch": 4.3875472796043065, + "flow_matching_loss": 0.013614681269973516, + "grad_norm": 0.07141207903623581, + "kept_token_count": 573.1484375, + "learning_rate": 1.6108466892237517e-05, + "loss": 0.017, + "mgd_cos_sim": 0.993261742591858, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067382776644080876, + "mgd_loss_weight": 0.49749991297721863, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.01531372070315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2621643066406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15080, + "valid_token_count": 818.30390625 + }, + { + "actual_token_mask_ratio": 0.3007225662469864, + "epoch": 4.390456793715449, + "flow_matching_loss": 0.011835049837827682, + "grad_norm": 0.06986317038536072, + "kept_token_count": 572.2890625, + "learning_rate": 1.609197673279132e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9933444619178772, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006655530538409948, + "mgd_loss_weight": 0.49716657400131226, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.9477783203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.17247314453124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15090, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.2996186077594757, + "epoch": 4.393366307826593, + "flow_matching_loss": 0.011724363919347525, + "grad_norm": 0.08517413586378098, + "kept_token_count": 573.00078125, + "learning_rate": 1.6075485246490428e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9930361449718476, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006963872537016869, + "mgd_loss_weight": 0.49683322608470915, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.3738708496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0508728027344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15100, + "valid_token_count": 818.12421875 + }, + { + "actual_token_mask_ratio": 0.30060323774814607, + "epoch": 4.396275821937737, + "flow_matching_loss": 0.01184395533055067, + "grad_norm": 0.07339180260896683, + "kept_token_count": 572.17265625, + "learning_rate": 1.605899245337355e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9933080017566681, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006692003691568971, + "mgd_loss_weight": 0.4964998811483383, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.0503875732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5835723876953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15110, + "valid_token_count": 818.08984375 + }, + { + "actual_token_mask_ratio": 0.29973814487457273, + "epoch": 4.3991853360488795, + "flow_matching_loss": 0.012955916300415992, + "grad_norm": 0.07599437236785889, + "kept_token_count": 573.1125, + "learning_rate": 1.6042498373480983e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9931784927845001, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068215093575418, + "mgd_loss_weight": 0.49616654217243195, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.80853881835935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1287811279297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15120, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.299638369679451, + "epoch": 4.402094850160023, + "flow_matching_loss": 0.012562703341245651, + "grad_norm": 0.06697738915681839, + "kept_token_count": 573.171875, + "learning_rate": 1.602600302685459e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9931226968765259, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00687732296064496, + "mgd_loss_weight": 0.49583319425582884, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.03247985839846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.19667358398436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15130, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.29962619245052335, + "epoch": 4.405004364271167, + "flow_matching_loss": 0.012855332344770432, + "grad_norm": 0.09012667834758759, + "kept_token_count": 573.0109375, + "learning_rate": 1.6009506433537774e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9929707705974579, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007029232475906611, + "mgd_loss_weight": 0.495499849319458, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.8593444824219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1510375976562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15140, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.2996619641780853, + "epoch": 4.40791387838231, + "flow_matching_loss": 0.012185138836503028, + "grad_norm": 0.07966164499521255, + "kept_token_count": 573.0796875, + "learning_rate": 1.599300861357545e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.993242061138153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006757943937554955, + "mgd_loss_weight": 0.49516651034355164, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.28179626464845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9258117675781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15150, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.3000454127788544, + "epoch": 4.4108233924934535, + "flow_matching_loss": 0.011756565142422914, + "grad_norm": 0.06165819615125656, + "kept_token_count": 572.6203125, + "learning_rate": 1.5976509587014032e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9930962324142456, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006903773872181773, + "mgd_loss_weight": 0.49483315646648407, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.87841796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5866333007813, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15160, + "valid_token_count": 818.08671875 + }, + { + "actual_token_mask_ratio": 0.30001826882362365, + "epoch": 4.413732906604597, + "flow_matching_loss": 0.012971948506310583, + "grad_norm": 0.07863325625658035, + "kept_token_count": 572.82890625, + "learning_rate": 1.5960009373901384e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9934598684310914, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006540123280137778, + "mgd_loss_weight": 0.4944998174905777, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.37811889648435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.59522399902346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15170, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.299774706363678, + "epoch": 4.416642420715741, + "flow_matching_loss": 0.012784666381776333, + "grad_norm": 0.07533009350299835, + "kept_token_count": 572.875, + "learning_rate": 1.5943507994286823e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.99308642745018, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006913578603416681, + "mgd_loss_weight": 0.4941664755344391, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.73703918457034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.40382690429686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15180, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.2996586084365845, + "epoch": 4.419551934826884, + "flow_matching_loss": 0.012034591380506754, + "grad_norm": 0.08807819336652756, + "kept_token_count": 573.08125, + "learning_rate": 1.5927005468221082e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9930710613727569, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006928946916013956, + "mgd_loss_weight": 0.49383312463760376, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.3375274658203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1679351806641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15190, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.2992686927318573, + "epoch": 4.422461448938027, + "flow_matching_loss": 0.012065718974918127, + "grad_norm": 0.08320382237434387, + "kept_token_count": 573.3421875, + "learning_rate": 1.5910501815756285e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9936542987823487, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006345714908093214, + "mgd_loss_weight": 0.4934997856616974, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.6751770019531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.3626678466797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15200, + "valid_token_count": 818.2125 + }, + { + "actual_token_mask_ratio": 0.2996751546859741, + "epoch": 4.425370963049171, + "flow_matching_loss": 0.014575369749218225, + "grad_norm": 0.08229606598615646, + "kept_token_count": 573.05078125, + "learning_rate": 1.589399705694593e-05, + "loss": 0.0179, + "mgd_cos_sim": 0.9932570815086365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006742908898741007, + "mgd_loss_weight": 0.4931664377450943, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.6085571289062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.33644104003906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15210, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.298877140879631, + "epoch": 4.428280477160314, + "flow_matching_loss": 0.01192712653428316, + "grad_norm": 0.0685049518942833, + "kept_token_count": 573.840625, + "learning_rate": 1.587749121184485e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9929967403411866, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007003255607560277, + "mgd_loss_weight": 0.49283309280872345, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.2552459716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4611877441406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15220, + "valid_token_count": 818.45234375 + }, + { + "actual_token_mask_ratio": 0.3002310752868652, + "epoch": 4.431189991271458, + "flow_matching_loss": 0.012605211045593024, + "grad_norm": 0.1149485781788826, + "kept_token_count": 572.628125, + "learning_rate": 1.5860984300509206e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9932288467884064, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006771144876256585, + "mgd_loss_weight": 0.4924997538328171, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.50279846191404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1130126953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15230, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.2993137091398239, + "epoch": 4.434099505382601, + "flow_matching_loss": 0.011246752599254251, + "grad_norm": 0.07126425951719284, + "kept_token_count": 573.371875, + "learning_rate": 1.5844476342996453e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9931433916091919, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006856612488627433, + "mgd_loss_weight": 0.49216640293598174, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.3472961425781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.91355285644534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15240, + "valid_token_count": 818.29609375 + }, + { + "actual_token_mask_ratio": 0.3012507945299149, + "epoch": 4.437009019493744, + "flow_matching_loss": 0.01123087303712964, + "grad_norm": 0.062217507511377335, + "kept_token_count": 571.94921875, + "learning_rate": 1.5827967359365306e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9930424869060517, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006957521801814437, + "mgd_loss_weight": 0.49183306097984314, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.2561920166016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5604644775391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15250, + "valid_token_count": 818.53203125 + }, + { + "actual_token_mask_ratio": 0.29992743134498595, + "epoch": 4.439918533604888, + "flow_matching_loss": 0.011564000556245447, + "grad_norm": 0.0808553397655487, + "kept_token_count": 572.8546875, + "learning_rate": 1.5811457369675758e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9933478474617005, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006652151281014085, + "mgd_loss_weight": 0.49149971902370454, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.35179748535154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.44517211914064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15260, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.3001003235578537, + "epoch": 4.442828047716032, + "flow_matching_loss": 0.012869366444647312, + "grad_norm": 0.08158465474843979, + "kept_token_count": 572.6796875, + "learning_rate": 1.5794946393988982e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9930577933788299, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00694221374578774, + "mgd_loss_weight": 0.4911663681268692, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.7257843017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.89117736816405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15270, + "valid_token_count": 818.23203125 + }, + { + "actual_token_mask_ratio": 0.30006239414215086, + "epoch": 4.445737561827175, + "flow_matching_loss": 0.01286917794495821, + "grad_norm": 0.08292505145072937, + "kept_token_count": 572.8015625, + "learning_rate": 1.5778434452367382e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9929383754730224, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007061615353450179, + "mgd_loss_weight": 0.49083302915096283, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.624072265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9591888427734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15280, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.2999578893184662, + "epoch": 4.448647075938318, + "flow_matching_loss": 0.011984088644385337, + "grad_norm": 0.07395964115858078, + "kept_token_count": 572.89609375, + "learning_rate": 1.576192156487452e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9932641744613647, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006735825398936868, + "mgd_loss_weight": 0.490499684214592, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.2016235351563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.993408203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15290, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.3000071495771408, + "epoch": 4.451556590049462, + "flow_matching_loss": 0.0123532778583467, + "grad_norm": 0.08366424590349197, + "kept_token_count": 572.76484375, + "learning_rate": 1.574540775157511e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9934540331363678, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006545948516577482, + "mgd_loss_weight": 0.4901663362979889, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.6676940917969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3763000488281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15300, + "valid_token_count": 818.23984375 + }, + { + "actual_token_mask_ratio": 0.30075172185897825, + "epoch": 4.4544661041606055, + "flow_matching_loss": 0.013377592619508504, + "grad_norm": 0.08617480844259262, + "kept_token_count": 572.22265625, + "learning_rate": 1.5728893032534998e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9930946528911591, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069053411483764645, + "mgd_loss_weight": 0.4898329973220825, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.706640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4496734619141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15310, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.2985995143651962, + "epoch": 4.457375618271748, + "flow_matching_loss": 0.011272275727242232, + "grad_norm": 0.06998316198587418, + "kept_token_count": 573.74140625, + "learning_rate": 1.571237742782112e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.993409413099289, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006590611068531871, + "mgd_loss_weight": 0.4894996464252472, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.8858947753906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6409393310547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15320, + "valid_token_count": 817.99609375 + }, + { + "actual_token_mask_ratio": 0.3001165211200714, + "epoch": 4.460285132382892, + "flow_matching_loss": 0.011613793391734361, + "grad_norm": 0.07371269911527634, + "kept_token_count": 572.80390625, + "learning_rate": 1.569586095750149e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9927064538002014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007293552346527577, + "mgd_loss_weight": 0.4891663044691086, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.2336730957031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5248626708984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15330, + "valid_token_count": 818.428125 + }, + { + "actual_token_mask_ratio": 0.3002583861351013, + "epoch": 4.463194646494036, + "flow_matching_loss": 0.013174074050039054, + "grad_norm": 0.09739914536476135, + "kept_token_count": 572.53515625, + "learning_rate": 1.5679343641645193e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9931663393974304, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006833666423335672, + "mgd_loss_weight": 0.4888329654932022, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.559228515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3295379638672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15340, + "valid_token_count": 818.21171875 + }, + { + "actual_token_mask_ratio": 0.3001825034618378, + "epoch": 4.466104160605179, + "flow_matching_loss": 0.013088422268629074, + "grad_norm": 0.08138392120599747, + "kept_token_count": 572.675, + "learning_rate": 1.5662825500322304e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9929431915283203, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007056812010705471, + "mgd_loss_weight": 0.48849961459636687, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.7411651611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.47642517089844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15350, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.29995855391025544, + "epoch": 4.469013674716322, + "flow_matching_loss": 0.014087534416466952, + "grad_norm": 0.09615359455347061, + "kept_token_count": 572.76640625, + "learning_rate": 1.5646306553603932e-05, + "loss": 0.0175, + "mgd_cos_sim": 0.9930738389492035, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006926167057827115, + "mgd_loss_weight": 0.48816627264022827, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.7599792480469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9030364990234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15360, + "valid_token_count": 818.19140625 + }, + { + "actual_token_mask_ratio": 0.3004345506429672, + "epoch": 4.471923188827466, + "flow_matching_loss": 0.012305388972163201, + "grad_norm": 0.072300024330616, + "kept_token_count": 572.446875, + "learning_rate": 1.5629786821562154e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9931809842586518, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006819010525941849, + "mgd_loss_weight": 0.48783293068408967, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.3313293457031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.6388854980469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15370, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.2998558759689331, + "epoch": 4.47483270293861, + "flow_matching_loss": 0.013401061948388814, + "grad_norm": 0.09411954879760742, + "kept_token_count": 573.07109375, + "learning_rate": 1.561326632427001e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.993044114112854, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006955887563526631, + "mgd_loss_weight": 0.48749957978725433, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.4526397705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3010559082031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15380, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.3001893937587738, + "epoch": 4.4777422170497525, + "flow_matching_loss": 0.012564524635672569, + "grad_norm": 0.07975422590970993, + "kept_token_count": 572.68828125, + "learning_rate": 1.5596745081801447e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9932675540447236, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006732447119429707, + "mgd_loss_weight": 0.48716624081134796, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.2117980957031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.663720703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15390, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.30059236884117124, + "epoch": 4.480651731160896, + "flow_matching_loss": 0.011562782153487206, + "grad_norm": 0.08116268366575241, + "kept_token_count": 572.1546875, + "learning_rate": 1.5580223114231348e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9927450537681579, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007254949817433953, + "mgd_loss_weight": 0.48683289587497713, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.78081665039065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.55617065429686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15400, + "valid_token_count": 818.0578125 + }, + { + "actual_token_mask_ratio": 0.299827516078949, + "epoch": 4.48356124527204, + "flow_matching_loss": 0.013228830974549055, + "grad_norm": 0.08183743059635162, + "kept_token_count": 572.88828125, + "learning_rate": 1.5563700441635445e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9924024641513824, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007597533939406276, + "mgd_loss_weight": 0.486499547958374, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 397.6616912841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2009979248047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15410, + "valid_token_count": 818.21796875 + }, + { + "actual_token_mask_ratio": 0.3001865267753601, + "epoch": 4.486470759383183, + "flow_matching_loss": 0.011939853429794312, + "grad_norm": 0.08958476781845093, + "kept_token_count": 572.78359375, + "learning_rate": 1.5547177084090358e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9926928639411926, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007307121669873596, + "mgd_loss_weight": 0.48616620898246765, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.3259674072266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9477935791016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15420, + "valid_token_count": 818.478125 + }, + { + "actual_token_mask_ratio": 0.29959012269973756, + "epoch": 4.4893802734943264, + "flow_matching_loss": 0.011547277402132749, + "grad_norm": 0.07184169441461563, + "kept_token_count": 573.23515625, + "learning_rate": 1.5530653061673514e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9933103561401367, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066896445583552126, + "mgd_loss_weight": 0.4858328580856323, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.9059753417969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.3985321044922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15430, + "valid_token_count": 818.42890625 + }, + { + "actual_token_mask_ratio": 0.3001323252916336, + "epoch": 4.49228978760547, + "flow_matching_loss": 0.011404246930032969, + "grad_norm": 0.08437494933605194, + "kept_token_count": 572.796875, + "learning_rate": 1.5514128394463167e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9930864632129669, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006913547636941075, + "mgd_loss_weight": 0.4854995161294937, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.4462005615234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.7578918457031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15440, + "valid_token_count": 818.44375 + }, + { + "actual_token_mask_ratio": 0.2998964756727219, + "epoch": 4.495199301716613, + "flow_matching_loss": 0.013907301984727383, + "grad_norm": 0.09065144509077072, + "kept_token_count": 573.00390625, + "learning_rate": 1.549760310253834e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9927540838718414, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007245927210897207, + "mgd_loss_weight": 0.48516617715358734, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.3587982177734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.46694030761716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15450, + "valid_token_count": 818.4578125 + }, + { + "actual_token_mask_ratio": 0.3000201880931854, + "epoch": 4.498108815827757, + "flow_matching_loss": 0.011875758599489927, + "grad_norm": 0.08591147512197495, + "kept_token_count": 572.89453125, + "learning_rate": 1.5481077205978816e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9932595789432526, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00674042496830225, + "mgd_loss_weight": 0.4848328232765198, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.44292907714845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1276794433594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15460, + "valid_token_count": 818.44765625 + }, + { + "actual_token_mask_ratio": 0.30117333829402926, + "epoch": 4.5010183299389, + "flow_matching_loss": 0.011988050770014524, + "grad_norm": 0.07968918234109879, + "kept_token_count": 571.76875, + "learning_rate": 1.5464550724865135e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9932060539722443, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006793946586549282, + "mgd_loss_weight": 0.4844994843006134, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.01398315429685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3337341308594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15470, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.3000207424163818, + "epoch": 4.503927844050043, + "flow_matching_loss": 0.013404296990484, + "grad_norm": 0.0932721421122551, + "kept_token_count": 572.925, + "learning_rate": 1.5448023679278517e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.992986673116684, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007013331726193428, + "mgd_loss_weight": 0.4841661393642426, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.4650360107422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.468701171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15480, + "valid_token_count": 818.4859375 + }, + { + "actual_token_mask_ratio": 0.30076920092105863, + "epoch": 4.506837358161187, + "flow_matching_loss": 0.011718090437352658, + "grad_norm": 0.10710594058036804, + "kept_token_count": 571.91328125, + "learning_rate": 1.543149608930088e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9930663883686066, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006933614285662771, + "mgd_loss_weight": 0.48383279144763947, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.70769958496095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1197235107422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15490, + "valid_token_count": 817.9203125 + }, + { + "actual_token_mask_ratio": 0.30076904892921447, + "epoch": 4.509746872272331, + "flow_matching_loss": 0.012707126513123512, + "grad_norm": 0.09214523434638977, + "kept_token_count": 572.03046875, + "learning_rate": 1.5414967975014805e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9934231579303742, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006576841231435537, + "mgd_loss_weight": 0.4834994524717331, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.228466796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4648956298828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15500, + "valid_token_count": 818.08671875 + }, + { + "actual_token_mask_ratio": 0.2998719274997711, + "epoch": 4.512656386383474, + "flow_matching_loss": 0.012595030479133129, + "grad_norm": 0.08898354321718216, + "kept_token_count": 573.01796875, + "learning_rate": 1.539843935650351e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9932810723781585, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006718937726691365, + "mgd_loss_weight": 0.48316610455513, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.72869262695315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.53827514648435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15510, + "valid_token_count": 818.44296875 + }, + { + "actual_token_mask_ratio": 0.2997107535600662, + "epoch": 4.515565900494617, + "flow_matching_loss": 0.012520382553339005, + "grad_norm": 0.10217861086130142, + "kept_token_count": 573.07265625, + "learning_rate": 1.5381910253850834e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9933884680271149, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006611524289473892, + "mgd_loss_weight": 0.48283275961875916, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.77161560058596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1546325683594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15520, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.3005198061466217, + "epoch": 4.518475414605761, + "flow_matching_loss": 0.013724250532686711, + "grad_norm": 0.07188605517148972, + "kept_token_count": 572.22578125, + "learning_rate": 1.5365380687141183e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9933196485042572, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006680347211658955, + "mgd_loss_weight": 0.4824994206428528, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.1010009765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.78709411621094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15530, + "valid_token_count": 818.07109375 + }, + { + "actual_token_mask_ratio": 0.2999421954154968, + "epoch": 4.521384928716905, + "flow_matching_loss": 0.012815950158983468, + "grad_norm": 0.10174369066953659, + "kept_token_count": 572.96015625, + "learning_rate": 1.5348850676459544e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9928240299224853, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007175955176353455, + "mgd_loss_weight": 0.4821660667657852, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.5063507080078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9541534423828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15540, + "valid_token_count": 818.44453125 + }, + { + "actual_token_mask_ratio": 0.2998396188020706, + "epoch": 4.524294442828047, + "flow_matching_loss": 0.012307167891412974, + "grad_norm": 0.07504288107156754, + "kept_token_count": 572.81171875, + "learning_rate": 1.5332320241891438e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9931927502155304, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006807232834398746, + "mgd_loss_weight": 0.48183272778987885, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.7868316650391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1863037109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15550, + "valid_token_count": 818.1109375 + }, + { + "actual_token_mask_ratio": 0.30076119005680085, + "epoch": 4.527203956939191, + "flow_matching_loss": 0.013446128647774457, + "grad_norm": 0.08641427010297775, + "kept_token_count": 572.18515625, + "learning_rate": 1.5315789403522908e-05, + "loss": 0.0167, + "mgd_cos_sim": 0.9932036578655243, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006796354381367564, + "mgd_loss_weight": 0.4814993888139725, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.45595397949216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7383636474609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15560, + "valid_token_count": 818.29609375 + }, + { + "actual_token_mask_ratio": 0.30044549107551577, + "epoch": 4.530113471050335, + "flow_matching_loss": 0.012929106038063765, + "grad_norm": 0.08499998599290848, + "kept_token_count": 572.4265625, + "learning_rate": 1.529925818144047e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9931342363357544, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006865756912156939, + "mgd_loss_weight": 0.4811660349369049, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.94603271484374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0061981201172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15570, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.2993171185255051, + "epoch": 4.5330229851614785, + "flow_matching_loss": 0.011508031003177165, + "grad_norm": 0.0770883709192276, + "kept_token_count": 573.47734375, + "learning_rate": 1.5282726595731126e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9930872023105621, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069128095638006926, + "mgd_loss_weight": 0.48083269596099854, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.0829559326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.159619140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15580, + "valid_token_count": 818.453125 + }, + { + "actual_token_mask_ratio": 0.2999164015054703, + "epoch": 4.535932499272621, + "flow_matching_loss": 0.012320182519033551, + "grad_norm": 0.09675179421901703, + "kept_token_count": 572.9453125, + "learning_rate": 1.5266194666482304e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9933192670345307, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006680721137672663, + "mgd_loss_weight": 0.4804993510246277, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.7014221191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.26349182128905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15590, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.30022822618484496, + "epoch": 4.538842013383765, + "flow_matching_loss": 0.012945900671184063, + "grad_norm": 0.09287720173597336, + "kept_token_count": 572.540625, + "learning_rate": 1.5249662413781866e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9934697926044465, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006530215125530958, + "mgd_loss_weight": 0.4801660031080246, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.59014282226565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.6756256103516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15600, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.30048506557941435, + "epoch": 4.541751527494909, + "flow_matching_loss": 0.012533223628997803, + "grad_norm": 0.11299102753400803, + "kept_token_count": 572.50390625, + "learning_rate": 1.5233129857718054e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9932082533836365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006791749643161893, + "mgd_loss_weight": 0.4798326641321182, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.7757507324219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8176635742187, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15610, + "valid_token_count": 818.42421875 + }, + { + "actual_token_mask_ratio": 0.2995013892650604, + "epoch": 4.5446610416060516, + "flow_matching_loss": 0.010739990696310996, + "grad_norm": 0.0721021518111229, + "kept_token_count": 573.0328125, + "learning_rate": 1.521659701837948e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9930601179599762, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006939894799143076, + "mgd_loss_weight": 0.4794993162155151, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.2288787841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.401416015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15620, + "valid_token_count": 818.03984375 + }, + { + "actual_token_mask_ratio": 0.29957040548324587, + "epoch": 4.547570555717195, + "flow_matching_loss": 0.012975187227129936, + "grad_norm": 0.08860577642917633, + "kept_token_count": 573.18359375, + "learning_rate": 1.5200063915855107e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9928067326545715, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007193245645612478, + "mgd_loss_weight": 0.4791659712791443, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.8767578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.35516357421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15630, + "valid_token_count": 818.33515625 + }, + { + "actual_token_mask_ratio": 0.29960175454616544, + "epoch": 4.550480069828339, + "flow_matching_loss": 0.012184417247772217, + "grad_norm": 0.08343596756458282, + "kept_token_count": 573.15078125, + "learning_rate": 1.5183530570234212e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.993392550945282, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006607458088546991, + "mgd_loss_weight": 0.4788326323032379, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.94737548828124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5142822265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15640, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.300544211268425, + "epoch": 4.553389583939482, + "flow_matching_loss": 0.012464826460927725, + "grad_norm": 0.06548494100570679, + "kept_token_count": 572.453125, + "learning_rate": 1.5166997001606361e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9932331442832947, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006766855577006936, + "mgd_loss_weight": 0.47849927842617035, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 445.2037689208984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9933807373047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15650, + "valid_token_count": 818.42890625 + }, + { + "actual_token_mask_ratio": 0.3000546872615814, + "epoch": 4.5562990980506255, + "flow_matching_loss": 0.011268515745177865, + "grad_norm": 0.08918432146310806, + "kept_token_count": 572.8, + "learning_rate": 1.5150463230061417e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9933070182800293, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006692966725677252, + "mgd_loss_weight": 0.478165939450264, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.03453063964844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.91636962890624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15660, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.2994785100221634, + "epoch": 4.559208612161769, + "flow_matching_loss": 0.01238416377454996, + "grad_norm": 0.07636178284883499, + "kept_token_count": 573.35703125, + "learning_rate": 1.5133929275689455e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9933024287223816, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006697562662884593, + "mgd_loss_weight": 0.4778325974941254, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.34051513671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.65553894042966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15670, + "valid_token_count": 818.471875 + }, + { + "actual_token_mask_ratio": 0.30075567662715913, + "epoch": 4.562118126272912, + "flow_matching_loss": 0.012427662685513496, + "grad_norm": 0.0857982262969017, + "kept_token_count": 572.240625, + "learning_rate": 1.5117395158580795e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9930866479873657, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006913357926532626, + "mgd_loss_weight": 0.47749924659729004, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 455.8392059326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.48740234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15680, + "valid_token_count": 818.37109375 + }, + { + "actual_token_mask_ratio": 0.30028350949287413, + "epoch": 4.565027640384056, + "flow_matching_loss": 0.013097687251865863, + "grad_norm": 0.08803581446409225, + "kept_token_count": 572.484375, + "learning_rate": 1.510086089882595e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9932005047798157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006799502298235893, + "mgd_loss_weight": 0.47716590762138367, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.0293273925781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8707336425781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15690, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.30020527839660643, + "epoch": 4.567937154495199, + "flow_matching_loss": 0.012103272788226604, + "grad_norm": 0.08650704473257065, + "kept_token_count": 572.490625, + "learning_rate": 1.5084326516515599e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9932851374149323, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006714864447712898, + "mgd_loss_weight": 0.47683255970478056, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.1600769042969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.52568969726565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15700, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.2997149914503098, + "epoch": 4.570846668606343, + "flow_matching_loss": 0.013385849446058274, + "grad_norm": 0.08099829405546188, + "kept_token_count": 572.88359375, + "learning_rate": 1.5067792031740582e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9931685507297516, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006831448059529066, + "mgd_loss_weight": 0.47649921476840973, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 447.2608337402344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7986328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15710, + "valid_token_count": 818.06953125 + }, + { + "actual_token_mask_ratio": 0.29952149987220766, + "epoch": 4.573756182717486, + "flow_matching_loss": 0.013171642180532217, + "grad_norm": 0.1110115796327591, + "kept_token_count": 572.92421875, + "learning_rate": 1.505125746459185e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9930344343185424, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069655782077461485, + "mgd_loss_weight": 0.47616587579250336, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 396.77077026367186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.68809814453124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15720, + "valid_token_count": 817.9 + }, + { + "actual_token_mask_ratio": 0.2999217391014099, + "epoch": 4.57666569682863, + "flow_matching_loss": 0.011848152056336403, + "grad_norm": 0.09821687638759613, + "kept_token_count": 572.82109375, + "learning_rate": 1.5034722835160466e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9927797317504883, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007220278074964881, + "mgd_loss_weight": 0.475832524895668, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.33751525878904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.3534301757812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15730, + "valid_token_count": 818.22265625 + }, + { + "actual_token_mask_ratio": 0.2988280862569809, + "epoch": 4.579575210939773, + "flow_matching_loss": 0.011286744195967913, + "grad_norm": 0.08566530793905258, + "kept_token_count": 573.77890625, + "learning_rate": 1.5018188163537566e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9933088898658753, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006691121077165007, + "mgd_loss_weight": 0.4754991829395294, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.46067810058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.12486877441404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15740, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.30055712163448334, + "epoch": 4.582484725050916, + "flow_matching_loss": 0.012081790622323752, + "grad_norm": 0.08512621372938156, + "kept_token_count": 572.3109375, + "learning_rate": 1.5001653469814331e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9932293057441711, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006770706130191683, + "mgd_loss_weight": 0.4751658409833908, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.0142456054688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0108245849609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15750, + "valid_token_count": 818.24140625 + }, + { + "actual_token_mask_ratio": 0.30046798586845397, + "epoch": 4.58539423916206, + "flow_matching_loss": 0.011852116510272025, + "grad_norm": 0.07366219907999039, + "kept_token_count": 572.36953125, + "learning_rate": 1.4985118774081978e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9931480646133423, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006851920858025551, + "mgd_loss_weight": 0.4748324900865555, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.7316436767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3020782470703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15760, + "valid_token_count": 818.22109375 + }, + { + "actual_token_mask_ratio": 0.300607430934906, + "epoch": 4.588303753273204, + "flow_matching_loss": 0.01184073993936181, + "grad_norm": 0.08624876290559769, + "kept_token_count": 572.24765625, + "learning_rate": 1.4968584096431708e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9934126675128937, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006587347434833646, + "mgd_loss_weight": 0.4744991511106491, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.55881958007814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9169891357422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15770, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.2998848736286163, + "epoch": 4.591213267384346, + "flow_matching_loss": 0.01164446547627449, + "grad_norm": 0.09028242528438568, + "kept_token_count": 573.02734375, + "learning_rate": 1.4952049456954731e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9936288952827453, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006371104577556253, + "mgd_loss_weight": 0.4741658091545105, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.6273559570312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0739318847656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15780, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.29924592971801756, + "epoch": 4.59412278149549, + "flow_matching_loss": 0.012459880951792002, + "grad_norm": 0.08065323531627655, + "kept_token_count": 573.47578125, + "learning_rate": 1.4935514875742182e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9933086574077606, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006691329553723336, + "mgd_loss_weight": 0.47383245825767517, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.5182159423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7395812988281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15790, + "valid_token_count": 818.37265625 + }, + { + "actual_token_mask_ratio": 0.2988205552101135, + "epoch": 4.597032295606634, + "flow_matching_loss": 0.011440170975401998, + "grad_norm": 0.09436231851577759, + "kept_token_count": 573.86875, + "learning_rate": 1.4918980372885143e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9934068024158478, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006593194790184498, + "mgd_loss_weight": 0.4734991192817688, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.71437072753906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.72413330078126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15800, + "valid_token_count": 818.43515625 + }, + { + "actual_token_mask_ratio": 0.3002362221479416, + "epoch": 4.599941809717777, + "flow_matching_loss": 0.012718828301876784, + "grad_norm": 0.07611358910799026, + "kept_token_count": 572.60625, + "learning_rate": 1.4902445968474591e-05, + "loss": 0.016, + "mgd_cos_sim": 0.993045300245285, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006954704131931067, + "mgd_loss_weight": 0.4731657713651657, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.7812133789063, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1711151123047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15810, + "valid_token_count": 818.28046875 + }, + { + "actual_token_mask_ratio": 0.30012679994106295, + "epoch": 4.60285132382892, + "flow_matching_loss": 0.012604235857725143, + "grad_norm": 0.08440844714641571, + "kept_token_count": 572.62265625, + "learning_rate": 1.4885911682601378e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9931127607822419, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006887239497154951, + "mgd_loss_weight": 0.47283242642879486, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.0480529785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8177154541016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15820, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.2999605506658554, + "epoch": 4.605760837940064, + "flow_matching_loss": 0.012075218558311462, + "grad_norm": 0.0700891837477684, + "kept_token_count": 572.8921875, + "learning_rate": 1.4869377535356235e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9934535801410675, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006546440999954939, + "mgd_loss_weight": 0.4724990874528885, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.2096435546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.81801452636716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15830, + "valid_token_count": 818.37265625 + }, + { + "actual_token_mask_ratio": 0.30015788674354554, + "epoch": 4.608670352051208, + "flow_matching_loss": 0.011329412460327148, + "grad_norm": 0.07097378373146057, + "kept_token_count": 572.52109375, + "learning_rate": 1.4852843546829704e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9934652626514435, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006534743821248412, + "mgd_loss_weight": 0.47216573655605315, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.87071533203124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3076873779297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15840, + "valid_token_count": 818.07265625 + }, + { + "actual_token_mask_ratio": 0.3007671147584915, + "epoch": 4.611579866162351, + "flow_matching_loss": 0.012480848096311092, + "grad_norm": 0.08128620684146881, + "kept_token_count": 572.19375, + "learning_rate": 1.483630973711214e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9930981814861297, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006901822611689567, + "mgd_loss_weight": 0.47183239459991455, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.8633575439453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.20787658691404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15850, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.29944452047348025, + "epoch": 4.614489380273494, + "flow_matching_loss": 0.012392463814467191, + "grad_norm": 0.1241554468870163, + "kept_token_count": 573.43984375, + "learning_rate": 1.4819776126293677e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9934881329536438, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006511870445683599, + "mgd_loss_weight": 0.47149905264377595, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.75762634277345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.04822692871096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15860, + "valid_token_count": 818.55078125 + }, + { + "actual_token_mask_ratio": 0.29973472058773043, + "epoch": 4.617398894384638, + "flow_matching_loss": 0.012299606902524829, + "grad_norm": 0.12802423536777496, + "kept_token_count": 572.94140625, + "learning_rate": 1.4803242734464214e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9932713627815246, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006728650536388159, + "mgd_loss_weight": 0.4711657017469406, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.5719970703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0373199462891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15870, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.29996584057807923, + "epoch": 4.620308408495781, + "flow_matching_loss": 0.011998192220926285, + "grad_norm": 0.06939028948545456, + "kept_token_count": 572.95625, + "learning_rate": 1.4786709581713383e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9934210300445556, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006578984716907143, + "mgd_loss_weight": 0.47083236277103424, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.75669555664064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5047302246094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15880, + "valid_token_count": 818.47265625 + }, + { + "actual_token_mask_ratio": 0.30062286257743837, + "epoch": 4.6232179226069245, + "flow_matching_loss": 0.013011866901069879, + "grad_norm": 0.08499739319086075, + "kept_token_count": 572.15703125, + "learning_rate": 1.4770176688130524e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9929755747318267, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007024438492953778, + "mgd_loss_weight": 0.4704990178346634, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.6360778808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3987609863281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15890, + "valid_token_count": 818.09296875 + }, + { + "actual_token_mask_ratio": 0.29975931644439696, + "epoch": 4.626127436718068, + "flow_matching_loss": 0.014396215230226517, + "grad_norm": 0.0865602046251297, + "kept_token_count": 572.9703125, + "learning_rate": 1.475364407380466e-05, + "loss": 0.0174, + "mgd_cos_sim": 0.9935499548912048, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064500485546886924, + "mgd_loss_weight": 0.4701656699180603, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.09361267089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1024169921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15900, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.30024552047252656, + "epoch": 4.629036950829212, + "flow_matching_loss": 0.012122406903654338, + "grad_norm": 0.08692227303981781, + "kept_token_count": 572.5390625, + "learning_rate": 1.4737111758824476e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9932375013828277, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006762511795386672, + "mgd_loss_weight": 0.46983233094215393, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.69102783203124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.33636779785155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15910, + "valid_token_count": 818.19765625 + }, + { + "actual_token_mask_ratio": 0.30007254183292387, + "epoch": 4.631946464940355, + "flow_matching_loss": 0.011977783124893903, + "grad_norm": 0.07070255279541016, + "kept_token_count": 572.6859375, + "learning_rate": 1.4720579763278289e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9934928297996521, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006507173785939813, + "mgd_loss_weight": 0.4694989800453186, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.3181457519531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.80296020507814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15920, + "valid_token_count": 818.21015625 + }, + { + "actual_token_mask_ratio": 0.29854980707168577, + "epoch": 4.6348559790514985, + "flow_matching_loss": 0.01352955112233758, + "grad_norm": 0.08282982558012009, + "kept_token_count": 574.0109375, + "learning_rate": 1.4704048107254042e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9930181980133057, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006981800310313702, + "mgd_loss_weight": 0.46916563808918, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.8368377685547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4208984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15930, + "valid_token_count": 818.32265625 + }, + { + "actual_token_mask_ratio": 0.29965728521347046, + "epoch": 4.637765493162642, + "flow_matching_loss": 0.012804881203919649, + "grad_norm": 0.08833421021699905, + "kept_token_count": 572.95625, + "learning_rate": 1.4687516810839246e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9930727303028106, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006927257357165218, + "mgd_loss_weight": 0.4688322991132736, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.12993774414065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3939270019531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15940, + "valid_token_count": 818.1078125 + }, + { + "actual_token_mask_ratio": 0.300835981965065, + "epoch": 4.640675007273785, + "flow_matching_loss": 0.013281662203371524, + "grad_norm": 0.07702866196632385, + "kept_token_count": 572.134375, + "learning_rate": 1.4670985894120997e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9933916807174683, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006608312018215656, + "mgd_loss_weight": 0.46849894523620605, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.3794738769531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2171203613281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15950, + "valid_token_count": 818.31640625 + }, + { + "actual_token_mask_ratio": 0.3005146920681, + "epoch": 4.643584521384929, + "flow_matching_loss": 0.012562212999910116, + "grad_norm": 0.08896191418170929, + "kept_token_count": 572.43515625, + "learning_rate": 1.465445537718591e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9928344130516052, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007165595702826977, + "mgd_loss_weight": 0.4681656062602997, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.5171203613281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.41631469726565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15960, + "valid_token_count": 818.37109375 + }, + { + "actual_token_mask_ratio": 0.29968206882476806, + "epoch": 4.646494035496072, + "flow_matching_loss": 0.011189950630068779, + "grad_norm": 0.08642589300870895, + "kept_token_count": 573.115625, + "learning_rate": 1.463792528012012e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.993287181854248, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006712819775566458, + "mgd_loss_weight": 0.46783226132392886, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.5867614746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.20927734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15970, + "valid_token_count": 818.36640625 + }, + { + "actual_token_mask_ratio": 0.2997221201658249, + "epoch": 4.649403549607215, + "flow_matching_loss": 0.014823956787586213, + "grad_norm": 0.06093735247850418, + "kept_token_count": 573.06953125, + "learning_rate": 1.4621395623009252e-05, + "loss": 0.018, + "mgd_cos_sim": 0.9932086884975433, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006791300838813186, + "mgd_loss_weight": 0.46749891340732574, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.7633758544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.37425231933594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15980, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.30018429160118104, + "epoch": 4.652313063718359, + "flow_matching_loss": 0.013131448253989219, + "grad_norm": 0.08398871123790741, + "kept_token_count": 572.571875, + "learning_rate": 1.4604866425938407e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9933109819889069, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006689021596685052, + "mgd_loss_weight": 0.4671655744314194, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.0786651611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5387847900391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 15990, + "valid_token_count": 818.17734375 + }, + { + "actual_token_mask_ratio": 0.3002839595079422, + "epoch": 4.655222577829503, + "flow_matching_loss": 0.012120956275612115, + "grad_norm": 0.07778912782669067, + "kept_token_count": 572.590625, + "learning_rate": 1.4588337708992115e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9932941734790802, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00670584561303258, + "mgd_loss_weight": 0.46683222949504855, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.77294921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.92481384277346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16000, + "valid_token_count": 818.32109375 + }, + { + "actual_token_mask_ratio": 0.29971585273742674, + "epoch": 4.6581320919406455, + "flow_matching_loss": 0.012515812646597624, + "grad_norm": 0.07746077328920364, + "kept_token_count": 573.14296875, + "learning_rate": 1.4571809492254318e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9932379066944123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006762099778279662, + "mgd_loss_weight": 0.46649888157844543, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.01491088867186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4838439941406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16010, + "valid_token_count": 818.44375 + }, + { + "actual_token_mask_ratio": 0.30043033361434934, + "epoch": 4.661041606051789, + "flow_matching_loss": 0.013662509061396122, + "grad_norm": 0.08285059034824371, + "kept_token_count": 572.5015625, + "learning_rate": 1.4555281795808371e-05, + "loss": 0.0168, + "mgd_cos_sim": 0.9933058500289917, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006694151507690549, + "mgd_loss_weight": 0.46616554260253906, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.30555725097656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6259063720703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16020, + "valid_token_count": 818.3671875 + }, + { + "actual_token_mask_ratio": 0.3004655748605728, + "epoch": 4.663951120162933, + "flow_matching_loss": 0.012622045073658227, + "grad_norm": 0.09929704666137695, + "kept_token_count": 572.46328125, + "learning_rate": 1.4538754639736966e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9932900607585907, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006709938356652856, + "mgd_loss_weight": 0.4658321917057037, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.3243804931641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1504241943359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16030, + "valid_token_count": 818.35234375 + }, + { + "actual_token_mask_ratio": 0.2994785219430923, + "epoch": 4.666860634274077, + "flow_matching_loss": 0.010866085719317197, + "grad_norm": 0.10465633124113083, + "kept_token_count": 573.28671875, + "learning_rate": 1.4522228044122174e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9934554994106293, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006544504035264254, + "mgd_loss_weight": 0.4654988497495651, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.6685852050781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.430615234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16040, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.3001690745353699, + "epoch": 4.669770148385219, + "flow_matching_loss": 0.013351953588426113, + "grad_norm": 0.0819782018661499, + "kept_token_count": 572.60859375, + "learning_rate": 1.4505702029045359e-05, + "loss": 0.0166, + "mgd_cos_sim": 0.9929308235645294, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007069167727604508, + "mgd_loss_weight": 0.46516551077365875, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.7279327392578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.38933715820315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16050, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.3002977192401886, + "epoch": 4.672679662496363, + "flow_matching_loss": 0.012931568920612336, + "grad_norm": 0.0842650756239891, + "kept_token_count": 572.68359375, + "learning_rate": 1.4489176614587189e-05, + "loss": 0.016, + "mgd_cos_sim": 0.9933119416236877, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006688065640628338, + "mgd_loss_weight": 0.4648321568965912, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.56281127929685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9859924316406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16060, + "valid_token_count": 818.4703125 + }, + { + "actual_token_mask_ratio": 0.299625763297081, + "epoch": 4.675589176607507, + "flow_matching_loss": 0.013058784045279026, + "grad_norm": 0.09519638866186142, + "kept_token_count": 573.05390625, + "learning_rate": 1.4472651820827595e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9934052944183349, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006594721972942352, + "mgd_loss_weight": 0.4644988179206848, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.23151245117185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1654907226563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16070, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.30039693117141725, + "epoch": 4.67849869071865, + "flow_matching_loss": 0.012331369798630476, + "grad_norm": 0.07619866728782654, + "kept_token_count": 572.36640625, + "learning_rate": 1.4456127667845766e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9930675148963928, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0069324867334216835, + "mgd_loss_weight": 0.464165472984314, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.5776397705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.56116943359376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16080, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.30022285878658295, + "epoch": 4.681408204829793, + "flow_matching_loss": 0.012502806074917317, + "grad_norm": 0.0770651176571846, + "kept_token_count": 572.42734375, + "learning_rate": 1.4439604175720104e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9933939635753631, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006606037588790059, + "mgd_loss_weight": 0.4638321250677109, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 445.2659149169922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4233093261719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16090, + "valid_token_count": 818.01640625 + }, + { + "actual_token_mask_ratio": 0.3004302144050598, + "epoch": 4.684317718940937, + "flow_matching_loss": 0.012108787149190902, + "grad_norm": 0.08050072193145752, + "kept_token_count": 572.53203125, + "learning_rate": 1.4423081364528212e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9930678367614746, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006932174181565642, + "mgd_loss_weight": 0.4634987860918045, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.8462280273437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.479052734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16100, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.29992382824420927, + "epoch": 4.687227233052081, + "flow_matching_loss": 0.012308114767074585, + "grad_norm": 0.09060937911272049, + "kept_token_count": 572.68515625, + "learning_rate": 1.4406559254346863e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9933580935001374, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006641904916614294, + "mgd_loss_weight": 0.4631654381752014, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.57838134765626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4106842041016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16110, + "valid_token_count": 818.03515625 + }, + { + "actual_token_mask_ratio": 0.2994468867778778, + "epoch": 4.690136747163224, + "flow_matching_loss": 0.011979542393237352, + "grad_norm": 0.07752751559019089, + "kept_token_count": 573.171875, + "learning_rate": 1.4390037865251978e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9933607399463653, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006639253068715334, + "mgd_loss_weight": 0.46283209323883057, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.09588928222655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9741668701172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16120, + "valid_token_count": 818.16953125 + }, + { + "actual_token_mask_ratio": 0.3001829355955124, + "epoch": 4.693046261274367, + "flow_matching_loss": 0.012387946434319019, + "grad_norm": 0.09916511178016663, + "kept_token_count": 572.7734375, + "learning_rate": 1.4373517217318597e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9935510814189911, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006448911037296057, + "mgd_loss_weight": 0.4624987542629242, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.19403076171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9094512939453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16130, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.2996911883354187, + "epoch": 4.69595577538551, + "flow_matching_loss": 0.011528560053557157, + "grad_norm": 0.11038928478956223, + "kept_token_count": 572.9109375, + "learning_rate": 1.4356997330620875e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9934575200080872, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006542479852214456, + "mgd_loss_weight": 0.46216540038585663, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.4366455078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.77533569335935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16140, + "valid_token_count": 818.0828125 + }, + { + "actual_token_mask_ratio": 0.29945737719535825, + "epoch": 4.698865289496654, + "flow_matching_loss": 0.012110690772533416, + "grad_norm": 0.09191429615020752, + "kept_token_count": 573.371875, + "learning_rate": 1.434047822523203e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9930039405822754, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006996062211692333, + "mgd_loss_weight": 0.46183206140995026, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.33974304199216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4885681152344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16150, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.2988515138626099, + "epoch": 4.7017748036077975, + "flow_matching_loss": 0.013986138626933098, + "grad_norm": 0.08160930126905441, + "kept_token_count": 573.7203125, + "learning_rate": 1.4323959921224322e-05, + "loss": 0.017, + "mgd_cos_sim": 0.9935709536075592, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006429035635665059, + "mgd_loss_weight": 0.46149871945381166, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.73659057617186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0580291748047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16160, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.30002023577690123, + "epoch": 4.704684317718941, + "flow_matching_loss": 0.011984542477875948, + "grad_norm": 0.0763290598988533, + "kept_token_count": 572.5234375, + "learning_rate": 1.4307442438669066e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9932670354843139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006732970476150513, + "mgd_loss_weight": 0.4611653685569763, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.2975708007813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6650787353516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16170, + "valid_token_count": 817.91640625 + }, + { + "actual_token_mask_ratio": 0.2988815188407898, + "epoch": 4.707593831830084, + "flow_matching_loss": 0.01234076488763094, + "grad_norm": 0.06931036710739136, + "kept_token_count": 573.63046875, + "learning_rate": 1.4290925797636546e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9932154595851899, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006784551683813334, + "mgd_loss_weight": 0.46083202958106995, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.4754241943359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8811950683594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16180, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.3006844907999039, + "epoch": 4.710503345941228, + "flow_matching_loss": 0.01261787572875619, + "grad_norm": 0.07080492377281189, + "kept_token_count": 572.159375, + "learning_rate": 1.4274410018196039e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9932655930519104, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006734411651268602, + "mgd_loss_weight": 0.46049868166446684, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4703735351562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.4855651855469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16190, + "valid_token_count": 818.17265625 + }, + { + "actual_token_mask_ratio": 0.300355526804924, + "epoch": 4.7134128600523715, + "flow_matching_loss": 0.013218318950384856, + "grad_norm": 0.07311806827783585, + "kept_token_count": 572.5875, + "learning_rate": 1.4257895120415783e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9934831380844116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00651685050688684, + "mgd_loss_weight": 0.460165336728096, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.18467102050784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.78688049316406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16200, + "valid_token_count": 818.39765625 + }, + { + "actual_token_mask_ratio": 0.2992302805185318, + "epoch": 4.716322374163514, + "flow_matching_loss": 0.013080334337428213, + "grad_norm": 0.09127897024154663, + "kept_token_count": 573.3828125, + "learning_rate": 1.4241381124362933e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9928214490413666, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007178551563993096, + "mgd_loss_weight": 0.45983199775218964, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.4558959960938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.11623229980466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16210, + "valid_token_count": 818.21484375 + }, + { + "actual_token_mask_ratio": 0.29972806572914124, + "epoch": 4.719231888274658, + "flow_matching_loss": 0.013992490526288748, + "grad_norm": 0.09641782194375992, + "kept_token_count": 572.9625, + "learning_rate": 1.4224868050103546e-05, + "loss": 0.0169, + "mgd_cos_sim": 0.9936200201511383, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006379981245845556, + "mgd_loss_weight": 0.4594986468553543, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.7744567871094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3355651855469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16220, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.300644850730896, + "epoch": 4.722141402385802, + "flow_matching_loss": 0.011736665666103364, + "grad_norm": 0.0717344656586647, + "kept_token_count": 572.21953125, + "learning_rate": 1.4208355917702573e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9929079115390778, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007092089112848044, + "mgd_loss_weight": 0.4591653048992157, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.1231750488281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.76051330566406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16230, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.3002266764640808, + "epoch": 4.725050916496945, + "flow_matching_loss": 0.012493746168911458, + "grad_norm": 0.08885765820741653, + "kept_token_count": 572.6078125, + "learning_rate": 1.41918447472238e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9934039235115051, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006596073601394891, + "mgd_loss_weight": 0.4588319659233093, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.70591125488284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8997100830078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16240, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.3001076549291611, + "epoch": 4.727960430608088, + "flow_matching_loss": 0.012815587688237429, + "grad_norm": 0.08540738373994827, + "kept_token_count": 572.45078125, + "learning_rate": 1.4175334558729869e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9931783616542816, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006821643700823188, + "mgd_loss_weight": 0.45849861204624176, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.5470275878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8923614501953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16250, + "valid_token_count": 817.90859375 + }, + { + "actual_token_mask_ratio": 0.29973170459270476, + "epoch": 4.730869944719232, + "flow_matching_loss": 0.013221720233559608, + "grad_norm": 0.10191946476697922, + "kept_token_count": 573.1078125, + "learning_rate": 1.4158825372282211e-05, + "loss": 0.0162, + "mgd_cos_sim": 0.9934366405010223, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00656335186213255, + "mgd_loss_weight": 0.4581652730703354, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.4782470703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6090362548828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16260, + "valid_token_count": 818.41171875 + }, + { + "actual_token_mask_ratio": 0.2996144384145737, + "epoch": 4.733779458830376, + "flow_matching_loss": 0.011884414637461305, + "grad_norm": 0.07761305570602417, + "kept_token_count": 573.0734375, + "learning_rate": 1.4142317207941042e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9932057619094848, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006794231059029698, + "mgd_loss_weight": 0.4578319311141968, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.9206787109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.7123199462891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16270, + "valid_token_count": 818.22265625 + }, + { + "actual_token_mask_ratio": 0.29922776520252226, + "epoch": 4.7366889729415185, + "flow_matching_loss": 0.011824478767812252, + "grad_norm": 0.05125740170478821, + "kept_token_count": 573.340625, + "learning_rate": 1.4125810085765339e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9937685430049896, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00623145243152976, + "mgd_loss_weight": 0.45749858021736145, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.4540710449219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.36118469238284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16280, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.3002872735261917, + "epoch": 4.739598487052662, + "flow_matching_loss": 0.01408302066847682, + "grad_norm": 0.07823308557271957, + "kept_token_count": 572.56484375, + "learning_rate": 1.410930402581281e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9933787167072297, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006621281057596207, + "mgd_loss_weight": 0.4571652412414551, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.6672882080078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.26719055175784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16290, + "valid_token_count": 818.29140625 + }, + { + "actual_token_mask_ratio": 0.30025648772716523, + "epoch": 4.742508001163806, + "flow_matching_loss": 0.012161599472165108, + "grad_norm": 0.09234154969453812, + "kept_token_count": 572.6015625, + "learning_rate": 1.4092799048139876e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9931560516357422, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006843950692564249, + "mgd_loss_weight": 0.45683189332485197, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.31642761230466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.442822265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16300, + "valid_token_count": 818.30234375 + }, + { + "actual_token_mask_ratio": 0.3000546008348465, + "epoch": 4.745417515274949, + "flow_matching_loss": 0.012368166726082564, + "grad_norm": 0.07982175052165985, + "kept_token_count": 572.79296875, + "learning_rate": 1.4076295172801647e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9931549310684205, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006845080619677901, + "mgd_loss_weight": 0.45649854838848114, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.46363220214846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.35039978027345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16310, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.3000710874795914, + "epoch": 4.748327029386092, + "flow_matching_loss": 0.01152330357581377, + "grad_norm": 0.0744452103972435, + "kept_token_count": 572.66796875, + "learning_rate": 1.4059792419851884e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9931939542293549, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068060555029660465, + "mgd_loss_weight": 0.45616520941257477, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.9116668701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0226715087891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16320, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.29978119432926176, + "epoch": 4.751236543497236, + "flow_matching_loss": 0.013315415661782026, + "grad_norm": 0.08993156254291534, + "kept_token_count": 572.9859375, + "learning_rate": 1.4043290809342984e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9929073214530945, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007092675426974893, + "mgd_loss_weight": 0.45583185851573943, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.1176727294922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4809112548828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16330, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.2993932098150253, + "epoch": 4.754146057608379, + "flow_matching_loss": 0.011816976498812437, + "grad_norm": 0.07741888612508774, + "kept_token_count": 573.12578125, + "learning_rate": 1.4026790361325958e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9931657195091248, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006834278907626867, + "mgd_loss_weight": 0.45549851655960083, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.86582946777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.4958160400391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16340, + "valid_token_count": 818.0421875 + }, + { + "actual_token_mask_ratio": 0.29958802461624146, + "epoch": 4.757055571719523, + "flow_matching_loss": 0.012151344120502472, + "grad_norm": 0.07780171185731888, + "kept_token_count": 573.10703125, + "learning_rate": 1.4010291095850415e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9932749271392822, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00672506419941783, + "mgd_loss_weight": 0.45516517460346223, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.6850128173828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5661560058594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16350, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.3002218008041382, + "epoch": 4.759965085830666, + "flow_matching_loss": 0.012472625728696585, + "grad_norm": 0.07833854109048843, + "kept_token_count": 572.45078125, + "learning_rate": 1.3993793032964515e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9931348741054535, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006865134881809354, + "mgd_loss_weight": 0.4548318237066269, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.4406799316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.85162353515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16360, + "valid_token_count": 818.053125 + }, + { + "actual_token_mask_ratio": 0.30008129179477694, + "epoch": 4.76287459994181, + "flow_matching_loss": 0.012480409443378448, + "grad_norm": 0.08570275455713272, + "kept_token_count": 572.7671875, + "learning_rate": 1.3977296192714956e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9931250631809234, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006874926062300802, + "mgd_loss_weight": 0.4544984847307205, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.5740570068359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.37851867675784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16370, + "valid_token_count": 818.33984375 + }, + { + "actual_token_mask_ratio": 0.30026374757289886, + "epoch": 4.765784114052953, + "flow_matching_loss": 0.01184499440714717, + "grad_norm": 0.0933246836066246, + "kept_token_count": 572.63984375, + "learning_rate": 1.396080059514696e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9932572543621063, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006742761190980673, + "mgd_loss_weight": 0.4541651397943497, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.8671936035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.70519104003904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16380, + "valid_token_count": 818.36484375 + }, + { + "actual_token_mask_ratio": 0.3000197857618332, + "epoch": 4.768693628164097, + "flow_matching_loss": 0.014011015091091395, + "grad_norm": 0.09152079373598099, + "kept_token_count": 572.8328125, + "learning_rate": 1.3944306260304232e-05, + "loss": 0.0171, + "mgd_cos_sim": 0.9931872725486756, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006812719348818064, + "mgd_loss_weight": 0.4538317918777466, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.9839172363281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3308074951172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16390, + "valid_token_count": 818.35703125 + }, + { + "actual_token_mask_ratio": 0.3000481963157654, + "epoch": 4.77160314227524, + "flow_matching_loss": 0.012554586213082076, + "grad_norm": 0.07705258578062057, + "kept_token_count": 572.6078125, + "learning_rate": 1.3927813208228936e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9935699522495269, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00643005408346653, + "mgd_loss_weight": 0.4534984529018402, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.88820190429686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0607055664062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16400, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.3002108782529831, + "epoch": 4.774512656386383, + "flow_matching_loss": 0.011391580291092395, + "grad_norm": 0.07847560197114944, + "kept_token_count": 572.68046875, + "learning_rate": 1.39113214589617e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.993819260597229, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006180737446993589, + "mgd_loss_weight": 0.45316510200500487, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.0176666259766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0819366455078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16410, + "valid_token_count": 818.36015625 + }, + { + "actual_token_mask_ratio": 0.3003930956125259, + "epoch": 4.777422170497527, + "flow_matching_loss": 0.011087426077574491, + "grad_norm": 0.07924111187458038, + "kept_token_count": 572.4390625, + "learning_rate": 1.3894831032541545e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9934871315956115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006512857181951404, + "mgd_loss_weight": 0.45283176004886627, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.41075134277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9311004638672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16420, + "valid_token_count": 818.22734375 + }, + { + "actual_token_mask_ratio": 0.29996084570884707, + "epoch": 4.7803316846086705, + "flow_matching_loss": 0.012168167252093553, + "grad_norm": 0.060946471989154816, + "kept_token_count": 572.69609375, + "learning_rate": 1.3878341949005895e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9934091329574585, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006590870441868901, + "mgd_loss_weight": 0.4524984210729599, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.26995239257815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6953094482422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16430, + "valid_token_count": 818.09296875 + }, + { + "actual_token_mask_ratio": 0.3002753764390945, + "epoch": 4.783241198719814, + "flow_matching_loss": 0.012411212548613548, + "grad_norm": 0.0676538273692131, + "kept_token_count": 572.4078125, + "learning_rate": 1.3861854228390546e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9928914606571198, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007108537247404456, + "mgd_loss_weight": 0.45216506719589233, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.49802551269534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2646820068359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16440, + "valid_token_count": 818.05390625 + }, + { + "actual_token_mask_ratio": 0.2998109430074692, + "epoch": 4.786150712830957, + "flow_matching_loss": 0.012653942499309779, + "grad_norm": 0.0825524553656578, + "kept_token_count": 572.91640625, + "learning_rate": 1.3845367890729621e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9932547092437745, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006745292292907834, + "mgd_loss_weight": 0.45183172821998596, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.8672576904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2619598388672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16450, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.2993762940168381, + "epoch": 4.789060226942101, + "flow_matching_loss": 0.012559075001627206, + "grad_norm": 0.09554509818553925, + "kept_token_count": 573.165625, + "learning_rate": 1.3828882956055591e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9931775629520416, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006822447199374438, + "mgd_loss_weight": 0.45149838626384736, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.6578308105469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5383605957031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16460, + "valid_token_count": 818.08359375 + }, + { + "actual_token_mask_ratio": 0.29927085638046264, + "epoch": 4.7919697410532445, + "flow_matching_loss": 0.011203332711011172, + "grad_norm": 0.0687514916062355, + "kept_token_count": 573.44453125, + "learning_rate": 1.38123994443992e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9932349920272827, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00676499824039638, + "mgd_loss_weight": 0.451165035367012, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.2138946533203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3865600585938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16470, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.29985358715057375, + "epoch": 4.794879255164387, + "flow_matching_loss": 0.01146317645907402, + "grad_norm": 0.08528267592191696, + "kept_token_count": 572.91484375, + "learning_rate": 1.3795917375789465e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.992973518371582, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007026471244171262, + "mgd_loss_weight": 0.45083169639110565, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.42669067382815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6881896972656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16480, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.3003442347049713, + "epoch": 4.797788769275531, + "flow_matching_loss": 0.012939125439152122, + "grad_norm": 0.0683806985616684, + "kept_token_count": 572.42109375, + "learning_rate": 1.3779436770253658e-05, + "loss": 0.016, + "mgd_cos_sim": 0.993252032995224, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006747981114313006, + "mgd_loss_weight": 0.4504983514547348, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.1897918701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.41298828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16490, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.30014302432537077, + "epoch": 4.800698283386675, + "flow_matching_loss": 0.011759018898010254, + "grad_norm": 0.07742315530776978, + "kept_token_count": 572.78359375, + "learning_rate": 1.3762957647817258e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9933395028114319, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006660486897453665, + "mgd_loss_weight": 0.4501650035381317, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.67552795410154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.43231506347655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16500, + "valid_token_count": 818.4296875 + }, + { + "actual_token_mask_ratio": 0.29953293800354003, + "epoch": 4.8036077974978175, + "flow_matching_loss": 0.011420516157522797, + "grad_norm": 0.06848544627428055, + "kept_token_count": 573.275, + "learning_rate": 1.374648002850397e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9930696725845337, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006930327275767922, + "mgd_loss_weight": 0.44983166456222534, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.9450958251953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.03932800292966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16510, + "valid_token_count": 818.41796875 + }, + { + "actual_token_mask_ratio": 0.2993212312459946, + "epoch": 4.806517311608961, + "flow_matching_loss": 0.01107998499646783, + "grad_norm": 0.07675747573375702, + "kept_token_count": 573.1578125, + "learning_rate": 1.3730003932335642e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9930935382843018, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006906450306996703, + "mgd_loss_weight": 0.44949831366539, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.17604370117186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.14576416015626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16520, + "valid_token_count": 818.00546875 + }, + { + "actual_token_mask_ratio": 0.30021447539329527, + "epoch": 4.809426825720105, + "flow_matching_loss": 0.011026273062452675, + "grad_norm": 0.0606725849211216, + "kept_token_count": 572.5125, + "learning_rate": 1.3713529379332292e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9931779205799103, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00682208901271224, + "mgd_loss_weight": 0.4491649717092514, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.6265502929688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6884796142578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16530, + "valid_token_count": 818.12734375 + }, + { + "actual_token_mask_ratio": 0.299978107213974, + "epoch": 4.812336339831248, + "flow_matching_loss": 0.011692937463521957, + "grad_norm": 0.07973097264766693, + "kept_token_count": 572.86875, + "learning_rate": 1.3697056389512054e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.993205726146698, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00679428344592452, + "mgd_loss_weight": 0.44883163273334503, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.261572265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0089477539062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16540, + "valid_token_count": 818.35703125 + }, + { + "actual_token_mask_ratio": 0.30014912486076356, + "epoch": 4.8152458539423915, + "flow_matching_loss": 0.012295865546911955, + "grad_norm": 0.08655986934900284, + "kept_token_count": 572.6328125, + "learning_rate": 1.3680584982891162e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9935067296028137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006493267556652427, + "mgd_loss_weight": 0.44849827885627747, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.27671813964844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7575378417969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16550, + "valid_token_count": 818.22265625 + }, + { + "actual_token_mask_ratio": 0.2996053248643875, + "epoch": 4.818155368053535, + "flow_matching_loss": 0.010789657896384596, + "grad_norm": 0.08361281454563141, + "kept_token_count": 573.084375, + "learning_rate": 1.3664115179483921e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9934114336967468, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006588573334738612, + "mgd_loss_weight": 0.4481649398803711, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.2659851074219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7091033935547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16560, + "valid_token_count": 818.23203125 + }, + { + "actual_token_mask_ratio": 0.30011979937553407, + "epoch": 4.821064882164679, + "flow_matching_loss": 0.011981652118265629, + "grad_norm": 0.06519795209169388, + "kept_token_count": 572.72265625, + "learning_rate": 1.364764699930271e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9934045314788819, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006595477415248752, + "mgd_loss_weight": 0.44783159494400027, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.8911651611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.67568054199216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16570, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.3005412817001343, + "epoch": 4.823974396275822, + "flow_matching_loss": 0.01089320071041584, + "grad_norm": 0.06692538410425186, + "kept_token_count": 572.35546875, + "learning_rate": 1.363118046235791e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9934662342071533, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006533764628693461, + "mgd_loss_weight": 0.44749824702739716, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.7175567626953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.57615051269534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16580, + "valid_token_count": 818.28828125 + }, + { + "actual_token_mask_ratio": 0.30081687569618226, + "epoch": 4.826883910386965, + "flow_matching_loss": 0.011449919687584043, + "grad_norm": 0.07066594809293747, + "kept_token_count": 572.26171875, + "learning_rate": 1.3614715588657921e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9936329841613769, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006367026455700397, + "mgd_loss_weight": 0.4471649080514908, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.3293426513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.898681640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16590, + "valid_token_count": 818.48125 + }, + { + "actual_token_mask_ratio": 0.30033851861953736, + "epoch": 4.829793424498109, + "flow_matching_loss": 0.01234765313565731, + "grad_norm": 0.08549607545137405, + "kept_token_count": 572.646875, + "learning_rate": 1.3598252398209118e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9931731879711151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006826802203431725, + "mgd_loss_weight": 0.4468315601348877, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.7421081542969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.40257568359374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16600, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.3006173551082611, + "epoch": 4.832702938609252, + "flow_matching_loss": 0.012998578511178494, + "grad_norm": 0.08767460286617279, + "kept_token_count": 572.2890625, + "learning_rate": 1.358179091101582e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.993129414319992, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006870587728917599, + "mgd_loss_weight": 0.44649821519851685, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.7722564697266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4086395263672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16610, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.30018663704395293, + "epoch": 4.835612452720396, + "flow_matching_loss": 0.012205845396965741, + "grad_norm": 0.07661128044128418, + "kept_token_count": 572.4984375, + "learning_rate": 1.3565331147080299e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9935067594051361, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006493243901059031, + "mgd_loss_weight": 0.4461648762226105, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.67775268554686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6737518310547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16620, + "valid_token_count": 818.07578125 + }, + { + "actual_token_mask_ratio": 0.3000066339969635, + "epoch": 4.838521966831539, + "flow_matching_loss": 0.013391916453838349, + "grad_norm": 0.07809526473283768, + "kept_token_count": 572.7765625, + "learning_rate": 1.3548873126402716e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9934657931327819, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065342225600034, + "mgd_loss_weight": 0.4458315223455429, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.43251647949216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.39794921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16630, + "valid_token_count": 818.26171875 + }, + { + "actual_token_mask_ratio": 0.29990596771240235, + "epoch": 4.841431480942683, + "flow_matching_loss": 0.011000571167096495, + "grad_norm": 0.08821288496255875, + "kept_token_count": 572.81171875, + "learning_rate": 1.3532416868981117e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9932344079017639, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067655963357537985, + "mgd_loss_weight": 0.44549818336963654, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 444.3020721435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2638397216797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16640, + "valid_token_count": 818.19609375 + }, + { + "actual_token_mask_ratio": 0.29981168210506437, + "epoch": 4.844340995053826, + "flow_matching_loss": 0.012822723481804132, + "grad_norm": 0.08313226699829102, + "kept_token_count": 573.01328125, + "learning_rate": 1.3515962394811401e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.993256539106369, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067434684373438355, + "mgd_loss_weight": 0.44516484141349794, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.4434844970703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.83684692382815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16650, + "valid_token_count": 818.36640625 + }, + { + "actual_token_mask_ratio": 0.30070944130420685, + "epoch": 4.84725050916497, + "flow_matching_loss": 0.012706278916448354, + "grad_norm": 0.08066537231206894, + "kept_token_count": 572.1390625, + "learning_rate": 1.3499509723887313e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9933461308479309, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006653881771489978, + "mgd_loss_weight": 0.4448314905166626, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.3835784912109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6965637207031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16660, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.300653350353241, + "epoch": 4.850160023276112, + "flow_matching_loss": 0.011761203221976758, + "grad_norm": 0.08039364963769913, + "kept_token_count": 572.0671875, + "learning_rate": 1.3483058876200396e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.993371993303299, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066280055325478315, + "mgd_loss_weight": 0.4444981515407562, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.0567108154297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.76817016601564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16670, + "valid_token_count": 818.00703125 + }, + { + "actual_token_mask_ratio": 0.3002068907022476, + "epoch": 4.853069537387256, + "flow_matching_loss": 0.012603904260322452, + "grad_norm": 0.07050485908985138, + "kept_token_count": 572.71171875, + "learning_rate": 1.3466609871739987e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9934465289115906, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065534652676433325, + "mgd_loss_weight": 0.4441648066043854, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.0172637939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.46922912597654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16680, + "valid_token_count": 818.40234375 + }, + { + "actual_token_mask_ratio": 0.2998716026544571, + "epoch": 4.8559790514984, + "flow_matching_loss": 0.01227091597393155, + "grad_norm": 0.09775448590517044, + "kept_token_count": 572.96484375, + "learning_rate": 1.3450162730493172e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9930366694927215, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006963323708623648, + "mgd_loss_weight": 0.4438314586877823, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.3570129394531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9916473388672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16690, + "valid_token_count": 818.37421875 + }, + { + "actual_token_mask_ratio": 0.3004028230905533, + "epoch": 4.8588885656095435, + "flow_matching_loss": 0.010524480883032084, + "grad_norm": 0.06588611006736755, + "kept_token_count": 572.5453125, + "learning_rate": 1.3433717472444781e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9938294529914856, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006170565215870738, + "mgd_loss_weight": 0.4434981197118759, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.13941650390626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0035003662109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16700, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.30031857788562777, + "epoch": 4.861798079720686, + "flow_matching_loss": 0.012029508780688047, + "grad_norm": 0.07793506234884262, + "kept_token_count": 572.55625, + "learning_rate": 1.3417274117577343e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9934858083724976, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006514194561168552, + "mgd_loss_weight": 0.4431647717952728, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.93711853027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2969055175781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16710, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.3002261698246002, + "epoch": 4.86470759383183, + "flow_matching_loss": 0.01273010903969407, + "grad_norm": 0.09277822822332382, + "kept_token_count": 572.74609375, + "learning_rate": 1.3400832685871102e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9935362339019775, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006463749753311277, + "mgd_loss_weight": 0.442831426858902, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.2814025878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7796997070312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16720, + "valid_token_count": 818.47734375 + }, + { + "actual_token_mask_ratio": 0.3000770747661591, + "epoch": 4.867617107942974, + "flow_matching_loss": 0.013009998109191656, + "grad_norm": 0.08592884242534637, + "kept_token_count": 572.73359375, + "learning_rate": 1.3384393197303936e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9933826506137848, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006617347057908773, + "mgd_loss_weight": 0.4424980878829956, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.7985900878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1947387695312, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16730, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.3008958578109741, + "epoch": 4.870526622054117, + "flow_matching_loss": 0.012692293711006641, + "grad_norm": 0.09458069503307343, + "kept_token_count": 572.121875, + "learning_rate": 1.336795567185138e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9933232605457306, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066767530515789986, + "mgd_loss_weight": 0.44216473400592804, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.45789794921876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1932312011719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16740, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.3000503212213516, + "epoch": 4.87343613616526, + "flow_matching_loss": 0.012308239936828613, + "grad_norm": 0.09240193665027618, + "kept_token_count": 572.85859375, + "learning_rate": 1.3351520129486573e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9935599744319916, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006440024776384235, + "mgd_loss_weight": 0.44183139503002167, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.4901153564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1121368408203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16750, + "valid_token_count": 818.42890625 + }, + { + "actual_token_mask_ratio": 0.29951071441173555, + "epoch": 4.876345650276404, + "flow_matching_loss": 0.01282715741544962, + "grad_norm": 0.0822342187166214, + "kept_token_count": 573.0984375, + "learning_rate": 1.3335086590180252e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9932617664337158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006738227372989058, + "mgd_loss_weight": 0.44149805307388307, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.1359100341797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0069580078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16760, + "valid_token_count": 818.14140625 + }, + { + "actual_token_mask_ratio": 0.30012897253036497, + "epoch": 4.879255164387548, + "flow_matching_loss": 0.012852212134748697, + "grad_norm": 0.0713464692234993, + "kept_token_count": 572.53515625, + "learning_rate": 1.3318655073900708e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9934692621231079, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065307293087244036, + "mgd_loss_weight": 0.44116470217704773, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.6775726318359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0673583984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16770, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.30005245208740233, + "epoch": 4.8821646784986905, + "flow_matching_loss": 0.011284645879641175, + "grad_norm": 0.06247946619987488, + "kept_token_count": 572.73515625, + "learning_rate": 1.3302225600613796e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9937727928161622, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006227198289707303, + "mgd_loss_weight": 0.44083136320114136, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.76645812988284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.88587646484376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16780, + "valid_token_count": 818.25546875 + }, + { + "actual_token_mask_ratio": 0.29962331652641294, + "epoch": 4.885074192609834, + "flow_matching_loss": 0.011514701787382365, + "grad_norm": 0.07025009393692017, + "kept_token_count": 573.0640625, + "learning_rate": 1.3285798190282869e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9932765364646912, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006723454315215349, + "mgd_loss_weight": 0.44049801528453825, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.62673034667966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3140014648437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16790, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.30018010139465334, + "epoch": 4.887983706720978, + "flow_matching_loss": 0.012311032693833112, + "grad_norm": 0.12555290758609772, + "kept_token_count": 572.8015625, + "learning_rate": 1.3269372862868776e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9933085918426514, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066914194729179146, + "mgd_loss_weight": 0.4401646703481674, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.64295654296876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0154571533203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16800, + "valid_token_count": 818.50703125 + }, + { + "actual_token_mask_ratio": 0.30035037100315093, + "epoch": 4.890893220832121, + "flow_matching_loss": 0.012024249415844679, + "grad_norm": 0.10620759427547455, + "kept_token_count": 572.534375, + "learning_rate": 1.3252949638329845e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9937832057476044, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006216798489913344, + "mgd_loss_weight": 0.43983133137226105, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.92579956054686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7969177246094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16810, + "valid_token_count": 818.3125 + }, + { + "actual_token_mask_ratio": 0.29954310357570646, + "epoch": 4.8938027349432645, + "flow_matching_loss": 0.012935450114309788, + "grad_norm": 0.09423979371786118, + "kept_token_count": 573.175, + "learning_rate": 1.3236528536621834e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9932910323143005, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006708969781175256, + "mgd_loss_weight": 0.4394979804754257, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.55324401855466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.25457458496095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16820, + "valid_token_count": 818.28515625 + }, + { + "actual_token_mask_ratio": 0.30038563907146454, + "epoch": 4.896712249054408, + "flow_matching_loss": 0.012992214784026146, + "grad_norm": 0.08579979836940765, + "kept_token_count": 572.28984375, + "learning_rate": 1.3220109577697943e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9934118568897248, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006588148884475231, + "mgd_loss_weight": 0.4391646385192871, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.3559234619141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8360260009766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16830, + "valid_token_count": 818.01015625 + }, + { + "actual_token_mask_ratio": 0.300162199139595, + "epoch": 4.899621763165551, + "flow_matching_loss": 0.012470546551048756, + "grad_norm": 0.09257042407989502, + "kept_token_count": 572.61484375, + "learning_rate": 1.3203692781508745e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9932105123996735, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006789485784247517, + "mgd_loss_weight": 0.4388312965631485, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.81329650878905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.19679870605466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16840, + "valid_token_count": 818.2125 + }, + { + "actual_token_mask_ratio": 0.30037849247455595, + "epoch": 4.902531277276695, + "flow_matching_loss": 0.012927943561226129, + "grad_norm": 0.08826201409101486, + "kept_token_count": 572.55703125, + "learning_rate": 1.3187278168002198e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.993257462978363, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006742540188133717, + "mgd_loss_weight": 0.43849794566631317, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.88133544921874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0331695556641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16850, + "valid_token_count": 818.38359375 + }, + { + "actual_token_mask_ratio": 0.29898368418216703, + "epoch": 4.905440791387838, + "flow_matching_loss": 0.012107883859425784, + "grad_norm": 0.12211304157972336, + "kept_token_count": 573.6171875, + "learning_rate": 1.3170865757123606e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9929981470108032, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007001849263906479, + "mgd_loss_weight": 0.4381646066904068, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.08644714355466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9151580810547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16860, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.30017929077148436, + "epoch": 4.908350305498981, + "flow_matching_loss": 0.011274809623137117, + "grad_norm": 0.06716288626194, + "kept_token_count": 572.690625, + "learning_rate": 1.3154455568815592e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9935336828231811, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006466332077980042, + "mgd_loss_weight": 0.43783126175403597, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.08250122070314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3089263916016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16870, + "valid_token_count": 818.33828125 + }, + { + "actual_token_mask_ratio": 0.2997828841209412, + "epoch": 4.911259819610125, + "flow_matching_loss": 0.011899993009865283, + "grad_norm": 0.06992371380329132, + "kept_token_count": 572.9921875, + "learning_rate": 1.3138047623018087e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9932036697864532, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006796343205496669, + "mgd_loss_weight": 0.43749791383743286, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 446.7211120605469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.30323181152346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16880, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.30033729076385496, + "epoch": 4.914169333721269, + "flow_matching_loss": 0.012048439122736454, + "grad_norm": 0.08960863202810287, + "kept_token_count": 572.59375, + "learning_rate": 1.3121641939668293e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9932834684848786, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006716553028672934, + "mgd_loss_weight": 0.4371645748615265, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.5358520507813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1781890869141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16890, + "valid_token_count": 818.38359375 + }, + { + "actual_token_mask_ratio": 0.29996800422668457, + "epoch": 4.917078847832412, + "flow_matching_loss": 0.011878594942390918, + "grad_norm": 0.08247583359479904, + "kept_token_count": 572.77734375, + "learning_rate": 1.3105238538700659e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9933714926242828, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066285109613090755, + "mgd_loss_weight": 0.4368312269449234, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.88832397460936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.07761840820314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16900, + "valid_token_count": 818.22109375 + }, + { + "actual_token_mask_ratio": 0.30048351287841796, + "epoch": 4.919988361943555, + "flow_matching_loss": 0.010805554874241351, + "grad_norm": 0.07118869572877884, + "kept_token_count": 572.49765625, + "learning_rate": 1.3088837440046866e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9939546942710876, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006045302236452698, + "mgd_loss_weight": 0.43649788200855255, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.38359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.61302795410154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16910, + "valid_token_count": 818.4140625 + }, + { + "actual_token_mask_ratio": 0.3000202924013138, + "epoch": 4.922897876054699, + "flow_matching_loss": 0.012079306319355965, + "grad_norm": 0.07565565407276154, + "kept_token_count": 572.78984375, + "learning_rate": 1.3072438663635787e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9933960855007171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006603914592415094, + "mgd_loss_weight": 0.4361645430326462, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.36012878417966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1909423828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16920, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.300227677822113, + "epoch": 4.925807390165843, + "flow_matching_loss": 0.011795436032116413, + "grad_norm": 0.10728280246257782, + "kept_token_count": 572.7234375, + "learning_rate": 1.3056042229393497e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9931964576244354, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006803546147421002, + "mgd_loss_weight": 0.43583119213581084, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.90683898925784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.84864501953126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16930, + "valid_token_count": 818.4453125 + }, + { + "actual_token_mask_ratio": 0.30053046643733977, + "epoch": 4.928716904276985, + "flow_matching_loss": 0.012503734230995179, + "grad_norm": 0.10371660441160202, + "kept_token_count": 572.38515625, + "learning_rate": 1.30396481572432e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9932910025119781, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006709015555679798, + "mgd_loss_weight": 0.43549785017967224, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.12095336914064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0585906982422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16940, + "valid_token_count": 818.31640625 + }, + { + "actual_token_mask_ratio": 0.29980394542217254, + "epoch": 4.931626418388129, + "flow_matching_loss": 0.014584357850253582, + "grad_norm": 0.08995471149682999, + "kept_token_count": 573.15859375, + "learning_rate": 1.3023256467105237e-05, + "loss": 0.0176, + "mgd_cos_sim": 0.9931143701076508, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00688563478179276, + "mgd_loss_weight": 0.43516450822353364, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.3839508056641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.03794250488284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16950, + "valid_token_count": 818.56953125 + }, + { + "actual_token_mask_ratio": 0.30007613003253936, + "epoch": 4.934535932499273, + "flow_matching_loss": 0.011433184146881104, + "grad_norm": 0.08748006075620651, + "kept_token_count": 572.68984375, + "learning_rate": 1.3006867178897067e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.99345081448555, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006549183418974281, + "mgd_loss_weight": 0.4348311573266983, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.11878967285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6036895751953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16960, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.30071194767951964, + "epoch": 4.9374454466104165, + "flow_matching_loss": 0.011620813794434071, + "grad_norm": 0.08826171606779099, + "kept_token_count": 572.1953125, + "learning_rate": 1.2990480312533213e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.993377149105072, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00662283762358129, + "mgd_loss_weight": 0.43449781835079193, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.3715362548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.20953369140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16970, + "valid_token_count": 818.25703125 + }, + { + "actual_token_mask_ratio": 0.29960229992866516, + "epoch": 4.940354960721559, + "flow_matching_loss": 0.013402201607823372, + "grad_norm": 0.08263148367404938, + "kept_token_count": 572.95546875, + "learning_rate": 1.2974095887925258e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9932255566120147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006774427788332105, + "mgd_loss_weight": 0.4341644734144211, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.8260528564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.154541015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16980, + "valid_token_count": 818.04140625 + }, + { + "actual_token_mask_ratio": 0.30033257603645325, + "epoch": 4.943264474832703, + "flow_matching_loss": 0.011869464255869388, + "grad_norm": 0.07976467162370682, + "kept_token_count": 572.60546875, + "learning_rate": 1.2957713924981837e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9935114443302154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006488547381013632, + "mgd_loss_weight": 0.433831125497818, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.83112182617185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4316741943359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 16990, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.30028635561466216, + "epoch": 4.946173988943847, + "flow_matching_loss": 0.011676982883363963, + "grad_norm": 0.07303363084793091, + "kept_token_count": 572.4125, + "learning_rate": 1.2941334443608572e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9935567796230316, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006443223264068365, + "mgd_loss_weight": 0.4334977865219116, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.0375274658203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.51324157714845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17000, + "valid_token_count": 818.06953125 + }, + { + "actual_token_mask_ratio": 0.2998526573181152, + "epoch": 4.94908350305499, + "flow_matching_loss": 0.011642814986407757, + "grad_norm": 0.06062136963009834, + "kept_token_count": 572.7921875, + "learning_rate": 1.2924957463708073e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.993250447511673, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067495577037334446, + "mgd_loss_weight": 0.4331644356250763, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.622998046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1444946289063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17010, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.30066035091876986, + "epoch": 4.951993017166133, + "flow_matching_loss": 0.011996131530031562, + "grad_norm": 0.07944254577159882, + "kept_token_count": 572.14375, + "learning_rate": 1.2908583005179924e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9936107218265533, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006389285204932094, + "mgd_loss_weight": 0.4328310936689377, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.05181274414065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6179565429687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17020, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.2995680123567581, + "epoch": 4.954902531277277, + "flow_matching_loss": 0.013677500374615192, + "grad_norm": 0.06514975428581238, + "kept_token_count": 573.1078125, + "learning_rate": 1.2892211087920624e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9934065103530884, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006593493651598692, + "mgd_loss_weight": 0.4324977546930313, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.5293212890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9064086914062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17030, + "valid_token_count": 818.21953125 + }, + { + "actual_token_mask_ratio": 0.2998769968748093, + "epoch": 4.95781204538842, + "flow_matching_loss": 0.009646081505343318, + "grad_norm": 0.06479652971029282, + "kept_token_count": 572.828125, + "learning_rate": 1.287584173182361e-05, + "loss": 0.0123, + "mgd_cos_sim": 0.9938009262084961, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006199084594845772, + "mgd_loss_weight": 0.43216440081596375, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.14369506835936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9139862060547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17040, + "valid_token_count": 818.18046875 + }, + { + "actual_token_mask_ratio": 0.29968403577804564, + "epoch": 4.9607215594995635, + "flow_matching_loss": 0.012753046117722988, + "grad_norm": 0.0763673186302185, + "kept_token_count": 573.00234375, + "learning_rate": 1.2859474956779188e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9930484235286713, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006951581221073866, + "mgd_loss_weight": 0.4318310618400574, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.5096130371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.15025634765624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17050, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.2992506206035614, + "epoch": 4.963631073610707, + "flow_matching_loss": 0.012091160845011472, + "grad_norm": 0.07353969663381577, + "kept_token_count": 573.384375, + "learning_rate": 1.2843110782674532e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9932919502258301, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00670805424451828, + "mgd_loss_weight": 0.43149771690368655, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 445.26575622558596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.2957397460938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17060, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.30014955103397367, + "epoch": 4.96654058772185, + "flow_matching_loss": 0.01177356531843543, + "grad_norm": 0.0731678232550621, + "kept_token_count": 572.50390625, + "learning_rate": 1.2826749229393655e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.993741261959076, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006258750893175602, + "mgd_loss_weight": 0.43116436898708344, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.8611358642578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2457580566406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17070, + "valid_token_count": 818.03203125 + }, + { + "actual_token_mask_ratio": 0.2997934579849243, + "epoch": 4.969450101832994, + "flow_matching_loss": 0.012987113278359174, + "grad_norm": 0.07489422708749771, + "kept_token_count": 572.91171875, + "learning_rate": 1.2810390316817381e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9937679052352906, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00623210109770298, + "mgd_loss_weight": 0.43083103001117706, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.7417388916016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0715850830078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17080, + "valid_token_count": 818.20234375 + }, + { + "actual_token_mask_ratio": 0.29947575330734255, + "epoch": 4.9723596159441374, + "flow_matching_loss": 0.011572943441569805, + "grad_norm": 0.08122875541448593, + "kept_token_count": 573.2109375, + "learning_rate": 1.2794034064823346e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9930510759353638, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006948933470994234, + "mgd_loss_weight": 0.43049768209457395, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.03484802246095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5277038574219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17090, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.30055168867111204, + "epoch": 4.975269130055281, + "flow_matching_loss": 0.011632209829986095, + "grad_norm": 0.08358224481344223, + "kept_token_count": 572.221875, + "learning_rate": 1.277768049328593e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9935518682003022, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064481259789317845, + "mgd_loss_weight": 0.4301643371582031, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.1572296142578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8293090820313, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17100, + "valid_token_count": 818.10859375 + }, + { + "actual_token_mask_ratio": 0.29993903934955596, + "epoch": 4.978178644166424, + "flow_matching_loss": 0.012632730416953564, + "grad_norm": 0.08394969254732132, + "kept_token_count": 573.01796875, + "learning_rate": 1.2761329622076269e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9935030817985535, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006496914708986878, + "mgd_loss_weight": 0.42983099818229675, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.8036254882812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.02089538574216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17110, + "valid_token_count": 818.528125 + }, + { + "actual_token_mask_ratio": 0.30059532523155214, + "epoch": 4.981088158277568, + "flow_matching_loss": 0.011352666094899178, + "grad_norm": 0.09366204589605331, + "kept_token_count": 572.36328125, + "learning_rate": 1.274498147106221e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9937259137630463, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006274082371965051, + "mgd_loss_weight": 0.4294976472854614, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.53758239746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.68297119140624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17120, + "valid_token_count": 818.3609375 + }, + { + "actual_token_mask_ratio": 0.30010354816913604, + "epoch": 4.983997672388711, + "flow_matching_loss": 0.011129462625831366, + "grad_norm": 0.08015558868646622, + "kept_token_count": 572.51484375, + "learning_rate": 1.272863606010829e-05, + "loss": 0.014, + "mgd_cos_sim": 0.993398267030716, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006601725937798619, + "mgd_loss_weight": 0.4291643053293228, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.2905578613281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.36890869140626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17130, + "valid_token_count": 818.00234375 + }, + { + "actual_token_mask_ratio": 0.29965971112251283, + "epoch": 4.986907186499854, + "flow_matching_loss": 0.012419088184833527, + "grad_norm": 0.0826055258512497, + "kept_token_count": 573.35, + "learning_rate": 1.2712293409075739e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9928214251995087, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0071785831823945045, + "mgd_loss_weight": 0.42883096635341644, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.9442901611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.78941040039064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17140, + "valid_token_count": 818.6765625 + }, + { + "actual_token_mask_ratio": 0.3005156546831131, + "epoch": 4.989816700610998, + "flow_matching_loss": 0.012710033543407917, + "grad_norm": 0.08294828981161118, + "kept_token_count": 572.41875, + "learning_rate": 1.2695953537822413e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9931818902492523, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006818108726292849, + "mgd_loss_weight": 0.4284976124763489, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 445.32862548828126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3589172363281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17150, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.300219801068306, + "epoch": 4.992726214722142, + "flow_matching_loss": 0.013500041700899601, + "grad_norm": 0.08767163008451462, + "kept_token_count": 572.4234375, + "learning_rate": 1.2679616466202792e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9933460295200348, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006653964193537832, + "mgd_loss_weight": 0.4281642735004425, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.36708374023436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3595764160156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17160, + "valid_token_count": 818.00625 + }, + { + "actual_token_mask_ratio": 0.2994641661643982, + "epoch": 4.995635728833285, + "flow_matching_loss": 0.01445689331740141, + "grad_norm": 0.08343211561441422, + "kept_token_count": 573.3609375, + "learning_rate": 1.2663282214067967e-05, + "loss": 0.0173, + "mgd_cos_sim": 0.9934650838375092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006534917186945677, + "mgd_loss_weight": 0.4278309285640717, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.8107940673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4688323974609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17170, + "valid_token_count": 818.45703125 + }, + { + "actual_token_mask_ratio": 0.2997468262910843, + "epoch": 4.998545242944428, + "flow_matching_loss": 0.01353301601484418, + "grad_norm": 0.09265527129173279, + "kept_token_count": 572.890625, + "learning_rate": 1.264695080126559e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9936039268970489, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00639608264900744, + "mgd_loss_weight": 0.42749758064746857, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.6361877441406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.92958068847656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17180, + "valid_token_count": 818.11640625 + }, + { + "actual_token_mask_ratio": 0.2995703160762787, + "epoch": 5.001454757055572, + "flow_matching_loss": 0.011142018716782331, + "grad_norm": 0.08565164357423782, + "kept_token_count": 573.1652526855469, + "learning_rate": 1.263062224763986e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9936172127723694, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006382805528119206, + "mgd_loss_weight": 0.4271642416715622, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.23427734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5162780761719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17190, + "valid_token_count": 818.3054016113281 + }, + { + "actual_token_mask_ratio": 0.30024364292621614, + "epoch": 5.004364271166716, + "flow_matching_loss": 0.012541995849460363, + "grad_norm": 0.08947587013244629, + "kept_token_count": 572.6265625, + "learning_rate": 1.2614296573031522e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9935461223125458, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006453886302188039, + "mgd_loss_weight": 0.4268308937549591, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.2087646484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.96214599609374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17200, + "valid_token_count": 818.32109375 + }, + { + "actual_token_mask_ratio": 0.29992645382881167, + "epoch": 5.007273785277858, + "flow_matching_loss": 0.01313941078260541, + "grad_norm": 0.08212016522884369, + "kept_token_count": 572.8984375, + "learning_rate": 1.259797379727781e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.993707925081253, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006292074033990502, + "mgd_loss_weight": 0.42649754881858826, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.96571655273436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1483489990234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17210, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.3000481158494949, + "epoch": 5.010183299389002, + "flow_matching_loss": 0.011235506739467382, + "grad_norm": 0.08836060017347336, + "kept_token_count": 572.77109375, + "learning_rate": 1.2581653940212423e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9930597484111786, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006940259970724583, + "mgd_loss_weight": 0.4261642098426819, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.8246185302734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.92763366699216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17220, + "valid_token_count": 818.30703125 + }, + { + "actual_token_mask_ratio": 0.3000279009342194, + "epoch": 5.013092813500146, + "flow_matching_loss": 0.012393777258694172, + "grad_norm": 0.08010286837816238, + "kept_token_count": 572.8015625, + "learning_rate": 1.2565337021665537e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9935609757900238, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00643901638686657, + "mgd_loss_weight": 0.4258308559656143, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.5727783203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.93182373046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17230, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.29989725053310395, + "epoch": 5.016002327611289, + "flow_matching_loss": 0.012195479311048984, + "grad_norm": 0.0764104574918747, + "kept_token_count": 572.92265625, + "learning_rate": 1.2549023061463739e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9937021732330322, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006297819968312979, + "mgd_loss_weight": 0.42549751698970795, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.1212982177734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.70179443359376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17240, + "valid_token_count": 818.33671875 + }, + { + "actual_token_mask_ratio": 0.30000223219394684, + "epoch": 5.018911841722432, + "flow_matching_loss": 0.011804243642836809, + "grad_norm": 0.08387462049722672, + "kept_token_count": 572.92734375, + "learning_rate": 1.2532712079430038e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9937245368957519, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00627547837793827, + "mgd_loss_weight": 0.42516417503356935, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.54786071777346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.83161010742185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17250, + "valid_token_count": 818.46796875 + }, + { + "actual_token_mask_ratio": 0.2994850784540176, + "epoch": 5.021821355833576, + "flow_matching_loss": 0.01351031931117177, + "grad_norm": 0.108846515417099, + "kept_token_count": 573.4015625, + "learning_rate": 1.2516404095383813e-05, + "loss": 0.0164, + "mgd_cos_sim": 0.9931441366672515, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0068558762781322, + "mgd_loss_weight": 0.424830824136734, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.24627685546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.25157775878904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17260, + "valid_token_count": 818.53828125 + }, + { + "actual_token_mask_ratio": 0.29964317977428434, + "epoch": 5.024730869944719, + "flow_matching_loss": 0.012766898190602661, + "grad_norm": 0.0907062515616417, + "kept_token_count": 573.06328125, + "learning_rate": 1.2500099129140796e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9934561491012573, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006543857930228114, + "mgd_loss_weight": 0.42449748516082764, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.5874908447266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.37355651855466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17270, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.2993248850107193, + "epoch": 5.0276403840558626, + "flow_matching_loss": 0.011619200743734837, + "grad_norm": 0.07695513963699341, + "kept_token_count": 573.2625, + "learning_rate": 1.2483797200513065e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9936889469623565, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063110476359725, + "mgd_loss_weight": 0.4241641372442245, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 447.43520202636716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.03777770996095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17280, + "valid_token_count": 818.1625 + }, + { + "actual_token_mask_ratio": 0.29939569532871246, + "epoch": 5.030549898167006, + "flow_matching_loss": 0.012331347540020942, + "grad_norm": 0.10017514228820801, + "kept_token_count": 573.23671875, + "learning_rate": 1.2467498329308992e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9933609187602996, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006639085663482547, + "mgd_loss_weight": 0.4238307923078537, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.873291015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.36229553222654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17290, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.2997163087129593, + "epoch": 5.03345941227815, + "flow_matching_loss": 0.013190904958173632, + "grad_norm": 0.0638647973537445, + "kept_token_count": 573.025, + "learning_rate": 1.2451202535333253e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9931689918041229, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006831002049148082, + "mgd_loss_weight": 0.4234974533319473, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.9309356689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2105773925781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17300, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.2998961240053177, + "epoch": 5.036368926389293, + "flow_matching_loss": 0.013427015859633684, + "grad_norm": 0.08789720386266708, + "kept_token_count": 572.784375, + "learning_rate": 1.2434909838386768e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.9932533800601959, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006746625388041138, + "mgd_loss_weight": 0.423164102435112, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.7188720703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1780975341797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17310, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.30008831322193147, + "epoch": 5.0392784405004365, + "flow_matching_loss": 0.011035440862178803, + "grad_norm": 0.06531691551208496, + "kept_token_count": 572.75703125, + "learning_rate": 1.2418620258266703e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9935659468173981, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064340588636696335, + "mgd_loss_weight": 0.4228307604789734, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.69854431152345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.45433044433594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17320, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.30052298307418823, + "epoch": 5.04218795461158, + "flow_matching_loss": 0.012846179492771625, + "grad_norm": 0.08974085748195648, + "kept_token_count": 572.3765625, + "learning_rate": 1.2402333814766432e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9933961749076843, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006603834219276905, + "mgd_loss_weight": 0.422497421503067, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.32615051269534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9552459716797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17330, + "valid_token_count": 818.28671875 + }, + { + "actual_token_mask_ratio": 0.29920929074287417, + "epoch": 5.045097468722723, + "flow_matching_loss": 0.011324225459247828, + "grad_norm": 0.07443885505199432, + "kept_token_count": 573.40703125, + "learning_rate": 1.2386050527675515e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9930929601192474, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006907055946066975, + "mgd_loss_weight": 0.42216406762599945, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.08050231933595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3647399902344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17340, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.29992267191410066, + "epoch": 5.048006982833867, + "flow_matching_loss": 0.010603250190615653, + "grad_norm": 0.06964915245771408, + "kept_token_count": 572.7265625, + "learning_rate": 1.2369770416779693e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9937353670597077, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006264624837785959, + "mgd_loss_weight": 0.4218307286500931, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 450.7513031005859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.90000610351564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17350, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.30005561113357543, + "epoch": 5.05091649694501, + "flow_matching_loss": 0.011385770607739687, + "grad_norm": 0.07661667466163635, + "kept_token_count": 572.96796875, + "learning_rate": 1.2353493501860828e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9935890913009644, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006410907255485654, + "mgd_loss_weight": 0.4214973866939545, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.8406219482422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6955902099609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17360, + "valid_token_count": 818.59296875 + }, + { + "actual_token_mask_ratio": 0.3001218169927597, + "epoch": 5.053826011056153, + "flow_matching_loss": 0.01292385496199131, + "grad_norm": 0.08955418318510056, + "kept_token_count": 572.765625, + "learning_rate": 1.2337219802696905e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9932258129119873, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006774189369753003, + "mgd_loss_weight": 0.42116403579711914, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.89023742675784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2580169677734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17370, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.2998281538486481, + "epoch": 5.056735525167297, + "flow_matching_loss": 0.012871404737234115, + "grad_norm": 0.06775586307048798, + "kept_token_count": 572.78828125, + "learning_rate": 1.2320949339062008e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9934104382991791, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006589570222422481, + "mgd_loss_weight": 0.42083069682121277, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.947509765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0706512451172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17380, + "valid_token_count": 818.06796875 + }, + { + "actual_token_mask_ratio": 0.29990915954113007, + "epoch": 5.059645039278441, + "flow_matching_loss": 0.012810400500893594, + "grad_norm": 0.10600795596837997, + "kept_token_count": 573.03125, + "learning_rate": 1.2304682130726281e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9930310785770416, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006968917837366462, + "mgd_loss_weight": 0.42049734890460966, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.8349609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3970092773437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17390, + "valid_token_count": 818.5140625 + }, + { + "actual_token_mask_ratio": 0.299885892868042, + "epoch": 5.0625545533895835, + "flow_matching_loss": 0.012599050998687744, + "grad_norm": 0.09283451735973358, + "kept_token_count": 572.90546875, + "learning_rate": 1.2288418197455913e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9934762239456176, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006523769954219461, + "mgd_loss_weight": 0.42016400396823883, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.0344177246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.09691162109374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17400, + "valid_token_count": 818.30390625 + }, + { + "actual_token_mask_ratio": 0.2998809278011322, + "epoch": 5.065464067500727, + "flow_matching_loss": 0.012258565053343773, + "grad_norm": 0.10013856738805771, + "kept_token_count": 572.853125, + "learning_rate": 1.2272157559013126e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9938170671463012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006182929407805205, + "mgd_loss_weight": 0.41983066499233246, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.4824676513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.16014709472654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17410, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.29961574971675875, + "epoch": 5.068373581611871, + "flow_matching_loss": 0.011322708707302809, + "grad_norm": 0.08705682307481766, + "kept_token_count": 573.328125, + "learning_rate": 1.2255900235156123e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9934856772422791, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065143255982547995, + "mgd_loss_weight": 0.4194973140954971, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.7974487304688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.36411437988284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17420, + "valid_token_count": 818.58984375 + }, + { + "actual_token_mask_ratio": 0.30037703216075895, + "epoch": 5.071283095723015, + "flow_matching_loss": 0.010957830958068371, + "grad_norm": 0.07487183064222336, + "kept_token_count": 572.4203125, + "learning_rate": 1.2239646245639087e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9939388573169708, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060611461289227005, + "mgd_loss_weight": 0.4191639721393585, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.72430419921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3829284667969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17430, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.3003957986831665, + "epoch": 5.074192609834157, + "flow_matching_loss": 0.011720223817974329, + "grad_norm": 0.09249655157327652, + "kept_token_count": 572.4703125, + "learning_rate": 1.2223395610212142e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9936773121356964, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006322694476693869, + "mgd_loss_weight": 0.4188306301832199, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.9693939208984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.31244506835935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17440, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.3000917464494705, + "epoch": 5.077102123945301, + "flow_matching_loss": 0.011995705217123032, + "grad_norm": 0.07719717174768448, + "kept_token_count": 572.60546875, + "learning_rate": 1.2207148348621348e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9929550290107727, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.007044955901801586, + "mgd_loss_weight": 0.4184972792863846, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.921630859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.5511474609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17450, + "valid_token_count": 818.11640625 + }, + { + "actual_token_mask_ratio": 0.29977609515190123, + "epoch": 5.080011638056445, + "flow_matching_loss": 0.011452930420637131, + "grad_norm": 0.06786525249481201, + "kept_token_count": 573.11875, + "learning_rate": 1.219090448060866e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.993399840593338, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006600168067961931, + "mgd_loss_weight": 0.4181639403104782, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.1568542480469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.85101318359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17460, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.2994675874710083, + "epoch": 5.082921152167588, + "flow_matching_loss": 0.013427641149610281, + "grad_norm": 0.08905039727687836, + "kept_token_count": 572.97421875, + "learning_rate": 1.217466402591191e-05, + "loss": 0.0163, + "mgd_cos_sim": 0.993118166923523, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006881851935759187, + "mgd_loss_weight": 0.4178305953741074, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.5623779296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.594970703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17470, + "valid_token_count": 817.915625 + }, + { + "actual_token_mask_ratio": 0.2996355801820755, + "epoch": 5.085830666278731, + "flow_matching_loss": 0.011168711632490159, + "grad_norm": 0.07203395664691925, + "kept_token_count": 573.303125, + "learning_rate": 1.2158427004264785e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9938740372657776, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006125948065891862, + "mgd_loss_weight": 0.4174972474575043, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.2372711181641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1892547607422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17480, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.3000414282083511, + "epoch": 5.088740180389875, + "flow_matching_loss": 0.012554288934916258, + "grad_norm": 0.08914890885353088, + "kept_token_count": 572.70859375, + "learning_rate": 1.2142193435396788e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9935646653175354, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006435323040932417, + "mgd_loss_weight": 0.4171639084815979, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.7568420410156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.52637939453126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17490, + "valid_token_count": 818.19765625 + }, + { + "actual_token_mask_ratio": 0.29974278509616853, + "epoch": 5.091649694501018, + "flow_matching_loss": 0.011395214963704348, + "grad_norm": 0.07973391562700272, + "kept_token_count": 573.05859375, + "learning_rate": 1.2125963339033234e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9935066878795624, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006493312073871493, + "mgd_loss_weight": 0.41683055758476256, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.2511810302734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.03248901367186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17500, + "valid_token_count": 818.35390625 + }, + { + "actual_token_mask_ratio": 0.29959748685359955, + "epoch": 5.094559208612162, + "flow_matching_loss": 0.011339453561231495, + "grad_norm": 0.08893005549907684, + "kept_token_count": 573.37578125, + "learning_rate": 1.2109736734895237e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9935551047325134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006444886326789856, + "mgd_loss_weight": 0.41649721562862396, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.8555969238281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8396270751953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17510, + "valid_token_count": 818.63359375 + }, + { + "actual_token_mask_ratio": 0.29972313344478607, + "epoch": 5.097468722723305, + "flow_matching_loss": 0.01244916459545493, + "grad_norm": 0.08743955194950104, + "kept_token_count": 573.15625, + "learning_rate": 1.2093513642699634e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9933079838752746, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066920276265591385, + "mgd_loss_weight": 0.4161638766527176, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.33733825683595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.39711303710936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17520, + "valid_token_count": 818.47109375 + }, + { + "actual_token_mask_ratio": 0.30001654028892516, + "epoch": 5.100378236834449, + "flow_matching_loss": 0.012161188386380672, + "grad_norm": 0.08887929469347, + "kept_token_count": 572.80234375, + "learning_rate": 1.2077294082159024e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9933652400970459, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006634759716689587, + "mgd_loss_weight": 0.41583052277565, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.0797973632813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.8204345703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17530, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.29990454614162443, + "epoch": 5.103287750945592, + "flow_matching_loss": 0.011611646320670844, + "grad_norm": 0.07888573408126831, + "kept_token_count": 572.8296875, + "learning_rate": 1.2061078072981697e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9930881381034851, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00691186785697937, + "mgd_loss_weight": 0.41549718379974365, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.3351776123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2012145996094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17540, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.30003166794776914, + "epoch": 5.1061972650567355, + "flow_matching_loss": 0.011481065349653364, + "grad_norm": 0.1162886768579483, + "kept_token_count": 572.8359375, + "learning_rate": 1.2044865634871626e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9934910297393799, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006508964160457253, + "mgd_loss_weight": 0.41516384184360505, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.4546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.87083435058594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17550, + "valid_token_count": 818.37109375 + }, + { + "actual_token_mask_ratio": 0.29998970329761504, + "epoch": 5.109106779167879, + "flow_matching_loss": 0.011909067817032336, + "grad_norm": 0.06858659535646439, + "kept_token_count": 572.98828125, + "learning_rate": 1.2028656787528468e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9930783927440643, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006921608000993729, + "mgd_loss_weight": 0.4148304909467697, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.18328247070315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4395446777344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17560, + "valid_token_count": 818.5421875 + }, + { + "actual_token_mask_ratio": 0.29970644414424896, + "epoch": 5.112016293279022, + "flow_matching_loss": 0.010242060339078308, + "grad_norm": 0.09260784089565277, + "kept_token_count": 572.95859375, + "learning_rate": 1.2012451550647491e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9935223698616028, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006477621803060174, + "mgd_loss_weight": 0.41449715197086334, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.70835571289064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.608935546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17570, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.3003688007593155, + "epoch": 5.114925807390166, + "flow_matching_loss": 0.011031573545187712, + "grad_norm": 0.07514488697052002, + "kept_token_count": 572.490625, + "learning_rate": 1.1996249943919589e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9937518060207366, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006248207669705153, + "mgd_loss_weight": 0.4141638070344925, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.8796020507813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2438720703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17580, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.29944789707660674, + "epoch": 5.1178353215013095, + "flow_matching_loss": 0.010273388493806124, + "grad_norm": 0.06560967862606049, + "kept_token_count": 573.25703125, + "learning_rate": 1.1980051987031237e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9936117649078369, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006388237001374364, + "mgd_loss_weight": 0.4138304591178894, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.4788513183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3123321533203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17590, + "valid_token_count": 818.28671875 + }, + { + "actual_token_mask_ratio": 0.2985595941543579, + "epoch": 5.120744835612452, + "flow_matching_loss": 0.010614406736567616, + "grad_norm": 0.0822068527340889, + "kept_token_count": 574.0, + "learning_rate": 1.1963857699664483e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9934628486633301, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006537163304165006, + "mgd_loss_weight": 0.41349712014198303, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.3625122070313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.110693359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17600, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.29956587553024294, + "epoch": 5.123654349723596, + "flow_matching_loss": 0.012952649779617787, + "grad_norm": 0.08741967380046844, + "kept_token_count": 573.28046875, + "learning_rate": 1.1947667101496911e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9935867190361023, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006413257634267211, + "mgd_loss_weight": 0.4131637692451477, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.6959503173828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1354156494141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17610, + "valid_token_count": 818.46796875 + }, + { + "actual_token_mask_ratio": 0.30041092336177827, + "epoch": 5.12656386383474, + "flow_matching_loss": 0.012529338151216507, + "grad_norm": 0.09625266492366791, + "kept_token_count": 572.6234375, + "learning_rate": 1.1931480212201633e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9936559617519378, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006344043090939522, + "mgd_loss_weight": 0.4128304272890091, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.6438293457031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.83119506835936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17620, + "valid_token_count": 818.52109375 + }, + { + "actual_token_mask_ratio": 0.29984829425811765, + "epoch": 5.129473377945883, + "flow_matching_loss": 0.010646647214889527, + "grad_norm": 0.11965834349393845, + "kept_token_count": 572.93046875, + "learning_rate": 1.1915297051447237e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9937020540237427, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006297949003055692, + "mgd_loss_weight": 0.4124970883131027, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.2008361816406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.94781188964845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17630, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.2993247926235199, + "epoch": 5.132382892057026, + "flow_matching_loss": 0.010922307102009653, + "grad_norm": 0.0749548077583313, + "kept_token_count": 573.20546875, + "learning_rate": 1.1899117638897793e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9933576047420501, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006642386363819241, + "mgd_loss_weight": 0.41216373443603516, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.87811279296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.99144287109374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17640, + "valid_token_count": 818.08046875 + }, + { + "actual_token_mask_ratio": 0.3000569701194763, + "epoch": 5.13529240616817, + "flow_matching_loss": 0.012233216036111116, + "grad_norm": 0.08886189758777618, + "kept_token_count": 572.7765625, + "learning_rate": 1.1882941994212811e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9937511563301087, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006248838920146227, + "mgd_loss_weight": 0.4118303954601288, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.3522705078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4962585449219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17650, + "valid_token_count": 818.31640625 + }, + { + "actual_token_mask_ratio": 0.29960119128227236, + "epoch": 5.138201920279314, + "flow_matching_loss": 0.011730512976646424, + "grad_norm": 0.08846698701381683, + "kept_token_count": 573.19765625, + "learning_rate": 1.1866770137047221e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9937021255493164, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062978836242109535, + "mgd_loss_weight": 0.41149705052375796, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.64366149902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4938232421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17660, + "valid_token_count": 818.38984375 + }, + { + "actual_token_mask_ratio": 0.3004756271839142, + "epoch": 5.1411114343904565, + "flow_matching_loss": 0.012485323566943407, + "grad_norm": 0.08709560334682465, + "kept_token_count": 572.5734375, + "learning_rate": 1.1850602087051363e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.9935295045375824, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064705026801675555, + "mgd_loss_weight": 0.41116370260715485, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.98350219726564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.78481140136716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17670, + "valid_token_count": 818.5125 + }, + { + "actual_token_mask_ratio": 0.3001020669937134, + "epoch": 5.1440209485016, + "flow_matching_loss": 0.011450299248099326, + "grad_norm": 0.07119929045438766, + "kept_token_count": 572.8296875, + "learning_rate": 1.183443786387094e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.993575930595398, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006424059346318245, + "mgd_loss_weight": 0.4108303636312485, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.7570739746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8246154785156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17680, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.29989539086818695, + "epoch": 5.146930462612744, + "flow_matching_loss": 0.012433448247611523, + "grad_norm": 0.06574647128582001, + "kept_token_count": 572.85234375, + "learning_rate": 1.1818277487147005e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9938368618488311, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006163128325715661, + "mgd_loss_weight": 0.41049701571464536, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.5393463134766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.6126220703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17690, + "valid_token_count": 818.23515625 + }, + { + "actual_token_mask_ratio": 0.3009543359279633, + "epoch": 5.149839976723887, + "flow_matching_loss": 0.011578218406066298, + "grad_norm": 0.0792127475142479, + "kept_token_count": 571.925, + "learning_rate": 1.180212097651594e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9935867190361023, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006413282733410597, + "mgd_loss_weight": 0.41016367077827454, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.707080078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.83470764160154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17700, + "valid_token_count": 818.15859375 + }, + { + "actual_token_mask_ratio": 0.30036846697330477, + "epoch": 5.15274949083503, + "flow_matching_loss": 0.013169378973543644, + "grad_norm": 0.07039369642734528, + "kept_token_count": 572.45703125, + "learning_rate": 1.1785968351609422e-05, + "loss": 0.0157, + "mgd_cos_sim": 0.9937553822994232, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006244623102247715, + "mgd_loss_weight": 0.40983033180236816, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.4987823486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0490966796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17710, + "valid_token_count": 818.22890625 + }, + { + "actual_token_mask_ratio": 0.301241061091423, + "epoch": 5.155659004946174, + "flow_matching_loss": 0.011776381498202681, + "grad_norm": 0.09208891540765762, + "kept_token_count": 571.71171875, + "learning_rate": 1.1769819632054427e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9937941908836365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062058089766651395, + "mgd_loss_weight": 0.4094969779253006, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.27976684570314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9317138671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17720, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.2999077081680298, + "epoch": 5.158568519057318, + "flow_matching_loss": 0.01136855110526085, + "grad_norm": 0.08981617540121078, + "kept_token_count": 572.915625, + "learning_rate": 1.1753674837473165e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9932995736598969, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006700430903583765, + "mgd_loss_weight": 0.4091636389493942, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.1286346435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8123748779297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17730, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.3001602470874786, + "epoch": 5.161478033168461, + "flow_matching_loss": 0.011223708232864737, + "grad_norm": 0.05162474513053894, + "kept_token_count": 572.821875, + "learning_rate": 1.1737533987483081e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9937501609325409, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006249841488897801, + "mgd_loss_weight": 0.4088302969932556, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.7789276123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3761413574219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17740, + "valid_token_count": 818.50625 + }, + { + "actual_token_mask_ratio": 0.30033695697784424, + "epoch": 5.164387547279604, + "flow_matching_loss": 0.011447014939039945, + "grad_norm": 0.06569245457649231, + "kept_token_count": 572.38359375, + "learning_rate": 1.1721397101696835e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.99332355260849, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006676458939909935, + "mgd_loss_weight": 0.4084969460964203, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.4012481689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8096588134766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17750, + "valid_token_count": 818.08828125 + }, + { + "actual_token_mask_ratio": 0.2990594953298569, + "epoch": 5.167297061390748, + "flow_matching_loss": 0.01146516315639019, + "grad_norm": 0.08254145085811615, + "kept_token_count": 573.52578125, + "learning_rate": 1.1705264199722257e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9934575498104096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006542439432814717, + "mgd_loss_weight": 0.4081636071205139, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.0604217529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.572900390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17760, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.29980559051036837, + "epoch": 5.170206575501891, + "flow_matching_loss": 0.012903467938303947, + "grad_norm": 0.07012003660202026, + "kept_token_count": 572.92890625, + "learning_rate": 1.1689135301162356e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9933462083339691, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006653800420463085, + "mgd_loss_weight": 0.4078302621841431, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.40479736328126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7605743408203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17770, + "valid_token_count": 818.24453125 + }, + { + "actual_token_mask_ratio": 0.300214883685112, + "epoch": 5.173116089613035, + "flow_matching_loss": 0.01102638514712453, + "grad_norm": 0.09484291076660156, + "kept_token_count": 572.671875, + "learning_rate": 1.1673010425615258e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9934204459190369, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006579558737576008, + "mgd_loss_weight": 0.40749691426754, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 440.68980712890624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.79278564453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17780, + "valid_token_count": 818.35859375 + }, + { + "actual_token_mask_ratio": 0.29994708895683286, + "epoch": 5.176025603724178, + "flow_matching_loss": 0.012902317196130752, + "grad_norm": 0.0714091807603836, + "kept_token_count": 572.85859375, + "learning_rate": 1.1656889592674208e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9933352291584014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066647693980485204, + "mgd_loss_weight": 0.4071635752916336, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.3245361328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.7824310302734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17790, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.3008063405752182, + "epoch": 5.178935117835321, + "flow_matching_loss": 0.012830171175301075, + "grad_norm": 0.08173715323209763, + "kept_token_count": 572.0578125, + "learning_rate": 1.1640772821927533e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.99356968998909, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006430316297337413, + "mgd_loss_weight": 0.4068302273750305, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.858642578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.93641967773436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17800, + "valid_token_count": 818.16640625 + }, + { + "actual_token_mask_ratio": 0.2996344745159149, + "epoch": 5.181844631946465, + "flow_matching_loss": 0.01180839203298092, + "grad_norm": 0.08020827174186707, + "kept_token_count": 573.1125, + "learning_rate": 1.162466013295864e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9933610260486603, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006638962868601083, + "mgd_loss_weight": 0.40649688243865967, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 449.5068756103516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.44471740722656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17810, + "valid_token_count": 818.30703125 + }, + { + "actual_token_mask_ratio": 0.3002511441707611, + "epoch": 5.1847541460576085, + "flow_matching_loss": 0.01133651090785861, + "grad_norm": 0.08080128580331802, + "kept_token_count": 572.65234375, + "learning_rate": 1.160855154534595e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9933864593505859, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006613551266491413, + "mgd_loss_weight": 0.4061635434627533, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.1721984863281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4999084472656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17820, + "valid_token_count": 818.36796875 + }, + { + "actual_token_mask_ratio": 0.3000775933265686, + "epoch": 5.187663660168752, + "flow_matching_loss": 0.011317570740357042, + "grad_norm": 0.07565264403820038, + "kept_token_count": 572.72109375, + "learning_rate": 1.1592447078662931e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9934650599956513, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006534955836832524, + "mgd_loss_weight": 0.40583018958568573, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.38047180175784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.91905822753904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17830, + "valid_token_count": 818.26015625 + }, + { + "actual_token_mask_ratio": 0.3001550227403641, + "epoch": 5.190573174279895, + "flow_matching_loss": 0.013114039693027735, + "grad_norm": 0.08661201596260071, + "kept_token_count": 572.5484375, + "learning_rate": 1.1576346752478025e-05, + "loss": 0.0158, + "mgd_cos_sim": 0.9934080123901368, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006591983744874597, + "mgd_loss_weight": 0.40549685060977936, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.95151977539064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2146728515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17840, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.29971071481704714, + "epoch": 5.193482688391039, + "flow_matching_loss": 0.011405498906970023, + "grad_norm": 0.08131549507379532, + "kept_token_count": 573.07578125, + "learning_rate": 1.1560250586354643e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9936093747615814, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006390612805262208, + "mgd_loss_weight": 0.40516350865364076, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.4070587158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.87229309082034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17850, + "valid_token_count": 818.34296875 + }, + { + "actual_token_mask_ratio": 0.3003457337617874, + "epoch": 5.1963922025021825, + "flow_matching_loss": 0.010603269934654236, + "grad_norm": 0.06344332545995712, + "kept_token_count": 572.5390625, + "learning_rate": 1.1544158599851146e-05, + "loss": 0.0131, + "mgd_cos_sim": 0.9938501060009003, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061498903669416904, + "mgd_loss_weight": 0.4048301577568054, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.1540466308594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8099884033203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17860, + "valid_token_count": 818.31640625 + }, + { + "actual_token_mask_ratio": 0.29962014853954316, + "epoch": 5.199301716613325, + "flow_matching_loss": 0.011330869421362877, + "grad_norm": 0.0932583212852478, + "kept_token_count": 573.03046875, + "learning_rate": 1.1528070812520809e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9932753622531891, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006724635977298022, + "mgd_loss_weight": 0.40449681878089905, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.8633605957031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.1385009765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17870, + "valid_token_count": 818.16875 + }, + { + "actual_token_mask_ratio": 0.3006268829107285, + "epoch": 5.202211230724469, + "flow_matching_loss": 0.0108633853495121, + "grad_norm": 0.0862269252538681, + "kept_token_count": 572.36015625, + "learning_rate": 1.1511987243911821e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9939051389694213, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006094869552180171, + "mgd_loss_weight": 0.40416347086429594, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.41224670410156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.48621520996096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17880, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.3002430319786072, + "epoch": 5.205120744835613, + "flow_matching_loss": 0.011300937924534082, + "grad_norm": 0.09464120119810104, + "kept_token_count": 572.6828125, + "learning_rate": 1.1495907913567228e-05, + "loss": 0.014, + "mgd_cos_sim": 0.993399840593338, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066001441329717634, + "mgd_loss_weight": 0.4038301259279251, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 444.89357604980466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3799194335937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17890, + "valid_token_count": 818.39921875 + }, + { + "actual_token_mask_ratio": 0.2994969695806503, + "epoch": 5.2080302589467555, + "flow_matching_loss": 0.0110049600712955, + "grad_norm": 0.07265052199363708, + "kept_token_count": 573.3609375, + "learning_rate": 1.1479832841024936e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9933879137039184, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006612079218029976, + "mgd_loss_weight": 0.40349678695201874, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.60719604492186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.34154052734374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17900, + "valid_token_count": 818.50078125 + }, + { + "actual_token_mask_ratio": 0.30059308707714083, + "epoch": 5.210939773057899, + "flow_matching_loss": 0.012178302835673093, + "grad_norm": 0.10853344202041626, + "kept_token_count": 572.42265625, + "learning_rate": 1.1463762045817675e-05, + "loss": 0.0149, + "mgd_cos_sim": 0.9933413982391357, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006658593192696571, + "mgd_loss_weight": 0.4031634360551834, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.9469055175781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7395782470703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17910, + "valid_token_count": 818.43828125 + }, + { + "actual_token_mask_ratio": 0.29996304512023925, + "epoch": 5.213849287169043, + "flow_matching_loss": 0.010499528795480727, + "grad_norm": 0.07022196054458618, + "kept_token_count": 572.87421875, + "learning_rate": 1.1447695547472964e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9936739861965179, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006326016178354621, + "mgd_loss_weight": 0.4028300940990448, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 449.5758026123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.83433837890624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17920, + "valid_token_count": 818.35078125 + }, + { + "actual_token_mask_ratio": 0.3001704037189484, + "epoch": 5.216758801280186, + "flow_matching_loss": 0.01143113998696208, + "grad_norm": 0.0727236196398735, + "kept_token_count": 572.62109375, + "learning_rate": 1.1431633365513132e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9933963000774384, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066037069540470835, + "mgd_loss_weight": 0.4024967521429062, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.31931457519534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.1367218017578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17930, + "valid_token_count": 818.23203125 + }, + { + "actual_token_mask_ratio": 0.30028010010719297, + "epoch": 5.2196683153913295, + "flow_matching_loss": 0.012054563220590352, + "grad_norm": 0.08424928784370422, + "kept_token_count": 572.47265625, + "learning_rate": 1.1415575519455235e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9932443201541901, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006755666760727763, + "mgd_loss_weight": 0.40216340124607086, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.9639923095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.51532287597655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17940, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.2999680459499359, + "epoch": 5.222577829502473, + "flow_matching_loss": 0.011101504834368826, + "grad_norm": 0.07974185049533844, + "kept_token_count": 572.83984375, + "learning_rate": 1.1399522028811069e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9937354981899261, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006264508701860905, + "mgd_loss_weight": 0.4018300622701645, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.5469024658203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5061065673828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17950, + "valid_token_count": 818.30390625 + }, + { + "actual_token_mask_ratio": 0.29974542260169984, + "epoch": 5.225487343613617, + "flow_matching_loss": 0.011993172392249108, + "grad_norm": 0.08337025344371796, + "kept_token_count": 573.053125, + "learning_rate": 1.1383472913087143e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9935665905475617, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00643340409733355, + "mgd_loss_weight": 0.40149671733379366, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.1619476318359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.51275634765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17960, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.30010446310043337, + "epoch": 5.22839685772476, + "flow_matching_loss": 0.010696344543248415, + "grad_norm": 0.10925529152154922, + "kept_token_count": 572.77734375, + "learning_rate": 1.1367428191784643e-05, + "loss": 0.0132, + "mgd_cos_sim": 0.9937295913696289, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00627040471881628, + "mgd_loss_weight": 0.40116336941719055, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.4550048828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.62889709472654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17970, + "valid_token_count": 818.38046875 + }, + { + "actual_token_mask_ratio": 0.2994543373584747, + "epoch": 5.231306371835903, + "flow_matching_loss": 0.012149435468018055, + "grad_norm": 0.0773434191942215, + "kept_token_count": 573.14296875, + "learning_rate": 1.1351387884399413e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.993684059381485, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006315934518352151, + "mgd_loss_weight": 0.4008300304412842, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.8052612304688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.9235015869141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17980, + "valid_token_count": 818.13828125 + }, + { + "actual_token_mask_ratio": 0.30056433379650116, + "epoch": 5.234215885947047, + "flow_matching_loss": 0.01218509660102427, + "grad_norm": 0.08882427215576172, + "kept_token_count": 572.2765625, + "learning_rate": 1.1335352010421953e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9934610962867737, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00653889486566186, + "mgd_loss_weight": 0.40049668252468107, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.34465942382815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.877001953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 17990, + "valid_token_count": 818.19921875 + }, + { + "actual_token_mask_ratio": 0.2996282190084457, + "epoch": 5.23712540005819, + "flow_matching_loss": 0.012324930634349585, + "grad_norm": 0.09200938791036606, + "kept_token_count": 573.0109375, + "learning_rate": 1.1319320589337352e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.993712866306305, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006287128943949938, + "mgd_loss_weight": 0.40016333758831024, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.5884674072266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.34491577148435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18000, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.29940687119960785, + "epoch": 5.240034914169334, + "flow_matching_loss": 0.011615362949669361, + "grad_norm": 0.08809921890497208, + "kept_token_count": 573.1984375, + "learning_rate": 1.1303293640625299e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9930581331253052, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006941865151748061, + "mgd_loss_weight": 0.39982999861240387, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.11749267578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.93517150878904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18010, + "valid_token_count": 818.1609375 + }, + { + "actual_token_mask_ratio": 0.29982776045799253, + "epoch": 5.242944428280477, + "flow_matching_loss": 0.011358501110225915, + "grad_norm": 0.06848353892564774, + "kept_token_count": 572.93046875, + "learning_rate": 1.1287271183760052e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9935721576213836, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006427826546132565, + "mgd_loss_weight": 0.39949664771556853, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.1814910888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3987121582031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18020, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.29991101026535033, + "epoch": 5.24585394239162, + "flow_matching_loss": 0.011888047493994237, + "grad_norm": 0.08188247680664062, + "kept_token_count": 572.8828125, + "learning_rate": 1.1271253238210398e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9936867713928222, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063132309820503, + "mgd_loss_weight": 0.39916330575942993, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.94342651367185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0538604736328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18030, + "valid_token_count": 818.30078125 + }, + { + "actual_token_mask_ratio": 0.2999258518218994, + "epoch": 5.248763456502764, + "flow_matching_loss": 0.01101594027131796, + "grad_norm": 0.08605360239744186, + "kept_token_count": 572.74140625, + "learning_rate": 1.1255239823439663e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9937653362751007, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006234676390886307, + "mgd_loss_weight": 0.39882996380329133, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.9130096435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1238098144531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18040, + "valid_token_count": 818.11171875 + }, + { + "actual_token_mask_ratio": 0.3001786470413208, + "epoch": 5.251672970613908, + "flow_matching_loss": 0.01127708088606596, + "grad_norm": 0.0845101848244667, + "kept_token_count": 572.6453125, + "learning_rate": 1.1239230958905652e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9936477899551391, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00635221628472209, + "mgd_loss_weight": 0.398496612906456, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.6439971923828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6300933837891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18050, + "valid_token_count": 818.27734375 + }, + { + "actual_token_mask_ratio": 0.299834668636322, + "epoch": 5.254582484725051, + "flow_matching_loss": 0.011963112372905016, + "grad_norm": 0.06766899675130844, + "kept_token_count": 572.80234375, + "learning_rate": 1.1223226664060641e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9934098303318024, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006590180564671755, + "mgd_loss_weight": 0.3981632739305496, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.37660522460936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.5623291015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18060, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.299861466884613, + "epoch": 5.257491998836194, + "flow_matching_loss": 0.010325475642457605, + "grad_norm": 0.08735979348421097, + "kept_token_count": 572.99140625, + "learning_rate": 1.1207226958351362e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9935836434364319, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006416349858045578, + "mgd_loss_weight": 0.3978299289941788, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.35536499023436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0811706542969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18070, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.29993732273578644, + "epoch": 5.260401512947338, + "flow_matching_loss": 0.01082564047537744, + "grad_norm": 0.05881132930517197, + "kept_token_count": 572.7703125, + "learning_rate": 1.1191231861218955e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9935981214046479, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006401872355490923, + "mgd_loss_weight": 0.3974965810775757, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.65465087890624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.94288940429686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18080, + "valid_token_count": 818.16796875 + }, + { + "actual_token_mask_ratio": 0.3005810022354126, + "epoch": 5.2633110270584815, + "flow_matching_loss": 0.01123164719901979, + "grad_norm": 0.06762374937534332, + "kept_token_count": 572.265625, + "learning_rate": 1.1175241392098988e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9936814129352569, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00631858422420919, + "mgd_loss_weight": 0.3971632421016693, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.9878845214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.6556335449219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18090, + "valid_token_count": 818.19609375 + }, + { + "actual_token_mask_ratio": 0.2991735428571701, + "epoch": 5.266220541169624, + "flow_matching_loss": 0.011163806449621916, + "grad_norm": 0.11156714707612991, + "kept_token_count": 573.52890625, + "learning_rate": 1.1159255570421376e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9939114570617675, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00608855620957911, + "mgd_loss_weight": 0.396829891204834, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.3442901611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1825927734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18100, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.3004180699586868, + "epoch": 5.269130055280768, + "flow_matching_loss": 0.011626254953444003, + "grad_norm": 0.0653262734413147, + "kept_token_count": 572.2890625, + "learning_rate": 1.1143274415610408e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9930389821529388, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006961019430309534, + "mgd_loss_weight": 0.3964965492486954, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.3300842285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.05970764160156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18110, + "valid_token_count": 818.04609375 + }, + { + "actual_token_mask_ratio": 0.2994253754615784, + "epoch": 5.272039569391912, + "flow_matching_loss": 0.012344063818454742, + "grad_norm": 0.06608334183692932, + "kept_token_count": 573.52890625, + "learning_rate": 1.1127297947084692e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9939141154289246, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006085881544277072, + "mgd_loss_weight": 0.396163210272789, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.9532440185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.23094482421874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18120, + "valid_token_count": 818.659375 + }, + { + "actual_token_mask_ratio": 0.2998231887817383, + "epoch": 5.274949083503055, + "flow_matching_loss": 0.012665988504886627, + "grad_norm": 0.07546661049127579, + "kept_token_count": 572.8421875, + "learning_rate": 1.1111326184257135e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9933641076087951, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006635881494730711, + "mgd_loss_weight": 0.39582985639572144, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.0549774169922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.94183654785155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18130, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.2989495813846588, + "epoch": 5.277858597614198, + "flow_matching_loss": 0.012875373288989068, + "grad_norm": 0.07898107916116714, + "kept_token_count": 573.56484375, + "learning_rate": 1.109535914653495e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9933917582035064, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006608243193477392, + "mgd_loss_weight": 0.39549651741981506, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.45232849121095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3420806884766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18140, + "valid_token_count": 818.15078125 + }, + { + "actual_token_mask_ratio": 0.30080181658267974, + "epoch": 5.280768111725342, + "flow_matching_loss": 0.01203925097361207, + "grad_norm": 0.07550644129514694, + "kept_token_count": 572.0171875, + "learning_rate": 1.1079396853319582e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9938495397567749, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006150462012737989, + "mgd_loss_weight": 0.39516317248344424, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.4174407958984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.02882385253906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18150, + "valid_token_count": 818.1015625 + }, + { + "actual_token_mask_ratio": 0.30058401823043823, + "epoch": 5.283677625836486, + "flow_matching_loss": 0.012156523019075393, + "grad_norm": 0.09864652156829834, + "kept_token_count": 572.34453125, + "learning_rate": 1.1063439324006724e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.9932927787303925, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006707236915826797, + "mgd_loss_weight": 0.3948298245668411, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.0076904296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0838256835938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18160, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.300063419342041, + "epoch": 5.2865871399476285, + "flow_matching_loss": 0.011995672713965178, + "grad_norm": 0.08439704775810242, + "kept_token_count": 572.7625, + "learning_rate": 1.1047486577986285e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9933293700218201, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006670636357739568, + "mgd_loss_weight": 0.39449648559093475, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.6988586425781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2102386474609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18170, + "valid_token_count": 818.30859375 + }, + { + "actual_token_mask_ratio": 0.2998866677284241, + "epoch": 5.289496654058772, + "flow_matching_loss": 0.011182577582076192, + "grad_norm": 0.07865118235349655, + "kept_token_count": 572.89296875, + "learning_rate": 1.1031538634642353e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9937703490257264, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006229668343439698, + "mgd_loss_weight": 0.39416313767433164, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.12046508789064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9494598388672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18180, + "valid_token_count": 818.28515625 + }, + { + "actual_token_mask_ratio": 0.30034179985523224, + "epoch": 5.292406168169916, + "flow_matching_loss": 0.01125612361356616, + "grad_norm": 0.06596358865499496, + "kept_token_count": 572.68203125, + "learning_rate": 1.1015595513353172e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9935770332813263, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006422980176284909, + "mgd_loss_weight": 0.3938297927379608, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.41534423828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3373291015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18190, + "valid_token_count": 818.51171875 + }, + { + "actual_token_mask_ratio": 0.3005185484886169, + "epoch": 5.295315682281059, + "flow_matching_loss": 0.012807875778526068, + "grad_norm": 0.0776531994342804, + "kept_token_count": 572.33515625, + "learning_rate": 1.0999657233491156e-05, + "loss": 0.0154, + "mgd_cos_sim": 0.9935069620609284, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006493042455986142, + "mgd_loss_weight": 0.39349645376205444, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.24321899414065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1949737548828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18200, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.29894773960113524, + "epoch": 5.2982251963922025, + "flow_matching_loss": 0.012642278801649809, + "grad_norm": 0.08903615176677704, + "kept_token_count": 573.6640625, + "learning_rate": 1.0983723814422808e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9933574557304382, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006642542826011777, + "mgd_loss_weight": 0.3931631028652191, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.9007263183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.66111145019534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18210, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.3000869393348694, + "epoch": 5.301134710503346, + "flow_matching_loss": 0.011357660498470068, + "grad_norm": 0.0810985267162323, + "kept_token_count": 572.66796875, + "learning_rate": 1.0967795275508735e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.993686181306839, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006313805840909481, + "mgd_loss_weight": 0.3928297609090805, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.3562072753906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.6253173828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18220, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.3009412556886673, + "epoch": 5.304044224614489, + "flow_matching_loss": 0.012450511194765567, + "grad_norm": 0.1011328399181366, + "kept_token_count": 572.04609375, + "learning_rate": 1.095187163610361e-05, + "loss": 0.015, + "mgd_cos_sim": 0.993448805809021, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006551193352788687, + "mgd_loss_weight": 0.39249642193317413, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.71702880859374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4514953613281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18230, + "valid_token_count": 818.30859375 + }, + { + "actual_token_mask_ratio": 0.30024260878562925, + "epoch": 5.306953738725633, + "flow_matching_loss": 0.011244824901223183, + "grad_norm": 0.0910000279545784, + "kept_token_count": 572.5359375, + "learning_rate": 1.0935952915556163e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9935759961605072, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006423996016383171, + "mgd_loss_weight": 0.39216306805610657, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.1315216064453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.45741577148436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18240, + "valid_token_count": 818.19296875 + }, + { + "actual_token_mask_ratio": 0.29962902665138247, + "epoch": 5.309863252836776, + "flow_matching_loss": 0.012756952084600925, + "grad_norm": 0.08823957294225693, + "kept_token_count": 572.8265625, + "learning_rate": 1.0920039133209137e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9937692105770111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062307889573276045, + "mgd_loss_weight": 0.3918297290802002, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.14504089355466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.41510314941405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18250, + "valid_token_count": 817.89296875 + }, + { + "actual_token_mask_ratio": 0.29968329071998595, + "epoch": 5.31277276694792, + "flow_matching_loss": 0.011589644383639098, + "grad_norm": 0.0826462060213089, + "kept_token_count": 573.10546875, + "learning_rate": 1.0904130308399283e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9937489092350006, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006251096026971936, + "mgd_loss_weight": 0.39149638414382937, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.87900695800784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1616149902344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18260, + "valid_token_count": 818.353125 + }, + { + "actual_token_mask_ratio": 0.299877205491066, + "epoch": 5.315682281059063, + "flow_matching_loss": 0.012211747467517853, + "grad_norm": 0.06975001096725464, + "kept_token_count": 572.93984375, + "learning_rate": 1.0888226460457318e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.99356609582901, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006433907896280289, + "mgd_loss_weight": 0.39116303622722626, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.17539978027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1758178710937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18270, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.3008210778236389, + "epoch": 5.318591795170207, + "flow_matching_loss": 0.01156202731654048, + "grad_norm": 0.08196784555912018, + "kept_token_count": 572.015625, + "learning_rate": 1.087232760870792e-05, + "loss": 0.014, + "mgd_cos_sim": 0.993682187795639, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006317793624475598, + "mgd_loss_weight": 0.3908296972513199, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.3534759521484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0792999267578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18280, + "valid_token_count": 818.1234375 + }, + { + "actual_token_mask_ratio": 0.3004658579826355, + "epoch": 5.32150130928135, + "flow_matching_loss": 0.011167371179908514, + "grad_norm": 0.09050388634204865, + "kept_token_count": 572.453125, + "learning_rate": 1.0856433772469686e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9936469495296478, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00635305312462151, + "mgd_loss_weight": 0.3904963493347168, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.5420379638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2511291503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18290, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.29884488582611085, + "epoch": 5.324410823392493, + "flow_matching_loss": 0.011849235324189068, + "grad_norm": 0.08375000208616257, + "kept_token_count": 573.79296875, + "learning_rate": 1.0840544971055135e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9935901939868927, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006409805174916983, + "mgd_loss_weight": 0.39016300439834595, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.43665771484376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.70496520996096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18300, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.2999097853899002, + "epoch": 5.327320337503637, + "flow_matching_loss": 0.010079416912049055, + "grad_norm": 0.07108669728040695, + "kept_token_count": 572.703125, + "learning_rate": 1.0824661223770655e-05, + "loss": 0.0126, + "mgd_cos_sim": 0.9934596598148346, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006540343957021832, + "mgd_loss_weight": 0.3898296654224396, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.79969177246096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.57398681640626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18310, + "valid_token_count": 818.04609375 + }, + { + "actual_token_mask_ratio": 0.30027370154857635, + "epoch": 5.330229851614781, + "flow_matching_loss": 0.013856751658022404, + "grad_norm": 0.09836610406637192, + "kept_token_count": 572.57578125, + "learning_rate": 1.0808782549916498e-05, + "loss": 0.0165, + "mgd_cos_sim": 0.9932963073253631, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006703681917861104, + "mgd_loss_weight": 0.389496311545372, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.34320678710935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5882598876953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18320, + "valid_token_count": 818.28515625 + }, + { + "actual_token_mask_ratio": 0.2994563192129135, + "epoch": 5.333139365725923, + "flow_matching_loss": 0.011587771959602833, + "grad_norm": 0.08036381006240845, + "kept_token_count": 573.34140625, + "learning_rate": 1.079290896878675e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9937749803066254, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006225026678293943, + "mgd_loss_weight": 0.38916297256946564, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.2067108154297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.15398559570315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18330, + "valid_token_count": 818.42265625 + }, + { + "actual_token_mask_ratio": 0.3001716434955597, + "epoch": 5.336048879837067, + "flow_matching_loss": 0.011840357165783643, + "grad_norm": 0.08788619190454483, + "kept_token_count": 572.63828125, + "learning_rate": 1.07770404996693e-05, + "loss": 0.0143, + "mgd_cos_sim": 0.9935501933097839, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006449799053370953, + "mgd_loss_weight": 0.38882963061332704, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.84678649902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.7532104492187, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18340, + "valid_token_count": 818.26171875 + }, + { + "actual_token_mask_ratio": 0.2994972139596939, + "epoch": 5.338958393948211, + "flow_matching_loss": 0.010161064937710762, + "grad_norm": 0.11883652955293655, + "kept_token_count": 573.153125, + "learning_rate": 1.0761177161845848e-05, + "loss": 0.0126, + "mgd_cos_sim": 0.9936289310455322, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006371081992983818, + "mgd_loss_weight": 0.3884962797164917, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.50440368652346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9613098144531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18350, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.2999125063419342, + "epoch": 5.3418679080593545, + "flow_matching_loss": 0.012242446467280389, + "grad_norm": 0.06952698528766632, + "kept_token_count": 572.82109375, + "learning_rate": 1.0745318974591841e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9938644647598267, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006135537009686232, + "mgd_loss_weight": 0.3881629407405853, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.485009765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.99907836914065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18360, + "valid_token_count": 818.20546875 + }, + { + "actual_token_mask_ratio": 0.3003946006298065, + "epoch": 5.344777422170497, + "flow_matching_loss": 0.01227257214486599, + "grad_norm": 0.07544151693582535, + "kept_token_count": 572.47578125, + "learning_rate": 1.0729465957176469e-05, + "loss": 0.0148, + "mgd_cos_sim": 0.993591719865799, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064082836732268335, + "mgd_loss_weight": 0.3878295928239822, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.6791717529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7234252929687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18370, + "valid_token_count": 818.2796875 + }, + { + "actual_token_mask_ratio": 0.3002886563539505, + "epoch": 5.347686936281641, + "flow_matching_loss": 0.010891927732154727, + "grad_norm": 0.06871985644102097, + "kept_token_count": 572.7390625, + "learning_rate": 1.0713618128862642e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.993448281288147, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006551710050553083, + "mgd_loss_weight": 0.3874962478876114, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.6093719482422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9698791503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18380, + "valid_token_count": 818.5328125 + }, + { + "actual_token_mask_ratio": 0.30024556517601014, + "epoch": 5.350596450392785, + "flow_matching_loss": 0.011571662593632936, + "grad_norm": 0.0708734542131424, + "kept_token_count": 572.41953125, + "learning_rate": 1.0697775508906969e-05, + "loss": 0.0141, + "mgd_cos_sim": 0.9935316264629364, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006468374421820045, + "mgd_loss_weight": 0.387162908911705, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.20599365234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1762939453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18390, + "valid_token_count": 818.0296875 + }, + { + "actual_token_mask_ratio": 0.2996209293603897, + "epoch": 5.353505964503928, + "flow_matching_loss": 0.011435576574876904, + "grad_norm": 0.05539451166987419, + "kept_token_count": 573.20625, + "learning_rate": 1.068193811655972e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9935723900794983, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006427602469921112, + "mgd_loss_weight": 0.3868295580148697, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.74029235839845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0526153564453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18400, + "valid_token_count": 818.42265625 + }, + { + "actual_token_mask_ratio": 0.3009326905012131, + "epoch": 5.356415478615071, + "flow_matching_loss": 0.012017331877723336, + "grad_norm": 0.09023706614971161, + "kept_token_count": 572.2421875, + "learning_rate": 1.0666105971064835e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9935282170772552, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00647180425003171, + "mgd_loss_weight": 0.3864962160587311, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.10370483398435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.33363037109376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18410, + "valid_token_count": 818.58046875 + }, + { + "actual_token_mask_ratio": 0.29943631291389466, + "epoch": 5.359324992726215, + "flow_matching_loss": 0.012027316074818373, + "grad_norm": 0.09297313541173935, + "kept_token_count": 573.1625, + "learning_rate": 1.0650279091659852e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9935317933559418, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006468211673200131, + "mgd_loss_weight": 0.3861628770828247, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.63019409179685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1520263671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18420, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.3001917690038681, + "epoch": 5.362234506837358, + "flow_matching_loss": 0.011274079652503132, + "grad_norm": 0.07615166157484055, + "kept_token_count": 572.5796875, + "learning_rate": 1.0634457497575926e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.993481022119522, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006518962839618325, + "mgd_loss_weight": 0.38582952320575714, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.6422576904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.74244689941406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18430, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.29993053078651427, + "epoch": 5.3651440209485015, + "flow_matching_loss": 0.011720722354948521, + "grad_norm": 0.06907711923122406, + "kept_token_count": 572.76171875, + "learning_rate": 1.061864120803778e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.9935167789459228, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006483226735144854, + "mgd_loss_weight": 0.38549618422985077, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.5793853759766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1783020019531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18440, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.29949377477169037, + "epoch": 5.368053535059645, + "flow_matching_loss": 0.01184635991230607, + "grad_norm": 0.06451206654310226, + "kept_token_count": 573.30625, + "learning_rate": 1.0602830242263701e-05, + "loss": 0.0142, + "mgd_cos_sim": 0.993849265575409, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061507370788604025, + "mgd_loss_weight": 0.38516284227371217, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.3690246582031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.86248779296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18450, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.3003047347068787, + "epoch": 5.370963049170788, + "flow_matching_loss": 0.010898258676752448, + "grad_norm": 0.07580960541963577, + "kept_token_count": 572.4328125, + "learning_rate": 1.0587024619465506e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9934679806232453, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006532019376754761, + "mgd_loss_weight": 0.38482949137687683, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.20176696777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1433959960938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18460, + "valid_token_count": 818.11640625 + }, + { + "actual_token_mask_ratio": 0.29977835416793824, + "epoch": 5.373872563281932, + "flow_matching_loss": 0.011482427874580026, + "grad_norm": 0.05873122066259384, + "kept_token_count": 573.01484375, + "learning_rate": 1.0571224358848517e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9936069548130035, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006393030611798167, + "mgd_loss_weight": 0.38449615240097046, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.4843688964844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2107299804687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18470, + "valid_token_count": 818.33203125 + }, + { + "actual_token_mask_ratio": 0.2996894508600235, + "epoch": 5.3767820773930755, + "flow_matching_loss": 0.012235208414494991, + "grad_norm": 0.07951024919748306, + "kept_token_count": 573.10390625, + "learning_rate": 1.0555429479611537e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9936912775039672, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006308719702064991, + "mgd_loss_weight": 0.38416280448436735, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.73769226074216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.46091613769534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18480, + "valid_token_count": 818.36015625 + }, + { + "actual_token_mask_ratio": 0.2998292773962021, + "epoch": 5.379691591504219, + "flow_matching_loss": 0.011595113296061754, + "grad_norm": 0.07415147125720978, + "kept_token_count": 572.99296875, + "learning_rate": 1.0539640000946835e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9937892913818359, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006210709735751152, + "mgd_loss_weight": 0.3838294595479965, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.66175842285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5779144287109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18490, + "valid_token_count": 818.3625 + }, + { + "actual_token_mask_ratio": 0.29995748996734617, + "epoch": 5.382601105615362, + "flow_matching_loss": 0.010052172187715768, + "grad_norm": 0.08036181330680847, + "kept_token_count": 572.7734375, + "learning_rate": 1.0523855942040111e-05, + "loss": 0.0125, + "mgd_cos_sim": 0.9936098039150238, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006390193291008472, + "mgd_loss_weight": 0.38349612057209015, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.4162292480469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5345001220703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18500, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.30051811337471007, + "epoch": 5.385510619726506, + "flow_matching_loss": 0.011132914246991277, + "grad_norm": 0.05112794041633606, + "kept_token_count": 572.43046875, + "learning_rate": 1.0508077322070496e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9936014831066131, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006398524297401309, + "mgd_loss_weight": 0.3831627696752548, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.4786773681641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.88462219238284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18510, + "valid_token_count": 818.36640625 + }, + { + "actual_token_mask_ratio": 0.30018052756786345, + "epoch": 5.388420133837649, + "flow_matching_loss": 0.011416512494906782, + "grad_norm": 0.08585899323225021, + "kept_token_count": 572.55625, + "learning_rate": 1.0492304160210495e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9939710378646851, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006028951471671462, + "mgd_loss_weight": 0.3828294277191162, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.46608276367186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.485693359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18520, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.30013868808746336, + "epoch": 5.391329647948792, + "flow_matching_loss": 0.012716194707900285, + "grad_norm": 0.08331704139709473, + "kept_token_count": 572.51484375, + "learning_rate": 1.0476536475625985e-05, + "loss": 0.0153, + "mgd_cos_sim": 0.9933118641376495, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006688141962513328, + "mgd_loss_weight": 0.3824960857629776, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.1380157470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8044128417969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18530, + "valid_token_count": 818.040625 + }, + { + "actual_token_mask_ratio": 0.29988019466400145, + "epoch": 5.394239162059936, + "flow_matching_loss": 0.013644586037844419, + "grad_norm": 0.09832495450973511, + "kept_token_count": 572.8015625, + "learning_rate": 1.0460774287476195e-05, + "loss": 0.0161, + "mgd_cos_sim": 0.9935739398002624, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006426057033240795, + "mgd_loss_weight": 0.3821627348661423, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.1990173339844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3742279052734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18540, + "valid_token_count": 818.15078125 + }, + { + "actual_token_mask_ratio": 0.2996418297290802, + "epoch": 5.39714867617108, + "flow_matching_loss": 0.011043997202068567, + "grad_norm": 0.07564817368984222, + "kept_token_count": 573.0578125, + "learning_rate": 1.044501761491366e-05, + "loss": 0.0136, + "mgd_cos_sim": 0.9932926774024964, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006707341223955154, + "mgd_loss_weight": 0.3818293958902359, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.63966979980466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9902099609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18550, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.30026639699935914, + "epoch": 5.400058190282223, + "flow_matching_loss": 0.01061010044068098, + "grad_norm": 0.08157272636890411, + "kept_token_count": 572.71875, + "learning_rate": 1.0429266477084235e-05, + "loss": 0.013, + "mgd_cos_sim": 0.9936452329158783, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006354763964191079, + "mgd_loss_weight": 0.3814960509538651, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.8007110595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5420745849609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18560, + "valid_token_count": 818.48515625 + }, + { + "actual_token_mask_ratio": 0.3001323163509369, + "epoch": 5.402967704393366, + "flow_matching_loss": 0.013010983075946569, + "grad_norm": 0.07779552042484283, + "kept_token_count": 572.6296875, + "learning_rate": 1.0413520893127034e-05, + "loss": 0.0155, + "mgd_cos_sim": 0.9934884309768677, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006511563109233976, + "mgd_loss_weight": 0.38116270303726196, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.44895935058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2301818847656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18570, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.2998039275407791, + "epoch": 5.40587721850451, + "flow_matching_loss": 0.01050764131359756, + "grad_norm": 0.0882333442568779, + "kept_token_count": 572.903125, + "learning_rate": 1.0397780882174424e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9937715768814087, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006228426890447736, + "mgd_loss_weight": 0.3808293640613556, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.2245330810547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.01844787597656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18580, + "valid_token_count": 818.20078125 + }, + { + "actual_token_mask_ratio": 0.30014516711235045, + "epoch": 5.408786732615654, + "flow_matching_loss": 0.01109215235337615, + "grad_norm": 0.06457095593214035, + "kept_token_count": 572.5296875, + "learning_rate": 1.0382046463351998e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9937929749488831, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006207024212926626, + "mgd_loss_weight": 0.38049601316452025, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.74949951171874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8119781494141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18590, + "valid_token_count": 818.07421875 + }, + { + "actual_token_mask_ratio": 0.2999383360147476, + "epoch": 5.411696246726796, + "flow_matching_loss": 0.012701751757413149, + "grad_norm": 0.07849949598312378, + "kept_token_count": 572.77265625, + "learning_rate": 1.0366317655778568e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9935404539108277, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006459536449983716, + "mgd_loss_weight": 0.38016267120838165, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.24745788574216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.45371704101564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18600, + "valid_token_count": 818.17578125 + }, + { + "actual_token_mask_ratio": 0.29977128803730013, + "epoch": 5.41460576083794, + "flow_matching_loss": 0.010316486936062574, + "grad_norm": 0.09833405166864395, + "kept_token_count": 573.00390625, + "learning_rate": 1.035059447856611e-05, + "loss": 0.0127, + "mgd_cos_sim": 0.9937115132808685, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006288478150963783, + "mgd_loss_weight": 0.3798293322324753, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.4609436035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.7412353515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18610, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.29968891441822054, + "epoch": 5.417515274949084, + "flow_matching_loss": 0.011374547425657511, + "grad_norm": 0.07949535548686981, + "kept_token_count": 573.1, + "learning_rate": 1.0334876950819777e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9936568439006805, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006343174539506435, + "mgd_loss_weight": 0.3794959783554077, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.85374450683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.18793029785155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18620, + "valid_token_count": 818.35078125 + }, + { + "actual_token_mask_ratio": 0.29927792847156526, + "epoch": 5.420424789060227, + "flow_matching_loss": 0.011275392584502697, + "grad_norm": 0.07355273514986038, + "kept_token_count": 573.4203125, + "learning_rate": 1.0319165091637837e-05, + "loss": 0.0137, + "mgd_cos_sim": 0.9936625838279725, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006337408069521188, + "mgd_loss_weight": 0.37916263937950134, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.96954040527345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.47829284667966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18630, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.30022644996643066, + "epoch": 5.42333430317137, + "flow_matching_loss": 0.011563637759536505, + "grad_norm": 0.06631281226873398, + "kept_token_count": 572.56796875, + "learning_rate": 1.0303458920111686e-05, + "loss": 0.014, + "mgd_cos_sim": 0.993526554107666, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006473454274237156, + "mgd_loss_weight": 0.3788292944431305, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.5866455078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.44309692382814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18640, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.2996246308088303, + "epoch": 5.426243817282514, + "flow_matching_loss": 0.012857567705214023, + "grad_norm": 0.07154868543148041, + "kept_token_count": 573.2109375, + "learning_rate": 1.0287758455325803e-05, + "loss": 0.0152, + "mgd_cos_sim": 0.9937869608402252, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006213025795295835, + "mgd_loss_weight": 0.3784959465265274, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.0455780029297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.36256103515626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18650, + "valid_token_count": 818.42890625 + }, + { + "actual_token_mask_ratio": 0.2996759355068207, + "epoch": 5.429153331393657, + "flow_matching_loss": 0.01110709561035037, + "grad_norm": 0.07268443703651428, + "kept_token_count": 573.22109375, + "learning_rate": 1.0272063716357729e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9942035734653473, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005796429002657533, + "mgd_loss_weight": 0.37816260755062103, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.6514190673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.90708923339844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18660, + "valid_token_count": 818.50234375 + }, + { + "actual_token_mask_ratio": 0.30011110603809354, + "epoch": 5.432062845504801, + "flow_matching_loss": 0.010592175181955099, + "grad_norm": 0.06196139007806778, + "kept_token_count": 572.48359375, + "learning_rate": 1.0256374722278057e-05, + "loss": 0.0129, + "mgd_cos_sim": 0.9938319206237793, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006168073927983642, + "mgd_loss_weight": 0.3778292626142502, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.5830474853516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.21828308105466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18670, + "valid_token_count": 817.96015625 + }, + { + "actual_token_mask_ratio": 0.30031820237636564, + "epoch": 5.434972359615944, + "flow_matching_loss": 0.012231663987040519, + "grad_norm": 0.0752568170428276, + "kept_token_count": 572.58984375, + "learning_rate": 1.0240691492150388e-05, + "loss": 0.0147, + "mgd_cos_sim": 0.9935879051685333, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006412088358774781, + "mgd_loss_weight": 0.3774959146976471, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.2604644775391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9048095703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18680, + "valid_token_count": 818.35859375 + }, + { + "actual_token_mask_ratio": 0.29986971616744995, + "epoch": 5.437881873727088, + "flow_matching_loss": 0.010808940976858139, + "grad_norm": 0.057104676961898804, + "kept_token_count": 572.74375, + "learning_rate": 1.0225014045031334e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9933829367160797, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006617040932178497, + "mgd_loss_weight": 0.3771625757217407, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.8517211914062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.45680847167966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18690, + "valid_token_count": 818.05546875 + }, + { + "actual_token_mask_ratio": 0.3008095592260361, + "epoch": 5.440791387838231, + "flow_matching_loss": 0.011082230415195227, + "grad_norm": 0.0737878829240799, + "kept_token_count": 572.13046875, + "learning_rate": 1.0209342399970468e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.9938804626464843, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006119518727064133, + "mgd_loss_weight": 0.3768292248249054, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.9670379638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3970916748047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18700, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.300628998875618, + "epoch": 5.4437009019493745, + "flow_matching_loss": 0.01102106375619769, + "grad_norm": 0.06474096328020096, + "kept_token_count": 572.27734375, + "learning_rate": 1.019367657601031e-05, + "loss": 0.0135, + "mgd_cos_sim": 0.9935398399829865, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064601520076394085, + "mgd_loss_weight": 0.3764958828687668, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.8166900634766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3288238525391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18710, + "valid_token_count": 818.27578125 + }, + { + "actual_token_mask_ratio": 0.2999301999807358, + "epoch": 5.446610416060518, + "flow_matching_loss": 0.011425207089632749, + "grad_norm": 0.0620122030377388, + "kept_token_count": 572.8875, + "learning_rate": 1.0178016592186323e-05, + "loss": 0.0138, + "mgd_cos_sim": 0.9937016367912292, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006298360601067543, + "mgd_loss_weight": 0.3761625438928604, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.9613311767578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.55364685058595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18720, + "valid_token_count": 818.32734375 + }, + { + "actual_token_mask_ratio": 0.29934294521808624, + "epoch": 5.449519930171661, + "flow_matching_loss": 0.012620151322335006, + "grad_norm": 0.06840653717517853, + "kept_token_count": 573.24921875, + "learning_rate": 1.0162362467526861e-05, + "loss": 0.015, + "mgd_cos_sim": 0.9937641441822052, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006235853116959333, + "mgd_loss_weight": 0.37582919001579285, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.23798217773435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9526336669922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18730, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.30089475214481354, + "epoch": 5.452429444282805, + "flow_matching_loss": 0.01139172213152051, + "grad_norm": 0.06350165605545044, + "kept_token_count": 572.228125, + "learning_rate": 1.0146714221053159e-05, + "loss": 0.0139, + "mgd_cos_sim": 0.9933662354946137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00663376972079277, + "mgd_loss_weight": 0.3754958510398865, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.71544494628904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.3390350341797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18740, + "valid_token_count": 818.5203125 + }, + { + "actual_token_mask_ratio": 0.30041358470916746, + "epoch": 5.4553389583939484, + "flow_matching_loss": 0.011619736393913626, + "grad_norm": 0.07940904051065445, + "kept_token_count": 572.5171875, + "learning_rate": 1.0131071871779314e-05, + "loss": 0.014, + "mgd_cos_sim": 0.9935516834259033, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006448318529874086, + "mgd_loss_weight": 0.37516250610351565, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.18585205078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3338562011719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18750, + "valid_token_count": 818.36640625 + }, + { + "actual_token_mask_ratio": 0.2995237559080124, + "epoch": 5.458248472505091, + "flow_matching_loss": 0.013404584117233753, + "grad_norm": 0.08435339480638504, + "kept_token_count": 573.08359375, + "learning_rate": 1.0115435438712249e-05, + "loss": 0.0159, + "mgd_cos_sim": 0.9933961927890778, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00660380283370614, + "mgd_loss_weight": 0.37482915818691254, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.9383819580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.21937561035156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18760, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.3005609452724457, + "epoch": 5.461157986616235, + "flow_matching_loss": 0.013244082313030958, + "grad_norm": 0.0837419405579567, + "kept_token_count": 572.2875, + "learning_rate": 1.0099804940851716e-05, + "loss": 0.0156, + "mgd_cos_sim": 0.9936499714851379, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006350032193586231, + "mgd_loss_weight": 0.37449581921100616, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.3356109619141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8785369873047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18770, + "valid_token_count": 818.20703125 + }, + { + "actual_token_mask_ratio": 0.2998965322971344, + "epoch": 5.464067500727379, + "flow_matching_loss": 0.010897805960848928, + "grad_norm": 0.06057242304086685, + "kept_token_count": 572.70078125, + "learning_rate": 1.0084180397190238e-05, + "loss": 0.0133, + "mgd_cos_sim": 0.9936017751693725, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006398219335824251, + "mgd_loss_weight": 0.37416247129440305, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.13926086425784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2063415527344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18780, + "valid_token_count": 818.02265625 + }, + { + "actual_token_mask_ratio": 0.3006874144077301, + "epoch": 5.4669770148385215, + "flow_matching_loss": 0.012644081842154264, + "grad_norm": 0.06522487848997116, + "kept_token_count": 572.3109375, + "learning_rate": 1.006856182671311e-05, + "loss": 0.0151, + "mgd_cos_sim": 0.993461686372757, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006538312137126923, + "mgd_loss_weight": 0.3738291263580322, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.96331481933595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9138519287109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18790, + "valid_token_count": 818.38671875 + }, + { + "actual_token_mask_ratio": 0.29912029206752777, + "epoch": 5.469886528949665, + "flow_matching_loss": 0.011988986656069756, + "grad_norm": 0.08424998819828033, + "kept_token_count": 573.5203125, + "learning_rate": 1.0052949248398367e-05, + "loss": 0.0144, + "mgd_cos_sim": 0.9934498071670532, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006550190504640341, + "mgd_loss_weight": 0.37349578738212585, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.9894134521484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9449066162109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18800, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.30046841502189636, + "epoch": 5.472796043060809, + "flow_matching_loss": 0.012128121266141535, + "grad_norm": 0.08030928671360016, + "kept_token_count": 572.5765625, + "learning_rate": 1.003734268121676e-05, + "loss": 0.0145, + "mgd_cos_sim": 0.9937000870704651, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006299921264871955, + "mgd_loss_weight": 0.3731624335050583, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.1932678222656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7186584472656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18810, + "valid_token_count": 818.51015625 + }, + { + "actual_token_mask_ratio": 0.30027155578136444, + "epoch": 5.475705557171953, + "flow_matching_loss": 0.011063891928642988, + "grad_norm": 0.06660174578428268, + "kept_token_count": 572.58984375, + "learning_rate": 1.0021742144131744e-05, + "loss": 0.0134, + "mgd_cos_sim": 0.993791264295578, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006208741711452604, + "mgd_loss_weight": 0.3728290945291519, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.8049072265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.84161071777345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18820, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.299691304564476, + "epoch": 5.478615071283095, + "flow_matching_loss": 0.012165590981021523, + "grad_norm": 0.08349911123514175, + "kept_token_count": 573.26640625, + "learning_rate": 1.000614765609944e-05, + "loss": 0.0146, + "mgd_cos_sim": 0.9935542702674866, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006445743469521403, + "mgd_loss_weight": 0.3724957525730133, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.0580718994141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8778778076172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18830, + "valid_token_count": 818.59453125 + }, + { + "actual_token_mask_ratio": 0.29994210600852966, + "epoch": 5.481524585394239, + "flow_matching_loss": 0.01057629669085145, + "grad_norm": 0.0761311948299408, + "kept_token_count": 572.7859375, + "learning_rate": 9.990559236068628e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9937294721603394, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006270530261099338, + "mgd_loss_weight": 0.372162401676178, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.4923461914062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6563323974609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18840, + "valid_token_count": 818.20078125 + }, + { + "actual_token_mask_ratio": 0.3002846330404282, + "epoch": 5.484434099505383, + "flow_matching_loss": 0.011255728174000978, + "grad_norm": 0.07729785144329071, + "kept_token_count": 572.53203125, + "learning_rate": 9.974976902980703e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9935618996620178, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006438096705824137, + "mgd_loss_weight": 0.3718290627002716, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.42760009765624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5443511962891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18850, + "valid_token_count": 818.23359375 + }, + { + "actual_token_mask_ratio": 0.2997289389371872, + "epoch": 5.487343613616526, + "flow_matching_loss": 0.012306836992502212, + "grad_norm": 0.06023546680808067, + "kept_token_count": 572.94921875, + "learning_rate": 9.959400675769668e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9933412194252014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006658782577142119, + "mgd_loss_weight": 0.3714957147836685, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.56346130371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.95623168945315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18860, + "valid_token_count": 818.18046875 + }, + { + "actual_token_mask_ratio": 0.2994279623031616, + "epoch": 5.490253127727669, + "flow_matching_loss": 0.012043858598917722, + "grad_norm": 0.0799187496304512, + "kept_token_count": 573.2125, + "learning_rate": 9.943830573362106e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9935013711452484, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064986416604369875, + "mgd_loss_weight": 0.37116236984729767, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.4223571777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0356170654297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18870, + "valid_token_count": 818.20546875 + }, + { + "actual_token_mask_ratio": 0.30014684796333313, + "epoch": 5.493162641838813, + "flow_matching_loss": 0.012503729946911336, + "grad_norm": 0.07018906623125076, + "kept_token_count": 572.70703125, + "learning_rate": 9.928266614677168e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.993492317199707, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006507693324238062, + "mgd_loss_weight": 0.3708290308713913, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.8678802490234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8188110351563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18880, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.3001666754484177, + "epoch": 5.496072155949957, + "flow_matching_loss": 0.0122827366925776, + "grad_norm": 0.062015101313591, + "kept_token_count": 572.75078125, + "learning_rate": 9.912708818626524e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9934897303581238, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006510264333337545, + "mgd_loss_weight": 0.3704956829547882, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.7259796142578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9907989501953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18890, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.2997904121875763, + "epoch": 5.4989816700611, + "flow_matching_loss": 0.011219463311135769, + "grad_norm": 0.08436606824398041, + "kept_token_count": 572.92578125, + "learning_rate": 9.897157204114369e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9940057933330536, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005994216306135058, + "mgd_loss_weight": 0.37016233801841736, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.7894226074219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.48761596679685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18900, + "valid_token_count": 818.22109375 + }, + { + "actual_token_mask_ratio": 0.2996865063905716, + "epoch": 5.501891184172243, + "flow_matching_loss": 0.010993981547653676, + "grad_norm": 0.1022607684135437, + "kept_token_count": 572.971875, + "learning_rate": 9.881611790037375e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9935398459434509, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064601405058056114, + "mgd_loss_weight": 0.369828999042511, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.4685852050781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6857940673828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18910, + "valid_token_count": 818.16484375 + }, + { + "actual_token_mask_ratio": 0.3005066096782684, + "epoch": 5.504800698283387, + "flow_matching_loss": 0.011765863094478846, + "grad_norm": 0.073643758893013, + "kept_token_count": 572.446875, + "learning_rate": 9.866072595284684e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9934040427207946, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006595960864797234, + "mgd_loss_weight": 0.3694956451654434, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.88686218261716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6196258544922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18920, + "valid_token_count": 818.3703125 + }, + { + "actual_token_mask_ratio": 0.30038184523582456, + "epoch": 5.50771021239453, + "flow_matching_loss": 0.011795329954475164, + "grad_norm": 0.06145423650741577, + "kept_token_count": 572.5234375, + "learning_rate": 9.850539638737894e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9931489765644074, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006851038662716746, + "mgd_loss_weight": 0.36916230618953705, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.00957946777345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2344055175781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18930, + "valid_token_count": 818.33515625 + }, + { + "actual_token_mask_ratio": 0.299769452214241, + "epoch": 5.5106197265056736, + "flow_matching_loss": 0.01262213783338666, + "grad_norm": 0.09800533950328827, + "kept_token_count": 573.09453125, + "learning_rate": 9.835012939271006e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.993343198299408, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006656813621520996, + "mgd_loss_weight": 0.36882896423339845, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.667724609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.67021484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18940, + "valid_token_count": 818.43125 + }, + { + "actual_token_mask_ratio": 0.2998886346817017, + "epoch": 5.513529240616817, + "flow_matching_loss": 0.010821118950843811, + "grad_norm": 0.07178322970867157, + "kept_token_count": 572.9984375, + "learning_rate": 9.819492515750418e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9935659945011139, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006433992832899094, + "mgd_loss_weight": 0.3684956133365631, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.5185852050781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.40331115722654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18950, + "valid_token_count": 818.44375 + }, + { + "actual_token_mask_ratio": 0.3000791400671005, + "epoch": 5.51643875472796, + "flow_matching_loss": 0.010862215980887413, + "grad_norm": 0.07842571288347244, + "kept_token_count": 572.85078125, + "learning_rate": 9.803978387034918e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9936176121234894, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006382377073168755, + "mgd_loss_weight": 0.36816227436065674, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.3031036376953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5192901611328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18960, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.299859219789505, + "epoch": 5.519348268839104, + "flow_matching_loss": 0.013616978656500578, + "grad_norm": 0.08746033906936646, + "kept_token_count": 572.903125, + "learning_rate": 9.788470571975624e-06, + "loss": 0.016, + "mgd_cos_sim": 0.9934939503669739, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006506043719127774, + "mgd_loss_weight": 0.3678289264440536, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.8789123535156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.102001953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18970, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.3004279285669327, + "epoch": 5.5222577829502475, + "flow_matching_loss": 0.010725435335189104, + "grad_norm": 0.07613854855298996, + "kept_token_count": 572.28125, + "learning_rate": 9.772969089416008e-06, + "loss": 0.013, + "mgd_cos_sim": 0.993808102607727, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061918905470520254, + "mgd_loss_weight": 0.3674955815076828, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.8026153564453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.31963500976565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18980, + "valid_token_count": 818.04140625 + }, + { + "actual_token_mask_ratio": 0.2996164858341217, + "epoch": 5.52516729706139, + "flow_matching_loss": 0.011401458596810698, + "grad_norm": 0.07533971220254898, + "kept_token_count": 573.290625, + "learning_rate": 9.757473958191822e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9934403777122498, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006559637188911438, + "mgd_loss_weight": 0.36716224253177643, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.5875183105469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2771697998047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 18990, + "valid_token_count": 818.52734375 + }, + { + "actual_token_mask_ratio": 0.29980801045894623, + "epoch": 5.528076811172534, + "flow_matching_loss": 0.010389988590031863, + "grad_norm": 0.077475905418396, + "kept_token_count": 572.840625, + "learning_rate": 9.741985197131118e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9938513576984406, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006148650031536818, + "mgd_loss_weight": 0.3668288916349411, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.1814910888672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.78659973144534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19000, + "valid_token_count": 818.11796875 + }, + { + "actual_token_mask_ratio": 0.3004901736974716, + "epoch": 5.530986325283678, + "flow_matching_loss": 0.009979232819750905, + "grad_norm": 0.07162734121084213, + "kept_token_count": 572.3, + "learning_rate": 9.726502825054195e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9936995923519134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006300410721451044, + "mgd_loss_weight": 0.3664955496788025, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.9949462890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.12957153320315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19010, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.29990961849689485, + "epoch": 5.533895839394821, + "flow_matching_loss": 0.01030686921440065, + "grad_norm": 0.07400213927030563, + "kept_token_count": 573.00078125, + "learning_rate": 9.711026860773595e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9939594447612763, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006040555750951171, + "mgd_loss_weight": 0.3661622077226639, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.3811065673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6615417480469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19020, + "valid_token_count": 818.46640625 + }, + { + "actual_token_mask_ratio": 0.2996016651391983, + "epoch": 5.536805353505964, + "flow_matching_loss": 0.011055416241288186, + "grad_norm": 0.08333735167980194, + "kept_token_count": 572.9625, + "learning_rate": 9.69555732309408e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9934569358825683, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065430637914687395, + "mgd_loss_weight": 0.36582885682582855, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.4906890869141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2170928955078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19030, + "valid_token_count": 818.0515625 + }, + { + "actual_token_mask_ratio": 0.3001322507858276, + "epoch": 5.539714867617108, + "flow_matching_loss": 0.012457999214529992, + "grad_norm": 0.09527970850467682, + "kept_token_count": 572.6453125, + "learning_rate": 9.68009423081259e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9938227891921997, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061772145796567205, + "mgd_loss_weight": 0.3654955178499222, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.0300598144531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.28147583007814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19040, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.3003238469362259, + "epoch": 5.542624381728252, + "flow_matching_loss": 0.011243397649377585, + "grad_norm": 0.06937180459499359, + "kept_token_count": 572.5671875, + "learning_rate": 9.664637602718249e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.99368234872818, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006317649781703949, + "mgd_loss_weight": 0.36516217291355135, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.63080139160155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3100067138672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19050, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.3003659784793854, + "epoch": 5.5455338958393945, + "flow_matching_loss": 0.011126477969810367, + "grad_norm": 0.08618732541799545, + "kept_token_count": 572.48671875, + "learning_rate": 9.649187457592313e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9933845579624176, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006615432910621166, + "mgd_loss_weight": 0.36482882499694824, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.4684204101562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.36070861816404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19060, + "valid_token_count": 818.26640625 + }, + { + "actual_token_mask_ratio": 0.3004184901714325, + "epoch": 5.548443409950538, + "flow_matching_loss": 0.011919931694865226, + "grad_norm": 0.08488604426383972, + "kept_token_count": 572.23984375, + "learning_rate": 9.63374381420816e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9935625910758972, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006437416886910796, + "mgd_loss_weight": 0.36449548602104187, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.4985046386719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.7088958740234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19070, + "valid_token_count": 817.9765625 + }, + { + "actual_token_mask_ratio": 0.29967363774776457, + "epoch": 5.551352924061682, + "flow_matching_loss": 0.0122257050126791, + "grad_norm": 0.08844752609729767, + "kept_token_count": 573.1015625, + "learning_rate": 9.618306691331279e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9934870421886444, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006512963445857167, + "mgd_loss_weight": 0.36416213512420653, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.9995513916016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.7070678710937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19080, + "valid_token_count": 818.33515625 + }, + { + "actual_token_mask_ratio": 0.29977545142173767, + "epoch": 5.554262438172826, + "flow_matching_loss": 0.012101226393133402, + "grad_norm": 0.06882021576166153, + "kept_token_count": 572.69609375, + "learning_rate": 9.602876107719228e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9940025269985199, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00599746941588819, + "mgd_loss_weight": 0.36382879316806793, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.708935546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.92982177734376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19090, + "valid_token_count": 817.87109375 + }, + { + "actual_token_mask_ratio": 0.30025399327278135, + "epoch": 5.557171952283968, + "flow_matching_loss": 0.011248981673270464, + "grad_norm": 0.07789470255374908, + "kept_token_count": 572.6734375, + "learning_rate": 9.587452082121621e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9937985658645629, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006201434135437012, + "mgd_loss_weight": 0.36349545419216156, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.93749084472654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.60400390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19100, + "valid_token_count": 818.40625 + }, + { + "actual_token_mask_ratio": 0.30008937418460846, + "epoch": 5.560081466395112, + "flow_matching_loss": 0.011503486568108201, + "grad_norm": 0.06841091811656952, + "kept_token_count": 572.67890625, + "learning_rate": 9.57203463328011e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9935245513916016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006475436920300126, + "mgd_loss_weight": 0.3631621032953262, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.7110992431641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.22588806152345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19110, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.3003283619880676, + "epoch": 5.562990980506255, + "flow_matching_loss": 0.01118373111821711, + "grad_norm": 0.06245448440313339, + "kept_token_count": 572.6546875, + "learning_rate": 9.556623779928341e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9935757875442505, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006424210779368877, + "mgd_loss_weight": 0.3628287613391876, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.0852813720703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3288299560547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19120, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.29930242002010343, + "epoch": 5.565900494617399, + "flow_matching_loss": 0.010604344075545669, + "grad_norm": 0.0772596076130867, + "kept_token_count": 573.33359375, + "learning_rate": 9.541219540791949e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9937188923358917, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062811087351292375, + "mgd_loss_weight": 0.362495419383049, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.9325378417969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9259552001953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19130, + "valid_token_count": 818.23515625 + }, + { + "actual_token_mask_ratio": 0.30010782182216644, + "epoch": 5.568810008728542, + "flow_matching_loss": 0.011971682123839856, + "grad_norm": 0.08002383261919022, + "kept_token_count": 572.75703125, + "learning_rate": 9.525821934588551e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9933409750461578, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066590080969035625, + "mgd_loss_weight": 0.3621620684862137, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.88978271484376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.16864318847655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19140, + "valid_token_count": 818.35390625 + }, + { + "actual_token_mask_ratio": 0.29954943656921384, + "epoch": 5.571719522839686, + "flow_matching_loss": 0.01173797519877553, + "grad_norm": 0.08244644850492477, + "kept_token_count": 573.09375, + "learning_rate": 9.510430980027684e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9936167538166046, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006383251957595348, + "mgd_loss_weight": 0.3618287295103073, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.6097808837891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.85870056152345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19150, + "valid_token_count": 818.17578125 + }, + { + "actual_token_mask_ratio": 0.3000094622373581, + "epoch": 5.574629036950829, + "flow_matching_loss": 0.011331204464659095, + "grad_norm": 0.05282296612858772, + "kept_token_count": 572.96640625, + "learning_rate": 9.495046695810809e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9938593864440918, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006140612158924341, + "mgd_loss_weight": 0.3614953845739365, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.11090393066405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3949279785156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19160, + "valid_token_count": 818.53125 + }, + { + "actual_token_mask_ratio": 0.2998738020658493, + "epoch": 5.577538551061973, + "flow_matching_loss": 0.01145780859515071, + "grad_norm": 0.08071529120206833, + "kept_token_count": 572.821875, + "learning_rate": 9.47966910063128e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.993798953294754, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006201048707589507, + "mgd_loss_weight": 0.3611620366573334, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.78929748535154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7736755371094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19170, + "valid_token_count": 818.16953125 + }, + { + "actual_token_mask_ratio": 0.3005418390035629, + "epoch": 5.580448065173116, + "flow_matching_loss": 0.011502055730670691, + "grad_norm": 0.08586666733026505, + "kept_token_count": 572.428125, + "learning_rate": 9.464298213174325e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9939544260501861, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006045566173270345, + "mgd_loss_weight": 0.360828697681427, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.53193359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.22637939453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19180, + "valid_token_count": 818.38828125 + }, + { + "actual_token_mask_ratio": 0.3003684550523758, + "epoch": 5.583357579284259, + "flow_matching_loss": 0.010479309968650341, + "grad_norm": 0.08107516169548035, + "kept_token_count": 572.50703125, + "learning_rate": 9.448934052117024e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.993450653553009, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006549346772953868, + "mgd_loss_weight": 0.36049534678459166, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.1511505126953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.04649353027344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19190, + "valid_token_count": 818.29765625 + }, + { + "actual_token_mask_ratio": 0.29994737207889555, + "epoch": 5.586267093395403, + "flow_matching_loss": 0.011625110916793347, + "grad_norm": 0.07740621268749237, + "kept_token_count": 572.84765625, + "learning_rate": 9.433576636128284e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9939774572849274, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006022539548575878, + "mgd_loss_weight": 0.36016200482845306, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 443.0421508789062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.10247497558595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19200, + "valid_token_count": 818.29609375 + }, + { + "actual_token_mask_ratio": 0.29965812265872954, + "epoch": 5.5891766075065465, + "flow_matching_loss": 0.010325937205925584, + "grad_norm": 0.07137511670589447, + "kept_token_count": 572.9171875, + "learning_rate": 9.41822598386881e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9938233017921447, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006176684424281121, + "mgd_loss_weight": 0.3598286658525467, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.398046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.23390808105466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19210, + "valid_token_count": 818.0546875 + }, + { + "actual_token_mask_ratio": 0.3001037538051605, + "epoch": 5.59208612161769, + "flow_matching_loss": 0.0126987480558455, + "grad_norm": 0.06905125081539154, + "kept_token_count": 572.7171875, + "learning_rate": 9.402882113991094e-06, + "loss": 0.0151, + "mgd_cos_sim": 0.9932384669780732, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006761533115059138, + "mgd_loss_weight": 0.3594953119754791, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 443.25661315917966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.55991516113284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19220, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.29995559751987455, + "epoch": 5.594995635728833, + "flow_matching_loss": 0.012813495472073555, + "grad_norm": 0.08193577826023102, + "kept_token_count": 572.9015625, + "learning_rate": 9.387545045139381e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9932777762413025, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006722220312803983, + "mgd_loss_weight": 0.35916197299957275, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.65325622558595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.4422668457031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19230, + "valid_token_count": 818.37890625 + }, + { + "actual_token_mask_ratio": 0.3001764744520187, + "epoch": 5.597905149839977, + "flow_matching_loss": 0.010675165522843599, + "grad_norm": 0.07274971157312393, + "kept_token_count": 572.86328125, + "learning_rate": 9.372214795949655e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9936481475830078, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063518471084535125, + "mgd_loss_weight": 0.3588286280632019, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 399.6182434082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.90268859863284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19240, + "valid_token_count": 818.58203125 + }, + { + "actual_token_mask_ratio": 0.2992553889751434, + "epoch": 5.6008146639511205, + "flow_matching_loss": 0.011965578887611628, + "grad_norm": 0.10000775754451752, + "kept_token_count": 573.39609375, + "learning_rate": 9.356891385049618e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9935430407524108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006456971494480967, + "mgd_loss_weight": 0.3584952801465988, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.5623291015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.58551025390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19250, + "valid_token_count": 818.26328125 + }, + { + "actual_token_mask_ratio": 0.3002893030643463, + "epoch": 5.603724178062263, + "flow_matching_loss": 0.011611841851845384, + "grad_norm": 0.06818930059671402, + "kept_token_count": 572.34765625, + "learning_rate": 9.341574831058664e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9934889376163483, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006511061033234, + "mgd_loss_weight": 0.35816194117069244, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.2156707763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2942413330078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19260, + "valid_token_count": 817.97109375 + }, + { + "actual_token_mask_ratio": 0.30064246952533724, + "epoch": 5.606633692173407, + "flow_matching_loss": 0.011689831595867872, + "grad_norm": 0.07246692478656769, + "kept_token_count": 572.3109375, + "learning_rate": 9.326265152587844e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9936744570732117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006325532402843237, + "mgd_loss_weight": 0.35782859325408933, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.1825378417969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 384.0146759033203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19270, + "valid_token_count": 818.33984375 + }, + { + "actual_token_mask_ratio": 0.29992619454860686, + "epoch": 5.609543206284551, + "flow_matching_loss": 0.011134331487119198, + "grad_norm": 0.06607566028833389, + "kept_token_count": 572.81796875, + "learning_rate": 9.310962368239861e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9932869017124176, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0067131098359823225, + "mgd_loss_weight": 0.3574952483177185, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.3080322265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.63604736328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19280, + "valid_token_count": 818.22578125 + }, + { + "actual_token_mask_ratio": 0.29962349236011504, + "epoch": 5.6124527203956935, + "flow_matching_loss": 0.012560648936778307, + "grad_norm": 0.0815054178237915, + "kept_token_count": 573.3109375, + "learning_rate": 9.295666496609036e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9935565114021301, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006443485477939248, + "mgd_loss_weight": 0.35716190934181213, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.5547180175781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7113830566406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19290, + "valid_token_count": 818.57890625 + }, + { + "actual_token_mask_ratio": 0.30027376413345336, + "epoch": 5.615362234506837, + "flow_matching_loss": 0.01107915760949254, + "grad_norm": 0.08703409135341644, + "kept_token_count": 572.58125, + "learning_rate": 9.280377556281306e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9936992347240448, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006300785252824426, + "mgd_loss_weight": 0.35682855546474457, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.87286987304685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8044036865234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19300, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.29947358965873716, + "epoch": 5.618271748617981, + "flow_matching_loss": 0.012986299488693476, + "grad_norm": 0.06566215306520462, + "kept_token_count": 573.12421875, + "learning_rate": 9.265095565834172e-06, + "loss": 0.0152, + "mgd_cos_sim": 0.9936920881271363, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006307896925136447, + "mgd_loss_weight": 0.3564952164888382, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.50391540527346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4678527832031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19310, + "valid_token_count": 818.13359375 + }, + { + "actual_token_mask_ratio": 0.29995146989822385, + "epoch": 5.621181262729124, + "flow_matching_loss": 0.010359179973602296, + "grad_norm": 0.09458748996257782, + "kept_token_count": 572.65234375, + "learning_rate": 9.249820543836688e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9936053812503814, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006394628575071693, + "mgd_loss_weight": 0.3561618775129318, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.3433502197266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8955078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19320, + "valid_token_count": 818.01484375 + }, + { + "actual_token_mask_ratio": 0.299906975030899, + "epoch": 5.6240907768402675, + "flow_matching_loss": 0.011774651799350977, + "grad_norm": 0.08589256554841995, + "kept_token_count": 572.76953125, + "learning_rate": 9.234552508849454e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9937180161476136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006281978450715542, + "mgd_loss_weight": 0.35582852363586426, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.63279113769534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0686096191406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19330, + "valid_token_count": 818.1328125 + }, + { + "actual_token_mask_ratio": 0.29901936948299407, + "epoch": 5.627000290951411, + "flow_matching_loss": 0.010538788791745902, + "grad_norm": 0.07188349217176437, + "kept_token_count": 573.76015625, + "learning_rate": 9.21929147942456e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9936535656452179, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006346435379236936, + "mgd_loss_weight": 0.3554951846599579, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.4184844970703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.03358459472656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19340, + "valid_token_count": 818.50390625 + }, + { + "actual_token_mask_ratio": 0.299909245967865, + "epoch": 5.629909805062555, + "flow_matching_loss": 0.010345186153426767, + "grad_norm": 0.06999395787715912, + "kept_token_count": 572.88984375, + "learning_rate": 9.204037474105611e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9939175605773926, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006082449806854129, + "mgd_loss_weight": 0.35516183972358706, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.1603271484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2457641601562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19350, + "valid_token_count": 818.3015625 + }, + { + "actual_token_mask_ratio": 0.30002636909484864, + "epoch": 5.632819319173698, + "flow_matching_loss": 0.011355128698050975, + "grad_norm": 0.06356323510408401, + "kept_token_count": 572.81953125, + "learning_rate": 9.188790511427653e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9934328377246857, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006567154219374061, + "mgd_loss_weight": 0.35482849180698395, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.6043426513672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4403381347656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19360, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.2992875099182129, + "epoch": 5.635728833284841, + "flow_matching_loss": 0.01136206705123186, + "grad_norm": 0.08517537266016006, + "kept_token_count": 573.50546875, + "learning_rate": 9.173550609917186e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9937497615814209, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006250241724774242, + "mgd_loss_weight": 0.3544951528310776, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.1851348876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1399230957031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19370, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.3001950651407242, + "epoch": 5.638638347395985, + "flow_matching_loss": 0.012499585701152682, + "grad_norm": 0.08157211542129517, + "kept_token_count": 572.49609375, + "learning_rate": 9.158317788092124e-06, + "loss": 0.0147, + "mgd_cos_sim": 0.9937695801258087, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006230411864817142, + "mgd_loss_weight": 0.35416180491447447, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.1267578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.973291015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19380, + "valid_token_count": 818.078125 + }, + { + "actual_token_mask_ratio": 0.29961723685264585, + "epoch": 5.641547861507128, + "flow_matching_loss": 0.011210598889738322, + "grad_norm": 0.06736494600772858, + "kept_token_count": 573.0109375, + "learning_rate": 9.143092064461783e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9932813704013824, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00671862093731761, + "mgd_loss_weight": 0.35382845997810364, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.0514892578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.6334747314453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19390, + "valid_token_count": 818.13671875 + }, + { + "actual_token_mask_ratio": 0.2996631473302841, + "epoch": 5.644457375618272, + "flow_matching_loss": 0.009995599556714297, + "grad_norm": 0.05108392983675003, + "kept_token_count": 573.02421875, + "learning_rate": 9.127873457526855e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9938311874866486, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006168812373653054, + "mgd_loss_weight": 0.35349512100219727, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.5509857177734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.31638488769534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19400, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.29959827065467837, + "epoch": 5.647366889729415, + "flow_matching_loss": 0.010841919388622046, + "grad_norm": 0.060289762914180756, + "kept_token_count": 573.0875, + "learning_rate": 9.112661985779382e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9938878178596496, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006112189032137394, + "mgd_loss_weight": 0.3531617671251297, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.8917175292969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6892517089844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19410, + "valid_token_count": 818.22734375 + }, + { + "actual_token_mask_ratio": 0.3001533567905426, + "epoch": 5.650276403840559, + "flow_matching_loss": 0.01258270237594843, + "grad_norm": 0.07551687955856323, + "kept_token_count": 572.68359375, + "learning_rate": 9.09745766770274e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9936790108680725, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006320993881672621, + "mgd_loss_weight": 0.3528284281492233, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.9718383789062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6871276855469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19420, + "valid_token_count": 818.30078125 + }, + { + "actual_token_mask_ratio": 0.30000483989715576, + "epoch": 5.653185917951702, + "flow_matching_loss": 0.011739444872364401, + "grad_norm": 0.07395312190055847, + "kept_token_count": 572.8390625, + "learning_rate": 9.082260521771601e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9931859135627746, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006814076378941536, + "mgd_loss_weight": 0.35249508619308473, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.52301330566405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3036804199219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19430, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.30078018009662627, + "epoch": 5.656095432062846, + "flow_matching_loss": 0.011483662528917193, + "grad_norm": 0.10187074542045593, + "kept_token_count": 572.2546875, + "learning_rate": 9.067070566451936e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9934806644916534, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065193539019674064, + "mgd_loss_weight": 0.3521617352962494, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.62835083007815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.62110290527346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19440, + "valid_token_count": 818.42109375 + }, + { + "actual_token_mask_ratio": 0.2999231666326523, + "epoch": 5.659004946173989, + "flow_matching_loss": 0.01203008317388594, + "grad_norm": 0.07968415319919586, + "kept_token_count": 573.01484375, + "learning_rate": 9.051887820200966e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9934814751148224, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006518526142463088, + "mgd_loss_weight": 0.351828396320343, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.87763061523435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5755584716797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19450, + "valid_token_count": 818.50390625 + }, + { + "actual_token_mask_ratio": 0.3004766255617142, + "epoch": 5.661914460285132, + "flow_matching_loss": 0.010655271355062723, + "grad_norm": 0.0668439120054245, + "kept_token_count": 572.36796875, + "learning_rate": 9.03671230146717e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9937326848506928, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006267299596220255, + "mgd_loss_weight": 0.3514950484037399, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.82781372070315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.09336547851564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19460, + "valid_token_count": 818.22890625 + }, + { + "actual_token_mask_ratio": 0.30040003061294557, + "epoch": 5.664823974396276, + "flow_matching_loss": 0.011470174416899681, + "grad_norm": 0.16120536625385284, + "kept_token_count": 572.3265625, + "learning_rate": 9.021544028690228e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9936924815177918, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006307526072487235, + "mgd_loss_weight": 0.3511617034673691, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.98244323730466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.998486328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19470, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.2995631545782089, + "epoch": 5.6677334885074195, + "flow_matching_loss": 0.010579349379986525, + "grad_norm": 0.07518875598907471, + "kept_token_count": 573.21875, + "learning_rate": 9.006383020301024e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.99376300573349, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062369951512664555, + "mgd_loss_weight": 0.3508283644914627, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.1311889648438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7821746826172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19480, + "valid_token_count": 818.36953125 + }, + { + "actual_token_mask_ratio": 0.30071179270744325, + "epoch": 5.670643002618562, + "flow_matching_loss": 0.010040445625782013, + "grad_norm": 0.07561884820461273, + "kept_token_count": 572.11953125, + "learning_rate": 8.991229294721613e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9934797167778016, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006520278099924326, + "mgd_loss_weight": 0.35049501359462737, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.6026580810547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.889306640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19490, + "valid_token_count": 818.15078125 + }, + { + "actual_token_mask_ratio": 0.2989063382148743, + "epoch": 5.673552516729706, + "flow_matching_loss": 0.012617651466280222, + "grad_norm": 0.07415249943733215, + "kept_token_count": 573.74296875, + "learning_rate": 8.976082870365191e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9938981235027313, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006101879291236401, + "mgd_loss_weight": 0.35016167163848877, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.55906982421874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.1021667480469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19500, + "valid_token_count": 818.35625 + }, + { + "actual_token_mask_ratio": 0.30048213005065916, + "epoch": 5.67646203084085, + "flow_matching_loss": 0.010311792558059096, + "grad_norm": 0.06631682068109512, + "kept_token_count": 572.36328125, + "learning_rate": 8.960943765636105e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9936904191970826, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006309568881988525, + "mgd_loss_weight": 0.34982832968235017, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.8584259033203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.17920837402346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19510, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.29998715221881866, + "epoch": 5.679371544951993, + "flow_matching_loss": 0.01089980723336339, + "grad_norm": 0.06860236078500748, + "kept_token_count": 572.828125, + "learning_rate": 8.945811998929792e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9937852323055267, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006214768346399069, + "mgd_loss_weight": 0.34949497878551483, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.0640502929688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1708618164063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19520, + "valid_token_count": 818.31015625 + }, + { + "actual_token_mask_ratio": 0.2996267944574356, + "epoch": 5.682281059063136, + "flow_matching_loss": 0.01057049110531807, + "grad_norm": 0.07127043604850769, + "kept_token_count": 572.9796875, + "learning_rate": 8.93068758863277e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9937069952487946, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006293004844337702, + "mgd_loss_weight": 0.34916163980960846, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.5238494873047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8872955322266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19530, + "valid_token_count": 818.10390625 + }, + { + "actual_token_mask_ratio": 0.29978553056716917, + "epoch": 5.68519057317428, + "flow_matching_loss": 0.01088533434085548, + "grad_norm": 0.05959031730890274, + "kept_token_count": 573.01953125, + "learning_rate": 8.915570553122628e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9935198962688446, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064800985157489775, + "mgd_loss_weight": 0.34882829785346986, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.63889465332034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.65302734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19540, + "valid_token_count": 818.34921875 + }, + { + "actual_token_mask_ratio": 0.3000138461589813, + "epoch": 5.688100087285424, + "flow_matching_loss": 0.010567926755174994, + "grad_norm": 0.053179360926151276, + "kept_token_count": 572.85078125, + "learning_rate": 8.900460910767982e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9938233077526093, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006176694622263312, + "mgd_loss_weight": 0.3484949469566345, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.24400634765624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9837280273438, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19550, + "valid_token_count": 818.37265625 + }, + { + "actual_token_mask_ratio": 0.3000248819589615, + "epoch": 5.6910096013965665, + "flow_matching_loss": 0.010850480711087584, + "grad_norm": 0.0649169534444809, + "kept_token_count": 572.65546875, + "learning_rate": 8.885358679928486e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9931293964385987, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006870605144649744, + "mgd_loss_weight": 0.34816160798072815, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.6319061279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.89562072753904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19560, + "valid_token_count": 818.10859375 + }, + { + "actual_token_mask_ratio": 0.30045575797557833, + "epoch": 5.69391911550771, + "flow_matching_loss": 0.010096576856449246, + "grad_norm": 0.07102181017398834, + "kept_token_count": 572.38515625, + "learning_rate": 8.870263878954767e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9936605572700501, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006339441239833832, + "mgd_loss_weight": 0.34782826006412504, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.26854248046874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5679046630859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19570, + "valid_token_count": 818.22421875 + }, + { + "actual_token_mask_ratio": 0.2997722774744034, + "epoch": 5.696828629618854, + "flow_matching_loss": 0.011521022953093051, + "grad_norm": 0.07177283614873886, + "kept_token_count": 573.0109375, + "learning_rate": 8.855176526188428e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9933015882968903, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006698403460904956, + "mgd_loss_weight": 0.3474949151277542, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.8887603759766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.67041015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19580, + "valid_token_count": 818.31640625 + }, + { + "actual_token_mask_ratio": 0.2996276289224625, + "epoch": 5.699738143729997, + "flow_matching_loss": 0.011730369459837676, + "grad_norm": 0.06456068158149719, + "kept_token_count": 573.00859375, + "learning_rate": 8.840096639962033e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9939833343029022, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00601666490547359, + "mgd_loss_weight": 0.34716157615184784, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.3316253662109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8564483642578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19590, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.3002373963594437, + "epoch": 5.7026476578411405, + "flow_matching_loss": 0.010789127368479968, + "grad_norm": 0.07934670895338058, + "kept_token_count": 572.5640625, + "learning_rate": 8.825024238599056e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9936582267284393, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006341777089983225, + "mgd_loss_weight": 0.3468282252550125, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.909912109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.10516357421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19600, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.29988707304000856, + "epoch": 5.705557171952284, + "flow_matching_loss": 0.012021061312407256, + "grad_norm": 0.09400874376296997, + "kept_token_count": 573.0171875, + "learning_rate": 8.809959340413891e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9935707807540893, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006429210212081671, + "mgd_loss_weight": 0.3464948832988739, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.34776916503904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.21543579101564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19610, + "valid_token_count": 818.46328125 + }, + { + "actual_token_mask_ratio": 0.2996315896511078, + "epoch": 5.708466686063428, + "flow_matching_loss": 0.011358830239623785, + "grad_norm": 0.07525376230478287, + "kept_token_count": 573.1953125, + "learning_rate": 8.79490196371181e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9937946617603302, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006205330416560173, + "mgd_loss_weight": 0.3461615413427353, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.7662384033203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0804473876953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19620, + "valid_token_count": 818.41796875 + }, + { + "actual_token_mask_ratio": 0.29964872300624845, + "epoch": 5.711376200174571, + "flow_matching_loss": 0.011406667437404394, + "grad_norm": 0.09584502130746841, + "kept_token_count": 573.12265625, + "learning_rate": 8.779852126788949e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.993878698348999, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006121295969933271, + "mgd_loss_weight": 0.34582819044589996, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.46669616699216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8713348388672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19630, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.29960426688194275, + "epoch": 5.714285714285714, + "flow_matching_loss": 0.01233352730050683, + "grad_norm": 0.06104659661650658, + "kept_token_count": 573.0109375, + "learning_rate": 8.764809847932277e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9936247229576111, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006375277834013105, + "mgd_loss_weight": 0.3454948514699936, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.1289001464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.55615234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19640, + "valid_token_count": 818.12421875 + }, + { + "actual_token_mask_ratio": 0.30040156841278076, + "epoch": 5.717195228396857, + "flow_matching_loss": 0.010578203108161688, + "grad_norm": 0.07053956389427185, + "kept_token_count": 572.446875, + "learning_rate": 8.749775145419577e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9934972286224365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006502771424129606, + "mgd_loss_weight": 0.34516150653362276, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.1946319580078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.85394897460935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19650, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.29964389503002165, + "epoch": 5.720104742508001, + "flow_matching_loss": 0.013516196608543396, + "grad_norm": 0.06954067945480347, + "kept_token_count": 573.12890625, + "learning_rate": 8.734748037519427e-06, + "loss": 0.0159, + "mgd_cos_sim": 0.9931191146373749, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006880894908681512, + "mgd_loss_weight": 0.34482815861701965, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.28114318847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1196044921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19660, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.3005419671535492, + "epoch": 5.723014256619145, + "flow_matching_loss": 0.012241119612008333, + "grad_norm": 0.07786216586828232, + "kept_token_count": 572.228125, + "learning_rate": 8.719728542491195e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9937419116497039, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006258083740249276, + "mgd_loss_weight": 0.3444948196411133, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.1394775390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2842254638672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19670, + "valid_token_count": 818.1 + }, + { + "actual_token_mask_ratio": 0.3003386348485947, + "epoch": 5.725923770730288, + "flow_matching_loss": 0.012111278716474771, + "grad_norm": 0.07467461377382278, + "kept_token_count": 572.50390625, + "learning_rate": 8.704716678584976e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9938831627368927, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006116823293268681, + "mgd_loss_weight": 0.34416146874427794, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.5426879882813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1903564453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19680, + "valid_token_count": 818.25859375 + }, + { + "actual_token_mask_ratio": 0.30076282024383544, + "epoch": 5.728833284841431, + "flow_matching_loss": 0.012293602339923382, + "grad_norm": 0.09512116014957428, + "kept_token_count": 572.21328125, + "learning_rate": 8.689712464041596e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9932604253292083, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006739559303969145, + "mgd_loss_weight": 0.34382812678813934, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.61952209472656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1058868408203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19690, + "valid_token_count": 818.34453125 + }, + { + "actual_token_mask_ratio": 0.29997186958789823, + "epoch": 5.731742798952575, + "flow_matching_loss": 0.011467583198100328, + "grad_norm": 0.09058059006929398, + "kept_token_count": 572.734375, + "learning_rate": 8.674715917092592e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9936493933200836, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006350601231679321, + "mgd_loss_weight": 0.34349478781223297, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.30480041503904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2829650878906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19700, + "valid_token_count": 818.15703125 + }, + { + "actual_token_mask_ratio": 0.29976804554462433, + "epoch": 5.734652313063719, + "flow_matching_loss": 0.011791271530091763, + "grad_norm": 0.07413889467716217, + "kept_token_count": 572.9859375, + "learning_rate": 8.65972705596018e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9941430449485779, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005856953514739871, + "mgd_loss_weight": 0.3431614339351654, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.1599853515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.14881591796876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19710, + "valid_token_count": 818.284375 + }, + { + "actual_token_mask_ratio": 0.3000944435596466, + "epoch": 5.737561827174861, + "flow_matching_loss": 0.011089175753295422, + "grad_norm": 0.07744968682527542, + "kept_token_count": 572.69296875, + "learning_rate": 8.64474589885725e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9937959730625152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006204023119062185, + "mgd_loss_weight": 0.34282809495925903, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.8083251953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8960723876953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19720, + "valid_token_count": 818.23828125 + }, + { + "actual_token_mask_ratio": 0.3003578633069992, + "epoch": 5.740471341286005, + "flow_matching_loss": 0.01158153866417706, + "grad_norm": 0.07282426208257675, + "kept_token_count": 572.4078125, + "learning_rate": 8.629772463987315e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9934368252754211, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006563176214694977, + "mgd_loss_weight": 0.3424947500228882, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 406.82098999023435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9252655029297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19730, + "valid_token_count": 818.14609375 + }, + { + "actual_token_mask_ratio": 0.3000405699014664, + "epoch": 5.743380855397149, + "flow_matching_loss": 0.01141882138326764, + "grad_norm": 0.07396069914102554, + "kept_token_count": 572.74609375, + "learning_rate": 8.614806769544507e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9938831210136414, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00611688937060535, + "mgd_loss_weight": 0.3421614021062851, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.02691345214845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8207305908203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19740, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.29913336634635923, + "epoch": 5.7462903695082925, + "flow_matching_loss": 0.009583336301147938, + "grad_norm": 0.09808561205863953, + "kept_token_count": 573.4984375, + "learning_rate": 8.59984883371356e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9935733854770661, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006426607817411422, + "mgd_loss_weight": 0.3418280631303787, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.46974182128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.32945861816404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19750, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.3004812091588974, + "epoch": 5.749199883619435, + "flow_matching_loss": 0.010862581199035048, + "grad_norm": 0.06554583460092545, + "kept_token_count": 572.27890625, + "learning_rate": 8.584898674669772e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9938369929790497, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006163018895313144, + "mgd_loss_weight": 0.3414947181940079, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.0991180419922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0698181152344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19760, + "valid_token_count": 818.10625 + }, + { + "actual_token_mask_ratio": 0.3012265503406525, + "epoch": 5.752109397730579, + "flow_matching_loss": 0.011216892581433057, + "grad_norm": 0.0885036513209343, + "kept_token_count": 571.759375, + "learning_rate": 8.569956310579001e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9933403730392456, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066596366930752994, + "mgd_loss_weight": 0.3411613702774048, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.6094604492188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1984497070313, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19770, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.2997919023036957, + "epoch": 5.755018911841723, + "flow_matching_loss": 0.011055855266749858, + "grad_norm": 0.08238116651773453, + "kept_token_count": 573.0296875, + "learning_rate": 8.555021759597627e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9938462913036347, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006153716519474983, + "mgd_loss_weight": 0.3408280313014984, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.44915771484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.37757873535156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19780, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.2999589771032333, + "epoch": 5.757928425952866, + "flow_matching_loss": 0.01292268862016499, + "grad_norm": 0.08209927380084991, + "kept_token_count": 572.76640625, + "learning_rate": 8.54009503987253e-06, + "loss": 0.015, + "mgd_cos_sim": 0.9939817726612091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006018221471458673, + "mgd_loss_weight": 0.3404946804046631, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.05182189941405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.148974609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19790, + "valid_token_count": 818.19140625 + }, + { + "actual_token_mask_ratio": 0.3008793920278549, + "epoch": 5.760837940064009, + "flow_matching_loss": 0.010144555708393455, + "grad_norm": 0.06459800153970718, + "kept_token_count": 571.92265625, + "learning_rate": 8.525176169541095e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9932549357414245, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006745069520547986, + "mgd_loss_weight": 0.3401613384485245, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.7362976074219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9904327392578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19800, + "valid_token_count": 818.06015625 + }, + { + "actual_token_mask_ratio": 0.29987822473049164, + "epoch": 5.763747454175153, + "flow_matching_loss": 0.011952568124979735, + "grad_norm": 0.06753042340278625, + "kept_token_count": 572.94296875, + "learning_rate": 8.510265166731151e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9935063779354095, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006493623647838831, + "mgd_loss_weight": 0.3398279994726181, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.56865234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9575622558594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19810, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.2998169422149658, + "epoch": 5.766656968286296, + "flow_matching_loss": 0.011937314411625266, + "grad_norm": 0.0722125694155693, + "kept_token_count": 572.846875, + "learning_rate": 8.495362049560964e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9939526081085205, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006047393381595612, + "mgd_loss_weight": 0.33949464559555054, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.55365600585935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.06907958984374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19820, + "valid_token_count": 818.13671875 + }, + { + "actual_token_mask_ratio": 0.29989951848983765, + "epoch": 5.7695664823974395, + "flow_matching_loss": 0.01133268466219306, + "grad_norm": 0.08974557369947433, + "kept_token_count": 572.675, + "learning_rate": 8.480466836139241e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9935330092906952, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006466990942135454, + "mgd_loss_weight": 0.33916130661964417, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.1621429443359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.35132751464846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19830, + "valid_token_count": 817.9890625 + }, + { + "actual_token_mask_ratio": 0.29999454617500304, + "epoch": 5.772475996508583, + "flow_matching_loss": 0.011116931773722171, + "grad_norm": 0.07577739655971527, + "kept_token_count": 572.73125, + "learning_rate": 8.465579544565066e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9938161373138428, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006183864688500762, + "mgd_loss_weight": 0.33882796168327334, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.2484588623047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1569396972656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19840, + "valid_token_count": 818.18359375 + }, + { + "actual_token_mask_ratio": 0.3003461301326752, + "epoch": 5.775385510619726, + "flow_matching_loss": 0.012758881226181984, + "grad_norm": 0.08284754306077957, + "kept_token_count": 572.5234375, + "learning_rate": 8.450700192927897e-06, + "loss": 0.0148, + "mgd_cos_sim": 0.9939215123653412, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006078480649739504, + "mgd_loss_weight": 0.3384946137666702, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.35322265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.16864013671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19850, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.3002242922782898, + "epoch": 5.77829502473087, + "flow_matching_loss": 0.010833105584606529, + "grad_norm": 0.07865804433822632, + "kept_token_count": 572.8140625, + "learning_rate": 8.435828799307551e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9937428295612335, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006257163779810071, + "mgd_loss_weight": 0.33816127479076385, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.67157592773435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9373413085938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19860, + "valid_token_count": 818.56015625 + }, + { + "actual_token_mask_ratio": 0.2999060809612274, + "epoch": 5.7812045388420135, + "flow_matching_loss": 0.011219904106110335, + "grad_norm": 0.07993057370185852, + "kept_token_count": 572.9046875, + "learning_rate": 8.420965381774172e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9933614194393158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006638577859848738, + "mgd_loss_weight": 0.33782792687416074, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.9732421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1765411376953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19870, + "valid_token_count": 818.32421875 + }, + { + "actual_token_mask_ratio": 0.3002810388803482, + "epoch": 5.784114052953157, + "flow_matching_loss": 0.011289227101951838, + "grad_norm": 0.07639918476343155, + "kept_token_count": 572.6328125, + "learning_rate": 8.406109958388222e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9939154267311097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006084591057151556, + "mgd_loss_weight": 0.3374945819377899, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.51063537597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.93556213378906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19880, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.30040291845798495, + "epoch": 5.7870235670643, + "flow_matching_loss": 0.011365941539406777, + "grad_norm": 0.07267750054597855, + "kept_token_count": 572.40859375, + "learning_rate": 8.391262547200434e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.993704330921173, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006295682583004236, + "mgd_loss_weight": 0.33716124296188354, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.742578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8862609863281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19890, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.29973621368408204, + "epoch": 5.789933081175444, + "flow_matching_loss": 0.01217379942536354, + "grad_norm": 0.08797014504671097, + "kept_token_count": 572.8140625, + "learning_rate": 8.376423166251816e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.993989247083664, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006010746210813522, + "mgd_loss_weight": 0.336827889084816, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.15798034667966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2560974121094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19900, + "valid_token_count": 817.99609375 + }, + { + "actual_token_mask_ratio": 0.2999363154172897, + "epoch": 5.792842595286587, + "flow_matching_loss": 0.011149276234209537, + "grad_norm": 0.08732225000858307, + "kept_token_count": 572.809375, + "learning_rate": 8.361591833573612e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9937326312065125, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006267357431352138, + "mgd_loss_weight": 0.3364945501089096, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.7188385009766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1280059814453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19910, + "valid_token_count": 818.22734375 + }, + { + "actual_token_mask_ratio": 0.29989500641822814, + "epoch": 5.79575210939773, + "flow_matching_loss": 0.011892574466764927, + "grad_norm": 0.07730259001255035, + "kept_token_count": 572.83046875, + "learning_rate": 8.346768567187287e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9938346147537231, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006165387993678451, + "mgd_loss_weight": 0.336161208152771, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.4283477783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.57685546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19920, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.29971531629562376, + "epoch": 5.798661623508874, + "flow_matching_loss": 0.01071631433442235, + "grad_norm": 0.07702703028917313, + "kept_token_count": 573.13984375, + "learning_rate": 8.331953385104516e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9937953531742096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006204665033146739, + "mgd_loss_weight": 0.33582785725593567, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.0905059814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7559753417969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19930, + "valid_token_count": 818.4359375 + }, + { + "actual_token_mask_ratio": 0.299578532576561, + "epoch": 5.801571137620018, + "flow_matching_loss": 0.010949266143143177, + "grad_norm": 0.09025603532791138, + "kept_token_count": 573.13046875, + "learning_rate": 8.317146305327134e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9938167214393616, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006183270737528801, + "mgd_loss_weight": 0.3354945182800293, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.8208862304688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.73240966796874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19940, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.3005897641181946, + "epoch": 5.804480651731161, + "flow_matching_loss": 0.012028120364993811, + "grad_norm": 0.06229160726070404, + "kept_token_count": 572.31796875, + "learning_rate": 8.302347345847146e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9938553035259247, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00614470150321722, + "mgd_loss_weight": 0.3351611703634262, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.5957733154297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.86607360839844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19950, + "valid_token_count": 818.28671875 + }, + { + "actual_token_mask_ratio": 0.30081748962402344, + "epoch": 5.807390165842304, + "flow_matching_loss": 0.011654919665306806, + "grad_norm": 0.06844808906316757, + "kept_token_count": 571.9984375, + "learning_rate": 8.287556524646683e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9932896792888641, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006710320012643933, + "mgd_loss_weight": 0.33482782542705536, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.8468505859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3895202636719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19960, + "valid_token_count": 818.09765625 + }, + { + "actual_token_mask_ratio": 0.2998737871646881, + "epoch": 5.810299679953448, + "flow_matching_loss": 0.010446025663986802, + "grad_norm": 0.06234511733055115, + "kept_token_count": 573.0671875, + "learning_rate": 8.272773859697978e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9938498497009277, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006150151370093226, + "mgd_loss_weight": 0.334494486451149, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.96490173339845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5456176757813, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19970, + "valid_token_count": 818.51953125 + }, + { + "actual_token_mask_ratio": 0.3003032594919205, + "epoch": 5.813209194064592, + "flow_matching_loss": 0.010242288932204247, + "grad_norm": 0.09381406009197235, + "kept_token_count": 572.5984375, + "learning_rate": 8.257999368963374e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9933690190315246, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006630977196618914, + "mgd_loss_weight": 0.3341611385345459, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.5782104492188, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.61399841308594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19980, + "valid_token_count": 818.34453125 + }, + { + "actual_token_mask_ratio": 0.30034383833408357, + "epoch": 5.816118708175734, + "flow_matching_loss": 0.010902974382042884, + "grad_norm": 0.07270459085702896, + "kept_token_count": 572.49140625, + "learning_rate": 8.243233070395266e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9936279475688934, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00637204609811306, + "mgd_loss_weight": 0.33382779359817505, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.2377227783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.87709655761716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 19990, + "valid_token_count": 818.24765625 + }, + { + "actual_token_mask_ratio": 0.3000833600759506, + "epoch": 5.819028222286878, + "flow_matching_loss": 0.011811060830950737, + "grad_norm": 0.07334659993648529, + "kept_token_count": 572.71875, + "learning_rate": 8.2284749819361e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9934639930725098, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00653601479716599, + "mgd_loss_weight": 0.3334944546222687, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.85296936035155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4255096435547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20000, + "valid_token_count": 818.26484375 + }, + { + "actual_token_mask_ratio": 0.29955987334251405, + "epoch": 5.821937736398022, + "flow_matching_loss": 0.012186832260340452, + "grad_norm": 0.08287933468818665, + "kept_token_count": 573.2234375, + "learning_rate": 8.213725121518346e-06, + "loss": 0.0142, + "mgd_cos_sim": 0.9940626859664917, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005937319621443749, + "mgd_loss_weight": 0.3331611007452011, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.2360107421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.26221618652346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20010, + "valid_token_count": 818.37890625 + }, + { + "actual_token_mask_ratio": 0.29993629455566406, + "epoch": 5.824847250509165, + "flow_matching_loss": 0.012416324391961098, + "grad_norm": 0.09932850301265717, + "kept_token_count": 572.79765625, + "learning_rate": 8.198983507064469e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9934516847133636, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006548315612599253, + "mgd_loss_weight": 0.33282776176929474, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.87373046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5083984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20020, + "valid_token_count": 818.2125 + }, + { + "actual_token_mask_ratio": 0.3000744253396988, + "epoch": 5.827756764620308, + "flow_matching_loss": 0.011926589254289865, + "grad_norm": 0.09331123530864716, + "kept_token_count": 572.81015625, + "learning_rate": 8.18425015648692e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9935760378837586, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006423950428143144, + "mgd_loss_weight": 0.33249441981315614, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.94280700683595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.12377014160154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20030, + "valid_token_count": 818.38984375 + }, + { + "actual_token_mask_ratio": 0.3005363166332245, + "epoch": 5.830666278731452, + "flow_matching_loss": 0.01125943623483181, + "grad_norm": 0.07442683726549149, + "kept_token_count": 572.328125, + "learning_rate": 8.169525087688122e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9938213467597962, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006178656220436096, + "mgd_loss_weight": 0.3321610689163208, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.48656005859374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7436492919922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20040, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.3002298265695572, + "epoch": 5.833575792842595, + "flow_matching_loss": 0.01228330391459167, + "grad_norm": 0.07384078949689865, + "kept_token_count": 572.56953125, + "learning_rate": 8.154808318560413e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9938039422035218, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006196052115410566, + "mgd_loss_weight": 0.33182772994041443, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.14013671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5037475585938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20050, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.2999027013778687, + "epoch": 5.836485306953739, + "flow_matching_loss": 0.010770367691293359, + "grad_norm": 0.06226833164691925, + "kept_token_count": 572.8296875, + "learning_rate": 8.140099866986058e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.993495762348175, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006504254136234522, + "mgd_loss_weight": 0.3314943820238113, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.82333984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4621215820313, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20060, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.2993886411190033, + "epoch": 5.839394821064882, + "flow_matching_loss": 0.010901590017601848, + "grad_norm": 0.0882171019911766, + "kept_token_count": 573.18046875, + "learning_rate": 8.125399750837202e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9937004208564758, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006299557350575924, + "mgd_loss_weight": 0.3311610370874405, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.0565704345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1798400878906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20070, + "valid_token_count": 818.11484375 + }, + { + "actual_token_mask_ratio": 0.3002815544605255, + "epoch": 5.842304335176026, + "flow_matching_loss": 0.011009199172258377, + "grad_norm": 0.07106735557317734, + "kept_token_count": 572.5625, + "learning_rate": 8.110707987975888e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9938578546047211, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006142158107832074, + "mgd_loss_weight": 0.3308276981115341, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.66908569335936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.952783203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20080, + "valid_token_count": 818.28203125 + }, + { + "actual_token_mask_ratio": 0.30003554821014405, + "epoch": 5.845213849287169, + "flow_matching_loss": 0.012002271134406328, + "grad_norm": 0.05814015492796898, + "kept_token_count": 572.74296875, + "learning_rate": 8.096024596253982e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9935961663722992, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006403831159695983, + "mgd_loss_weight": 0.3304943472146988, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 441.15979919433596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.57154235839846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20090, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.30000675320625303, + "epoch": 5.8481233633983125, + "flow_matching_loss": 0.01058085192926228, + "grad_norm": 0.09350869804620743, + "kept_token_count": 572.8578125, + "learning_rate": 8.081349593513197e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9939427196979522, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006057284073904157, + "mgd_loss_weight": 0.3301610052585602, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.5748657226562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3630737304687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20100, + "valid_token_count": 818.378125 + }, + { + "actual_token_mask_ratio": 0.29959882199764254, + "epoch": 5.851032877509456, + "flow_matching_loss": 0.010511578060686588, + "grad_norm": 0.0721055269241333, + "kept_token_count": 573.02109375, + "learning_rate": 8.066682997585045e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9936654567718506, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006334545649588108, + "mgd_loss_weight": 0.3298276633024216, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.916455078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.09071044921876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20110, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.2995157986879349, + "epoch": 5.853942391620599, + "flow_matching_loss": 0.010523962043225765, + "grad_norm": 0.07530540227890015, + "kept_token_count": 573.03125, + "learning_rate": 8.052024826290819e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9939142763614655, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060857359785586596, + "mgd_loss_weight": 0.32949431240558624, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.71099243164065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5658721923828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20120, + "valid_token_count": 818.05390625 + }, + { + "actual_token_mask_ratio": 0.2999647676944733, + "epoch": 5.856851905731743, + "flow_matching_loss": 0.012507011694833637, + "grad_norm": 0.08674941956996918, + "kept_token_count": 572.79296875, + "learning_rate": 8.037375097441574e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9938451945781708, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006154797086492181, + "mgd_loss_weight": 0.32916097342967987, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.1671081542969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.72454223632815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20130, + "valid_token_count": 818.23515625 + }, + { + "actual_token_mask_ratio": 0.2995296746492386, + "epoch": 5.8597614198428865, + "flow_matching_loss": 0.01103390771895647, + "grad_norm": 0.05480261147022247, + "kept_token_count": 573.2015625, + "learning_rate": 8.022733828838124e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.993850839138031, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006149164633825421, + "mgd_loss_weight": 0.32882762849330904, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.9833648681641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.12665100097655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20140, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.2991969853639603, + "epoch": 5.86267093395403, + "flow_matching_loss": 0.01186791630461812, + "grad_norm": 0.08032936602830887, + "kept_token_count": 573.4359375, + "learning_rate": 8.008101038270988e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9934490144252777, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00655098888091743, + "mgd_loss_weight": 0.32849428057670593, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.49840393066404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3726745605469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20150, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.2998010337352753, + "epoch": 5.865580448065173, + "flow_matching_loss": 0.012052909983322025, + "grad_norm": 0.11427802592515945, + "kept_token_count": 572.78984375, + "learning_rate": 7.99347674352038e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9937620043754578, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006237997347489, + "mgd_loss_weight": 0.32816094160079956, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.49635620117186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.21754760742186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20160, + "valid_token_count": 818.04375 + }, + { + "actual_token_mask_ratio": 0.29925687313079835, + "epoch": 5.868489962176317, + "flow_matching_loss": 0.010617707017809152, + "grad_norm": 0.05662163347005844, + "kept_token_count": 573.16484375, + "learning_rate": 7.978860962356202e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9937471985816956, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00625281292013824, + "mgd_loss_weight": 0.3278275907039642, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.0937561035156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8359405517578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20170, + "valid_token_count": 817.940625 + }, + { + "actual_token_mask_ratio": 0.29948112964630125, + "epoch": 5.8713994762874595, + "flow_matching_loss": 0.010962872579693795, + "grad_norm": 0.08194281160831451, + "kept_token_count": 573.19453125, + "learning_rate": 7.964253712537997e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9940723538398742, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005927640991285443, + "mgd_loss_weight": 0.3274942487478256, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.9871795654297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6451782226562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20180, + "valid_token_count": 818.23984375 + }, + { + "actual_token_mask_ratio": 0.2998574525117874, + "epoch": 5.874308990398603, + "flow_matching_loss": 0.010857793502509594, + "grad_norm": 0.07557865232229233, + "kept_token_count": 572.78828125, + "learning_rate": 7.949655011814968e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.993847131729126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00615286179818213, + "mgd_loss_weight": 0.32716090977191925, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.2593109130859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7362976074219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20190, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.30001200139522555, + "epoch": 5.877218504509747, + "flow_matching_loss": 0.010790162067860366, + "grad_norm": 0.07234112918376923, + "kept_token_count": 572.809375, + "learning_rate": 7.9350648779259e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9935039937496185, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006496008951216936, + "mgd_loss_weight": 0.3268275588750839, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.64072875976564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9001861572266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20200, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.3002327263355255, + "epoch": 5.880128018620891, + "flow_matching_loss": 0.010769706033170224, + "grad_norm": 0.06721478700637817, + "kept_token_count": 572.496875, + "learning_rate": 7.92048332859919e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9939605116844177, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006039477465674281, + "mgd_loss_weight": 0.3264942169189453, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.5869873046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.33836669921874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20210, + "valid_token_count": 818.12109375 + }, + { + "actual_token_mask_ratio": 0.3004975199699402, + "epoch": 5.8830375327320334, + "flow_matching_loss": 0.012173452693969012, + "grad_norm": 0.08589829504489899, + "kept_token_count": 572.3734375, + "learning_rate": 7.90591038155278e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9934634208679199, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006536571076139808, + "mgd_loss_weight": 0.3261608749628067, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.9395812988281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0004577636719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20220, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.29972347617149353, + "epoch": 5.885947046843177, + "flow_matching_loss": 0.01017321813851595, + "grad_norm": 0.059290818870067596, + "kept_token_count": 572.996875, + "learning_rate": 7.891346054494194e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9936068773269653, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006393121182918548, + "mgd_loss_weight": 0.3258275240659714, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.7237548828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0069213867188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20230, + "valid_token_count": 818.24453125 + }, + { + "actual_token_mask_ratio": 0.30078466832637785, + "epoch": 5.888856560954321, + "flow_matching_loss": 0.011311986995860935, + "grad_norm": 0.05737604200839996, + "kept_token_count": 572.1515625, + "learning_rate": 7.876790365120447e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9937385499477387, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062614426016807554, + "mgd_loss_weight": 0.325494185090065, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.0907867431641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9345947265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20240, + "valid_token_count": 818.28203125 + }, + { + "actual_token_mask_ratio": 0.30130939483642577, + "epoch": 5.891766075065464, + "flow_matching_loss": 0.01114671523682773, + "grad_norm": 0.09988988190889359, + "kept_token_count": 571.6203125, + "learning_rate": 7.862243331118088e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9933651864528656, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006634800741448998, + "mgd_loss_weight": 0.3251608401536942, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 447.02679443359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.75181274414064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20250, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.2993013858795166, + "epoch": 5.894675589176607, + "flow_matching_loss": 0.01018179403617978, + "grad_norm": 0.058462824672460556, + "kept_token_count": 573.38046875, + "learning_rate": 7.847704970163127e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9938945889472961, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061054055113345385, + "mgd_loss_weight": 0.32482749223709106, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.85340881347656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1476135253906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20260, + "valid_token_count": 818.30078125 + }, + { + "actual_token_mask_ratio": 0.30024356245994566, + "epoch": 5.897585103287751, + "flow_matching_loss": 0.011145165795460344, + "grad_norm": 0.09174038469791412, + "kept_token_count": 572.709375, + "learning_rate": 7.833175299921046e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9934607863426208, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006539201969280839, + "mgd_loss_weight": 0.3244941532611847, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.0901672363281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9013397216797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20270, + "valid_token_count": 818.43671875 + }, + { + "actual_token_mask_ratio": 0.3003327786922455, + "epoch": 5.900494617398895, + "flow_matching_loss": 0.010873712133616209, + "grad_norm": 0.07715019583702087, + "kept_token_count": 572.540625, + "learning_rate": 7.81865433804676e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.993620228767395, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00637975987046957, + "mgd_loss_weight": 0.32416080236434935, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.6710662841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.30873413085936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20280, + "valid_token_count": 818.30703125 + }, + { + "actual_token_mask_ratio": 0.2998621970415115, + "epoch": 5.903404131510038, + "flow_matching_loss": 0.011275323573499918, + "grad_norm": 0.086879663169384, + "kept_token_count": 572.9734375, + "learning_rate": 7.80414210218461e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9933681190013885, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006631868798285723, + "mgd_loss_weight": 0.32382746040821075, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.3142486572266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.44456787109374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20290, + "valid_token_count": 818.365625 + }, + { + "actual_token_mask_ratio": 0.30061024725437163, + "epoch": 5.906313645621181, + "flow_matching_loss": 0.010759176732972265, + "grad_norm": 0.08949145674705505, + "kept_token_count": 572.303125, + "learning_rate": 7.789638609968332e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.993444412946701, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00655558668076992, + "mgd_loss_weight": 0.3234941214323044, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.9201721191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.17665405273436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20300, + "valid_token_count": 818.28828125 + }, + { + "actual_token_mask_ratio": 0.29985488653182985, + "epoch": 5.909223159732325, + "flow_matching_loss": 0.01199712185189128, + "grad_norm": 0.07600690424442291, + "kept_token_count": 572.95, + "learning_rate": 7.775143879021037e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9936803698539733, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063196144066751, + "mgd_loss_weight": 0.3231607675552368, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.6126342773438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.17388305664065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20310, + "valid_token_count": 818.32734375 + }, + { + "actual_token_mask_ratio": 0.2993831396102905, + "epoch": 5.912132673843468, + "flow_matching_loss": 0.011389537760987879, + "grad_norm": 0.08276503533124924, + "kept_token_count": 573.1734375, + "learning_rate": 7.760657926955187e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9937862575054168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062137455213814976, + "mgd_loss_weight": 0.32282742857933044, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.9558502197266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.23125915527345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20320, + "valid_token_count": 818.1 + }, + { + "actual_token_mask_ratio": 0.2999826282262802, + "epoch": 5.915042187954612, + "flow_matching_loss": 0.010984756052494049, + "grad_norm": 0.09813408553600311, + "kept_token_count": 572.83828125, + "learning_rate": 7.746180771372584e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9934579610824585, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006542032351717353, + "mgd_loss_weight": 0.3224940836429596, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.5095642089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.30332641601564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20330, + "valid_token_count": 818.32265625 + }, + { + "actual_token_mask_ratio": 0.30020602643489835, + "epoch": 5.917951702065755, + "flow_matching_loss": 0.011008327407762409, + "grad_norm": 0.08008319139480591, + "kept_token_count": 572.66328125, + "learning_rate": 7.731712429864325e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9934662938117981, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006533716293051839, + "mgd_loss_weight": 0.3221607357263565, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.3478118896484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.55018310546876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20340, + "valid_token_count": 818.3375 + }, + { + "actual_token_mask_ratio": 0.29946569800376893, + "epoch": 5.920861216176898, + "flow_matching_loss": 0.011100087221711874, + "grad_norm": 0.06815819442272186, + "kept_token_count": 573.34765625, + "learning_rate": 7.717252920010828e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9939843535423278, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006015644920989871, + "mgd_loss_weight": 0.32182739675045013, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.36756286621096, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.84052124023435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20350, + "valid_token_count": 818.44140625 + }, + { + "actual_token_mask_ratio": 0.29996141195297243, + "epoch": 5.923770730288042, + "flow_matching_loss": 0.012233744747936726, + "grad_norm": 0.07708205282688141, + "kept_token_count": 572.78671875, + "learning_rate": 7.702802259381742e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9936214089393616, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063785785809159275, + "mgd_loss_weight": 0.321494048833847, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.4526611328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.347119140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20360, + "valid_token_count": 818.22421875 + }, + { + "actual_token_mask_ratio": 0.3004931092262268, + "epoch": 5.9266802443991855, + "flow_matching_loss": 0.011698713060468435, + "grad_norm": 0.07362585514783859, + "kept_token_count": 572.3625, + "learning_rate": 7.688360465535999e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9936127483844757, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006387255666777491, + "mgd_loss_weight": 0.3211607038974762, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.4859985351562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1340728759766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20370, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.2999460488557816, + "epoch": 5.929589758510328, + "flow_matching_loss": 0.011322792898863555, + "grad_norm": 0.08260619640350342, + "kept_token_count": 572.9, + "learning_rate": 7.673927556021734e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9939584791660309, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006041530426591635, + "mgd_loss_weight": 0.3208273649215698, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.7413604736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8264556884766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20380, + "valid_token_count": 818.365625 + }, + { + "actual_token_mask_ratio": 0.3001559227705002, + "epoch": 5.932499272621472, + "flow_matching_loss": 0.012039536889642478, + "grad_norm": 0.09256479144096375, + "kept_token_count": 572.8640625, + "learning_rate": 7.659503548376284e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9936170995235443, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006382909463718533, + "mgd_loss_weight": 0.32049401104450226, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.84681091308596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7689910888672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20390, + "valid_token_count": 818.55859375 + }, + { + "actual_token_mask_ratio": 0.30001193284988403, + "epoch": 5.935408786732616, + "flow_matching_loss": 0.010144712310284376, + "grad_norm": 0.10510845482349396, + "kept_token_count": 573.02890625, + "learning_rate": 7.645088460126193e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9936217784881591, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006378218811005354, + "mgd_loss_weight": 0.3201606720685959, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.1902282714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9285125732422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20400, + "valid_token_count": 818.62421875 + }, + { + "actual_token_mask_ratio": 0.3000218838453293, + "epoch": 5.9383183008437594, + "flow_matching_loss": 0.011536234710365533, + "grad_norm": 0.07453994452953339, + "kept_token_count": 572.71015625, + "learning_rate": 7.630682308787145e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.993771904706955, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006228099763393402, + "mgd_loss_weight": 0.3198273330926895, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.0857452392578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.81206665039065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20410, + "valid_token_count": 818.18359375 + }, + { + "actual_token_mask_ratio": 0.29990247189998626, + "epoch": 5.941227814954902, + "flow_matching_loss": 0.01138515374623239, + "grad_norm": 0.06877856701612473, + "kept_token_count": 572.87109375, + "learning_rate": 7.6162851118639754e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9938407242298126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006159282987937331, + "mgd_loss_weight": 0.31949397921562195, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.68329467773435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5671447753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20420, + "valid_token_count": 818.27421875 + }, + { + "actual_token_mask_ratio": 0.3003089964389801, + "epoch": 5.944137329066046, + "flow_matching_loss": 0.010283705173060298, + "grad_norm": 0.08514036983251572, + "kept_token_count": 572.4453125, + "learning_rate": 7.601896886850632e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.993606698513031, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006393305398523808, + "mgd_loss_weight": 0.3191606402397156, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.41807556152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1931915283203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20430, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.29884772300720214, + "epoch": 5.94704684317719, + "flow_matching_loss": 0.012095072912052274, + "grad_norm": 0.07742153853178024, + "kept_token_count": 573.86328125, + "learning_rate": 7.587517651230168e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.993607473373413, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006392532493919134, + "mgd_loss_weight": 0.31882729530334475, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.72547912597656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.14781494140624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20440, + "valid_token_count": 818.45234375 + }, + { + "actual_token_mask_ratio": 0.30026782751083375, + "epoch": 5.9499563572883325, + "flow_matching_loss": 0.011529228556901217, + "grad_norm": 0.09260149300098419, + "kept_token_count": 572.64453125, + "learning_rate": 7.573147422474702e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9937388002872467, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006261195614933967, + "mgd_loss_weight": 0.31849394738674164, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.57789916992186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6243072509766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20450, + "valid_token_count": 818.3796875 + }, + { + "actual_token_mask_ratio": 0.29998632669448855, + "epoch": 5.952865871399476, + "flow_matching_loss": 0.010382347740232944, + "grad_norm": 0.059601571410894394, + "kept_token_count": 572.8890625, + "learning_rate": 7.55878621804543e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9940160512924194, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005983953969553113, + "mgd_loss_weight": 0.31816060841083527, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.37042541503905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.16282958984374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20460, + "valid_token_count": 818.3921875 + }, + { + "actual_token_mask_ratio": 0.30062885880470275, + "epoch": 5.95577538551062, + "flow_matching_loss": 0.010958776948973537, + "grad_norm": 0.061576735228300095, + "kept_token_count": 572.3546875, + "learning_rate": 7.544434055392564e-06, + "loss": 0.013, + "mgd_cos_sim": 0.993679016828537, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006320991367101669, + "mgd_loss_weight": 0.31782726049423216, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.8788116455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.86762084960935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20470, + "valid_token_count": 818.38203125 + }, + { + "actual_token_mask_ratio": 0.2998837471008301, + "epoch": 5.958684899621764, + "flow_matching_loss": 0.011163070891052484, + "grad_norm": 0.0604521781206131, + "kept_token_count": 573.0640625, + "learning_rate": 7.530090951955334e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9938037276268006, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00619628606364131, + "mgd_loss_weight": 0.31749391555786133, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.75987243652344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8399383544922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20480, + "valid_token_count": 818.5296875 + }, + { + "actual_token_mask_ratio": 0.3007557034492493, + "epoch": 5.961594413732906, + "flow_matching_loss": 0.011614682897925378, + "grad_norm": 0.06872782111167908, + "kept_token_count": 572.2375, + "learning_rate": 7.515756925161959e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9938186645507813, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061813437845557925, + "mgd_loss_weight": 0.31716057658195496, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.8995056152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2860382080078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20490, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.29956044554710387, + "epoch": 5.96450392784405, + "flow_matching_loss": 0.010847051860764623, + "grad_norm": 0.08978058397769928, + "kept_token_count": 573.1796875, + "learning_rate": 7.501431992429643e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9937556207180023, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006244379281997681, + "mgd_loss_weight": 0.3168272227048874, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.0824310302734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.29344787597654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20500, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.29984281659126283, + "epoch": 5.967413441955194, + "flow_matching_loss": 0.011135624442249536, + "grad_norm": 0.06350160390138626, + "kept_token_count": 572.8453125, + "learning_rate": 7.487116171164519e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9936589419841766, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006341063184663653, + "mgd_loss_weight": 0.316493883728981, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.94935913085936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8943603515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20510, + "valid_token_count": 818.16484375 + }, + { + "actual_token_mask_ratio": 0.2996444761753082, + "epoch": 5.970322956066337, + "flow_matching_loss": 0.011482763662934303, + "grad_norm": 0.08323454111814499, + "kept_token_count": 572.9671875, + "learning_rate": 7.472809478761672e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9937821447849273, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006217868952080608, + "mgd_loss_weight": 0.3161605417728424, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.46396789550784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.38413391113284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20520, + "valid_token_count": 818.11015625 + }, + { + "actual_token_mask_ratio": 0.30075863003730774, + "epoch": 5.97323247017748, + "flow_matching_loss": 0.010854514176025986, + "grad_norm": 0.06660028547048569, + "kept_token_count": 572.2546875, + "learning_rate": 7.4585119326050775e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9940764248371124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005923578981310129, + "mgd_loss_weight": 0.3158271908760071, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.01332397460936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4509735107422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20530, + "valid_token_count": 818.3890625 + }, + { + "actual_token_mask_ratio": 0.30057321190834047, + "epoch": 5.976141984288624, + "flow_matching_loss": 0.010427941661328077, + "grad_norm": 0.07132350653409958, + "kept_token_count": 572.3875, + "learning_rate": 7.444223550067599e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9934007525444031, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065992476884275675, + "mgd_loss_weight": 0.3154938519001007, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.35570068359374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8391479492187, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20540, + "valid_token_count": 818.36484375 + }, + { + "actual_token_mask_ratio": 0.3006428211927414, + "epoch": 5.979051498399767, + "flow_matching_loss": 0.010601214179769159, + "grad_norm": 0.075096994638443, + "kept_token_count": 572.284375, + "learning_rate": 7.429944348510964e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.993855893611908, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00614411891438067, + "mgd_loss_weight": 0.3151605039834976, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.13092346191405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9643127441406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20550, + "valid_token_count": 818.29765625 + }, + { + "actual_token_mask_ratio": 0.30053464472293856, + "epoch": 5.981961012510911, + "flow_matching_loss": 0.01213545948266983, + "grad_norm": 0.0811978131532669, + "kept_token_count": 572.19375, + "learning_rate": 7.415674345285763e-06, + "loss": 0.0141, + "mgd_cos_sim": 0.9936405718326569, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006359423836693167, + "mgd_loss_weight": 0.31482715904712677, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.8683135986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2013641357422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20560, + "valid_token_count": 818.04453125 + }, + { + "actual_token_mask_ratio": 0.30012224018573763, + "epoch": 5.984870526622054, + "flow_matching_loss": 0.010987965296953917, + "grad_norm": 0.0868815928697586, + "kept_token_count": 572.6484375, + "learning_rate": 7.401413557731387e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9933909595012664, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006609037751331925, + "mgd_loss_weight": 0.3144938200712204, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.0251525878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2257873535156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20570, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.29989425241947176, + "epoch": 5.987780040733197, + "flow_matching_loss": 0.010337783163413406, + "grad_norm": 0.07013809680938721, + "kept_token_count": 572.96953125, + "learning_rate": 7.387162003176037e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9938437879085541, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006156206969171763, + "mgd_loss_weight": 0.31416046917438506, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.51690979003905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.25852661132814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20580, + "valid_token_count": 818.40390625 + }, + { + "actual_token_mask_ratio": 0.300201877951622, + "epoch": 5.990689554844341, + "flow_matching_loss": 0.011219706013798713, + "grad_norm": 0.08748108148574829, + "kept_token_count": 572.62734375, + "learning_rate": 7.372919698936697e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.993318235874176, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006681772507727146, + "mgd_loss_weight": 0.31382712721824646, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.59157409667966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8435974121094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20590, + "valid_token_count": 818.27421875 + }, + { + "actual_token_mask_ratio": 0.30052714943885805, + "epoch": 5.9935990689554846, + "flow_matching_loss": 0.012480244552716614, + "grad_norm": 0.0788656696677208, + "kept_token_count": 572.40234375, + "learning_rate": 7.358686662319104e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9938344836235047, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006165514187887311, + "mgd_loss_weight": 0.31349378526210786, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.0263244628906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3961975097656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20600, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.3006440162658691, + "epoch": 5.996508583066628, + "flow_matching_loss": 0.010621215589344502, + "grad_norm": 0.07822787761688232, + "kept_token_count": 572.37265625, + "learning_rate": 7.344462910617752e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9935938835144043, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006406114157289267, + "mgd_loss_weight": 0.3131604343652725, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.20062255859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.10480041503905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20610, + "valid_token_count": 818.4296875 + }, + { + "actual_token_mask_ratio": 0.29961768388748167, + "epoch": 5.999418097177771, + "flow_matching_loss": 0.01054895669221878, + "grad_norm": 0.1338951140642166, + "kept_token_count": 572.97265625, + "learning_rate": 7.330248461115832e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9935655057430267, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064344910439103845, + "mgd_loss_weight": 0.31282709538936615, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.24412536621094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.88897705078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20620, + "valid_token_count": 818.09296875 + }, + { + "actual_token_mask_ratio": 0.3005692958831787, + "epoch": 6.002327611288915, + "flow_matching_loss": 0.01064643245190382, + "grad_norm": 0.07192042469978333, + "kept_token_count": 572.2891967773437, + "learning_rate": 7.316043331085242e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9939044713973999, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006095510628074408, + "mgd_loss_weight": 0.31249375343322755, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.4535369873047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.6781524658203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20630, + "valid_token_count": 818.2268005371094 + }, + { + "actual_token_mask_ratio": 0.29960985779762267, + "epoch": 6.0052371254000585, + "flow_matching_loss": 0.011914541386067868, + "grad_norm": 0.08124234527349472, + "kept_token_count": 573.11171875, + "learning_rate": 7.30184753778655e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9932915806770325, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006708409637212753, + "mgd_loss_weight": 0.3121604025363922, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.43336181640626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.1530303955078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20640, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.30062074959278107, + "epoch": 6.008146639511201, + "flow_matching_loss": 0.011465153796598316, + "grad_norm": 0.0849519893527031, + "kept_token_count": 572.26796875, + "learning_rate": 7.2876610984689935e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9937621057033539, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006237888522446155, + "mgd_loss_weight": 0.31182706356048584, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.3003265380859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7996398925781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20650, + "valid_token_count": 818.25390625 + }, + { + "actual_token_mask_ratio": 0.29946031868457795, + "epoch": 6.011056153622345, + "flow_matching_loss": 0.011169994063675404, + "grad_norm": 0.07755821198225021, + "kept_token_count": 573.246875, + "learning_rate": 7.273484030370423e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9937744736671448, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006225514179095626, + "mgd_loss_weight": 0.31149371564388273, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.7098419189453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7549133300781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20660, + "valid_token_count": 818.29140625 + }, + { + "actual_token_mask_ratio": 0.2999468296766281, + "epoch": 6.013965667733489, + "flow_matching_loss": 0.011464084219187497, + "grad_norm": 0.07660321891307831, + "kept_token_count": 573.1171875, + "learning_rate": 7.259316350717324e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9937955975532532, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006204412551596761, + "mgd_loss_weight": 0.3111603707075119, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.0445556640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1619903564453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20670, + "valid_token_count": 818.684375 + }, + { + "actual_token_mask_ratio": 0.30008259117603303, + "epoch": 6.0168751818446315, + "flow_matching_loss": 0.010482214717194438, + "grad_norm": 0.06534358859062195, + "kept_token_count": 572.82734375, + "learning_rate": 7.245158076724757e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9939984858036042, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006001517362892628, + "mgd_loss_weight": 0.31082703173160553, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.48728332519534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.14873962402345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20680, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.2995750606060028, + "epoch": 6.019784695955775, + "flow_matching_loss": 0.010949086491018533, + "grad_norm": 0.0562606155872345, + "kept_token_count": 573.24296875, + "learning_rate": 7.23100922559636e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9937124252319336, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00628758231177926, + "mgd_loss_weight": 0.3104936808347702, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.41576232910154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0951690673828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20690, + "valid_token_count": 818.42734375 + }, + { + "actual_token_mask_ratio": 0.30033330619335175, + "epoch": 6.022694210066919, + "flow_matching_loss": 0.011906041018664836, + "grad_norm": 0.07403728365898132, + "kept_token_count": 572.63984375, + "learning_rate": 7.216869814524323e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9936582446098328, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006341741420328617, + "mgd_loss_weight": 0.3101603388786316, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.2163360595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8513427734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20700, + "valid_token_count": 818.44453125 + }, + { + "actual_token_mask_ratio": 0.29997304975986483, + "epoch": 6.025603724178063, + "flow_matching_loss": 0.010886959452182054, + "grad_norm": 0.06268255412578583, + "kept_token_count": 572.93046875, + "learning_rate": 7.202739860689357e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9935752868652343, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006424714252352715, + "mgd_loss_weight": 0.309826996922493, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.57615661621094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9198333740234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20710, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.3000580370426178, + "epoch": 6.0285132382892055, + "flow_matching_loss": 0.010296760872006417, + "grad_norm": 0.0669282078742981, + "kept_token_count": 572.63203125, + "learning_rate": 7.188619381260698e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9937277972698212, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006272215023636818, + "mgd_loss_weight": 0.30949364602565765, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.47825927734374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0687683105469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20720, + "valid_token_count": 818.109375 + }, + { + "actual_token_mask_ratio": 0.30081490576267245, + "epoch": 6.031422752400349, + "flow_matching_loss": 0.011866270657628774, + "grad_norm": 0.07262624055147171, + "kept_token_count": 572.0546875, + "learning_rate": 7.174508393396057e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9939648449420929, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060351453255862, + "mgd_loss_weight": 0.3091603070497513, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.93323669433596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.309765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20730, + "valid_token_count": 818.17265625 + }, + { + "actual_token_mask_ratio": 0.3005917012691498, + "epoch": 6.034332266511493, + "flow_matching_loss": 0.01108025098219514, + "grad_norm": 0.09202215075492859, + "kept_token_count": 572.25234375, + "learning_rate": 7.160406914241615e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9937241494655609, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006275868834927678, + "mgd_loss_weight": 0.30882696211338045, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.0683624267578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.986083984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20740, + "valid_token_count": 818.19609375 + }, + { + "actual_token_mask_ratio": 0.2999814093112946, + "epoch": 6.037241780622636, + "flow_matching_loss": 0.010548272170126438, + "grad_norm": 0.06502711772918701, + "kept_token_count": 572.8890625, + "learning_rate": 7.146314960931996e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9937187731266022, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006281225010752678, + "mgd_loss_weight": 0.30849361419677734, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.3305999755859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0964599609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20750, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.3001188635826111, + "epoch": 6.040151294733779, + "flow_matching_loss": 0.010652086604386568, + "grad_norm": 0.07711414247751236, + "kept_token_count": 572.6890625, + "learning_rate": 7.132232550590247e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9936121165752411, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006387878069654107, + "mgd_loss_weight": 0.30816027522087097, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.83603515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2286834716797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20760, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.3000844746828079, + "epoch": 6.043060808844923, + "flow_matching_loss": 0.010878533124923706, + "grad_norm": 0.06775128841400146, + "kept_token_count": 572.628125, + "learning_rate": 7.118159700327839e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9939321994781494, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006067805131897331, + "mgd_loss_weight": 0.30782692432403563, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.66062622070314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1787170410156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20770, + "valid_token_count": 818.14609375 + }, + { + "actual_token_mask_ratio": 0.2995046108961105, + "epoch": 6.045970322956066, + "flow_matching_loss": 0.01062195273116231, + "grad_norm": 0.06770741194486618, + "kept_token_count": 573.20859375, + "learning_rate": 7.104096427244603e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9934795379638672, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0065204652957618235, + "mgd_loss_weight": 0.30749358236789703, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.9638244628906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7217132568359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20780, + "valid_token_count": 818.29140625 + }, + { + "actual_token_mask_ratio": 0.2996883988380432, + "epoch": 6.04887983706721, + "flow_matching_loss": 0.011553592234849929, + "grad_norm": 0.09191600978374481, + "kept_token_count": 572.975, + "learning_rate": 7.090042748428737e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9939467012882233, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006053302250802517, + "mgd_loss_weight": 0.30716024339199066, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.66331481933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0718200683594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20790, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.2994033545255661, + "epoch": 6.051789351178353, + "flow_matching_loss": 0.010224373079836369, + "grad_norm": 0.07595989108085632, + "kept_token_count": 573.103125, + "learning_rate": 7.075998680956797e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9939724385738373, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006027560867369175, + "mgd_loss_weight": 0.3068268895149231, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.2565216064453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.32464904785155, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20800, + "valid_token_count": 818.0203125 + }, + { + "actual_token_mask_ratio": 0.29940215349197385, + "epoch": 6.054698865289497, + "flow_matching_loss": 0.009781909920275212, + "grad_norm": 0.07867738604545593, + "kept_token_count": 573.2625, + "learning_rate": 7.061964241893639e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9940830528736114, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005916947964578867, + "mgd_loss_weight": 0.3064935505390167, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.94171142578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3772247314453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20810, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.29979748725891114, + "epoch": 6.05760837940064, + "flow_matching_loss": 0.012109900079667568, + "grad_norm": 0.07805078476667404, + "kept_token_count": 573.04453125, + "learning_rate": 7.047939448292442e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9936699807643891, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006330013507977128, + "mgd_loss_weight": 0.3061602056026459, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.4583038330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.54603576660156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20820, + "valid_token_count": 818.4078125 + }, + { + "actual_token_mask_ratio": 0.3000959545373917, + "epoch": 6.060517893511784, + "flow_matching_loss": 0.010363908065482975, + "grad_norm": 0.0690731629729271, + "kept_token_count": 572.60078125, + "learning_rate": 7.033924317194644e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9936323285102844, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006367671862244606, + "mgd_loss_weight": 0.3058268576860428, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.6893585205078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.38352355957034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20830, + "valid_token_count": 818.1140625 + }, + { + "actual_token_mask_ratio": 0.3009469747543335, + "epoch": 6.063427407622927, + "flow_matching_loss": 0.010989739699289203, + "grad_norm": 0.07641322165727615, + "kept_token_count": 571.9234375, + "learning_rate": 7.0199188656299524e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9940135538578033, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005986457131803036, + "mgd_loss_weight": 0.3054935187101364, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.2646881103516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6398468017578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20840, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.2999622941017151, + "epoch": 6.06633692173407, + "flow_matching_loss": 0.010190214030444622, + "grad_norm": 0.06932828575372696, + "kept_token_count": 572.7578125, + "learning_rate": 7.005923110616313e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9934607863426208, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00653922506608069, + "mgd_loss_weight": 0.3051601707935333, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.71043090820314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.69297485351564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20850, + "valid_token_count": 818.18046875 + }, + { + "actual_token_mask_ratio": 0.3004938781261444, + "epoch": 6.069246435845214, + "flow_matching_loss": 0.010924496967345477, + "grad_norm": 0.07923614978790283, + "kept_token_count": 572.36796875, + "learning_rate": 6.9919370691598855e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9936130285263062, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006386969331651926, + "mgd_loss_weight": 0.3048268258571625, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.8115234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.48199768066405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20860, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.29941296577453613, + "epoch": 6.0721559499563575, + "flow_matching_loss": 0.010058399057015777, + "grad_norm": 0.06337424367666245, + "kept_token_count": 573.40859375, + "learning_rate": 6.977960758255024e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9936242461204529, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006375752855092287, + "mgd_loss_weight": 0.3044934868812561, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.13067626953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.55938415527345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20870, + "valid_token_count": 818.46640625 + }, + { + "actual_token_mask_ratio": 0.29959025979042053, + "epoch": 6.0750654640675, + "flow_matching_loss": 0.010768401622772216, + "grad_norm": 0.057007212191820145, + "kept_token_count": 573.1859375, + "learning_rate": 6.963994194884276e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.99393909573555, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006060901563614607, + "mgd_loss_weight": 0.30416013598442077, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.8892578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.07785949707034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20880, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.299619922041893, + "epoch": 6.077974978178644, + "flow_matching_loss": 0.011534259980544448, + "grad_norm": 0.07821928709745407, + "kept_token_count": 573.0734375, + "learning_rate": 6.950037396018328e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9936629593372345, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063370468094944956, + "mgd_loss_weight": 0.30382679402828217, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.3778594970703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.687646484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20890, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.29995315670967104, + "epoch": 6.080884492289788, + "flow_matching_loss": 0.011918793059885502, + "grad_norm": 0.0963151752948761, + "kept_token_count": 572.79296875, + "learning_rate": 6.9360903786160056e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9934402525424957, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006559750391170382, + "mgd_loss_weight": 0.3034934550523758, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.26439514160154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7587463378906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20900, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.30069929361343384, + "epoch": 6.083794006400931, + "flow_matching_loss": 0.010459432657808066, + "grad_norm": 0.08130116015672684, + "kept_token_count": 572.43203125, + "learning_rate": 6.9221531596242485e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9938077688217163, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006192233180627227, + "mgd_loss_weight": 0.3031601011753082, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.65684814453124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5637573242187, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20910, + "valid_token_count": 818.58046875 + }, + { + "actual_token_mask_ratio": 0.2988367348909378, + "epoch": 6.086703520512074, + "flow_matching_loss": 0.011590569000691175, + "grad_norm": 0.0661664679646492, + "kept_token_count": 573.73125, + "learning_rate": 6.908225755978089e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9939658939838409, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006034103315323591, + "mgd_loss_weight": 0.30282676219940186, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.51331787109376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.53062744140624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20920, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.2993671327829361, + "epoch": 6.089613034623218, + "flow_matching_loss": 0.010671762190759182, + "grad_norm": 0.08674667775630951, + "kept_token_count": 573.2984375, + "learning_rate": 6.8943081846006466e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9937120914459229, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006287913024425507, + "mgd_loss_weight": 0.30249341726303103, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.1412811279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.7472320556641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20930, + "valid_token_count": 818.25546875 + }, + { + "actual_token_mask_ratio": 0.29999233186244967, + "epoch": 6.092522548734362, + "flow_matching_loss": 0.011771371588110924, + "grad_norm": 0.09709430485963821, + "kept_token_count": 572.92265625, + "learning_rate": 6.88040046240307e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9934356808662415, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006564331147819758, + "mgd_loss_weight": 0.3021600693464279, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.3285705566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.88108215332034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20940, + "valid_token_count": 818.45234375 + }, + { + "actual_token_mask_ratio": 0.2999158650636673, + "epoch": 6.0954320628455045, + "flow_matching_loss": 0.010834109457209706, + "grad_norm": 0.08961021900177002, + "kept_token_count": 572.87265625, + "learning_rate": 6.866502606284566e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9940295040607452, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005970497149974108, + "mgd_loss_weight": 0.30182673037052155, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.46551818847655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4198303222656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20950, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.30014258325099946, + "epoch": 6.098341576956648, + "flow_matching_loss": 0.011215866217389703, + "grad_norm": 0.06622719019651413, + "kept_token_count": 572.6921875, + "learning_rate": 6.852614633132335e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9935010254383088, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006498963758349419, + "mgd_loss_weight": 0.30149338245391843, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.9732147216797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4195159912109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20960, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.2995686918497086, + "epoch": 6.101251091067792, + "flow_matching_loss": 0.011243579536676407, + "grad_norm": 0.07163788378238678, + "kept_token_count": 573.2359375, + "learning_rate": 6.8387365598215665e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9935492396354675, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00645075305365026, + "mgd_loss_weight": 0.3011600375175476, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.5368621826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.87367858886716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20970, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.30071710646152494, + "epoch": 6.104160605178935, + "flow_matching_loss": 0.0118942866101861, + "grad_norm": 0.08257906883955002, + "kept_token_count": 572.12890625, + "learning_rate": 6.82486840321544e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9935059487819672, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006494041066616774, + "mgd_loss_weight": 0.30082669854164124, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.75429077148436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1681304931641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20980, + "valid_token_count": 818.16171875 + }, + { + "actual_token_mask_ratio": 0.3006030708551407, + "epoch": 6.1070701192900785, + "flow_matching_loss": 0.012010886799544096, + "grad_norm": 0.07733793556690216, + "kept_token_count": 572.44296875, + "learning_rate": 6.811010180165069e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9937140047550201, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062860044650733474, + "mgd_loss_weight": 0.30049334466457367, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.32244567871095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9555694580078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 20990, + "valid_token_count": 818.47890625 + }, + { + "actual_token_mask_ratio": 0.2994963347911835, + "epoch": 6.109979633401222, + "flow_matching_loss": 0.010225282469764351, + "grad_norm": 0.08971462398767471, + "kept_token_count": 573.275, + "learning_rate": 6.797161907509499e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9936482787132264, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006351726595312357, + "mgd_loss_weight": 0.3001600056886673, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.3867218017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.40191650390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21000, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.2999829024076462, + "epoch": 6.112889147512365, + "flow_matching_loss": 0.010293244244530797, + "grad_norm": 0.07269442826509476, + "kept_token_count": 572.91953125, + "learning_rate": 6.783323602075688e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9936396718025208, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006360326893627644, + "mgd_loss_weight": 0.2998266637325287, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.42112426757814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.65325927734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21010, + "valid_token_count": 818.4375 + }, + { + "actual_token_mask_ratio": 0.30040728449821474, + "epoch": 6.115798661623509, + "flow_matching_loss": 0.009619570383802057, + "grad_norm": 0.08025580644607544, + "kept_token_count": 572.509375, + "learning_rate": 6.7694952806784784e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9939634144306183, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060365934856235984, + "mgd_loss_weight": 0.29949331283569336, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.3771728515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2736572265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21020, + "valid_token_count": 818.3484375 + }, + { + "actual_token_mask_ratio": 0.3006090193986893, + "epoch": 6.118708175734652, + "flow_matching_loss": 0.011213694978505374, + "grad_norm": 0.08205360919237137, + "kept_token_count": 572.31796875, + "learning_rate": 6.7556769601205944e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9938464999198914, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006153498869389295, + "mgd_loss_weight": 0.299159973859787, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.1830657958984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.27887268066405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21030, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.2998510032892227, + "epoch": 6.121617689845796, + "flow_matching_loss": 0.012547798128798604, + "grad_norm": 0.0655047744512558, + "kept_token_count": 572.76484375, + "learning_rate": 6.741868657192594e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9933491945266724, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00665079802274704, + "mgd_loss_weight": 0.2988266259431839, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.69019470214846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3066925048828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21040, + "valid_token_count": 818.0625 + }, + { + "actual_token_mask_ratio": 0.3003826975822449, + "epoch": 6.124527203956939, + "flow_matching_loss": 0.010951343411579727, + "grad_norm": 0.062143534421920776, + "kept_token_count": 572.565625, + "learning_rate": 6.728070388672864e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9936794757843017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006320533156394958, + "mgd_loss_weight": 0.29849328100681305, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.6868408203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.48475341796876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21050, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.29998851120471953, + "epoch": 6.127436718068083, + "flow_matching_loss": 0.010589044960215687, + "grad_norm": 0.05590308830142021, + "kept_token_count": 572.878125, + "learning_rate": 6.714282171327604e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9941075921058655, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005892405146732926, + "mgd_loss_weight": 0.2981599420309067, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.44482421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.63657836914064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21060, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.29974002242088316, + "epoch": 6.130346232179226, + "flow_matching_loss": 0.010372066823765635, + "grad_norm": 0.08266465365886688, + "kept_token_count": 573.22421875, + "learning_rate": 6.700504021910791e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9937283873558045, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062716030050069095, + "mgd_loss_weight": 0.29782659113407134, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.32496032714846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.27513732910154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21070, + "valid_token_count": 818.5890625 + }, + { + "actual_token_mask_ratio": 0.299567386507988, + "epoch": 6.133255746290369, + "flow_matching_loss": 0.009779440285637975, + "grad_norm": 0.07546624541282654, + "kept_token_count": 573.1859375, + "learning_rate": 6.68673595716419e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9940612316131592, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005938765406608581, + "mgd_loss_weight": 0.29749324917793274, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.1113708496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.97783203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21080, + "valid_token_count": 818.33046875 + }, + { + "actual_token_mask_ratio": 0.300093537569046, + "epoch": 6.136165260401513, + "flow_matching_loss": 0.011355542112141848, + "grad_norm": 0.06533528864383698, + "kept_token_count": 572.66640625, + "learning_rate": 6.6729779938172856e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9940608978271485, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059391212649643425, + "mgd_loss_weight": 0.29715991020202637, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.8116027832031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8416046142578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21090, + "valid_token_count": 818.2 + }, + { + "actual_token_mask_ratio": 0.29987596571445463, + "epoch": 6.139074774512657, + "flow_matching_loss": 0.01038518869318068, + "grad_norm": 0.07813818007707596, + "kept_token_count": 572.85390625, + "learning_rate": 6.659230148587311e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9934764385223389, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006523569533601403, + "mgd_loss_weight": 0.2968265563249588, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.3636810302734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.204052734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21100, + "valid_token_count": 818.22109375 + }, + { + "actual_token_mask_ratio": 0.3007997453212738, + "epoch": 6.141984288623799, + "flow_matching_loss": 0.01058632773347199, + "grad_norm": 0.08095894008874893, + "kept_token_count": 572.11328125, + "learning_rate": 6.645492438179188e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.993845957517624, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006154040107503533, + "mgd_loss_weight": 0.29649321734905243, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 439.8665771484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.83807373046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21110, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.29970612525939944, + "epoch": 6.144893802734943, + "flow_matching_loss": 0.010470710648223757, + "grad_norm": 0.06469423323869705, + "kept_token_count": 573.06015625, + "learning_rate": 6.631764879285531e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9937279045581817, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062721104361116884, + "mgd_loss_weight": 0.29615987539291383, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.1731414794922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.64381408691406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21120, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.29979959726333616, + "epoch": 6.147803316846087, + "flow_matching_loss": 0.010482519585639238, + "grad_norm": 0.08334849774837494, + "kept_token_count": 572.98984375, + "learning_rate": 6.618047488586614e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9941109955310822, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005889010103419423, + "mgd_loss_weight": 0.2958265244960785, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.0256805419922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.36952819824216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21130, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.300199830532074, + "epoch": 6.1507128309572305, + "flow_matching_loss": 0.01210370846092701, + "grad_norm": 0.06603080779314041, + "kept_token_count": 572.609375, + "learning_rate": 6.604340282750371e-06, + "loss": 0.014, + "mgd_cos_sim": 0.9935901939868927, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006409792415797711, + "mgd_loss_weight": 0.2954931855201721, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.6707305908203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6305389404297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21140, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.3006801098585129, + "epoch": 6.153622345068373, + "flow_matching_loss": 0.010670073563233018, + "grad_norm": 0.0554928295314312, + "kept_token_count": 572.2046875, + "learning_rate": 6.590643278432342e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9942012190818786, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005798775004222989, + "mgd_loss_weight": 0.295159837603569, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.57203674316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.66103515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21150, + "valid_token_count": 818.23125 + }, + { + "actual_token_mask_ratio": 0.2988685369491577, + "epoch": 6.156531859179517, + "flow_matching_loss": 0.011049536149948835, + "grad_norm": 0.07435163110494614, + "kept_token_count": 573.6828125, + "learning_rate": 6.576956492275682e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9941137194633484, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005886284075677395, + "mgd_loss_weight": 0.2948264926671982, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.505322265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0471923828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21160, + "valid_token_count": 818.22578125 + }, + { + "actual_token_mask_ratio": 0.3000166594982147, + "epoch": 6.159441373290661, + "flow_matching_loss": 0.011528433207422495, + "grad_norm": 0.07806994020938873, + "kept_token_count": 572.89296875, + "learning_rate": 6.563279940911124e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9939041674137116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006095827696844936, + "mgd_loss_weight": 0.2944931536912918, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.6513366699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.18817443847655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21170, + "valid_token_count": 818.43359375 + }, + { + "actual_token_mask_ratio": 0.3001873046159744, + "epoch": 6.162350887401804, + "flow_matching_loss": 0.009686227794736624, + "grad_norm": 0.08266045153141022, + "kept_token_count": 572.5984375, + "learning_rate": 6.549613640956963e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9937147498130798, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062852462753653525, + "mgd_loss_weight": 0.29415980279445647, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.6679290771484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.35943908691405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21180, + "valid_token_count": 818.21640625 + }, + { + "actual_token_mask_ratio": 0.2998143434524536, + "epoch": 6.165260401512947, + "flow_matching_loss": 0.011862288322299719, + "grad_norm": 0.0749034658074379, + "kept_token_count": 573.01171875, + "learning_rate": 6.535957609019053e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9937941193580627, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006205881899222732, + "mgd_loss_weight": 0.29382646083831787, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.3084289550781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7399536132813, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21190, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.29905094504356383, + "epoch": 6.168169915624091, + "flow_matching_loss": 0.011155919497832656, + "grad_norm": 0.0778762549161911, + "kept_token_count": 573.43828125, + "learning_rate": 6.5223118616907545e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9938633620738984, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006136644957587123, + "mgd_loss_weight": 0.29349311888217927, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.3165252685547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4770446777344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21200, + "valid_token_count": 818.0875 + }, + { + "actual_token_mask_ratio": 0.299359130859375, + "epoch": 6.171079429735234, + "flow_matching_loss": 0.011409405246376991, + "grad_norm": 0.09116967022418976, + "kept_token_count": 573.26015625, + "learning_rate": 6.508676415552939e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9938966512680054, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006103352550417185, + "mgd_loss_weight": 0.29315976798534393, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.50371704101565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4033660888672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21210, + "valid_token_count": 818.19609375 + }, + { + "actual_token_mask_ratio": 0.30053446590900423, + "epoch": 6.1739889438463775, + "flow_matching_loss": 0.012781368661671877, + "grad_norm": 0.10619403421878815, + "kept_token_count": 572.1984375, + "learning_rate": 6.495051287173952e-06, + "loss": 0.0145, + "mgd_cos_sim": 0.9941771328449249, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005822868924587965, + "mgd_loss_weight": 0.29282642900943756, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.5657501220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.48851013183594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21220, + "valid_token_count": 818.0484375 + }, + { + "actual_token_mask_ratio": 0.2996545761823654, + "epoch": 6.176898457957521, + "flow_matching_loss": 0.010901580890640616, + "grad_norm": 0.061294302344322205, + "kept_token_count": 573.0953125, + "learning_rate": 6.481436493109621e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9938439786434173, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006156022939831018, + "mgd_loss_weight": 0.29249308407306673, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.326318359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.35955810546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21230, + "valid_token_count": 818.29921875 + }, + { + "actual_token_mask_ratio": 0.29946031272411344, + "epoch": 6.179807972068665, + "flow_matching_loss": 0.011060041934251785, + "grad_norm": 0.06794647127389908, + "kept_token_count": 573.2796875, + "learning_rate": 6.467832049903194e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9938779354095459, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006122062122449279, + "mgd_loss_weight": 0.2921597361564636, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.10430297851565, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3660095214844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21240, + "valid_token_count": 818.34296875 + }, + { + "actual_token_mask_ratio": 0.30032635033130645, + "epoch": 6.182717486179808, + "flow_matching_loss": 0.012595629971474408, + "grad_norm": 0.0873430073261261, + "kept_token_count": 572.553125, + "learning_rate": 6.454237974085361e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9937069237232208, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006293071946129203, + "mgd_loss_weight": 0.29182639718055725, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.5566009521484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3682006835937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21250, + "valid_token_count": 818.3140625 + }, + { + "actual_token_mask_ratio": 0.3004685938358307, + "epoch": 6.1856270002909515, + "flow_matching_loss": 0.009681224077939986, + "grad_norm": 0.06726337224245071, + "kept_token_count": 572.384375, + "learning_rate": 6.440654282174203e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9936766743659973, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006323323585093021, + "mgd_loss_weight": 0.2914930462837219, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.2593994140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7016540527344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21260, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.29969091415405275, + "epoch": 6.188536514402095, + "flow_matching_loss": 0.011751217953860759, + "grad_norm": 0.11929036676883698, + "kept_token_count": 573.2328125, + "learning_rate": 6.427080990675187e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9938595831394196, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006140423892065883, + "mgd_loss_weight": 0.2911597043275833, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.8793701171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.72218627929686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21270, + "valid_token_count": 818.54609375 + }, + { + "actual_token_mask_ratio": 0.2995895564556122, + "epoch": 6.191446028513238, + "flow_matching_loss": 0.011924591893330217, + "grad_norm": 0.07573722302913666, + "kept_token_count": 572.88125, + "learning_rate": 6.41351811608114e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9939976334571838, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006002366077154875, + "mgd_loss_weight": 0.29082636535167694, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.0403991699219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.65076599121096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21280, + "valid_token_count": 817.92421875 + }, + { + "actual_token_mask_ratio": 0.3001497477293015, + "epoch": 6.194355542624382, + "flow_matching_loss": 0.010383921349421144, + "grad_norm": 0.07403408735990524, + "kept_token_count": 572.74921875, + "learning_rate": 6.399965674872228e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9938151240348816, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006184875825420022, + "mgd_loss_weight": 0.2904930114746094, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.4205841064453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.61155700683594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21290, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.2998443186283112, + "epoch": 6.197265056735525, + "flow_matching_loss": 0.011777797061949969, + "grad_norm": 0.06260353326797485, + "kept_token_count": 572.78984375, + "learning_rate": 6.386423683515957e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9937017500400543, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006298253545537591, + "mgd_loss_weight": 0.290159672498703, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.945751953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8042358398437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21300, + "valid_token_count": 818.09296875 + }, + { + "actual_token_mask_ratio": 0.30011254251003266, + "epoch": 6.200174570846668, + "flow_matching_loss": 0.011040781904011965, + "grad_norm": 0.08811786025762558, + "kept_token_count": 572.78515625, + "learning_rate": 6.372892158467119e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9936702072620391, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006329785287380219, + "mgd_loss_weight": 0.2898263305425644, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.71337890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5182250976562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21310, + "valid_token_count": 818.39296875 + }, + { + "actual_token_mask_ratio": 0.3001540511846542, + "epoch": 6.203084084957812, + "flow_matching_loss": 0.011442004423588514, + "grad_norm": 0.08538016676902771, + "kept_token_count": 572.65859375, + "learning_rate": 6.359371116167791e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9937944710254669, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00620553195476532, + "mgd_loss_weight": 0.28949297964572906, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.3126525878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.49171142578126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21320, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.3006461501121521, + "epoch": 6.205993599068956, + "flow_matching_loss": 0.009886470064520836, + "grad_norm": 0.07352609932422638, + "kept_token_count": 572.325, + "learning_rate": 6.345860573047314e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9938950777053833, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006104934541508555, + "mgd_loss_weight": 0.2891596406698227, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.8263824462891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.7850738525391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21330, + "valid_token_count": 818.36484375 + }, + { + "actual_token_mask_ratio": 0.299662584066391, + "epoch": 6.208903113180099, + "flow_matching_loss": 0.011605187505483627, + "grad_norm": 0.06577607989311218, + "kept_token_count": 572.9515625, + "learning_rate": 6.332360545522268e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9935925304889679, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006407470721751451, + "mgd_loss_weight": 0.28882629573345187, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.2554504394531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1944152832031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21340, + "valid_token_count": 818.11015625 + }, + { + "actual_token_mask_ratio": 0.2997654914855957, + "epoch": 6.211812627291242, + "flow_matching_loss": 0.011114971153438092, + "grad_norm": 0.12884968519210815, + "kept_token_count": 573.12109375, + "learning_rate": 6.31887104999647e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.993644779920578, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006355214258655905, + "mgd_loss_weight": 0.28849294781684875, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.5531341552734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4551055908203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21350, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.2992656111717224, + "epoch": 6.214722141402386, + "flow_matching_loss": 0.010534413903951646, + "grad_norm": 0.07102102786302567, + "kept_token_count": 573.36484375, + "learning_rate": 6.305392102860924e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9939708113670349, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006029194593429566, + "mgd_loss_weight": 0.2881596088409424, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.5082183837891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7559020996094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21360, + "valid_token_count": 818.23828125 + }, + { + "actual_token_mask_ratio": 0.29951347410678864, + "epoch": 6.21763165551353, + "flow_matching_loss": 0.010286729643121361, + "grad_norm": 0.07195078581571579, + "kept_token_count": 573.22890625, + "learning_rate": 6.291923720493814e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9937431991100312, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006256796652451158, + "mgd_loss_weight": 0.28782625794410704, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.66888122558595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.56800842285156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21370, + "valid_token_count": 818.32890625 + }, + { + "actual_token_mask_ratio": 0.30107995867729187, + "epoch": 6.220541169624672, + "flow_matching_loss": 0.011091077886521816, + "grad_norm": 0.07337956875562668, + "kept_token_count": 571.82421875, + "learning_rate": 6.278465919260507e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.993550443649292, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006449572136625647, + "mgd_loss_weight": 0.28749291598796844, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.4278503417969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2403625488281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21380, + "valid_token_count": 818.15390625 + }, + { + "actual_token_mask_ratio": 0.30011468529701235, + "epoch": 6.223450683735816, + "flow_matching_loss": 0.009887955291196704, + "grad_norm": 0.08410823345184326, + "kept_token_count": 572.66796875, + "learning_rate": 6.26501871551349e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9937101244926453, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006289866147562861, + "mgd_loss_weight": 0.2871595770120621, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.6734130859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0616882324219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21390, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.29979705810546875, + "epoch": 6.22636019784696, + "flow_matching_loss": 0.010275981295853854, + "grad_norm": 0.09878027439117432, + "kept_token_count": 572.95234375, + "learning_rate": 6.251582125592394e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9938700556755066, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006129946559667587, + "mgd_loss_weight": 0.2868262231349945, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.83506164550784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4478240966797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21400, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.3003750205039978, + "epoch": 6.229269711958103, + "flow_matching_loss": 0.010467879148200155, + "grad_norm": 0.0878947526216507, + "kept_token_count": 572.48671875, + "learning_rate": 6.2381561658239386e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9936047613620758, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006395221129059792, + "mgd_loss_weight": 0.28649288415908813, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.4619659423828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5805603027344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21410, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.29996397793293, + "epoch": 6.232179226069246, + "flow_matching_loss": 0.009827966662123799, + "grad_norm": 0.18763945996761322, + "kept_token_count": 572.78203125, + "learning_rate": 6.2247408525219276e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9939707219600677, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006029272451996803, + "mgd_loss_weight": 0.2861595392227173, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.6899932861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.39524230957034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21420, + "valid_token_count": 818.22109375 + }, + { + "actual_token_mask_ratio": 0.30105802714824675, + "epoch": 6.23508874018039, + "flow_matching_loss": 0.010571238026022911, + "grad_norm": 0.05848930403590202, + "kept_token_count": 571.84296875, + "learning_rate": 6.2113362019872335e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9933368265628815, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006663172878324986, + "mgd_loss_weight": 0.2858261913061142, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.5095184326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.44318237304685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21430, + "valid_token_count": 818.1546875 + }, + { + "actual_token_mask_ratio": 0.3010024011135101, + "epoch": 6.237998254291533, + "flow_matching_loss": 0.010008413810282946, + "grad_norm": 0.07311441004276276, + "kept_token_count": 572.1515625, + "learning_rate": 6.197942230507772e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9940064966678619, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005993507103994489, + "mgd_loss_weight": 0.2854928523302078, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.89313354492185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3491638183594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21440, + "valid_token_count": 818.5296875 + }, + { + "actual_token_mask_ratio": 0.30064116716384887, + "epoch": 6.240907768402677, + "flow_matching_loss": 0.011131230276077986, + "grad_norm": 0.08625360578298569, + "kept_token_count": 572.33046875, + "learning_rate": 6.184558954358475e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9934110820293427, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006588916759938002, + "mgd_loss_weight": 0.2851595044136047, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4076721191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.21973876953126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21450, + "valid_token_count": 818.36328125 + }, + { + "actual_token_mask_ratio": 0.2997565746307373, + "epoch": 6.24381728251382, + "flow_matching_loss": 0.01179728708229959, + "grad_norm": 0.06862115114927292, + "kept_token_count": 572.971875, + "learning_rate": 6.1711863898012965e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9938023269176484, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006197673128917813, + "mgd_loss_weight": 0.2848261594772339, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.6199951171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.05066833496096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21460, + "valid_token_count": 818.24609375 + }, + { + "actual_token_mask_ratio": 0.2995845228433609, + "epoch": 6.246726796624964, + "flow_matching_loss": 0.010464677959680558, + "grad_norm": 0.07860663533210754, + "kept_token_count": 573.06171875, + "learning_rate": 6.157824553085155e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9939709484577179, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006029062205925584, + "mgd_loss_weight": 0.2844928205013275, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.008544921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.89563903808596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21470, + "valid_token_count": 818.17890625 + }, + { + "actual_token_mask_ratio": 0.29963074922561644, + "epoch": 6.249636310736107, + "flow_matching_loss": 0.010802190564572812, + "grad_norm": 0.06977556645870209, + "kept_token_count": 573.12578125, + "learning_rate": 6.144473460445945e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9935539662837982, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00644603967666626, + "mgd_loss_weight": 0.28415946662425995, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.30480041503904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.72247924804685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21480, + "valid_token_count": 818.3203125 + }, + { + "actual_token_mask_ratio": 0.3000739961862564, + "epoch": 6.2525458248472505, + "flow_matching_loss": 0.010810091579332947, + "grad_norm": 0.1410294473171234, + "kept_token_count": 572.484375, + "learning_rate": 6.131133128106503e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9937354564666748, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00626455107703805, + "mgd_loss_weight": 0.2838261276483536, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.09478454589845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6967010498047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21490, + "valid_token_count": 817.92265625 + }, + { + "actual_token_mask_ratio": 0.2999163955450058, + "epoch": 6.255455338958394, + "flow_matching_loss": 0.01062921117991209, + "grad_norm": 0.08289118856191635, + "kept_token_count": 572.75, + "learning_rate": 6.117803572276584e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9937199473381042, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00628004614263773, + "mgd_loss_weight": 0.283492785692215, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.64019775390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1296752929687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21500, + "valid_token_count": 818.12265625 + }, + { + "actual_token_mask_ratio": 0.299845415353775, + "epoch": 6.258364853069537, + "flow_matching_loss": 0.011036740569397806, + "grad_norm": 0.09072312712669373, + "kept_token_count": 572.834375, + "learning_rate": 6.104484809152866e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9936690628528595, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006330938870087266, + "mgd_loss_weight": 0.28315943479537964, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.4162567138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.10166320800784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21510, + "valid_token_count": 818.15703125 + }, + { + "actual_token_mask_ratio": 0.299861541390419, + "epoch": 6.261274367180681, + "flow_matching_loss": 0.012163979653269052, + "grad_norm": 0.09321042895317078, + "kept_token_count": 572.94296875, + "learning_rate": 6.091176854918892e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9937896966934204, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006210297439247369, + "mgd_loss_weight": 0.28282609581947327, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.87864685058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4041290283203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21520, + "valid_token_count": 818.32734375 + }, + { + "actual_token_mask_ratio": 0.2999750018119812, + "epoch": 6.2641838812918245, + "flow_matching_loss": 0.011322080809623002, + "grad_norm": 0.05747690051794052, + "kept_token_count": 572.6796875, + "learning_rate": 6.077879725745092e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9937638521194458, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006236159056425094, + "mgd_loss_weight": 0.28249275088310244, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.1296051025391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7344604492188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21530, + "valid_token_count": 818.08671875 + }, + { + "actual_token_mask_ratio": 0.29999410808086396, + "epoch": 6.267093395402968, + "flow_matching_loss": 0.011317645478993655, + "grad_norm": 0.05380995571613312, + "kept_token_count": 572.7671875, + "learning_rate": 6.064593437788727e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9934987723827362, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00650122850202024, + "mgd_loss_weight": 0.28215940296649933, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.049169921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.07045288085936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21540, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.2998552471399307, + "epoch": 6.270002909514111, + "flow_matching_loss": 0.010499664675444364, + "grad_norm": 0.0779557153582573, + "kept_token_count": 572.89375, + "learning_rate": 6.051318007193881e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9938642680644989, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006135736824944615, + "mgd_loss_weight": 0.28182606399059296, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 442.8304412841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3940032958984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21550, + "valid_token_count": 818.24609375 + }, + { + "actual_token_mask_ratio": 0.3009229481220245, + "epoch": 6.272912423625255, + "flow_matching_loss": 0.010219722334295512, + "grad_norm": 0.06706223636865616, + "kept_token_count": 572.05, + "learning_rate": 6.038053450091469e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9937829434871673, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006217058096081018, + "mgd_loss_weight": 0.28149271607398985, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.3641876220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.99876098632814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21560, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.30012844502925873, + "epoch": 6.275821937736398, + "flow_matching_loss": 0.010888977581635118, + "grad_norm": 0.0712142363190651, + "kept_token_count": 572.6140625, + "learning_rate": 6.0247997825991705e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9938890933990479, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006110914470627904, + "mgd_loss_weight": 0.281159371137619, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.6035614013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8858367919922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21570, + "valid_token_count": 818.17265625 + }, + { + "actual_token_mask_ratio": 0.3004466027021408, + "epoch": 6.278731451847541, + "flow_matching_loss": 0.011194049846380949, + "grad_norm": 0.0834413394331932, + "kept_token_count": 572.44296875, + "learning_rate": 6.011557020821444e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9937492847442627, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006250716373324394, + "mgd_loss_weight": 0.28082603216171265, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.2721008300781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.83606872558596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21580, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.29947261810302733, + "epoch": 6.281640965958685, + "flow_matching_loss": 0.011144689749926329, + "grad_norm": 0.05849365517497063, + "kept_token_count": 573.14375, + "learning_rate": 5.99832518084949e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9938923954963684, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006107613630592823, + "mgd_loss_weight": 0.2804926782846451, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.9226348876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6700103759766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21590, + "valid_token_count": 818.15859375 + }, + { + "actual_token_mask_ratio": 0.29962668716907503, + "epoch": 6.284550480069829, + "flow_matching_loss": 0.01133418846875429, + "grad_norm": 0.0782986730337143, + "kept_token_count": 573.1859375, + "learning_rate": 5.98510427876124e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.993863183259964, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006136818695813418, + "mgd_loss_weight": 0.2801593393087387, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.5200714111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.43011169433595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21600, + "valid_token_count": 818.4015625 + }, + { + "actual_token_mask_ratio": 0.3004894733428955, + "epoch": 6.2874599941809715, + "flow_matching_loss": 0.010905604716390372, + "grad_norm": 0.0948733314871788, + "kept_token_count": 572.39140625, + "learning_rate": 5.971894330621344e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9938511729240418, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006148824468255043, + "mgd_loss_weight": 0.2798259973526001, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.00160827636716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5234619140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21610, + "valid_token_count": 818.27421875 + }, + { + "actual_token_mask_ratio": 0.3001824676990509, + "epoch": 6.290369508292115, + "flow_matching_loss": 0.01169779421761632, + "grad_norm": 0.06570582091808319, + "kept_token_count": 572.3890625, + "learning_rate": 5.958695352481134e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9939822018146515, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006017811829224229, + "mgd_loss_weight": 0.27949264645576477, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.50812377929685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1751983642578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21620, + "valid_token_count": 817.909375 + }, + { + "actual_token_mask_ratio": 0.3003107726573944, + "epoch": 6.293279022403259, + "flow_matching_loss": 0.009876599814742804, + "grad_norm": 0.07174824923276901, + "kept_token_count": 572.546875, + "learning_rate": 5.945507360378608e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9938941597938538, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061058422550559046, + "mgd_loss_weight": 0.2791593074798584, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.3883972167969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.67279052734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21630, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.2992627680301666, + "epoch": 6.296188536514402, + "flow_matching_loss": 0.012536665890365839, + "grad_norm": 0.07206975668668747, + "kept_token_count": 573.54140625, + "learning_rate": 5.9323303703384225e-06, + "loss": 0.0143, + "mgd_cos_sim": 0.9935401737689972, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006459822552278638, + "mgd_loss_weight": 0.2788259595632553, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.4355773925781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3731414794922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21640, + "valid_token_count": 818.48046875 + }, + { + "actual_token_mask_ratio": 0.2996323674917221, + "epoch": 6.299098050625545, + "flow_matching_loss": 0.011082890536636114, + "grad_norm": 0.07617342472076416, + "kept_token_count": 573.09453125, + "learning_rate": 5.919164398371857e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.993611091375351, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006388908065855503, + "mgd_loss_weight": 0.27849261462688446, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.99851989746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.18271484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21650, + "valid_token_count": 818.27734375 + }, + { + "actual_token_mask_ratio": 0.3006873488426208, + "epoch": 6.302007564736689, + "flow_matching_loss": 0.010489609697833658, + "grad_norm": 0.08395874500274658, + "kept_token_count": 572.3796875, + "learning_rate": 5.906009460476818e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9938097178936005, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00619028378278017, + "mgd_loss_weight": 0.2781592756509781, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.6114868164062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5547729492188, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21660, + "valid_token_count": 818.48984375 + }, + { + "actual_token_mask_ratio": 0.29983791410923005, + "epoch": 6.304917078847833, + "flow_matching_loss": 0.011699206568300724, + "grad_norm": 0.10126480460166931, + "kept_token_count": 572.86796875, + "learning_rate": 5.892865572637789e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9937382102012634, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006261791381984949, + "mgd_loss_weight": 0.27782592475414275, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.9394012451172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0611267089844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21670, + "valid_token_count": 818.19140625 + }, + { + "actual_token_mask_ratio": 0.3003310114145279, + "epoch": 6.307826592958976, + "flow_matching_loss": 0.01067003714852035, + "grad_norm": 0.06486072391271591, + "kept_token_count": 572.325, + "learning_rate": 5.8797327508258396e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9937558829784393, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006244120048359036, + "mgd_loss_weight": 0.27749258279800415, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.0211883544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0330749511719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21680, + "valid_token_count": 817.99375 + }, + { + "actual_token_mask_ratio": 0.30043193995952605, + "epoch": 6.310736107070119, + "flow_matching_loss": 0.011156841553747654, + "grad_norm": 0.06985784322023392, + "kept_token_count": 572.4046875, + "learning_rate": 5.866611010998584e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9934962272644043, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00650378386490047, + "mgd_loss_weight": 0.27715924084186555, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.2651672363281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.9221252441406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21690, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.30038611590862274, + "epoch": 6.313645621181263, + "flow_matching_loss": 0.011404884047806263, + "grad_norm": 0.06215968355536461, + "kept_token_count": 572.490625, + "learning_rate": 5.853500369100172e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9939958989620209, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060040952637791635, + "mgd_loss_weight": 0.2768258899450302, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.5096038818359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.58465881347655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21700, + "valid_token_count": 818.30078125 + }, + { + "actual_token_mask_ratio": 0.30026847422122954, + "epoch": 6.316555135292406, + "flow_matching_loss": 0.00975304408930242, + "grad_norm": 0.07060591131448746, + "kept_token_count": 572.534375, + "learning_rate": 5.840400841061265e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9938382208347321, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006161783030256629, + "mgd_loss_weight": 0.27649255096912384, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.8418212890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3550750732422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21710, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.30023249983787537, + "epoch": 6.31946464940355, + "flow_matching_loss": 0.010148484027013183, + "grad_norm": 0.09931361675262451, + "kept_token_count": 572.56953125, + "learning_rate": 5.827312442799034e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.993991219997406, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006008786102756858, + "mgd_loss_weight": 0.276159206032753, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.1957763671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.96986694335936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21720, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.29987071454524994, + "epoch": 6.322374163514693, + "flow_matching_loss": 0.009600699925795198, + "grad_norm": 0.07096727192401886, + "kept_token_count": 572.928125, + "learning_rate": 5.814235190217114e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9938124597072602, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006187541456893086, + "mgd_loss_weight": 0.2758258581161499, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.7127288818359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.86907653808595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21730, + "valid_token_count": 818.315625 + }, + { + "actual_token_mask_ratio": 0.30007216334342957, + "epoch": 6.325283677625836, + "flow_matching_loss": 0.010252416413277388, + "grad_norm": 0.0793129950761795, + "kept_token_count": 572.64296875, + "learning_rate": 5.801169099205599e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.993901151418686, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006098862178623677, + "mgd_loss_weight": 0.27549251914024353, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.5704376220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6710601806641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21740, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.30022028684616087, + "epoch": 6.32819319173698, + "flow_matching_loss": 0.010237955302000046, + "grad_norm": 0.07298800349235535, + "kept_token_count": 572.409375, + "learning_rate": 5.788114185641022e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9938288927078247, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006171095045283436, + "mgd_loss_weight": 0.2751591712236404, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.6399719238281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.58392028808595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21750, + "valid_token_count": 817.9828125 + }, + { + "actual_token_mask_ratio": 0.2999568611383438, + "epoch": 6.3311027058481235, + "flow_matching_loss": 0.011162499710917474, + "grad_norm": 0.07726593315601349, + "kept_token_count": 572.915625, + "learning_rate": 5.775070465386328e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9937246561050415, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006275343336164951, + "mgd_loss_weight": 0.2748258262872696, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.33095703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.86928405761716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21760, + "valid_token_count": 818.39765625 + }, + { + "actual_token_mask_ratio": 0.30019897818565366, + "epoch": 6.334012219959266, + "flow_matching_loss": 0.011797798145562411, + "grad_norm": 0.104599229991436, + "kept_token_count": 572.628125, + "learning_rate": 5.7620379542908765e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9937271118164063, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062728967051953076, + "mgd_loss_weight": 0.2744924873113632, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.9571258544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.79493713378906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21770, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.30022969245910647, + "epoch": 6.33692173407041, + "flow_matching_loss": 0.011095707025378942, + "grad_norm": 0.09520632028579712, + "kept_token_count": 572.6125, + "learning_rate": 5.749016668190395e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9939443051815033, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006055695423856378, + "mgd_loss_weight": 0.2741591364145279, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.7255157470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0852355957031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21780, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.29987485706806183, + "epoch": 6.339831248181554, + "flow_matching_loss": 0.011666198400780559, + "grad_norm": 0.06559114158153534, + "kept_token_count": 572.85078125, + "learning_rate": 5.736006622906973e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9936678826808929, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006332116527482867, + "mgd_loss_weight": 0.2738257944583893, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.0268829345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6881866455078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21790, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.30043556690216067, + "epoch": 6.3427407622926975, + "flow_matching_loss": 0.01182982288300991, + "grad_norm": 0.0745345875620842, + "kept_token_count": 572.34453125, + "learning_rate": 5.723007834249034e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.9938400387763977, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006159960245713592, + "mgd_loss_weight": 0.2734924525022507, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.3114013671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5505798339844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21800, + "valid_token_count": 818.14453125 + }, + { + "actual_token_mask_ratio": 0.30010595321655276, + "epoch": 6.34565027640384, + "flow_matching_loss": 0.01168473195284605, + "grad_norm": 0.06232859194278717, + "kept_token_count": 572.725, + "learning_rate": 5.710020318011346e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9940243482589721, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005975661426782608, + "mgd_loss_weight": 0.27315910160541534, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.32545471191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.07765502929686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21810, + "valid_token_count": 818.30390625 + }, + { + "actual_token_mask_ratio": 0.3005859375, + "epoch": 6.348559790514984, + "flow_matching_loss": 0.011624120594933628, + "grad_norm": 0.06096884235739708, + "kept_token_count": 572.365625, + "learning_rate": 5.697044089974955e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9940184593200684, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005981532530859113, + "mgd_loss_weight": 0.27282576262950897, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.40162353515626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.84169311523436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21820, + "valid_token_count": 818.34609375 + }, + { + "actual_token_mask_ratio": 0.29997922778129577, + "epoch": 6.351469304626128, + "flow_matching_loss": 0.011070086061954499, + "grad_norm": 0.06508708745241165, + "kept_token_count": 572.803125, + "learning_rate": 5.68407916590721e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9936684668064117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006331539060920477, + "mgd_loss_weight": 0.27249241769313814, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.46868591308595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.256298828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21830, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.29942276775836946, + "epoch": 6.3543788187372705, + "flow_matching_loss": 0.00975158833898604, + "grad_norm": 0.06505842506885529, + "kept_token_count": 573.26484375, + "learning_rate": 5.671125561561713e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9936522781848908, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006347713666036725, + "mgd_loss_weight": 0.27215906977653503, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.9518585205078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5550537109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21840, + "valid_token_count": 818.27734375 + }, + { + "actual_token_mask_ratio": 0.29997136890888215, + "epoch": 6.357288332848414, + "flow_matching_loss": 0.0115998444147408, + "grad_norm": 0.058985356241464615, + "kept_token_count": 573.02578125, + "learning_rate": 5.658183292678315e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9936876595020294, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063123426400125025, + "mgd_loss_weight": 0.27182573080062866, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.2031677246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.75281677246096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21850, + "valid_token_count": 818.5734375 + }, + { + "actual_token_mask_ratio": 0.3003381937742233, + "epoch": 6.360197846959558, + "flow_matching_loss": 0.010037966817617417, + "grad_norm": 0.0563875213265419, + "kept_token_count": 572.5140625, + "learning_rate": 5.645252374983091e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9938192307949066, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006180765712633729, + "mgd_loss_weight": 0.2714923799037933, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.1138214111328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2746643066406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21860, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.29961508214473725, + "epoch": 6.363107361070702, + "flow_matching_loss": 0.011164325522258878, + "grad_norm": 0.060363270342350006, + "kept_token_count": 572.94140625, + "learning_rate": 5.6323328241883224e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9938668966293335, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006133108679205179, + "mgd_loss_weight": 0.2711590379476547, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.8487518310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8236846923828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21870, + "valid_token_count": 818.034375 + }, + { + "actual_token_mask_ratio": 0.29925312995910647, + "epoch": 6.3660168751818444, + "flow_matching_loss": 0.010747120482847095, + "grad_norm": 0.09647394716739655, + "kept_token_count": 573.37890625, + "learning_rate": 5.61942465599249e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9936221301555633, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006377868354320526, + "mgd_loss_weight": 0.27082569897174835, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.15171813964844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4906433105469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21880, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.29978167116641996, + "epoch": 6.368926389292988, + "flow_matching_loss": 0.011477652005851269, + "grad_norm": 0.08261404186487198, + "kept_token_count": 573.0265625, + "learning_rate": 5.60652788608023e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9939467906951904, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006053206324577331, + "mgd_loss_weight": 0.2704923450946808, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.48143310546874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.89317321777344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21890, + "valid_token_count": 818.35390625 + }, + { + "actual_token_mask_ratio": 0.2993278056383133, + "epoch": 6.371835903404132, + "flow_matching_loss": 0.010046089859679341, + "grad_norm": 0.06140768155455589, + "kept_token_count": 573.29140625, + "learning_rate": 5.593642530122336e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9938883066177369, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006111706420779228, + "mgd_loss_weight": 0.2701590061187744, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.1259521484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6173095703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21900, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.3001383453607559, + "epoch": 6.374745417515275, + "flow_matching_loss": 0.013196800323203205, + "grad_norm": 0.0859949067234993, + "kept_token_count": 572.478125, + "learning_rate": 5.58076860377573e-06, + "loss": 0.0149, + "mgd_cos_sim": 0.9937701880931854, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006229810183867812, + "mgd_loss_weight": 0.2698256611824036, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.44627075195314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.868359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21910, + "valid_token_count": 817.990625 + }, + { + "actual_token_mask_ratio": 0.3006512522697449, + "epoch": 6.377654931626418, + "flow_matching_loss": 0.010332803335040807, + "grad_norm": 0.08469067513942719, + "kept_token_count": 572.31484375, + "learning_rate": 5.567906122683439e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9936830282211304, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006316965399309993, + "mgd_loss_weight": 0.2694923132658005, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.7972351074219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8921783447266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21920, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.29891214072704314, + "epoch": 6.380564445737562, + "flow_matching_loss": 0.012108866777271032, + "grad_norm": 0.07318778336048126, + "kept_token_count": 573.68671875, + "learning_rate": 5.555055102474604e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9935853183269501, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00641469513066113, + "mgd_loss_weight": 0.2691589742898941, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.67308349609374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1437072753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21930, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.30047337114810946, + "epoch": 6.383473959848705, + "flow_matching_loss": 0.010537008801475168, + "grad_norm": 0.06476966291666031, + "kept_token_count": 572.65390625, + "learning_rate": 5.542215558764421e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9935008883476257, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006499106390401721, + "mgd_loss_weight": 0.268825626373291, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.33336181640624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.82976989746095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21940, + "valid_token_count": 818.634375 + }, + { + "actual_token_mask_ratio": 0.29927447736263274, + "epoch": 6.386383473959849, + "flow_matching_loss": 0.010875984327867627, + "grad_norm": 0.07465936988592148, + "kept_token_count": 573.4828125, + "learning_rate": 5.529387507154138e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9941880166530609, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005811985582113266, + "mgd_loss_weight": 0.26849228143692017, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.54115905761716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7134948730469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21950, + "valid_token_count": 818.40859375 + }, + { + "actual_token_mask_ratio": 0.3005138039588928, + "epoch": 6.389292988070992, + "flow_matching_loss": 0.01222343733534217, + "grad_norm": 0.07139396667480469, + "kept_token_count": 572.34375, + "learning_rate": 5.516570963231058e-06, + "loss": 0.0139, + "mgd_cos_sim": 0.9937102615833282, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006289722537621856, + "mgd_loss_weight": 0.2681589424610138, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.70953063964845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.35390014648436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21960, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.30059294402599335, + "epoch": 6.392202502182135, + "flow_matching_loss": 0.010318622272461653, + "grad_norm": 0.0664498582482338, + "kept_token_count": 572.24609375, + "learning_rate": 5.503765942568484e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9936008155345917, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006399172777310014, + "mgd_loss_weight": 0.26782559156417846, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.30194702148435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0427490234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21970, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.30004303753376005, + "epoch": 6.395112016293279, + "flow_matching_loss": 0.01082070404663682, + "grad_norm": 0.09387531131505966, + "kept_token_count": 572.97734375, + "learning_rate": 5.49097246072573e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9935235857963562, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006476422771811485, + "mgd_loss_weight": 0.26749224960803986, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.7556640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6322998046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21980, + "valid_token_count": 818.59296875 + }, + { + "actual_token_mask_ratio": 0.3004842311143875, + "epoch": 6.398021530404423, + "flow_matching_loss": 0.010624524718150497, + "grad_norm": 0.08216728270053864, + "kept_token_count": 572.38125, + "learning_rate": 5.4781905332480805e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9940797090530396, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005920294858515263, + "mgd_loss_weight": 0.2671589106321335, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.32041320800784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.18983154296876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 21990, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.29975546300411227, + "epoch": 6.400931044515566, + "flow_matching_loss": 0.010992303024977446, + "grad_norm": 0.06863310933113098, + "kept_token_count": 572.93203125, + "learning_rate": 5.4654201756667795e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9936998903751373, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006300110602751374, + "mgd_loss_weight": 0.2668255567550659, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.59539794921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.07647705078125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22000, + "valid_token_count": 818.18984375 + }, + { + "actual_token_mask_ratio": 0.2993086397647858, + "epoch": 6.403840558626709, + "flow_matching_loss": 0.009259240375831724, + "grad_norm": 0.07667765766382217, + "kept_token_count": 573.22734375, + "learning_rate": 5.452661403499017e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9935820937156677, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006417908612638712, + "mgd_loss_weight": 0.26649221777915955, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.1852142333984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1214233398438, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22010, + "valid_token_count": 818.08203125 + }, + { + "actual_token_mask_ratio": 0.29989163875579833, + "epoch": 6.406750072737853, + "flow_matching_loss": 0.010123284021392464, + "grad_norm": 0.07461430132389069, + "kept_token_count": 572.92578125, + "learning_rate": 5.4399142322479005e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9939310610294342, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006068947352468967, + "mgd_loss_weight": 0.2661588728427887, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.2414794921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9864959716797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22020, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.30026805996894834, + "epoch": 6.4096595868489965, + "flow_matching_loss": 0.010958223789930343, + "grad_norm": 0.07617184519767761, + "kept_token_count": 572.38828125, + "learning_rate": 5.427178677402452e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9938718557357789, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006128129130229354, + "mgd_loss_weight": 0.2658255249261856, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.9378692626953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6267517089844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22030, + "valid_token_count": 818.015625 + }, + { + "actual_token_mask_ratio": 0.29940800070762635, + "epoch": 6.412569100960139, + "flow_matching_loss": 0.010738195478916168, + "grad_norm": 0.09024089574813843, + "kept_token_count": 573.203125, + "learning_rate": 5.41445475443757e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9939725458621979, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006027441332116723, + "mgd_loss_weight": 0.26549218595027924, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.71741638183596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7286102294922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22040, + "valid_token_count": 818.17109375 + }, + { + "actual_token_mask_ratio": 0.29970611333847047, + "epoch": 6.415478615071283, + "flow_matching_loss": 0.010236953059211374, + "grad_norm": 0.0757087990641594, + "kept_token_count": 573.22578125, + "learning_rate": 5.4017424788140194e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9939729154109955, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006027087336406112, + "mgd_loss_weight": 0.2651588380336761, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.86904296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6355041503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22050, + "valid_token_count": 818.55078125 + }, + { + "actual_token_mask_ratio": 0.3004215657711029, + "epoch": 6.418388129182427, + "flow_matching_loss": 0.01213566530495882, + "grad_norm": 0.08582047373056412, + "kept_token_count": 572.45546875, + "learning_rate": 5.389041865978414e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9936461031436921, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006353888846933842, + "mgd_loss_weight": 0.2648254930973053, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.5005279541016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3528564453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22060, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.2999108493328094, + "epoch": 6.4212976432935704, + "flow_matching_loss": 0.010807304549962283, + "grad_norm": 0.076493039727211, + "kept_token_count": 572.871875, + "learning_rate": 5.37635293136319e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9937689244747162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006231060344725847, + "mgd_loss_weight": 0.2644921541213989, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.8321472167969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7289672851563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22070, + "valid_token_count": 818.28359375 + }, + { + "actual_token_mask_ratio": 0.3008274585008621, + "epoch": 6.424207157404713, + "flow_matching_loss": 0.012229265738278627, + "grad_norm": 0.09475965797901154, + "kept_token_count": 572.06171875, + "learning_rate": 5.363675690386609e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.994020265340805, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00597972827963531, + "mgd_loss_weight": 0.26415880024433136, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.3126281738281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8566925048828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22080, + "valid_token_count": 818.19921875 + }, + { + "actual_token_mask_ratio": 0.30046104192733764, + "epoch": 6.427116671515857, + "flow_matching_loss": 0.01138119432143867, + "grad_norm": 0.07334782183170319, + "kept_token_count": 572.37734375, + "learning_rate": 5.351010158452708e-06, + "loss": 0.0131, + "mgd_cos_sim": 0.9936254620552063, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006374525791034103, + "mgd_loss_weight": 0.263825461268425, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.61016845703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.95836486816404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22090, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.2999522745609283, + "epoch": 6.430026185627001, + "flow_matching_loss": 0.011026301328092814, + "grad_norm": 0.09741006046533585, + "kept_token_count": 572.91171875, + "learning_rate": 5.338356350951308e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9933209955692291, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0066790015436708925, + "mgd_loss_weight": 0.2634921193122864, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.3477294921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.00726318359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22100, + "valid_token_count": 818.384375 + }, + { + "actual_token_mask_ratio": 0.29980392158031466, + "epoch": 6.4329356997381435, + "flow_matching_loss": 0.00981481079943478, + "grad_norm": 0.09790944308042526, + "kept_token_count": 572.734375, + "learning_rate": 5.325714283257974e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.993897670507431, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006102344766259193, + "mgd_loss_weight": 0.26315876841545105, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.3453338623047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.5463806152344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22110, + "valid_token_count": 817.96484375 + }, + { + "actual_token_mask_ratio": 0.2991762638092041, + "epoch": 6.435845213849287, + "flow_matching_loss": 0.01160767674446106, + "grad_norm": 0.07365420460700989, + "kept_token_count": 573.51484375, + "learning_rate": 5.313083970734013e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.9933686256408691, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00663136919029057, + "mgd_loss_weight": 0.2628254294395447, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.78224487304686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1058624267578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22120, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.3007792741060257, + "epoch": 6.438754727960431, + "flow_matching_loss": 0.013001600699499249, + "grad_norm": 0.08228171616792679, + "kept_token_count": 571.96171875, + "learning_rate": 5.300465428726439e-06, + "loss": 0.0146, + "mgd_cos_sim": 0.9937871038913727, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006212910544127226, + "mgd_loss_weight": 0.26249208152294157, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.3414367675781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.09568786621094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22130, + "valid_token_count": 817.99765625 + }, + { + "actual_token_mask_ratio": 0.3000252336263657, + "epoch": 6.441664242071574, + "flow_matching_loss": 0.011132622743025422, + "grad_norm": 0.07932594418525696, + "kept_token_count": 572.80703125, + "learning_rate": 5.287858672567979e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9940625011920929, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005937497271224857, + "mgd_loss_weight": 0.26215873658657074, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.90715942382815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6320373535156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22140, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.300032639503479, + "epoch": 6.444573756182717, + "flow_matching_loss": 0.011124295648187399, + "grad_norm": 0.07385016977787018, + "kept_token_count": 572.71328125, + "learning_rate": 5.275263717577028e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9937131226062774, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062868828419595955, + "mgd_loss_weight": 0.26182539761066437, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.70338745117186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.72022399902346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22150, + "valid_token_count": 818.20546875 + }, + { + "actual_token_mask_ratio": 0.2998420476913452, + "epoch": 6.447483270293861, + "flow_matching_loss": 0.010245522530749441, + "grad_norm": 0.07197077572345734, + "kept_token_count": 572.8984375, + "learning_rate": 5.2626805790576435e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9937210857868195, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062789182178676125, + "mgd_loss_weight": 0.26149204671382903, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.58501892089845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.26189575195315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22160, + "valid_token_count": 818.24140625 + }, + { + "actual_token_mask_ratio": 0.2994808167219162, + "epoch": 6.450392784405004, + "flow_matching_loss": 0.010876953508704901, + "grad_norm": 0.08644844591617584, + "kept_token_count": 573.1171875, + "learning_rate": 5.250109272299523e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9938242077827454, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00617579179815948, + "mgd_loss_weight": 0.26115870475769043, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.52520751953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4086944580078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22170, + "valid_token_count": 818.13046875 + }, + { + "actual_token_mask_ratio": 0.30054105520248414, + "epoch": 6.453302298516148, + "flow_matching_loss": 0.010318488255143166, + "grad_norm": 0.10078122466802597, + "kept_token_count": 572.296875, + "learning_rate": 5.237549812577987e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9940101683139801, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005989833502098918, + "mgd_loss_weight": 0.26082536578178406, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.6954711914062, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8742218017578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22180, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.2999164193868637, + "epoch": 6.456211812627291, + "flow_matching_loss": 0.011626521637663246, + "grad_norm": 0.07848189026117325, + "kept_token_count": 572.8625, + "learning_rate": 5.225002215153973e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9936134874820709, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006386508792638778, + "mgd_loss_weight": 0.2604920119047165, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.0735687255859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.58045349121096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22190, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.2999629616737366, + "epoch": 6.459121326738435, + "flow_matching_loss": 0.010495324200019241, + "grad_norm": 0.058927807956933975, + "kept_token_count": 572.60546875, + "learning_rate": 5.21246649527399e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9938912153244018, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006108791753649712, + "mgd_loss_weight": 0.2601586729288101, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.4409545898437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4373077392578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22200, + "valid_token_count": 817.9640625 + }, + { + "actual_token_mask_ratio": 0.29980733394622805, + "epoch": 6.462030840849578, + "flow_matching_loss": 0.009318865090608596, + "grad_norm": 0.05212217569351196, + "kept_token_count": 572.98984375, + "learning_rate": 5.199942668170118e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9938220262527466, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006177973747253418, + "mgd_loss_weight": 0.2598253309726715, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.3676818847656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.60284423828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22210, + "valid_token_count": 818.33203125 + }, + { + "actual_token_mask_ratio": 0.3003489553928375, + "epoch": 6.464940354960722, + "flow_matching_loss": 0.010612480249255896, + "grad_norm": 0.05886290967464447, + "kept_token_count": 572.47578125, + "learning_rate": 5.187430749059986e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9937416255474091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062583740800619125, + "mgd_loss_weight": 0.2594919800758362, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.167529296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.26280822753904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22220, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.3004219174385071, + "epoch": 6.467849869071865, + "flow_matching_loss": 0.011807601898908615, + "grad_norm": 0.08655233681201935, + "kept_token_count": 572.29609375, + "learning_rate": 5.174930753146763e-06, + "loss": 0.0135, + "mgd_cos_sim": 0.993648785352707, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063512143678963184, + "mgd_loss_weight": 0.2591586410999298, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.18214111328126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.79319458007814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22230, + "valid_token_count": 818.0640625 + }, + { + "actual_token_mask_ratio": 0.30032900273799895, + "epoch": 6.470759383183008, + "flow_matching_loss": 0.009438553638756276, + "grad_norm": 0.05894403159618378, + "kept_token_count": 572.59765625, + "learning_rate": 5.162442695619114e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9938282370567322, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006171772815287113, + "mgd_loss_weight": 0.2588252931833267, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.5193664550781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.88624572753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22240, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.3005424439907074, + "epoch": 6.473668897294152, + "flow_matching_loss": 0.010068163368850946, + "grad_norm": 0.08695744723081589, + "kept_token_count": 572.33359375, + "learning_rate": 5.149966591651215e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9937836587429046, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006216349871829152, + "mgd_loss_weight": 0.25849194824695587, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.13743286132814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7461883544922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22250, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.2996830463409424, + "epoch": 6.4765784114052956, + "flow_matching_loss": 0.01147326985374093, + "grad_norm": 0.07119696587324142, + "kept_token_count": 573.08828125, + "learning_rate": 5.137502456402705e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9941229164600373, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005877089314162731, + "mgd_loss_weight": 0.2581586092710495, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.79005432128906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6531066894531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22260, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.2995628625154495, + "epoch": 6.479487925516438, + "flow_matching_loss": 0.010970636503770947, + "grad_norm": 0.0666658878326416, + "kept_token_count": 573.01171875, + "learning_rate": 5.125050305018683e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9937305271625518, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006269470555707812, + "mgd_loss_weight": 0.25782525837421416, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.75237731933595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.04363403320315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22270, + "valid_token_count": 818.08125 + }, + { + "actual_token_mask_ratio": 0.29991884231567384, + "epoch": 6.482397439627582, + "flow_matching_loss": 0.010991853848099708, + "grad_norm": 0.062361493706703186, + "kept_token_count": 572.99609375, + "learning_rate": 5.112610152629687e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9938063383102417, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006193664390593767, + "mgd_loss_weight": 0.25749191641807556, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.1651184082031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.14708251953124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22280, + "valid_token_count": 818.46796875 + }, + { + "actual_token_mask_ratio": 0.29936357736587527, + "epoch": 6.485306953738726, + "flow_matching_loss": 0.012243654066696762, + "grad_norm": 0.05801454186439514, + "kept_token_count": 573.4328125, + "learning_rate": 5.10018201435167e-06, + "loss": 0.0138, + "mgd_cos_sim": 0.9938638269901275, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006136170960962773, + "mgd_loss_weight": 0.25715857446193696, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.54076232910154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.92926025390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22290, + "valid_token_count": 818.44609375 + }, + { + "actual_token_mask_ratio": 0.29985882341861725, + "epoch": 6.488216467849869, + "flow_matching_loss": 0.009576447075232863, + "grad_norm": 0.0654684528708458, + "kept_token_count": 572.8015625, + "learning_rate": 5.087765905286004e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9939469277858735, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006053055729717016, + "mgd_loss_weight": 0.2568252235651016, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.6942199707031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4712249755859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22300, + "valid_token_count": 818.121875 + }, + { + "actual_token_mask_ratio": 0.29928877055644987, + "epoch": 6.491125981961012, + "flow_matching_loss": 0.01002224455587566, + "grad_norm": 0.07184648513793945, + "kept_token_count": 573.47265625, + "learning_rate": 5.0753618405194245e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9937972009181977, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006202808022499085, + "mgd_loss_weight": 0.25649188458919525, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.9770812988281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5291687011719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22310, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.29964359998703005, + "epoch": 6.494035496072156, + "flow_matching_loss": 0.010308222332969307, + "grad_norm": 0.05985356494784355, + "kept_token_count": 573.0484375, + "learning_rate": 5.062969835124042e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9937098085880279, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006290185172110796, + "mgd_loss_weight": 0.2561585396528244, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 403.90601501464846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.37046508789064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22320, + "valid_token_count": 818.22890625 + }, + { + "actual_token_mask_ratio": 0.2999163120985031, + "epoch": 6.4969450101833, + "flow_matching_loss": 0.010285761207342148, + "grad_norm": 0.057989221066236496, + "kept_token_count": 572.95234375, + "learning_rate": 5.05058990415731e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9937682807445526, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006231720745563507, + "mgd_loss_weight": 0.2558251917362213, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.64536437988284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.59381713867185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22330, + "valid_token_count": 818.40390625 + }, + { + "actual_token_mask_ratio": 0.30151225328445436, + "epoch": 6.4998545242944425, + "flow_matching_loss": 0.010260187881067395, + "grad_norm": 0.08233875036239624, + "kept_token_count": 571.75625, + "learning_rate": 5.038222062662007e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9937566518783569, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006243356503546238, + "mgd_loss_weight": 0.25549185276031494, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.9618347167969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9238220214844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22340, + "valid_token_count": 818.5671875 + }, + { + "actual_token_mask_ratio": 0.29932028651237486, + "epoch": 6.502764038405586, + "flow_matching_loss": 0.01210155370645225, + "grad_norm": 0.0655151829123497, + "kept_token_count": 573.296875, + "learning_rate": 5.025866325666237e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.994023984670639, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005976025108247995, + "mgd_loss_weight": 0.2551585018634796, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.74022216796874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.00229797363284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22350, + "valid_token_count": 818.19609375 + }, + { + "actual_token_mask_ratio": 0.30037657618522645, + "epoch": 6.50567355251673, + "flow_matching_loss": 0.011648423131555318, + "grad_norm": 0.08290112018585205, + "kept_token_count": 572.584375, + "learning_rate": 5.013522708183383e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9938024580478668, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006197549169883132, + "mgd_loss_weight": 0.254825159907341, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.3544097900391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6763061523437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22360, + "valid_token_count": 818.4203125 + }, + { + "actual_token_mask_ratio": 0.29977365136146544, + "epoch": 6.508583066627873, + "flow_matching_loss": 0.01190927871502936, + "grad_norm": 0.0677419975399971, + "kept_token_count": 572.9515625, + "learning_rate": 5.001191225212096e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9935043036937714, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006495690904557705, + "mgd_loss_weight": 0.25449182093143463, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.87383728027345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.901611328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22370, + "valid_token_count": 818.23671875 + }, + { + "actual_token_mask_ratio": 0.2998641341924667, + "epoch": 6.5114925807390165, + "flow_matching_loss": 0.0101151829585433, + "grad_norm": 0.07651244103908539, + "kept_token_count": 572.9703125, + "learning_rate": 4.9888718917363e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9938672661781311, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006132729770615697, + "mgd_loss_weight": 0.25415846705436707, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.42991638183594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8944030761719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22380, + "valid_token_count": 818.36796875 + }, + { + "actual_token_mask_ratio": 0.30016528964042666, + "epoch": 6.51440209485016, + "flow_matching_loss": 0.010085269901901483, + "grad_norm": 0.0717889741063118, + "kept_token_count": 572.7171875, + "learning_rate": 4.9765647227251395e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9940615117549896, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005938486661761999, + "mgd_loss_weight": 0.2538251280784607, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.900048828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.66090698242186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22390, + "valid_token_count": 818.36015625 + }, + { + "actual_token_mask_ratio": 0.2998257577419281, + "epoch": 6.517311608961304, + "flow_matching_loss": 0.009411622816696764, + "grad_norm": 0.05589482560753822, + "kept_token_count": 572.9296875, + "learning_rate": 4.964269733132992e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9940149843692779, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00598501143977046, + "mgd_loss_weight": 0.2534917861223221, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.8706115722656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.02240600585935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22400, + "valid_token_count": 818.27109375 + }, + { + "actual_token_mask_ratio": 0.29982709884643555, + "epoch": 6.520221123072447, + "flow_matching_loss": 0.010952346492558718, + "grad_norm": 0.06811263412237167, + "kept_token_count": 573.07578125, + "learning_rate": 4.951986937899428e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9938105404376983, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006189460167661309, + "mgd_loss_weight": 0.25315843522548676, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.3518768310547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0530578613281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22410, + "valid_token_count": 818.48046875 + }, + { + "actual_token_mask_ratio": 0.3007503032684326, + "epoch": 6.52313063718359, + "flow_matching_loss": 0.010916099231690169, + "grad_norm": 0.09700514376163483, + "kept_token_count": 572.190625, + "learning_rate": 4.9397163519492e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9939346849918366, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006065297266468405, + "mgd_loss_weight": 0.2528250962495804, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.5598449707031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6008331298828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22420, + "valid_token_count": 818.29453125 + }, + { + "actual_token_mask_ratio": 0.2999907165765762, + "epoch": 6.526040151294734, + "flow_matching_loss": 0.010351746389642359, + "grad_norm": 0.0663171038031578, + "kept_token_count": 572.69140625, + "learning_rate": 4.9274579901922275e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9935966074466706, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006403400981798768, + "mgd_loss_weight": 0.25249175131320956, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.8500732421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.83094482421876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22430, + "valid_token_count": 818.12265625 + }, + { + "actual_token_mask_ratio": 0.3009296804666519, + "epoch": 6.528949665405877, + "flow_matching_loss": 0.00918576428666711, + "grad_norm": 0.07670973986387253, + "kept_token_count": 572.06640625, + "learning_rate": 4.9152118675235705e-06, + "loss": 0.0108, + "mgd_cos_sim": 0.9936063706874847, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063936143647879366, + "mgd_loss_weight": 0.25215840339660645, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.6305358886719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0127380371094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22440, + "valid_token_count": 818.32265625 + }, + { + "actual_token_mask_ratio": 0.3000826358795166, + "epoch": 6.531859179517021, + "flow_matching_loss": 0.010947762290015817, + "grad_norm": 0.07780329138040543, + "kept_token_count": 572.72890625, + "learning_rate": 4.902977998823432e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9940932035446167, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005906810564920306, + "mgd_loss_weight": 0.2518250644207001, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.0007598876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.16712646484376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22450, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.300109276175499, + "epoch": 6.534768693628164, + "flow_matching_loss": 0.01076087523251772, + "grad_norm": 0.061990514397621155, + "kept_token_count": 572.70234375, + "learning_rate": 4.890756398957114e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9942701995372772, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005729795014485717, + "mgd_loss_weight": 0.25149171352386473, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.1540893554687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.09960021972654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22460, + "valid_token_count": 818.27265625 + }, + { + "actual_token_mask_ratio": 0.29947671890258787, + "epoch": 6.537678207739307, + "flow_matching_loss": 0.010221983771771193, + "grad_norm": 0.06495460122823715, + "kept_token_count": 573.17265625, + "learning_rate": 4.8785470827750075e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9938801825046539, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061198108363896605, + "mgd_loss_weight": 0.25115837156772614, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.9165985107422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.26006774902345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22470, + "valid_token_count": 818.20390625 + }, + { + "actual_token_mask_ratio": 0.2998882859945297, + "epoch": 6.540587721850451, + "flow_matching_loss": 0.010204431368038058, + "grad_norm": 0.06253832578659058, + "kept_token_count": 572.73515625, + "learning_rate": 4.866350065112588e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9940023839473724, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005997600359842181, + "mgd_loss_weight": 0.25082503259181976, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.4282989501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7602111816406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22480, + "valid_token_count": 818.06015625 + }, + { + "actual_token_mask_ratio": 0.3004168689250946, + "epoch": 6.543497235961595, + "flow_matching_loss": 0.009365748707205057, + "grad_norm": 0.08682353049516678, + "kept_token_count": 572.41875, + "learning_rate": 4.854165360790381e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9938930094242096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061069906689226626, + "mgd_loss_weight": 0.2504916787147522, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.91407165527346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8283752441406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22490, + "valid_token_count": 818.22890625 + }, + { + "actual_token_mask_ratio": 0.2997942119836807, + "epoch": 6.546406750072737, + "flow_matching_loss": 0.00997958518564701, + "grad_norm": 0.0822925940155983, + "kept_token_count": 572.9109375, + "learning_rate": 4.8419929846139425e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9935687124729157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006431295583024621, + "mgd_loss_weight": 0.2501583397388458, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.5534973144531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0337158203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22500, + "valid_token_count": 818.203125 + }, + { + "actual_token_mask_ratio": 0.2993078351020813, + "epoch": 6.549316264183881, + "flow_matching_loss": 0.011427054088562727, + "grad_norm": 0.07481537759304047, + "kept_token_count": 573.38125, + "learning_rate": 4.829832951373872e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9941357433795929, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005864256899803877, + "mgd_loss_weight": 0.24982499331235886, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.1269104003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3416778564453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22510, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.30004789531230924, + "epoch": 6.552225778295025, + "flow_matching_loss": 0.01087102210149169, + "grad_norm": 0.0740644708275795, + "kept_token_count": 572.71328125, + "learning_rate": 4.817685275845746e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9940088033676148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005991188809275627, + "mgd_loss_weight": 0.24949165433645248, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.22191467285154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8558044433594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22520, + "valid_token_count": 818.2109375 + }, + { + "actual_token_mask_ratio": 0.3004326164722443, + "epoch": 6.5551352924061685, + "flow_matching_loss": 0.011569209955632686, + "grad_norm": 0.07644712179899216, + "kept_token_count": 572.2359375, + "learning_rate": 4.805549972790143e-06, + "loss": 0.013, + "mgd_cos_sim": 0.994274914264679, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005725093279033899, + "mgd_loss_weight": 0.24915830045938492, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.7506042480469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.13121643066404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22530, + "valid_token_count": 817.98203125 + }, + { + "actual_token_mask_ratio": 0.2997427672147751, + "epoch": 6.558044806517311, + "flow_matching_loss": 0.010280196694657207, + "grad_norm": 0.07190106809139252, + "kept_token_count": 573.0734375, + "learning_rate": 4.7934270569525966e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.994157749414444, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00584225095808506, + "mgd_loss_weight": 0.24882496148347855, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.1372467041016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.3635314941406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22540, + "valid_token_count": 818.38359375 + }, + { + "actual_token_mask_ratio": 0.2993281245231628, + "epoch": 6.560954320628455, + "flow_matching_loss": 0.010404905397444963, + "grad_norm": 0.0865972489118576, + "kept_token_count": 573.4265625, + "learning_rate": 4.7813165430635874e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9940177917480468, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005982204992324114, + "mgd_loss_weight": 0.2484916180372238, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.8948272705078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5604553222656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22550, + "valid_token_count": 818.39375 + }, + { + "actual_token_mask_ratio": 0.29995736479759216, + "epoch": 6.563863834739599, + "flow_matching_loss": 0.010954535426571965, + "grad_norm": 0.08559110760688782, + "kept_token_count": 572.80859375, + "learning_rate": 4.769218445838544e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9937955856323242, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006204418186098337, + "mgd_loss_weight": 0.2481582686305046, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.5134582519531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.21250305175784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22560, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.2997695654630661, + "epoch": 6.566773348850742, + "flow_matching_loss": 0.010954609140753746, + "grad_norm": 0.07131433486938477, + "kept_token_count": 573.1453125, + "learning_rate": 4.757132779977787e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9938877880573272, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006112197693437338, + "mgd_loss_weight": 0.24782492965459824, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.6999816894531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.41640014648436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22570, + "valid_token_count": 818.51015625 + }, + { + "actual_token_mask_ratio": 0.2999669581651688, + "epoch": 6.569682862961885, + "flow_matching_loss": 0.009592374553903937, + "grad_norm": 0.07389364391565323, + "kept_token_count": 572.75078125, + "learning_rate": 4.745059560166542e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9937470436096192, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00625296556390822, + "mgd_loss_weight": 0.24749158322811127, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.56358947753904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.76536254882814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22580, + "valid_token_count": 818.18046875 + }, + { + "actual_token_mask_ratio": 0.30049356520175935, + "epoch": 6.572592377073029, + "flow_matching_loss": 0.009856903972104192, + "grad_norm": 0.08184609562158585, + "kept_token_count": 572.25703125, + "learning_rate": 4.732998801074911e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9939482271671295, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006051767664030194, + "mgd_loss_weight": 0.2471582368016243, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.8992431640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.71702575683594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22590, + "valid_token_count": 818.084375 + }, + { + "actual_token_mask_ratio": 0.29968743622303007, + "epoch": 6.575501891184173, + "flow_matching_loss": 0.010146331414580346, + "grad_norm": 0.0849703699350357, + "kept_token_count": 573.09375, + "learning_rate": 4.7209505173578445e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9938923299312592, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006107671419158578, + "mgd_loss_weight": 0.24682489782571793, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.1316802978516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.26605224609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22600, + "valid_token_count": 818.340625 + }, + { + "actual_token_mask_ratio": 0.29984105825424195, + "epoch": 6.5784114052953155, + "flow_matching_loss": 0.011701003881171346, + "grad_norm": 0.1020822823047638, + "kept_token_count": 572.9078125, + "learning_rate": 4.7089147236551555e-06, + "loss": 0.0133, + "mgd_cos_sim": 0.9935132920742035, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00648671411909163, + "mgd_loss_weight": 0.2464915469288826, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.5017059326172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0715393066406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22610, + "valid_token_count": 818.25234375 + }, + { + "actual_token_mask_ratio": 0.3002148926258087, + "epoch": 6.581320919406459, + "flow_matching_loss": 0.009807229368016123, + "grad_norm": 0.06809519976377487, + "kept_token_count": 572.6515625, + "learning_rate": 4.696891434591464e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9933063089847565, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006693667033687234, + "mgd_loss_weight": 0.246158204972744, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.1110473632813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4663482666016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22620, + "valid_token_count": 818.32890625 + }, + { + "actual_token_mask_ratio": 0.30035998225212096, + "epoch": 6.584230433517602, + "flow_matching_loss": 0.011525177117437124, + "grad_norm": 0.13366878032684326, + "kept_token_count": 572.4125, + "learning_rate": 4.684880664776199e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9938327968120575, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006167203327640891, + "mgd_loss_weight": 0.2458248645067215, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.6167846679688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5933807373047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22630, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.30064069032669066, + "epoch": 6.587139947628746, + "flow_matching_loss": 0.012036432698369025, + "grad_norm": 0.09144806861877441, + "kept_token_count": 572.2625, + "learning_rate": 4.6728824288035784e-06, + "loss": 0.0136, + "mgd_cos_sim": 0.9937463164329529, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062536787241697315, + "mgd_loss_weight": 0.24549151211977005, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.2647277832031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0181427001953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22640, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.29970456659793854, + "epoch": 6.5900494617398895, + "flow_matching_loss": 0.011455403361469507, + "grad_norm": 0.06683564186096191, + "kept_token_count": 572.8984375, + "learning_rate": 4.660896741252586e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9941006898880005, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005899311462417245, + "mgd_loss_weight": 0.24515817314386368, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.45814208984376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8289001464844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22650, + "valid_token_count": 818.08046875 + }, + { + "actual_token_mask_ratio": 0.30029481947422026, + "epoch": 6.592958975851033, + "flow_matching_loss": 0.010998565703630447, + "grad_norm": 0.07092224061489105, + "kept_token_count": 572.64453125, + "learning_rate": 4.6489236166869714e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9937945485115052, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062054447829723355, + "mgd_loss_weight": 0.24482482820749282, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.96942443847655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.06581115722656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22660, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.299459707736969, + "epoch": 6.595868489962176, + "flow_matching_loss": 0.008937644260004162, + "grad_norm": 0.07577329874038696, + "kept_token_count": 573.22578125, + "learning_rate": 4.6369630696552e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9938584923744201, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006141511816531419, + "mgd_loss_weight": 0.24449148029088974, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.4641510009766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4665100097656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22670, + "valid_token_count": 818.26171875 + }, + { + "actual_token_mask_ratio": 0.29964557886123655, + "epoch": 6.59877800407332, + "flow_matching_loss": 0.01076484345830977, + "grad_norm": 0.0671355277299881, + "kept_token_count": 573.08359375, + "learning_rate": 4.625015114690474e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9937222242355347, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006277768267318606, + "mgd_loss_weight": 0.24415814131498337, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.87847900390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.01427001953124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22680, + "valid_token_count": 818.26953125 + }, + { + "actual_token_mask_ratio": 0.2997918963432312, + "epoch": 6.601687518184463, + "flow_matching_loss": 0.009938058909028769, + "grad_norm": 0.0716327428817749, + "kept_token_count": 572.9796875, + "learning_rate": 4.613079766310682e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9939693510532379, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006030651647597552, + "mgd_loss_weight": 0.24382479339838029, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.38646240234374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2966339111328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22690, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.3002728074789047, + "epoch": 6.604597032295606, + "flow_matching_loss": 0.010285191051661969, + "grad_norm": 0.07063740491867065, + "kept_token_count": 572.5875, + "learning_rate": 4.601157039018396e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9940062701702118, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005993714882060885, + "mgd_loss_weight": 0.24349144846200943, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.41143798828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4757568359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22700, + "valid_token_count": 818.3 + }, + { + "actual_token_mask_ratio": 0.3001304596662521, + "epoch": 6.60750654640675, + "flow_matching_loss": 0.009106161585077644, + "grad_norm": 0.06721111387014389, + "kept_token_count": 572.625, + "learning_rate": 4.589246947300849e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9936285078525543, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006371484743431211, + "mgd_loss_weight": 0.24315810948610306, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.5323425292969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2789001464844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22710, + "valid_token_count": 818.18671875 + }, + { + "actual_token_mask_ratio": 0.2997980922460556, + "epoch": 6.610416060517894, + "flow_matching_loss": 0.010206322930753232, + "grad_norm": 0.0514913834631443, + "kept_token_count": 572.9640625, + "learning_rate": 4.577349505629936e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.993742311000824, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006257695239037275, + "mgd_loss_weight": 0.2428247570991516, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.02431640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.38975524902344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22720, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.2997483164072037, + "epoch": 6.613325574629037, + "flow_matching_loss": 0.010864907596260309, + "grad_norm": 0.0739419236779213, + "kept_token_count": 573.009375, + "learning_rate": 4.565464728462165e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9937429010868073, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006257103569805622, + "mgd_loss_weight": 0.24249141663312912, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.3756591796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5154693603516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22730, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.29948394298553466, + "epoch": 6.61623508874018, + "flow_matching_loss": 0.011430455092340707, + "grad_norm": 0.061499595642089844, + "kept_token_count": 573.27265625, + "learning_rate": 4.553592630238663e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9937000751495362, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006299922242760658, + "mgd_loss_weight": 0.24215807467699052, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.1306518554687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.33738708496094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22740, + "valid_token_count": 818.35703125 + }, + { + "actual_token_mask_ratio": 0.29995274245738984, + "epoch": 6.619144602851324, + "flow_matching_loss": 0.010340344114229082, + "grad_norm": 0.08590958267450333, + "kept_token_count": 572.98125, + "learning_rate": 4.541733225385146e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9937330663204194, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006266948394477367, + "mgd_loss_weight": 0.24182472378015518, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.79860534667966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.98651428222655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22750, + "valid_token_count": 818.48984375 + }, + { + "actual_token_mask_ratio": 0.2998057812452316, + "epoch": 6.622054116962468, + "flow_matching_loss": 0.010139535274356603, + "grad_norm": 0.08243528008460999, + "kept_token_count": 572.9828125, + "learning_rate": 4.529886528311905e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9939978420734406, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006002161093056202, + "mgd_loss_weight": 0.2414913848042488, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.7246795654297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4120819091797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22760, + "valid_token_count": 818.32265625 + }, + { + "actual_token_mask_ratio": 0.3005924552679062, + "epoch": 6.62496363107361, + "flow_matching_loss": 0.011054898938164115, + "grad_norm": 0.06145981326699257, + "kept_token_count": 572.2796875, + "learning_rate": 4.5180525534138045e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.993828934431076, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006171062542125583, + "mgd_loss_weight": 0.24115803837776184, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.9267852783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.20350341796876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22770, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.30048738718032836, + "epoch": 6.627873145184754, + "flow_matching_loss": 0.011528067803010345, + "grad_norm": 0.06438632309436798, + "kept_token_count": 572.2578125, + "learning_rate": 4.506231315070235e-06, + "loss": 0.013, + "mgd_cos_sim": 0.99374138712883, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006258618738502264, + "mgd_loss_weight": 0.24082469195127487, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.7496612548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.75709533691406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22780, + "valid_token_count": 818.07890625 + }, + { + "actual_token_mask_ratio": 0.29933922290802, + "epoch": 6.630782659295898, + "flow_matching_loss": 0.010683063883334398, + "grad_norm": 0.08454517275094986, + "kept_token_count": 573.296875, + "learning_rate": 4.494422827645115e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.994184285402298, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0058157230261713265, + "mgd_loss_weight": 0.2404913529753685, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.9639831542969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8938812255859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22790, + "valid_token_count": 818.21796875 + }, + { + "actual_token_mask_ratio": 0.2988906383514404, + "epoch": 6.633692173407041, + "flow_matching_loss": 0.010054138721898198, + "grad_norm": 0.09248477220535278, + "kept_token_count": 573.628125, + "learning_rate": 4.482627105486865e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9939023077487945, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006097695417702198, + "mgd_loss_weight": 0.2401580035686493, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.14873962402345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.98541870117185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22800, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.30000662207603457, + "epoch": 6.636601687518184, + "flow_matching_loss": 0.010195054300129413, + "grad_norm": 0.06793753057718277, + "kept_token_count": 572.7, + "learning_rate": 4.470844162928412e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9938853859901429, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006114615546539426, + "mgd_loss_weight": 0.23982466012239456, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.752880859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.65232238769534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22810, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.3000239968299866, + "epoch": 6.639511201629328, + "flow_matching_loss": 0.010261870920658112, + "grad_norm": 0.08077941834926605, + "kept_token_count": 572.77890625, + "learning_rate": 4.459074014287129e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9940477073192596, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005952266743406653, + "mgd_loss_weight": 0.2394913211464882, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.6663024902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.29331970214844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22820, + "valid_token_count": 818.28515625 + }, + { + "actual_token_mask_ratio": 0.29984256625175476, + "epoch": 6.642420715740471, + "flow_matching_loss": 0.010938902804628014, + "grad_norm": 0.15176020562648773, + "kept_token_count": 572.93046875, + "learning_rate": 4.447316673864869e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9937080681324005, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006291929399594665, + "mgd_loss_weight": 0.23915796726942062, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.38262939453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.67485961914065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22830, + "valid_token_count": 818.29140625 + }, + { + "actual_token_mask_ratio": 0.29995745718479155, + "epoch": 6.645330229851615, + "flow_matching_loss": 0.010217520827427506, + "grad_norm": 0.08041997998952866, + "kept_token_count": 572.71328125, + "learning_rate": 4.435572155947903e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9937533676624298, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062466227449476715, + "mgd_loss_weight": 0.23882462829351425, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.7427978515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2715759277344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22840, + "valid_token_count": 818.11796875 + }, + { + "actual_token_mask_ratio": 0.3004305064678192, + "epoch": 6.648239743962758, + "flow_matching_loss": 0.010379320802167058, + "grad_norm": 0.09054520726203918, + "kept_token_count": 572.46640625, + "learning_rate": 4.423840474806929e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9935827553272247, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006417241645976901, + "mgd_loss_weight": 0.2384912848472595, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.9649658203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8198638916016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22850, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.29986492693424227, + "epoch": 6.651149258073902, + "flow_matching_loss": 0.011127088870853185, + "grad_norm": 0.07172489166259766, + "kept_token_count": 572.9640625, + "learning_rate": 4.412121644697043e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9937744915485383, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006225502584129572, + "mgd_loss_weight": 0.2381579354405403, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.26279296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.78759765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22860, + "valid_token_count": 818.36796875 + }, + { + "actual_token_mask_ratio": 0.2999331057071686, + "epoch": 6.654058772185045, + "flow_matching_loss": 0.00990416742861271, + "grad_norm": 0.09400661289691925, + "kept_token_count": 572.6125, + "learning_rate": 4.400415679857727e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9936791241168976, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006320884916931391, + "mgd_loss_weight": 0.23782459646463394, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.1116943359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.46563110351565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22870, + "valid_token_count": 817.9375 + }, + { + "actual_token_mask_ratio": 0.3000173598527908, + "epoch": 6.6569682862961885, + "flow_matching_loss": 0.011169072706252337, + "grad_norm": 0.09250038862228394, + "kept_token_count": 572.70625, + "learning_rate": 4.388722594512838e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9935028612613678, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064971389714628455, + "mgd_loss_weight": 0.23749124854803086, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.8388244628906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2593078613281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22880, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.29992611408233644, + "epoch": 6.659877800407332, + "flow_matching_loss": 0.010908689210191369, + "grad_norm": 0.07964923232793808, + "kept_token_count": 572.83984375, + "learning_rate": 4.377042402870577e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9937773227691651, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006222674017772078, + "mgd_loss_weight": 0.23715790361166, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.0663787841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.85007629394534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22890, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.29948092103004453, + "epoch": 6.662787314518475, + "flow_matching_loss": 0.009564476646482944, + "grad_norm": 0.07716059684753418, + "kept_token_count": 573.20234375, + "learning_rate": 4.365375119123476e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9940399527549744, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005960055347532034, + "mgd_loss_weight": 0.23682456463575363, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.73020324707034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3748809814453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22900, + "valid_token_count": 818.25078125 + }, + { + "actual_token_mask_ratio": 0.29965580999851227, + "epoch": 6.665696828629619, + "flow_matching_loss": 0.009584633447229863, + "grad_norm": 0.07722853124141693, + "kept_token_count": 573.2859375, + "learning_rate": 4.353720757448383e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9937844634056091, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062155370134860275, + "mgd_loss_weight": 0.2364912137389183, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.1251922607422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4496154785156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22910, + "valid_token_count": 818.578125 + }, + { + "actual_token_mask_ratio": 0.299979966878891, + "epoch": 6.6686063427407625, + "flow_matching_loss": 0.01019150260835886, + "grad_norm": 0.0848519429564476, + "kept_token_count": 572.97734375, + "learning_rate": 4.342079332006445e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9939254999160767, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006074495799839497, + "mgd_loss_weight": 0.2361578717827797, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.3245513916016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2011474609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22920, + "valid_token_count": 818.51328125 + }, + { + "actual_token_mask_ratio": 0.3003125756978989, + "epoch": 6.671515856851906, + "flow_matching_loss": 0.010356157505884766, + "grad_norm": 0.08696092665195465, + "kept_token_count": 572.5296875, + "learning_rate": 4.330450856943101e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9937027394771576, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006297259451821446, + "mgd_loss_weight": 0.2358245313167572, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.76499328613284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.65545959472655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22930, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.30070233047008516, + "epoch": 6.674425370963049, + "flow_matching_loss": 0.01011166381649673, + "grad_norm": 0.06488236784934998, + "kept_token_count": 572.29453125, + "learning_rate": 4.318835346388036e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9939416289329529, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006058369716629386, + "mgd_loss_weight": 0.23549117892980576, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.2445404052734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4571929931641, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22940, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.30006611049175264, + "epoch": 6.677334885074193, + "flow_matching_loss": 0.01000360669568181, + "grad_norm": 0.07276339828968048, + "kept_token_count": 572.815625, + "learning_rate": 4.307232814455189e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.994025456905365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005974553572013974, + "mgd_loss_weight": 0.23515783995389938, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.6236572265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9585479736328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22950, + "valid_token_count": 818.38828125 + }, + { + "actual_token_mask_ratio": 0.300008162856102, + "epoch": 6.680244399185336, + "flow_matching_loss": 0.010439067101106047, + "grad_norm": 0.09685719013214111, + "kept_token_count": 572.80234375, + "learning_rate": 4.29564327524274e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9939041376113892, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060958590824157, + "mgd_loss_weight": 0.23482449501752853, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.9413360595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5586883544922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22960, + "valid_token_count": 818.29765625 + }, + { + "actual_token_mask_ratio": 0.3003692775964737, + "epoch": 6.683153913296479, + "flow_matching_loss": 0.01111569907516241, + "grad_norm": 0.06422224640846252, + "kept_token_count": 572.46171875, + "learning_rate": 4.284066742833061e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9940785706043244, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005921424087136984, + "mgd_loss_weight": 0.23449114710092545, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.986083984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4203216552734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22970, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.29993997514247894, + "epoch": 6.686063427407623, + "flow_matching_loss": 0.010703627625480295, + "grad_norm": 0.12964212894439697, + "kept_token_count": 572.7578125, + "learning_rate": 4.272503231292741e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9937472403049469, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006252761883661151, + "mgd_loss_weight": 0.23415780812501907, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 436.06463928222655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3325988769531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22980, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.29972596764564513, + "epoch": 6.688972941518767, + "flow_matching_loss": 0.011298395739868284, + "grad_norm": 0.08842388540506363, + "kept_token_count": 572.9109375, + "learning_rate": 4.260952754672532e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9939438998699188, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006056110141798854, + "mgd_loss_weight": 0.23382445871829988, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.22141723632814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7946472167969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 22990, + "valid_token_count": 818.1296875 + }, + { + "actual_token_mask_ratio": 0.3002563685178757, + "epoch": 6.6918824556299095, + "flow_matching_loss": 0.011406027199700474, + "grad_norm": 0.07315495610237122, + "kept_token_count": 572.4734375, + "learning_rate": 4.249415327007353e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9934366166591644, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006563384877517819, + "mgd_loss_weight": 0.23349111527204514, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.2563995361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.00575256347656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23000, + "valid_token_count": 818.1171875 + }, + { + "actual_token_mask_ratio": 0.30012274980545045, + "epoch": 6.694791969741053, + "flow_matching_loss": 0.011084899818524719, + "grad_norm": 0.07882509380578995, + "kept_token_count": 572.525, + "learning_rate": 4.237890962316264e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9939314067363739, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006068598385900259, + "mgd_loss_weight": 0.23315777629613876, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.45103454589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1612060546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23010, + "valid_token_count": 818.03828125 + }, + { + "actual_token_mask_ratio": 0.2998686760663986, + "epoch": 6.697701483852197, + "flow_matching_loss": 0.010802445374429226, + "grad_norm": 0.0945766344666481, + "kept_token_count": 572.9484375, + "learning_rate": 4.226379674602452e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9941446304321289, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0058553605806082485, + "mgd_loss_weight": 0.2328244239091873, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 437.18626403808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3388610839844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23020, + "valid_token_count": 818.34453125 + }, + { + "actual_token_mask_ratio": 0.29975156486034393, + "epoch": 6.70061099796334, + "flow_matching_loss": 0.011216255417093635, + "grad_norm": 0.08079510927200317, + "kept_token_count": 572.915625, + "learning_rate": 4.214881477853227e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9937972962856293, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006202700966969133, + "mgd_loss_weight": 0.23249108344316483, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 438.0593200683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.05735168457034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23030, + "valid_token_count": 818.15859375 + }, + { + "actual_token_mask_ratio": 0.2999885410070419, + "epoch": 6.703520512074483, + "flow_matching_loss": 0.010449646785855294, + "grad_norm": 0.052907198667526245, + "kept_token_count": 572.74296875, + "learning_rate": 4.2033963860399745e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.993680065870285, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006319935619831085, + "mgd_loss_weight": 0.23215774148702623, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.5100463867187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4867370605469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23040, + "valid_token_count": 818.19140625 + }, + { + "actual_token_mask_ratio": 0.3004355698823929, + "epoch": 6.706430026185627, + "flow_matching_loss": 0.008952199062332512, + "grad_norm": 0.08908766508102417, + "kept_token_count": 572.35234375, + "learning_rate": 4.1919244131181696e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9940505564212799, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005949454801157117, + "mgd_loss_weight": 0.2318243905901909, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.6447296142578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.50518188476565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23050, + "valid_token_count": 818.15703125 + }, + { + "actual_token_mask_ratio": 0.3003302216529846, + "epoch": 6.709339540296771, + "flow_matching_loss": 0.010367145016789436, + "grad_norm": 0.06175202503800392, + "kept_token_count": 572.6421875, + "learning_rate": 4.180465573027335e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9941160261631012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0058839738368988035, + "mgd_loss_weight": 0.23149105161428452, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.2898345947266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1815887451172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23060, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.29927540123462676, + "epoch": 6.712249054407914, + "flow_matching_loss": 0.008936766069382429, + "grad_norm": 0.07044392824172974, + "kept_token_count": 573.31015625, + "learning_rate": 4.169019879691046e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9936833918094635, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006316616665571928, + "mgd_loss_weight": 0.23115770518779755, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.1247894287109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0149383544922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23070, + "valid_token_count": 818.17265625 + }, + { + "actual_token_mask_ratio": 0.30032444298267363, + "epoch": 6.715158568519057, + "flow_matching_loss": 0.010426302114501596, + "grad_norm": 0.07628566771745682, + "kept_token_count": 572.61875, + "learning_rate": 4.157587347016891e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9941109120845795, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005889092991128564, + "mgd_loss_weight": 0.23082435876131058, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.71144409179686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.05519104003906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23080, + "valid_token_count": 818.40703125 + }, + { + "actual_token_mask_ratio": 0.2999100387096405, + "epoch": 6.718068082630201, + "flow_matching_loss": 0.011059652687981725, + "grad_norm": 0.06295279413461685, + "kept_token_count": 572.81796875, + "learning_rate": 4.14616798889649e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9940033257007599, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059966896660625935, + "mgd_loss_weight": 0.2304910197854042, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.9762664794922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.78676452636716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23090, + "valid_token_count": 818.21015625 + }, + { + "actual_token_mask_ratio": 0.30016350746154785, + "epoch": 6.720977596741344, + "flow_matching_loss": 0.00919338883832097, + "grad_norm": 0.07309964299201965, + "kept_token_count": 572.7265625, + "learning_rate": 4.134761819205425e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.9936812996864319, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006318707391619683, + "mgd_loss_weight": 0.23015766888856887, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.09897766113284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.39794006347654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23100, + "valid_token_count": 818.37109375 + }, + { + "actual_token_mask_ratio": 0.29950729608535764, + "epoch": 6.723887110852488, + "flow_matching_loss": 0.010054918937385082, + "grad_norm": 0.07338729500770569, + "kept_token_count": 573.1828125, + "learning_rate": 4.1233688518032794e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9938056826591491, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006194307468831539, + "mgd_loss_weight": 0.22982432693243027, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.7683837890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3744201660156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23110, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.29962447881698606, + "epoch": 6.726796624963631, + "flow_matching_loss": 0.010010181507095695, + "grad_norm": 0.06539848446846008, + "kept_token_count": 573.2, + "learning_rate": 4.1119891005335795e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9940160274505615, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005983983725309372, + "mgd_loss_weight": 0.2294909879565239, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.1874298095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0445587158203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23120, + "valid_token_count": 818.41953125 + }, + { + "actual_token_mask_ratio": 0.2992144823074341, + "epoch": 6.729706139074775, + "flow_matching_loss": 0.009413896827027202, + "grad_norm": 0.08854766190052032, + "kept_token_count": 573.3390625, + "learning_rate": 4.1006225792237895e-06, + "loss": 0.0108, + "mgd_cos_sim": 0.9940809428691864, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005919048748910427, + "mgd_loss_weight": 0.22915763407945633, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.2743682861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.22901611328126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23130, + "valid_token_count": 818.1421875 + }, + { + "actual_token_mask_ratio": 0.2996213436126709, + "epoch": 6.732615653185918, + "flow_matching_loss": 0.010046668257564307, + "grad_norm": 0.059531070291996, + "kept_token_count": 573.02265625, + "learning_rate": 4.089269301685314e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9938540875911712, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006145903747528792, + "mgd_loss_weight": 0.22882429510354996, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 435.0460601806641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5615966796875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23140, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.2998559236526489, + "epoch": 6.7355251672970615, + "flow_matching_loss": 0.009259972348809242, + "grad_norm": 0.05947510898113251, + "kept_token_count": 572.88125, + "learning_rate": 4.077929281713453e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.9941512227058411, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005848770588636398, + "mgd_loss_weight": 0.22849095165729522, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.3632385253906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3186920166016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23150, + "valid_token_count": 818.23359375 + }, + { + "actual_token_mask_ratio": 0.2999052941799164, + "epoch": 6.738434681408204, + "flow_matching_loss": 0.010956945549696683, + "grad_norm": 0.07533618062734604, + "kept_token_count": 572.88671875, + "learning_rate": 4.066602533087398e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9936894178390503, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006310591427609325, + "mgd_loss_weight": 0.22815760225057602, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 404.8654418945313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7257843017578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23160, + "valid_token_count": 818.29921875 + }, + { + "actual_token_mask_ratio": 0.300362166762352, + "epoch": 6.741344195519348, + "flow_matching_loss": 0.01047995239496231, + "grad_norm": 0.07040742039680481, + "kept_token_count": 572.49375, + "learning_rate": 4.0552890695702175e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9939053356647491, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006094659818336368, + "mgd_loss_weight": 0.22782426327466965, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.5506530761719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7342529296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23170, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.30001612901687624, + "epoch": 6.744253709630492, + "flow_matching_loss": 0.010028928657993675, + "grad_norm": 0.08435887843370438, + "kept_token_count": 572.78984375, + "learning_rate": 4.04398890490883e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9938513338565826, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006148668052628636, + "mgd_loss_weight": 0.22749091535806656, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.13600158691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8005096435547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23180, + "valid_token_count": 818.29375 + }, + { + "actual_token_mask_ratio": 0.3003817081451416, + "epoch": 6.7471632237416355, + "flow_matching_loss": 0.010281465016305447, + "grad_norm": 0.07489631325006485, + "kept_token_count": 572.6125, + "learning_rate": 4.032702052834013e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9940135300159454, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005986476456746459, + "mgd_loss_weight": 0.2271575704216957, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.70009765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3101440429688, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23190, + "valid_token_count": 818.4640625 + }, + { + "actual_token_mask_ratio": 0.30022079646587374, + "epoch": 6.750072737852778, + "flow_matching_loss": 0.010343708377331496, + "grad_norm": 0.10472963005304337, + "kept_token_count": 572.728125, + "learning_rate": 4.021428527060352e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9939695537090302, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006030445406213403, + "mgd_loss_weight": 0.22682423144578934, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.7288116455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8658813476562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23200, + "valid_token_count": 818.43515625 + }, + { + "actual_token_mask_ratio": 0.3000446975231171, + "epoch": 6.752982251963922, + "flow_matching_loss": 0.010900179203599692, + "grad_norm": 0.08480072021484375, + "kept_token_count": 572.71171875, + "learning_rate": 4.010168341286241e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9936534523963928, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063465629704296585, + "mgd_loss_weight": 0.22649087905883789, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.05523986816405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4877532958984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23210, + "valid_token_count": 818.20703125 + }, + { + "actual_token_mask_ratio": 0.30018527805805206, + "epoch": 6.755891766075066, + "flow_matching_loss": 0.010285377921536564, + "grad_norm": 0.11700256913900375, + "kept_token_count": 572.5765625, + "learning_rate": 3.998921509193868e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9938689291477203, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006131086498498917, + "mgd_loss_weight": 0.2261575385928154, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.55909729003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3026123046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23220, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.30068525671958923, + "epoch": 6.7588012801862085, + "flow_matching_loss": 0.010938500240445136, + "grad_norm": 0.07498282939195633, + "kept_token_count": 572.21484375, + "learning_rate": 3.987688044449193e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9939435243606567, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006056457292288542, + "mgd_loss_weight": 0.2258241981267929, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.72755737304686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.46275024414064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23230, + "valid_token_count": 818.24296875 + }, + { + "actual_token_mask_ratio": 0.2995401084423065, + "epoch": 6.761710794297352, + "flow_matching_loss": 0.012306642485782504, + "grad_norm": 0.07972784340381622, + "kept_token_count": 573.203125, + "learning_rate": 3.9764679607019385e-06, + "loss": 0.0137, + "mgd_cos_sim": 0.9940031051635743, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005996899632737041, + "mgd_loss_weight": 0.22549084573984146, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.46160583496095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9003936767578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23240, + "valid_token_count": 818.321875 + }, + { + "actual_token_mask_ratio": 0.3003484010696411, + "epoch": 6.764620308408496, + "flow_matching_loss": 0.010660809464752674, + "grad_norm": 0.08344599604606628, + "kept_token_count": 572.415625, + "learning_rate": 3.965261271585561e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9942512810230255, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005748707056045532, + "mgd_loss_weight": 0.2251575067639351, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.942236328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4238616943359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23250, + "valid_token_count": 818.1453125 + }, + { + "actual_token_mask_ratio": 0.300260654091835, + "epoch": 6.76752982251964, + "flow_matching_loss": 0.010073263803496957, + "grad_norm": 0.07610130310058594, + "kept_token_count": 572.44375, + "learning_rate": 3.954067990717251e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9933787882328033, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006621214468032122, + "mgd_loss_weight": 0.22482416182756423, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.90614929199216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1044982910156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23260, + "valid_token_count": 818.08359375 + }, + { + "actual_token_mask_ratio": 0.29984923005104064, + "epoch": 6.7704393366307825, + "flow_matching_loss": 0.009900212287902832, + "grad_norm": 0.07800524681806564, + "kept_token_count": 572.9609375, + "learning_rate": 3.9428881316978935e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9938095390796662, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006190461898222565, + "mgd_loss_weight": 0.22449081391096115, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.20819396972655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1142822265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23270, + "valid_token_count": 818.33671875 + }, + { + "actual_token_mask_ratio": 0.29998658299446107, + "epoch": 6.773348850741926, + "flow_matching_loss": 0.011122400453314186, + "grad_norm": 0.06181173026561737, + "kept_token_count": 572.6078125, + "learning_rate": 3.931721708112075e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9935113787651062, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006488625286146998, + "mgd_loss_weight": 0.22415747493505478, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.258056640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0421478271484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23280, + "valid_token_count": 817.996875 + }, + { + "actual_token_mask_ratio": 0.3001531571149826, + "epoch": 6.77625836485307, + "flow_matching_loss": 0.009575322130694985, + "grad_norm": 0.06965421140193939, + "kept_token_count": 572.85234375, + "learning_rate": 3.920568733528047e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9937390148639679, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006260986439883709, + "mgd_loss_weight": 0.22382412552833558, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.7770172119141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.61220703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23290, + "valid_token_count": 818.540625 + }, + { + "actual_token_mask_ratio": 0.30006619691848757, + "epoch": 6.779167878964213, + "flow_matching_loss": 0.010248426487669349, + "grad_norm": 0.0627995952963829, + "kept_token_count": 572.6625, + "learning_rate": 3.909429221497734e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9938032627105713, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006196738593280316, + "mgd_loss_weight": 0.22349078208208084, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.45274047851564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.295166015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23300, + "valid_token_count": 818.16796875 + }, + { + "actual_token_mask_ratio": 0.300016513466835, + "epoch": 6.782077393075356, + "flow_matching_loss": 0.010280262166634201, + "grad_norm": 0.07379976660013199, + "kept_token_count": 572.68671875, + "learning_rate": 3.898303185556689e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9936664223670959, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006333581870421767, + "mgd_loss_weight": 0.22315744310617447, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.45009155273436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.04224853515626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23310, + "valid_token_count": 818.1390625 + }, + { + "actual_token_mask_ratio": 0.2996005266904831, + "epoch": 6.7849869071865, + "flow_matching_loss": 0.01089261299930513, + "grad_norm": 0.08739498257637024, + "kept_token_count": 573.34375, + "learning_rate": 3.887190639224097e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9935744404792786, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006425550626590848, + "mgd_loss_weight": 0.2228240892291069, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.3263641357422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8358673095703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23320, + "valid_token_count": 818.6 + }, + { + "actual_token_mask_ratio": 0.30020029544830323, + "epoch": 6.787896421297643, + "flow_matching_loss": 0.009872946003451944, + "grad_norm": 0.0656718835234642, + "kept_token_count": 572.76484375, + "learning_rate": 3.876091596002746e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.994132661819458, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005867346795275808, + "mgd_loss_weight": 0.22249075025320053, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.8084747314453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0298248291016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23330, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.2996238708496094, + "epoch": 6.790805935408787, + "flow_matching_loss": 0.010909897135570645, + "grad_norm": 0.08072823286056519, + "kept_token_count": 573.15703125, + "learning_rate": 3.865006069379019e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.993531060218811, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006468934658914805, + "mgd_loss_weight": 0.22215740829706193, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.3030548095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.800048828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23340, + "valid_token_count": 818.36015625 + }, + { + "actual_token_mask_ratio": 0.3000066041946411, + "epoch": 6.79371544951993, + "flow_matching_loss": 0.009560509491711855, + "grad_norm": 0.07186267524957657, + "kept_token_count": 572.88515625, + "learning_rate": 3.853934072822881e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9941613256931305, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005838675703853369, + "mgd_loss_weight": 0.2218240574002266, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.1797119140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.76917419433596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23350, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.3003662794828415, + "epoch": 6.796624963631073, + "flow_matching_loss": 0.009706883551552891, + "grad_norm": 0.08864068239927292, + "kept_token_count": 572.5828125, + "learning_rate": 3.8428756197878514e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9939022123813629, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006097780680283904, + "mgd_loss_weight": 0.22149071842432022, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.49065856933595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.0179962158203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23360, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.2994713425636292, + "epoch": 6.799534477742217, + "flow_matching_loss": 0.010764645412564278, + "grad_norm": 0.06396064162254333, + "kept_token_count": 573.3359375, + "learning_rate": 3.831830723710992e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9938734889030456, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006126507790759206, + "mgd_loss_weight": 0.22115737199783325, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.9095916748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.61383056640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23370, + "valid_token_count": 818.4390625 + }, + { + "actual_token_mask_ratio": 0.30096615552902223, + "epoch": 6.802443991853361, + "flow_matching_loss": 0.00992739903740585, + "grad_norm": 0.0691530704498291, + "kept_token_count": 571.93046875, + "learning_rate": 3.820799398012889e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9940058648586273, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00599412894807756, + "mgd_loss_weight": 0.22082402557134628, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.9716796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.54083251953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23380, + "valid_token_count": 818.17578125 + }, + { + "actual_token_mask_ratio": 0.3008991479873657, + "epoch": 6.805353505964504, + "flow_matching_loss": 0.010619972506538034, + "grad_norm": 0.08759129047393799, + "kept_token_count": 571.9265625, + "learning_rate": 3.80978165609765e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9939722001552582, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006027804873883724, + "mgd_loss_weight": 0.2204906865954399, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.1889343261719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4361999511719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23390, + "valid_token_count": 818.08828125 + }, + { + "actual_token_mask_ratio": 0.3003277748823166, + "epoch": 6.808263020075647, + "flow_matching_loss": 0.011674165818840265, + "grad_norm": 0.08232855796813965, + "kept_token_count": 572.45390625, + "learning_rate": 3.798777511352864e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9938281178474426, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006171883922070265, + "mgd_loss_weight": 0.22015733569860457, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.24598999023436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4693176269531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23400, + "valid_token_count": 818.17734375 + }, + { + "actual_token_mask_ratio": 0.3004380017518997, + "epoch": 6.811172534186791, + "flow_matching_loss": 0.010742456978186965, + "grad_norm": 0.06746367365121841, + "kept_token_count": 572.26953125, + "learning_rate": 3.787786977149613e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9937752068042756, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006224792916327715, + "mgd_loss_weight": 0.21982399374246597, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.43740234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6003143310547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23410, + "valid_token_count": 818.03828125 + }, + { + "actual_token_mask_ratio": 0.3004774868488312, + "epoch": 6.8140820482979345, + "flow_matching_loss": 0.0102489048615098, + "grad_norm": 0.07795066386461258, + "kept_token_count": 572.43984375, + "learning_rate": 3.776810066842427e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9935229420661926, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064770511351525785, + "mgd_loss_weight": 0.2194906532764435, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.61436157226564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.16698303222654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23420, + "valid_token_count": 818.32890625 + }, + { + "actual_token_mask_ratio": 0.3002776652574539, + "epoch": 6.816991562409077, + "flow_matching_loss": 0.010029842797666787, + "grad_norm": 0.06536580622196198, + "kept_token_count": 572.6296875, + "learning_rate": 3.7658467937692866e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9937695801258087, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006230427278205752, + "mgd_loss_weight": 0.21915730088949203, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.6976654052734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.91474609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23430, + "valid_token_count": 818.36484375 + }, + { + "actual_token_mask_ratio": 0.2992425441741943, + "epoch": 6.819901076520221, + "flow_matching_loss": 0.009611835144460202, + "grad_norm": 0.06637734919786453, + "kept_token_count": 573.51640625, + "learning_rate": 3.7548971712516e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9938286900520324, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006171303428709507, + "mgd_loss_weight": 0.21882396191358566, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.80096740722655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.65391845703124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23440, + "valid_token_count": 818.41875 + }, + { + "actual_token_mask_ratio": 0.3000862032175064, + "epoch": 6.822810590631365, + "flow_matching_loss": 0.010617024451494216, + "grad_norm": 0.077914297580719, + "kept_token_count": 572.765625, + "learning_rate": 3.743961212594197e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9939068615436554, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060931481420993805, + "mgd_loss_weight": 0.21849061846733092, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.2687255859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.34227600097654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23450, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.30002237260341647, + "epoch": 6.8257201047425085, + "flow_matching_loss": 0.010358828119933605, + "grad_norm": 0.07780610769987106, + "kept_token_count": 572.5953125, + "learning_rate": 3.7330389310852975e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9939516186714172, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006048391712829471, + "mgd_loss_weight": 0.21815726906061172, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.5147735595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6186889648437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23460, + "valid_token_count": 818.021875 + }, + { + "actual_token_mask_ratio": 0.29955360293388367, + "epoch": 6.828629618853651, + "flow_matching_loss": 0.011854986194521189, + "grad_norm": 0.08979327231645584, + "kept_token_count": 573.08515625, + "learning_rate": 3.7221303399965008e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.993843287229538, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006156718684360385, + "mgd_loss_weight": 0.21782393008470535, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.8976318359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.13356018066406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23470, + "valid_token_count": 818.17109375 + }, + { + "actual_token_mask_ratio": 0.30055502951145174, + "epoch": 6.831539132964795, + "flow_matching_loss": 0.010033410973846912, + "grad_norm": 0.10072649270296097, + "kept_token_count": 572.42890625, + "learning_rate": 3.711235452582775e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9940006673336029, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059993165545165535, + "mgd_loss_weight": 0.21749058216810227, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.88085021972654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4521057128906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23480, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.29931595325469973, + "epoch": 6.834448647075939, + "flow_matching_loss": 0.009924397664144635, + "grad_norm": 0.07809685915708542, + "kept_token_count": 573.36328125, + "learning_rate": 3.7003542820824347e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9936073541641235, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006392639083787799, + "mgd_loss_weight": 0.21715723723173141, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.34681396484376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4403045654297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23490, + "valid_token_count": 818.296875 + }, + { + "actual_token_mask_ratio": 0.3002253919839859, + "epoch": 6.8373581611870815, + "flow_matching_loss": 0.011321348510682583, + "grad_norm": 0.07247782498598099, + "kept_token_count": 572.5140625, + "learning_rate": 3.6894868417171252e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9938163161277771, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006183670181781053, + "mgd_loss_weight": 0.21682389825582504, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.9735137939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.0674163818359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23500, + "valid_token_count": 818.14296875 + }, + { + "actual_token_mask_ratio": 0.2996890515089035, + "epoch": 6.840267675298225, + "flow_matching_loss": 0.010632065776735544, + "grad_norm": 0.07652849704027176, + "kept_token_count": 572.9828125, + "learning_rate": 3.678633144691818e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9937900543212891, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006209940742701292, + "mgd_loss_weight": 0.2164905458688736, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.08494567871094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6115417480469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23510, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.29952956140041354, + "epoch": 6.843177189409369, + "flow_matching_loss": 0.01030676532536745, + "grad_norm": 0.07081158459186554, + "kept_token_count": 573.0390625, + "learning_rate": 3.667793204194776e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9937635719776153, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006236422760412097, + "mgd_loss_weight": 0.2161572054028511, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.5589904785156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.45226135253904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23520, + "valid_token_count": 818.0734375 + }, + { + "actual_token_mask_ratio": 0.29970757365226747, + "epoch": 6.846086703520512, + "flow_matching_loss": 0.008434869535267352, + "grad_norm": 0.05876701697707176, + "kept_token_count": 572.9546875, + "learning_rate": 3.656967033397544e-06, + "loss": 0.0098, + "mgd_cos_sim": 0.9935991644859314, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006400840450078249, + "mgd_loss_weight": 0.2158238634467125, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.0174591064453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4626892089844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23530, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.2998328059911728, + "epoch": 6.8489962176316554, + "flow_matching_loss": 0.01119143208488822, + "grad_norm": 0.10846209526062012, + "kept_token_count": 572.9328125, + "learning_rate": 3.646154645454948e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9937467098236084, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006253302888944745, + "mgd_loss_weight": 0.21549051254987717, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.4623596191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.24449768066404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23540, + "valid_token_count": 818.27734375 + }, + { + "actual_token_mask_ratio": 0.30088576674461365, + "epoch": 6.851905731742799, + "flow_matching_loss": 0.009956429526209832, + "grad_norm": 0.06311891227960587, + "kept_token_count": 572.18046875, + "learning_rate": 3.6353560535050524e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9938564002513885, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006143592623993754, + "mgd_loss_weight": 0.2151571735739708, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.56053466796874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4664306640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23550, + "valid_token_count": 818.44296875 + }, + { + "actual_token_mask_ratio": 0.2998735040426254, + "epoch": 6.854815245853942, + "flow_matching_loss": 0.009981692256405949, + "grad_norm": 0.09330127388238907, + "kept_token_count": 572.91796875, + "learning_rate": 3.624571270669172e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.993726646900177, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006273353938013315, + "mgd_loss_weight": 0.21482382863759994, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.0523406982422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.61102600097655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23560, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.29944262802600863, + "epoch": 6.857724759965086, + "flow_matching_loss": 0.010513351345434786, + "grad_norm": 0.07695411145687103, + "kept_token_count": 573.4328125, + "learning_rate": 3.6138003100518296e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.994109708070755, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00589029798284173, + "mgd_loss_weight": 0.21449048072099686, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.76198120117186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.95054931640624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23570, + "valid_token_count": 818.53359375 + }, + { + "actual_token_mask_ratio": 0.30092852413654325, + "epoch": 6.860634274076229, + "flow_matching_loss": 0.010092102456837893, + "grad_norm": 0.05570763722062111, + "kept_token_count": 572.02109375, + "learning_rate": 3.6030431847407595e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9939441084861755, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006055873818695545, + "mgd_loss_weight": 0.21415714174509048, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.2353240966797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.44166870117186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23580, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.2995576560497284, + "epoch": 6.863543788187373, + "flow_matching_loss": 0.011602898687124252, + "grad_norm": 0.08548872172832489, + "kept_token_count": 573.3078125, + "learning_rate": 3.592299907806881e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9938324689865112, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006167533248662949, + "mgd_loss_weight": 0.2138237923383713, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.75648803710936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.13018493652345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23590, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.3005735605955124, + "epoch": 6.866453302298516, + "flow_matching_loss": 0.00993008054792881, + "grad_norm": 0.06799062341451645, + "kept_token_count": 572.38203125, + "learning_rate": 3.5815704923042892e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9939103543758392, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00608964073471725, + "mgd_loss_weight": 0.21349044889211655, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.1619415283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.43349304199216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23600, + "valid_token_count": 818.36171875 + }, + { + "actual_token_mask_ratio": 0.2997605115175247, + "epoch": 6.86936281640966, + "flow_matching_loss": 0.010893193446099758, + "grad_norm": 0.08095502853393555, + "kept_token_count": 572.88671875, + "learning_rate": 3.5708549512702392e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9939061462879181, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006093855574727058, + "mgd_loss_weight": 0.21315710991621017, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.9446258544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.531982421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23610, + "valid_token_count": 818.13046875 + }, + { + "actual_token_mask_ratio": 0.3003492593765259, + "epoch": 6.872272330520803, + "flow_matching_loss": 0.010759471450001001, + "grad_norm": 0.06878205388784409, + "kept_token_count": 572.4859375, + "learning_rate": 3.5601532977251217e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9939298808574677, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006070140656083823, + "mgd_loss_weight": 0.2128237560391426, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.0723846435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.95978088378905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23620, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.30061989426612856, + "epoch": 6.875181844631946, + "flow_matching_loss": 0.009810247551649809, + "grad_norm": 0.09788589179515839, + "kept_token_count": 572.28984375, + "learning_rate": 3.5494655446724544e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9938031792640686, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006196814170107245, + "mgd_loss_weight": 0.21249041706323624, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.0078643798828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.80247497558594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23630, + "valid_token_count": 818.27578125 + }, + { + "actual_token_mask_ratio": 0.2996149480342865, + "epoch": 6.87809135874309, + "flow_matching_loss": 0.010902353376150132, + "grad_norm": 0.05758732557296753, + "kept_token_count": 573.08828125, + "learning_rate": 3.538791705098866e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9939930260181427, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006006986601278186, + "mgd_loss_weight": 0.2121570736169815, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.0000885009766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4305114746094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23640, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.2994659751653671, + "epoch": 6.881000872854234, + "flow_matching_loss": 0.010040164180099963, + "grad_norm": 0.07526575773954391, + "kept_token_count": 573.38125, + "learning_rate": 3.5281317919740718e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9938687682151794, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006131222797557711, + "mgd_loss_weight": 0.2118237242102623, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.65379943847654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5484161376953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23650, + "valid_token_count": 818.49296875 + }, + { + "actual_token_mask_ratio": 0.2990638703107834, + "epoch": 6.883910386965377, + "flow_matching_loss": 0.00880212914198637, + "grad_norm": 0.07153334468603134, + "kept_token_count": 573.39296875, + "learning_rate": 3.5174858182508825e-06, + "loss": 0.0101, + "mgd_cos_sim": 0.9940726339817048, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005927369277924299, + "mgd_loss_weight": 0.21149038523435593, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.3015625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2865753173828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23660, + "valid_token_count": 818.03671875 + }, + { + "actual_token_mask_ratio": 0.2995642751455307, + "epoch": 6.88681990107652, + "flow_matching_loss": 0.011503249919041992, + "grad_norm": 0.11202472448348999, + "kept_token_count": 573.434375, + "learning_rate": 3.506853796865153e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9937065601348877, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006293429434299469, + "mgd_loss_weight": 0.21115703880786896, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.1035125732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1356170654297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23670, + "valid_token_count": 818.67734375 + }, + { + "actual_token_mask_ratio": 0.29897006750106814, + "epoch": 6.889729415187664, + "flow_matching_loss": 0.010527052124962211, + "grad_norm": 0.07224710285663605, + "kept_token_count": 573.62578125, + "learning_rate": 3.496235740735789e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9938387334346771, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006161268707364798, + "mgd_loss_weight": 0.210823692381382, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.3292633056641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2423034667969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23680, + "valid_token_count": 818.2625 + }, + { + "actual_token_mask_ratio": 0.3002451151609421, + "epoch": 6.892638929298807, + "flow_matching_loss": 0.010616458719596267, + "grad_norm": 0.08315945416688919, + "kept_token_count": 572.48125, + "learning_rate": 3.485631662764738e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9941674709320069, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005832524178549648, + "mgd_loss_weight": 0.21049035340547562, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 434.5651489257813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8548126220703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23690, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.30008842051029205, + "epoch": 6.89554844340995, + "flow_matching_loss": 0.010430598584935068, + "grad_norm": 0.06893283873796463, + "kept_token_count": 572.80703125, + "learning_rate": 3.475041575836952e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9937375783920288, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062624157872051, + "mgd_loss_weight": 0.21015700250864028, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.72674560546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9159637451172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23700, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.3006395697593689, + "epoch": 6.898457957521094, + "flow_matching_loss": 0.009925680328160524, + "grad_norm": 0.0735948458313942, + "kept_token_count": 572.371875, + "learning_rate": 3.4644654928203796e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9939573109149933, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006042701099067927, + "mgd_loss_weight": 0.20982366055250168, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.0455291748047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.05979614257814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23710, + "valid_token_count": 818.4171875 + }, + { + "actual_token_mask_ratio": 0.3001430928707123, + "epoch": 6.901367471632238, + "flow_matching_loss": 0.009628144511952996, + "grad_norm": 0.05493362993001938, + "kept_token_count": 572.671875, + "learning_rate": 3.4539034265659663e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9940243065357208, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00597569583915174, + "mgd_loss_weight": 0.2094903200864792, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.335205078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7131042480469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23720, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.300633105635643, + "epoch": 6.904276985743381, + "flow_matching_loss": 0.010276600020006299, + "grad_norm": 0.08180596679449081, + "kept_token_count": 572.19296875, + "learning_rate": 3.4433553899076163e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9939506471157074, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006049334025010467, + "mgd_loss_weight": 0.20915696769952774, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.9816497802734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5763885498047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23730, + "valid_token_count": 818.15859375 + }, + { + "actual_token_mask_ratio": 0.2999173790216446, + "epoch": 6.907186499854524, + "flow_matching_loss": 0.010486825928092003, + "grad_norm": 0.07005970925092697, + "kept_token_count": 572.8015625, + "learning_rate": 3.432821395662189e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.993633508682251, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063665035180747506, + "mgd_loss_weight": 0.20882362872362137, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.52021484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4592346191406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23740, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.30001559257507326, + "epoch": 6.910096013965668, + "flow_matching_loss": 0.0109098840970546, + "grad_norm": 0.07388018816709518, + "kept_token_count": 572.7453125, + "learning_rate": 3.4223014566294785e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9937597215175629, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006240289704874158, + "mgd_loss_weight": 0.2084902837872505, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 433.21568908691404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1089263916016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23750, + "valid_token_count": 818.22421875 + }, + { + "actual_token_mask_ratio": 0.3000280290842056, + "epoch": 6.913005528076811, + "flow_matching_loss": 0.011815681401640177, + "grad_norm": 0.07870147377252579, + "kept_token_count": 572.88046875, + "learning_rate": 3.4117955855922e-06, + "loss": 0.0132, + "mgd_cos_sim": 0.9935566842556, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006443305825814605, + "mgd_loss_weight": 0.20815693587064743, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.0117614746094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2881744384766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23760, + "valid_token_count": 818.43203125 + }, + { + "actual_token_mask_ratio": 0.3004447132349014, + "epoch": 6.9159150421879545, + "flow_matching_loss": 0.010169494058936834, + "grad_norm": 0.05881081894040108, + "kept_token_count": 572.459375, + "learning_rate": 3.4013037953159855e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9936817824840546, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006318224966526032, + "mgd_loss_weight": 0.20782359689474106, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.7055236816406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4593231201172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23770, + "valid_token_count": 818.31796875 + }, + { + "actual_token_mask_ratio": 0.2994142770767212, + "epoch": 6.918824556299098, + "flow_matching_loss": 0.011342743597924709, + "grad_norm": 0.06955400854349136, + "kept_token_count": 573.375, + "learning_rate": 3.390826098549345e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9937654554843902, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062345522455871105, + "mgd_loss_weight": 0.20749024897813798, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.53289794921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8600128173828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23780, + "valid_token_count": 818.42109375 + }, + { + "actual_token_mask_ratio": 0.3004441767930984, + "epoch": 6.921734070410242, + "flow_matching_loss": 0.009463638346642256, + "grad_norm": 0.06145612150430679, + "kept_token_count": 572.5109375, + "learning_rate": 3.3803625080236684e-06, + "loss": 0.0108, + "mgd_cos_sim": 0.9935574769973755, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006442514713853598, + "mgd_loss_weight": 0.20715690404176712, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.7974487304688, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9573059082031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23790, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.3003924906253815, + "epoch": 6.924643584521385, + "flow_matching_loss": 0.013123873714357614, + "grad_norm": 0.145218163728714, + "kept_token_count": 572.45625, + "learning_rate": 3.369913036453205e-06, + "loss": 0.0144, + "mgd_cos_sim": 0.9938754975795746, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006124491803348064, + "mgd_loss_weight": 0.20682356506586075, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.30274963378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9345764160156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23800, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.29956925511360166, + "epoch": 6.927553098632528, + "flow_matching_loss": 0.010505446419119834, + "grad_norm": 0.08190453797578812, + "kept_token_count": 573.06328125, + "learning_rate": 3.3594776965350434e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9938901245594025, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006109871063381433, + "mgd_loss_weight": 0.2064902126789093, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.63426513671874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3627960205078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23810, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.300815150141716, + "epoch": 6.930462612743672, + "flow_matching_loss": 0.010912036476656794, + "grad_norm": 0.062424708157777786, + "kept_token_count": 572.12578125, + "learning_rate": 3.34905650094911e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.994060081243515, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059399050660431385, + "mgd_loss_weight": 0.2061568722128868, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.1191345214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1620178222656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23820, + "valid_token_count": 818.27421875 + }, + { + "actual_token_mask_ratio": 0.2999443680047989, + "epoch": 6.933372126854815, + "flow_matching_loss": 0.010128088295459747, + "grad_norm": 0.08117002248764038, + "kept_token_count": 572.77109375, + "learning_rate": 3.338649462358145e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9940034449100494, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059965604450553656, + "mgd_loss_weight": 0.2058235302567482, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.06715393066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3503143310547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23830, + "valid_token_count": 818.18125 + }, + { + "actual_token_mask_ratio": 0.2995548903942108, + "epoch": 6.936281640965959, + "flow_matching_loss": 0.010046202456578612, + "grad_norm": 0.05125366523861885, + "kept_token_count": 573.31875, + "learning_rate": 3.3282565934076765e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9936300575733185, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00636993432417512, + "mgd_loss_weight": 0.20549017935991287, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.2086242675781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.79583740234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23840, + "valid_token_count": 818.5078125 + }, + { + "actual_token_mask_ratio": 0.299196195602417, + "epoch": 6.939191155077102, + "flow_matching_loss": 0.010414724424481391, + "grad_norm": 0.11767544597387314, + "kept_token_count": 573.325, + "learning_rate": 3.31787790672602e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9937135696411132, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006286422489210963, + "mgd_loss_weight": 0.2051568403840065, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 402.8409454345703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.73951416015626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23850, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.30044967830181124, + "epoch": 6.942100669188245, + "flow_matching_loss": 0.00938415341079235, + "grad_norm": 0.0733611211180687, + "kept_token_count": 572.3484375, + "learning_rate": 3.307513414924255e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.9942104697227478, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005789525248110295, + "mgd_loss_weight": 0.20482349395751953, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.99114379882815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5528167724609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23860, + "valid_token_count": 818.1640625 + }, + { + "actual_token_mask_ratio": 0.2993336349725723, + "epoch": 6.945010183299389, + "flow_matching_loss": 0.010865889163687826, + "grad_norm": 0.05828622728586197, + "kept_token_count": 573.41015625, + "learning_rate": 3.2971631305962246e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.993586540222168, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006413454515859484, + "mgd_loss_weight": 0.20449014753103256, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.1388854980469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7203857421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23870, + "valid_token_count": 818.37265625 + }, + { + "actual_token_mask_ratio": 0.300106480717659, + "epoch": 6.947919697410533, + "flow_matching_loss": 0.009986452385783195, + "grad_norm": 0.05298667028546333, + "kept_token_count": 572.78125, + "learning_rate": 3.2868270663184964e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9939148306846619, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006085159769281745, + "mgd_loss_weight": 0.2041568085551262, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.0678741455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.009375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23880, + "valid_token_count": 818.38046875 + }, + { + "actual_token_mask_ratio": 0.2998227059841156, + "epoch": 6.950829211521675, + "flow_matching_loss": 0.011559566017240286, + "grad_norm": 0.07209689170122147, + "kept_token_count": 572.97421875, + "learning_rate": 3.276505234650362e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9939215123653412, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006078495085239411, + "mgd_loss_weight": 0.203823459148407, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.0172119140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0028533935547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23890, + "valid_token_count": 818.33046875 + }, + { + "actual_token_mask_ratio": 0.2994387954473495, + "epoch": 6.953738725632819, + "flow_matching_loss": 0.010494278185069561, + "grad_norm": 0.06627198308706284, + "kept_token_count": 573.165625, + "learning_rate": 3.266197648133821e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9938389301300049, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006161070009693503, + "mgd_loss_weight": 0.20349011570215225, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.3897735595703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2409210205078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23900, + "valid_token_count": 818.15546875 + }, + { + "actual_token_mask_ratio": 0.3002262681722641, + "epoch": 6.956648239743963, + "flow_matching_loss": 0.010500900680199265, + "grad_norm": 0.08982692658901215, + "kept_token_count": 572.528125, + "learning_rate": 3.255904319293562e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9939850032329559, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006014993786811829, + "mgd_loss_weight": 0.20315677672624588, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.18504333496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8726287841797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23910, + "valid_token_count": 818.1578125 + }, + { + "actual_token_mask_ratio": 0.30048651099205015, + "epoch": 6.9595577538551066, + "flow_matching_loss": 0.009693872462958098, + "grad_norm": 0.059800323098897934, + "kept_token_count": 572.4546875, + "learning_rate": 3.2456252606369485e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9940357565879822, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005964235542342067, + "mgd_loss_weight": 0.20282342284917831, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.6836730957031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.96226196289064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23920, + "valid_token_count": 818.36640625 + }, + { + "actual_token_mask_ratio": 0.30035645663738253, + "epoch": 6.962467267966249, + "flow_matching_loss": 0.009448980493471027, + "grad_norm": 0.1204603984951973, + "kept_token_count": 572.5, + "learning_rate": 3.235360484654011e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.9940876364707947, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059123602230101826, + "mgd_loss_weight": 0.20249008387327194, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.381787109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.77093505859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23930, + "valid_token_count": 818.27265625 + }, + { + "actual_token_mask_ratio": 0.3005271881818771, + "epoch": 6.965376782077393, + "flow_matching_loss": 0.0106589344330132, + "grad_norm": 0.061076171696186066, + "kept_token_count": 572.61171875, + "learning_rate": 3.225110003817418e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9940981686115264, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00590182370506227, + "mgd_loss_weight": 0.2021567404270172, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.9608551025391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.97962951660156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23940, + "valid_token_count": 818.63515625 + }, + { + "actual_token_mask_ratio": 0.29964175522327424, + "epoch": 6.968286296188537, + "flow_matching_loss": 0.011155073903501034, + "grad_norm": 0.10257071256637573, + "kept_token_count": 573.01171875, + "learning_rate": 3.214873830582467e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.994016021490097, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005983985587954521, + "mgd_loss_weight": 0.201823391020298, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.39173278808596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.68283996582034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23950, + "valid_token_count": 818.16640625 + }, + { + "actual_token_mask_ratio": 0.3005324065685272, + "epoch": 6.97119581029968, + "flow_matching_loss": 0.010565112996846437, + "grad_norm": 0.06613349169492722, + "kept_token_count": 572.35, + "learning_rate": 3.2046519773870803e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9938418388366699, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006158156460151077, + "mgd_loss_weight": 0.20149005204439163, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.0310882568359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.42473754882815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23960, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.2999217748641968, + "epoch": 6.974105324410823, + "flow_matching_loss": 0.01090946514159441, + "grad_norm": 0.07045379281044006, + "kept_token_count": 572.78125, + "learning_rate": 3.1944444566517654e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9938964128494263, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061035874299705025, + "mgd_loss_weight": 0.20115670412778855, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.60264587402344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4886199951172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23970, + "valid_token_count": 818.16640625 + }, + { + "actual_token_mask_ratio": 0.29987032413482667, + "epoch": 6.977014838521967, + "flow_matching_loss": 0.011028059013187885, + "grad_norm": 0.0774400606751442, + "kept_token_count": 572.91484375, + "learning_rate": 3.18425128077963e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9939284324645996, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006071565626189113, + "mgd_loss_weight": 0.2008233591914177, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.5353698730469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.09658203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23980, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.29980718791484834, + "epoch": 6.979924352633111, + "flow_matching_loss": 0.011287278681993484, + "grad_norm": 0.09535343199968338, + "kept_token_count": 573.0796875, + "learning_rate": 3.174072462156342e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9938762843608856, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00612372444011271, + "mgd_loss_weight": 0.20049002021551132, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.82347412109374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.01402282714844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 23990, + "valid_token_count": 818.45859375 + }, + { + "actual_token_mask_ratio": 0.2995390176773071, + "epoch": 6.9828338667442535, + "flow_matching_loss": 0.009231869364157319, + "grad_norm": 0.07870370149612427, + "kept_token_count": 573.190625, + "learning_rate": 3.163908013150124e-06, + "loss": 0.0105, + "mgd_cos_sim": 0.9938686072826386, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006131390109658241, + "mgd_loss_weight": 0.20015666931867598, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.7649780273438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8011444091797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24000, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.30005858838558197, + "epoch": 6.985743380855397, + "flow_matching_loss": 0.009833789756521582, + "grad_norm": 0.05648857355117798, + "kept_token_count": 572.85546875, + "learning_rate": 3.153757946111741e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9938489079475403, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061510919593274595, + "mgd_loss_weight": 0.19982332736253738, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.02345886230466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7060272216797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24010, + "valid_token_count": 818.43359375 + }, + { + "actual_token_mask_ratio": 0.2995370477437973, + "epoch": 6.988652894966541, + "flow_matching_loss": 0.009756442904472352, + "grad_norm": 0.06536144018173218, + "kept_token_count": 573.13515625, + "learning_rate": 3.1436222733744776e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9939140617847443, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006085943058133125, + "mgd_loss_weight": 0.1994899868965149, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.3066375732422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.89967651367186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24020, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.3002933919429779, + "epoch": 6.991562409077684, + "flow_matching_loss": 0.009361500479280948, + "grad_norm": 0.06859784573316574, + "kept_token_count": 572.525, + "learning_rate": 3.133501007254138e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.9935925006866455, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006407497217878699, + "mgd_loss_weight": 0.19915663450956345, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.1678436279297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6676086425781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24030, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.3003568321466446, + "epoch": 6.9944719231888275, + "flow_matching_loss": 0.01093978676944971, + "grad_norm": 0.08274974673986435, + "kept_token_count": 572.5109375, + "learning_rate": 3.1233941600490123e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9939050197601318, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00609497451223433, + "mgd_loss_weight": 0.19882329553365707, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.57526245117185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.97035217285156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24040, + "valid_token_count": 818.29140625 + }, + { + "actual_token_mask_ratio": 0.2997754544019699, + "epoch": 6.997381437299971, + "flow_matching_loss": 0.010547151602804661, + "grad_norm": 0.08300282061100006, + "kept_token_count": 572.91015625, + "learning_rate": 3.113301744039875e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9937277734279633, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006272218329831958, + "mgd_loss_weight": 0.19848995059728622, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.9013702392578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.42209777832034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24050, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.29921780824661254, + "epoch": 7.000290951411114, + "flow_matching_loss": 0.010306518943980335, + "grad_norm": 0.06735968589782715, + "kept_token_count": 573.4323059082031, + "learning_rate": 3.103223771489959e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9942112624645233, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005788729107007384, + "mgd_loss_weight": 0.19815660268068314, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.809423828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.29651794433596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24060, + "valid_token_count": 818.2808776855469 + }, + { + "actual_token_mask_ratio": 0.30017809867858886, + "epoch": 7.003200465522258, + "flow_matching_loss": 0.009137410530820489, + "grad_norm": 0.07865437865257263, + "kept_token_count": 572.6234375, + "learning_rate": 3.09316025464495e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9937959015369415, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006204100139439106, + "mgd_loss_weight": 0.19782326370477676, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.2369781494141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.34673767089845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24070, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.29941248893737793, + "epoch": 7.006109979633401, + "flow_matching_loss": 0.010623449739068747, + "grad_norm": 0.06767246872186661, + "kept_token_count": 573.18671875, + "learning_rate": 3.0831112057329775e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9940128922462463, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005987110454589129, + "mgd_loss_weight": 0.19748991429805757, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.29962768554685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1951019287109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24080, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.3001786321401596, + "epoch": 7.009019493744544, + "flow_matching_loss": 0.010606492636725307, + "grad_norm": 0.06796804815530777, + "kept_token_count": 572.571875, + "learning_rate": 3.073076636964579e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9938719153404236, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006128078326582909, + "mgd_loss_weight": 0.19715657085180283, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.49033508300784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.4793182373047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24090, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.30010777711868286, + "epoch": 7.011929007855688, + "flow_matching_loss": 0.009699534252285957, + "grad_norm": 0.06819725036621094, + "kept_token_count": 572.7359375, + "learning_rate": 3.063056560532698e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.993979948759079, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006020038994029165, + "mgd_loss_weight": 0.19682323187589645, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.15655822753905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5776794433594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24100, + "valid_token_count": 818.32109375 + }, + { + "actual_token_mask_ratio": 0.30012241899967196, + "epoch": 7.014838521966832, + "flow_matching_loss": 0.00969609091989696, + "grad_norm": 0.06777489930391312, + "kept_token_count": 572.77265625, + "learning_rate": 3.0530509886126807e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9938312709331513, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006168735353276134, + "mgd_loss_weight": 0.196489879488945, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.03110961914064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.62542724609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24110, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.29934796392917634, + "epoch": 7.017748036077975, + "flow_matching_loss": 0.0095413435716182, + "grad_norm": 0.058981504291296005, + "kept_token_count": 573.1328125, + "learning_rate": 3.043059933362236e-06, + "loss": 0.0108, + "mgd_cos_sim": 0.9938198983669281, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006180097348988056, + "mgd_loss_weight": 0.19615653902292252, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.44334716796874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.83112182617185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24120, + "valid_token_count": 818.0015625 + }, + { + "actual_token_mask_ratio": 0.30047766864299774, + "epoch": 7.020657550189118, + "flow_matching_loss": 0.01076448941603303, + "grad_norm": 0.08672698587179184, + "kept_token_count": 572.36953125, + "learning_rate": 3.0330834069214337e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9940481305122375, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005951880011707544, + "mgd_loss_weight": 0.19582319706678392, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.92552490234374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.05381164550784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24130, + "valid_token_count": 818.2328125 + }, + { + "actual_token_mask_ratio": 0.3003435075283051, + "epoch": 7.023567064300262, + "flow_matching_loss": 0.010638819402083754, + "grad_norm": 0.09121520072221756, + "kept_token_count": 572.4, + "learning_rate": 3.023121421412703e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9936845242977143, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006315472535789013, + "mgd_loss_weight": 0.19548984616994858, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.08098754882815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0339019775391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24140, + "valid_token_count": 818.11171875 + }, + { + "actual_token_mask_ratio": 0.2999905586242676, + "epoch": 7.026476578411406, + "flow_matching_loss": 0.010727101983502508, + "grad_norm": 0.06961533427238464, + "kept_token_count": 572.8421875, + "learning_rate": 3.0131739889407896e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9938876211643219, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006112377298995853, + "mgd_loss_weight": 0.1951565071940422, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.1267120361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.36731872558596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24150, + "valid_token_count": 818.33515625 + }, + { + "actual_token_mask_ratio": 0.2996636241674423, + "epoch": 7.029386092522548, + "flow_matching_loss": 0.010659833066165447, + "grad_norm": 0.07066170126199722, + "kept_token_count": 573.0359375, + "learning_rate": 3.0032411215927645e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.994271194934845, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005728800548240542, + "mgd_loss_weight": 0.19482316076755524, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.913427734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.88433837890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24160, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.2996632128953934, + "epoch": 7.032295606633692, + "flow_matching_loss": 0.01030608480796218, + "grad_norm": 0.06606818735599518, + "kept_token_count": 572.9046875, + "learning_rate": 2.9933228314379975e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9940875947475434, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005912390211597085, + "mgd_loss_weight": 0.19448981434106827, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.279736328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4781555175781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24170, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.30024186968803407, + "epoch": 7.035205120744836, + "flow_matching_loss": 0.009374928940087558, + "grad_norm": 0.0686914399266243, + "kept_token_count": 572.45390625, + "learning_rate": 2.9834191305281405e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.993927538394928, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00607245652936399, + "mgd_loss_weight": 0.1941564753651619, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.2826385498047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.25323791503905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24180, + "valid_token_count": 818.07578125 + }, + { + "actual_token_mask_ratio": 0.3000962227582932, + "epoch": 7.038114634855979, + "flow_matching_loss": 0.010561170335859061, + "grad_norm": 0.06682723760604858, + "kept_token_count": 572.571875, + "learning_rate": 2.973530030897135e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9936442136764526, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006355798291042447, + "mgd_loss_weight": 0.19382312446832656, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.6646209716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8724700927734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24190, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.29944474101066587, + "epoch": 7.041024148967122, + "flow_matching_loss": 0.009406709484755993, + "grad_norm": 0.08270436525344849, + "kept_token_count": 573.26796875, + "learning_rate": 2.9636555445611646e-06, + "loss": 0.0105, + "mgd_cos_sim": 0.9941189885139465, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0058810147922486065, + "mgd_loss_weight": 0.19348978251218796, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.9987274169922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7099548339844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24200, + "valid_token_count": 818.30390625 + }, + { + "actual_token_mask_ratio": 0.30019454658031464, + "epoch": 7.043933663078266, + "flow_matching_loss": 0.010004780581220984, + "grad_norm": 0.0877746194601059, + "kept_token_count": 572.63203125, + "learning_rate": 2.953795683518659e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9938171684741974, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006182843260467053, + "mgd_loss_weight": 0.19315644353628159, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.51292724609374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3883453369141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24210, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.3003873735666275, + "epoch": 7.04684317718941, + "flow_matching_loss": 0.010430366639047861, + "grad_norm": 0.0910266563296318, + "kept_token_count": 572.5171875, + "learning_rate": 2.9439504597502815e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9936277806758881, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006372218718752265, + "mgd_loss_weight": 0.19282308965921402, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.1192657470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0367431640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24220, + "valid_token_count": 818.33515625 + }, + { + "actual_token_mask_ratio": 0.29942726790905, + "epoch": 7.049752691300553, + "flow_matching_loss": 0.011036988766863941, + "grad_norm": 0.061375249177217484, + "kept_token_count": 573.2828125, + "learning_rate": 2.934119885218902e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9941505789756775, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005849415529519319, + "mgd_loss_weight": 0.19248975068330765, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.9949890136719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0669403076172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24230, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.3002824693918228, + "epoch": 7.052662205411696, + "flow_matching_loss": 0.00898924279026687, + "grad_norm": 0.08169211447238922, + "kept_token_count": 572.7421875, + "learning_rate": 2.9243039718696045e-06, + "loss": 0.0101, + "mgd_cos_sim": 0.9940220832824707, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005977913970127702, + "mgd_loss_weight": 0.1921564072370529, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.6569885253906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2408935546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24240, + "valid_token_count": 818.53125 + }, + { + "actual_token_mask_ratio": 0.29957868456840514, + "epoch": 7.05557171952284, + "flow_matching_loss": 0.009176828758791089, + "grad_norm": 0.057242389768362045, + "kept_token_count": 573.13203125, + "learning_rate": 2.914502731629641e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9940340578556061, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005965940793976188, + "mgd_loss_weight": 0.1918230578303337, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.492578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8449279785156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24250, + "valid_token_count": 818.26640625 + }, + { + "actual_token_mask_ratio": 0.30006087720394137, + "epoch": 7.058481233633983, + "flow_matching_loss": 0.011449975706636906, + "grad_norm": 0.07443539798259735, + "kept_token_count": 572.73828125, + "learning_rate": 2.904716176408451e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.994164091348648, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005835899850353599, + "mgd_loss_weight": 0.19148971885442734, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.8772277832031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1268676757812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24260, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.2998062551021576, + "epoch": 7.0613907477451265, + "flow_matching_loss": 0.010300432564690709, + "grad_norm": 0.06334422528743744, + "kept_token_count": 572.85625, + "learning_rate": 2.8949443180976172e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9937840580940247, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006215952709317208, + "mgd_loss_weight": 0.19115637093782425, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 430.52208251953124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9107177734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24270, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.30075536072254183, + "epoch": 7.06430026185627, + "flow_matching_loss": 0.010271114995703101, + "grad_norm": 0.06826535612344742, + "kept_token_count": 572.27109375, + "learning_rate": 2.885187168570864e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9936893105506897, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006310682836920023, + "mgd_loss_weight": 0.1908230260014534, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.87738342285155, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.33912353515626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24280, + "valid_token_count": 818.41953125 + }, + { + "actual_token_mask_ratio": 0.299398797750473, + "epoch": 7.067209775967413, + "flow_matching_loss": 0.011757140094414354, + "grad_norm": 0.08377774804830551, + "kept_token_count": 573.246875, + "learning_rate": 2.875444739684058e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9937946438789368, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006205355376005172, + "mgd_loss_weight": 0.19048968702554703, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.5521606445312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7262237548828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24290, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.29912079870700836, + "epoch": 7.070119290078557, + "flow_matching_loss": 0.00978102656081319, + "grad_norm": 0.08599833399057388, + "kept_token_count": 573.35, + "learning_rate": 2.865717043275165e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9939346432685852, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006065342715010047, + "mgd_loss_weight": 0.19015633463859558, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.5931732177734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2539459228516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24300, + "valid_token_count": 818.0515625 + }, + { + "actual_token_mask_ratio": 0.300167578458786, + "epoch": 7.0730288041897005, + "flow_matching_loss": 0.011448138579726218, + "grad_norm": 0.10047376155853271, + "kept_token_count": 572.790625, + "learning_rate": 2.8560040911642528e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.9937742471694946, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00622576242312789, + "mgd_loss_weight": 0.1898229941725731, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.7358154296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.00787658691405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24310, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.30071853995323183, + "epoch": 7.075938318300844, + "flow_matching_loss": 0.01063527651131153, + "grad_norm": 0.07136092334985733, + "kept_token_count": 572.25078125, + "learning_rate": 2.846305895153472e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.99382763504982, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006172374403104186, + "mgd_loss_weight": 0.1894896522164345, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.3686065673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.07468566894534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24320, + "valid_token_count": 818.34453125 + }, + { + "actual_token_mask_ratio": 0.3004727870225906, + "epoch": 7.078847832411987, + "flow_matching_loss": 0.011577366665005685, + "grad_norm": 0.09840402752161026, + "kept_token_count": 572.4546875, + "learning_rate": 2.8366224670270487e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9938009083271027, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061990886926651, + "mgd_loss_weight": 0.18915630131959915, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.6239929199219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.55788269042966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24330, + "valid_token_count": 818.34609375 + }, + { + "actual_token_mask_ratio": 0.29986165165901185, + "epoch": 7.081757346523131, + "flow_matching_loss": 0.008988401899114252, + "grad_norm": 0.08714614063501358, + "kept_token_count": 573.028125, + "learning_rate": 2.8269538185512535e-06, + "loss": 0.01, + "mgd_cos_sim": 0.9944125235080719, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005587476631626486, + "mgd_loss_weight": 0.18882296234369278, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.6567321777344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.102880859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24340, + "valid_token_count": 818.4484375 + }, + { + "actual_token_mask_ratio": 0.3001795560121536, + "epoch": 7.084666860634274, + "flow_matching_loss": 0.00991825102828443, + "grad_norm": 0.09774893522262573, + "kept_token_count": 572.6953125, + "learning_rate": 2.817299961474417e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9940272390842437, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059727620333433155, + "mgd_loss_weight": 0.18848961740732192, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.43681640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6828186035156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24350, + "valid_token_count": 818.35 + }, + { + "actual_token_mask_ratio": 0.2998412877321243, + "epoch": 7.087576374745417, + "flow_matching_loss": 0.010350798815488815, + "grad_norm": 0.07378052920103073, + "kept_token_count": 573.078125, + "learning_rate": 2.8076609075268816e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9941838145256042, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005816195718944072, + "mgd_loss_weight": 0.18815626949071884, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.22198181152345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.346533203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24360, + "valid_token_count": 818.49609375 + }, + { + "actual_token_mask_ratio": 0.30003734230995177, + "epoch": 7.090485888856561, + "flow_matching_loss": 0.011575150862336158, + "grad_norm": 0.07653500884771347, + "kept_token_count": 572.784375, + "learning_rate": 2.798036668421006e-06, + "loss": 0.0128, + "mgd_cos_sim": 0.9934482395648956, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006551765976473689, + "mgd_loss_weight": 0.18782293051481247, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.6279693603516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0097900390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24370, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.29952715933322904, + "epoch": 7.093395402967705, + "flow_matching_loss": 0.00977907688356936, + "grad_norm": 0.07548600435256958, + "kept_token_count": 573.25390625, + "learning_rate": 2.7884272558511433e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9939639031887054, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060360911302268505, + "mgd_loss_weight": 0.18748958110809327, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.9505249023438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9182983398438, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24380, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.2993149757385254, + "epoch": 7.0963049170788475, + "flow_matching_loss": 0.010617093741893768, + "grad_norm": 0.06030399724841118, + "kept_token_count": 573.2359375, + "learning_rate": 2.7788326814936456e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9939152061939239, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006084795016795397, + "mgd_loss_weight": 0.18715623766183853, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.4385589599609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.58559265136716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24390, + "valid_token_count": 818.10625 + }, + { + "actual_token_mask_ratio": 0.2995823472738266, + "epoch": 7.099214431189991, + "flow_matching_loss": 0.011512633878737688, + "grad_norm": 0.06243067979812622, + "kept_token_count": 573.0765625, + "learning_rate": 2.7692529570068176e-06, + "loss": 0.0126, + "mgd_cos_sim": 0.994217723608017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005782274436205626, + "mgd_loss_weight": 0.18682289868593216, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.4587707519531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9646820068359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24400, + "valid_token_count": 818.19140625 + }, + { + "actual_token_mask_ratio": 0.29990138709545133, + "epoch": 7.102123945301135, + "flow_matching_loss": 0.011813239380717277, + "grad_norm": 0.07704108953475952, + "kept_token_count": 572.95078125, + "learning_rate": 2.759688094030933e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9937901198863983, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062099003698676825, + "mgd_loss_weight": 0.1864895448088646, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.17530212402346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.47149353027345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24410, + "valid_token_count": 818.38125 + }, + { + "actual_token_mask_ratio": 0.3008807361125946, + "epoch": 7.105033459412278, + "flow_matching_loss": 0.010545047372579575, + "grad_norm": 0.07966019958257675, + "kept_token_count": 572.08984375, + "learning_rate": 2.750138104188201e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9941664814949036, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005833513988181949, + "mgd_loss_weight": 0.18615620583295822, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.99449157714844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7836853027344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24420, + "valid_token_count": 818.30234375 + }, + { + "actual_token_mask_ratio": 0.29982084035873413, + "epoch": 7.107942973523421, + "flow_matching_loss": 0.00946962391026318, + "grad_norm": 0.06554200500249863, + "kept_token_count": 573.034375, + "learning_rate": 2.740602999082759e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.9940774440765381, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059225557837635275, + "mgd_loss_weight": 0.18582286238670348, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.32957763671874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0974182128906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24430, + "valid_token_count": 818.41484375 + }, + { + "actual_token_mask_ratio": 0.3008363902568817, + "epoch": 7.110852487634565, + "flow_matching_loss": 0.010008963290601969, + "grad_norm": 0.05868677794933319, + "kept_token_count": 572.29375, + "learning_rate": 2.731082790300654e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9942860126495361, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0057139988522976635, + "mgd_loss_weight": 0.18548951297998428, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.4645080566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.08362731933596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24440, + "valid_token_count": 818.5421875 + }, + { + "actual_token_mask_ratio": 0.3006715714931488, + "epoch": 7.113762001745709, + "flow_matching_loss": 0.010410834802314639, + "grad_norm": 0.06745339184999466, + "kept_token_count": 572.3109375, + "learning_rate": 2.7215774894098474e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9938538372516632, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061461441218853, + "mgd_loss_weight": 0.1851561740040779, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.08525390625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.03875122070315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24450, + "valid_token_count": 818.37421875 + }, + { + "actual_token_mask_ratio": 0.29989257752895354, + "epoch": 7.116671515856852, + "flow_matching_loss": 0.008698369190096856, + "grad_norm": 0.05955169349908829, + "kept_token_count": 572.98671875, + "learning_rate": 2.7120871079601696e-06, + "loss": 0.0098, + "mgd_cos_sim": 0.9940457701683044, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005954226385802031, + "mgd_loss_weight": 0.18482282757759094, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.0051788330078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.74400024414064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24460, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.2999306291341782, + "epoch": 7.119581029967995, + "flow_matching_loss": 0.010365764517337084, + "grad_norm": 0.0695519745349884, + "kept_token_count": 572.8296875, + "learning_rate": 2.7026116574833314e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9937572836875915, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006242701504379511, + "mgd_loss_weight": 0.18448948115110397, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.09044189453124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5741363525391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24470, + "valid_token_count": 818.24609375 + }, + { + "actual_token_mask_ratio": 0.2998531937599182, + "epoch": 7.122490544079139, + "flow_matching_loss": 0.010006502969190478, + "grad_norm": 0.07712678611278534, + "kept_token_count": 572.98671875, + "learning_rate": 2.6931511494928953e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9937635004520416, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006236500665545463, + "mgd_loss_weight": 0.1841561421751976, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.3561584472656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8475799560547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24480, + "valid_token_count": 818.38203125 + }, + { + "actual_token_mask_ratio": 0.3002884268760681, + "epoch": 7.125400058190282, + "flow_matching_loss": 0.010220870189368726, + "grad_norm": 0.06948069483041763, + "kept_token_count": 572.6625, + "learning_rate": 2.683705595484275e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9937787592411041, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006221228139474988, + "mgd_loss_weight": 0.18382279127836226, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4117004394531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6109985351562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24490, + "valid_token_count": 818.42578125 + }, + { + "actual_token_mask_ratio": 0.29994182884693144, + "epoch": 7.128309572301426, + "flow_matching_loss": 0.010914943506941199, + "grad_norm": 0.06918906420469284, + "kept_token_count": 572.75, + "learning_rate": 2.6742750069347014e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9937355875968933, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006264408258721233, + "mgd_loss_weight": 0.18348944932222366, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.0313446044922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.154736328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24500, + "valid_token_count": 818.146875 + }, + { + "actual_token_mask_ratio": 0.3001537352800369, + "epoch": 7.131219086412569, + "flow_matching_loss": 0.011004356713965536, + "grad_norm": 0.06598277390003204, + "kept_token_count": 572.77734375, + "learning_rate": 2.664859395303239e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9937586545944214, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006241340562701226, + "mgd_loss_weight": 0.18315610885620118, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.8119842529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3727325439453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24510, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.3006025433540344, + "epoch": 7.134128600523712, + "flow_matching_loss": 0.012344497302547098, + "grad_norm": 0.15189288556575775, + "kept_token_count": 572.34609375, + "learning_rate": 2.655458772030739e-06, + "loss": 0.0134, + "mgd_cos_sim": 0.994124811887741, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0058751905802637335, + "mgd_loss_weight": 0.18282275646924973, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.2841857910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.38963012695314, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24520, + "valid_token_count": 818.3359375 + }, + { + "actual_token_mask_ratio": 0.2992671072483063, + "epoch": 7.137038114634856, + "flow_matching_loss": 0.009562257677316666, + "grad_norm": 0.07450947165489197, + "kept_token_count": 573.33359375, + "learning_rate": 2.6460731485398403e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9940083444118499, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005991652142256498, + "mgd_loss_weight": 0.18248941749334335, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.3824768066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.68573303222655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24530, + "valid_token_count": 818.1890625 + }, + { + "actual_token_mask_ratio": 0.2998883605003357, + "epoch": 7.1399476287459995, + "flow_matching_loss": 0.010819588555023074, + "grad_norm": 0.09032542258501053, + "kept_token_count": 572.85546875, + "learning_rate": 2.6367025362349695e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9935226023197175, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006477399123832583, + "mgd_loss_weight": 0.1821560725569725, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.0850036621094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.85144958496096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24540, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.29948111474514005, + "epoch": 7.142857142857143, + "flow_matching_loss": 0.01018754243850708, + "grad_norm": 0.058466773480176926, + "kept_token_count": 573.175, + "learning_rate": 2.6273469465022965e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9938177525997162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006182236224412918, + "mgd_loss_weight": 0.18182272464036942, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.17629089355466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 383.2648132324219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24550, + "valid_token_count": 818.2125 + }, + { + "actual_token_mask_ratio": 0.3001192182302475, + "epoch": 7.145766656968286, + "flow_matching_loss": 0.010363072436302901, + "grad_norm": 0.07063386589288712, + "kept_token_count": 572.96171875, + "learning_rate": 2.618006390709753e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9939553380012512, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006044662650674581, + "mgd_loss_weight": 0.18148938566446304, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.8330474853516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24560, + "valid_token_count": 818.65859375 + }, + { + "actual_token_mask_ratio": 0.2997064530849457, + "epoch": 7.14867617107943, + "flow_matching_loss": 0.009466110123321414, + "grad_norm": 0.05871272832155228, + "kept_token_count": 572.9984375, + "learning_rate": 2.60868088020699e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.9939077198505402, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006092286016792059, + "mgd_loss_weight": 0.18115603774785996, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.7744598388672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.33643493652346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24570, + "valid_token_count": 818.22421875 + }, + { + "actual_token_mask_ratio": 0.299780997633934, + "epoch": 7.1515856851905735, + "flow_matching_loss": 0.009968789620324969, + "grad_norm": 0.08398637175559998, + "kept_token_count": 572.99765625, + "learning_rate": 2.599370426325382e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9938896775245667, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006110324105247855, + "mgd_loss_weight": 0.1808226928114891, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.843505859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5264038085937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24580, + "valid_token_count": 818.30703125 + }, + { + "actual_token_mask_ratio": 0.2994183599948883, + "epoch": 7.154495199301716, + "flow_matching_loss": 0.011385040218010544, + "grad_norm": 0.07220252603292465, + "kept_token_count": 573.25390625, + "learning_rate": 2.5900750403780093e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9938512325286866, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006148766214028001, + "mgd_loss_weight": 0.18048935383558273, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.21373596191404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8886291503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24590, + "valid_token_count": 818.25625 + }, + { + "actual_token_mask_ratio": 0.2994257062673569, + "epoch": 7.15740471341286, + "flow_matching_loss": 0.009987697610631585, + "grad_norm": 0.05933457985520363, + "kept_token_count": 573.346875, + "learning_rate": 2.580794733659639e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9936738193035126, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00632618092931807, + "mgd_loss_weight": 0.18015600144863128, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.1599609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.89931335449216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24600, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.29973631501197817, + "epoch": 7.160314227524004, + "flow_matching_loss": 0.011449121823534369, + "grad_norm": 0.06469796597957611, + "kept_token_count": 573.01328125, + "learning_rate": 2.5715295174467236e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9940194070339203, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005980603769421578, + "mgd_loss_weight": 0.1798226609826088, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.49757995605466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.67218627929685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24610, + "valid_token_count": 818.27890625 + }, + { + "actual_token_mask_ratio": 0.29946697056293486, + "epoch": 7.1632237416351465, + "flow_matching_loss": 0.009324452374130487, + "grad_norm": 0.06123390793800354, + "kept_token_count": 573.26328125, + "learning_rate": 2.5622794029973746e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9941459953784942, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00585400746203959, + "mgd_loss_weight": 0.1794893190264702, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.0330871582031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.70684814453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24620, + "valid_token_count": 818.31875 + }, + { + "actual_token_mask_ratio": 0.2994764238595963, + "epoch": 7.16613325574629, + "flow_matching_loss": 0.010784535761922598, + "grad_norm": 0.06681092828512192, + "kept_token_count": 573.2796875, + "learning_rate": 2.553044401551351e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9941245436668396, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00587546587921679, + "mgd_loss_weight": 0.17915596812963486, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.2399871826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1236053466797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24630, + "valid_token_count": 818.36015625 + }, + { + "actual_token_mask_ratio": 0.3000976026058197, + "epoch": 7.169042769857434, + "flow_matching_loss": 0.011347768735140562, + "grad_norm": 0.06604595482349396, + "kept_token_count": 572.7046875, + "learning_rate": 2.54382452433005e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.993628978729248, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006371026718989014, + "mgd_loss_weight": 0.17882262915372849, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.84166564941404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8834747314453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24640, + "valid_token_count": 818.26171875 + }, + { + "actual_token_mask_ratio": 0.30063918232917786, + "epoch": 7.171952283968578, + "flow_matching_loss": 0.0098592069465667, + "grad_norm": 0.08246071636676788, + "kept_token_count": 572.23203125, + "learning_rate": 2.53461978253649e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9937786102294922, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006221382785588503, + "mgd_loss_weight": 0.17848928272724152, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.3864349365234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.46326599121096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24650, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.29952490925788877, + "epoch": 7.1748617980797205, + "flow_matching_loss": 0.0097980045247823, + "grad_norm": 0.07871396839618683, + "kept_token_count": 573.259375, + "learning_rate": 2.5254301873553065e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9937903225421906, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00620967810973525, + "mgd_loss_weight": 0.17815593630075455, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.2665771484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2083709716797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24660, + "valid_token_count": 818.3875 + }, + { + "actual_token_mask_ratio": 0.2999080061912537, + "epoch": 7.177771312190864, + "flow_matching_loss": 0.00997355142608285, + "grad_norm": 0.06724995374679565, + "kept_token_count": 572.91015625, + "learning_rate": 2.5162557499527184e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9939114630222321, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006088545126840472, + "mgd_loss_weight": 0.17782259732484818, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.6507141113281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.82424011230466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24670, + "valid_token_count": 818.33671875 + }, + { + "actual_token_mask_ratio": 0.29956655502319335, + "epoch": 7.180680826302008, + "flow_matching_loss": 0.01018137950450182, + "grad_norm": 0.1172712966799736, + "kept_token_count": 573.13046875, + "learning_rate": 2.5070964814765335e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9939671993255615, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006032785400748253, + "mgd_loss_weight": 0.17748924791812898, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 431.6782989501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6700714111328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24680, + "valid_token_count": 818.25390625 + }, + { + "actual_token_mask_ratio": 0.30002548098564147, + "epoch": 7.183590340413151, + "flow_matching_loss": 0.010586923593655229, + "grad_norm": 0.07084252685308456, + "kept_token_count": 572.69765625, + "learning_rate": 2.4979523930561295e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9939913511276245, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006008647568523884, + "mgd_loss_weight": 0.17715590447187424, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.7768890380859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6864318847656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24690, + "valid_token_count": 818.16484375 + }, + { + "actual_token_mask_ratio": 0.29941657185554504, + "epoch": 7.186499854524294, + "flow_matching_loss": 0.008861614065244794, + "grad_norm": 0.0672004446387291, + "kept_token_count": 573.2640625, + "learning_rate": 2.4888234958024333e-06, + "loss": 0.0099, + "mgd_cos_sim": 0.9943342506885529, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005665749916806817, + "mgd_loss_weight": 0.17682256549596786, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.782470703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1709014892578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24700, + "valid_token_count": 818.26640625 + }, + { + "actual_token_mask_ratio": 0.3004055470228195, + "epoch": 7.189409368635438, + "flow_matching_loss": 0.011025181272998452, + "grad_norm": 0.06452473998069763, + "kept_token_count": 572.40390625, + "learning_rate": 2.479709800807915e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9935604631900787, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006439553946256638, + "mgd_loss_weight": 0.1764892116189003, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.5727996826172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3720764160156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24710, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.29925583600997924, + "epoch": 7.192318882746581, + "flow_matching_loss": 0.010523526696488261, + "grad_norm": 0.0814247876405716, + "kept_token_count": 573.38828125, + "learning_rate": 2.4706113191465767e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9938771247863769, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006122865574434399, + "mgd_loss_weight": 0.17615587264299393, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.6311279296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5614013671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24720, + "valid_token_count": 818.25234375 + }, + { + "actual_token_mask_ratio": 0.3001029431819916, + "epoch": 7.195228396857725, + "flow_matching_loss": 0.011013155290856957, + "grad_norm": 0.08972405642271042, + "kept_token_count": 572.82578125, + "learning_rate": 2.46152806187393e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9939618349075318, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006038142880424857, + "mgd_loss_weight": 0.17582252919673919, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.66001281738284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5279846191406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24730, + "valid_token_count": 818.4453125 + }, + { + "actual_token_mask_ratio": 0.30047628879547117, + "epoch": 7.198137910968868, + "flow_matching_loss": 0.009762455755844713, + "grad_norm": 0.08017634600400925, + "kept_token_count": 572.27578125, + "learning_rate": 2.452460040026989e-06, + "loss": 0.0108, + "mgd_cos_sim": 0.9940976500511169, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059023459907621145, + "mgd_loss_weight": 0.17548917979002, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 427.67249755859376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5856048583984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24740, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.30023818612098696, + "epoch": 7.201047425080012, + "flow_matching_loss": 0.011194363282993437, + "grad_norm": 0.11891049891710281, + "kept_token_count": 572.7625, + "learning_rate": 2.4434072646242555e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.994104015827179, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005895982868969441, + "mgd_loss_weight": 0.17515584081411362, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.7519927978516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.68841247558595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24750, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.30002195239067075, + "epoch": 7.203956939191155, + "flow_matching_loss": 0.00975659554824233, + "grad_norm": 0.09442217648029327, + "kept_token_count": 572.66015625, + "learning_rate": 2.4343697466657015e-06, + "loss": 0.0108, + "mgd_cos_sim": 0.9939448177814484, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006055188551545143, + "mgd_loss_weight": 0.17482249289751053, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.8164520263672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.67051391601564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24760, + "valid_token_count": 818.10625 + }, + { + "actual_token_mask_ratio": 0.2995716243982315, + "epoch": 7.206866453302299, + "flow_matching_loss": 0.00884034289047122, + "grad_norm": 0.0730750560760498, + "kept_token_count": 573.23515625, + "learning_rate": 2.42534749713277e-06, + "loss": 0.0099, + "mgd_cos_sim": 0.9937323987483978, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00626760208979249, + "mgd_loss_weight": 0.17448914796113968, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.4083526611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3535552978516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24770, + "valid_token_count": 818.40390625 + }, + { + "actual_token_mask_ratio": 0.30006695687770846, + "epoch": 7.209775967413442, + "flow_matching_loss": 0.010563275218009949, + "grad_norm": 0.055962808430194855, + "kept_token_count": 572.92265625, + "learning_rate": 2.4163405269883425e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9938878834247589, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006112116714939475, + "mgd_loss_weight": 0.1741558089852333, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.55718383789065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.641064453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24780, + "valid_token_count": 818.54609375 + }, + { + "actual_token_mask_ratio": 0.2996458917856216, + "epoch": 7.212685481524585, + "flow_matching_loss": 0.010556748928502202, + "grad_norm": 0.05984252318739891, + "kept_token_count": 573.21015625, + "learning_rate": 2.407348847176735e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9938118577003479, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061881494708359245, + "mgd_loss_weight": 0.17382245808839797, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.6175903320312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.93101806640624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24790, + "valid_token_count": 818.45859375 + }, + { + "actual_token_mask_ratio": 0.2991281896829605, + "epoch": 7.215594995635729, + "flow_matching_loss": 0.01086807120591402, + "grad_norm": 0.0823216438293457, + "kept_token_count": 573.62734375, + "learning_rate": 2.398372468623688e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9938023924827576, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006197600113227964, + "mgd_loss_weight": 0.17348911613225937, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.24280700683596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3137237548828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24800, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.29966155290603635, + "epoch": 7.2185045097468725, + "flow_matching_loss": 0.009667446697130799, + "grad_norm": 0.049389615654945374, + "kept_token_count": 573.18828125, + "learning_rate": 2.3894114022363446e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9940651476383209, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059348615817725655, + "mgd_loss_weight": 0.17315577566623688, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.121630859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.61217041015624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24810, + "valid_token_count": 818.43984375 + }, + { + "actual_token_mask_ratio": 0.2999293565750122, + "epoch": 7.221414023858015, + "flow_matching_loss": 0.009934061951935292, + "grad_norm": 0.07511084526777267, + "kept_token_count": 573.0515625, + "learning_rate": 2.3804656589032514e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9940535187721252, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005946486862376333, + "mgd_loss_weight": 0.17282242327928543, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.66395263671876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0332275390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24820, + "valid_token_count": 818.56015625 + }, + { + "actual_token_mask_ratio": 0.2996627390384674, + "epoch": 7.224323537969159, + "flow_matching_loss": 0.00979608679190278, + "grad_norm": 0.06791269779205322, + "kept_token_count": 573.14765625, + "learning_rate": 2.3715352494943237e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9938081681728363, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006191823445260525, + "mgd_loss_weight": 0.17248908430337906, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.1612274169922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5071136474609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24830, + "valid_token_count": 818.38984375 + }, + { + "actual_token_mask_ratio": 0.300290521979332, + "epoch": 7.227233052080303, + "flow_matching_loss": 0.010182980401441454, + "grad_norm": 0.07470990717411041, + "kept_token_count": 572.44140625, + "learning_rate": 2.3626201848608582e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.994152557849884, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005847440846264362, + "mgd_loss_weight": 0.1721557393670082, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.8171020507813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7001678466797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24840, + "valid_token_count": 818.11484375 + }, + { + "actual_token_mask_ratio": 0.29988677203655245, + "epoch": 7.2301425661914465, + "flow_matching_loss": 0.010689091589301824, + "grad_norm": 0.0788988545536995, + "kept_token_count": 572.784375, + "learning_rate": 2.353720475835494e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9940128803253174, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005987114133313298, + "mgd_loss_weight": 0.17182239145040512, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.0357727050781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.47266845703126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24850, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.30038420855998993, + "epoch": 7.233052080302589, + "flow_matching_loss": 0.009025881253182887, + "grad_norm": 0.08427184075117111, + "kept_token_count": 572.53984375, + "learning_rate": 2.344836133232215e-06, + "loss": 0.01, + "mgd_cos_sim": 0.9941382348537445, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005861782934516668, + "mgd_loss_weight": 0.17148905247449875, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4071807861328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.76011962890624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24860, + "valid_token_count": 818.36796875 + }, + { + "actual_token_mask_ratio": 0.2998581975698471, + "epoch": 7.235961594413733, + "flow_matching_loss": 0.010468839854001998, + "grad_norm": 0.06875169277191162, + "kept_token_count": 573.07265625, + "learning_rate": 2.335967167846342e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9941109240055084, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0058890807442367075, + "mgd_loss_weight": 0.17115570306777955, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.85435485839844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4893798828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24870, + "valid_token_count": 818.509375 + }, + { + "actual_token_mask_ratio": 0.29951720535755155, + "epoch": 7.238871108524877, + "flow_matching_loss": 0.010213358234614134, + "grad_norm": 0.07307920604944229, + "kept_token_count": 573.04765625, + "learning_rate": 2.3271135904544993e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9937564015388489, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062435805331915615, + "mgd_loss_weight": 0.1708223596215248, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.50868835449216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7414031982422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24880, + "valid_token_count": 818.07734375 + }, + { + "actual_token_mask_ratio": 0.3007055759429932, + "epoch": 7.2417806226360195, + "flow_matching_loss": 0.009203980537131428, + "grad_norm": 0.07415574043989182, + "kept_token_count": 572.34375, + "learning_rate": 2.3182754118146206e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9937946319580078, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006205362966284156, + "mgd_loss_weight": 0.17048902064561844, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.6683410644531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7382293701172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24890, + "valid_token_count": 818.459375 + }, + { + "actual_token_mask_ratio": 0.29992047846317293, + "epoch": 7.244690136747163, + "flow_matching_loss": 0.00977912344969809, + "grad_norm": 0.06514289975166321, + "kept_token_count": 572.79375, + "learning_rate": 2.309452642665923e-06, + "loss": 0.0108, + "mgd_cos_sim": 0.994028913974762, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005971084022894502, + "mgd_loss_weight": 0.170155668258667, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.4969207763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.161962890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24900, + "valid_token_count": 818.18046875 + }, + { + "actual_token_mask_ratio": 0.29977706670761106, + "epoch": 7.247599650858307, + "flow_matching_loss": 0.00993119003251195, + "grad_norm": 0.08166977763175964, + "kept_token_count": 573.04453125, + "learning_rate": 2.3006452937289065e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9942155539989471, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005784439761191607, + "mgd_loss_weight": 0.1698223277926445, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.1063568115234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.2476043701172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24910, + "valid_token_count": 818.37578125 + }, + { + "actual_token_mask_ratio": 0.3003877878189087, + "epoch": 7.25050916496945, + "flow_matching_loss": 0.010294524114578962, + "grad_norm": 0.0623989999294281, + "kept_token_count": 572.334375, + "learning_rate": 2.291853375705327e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9942494750022888, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005750531703233719, + "mgd_loss_weight": 0.1694889858365059, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.6757873535156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.67109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24920, + "valid_token_count": 818.0671875 + }, + { + "actual_token_mask_ratio": 0.2999112904071808, + "epoch": 7.2534186790805935, + "flow_matching_loss": 0.010764641081914305, + "grad_norm": 0.07493360340595245, + "kept_token_count": 572.83125, + "learning_rate": 2.283076899278197e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9938704371452332, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006129568023607135, + "mgd_loss_weight": 0.16915563493967056, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.1816680908203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8784423828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24930, + "valid_token_count": 818.22734375 + }, + { + "actual_token_mask_ratio": 0.30041220486164094, + "epoch": 7.256328193191737, + "flow_matching_loss": 0.01072638975456357, + "grad_norm": 0.06026334688067436, + "kept_token_count": 572.3671875, + "learning_rate": 2.274315875111763e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9939164340496063, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006083573680371046, + "mgd_loss_weight": 0.1688222959637642, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.2367858886719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2842071533203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24940, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.2997091799974442, + "epoch": 7.25923770730288, + "flow_matching_loss": 0.009804358053952456, + "grad_norm": 0.061004944145679474, + "kept_token_count": 573.06796875, + "learning_rate": 2.265570313851493e-06, + "loss": 0.0108, + "mgd_cos_sim": 0.9939594388008117, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006040560221299529, + "mgd_loss_weight": 0.16848894953727722, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.7192047119141, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9720458984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24950, + "valid_token_count": 818.33203125 + }, + { + "actual_token_mask_ratio": 0.2999307751655579, + "epoch": 7.262147221414024, + "flow_matching_loss": 0.01042288145981729, + "grad_norm": 0.07835868746042252, + "kept_token_count": 572.74296875, + "learning_rate": 2.2568402261240685e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9940141975879669, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005985809210687876, + "mgd_loss_weight": 0.16815560311079025, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.353564453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.04124145507814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24960, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.3005118727684021, + "epoch": 7.265056735525167, + "flow_matching_loss": 0.009821571782231331, + "grad_norm": 0.06958071887493134, + "kept_token_count": 572.2984375, + "learning_rate": 2.248125622537373e-06, + "loss": 0.0108, + "mgd_cos_sim": 0.9940735161304474, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005926474649459124, + "mgd_loss_weight": 0.16782226413488388, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.2578491210937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.36712341308595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24970, + "valid_token_count": 818.1671875 + }, + { + "actual_token_mask_ratio": 0.29996026754379274, + "epoch": 7.267966249636311, + "flow_matching_loss": 0.011922165844589471, + "grad_norm": 0.07450820505619049, + "kept_token_count": 572.81171875, + "learning_rate": 2.239426513680467e-06, + "loss": 0.0129, + "mgd_cos_sim": 0.9940904796123504, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005909525044262409, + "mgd_loss_weight": 0.16748891323804854, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.7200927734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.53774108886716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24980, + "valid_token_count": 818.2578125 + }, + { + "actual_token_mask_ratio": 0.2993035167455673, + "epoch": 7.270875763747454, + "flow_matching_loss": 0.008433366194367409, + "grad_norm": 0.07636392116546631, + "kept_token_count": 573.08203125, + "learning_rate": 2.2307429101235915e-06, + "loss": 0.0095, + "mgd_cos_sim": 0.9937720119953155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006227984605357051, + "mgd_loss_weight": 0.16715557128190994, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.7450927734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.80405578613284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 24990, + "valid_token_count": 817.87578125 + }, + { + "actual_token_mask_ratio": 0.2996959775686264, + "epoch": 7.273785277858598, + "flow_matching_loss": 0.010760145215317606, + "grad_norm": 0.07322844117879868, + "kept_token_count": 572.99609375, + "learning_rate": 2.222074822418145e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9938335955142975, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006166416220366955, + "mgd_loss_weight": 0.16682223230600357, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.676318359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.58753356933596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25000, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.3000164121389389, + "epoch": 7.276694791969741, + "flow_matching_loss": 0.010352202598005534, + "grad_norm": 0.07343009859323502, + "kept_token_count": 572.75546875, + "learning_rate": 2.213422261096668e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9937926948070526, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006207305379211903, + "mgd_loss_weight": 0.166488878428936, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.6049530029297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.13173217773436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25010, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.3001988768577576, + "epoch": 7.279604306080884, + "flow_matching_loss": 0.009776618285104632, + "grad_norm": 0.0688406452536583, + "kept_token_count": 572.54140625, + "learning_rate": 2.204785236672835e-06, + "loss": 0.0108, + "mgd_cos_sim": 0.994131076335907, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005868927855044603, + "mgd_loss_weight": 0.16615553945302963, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.6120849609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.11309509277345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25020, + "valid_token_count": 818.14921875 + }, + { + "actual_token_mask_ratio": 0.29938073456287384, + "epoch": 7.282513820192028, + "flow_matching_loss": 0.010623388085514307, + "grad_norm": 0.08246102929115295, + "kept_token_count": 573.09453125, + "learning_rate": 2.1961637596414536e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9935369491577148, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006463042227551341, + "mgd_loss_weight": 0.1658221960067749, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.47142639160154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6533630371094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25030, + "valid_token_count": 817.98671875 + }, + { + "actual_token_mask_ratio": 0.2999481946229935, + "epoch": 7.285423334303172, + "flow_matching_loss": 0.01146345748566091, + "grad_norm": 0.08200856298208237, + "kept_token_count": 572.74765625, + "learning_rate": 2.1875578404784275e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9938511729240418, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061488425824791195, + "mgd_loss_weight": 0.1654888466000557, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.9629302978516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.00969848632815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25040, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.30062656104564667, + "epoch": 7.288332848414315, + "flow_matching_loss": 0.009645178355276585, + "grad_norm": 0.056064486503601074, + "kept_token_count": 572.425, + "learning_rate": 2.1789674896407583e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.9944036424160003, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005596349574625492, + "mgd_loss_weight": 0.16515550762414932, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.85224609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.91468505859376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25050, + "valid_token_count": 818.4859375 + }, + { + "actual_token_mask_ratio": 0.29966804683208464, + "epoch": 7.291242362525458, + "flow_matching_loss": 0.010143583873286843, + "grad_norm": 0.06337190419435501, + "kept_token_count": 573.14375, + "learning_rate": 2.170392717566534e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9940050423145295, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005994958896189928, + "mgd_loss_weight": 0.16482215970754624, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.0289337158203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.02281188964844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25060, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.29943961203098296, + "epoch": 7.294151876636602, + "flow_matching_loss": 0.01022039926610887, + "grad_norm": 0.047459185123443604, + "kept_token_count": 573.37421875, + "learning_rate": 2.161833534674909e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9938817143440246, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061182908713817595, + "mgd_loss_weight": 0.16448881477117538, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.3518310546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2628509521484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25070, + "valid_token_count": 818.4546875 + }, + { + "actual_token_mask_ratio": 0.3000615805387497, + "epoch": 7.2970613907477455, + "flow_matching_loss": 0.010454490967094898, + "grad_norm": 0.058651674538850784, + "kept_token_count": 572.80546875, + "learning_rate": 2.1532899513661015e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9936108469963074, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006389154586941004, + "mgd_loss_weight": 0.164155475795269, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.5413757324219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7865783691406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25080, + "valid_token_count": 818.365625 + }, + { + "actual_token_mask_ratio": 0.3002077043056488, + "epoch": 7.299970904858888, + "flow_matching_loss": 0.010272328834980726, + "grad_norm": 0.06719950586557388, + "kept_token_count": 572.6609375, + "learning_rate": 2.1447619780213682e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9941579043865204, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005842098873108626, + "mgd_loss_weight": 0.16382212340831756, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.3946197509766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.98585510253906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25090, + "valid_token_count": 818.32734375 + }, + { + "actual_token_mask_ratio": 0.3005330562591553, + "epoch": 7.302880418970032, + "flow_matching_loss": 0.009973758039996028, + "grad_norm": 0.06218327209353447, + "kept_token_count": 572.165625, + "learning_rate": 2.136249625003001e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9937212288379669, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006278762919828295, + "mgd_loss_weight": 0.16348878294229507, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.1158020019531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.12957763671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25100, + "valid_token_count": 818.00234375 + }, + { + "actual_token_mask_ratio": 0.30038736164569857, + "epoch": 7.305789933081176, + "flow_matching_loss": 0.010398675920441747, + "grad_norm": 0.05246853828430176, + "kept_token_count": 572.37578125, + "learning_rate": 2.1277529026543107e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.993957906961441, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060420953668653965, + "mgd_loss_weight": 0.1631554424762726, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.0872375488281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.42003173828124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25110, + "valid_token_count": 818.13203125 + }, + { + "actual_token_mask_ratio": 0.29956803321838377, + "epoch": 7.308699447192319, + "flow_matching_loss": 0.010567399626597761, + "grad_norm": 0.06232471391558647, + "kept_token_count": 573.075, + "learning_rate": 2.1192718212996188e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.994161456823349, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005838542943820357, + "mgd_loss_weight": 0.16282209008932114, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.0906188964844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3221008300781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25120, + "valid_token_count": 818.17265625 + }, + { + "actual_token_mask_ratio": 0.3003705531358719, + "epoch": 7.311608961303462, + "flow_matching_loss": 0.008919106610119343, + "grad_norm": 0.059758760035037994, + "kept_token_count": 572.35234375, + "learning_rate": 2.110806391244235e-06, + "loss": 0.0099, + "mgd_cos_sim": 0.9940715730190277, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005928426003083587, + "mgd_loss_weight": 0.16248875111341476, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.47413330078126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.3946868896484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25130, + "valid_token_count": 818.08046875 + }, + { + "actual_token_mask_ratio": 0.29987328946590425, + "epoch": 7.314518475414606, + "flow_matching_loss": 0.010872122179716826, + "grad_norm": 0.07809996604919434, + "kept_token_count": 572.80703125, + "learning_rate": 2.1023566227744596e-06, + "loss": 0.0118, + "mgd_cos_sim": 0.9940843760967255, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005915614264085889, + "mgd_loss_weight": 0.1621554061770439, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.90476684570314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8750305175781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25140, + "valid_token_count": 818.14453125 + }, + { + "actual_token_mask_ratio": 0.3000421106815338, + "epoch": 7.317427989525749, + "flow_matching_loss": 0.010696198605000973, + "grad_norm": 0.06863370537757874, + "kept_token_count": 572.578125, + "learning_rate": 2.0939225261575563e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9936973989009857, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006302599562332034, + "mgd_loss_weight": 0.16182205826044083, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.91558227539065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9674591064453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25150, + "valid_token_count": 818.02109375 + }, + { + "actual_token_mask_ratio": 0.30023506879806516, + "epoch": 7.3203375036368925, + "flow_matching_loss": 0.00933722611516714, + "grad_norm": 0.08674698323011398, + "kept_token_count": 572.58125, + "learning_rate": 2.0855041116417473e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9940350115299225, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005964977154508233, + "mgd_loss_weight": 0.16148871928453445, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.0384033203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6464569091797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25160, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.29917273223400115, + "epoch": 7.323247017748036, + "flow_matching_loss": 0.009726791642606258, + "grad_norm": 0.0819072276353836, + "kept_token_count": 573.4015625, + "learning_rate": 2.0771013894562003e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.994201374053955, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005798632232472301, + "mgd_loss_weight": 0.16115536987781526, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.26683349609374, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5707122802734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25170, + "valid_token_count": 818.17890625 + }, + { + "actual_token_mask_ratio": 0.2997723639011383, + "epoch": 7.32615653185918, + "flow_matching_loss": 0.009466558927670121, + "grad_norm": 0.07741384208202362, + "kept_token_count": 573.12578125, + "learning_rate": 2.0687143698110113e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9940638065338134, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005936182942241431, + "mgd_loss_weight": 0.16082202643156052, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.80152893066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6350067138672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25180, + "valid_token_count": 818.48046875 + }, + { + "actual_token_mask_ratio": 0.3003138452768326, + "epoch": 7.329066045970323, + "flow_matching_loss": 0.010305725410580635, + "grad_norm": 0.08977211266756058, + "kept_token_count": 572.4515625, + "learning_rate": 2.060343062897208e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9937137484550476, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006286247679963708, + "mgd_loss_weight": 0.16048868745565414, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.62860107421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3345092773437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25190, + "valid_token_count": 818.15625 + }, + { + "actual_token_mask_ratio": 0.2996089071035385, + "epoch": 7.3319755600814664, + "flow_matching_loss": 0.009783803392201662, + "grad_norm": 0.06532015651464462, + "kept_token_count": 573.2390625, + "learning_rate": 2.0519874788867127e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9940097510814667, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059902490582317116, + "mgd_loss_weight": 0.16015533357858658, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.35341796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8888732910156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25200, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.3008837133646011, + "epoch": 7.33488507419261, + "flow_matching_loss": 0.011273084208369255, + "grad_norm": 0.09979204833507538, + "kept_token_count": 572.04609375, + "learning_rate": 2.043647627932347e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9940230250358582, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005976985488086939, + "mgd_loss_weight": 0.1598219946026802, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.02655029296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.404150390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25210, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.3002581626176834, + "epoch": 7.337794588303753, + "flow_matching_loss": 0.010005530528724193, + "grad_norm": 0.0656522884964943, + "kept_token_count": 572.59296875, + "learning_rate": 2.0353235201678193e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9938808441162109, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00611917139030993, + "mgd_loss_weight": 0.1594886526465416, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.0333282470703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.31731567382815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25220, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.29938264191150665, + "epoch": 7.340704102414897, + "flow_matching_loss": 0.009420477040112018, + "grad_norm": 0.06186295300722122, + "kept_token_count": 573.365625, + "learning_rate": 2.0270151657076995e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9940381586551666, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005961827654391527, + "mgd_loss_weight": 0.15915530174970627, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.69353942871095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.42882080078124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25230, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.3001684665679932, + "epoch": 7.34361361652604, + "flow_matching_loss": 0.011779774026945234, + "grad_norm": 0.07774543017148972, + "kept_token_count": 572.72265625, + "learning_rate": 2.0187225746474286e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9939915239810944, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006008476763963699, + "mgd_loss_weight": 0.1588219627737999, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.1247894287109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8789764404297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25240, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.30020929872989655, + "epoch": 7.346523130637183, + "flow_matching_loss": 0.010012292861938476, + "grad_norm": 0.05922162905335426, + "kept_token_count": 572.6140625, + "learning_rate": 2.0104457570632845e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9942510664463043, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0057489394675940275, + "mgd_loss_weight": 0.15848861634731293, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.8869262695313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9524871826172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25250, + "valid_token_count": 818.2671875 + }, + { + "actual_token_mask_ratio": 0.30000723004341123, + "epoch": 7.349432644748327, + "flow_matching_loss": 0.00930280378088355, + "grad_norm": 0.09776221215724945, + "kept_token_count": 572.7296875, + "learning_rate": 2.0021847230123775e-06, + "loss": 0.0102, + "mgd_cos_sim": 0.9942182242870331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005781787866726518, + "mgd_loss_weight": 0.15815526992082596, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.4565032958984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2883056640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25260, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.29993617832660674, + "epoch": 7.352342158859471, + "flow_matching_loss": 0.009366183262318372, + "grad_norm": 0.055315300822257996, + "kept_token_count": 572.959375, + "learning_rate": 1.9939394825326494e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9940458595752716, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005954137165099382, + "mgd_loss_weight": 0.1578219309449196, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.5506164550781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4776214599609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25270, + "valid_token_count": 818.440625 + }, + { + "actual_token_mask_ratio": 0.3008417457342148, + "epoch": 7.355251672970613, + "flow_matching_loss": 0.009212123928591609, + "grad_norm": 0.05087278038263321, + "kept_token_count": 572.19296875, + "learning_rate": 1.9857100456428385e-06, + "loss": 0.0101, + "mgd_cos_sim": 0.9941706895828247, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0058293189387768505, + "mgd_loss_weight": 0.15748858004808425, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 429.2654266357422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2259582519531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25280, + "valid_token_count": 818.40390625 + }, + { + "actual_token_mask_ratio": 0.3001181066036224, + "epoch": 7.358161187081757, + "flow_matching_loss": 0.010504335816949605, + "grad_norm": 0.0688079372048378, + "kept_token_count": 572.66953125, + "learning_rate": 1.977496422342492e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9940643072128296, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005935680028051138, + "mgd_loss_weight": 0.15715523809194565, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.9209930419922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8902557373047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25290, + "valid_token_count": 818.23671875 + }, + { + "actual_token_mask_ratio": 0.2999504119157791, + "epoch": 7.361070701192901, + "flow_matching_loss": 0.0097602519672364, + "grad_norm": 0.0793321505188942, + "kept_token_count": 572.725, + "learning_rate": 1.9692986226119345e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9940152764320374, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00598471756093204, + "mgd_loss_weight": 0.15682189762592316, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.8484405517578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.9781768798828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25300, + "valid_token_count": 818.11796875 + }, + { + "actual_token_mask_ratio": 0.3005712151527405, + "epoch": 7.363980215304045, + "flow_matching_loss": 0.011981021612882614, + "grad_norm": 0.08075914531946182, + "kept_token_count": 572.22265625, + "learning_rate": 1.9611166564122658e-06, + "loss": 0.013, + "mgd_cos_sim": 0.9937869787216187, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006213024258613586, + "mgd_loss_weight": 0.1564885452389717, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.5409912109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.267333984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25310, + "valid_token_count": 818.13046875 + }, + { + "actual_token_mask_ratio": 0.299747622013092, + "epoch": 7.366889729415187, + "flow_matching_loss": 0.01034745662473142, + "grad_norm": 0.06971751898527145, + "kept_token_count": 573.05234375, + "learning_rate": 1.9529505336853456e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9937617123126984, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006238284800201654, + "mgd_loss_weight": 0.15615520626306534, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.1790252685547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1135284423828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25320, + "valid_token_count": 818.35390625 + }, + { + "actual_token_mask_ratio": 0.29939102828502656, + "epoch": 7.369799243526331, + "flow_matching_loss": 0.011565587157383561, + "grad_norm": 0.05526053532958031, + "kept_token_count": 573.3234375, + "learning_rate": 1.9448002643537833e-06, + "loss": 0.0125, + "mgd_cos_sim": 0.9940191745758057, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005980811407789588, + "mgd_loss_weight": 0.1558218628168106, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.7306671142578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.00884704589845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25330, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.2996685057878494, + "epoch": 7.372708757637475, + "flow_matching_loss": 0.008825615653768182, + "grad_norm": 0.07403916120529175, + "kept_token_count": 572.9875, + "learning_rate": 1.9366658583209205e-06, + "loss": 0.0097, + "mgd_cos_sim": 0.9940877437591553, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00591226639226079, + "mgd_loss_weight": 0.1554885134100914, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.5674621582031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.20367431640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25340, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.2998844087123871, + "epoch": 7.375618271748618, + "flow_matching_loss": 0.009097789414227009, + "grad_norm": 0.07648375630378723, + "kept_token_count": 572.85234375, + "learning_rate": 1.9285473254708363e-06, + "loss": 0.0101, + "mgd_cos_sim": 0.9937893688678742, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006210636300966144, + "mgd_loss_weight": 0.15515517443418503, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.96387939453126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.67592468261716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25350, + "valid_token_count": 818.221875 + }, + { + "actual_token_mask_ratio": 0.3005296796560287, + "epoch": 7.378527785859761, + "flow_matching_loss": 0.010628471244126558, + "grad_norm": 0.06345345079898834, + "kept_token_count": 572.37421875, + "learning_rate": 1.9204446756683092e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9938906013965607, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00610939827747643, + "mgd_loss_weight": 0.15482182651758195, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.2447540283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.79938354492185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25360, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.3005844444036484, + "epoch": 7.381437299970905, + "flow_matching_loss": 0.009884716104716062, + "grad_norm": 0.05685977637767792, + "kept_token_count": 572.2921875, + "learning_rate": 1.912357918758822e-06, + "loss": 0.0108, + "mgd_cos_sim": 0.9939326286315918, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060673804022371765, + "mgd_loss_weight": 0.1544884815812111, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.8840057373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4853271484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25370, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.2998281419277191, + "epoch": 7.384346814082049, + "flow_matching_loss": 0.009270105324685574, + "grad_norm": 0.06000018119812012, + "kept_token_count": 572.91015625, + "learning_rate": 1.9042870645685478e-06, + "loss": 0.0102, + "mgd_cos_sim": 0.9939201593399047, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006079853978008032, + "mgd_loss_weight": 0.15415514260530472, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.79591674804686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.18967590332034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25380, + "valid_token_count": 818.23515625 + }, + { + "actual_token_mask_ratio": 0.2996948927640915, + "epoch": 7.387256328193192, + "flow_matching_loss": 0.010610855789855123, + "grad_norm": 0.07751180231571198, + "kept_token_count": 573.06796875, + "learning_rate": 1.8962321229043322e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9939354300498963, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006064563151448965, + "mgd_loss_weight": 0.15382179021835327, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.73302001953124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4564636230469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25390, + "valid_token_count": 818.31328125 + }, + { + "actual_token_mask_ratio": 0.29949831068515775, + "epoch": 7.390165842304335, + "flow_matching_loss": 0.011484029656276106, + "grad_norm": 0.08170925825834274, + "kept_token_count": 573.22109375, + "learning_rate": 1.8881931035536943e-06, + "loss": 0.0124, + "mgd_cos_sim": 0.9939452648162842, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006054733041673898, + "mgd_loss_weight": 0.15348844975233078, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.25965576171876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1289489746094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25400, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.30030763149261475, + "epoch": 7.393075356415479, + "flow_matching_loss": 0.009997680643573403, + "grad_norm": 0.05870679393410683, + "kept_token_count": 572.3453125, + "learning_rate": 1.8801700162847952e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9939892590045929, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00601075547747314, + "mgd_loss_weight": 0.15315510779619218, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.17198486328124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2665283203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25410, + "valid_token_count": 817.99296875 + }, + { + "actual_token_mask_ratio": 0.30006581246852876, + "epoch": 7.395984870526622, + "flow_matching_loss": 0.011211140546947718, + "grad_norm": 0.09466851502656937, + "kept_token_count": 572.59453125, + "learning_rate": 1.87216287084645e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.993830269575119, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006169730750843883, + "mgd_loss_weight": 0.15282175689935684, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.9030426025391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7019409179687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25420, + "valid_token_count": 818.075 + }, + { + "actual_token_mask_ratio": 0.3002722799777985, + "epoch": 7.3988943846377655, + "flow_matching_loss": 0.010358566511422395, + "grad_norm": 0.05840832740068436, + "kept_token_count": 572.528125, + "learning_rate": 1.8641716769680895e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9937641322612762, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006235876074060798, + "mgd_loss_weight": 0.15248841792345047, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.68927612304685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6132446289063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25430, + "valid_token_count": 818.215625 + }, + { + "actual_token_mask_ratio": 0.3003369629383087, + "epoch": 7.401803898748909, + "flow_matching_loss": 0.009562843106687069, + "grad_norm": 0.06176323443651199, + "kept_token_count": 572.51015625, + "learning_rate": 1.8561964443597662e-06, + "loss": 0.0105, + "mgd_cos_sim": 0.9937674462795257, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006232546921819448, + "mgd_loss_weight": 0.15215507298707961, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.2590576171875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.36240234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25440, + "valid_token_count": 818.26171875 + }, + { + "actual_token_mask_ratio": 0.29994178712368014, + "epoch": 7.404713412860052, + "flow_matching_loss": 0.010574967646971346, + "grad_norm": 0.07564952969551086, + "kept_token_count": 572.75546875, + "learning_rate": 1.848237182712147e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9940901279449463, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005909872660413385, + "mgd_loss_weight": 0.15182172507047653, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.1432220458984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9355438232422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25450, + "valid_token_count": 818.14921875 + }, + { + "actual_token_mask_ratio": 0.29977898597717284, + "epoch": 7.407622926971196, + "flow_matching_loss": 0.00926970667205751, + "grad_norm": 0.061979684978723526, + "kept_token_count": 572.9171875, + "learning_rate": 1.8402939016964788e-06, + "loss": 0.0102, + "mgd_cos_sim": 0.9938409626483917, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061590300872921945, + "mgd_loss_weight": 0.15148838609457016, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.95188903808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.34156799316406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25460, + "valid_token_count": 818.19375 + }, + { + "actual_token_mask_ratio": 0.30005564987659455, + "epoch": 7.410532441082339, + "flow_matching_loss": 0.010577093437314033, + "grad_norm": 0.056698884814977646, + "kept_token_count": 572.81484375, + "learning_rate": 1.832366610964601e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9937705934047699, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006229405198246241, + "mgd_loss_weight": 0.15115503668785096, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.87827758789064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9031707763672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25470, + "valid_token_count": 818.371875 + }, + { + "actual_token_mask_ratio": 0.3004601687192917, + "epoch": 7.413441955193482, + "flow_matching_loss": 0.009882458252832294, + "grad_norm": 0.051801443099975586, + "kept_token_count": 572.4, + "learning_rate": 1.8244553201489166e-06, + "loss": 0.0108, + "mgd_cos_sim": 0.9938730597496033, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00612693028524518, + "mgd_loss_weight": 0.15082169324159622, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.61138610839845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.4443695068359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25480, + "valid_token_count": 818.2484375 + }, + { + "actual_token_mask_ratio": 0.30000032782554625, + "epoch": 7.416351469304626, + "flow_matching_loss": 0.009676014631986618, + "grad_norm": 0.08042281121015549, + "kept_token_count": 572.72265625, + "learning_rate": 1.8165600388623882e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.9939730405807495, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006026964820921421, + "mgd_loss_weight": 0.15048835426568985, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.21971435546874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6960418701172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25490, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.29994216561317444, + "epoch": 7.41926098341577, + "flow_matching_loss": 0.010624109581112862, + "grad_norm": 0.07879775762557983, + "kept_token_count": 572.81640625, + "learning_rate": 1.8086807766985331e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9935301184654236, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006469880882650614, + "mgd_loss_weight": 0.15015500038862228, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.72157897949216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0808471679687, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25500, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.29991795122623444, + "epoch": 7.422170497526913, + "flow_matching_loss": 0.009484363766387106, + "grad_norm": 0.07327237725257874, + "kept_token_count": 572.71953125, + "learning_rate": 1.8008175432313946e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9942303419113159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005769658600911498, + "mgd_loss_weight": 0.1498216614127159, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.3371612548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.70191650390626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25510, + "valid_token_count": 818.07734375 + }, + { + "actual_token_mask_ratio": 0.3003439992666245, + "epoch": 7.425080011638056, + "flow_matching_loss": 0.00966085116378963, + "grad_norm": 0.08731098473072052, + "kept_token_count": 572.43828125, + "learning_rate": 1.7929703480155429e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.9939465999603272, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006053409306332469, + "mgd_loss_weight": 0.14948831796646117, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.9985107421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3807769775391, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25520, + "valid_token_count": 818.171875 + }, + { + "actual_token_mask_ratio": 0.2998553514480591, + "epoch": 7.4279895257492, + "flow_matching_loss": 0.011137067759409546, + "grad_norm": 0.07786829024553299, + "kept_token_count": 572.82421875, + "learning_rate": 1.7851392005860595e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9938309013843536, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006169109977781773, + "mgd_loss_weight": 0.14915496855974197, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.2282440185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.04481201171876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25530, + "valid_token_count": 818.15390625 + }, + { + "actual_token_mask_ratio": 0.29950169920921327, + "epoch": 7.430899039860344, + "flow_matching_loss": 0.009964241832494735, + "grad_norm": 0.0553257018327713, + "kept_token_count": 573.27578125, + "learning_rate": 1.7773241104585243e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9940050899982452, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005994918802753091, + "mgd_loss_weight": 0.1488216295838356, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.6805755615234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3916473388672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25540, + "valid_token_count": 818.38515625 + }, + { + "actual_token_mask_ratio": 0.2998633921146393, + "epoch": 7.433808553971486, + "flow_matching_loss": 0.009641924221068621, + "grad_norm": 0.08374449610710144, + "kept_token_count": 572.8765625, + "learning_rate": 1.7695250871290108e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.9938400328159332, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006159971468150616, + "mgd_loss_weight": 0.14848828315734863, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.2822265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.07923278808596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25550, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.29946381151676177, + "epoch": 7.43671806808263, + "flow_matching_loss": 0.010284574097022414, + "grad_norm": 0.07426536828279495, + "kept_token_count": 573.15546875, + "learning_rate": 1.761742140074074e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9939913928508759, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00600862530991435, + "mgd_loss_weight": 0.14815493673086166, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.50514526367186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.68369750976564, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25560, + "valid_token_count": 818.17109375 + }, + { + "actual_token_mask_ratio": 0.3000116139650345, + "epoch": 7.439627582193774, + "flow_matching_loss": 0.009248580643907189, + "grad_norm": 0.06193587929010391, + "kept_token_count": 572.70625, + "learning_rate": 1.7539752787507208e-06, + "loss": 0.0102, + "mgd_cos_sim": 0.993841552734375, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006158452806994319, + "mgd_loss_weight": 0.1478215977549553, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.37116088867185, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6121398925781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25570, + "valid_token_count": 818.16796875 + }, + { + "actual_token_mask_ratio": 0.29966886043548585, + "epoch": 7.442537096304917, + "flow_matching_loss": 0.011278187297284603, + "grad_norm": 0.06280305981636047, + "kept_token_count": 573.18828125, + "learning_rate": 1.7462245125964233e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9940418839454651, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059581249486655, + "mgd_loss_weight": 0.14748824685811995, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.0200531005859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.07636108398435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25580, + "valid_token_count": 818.45703125 + }, + { + "actual_token_mask_ratio": 0.29998522698879243, + "epoch": 7.44544661041606, + "flow_matching_loss": 0.01175213665701449, + "grad_norm": 0.07614526152610779, + "kept_token_count": 572.78203125, + "learning_rate": 1.7384898510290931e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9938977479934692, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006102242041379213, + "mgd_loss_weight": 0.14715490490198135, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.4261871337891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9660919189453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25590, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.29987633526325225, + "epoch": 7.448356124527204, + "flow_matching_loss": 0.010502904932945967, + "grad_norm": 0.06625496596097946, + "kept_token_count": 572.8296875, + "learning_rate": 1.730771303447069e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9938339948654175, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006165982317179441, + "mgd_loss_weight": 0.14682156443595887, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 432.33507385253904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0624542236328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25600, + "valid_token_count": 818.18203125 + }, + { + "actual_token_mask_ratio": 0.2993245333433151, + "epoch": 7.451265638638348, + "flow_matching_loss": 0.009661586442962288, + "grad_norm": 0.05532580241560936, + "kept_token_count": 573.2453125, + "learning_rate": 1.7230688792291204e-06, + "loss": 0.0105, + "mgd_cos_sim": 0.993999183177948, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00600082240998745, + "mgd_loss_weight": 0.14648821204900742, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 425.5926239013672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7392944335937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25610, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.3000793308019638, + "epoch": 7.454175152749491, + "flow_matching_loss": 0.011777406837791205, + "grad_norm": 0.1330539882183075, + "kept_token_count": 572.821875, + "learning_rate": 1.7153825877344175e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9940174639225006, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005982541339471936, + "mgd_loss_weight": 0.14615487307310104, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.0827239990234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.26909790039065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25620, + "valid_token_count": 818.4125 + }, + { + "actual_token_mask_ratio": 0.30010361671447755, + "epoch": 7.457084666860634, + "flow_matching_loss": 0.01004431415349245, + "grad_norm": 0.0609971322119236, + "kept_token_count": 572.83203125, + "learning_rate": 1.7077124383025278e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9939737021923065, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006026316108182073, + "mgd_loss_weight": 0.1458215281367302, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.49110107421876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8708953857422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25630, + "valid_token_count": 818.45390625 + }, + { + "actual_token_mask_ratio": 0.2998504102230072, + "epoch": 7.459994180971778, + "flow_matching_loss": 0.00950172683224082, + "grad_norm": 0.08013124018907547, + "kept_token_count": 573.03515625, + "learning_rate": 1.7000584402534058e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9938319623470306, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006168048223480582, + "mgd_loss_weight": 0.1454881802201271, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.99766540527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.93518981933596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25640, + "valid_token_count": 818.4515625 + }, + { + "actual_token_mask_ratio": 0.29950514137744905, + "epoch": 7.462903695082921, + "flow_matching_loss": 0.00995397507213056, + "grad_norm": 0.06937383115291595, + "kept_token_count": 573.19453125, + "learning_rate": 1.6924206028873774e-06, + "loss": 0.0108, + "mgd_cos_sim": 0.9938660681247711, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006133917486295104, + "mgd_loss_weight": 0.14515484124422073, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.8079010009766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8965148925781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25650, + "valid_token_count": 818.26953125 + }, + { + "actual_token_mask_ratio": 0.2992885559797287, + "epoch": 7.4658132091940645, + "flow_matching_loss": 0.010283038020133972, + "grad_norm": 0.06780827790498734, + "kept_token_count": 573.3484375, + "learning_rate": 1.6847989354851434e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9939892292022705, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006010768143460155, + "mgd_loss_weight": 0.14482149332761765, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.99774169921875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.43553771972654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25660, + "valid_token_count": 818.23359375 + }, + { + "actual_token_mask_ratio": 0.29955974817276, + "epoch": 7.468722723305208, + "flow_matching_loss": 0.00883628949522972, + "grad_norm": 0.05458337441086769, + "kept_token_count": 573.02109375, + "learning_rate": 1.6771934473077443e-06, + "loss": 0.0097, + "mgd_cos_sim": 0.9939305067062378, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006069489847868681, + "mgd_loss_weight": 0.1444881483912468, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.8350524902344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6260925292969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25670, + "valid_token_count": 818.0828125 + }, + { + "actual_token_mask_ratio": 0.3003279834985733, + "epoch": 7.471632237416351, + "flow_matching_loss": 0.010220269579440356, + "grad_norm": 0.06499263644218445, + "kept_token_count": 572.47734375, + "learning_rate": 1.669604147596563e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9934616684913635, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00653832065872848, + "mgd_loss_weight": 0.14415480941534042, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.1307098388672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4102813720703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25680, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.3000193923711777, + "epoch": 7.474541751527495, + "flow_matching_loss": 0.009701386420056224, + "grad_norm": 0.08123143017292023, + "kept_token_count": 572.8359375, + "learning_rate": 1.662031045573319e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.9940308570861817, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005969140725210309, + "mgd_loss_weight": 0.14382145702838897, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.00733642578126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.18369140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25690, + "valid_token_count": 818.36171875 + }, + { + "actual_token_mask_ratio": 0.2995984792709351, + "epoch": 7.4774512656386385, + "flow_matching_loss": 0.01044928263872862, + "grad_norm": 0.06307267397642136, + "kept_token_count": 573.13828125, + "learning_rate": 1.6544741504400402e-06, + "loss": 0.0113, + "mgd_cos_sim": 0.9940713882446289, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005928601091727615, + "mgd_loss_weight": 0.14348811656236649, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.6537231445312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.72182312011716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25700, + "valid_token_count": 818.29921875 + }, + { + "actual_token_mask_ratio": 0.2998785346746445, + "epoch": 7.480360779749782, + "flow_matching_loss": 0.008884273981675506, + "grad_norm": 0.09880264848470688, + "kept_token_count": 572.73984375, + "learning_rate": 1.6469334713790712e-06, + "loss": 0.0097, + "mgd_cos_sim": 0.9939539670944214, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060460417531430725, + "mgd_loss_weight": 0.14315477460622789, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.3752716064453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6684814453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25710, + "valid_token_count": 818.05390625 + }, + { + "actual_token_mask_ratio": 0.2994789987802505, + "epoch": 7.483270293860925, + "flow_matching_loss": 0.011096622981131076, + "grad_norm": 0.066552072763443, + "kept_token_count": 573.2390625, + "learning_rate": 1.639409017553048e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9943119943141937, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0056880178395658735, + "mgd_loss_weight": 0.14282142370939255, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.8567138671875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1647094726562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25720, + "valid_token_count": 818.30703125 + }, + { + "actual_token_mask_ratio": 0.29989429712295534, + "epoch": 7.486179807972069, + "flow_matching_loss": 0.00953184966929257, + "grad_norm": 0.055624306201934814, + "kept_token_count": 572.8421875, + "learning_rate": 1.6319007981048884e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9937311172485351, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00626889169216156, + "mgd_loss_weight": 0.14248808473348618, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.6250885009766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.73680419921874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25730, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.3005033880472183, + "epoch": 7.489089322083212, + "flow_matching_loss": 0.009567687986418605, + "grad_norm": 0.06322070211172104, + "kept_token_count": 572.3125, + "learning_rate": 1.6244088221577896e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9942852199077606, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005714773945510387, + "mgd_loss_weight": 0.1421547383069992, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.65819396972654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1653198242187, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25740, + "valid_token_count": 818.178125 + }, + { + "actual_token_mask_ratio": 0.30013352930545806, + "epoch": 7.491998836194355, + "flow_matching_loss": 0.011134608928114176, + "grad_norm": 0.07113077491521835, + "kept_token_count": 572.71953125, + "learning_rate": 1.6169330988152081e-06, + "loss": 0.012, + "mgd_cos_sim": 0.9939429879188537, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006057011103257537, + "mgd_loss_weight": 0.14182139188051224, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.51311645507815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.3194946289062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25750, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.3004104673862457, + "epoch": 7.494908350305499, + "flow_matching_loss": 0.009643102716654538, + "grad_norm": 0.08022255450487137, + "kept_token_count": 572.3984375, + "learning_rate": 1.6094736371608475e-06, + "loss": 0.0105, + "mgd_cos_sim": 0.9940693378448486, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005930650420486927, + "mgd_loss_weight": 0.14148805290460587, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.9691467285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.16002502441404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25760, + "valid_token_count": 818.19296875 + }, + { + "actual_token_mask_ratio": 0.29960252046585084, + "epoch": 7.497817864416643, + "flow_matching_loss": 0.009849266102537512, + "grad_norm": 0.07198453694581985, + "kept_token_count": 573.0671875, + "learning_rate": 1.602030446258666e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9937623798847198, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006237606052309275, + "mgd_loss_weight": 0.14115470349788667, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.8894378662109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7462188720703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25770, + "valid_token_count": 818.20078125 + }, + { + "actual_token_mask_ratio": 0.30012422800064087, + "epoch": 7.5007273785277855, + "flow_matching_loss": 0.01016032719053328, + "grad_norm": 0.07119804620742798, + "kept_token_count": 572.7796875, + "learning_rate": 1.5946035351528388e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9940366923809052, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005963311856612563, + "mgd_loss_weight": 0.14082136005163193, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.18036193847655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.46782836914065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25780, + "valid_token_count": 818.4046875 + }, + { + "actual_token_mask_ratio": 0.30010672807693484, + "epoch": 7.503636892638929, + "flow_matching_loss": 0.008844333607703448, + "grad_norm": 0.06493385881185532, + "kept_token_count": 572.7921875, + "learning_rate": 1.5871929128677621e-06, + "loss": 0.0097, + "mgd_cos_sim": 0.9941141605377197, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005885833548381925, + "mgd_loss_weight": 0.14048802107572556, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.8003295898437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9670166015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25790, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.299809655547142, + "epoch": 7.506546406750073, + "flow_matching_loss": 0.010110098961740732, + "grad_norm": 0.08217469602823257, + "kept_token_count": 573.10234375, + "learning_rate": 1.5797985884080412e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.994039672613144, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005960325058549642, + "mgd_loss_weight": 0.140154667198658, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.6132568359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1993896484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25800, + "valid_token_count": 818.496875 + }, + { + "actual_token_mask_ratio": 0.29945975840091704, + "epoch": 7.509455920861216, + "flow_matching_loss": 0.010485711926594376, + "grad_norm": 0.07419691979885101, + "kept_token_count": 573.23828125, + "learning_rate": 1.572420570758476e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.993750524520874, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006249461695551873, + "mgd_loss_weight": 0.13982132822275162, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.5859771728516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9390014648437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25810, + "valid_token_count": 818.28359375 + }, + { + "actual_token_mask_ratio": 0.29991117417812346, + "epoch": 7.512365434972359, + "flow_matching_loss": 0.01186474310234189, + "grad_norm": 0.07312896847724915, + "kept_token_count": 572.9609375, + "learning_rate": 1.5650588688840567e-06, + "loss": 0.0127, + "mgd_cos_sim": 0.9941748738288879, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005825144238770008, + "mgd_loss_weight": 0.13948798477649688, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.0355499267578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.56939697265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25820, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.29985205829143524, + "epoch": 7.515274949083503, + "flow_matching_loss": 0.011336007434874774, + "grad_norm": 0.06960202008485794, + "kept_token_count": 573.08984375, + "learning_rate": 1.5577134917299413e-06, + "loss": 0.0122, + "mgd_cos_sim": 0.9936359763145447, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006364019215106964, + "mgd_loss_weight": 0.13915463536977768, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.9823333740234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.22588500976565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25830, + "valid_token_count": 818.52734375 + }, + { + "actual_token_mask_ratio": 0.3001546382904053, + "epoch": 7.518184463194647, + "flow_matching_loss": 0.00940512390807271, + "grad_norm": 0.08506874740123749, + "kept_token_count": 572.696875, + "learning_rate": 1.5503844482214618e-06, + "loss": 0.0102, + "mgd_cos_sim": 0.9940958142280578, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059041988104581835, + "mgd_loss_weight": 0.1388212963938713, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.9252014160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1908843994141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25840, + "valid_token_count": 818.31796875 + }, + { + "actual_token_mask_ratio": 0.30015099942684176, + "epoch": 7.52109397730579, + "flow_matching_loss": 0.009290114603936672, + "grad_norm": 0.06350487470626831, + "kept_token_count": 572.61875, + "learning_rate": 1.5430717472640932e-06, + "loss": 0.0101, + "mgd_cos_sim": 0.994217985868454, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005782013898715377, + "mgd_loss_weight": 0.13848794847726822, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.919873046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.23545837402344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25850, + "valid_token_count": 818.20234375 + }, + { + "actual_token_mask_ratio": 0.300004380941391, + "epoch": 7.524003491416933, + "flow_matching_loss": 0.010521059762686491, + "grad_norm": 0.08462678641080856, + "kept_token_count": 572.67109375, + "learning_rate": 1.5357753977434541e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9935876190662384, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006412384426221251, + "mgd_loss_weight": 0.13815460354089737, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.860693359375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3188812255859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25860, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.2997181475162506, + "epoch": 7.526913005528077, + "flow_matching_loss": 0.010298666637390852, + "grad_norm": 0.07094895094633102, + "kept_token_count": 573.04609375, + "learning_rate": 1.5284954085253043e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9938309252262115, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006169075239449739, + "mgd_loss_weight": 0.137821264564991, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.38421020507815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7457244873047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25870, + "valid_token_count": 818.309375 + }, + { + "actual_token_mask_ratio": 0.30044851303100584, + "epoch": 7.52982251963922, + "flow_matching_loss": 0.009845318924635648, + "grad_norm": 0.06906910985708237, + "kept_token_count": 572.32890625, + "learning_rate": 1.5212317884555115e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9938122570514679, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006187735171988606, + "mgd_loss_weight": 0.13748791366815566, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.89638061523436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.6566864013672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25880, + "valid_token_count": 818.14140625 + }, + { + "actual_token_mask_ratio": 0.29966836273670194, + "epoch": 7.532732033750364, + "flow_matching_loss": 0.009454055922105908, + "grad_norm": 0.07187951356172562, + "kept_token_count": 572.921875, + "learning_rate": 1.5139845463600605e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9940381944179535, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005961809540167451, + "mgd_loss_weight": 0.13715457171201706, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.82000427246095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1582824707031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25890, + "valid_token_count": 818.07421875 + }, + { + "actual_token_mask_ratio": 0.29991087317466736, + "epoch": 7.535641547861507, + "flow_matching_loss": 0.009655933640897275, + "grad_norm": 0.06908000260591507, + "kept_token_count": 572.9375, + "learning_rate": 1.5067536910450352e-06, + "loss": 0.0105, + "mgd_cos_sim": 0.9940978050231933, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005902186967432499, + "mgd_loss_weight": 0.13682123124599457, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.1236602783203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.8023193359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25900, + "valid_token_count": 818.37734375 + }, + { + "actual_token_mask_ratio": 0.3000008165836334, + "epoch": 7.538551061972651, + "flow_matching_loss": 0.007886045146733522, + "grad_norm": 0.06129864230751991, + "kept_token_count": 572.66875, + "learning_rate": 1.4995392312966044e-06, + "loss": 0.0087, + "mgd_cos_sim": 0.9939938724040985, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006006127223372459, + "mgd_loss_weight": 0.13648787885904312, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.2493469238281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9580322265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25910, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.30064975917339326, + "epoch": 7.541460576083794, + "flow_matching_loss": 0.00959382951259613, + "grad_norm": 0.06586544960737228, + "kept_token_count": 572.25859375, + "learning_rate": 1.492341175881014e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9938213706016541, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006178625533357263, + "mgd_loss_weight": 0.13615453988313675, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.09917907714845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9858367919922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25920, + "valid_token_count": 818.27421875 + }, + { + "actual_token_mask_ratio": 0.29955620467662813, + "epoch": 7.5443700901949375, + "flow_matching_loss": 0.010114470077678562, + "grad_norm": 0.049678146839141846, + "kept_token_count": 573.228125, + "learning_rate": 1.4851595335445895e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9939354956150055, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006064507737755775, + "mgd_loss_weight": 0.1358211949467659, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.7947082519531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.43468017578124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25930, + "valid_token_count": 818.37734375 + }, + { + "actual_token_mask_ratio": 0.299807271361351, + "epoch": 7.547279604306081, + "flow_matching_loss": 0.010312627675011754, + "grad_norm": 0.06246088445186615, + "kept_token_count": 572.92890625, + "learning_rate": 1.4779943130136953e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9941091120243073, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0058908812701702114, + "mgd_loss_weight": 0.1354878470301628, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.3932586669922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7019958496094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25940, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.2993207186460495, + "epoch": 7.550189118417224, + "flow_matching_loss": 0.00898056598380208, + "grad_norm": 0.06613822281360626, + "kept_token_count": 573.15625, + "learning_rate": 1.4708455229947537e-06, + "loss": 0.0098, + "mgd_cos_sim": 0.9941649258136749, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0058350673411041495, + "mgd_loss_weight": 0.13515450805425644, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.82337341308596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3562744140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25950, + "valid_token_count": 817.99765625 + }, + { + "actual_token_mask_ratio": 0.299695149064064, + "epoch": 7.553098632528368, + "flow_matching_loss": 0.010144416661933064, + "grad_norm": 0.07942482829093933, + "kept_token_count": 573.05703125, + "learning_rate": 1.4637131721742148e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9941417574882507, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005858237668871879, + "mgd_loss_weight": 0.13482115864753724, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.12207946777346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.07574768066405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25960, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.30029689371585844, + "epoch": 7.5560081466395115, + "flow_matching_loss": 0.009448057040572166, + "grad_norm": 0.0567006841301918, + "kept_token_count": 572.653125, + "learning_rate": 1.4565972692185626e-06, + "loss": 0.0102, + "mgd_cos_sim": 0.994296270608902, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005703729018568993, + "mgd_loss_weight": 0.1344878152012825, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.71453552246095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4967010498047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25970, + "valid_token_count": 818.421875 + }, + { + "actual_token_mask_ratio": 0.29898491203784944, + "epoch": 7.558917660750654, + "flow_matching_loss": 0.009702266752719879, + "grad_norm": 0.14390069246292114, + "kept_token_count": 573.6109375, + "learning_rate": 1.4494978227742855e-06, + "loss": 0.0105, + "mgd_cos_sim": 0.9942461729049683, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005753833381459117, + "mgd_loss_weight": 0.13415447622537613, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.5471496582031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.72198181152345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25980, + "valid_token_count": 818.25703125 + }, + { + "actual_token_mask_ratio": 0.299648842215538, + "epoch": 7.561827174861798, + "flow_matching_loss": 0.010729235038161278, + "grad_norm": 0.05409117788076401, + "kept_token_count": 573.0046875, + "learning_rate": 1.442414841467885e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9938955068588257, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006104496447369457, + "mgd_loss_weight": 0.13382112383842468, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.05823974609376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.64450988769534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 25990, + "valid_token_count": 818.16953125 + }, + { + "actual_token_mask_ratio": 0.2992403209209442, + "epoch": 7.564736688972942, + "flow_matching_loss": 0.010203879047185182, + "grad_norm": 0.07903089374303818, + "kept_token_count": 573.60546875, + "learning_rate": 1.4353483339058498e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9938008904457092, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00619911290705204, + "mgd_loss_weight": 0.1334877833724022, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.76067810058595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.521826171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26000, + "valid_token_count": 818.54453125 + }, + { + "actual_token_mask_ratio": 0.30090962052345277, + "epoch": 7.5676462030840845, + "flow_matching_loss": 0.009563058475032448, + "grad_norm": 0.06600578129291534, + "kept_token_count": 572.07265625, + "learning_rate": 1.4282983086746498e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9936483442783356, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00635165018029511, + "mgd_loss_weight": 0.1331544414162636, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.9033966064453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3079437255859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26010, + "valid_token_count": 818.3078125 + }, + { + "actual_token_mask_ratio": 0.3002181380987167, + "epoch": 7.570555717195228, + "flow_matching_loss": 0.00888048498891294, + "grad_norm": 0.058036789298057556, + "kept_token_count": 572.61953125, + "learning_rate": 1.4212647743407325e-06, + "loss": 0.0097, + "mgd_cos_sim": 0.9940410137176514, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005958999553695321, + "mgd_loss_weight": 0.13282109051942825, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.89426574707034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.80711975097654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26020, + "valid_token_count": 818.28828125 + }, + { + "actual_token_mask_ratio": 0.30013019442558286, + "epoch": 7.573465231306372, + "flow_matching_loss": 0.010104168253019453, + "grad_norm": 0.07454611361026764, + "kept_token_count": 572.68359375, + "learning_rate": 1.4142477394505065e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9942057311534882, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005794279044494033, + "mgd_loss_weight": 0.13248775154352188, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.7407257080078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2835174560547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26030, + "valid_token_count": 818.27109375 + }, + { + "actual_token_mask_ratio": 0.29941472709178923, + "epoch": 7.576374745417516, + "flow_matching_loss": 0.010051826294511557, + "grad_norm": 0.09366306662559509, + "kept_token_count": 573.3421875, + "learning_rate": 1.4072472125303299e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.993744534254074, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006255465932190419, + "mgd_loss_weight": 0.1321544051170349, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.4378387451172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.43934020996096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26040, + "valid_token_count": 818.375 + }, + { + "actual_token_mask_ratio": 0.29967611134052274, + "epoch": 7.5792842595286585, + "flow_matching_loss": 0.009913302026689053, + "grad_norm": 0.07004699110984802, + "kept_token_count": 573.04921875, + "learning_rate": 1.4002632020865031e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9937992036342621, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006200785655528307, + "mgd_loss_weight": 0.13182105869054794, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.81512756347655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.41387329101565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26050, + "valid_token_count": 818.26640625 + }, + { + "actual_token_mask_ratio": 0.2999077320098877, + "epoch": 7.582193773639802, + "flow_matching_loss": 0.010010154033079744, + "grad_norm": 0.07178133726119995, + "kept_token_count": 572.90390625, + "learning_rate": 1.393295716605258e-06, + "loss": 0.0108, + "mgd_cos_sim": 0.993634307384491, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006365682138130069, + "mgd_loss_weight": 0.13148771971464157, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.9048278808594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.35911254882814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26060, + "valid_token_count": 818.33359375 + }, + { + "actual_token_mask_ratio": 0.3000533491373062, + "epoch": 7.585103287750946, + "flow_matching_loss": 0.010135544231161475, + "grad_norm": 0.06381040811538696, + "kept_token_count": 572.85703125, + "learning_rate": 1.386344764552742e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9939331352710724, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006066875206306577, + "mgd_loss_weight": 0.13115436881780623, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.9207244873047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1326080322266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26070, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.300344318151474, + "epoch": 7.588012801862089, + "flow_matching_loss": 0.009126218408346176, + "grad_norm": 0.06555306911468506, + "kept_token_count": 572.40703125, + "learning_rate": 1.3794103543750246e-06, + "loss": 0.0099, + "mgd_cos_sim": 0.9941390573978424, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005860947910696268, + "mgd_loss_weight": 0.13082102686166763, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.9768005371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.03020629882815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26080, + "valid_token_count": 818.13359375 + }, + { + "actual_token_mask_ratio": 0.29979732632637024, + "epoch": 7.590922315973232, + "flow_matching_loss": 0.00886404449120164, + "grad_norm": 0.057498421519994736, + "kept_token_count": 572.96796875, + "learning_rate": 1.3724924944980637e-06, + "loss": 0.0096, + "mgd_cos_sim": 0.9941105484962464, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005889451736584306, + "mgd_loss_weight": 0.13048768788576126, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.5452819824219, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4463348388672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26090, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.30009130835533143, + "epoch": 7.593831830084376, + "flow_matching_loss": 0.008975555188953876, + "grad_norm": 0.05702831968665123, + "kept_token_count": 572.6171875, + "learning_rate": 1.365591193327712e-06, + "loss": 0.0097, + "mgd_cos_sim": 0.9940695524215698, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005930450791493058, + "mgd_loss_weight": 0.1301543340086937, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 423.1136474609375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.03675537109376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26100, + "valid_token_count": 818.134375 + }, + { + "actual_token_mask_ratio": 0.2993889480829239, + "epoch": 7.59674134419552, + "flow_matching_loss": 0.010815661819651722, + "grad_norm": 0.061680927872657776, + "kept_token_count": 573.13125, + "learning_rate": 1.3587064592496984e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9941490828990937, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005850912630558014, + "mgd_loss_weight": 0.12982099503278732, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.0190185546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5865417480469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26110, + "valid_token_count": 818.04453125 + }, + { + "actual_token_mask_ratio": 0.29970974326133726, + "epoch": 7.599650858306663, + "flow_matching_loss": 0.011322373151779174, + "grad_norm": 0.06339993327856064, + "kept_token_count": 573.0390625, + "learning_rate": 1.3518383006296292e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9938767433166504, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006123265158385039, + "mgd_loss_weight": 0.12948765158653258, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.7864593505859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1139404296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26120, + "valid_token_count": 818.290625 + }, + { + "actual_token_mask_ratio": 0.2991960793733597, + "epoch": 7.602560372417806, + "flow_matching_loss": 0.008970247069373726, + "grad_norm": 0.07437415421009064, + "kept_token_count": 573.48046875, + "learning_rate": 1.344986725812959e-06, + "loss": 0.0097, + "mgd_cos_sim": 0.9943765282630921, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005623477743938565, + "mgd_loss_weight": 0.12915430217981339, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.2677062988281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2549133300781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26130, + "valid_token_count": 818.3171875 + }, + { + "actual_token_mask_ratio": 0.3009042114019394, + "epoch": 7.605469886528949, + "flow_matching_loss": 0.00915148127824068, + "grad_norm": 0.058664027601480484, + "kept_token_count": 572.11640625, + "learning_rate": 1.3381517431250046e-06, + "loss": 0.0099, + "mgd_cos_sim": 0.9942901849746704, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005709807900711894, + "mgd_loss_weight": 0.128820963203907, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.9585723876953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.30426330566405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26140, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.300120809674263, + "epoch": 7.608379400640093, + "flow_matching_loss": 0.010295244865119457, + "grad_norm": 0.05573809891939163, + "kept_token_count": 572.84453125, + "learning_rate": 1.3313333608709095e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.9940858542919159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005914154509082436, + "mgd_loss_weight": 0.12848761528730393, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.52024536132814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8354919433594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26150, + "valid_token_count": 818.490625 + }, + { + "actual_token_mask_ratio": 0.30058004558086393, + "epoch": 7.611288914751237, + "flow_matching_loss": 0.010165241546928883, + "grad_norm": 0.09018033742904663, + "kept_token_count": 572.23046875, + "learning_rate": 1.3245315873356528e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9939880490303039, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006011962238699198, + "mgd_loss_weight": 0.12815427035093307, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.01056213378905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.58412475585936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26160, + "valid_token_count": 818.14921875 + }, + { + "actual_token_mask_ratio": 0.29921122193336486, + "epoch": 7.61419842886238, + "flow_matching_loss": 0.009314713766798377, + "grad_norm": 0.10424046963453293, + "kept_token_count": 573.45078125, + "learning_rate": 1.3177464307840304e-06, + "loss": 0.0101, + "mgd_cos_sim": 0.9939621984958649, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006037798197939992, + "mgd_loss_weight": 0.1278209313750267, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.6851043701172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.2930511474609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26170, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.30018622875213624, + "epoch": 7.617107942973523, + "flow_matching_loss": 0.00961956693790853, + "grad_norm": 0.0785268023610115, + "kept_token_count": 572.8078125, + "learning_rate": 1.310977899460642e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9937819838523865, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006218014284968376, + "mgd_loss_weight": 0.12748757898807525, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.34002990722655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9417266845703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26180, + "valid_token_count": 818.51640625 + }, + { + "actual_token_mask_ratio": 0.29993943870067596, + "epoch": 7.620017457084667, + "flow_matching_loss": 0.010221201879903673, + "grad_norm": 0.05419721081852913, + "kept_token_count": 572.940625, + "learning_rate": 1.3042260015899006e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9939391195774079, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006060882098972797, + "mgd_loss_weight": 0.12715423852205276, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.1934356689453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.28359069824216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26190, + "valid_token_count": 818.41328125 + }, + { + "actual_token_mask_ratio": 0.3003739893436432, + "epoch": 7.6229269711958105, + "flow_matching_loss": 0.011110199755057693, + "grad_norm": 0.07966816425323486, + "kept_token_count": 572.471875, + "learning_rate": 1.2974907453759948e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9940214693546295, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059785332530736925, + "mgd_loss_weight": 0.12682089805603028, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.51976623535154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6079956054688, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26200, + "valid_token_count": 818.2515625 + }, + { + "actual_token_mask_ratio": 0.30041420459747314, + "epoch": 7.625836485306953, + "flow_matching_loss": 0.011099659278988839, + "grad_norm": 0.07014495134353638, + "kept_token_count": 572.4015625, + "learning_rate": 1.2907721390028937e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9940427243709564, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005957268271595239, + "mgd_loss_weight": 0.12648754566907883, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.96666259765624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.23130493164064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26210, + "valid_token_count": 818.196875 + }, + { + "actual_token_mask_ratio": 0.3002326816320419, + "epoch": 7.628745999418097, + "flow_matching_loss": 0.010731024015694857, + "grad_norm": 0.07327164709568024, + "kept_token_count": 572.68359375, + "learning_rate": 1.2840701906343417e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9938202202320099, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00617975746281445, + "mgd_loss_weight": 0.12615420669317245, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.5764129638672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9230499267578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26220, + "valid_token_count": 818.38828125 + }, + { + "actual_token_mask_ratio": 0.3004931896924973, + "epoch": 7.631655513529241, + "flow_matching_loss": 0.01070245886221528, + "grad_norm": 0.07132497429847717, + "kept_token_count": 572.42734375, + "learning_rate": 1.2773849084138312e-06, + "loss": 0.0115, + "mgd_cos_sim": 0.9939022362232208, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006097762752324343, + "mgd_loss_weight": 0.1258208617568016, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.8346801757813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8572509765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26230, + "valid_token_count": 818.328125 + }, + { + "actual_token_mask_ratio": 0.30001622140407563, + "epoch": 7.6345650276403845, + "flow_matching_loss": 0.009015681222081184, + "grad_norm": 0.07440773397684097, + "kept_token_count": 572.73046875, + "learning_rate": 1.2707163004646177e-06, + "loss": 0.0098, + "mgd_cos_sim": 0.9940685391426086, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005931456666439771, + "mgd_loss_weight": 0.12548751384019852, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.6804626464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0988525390625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26240, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.2999728262424469, + "epoch": 7.637474541751527, + "flow_matching_loss": 0.009572496917098761, + "grad_norm": 0.08755095303058624, + "kept_token_count": 572.775, + "learning_rate": 1.2640643748896879e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9938615500926972, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006138452794402838, + "mgd_loss_weight": 0.12515417486429214, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.47300415039064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.53724365234376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26250, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.3001550167798996, + "epoch": 7.640384055862671, + "flow_matching_loss": 0.009493451612070204, + "grad_norm": 0.04876488074660301, + "kept_token_count": 572.63203125, + "learning_rate": 1.257429139771753e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9938540875911712, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006145915715023875, + "mgd_loss_weight": 0.12482082843780518, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.86900024414064, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.65956726074216, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26260, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.3007435142993927, + "epoch": 7.643293569973815, + "flow_matching_loss": 0.009692127676680684, + "grad_norm": 0.061106257140636444, + "kept_token_count": 572.19453125, + "learning_rate": 1.2508106031732553e-06, + "loss": 0.0105, + "mgd_cos_sim": 0.9938926696777344, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006107335491105914, + "mgd_loss_weight": 0.1244874820113182, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.2398254394531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3591247558594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26270, + "valid_token_count": 818.2921875 + }, + { + "actual_token_mask_ratio": 0.29934062361717223, + "epoch": 7.6462030840849575, + "flow_matching_loss": 0.009439601236954332, + "grad_norm": 0.06514393538236618, + "kept_token_count": 573.2140625, + "learning_rate": 1.2442087731363366e-06, + "loss": 0.0102, + "mgd_cos_sim": 0.993638277053833, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063617265783250335, + "mgd_loss_weight": 0.12415413782000542, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.55340270996095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.16322937011716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26280, + "valid_token_count": 818.10390625 + }, + { + "actual_token_mask_ratio": 0.2995433181524277, + "epoch": 7.649112598196101, + "flow_matching_loss": 0.01042864639312029, + "grad_norm": 0.07979927211999893, + "kept_token_count": 573.3, + "learning_rate": 1.237623657682847e-06, + "loss": 0.0112, + "mgd_cos_sim": 0.9940446376800537, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005955361109226942, + "mgd_loss_weight": 0.12382079437375068, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.3841949462891, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 376.8711730957031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26290, + "valid_token_count": 818.46953125 + }, + { + "actual_token_mask_ratio": 0.3005124717950821, + "epoch": 7.652022112307245, + "flow_matching_loss": 0.01000648457556963, + "grad_norm": 0.0805821493268013, + "kept_token_count": 572.4921875, + "learning_rate": 1.2310552648143208e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9942144691944123, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005785526847466827, + "mgd_loss_weight": 0.1234874501824379, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.59715576171874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.21356506347655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26300, + "valid_token_count": 818.4453125 + }, + { + "actual_token_mask_ratio": 0.30009637475013734, + "epoch": 7.654931626418388, + "flow_matching_loss": 0.009500230988487601, + "grad_norm": 0.07439962029457092, + "kept_token_count": 572.715625, + "learning_rate": 1.2245036025119738e-06, + "loss": 0.0102, + "mgd_cos_sim": 0.9940873861312867, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005912625137716532, + "mgd_loss_weight": 0.12315410450100898, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.8280487060547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.57237548828124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26310, + "valid_token_count": 818.28125 + }, + { + "actual_token_mask_ratio": 0.2995452553033829, + "epoch": 7.6578411405295315, + "flow_matching_loss": 0.009930737502872943, + "grad_norm": 0.0903518870472908, + "kept_token_count": 573.22265625, + "learning_rate": 1.2179686787366962e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.9944447159767151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005555288447067141, + "mgd_loss_weight": 0.1228207603096962, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 426.7810821533203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.6152587890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26320, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.29941740334033967, + "epoch": 7.660750654640675, + "flow_matching_loss": 0.010940278926864267, + "grad_norm": 0.10408370941877365, + "kept_token_count": 573.31953125, + "learning_rate": 1.2114505014290329e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9938195168972015, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006180478259921074, + "mgd_loss_weight": 0.12248741686344147, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.2571533203125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2347015380859, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26330, + "valid_token_count": 818.346875 + }, + { + "actual_token_mask_ratio": 0.29983003437519073, + "epoch": 7.663660168751818, + "flow_matching_loss": 0.011131684482097625, + "grad_norm": 0.07764046639204025, + "kept_token_count": 573.03515625, + "learning_rate": 1.2049490785091833e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9938384413719177, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00616156505420804, + "mgd_loss_weight": 0.12215407192707062, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.2170104980469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9784210205078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26340, + "valid_token_count": 818.4234375 + }, + { + "actual_token_mask_ratio": 0.29996230602264407, + "epoch": 7.666569682862962, + "flow_matching_loss": 0.011344062630087137, + "grad_norm": 0.08257880061864853, + "kept_token_count": 572.9390625, + "learning_rate": 1.1984644178769949e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9935667932033538, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006433218205347657, + "mgd_loss_weight": 0.12182072699069976, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.91090393066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6613830566406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26350, + "valid_token_count": 818.43828125 + }, + { + "actual_token_mask_ratio": 0.30047890841960906, + "epoch": 7.669479196974105, + "flow_matching_loss": 0.009821869758889079, + "grad_norm": 0.06781774759292603, + "kept_token_count": 572.55390625, + "learning_rate": 1.1919965274119381e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.994003564119339, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059964414685964584, + "mgd_loss_weight": 0.12148738279938698, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.7768615722656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3989196777344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26360, + "valid_token_count": 818.4953125 + }, + { + "actual_token_mask_ratio": 0.29957208335399627, + "epoch": 7.672388711085249, + "flow_matching_loss": 0.009889694768935442, + "grad_norm": 0.10837927460670471, + "kept_token_count": 573.0953125, + "learning_rate": 1.18554541497311e-06, + "loss": 0.0106, + "mgd_cos_sim": 0.9940941274166107, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005905864713713527, + "mgd_loss_weight": 0.12115403935313225, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.8049560546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1927215576172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26370, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.29876416325569155, + "epoch": 7.675298225196392, + "flow_matching_loss": 0.010170472040772437, + "grad_norm": 0.07406127452850342, + "kept_token_count": 573.775, + "learning_rate": 1.1791110883992213e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9941113948822021, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005888605862855912, + "mgd_loss_weight": 0.12082069367170334, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.6308837890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.20489196777345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26380, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.30052644312381743, + "epoch": 7.678207739307536, + "flow_matching_loss": 0.009688736032694578, + "grad_norm": 0.06085342913866043, + "kept_token_count": 572.44921875, + "learning_rate": 1.1726935555085817e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9942377507686615, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00576223754324019, + "mgd_loss_weight": 0.12048734873533248, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 428.4958099365234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1081024169922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26390, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.2997952789068222, + "epoch": 7.681117253418679, + "flow_matching_loss": 0.010239403788000345, + "grad_norm": 0.07237427681684494, + "kept_token_count": 572.959375, + "learning_rate": 1.166292824099105e-06, + "loss": 0.011, + "mgd_cos_sim": 0.9936060845851898, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063939311541616915, + "mgd_loss_weight": 0.12015400528907776, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.6294647216797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6327697753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26400, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.2994945704936981, + "epoch": 7.684026767529822, + "flow_matching_loss": 0.009201107546687126, + "grad_norm": 0.057281430810689926, + "kept_token_count": 573.32109375, + "learning_rate": 1.1599089019482768e-06, + "loss": 0.0099, + "mgd_cos_sim": 0.9938552856445313, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006144712911918759, + "mgd_loss_weight": 0.11982066184282303, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.1598205566406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.81285705566404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26410, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.29978181719779967, + "epoch": 7.686936281640966, + "flow_matching_loss": 0.011008877959102392, + "grad_norm": 0.06921914219856262, + "kept_token_count": 573.10390625, + "learning_rate": 1.1535417968131718e-06, + "loss": 0.0117, + "mgd_cos_sim": 0.9938410580158233, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006158948550000787, + "mgd_loss_weight": 0.11948731541633606, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.9743133544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8415802001953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26420, + "valid_token_count": 818.46484375 + }, + { + "actual_token_mask_ratio": 0.30053838789463044, + "epoch": 7.68984579575211, + "flow_matching_loss": 0.009989190148189664, + "grad_norm": 0.06646915525197983, + "kept_token_count": 572.18046875, + "learning_rate": 1.147191516430418e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9941030740737915, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0058969234116375445, + "mgd_loss_weight": 0.11915397122502328, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.9922760009766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.09270629882815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26430, + "valid_token_count": 818.03046875 + }, + { + "actual_token_mask_ratio": 0.3003827005624771, + "epoch": 7.692755309863253, + "flow_matching_loss": 0.009172415174543858, + "grad_norm": 0.06510555744171143, + "kept_token_count": 572.46953125, + "learning_rate": 1.140858068516205e-06, + "loss": 0.0099, + "mgd_cos_sim": 0.9938227832317352, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006177217699587345, + "mgd_loss_weight": 0.11882062777876853, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.45714111328124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.86826477050784, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26440, + "valid_token_count": 818.265625 + }, + { + "actual_token_mask_ratio": 0.30053088068962097, + "epoch": 7.695664823974396, + "flow_matching_loss": 0.010149935074150563, + "grad_norm": 0.07090956717729568, + "kept_token_count": 572.27421875, + "learning_rate": 1.1345414607662736e-06, + "loss": 0.0109, + "mgd_cos_sim": 0.9939082205295563, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006091783102601766, + "mgd_loss_weight": 0.11848728358745575, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.2236083984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8862243652344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26450, + "valid_token_count": 818.15390625 + }, + { + "actual_token_mask_ratio": 0.2997361570596695, + "epoch": 7.69857433808554, + "flow_matching_loss": 0.010858549643307925, + "grad_norm": 0.056831471621990204, + "kept_token_count": 573.00078125, + "learning_rate": 1.1282417008558937e-06, + "loss": 0.0116, + "mgd_cos_sim": 0.9937830150127411, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006216968689113855, + "mgd_loss_weight": 0.11815393790602684, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.3275207519531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0044860839844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26460, + "valid_token_count": 818.26484375 + }, + { + "actual_token_mask_ratio": 0.30028087496757505, + "epoch": 7.7014838521966835, + "flow_matching_loss": 0.008530229376628996, + "grad_norm": 0.06846863031387329, + "kept_token_count": 572.46953125, + "learning_rate": 1.1219587964398674e-06, + "loss": 0.0092, + "mgd_cos_sim": 0.993994665145874, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006005324702709913, + "mgd_loss_weight": 0.11782059371471405, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.3667236328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5442779541016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26470, + "valid_token_count": 818.1453125 + }, + { + "actual_token_mask_ratio": 0.30036530196666716, + "epoch": 7.704393366307826, + "flow_matching_loss": 0.011206195782870054, + "grad_norm": 0.08630924671888351, + "kept_token_count": 572.36484375, + "learning_rate": 1.1156927551525198e-06, + "loss": 0.0119, + "mgd_cos_sim": 0.9939789235591888, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006021088035777211, + "mgd_loss_weight": 0.11748725026845933, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.14003295898436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.62782592773436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26480, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.2997298717498779, + "epoch": 7.70730288041897, + "flow_matching_loss": 0.010050677042454482, + "grad_norm": 0.07744117081165314, + "kept_token_count": 572.96171875, + "learning_rate": 1.109443584607675e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9941230714321136, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005876936158165336, + "mgd_loss_weight": 0.11715390533208847, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.31337890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.53619995117185, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26490, + "valid_token_count": 818.20078125 + }, + { + "actual_token_mask_ratio": 0.3003718703985214, + "epoch": 7.710212394530114, + "flow_matching_loss": 0.010058103268966079, + "grad_norm": 0.08180331438779831, + "kept_token_count": 572.55859375, + "learning_rate": 1.1032112923986708e-06, + "loss": 0.0108, + "mgd_cos_sim": 0.9936765730381012, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006323425378650427, + "mgd_loss_weight": 0.11682055965065956, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.295751953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8902557373047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26500, + "valid_token_count": 818.3765625 + }, + { + "actual_token_mask_ratio": 0.3006025731563568, + "epoch": 7.713121908641257, + "flow_matching_loss": 0.008754371153190731, + "grad_norm": 0.06324496865272522, + "kept_token_count": 572.103125, + "learning_rate": 1.0969958860983299e-06, + "loss": 0.0094, + "mgd_cos_sim": 0.9940428614616394, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059571424033492805, + "mgd_loss_weight": 0.11648721620440483, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.3014373779297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.7595916748047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26510, + "valid_token_count": 818.0 + }, + { + "actual_token_mask_ratio": 0.3001260578632355, + "epoch": 7.7160314227524, + "flow_matching_loss": 0.01039242879487574, + "grad_norm": 0.06901992857456207, + "kept_token_count": 572.66796875, + "learning_rate": 1.0907973732589537e-06, + "loss": 0.0111, + "mgd_cos_sim": 0.994030886888504, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005969117116183043, + "mgd_loss_weight": 0.1161538727581501, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.7655517578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.77857666015626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26520, + "valid_token_count": 818.2453125 + }, + { + "actual_token_mask_ratio": 0.2984771549701691, + "epoch": 7.718940936863544, + "flow_matching_loss": 0.011397866625338792, + "grad_norm": 0.07061687111854553, + "kept_token_count": 573.8359375, + "learning_rate": 1.0846157614123215e-06, + "loss": 0.0121, + "mgd_cos_sim": 0.9939518749713898, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006048132246360183, + "mgd_loss_weight": 0.11582052707672119, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.05393676757814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3262084960937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26530, + "valid_token_count": 817.98984375 + }, + { + "actual_token_mask_ratio": 0.2996458142995834, + "epoch": 7.721850450974687, + "flow_matching_loss": 0.009971797419711947, + "grad_norm": 0.07347835600376129, + "kept_token_count": 573.05859375, + "learning_rate": 1.078451058069671e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9940537631511688, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005946244858205318, + "mgd_loss_weight": 0.11548718214035034, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.36448364257814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.00315856933594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26540, + "valid_token_count": 818.24453125 + }, + { + "actual_token_mask_ratio": 0.30087631940841675, + "epoch": 7.7247599650858305, + "flow_matching_loss": 0.009089418407529592, + "grad_norm": 0.07666988670825958, + "kept_token_count": 572.20703125, + "learning_rate": 1.0723032707217056e-06, + "loss": 0.0098, + "mgd_cos_sim": 0.993973171710968, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006026821350678802, + "mgd_loss_weight": 0.11515383869409561, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.7333221435547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7036865234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26550, + "valid_token_count": 818.46015625 + }, + { + "actual_token_mask_ratio": 0.3000603049993515, + "epoch": 7.727669479196974, + "flow_matching_loss": 0.010043029580265284, + "grad_norm": 0.07844694703817368, + "kept_token_count": 572.7203125, + "learning_rate": 1.066172406838558e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.9944457590579987, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0055542378686368465, + "mgd_loss_weight": 0.11482049524784088, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.79953918457034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.96148681640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26560, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.3001507997512817, + "epoch": 7.730578993308118, + "flow_matching_loss": 0.010707445302978158, + "grad_norm": 0.06680604070425034, + "kept_token_count": 572.7234375, + "learning_rate": 1.0600584738698127e-06, + "loss": 0.0114, + "mgd_cos_sim": 0.9941043853759766, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005895619513466954, + "mgd_loss_weight": 0.11448714882135391, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.0563018798828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4351348876953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26570, + "valid_token_count": 818.34765625 + }, + { + "actual_token_mask_ratio": 0.29986720383167265, + "epoch": 7.733488507419261, + "flow_matching_loss": 0.008785658236593007, + "grad_norm": 0.07901802659034729, + "kept_token_count": 572.83359375, + "learning_rate": 1.0539614792444735e-06, + "loss": 0.0095, + "mgd_cos_sim": 0.9940966546535492, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005903347255662083, + "mgd_loss_weight": 0.11415380463004113, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.68573303222655, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.01011657714844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26580, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.2994944930076599, + "epoch": 7.7363980215304045, + "flow_matching_loss": 0.009579211939126253, + "grad_norm": 0.08287423849105835, + "kept_token_count": 573.1734375, + "learning_rate": 1.0478814303709627e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9936888158321381, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006311180489137769, + "mgd_loss_weight": 0.11382046118378639, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.52477111816404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4209228515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26590, + "valid_token_count": 818.2296875 + }, + { + "actual_token_mask_ratio": 0.30021469593048095, + "epoch": 7.739307535641548, + "flow_matching_loss": 0.009687727037817239, + "grad_norm": 0.0641300156712532, + "kept_token_count": 572.5734375, + "learning_rate": 1.041818334637109e-06, + "loss": 0.0104, + "mgd_cos_sim": 0.9940558731555938, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005944125959649682, + "mgd_loss_weight": 0.1134871169924736, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.4277069091797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.6125732421875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26600, + "valid_token_count": 818.20859375 + }, + { + "actual_token_mask_ratio": 0.3000773310661316, + "epoch": 7.742217049752691, + "flow_matching_loss": 0.009979551844298839, + "grad_norm": 0.07333307713270187, + "kept_token_count": 572.69453125, + "learning_rate": 1.0357721994101526e-06, + "loss": 0.0107, + "mgd_cos_sim": 0.99406298995018, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005937004322186112, + "mgd_loss_weight": 0.11315377056598663, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.00047302246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.4021301269531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26610, + "valid_token_count": 818.22890625 + }, + { + "actual_token_mask_ratio": 0.30073363780975343, + "epoch": 7.745126563863835, + "flow_matching_loss": 0.009248449560254812, + "grad_norm": 0.06391684710979462, + "kept_token_count": 572.10546875, + "learning_rate": 1.0297430320367135e-06, + "loss": 0.0099, + "mgd_cos_sim": 0.9942808091640473, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005719196377322077, + "mgd_loss_weight": 0.1128204271197319, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.4337554931641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.66831970214844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26620, + "valid_token_count": 818.14921875 + }, + { + "actual_token_mask_ratio": 0.3003910332918167, + "epoch": 7.748036077974978, + "flow_matching_loss": 0.011634607613086701, + "grad_norm": 0.08101952075958252, + "kept_token_count": 572.48828125, + "learning_rate": 1.0237308398428014e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9939713716506958, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006028626346960664, + "mgd_loss_weight": 0.11248708367347718, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.27768859863284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7445343017578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26630, + "valid_token_count": 818.29609375 + }, + { + "actual_token_mask_ratio": 0.29977667927742, + "epoch": 7.750945592086122, + "flow_matching_loss": 0.009630817268043757, + "grad_norm": 0.071883425116539, + "kept_token_count": 573.03828125, + "learning_rate": 1.0177356301337925e-06, + "loss": 0.0103, + "mgd_cos_sim": 0.9940917074680329, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005908297747373581, + "mgd_loss_weight": 0.11215373873710632, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.2665740966797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.21660766601565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26640, + "valid_token_count": 818.36171875 + }, + { + "actual_token_mask_ratio": 0.3006525993347168, + "epoch": 7.753855106197265, + "flow_matching_loss": 0.009808247210457921, + "grad_norm": 0.12477325648069382, + "kept_token_count": 572.271875, + "learning_rate": 1.0117574101944348e-06, + "loss": 0.0105, + "mgd_cos_sim": 0.993739104270935, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006260888651013374, + "mgd_loss_weight": 0.11182039305567741, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.1510833740234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8240020751953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26650, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.3003187537193298, + "epoch": 7.756764620308409, + "flow_matching_loss": 0.01161624165251851, + "grad_norm": 0.09423435479402542, + "kept_token_count": 572.37421875, + "learning_rate": 1.0057961872888293e-06, + "loss": 0.0123, + "mgd_cos_sim": 0.9938105165958404, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006189493834972382, + "mgd_loss_weight": 0.11148704960942268, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4066162109375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9412872314453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26660, + "valid_token_count": 818.0515625 + }, + { + "actual_token_mask_ratio": 0.29992020726203916, + "epoch": 7.7596741344195515, + "flow_matching_loss": 0.01139222914353013, + "grad_norm": 0.06320817768573761, + "kept_token_count": 572.74765625, + "learning_rate": 9.998519686604235e-07, + "loss": 0.0121, + "mgd_cos_sim": 0.9939040243625641, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006095975218340754, + "mgd_loss_weight": 0.11115370616316796, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.4170227050781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3425689697266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26670, + "valid_token_count": 818.115625 + }, + { + "actual_token_mask_ratio": 0.29963430762290955, + "epoch": 7.762583648530695, + "flow_matching_loss": 0.010273638088256121, + "grad_norm": 0.06487379223108292, + "kept_token_count": 573.05078125, + "learning_rate": 9.939247615320047e-07, + "loss": 0.0109, + "mgd_cos_sim": 0.9941810429096222, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005818957928568125, + "mgd_loss_weight": 0.11082036048173904, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.1941619873047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.190625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26680, + "valid_token_count": 818.2140625 + }, + { + "actual_token_mask_ratio": 0.30045635998249054, + "epoch": 7.765493162641839, + "flow_matching_loss": 0.010299950558692216, + "grad_norm": 0.05668337643146515, + "kept_token_count": 572.375, + "learning_rate": 9.880145731056834e-07, + "loss": 0.011, + "mgd_cos_sim": 0.9940844357013703, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059155605267733335, + "mgd_loss_weight": 0.11048701554536819, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.5222503662109, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.73211669921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26690, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.30024746656417844, + "epoch": 7.768402676752983, + "flow_matching_loss": 0.009358563087880611, + "grad_norm": 0.07379001379013062, + "kept_token_count": 572.63359375, + "learning_rate": 9.821214105629034e-07, + "loss": 0.01, + "mgd_cos_sim": 0.994108009338379, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005891980975866318, + "mgd_loss_weight": 0.11015367209911346, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.99298095703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.6650756835937, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26700, + "valid_token_count": 818.33203125 + }, + { + "actual_token_mask_ratio": 0.3001392185688019, + "epoch": 7.771312190864125, + "flow_matching_loss": 0.010206945706158876, + "grad_norm": 0.06338127702474594, + "kept_token_count": 572.78671875, + "learning_rate": 9.762452810644068e-07, + "loss": 0.0109, + "mgd_cos_sim": 0.9939722716808319, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006027730787172914, + "mgd_loss_weight": 0.10982032790780068, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.5736358642578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.04912719726565, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26710, + "valid_token_count": 818.43671875 + }, + { + "actual_token_mask_ratio": 0.29948821663856506, + "epoch": 7.774221704975269, + "flow_matching_loss": 0.009708013338968157, + "grad_norm": 0.06287132203578949, + "kept_token_count": 573.2859375, + "learning_rate": 9.703861917502487e-07, + "loss": 0.0104, + "mgd_cos_sim": 0.994130939245224, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005869065457955003, + "mgd_loss_weight": 0.10948698222637177, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.65232543945314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5892730712891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26720, + "valid_token_count": 818.3859375 + }, + { + "actual_token_mask_ratio": 0.2993608623743057, + "epoch": 7.777131219086413, + "flow_matching_loss": 0.009388872236013413, + "grad_norm": 0.05199216678738594, + "kept_token_count": 573.42265625, + "learning_rate": 9.645441497397744e-07, + "loss": 0.01, + "mgd_cos_sim": 0.9942767739295959, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005723233334720135, + "mgd_loss_weight": 0.10915363803505898, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.7309143066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7947479248047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26730, + "valid_token_count": 818.4328125 + }, + { + "actual_token_mask_ratio": 0.3005278199911118, + "epoch": 7.780040733197556, + "flow_matching_loss": 0.010054887272417545, + "grad_norm": 0.08255058526992798, + "kept_token_count": 572.27734375, + "learning_rate": 9.587191621316154e-07, + "loss": 0.0107, + "mgd_cos_sim": 0.9943983495235443, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005601651733741164, + "mgd_loss_weight": 0.10882029458880424, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.8501342773437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.60625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26740, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.300400373339653, + "epoch": 7.782950247308699, + "flow_matching_loss": 0.009706475771963596, + "grad_norm": 0.06877508759498596, + "kept_token_count": 572.50078125, + "learning_rate": 9.529112360036802e-07, + "loss": 0.0103, + "mgd_cos_sim": 0.9941264629364014, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005873546469956636, + "mgd_loss_weight": 0.10848695039749146, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.86667785644534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6528747558594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26750, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.2998602867126465, + "epoch": 7.785859761419843, + "flow_matching_loss": 0.009747401485219598, + "grad_norm": 0.07163394242525101, + "kept_token_count": 572.86953125, + "learning_rate": 9.471203784131449e-07, + "loss": 0.0104, + "mgd_cos_sim": 0.9938901424407959, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006109867757186293, + "mgd_loss_weight": 0.10815360397100449, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.2635498046875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.1549346923828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26760, + "valid_token_count": 818.21875 + }, + { + "actual_token_mask_ratio": 0.3005607485771179, + "epoch": 7.788769275530987, + "flow_matching_loss": 0.009453083528205753, + "grad_norm": 0.06338860839605331, + "kept_token_count": 572.253125, + "learning_rate": 9.41346596396453e-07, + "loss": 0.0101, + "mgd_cos_sim": 0.9940253436565399, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005974646983668208, + "mgd_loss_weight": 0.10782026052474976, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.502001953125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.29949645996095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26770, + "valid_token_count": 818.15703125 + }, + { + "actual_token_mask_ratio": 0.30143878757953646, + "epoch": 7.79167878964213, + "flow_matching_loss": 0.010615269746631384, + "grad_norm": 0.07862836122512817, + "kept_token_count": 571.58046875, + "learning_rate": 9.355898969692889e-07, + "loss": 0.0113, + "mgd_cos_sim": 0.9938295841217041, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006170426961034536, + "mgd_loss_weight": 0.10748691707849503, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.19507751464846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7563751220703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26780, + "valid_token_count": 818.22265625 + }, + { + "actual_token_mask_ratio": 0.3001790106296539, + "epoch": 7.794588303753273, + "flow_matching_loss": 0.009867267962545157, + "grad_norm": 0.05477432906627655, + "kept_token_count": 572.8078125, + "learning_rate": 9.29850287126589e-07, + "loss": 0.0105, + "mgd_cos_sim": 0.994177120923996, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0058228781446814535, + "mgd_loss_weight": 0.10715357214212418, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.15285339355466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.79039306640624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26790, + "valid_token_count": 818.503125 + }, + { + "actual_token_mask_ratio": 0.2992843508720398, + "epoch": 7.797497817864417, + "flow_matching_loss": 0.01037036399357021, + "grad_norm": 0.08174606412649155, + "kept_token_count": 573.45546875, + "learning_rate": 9.24127773842518e-07, + "loss": 0.011, + "mgd_cos_sim": 0.99394451379776, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006055492023006081, + "mgd_loss_weight": 0.10682022646069526, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.37183837890626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.26344604492186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26800, + "valid_token_count": 818.38359375 + }, + { + "actual_token_mask_ratio": 0.30078292787075045, + "epoch": 7.80040733197556, + "flow_matching_loss": 0.009822205873206258, + "grad_norm": 0.09784334152936935, + "kept_token_count": 572.24453125, + "learning_rate": 9.18422364070467e-07, + "loss": 0.0105, + "mgd_cos_sim": 0.9940533995628357, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005946601927280426, + "mgd_loss_weight": 0.10648688301444054, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.41697998046874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5453643798828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26810, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.29962922632694244, + "epoch": 7.8033168460867035, + "flow_matching_loss": 0.01033829441294074, + "grad_norm": 0.06086823716759682, + "kept_token_count": 572.97109375, + "learning_rate": 9.127340647430538e-07, + "loss": 0.011, + "mgd_cos_sim": 0.9941202104091644, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005879788333550096, + "mgd_loss_weight": 0.10615353882312775, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.4014953613281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0080230712891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26820, + "valid_token_count": 818.0921875 + }, + { + "actual_token_mask_ratio": 0.3000392526388168, + "epoch": 7.806226360197847, + "flow_matching_loss": 0.008939984999597073, + "grad_norm": 0.10376866161823273, + "kept_token_count": 572.84296875, + "learning_rate": 9.070628827720929e-07, + "loss": 0.0096, + "mgd_cos_sim": 0.9941737592220307, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005826251255348325, + "mgd_loss_weight": 0.1058201938867569, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.26087646484376, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.81487426757815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26830, + "valid_token_count": 818.390625 + }, + { + "actual_token_mask_ratio": 0.3010231852531433, + "epoch": 7.80913587430899, + "flow_matching_loss": 0.0103045416995883, + "grad_norm": 0.07694811373949051, + "kept_token_count": 571.93671875, + "learning_rate": 9.014088250486052e-07, + "loss": 0.011, + "mgd_cos_sim": 0.9937291741371155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006270811334252358, + "mgd_loss_weight": 0.10548684895038604, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.09961853027346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.58307189941405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26840, + "valid_token_count": 818.24921875 + }, + { + "actual_token_mask_ratio": 0.3000237911939621, + "epoch": 7.812045388420134, + "flow_matching_loss": 0.009321485087275505, + "grad_norm": 0.052915867418050766, + "kept_token_count": 572.82734375, + "learning_rate": 8.95771898442807e-07, + "loss": 0.0099, + "mgd_cos_sim": 0.9941214025020599, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005878595402464271, + "mgd_loss_weight": 0.10515350550413131, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.1500274658203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.7088897705078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26850, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.30016697347164156, + "epoch": 7.8149549025312774, + "flow_matching_loss": 0.009720451664179564, + "grad_norm": 0.07378201931715012, + "kept_token_count": 572.640625, + "learning_rate": 8.90152109804091e-07, + "loss": 0.0103, + "mgd_cos_sim": 0.994074010848999, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005926010105758906, + "mgd_loss_weight": 0.10482016131281853, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.1636047363281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.42372436523436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26860, + "valid_token_count": 818.25234375 + }, + { + "actual_token_mask_ratio": 0.29967454075813293, + "epoch": 7.81786441664242, + "flow_matching_loss": 0.009643526375293731, + "grad_norm": 0.06800888478755951, + "kept_token_count": 573.16953125, + "learning_rate": 8.845494659610365e-07, + "loss": 0.0103, + "mgd_cos_sim": 0.9940309703350068, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005969041772186756, + "mgd_loss_weight": 0.10448681563138962, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.61304931640626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3006622314453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26870, + "valid_token_count": 818.43671875 + }, + { + "actual_token_mask_ratio": 0.30054168701171874, + "epoch": 7.820773930753564, + "flow_matching_loss": 0.009598847385495901, + "grad_norm": 0.057000577449798584, + "kept_token_count": 572.47265625, + "learning_rate": 8.789639737213812e-07, + "loss": 0.0102, + "mgd_cos_sim": 0.9940044462680817, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005995561927556991, + "mgd_loss_weight": 0.10415347144007683, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.3061798095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.5680877685547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26880, + "valid_token_count": 818.45234375 + }, + { + "actual_token_mask_ratio": 0.2998322457075119, + "epoch": 7.823683444864708, + "flow_matching_loss": 0.011550391372293235, + "grad_norm": 0.11456876993179321, + "kept_token_count": 573.09765625, + "learning_rate": 8.733956398720244e-07, + "loss": 0.0122, + "mgd_cos_sim": 0.9939189493656159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006081044720485806, + "mgd_loss_weight": 0.10382012799382209, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.32762145996094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7093811035156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26890, + "valid_token_count": 818.51328125 + }, + { + "actual_token_mask_ratio": 0.29925361573696135, + "epoch": 7.826592958975851, + "flow_matching_loss": 0.00893667726777494, + "grad_norm": 0.05194089934229851, + "kept_token_count": 573.340625, + "learning_rate": 8.678444711790157e-07, + "loss": 0.0096, + "mgd_cos_sim": 0.9939764261245727, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00602356162853539, + "mgd_loss_weight": 0.10348678380250931, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.97447814941404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.289208984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26900, + "valid_token_count": 818.18671875 + }, + { + "actual_token_mask_ratio": 0.29948855340480807, + "epoch": 7.829502473086994, + "flow_matching_loss": 0.010781247075647116, + "grad_norm": 0.08574552834033966, + "kept_token_count": 573.17578125, + "learning_rate": 8.623104743875493e-07, + "loss": 0.0114, + "mgd_cos_sim": 0.9938802421092987, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006119777215644717, + "mgd_loss_weight": 0.10315343737602234, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.77260131835936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.6113739013672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26910, + "valid_token_count": 818.225 + }, + { + "actual_token_mask_ratio": 0.2994265168905258, + "epoch": 7.832411987198138, + "flow_matching_loss": 0.009018037375062704, + "grad_norm": 0.07025308907032013, + "kept_token_count": 573.328125, + "learning_rate": 8.567936562219547e-07, + "loss": 0.0096, + "mgd_cos_sim": 0.9942450225353241, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005754971830174327, + "mgd_loss_weight": 0.10282009392976761, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.372314453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9371337890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26920, + "valid_token_count": 818.36953125 + }, + { + "actual_token_mask_ratio": 0.2999707579612732, + "epoch": 7.835321501309282, + "flow_matching_loss": 0.009833498299121857, + "grad_norm": 0.060772016644477844, + "kept_token_count": 572.684375, + "learning_rate": 8.512940233856847e-07, + "loss": 0.0105, + "mgd_cos_sim": 0.9939460635185242, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006053941929712892, + "mgd_loss_weight": 0.10248674973845481, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.92457275390626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4607177734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26930, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.29982784688472747, + "epoch": 7.838231015420424, + "flow_matching_loss": 0.010464094253256917, + "grad_norm": 0.07431396842002869, + "kept_token_count": 572.90546875, + "learning_rate": 8.458115825613122e-07, + "loss": 0.0111, + "mgd_cos_sim": 0.9940561652183533, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005943831149488687, + "mgd_loss_weight": 0.10215340554714203, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.3880187988281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0364166259766, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26940, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.29984001219272616, + "epoch": 7.841140529531568, + "flow_matching_loss": 0.010375806270167232, + "grad_norm": 0.06175671145319939, + "kept_token_count": 572.8375, + "learning_rate": 8.403463404105199e-07, + "loss": 0.011, + "mgd_cos_sim": 0.9940688788890839, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005931127071380615, + "mgd_loss_weight": 0.10182005986571312, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.3929718017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.50108337402344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26950, + "valid_token_count": 818.1515625 + }, + { + "actual_token_mask_ratio": 0.3007077515125275, + "epoch": 7.844050043642712, + "flow_matching_loss": 0.010693209432065486, + "grad_norm": 0.09033224731683731, + "kept_token_count": 572.15078125, + "learning_rate": 8.348983035740926e-07, + "loss": 0.0113, + "mgd_cos_sim": 0.993714326620102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006285665230825544, + "mgd_loss_weight": 0.10148671641945839, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.9320983886719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.852587890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26960, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.29979753494262695, + "epoch": 7.846959557753856, + "flow_matching_loss": 0.010496333381161094, + "grad_norm": 0.06324920058250427, + "kept_token_count": 572.91953125, + "learning_rate": 8.294674786719064e-07, + "loss": 0.0111, + "mgd_cos_sim": 0.9938705503940582, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006129448208957911, + "mgd_loss_weight": 0.1011533722281456, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.59263610839844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.76665649414065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26970, + "valid_token_count": 818.2171875 + }, + { + "actual_token_mask_ratio": 0.2999876856803894, + "epoch": 7.849869071864998, + "flow_matching_loss": 0.009569356916472316, + "grad_norm": 0.0543217808008194, + "kept_token_count": 572.82890625, + "learning_rate": 8.240538723029295e-07, + "loss": 0.0102, + "mgd_cos_sim": 0.9939915001392364, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006008497392758727, + "mgd_loss_weight": 0.10082002729177475, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.4392944335938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.15159606933594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26980, + "valid_token_count": 818.31484375 + }, + { + "actual_token_mask_ratio": 0.30024146735668183, + "epoch": 7.852778585976142, + "flow_matching_loss": 0.011147952126339077, + "grad_norm": 0.08258429914712906, + "kept_token_count": 572.68203125, + "learning_rate": 8.186574910451983e-07, + "loss": 0.0118, + "mgd_cos_sim": 0.9937770903110504, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006222907453775406, + "mgd_loss_weight": 0.1004866823554039, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.82672119140625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.633203125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 26990, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.29978452026844027, + "epoch": 7.855688100087286, + "flow_matching_loss": 0.010458363452926278, + "grad_norm": 0.07187816500663757, + "kept_token_count": 572.91796875, + "learning_rate": 8.132783414558326e-07, + "loss": 0.0111, + "mgd_cos_sim": 0.9940852344036102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005914766388013959, + "mgd_loss_weight": 0.10015333890914917, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 424.3408630371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6343078613281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27000, + "valid_token_count": 818.1984375 + }, + { + "actual_token_mask_ratio": 0.2999710589647293, + "epoch": 7.858597614198429, + "flow_matching_loss": 0.009862978430464863, + "grad_norm": 0.04846183583140373, + "kept_token_count": 572.75234375, + "learning_rate": 8.079164300710023e-07, + "loss": 0.0105, + "mgd_cos_sim": 0.9940361380577087, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00596386413089931, + "mgd_loss_weight": 0.09981999471783638, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.8038391113281, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3110656738281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27010, + "valid_token_count": 818.1828125 + }, + { + "actual_token_mask_ratio": 0.3005264699459076, + "epoch": 7.861507128309572, + "flow_matching_loss": 0.010338117461651563, + "grad_norm": 0.06346259266138077, + "kept_token_count": 572.5046875, + "learning_rate": 8.025717634059326e-07, + "loss": 0.0109, + "mgd_cos_sim": 0.99386767745018, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006132323481142521, + "mgd_loss_weight": 0.09948664903640747, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.2071228027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.49524841308596, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27020, + "valid_token_count": 818.4796875 + }, + { + "actual_token_mask_ratio": 0.2999835669994354, + "epoch": 7.864416642420716, + "flow_matching_loss": 0.0095667929854244, + "grad_norm": 0.058560218662023544, + "kept_token_count": 572.63828125, + "learning_rate": 7.972443479549035e-07, + "loss": 0.0101, + "mgd_cos_sim": 0.9942193984985351, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005780609883368015, + "mgd_loss_weight": 0.09915330484509469, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.843505859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1214233398438, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27030, + "valid_token_count": 818.03515625 + }, + { + "actual_token_mask_ratio": 0.3004889190196991, + "epoch": 7.867326156531859, + "flow_matching_loss": 0.010274358559399843, + "grad_norm": 0.07544386386871338, + "kept_token_count": 572.4328125, + "learning_rate": 7.919341901912236e-07, + "loss": 0.0109, + "mgd_cos_sim": 0.994076156616211, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005923852836713195, + "mgd_loss_weight": 0.09881996065378189, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.31748657226564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.17335510253906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27040, + "valid_token_count": 818.334375 + }, + { + "actual_token_mask_ratio": 0.29991022646427157, + "epoch": 7.870235670643003, + "flow_matching_loss": 0.010154149495065212, + "grad_norm": 0.06353039294481277, + "kept_token_count": 572.965625, + "learning_rate": 7.866412965672365e-07, + "loss": 0.0107, + "mgd_cos_sim": 0.9942391514778137, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005760842561721801, + "mgd_loss_weight": 0.09848661720752716, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.4804229736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5532470703125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27050, + "valid_token_count": 818.415625 + }, + { + "actual_token_mask_ratio": 0.29979259669780733, + "epoch": 7.873145184754146, + "flow_matching_loss": 0.010531292064115405, + "grad_norm": 0.05873200297355652, + "kept_token_count": 573.1515625, + "learning_rate": 7.813656735143044e-07, + "loss": 0.0111, + "mgd_cos_sim": 0.9940978348255157, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005902170529589057, + "mgd_loss_weight": 0.09815327078104019, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.97031860351564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.7205535888672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27060, + "valid_token_count": 818.54296875 + }, + { + "actual_token_mask_ratio": 0.2997400999069214, + "epoch": 7.876054698865289, + "flow_matching_loss": 0.011820242460817098, + "grad_norm": 0.07862067222595215, + "kept_token_count": 572.97890625, + "learning_rate": 7.761073274428076e-07, + "loss": 0.0125, + "mgd_cos_sim": 0.9935252666473389, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006474731164053083, + "mgd_loss_weight": 0.09781992733478546, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.1881408691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3489532470703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27070, + "valid_token_count": 818.2359375 + }, + { + "actual_token_mask_ratio": 0.29978619515895844, + "epoch": 7.878964212976433, + "flow_matching_loss": 0.009516131365671754, + "grad_norm": 0.0680142417550087, + "kept_token_count": 573.0734375, + "learning_rate": 7.708662647421355e-07, + "loss": 0.0101, + "mgd_cos_sim": 0.9939017176628113, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006098289834335446, + "mgd_loss_weight": 0.09748658314347267, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.01378173828124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1645874023437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27080, + "valid_token_count": 818.42109375 + }, + { + "actual_token_mask_ratio": 0.3000826478004456, + "epoch": 7.8818737270875765, + "flow_matching_loss": 0.010652663744986057, + "grad_norm": 0.06357044726610184, + "kept_token_count": 572.88515625, + "learning_rate": 7.65642491780672e-07, + "loss": 0.0113, + "mgd_cos_sim": 0.9935999393463135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064000572543591264, + "mgd_loss_weight": 0.09715323895215988, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.48537292480466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9676116943359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27090, + "valid_token_count": 818.50859375 + }, + { + "actual_token_mask_ratio": 0.3004527598619461, + "epoch": 7.88478324119872, + "flow_matching_loss": 0.008612558990716935, + "grad_norm": 0.08861103653907776, + "kept_token_count": 572.33203125, + "learning_rate": 7.604360149057916e-07, + "loss": 0.0092, + "mgd_cos_sim": 0.9939493000507355, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060506985988467935, + "mgd_loss_weight": 0.09681989327073097, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.6664093017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.01560363769534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27100, + "valid_token_count": 818.14375 + }, + { + "actual_token_mask_ratio": 0.2999733805656433, + "epoch": 7.887692755309863, + "flow_matching_loss": 0.009979246510192752, + "grad_norm": 0.04671040549874306, + "kept_token_count": 572.76328125, + "learning_rate": 7.552468404438568e-07, + "loss": 0.0106, + "mgd_cos_sim": 0.9936086893081665, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006391321308910847, + "mgd_loss_weight": 0.09648654982447624, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.82501220703125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.96901550292966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27110, + "valid_token_count": 818.2015625 + }, + { + "actual_token_mask_ratio": 0.2995732963085175, + "epoch": 7.890602269421007, + "flow_matching_loss": 0.008992500603199005, + "grad_norm": 0.09137700498104095, + "kept_token_count": 573.27421875, + "learning_rate": 7.500749747002022e-07, + "loss": 0.0095, + "mgd_cos_sim": 0.994568532705307, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005431484710425138, + "mgd_loss_weight": 0.09615320563316346, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.1527893066406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0025146484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27120, + "valid_token_count": 818.4625 + }, + { + "actual_token_mask_ratio": 0.2993538498878479, + "epoch": 7.89351178353215, + "flow_matching_loss": 0.010794037673622369, + "grad_norm": 0.04869851469993591, + "kept_token_count": 573.32109375, + "learning_rate": 7.449204239591345e-07, + "loss": 0.0114, + "mgd_cos_sim": 0.9940500617027282, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005949945328757167, + "mgd_loss_weight": 0.0958198606967926, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.2896484375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.21425170898436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27130, + "valid_token_count": 818.27734375 + }, + { + "actual_token_mask_ratio": 0.30077266991138457, + "epoch": 7.896421297643293, + "flow_matching_loss": 0.009318853123113513, + "grad_norm": 0.07198414951562881, + "kept_token_count": 572.0328125, + "learning_rate": 7.397831944839167e-07, + "loss": 0.0099, + "mgd_cos_sim": 0.9940089583396912, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059910538140684364, + "mgd_loss_weight": 0.09548651576042175, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.1106292724609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.82567138671874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27140, + "valid_token_count": 818.09375 + }, + { + "actual_token_mask_ratio": 0.30019613206386564, + "epoch": 7.899330811754437, + "flow_matching_loss": 0.00980276195332408, + "grad_norm": 0.05710418522357941, + "kept_token_count": 572.715625, + "learning_rate": 7.346632925167706e-07, + "loss": 0.0104, + "mgd_cos_sim": 0.9937927067279816, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006207276228815317, + "mgd_loss_weight": 0.09515317156910896, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.1918060302734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1912109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27150, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.2999676525592804, + "epoch": 7.902240325865581, + "flow_matching_loss": 0.009273062506690622, + "grad_norm": 0.06697714328765869, + "kept_token_count": 572.9609375, + "learning_rate": 7.295607242788566e-07, + "loss": 0.0099, + "mgd_cos_sim": 0.9939153492450714, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006084652338176966, + "mgd_loss_weight": 0.09481982812285424, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.3878448486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0517211914063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27160, + "valid_token_count": 818.48515625 + }, + { + "actual_token_mask_ratio": 0.30017244517803193, + "epoch": 7.905149839976724, + "flow_matching_loss": 0.010679405741393566, + "grad_norm": 0.08960755914449692, + "kept_token_count": 572.55078125, + "learning_rate": 7.244754959702765e-07, + "loss": 0.0112, + "mgd_cos_sim": 0.9942196190357209, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0057803862262517216, + "mgd_loss_weight": 0.09448648244142532, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.66321716308596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.21959533691404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27170, + "valid_token_count": 818.13203125 + }, + { + "actual_token_mask_ratio": 0.30037170052528384, + "epoch": 7.908059354087867, + "flow_matching_loss": 0.010579423047602177, + "grad_norm": 0.0907476618885994, + "kept_token_count": 572.60625, + "learning_rate": 7.194076137700628e-07, + "loss": 0.0112, + "mgd_cos_sim": 0.9938835263252258, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006116481777280569, + "mgd_loss_weight": 0.09415313825011254, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.69020690917966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.5938751220703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27180, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.30046246349811556, + "epoch": 7.910968868199011, + "flow_matching_loss": 0.009487853338941931, + "grad_norm": 0.06589949876070023, + "kept_token_count": 572.4484375, + "learning_rate": 7.143570838361696e-07, + "loss": 0.0101, + "mgd_cos_sim": 0.993946784734726, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006053215591236949, + "mgd_loss_weight": 0.09381979405879974, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.0575408935547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.87012634277346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27190, + "valid_token_count": 818.32265625 + }, + { + "actual_token_mask_ratio": 0.30040143728256224, + "epoch": 7.913878382310154, + "flow_matching_loss": 0.008807464735582471, + "grad_norm": 0.074674591422081, + "kept_token_count": 572.415625, + "learning_rate": 7.093239123054696e-07, + "loss": 0.0094, + "mgd_cos_sim": 0.9940115094184876, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005988485738635063, + "mgd_loss_weight": 0.09348645061254501, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.5251190185547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.826708984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27200, + "valid_token_count": 818.2078125 + }, + { + "actual_token_mask_ratio": 0.30032147765159606, + "epoch": 7.916787896421297, + "flow_matching_loss": 0.00990948169492185, + "grad_norm": 0.06631717085838318, + "kept_token_count": 572.5421875, + "learning_rate": 7.043081052937389e-07, + "loss": 0.0105, + "mgd_cos_sim": 0.9937776446342468, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062223545275628565, + "mgd_loss_weight": 0.09315310418605804, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.7591156005859, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0037078857422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27210, + "valid_token_count": 818.29609375 + }, + { + "actual_token_mask_ratio": 0.29988121092319486, + "epoch": 7.919697410532441, + "flow_matching_loss": 0.010049715777859092, + "grad_norm": 0.07530006021261215, + "kept_token_count": 573.1453125, + "learning_rate": 6.993096688956558e-07, + "loss": 0.0106, + "mgd_cos_sim": 0.993750411272049, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006249576807022095, + "mgd_loss_weight": 0.09281975999474526, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 405.0565216064453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3039794921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27220, + "valid_token_count": 818.6390625 + }, + { + "actual_token_mask_ratio": 0.29957271814346315, + "epoch": 7.922606924643585, + "flow_matching_loss": 0.01103905811905861, + "grad_norm": 0.06430376321077347, + "kept_token_count": 573.14296875, + "learning_rate": 6.943286091847884e-07, + "loss": 0.0116, + "mgd_cos_sim": 0.9939516007900238, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060483993962407116, + "mgd_loss_weight": 0.09248641654849052, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.15282287597654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.94736328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27230, + "valid_token_count": 818.27734375 + }, + { + "actual_token_mask_ratio": 0.2993245512247086, + "epoch": 7.925516438754728, + "flow_matching_loss": 0.009949815273284913, + "grad_norm": 0.07950831949710846, + "kept_token_count": 573.50390625, + "learning_rate": 6.89364932213602e-07, + "loss": 0.0105, + "mgd_cos_sim": 0.9940149247646332, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005985078960657119, + "mgd_loss_weight": 0.09215307235717773, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.0901824951172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.99664306640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27240, + "valid_token_count": 818.49921875 + }, + { + "actual_token_mask_ratio": 0.2999925047159195, + "epoch": 7.928425952865871, + "flow_matching_loss": 0.010157987289130688, + "grad_norm": 0.0692518875002861, + "kept_token_count": 572.71015625, + "learning_rate": 6.844186440134253e-07, + "loss": 0.0107, + "mgd_cos_sim": 0.9941106081008911, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0058893953450024124, + "mgd_loss_weight": 0.09181972667574882, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.2313995361328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5862548828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27250, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.29966541826725007, + "epoch": 7.931335466977015, + "flow_matching_loss": 0.0091943739913404, + "grad_norm": 0.06481482088565826, + "kept_token_count": 573.05546875, + "learning_rate": 6.794897505944692e-07, + "loss": 0.0097, + "mgd_cos_sim": 0.9942705392837524, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005729460716247558, + "mgd_loss_weight": 0.09148638248443604, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.2862548828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.0214141845703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27260, + "valid_token_count": 818.25703125 + }, + { + "actual_token_mask_ratio": 0.30011325478553774, + "epoch": 7.934244981088158, + "flow_matching_loss": 0.009261667123064398, + "grad_norm": 0.0635635182261467, + "kept_token_count": 572.70234375, + "learning_rate": 6.745782579457993e-07, + "loss": 0.0098, + "mgd_cos_sim": 0.9937777578830719, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006222240254282951, + "mgd_loss_weight": 0.09115303903818131, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.1693634033203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.10711059570315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27270, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.30015051960945127, + "epoch": 7.937154495199302, + "flow_matching_loss": 0.009902630653232335, + "grad_norm": 0.0735534280538559, + "kept_token_count": 572.7875, + "learning_rate": 6.696841720353453e-07, + "loss": 0.0105, + "mgd_cos_sim": 0.9939272046089173, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006072786729782819, + "mgd_loss_weight": 0.09081969410181046, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.4049896240234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.19239807128906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27280, + "valid_token_count": 818.4421875 + }, + { + "actual_token_mask_ratio": 0.3007117211818695, + "epoch": 7.940064009310445, + "flow_matching_loss": 0.009881410980597138, + "grad_norm": 0.07069293409585953, + "kept_token_count": 572.25625, + "learning_rate": 6.648074988098834e-07, + "loss": 0.0104, + "mgd_cos_sim": 0.9940452456474305, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00595476757735014, + "mgd_loss_weight": 0.0904863491654396, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.56829223632815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.7582733154297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27290, + "valid_token_count": 818.3421875 + }, + { + "actual_token_mask_ratio": 0.3000251442193985, + "epoch": 7.942973523421589, + "flow_matching_loss": 0.009569071279838682, + "grad_norm": 0.0802290067076683, + "kept_token_count": 572.76484375, + "learning_rate": 6.599482441950322e-07, + "loss": 0.0101, + "mgd_cos_sim": 0.9938578069210052, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061421982478350404, + "mgd_loss_weight": 0.09015300497412682, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.0781677246094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.38478393554686, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27300, + "valid_token_count": 818.26328125 + }, + { + "actual_token_mask_ratio": 0.3003507614135742, + "epoch": 7.945883037532732, + "flow_matching_loss": 0.010641548037528991, + "grad_norm": 0.06547287106513977, + "kept_token_count": 572.4328125, + "learning_rate": 6.551064140952417e-07, + "loss": 0.0112, + "mgd_cos_sim": 0.9939809739589691, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00601901076734066, + "mgd_loss_weight": 0.08981966152787209, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.6963195800781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9507110595703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27310, + "valid_token_count": 818.1765625 + }, + { + "actual_token_mask_ratio": 0.3004328697919846, + "epoch": 7.9487925516438755, + "flow_matching_loss": 0.00977000785060227, + "grad_norm": 0.07637304067611694, + "kept_token_count": 572.396875, + "learning_rate": 6.502820143937943e-07, + "loss": 0.0103, + "mgd_cos_sim": 0.9939417123794556, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060582818929106, + "mgd_loss_weight": 0.08948631584644318, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.4428283691406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.37757568359376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27320, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.30013874769210813, + "epoch": 7.951702065755019, + "flow_matching_loss": 0.010462911427021026, + "grad_norm": 0.09386486560106277, + "kept_token_count": 572.57734375, + "learning_rate": 6.454750509527874e-07, + "loss": 0.011, + "mgd_cos_sim": 0.993755841255188, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006244156369939446, + "mgd_loss_weight": 0.08915297091007232, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.90240173339845, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0074127197266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27330, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.3001108705997467, + "epoch": 7.954611579866162, + "flow_matching_loss": 0.009757715230807661, + "grad_norm": 0.08126954734325409, + "kept_token_count": 572.753125, + "learning_rate": 6.406855296131387e-07, + "loss": 0.0103, + "mgd_cos_sim": 0.994079840183258, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005920167127624154, + "mgd_loss_weight": 0.0888196274638176, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.28583984375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.89678649902345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27340, + "valid_token_count": 818.35078125 + }, + { + "actual_token_mask_ratio": 0.3002460688352585, + "epoch": 7.957521093977306, + "flow_matching_loss": 0.009964514430612326, + "grad_norm": 0.06897258013486862, + "kept_token_count": 572.434375, + "learning_rate": 6.359134561945662e-07, + "loss": 0.0105, + "mgd_cos_sim": 0.9938443839550019, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006155615346506238, + "mgd_loss_weight": 0.08848628401756287, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.45119323730466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.19305419921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27350, + "valid_token_count": 818.05234375 + }, + { + "actual_token_mask_ratio": 0.3001899540424347, + "epoch": 7.9604306080884495, + "flow_matching_loss": 0.00944364294409752, + "grad_norm": 0.07885332405567169, + "kept_token_count": 572.6046875, + "learning_rate": 6.311588364955883e-07, + "loss": 0.01, + "mgd_cos_sim": 0.9937059879302979, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006294014351442456, + "mgd_loss_weight": 0.0881529375910759, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.6918487548828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.1573974609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27360, + "valid_token_count": 818.228125 + }, + { + "actual_token_mask_ratio": 0.30061176121234895, + "epoch": 7.963340122199592, + "flow_matching_loss": 0.009044451778754592, + "grad_norm": 0.12864433228969574, + "kept_token_count": 572.225, + "learning_rate": 6.264216762935187e-07, + "loss": 0.0096, + "mgd_cos_sim": 0.9939621806144714, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006037827860563993, + "mgd_loss_weight": 0.08781959339976311, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.3844482421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.77088317871096, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27370, + "valid_token_count": 818.17734375 + }, + { + "actual_token_mask_ratio": 0.3006211489439011, + "epoch": 7.966249636310736, + "flow_matching_loss": 0.010228591039776802, + "grad_norm": 0.06443148851394653, + "kept_token_count": 572.20625, + "learning_rate": 6.21701981344453e-07, + "loss": 0.0108, + "mgd_cos_sim": 0.993885463476181, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00611452623270452, + "mgd_loss_weight": 0.08748624995350837, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.9694549560547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9303802490234, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27380, + "valid_token_count": 818.165625 + }, + { + "actual_token_mask_ratio": 0.2994116634130478, + "epoch": 7.96915915042188, + "flow_matching_loss": 0.010770188644528389, + "grad_norm": 0.052072614431381226, + "kept_token_count": 573.4046875, + "learning_rate": 6.16999757383262e-07, + "loss": 0.0113, + "mgd_cos_sim": 0.9939123094081879, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006087680673226714, + "mgd_loss_weight": 0.08715290576219559, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.7184692382813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8490020751953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27390, + "valid_token_count": 818.4609375 + }, + { + "actual_token_mask_ratio": 0.29959493279457095, + "epoch": 7.9720686645330225, + "flow_matching_loss": 0.008932623080909252, + "grad_norm": 0.06577331572771072, + "kept_token_count": 573.11953125, + "learning_rate": 6.123150101235952e-07, + "loss": 0.0094, + "mgd_cos_sim": 0.9940544247627259, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005945589765906334, + "mgd_loss_weight": 0.08681956008076668, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.59503479003905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0004943847656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27400, + "valid_token_count": 818.26953125 + }, + { + "actual_token_mask_ratio": 0.3003071427345276, + "epoch": 7.974978178644166, + "flow_matching_loss": 0.010157679812982679, + "grad_norm": 0.0646962821483612, + "kept_token_count": 572.56953125, + "learning_rate": 6.076477452578594e-07, + "loss": 0.0107, + "mgd_cos_sim": 0.9937623977661133, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006237602699548006, + "mgd_loss_weight": 0.08648621588945389, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.0872314453125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5829162597656, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27410, + "valid_token_count": 818.31796875 + }, + { + "actual_token_mask_ratio": 0.2999618411064148, + "epoch": 7.97788769275531, + "flow_matching_loss": 0.010868643969297409, + "grad_norm": 0.08558405190706253, + "kept_token_count": 572.65703125, + "learning_rate": 6.029979684572262e-07, + "loss": 0.0114, + "mgd_cos_sim": 0.993908828496933, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006091175461187958, + "mgd_loss_weight": 0.08615287244319916, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.74005432128905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4048614501953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27420, + "valid_token_count": 818.03203125 + }, + { + "actual_token_mask_ratio": 0.3007804274559021, + "epoch": 7.980797206866454, + "flow_matching_loss": 0.008966911025345326, + "grad_norm": 0.08075396716594696, + "kept_token_count": 572.1546875, + "learning_rate": 5.983656853716097e-07, + "loss": 0.0095, + "mgd_cos_sim": 0.9941279292106628, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0058720741420984265, + "mgd_loss_weight": 0.08581952750682831, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.63426513671874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.047998046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27430, + "valid_token_count": 818.27734375 + }, + { + "actual_token_mask_ratio": 0.3001518756151199, + "epoch": 7.9837067209775965, + "flow_matching_loss": 0.01033459478057921, + "grad_norm": 0.10834997147321701, + "kept_token_count": 572.69296875, + "learning_rate": 5.937509016296689e-07, + "loss": 0.0109, + "mgd_cos_sim": 0.9938570618629455, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006142922816798091, + "mgd_loss_weight": 0.0854861818253994, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.8507385253906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.97239685058594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27440, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.29979555010795594, + "epoch": 7.98661623508874, + "flow_matching_loss": 0.010205463320016862, + "grad_norm": 0.09477737545967102, + "kept_token_count": 572.909375, + "learning_rate": 5.89153622838805e-07, + "loss": 0.0107, + "mgd_cos_sim": 0.994272643327713, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005727357463911176, + "mgd_loss_weight": 0.08515283837914467, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.1459503173828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9138580322266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27450, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.30004558265209197, + "epoch": 7.989525749199884, + "flow_matching_loss": 0.009010005602613091, + "grad_norm": 0.07890121638774872, + "kept_token_count": 572.6765625, + "learning_rate": 5.845738545851431e-07, + "loss": 0.0095, + "mgd_cos_sim": 0.9939764738082886, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006023539369925857, + "mgd_loss_weight": 0.08481949493288994, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.9413146972656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3722198486328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27460, + "valid_token_count": 818.16171875 + }, + { + "actual_token_mask_ratio": 0.29944812059402465, + "epoch": 7.992435263311027, + "flow_matching_loss": 0.010672592138871551, + "grad_norm": 0.06728491932153702, + "kept_token_count": 573.2484375, + "learning_rate": 5.800116024335334e-07, + "loss": 0.0112, + "mgd_cos_sim": 0.9941124677658081, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005887535912916064, + "mgd_loss_weight": 0.08448614925146103, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.9320343017578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0648468017578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27470, + "valid_token_count": 818.28359375 + }, + { + "actual_token_mask_ratio": 0.3002313733100891, + "epoch": 7.99534477742217, + "flow_matching_loss": 0.010426911059767008, + "grad_norm": 0.10387665033340454, + "kept_token_count": 572.48515625, + "learning_rate": 5.75466871927543e-07, + "loss": 0.0109, + "mgd_cos_sim": 0.9939090490341187, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060909480787813665, + "mgd_loss_weight": 0.08415280431509017, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.5293304443359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.37730102539064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27480, + "valid_token_count": 818.10234375 + }, + { + "actual_token_mask_ratio": 0.3006665676832199, + "epoch": 7.998254291533314, + "flow_matching_loss": 0.009751614462584258, + "grad_norm": 0.08941088616847992, + "kept_token_count": 572.209375, + "learning_rate": 5.709396685894463e-07, + "loss": 0.0103, + "mgd_cos_sim": 0.9938682317733765, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006131768645718694, + "mgd_loss_weight": 0.08381946086883545, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.52557067871095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0033447265625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27490, + "valid_token_count": 818.22578125 + }, + { + "actual_token_mask_ratio": 0.3008122593164444, + "epoch": 8.001163805644458, + "flow_matching_loss": 0.008517815731465816, + "grad_norm": 0.0647219568490982, + "kept_token_count": 572.2457885742188, + "learning_rate": 5.664299979202264e-07, + "loss": 0.009, + "mgd_cos_sim": 0.9941650211811066, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005834981473162771, + "mgd_loss_weight": 0.08348611742258072, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.67043151855466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.74855041503906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27500, + "valid_token_count": 818.4460632324219 + }, + { + "actual_token_mask_ratio": 0.30014669597148896, + "epoch": 8.004073319755602, + "flow_matching_loss": 0.0105460817925632, + "grad_norm": 0.07142964005470276, + "kept_token_count": 572.6734375, + "learning_rate": 5.61937865399555e-07, + "loss": 0.0111, + "mgd_cos_sim": 0.9936281144618988, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006371890660375357, + "mgd_loss_weight": 0.08315277099609375, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.67149353027344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.48297424316405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27510, + "valid_token_count": 818.278125 + }, + { + "actual_token_mask_ratio": 0.29909024238586424, + "epoch": 8.006982833866743, + "flow_matching_loss": 0.010646941233426332, + "grad_norm": 0.06889656186103821, + "kept_token_count": 573.4515625, + "learning_rate": 5.574632764857974e-07, + "loss": 0.0112, + "mgd_cos_sim": 0.9938717544078827, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006128232507035136, + "mgd_loss_weight": 0.08281942680478097, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.8831481933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4428741455078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27520, + "valid_token_count": 818.15859375 + }, + { + "actual_token_mask_ratio": 0.29961864054203036, + "epoch": 8.009892347977887, + "flow_matching_loss": 0.010416920576244593, + "grad_norm": 0.0668485164642334, + "kept_token_count": 573.06875, + "learning_rate": 5.53006236616001e-07, + "loss": 0.0109, + "mgd_cos_sim": 0.9940387368202209, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005961267510429025, + "mgd_loss_weight": 0.08248608335852622, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.35932006835935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2117980957031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27530, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.3004141926765442, + "epoch": 8.012801862089031, + "flow_matching_loss": 0.009680722886696459, + "grad_norm": 0.061354659497737885, + "kept_token_count": 572.36171875, + "learning_rate": 5.485667512058873e-07, + "loss": 0.0102, + "mgd_cos_sim": 0.9939364314079284, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060635741334408525, + "mgd_loss_weight": 0.08215273916721344, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.6014739990234, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.51661071777346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27540, + "valid_token_count": 818.140625 + }, + { + "actual_token_mask_ratio": 0.2993587404489517, + "epoch": 8.015711376200175, + "flow_matching_loss": 0.010172323463484645, + "grad_norm": 0.06317845731973648, + "kept_token_count": 573.4765625, + "learning_rate": 5.441448256498544e-07, + "loss": 0.0107, + "mgd_cos_sim": 0.9941332578659058, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0058667480014264585, + "mgd_loss_weight": 0.08181939274072647, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.6287475585938, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.21690979003904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27550, + "valid_token_count": 818.5015625 + }, + { + "actual_token_mask_ratio": 0.30010606944561, + "epoch": 8.018620890311318, + "flow_matching_loss": 0.00948881721124053, + "grad_norm": 0.09486053884029388, + "kept_token_count": 572.67890625, + "learning_rate": 5.397404653209559e-07, + "loss": 0.01, + "mgd_cos_sim": 0.993626719713211, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006373286200687289, + "mgd_loss_weight": 0.08148604929447174, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.4656036376953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4405029296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27560, + "valid_token_count": 818.2375 + }, + { + "actual_token_mask_ratio": 0.2999782651662827, + "epoch": 8.021530404422462, + "flow_matching_loss": 0.010643631359562278, + "grad_norm": 0.06069819629192352, + "kept_token_count": 572.77265625, + "learning_rate": 5.35353675570906e-07, + "loss": 0.0112, + "mgd_cos_sim": 0.9936375439167022, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00636246595531702, + "mgd_loss_weight": 0.08115270584821702, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.8215087890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.3325561523437, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27570, + "valid_token_count": 818.2203125 + }, + { + "actual_token_mask_ratio": 0.30019248425960543, + "epoch": 8.024439918533606, + "flow_matching_loss": 0.009872223250567913, + "grad_norm": 0.07534807175397873, + "kept_token_count": 572.5890625, + "learning_rate": 5.309844617300674e-07, + "loss": 0.0104, + "mgd_cos_sim": 0.9939240753650666, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060759116895496845, + "mgd_loss_weight": 0.08081936091184616, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.8456604003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.15663452148436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27580, + "valid_token_count": 818.2125 + }, + { + "actual_token_mask_ratio": 0.2994853526353836, + "epoch": 8.027349432644748, + "flow_matching_loss": 0.009424732066690922, + "grad_norm": 0.07558830827474594, + "kept_token_count": 573.23515625, + "learning_rate": 5.266328291074468e-07, + "loss": 0.0099, + "mgd_cos_sim": 0.9942135632038116, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005786425573751331, + "mgd_loss_weight": 0.08048601523041725, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.4144348144531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0836547851562, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27590, + "valid_token_count": 818.30703125 + }, + { + "actual_token_mask_ratio": 0.29999497532844543, + "epoch": 8.030258946755891, + "flow_matching_loss": 0.011021843738853931, + "grad_norm": 0.06831004470586777, + "kept_token_count": 572.871875, + "learning_rate": 5.222987829906861e-07, + "loss": 0.0116, + "mgd_cos_sim": 0.993198162317276, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006801838520914316, + "mgd_loss_weight": 0.08015267178416252, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 407.29722900390624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.7167114257812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27600, + "valid_token_count": 818.38203125 + }, + { + "actual_token_mask_ratio": 0.2994005113840103, + "epoch": 8.033168460867035, + "flow_matching_loss": 0.009998772479593753, + "grad_norm": 0.06625663489103317, + "kept_token_count": 573.0015625, + "learning_rate": 5.179823286460644e-07, + "loss": 0.0105, + "mgd_cos_sim": 0.9937537312507629, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062462640926241875, + "mgd_loss_weight": 0.0798193283379078, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.1619079589844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0400848388672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27610, + "valid_token_count": 817.87734375 + }, + { + "actual_token_mask_ratio": 0.3001994341611862, + "epoch": 8.036077974978179, + "flow_matching_loss": 0.010583771765232087, + "grad_norm": 0.06613364070653915, + "kept_token_count": 572.76953125, + "learning_rate": 5.13683471318478e-07, + "loss": 0.0111, + "mgd_cos_sim": 0.9939957082271575, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060042859055101875, + "mgd_loss_weight": 0.07948598265647888, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.0759643554687, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.57533874511716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27620, + "valid_token_count": 818.47109375 + }, + { + "actual_token_mask_ratio": 0.3005002409219742, + "epoch": 8.038987489089322, + "flow_matching_loss": 0.009638367546722293, + "grad_norm": 0.06721439957618713, + "kept_token_count": 572.37265625, + "learning_rate": 5.094022162314449e-07, + "loss": 0.0101, + "mgd_cos_sim": 0.9942384123802185, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0057615727186203, + "mgd_loss_weight": 0.07915263772010803, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.31470947265626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.812109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27630, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.29982076585292816, + "epoch": 8.041897003200466, + "flow_matching_loss": 0.009022361971437931, + "grad_norm": 0.10945206880569458, + "kept_token_count": 572.88125, + "learning_rate": 5.051385685870924e-07, + "loss": 0.0095, + "mgd_cos_sim": 0.9941535413265228, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005846461420878768, + "mgd_loss_weight": 0.0788192942738533, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.871142578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.73935546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27640, + "valid_token_count": 818.1921875 + }, + { + "actual_token_mask_ratio": 0.3007183909416199, + "epoch": 8.044806517311608, + "flow_matching_loss": 0.010181357292458415, + "grad_norm": 0.07892516255378723, + "kept_token_count": 572.1, + "learning_rate": 5.008925335661546e-07, + "loss": 0.0107, + "mgd_cos_sim": 0.9938432335853576, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006156766368076205, + "mgd_loss_weight": 0.07848595008254051, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.19382629394534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.26243896484374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27650, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.300397589802742, + "epoch": 8.047716031422752, + "flow_matching_loss": 0.0108909429050982, + "grad_norm": 0.08310779184103012, + "kept_token_count": 572.49375, + "learning_rate": 4.966641163279662e-07, + "loss": 0.0113, + "mgd_cos_sim": 0.9941914975643158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005808503646403551, + "mgd_loss_weight": 0.0781526044011116, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.46888122558596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9558624267578, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27660, + "valid_token_count": 818.3109375 + }, + { + "actual_token_mask_ratio": 0.3002324640750885, + "epoch": 8.050625545533896, + "flow_matching_loss": 0.009554454870522023, + "grad_norm": 0.08396144956350327, + "kept_token_count": 572.44296875, + "learning_rate": 4.924533220104521e-07, + "loss": 0.01, + "mgd_cos_sim": 0.9939590632915497, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060409399215131995, + "mgd_loss_weight": 0.07781926020979882, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.26319580078126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3306945800781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27670, + "valid_token_count": 818.0421875 + }, + { + "actual_token_mask_ratio": 0.2999306708574295, + "epoch": 8.05353505964504, + "flow_matching_loss": 0.010169249214231967, + "grad_norm": 0.08247242867946625, + "kept_token_count": 572.73125, + "learning_rate": 4.882601557301231e-07, + "loss": 0.0106, + "mgd_cos_sim": 0.9938693344593048, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006130659347400069, + "mgd_loss_weight": 0.07748591676354408, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.1018310546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2524932861328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27680, + "valid_token_count": 818.1046875 + }, + { + "actual_token_mask_ratio": 0.30054285526275637, + "epoch": 8.056444573756183, + "flow_matching_loss": 0.009801173396408559, + "grad_norm": 0.05311458185315132, + "kept_token_count": 572.42421875, + "learning_rate": 4.840846225820717e-07, + "loss": 0.0102, + "mgd_cos_sim": 0.9942869186401367, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005713085690513253, + "mgd_loss_weight": 0.07715257257223129, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.69309997558594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8302001953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27690, + "valid_token_count": 818.38671875 + }, + { + "actual_token_mask_ratio": 0.3004006385803223, + "epoch": 8.059354087867327, + "flow_matching_loss": 0.010717620560899376, + "grad_norm": 0.06584852933883667, + "kept_token_count": 572.5, + "learning_rate": 4.79926727639966e-07, + "loss": 0.0112, + "mgd_cos_sim": 0.993777734041214, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006222252082079649, + "mgd_loss_weight": 0.07681922614574432, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.91710510253904, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.12884521484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27700, + "valid_token_count": 818.32578125 + }, + { + "actual_token_mask_ratio": 0.30046319365501406, + "epoch": 8.06226360197847, + "flow_matching_loss": 0.011837603710591792, + "grad_norm": 0.10628306865692139, + "kept_token_count": 572.446875, + "learning_rate": 4.7578647595603817e-07, + "loss": 0.0123, + "mgd_cos_sim": 0.993925827741623, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006074167275801301, + "mgd_loss_weight": 0.0764858826994896, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.3496368408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.40460510253905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27710, + "valid_token_count": 818.3265625 + }, + { + "actual_token_mask_ratio": 0.2999122440814972, + "epoch": 8.065173116089612, + "flow_matching_loss": 0.00876374957151711, + "grad_norm": 0.05324616655707359, + "kept_token_count": 572.925, + "learning_rate": 4.716638725610889e-07, + "loss": 0.0092, + "mgd_cos_sim": 0.9938462018966675, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006153796287253499, + "mgd_loss_weight": 0.07615253925323487, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.9936798095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2294128417969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27720, + "valid_token_count": 818.35859375 + }, + { + "actual_token_mask_ratio": 0.29998699128627776, + "epoch": 8.068082630200756, + "flow_matching_loss": 0.008222024608403444, + "grad_norm": 0.07333754748106003, + "kept_token_count": 572.9234375, + "learning_rate": 4.6755892246446606e-07, + "loss": 0.0087, + "mgd_cos_sim": 0.9940014839172363, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005998530890792608, + "mgd_loss_weight": 0.07581919431686401, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 422.10247802734375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0649047851563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27730, + "valid_token_count": 818.44609375 + }, + { + "actual_token_mask_ratio": 0.30010946393013, + "epoch": 8.0709921443119, + "flow_matching_loss": 0.00985496244393289, + "grad_norm": 0.06510623544454575, + "kept_token_count": 573.01640625, + "learning_rate": 4.634716306540715e-07, + "loss": 0.0103, + "mgd_cos_sim": 0.9937035024166108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006296502845361829, + "mgd_loss_weight": 0.0754858486354351, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.8267395019531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.7807159423828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27740, + "valid_token_count": 818.7234375 + }, + { + "actual_token_mask_ratio": 0.2999833583831787, + "epoch": 8.073901658423043, + "flow_matching_loss": 0.010011468222364784, + "grad_norm": 0.07280506938695908, + "kept_token_count": 572.821875, + "learning_rate": 4.5940200209635244e-07, + "loss": 0.0105, + "mgd_cos_sim": 0.993858027458191, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061419826000928875, + "mgd_loss_weight": 0.07515250518918037, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.7300262451172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9760437011719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27750, + "valid_token_count": 818.29765625 + }, + { + "actual_token_mask_ratio": 0.30035560131072997, + "epoch": 8.076811172534187, + "flow_matching_loss": 0.010622528148815036, + "grad_norm": 0.09900770336389542, + "kept_token_count": 572.57421875, + "learning_rate": 4.5535004173628657e-07, + "loss": 0.0111, + "mgd_cos_sim": 0.9936828792095185, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006317136716097594, + "mgd_loss_weight": 0.07481916099786759, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.29131774902345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0201416015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27760, + "valid_token_count": 818.38828125 + }, + { + "actual_token_mask_ratio": 0.3007500022649765, + "epoch": 8.07972068664533, + "flow_matching_loss": 0.009829688351601362, + "grad_norm": 0.06173962354660034, + "kept_token_count": 572.0484375, + "learning_rate": 4.5131575449739225e-07, + "loss": 0.0103, + "mgd_cos_sim": 0.9934839546680451, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006516035832464695, + "mgd_loss_weight": 0.07448581606149673, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.9473419189453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1512817382812, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27770, + "valid_token_count": 818.0890625 + }, + { + "actual_token_mask_ratio": 0.2997113078832626, + "epoch": 8.082630200756475, + "flow_matching_loss": 0.00981317819096148, + "grad_norm": 0.08602485805749893, + "kept_token_count": 572.98984375, + "learning_rate": 4.472991452817099e-07, + "loss": 0.0103, + "mgd_cos_sim": 0.9940067052841186, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005993297649547457, + "mgd_loss_weight": 0.07415247112512588, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.42344360351564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.32286682128904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27780, + "valid_token_count": 818.21640625 + }, + { + "actual_token_mask_ratio": 0.300067201256752, + "epoch": 8.085539714867616, + "flow_matching_loss": 0.010034664999693632, + "grad_norm": 0.08182031661272049, + "kept_token_count": 572.7171875, + "learning_rate": 4.433002189697938e-07, + "loss": 0.0105, + "mgd_cos_sim": 0.9938000023365021, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006199999619275332, + "mgd_loss_weight": 0.07381912767887115, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.8456756591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.86733093261716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27790, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.30025065541267393, + "epoch": 8.08844922897876, + "flow_matching_loss": 0.009758034627884627, + "grad_norm": 0.08760201185941696, + "kept_token_count": 572.29609375, + "learning_rate": 4.393189804207187e-07, + "loss": 0.0102, + "mgd_cos_sim": 0.9941601157188416, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005839883629232645, + "mgd_loss_weight": 0.07348578348755837, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.4031066894531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.61553649902345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27800, + "valid_token_count": 817.86015625 + }, + { + "actual_token_mask_ratio": 0.3002143412828445, + "epoch": 8.091358743089904, + "flow_matching_loss": 0.01081466106697917, + "grad_norm": 0.06297440826892853, + "kept_token_count": 572.55859375, + "learning_rate": 4.353554344720651e-07, + "loss": 0.0112, + "mgd_cos_sim": 0.9943531155586243, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005646893475204706, + "mgd_loss_weight": 0.07315243780612946, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.2787841796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.290087890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27810, + "valid_token_count": 818.18828125 + }, + { + "actual_token_mask_ratio": 0.30000956952571867, + "epoch": 8.094268257201048, + "flow_matching_loss": 0.010784484120085836, + "grad_norm": 0.08056232333183289, + "kept_token_count": 572.6484375, + "learning_rate": 4.314095859399153e-07, + "loss": 0.0112, + "mgd_cos_sim": 0.9936470270156861, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006352972565218806, + "mgd_loss_weight": 0.07281909361481667, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.1256866455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6795349121094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27820, + "valid_token_count": 818.0734375 + }, + { + "actual_token_mask_ratio": 0.29940317273139955, + "epoch": 8.097177771312191, + "flow_matching_loss": 0.00977029330097139, + "grad_norm": 0.08345057815313339, + "kept_token_count": 573.1578125, + "learning_rate": 4.274814396188459e-07, + "loss": 0.0102, + "mgd_cos_sim": 0.9937349140644074, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006265095062553882, + "mgd_loss_weight": 0.07248575016856193, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.6815979003906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5879425048828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27830, + "valid_token_count": 818.103125 + }, + { + "actual_token_mask_ratio": 0.2999140411615372, + "epoch": 8.100087285423335, + "flow_matching_loss": 0.009942208137363195, + "grad_norm": 0.0591861717402935, + "kept_token_count": 572.8125, + "learning_rate": 4.235710002819271e-07, + "loss": 0.0104, + "mgd_cos_sim": 0.993987214565277, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006012787017971278, + "mgd_loss_weight": 0.07215240597724915, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.0093200683594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.17351684570315, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27840, + "valid_token_count": 818.2046875 + }, + { + "actual_token_mask_ratio": 0.30041840076446535, + "epoch": 8.102996799534477, + "flow_matching_loss": 0.01007961891591549, + "grad_norm": 0.05928880348801613, + "kept_token_count": 572.3875, + "learning_rate": 4.196782726807097e-07, + "loss": 0.0105, + "mgd_cos_sim": 0.9942230820655823, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005776909226551652, + "mgd_loss_weight": 0.07181905955076218, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.0628326416016, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4614562988281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27850, + "valid_token_count": 818.18828125 + }, + { + "actual_token_mask_ratio": 0.2995767205953598, + "epoch": 8.10590631364562, + "flow_matching_loss": 0.009838398545980453, + "grad_norm": 0.06534652411937714, + "kept_token_count": 573.07421875, + "learning_rate": 4.158032615452251e-07, + "loss": 0.0102, + "mgd_cos_sim": 0.9942838966846466, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005716105410829187, + "mgd_loss_weight": 0.07148571610450745, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.69337768554686, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.93397827148436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27860, + "valid_token_count": 818.18203125 + }, + { + "actual_token_mask_ratio": 0.2994101315736771, + "epoch": 8.108815827756764, + "flow_matching_loss": 0.011125530116260052, + "grad_norm": 0.06703834235668182, + "kept_token_count": 573.32578125, + "learning_rate": 4.1194597158397853e-07, + "loss": 0.0116, + "mgd_cos_sim": 0.9938780486583709, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006121958140283823, + "mgd_loss_weight": 0.07115237191319465, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.9095489501953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8105804443359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27870, + "valid_token_count": 818.34609375 + }, + { + "actual_token_mask_ratio": 0.3000223696231842, + "epoch": 8.111725341867908, + "flow_matching_loss": 0.01014930964447558, + "grad_norm": 0.06603679805994034, + "kept_token_count": 572.8453125, + "learning_rate": 4.081064074839408e-07, + "loss": 0.0106, + "mgd_cos_sim": 0.9940914452075958, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005908552743494511, + "mgd_loss_weight": 0.07081902772188187, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.61940612792966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2677032470703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27880, + "valid_token_count": 818.37109375 + }, + { + "actual_token_mask_ratio": 0.2998647540807724, + "epoch": 8.114634855979052, + "flow_matching_loss": 0.010121897188946604, + "grad_norm": 0.07943391054868698, + "kept_token_count": 572.98046875, + "learning_rate": 4.042845739105433e-07, + "loss": 0.0105, + "mgd_cos_sim": 0.9940467953681946, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059531938284635546, + "mgd_loss_weight": 0.07048568204045295, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.20787658691404, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.19390869140625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27890, + "valid_token_count": 818.37890625 + }, + { + "actual_token_mask_ratio": 0.3003407776355743, + "epoch": 8.117544370090195, + "flow_matching_loss": 0.009823138173669577, + "grad_norm": 0.08023025095462799, + "kept_token_count": 572.47890625, + "learning_rate": 4.004804755076746e-07, + "loss": 0.0102, + "mgd_cos_sim": 0.9940261840820312, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005973820388317108, + "mgd_loss_weight": 0.07015233859419823, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.0993286132813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3130157470703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27900, + "valid_token_count": 818.2234375 + }, + { + "actual_token_mask_ratio": 0.29969231188297274, + "epoch": 8.12045388420134, + "flow_matching_loss": 0.00911385086365044, + "grad_norm": 0.0654577985405922, + "kept_token_count": 573.0, + "learning_rate": 3.9669411689767063e-07, + "loss": 0.0095, + "mgd_cos_sim": 0.9938978612422943, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006102129677310586, + "mgd_loss_weight": 0.06981899440288544, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.8550323486328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2471405029297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27910, + "valid_token_count": 818.21015625 + }, + { + "actual_token_mask_ratio": 0.2998261660337448, + "epoch": 8.123363398312481, + "flow_matching_loss": 0.009454913018271328, + "grad_norm": 0.08014268428087234, + "kept_token_count": 573.00390625, + "learning_rate": 3.929255026813194e-07, + "loss": 0.0099, + "mgd_cos_sim": 0.9941040217876435, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0058959817048162225, + "mgd_loss_weight": 0.06948564946651459, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.4940887451172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4861785888672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27920, + "valid_token_count": 818.3734375 + }, + { + "actual_token_mask_ratio": 0.3001160860061646, + "epoch": 8.126272912423625, + "flow_matching_loss": 0.009276112355291844, + "grad_norm": 0.06413637101650238, + "kept_token_count": 572.709375, + "learning_rate": 3.8917463743783797e-07, + "loss": 0.0097, + "mgd_cos_sim": 0.994036066532135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005963939242064953, + "mgd_loss_weight": 0.06915230453014373, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.17872619628906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.45091247558594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27930, + "valid_token_count": 818.2890625 + }, + { + "actual_token_mask_ratio": 0.2997812032699585, + "epoch": 8.129182426534769, + "flow_matching_loss": 0.01024277051910758, + "grad_norm": 0.11619797348976135, + "kept_token_count": 572.971875, + "learning_rate": 3.854415257248839e-07, + "loss": 0.0107, + "mgd_cos_sim": 0.9937198460102081, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062801492866128685, + "mgd_loss_weight": 0.068818961083889, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.6134887695313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.211376953125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27940, + "valid_token_count": 818.2734375 + }, + { + "actual_token_mask_ratio": 0.3004299521446228, + "epoch": 8.132091940645912, + "flow_matching_loss": 0.008889161050319672, + "grad_norm": 0.07081720232963562, + "kept_token_count": 572.52578125, + "learning_rate": 3.8172617207853864e-07, + "loss": 0.0093, + "mgd_cos_sim": 0.9944046258926391, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005595368752256036, + "mgd_loss_weight": 0.06848561689257622, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.5379333496094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9874572753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27950, + "valid_token_count": 818.3953125 + }, + { + "actual_token_mask_ratio": 0.3000662624835968, + "epoch": 8.135001454757056, + "flow_matching_loss": 0.009603909403085708, + "grad_norm": 0.06711063534021378, + "kept_token_count": 572.65234375, + "learning_rate": 3.7802858101330773e-07, + "loss": 0.01, + "mgd_cos_sim": 0.9939675569534302, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060324535705149175, + "mgd_loss_weight": 0.06815227121114731, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.4072540283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8804931640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27960, + "valid_token_count": 818.15234375 + }, + { + "actual_token_mask_ratio": 0.2996863633394241, + "epoch": 8.1379109688682, + "flow_matching_loss": 0.010474079241976141, + "grad_norm": 0.08257749676704407, + "kept_token_count": 572.959375, + "learning_rate": 3.743487570221121e-07, + "loss": 0.0109, + "mgd_cos_sim": 0.9942006468772888, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005799334635958076, + "mgd_loss_weight": 0.06781892701983452, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.27918701171876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4161865234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27970, + "valid_token_count": 818.14765625 + }, + { + "actual_token_mask_ratio": 0.3001302719116211, + "epoch": 8.140820482979342, + "flow_matching_loss": 0.009903118247166276, + "grad_norm": 0.07335013896226883, + "kept_token_count": 572.5890625, + "learning_rate": 3.7068670457628517e-07, + "loss": 0.0103, + "mgd_cos_sim": 0.9939772725105286, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006022721622139215, + "mgd_loss_weight": 0.06748558282852173, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.2757080078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8988006591797, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27980, + "valid_token_count": 818.13984375 + }, + { + "actual_token_mask_ratio": 0.300760555267334, + "epoch": 8.143729997090485, + "flow_matching_loss": 0.01038942881859839, + "grad_norm": 0.07622578740119934, + "kept_token_count": 572.26796875, + "learning_rate": 3.6704242812556576e-07, + "loss": 0.0108, + "mgd_cos_sim": 0.9937749564647674, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062250358983874324, + "mgd_loss_weight": 0.067152239382267, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.6066467285156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6258514404297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 27990, + "valid_token_count": 818.41484375 + }, + { + "actual_token_mask_ratio": 0.30065150558948517, + "epoch": 8.146639511201629, + "flow_matching_loss": 0.009880666434764863, + "grad_norm": 0.08759494125843048, + "kept_token_count": 572.33671875, + "learning_rate": 3.634159320980951e-07, + "loss": 0.0103, + "mgd_cos_sim": 0.9939812839031219, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006018720241263509, + "mgd_loss_weight": 0.06681889295578003, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.5978302001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.44166564941406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28000, + "valid_token_count": 818.38671875 + }, + { + "actual_token_mask_ratio": 0.2991171985864639, + "epoch": 8.149549025312773, + "flow_matching_loss": 0.01022108132019639, + "grad_norm": 0.05839881673455238, + "kept_token_count": 573.609375, + "learning_rate": 3.5980722090040663e-07, + "loss": 0.0106, + "mgd_cos_sim": 0.9938511073589325, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006148904841393232, + "mgd_loss_weight": 0.0664855495095253, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.5704040527344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5184265136719, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28010, + "valid_token_count": 818.4109375 + }, + { + "actual_token_mask_ratio": 0.30004872679710387, + "epoch": 8.152458539423916, + "flow_matching_loss": 0.009658019850030541, + "grad_norm": 0.08256764709949493, + "kept_token_count": 572.66171875, + "learning_rate": 3.5621629891742447e-07, + "loss": 0.01, + "mgd_cos_sim": 0.9941419184207916, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0058580843266099695, + "mgd_loss_weight": 0.0661522053182125, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.7007537841797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1876647949219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28020, + "valid_token_count": 818.14296875 + }, + { + "actual_token_mask_ratio": 0.29956085979938507, + "epoch": 8.15536805353506, + "flow_matching_loss": 0.012229623645544052, + "grad_norm": 0.08272276818752289, + "kept_token_count": 573.1921875, + "learning_rate": 3.526431705124583e-07, + "loss": 0.0126, + "mgd_cos_sim": 0.9940567314624786, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005943271005526185, + "mgd_loss_weight": 0.06581886112689972, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.2353759765625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8385498046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28030, + "valid_token_count": 818.33203125 + }, + { + "actual_token_mask_ratio": 0.3001368135213852, + "epoch": 8.158277567646204, + "flow_matching_loss": 0.011603662185370922, + "grad_norm": 0.086448535323143, + "kept_token_count": 572.575, + "learning_rate": 3.4908784002719517e-07, + "loss": 0.012, + "mgd_cos_sim": 0.9939004719257355, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006099525792524219, + "mgd_loss_weight": 0.06548551544547081, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.94446716308596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.07330017089845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28040, + "valid_token_count": 818.128125 + }, + { + "actual_token_mask_ratio": 0.2998820096254349, + "epoch": 8.161187081757346, + "flow_matching_loss": 0.010232607694342732, + "grad_norm": 0.07259451597929001, + "kept_token_count": 572.94609375, + "learning_rate": 3.4555031178169773e-07, + "loss": 0.0106, + "mgd_cos_sim": 0.9936117589473724, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006388244265690446, + "mgd_loss_weight": 0.06515217199921608, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.7014862060547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.07861328125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28050, + "valid_token_count": 818.359375 + }, + { + "actual_token_mask_ratio": 0.30018925964832305, + "epoch": 8.16409659586849, + "flow_matching_loss": 0.010293481405824423, + "grad_norm": 0.07755604386329651, + "kept_token_count": 572.5984375, + "learning_rate": 3.420305900743942e-07, + "loss": 0.0107, + "mgd_cos_sim": 0.9939955294132232, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006004461087286472, + "mgd_loss_weight": 0.0648188278079033, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.9408325195312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.68003845214844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28060, + "valid_token_count": 818.21328125 + }, + { + "actual_token_mask_ratio": 0.30031933486461637, + "epoch": 8.167006109979633, + "flow_matching_loss": 0.010318532539531588, + "grad_norm": 0.06725788116455078, + "kept_token_count": 572.5859375, + "learning_rate": 3.385286791820802e-07, + "loss": 0.0107, + "mgd_cos_sim": 0.9939063906669616, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006093595642596483, + "mgd_loss_weight": 0.06448548287153244, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.3451782226563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.3346893310547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28070, + "valid_token_count": 818.35234375 + }, + { + "actual_token_mask_ratio": 0.3002335995435715, + "epoch": 8.169915624090777, + "flow_matching_loss": 0.011307587334886193, + "grad_norm": 0.16938962042331696, + "kept_token_count": 572.671875, + "learning_rate": 3.350445833599086e-07, + "loss": 0.0117, + "mgd_cos_sim": 0.9937796413898468, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006220371928066015, + "mgd_loss_weight": 0.06415213793516159, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 408.26930847167966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4773773193359, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28080, + "valid_token_count": 818.37734375 + }, + { + "actual_token_mask_ratio": 0.2997549593448639, + "epoch": 8.17282513820192, + "flow_matching_loss": 0.009003295004367829, + "grad_norm": 0.072465680539608, + "kept_token_count": 572.9234375, + "learning_rate": 3.3157830684138303e-07, + "loss": 0.0094, + "mgd_cos_sim": 0.9939235866069793, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060764208901673555, + "mgd_loss_weight": 0.0638187937438488, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.4467834472656, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.75206604003904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28090, + "valid_token_count": 818.1734375 + }, + { + "actual_token_mask_ratio": 0.29937836825847625, + "epoch": 8.175734652313064, + "flow_matching_loss": 0.009696670714765787, + "grad_norm": 0.09757251292467117, + "kept_token_count": 573.38828125, + "learning_rate": 3.281298538383559e-07, + "loss": 0.0101, + "mgd_cos_sim": 0.9943057954311371, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0056942142080515625, + "mgd_loss_weight": 0.06348545029759407, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.6751251220703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0749206542969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28100, + "valid_token_count": 818.403125 + }, + { + "actual_token_mask_ratio": 0.2992480546236038, + "epoch": 8.178644166424208, + "flow_matching_loss": 0.010313367703929543, + "grad_norm": 0.07181833684444427, + "kept_token_count": 573.39453125, + "learning_rate": 3.2469922854102387e-07, + "loss": 0.0107, + "mgd_cos_sim": 0.9936483204364777, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006351686222478748, + "mgd_loss_weight": 0.06315210461616516, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.11854248046876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.96587829589845, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28110, + "valid_token_count": 818.25546875 + }, + { + "actual_token_mask_ratio": 0.3002230793237686, + "epoch": 8.18155368053535, + "flow_matching_loss": 0.011505938740447164, + "grad_norm": 0.07682354748249054, + "kept_token_count": 572.58125, + "learning_rate": 3.21286435117919e-07, + "loss": 0.0119, + "mgd_cos_sim": 0.993652093410492, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006347897369414568, + "mgd_loss_weight": 0.06281876042485238, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.71088256835935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8662048339844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28120, + "valid_token_count": 818.234375 + }, + { + "actual_token_mask_ratio": 0.29992042779922484, + "epoch": 8.184463194646494, + "flow_matching_loss": 0.009516542637720704, + "grad_norm": 0.06490957736968994, + "kept_token_count": 572.85625, + "learning_rate": 3.1789147771590754e-07, + "loss": 0.0099, + "mgd_cos_sim": 0.9938583016395569, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00614170148037374, + "mgd_loss_weight": 0.06248541735112667, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.9560882568359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.80002746582034, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28130, + "valid_token_count": 818.27578125 + }, + { + "actual_token_mask_ratio": 0.30009894669055937, + "epoch": 8.187372708757637, + "flow_matching_loss": 0.010755649721249937, + "grad_norm": 0.045559145510196686, + "kept_token_count": 572.92734375, + "learning_rate": 3.14514360460183e-07, + "loss": 0.0111, + "mgd_cos_sim": 0.9939309358596802, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060690675862133505, + "mgd_loss_weight": 0.06215207129716873, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.2476104736328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.89833984375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28140, + "valid_token_count": 818.5765625 + }, + { + "actual_token_mask_ratio": 0.30016416013240815, + "epoch": 8.190282222868781, + "flow_matching_loss": 0.009782345173880458, + "grad_norm": 0.06884779036045074, + "kept_token_count": 572.4671875, + "learning_rate": 3.1115508745426293e-07, + "loss": 0.0102, + "mgd_cos_sim": 0.9938110053539276, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006188985519111156, + "mgd_loss_weight": 0.061818727478384974, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.98094787597654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8536895751953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28150, + "valid_token_count": 817.99765625 + }, + { + "actual_token_mask_ratio": 0.30046428442001344, + "epoch": 8.193191736979925, + "flow_matching_loss": 0.010990609508007764, + "grad_norm": 0.08201050758361816, + "kept_token_count": 572.34609375, + "learning_rate": 3.0781366277997716e-07, + "loss": 0.0114, + "mgd_cos_sim": 0.9938863515853882, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00611365744844079, + "mgd_loss_weight": 0.06148538254201412, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.46441040039065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6768737792969, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28160, + "valid_token_count": 818.1859375 + }, + { + "actual_token_mask_ratio": 0.29943405389785765, + "epoch": 8.196101251091068, + "flow_matching_loss": 0.011619640933349728, + "grad_norm": 0.06214906647801399, + "kept_token_count": 573.3015625, + "learning_rate": 3.044900904974762e-07, + "loss": 0.012, + "mgd_cos_sim": 0.9939529299736023, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006047075754031539, + "mgd_loss_weight": 0.06115203872323036, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.439111328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3878692626953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28170, + "valid_token_count": 818.34453125 + }, + { + "actual_token_mask_ratio": 0.3002466678619385, + "epoch": 8.19901076520221, + "flow_matching_loss": 0.010214727511629463, + "grad_norm": 0.06399282068014145, + "kept_token_count": 572.54765625, + "learning_rate": 3.011843746452081e-07, + "loss": 0.0106, + "mgd_cos_sim": 0.9942207753658294, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005779227381572128, + "mgd_loss_weight": 0.060818693786859515, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.64040222167966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4732666015625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28180, + "valid_token_count": 818.21953125 + }, + { + "actual_token_mask_ratio": 0.3005491256713867, + "epoch": 8.201920279313354, + "flow_matching_loss": 0.011507660476490855, + "grad_norm": 0.05730153247714043, + "kept_token_count": 572.47890625, + "learning_rate": 2.9789651923993464e-07, + "loss": 0.0119, + "mgd_cos_sim": 0.9939261913299561, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006073798611760139, + "mgd_loss_weight": 0.060485349595546724, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.96271362304685, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.06235961914064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28190, + "valid_token_count": 818.46875 + }, + { + "actual_token_mask_ratio": 0.30030869841575625, + "epoch": 8.204829793424498, + "flow_matching_loss": 0.009330901503562927, + "grad_norm": 0.05041227117180824, + "kept_token_count": 572.484375, + "learning_rate": 2.9462652827670855e-07, + "loss": 0.0097, + "mgd_cos_sim": 0.9942805469036102, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005719430232420564, + "mgd_loss_weight": 0.060152005031704904, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.56007385253906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.655810546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28200, + "valid_token_count": 818.19765625 + }, + { + "actual_token_mask_ratio": 0.30033750236034396, + "epoch": 8.207739307535642, + "flow_matching_loss": 0.010220313351601363, + "grad_norm": 0.08525852113962173, + "kept_token_count": 572.5, + "learning_rate": 2.913744057288764e-07, + "loss": 0.0106, + "mgd_cos_sim": 0.9942092359066009, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005790744768455624, + "mgd_loss_weight": 0.05981866084039211, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.91542358398436, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1753814697266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28210, + "valid_token_count": 818.246875 + }, + { + "actual_token_mask_ratio": 0.30054239332675936, + "epoch": 8.210648821646785, + "flow_matching_loss": 0.011229633865877987, + "grad_norm": 0.11042515933513641, + "kept_token_count": 572.2859375, + "learning_rate": 2.8814015554807396e-07, + "loss": 0.0116, + "mgd_cos_sim": 0.9939857423305511, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006014254735782742, + "mgd_loss_weight": 0.059485315904021265, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 409.28672485351564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0124450683594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28220, + "valid_token_count": 818.18828125 + }, + { + "actual_token_mask_ratio": 0.29999034106731415, + "epoch": 8.213558335757929, + "flow_matching_loss": 0.010223965905606747, + "grad_norm": 0.0816899761557579, + "kept_token_count": 572.76640625, + "learning_rate": 2.849237816642192e-07, + "loss": 0.0106, + "mgd_cos_sim": 0.9940980315208435, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005901956046000123, + "mgd_loss_weight": 0.05915197171270847, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.82183837890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.5785888671875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28230, + "valid_token_count": 818.22890625 + }, + { + "actual_token_mask_ratio": 0.30034333765506743, + "epoch": 8.216467849869073, + "flow_matching_loss": 0.009614317631348967, + "grad_norm": 0.049192897975444794, + "kept_token_count": 572.38125, + "learning_rate": 2.817252879855109e-07, + "loss": 0.01, + "mgd_cos_sim": 0.9938870370388031, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061129707843065265, + "mgd_loss_weight": 0.05881862752139568, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.5970428466797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.97267150878906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28240, + "valid_token_count": 818.0953125 + }, + { + "actual_token_mask_ratio": 0.30044569075107574, + "epoch": 8.219377363980215, + "flow_matching_loss": 0.009108433127403259, + "grad_norm": 0.07239875197410583, + "kept_token_count": 572.6046875, + "learning_rate": 2.785446783984219e-07, + "loss": 0.0095, + "mgd_cos_sim": 0.9940067529678345, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059932509902864695, + "mgd_loss_weight": 0.058485282585024834, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.3011169433594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0570770263672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28250, + "valid_token_count": 818.5234375 + }, + { + "actual_token_mask_ratio": 0.29986388683319093, + "epoch": 8.222286878091358, + "flow_matching_loss": 0.008267336059361696, + "grad_norm": 0.06657763570547104, + "kept_token_count": 572.7703125, + "learning_rate": 2.7538195676769076e-07, + "loss": 0.0086, + "mgd_cos_sim": 0.9940743625164032, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005925642792135477, + "mgd_loss_weight": 0.05815193839371204, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.3376892089844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6272308349609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28260, + "valid_token_count": 818.0828125 + }, + { + "actual_token_mask_ratio": 0.29971199929714204, + "epoch": 8.225196392202502, + "flow_matching_loss": 0.010957116726785898, + "grad_norm": 0.059288810938596725, + "kept_token_count": 573.00390625, + "learning_rate": 2.722371269363216e-07, + "loss": 0.0113, + "mgd_cos_sim": 0.9939947903156281, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060052121989429, + "mgd_loss_weight": 0.05781859382987022, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.0481323242187, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.44781799316405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28270, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.2993417978286743, + "epoch": 8.228105906313646, + "flow_matching_loss": 0.010278696427121758, + "grad_norm": 0.06526695936918259, + "kept_token_count": 573.159375, + "learning_rate": 2.6911019272558125e-07, + "loss": 0.0106, + "mgd_cos_sim": 0.994265478849411, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0057345082517713305, + "mgd_loss_weight": 0.05748525001108647, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.77772216796876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5883056640625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28280, + "valid_token_count": 818.0296875 + }, + { + "actual_token_mask_ratio": 0.2998976320028305, + "epoch": 8.23101542042479, + "flow_matching_loss": 0.00882067522034049, + "grad_norm": 0.06365179270505905, + "kept_token_count": 572.7109375, + "learning_rate": 2.6600115793498857e-07, + "loss": 0.0092, + "mgd_cos_sim": 0.9939787685871124, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006021235836669803, + "mgd_loss_weight": 0.05715190470218658, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.6818084716797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.8804870605469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28290, + "valid_token_count": 818.0390625 + }, + { + "actual_token_mask_ratio": 0.2996610552072525, + "epoch": 8.233924934535933, + "flow_matching_loss": 0.010213427292183042, + "grad_norm": 0.07670968770980835, + "kept_token_count": 573.165625, + "learning_rate": 2.629100263423151e-07, + "loss": 0.0106, + "mgd_cos_sim": 0.9936978280544281, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063021636568009855, + "mgd_loss_weight": 0.05681856088340283, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.73984985351564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6769134521484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28300, + "valid_token_count": 818.409375 + }, + { + "actual_token_mask_ratio": 0.29985949099063874, + "epoch": 8.236834448647077, + "flow_matching_loss": 0.00855405405163765, + "grad_norm": 0.06961393356323242, + "kept_token_count": 572.759375, + "learning_rate": 2.598368017035763e-07, + "loss": 0.0089, + "mgd_cos_sim": 0.9937768638134002, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006223122915253043, + "mgd_loss_weight": 0.05648521594703197, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.27455139160156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8686187744141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28310, + "valid_token_count": 818.06796875 + }, + { + "actual_token_mask_ratio": 0.3002629429101944, + "epoch": 8.239743962758219, + "flow_matching_loss": 0.010505665140226484, + "grad_norm": 0.05666295066475868, + "kept_token_count": 572.67578125, + "learning_rate": 2.5678148775303e-07, + "loss": 0.0108, + "mgd_cos_sim": 0.9939309537410737, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006069039180874825, + "mgd_loss_weight": 0.056151872128248216, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.93128356933596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4006103515625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28320, + "valid_token_count": 818.41640625 + }, + { + "actual_token_mask_ratio": 0.29992068409919737, + "epoch": 8.242653476869362, + "flow_matching_loss": 0.010078654484823346, + "grad_norm": 0.07652249187231064, + "kept_token_count": 572.71484375, + "learning_rate": 2.537440882031683e-07, + "loss": 0.0104, + "mgd_cos_sim": 0.9939450263977051, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006054962333291769, + "mgd_loss_weight": 0.05581852681934833, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.7718933105469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9543762207031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28330, + "valid_token_count": 818.06796875 + }, + { + "actual_token_mask_ratio": 0.2999396711587906, + "epoch": 8.245562990980506, + "flow_matching_loss": 0.010773575305938721, + "grad_norm": 0.06910889595746994, + "kept_token_count": 572.8640625, + "learning_rate": 2.5072460674472385e-07, + "loss": 0.0111, + "mgd_cos_sim": 0.9940176784992218, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00598232913762331, + "mgd_loss_weight": 0.05548518300056458, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.4555633544922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.2901947021484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28340, + "valid_token_count": 818.30390625 + }, + { + "actual_token_mask_ratio": 0.2995790302753448, + "epoch": 8.24847250509165, + "flow_matching_loss": 0.008746144641190768, + "grad_norm": 0.1085871160030365, + "kept_token_count": 573.03046875, + "learning_rate": 2.4772304704664683e-07, + "loss": 0.0091, + "mgd_cos_sim": 0.9940972626209259, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005902737798169255, + "mgd_loss_weight": 0.05515183843672276, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.6004211425781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.79710998535154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28350, + "valid_token_count": 818.1203125 + }, + { + "actual_token_mask_ratio": 0.29964234828948977, + "epoch": 8.251382019202794, + "flow_matching_loss": 0.009577668784186245, + "grad_norm": 0.0732724666595459, + "kept_token_count": 573.0359375, + "learning_rate": 2.447394127561181e-07, + "loss": 0.0099, + "mgd_cos_sim": 0.9937717378139496, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062282771803438665, + "mgd_loss_weight": 0.05481849387288094, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.2925048828125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.39649658203126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28360, + "valid_token_count": 818.20390625 + }, + { + "actual_token_mask_ratio": 0.2995594501495361, + "epoch": 8.254291533313937, + "flow_matching_loss": 0.010518626868724823, + "grad_norm": 0.0747307762503624, + "kept_token_count": 573.12890625, + "learning_rate": 2.41773707498531e-07, + "loss": 0.0109, + "mgd_cos_sim": 0.99370836019516, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006291639897972346, + "mgd_loss_weight": 0.05448514930903912, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.2975738525391, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4259918212891, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28370, + "valid_token_count": 818.2421875 + }, + { + "actual_token_mask_ratio": 0.30053846836090087, + "epoch": 8.25720104742508, + "flow_matching_loss": 0.009533291961997748, + "grad_norm": 0.06774912029504776, + "kept_token_count": 572.2671875, + "learning_rate": 2.388259348774996e-07, + "loss": 0.0099, + "mgd_cos_sim": 0.9939452290534974, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006054774625226855, + "mgd_loss_weight": 0.054151805117726326, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.07016906738284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.01807861328126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28380, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.30047839879989624, + "epoch": 8.260110561536223, + "flow_matching_loss": 0.009844665788114071, + "grad_norm": 0.06792343407869339, + "kept_token_count": 572.38515625, + "learning_rate": 2.358960984748454e-07, + "loss": 0.0102, + "mgd_cos_sim": 0.9939119875431061, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006088031781837344, + "mgd_loss_weight": 0.053818460926413535, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.53106384277345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8970123291016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28390, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.29954981803894043, + "epoch": 8.263020075647367, + "flow_matching_loss": 0.008601206727325917, + "grad_norm": 0.05930468812584877, + "kept_token_count": 573.1484375, + "learning_rate": 2.3298420185059565e-07, + "loss": 0.0089, + "mgd_cos_sim": 0.9943024694919587, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005697536980733275, + "mgd_loss_weight": 0.053485115990042686, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.2947235107422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.03464660644534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28400, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.30077214241027833, + "epoch": 8.26592958975851, + "flow_matching_loss": 0.01053175306878984, + "grad_norm": 0.07641749083995819, + "kept_token_count": 572.13203125, + "learning_rate": 2.3009024854297844e-07, + "loss": 0.0108, + "mgd_cos_sim": 0.9940989255905152, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005901078833267093, + "mgd_loss_weight": 0.053151771798729895, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.6764221191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.21300354003904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28410, + "valid_token_count": 818.23359375 + }, + { + "actual_token_mask_ratio": 0.300017973780632, + "epoch": 8.268839103869654, + "flow_matching_loss": 0.009384648222476245, + "grad_norm": 0.08064619451761246, + "kept_token_count": 572.64296875, + "learning_rate": 2.272142420684209e-07, + "loss": 0.0097, + "mgd_cos_sim": 0.9939060747623444, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060939304530620575, + "mgd_loss_weight": 0.052818427234888075, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.9672576904297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3925384521484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28420, + "valid_token_count": 818.08359375 + }, + { + "actual_token_mask_ratio": 0.300225630402565, + "epoch": 8.271748617980798, + "flow_matching_loss": 0.01061623073183, + "grad_norm": 0.07580302655696869, + "kept_token_count": 572.49140625, + "learning_rate": 2.243561859215426e-07, + "loss": 0.0109, + "mgd_cos_sim": 0.9939808309078216, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006019166065379977, + "mgd_loss_weight": 0.052485083043575284, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.5325012207031, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1710235595703, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28430, + "valid_token_count": 818.10625 + }, + { + "actual_token_mask_ratio": 0.30109214782714844, + "epoch": 8.274658132091941, + "flow_matching_loss": 0.009816049132496119, + "grad_norm": 0.05469954386353493, + "kept_token_count": 571.82421875, + "learning_rate": 2.2151608357514897e-07, + "loss": 0.0102, + "mgd_cos_sim": 0.9935601592063904, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006439833110198379, + "mgd_loss_weight": 0.052151738107204436, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.79795227050784, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1715057373047, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28440, + "valid_token_count": 818.16484375 + }, + { + "actual_token_mask_ratio": 0.3004545569419861, + "epoch": 8.277567646203083, + "flow_matching_loss": 0.009368537040427328, + "grad_norm": 0.05851605907082558, + "kept_token_count": 572.65078125, + "learning_rate": 2.186939384802361e-07, + "loss": 0.0097, + "mgd_cos_sim": 0.9939322948455811, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006067702081054449, + "mgd_loss_weight": 0.05181839428842068, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.37205505371094, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.9684356689453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28450, + "valid_token_count": 818.6078125 + }, + { + "actual_token_mask_ratio": 0.29979380369186404, + "epoch": 8.280477160314227, + "flow_matching_loss": 0.009498688858002424, + "grad_norm": 0.06626985222101212, + "kept_token_count": 572.9765625, + "learning_rate": 2.1588975406597266e-07, + "loss": 0.0098, + "mgd_cos_sim": 0.9936965346336365, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006303474772721529, + "mgd_loss_weight": 0.051485049352049825, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.0104949951172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.11187133789065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28460, + "valid_token_count": 818.29765625 + }, + { + "actual_token_mask_ratio": 0.2994048446416855, + "epoch": 8.28338667442537, + "flow_matching_loss": 0.009711607964709401, + "grad_norm": 0.06897349655628204, + "kept_token_count": 573.275, + "learning_rate": 2.131035337397097e-07, + "loss": 0.01, + "mgd_cos_sim": 0.994134658575058, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005865326104685664, + "mgd_loss_weight": 0.05115170516073704, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.4256317138672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9908508300781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28470, + "valid_token_count": 818.2703125 + }, + { + "actual_token_mask_ratio": 0.29955979585647585, + "epoch": 8.286296188536515, + "flow_matching_loss": 0.010425103921443224, + "grad_norm": 0.07120785117149353, + "kept_token_count": 573.2015625, + "learning_rate": 2.1033528088696575e-07, + "loss": 0.0108, + "mgd_cos_sim": 0.9935660243034363, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006433997955173254, + "mgd_loss_weight": 0.050818360224366185, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.2814636230469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.5116821289063, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28480, + "valid_token_count": 818.3453125 + }, + { + "actual_token_mask_ratio": 0.30016333758831026, + "epoch": 8.289205702647658, + "flow_matching_loss": 0.009621949633583426, + "grad_norm": 0.06850314885377884, + "kept_token_count": 572.64140625, + "learning_rate": 2.0758499887143012e-07, + "loss": 0.0099, + "mgd_cos_sim": 0.9938449919223785, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006155008403584361, + "mgd_loss_weight": 0.05048501640558243, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.92633361816405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0405670166016, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28490, + "valid_token_count": 818.24921875 + }, + { + "actual_token_mask_ratio": 0.30027145743370054, + "epoch": 8.292115216758802, + "flow_matching_loss": 0.00998902809806168, + "grad_norm": 0.06409192830324173, + "kept_token_count": 572.546875, + "learning_rate": 2.0485269103495462e-07, + "loss": 0.0103, + "mgd_cos_sim": 0.9937776148319244, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062223758548498155, + "mgd_loss_weight": 0.05015167184174061, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.71009826660156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9839782714844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28500, + "valid_token_count": 818.2390625 + }, + { + "actual_token_mask_ratio": 0.3004023492336273, + "epoch": 8.295024730869944, + "flow_matching_loss": 0.009729495598003268, + "grad_norm": 0.06824012845754623, + "kept_token_count": 572.56875, + "learning_rate": 2.021383606975502e-07, + "loss": 0.01, + "mgd_cos_sim": 0.9940390467643738, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005960944015532732, + "mgd_loss_weight": 0.04981832727789879, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.06251220703126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.6892120361328, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28510, + "valid_token_count": 818.4265625 + }, + { + "actual_token_mask_ratio": 0.29937881231307983, + "epoch": 8.297934244981088, + "flow_matching_loss": 0.009101786045357585, + "grad_norm": 0.06556191295385361, + "kept_token_count": 573.08203125, + "learning_rate": 1.9944201115738524e-07, + "loss": 0.0094, + "mgd_cos_sim": 0.9939276456832886, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060723601840436455, + "mgd_loss_weight": 0.04948498271405697, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.47170104980466, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.18182373046875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28520, + "valid_token_count": 817.96875 + }, + { + "actual_token_mask_ratio": 0.29996263086795805, + "epoch": 8.300843759092231, + "flow_matching_loss": 0.010223600175231695, + "grad_norm": 0.057627081871032715, + "kept_token_count": 572.88515625, + "learning_rate": 1.9676364569077732e-07, + "loss": 0.0105, + "mgd_cos_sim": 0.9939610302448273, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00603897413238883, + "mgd_loss_weight": 0.04915163852274418, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.51231384277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3943206787109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28530, + "valid_token_count": 818.35859375 + }, + { + "actual_token_mask_ratio": 0.2996535062789917, + "epoch": 8.303753273203375, + "flow_matching_loss": 0.009686964144930244, + "grad_norm": 0.07980683445930481, + "kept_token_count": 573.27265625, + "learning_rate": 1.9410326755219477e-07, + "loss": 0.01, + "mgd_cos_sim": 0.9938896834850312, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006110312510281801, + "mgd_loss_weight": 0.04881829395890236, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.21556701660154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.89080505371095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28540, + "valid_token_count": 818.55859375 + }, + { + "actual_token_mask_ratio": 0.3002362847328186, + "epoch": 8.306662787314519, + "flow_matching_loss": 0.009390911692753435, + "grad_norm": 0.09130669385194778, + "kept_token_count": 572.81796875, + "learning_rate": 1.9146087997424677e-07, + "loss": 0.0097, + "mgd_cos_sim": 0.9938159584999084, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006184036284685135, + "mgd_loss_weight": 0.04848494939506054, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.6795227050781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.26642150878905, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28550, + "valid_token_count": 818.58671875 + }, + { + "actual_token_mask_ratio": 0.2996431440114975, + "epoch": 8.309572301425662, + "flow_matching_loss": 0.010684772999957204, + "grad_norm": 0.0689656212925911, + "kept_token_count": 573.06796875, + "learning_rate": 1.8883648616768335e-07, + "loss": 0.011, + "mgd_cos_sim": 0.9936915874481201, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00630841888487339, + "mgd_loss_weight": 0.04815160520374775, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.0387725830078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.66295166015624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28560, + "valid_token_count": 818.25078125 + }, + { + "actual_token_mask_ratio": 0.2994756788015366, + "epoch": 8.312481815536806, + "flow_matching_loss": 0.009559794003143907, + "grad_norm": 0.10089007765054703, + "kept_token_count": 573.23046875, + "learning_rate": 1.862300893213903e-07, + "loss": 0.0098, + "mgd_cos_sim": 0.9939904987812043, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006009479612112045, + "mgd_loss_weight": 0.04781826063990593, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.90790405273435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.71848754882814, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28570, + "valid_token_count": 818.28671875 + }, + { + "actual_token_mask_ratio": 0.3001377433538437, + "epoch": 8.315391329647948, + "flow_matching_loss": 0.00907090730033815, + "grad_norm": 0.0751652717590332, + "kept_token_count": 572.60625, + "learning_rate": 1.8364169260238595e-07, + "loss": 0.0094, + "mgd_cos_sim": 0.993958979845047, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060410141479223965, + "mgd_loss_weight": 0.04748491607606411, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.34807739257815, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.0706512451172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28580, + "valid_token_count": 818.1703125 + }, + { + "actual_token_mask_ratio": 0.30015321969985964, + "epoch": 8.318300843759092, + "flow_matching_loss": 0.01039059804752469, + "grad_norm": 0.06605572998523712, + "kept_token_count": 572.715625, + "learning_rate": 1.8107129915581777e-07, + "loss": 0.0107, + "mgd_cos_sim": 0.9940691232681275, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005930886277928949, + "mgd_loss_weight": 0.04715157151222229, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.4760437011719, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.12420349121095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28590, + "valid_token_count": 818.35234375 + }, + { + "actual_token_mask_ratio": 0.2996160864830017, + "epoch": 8.321210357870235, + "flow_matching_loss": 0.00939859845675528, + "grad_norm": 0.0622875802218914, + "kept_token_count": 572.96328125, + "learning_rate": 1.785189121049541e-07, + "loss": 0.0097, + "mgd_cos_sim": 0.9938378512859345, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006162140052765608, + "mgd_loss_weight": 0.04681822769343853, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.91517944335936, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2299285888672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28600, + "valid_token_count": 818.07109375 + }, + { + "actual_token_mask_ratio": 0.3001125305891037, + "epoch": 8.32411987198138, + "flow_matching_loss": 0.009830823168158531, + "grad_norm": 0.06965535134077072, + "kept_token_count": 572.63359375, + "learning_rate": 1.7598453455118567e-07, + "loss": 0.0101, + "mgd_cos_sim": 0.9941229224205017, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005877073341980577, + "mgd_loss_weight": 0.04648488238453865, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.31792602539065, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.67096252441405, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28610, + "valid_token_count": 818.1796875 + }, + { + "actual_token_mask_ratio": 0.29980174601078036, + "epoch": 8.327029386092523, + "flow_matching_loss": 0.011006917990744113, + "grad_norm": 0.0708344578742981, + "kept_token_count": 572.8140625, + "learning_rate": 1.7346816957402256e-07, + "loss": 0.0113, + "mgd_cos_sim": 0.9939202904701233, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006079724757000804, + "mgd_loss_weight": 0.04615153856575489, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.4767822265625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.1541442871094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28620, + "valid_token_count": 818.0671875 + }, + { + "actual_token_mask_ratio": 0.30030917525291445, + "epoch": 8.329938900203667, + "flow_matching_loss": 0.009325246885418892, + "grad_norm": 0.07213705033063889, + "kept_token_count": 572.7921875, + "learning_rate": 1.7096982023108389e-07, + "loss": 0.0096, + "mgd_cos_sim": 0.9942388713359833, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005761139141395688, + "mgd_loss_weight": 0.04581819362938404, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.14528198242186, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3303558349609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28630, + "valid_token_count": 818.6375 + }, + { + "actual_token_mask_ratio": 0.3002126902341843, + "epoch": 8.33284841431481, + "flow_matching_loss": 0.00953470729291439, + "grad_norm": 0.04424956813454628, + "kept_token_count": 572.64453125, + "learning_rate": 1.6848948955809963e-07, + "loss": 0.0098, + "mgd_cos_sim": 0.9936225950717926, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006377412891015411, + "mgd_loss_weight": 0.04548484981060028, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.6357452392578, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9029571533203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28640, + "valid_token_count": 818.3046875 + }, + { + "actual_token_mask_ratio": 0.2997624635696411, + "epoch": 8.335757928425952, + "flow_matching_loss": 0.010393355879932642, + "grad_norm": 0.08839250355958939, + "kept_token_count": 572.98359375, + "learning_rate": 1.66027180568904e-07, + "loss": 0.0107, + "mgd_cos_sim": 0.9943066358566284, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005693362047895789, + "mgd_loss_weight": 0.045151504874229434, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.56842346191405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.36566772460935, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28650, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.3004460871219635, + "epoch": 8.338667442537096, + "flow_matching_loss": 0.01001648181118071, + "grad_norm": 0.04964075982570648, + "kept_token_count": 572.23359375, + "learning_rate": 1.6358289625543865e-07, + "loss": 0.0103, + "mgd_cos_sim": 0.9941117525100708, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005888233846053481, + "mgd_loss_weight": 0.04481816068291664, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.91405334472654, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.68479919433594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28660, + "valid_token_count": 817.9984375 + }, + { + "actual_token_mask_ratio": 0.29942054450511935, + "epoch": 8.34157695664824, + "flow_matching_loss": 0.00978551316075027, + "grad_norm": 0.0633196011185646, + "kept_token_count": 573.28671875, + "learning_rate": 1.6115663958773618e-07, + "loss": 0.0101, + "mgd_cos_sim": 0.9935638129711151, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006436179578304291, + "mgd_loss_weight": 0.04448481611907482, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.8685668945312, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1217102050781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28670, + "valid_token_count": 818.2984375 + }, + { + "actual_token_mask_ratio": 0.3004224509000778, + "epoch": 8.344486470759383, + "flow_matching_loss": 0.010392466001212597, + "grad_norm": 0.06382670253515244, + "kept_token_count": 572.40234375, + "learning_rate": 1.5874841351392832e-07, + "loss": 0.0107, + "mgd_cos_sim": 0.9939417004585266, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006058285944163799, + "mgd_loss_weight": 0.04415147192776203, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.4899017333984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.27630615234375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28680, + "valid_token_count": 818.2125 + }, + { + "actual_token_mask_ratio": 0.2998449742794037, + "epoch": 8.347395984870527, + "flow_matching_loss": 0.012478657066822052, + "grad_norm": 0.07363777607679367, + "kept_token_count": 572.8765625, + "learning_rate": 1.56358220960236e-07, + "loss": 0.0127, + "mgd_cos_sim": 0.9942707300186158, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005729266395792365, + "mgd_loss_weight": 0.04381812699139118, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4816528320313, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.98777770996094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28690, + "valid_token_count": 818.209375 + }, + { + "actual_token_mask_ratio": 0.2997004151344299, + "epoch": 8.35030549898167, + "flow_matching_loss": 0.009735916974022985, + "grad_norm": 0.075860895216465, + "kept_token_count": 573.03828125, + "learning_rate": 1.5398606483096932e-07, + "loss": 0.01, + "mgd_cos_sim": 0.9938099145889282, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006190085597336292, + "mgd_loss_weight": 0.04348478280007839, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.04737548828126, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3848571777344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28700, + "valid_token_count": 818.27578125 + }, + { + "actual_token_mask_ratio": 0.2993279695510864, + "epoch": 8.353215013092813, + "flow_matching_loss": 0.01116121867671609, + "grad_norm": 0.08688245713710785, + "kept_token_count": 573.38046875, + "learning_rate": 1.5163194800852265e-07, + "loss": 0.0114, + "mgd_cos_sim": 0.9941309869289399, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005869004456326365, + "mgd_loss_weight": 0.0431514386087656, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.16391906738284, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.26842041015624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28710, + "valid_token_count": 818.33359375 + }, + { + "actual_token_mask_ratio": 0.30060724914073944, + "epoch": 8.356124527203956, + "flow_matching_loss": 0.011516337702050804, + "grad_norm": 0.08333424478769302, + "kept_token_count": 572.20546875, + "learning_rate": 1.4929587335336614e-07, + "loss": 0.0118, + "mgd_cos_sim": 0.993825089931488, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006174919195473194, + "mgd_loss_weight": 0.04281809367239475, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.06299743652346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.00291748046874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28720, + "valid_token_count": 818.14609375 + }, + { + "actual_token_mask_ratio": 0.3004763096570969, + "epoch": 8.3590340413151, + "flow_matching_loss": 0.010152529831975699, + "grad_norm": 0.07782169431447983, + "kept_token_count": 572.17421875, + "learning_rate": 1.4697784370405587e-07, + "loss": 0.0104, + "mgd_cos_sim": 0.9939689218997956, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006031099101528525, + "mgd_loss_weight": 0.04248474948108196, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.96640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7379425048828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28730, + "valid_token_count": 817.94921875 + }, + { + "actual_token_mask_ratio": 0.29961180686950684, + "epoch": 8.361943555426244, + "flow_matching_loss": 0.01010993095114827, + "grad_norm": 0.07217292487621307, + "kept_token_count": 573.11953125, + "learning_rate": 1.4467786187721376e-07, + "loss": 0.0104, + "mgd_cos_sim": 0.9941180527210236, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005881946906447411, + "mgd_loss_weight": 0.04215140491724014, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.99729309082034, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.50877685546874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28740, + "valid_token_count": 818.2859375 + }, + { + "actual_token_mask_ratio": 0.2998366683721542, + "epoch": 8.364853069537387, + "flow_matching_loss": 0.00968235805630684, + "grad_norm": 0.10574992001056671, + "kept_token_count": 572.90390625, + "learning_rate": 1.4239593066753265e-07, + "loss": 0.0099, + "mgd_cos_sim": 0.9941702663898468, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005829741479828954, + "mgd_loss_weight": 0.041818061098456386, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.172607421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2345306396484, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28750, + "valid_token_count": 818.23984375 + }, + { + "actual_token_mask_ratio": 0.3001610994338989, + "epoch": 8.367762583648531, + "flow_matching_loss": 0.00917446380481124, + "grad_norm": 0.06529247760772705, + "kept_token_count": 572.66171875, + "learning_rate": 1.4013205284777953e-07, + "loss": 0.0094, + "mgd_cos_sim": 0.993830394744873, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061696108896285295, + "mgd_loss_weight": 0.0414847157895565, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.1306945800781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.65502014160154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28760, + "valid_token_count": 818.27890625 + }, + { + "actual_token_mask_ratio": 0.30038235187530515, + "epoch": 8.370672097759675, + "flow_matching_loss": 0.009902820270508528, + "grad_norm": 0.07343120127916336, + "kept_token_count": 572.3078125, + "learning_rate": 1.3788623116877397e-07, + "loss": 0.0101, + "mgd_cos_sim": 0.994099748134613, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00590024571865797, + "mgd_loss_weight": 0.041151371970772746, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.7604217529297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1840087890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28770, + "valid_token_count": 818.03125 + }, + { + "actual_token_mask_ratio": 0.29975679218769075, + "epoch": 8.373581611870817, + "flow_matching_loss": 0.009431817196309566, + "grad_norm": 0.060220152139663696, + "kept_token_count": 573.0015625, + "learning_rate": 1.3565846835940477e-07, + "loss": 0.0097, + "mgd_cos_sim": 0.9939605116844177, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006039491668343544, + "mgd_loss_weight": 0.04081802703440189, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.45228576660156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4510192871094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28780, + "valid_token_count": 818.28984375 + }, + { + "actual_token_mask_ratio": 0.29974907338619233, + "epoch": 8.37649112598196, + "flow_matching_loss": 0.010173796443268657, + "grad_norm": 0.07172124087810516, + "kept_token_count": 573.0859375, + "learning_rate": 1.334487671266116e-07, + "loss": 0.0104, + "mgd_cos_sim": 0.9939801812171936, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060198221821337935, + "mgd_loss_weight": 0.040484683215618135, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.31385498046876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7983001708984, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28790, + "valid_token_count": 818.39765625 + }, + { + "actual_token_mask_ratio": 0.29976093769073486, + "epoch": 8.379400640093104, + "flow_matching_loss": 0.00931561691686511, + "grad_norm": 0.07283526659011841, + "kept_token_count": 572.95703125, + "learning_rate": 1.3125713015538998e-07, + "loss": 0.0096, + "mgd_cos_sim": 0.9941003382205963, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005899685993790626, + "mgd_loss_weight": 0.04015133790671825, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.63316345214844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.82545471191406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28800, + "valid_token_count": 818.23203125 + }, + { + "actual_token_mask_ratio": 0.29937746822834016, + "epoch": 8.382310154204248, + "flow_matching_loss": 0.009833482978865504, + "grad_norm": 0.06880418211221695, + "kept_token_count": 573.32890625, + "learning_rate": 1.290835601087864e-07, + "loss": 0.0101, + "mgd_cos_sim": 0.9938560247421264, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006143971020355821, + "mgd_loss_weight": 0.039817994087934495, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.87928466796876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2877136230469, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28810, + "valid_token_count": 818.30625 + }, + { + "actual_token_mask_ratio": 0.29916462004184724, + "epoch": 8.385219668315392, + "flow_matching_loss": 0.009356471244245768, + "grad_norm": 0.07700514793395996, + "kept_token_count": 573.39765625, + "learning_rate": 1.2692805962789323e-07, + "loss": 0.0096, + "mgd_cos_sim": 0.9941298425197601, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005870163580402732, + "mgd_loss_weight": 0.039484649524092676, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.6294860839844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.7311492919922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28820, + "valid_token_count": 818.16484375 + }, + { + "actual_token_mask_ratio": 0.29931462109088897, + "epoch": 8.388129182426535, + "flow_matching_loss": 0.009588324883952736, + "grad_norm": 0.051428649574518204, + "kept_token_count": 573.50859375, + "learning_rate": 1.2479063133184698e-07, + "loss": 0.0098, + "mgd_cos_sim": 0.993830019235611, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006169981230050326, + "mgd_loss_weight": 0.039151304960250856, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.2195617675781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.23198852539065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28830, + "valid_token_count": 818.49921875 + }, + { + "actual_token_mask_ratio": 0.2992189347743988, + "epoch": 8.391038696537677, + "flow_matching_loss": 0.009377100551500917, + "grad_norm": 0.08693193644285202, + "kept_token_count": 573.52578125, + "learning_rate": 1.2267127781782516e-07, + "loss": 0.0096, + "mgd_cos_sim": 0.9940289735794068, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005971025256440043, + "mgd_loss_weight": 0.038817960396409036, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.5398376464844, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1420654296875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28840, + "valid_token_count": 818.41171875 + }, + { + "actual_token_mask_ratio": 0.2990329027175903, + "epoch": 8.393948210648821, + "flow_matching_loss": 0.009773602057248355, + "grad_norm": 0.11993909627199173, + "kept_token_count": 573.53984375, + "learning_rate": 1.205700016610395e-07, + "loss": 0.01, + "mgd_cos_sim": 0.9939966917037963, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006003311136737466, + "mgd_loss_weight": 0.038484616205096245, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.4563049316406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.42612915039064, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28850, + "valid_token_count": 818.21328125 + }, + { + "actual_token_mask_ratio": 0.3002693593502045, + "epoch": 8.396857724759965, + "flow_matching_loss": 0.009862388670444488, + "grad_norm": 0.06017151474952698, + "kept_token_count": 572.49453125, + "learning_rate": 1.1848680541474088e-07, + "loss": 0.0101, + "mgd_cos_sim": 0.9941837191581726, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005816275347024202, + "mgd_loss_weight": 0.038151272013783454, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.5700408935547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.45325927734376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28860, + "valid_token_count": 818.16796875 + }, + { + "actual_token_mask_ratio": 0.3001409977674484, + "epoch": 8.399767238871108, + "flow_matching_loss": 0.00944917551241815, + "grad_norm": 0.0536467470228672, + "kept_token_count": 572.7453125, + "learning_rate": 1.1642169161020955e-07, + "loss": 0.0097, + "mgd_cos_sim": 0.9941576182842254, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005842371238395571, + "mgd_loss_weight": 0.037817927077412605, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.7395477294922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5395538330078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28870, + "valid_token_count": 818.36875 + }, + { + "actual_token_mask_ratio": 0.2998346447944641, + "epoch": 8.402676752982252, + "flow_matching_loss": 0.009308456722646952, + "grad_norm": 0.08579215407371521, + "kept_token_count": 572.8765625, + "learning_rate": 1.1437466275675156e-07, + "loss": 0.0095, + "mgd_cos_sim": 0.9939164340496063, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006083567533642054, + "mgd_loss_weight": 0.037484582886099814, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.0504852294922, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7184326171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28880, + "valid_token_count": 818.20078125 + }, + { + "actual_token_mask_ratio": 0.29921739399433134, + "epoch": 8.405586267093396, + "flow_matching_loss": 0.010296686133369804, + "grad_norm": 0.06539381295442581, + "kept_token_count": 573.33515625, + "learning_rate": 1.1234572134170229e-07, + "loss": 0.0105, + "mgd_cos_sim": 0.9934542655944825, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006545743718743324, + "mgd_loss_weight": 0.037151238322257994, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.38177490234375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.99000244140626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28890, + "valid_token_count": 818.13125 + }, + { + "actual_token_mask_ratio": 0.30013962984085085, + "epoch": 8.40849578120454, + "flow_matching_loss": 0.009152676723897458, + "grad_norm": 0.06716343015432358, + "kept_token_count": 572.6796875, + "learning_rate": 1.1033486983041463e-07, + "loss": 0.0094, + "mgd_cos_sim": 0.9939739406108856, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006026063906028866, + "mgd_loss_weight": 0.0368178941309452, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.22381286621095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.0915496826172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28900, + "valid_token_count": 818.271875 + }, + { + "actual_token_mask_ratio": 0.3003580182790756, + "epoch": 8.411405295315681, + "flow_matching_loss": 0.009795419638976455, + "grad_norm": 0.0779128298163414, + "kept_token_count": 572.515625, + "learning_rate": 1.0834211066626409e-07, + "loss": 0.01, + "mgd_cos_sim": 0.9941695749759674, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00583043503575027, + "mgd_loss_weight": 0.036484549194574355, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.49338989257814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.30992736816404, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28910, + "valid_token_count": 818.29296875 + }, + { + "actual_token_mask_ratio": 0.2998332053422928, + "epoch": 8.414314809426825, + "flow_matching_loss": 0.008958851173520088, + "grad_norm": 0.06591514497995377, + "kept_token_count": 572.6875, + "learning_rate": 1.0636744627064044e-07, + "loss": 0.0092, + "mgd_cos_sim": 0.9940848529338837, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005915143666788936, + "mgd_loss_weight": 0.0361512053757906, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.3180786132813, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8026397705078, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28920, + "valid_token_count": 817.93046875 + }, + { + "actual_token_mask_ratio": 0.3004133403301239, + "epoch": 8.417224323537969, + "flow_matching_loss": 0.01066279043443501, + "grad_norm": 0.061733428388834, + "kept_token_count": 572.36953125, + "learning_rate": 1.0441087904294766e-07, + "loss": 0.0109, + "mgd_cos_sim": 0.9937541127204895, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006245893379673362, + "mgd_loss_weight": 0.035817860439419744, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.8084381103516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.41033935546875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28930, + "valid_token_count": 818.15234375 + }, + { + "actual_token_mask_ratio": 0.299202024936676, + "epoch": 8.420133837649113, + "flow_matching_loss": 0.008471819292753934, + "grad_norm": 0.06586987525224686, + "kept_token_count": 573.359375, + "learning_rate": 1.0247241136059904e-07, + "loss": 0.0087, + "mgd_cos_sim": 0.993886137008667, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061138535384088755, + "mgd_loss_weight": 0.03548451624810696, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.89259338378906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.3837951660156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28940, + "valid_token_count": 818.15234375 + }, + { + "actual_token_mask_ratio": 0.3002294212579727, + "epoch": 8.423043351760256, + "flow_matching_loss": 0.009657525131478906, + "grad_norm": 0.06952618807554245, + "kept_token_count": 572.50859375, + "learning_rate": 1.0055204557901709e-07, + "loss": 0.0099, + "mgd_cos_sim": 0.9938133776187896, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006186618329957127, + "mgd_loss_weight": 0.035151171311736104, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.27643127441405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.74432678222655, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28950, + "valid_token_count": 818.1375 + }, + { + "actual_token_mask_ratio": 0.30021682679653167, + "epoch": 8.4259528658714, + "flow_matching_loss": 0.010613431129604578, + "grad_norm": 0.07961950451135635, + "kept_token_count": 572.61484375, + "learning_rate": 9.864978403162361e-08, + "loss": 0.0108, + "mgd_cos_sim": 0.9939443469047546, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006055662781000138, + "mgd_loss_weight": 0.03481782749295235, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.7842681884766, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.08017578125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28960, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.2992885112762451, + "epoch": 8.428862379982544, + "flow_matching_loss": 0.009779749158769845, + "grad_norm": 0.05901376157999039, + "kept_token_count": 573.28828125, + "learning_rate": 9.676562902984965e-08, + "loss": 0.01, + "mgd_cos_sim": 0.9939699232578277, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006030081445351243, + "mgd_loss_weight": 0.03448448292911053, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.4050567626953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2609008789062, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28970, + "valid_token_count": 818.15 + }, + { + "actual_token_mask_ratio": 0.29887051582336427, + "epoch": 8.431771894093686, + "flow_matching_loss": 0.008979959692806006, + "grad_norm": 0.055424146354198456, + "kept_token_count": 573.69609375, + "learning_rate": 9.489958286311883e-08, + "loss": 0.0092, + "mgd_cos_sim": 0.9941327333450317, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00586726083420217, + "mgd_loss_weight": 0.03415113836526871, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.6708129882812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.14317626953124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28980, + "valid_token_count": 818.24375 + }, + { + "actual_token_mask_ratio": 0.3000608056783676, + "epoch": 8.43468140820483, + "flow_matching_loss": 0.009485973743721843, + "grad_norm": 0.06441959738731384, + "kept_token_count": 572.93359375, + "learning_rate": 9.30516477988541e-08, + "loss": 0.0097, + "mgd_cos_sim": 0.9939166605472565, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006083335215225816, + "mgd_loss_weight": 0.03381779380142689, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.82818908691405, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.85857849121095, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 28990, + "valid_token_count": 818.5484375 + }, + { + "actual_token_mask_ratio": 0.2993915379047394, + "epoch": 8.437590922315973, + "flow_matching_loss": 0.009401965932920576, + "grad_norm": 0.09470264613628387, + "kept_token_count": 573.225, + "learning_rate": 9.122182608247098e-08, + "loss": 0.0096, + "mgd_cos_sim": 0.9936302781105042, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006369711179286241, + "mgd_loss_weight": 0.0334844496101141, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.7141082763672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.4733856201172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29000, + "valid_token_count": 818.184375 + }, + { + "actual_token_mask_ratio": 0.3002982050180435, + "epoch": 8.440500436427117, + "flow_matching_loss": 0.009321665484458208, + "grad_norm": 0.06838108599185944, + "kept_token_count": 572.653125, + "learning_rate": 8.94101199373759e-08, + "loss": 0.0095, + "mgd_cos_sim": 0.9940616667270661, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005938344541937113, + "mgd_loss_weight": 0.03315110504627228, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.37986145019534, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.99879760742186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29010, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.3005917459726334, + "epoch": 8.44340995053826, + "flow_matching_loss": 0.008891465375199914, + "grad_norm": 0.05008366331458092, + "kept_token_count": 572.31015625, + "learning_rate": 8.761653156495964e-08, + "loss": 0.0091, + "mgd_cos_sim": 0.9940334141254425, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005966588482260704, + "mgd_loss_weight": 0.03281776048243046, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.4332733154297, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.40819091796874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29020, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.2997386485338211, + "epoch": 8.446319464649404, + "flow_matching_loss": 0.0101965076290071, + "grad_norm": 0.0811019241809845, + "kept_token_count": 572.99921875, + "learning_rate": 8.584106314460549e-08, + "loss": 0.0104, + "mgd_cos_sim": 0.9941033363342285, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005896651744842529, + "mgd_loss_weight": 0.03248441629111767, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.7487060546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2378387451172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29030, + "valid_token_count": 818.2609375 + }, + { + "actual_token_mask_ratio": 0.29921787679195405, + "epoch": 8.449228978760548, + "flow_matching_loss": 0.010564792761579156, + "grad_norm": 0.08766581863164902, + "kept_token_count": 573.3359375, + "learning_rate": 8.408371683367111e-08, + "loss": 0.0108, + "mgd_cos_sim": 0.9939016044139862, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006098399963229894, + "mgd_loss_weight": 0.03215107172727585, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.3496368408203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.64599609375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29040, + "valid_token_count": 818.13515625 + }, + { + "actual_token_mask_ratio": 0.3007204711437225, + "epoch": 8.45213849287169, + "flow_matching_loss": 0.008977949665859342, + "grad_norm": 0.061765819787979126, + "kept_token_count": 572.22109375, + "learning_rate": 8.234449476750172e-08, + "loss": 0.0092, + "mgd_cos_sim": 0.9938889980316162, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006111018452793359, + "mgd_loss_weight": 0.03181772716343403, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.8317810058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2439819335938, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29050, + "valid_token_count": 818.29453125 + }, + { + "actual_token_mask_ratio": 0.2997033208608627, + "epoch": 8.455048006982834, + "flow_matching_loss": 0.010300722857937216, + "grad_norm": 0.06969185918569565, + "kept_token_count": 572.99609375, + "learning_rate": 8.062339905941351e-08, + "loss": 0.0105, + "mgd_cos_sim": 0.9940568149089813, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005943185416981578, + "mgd_loss_weight": 0.03148438259959221, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.38466491699216, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.76539916992186, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29060, + "valid_token_count": 818.21953125 + }, + { + "actual_token_mask_ratio": 0.2994959056377411, + "epoch": 8.457957521093977, + "flow_matching_loss": 0.010405092872679234, + "grad_norm": 0.06719283759593964, + "kept_token_count": 573.23046875, + "learning_rate": 7.89204318007003e-08, + "loss": 0.0106, + "mgd_cos_sim": 0.9936080992221832, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006391885643824935, + "mgd_loss_weight": 0.031151038594543934, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.47804260253906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0954071044922, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29070, + "valid_token_count": 818.31015625 + }, + { + "actual_token_mask_ratio": 0.29908024370670316, + "epoch": 8.460867035205121, + "flow_matching_loss": 0.009071670332923531, + "grad_norm": 0.051812794059515, + "kept_token_count": 573.603125, + "learning_rate": 7.723559506062849e-08, + "loss": 0.0093, + "mgd_cos_sim": 0.993759274482727, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006240716017782688, + "mgd_loss_weight": 0.0308176938444376, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.5948089599609, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3236145019531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29080, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.30001119077205657, + "epoch": 8.463776549316265, + "flow_matching_loss": 0.010189805179834366, + "grad_norm": 0.04363810271024704, + "kept_token_count": 572.7953125, + "learning_rate": 7.556889088643381e-08, + "loss": 0.0104, + "mgd_cos_sim": 0.9939183592796326, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006081644166260958, + "mgd_loss_weight": 0.030484349466860294, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.1619415283203, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.54710693359374, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29090, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.30073559284210205, + "epoch": 8.466686063427408, + "flow_matching_loss": 0.010973784606903791, + "grad_norm": 0.07285910099744797, + "kept_token_count": 572.18125, + "learning_rate": 7.392032130331793e-08, + "loss": 0.0112, + "mgd_cos_sim": 0.9938114404678344, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006188570708036423, + "mgd_loss_weight": 0.03015100508928299, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.21404724121095, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1035980224609, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29100, + "valid_token_count": 818.26484375 + }, + { + "actual_token_mask_ratio": 0.30042138397693635, + "epoch": 8.46959557753855, + "flow_matching_loss": 0.012134781526401638, + "grad_norm": 0.07286175340414047, + "kept_token_count": 572.496875, + "learning_rate": 7.228988831445015e-08, + "loss": 0.0123, + "mgd_cos_sim": 0.9937798500061035, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006220148550346494, + "mgd_loss_weight": 0.02981766052544117, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.7003692626953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.30907287597654, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29110, + "valid_token_count": 818.34375 + }, + { + "actual_token_mask_ratio": 0.3001216292381287, + "epoch": 8.472505091649694, + "flow_matching_loss": 0.010607494181022048, + "grad_norm": 0.07163465768098831, + "kept_token_count": 572.60234375, + "learning_rate": 7.06775939009574e-08, + "loss": 0.0108, + "mgd_cos_sim": 0.9941263377666474, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005873676994815469, + "mgd_loss_weight": 0.029484316147863864, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.5195770263672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.40004577636716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29120, + "valid_token_count": 818.14453125 + }, + { + "actual_token_mask_ratio": 0.2995211869478226, + "epoch": 8.475414605760838, + "flow_matching_loss": 0.010044399136677384, + "grad_norm": 0.05173156037926674, + "kept_token_count": 573.05078125, + "learning_rate": 6.908344002193256e-08, + "loss": 0.0102, + "mgd_cos_sim": 0.9938610792160034, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006138916499912739, + "mgd_loss_weight": 0.029150971584022044, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.3581939697266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.12111511230466, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29130, + "valid_token_count": 818.0828125 + }, + { + "actual_token_mask_ratio": 0.300078871846199, + "epoch": 8.478324119871981, + "flow_matching_loss": 0.008911935286596418, + "grad_norm": 0.05528731644153595, + "kept_token_count": 572.78828125, + "learning_rate": 6.750742861442116e-08, + "loss": 0.0091, + "mgd_cos_sim": 0.99400315284729, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00599683984182775, + "mgd_loss_weight": 0.028817627392709256, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.7450775146484, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.58558654785156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29140, + "valid_token_count": 818.3640625 + }, + { + "actual_token_mask_ratio": 0.3000641971826553, + "epoch": 8.481233633983125, + "flow_matching_loss": 0.009347655205056071, + "grad_norm": 0.06567620486021042, + "kept_token_count": 572.915625, + "learning_rate": 6.594956159342636e-08, + "loss": 0.0095, + "mgd_cos_sim": 0.9938089549541473, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061910419259220365, + "mgd_loss_weight": 0.028484282828867436, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.95826110839846, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.9715637207031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29150, + "valid_token_count": 818.528125 + }, + { + "actual_token_mask_ratio": 0.2998764872550964, + "epoch": 8.484143148094269, + "flow_matching_loss": 0.011111629195511341, + "grad_norm": 0.05652421712875366, + "kept_token_count": 572.93515625, + "learning_rate": 6.440984085190394e-08, + "loss": 0.0113, + "mgd_cos_sim": 0.993703418970108, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062965695746243, + "mgd_loss_weight": 0.02815093845129013, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.56764831542966, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.05662841796874, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29160, + "valid_token_count": 818.3296875 + }, + { + "actual_token_mask_ratio": 0.29962935745716096, + "epoch": 8.487052662205413, + "flow_matching_loss": 0.009880383033305407, + "grad_norm": 0.0562664195895195, + "kept_token_count": 573.153125, + "learning_rate": 6.288826826075733e-08, + "loss": 0.0101, + "mgd_cos_sim": 0.993841576576233, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061584276147186754, + "mgd_loss_weight": 0.02781759388744831, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.2874298095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.9120574951172, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29170, + "valid_token_count": 818.36328125 + }, + { + "actual_token_mask_ratio": 0.2999816596508026, + "epoch": 8.489962176316554, + "flow_matching_loss": 0.010191093804314733, + "grad_norm": 0.0663834735751152, + "kept_token_count": 572.76484375, + "learning_rate": 6.138484566884261e-08, + "loss": 0.0104, + "mgd_cos_sim": 0.9937761664390564, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006223828112706542, + "mgd_loss_weight": 0.027484249509871005, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.0302703857422, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.6746490478516, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29180, + "valid_token_count": 818.21796875 + }, + { + "actual_token_mask_ratio": 0.29992853105068207, + "epoch": 8.492871690427698, + "flow_matching_loss": 0.009546761028468609, + "grad_norm": 0.06368713080883026, + "kept_token_count": 572.9, + "learning_rate": 5.989957490296183e-08, + "loss": 0.0097, + "mgd_cos_sim": 0.9940151870250702, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005984816933050752, + "mgd_loss_weight": 0.027150904946029186, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.2677917480469, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.358544921875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29190, + "valid_token_count": 818.3390625 + }, + { + "actual_token_mask_ratio": 0.30005806386470796, + "epoch": 8.495781204538842, + "flow_matching_loss": 0.009290799917653202, + "grad_norm": 0.05058025196194649, + "kept_token_count": 572.72890625, + "learning_rate": 5.843245776785466e-08, + "loss": 0.0094, + "mgd_cos_sim": 0.9940662503242492, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005933750746771693, + "mgd_loss_weight": 0.02681756056845188, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.709228515625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.8587646484375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29200, + "valid_token_count": 818.25234375 + }, + { + "actual_token_mask_ratio": 0.3006919533014297, + "epoch": 8.498690718649986, + "flow_matching_loss": 0.00912487474270165, + "grad_norm": 0.06894899904727936, + "kept_token_count": 572.23828125, + "learning_rate": 5.6983496046210115e-08, + "loss": 0.0093, + "mgd_cos_sim": 0.9939918994903565, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006008108099922538, + "mgd_loss_weight": 0.026484216190874578, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.01316528320314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.23700561523435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29210, + "valid_token_count": 818.2953125 + }, + { + "actual_token_mask_ratio": 0.29938497543334963, + "epoch": 8.50160023276113, + "flow_matching_loss": 0.0096902787219733, + "grad_norm": 0.061387743800878525, + "kept_token_count": 573.30078125, + "learning_rate": 5.5552691498653165e-08, + "loss": 0.0098, + "mgd_cos_sim": 0.994102668762207, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005897319223731756, + "mgd_loss_weight": 0.026150871627032758, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.50904846191406, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.83445739746094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29220, + "valid_token_count": 818.28359375 + }, + { + "actual_token_mask_ratio": 0.30043596029281616, + "epoch": 8.504509746872273, + "flow_matching_loss": 0.009452642407268286, + "grad_norm": 0.054158493876457214, + "kept_token_count": 572.41484375, + "learning_rate": 5.4140045863743106e-08, + "loss": 0.0096, + "mgd_cos_sim": 0.9939269125461578, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006073078839108348, + "mgd_loss_weight": 0.025817527249455453, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.8138732910156, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2709625244141, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29230, + "valid_token_count": 818.25078125 + }, + { + "actual_token_mask_ratio": 0.30011827647686007, + "epoch": 8.507419260983415, + "flow_matching_loss": 0.010273050284013152, + "grad_norm": 0.05846702307462692, + "kept_token_count": 572.78203125, + "learning_rate": 5.274556085798188e-08, + "loss": 0.0104, + "mgd_cos_sim": 0.9940964221954346, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005903574312105775, + "mgd_loss_weight": 0.025484182685613633, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.2412841796875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.149755859375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29240, + "valid_token_count": 818.396875 + }, + { + "actual_token_mask_ratio": 0.2992758840322495, + "epoch": 8.510328775094559, + "flow_matching_loss": 0.009420090215280652, + "grad_norm": 0.054925587028265, + "kept_token_count": 573.3546875, + "learning_rate": 5.136923817579742e-08, + "loss": 0.0096, + "mgd_cos_sim": 0.993734472990036, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00626552333123982, + "mgd_loss_weight": 0.02515083849430084, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.3066131591797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2236022949219, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29250, + "valid_token_count": 818.23046875 + }, + { + "actual_token_mask_ratio": 0.298984757065773, + "epoch": 8.513238289205702, + "flow_matching_loss": 0.00978866433724761, + "grad_norm": 0.06256508082151413, + "kept_token_count": 573.62109375, + "learning_rate": 5.001107948955197e-08, + "loss": 0.0099, + "mgd_cos_sim": 0.9940072774887085, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005992721952497959, + "mgd_loss_weight": 0.024817493930459022, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.05601806640624, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.00356140136716, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29260, + "valid_token_count": 818.275 + }, + { + "actual_token_mask_ratio": 0.29946996569633483, + "epoch": 8.516147803316846, + "flow_matching_loss": 0.009453663136810064, + "grad_norm": 0.0763644203543663, + "kept_token_count": 573.153125, + "learning_rate": 4.867108644953711e-08, + "loss": 0.0096, + "mgd_cos_sim": 0.9940383553504943, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005961649818345904, + "mgd_loss_weight": 0.024484149552881716, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.9712677001953, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.63616333007815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29270, + "valid_token_count": 818.16796875 + }, + { + "actual_token_mask_ratio": 0.30020089745521544, + "epoch": 8.51905731742799, + "flow_matching_loss": 0.011482935445383192, + "grad_norm": 0.071759432554245, + "kept_token_count": 572.675, + "learning_rate": 4.734926068397038e-08, + "loss": 0.0116, + "mgd_cos_sim": 0.9939706563949585, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006029360508546233, + "mgd_loss_weight": 0.024150804989039897, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.3866241455078, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.29454650878904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29280, + "valid_token_count": 818.34296875 + }, + { + "actual_token_mask_ratio": 0.2992423504590988, + "epoch": 8.521966831539133, + "flow_matching_loss": 0.010769474878907203, + "grad_norm": 0.05619751289486885, + "kept_token_count": 573.36484375, + "learning_rate": 4.604560379899536e-08, + "loss": 0.0109, + "mgd_cos_sim": 0.9941674053668976, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00583259966224432, + "mgd_loss_weight": 0.023817460611462594, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.9806732177734, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.35912170410154, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29290, + "valid_token_count": 818.20625 + }, + { + "actual_token_mask_ratio": 0.3001439183950424, + "epoch": 8.524876345650277, + "flow_matching_loss": 0.00998760350048542, + "grad_norm": 0.07246925681829453, + "kept_token_count": 572.63359375, + "learning_rate": 4.476011737867658e-08, + "loss": 0.0101, + "mgd_cos_sim": 0.9938088953495026, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006191102741286159, + "mgd_loss_weight": 0.02348411623388529, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.1904663085937, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.11839294433594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29300, + "valid_token_count": 818.21953125 + }, + { + "actual_token_mask_ratio": 0.2999327301979065, + "epoch": 8.52778585976142, + "flow_matching_loss": 0.010237274738028646, + "grad_norm": 0.05863725766539574, + "kept_token_count": 572.746875, + "learning_rate": 4.349280298500291e-08, + "loss": 0.0104, + "mgd_cos_sim": 0.9938460707664489, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061539123766124245, + "mgd_loss_weight": 0.02315077167004347, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.9780731201172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.5910247802734, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29310, + "valid_token_count": 818.13046875 + }, + { + "actual_token_mask_ratio": 0.30048583447933197, + "epoch": 8.530695373872563, + "flow_matching_loss": 0.009698179503902792, + "grad_norm": 0.0993259847164154, + "kept_token_count": 572.528125, + "learning_rate": 4.224366215787923e-08, + "loss": 0.0098, + "mgd_cos_sim": 0.9939655542373658, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006034435098990798, + "mgd_loss_weight": 0.022817427292466164, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.7952362060547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 382.5012237548828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29320, + "valid_token_count": 818.46640625 + }, + { + "actual_token_mask_ratio": 0.3000591665506363, + "epoch": 8.533604887983707, + "flow_matching_loss": 0.008540182420983911, + "grad_norm": 0.07534698396921158, + "kept_token_count": 572.86640625, + "learning_rate": 4.101269641513139e-08, + "loss": 0.0087, + "mgd_cos_sim": 0.9937236428260803, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006276353495195508, + "mgd_loss_weight": 0.022484082728624344, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.0378387451172, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.8077697753906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29330, + "valid_token_count": 818.453125 + }, + { + "actual_token_mask_ratio": 0.3001907229423523, + "epoch": 8.53651440209485, + "flow_matching_loss": 0.011038877721875905, + "grad_norm": 0.06721630692481995, + "kept_token_count": 572.62265625, + "learning_rate": 3.979990725249627e-08, + "loss": 0.0112, + "mgd_cos_sim": 0.9940779507160187, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059220625553280115, + "mgd_loss_weight": 0.02215073835104704, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.6835205078125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.29925537109375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29340, + "valid_token_count": 818.2546875 + }, + { + "actual_token_mask_ratio": 0.29977051019668577, + "epoch": 8.539423916205994, + "flow_matching_loss": 0.01060613919980824, + "grad_norm": 0.076894111931324, + "kept_token_count": 572.934375, + "learning_rate": 3.860529614363006e-08, + "loss": 0.0107, + "mgd_cos_sim": 0.9943274319171905, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005672561889514327, + "mgd_loss_weight": 0.02181739378720522, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.0686798095703, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.88344116210936, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29350, + "valid_token_count": 818.21015625 + }, + { + "actual_token_mask_ratio": 0.300752666592598, + "epoch": 8.542333430317138, + "flow_matching_loss": 0.009951673680916429, + "grad_norm": 0.05543201044201851, + "kept_token_count": 572.2328125, + "learning_rate": 3.742886454009664e-08, + "loss": 0.0101, + "mgd_cos_sim": 0.9940261781215668, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059738061390817165, + "mgd_loss_weight": 0.021484049409627913, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.05858154296874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.715625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29360, + "valid_token_count": 818.35546875 + }, + { + "actual_token_mask_ratio": 0.2997681885957718, + "epoch": 8.545242944428281, + "flow_matching_loss": 0.008768083062022924, + "grad_norm": 0.0642470270395279, + "kept_token_count": 572.965625, + "learning_rate": 3.627061387137087e-08, + "loss": 0.0089, + "mgd_cos_sim": 0.9940145611763, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059854339342564344, + "mgd_loss_weight": 0.02115070503205061, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.6961334228516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.2595550537109, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29370, + "valid_token_count": 818.25703125 + }, + { + "actual_token_mask_ratio": 0.300033500790596, + "epoch": 8.548152458539423, + "flow_matching_loss": 0.010942619666457177, + "grad_norm": 0.07461626827716827, + "kept_token_count": 572.6046875, + "learning_rate": 3.5130545544835305e-08, + "loss": 0.0111, + "mgd_cos_sim": 0.9939130127429963, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006086983112618327, + "mgd_loss_weight": 0.020817360654473305, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.1279815673828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.28986206054685, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29380, + "valid_token_count": 818.046875 + }, + { + "actual_token_mask_ratio": 0.30002360641956327, + "epoch": 8.551061972650567, + "flow_matching_loss": 0.01062203561887145, + "grad_norm": 0.07981348037719727, + "kept_token_count": 572.84765625, + "learning_rate": 3.400866094578348e-08, + "loss": 0.0107, + "mgd_cos_sim": 0.9938228845596313, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006177127594128251, + "mgd_loss_weight": 0.020484016090631486, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.592431640625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.84974365234376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29390, + "valid_token_count": 818.38203125 + }, + { + "actual_token_mask_ratio": 0.29966663420200346, + "epoch": 8.55397148676171, + "flow_matching_loss": 0.010326190665364266, + "grad_norm": 0.059381868690252304, + "kept_token_count": 572.93828125, + "learning_rate": 3.290496143740829e-08, + "loss": 0.0105, + "mgd_cos_sim": 0.9936411261558533, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0063588694669306275, + "mgd_loss_weight": 0.02015067171305418, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.9198760986328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.31329040527345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29400, + "valid_token_count": 818.09453125 + }, + { + "actual_token_mask_ratio": 0.29988173246383665, + "epoch": 8.556881000872854, + "flow_matching_loss": 0.011840332020074128, + "grad_norm": 0.057493165135383606, + "kept_token_count": 572.8828125, + "learning_rate": 3.18194483608103e-08, + "loss": 0.012, + "mgd_cos_sim": 0.9941559731960297, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0058440117631107565, + "mgd_loss_weight": 0.019817327335476875, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 419.4602081298828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.5108612060547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29410, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.30113633871078493, + "epoch": 8.559790514983998, + "flow_matching_loss": 0.009314024960622192, + "grad_norm": 0.04699324071407318, + "kept_token_count": 571.7734375, + "learning_rate": 3.075212303498942e-08, + "loss": 0.0094, + "mgd_cos_sim": 0.9940196394920349, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005980346258729697, + "mgd_loss_weight": 0.019483982771635055, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.58080139160154, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.14331970214846, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29420, + "valid_token_count": 818.14765625 + }, + { + "actual_token_mask_ratio": 0.2999047577381134, + "epoch": 8.562700029095142, + "flow_matching_loss": 0.009813840594142675, + "grad_norm": 0.06893975287675858, + "kept_token_count": 572.8140625, + "learning_rate": 2.9702986756846574e-08, + "loss": 0.0099, + "mgd_cos_sim": 0.993812495470047, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061874994542449714, + "mgd_loss_weight": 0.01915063839405775, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.03770751953124, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0955841064453, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29430, + "valid_token_count": 818.19140625 + }, + { + "actual_token_mask_ratio": 0.30012536942958834, + "epoch": 8.565609543206284, + "flow_matching_loss": 0.00816369247622788, + "grad_norm": 0.05835371091961861, + "kept_token_count": 572.79140625, + "learning_rate": 2.867204080118202e-08, + "loss": 0.0083, + "mgd_cos_sim": 0.9941504597663879, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005849542887881398, + "mgd_loss_weight": 0.01881729383021593, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.79653015136716, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0150848388672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29440, + "valid_token_count": 818.425 + }, + { + "actual_token_mask_ratio": 0.3001519709825516, + "epoch": 8.568519057317427, + "flow_matching_loss": 0.01076224548742175, + "grad_norm": 0.09207199513912201, + "kept_token_count": 572.909375, + "learning_rate": 2.7659286420693708e-08, + "loss": 0.0109, + "mgd_cos_sim": 0.9940666139125824, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0059333907440304754, + "mgd_loss_weight": 0.018483949452638627, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.92503051757814, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.46856689453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29450, + "valid_token_count": 818.62109375 + }, + { + "actual_token_mask_ratio": 0.3005695313215256, + "epoch": 8.571428571428571, + "flow_matching_loss": 0.00986104616895318, + "grad_norm": 0.07646405696868896, + "kept_token_count": 572.36015625, + "learning_rate": 2.6664724845968934e-08, + "loss": 0.01, + "mgd_cos_sim": 0.9937124073505401, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062876065261662, + "mgd_loss_weight": 0.018150605075061322, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.16449584960935, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7933044433594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29460, + "valid_token_count": 818.32734375 + }, + { + "actual_token_mask_ratio": 0.30030601322650907, + "epoch": 8.574338085539715, + "flow_matching_loss": 0.010404308373108507, + "grad_norm": 0.06861188262701035, + "kept_token_count": 572.5609375, + "learning_rate": 2.5688357285499343e-08, + "loss": 0.0105, + "mgd_cos_sim": 0.9939728796482086, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006027119373902679, + "mgd_loss_weight": 0.017817260511219502, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.21465759277345, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.7417297363281, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29470, + "valid_token_count": 818.303125 + }, + { + "actual_token_mask_ratio": 0.29994709193706515, + "epoch": 8.577247599650859, + "flow_matching_loss": 0.011222650203853846, + "grad_norm": 0.062143340706825256, + "kept_token_count": 572.959375, + "learning_rate": 2.473018492565926e-08, + "loss": 0.0113, + "mgd_cos_sim": 0.9938464939594269, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006153511861339211, + "mgd_loss_weight": 0.017483916133642197, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.49639892578125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.59470520019534, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29480, + "valid_token_count": 818.44296875 + }, + { + "actual_token_mask_ratio": 0.2998436450958252, + "epoch": 8.580157113762002, + "flow_matching_loss": 0.00852427128702402, + "grad_norm": 0.07296676188707352, + "kept_token_count": 572.86640625, + "learning_rate": 2.3790208930720702e-08, + "loss": 0.0086, + "mgd_cos_sim": 0.9940153121948242, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005984682403504848, + "mgd_loss_weight": 0.01715057175606489, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.0996490478516, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.821337890625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29490, + "valid_token_count": 818.1953125 + }, + { + "actual_token_mask_ratio": 0.3005076825618744, + "epoch": 8.583066627873146, + "flow_matching_loss": 0.009683550568297505, + "grad_norm": 0.08916730433702469, + "kept_token_count": 572.4109375, + "learning_rate": 2.2868430442843368e-08, + "loss": 0.0098, + "mgd_cos_sim": 0.9940366983413697, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005963303800672293, + "mgd_loss_weight": 0.01681722719222307, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.10675354003905, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.14964294433594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29500, + "valid_token_count": 818.32421875 + }, + { + "actual_token_mask_ratio": 0.29980850517749785, + "epoch": 8.585976141984288, + "flow_matching_loss": 0.008775502583011985, + "grad_norm": 0.08313173055648804, + "kept_token_count": 573.090625, + "learning_rate": 2.196485058207298e-08, + "loss": 0.0089, + "mgd_cos_sim": 0.9941264450550079, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005873555224388838, + "mgd_loss_weight": 0.016483882814645766, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.7240844726563, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1893005371094, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29510, + "valid_token_count": 818.47890625 + }, + { + "actual_token_mask_ratio": 0.3001784682273865, + "epoch": 8.588885656095432, + "flow_matching_loss": 0.009516884759068488, + "grad_norm": 0.0661381259560585, + "kept_token_count": 572.50390625, + "learning_rate": 2.1079470446346282e-08, + "loss": 0.0096, + "mgd_cos_sim": 0.994096177816391, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005903811939060688, + "mgd_loss_weight": 0.016150538437068464, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.6287445068359, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.646875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29520, + "valid_token_count": 818.0703125 + }, + { + "actual_token_mask_ratio": 0.29987181127071383, + "epoch": 8.591795170206575, + "flow_matching_loss": 0.010728887096047402, + "grad_norm": 0.05125971511006355, + "kept_token_count": 573.01640625, + "learning_rate": 2.0212291111482707e-08, + "loss": 0.0108, + "mgd_cos_sim": 0.9938435077667236, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006156480032950639, + "mgd_loss_weight": 0.015817193873226644, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.88251342773435, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.1742248535156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29530, + "valid_token_count": 818.44765625 + }, + { + "actual_token_mask_ratio": 0.3004293143749237, + "epoch": 8.594704684317719, + "flow_matching_loss": 0.009772537928074598, + "grad_norm": 0.0615326352417469, + "kept_token_count": 572.5125, + "learning_rate": 1.9363313631187707e-08, + "loss": 0.0099, + "mgd_cos_sim": 0.9941806077957154, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005819410504773259, + "mgd_loss_weight": 0.015483849495649338, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.10859375, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.2021423339844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29540, + "valid_token_count": 818.38046875 + }, + { + "actual_token_mask_ratio": 0.2994013577699661, + "epoch": 8.597614198428863, + "flow_matching_loss": 0.011023408081382514, + "grad_norm": 0.06589449942111969, + "kept_token_count": 572.90859375, + "learning_rate": 1.853253903704777e-08, + "loss": 0.0111, + "mgd_cos_sim": 0.9940627634525299, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005937229283154011, + "mgd_loss_weight": 0.015150505024939776, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.2570526123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.8222595214844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29550, + "valid_token_count": 817.74609375 + }, + { + "actual_token_mask_ratio": 0.29991391599178313, + "epoch": 8.600523712540006, + "flow_matching_loss": 0.008942558150738477, + "grad_norm": 0.07603824883699417, + "kept_token_count": 573.003125, + "learning_rate": 1.7719968338533733e-08, + "loss": 0.009, + "mgd_cos_sim": 0.9937464952468872, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006253511644899845, + "mgd_loss_weight": 0.014817160554230213, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.64854431152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8979125976563, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29560, + "valid_token_count": 818.47734375 + }, + { + "actual_token_mask_ratio": 0.30016623735427855, + "epoch": 8.603433226651148, + "flow_matching_loss": 0.009303623531013726, + "grad_norm": 0.059823907911777496, + "kept_token_count": 572.55390625, + "learning_rate": 1.6925602522992467e-08, + "loss": 0.0094, + "mgd_cos_sim": 0.9940487027168274, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005951295793056488, + "mgd_loss_weight": 0.014483816176652908, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.90406494140626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9257781982422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29570, + "valid_token_count": 818.1265625 + }, + { + "actual_token_mask_ratio": 0.30022110641002653, + "epoch": 8.606342740762292, + "flow_matching_loss": 0.009336877102032303, + "grad_norm": 0.0723387748003006, + "kept_token_count": 572.46015625, + "learning_rate": 1.6149442555658533e-08, + "loss": 0.0094, + "mgd_cos_sim": 0.9941093862056732, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005890617985278368, + "mgd_loss_weight": 0.014150471705943347, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.5480133056641, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.62530212402345, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29580, + "valid_token_count": 818.059375 + }, + { + "actual_token_mask_ratio": 0.30007552802562715, + "epoch": 8.609252254873436, + "flow_matching_loss": 0.009962826315313578, + "grad_norm": 0.06043803691864014, + "kept_token_count": 572.71953125, + "learning_rate": 1.5391489379635858e-08, + "loss": 0.01, + "mgd_cos_sim": 0.9939993679523468, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0060006142128258945, + "mgd_loss_weight": 0.013817127235233784, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.6175170898438, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7931427001953, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29590, + "valid_token_count": 818.259375 + }, + { + "actual_token_mask_ratio": 0.30004564225673674, + "epoch": 8.61216176898458, + "flow_matching_loss": 0.009771350352093578, + "grad_norm": 0.053948044776916504, + "kept_token_count": 572.609375, + "learning_rate": 1.4651743915909399e-08, + "loss": 0.0099, + "mgd_cos_sim": 0.9940876662731171, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005912312725558877, + "mgd_loss_weight": 0.013483782764524221, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.1965087890625, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.86143798828124, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29600, + "valid_token_count": 818.06875 + }, + { + "actual_token_mask_ratio": 0.30042438507080077, + "epoch": 8.615071283095723, + "flow_matching_loss": 0.011497992416843772, + "grad_norm": 0.07019548118114471, + "kept_token_count": 572.54765625, + "learning_rate": 1.3930207063340139e-08, + "loss": 0.0116, + "mgd_cos_sim": 0.9939339518547058, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00606606206856668, + "mgd_loss_weight": 0.013150438386946916, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.10997009277344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.7495910644531, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29610, + "valid_token_count": 818.42578125 + }, + { + "actual_token_mask_ratio": 0.2998717695474625, + "epoch": 8.617980797206867, + "flow_matching_loss": 0.009387532388791442, + "grad_norm": 0.04745718464255333, + "kept_token_count": 572.80859375, + "learning_rate": 1.322687969866343e-08, + "loss": 0.0095, + "mgd_cos_sim": 0.9936802864074707, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006319711171090603, + "mgd_loss_weight": 0.012817093823105096, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.65569763183595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.87509765625, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29620, + "valid_token_count": 818.1484375 + }, + { + "actual_token_mask_ratio": 0.3000748872756958, + "epoch": 8.62089031131801, + "flow_matching_loss": 0.008972941711544991, + "grad_norm": 0.10351473093032837, + "kept_token_count": 572.81953125, + "learning_rate": 1.2541762676488989e-08, + "loss": 0.009, + "mgd_cos_sim": 0.9938342571258545, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0061657516285777095, + "mgd_loss_weight": 0.012483749445527792, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 411.4637817382812, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.218359375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29630, + "valid_token_count": 818.3984375 + }, + { + "actual_token_mask_ratio": 0.29969242215156555, + "epoch": 8.623799825429153, + "flow_matching_loss": 0.008981512067839504, + "grad_norm": 0.06945096701383591, + "kept_token_count": 572.90390625, + "learning_rate": 1.1874856829297564e-08, + "loss": 0.0091, + "mgd_cos_sim": 0.9939761281013488, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00602386835962534, + "mgd_loss_weight": 0.012150405067950487, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.99412231445314, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.05604858398436, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29640, + "valid_token_count": 818.0765625 + }, + { + "actual_token_mask_ratio": 0.30064927637577055, + "epoch": 8.626709339540296, + "flow_matching_loss": 0.010793477529659867, + "grad_norm": 0.057420503348112106, + "kept_token_count": 572.2984375, + "learning_rate": 1.1226162967442611e-08, + "loss": 0.0109, + "mgd_cos_sim": 0.9939966797828674, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006003333255648613, + "mgd_loss_weight": 0.011817060597240924, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.6114776611328, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.68189697265626, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29650, + "valid_token_count": 818.3234375 + }, + { + "actual_token_mask_ratio": 0.3007348209619522, + "epoch": 8.62961885365144, + "flow_matching_loss": 0.00867057521827519, + "grad_norm": 0.08854135870933533, + "kept_token_count": 572.22578125, + "learning_rate": 1.059568187914861e-08, + "loss": 0.0087, + "mgd_cos_sim": 0.9937809884548188, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006219013454392552, + "mgd_loss_weight": 0.011483716126531363, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 420.4228973388672, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 377.88397827148435, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29660, + "valid_token_count": 818.325 + }, + { + "actual_token_mask_ratio": 0.3005328387022018, + "epoch": 8.632528367762584, + "flow_matching_loss": 0.009176822006702423, + "grad_norm": 0.07135645300149918, + "kept_token_count": 572.4203125, + "learning_rate": 9.983414330511087e-09, + "loss": 0.0092, + "mgd_cos_sim": 0.9937968790531159, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0062031146138906475, + "mgd_loss_weight": 0.0111503716558218, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.12420654296875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.80381774902344, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29670, + "valid_token_count": 818.36328125 + }, + { + "actual_token_mask_ratio": 0.3004635900259018, + "epoch": 8.635437881873727, + "flow_matching_loss": 0.010484907915815711, + "grad_norm": 0.08676669746637344, + "kept_token_count": 572.3421875, + "learning_rate": 9.389361065489932e-09, + "loss": 0.0105, + "mgd_cos_sim": 0.9940049231052399, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005995093518868089, + "mgd_loss_weight": 0.010817027185112238, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.5034606933594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.52906799316406, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29680, + "valid_token_count": 818.17265625 + }, + { + "actual_token_mask_ratio": 0.30044832527637483, + "epoch": 8.638347395984871, + "flow_matching_loss": 0.010052533587440848, + "grad_norm": 0.07933400571346283, + "kept_token_count": 572.49140625, + "learning_rate": 8.81352280591774e-09, + "loss": 0.0101, + "mgd_cos_sim": 0.9939328074455261, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006067188782617449, + "mgd_loss_weight": 0.010483682807534932, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.04795532226564, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.33880310058595, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29690, + "valid_token_count": 818.37109375 + }, + { + "actual_token_mask_ratio": 0.3002248078584671, + "epoch": 8.641256910096015, + "flow_matching_loss": 0.008251765044406056, + "grad_norm": 0.04930799454450607, + "kept_token_count": 572.5203125, + "learning_rate": 8.255900251491478e-09, + "loss": 0.0083, + "mgd_cos_sim": 0.9938937485218048, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006106246495619416, + "mgd_loss_weight": 0.010150338336825371, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.4724060058594, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.2840576171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29700, + "valid_token_count": 818.153125 + }, + { + "actual_token_mask_ratio": 0.2994094491004944, + "epoch": 8.644166424207157, + "flow_matching_loss": 0.011916135996580124, + "grad_norm": 0.06702525913715363, + "kept_token_count": 573.26484375, + "learning_rate": 7.716494079775816e-09, + "loss": 0.012, + "mgd_cos_sim": 0.9936870276927948, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006312970258295536, + "mgd_loss_weight": 0.009816993866115809, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.5867492675781, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.3087951660156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29710, + "valid_token_count": 818.25703125 + }, + { + "actual_token_mask_ratio": 0.3002495586872101, + "epoch": 8.6470759383183, + "flow_matching_loss": 0.00969092226587236, + "grad_norm": 0.06731679290533066, + "kept_token_count": 572.70234375, + "learning_rate": 7.195304946199799e-09, + "loss": 0.0097, + "mgd_cos_sim": 0.9939861595630646, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006013840110972524, + "mgd_loss_weight": 0.009483649488538503, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.4167205810547, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.09815673828126, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29720, + "valid_token_count": 818.43359375 + }, + { + "actual_token_mask_ratio": 0.2998644530773163, + "epoch": 8.649985452429444, + "flow_matching_loss": 0.009718065755441785, + "grad_norm": 0.06073537841439247, + "kept_token_count": 572.9265625, + "learning_rate": 6.692333484056845e-09, + "loss": 0.0098, + "mgd_cos_sim": 0.9941666305065155, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00583337564021349, + "mgd_loss_weight": 0.00915030501782894, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.9540740966797, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.7578918457031, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29730, + "valid_token_count": 818.31484375 + }, + { + "actual_token_mask_ratio": 0.30035197734832764, + "epoch": 8.652894966540588, + "flow_matching_loss": 0.01060265782289207, + "grad_norm": 0.06818682700395584, + "kept_token_count": 572.49609375, + "learning_rate": 6.207580304506411e-09, + "loss": 0.0107, + "mgd_cos_sim": 0.9935855388641357, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006414449820294976, + "mgd_loss_weight": 0.00881696054711938, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.7758026123047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.79847412109376, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29740, + "valid_token_count": 818.2640625 + }, + { + "actual_token_mask_ratio": 0.2998833000659943, + "epoch": 8.655804480651732, + "flow_matching_loss": 0.009940907498821617, + "grad_norm": 0.10303390771150589, + "kept_token_count": 572.884375, + "learning_rate": 5.741045996567329e-09, + "loss": 0.01, + "mgd_cos_sim": 0.9939819931983948, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006017995532602072, + "mgd_loss_weight": 0.008483616169542074, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.79127197265626, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4998321533203, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29750, + "valid_token_count": 818.26875 + }, + { + "actual_token_mask_ratio": 0.2999373525381088, + "epoch": 8.658713994762875, + "flow_matching_loss": 0.009937589010223746, + "grad_norm": 0.10941930115222931, + "kept_token_count": 572.98046875, + "learning_rate": 5.2927311271244725e-09, + "loss": 0.01, + "mgd_cos_sim": 0.9938926935195923, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006107296701520682, + "mgd_loss_weight": 0.008150271698832512, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.7842559814453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.38685607910156, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29760, + "valid_token_count": 818.47265625 + }, + { + "actual_token_mask_ratio": 0.30001780688762664, + "epoch": 8.661623508874017, + "flow_matching_loss": 0.01073409365490079, + "grad_norm": 0.060541342943906784, + "kept_token_count": 572.89921875, + "learning_rate": 4.8626362409204265e-09, + "loss": 0.0108, + "mgd_cos_sim": 0.9942324757575989, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0057675217743963, + "mgd_loss_weight": 0.007816927228122949, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.3482421875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.23961181640624, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29770, + "valid_token_count": 818.45 + }, + { + "actual_token_mask_ratio": 0.3005746066570282, + "epoch": 8.664533022985161, + "flow_matching_loss": 0.010669754166156054, + "grad_norm": 0.06769995391368866, + "kept_token_count": 572.240625, + "learning_rate": 4.4507618605638126e-09, + "loss": 0.0107, + "mgd_cos_sim": 0.9939356565475463, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006064337817952037, + "mgd_loss_weight": 0.007483582803979516, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.8095275878906, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.0247436523438, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29780, + "valid_token_count": 818.15703125 + }, + { + "actual_token_mask_ratio": 0.3000674158334732, + "epoch": 8.667442537096305, + "flow_matching_loss": 0.010915022622793914, + "grad_norm": 0.07698554545640945, + "kept_token_count": 572.74921875, + "learning_rate": 4.057108486519301e-09, + "loss": 0.011, + "mgd_cos_sim": 0.9940970599651336, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005902943294495344, + "mgd_loss_weight": 0.0071502383798360825, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.7975769042969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.5376922607422, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29790, + "valid_token_count": 818.2875 + }, + { + "actual_token_mask_ratio": 0.2998612582683563, + "epoch": 8.670352051207448, + "flow_matching_loss": 0.010095319896936416, + "grad_norm": 0.0711047425866127, + "kept_token_count": 572.87421875, + "learning_rate": 3.6816765971126043e-09, + "loss": 0.0101, + "mgd_cos_sim": 0.9938464283943176, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006153574213385582, + "mgd_loss_weight": 0.006816893862560391, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.3825988769531, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9810485839844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29800, + "valid_token_count": 818.2265625 + }, + { + "actual_token_mask_ratio": 0.30003831088542937, + "epoch": 8.673261565318592, + "flow_matching_loss": 0.010040307557210327, + "grad_norm": 0.06569454818964005, + "kept_token_count": 572.79296875, + "learning_rate": 3.324466648528812e-09, + "loss": 0.0101, + "mgd_cos_sim": 0.9940189182758331, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005981083353981375, + "mgd_loss_weight": 0.006483549438416958, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.46502685546875, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.09018859863284, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29810, + "valid_token_count": 818.31953125 + }, + { + "actual_token_mask_ratio": 0.2999120831489563, + "epoch": 8.676171079429736, + "flow_matching_loss": 0.010656268754974007, + "grad_norm": 0.06635844707489014, + "kept_token_count": 572.865625, + "learning_rate": 2.9854790748123917e-09, + "loss": 0.0107, + "mgd_cos_sim": 0.9935350716114044, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0064649222884327175, + "mgd_loss_weight": 0.006150205014273524, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.26604919433595, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.86712036132815, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29820, + "valid_token_count": 818.2765625 + }, + { + "actual_token_mask_ratio": 0.3001148909330368, + "epoch": 8.67908059354088, + "flow_matching_loss": 0.010198926273733377, + "grad_norm": 0.05706186220049858, + "kept_token_count": 572.57109375, + "learning_rate": 2.664714287865522e-09, + "loss": 0.0102, + "mgd_cos_sim": 0.9939776241779328, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.00602238504216075, + "mgd_loss_weight": 0.005816860590130091, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.01463623046874, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.2453582763672, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29830, + "valid_token_count": 818.090625 + }, + { + "actual_token_mask_ratio": 0.29941575825214384, + "epoch": 8.681990107652021, + "flow_matching_loss": 0.010037048067897559, + "grad_norm": 0.07685030996799469, + "kept_token_count": 573.32890625, + "learning_rate": 2.362172677446428e-09, + "loss": 0.0101, + "mgd_cos_sim": 0.9938197612762452, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006180249247699976, + "mgd_loss_weight": 0.0054835160728543995, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.6736236572266, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8762939453125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29840, + "valid_token_count": 818.3578125 + }, + { + "actual_token_mask_ratio": 0.30004860162734986, + "epoch": 8.684899621763165, + "flow_matching_loss": 0.009017485193908215, + "grad_norm": 0.0677274614572525, + "kept_token_count": 572.48203125, + "learning_rate": 2.0778546111727137e-09, + "loss": 0.009, + "mgd_cos_sim": 0.9941406607627868, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005859342822805047, + "mgd_loss_weight": 0.005150171695277095, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.8342712402344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.05115966796876, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29850, + "valid_token_count": 817.88515625 + }, + { + "actual_token_mask_ratio": 0.29954981803894043, + "epoch": 8.687809135874309, + "flow_matching_loss": 0.00946440864354372, + "grad_norm": 0.07483687996864319, + "kept_token_count": 573.1578125, + "learning_rate": 1.811760434516363e-09, + "loss": 0.0095, + "mgd_cos_sim": 0.9937521815299988, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006247824709862471, + "mgd_loss_weight": 0.004816827224567532, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 410.8628295898437, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.95220336914065, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29860, + "valid_token_count": 818.26796875 + }, + { + "actual_token_mask_ratio": 0.3003517150878906, + "epoch": 8.690718649985453, + "flow_matching_loss": 0.009129501786082983, + "grad_norm": 0.05652337148785591, + "kept_token_count": 572.5140625, + "learning_rate": 1.5638904708087377e-09, + "loss": 0.0092, + "mgd_cos_sim": 0.9940657079219818, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005934282718226313, + "mgd_loss_weight": 0.004483482800424099, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.4141143798828, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.8095642089844, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29870, + "valid_token_count": 818.28515625 + }, + { + "actual_token_mask_ratio": 0.2996516853570938, + "epoch": 8.693628164096596, + "flow_matching_loss": 0.008995380718261003, + "grad_norm": 0.05619153380393982, + "kept_token_count": 573.06015625, + "learning_rate": 1.3342450212355805e-09, + "loss": 0.009, + "mgd_cos_sim": 0.9939245522022248, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006075440160930157, + "mgd_loss_weight": 0.004150138329714536, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.6496307373047, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.56798400878904, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29880, + "valid_token_count": 818.25234375 + }, + { + "actual_token_mask_ratio": 0.29974517822265623, + "epoch": 8.69653767820774, + "flow_matching_loss": 0.010682418616488576, + "grad_norm": 0.07347512990236282, + "kept_token_count": 572.925, + "learning_rate": 1.122824364835351e-09, + "loss": 0.0107, + "mgd_cos_sim": 0.9940526902675628, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005947311315685511, + "mgd_loss_weight": 0.003816793905571103, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 415.26917419433596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.48919677734375, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29890, + "valid_token_count": 818.16171875 + }, + { + "actual_token_mask_ratio": 0.30003270208835603, + "epoch": 8.699447192318882, + "flow_matching_loss": 0.009219792718067766, + "grad_norm": 0.06355679035186768, + "kept_token_count": 572.8265625, + "learning_rate": 9.296287585042196e-10, + "loss": 0.0092, + "mgd_cos_sim": 0.9942307114601135, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0057692818343639376, + "mgd_loss_weight": 0.003483449458144605, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 416.7307861328125, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 381.2108123779297, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29900, + "valid_token_count": 818.36484375 + }, + { + "actual_token_mask_ratio": 0.29980183243751524, + "epoch": 8.702356706430026, + "flow_matching_loss": 0.009649907844141126, + "grad_norm": 0.0879652202129364, + "kept_token_count": 573.1296875, + "learning_rate": 7.546584369960697e-10, + "loss": 0.0097, + "mgd_cos_sim": 0.9942245483398438, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005775444116443396, + "mgd_loss_weight": 0.003150104987435043, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.44151306152344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3121002197266, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29910, + "valid_token_count": 818.5203125 + }, + { + "actual_token_mask_ratio": 0.300407862663269, + "epoch": 8.70526622054117, + "flow_matching_loss": 0.010843981197103858, + "grad_norm": 0.08402201533317566, + "kept_token_count": 572.51015625, + "learning_rate": 5.97913612912504e-10, + "loss": 0.0109, + "mgd_cos_sim": 0.9936188638210297, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006381131149828434, + "mgd_loss_weight": 0.0028167605632916093, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 414.9057220458984, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.3047180175781, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29920, + "valid_token_count": 818.3515625 + }, + { + "actual_token_mask_ratio": 0.3003063201904297, + "epoch": 8.708175734652313, + "flow_matching_loss": 0.009109922312200069, + "grad_norm": 0.07634323090314865, + "kept_token_count": 572.60078125, + "learning_rate": 4.593944767161684e-10, + "loss": 0.0091, + "mgd_cos_sim": 0.9939829051494599, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006017091823741794, + "mgd_loss_weight": 0.0024834161158651114, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.67937622070315, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.1637908935547, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29930, + "valid_token_count": 818.3546875 + }, + { + "actual_token_mask_ratio": 0.30030012130737305, + "epoch": 8.711085248763457, + "flow_matching_loss": 0.010873437765985728, + "grad_norm": 0.07228317856788635, + "kept_token_count": 572.52734375, + "learning_rate": 3.391011967207591e-10, + "loss": 0.0109, + "mgd_cos_sim": 0.9941826581954956, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005817341199144721, + "mgd_loss_weight": 0.0021500716684386134, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 418.55506591796876, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.72657165527346, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29940, + "valid_token_count": 818.240625 + }, + { + "actual_token_mask_ratio": 0.29909551441669463, + "epoch": 8.7139947628746, + "flow_matching_loss": 0.009180960711091757, + "grad_norm": 0.07321898639202118, + "kept_token_count": 573.46953125, + "learning_rate": 2.370339190910231e-10, + "loss": 0.0092, + "mgd_cos_sim": 0.994273567199707, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005726442160084844, + "mgd_loss_weight": 0.0018167272326536476, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.69056091308596, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.4897888183594, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29950, + "valid_token_count": 818.18359375 + }, + { + "actual_token_mask_ratio": 0.30046276450157167, + "epoch": 8.716904276985744, + "flow_matching_loss": 0.009698011633008718, + "grad_norm": 0.05046404153108597, + "kept_token_count": 572.221875, + "learning_rate": 1.531927678510847e-10, + "loss": 0.0097, + "mgd_cos_sim": 0.994242376089096, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005757621582597494, + "mgd_loss_weight": 0.0014833827735856175, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 421.4071472167969, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.271923828125, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29960, + "valid_token_count": 818.003125 + }, + { + "actual_token_mask_ratio": 0.2996538460254669, + "epoch": 8.719813791096886, + "flow_matching_loss": 0.01110967262648046, + "grad_norm": 0.07410960644483566, + "kept_token_count": 573.18046875, + "learning_rate": 8.757784487611886e-11, + "loss": 0.0111, + "mgd_cos_sim": 0.9936575531959534, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006342452298849821, + "mgd_loss_weight": 0.0011500383378006517, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.41114196777346, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 380.9169494628906, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29970, + "valid_token_count": 818.43046875 + }, + { + "actual_token_mask_ratio": 0.30010211169719697, + "epoch": 8.72272330520803, + "flow_matching_loss": 0.01025700536556542, + "grad_norm": 0.0715942531824112, + "kept_token_count": 572.8515625, + "learning_rate": 4.018922989235119e-11, + "loss": 0.0103, + "mgd_cos_sim": 0.99431391954422, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.0056860969401896, + "mgd_loss_weight": 0.0008166938845533877, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 417.5992797851562, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 378.5934326171875, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29980, + "valid_token_count": 818.4828125 + }, + { + "actual_token_mask_ratio": 0.3000274032354355, + "epoch": 8.725632819319173, + "flow_matching_loss": 0.010529120406135916, + "grad_norm": 0.06420035660266876, + "kept_token_count": 572.70078125, + "learning_rate": 1.1026980482053972e-11, + "loss": 0.0105, + "mgd_cos_sim": 0.9939523935317993, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.006047593941912055, + "mgd_loss_weight": 0.00048334944585803896, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 413.49717712402344, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.01296081542966, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 29990, + "valid_token_count": 818.17578125 + }, + { + "actual_token_mask_ratio": 0.299853453040123, + "epoch": 8.728542333430317, + "flow_matching_loss": 0.009665953135117889, + "grad_norm": 0.06212766095995903, + "kept_token_count": 572.91953125, + "learning_rate": 9.113208188082922e-14, + "loss": 0.0097, + "mgd_cos_sim": 0.9940909683704376, + "mgd_fm_loss_weight": 1.0, + "mgd_loss": 0.005909041222184897, + "mgd_loss_weight": 0.0001500049969763495, + "mgd_pred_norm_post": 1.0, + "mgd_pred_norm_raw": 412.6203887939453, + "mgd_target_norm_post": 1.0, + "mgd_target_norm_raw": 379.0520050048828, + "mgd_token_mask_ratio_cfg": 0.30000001192092896, + "step": 30000, + "valid_token_count": 818.28671875 + }, + { + "epoch": 8.728542333430317, + "step": 30000, + "total_flos": 0.0, + "train_loss": 0.016984580224504074, + "train_runtime": 48570.3725, + "train_samples_per_second": 79.061, + "train_steps_per_second": 0.618 + } + ], + "logging_steps": 10, + "max_steps": 30000, + "num_input_tokens_seen": 0, + "num_train_epochs": 9, + "save_steps": 1, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": false, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 0.0, + "train_batch_size": 128, + "trial_name": null, + "trial_params": null +} diff --git a/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/resolved_config.yaml b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/resolved_config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..9753cf8751b0227a6a1d05451a55879fcea7e067 --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_aux_linear_decay__mgd_token_mask_ratio_0.3/resolved_config.yaml @@ -0,0 +1,36 @@ +experiment_name: mgd_phase3_aux_linear_decay +policy_type: groot_mgd +base_model_path: ${HOME}/groot_robocasa/robocasa_v2/MGD_Processing_line_Only/mgd_token_mask_ratio_0.3/phase2 +output_root: ${HOME}/groot_robocasa/robocasa_v2/processing_line_only/loss_strategy_ablation/mgd +sweep: + loss_strategy: + - aux_linear_decay + mgd_token_mask_ratio: + - 0.3 +dataset: + dataset_soup: my_atomic26_human +training: + num_gpus: 1 + batch_size: 128 +model: null +phases: +- name: phase3_fm_mgd_linear_decay + max_steps: 30000 + save_steps: 0 + trainable: + tune_llm: false + tune_visual: false + tune_projector: true + tune_diffusion_model: true + losses: + mgd_enabled: true + mgd_fm_loss_weight: 1.0 + mgd_loss_weight_schedule: linear_decay + mgd_loss_weight_start: 1.0 + mgd_loss_weight_end: 0.0 + mgd_token_mask_ratio: 0.3 + mgd_use_cosine_loss: true + mgd_use_mse_loss: false +resolved_sweep: + loss_strategy: aux_linear_decay + mgd_token_mask_ratio: 0.3 diff --git a/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/mgd_phase3_fm_only.yaml b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/mgd_phase3_fm_only.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b7acf66a162ed8b1539680bf81ee78b1868f8d39 --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/mgd_phase3_fm_only.yaml @@ -0,0 +1,31 @@ +experiment_name: mgd_phase3_fm_only +policy_type: groot_mgd +base_model_path: ${HOME}/groot_robocasa/robocasa_v2/MGD_Processing_line_Only/mgd_token_mask_ratio_${sweep.mgd_token_mask_ratio}/phase2 +output_root: ${HOME}/groot_robocasa/robocasa_v2/processing_line_only/loss_strategy_ablation/mgd +sweep: + loss_strategy: + - fm_only + mgd_token_mask_ratio: + - 0.3 +dataset: + dataset_soup: my_atomic26_human +training: + num_gpus: 1 + batch_size: 128 +model: null +phases: +- name: phase3_fm_only + max_steps: 30000 + save_steps: 0 + trainable: + tune_llm: false + tune_visual: false + tune_projector: true + tune_diffusion_model: true + losses: + mgd_enabled: false + mgd_fm_loss_weight: 1.0 + mgd_loss_weight: 0.0 + mgd_token_mask_ratio: ${sweep.mgd_token_mask_ratio} + mgd_use_cosine_loss: true + mgd_use_mse_loss: false diff --git a/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/phase3/config.json b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/phase3/config.json new file mode 100644 index 0000000000000000000000000000000000000000..114b2f28308b433790b25a434759dde354ec90c3 --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/phase3/config.json @@ -0,0 +1,78 @@ +{ + "action_dim": 32, + "action_head_cfg": { + "action_dim": 32, + "action_horizon": 16, + "add_pos_embed": true, + "backbone_embedding_dim": 2048, + "diffusion_model_cfg": { + "attention_head_dim": 48, + "cross_attention_dim": 2048, + "dropout": 0.2, + "final_dropout": true, + "interleave_self_attention": true, + "norm_type": "ada_norm", + "num_attention_heads": 32, + "num_layers": 16, + "output_dim": 1024, + "positional_embeddings": null + }, + "hidden_size": 1024, + "input_embedding_dim": 1536, + "max_action_dim": 32, + "max_state_dim": 64, + "model_dtype": "float32", + "noise_beta_alpha": 1.5, + "noise_beta_beta": 1.0, + "noise_s": 0.999, + "num_inference_timesteps": 4, + "num_target_vision_tokens": 32, + "num_timestep_buckets": 1000, + "tune_diffusion_model": true, + "tune_projector": true, + "use_vlln": true, + "vl_self_attention_cfg": { + "attention_head_dim": 64, + "dropout": 0.2, + "final_dropout": true, + "num_attention_heads": 32, + "num_layers": 4, + "positional_embeddings": null + } + }, + "action_horizon": 16, + "architectures": [ + "GR00T_N1_5_MGD" + ], + "attn_implementation": null, + "backbone_cfg": { + "eagle_path": "NVEagle/eagle_er-qwen3_1_7B-Siglip2_400M_stage1_5_128gpu_er_v7_1mlp_nops", + "load_bf16": false, + "project_to_dim": null, + "reproject_vision": false, + "select_layer": 12, + "tune_llm": false, + "tune_visual": true, + "use_flash_attention": true + }, + "compute_dtype": "bfloat16", + "hidden_size": 2048, + "mgd_enabled": false, + "mgd_fm_loss_weight": 1.0, + "mgd_loss_weight": 0.0, + "mgd_loss_weight_end": 0.0, + "mgd_loss_weight_schedule": null, + "mgd_loss_weight_start": 0.05, + "mgd_pretrained_projector_path": null, + "mgd_sequence_hidden_dim": 512, + "mgd_target_dim": 512, + "mgd_target_pooling": "flatten", + "mgd_target_projection": "frozen_random", + "mgd_token_mask_ratio": 0.3, + "mgd_use_cosine_loss": true, + "mgd_use_mse_loss": false, + "model_dtype": "float32", + "model_type": "gr00t_n1_5", + "torch_dtype": "bfloat16", + "transformers_version": "4.51.3" +} diff --git a/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/phase3/experiment_cfg/metadata.json b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/phase3/experiment_cfg/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..300eb070963337016e16465f50ba8351ea2eb25c --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/phase3/experiment_cfg/metadata.json @@ -0,0 +1,431 @@ +{ + "new_embodiment": { + "statistics": { + "state": { + "base_position": { + "max": [ + 7.3139495849609375, + 0.4876587688922882, + 0.7196521759033203 + ], + "min": [ + -4.94293737411499, + -6.890198230743408, + 0.6996827125549316 + ], + "mean": [ + 2.749288365724937, + -2.0455216239909983, + 0.700532551020533 + ], + "std": [ + 1.6786295897046262, + 1.312897969434244, + 0.00133751758906258 + ], + "q01": [ + -1.4241907881148037, + -5.1716547935950885, + 0.6999670898768614 + ], + "q99": [ + 6.198111545831555, + -0.5873531334104489, + 0.7038833014242587 + ] + }, + "base_rotation": { + "max": [ + 0.0, + 0.0, + 1.0, + 1.0 + ], + "min": [ + 0.0, + 0.0, + -1.0, + 0.0 + ], + "mean": [ + 0.0, + 0.0, + 0.23084999587450247, + 0.6238431088581847 + ], + "std": [ + 0.0, + 0.0, + 0.6556404371644047, + 0.3572252105732042 + ], + "q01": [ + 0.0, + 0.0, + -0.9999999999999999, + 1.7482354593273349e-06 + ], + "q99": [ + 0.0, + 0.0, + 0.9999999999999999, + 0.9999999999999999 + ] + }, + "end_effector_position_relative": { + "max": [ + 0.9014528393745422, + 0.8003877401351929, + 0.9829942584037781 + ], + "min": [ + -0.37471598386764526, + -0.8472502827644348, + -0.25070279836654663 + ], + "mean": [ + 0.28667332601863704, + -0.038315473170876704, + 0.45974658906777444 + ], + "std": [ + 0.17067058828305154, + 0.23569952037784195, + 0.21950702354181487 + ], + "q01": [ + 0.015172948963784228, + -0.44450502063220615, + 0.23314444103329338 + ], + "q99": [ + 0.5609069689572412, + 0.38454179128009547, + 0.7028102736473852 + ] + }, + "end_effector_rotation_relative": { + "max": [ + 0.9999998807907104, + 0.9984455108642578, + 0.9434727430343628, + 0.9062229990959167 + ], + "min": [ + -0.9999930262565613, + -0.9988337755203247, + -0.9618149995803833, + 2.1872274658107926e-07 + ], + "mean": [ + -0.25672297401336, + 0.024425335806687216, + -0.08740977164483847, + 0.16608739644018397 + ], + "std": [ + 0.7932585208392383, + 0.3102260195580416, + 0.3772472576315148, + 0.17451959300158773 + ], + "q01": [ + -0.99379006987882, + -0.5478102050038828, + -0.6101777952971636, + 0.002274085014134748 + ], + "q99": [ + 0.8804856038896288, + 0.5910611092873037, + 0.5216058042993562, + 0.5231965974012255 + ] + }, + "gripper_qpos": { + "max": [ + 0.055869169533252716, + 0.010916369967162609 + ], + "min": [ + -0.011436971835792065, + -0.05664053186774254 + ], + "mean": [ + 0.031651551516052555, + -0.03161481387920421 + ], + "std": [ + 0.013119769222217526, + 0.01306078225192402 + ], + "q01": [ + 0.0060999246446777336, + -0.04062601653964198 + ], + "q99": [ + 0.04054891621517283, + -0.0059163567113456345 + ] + } + }, + "action": { + "base_motion": { + "max": [ + 1.0, + 1.0, + 1.0, + 0.0 + ], + "min": [ + -1.0, + -1.0, + -1.0, + 0.0 + ], + "mean": [ + 0.008144896753718373, + -0.00018893135146078263, + -0.0008739062727221845, + 0.0 + ], + "std": [ + 0.11030045424364411, + 0.10148082570313594, + 0.08975698373467506, + 0.0 + ], + "q01": [ + -0.07287373067581518, + -0.09446994199569948, + -0.07702818259249572, + 0.0 + ], + "q99": [ + 0.04485485414407898, + 0.09626941605965177, + 0.07610665906122742, + 0.0 + ] + }, + "control_mode": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.9221974313425939 + ], + "std": [ + 0.38670194342612674 + ], + "q01": [ + -1.0 + ], + "q99": [ + -0.384693946883099 + ] + }, + "end_effector_position": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + 0.01976129923017491, + -0.020870631344490034, + -0.05993476517349181 + ], + "std": [ + 0.4347024990804053, + 0.4221662976569997, + 0.3845506843044066 + ], + "q01": [ + -0.8835792939767807, + -0.9280919251713778, + -0.783361071470956 + ], + "q99": [ + 0.7622084438449307, + 0.8625125252609077, + 0.7470974753250706 + ] + }, + "end_effector_rotation": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + 0.008089673068544335, + -0.026498254627927348, + 0.0029083551743059005 + ], + "std": [ + 0.11216734503733773, + 0.13206002613798629, + 0.12615870412692864 + ], + "q01": [ + -0.2814627972846672, + -0.4342386516115439, + -0.34489755628340574 + ], + "q99": [ + 0.31879353266939386, + 0.28411797683220563, + 0.3597718671669894 + ] + }, + "gripper_close": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.3824228874769045 + ], + "std": [ + 0.9240015096469786 + ], + "q01": [ + -1.0 + ], + "q99": [ + 0.5597272305813592 + ] + } + } + }, + "modalities": { + "video": { + "robot0_eye_in_hand": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + }, + "robot0_agentview_left": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + }, + "robot0_agentview_right": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + } + }, + "state": { + "base_position": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "base_rotation": { + "absolute": true, + "rotation_type": "quaternion", + "shape": [ + 4 + ], + "continuous": true + }, + "end_effector_position_relative": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "end_effector_rotation_relative": { + "absolute": true, + "rotation_type": "quaternion", + "shape": [ + 4 + ], + "continuous": true + }, + "gripper_qpos": { + "absolute": true, + "rotation_type": null, + "shape": [ + 2 + ], + "continuous": true + } + }, + "action": { + "base_motion": { + "absolute": true, + "rotation_type": null, + "shape": [ + 4 + ], + "continuous": true + }, + "control_mode": { + "absolute": true, + "rotation_type": null, + "shape": [ + 1 + ], + "continuous": true + }, + "end_effector_position": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "end_effector_rotation": { + "absolute": true, + "rotation_type": "axis_angle", + "shape": [ + 3 + ], + "continuous": true + }, + "gripper_close": { + "absolute": true, + "rotation_type": null, + "shape": [ + 1 + ], + "continuous": true + } + } + }, + "embodiment_tag": "new_embodiment" + } +} \ No newline at end of file diff --git a/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/phase3/model.safetensors.index.json b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/phase3/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..579ccf56b744c24c73f571cce697c12343122f56 --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/phase3/model.safetensors.index.json @@ -0,0 +1,916 @@ +{ + "metadata": { + "total_size": 7642572676 + }, + "weight_map": { + "action_head.action_decoder.layer1.W": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer1.b": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer2.W": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer2.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W1.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W1.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W2.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W2.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W3.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W3.b": "model-00002-of-00002.safetensors", + "action_head.future_tokens.weight": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_1.bias": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_1.weight": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_2.bias": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_2.weight": "model-00002-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_1.bias": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_1.weight": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_2.bias": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.position_embedding.weight": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer1.W": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer1.b": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer2.W": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer2.b": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vlln.bias": "model-00002-of-00002.safetensors", + "action_head.vlln.weight": "model-00002-of-00002.safetensors", + "action_target_projector.projector.bias": "model-00002-of-00002.safetensors", + "action_target_projector.projector.weight": "model-00002-of-00002.safetensors", + "backbone.eagle_model.language_model.lm_head.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.embed_tokens.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.mlp1.0.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.mlp1.0.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.patch_embedding.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.patch_embedding.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.position_embedding.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.in_proj_bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.in_proj_weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.layernorm.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.probe": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.post_layernorm.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.post_layernorm.weight": "model-00001-of-00002.safetensors", + "sequence_mgd_head.attn_score.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.attn_score.weight": "model-00002-of-00002.safetensors", + "sequence_mgd_head.out_proj.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.out_proj.weight": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.0.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.0.weight": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.2.bias": "model-00002-of-00002.safetensors", + "sequence_mgd_head.token_mlp.2.weight": "model-00002-of-00002.safetensors" + } +} diff --git a/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/resolved_config.yaml b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/resolved_config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..4ed7f4fa95a16ea476cdfc56fab77cedeb43af69 --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/mgd/loss_strategy_fm_only__mgd_token_mask_ratio_0.3/resolved_config.yaml @@ -0,0 +1,34 @@ +experiment_name: mgd_phase3_fm_only +policy_type: groot_mgd +base_model_path: ${HOME}/groot_robocasa/robocasa_v2/MGD_Processing_line_Only/mgd_token_mask_ratio_0.3/phase2 +output_root: ${HOME}/groot_robocasa/robocasa_v2/processing_line_only/loss_strategy_ablation/mgd +sweep: + loss_strategy: + - fm_only + mgd_token_mask_ratio: + - 0.3 +dataset: + dataset_soup: my_atomic26_human +training: + num_gpus: 1 + batch_size: 128 +model: null +phases: +- name: phase3_fm_only + max_steps: 30000 + save_steps: 0 + trainable: + tune_llm: false + tune_visual: false + tune_projector: true + tune_diffusion_model: true + losses: + mgd_enabled: false + mgd_fm_loss_weight: 1.0 + mgd_loss_weight: 0.0 + mgd_token_mask_ratio: 0.3 + mgd_use_cosine_loss: true + mgd_use_mse_loss: false +resolved_sweep: + loss_strategy: fm_only + mgd_token_mask_ratio: 0.3 diff --git a/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/phase3/config.json b/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/phase3/config.json new file mode 100644 index 0000000000000000000000000000000000000000..27b0fe5c63e5caf9fb7651415448c778f194f459 --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/phase3/config.json @@ -0,0 +1,73 @@ +{ + "action_dim": 32, + "action_head_cfg": { + "action_dim": 32, + "action_horizon": 16, + "add_pos_embed": true, + "backbone_embedding_dim": 2048, + "diffusion_model_cfg": { + "attention_head_dim": 48, + "cross_attention_dim": 2048, + "dropout": 0.2, + "final_dropout": true, + "interleave_self_attention": true, + "norm_type": "ada_norm", + "num_attention_heads": 32, + "num_layers": 16, + "output_dim": 1024, + "positional_embeddings": null + }, + "hidden_size": 1024, + "input_embedding_dim": 1536, + "max_action_dim": 32, + "max_state_dim": 64, + "model_dtype": "float32", + "noise_beta_alpha": 1.5, + "noise_beta_beta": 1.0, + "noise_s": 0.999, + "num_inference_timesteps": 4, + "num_target_vision_tokens": 32, + "num_timestep_buckets": 1000, + "tune_diffusion_model": true, + "tune_projector": true, + "use_vlln": true, + "vl_self_attention_cfg": { + "attention_head_dim": 64, + "dropout": 0.2, + "final_dropout": true, + "num_attention_heads": 32, + "num_layers": 4, + "positional_embeddings": null + } + }, + "action_horizon": 16, + "architectures": [ + "GR00T_N1_5_RKD" + ], + "attn_implementation": null, + "backbone_cfg": { + "eagle_path": "NVEagle/eagle_er-qwen3_1_7B-Siglip2_400M_stage1_5_128gpu_er_v7_1mlp_nops", + "load_bf16": false, + "project_to_dim": null, + "reproject_vision": false, + "select_layer": 12, + "tune_llm": false, + "tune_visual": true, + "use_flash_attention": true + }, + "compute_dtype": "bfloat16", + "hidden_size": 2048, + "model_dtype": "float32", + "model_type": "gr00t_n1_5", + "rkd_action_temp": 0.1, + "rkd_enabled": true, + "rkd_fm_loss_weight": 1.0, + "rkd_loss_weight": 0.0, + "rkd_loss_weight_end": 0.0, + "rkd_loss_weight_schedule": "linear_decay", + "rkd_loss_weight_start": 1.0, + "rkd_relation_mode": "token_pair_mean", + "rkd_vlm_temp": 0.1, + "torch_dtype": "bfloat16", + "transformers_version": "4.51.3" +} diff --git a/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/phase3/experiment_cfg/metadata.json b/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/phase3/experiment_cfg/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..300eb070963337016e16465f50ba8351ea2eb25c --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/phase3/experiment_cfg/metadata.json @@ -0,0 +1,431 @@ +{ + "new_embodiment": { + "statistics": { + "state": { + "base_position": { + "max": [ + 7.3139495849609375, + 0.4876587688922882, + 0.7196521759033203 + ], + "min": [ + -4.94293737411499, + -6.890198230743408, + 0.6996827125549316 + ], + "mean": [ + 2.749288365724937, + -2.0455216239909983, + 0.700532551020533 + ], + "std": [ + 1.6786295897046262, + 1.312897969434244, + 0.00133751758906258 + ], + "q01": [ + -1.4241907881148037, + -5.1716547935950885, + 0.6999670898768614 + ], + "q99": [ + 6.198111545831555, + -0.5873531334104489, + 0.7038833014242587 + ] + }, + "base_rotation": { + "max": [ + 0.0, + 0.0, + 1.0, + 1.0 + ], + "min": [ + 0.0, + 0.0, + -1.0, + 0.0 + ], + "mean": [ + 0.0, + 0.0, + 0.23084999587450247, + 0.6238431088581847 + ], + "std": [ + 0.0, + 0.0, + 0.6556404371644047, + 0.3572252105732042 + ], + "q01": [ + 0.0, + 0.0, + -0.9999999999999999, + 1.7482354593273349e-06 + ], + "q99": [ + 0.0, + 0.0, + 0.9999999999999999, + 0.9999999999999999 + ] + }, + "end_effector_position_relative": { + "max": [ + 0.9014528393745422, + 0.8003877401351929, + 0.9829942584037781 + ], + "min": [ + -0.37471598386764526, + -0.8472502827644348, + -0.25070279836654663 + ], + "mean": [ + 0.28667332601863704, + -0.038315473170876704, + 0.45974658906777444 + ], + "std": [ + 0.17067058828305154, + 0.23569952037784195, + 0.21950702354181487 + ], + "q01": [ + 0.015172948963784228, + -0.44450502063220615, + 0.23314444103329338 + ], + "q99": [ + 0.5609069689572412, + 0.38454179128009547, + 0.7028102736473852 + ] + }, + "end_effector_rotation_relative": { + "max": [ + 0.9999998807907104, + 0.9984455108642578, + 0.9434727430343628, + 0.9062229990959167 + ], + "min": [ + -0.9999930262565613, + -0.9988337755203247, + -0.9618149995803833, + 2.1872274658107926e-07 + ], + "mean": [ + -0.25672297401336, + 0.024425335806687216, + -0.08740977164483847, + 0.16608739644018397 + ], + "std": [ + 0.7932585208392383, + 0.3102260195580416, + 0.3772472576315148, + 0.17451959300158773 + ], + "q01": [ + -0.99379006987882, + -0.5478102050038828, + -0.6101777952971636, + 0.002274085014134748 + ], + "q99": [ + 0.8804856038896288, + 0.5910611092873037, + 0.5216058042993562, + 0.5231965974012255 + ] + }, + "gripper_qpos": { + "max": [ + 0.055869169533252716, + 0.010916369967162609 + ], + "min": [ + -0.011436971835792065, + -0.05664053186774254 + ], + "mean": [ + 0.031651551516052555, + -0.03161481387920421 + ], + "std": [ + 0.013119769222217526, + 0.01306078225192402 + ], + "q01": [ + 0.0060999246446777336, + -0.04062601653964198 + ], + "q99": [ + 0.04054891621517283, + -0.0059163567113456345 + ] + } + }, + "action": { + "base_motion": { + "max": [ + 1.0, + 1.0, + 1.0, + 0.0 + ], + "min": [ + -1.0, + -1.0, + -1.0, + 0.0 + ], + "mean": [ + 0.008144896753718373, + -0.00018893135146078263, + -0.0008739062727221845, + 0.0 + ], + "std": [ + 0.11030045424364411, + 0.10148082570313594, + 0.08975698373467506, + 0.0 + ], + "q01": [ + -0.07287373067581518, + -0.09446994199569948, + -0.07702818259249572, + 0.0 + ], + "q99": [ + 0.04485485414407898, + 0.09626941605965177, + 0.07610665906122742, + 0.0 + ] + }, + "control_mode": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.9221974313425939 + ], + "std": [ + 0.38670194342612674 + ], + "q01": [ + -1.0 + ], + "q99": [ + -0.384693946883099 + ] + }, + "end_effector_position": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + 0.01976129923017491, + -0.020870631344490034, + -0.05993476517349181 + ], + "std": [ + 0.4347024990804053, + 0.4221662976569997, + 0.3845506843044066 + ], + "q01": [ + -0.8835792939767807, + -0.9280919251713778, + -0.783361071470956 + ], + "q99": [ + 0.7622084438449307, + 0.8625125252609077, + 0.7470974753250706 + ] + }, + "end_effector_rotation": { + "max": [ + 1.0, + 1.0, + 1.0 + ], + "min": [ + -1.0, + -1.0, + -1.0 + ], + "mean": [ + 0.008089673068544335, + -0.026498254627927348, + 0.0029083551743059005 + ], + "std": [ + 0.11216734503733773, + 0.13206002613798629, + 0.12615870412692864 + ], + "q01": [ + -0.2814627972846672, + -0.4342386516115439, + -0.34489755628340574 + ], + "q99": [ + 0.31879353266939386, + 0.28411797683220563, + 0.3597718671669894 + ] + }, + "gripper_close": { + "max": [ + 1.0 + ], + "min": [ + -1.0 + ], + "mean": [ + -0.3824228874769045 + ], + "std": [ + 0.9240015096469786 + ], + "q01": [ + -1.0 + ], + "q99": [ + 0.5597272305813592 + ] + } + } + }, + "modalities": { + "video": { + "robot0_eye_in_hand": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + }, + "robot0_agentview_left": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + }, + "robot0_agentview_right": { + "resolution": [ + 256, + 256 + ], + "channels": 3, + "fps": 20.0 + } + }, + "state": { + "base_position": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "base_rotation": { + "absolute": true, + "rotation_type": "quaternion", + "shape": [ + 4 + ], + "continuous": true + }, + "end_effector_position_relative": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "end_effector_rotation_relative": { + "absolute": true, + "rotation_type": "quaternion", + "shape": [ + 4 + ], + "continuous": true + }, + "gripper_qpos": { + "absolute": true, + "rotation_type": null, + "shape": [ + 2 + ], + "continuous": true + } + }, + "action": { + "base_motion": { + "absolute": true, + "rotation_type": null, + "shape": [ + 4 + ], + "continuous": true + }, + "control_mode": { + "absolute": true, + "rotation_type": null, + "shape": [ + 1 + ], + "continuous": true + }, + "end_effector_position": { + "absolute": true, + "rotation_type": null, + "shape": [ + 3 + ], + "continuous": true + }, + "end_effector_rotation": { + "absolute": true, + "rotation_type": "axis_angle", + "shape": [ + 3 + ], + "continuous": true + }, + "gripper_close": { + "absolute": true, + "rotation_type": null, + "shape": [ + 1 + ], + "continuous": true + } + } + }, + "embodiment_tag": "new_embodiment" + } +} \ No newline at end of file diff --git a/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/phase3/model.safetensors.index.json b/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/phase3/model.safetensors.index.json new file mode 100644 index 0000000000000000000000000000000000000000..5ef5f247b61eb93e89601c7da8334fc087e19750 --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/phase3/model.safetensors.index.json @@ -0,0 +1,906 @@ +{ + "metadata": { + "total_size": 7585939328 + }, + "weight_map": { + "action_head.action_decoder.layer1.W": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer1.b": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer2.W": "model-00002-of-00002.safetensors", + "action_head.action_decoder.layer2.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W1.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W1.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W2.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W2.b": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W3.W": "model-00002-of-00002.safetensors", + "action_head.action_encoder.W3.b": "model-00002-of-00002.safetensors", + "action_head.future_tokens.weight": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_1.bias": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_1.weight": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_2.bias": "model-00002-of-00002.safetensors", + "action_head.model.proj_out_2.weight": "model-00002-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_1.bias": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_1.weight": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_2.bias": "model-00001-of-00002.safetensors", + "action_head.model.timestep_encoder.timestep_embedder.linear_2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.0.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.1.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.10.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.11.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.12.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.12.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.13.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.14.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.norm1.linear.bias": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.15.norm1.linear.weight": "model-00002-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.2.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.3.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.4.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.5.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.6.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.7.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.8.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_k.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_k.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_out.0.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_out.0.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_q.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_q.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_v.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.attn1.to_v.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.0.proj.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.0.proj.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.2.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.ff.net.2.weight": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.norm1.linear.bias": "model-00001-of-00002.safetensors", + "action_head.model.transformer_blocks.9.norm1.linear.weight": "model-00001-of-00002.safetensors", + "action_head.position_embedding.weight": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer1.W": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer1.b": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer2.W": "model-00002-of-00002.safetensors", + "action_head.state_encoder.layer2.b": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.0.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.1.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.2.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_k.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_k.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_out.0.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_out.0.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_q.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_q.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_v.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.attn1.to_v.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.0.proj.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.0.proj.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.2.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.ff.net.2.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm1.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm1.weight": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm3.bias": "model-00002-of-00002.safetensors", + "action_head.vl_self_attention.transformer_blocks.3.norm3.weight": "model-00002-of-00002.safetensors", + "action_head.vlln.bias": "model-00002-of-00002.safetensors", + "action_head.vlln.weight": "model-00002-of-00002.safetensors", + "backbone.eagle_model.language_model.lm_head.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.embed_tokens.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.down_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.gate_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.mlp.up_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.k_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.o_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.q_norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.language_model.model.norm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.mlp1.0.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.mlp1.0.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.patch_embedding.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.patch_embedding.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.embeddings.position_embedding.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.0.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.1.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.10.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.11.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.12.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.13.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.14.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.15.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.16.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.17.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.18.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.19.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.2.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.20.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.21.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.22.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.23.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.24.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.25.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.26.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.3.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.4.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.5.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.6.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.7.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.8.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.layer_norm2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.k_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.k_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.q_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.q_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.v_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.encoder.layers.9.self_attn.v_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.in_proj_bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.in_proj_weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.out_proj.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.attention.out_proj.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.layernorm.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.layernorm.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc1.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc1.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc2.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.mlp.fc2.weight": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.head.probe": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.post_layernorm.bias": "model-00001-of-00002.safetensors", + "backbone.eagle_model.vision_model.vision_model.post_layernorm.weight": "model-00001-of-00002.safetensors" + } +} diff --git a/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/phase3/trainer_state.json b/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/phase3/trainer_state.json new file mode 100644 index 0000000000000000000000000000000000000000..53cc50b42e6545c0e44e4db30e933e986ea2162b --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/phase3/trainer_state.json @@ -0,0 +1,51043 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 8.728542333430317, + "eval_steps": 500, + "global_step": 30000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.002909514111143439, + "flow_matching_loss": 0.14025882333517076, + "grad_norm": 0.96406489610672, + "learning_rate": 1.8e-07, + "loss": 0.1503, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010033373860642314, + "rkd_loss_weight": 0.9998499929904938, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.42724871039390566, + "rkd_teacher_entropy": 3.7180015563964846, + "rkd_teacher_similarity_mean": 0.7722879409790039, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10 + }, + { + "epoch": 0.005819028222286878, + "flow_matching_loss": 0.13359030038118364, + "grad_norm": 0.9579328298568726, + "learning_rate": 3.8e-07, + "loss": 0.1429, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009341436438262463, + "rkd_loss_weight": 0.9995166480541229, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.43852487206459045, + "rkd_teacher_entropy": 3.739339256286621, + "rkd_teacher_similarity_mean": 0.7829276084899902, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20 + }, + { + "epoch": 0.008728542333430317, + "flow_matching_loss": 0.13304784297943115, + "grad_norm": 1.0542857646942139, + "learning_rate": 5.800000000000001e-07, + "loss": 0.1436, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010525110596790909, + "rkd_loss_weight": 0.9991833090782165, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.43663943707942965, + "rkd_teacher_entropy": 3.72597336769104, + "rkd_teacher_similarity_mean": 0.7821591317653656, + "rkd_vlm_temp": 0.10000000149011612, + "step": 30 + }, + { + "epoch": 0.011638056444573755, + "flow_matching_loss": 0.09897692129015923, + "grad_norm": 0.5641133189201355, + "learning_rate": 7.799999999999999e-07, + "loss": 0.1085, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009534950833767653, + "rkd_loss_weight": 0.9988499581813812, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.43553556501865387, + "rkd_teacher_entropy": 3.7382612943649294, + "rkd_teacher_similarity_mean": 0.7814847946166992, + "rkd_vlm_temp": 0.10000000149011612, + "step": 40 + }, + { + "epoch": 0.014547570555717195, + "flow_matching_loss": 0.09554776214063168, + "grad_norm": 0.5771467089653015, + "learning_rate": 9.8e-07, + "loss": 0.1039, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008347280416637659, + "rkd_loss_weight": 0.9985166192054749, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.43164077699184417, + "rkd_teacher_entropy": 3.7146430015563965, + "rkd_teacher_similarity_mean": 0.7777485966682434, + "rkd_vlm_temp": 0.10000000149011612, + "step": 50 + }, + { + "epoch": 0.017457084666860634, + "flow_matching_loss": 0.10441746711730956, + "grad_norm": 0.8603259325027466, + "learning_rate": 1.18e-06, + "loss": 0.1133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008933236077427863, + "rkd_loss_weight": 0.998183274269104, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4353631198406219, + "rkd_teacher_entropy": 3.742427635192871, + "rkd_teacher_similarity_mean": 0.7820838630199433, + "rkd_vlm_temp": 0.10000000149011612, + "step": 60 + }, + { + "epoch": 0.020366598778004074, + "flow_matching_loss": 0.10942962542176246, + "grad_norm": 0.9414843916893005, + "learning_rate": 1.38e-06, + "loss": 0.1183, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008873872319236397, + "rkd_loss_weight": 0.9978499233722686, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4381743609905243, + "rkd_teacher_entropy": 3.7436843633651735, + "rkd_teacher_similarity_mean": 0.7858612179756165, + "rkd_vlm_temp": 0.10000000149011612, + "step": 70 + }, + { + "epoch": 0.02327611288914751, + "flow_matching_loss": 0.08123084902763367, + "grad_norm": 0.5137049555778503, + "learning_rate": 1.5800000000000001e-06, + "loss": 0.0891, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007894788961857557, + "rkd_loss_weight": 0.9975165903568268, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4428290337324142, + "rkd_teacher_entropy": 3.7553800106048585, + "rkd_teacher_similarity_mean": 0.790579479932785, + "rkd_vlm_temp": 0.10000000149011612, + "step": 80 + }, + { + "epoch": 0.02618562700029095, + "flow_matching_loss": 0.08918447270989419, + "grad_norm": 0.6402997970581055, + "learning_rate": 1.7800000000000001e-06, + "loss": 0.0983, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00910939248278737, + "rkd_loss_weight": 0.997183233499527, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.44081381261348723, + "rkd_teacher_entropy": 3.7617879629135134, + "rkd_teacher_similarity_mean": 0.7880800902843476, + "rkd_vlm_temp": 0.10000000149011612, + "step": 90 + }, + { + "epoch": 0.02909514111143439, + "flow_matching_loss": 0.07274332121014596, + "grad_norm": 0.30616456270217896, + "learning_rate": 1.98e-06, + "loss": 0.083, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010291284276172518, + "rkd_loss_weight": 0.9968498945236206, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4329862654209137, + "rkd_teacher_entropy": 3.7252931118011476, + "rkd_teacher_similarity_mean": 0.779593825340271, + "rkd_vlm_temp": 0.10000000149011612, + "step": 100 + }, + { + "epoch": 0.03200465522257783, + "flow_matching_loss": 0.08191884830594062, + "grad_norm": 0.4191439747810364, + "learning_rate": 2.1800000000000003e-06, + "loss": 0.0911, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00924587226472795, + "rkd_loss_weight": 0.9965165555477142, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4299120754003525, + "rkd_teacher_entropy": 3.730306100845337, + "rkd_teacher_similarity_mean": 0.7771810472011567, + "rkd_vlm_temp": 0.10000000149011612, + "step": 110 + }, + { + "epoch": 0.03491416933372127, + "flow_matching_loss": 0.06738889999687672, + "grad_norm": 0.3135012090206146, + "learning_rate": 2.38e-06, + "loss": 0.0786, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011264462303370238, + "rkd_loss_weight": 0.9961831986904144, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.43083401918411257, + "rkd_teacher_entropy": 3.726189637184143, + "rkd_teacher_similarity_mean": 0.7780045986175537, + "rkd_vlm_temp": 0.10000000149011612, + "step": 120 + }, + { + "epoch": 0.037823683444864704, + "flow_matching_loss": 0.06838675327599049, + "grad_norm": 0.35055360198020935, + "learning_rate": 2.58e-06, + "loss": 0.0775, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009191860072314739, + "rkd_loss_weight": 0.9958498656749726, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.42816458344459535, + "rkd_teacher_entropy": 3.733244276046753, + "rkd_teacher_similarity_mean": 0.7779533863067627, + "rkd_vlm_temp": 0.10000000149011612, + "step": 130 + }, + { + "epoch": 0.04073319755600815, + "flow_matching_loss": 0.07096878439188004, + "grad_norm": 0.23710861802101135, + "learning_rate": 2.78e-06, + "loss": 0.0802, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009291068743914365, + "rkd_loss_weight": 0.9955165147781372, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4317079663276672, + "rkd_teacher_entropy": 3.7422439098358153, + "rkd_teacher_similarity_mean": 0.7806917548179626, + "rkd_vlm_temp": 0.10000000149011612, + "step": 140 + }, + { + "epoch": 0.043642711667151585, + "flow_matching_loss": 0.07115022018551827, + "grad_norm": 0.5185287594795227, + "learning_rate": 2.98e-06, + "loss": 0.0805, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009346059802919627, + "rkd_loss_weight": 0.9951831698417664, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4246551513671875, + "rkd_teacher_entropy": 3.719101619720459, + "rkd_teacher_similarity_mean": 0.7753975629806519, + "rkd_vlm_temp": 0.10000000149011612, + "step": 150 + }, + { + "epoch": 0.04655222577829502, + "flow_matching_loss": 0.06526296474039554, + "grad_norm": 0.3857128322124481, + "learning_rate": 3.18e-06, + "loss": 0.0754, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010148151777684689, + "rkd_loss_weight": 0.99484983086586, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.42836311757564544, + "rkd_teacher_entropy": 3.7381389856338503, + "rkd_teacher_similarity_mean": 0.7802381098270417, + "rkd_vlm_temp": 0.10000000149011612, + "step": 160 + }, + { + "epoch": 0.049461739889438465, + "flow_matching_loss": 0.05379932001233101, + "grad_norm": 0.40545976161956787, + "learning_rate": 3.38e-06, + "loss": 0.0652, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01143162059597671, + "rkd_loss_weight": 0.9945164799690247, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4198562830686569, + "rkd_teacher_entropy": 3.6976291656494142, + "rkd_teacher_similarity_mean": 0.7705481946468353, + "rkd_vlm_temp": 0.10000000149011612, + "step": 170 + }, + { + "epoch": 0.0523712540005819, + "flow_matching_loss": 0.05084514990448952, + "grad_norm": 0.3342762589454651, + "learning_rate": 3.58e-06, + "loss": 0.0594, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008612280339002609, + "rkd_loss_weight": 0.9941831409931183, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.42157893180847167, + "rkd_teacher_entropy": 3.7126569509506226, + "rkd_teacher_similarity_mean": 0.7739878952503204, + "rkd_vlm_temp": 0.10000000149011612, + "step": 180 + }, + { + "epoch": 0.055280768111725345, + "flow_matching_loss": 0.05238770171999931, + "grad_norm": 0.36431318521499634, + "learning_rate": 3.7800000000000002e-06, + "loss": 0.0632, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010865884739905596, + "rkd_loss_weight": 0.9938497960567474, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4168315589427948, + "rkd_teacher_entropy": 3.6951599597930906, + "rkd_teacher_similarity_mean": 0.7705600500106812, + "rkd_vlm_temp": 0.10000000149011612, + "step": 190 + }, + { + "epoch": 0.05819028222286878, + "flow_matching_loss": 0.051006301492452624, + "grad_norm": 0.300343781709671, + "learning_rate": 3.98e-06, + "loss": 0.0606, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00961840059608221, + "rkd_loss_weight": 0.9935164451599121, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.42619766891002653, + "rkd_teacher_entropy": 3.7344072103500365, + "rkd_teacher_similarity_mean": 0.7815191864967346, + "rkd_vlm_temp": 0.10000000149011612, + "step": 200 + }, + { + "epoch": 0.06109979633401222, + "flow_matching_loss": 0.05319566689431667, + "grad_norm": 0.4741889536380768, + "learning_rate": 4.18e-06, + "loss": 0.0679, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014807002153247595, + "rkd_loss_weight": 0.9931831061840057, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4218900740146637, + "rkd_teacher_entropy": 3.7323689222335816, + "rkd_teacher_similarity_mean": 0.778412526845932, + "rkd_vlm_temp": 0.10000000149011612, + "step": 210 + }, + { + "epoch": 0.06400931044515566, + "flow_matching_loss": 0.04879615493118763, + "grad_norm": 0.2915203869342804, + "learning_rate": 4.3799999999999996e-06, + "loss": 0.0597, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010936688538640738, + "rkd_loss_weight": 0.9928497612476349, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4186790943145752, + "rkd_teacher_entropy": 3.719696879386902, + "rkd_teacher_similarity_mean": 0.7753785371780395, + "rkd_vlm_temp": 0.10000000149011612, + "step": 220 + }, + { + "epoch": 0.06691882455629909, + "flow_matching_loss": 0.053802769631147385, + "grad_norm": 0.3287517726421356, + "learning_rate": 4.58e-06, + "loss": 0.0638, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010106049990281463, + "rkd_loss_weight": 0.9925164163112641, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4077430427074432, + "rkd_teacher_entropy": 3.6891262531280518, + "rkd_teacher_similarity_mean": 0.7658524930477142, + "rkd_vlm_temp": 0.10000000149011612, + "step": 230 + }, + { + "epoch": 0.06982833866744254, + "flow_matching_loss": 0.05374416634440422, + "grad_norm": 0.32534557580947876, + "learning_rate": 4.78e-06, + "loss": 0.063, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009322842815890908, + "rkd_loss_weight": 0.9921830713748931, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4143189311027527, + "rkd_teacher_entropy": 3.706947135925293, + "rkd_teacher_similarity_mean": 0.7720498502254486, + "rkd_vlm_temp": 0.10000000149011612, + "step": 240 + }, + { + "epoch": 0.07273785277858598, + "flow_matching_loss": 0.047470588237047195, + "grad_norm": 0.5180434584617615, + "learning_rate": 4.980000000000001e-06, + "loss": 0.0582, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010808159690350294, + "rkd_loss_weight": 0.9918497323989868, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.41129393577575685, + "rkd_teacher_entropy": 3.7097446441650392, + "rkd_teacher_similarity_mean": 0.7715014398097992, + "rkd_vlm_temp": 0.10000000149011612, + "step": 250 + }, + { + "epoch": 0.07564736688972941, + "flow_matching_loss": 0.04419988039880991, + "grad_norm": 0.20404763519763947, + "learning_rate": 5.18e-06, + "loss": 0.0541, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010003960598260165, + "rkd_loss_weight": 0.9915163815021515, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4135258704423904, + "rkd_teacher_entropy": 3.7064817190170287, + "rkd_teacher_similarity_mean": 0.7746111214160919, + "rkd_vlm_temp": 0.10000000149011612, + "step": 260 + }, + { + "epoch": 0.07855688100087285, + "flow_matching_loss": 0.048570461198687555, + "grad_norm": 0.3614705502986908, + "learning_rate": 5.38e-06, + "loss": 0.0595, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011060095857828856, + "rkd_loss_weight": 0.9911830425262451, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4171010613441467, + "rkd_teacher_entropy": 3.7299842119216917, + "rkd_teacher_similarity_mean": 0.7773393511772155, + "rkd_vlm_temp": 0.10000000149011612, + "step": 270 + }, + { + "epoch": 0.0814663951120163, + "flow_matching_loss": 0.043953803926706315, + "grad_norm": 0.24692820012569427, + "learning_rate": 5.58e-06, + "loss": 0.0548, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01093492629006505, + "rkd_loss_weight": 0.9908496975898743, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.41139804720878603, + "rkd_teacher_entropy": 3.701107931137085, + "rkd_teacher_similarity_mean": 0.7720763444900512, + "rkd_vlm_temp": 0.10000000149011612, + "step": 280 + }, + { + "epoch": 0.08437590922315973, + "flow_matching_loss": 0.043959983997046946, + "grad_norm": 0.19703049957752228, + "learning_rate": 5.7800000000000006e-06, + "loss": 0.0526, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008759035542607308, + "rkd_loss_weight": 0.9905163466930389, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.42213018238544464, + "rkd_teacher_entropy": 3.7294888734817504, + "rkd_teacher_similarity_mean": 0.7830714225769043, + "rkd_vlm_temp": 0.10000000149011612, + "step": 290 + }, + { + "epoch": 0.08728542333430317, + "flow_matching_loss": 0.04915007837116718, + "grad_norm": 0.27172303199768066, + "learning_rate": 5.98e-06, + "loss": 0.0601, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011092372611165047, + "rkd_loss_weight": 0.9901830136775971, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.41086157262325285, + "rkd_teacher_entropy": 3.7085954666137697, + "rkd_teacher_similarity_mean": 0.7708386182785034, + "rkd_vlm_temp": 0.10000000149011612, + "step": 300 + }, + { + "epoch": 0.09019493744544661, + "flow_matching_loss": 0.045510361343622206, + "grad_norm": 0.25799813866615295, + "learning_rate": 6.18e-06, + "loss": 0.0555, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010054818354547024, + "rkd_loss_weight": 0.9898496568202972, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4111660659313202, + "rkd_teacher_entropy": 3.7062599897384643, + "rkd_teacher_similarity_mean": 0.7712823808193207, + "rkd_vlm_temp": 0.10000000149011612, + "step": 310 + }, + { + "epoch": 0.09310445155659004, + "flow_matching_loss": 0.035994429513812064, + "grad_norm": 0.3375902473926544, + "learning_rate": 6.38e-06, + "loss": 0.0456, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009748045587912202, + "rkd_loss_weight": 0.9895163178443909, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.41193203032016756, + "rkd_teacher_entropy": 3.6975343942642214, + "rkd_teacher_similarity_mean": 0.7731882333755493, + "rkd_vlm_temp": 0.10000000149011612, + "step": 320 + }, + { + "epoch": 0.09601396566773349, + "flow_matching_loss": 0.04519069716334343, + "grad_norm": 0.25600525736808777, + "learning_rate": 6.58e-06, + "loss": 0.0576, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012505038874223828, + "rkd_loss_weight": 0.9891829788684845, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4113554507493973, + "rkd_teacher_entropy": 3.7152222633361816, + "rkd_teacher_similarity_mean": 0.7739176750183105, + "rkd_vlm_temp": 0.10000000149011612, + "step": 330 + }, + { + "epoch": 0.09892347977887693, + "flow_matching_loss": 0.049883341416716576, + "grad_norm": 0.33594194054603577, + "learning_rate": 6.78e-06, + "loss": 0.0593, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009500925382599234, + "rkd_loss_weight": 0.9888496220111846, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.40974573493003846, + "rkd_teacher_entropy": 3.7088528156280516, + "rkd_teacher_similarity_mean": 0.7711690962314606, + "rkd_vlm_temp": 0.10000000149011612, + "step": 340 + }, + { + "epoch": 0.10183299389002037, + "flow_matching_loss": 0.0393703255802393, + "grad_norm": 0.2679912745952606, + "learning_rate": 6.98e-06, + "loss": 0.049, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009742550505325197, + "rkd_loss_weight": 0.9885162889957428, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.413285431265831, + "rkd_teacher_entropy": 3.709138059616089, + "rkd_teacher_similarity_mean": 0.7741859436035157, + "rkd_vlm_temp": 0.10000000149011612, + "step": 350 + }, + { + "epoch": 0.1047425080011638, + "flow_matching_loss": 0.04411491490900517, + "grad_norm": 0.2521236836910248, + "learning_rate": 7.180000000000001e-06, + "loss": 0.0542, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01018662927672267, + "rkd_loss_weight": 0.9881829380989074, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.40720858573913576, + "rkd_teacher_entropy": 3.6850615978240966, + "rkd_teacher_similarity_mean": 0.7687876701354981, + "rkd_vlm_temp": 0.10000000149011612, + "step": 360 + }, + { + "epoch": 0.10765202211230725, + "flow_matching_loss": 0.03959639649838209, + "grad_norm": 0.20597384870052338, + "learning_rate": 7.3800000000000005e-06, + "loss": 0.0496, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010091595025733114, + "rkd_loss_weight": 0.9878495931625366, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4056540936231613, + "rkd_teacher_entropy": 3.6962419033050535, + "rkd_teacher_similarity_mean": 0.7687727630138397, + "rkd_vlm_temp": 0.10000000149011612, + "step": 370 + }, + { + "epoch": 0.11056153622345069, + "flow_matching_loss": 0.03867247961461544, + "grad_norm": 0.26363131403923035, + "learning_rate": 7.5799999999999994e-06, + "loss": 0.0496, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011085089482367038, + "rkd_loss_weight": 0.9875162541866302, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4098743677139282, + "rkd_teacher_entropy": 3.7079479932785033, + "rkd_teacher_similarity_mean": 0.7735271036624909, + "rkd_vlm_temp": 0.10000000149011612, + "step": 380 + }, + { + "epoch": 0.11347105033459412, + "flow_matching_loss": 0.03747809287160635, + "grad_norm": 0.24334818124771118, + "learning_rate": 7.780000000000002e-06, + "loss": 0.0479, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010515644680708647, + "rkd_loss_weight": 0.987182903289795, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.42303854823112486, + "rkd_teacher_entropy": 3.74943265914917, + "rkd_teacher_similarity_mean": 0.7870156347751618, + "rkd_vlm_temp": 0.10000000149011612, + "step": 390 + }, + { + "epoch": 0.11638056444573756, + "flow_matching_loss": 0.039221318997442724, + "grad_norm": 0.2434045970439911, + "learning_rate": 7.98e-06, + "loss": 0.0494, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010268703708425165, + "rkd_loss_weight": 0.9868495643138886, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.40889482796192167, + "rkd_teacher_entropy": 3.7118878602981566, + "rkd_teacher_similarity_mean": 0.7726708173751831, + "rkd_vlm_temp": 0.10000000149011612, + "step": 400 + }, + { + "epoch": 0.11929007855688101, + "flow_matching_loss": 0.04182341694831848, + "grad_norm": 0.34865933656692505, + "learning_rate": 8.18e-06, + "loss": 0.0539, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012192778754979372, + "rkd_loss_weight": 0.9865162193775177, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.41697209775447847, + "rkd_teacher_entropy": 3.735508894920349, + "rkd_teacher_similarity_mean": 0.7803588271141052, + "rkd_vlm_temp": 0.10000000149011612, + "step": 410 + }, + { + "epoch": 0.12219959266802444, + "flow_matching_loss": 0.037792216427624224, + "grad_norm": 0.2213912010192871, + "learning_rate": 8.38e-06, + "loss": 0.0473, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009630964882671834, + "rkd_loss_weight": 0.9861828684806824, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4105775892734528, + "rkd_teacher_entropy": 3.7052868127822878, + "rkd_teacher_similarity_mean": 0.7742617011070252, + "rkd_vlm_temp": 0.10000000149011612, + "step": 420 + }, + { + "epoch": 0.12510910677916787, + "flow_matching_loss": 0.04301798343658447, + "grad_norm": 0.29437413811683655, + "learning_rate": 8.58e-06, + "loss": 0.0541, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01122089670971036, + "rkd_loss_weight": 0.985849529504776, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.40476627051830294, + "rkd_teacher_entropy": 3.7088760137557983, + "rkd_teacher_similarity_mean": 0.7690805077552796, + "rkd_vlm_temp": 0.10000000149011612, + "step": 430 + }, + { + "epoch": 0.12801862089031132, + "flow_matching_loss": 0.03877781666815281, + "grad_norm": 0.2545859217643738, + "learning_rate": 8.78e-06, + "loss": 0.0498, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01114336559548974, + "rkd_loss_weight": 0.9855161845684052, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4005724549293518, + "rkd_teacher_entropy": 3.6904345989227294, + "rkd_teacher_similarity_mean": 0.7649206995964051, + "rkd_vlm_temp": 0.10000000149011612, + "step": 440 + }, + { + "epoch": 0.13092813500145475, + "flow_matching_loss": 0.03672951515763998, + "grad_norm": 0.26151135563850403, + "learning_rate": 8.98e-06, + "loss": 0.0461, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009553957544267178, + "rkd_loss_weight": 0.9851828396320343, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4136164397001266, + "rkd_teacher_entropy": 3.7199615955352785, + "rkd_teacher_similarity_mean": 0.7779690504074097, + "rkd_vlm_temp": 0.10000000149011612, + "step": 450 + }, + { + "epoch": 0.13383764911259818, + "flow_matching_loss": 0.035442753322422506, + "grad_norm": 0.23165753483772278, + "learning_rate": 9.18e-06, + "loss": 0.0477, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012425907049328088, + "rkd_loss_weight": 0.9848494946956634, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.40634443759918215, + "rkd_teacher_entropy": 3.7055842161178587, + "rkd_teacher_similarity_mean": 0.7707255184650421, + "rkd_vlm_temp": 0.10000000149011612, + "step": 460 + }, + { + "epoch": 0.13674716322374164, + "flow_matching_loss": 0.03637669421732426, + "grad_norm": 0.2221585065126419, + "learning_rate": 9.38e-06, + "loss": 0.0474, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011214674962684513, + "rkd_loss_weight": 0.9845161497592926, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3965310513973236, + "rkd_teacher_entropy": 3.6819796323776246, + "rkd_teacher_similarity_mean": 0.7622992157936096, + "rkd_vlm_temp": 0.10000000149011612, + "step": 470 + }, + { + "epoch": 0.13965667733488507, + "flow_matching_loss": 0.03793396912515164, + "grad_norm": 0.4952162802219391, + "learning_rate": 9.580000000000001e-06, + "loss": 0.0485, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01076717278920114, + "rkd_loss_weight": 0.9841828048229218, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.41278139054775237, + "rkd_teacher_entropy": 3.7378421306610106, + "rkd_teacher_similarity_mean": 0.7797256410121918, + "rkd_vlm_temp": 0.10000000149011612, + "step": 480 + }, + { + "epoch": 0.1425661914460285, + "flow_matching_loss": 0.03699308484792709, + "grad_norm": 0.2771536409854889, + "learning_rate": 9.780000000000001e-06, + "loss": 0.0484, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011629818007349968, + "rkd_loss_weight": 0.9838494658470154, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4067473769187927, + "rkd_teacher_entropy": 3.7141572713851927, + "rkd_teacher_similarity_mean": 0.7733731150627137, + "rkd_vlm_temp": 0.10000000149011612, + "step": 490 + }, + { + "epoch": 0.14547570555717196, + "flow_matching_loss": 0.034722713939845565, + "grad_norm": 0.19447974860668182, + "learning_rate": 9.980000000000001e-06, + "loss": 0.0441, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00957130566239357, + "rkd_loss_weight": 0.9835161149501801, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4091553121805191, + "rkd_teacher_entropy": 3.7150797128677366, + "rkd_teacher_similarity_mean": 0.7763316333293915, + "rkd_vlm_temp": 0.10000000149011612, + "step": 500 + }, + { + "epoch": 0.1483852196683154, + "flow_matching_loss": 0.033097234554588795, + "grad_norm": 0.24570804834365845, + "learning_rate": 1.0179999999999999e-05, + "loss": 0.0431, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01013768557459116, + "rkd_loss_weight": 0.9831827700138092, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4149689793586731, + "rkd_teacher_entropy": 3.73204824924469, + "rkd_teacher_similarity_mean": 0.7812315404415131, + "rkd_vlm_temp": 0.10000000149011612, + "step": 510 + }, + { + "epoch": 0.15129473377945882, + "flow_matching_loss": 0.03631176259368658, + "grad_norm": 0.2503458261489868, + "learning_rate": 1.0379999999999999e-05, + "loss": 0.046, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00989700579084456, + "rkd_loss_weight": 0.9828494310379028, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3948009073734283, + "rkd_teacher_entropy": 3.6638636112213137, + "rkd_teacher_similarity_mean": 0.759139621257782, + "rkd_vlm_temp": 0.10000000149011612, + "step": 520 + }, + { + "epoch": 0.15420424789060228, + "flow_matching_loss": 0.03491205722093582, + "grad_norm": 0.2675885558128357, + "learning_rate": 1.058e-05, + "loss": 0.0446, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009894777508452534, + "rkd_loss_weight": 0.9825160801410675, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4074894040822983, + "rkd_teacher_entropy": 3.7147634506225584, + "rkd_teacher_similarity_mean": 0.7746267855167389, + "rkd_vlm_temp": 0.10000000149011612, + "step": 530 + }, + { + "epoch": 0.1571137620017457, + "flow_matching_loss": 0.037563922442495824, + "grad_norm": 0.28288885951042175, + "learning_rate": 1.078e-05, + "loss": 0.0486, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01118617933243513, + "rkd_loss_weight": 0.9821827411651611, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4031944841146469, + "rkd_teacher_entropy": 3.7097357511520386, + "rkd_teacher_similarity_mean": 0.7700550317764282, + "rkd_vlm_temp": 0.10000000149011612, + "step": 540 + }, + { + "epoch": 0.16002327611288913, + "flow_matching_loss": 0.03027750626206398, + "grad_norm": 0.25570809841156006, + "learning_rate": 1.098e-05, + "loss": 0.0438, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013766143843531609, + "rkd_loss_weight": 0.9818493962287903, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4013044983148575, + "rkd_teacher_entropy": 3.702469515800476, + "rkd_teacher_similarity_mean": 0.7684018790721894, + "rkd_vlm_temp": 0.10000000149011612, + "step": 550 + }, + { + "epoch": 0.1629327902240326, + "flow_matching_loss": 0.03443137351423502, + "grad_norm": 0.2387249767780304, + "learning_rate": 1.118e-05, + "loss": 0.046, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01182059538550675, + "rkd_loss_weight": 0.9815160453319549, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.40481172800064086, + "rkd_teacher_entropy": 3.7210259437561035, + "rkd_teacher_similarity_mean": 0.7723895311355591, + "rkd_vlm_temp": 0.10000000149011612, + "step": 560 + }, + { + "epoch": 0.16584230433517602, + "flow_matching_loss": 0.033784153312444686, + "grad_norm": 0.29336708784103394, + "learning_rate": 1.1380000000000001e-05, + "loss": 0.0436, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010018761875107885, + "rkd_loss_weight": 0.9811827123165131, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39847474694252016, + "rkd_teacher_entropy": 3.687532639503479, + "rkd_teacher_similarity_mean": 0.7693351149559021, + "rkd_vlm_temp": 0.10000000149011612, + "step": 570 + }, + { + "epoch": 0.16875181844631945, + "flow_matching_loss": 0.0333418482914567, + "grad_norm": 0.3124074935913086, + "learning_rate": 1.1580000000000001e-05, + "loss": 0.0442, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011033209739252925, + "rkd_loss_weight": 0.9808493554592133, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4022810131311417, + "rkd_teacher_entropy": 3.691006803512573, + "rkd_teacher_similarity_mean": 0.771075439453125, + "rkd_vlm_temp": 0.10000000149011612, + "step": 580 + }, + { + "epoch": 0.1716613325574629, + "flow_matching_loss": 0.033775914646685126, + "grad_norm": 0.27104371786117554, + "learning_rate": 1.178e-05, + "loss": 0.045, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01139905429445207, + "rkd_loss_weight": 0.9805160164833069, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4049977332353592, + "rkd_teacher_entropy": 3.7233808994293214, + "rkd_teacher_similarity_mean": 0.7738442897796631, + "rkd_vlm_temp": 0.10000000149011612, + "step": 590 + }, + { + "epoch": 0.17457084666860634, + "flow_matching_loss": 0.0320875458419323, + "grad_norm": 0.24116182327270508, + "learning_rate": 1.198e-05, + "loss": 0.0417, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009850285807624459, + "rkd_loss_weight": 0.9801826775074005, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.40463078320026397, + "rkd_teacher_entropy": 3.708412790298462, + "rkd_teacher_similarity_mean": 0.7734229326248169, + "rkd_vlm_temp": 0.10000000149011612, + "step": 600 + }, + { + "epoch": 0.17748036077974977, + "flow_matching_loss": 0.030535879731178283, + "grad_norm": 0.3063070476055145, + "learning_rate": 1.2180000000000002e-05, + "loss": 0.0412, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010905801597982646, + "rkd_loss_weight": 0.9798493206501007, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4163065642118454, + "rkd_teacher_entropy": 3.739304518699646, + "rkd_teacher_similarity_mean": 0.7859081506729126, + "rkd_vlm_temp": 0.10000000149011612, + "step": 610 + }, + { + "epoch": 0.18038987489089323, + "flow_matching_loss": 0.03186215329915285, + "grad_norm": 0.20207229256629944, + "learning_rate": 1.238e-05, + "loss": 0.0423, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010648848814889788, + "rkd_loss_weight": 0.9795159876346589, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.41019182801246645, + "rkd_teacher_entropy": 3.7391005754470825, + "rkd_teacher_similarity_mean": 0.7803868651390076, + "rkd_vlm_temp": 0.10000000149011612, + "step": 620 + }, + { + "epoch": 0.18329938900203666, + "flow_matching_loss": 0.03326521534472704, + "grad_norm": 0.2564499080181122, + "learning_rate": 1.258e-05, + "loss": 0.0445, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011519324639812112, + "rkd_loss_weight": 0.9791826367378235, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4085140526294708, + "rkd_teacher_entropy": 3.728926491737366, + "rkd_teacher_similarity_mean": 0.7789148390293121, + "rkd_vlm_temp": 0.10000000149011612, + "step": 630 + }, + { + "epoch": 0.18620890311318009, + "flow_matching_loss": 0.03275956176221371, + "grad_norm": 0.3363300859928131, + "learning_rate": 1.278e-05, + "loss": 0.0441, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011606285581365228, + "rkd_loss_weight": 0.9788492918014526, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4062494307756424, + "rkd_teacher_entropy": 3.7155901908874513, + "rkd_teacher_similarity_mean": 0.7758889019489288, + "rkd_vlm_temp": 0.10000000149011612, + "step": 640 + }, + { + "epoch": 0.18911841722432354, + "flow_matching_loss": 0.03243814259767532, + "grad_norm": 0.25433385372161865, + "learning_rate": 1.298e-05, + "loss": 0.0435, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011312606651335955, + "rkd_loss_weight": 0.9785159528255463, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39958829879760743, + "rkd_teacher_entropy": 3.703135919570923, + "rkd_teacher_similarity_mean": 0.7719600021839141, + "rkd_vlm_temp": 0.10000000149011612, + "step": 650 + }, + { + "epoch": 0.19202793133546697, + "flow_matching_loss": 0.03201634231954813, + "grad_norm": 0.29359012842178345, + "learning_rate": 1.3180000000000001e-05, + "loss": 0.0422, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01040339395403862, + "rkd_loss_weight": 0.978182601928711, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39690669178962706, + "rkd_teacher_entropy": 3.6966707706451416, + "rkd_teacher_similarity_mean": 0.7706521868705749, + "rkd_vlm_temp": 0.10000000149011612, + "step": 660 + }, + { + "epoch": 0.1949374454466104, + "flow_matching_loss": 0.02970324121415615, + "grad_norm": 0.23565120995044708, + "learning_rate": 1.338e-05, + "loss": 0.0392, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009703474771231412, + "rkd_loss_weight": 0.9778492629528046, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4026197373867035, + "rkd_teacher_entropy": 3.7216519594192503, + "rkd_teacher_similarity_mean": 0.7770686686038971, + "rkd_vlm_temp": 0.10000000149011612, + "step": 670 + }, + { + "epoch": 0.19784695955775386, + "flow_matching_loss": 0.029494402557611467, + "grad_norm": 0.26101428270339966, + "learning_rate": 1.358e-05, + "loss": 0.0394, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010087478021159769, + "rkd_loss_weight": 0.9775159180164337, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4017327964305878, + "rkd_teacher_entropy": 3.714636540412903, + "rkd_teacher_similarity_mean": 0.7748876929283142, + "rkd_vlm_temp": 0.10000000149011612, + "step": 680 + }, + { + "epoch": 0.2007564736688973, + "flow_matching_loss": 0.026701010763645172, + "grad_norm": 0.1768275499343872, + "learning_rate": 1.378e-05, + "loss": 0.0361, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009662842471152544, + "rkd_loss_weight": 0.9771825730800628, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4010465979576111, + "rkd_teacher_entropy": 3.7015474796295167, + "rkd_teacher_similarity_mean": 0.7733925104141235, + "rkd_vlm_temp": 0.10000000149011612, + "step": 690 + }, + { + "epoch": 0.20366598778004075, + "flow_matching_loss": 0.028723240457475187, + "grad_norm": 0.3563152551651001, + "learning_rate": 1.3980000000000002e-05, + "loss": 0.0397, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011212308704853059, + "rkd_loss_weight": 0.976849228143692, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39616980850696565, + "rkd_teacher_entropy": 3.694403958320618, + "rkd_teacher_similarity_mean": 0.7691696047782898, + "rkd_vlm_temp": 0.10000000149011612, + "step": 700 + }, + { + "epoch": 0.20657550189118418, + "flow_matching_loss": 0.030869448743760585, + "grad_norm": 0.2516206204891205, + "learning_rate": 1.4180000000000001e-05, + "loss": 0.0411, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01048455098643899, + "rkd_loss_weight": 0.9765158891677856, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4060651123523712, + "rkd_teacher_entropy": 3.7305336475372313, + "rkd_teacher_similarity_mean": 0.7794168889522552, + "rkd_vlm_temp": 0.10000000149011612, + "step": 710 + }, + { + "epoch": 0.2094850160023276, + "flow_matching_loss": 0.027737780287861825, + "grad_norm": 0.21285080909729004, + "learning_rate": 1.438e-05, + "loss": 0.04, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012558154622092843, + "rkd_loss_weight": 0.9761825382709504, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4014017999172211, + "rkd_teacher_entropy": 3.7125699520111084, + "rkd_teacher_similarity_mean": 0.7728373587131501, + "rkd_vlm_temp": 0.10000000149011612, + "step": 720 + }, + { + "epoch": 0.21239453011347106, + "flow_matching_loss": 0.029376295767724513, + "grad_norm": 0.19624243676662445, + "learning_rate": 1.458e-05, + "loss": 0.0405, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011385175259783864, + "rkd_loss_weight": 0.9758491933345794, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4032266855239868, + "rkd_teacher_entropy": 3.715323877334595, + "rkd_teacher_similarity_mean": 0.7757520139217376, + "rkd_vlm_temp": 0.10000000149011612, + "step": 730 + }, + { + "epoch": 0.2153040442246145, + "flow_matching_loss": 0.03188369628041983, + "grad_norm": 0.23305751383304596, + "learning_rate": 1.4779999999999999e-05, + "loss": 0.0423, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010670844232663512, + "rkd_loss_weight": 0.975515854358673, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3988164275884628, + "rkd_teacher_entropy": 3.709433388710022, + "rkd_teacher_similarity_mean": 0.7720642149448395, + "rkd_vlm_temp": 0.10000000149011612, + "step": 740 + }, + { + "epoch": 0.21821355833575792, + "flow_matching_loss": 0.03091402966529131, + "grad_norm": 0.26065465807914734, + "learning_rate": 1.498e-05, + "loss": 0.0418, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011151475319638848, + "rkd_loss_weight": 0.9751825034618378, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.40101061165332796, + "rkd_teacher_entropy": 3.72193238735199, + "rkd_teacher_similarity_mean": 0.7760172665119172, + "rkd_vlm_temp": 0.10000000149011612, + "step": 750 + }, + { + "epoch": 0.22112307244690138, + "flow_matching_loss": 0.028676371648907663, + "grad_norm": 0.27041494846343994, + "learning_rate": 1.518e-05, + "loss": 0.0381, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009633925184607506, + "rkd_loss_weight": 0.9748491644859314, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.40062313377857206, + "rkd_teacher_entropy": 3.7190659046173096, + "rkd_teacher_similarity_mean": 0.7768916547298431, + "rkd_vlm_temp": 0.10000000149011612, + "step": 760 + }, + { + "epoch": 0.2240325865580448, + "flow_matching_loss": 0.03318574670702219, + "grad_norm": 0.2528550326824188, + "learning_rate": 1.5380000000000002e-05, + "loss": 0.0454, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012555446568876505, + "rkd_loss_weight": 0.9745158195495606, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3883216708898544, + "rkd_teacher_entropy": 3.678701329231262, + "rkd_teacher_similarity_mean": 0.7615839183330536, + "rkd_vlm_temp": 0.10000000149011612, + "step": 770 + }, + { + "epoch": 0.22694210066918824, + "flow_matching_loss": 0.028563840501010417, + "grad_norm": 0.18089556694030762, + "learning_rate": 1.558e-05, + "loss": 0.0392, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010910950414836407, + "rkd_loss_weight": 0.9741824686527252, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39545797407627103, + "rkd_teacher_entropy": 3.711810278892517, + "rkd_teacher_similarity_mean": 0.7725248992443084, + "rkd_vlm_temp": 0.10000000149011612, + "step": 780 + }, + { + "epoch": 0.2298516147803317, + "flow_matching_loss": 0.026522436551749705, + "grad_norm": 0.1771615892648697, + "learning_rate": 1.578e-05, + "loss": 0.0358, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009486928628757597, + "rkd_loss_weight": 0.9738491356372834, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39528937339782716, + "rkd_teacher_entropy": 3.717515707015991, + "rkd_teacher_similarity_mean": 0.7744565665721893, + "rkd_vlm_temp": 0.10000000149011612, + "step": 790 + }, + { + "epoch": 0.23276112889147513, + "flow_matching_loss": 0.028337953239679338, + "grad_norm": 0.19963528215885162, + "learning_rate": 1.598e-05, + "loss": 0.0391, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011037854943424463, + "rkd_loss_weight": 0.9735157787799835, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3911569446325302, + "rkd_teacher_entropy": 3.700474762916565, + "rkd_teacher_similarity_mean": 0.7697933316230774, + "rkd_vlm_temp": 0.10000000149011612, + "step": 800 + }, + { + "epoch": 0.23567064300261856, + "flow_matching_loss": 0.02478548139333725, + "grad_norm": 0.16958816349506378, + "learning_rate": 1.618e-05, + "loss": 0.0349, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010405555553734302, + "rkd_loss_weight": 0.9731824398040771, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3933457762002945, + "rkd_teacher_entropy": 3.7087790966033936, + "rkd_teacher_similarity_mean": 0.7728844463825226, + "rkd_vlm_temp": 0.10000000149011612, + "step": 810 + }, + { + "epoch": 0.23858015711376201, + "flow_matching_loss": 0.02692473102360964, + "grad_norm": 0.2848718464374542, + "learning_rate": 1.6380000000000002e-05, + "loss": 0.0393, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012678466690704226, + "rkd_loss_weight": 0.9728491008281708, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3859833747148514, + "rkd_teacher_entropy": 3.695397424697876, + "rkd_teacher_similarity_mean": 0.7662287890911103, + "rkd_vlm_temp": 0.10000000149011612, + "step": 820 + }, + { + "epoch": 0.24148967122490544, + "flow_matching_loss": 0.02671426422894001, + "grad_norm": 0.18642382323741913, + "learning_rate": 1.658e-05, + "loss": 0.0372, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01073972750455141, + "rkd_loss_weight": 0.9725157439708709, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3960035592317581, + "rkd_teacher_entropy": 3.7169003963470457, + "rkd_teacher_similarity_mean": 0.7762486040592194, + "rkd_vlm_temp": 0.10000000149011612, + "step": 830 + }, + { + "epoch": 0.24439918533604887, + "flow_matching_loss": 0.028486189618706702, + "grad_norm": 0.21947810053825378, + "learning_rate": 1.6780000000000002e-05, + "loss": 0.0399, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011711326707154513, + "rkd_loss_weight": 0.9721824109554291, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38824307918548584, + "rkd_teacher_entropy": 3.697144317626953, + "rkd_teacher_similarity_mean": 0.768361109495163, + "rkd_vlm_temp": 0.10000000149011612, + "step": 840 + }, + { + "epoch": 0.24730869944719233, + "flow_matching_loss": 0.029697749949991702, + "grad_norm": 0.22763903439044952, + "learning_rate": 1.698e-05, + "loss": 0.0421, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0127546482719481, + "rkd_loss_weight": 0.9718490600585937, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39245774745941164, + "rkd_teacher_entropy": 3.720650577545166, + "rkd_teacher_similarity_mean": 0.7733542084693908, + "rkd_vlm_temp": 0.10000000149011612, + "step": 850 + }, + { + "epoch": 0.25021821355833573, + "flow_matching_loss": 0.028221335262060165, + "grad_norm": 0.1971784383058548, + "learning_rate": 1.718e-05, + "loss": 0.0374, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009436330571770669, + "rkd_loss_weight": 0.9715157151222229, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.40081807076931, + "rkd_teacher_entropy": 3.745291519165039, + "rkd_teacher_similarity_mean": 0.7838566780090332, + "rkd_vlm_temp": 0.10000000149011612, + "step": 860 + }, + { + "epoch": 0.2531277276694792, + "flow_matching_loss": 0.02860519476234913, + "grad_norm": 0.2823043465614319, + "learning_rate": 1.7380000000000003e-05, + "loss": 0.0414, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013129871990531682, + "rkd_loss_weight": 0.9711823761463165, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39701139330863955, + "rkd_teacher_entropy": 3.7396554470062258, + "rkd_teacher_similarity_mean": 0.7806062221527099, + "rkd_vlm_temp": 0.10000000149011612, + "step": 870 + }, + { + "epoch": 0.25603724178062265, + "flow_matching_loss": 0.028918845765292644, + "grad_norm": 0.3066824972629547, + "learning_rate": 1.758e-05, + "loss": 0.0401, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011550641804933547, + "rkd_loss_weight": 0.9708490252494812, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39634667336940765, + "rkd_teacher_entropy": 3.7341068029403686, + "rkd_teacher_similarity_mean": 0.7787505209445953, + "rkd_vlm_temp": 0.10000000149011612, + "step": 880 + }, + { + "epoch": 0.2589467558917661, + "flow_matching_loss": 0.02654002495110035, + "grad_norm": 0.23073555529117584, + "learning_rate": 1.778e-05, + "loss": 0.0385, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012279006931930781, + "rkd_loss_weight": 0.9705156862735749, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3959808737039566, + "rkd_teacher_entropy": 3.7232213020324707, + "rkd_teacher_similarity_mean": 0.7799028694629669, + "rkd_vlm_temp": 0.10000000149011612, + "step": 890 + }, + { + "epoch": 0.2618562700029095, + "flow_matching_loss": 0.029312468133866787, + "grad_norm": 0.2294447273015976, + "learning_rate": 1.798e-05, + "loss": 0.0393, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010297268349677324, + "rkd_loss_weight": 0.9701823413372039, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3964372634887695, + "rkd_teacher_entropy": 3.7464475870132445, + "rkd_teacher_similarity_mean": 0.7832026243209839, + "rkd_vlm_temp": 0.10000000149011612, + "step": 900 + }, + { + "epoch": 0.26476578411405294, + "flow_matching_loss": 0.0274077158421278, + "grad_norm": 0.18329332768917084, + "learning_rate": 1.818e-05, + "loss": 0.039, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011996620334684849, + "rkd_loss_weight": 0.9698489904403687, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39246341288089753, + "rkd_teacher_entropy": 3.7354109287261963, + "rkd_teacher_similarity_mean": 0.7796624422073364, + "rkd_vlm_temp": 0.10000000149011612, + "step": 910 + }, + { + "epoch": 0.26767529822519637, + "flow_matching_loss": 0.02631158158183098, + "grad_norm": 0.23236750066280365, + "learning_rate": 1.838e-05, + "loss": 0.0376, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011618571635335685, + "rkd_loss_weight": 0.9695156514644623, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3921821802854538, + "rkd_teacher_entropy": 3.724597382545471, + "rkd_teacher_similarity_mean": 0.779946768283844, + "rkd_vlm_temp": 0.10000000149011612, + "step": 920 + }, + { + "epoch": 0.27058481233633985, + "flow_matching_loss": 0.02847759835422039, + "grad_norm": 0.20664706826210022, + "learning_rate": 1.8579999999999998e-05, + "loss": 0.0409, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012838017009198666, + "rkd_loss_weight": 0.9691823065280915, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39097768664360044, + "rkd_teacher_entropy": 3.729241132736206, + "rkd_teacher_similarity_mean": 0.7765923500061035, + "rkd_vlm_temp": 0.10000000149011612, + "step": 930 + }, + { + "epoch": 0.2734943264474833, + "flow_matching_loss": 0.02802292723208666, + "grad_norm": 0.24584190547466278, + "learning_rate": 1.878e-05, + "loss": 0.0426, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015008683502674102, + "rkd_loss_weight": 0.9688489615917206, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3866930454969406, + "rkd_teacher_entropy": 3.7266064643859864, + "rkd_teacher_similarity_mean": 0.7758705854415894, + "rkd_vlm_temp": 0.10000000149011612, + "step": 940 + }, + { + "epoch": 0.2764038405586267, + "flow_matching_loss": 0.026353060454130172, + "grad_norm": 0.23106597363948822, + "learning_rate": 1.898e-05, + "loss": 0.0373, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011301421700045467, + "rkd_loss_weight": 0.9685156166553497, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3922030210494995, + "rkd_teacher_entropy": 3.7233094692230226, + "rkd_teacher_similarity_mean": 0.7791154563426972, + "rkd_vlm_temp": 0.10000000149011612, + "step": 950 + }, + { + "epoch": 0.27931335466977014, + "flow_matching_loss": 0.026662572845816614, + "grad_norm": 0.20021259784698486, + "learning_rate": 1.918e-05, + "loss": 0.037, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01066466998308897, + "rkd_loss_weight": 0.9681822717189789, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3937453627586365, + "rkd_teacher_entropy": 3.7379182100296022, + "rkd_teacher_similarity_mean": 0.7833275616168975, + "rkd_vlm_temp": 0.10000000149011612, + "step": 960 + }, + { + "epoch": 0.28222286878091357, + "flow_matching_loss": 0.027807554230093956, + "grad_norm": 0.2051408886909485, + "learning_rate": 1.938e-05, + "loss": 0.0408, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01341025298461318, + "rkd_loss_weight": 0.967848926782608, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3835958868265152, + "rkd_teacher_entropy": 3.7100881576538085, + "rkd_teacher_similarity_mean": 0.7727314174175263, + "rkd_vlm_temp": 0.10000000149011612, + "step": 970 + }, + { + "epoch": 0.285132382892057, + "flow_matching_loss": 0.023342322558164597, + "grad_norm": 0.16643817722797394, + "learning_rate": 1.958e-05, + "loss": 0.0328, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009742916096001863, + "rkd_loss_weight": 0.9675155878067017, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3913049131631851, + "rkd_teacher_entropy": 3.7121846437454225, + "rkd_teacher_similarity_mean": 0.7785353243350983, + "rkd_vlm_temp": 0.10000000149011612, + "step": 980 + }, + { + "epoch": 0.2880418970032005, + "flow_matching_loss": 0.03023862522095442, + "grad_norm": 0.20613476634025574, + "learning_rate": 1.978e-05, + "loss": 0.0435, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01372643169015646, + "rkd_loss_weight": 0.9671822369098664, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37823342680931094, + "rkd_teacher_entropy": 3.68140332698822, + "rkd_teacher_similarity_mean": 0.7660344421863556, + "rkd_vlm_temp": 0.10000000149011612, + "step": 990 + }, + { + "epoch": 0.2909514111143439, + "flow_matching_loss": 0.025883745402097702, + "grad_norm": 0.19277477264404297, + "learning_rate": 1.9980000000000002e-05, + "loss": 0.0375, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012025629077106715, + "rkd_loss_weight": 0.9668488919734954, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3802745759487152, + "rkd_teacher_entropy": 3.6991686105728148, + "rkd_teacher_similarity_mean": 0.7709045469760895, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1000 + }, + { + "epoch": 0.29386092522548735, + "flow_matching_loss": 0.025592799484729766, + "grad_norm": 0.2524299919605255, + "learning_rate": 2.018e-05, + "loss": 0.0388, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01365748206153512, + "rkd_loss_weight": 0.9665155529975891, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3928110718727112, + "rkd_teacher_entropy": 3.7245322942733763, + "rkd_teacher_similarity_mean": 0.77865149974823, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1010 + }, + { + "epoch": 0.2967704393366308, + "flow_matching_loss": 0.02998650595545769, + "grad_norm": 0.22138845920562744, + "learning_rate": 2.038e-05, + "loss": 0.0414, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011853776406496763, + "rkd_loss_weight": 0.9661822021007538, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39619932770729066, + "rkd_teacher_entropy": 3.746971583366394, + "rkd_teacher_similarity_mean": 0.7841783046722413, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1020 + }, + { + "epoch": 0.2996799534477742, + "flow_matching_loss": 0.022943131253123285, + "grad_norm": 0.24928595125675201, + "learning_rate": 2.0580000000000003e-05, + "loss": 0.033, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010431933170184492, + "rkd_loss_weight": 0.9658488631248474, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.40589559376239776, + "rkd_teacher_entropy": 3.7649163484573362, + "rkd_teacher_similarity_mean": 0.793960440158844, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1030 + }, + { + "epoch": 0.30258946755891764, + "flow_matching_loss": 0.02469215765595436, + "grad_norm": 0.31996816396713257, + "learning_rate": 2.078e-05, + "loss": 0.0417, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01757351099513471, + "rkd_loss_weight": 0.9655155181884766, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3870132386684418, + "rkd_teacher_entropy": 3.7220019340515136, + "rkd_teacher_similarity_mean": 0.7750037133693695, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1040 + }, + { + "epoch": 0.3054989816700611, + "flow_matching_loss": 0.023904903046786784, + "grad_norm": 0.27869120240211487, + "learning_rate": 2.0980000000000002e-05, + "loss": 0.0342, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010648482665419578, + "rkd_loss_weight": 0.9651821672916412, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39225933253765105, + "rkd_teacher_entropy": 3.728520941734314, + "rkd_teacher_similarity_mean": 0.7784025967121124, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1050 + }, + { + "epoch": 0.30840849578120455, + "flow_matching_loss": 0.022477711737155914, + "grad_norm": 0.2526923418045044, + "learning_rate": 2.118e-05, + "loss": 0.034, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011958799976855516, + "rkd_loss_weight": 0.9648488342761994, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39747937917709353, + "rkd_teacher_entropy": 3.747969388961792, + "rkd_teacher_similarity_mean": 0.785209572315216, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1060 + }, + { + "epoch": 0.311318009892348, + "flow_matching_loss": 0.025620586052536965, + "grad_norm": 0.23914861679077148, + "learning_rate": 2.1380000000000002e-05, + "loss": 0.0364, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011172611359506845, + "rkd_loss_weight": 0.9645154774188995, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39380207657814026, + "rkd_teacher_entropy": 3.7388128995895387, + "rkd_teacher_similarity_mean": 0.7834330320358276, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1070 + }, + { + "epoch": 0.3142275240034914, + "flow_matching_loss": 0.02597023406997323, + "grad_norm": 0.2861296832561493, + "learning_rate": 2.1580000000000003e-05, + "loss": 0.039, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013560173939913512, + "rkd_loss_weight": 0.9641821384429932, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3887799888849258, + "rkd_teacher_entropy": 3.7202968120574953, + "rkd_teacher_similarity_mean": 0.776826697587967, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1080 + }, + { + "epoch": 0.31713703811463484, + "flow_matching_loss": 0.02387774670496583, + "grad_norm": 0.25812995433807373, + "learning_rate": 2.178e-05, + "loss": 0.0361, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01266304710879922, + "rkd_loss_weight": 0.9638487994670868, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38547921776771543, + "rkd_teacher_entropy": 3.706696319580078, + "rkd_teacher_similarity_mean": 0.7737352848052979, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1090 + }, + { + "epoch": 0.32004655222577827, + "flow_matching_loss": 0.02598208412528038, + "grad_norm": 0.2696840167045593, + "learning_rate": 2.198e-05, + "loss": 0.0375, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011941988859325648, + "rkd_loss_weight": 0.9635154426097869, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3836170732975006, + "rkd_teacher_entropy": 3.698672318458557, + "rkd_teacher_similarity_mean": 0.7715915679931641, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1100 + }, + { + "epoch": 0.32295606633692175, + "flow_matching_loss": 0.0243441641330719, + "grad_norm": 0.19193029403686523, + "learning_rate": 2.2179999999999998e-05, + "loss": 0.0376, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013770559895783662, + "rkd_loss_weight": 0.9631821095943451, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3894036650657654, + "rkd_teacher_entropy": 3.7211113929748536, + "rkd_teacher_similarity_mean": 0.775967812538147, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1110 + }, + { + "epoch": 0.3258655804480652, + "flow_matching_loss": 0.024248337745666503, + "grad_norm": 0.31007564067840576, + "learning_rate": 2.238e-05, + "loss": 0.0344, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010587883833795785, + "rkd_loss_weight": 0.9628487646579742, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3898514211177826, + "rkd_teacher_entropy": 3.7295132875442505, + "rkd_teacher_similarity_mean": 0.7779419422149658, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1120 + }, + { + "epoch": 0.3287750945592086, + "flow_matching_loss": 0.02704290710389614, + "grad_norm": 0.2127709835767746, + "learning_rate": 2.258e-05, + "loss": 0.0391, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012564101349562406, + "rkd_loss_weight": 0.9625154137611389, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39133801162242887, + "rkd_teacher_entropy": 3.7245944499969483, + "rkd_teacher_similarity_mean": 0.7792308509349823, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1130 + }, + { + "epoch": 0.33168460867035204, + "flow_matching_loss": 0.025922723487019537, + "grad_norm": 0.28710129857063293, + "learning_rate": 2.278e-05, + "loss": 0.0396, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014223250746726989, + "rkd_loss_weight": 0.9621820747852325, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37915074825286865, + "rkd_teacher_entropy": 3.702385354042053, + "rkd_teacher_similarity_mean": 0.7678449392318726, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1140 + }, + { + "epoch": 0.3345941227814955, + "flow_matching_loss": 0.022808071970939637, + "grad_norm": 0.1876709908246994, + "learning_rate": 2.298e-05, + "loss": 0.0338, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011436624825000763, + "rkd_loss_weight": 0.9618487298488617, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3780492216348648, + "rkd_teacher_entropy": 3.6956515073776246, + "rkd_teacher_similarity_mean": 0.769874382019043, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1150 + }, + { + "epoch": 0.3375036368926389, + "flow_matching_loss": 0.02521066563203931, + "grad_norm": 0.22150033712387085, + "learning_rate": 2.3179999999999998e-05, + "loss": 0.0386, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013923092093318701, + "rkd_loss_weight": 0.9615153849124909, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38893991708755493, + "rkd_teacher_entropy": 3.72441508769989, + "rkd_teacher_similarity_mean": 0.7768840253353119, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1160 + }, + { + "epoch": 0.3404131510037824, + "flow_matching_loss": 0.027532869204878806, + "grad_norm": 0.28858888149261475, + "learning_rate": 2.338e-05, + "loss": 0.041, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014059560559689998, + "rkd_loss_weight": 0.96118203997612, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3856250554323196, + "rkd_teacher_entropy": 3.712056303024292, + "rkd_teacher_similarity_mean": 0.7733197808265686, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1170 + }, + { + "epoch": 0.3433226651149258, + "flow_matching_loss": 0.02708932217210531, + "grad_norm": 0.21611164510250092, + "learning_rate": 2.358e-05, + "loss": 0.0449, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.018514181673526763, + "rkd_loss_weight": 0.9608486950397491, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3794753313064575, + "rkd_teacher_entropy": 3.7099194049835207, + "rkd_teacher_similarity_mean": 0.7702456414699554, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1180 + }, + { + "epoch": 0.34623217922606925, + "flow_matching_loss": 0.02302883639931679, + "grad_norm": 0.29488828778266907, + "learning_rate": 2.378e-05, + "loss": 0.0369, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01448048185557127, + "rkd_loss_weight": 0.9605153501033783, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38712741434574127, + "rkd_teacher_entropy": 3.732503294944763, + "rkd_teacher_similarity_mean": 0.7805300414562225, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1190 + }, + { + "epoch": 0.3491416933372127, + "flow_matching_loss": 0.02747269757091999, + "grad_norm": 0.24309180676937103, + "learning_rate": 2.398e-05, + "loss": 0.0442, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017430036328732967, + "rkd_loss_weight": 0.9601820111274719, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38402228951454165, + "rkd_teacher_entropy": 3.717808485031128, + "rkd_teacher_similarity_mean": 0.7746878743171692, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1200 + }, + { + "epoch": 0.3520512074483561, + "flow_matching_loss": 0.023619108460843564, + "grad_norm": 0.28195035457611084, + "learning_rate": 2.4180000000000002e-05, + "loss": 0.0358, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012661275360733271, + "rkd_loss_weight": 0.9598486602306366, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3808061063289642, + "rkd_teacher_entropy": 3.6989772796630858, + "rkd_teacher_similarity_mean": 0.7716823995113373, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1210 + }, + { + "epoch": 0.35496072155949954, + "flow_matching_loss": 0.025398100912570953, + "grad_norm": 0.27987340092658997, + "learning_rate": 2.438e-05, + "loss": 0.0389, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01412323284894228, + "rkd_loss_weight": 0.9595153152942657, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38053129613399506, + "rkd_teacher_entropy": 3.700506639480591, + "rkd_teacher_similarity_mean": 0.7715128004550934, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1220 + }, + { + "epoch": 0.357870235670643, + "flow_matching_loss": 0.025806737877428532, + "grad_norm": 0.2388022094964981, + "learning_rate": 2.4580000000000002e-05, + "loss": 0.0394, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014154891856014729, + "rkd_loss_weight": 0.9591819763183593, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.380745667219162, + "rkd_teacher_entropy": 3.7011789083480835, + "rkd_teacher_similarity_mean": 0.7688676834106445, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1230 + }, + { + "epoch": 0.36077974978178645, + "flow_matching_loss": 0.023788285814225672, + "grad_norm": 0.2440781593322754, + "learning_rate": 2.478e-05, + "loss": 0.0364, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013187664281576871, + "rkd_loss_weight": 0.958848625421524, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.385807591676712, + "rkd_teacher_entropy": 3.7105973958969116, + "rkd_teacher_similarity_mean": 0.7743953168392181, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1240 + }, + { + "epoch": 0.3636892638929299, + "flow_matching_loss": 0.02400032114237547, + "grad_norm": 0.24247409403324127, + "learning_rate": 2.498e-05, + "loss": 0.0391, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015718870516866445, + "rkd_loss_weight": 0.9585152864456177, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3913044512271881, + "rkd_teacher_entropy": 3.7225290536880493, + "rkd_teacher_similarity_mean": 0.7787424981594085, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1250 + }, + { + "epoch": 0.3665987780040733, + "flow_matching_loss": 0.024665779247879983, + "grad_norm": 0.1990078240633011, + "learning_rate": 2.5180000000000003e-05, + "loss": 0.0381, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014015414519235492, + "rkd_loss_weight": 0.9581819415092468, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38821350038051605, + "rkd_teacher_entropy": 3.7249312162399293, + "rkd_teacher_similarity_mean": 0.7791846096515656, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1260 + }, + { + "epoch": 0.36950829211521674, + "flow_matching_loss": 0.023447480984032154, + "grad_norm": 0.31095314025878906, + "learning_rate": 2.538e-05, + "loss": 0.0357, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012752013746649027, + "rkd_loss_weight": 0.9578485906124115, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3852444291114807, + "rkd_teacher_entropy": 3.720211052894592, + "rkd_teacher_similarity_mean": 0.7747839570045472, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1270 + }, + { + "epoch": 0.37241780622636017, + "flow_matching_loss": 0.02403263244777918, + "grad_norm": 0.15724784135818481, + "learning_rate": 2.5580000000000002e-05, + "loss": 0.0359, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012365718418732286, + "rkd_loss_weight": 0.9575152575969696, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39747401475906374, + "rkd_teacher_entropy": 3.7507396221160887, + "rkd_teacher_similarity_mean": 0.7867114901542663, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1280 + }, + { + "epoch": 0.37532732033750366, + "flow_matching_loss": 0.02445501871407032, + "grad_norm": 0.18285474181175232, + "learning_rate": 2.578e-05, + "loss": 0.0378, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01389182722195983, + "rkd_loss_weight": 0.9571819007396698, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3852704882621765, + "rkd_teacher_entropy": 3.7056650161743163, + "rkd_teacher_similarity_mean": 0.7741279065608978, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1290 + }, + { + "epoch": 0.3782368344486471, + "flow_matching_loss": 0.02410435229539871, + "grad_norm": 0.2828579246997833, + "learning_rate": 2.5980000000000002e-05, + "loss": 0.0397, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01634847531095147, + "rkd_loss_weight": 0.9568485617637634, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3875483751296997, + "rkd_teacher_entropy": 3.7275431871414186, + "rkd_teacher_similarity_mean": 0.7777420699596405, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1300 + }, + { + "epoch": 0.3811463485597905, + "flow_matching_loss": 0.021692500170320272, + "grad_norm": 0.2464597523212433, + "learning_rate": 2.618e-05, + "loss": 0.0339, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012758578360080718, + "rkd_loss_weight": 0.9565152227878571, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3867333471775055, + "rkd_teacher_entropy": 3.7293481826782227, + "rkd_teacher_similarity_mean": 0.7787551760673523, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1310 + }, + { + "epoch": 0.38405586267093395, + "flow_matching_loss": 0.02018053513020277, + "grad_norm": 0.19480767846107483, + "learning_rate": 2.6379999999999998e-05, + "loss": 0.0342, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014619212411344052, + "rkd_loss_weight": 0.9561818659305572, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4000688254833221, + "rkd_teacher_entropy": 3.763897967338562, + "rkd_teacher_similarity_mean": 0.7905187845230103, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1320 + }, + { + "epoch": 0.3869653767820774, + "flow_matching_loss": 0.028174177184700964, + "grad_norm": 0.24475997686386108, + "learning_rate": 2.658e-05, + "loss": 0.0423, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014807805884629488, + "rkd_loss_weight": 0.9558485329151154, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3873999029397964, + "rkd_teacher_entropy": 3.7199406147003176, + "rkd_teacher_similarity_mean": 0.7746494531631469, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1330 + }, + { + "epoch": 0.3898748908932208, + "flow_matching_loss": 0.024973647296428682, + "grad_norm": 0.22520962357521057, + "learning_rate": 2.678e-05, + "loss": 0.0426, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.018467433378100397, + "rkd_loss_weight": 0.95551518201828, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3971333235502243, + "rkd_teacher_entropy": 3.7615918636322023, + "rkd_teacher_similarity_mean": 0.7881286084651947, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1340 + }, + { + "epoch": 0.3927844050043643, + "flow_matching_loss": 0.024894135631620885, + "grad_norm": 0.20424515008926392, + "learning_rate": 2.698e-05, + "loss": 0.037, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012706672959029674, + "rkd_loss_weight": 0.9551818370819092, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3830201327800751, + "rkd_teacher_entropy": 3.724007272720337, + "rkd_teacher_similarity_mean": 0.775218379497528, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1350 + }, + { + "epoch": 0.3956939191155077, + "flow_matching_loss": 0.024654229171574116, + "grad_norm": 0.24666552245616913, + "learning_rate": 2.718e-05, + "loss": 0.0362, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012107907980680465, + "rkd_loss_weight": 0.9548484981060028, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.387481489777565, + "rkd_teacher_entropy": 3.7341129779815674, + "rkd_teacher_similarity_mean": 0.7790236353874207, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1360 + }, + { + "epoch": 0.39860343322665115, + "flow_matching_loss": 0.025044144690036775, + "grad_norm": 0.21934907138347626, + "learning_rate": 2.738e-05, + "loss": 0.0378, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01340948399156332, + "rkd_loss_weight": 0.9545151472091675, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38593626022338867, + "rkd_teacher_entropy": 3.7284576654434205, + "rkd_teacher_similarity_mean": 0.7761586487293244, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1370 + }, + { + "epoch": 0.4015129473377946, + "flow_matching_loss": 0.021454064920544625, + "grad_norm": 0.2402636706829071, + "learning_rate": 2.758e-05, + "loss": 0.0334, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01247716061770916, + "rkd_loss_weight": 0.9541818082332612, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3805089950561523, + "rkd_teacher_entropy": 3.698855924606323, + "rkd_teacher_similarity_mean": 0.7744203686714173, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1380 + }, + { + "epoch": 0.404422461448938, + "flow_matching_loss": 0.023247806169092656, + "grad_norm": 0.19417425990104675, + "learning_rate": 2.778e-05, + "loss": 0.0374, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014886298682540655, + "rkd_loss_weight": 0.9538484632968902, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38834960758686066, + "rkd_teacher_entropy": 3.736691451072693, + "rkd_teacher_similarity_mean": 0.7798957824707031, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1390 + }, + { + "epoch": 0.4073319755600815, + "flow_matching_loss": 0.022371009271591902, + "grad_norm": 0.2097790539264679, + "learning_rate": 2.798e-05, + "loss": 0.0346, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012789490725845098, + "rkd_loss_weight": 0.9535151124000549, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3812498182058334, + "rkd_teacher_entropy": 3.6992570400238036, + "rkd_teacher_similarity_mean": 0.7739984452724457, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1400 + }, + { + "epoch": 0.4102414896712249, + "flow_matching_loss": 0.023044230602681636, + "grad_norm": 0.23980161547660828, + "learning_rate": 2.818e-05, + "loss": 0.0373, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014928948599845171, + "rkd_loss_weight": 0.9531817734241486, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3866866439580917, + "rkd_teacher_entropy": 3.7326053857803343, + "rkd_teacher_similarity_mean": 0.7801648855209351, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1410 + }, + { + "epoch": 0.41315100378236835, + "flow_matching_loss": 0.024837481416761877, + "grad_norm": 0.2179541289806366, + "learning_rate": 2.838e-05, + "loss": 0.0404, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016327724140137435, + "rkd_loss_weight": 0.9528484284877777, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3826770097017288, + "rkd_teacher_entropy": 3.72520854473114, + "rkd_teacher_similarity_mean": 0.7759030640125275, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1420 + }, + { + "epoch": 0.4160605178935118, + "flow_matching_loss": 0.022835460491478443, + "grad_norm": 0.1689184606075287, + "learning_rate": 2.858e-05, + "loss": 0.0365, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014345226809382439, + "rkd_loss_weight": 0.9525150835514069, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3853329658508301, + "rkd_teacher_entropy": 3.7215634107589723, + "rkd_teacher_similarity_mean": 0.7775545597076416, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1430 + }, + { + "epoch": 0.4189700320046552, + "flow_matching_loss": 0.021487159468233586, + "grad_norm": 0.21842902898788452, + "learning_rate": 2.8780000000000002e-05, + "loss": 0.034, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013156270142644644, + "rkd_loss_weight": 0.952181738615036, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3805884957313538, + "rkd_teacher_entropy": 3.711973237991333, + "rkd_teacher_similarity_mean": 0.7762383818626404, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1440 + }, + { + "epoch": 0.42187954611579864, + "flow_matching_loss": 0.022558098286390306, + "grad_norm": 0.20280282199382782, + "learning_rate": 2.898e-05, + "loss": 0.0384, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016614855360239746, + "rkd_loss_weight": 0.9518483936786651, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38522306382656096, + "rkd_teacher_entropy": 3.7202538013458253, + "rkd_teacher_similarity_mean": 0.7794477999210357, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1450 + }, + { + "epoch": 0.42478906022694213, + "flow_matching_loss": 0.0277953939512372, + "grad_norm": 0.2065363973379135, + "learning_rate": 2.9180000000000002e-05, + "loss": 0.0435, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016494630370289087, + "rkd_loss_weight": 0.9515150487422943, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38178289532661436, + "rkd_teacher_entropy": 3.71258430480957, + "rkd_teacher_similarity_mean": 0.7716500997543335, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1460 + }, + { + "epoch": 0.42769857433808556, + "flow_matching_loss": 0.0223079776391387, + "grad_norm": 0.15341266989707947, + "learning_rate": 2.938e-05, + "loss": 0.0367, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015152645204216242, + "rkd_loss_weight": 0.9511817097663879, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38432141542434695, + "rkd_teacher_entropy": 3.7077661037445067, + "rkd_teacher_similarity_mean": 0.7745723366737366, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1470 + }, + { + "epoch": 0.430608088449229, + "flow_matching_loss": 0.023663434386253356, + "grad_norm": 0.24755236506462097, + "learning_rate": 2.958e-05, + "loss": 0.0371, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014153410121798516, + "rkd_loss_weight": 0.9508483588695527, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3881279528141022, + "rkd_teacher_entropy": 3.719397783279419, + "rkd_teacher_similarity_mean": 0.7795057117938995, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1480 + }, + { + "epoch": 0.4335176025603724, + "flow_matching_loss": 0.023816115409135818, + "grad_norm": 0.31216514110565186, + "learning_rate": 2.9780000000000003e-05, + "loss": 0.038, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014880159124732017, + "rkd_loss_weight": 0.9505150139331817, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38174316585063933, + "rkd_teacher_entropy": 3.714092659950256, + "rkd_teacher_similarity_mean": 0.7726036012172699, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1490 + }, + { + "epoch": 0.43642711667151585, + "flow_matching_loss": 0.023210758157074452, + "grad_norm": 0.34984350204467773, + "learning_rate": 2.998e-05, + "loss": 0.038, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015565278567373752, + "rkd_loss_weight": 0.9501816749572753, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38260034620761874, + "rkd_teacher_entropy": 3.7210936546325684, + "rkd_teacher_similarity_mean": 0.7746342480182647, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1500 + }, + { + "epoch": 0.4393366307826593, + "flow_matching_loss": 0.024646736122667788, + "grad_norm": 0.20392094552516937, + "learning_rate": 2.9999992618302022e-05, + "loss": 0.0412, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01746482402086258, + "rkd_loss_weight": 0.9498483240604401, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37469092607498167, + "rkd_teacher_entropy": 3.6892024278640747, + "rkd_teacher_similarity_mean": 0.7655722916126251, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1510 + }, + { + "epoch": 0.44224614489380276, + "flow_matching_loss": 0.02098074769601226, + "grad_norm": 0.29543402791023254, + "learning_rate": 2.999996710133069e-05, + "loss": 0.0347, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01445854278281331, + "rkd_loss_weight": 0.9495149850845337, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37954085469245913, + "rkd_teacher_entropy": 3.7054129600524903, + "rkd_teacher_similarity_mean": 0.7694184720516205, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1520 + }, + { + "epoch": 0.4451556590049462, + "flow_matching_loss": 0.02327049374580383, + "grad_norm": 0.19421762228012085, + "learning_rate": 2.9999923357984923e-05, + "loss": 0.0365, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013932699337601662, + "rkd_loss_weight": 0.9491816401481629, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3870362192392349, + "rkd_teacher_entropy": 3.725771594047546, + "rkd_teacher_similarity_mean": 0.7789050698280334, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1530 + }, + { + "epoch": 0.4480651731160896, + "flow_matching_loss": 0.02474696412682533, + "grad_norm": 0.3505500853061676, + "learning_rate": 2.9999861388317874e-05, + "loss": 0.039, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01503928741440177, + "rkd_loss_weight": 0.9488482892513275, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37294261157512665, + "rkd_teacher_entropy": 3.6893558263778687, + "rkd_teacher_similarity_mean": 0.7666844189167022, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1540 + }, + { + "epoch": 0.45097468722723305, + "flow_matching_loss": 0.024324307218194008, + "grad_norm": 0.23693867027759552, + "learning_rate": 2.9999781192404847e-05, + "loss": 0.0406, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01711007235571742, + "rkd_loss_weight": 0.9485149562358857, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.386845862865448, + "rkd_teacher_entropy": 3.738472056388855, + "rkd_teacher_similarity_mean": 0.7791788399219512, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1550 + }, + { + "epoch": 0.4538842013383765, + "flow_matching_loss": 0.02545690257102251, + "grad_norm": 0.2052583396434784, + "learning_rate": 2.9999682770343283e-05, + "loss": 0.0398, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01511084483936429, + "rkd_loss_weight": 0.9481816053390503, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3840500831604004, + "rkd_teacher_entropy": 3.718096709251404, + "rkd_teacher_similarity_mean": 0.7730439960956573, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1560 + }, + { + "epoch": 0.4567937154495199, + "flow_matching_loss": 0.025049092434346675, + "grad_norm": 0.22012750804424286, + "learning_rate": 2.9999566122252777e-05, + "loss": 0.0397, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015405100304633379, + "rkd_loss_weight": 0.9478482604026794, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38662797808647154, + "rkd_teacher_entropy": 3.7212651014328, + "rkd_teacher_similarity_mean": 0.7779876410961151, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1570 + }, + { + "epoch": 0.4597032295606634, + "flow_matching_loss": 0.023769733775407077, + "grad_norm": 0.24316056072711945, + "learning_rate": 2.9999431248275065e-05, + "loss": 0.0396, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01674623480066657, + "rkd_loss_weight": 0.9475149214267731, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37340920567512514, + "rkd_teacher_entropy": 3.703950071334839, + "rkd_teacher_similarity_mean": 0.7665853083133698, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1580 + }, + { + "epoch": 0.4626127436718068, + "flow_matching_loss": 0.024648249708116056, + "grad_norm": 0.21809929609298706, + "learning_rate": 2.9999278148574027e-05, + "loss": 0.0464, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.02296052286401391, + "rkd_loss_weight": 0.9471815705299378, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37695316672325135, + "rkd_teacher_entropy": 3.706635570526123, + "rkd_teacher_similarity_mean": 0.7676470279693604, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1590 + }, + { + "epoch": 0.46552225778295026, + "flow_matching_loss": 0.025078984536230566, + "grad_norm": 0.2355981171131134, + "learning_rate": 2.9999106823335705e-05, + "loss": 0.0402, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0159309814684093, + "rkd_loss_weight": 0.9468482315540314, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3870518296957016, + "rkd_teacher_entropy": 3.723440480232239, + "rkd_teacher_similarity_mean": 0.7773932635784149, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1600 + }, + { + "epoch": 0.4684317718940937, + "flow_matching_loss": 0.024373055994510652, + "grad_norm": 0.24106013774871826, + "learning_rate": 2.999891727276827e-05, + "loss": 0.0392, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015631617140024902, + "rkd_loss_weight": 0.9465148866176605, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38500000536441803, + "rkd_teacher_entropy": 3.7453314542770384, + "rkd_teacher_similarity_mean": 0.7789949238300323, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1610 + }, + { + "epoch": 0.4713412860052371, + "flow_matching_loss": 0.02277332078665495, + "grad_norm": 0.3045925199985504, + "learning_rate": 2.9998709497102037e-05, + "loss": 0.0406, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.018858731165528296, + "rkd_loss_weight": 0.9461815357208252, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3813613533973694, + "rkd_teacher_entropy": 3.715682625770569, + "rkd_teacher_similarity_mean": 0.7740076899528503, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1620 + }, + { + "epoch": 0.47425080011638054, + "flow_matching_loss": 0.02563936822116375, + "grad_norm": 0.20541150867938995, + "learning_rate": 2.9998483496589485e-05, + "loss": 0.0409, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016121948417276145, + "rkd_loss_weight": 0.9458481967449188, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.383531790971756, + "rkd_teacher_entropy": 3.731403660774231, + "rkd_teacher_similarity_mean": 0.7781675696372986, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1630 + }, + { + "epoch": 0.47716031422752403, + "flow_matching_loss": 0.0217175904661417, + "grad_norm": 0.2621610164642334, + "learning_rate": 2.9998239271505217e-05, + "loss": 0.0337, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012689131684601307, + "rkd_loss_weight": 0.945514851808548, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3827574193477631, + "rkd_teacher_entropy": 3.736353611946106, + "rkd_teacher_similarity_mean": 0.7802532851696015, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1640 + }, + { + "epoch": 0.48006982833866746, + "flow_matching_loss": 0.023066156916320323, + "grad_norm": 0.2741584777832031, + "learning_rate": 2.9997976822145995e-05, + "loss": 0.0396, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017463109083473684, + "rkd_loss_weight": 0.9451815068721772, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37959596514701843, + "rkd_teacher_entropy": 3.73248872756958, + "rkd_teacher_similarity_mean": 0.775779390335083, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1650 + }, + { + "epoch": 0.4829793424498109, + "flow_matching_loss": 0.0205201999284327, + "grad_norm": 0.22289462387561798, + "learning_rate": 2.9997696148830714e-05, + "loss": 0.0359, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016266565956175327, + "rkd_loss_weight": 0.9448481619358062, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3774845778942108, + "rkd_teacher_entropy": 3.7196330785751344, + "rkd_teacher_similarity_mean": 0.7745114386081695, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1660 + }, + { + "epoch": 0.4858888565609543, + "flow_matching_loss": 0.023907561134546996, + "grad_norm": 0.2585909962654114, + "learning_rate": 2.9997397251900423e-05, + "loss": 0.0396, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016601407248526812, + "rkd_loss_weight": 0.9445148169994354, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3758931905031204, + "rkd_teacher_entropy": 3.711072492599487, + "rkd_teacher_similarity_mean": 0.774362200498581, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1670 + }, + { + "epoch": 0.48879837067209775, + "flow_matching_loss": 0.023397716227918863, + "grad_norm": 0.2716364562511444, + "learning_rate": 2.999708013171831e-05, + "loss": 0.038, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015477201901376247, + "rkd_loss_weight": 0.9441814720630646, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3861248940229416, + "rkd_teacher_entropy": 3.732848024368286, + "rkd_teacher_similarity_mean": 0.7817353308200836, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1680 + }, + { + "epoch": 0.4917078847832412, + "flow_matching_loss": 0.026524961553514005, + "grad_norm": 0.19247807562351227, + "learning_rate": 2.9996744788669705e-05, + "loss": 0.0466, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0213051057420671, + "rkd_loss_weight": 0.9438481330871582, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37206536829471587, + "rkd_teacher_entropy": 3.7004411935806276, + "rkd_teacher_similarity_mean": 0.7682225465774536, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1690 + }, + { + "epoch": 0.49461739889438466, + "flow_matching_loss": 0.026639231108129026, + "grad_norm": 0.20530636608600616, + "learning_rate": 2.999639122316208e-05, + "loss": 0.0403, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014456177782267331, + "rkd_loss_weight": 0.9435147821903229, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3753638505935669, + "rkd_teacher_entropy": 3.698908543586731, + "rkd_teacher_similarity_mean": 0.7715628564357757, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1700 + }, + { + "epoch": 0.4975269130055281, + "flow_matching_loss": 0.023296541534364223, + "grad_norm": 0.41209131479263306, + "learning_rate": 2.999601943562505e-05, + "loss": 0.0397, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01743079675361514, + "rkd_loss_weight": 0.943181437253952, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38070963621139525, + "rkd_teacher_entropy": 3.708251619338989, + "rkd_teacher_similarity_mean": 0.7756233036518096, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1710 + }, + { + "epoch": 0.5004364271166715, + "flow_matching_loss": 0.022754223458468913, + "grad_norm": 0.2337818443775177, + "learning_rate": 2.9995629426510377e-05, + "loss": 0.0394, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017662063147872685, + "rkd_loss_weight": 0.9428480982780456, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38116995990276337, + "rkd_teacher_entropy": 3.7146326541900634, + "rkd_teacher_similarity_mean": 0.7745146512985229, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1720 + }, + { + "epoch": 0.503345941227815, + "flow_matching_loss": 0.022803100012242793, + "grad_norm": 0.19463299214839935, + "learning_rate": 2.9995221196291952e-05, + "loss": 0.0378, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015896375849843027, + "rkd_loss_weight": 0.9425147473812103, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37621262967586516, + "rkd_teacher_entropy": 3.7119781017303466, + "rkd_teacher_similarity_mean": 0.7718587279319763, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1730 + }, + { + "epoch": 0.5062554553389584, + "flow_matching_loss": 0.02304721288383007, + "grad_norm": 0.4143763482570648, + "learning_rate": 2.999479474546582e-05, + "loss": 0.041, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01905381614342332, + "rkd_loss_weight": 0.942181408405304, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38123898208141327, + "rkd_teacher_entropy": 3.714391994476318, + "rkd_teacher_similarity_mean": 0.7732662856578827, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1740 + }, + { + "epoch": 0.5091649694501018, + "flow_matching_loss": 0.025273722410202027, + "grad_norm": 0.20379798114299774, + "learning_rate": 2.9994350074550153e-05, + "loss": 0.0403, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01592886755242944, + "rkd_loss_weight": 0.9418480634689331, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3755435973405838, + "rkd_teacher_entropy": 3.707622194290161, + "rkd_teacher_similarity_mean": 0.7693977296352387, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1750 + }, + { + "epoch": 0.5120744835612453, + "flow_matching_loss": 0.0226708373054862, + "grad_norm": 0.27924883365631104, + "learning_rate": 2.9993887184085273e-05, + "loss": 0.0372, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015403318870812654, + "rkd_loss_weight": 0.9415147125720977, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3899535179138184, + "rkd_teacher_entropy": 3.7421849012374877, + "rkd_teacher_similarity_mean": 0.7851236641407013, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1760 + }, + { + "epoch": 0.5149839976723887, + "flow_matching_loss": 0.025137345120310784, + "grad_norm": 0.30311718583106995, + "learning_rate": 2.9993406074633628e-05, + "loss": 0.0422, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.018088831193745138, + "rkd_loss_weight": 0.9411813795566559, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37615360617637633, + "rkd_teacher_entropy": 3.696899580955505, + "rkd_teacher_similarity_mean": 0.7706542491912842, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1770 + }, + { + "epoch": 0.5178935117835322, + "flow_matching_loss": 0.02260178532451391, + "grad_norm": 0.265865683555603, + "learning_rate": 2.9992906746779824e-05, + "loss": 0.0368, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015099669340997934, + "rkd_loss_weight": 0.9408480226993561, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37363339960575104, + "rkd_teacher_entropy": 3.6964890480041506, + "rkd_teacher_similarity_mean": 0.7701687335968017, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1780 + }, + { + "epoch": 0.5208030258946756, + "flow_matching_loss": 0.02345808958634734, + "grad_norm": 0.2114669233560562, + "learning_rate": 2.9992389201130576e-05, + "loss": 0.0357, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012986898422241211, + "rkd_loss_weight": 0.9405146837234497, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38214167654514314, + "rkd_teacher_entropy": 3.7187749624252318, + "rkd_teacher_similarity_mean": 0.7791468739509583, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1790 + }, + { + "epoch": 0.523712540005819, + "flow_matching_loss": 0.02487478144466877, + "grad_norm": 0.35661596059799194, + "learning_rate": 2.999185343831476e-05, + "loss": 0.0453, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.02174576325342059, + "rkd_loss_weight": 0.9401813447475433, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3804393708705902, + "rkd_teacher_entropy": 3.7221697330474854, + "rkd_teacher_similarity_mean": 0.7743526875972748, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1800 + }, + { + "epoch": 0.5266220541169625, + "flow_matching_loss": 0.02117874063551426, + "grad_norm": 0.3089236915111542, + "learning_rate": 2.999129945898338e-05, + "loss": 0.0369, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016713081300258635, + "rkd_loss_weight": 0.9398479878902435, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38409359455108644, + "rkd_teacher_entropy": 3.741469645500183, + "rkd_teacher_similarity_mean": 0.7812611103057862, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1810 + }, + { + "epoch": 0.5295315682281059, + "flow_matching_loss": 0.023422961123287678, + "grad_norm": 0.21379360556602478, + "learning_rate": 2.999072726380957e-05, + "loss": 0.0403, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0179595704190433, + "rkd_loss_weight": 0.9395146548748017, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37269283533096315, + "rkd_teacher_entropy": 3.7056435108184815, + "rkd_teacher_similarity_mean": 0.7693736016750335, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1820 + }, + { + "epoch": 0.5324410823392494, + "flow_matching_loss": 0.020683289505541325, + "grad_norm": 0.19203785061836243, + "learning_rate": 2.9990136853488603e-05, + "loss": 0.0364, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01675354614853859, + "rkd_loss_weight": 0.9391813039779663, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38186830580234526, + "rkd_teacher_entropy": 3.7126380443572997, + "rkd_teacher_similarity_mean": 0.7750930309295654, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1830 + }, + { + "epoch": 0.5353505964503927, + "flow_matching_loss": 0.02532524149864912, + "grad_norm": 0.4707542657852173, + "learning_rate": 2.9989528228737878e-05, + "loss": 0.0423, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01808991711586714, + "rkd_loss_weight": 0.9388479590415955, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.385396009683609, + "rkd_teacher_entropy": 3.7307719707489015, + "rkd_teacher_similarity_mean": 0.7803231954574585, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1840 + }, + { + "epoch": 0.5382601105615362, + "flow_matching_loss": 0.023522717971354724, + "grad_norm": 0.2313697338104248, + "learning_rate": 2.9988901390296935e-05, + "loss": 0.0376, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014972078055143357, + "rkd_loss_weight": 0.9385146200656891, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3794202357530594, + "rkd_teacher_entropy": 3.7222827434539796, + "rkd_teacher_similarity_mean": 0.7726526081562042, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1850 + }, + { + "epoch": 0.5411696246726797, + "flow_matching_loss": 0.02412004005163908, + "grad_norm": 0.23818349838256836, + "learning_rate": 2.9988256338927442e-05, + "loss": 0.0384, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015224925614893437, + "rkd_loss_weight": 0.9381812691688538, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3836183398962021, + "rkd_teacher_entropy": 3.735090231895447, + "rkd_teacher_similarity_mean": 0.7794256746768952, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1860 + }, + { + "epoch": 0.5440791387838231, + "flow_matching_loss": 0.022074596397578715, + "grad_norm": 0.19191282987594604, + "learning_rate": 2.9987593075413198e-05, + "loss": 0.0371, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016043000109493732, + "rkd_loss_weight": 0.9378479301929474, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37757840752601624, + "rkd_teacher_entropy": 3.711268353462219, + "rkd_teacher_similarity_mean": 0.7751650989055634, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1870 + }, + { + "epoch": 0.5469886528949666, + "flow_matching_loss": 0.02183581218123436, + "grad_norm": 0.24826966226100922, + "learning_rate": 2.9986911600560128e-05, + "loss": 0.035, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014070738479495049, + "rkd_loss_weight": 0.9375145852565765, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37377003133296965, + "rkd_teacher_entropy": 3.7029200553894044, + "rkd_teacher_similarity_mean": 0.7706274509429931, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1880 + }, + { + "epoch": 0.5498981670061099, + "flow_matching_loss": 0.025321516394615173, + "grad_norm": 0.22154001891613007, + "learning_rate": 2.998621191519629e-05, + "loss": 0.0395, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015159729961305856, + "rkd_loss_weight": 0.9371812343597412, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36958602368831633, + "rkd_teacher_entropy": 3.69249906539917, + "rkd_teacher_similarity_mean": 0.7654386460781097, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1890 + }, + { + "epoch": 0.5528076811172534, + "flow_matching_loss": 0.022361057717353104, + "grad_norm": 0.20672020316123962, + "learning_rate": 2.9985494020171873e-05, + "loss": 0.0392, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0179646129719913, + "rkd_loss_weight": 0.9368478953838348, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37249798476696017, + "rkd_teacher_entropy": 3.708869957923889, + "rkd_teacher_similarity_mean": 0.7703764975070954, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1900 + }, + { + "epoch": 0.5557171952283969, + "flow_matching_loss": 0.023508424870669842, + "grad_norm": 0.19209878146648407, + "learning_rate": 2.9984757916359176e-05, + "loss": 0.0376, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015007772762328387, + "rkd_loss_weight": 0.936514550447464, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38714278042316436, + "rkd_teacher_entropy": 3.7505032777786256, + "rkd_teacher_similarity_mean": 0.7831110954284668, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1910 + }, + { + "epoch": 0.5586267093395403, + "flow_matching_loss": 0.023464873991906644, + "grad_norm": 0.17004339396953583, + "learning_rate": 2.998400360465264e-05, + "loss": 0.0394, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016969190817326306, + "rkd_loss_weight": 0.9361812055110932, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3842951595783234, + "rkd_teacher_entropy": 3.7360724925994875, + "rkd_teacher_similarity_mean": 0.7784012138843537, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1920 + }, + { + "epoch": 0.5615362234506838, + "flow_matching_loss": 0.023654601722955703, + "grad_norm": 0.2843477725982666, + "learning_rate": 2.998323108596883e-05, + "loss": 0.0384, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01573750451207161, + "rkd_loss_weight": 0.9358478605747222, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3768459439277649, + "rkd_teacher_entropy": 3.7278939485549927, + "rkd_teacher_similarity_mean": 0.7742054641246796, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1930 + }, + { + "epoch": 0.5644457375618271, + "flow_matching_loss": 0.02305780295282602, + "grad_norm": 0.2615198493003845, + "learning_rate": 2.9982440361246422e-05, + "loss": 0.0377, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01567263901233673, + "rkd_loss_weight": 0.9355145156383514, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3737346976995468, + "rkd_teacher_entropy": 3.711718797683716, + "rkd_teacher_similarity_mean": 0.7739293336868286, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1940 + }, + { + "epoch": 0.5673552516729706, + "flow_matching_loss": 0.021672856993973255, + "grad_norm": 0.21484988927841187, + "learning_rate": 2.998163143144622e-05, + "loss": 0.0347, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013946205750107766, + "rkd_loss_weight": 0.9351811707019806, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3818594992160797, + "rkd_teacher_entropy": 3.745482397079468, + "rkd_teacher_similarity_mean": 0.7818923532962799, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1950 + }, + { + "epoch": 0.570264765784114, + "flow_matching_loss": 0.021818689070641994, + "grad_norm": 0.22110293805599213, + "learning_rate": 2.9980804297551155e-05, + "loss": 0.0361, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015228192694485188, + "rkd_loss_weight": 0.9348478317260742, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38337517380714414, + "rkd_teacher_entropy": 3.7309208869934083, + "rkd_teacher_similarity_mean": 0.7802725195884704, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1960 + }, + { + "epoch": 0.5731742798952575, + "flow_matching_loss": 0.02324235923588276, + "grad_norm": 0.5232953429222107, + "learning_rate": 2.997995896056627e-05, + "loss": 0.0394, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017284761182963848, + "rkd_loss_weight": 0.9345144808292389, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37570367455482484, + "rkd_teacher_entropy": 3.713979887962341, + "rkd_teacher_similarity_mean": 0.7734692931175232, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1970 + }, + { + "epoch": 0.576083794006401, + "flow_matching_loss": 0.023306697607040405, + "grad_norm": 0.22234342992305756, + "learning_rate": 2.997909542151873e-05, + "loss": 0.0377, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015404222905635834, + "rkd_loss_weight": 0.934181135892868, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3680474698543549, + "rkd_teacher_entropy": 3.6919491052627564, + "rkd_teacher_similarity_mean": 0.7673170328140259, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1980 + }, + { + "epoch": 0.5789933081175443, + "flow_matching_loss": 0.023911610059440137, + "grad_norm": 0.26564690470695496, + "learning_rate": 2.9978213681457823e-05, + "loss": 0.0404, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017681425809860228, + "rkd_loss_weight": 0.9338478028774262, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36456931233406065, + "rkd_teacher_entropy": 3.6927927255630495, + "rkd_teacher_similarity_mean": 0.7659428000450135, + "rkd_vlm_temp": 0.10000000149011612, + "step": 1990 + }, + { + "epoch": 0.5819028222286878, + "flow_matching_loss": 0.021188224852085113, + "grad_norm": 0.23278780281543732, + "learning_rate": 2.9977313741454933e-05, + "loss": 0.0365, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01639427188783884, + "rkd_loss_weight": 0.9335144460201263, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3842614650726318, + "rkd_teacher_entropy": 3.742815113067627, + "rkd_teacher_similarity_mean": 0.7855409741401672, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2000 + }, + { + "epoch": 0.5848123363398312, + "flow_matching_loss": 0.02150292955338955, + "grad_norm": 0.22806130349636078, + "learning_rate": 2.997639560260358e-05, + "loss": 0.036, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015516559220850467, + "rkd_loss_weight": 0.93318110704422, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.364029136300087, + "rkd_teacher_entropy": 3.67610080242157, + "rkd_teacher_similarity_mean": 0.7644607961177826, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2010 + }, + { + "epoch": 0.5877218504509747, + "flow_matching_loss": 0.02388681098818779, + "grad_norm": 0.29138240218162537, + "learning_rate": 2.9975459266019387e-05, + "loss": 0.0401, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017377761472016574, + "rkd_loss_weight": 0.9328477680683136, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3802516311407089, + "rkd_teacher_entropy": 3.737135648727417, + "rkd_teacher_similarity_mean": 0.7820790112018585, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2020 + }, + { + "epoch": 0.5906313645621182, + "flow_matching_loss": 0.02419343376532197, + "grad_norm": 0.15551549196243286, + "learning_rate": 2.9974504732840092e-05, + "loss": 0.0412, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.018232410214841364, + "rkd_loss_weight": 0.9325144112110137, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3797155797481537, + "rkd_teacher_entropy": 3.7200950145721436, + "rkd_teacher_similarity_mean": 0.7762146651744842, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2030 + }, + { + "epoch": 0.5935408786732616, + "flow_matching_loss": 0.021793314814567567, + "grad_norm": 0.2094310075044632, + "learning_rate": 2.997353200422555e-05, + "loss": 0.0372, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0165372870862484, + "rkd_loss_weight": 0.9321810781955719, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3819990545511246, + "rkd_teacher_entropy": 3.7329089879989623, + "rkd_teacher_similarity_mean": 0.7800082802772522, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2040 + }, + { + "epoch": 0.596450392784405, + "flow_matching_loss": 0.025137214548885824, + "grad_norm": 0.25196507573127747, + "learning_rate": 2.99725410813577e-05, + "loss": 0.0403, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016223120503127575, + "rkd_loss_weight": 0.9318477272987366, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37543629109859467, + "rkd_teacher_entropy": 3.7009880781173705, + "rkd_teacher_similarity_mean": 0.7715258181095124, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2050 + }, + { + "epoch": 0.5993599068955484, + "flow_matching_loss": 0.02258778512477875, + "grad_norm": 0.24170750379562378, + "learning_rate": 2.997153196544063e-05, + "loss": 0.038, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016493461560457945, + "rkd_loss_weight": 0.9315143823623657, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3823054552078247, + "rkd_teacher_entropy": 3.731314539909363, + "rkd_teacher_similarity_mean": 0.7815723955631256, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2060 + }, + { + "epoch": 0.6022694210066919, + "flow_matching_loss": 0.025636531971395017, + "grad_norm": 0.21801340579986572, + "learning_rate": 2.997050465770049e-05, + "loss": 0.0421, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017731573432683945, + "rkd_loss_weight": 0.9311810433864594, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36732528507709505, + "rkd_teacher_entropy": 3.699278235435486, + "rkd_teacher_similarity_mean": 0.7656932651996613, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2070 + }, + { + "epoch": 0.6051789351178353, + "flow_matching_loss": 0.020135218929499387, + "grad_norm": 0.18343214690685272, + "learning_rate": 2.9969459159385562e-05, + "loss": 0.0377, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.018877501226961614, + "rkd_loss_weight": 0.9308476924896241, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37860189378261566, + "rkd_teacher_entropy": 3.7236247777938845, + "rkd_teacher_similarity_mean": 0.7761804342269898, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2080 + }, + { + "epoch": 0.6080884492289788, + "flow_matching_loss": 0.0240546440705657, + "grad_norm": 0.1748870611190796, + "learning_rate": 2.9968395471766237e-05, + "loss": 0.0386, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015652068052440883, + "rkd_loss_weight": 0.9305143535137177, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3837636411190033, + "rkd_teacher_entropy": 3.72776255607605, + "rkd_teacher_similarity_mean": 0.7802671611309051, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2090 + }, + { + "epoch": 0.6109979633401222, + "flow_matching_loss": 0.023156100511550905, + "grad_norm": 0.20181629061698914, + "learning_rate": 2.996731359613498e-05, + "loss": 0.0384, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016357611026614906, + "rkd_loss_weight": 0.9301810085773468, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36832368075847627, + "rkd_teacher_entropy": 3.693626618385315, + "rkd_teacher_similarity_mean": 0.7681727588176728, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2100 + }, + { + "epoch": 0.6139074774512656, + "flow_matching_loss": 0.020885653048753738, + "grad_norm": 0.3095831871032715, + "learning_rate": 2.996621353380638e-05, + "loss": 0.038, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.018447270151227713, + "rkd_loss_weight": 0.9298476576805115, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.377843314409256, + "rkd_teacher_entropy": 3.7134514808654786, + "rkd_teacher_similarity_mean": 0.7724238693714142, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2110 + }, + { + "epoch": 0.6168169915624091, + "flow_matching_loss": 0.02155868737027049, + "grad_norm": 0.16598019003868103, + "learning_rate": 2.9965095286117113e-05, + "loss": 0.035, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014490641467273236, + "rkd_loss_weight": 0.9295143187046051, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3871335029602051, + "rkd_teacher_entropy": 3.7384658813476563, + "rkd_teacher_similarity_mean": 0.7817850172519684, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2120 + }, + { + "epoch": 0.6197265056735525, + "flow_matching_loss": 0.02536831647157669, + "grad_norm": 0.15984952449798584, + "learning_rate": 2.996395885442596e-05, + "loss": 0.0398, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015554926171898843, + "rkd_loss_weight": 0.9291809737682343, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3807888299226761, + "rkd_teacher_entropy": 3.7202908992767334, + "rkd_teacher_similarity_mean": 0.7768122136592865, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2130 + }, + { + "epoch": 0.622636019784696, + "flow_matching_loss": 0.020942688547074796, + "grad_norm": 0.24405454099178314, + "learning_rate": 2.996280424011379e-05, + "loss": 0.037, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017318658716976642, + "rkd_loss_weight": 0.9288476288318634, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3743843466043472, + "rkd_teacher_entropy": 3.6996456384658813, + "rkd_teacher_similarity_mean": 0.771442985534668, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2140 + }, + { + "epoch": 0.6255455338958394, + "flow_matching_loss": 0.022330766916275023, + "grad_norm": 0.34812870621681213, + "learning_rate": 2.9961631444583565e-05, + "loss": 0.0375, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016338256374001502, + "rkd_loss_weight": 0.9285142838954925, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38677569329738615, + "rkd_teacher_entropy": 3.7293797731399536, + "rkd_teacher_similarity_mean": 0.7821490466594696, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2150 + }, + { + "epoch": 0.6284550480069828, + "flow_matching_loss": 0.020399874076247215, + "grad_norm": 0.34433314204216003, + "learning_rate": 2.9960440469260347e-05, + "loss": 0.0371, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.018026379216462374, + "rkd_loss_weight": 0.9281809389591217, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38849562108516694, + "rkd_teacher_entropy": 3.745758366584778, + "rkd_teacher_similarity_mean": 0.7835495114326477, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2160 + }, + { + "epoch": 0.6313645621181263, + "flow_matching_loss": 0.020205985009670257, + "grad_norm": 0.23297931253910065, + "learning_rate": 2.995923131559128e-05, + "loss": 0.0333, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014134494494646788, + "rkd_loss_weight": 0.9278475940227509, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37607690691947937, + "rkd_teacher_entropy": 3.7251062870025633, + "rkd_teacher_similarity_mean": 0.7785454392433167, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2170 + }, + { + "epoch": 0.6342740762292697, + "flow_matching_loss": 0.021041371300816535, + "grad_norm": 0.19633305072784424, + "learning_rate": 2.99580039850456e-05, + "loss": 0.0368, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017014738358557224, + "rkd_loss_weight": 0.9275142550468445, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37202632427215576, + "rkd_teacher_entropy": 3.7160106897354126, + "rkd_teacher_similarity_mean": 0.7738932311534882, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2180 + }, + { + "epoch": 0.6371835903404132, + "flow_matching_loss": 0.020870907884091138, + "grad_norm": 0.1772083342075348, + "learning_rate": 2.9956758479114635e-05, + "loss": 0.0331, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013151598908007146, + "rkd_loss_weight": 0.9271809041500092, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37745487987995147, + "rkd_teacher_entropy": 3.723833155632019, + "rkd_teacher_similarity_mean": 0.7750653564929962, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2190 + }, + { + "epoch": 0.6400931044515565, + "flow_matching_loss": 0.02697074003517628, + "grad_norm": 0.2544708251953125, + "learning_rate": 2.9955494799311787e-05, + "loss": 0.0426, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016831567604094745, + "rkd_loss_weight": 0.9268475592136383, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37492574751377106, + "rkd_teacher_entropy": 3.7096842765808105, + "rkd_teacher_similarity_mean": 0.7729879260063172, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2200 + }, + { + "epoch": 0.6430026185627, + "flow_matching_loss": 0.02286628885194659, + "grad_norm": 0.2949284613132477, + "learning_rate": 2.995421294717254e-05, + "loss": 0.0388, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017241451982408763, + "rkd_loss_weight": 0.9265142202377319, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37794322669506075, + "rkd_teacher_entropy": 3.7181253910064695, + "rkd_teacher_similarity_mean": 0.7790391027927399, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2210 + }, + { + "epoch": 0.6459121326738435, + "flow_matching_loss": 0.023340736329555512, + "grad_norm": 0.27659130096435547, + "learning_rate": 2.995291292425448e-05, + "loss": 0.0378, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015560544934123755, + "rkd_loss_weight": 0.9261808693408966, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3807608336210251, + "rkd_teacher_entropy": 3.7316509962081907, + "rkd_teacher_similarity_mean": 0.7810106396675109, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2220 + }, + { + "epoch": 0.6488216467849869, + "flow_matching_loss": 0.02170895887538791, + "grad_norm": 0.22449801862239838, + "learning_rate": 2.9951594732137248e-05, + "loss": 0.0402, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.019945940282195807, + "rkd_loss_weight": 0.9258475303649902, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36920385956764223, + "rkd_teacher_entropy": 3.696025514602661, + "rkd_teacher_similarity_mean": 0.767835009098053, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2230 + }, + { + "epoch": 0.6517311608961304, + "flow_matching_loss": 0.022094752453267575, + "grad_norm": 0.3292069137096405, + "learning_rate": 2.995025837242257e-05, + "loss": 0.0392, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01853440934792161, + "rkd_loss_weight": 0.9255141854286194, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37553618252277376, + "rkd_teacher_entropy": 3.714640974998474, + "rkd_teacher_similarity_mean": 0.7744473814964294, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2240 + }, + { + "epoch": 0.6546406750072737, + "flow_matching_loss": 0.021946728229522705, + "grad_norm": 0.2549424469470978, + "learning_rate": 2.994890384673426e-05, + "loss": 0.0395, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01893097646534443, + "rkd_loss_weight": 0.925180834531784, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3740757167339325, + "rkd_teacher_entropy": 3.7187893390655518, + "rkd_teacher_similarity_mean": 0.7741801798343658, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2250 + }, + { + "epoch": 0.6575501891184172, + "flow_matching_loss": 0.02246767785400152, + "grad_norm": 0.19909997284412384, + "learning_rate": 2.9947531156718174e-05, + "loss": 0.0369, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015609298646450043, + "rkd_loss_weight": 0.9248475015163422, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3802679508924484, + "rkd_teacher_entropy": 3.7429622173309327, + "rkd_teacher_similarity_mean": 0.782190603017807, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2260 + }, + { + "epoch": 0.6604597032295607, + "flow_matching_loss": 0.02080303244292736, + "grad_norm": 0.16586551070213318, + "learning_rate": 2.994614030404228e-05, + "loss": 0.0357, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01607184251770377, + "rkd_loss_weight": 0.9245141446590424, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37697080969810487, + "rkd_teacher_entropy": 3.713635563850403, + "rkd_teacher_similarity_mean": 0.7761223912239075, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2270 + }, + { + "epoch": 0.6633692173407041, + "flow_matching_loss": 0.019681403413414955, + "grad_norm": 0.1893588751554489, + "learning_rate": 2.994473129039659e-05, + "loss": 0.0343, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015784653648734093, + "rkd_loss_weight": 0.924180805683136, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38067491054534913, + "rkd_teacher_entropy": 3.7192399978637694, + "rkd_teacher_similarity_mean": 0.7764714539051056, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2280 + }, + { + "epoch": 0.6662787314518476, + "flow_matching_loss": 0.020718936715275048, + "grad_norm": 0.22965189814567566, + "learning_rate": 2.9943304117493183e-05, + "loss": 0.0369, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01750639537349343, + "rkd_loss_weight": 0.9238474667072296, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37651908695697783, + "rkd_teacher_entropy": 3.714120864868164, + "rkd_teacher_similarity_mean": 0.7741123557090759, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2290 + }, + { + "epoch": 0.669188245562991, + "flow_matching_loss": 0.02245662547647953, + "grad_norm": 0.27718105912208557, + "learning_rate": 2.9941858787066207e-05, + "loss": 0.0385, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01738027362152934, + "rkd_loss_weight": 0.9235141098499298, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37556153535842896, + "rkd_teacher_entropy": 3.7166078090667725, + "rkd_teacher_similarity_mean": 0.7732244968414307, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2300 + }, + { + "epoch": 0.6720977596741344, + "flow_matching_loss": 0.02206927640363574, + "grad_norm": 0.20092689990997314, + "learning_rate": 2.9940395300871888e-05, + "loss": 0.0366, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0157824688591063, + "rkd_loss_weight": 0.923180776834488, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3722349852323532, + "rkd_teacher_entropy": 3.699709725379944, + "rkd_teacher_similarity_mean": 0.7685117721557617, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2310 + }, + { + "epoch": 0.6750072737852778, + "flow_matching_loss": 0.021957069262862207, + "grad_norm": 0.3584940731525421, + "learning_rate": 2.993891366068849e-05, + "loss": 0.0377, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01710853260010481, + "rkd_loss_weight": 0.9228474259376526, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3841800779104233, + "rkd_teacher_entropy": 3.733253574371338, + "rkd_teacher_similarity_mean": 0.7807928919792175, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2320 + }, + { + "epoch": 0.6779167878964213, + "flow_matching_loss": 0.0254778241738677, + "grad_norm": 0.2837698757648468, + "learning_rate": 2.993741386831634e-05, + "loss": 0.0409, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01673399582505226, + "rkd_loss_weight": 0.9225140810012817, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37846958041191103, + "rkd_teacher_entropy": 3.729150915145874, + "rkd_teacher_similarity_mean": 0.7785368621349334, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2330 + }, + { + "epoch": 0.6808263020075648, + "flow_matching_loss": 0.020788709819316863, + "grad_norm": 0.37056419253349304, + "learning_rate": 2.9935895925577837e-05, + "loss": 0.0391, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01989773567765951, + "rkd_loss_weight": 0.9221807420253754, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3762740522623062, + "rkd_teacher_entropy": 3.728468084335327, + "rkd_teacher_similarity_mean": 0.7745156824588776, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2340 + }, + { + "epoch": 0.6837358161187082, + "flow_matching_loss": 0.02517085550352931, + "grad_norm": 0.23666515946388245, + "learning_rate": 2.9934359834317425e-05, + "loss": 0.0448, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.02127314554527402, + "rkd_loss_weight": 0.9218473911285401, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37864709794521334, + "rkd_teacher_entropy": 3.710331439971924, + "rkd_teacher_similarity_mean": 0.7738007783889771, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2350 + }, + { + "epoch": 0.6866453302298516, + "flow_matching_loss": 0.02271494045853615, + "grad_norm": 0.2611723840236664, + "learning_rate": 2.993280559640159e-05, + "loss": 0.0374, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015887008141726257, + "rkd_loss_weight": 0.9215140521526337, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3764863520860672, + "rkd_teacher_entropy": 3.7176710844039915, + "rkd_teacher_similarity_mean": 0.7773466765880584, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2360 + }, + { + "epoch": 0.689554844340995, + "flow_matching_loss": 0.020802856422960758, + "grad_norm": 0.25089532136917114, + "learning_rate": 2.9931233213718887e-05, + "loss": 0.0352, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015591195411980152, + "rkd_loss_weight": 0.9211807072162628, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37268540263175964, + "rkd_teacher_entropy": 3.7073514223098756, + "rkd_teacher_similarity_mean": 0.7717595338821411, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2370 + }, + { + "epoch": 0.6924643584521385, + "flow_matching_loss": 0.02043521888554096, + "grad_norm": 0.16675272583961487, + "learning_rate": 2.992964268817991e-05, + "loss": 0.0331, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013727477844804525, + "rkd_loss_weight": 0.9208473563194275, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3772790372371674, + "rkd_teacher_entropy": 3.714387631416321, + "rkd_teacher_similarity_mean": 0.777458769083023, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2380 + }, + { + "epoch": 0.695373872563282, + "flow_matching_loss": 0.022066435124725104, + "grad_norm": 0.19668002426624298, + "learning_rate": 2.9928034021717286e-05, + "loss": 0.0365, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015718644484877586, + "rkd_loss_weight": 0.9205140173435211, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3778013676404953, + "rkd_teacher_entropy": 3.7165149211883546, + "rkd_teacher_similarity_mean": 0.7775910496711731, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2390 + }, + { + "epoch": 0.6982833866744254, + "flow_matching_loss": 0.022389063984155654, + "grad_norm": 0.15821661055088043, + "learning_rate": 2.9926407216285707e-05, + "loss": 0.0418, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.02114802384749055, + "rkd_loss_weight": 0.9201806724071503, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3780795902013779, + "rkd_teacher_entropy": 3.7299895763397215, + "rkd_teacher_similarity_mean": 0.7774544298648834, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2400 + }, + { + "epoch": 0.7011929007855688, + "flow_matching_loss": 0.019828521087765692, + "grad_norm": 0.19625839591026306, + "learning_rate": 2.9924762273861894e-05, + "loss": 0.0323, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013547401130199432, + "rkd_loss_weight": 0.9198473274707795, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3870637059211731, + "rkd_teacher_entropy": 3.7388537406921385, + "rkd_teacher_similarity_mean": 0.783974701166153, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2410 + }, + { + "epoch": 0.7041024148967122, + "flow_matching_loss": 0.026062578801065682, + "grad_norm": 0.19469773769378662, + "learning_rate": 2.992309919644461e-05, + "loss": 0.0396, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014678260311484337, + "rkd_loss_weight": 0.9195139825344085, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37928621768951415, + "rkd_teacher_entropy": 3.7183714151382445, + "rkd_teacher_similarity_mean": 0.7752136886119843, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2420 + }, + { + "epoch": 0.7070119290078557, + "flow_matching_loss": 0.022567562852054836, + "grad_norm": 0.2825496196746826, + "learning_rate": 2.9921417986054633e-05, + "loss": 0.0423, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.02142473664134741, + "rkd_loss_weight": 0.9191806375980377, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3700226366519928, + "rkd_teacher_entropy": 3.700141978263855, + "rkd_teacher_similarity_mean": 0.7688384294509888, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2430 + }, + { + "epoch": 0.7099214431189991, + "flow_matching_loss": 0.021479006484150888, + "grad_norm": 0.24755221605300903, + "learning_rate": 2.9919718644734815e-05, + "loss": 0.0356, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015391422901302576, + "rkd_loss_weight": 0.9188472926616669, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3763744682073593, + "rkd_teacher_entropy": 3.71338050365448, + "rkd_teacher_similarity_mean": 0.7772617697715759, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2440 + }, + { + "epoch": 0.7128309572301426, + "flow_matching_loss": 0.02296636439859867, + "grad_norm": 0.3118444085121155, + "learning_rate": 2.991800117455e-05, + "loss": 0.0388, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017289530579000713, + "rkd_loss_weight": 0.9185139536857605, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37716975510120393, + "rkd_teacher_entropy": 3.720667767524719, + "rkd_teacher_similarity_mean": 0.7759109914302826, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2450 + }, + { + "epoch": 0.715740471341286, + "flow_matching_loss": 0.019885686878114938, + "grad_norm": 0.21054968237876892, + "learning_rate": 2.9916265577587085e-05, + "loss": 0.0337, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014996589068323374, + "rkd_loss_weight": 0.9181806087493897, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3748540788888931, + "rkd_teacher_entropy": 3.7247845411300657, + "rkd_teacher_similarity_mean": 0.772482818365097, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2460 + }, + { + "epoch": 0.7186499854524294, + "flow_matching_loss": 0.021917753946036102, + "grad_norm": 0.22752733528614044, + "learning_rate": 2.9914511855954986e-05, + "loss": 0.0372, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01660864893347025, + "rkd_loss_weight": 0.9178472578525543, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3658324539661407, + "rkd_teacher_entropy": 3.6748165130615233, + "rkd_teacher_similarity_mean": 0.765330970287323, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2470 + }, + { + "epoch": 0.7215594995635729, + "flow_matching_loss": 0.019476377870887518, + "grad_norm": 0.15797263383865356, + "learning_rate": 2.9912740011784626e-05, + "loss": 0.0356, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01760432794690132, + "rkd_loss_weight": 0.9175139248371125, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37861802279949186, + "rkd_teacher_entropy": 3.7205947160720827, + "rkd_teacher_similarity_mean": 0.7790525138378144, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2480 + }, + { + "epoch": 0.7244690136747163, + "flow_matching_loss": 0.02410769518464804, + "grad_norm": 0.22827760875225067, + "learning_rate": 2.991095004722898e-05, + "loss": 0.0357, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012644620519131422, + "rkd_loss_weight": 0.9171805679798126, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3859380006790161, + "rkd_teacher_entropy": 3.7379931688308714, + "rkd_teacher_similarity_mean": 0.7852692902088165, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2490 + }, + { + "epoch": 0.7273785277858598, + "flow_matching_loss": 0.02257515201345086, + "grad_norm": 0.2214665561914444, + "learning_rate": 2.9909141964463017e-05, + "loss": 0.0409, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.019980346597731114, + "rkd_loss_weight": 0.9168472290039062, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37656254768371583, + "rkd_teacher_entropy": 3.7156179189682006, + "rkd_teacher_similarity_mean": 0.7728223443031311, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2500 + }, + { + "epoch": 0.7302880418970032, + "flow_matching_loss": 0.02330344757065177, + "grad_norm": 0.2862389087677002, + "learning_rate": 2.990731576568372e-05, + "loss": 0.0391, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017249741405248643, + "rkd_loss_weight": 0.9165138900279999, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37623140811920164, + "rkd_teacher_entropy": 3.722564125061035, + "rkd_teacher_similarity_mean": 0.7753551363945007, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2510 + }, + { + "epoch": 0.7331975560081466, + "flow_matching_loss": 0.021762381400913, + "grad_norm": 0.30380165576934814, + "learning_rate": 2.99054714531101e-05, + "loss": 0.0356, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015103768557310104, + "rkd_loss_weight": 0.9161805331707, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37575105428695676, + "rkd_teacher_entropy": 3.721630907058716, + "rkd_teacher_similarity_mean": 0.7742697119712829, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2520 + }, + { + "epoch": 0.7361070701192901, + "flow_matching_loss": 0.02256292486563325, + "grad_norm": 0.2446414828300476, + "learning_rate": 2.9903609028983178e-05, + "loss": 0.0395, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.018525015003979205, + "rkd_loss_weight": 0.9158472001552582, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.378590852022171, + "rkd_teacher_entropy": 3.716808247566223, + "rkd_teacher_similarity_mean": 0.7749502778053283, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2530 + }, + { + "epoch": 0.7390165842304335, + "flow_matching_loss": 0.0200792427174747, + "grad_norm": 0.38833919167518616, + "learning_rate": 2.990172849556596e-05, + "loss": 0.035, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016318145301193, + "rkd_loss_weight": 0.9155138492584228, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3811863660812378, + "rkd_teacher_entropy": 3.720296597480774, + "rkd_teacher_similarity_mean": 0.7757275342941284, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2540 + }, + { + "epoch": 0.741926098341577, + "flow_matching_loss": 0.023030294664204122, + "grad_norm": 0.24040304124355316, + "learning_rate": 2.989982985514348e-05, + "loss": 0.0409, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.019573326874524354, + "rkd_loss_weight": 0.915180504322052, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36399958431720736, + "rkd_teacher_entropy": 3.6652109622955322, + "rkd_teacher_similarity_mean": 0.7595776855945587, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2550 + }, + { + "epoch": 0.7448356124527203, + "flow_matching_loss": 0.023622667510062458, + "grad_norm": 0.2545727491378784, + "learning_rate": 2.9897913110022766e-05, + "loss": 0.0398, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017729825153946877, + "rkd_loss_weight": 0.9148471653461456, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3741179883480072, + "rkd_teacher_entropy": 3.7145177125930786, + "rkd_teacher_similarity_mean": 0.7712162792682647, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2560 + }, + { + "epoch": 0.7477451265638638, + "flow_matching_loss": 0.0200529458001256, + "grad_norm": 0.22675728797912598, + "learning_rate": 2.989597826253284e-05, + "loss": 0.0354, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016781446803361178, + "rkd_loss_weight": 0.9145138144493103, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3822296172380447, + "rkd_teacher_entropy": 3.734745740890503, + "rkd_teacher_similarity_mean": 0.7773965716361999, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2570 + }, + { + "epoch": 0.7506546406750073, + "flow_matching_loss": 0.022935482300817968, + "grad_norm": 0.2150903046131134, + "learning_rate": 2.989402531502473e-05, + "loss": 0.0395, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.018151788972318172, + "rkd_loss_weight": 0.914180475473404, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38176205158233645, + "rkd_teacher_entropy": 3.731981873512268, + "rkd_teacher_similarity_mean": 0.7782416880130768, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2580 + }, + { + "epoch": 0.7535641547861507, + "flow_matching_loss": 0.018541551567614077, + "grad_norm": 0.1883399784564972, + "learning_rate": 2.989205426987144e-05, + "loss": 0.0327, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015536444447934627, + "rkd_loss_weight": 0.913847130537033, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3766990631818771, + "rkd_teacher_entropy": 3.716141200065613, + "rkd_teacher_similarity_mean": 0.7733087360858917, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2590 + }, + { + "epoch": 0.7564736688972942, + "flow_matching_loss": 0.023473278060555457, + "grad_norm": 0.216995969414711, + "learning_rate": 2.9890065129467988e-05, + "loss": 0.0363, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014091093745082616, + "rkd_loss_weight": 0.9135137796401978, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36823610663414, + "rkd_teacher_entropy": 3.6804345846176147, + "rkd_teacher_similarity_mean": 0.765192049741745, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2600 + }, + { + "epoch": 0.7593831830084375, + "flow_matching_loss": 0.021280353888869286, + "grad_norm": 0.2240721881389618, + "learning_rate": 2.9888057896231364e-05, + "loss": 0.0375, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017739271558821202, + "rkd_loss_weight": 0.9131804406642914, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37686572372913363, + "rkd_teacher_entropy": 3.7039923906326293, + "rkd_teacher_similarity_mean": 0.773543655872345, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2610 + }, + { + "epoch": 0.762292697119581, + "flow_matching_loss": 0.02182421777397394, + "grad_norm": 0.23747050762176514, + "learning_rate": 2.988603257260054e-05, + "loss": 0.0354, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01481951791793108, + "rkd_loss_weight": 0.9128470957279206, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3804984509944916, + "rkd_teacher_entropy": 3.7317757844924926, + "rkd_teacher_similarity_mean": 0.7784815013408661, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2620 + }, + { + "epoch": 0.7652022112307245, + "flow_matching_loss": 0.02086776401847601, + "grad_norm": 0.35758310556411743, + "learning_rate": 2.988398916103648e-05, + "loss": 0.0365, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017154190689325333, + "rkd_loss_weight": 0.9125137507915497, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3766995906829834, + "rkd_teacher_entropy": 3.7186562538146974, + "rkd_teacher_similarity_mean": 0.7754173398017883, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2630 + }, + { + "epoch": 0.7681117253418679, + "flow_matching_loss": 0.02175281587988138, + "grad_norm": 0.1637771725654602, + "learning_rate": 2.988192766402212e-05, + "loss": 0.0348, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014344534557312726, + "rkd_loss_weight": 0.9121804058551788, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3835303604602814, + "rkd_teacher_entropy": 3.7299055814743043, + "rkd_teacher_similarity_mean": 0.7785704076290131, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2640 + }, + { + "epoch": 0.7710212394530114, + "flow_matching_loss": 0.022756517771631478, + "grad_norm": 0.3708428740501404, + "learning_rate": 2.9879848084062378e-05, + "loss": 0.0371, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01570715196430683, + "rkd_loss_weight": 0.911847060918808, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37558688819408415, + "rkd_teacher_entropy": 3.697418546676636, + "rkd_teacher_similarity_mean": 0.7690761268138886, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2650 + }, + { + "epoch": 0.7739307535641547, + "flow_matching_loss": 0.02189352661371231, + "grad_norm": 0.354775995016098, + "learning_rate": 2.9877750423684132e-05, + "loss": 0.0365, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016017126478254796, + "rkd_loss_weight": 0.9115137159824371, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37340347170829774, + "rkd_teacher_entropy": 3.7064319849014282, + "rkd_teacher_similarity_mean": 0.7700973331928254, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2660 + }, + { + "epoch": 0.7768402676752982, + "flow_matching_loss": 0.02224481077864766, + "grad_norm": 0.21931825578212738, + "learning_rate": 2.987563468543624e-05, + "loss": 0.0399, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.019324551802128553, + "rkd_loss_weight": 0.9111803770065308, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37233260571956633, + "rkd_teacher_entropy": 3.701659059524536, + "rkd_teacher_similarity_mean": 0.7680107593536377, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2670 + }, + { + "epoch": 0.7797497817864416, + "flow_matching_loss": 0.022693347185850143, + "grad_norm": 0.2752886414527893, + "learning_rate": 2.9873500871889533e-05, + "loss": 0.0408, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.019892242178320883, + "rkd_loss_weight": 0.9108470261096955, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37794829308986666, + "rkd_teacher_entropy": 3.720204496383667, + "rkd_teacher_similarity_mean": 0.7752927362918853, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2680 + }, + { + "epoch": 0.7826592958975851, + "flow_matching_loss": 0.022279214207082988, + "grad_norm": 0.1851450502872467, + "learning_rate": 2.9871348985636778e-05, + "loss": 0.0364, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015510813612490893, + "rkd_loss_weight": 0.9105136811733245, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3833311080932617, + "rkd_teacher_entropy": 3.719673728942871, + "rkd_teacher_similarity_mean": 0.7807234466075897, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2690 + }, + { + "epoch": 0.7855688100087286, + "flow_matching_loss": 0.022414901480078697, + "grad_norm": 0.2579030990600586, + "learning_rate": 2.986917902929273e-05, + "loss": 0.0422, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.021716721542179584, + "rkd_loss_weight": 0.9101803421974182, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37604604065418246, + "rkd_teacher_entropy": 3.719306755065918, + "rkd_teacher_similarity_mean": 0.7739633858203888, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2700 + }, + { + "epoch": 0.788478324119872, + "flow_matching_loss": 0.021670648269355296, + "grad_norm": 0.284196138381958, + "learning_rate": 2.9866991005494095e-05, + "loss": 0.0395, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01957749016582966, + "rkd_loss_weight": 0.9098469913005829, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3783859461545944, + "rkd_teacher_entropy": 3.7204967021942137, + "rkd_teacher_similarity_mean": 0.7765630722045899, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2710 + }, + { + "epoch": 0.7913878382310154, + "flow_matching_loss": 0.02075803643092513, + "grad_norm": 0.22673726081848145, + "learning_rate": 2.986478491689952e-05, + "loss": 0.0356, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01627278169617057, + "rkd_loss_weight": 0.9095136523246765, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37400893270969393, + "rkd_teacher_entropy": 3.728682589530945, + "rkd_teacher_similarity_mean": 0.7745314836502075, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2720 + }, + { + "epoch": 0.7942973523421588, + "flow_matching_loss": 0.021366867143660782, + "grad_norm": 0.24494992196559906, + "learning_rate": 2.986256076618961e-05, + "loss": 0.0391, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.019556437619030476, + "rkd_loss_weight": 0.9091803073883057, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36733774542808534, + "rkd_teacher_entropy": 3.6916805505752563, + "rkd_teacher_similarity_mean": 0.7663785696029664, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2730 + }, + { + "epoch": 0.7972068664533023, + "flow_matching_loss": 0.019534143153578043, + "grad_norm": 0.38684606552124023, + "learning_rate": 2.986031855606692e-05, + "loss": 0.0339, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015770553424954415, + "rkd_loss_weight": 0.9088469564914703, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3734954923391342, + "rkd_teacher_entropy": 3.702902150154114, + "rkd_teacher_similarity_mean": 0.7691693842411041, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2740 + }, + { + "epoch": 0.8001163805644458, + "flow_matching_loss": 0.019094485510140657, + "grad_norm": 0.20532383024692535, + "learning_rate": 2.9858058289255954e-05, + "loss": 0.0376, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.020405642129480837, + "rkd_loss_weight": 0.9085136234760285, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.373860576748848, + "rkd_teacher_entropy": 3.7079575777053835, + "rkd_teacher_similarity_mean": 0.7710104823112488, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2750 + }, + { + "epoch": 0.8030258946755892, + "flow_matching_loss": 0.02052140301093459, + "grad_norm": 0.14627580344676971, + "learning_rate": 2.985577996850314e-05, + "loss": 0.0357, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01670566825196147, + "rkd_loss_weight": 0.9081802666187286, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3719192296266556, + "rkd_teacher_entropy": 3.692878746986389, + "rkd_teacher_similarity_mean": 0.7686497688293457, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2760 + }, + { + "epoch": 0.8059354087867326, + "flow_matching_loss": 0.022434815019369125, + "grad_norm": 0.21965086460113525, + "learning_rate": 2.9853483596576856e-05, + "loss": 0.0381, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01724703749641776, + "rkd_loss_weight": 0.9078469276428223, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3761195123195648, + "rkd_teacher_entropy": 3.6995809078216553, + "rkd_teacher_similarity_mean": 0.7712027192115783, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2770 + }, + { + "epoch": 0.808844922897876, + "flow_matching_loss": 0.020978328585624696, + "grad_norm": 0.24813099205493927, + "learning_rate": 2.9851169176267412e-05, + "loss": 0.0352, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015622197557240725, + "rkd_loss_weight": 0.9075135886669159, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3752696573734283, + "rkd_teacher_entropy": 3.716251015663147, + "rkd_teacher_similarity_mean": 0.7738028287887573, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2780 + }, + { + "epoch": 0.8117544370090195, + "flow_matching_loss": 0.022000928595662116, + "grad_norm": 0.2630294859409332, + "learning_rate": 2.9848836710387044e-05, + "loss": 0.0412, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.021149823814630507, + "rkd_loss_weight": 0.907180231809616, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3744067341089249, + "rkd_teacher_entropy": 3.703455901145935, + "rkd_teacher_similarity_mean": 0.7708970069885254, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2790 + }, + { + "epoch": 0.814663951120163, + "flow_matching_loss": 0.021068753302097322, + "grad_norm": 0.2642725110054016, + "learning_rate": 2.9846486201769917e-05, + "loss": 0.0339, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014104197267442942, + "rkd_loss_weight": 0.9068468987941742, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3894589513540268, + "rkd_teacher_entropy": 3.7535003185272218, + "rkd_teacher_similarity_mean": 0.786253136396408, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2800 + }, + { + "epoch": 0.8175734652313064, + "flow_matching_loss": 0.0245265144854784, + "grad_norm": 0.23937028646469116, + "learning_rate": 2.9844117653272122e-05, + "loss": 0.0411, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01831063497811556, + "rkd_loss_weight": 0.9065135478973388, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3755494147539139, + "rkd_teacher_entropy": 3.707147145271301, + "rkd_teacher_similarity_mean": 0.7706023752689362, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2810 + }, + { + "epoch": 0.8204829793424498, + "flow_matching_loss": 0.02225257605314255, + "grad_norm": 0.2111952304840088, + "learning_rate": 2.9841731067771676e-05, + "loss": 0.0399, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01943623973056674, + "rkd_loss_weight": 0.906180202960968, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3698846220970154, + "rkd_teacher_entropy": 3.691010570526123, + "rkd_teacher_similarity_mean": 0.7680344164371491, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2820 + }, + { + "epoch": 0.8233924934535932, + "flow_matching_loss": 0.019630792271345853, + "grad_norm": 0.23405013978481293, + "learning_rate": 2.9839326448168492e-05, + "loss": 0.0346, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016570462752133606, + "rkd_loss_weight": 0.9058468639850616, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3764860391616821, + "rkd_teacher_entropy": 3.720901298522949, + "rkd_teacher_similarity_mean": 0.7767000079154969, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2830 + }, + { + "epoch": 0.8263020075647367, + "flow_matching_loss": 0.020564391929656266, + "grad_norm": 0.23352648317813873, + "learning_rate": 2.983690379738442e-05, + "loss": 0.0387, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.019985788222402334, + "rkd_loss_weight": 0.9055135130882264, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.386385378241539, + "rkd_teacher_entropy": 3.7486836671829225, + "rkd_teacher_similarity_mean": 0.7859152257442474, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2840 + }, + { + "epoch": 0.8292115216758801, + "flow_matching_loss": 0.01977533670142293, + "grad_norm": 0.18135900795459747, + "learning_rate": 2.9834463118363204e-05, + "loss": 0.0338, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015454100165516137, + "rkd_loss_weight": 0.90518017411232, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38027210235595704, + "rkd_teacher_entropy": 3.71722469329834, + "rkd_teacher_similarity_mean": 0.7790140092372895, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2850 + }, + { + "epoch": 0.8321210357870236, + "flow_matching_loss": 0.023939158022403716, + "grad_norm": 0.17986086010932922, + "learning_rate": 2.98320044140705e-05, + "loss": 0.0375, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015035375580191612, + "rkd_loss_weight": 0.904846829175949, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3775987893342972, + "rkd_teacher_entropy": 3.718233895301819, + "rkd_teacher_similarity_mean": 0.7769046902656556, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2860 + }, + { + "epoch": 0.835030549898167, + "flow_matching_loss": 0.021576057374477386, + "grad_norm": 0.19923748075962067, + "learning_rate": 2.9829527687493868e-05, + "loss": 0.0351, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015005903225392104, + "rkd_loss_weight": 0.9045134782791138, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38454224467277526, + "rkd_teacher_entropy": 3.723585367202759, + "rkd_teacher_similarity_mean": 0.7801239669322968, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2870 + }, + { + "epoch": 0.8379400640093104, + "flow_matching_loss": 0.020965270604938267, + "grad_norm": 0.16983096301555634, + "learning_rate": 2.9827032941642765e-05, + "loss": 0.0349, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015413758903741836, + "rkd_loss_weight": 0.9041801393032074, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38103680312633514, + "rkd_teacher_entropy": 3.7185941696166993, + "rkd_teacher_similarity_mean": 0.776253879070282, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2880 + }, + { + "epoch": 0.8408495781204539, + "flow_matching_loss": 0.022592691145837306, + "grad_norm": 0.1804000586271286, + "learning_rate": 2.9824520179548535e-05, + "loss": 0.0373, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01622123410925269, + "rkd_loss_weight": 0.903846800327301, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3725450813770294, + "rkd_teacher_entropy": 3.7060416460037233, + "rkd_teacher_similarity_mean": 0.7687736034393311, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2890 + }, + { + "epoch": 0.8437590922315973, + "flow_matching_loss": 0.02123790271580219, + "grad_norm": 0.20799818634986877, + "learning_rate": 2.9821989404264428e-05, + "loss": 0.0335, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01362139154225588, + "rkd_loss_weight": 0.9035134494304657, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38463216125965116, + "rkd_teacher_entropy": 3.7297805309295655, + "rkd_teacher_similarity_mean": 0.7805473685264588, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2900 + }, + { + "epoch": 0.8466686063427408, + "flow_matching_loss": 0.02066362965852022, + "grad_norm": 0.5710117816925049, + "learning_rate": 2.9819440618865572e-05, + "loss": 0.0378, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.018958767782896756, + "rkd_loss_weight": 0.9031801044940948, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3793178617954254, + "rkd_teacher_entropy": 3.718584585189819, + "rkd_teacher_similarity_mean": 0.7749667406082154, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2910 + }, + { + "epoch": 0.8495781204538843, + "flow_matching_loss": 0.022759503033012152, + "grad_norm": 0.20023953914642334, + "learning_rate": 2.981687382644898e-05, + "loss": 0.0391, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.018077775929123165, + "rkd_loss_weight": 0.9028467655181884, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3798105716705322, + "rkd_teacher_entropy": 3.716178846359253, + "rkd_teacher_similarity_mean": 0.7758917331695556, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2920 + }, + { + "epoch": 0.8524876345650276, + "flow_matching_loss": 0.02209971398115158, + "grad_norm": 0.20419879257678986, + "learning_rate": 2.9814289030133556e-05, + "loss": 0.0379, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01753139588981867, + "rkd_loss_weight": 0.9025134146213531, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37731626629829407, + "rkd_teacher_entropy": 3.714985466003418, + "rkd_teacher_similarity_mean": 0.7749333620071411, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2930 + }, + { + "epoch": 0.8553971486761711, + "flow_matching_loss": 0.023786648735404016, + "grad_norm": 0.24411091208457947, + "learning_rate": 2.9811686233060058e-05, + "loss": 0.0383, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01612147530540824, + "rkd_loss_weight": 0.9021800756454468, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3728377789258957, + "rkd_teacher_entropy": 3.6892472743988036, + "rkd_teacher_similarity_mean": 0.7684978306293487, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2940 + }, + { + "epoch": 0.8583066627873145, + "flow_matching_loss": 0.02089438308030367, + "grad_norm": 0.2745361030101776, + "learning_rate": 2.9809065438391138e-05, + "loss": 0.0366, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017387271765619515, + "rkd_loss_weight": 0.901846730709076, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3844774216413498, + "rkd_teacher_entropy": 3.7347646951675415, + "rkd_teacher_similarity_mean": 0.7811690628528595, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2950 + }, + { + "epoch": 0.861216176898458, + "flow_matching_loss": 0.021434941980987787, + "grad_norm": 0.4448002278804779, + "learning_rate": 2.9806426649311304e-05, + "loss": 0.038, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.018321879487484694, + "rkd_loss_weight": 0.9015133798122406, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37858544290065765, + "rkd_teacher_entropy": 3.7146756649017334, + "rkd_teacher_similarity_mean": 0.775092726945877, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2960 + }, + { + "epoch": 0.8641256910096013, + "flow_matching_loss": 0.020273437723517418, + "grad_norm": 0.2332616150379181, + "learning_rate": 2.980376986902694e-05, + "loss": 0.0339, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015148940589278936, + "rkd_loss_weight": 0.9011800467967988, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3771302491426468, + "rkd_teacher_entropy": 3.7110060930252073, + "rkd_teacher_similarity_mean": 0.7745823681354522, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2970 + }, + { + "epoch": 0.8670352051207448, + "flow_matching_loss": 0.020679844729602336, + "grad_norm": 0.23219895362854004, + "learning_rate": 2.9801095100766276e-05, + "loss": 0.0361, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017131804022938013, + "rkd_loss_weight": 0.9008466899394989, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37512710094451907, + "rkd_teacher_entropy": 3.706352186203003, + "rkd_teacher_similarity_mean": 0.7703518807888031, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2980 + }, + { + "epoch": 0.8699447192318883, + "flow_matching_loss": 0.019544066861271858, + "grad_norm": 0.20626963675022125, + "learning_rate": 2.979840234777942e-05, + "loss": 0.0344, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01650428753346205, + "rkd_loss_weight": 0.9005133509635925, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3760264217853546, + "rkd_teacher_entropy": 3.7083136796951295, + "rkd_teacher_similarity_mean": 0.7727811992168426, + "rkd_vlm_temp": 0.10000000149011612, + "step": 2990 + }, + { + "epoch": 0.8728542333430317, + "flow_matching_loss": 0.019731988292187454, + "grad_norm": 0.1993037611246109, + "learning_rate": 2.979569161333831e-05, + "loss": 0.0363, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.018354375008493663, + "rkd_loss_weight": 0.9001800119876862, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3863524317741394, + "rkd_teacher_entropy": 3.7596565008163454, + "rkd_teacher_similarity_mean": 0.7869907081127167, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3000 + }, + { + "epoch": 0.8757637474541752, + "flow_matching_loss": 0.01971118887886405, + "grad_norm": 0.292004257440567, + "learning_rate": 2.9792962900736744e-05, + "loss": 0.0357, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01774954656139016, + "rkd_loss_weight": 0.8998466551303863, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.373220095038414, + "rkd_teacher_entropy": 3.711384725570679, + "rkd_teacher_similarity_mean": 0.7718468904495239, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3010 + }, + { + "epoch": 0.8786732615653186, + "flow_matching_loss": 0.021474307216703892, + "grad_norm": 0.2512216567993164, + "learning_rate": 2.9790216213290372e-05, + "loss": 0.0441, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.025136415008455516, + "rkd_loss_weight": 0.8995133221149445, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3703026294708252, + "rkd_teacher_entropy": 3.7037659883499146, + "rkd_teacher_similarity_mean": 0.7682035028934479, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3020 + }, + { + "epoch": 0.881582775676462, + "flow_matching_loss": 0.018289437517523766, + "grad_norm": 0.21721874177455902, + "learning_rate": 2.9787451554336672e-05, + "loss": 0.0316, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014787649922072887, + "rkd_loss_weight": 0.8991799712181091, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3860099673271179, + "rkd_teacher_entropy": 3.758297848701477, + "rkd_teacher_similarity_mean": 0.7846527695655823, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3030 + }, + { + "epoch": 0.8844922897876055, + "flow_matching_loss": 0.02187084686011076, + "grad_norm": 0.17229144275188446, + "learning_rate": 2.9784668927234972e-05, + "loss": 0.0383, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01826531570404768, + "rkd_loss_weight": 0.8988466262817383, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37639175057411195, + "rkd_teacher_entropy": 3.719696068763733, + "rkd_teacher_similarity_mean": 0.7746940195560456, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3040 + }, + { + "epoch": 0.8874018038987489, + "flow_matching_loss": 0.018704318441450596, + "grad_norm": 0.3009820580482483, + "learning_rate": 2.9781868335366418e-05, + "loss": 0.0332, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016093359794467688, + "rkd_loss_weight": 0.8985132873058319, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37462512254714964, + "rkd_teacher_entropy": 3.694365358352661, + "rkd_teacher_similarity_mean": 0.7707667708396911, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3050 + }, + { + "epoch": 0.8903113180098924, + "flow_matching_loss": 0.021211790572851896, + "grad_norm": 0.2892228662967682, + "learning_rate": 2.9779049782134e-05, + "loss": 0.0379, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.018559802789241076, + "rkd_loss_weight": 0.8981799364089966, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3649192780256271, + "rkd_teacher_entropy": 3.684034752845764, + "rkd_teacher_similarity_mean": 0.7643711626529693, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3060 + }, + { + "epoch": 0.8932208321210358, + "flow_matching_loss": 0.01944883158430457, + "grad_norm": 0.21430887281894684, + "learning_rate": 2.9776213270962523e-05, + "loss": 0.0335, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015632299613207577, + "rkd_loss_weight": 0.8978465974330903, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38190936148166654, + "rkd_teacher_entropy": 3.72374906539917, + "rkd_teacher_similarity_mean": 0.7795999348163605, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3070 + }, + { + "epoch": 0.8961303462321792, + "flow_matching_loss": 0.02085421197116375, + "grad_norm": 0.23774777352809906, + "learning_rate": 2.977335880529861e-05, + "loss": 0.0334, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013947513047605753, + "rkd_loss_weight": 0.8975132524967193, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37946425974369047, + "rkd_teacher_entropy": 3.717381811141968, + "rkd_teacher_similarity_mean": 0.774268501996994, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3080 + }, + { + "epoch": 0.8990398603433226, + "flow_matching_loss": 0.02140059173107147, + "grad_norm": 0.3623819649219513, + "learning_rate": 2.977048638861072e-05, + "loss": 0.0351, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015243022330105304, + "rkd_loss_weight": 0.897179901599884, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38491667807102203, + "rkd_teacher_entropy": 3.7220632553100588, + "rkd_teacher_similarity_mean": 0.7799607515335083, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3090 + }, + { + "epoch": 0.9019493744544661, + "flow_matching_loss": 0.02183159636333585, + "grad_norm": 0.1826540231704712, + "learning_rate": 2.97675960243891e-05, + "loss": 0.0347, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014302191045135259, + "rkd_loss_weight": 0.8968465626239777, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3698392152786255, + "rkd_teacher_entropy": 3.691618633270264, + "rkd_teacher_similarity_mean": 0.7668278336524963, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3100 + }, + { + "epoch": 0.9048588885656096, + "flow_matching_loss": 0.019644910283386708, + "grad_norm": 0.19239062070846558, + "learning_rate": 2.976468771614582e-05, + "loss": 0.0358, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017970732413232327, + "rkd_loss_weight": 0.8965132176876068, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37243176698684693, + "rkd_teacher_entropy": 3.6907517433166506, + "rkd_teacher_similarity_mean": 0.7677657544612885, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3110 + }, + { + "epoch": 0.907768402676753, + "flow_matching_loss": 0.020704889297485353, + "grad_norm": 0.22079487144947052, + "learning_rate": 2.9761761467414736e-05, + "loss": 0.0327, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013434414379298687, + "rkd_loss_weight": 0.896179872751236, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3790281444787979, + "rkd_teacher_entropy": 3.7267746925354004, + "rkd_teacher_similarity_mean": 0.7765112698078156, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3120 + }, + { + "epoch": 0.9106779167878964, + "flow_matching_loss": 0.022432346269488333, + "grad_norm": 0.26864972710609436, + "learning_rate": 2.9758817281751538e-05, + "loss": 0.0375, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01684987246990204, + "rkd_loss_weight": 0.8958465278148651, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37186062932014463, + "rkd_teacher_entropy": 3.6875205278396606, + "rkd_teacher_similarity_mean": 0.7672874987125397, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3130 + }, + { + "epoch": 0.9135874308990398, + "flow_matching_loss": 0.020966415107250214, + "grad_norm": 0.2057720422744751, + "learning_rate": 2.975585516273367e-05, + "loss": 0.0345, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015095022972673178, + "rkd_loss_weight": 0.8955131828784942, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3751755774021149, + "rkd_teacher_entropy": 3.7074207067489624, + "rkd_teacher_similarity_mean": 0.7723685801029205, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3140 + }, + { + "epoch": 0.9164969450101833, + "flow_matching_loss": 0.020067215245217086, + "grad_norm": 0.1836581528186798, + "learning_rate": 2.975287511396039e-05, + "loss": 0.0333, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014727825578302146, + "rkd_loss_weight": 0.8951798379421234, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3796973764896393, + "rkd_teacher_entropy": 3.7161335945129395, + "rkd_teacher_similarity_mean": 0.7762080252170562, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3150 + }, + { + "epoch": 0.9194064591213268, + "flow_matching_loss": 0.019998880941420792, + "grad_norm": 0.2813730835914612, + "learning_rate": 2.9749877139052745e-05, + "loss": 0.0372, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.019257364235818385, + "rkd_loss_weight": 0.894846498966217, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3748126834630966, + "rkd_teacher_entropy": 3.7061303853988647, + "rkd_teacher_similarity_mean": 0.7708778738975525, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3160 + }, + { + "epoch": 0.9223159732324702, + "flow_matching_loss": 0.01851281337440014, + "grad_norm": 0.1758991926908493, + "learning_rate": 2.9746861241653555e-05, + "loss": 0.0323, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015426819305866956, + "rkd_loss_weight": 0.8945131480693818, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3837073862552643, + "rkd_teacher_entropy": 3.734856677055359, + "rkd_teacher_similarity_mean": 0.7801495134830475, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3170 + }, + { + "epoch": 0.9252254873436137, + "flow_matching_loss": 0.01951051438227296, + "grad_norm": 0.16904893517494202, + "learning_rate": 2.9743827425427416e-05, + "loss": 0.0337, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015844588819891214, + "rkd_loss_weight": 0.8941798031330108, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38054288923740387, + "rkd_teacher_entropy": 3.723429799079895, + "rkd_teacher_similarity_mean": 0.7779239892959595, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3180 + }, + { + "epoch": 0.928135001454757, + "flow_matching_loss": 0.0205963677726686, + "grad_norm": 0.2615310251712799, + "learning_rate": 2.9740775694060704e-05, + "loss": 0.0365, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01777731077745557, + "rkd_loss_weight": 0.8938464641571044, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3714138329029083, + "rkd_teacher_entropy": 3.6982248306274412, + "rkd_teacher_similarity_mean": 0.768273901939392, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3190 + }, + { + "epoch": 0.9310445155659005, + "flow_matching_loss": 0.020448963157832623, + "grad_norm": 0.19004230201244354, + "learning_rate": 2.973770605126156e-05, + "loss": 0.0336, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01474605854600668, + "rkd_loss_weight": 0.8935131132602692, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3913265436887741, + "rkd_teacher_entropy": 3.76165509223938, + "rkd_teacher_similarity_mean": 0.7873181104660034, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3200 + }, + { + "epoch": 0.9339540296770439, + "flow_matching_loss": 0.021040507964789866, + "grad_norm": 0.6403438448905945, + "learning_rate": 2.9734618500759886e-05, + "loss": 0.0362, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01697912784293294, + "rkd_loss_weight": 0.8931797742843628, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3827075302600861, + "rkd_teacher_entropy": 3.7258423805236816, + "rkd_teacher_similarity_mean": 0.7778495073318481, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3210 + }, + { + "epoch": 0.9368635437881874, + "flow_matching_loss": 0.019751848932355643, + "grad_norm": 0.2540021240711212, + "learning_rate": 2.9731513046307356e-05, + "loss": 0.0337, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015581326093524694, + "rkd_loss_weight": 0.892846429347992, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.377691438794136, + "rkd_teacher_entropy": 3.7309195756912232, + "rkd_teacher_similarity_mean": 0.779142963886261, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3220 + }, + { + "epoch": 0.9397730578993309, + "flow_matching_loss": 0.017677047941833734, + "grad_norm": 0.22104975581169128, + "learning_rate": 2.972838969167738e-05, + "loss": 0.0306, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014484201837331057, + "rkd_loss_weight": 0.8925130784511566, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37969958782196045, + "rkd_teacher_entropy": 3.736437439918518, + "rkd_teacher_similarity_mean": 0.783122181892395, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3230 + }, + { + "epoch": 0.9426825720104742, + "flow_matching_loss": 0.020523226261138915, + "grad_norm": 0.31322577595710754, + "learning_rate": 2.972524844066513e-05, + "loss": 0.0351, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016378701757639648, + "rkd_loss_weight": 0.8921797454357148, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.379490265250206, + "rkd_teacher_entropy": 3.728461980819702, + "rkd_teacher_similarity_mean": 0.7800122141838074, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3240 + }, + { + "epoch": 0.9455920861216177, + "flow_matching_loss": 0.019952893909066915, + "grad_norm": 0.23385240137577057, + "learning_rate": 2.9722089297087524e-05, + "loss": 0.0349, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0167588422074914, + "rkd_loss_weight": 0.8918463885784149, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3721825063228607, + "rkd_teacher_entropy": 3.7175161361694338, + "rkd_teacher_similarity_mean": 0.7722346723079682, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3250 + }, + { + "epoch": 0.9485016002327611, + "flow_matching_loss": 0.021359178889542817, + "grad_norm": 0.14514945447444916, + "learning_rate": 2.971891226478322e-05, + "loss": 0.0378, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.018446282483637333, + "rkd_loss_weight": 0.8915130496025085, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3730697363615036, + "rkd_teacher_entropy": 3.7009692668914793, + "rkd_teacher_similarity_mean": 0.7695068418979645, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3260 + }, + { + "epoch": 0.9514111143439046, + "flow_matching_loss": 0.019913725648075344, + "grad_norm": 0.22964127361774445, + "learning_rate": 2.9715717347612615e-05, + "loss": 0.0314, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012910183239728212, + "rkd_loss_weight": 0.8911797106266022, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37817307710647585, + "rkd_teacher_entropy": 3.7235800981521607, + "rkd_teacher_similarity_mean": 0.7774867415428162, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3270 + }, + { + "epoch": 0.9543206284550481, + "flow_matching_loss": 0.021304117888212203, + "grad_norm": 0.19974996149539948, + "learning_rate": 2.971250454945783e-05, + "loss": 0.0369, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017533960565924645, + "rkd_loss_weight": 0.8908463537693023, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.372209432721138, + "rkd_teacher_entropy": 3.707197642326355, + "rkd_teacher_similarity_mean": 0.7687856376171112, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3280 + }, + { + "epoch": 0.9572301425661914, + "flow_matching_loss": 0.02531222067773342, + "grad_norm": 0.2155131846666336, + "learning_rate": 2.970927387422272e-05, + "loss": 0.0401, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016625000257045032, + "rkd_loss_weight": 0.8905130207538605, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3718187987804413, + "rkd_teacher_entropy": 3.7091668605804444, + "rkd_teacher_similarity_mean": 0.768878048658371, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3290 + }, + { + "epoch": 0.9601396566773349, + "flow_matching_loss": 0.01950321178883314, + "grad_norm": 0.20493513345718384, + "learning_rate": 2.9706025325832858e-05, + "loss": 0.0326, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01469958508387208, + "rkd_loss_weight": 0.8901796698570251, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3782210975885391, + "rkd_teacher_entropy": 3.719260597229004, + "rkd_teacher_similarity_mean": 0.7759594261646271, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3300 + }, + { + "epoch": 0.9630491707884783, + "flow_matching_loss": 0.019921387545764446, + "grad_norm": 0.25216153264045715, + "learning_rate": 2.9702758908235534e-05, + "loss": 0.0331, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014775821194052696, + "rkd_loss_weight": 0.8898463249206543, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38036724328994753, + "rkd_teacher_entropy": 3.7240681409835816, + "rkd_teacher_similarity_mean": 0.7780808866024017, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3310 + }, + { + "epoch": 0.9659586848996218, + "flow_matching_loss": 0.02233472652733326, + "grad_norm": 0.23679834604263306, + "learning_rate": 2.9699474625399765e-05, + "loss": 0.0357, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01506697228178382, + "rkd_loss_weight": 0.8895129859447479, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3788838326931, + "rkd_teacher_entropy": 3.7197112321853636, + "rkd_teacher_similarity_mean": 0.776225483417511, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3320 + }, + { + "epoch": 0.9688681990107652, + "flow_matching_loss": 0.019999795127660036, + "grad_norm": 0.20186224579811096, + "learning_rate": 2.9696172481316246e-05, + "loss": 0.0348, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01667084265500307, + "rkd_loss_weight": 0.8891796410083771, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3712177246809006, + "rkd_teacher_entropy": 3.7080034255981444, + "rkd_teacher_similarity_mean": 0.7699445128440857, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3330 + }, + { + "epoch": 0.9717777131219086, + "flow_matching_loss": 0.020818234700709582, + "grad_norm": 0.2203471064567566, + "learning_rate": 2.9692852479997407e-05, + "loss": 0.0365, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017587726470083, + "rkd_loss_weight": 0.8888462960720063, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3793377190828323, + "rkd_teacher_entropy": 3.717177939414978, + "rkd_teacher_similarity_mean": 0.7766895890235901, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3340 + }, + { + "epoch": 0.9746872272330521, + "flow_matching_loss": 0.020586607325822116, + "grad_norm": 0.20187993347644806, + "learning_rate": 2.9689514625477364e-05, + "loss": 0.0358, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01715637296438217, + "rkd_loss_weight": 0.8885129511356353, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3828275889158249, + "rkd_teacher_entropy": 3.7304853439331054, + "rkd_teacher_similarity_mean": 0.7768496692180633, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3350 + }, + { + "epoch": 0.9775967413441955, + "flow_matching_loss": 0.020499402843415736, + "grad_norm": 0.2604275941848755, + "learning_rate": 2.968615892181192e-05, + "loss": 0.0337, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014906145818531513, + "rkd_loss_weight": 0.8881796061992645, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37563939988613126, + "rkd_teacher_entropy": 3.704601359367371, + "rkd_teacher_similarity_mean": 0.7721707105636597, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3360 + }, + { + "epoch": 0.980506255455339, + "flow_matching_loss": 0.021015137806534766, + "grad_norm": 0.2169305682182312, + "learning_rate": 2.9682785373078567e-05, + "loss": 0.0355, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0163703135214746, + "rkd_loss_weight": 0.8878462612628937, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3802900522947311, + "rkd_teacher_entropy": 3.7282402992248533, + "rkd_teacher_similarity_mean": 0.7774003982543946, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3370 + }, + { + "epoch": 0.9834157695664824, + "flow_matching_loss": 0.018562513403594494, + "grad_norm": 0.37753722071647644, + "learning_rate": 2.9679393983376493e-05, + "loss": 0.0324, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015556371491402387, + "rkd_loss_weight": 0.8875129222869873, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3708679646253586, + "rkd_teacher_entropy": 3.6959417343139647, + "rkd_teacher_similarity_mean": 0.7696940124034881, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3380 + }, + { + "epoch": 0.9863252836776258, + "flow_matching_loss": 0.01789107369259, + "grad_norm": 0.27700236439704895, + "learning_rate": 2.9675984756826545e-05, + "loss": 0.036, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.020463995542377232, + "rkd_loss_weight": 0.887179571390152, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38662601113319395, + "rkd_teacher_entropy": 3.745414209365845, + "rkd_teacher_similarity_mean": 0.7834064722061157, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3390 + }, + { + "epoch": 0.9892347977887693, + "flow_matching_loss": 0.022486172337085008, + "grad_norm": 0.2128812074661255, + "learning_rate": 2.9672557697571273e-05, + "loss": 0.038, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017530356347560883, + "rkd_loss_weight": 0.8868462264537811, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3814424365758896, + "rkd_teacher_entropy": 3.7252819776535033, + "rkd_teacher_similarity_mean": 0.7758854806423188, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3400 + }, + { + "epoch": 0.9921443118999127, + "flow_matching_loss": 0.017335179541260005, + "grad_norm": 0.22496908903121948, + "learning_rate": 2.966911280977486e-05, + "loss": 0.0313, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01577185671776533, + "rkd_loss_weight": 0.8865128874778747, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3731997817754745, + "rkd_teacher_entropy": 3.69508159160614, + "rkd_teacher_similarity_mean": 0.7694773077964783, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3410 + }, + { + "epoch": 0.9950538260110562, + "flow_matching_loss": 0.017254215478897095, + "grad_norm": 0.21065016090869904, + "learning_rate": 2.966565009762318e-05, + "loss": 0.0312, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015782902017235756, + "rkd_loss_weight": 0.8861795365810394, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3792996257543564, + "rkd_teacher_entropy": 3.7304594993591307, + "rkd_teacher_similarity_mean": 0.778365021944046, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3420 + }, + { + "epoch": 0.9979633401221996, + "flow_matching_loss": 0.021506056282669307, + "grad_norm": 0.24577680230140686, + "learning_rate": 2.966216956532375e-05, + "loss": 0.0355, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015753882378339766, + "rkd_loss_weight": 0.8858461976051331, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3710776060819626, + "rkd_teacher_entropy": 3.697306489944458, + "rkd_teacher_similarity_mean": 0.7681488394737244, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3430 + }, + { + "epoch": 1.000872854233343, + "flow_matching_loss": 0.023907487280666828, + "grad_norm": 0.1630074679851532, + "learning_rate": 2.9658671217105744e-05, + "loss": 0.043, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.02159448079764843, + "rkd_loss_weight": 0.8855128526687622, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3811010420322418, + "rkd_teacher_entropy": 3.7248361110687256, + "rkd_teacher_similarity_mean": 0.7769972085952759, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3440 + }, + { + "epoch": 1.0037823683444864, + "flow_matching_loss": 0.02115125861018896, + "grad_norm": 0.16992346942424774, + "learning_rate": 2.965515505721999e-05, + "loss": 0.0365, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017288437485694884, + "rkd_loss_weight": 0.8851795017719268, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3725295841693878, + "rkd_teacher_entropy": 3.707169198989868, + "rkd_teacher_similarity_mean": 0.7708327174186707, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3450 + }, + { + "epoch": 1.00669188245563, + "flow_matching_loss": 0.022746523097157478, + "grad_norm": 0.24638360738754272, + "learning_rate": 2.9651621089938948e-05, + "loss": 0.0371, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01619425667449832, + "rkd_loss_weight": 0.884846168756485, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37470753490924835, + "rkd_teacher_entropy": 3.721055579185486, + "rkd_teacher_similarity_mean": 0.7764740943908691, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3460 + }, + { + "epoch": 1.0096013965667734, + "flow_matching_loss": 0.019275028631091116, + "grad_norm": 0.34525105357170105, + "learning_rate": 2.9648069319556725e-05, + "loss": 0.0361, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0190736654214561, + "rkd_loss_weight": 0.8845128118991852, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37264293134212495, + "rkd_teacher_entropy": 3.7019579887390135, + "rkd_teacher_similarity_mean": 0.770649379491806, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3470 + }, + { + "epoch": 1.0125109106779169, + "flow_matching_loss": 0.020681661367416383, + "grad_norm": 0.20416034758090973, + "learning_rate": 2.9644499750389057e-05, + "loss": 0.0324, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01323088612407446, + "rkd_loss_weight": 0.8841794729232788, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37440618574619294, + "rkd_teacher_entropy": 3.7018069505691527, + "rkd_teacher_similarity_mean": 0.7698211967945099, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3480 + }, + { + "epoch": 1.0154204247890601, + "flow_matching_loss": 0.019071735348552464, + "grad_norm": 0.22457407414913177, + "learning_rate": 2.9640912386773305e-05, + "loss": 0.0317, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01429501618258655, + "rkd_loss_weight": 0.8838461339473724, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3776111304759979, + "rkd_teacher_entropy": 3.7210329294204714, + "rkd_teacher_similarity_mean": 0.7736182391643525, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3490 + }, + { + "epoch": 1.0183299389002036, + "flow_matching_loss": 0.019966741371899844, + "grad_norm": 0.21742022037506104, + "learning_rate": 2.9637307233068455e-05, + "loss": 0.0346, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01659308737143874, + "rkd_loss_weight": 0.8835127770900726, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3795544594526291, + "rkd_teacher_entropy": 3.6997851133346558, + "rkd_teacher_similarity_mean": 0.7741523206233978, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3500 + }, + { + "epoch": 1.021239453011347, + "flow_matching_loss": 0.019649996515363456, + "grad_norm": 0.1985870748758316, + "learning_rate": 2.9633684293655108e-05, + "loss": 0.0337, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01594329606741667, + "rkd_loss_weight": 0.8831794440746308, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37009976506233216, + "rkd_teacher_entropy": 3.6811540365219115, + "rkd_teacher_similarity_mean": 0.7645905613899231, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3510 + }, + { + "epoch": 1.0241489671224906, + "flow_matching_loss": 0.021244600415229797, + "grad_norm": 0.1768382340669632, + "learning_rate": 2.963004357293548e-05, + "loss": 0.0339, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014317583665251732, + "rkd_loss_weight": 0.8828460931777954, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38013501167297364, + "rkd_teacher_entropy": 3.7123718738555906, + "rkd_teacher_similarity_mean": 0.7759054601192474, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3520 + }, + { + "epoch": 1.027058481233634, + "flow_matching_loss": 0.02226080223917961, + "grad_norm": 0.20960861444473267, + "learning_rate": 2.962638507533339e-05, + "loss": 0.0359, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01540805185213685, + "rkd_loss_weight": 0.8825127482414246, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38037073612213135, + "rkd_teacher_entropy": 3.728463912010193, + "rkd_teacher_similarity_mean": 0.7757183194160462, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3530 + }, + { + "epoch": 1.0299679953447773, + "flow_matching_loss": 0.01991554917767644, + "grad_norm": 0.2237931340932846, + "learning_rate": 2.962270880529425e-05, + "loss": 0.0348, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016903635673224927, + "rkd_loss_weight": 0.8821794092655182, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3936391890048981, + "rkd_teacher_entropy": 3.7700990200042725, + "rkd_teacher_similarity_mean": 0.78894322514534, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3540 + }, + { + "epoch": 1.0328775094559208, + "flow_matching_loss": 0.021216181572526693, + "grad_norm": 0.21400316059589386, + "learning_rate": 2.9619014767285084e-05, + "loss": 0.0355, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016156243346631526, + "rkd_loss_weight": 0.8818460583686829, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.371984001994133, + "rkd_teacher_entropy": 3.699299955368042, + "rkd_teacher_similarity_mean": 0.7702839016914368, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3550 + }, + { + "epoch": 1.0357870235670643, + "flow_matching_loss": 0.0209825050085783, + "grad_norm": 0.22254513204097748, + "learning_rate": 2.9615302965794488e-05, + "loss": 0.037, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.018122368212789296, + "rkd_loss_weight": 0.8815127193927765, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38305011093616487, + "rkd_teacher_entropy": 3.752386736869812, + "rkd_teacher_similarity_mean": 0.7832138359546661, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3560 + }, + { + "epoch": 1.0386965376782078, + "flow_matching_loss": 0.02128041423857212, + "grad_norm": 0.18414464592933655, + "learning_rate": 2.961157340533266e-05, + "loss": 0.0346, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01515639191493392, + "rkd_loss_weight": 0.8811793744564056, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3880684942007065, + "rkd_teacher_entropy": 3.7474391937255858, + "rkd_teacher_similarity_mean": 0.7852696716785431, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3570 + }, + { + "epoch": 1.0416060517893513, + "flow_matching_loss": 0.022826997842639685, + "grad_norm": 0.1851673722267151, + "learning_rate": 2.9607826090431364e-05, + "loss": 0.0362, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015206352341920137, + "rkd_loss_weight": 0.8808460235595703, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3791158676147461, + "rkd_teacher_entropy": 3.737602949142456, + "rkd_teacher_similarity_mean": 0.7772335112094879, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3580 + }, + { + "epoch": 1.0445155659004945, + "flow_matching_loss": 0.02079937160015106, + "grad_norm": 0.22498026490211487, + "learning_rate": 2.9604061025643936e-05, + "loss": 0.0375, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.018979437369853257, + "rkd_loss_weight": 0.8805126845836639, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3708818882703781, + "rkd_teacher_entropy": 3.690088653564453, + "rkd_teacher_similarity_mean": 0.7653691470623016, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3590 + }, + { + "epoch": 1.047425080011638, + "flow_matching_loss": 0.02136858906596899, + "grad_norm": 0.2160467952489853, + "learning_rate": 2.9600278215545292e-05, + "loss": 0.0353, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01585658909752965, + "rkd_loss_weight": 0.8801793396472931, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38132860958576204, + "rkd_teacher_entropy": 3.7357521057128906, + "rkd_teacher_similarity_mean": 0.7783149600028991, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3600 + }, + { + "epoch": 1.0503345941227815, + "flow_matching_loss": 0.01921774186193943, + "grad_norm": 0.20873971283435822, + "learning_rate": 2.95964776647319e-05, + "loss": 0.0307, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01301453886553645, + "rkd_loss_weight": 0.8798459947109223, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37746126353740694, + "rkd_teacher_entropy": 3.718003034591675, + "rkd_teacher_similarity_mean": 0.7751201093196869, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3610 + }, + { + "epoch": 1.053244108233925, + "flow_matching_loss": 0.018984951730817556, + "grad_norm": 0.21010220050811768, + "learning_rate": 2.9592659377821787e-05, + "loss": 0.0309, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013554212637245655, + "rkd_loss_weight": 0.8795126497745513, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3779489815235138, + "rkd_teacher_entropy": 3.7184102535247803, + "rkd_teacher_similarity_mean": 0.7754595935344696, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3620 + }, + { + "epoch": 1.0561536223450685, + "flow_matching_loss": 0.018644921574741603, + "grad_norm": 0.2441609501838684, + "learning_rate": 2.9588823359454528e-05, + "loss": 0.0345, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.018019397649914025, + "rkd_loss_weight": 0.8791793048381805, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3752518504858017, + "rkd_teacher_entropy": 3.708041262626648, + "rkd_teacher_similarity_mean": 0.7712159276008606, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3630 + }, + { + "epoch": 1.0590631364562118, + "flow_matching_loss": 0.019953274074941872, + "grad_norm": 0.19330982863903046, + "learning_rate": 2.958496961429126e-05, + "loss": 0.0329, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014682445023208857, + "rkd_loss_weight": 0.8788459599018097, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37611511051654817, + "rkd_teacher_entropy": 3.7057165145874023, + "rkd_teacher_similarity_mean": 0.7717623233795166, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3640 + }, + { + "epoch": 1.0619726505673552, + "flow_matching_loss": 0.021998417004942893, + "grad_norm": 0.2747575342655182, + "learning_rate": 2.9581098147014628e-05, + "loss": 0.0365, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016506341565400363, + "rkd_loss_weight": 0.8785126209259033, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37180737853050233, + "rkd_teacher_entropy": 3.7001866579055784, + "rkd_teacher_similarity_mean": 0.7694468259811401, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3650 + }, + { + "epoch": 1.0648821646784987, + "flow_matching_loss": 0.020624257065355776, + "grad_norm": 0.24998891353607178, + "learning_rate": 2.9577208962328843e-05, + "loss": 0.0336, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014754900895059108, + "rkd_loss_weight": 0.878179270029068, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3707940250635147, + "rkd_teacher_entropy": 3.705404043197632, + "rkd_teacher_similarity_mean": 0.7687825798988343, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3660 + }, + { + "epoch": 1.0677916787896422, + "flow_matching_loss": 0.020340539049357176, + "grad_norm": 0.20437556505203247, + "learning_rate": 2.9573302064959633e-05, + "loss": 0.0376, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.019679914973676205, + "rkd_loss_weight": 0.8778459250926971, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37737413942813874, + "rkd_teacher_entropy": 3.709873676300049, + "rkd_teacher_similarity_mean": 0.7727562367916108, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3670 + }, + { + "epoch": 1.0707011929007855, + "flow_matching_loss": 0.018933771830052137, + "grad_norm": 0.27800843119621277, + "learning_rate": 2.9569377459654235e-05, + "loss": 0.0345, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01774178072810173, + "rkd_loss_weight": 0.8775125861167907, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3745639741420746, + "rkd_teacher_entropy": 3.70768141746521, + "rkd_teacher_similarity_mean": 0.7694581568241119, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3680 + }, + { + "epoch": 1.073610707011929, + "flow_matching_loss": 0.02110766591504216, + "grad_norm": 0.37769174575805664, + "learning_rate": 2.9565435151181424e-05, + "loss": 0.0369, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01800797013565898, + "rkd_loss_weight": 0.8771792352199554, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3769204646348953, + "rkd_teacher_entropy": 3.714835214614868, + "rkd_teacher_similarity_mean": 0.7738233387470246, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3690 + }, + { + "epoch": 1.0765202211230724, + "flow_matching_loss": 0.019519795011729003, + "grad_norm": 0.3955651819705963, + "learning_rate": 2.9561475144331473e-05, + "loss": 0.0333, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01570251155644655, + "rkd_loss_weight": 0.8768458962440491, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37705683410167695, + "rkd_teacher_entropy": 3.6981968879699707, + "rkd_teacher_similarity_mean": 0.7720548808574677, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3700 + }, + { + "epoch": 1.079429735234216, + "flow_matching_loss": 0.02082753600552678, + "grad_norm": 0.18521493673324585, + "learning_rate": 2.9557497443916162e-05, + "loss": 0.0337, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01471776096150279, + "rkd_loss_weight": 0.8765125513076782, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3761585414409637, + "rkd_teacher_entropy": 3.7074289560317992, + "rkd_teacher_similarity_mean": 0.7706235826015473, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3710 + }, + { + "epoch": 1.0823392493453594, + "flow_matching_loss": 0.022006253246217966, + "grad_norm": 0.20209065079689026, + "learning_rate": 2.9553502054768774e-05, + "loss": 0.0368, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016847761627286673, + "rkd_loss_weight": 0.8761792004108429, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37483565509319305, + "rkd_teacher_entropy": 3.7200506925582886, + "rkd_teacher_similarity_mean": 0.7740749537944793, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3720 + }, + { + "epoch": 1.0852487634565027, + "flow_matching_loss": 0.020892789121717215, + "grad_norm": 0.26355311274528503, + "learning_rate": 2.954948898174409e-05, + "loss": 0.0353, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01639304282143712, + "rkd_loss_weight": 0.875845867395401, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3855020523071289, + "rkd_teacher_entropy": 3.738950753211975, + "rkd_teacher_similarity_mean": 0.7815780401229858, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3730 + }, + { + "epoch": 1.0881582775676462, + "flow_matching_loss": 0.020357936434447766, + "grad_norm": 0.20825594663619995, + "learning_rate": 2.954545822971836e-05, + "loss": 0.0357, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017565327417105436, + "rkd_loss_weight": 0.8755125105381012, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3761936455965042, + "rkd_teacher_entropy": 3.7230560064315794, + "rkd_teacher_similarity_mean": 0.77378009557724, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3740 + }, + { + "epoch": 1.0910677916787896, + "flow_matching_loss": 0.022078982088714838, + "grad_norm": 0.2226894199848175, + "learning_rate": 2.9541409803589335e-05, + "loss": 0.0391, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.019419978745281695, + "rkd_loss_weight": 0.8751791715621948, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3739885717630386, + "rkd_teacher_entropy": 3.7033994674682615, + "rkd_teacher_similarity_mean": 0.7716209053993225, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3750 + }, + { + "epoch": 1.0939773057899331, + "flow_matching_loss": 0.021815265715122222, + "grad_norm": 0.2638004422187805, + "learning_rate": 2.9537343708276235e-05, + "loss": 0.0365, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01682590236887336, + "rkd_loss_weight": 0.8748458325862885, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3711680293083191, + "rkd_teacher_entropy": 3.7105440378189085, + "rkd_teacher_similarity_mean": 0.7720076501369476, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3760 + }, + { + "epoch": 1.0968868199010766, + "flow_matching_loss": 0.019695974234491587, + "grad_norm": 0.22639968991279602, + "learning_rate": 2.953325994871975e-05, + "loss": 0.0318, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013805981539189816, + "rkd_loss_weight": 0.8745124816894532, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38460024893283845, + "rkd_teacher_entropy": 3.7508121013641356, + "rkd_teacher_similarity_mean": 0.7826421022415161, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3770 + }, + { + "epoch": 1.0997963340122199, + "flow_matching_loss": 0.018856427259743215, + "grad_norm": 0.1861996054649353, + "learning_rate": 2.9529158529882023e-05, + "loss": 0.0339, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01721214847639203, + "rkd_loss_weight": 0.8741791427135468, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3773783385753632, + "rkd_teacher_entropy": 3.7050771474838258, + "rkd_teacher_similarity_mean": 0.7715902209281922, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3780 + }, + { + "epoch": 1.1027058481233634, + "flow_matching_loss": 0.020532514434307814, + "grad_norm": 0.18965275585651398, + "learning_rate": 2.9525039456746677e-05, + "loss": 0.0352, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016838926076889037, + "rkd_loss_weight": 0.8738457977771759, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3740159660577774, + "rkd_teacher_entropy": 3.6996373414993284, + "rkd_teacher_similarity_mean": 0.7734486222267151, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3790 + }, + { + "epoch": 1.1056153622345068, + "flow_matching_loss": 0.02172482740134001, + "grad_norm": 0.21370670199394226, + "learning_rate": 2.952090273431877e-05, + "loss": 0.0357, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01597529025748372, + "rkd_loss_weight": 0.8735124468803406, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3737410485744476, + "rkd_teacher_entropy": 3.7106195211410524, + "rkd_teacher_similarity_mean": 0.7709264636039734, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3800 + }, + { + "epoch": 1.1085248763456503, + "flow_matching_loss": 0.018372091744095088, + "grad_norm": 0.16131268441677094, + "learning_rate": 2.9516748367624808e-05, + "loss": 0.034, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01787356473505497, + "rkd_loss_weight": 0.8731791079044342, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3669866114854813, + "rkd_teacher_entropy": 3.6962432146072386, + "rkd_teacher_similarity_mean": 0.7638638079166412, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3810 + }, + { + "epoch": 1.1114343904567938, + "flow_matching_loss": 0.019617373123764992, + "grad_norm": 0.25405311584472656, + "learning_rate": 2.951257636171274e-05, + "loss": 0.0341, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016538580041378738, + "rkd_loss_weight": 0.8728457629680634, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3728520184755325, + "rkd_teacher_entropy": 3.708735728263855, + "rkd_teacher_similarity_mean": 0.7716537833213806, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3820 + }, + { + "epoch": 1.114343904567937, + "flow_matching_loss": 0.020041238516569138, + "grad_norm": 0.2711043953895569, + "learning_rate": 2.9508386721651944e-05, + "loss": 0.0352, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017419697903096675, + "rkd_loss_weight": 0.8725124180316925, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37778967022895815, + "rkd_teacher_entropy": 3.7174421548843384, + "rkd_teacher_similarity_mean": 0.77586989402771, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3830 + }, + { + "epoch": 1.1172534186790806, + "flow_matching_loss": 0.02020737947896123, + "grad_norm": 0.22628584504127502, + "learning_rate": 2.9504179452533228e-05, + "loss": 0.033, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014698131382465363, + "rkd_loss_weight": 0.8721790730953216, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.372541144490242, + "rkd_teacher_entropy": 3.7103185176849367, + "rkd_teacher_similarity_mean": 0.7707604110240937, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3840 + }, + { + "epoch": 1.120162932790224, + "flow_matching_loss": 0.018706713896244766, + "grad_norm": 0.24994246661663055, + "learning_rate": 2.9499954559468826e-05, + "loss": 0.0345, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.018078646436333657, + "rkd_loss_weight": 0.8718457281589508, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37532668113708495, + "rkd_teacher_entropy": 3.7248347282409666, + "rkd_teacher_similarity_mean": 0.7766251504421234, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3850 + }, + { + "epoch": 1.1230724469013675, + "flow_matching_loss": 0.019930611085146665, + "grad_norm": 0.2202467918395996, + "learning_rate": 2.9495712047592378e-05, + "loss": 0.0345, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01669225413352251, + "rkd_loss_weight": 0.87151238322258, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36539641618728635, + "rkd_teacher_entropy": 3.697703242301941, + "rkd_teacher_similarity_mean": 0.7679303526878357, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3860 + }, + { + "epoch": 1.1259819610125108, + "flow_matching_loss": 0.021236061304807662, + "grad_norm": 0.19473133981227875, + "learning_rate": 2.949145192205893e-05, + "loss": 0.0352, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01607605703175068, + "rkd_loss_weight": 0.8711790442466736, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3788133472204208, + "rkd_teacher_entropy": 3.724376392364502, + "rkd_teacher_similarity_mean": 0.777026242017746, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3870 + }, + { + "epoch": 1.1288914751236543, + "flow_matching_loss": 0.01867344370111823, + "grad_norm": 0.17704571783542633, + "learning_rate": 2.9487174188044953e-05, + "loss": 0.0299, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012926889955997467, + "rkd_loss_weight": 0.8708456933498383, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37639225125312803, + "rkd_teacher_entropy": 3.7123181581497193, + "rkd_teacher_similarity_mean": 0.7751086115837097, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3880 + }, + { + "epoch": 1.1318009892347978, + "flow_matching_loss": 0.018718103505671026, + "grad_norm": 0.2581542730331421, + "learning_rate": 2.9482878850748276e-05, + "loss": 0.0331, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01647118842229247, + "rkd_loss_weight": 0.8705123484134674, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37333810031414033, + "rkd_teacher_entropy": 3.713718819618225, + "rkd_teacher_similarity_mean": 0.7705463469028473, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3890 + }, + { + "epoch": 1.1347105033459413, + "flow_matching_loss": 0.020084941200911997, + "grad_norm": 0.27005696296691895, + "learning_rate": 2.9478565915388157e-05, + "loss": 0.0351, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017216877080500126, + "rkd_loss_weight": 0.870179009437561, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.376643905043602, + "rkd_teacher_entropy": 3.7151729345321653, + "rkd_teacher_similarity_mean": 0.7722509443759918, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3900 + }, + { + "epoch": 1.1376200174570847, + "flow_matching_loss": 0.022271097637712955, + "grad_norm": 0.19109822809696198, + "learning_rate": 2.9474235387205207e-05, + "loss": 0.0339, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013359248638153076, + "rkd_loss_weight": 0.8698456585407257, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3723158687353134, + "rkd_teacher_entropy": 3.7076321125030516, + "rkd_teacher_similarity_mean": 0.7682428359985352, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3910 + }, + { + "epoch": 1.140529531568228, + "flow_matching_loss": 0.02223889110609889, + "grad_norm": 0.19434277713298798, + "learning_rate": 2.946988727146143e-05, + "loss": 0.0355, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015295556839555502, + "rkd_loss_weight": 0.8695123195648193, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3717121809720993, + "rkd_teacher_entropy": 3.6854774951934814, + "rkd_teacher_similarity_mean": 0.7661900758743286, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3920 + }, + { + "epoch": 1.1434390456793715, + "flow_matching_loss": 0.01749715143814683, + "grad_norm": 0.3068516254425049, + "learning_rate": 2.9465521573440196e-05, + "loss": 0.0291, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013379123900085688, + "rkd_loss_weight": 0.8691789746284485, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3766183823347092, + "rkd_teacher_entropy": 3.7093430757522583, + "rkd_teacher_similarity_mean": 0.7721340715885162, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3930 + }, + { + "epoch": 1.146348559790515, + "flow_matching_loss": 0.020015545189380646, + "grad_norm": 0.3251587748527527, + "learning_rate": 2.946113829844624e-05, + "loss": 0.0353, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017590552382171155, + "rkd_loss_weight": 0.8688456237316131, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3802037537097931, + "rkd_teacher_entropy": 3.7177348375320434, + "rkd_teacher_similarity_mean": 0.7750738143920899, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3940 + }, + { + "epoch": 1.1492580739016585, + "flow_matching_loss": 0.018447738699615003, + "grad_norm": 0.1991017460823059, + "learning_rate": 2.945673745180566e-05, + "loss": 0.0311, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01456941431388259, + "rkd_loss_weight": 0.8685122907161713, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3758204609155655, + "rkd_teacher_entropy": 3.6953845739364626, + "rkd_teacher_similarity_mean": 0.7706306397914886, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3950 + }, + { + "epoch": 1.152167588012802, + "flow_matching_loss": 0.021070570778101684, + "grad_norm": 0.20565539598464966, + "learning_rate": 2.945231903886589e-05, + "loss": 0.0366, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017848302237689496, + "rkd_loss_weight": 0.8681789338588715, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3816089600324631, + "rkd_teacher_entropy": 3.7155566215515137, + "rkd_teacher_similarity_mean": 0.7770087063312531, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3960 + }, + { + "epoch": 1.1550771021239452, + "flow_matching_loss": 0.020424645207822324, + "grad_norm": 0.26689383387565613, + "learning_rate": 2.9447883064995728e-05, + "loss": 0.0354, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017272952292114497, + "rkd_loss_weight": 0.8678455948829651, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37148923277854917, + "rkd_teacher_entropy": 3.6940998315811155, + "rkd_teacher_similarity_mean": 0.7675588369369507, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3970 + }, + { + "epoch": 1.1579866162350887, + "flow_matching_loss": 0.024431699141860008, + "grad_norm": 0.2788066565990448, + "learning_rate": 2.9443429535585288e-05, + "loss": 0.0382, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01591830337420106, + "rkd_loss_weight": 0.8675122559070587, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37948840856552124, + "rkd_teacher_entropy": 3.723736619949341, + "rkd_teacher_similarity_mean": 0.7753281474113465, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3980 + }, + { + "epoch": 1.1608961303462322, + "flow_matching_loss": 0.024061963707208634, + "grad_norm": 0.22554168105125427, + "learning_rate": 2.9438958456046044e-05, + "loss": 0.036, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013816832005977631, + "rkd_loss_weight": 0.8671788990497589, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37443905770778657, + "rkd_teacher_entropy": 3.709714484214783, + "rkd_teacher_similarity_mean": 0.7723363876342774, + "rkd_vlm_temp": 0.10000000149011612, + "step": 3990 + }, + { + "epoch": 1.1638056444573757, + "flow_matching_loss": 0.020089432410895825, + "grad_norm": 0.19009467959403992, + "learning_rate": 2.9434469831810765e-05, + "loss": 0.0322, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013948123436421156, + "rkd_loss_weight": 0.8668455660343171, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3821175843477249, + "rkd_teacher_entropy": 3.733018183708191, + "rkd_teacher_similarity_mean": 0.7808993339538575, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4000 + }, + { + "epoch": 1.1667151585685192, + "flow_matching_loss": 0.02142559913918376, + "grad_norm": 0.272401362657547, + "learning_rate": 2.9429963668333566e-05, + "loss": 0.0336, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014029685966670513, + "rkd_loss_weight": 0.8665122151374817, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3736222118139267, + "rkd_teacher_entropy": 3.7085533618927, + "rkd_teacher_similarity_mean": 0.7723472774028778, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4010 + }, + { + "epoch": 1.1696246726796624, + "flow_matching_loss": 0.019151077978312968, + "grad_norm": 0.19788187742233276, + "learning_rate": 2.9425439971089845e-05, + "loss": 0.0303, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012876258697360754, + "rkd_loss_weight": 0.8661788702011108, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3706649988889694, + "rkd_teacher_entropy": 3.710145139694214, + "rkd_teacher_similarity_mean": 0.7700979948043823, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4020 + }, + { + "epoch": 1.172534186790806, + "flow_matching_loss": 0.019108899030834438, + "grad_norm": 0.22630977630615234, + "learning_rate": 2.9420898745576336e-05, + "loss": 0.0308, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013455904461443425, + "rkd_loss_weight": 0.8658455312252045, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3785705238580704, + "rkd_teacher_entropy": 3.725653386116028, + "rkd_teacher_similarity_mean": 0.7771662473678589, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4030 + }, + { + "epoch": 1.1754437009019494, + "flow_matching_loss": 0.019146770425140857, + "grad_norm": 0.2354799062013626, + "learning_rate": 2.9416339997311044e-05, + "loss": 0.0344, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01761604016646743, + "rkd_loss_weight": 0.8655121803283692, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3845237523317337, + "rkd_teacher_entropy": 3.739044737815857, + "rkd_teacher_similarity_mean": 0.7824091970920563, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4040 + }, + { + "epoch": 1.1783532150130929, + "flow_matching_loss": 0.017757477518171073, + "grad_norm": 0.2205667942762375, + "learning_rate": 2.941176373183329e-05, + "loss": 0.0303, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014465859439224005, + "rkd_loss_weight": 0.8651788413524628, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37648465037345885, + "rkd_teacher_entropy": 3.7029669523239135, + "rkd_teacher_similarity_mean": 0.7706955015659332, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4050 + }, + { + "epoch": 1.1812627291242364, + "flow_matching_loss": 0.020292120333760977, + "grad_norm": 0.3729388117790222, + "learning_rate": 2.9407169954703653e-05, + "loss": 0.0331, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01476273899897933, + "rkd_loss_weight": 0.8648454964160919, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3821453988552094, + "rkd_teacher_entropy": 3.725294351577759, + "rkd_teacher_similarity_mean": 0.7782662034034729, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4060 + }, + { + "epoch": 1.1841722432353796, + "flow_matching_loss": 0.019568940717726947, + "grad_norm": 0.18414387106895447, + "learning_rate": 2.9402558671504018e-05, + "loss": 0.0328, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015333972219377757, + "rkd_loss_weight": 0.8645121455192566, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37356777787208556, + "rkd_teacher_entropy": 3.7102524995803834, + "rkd_teacher_similarity_mean": 0.7721898078918457, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4070 + }, + { + "epoch": 1.187081757346523, + "flow_matching_loss": 0.020072111487388612, + "grad_norm": 0.17985081672668457, + "learning_rate": 2.9397929887837524e-05, + "loss": 0.0335, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015588747616857291, + "rkd_loss_weight": 0.8641788065433502, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38880142867565154, + "rkd_teacher_entropy": 3.741062045097351, + "rkd_teacher_similarity_mean": 0.7833883225917816, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4080 + }, + { + "epoch": 1.1899912714576666, + "flow_matching_loss": 0.01968375165015459, + "grad_norm": 0.3235985338687897, + "learning_rate": 2.939328360932858e-05, + "loss": 0.0333, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01574700865894556, + "rkd_loss_weight": 0.8638454616069794, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37538603842258456, + "rkd_teacher_entropy": 3.713528299331665, + "rkd_teacher_similarity_mean": 0.7721310913562774, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4090 + }, + { + "epoch": 1.19290078556881, + "flow_matching_loss": 0.019684395752847195, + "grad_norm": 0.2097564935684204, + "learning_rate": 2.938861984162285e-05, + "loss": 0.0335, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015944771096110345, + "rkd_loss_weight": 0.8635121166706086, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37785992920398714, + "rkd_teacher_entropy": 3.728734564781189, + "rkd_teacher_similarity_mean": 0.7775693714618683, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4100 + }, + { + "epoch": 1.1958102996799536, + "flow_matching_loss": 0.020138243027031422, + "grad_norm": 0.15489701926708221, + "learning_rate": 2.938393859038726e-05, + "loss": 0.0333, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015204781293869018, + "rkd_loss_weight": 0.8631787717342376, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3770127981901169, + "rkd_teacher_entropy": 3.7247455596923826, + "rkd_teacher_similarity_mean": 0.7766978859901428, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4110 + }, + { + "epoch": 1.1987198137910968, + "flow_matching_loss": 0.01980291623622179, + "grad_norm": 0.15830202400684357, + "learning_rate": 2.9379239861309964e-05, + "loss": 0.0336, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016046002693474292, + "rkd_loss_weight": 0.8628454267978668, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38003878593444823, + "rkd_teacher_entropy": 3.7267027854919434, + "rkd_teacher_similarity_mean": 0.7800451278686523, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4120 + }, + { + "epoch": 1.2016293279022403, + "flow_matching_loss": 0.017358965147286652, + "grad_norm": 0.1735105663537979, + "learning_rate": 2.937452366010036e-05, + "loss": 0.0293, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013791413139551878, + "rkd_loss_weight": 0.862512081861496, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3820036739110947, + "rkd_teacher_entropy": 3.738743305206299, + "rkd_teacher_similarity_mean": 0.7818490028381347, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4130 + }, + { + "epoch": 1.2045388420133838, + "flow_matching_loss": 0.02284441199153662, + "grad_norm": 0.17643864452838898, + "learning_rate": 2.936978999248908e-05, + "loss": 0.0343, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013238626439124346, + "rkd_loss_weight": 0.8621787428855896, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3723410278558731, + "rkd_teacher_entropy": 3.712397313117981, + "rkd_teacher_similarity_mean": 0.7719934284687042, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4140 + }, + { + "epoch": 1.2074483561245273, + "flow_matching_loss": 0.020541212055832148, + "grad_norm": 0.3468073904514313, + "learning_rate": 2.936503886422798e-05, + "loss": 0.0346, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016334587894380092, + "rkd_loss_weight": 0.8618453919887543, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3713737577199936, + "rkd_teacher_entropy": 3.702675485610962, + "rkd_teacher_similarity_mean": 0.7693481802940368, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4150 + }, + { + "epoch": 1.2103578702356708, + "flow_matching_loss": 0.02138378992676735, + "grad_norm": 0.1934867650270462, + "learning_rate": 2.9360270281090122e-05, + "loss": 0.0336, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014212072174996138, + "rkd_loss_weight": 0.8615120470523834, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3816572278738022, + "rkd_teacher_entropy": 3.735317850112915, + "rkd_teacher_similarity_mean": 0.7796253263950348, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4160 + }, + { + "epoch": 1.213267384346814, + "flow_matching_loss": 0.020467781461775303, + "grad_norm": 0.22537721693515778, + "learning_rate": 2.935548424886979e-05, + "loss": 0.0321, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01349943969398737, + "rkd_loss_weight": 0.861178708076477, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37888910472393034, + "rkd_teacher_entropy": 3.72514750957489, + "rkd_teacher_similarity_mean": 0.7768618583679199, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4170 + }, + { + "epoch": 1.2161768984579575, + "flow_matching_loss": 0.019406792242079973, + "grad_norm": 0.24269062280654907, + "learning_rate": 2.935068077338246e-05, + "loss": 0.0319, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0145069545134902, + "rkd_loss_weight": 0.8608453571796417, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38638884723186495, + "rkd_teacher_entropy": 3.7489392518997193, + "rkd_teacher_similarity_mean": 0.7849957168102264, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4180 + }, + { + "epoch": 1.219086412569101, + "flow_matching_loss": 0.018850005138665436, + "grad_norm": 0.1666533648967743, + "learning_rate": 2.934585986046481e-05, + "loss": 0.0337, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017301690950989723, + "rkd_loss_weight": 0.8605120182037354, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37685911953449247, + "rkd_teacher_entropy": 3.7228496789932253, + "rkd_teacher_similarity_mean": 0.7747901320457459, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4190 + }, + { + "epoch": 1.2219959266802445, + "flow_matching_loss": 0.01834499593824148, + "grad_norm": 0.203568696975708, + "learning_rate": 2.93410215159747e-05, + "loss": 0.0346, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01894211145117879, + "rkd_loss_weight": 0.860178679227829, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3750350117683411, + "rkd_teacher_entropy": 3.7127312898635862, + "rkd_teacher_similarity_mean": 0.7750449478626251, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4200 + }, + { + "epoch": 1.2249054407913877, + "flow_matching_loss": 0.018870460987091064, + "grad_norm": 0.2668931782245636, + "learning_rate": 2.933616574579118e-05, + "loss": 0.0317, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014866839163005352, + "rkd_loss_weight": 0.8598453223705291, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3802117794752121, + "rkd_teacher_entropy": 3.740868163108826, + "rkd_teacher_similarity_mean": 0.7807983517646789, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4210 + }, + { + "epoch": 1.2278149549025312, + "flow_matching_loss": 0.018270390667021276, + "grad_norm": 0.21648187935352325, + "learning_rate": 2.933129255581447e-05, + "loss": 0.0334, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017572143767029046, + "rkd_loss_weight": 0.8595119893550873, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37587508261203767, + "rkd_teacher_entropy": 3.726883387565613, + "rkd_teacher_similarity_mean": 0.7773212313652038, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4220 + }, + { + "epoch": 1.2307244690136747, + "flow_matching_loss": 0.018378487322479486, + "grad_norm": 0.23585492372512817, + "learning_rate": 2.9326401951965954e-05, + "loss": 0.0322, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016118996404111387, + "rkd_loss_weight": 0.8591786384582519, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3770058751106262, + "rkd_teacher_entropy": 3.7162421226501463, + "rkd_teacher_similarity_mean": 0.7747301816940307, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4230 + }, + { + "epoch": 1.2336339831248182, + "flow_matching_loss": 0.02074527470394969, + "grad_norm": 0.20908603072166443, + "learning_rate": 2.9321493940188165e-05, + "loss": 0.0327, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013880273792892694, + "rkd_loss_weight": 0.8588452935218811, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38513397574424746, + "rkd_teacher_entropy": 3.749261474609375, + "rkd_teacher_similarity_mean": 0.7849096179008483, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4240 + }, + { + "epoch": 1.2365434972359617, + "flow_matching_loss": 0.01881406707689166, + "grad_norm": 0.2779223322868347, + "learning_rate": 2.931656852644482e-05, + "loss": 0.0303, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013387797446921468, + "rkd_loss_weight": 0.8585119545459747, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37849338054656984, + "rkd_teacher_entropy": 3.718817949295044, + "rkd_teacher_similarity_mean": 0.777393388748169, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4250 + }, + { + "epoch": 1.239453011347105, + "flow_matching_loss": 0.019073211029171942, + "grad_norm": 0.20375315845012665, + "learning_rate": 2.9311625716720748e-05, + "loss": 0.032, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015051155164837838, + "rkd_loss_weight": 0.8581786036491394, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3780130684375763, + "rkd_teacher_entropy": 3.7113848686218263, + "rkd_teacher_similarity_mean": 0.7749439239501953, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4260 + }, + { + "epoch": 1.2423625254582484, + "flow_matching_loss": 0.016787401959300043, + "grad_norm": 0.21586480736732483, + "learning_rate": 2.9306665517021927e-05, + "loss": 0.0301, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015540078002959489, + "rkd_loss_weight": 0.8578452646732331, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37306455373764036, + "rkd_teacher_entropy": 3.698967790603638, + "rkd_teacher_similarity_mean": 0.7714496195316315, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4270 + }, + { + "epoch": 1.245272039569392, + "flow_matching_loss": 0.019674633629620075, + "grad_norm": 0.2587047219276428, + "learning_rate": 2.9301687933375483e-05, + "loss": 0.0323, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014764299429953099, + "rkd_loss_weight": 0.8575119197368621, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3778370827436447, + "rkd_teacher_entropy": 3.7204824447631837, + "rkd_teacher_similarity_mean": 0.7761914014816285, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4280 + }, + { + "epoch": 1.2481815536805354, + "flow_matching_loss": 0.020146562438458204, + "grad_norm": 0.18092580139636993, + "learning_rate": 2.929669297182963e-05, + "loss": 0.034, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016167571675032376, + "rkd_loss_weight": 0.8571785688400269, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3795346736907959, + "rkd_teacher_entropy": 3.7323646068573, + "rkd_teacher_similarity_mean": 0.7749876260757447, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4290 + }, + { + "epoch": 1.2510910677916787, + "flow_matching_loss": 0.0203416901640594, + "grad_norm": 0.16947194933891296, + "learning_rate": 2.929168063845373e-05, + "loss": 0.032, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013630767446011305, + "rkd_loss_weight": 0.8568452298641205, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3779654622077942, + "rkd_teacher_entropy": 3.721381092071533, + "rkd_teacher_similarity_mean": 0.7736253857612609, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4300 + }, + { + "epoch": 1.2540005819028222, + "flow_matching_loss": 0.020954694226384164, + "grad_norm": 0.2185744047164917, + "learning_rate": 2.9286650939338236e-05, + "loss": 0.0364, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.018089354131370784, + "rkd_loss_weight": 0.8565118849277497, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3674547404050827, + "rkd_teacher_entropy": 3.7072983503341677, + "rkd_teacher_similarity_mean": 0.768258249759674, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4310 + }, + { + "epoch": 1.2569100960139656, + "flow_matching_loss": 0.017990300338715313, + "grad_norm": 0.2050815373659134, + "learning_rate": 2.9281603880594706e-05, + "loss": 0.0316, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01583975926041603, + "rkd_loss_weight": 0.8561785399913788, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3765593856573105, + "rkd_teacher_entropy": 3.720667767524719, + "rkd_teacher_similarity_mean": 0.7754500389099122, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4320 + }, + { + "epoch": 1.2598196101251091, + "flow_matching_loss": 0.020605135522782803, + "grad_norm": 0.36962246894836426, + "learning_rate": 2.9276539468355796e-05, + "loss": 0.0335, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01501880045980215, + "rkd_loss_weight": 0.8558451950550079, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36839568614959717, + "rkd_teacher_entropy": 3.691095566749573, + "rkd_teacher_similarity_mean": 0.7669952869415283, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4330 + }, + { + "epoch": 1.2627291242362526, + "flow_matching_loss": 0.021249614004045726, + "grad_norm": 0.26275357604026794, + "learning_rate": 2.9271457708775244e-05, + "loss": 0.0363, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017638353258371355, + "rkd_loss_weight": 0.8555118501186371, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37306385338306425, + "rkd_teacher_entropy": 3.724526381492615, + "rkd_teacher_similarity_mean": 0.773382306098938, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4340 + }, + { + "epoch": 1.2656386383473959, + "flow_matching_loss": 0.018823277857154607, + "grad_norm": 0.21637187898159027, + "learning_rate": 2.9266358608027863e-05, + "loss": 0.0307, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01384894335642457, + "rkd_loss_weight": 0.8551785051822662, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3716336339712143, + "rkd_teacher_entropy": 3.699429750442505, + "rkd_teacher_similarity_mean": 0.7685551524162293, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4350 + }, + { + "epoch": 1.2685481524585394, + "flow_matching_loss": 0.01886363383382559, + "grad_norm": 0.21911033987998962, + "learning_rate": 2.9261242172309543e-05, + "loss": 0.0329, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01642648857086897, + "rkd_loss_weight": 0.8548451662063599, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3804388612508774, + "rkd_teacher_entropy": 3.7291951179504395, + "rkd_teacher_similarity_mean": 0.7766258358955384, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4360 + }, + { + "epoch": 1.2714576665696828, + "flow_matching_loss": 0.018992546387016772, + "grad_norm": 0.19700083136558533, + "learning_rate": 2.925610840783723e-05, + "loss": 0.0302, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013157183770090342, + "rkd_loss_weight": 0.8545118153095246, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3735150247812271, + "rkd_teacher_entropy": 3.6947036266326903, + "rkd_teacher_similarity_mean": 0.7707523226737976, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4370 + }, + { + "epoch": 1.2743671806808263, + "flow_matching_loss": 0.022450137231498957, + "grad_norm": 0.19238710403442383, + "learning_rate": 2.9250957320848946e-05, + "loss": 0.0355, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015247428324073553, + "rkd_loss_weight": 0.8541784703731536, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37084453701972964, + "rkd_teacher_entropy": 3.696952986717224, + "rkd_teacher_similarity_mean": 0.7677396535873413, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4380 + }, + { + "epoch": 1.2772766947919698, + "flow_matching_loss": 0.02041304837912321, + "grad_norm": 0.19221235811710358, + "learning_rate": 2.924578891760374e-05, + "loss": 0.032, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013566946052014828, + "rkd_loss_weight": 0.8538451313972473, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3760466337203979, + "rkd_teacher_entropy": 3.7166522979736327, + "rkd_teacher_similarity_mean": 0.7739440977573395, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4390 + }, + { + "epoch": 1.280186208903113, + "flow_matching_loss": 0.021482571959495544, + "grad_norm": 0.22455161809921265, + "learning_rate": 2.924060320438171e-05, + "loss": 0.0334, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013910254649817943, + "rkd_loss_weight": 0.853511780500412, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37107549905776976, + "rkd_teacher_entropy": 3.711002969741821, + "rkd_teacher_similarity_mean": 0.7693298876285553, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4400 + }, + { + "epoch": 1.2830957230142566, + "flow_matching_loss": 0.020669337175786496, + "grad_norm": 0.17002777755260468, + "learning_rate": 2.9235400187483978e-05, + "loss": 0.0325, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01385216386988759, + "rkd_loss_weight": 0.8531784415245056, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37612099945545197, + "rkd_teacher_entropy": 3.7023067474365234, + "rkd_teacher_similarity_mean": 0.7686245501041412, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4410 + }, + { + "epoch": 1.2860052371254, + "flow_matching_loss": 0.020838850177824497, + "grad_norm": 0.23826029896736145, + "learning_rate": 2.9230179873232716e-05, + "loss": 0.0359, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017717252485454082, + "rkd_loss_weight": 0.8528450965881348, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3780520558357239, + "rkd_teacher_entropy": 3.724446654319763, + "rkd_teacher_similarity_mean": 0.7746546626091003, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4420 + }, + { + "epoch": 1.2889147512365435, + "flow_matching_loss": 0.02128158500418067, + "grad_norm": 0.29871439933776855, + "learning_rate": 2.922494226797108e-05, + "loss": 0.0345, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015522504039108754, + "rkd_loss_weight": 0.8525117456912994, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37694706320762633, + "rkd_teacher_entropy": 3.7123141288757324, + "rkd_teacher_similarity_mean": 0.7733358502388, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4430 + }, + { + "epoch": 1.291824265347687, + "flow_matching_loss": 0.01922502350062132, + "grad_norm": 0.18005406856536865, + "learning_rate": 2.9219687378063275e-05, + "loss": 0.0311, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01389224585145712, + "rkd_loss_weight": 0.8521784126758576, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3814809024333954, + "rkd_teacher_entropy": 3.7077762126922607, + "rkd_teacher_similarity_mean": 0.7758466303348541, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4440 + }, + { + "epoch": 1.2947337794588303, + "flow_matching_loss": 0.01949816271662712, + "grad_norm": 0.18269364535808563, + "learning_rate": 2.921441520989447e-05, + "loss": 0.0343, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01738750459626317, + "rkd_loss_weight": 0.8518450558185577, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38214552104473115, + "rkd_teacher_entropy": 3.714300298690796, + "rkd_teacher_similarity_mean": 0.7751463949680328, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4450 + }, + { + "epoch": 1.2976432935699738, + "flow_matching_loss": 0.018466565012931823, + "grad_norm": 0.210993692278862, + "learning_rate": 2.9209125769870854e-05, + "loss": 0.0336, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017795172426849602, + "rkd_loss_weight": 0.8515117168426514, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37424195408821104, + "rkd_teacher_entropy": 3.7020673036575316, + "rkd_teacher_similarity_mean": 0.7691540122032166, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4460 + }, + { + "epoch": 1.3005528076811173, + "flow_matching_loss": 0.018130764737725256, + "grad_norm": 0.19576773047447205, + "learning_rate": 2.92038190644196e-05, + "loss": 0.0323, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01669871574267745, + "rkd_loss_weight": 0.851178377866745, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37595327496528624, + "rkd_teacher_entropy": 3.70582914352417, + "rkd_teacher_similarity_mean": 0.7717800557613372, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4470 + }, + { + "epoch": 1.3034623217922607, + "flow_matching_loss": 0.020917263999581336, + "grad_norm": 0.2890095114707947, + "learning_rate": 2.9198495099988843e-05, + "loss": 0.0346, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016026438027620316, + "rkd_loss_weight": 0.8508450210094451, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37055588960647584, + "rkd_teacher_entropy": 3.6990477561950685, + "rkd_teacher_similarity_mean": 0.7695429205894471, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4480 + }, + { + "epoch": 1.3063718359034042, + "flow_matching_loss": 0.019930726476013662, + "grad_norm": 0.2718726694583893, + "learning_rate": 2.9193153883047712e-05, + "loss": 0.0314, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013517937902361155, + "rkd_loss_weight": 0.8505116879940033, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37521655559539796, + "rkd_teacher_entropy": 3.707182598114014, + "rkd_teacher_similarity_mean": 0.7728054225444794, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4490 + }, + { + "epoch": 1.3092813500145475, + "flow_matching_loss": 0.02037877161055803, + "grad_norm": 0.19713369011878967, + "learning_rate": 2.9187795420086287e-05, + "loss": 0.0333, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015204787719994783, + "rkd_loss_weight": 0.850178337097168, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3663002818822861, + "rkd_teacher_entropy": 3.67777361869812, + "rkd_teacher_similarity_mean": 0.7651518046855926, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4500 + }, + { + "epoch": 1.312190864125691, + "flow_matching_loss": 0.015900666266679762, + "grad_norm": 0.14872825145721436, + "learning_rate": 2.9182419717615602e-05, + "loss": 0.0297, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016252581775188447, + "rkd_loss_weight": 0.8498449921607971, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37402623891830444, + "rkd_teacher_entropy": 3.7103978872299193, + "rkd_teacher_similarity_mean": 0.7742419540882111, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4510 + }, + { + "epoch": 1.3151003782368345, + "flow_matching_loss": 0.018281987961381675, + "grad_norm": 0.1852361559867859, + "learning_rate": 2.917702678216765e-05, + "loss": 0.0329, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017233865242451428, + "rkd_loss_weight": 0.8495116531848907, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3681177645921707, + "rkd_teacher_entropy": 3.6812939405441285, + "rkd_teacher_similarity_mean": 0.7658869802951813, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4520 + }, + { + "epoch": 1.318009892347978, + "flow_matching_loss": 0.018744277954101562, + "grad_norm": 0.23518380522727966, + "learning_rate": 2.9171616620295347e-05, + "loss": 0.0323, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01596949053928256, + "rkd_loss_weight": 0.8491783022880555, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.361523249745369, + "rkd_teacher_entropy": 3.667899513244629, + "rkd_teacher_similarity_mean": 0.7631741762161255, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4530 + }, + { + "epoch": 1.3209194064591214, + "flow_matching_loss": 0.01875256495550275, + "grad_norm": 0.21346122026443481, + "learning_rate": 2.9166189238572554e-05, + "loss": 0.0321, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015673823188990355, + "rkd_loss_weight": 0.8488449633121491, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.376146125793457, + "rkd_teacher_entropy": 3.7355056285858153, + "rkd_teacher_similarity_mean": 0.7787266612052918, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4540 + }, + { + "epoch": 1.3238289205702647, + "flow_matching_loss": 0.019714823737740517, + "grad_norm": 0.18590372800827026, + "learning_rate": 2.9160744643594052e-05, + "loss": 0.0338, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016599944885820152, + "rkd_loss_weight": 0.8485116183757782, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3777234137058258, + "rkd_teacher_entropy": 3.7224852800369264, + "rkd_teacher_similarity_mean": 0.7726839363574982, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4550 + }, + { + "epoch": 1.3267384346814082, + "flow_matching_loss": 0.0198115311563015, + "grad_norm": 0.2655235230922699, + "learning_rate": 2.915528284197554e-05, + "loss": 0.0329, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015387212857604027, + "rkd_loss_weight": 0.8481782674789429, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3812467098236084, + "rkd_teacher_entropy": 3.72605721950531, + "rkd_teacher_similarity_mean": 0.7801589608192444, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4560 + }, + { + "epoch": 1.3296479487925517, + "flow_matching_loss": 0.01741666253656149, + "grad_norm": 0.1930238902568817, + "learning_rate": 2.914980384035362e-05, + "loss": 0.0311, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016185384895652532, + "rkd_loss_weight": 0.8478449285030365, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3754947692155838, + "rkd_teacher_entropy": 3.722043180465698, + "rkd_teacher_similarity_mean": 0.7759804248809814, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4570 + }, + { + "epoch": 1.3325574629036951, + "flow_matching_loss": 0.02177420062944293, + "grad_norm": 0.18482503294944763, + "learning_rate": 2.9144307645385787e-05, + "loss": 0.0343, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014791782386600972, + "rkd_loss_weight": 0.8475115835666657, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3731850624084473, + "rkd_teacher_entropy": 3.710199570655823, + "rkd_teacher_similarity_mean": 0.7716832101345062, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4580 + }, + { + "epoch": 1.3354669770148386, + "flow_matching_loss": 0.022444938868284227, + "grad_norm": 0.1657591015100479, + "learning_rate": 2.9138794263750454e-05, + "loss": 0.0378, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.018085608910769223, + "rkd_loss_weight": 0.8471782386302948, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37835306823253634, + "rkd_teacher_entropy": 3.730103945732117, + "rkd_teacher_similarity_mean": 0.7760148882865906, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4590 + }, + { + "epoch": 1.338376491125982, + "flow_matching_loss": 0.018151776678860188, + "grad_norm": 0.22324775159358978, + "learning_rate": 2.913326370214688e-05, + "loss": 0.0297, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013615761324763298, + "rkd_loss_weight": 0.8468448936939239, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3817181348800659, + "rkd_teacher_entropy": 3.725617551803589, + "rkd_teacher_similarity_mean": 0.7804491460323334, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4600 + }, + { + "epoch": 1.3412860052371254, + "flow_matching_loss": 0.022763746231794356, + "grad_norm": 0.24317678809165955, + "learning_rate": 2.9127715967295234e-05, + "loss": 0.0346, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013968766946345568, + "rkd_loss_weight": 0.8465115487575531, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.371809658408165, + "rkd_teacher_entropy": 3.7023159980773928, + "rkd_teacher_similarity_mean": 0.7707738697528839, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4610 + }, + { + "epoch": 1.3441955193482689, + "flow_matching_loss": 0.01895857062190771, + "grad_norm": 0.3573990762233734, + "learning_rate": 2.912215106593653e-05, + "loss": 0.0329, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01649667900055647, + "rkd_loss_weight": 0.8461782038211823, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3814451277256012, + "rkd_teacher_entropy": 3.72841477394104, + "rkd_teacher_similarity_mean": 0.776833838224411, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4620 + }, + { + "epoch": 1.3471050334594123, + "flow_matching_loss": 0.019033401366323234, + "grad_norm": 0.14760272204875946, + "learning_rate": 2.911656900483265e-05, + "loss": 0.0335, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017071243561804294, + "rkd_loss_weight": 0.8458448648452759, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3792101114988327, + "rkd_teacher_entropy": 3.7226582288742067, + "rkd_teacher_similarity_mean": 0.7766776561737061, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4630 + }, + { + "epoch": 1.3500145475705558, + "flow_matching_loss": 0.01755672777071595, + "grad_norm": 0.15027116239070892, + "learning_rate": 2.9110969790766322e-05, + "loss": 0.0292, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013750522583723068, + "rkd_loss_weight": 0.8455115139484406, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37576046884059905, + "rkd_teacher_entropy": 3.7174529075622558, + "rkd_teacher_similarity_mean": 0.776185667514801, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4640 + }, + { + "epoch": 1.352924061681699, + "flow_matching_loss": 0.01923471540212631, + "grad_norm": 0.17949260771274567, + "learning_rate": 2.910535343054112e-05, + "loss": 0.0316, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01461517447605729, + "rkd_loss_weight": 0.8451781690120697, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3753079980611801, + "rkd_teacher_entropy": 3.718987321853638, + "rkd_teacher_similarity_mean": 0.7746521472930908, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4650 + }, + { + "epoch": 1.3558335757928426, + "flow_matching_loss": 0.018580315075814723, + "grad_norm": 0.23945662379264832, + "learning_rate": 2.909971993098145e-05, + "loss": 0.0308, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014505346491932868, + "rkd_loss_weight": 0.8448448359966279, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3771391361951828, + "rkd_teacher_entropy": 3.709166741371155, + "rkd_teacher_similarity_mean": 0.773827975988388, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4660 + }, + { + "epoch": 1.358743089903986, + "flow_matching_loss": 0.020153945218771695, + "grad_norm": 0.16534049808979034, + "learning_rate": 2.9094069298932552e-05, + "loss": 0.0298, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011412954609841108, + "rkd_loss_weight": 0.844511479139328, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3769552856683731, + "rkd_teacher_entropy": 3.7055046796798705, + "rkd_teacher_similarity_mean": 0.7731925249099731, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4670 + }, + { + "epoch": 1.3616526040151296, + "flow_matching_loss": 0.01927756201475859, + "grad_norm": 0.22904329001903534, + "learning_rate": 2.908840154126047e-05, + "loss": 0.0312, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014150828029960394, + "rkd_loss_weight": 0.8441781401634216, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37067234218120576, + "rkd_teacher_entropy": 3.7014949560165404, + "rkd_teacher_similarity_mean": 0.769942706823349, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4680 + }, + { + "epoch": 1.364562118126273, + "flow_matching_loss": 0.0213177727535367, + "grad_norm": 0.20869646966457367, + "learning_rate": 2.9082716664852063e-05, + "loss": 0.0353, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016585759166628123, + "rkd_loss_weight": 0.8438448011875153, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3789159685373306, + "rkd_teacher_entropy": 3.722491216659546, + "rkd_teacher_similarity_mean": 0.7760435819625855, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4690 + }, + { + "epoch": 1.3674716322374163, + "flow_matching_loss": 0.021145828161388637, + "grad_norm": 0.22364076972007751, + "learning_rate": 2.9077014676615e-05, + "loss": 0.0329, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013920049369335174, + "rkd_loss_weight": 0.8435114443302154, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3724900096654892, + "rkd_teacher_entropy": 3.7148394346237184, + "rkd_teacher_similarity_mean": 0.7712933242321014, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4700 + }, + { + "epoch": 1.3703811463485598, + "flow_matching_loss": 0.019533657189458608, + "grad_norm": 0.2982924282550812, + "learning_rate": 2.9071295583477724e-05, + "loss": 0.0346, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017911224719136955, + "rkd_loss_weight": 0.8431781113147736, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3834151715040207, + "rkd_teacher_entropy": 3.7372787237167358, + "rkd_teacher_similarity_mean": 0.7830440819263458, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4710 + }, + { + "epoch": 1.3732906604597033, + "flow_matching_loss": 0.02062594760209322, + "grad_norm": 0.24642252922058105, + "learning_rate": 2.906555939238948e-05, + "loss": 0.0369, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.019326108787208796, + "rkd_loss_weight": 0.8428447604179382, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38234783411026, + "rkd_teacher_entropy": 3.7219348907470704, + "rkd_teacher_similarity_mean": 0.779549103975296, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4720 + }, + { + "epoch": 1.3762001745708465, + "flow_matching_loss": 0.019282583985477685, + "grad_norm": 0.16565388441085815, + "learning_rate": 2.905980611032028e-05, + "loss": 0.0304, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01316548241302371, + "rkd_loss_weight": 0.8425114154815674, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3789342284202576, + "rkd_teacher_entropy": 3.72038152217865, + "rkd_teacher_similarity_mean": 0.7772613942623139, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4730 + }, + { + "epoch": 1.3791096886819902, + "flow_matching_loss": 0.019911481812596322, + "grad_norm": 0.2222883254289627, + "learning_rate": 2.9054035744260904e-05, + "loss": 0.0313, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013558639492839576, + "rkd_loss_weight": 0.842178076505661, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38387741446495055, + "rkd_teacher_entropy": 3.7337488412857054, + "rkd_teacher_similarity_mean": 0.7813285052776336, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4740 + }, + { + "epoch": 1.3820192027931335, + "flow_matching_loss": 0.021486094780266286, + "grad_norm": 0.24982328712940216, + "learning_rate": 2.9048248301222885e-05, + "loss": 0.0365, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017786145862191915, + "rkd_loss_weight": 0.8418447256088257, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3763726055622101, + "rkd_teacher_entropy": 3.712601351737976, + "rkd_teacher_similarity_mean": 0.7712652087211609, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4750 + }, + { + "epoch": 1.384928716904277, + "flow_matching_loss": 0.020451419986784458, + "grad_norm": 0.19643700122833252, + "learning_rate": 2.904244378823852e-05, + "loss": 0.0326, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014443173073232174, + "rkd_loss_weight": 0.8415113866329194, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3703833341598511, + "rkd_teacher_entropy": 3.687127208709717, + "rkd_teacher_similarity_mean": 0.7683172881603241, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4760 + }, + { + "epoch": 1.3878382310154205, + "flow_matching_loss": 0.019805411342531443, + "grad_norm": 0.339511513710022, + "learning_rate": 2.9036622212360836e-05, + "loss": 0.035, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.018118733912706374, + "rkd_loss_weight": 0.8411780416965484, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3786584049463272, + "rkd_teacher_entropy": 3.718133544921875, + "rkd_teacher_similarity_mean": 0.7755283236503601, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4770 + }, + { + "epoch": 1.3907477451265637, + "flow_matching_loss": 0.022305726073682308, + "grad_norm": 0.337769091129303, + "learning_rate": 2.9030783580663596e-05, + "loss": 0.0357, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015915641374886037, + "rkd_loss_weight": 0.8408446907997131, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3758549094200134, + "rkd_teacher_entropy": 3.7153203964233397, + "rkd_teacher_similarity_mean": 0.7727798879146576, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4780 + }, + { + "epoch": 1.3936572592377072, + "flow_matching_loss": 0.020724315382540225, + "grad_norm": 0.22463640570640564, + "learning_rate": 2.902492790024129e-05, + "loss": 0.0336, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015312867891043424, + "rkd_loss_weight": 0.8405113518238068, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3799288600683212, + "rkd_teacher_entropy": 3.721843409538269, + "rkd_teacher_similarity_mean": 0.7770405590534211, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4790 + }, + { + "epoch": 1.3965667733488507, + "flow_matching_loss": 0.017811060603708028, + "grad_norm": 0.20177844166755676, + "learning_rate": 2.901905517820912e-05, + "loss": 0.0315, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016262090764939785, + "rkd_loss_weight": 0.8401780068874359, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3745445877313614, + "rkd_teacher_entropy": 3.7084158182144167, + "rkd_teacher_similarity_mean": 0.7719341933727264, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4800 + }, + { + "epoch": 1.3994762874599942, + "flow_matching_loss": 0.022401710227131843, + "grad_norm": 0.26682958006858826, + "learning_rate": 2.9013165421703e-05, + "loss": 0.0332, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012886296026408672, + "rkd_loss_weight": 0.8398446619510651, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37774854302406313, + "rkd_teacher_entropy": 3.7287753343582155, + "rkd_teacher_similarity_mean": 0.7783403337001801, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4810 + }, + { + "epoch": 1.4023858015711377, + "flow_matching_loss": 0.020601065829396248, + "grad_norm": 0.18531309068202972, + "learning_rate": 2.9007258637879534e-05, + "loss": 0.0325, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014212483353912831, + "rkd_loss_weight": 0.8395113170146942, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36721789836883545, + "rkd_teacher_entropy": 3.6844611167907715, + "rkd_teacher_similarity_mean": 0.7653422236442566, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4820 + }, + { + "epoch": 1.405295315682281, + "flow_matching_loss": 0.0198162280023098, + "grad_norm": 0.17000925540924072, + "learning_rate": 2.9001334833916025e-05, + "loss": 0.0332, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01598474457859993, + "rkd_loss_weight": 0.8391779720783233, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3820540815591812, + "rkd_teacher_entropy": 3.741934370994568, + "rkd_teacher_similarity_mean": 0.7798808991909028, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4830 + }, + { + "epoch": 1.4082048297934244, + "flow_matching_loss": 0.019903726037591697, + "grad_norm": 0.1969371885061264, + "learning_rate": 2.899539401701046e-05, + "loss": 0.0323, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014718263037502765, + "rkd_loss_weight": 0.8388446271419525, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3760757476091385, + "rkd_teacher_entropy": 3.7142275094985964, + "rkd_teacher_similarity_mean": 0.7735866248607636, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4840 + }, + { + "epoch": 1.411114343904568, + "flow_matching_loss": 0.017972411401569844, + "grad_norm": 0.2142428606748581, + "learning_rate": 2.8989436194381474e-05, + "loss": 0.0297, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014027241058647633, + "rkd_loss_weight": 0.8385112881660461, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37653017938137057, + "rkd_teacher_entropy": 3.7149398803710936, + "rkd_teacher_similarity_mean": 0.7752100586891174, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4850 + }, + { + "epoch": 1.4140238580157114, + "flow_matching_loss": 0.021269003767520188, + "grad_norm": 0.17401805520057678, + "learning_rate": 2.8983461373268404e-05, + "loss": 0.0339, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015024269279092551, + "rkd_loss_weight": 0.8381779372692109, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3776071906089783, + "rkd_teacher_entropy": 3.7109790325164793, + "rkd_teacher_similarity_mean": 0.772808974981308, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4860 + }, + { + "epoch": 1.4169333721268549, + "flow_matching_loss": 0.020832088869065048, + "grad_norm": 0.18140685558319092, + "learning_rate": 2.8977469560931205e-05, + "loss": 0.0371, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.019373126048594713, + "rkd_loss_weight": 0.8378445923328399, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3810790449380875, + "rkd_teacher_entropy": 3.7268138408660887, + "rkd_teacher_similarity_mean": 0.7785545349121094, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4870 + }, + { + "epoch": 1.4198428862379981, + "flow_matching_loss": 0.019990904070436955, + "grad_norm": 0.19442881643772125, + "learning_rate": 2.8971460764650502e-05, + "loss": 0.0313, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013512632343918085, + "rkd_loss_weight": 0.8375112533569335, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36687411069869996, + "rkd_teacher_entropy": 3.6863688945770265, + "rkd_teacher_similarity_mean": 0.7647911608219147, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4880 + }, + { + "epoch": 1.4227524003491416, + "flow_matching_loss": 0.01749962382018566, + "grad_norm": 0.13518744707107544, + "learning_rate": 2.8965434991727545e-05, + "loss": 0.0321, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01744557898491621, + "rkd_loss_weight": 0.8371779024600983, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3828879356384277, + "rkd_teacher_entropy": 3.729045104980469, + "rkd_teacher_similarity_mean": 0.779989892244339, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4890 + }, + { + "epoch": 1.4256619144602851, + "flow_matching_loss": 0.019242529384791852, + "grad_norm": 0.16946764290332794, + "learning_rate": 2.895939224948422e-05, + "loss": 0.0318, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014971032924950122, + "rkd_loss_weight": 0.8368445634841919, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38003213703632355, + "rkd_teacher_entropy": 3.7200876235961915, + "rkd_teacher_similarity_mean": 0.7761871218681335, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4900 + }, + { + "epoch": 1.4285714285714286, + "flow_matching_loss": 0.019709511939436198, + "grad_norm": 0.1924055814743042, + "learning_rate": 2.895333254526303e-05, + "loss": 0.031, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01355623472481966, + "rkd_loss_weight": 0.8365112185478211, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.373407644033432, + "rkd_teacher_entropy": 3.7148253202438353, + "rkd_teacher_similarity_mean": 0.7706676363945008, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4910 + }, + { + "epoch": 1.431480942682572, + "flow_matching_loss": 0.02249392746016383, + "grad_norm": 0.1975300908088684, + "learning_rate": 2.8947255886427084e-05, + "loss": 0.0361, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01623530387878418, + "rkd_loss_weight": 0.8361778676509857, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37864467203617097, + "rkd_teacher_entropy": 3.719663119316101, + "rkd_teacher_similarity_mean": 0.774185174703598, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4920 + }, + { + "epoch": 1.4343904567937154, + "flow_matching_loss": 0.021094563975930215, + "grad_norm": 0.21034087240695953, + "learning_rate": 2.8941162280360095e-05, + "loss": 0.0322, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013332857098430396, + "rkd_loss_weight": 0.8358445346355439, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3850958913564682, + "rkd_teacher_entropy": 3.7318463802337645, + "rkd_teacher_similarity_mean": 0.7808663487434387, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4930 + }, + { + "epoch": 1.4372999709048588, + "flow_matching_loss": 0.018631737399846315, + "grad_norm": 0.32069653272628784, + "learning_rate": 2.8935051734466368e-05, + "loss": 0.032, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01604907512664795, + "rkd_loss_weight": 0.835511177778244, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.377356493473053, + "rkd_teacher_entropy": 3.7199040174484255, + "rkd_teacher_similarity_mean": 0.7732920348644257, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4940 + }, + { + "epoch": 1.4402094850160023, + "flow_matching_loss": 0.019767794758081436, + "grad_norm": 0.2148829847574234, + "learning_rate": 2.8928924256170792e-05, + "loss": 0.0309, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013315003365278244, + "rkd_loss_weight": 0.8351778388023376, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38386168479919436, + "rkd_teacher_entropy": 3.719732737541199, + "rkd_teacher_similarity_mean": 0.7799457967281341, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4950 + }, + { + "epoch": 1.4431189991271458, + "flow_matching_loss": 0.019967736769467594, + "grad_norm": 0.1344568431377411, + "learning_rate": 2.8922779852918836e-05, + "loss": 0.0302, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012298326380550862, + "rkd_loss_weight": 0.8348444998264313, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3810773342847824, + "rkd_teacher_entropy": 3.7194835424423216, + "rkd_teacher_similarity_mean": 0.7748064041137696, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4960 + }, + { + "epoch": 1.4460285132382893, + "flow_matching_loss": 0.020421620365232228, + "grad_norm": 0.19753117859363556, + "learning_rate": 2.8916618532176526e-05, + "loss": 0.0331, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01523694284260273, + "rkd_loss_weight": 0.8345111429691314, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37275839149951934, + "rkd_teacher_entropy": 3.687858819961548, + "rkd_teacher_similarity_mean": 0.7662282764911652, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4970 + }, + { + "epoch": 1.4489380273494326, + "flow_matching_loss": 0.020098497346043586, + "grad_norm": 0.1758008897304535, + "learning_rate": 2.891044030143045e-05, + "loss": 0.031, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013054818380624056, + "rkd_loss_weight": 0.8341778099536896, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37966213524341585, + "rkd_teacher_entropy": 3.7263168811798097, + "rkd_teacher_similarity_mean": 0.7778084516525269, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4980 + }, + { + "epoch": 1.451847541460576, + "flow_matching_loss": 0.019761575385928154, + "grad_norm": 0.2677554488182068, + "learning_rate": 2.890424516818774e-05, + "loss": 0.0324, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01520776255056262, + "rkd_loss_weight": 0.8338444590568542, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37212757468223573, + "rkd_teacher_entropy": 3.7043243646621704, + "rkd_teacher_similarity_mean": 0.7707508265972137, + "rkd_vlm_temp": 0.10000000149011612, + "step": 4990 + }, + { + "epoch": 1.4547570555717195, + "flow_matching_loss": 0.020418715197592974, + "grad_norm": 0.20446032285690308, + "learning_rate": 2.8898033139976057e-05, + "loss": 0.0309, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012613671272993088, + "rkd_loss_weight": 0.8335111141204834, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3816624879837036, + "rkd_teacher_entropy": 3.7372210025787354, + "rkd_teacher_similarity_mean": 0.7787380039691925, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5000 + }, + { + "epoch": 1.457666569682863, + "flow_matching_loss": 0.019901585951447486, + "grad_norm": 0.1619981974363327, + "learning_rate": 2.889180422434362e-05, + "loss": 0.0325, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015109512209892272, + "rkd_loss_weight": 0.833177775144577, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37579452991485596, + "rkd_teacher_entropy": 3.7054341316223143, + "rkd_teacher_similarity_mean": 0.7711282908916474, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5010 + }, + { + "epoch": 1.4605760837940065, + "flow_matching_loss": 0.0205977245233953, + "grad_norm": 0.2657514214515686, + "learning_rate": 2.8885558428859136e-05, + "loss": 0.0322, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01396731436252594, + "rkd_loss_weight": 0.8328444242477417, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3757439285516739, + "rkd_teacher_entropy": 3.7076183795928954, + "rkd_teacher_similarity_mean": 0.7710569560527801, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5020 + }, + { + "epoch": 1.4634855979051498, + "flow_matching_loss": 0.02115632249042392, + "grad_norm": 0.3063185214996338, + "learning_rate": 2.8879295761111835e-05, + "loss": 0.0329, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014065569173544646, + "rkd_loss_weight": 0.8325110852718354, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38170196712017057, + "rkd_teacher_entropy": 3.7292377233505247, + "rkd_teacher_similarity_mean": 0.7781385838985443, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5030 + }, + { + "epoch": 1.4663951120162932, + "flow_matching_loss": 0.019333341624587774, + "grad_norm": 0.1462094485759735, + "learning_rate": 2.8873016228711452e-05, + "loss": 0.0308, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013767745997756719, + "rkd_loss_weight": 0.8321777403354644, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38656589984893797, + "rkd_teacher_entropy": 3.733731508255005, + "rkd_teacher_similarity_mean": 0.7798082709312439, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5040 + }, + { + "epoch": 1.4693046261274367, + "flow_matching_loss": 0.017785709444433452, + "grad_norm": 0.16566720604896545, + "learning_rate": 2.8866719839288215e-05, + "loss": 0.0291, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013630218151956797, + "rkd_loss_weight": 0.8318443894386292, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38448272049427035, + "rkd_teacher_entropy": 3.7459094524383545, + "rkd_teacher_similarity_mean": 0.7795694649219513, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5050 + }, + { + "epoch": 1.4722141402385802, + "flow_matching_loss": 0.01885467180982232, + "grad_norm": 0.1790178269147873, + "learning_rate": 2.886040660049283e-05, + "loss": 0.0295, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012767676915973424, + "rkd_loss_weight": 0.8315110504627228, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38375439345836637, + "rkd_teacher_entropy": 3.7265090227127073, + "rkd_teacher_similarity_mean": 0.7778000354766845, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5060 + }, + { + "epoch": 1.4751236543497237, + "flow_matching_loss": 0.018647228460758926, + "grad_norm": 0.18849073350429535, + "learning_rate": 2.8854076519996467e-05, + "loss": 0.0299, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013513791840523481, + "rkd_loss_weight": 0.831177705526352, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37690464556217196, + "rkd_teacher_entropy": 3.7169231176376343, + "rkd_teacher_similarity_mean": 0.7728834569454193, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5070 + }, + { + "epoch": 1.478033168460867, + "flow_matching_loss": 0.018482788931578398, + "grad_norm": 0.1868232935667038, + "learning_rate": 2.884772960549078e-05, + "loss": 0.0312, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015262315887957812, + "rkd_loss_weight": 0.8308443605899811, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3774959981441498, + "rkd_teacher_entropy": 3.700615954399109, + "rkd_teacher_similarity_mean": 0.7722852230072021, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5080 + }, + { + "epoch": 1.4809426825720104, + "flow_matching_loss": 0.019901997689157725, + "grad_norm": 0.18627341091632843, + "learning_rate": 2.8841365864687867e-05, + "loss": 0.0318, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014385319780558348, + "rkd_loss_weight": 0.8305110156536102, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3821996092796326, + "rkd_teacher_entropy": 3.7339192152023317, + "rkd_teacher_similarity_mean": 0.7814014852046967, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5090 + }, + { + "epoch": 1.483852196683154, + "flow_matching_loss": 0.019537121430039404, + "grad_norm": 0.1749638468027115, + "learning_rate": 2.8834985305320276e-05, + "loss": 0.0328, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01593980989418924, + "rkd_loss_weight": 0.8301776766777038, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37439886033535, + "rkd_teacher_entropy": 3.7218552589416505, + "rkd_teacher_similarity_mean": 0.7730808794498444, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5100 + }, + { + "epoch": 1.4867617107942974, + "flow_matching_loss": 0.021309428848326206, + "grad_norm": 0.26129627227783203, + "learning_rate": 2.882858793514098e-05, + "loss": 0.0381, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.02019144603982568, + "rkd_loss_weight": 0.8298443257808685, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37543632090091705, + "rkd_teacher_entropy": 3.713186478614807, + "rkd_teacher_similarity_mean": 0.7735533058643341, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5110 + }, + { + "epoch": 1.489671224905441, + "flow_matching_loss": 0.019750799983739853, + "grad_norm": 0.32260602712631226, + "learning_rate": 2.8822173761923398e-05, + "loss": 0.035, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01844127755612135, + "rkd_loss_weight": 0.8295109868049622, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38164438009262086, + "rkd_teacher_entropy": 3.72546501159668, + "rkd_teacher_similarity_mean": 0.7763604462146759, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5120 + }, + { + "epoch": 1.4925807390165842, + "flow_matching_loss": 0.019692876655608416, + "grad_norm": 0.17951548099517822, + "learning_rate": 2.8815742793461343e-05, + "loss": 0.0307, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013217638339847326, + "rkd_loss_weight": 0.8291776418685913, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3863978832960129, + "rkd_teacher_entropy": 3.737503242492676, + "rkd_teacher_similarity_mean": 0.7825826346874237, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5130 + }, + { + "epoch": 1.4954902531277277, + "flow_matching_loss": 0.022900464478880167, + "grad_norm": 0.19472089409828186, + "learning_rate": 2.880929503756906e-05, + "loss": 0.0361, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015883020218461753, + "rkd_loss_weight": 0.8288442909717559, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37620066702365873, + "rkd_teacher_entropy": 3.7324826240539553, + "rkd_teacher_similarity_mean": 0.7724596202373505, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5140 + }, + { + "epoch": 1.4983997672388711, + "flow_matching_loss": 0.02046155259013176, + "grad_norm": 0.199837327003479, + "learning_rate": 2.880283050208117e-05, + "loss": 0.0323, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014332521706819534, + "rkd_loss_weight": 0.8285109579563141, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3745988249778748, + "rkd_teacher_entropy": 3.7235225439071655, + "rkd_teacher_similarity_mean": 0.7728391528129578, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5150 + }, + { + "epoch": 1.5013092813500144, + "flow_matching_loss": 0.020053804386407137, + "grad_norm": 0.20147834718227386, + "learning_rate": 2.8796349194852704e-05, + "loss": 0.0344, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017342337779700755, + "rkd_loss_weight": 0.8281776010990143, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37325483560562134, + "rkd_teacher_entropy": 3.7085301876068115, + "rkd_teacher_similarity_mean": 0.7697937369346619, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5160 + }, + { + "epoch": 1.504218795461158, + "flow_matching_loss": 0.018454787135124207, + "grad_norm": 0.19510489702224731, + "learning_rate": 2.8789851123759052e-05, + "loss": 0.03, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013902055937796831, + "rkd_loss_weight": 0.8278442621231079, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37786532938480377, + "rkd_teacher_entropy": 3.7176716804504393, + "rkd_teacher_similarity_mean": 0.7756100416183471, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5170 + }, + { + "epoch": 1.5071283095723014, + "flow_matching_loss": 0.017016348894685507, + "grad_norm": 0.16627191007137299, + "learning_rate": 2.8783336296695985e-05, + "loss": 0.0294, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015017440263181924, + "rkd_loss_weight": 0.8275109231472015, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37225112318992615, + "rkd_teacher_entropy": 3.6981772661209105, + "rkd_teacher_similarity_mean": 0.7695075213909149, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5180 + }, + { + "epoch": 1.5100378236834449, + "flow_matching_loss": 0.021388547495007515, + "grad_norm": 0.24615488946437836, + "learning_rate": 2.877680472157964e-05, + "loss": 0.033, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014003422390669585, + "rkd_loss_weight": 0.8271775662899017, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37890752255916593, + "rkd_teacher_entropy": 3.7204819917678833, + "rkd_teacher_similarity_mean": 0.7761666655540467, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5190 + }, + { + "epoch": 1.5129473377945883, + "flow_matching_loss": 0.018247734941542148, + "grad_norm": 0.25016647577285767, + "learning_rate": 2.8770256406346483e-05, + "loss": 0.0299, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01410986389964819, + "rkd_loss_weight": 0.8268442332744599, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38477596640586853, + "rkd_teacher_entropy": 3.740004801750183, + "rkd_teacher_similarity_mean": 0.7796512901782989, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5200 + }, + { + "epoch": 1.5158568519057316, + "flow_matching_loss": 0.018812813702970742, + "grad_norm": 0.22454825043678284, + "learning_rate": 2.8763691358953348e-05, + "loss": 0.0302, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013836309500038623, + "rkd_loss_weight": 0.8265108823776245, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3784813493490219, + "rkd_teacher_entropy": 3.7191306591033935, + "rkd_teacher_similarity_mean": 0.7735161542892456, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5210 + }, + { + "epoch": 1.5187663660168753, + "flow_matching_loss": 0.018576749693602325, + "grad_norm": 0.22734251618385315, + "learning_rate": 2.8757109587377377e-05, + "loss": 0.0309, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01487552598118782, + "rkd_loss_weight": 0.8261775374412537, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37832321524620055, + "rkd_teacher_entropy": 3.7138511419296263, + "rkd_teacher_similarity_mean": 0.7747737109661103, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5220 + }, + { + "epoch": 1.5216758801280186, + "flow_matching_loss": 0.01886829249560833, + "grad_norm": 0.1749114841222763, + "learning_rate": 2.8750511099616047e-05, + "loss": 0.0291, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01241185124963522, + "rkd_loss_weight": 0.8258441984653473, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38236938416957855, + "rkd_teacher_entropy": 3.742629647254944, + "rkd_teacher_similarity_mean": 0.7779726982116699, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5230 + }, + { + "epoch": 1.524585394239162, + "flow_matching_loss": 0.01853857720270753, + "grad_norm": 0.2554599940776825, + "learning_rate": 2.8743895903687143e-05, + "loss": 0.0318, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016104016173630952, + "rkd_loss_weight": 0.825510847568512, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3821983069181442, + "rkd_teacher_entropy": 3.73279983997345, + "rkd_teacher_similarity_mean": 0.7776747107505798, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5240 + }, + { + "epoch": 1.5274949083503055, + "flow_matching_loss": 0.019304281566292048, + "grad_norm": 0.20141476392745972, + "learning_rate": 2.8737264007628753e-05, + "loss": 0.0297, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012592472229152918, + "rkd_loss_weight": 0.8251775085926056, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3740054130554199, + "rkd_teacher_entropy": 3.6996612787246703, + "rkd_teacher_similarity_mean": 0.7688358545303344, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5250 + }, + { + "epoch": 1.5304044224614488, + "flow_matching_loss": 0.020708415284752847, + "grad_norm": 0.20998477935791016, + "learning_rate": 2.8730615419499247e-05, + "loss": 0.0335, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015459050983190536, + "rkd_loss_weight": 0.8248441636562347, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37786470353603363, + "rkd_teacher_entropy": 3.7231579065322875, + "rkd_teacher_similarity_mean": 0.7740902602672577, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5260 + }, + { + "epoch": 1.5333139365725925, + "flow_matching_loss": 0.018839829508215188, + "grad_norm": 0.19197948276996613, + "learning_rate": 2.8723950147377303e-05, + "loss": 0.0303, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013861553091555835, + "rkd_loss_weight": 0.8245108127593994, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3761488288640976, + "rkd_teacher_entropy": 3.7049872398376467, + "rkd_teacher_similarity_mean": 0.7725997269153595, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5270 + }, + { + "epoch": 1.5362234506837358, + "flow_matching_loss": 0.018818091228604318, + "grad_norm": 0.27467960119247437, + "learning_rate": 2.871726819936184e-05, + "loss": 0.0318, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015811979491263627, + "rkd_loss_weight": 0.824177473783493, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3749155253171921, + "rkd_teacher_entropy": 3.7044729471206663, + "rkd_teacher_similarity_mean": 0.770712035894394, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5280 + }, + { + "epoch": 1.5391329647948793, + "flow_matching_loss": 0.017878624610602856, + "grad_norm": 0.1373090147972107, + "learning_rate": 2.8710569583572068e-05, + "loss": 0.0323, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017518373485654593, + "rkd_loss_weight": 0.8238441288471222, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3888072520494461, + "rkd_teacher_entropy": 3.736817407608032, + "rkd_teacher_similarity_mean": 0.7819257020950318, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5290 + }, + { + "epoch": 1.5420424789060228, + "flow_matching_loss": 0.019107543770223855, + "grad_norm": 0.1724090278148651, + "learning_rate": 2.870385430814742e-05, + "loss": 0.03, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013199077174067497, + "rkd_loss_weight": 0.8235107839107514, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3853547483682632, + "rkd_teacher_entropy": 3.7541558504104615, + "rkd_teacher_similarity_mean": 0.7828195035457611, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5300 + }, + { + "epoch": 1.544951993017166, + "flow_matching_loss": 0.0169873408973217, + "grad_norm": 0.1804705709218979, + "learning_rate": 2.8697122381247605e-05, + "loss": 0.0279, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013257636316120624, + "rkd_loss_weight": 0.8231774389743804, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3838395029306412, + "rkd_teacher_entropy": 3.7384379625320436, + "rkd_teacher_similarity_mean": 0.7817961871623993, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5310 + }, + { + "epoch": 1.5478615071283097, + "flow_matching_loss": 0.019650589767843486, + "grad_norm": 0.23282863199710846, + "learning_rate": 2.8690373811052545e-05, + "loss": 0.0324, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01548475781455636, + "rkd_loss_weight": 0.8228440940380096, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3781481713056564, + "rkd_teacher_entropy": 3.7208619594573973, + "rkd_teacher_similarity_mean": 0.7756295680999756, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5320 + }, + { + "epoch": 1.550771021239453, + "flow_matching_loss": 0.018379591405391693, + "grad_norm": 0.16905857622623444, + "learning_rate": 2.8683608605762378e-05, + "loss": 0.0305, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01477826051414013, + "rkd_loss_weight": 0.8225107491016388, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37831591069698334, + "rkd_teacher_entropy": 3.718291473388672, + "rkd_teacher_similarity_mean": 0.7750795006752014, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5330 + }, + { + "epoch": 1.5536805353505965, + "flow_matching_loss": 0.018449717946350575, + "grad_norm": 0.22074614465236664, + "learning_rate": 2.8676826773597485e-05, + "loss": 0.0312, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015448591765016317, + "rkd_loss_weight": 0.8221774101257324, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3765763908624649, + "rkd_teacher_entropy": 3.712827372550964, + "rkd_teacher_similarity_mean": 0.7748854219913482, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5340 + }, + { + "epoch": 1.55659004946174, + "flow_matching_loss": 0.017974628880620003, + "grad_norm": 0.1569817066192627, + "learning_rate": 2.8670028322798413e-05, + "loss": 0.0286, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01291329376399517, + "rkd_loss_weight": 0.8218440592288971, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3840701937675476, + "rkd_teacher_entropy": 3.7354963064193725, + "rkd_teacher_similarity_mean": 0.7816467940807342, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5350 + }, + { + "epoch": 1.5594995635728832, + "flow_matching_loss": 0.01935545802116394, + "grad_norm": 0.24757707118988037, + "learning_rate": 2.8663213261625932e-05, + "loss": 0.0317, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015011664573103189, + "rkd_loss_weight": 0.8215107142925262, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37723633646965027, + "rkd_teacher_entropy": 3.7242414951324463, + "rkd_teacher_similarity_mean": 0.7760546207427979, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5360 + }, + { + "epoch": 1.562409077684027, + "flow_matching_loss": 0.018547193426638842, + "grad_norm": 0.20439478754997253, + "learning_rate": 2.865638159836098e-05, + "loss": 0.0311, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01527619818225503, + "rkd_loss_weight": 0.8211773753166198, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3736391067504883, + "rkd_teacher_entropy": 3.728296899795532, + "rkd_teacher_similarity_mean": 0.775154036283493, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5370 + }, + { + "epoch": 1.5653185917951702, + "flow_matching_loss": 0.019449665676802397, + "grad_norm": 0.15097776055335999, + "learning_rate": 2.8649533341304677e-05, + "loss": 0.031, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01409512385725975, + "rkd_loss_weight": 0.8208440244197845, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3772146373987198, + "rkd_teacher_entropy": 3.7075724601745605, + "rkd_teacher_similarity_mean": 0.773537278175354, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5380 + }, + { + "epoch": 1.5682281059063137, + "flow_matching_loss": 0.02043049605563283, + "grad_norm": 0.31526196002960205, + "learning_rate": 2.8642668498778297e-05, + "loss": 0.0328, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015066707413643599, + "rkd_loss_weight": 0.8205106854438782, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3795627772808075, + "rkd_teacher_entropy": 3.7400415182113647, + "rkd_teacher_similarity_mean": 0.7770272016525268, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5390 + }, + { + "epoch": 1.5711376200174572, + "flow_matching_loss": 0.019173471443355083, + "grad_norm": 0.19531510770320892, + "learning_rate": 2.8635787079123268e-05, + "loss": 0.0299, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013066610787063837, + "rkd_loss_weight": 0.8201773405075073, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3742097347974777, + "rkd_teacher_entropy": 3.7070476770401, + "rkd_teacher_similarity_mean": 0.7734617710113525, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5400 + }, + { + "epoch": 1.5740471341286004, + "flow_matching_loss": 0.019866212084889413, + "grad_norm": 0.16749131679534912, + "learning_rate": 2.8628889090701175e-05, + "loss": 0.0347, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0181291900575161, + "rkd_loss_weight": 0.819843989610672, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3850123107433319, + "rkd_teacher_entropy": 3.7381860494613646, + "rkd_teacher_similarity_mean": 0.7818447887897492, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5410 + }, + { + "epoch": 1.576956648239744, + "flow_matching_loss": 0.019342986214905978, + "grad_norm": 0.1791512668132782, + "learning_rate": 2.862197454189371e-05, + "loss": 0.0318, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015241857152432203, + "rkd_loss_weight": 0.8195106565952301, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3848776310682297, + "rkd_teacher_entropy": 3.732759189605713, + "rkd_teacher_similarity_mean": 0.7803104817867279, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5420 + }, + { + "epoch": 1.5798661623508874, + "flow_matching_loss": 0.017977258004248143, + "grad_norm": 0.18886636197566986, + "learning_rate": 2.8615043441102708e-05, + "loss": 0.0298, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0144367853179574, + "rkd_loss_weight": 0.8191772997379303, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.381242048740387, + "rkd_teacher_entropy": 3.7333101272583007, + "rkd_teacher_similarity_mean": 0.7769344091415405, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5430 + }, + { + "epoch": 1.5827756764620309, + "flow_matching_loss": 0.017450916487723588, + "grad_norm": 0.21136584877967834, + "learning_rate": 2.860809579675011e-05, + "loss": 0.0306, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016008291952311994, + "rkd_loss_weight": 0.8188439607620239, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3902402400970459, + "rkd_teacher_entropy": 3.747873854637146, + "rkd_teacher_similarity_mean": 0.7880598723888397, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5440 + }, + { + "epoch": 1.5856851905731744, + "flow_matching_loss": 0.018098975531756878, + "grad_norm": 0.3597829341888428, + "learning_rate": 2.860113161727796e-05, + "loss": 0.0306, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015333814546465874, + "rkd_loss_weight": 0.8185106217861176, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38768965303897857, + "rkd_teacher_entropy": 3.7436794996261598, + "rkd_teacher_similarity_mean": 0.7833721458911895, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5450 + }, + { + "epoch": 1.5885947046843176, + "flow_matching_loss": 0.01677928101271391, + "grad_norm": 0.22840635478496552, + "learning_rate": 2.8594150911148393e-05, + "loss": 0.027, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01248972089961171, + "rkd_loss_weight": 0.8181772649288177, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3808591514825821, + "rkd_teacher_entropy": 3.727442741394043, + "rkd_teacher_similarity_mean": 0.7766424715518951, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5460 + }, + { + "epoch": 1.5915042187954611, + "flow_matching_loss": 0.01955587789416313, + "grad_norm": 0.18181663751602173, + "learning_rate": 2.8587153686843626e-05, + "loss": 0.0323, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015625439956784247, + "rkd_loss_weight": 0.8178439319133759, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37963123321533204, + "rkd_teacher_entropy": 3.7257237434387207, + "rkd_teacher_similarity_mean": 0.7756927967071533, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5470 + }, + { + "epoch": 1.5944137329066046, + "flow_matching_loss": 0.01994113065302372, + "grad_norm": 0.19424234330654144, + "learning_rate": 2.858013995286594e-05, + "loss": 0.0327, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015600053127855062, + "rkd_loss_weight": 0.8175105810165405, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3869525372982025, + "rkd_teacher_entropy": 3.7424909591674806, + "rkd_teacher_similarity_mean": 0.7824862480163575, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5480 + }, + { + "epoch": 1.597323247017748, + "flow_matching_loss": 0.018846902716904878, + "grad_norm": 0.18930743634700775, + "learning_rate": 2.857310971773769e-05, + "loss": 0.0303, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014025552384555339, + "rkd_loss_weight": 0.8171772360801697, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3800706028938293, + "rkd_teacher_entropy": 3.7168476819992065, + "rkd_teacher_similarity_mean": 0.7754452526569366, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5490 + }, + { + "epoch": 1.6002327611288916, + "flow_matching_loss": 0.022516210190951826, + "grad_norm": 0.17686934769153595, + "learning_rate": 2.8566062990001274e-05, + "loss": 0.0368, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01751694893464446, + "rkd_loss_weight": 0.8168438971042633, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36771261394023896, + "rkd_teacher_entropy": 3.687391185760498, + "rkd_teacher_similarity_mean": 0.7625873684883118, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5500 + }, + { + "epoch": 1.6031422752400348, + "flow_matching_loss": 0.017704290617257357, + "grad_norm": 0.18728621304035187, + "learning_rate": 2.8558999778219126e-05, + "loss": 0.0286, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013340989500284195, + "rkd_loss_weight": 0.816510546207428, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38185423612594604, + "rkd_teacher_entropy": 3.7291337490081786, + "rkd_teacher_similarity_mean": 0.7778617084026337, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5510 + }, + { + "epoch": 1.6060517893511783, + "flow_matching_loss": 0.01847217567265034, + "grad_norm": 0.21726909279823303, + "learning_rate": 2.8551920090973722e-05, + "loss": 0.031, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01530064009130001, + "rkd_loss_weight": 0.8161772072315217, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3814694881439209, + "rkd_teacher_entropy": 3.719847893714905, + "rkd_teacher_similarity_mean": 0.7769156754016876, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5520 + }, + { + "epoch": 1.6089613034623218, + "flow_matching_loss": 0.016554466262459756, + "grad_norm": 0.2651967704296112, + "learning_rate": 2.854482393686754e-05, + "loss": 0.0279, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013884566817432642, + "rkd_loss_weight": 0.8158438622951507, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3884588360786438, + "rkd_teacher_entropy": 3.754813313484192, + "rkd_teacher_similarity_mean": 0.7843117713928223, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5530 + }, + { + "epoch": 1.6118708175734653, + "flow_matching_loss": 0.020902318600565194, + "grad_norm": 0.20716924965381622, + "learning_rate": 2.8537711324523086e-05, + "loss": 0.0342, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016351579688489437, + "rkd_loss_weight": 0.8155105173587799, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3843341499567032, + "rkd_teacher_entropy": 3.74598970413208, + "rkd_teacher_similarity_mean": 0.779675567150116, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5540 + }, + { + "epoch": 1.6147803316846088, + "flow_matching_loss": 0.019414758868515493, + "grad_norm": 0.21237392723560333, + "learning_rate": 2.8530582262582845e-05, + "loss": 0.0308, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01397464368492365, + "rkd_loss_weight": 0.8151771724224091, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37290380597114564, + "rkd_teacher_entropy": 3.704784059524536, + "rkd_teacher_similarity_mean": 0.770972853899002, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5550 + }, + { + "epoch": 1.617689845795752, + "flow_matching_loss": 0.022127147391438486, + "grad_norm": 0.22753286361694336, + "learning_rate": 2.8523436759709306e-05, + "loss": 0.0349, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01566165331751108, + "rkd_loss_weight": 0.8148438334465027, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3695115983486176, + "rkd_teacher_entropy": 3.696432900428772, + "rkd_teacher_similarity_mean": 0.7658731579780579, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5560 + }, + { + "epoch": 1.6205993599068955, + "flow_matching_loss": 0.01929787304252386, + "grad_norm": 0.17943601310253143, + "learning_rate": 2.8516274824584926e-05, + "loss": 0.0297, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012816781364381313, + "rkd_loss_weight": 0.8145104825496674, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39090431928634645, + "rkd_teacher_entropy": 3.761083722114563, + "rkd_teacher_similarity_mean": 0.7907997310161591, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5570 + }, + { + "epoch": 1.623508874018039, + "flow_matching_loss": 0.022440522722899914, + "grad_norm": 0.15586018562316895, + "learning_rate": 2.8509096465912136e-05, + "loss": 0.0358, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016368631087243558, + "rkd_loss_weight": 0.8141771376132965, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38507889211177826, + "rkd_teacher_entropy": 3.725336289405823, + "rkd_teacher_similarity_mean": 0.7787647724151612, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5580 + }, + { + "epoch": 1.6264183881291823, + "flow_matching_loss": 0.018218608107417822, + "grad_norm": 0.23474662005901337, + "learning_rate": 2.8501901692413315e-05, + "loss": 0.0289, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013163407612591983, + "rkd_loss_weight": 0.8138437986373901, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38299946784973143, + "rkd_teacher_entropy": 3.734853458404541, + "rkd_teacher_similarity_mean": 0.7771531701087951, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5590 + }, + { + "epoch": 1.629327902240326, + "flow_matching_loss": 0.020829881262034177, + "grad_norm": 0.17167524993419647, + "learning_rate": 2.8494690512830788e-05, + "loss": 0.0323, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01410979051142931, + "rkd_loss_weight": 0.8135104477405548, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3776038110256195, + "rkd_teacher_entropy": 3.728253197669983, + "rkd_teacher_similarity_mean": 0.775565755367279, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5600 + }, + { + "epoch": 1.6322374163514692, + "flow_matching_loss": 0.02081481320783496, + "grad_norm": 0.15471293032169342, + "learning_rate": 2.8487462935926823e-05, + "loss": 0.033, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015005746856331825, + "rkd_loss_weight": 0.8131771087646484, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3831695854663849, + "rkd_teacher_entropy": 3.742022466659546, + "rkd_teacher_similarity_mean": 0.7783325910568237, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5610 + }, + { + "epoch": 1.6351469304626127, + "flow_matching_loss": 0.01806236822158098, + "grad_norm": 0.14693500101566315, + "learning_rate": 2.8480218970483606e-05, + "loss": 0.0292, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013642036821693182, + "rkd_loss_weight": 0.8128437638282776, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38255807757377625, + "rkd_teacher_entropy": 3.718457818031311, + "rkd_teacher_similarity_mean": 0.7773493707180024, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5620 + }, + { + "epoch": 1.6380564445737562, + "flow_matching_loss": 0.020929934363812207, + "grad_norm": 0.2579675316810608, + "learning_rate": 2.847295862530324e-05, + "loss": 0.0338, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015883150976151228, + "rkd_loss_weight": 0.8125104129314422, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38238413333892823, + "rkd_teacher_entropy": 3.7250109910964966, + "rkd_teacher_similarity_mean": 0.777712368965149, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5630 + }, + { + "epoch": 1.6409659586848995, + "flow_matching_loss": 0.01807741243392229, + "grad_norm": 0.17487658560276031, + "learning_rate": 2.846568190920773e-05, + "loss": 0.032, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017120875976979734, + "rkd_loss_weight": 0.8121770799160004, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37779666781425475, + "rkd_teacher_entropy": 3.7219066619873047, + "rkd_teacher_similarity_mean": 0.7757542133331299, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5640 + }, + { + "epoch": 1.6438754727960432, + "flow_matching_loss": 0.02053183550015092, + "grad_norm": 0.16633376479148865, + "learning_rate": 2.8458388831038965e-05, + "loss": 0.0324, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014634303189814091, + "rkd_loss_weight": 0.8118437230587006, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3715886563062668, + "rkd_teacher_entropy": 3.6956079244613647, + "rkd_teacher_similarity_mean": 0.7682888627052307, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5650 + }, + { + "epoch": 1.6467849869071864, + "flow_matching_loss": 0.019399714749306442, + "grad_norm": 0.24492450058460236, + "learning_rate": 2.8451079399658725e-05, + "loss": 0.0339, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017866753693670034, + "rkd_loss_weight": 0.8115103840827942, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37294589877128603, + "rkd_teacher_entropy": 3.7210729837417604, + "rkd_teacher_similarity_mean": 0.7741088330745697, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5660 + }, + { + "epoch": 1.64969450101833, + "flow_matching_loss": 0.02192228315398097, + "grad_norm": 0.2550716698169708, + "learning_rate": 2.8443753623948665e-05, + "loss": 0.0334, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014143660385161638, + "rkd_loss_weight": 0.8111770451068878, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38039388954639436, + "rkd_teacher_entropy": 3.7363096952438353, + "rkd_teacher_similarity_mean": 0.7807351529598237, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5670 + }, + { + "epoch": 1.6526040151294734, + "flow_matching_loss": 0.017377471458166836, + "grad_norm": 0.1692216545343399, + "learning_rate": 2.8436411512810288e-05, + "loss": 0.0299, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015396180562675, + "rkd_loss_weight": 0.810843688249588, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37027662098407743, + "rkd_teacher_entropy": 3.7079463243484496, + "rkd_teacher_similarity_mean": 0.7691720247268676, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5680 + }, + { + "epoch": 1.6555135292406167, + "flow_matching_loss": 0.021131620090454818, + "grad_norm": 0.18171069025993347, + "learning_rate": 2.842905307516495e-05, + "loss": 0.0333, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015074332803487777, + "rkd_loss_weight": 0.8105103552341462, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3749120831489563, + "rkd_teacher_entropy": 3.7393825531005858, + "rkd_teacher_similarity_mean": 0.7757578372955323, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5690 + }, + { + "epoch": 1.6584230433517604, + "flow_matching_loss": 0.016702421847730874, + "grad_norm": 0.16553060710430145, + "learning_rate": 2.8421678319953854e-05, + "loss": 0.0294, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015730483643710613, + "rkd_loss_weight": 0.8101770043373108, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3889081120491028, + "rkd_teacher_entropy": 3.7540066719055174, + "rkd_teacher_similarity_mean": 0.787485545873642, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5700 + }, + { + "epoch": 1.6613325574629036, + "flow_matching_loss": 0.020988202095031737, + "grad_norm": 0.21577569842338562, + "learning_rate": 2.841428725613801e-05, + "loss": 0.0311, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012463998887687922, + "rkd_loss_weight": 0.8098436594009399, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37488495409488676, + "rkd_teacher_entropy": 3.7143063306808473, + "rkd_teacher_similarity_mean": 0.7732667863368988, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5710 + }, + { + "epoch": 1.6642420715740471, + "flow_matching_loss": 0.01919092461466789, + "grad_norm": 0.19472534954547882, + "learning_rate": 2.8406879892698272e-05, + "loss": 0.0287, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011779259610921144, + "rkd_loss_weight": 0.8095103204250336, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38828144371509554, + "rkd_teacher_entropy": 3.7463204145431517, + "rkd_teacher_similarity_mean": 0.7844540357589722, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5720 + }, + { + "epoch": 1.6671515856851906, + "flow_matching_loss": 0.0235688716173172, + "grad_norm": 0.2352854311466217, + "learning_rate": 2.8399456238635274e-05, + "loss": 0.0381, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017944364063441753, + "rkd_loss_weight": 0.8091769695281983, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38105492293834686, + "rkd_teacher_entropy": 3.7339146375656127, + "rkd_teacher_similarity_mean": 0.777901554107666, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5730 + }, + { + "epoch": 1.6700610997963339, + "flow_matching_loss": 0.020117423590272666, + "grad_norm": 0.22796399891376495, + "learning_rate": 2.8392016302969457e-05, + "loss": 0.0316, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014195432234555484, + "rkd_loss_weight": 0.8088436305522919, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37536920309066774, + "rkd_teacher_entropy": 3.708636164665222, + "rkd_teacher_similarity_mean": 0.7695571482181549, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5740 + }, + { + "epoch": 1.6729706139074776, + "flow_matching_loss": 0.017616425547748803, + "grad_norm": 0.3091526925563812, + "learning_rate": 2.8384560094741055e-05, + "loss": 0.0285, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013431841228157281, + "rkd_loss_weight": 0.808510285615921, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3798834979534149, + "rkd_teacher_entropy": 3.7327314615249634, + "rkd_teacher_similarity_mean": 0.7780246496200561, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5750 + }, + { + "epoch": 1.6758801280186209, + "flow_matching_loss": 0.017323651909828187, + "grad_norm": 0.17506904900074005, + "learning_rate": 2.8377087623010054e-05, + "loss": 0.0284, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01370399296283722, + "rkd_loss_weight": 0.8081769347190857, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3855692893266678, + "rkd_teacher_entropy": 3.746884846687317, + "rkd_teacher_similarity_mean": 0.7826797306537628, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5760 + }, + { + "epoch": 1.6787896421297643, + "flow_matching_loss": 0.018134128581732512, + "grad_norm": 0.15470220148563385, + "learning_rate": 2.8369598896856213e-05, + "loss": 0.0276, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011770824901759624, + "rkd_loss_weight": 0.8078435957431793, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37734395563602446, + "rkd_teacher_entropy": 3.7220016717910767, + "rkd_teacher_similarity_mean": 0.7764519274234771, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5770 + }, + { + "epoch": 1.6816991562409078, + "flow_matching_loss": 0.020437431428581475, + "grad_norm": 0.20197556912899017, + "learning_rate": 2.836209392537904e-05, + "loss": 0.0305, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012474694196134806, + "rkd_loss_weight": 0.8075102508068085, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37653082609176636, + "rkd_teacher_entropy": 3.7163825750350954, + "rkd_teacher_similarity_mean": 0.775563770532608, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5780 + }, + { + "epoch": 1.684608670352051, + "flow_matching_loss": 0.01718769483268261, + "grad_norm": 0.18221665918827057, + "learning_rate": 2.8354572717697787e-05, + "loss": 0.0298, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015572066977620124, + "rkd_loss_weight": 0.8071769058704377, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3798433989286423, + "rkd_teacher_entropy": 3.72267370223999, + "rkd_teacher_similarity_mean": 0.776359885931015, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5790 + }, + { + "epoch": 1.6875181844631948, + "flow_matching_loss": 0.01907844841480255, + "grad_norm": 0.2600548267364502, + "learning_rate": 2.8347035282951428e-05, + "loss": 0.0311, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014857002440840006, + "rkd_loss_weight": 0.8068435609340667, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3831911742687225, + "rkd_teacher_entropy": 3.743376684188843, + "rkd_teacher_similarity_mean": 0.7803894400596618, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5800 + }, + { + "epoch": 1.690427698574338, + "flow_matching_loss": 0.018218605127185584, + "grad_norm": 0.2759881317615509, + "learning_rate": 2.8339481630298653e-05, + "loss": 0.0302, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014835442416369915, + "rkd_loss_weight": 0.8065102159976959, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38572876155376434, + "rkd_teacher_entropy": 3.754758596420288, + "rkd_teacher_similarity_mean": 0.7863865196704865, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5810 + }, + { + "epoch": 1.6933372126854815, + "flow_matching_loss": 0.017668856773525476, + "grad_norm": 0.15931977331638336, + "learning_rate": 2.833191176891787e-05, + "loss": 0.028, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012833892367780209, + "rkd_loss_weight": 0.8061768710613251, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3814793199300766, + "rkd_teacher_entropy": 3.738438272476196, + "rkd_teacher_similarity_mean": 0.7819328367710113, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5820 + }, + { + "epoch": 1.696246726796625, + "flow_matching_loss": 0.018753972370177507, + "grad_norm": 0.1720026135444641, + "learning_rate": 2.832432570800717e-05, + "loss": 0.0318, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016187153104692698, + "rkd_loss_weight": 0.8058435320854187, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37338771522045133, + "rkd_teacher_entropy": 3.7150880336761474, + "rkd_teacher_similarity_mean": 0.7728146135807037, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5830 + }, + { + "epoch": 1.6991562409077683, + "flow_matching_loss": 0.016507747769355773, + "grad_norm": 0.23416748642921448, + "learning_rate": 2.8316723456784332e-05, + "loss": 0.0291, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015615282207727432, + "rkd_loss_weight": 0.8055101811885834, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3801446110010147, + "rkd_teacher_entropy": 3.7210307121276855, + "rkd_teacher_similarity_mean": 0.7821361362934113, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5840 + }, + { + "epoch": 1.702065755018912, + "flow_matching_loss": 0.019101509358733892, + "grad_norm": 0.20548485219478607, + "learning_rate": 2.8309105024486808e-05, + "loss": 0.0305, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014114806521683931, + "rkd_loss_weight": 0.8051768362522125, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.376276695728302, + "rkd_teacher_entropy": 3.7243891954421997, + "rkd_teacher_similarity_mean": 0.7733984112739563, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5850 + }, + { + "epoch": 1.7049752691300553, + "flow_matching_loss": 0.019279235880821942, + "grad_norm": 0.1747581660747528, + "learning_rate": 2.830147042037172e-05, + "loss": 0.0306, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01408785507082939, + "rkd_loss_weight": 0.8048434972763061, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37762522995471953, + "rkd_teacher_entropy": 3.7224286556243897, + "rkd_teacher_similarity_mean": 0.7760672807693482, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5860 + }, + { + "epoch": 1.7078847832411987, + "flow_matching_loss": 0.018043364863842726, + "grad_norm": 0.3141026794910431, + "learning_rate": 2.8293819653715816e-05, + "loss": 0.0321, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017459418252110483, + "rkd_loss_weight": 0.8045101463794708, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37733252346515656, + "rkd_teacher_entropy": 3.720308017730713, + "rkd_teacher_similarity_mean": 0.7755645096302033, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5870 + }, + { + "epoch": 1.7107942973523422, + "flow_matching_loss": 0.019564095232635737, + "grad_norm": 0.28569748997688293, + "learning_rate": 2.828615273381551e-05, + "loss": 0.0308, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014026863127946853, + "rkd_loss_weight": 0.8041768074035645, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3809512466192245, + "rkd_teacher_entropy": 3.729235577583313, + "rkd_teacher_similarity_mean": 0.7772120535373688, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5880 + }, + { + "epoch": 1.7137038114634855, + "flow_matching_loss": 0.0163024784065783, + "grad_norm": 0.19466035068035126, + "learning_rate": 2.827846966998683e-05, + "loss": 0.0282, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014787133596837521, + "rkd_loss_weight": 0.8038434624671936, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3817946881055832, + "rkd_teacher_entropy": 3.728414845466614, + "rkd_teacher_similarity_mean": 0.778993546962738, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5890 + }, + { + "epoch": 1.7166133255746292, + "flow_matching_loss": 0.01798307094722986, + "grad_norm": 0.15818725526332855, + "learning_rate": 2.8270770471565422e-05, + "loss": 0.0287, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013297533709555864, + "rkd_loss_weight": 0.8035101115703582, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3827203452587128, + "rkd_teacher_entropy": 3.7399893045425414, + "rkd_teacher_similarity_mean": 0.782378488779068, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5900 + }, + { + "epoch": 1.7195228396857725, + "flow_matching_loss": 0.01837499514222145, + "grad_norm": 0.16812783479690552, + "learning_rate": 2.826305514790654e-05, + "loss": 0.0298, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014208063203841447, + "rkd_loss_weight": 0.8031767785549164, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3844402402639389, + "rkd_teacher_entropy": 3.746543049812317, + "rkd_teacher_similarity_mean": 0.7830708920955658, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5910 + }, + { + "epoch": 1.722432353796916, + "flow_matching_loss": 0.01994284149259329, + "grad_norm": 0.20610564947128296, + "learning_rate": 2.825532370838503e-05, + "loss": 0.031, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013802686240524054, + "rkd_loss_weight": 0.8028434216976166, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37486677467823026, + "rkd_teacher_entropy": 3.721963882446289, + "rkd_teacher_similarity_mean": 0.7728199481964111, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5920 + }, + { + "epoch": 1.7253418679080594, + "flow_matching_loss": 0.02146736029535532, + "grad_norm": 0.23086801171302795, + "learning_rate": 2.824757616239532e-05, + "loss": 0.0378, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.020370572432875635, + "rkd_loss_weight": 0.8025100827217102, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3796080142259598, + "rkd_teacher_entropy": 3.728911781311035, + "rkd_teacher_similarity_mean": 0.7795083701610566, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5930 + }, + { + "epoch": 1.7282513820192027, + "flow_matching_loss": 0.0199979767203331, + "grad_norm": 0.14044015109539032, + "learning_rate": 2.8239812519351404e-05, + "loss": 0.0322, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01515177208930254, + "rkd_loss_weight": 0.8021767437458038, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39060530066490173, + "rkd_teacher_entropy": 3.7590489864349363, + "rkd_teacher_similarity_mean": 0.7869412481784821, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5940 + }, + { + "epoch": 1.7311608961303462, + "flow_matching_loss": 0.020409602019935847, + "grad_norm": 0.19024904072284698, + "learning_rate": 2.823203278868685e-05, + "loss": 0.0322, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014698547217994929, + "rkd_loss_weight": 0.801843386888504, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37801676988601685, + "rkd_teacher_entropy": 3.7244333028793335, + "rkd_teacher_similarity_mean": 0.7759640157222748, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5950 + }, + { + "epoch": 1.7340704102414897, + "flow_matching_loss": 0.019919699523597956, + "grad_norm": 0.22566145658493042, + "learning_rate": 2.8224236979854752e-05, + "loss": 0.0309, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013661792129278183, + "rkd_loss_weight": 0.8015100538730622, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3798628330230713, + "rkd_teacher_entropy": 3.74082407951355, + "rkd_teacher_similarity_mean": 0.779861569404602, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5960 + }, + { + "epoch": 1.7369799243526332, + "flow_matching_loss": 0.01849620584398508, + "grad_norm": 0.13486647605895996, + "learning_rate": 2.821642510232776e-05, + "loss": 0.0285, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012469311151653529, + "rkd_loss_weight": 0.8011767089366912, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38388609886169434, + "rkd_teacher_entropy": 3.7333094596862795, + "rkd_teacher_similarity_mean": 0.7804243624210357, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5970 + }, + { + "epoch": 1.7398894384637766, + "flow_matching_loss": 0.017030812986195087, + "grad_norm": 0.17106783390045166, + "learning_rate": 2.8208597165598046e-05, + "loss": 0.0277, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013312855549156666, + "rkd_loss_weight": 0.800843358039856, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.382710599899292, + "rkd_teacher_entropy": 3.7212512493133545, + "rkd_teacher_similarity_mean": 0.7804597556591034, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5980 + }, + { + "epoch": 1.74279895257492, + "flow_matching_loss": 0.019764717388898134, + "grad_norm": 0.19747568666934967, + "learning_rate": 2.8200753179177285e-05, + "loss": 0.0305, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0133566215634346, + "rkd_loss_weight": 0.8005100190639496, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3793828248977661, + "rkd_teacher_entropy": 3.724960517883301, + "rkd_teacher_similarity_mean": 0.7763653576374054, + "rkd_vlm_temp": 0.10000000149011612, + "step": 5990 + }, + { + "epoch": 1.7457084666860634, + "flow_matching_loss": 0.018351767119020222, + "grad_norm": 0.20491062104701996, + "learning_rate": 2.819289315259666e-05, + "loss": 0.0302, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01477863835170865, + "rkd_loss_weight": 0.8001766741275788, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.378299754858017, + "rkd_teacher_entropy": 3.721347761154175, + "rkd_teacher_similarity_mean": 0.7730456173419953, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6000 + }, + { + "epoch": 1.7486179807972069, + "flow_matching_loss": 0.020291802380234, + "grad_norm": 0.25596997141838074, + "learning_rate": 2.8185017095406845e-05, + "loss": 0.0324, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015142156835645437, + "rkd_loss_weight": 0.7998433291912079, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3817197471857071, + "rkd_teacher_entropy": 3.7208141803741457, + "rkd_teacher_similarity_mean": 0.7775381326675415, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6010 + }, + { + "epoch": 1.7515274949083504, + "flow_matching_loss": 0.01818032767623663, + "grad_norm": 0.15304826200008392, + "learning_rate": 2.8177125017177998e-05, + "loss": 0.0292, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013810117729008198, + "rkd_loss_weight": 0.799509984254837, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3781019687652588, + "rkd_teacher_entropy": 3.710804510116577, + "rkd_teacher_similarity_mean": 0.7732566237449646, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6020 + }, + { + "epoch": 1.7544370090194938, + "flow_matching_loss": 0.021443308982998132, + "grad_norm": 0.18371978402137756, + "learning_rate": 2.8169216927499727e-05, + "loss": 0.0334, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014953150134533643, + "rkd_loss_weight": 0.7991766393184662, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37944156229496, + "rkd_teacher_entropy": 3.722094464302063, + "rkd_teacher_similarity_mean": 0.774479067325592, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6030 + }, + { + "epoch": 1.757346523130637, + "flow_matching_loss": 0.017031017038971186, + "grad_norm": 0.18435943126678467, + "learning_rate": 2.816129283598112e-05, + "loss": 0.0286, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014467519894242286, + "rkd_loss_weight": 0.7988432943820953, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38855618238449097, + "rkd_teacher_entropy": 3.751187491416931, + "rkd_teacher_similarity_mean": 0.7843833208084107, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6040 + }, + { + "epoch": 1.7602560372417806, + "flow_matching_loss": 0.01857361663132906, + "grad_norm": 0.192178413271904, + "learning_rate": 2.815335275225068e-05, + "loss": 0.0299, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014230541791766882, + "rkd_loss_weight": 0.798509955406189, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37928769290447234, + "rkd_teacher_entropy": 3.7205651998519897, + "rkd_teacher_similarity_mean": 0.7742977142333984, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6050 + }, + { + "epoch": 1.763165551352924, + "flow_matching_loss": 0.019015211146324873, + "grad_norm": 0.2503959536552429, + "learning_rate": 2.8145396685956374e-05, + "loss": 0.0334, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.018035999033600092, + "rkd_loss_weight": 0.7981766045093537, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3808060050010681, + "rkd_teacher_entropy": 3.7258326053619384, + "rkd_teacher_similarity_mean": 0.7776096165180206, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6060 + }, + { + "epoch": 1.7660750654640673, + "flow_matching_loss": 0.017528396658599378, + "grad_norm": 0.15941523015499115, + "learning_rate": 2.813742464676556e-05, + "loss": 0.0287, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014063623268157244, + "rkd_loss_weight": 0.7978432595729827, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3805283635854721, + "rkd_teacher_entropy": 3.7228830814361573, + "rkd_teacher_similarity_mean": 0.774693900346756, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6070 + }, + { + "epoch": 1.768984579575211, + "flow_matching_loss": 0.02241280097514391, + "grad_norm": 0.1999702900648117, + "learning_rate": 2.8129436644365025e-05, + "loss": 0.0324, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012540300749242306, + "rkd_loss_weight": 0.7975099205970764, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3785398691892624, + "rkd_teacher_entropy": 3.7222930431365966, + "rkd_teacher_similarity_mean": 0.7743019938468934, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6080 + }, + { + "epoch": 1.7718940936863543, + "flow_matching_loss": 0.019523031450808047, + "grad_norm": 0.25091567635536194, + "learning_rate": 2.8121432688460938e-05, + "loss": 0.0333, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.017245302721858025, + "rkd_loss_weight": 0.7971765697002411, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37326371669769287, + "rkd_teacher_entropy": 3.7006197452545164, + "rkd_teacher_similarity_mean": 0.7699838876724243, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6090 + }, + { + "epoch": 1.7748036077974978, + "flow_matching_loss": 0.018917490728199482, + "grad_norm": 0.18683110177516937, + "learning_rate": 2.8113412788778867e-05, + "loss": 0.0288, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012436060002073646, + "rkd_loss_weight": 0.7968432307243347, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39306145906448364, + "rkd_teacher_entropy": 3.758553671836853, + "rkd_teacher_similarity_mean": 0.7872677326202393, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6100 + }, + { + "epoch": 1.7777131219086413, + "flow_matching_loss": 0.01902142772451043, + "grad_norm": 0.20931293070316315, + "learning_rate": 2.8105376955063744e-05, + "loss": 0.0307, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01464278670027852, + "rkd_loss_weight": 0.7965098857879639, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3788837730884552, + "rkd_teacher_entropy": 3.7238940238952636, + "rkd_teacher_similarity_mean": 0.7749777734279633, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6110 + }, + { + "epoch": 1.7806226360197845, + "flow_matching_loss": 0.020305990427732467, + "grad_norm": 0.16876329481601715, + "learning_rate": 2.8097325197079862e-05, + "loss": 0.0297, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01180649185553193, + "rkd_loss_weight": 0.7961765348911285, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38284246921539306, + "rkd_teacher_entropy": 3.73650062084198, + "rkd_teacher_similarity_mean": 0.779411768913269, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6120 + }, + { + "epoch": 1.7835321501309283, + "flow_matching_loss": 0.019932676572352648, + "grad_norm": 0.2352622151374817, + "learning_rate": 2.808925752461087e-05, + "loss": 0.0313, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014249060861766338, + "rkd_loss_weight": 0.7958432018756867, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3834924906492233, + "rkd_teacher_entropy": 3.740513014793396, + "rkd_teacher_similarity_mean": 0.7800702929496766, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6130 + }, + { + "epoch": 1.7864416642420715, + "flow_matching_loss": 0.019053488597273828, + "grad_norm": 0.20881079137325287, + "learning_rate": 2.808117394745975e-05, + "loss": 0.0314, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015510247834026814, + "rkd_loss_weight": 0.7955098450183868, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3811270773410797, + "rkd_teacher_entropy": 3.725534534454346, + "rkd_teacher_similarity_mean": 0.7768575549125671, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6140 + }, + { + "epoch": 1.789351178353215, + "flow_matching_loss": 0.017185825388878584, + "grad_norm": 0.18391939997673035, + "learning_rate": 2.807307447544881e-05, + "loss": 0.0291, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01496571721509099, + "rkd_loss_weight": 0.7951765060424805, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3809081941843033, + "rkd_teacher_entropy": 3.7192243576049804, + "rkd_teacher_similarity_mean": 0.7773445606231689, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6150 + }, + { + "epoch": 1.7922606924643585, + "flow_matching_loss": 0.018550318479537965, + "grad_norm": 0.1644180566072464, + "learning_rate": 2.806495911841967e-05, + "loss": 0.0306, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01511178445070982, + "rkd_loss_weight": 0.7948431670665741, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37907699644565584, + "rkd_teacher_entropy": 3.740510416030884, + "rkd_teacher_similarity_mean": 0.7775200366973877, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6160 + }, + { + "epoch": 1.7951702065755017, + "flow_matching_loss": 0.016153142787516117, + "grad_norm": 0.1646028757095337, + "learning_rate": 2.8056827886233255e-05, + "loss": 0.027, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013628901727497578, + "rkd_loss_weight": 0.7945098102092742, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39186743795871737, + "rkd_teacher_entropy": 3.751999616622925, + "rkd_teacher_similarity_mean": 0.7855335354804993, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6170 + }, + { + "epoch": 1.7980797206866455, + "flow_matching_loss": 0.0178585397079587, + "grad_norm": 0.2398061901330948, + "learning_rate": 2.8048680788769783e-05, + "loss": 0.0307, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01611931622028351, + "rkd_loss_weight": 0.7941764771938324, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3827651083469391, + "rkd_teacher_entropy": 3.7249526739120484, + "rkd_teacher_similarity_mean": 0.7772353827953339, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6180 + }, + { + "epoch": 1.8009892347977887, + "flow_matching_loss": 0.01889458317309618, + "grad_norm": 0.19041217863559723, + "learning_rate": 2.8040517835928742e-05, + "loss": 0.0299, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013851233385503292, + "rkd_loss_weight": 0.793843126296997, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3855851322412491, + "rkd_teacher_entropy": 3.750024843215942, + "rkd_teacher_similarity_mean": 0.782490712404251, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6190 + }, + { + "epoch": 1.8038987489089322, + "flow_matching_loss": 0.018861327785998583, + "grad_norm": 0.1997249722480774, + "learning_rate": 2.803233903762889e-05, + "loss": 0.0296, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013538509141653776, + "rkd_loss_weight": 0.7935097813606262, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3839034974575043, + "rkd_teacher_entropy": 3.728865718841553, + "rkd_teacher_similarity_mean": 0.77935289144516, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6200 + }, + { + "epoch": 1.8068082630200757, + "flow_matching_loss": 0.019414264429360627, + "grad_norm": 0.19979245960712433, + "learning_rate": 2.8024144403808235e-05, + "loss": 0.0319, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015707917418330906, + "rkd_loss_weight": 0.7931764423847198, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3757191628217697, + "rkd_teacher_entropy": 3.704569673538208, + "rkd_teacher_similarity_mean": 0.7700340330600739, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6210 + }, + { + "epoch": 1.809717777131219, + "flow_matching_loss": 0.01927773207426071, + "grad_norm": 0.17938676476478577, + "learning_rate": 2.8015933944424037e-05, + "loss": 0.0303, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013927892688661814, + "rkd_loss_weight": 0.7928430914878846, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3867562860250473, + "rkd_teacher_entropy": 3.748392105102539, + "rkd_teacher_similarity_mean": 0.7822883486747741, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6220 + }, + { + "epoch": 1.8126272912423627, + "flow_matching_loss": 0.018715405091643333, + "grad_norm": 0.3801938593387604, + "learning_rate": 2.800770766945277e-05, + "loss": 0.0325, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01745175300166011, + "rkd_loss_weight": 0.7925097525119782, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3891234666109085, + "rkd_teacher_entropy": 3.7514795541763304, + "rkd_teacher_similarity_mean": 0.7858203470706939, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6230 + }, + { + "epoch": 1.815536805353506, + "flow_matching_loss": 0.01831489084288478, + "grad_norm": 0.14149047434329987, + "learning_rate": 2.7999465588890135e-05, + "loss": 0.0283, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01258147619664669, + "rkd_loss_weight": 0.7921764075756073, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37775316536426545, + "rkd_teacher_entropy": 3.7129774570465086, + "rkd_teacher_similarity_mean": 0.7723265826702118, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6240 + }, + { + "epoch": 1.8184463194646494, + "flow_matching_loss": 0.018034793343394993, + "grad_norm": 0.16815195977687836, + "learning_rate": 2.799120771275105e-05, + "loss": 0.0288, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013553793914616108, + "rkd_loss_weight": 0.791843056678772, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3836155205965042, + "rkd_teacher_entropy": 3.7154126882553102, + "rkd_teacher_similarity_mean": 0.7782381653785706, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6250 + }, + { + "epoch": 1.821355833575793, + "flow_matching_loss": 0.018034701701253653, + "grad_norm": 0.14318247139453888, + "learning_rate": 2.7982934051069597e-05, + "loss": 0.028, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012556720245629548, + "rkd_loss_weight": 0.7915097177028656, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3825232028961182, + "rkd_teacher_entropy": 3.7242870569229125, + "rkd_teacher_similarity_mean": 0.7767672657966613, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6260 + }, + { + "epoch": 1.8242653476869362, + "flow_matching_loss": 0.0169092757627368, + "grad_norm": 0.1465601772069931, + "learning_rate": 2.7974644613899063e-05, + "loss": 0.027, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012703144270926715, + "rkd_loss_weight": 0.7911763727664948, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38113549947738645, + "rkd_teacher_entropy": 3.72431275844574, + "rkd_teacher_similarity_mean": 0.7768309652805329, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6270 + }, + { + "epoch": 1.8271748617980799, + "flow_matching_loss": 0.016656945180147886, + "grad_norm": 0.21422846615314484, + "learning_rate": 2.796633941131189e-05, + "loss": 0.0277, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013953728089109062, + "rkd_loss_weight": 0.790843027830124, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3866376280784607, + "rkd_teacher_entropy": 3.7439542531967165, + "rkd_teacher_similarity_mean": 0.7832771062850952, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6280 + }, + { + "epoch": 1.8300843759092231, + "flow_matching_loss": 0.019902670476585626, + "grad_norm": 0.22709479928016663, + "learning_rate": 2.795801845339969e-05, + "loss": 0.0318, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015052001550793647, + "rkd_loss_weight": 0.790509682893753, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37945052683353425, + "rkd_teacher_entropy": 3.71864709854126, + "rkd_teacher_similarity_mean": 0.7746164798736572, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6290 + }, + { + "epoch": 1.8329938900203666, + "flow_matching_loss": 0.017704437393695115, + "grad_norm": 0.17365500330924988, + "learning_rate": 2.7949681750273208e-05, + "loss": 0.0267, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011425827536731959, + "rkd_loss_weight": 0.7901763379573822, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3852950394153595, + "rkd_teacher_entropy": 3.737942337989807, + "rkd_teacher_similarity_mean": 0.7800416767597198, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6300 + }, + { + "epoch": 1.83590340413151, + "flow_matching_loss": 0.01726399203762412, + "grad_norm": 0.15971055626869202, + "learning_rate": 2.7941329312062322e-05, + "loss": 0.0266, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011855719704180956, + "rkd_loss_weight": 0.7898429930210114, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3807476282119751, + "rkd_teacher_entropy": 3.7163588285446165, + "rkd_teacher_similarity_mean": 0.7781699657440185, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6310 + }, + { + "epoch": 1.8388129182426534, + "flow_matching_loss": 0.017196604888886212, + "grad_norm": 0.1616186946630478, + "learning_rate": 2.7932961148916037e-05, + "loss": 0.0286, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01440110830590129, + "rkd_loss_weight": 0.789509654045105, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3749134510755539, + "rkd_teacher_entropy": 3.71610050201416, + "rkd_teacher_similarity_mean": 0.7728530287742614, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6320 + }, + { + "epoch": 1.841722432353797, + "flow_matching_loss": 0.02015474671497941, + "grad_norm": 0.29234617948532104, + "learning_rate": 2.7924577271002456e-05, + "loss": 0.0322, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015311151277273893, + "rkd_loss_weight": 0.7891763031482697, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38667419254779817, + "rkd_teacher_entropy": 3.7556328535079957, + "rkd_teacher_similarity_mean": 0.7832113564014435, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6330 + }, + { + "epoch": 1.8446319464649403, + "flow_matching_loss": 0.017690077982842924, + "grad_norm": 0.23675605654716492, + "learning_rate": 2.7916177688508782e-05, + "loss": 0.0312, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0171514512039721, + "rkd_loss_weight": 0.7888429582118988, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3867544740438461, + "rkd_teacher_entropy": 3.7481605052947997, + "rkd_teacher_similarity_mean": 0.7848730027675629, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6340 + }, + { + "epoch": 1.8475414605760838, + "flow_matching_loss": 0.021276113018393515, + "grad_norm": 0.25295770168304443, + "learning_rate": 2.7907762411641302e-05, + "loss": 0.0339, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01603808831423521, + "rkd_loss_weight": 0.7885096192359924, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3846159398555756, + "rkd_teacher_entropy": 3.743904972076416, + "rkd_teacher_similarity_mean": 0.7810087740421295, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6350 + }, + { + "epoch": 1.8504509746872273, + "flow_matching_loss": 0.017039876244962216, + "grad_norm": 0.16208115220069885, + "learning_rate": 2.7899331450625367e-05, + "loss": 0.0291, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01526174321770668, + "rkd_loss_weight": 0.7881762683391571, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3766286909580231, + "rkd_teacher_entropy": 3.728111433982849, + "rkd_teacher_similarity_mean": 0.774427056312561, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6360 + }, + { + "epoch": 1.8533604887983706, + "flow_matching_loss": 0.0177853899076581, + "grad_norm": 0.22905386984348297, + "learning_rate": 2.7890884815705397e-05, + "loss": 0.0296, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015029634069651365, + "rkd_loss_weight": 0.7878429293632507, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3775303870439529, + "rkd_teacher_entropy": 3.7208590269088746, + "rkd_teacher_similarity_mean": 0.7760675132274628, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6370 + }, + { + "epoch": 1.8562700029095143, + "flow_matching_loss": 0.018394388630986212, + "grad_norm": 0.37617027759552, + "learning_rate": 2.788242251714484e-05, + "loss": 0.0316, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01677700038999319, + "rkd_loss_weight": 0.7875095844268799, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37879684567451477, + "rkd_teacher_entropy": 3.725719356536865, + "rkd_teacher_similarity_mean": 0.77441565990448, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6380 + }, + { + "epoch": 1.8591795170206575, + "flow_matching_loss": 0.019855966977775096, + "grad_norm": 0.16417895257472992, + "learning_rate": 2.787394456522619e-05, + "loss": 0.031, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014096865430474281, + "rkd_loss_weight": 0.7871762335300445, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37661308646202085, + "rkd_teacher_entropy": 3.7199734926223753, + "rkd_teacher_similarity_mean": 0.7756659507751464, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6390 + }, + { + "epoch": 1.862089031131801, + "flow_matching_loss": 0.02146653737872839, + "grad_norm": 0.24596232175827026, + "learning_rate": 2.7865450970250965e-05, + "loss": 0.0331, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014724324829876423, + "rkd_loss_weight": 0.7868429005146027, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38428168296813964, + "rkd_teacher_entropy": 3.750639247894287, + "rkd_teacher_similarity_mean": 0.784703540802002, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6400 + }, + { + "epoch": 1.8649985452429445, + "flow_matching_loss": 0.020312971901148558, + "grad_norm": 0.17710132896900177, + "learning_rate": 2.7856941742539676e-05, + "loss": 0.031, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013553835358470679, + "rkd_loss_weight": 0.7865095496177673, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38134339451789856, + "rkd_teacher_entropy": 3.7273818969726564, + "rkd_teacher_similarity_mean": 0.7810344576835633, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6410 + }, + { + "epoch": 1.8679080593540878, + "flow_matching_loss": 0.01926034539937973, + "grad_norm": 0.2151990532875061, + "learning_rate": 2.784841689243184e-05, + "loss": 0.0293, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012810147739946843, + "rkd_loss_weight": 0.7861762046813965, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.380506294965744, + "rkd_teacher_entropy": 3.726218509674072, + "rkd_teacher_similarity_mean": 0.7788707971572876, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6420 + }, + { + "epoch": 1.8708175734652313, + "flow_matching_loss": 0.018498364370316267, + "grad_norm": 0.22631949186325073, + "learning_rate": 2.783987643028595e-05, + "loss": 0.0287, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013034339156001806, + "rkd_loss_weight": 0.7858428657054901, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38221484124660493, + "rkd_teacher_entropy": 3.736194205284119, + "rkd_teacher_similarity_mean": 0.779453444480896, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6430 + }, + { + "epoch": 1.8737270875763747, + "flow_matching_loss": 0.019722925126552583, + "grad_norm": 0.22309042513370514, + "learning_rate": 2.783132036647948e-05, + "loss": 0.0307, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014020904898643494, + "rkd_loss_weight": 0.7855095148086548, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.381502041220665, + "rkd_teacher_entropy": 3.734965515136719, + "rkd_teacher_similarity_mean": 0.7793893039226532, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6440 + }, + { + "epoch": 1.8766366016875182, + "flow_matching_loss": 0.018488019332289697, + "grad_norm": 0.23241476714611053, + "learning_rate": 2.7822748711408852e-05, + "loss": 0.0295, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014012584462761879, + "rkd_loss_weight": 0.7851761758327485, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3761905819177628, + "rkd_teacher_entropy": 3.7291671514511107, + "rkd_teacher_similarity_mean": 0.7746638715267181, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6450 + }, + { + "epoch": 1.8795461157986617, + "flow_matching_loss": 0.019828020222485067, + "grad_norm": 0.20798739790916443, + "learning_rate": 2.7814161475489428e-05, + "loss": 0.0314, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014725002506747843, + "rkd_loss_weight": 0.7848428308963775, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3809466391801834, + "rkd_teacher_entropy": 3.7270237684249876, + "rkd_teacher_similarity_mean": 0.7779826760292053, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6460 + }, + { + "epoch": 1.882455629909805, + "flow_matching_loss": 0.016951591614633797, + "grad_norm": 0.21925967931747437, + "learning_rate": 2.780555866915552e-05, + "loss": 0.0285, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01473532384261489, + "rkd_loss_weight": 0.7845094799995422, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38634092807769777, + "rkd_teacher_entropy": 3.741237211227417, + "rkd_teacher_similarity_mean": 0.7827922165393829, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6470 + }, + { + "epoch": 1.8853651440209485, + "flow_matching_loss": 0.01825336329638958, + "grad_norm": 0.20194639265537262, + "learning_rate": 2.7796940302860338e-05, + "loss": 0.0264, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01044999696314335, + "rkd_loss_weight": 0.7841761410236359, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38294604122638704, + "rkd_teacher_entropy": 3.7430062770843504, + "rkd_teacher_similarity_mean": 0.7814508318901062, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6480 + }, + { + "epoch": 1.888274658132092, + "flow_matching_loss": 0.01976033514365554, + "grad_norm": 0.26111963391304016, + "learning_rate": 2.7788306387076018e-05, + "loss": 0.0306, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013850660249590873, + "rkd_loss_weight": 0.783842796087265, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38056533336639403, + "rkd_teacher_entropy": 3.7243855476379393, + "rkd_teacher_similarity_mean": 0.779969972372055, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6490 + }, + { + "epoch": 1.8911841722432354, + "flow_matching_loss": 0.0196204187348485, + "grad_norm": 0.17340396344661713, + "learning_rate": 2.7779656932293584e-05, + "loss": 0.0292, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012177587766200305, + "rkd_loss_weight": 0.7835094511508942, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37510640621185304, + "rkd_teacher_entropy": 3.7108415365219116, + "rkd_teacher_similarity_mean": 0.7711569368839264, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6500 + }, + { + "epoch": 1.894093686354379, + "flow_matching_loss": 0.021299185790121557, + "grad_norm": 0.15958207845687866, + "learning_rate": 2.777099194902293e-05, + "loss": 0.0325, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014350730739533902, + "rkd_loss_weight": 0.7831761062145233, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3815311253070831, + "rkd_teacher_entropy": 3.7303401470184325, + "rkd_teacher_similarity_mean": 0.7778044581413269, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6510 + }, + { + "epoch": 1.8970032004655222, + "flow_matching_loss": 0.018878408521413804, + "grad_norm": 0.15348592400550842, + "learning_rate": 2.776231144779283e-05, + "loss": 0.0297, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01387281371280551, + "rkd_loss_weight": 0.7828427612781524, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36880584955215456, + "rkd_teacher_entropy": 3.7023813486099244, + "rkd_teacher_similarity_mean": 0.7677814245224, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6520 + }, + { + "epoch": 1.8999127145766657, + "flow_matching_loss": 0.0184234619140625, + "grad_norm": 0.1613437384366989, + "learning_rate": 2.775361543915091e-05, + "loss": 0.0325, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01802350990474224, + "rkd_loss_weight": 0.7825094163417816, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3778180956840515, + "rkd_teacher_entropy": 3.708527660369873, + "rkd_teacher_similarity_mean": 0.7726912021636962, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6530 + }, + { + "epoch": 1.9028222286878091, + "flow_matching_loss": 0.01649903943762183, + "grad_norm": 0.2414642721414566, + "learning_rate": 2.774490393366365e-05, + "loss": 0.0275, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01408064910210669, + "rkd_loss_weight": 0.7821760773658752, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3832077354192734, + "rkd_teacher_entropy": 3.7292712450027468, + "rkd_teacher_similarity_mean": 0.7762354373931885, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6540 + }, + { + "epoch": 1.9057317427989524, + "flow_matching_loss": 0.01855020588263869, + "grad_norm": 0.1989041268825531, + "learning_rate": 2.7736176941916346e-05, + "loss": 0.0282, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01231165835633874, + "rkd_loss_weight": 0.78184272646904, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3860775977373123, + "rkd_teacher_entropy": 3.7435108423233032, + "rkd_teacher_similarity_mean": 0.7836488008499145, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6550 + }, + { + "epoch": 1.9086412569100961, + "flow_matching_loss": 0.018809772189706565, + "grad_norm": 0.24272866547107697, + "learning_rate": 2.772743447451311e-05, + "loss": 0.0307, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015256730187684298, + "rkd_loss_weight": 0.781509381532669, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3733151078224182, + "rkd_teacher_entropy": 3.7195173501968384, + "rkd_teacher_similarity_mean": 0.7709245085716248, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6560 + }, + { + "epoch": 1.9115507710212394, + "flow_matching_loss": 0.019439257308840753, + "grad_norm": 0.14770029485225677, + "learning_rate": 2.7718676542076872e-05, + "loss": 0.0301, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013708905037492513, + "rkd_loss_weight": 0.7811760425567627, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3737301528453827, + "rkd_teacher_entropy": 3.7157571077346803, + "rkd_teacher_similarity_mean": 0.7708299577236175, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6570 + }, + { + "epoch": 1.9144602851323829, + "flow_matching_loss": 0.017081781383603813, + "grad_norm": 0.18707415461540222, + "learning_rate": 2.770990315524934e-05, + "loss": 0.0277, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013653972651809454, + "rkd_loss_weight": 0.7808426916599274, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39515613913536074, + "rkd_teacher_entropy": 3.7696252346038817, + "rkd_teacher_similarity_mean": 0.7914789319038391, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6580 + }, + { + "epoch": 1.9173697992435264, + "flow_matching_loss": 0.01892915042117238, + "grad_norm": 0.21789413690567017, + "learning_rate": 2.7701114324691015e-05, + "loss": 0.0311, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015534810908138751, + "rkd_loss_weight": 0.780509352684021, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.384135901927948, + "rkd_teacher_entropy": 3.726359248161316, + "rkd_teacher_similarity_mean": 0.7802218914031982, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6590 + }, + { + "epoch": 1.9202793133546696, + "flow_matching_loss": 0.01961874235421419, + "grad_norm": 0.16213558614253998, + "learning_rate": 2.7692310061081145e-05, + "loss": 0.0305, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013979666214436293, + "rkd_loss_weight": 0.7801760077476502, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.383752366900444, + "rkd_teacher_entropy": 3.738799524307251, + "rkd_teacher_similarity_mean": 0.7800216257572175, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6600 + }, + { + "epoch": 1.9231888274658133, + "flow_matching_loss": 0.018673319555819034, + "grad_norm": 0.1841641217470169, + "learning_rate": 2.7683490375117742e-05, + "loss": 0.028, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012022992875427008, + "rkd_loss_weight": 0.7798426568508148, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3926300048828125, + "rkd_teacher_entropy": 3.742633748054504, + "rkd_teacher_similarity_mean": 0.785076904296875, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6610 + }, + { + "epoch": 1.9260983415769566, + "flow_matching_loss": 0.021405904553830624, + "grad_norm": 0.31153321266174316, + "learning_rate": 2.7674655277517556e-05, + "loss": 0.0336, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01558441985398531, + "rkd_loss_weight": 0.779509323835373, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3887563765048981, + "rkd_teacher_entropy": 3.740591287612915, + "rkd_teacher_similarity_mean": 0.7799501836299896, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6620 + }, + { + "epoch": 1.9290078556881, + "flow_matching_loss": 0.01965391533449292, + "grad_norm": 0.17914050817489624, + "learning_rate": 2.7665804779016072e-05, + "loss": 0.0291, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012164821848273277, + "rkd_loss_weight": 0.7791759669780731, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3878941863775253, + "rkd_teacher_entropy": 3.732085680961609, + "rkd_teacher_similarity_mean": 0.7820557951927185, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6630 + }, + { + "epoch": 1.9319173697992436, + "flow_matching_loss": 0.020746017061173915, + "grad_norm": 0.227333202958107, + "learning_rate": 2.765693889036747e-05, + "loss": 0.0327, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015371199324727058, + "rkd_loss_weight": 0.7788426280021667, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39250619113445284, + "rkd_teacher_entropy": 3.7550545930862427, + "rkd_teacher_similarity_mean": 0.7869006931781769, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6640 + }, + { + "epoch": 1.9348268839103868, + "flow_matching_loss": 0.022159754764288664, + "grad_norm": 0.23626522719860077, + "learning_rate": 2.764805762234465e-05, + "loss": 0.0322, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012901846505701541, + "rkd_loss_weight": 0.7785092890262604, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3892532676458359, + "rkd_teacher_entropy": 3.7522701263427733, + "rkd_teacher_similarity_mean": 0.7846925020217895, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6650 + }, + { + "epoch": 1.9377363980215305, + "flow_matching_loss": 0.01657838635146618, + "grad_norm": 0.28075453639030457, + "learning_rate": 2.7639160985739182e-05, + "loss": 0.0268, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0130913439206779, + "rkd_loss_weight": 0.7781759321689605, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3897120863199234, + "rkd_teacher_entropy": 3.7495099544525146, + "rkd_teacher_similarity_mean": 0.7838733851909637, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6660 + }, + { + "epoch": 1.9406459121326738, + "flow_matching_loss": 0.019984566513448952, + "grad_norm": 0.1755090355873108, + "learning_rate": 2.7630248991361317e-05, + "loss": 0.03, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012815549224615096, + "rkd_loss_weight": 0.7778425991535187, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38882575631141664, + "rkd_teacher_entropy": 3.754168891906738, + "rkd_teacher_similarity_mean": 0.785148274898529, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6670 + }, + { + "epoch": 1.9435554262438173, + "flow_matching_loss": 0.0163737160153687, + "grad_norm": 0.17421889305114746, + "learning_rate": 2.762132165003998e-05, + "loss": 0.0277, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014516855217516422, + "rkd_loss_weight": 0.7775092482566833, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3815676003694534, + "rkd_teacher_entropy": 3.7057761192321776, + "rkd_teacher_similarity_mean": 0.773514312505722, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6680 + }, + { + "epoch": 1.9464649403549608, + "flow_matching_loss": 0.017208839207887648, + "grad_norm": 0.2579927146434784, + "learning_rate": 2.7612378972622722e-05, + "loss": 0.0275, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013305100984871388, + "rkd_loss_weight": 0.7771759033203125, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3876975476741791, + "rkd_teacher_entropy": 3.7504640579223634, + "rkd_teacher_similarity_mean": 0.7834380447864533, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6690 + }, + { + "epoch": 1.949374454466104, + "flow_matching_loss": 0.017322116065770388, + "grad_norm": 0.13409432768821716, + "learning_rate": 2.7603420969975743e-05, + "loss": 0.0266, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01194694871082902, + "rkd_loss_weight": 0.7768425643444061, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3828670740127563, + "rkd_teacher_entropy": 3.7360432624816893, + "rkd_teacher_similarity_mean": 0.7780452132225036, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6700 + }, + { + "epoch": 1.9522839685772477, + "flow_matching_loss": 0.02063147835433483, + "grad_norm": 0.1604330837726593, + "learning_rate": 2.7594447652983862e-05, + "loss": 0.0317, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014236043952405453, + "rkd_loss_weight": 0.7765092134475708, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38112453520298006, + "rkd_teacher_entropy": 3.7251381158828734, + "rkd_teacher_similarity_mean": 0.7761844873428345, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6710 + }, + { + "epoch": 1.955193482688391, + "flow_matching_loss": 0.023121032491326333, + "grad_norm": 0.15018832683563232, + "learning_rate": 2.7585459032550497e-05, + "loss": 0.0326, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012237191386520862, + "rkd_loss_weight": 0.7761758744716645, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37738923728466034, + "rkd_teacher_entropy": 3.7134454965591432, + "rkd_teacher_similarity_mean": 0.7715230941772461, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6720 + }, + { + "epoch": 1.9581029967995345, + "flow_matching_loss": 0.019731630105525255, + "grad_norm": 0.20777533948421478, + "learning_rate": 2.7576455119597684e-05, + "loss": 0.0301, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013316040392965078, + "rkd_loss_weight": 0.7758425295352935, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3831041991710663, + "rkd_teacher_entropy": 3.7387229204177856, + "rkd_teacher_similarity_mean": 0.7774753749370575, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6730 + }, + { + "epoch": 1.961012510910678, + "flow_matching_loss": 0.017677260376513005, + "grad_norm": 0.17337730526924133, + "learning_rate": 2.7567435925066012e-05, + "loss": 0.0275, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012613296369090676, + "rkd_loss_weight": 0.7755091786384583, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3864311307668686, + "rkd_teacher_entropy": 3.7282896041870117, + "rkd_teacher_similarity_mean": 0.7790946900844574, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6740 + }, + { + "epoch": 1.9639220250218212, + "flow_matching_loss": 0.021860696468502284, + "grad_norm": 0.22649475932121277, + "learning_rate": 2.7558401459914665e-05, + "loss": 0.0313, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012193948123604059, + "rkd_loss_weight": 0.7751758396625519, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3878162920475006, + "rkd_teacher_entropy": 3.7475479364395143, + "rkd_teacher_similarity_mean": 0.7818822741508484, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6750 + }, + { + "epoch": 1.966831539132965, + "flow_matching_loss": 0.018865352869033812, + "grad_norm": 0.14979664981365204, + "learning_rate": 2.7549351735121362e-05, + "loss": 0.0304, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014889761339873075, + "rkd_loss_weight": 0.774842494726181, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38330390453338625, + "rkd_teacher_entropy": 3.7235904216766356, + "rkd_teacher_similarity_mean": 0.7781642556190491, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6760 + }, + { + "epoch": 1.9697410532441082, + "flow_matching_loss": 0.018226588051766156, + "grad_norm": 0.14835134148597717, + "learning_rate": 2.7540286761682377e-05, + "loss": 0.0296, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014664290938526392, + "rkd_loss_weight": 0.7745091497898102, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39315774142742155, + "rkd_teacher_entropy": 3.7531044483184814, + "rkd_teacher_similarity_mean": 0.7859718084335328, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6770 + }, + { + "epoch": 1.9726505673552517, + "flow_matching_loss": 0.01969079812988639, + "grad_norm": 0.16618113219738007, + "learning_rate": 2.75312065506125e-05, + "loss": 0.0305, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013909244816750288, + "rkd_loss_weight": 0.7741758048534393, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38785227835178376, + "rkd_teacher_entropy": 3.729623055458069, + "rkd_teacher_similarity_mean": 0.7799700796604156, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6780 + }, + { + "epoch": 1.9755600814663952, + "flow_matching_loss": 0.01880234098061919, + "grad_norm": 0.24162557721138, + "learning_rate": 2.7522111112945052e-05, + "loss": 0.0285, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012535418570041656, + "rkd_loss_weight": 0.7738424599170685, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.383782222867012, + "rkd_teacher_entropy": 3.7295445919036867, + "rkd_teacher_similarity_mean": 0.7778990268707275, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6790 + }, + { + "epoch": 1.9784695955775384, + "flow_matching_loss": 0.01894716490060091, + "grad_norm": 0.16715922951698303, + "learning_rate": 2.751300045973184e-05, + "loss": 0.0283, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012151260767132044, + "rkd_loss_weight": 0.7735091149806976, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37702657878398893, + "rkd_teacher_entropy": 3.725097155570984, + "rkd_teacher_similarity_mean": 0.7727644622325898, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6800 + }, + { + "epoch": 1.9813791096886821, + "flow_matching_loss": 0.02041556453332305, + "grad_norm": 0.17834526300430298, + "learning_rate": 2.750387460204317e-05, + "loss": 0.0311, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013827449921518565, + "rkd_loss_weight": 0.7731757760047913, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3820954740047455, + "rkd_teacher_entropy": 3.7379878520965577, + "rkd_teacher_similarity_mean": 0.7768990337848664, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6810 + }, + { + "epoch": 1.9842886237998254, + "flow_matching_loss": 0.02098681339994073, + "grad_norm": 0.3101232051849365, + "learning_rate": 2.7494733550967827e-05, + "loss": 0.033, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015564354322850703, + "rkd_loss_weight": 0.772842425107956, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37798420786857606, + "rkd_teacher_entropy": 3.7127289295196535, + "rkd_teacher_similarity_mean": 0.7733750641345978, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6820 + }, + { + "epoch": 1.9871981379109689, + "flow_matching_loss": 0.016908400785177947, + "grad_norm": 0.15741705894470215, + "learning_rate": 2.748557731761304e-05, + "loss": 0.0267, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012617106456309557, + "rkd_loss_weight": 0.772509080171585, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38789002597332, + "rkd_teacher_entropy": 3.748404932022095, + "rkd_teacher_similarity_mean": 0.7837812721729278, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6830 + }, + { + "epoch": 1.9901076520221124, + "flow_matching_loss": 0.019414545595645906, + "grad_norm": 0.15926964581012726, + "learning_rate": 2.74764059131045e-05, + "loss": 0.0296, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013253656961023808, + "rkd_loss_weight": 0.7721757471561432, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3836622416973114, + "rkd_teacher_entropy": 3.7462480068206787, + "rkd_teacher_similarity_mean": 0.7789127945899963, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6840 + }, + { + "epoch": 1.9930171661332556, + "flow_matching_loss": 0.018073254264891148, + "grad_norm": 0.19025345146656036, + "learning_rate": 2.7467219348586322e-05, + "loss": 0.0289, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013992709526792168, + "rkd_loss_weight": 0.7718423902988434, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3879652827978134, + "rkd_teacher_entropy": 3.7356940269470216, + "rkd_teacher_similarity_mean": 0.7806318044662476, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6850 + }, + { + "epoch": 1.9959266802443993, + "flow_matching_loss": 0.019542469922453166, + "grad_norm": 0.14923202991485596, + "learning_rate": 2.7458017635221056e-05, + "loss": 0.0303, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013923617918044329, + "rkd_loss_weight": 0.771509051322937, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37958577573299407, + "rkd_teacher_entropy": 3.7306424617767333, + "rkd_teacher_similarity_mean": 0.7780254542827606, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6860 + }, + { + "epoch": 1.9988361943555426, + "flow_matching_loss": 0.01785888746380806, + "grad_norm": 0.16855919361114502, + "learning_rate": 2.7448800784189652e-05, + "loss": 0.0297, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015347565524280071, + "rkd_loss_weight": 0.7711757123470306, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38543625771999357, + "rkd_teacher_entropy": 3.746114420890808, + "rkd_teacher_similarity_mean": 0.7844165325164795, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6870 + }, + { + "epoch": 2.001745708466686, + "flow_matching_loss": 0.01887637274339795, + "grad_norm": 0.270055890083313, + "learning_rate": 2.7439568806691448e-05, + "loss": 0.0286, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012610951624810695, + "rkd_loss_weight": 0.7708423554897308, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37956094145774844, + "rkd_teacher_entropy": 3.7246819972991942, + "rkd_teacher_similarity_mean": 0.7772304773330688, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6880 + }, + { + "epoch": 2.0046552225778296, + "flow_matching_loss": 0.018653354980051516, + "grad_norm": 0.17150263488292694, + "learning_rate": 2.7430321713944167e-05, + "loss": 0.0291, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013601904921233653, + "rkd_loss_weight": 0.770509022474289, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38202131986618043, + "rkd_teacher_entropy": 3.7389780044555665, + "rkd_teacher_similarity_mean": 0.7784753680229187, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6890 + }, + { + "epoch": 2.007564736688973, + "flow_matching_loss": 0.019971953798085453, + "grad_norm": 0.17802868783473969, + "learning_rate": 2.7421059517183906e-05, + "loss": 0.0295, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012379074096679687, + "rkd_loss_weight": 0.7701756715774536, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38118250072002413, + "rkd_teacher_entropy": 3.735322856903076, + "rkd_teacher_similarity_mean": 0.7785012662410736, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6900 + }, + { + "epoch": 2.0104742508001165, + "flow_matching_loss": 0.020192697923630477, + "grad_norm": 0.16916611790657043, + "learning_rate": 2.7411782227665097e-05, + "loss": 0.0318, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015090929344296455, + "rkd_loss_weight": 0.7698423266410828, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38973182141780854, + "rkd_teacher_entropy": 3.7549455881118776, + "rkd_teacher_similarity_mean": 0.7852426171302795, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6910 + }, + { + "epoch": 2.01338376491126, + "flow_matching_loss": 0.01878892183303833, + "grad_norm": 0.17883911728858948, + "learning_rate": 2.7402489856660528e-05, + "loss": 0.0277, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011626727785915137, + "rkd_loss_weight": 0.7695089876651764, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38994555473327636, + "rkd_teacher_entropy": 3.7513335704803468, + "rkd_teacher_similarity_mean": 0.7850166082382202, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6920 + }, + { + "epoch": 2.016293279022403, + "flow_matching_loss": 0.01895835530012846, + "grad_norm": 0.19823165237903595, + "learning_rate": 2.7393182415461307e-05, + "loss": 0.0297, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013916361145675182, + "rkd_loss_weight": 0.7691756367683411, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37560135722160337, + "rkd_teacher_entropy": 3.716739869117737, + "rkd_teacher_similarity_mean": 0.7722814381122589, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6930 + }, + { + "epoch": 2.019202793133547, + "flow_matching_loss": 0.017685891594737768, + "grad_norm": 0.16654525697231293, + "learning_rate": 2.7383859915376853e-05, + "loss": 0.0296, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015543736703693866, + "rkd_loss_weight": 0.7688422977924347, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39004878997802733, + "rkd_teacher_entropy": 3.7582547664642334, + "rkd_teacher_similarity_mean": 0.7872194290161133, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6940 + }, + { + "epoch": 2.02211230724469, + "flow_matching_loss": 0.01936702271923423, + "grad_norm": 0.22629298269748688, + "learning_rate": 2.7374522367734875e-05, + "loss": 0.0292, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012793775368481874, + "rkd_loss_weight": 0.7685089528560638, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38252575099468233, + "rkd_teacher_entropy": 3.73043532371521, + "rkd_teacher_similarity_mean": 0.7792853355407715, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6950 + }, + { + "epoch": 2.0250218213558338, + "flow_matching_loss": 0.018960215244442226, + "grad_norm": 0.2541102170944214, + "learning_rate": 2.7365169783881378e-05, + "loss": 0.0286, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012514639459550381, + "rkd_loss_weight": 0.7681756019592285, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37842405438423155, + "rkd_teacher_entropy": 3.716379737854004, + "rkd_teacher_similarity_mean": 0.7757084846496582, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6960 + }, + { + "epoch": 2.027931335466977, + "flow_matching_loss": 0.018814549036324024, + "grad_norm": 0.13229793310165405, + "learning_rate": 2.735580217518064e-05, + "loss": 0.0292, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013584792800247669, + "rkd_loss_weight": 0.7678422629833221, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3836944282054901, + "rkd_teacher_entropy": 3.734186315536499, + "rkd_teacher_similarity_mean": 0.7800632238388061, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6970 + }, + { + "epoch": 2.0308408495781203, + "flow_matching_loss": 0.019729462172836065, + "grad_norm": 0.20751579105854034, + "learning_rate": 2.7346419553015175e-05, + "loss": 0.0287, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011710136663168669, + "rkd_loss_weight": 0.7675089180469513, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38751463294029237, + "rkd_teacher_entropy": 3.747091770172119, + "rkd_teacher_similarity_mean": 0.783930116891861, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6980 + }, + { + "epoch": 2.033750363689264, + "flow_matching_loss": 0.017999732308089732, + "grad_norm": 0.19129054248332977, + "learning_rate": 2.733702192878577e-05, + "loss": 0.0278, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012823153473436832, + "rkd_loss_weight": 0.7671755731105805, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38241797089576723, + "rkd_teacher_entropy": 3.7330538749694826, + "rkd_teacher_similarity_mean": 0.7792617082595825, + "rkd_vlm_temp": 0.10000000149011612, + "step": 6990 + }, + { + "epoch": 2.0366598778004072, + "flow_matching_loss": 0.017371540330350398, + "grad_norm": 0.1745273470878601, + "learning_rate": 2.732760931391141e-05, + "loss": 0.0276, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013286736700683833, + "rkd_loss_weight": 0.7668422281742096, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38018331825733187, + "rkd_teacher_entropy": 3.74027099609375, + "rkd_teacher_similarity_mean": 0.7782812714576721, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7000 + }, + { + "epoch": 2.039569391911551, + "flow_matching_loss": 0.019652044959366323, + "grad_norm": 0.19301696121692657, + "learning_rate": 2.731818171982932e-05, + "loss": 0.029, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012235177122056485, + "rkd_loss_weight": 0.7665088832378387, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3801992177963257, + "rkd_teacher_entropy": 3.72746205329895, + "rkd_teacher_similarity_mean": 0.7782694578170777, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7010 + }, + { + "epoch": 2.042478906022694, + "flow_matching_loss": 0.01755335573107004, + "grad_norm": 0.13675673305988312, + "learning_rate": 2.730873915799491e-05, + "loss": 0.0278, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013360243663191796, + "rkd_loss_weight": 0.7661755383014679, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3847648948431015, + "rkd_teacher_entropy": 3.735346794128418, + "rkd_teacher_similarity_mean": 0.7838184118270874, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7020 + }, + { + "epoch": 2.0453884201338375, + "flow_matching_loss": 0.01641121841967106, + "grad_norm": 0.1578187793493271, + "learning_rate": 2.7299281639881783e-05, + "loss": 0.0263, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012920451443642378, + "rkd_loss_weight": 0.7658421993255615, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3894356399774551, + "rkd_teacher_entropy": 3.755241942405701, + "rkd_teacher_similarity_mean": 0.7869186580181122, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7030 + }, + { + "epoch": 2.048297934244981, + "flow_matching_loss": 0.018498414475470782, + "grad_norm": 0.1496049165725708, + "learning_rate": 2.728980917698172e-05, + "loss": 0.0274, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011680716276168823, + "rkd_loss_weight": 0.7655088484287262, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38673625886440277, + "rkd_teacher_entropy": 3.7442235469818117, + "rkd_teacher_similarity_mean": 0.7834904074668885, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7040 + }, + { + "epoch": 2.0512074483561245, + "flow_matching_loss": 0.017757115699350835, + "grad_norm": 0.18100771307945251, + "learning_rate": 2.7280321780804655e-05, + "loss": 0.0264, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011299339681863784, + "rkd_loss_weight": 0.7651755034923553, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3802807092666626, + "rkd_teacher_entropy": 3.729355525970459, + "rkd_teacher_similarity_mean": 0.7774268567562104, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7050 + }, + { + "epoch": 2.054116962467268, + "flow_matching_loss": 0.016397663019597532, + "grad_norm": 0.30081334710121155, + "learning_rate": 2.7270819462878666e-05, + "loss": 0.0268, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013570921868085862, + "rkd_loss_weight": 0.7648421645164489, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3766243040561676, + "rkd_teacher_entropy": 3.7151686668396, + "rkd_teacher_similarity_mean": 0.7745634138584137, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7060 + }, + { + "epoch": 2.0570264765784114, + "flow_matching_loss": 0.020781490299850703, + "grad_norm": 0.27922895550727844, + "learning_rate": 2.7261302234749965e-05, + "loss": 0.0325, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015357587393373251, + "rkd_loss_weight": 0.7645088136196136, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38192006647586824, + "rkd_teacher_entropy": 3.741830921173096, + "rkd_teacher_similarity_mean": 0.7809388339519501, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7070 + }, + { + "epoch": 2.0599359906895547, + "flow_matching_loss": 0.016563829872757195, + "grad_norm": 0.15199126303195953, + "learning_rate": 2.7251770107982888e-05, + "loss": 0.0254, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011534148082137107, + "rkd_loss_weight": 0.7641754746437073, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38397310972213744, + "rkd_teacher_entropy": 3.7398201704025267, + "rkd_teacher_similarity_mean": 0.7811244904994965, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7080 + }, + { + "epoch": 2.0628455048006984, + "flow_matching_loss": 0.01782287023961544, + "grad_norm": 0.2102210521697998, + "learning_rate": 2.7242223094159856e-05, + "loss": 0.0283, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013655513571575285, + "rkd_loss_weight": 0.7638421297073364, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37901287972927095, + "rkd_teacher_entropy": 3.7200724363327025, + "rkd_teacher_similarity_mean": 0.7743218004703522, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7090 + }, + { + "epoch": 2.0657550189118417, + "flow_matching_loss": 0.0173844862729311, + "grad_norm": 0.16327613592147827, + "learning_rate": 2.7232661204881408e-05, + "loss": 0.0283, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01436009956523776, + "rkd_loss_weight": 0.763508778810501, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3771268457174301, + "rkd_teacher_entropy": 3.7188417673110963, + "rkd_teacher_similarity_mean": 0.7743833124637604, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7100 + }, + { + "epoch": 2.0686645330229854, + "flow_matching_loss": 0.018913179356604814, + "grad_norm": 0.15429158508777618, + "learning_rate": 2.7223084451766123e-05, + "loss": 0.028, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011914919596165419, + "rkd_loss_weight": 0.7631754457950592, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38985691368579867, + "rkd_teacher_entropy": 3.742857575416565, + "rkd_teacher_similarity_mean": 0.7886037707328797, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7110 + }, + { + "epoch": 2.0715740471341286, + "flow_matching_loss": 0.019002045132219793, + "grad_norm": 0.1566379815340042, + "learning_rate": 2.721349284645067e-05, + "loss": 0.0289, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012972262967377902, + "rkd_loss_weight": 0.7628420889377594, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38218545019626615, + "rkd_teacher_entropy": 3.733724069595337, + "rkd_teacher_similarity_mean": 0.779089093208313, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7120 + }, + { + "epoch": 2.074483561245272, + "flow_matching_loss": 0.01910590846091509, + "grad_norm": 0.1676478087902069, + "learning_rate": 2.7203886400589752e-05, + "loss": 0.0298, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013982066232711077, + "rkd_loss_weight": 0.762508749961853, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38338465988636017, + "rkd_teacher_entropy": 3.739035987854004, + "rkd_teacher_similarity_mean": 0.78028244972229, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7130 + }, + { + "epoch": 2.0773930753564156, + "flow_matching_loss": 0.017837692610919477, + "grad_norm": 0.1599939465522766, + "learning_rate": 2.7194265125856105e-05, + "loss": 0.0267, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011574844643473626, + "rkd_loss_weight": 0.7621754109859467, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3809997886419296, + "rkd_teacher_entropy": 3.7276219844818117, + "rkd_teacher_similarity_mean": 0.7790902137756348, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7140 + }, + { + "epoch": 2.080302589467559, + "flow_matching_loss": 0.016294002905488013, + "grad_norm": 0.17574025690555573, + "learning_rate": 2.718462903394049e-05, + "loss": 0.0266, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01354372026398778, + "rkd_loss_weight": 0.7618420541286468, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3766185134649277, + "rkd_teacher_entropy": 3.722101330757141, + "rkd_teacher_similarity_mean": 0.7755940556526184, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7150 + }, + { + "epoch": 2.0832121035787026, + "flow_matching_loss": 0.01943946136161685, + "grad_norm": 0.19569526612758636, + "learning_rate": 2.7174978136551662e-05, + "loss": 0.0295, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013163985777646303, + "rkd_loss_weight": 0.761508721113205, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3797566771507263, + "rkd_teacher_entropy": 3.7348681688308716, + "rkd_teacher_similarity_mean": 0.7807382881641388, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7160 + }, + { + "epoch": 2.086121617689846, + "flow_matching_loss": 0.02008408959954977, + "grad_norm": 0.20963171124458313, + "learning_rate": 2.716531244541637e-05, + "loss": 0.0303, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01337000010535121, + "rkd_loss_weight": 0.7611753702163696, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38402168452739716, + "rkd_teacher_entropy": 3.7466973066329956, + "rkd_teacher_similarity_mean": 0.78230602145195, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7170 + }, + { + "epoch": 2.089031131800989, + "flow_matching_loss": 0.019584843330085276, + "grad_norm": 0.17918655276298523, + "learning_rate": 2.715563197227934e-05, + "loss": 0.0295, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01305175581946969, + "rkd_loss_weight": 0.7608420252799988, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38971503376960753, + "rkd_teacher_entropy": 3.7552907466888428, + "rkd_teacher_similarity_mean": 0.7861795961856842, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7180 + }, + { + "epoch": 2.091940645912133, + "flow_matching_loss": 0.019581091497093438, + "grad_norm": 0.15820130705833435, + "learning_rate": 2.714593672890327e-05, + "loss": 0.0301, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013822425529360772, + "rkd_loss_weight": 0.7605086863040924, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37494221031665803, + "rkd_teacher_entropy": 3.7251227140426635, + "rkd_teacher_similarity_mean": 0.7747707605361939, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7190 + }, + { + "epoch": 2.094850160023276, + "flow_matching_loss": 0.017794495541602375, + "grad_norm": 0.20260705053806305, + "learning_rate": 2.7136226727068783e-05, + "loss": 0.0273, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01245317943394184, + "rkd_loss_weight": 0.7601753354072571, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37527925968170167, + "rkd_teacher_entropy": 3.7163607120513915, + "rkd_teacher_similarity_mean": 0.776327782869339, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7200 + }, + { + "epoch": 2.0977596741344193, + "flow_matching_loss": 0.017753463704138994, + "grad_norm": 0.24347081780433655, + "learning_rate": 2.712650197857445e-05, + "loss": 0.0276, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012946561258286238, + "rkd_loss_weight": 0.7598419964313508, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3827818214893341, + "rkd_teacher_entropy": 3.7395317792892455, + "rkd_teacher_similarity_mean": 0.7836184144020081, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7210 + }, + { + "epoch": 2.100669188245563, + "flow_matching_loss": 0.015556391887366771, + "grad_norm": 0.2172735631465912, + "learning_rate": 2.7116762495236763e-05, + "loss": 0.0265, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014385324809700251, + "rkd_loss_weight": 0.7595086514949798, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3786837697029114, + "rkd_teacher_entropy": 3.7274251699447634, + "rkd_teacher_similarity_mean": 0.7797669887542724, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7220 + }, + { + "epoch": 2.1035787023567063, + "flow_matching_loss": 0.019350775703787803, + "grad_norm": 0.14567583799362183, + "learning_rate": 2.7107008288890106e-05, + "loss": 0.0279, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011237612087279558, + "rkd_loss_weight": 0.7591753005981445, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37592721581459043, + "rkd_teacher_entropy": 3.714743494987488, + "rkd_teacher_similarity_mean": 0.775417011976242, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7230 + }, + { + "epoch": 2.10648821646785, + "flow_matching_loss": 0.020336301904171704, + "grad_norm": 0.1716751903295517, + "learning_rate": 2.709723937138676e-05, + "loss": 0.029, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011442845035344363, + "rkd_loss_weight": 0.7588419616222382, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3832948446273804, + "rkd_teacher_entropy": 3.73614387512207, + "rkd_teacher_similarity_mean": 0.7818938970565796, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7240 + }, + { + "epoch": 2.1093977305789933, + "flow_matching_loss": 0.01611080188304186, + "grad_norm": 0.15972168743610382, + "learning_rate": 2.7087455754596883e-05, + "loss": 0.025, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011653334274888039, + "rkd_loss_weight": 0.7585086166858673, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38508919477462766, + "rkd_teacher_entropy": 3.746321105957031, + "rkd_teacher_similarity_mean": 0.7846007764339447, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7250 + }, + { + "epoch": 2.112307244690137, + "flow_matching_loss": 0.016491200122982264, + "grad_norm": 0.15650713443756104, + "learning_rate": 2.7077657450408498e-05, + "loss": 0.0255, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011946344934403896, + "rkd_loss_weight": 0.7581752717494965, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3799032777547836, + "rkd_teacher_entropy": 3.7306453943252564, + "rkd_teacher_similarity_mean": 0.7792643904685974, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7260 + }, + { + "epoch": 2.1152167588012802, + "flow_matching_loss": 0.018839637376368047, + "grad_norm": 0.25954321026802063, + "learning_rate": 2.7067844470727456e-05, + "loss": 0.0298, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014409559685736895, + "rkd_loss_weight": 0.7578419268131256, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38217505216598513, + "rkd_teacher_entropy": 3.7252942323684692, + "rkd_teacher_similarity_mean": 0.7776755392551422, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7270 + }, + { + "epoch": 2.1181262729124235, + "flow_matching_loss": 0.019014006946235895, + "grad_norm": 0.2005099356174469, + "learning_rate": 2.7058016827477455e-05, + "loss": 0.0295, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013782992027699947, + "rkd_loss_weight": 0.7575085818767547, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.380786395072937, + "rkd_teacher_entropy": 3.7328541755676268, + "rkd_teacher_similarity_mean": 0.7791889548301697, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7280 + }, + { + "epoch": 2.121035787023567, + "flow_matching_loss": 0.02031594067811966, + "grad_norm": 0.24143759906291962, + "learning_rate": 2.7048174532600018e-05, + "loss": 0.0308, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01383385444059968, + "rkd_loss_weight": 0.7571752369403839, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38260661959648135, + "rkd_teacher_entropy": 3.7452654123306273, + "rkd_teacher_similarity_mean": 0.7803253531455994, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7290 + }, + { + "epoch": 2.1239453011347105, + "flow_matching_loss": 0.01697419714182615, + "grad_norm": 0.1234879195690155, + "learning_rate": 2.7038317598054455e-05, + "loss": 0.0256, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011443178914487363, + "rkd_loss_weight": 0.7568418979644775, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3923561930656433, + "rkd_teacher_entropy": 3.751757025718689, + "rkd_teacher_similarity_mean": 0.787241268157959, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7300 + }, + { + "epoch": 2.1268548152458537, + "flow_matching_loss": 0.01871217740699649, + "grad_norm": 0.20135900378227234, + "learning_rate": 2.702844603581787e-05, + "loss": 0.0293, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01396530270576477, + "rkd_loss_weight": 0.7565085530281067, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.387032550573349, + "rkd_teacher_entropy": 3.7381464004516602, + "rkd_teacher_similarity_mean": 0.7806186497211456, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7310 + }, + { + "epoch": 2.1297643293569974, + "flow_matching_loss": 0.020930591505020857, + "grad_norm": 0.14572368562221527, + "learning_rate": 2.701855985788514e-05, + "loss": 0.0317, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014188769552856684, + "rkd_loss_weight": 0.7561752021312713, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3821777641773224, + "rkd_teacher_entropy": 3.7168276071548463, + "rkd_teacher_similarity_mean": 0.7779447436332703, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7320 + }, + { + "epoch": 2.1326738434681407, + "flow_matching_loss": 0.01793085541576147, + "grad_norm": 0.15852566063404083, + "learning_rate": 2.7008659076268912e-05, + "loss": 0.0274, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01258913604542613, + "rkd_loss_weight": 0.7558418691158295, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38508881628513336, + "rkd_teacher_entropy": 3.734599256515503, + "rkd_teacher_similarity_mean": 0.7792636394500733, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7330 + }, + { + "epoch": 2.1355833575792844, + "flow_matching_loss": 0.019149594474583864, + "grad_norm": 0.1808469444513321, + "learning_rate": 2.6998743702999562e-05, + "loss": 0.0296, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013891409803181886, + "rkd_loss_weight": 0.7555085122585297, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3838393211364746, + "rkd_teacher_entropy": 3.738522005081177, + "rkd_teacher_similarity_mean": 0.7786450088024139, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7340 + }, + { + "epoch": 2.1384928716904277, + "flow_matching_loss": 0.018128317873924972, + "grad_norm": 0.2289903163909912, + "learning_rate": 2.6988813750125202e-05, + "loss": 0.0295, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015036764740943908, + "rkd_loss_weight": 0.7551751732826233, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38991723358631136, + "rkd_teacher_entropy": 3.750408744812012, + "rkd_teacher_similarity_mean": 0.7837277591228485, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7350 + }, + { + "epoch": 2.141402385801571, + "flow_matching_loss": 0.02272695116698742, + "grad_norm": 0.21160049736499786, + "learning_rate": 2.6978869229711668e-05, + "loss": 0.0341, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015068430360406638, + "rkd_loss_weight": 0.7548418343067169, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3843601316213608, + "rkd_teacher_entropy": 3.7381548166275023, + "rkd_teacher_similarity_mean": 0.7794408798217773, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7360 + }, + { + "epoch": 2.1443118999127146, + "flow_matching_loss": 0.02189913783222437, + "grad_norm": 0.1507946401834488, + "learning_rate": 2.6968910153842484e-05, + "loss": 0.0337, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01567315785214305, + "rkd_loss_weight": 0.7545084774494171, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38722731471061705, + "rkd_teacher_entropy": 3.7337464809417726, + "rkd_teacher_similarity_mean": 0.7803154945373535, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7370 + }, + { + "epoch": 2.147221414023858, + "flow_matching_loss": 0.019126459211111068, + "grad_norm": 0.2310609519481659, + "learning_rate": 2.6958936534618874e-05, + "loss": 0.0316, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016515968181192876, + "rkd_loss_weight": 0.7541751444339753, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3773760825395584, + "rkd_teacher_entropy": 3.722926306724548, + "rkd_teacher_similarity_mean": 0.7740435481071473, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7380 + }, + { + "epoch": 2.1501309281350016, + "flow_matching_loss": 0.020434887241572143, + "grad_norm": 0.1717895120382309, + "learning_rate": 2.694894838415971e-05, + "loss": 0.0306, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013504886813461781, + "rkd_loss_weight": 0.7538417935371399, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37799360752105715, + "rkd_teacher_entropy": 3.716721200942993, + "rkd_teacher_similarity_mean": 0.773434865474701, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7390 + }, + { + "epoch": 2.153040442246145, + "flow_matching_loss": 0.016922960057854652, + "grad_norm": 0.1460474729537964, + "learning_rate": 2.6938945714601555e-05, + "loss": 0.0275, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01402158997952938, + "rkd_loss_weight": 0.753508448600769, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38903714418411256, + "rkd_teacher_entropy": 3.753339147567749, + "rkd_teacher_similarity_mean": 0.7842981338500976, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7400 + }, + { + "epoch": 2.155949956357288, + "flow_matching_loss": 0.018189989030361176, + "grad_norm": 0.15603525936603546, + "learning_rate": 2.6928928538098587e-05, + "loss": 0.0276, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012446517311036586, + "rkd_loss_weight": 0.7531751096248627, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38906604051589966, + "rkd_teacher_entropy": 3.7389071464538572, + "rkd_teacher_similarity_mean": 0.7855085730552673, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7410 + }, + { + "epoch": 2.158859470468432, + "flow_matching_loss": 0.01793726719915867, + "grad_norm": 0.1630019247531891, + "learning_rate": 2.6918896866822615e-05, + "loss": 0.0261, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010859794728457928, + "rkd_loss_weight": 0.7528417587280274, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38620219230651853, + "rkd_teacher_entropy": 3.7397778987884522, + "rkd_teacher_similarity_mean": 0.7832468628883362, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7420 + }, + { + "epoch": 2.161768984579575, + "flow_matching_loss": 0.017499422281980516, + "grad_norm": 0.17836935818195343, + "learning_rate": 2.690885071296308e-05, + "loss": 0.0259, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011156484205275773, + "rkd_loss_weight": 0.752508419752121, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38360072374343873, + "rkd_teacher_entropy": 3.7344048738479616, + "rkd_teacher_similarity_mean": 0.781316488981247, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7430 + }, + { + "epoch": 2.164678498690719, + "flow_matching_loss": 0.017911468632519245, + "grad_norm": 0.16043174266815186, + "learning_rate": 2.6898790088726993e-05, + "loss": 0.0269, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011933855898678302, + "rkd_loss_weight": 0.7521750748157501, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39318758249282837, + "rkd_teacher_entropy": 3.7640804052352905, + "rkd_teacher_similarity_mean": 0.7908761858940124, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7440 + }, + { + "epoch": 2.167588012801862, + "flow_matching_loss": 0.01750186588615179, + "grad_norm": 0.1547698974609375, + "learning_rate": 2.6888715006338968e-05, + "loss": 0.0274, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013111566845327615, + "rkd_loss_weight": 0.7518417239189148, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3907678633928299, + "rkd_teacher_entropy": 3.7615471363067625, + "rkd_teacher_similarity_mean": 0.787509423494339, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7450 + }, + { + "epoch": 2.1704975269130053, + "flow_matching_loss": 0.020058987196534872, + "grad_norm": 0.16721336543560028, + "learning_rate": 2.6878625478041172e-05, + "loss": 0.0294, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012491014739498495, + "rkd_loss_weight": 0.7515083849430084, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3844289630651474, + "rkd_teacher_entropy": 3.733153200149536, + "rkd_teacher_similarity_mean": 0.7803512334823608, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7460 + }, + { + "epoch": 2.173407041024149, + "flow_matching_loss": 0.018428358994424344, + "grad_norm": 0.16069722175598145, + "learning_rate": 2.6868521516093342e-05, + "loss": 0.0273, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011852994188666343, + "rkd_loss_weight": 0.7511750400066376, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38228921592235565, + "rkd_teacher_entropy": 3.7381947517395018, + "rkd_teacher_similarity_mean": 0.7813246667385101, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7470 + }, + { + "epoch": 2.1763165551352923, + "flow_matching_loss": 0.01756239412352443, + "grad_norm": 0.19797196984291077, + "learning_rate": 2.6858403132772745e-05, + "loss": 0.0266, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011996607203036547, + "rkd_loss_weight": 0.7508416950702668, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3928954929113388, + "rkd_teacher_entropy": 3.777308201789856, + "rkd_teacher_similarity_mean": 0.7928188860416412, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7480 + }, + { + "epoch": 2.179226069246436, + "flow_matching_loss": 0.01570795075967908, + "grad_norm": 0.15178132057189941, + "learning_rate": 2.6848270340374162e-05, + "loss": 0.0258, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013470074068754912, + "rkd_loss_weight": 0.7505083501338958, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37528685927391053, + "rkd_teacher_entropy": 3.705448126792908, + "rkd_teacher_similarity_mean": 0.7705934464931488, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7490 + }, + { + "epoch": 2.1821355833575793, + "flow_matching_loss": 0.020377718191593886, + "grad_norm": 0.1522691696882248, + "learning_rate": 2.68381231512099e-05, + "loss": 0.0308, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013932311907410622, + "rkd_loss_weight": 0.750175005197525, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3776364713907242, + "rkd_teacher_entropy": 3.7126283407211305, + "rkd_teacher_similarity_mean": 0.7714143335819245, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7500 + }, + { + "epoch": 2.1850450974687226, + "flow_matching_loss": 0.017920018825680017, + "grad_norm": 0.34090179204940796, + "learning_rate": 2.6827961577609745e-05, + "loss": 0.0285, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014174133352935315, + "rkd_loss_weight": 0.7498416602611542, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37827340662479403, + "rkd_teacher_entropy": 3.7356621980667115, + "rkd_teacher_similarity_mean": 0.7778164923191071, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7510 + }, + { + "epoch": 2.1879546115798663, + "flow_matching_loss": 0.021243251208215953, + "grad_norm": 0.24483944475650787, + "learning_rate": 2.681778563192097e-05, + "loss": 0.0327, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015294513478875161, + "rkd_loss_weight": 0.7495083212852478, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39061145186424256, + "rkd_teacher_entropy": 3.7718693733215334, + "rkd_teacher_similarity_mean": 0.7885009706020355, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7520 + }, + { + "epoch": 2.1908641256910095, + "flow_matching_loss": 0.01736484095454216, + "grad_norm": 0.16771511733531952, + "learning_rate": 2.6807595326508308e-05, + "loss": 0.0278, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013916752766817808, + "rkd_loss_weight": 0.7491749703884125, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3701164871454239, + "rkd_teacher_entropy": 3.6904604911804197, + "rkd_teacher_similarity_mean": 0.7673202872276306, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7530 + }, + { + "epoch": 2.1937736398021532, + "flow_matching_loss": 0.017297479230910538, + "grad_norm": 0.17688392102718353, + "learning_rate": 2.679739067375394e-05, + "loss": 0.0259, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011526377499103546, + "rkd_loss_weight": 0.7488416254520416, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3841901898384094, + "rkd_teacher_entropy": 3.753845143318176, + "rkd_teacher_similarity_mean": 0.7839423835277557, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7540 + }, + { + "epoch": 2.1966831539132965, + "flow_matching_loss": 0.018084768578410147, + "grad_norm": 0.2297622561454773, + "learning_rate": 2.6787171686057487e-05, + "loss": 0.0289, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014396316278725863, + "rkd_loss_weight": 0.7485082864761352, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38088143765926363, + "rkd_teacher_entropy": 3.7290563106536867, + "rkd_teacher_similarity_mean": 0.7801232755184173, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7550 + }, + { + "epoch": 2.1995926680244398, + "flow_matching_loss": 0.020635624043643476, + "grad_norm": 0.191923126578331, + "learning_rate": 2.677693837583597e-05, + "loss": 0.0301, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012696328572928905, + "rkd_loss_weight": 0.7481749355792999, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3835279017686844, + "rkd_teacher_entropy": 3.7376479625701906, + "rkd_teacher_similarity_mean": 0.7817672669887543, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7560 + }, + { + "epoch": 2.2025021821355835, + "flow_matching_loss": 0.016393035277724267, + "grad_norm": 0.16806794703006744, + "learning_rate": 2.6766690755523836e-05, + "loss": 0.026, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012779910489916802, + "rkd_loss_weight": 0.7478415966033936, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3829520374536514, + "rkd_teacher_entropy": 3.719150996208191, + "rkd_teacher_similarity_mean": 0.7781254947185516, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7570 + }, + { + "epoch": 2.2054116962467267, + "flow_matching_loss": 0.016937372647225855, + "grad_norm": 0.30143022537231445, + "learning_rate": 2.675642883757291e-05, + "loss": 0.0265, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012775290943682193, + "rkd_loss_weight": 0.7475082516670227, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38421749472618105, + "rkd_teacher_entropy": 3.7449097871780395, + "rkd_teacher_similarity_mean": 0.7814590871334076, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7580 + }, + { + "epoch": 2.2083212103578704, + "flow_matching_loss": 0.01790853925049305, + "grad_norm": 0.27450886368751526, + "learning_rate": 2.674615263445239e-05, + "loss": 0.0269, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0120752464979887, + "rkd_loss_weight": 0.7471749007701873, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.386181578040123, + "rkd_teacher_entropy": 3.745365524291992, + "rkd_teacher_similarity_mean": 0.7846297323703766, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7590 + }, + { + "epoch": 2.2112307244690137, + "flow_matching_loss": 0.018053863663226366, + "grad_norm": 0.2074274867773056, + "learning_rate": 2.673586215864882e-05, + "loss": 0.0276, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012789921741932631, + "rkd_loss_weight": 0.7468415677547455, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3855713099241257, + "rkd_teacher_entropy": 3.736197066307068, + "rkd_teacher_similarity_mean": 0.7824857532978058, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7600 + }, + { + "epoch": 2.214140238580157, + "flow_matching_loss": 0.017040489055216313, + "grad_norm": 0.1810796707868576, + "learning_rate": 2.6725557422666113e-05, + "loss": 0.0271, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013505901396274566, + "rkd_loss_weight": 0.7465082108974457, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3877188146114349, + "rkd_teacher_entropy": 3.752838945388794, + "rkd_teacher_similarity_mean": 0.7842600166797637, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7610 + }, + { + "epoch": 2.2170497526913007, + "flow_matching_loss": 0.017254010401666166, + "grad_norm": 0.1495549976825714, + "learning_rate": 2.6715238439025493e-05, + "loss": 0.0271, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013167501986026764, + "rkd_loss_weight": 0.7461748719215393, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3909841239452362, + "rkd_teacher_entropy": 3.749372100830078, + "rkd_teacher_similarity_mean": 0.7866510689258576, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7620 + }, + { + "epoch": 2.219959266802444, + "flow_matching_loss": 0.019025284424424173, + "grad_norm": 0.1881931871175766, + "learning_rate": 2.6704905220265493e-05, + "loss": 0.0299, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01453378377482295, + "rkd_loss_weight": 0.7458415329456329, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38359035849571227, + "rkd_teacher_entropy": 3.731373929977417, + "rkd_teacher_similarity_mean": 0.7800119102001191, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7630 + }, + { + "epoch": 2.2228687809135876, + "flow_matching_loss": 0.01823749653995037, + "grad_norm": 0.1953657865524292, + "learning_rate": 2.6694557778941947e-05, + "loss": 0.0278, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01286142636090517, + "rkd_loss_weight": 0.7455081760883331, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38893520534038545, + "rkd_teacher_entropy": 3.7565080404281614, + "rkd_teacher_similarity_mean": 0.787079393863678, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7640 + }, + { + "epoch": 2.225778295024731, + "flow_matching_loss": 0.01869949037209153, + "grad_norm": 0.2937917411327362, + "learning_rate": 2.6684196127627977e-05, + "loss": 0.0298, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014842665987089276, + "rkd_loss_weight": 0.7451748430728913, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38693633377552034, + "rkd_teacher_entropy": 3.742464280128479, + "rkd_teacher_similarity_mean": 0.7835041403770446, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7650 + }, + { + "epoch": 2.228687809135874, + "flow_matching_loss": 0.022506882436573505, + "grad_norm": 0.2064017504453659, + "learning_rate": 2.6673820278913963e-05, + "loss": 0.031, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011352307908236981, + "rkd_loss_weight": 0.7448414921760559, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38505671322345736, + "rkd_teacher_entropy": 3.751017689704895, + "rkd_teacher_similarity_mean": 0.7818023264408112, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7660 + }, + { + "epoch": 2.231597323247018, + "flow_matching_loss": 0.018042841460555792, + "grad_norm": 0.18285033106803894, + "learning_rate": 2.6663430245407544e-05, + "loss": 0.0274, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01255797203630209, + "rkd_loss_weight": 0.7445081472396851, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38632768392562866, + "rkd_teacher_entropy": 3.744926428794861, + "rkd_teacher_similarity_mean": 0.7830249667167664, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7670 + }, + { + "epoch": 2.234506837358161, + "flow_matching_loss": 0.017659199703484774, + "grad_norm": 0.15429939329624176, + "learning_rate": 2.6653026039733585e-05, + "loss": 0.0266, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011973702255636454, + "rkd_loss_weight": 0.7441748082637787, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.374105429649353, + "rkd_teacher_entropy": 3.707072901725769, + "rkd_teacher_similarity_mean": 0.7695008456707001, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7680 + }, + { + "epoch": 2.237416351469305, + "flow_matching_loss": 0.016706874780356883, + "grad_norm": 0.161892831325531, + "learning_rate": 2.6642607674534176e-05, + "loss": 0.0265, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013203760795295238, + "rkd_loss_weight": 0.7438414573669434, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3774045020341873, + "rkd_teacher_entropy": 3.7070250511169434, + "rkd_teacher_similarity_mean": 0.7720758497714997, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7690 + }, + { + "epoch": 2.240325865580448, + "flow_matching_loss": 0.021946324314922095, + "grad_norm": 0.16284510493278503, + "learning_rate": 2.663217516246862e-05, + "loss": 0.0329, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014796940423548222, + "rkd_loss_weight": 0.743508118391037, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.385759973526001, + "rkd_teacher_entropy": 3.732961368560791, + "rkd_teacher_similarity_mean": 0.7822143733501434, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7700 + }, + { + "epoch": 2.2432353796915914, + "flow_matching_loss": 0.018916599545627834, + "grad_norm": 0.2531246542930603, + "learning_rate": 2.66217285162134e-05, + "loss": 0.0295, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014235298428684474, + "rkd_loss_weight": 0.7431747734546661, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3800871789455414, + "rkd_teacher_entropy": 3.719753456115723, + "rkd_teacher_similarity_mean": 0.7748778283596038, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7710 + }, + { + "epoch": 2.246144893802735, + "flow_matching_loss": 0.016887657903134824, + "grad_norm": 0.1680547297000885, + "learning_rate": 2.6611267748462175e-05, + "loss": 0.0262, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01249928055331111, + "rkd_loss_weight": 0.7428414225578308, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3829909563064575, + "rkd_teacher_entropy": 3.7249913930892946, + "rkd_teacher_similarity_mean": 0.7794267117977143, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7720 + }, + { + "epoch": 2.2490544079138783, + "flow_matching_loss": 0.018757448997348547, + "grad_norm": 0.20050904154777527, + "learning_rate": 2.6600792871925764e-05, + "loss": 0.029, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013788842596113682, + "rkd_loss_weight": 0.7425080835819244, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38307216167449953, + "rkd_teacher_entropy": 3.727544736862183, + "rkd_teacher_similarity_mean": 0.7777710318565368, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7730 + }, + { + "epoch": 2.2519639220250216, + "flow_matching_loss": 0.017616415116935968, + "grad_norm": 0.19016481935977936, + "learning_rate": 2.659030389933213e-05, + "loss": 0.0275, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013258658070117234, + "rkd_loss_weight": 0.7421747446060181, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3867697209119797, + "rkd_teacher_entropy": 3.7484647274017333, + "rkd_teacher_similarity_mean": 0.785621190071106, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7740 + }, + { + "epoch": 2.2548734361361653, + "flow_matching_loss": 0.017975373845547437, + "grad_norm": 0.21084663271903992, + "learning_rate": 2.6579800843426358e-05, + "loss": 0.0285, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014237384870648384, + "rkd_loss_weight": 0.7418413937091828, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39172492623329164, + "rkd_teacher_entropy": 3.770625042915344, + "rkd_teacher_similarity_mean": 0.7889850318431855, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7750 + }, + { + "epoch": 2.2577829502473086, + "flow_matching_loss": 0.018151620496064426, + "grad_norm": 0.1692495197057724, + "learning_rate": 2.656928371697067e-05, + "loss": 0.0284, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013790252339094877, + "rkd_loss_weight": 0.7415080487728118, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3874436467885971, + "rkd_teacher_entropy": 3.7587918519973753, + "rkd_teacher_similarity_mean": 0.7842287003993988, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7760 + }, + { + "epoch": 2.2606924643584523, + "flow_matching_loss": 0.023463150300085545, + "grad_norm": 0.1662845015525818, + "learning_rate": 2.6558752532744346e-05, + "loss": 0.0325, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012169005535542964, + "rkd_loss_weight": 0.7411747097969055, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38188145458698275, + "rkd_teacher_entropy": 3.732459235191345, + "rkd_teacher_similarity_mean": 0.7790606856346131, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7770 + }, + { + "epoch": 2.2636019784695955, + "flow_matching_loss": 0.01666386453434825, + "grad_norm": 0.17014403641223907, + "learning_rate": 2.654820730354377e-05, + "loss": 0.0273, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014384075999259949, + "rkd_loss_weight": 0.7408413589000702, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3860397577285767, + "rkd_teacher_entropy": 3.749503254890442, + "rkd_teacher_similarity_mean": 0.7834955453872681, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7780 + }, + { + "epoch": 2.2665114925807393, + "flow_matching_loss": 0.016561725176870823, + "grad_norm": 0.16845044493675232, + "learning_rate": 2.6537648042182406e-05, + "loss": 0.0268, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01380744930356741, + "rkd_loss_weight": 0.7405080199241638, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38083186745643616, + "rkd_teacher_entropy": 3.7383872509002685, + "rkd_teacher_similarity_mean": 0.7779848039150238, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7790 + }, + { + "epoch": 2.2694210066918825, + "flow_matching_loss": 0.017542120162397624, + "grad_norm": 0.26451733708381653, + "learning_rate": 2.6527074761490735e-05, + "loss": 0.0274, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01329907993786037, + "rkd_loss_weight": 0.740174674987793, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3826435685157776, + "rkd_teacher_entropy": 3.740646004676819, + "rkd_teacher_similarity_mean": 0.7809413790702819, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7800 + }, + { + "epoch": 2.2723305208030258, + "flow_matching_loss": 0.018340577837079763, + "grad_norm": 0.1601857990026474, + "learning_rate": 2.6516487474316303e-05, + "loss": 0.0292, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01462745675817132, + "rkd_loss_weight": 0.7398413240909576, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38883050382137296, + "rkd_teacher_entropy": 3.7558820486068725, + "rkd_teacher_similarity_mean": 0.7861242294311523, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7810 + }, + { + "epoch": 2.2752400349141695, + "flow_matching_loss": 0.01657079951837659, + "grad_norm": 0.3122532367706299, + "learning_rate": 2.650588619352365e-05, + "loss": 0.0276, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014891423936933278, + "rkd_loss_weight": 0.7395079910755158, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3801494002342224, + "rkd_teacher_entropy": 3.7211596965789795, + "rkd_teacher_similarity_mean": 0.7770409226417542, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7820 + }, + { + "epoch": 2.2781495490253127, + "flow_matching_loss": 0.01978601822629571, + "grad_norm": 0.1744370460510254, + "learning_rate": 2.649527093199434e-05, + "loss": 0.0303, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01427892604842782, + "rkd_loss_weight": 0.7391746342182159, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3724718689918518, + "rkd_teacher_entropy": 3.7185746908187864, + "rkd_teacher_similarity_mean": 0.7719177722930908, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7830 + }, + { + "epoch": 2.281059063136456, + "flow_matching_loss": 0.01709540272131562, + "grad_norm": 0.15424759685993195, + "learning_rate": 2.6484641702626914e-05, + "loss": 0.0258, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011842411197721957, + "rkd_loss_weight": 0.7388412952423096, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38293963074684145, + "rkd_teacher_entropy": 3.7493773460388184, + "rkd_teacher_similarity_mean": 0.7836759150028229, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7840 + }, + { + "epoch": 2.2839685772475997, + "flow_matching_loss": 0.016926297545433046, + "grad_norm": 0.18688443303108215, + "learning_rate": 2.647399851833689e-05, + "loss": 0.0272, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01397733255289495, + "rkd_loss_weight": 0.7385079562664032, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3915732830762863, + "rkd_teacher_entropy": 3.764321970939636, + "rkd_teacher_similarity_mean": 0.7901957273483277, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7850 + }, + { + "epoch": 2.286878091358743, + "flow_matching_loss": 0.017908591590821744, + "grad_norm": 0.13128238916397095, + "learning_rate": 2.6463341392056735e-05, + "loss": 0.0268, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012034815549850465, + "rkd_loss_weight": 0.7381745994091033, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37489859461784364, + "rkd_teacher_entropy": 3.7201905727386473, + "rkd_teacher_similarity_mean": 0.7738696932792664, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7860 + }, + { + "epoch": 2.2897876054698867, + "flow_matching_loss": 0.0171006572432816, + "grad_norm": 0.4052172899246216, + "learning_rate": 2.6452670336735867e-05, + "loss": 0.0288, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01590062789618969, + "rkd_loss_weight": 0.7378412663936615, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3853732168674469, + "rkd_teacher_entropy": 3.7535486936569216, + "rkd_teacher_similarity_mean": 0.783671647310257, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7870 + }, + { + "epoch": 2.29269711958103, + "flow_matching_loss": 0.0176263814792037, + "grad_norm": 0.15760080516338348, + "learning_rate": 2.644198536534063e-05, + "loss": 0.0261, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011484840791672469, + "rkd_loss_weight": 0.7375079154968261, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.378249853849411, + "rkd_teacher_entropy": 3.735841464996338, + "rkd_teacher_similarity_mean": 0.7788523852825164, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7880 + }, + { + "epoch": 2.295606633692173, + "flow_matching_loss": 0.02010435163974762, + "grad_norm": 0.18171928822994232, + "learning_rate": 2.6431286490854255e-05, + "loss": 0.0283, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011175769474357366, + "rkd_loss_weight": 0.7371745705604553, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39034366309642793, + "rkd_teacher_entropy": 3.763974142074585, + "rkd_teacher_similarity_mean": 0.787211400270462, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7890 + }, + { + "epoch": 2.298516147803317, + "flow_matching_loss": 0.018701531551778318, + "grad_norm": 0.2003522366285324, + "learning_rate": 2.6420573726276903e-05, + "loss": 0.0277, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01217447817325592, + "rkd_loss_weight": 0.736841231584549, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37636898458004, + "rkd_teacher_entropy": 3.70139684677124, + "rkd_teacher_similarity_mean": 0.7730632722377777, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7900 + }, + { + "epoch": 2.30142566191446, + "flow_matching_loss": 0.018015774246305228, + "grad_norm": 0.14970336854457855, + "learning_rate": 2.6409847084625575e-05, + "loss": 0.0276, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013006838317960501, + "rkd_loss_weight": 0.7365078806877137, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37826106548309324, + "rkd_teacher_entropy": 3.7186822652816773, + "rkd_teacher_similarity_mean": 0.7769259572029114, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7910 + }, + { + "epoch": 2.304335176025604, + "flow_matching_loss": 0.017508001066744328, + "grad_norm": 0.12881678342819214, + "learning_rate": 2.6399106578934167e-05, + "loss": 0.0264, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012033085525035857, + "rkd_loss_weight": 0.7361745417118073, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3826436251401901, + "rkd_teacher_entropy": 3.7343642711639404, + "rkd_teacher_similarity_mean": 0.7811158835887909, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7920 + }, + { + "epoch": 2.307244690136747, + "flow_matching_loss": 0.01840859614312649, + "grad_norm": 0.17564581334590912, + "learning_rate": 2.63883522222534e-05, + "loss": 0.0279, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012900380231440068, + "rkd_loss_weight": 0.7358411967754364, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3909869074821472, + "rkd_teacher_entropy": 3.7611470222473145, + "rkd_teacher_similarity_mean": 0.7873454630374909, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7930 + }, + { + "epoch": 2.3101542042478904, + "flow_matching_loss": 0.01931327721104026, + "grad_norm": 0.19855289161205292, + "learning_rate": 2.6377584027650827e-05, + "loss": 0.0295, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01386629557237029, + "rkd_loss_weight": 0.7355078458786011, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39018738865852354, + "rkd_teacher_entropy": 3.75549578666687, + "rkd_teacher_similarity_mean": 0.7850336611270905, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7940 + }, + { + "epoch": 2.313063718359034, + "flow_matching_loss": 0.018324535246938466, + "grad_norm": 0.21822437644004822, + "learning_rate": 2.636680200821083e-05, + "loss": 0.0276, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012606700602918863, + "rkd_loss_weight": 0.7351745069026947, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38008909821510317, + "rkd_teacher_entropy": 3.721890592575073, + "rkd_teacher_similarity_mean": 0.7758349239826202, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7950 + }, + { + "epoch": 2.3159732324701774, + "flow_matching_loss": 0.019810855761170386, + "grad_norm": 0.18672563135623932, + "learning_rate": 2.635600617703457e-05, + "loss": 0.0286, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012002452090382577, + "rkd_loss_weight": 0.7348411619663239, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3916609823703766, + "rkd_teacher_entropy": 3.7564741373062134, + "rkd_teacher_similarity_mean": 0.7890559017658234, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7960 + }, + { + "epoch": 2.318882746581321, + "flow_matching_loss": 0.02081832615658641, + "grad_norm": 0.17520380020141602, + "learning_rate": 2.6345196547240007e-05, + "loss": 0.0313, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014296324457973242, + "rkd_loss_weight": 0.734507817029953, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38105728924274446, + "rkd_teacher_entropy": 3.7223450183868407, + "rkd_teacher_similarity_mean": 0.7754047930240631, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7970 + }, + { + "epoch": 2.3217922606924644, + "flow_matching_loss": 0.01740659065544605, + "grad_norm": 0.21087802946567535, + "learning_rate": 2.6334373131961854e-05, + "loss": 0.0264, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01229013567790389, + "rkd_loss_weight": 0.7341744720935821, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3868465930223465, + "rkd_teacher_entropy": 3.7332995414733885, + "rkd_teacher_similarity_mean": 0.7827196180820465, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7980 + }, + { + "epoch": 2.3247017748036076, + "flow_matching_loss": 0.019663813151419164, + "grad_norm": 0.22147467732429504, + "learning_rate": 2.6323535944351583e-05, + "loss": 0.0306, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014964670594781636, + "rkd_loss_weight": 0.7338411271572113, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38652199804782866, + "rkd_teacher_entropy": 3.7408468246459963, + "rkd_teacher_similarity_mean": 0.780219167470932, + "rkd_vlm_temp": 0.10000000149011612, + "step": 7990 + }, + { + "epoch": 2.3276112889147513, + "flow_matching_loss": 0.018550648260861634, + "grad_norm": 0.26363566517829895, + "learning_rate": 2.63126849975774e-05, + "loss": 0.0301, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015781467594206334, + "rkd_loss_weight": 0.7335077822208405, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3835561603307724, + "rkd_teacher_entropy": 3.73585422039032, + "rkd_teacher_similarity_mean": 0.7775997161865235, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8000 + }, + { + "epoch": 2.3305208030258946, + "flow_matching_loss": 0.016775253508239985, + "grad_norm": 0.18599452078342438, + "learning_rate": 2.6301820304824234e-05, + "loss": 0.0283, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015740766748785974, + "rkd_loss_weight": 0.7331744432449341, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3846834808588028, + "rkd_teacher_entropy": 3.736824083328247, + "rkd_teacher_similarity_mean": 0.7826913237571717, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8010 + }, + { + "epoch": 2.3334303171370383, + "flow_matching_loss": 0.01755594350397587, + "grad_norm": 0.20267760753631592, + "learning_rate": 2.6290941879293703e-05, + "loss": 0.0268, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012650837190449238, + "rkd_loss_weight": 0.7328410923480988, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38397417664527894, + "rkd_teacher_entropy": 3.732296919822693, + "rkd_teacher_similarity_mean": 0.7816109895706177, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8020 + }, + { + "epoch": 2.3363398312481816, + "flow_matching_loss": 0.018714144453406335, + "grad_norm": 0.15005019307136536, + "learning_rate": 2.628004973420412e-05, + "loss": 0.0276, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012119608139619232, + "rkd_loss_weight": 0.7325077474117279, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37758969366550443, + "rkd_teacher_entropy": 3.729563760757446, + "rkd_teacher_similarity_mean": 0.7750311553478241, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8030 + }, + { + "epoch": 2.339249345359325, + "flow_matching_loss": 0.01945664267987013, + "grad_norm": 0.28966954350471497, + "learning_rate": 2.626914388279047e-05, + "loss": 0.0306, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015186137519776821, + "rkd_loss_weight": 0.7321744084358215, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38335191607475283, + "rkd_teacher_entropy": 3.7340717792510985, + "rkd_teacher_similarity_mean": 0.7792138636112214, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8040 + }, + { + "epoch": 2.3421588594704685, + "flow_matching_loss": 0.018895635195076466, + "grad_norm": 0.18041348457336426, + "learning_rate": 2.6258224338304394e-05, + "loss": 0.0296, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014633973874151707, + "rkd_loss_weight": 0.7318410575389862, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3885341137647629, + "rkd_teacher_entropy": 3.7502320289611815, + "rkd_teacher_similarity_mean": 0.7846149981021882, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8050 + }, + { + "epoch": 2.345068373581612, + "flow_matching_loss": 0.01901356680318713, + "grad_norm": 0.17201712727546692, + "learning_rate": 2.6247291114014166e-05, + "loss": 0.0283, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012667994387447834, + "rkd_loss_weight": 0.7315077185630798, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37790550887584684, + "rkd_teacher_entropy": 3.7157708406448364, + "rkd_teacher_similarity_mean": 0.7759126722812653, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8060 + }, + { + "epoch": 2.347977887692755, + "flow_matching_loss": 0.01820319416001439, + "grad_norm": 0.13451887667179108, + "learning_rate": 2.6236344223204683e-05, + "loss": 0.028, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01333870030939579, + "rkd_loss_weight": 0.731174373626709, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38012992441654203, + "rkd_teacher_entropy": 3.730945920944214, + "rkd_teacher_similarity_mean": 0.7779347896575928, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8070 + }, + { + "epoch": 2.3508874018038988, + "flow_matching_loss": 0.01944852378219366, + "grad_norm": 0.18100854754447937, + "learning_rate": 2.6225383679177453e-05, + "loss": 0.0291, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013247731700539589, + "rkd_loss_weight": 0.7308410227298736, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37616429924964906, + "rkd_teacher_entropy": 3.7099280834197996, + "rkd_teacher_similarity_mean": 0.7717254817485809, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8080 + }, + { + "epoch": 2.353796915915042, + "flow_matching_loss": 0.018503978103399276, + "grad_norm": 0.15561316907405853, + "learning_rate": 2.6214409495250568e-05, + "loss": 0.029, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014344207011163235, + "rkd_loss_weight": 0.7305076897144318, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.390166711807251, + "rkd_teacher_entropy": 3.7506144285202025, + "rkd_teacher_similarity_mean": 0.7854730188846588, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8090 + }, + { + "epoch": 2.3567064300261857, + "flow_matching_loss": 0.019924837723374368, + "grad_norm": 0.199973925948143, + "learning_rate": 2.620342168475869e-05, + "loss": 0.0338, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.019001664966344832, + "rkd_loss_weight": 0.730174332857132, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38846955001354216, + "rkd_teacher_entropy": 3.755199980735779, + "rkd_teacher_similarity_mean": 0.7848584473133087, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8100 + }, + { + "epoch": 2.359615944137329, + "flow_matching_loss": 0.01716543696820736, + "grad_norm": 0.17165106534957886, + "learning_rate": 2.619242026105306e-05, + "loss": 0.0261, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012202681135386228, + "rkd_loss_weight": 0.7298409938812256, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37765795886516573, + "rkd_teacher_entropy": 3.7231828689575197, + "rkd_teacher_similarity_mean": 0.7739764153957367, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8110 + }, + { + "epoch": 2.3625254582484727, + "flow_matching_loss": 0.019272705540060997, + "grad_norm": 0.21392016112804413, + "learning_rate": 2.618140523750143e-05, + "loss": 0.0314, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.016594146192073823, + "rkd_loss_weight": 0.7295076549053192, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3897701323032379, + "rkd_teacher_entropy": 3.755218267440796, + "rkd_teacher_similarity_mean": 0.786519980430603, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8120 + }, + { + "epoch": 2.365434972359616, + "flow_matching_loss": 0.01890130275860429, + "grad_norm": 0.18326157331466675, + "learning_rate": 2.6170376627488103e-05, + "loss": 0.027, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011097334930673242, + "rkd_loss_weight": 0.7291742980480194, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3850056707859039, + "rkd_teacher_entropy": 3.7337526082992554, + "rkd_teacher_similarity_mean": 0.7810908615589142, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8130 + }, + { + "epoch": 2.3683444864707592, + "flow_matching_loss": 0.01735866256058216, + "grad_norm": 0.13998344540596008, + "learning_rate": 2.6159334444413873e-05, + "loss": 0.0268, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013014594558626413, + "rkd_loss_weight": 0.7288409650325776, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38087272047996523, + "rkd_teacher_entropy": 3.7375163078308105, + "rkd_teacher_similarity_mean": 0.7785791456699371, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8140 + }, + { + "epoch": 2.371254000581903, + "flow_matching_loss": 0.019211459904909134, + "grad_norm": 0.20343172550201416, + "learning_rate": 2.614827870169604e-05, + "loss": 0.0293, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013823393872007728, + "rkd_loss_weight": 0.7285076141357422, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38947294354438783, + "rkd_teacher_entropy": 3.74032609462738, + "rkd_teacher_similarity_mean": 0.7860030949115753, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8150 + }, + { + "epoch": 2.374163514693046, + "flow_matching_loss": 0.019648867100477217, + "grad_norm": 0.2577209770679474, + "learning_rate": 2.6137209412768367e-05, + "loss": 0.0293, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013212026096880436, + "rkd_loss_weight": 0.7281742691993713, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39199581146240237, + "rkd_teacher_entropy": 3.7580047845840454, + "rkd_teacher_similarity_mean": 0.7872322082519532, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8160 + }, + { + "epoch": 2.3770730288041895, + "flow_matching_loss": 0.01912612970918417, + "grad_norm": 0.24909719824790955, + "learning_rate": 2.612612659108109e-05, + "loss": 0.0288, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013243908528238535, + "rkd_loss_weight": 0.727840930223465, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3820773154497147, + "rkd_teacher_entropy": 3.7353650093078614, + "rkd_teacher_similarity_mean": 0.7806888222694397, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8170 + }, + { + "epoch": 2.379982542915333, + "flow_matching_loss": 0.017229176498949526, + "grad_norm": 0.17636574804782867, + "learning_rate": 2.611503025010088e-05, + "loss": 0.0259, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011924209026619791, + "rkd_loss_weight": 0.7275075852870941, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39400299191474913, + "rkd_teacher_entropy": 3.766728091239929, + "rkd_teacher_similarity_mean": 0.7926566660404205, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8180 + }, + { + "epoch": 2.3828920570264764, + "flow_matching_loss": 0.02062131240963936, + "grad_norm": 0.2461921125650406, + "learning_rate": 2.6103920403310832e-05, + "loss": 0.0301, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013067641528323293, + "rkd_loss_weight": 0.7271742403507233, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37909104824066164, + "rkd_teacher_entropy": 3.723238492012024, + "rkd_teacher_similarity_mean": 0.7745481371879578, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8190 + }, + { + "epoch": 2.38580157113762, + "flow_matching_loss": 0.018042491842061283, + "grad_norm": 0.2475343644618988, + "learning_rate": 2.6092797064210465e-05, + "loss": 0.028, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01375831626355648, + "rkd_loss_weight": 0.7268408954143524, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38334660828113554, + "rkd_teacher_entropy": 3.7385896682739257, + "rkd_teacher_similarity_mean": 0.7808352768421173, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8200 + }, + { + "epoch": 2.3887110852487634, + "flow_matching_loss": 0.018326980341225863, + "grad_norm": 0.15720908343791962, + "learning_rate": 2.6081660246315685e-05, + "loss": 0.0286, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014099675975739957, + "rkd_loss_weight": 0.7265075504779815, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3768814504146576, + "rkd_teacher_entropy": 3.729821801185608, + "rkd_teacher_similarity_mean": 0.7767616748809815, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8210 + }, + { + "epoch": 2.391620599359907, + "flow_matching_loss": 0.01657602293416858, + "grad_norm": 0.25920966267585754, + "learning_rate": 2.6070509963158774e-05, + "loss": 0.0253, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012000379711389541, + "rkd_loss_weight": 0.7261742055416107, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3927830159664154, + "rkd_teacher_entropy": 3.7715575218200685, + "rkd_teacher_similarity_mean": 0.7911542892456055, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8220 + }, + { + "epoch": 2.3945301134710504, + "flow_matching_loss": 0.01709804581478238, + "grad_norm": 0.1951572448015213, + "learning_rate": 2.6059346228288375e-05, + "loss": 0.0261, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012435140181332826, + "rkd_loss_weight": 0.7258408665657043, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38674250841140745, + "rkd_teacher_entropy": 3.7527935028076174, + "rkd_teacher_similarity_mean": 0.7838770687580109, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8230 + }, + { + "epoch": 2.3974396275821936, + "flow_matching_loss": 0.018683134764432906, + "grad_norm": 0.17486706376075745, + "learning_rate": 2.604816905526948e-05, + "loss": 0.0283, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013277212157845497, + "rkd_loss_weight": 0.7255075156688691, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3792766958475113, + "rkd_teacher_entropy": 3.724137234687805, + "rkd_teacher_similarity_mean": 0.778274130821228, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8240 + }, + { + "epoch": 2.4003491416933374, + "flow_matching_loss": 0.0188818053342402, + "grad_norm": 0.16321948170661926, + "learning_rate": 2.603697845768341e-05, + "loss": 0.0278, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012302345130592585, + "rkd_loss_weight": 0.7251741707324981, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3810563564300537, + "rkd_teacher_entropy": 3.7426843881607055, + "rkd_teacher_similarity_mean": 0.7814385414123535, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8250 + }, + { + "epoch": 2.4032586558044806, + "flow_matching_loss": 0.0186331519857049, + "grad_norm": 0.24100103974342346, + "learning_rate": 2.6025774449127792e-05, + "loss": 0.0287, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013832775875926018, + "rkd_loss_weight": 0.7248408317565918, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3826424181461334, + "rkd_teacher_entropy": 3.7468353509902954, + "rkd_teacher_similarity_mean": 0.7812629997730255, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8260 + }, + { + "epoch": 2.406168169915624, + "flow_matching_loss": 0.017747949808835983, + "grad_norm": 0.13695122301578522, + "learning_rate": 2.6014557043216562e-05, + "loss": 0.0258, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011155752185732126, + "rkd_loss_weight": 0.7245074808597565, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3768328070640564, + "rkd_teacher_entropy": 3.724242067337036, + "rkd_teacher_similarity_mean": 0.7741575241088867, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8270 + }, + { + "epoch": 2.4090776840267676, + "flow_matching_loss": 0.017999306693673132, + "grad_norm": 0.16062003374099731, + "learning_rate": 2.6003326253579913e-05, + "loss": 0.026, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011097407573834062, + "rkd_loss_weight": 0.7241741418838501, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38569627702236176, + "rkd_teacher_entropy": 3.7500746726989744, + "rkd_teacher_similarity_mean": 0.7858865559101105, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8280 + }, + { + "epoch": 2.411987198137911, + "flow_matching_loss": 0.019121417496353387, + "grad_norm": 0.18193106353282928, + "learning_rate": 2.599208209386432e-05, + "loss": 0.0282, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012480533309280872, + "rkd_loss_weight": 0.7238407969474793, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3838197022676468, + "rkd_teacher_entropy": 3.750657081604004, + "rkd_teacher_similarity_mean": 0.7823543131351471, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8290 + }, + { + "epoch": 2.4148967122490546, + "flow_matching_loss": 0.017145617678761483, + "grad_norm": 0.28946036100387573, + "learning_rate": 2.5980824577732502e-05, + "loss": 0.0267, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013176233414560557, + "rkd_loss_weight": 0.7235074460506439, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3782353550195694, + "rkd_teacher_entropy": 3.709608554840088, + "rkd_teacher_similarity_mean": 0.772143691778183, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8300 + }, + { + "epoch": 2.417806226360198, + "flow_matching_loss": 0.019739540852606298, + "grad_norm": 0.18827955424785614, + "learning_rate": 2.5969553718863395e-05, + "loss": 0.0279, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011311911791563035, + "rkd_loss_weight": 0.7231741130352021, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3846290946006775, + "rkd_teacher_entropy": 3.7488078117370605, + "rkd_teacher_similarity_mean": 0.7827420055866241, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8310 + }, + { + "epoch": 2.4207157404713415, + "flow_matching_loss": 0.01844852799549699, + "grad_norm": 0.17492301762104034, + "learning_rate": 2.595826953095216e-05, + "loss": 0.0282, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013544977642595768, + "rkd_loss_weight": 0.7228407561779022, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3809160977602005, + "rkd_teacher_entropy": 3.725356698036194, + "rkd_teacher_similarity_mean": 0.7786648154258728, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8320 + }, + { + "epoch": 2.423625254582485, + "flow_matching_loss": 0.018937273323535918, + "grad_norm": 0.16263137757778168, + "learning_rate": 2.594697202771014e-05, + "loss": 0.0278, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012314414232969284, + "rkd_loss_weight": 0.7225074172019958, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3818497359752655, + "rkd_teacher_entropy": 3.72094521522522, + "rkd_teacher_similarity_mean": 0.7769818127155304, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8330 + }, + { + "epoch": 2.426534768693628, + "flow_matching_loss": 0.01809562873095274, + "grad_norm": 0.19122973084449768, + "learning_rate": 2.5935661222864888e-05, + "loss": 0.0272, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01265520639717579, + "rkd_loss_weight": 0.7221740782260895, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3812875360250473, + "rkd_teacher_entropy": 3.7353700160980225, + "rkd_teacher_similarity_mean": 0.7791438698768616, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8340 + }, + { + "epoch": 2.4294442828047718, + "flow_matching_loss": 0.02044216599315405, + "grad_norm": 0.22249920666217804, + "learning_rate": 2.592433713016008e-05, + "loss": 0.0301, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013371307961642742, + "rkd_loss_weight": 0.7218407213687896, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3739199429750443, + "rkd_teacher_entropy": 3.7189032077789306, + "rkd_teacher_similarity_mean": 0.7720774710178375, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8350 + }, + { + "epoch": 2.432353796915915, + "flow_matching_loss": 0.017953427322208882, + "grad_norm": 0.1722015142440796, + "learning_rate": 2.5912999763355553e-05, + "loss": 0.0268, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012318678665906191, + "rkd_loss_weight": 0.7215073883533478, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38267157077789304, + "rkd_teacher_entropy": 3.7365429401397705, + "rkd_teacher_similarity_mean": 0.781508082151413, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8360 + }, + { + "epoch": 2.4352633110270583, + "flow_matching_loss": 0.015214350447058678, + "grad_norm": 0.1404884308576584, + "learning_rate": 2.590164913622729e-05, + "loss": 0.025, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013618897181004286, + "rkd_loss_weight": 0.7211740374565124, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3815473735332489, + "rkd_teacher_entropy": 3.7344226360321047, + "rkd_teacher_similarity_mean": 0.7785586237907409, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8370 + }, + { + "epoch": 2.438172825138202, + "flow_matching_loss": 0.01801345469430089, + "grad_norm": 0.14659447968006134, + "learning_rate": 2.589028526256737e-05, + "loss": 0.0267, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011996197048574686, + "rkd_loss_weight": 0.7208406925201416, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39006083011627196, + "rkd_teacher_entropy": 3.7596494436264036, + "rkd_teacher_similarity_mean": 0.784942239522934, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8380 + }, + { + "epoch": 2.4410823392493453, + "flow_matching_loss": 0.017158571537584065, + "grad_norm": 0.12227033078670502, + "learning_rate": 2.587890815618397e-05, + "loss": 0.0269, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013540877494961023, + "rkd_loss_weight": 0.7205073535442352, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38788300156593325, + "rkd_teacher_entropy": 3.7486963748931883, + "rkd_teacher_similarity_mean": 0.7844990789890289, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8390 + }, + { + "epoch": 2.443991853360489, + "flow_matching_loss": 0.016668780986219645, + "grad_norm": 0.1638348549604416, + "learning_rate": 2.5867517830901343e-05, + "loss": 0.0248, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01126501876860857, + "rkd_loss_weight": 0.7201740026474, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38388989865779877, + "rkd_teacher_entropy": 3.727070426940918, + "rkd_teacher_similarity_mean": 0.7789360880851746, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8400 + }, + { + "epoch": 2.4469013674716322, + "flow_matching_loss": 0.018977110739797354, + "grad_norm": 0.20866483449935913, + "learning_rate": 2.5856114300559812e-05, + "loss": 0.0277, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01216010032221675, + "rkd_loss_weight": 0.7198406636714936, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3916279703378677, + "rkd_teacher_entropy": 3.762866997718811, + "rkd_teacher_similarity_mean": 0.7891597986221314, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8410 + }, + { + "epoch": 2.4498108815827755, + "flow_matching_loss": 0.018174911476671697, + "grad_norm": 0.2187003791332245, + "learning_rate": 2.584469757901574e-05, + "loss": 0.0263, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011258350498974324, + "rkd_loss_weight": 0.7195073187351226, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3796356111764908, + "rkd_teacher_entropy": 3.721190905570984, + "rkd_teacher_similarity_mean": 0.7758416891098022, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8420 + }, + { + "epoch": 2.452720395693919, + "flow_matching_loss": 0.017044569924473764, + "grad_norm": 0.1556888073682785, + "learning_rate": 2.5833267680141526e-05, + "loss": 0.0263, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012895033787935972, + "rkd_loss_weight": 0.7191739678382874, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38091676533222196, + "rkd_teacher_entropy": 3.730309081077576, + "rkd_teacher_similarity_mean": 0.7762748956680298, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8430 + }, + { + "epoch": 2.4556299098050625, + "flow_matching_loss": 0.019310108665376902, + "grad_norm": 0.1632586270570755, + "learning_rate": 2.5821824617825564e-05, + "loss": 0.0279, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011990048736333848, + "rkd_loss_weight": 0.718840628862381, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38329221606254577, + "rkd_teacher_entropy": 3.73799090385437, + "rkd_teacher_similarity_mean": 0.7806476294994354, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8440 + }, + { + "epoch": 2.458539423916206, + "flow_matching_loss": 0.021966111566871404, + "grad_norm": 0.17128463089466095, + "learning_rate": 2.5810368405972265e-05, + "loss": 0.0303, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011635812651365995, + "rkd_loss_weight": 0.7185072839260102, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38463248312473297, + "rkd_teacher_entropy": 3.750053811073303, + "rkd_teacher_similarity_mean": 0.7835141539573669, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8450 + }, + { + "epoch": 2.4614489380273494, + "flow_matching_loss": 0.018714682664722204, + "grad_norm": 0.173572838306427, + "learning_rate": 2.5798899058501996e-05, + "loss": 0.0283, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013330879528075457, + "rkd_loss_weight": 0.7181739389896393, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3823684871196747, + "rkd_teacher_entropy": 3.742889928817749, + "rkd_teacher_similarity_mean": 0.7795866131782532, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8460 + }, + { + "epoch": 2.4643584521384927, + "flow_matching_loss": 0.018303281534463166, + "grad_norm": 0.1917596161365509, + "learning_rate": 2.578741658935111e-05, + "loss": 0.0275, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012810674868524075, + "rkd_loss_weight": 0.7178405940532684, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3779431253671646, + "rkd_teacher_entropy": 3.71656174659729, + "rkd_teacher_similarity_mean": 0.7741257131099701, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8470 + }, + { + "epoch": 2.4672679662496364, + "flow_matching_loss": 0.018189545441418885, + "grad_norm": 0.1586061269044876, + "learning_rate": 2.5775921012471875e-05, + "loss": 0.0271, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01242821617051959, + "rkd_loss_weight": 0.7175072491168976, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38592506349086764, + "rkd_teacher_entropy": 3.754181909561157, + "rkd_teacher_similarity_mean": 0.7838948667049408, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8480 + }, + { + "epoch": 2.4701774803607797, + "flow_matching_loss": 0.017402077838778496, + "grad_norm": 0.24778270721435547, + "learning_rate": 2.5764412341832505e-05, + "loss": 0.0283, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015242537204176187, + "rkd_loss_weight": 0.7171739041805267, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3917230486869812, + "rkd_teacher_entropy": 3.7508103370666506, + "rkd_teacher_similarity_mean": 0.786679595708847, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8490 + }, + { + "epoch": 2.4730869944719234, + "flow_matching_loss": 0.020523922611027957, + "grad_norm": 0.17943724989891052, + "learning_rate": 2.5752890591417125e-05, + "loss": 0.0296, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012703771889209747, + "rkd_loss_weight": 0.7168405652046204, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38938920199871063, + "rkd_teacher_entropy": 3.75075421333313, + "rkd_teacher_similarity_mean": 0.7848558366298676, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8500 + }, + { + "epoch": 2.4759965085830666, + "flow_matching_loss": 0.017556928750127553, + "grad_norm": 0.16658055782318115, + "learning_rate": 2.5741355775225744e-05, + "loss": 0.0264, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012407641531899572, + "rkd_loss_weight": 0.7165072143077851, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39154318869113924, + "rkd_teacher_entropy": 3.761663627624512, + "rkd_teacher_similarity_mean": 0.7872033059597016, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8510 + }, + { + "epoch": 2.47890602269421, + "flow_matching_loss": 0.017480165138840674, + "grad_norm": 0.16866299510002136, + "learning_rate": 2.5729807907274254e-05, + "loss": 0.0276, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014144924841821194, + "rkd_loss_weight": 0.7161738693714141, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38580514788627623, + "rkd_teacher_entropy": 3.754978060722351, + "rkd_teacher_similarity_mean": 0.7828546404838562, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8520 + }, + { + "epoch": 2.4818155368053536, + "flow_matching_loss": 0.01734958179295063, + "grad_norm": 0.173502117395401, + "learning_rate": 2.5718247001594403e-05, + "loss": 0.0245, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009976821206510066, + "rkd_loss_weight": 0.7158405303955078, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38296072483062743, + "rkd_teacher_entropy": 3.728889489173889, + "rkd_teacher_similarity_mean": 0.7789979934692383, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8530 + }, + { + "epoch": 2.484725050916497, + "flow_matching_loss": 0.019890543725341558, + "grad_norm": 0.208493173122406, + "learning_rate": 2.5706673072233776e-05, + "loss": 0.0292, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013080185186117888, + "rkd_loss_weight": 0.7155071794986725, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3757369488477707, + "rkd_teacher_entropy": 3.716268229484558, + "rkd_teacher_similarity_mean": 0.7725661754608154, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8540 + }, + { + "epoch": 2.4876345650276406, + "flow_matching_loss": 0.01643528938293457, + "grad_norm": 0.20011308789253235, + "learning_rate": 2.5695086133255795e-05, + "loss": 0.024, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010571740055456757, + "rkd_loss_weight": 0.7151738405227661, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38587100207805636, + "rkd_teacher_entropy": 3.747970986366272, + "rkd_teacher_similarity_mean": 0.7822589755058289, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8550 + }, + { + "epoch": 2.490544079138784, + "flow_matching_loss": 0.01882321946322918, + "grad_norm": 0.23457346856594086, + "learning_rate": 2.5683486198739687e-05, + "loss": 0.0281, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012956405058503151, + "rkd_loss_weight": 0.7148404955863953, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3849461078643799, + "rkd_teacher_entropy": 3.7412002325057983, + "rkd_teacher_similarity_mean": 0.7794536709785461, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8560 + }, + { + "epoch": 2.493453593249927, + "flow_matching_loss": 0.0217879525385797, + "grad_norm": 0.12702378630638123, + "learning_rate": 2.5671873282780452e-05, + "loss": 0.0316, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01373451789841056, + "rkd_loss_weight": 0.7145071446895599, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.383145272731781, + "rkd_teacher_entropy": 3.736411476135254, + "rkd_teacher_similarity_mean": 0.7805662035942078, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8570 + }, + { + "epoch": 2.496363107361071, + "flow_matching_loss": 0.017897674720734356, + "grad_norm": 0.2037818729877472, + "learning_rate": 2.5660247399488887e-05, + "loss": 0.0264, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01187034291215241, + "rkd_loss_weight": 0.7141738116741181, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3893995344638824, + "rkd_teacher_entropy": 3.7557517528533935, + "rkd_teacher_similarity_mean": 0.7871295630931854, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8580 + }, + { + "epoch": 2.499272621472214, + "flow_matching_loss": 0.0175727479159832, + "grad_norm": 0.16885127127170563, + "learning_rate": 2.564860856299154e-05, + "loss": 0.0279, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01442914865911007, + "rkd_loss_weight": 0.7138404548168182, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38443140387535096, + "rkd_teacher_entropy": 3.7176629066467286, + "rkd_teacher_similarity_mean": 0.7791047275066376, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8590 + }, + { + "epoch": 2.5021821355833573, + "flow_matching_loss": 0.017179602477699518, + "grad_norm": 0.24309663474559784, + "learning_rate": 2.5636956787430693e-05, + "loss": 0.0271, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013930595014244318, + "rkd_loss_weight": 0.7135071158409119, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38299046754837035, + "rkd_teacher_entropy": 3.7363807201385497, + "rkd_teacher_similarity_mean": 0.7776327788829803, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8600 + }, + { + "epoch": 2.505091649694501, + "flow_matching_loss": 0.019574340526014568, + "grad_norm": 0.20723001658916473, + "learning_rate": 2.5625292086964346e-05, + "loss": 0.029, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013286503218114375, + "rkd_loss_weight": 0.7131737768650055, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38021266758441924, + "rkd_teacher_entropy": 3.722288656234741, + "rkd_teacher_similarity_mean": 0.7762866199016571, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8610 + }, + { + "epoch": 2.5080011638056443, + "flow_matching_loss": 0.018412729352712633, + "grad_norm": 0.14543725550174713, + "learning_rate": 2.5613614475766212e-05, + "loss": 0.0295, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015518234204500913, + "rkd_loss_weight": 0.7128404259681702, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39641194939613345, + "rkd_teacher_entropy": 3.774184226989746, + "rkd_teacher_similarity_mean": 0.789321768283844, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8620 + }, + { + "epoch": 2.510910677916788, + "flow_matching_loss": 0.01920371549203992, + "grad_norm": 0.13449640572071075, + "learning_rate": 2.5601923968025694e-05, + "loss": 0.0282, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012675356771796941, + "rkd_loss_weight": 0.7125070869922638, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38669543266296386, + "rkd_teacher_entropy": 3.726621150970459, + "rkd_teacher_similarity_mean": 0.7800732731819153, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8630 + }, + { + "epoch": 2.5138201920279313, + "flow_matching_loss": 0.01946018012240529, + "grad_norm": 0.15766017138957977, + "learning_rate": 2.5590220577947862e-05, + "loss": 0.0297, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01441351156681776, + "rkd_loss_weight": 0.7121737420558929, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38454064130783083, + "rkd_teacher_entropy": 3.732128405570984, + "rkd_teacher_similarity_mean": 0.7785231292247772, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8640 + }, + { + "epoch": 2.516729706139075, + "flow_matching_loss": 0.01808817544952035, + "grad_norm": 0.16538676619529724, + "learning_rate": 2.5578504319753427e-05, + "loss": 0.0253, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010127993952482938, + "rkd_loss_weight": 0.7118403911590576, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3852589398622513, + "rkd_teacher_entropy": 3.738128352165222, + "rkd_teacher_similarity_mean": 0.781784611940384, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8650 + }, + { + "epoch": 2.5196392202502182, + "flow_matching_loss": 0.01745551759377122, + "grad_norm": 0.16225825250148773, + "learning_rate": 2.556677520767876e-05, + "loss": 0.0255, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011368276737630367, + "rkd_loss_weight": 0.7115070521831512, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3848963737487793, + "rkd_teacher_entropy": 3.7336968421936034, + "rkd_teacher_similarity_mean": 0.7797097623348236, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8660 + }, + { + "epoch": 2.5225487343613615, + "flow_matching_loss": 0.019519472122192384, + "grad_norm": 0.21573449671268463, + "learning_rate": 2.5555033255975835e-05, + "loss": 0.0297, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014253245014697314, + "rkd_loss_weight": 0.7111737072467804, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37502020299434663, + "rkd_teacher_entropy": 3.69847776889801, + "rkd_teacher_similarity_mean": 0.767296576499939, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8670 + }, + { + "epoch": 2.525458248472505, + "flow_matching_loss": 0.019564421754330395, + "grad_norm": 0.17525547742843628, + "learning_rate": 2.5543278478912228e-05, + "loss": 0.0287, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012798640318214894, + "rkd_loss_weight": 0.7108403623104096, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3788935273885727, + "rkd_teacher_entropy": 3.7134981632232664, + "rkd_teacher_similarity_mean": 0.7732724666595459, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8680 + }, + { + "epoch": 2.5283677625836485, + "flow_matching_loss": 0.016339343413710593, + "grad_norm": 0.17088869214057922, + "learning_rate": 2.55315108907711e-05, + "loss": 0.0243, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011153245717287064, + "rkd_loss_weight": 0.7105070173740387, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38311020731925965, + "rkd_teacher_entropy": 3.7293068885803224, + "rkd_teacher_similarity_mean": 0.7772112905979156, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8690 + }, + { + "epoch": 2.5312772766947917, + "flow_matching_loss": 0.01578378854319453, + "grad_norm": 0.16493678092956543, + "learning_rate": 2.551973050585119e-05, + "loss": 0.0261, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014542076969519258, + "rkd_loss_weight": 0.7101736724376678, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3844698011875153, + "rkd_teacher_entropy": 3.7232231378555296, + "rkd_teacher_similarity_mean": 0.7791569113731385, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8700 + }, + { + "epoch": 2.5341867908059355, + "flow_matching_loss": 0.02021712874993682, + "grad_norm": 0.18827159702777863, + "learning_rate": 2.5507937338466768e-05, + "loss": 0.0284, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011510658822953702, + "rkd_loss_weight": 0.709840327501297, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3822333961725235, + "rkd_teacher_entropy": 3.7364559888839723, + "rkd_teacher_similarity_mean": 0.7776750385761261, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8710 + }, + { + "epoch": 2.5370963049170787, + "flow_matching_loss": 0.01859250757843256, + "grad_norm": 0.13888831436634064, + "learning_rate": 2.549613140294764e-05, + "loss": 0.0297, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01566348378546536, + "rkd_loss_weight": 0.7095069885253906, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.383224293589592, + "rkd_teacher_entropy": 3.7243700504302977, + "rkd_teacher_similarity_mean": 0.7778166830539703, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8720 + }, + { + "epoch": 2.5400058190282224, + "flow_matching_loss": 0.016992334090173245, + "grad_norm": 0.2037404179573059, + "learning_rate": 2.548431271363914e-05, + "loss": 0.0256, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01210123640485108, + "rkd_loss_weight": 0.7091736376285553, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.386194697022438, + "rkd_teacher_entropy": 3.7454323768615723, + "rkd_teacher_similarity_mean": 0.7816722333431244, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8730 + }, + { + "epoch": 2.5429153331393657, + "flow_matching_loss": 0.017825005855411292, + "grad_norm": 0.13893504440784454, + "learning_rate": 2.5472481284902092e-05, + "loss": 0.0248, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009865587949752808, + "rkd_loss_weight": 0.7088402926921844, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3886842876672745, + "rkd_teacher_entropy": 3.759353184700012, + "rkd_teacher_similarity_mean": 0.7859561681747437, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8740 + }, + { + "epoch": 2.5458248472505094, + "flow_matching_loss": 0.01624065488576889, + "grad_norm": 0.17925061285495758, + "learning_rate": 2.5460637131112788e-05, + "loss": 0.0269, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015060307458043098, + "rkd_loss_weight": 0.708506953716278, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3891061395406723, + "rkd_teacher_entropy": 3.762107491493225, + "rkd_teacher_similarity_mean": 0.785437422990799, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8750 + }, + { + "epoch": 2.5487343613616527, + "flow_matching_loss": 0.015736687183380126, + "grad_norm": 0.17210283875465393, + "learning_rate": 2.5448780266662995e-05, + "loss": 0.0243, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012053737509995699, + "rkd_loss_weight": 0.7081736028194427, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39490775763988495, + "rkd_teacher_entropy": 3.762908911705017, + "rkd_teacher_similarity_mean": 0.790894627571106, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8760 + }, + { + "epoch": 2.551643875472796, + "flow_matching_loss": 0.02043408704921603, + "grad_norm": 0.1808985024690628, + "learning_rate": 2.5436910705959926e-05, + "loss": 0.029, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012151595205068588, + "rkd_loss_weight": 0.7078402638435364, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.390019890666008, + "rkd_teacher_entropy": 3.7628681898117065, + "rkd_teacher_similarity_mean": 0.7872962176799774, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8770 + }, + { + "epoch": 2.5545533895839396, + "flow_matching_loss": 0.018628019839525223, + "grad_norm": 0.1522637903690338, + "learning_rate": 2.542502846342621e-05, + "loss": 0.0284, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013757362868636847, + "rkd_loss_weight": 0.7075069189071655, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38860071897506715, + "rkd_teacher_entropy": 3.7535714864730836, + "rkd_teacher_similarity_mean": 0.782898223400116, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8780 + }, + { + "epoch": 2.557462903695083, + "flow_matching_loss": 0.018865940533578394, + "grad_norm": 0.19024744629859924, + "learning_rate": 2.5413133553499892e-05, + "loss": 0.0276, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012352557852864265, + "rkd_loss_weight": 0.7071735680103302, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38052729070186614, + "rkd_teacher_entropy": 3.7293768167495727, + "rkd_teacher_similarity_mean": 0.7755916893482209, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8790 + }, + { + "epoch": 2.560372417806226, + "flow_matching_loss": 0.015573482681065797, + "grad_norm": 0.20993945002555847, + "learning_rate": 2.540122599063442e-05, + "loss": 0.0262, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01497627217322588, + "rkd_loss_weight": 0.7068402349948884, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3852348357439041, + "rkd_teacher_entropy": 3.7308809995651244, + "rkd_teacher_similarity_mean": 0.7802899122238159, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8800 + }, + { + "epoch": 2.56328193191737, + "flow_matching_loss": 0.016287091281265022, + "grad_norm": 0.1346064954996109, + "learning_rate": 2.538930578929859e-05, + "loss": 0.0251, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012457764241844416, + "rkd_loss_weight": 0.7065068781375885, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3748993694782257, + "rkd_teacher_entropy": 3.7021719932556154, + "rkd_teacher_similarity_mean": 0.7680875480175018, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8810 + }, + { + "epoch": 2.566191446028513, + "flow_matching_loss": 0.018320351000875236, + "grad_norm": 0.17432266473770142, + "learning_rate": 2.537737296397658e-05, + "loss": 0.0271, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012457852298393846, + "rkd_loss_weight": 0.7061735391616821, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3799802988767624, + "rkd_teacher_entropy": 3.7307323694229124, + "rkd_teacher_similarity_mean": 0.7761058628559112, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8820 + }, + { + "epoch": 2.569100960139657, + "flow_matching_loss": 0.018077059742063285, + "grad_norm": 0.2351449429988861, + "learning_rate": 2.5365427529167903e-05, + "loss": 0.0267, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01216946537606418, + "rkd_loss_weight": 0.7058402001857758, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38105922043323515, + "rkd_teacher_entropy": 3.7193631887435914, + "rkd_teacher_similarity_mean": 0.7770646333694458, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8830 + }, + { + "epoch": 2.5720104742508, + "flow_matching_loss": 0.020685928501188756, + "grad_norm": 0.1425979882478714, + "learning_rate": 2.5353469499387375e-05, + "loss": 0.0286, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011245740205049514, + "rkd_loss_weight": 0.7055068433284759, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3784038066864014, + "rkd_teacher_entropy": 3.7203436136245727, + "rkd_teacher_similarity_mean": 0.773213654756546, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8840 + }, + { + "epoch": 2.574919988361944, + "flow_matching_loss": 0.019729139190167187, + "grad_norm": 0.16533653438091278, + "learning_rate": 2.534149888916514e-05, + "loss": 0.0284, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012332318164408208, + "rkd_loss_weight": 0.7051735103130341, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3872143507003784, + "rkd_teacher_entropy": 3.747303009033203, + "rkd_teacher_similarity_mean": 0.7820237755775452, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8850 + }, + { + "epoch": 2.577829502473087, + "flow_matching_loss": 0.018265264108777046, + "grad_norm": 0.1896127462387085, + "learning_rate": 2.5329515713046616e-05, + "loss": 0.0275, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013056747848168015, + "rkd_loss_weight": 0.7048401594161987, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38424421548843385, + "rkd_teacher_entropy": 3.7253708839416504, + "rkd_teacher_similarity_mean": 0.7782369315624237, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8860 + }, + { + "epoch": 2.5807390165842303, + "flow_matching_loss": 0.02018498545512557, + "grad_norm": 0.22727365791797638, + "learning_rate": 2.531751998559249e-05, + "loss": 0.0295, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01317142378538847, + "rkd_loss_weight": 0.7045068144798279, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38450071811676023, + "rkd_teacher_entropy": 3.744777250289917, + "rkd_teacher_similarity_mean": 0.7810831725597381, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8870 + }, + { + "epoch": 2.583648530695374, + "flow_matching_loss": 0.017385839484632015, + "grad_norm": 0.1912308931350708, + "learning_rate": 2.5305511721378703e-05, + "loss": 0.0263, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012690640892833472, + "rkd_loss_weight": 0.7041734755039215, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38685753047466276, + "rkd_teacher_entropy": 3.7492602348327635, + "rkd_teacher_similarity_mean": 0.7818757951259613, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8880 + }, + { + "epoch": 2.5865580448065173, + "flow_matching_loss": 0.018408370297402145, + "grad_norm": 0.3590984046459198, + "learning_rate": 2.529349093499642e-05, + "loss": 0.027, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012274744873866438, + "rkd_loss_weight": 0.7038401246070862, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3875593483448029, + "rkd_teacher_entropy": 3.745058512687683, + "rkd_teacher_similarity_mean": 0.7820533156394959, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8890 + }, + { + "epoch": 2.5894675589176606, + "flow_matching_loss": 0.018518927693367004, + "grad_norm": 0.14522817730903625, + "learning_rate": 2.528145764105205e-05, + "loss": 0.0261, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010834657587110996, + "rkd_loss_weight": 0.7035067856311799, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3921065777540207, + "rkd_teacher_entropy": 3.7713464736938476, + "rkd_teacher_similarity_mean": 0.7873857140541076, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8900 + }, + { + "epoch": 2.5923770730288043, + "flow_matching_loss": 0.01723527368158102, + "grad_norm": 0.1872747242450714, + "learning_rate": 2.5269411854167166e-05, + "loss": 0.0268, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013600823748856784, + "rkd_loss_weight": 0.7031734406948089, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3907622009515762, + "rkd_teacher_entropy": 3.7523523330688477, + "rkd_teacher_similarity_mean": 0.7854729175567627, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8910 + }, + { + "epoch": 2.5952865871399475, + "flow_matching_loss": 0.018714037258177996, + "grad_norm": 0.15581005811691284, + "learning_rate": 2.5257353588978534e-05, + "loss": 0.0281, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013322603330016137, + "rkd_loss_weight": 0.7028400897979736, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3834346681833267, + "rkd_teacher_entropy": 3.728838014602661, + "rkd_teacher_similarity_mean": 0.7795731663703919, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8920 + }, + { + "epoch": 2.598196101251091, + "flow_matching_loss": 0.02119495989754796, + "grad_norm": 0.16547076404094696, + "learning_rate": 2.5245282860138083e-05, + "loss": 0.0299, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01242131246253848, + "rkd_loss_weight": 0.7025067508220673, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38573433756828307, + "rkd_teacher_entropy": 3.7451585292816163, + "rkd_teacher_similarity_mean": 0.7798103272914887, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8930 + }, + { + "epoch": 2.6011056153622345, + "flow_matching_loss": 0.019671991001814604, + "grad_norm": 0.20767320692539215, + "learning_rate": 2.5233199682312894e-05, + "loss": 0.0275, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011142675578594208, + "rkd_loss_weight": 0.7021734058856964, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38538992404937744, + "rkd_teacher_entropy": 3.7519113302230833, + "rkd_teacher_similarity_mean": 0.7833747088909149, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8940 + }, + { + "epoch": 2.604015129473378, + "flow_matching_loss": 0.01721403105184436, + "grad_norm": 0.1473170965909958, + "learning_rate": 2.5221104070185163e-05, + "loss": 0.0247, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010683509707450866, + "rkd_loss_weight": 0.7018400609493256, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38450529873371125, + "rkd_teacher_entropy": 3.744655466079712, + "rkd_teacher_similarity_mean": 0.7835772097110748, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8950 + }, + { + "epoch": 2.6069246435845215, + "flow_matching_loss": 0.02011346761137247, + "grad_norm": 0.14094766974449158, + "learning_rate": 2.5208996038452204e-05, + "loss": 0.0276, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010689407680183649, + "rkd_loss_weight": 0.7015067160129547, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3863088756799698, + "rkd_teacher_entropy": 3.7397205114364622, + "rkd_teacher_similarity_mean": 0.782031100988388, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8960 + }, + { + "epoch": 2.6098341576956647, + "flow_matching_loss": 0.019003946520388126, + "grad_norm": 0.1448926031589508, + "learning_rate": 2.5196875601826403e-05, + "loss": 0.0273, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011795808188617229, + "rkd_loss_weight": 0.7011733710765838, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38308608531951904, + "rkd_teacher_entropy": 3.731994390487671, + "rkd_teacher_similarity_mean": 0.7781138300895691, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8970 + }, + { + "epoch": 2.6127436718068084, + "flow_matching_loss": 0.018424658197909595, + "grad_norm": 0.17809125781059265, + "learning_rate": 2.518474277503525e-05, + "loss": 0.0278, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013380550825968385, + "rkd_loss_weight": 0.700840026140213, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38356623947620394, + "rkd_teacher_entropy": 3.7381360054016115, + "rkd_teacher_similarity_mean": 0.7794567227363587, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8980 + }, + { + "epoch": 2.6156531859179517, + "flow_matching_loss": 0.019505139905959368, + "grad_norm": 0.12734650075435638, + "learning_rate": 2.5172597572821266e-05, + "loss": 0.0291, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01375710591673851, + "rkd_loss_weight": 0.7005066871643066, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38751098811626433, + "rkd_teacher_entropy": 3.736536383628845, + "rkd_teacher_similarity_mean": 0.7805849730968475, + "rkd_vlm_temp": 0.10000000149011612, + "step": 8990 + }, + { + "epoch": 2.618562700029095, + "flow_matching_loss": 0.018095537833869458, + "grad_norm": 0.14268432557582855, + "learning_rate": 2.5160440009942015e-05, + "loss": 0.0256, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010710478480905294, + "rkd_loss_weight": 0.7001733362674714, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39250577688217164, + "rkd_teacher_entropy": 3.754500079154968, + "rkd_teacher_similarity_mean": 0.7876978397369385, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9000 + }, + { + "epoch": 2.6214722141402387, + "flow_matching_loss": 0.019387014769017698, + "grad_norm": 0.1913038045167923, + "learning_rate": 2.514827010117009e-05, + "loss": 0.0271, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01107975160703063, + "rkd_loss_weight": 0.6998399913311004, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3833264082670212, + "rkd_teacher_entropy": 3.732564091682434, + "rkd_teacher_similarity_mean": 0.7753001272678375, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9010 + }, + { + "epoch": 2.624381728251382, + "flow_matching_loss": 0.018616623897105455, + "grad_norm": 0.21900847554206848, + "learning_rate": 2.5136087861293068e-05, + "loss": 0.0278, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013089630566537381, + "rkd_loss_weight": 0.699506652355194, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39111179709434507, + "rkd_teacher_entropy": 3.7601494550704957, + "rkd_teacher_similarity_mean": 0.7864228963851929, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9020 + }, + { + "epoch": 2.627291242362525, + "flow_matching_loss": 0.019661280419677497, + "grad_norm": 0.212416872382164, + "learning_rate": 2.5123893305113525e-05, + "loss": 0.0291, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01348068630322814, + "rkd_loss_weight": 0.6991733014583588, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3821627736091614, + "rkd_teacher_entropy": 3.7348875045776366, + "rkd_teacher_similarity_mean": 0.7783466041088104, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9030 + }, + { + "epoch": 2.630200756473669, + "flow_matching_loss": 0.017457567621022463, + "grad_norm": 0.17070814967155457, + "learning_rate": 2.5111686447448997e-05, + "loss": 0.0259, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012036937149241566, + "rkd_loss_weight": 0.6988399624824524, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38088160157203677, + "rkd_teacher_entropy": 3.724839973449707, + "rkd_teacher_similarity_mean": 0.7748346328735352, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9040 + }, + { + "epoch": 2.633110270584812, + "flow_matching_loss": 0.017763434536755086, + "grad_norm": 0.12470129877328873, + "learning_rate": 2.5099467303131966e-05, + "loss": 0.0264, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012359178625047207, + "rkd_loss_weight": 0.698506623506546, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39283052980899813, + "rkd_teacher_entropy": 3.770961380004883, + "rkd_teacher_similarity_mean": 0.7881782114505768, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9050 + }, + { + "epoch": 2.636019784695956, + "flow_matching_loss": 0.019270456116646528, + "grad_norm": 0.1634802371263504, + "learning_rate": 2.508723588700984e-05, + "loss": 0.028, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012530304864048958, + "rkd_loss_weight": 0.6981732666492462, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3815161049365997, + "rkd_teacher_entropy": 3.721452331542969, + "rkd_teacher_similarity_mean": 0.7762481093406677, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9060 + }, + { + "epoch": 2.638929298807099, + "flow_matching_loss": 0.019247034471482037, + "grad_norm": 0.15694789588451385, + "learning_rate": 2.5074992213944952e-05, + "loss": 0.0285, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013220001198351383, + "rkd_loss_weight": 0.6978399336338044, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3874123334884644, + "rkd_teacher_entropy": 3.7493112802505495, + "rkd_teacher_similarity_mean": 0.7828454792499542, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9070 + }, + { + "epoch": 2.641838812918243, + "flow_matching_loss": 0.018079957645386457, + "grad_norm": 0.13455571234226227, + "learning_rate": 2.506273629881451e-05, + "loss": 0.028, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014186157565563916, + "rkd_loss_weight": 0.697506582736969, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38575644195079806, + "rkd_teacher_entropy": 3.7413536310195923, + "rkd_teacher_similarity_mean": 0.7823214173316956, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9080 + }, + { + "epoch": 2.644748327029386, + "flow_matching_loss": 0.017362204566597938, + "grad_norm": 0.1687612533569336, + "learning_rate": 2.5050468156510617e-05, + "loss": 0.0259, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012299739010632037, + "rkd_loss_weight": 0.6971732378005981, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3861160397529602, + "rkd_teacher_entropy": 3.7382900953292846, + "rkd_teacher_similarity_mean": 0.7800210833549499, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9090 + }, + { + "epoch": 2.6476578411405294, + "flow_matching_loss": 0.01789680039510131, + "grad_norm": 0.1190759614109993, + "learning_rate": 2.5038187801940215e-05, + "loss": 0.0256, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011105595808476209, + "rkd_loss_weight": 0.6968398988246918, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38010722398757935, + "rkd_teacher_entropy": 3.7265729188919066, + "rkd_teacher_similarity_mean": 0.774831372499466, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9100 + }, + { + "epoch": 2.650567355251673, + "flow_matching_loss": 0.016803851816803217, + "grad_norm": 0.18703137338161469, + "learning_rate": 2.5025895250025088e-05, + "loss": 0.0258, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012914452888071536, + "rkd_loss_weight": 0.6965065479278565, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38728774189949033, + "rkd_teacher_entropy": 3.7462764978408813, + "rkd_teacher_similarity_mean": 0.783622008562088, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9110 + }, + { + "epoch": 2.6534768693628163, + "flow_matching_loss": 0.018263907451182605, + "grad_norm": 0.17521046102046967, + "learning_rate": 2.501359051570186e-05, + "loss": 0.0261, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011321680760011077, + "rkd_loss_weight": 0.6961732089519501, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38747637569904325, + "rkd_teacher_entropy": 3.748961353302002, + "rkd_teacher_similarity_mean": 0.7829337239265441, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9120 + }, + { + "epoch": 2.6563863834739596, + "flow_matching_loss": 0.016717770230025054, + "grad_norm": 0.16776986420154572, + "learning_rate": 2.500127361392193e-05, + "loss": 0.0247, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011511513497680426, + "rkd_loss_weight": 0.6958398640155792, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38344558477401736, + "rkd_teacher_entropy": 3.7297710180282593, + "rkd_teacher_similarity_mean": 0.7785124242305755, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9130 + }, + { + "epoch": 2.6592958975851033, + "flow_matching_loss": 0.016759178880602122, + "grad_norm": 0.16128383576869965, + "learning_rate": 2.49889445596515e-05, + "loss": 0.0261, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013445383217185736, + "rkd_loss_weight": 0.6955065131187439, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3766615867614746, + "rkd_teacher_entropy": 3.7124990224838257, + "rkd_teacher_similarity_mean": 0.7730571866035462, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9140 + }, + { + "epoch": 2.6622054116962466, + "flow_matching_loss": 0.021123645547777414, + "grad_norm": 0.173654705286026, + "learning_rate": 2.497660336787154e-05, + "loss": 0.0296, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012143122870475054, + "rkd_loss_weight": 0.6951731741428375, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3863146364688873, + "rkd_teacher_entropy": 3.7492010831832885, + "rkd_teacher_similarity_mean": 0.7800174593925476, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9150 + }, + { + "epoch": 2.6651149258073903, + "flow_matching_loss": 0.016058863047510384, + "grad_norm": 0.15587033331394196, + "learning_rate": 2.496425005357775e-05, + "loss": 0.0249, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012760044354945422, + "rkd_loss_weight": 0.6948398292064667, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3858017146587372, + "rkd_teacher_entropy": 3.749003839492798, + "rkd_teacher_similarity_mean": 0.7794944465160369, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9160 + }, + { + "epoch": 2.6680244399185336, + "flow_matching_loss": 0.01941660512238741, + "grad_norm": 0.16323056817054749, + "learning_rate": 2.4951884631780578e-05, + "loss": 0.028, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012287008576095104, + "rkd_loss_weight": 0.6945064842700959, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39467082619667054, + "rkd_teacher_entropy": 3.768969988822937, + "rkd_teacher_similarity_mean": 0.789600545167923, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9170 + }, + { + "epoch": 2.6709339540296773, + "flow_matching_loss": 0.01688202004879713, + "grad_norm": 0.17395518720149994, + "learning_rate": 2.493950711750518e-05, + "loss": 0.0251, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011900447402149439, + "rkd_loss_weight": 0.6941731393337249, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3808446675539017, + "rkd_teacher_entropy": 3.7309097051620483, + "rkd_teacher_similarity_mean": 0.7774238169193268, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9180 + }, + { + "epoch": 2.6738434681408205, + "flow_matching_loss": 0.02016874309629202, + "grad_norm": 0.18813353776931763, + "learning_rate": 2.49271175257914e-05, + "loss": 0.0291, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012810384761542082, + "rkd_loss_weight": 0.6938397943973541, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38863938450813296, + "rkd_teacher_entropy": 3.7422186613082884, + "rkd_teacher_similarity_mean": 0.7830451130867004, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9190 + }, + { + "epoch": 2.676752982251964, + "flow_matching_loss": 0.018648456409573556, + "grad_norm": 0.30030325055122375, + "learning_rate": 2.4914715871693757e-05, + "loss": 0.028, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013446255866438151, + "rkd_loss_weight": 0.6935064494609833, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3908844619989395, + "rkd_teacher_entropy": 3.7462102651596068, + "rkd_teacher_similarity_mean": 0.782121205329895, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9200 + }, + { + "epoch": 2.6796624963631075, + "flow_matching_loss": 0.018414110597223043, + "grad_norm": 0.2808041572570801, + "learning_rate": 2.4902302170281438e-05, + "loss": 0.0269, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012219140632078051, + "rkd_loss_weight": 0.6931731104850769, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38160785734653474, + "rkd_teacher_entropy": 3.7360048294067383, + "rkd_teacher_similarity_mean": 0.7758565962314605, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9210 + }, + { + "epoch": 2.6825720104742508, + "flow_matching_loss": 0.018729236070066692, + "grad_norm": 0.18278363347053528, + "learning_rate": 2.4889876436638252e-05, + "loss": 0.028, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013439127430319785, + "rkd_loss_weight": 0.6928397595882416, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37810832262039185, + "rkd_teacher_entropy": 3.726880359649658, + "rkd_teacher_similarity_mean": 0.7749143004417419, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9220 + }, + { + "epoch": 2.685481524585394, + "flow_matching_loss": 0.018875785917043687, + "grad_norm": 0.16332276165485382, + "learning_rate": 2.4877438685862643e-05, + "loss": 0.0269, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011585579719394445, + "rkd_loss_weight": 0.6925064146518707, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3867544710636139, + "rkd_teacher_entropy": 3.759402060508728, + "rkd_teacher_similarity_mean": 0.7840169429779053, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9230 + }, + { + "epoch": 2.6883910386965377, + "flow_matching_loss": 0.01674517337232828, + "grad_norm": 0.23612266778945923, + "learning_rate": 2.486498893306765e-05, + "loss": 0.0263, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013851361721754074, + "rkd_loss_weight": 0.6921730756759643, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3968571752309799, + "rkd_teacher_entropy": 3.770274591445923, + "rkd_teacher_similarity_mean": 0.7916067540645599, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9240 + }, + { + "epoch": 2.691300552807681, + "flow_matching_loss": 0.018311017658561468, + "grad_norm": 0.19276967644691467, + "learning_rate": 2.485252719338089e-05, + "loss": 0.0262, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011466168612241746, + "rkd_loss_weight": 0.691839724779129, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37704563438892363, + "rkd_teacher_entropy": 3.703796458244324, + "rkd_teacher_similarity_mean": 0.7730741620063781, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9250 + }, + { + "epoch": 2.6942100669188247, + "flow_matching_loss": 0.01892481492832303, + "grad_norm": 0.12180554121732712, + "learning_rate": 2.4840053481944563e-05, + "loss": 0.0263, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01069770846515894, + "rkd_loss_weight": 0.6915063858032227, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3782868266105652, + "rkd_teacher_entropy": 3.731794738769531, + "rkd_teacher_similarity_mean": 0.7766677439212799, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9260 + }, + { + "epoch": 2.697119581029968, + "flow_matching_loss": 0.020662061776965855, + "grad_norm": 0.18139499425888062, + "learning_rate": 2.4827567813915398e-05, + "loss": 0.0289, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011935388343408704, + "rkd_loss_weight": 0.6911730408668518, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39166775345802307, + "rkd_teacher_entropy": 3.770013952255249, + "rkd_teacher_similarity_mean": 0.7893359422683716, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9270 + }, + { + "epoch": 2.7000290951411117, + "flow_matching_loss": 0.018899262137711047, + "grad_norm": 0.16972225904464722, + "learning_rate": 2.481507020446466e-05, + "loss": 0.0267, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011324713006615639, + "rkd_loss_weight": 0.6908396899700164, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3930594652891159, + "rkd_teacher_entropy": 3.7694265842437744, + "rkd_teacher_similarity_mean": 0.7898699045181274, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9280 + }, + { + "epoch": 2.702938609252255, + "flow_matching_loss": 0.01649754885584116, + "grad_norm": 0.1546514928340912, + "learning_rate": 2.4802560668778122e-05, + "loss": 0.0236, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010259246174246072, + "rkd_loss_weight": 0.6905063569545746, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3865939140319824, + "rkd_teacher_entropy": 3.7414467334747314, + "rkd_teacher_similarity_mean": 0.7843884050846099, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9290 + }, + { + "epoch": 2.705848123363398, + "flow_matching_loss": 0.01859353929758072, + "grad_norm": 0.15019026398658752, + "learning_rate": 2.479003922205605e-05, + "loss": 0.0254, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009883117722347378, + "rkd_loss_weight": 0.6901730000972748, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3810259163379669, + "rkd_teacher_entropy": 3.7295937299728394, + "rkd_teacher_similarity_mean": 0.7772791624069214, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9300 + }, + { + "epoch": 2.708757637474542, + "flow_matching_loss": 0.01760333627462387, + "grad_norm": 0.1553613841533661, + "learning_rate": 2.4777505879513183e-05, + "loss": 0.025, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010791652323678135, + "rkd_loss_weight": 0.6898396611213684, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37466945350170133, + "rkd_teacher_entropy": 3.712295651435852, + "rkd_teacher_similarity_mean": 0.770207554101944, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9310 + }, + { + "epoch": 2.711667151585685, + "flow_matching_loss": 0.015931012108922005, + "grad_norm": 0.15710484981536865, + "learning_rate": 2.4764960656378714e-05, + "loss": 0.0244, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012291456758975982, + "rkd_loss_weight": 0.689506322145462, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39834682941436766, + "rkd_teacher_entropy": 3.7678990602493285, + "rkd_teacher_similarity_mean": 0.7931482255458832, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9320 + }, + { + "epoch": 2.7145766656968284, + "flow_matching_loss": 0.017354662343859673, + "grad_norm": 0.27742999792099, + "learning_rate": 2.4752403567896274e-05, + "loss": 0.0274, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014514475502073765, + "rkd_loss_weight": 0.6891729652881622, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3830423802137375, + "rkd_teacher_entropy": 3.7306634187698364, + "rkd_teacher_similarity_mean": 0.7793921053409576, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9330 + }, + { + "epoch": 2.717486179807972, + "flow_matching_loss": 0.020550470799207687, + "grad_norm": 0.17207121849060059, + "learning_rate": 2.47398346293239e-05, + "loss": 0.0285, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011563338804990053, + "rkd_loss_weight": 0.6888396322727204, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3865086793899536, + "rkd_teacher_entropy": 3.7458082675933837, + "rkd_teacher_similarity_mean": 0.7831831037998199, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9340 + }, + { + "epoch": 2.7203956939191154, + "flow_matching_loss": 0.018701729364693165, + "grad_norm": 0.13885842263698578, + "learning_rate": 2.4727253855934055e-05, + "loss": 0.0271, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012157618626952171, + "rkd_loss_weight": 0.688506281375885, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3852247208356857, + "rkd_teacher_entropy": 3.7366172790527346, + "rkd_teacher_similarity_mean": 0.778994232416153, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9350 + }, + { + "epoch": 2.723305208030259, + "flow_matching_loss": 0.017185181099921464, + "grad_norm": 0.28971680998802185, + "learning_rate": 2.4714661263013548e-05, + "loss": 0.0253, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011726946011185646, + "rkd_loss_weight": 0.6881729364395142, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3940970182418823, + "rkd_teacher_entropy": 3.7665310859680177, + "rkd_teacher_similarity_mean": 0.7905450582504272, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9360 + }, + { + "epoch": 2.7262147221414024, + "flow_matching_loss": 0.01727288356050849, + "grad_norm": 0.20071834325790405, + "learning_rate": 2.4702056865863575e-05, + "loss": 0.027, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01410171166062355, + "rkd_loss_weight": 0.6878395974636078, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3848879188299179, + "rkd_teacher_entropy": 3.744809937477112, + "rkd_teacher_similarity_mean": 0.7824682593345642, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9370 + }, + { + "epoch": 2.729124236252546, + "flow_matching_loss": 0.014880501478910447, + "grad_norm": 0.16237196326255798, + "learning_rate": 2.4689440679799672e-05, + "loss": 0.0229, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011733915563672781, + "rkd_loss_weight": 0.6875062465667725, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3904766499996185, + "rkd_teacher_entropy": 3.7527274608612062, + "rkd_teacher_similarity_mean": 0.7861459791660309, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9380 + }, + { + "epoch": 2.7320337503636893, + "flow_matching_loss": 0.01790269622579217, + "grad_norm": 0.19553376734256744, + "learning_rate": 2.4676812720151678e-05, + "loss": 0.0261, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011950154695659876, + "rkd_loss_weight": 0.6871729075908661, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38457719683647157, + "rkd_teacher_entropy": 3.736064600944519, + "rkd_teacher_similarity_mean": 0.7796039044857025, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9390 + }, + { + "epoch": 2.7349432644748326, + "flow_matching_loss": 0.01869045216590166, + "grad_norm": 0.19094575941562653, + "learning_rate": 2.466417300226377e-05, + "loss": 0.0263, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011073655355721712, + "rkd_loss_weight": 0.6868395626544952, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3807124733924866, + "rkd_teacher_entropy": 3.73516161441803, + "rkd_teacher_similarity_mean": 0.7759599387645721, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9400 + }, + { + "epoch": 2.7378527785859763, + "flow_matching_loss": 0.018983042053878307, + "grad_norm": 0.23472312092781067, + "learning_rate": 2.4651521541494394e-05, + "loss": 0.0279, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01297315713018179, + "rkd_loss_weight": 0.6865062117576599, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.386382195353508, + "rkd_teacher_entropy": 3.7360289096832275, + "rkd_teacher_similarity_mean": 0.7797063589096069, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9410 + }, + { + "epoch": 2.7407622926971196, + "flow_matching_loss": 0.015795184578746558, + "grad_norm": 0.1530633270740509, + "learning_rate": 2.4638858353216267e-05, + "loss": 0.0242, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012311148922890425, + "rkd_loss_weight": 0.6861728727817535, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3969403594732285, + "rkd_teacher_entropy": 3.764282751083374, + "rkd_teacher_similarity_mean": 0.7920947134494781, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9420 + }, + { + "epoch": 2.743671806808263, + "flow_matching_loss": 0.01718572210520506, + "grad_norm": 0.15059031546115875, + "learning_rate": 2.4626183452816355e-05, + "loss": 0.0248, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011129537597298622, + "rkd_loss_weight": 0.6858395278453827, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.385479599237442, + "rkd_teacher_entropy": 3.7377293586730955, + "rkd_teacher_similarity_mean": 0.7817018508911133, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9430 + }, + { + "epoch": 2.7465813209194065, + "flow_matching_loss": 0.017811206076294182, + "grad_norm": 0.16295993328094482, + "learning_rate": 2.4613496855695855e-05, + "loss": 0.0259, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011746774660423398, + "rkd_loss_weight": 0.6855061829090119, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3873810112476349, + "rkd_teacher_entropy": 3.7479687690734864, + "rkd_teacher_similarity_mean": 0.7820706844329834, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9440 + }, + { + "epoch": 2.74949083503055, + "flow_matching_loss": 0.01805741563439369, + "grad_norm": 0.19110846519470215, + "learning_rate": 2.4600798577270183e-05, + "loss": 0.0269, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012903957813978194, + "rkd_loss_weight": 0.685172837972641, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38006560802459716, + "rkd_teacher_entropy": 3.7247447967529297, + "rkd_teacher_similarity_mean": 0.7742421209812165, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9450 + }, + { + "epoch": 2.752400349141693, + "flow_matching_loss": 0.019859129562973976, + "grad_norm": 0.25402942299842834, + "learning_rate": 2.4588088632968944e-05, + "loss": 0.0294, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013865676103159785, + "rkd_loss_weight": 0.6848394930362701, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3842185169458389, + "rkd_teacher_entropy": 3.741113615036011, + "rkd_teacher_similarity_mean": 0.7797388017177582, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9460 + }, + { + "epoch": 2.7553098632528368, + "flow_matching_loss": 0.018477936647832392, + "grad_norm": 0.13020098209381104, + "learning_rate": 2.4575367038235916e-05, + "loss": 0.0262, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011289618443697691, + "rkd_loss_weight": 0.6845061480998993, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38327291011810305, + "rkd_teacher_entropy": 3.7436386823654173, + "rkd_teacher_similarity_mean": 0.779334670305252, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9470 + }, + { + "epoch": 2.7582193773639805, + "flow_matching_loss": 0.017993636522442104, + "grad_norm": 0.20134337246418, + "learning_rate": 2.4562633808529043e-05, + "loss": 0.027, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013182141445577144, + "rkd_loss_weight": 0.6841728091239929, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3945482850074768, + "rkd_teacher_entropy": 3.761246132850647, + "rkd_teacher_similarity_mean": 0.7869853794574737, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9480 + }, + { + "epoch": 2.7611288914751237, + "flow_matching_loss": 0.01702094851061702, + "grad_norm": 0.16719238460063934, + "learning_rate": 2.4549888959320396e-05, + "loss": 0.0261, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013260004855692387, + "rkd_loss_weight": 0.6838394582271576, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3837919056415558, + "rkd_teacher_entropy": 3.739100980758667, + "rkd_teacher_similarity_mean": 0.7786635279655456, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9490 + }, + { + "epoch": 2.764038405586267, + "flow_matching_loss": 0.01516501121222973, + "grad_norm": 0.16447964310646057, + "learning_rate": 2.4537132506096162e-05, + "loss": 0.0229, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011343973688781261, + "rkd_loss_weight": 0.6835061132907867, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3836254894733429, + "rkd_teacher_entropy": 3.733639144897461, + "rkd_teacher_similarity_mean": 0.7799869358539582, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9500 + }, + { + "epoch": 2.7669479196974107, + "flow_matching_loss": 0.019037283305078746, + "grad_norm": 0.16127541661262512, + "learning_rate": 2.452436446435664e-05, + "loss": 0.0257, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00979663641192019, + "rkd_loss_weight": 0.6831727802753449, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.385939285159111, + "rkd_teacher_entropy": 3.727817511558533, + "rkd_teacher_similarity_mean": 0.780520248413086, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9510 + }, + { + "epoch": 2.769857433808554, + "flow_matching_loss": 0.019532474875450134, + "grad_norm": 0.19791245460510254, + "learning_rate": 2.451158484961621e-05, + "loss": 0.0291, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014000425254926086, + "rkd_loss_weight": 0.682839423418045, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3847504884004593, + "rkd_teacher_entropy": 3.743904161453247, + "rkd_teacher_similarity_mean": 0.7802191913127899, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9520 + }, + { + "epoch": 2.7727669479196972, + "flow_matching_loss": 0.015380450617522001, + "grad_norm": 0.16288000345230103, + "learning_rate": 2.4498793677403304e-05, + "loss": 0.0243, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013123931270092726, + "rkd_loss_weight": 0.6825060844421387, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38768257200717926, + "rkd_teacher_entropy": 3.7459491968154905, + "rkd_teacher_similarity_mean": 0.7838925123214722, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9530 + }, + { + "epoch": 2.775676462030841, + "flow_matching_loss": 0.017927868850529195, + "grad_norm": 0.1467403918504715, + "learning_rate": 2.4485990963260398e-05, + "loss": 0.0249, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010225470922887326, + "rkd_loss_weight": 0.6821727454662323, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3870130628347397, + "rkd_teacher_entropy": 3.751745581626892, + "rkd_teacher_similarity_mean": 0.7842716157436371, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9540 + }, + { + "epoch": 2.778585976141984, + "flow_matching_loss": 0.017437166441231965, + "grad_norm": 0.14987008273601532, + "learning_rate": 2.4473176722744e-05, + "loss": 0.0254, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011646169796586036, + "rkd_loss_weight": 0.6818393886089325, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3887144267559052, + "rkd_teacher_entropy": 3.739928460121155, + "rkd_teacher_similarity_mean": 0.7831940472126007, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9550 + }, + { + "epoch": 2.7814954902531275, + "flow_matching_loss": 0.015426218137145042, + "grad_norm": 0.1466987431049347, + "learning_rate": 2.446035097142463e-05, + "loss": 0.023, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011062025371938945, + "rkd_loss_weight": 0.6815060555934906, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3949295669794083, + "rkd_teacher_entropy": 3.76624915599823, + "rkd_teacher_similarity_mean": 0.7909410119056701, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9560 + }, + { + "epoch": 2.784405004364271, + "flow_matching_loss": 0.017984882835298775, + "grad_norm": 0.23628519475460052, + "learning_rate": 2.444751372488677e-05, + "loss": 0.0243, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0093096312135458, + "rkd_loss_weight": 0.6811727046966553, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3881399750709534, + "rkd_teacher_entropy": 3.765193819999695, + "rkd_teacher_similarity_mean": 0.78607177734375, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9570 + }, + { + "epoch": 2.7873145184754144, + "flow_matching_loss": 0.019286026153713463, + "grad_norm": 0.1989506036043167, + "learning_rate": 2.443466499872889e-05, + "loss": 0.0273, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011811196710914374, + "rkd_loss_weight": 0.6808393597602844, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38409750163555145, + "rkd_teacher_entropy": 3.7372156143188477, + "rkd_teacher_similarity_mean": 0.7803955018520355, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9580 + }, + { + "epoch": 2.790224032586558, + "flow_matching_loss": 0.017539361119270326, + "grad_norm": 0.19936051964759827, + "learning_rate": 2.4421804808563413e-05, + "loss": 0.0258, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012081494927406311, + "rkd_loss_weight": 0.680506020784378, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3834814757108688, + "rkd_teacher_entropy": 3.737623596191406, + "rkd_teacher_similarity_mean": 0.7782076060771942, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9590 + }, + { + "epoch": 2.7931335466977014, + "flow_matching_loss": 0.01806072425097227, + "grad_norm": 0.2217656970024109, + "learning_rate": 2.440893317001667e-05, + "loss": 0.0274, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013747924659401178, + "rkd_loss_weight": 0.6801726698875428, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3885666787624359, + "rkd_teacher_entropy": 3.74503755569458, + "rkd_teacher_similarity_mean": 0.7838293433189392, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9600 + }, + { + "epoch": 2.796043060808845, + "flow_matching_loss": 0.016399416979402302, + "grad_norm": 0.14946849644184113, + "learning_rate": 2.4396050098728925e-05, + "loss": 0.0236, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010572574101388454, + "rkd_loss_weight": 0.6798393309116364, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38022843599319456, + "rkd_teacher_entropy": 3.7205769300460814, + "rkd_teacher_similarity_mean": 0.7762082695960999, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9610 + }, + { + "epoch": 2.7989525749199884, + "flow_matching_loss": 0.017927495203912257, + "grad_norm": 0.1591920256614685, + "learning_rate": 2.4383155610354314e-05, + "loss": 0.0275, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014023333229124546, + "rkd_loss_weight": 0.6795059859752655, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38592917621135714, + "rkd_teacher_entropy": 3.7468832969665526, + "rkd_teacher_similarity_mean": 0.7821976542472839, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9620 + }, + { + "epoch": 2.8018620890311317, + "flow_matching_loss": 0.016865809727460147, + "grad_norm": 0.1597060263156891, + "learning_rate": 2.437024972056087e-05, + "loss": 0.024, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010571593930944801, + "rkd_loss_weight": 0.6791726350784302, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39411160051822663, + "rkd_teacher_entropy": 3.751075839996338, + "rkd_teacher_similarity_mean": 0.7857027590274811, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9630 + }, + { + "epoch": 2.8047716031422754, + "flow_matching_loss": 0.018372327089309692, + "grad_norm": 0.1393096148967743, + "learning_rate": 2.4357332445030452e-05, + "loss": 0.0263, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01174205276183784, + "rkd_loss_weight": 0.6788392961025238, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.40152994394302366, + "rkd_teacher_entropy": 3.7848456144332885, + "rkd_teacher_similarity_mean": 0.7945121169090271, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9640 + }, + { + "epoch": 2.8076811172534186, + "flow_matching_loss": 0.017336640693247318, + "grad_norm": 0.15658098459243774, + "learning_rate": 2.434440379945878e-05, + "loss": 0.0255, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011979099130257964, + "rkd_loss_weight": 0.678505951166153, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3847231924533844, + "rkd_teacher_entropy": 3.735755467414856, + "rkd_teacher_similarity_mean": 0.7779786825180054, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9650 + }, + { + "epoch": 2.810590631364562, + "flow_matching_loss": 0.0152107248082757, + "grad_norm": 0.1358291357755661, + "learning_rate": 2.4331463799555366e-05, + "loss": 0.0227, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01106001827865839, + "rkd_loss_weight": 0.6781726062297821, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3901958465576172, + "rkd_teacher_entropy": 3.7476892948150633, + "rkd_teacher_similarity_mean": 0.7843361377716065, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9660 + }, + { + "epoch": 2.8135001454757056, + "flow_matching_loss": 0.016806190926581623, + "grad_norm": 0.12448987364768982, + "learning_rate": 2.4318512461043544e-05, + "loss": 0.0238, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010383629193529486, + "rkd_loss_weight": 0.6778392612934112, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3900000065565109, + "rkd_teacher_entropy": 3.754815363883972, + "rkd_teacher_similarity_mean": 0.7855409383773804, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9670 + }, + { + "epoch": 2.816409659586849, + "flow_matching_loss": 0.015311282221227884, + "grad_norm": 0.15503032505512238, + "learning_rate": 2.4305549799660397e-05, + "loss": 0.0227, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010893376637250185, + "rkd_loss_weight": 0.6775059163570404, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3876234799623489, + "rkd_teacher_entropy": 3.7417144536972047, + "rkd_teacher_similarity_mean": 0.7827082097530365, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9680 + }, + { + "epoch": 2.8193191736979926, + "flow_matching_loss": 0.016169856954365968, + "grad_norm": 0.2874017655849457, + "learning_rate": 2.429257583115679e-05, + "loss": 0.0242, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011898940335959196, + "rkd_loss_weight": 0.6771725714206696, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39463716745376587, + "rkd_teacher_entropy": 3.755487632751465, + "rkd_teacher_similarity_mean": 0.7873702824115754, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9690 + }, + { + "epoch": 2.822228687809136, + "flow_matching_loss": 0.01780065270140767, + "grad_norm": 0.13297872245311737, + "learning_rate": 2.4279590571297316e-05, + "loss": 0.0259, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011995249195024371, + "rkd_loss_weight": 0.6768392324447632, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3829847455024719, + "rkd_teacher_entropy": 3.731919860839844, + "rkd_teacher_similarity_mean": 0.7773659348487854, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9700 + }, + { + "epoch": 2.8251382019202795, + "flow_matching_loss": 0.017412344366312026, + "grad_norm": 0.17661041021347046, + "learning_rate": 2.4266594035860288e-05, + "loss": 0.0263, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013175168959423899, + "rkd_loss_weight": 0.6765058815479279, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3904698550701141, + "rkd_teacher_entropy": 3.7533575534820556, + "rkd_teacher_similarity_mean": 0.783875685930252, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9710 + }, + { + "epoch": 2.828047716031423, + "flow_matching_loss": 0.0174109804444015, + "grad_norm": 0.14876431226730347, + "learning_rate": 2.425358624063773e-05, + "loss": 0.0253, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01162271834909916, + "rkd_loss_weight": 0.676172536611557, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38390971422195436, + "rkd_teacher_entropy": 3.734575867652893, + "rkd_teacher_similarity_mean": 0.7783501267433166, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9720 + }, + { + "epoch": 2.830957230142566, + "flow_matching_loss": 0.015023607760667801, + "grad_norm": 0.1715191751718521, + "learning_rate": 2.424056720143532e-05, + "loss": 0.0233, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012268138956278563, + "rkd_loss_weight": 0.6758391976356506, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39158505499362944, + "rkd_teacher_entropy": 3.749839425086975, + "rkd_teacher_similarity_mean": 0.7871481955051423, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9730 + }, + { + "epoch": 2.8338667442537098, + "flow_matching_loss": 0.020494789350777864, + "grad_norm": 0.32615000009536743, + "learning_rate": 2.422753693407244e-05, + "loss": 0.0293, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013000571448355914, + "rkd_loss_weight": 0.6755058467388153, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39132251143455504, + "rkd_teacher_entropy": 3.754395079612732, + "rkd_teacher_similarity_mean": 0.7867206335067749, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9740 + }, + { + "epoch": 2.836776258364853, + "flow_matching_loss": 0.019240039587020873, + "grad_norm": 0.18154612183570862, + "learning_rate": 2.421449545438209e-05, + "loss": 0.0274, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01214338354766369, + "rkd_loss_weight": 0.6751725077629089, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3933902204036713, + "rkd_teacher_entropy": 3.7541319608688353, + "rkd_teacher_similarity_mean": 0.78655925989151, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9750 + }, + { + "epoch": 2.8396857724759963, + "flow_matching_loss": 0.017460347898304463, + "grad_norm": 0.12424468994140625, + "learning_rate": 2.4201442778210887e-05, + "loss": 0.0276, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015010873414576053, + "rkd_loss_weight": 0.6748391628265381, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3822584360837936, + "rkd_teacher_entropy": 3.7151095151901243, + "rkd_teacher_similarity_mean": 0.7730722665786743, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9760 + }, + { + "epoch": 2.84259528658714, + "flow_matching_loss": 0.01791738597676158, + "grad_norm": 0.13130776584148407, + "learning_rate": 2.418837892141907e-05, + "loss": 0.0259, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01184681747108698, + "rkd_loss_weight": 0.6745058119297027, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39433135390281676, + "rkd_teacher_entropy": 3.7585681438446046, + "rkd_teacher_similarity_mean": 0.7862258017063141, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9770 + }, + { + "epoch": 2.8455048006982833, + "flow_matching_loss": 0.01965144807472825, + "grad_norm": 0.18579213321208954, + "learning_rate": 2.4175303899880458e-05, + "loss": 0.0276, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011802055686712266, + "rkd_loss_weight": 0.6741724789142609, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38434564471244814, + "rkd_teacher_entropy": 3.727480888366699, + "rkd_teacher_similarity_mean": 0.7762387096881866, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9780 + }, + { + "epoch": 2.848414314809427, + "flow_matching_loss": 0.019096874725073577, + "grad_norm": 0.18672321736812592, + "learning_rate": 2.4162217729482427e-05, + "loss": 0.0278, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012969043664634227, + "rkd_loss_weight": 0.6738391220569611, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3878586173057556, + "rkd_teacher_entropy": 3.7426086902618407, + "rkd_teacher_similarity_mean": 0.7816458284854889, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9790 + }, + { + "epoch": 2.8513238289205702, + "flow_matching_loss": 0.01540848622098565, + "grad_norm": 0.13104751706123352, + "learning_rate": 2.4149120426125918e-05, + "loss": 0.0231, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011428983369842171, + "rkd_loss_weight": 0.6735057830810547, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3903628498315811, + "rkd_teacher_entropy": 3.7407423734664915, + "rkd_teacher_similarity_mean": 0.7822753608226776, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9800 + }, + { + "epoch": 2.854233343031714, + "flow_matching_loss": 0.015469806548207999, + "grad_norm": 0.1873980313539505, + "learning_rate": 2.413601200572538e-05, + "loss": 0.0234, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011846188921481371, + "rkd_loss_weight": 0.6731724441051483, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3903420686721802, + "rkd_teacher_entropy": 3.758308672904968, + "rkd_teacher_similarity_mean": 0.7827643930912018, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9810 + }, + { + "epoch": 2.857142857142857, + "flow_matching_loss": 0.01613344792276621, + "grad_norm": 0.23818723857402802, + "learning_rate": 2.4122892484208794e-05, + "loss": 0.0256, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014131461456418037, + "rkd_loss_weight": 0.6728390872478485, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38959103524684907, + "rkd_teacher_entropy": 3.753969740867615, + "rkd_teacher_similarity_mean": 0.7816823899745942, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9820 + }, + { + "epoch": 2.8600523712540005, + "flow_matching_loss": 0.017205263208597898, + "grad_norm": 0.19451117515563965, + "learning_rate": 2.4109761877517603e-05, + "loss": 0.0248, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011355661135166884, + "rkd_loss_weight": 0.6725057542324067, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3914614200592041, + "rkd_teacher_entropy": 3.752730870246887, + "rkd_teacher_similarity_mean": 0.7849747240543365, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9830 + }, + { + "epoch": 2.862961885365144, + "flow_matching_loss": 0.01812290111556649, + "grad_norm": 0.14297708868980408, + "learning_rate": 2.4096620201606737e-05, + "loss": 0.027, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013169701769948005, + "rkd_loss_weight": 0.6721724033355713, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3882122367620468, + "rkd_teacher_entropy": 3.739221429824829, + "rkd_teacher_similarity_mean": 0.7822733521461487, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9840 + }, + { + "epoch": 2.8658713994762874, + "flow_matching_loss": 0.019764697644859552, + "grad_norm": 0.1743268072605133, + "learning_rate": 2.4083467472444573e-05, + "loss": 0.0273, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01118990322574973, + "rkd_loss_weight": 0.6718390583992004, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38519267439842225, + "rkd_teacher_entropy": 3.7254276990890505, + "rkd_teacher_similarity_mean": 0.7778976857662201, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9850 + }, + { + "epoch": 2.8687809135874307, + "flow_matching_loss": 0.01710573462769389, + "grad_norm": 0.2508334517478943, + "learning_rate": 2.4070303706012912e-05, + "loss": 0.0253, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012213725037872792, + "rkd_loss_weight": 0.6715057194232941, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.384654900431633, + "rkd_teacher_entropy": 3.7260900735855103, + "rkd_teacher_similarity_mean": 0.7781834185123444, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9860 + }, + { + "epoch": 2.8716904276985744, + "flow_matching_loss": 0.018247796315699817, + "grad_norm": 0.15504172444343567, + "learning_rate": 2.4057128918306975e-05, + "loss": 0.0272, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013360936474055052, + "rkd_loss_weight": 0.6711723685264588, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3845960110425949, + "rkd_teacher_entropy": 3.7259506225585937, + "rkd_teacher_similarity_mean": 0.7780405342578888, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9870 + }, + { + "epoch": 2.8745999418097177, + "flow_matching_loss": 0.018036810494959355, + "grad_norm": 0.265826940536499, + "learning_rate": 2.4043943125335373e-05, + "loss": 0.0267, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012888970784842968, + "rkd_loss_weight": 0.6708390295505524, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38571132719516754, + "rkd_teacher_entropy": 3.743310284614563, + "rkd_teacher_similarity_mean": 0.7811883509159088, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9880 + }, + { + "epoch": 2.8775094559208614, + "flow_matching_loss": 0.01863147532567382, + "grad_norm": 0.25061964988708496, + "learning_rate": 2.403074634312008e-05, + "loss": 0.0282, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01432628957554698, + "rkd_loss_weight": 0.6705056846141815, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39031540155410765, + "rkd_teacher_entropy": 3.7553322315216064, + "rkd_teacher_similarity_mean": 0.7847499966621398, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9890 + }, + { + "epoch": 2.8804189700320046, + "flow_matching_loss": 0.01739287143573165, + "grad_norm": 0.16509155929088593, + "learning_rate": 2.4017538587696442e-05, + "loss": 0.0257, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012376392912119626, + "rkd_loss_weight": 0.6701723337173462, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38519746661186216, + "rkd_teacher_entropy": 3.7299861192703245, + "rkd_teacher_similarity_mean": 0.7785587549209595, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9900 + }, + { + "epoch": 2.8833284841431484, + "flow_matching_loss": 0.017393136490136386, + "grad_norm": 0.16625139117240906, + "learning_rate": 2.4004319875113118e-05, + "loss": 0.0263, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013335887901484966, + "rkd_loss_weight": 0.6698389947414398, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38981216549873354, + "rkd_teacher_entropy": 3.737650752067566, + "rkd_teacher_similarity_mean": 0.7828519284725189, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9910 + }, + { + "epoch": 2.8862379982542916, + "flow_matching_loss": 0.021124158333987, + "grad_norm": 0.14285437762737274, + "learning_rate": 2.3991090221432096e-05, + "loss": 0.0294, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012368367239832877, + "rkd_loss_weight": 0.669505649805069, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38365220129489896, + "rkd_teacher_entropy": 3.7296054363250732, + "rkd_teacher_similarity_mean": 0.7767890870571137, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9920 + }, + { + "epoch": 2.889147512365435, + "flow_matching_loss": 0.022364661376923323, + "grad_norm": 0.19813787937164307, + "learning_rate": 2.397784964272865e-05, + "loss": 0.0317, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013905218336731195, + "rkd_loss_weight": 0.6691723048686982, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.382055389881134, + "rkd_teacher_entropy": 3.730531191825867, + "rkd_teacher_similarity_mean": 0.776715886592865, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9930 + }, + { + "epoch": 2.8920570264765786, + "flow_matching_loss": 0.01781678618863225, + "grad_norm": 0.19519859552383423, + "learning_rate": 2.3964598155091335e-05, + "loss": 0.0264, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012873793672770261, + "rkd_loss_weight": 0.6688389599323272, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38900490403175353, + "rkd_teacher_entropy": 3.7377236127853393, + "rkd_teacher_similarity_mean": 0.7805819272994995, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9940 + }, + { + "epoch": 2.894966540587722, + "flow_matching_loss": 0.018547466769814492, + "grad_norm": 0.182347372174263, + "learning_rate": 2.3951335774621952e-05, + "loss": 0.0258, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010890380386263132, + "rkd_loss_weight": 0.6685056209564209, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.387670961022377, + "rkd_teacher_entropy": 3.7474266052246095, + "rkd_teacher_similarity_mean": 0.7836907744407654, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9950 + }, + { + "epoch": 2.897876054698865, + "flow_matching_loss": 0.01672778818756342, + "grad_norm": 0.13661816716194153, + "learning_rate": 2.3938062517435548e-05, + "loss": 0.0245, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01161985294893384, + "rkd_loss_weight": 0.6681722700595856, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38753949403762816, + "rkd_teacher_entropy": 3.751998209953308, + "rkd_teacher_similarity_mean": 0.78150754570961, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9960 + }, + { + "epoch": 2.900785568810009, + "flow_matching_loss": 0.017258697468787432, + "grad_norm": 0.13937239348888397, + "learning_rate": 2.392477839966038e-05, + "loss": 0.0263, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013607059512287379, + "rkd_loss_weight": 0.6678389310836792, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3826844334602356, + "rkd_teacher_entropy": 3.7303213357925413, + "rkd_teacher_similarity_mean": 0.7768002867698669, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9970 + }, + { + "epoch": 2.903695082921152, + "flow_matching_loss": 0.018064788077026606, + "grad_norm": 0.1466348022222519, + "learning_rate": 2.3911483437437904e-05, + "loss": 0.0268, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013066128082573413, + "rkd_loss_weight": 0.6675055861473084, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3894209176301956, + "rkd_teacher_entropy": 3.746876049041748, + "rkd_teacher_similarity_mean": 0.7832213878631592, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9980 + }, + { + "epoch": 2.9066045970322953, + "flow_matching_loss": 0.015858082566410303, + "grad_norm": 0.12785492837429047, + "learning_rate": 2.3898177646922757e-05, + "loss": 0.0228, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010389547841623426, + "rkd_loss_weight": 0.667172235250473, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3796347796916962, + "rkd_teacher_entropy": 3.7172295570373537, + "rkd_teacher_similarity_mean": 0.7747010290622711, + "rkd_vlm_temp": 0.10000000149011612, + "step": 9990 + }, + { + "epoch": 2.909514111143439, + "flow_matching_loss": 0.017038852069526912, + "grad_norm": 0.14470350742340088, + "learning_rate": 2.3884861044282722e-05, + "loss": 0.0275, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015697213355451822, + "rkd_loss_weight": 0.6668389022350312, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3845343142747879, + "rkd_teacher_entropy": 3.737260866165161, + "rkd_teacher_similarity_mean": 0.7800199151039123, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10000 + }, + { + "epoch": 2.9124236252545828, + "flow_matching_loss": 0.016266233380883934, + "grad_norm": 0.15395009517669678, + "learning_rate": 2.3871533645698732e-05, + "loss": 0.0252, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013337124232202768, + "rkd_loss_weight": 0.6665055453777313, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3861635088920593, + "rkd_teacher_entropy": 3.740014123916626, + "rkd_teacher_similarity_mean": 0.7798472940921783, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10010 + }, + { + "epoch": 2.915333139365726, + "flow_matching_loss": 0.019017838034778835, + "grad_norm": 0.20333950221538544, + "learning_rate": 2.3858195467364833e-05, + "loss": 0.0267, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01156088849529624, + "rkd_loss_weight": 0.666172206401825, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3824995756149292, + "rkd_teacher_entropy": 3.716093373298645, + "rkd_teacher_similarity_mean": 0.7759766101837158, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10020 + }, + { + "epoch": 2.9182426534768693, + "flow_matching_loss": 0.01820995258167386, + "grad_norm": 0.202691912651062, + "learning_rate": 2.3844846525488166e-05, + "loss": 0.0275, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013999468553811312, + "rkd_loss_weight": 0.6658388674259186, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.393659844994545, + "rkd_teacher_entropy": 3.7577585220336913, + "rkd_teacher_similarity_mean": 0.7869181931018829, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10030 + }, + { + "epoch": 2.921152167588013, + "flow_matching_loss": 0.016290524508804083, + "grad_norm": 0.1465689092874527, + "learning_rate": 2.3831486836288957e-05, + "loss": 0.0238, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01124941473826766, + "rkd_loss_weight": 0.6655055105686187, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3954494267702103, + "rkd_teacher_entropy": 3.7573774337768553, + "rkd_teacher_similarity_mean": 0.7904290437698365, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10040 + }, + { + "epoch": 2.9240616816991563, + "flow_matching_loss": 0.016555683687329292, + "grad_norm": 0.1656964123249054, + "learning_rate": 2.381811641600048e-05, + "loss": 0.0248, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012417755369096995, + "rkd_loss_weight": 0.6651721775531769, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38211490213871, + "rkd_teacher_entropy": 3.717779016494751, + "rkd_teacher_similarity_mean": 0.7759050130844116, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10050 + }, + { + "epoch": 2.9269711958102995, + "flow_matching_loss": 0.01610120702534914, + "grad_norm": 0.13355547189712524, + "learning_rate": 2.3804735280869073e-05, + "loss": 0.0251, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01358925262466073, + "rkd_loss_weight": 0.6648388266563415, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3881184905767441, + "rkd_teacher_entropy": 3.749654698371887, + "rkd_teacher_similarity_mean": 0.7820021331310272, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10060 + }, + { + "epoch": 2.9298807099214432, + "flow_matching_loss": 0.01670389110222459, + "grad_norm": 0.12943299114704132, + "learning_rate": 2.3791343447154064e-05, + "loss": 0.0249, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012333360966295004, + "rkd_loss_weight": 0.6645054817199707, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3796939581632614, + "rkd_teacher_entropy": 3.7190472364425657, + "rkd_teacher_similarity_mean": 0.7732226073741912, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10070 + }, + { + "epoch": 2.9327902240325865, + "flow_matching_loss": 0.01865198416635394, + "grad_norm": 0.165034681558609, + "learning_rate": 2.37779409311278e-05, + "loss": 0.0254, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010149851348251105, + "rkd_loss_weight": 0.6641721427440643, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3892437607049942, + "rkd_teacher_entropy": 3.7538233757019044, + "rkd_teacher_similarity_mean": 0.7833300530910492, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10080 + }, + { + "epoch": 2.9356997381437298, + "flow_matching_loss": 0.017096691019833087, + "grad_norm": 0.13711780309677124, + "learning_rate": 2.3764527749075596e-05, + "loss": 0.0251, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011986644472926855, + "rkd_loss_weight": 0.663838791847229, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3889825165271759, + "rkd_teacher_entropy": 3.7478973150253294, + "rkd_teacher_similarity_mean": 0.7837580263614654, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10090 + }, + { + "epoch": 2.9386092522548735, + "flow_matching_loss": 0.018575351685285568, + "grad_norm": 0.13226082921028137, + "learning_rate": 2.3751103917295735e-05, + "loss": 0.0255, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010403595864772797, + "rkd_loss_weight": 0.6635054528713227, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39014569520950315, + "rkd_teacher_entropy": 3.7426661014556886, + "rkd_teacher_similarity_mean": 0.7837404072284698, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10100 + }, + { + "epoch": 2.9415187663660167, + "flow_matching_loss": 0.0190000812523067, + "grad_norm": 0.18153917789459229, + "learning_rate": 2.373766945209944e-05, + "loss": 0.0271, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012276169192045928, + "rkd_loss_weight": 0.6631721079349517, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3839076966047287, + "rkd_teacher_entropy": 3.728600549697876, + "rkd_teacher_similarity_mean": 0.7767038166522979, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10110 + }, + { + "epoch": 2.9444282804771604, + "flow_matching_loss": 0.017466360051184893, + "grad_norm": 0.21894671022891998, + "learning_rate": 2.3724224369810856e-05, + "loss": 0.0269, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014291625563055277, + "rkd_loss_weight": 0.6628387570381165, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38023545742034914, + "rkd_teacher_entropy": 3.7129583835601805, + "rkd_teacher_similarity_mean": 0.7726898550987243, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10120 + }, + { + "epoch": 2.9473377945883037, + "flow_matching_loss": 0.01679507680237293, + "grad_norm": 0.14439059793949127, + "learning_rate": 2.3710768686767015e-05, + "loss": 0.0247, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011872299341484904, + "rkd_loss_weight": 0.6625054180622101, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3877204179763794, + "rkd_teacher_entropy": 3.73597457408905, + "rkd_teacher_similarity_mean": 0.7828685343265533, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10130 + }, + { + "epoch": 2.9502473086994474, + "flow_matching_loss": 0.017172434087842702, + "grad_norm": 0.17005880177021027, + "learning_rate": 2.3697302419317856e-05, + "loss": 0.0248, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011515324097126722, + "rkd_loss_weight": 0.6621720731258393, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3922212690114975, + "rkd_teacher_entropy": 3.758710527420044, + "rkd_teacher_similarity_mean": 0.7860635340213775, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10140 + }, + { + "epoch": 2.9531568228105907, + "flow_matching_loss": 0.014701136015355587, + "grad_norm": 0.1593327671289444, + "learning_rate": 2.3683825583826146e-05, + "loss": 0.0226, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011874253675341607, + "rkd_loss_weight": 0.6618387281894684, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3829354792833328, + "rkd_teacher_entropy": 3.7271104574203493, + "rkd_teacher_similarity_mean": 0.7772333085536957, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10150 + }, + { + "epoch": 2.956066336921734, + "flow_matching_loss": 0.016903217229992152, + "grad_norm": 0.1761886179447174, + "learning_rate": 2.3670338196667528e-05, + "loss": 0.0259, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01354126976802945, + "rkd_loss_weight": 0.6615053832530975, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3908939719200134, + "rkd_teacher_entropy": 3.757301163673401, + "rkd_teacher_similarity_mean": 0.7859590411186218, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10160 + }, + { + "epoch": 2.9589758510328776, + "flow_matching_loss": 0.020162956696003674, + "grad_norm": 0.159865602850914, + "learning_rate": 2.3656840274230443e-05, + "loss": 0.0295, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014177128579467535, + "rkd_loss_weight": 0.6611720383167267, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3818633884191513, + "rkd_teacher_entropy": 3.717906284332275, + "rkd_teacher_similarity_mean": 0.7749974250793457, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10170 + }, + { + "epoch": 2.961885365144021, + "flow_matching_loss": 0.01706913476809859, + "grad_norm": 0.163112074136734, + "learning_rate": 2.3643331832916133e-05, + "loss": 0.0243, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011001302441582084, + "rkd_loss_weight": 0.6608386933803558, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38287157714366915, + "rkd_teacher_entropy": 3.736213517189026, + "rkd_teacher_similarity_mean": 0.7775068759918213, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10180 + }, + { + "epoch": 2.964794879255164, + "flow_matching_loss": 0.014932018984109164, + "grad_norm": 0.1564115285873413, + "learning_rate": 2.362981288913863e-05, + "loss": 0.0226, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011601447593420744, + "rkd_loss_weight": 0.6605053544044495, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3818147897720337, + "rkd_teacher_entropy": 3.7252434730529784, + "rkd_teacher_similarity_mean": 0.77584188580513, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10190 + }, + { + "epoch": 2.967704393366308, + "flow_matching_loss": 0.01614689715206623, + "grad_norm": 0.14628386497497559, + "learning_rate": 2.3616283459324737e-05, + "loss": 0.027, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01648045899346471, + "rkd_loss_weight": 0.6601720035076142, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39202064275741577, + "rkd_teacher_entropy": 3.748187232017517, + "rkd_teacher_similarity_mean": 0.7847728729248047, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10200 + }, + { + "epoch": 2.970613907477451, + "flow_matching_loss": 0.01728510968387127, + "grad_norm": 0.11071038991212845, + "learning_rate": 2.3602743559913973e-05, + "loss": 0.0253, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012214872054755688, + "rkd_loss_weight": 0.6598386585712432, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38857988715171815, + "rkd_teacher_entropy": 3.7481662750244142, + "rkd_teacher_similarity_mean": 0.783755874633789, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10210 + }, + { + "epoch": 2.973523421588595, + "flow_matching_loss": 0.019267561193555594, + "grad_norm": 0.1681351661682129, + "learning_rate": 2.35891932073586e-05, + "loss": 0.0269, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011525119561702013, + "rkd_loss_weight": 0.6595053195953369, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3830097258090973, + "rkd_teacher_entropy": 3.7192607164382934, + "rkd_teacher_similarity_mean": 0.7770178496837616, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10220 + }, + { + "epoch": 2.976432935699738, + "flow_matching_loss": 0.01623391481116414, + "grad_norm": 0.1642017364501953, + "learning_rate": 2.357563241812357e-05, + "loss": 0.0246, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012656236626207828, + "rkd_loss_weight": 0.6591719686985016, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38615486919879916, + "rkd_teacher_entropy": 3.728906750679016, + "rkd_teacher_similarity_mean": 0.7821174561977386, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10230 + }, + { + "epoch": 2.979342449810882, + "flow_matching_loss": 0.016418304387480022, + "grad_norm": 0.1639064997434616, + "learning_rate": 2.3562061208686522e-05, + "loss": 0.0238, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011207783967256546, + "rkd_loss_weight": 0.6588386297225952, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38392957746982576, + "rkd_teacher_entropy": 3.734687924385071, + "rkd_teacher_similarity_mean": 0.7807462275028229, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10240 + }, + { + "epoch": 2.982251963922025, + "flow_matching_loss": 0.015465098898857832, + "grad_norm": 0.24994607269763947, + "learning_rate": 2.354847959553776e-05, + "loss": 0.0244, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013644452393054961, + "rkd_loss_weight": 0.6585052847862244, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38756422102451327, + "rkd_teacher_entropy": 3.756576180458069, + "rkd_teacher_similarity_mean": 0.7863644182682037, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10250 + }, + { + "epoch": 2.9851614780331683, + "flow_matching_loss": 0.018093475792557, + "grad_norm": 0.1495356559753418, + "learning_rate": 2.353488759518022e-05, + "loss": 0.0257, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011534689459949732, + "rkd_loss_weight": 0.658171933889389, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38628058731555937, + "rkd_teacher_entropy": 3.7574617862701416, + "rkd_teacher_similarity_mean": 0.7865418374538422, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10260 + }, + { + "epoch": 2.988070992144312, + "flow_matching_loss": 0.017815558053553105, + "grad_norm": 0.14751626551151276, + "learning_rate": 2.352128522412946e-05, + "loss": 0.0257, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012025352474302054, + "rkd_loss_weight": 0.6578386008739472, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3749242812395096, + "rkd_teacher_entropy": 3.7162151098251344, + "rkd_teacher_similarity_mean": 0.7728440642356873, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10270 + }, + { + "epoch": 2.9909805062554553, + "flow_matching_loss": 0.018374893441796304, + "grad_norm": 0.16582545638084412, + "learning_rate": 2.3507672498913653e-05, + "loss": 0.0279, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014507231302559376, + "rkd_loss_weight": 0.6575052440166473, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3822404533624649, + "rkd_teacher_entropy": 3.73668258190155, + "rkd_teacher_similarity_mean": 0.7788557052612305, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10280 + }, + { + "epoch": 2.9938900203665986, + "flow_matching_loss": 0.017045415844768284, + "grad_norm": 0.14392361044883728, + "learning_rate": 2.3494049436073537e-05, + "loss": 0.0252, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012444437108933925, + "rkd_loss_weight": 0.657171905040741, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3859290271997452, + "rkd_teacher_entropy": 3.7399601459503176, + "rkd_teacher_similarity_mean": 0.7831000447273254, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10290 + }, + { + "epoch": 2.9967995344777423, + "flow_matching_loss": 0.01667530685663223, + "grad_norm": 0.19402773678302765, + "learning_rate": 2.348041605216242e-05, + "loss": 0.0256, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013577969698235392, + "rkd_loss_weight": 0.6568385660648346, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3900995284318924, + "rkd_teacher_entropy": 3.7577874660491943, + "rkd_teacher_similarity_mean": 0.7873945116996766, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10300 + }, + { + "epoch": 2.9997090485888855, + "flow_matching_loss": 0.01848538750782609, + "grad_norm": 0.16111914813518524, + "learning_rate": 2.3466772363746146e-05, + "loss": 0.0308, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.018777088820934297, + "rkd_loss_weight": 0.6565052092075347, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38361251056194307, + "rkd_teacher_entropy": 3.730931258201599, + "rkd_teacher_similarity_mean": 0.7793646931648255, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10310 + }, + { + "epoch": 3.0026185627000292, + "flow_matching_loss": 0.01532314931973815, + "grad_norm": 0.14481006562709808, + "learning_rate": 2.345311838740309e-05, + "loss": 0.0237, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012719385419040918, + "rkd_loss_weight": 0.6561718761920929, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.388114920258522, + "rkd_teacher_entropy": 3.7469650745391845, + "rkd_teacher_similarity_mean": 0.781953650712967, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10320 + }, + { + "epoch": 3.0055280768111725, + "flow_matching_loss": 0.018921857327222825, + "grad_norm": 0.15596336126327515, + "learning_rate": 2.343945413972411e-05, + "loss": 0.0289, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015166151709854603, + "rkd_loss_weight": 0.6558385252952575, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39290265142917635, + "rkd_teacher_entropy": 3.7582918405532837, + "rkd_teacher_similarity_mean": 0.7871361315250397, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10330 + }, + { + "epoch": 3.0084375909223158, + "flow_matching_loss": 0.016931130085140467, + "grad_norm": 0.19863052666187286, + "learning_rate": 2.3425779637312568e-05, + "loss": 0.0259, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013647411298006774, + "rkd_loss_weight": 0.6555051803588867, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38104387521743777, + "rkd_teacher_entropy": 3.728743314743042, + "rkd_teacher_similarity_mean": 0.7762851715087891, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10340 + }, + { + "epoch": 3.0113471050334595, + "flow_matching_loss": 0.018730608653277157, + "grad_norm": 0.16698810458183289, + "learning_rate": 2.3412094896784264e-05, + "loss": 0.0268, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012268327409401536, + "rkd_loss_weight": 0.6551718413829803, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3778208762407303, + "rkd_teacher_entropy": 3.717124819755554, + "rkd_teacher_similarity_mean": 0.7739063262939453, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10350 + }, + { + "epoch": 3.0142566191446027, + "flow_matching_loss": 0.017608135007321835, + "grad_norm": 0.182255819439888, + "learning_rate": 2.339839993476745e-05, + "loss": 0.0253, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01176717495545745, + "rkd_loss_weight": 0.6548384904861451, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38989475965499876, + "rkd_teacher_entropy": 3.748808431625366, + "rkd_teacher_similarity_mean": 0.7864301025867462, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10360 + }, + { + "epoch": 3.0171661332557465, + "flow_matching_loss": 0.01593589885160327, + "grad_norm": 0.1364821046590805, + "learning_rate": 2.3384694767902804e-05, + "loss": 0.0224, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009930986911058426, + "rkd_loss_weight": 0.6545051515102387, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38482820689678193, + "rkd_teacher_entropy": 3.7339629411697386, + "rkd_teacher_similarity_mean": 0.7810814023017884, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10370 + }, + { + "epoch": 3.0200756473668897, + "flow_matching_loss": 0.016885956469923257, + "grad_norm": 0.22870539128780365, + "learning_rate": 2.3370979412843384e-05, + "loss": 0.0258, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01361315455287695, + "rkd_loss_weight": 0.6541718065738678, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.390008756518364, + "rkd_teacher_entropy": 3.759810757637024, + "rkd_teacher_similarity_mean": 0.7839647710323334, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10380 + }, + { + "epoch": 3.022985161478033, + "flow_matching_loss": 0.017975575756281614, + "grad_norm": 0.14205241203308105, + "learning_rate": 2.3357253886254648e-05, + "loss": 0.026, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012207769975066184, + "rkd_loss_weight": 0.6538384616374969, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38411434888839724, + "rkd_teacher_entropy": 3.728217887878418, + "rkd_teacher_similarity_mean": 0.7792082786560058, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10390 + }, + { + "epoch": 3.0258946755891767, + "flow_matching_loss": 0.01763251507654786, + "grad_norm": 0.1755082905292511, + "learning_rate": 2.3343518204814407e-05, + "loss": 0.0245, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010512345284223557, + "rkd_loss_weight": 0.6535051167011261, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3874636501073837, + "rkd_teacher_entropy": 3.7515247106552123, + "rkd_teacher_similarity_mean": 0.7829421699047089, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10400 + }, + { + "epoch": 3.02880418970032, + "flow_matching_loss": 0.017749020643532277, + "grad_norm": 0.1426272988319397, + "learning_rate": 2.3329772385212798e-05, + "loss": 0.0264, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013224635273218155, + "rkd_loss_weight": 0.6531717777252197, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3814480245113373, + "rkd_teacher_entropy": 3.7280680179595946, + "rkd_teacher_similarity_mean": 0.7776778221130372, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10410 + }, + { + "epoch": 3.0317137038114637, + "flow_matching_loss": 0.01654777554795146, + "grad_norm": 0.17245639860630035, + "learning_rate": 2.3316016444152296e-05, + "loss": 0.0248, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012677379511296749, + "rkd_loss_weight": 0.6528384268283844, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3851331800222397, + "rkd_teacher_entropy": 3.741252326965332, + "rkd_teacher_similarity_mean": 0.7804071307182312, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10420 + }, + { + "epoch": 3.034623217922607, + "flow_matching_loss": 0.014759779628366231, + "grad_norm": 0.15695273876190186, + "learning_rate": 2.3302250398347668e-05, + "loss": 0.0222, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01134498007595539, + "rkd_loss_weight": 0.6525050818920135, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38111159205436707, + "rkd_teacher_entropy": 3.7351982831954955, + "rkd_teacher_similarity_mean": 0.7768213272094726, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10430 + }, + { + "epoch": 3.03753273203375, + "flow_matching_loss": 0.016330500412732363, + "grad_norm": 0.15533171594142914, + "learning_rate": 2.328847426452595e-05, + "loss": 0.0245, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012558474950492383, + "rkd_loss_weight": 0.6521717429161071, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39006958305835726, + "rkd_teacher_entropy": 3.7564226388931274, + "rkd_teacher_similarity_mean": 0.7888757646083832, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10440 + }, + { + "epoch": 3.040442246144894, + "flow_matching_loss": 0.016726392693817614, + "grad_norm": 0.17133258283138275, + "learning_rate": 2.3274688059426456e-05, + "loss": 0.0237, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010721101751551032, + "rkd_loss_weight": 0.6518383920192719, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3833125323057175, + "rkd_teacher_entropy": 3.741901969909668, + "rkd_teacher_similarity_mean": 0.7786294996738434, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10450 + }, + { + "epoch": 3.043351760256037, + "flow_matching_loss": 0.01876667160540819, + "grad_norm": 0.20177768170833588, + "learning_rate": 2.3260891799800715e-05, + "loss": 0.0265, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011799605190753936, + "rkd_loss_weight": 0.6515050530433655, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38855260908603667, + "rkd_teacher_entropy": 3.745826768875122, + "rkd_teacher_similarity_mean": 0.7850810945034027, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10460 + }, + { + "epoch": 3.046261274367181, + "flow_matching_loss": 0.015239734295755624, + "grad_norm": 0.143587127327919, + "learning_rate": 2.3247085502412482e-05, + "loss": 0.0218, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010103135462850332, + "rkd_loss_weight": 0.6511717081069947, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39886229336261747, + "rkd_teacher_entropy": 3.763743281364441, + "rkd_teacher_similarity_mean": 0.7928139567375183, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10470 + }, + { + "epoch": 3.049170788478324, + "flow_matching_loss": 0.01708298586308956, + "grad_norm": 0.14164236187934875, + "learning_rate": 2.3233269184037717e-05, + "loss": 0.0247, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011737679690122604, + "rkd_loss_weight": 0.6508383572101593, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3856168955564499, + "rkd_teacher_entropy": 3.7355026960372926, + "rkd_teacher_similarity_mean": 0.7812331557273865, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10480 + }, + { + "epoch": 3.0520803025894674, + "flow_matching_loss": 0.01872971635311842, + "grad_norm": 0.1758045107126236, + "learning_rate": 2.3219442861464547e-05, + "loss": 0.0259, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010957844369113446, + "rkd_loss_weight": 0.6505050241947175, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3935443878173828, + "rkd_teacher_entropy": 3.767397236824036, + "rkd_teacher_similarity_mean": 0.7886037826538086, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10490 + }, + { + "epoch": 3.054989816700611, + "flow_matching_loss": 0.01954898899421096, + "grad_norm": 0.15185725688934326, + "learning_rate": 2.3205606551493255e-05, + "loss": 0.027, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01146606421098113, + "rkd_loss_weight": 0.6501716673374176, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3780395120382309, + "rkd_teacher_entropy": 3.719839358329773, + "rkd_teacher_similarity_mean": 0.7724733710289001, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10500 + }, + { + "epoch": 3.0578993308117544, + "flow_matching_loss": 0.015511761140078306, + "grad_norm": 0.16549283266067505, + "learning_rate": 2.319176027093627e-05, + "loss": 0.0223, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010462925722822546, + "rkd_loss_weight": 0.6498383283615112, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3849914252758026, + "rkd_teacher_entropy": 3.747821402549744, + "rkd_teacher_similarity_mean": 0.7794607043266296, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10510 + }, + { + "epoch": 3.060808844922898, + "flow_matching_loss": 0.016012375708669424, + "grad_norm": 0.18217462301254272, + "learning_rate": 2.3177904036618117e-05, + "loss": 0.0234, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011322597227990627, + "rkd_loss_weight": 0.6495049893856049, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38494682908058164, + "rkd_teacher_entropy": 3.7502658367156982, + "rkd_teacher_similarity_mean": 0.7807362675666809, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10520 + }, + { + "epoch": 3.0637183590340413, + "flow_matching_loss": 0.01903521427884698, + "grad_norm": 0.19208504259586334, + "learning_rate": 2.316403786537544e-05, + "loss": 0.0262, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011107403691858054, + "rkd_loss_weight": 0.649171632528305, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38177012503147123, + "rkd_teacher_entropy": 3.735566329956055, + "rkd_teacher_similarity_mean": 0.7780948340892792, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10530 + }, + { + "epoch": 3.0666278731451846, + "flow_matching_loss": 0.017859334032982586, + "grad_norm": 0.14408385753631592, + "learning_rate": 2.3150161774056933e-05, + "loss": 0.0253, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011514544300734997, + "rkd_loss_weight": 0.6488382995128632, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39378444850444794, + "rkd_teacher_entropy": 3.7600104808807373, + "rkd_teacher_similarity_mean": 0.7877586305141449, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10540 + }, + { + "epoch": 3.0695373872563283, + "flow_matching_loss": 0.017962956242263316, + "grad_norm": 0.2038799524307251, + "learning_rate": 2.313627577952336e-05, + "loss": 0.026, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012454634252935647, + "rkd_loss_weight": 0.6485049486160278, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3883085370063782, + "rkd_teacher_entropy": 3.760346841812134, + "rkd_teacher_similarity_mean": 0.7835793554782867, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10550 + }, + { + "epoch": 3.0724469013674716, + "flow_matching_loss": 0.015558692533522845, + "grad_norm": 0.17766501009464264, + "learning_rate": 2.312237989864752e-05, + "loss": 0.0217, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009504569042474032, + "rkd_loss_weight": 0.648171603679657, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38583381474018097, + "rkd_teacher_entropy": 3.733658528327942, + "rkd_teacher_similarity_mean": 0.7799773573875427, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10560 + }, + { + "epoch": 3.0753564154786153, + "flow_matching_loss": 0.0172104231081903, + "grad_norm": 0.16774959862232208, + "learning_rate": 2.3108474148314205e-05, + "loss": 0.0261, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013701723096892238, + "rkd_loss_weight": 0.6478382647037506, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3829788684844971, + "rkd_teacher_entropy": 3.731401824951172, + "rkd_teacher_similarity_mean": 0.7784486234188079, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10570 + }, + { + "epoch": 3.0782659295897585, + "flow_matching_loss": 0.01478834468871355, + "grad_norm": 0.17144335806369781, + "learning_rate": 2.309455854542023e-05, + "loss": 0.0215, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010359390638768673, + "rkd_loss_weight": 0.6475049138069153, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3884860187768936, + "rkd_teacher_entropy": 3.744316244125366, + "rkd_teacher_similarity_mean": 0.7839914441108704, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10580 + }, + { + "epoch": 3.081175443700902, + "flow_matching_loss": 0.01684037148952484, + "grad_norm": 0.1761188805103302, + "learning_rate": 2.308063310687435e-05, + "loss": 0.0247, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012079469161108136, + "rkd_loss_weight": 0.647171574831009, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38556026220321654, + "rkd_teacher_entropy": 3.7445241451263427, + "rkd_teacher_similarity_mean": 0.7802174031734467, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10590 + }, + { + "epoch": 3.0840849578120455, + "flow_matching_loss": 0.01671651201322675, + "grad_norm": 0.16885124146938324, + "learning_rate": 2.306669784959729e-05, + "loss": 0.025, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01286209924146533, + "rkd_loss_weight": 0.646838229894638, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3845379054546356, + "rkd_teacher_entropy": 3.7465086936950684, + "rkd_teacher_similarity_mean": 0.7812074959278107, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10600 + }, + { + "epoch": 3.0869944719231888, + "flow_matching_loss": 0.015438118670135736, + "grad_norm": 0.12437096238136292, + "learning_rate": 2.3052752790521703e-05, + "loss": 0.0227, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011158321611583233, + "rkd_loss_weight": 0.6465048789978027, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38340153694152834, + "rkd_teacher_entropy": 3.724165439605713, + "rkd_teacher_similarity_mean": 0.7794331252574921, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10610 + }, + { + "epoch": 3.0899039860343325, + "flow_matching_loss": 0.018247523065656423, + "grad_norm": 0.16984866559505463, + "learning_rate": 2.3038797946592152e-05, + "loss": 0.0255, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011245842091739178, + "rkd_loss_weight": 0.6461715400218964, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37414298951625824, + "rkd_teacher_entropy": 3.7149710178375246, + "rkd_teacher_similarity_mean": 0.7718745350837708, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10620 + }, + { + "epoch": 3.0928135001454757, + "flow_matching_loss": 0.017494655586779116, + "grad_norm": 0.14904624223709106, + "learning_rate": 2.3024833334765087e-05, + "loss": 0.0249, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011466848477721215, + "rkd_loss_weight": 0.6458381950855255, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38682331144809723, + "rkd_teacher_entropy": 3.743308687210083, + "rkd_teacher_similarity_mean": 0.7837560892105102, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10630 + }, + { + "epoch": 3.095723014256619, + "flow_matching_loss": 0.019700941815972327, + "grad_norm": 0.09892714023590088, + "learning_rate": 2.301085897200883e-05, + "loss": 0.0283, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013333364762365818, + "rkd_loss_weight": 0.6455048501491547, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3916597247123718, + "rkd_teacher_entropy": 3.762343239784241, + "rkd_teacher_similarity_mean": 0.7878194510936737, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10640 + }, + { + "epoch": 3.0986325283677627, + "flow_matching_loss": 0.018050091993063688, + "grad_norm": 0.17129267752170563, + "learning_rate": 2.299687487530354e-05, + "loss": 0.025, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01079693716019392, + "rkd_loss_weight": 0.6451715052127838, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38461431562900544, + "rkd_teacher_entropy": 3.7326796531677244, + "rkd_teacher_similarity_mean": 0.7799881160259247, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10650 + }, + { + "epoch": 3.101542042478906, + "flow_matching_loss": 0.014123800955712796, + "grad_norm": 0.1841300129890442, + "learning_rate": 2.2982881061641228e-05, + "loss": 0.0211, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010787365678697824, + "rkd_loss_weight": 0.6448381602764129, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3866382032632828, + "rkd_teacher_entropy": 3.748477911949158, + "rkd_teacher_similarity_mean": 0.7822773516178131, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10660 + }, + { + "epoch": 3.1044515565900497, + "flow_matching_loss": 0.01684667207300663, + "grad_norm": 0.14583805203437805, + "learning_rate": 2.296887754802569e-05, + "loss": 0.0265, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014950532559305429, + "rkd_loss_weight": 0.6445048153400421, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3842517614364624, + "rkd_teacher_entropy": 3.7327388048172, + "rkd_teacher_similarity_mean": 0.7793650269508362, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10670 + }, + { + "epoch": 3.107361070701193, + "flow_matching_loss": 0.018288742657750844, + "grad_norm": 0.18398620188236237, + "learning_rate": 2.295486435147251e-05, + "loss": 0.0264, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012627397663891315, + "rkd_loss_weight": 0.6441714763641357, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3790604412555695, + "rkd_teacher_entropy": 3.7111181020736694, + "rkd_teacher_similarity_mean": 0.7708229005336762, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10680 + }, + { + "epoch": 3.110270584812336, + "flow_matching_loss": 0.018035299703478815, + "grad_norm": 0.1766490638256073, + "learning_rate": 2.294084148900905e-05, + "loss": 0.0257, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011891122488304973, + "rkd_loss_weight": 0.6438381254673005, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3786309003829956, + "rkd_teacher_entropy": 3.717085933685303, + "rkd_teacher_similarity_mean": 0.7733356416225433, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10690 + }, + { + "epoch": 3.11318009892348, + "flow_matching_loss": 0.017189454287290573, + "grad_norm": 0.11521129310131073, + "learning_rate": 2.292680897767441e-05, + "loss": 0.0244, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011169989593327045, + "rkd_loss_weight": 0.6435047805309295, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38963864743709564, + "rkd_teacher_entropy": 3.750740671157837, + "rkd_teacher_similarity_mean": 0.7853134453296662, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10700 + }, + { + "epoch": 3.116089613034623, + "flow_matching_loss": 0.017643094621598722, + "grad_norm": 0.15014751255512238, + "learning_rate": 2.291276683451941e-05, + "loss": 0.0234, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009027444710955023, + "rkd_loss_weight": 0.6431714415550231, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3798526138067245, + "rkd_teacher_entropy": 3.719565010070801, + "rkd_teacher_similarity_mean": 0.7743006110191345, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10710 + }, + { + "epoch": 3.1189991271457664, + "flow_matching_loss": 0.018105142191052435, + "grad_norm": 0.14146864414215088, + "learning_rate": 2.2898715076606584e-05, + "loss": 0.0253, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011127169243991375, + "rkd_loss_weight": 0.6428380906581879, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38711909353733065, + "rkd_teacher_entropy": 3.749300646781921, + "rkd_teacher_similarity_mean": 0.783002358675003, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10720 + }, + { + "epoch": 3.12190864125691, + "flow_matching_loss": 0.017442796379327774, + "grad_norm": 0.28545060753822327, + "learning_rate": 2.2884653721010135e-05, + "loss": 0.0251, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011969769466668367, + "rkd_loss_weight": 0.6425047516822815, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3889099955558777, + "rkd_teacher_entropy": 3.74605507850647, + "rkd_teacher_similarity_mean": 0.7831062495708465, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10730 + }, + { + "epoch": 3.1248181553680534, + "flow_matching_loss": 0.017408666387200355, + "grad_norm": 0.12007831782102585, + "learning_rate": 2.2870582784815946e-05, + "loss": 0.0236, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00962788905017078, + "rkd_loss_weight": 0.6421714067459107, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3861252874135971, + "rkd_teacher_entropy": 3.7463993072509765, + "rkd_teacher_similarity_mean": 0.7815783977508545, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10740 + }, + { + "epoch": 3.127727669479197, + "flow_matching_loss": 0.017149493843317033, + "grad_norm": 0.21054258942604065, + "learning_rate": 2.2856502285121525e-05, + "loss": 0.0235, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009956760657951235, + "rkd_loss_weight": 0.6418380558490753, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39232295751571655, + "rkd_teacher_entropy": 3.75872163772583, + "rkd_teacher_similarity_mean": 0.7860643029212951, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10750 + }, + { + "epoch": 3.1306371835903404, + "flow_matching_loss": 0.01754650576040149, + "grad_norm": 0.16641564667224884, + "learning_rate": 2.2842412239036004e-05, + "loss": 0.0256, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012574689090251922, + "rkd_loss_weight": 0.6415047228336335, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38662591576576233, + "rkd_teacher_entropy": 3.7379306077957155, + "rkd_teacher_similarity_mean": 0.7788321554660798, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10760 + }, + { + "epoch": 3.133546697701484, + "flow_matching_loss": 0.01558292331174016, + "grad_norm": 0.1541357785463333, + "learning_rate": 2.2828312663680125e-05, + "loss": 0.0225, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010740065155550838, + "rkd_loss_weight": 0.6411713659763336, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39395509362220765, + "rkd_teacher_entropy": 3.7630926609039306, + "rkd_teacher_similarity_mean": 0.7882080376148224, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10770 + }, + { + "epoch": 3.1364562118126273, + "flow_matching_loss": 0.013893318176269532, + "grad_norm": 0.10407712310552597, + "learning_rate": 2.2814203576186194e-05, + "loss": 0.0211, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011232395889237523, + "rkd_loss_weight": 0.6408380270004272, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3837391197681427, + "rkd_teacher_entropy": 3.7221672773361205, + "rkd_teacher_similarity_mean": 0.7754381895065308, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10780 + }, + { + "epoch": 3.1393657259237706, + "flow_matching_loss": 0.015058291982859373, + "grad_norm": 0.15075957775115967, + "learning_rate": 2.2800084993698093e-05, + "loss": 0.0221, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010967167234048248, + "rkd_loss_weight": 0.6405046880245209, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39513094127178194, + "rkd_teacher_entropy": 3.7489492654800416, + "rkd_teacher_similarity_mean": 0.7861159920692444, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10790 + }, + { + "epoch": 3.1422752400349143, + "flow_matching_loss": 0.018512946646660568, + "grad_norm": 0.1761058121919632, + "learning_rate": 2.2785956933371224e-05, + "loss": 0.0265, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012463211873546242, + "rkd_loss_weight": 0.640171331167221, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3829755634069443, + "rkd_teacher_entropy": 3.741734433174133, + "rkd_teacher_similarity_mean": 0.7740961015224457, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10800 + }, + { + "epoch": 3.1451847541460576, + "flow_matching_loss": 0.01850140979513526, + "grad_norm": 0.1728128045797348, + "learning_rate": 2.2771819412372505e-05, + "loss": 0.0252, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010397023009136318, + "rkd_loss_weight": 0.6398379981517792, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38800583481788636, + "rkd_teacher_entropy": 3.7259360790252685, + "rkd_teacher_similarity_mean": 0.7793304204940796, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10810 + }, + { + "epoch": 3.148094268257201, + "flow_matching_loss": 0.017310132458806037, + "grad_norm": 0.14593851566314697, + "learning_rate": 2.2757672447880374e-05, + "loss": 0.0236, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009814719809219241, + "rkd_loss_weight": 0.6395046532154083, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.389096611738205, + "rkd_teacher_entropy": 3.7348784446716308, + "rkd_teacher_similarity_mean": 0.7810262084007263, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10820 + }, + { + "epoch": 3.1510037823683446, + "flow_matching_loss": 0.015792532451450826, + "grad_norm": 0.13569365441799164, + "learning_rate": 2.2743516057084712e-05, + "loss": 0.0233, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01181302135810256, + "rkd_loss_weight": 0.639171302318573, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3801282286643982, + "rkd_teacher_entropy": 3.715265655517578, + "rkd_teacher_similarity_mean": 0.7734374761581421, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10830 + }, + { + "epoch": 3.153913296479488, + "flow_matching_loss": 0.017197334486991168, + "grad_norm": 0.15259817242622375, + "learning_rate": 2.2729350257186877e-05, + "loss": 0.025, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012232630979269743, + "rkd_loss_weight": 0.6388379633426666, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38317736983299255, + "rkd_teacher_entropy": 3.724532890319824, + "rkd_teacher_similarity_mean": 0.7759342789649963, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10840 + }, + { + "epoch": 3.1568228105906315, + "flow_matching_loss": 0.017700184229761363, + "grad_norm": 0.17589175701141357, + "learning_rate": 2.2715175065399655e-05, + "loss": 0.0241, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009980657370761037, + "rkd_loss_weight": 0.6385046184062958, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39768085479736326, + "rkd_teacher_entropy": 3.7685636281967163, + "rkd_teacher_similarity_mean": 0.7904402554035187, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10850 + }, + { + "epoch": 3.159732324701775, + "flow_matching_loss": 0.016454870719462632, + "grad_norm": 0.20878398418426514, + "learning_rate": 2.270099049894723e-05, + "loss": 0.0244, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01238852646201849, + "rkd_loss_weight": 0.638171273469925, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39094178974628446, + "rkd_teacher_entropy": 3.7576477766036986, + "rkd_teacher_similarity_mean": 0.786439836025238, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10860 + }, + { + "epoch": 3.162641838812918, + "flow_matching_loss": 0.01708325371146202, + "grad_norm": 0.155100479722023, + "learning_rate": 2.268679657506519e-05, + "loss": 0.0241, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010965378861874342, + "rkd_loss_weight": 0.637837928533554, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38992711901664734, + "rkd_teacher_entropy": 3.7534733772277833, + "rkd_teacher_similarity_mean": 0.7857793748378754, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10870 + }, + { + "epoch": 3.1655513529240618, + "flow_matching_loss": 0.016811232827603818, + "grad_norm": 0.15559618175029755, + "learning_rate": 2.2672593311000498e-05, + "loss": 0.024, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011253982642665505, + "rkd_loss_weight": 0.6375045835971832, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3877961367368698, + "rkd_teacher_entropy": 3.737868618965149, + "rkd_teacher_similarity_mean": 0.7812384605407715, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10880 + }, + { + "epoch": 3.168460867035205, + "flow_matching_loss": 0.016536066308617592, + "grad_norm": 0.16143955290317535, + "learning_rate": 2.2658380724011452e-05, + "loss": 0.0236, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011116801481693982, + "rkd_loss_weight": 0.6371712386608124, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3920976907014847, + "rkd_teacher_entropy": 3.757705330848694, + "rkd_teacher_similarity_mean": 0.7851015567779541, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10890 + }, + { + "epoch": 3.1713703811463487, + "flow_matching_loss": 0.016076804231852294, + "grad_norm": 0.17058715224266052, + "learning_rate": 2.264415883136769e-05, + "loss": 0.023, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010832090256735682, + "rkd_loss_weight": 0.636837899684906, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39240422546863557, + "rkd_teacher_entropy": 3.7576799392700195, + "rkd_teacher_similarity_mean": 0.7877068400382996, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10900 + }, + { + "epoch": 3.174279895257492, + "flow_matching_loss": 0.01693864883854985, + "grad_norm": 0.1355966478586197, + "learning_rate": 2.262992765035015e-05, + "loss": 0.0237, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010632106568664312, + "rkd_loss_weight": 0.6365045487880707, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3868690222501755, + "rkd_teacher_entropy": 3.7419362545013426, + "rkd_teacher_similarity_mean": 0.7799082934856415, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10910 + }, + { + "epoch": 3.1771894093686353, + "flow_matching_loss": 0.01575948754325509, + "grad_norm": 0.15265585482120514, + "learning_rate": 2.2615687198251062e-05, + "loss": 0.0226, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010717552062124014, + "rkd_loss_weight": 0.6361712038516998, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39111167192459106, + "rkd_teacher_entropy": 3.744927501678467, + "rkd_teacher_similarity_mean": 0.783619862794876, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10920 + }, + { + "epoch": 3.180098923479779, + "flow_matching_loss": 0.017850722931325436, + "grad_norm": 0.2966863214969635, + "learning_rate": 2.2601437492373914e-05, + "loss": 0.0269, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01423080200329423, + "rkd_loss_weight": 0.6358378648757934, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3955475836992264, + "rkd_teacher_entropy": 3.759894108772278, + "rkd_teacher_similarity_mean": 0.7863038778305054, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10930 + }, + { + "epoch": 3.1830084375909222, + "flow_matching_loss": 0.016600665543228386, + "grad_norm": 0.15191422402858734, + "learning_rate": 2.2587178550033444e-05, + "loss": 0.0236, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010960574494674802, + "rkd_loss_weight": 0.6355045139789581, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39238450229167937, + "rkd_teacher_entropy": 3.736569952964783, + "rkd_teacher_similarity_mean": 0.7837826609611511, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10940 + }, + { + "epoch": 3.185917951702066, + "flow_matching_loss": 0.015479742642492056, + "grad_norm": 0.13541944324970245, + "learning_rate": 2.2572910388555615e-05, + "loss": 0.0224, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010839787870645523, + "rkd_loss_weight": 0.6351711750030518, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3820787727832794, + "rkd_teacher_entropy": 3.7316967487335204, + "rkd_teacher_similarity_mean": 0.775322437286377, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10950 + }, + { + "epoch": 3.188827465813209, + "flow_matching_loss": 0.015541072655469178, + "grad_norm": 0.14996300637722015, + "learning_rate": 2.255863302527758e-05, + "loss": 0.0232, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01212801062501967, + "rkd_loss_weight": 0.6348378300666809, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38938689827919004, + "rkd_teacher_entropy": 3.747868061065674, + "rkd_teacher_similarity_mean": 0.7819258272647858, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10960 + }, + { + "epoch": 3.1917369799243525, + "flow_matching_loss": 0.019048788119107485, + "grad_norm": 0.14305375516414642, + "learning_rate": 2.2544346477547685e-05, + "loss": 0.0267, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012124477792531252, + "rkd_loss_weight": 0.6345044791698455, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38512001037597654, + "rkd_teacher_entropy": 3.7249869108200073, + "rkd_teacher_similarity_mean": 0.7791773200035095, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10970 + }, + { + "epoch": 3.194646494035496, + "flow_matching_loss": 0.015618563164025545, + "grad_norm": 0.15105904638767242, + "learning_rate": 2.253005076272544e-05, + "loss": 0.0232, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01202785475179553, + "rkd_loss_weight": 0.6341711461544037, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.386173403263092, + "rkd_teacher_entropy": 3.737831878662109, + "rkd_teacher_similarity_mean": 0.7803730547428132, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10980 + }, + { + "epoch": 3.1975560081466394, + "flow_matching_loss": 0.019244041293859482, + "grad_norm": 0.16925689578056335, + "learning_rate": 2.2515745898181477e-05, + "loss": 0.028, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013814286841079593, + "rkd_loss_weight": 0.6338377892971039, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39134745597839354, + "rkd_teacher_entropy": 3.7625098705291746, + "rkd_teacher_similarity_mean": 0.786723929643631, + "rkd_vlm_temp": 0.10000000149011612, + "step": 10990 + }, + { + "epoch": 3.200465522257783, + "flow_matching_loss": 0.01630673622712493, + "grad_norm": 0.1612452268600464, + "learning_rate": 2.250143190129756e-05, + "loss": 0.0234, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011236261529847979, + "rkd_loss_weight": 0.6335044503211975, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3875582695007324, + "rkd_teacher_entropy": 3.7396472454071046, + "rkd_teacher_similarity_mean": 0.7820027649402619, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11000 + }, + { + "epoch": 3.2033750363689264, + "flow_matching_loss": 0.017308276146650314, + "grad_norm": 0.1720784306526184, + "learning_rate": 2.2487108789466547e-05, + "loss": 0.025, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012115218210965395, + "rkd_loss_weight": 0.6331711113452911, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3882460057735443, + "rkd_teacher_entropy": 3.7473923683166506, + "rkd_teacher_similarity_mean": 0.783160787820816, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11010 + }, + { + "epoch": 3.2062845504800697, + "flow_matching_loss": 0.017311785090714692, + "grad_norm": 0.19802291691303253, + "learning_rate": 2.247277658009236e-05, + "loss": 0.0245, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01131708393804729, + "rkd_loss_weight": 0.6328377544879913, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38347095549106597, + "rkd_teacher_entropy": 3.7372856378555297, + "rkd_teacher_similarity_mean": 0.7791749596595764, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11020 + }, + { + "epoch": 3.2091940645912134, + "flow_matching_loss": 0.016449300665408374, + "grad_norm": 0.1495160460472107, + "learning_rate": 2.245843529059e-05, + "loss": 0.0235, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011160812061280011, + "rkd_loss_weight": 0.6325044214725495, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3806987822055817, + "rkd_teacher_entropy": 3.720437502861023, + "rkd_teacher_similarity_mean": 0.7743231534957886, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11030 + }, + { + "epoch": 3.2121035787023566, + "flow_matching_loss": 0.015801646932959556, + "grad_norm": 0.16264475882053375, + "learning_rate": 2.244408493838547e-05, + "loss": 0.0232, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011628643702715635, + "rkd_loss_weight": 0.6321710705757141, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39239626526832583, + "rkd_teacher_entropy": 3.7521105766296388, + "rkd_teacher_similarity_mean": 0.7855733275413513, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11040 + }, + { + "epoch": 3.2150130928135003, + "flow_matching_loss": 0.017693298775702714, + "grad_norm": 0.165726900100708, + "learning_rate": 2.242972554091581e-05, + "loss": 0.0255, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012341765593737364, + "rkd_loss_weight": 0.6318377256393433, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3867313265800476, + "rkd_teacher_entropy": 3.750403571128845, + "rkd_teacher_similarity_mean": 0.7820913195610046, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11050 + }, + { + "epoch": 3.2179226069246436, + "flow_matching_loss": 0.016449771448969842, + "grad_norm": 0.1287887543439865, + "learning_rate": 2.2415357115629045e-05, + "loss": 0.0242, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012281953170895577, + "rkd_loss_weight": 0.6315043866634369, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38385416865348815, + "rkd_teacher_entropy": 3.731338930130005, + "rkd_teacher_similarity_mean": 0.7770849227905273, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11060 + }, + { + "epoch": 3.220832121035787, + "flow_matching_loss": 0.016010722238570453, + "grad_norm": 0.16439589858055115, + "learning_rate": 2.2400979679984162e-05, + "loss": 0.0236, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012017310131341219, + "rkd_loss_weight": 0.6311710357666016, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3882849633693695, + "rkd_teacher_entropy": 3.745312142372131, + "rkd_teacher_similarity_mean": 0.7827070415019989, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11070 + }, + { + "epoch": 3.2237416351469306, + "flow_matching_loss": 0.01737408507615328, + "grad_norm": 0.17141661047935486, + "learning_rate": 2.2386593251451096e-05, + "loss": 0.0242, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010748044680804014, + "rkd_loss_weight": 0.6308376967906952, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3852545738220215, + "rkd_teacher_entropy": 3.723677372932434, + "rkd_teacher_similarity_mean": 0.7797451913356781, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11080 + }, + { + "epoch": 3.226651149258074, + "flow_matching_loss": 0.01615849370136857, + "grad_norm": 0.1626216024160385, + "learning_rate": 2.2372197847510722e-05, + "loss": 0.0236, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011830829549580812, + "rkd_loss_weight": 0.6305043518543243, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.380439031124115, + "rkd_teacher_entropy": 3.719100069999695, + "rkd_teacher_similarity_mean": 0.7746361553668976, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11090 + }, + { + "epoch": 3.2295606633692175, + "flow_matching_loss": 0.017386616859585046, + "grad_norm": 0.1280246376991272, + "learning_rate": 2.2357793485654813e-05, + "loss": 0.0242, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010767446551471948, + "rkd_loss_weight": 0.630171000957489, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37946549952030184, + "rkd_teacher_entropy": 3.7216002225875853, + "rkd_teacher_similarity_mean": 0.7751791894435882, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11100 + }, + { + "epoch": 3.232470177480361, + "flow_matching_loss": 0.016325320582836866, + "grad_norm": 0.16162382066249847, + "learning_rate": 2.2343380183386017e-05, + "loss": 0.0234, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011291976971551775, + "rkd_loss_weight": 0.6298376619815826, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39274920523166656, + "rkd_teacher_entropy": 3.7603158473968508, + "rkd_teacher_similarity_mean": 0.7879452228546142, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11110 + }, + { + "epoch": 3.235379691591504, + "flow_matching_loss": 0.017542296182364227, + "grad_norm": 0.1558889001607895, + "learning_rate": 2.232895795821787e-05, + "loss": 0.0242, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01052694795653224, + "rkd_loss_weight": 0.6295043170452118, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3874602347612381, + "rkd_teacher_entropy": 3.7449076414108275, + "rkd_teacher_similarity_mean": 0.7833570122718811, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11120 + }, + { + "epoch": 3.2382892057026478, + "flow_matching_loss": 0.018643393646925688, + "grad_norm": 0.13588443398475647, + "learning_rate": 2.2314526827674723e-05, + "loss": 0.0267, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012834086362272502, + "rkd_loss_weight": 0.629170972108841, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.384564071893692, + "rkd_teacher_entropy": 3.737543058395386, + "rkd_teacher_similarity_mean": 0.7776406705379486, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11130 + }, + { + "epoch": 3.241198719813791, + "flow_matching_loss": 0.015871358662843706, + "grad_norm": 0.12866896390914917, + "learning_rate": 2.230008680929177e-05, + "loss": 0.0226, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01066657779738307, + "rkd_loss_weight": 0.62883762717247, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38453627228736875, + "rkd_teacher_entropy": 3.7363085985183715, + "rkd_teacher_similarity_mean": 0.7798204958438874, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11140 + }, + { + "epoch": 3.2441082339249343, + "flow_matching_loss": 0.014459114987403154, + "grad_norm": 0.1179862841963768, + "learning_rate": 2.2285637920614987e-05, + "loss": 0.0203, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009353065025061369, + "rkd_loss_weight": 0.6285042822360992, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38705967664718627, + "rkd_teacher_entropy": 3.737805438041687, + "rkd_teacher_similarity_mean": 0.7805473566055298, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11150 + }, + { + "epoch": 3.247017748036078, + "flow_matching_loss": 0.01788745988160372, + "grad_norm": 0.16794069111347198, + "learning_rate": 2.2271180179201143e-05, + "loss": 0.024, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009705005586147309, + "rkd_loss_weight": 0.6281709372997284, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38726518750190736, + "rkd_teacher_entropy": 3.736992049217224, + "rkd_teacher_similarity_mean": 0.7806683659553528, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11160 + }, + { + "epoch": 3.2499272621472213, + "flow_matching_loss": 0.017688824981451034, + "grad_norm": 0.1267317533493042, + "learning_rate": 2.225671360261775e-05, + "loss": 0.0242, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010345373814925552, + "rkd_loss_weight": 0.627837598323822, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3904252976179123, + "rkd_teacher_entropy": 3.7540201902389527, + "rkd_teacher_similarity_mean": 0.7861678779125214, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11170 + }, + { + "epoch": 3.252836776258365, + "flow_matching_loss": 0.01490089176222682, + "grad_norm": 0.13456200063228607, + "learning_rate": 2.2242238208443068e-05, + "loss": 0.0222, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011554892454296351, + "rkd_loss_weight": 0.6275042474269867, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3886301159858704, + "rkd_teacher_entropy": 3.747485566139221, + "rkd_teacher_similarity_mean": 0.7833233654499054, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11180 + }, + { + "epoch": 3.2557462903695082, + "flow_matching_loss": 0.019579137954860926, + "grad_norm": 0.11868266761302948, + "learning_rate": 2.2227754014266063e-05, + "loss": 0.0276, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01272027064114809, + "rkd_loss_weight": 0.6271709024906158, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3755675137042999, + "rkd_teacher_entropy": 3.716787242889404, + "rkd_teacher_similarity_mean": 0.7713437616825104, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11190 + }, + { + "epoch": 3.258655804480652, + "flow_matching_loss": 0.017323657963424922, + "grad_norm": 0.17655706405639648, + "learning_rate": 2.22132610376864e-05, + "loss": 0.0233, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009517045738175512, + "rkd_loss_weight": 0.6268375635147094, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3856132537126541, + "rkd_teacher_entropy": 3.7376088857650758, + "rkd_teacher_similarity_mean": 0.780953311920166, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11200 + }, + { + "epoch": 3.261565318591795, + "flow_matching_loss": 0.018193596228957176, + "grad_norm": 0.24757502973079681, + "learning_rate": 2.219875929631441e-05, + "loss": 0.0259, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012240152154117823, + "rkd_loss_weight": 0.6265042126178741, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39082875549793245, + "rkd_teacher_entropy": 3.750091528892517, + "rkd_teacher_similarity_mean": 0.7842172622680664, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11210 + }, + { + "epoch": 3.2644748327029385, + "flow_matching_loss": 0.014998449105769395, + "grad_norm": 0.11410510540008545, + "learning_rate": 2.2184248807771074e-05, + "loss": 0.0208, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009227023040875793, + "rkd_loss_weight": 0.6261708736419678, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3891926288604736, + "rkd_teacher_entropy": 3.7441224813461305, + "rkd_teacher_similarity_mean": 0.7829852521419525, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11220 + }, + { + "epoch": 3.267384346814082, + "flow_matching_loss": 0.020681681111454964, + "grad_norm": 0.15586717426776886, + "learning_rate": 2.216972958968801e-05, + "loss": 0.0287, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012886155024170875, + "rkd_loss_weight": 0.625837528705597, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38795039653778074, + "rkd_teacher_entropy": 3.750031900405884, + "rkd_teacher_similarity_mean": 0.7833163380622864, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11230 + }, + { + "epoch": 3.2702938609252254, + "flow_matching_loss": 0.01704757222905755, + "grad_norm": 0.13380026817321777, + "learning_rate": 2.215520165970743e-05, + "loss": 0.0267, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01539902100339532, + "rkd_loss_weight": 0.6255041778087616, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38498568832874297, + "rkd_teacher_entropy": 3.7372222423553465, + "rkd_teacher_similarity_mean": 0.7772565126419068, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11240 + }, + { + "epoch": 3.2732033750363687, + "flow_matching_loss": 0.017864474561065435, + "grad_norm": 0.13479965925216675, + "learning_rate": 2.2140665035482143e-05, + "loss": 0.0275, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.015490110591053962, + "rkd_loss_weight": 0.6251708447933197, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3869875520467758, + "rkd_teacher_entropy": 3.738211226463318, + "rkd_teacher_similarity_mean": 0.7807332396507263, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11250 + }, + { + "epoch": 3.2761128891475124, + "flow_matching_loss": 0.01659601414576173, + "grad_norm": 0.2397838830947876, + "learning_rate": 2.2126119734675518e-05, + "loss": 0.0248, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013116379268467426, + "rkd_loss_weight": 0.6248374938964844, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3792066961526871, + "rkd_teacher_entropy": 3.7151889562606812, + "rkd_teacher_similarity_mean": 0.7741345703601837, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11260 + }, + { + "epoch": 3.2790224032586557, + "flow_matching_loss": 0.016200414393097162, + "grad_norm": 0.138925701379776, + "learning_rate": 2.2111565774961463e-05, + "loss": 0.0227, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010437358729541301, + "rkd_loss_weight": 0.6245041489601135, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3816437393426895, + "rkd_teacher_entropy": 3.7239464282989503, + "rkd_teacher_similarity_mean": 0.7758231461048126, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11270 + }, + { + "epoch": 3.2819319173697994, + "flow_matching_loss": 0.015554889012128115, + "grad_norm": 0.14352573454380035, + "learning_rate": 2.2097003174024417e-05, + "loss": 0.0239, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013413448166102172, + "rkd_loss_weight": 0.6241708099842072, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.394639652967453, + "rkd_teacher_entropy": 3.7556442975997926, + "rkd_teacher_similarity_mean": 0.7871138155460358, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11280 + }, + { + "epoch": 3.2848414314809427, + "flow_matching_loss": 0.014395058900117875, + "grad_norm": 0.15217486023902893, + "learning_rate": 2.2082431949559315e-05, + "loss": 0.0205, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009716271329671145, + "rkd_loss_weight": 0.6238374590873719, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3902548760175705, + "rkd_teacher_entropy": 3.7425126791000367, + "rkd_teacher_similarity_mean": 0.782868480682373, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11290 + }, + { + "epoch": 3.2877509455920864, + "flow_matching_loss": 0.015967493504285814, + "grad_norm": 0.16952024400234222, + "learning_rate": 2.2067852119271558e-05, + "loss": 0.0245, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013756373152136802, + "rkd_loss_weight": 0.6235041201114655, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3877974718809128, + "rkd_teacher_entropy": 3.7405576705932617, + "rkd_teacher_similarity_mean": 0.7826291084289551, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11300 + }, + { + "epoch": 3.2906604597032296, + "flow_matching_loss": 0.016387046594172717, + "grad_norm": 0.14367839694023132, + "learning_rate": 2.2053263700877026e-05, + "loss": 0.0245, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012951045390218496, + "rkd_loss_weight": 0.6231707751750946, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38465302586555483, + "rkd_teacher_entropy": 3.739955925941467, + "rkd_teacher_similarity_mean": 0.7799034655094147, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11310 + }, + { + "epoch": 3.293569973814373, + "flow_matching_loss": 0.015947797521948814, + "grad_norm": 0.12445977330207825, + "learning_rate": 2.2038666712102015e-05, + "loss": 0.0227, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010829831752926112, + "rkd_loss_weight": 0.6228374242782593, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3778146028518677, + "rkd_teacher_entropy": 3.7017706871032714, + "rkd_teacher_similarity_mean": 0.7722583293914795, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11320 + }, + { + "epoch": 3.2964794879255166, + "flow_matching_loss": 0.016011057142168283, + "grad_norm": 0.11094935983419418, + "learning_rate": 2.2024061170683246e-05, + "loss": 0.0229, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011140075884759426, + "rkd_loss_weight": 0.6225040853023529, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38544503450393675, + "rkd_teacher_entropy": 3.7396785974502564, + "rkd_teacher_similarity_mean": 0.7804378867149353, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11330 + }, + { + "epoch": 3.29938900203666, + "flow_matching_loss": 0.018949488922953606, + "grad_norm": 0.14711372554302216, + "learning_rate": 2.200944709436783e-05, + "loss": 0.0279, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014439681265503167, + "rkd_loss_weight": 0.6221707403659821, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38073945343494414, + "rkd_teacher_entropy": 3.7383256435394285, + "rkd_teacher_similarity_mean": 0.7753138780593872, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11340 + }, + { + "epoch": 3.302298516147803, + "flow_matching_loss": 0.018098848406225443, + "grad_norm": 0.1894487887620926, + "learning_rate": 2.1994824500913246e-05, + "loss": 0.025, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011044952832162381, + "rkd_loss_weight": 0.6218373954296113, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38306685090065, + "rkd_teacher_entropy": 3.728051280975342, + "rkd_teacher_similarity_mean": 0.7797475576400756, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11350 + }, + { + "epoch": 3.305208030258947, + "flow_matching_loss": 0.0184167405590415, + "grad_norm": 0.2259497344493866, + "learning_rate": 2.1980193408087322e-05, + "loss": 0.0268, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013445660192519426, + "rkd_loss_weight": 0.6215040504932403, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3889976114034653, + "rkd_teacher_entropy": 3.76297926902771, + "rkd_teacher_similarity_mean": 0.7850002765655517, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11360 + }, + { + "epoch": 3.30811754437009, + "flow_matching_loss": 0.016733247227966784, + "grad_norm": 0.11104997992515564, + "learning_rate": 2.1965553833668224e-05, + "loss": 0.024, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01166740171611309, + "rkd_loss_weight": 0.6211707055568695, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39145558774471284, + "rkd_teacher_entropy": 3.751924729347229, + "rkd_teacher_similarity_mean": 0.7865046739578248, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11370 + }, + { + "epoch": 3.311027058481234, + "flow_matching_loss": 0.01758384667336941, + "grad_norm": 0.13903382420539856, + "learning_rate": 2.195090579544441e-05, + "loss": 0.025, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011934997607022524, + "rkd_loss_weight": 0.6208373606204987, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38721889853477476, + "rkd_teacher_entropy": 3.7355766534805297, + "rkd_teacher_similarity_mean": 0.782219922542572, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11380 + }, + { + "epoch": 3.313936572592377, + "flow_matching_loss": 0.017804537340998648, + "grad_norm": 0.20593373477458954, + "learning_rate": 2.193624931121462e-05, + "loss": 0.025, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011609384417533874, + "rkd_loss_weight": 0.6205040216445923, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3889350265264511, + "rkd_teacher_entropy": 3.755243706703186, + "rkd_teacher_similarity_mean": 0.782799357175827, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11390 + }, + { + "epoch": 3.3168460867035203, + "flow_matching_loss": 0.01729750642552972, + "grad_norm": 0.1351495385169983, + "learning_rate": 2.1921584398787872e-05, + "loss": 0.0234, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009832009766250848, + "rkd_loss_weight": 0.620170670747757, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3884804040193558, + "rkd_teacher_entropy": 3.7450664758682253, + "rkd_teacher_similarity_mean": 0.783700692653656, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11400 + }, + { + "epoch": 3.319755600814664, + "flow_matching_loss": 0.018015977833420037, + "grad_norm": 0.1547149419784546, + "learning_rate": 2.1906911075983418e-05, + "loss": 0.0255, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012101259268820285, + "rkd_loss_weight": 0.6198373258113861, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38454827964305877, + "rkd_teacher_entropy": 3.7420929670333862, + "rkd_teacher_similarity_mean": 0.7818818330764771, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11410 + }, + { + "epoch": 3.3226651149258073, + "flow_matching_loss": 0.015578328911215068, + "grad_norm": 0.1571439653635025, + "learning_rate": 2.1892229360630728e-05, + "loss": 0.022, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010423523839563131, + "rkd_loss_weight": 0.6195039868354797, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3782970517873764, + "rkd_teacher_entropy": 3.7141685485839844, + "rkd_teacher_similarity_mean": 0.7756404280662537, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11420 + }, + { + "epoch": 3.325574629036951, + "flow_matching_loss": 0.015309355873614549, + "grad_norm": 0.14669150114059448, + "learning_rate": 2.187753927056947e-05, + "loss": 0.0211, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009322745352983474, + "rkd_loss_weight": 0.6191706359386444, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38984526693820953, + "rkd_teacher_entropy": 3.7431373834609984, + "rkd_teacher_similarity_mean": 0.785960602760315, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11430 + }, + { + "epoch": 3.3284841431480943, + "flow_matching_loss": 0.01499863201752305, + "grad_norm": 0.13968373835086823, + "learning_rate": 2.186284082364949e-05, + "loss": 0.0219, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011130170617252588, + "rkd_loss_weight": 0.618837296962738, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39231039583683014, + "rkd_teacher_entropy": 3.757830333709717, + "rkd_teacher_similarity_mean": 0.7883582234382629, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11440 + }, + { + "epoch": 3.3313936572592375, + "flow_matching_loss": 0.016768183279782533, + "grad_norm": 0.11903808265924454, + "learning_rate": 2.1848134037730786e-05, + "loss": 0.0236, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01103258365765214, + "rkd_loss_weight": 0.6185039520263672, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38100006282329557, + "rkd_teacher_entropy": 3.734152007102966, + "rkd_teacher_similarity_mean": 0.776258772611618, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11450 + }, + { + "epoch": 3.3343031713703812, + "flow_matching_loss": 0.01666437666863203, + "grad_norm": 0.15806445479393005, + "learning_rate": 2.183341893068349e-05, + "loss": 0.0236, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011232402781024576, + "rkd_loss_weight": 0.6181706011295318, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3831387430429459, + "rkd_teacher_entropy": 3.7421175718307493, + "rkd_teacher_similarity_mean": 0.779647147655487, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11460 + }, + { + "epoch": 3.3372126854815245, + "flow_matching_loss": 0.015075957216322423, + "grad_norm": 0.15107066929340363, + "learning_rate": 2.1818695520387842e-05, + "loss": 0.0213, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010005532111972571, + "rkd_loss_weight": 0.61783726811409, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38703481256961825, + "rkd_teacher_entropy": 3.7362842559814453, + "rkd_teacher_similarity_mean": 0.7806992530822754, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11470 + }, + { + "epoch": 3.340122199592668, + "flow_matching_loss": 0.015604157466441393, + "grad_norm": 0.13678349554538727, + "learning_rate": 2.1803963824734173e-05, + "loss": 0.0226, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011321057472378016, + "rkd_loss_weight": 0.6175039112567902, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39250929951667785, + "rkd_teacher_entropy": 3.7545600652694704, + "rkd_teacher_similarity_mean": 0.7865493595600128, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11480 + }, + { + "epoch": 3.3430317137038115, + "flow_matching_loss": 0.016445311531424523, + "grad_norm": 0.14682410657405853, + "learning_rate": 2.1789223861622887e-05, + "loss": 0.023, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010609012003988028, + "rkd_loss_weight": 0.6171705722808838, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39037765860557555, + "rkd_teacher_entropy": 3.755816197395325, + "rkd_teacher_similarity_mean": 0.7854897677898407, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11490 + }, + { + "epoch": 3.3459412278149547, + "flow_matching_loss": 0.016773318219929934, + "grad_norm": 0.19102510809898376, + "learning_rate": 2.1774475648964428e-05, + "loss": 0.0242, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011991688655689359, + "rkd_loss_weight": 0.6168372333049774, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3869593769311905, + "rkd_teacher_entropy": 3.746196985244751, + "rkd_teacher_similarity_mean": 0.7799651622772217, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11500 + }, + { + "epoch": 3.3488507419260984, + "flow_matching_loss": 0.017290299665182828, + "grad_norm": 0.16084758937358856, + "learning_rate": 2.1759719204679267e-05, + "loss": 0.0242, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011199614219367503, + "rkd_loss_weight": 0.6165038764476776, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39484699070453644, + "rkd_teacher_entropy": 3.7517099380493164, + "rkd_teacher_similarity_mean": 0.7890933454036713, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11510 + }, + { + "epoch": 3.3517602560372417, + "flow_matching_loss": 0.019049639534205198, + "grad_norm": 0.13868114352226257, + "learning_rate": 2.174495454669787e-05, + "loss": 0.027, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012975850887596608, + "rkd_loss_weight": 0.6161705434322358, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3770891338586807, + "rkd_teacher_entropy": 3.7243773698806764, + "rkd_teacher_similarity_mean": 0.773527455329895, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11520 + }, + { + "epoch": 3.3546697701483854, + "flow_matching_loss": 0.018991153966635466, + "grad_norm": 0.11370683461427689, + "learning_rate": 2.1730181692960692e-05, + "loss": 0.0248, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009359644539654256, + "rkd_loss_weight": 0.6158371925354004, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3896838784217834, + "rkd_teacher_entropy": 3.741495656967163, + "rkd_teacher_similarity_mean": 0.7834163129329681, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11530 + }, + { + "epoch": 3.3575792842595287, + "flow_matching_loss": 0.01982291219756007, + "grad_norm": 0.4867572784423828, + "learning_rate": 2.171540066141814e-05, + "loss": 0.0268, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011283691413700581, + "rkd_loss_weight": 0.6155038475990295, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3899905413389206, + "rkd_teacher_entropy": 3.7413751602172853, + "rkd_teacher_similarity_mean": 0.7809542298316956, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11540 + }, + { + "epoch": 3.360488798370672, + "flow_matching_loss": 0.018613770697265865, + "grad_norm": 0.14463210105895996, + "learning_rate": 2.170061147003056e-05, + "loss": 0.0252, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010734691470861434, + "rkd_loss_weight": 0.6151705086231232, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.381685209274292, + "rkd_teacher_entropy": 3.72702317237854, + "rkd_teacher_similarity_mean": 0.7755578219890594, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11550 + }, + { + "epoch": 3.3633983124818156, + "flow_matching_loss": 0.017749641463160513, + "grad_norm": 0.150318443775177, + "learning_rate": 2.1685814136768225e-05, + "loss": 0.0239, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010023936815559863, + "rkd_loss_weight": 0.6148371577262879, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3928703159093857, + "rkd_teacher_entropy": 3.7548454284667967, + "rkd_teacher_similarity_mean": 0.7867996096611023, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11560 + }, + { + "epoch": 3.366307826592959, + "flow_matching_loss": 0.01633913805708289, + "grad_norm": 0.27438464760780334, + "learning_rate": 2.167100867961128e-05, + "loss": 0.0236, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011744142603129148, + "rkd_loss_weight": 0.6145038187503815, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3921250104904175, + "rkd_teacher_entropy": 3.751214337348938, + "rkd_teacher_similarity_mean": 0.7862878024578095, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11570 + }, + { + "epoch": 3.369217340704102, + "flow_matching_loss": 0.018178334180265664, + "grad_norm": 0.10678394138813019, + "learning_rate": 2.1656195116549756e-05, + "loss": 0.0243, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009974132385104894, + "rkd_loss_weight": 0.6141704738140106, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39880686104297636, + "rkd_teacher_entropy": 3.7683680295944213, + "rkd_teacher_similarity_mean": 0.7907957673072815, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11580 + }, + { + "epoch": 3.372126854815246, + "flow_matching_loss": 0.017661937978118658, + "grad_norm": 0.10618361085653305, + "learning_rate": 2.1641373465583527e-05, + "loss": 0.025, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011924225883558393, + "rkd_loss_weight": 0.6138371229171753, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38085164725780485, + "rkd_teacher_entropy": 3.7190707683563233, + "rkd_teacher_similarity_mean": 0.7738887012004853, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11590 + }, + { + "epoch": 3.375036368926389, + "flow_matching_loss": 0.01551748663187027, + "grad_norm": 0.17266884446144104, + "learning_rate": 2.1626543744722292e-05, + "loss": 0.0218, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010313973389565944, + "rkd_loss_weight": 0.6135037839412689, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38691252171993257, + "rkd_teacher_entropy": 3.747749948501587, + "rkd_teacher_similarity_mean": 0.7823089241981507, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11600 + }, + { + "epoch": 3.377945883037533, + "flow_matching_loss": 0.01921740248799324, + "grad_norm": 0.15687499940395355, + "learning_rate": 2.161170597198557e-05, + "loss": 0.0274, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013342340663075446, + "rkd_loss_weight": 0.6131704390048981, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3887634247541428, + "rkd_teacher_entropy": 3.7483399629592897, + "rkd_teacher_similarity_mean": 0.7840461015701294, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11610 + }, + { + "epoch": 3.380855397148676, + "flow_matching_loss": 0.019705722387880088, + "grad_norm": 0.14315447211265564, + "learning_rate": 2.1596860165402645e-05, + "loss": 0.0281, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013762601977214217, + "rkd_loss_weight": 0.6128370940685273, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39041311144828794, + "rkd_teacher_entropy": 3.751038980484009, + "rkd_teacher_similarity_mean": 0.783612585067749, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11620 + }, + { + "epoch": 3.38376491125982, + "flow_matching_loss": 0.020822193566709758, + "grad_norm": 0.1989155262708664, + "learning_rate": 2.1582006343012576e-05, + "loss": 0.028, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011667505744844675, + "rkd_loss_weight": 0.6125037491321563, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39581328332424165, + "rkd_teacher_entropy": 3.7686410427093504, + "rkd_teacher_similarity_mean": 0.7898139178752899, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11630 + }, + { + "epoch": 3.386674425370963, + "flow_matching_loss": 0.01660683797672391, + "grad_norm": 0.19902262091636658, + "learning_rate": 2.156714452286416e-05, + "loss": 0.023, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010367634985595942, + "rkd_loss_weight": 0.6121704041957855, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3888285726308823, + "rkd_teacher_entropy": 3.7487783908843992, + "rkd_teacher_similarity_mean": 0.7829155504703522, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11640 + }, + { + "epoch": 3.3895839394821063, + "flow_matching_loss": 0.017037076223641635, + "grad_norm": 0.12586039304733276, + "learning_rate": 2.155227472301591e-05, + "loss": 0.0227, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009255506983026861, + "rkd_loss_weight": 0.6118370592594147, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38883295059204104, + "rkd_teacher_entropy": 3.7459510564804077, + "rkd_teacher_similarity_mean": 0.7845112383365631, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11650 + }, + { + "epoch": 3.39249345359325, + "flow_matching_loss": 0.014907645247876643, + "grad_norm": 0.10612862557172775, + "learning_rate": 2.1537396961536027e-05, + "loss": 0.0216, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010917597636580467, + "rkd_loss_weight": 0.6115037202835083, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38948037326335905, + "rkd_teacher_entropy": 3.7440961360931397, + "rkd_teacher_similarity_mean": 0.7854515671730041, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11660 + }, + { + "epoch": 3.3954029677043933, + "flow_matching_loss": 0.01626794533804059, + "grad_norm": 0.14748068153858185, + "learning_rate": 2.1522511256502403e-05, + "loss": 0.0235, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011757027823477983, + "rkd_loss_weight": 0.611170369386673, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3853019803762436, + "rkd_teacher_entropy": 3.7387855529785154, + "rkd_teacher_similarity_mean": 0.7795323491096496, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11670 + }, + { + "epoch": 3.3983124818155366, + "flow_matching_loss": 0.015964841190725564, + "grad_norm": 0.14940553903579712, + "learning_rate": 2.1507617626002566e-05, + "loss": 0.0227, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011059872433543205, + "rkd_loss_weight": 0.6108370244503021, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3918877631425858, + "rkd_teacher_entropy": 3.7583644866943358, + "rkd_teacher_similarity_mean": 0.7865488946437835, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11680 + }, + { + "epoch": 3.4012219959266803, + "flow_matching_loss": 0.016830604430288075, + "grad_norm": 0.16698743402957916, + "learning_rate": 2.1492716088133685e-05, + "loss": 0.0236, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011079202685505152, + "rkd_loss_weight": 0.6105036854743957, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3836578667163849, + "rkd_teacher_entropy": 3.7334917068481444, + "rkd_teacher_similarity_mean": 0.7782432496547699, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11690 + }, + { + "epoch": 3.4041315100378235, + "flow_matching_loss": 0.01846001036465168, + "grad_norm": 0.17624881863594055, + "learning_rate": 2.1477806661002536e-05, + "loss": 0.0248, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010383669473230838, + "rkd_loss_weight": 0.6101703345775604, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39234966933727267, + "rkd_teacher_entropy": 3.7724159479141237, + "rkd_teacher_similarity_mean": 0.7884582161903382, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11700 + }, + { + "epoch": 3.4070410241489673, + "flow_matching_loss": 0.016376413963735105, + "grad_norm": 0.16449151933193207, + "learning_rate": 2.146288936272548e-05, + "loss": 0.0223, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009662346681579947, + "rkd_loss_weight": 0.609836995601654, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3915747344493866, + "rkd_teacher_entropy": 3.7563950777053834, + "rkd_teacher_similarity_mean": 0.7862485885620117, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11710 + }, + { + "epoch": 3.4099505382601105, + "flow_matching_loss": 0.015433330554515123, + "grad_norm": 0.15962202847003937, + "learning_rate": 2.1447964211428433e-05, + "loss": 0.0221, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0109846162609756, + "rkd_loss_weight": 0.6095036566257477, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39682166576385497, + "rkd_teacher_entropy": 3.774093508720398, + "rkd_teacher_similarity_mean": 0.792992115020752, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11720 + }, + { + "epoch": 3.4128600523712542, + "flow_matching_loss": 0.017830533254891635, + "grad_norm": 0.15140998363494873, + "learning_rate": 2.1433031225246865e-05, + "loss": 0.0246, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011056317202746867, + "rkd_loss_weight": 0.6091702997684478, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3845615655183792, + "rkd_teacher_entropy": 3.735339641571045, + "rkd_teacher_similarity_mean": 0.780362355709076, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11730 + }, + { + "epoch": 3.4157695664823975, + "flow_matching_loss": 0.017269314266741276, + "grad_norm": 0.1496421992778778, + "learning_rate": 2.1418090422325766e-05, + "loss": 0.0234, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010094396024942397, + "rkd_loss_weight": 0.608836966753006, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3875363081693649, + "rkd_teacher_entropy": 3.7459378004074098, + "rkd_teacher_similarity_mean": 0.7822490453720092, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11740 + }, + { + "epoch": 3.4186790805935408, + "flow_matching_loss": 0.018885674700140954, + "grad_norm": 0.12815068662166595, + "learning_rate": 2.1403141820819623e-05, + "loss": 0.0254, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010660990560427307, + "rkd_loss_weight": 0.6085036158561706, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3852160006761551, + "rkd_teacher_entropy": 3.734504556655884, + "rkd_teacher_similarity_mean": 0.7789099514484406, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11750 + }, + { + "epoch": 3.4215885947046845, + "flow_matching_loss": 0.017782004456967114, + "grad_norm": 0.15275578200817108, + "learning_rate": 2.1388185438892377e-05, + "loss": 0.026, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013435437856242061, + "rkd_loss_weight": 0.6081702709197998, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.384446319937706, + "rkd_teacher_entropy": 3.73158061504364, + "rkd_teacher_similarity_mean": 0.7794990301132202, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11760 + }, + { + "epoch": 3.4244981088158277, + "flow_matching_loss": 0.017528201546519996, + "grad_norm": 0.1354130655527115, + "learning_rate": 2.1373221294717465e-05, + "loss": 0.0242, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011021713586524129, + "rkd_loss_weight": 0.6078369319438934, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3813409149646759, + "rkd_teacher_entropy": 3.72776882648468, + "rkd_teacher_similarity_mean": 0.7757251679897308, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11770 + }, + { + "epoch": 3.427407622926971, + "flow_matching_loss": 0.014121672045439482, + "grad_norm": 0.11059525609016418, + "learning_rate": 2.1358249406477724e-05, + "loss": 0.0202, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010001720115542411, + "rkd_loss_weight": 0.6075035810470581, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3780858814716339, + "rkd_teacher_entropy": 3.711706209182739, + "rkd_teacher_similarity_mean": 0.773498922586441, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11780 + }, + { + "epoch": 3.4303171370381147, + "flow_matching_loss": 0.017450144793838263, + "grad_norm": 0.20083147287368774, + "learning_rate": 2.1343269792365408e-05, + "loss": 0.0242, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011112158559262753, + "rkd_loss_weight": 0.6071702420711518, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38208668529987333, + "rkd_teacher_entropy": 3.734236240386963, + "rkd_teacher_similarity_mean": 0.7778810679912567, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11790 + }, + { + "epoch": 3.433226651149258, + "flow_matching_loss": 0.01730268644168973, + "grad_norm": 0.13813354074954987, + "learning_rate": 2.132828247058217e-05, + "loss": 0.0236, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010366832651197911, + "rkd_loss_weight": 0.6068368971347808, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38778439462184905, + "rkd_teacher_entropy": 3.7456477642059327, + "rkd_teacher_similarity_mean": 0.7822954595088959, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11800 + }, + { + "epoch": 3.4361361652604017, + "flow_matching_loss": 0.017788456566631795, + "grad_norm": 0.16887988150119781, + "learning_rate": 2.1313287459339004e-05, + "loss": 0.0242, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010600805282592773, + "rkd_loss_weight": 0.6065035462379456, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39077328741550443, + "rkd_teacher_entropy": 3.759382438659668, + "rkd_teacher_similarity_mean": 0.7844390749931336, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11810 + }, + { + "epoch": 3.439045679371545, + "flow_matching_loss": 0.01792929572984576, + "grad_norm": 0.1744762361049652, + "learning_rate": 2.1298284776856276e-05, + "loss": 0.0253, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012099288171157241, + "rkd_loss_weight": 0.6061702072620392, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.380137225985527, + "rkd_teacher_entropy": 3.7149277687072755, + "rkd_teacher_similarity_mean": 0.7723671436309815, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11820 + }, + { + "epoch": 3.4419551934826886, + "flow_matching_loss": 0.018352210894227027, + "grad_norm": 0.13449309766292572, + "learning_rate": 2.128327444136365e-05, + "loss": 0.0259, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012485641706734895, + "rkd_loss_weight": 0.6058368623256684, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38169653713703156, + "rkd_teacher_entropy": 3.7281877994537354, + "rkd_teacher_similarity_mean": 0.7742230057716369, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11830 + }, + { + "epoch": 3.444864707593832, + "flow_matching_loss": 0.018361797835677863, + "grad_norm": 0.1675049066543579, + "learning_rate": 2.1268256471100105e-05, + "loss": 0.0247, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01040701288729906, + "rkd_loss_weight": 0.6055035173892975, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3855190485715866, + "rkd_teacher_entropy": 3.7438304901123045, + "rkd_teacher_similarity_mean": 0.7819435834884644, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11840 + }, + { + "epoch": 3.447774221704975, + "flow_matching_loss": 0.01777125084772706, + "grad_norm": 0.1832544356584549, + "learning_rate": 2.1253230884313893e-05, + "loss": 0.0256, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012970731779932975, + "rkd_loss_weight": 0.6051701724529266, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3764992743730545, + "rkd_teacher_entropy": 3.7258355379104615, + "rkd_teacher_similarity_mean": 0.7707186102867126, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11850 + }, + { + "epoch": 3.450683735816119, + "flow_matching_loss": 0.01579666044563055, + "grad_norm": 0.31348010897636414, + "learning_rate": 2.123819769926251e-05, + "loss": 0.0232, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012210222892463207, + "rkd_loss_weight": 0.6048368275165558, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3930436849594116, + "rkd_teacher_entropy": 3.7626133441925047, + "rkd_teacher_similarity_mean": 0.7868388414382934, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11860 + }, + { + "epoch": 3.453593249927262, + "flow_matching_loss": 0.0182443636469543, + "grad_norm": 0.13167937099933624, + "learning_rate": 2.1223156934212702e-05, + "loss": 0.0253, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011746285995468498, + "rkd_loss_weight": 0.6045034825801849, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3833127677440643, + "rkd_teacher_entropy": 3.719094467163086, + "rkd_teacher_similarity_mean": 0.7782960653305053, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11870 + }, + { + "epoch": 3.4565027640384054, + "flow_matching_loss": 0.01815000930801034, + "grad_norm": 0.15573792159557343, + "learning_rate": 2.1208108607440418e-05, + "loss": 0.0248, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010943195037543773, + "rkd_loss_weight": 0.6041701436042786, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38637098073959353, + "rkd_teacher_entropy": 3.7414710998535154, + "rkd_teacher_similarity_mean": 0.7813052594661712, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11880 + }, + { + "epoch": 3.459412278149549, + "flow_matching_loss": 0.016657885815948247, + "grad_norm": 0.170880988240242, + "learning_rate": 2.119305273723078e-05, + "loss": 0.024, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012208596710115672, + "rkd_loss_weight": 0.6038367927074433, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3727321267127991, + "rkd_teacher_entropy": 3.7164146900177, + "rkd_teacher_similarity_mean": 0.7696920692920685, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11890 + }, + { + "epoch": 3.4623217922606924, + "flow_matching_loss": 0.016776107996702195, + "grad_norm": 0.15030397474765778, + "learning_rate": 2.1177989341878106e-05, + "loss": 0.0246, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013038459233939648, + "rkd_loss_weight": 0.6035034477710723, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38172935843467715, + "rkd_teacher_entropy": 3.7242783069610597, + "rkd_teacher_similarity_mean": 0.7772550463676453, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11900 + }, + { + "epoch": 3.465231306371836, + "flow_matching_loss": 0.017275407537817956, + "grad_norm": 0.14187349379062653, + "learning_rate": 2.116291843968583e-05, + "loss": 0.0241, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011279290728271007, + "rkd_loss_weight": 0.603170108795166, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.386843740940094, + "rkd_teacher_entropy": 3.7523247957229615, + "rkd_teacher_similarity_mean": 0.7846391201019287, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11910 + }, + { + "epoch": 3.4681408204829793, + "flow_matching_loss": 0.01716323336586356, + "grad_norm": 0.18470121920108795, + "learning_rate": 2.114784004896653e-05, + "loss": 0.0242, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011662652809172868, + "rkd_loss_weight": 0.6028367578983307, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3844642013311386, + "rkd_teacher_entropy": 3.7443861246109007, + "rkd_teacher_similarity_mean": 0.7808931827545166, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11920 + }, + { + "epoch": 3.4710503345941226, + "flow_matching_loss": 0.020559743326157332, + "grad_norm": 0.16960495710372925, + "learning_rate": 2.113275418804187e-05, + "loss": 0.0267, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01018848717212677, + "rkd_loss_weight": 0.6025034189224243, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38679059147834777, + "rkd_teacher_entropy": 3.742726683616638, + "rkd_teacher_similarity_mean": 0.7830457806587219, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11930 + }, + { + "epoch": 3.4739598487052663, + "flow_matching_loss": 0.021241496317088604, + "grad_norm": 0.19560249149799347, + "learning_rate": 2.1117660875242583e-05, + "loss": 0.0279, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011033994331955909, + "rkd_loss_weight": 0.6021700739860535, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3855499893426895, + "rkd_teacher_entropy": 3.7511736392974853, + "rkd_teacher_similarity_mean": 0.7823639929294586, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11940 + }, + { + "epoch": 3.4768693628164096, + "flow_matching_loss": 0.016205182392150165, + "grad_norm": 0.14366857707500458, + "learning_rate": 2.1102560128908478e-05, + "loss": 0.0229, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011152021121233702, + "rkd_loss_weight": 0.6018367230892181, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3809219181537628, + "rkd_teacher_entropy": 3.735102343559265, + "rkd_teacher_similarity_mean": 0.7794345319271088, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11950 + }, + { + "epoch": 3.4797788769275533, + "flow_matching_loss": 0.019637547712773085, + "grad_norm": 0.15729756653308868, + "learning_rate": 2.1087451967388384e-05, + "loss": 0.026, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010572240175679325, + "rkd_loss_weight": 0.6015033900737763, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3838327020406723, + "rkd_teacher_entropy": 3.7436389923095703, + "rkd_teacher_similarity_mean": 0.779504370689392, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11960 + }, + { + "epoch": 3.4826883910386965, + "flow_matching_loss": 0.016883899364620446, + "grad_norm": 0.12921582162380219, + "learning_rate": 2.1072336409040135e-05, + "loss": 0.0234, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010762709425762296, + "rkd_loss_weight": 0.6011700332164764, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3952589750289917, + "rkd_teacher_entropy": 3.769771933555603, + "rkd_teacher_similarity_mean": 0.7905411720275879, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11970 + }, + { + "epoch": 3.48559790514984, + "flow_matching_loss": 0.016588738281279804, + "grad_norm": 0.15015296638011932, + "learning_rate": 2.1057213472230573e-05, + "loss": 0.0229, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010508549446240068, + "rkd_loss_weight": 0.6008366942405701, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38483180701732633, + "rkd_teacher_entropy": 3.742851424217224, + "rkd_teacher_similarity_mean": 0.7800000727176666, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11980 + }, + { + "epoch": 3.4885074192609835, + "flow_matching_loss": 0.01565094217658043, + "grad_norm": 0.176306813955307, + "learning_rate": 2.1042083175335478e-05, + "loss": 0.0218, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010265918541699648, + "rkd_loss_weight": 0.6005033552646637, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3928330600261688, + "rkd_teacher_entropy": 3.7611523866653442, + "rkd_teacher_similarity_mean": 0.7886550843715667, + "rkd_vlm_temp": 0.10000000149011612, + "step": 11990 + }, + { + "epoch": 3.4914169333721268, + "flow_matching_loss": 0.01833558464422822, + "grad_norm": 0.14668481051921844, + "learning_rate": 2.1026945536739595e-05, + "loss": 0.0253, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011575884977355599, + "rkd_loss_weight": 0.6001699984073638, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39545040726661684, + "rkd_teacher_entropy": 3.761690783500671, + "rkd_teacher_similarity_mean": 0.7912209928035736, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12000 + }, + { + "epoch": 3.4943264474832705, + "flow_matching_loss": 0.018749997951090336, + "grad_norm": 0.15482573211193085, + "learning_rate": 2.1011800574836586e-05, + "loss": 0.0248, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010066585056483746, + "rkd_loss_weight": 0.599836665391922, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3836606562137604, + "rkd_teacher_entropy": 3.741389584541321, + "rkd_teacher_similarity_mean": 0.7794898092746735, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12010 + }, + { + "epoch": 3.4972359615944137, + "flow_matching_loss": 0.01533700991421938, + "grad_norm": 0.16298387944698334, + "learning_rate": 2.0996648308028992e-05, + "loss": 0.0217, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010547023639082909, + "rkd_loss_weight": 0.5995033144950866, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38752131164073944, + "rkd_teacher_entropy": 3.741019082069397, + "rkd_teacher_similarity_mean": 0.7827386319637298, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12020 + }, + { + "epoch": 3.500145475705557, + "flow_matching_loss": 0.016410434525460003, + "grad_norm": 0.16782139241695404, + "learning_rate": 2.0981488754728266e-05, + "loss": 0.0233, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011520369024947286, + "rkd_loss_weight": 0.5991699695587158, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.381339356303215, + "rkd_teacher_entropy": 3.7315431594848634, + "rkd_teacher_similarity_mean": 0.7778219401836395, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12030 + }, + { + "epoch": 3.5030549898167007, + "flow_matching_loss": 0.016362849902361632, + "grad_norm": 0.1348849982023239, + "learning_rate": 2.0966321933354685e-05, + "loss": 0.0226, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010361134633421898, + "rkd_loss_weight": 0.5988366305828094, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3881185293197632, + "rkd_teacher_entropy": 3.75207839012146, + "rkd_teacher_similarity_mean": 0.7853310585021973, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12040 + }, + { + "epoch": 3.505964503927844, + "flow_matching_loss": 0.016868248116225004, + "grad_norm": 0.13941574096679688, + "learning_rate": 2.095114786233736e-05, + "loss": 0.0223, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009154630592092872, + "rkd_loss_weight": 0.5985032796859742, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39356123507022855, + "rkd_teacher_entropy": 3.7643898010253904, + "rkd_teacher_similarity_mean": 0.7899097740650177, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12050 + }, + { + "epoch": 3.5088740180389877, + "flow_matching_loss": 0.01801359197124839, + "grad_norm": 0.227393239736557, + "learning_rate": 2.0935966560114236e-05, + "loss": 0.0249, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011482240771874786, + "rkd_loss_weight": 0.5981699407100678, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3905412435531616, + "rkd_teacher_entropy": 3.752939462661743, + "rkd_teacher_similarity_mean": 0.7860037207603454, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12060 + }, + { + "epoch": 3.511783532150131, + "flow_matching_loss": 0.015391553658992052, + "grad_norm": 0.1407332867383957, + "learning_rate": 2.092077804513201e-05, + "loss": 0.0218, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010731169208884239, + "rkd_loss_weight": 0.5978365957736969, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3803348749876022, + "rkd_teacher_entropy": 3.7313684463500976, + "rkd_teacher_similarity_mean": 0.7750015079975128, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12070 + }, + { + "epoch": 3.514693046261274, + "flow_matching_loss": 0.01634513381868601, + "grad_norm": 0.16911476850509644, + "learning_rate": 2.0905582335846167e-05, + "loss": 0.0226, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01050488161854446, + "rkd_loss_weight": 0.5975032448768616, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3876091629266739, + "rkd_teacher_entropy": 3.7463369607925414, + "rkd_teacher_similarity_mean": 0.7813027560710907, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12080 + }, + { + "epoch": 3.517602560372418, + "flow_matching_loss": 0.01608001533895731, + "grad_norm": 0.16198945045471191, + "learning_rate": 2.0890379450720927e-05, + "loss": 0.0222, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010252418089658022, + "rkd_loss_weight": 0.5971699059009552, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3868269056081772, + "rkd_teacher_entropy": 3.749167776107788, + "rkd_teacher_similarity_mean": 0.781314903497696, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12090 + }, + { + "epoch": 3.520512074483561, + "flow_matching_loss": 0.01842402359470725, + "grad_norm": 0.14188401401042938, + "learning_rate": 2.087516940822923e-05, + "loss": 0.0246, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01027931710705161, + "rkd_loss_weight": 0.5968365609645844, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38348461389541627, + "rkd_teacher_entropy": 3.742483639717102, + "rkd_teacher_similarity_mean": 0.7768319487571717, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12100 + }, + { + "epoch": 3.5234215885947044, + "flow_matching_loss": 0.016166370920836927, + "grad_norm": 0.13776646554470062, + "learning_rate": 2.085995222685271e-05, + "loss": 0.0233, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011885342188179493, + "rkd_loss_weight": 0.5965032160282135, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38750178515911105, + "rkd_teacher_entropy": 3.750076651573181, + "rkd_teacher_similarity_mean": 0.7814770936965942, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12110 + }, + { + "epoch": 3.526331102705848, + "flow_matching_loss": 0.016501712333410978, + "grad_norm": 0.16504158079624176, + "learning_rate": 2.084472792508167e-05, + "loss": 0.0251, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014363988861441612, + "rkd_loss_weight": 0.5961698710918426, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3963334858417511, + "rkd_teacher_entropy": 3.7688020706176757, + "rkd_teacher_similarity_mean": 0.7920501410961152, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12120 + }, + { + "epoch": 3.5292406168169914, + "flow_matching_loss": 0.01783650815486908, + "grad_norm": 0.20613279938697815, + "learning_rate": 2.082949652141508e-05, + "loss": 0.0251, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01222078651189804, + "rkd_loss_weight": 0.5958365261554718, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38871740400791166, + "rkd_teacher_entropy": 3.7446685552597048, + "rkd_teacher_similarity_mean": 0.7808901011943817, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12130 + }, + { + "epoch": 3.532150130928135, + "flow_matching_loss": 0.016623678803443908, + "grad_norm": 0.1357162594795227, + "learning_rate": 2.081425803436054e-05, + "loss": 0.0241, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012608036072924734, + "rkd_loss_weight": 0.595503181219101, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38821530938148496, + "rkd_teacher_entropy": 3.7416056871414183, + "rkd_teacher_similarity_mean": 0.7822644710540771, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12140 + }, + { + "epoch": 3.5350596450392784, + "flow_matching_loss": 0.015438038669526578, + "grad_norm": 0.1318046897649765, + "learning_rate": 2.079901248243423e-05, + "loss": 0.0213, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009775581955909728, + "rkd_loss_weight": 0.5951698422431946, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.391833558678627, + "rkd_teacher_entropy": 3.7557226181030274, + "rkd_teacher_similarity_mean": 0.7862212121486664, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12150 + }, + { + "epoch": 3.537969159150422, + "flow_matching_loss": 0.016883029229938985, + "grad_norm": 0.13903245329856873, + "learning_rate": 2.078375988416095e-05, + "loss": 0.0226, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009669003495946526, + "rkd_loss_weight": 0.5948364973068238, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37840896248817446, + "rkd_teacher_entropy": 3.7102973461151123, + "rkd_teacher_similarity_mean": 0.7734075546264648, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12160 + }, + { + "epoch": 3.5408786732615654, + "flow_matching_loss": 0.01610993891954422, + "grad_norm": 0.1576744019985199, + "learning_rate": 2.076850025807404e-05, + "loss": 0.0225, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01079076835885644, + "rkd_loss_weight": 0.5945031464099884, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3938495099544525, + "rkd_teacher_entropy": 3.7578933000564576, + "rkd_teacher_similarity_mean": 0.7859066545963287, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12170 + }, + { + "epoch": 3.5437881873727086, + "flow_matching_loss": 0.01894139386713505, + "grad_norm": 0.19414861500263214, + "learning_rate": 2.0753233622715374e-05, + "loss": 0.0255, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010960429906845093, + "rkd_loss_weight": 0.5941698133945466, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3851477265357971, + "rkd_teacher_entropy": 3.730500340461731, + "rkd_teacher_similarity_mean": 0.7784151911735535, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12180 + }, + { + "epoch": 3.5466977014838523, + "flow_matching_loss": 0.015961666684597732, + "grad_norm": 0.1770491600036621, + "learning_rate": 2.073795999663536e-05, + "loss": 0.0243, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.014024365041404963, + "rkd_loss_weight": 0.5938364565372467, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38913391828536986, + "rkd_teacher_entropy": 3.767058849334717, + "rkd_teacher_similarity_mean": 0.7841499447822571, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12190 + }, + { + "epoch": 3.5496072155949956, + "flow_matching_loss": 0.017923811357468365, + "grad_norm": 0.23654958605766296, + "learning_rate": 2.07226793983929e-05, + "loss": 0.0252, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01220523240044713, + "rkd_loss_weight": 0.5935031175613403, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38742064535617826, + "rkd_teacher_entropy": 3.7420053243637086, + "rkd_teacher_similarity_mean": 0.7812840461730957, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12200 + }, + { + "epoch": 3.552516729706139, + "flow_matching_loss": 0.02027846034616232, + "grad_norm": 0.17929047346115112, + "learning_rate": 2.070739184655536e-05, + "loss": 0.0282, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013315849471837282, + "rkd_loss_weight": 0.593169778585434, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3904606133699417, + "rkd_teacher_entropy": 3.745377230644226, + "rkd_teacher_similarity_mean": 0.7823914289474487, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12210 + }, + { + "epoch": 3.5554262438172826, + "flow_matching_loss": 0.014582331385463476, + "grad_norm": 0.16344785690307617, + "learning_rate": 2.069209735969855e-05, + "loss": 0.0206, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01014881138689816, + "rkd_loss_weight": 0.5928364217281341, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3879755973815918, + "rkd_teacher_entropy": 3.7343230485916137, + "rkd_teacher_similarity_mean": 0.7790598273277283, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12220 + }, + { + "epoch": 3.558335757928426, + "flow_matching_loss": 0.01670935656875372, + "grad_norm": 0.1190183088183403, + "learning_rate": 2.067679595640672e-05, + "loss": 0.0223, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00935559025965631, + "rkd_loss_weight": 0.5925030887126923, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3872331976890564, + "rkd_teacher_entropy": 3.74196343421936, + "rkd_teacher_similarity_mean": 0.7824367225170136, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12230 + }, + { + "epoch": 3.5612452720395695, + "flow_matching_loss": 0.016870092321187257, + "grad_norm": 0.158208966255188, + "learning_rate": 2.066148765527252e-05, + "loss": 0.0233, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010937013477087022, + "rkd_loss_weight": 0.5921697378158569, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3915320932865143, + "rkd_teacher_entropy": 3.7505359649658203, + "rkd_teacher_similarity_mean": 0.7857295095920562, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12240 + }, + { + "epoch": 3.564154786150713, + "flow_matching_loss": 0.014597517251968384, + "grad_norm": 0.12323864549398422, + "learning_rate": 2.0646172474896976e-05, + "loss": 0.0207, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010297041246667504, + "rkd_loss_weight": 0.5918363928794861, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.386125311255455, + "rkd_teacher_entropy": 3.725374150276184, + "rkd_teacher_similarity_mean": 0.7796672523021698, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12250 + }, + { + "epoch": 3.5670643002618565, + "flow_matching_loss": 0.017199706006795167, + "grad_norm": 0.15904533863067627, + "learning_rate": 2.063085043388947e-05, + "loss": 0.024, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011416676361113787, + "rkd_loss_weight": 0.5915030539035797, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3910806059837341, + "rkd_teacher_entropy": 3.747299408912659, + "rkd_teacher_similarity_mean": 0.7852835297584534, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12260 + }, + { + "epoch": 3.5699738143729998, + "flow_matching_loss": 0.01856490345671773, + "grad_norm": 0.11274620145559311, + "learning_rate": 2.0615521550867732e-05, + "loss": 0.0259, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012413277430459857, + "rkd_loss_weight": 0.5911697030067444, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3852745622396469, + "rkd_teacher_entropy": 3.74106764793396, + "rkd_teacher_similarity_mean": 0.7802099466323853, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12270 + }, + { + "epoch": 3.572883328484143, + "flow_matching_loss": 0.014721281174570323, + "grad_norm": 0.14427019655704498, + "learning_rate": 2.0600185844457803e-05, + "loss": 0.0199, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008846459630876779, + "rkd_loss_weight": 0.5908363640308381, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3845567137002945, + "rkd_teacher_entropy": 3.742778944969177, + "rkd_teacher_similarity_mean": 0.7806984126567841, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12280 + }, + { + "epoch": 3.5757928425952867, + "flow_matching_loss": 0.013742051925510168, + "grad_norm": 0.15554161369800568, + "learning_rate": 2.058484333329401e-05, + "loss": 0.0197, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010119642224162817, + "rkd_loss_weight": 0.5905030190944671, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38970872163772585, + "rkd_teacher_entropy": 3.748994469642639, + "rkd_teacher_similarity_mean": 0.7833202242851257, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12290 + }, + { + "epoch": 3.57870235670643, + "flow_matching_loss": 0.01917766761034727, + "grad_norm": 0.21082186698913574, + "learning_rate": 2.0569494036018955e-05, + "loss": 0.0248, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009582884190604091, + "rkd_loss_weight": 0.5901696681976318, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3834479123353958, + "rkd_teacher_entropy": 3.73054039478302, + "rkd_teacher_similarity_mean": 0.7780939996242523, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12300 + }, + { + "epoch": 3.5816118708175733, + "flow_matching_loss": 0.018010766804218294, + "grad_norm": 0.16777794063091278, + "learning_rate": 2.055413797128347e-05, + "loss": 0.0247, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011306673428043723, + "rkd_loss_weight": 0.5898363292217255, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3834805816411972, + "rkd_teacher_entropy": 3.737807846069336, + "rkd_teacher_similarity_mean": 0.778156852722168, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12310 + }, + { + "epoch": 3.584521384928717, + "flow_matching_loss": 0.016410776413977146, + "grad_norm": 0.1677381992340088, + "learning_rate": 2.0538775157746634e-05, + "loss": 0.0228, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010918162018060684, + "rkd_loss_weight": 0.5895029842853546, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3835473835468292, + "rkd_teacher_entropy": 3.7275697946548463, + "rkd_teacher_similarity_mean": 0.7777031481266021, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12320 + }, + { + "epoch": 3.5874308990398602, + "flow_matching_loss": 0.015149759221822023, + "grad_norm": 0.19100813567638397, + "learning_rate": 2.0523405614075706e-05, + "loss": 0.0224, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01227299878373742, + "rkd_loss_weight": 0.5891696393489838, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38886958360671997, + "rkd_teacher_entropy": 3.739589881896973, + "rkd_teacher_similarity_mean": 0.7834153771400452, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12330 + }, + { + "epoch": 3.590340413151004, + "flow_matching_loss": 0.016637739539146424, + "grad_norm": 0.15330258011817932, + "learning_rate": 2.0508029358946132e-05, + "loss": 0.0231, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011014976212754846, + "rkd_loss_weight": 0.5888362944126129, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38672326803207396, + "rkd_teacher_entropy": 3.7381970167160032, + "rkd_teacher_similarity_mean": 0.7810907602310181, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12340 + }, + { + "epoch": 3.593249927262147, + "flow_matching_loss": 0.017719034105539322, + "grad_norm": 0.12256602942943573, + "learning_rate": 2.0492646411041515e-05, + "loss": 0.0237, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010095057543367147, + "rkd_loss_weight": 0.588502949476242, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.389725935459137, + "rkd_teacher_entropy": 3.748305320739746, + "rkd_teacher_similarity_mean": 0.7845160722732544, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12350 + }, + { + "epoch": 3.596159441373291, + "flow_matching_loss": 0.016467771120369436, + "grad_norm": 0.1358654946088791, + "learning_rate": 2.0477256789053587e-05, + "loss": 0.0228, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01071688560768962, + "rkd_loss_weight": 0.5881696045398712, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3789651572704315, + "rkd_teacher_entropy": 3.7200151920318603, + "rkd_teacher_similarity_mean": 0.772348964214325, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12360 + }, + { + "epoch": 3.599068955484434, + "flow_matching_loss": 0.015191418305039405, + "grad_norm": 0.11362849175930023, + "learning_rate": 2.0461860511682194e-05, + "loss": 0.0221, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011736913910135626, + "rkd_loss_weight": 0.5878362655639648, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3917147159576416, + "rkd_teacher_entropy": 3.7593649864196776, + "rkd_teacher_similarity_mean": 0.7864554643630981, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12370 + }, + { + "epoch": 3.6019784695955774, + "flow_matching_loss": 0.014710365794599056, + "grad_norm": 0.11852037161588669, + "learning_rate": 2.0446457597635254e-05, + "loss": 0.0203, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009551047161221505, + "rkd_loss_weight": 0.5875029146671296, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3910679280757904, + "rkd_teacher_entropy": 3.7527389764785766, + "rkd_teacher_similarity_mean": 0.7845878064632416, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12380 + }, + { + "epoch": 3.604887983706721, + "flow_matching_loss": 0.01584760593250394, + "grad_norm": 0.1500435322523117, + "learning_rate": 2.0431048065628763e-05, + "loss": 0.0219, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010235288040712475, + "rkd_loss_weight": 0.5871695697307586, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3873632848262787, + "rkd_teacher_entropy": 3.744417190551758, + "rkd_teacher_similarity_mean": 0.7818353354930878, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12390 + }, + { + "epoch": 3.6077974978178644, + "flow_matching_loss": 0.016097019612789153, + "grad_norm": 0.1998249590396881, + "learning_rate": 2.0415631934386764e-05, + "loss": 0.0225, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010900269728153944, + "rkd_loss_weight": 0.5868362307548523, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38895107209682467, + "rkd_teacher_entropy": 3.752540183067322, + "rkd_teacher_similarity_mean": 0.7853648424148559, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12400 + }, + { + "epoch": 3.6107070119290077, + "flow_matching_loss": 0.014091319777071476, + "grad_norm": 0.18957915902137756, + "learning_rate": 2.04002092226413e-05, + "loss": 0.0202, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010398129886016249, + "rkd_loss_weight": 0.586502879858017, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3861166089773178, + "rkd_teacher_entropy": 3.721928834915161, + "rkd_teacher_similarity_mean": 0.7814932405948639, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12410 + }, + { + "epoch": 3.6136165260401514, + "flow_matching_loss": 0.016118127014487983, + "grad_norm": 0.1364223212003708, + "learning_rate": 2.038477994913242e-05, + "loss": 0.0225, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010824891924858093, + "rkd_loss_weight": 0.5861695408821106, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3877732217311859, + "rkd_teacher_entropy": 3.748839831352234, + "rkd_teacher_similarity_mean": 0.7833598136901856, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12420 + }, + { + "epoch": 3.6165260401512946, + "flow_matching_loss": 0.018498164787888526, + "grad_norm": 0.21142055094242096, + "learning_rate": 2.036934413260816e-05, + "loss": 0.0256, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012118501029908657, + "rkd_loss_weight": 0.5858361959457398, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3817235380411148, + "rkd_teacher_entropy": 3.727872347831726, + "rkd_teacher_similarity_mean": 0.7749469935894012, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12430 + }, + { + "epoch": 3.619435554262438, + "flow_matching_loss": 0.014211959391832351, + "grad_norm": 0.15939666330814362, + "learning_rate": 2.0353901791824473e-05, + "loss": 0.0197, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009436976304277778, + "rkd_loss_weight": 0.5855028450489044, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39678657352924346, + "rkd_teacher_entropy": 3.7693343639373778, + "rkd_teacher_similarity_mean": 0.7912981450557709, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12440 + }, + { + "epoch": 3.6223450683735816, + "flow_matching_loss": 0.015900040883570908, + "grad_norm": 0.17728354036808014, + "learning_rate": 2.0338452945545275e-05, + "loss": 0.0222, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010788806015625596, + "rkd_loss_weight": 0.5851695120334626, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3938012093305588, + "rkd_teacher_entropy": 3.7710463047027587, + "rkd_teacher_similarity_mean": 0.78815456032753, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12450 + }, + { + "epoch": 3.6252545824847253, + "flow_matching_loss": 0.013990860804915429, + "grad_norm": 0.15029188990592957, + "learning_rate": 2.0322997612542366e-05, + "loss": 0.0196, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00955571122467518, + "rkd_loss_weight": 0.5848361551761627, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3891842871904373, + "rkd_teacher_entropy": 3.7374592542648317, + "rkd_teacher_similarity_mean": 0.7813698828220368, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12460 + }, + { + "epoch": 3.6281640965958686, + "flow_matching_loss": 0.017467417195439337, + "grad_norm": 0.14302268624305725, + "learning_rate": 2.0307535811595427e-05, + "loss": 0.0237, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010691831074655055, + "rkd_loss_weight": 0.5845028162002563, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3949605464935303, + "rkd_teacher_entropy": 3.7563228368759156, + "rkd_teacher_similarity_mean": 0.7880407512187958, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12470 + }, + { + "epoch": 3.631073610707012, + "flow_matching_loss": 0.015230520628392696, + "grad_norm": 0.1318579763174057, + "learning_rate": 2.0292067561492013e-05, + "loss": 0.0213, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010351557051762938, + "rkd_loss_weight": 0.58416947722435, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3858999162912369, + "rkd_teacher_entropy": 3.7421982526779174, + "rkd_teacher_similarity_mean": 0.7803709328174591, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12480 + }, + { + "epoch": 3.6339831248181556, + "flow_matching_loss": 0.016652677208185196, + "grad_norm": 0.10457063466310501, + "learning_rate": 2.02765928810275e-05, + "loss": 0.0222, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009464746154844762, + "rkd_loss_weight": 0.5838361203670501, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3876893728971481, + "rkd_teacher_entropy": 3.739657497406006, + "rkd_teacher_similarity_mean": 0.7810539782047272, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12490 + }, + { + "epoch": 3.636892638929299, + "flow_matching_loss": 0.015104650799185038, + "grad_norm": 0.18164502084255219, + "learning_rate": 2.0261111789005087e-05, + "loss": 0.0216, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011102675646543502, + "rkd_loss_weight": 0.5835027873516083, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38792336285114287, + "rkd_teacher_entropy": 3.742990016937256, + "rkd_teacher_similarity_mean": 0.7799195170402526, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12500 + }, + { + "epoch": 3.639802153040442, + "flow_matching_loss": 0.017537414096295834, + "grad_norm": 0.21790912747383118, + "learning_rate": 2.024562430423576e-05, + "loss": 0.0247, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012310887314379215, + "rkd_loss_weight": 0.5831694364547729, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.394584921002388, + "rkd_teacher_entropy": 3.7548553943634033, + "rkd_teacher_similarity_mean": 0.7870738625526428, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12510 + }, + { + "epoch": 3.642711667151586, + "flow_matching_loss": 0.01447230214253068, + "grad_norm": 0.15491564571857452, + "learning_rate": 2.0230130445538273e-05, + "loss": 0.0208, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010866539832204581, + "rkd_loss_weight": 0.5828360915184021, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3912324130535126, + "rkd_teacher_entropy": 3.7506547212600707, + "rkd_teacher_similarity_mean": 0.785632586479187, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12520 + }, + { + "epoch": 3.645621181262729, + "flow_matching_loss": 0.017461953964084388, + "grad_norm": 0.15478231012821198, + "learning_rate": 2.0214630231739127e-05, + "loss": 0.0236, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010547052230685949, + "rkd_loss_weight": 0.5825027525424957, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38444418013095855, + "rkd_teacher_entropy": 3.738406229019165, + "rkd_teacher_similarity_mean": 0.7784962892532349, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12530 + }, + { + "epoch": 3.6485306953738723, + "flow_matching_loss": 0.014885784778743982, + "grad_norm": 0.17678417265415192, + "learning_rate": 2.0199123681672543e-05, + "loss": 0.0215, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011329382890835404, + "rkd_loss_weight": 0.5821694016456604, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3876371204853058, + "rkd_teacher_entropy": 3.759446620941162, + "rkd_teacher_similarity_mean": 0.782075297832489, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12540 + }, + { + "epoch": 3.651440209485016, + "flow_matching_loss": 0.01663851095363498, + "grad_norm": 0.21684123575687408, + "learning_rate": 2.0183610814180437e-05, + "loss": 0.0229, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01074509397149086, + "rkd_loss_weight": 0.5818360626697541, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39510629177093504, + "rkd_teacher_entropy": 3.7659378051757812, + "rkd_teacher_similarity_mean": 0.7886342167854309, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12550 + }, + { + "epoch": 3.6543497235961593, + "flow_matching_loss": 0.018259890843182802, + "grad_norm": 0.13902664184570312, + "learning_rate": 2.016809164811241e-05, + "loss": 0.024, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009834661986678839, + "rkd_loss_weight": 0.5815027177333831, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3861401557922363, + "rkd_teacher_entropy": 3.7302481651306154, + "rkd_teacher_similarity_mean": 0.7798066198825836, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12560 + }, + { + "epoch": 3.657259237707303, + "flow_matching_loss": 0.01649058861657977, + "grad_norm": 0.18980449438095093, + "learning_rate": 2.015256620232571e-05, + "loss": 0.0226, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010459127742797137, + "rkd_loss_weight": 0.5811693668365479, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38290233314037325, + "rkd_teacher_entropy": 3.7327670097351073, + "rkd_teacher_similarity_mean": 0.7756226003170014, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12570 + }, + { + "epoch": 3.6601687518184463, + "flow_matching_loss": 0.017864481080323458, + "grad_norm": 0.1385728120803833, + "learning_rate": 2.0137034495685222e-05, + "loss": 0.0246, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011670035775750875, + "rkd_loss_weight": 0.5808360278606415, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38699751496315005, + "rkd_teacher_entropy": 3.7304296016693117, + "rkd_teacher_similarity_mean": 0.7785439312458038, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12580 + }, + { + "epoch": 3.66307826592959, + "flow_matching_loss": 0.01582524450495839, + "grad_norm": 0.118413545191288, + "learning_rate": 2.012149654706343e-05, + "loss": 0.0216, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00997004103846848, + "rkd_loss_weight": 0.5805026888847351, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3779824197292328, + "rkd_teacher_entropy": 3.7227651596069338, + "rkd_teacher_similarity_mean": 0.7718213379383088, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12590 + }, + { + "epoch": 3.6659877800407332, + "flow_matching_loss": 0.015089416410773993, + "grad_norm": 0.1301911324262619, + "learning_rate": 2.010595237534041e-05, + "loss": 0.0214, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010912671498954297, + "rkd_loss_weight": 0.5801693379878998, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3943567931652069, + "rkd_teacher_entropy": 3.7627771139144897, + "rkd_teacher_similarity_mean": 0.7893946945667267, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12600 + }, + { + "epoch": 3.6688972941518765, + "flow_matching_loss": 0.013822116516530513, + "grad_norm": 0.13928809762001038, + "learning_rate": 2.0090401999403793e-05, + "loss": 0.02, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010570927103981376, + "rkd_loss_weight": 0.5798359930515289, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3750214993953705, + "rkd_teacher_entropy": 3.7156723499298097, + "rkd_teacher_similarity_mean": 0.7703317403793335, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12610 + }, + { + "epoch": 3.67180680826302, + "flow_matching_loss": 0.01596508249640465, + "grad_norm": 0.21802115440368652, + "learning_rate": 2.007484543814875e-05, + "loss": 0.0219, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010276801930740476, + "rkd_loss_weight": 0.5795026540756225, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3829329639673233, + "rkd_teacher_entropy": 3.7334771156311035, + "rkd_teacher_similarity_mean": 0.7778987765312195, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12620 + }, + { + "epoch": 3.6747163223741635, + "flow_matching_loss": 0.0166568985208869, + "grad_norm": 0.19036102294921875, + "learning_rate": 2.0059282710477972e-05, + "loss": 0.0218, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008894608961418271, + "rkd_loss_weight": 0.5791693031787872, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38664776682853697, + "rkd_teacher_entropy": 3.7418497085571287, + "rkd_teacher_similarity_mean": 0.7816833019256592, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12630 + }, + { + "epoch": 3.6776258364853067, + "flow_matching_loss": 0.018387911934405567, + "grad_norm": 0.1353033185005188, + "learning_rate": 2.0043713835301647e-05, + "loss": 0.0242, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010092912055552006, + "rkd_loss_weight": 0.5788359642028809, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37561582028865814, + "rkd_teacher_entropy": 3.6927164793014526, + "rkd_teacher_similarity_mean": 0.7697081685066223, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12640 + }, + { + "epoch": 3.6805353505964504, + "flow_matching_loss": 0.01575953820720315, + "grad_norm": 0.13915909826755524, + "learning_rate": 2.0028138831537417e-05, + "loss": 0.0217, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010266422294080257, + "rkd_loss_weight": 0.57850261926651, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.389010015130043, + "rkd_teacher_entropy": 3.747674655914307, + "rkd_teacher_similarity_mean": 0.7839124858379364, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12650 + }, + { + "epoch": 3.6834448647075937, + "flow_matching_loss": 0.016020684409886597, + "grad_norm": 0.1431683450937271, + "learning_rate": 2.0012557718110387e-05, + "loss": 0.0217, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009820045111700893, + "rkd_loss_weight": 0.5781692683696746, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38307914733886717, + "rkd_teacher_entropy": 3.72769980430603, + "rkd_teacher_similarity_mean": 0.7768272221088409, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12660 + }, + { + "epoch": 3.6863543788187374, + "flow_matching_loss": 0.01763633033260703, + "grad_norm": 0.1596502810716629, + "learning_rate": 1.9996970513953077e-05, + "loss": 0.0237, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010574593301862478, + "rkd_loss_weight": 0.5778359353542328, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3879896342754364, + "rkd_teacher_entropy": 3.7521589756011964, + "rkd_teacher_similarity_mean": 0.7839715480804443, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12670 + }, + { + "epoch": 3.6892638929298807, + "flow_matching_loss": 0.01583824846893549, + "grad_norm": 0.14017558097839355, + "learning_rate": 1.9981377238005405e-05, + "loss": 0.022, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010746796708554029, + "rkd_loss_weight": 0.577502578496933, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39109828770160676, + "rkd_teacher_entropy": 3.7601780891418457, + "rkd_teacher_similarity_mean": 0.786212170124054, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12680 + }, + { + "epoch": 3.6921734070410244, + "flow_matching_loss": 0.01748856930062175, + "grad_norm": 0.15661220252513885, + "learning_rate": 1.9965777909214674e-05, + "loss": 0.024, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011281977687031031, + "rkd_loss_weight": 0.5771692395210266, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39383518099784853, + "rkd_teacher_entropy": 3.7620455265045165, + "rkd_teacher_similarity_mean": 0.7879540920257568, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12690 + }, + { + "epoch": 3.6950829211521676, + "flow_matching_loss": 0.015512090362608432, + "grad_norm": 0.13528814911842346, + "learning_rate": 1.9950172546535545e-05, + "loss": 0.0213, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009969404293224215, + "rkd_loss_weight": 0.5768359005451202, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38968858420848845, + "rkd_teacher_entropy": 3.73136146068573, + "rkd_teacher_similarity_mean": 0.7803931832313538, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12700 + }, + { + "epoch": 3.697992435263311, + "flow_matching_loss": 0.016137535404413938, + "grad_norm": 0.304132342338562, + "learning_rate": 1.9934561168930002e-05, + "loss": 0.023, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011951983813196421, + "rkd_loss_weight": 0.5765025436878204, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3869414091110229, + "rkd_teacher_entropy": 3.7322760820388794, + "rkd_teacher_similarity_mean": 0.7799836575984955, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12710 + }, + { + "epoch": 3.7009019493744546, + "flow_matching_loss": 0.017855302337557076, + "grad_norm": 0.13299812376499176, + "learning_rate": 1.991894379536734e-05, + "loss": 0.024, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010699552111327648, + "rkd_loss_weight": 0.5761692106723786, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39174762964248655, + "rkd_teacher_entropy": 3.7560920476913453, + "rkd_teacher_similarity_mean": 0.7858011782169342, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12720 + }, + { + "epoch": 3.703811463485598, + "flow_matching_loss": 0.018556949123740195, + "grad_norm": 0.15944287180900574, + "learning_rate": 1.9903320444824137e-05, + "loss": 0.0254, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011957023199647665, + "rkd_loss_weight": 0.5758358597755432, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3905579298734665, + "rkd_teacher_entropy": 3.7490829706192015, + "rkd_teacher_similarity_mean": 0.7827381789684296, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12730 + }, + { + "epoch": 3.706720977596741, + "flow_matching_loss": 0.014340752735733986, + "grad_norm": 0.1249518170952797, + "learning_rate": 1.9887691136284245e-05, + "loss": 0.0192, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008529054978862406, + "rkd_loss_weight": 0.5755025148391724, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3931089460849762, + "rkd_teacher_entropy": 3.754074478149414, + "rkd_teacher_similarity_mean": 0.7863570749759674, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12740 + }, + { + "epoch": 3.709630491707885, + "flow_matching_loss": 0.015973377600312234, + "grad_norm": 0.14314039051532745, + "learning_rate": 1.9872055888738745e-05, + "loss": 0.0218, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010049565695226192, + "rkd_loss_weight": 0.575169175863266, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3869696170091629, + "rkd_teacher_entropy": 3.740568518638611, + "rkd_teacher_similarity_mean": 0.7795029282569885, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12750 + }, + { + "epoch": 3.712540005819028, + "flow_matching_loss": 0.01838002111762762, + "grad_norm": 0.13662227988243103, + "learning_rate": 1.985641472118593e-05, + "loss": 0.0238, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009455543151125312, + "rkd_loss_weight": 0.5748358249664307, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38278440833091737, + "rkd_teacher_entropy": 3.7272249698638915, + "rkd_teacher_similarity_mean": 0.7770208835601806, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12760 + }, + { + "epoch": 3.715449519930172, + "flow_matching_loss": 0.01817467100918293, + "grad_norm": 0.16266299784183502, + "learning_rate": 1.9840767652631294e-05, + "loss": 0.0243, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010674463305622339, + "rkd_loss_weight": 0.5745024859905243, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3810086488723755, + "rkd_teacher_entropy": 3.7214396476745604, + "rkd_teacher_similarity_mean": 0.775000524520874, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12770 + }, + { + "epoch": 3.718359034041315, + "flow_matching_loss": 0.017004005797207355, + "grad_norm": 0.15475396811962128, + "learning_rate": 1.9825114702087513e-05, + "loss": 0.0225, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009503857139497996, + "rkd_loss_weight": 0.5741691410541534, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3856954127550125, + "rkd_teacher_entropy": 3.735055685043335, + "rkd_teacher_similarity_mean": 0.7807457566261291, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12780 + }, + { + "epoch": 3.7212685481524588, + "flow_matching_loss": 0.017138451058417558, + "grad_norm": 0.23445294797420502, + "learning_rate": 1.9809455888574388e-05, + "loss": 0.0233, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010754669550806284, + "rkd_loss_weight": 0.5738357901573181, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.380136975646019, + "rkd_teacher_entropy": 3.7317983150482177, + "rkd_teacher_similarity_mean": 0.7750640392303467, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12790 + }, + { + "epoch": 3.724178062263602, + "flow_matching_loss": 0.016640072222799063, + "grad_norm": 0.14862066507339478, + "learning_rate": 1.9793791231118864e-05, + "loss": 0.0219, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0091221883893013, + "rkd_loss_weight": 0.5735024511814117, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3861099869012833, + "rkd_teacher_entropy": 3.7394461154937746, + "rkd_teacher_similarity_mean": 0.7800759851932526, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12800 + }, + { + "epoch": 3.7270875763747453, + "flow_matching_loss": 0.015043993294239045, + "grad_norm": 0.1125853881239891, + "learning_rate": 1.9778120748754958e-05, + "loss": 0.0216, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011471627652645111, + "rkd_loss_weight": 0.5731691062450409, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3969193369150162, + "rkd_teacher_entropy": 3.7696385860443113, + "rkd_teacher_similarity_mean": 0.7919695556163788, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12810 + }, + { + "epoch": 3.729997090485889, + "flow_matching_loss": 0.016168993059545757, + "grad_norm": 0.12123750150203705, + "learning_rate": 1.97624444605238e-05, + "loss": 0.0219, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009991578012704849, + "rkd_loss_weight": 0.5728357613086701, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3831066906452179, + "rkd_teacher_entropy": 3.735363984107971, + "rkd_teacher_similarity_mean": 0.7773921668529511, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12820 + }, + { + "epoch": 3.7329066045970323, + "flow_matching_loss": 0.01613922230899334, + "grad_norm": 0.34889334440231323, + "learning_rate": 1.9746762385473552e-05, + "loss": 0.0229, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011814844282343983, + "rkd_loss_weight": 0.5725024163722991, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3859322637319565, + "rkd_teacher_entropy": 3.741935682296753, + "rkd_teacher_similarity_mean": 0.7812684059143067, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12830 + }, + { + "epoch": 3.7358161187081755, + "flow_matching_loss": 0.015514307469129563, + "grad_norm": 0.1528555452823639, + "learning_rate": 1.9731074542659414e-05, + "loss": 0.0224, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012106281612068415, + "rkd_loss_weight": 0.5721690714359283, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3850843280553818, + "rkd_teacher_entropy": 3.7411158323287963, + "rkd_teacher_similarity_mean": 0.7803699970245361, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12840 + }, + { + "epoch": 3.7387256328193192, + "flow_matching_loss": 0.018406127113848925, + "grad_norm": 0.18036730587482452, + "learning_rate": 1.9715380951143603e-05, + "loss": 0.0246, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010845934692770242, + "rkd_loss_weight": 0.5718357264995575, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38463265299797056, + "rkd_teacher_entropy": 3.7332703113555907, + "rkd_teacher_similarity_mean": 0.7779872596263886, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12850 + }, + { + "epoch": 3.7416351469304625, + "flow_matching_loss": 0.01643949681892991, + "grad_norm": 0.1330118477344513, + "learning_rate": 1.9699681629995295e-05, + "loss": 0.0229, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011348777590319515, + "rkd_loss_weight": 0.5715023875236511, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.391739022731781, + "rkd_teacher_entropy": 3.7545933723449707, + "rkd_teacher_similarity_mean": 0.7867319166660309, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12860 + }, + { + "epoch": 3.744544661041606, + "flow_matching_loss": 0.01712703686207533, + "grad_norm": 0.17985697090625763, + "learning_rate": 1.968397659829066e-05, + "loss": 0.0237, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011424683034420013, + "rkd_loss_weight": 0.5711690366268158, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38805108666419985, + "rkd_teacher_entropy": 3.742787766456604, + "rkd_teacher_similarity_mean": 0.784423154592514, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12870 + }, + { + "epoch": 3.7474541751527495, + "flow_matching_loss": 0.015501459501683712, + "grad_norm": 0.11531278491020203, + "learning_rate": 1.966826587511279e-05, + "loss": 0.021, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00961170275695622, + "rkd_loss_weight": 0.5708356916904449, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3875621110200882, + "rkd_teacher_entropy": 3.7536415576934816, + "rkd_teacher_similarity_mean": 0.7838643968105317, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12880 + }, + { + "epoch": 3.750363689263893, + "flow_matching_loss": 0.017566777113825083, + "grad_norm": 0.16455475986003876, + "learning_rate": 1.965254947955169e-05, + "loss": 0.0236, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010523789562284946, + "rkd_loss_weight": 0.5705023527145385, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38447438180446625, + "rkd_teacher_entropy": 3.742937469482422, + "rkd_teacher_similarity_mean": 0.7812733054161072, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12890 + }, + { + "epoch": 3.7532732033750364, + "flow_matching_loss": 0.01479962132871151, + "grad_norm": 0.11221630126237869, + "learning_rate": 1.963682743070427e-05, + "loss": 0.0198, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008786582294851542, + "rkd_loss_weight": 0.5701690018177032, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38530613780021666, + "rkd_teacher_entropy": 3.7446258306503295, + "rkd_teacher_similarity_mean": 0.7794397115707398, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12900 + }, + { + "epoch": 3.7561827174861797, + "flow_matching_loss": 0.015493093710392713, + "grad_norm": 0.17041024565696716, + "learning_rate": 1.962109974767429e-05, + "loss": 0.0212, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009968857094645501, + "rkd_loss_weight": 0.5698356628417969, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3897756218910217, + "rkd_teacher_entropy": 3.752653789520264, + "rkd_teacher_similarity_mean": 0.7851400792598724, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12910 + }, + { + "epoch": 3.7590922315973234, + "flow_matching_loss": 0.016014330554753542, + "grad_norm": 0.11275232583284378, + "learning_rate": 1.960536644957238e-05, + "loss": 0.0218, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010225034318864346, + "rkd_loss_weight": 0.569502317905426, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38531701266765594, + "rkd_teacher_entropy": 3.728294563293457, + "rkd_teacher_similarity_mean": 0.7788093328475952, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12920 + }, + { + "epoch": 3.7620017457084667, + "flow_matching_loss": 0.018808699492365123, + "grad_norm": 0.14638406038284302, + "learning_rate": 1.9589627555515974e-05, + "loss": 0.0255, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011810178915038706, + "rkd_loss_weight": 0.5691689670085907, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38608448803424833, + "rkd_teacher_entropy": 3.7599076509475706, + "rkd_teacher_similarity_mean": 0.7815551996231079, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12930 + }, + { + "epoch": 3.76491125981961, + "flow_matching_loss": 0.016343070194125175, + "grad_norm": 0.14580915868282318, + "learning_rate": 1.9573883084629317e-05, + "loss": 0.0215, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009116399195045232, + "rkd_loss_weight": 0.5688356339931488, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.380817773938179, + "rkd_teacher_entropy": 3.7317614793777465, + "rkd_teacher_similarity_mean": 0.7760476171970367, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12940 + }, + { + "epoch": 3.7678207739307537, + "flow_matching_loss": 0.01773016694933176, + "grad_norm": 0.13296520709991455, + "learning_rate": 1.9558133056043422e-05, + "loss": 0.0239, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010913202073425055, + "rkd_loss_weight": 0.568502277135849, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3921300143003464, + "rkd_teacher_entropy": 3.762964606285095, + "rkd_teacher_similarity_mean": 0.7870687484741211, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12950 + }, + { + "epoch": 3.770730288041897, + "flow_matching_loss": 0.016553179081529378, + "grad_norm": 0.11697728931903839, + "learning_rate": 1.9542377488896062e-05, + "loss": 0.0226, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010558768967166543, + "rkd_loss_weight": 0.5681689381599426, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38585489690303804, + "rkd_teacher_entropy": 3.749977135658264, + "rkd_teacher_similarity_mean": 0.7815354764461517, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12960 + }, + { + "epoch": 3.77363980215304, + "flow_matching_loss": 0.015583399031311274, + "grad_norm": 0.12041793018579483, + "learning_rate": 1.9526616402331733e-05, + "loss": 0.021, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009620442846789956, + "rkd_loss_weight": 0.5678355991840363, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3865932673215866, + "rkd_teacher_entropy": 3.7357717037200926, + "rkd_teacher_similarity_mean": 0.7819673299789429, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12970 + }, + { + "epoch": 3.776549316264184, + "flow_matching_loss": 0.015142019931226968, + "grad_norm": 0.12570255994796753, + "learning_rate": 1.9510849815501638e-05, + "loss": 0.0202, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0089634639210999, + "rkd_loss_weight": 0.5675022423267364, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3904029905796051, + "rkd_teacher_entropy": 3.754174327850342, + "rkd_teacher_similarity_mean": 0.7863447844982148, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12980 + }, + { + "epoch": 3.7794588303753276, + "flow_matching_loss": 0.016178760305047037, + "grad_norm": 0.15510162711143494, + "learning_rate": 1.949507774756368e-05, + "loss": 0.0222, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010604238044470549, + "rkd_loss_weight": 0.5671689093112946, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39747016727924345, + "rkd_teacher_entropy": 3.7777004480361938, + "rkd_teacher_similarity_mean": 0.7932454526424408, + "rkd_vlm_temp": 0.10000000149011612, + "step": 12990 + }, + { + "epoch": 3.782368344486471, + "flow_matching_loss": 0.018286052998155354, + "grad_norm": 0.15566949546337128, + "learning_rate": 1.9479300217682394e-05, + "loss": 0.0251, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011945134913548826, + "rkd_loss_weight": 0.5668355584144592, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38640803694725034, + "rkd_teacher_entropy": 3.751129651069641, + "rkd_teacher_similarity_mean": 0.7812193334102631, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13000 + }, + { + "epoch": 3.785277858597614, + "flow_matching_loss": 0.01697117304429412, + "grad_norm": 0.17298685014247894, + "learning_rate": 1.946351724502898e-05, + "loss": 0.0238, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011969800386577845, + "rkd_loss_weight": 0.5665022134780884, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39294540584087373, + "rkd_teacher_entropy": 3.7628824949264525, + "rkd_teacher_similarity_mean": 0.7873679816722869, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13010 + }, + { + "epoch": 3.788187372708758, + "flow_matching_loss": 0.01672116694971919, + "grad_norm": 0.14099648594856262, + "learning_rate": 1.9447728848781227e-05, + "loss": 0.0228, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010778781725093722, + "rkd_loss_weight": 0.566168874502182, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3907987713813782, + "rkd_teacher_entropy": 3.7544252634048463, + "rkd_teacher_similarity_mean": 0.7860929667949677, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13020 + }, + { + "epoch": 3.791096886819901, + "flow_matching_loss": 0.016009471472352745, + "grad_norm": 0.1491777002811432, + "learning_rate": 1.943193504812353e-05, + "loss": 0.0213, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009282778203487396, + "rkd_loss_weight": 0.5658355295658112, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39272057414054873, + "rkd_teacher_entropy": 3.7649544954299925, + "rkd_teacher_similarity_mean": 0.7884357213973999, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13030 + }, + { + "epoch": 3.7940064009310444, + "flow_matching_loss": 0.015755610726773738, + "grad_norm": 0.12029265612363815, + "learning_rate": 1.9416135862246846e-05, + "loss": 0.0233, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.013310310803353787, + "rkd_loss_weight": 0.5655021846294404, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39409870505332945, + "rkd_teacher_entropy": 3.760671520233154, + "rkd_teacher_similarity_mean": 0.7881351947784424, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13040 + }, + { + "epoch": 3.796915915042188, + "flow_matching_loss": 0.018088317662477493, + "grad_norm": 0.17134246230125427, + "learning_rate": 1.9400331310348673e-05, + "loss": 0.0233, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009134124824777245, + "rkd_loss_weight": 0.5651688396930694, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3871101289987564, + "rkd_teacher_entropy": 3.7332090616226195, + "rkd_teacher_similarity_mean": 0.7802634477615357, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13050 + }, + { + "epoch": 3.7998254291533313, + "flow_matching_loss": 0.016046538576483728, + "grad_norm": 0.14028187096118927, + "learning_rate": 1.9384521411633033e-05, + "loss": 0.0218, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010257471725344658, + "rkd_loss_weight": 0.5648354947566986, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3902104556560516, + "rkd_teacher_entropy": 3.7385280609130858, + "rkd_teacher_similarity_mean": 0.7835279643535614, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13060 + }, + { + "epoch": 3.8027349432644746, + "flow_matching_loss": 0.01396969985216856, + "grad_norm": 0.1321258395910263, + "learning_rate": 1.936870618531045e-05, + "loss": 0.0192, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009330721152946354, + "rkd_loss_weight": 0.5645021498203278, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38367733657360076, + "rkd_teacher_entropy": 3.7340871334075927, + "rkd_teacher_similarity_mean": 0.7771281719207763, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13070 + }, + { + "epoch": 3.8056444573756183, + "flow_matching_loss": 0.017875214386731387, + "grad_norm": 0.130575031042099, + "learning_rate": 1.9352885650597904e-05, + "loss": 0.0241, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011038825195282698, + "rkd_loss_weight": 0.5641688108444214, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3933953493833542, + "rkd_teacher_entropy": 3.7559643507003786, + "rkd_teacher_similarity_mean": 0.7870792925357819, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13080 + }, + { + "epoch": 3.8085539714867616, + "flow_matching_loss": 0.01731226360425353, + "grad_norm": 0.15398791432380676, + "learning_rate": 1.9337059826718847e-05, + "loss": 0.0232, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010458429902791977, + "rkd_loss_weight": 0.5638354599475861, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3913642466068268, + "rkd_teacher_entropy": 3.743945503234863, + "rkd_teacher_similarity_mean": 0.7857958495616912, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13090 + }, + { + "epoch": 3.8114634855979053, + "flow_matching_loss": 0.015478010196238757, + "grad_norm": 0.1677735447883606, + "learning_rate": 1.932122873290314e-05, + "loss": 0.0205, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008921240223571658, + "rkd_loss_weight": 0.5635021150112152, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3937062531709671, + "rkd_teacher_entropy": 3.7660845041275026, + "rkd_teacher_similarity_mean": 0.787337201833725, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13100 + }, + { + "epoch": 3.8143729997090485, + "flow_matching_loss": 0.017698917258530854, + "grad_norm": 0.14007766544818878, + "learning_rate": 1.9305392388387054e-05, + "loss": 0.0249, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012874445086345077, + "rkd_loss_weight": 0.5631687760353088, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3787721335887909, + "rkd_teacher_entropy": 3.7111385107040404, + "rkd_teacher_similarity_mean": 0.7718858838081359, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13110 + }, + { + "epoch": 3.8172825138201922, + "flow_matching_loss": 0.015340095665305854, + "grad_norm": 0.14289098978042603, + "learning_rate": 1.928955081241325e-05, + "loss": 0.0214, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010749618895351887, + "rkd_loss_weight": 0.5628354251384735, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37933054864406585, + "rkd_teacher_entropy": 3.707157826423645, + "rkd_teacher_similarity_mean": 0.7730563700199127, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13120 + }, + { + "epoch": 3.8201920279313355, + "flow_matching_loss": 0.016646271478384732, + "grad_norm": 0.13323181867599487, + "learning_rate": 1.9273704024230725e-05, + "loss": 0.0226, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010535323899239302, + "rkd_loss_weight": 0.5625020861625671, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39024797081947327, + "rkd_teacher_entropy": 3.7533586263656615, + "rkd_teacher_similarity_mean": 0.7846658110618592, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13130 + }, + { + "epoch": 3.8231015420424788, + "flow_matching_loss": 0.014744718745350837, + "grad_norm": 0.15230552852153778, + "learning_rate": 1.925785204309483e-05, + "loss": 0.0199, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009213961334899068, + "rkd_loss_weight": 0.5621687412261963, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38895292580127716, + "rkd_teacher_entropy": 3.7459087133407594, + "rkd_teacher_similarity_mean": 0.7831291854381561, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13140 + }, + { + "epoch": 3.8260110561536225, + "flow_matching_loss": 0.01632644822821021, + "grad_norm": 0.14859187602996826, + "learning_rate": 1.9241994888267212e-05, + "loss": 0.0221, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010243611317127943, + "rkd_loss_weight": 0.5618353903293609, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39066720604896543, + "rkd_teacher_entropy": 3.7461074352264405, + "rkd_teacher_similarity_mean": 0.7854294836521148, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13150 + }, + { + "epoch": 3.8289205702647657, + "flow_matching_loss": 0.015008402895182371, + "grad_norm": 0.11918096244335175, + "learning_rate": 1.9226132579015815e-05, + "loss": 0.0202, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009236089233309031, + "rkd_loss_weight": 0.5615020573139191, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.381158384680748, + "rkd_teacher_entropy": 3.7328386068344117, + "rkd_teacher_similarity_mean": 0.7752498984336853, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13160 + }, + { + "epoch": 3.831830084375909, + "flow_matching_loss": 0.016315168514847756, + "grad_norm": 0.10899753868579865, + "learning_rate": 1.9210265134614832e-05, + "loss": 0.0214, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009026558883488179, + "rkd_loss_weight": 0.5611687004566193, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3883275330066681, + "rkd_teacher_entropy": 3.7591461420059202, + "rkd_teacher_similarity_mean": 0.7847889125347137, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13170 + }, + { + "epoch": 3.8347395984870527, + "flow_matching_loss": 0.015375897195190192, + "grad_norm": 0.14033609628677368, + "learning_rate": 1.9194392574344712e-05, + "loss": 0.0216, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01107148788869381, + "rkd_loss_weight": 0.5608353614807129, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38422935307025907, + "rkd_teacher_entropy": 3.7358104944229127, + "rkd_teacher_similarity_mean": 0.7802206993103027, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13180 + }, + { + "epoch": 3.837649112598196, + "flow_matching_loss": 0.01883981879800558, + "grad_norm": 0.18795561790466309, + "learning_rate": 1.9178514917492103e-05, + "loss": 0.0249, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01083438703790307, + "rkd_loss_weight": 0.5605020225048065, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3901412069797516, + "rkd_teacher_entropy": 3.758239579200745, + "rkd_teacher_similarity_mean": 0.7853290855884552, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13190 + }, + { + "epoch": 3.8405586267093397, + "flow_matching_loss": 0.014361642114818097, + "grad_norm": 0.1507813185453415, + "learning_rate": 1.9162632183349866e-05, + "loss": 0.0199, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0098555869422853, + "rkd_loss_weight": 0.5601686656475067, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38698300123214724, + "rkd_teacher_entropy": 3.750980281829834, + "rkd_teacher_similarity_mean": 0.7824545919895172, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13200 + }, + { + "epoch": 3.843468140820483, + "flow_matching_loss": 0.016360741667449476, + "grad_norm": 0.12194015830755234, + "learning_rate": 1.914674439121701e-05, + "loss": 0.0225, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010891510965302586, + "rkd_loss_weight": 0.5598353326320649, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38324193060398104, + "rkd_teacher_entropy": 3.729469156265259, + "rkd_teacher_similarity_mean": 0.7785851836204529, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13210 + }, + { + "epoch": 3.8463776549316266, + "flow_matching_loss": 0.01651608692482114, + "grad_norm": 0.12852373719215393, + "learning_rate": 1.9130851560398715e-05, + "loss": 0.0221, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00997392563149333, + "rkd_loss_weight": 0.5595019817352295, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3890842586755753, + "rkd_teacher_entropy": 3.7499625444412232, + "rkd_teacher_similarity_mean": 0.7833864867687226, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13220 + }, + { + "epoch": 3.84928716904277, + "flow_matching_loss": 0.016590669006109237, + "grad_norm": 0.12296545505523682, + "learning_rate": 1.911495371020625e-05, + "loss": 0.0213, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00847500106319785, + "rkd_loss_weight": 0.5591686367988586, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38130720257759093, + "rkd_teacher_entropy": 3.7167317867279053, + "rkd_teacher_similarity_mean": 0.7775058805942535, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13230 + }, + { + "epoch": 3.852196683153913, + "flow_matching_loss": 0.014291033428162337, + "grad_norm": 0.12798956036567688, + "learning_rate": 1.909905085995702e-05, + "loss": 0.0198, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009912232775241137, + "rkd_loss_weight": 0.5588352978229523, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38205197751522063, + "rkd_teacher_entropy": 3.743632364273071, + "rkd_teacher_similarity_mean": 0.7770219802856445, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13240 + }, + { + "epoch": 3.855106197265057, + "flow_matching_loss": 0.01590652335435152, + "grad_norm": 0.10572158545255661, + "learning_rate": 1.908314302897448e-05, + "loss": 0.0224, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011539193196222186, + "rkd_loss_weight": 0.558501946926117, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3882107824087143, + "rkd_teacher_entropy": 3.745874524116516, + "rkd_teacher_similarity_mean": 0.7837092280387878, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13250 + }, + { + "epoch": 3.8580157113762, + "flow_matching_loss": 0.01691780937835574, + "grad_norm": 0.09550908952951431, + "learning_rate": 1.9067230236588142e-05, + "loss": 0.0221, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009338005119934678, + "rkd_loss_weight": 0.5581686079502106, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3937837690114975, + "rkd_teacher_entropy": 3.7615199089050293, + "rkd_teacher_similarity_mean": 0.7880826950073242, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13260 + }, + { + "epoch": 3.8609252254873434, + "flow_matching_loss": 0.015563318552449346, + "grad_norm": 0.1102750301361084, + "learning_rate": 1.905131250213355e-05, + "loss": 0.0209, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009590072371065617, + "rkd_loss_weight": 0.5578352630138397, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38232971727848053, + "rkd_teacher_entropy": 3.723012614250183, + "rkd_teacher_similarity_mean": 0.7765276670455933, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13270 + }, + { + "epoch": 3.863834739598487, + "flow_matching_loss": 0.016057403478771447, + "grad_norm": 0.14229919016361237, + "learning_rate": 1.9035389844952256e-05, + "loss": 0.0208, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008443730417639017, + "rkd_loss_weight": 0.5575019121170044, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3954047620296478, + "rkd_teacher_entropy": 3.770269513130188, + "rkd_teacher_similarity_mean": 0.7921039044857026, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13280 + }, + { + "epoch": 3.8667442537096304, + "flow_matching_loss": 0.017695362307131292, + "grad_norm": 0.15767377614974976, + "learning_rate": 1.901946228439179e-05, + "loss": 0.0238, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0108936064876616, + "rkd_loss_weight": 0.557168573141098, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38812435865402223, + "rkd_teacher_entropy": 3.748405432701111, + "rkd_teacher_similarity_mean": 0.7827330768108368, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13290 + }, + { + "epoch": 3.869653767820774, + "flow_matching_loss": 0.014641310926526785, + "grad_norm": 0.14189903438091278, + "learning_rate": 1.9003529839805637e-05, + "loss": 0.0195, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008751361072063446, + "rkd_loss_weight": 0.5568352282047272, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3791756421327591, + "rkd_teacher_entropy": 3.7162570476531984, + "rkd_teacher_similarity_mean": 0.7750256776809692, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13300 + }, + { + "epoch": 3.8725632819319173, + "flow_matching_loss": 0.016010104678571224, + "grad_norm": 0.16955488920211792, + "learning_rate": 1.8987592530553217e-05, + "loss": 0.0224, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0115061413962394, + "rkd_loss_weight": 0.5565018832683564, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3895725905895233, + "rkd_teacher_entropy": 3.744864821434021, + "rkd_teacher_similarity_mean": 0.7837300717830658, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13310 + }, + { + "epoch": 3.875472796043061, + "flow_matching_loss": 0.019805910624563695, + "grad_norm": 0.16809090971946716, + "learning_rate": 1.897165037599987e-05, + "loss": 0.0249, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009198209457099437, + "rkd_loss_weight": 0.5561685383319854, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38337494134902955, + "rkd_teacher_entropy": 3.735206866264343, + "rkd_teacher_similarity_mean": 0.7769678354263305, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13320 + }, + { + "epoch": 3.8783823101542043, + "flow_matching_loss": 0.017212343122810125, + "grad_norm": 0.1056973859667778, + "learning_rate": 1.8955703395516814e-05, + "loss": 0.0226, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00976336789317429, + "rkd_loss_weight": 0.5558351933956146, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3798086553812027, + "rkd_teacher_entropy": 3.728218126296997, + "rkd_teacher_similarity_mean": 0.7737678468227387, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13330 + }, + { + "epoch": 3.8812918242653476, + "flow_matching_loss": 0.015522543899714946, + "grad_norm": 0.16882435977458954, + "learning_rate": 1.8939751608481134e-05, + "loss": 0.0209, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009640775062143803, + "rkd_loss_weight": 0.5555018484592438, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3865471601486206, + "rkd_teacher_entropy": 3.7518231868743896, + "rkd_teacher_similarity_mean": 0.7828288495540618, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13340 + }, + { + "epoch": 3.8842013383764913, + "flow_matching_loss": 0.015710464771836997, + "grad_norm": 0.20931315422058105, + "learning_rate": 1.892379503427575e-05, + "loss": 0.0211, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00974478768184781, + "rkd_loss_weight": 0.5551685094833374, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3932844907045364, + "rkd_teacher_entropy": 3.75534930229187, + "rkd_teacher_similarity_mean": 0.7868433475494385, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13350 + }, + { + "epoch": 3.8871108524876345, + "flow_matching_loss": 0.019308330863714217, + "grad_norm": 0.16801010072231293, + "learning_rate": 1.8907833692289412e-05, + "loss": 0.0259, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011902076797559858, + "rkd_loss_weight": 0.5548351585865021, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38605842292308806, + "rkd_teacher_entropy": 3.7287245988845825, + "rkd_teacher_similarity_mean": 0.7813280045986175, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13360 + }, + { + "epoch": 3.890020366598778, + "flow_matching_loss": 0.013640288077294827, + "grad_norm": 0.12186131626367569, + "learning_rate": 1.889186760191666e-05, + "loss": 0.0179, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00768857691437006, + "rkd_loss_weight": 0.5545018136501312, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39725437164306643, + "rkd_teacher_entropy": 3.7786184549331665, + "rkd_teacher_similarity_mean": 0.7937830448150635, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13370 + }, + { + "epoch": 3.8929298807099215, + "flow_matching_loss": 0.015438609197735787, + "grad_norm": 0.14511533081531525, + "learning_rate": 1.887589678255779e-05, + "loss": 0.0203, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00879208892583847, + "rkd_loss_weight": 0.5541684746742248, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3864015400409698, + "rkd_teacher_entropy": 3.740884566307068, + "rkd_teacher_similarity_mean": 0.7818450272083283, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13380 + }, + { + "epoch": 3.895839394821065, + "flow_matching_loss": 0.01545423809438944, + "grad_norm": 0.1570306271314621, + "learning_rate": 1.8859921253618852e-05, + "loss": 0.0214, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010823759576305747, + "rkd_loss_weight": 0.5538351237773895, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3937904447317123, + "rkd_teacher_entropy": 3.752210998535156, + "rkd_teacher_similarity_mean": 0.7879610061645508, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13390 + }, + { + "epoch": 3.898748908932208, + "flow_matching_loss": 0.016391276754438876, + "grad_norm": 0.12857049703598022, + "learning_rate": 1.884394103451163e-05, + "loss": 0.0224, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010897449869662523, + "rkd_loss_weight": 0.5535017848014832, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39033598005771636, + "rkd_teacher_entropy": 3.7529006004333496, + "rkd_teacher_similarity_mean": 0.7842790424823761, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13400 + }, + { + "epoch": 3.9016584230433518, + "flow_matching_loss": 0.016202117223292588, + "grad_norm": 0.12883441150188446, + "learning_rate": 1.8827956144653583e-05, + "loss": 0.0227, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011742738354951144, + "rkd_loss_weight": 0.5531684398651123, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3959999293088913, + "rkd_teacher_entropy": 3.7609389066696166, + "rkd_teacher_similarity_mean": 0.7874560952186584, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13410 + }, + { + "epoch": 3.9045679371544955, + "flow_matching_loss": 0.01835427088662982, + "grad_norm": 0.18888358771800995, + "learning_rate": 1.8811966603467874e-05, + "loss": 0.0234, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009068696852773428, + "rkd_loss_weight": 0.5528350889682769, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.395158463716507, + "rkd_teacher_entropy": 3.7545048236846923, + "rkd_teacher_similarity_mean": 0.7870621025562287, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13420 + }, + { + "epoch": 3.9074774512656387, + "flow_matching_loss": 0.01783349197357893, + "grad_norm": 0.14124229550361633, + "learning_rate": 1.8795972430383293e-05, + "loss": 0.0227, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008717895764857531, + "rkd_loss_weight": 0.5525017559528351, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39770875573158265, + "rkd_teacher_entropy": 3.7649896144866943, + "rkd_teacher_similarity_mean": 0.7914419591426849, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13430 + }, + { + "epoch": 3.910386965376782, + "flow_matching_loss": 0.016814519930630923, + "grad_norm": 0.13207006454467773, + "learning_rate": 1.8779973644834273e-05, + "loss": 0.0214, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008217021403834224, + "rkd_loss_weight": 0.5521683990955353, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3927711546421051, + "rkd_teacher_entropy": 3.7520440816879272, + "rkd_teacher_similarity_mean": 0.7873454332351685, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13440 + }, + { + "epoch": 3.9132964794879257, + "flow_matching_loss": 0.014207857102155686, + "grad_norm": 0.12709900736808777, + "learning_rate": 1.876397026626085e-05, + "loss": 0.0199, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010341810109093786, + "rkd_loss_weight": 0.5518350601196289, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3886430233716965, + "rkd_teacher_entropy": 3.7488327264785766, + "rkd_teacher_similarity_mean": 0.7831282615661621, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13450 + }, + { + "epoch": 3.916205993599069, + "flow_matching_loss": 0.016147048398852348, + "grad_norm": 0.14441949129104614, + "learning_rate": 1.8747962314108633e-05, + "loss": 0.0212, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009216850716620683, + "rkd_loss_weight": 0.5515017211437225, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38979889154434205, + "rkd_teacher_entropy": 3.7426537752151487, + "rkd_teacher_similarity_mean": 0.7828722894191742, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13460 + }, + { + "epoch": 3.919115507710212, + "flow_matching_loss": 0.016409459337592126, + "grad_norm": 0.16556930541992188, + "learning_rate": 1.873194980782879e-05, + "loss": 0.0225, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01108158528804779, + "rkd_loss_weight": 0.5511683702468873, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38197643458843233, + "rkd_teacher_entropy": 3.721217393875122, + "rkd_teacher_similarity_mean": 0.774132490158081, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13470 + }, + { + "epoch": 3.922025021821356, + "flow_matching_loss": 0.01687170611694455, + "grad_norm": 0.11704910546541214, + "learning_rate": 1.871593276687804e-05, + "loss": 0.0229, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010916345193982125, + "rkd_loss_weight": 0.5508350312709809, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38658786714077, + "rkd_teacher_entropy": 3.742514395713806, + "rkd_teacher_similarity_mean": 0.7816598653793335, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13480 + }, + { + "epoch": 3.924934535932499, + "flow_matching_loss": 0.014708083588629961, + "grad_norm": 0.10229820013046265, + "learning_rate": 1.869991121071858e-05, + "loss": 0.0198, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009267277270555496, + "rkd_loss_weight": 0.5505016863346099, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3759728342294693, + "rkd_teacher_entropy": 3.7149418115615847, + "rkd_teacher_similarity_mean": 0.7711698591709137, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13490 + }, + { + "epoch": 3.9278440500436425, + "flow_matching_loss": 0.016493247356265783, + "grad_norm": 0.12882757186889648, + "learning_rate": 1.8683885158818117e-05, + "loss": 0.022, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009939222550019621, + "rkd_loss_weight": 0.5501683354377747, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3941461890935898, + "rkd_teacher_entropy": 3.7670345306396484, + "rkd_teacher_similarity_mean": 0.7895209610462188, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13500 + }, + { + "epoch": 3.930753564154786, + "flow_matching_loss": 0.01808543559163809, + "grad_norm": 0.15820783376693726, + "learning_rate": 1.8667854630649825e-05, + "loss": 0.0237, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010248315148055553, + "rkd_loss_weight": 0.5498349964618683, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38405910730361936, + "rkd_teacher_entropy": 3.7421767950057983, + "rkd_teacher_similarity_mean": 0.780545037984848, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13510 + }, + { + "epoch": 3.9336630782659294, + "flow_matching_loss": 0.014073537103831768, + "grad_norm": 0.2284032702445984, + "learning_rate": 1.86518196456923e-05, + "loss": 0.0202, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01107854750007391, + "rkd_loss_weight": 0.5495016515254975, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38997324705123904, + "rkd_teacher_entropy": 3.745179319381714, + "rkd_teacher_similarity_mean": 0.7855203151702881, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13520 + }, + { + "epoch": 3.936572592377073, + "flow_matching_loss": 0.015483268164098262, + "grad_norm": 0.11539582163095474, + "learning_rate": 1.8635780223429565e-05, + "loss": 0.0214, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010746293794363737, + "rkd_loss_weight": 0.5491683065891266, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38539025783538816, + "rkd_teacher_entropy": 3.7360085010528565, + "rkd_teacher_similarity_mean": 0.78092360496521, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13530 + }, + { + "epoch": 3.9394821064882164, + "flow_matching_loss": 0.015622918400913477, + "grad_norm": 0.1833898425102234, + "learning_rate": 1.8619736383351017e-05, + "loss": 0.0212, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010086444718763233, + "rkd_loss_weight": 0.5488349616527557, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3900896102190018, + "rkd_teacher_entropy": 3.7448859930038454, + "rkd_teacher_similarity_mean": 0.7854043543338776, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13540 + }, + { + "epoch": 3.94239162059936, + "flow_matching_loss": 0.01723289843648672, + "grad_norm": 0.11615894734859467, + "learning_rate": 1.8603688144951454e-05, + "loss": 0.0221, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00885753594338894, + "rkd_loss_weight": 0.5485016167163849, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3921861618757248, + "rkd_teacher_entropy": 3.758550834655762, + "rkd_teacher_similarity_mean": 0.7862195253372193, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13550 + }, + { + "epoch": 3.9453011347105034, + "flow_matching_loss": 0.016588410641998053, + "grad_norm": 0.12308263778686523, + "learning_rate": 1.8587635527730983e-05, + "loss": 0.023, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01169573999941349, + "rkd_loss_weight": 0.548168271780014, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3845218449831009, + "rkd_teacher_entropy": 3.743248152732849, + "rkd_teacher_similarity_mean": 0.7780365109443664, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13560 + }, + { + "epoch": 3.9482106488216466, + "flow_matching_loss": 0.01804502885788679, + "grad_norm": 0.11426875740289688, + "learning_rate": 1.8571578551195056e-05, + "loss": 0.0234, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009858380863443017, + "rkd_loss_weight": 0.5478349328041077, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39360980689525604, + "rkd_teacher_entropy": 3.771501302719116, + "rkd_teacher_similarity_mean": 0.7904372990131379, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13570 + }, + { + "epoch": 3.9511201629327903, + "flow_matching_loss": 0.01673600561916828, + "grad_norm": 0.13694262504577637, + "learning_rate": 1.855551723485441e-05, + "loss": 0.0223, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010164478141814471, + "rkd_loss_weight": 0.5475015819072724, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3835152149200439, + "rkd_teacher_entropy": 3.729580545425415, + "rkd_teacher_similarity_mean": 0.7785778105258941, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13580 + }, + { + "epoch": 3.9540296770439336, + "flow_matching_loss": 0.01887253802269697, + "grad_norm": 0.14903588593006134, + "learning_rate": 1.8539451598225066e-05, + "loss": 0.0254, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01201608618721366, + "rkd_loss_weight": 0.5471682369709014, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38576588332653045, + "rkd_teacher_entropy": 3.731005072593689, + "rkd_teacher_similarity_mean": 0.7795698702335357, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13590 + }, + { + "epoch": 3.956939191155077, + "flow_matching_loss": 0.015227486938238144, + "grad_norm": 0.12837618589401245, + "learning_rate": 1.8523381660828276e-05, + "loss": 0.0207, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010071614850312471, + "rkd_loss_weight": 0.5468348979949951, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39779540002346037, + "rkd_teacher_entropy": 3.7745291709899904, + "rkd_teacher_similarity_mean": 0.7908002734184265, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13600 + }, + { + "epoch": 3.9598487052662206, + "flow_matching_loss": 0.01427264353260398, + "grad_norm": 0.16140338778495789, + "learning_rate": 1.850730744219054e-05, + "loss": 0.02, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010518785798922181, + "rkd_loss_weight": 0.5465015470981598, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3797468185424805, + "rkd_teacher_entropy": 3.723392295837402, + "rkd_teacher_similarity_mean": 0.7736671030521393, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13610 + }, + { + "epoch": 3.962758219377364, + "flow_matching_loss": 0.014924184139817952, + "grad_norm": 0.1446857750415802, + "learning_rate": 1.849122896184354e-05, + "loss": 0.0215, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012097471114248037, + "rkd_loss_weight": 0.5461682081222534, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4003534972667694, + "rkd_teacher_entropy": 3.7784716844558717, + "rkd_teacher_similarity_mean": 0.7941587865352631, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13620 + }, + { + "epoch": 3.9656677334885075, + "flow_matching_loss": 0.014954723510891199, + "grad_norm": 0.147629052400589, + "learning_rate": 1.8475146239324146e-05, + "loss": 0.0199, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008981773117557167, + "rkd_loss_weight": 0.5458348631858826, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3968439906835556, + "rkd_teacher_entropy": 3.769197773933411, + "rkd_teacher_similarity_mean": 0.7903623461723328, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13630 + }, + { + "epoch": 3.968577247599651, + "flow_matching_loss": 0.014593501202762127, + "grad_norm": 0.1273132711648941, + "learning_rate": 1.8459059294174396e-05, + "loss": 0.0199, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009723572665825487, + "rkd_loss_weight": 0.5455015122890472, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39598774909973145, + "rkd_teacher_entropy": 3.77399480342865, + "rkd_teacher_similarity_mean": 0.7897404253482818, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13640 + }, + { + "epoch": 3.9714867617107945, + "flow_matching_loss": 0.0169042787514627, + "grad_norm": 0.16176992654800415, + "learning_rate": 1.844296814594143e-05, + "loss": 0.0224, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010064935777336359, + "rkd_loss_weight": 0.5451681792736054, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39188025891780853, + "rkd_teacher_entropy": 3.7575806617736816, + "rkd_teacher_similarity_mean": 0.7858556389808655, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13650 + }, + { + "epoch": 3.9743962758219378, + "flow_matching_loss": 0.014241230208426714, + "grad_norm": 0.1294633150100708, + "learning_rate": 1.8426872814177516e-05, + "loss": 0.0196, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009923039935529232, + "rkd_loss_weight": 0.5448348224163055, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38742059767246245, + "rkd_teacher_entropy": 3.7322062730789183, + "rkd_teacher_similarity_mean": 0.7817445218563079, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13660 + }, + { + "epoch": 3.977305789933081, + "flow_matching_loss": 0.01603022972121835, + "grad_norm": 0.12823808193206787, + "learning_rate": 1.8410773318440003e-05, + "loss": 0.0229, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012567473109811544, + "rkd_loss_weight": 0.5445014834403992, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39100308120250704, + "rkd_teacher_entropy": 3.753385639190674, + "rkd_teacher_similarity_mean": 0.784778606891632, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13670 + }, + { + "epoch": 3.9802153040442247, + "flow_matching_loss": 0.019087345339357853, + "grad_norm": 0.15023000538349152, + "learning_rate": 1.839466967829129e-05, + "loss": 0.0249, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01072408314794302, + "rkd_loss_weight": 0.5441681444644928, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3884535312652588, + "rkd_teacher_entropy": 3.7410848140716553, + "rkd_teacher_similarity_mean": 0.7800574004650116, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13680 + }, + { + "epoch": 3.983124818155368, + "flow_matching_loss": 0.016445128247141838, + "grad_norm": 0.1763187199831009, + "learning_rate": 1.8378561913298822e-05, + "loss": 0.0222, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010568860266357661, + "rkd_loss_weight": 0.543834787607193, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3835393488407135, + "rkd_teacher_entropy": 3.743178462982178, + "rkd_teacher_similarity_mean": 0.7765908300876617, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13690 + }, + { + "epoch": 3.9860343322665113, + "flow_matching_loss": 0.018242126423865558, + "grad_norm": 0.1309187263250351, + "learning_rate": 1.8362450043035053e-05, + "loss": 0.0236, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00980870989151299, + "rkd_loss_weight": 0.5435014545917511, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38631761372089385, + "rkd_teacher_entropy": 3.735590934753418, + "rkd_teacher_similarity_mean": 0.780103611946106, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13700 + }, + { + "epoch": 3.988943846377655, + "flow_matching_loss": 0.016791431978344918, + "grad_norm": 0.19398818910121918, + "learning_rate": 1.834633408707742e-05, + "loss": 0.0226, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010780706442892552, + "rkd_loss_weight": 0.5431681036949157, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3834777593612671, + "rkd_teacher_entropy": 3.720545196533203, + "rkd_teacher_similarity_mean": 0.7752135992050171, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13710 + }, + { + "epoch": 3.9918533604887982, + "flow_matching_loss": 0.014491152204573154, + "grad_norm": 0.17591214179992676, + "learning_rate": 1.8330214065008336e-05, + "loss": 0.0214, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01275276644155383, + "rkd_loss_weight": 0.5428347587585449, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39004749059677124, + "rkd_teacher_entropy": 3.7434105396270754, + "rkd_teacher_similarity_mean": 0.7826881885528565, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13720 + }, + { + "epoch": 3.994762874599942, + "flow_matching_loss": 0.018950939550995825, + "grad_norm": 0.14899256825447083, + "learning_rate": 1.831408999641514e-05, + "loss": 0.0239, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009205890679731965, + "rkd_loss_weight": 0.5425014197826385, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3884757846593857, + "rkd_teacher_entropy": 3.7533122539520263, + "rkd_teacher_similarity_mean": 0.7845056414604187, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13730 + }, + { + "epoch": 3.997672388711085, + "flow_matching_loss": 0.01535683823749423, + "grad_norm": 0.12835052609443665, + "learning_rate": 1.8297961900890107e-05, + "loss": 0.0199, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008414192544296383, + "rkd_loss_weight": 0.5421680688858033, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39001300632953645, + "rkd_teacher_entropy": 3.7511272192001344, + "rkd_teacher_similarity_mean": 0.7864390790462494, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13740 + }, + { + "epoch": 4.000581902822229, + "flow_matching_loss": 0.018905821815133096, + "grad_norm": 0.1558203399181366, + "learning_rate": 1.8281829798030385e-05, + "loss": 0.0238, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00904401894658804, + "rkd_loss_weight": 0.5418347299098969, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3895399808883667, + "rkd_teacher_entropy": 3.7479530811309814, + "rkd_teacher_similarity_mean": 0.7851445138454437, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13750 + }, + { + "epoch": 4.003491416933372, + "flow_matching_loss": 0.01799536095932126, + "grad_norm": 0.17875467240810394, + "learning_rate": 1.8265693707438002e-05, + "loss": 0.0233, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00976851056329906, + "rkd_loss_weight": 0.541501384973526, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38389774560928347, + "rkd_teacher_entropy": 3.742908263206482, + "rkd_teacher_similarity_mean": 0.7793477892875671, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13760 + }, + { + "epoch": 4.0064009310445154, + "flow_matching_loss": 0.013376928912475706, + "grad_norm": 0.14787405729293823, + "learning_rate": 1.824955364871982e-05, + "loss": 0.0185, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009547748137265444, + "rkd_loss_weight": 0.5411680340766907, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39705370366573334, + "rkd_teacher_entropy": 3.764028716087341, + "rkd_teacher_similarity_mean": 0.7924373686313629, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13770 + }, + { + "epoch": 4.009310445155659, + "flow_matching_loss": 0.01494171777740121, + "grad_norm": 0.1389983892440796, + "learning_rate": 1.8233409641487553e-05, + "loss": 0.0201, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00945160910487175, + "rkd_loss_weight": 0.5408346951007843, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38768757283687594, + "rkd_teacher_entropy": 3.737738299369812, + "rkd_teacher_similarity_mean": 0.7806459784507751, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13780 + }, + { + "epoch": 4.012219959266803, + "flow_matching_loss": 0.015288703050464392, + "grad_norm": 0.12233857810497284, + "learning_rate": 1.8217261705357677e-05, + "loss": 0.0202, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009140941873192787, + "rkd_loss_weight": 0.5405013501644135, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3896574079990387, + "rkd_teacher_entropy": 3.7496530771255494, + "rkd_teacher_similarity_mean": 0.7841566205024719, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13790 + }, + { + "epoch": 4.015129473377946, + "flow_matching_loss": 0.014464330486953258, + "grad_norm": 0.11901683360338211, + "learning_rate": 1.8201109859951462e-05, + "loss": 0.0201, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010465431353077292, + "rkd_loss_weight": 0.5401680052280426, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3805020600557327, + "rkd_teacher_entropy": 3.7161735773086546, + "rkd_teacher_similarity_mean": 0.7753985106945038, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13800 + }, + { + "epoch": 4.018038987489089, + "flow_matching_loss": 0.015110189374536277, + "grad_norm": 0.12777332961559296, + "learning_rate": 1.818495412489492e-05, + "loss": 0.0207, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01039682812988758, + "rkd_loss_weight": 0.5398346602916717, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38878165781497953, + "rkd_teacher_entropy": 3.7533700227737428, + "rkd_teacher_similarity_mean": 0.7833024621009826, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13810 + }, + { + "epoch": 4.020948501600233, + "flow_matching_loss": 0.016707618534564973, + "grad_norm": 0.14299792051315308, + "learning_rate": 1.8168794519818802e-05, + "loss": 0.022, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009818619629368187, + "rkd_loss_weight": 0.5395013153553009, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38438917100429537, + "rkd_teacher_entropy": 3.7395177841186524, + "rkd_teacher_similarity_mean": 0.7775623619556427, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13820 + }, + { + "epoch": 4.023858015711376, + "flow_matching_loss": 0.018030559457838536, + "grad_norm": 0.12192855775356293, + "learning_rate": 1.8152631064358537e-05, + "loss": 0.0227, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008734193723648786, + "rkd_loss_weight": 0.53916797041893, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38642307221889494, + "rkd_teacher_entropy": 3.740230989456177, + "rkd_teacher_similarity_mean": 0.7803746640682221, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13830 + }, + { + "epoch": 4.02676752982252, + "flow_matching_loss": 0.018067200388759374, + "grad_norm": 0.12706208229064941, + "learning_rate": 1.8136463778154258e-05, + "loss": 0.0232, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009566375939175486, + "rkd_loss_weight": 0.5388346314430237, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3810995936393738, + "rkd_teacher_entropy": 3.7290627717971803, + "rkd_teacher_similarity_mean": 0.7759389698505401, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13840 + }, + { + "epoch": 4.029677043933663, + "flow_matching_loss": 0.015214463975280524, + "grad_norm": 0.12932102382183075, + "learning_rate": 1.812029268085074e-05, + "loss": 0.0211, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010849865805357694, + "rkd_loss_weight": 0.5385012805461884, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3823292195796967, + "rkd_teacher_entropy": 3.7184029102325438, + "rkd_teacher_similarity_mean": 0.7766475737094879, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13850 + }, + { + "epoch": 4.032586558044806, + "flow_matching_loss": 0.013692414481192827, + "grad_norm": 0.14172078669071198, + "learning_rate": 1.8104117792097387e-05, + "loss": 0.0193, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010451259277760982, + "rkd_loss_weight": 0.5381679356098175, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3952883183956146, + "rkd_teacher_entropy": 3.7583162784576416, + "rkd_teacher_similarity_mean": 0.7885973334312439, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13860 + }, + { + "epoch": 4.03549607215595, + "flow_matching_loss": 0.018613770231604577, + "grad_norm": 0.1435033231973648, + "learning_rate": 1.808793913154822e-05, + "loss": 0.0243, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010596063733100892, + "rkd_loss_weight": 0.5378345966339111, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38891597986221316, + "rkd_teacher_entropy": 3.7443172216415403, + "rkd_teacher_similarity_mean": 0.7820828378200531, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13870 + }, + { + "epoch": 4.038405586267094, + "flow_matching_loss": 0.01715138740837574, + "grad_norm": 0.14882609248161316, + "learning_rate": 1.8071756718861838e-05, + "loss": 0.0221, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009244820568710566, + "rkd_loss_weight": 0.5375012457370758, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38167236745357513, + "rkd_teacher_entropy": 3.7223657369613647, + "rkd_teacher_similarity_mean": 0.7759829640388489, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13880 + }, + { + "epoch": 4.041315100378237, + "flow_matching_loss": 0.017302628187462688, + "grad_norm": 0.1225898414850235, + "learning_rate": 1.805557057370139e-05, + "loss": 0.0231, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010792913101613521, + "rkd_loss_weight": 0.5371679067611694, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38425567746162415, + "rkd_teacher_entropy": 3.734106683731079, + "rkd_teacher_similarity_mean": 0.77734734416008, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13890 + }, + { + "epoch": 4.04422461448938, + "flow_matching_loss": 0.016850190702825785, + "grad_norm": 0.13682669401168823, + "learning_rate": 1.8039380715734574e-05, + "loss": 0.0222, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009884926024824381, + "rkd_loss_weight": 0.5368345618247986, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3884776532649994, + "rkd_teacher_entropy": 3.7475330352783205, + "rkd_teacher_similarity_mean": 0.780592429637909, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13900 + }, + { + "epoch": 4.047134128600524, + "flow_matching_loss": 0.016194915771484374, + "grad_norm": 0.17196024954319, + "learning_rate": 1.802318716463359e-05, + "loss": 0.0209, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008830292895436288, + "rkd_loss_weight": 0.5365012109279632, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.390071713924408, + "rkd_teacher_entropy": 3.7476386547088625, + "rkd_teacher_similarity_mean": 0.7828845083713531, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13910 + }, + { + "epoch": 4.0500436427116675, + "flow_matching_loss": 0.01575913904234767, + "grad_norm": 0.13204129040241241, + "learning_rate": 1.8006989940075138e-05, + "loss": 0.0209, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00949929617345333, + "rkd_loss_weight": 0.5361678779125214, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39475836157798766, + "rkd_teacher_entropy": 3.7637684106826783, + "rkd_teacher_similarity_mean": 0.7872779071331024, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13920 + }, + { + "epoch": 4.05295315682281, + "flow_matching_loss": 0.013506767526268958, + "grad_norm": 0.12406972795724869, + "learning_rate": 1.7990789061740368e-05, + "loss": 0.0177, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007821542955935002, + "rkd_loss_weight": 0.535834527015686, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3861511737108231, + "rkd_teacher_entropy": 3.7335283041000364, + "rkd_teacher_similarity_mean": 0.7792687892913819, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13930 + }, + { + "epoch": 4.055862670933954, + "flow_matching_loss": 0.014694240503013134, + "grad_norm": 0.11940757930278778, + "learning_rate": 1.7974584549314875e-05, + "loss": 0.0201, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010049404576420784, + "rkd_loss_weight": 0.5355011820793152, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3830118626356125, + "rkd_teacher_entropy": 3.738807511329651, + "rkd_teacher_similarity_mean": 0.7792766034603119, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13940 + }, + { + "epoch": 4.058772185045098, + "flow_matching_loss": 0.01666394863277674, + "grad_norm": 0.11442584544420242, + "learning_rate": 1.7958376422488672e-05, + "loss": 0.0225, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010964788030833006, + "rkd_loss_weight": 0.5351678431034088, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.386908620595932, + "rkd_teacher_entropy": 3.752407741546631, + "rkd_teacher_similarity_mean": 0.7824332535266876, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13950 + }, + { + "epoch": 4.0616816991562406, + "flow_matching_loss": 0.015607030410319567, + "grad_norm": 0.11395110934972763, + "learning_rate": 1.7942164700956156e-05, + "loss": 0.0207, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009564681770280003, + "rkd_loss_weight": 0.5348344922065735, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3842126190662384, + "rkd_teacher_entropy": 3.721808385848999, + "rkd_teacher_similarity_mean": 0.7778428137302399, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13960 + }, + { + "epoch": 4.064591213267384, + "flow_matching_loss": 0.017671321611851455, + "grad_norm": 0.14550895988941193, + "learning_rate": 1.79259494044161e-05, + "loss": 0.0226, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009302606526762246, + "rkd_loss_weight": 0.5345011532306672, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.390451991558075, + "rkd_teacher_entropy": 3.752815079689026, + "rkd_teacher_similarity_mean": 0.783705860376358, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13970 + }, + { + "epoch": 4.067500727378528, + "flow_matching_loss": 0.014958348404616118, + "grad_norm": 0.12281741946935654, + "learning_rate": 1.7909730552571617e-05, + "loss": 0.0205, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010310828872025012, + "rkd_loss_weight": 0.5341678082942962, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38343391120433806, + "rkd_teacher_entropy": 3.723976397514343, + "rkd_teacher_similarity_mean": 0.7780095875263214, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13980 + }, + { + "epoch": 4.070410241489671, + "flow_matching_loss": 0.015811497718095778, + "grad_norm": 0.20535443723201752, + "learning_rate": 1.7893508165130143e-05, + "loss": 0.0213, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010207143612205982, + "rkd_loss_weight": 0.5338344573974609, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3929052114486694, + "rkd_teacher_entropy": 3.769338774681091, + "rkd_teacher_similarity_mean": 0.7870662748813629, + "rkd_vlm_temp": 0.10000000149011612, + "step": 13990 + }, + { + "epoch": 4.0733197556008145, + "flow_matching_loss": 0.01322949891909957, + "grad_norm": 0.11442889273166656, + "learning_rate": 1.787728226180341e-05, + "loss": 0.0188, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010431859456002713, + "rkd_loss_weight": 0.5335011184215546, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38985384404659273, + "rkd_teacher_entropy": 3.749570941925049, + "rkd_teacher_similarity_mean": 0.783407062292099, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14000 + }, + { + "epoch": 4.076229269711958, + "flow_matching_loss": 0.0162186480127275, + "grad_norm": 0.1467273086309433, + "learning_rate": 1.786105286230742e-05, + "loss": 0.0211, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00911079766228795, + "rkd_loss_weight": 0.5331677734851837, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3888168752193451, + "rkd_teacher_entropy": 3.745684099197388, + "rkd_teacher_similarity_mean": 0.7814636886119842, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14010 + }, + { + "epoch": 4.079138783823102, + "flow_matching_loss": 0.01628327025100589, + "grad_norm": 0.11882033199071884, + "learning_rate": 1.784481998636242e-05, + "loss": 0.0213, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009333521081134676, + "rkd_loss_weight": 0.5328344285488129, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3959232747554779, + "rkd_teacher_entropy": 3.768633818626404, + "rkd_teacher_similarity_mean": 0.790285462141037, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14020 + }, + { + "epoch": 4.082048297934245, + "flow_matching_loss": 0.014368782751262188, + "grad_norm": 0.13446861505508423, + "learning_rate": 1.782858365369289e-05, + "loss": 0.0192, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009098028857260942, + "rkd_loss_weight": 0.532501083612442, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.387779700756073, + "rkd_teacher_entropy": 3.7422871589660645, + "rkd_teacher_similarity_mean": 0.7809910595417022, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14030 + }, + { + "epoch": 4.084957812045388, + "flow_matching_loss": 0.013795493077486754, + "grad_norm": 0.11432575434446335, + "learning_rate": 1.78123438840275e-05, + "loss": 0.0187, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009169285604730248, + "rkd_loss_weight": 0.5321677386760711, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3922613263130188, + "rkd_teacher_entropy": 3.750457739830017, + "rkd_teacher_similarity_mean": 0.784782099723816, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14040 + }, + { + "epoch": 4.087867326156532, + "flow_matching_loss": 0.01848144056275487, + "grad_norm": 0.13935355842113495, + "learning_rate": 1.779610069709911e-05, + "loss": 0.023, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008494235901162028, + "rkd_loss_weight": 0.5318343937397003, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38809171319007874, + "rkd_teacher_entropy": 3.73672411441803, + "rkd_teacher_similarity_mean": 0.7807168245315552, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14050 + }, + { + "epoch": 4.090776840267675, + "flow_matching_loss": 0.01599872186779976, + "grad_norm": 0.12214263528585434, + "learning_rate": 1.7779854112644713e-05, + "loss": 0.0215, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010355788562446833, + "rkd_loss_weight": 0.531501054763794, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38639668822288514, + "rkd_teacher_entropy": 3.7405397415161135, + "rkd_teacher_similarity_mean": 0.7808309853076935, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14060 + }, + { + "epoch": 4.093686354378819, + "flow_matching_loss": 0.012821591831743718, + "grad_norm": 0.14130799472332, + "learning_rate": 1.7763604150405446e-05, + "loss": 0.0185, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010665113478899002, + "rkd_loss_weight": 0.5311677038669587, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.391752290725708, + "rkd_teacher_entropy": 3.7572280406951903, + "rkd_teacher_similarity_mean": 0.7860895395278931, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14070 + }, + { + "epoch": 4.096595868489962, + "flow_matching_loss": 0.014895072486251593, + "grad_norm": 0.1554780751466751, + "learning_rate": 1.7747350830126548e-05, + "loss": 0.0202, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010033432301133871, + "rkd_loss_weight": 0.5308343589305877, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38710587918758393, + "rkd_teacher_entropy": 3.7491020441055296, + "rkd_teacher_similarity_mean": 0.7805100619792938, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14080 + }, + { + "epoch": 4.099505382601105, + "flow_matching_loss": 0.01501210844144225, + "grad_norm": 0.12579144537448883, + "learning_rate": 1.7731094171557332e-05, + "loss": 0.0207, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010793511103838683, + "rkd_loss_weight": 0.5305010199546814, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39420745372772215, + "rkd_teacher_entropy": 3.769609212875366, + "rkd_teacher_similarity_mean": 0.7875656068325043, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14090 + }, + { + "epoch": 4.102414896712249, + "flow_matching_loss": 0.014695442561060191, + "grad_norm": 0.14418914914131165, + "learning_rate": 1.771483419445117e-05, + "loss": 0.0201, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010237302444875241, + "rkd_loss_weight": 0.5301676690578461, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38290072083473203, + "rkd_teacher_entropy": 3.7204938411712645, + "rkd_teacher_similarity_mean": 0.7765249729156494, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14100 + }, + { + "epoch": 4.105324410823393, + "flow_matching_loss": 0.015200815629214048, + "grad_norm": 0.16046816110610962, + "learning_rate": 1.7698570918565474e-05, + "loss": 0.0202, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00946721127256751, + "rkd_loss_weight": 0.5298343300819397, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39149667918682096, + "rkd_teacher_entropy": 3.7444339036941527, + "rkd_teacher_similarity_mean": 0.7873354911804199, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14110 + }, + { + "epoch": 4.108233924934536, + "flow_matching_loss": 0.016616802662611008, + "grad_norm": 0.11920808255672455, + "learning_rate": 1.768230436366165e-05, + "loss": 0.0211, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008490697760134936, + "rkd_loss_weight": 0.5295009851455689, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38716049790382384, + "rkd_teacher_entropy": 3.748089909553528, + "rkd_teacher_similarity_mean": 0.7835131585597992, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14120 + }, + { + "epoch": 4.111143439045679, + "flow_matching_loss": 0.016015436500310898, + "grad_norm": 0.11240214109420776, + "learning_rate": 1.7666034549505102e-05, + "loss": 0.0212, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009812888130545616, + "rkd_loss_weight": 0.5291676342487335, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3900479435920715, + "rkd_teacher_entropy": 3.7457071781158446, + "rkd_teacher_similarity_mean": 0.7846909761428833, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14130 + }, + { + "epoch": 4.114052953156823, + "flow_matching_loss": 0.014424726180732251, + "grad_norm": 0.13077332079410553, + "learning_rate": 1.7649761495865185e-05, + "loss": 0.0198, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010077047068625689, + "rkd_loss_weight": 0.5288343012332917, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3849096864461899, + "rkd_teacher_entropy": 3.733624982833862, + "rkd_teacher_similarity_mean": 0.7787589192390442, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14140 + }, + { + "epoch": 4.1169624672679666, + "flow_matching_loss": 0.01374445129185915, + "grad_norm": 0.1677834689617157, + "learning_rate": 1.7633485222515188e-05, + "loss": 0.0194, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01070856163278222, + "rkd_loss_weight": 0.5285009443759918, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39180195331573486, + "rkd_teacher_entropy": 3.748873567581177, + "rkd_teacher_similarity_mean": 0.7864714443683625, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14150 + }, + { + "epoch": 4.119871981379109, + "flow_matching_loss": 0.015603217668831349, + "grad_norm": 0.14102190732955933, + "learning_rate": 1.761720574923233e-05, + "loss": 0.0218, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011638250853866338, + "rkd_loss_weight": 0.5281676054000854, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38276738822460177, + "rkd_teacher_entropy": 3.73365261554718, + "rkd_teacher_similarity_mean": 0.7785238564014435, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14160 + }, + { + "epoch": 4.122781495490253, + "flow_matching_loss": 0.016318903490900993, + "grad_norm": 0.18068024516105652, + "learning_rate": 1.7600923095797704e-05, + "loss": 0.0207, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00825060815550387, + "rkd_loss_weight": 0.5278342664241791, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3870002835988998, + "rkd_teacher_entropy": 3.7461180686950684, + "rkd_teacher_similarity_mean": 0.7817069947719574, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14170 + }, + { + "epoch": 4.125691009601397, + "flow_matching_loss": 0.015400339011102914, + "grad_norm": 0.122889444231987, + "learning_rate": 1.758463728199626e-05, + "loss": 0.0203, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009295509709045292, + "rkd_loss_weight": 0.5275009095668792, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3836555272340775, + "rkd_teacher_entropy": 3.7288311958312987, + "rkd_teacher_similarity_mean": 0.7801561772823333, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14180 + }, + { + "epoch": 4.12860052371254, + "flow_matching_loss": 0.016918944846838712, + "grad_norm": 0.13968345522880554, + "learning_rate": 1.7568348327616803e-05, + "loss": 0.0221, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009779861336573959, + "rkd_loss_weight": 0.5271675765514374, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38524084389209745, + "rkd_teacher_entropy": 3.7473979949951173, + "rkd_teacher_similarity_mean": 0.7822755813598633, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14190 + }, + { + "epoch": 4.131510037823683, + "flow_matching_loss": 0.015321470983326434, + "grad_norm": 0.11374509334564209, + "learning_rate": 1.755205625245195e-05, + "loss": 0.0203, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009414679044857622, + "rkd_loss_weight": 0.526834225654602, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3862040251493454, + "rkd_teacher_entropy": 3.748544454574585, + "rkd_teacher_similarity_mean": 0.781298702955246, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14200 + }, + { + "epoch": 4.134419551934827, + "flow_matching_loss": 0.017215089034289123, + "grad_norm": 0.15482109785079956, + "learning_rate": 1.7535761076298104e-05, + "loss": 0.0221, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009204978495836258, + "rkd_loss_weight": 0.5265008807182312, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.4017797350883484, + "rkd_teacher_entropy": 3.7911744117736816, + "rkd_teacher_similarity_mean": 0.7981953740119934, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14210 + }, + { + "epoch": 4.137329066045971, + "flow_matching_loss": 0.01672039907425642, + "grad_norm": 0.14624837040901184, + "learning_rate": 1.7519462818955448e-05, + "loss": 0.022, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009981325641274453, + "rkd_loss_weight": 0.5261675417423248, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38296698331832885, + "rkd_teacher_entropy": 3.7297048807144164, + "rkd_teacher_similarity_mean": 0.7775816082954407, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14220 + }, + { + "epoch": 4.1402385801571135, + "flow_matching_loss": 0.016368712484836578, + "grad_norm": 0.1685892939567566, + "learning_rate": 1.7503161500227896e-05, + "loss": 0.021, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00886894748546183, + "rkd_loss_weight": 0.5258341908454895, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3782899022102356, + "rkd_teacher_entropy": 3.7135255336761475, + "rkd_teacher_similarity_mean": 0.77260822057724, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14230 + }, + { + "epoch": 4.143148094268257, + "flow_matching_loss": 0.015018103178590536, + "grad_norm": 0.14830726385116577, + "learning_rate": 1.7486857139923093e-05, + "loss": 0.0204, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010235450090840459, + "rkd_loss_weight": 0.5255008518695832, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38218350410461427, + "rkd_teacher_entropy": 3.7281824350357056, + "rkd_teacher_similarity_mean": 0.7785516679286957, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14240 + }, + { + "epoch": 4.146057608379401, + "flow_matching_loss": 0.016797901690006257, + "grad_norm": 0.1328965723514557, + "learning_rate": 1.747054975785236e-05, + "loss": 0.0226, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010999525664374232, + "rkd_loss_weight": 0.5251675069332122, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38598203361034394, + "rkd_teacher_entropy": 3.7322120904922484, + "rkd_teacher_similarity_mean": 0.780205637216568, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14250 + }, + { + "epoch": 4.148967122490544, + "flow_matching_loss": 0.015074805915355682, + "grad_norm": 0.15606920421123505, + "learning_rate": 1.745423937383072e-05, + "loss": 0.0206, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010553135862573981, + "rkd_loss_weight": 0.524834156036377, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3898384839296341, + "rkd_teacher_entropy": 3.7524853706359864, + "rkd_teacher_similarity_mean": 0.7828916132450103, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14260 + }, + { + "epoch": 4.1518766366016875, + "flow_matching_loss": 0.016631006728857756, + "grad_norm": 0.1408776342868805, + "learning_rate": 1.7437926007676825e-05, + "loss": 0.0219, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009997036214917899, + "rkd_loss_weight": 0.5245008170604706, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3848165303468704, + "rkd_teacher_entropy": 3.728816604614258, + "rkd_teacher_similarity_mean": 0.7797871232032776, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14270 + }, + { + "epoch": 4.154786150712831, + "flow_matching_loss": 0.015089895576238632, + "grad_norm": 0.14198635518550873, + "learning_rate": 1.742160967921294e-05, + "loss": 0.0204, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010085524944588541, + "rkd_loss_weight": 0.5241674721240998, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38949035108089447, + "rkd_teacher_entropy": 3.765627861022949, + "rkd_teacher_similarity_mean": 0.7857569396495819, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14280 + }, + { + "epoch": 4.157695664823974, + "flow_matching_loss": 0.014707687450572849, + "grad_norm": 0.13207541406154633, + "learning_rate": 1.7405290408264954e-05, + "loss": 0.0196, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00940228346735239, + "rkd_loss_weight": 0.5238341271877289, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3872263222932816, + "rkd_teacher_entropy": 3.7414490222930907, + "rkd_teacher_similarity_mean": 0.7825487613677978, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14290 + }, + { + "epoch": 4.160605178935118, + "flow_matching_loss": 0.01622771443799138, + "grad_norm": 0.13388162851333618, + "learning_rate": 1.738896821466232e-05, + "loss": 0.0207, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00846427450887859, + "rkd_loss_weight": 0.523500782251358, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3935294449329376, + "rkd_teacher_entropy": 3.7620716094970703, + "rkd_teacher_similarity_mean": 0.787447714805603, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14300 + }, + { + "epoch": 4.163514693046261, + "flow_matching_loss": 0.018212629947811365, + "grad_norm": 0.1317088007926941, + "learning_rate": 1.737264311823804e-05, + "loss": 0.0235, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010082203894853592, + "rkd_loss_weight": 0.5231674373149872, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38825843930244447, + "rkd_teacher_entropy": 3.7552236318588257, + "rkd_teacher_similarity_mean": 0.7832855165004731, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14310 + }, + { + "epoch": 4.166424207157405, + "flow_matching_loss": 0.015653409902006386, + "grad_norm": 0.1579591929912567, + "learning_rate": 1.7356315138828645e-05, + "loss": 0.0201, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008510035648941993, + "rkd_loss_weight": 0.5228340923786163, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3855924516916275, + "rkd_teacher_entropy": 3.746517705917358, + "rkd_teacher_similarity_mean": 0.7832855522632599, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14320 + }, + { + "epoch": 4.169333721268548, + "flow_matching_loss": 0.01412602597847581, + "grad_norm": 0.14979396760463715, + "learning_rate": 1.7339984296274174e-05, + "loss": 0.0199, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011021946091204882, + "rkd_loss_weight": 0.52250075340271, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3870076179504395, + "rkd_teacher_entropy": 3.751571536064148, + "rkd_teacher_similarity_mean": 0.7829826772212982, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14330 + }, + { + "epoch": 4.172243235379692, + "flow_matching_loss": 0.017820207122713327, + "grad_norm": 0.138068288564682, + "learning_rate": 1.7323650610418134e-05, + "loss": 0.0223, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008617088058963418, + "rkd_loss_weight": 0.5221674025058747, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37966109812259674, + "rkd_teacher_entropy": 3.7207091093063354, + "rkd_teacher_similarity_mean": 0.7744242370128631, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14340 + }, + { + "epoch": 4.175152749490835, + "flow_matching_loss": 0.013887672312557697, + "grad_norm": 0.11490844190120697, + "learning_rate": 1.730731410110749e-05, + "loss": 0.0182, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008313080808147789, + "rkd_loss_weight": 0.5218340575695037, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38998137712478637, + "rkd_teacher_entropy": 3.749375891685486, + "rkd_teacher_similarity_mean": 0.7843608140945435, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14350 + }, + { + "epoch": 4.178062263601978, + "flow_matching_loss": 0.01535833002999425, + "grad_norm": 0.10345927625894547, + "learning_rate": 1.7290974788192655e-05, + "loss": 0.0206, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010032167471945286, + "rkd_loss_weight": 0.5215007245540619, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38553199768066404, + "rkd_teacher_entropy": 3.743841028213501, + "rkd_teacher_similarity_mean": 0.780822879076004, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14360 + }, + { + "epoch": 4.180971777713122, + "flow_matching_loss": 0.014508702978491782, + "grad_norm": 0.09786123037338257, + "learning_rate": 1.727463269152743e-05, + "loss": 0.0197, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00994429662823677, + "rkd_loss_weight": 0.5211673676967621, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3871102213859558, + "rkd_teacher_entropy": 3.744387722015381, + "rkd_teacher_similarity_mean": 0.7817184150218963, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14370 + }, + { + "epoch": 4.183881291824266, + "flow_matching_loss": 0.015308840107172728, + "grad_norm": 0.14013902842998505, + "learning_rate": 1.7258287830968996e-05, + "loss": 0.0205, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009892114531248808, + "rkd_loss_weight": 0.5208340287208557, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39172687828540803, + "rkd_teacher_entropy": 3.7598848819732664, + "rkd_teacher_similarity_mean": 0.78806933760643, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14380 + }, + { + "epoch": 4.186790805935408, + "flow_matching_loss": 0.017739410046488047, + "grad_norm": 0.13013845682144165, + "learning_rate": 1.7241940226377908e-05, + "loss": 0.0238, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011578104551881551, + "rkd_loss_weight": 0.5205006897449493, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38790206909179686, + "rkd_teacher_entropy": 3.7604400873184205, + "rkd_teacher_similarity_mean": 0.7854138791561127, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14390 + }, + { + "epoch": 4.189700320046552, + "flow_matching_loss": 0.015465872455388307, + "grad_norm": 0.15112583339214325, + "learning_rate": 1.7225589897618043e-05, + "loss": 0.02, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008709522197023034, + "rkd_loss_weight": 0.5201673328876495, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3969861000776291, + "rkd_teacher_entropy": 3.775694465637207, + "rkd_teacher_similarity_mean": 0.7934878587722778, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14400 + }, + { + "epoch": 4.192609834157696, + "flow_matching_loss": 0.013590610586106777, + "grad_norm": 0.11525988578796387, + "learning_rate": 1.7209236864556588e-05, + "loss": 0.0183, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009132977481931448, + "rkd_loss_weight": 0.5198339998722077, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38364542126655576, + "rkd_teacher_entropy": 3.7331717729568483, + "rkd_teacher_similarity_mean": 0.7800330638885498, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14410 + }, + { + "epoch": 4.195519348268839, + "flow_matching_loss": 0.013134629838168621, + "grad_norm": 0.09323695302009583, + "learning_rate": 1.719288114706403e-05, + "loss": 0.0183, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009944321168586612, + "rkd_loss_weight": 0.5195006489753723, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.388714587688446, + "rkd_teacher_entropy": 3.7536903858184814, + "rkd_teacher_similarity_mean": 0.7856727004051208, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14420 + }, + { + "epoch": 4.198428862379982, + "flow_matching_loss": 0.015964481607079507, + "grad_norm": 0.16386893391609192, + "learning_rate": 1.7176522765014104e-05, + "loss": 0.021, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009726041508838535, + "rkd_loss_weight": 0.5191673040390015, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38451671302318574, + "rkd_teacher_entropy": 3.7440497159957884, + "rkd_teacher_similarity_mean": 0.7799383699893951, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14430 + }, + { + "epoch": 4.201338376491126, + "flow_matching_loss": 0.017382190190255643, + "grad_norm": 0.14151950180530548, + "learning_rate": 1.7160161738283782e-05, + "loss": 0.022, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008954036328941584, + "rkd_loss_weight": 0.5188339650630951, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38504915237426757, + "rkd_teacher_entropy": 3.734784412384033, + "rkd_teacher_similarity_mean": 0.7818918704986573, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14440 + }, + { + "epoch": 4.20424789060227, + "flow_matching_loss": 0.013892145548015832, + "grad_norm": 0.12841740250587463, + "learning_rate": 1.714379808675326e-05, + "loss": 0.0185, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00893460800871253, + "rkd_loss_weight": 0.5185006141662598, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39557033479213716, + "rkd_teacher_entropy": 3.7694579362869263, + "rkd_teacher_similarity_mean": 0.7922647893428802, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14450 + }, + { + "epoch": 4.207157404713413, + "flow_matching_loss": 0.01376346992328763, + "grad_norm": 0.1329321563243866, + "learning_rate": 1.712743183030592e-05, + "loss": 0.0185, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009190293122082948, + "rkd_loss_weight": 0.5181672751903534, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3917644828557968, + "rkd_teacher_entropy": 3.754420351982117, + "rkd_teacher_similarity_mean": 0.7891156017780304, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14460 + }, + { + "epoch": 4.210066918824556, + "flow_matching_loss": 0.014513122849166393, + "grad_norm": 0.16410240530967712, + "learning_rate": 1.7111062988828294e-05, + "loss": 0.0196, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009784694202244282, + "rkd_loss_weight": 0.5178339302539825, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38337385654449463, + "rkd_teacher_entropy": 3.7341525077819826, + "rkd_teacher_similarity_mean": 0.7798684954643249, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14470 + }, + { + "epoch": 4.2129764329357, + "flow_matching_loss": 0.013717100210487842, + "grad_norm": 0.18136118352413177, + "learning_rate": 1.709469158221008e-05, + "loss": 0.0198, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01168991457670927, + "rkd_loss_weight": 0.5175005793571472, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3905797302722931, + "rkd_teacher_entropy": 3.762316036224365, + "rkd_teacher_similarity_mean": 0.7864171206951142, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14480 + }, + { + "epoch": 4.215885947046843, + "flow_matching_loss": 0.016092297434806824, + "grad_norm": 0.11899459362030029, + "learning_rate": 1.7078317630344083e-05, + "loss": 0.0219, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011275166925042868, + "rkd_loss_weight": 0.5171672403812408, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3777416318655014, + "rkd_teacher_entropy": 3.722160506248474, + "rkd_teacher_similarity_mean": 0.7735595762729645, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14490 + }, + { + "epoch": 4.2187954611579865, + "flow_matching_loss": 0.014854240417480468, + "grad_norm": 0.11129783093929291, + "learning_rate": 1.706194115312618e-05, + "loss": 0.0204, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010744127631187438, + "rkd_loss_weight": 0.51683389544487, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3852448374032974, + "rkd_teacher_entropy": 3.7341422557830812, + "rkd_teacher_similarity_mean": 0.7817730486392975, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14500 + }, + { + "epoch": 4.22170497526913, + "flow_matching_loss": 0.01704372987151146, + "grad_norm": 0.1608160138130188, + "learning_rate": 1.704556217045536e-05, + "loss": 0.0216, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00891810618340969, + "rkd_loss_weight": 0.5165005505084992, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38126522600650786, + "rkd_teacher_entropy": 3.749091649055481, + "rkd_teacher_similarity_mean": 0.7797379851341247, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14510 + }, + { + "epoch": 4.224614489380274, + "flow_matching_loss": 0.015220005717128515, + "grad_norm": 0.1644209325313568, + "learning_rate": 1.7029180702233612e-05, + "loss": 0.021, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011128711234778165, + "rkd_loss_weight": 0.5161672055721283, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3807022929191589, + "rkd_teacher_entropy": 3.7277052879333494, + "rkd_teacher_similarity_mean": 0.7773911774158477, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14520 + }, + { + "epoch": 4.227524003491417, + "flow_matching_loss": 0.015231904201209545, + "grad_norm": 0.12019128352403641, + "learning_rate": 1.7012796768365975e-05, + "loss": 0.02, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009322010492905975, + "rkd_loss_weight": 0.5158338606357574, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3891221463680267, + "rkd_teacher_entropy": 3.760155439376831, + "rkd_teacher_similarity_mean": 0.7859695792198181, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14530 + }, + { + "epoch": 4.2304335176025605, + "flow_matching_loss": 0.014623101521283387, + "grad_norm": 0.14922724664211273, + "learning_rate": 1.6996410388760465e-05, + "loss": 0.0197, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009897336270660162, + "rkd_loss_weight": 0.5155005156993866, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37561908066272737, + "rkd_teacher_entropy": 3.71531081199646, + "rkd_teacher_similarity_mean": 0.7722173988819122, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14540 + }, + { + "epoch": 4.233343031713704, + "flow_matching_loss": 0.017668883316218853, + "grad_norm": 0.11430481821298599, + "learning_rate": 1.698002158332809e-05, + "loss": 0.023, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01031728545203805, + "rkd_loss_weight": 0.5151671767234802, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3793355941772461, + "rkd_teacher_entropy": 3.7337724924087525, + "rkd_teacher_similarity_mean": 0.7783281028270721, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14550 + }, + { + "epoch": 4.236252545824847, + "flow_matching_loss": 0.016202810406684875, + "grad_norm": 0.13339205086231232, + "learning_rate": 1.6963630371982783e-05, + "loss": 0.0208, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008958415733650327, + "rkd_loss_weight": 0.5148338258266449, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.382657578587532, + "rkd_teacher_entropy": 3.7397552490234376, + "rkd_teacher_similarity_mean": 0.7809366524219513, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14560 + }, + { + "epoch": 4.239162059935991, + "flow_matching_loss": 0.015049829613417387, + "grad_norm": 0.09989698976278305, + "learning_rate": 1.6947236774641416e-05, + "loss": 0.02, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009526043524965644, + "rkd_loss_weight": 0.514500480890274, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3953002095222473, + "rkd_teacher_entropy": 3.7790574550628664, + "rkd_teacher_similarity_mean": 0.7932335674762726, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14570 + }, + { + "epoch": 4.242071574047134, + "flow_matching_loss": 0.013668843638151884, + "grad_norm": 0.1143992692232132, + "learning_rate": 1.6930840811223755e-05, + "loss": 0.0199, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01217083060182631, + "rkd_loss_weight": 0.5141671419143676, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38383273780345917, + "rkd_teacher_entropy": 3.744377303123474, + "rkd_teacher_similarity_mean": 0.7805109322071075, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14580 + }, + { + "epoch": 4.244981088158277, + "flow_matching_loss": 0.014738005120307207, + "grad_norm": 0.16986216604709625, + "learning_rate": 1.6914442501652448e-05, + "loss": 0.019, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008269986975938082, + "rkd_loss_weight": 0.5138337910175323, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3910281121730804, + "rkd_teacher_entropy": 3.764383411407471, + "rkd_teacher_similarity_mean": 0.7880271792411804, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14590 + }, + { + "epoch": 4.247890602269421, + "flow_matching_loss": 0.013905453914776444, + "grad_norm": 0.1042054295539856, + "learning_rate": 1.6898041865852974e-05, + "loss": 0.0184, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008764063054695726, + "rkd_loss_weight": 0.513500452041626, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39556888341903684, + "rkd_teacher_entropy": 3.770714783668518, + "rkd_teacher_similarity_mean": 0.7927272617816925, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14600 + }, + { + "epoch": 4.250800116380565, + "flow_matching_loss": 0.015684608649462462, + "grad_norm": 0.11266862601041794, + "learning_rate": 1.6881638923753663e-05, + "loss": 0.0212, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010805617412552237, + "rkd_loss_weight": 0.5131671071052551, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3884266287088394, + "rkd_teacher_entropy": 3.74924099445343, + "rkd_teacher_similarity_mean": 0.783335417509079, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14610 + }, + { + "epoch": 4.2537096304917075, + "flow_matching_loss": 0.01437581991776824, + "grad_norm": 0.12732474505901337, + "learning_rate": 1.6865233695285632e-05, + "loss": 0.0193, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00952702364884317, + "rkd_loss_weight": 0.5128337562084198, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.388440802693367, + "rkd_teacher_entropy": 3.75381920337677, + "rkd_teacher_similarity_mean": 0.7843519151210785, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14620 + }, + { + "epoch": 4.256619144602851, + "flow_matching_loss": 0.014568849094212055, + "grad_norm": 0.11239607632160187, + "learning_rate": 1.6848826200382782e-05, + "loss": 0.0191, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008758102729916572, + "rkd_loss_weight": 0.512500423192978, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3973827391862869, + "rkd_teacher_entropy": 3.776046800613403, + "rkd_teacher_similarity_mean": 0.7936823248863221, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14630 + }, + { + "epoch": 4.259528658713995, + "flow_matching_loss": 0.015150590427219867, + "grad_norm": 0.09425687044858932, + "learning_rate": 1.6832416458981762e-05, + "loss": 0.0198, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009157165186479688, + "rkd_loss_weight": 0.5121670663356781, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3861189126968384, + "rkd_teacher_entropy": 3.7404683589935304, + "rkd_teacher_similarity_mean": 0.7821705341339111, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14640 + }, + { + "epoch": 4.262438172825139, + "flow_matching_loss": 0.013996420241892338, + "grad_norm": 0.1269286423921585, + "learning_rate": 1.681600449102196e-05, + "loss": 0.02, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011644022166728973, + "rkd_loss_weight": 0.5118337273597717, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3843249619007111, + "rkd_teacher_entropy": 3.75328369140625, + "rkd_teacher_similarity_mean": 0.7806979537010192, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14650 + }, + { + "epoch": 4.265347686936281, + "flow_matching_loss": 0.01537253512069583, + "grad_norm": 0.14474773406982422, + "learning_rate": 1.6799590316445468e-05, + "loss": 0.0217, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012286472506821156, + "rkd_loss_weight": 0.5115003883838654, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38304975926876067, + "rkd_teacher_entropy": 3.7344931840896605, + "rkd_teacher_similarity_mean": 0.7779374361038208, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14660 + }, + { + "epoch": 4.268257201047425, + "flow_matching_loss": 0.017970755137503146, + "grad_norm": 0.14672444760799408, + "learning_rate": 1.678317395519705e-05, + "loss": 0.0236, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010917302034795285, + "rkd_loss_weight": 0.5111670315265655, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38185561895370485, + "rkd_teacher_entropy": 3.7430189847946167, + "rkd_teacher_similarity_mean": 0.7783023416996002, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14670 + }, + { + "epoch": 4.271166715158569, + "flow_matching_loss": 0.015545608848333359, + "grad_norm": 0.13882356882095337, + "learning_rate": 1.676675542722413e-05, + "loss": 0.021, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010585390916094183, + "rkd_loss_weight": 0.5108336985111237, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39015314280986785, + "rkd_teacher_entropy": 3.753235340118408, + "rkd_teacher_similarity_mean": 0.7846302092075348, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14680 + }, + { + "epoch": 4.274076229269712, + "flow_matching_loss": 0.01429811753332615, + "grad_norm": 0.1350121945142746, + "learning_rate": 1.6750334752476776e-05, + "loss": 0.0189, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008954100403934717, + "rkd_loss_weight": 0.5105003476142883, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38275872766971586, + "rkd_teacher_entropy": 3.732780385017395, + "rkd_teacher_similarity_mean": 0.7786784112453461, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14690 + }, + { + "epoch": 4.276985743380855, + "flow_matching_loss": 0.017484317626804115, + "grad_norm": 0.19034439325332642, + "learning_rate": 1.673391195090765e-05, + "loss": 0.0226, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01011572228744626, + "rkd_loss_weight": 0.5101670026779175, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3763954699039459, + "rkd_teacher_entropy": 3.719337272644043, + "rkd_teacher_similarity_mean": 0.7727958023548126, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14700 + }, + { + "epoch": 4.279895257491999, + "flow_matching_loss": 0.01465861676260829, + "grad_norm": 0.1401083618402481, + "learning_rate": 1.6717487042472006e-05, + "loss": 0.0196, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009643987752497196, + "rkd_loss_weight": 0.5098336637020111, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3832811236381531, + "rkd_teacher_entropy": 3.734942579269409, + "rkd_teacher_similarity_mean": 0.7790607452392578, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14710 + }, + { + "epoch": 4.282804771603142, + "flow_matching_loss": 0.01555356876924634, + "grad_norm": 0.1484406292438507, + "learning_rate": 1.6701060047127655e-05, + "loss": 0.0202, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009192977985367178, + "rkd_loss_weight": 0.5095003128051758, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37912428975105283, + "rkd_teacher_entropy": 3.737939453125, + "rkd_teacher_similarity_mean": 0.7769925713539123, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14720 + }, + { + "epoch": 4.285714285714286, + "flow_matching_loss": 0.01717069735750556, + "grad_norm": 0.13402202725410461, + "learning_rate": 1.6684630984834947e-05, + "loss": 0.0216, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008675994537770747, + "rkd_loss_weight": 0.5091669738292695, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39360325038433075, + "rkd_teacher_entropy": 3.7731231689453124, + "rkd_teacher_similarity_mean": 0.7911084532737732, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14730 + }, + { + "epoch": 4.288623799825429, + "flow_matching_loss": 0.014422889146953822, + "grad_norm": 0.11333892494440079, + "learning_rate": 1.666819987555675e-05, + "loss": 0.0191, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00926144034601748, + "rkd_loss_weight": 0.5088336288928985, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3871480405330658, + "rkd_teacher_entropy": 3.748008918762207, + "rkd_teacher_similarity_mean": 0.7818977653980255, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14740 + }, + { + "epoch": 4.291533313936573, + "flow_matching_loss": 0.013377172593027353, + "grad_norm": 0.11099641770124435, + "learning_rate": 1.6651766739258392e-05, + "loss": 0.0179, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008978806296363473, + "rkd_loss_weight": 0.5085002779960632, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3843231827020645, + "rkd_teacher_entropy": 3.732310962677002, + "rkd_teacher_similarity_mean": 0.778970229625702, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14750 + }, + { + "epoch": 4.294442828047716, + "flow_matching_loss": 0.01657152036204934, + "grad_norm": 0.1701464056968689, + "learning_rate": 1.6635331595907704e-05, + "loss": 0.0217, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010011755209416152, + "rkd_loss_weight": 0.5081669390201569, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3952674388885498, + "rkd_teacher_entropy": 3.764198350906372, + "rkd_teacher_similarity_mean": 0.7908576905727387, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14760 + }, + { + "epoch": 4.2973523421588595, + "flow_matching_loss": 0.015237564034759999, + "grad_norm": 0.1190781444311142, + "learning_rate": 1.6618894465474923e-05, + "loss": 0.0194, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008200706588104368, + "rkd_loss_weight": 0.507833594083786, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3989512652158737, + "rkd_teacher_entropy": 3.78740017414093, + "rkd_teacher_similarity_mean": 0.7943663656711578, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14770 + }, + { + "epoch": 4.300261856270003, + "flow_matching_loss": 0.016463557351380588, + "grad_norm": 0.13663843274116516, + "learning_rate": 1.660245536793271e-05, + "loss": 0.0214, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009755481267347932, + "rkd_loss_weight": 0.5075002491474152, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38163132667541505, + "rkd_teacher_entropy": 3.7247625827789306, + "rkd_teacher_similarity_mean": 0.775992614030838, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14780 + }, + { + "epoch": 4.303171370381146, + "flow_matching_loss": 0.01393148973584175, + "grad_norm": 0.12914356589317322, + "learning_rate": 1.6586014323256126e-05, + "loss": 0.0192, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01041093929670751, + "rkd_loss_weight": 0.5071669042110443, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38898479342460635, + "rkd_teacher_entropy": 3.7576964139938354, + "rkd_teacher_similarity_mean": 0.7855288863182068, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14790 + }, + { + "epoch": 4.30608088449229, + "flow_matching_loss": 0.015452029090374709, + "grad_norm": 0.12175602465867996, + "learning_rate": 1.656957135142259e-05, + "loss": 0.0199, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008823601203039289, + "rkd_loss_weight": 0.5068335652351379, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38892472386360166, + "rkd_teacher_entropy": 3.7527488470077515, + "rkd_teacher_similarity_mean": 0.7852052271366119, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14800 + }, + { + "epoch": 4.3089903986034335, + "flow_matching_loss": 0.01589509928599, + "grad_norm": 0.20218676328659058, + "learning_rate": 1.655312647241186e-05, + "loss": 0.0216, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011334682069718837, + "rkd_loss_weight": 0.5065002143383026, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38567071259021757, + "rkd_teacher_entropy": 3.7350364208221434, + "rkd_teacher_similarity_mean": 0.7797099351882935, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14810 + }, + { + "epoch": 4.311899912714576, + "flow_matching_loss": 0.013508064579218626, + "grad_norm": 0.136082723736763, + "learning_rate": 1.6536679706206015e-05, + "loss": 0.0179, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008585731917992234, + "rkd_loss_weight": 0.5061668753623962, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3932671815156937, + "rkd_teacher_entropy": 3.768799591064453, + "rkd_teacher_similarity_mean": 0.7884999454021454, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14820 + }, + { + "epoch": 4.31480942682572, + "flow_matching_loss": 0.01697949841618538, + "grad_norm": 0.11321727186441422, + "learning_rate": 1.6520231072789427e-05, + "loss": 0.0223, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010600242065265775, + "rkd_loss_weight": 0.5058335304260254, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.391046991944313, + "rkd_teacher_entropy": 3.7578126192092896, + "rkd_teacher_similarity_mean": 0.7859881937503814, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14830 + }, + { + "epoch": 4.317718940936864, + "flow_matching_loss": 0.015898262429982424, + "grad_norm": 0.1455542892217636, + "learning_rate": 1.6503780592148737e-05, + "loss": 0.0206, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009209226816892624, + "rkd_loss_weight": 0.50550017952919, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39451451003551485, + "rkd_teacher_entropy": 3.7657355308532714, + "rkd_teacher_similarity_mean": 0.7894185304641723, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14840 + }, + { + "epoch": 4.320628455048007, + "flow_matching_loss": 0.013668027333915233, + "grad_norm": 0.11692952364683151, + "learning_rate": 1.648732828427283e-05, + "loss": 0.0184, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009319999441504478, + "rkd_loss_weight": 0.5051668465137482, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39332629144191744, + "rkd_teacher_entropy": 3.7604864835739136, + "rkd_teacher_similarity_mean": 0.7892859697341919, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14850 + }, + { + "epoch": 4.32353796915915, + "flow_matching_loss": 0.01559792710468173, + "grad_norm": 0.20791926980018616, + "learning_rate": 1.647087416915281e-05, + "loss": 0.0204, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009421018790453673, + "rkd_loss_weight": 0.5048334896564484, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38636762499809263, + "rkd_teacher_entropy": 3.7369882822036744, + "rkd_teacher_similarity_mean": 0.7839009642601014, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14860 + }, + { + "epoch": 4.326447483270294, + "flow_matching_loss": 0.016080070845782758, + "grad_norm": 0.13953785598278046, + "learning_rate": 1.6454418266781987e-05, + "loss": 0.0207, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009170470014214515, + "rkd_loss_weight": 0.504500150680542, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3929387599229813, + "rkd_teacher_entropy": 3.7715436458587646, + "rkd_teacher_similarity_mean": 0.7890156984329224, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14870 + }, + { + "epoch": 4.329356997381438, + "flow_matching_loss": 0.01699187783524394, + "grad_norm": 0.14799335598945618, + "learning_rate": 1.6437960597155824e-05, + "loss": 0.0221, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010114762838929892, + "rkd_loss_weight": 0.5041668117046356, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3901684135198593, + "rkd_teacher_entropy": 3.753424859046936, + "rkd_teacher_similarity_mean": 0.7849898159503936, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14880 + }, + { + "epoch": 4.3322665114925805, + "flow_matching_loss": 0.014265009481459857, + "grad_norm": 0.12114934623241425, + "learning_rate": 1.642150118027194e-05, + "loss": 0.0188, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009038659371435642, + "rkd_loss_weight": 0.5038334548473358, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39227353036403656, + "rkd_teacher_entropy": 3.7609155893325807, + "rkd_teacher_similarity_mean": 0.7875163793563843, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14890 + }, + { + "epoch": 4.335176025603724, + "flow_matching_loss": 0.017813057545572518, + "grad_norm": 0.14366529881954193, + "learning_rate": 1.640504003613009e-05, + "loss": 0.0226, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009476035553961992, + "rkd_loss_weight": 0.503500121831894, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38703483045101167, + "rkd_teacher_entropy": 3.7484612464904785, + "rkd_teacher_similarity_mean": 0.7842558443546295, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14900 + }, + { + "epoch": 4.338085539714868, + "flow_matching_loss": 0.016894249152392148, + "grad_norm": 0.13450446724891663, + "learning_rate": 1.6388577184732107e-05, + "loss": 0.0219, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009906800836324692, + "rkd_loss_weight": 0.5031667709350586, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38962445557117464, + "rkd_teacher_entropy": 3.758418107032776, + "rkd_teacher_similarity_mean": 0.7863789021968841, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14910 + }, + { + "epoch": 4.340995053826011, + "flow_matching_loss": 0.01604188196361065, + "grad_norm": 0.13382014632225037, + "learning_rate": 1.6372112646081907e-05, + "loss": 0.0209, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00974812898784876, + "rkd_loss_weight": 0.5028334259986877, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3875406265258789, + "rkd_teacher_entropy": 3.7429740905761717, + "rkd_teacher_similarity_mean": 0.7845647215843201, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14920 + }, + { + "epoch": 4.343904567937154, + "flow_matching_loss": 0.017814129032194613, + "grad_norm": 0.1777607798576355, + "learning_rate": 1.6355646440185464e-05, + "loss": 0.0231, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010523309418931603, + "rkd_loss_weight": 0.5025000870227814, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3813187897205353, + "rkd_teacher_entropy": 3.7380271911621095, + "rkd_teacher_similarity_mean": 0.7776569187641144, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14930 + }, + { + "epoch": 4.346814082048298, + "flow_matching_loss": 0.01606729459017515, + "grad_norm": 0.1269378513097763, + "learning_rate": 1.6339178587050764e-05, + "loss": 0.021, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009882614389061928, + "rkd_loss_weight": 0.5021667361259461, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3924069792032242, + "rkd_teacher_entropy": 3.7607086658477784, + "rkd_teacher_similarity_mean": 0.7889501214027405, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14940 + }, + { + "epoch": 4.349723596159441, + "flow_matching_loss": 0.01510247541591525, + "grad_norm": 0.15709616243839264, + "learning_rate": 1.632270910668781e-05, + "loss": 0.0201, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009924743371084333, + "rkd_loss_weight": 0.5018333971500397, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39491123557090757, + "rkd_teacher_entropy": 3.767144227027893, + "rkd_teacher_similarity_mean": 0.7904297947883606, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14950 + }, + { + "epoch": 4.352633110270585, + "flow_matching_loss": 0.013496332755312323, + "grad_norm": 0.15099844336509705, + "learning_rate": 1.6306238019108562e-05, + "loss": 0.0183, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009534108731895686, + "rkd_loss_weight": 0.5015000522136688, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3816094994544983, + "rkd_teacher_entropy": 3.7339943647384644, + "rkd_teacher_similarity_mean": 0.7779581129550934, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14960 + }, + { + "epoch": 4.355542624381728, + "flow_matching_loss": 0.015463450737297536, + "grad_norm": 0.12418626993894577, + "learning_rate": 1.6289765344326952e-05, + "loss": 0.0193, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007695357780903578, + "rkd_loss_weight": 0.5011667013168335, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3877167165279388, + "rkd_teacher_entropy": 3.753309655189514, + "rkd_teacher_similarity_mean": 0.7829567790031433, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14970 + }, + { + "epoch": 4.358452138492872, + "flow_matching_loss": 0.016587523370981218, + "grad_norm": 0.11634265631437302, + "learning_rate": 1.6273291102358833e-05, + "loss": 0.0209, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008623846434056759, + "rkd_loss_weight": 0.5008333623409271, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.385781791806221, + "rkd_teacher_entropy": 3.747362971305847, + "rkd_teacher_similarity_mean": 0.7820015549659729, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14980 + }, + { + "epoch": 4.361361652604015, + "flow_matching_loss": 0.014708815049380064, + "grad_norm": 0.13096071779727936, + "learning_rate": 1.6256815313221963e-05, + "loss": 0.0194, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009419192560017109, + "rkd_loss_weight": 0.5005000174045563, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3930469214916229, + "rkd_teacher_entropy": 3.7586756467819216, + "rkd_teacher_similarity_mean": 0.7883979439735412, + "rkd_vlm_temp": 0.10000000149011612, + "step": 14990 + }, + { + "epoch": 4.364271166715159, + "flow_matching_loss": 0.015109305828809738, + "grad_norm": 0.09433169662952423, + "learning_rate": 1.6240337996935972e-05, + "loss": 0.0199, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00961321322247386, + "rkd_loss_weight": 0.5001666724681855, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38350779712200167, + "rkd_teacher_entropy": 3.740031909942627, + "rkd_teacher_similarity_mean": 0.780010461807251, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15000 + }, + { + "epoch": 4.367180680826302, + "flow_matching_loss": 0.016325947642326356, + "grad_norm": 0.1392330825328827, + "learning_rate": 1.6223859173522367e-05, + "loss": 0.0205, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008323422260582447, + "rkd_loss_weight": 0.4998333305120468, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38353601694107053, + "rkd_teacher_entropy": 3.7367622375488283, + "rkd_teacher_similarity_mean": 0.7801901996135712, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15010 + }, + { + "epoch": 4.370090194937445, + "flow_matching_loss": 0.01639810996130109, + "grad_norm": 0.1228889524936676, + "learning_rate": 1.6207378863004455e-05, + "loss": 0.0203, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007853053603321314, + "rkd_loss_weight": 0.4994999825954437, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3816773682832718, + "rkd_teacher_entropy": 3.732175278663635, + "rkd_teacher_similarity_mean": 0.7772081673145295, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15020 + }, + { + "epoch": 4.372999709048589, + "flow_matching_loss": 0.013525386992841958, + "grad_norm": 0.1529306173324585, + "learning_rate": 1.6190897085407382e-05, + "loss": 0.0178, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00859358087182045, + "rkd_loss_weight": 0.4991666376590729, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3896830528974533, + "rkd_teacher_entropy": 3.757411742210388, + "rkd_teacher_similarity_mean": 0.7861161470413208, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15030 + }, + { + "epoch": 4.3759092231597325, + "flow_matching_loss": 0.014605278335511684, + "grad_norm": 0.1598024219274521, + "learning_rate": 1.617441386075805e-05, + "loss": 0.0193, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00939357653260231, + "rkd_loss_weight": 0.4988332986831665, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38177417814731596, + "rkd_teacher_entropy": 3.735437250137329, + "rkd_teacher_similarity_mean": 0.7782414853572845, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15040 + }, + { + "epoch": 4.378818737270876, + "flow_matching_loss": 0.015565304737538099, + "grad_norm": 0.18043994903564453, + "learning_rate": 1.6157929209085142e-05, + "loss": 0.02, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008903558598831296, + "rkd_loss_weight": 0.49849994480609894, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37963636219501495, + "rkd_teacher_entropy": 3.7230520248413086, + "rkd_teacher_similarity_mean": 0.7772915601730347, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15050 + }, + { + "epoch": 4.381728251382019, + "flow_matching_loss": 0.015043562091886998, + "grad_norm": 0.11999887228012085, + "learning_rate": 1.614144315041906e-05, + "loss": 0.0191, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008076206920668483, + "rkd_loss_weight": 0.49816660583019257, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3809766948223114, + "rkd_teacher_entropy": 3.7313637018203734, + "rkd_teacher_similarity_mean": 0.779605609178543, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15060 + }, + { + "epoch": 4.384637765493163, + "flow_matching_loss": 0.016551261488348245, + "grad_norm": 0.1555318534374237, + "learning_rate": 1.612495570479192e-05, + "loss": 0.0205, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007958491472527385, + "rkd_loss_weight": 0.49783326387405397, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3796262502670288, + "rkd_teacher_entropy": 3.731432008743286, + "rkd_teacher_similarity_mean": 0.776610654592514, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15070 + }, + { + "epoch": 4.3875472796043065, + "flow_matching_loss": 0.018815217539668083, + "grad_norm": 0.15625986456871033, + "learning_rate": 1.6108466892237517e-05, + "loss": 0.023, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008314196951687336, + "rkd_loss_weight": 0.49749991297721863, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38129649460315707, + "rkd_teacher_entropy": 3.736818623542786, + "rkd_teacher_similarity_mean": 0.7782282054424285, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15080 + }, + { + "epoch": 4.390456793715449, + "flow_matching_loss": 0.017810961604118346, + "grad_norm": 0.14013230800628662, + "learning_rate": 1.609197673279132e-05, + "loss": 0.0221, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008581740409135818, + "rkd_loss_weight": 0.49716657400131226, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3917904019355774, + "rkd_teacher_entropy": 3.7685622692108156, + "rkd_teacher_similarity_mean": 0.7899438440799713, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15090 + }, + { + "epoch": 4.393366307826593, + "flow_matching_loss": 0.01630017561838031, + "grad_norm": 0.193184033036232, + "learning_rate": 1.6075485246490428e-05, + "loss": 0.0207, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00882588755339384, + "rkd_loss_weight": 0.49683322608470915, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3880681574344635, + "rkd_teacher_entropy": 3.7440359354019166, + "rkd_teacher_similarity_mean": 0.7856143057346344, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15100 + }, + { + "epoch": 4.396275821937737, + "flow_matching_loss": 0.01707146461121738, + "grad_norm": 0.11799610406160355, + "learning_rate": 1.605899245337355e-05, + "loss": 0.0213, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008425839804112911, + "rkd_loss_weight": 0.4964998811483383, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3766784191131592, + "rkd_teacher_entropy": 3.714428925514221, + "rkd_teacher_similarity_mean": 0.772972697019577, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15110 + }, + { + "epoch": 4.3991853360488795, + "flow_matching_loss": 0.01547321304678917, + "grad_norm": 0.1399538367986679, + "learning_rate": 1.6042498373480983e-05, + "loss": 0.0195, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008181653544306755, + "rkd_loss_weight": 0.49616654217243195, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39316610991954803, + "rkd_teacher_entropy": 3.759928750991821, + "rkd_teacher_similarity_mean": 0.7907270431518555, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15120 + }, + { + "epoch": 4.402094850160023, + "flow_matching_loss": 0.014418426994234323, + "grad_norm": 0.09918193519115448, + "learning_rate": 1.602600302685459e-05, + "loss": 0.019, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009319509193301202, + "rkd_loss_weight": 0.49583319425582884, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3884526759386063, + "rkd_teacher_entropy": 3.754243278503418, + "rkd_teacher_similarity_mean": 0.7853314518928528, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15130 + }, + { + "epoch": 4.405004364271167, + "flow_matching_loss": 0.016520039830356836, + "grad_norm": 0.17625653743743896, + "learning_rate": 1.6009506433537774e-05, + "loss": 0.0212, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009491849038749934, + "rkd_loss_weight": 0.495499849319458, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38933991491794584, + "rkd_teacher_entropy": 3.7611085414886474, + "rkd_teacher_similarity_mean": 0.7860175609588623, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15140 + }, + { + "epoch": 4.40791387838231, + "flow_matching_loss": 0.013104847632348537, + "grad_norm": 0.09658341109752655, + "learning_rate": 1.599300861357545e-05, + "loss": 0.0179, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009634802024811507, + "rkd_loss_weight": 0.49516651034355164, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38881371915340424, + "rkd_teacher_entropy": 3.753363919258118, + "rkd_teacher_similarity_mean": 0.7856534063816071, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15150 + }, + { + "epoch": 4.4108233924934535, + "flow_matching_loss": 0.01461921725422144, + "grad_norm": 0.17315468192100525, + "learning_rate": 1.5976509587014032e-05, + "loss": 0.0201, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011116880178451537, + "rkd_loss_weight": 0.49483315646648407, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3867286115884781, + "rkd_teacher_entropy": 3.757890748977661, + "rkd_teacher_similarity_mean": 0.7835926234722137, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15160 + }, + { + "epoch": 4.413732906604597, + "flow_matching_loss": 0.015518915560096502, + "grad_norm": 0.11212094128131866, + "learning_rate": 1.5960009373901384e-05, + "loss": 0.02, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00897689643315971, + "rkd_loss_weight": 0.4944998174905777, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38448616564273835, + "rkd_teacher_entropy": 3.754997992515564, + "rkd_teacher_similarity_mean": 0.7819453597068786, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15170 + }, + { + "epoch": 4.416642420715741, + "flow_matching_loss": 0.015400576125830411, + "grad_norm": 0.13960330188274384, + "learning_rate": 1.5943507994286823e-05, + "loss": 0.02, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00926815285347402, + "rkd_loss_weight": 0.4941664755344391, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3849134027957916, + "rkd_teacher_entropy": 3.746153235435486, + "rkd_teacher_similarity_mean": 0.7817315280437469, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15180 + }, + { + "epoch": 4.419551934826884, + "flow_matching_loss": 0.017910159938037396, + "grad_norm": 0.11544210463762283, + "learning_rate": 1.5927005468221082e-05, + "loss": 0.0237, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01181841092184186, + "rkd_loss_weight": 0.49383312463760376, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3854218155145645, + "rkd_teacher_entropy": 3.755238890647888, + "rkd_teacher_similarity_mean": 0.7817229628562927, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15190 + }, + { + "epoch": 4.422461448938027, + "flow_matching_loss": 0.015196772571653128, + "grad_norm": 0.09739126265048981, + "learning_rate": 1.5910501815756285e-05, + "loss": 0.0194, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008530354686081409, + "rkd_loss_weight": 0.4934997856616974, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3807384967803955, + "rkd_teacher_entropy": 3.73053297996521, + "rkd_teacher_similarity_mean": 0.7783098220825195, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15200 + }, + { + "epoch": 4.425370963049171, + "flow_matching_loss": 0.014768172614276409, + "grad_norm": 0.1084161028265953, + "learning_rate": 1.589399705694593e-05, + "loss": 0.0195, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009573472756892442, + "rkd_loss_weight": 0.4931664377450943, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3851009428501129, + "rkd_teacher_entropy": 3.7412758111953734, + "rkd_teacher_similarity_mean": 0.7815673351287842, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15210 + }, + { + "epoch": 4.428280477160314, + "flow_matching_loss": 0.01528178844600916, + "grad_norm": 0.12570765614509583, + "learning_rate": 1.587749121184485e-05, + "loss": 0.0193, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008224937319755554, + "rkd_loss_weight": 0.49283309280872345, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3887781918048859, + "rkd_teacher_entropy": 3.756561541557312, + "rkd_teacher_similarity_mean": 0.7859101235866547, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15220 + }, + { + "epoch": 4.431189991271458, + "flow_matching_loss": 0.015121325571089982, + "grad_norm": 0.0999971404671669, + "learning_rate": 1.5860984300509206e-05, + "loss": 0.0194, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00861050239764154, + "rkd_loss_weight": 0.4924997538328171, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3862706869840622, + "rkd_teacher_entropy": 3.756147003173828, + "rkd_teacher_similarity_mean": 0.7839293062686921, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15230 + }, + { + "epoch": 4.434099505382601, + "flow_matching_loss": 0.01623823894187808, + "grad_norm": 0.1771387904882431, + "learning_rate": 1.5844476342996453e-05, + "loss": 0.021, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009622760815545917, + "rkd_loss_weight": 0.49216640293598174, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3840878874063492, + "rkd_teacher_entropy": 3.742386221885681, + "rkd_teacher_similarity_mean": 0.780803096294403, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15240 + }, + { + "epoch": 4.437009019493744, + "flow_matching_loss": 0.015225897822529078, + "grad_norm": 0.11699093133211136, + "learning_rate": 1.5827967359365306e-05, + "loss": 0.0202, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01015540617518127, + "rkd_loss_weight": 0.49183306097984314, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39237635731697085, + "rkd_teacher_entropy": 3.764909291267395, + "rkd_teacher_similarity_mean": 0.7892781853675842, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15250 + }, + { + "epoch": 4.439918533604888, + "flow_matching_loss": 0.01524876281619072, + "grad_norm": 0.1348302960395813, + "learning_rate": 1.5811457369675758e-05, + "loss": 0.0195, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008725630585104227, + "rkd_loss_weight": 0.49149971902370454, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.40327560901641846, + "rkd_teacher_entropy": 3.800950264930725, + "rkd_teacher_similarity_mean": 0.8000578701496124, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15260 + }, + { + "epoch": 4.442828047716032, + "flow_matching_loss": 0.018892494030296803, + "grad_norm": 0.12884894013404846, + "learning_rate": 1.5794946393988982e-05, + "loss": 0.0235, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009403334697708488, + "rkd_loss_weight": 0.4911663681268692, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38264192938804625, + "rkd_teacher_entropy": 3.7348220586776733, + "rkd_teacher_similarity_mean": 0.7774173676967621, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15270 + }, + { + "epoch": 4.445737561827175, + "flow_matching_loss": 0.015849278215318917, + "grad_norm": 0.10508684813976288, + "learning_rate": 1.5778434452367382e-05, + "loss": 0.02, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008475154545158148, + "rkd_loss_weight": 0.49083302915096283, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39060859084129335, + "rkd_teacher_entropy": 3.758342242240906, + "rkd_teacher_similarity_mean": 0.7866597890853881, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15280 + }, + { + "epoch": 4.448647075938318, + "flow_matching_loss": 0.014985739719122648, + "grad_norm": 0.12151330709457397, + "learning_rate": 1.576192156487452e-05, + "loss": 0.0197, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009575061267241836, + "rkd_loss_weight": 0.490499684214592, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3858360469341278, + "rkd_teacher_entropy": 3.740119123458862, + "rkd_teacher_similarity_mean": 0.7820057988166809, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15290 + }, + { + "epoch": 4.451556590049462, + "flow_matching_loss": 0.017464340198785067, + "grad_norm": 0.15510156750679016, + "learning_rate": 1.574540775157511e-05, + "loss": 0.022, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00918537899851799, + "rkd_loss_weight": 0.4901663362979889, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3800637423992157, + "rkd_teacher_entropy": 3.742747497558594, + "rkd_teacher_similarity_mean": 0.7776907324790955, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15300 + }, + { + "epoch": 4.4544661041606055, + "flow_matching_loss": 0.014621646609157324, + "grad_norm": 0.1611538827419281, + "learning_rate": 1.5728893032534998e-05, + "loss": 0.0198, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010561424074694514, + "rkd_loss_weight": 0.4898329973220825, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38844447433948515, + "rkd_teacher_entropy": 3.7583329916000365, + "rkd_teacher_similarity_mean": 0.7857670366764069, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15310 + }, + { + "epoch": 4.457375618271748, + "flow_matching_loss": 0.012795846350491047, + "grad_norm": 0.12889651954174042, + "learning_rate": 1.571237742782112e-05, + "loss": 0.0178, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010305100539699196, + "rkd_loss_weight": 0.4894996464252472, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3881300866603851, + "rkd_teacher_entropy": 3.7523336172103883, + "rkd_teacher_similarity_mean": 0.7834143698215484, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15320 + }, + { + "epoch": 4.460285132382892, + "flow_matching_loss": 0.015361307561397553, + "grad_norm": 0.11237585544586182, + "learning_rate": 1.569586095750149e-05, + "loss": 0.0197, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00885031153447926, + "rkd_loss_weight": 0.4891663044691086, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38269701302051545, + "rkd_teacher_entropy": 3.733975863456726, + "rkd_teacher_similarity_mean": 0.7782307088375091, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15330 + }, + { + "epoch": 4.463194646494036, + "flow_matching_loss": 0.013720433879643678, + "grad_norm": 0.10476595908403397, + "learning_rate": 1.5679343641645193e-05, + "loss": 0.0178, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008412842778488994, + "rkd_loss_weight": 0.4888329654932022, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.389948245882988, + "rkd_teacher_entropy": 3.757446527481079, + "rkd_teacher_similarity_mean": 0.7876951992511749, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15340 + }, + { + "epoch": 4.466104160605179, + "flow_matching_loss": 0.015367629658430814, + "grad_norm": 0.1183597669005394, + "learning_rate": 1.5662825500322304e-05, + "loss": 0.0199, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00936132981441915, + "rkd_loss_weight": 0.48849961459636687, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38744508028030394, + "rkd_teacher_entropy": 3.7530284881591798, + "rkd_teacher_similarity_mean": 0.7839592099189758, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15350 + }, + { + "epoch": 4.469013674716322, + "flow_matching_loss": 0.016907027922570705, + "grad_norm": 0.13894125819206238, + "learning_rate": 1.5646306553603932e-05, + "loss": 0.021, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008438515430316328, + "rkd_loss_weight": 0.48816627264022827, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38856152892112733, + "rkd_teacher_entropy": 3.7412392139434814, + "rkd_teacher_similarity_mean": 0.7841209471225739, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15360 + }, + { + "epoch": 4.471923188827466, + "flow_matching_loss": 0.01730038281530142, + "grad_norm": 0.13464587926864624, + "learning_rate": 1.5629786821562154e-05, + "loss": 0.0215, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008581011230126024, + "rkd_loss_weight": 0.48783293068408967, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39345550537109375, + "rkd_teacher_entropy": 3.7628000020980834, + "rkd_teacher_similarity_mean": 0.7887833297252655, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15370 + }, + { + "epoch": 4.47483270293861, + "flow_matching_loss": 0.01405011061578989, + "grad_norm": 0.15630094707012177, + "learning_rate": 1.561326632427001e-05, + "loss": 0.0188, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009735217364504934, + "rkd_loss_weight": 0.48749957978725433, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.396638897061348, + "rkd_teacher_entropy": 3.7738448858261107, + "rkd_teacher_similarity_mean": 0.7921802222728729, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15380 + }, + { + "epoch": 4.4777422170497525, + "flow_matching_loss": 0.016017878241837025, + "grad_norm": 0.11999255418777466, + "learning_rate": 1.5596745081801447e-05, + "loss": 0.0198, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007740007480606437, + "rkd_loss_weight": 0.48716624081134796, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3886969268321991, + "rkd_teacher_entropy": 3.7483836889266966, + "rkd_teacher_similarity_mean": 0.783832436800003, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15390 + }, + { + "epoch": 4.480651731160896, + "flow_matching_loss": 0.014658907242119313, + "grad_norm": 0.12422582507133484, + "learning_rate": 1.5580223114231348e-05, + "loss": 0.0189, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00869869189336896, + "rkd_loss_weight": 0.48683289587497713, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3993815124034882, + "rkd_teacher_entropy": 3.7858745574951174, + "rkd_teacher_similarity_mean": 0.7952039182186127, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15400 + }, + { + "epoch": 4.48356124527204, + "flow_matching_loss": 0.0137765409424901, + "grad_norm": 0.12194593995809555, + "learning_rate": 1.5563700441635445e-05, + "loss": 0.0183, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009202834172174334, + "rkd_loss_weight": 0.486499547958374, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38623646795749667, + "rkd_teacher_entropy": 3.745773696899414, + "rkd_teacher_similarity_mean": 0.7812827825546265, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15410 + }, + { + "epoch": 4.486470759383183, + "flow_matching_loss": 0.0173289286904037, + "grad_norm": 0.11281583458185196, + "learning_rate": 1.5547177084090358e-05, + "loss": 0.022, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009645675495266914, + "rkd_loss_weight": 0.48616620898246765, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38346913158893586, + "rkd_teacher_entropy": 3.7394842386245726, + "rkd_teacher_similarity_mean": 0.7793325841426849, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15420 + }, + { + "epoch": 4.4893802734943264, + "flow_matching_loss": 0.014569248631596565, + "grad_norm": 0.14673998951911926, + "learning_rate": 1.5530653061673514e-05, + "loss": 0.0189, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008982635382562876, + "rkd_loss_weight": 0.4858328580856323, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38895736932754515, + "rkd_teacher_entropy": 3.753132629394531, + "rkd_teacher_similarity_mean": 0.7867584884166717, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15430 + }, + { + "epoch": 4.49228978760547, + "flow_matching_loss": 0.016710819210857153, + "grad_norm": 0.1425517350435257, + "learning_rate": 1.5514128394463167e-05, + "loss": 0.0214, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009618447814136743, + "rkd_loss_weight": 0.4854995161294937, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3958801716566086, + "rkd_teacher_entropy": 3.769186496734619, + "rkd_teacher_similarity_mean": 0.7915683090686798, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15440 + }, + { + "epoch": 4.495199301716613, + "flow_matching_loss": 0.014702636655420065, + "grad_norm": 0.1349312961101532, + "learning_rate": 1.549760310253834e-05, + "loss": 0.0204, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011799124907702208, + "rkd_loss_weight": 0.48516617715358734, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3891328811645508, + "rkd_teacher_entropy": 3.7541977405548095, + "rkd_teacher_similarity_mean": 0.7834421038627625, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15450 + }, + { + "epoch": 4.498108815827757, + "flow_matching_loss": 0.01451376611366868, + "grad_norm": 0.11700194329023361, + "learning_rate": 1.5481077205978816e-05, + "loss": 0.0192, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00971264922991395, + "rkd_loss_weight": 0.4848328232765198, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3858819454908371, + "rkd_teacher_entropy": 3.741065168380737, + "rkd_teacher_similarity_mean": 0.7804217636585236, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15460 + }, + { + "epoch": 4.5010183299389, + "flow_matching_loss": 0.012801633728668094, + "grad_norm": 0.10525963455438614, + "learning_rate": 1.5464550724865135e-05, + "loss": 0.0175, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00978316469117999, + "rkd_loss_weight": 0.4844994843006134, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3858676850795746, + "rkd_teacher_entropy": 3.7465219497680664, + "rkd_teacher_similarity_mean": 0.7826243996620178, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15470 + }, + { + "epoch": 4.503927844050043, + "flow_matching_loss": 0.01683522863313556, + "grad_norm": 0.12270151823759079, + "learning_rate": 1.5448023679278517e-05, + "loss": 0.0212, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009098393470048904, + "rkd_loss_weight": 0.4841661393642426, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3781444847583771, + "rkd_teacher_entropy": 3.7228007078170777, + "rkd_teacher_similarity_mean": 0.7734023094177246, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15480 + }, + { + "epoch": 4.506837358161187, + "flow_matching_loss": 0.016048336774110793, + "grad_norm": 0.11966320872306824, + "learning_rate": 1.543149608930088e-05, + "loss": 0.0206, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00935218040831387, + "rkd_loss_weight": 0.48383279144763947, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38537276685237887, + "rkd_teacher_entropy": 3.7326302766799926, + "rkd_teacher_similarity_mean": 0.7807897210121155, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15490 + }, + { + "epoch": 4.509746872272331, + "flow_matching_loss": 0.017422986682504416, + "grad_norm": 0.16817176342010498, + "learning_rate": 1.5414967975014805e-05, + "loss": 0.0219, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009300877247005701, + "rkd_loss_weight": 0.4834994524717331, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3850229889154434, + "rkd_teacher_entropy": 3.753816342353821, + "rkd_teacher_similarity_mean": 0.7816944360733032, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15500 + }, + { + "epoch": 4.512656386383474, + "flow_matching_loss": 0.014671436790376902, + "grad_norm": 0.13853326439857483, + "learning_rate": 1.539843935650351e-05, + "loss": 0.0196, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010234497115015983, + "rkd_loss_weight": 0.48316610455513, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39524413645267487, + "rkd_teacher_entropy": 3.7688517570495605, + "rkd_teacher_similarity_mean": 0.7920857191085815, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15510 + }, + { + "epoch": 4.515565900494617, + "flow_matching_loss": 0.01561178918927908, + "grad_norm": 0.20474830269813538, + "learning_rate": 1.5381910253850834e-05, + "loss": 0.0211, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011353666195645928, + "rkd_loss_weight": 0.48283275961875916, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38075690567493437, + "rkd_teacher_entropy": 3.719246029853821, + "rkd_teacher_similarity_mean": 0.777092570066452, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15520 + }, + { + "epoch": 4.518475414605761, + "flow_matching_loss": 0.013856042828410863, + "grad_norm": 0.17995791137218475, + "learning_rate": 1.5365380687141183e-05, + "loss": 0.0177, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008036299655213953, + "rkd_loss_weight": 0.4824994206428528, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3832302391529083, + "rkd_teacher_entropy": 3.737690782546997, + "rkd_teacher_similarity_mean": 0.7815475285053253, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15530 + }, + { + "epoch": 4.521384928716905, + "flow_matching_loss": 0.015708901546895505, + "grad_norm": 0.12771253287792206, + "learning_rate": 1.5348850676459544e-05, + "loss": 0.0212, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01146266758441925, + "rkd_loss_weight": 0.4821660667657852, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3824955314397812, + "rkd_teacher_entropy": 3.742147135734558, + "rkd_teacher_similarity_mean": 0.7810673117637634, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15540 + }, + { + "epoch": 4.524294442828047, + "flow_matching_loss": 0.01571432463824749, + "grad_norm": 0.08966214209794998, + "learning_rate": 1.5332320241891438e-05, + "loss": 0.0194, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00773379891179502, + "rkd_loss_weight": 0.48183272778987885, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39196679890155794, + "rkd_teacher_entropy": 3.7668998718261717, + "rkd_teacher_similarity_mean": 0.7892067492008209, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15550 + }, + { + "epoch": 4.527203956939191, + "flow_matching_loss": 0.01661347383633256, + "grad_norm": 0.13006922602653503, + "learning_rate": 1.5315789403522908e-05, + "loss": 0.0202, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007445191266015172, + "rkd_loss_weight": 0.4814993888139725, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3905827045440674, + "rkd_teacher_entropy": 3.7636006593704225, + "rkd_teacher_similarity_mean": 0.7872957170009613, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15560 + }, + { + "epoch": 4.530113471050335, + "flow_matching_loss": 0.01750286892056465, + "grad_norm": 0.15229573845863342, + "learning_rate": 1.529925818144047e-05, + "loss": 0.0216, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008607040438801051, + "rkd_loss_weight": 0.4811660349369049, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38795130550861356, + "rkd_teacher_entropy": 3.753757905960083, + "rkd_teacher_similarity_mean": 0.7851765275001525, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15570 + }, + { + "epoch": 4.5330229851614785, + "flow_matching_loss": 0.016993778850883247, + "grad_norm": 0.0973297506570816, + "learning_rate": 1.5282726595731126e-05, + "loss": 0.0217, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009861670481041074, + "rkd_loss_weight": 0.48083269596099854, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.388843235373497, + "rkd_teacher_entropy": 3.7559388399124147, + "rkd_teacher_similarity_mean": 0.7856887698173523, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15580 + }, + { + "epoch": 4.535932499272621, + "flow_matching_loss": 0.013811699207872152, + "grad_norm": 0.1163967028260231, + "learning_rate": 1.5266194666482304e-05, + "loss": 0.018, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008712674770504237, + "rkd_loss_weight": 0.4804993510246277, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39172652959823606, + "rkd_teacher_entropy": 3.758353281021118, + "rkd_teacher_similarity_mean": 0.7885165989398957, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15590 + }, + { + "epoch": 4.538842013383765, + "flow_matching_loss": 0.01647224258631468, + "grad_norm": 0.15008030831813812, + "learning_rate": 1.5249662413781866e-05, + "loss": 0.0209, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00926026194356382, + "rkd_loss_weight": 0.4801660031080246, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38868752419948577, + "rkd_teacher_entropy": 3.758363223075867, + "rkd_teacher_similarity_mean": 0.7846027314662933, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15600 + }, + { + "epoch": 4.541751527494909, + "flow_matching_loss": 0.015516362665221096, + "grad_norm": 0.17337903380393982, + "learning_rate": 1.5233129857718054e-05, + "loss": 0.0207, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010819049645215273, + "rkd_loss_weight": 0.4798326641321182, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39257214963436127, + "rkd_teacher_entropy": 3.7647876739501953, + "rkd_teacher_similarity_mean": 0.7885352849960328, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15610 + }, + { + "epoch": 4.5446610416060516, + "flow_matching_loss": 0.014151031244546175, + "grad_norm": 0.12298211455345154, + "learning_rate": 1.521659701837948e-05, + "loss": 0.0186, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009307984402403236, + "rkd_loss_weight": 0.4794993162155151, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3833700090646744, + "rkd_teacher_entropy": 3.733708310127258, + "rkd_teacher_similarity_mean": 0.7804021418094635, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15620 + }, + { + "epoch": 4.547570555717195, + "flow_matching_loss": 0.014524477813392878, + "grad_norm": 0.11551017314195633, + "learning_rate": 1.5200063915855107e-05, + "loss": 0.0187, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00873218085616827, + "rkd_loss_weight": 0.4791659712791443, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38592398464679717, + "rkd_teacher_entropy": 3.7472591161727906, + "rkd_teacher_similarity_mean": 0.7828995764255524, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15630 + }, + { + "epoch": 4.550480069828339, + "flow_matching_loss": 0.015806225966662167, + "grad_norm": 0.1630716621875763, + "learning_rate": 1.5183530570234212e-05, + "loss": 0.0204, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009559336956590414, + "rkd_loss_weight": 0.4788326323032379, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37950659096240996, + "rkd_teacher_entropy": 3.7407068252563476, + "rkd_teacher_similarity_mean": 0.7773817777633667, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15640 + }, + { + "epoch": 4.553389583939482, + "flow_matching_loss": 0.014411146100610495, + "grad_norm": 0.13257300853729248, + "learning_rate": 1.5166997001606361e-05, + "loss": 0.0185, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008512074081227183, + "rkd_loss_weight": 0.47849927842617035, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38948899805545806, + "rkd_teacher_entropy": 3.756593346595764, + "rkd_teacher_similarity_mean": 0.7868495941162109, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15650 + }, + { + "epoch": 4.5562990980506255, + "flow_matching_loss": 0.01597290001809597, + "grad_norm": 0.13570696115493774, + "learning_rate": 1.5150463230061417e-05, + "loss": 0.0204, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009314490342512726, + "rkd_loss_weight": 0.478165939450264, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3896933913230896, + "rkd_teacher_entropy": 3.753093194961548, + "rkd_teacher_similarity_mean": 0.7871554434299469, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15660 + }, + { + "epoch": 4.559208612161769, + "flow_matching_loss": 0.013734109047800303, + "grad_norm": 0.13681532442569733, + "learning_rate": 1.5133929275689455e-05, + "loss": 0.0175, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007900664675980806, + "rkd_loss_weight": 0.4778325974941254, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39436588883399964, + "rkd_teacher_entropy": 3.766768479347229, + "rkd_teacher_similarity_mean": 0.790648078918457, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15670 + }, + { + "epoch": 4.562118126272912, + "flow_matching_loss": 0.013170136604458094, + "grad_norm": 0.12010776996612549, + "learning_rate": 1.5117395158580795e-05, + "loss": 0.0182, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010539603559300304, + "rkd_loss_weight": 0.47749924659729004, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3862541884183884, + "rkd_teacher_entropy": 3.743207025527954, + "rkd_teacher_similarity_mean": 0.7826350390911102, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15680 + }, + { + "epoch": 4.565027640384056, + "flow_matching_loss": 0.013428636500611902, + "grad_norm": 0.1345667839050293, + "learning_rate": 1.510086089882595e-05, + "loss": 0.0176, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008832731377333402, + "rkd_loss_weight": 0.47716590762138367, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38196093440055845, + "rkd_teacher_entropy": 3.743696188926697, + "rkd_teacher_similarity_mean": 0.7790040493011474, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15690 + }, + { + "epoch": 4.567937154495199, + "flow_matching_loss": 0.016379670705646276, + "grad_norm": 0.11554092168807983, + "learning_rate": 1.5084326516515599e-05, + "loss": 0.021, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009748909575864672, + "rkd_loss_weight": 0.47683255970478056, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3796779990196228, + "rkd_teacher_entropy": 3.74106080532074, + "rkd_teacher_similarity_mean": 0.7769494116306305, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15700 + }, + { + "epoch": 4.570846668606343, + "flow_matching_loss": 0.015087543288245797, + "grad_norm": 0.11877422779798508, + "learning_rate": 1.5067792031740582e-05, + "loss": 0.0194, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009135852521285414, + "rkd_loss_weight": 0.47649921476840973, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38364273607730864, + "rkd_teacher_entropy": 3.7434067249298097, + "rkd_teacher_similarity_mean": 0.7814121782779694, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15710 + }, + { + "epoch": 4.573756182717486, + "flow_matching_loss": 0.013543921522796154, + "grad_norm": 0.16605933010578156, + "learning_rate": 1.505125746459185e-05, + "loss": 0.0179, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00917038400657475, + "rkd_loss_weight": 0.47616587579250336, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3853742778301239, + "rkd_teacher_entropy": 3.7338085412979125, + "rkd_teacher_similarity_mean": 0.7816045105457305, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15720 + }, + { + "epoch": 4.57666569682863, + "flow_matching_loss": 0.014815860334783793, + "grad_norm": 0.1533609926700592, + "learning_rate": 1.5034722835160466e-05, + "loss": 0.0204, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011666367715224623, + "rkd_loss_weight": 0.475832524895668, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3826103150844574, + "rkd_teacher_entropy": 3.728675365447998, + "rkd_teacher_similarity_mean": 0.777935528755188, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15730 + }, + { + "epoch": 4.579575210939773, + "flow_matching_loss": 0.014999578148126603, + "grad_norm": 0.16014167666435242, + "learning_rate": 1.5018188163537566e-05, + "loss": 0.0191, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008662992622703314, + "rkd_loss_weight": 0.4754991829395294, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3876750975847244, + "rkd_teacher_entropy": 3.7470038890838624, + "rkd_teacher_similarity_mean": 0.783454567193985, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15740 + }, + { + "epoch": 4.582484725050916, + "flow_matching_loss": 0.016223638877272607, + "grad_norm": 0.1129017099738121, + "learning_rate": 1.5001653469814331e-05, + "loss": 0.0204, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008771740458905697, + "rkd_loss_weight": 0.4751658409833908, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39430112540721896, + "rkd_teacher_entropy": 3.7686269998550417, + "rkd_teacher_similarity_mean": 0.7918965280056, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15750 + }, + { + "epoch": 4.58539423916206, + "flow_matching_loss": 0.01575995432212949, + "grad_norm": 0.1942344754934311, + "learning_rate": 1.4985118774081978e-05, + "loss": 0.0203, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009616471966728568, + "rkd_loss_weight": 0.4748324900865555, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38585810661315917, + "rkd_teacher_entropy": 3.7571634531021116, + "rkd_teacher_similarity_mean": 0.7821108877658844, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15760 + }, + { + "epoch": 4.588303753273204, + "flow_matching_loss": 0.016159482393413784, + "grad_norm": 0.14286689460277557, + "learning_rate": 1.4968584096431708e-05, + "loss": 0.0199, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007846231572329998, + "rkd_loss_weight": 0.4744991511106491, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3918425112962723, + "rkd_teacher_entropy": 3.759456419944763, + "rkd_teacher_similarity_mean": 0.7876537084579468, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15770 + }, + { + "epoch": 4.591213267384346, + "flow_matching_loss": 0.015584421390667558, + "grad_norm": 0.13568924367427826, + "learning_rate": 1.4952049456954731e-05, + "loss": 0.0204, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010060069663450123, + "rkd_loss_weight": 0.4741658091545105, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3872187972068787, + "rkd_teacher_entropy": 3.7414791345596314, + "rkd_teacher_similarity_mean": 0.7830815255641937, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15780 + }, + { + "epoch": 4.59412278149549, + "flow_matching_loss": 0.016236088704317808, + "grad_norm": 0.11733905971050262, + "learning_rate": 1.4935514875742182e-05, + "loss": 0.0206, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009310000063851475, + "rkd_loss_weight": 0.47383245825767517, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38461747765541077, + "rkd_teacher_entropy": 3.7277115106582643, + "rkd_teacher_similarity_mean": 0.7807893455028534, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15790 + }, + { + "epoch": 4.597032295606634, + "flow_matching_loss": 0.016070273239165545, + "grad_norm": 0.22102251648902893, + "learning_rate": 1.4918980372885143e-05, + "loss": 0.0206, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009522276977077127, + "rkd_loss_weight": 0.4734991192817688, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3928302347660065, + "rkd_teacher_entropy": 3.7598980903625487, + "rkd_teacher_similarity_mean": 0.7904703557491303, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15800 + }, + { + "epoch": 4.599941809717777, + "flow_matching_loss": 0.016004296950995922, + "grad_norm": 0.15523357689380646, + "learning_rate": 1.4902445968474591e-05, + "loss": 0.0208, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010120260016992687, + "rkd_loss_weight": 0.4731657713651657, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38092235922813417, + "rkd_teacher_entropy": 3.743788647651672, + "rkd_teacher_similarity_mean": 0.7784821689128876, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15810 + }, + { + "epoch": 4.60285132382892, + "flow_matching_loss": 0.014969139080494642, + "grad_norm": 0.11034217476844788, + "learning_rate": 1.4885911682601378e-05, + "loss": 0.0194, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009295145608484745, + "rkd_loss_weight": 0.47283242642879486, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38796730935573576, + "rkd_teacher_entropy": 3.7498926639556887, + "rkd_teacher_similarity_mean": 0.7851293683052063, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15820 + }, + { + "epoch": 4.605760837940064, + "flow_matching_loss": 0.014031477179378271, + "grad_norm": 0.1334950476884842, + "learning_rate": 1.4869377535356235e-05, + "loss": 0.0183, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009045608853921295, + "rkd_loss_weight": 0.4724990874528885, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38533680737018583, + "rkd_teacher_entropy": 3.744998574256897, + "rkd_teacher_similarity_mean": 0.7814902901649475, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15830 + }, + { + "epoch": 4.608670352051208, + "flow_matching_loss": 0.014559702388942242, + "grad_norm": 0.1339893639087677, + "learning_rate": 1.4852843546829704e-05, + "loss": 0.0188, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009025279898196459, + "rkd_loss_weight": 0.47216573655605315, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3941929191350937, + "rkd_teacher_entropy": 3.773493838310242, + "rkd_teacher_similarity_mean": 0.7893116116523743, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15840 + }, + { + "epoch": 4.611579866162351, + "flow_matching_loss": 0.016869272105395794, + "grad_norm": 0.13997703790664673, + "learning_rate": 1.483630973711214e-05, + "loss": 0.0219, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010598790552467107, + "rkd_loss_weight": 0.47183239459991455, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38391758501529694, + "rkd_teacher_entropy": 3.7367352485656737, + "rkd_teacher_similarity_mean": 0.7811962902545929, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15850 + }, + { + "epoch": 4.614489380273494, + "flow_matching_loss": 0.01741286963224411, + "grad_norm": 0.16469904780387878, + "learning_rate": 1.4819776126293677e-05, + "loss": 0.0228, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011334889568388462, + "rkd_loss_weight": 0.47149905264377595, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3813831627368927, + "rkd_teacher_entropy": 3.7301939010620115, + "rkd_teacher_similarity_mean": 0.7771961569786072, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15860 + }, + { + "epoch": 4.617398894384638, + "flow_matching_loss": 0.015458670351654291, + "grad_norm": 0.12752178311347961, + "learning_rate": 1.4803242734464214e-05, + "loss": 0.0204, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010474374005571007, + "rkd_loss_weight": 0.4711657017469406, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3986332565546036, + "rkd_teacher_entropy": 3.781224799156189, + "rkd_teacher_similarity_mean": 0.7950849294662475, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15870 + }, + { + "epoch": 4.620308408495781, + "flow_matching_loss": 0.014980649668723344, + "grad_norm": 0.1328587830066681, + "learning_rate": 1.4786709581713383e-05, + "loss": 0.0197, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010052607534453273, + "rkd_loss_weight": 0.47083236277103424, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3858889490365982, + "rkd_teacher_entropy": 3.734410810470581, + "rkd_teacher_similarity_mean": 0.782509845495224, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15880 + }, + { + "epoch": 4.6232179226069245, + "flow_matching_loss": 0.01761226076632738, + "grad_norm": 0.1327252835035324, + "learning_rate": 1.4770176688130524e-05, + "loss": 0.0215, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008270859811455011, + "rkd_loss_weight": 0.4704990178346634, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.388874351978302, + "rkd_teacher_entropy": 3.752024602890015, + "rkd_teacher_similarity_mean": 0.7861478924751282, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15890 + }, + { + "epoch": 4.626127436718068, + "flow_matching_loss": 0.014790813252329827, + "grad_norm": 0.12522055208683014, + "learning_rate": 1.475364407380466e-05, + "loss": 0.0194, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009829806629568338, + "rkd_loss_weight": 0.4701656699180603, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.381326699256897, + "rkd_teacher_entropy": 3.7417264699935915, + "rkd_teacher_similarity_mean": 0.777870637178421, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15900 + }, + { + "epoch": 4.629036950829212, + "flow_matching_loss": 0.014268972724676133, + "grad_norm": 0.12124690413475037, + "learning_rate": 1.4737111758824476e-05, + "loss": 0.0186, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009174773329868912, + "rkd_loss_weight": 0.46983233094215393, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3925314784049988, + "rkd_teacher_entropy": 3.758475184440613, + "rkd_teacher_similarity_mean": 0.7870211005210876, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15910 + }, + { + "epoch": 4.631946464940355, + "flow_matching_loss": 0.014313885755836964, + "grad_norm": 0.11979041993618011, + "learning_rate": 1.4720579763278289e-05, + "loss": 0.0187, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009263508347794413, + "rkd_loss_weight": 0.4694989800453186, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39268718659877777, + "rkd_teacher_entropy": 3.7614006280899046, + "rkd_teacher_similarity_mean": 0.7869007527828217, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15920 + }, + { + "epoch": 4.6348559790514985, + "flow_matching_loss": 0.015588729362934827, + "grad_norm": 0.16551275551319122, + "learning_rate": 1.4704048107254042e-05, + "loss": 0.0203, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010014119790866971, + "rkd_loss_weight": 0.46916563808918, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38834275007247926, + "rkd_teacher_entropy": 3.7563190698623656, + "rkd_teacher_similarity_mean": 0.783401370048523, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15930 + }, + { + "epoch": 4.637765493162642, + "flow_matching_loss": 0.014772607339546084, + "grad_norm": 0.1373894363641739, + "learning_rate": 1.4687516810839246e-05, + "loss": 0.0185, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007953368593007326, + "rkd_loss_weight": 0.4688322991132736, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38972023129463196, + "rkd_teacher_entropy": 3.747688865661621, + "rkd_teacher_similarity_mean": 0.784941452741623, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15940 + }, + { + "epoch": 4.640675007273785, + "flow_matching_loss": 0.01614005034789443, + "grad_norm": 0.17019839584827423, + "learning_rate": 1.4670985894120997e-05, + "loss": 0.0202, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008745534811168909, + "rkd_loss_weight": 0.46849894523620605, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38234101235866547, + "rkd_teacher_entropy": 3.7269593238830567, + "rkd_teacher_similarity_mean": 0.7778266549110413, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15950 + }, + { + "epoch": 4.643584521384929, + "flow_matching_loss": 0.014669785741716623, + "grad_norm": 0.09559635072946548, + "learning_rate": 1.465445537718591e-05, + "loss": 0.0192, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009597699204459787, + "rkd_loss_weight": 0.4681656062602997, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3808588832616806, + "rkd_teacher_entropy": 3.7344644546508787, + "rkd_teacher_similarity_mean": 0.7766365587711335, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15960 + }, + { + "epoch": 4.646494035496072, + "flow_matching_loss": 0.015406351443380118, + "grad_norm": 0.16342639923095703, + "learning_rate": 1.463792528012012e-05, + "loss": 0.0202, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010189609741792083, + "rkd_loss_weight": 0.46783226132392886, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3880521297454834, + "rkd_teacher_entropy": 3.7567779779434205, + "rkd_teacher_similarity_mean": 0.7847238302230835, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15970 + }, + { + "epoch": 4.649403549607215, + "flow_matching_loss": 0.016491254419088365, + "grad_norm": 0.1731080263853073, + "learning_rate": 1.4621395623009252e-05, + "loss": 0.0209, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009537659911438822, + "rkd_loss_weight": 0.46749891340732574, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38800967633724215, + "rkd_teacher_entropy": 3.752089095115662, + "rkd_teacher_similarity_mean": 0.7845723867416382, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15980 + }, + { + "epoch": 4.652313063718359, + "flow_matching_loss": 0.015950860828161238, + "grad_norm": 0.13111376762390137, + "learning_rate": 1.4604866425938407e-05, + "loss": 0.0207, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010116778314113617, + "rkd_loss_weight": 0.4671655744314194, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.383954843878746, + "rkd_teacher_entropy": 3.731543445587158, + "rkd_teacher_similarity_mean": 0.7786641418933868, + "rkd_vlm_temp": 0.10000000149011612, + "step": 15990 + }, + { + "epoch": 4.655222577829503, + "flow_matching_loss": 0.014641396049410105, + "grad_norm": 0.10180510580539703, + "learning_rate": 1.4588337708992115e-05, + "loss": 0.019, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00943245617672801, + "rkd_loss_weight": 0.46683222949504855, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39050496816635133, + "rkd_teacher_entropy": 3.752620053291321, + "rkd_teacher_similarity_mean": 0.7866589784622192, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16000 + }, + { + "epoch": 4.6581320919406455, + "flow_matching_loss": 0.013547447696328162, + "grad_norm": 0.11728070676326752, + "learning_rate": 1.4571809492254318e-05, + "loss": 0.0175, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0084456501994282, + "rkd_loss_weight": 0.46649888157844543, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3847105592489243, + "rkd_teacher_entropy": 3.7396551370620728, + "rkd_teacher_similarity_mean": 0.779957902431488, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16010 + }, + { + "epoch": 4.661041606051789, + "flow_matching_loss": 0.016457556188106535, + "grad_norm": 0.16115297377109528, + "learning_rate": 1.4555281795808371e-05, + "loss": 0.0215, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010789966862648725, + "rkd_loss_weight": 0.46616554260253906, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3784129679203033, + "rkd_teacher_entropy": 3.7331012010574343, + "rkd_teacher_similarity_mean": 0.7751053512096405, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16020 + }, + { + "epoch": 4.663951120162933, + "flow_matching_loss": 0.014061814174056053, + "grad_norm": 0.11318398267030716, + "learning_rate": 1.4538754639736966e-05, + "loss": 0.0184, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009391969349235297, + "rkd_loss_weight": 0.4658321917057037, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.391093248128891, + "rkd_teacher_entropy": 3.7588447093963624, + "rkd_teacher_similarity_mean": 0.7872952163219452, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16030 + }, + { + "epoch": 4.666860634274077, + "flow_matching_loss": 0.015079006459563971, + "grad_norm": 0.11125119030475616, + "learning_rate": 1.4522228044122174e-05, + "loss": 0.0192, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008854327024891973, + "rkd_loss_weight": 0.4654988497495651, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3916670322418213, + "rkd_teacher_entropy": 3.755782389640808, + "rkd_teacher_similarity_mean": 0.785854697227478, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16040 + }, + { + "epoch": 4.669770148385219, + "flow_matching_loss": 0.015669478476047514, + "grad_norm": 0.13749662041664124, + "learning_rate": 1.4505702029045359e-05, + "loss": 0.0198, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008884645393118262, + "rkd_loss_weight": 0.46516551077365875, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.385174024105072, + "rkd_teacher_entropy": 3.741826963424683, + "rkd_teacher_similarity_mean": 0.7811267733573913, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16050 + }, + { + "epoch": 4.672679662496363, + "flow_matching_loss": 0.016701365262269972, + "grad_norm": 0.12379880249500275, + "learning_rate": 1.4489176614587189e-05, + "loss": 0.0214, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010053586587309838, + "rkd_loss_weight": 0.4648321568965912, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3859261482954025, + "rkd_teacher_entropy": 3.74050395488739, + "rkd_teacher_similarity_mean": 0.7824146807193756, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16060 + }, + { + "epoch": 4.675589176607507, + "flow_matching_loss": 0.014359341096132993, + "grad_norm": 0.11120583117008209, + "learning_rate": 1.4472651820827595e-05, + "loss": 0.0188, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009657080844044685, + "rkd_loss_weight": 0.4644988179206848, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3889162719249725, + "rkd_teacher_entropy": 3.7589066505432127, + "rkd_teacher_similarity_mean": 0.7860940396785736, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16070 + }, + { + "epoch": 4.67849869071865, + "flow_matching_loss": 0.015999090485274793, + "grad_norm": 0.131313756108284, + "learning_rate": 1.4456127667845766e-05, + "loss": 0.0202, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009024853399023414, + "rkd_loss_weight": 0.464165472984314, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3937512248754501, + "rkd_teacher_entropy": 3.7791749238967896, + "rkd_teacher_similarity_mean": 0.790458744764328, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16080 + }, + { + "epoch": 4.681408204829793, + "flow_matching_loss": 0.01786069478839636, + "grad_norm": 0.1166650578379631, + "learning_rate": 1.4439604175720104e-05, + "loss": 0.0219, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008766274154186248, + "rkd_loss_weight": 0.4638321250677109, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3856325179338455, + "rkd_teacher_entropy": 3.74450626373291, + "rkd_teacher_similarity_mean": 0.7823145270347596, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16090 + }, + { + "epoch": 4.684317718940937, + "flow_matching_loss": 0.01227900618687272, + "grad_norm": 0.1044958233833313, + "learning_rate": 1.4423081364528212e-05, + "loss": 0.0167, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009574644081294537, + "rkd_loss_weight": 0.4634987860918045, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39195155203342436, + "rkd_teacher_entropy": 3.7603175163269045, + "rkd_teacher_similarity_mean": 0.7891174733638764, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16100 + }, + { + "epoch": 4.687227233052081, + "flow_matching_loss": 0.014885987248271703, + "grad_norm": 0.1193845346570015, + "learning_rate": 1.4406559254346863e-05, + "loss": 0.0192, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00932338610291481, + "rkd_loss_weight": 0.4631654381752014, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.383341908454895, + "rkd_teacher_entropy": 3.738085627555847, + "rkd_teacher_similarity_mean": 0.7796183466911316, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16110 + }, + { + "epoch": 4.690136747163224, + "flow_matching_loss": 0.014013688359409571, + "grad_norm": 0.11129748821258545, + "learning_rate": 1.4390037865251978e-05, + "loss": 0.0184, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009428019169718027, + "rkd_loss_weight": 0.46283209323883057, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38636578917503356, + "rkd_teacher_entropy": 3.7478683471679686, + "rkd_teacher_similarity_mean": 0.7831884980201721, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16120 + }, + { + "epoch": 4.693046261274367, + "flow_matching_loss": 0.015427169669419528, + "grad_norm": 0.16761688888072968, + "learning_rate": 1.4373517217318597e-05, + "loss": 0.0198, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009529882995411753, + "rkd_loss_weight": 0.4624987542629242, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38389811515808103, + "rkd_teacher_entropy": 3.7366447925567625, + "rkd_teacher_similarity_mean": 0.7805927038192749, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16130 + }, + { + "epoch": 4.69595577538551, + "flow_matching_loss": 0.0158530879765749, + "grad_norm": 0.1331115961074829, + "learning_rate": 1.4356997330620875e-05, + "loss": 0.0192, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007144275773316622, + "rkd_loss_weight": 0.46216540038585663, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3807265341281891, + "rkd_teacher_entropy": 3.734182095527649, + "rkd_teacher_similarity_mean": 0.7784373521804809, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16140 + }, + { + "epoch": 4.698865289496654, + "flow_matching_loss": 0.015518607292324304, + "grad_norm": 0.18015097081661224, + "learning_rate": 1.434047822523203e-05, + "loss": 0.0198, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009374215686693787, + "rkd_loss_weight": 0.46183206140995026, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37973888516426085, + "rkd_teacher_entropy": 3.7239936113357546, + "rkd_teacher_similarity_mean": 0.7761738896369934, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16150 + }, + { + "epoch": 4.7017748036077975, + "flow_matching_loss": 0.014299721550196409, + "grad_norm": 0.1613825261592865, + "learning_rate": 1.4323959921224322e-05, + "loss": 0.0179, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007844665134325624, + "rkd_loss_weight": 0.46149871945381166, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3915173977613449, + "rkd_teacher_entropy": 3.7673641443252563, + "rkd_teacher_similarity_mean": 0.7894605100154877, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16160 + }, + { + "epoch": 4.704684317718941, + "flow_matching_loss": 0.014115851931273937, + "grad_norm": 0.13650326430797577, + "learning_rate": 1.4307442438669066e-05, + "loss": 0.0181, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008727483870461582, + "rkd_loss_weight": 0.4611653685569763, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.394605764746666, + "rkd_teacher_entropy": 3.7613049030303953, + "rkd_teacher_similarity_mean": 0.7905068576335907, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16170 + }, + { + "epoch": 4.707593831830084, + "flow_matching_loss": 0.016097024269402026, + "grad_norm": 0.19529752433300018, + "learning_rate": 1.4290925797636546e-05, + "loss": 0.0208, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010162109881639481, + "rkd_loss_weight": 0.46083202958106995, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37859717905521395, + "rkd_teacher_entropy": 3.722956085205078, + "rkd_teacher_similarity_mean": 0.7749413669109344, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16180 + }, + { + "epoch": 4.710503345941228, + "flow_matching_loss": 0.015759071428328754, + "grad_norm": 0.15596133470535278, + "learning_rate": 1.4274410018196039e-05, + "loss": 0.0193, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007750593591481447, + "rkd_loss_weight": 0.46049868166446684, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3829242318868637, + "rkd_teacher_entropy": 3.7323686838150025, + "rkd_teacher_similarity_mean": 0.7802250444889068, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16190 + }, + { + "epoch": 4.7134128600523715, + "flow_matching_loss": 0.017053802590817214, + "grad_norm": 0.11767198145389557, + "learning_rate": 1.4257895120415783e-05, + "loss": 0.0213, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009200500836595894, + "rkd_loss_weight": 0.460165336728096, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38109962046146395, + "rkd_teacher_entropy": 3.733890676498413, + "rkd_teacher_similarity_mean": 0.7788096129894256, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16200 + }, + { + "epoch": 4.716322374163514, + "flow_matching_loss": 0.016675823740661143, + "grad_norm": 0.1258300393819809, + "learning_rate": 1.4241381124362933e-05, + "loss": 0.0206, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008593093883246184, + "rkd_loss_weight": 0.45983199775218964, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38134238719940183, + "rkd_teacher_entropy": 3.7455288410186767, + "rkd_teacher_similarity_mean": 0.7802065134048461, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16210 + }, + { + "epoch": 4.719231888274658, + "flow_matching_loss": 0.01628963677212596, + "grad_norm": 0.11948138475418091, + "learning_rate": 1.4224868050103546e-05, + "loss": 0.0202, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00854114694520831, + "rkd_loss_weight": 0.4594986468553543, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3856598436832428, + "rkd_teacher_entropy": 3.742602586746216, + "rkd_teacher_similarity_mean": 0.7828816771507263, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16220 + }, + { + "epoch": 4.722141402385802, + "flow_matching_loss": 0.013469352945685386, + "grad_norm": 0.10416506230831146, + "learning_rate": 1.4208355917702573e-05, + "loss": 0.0172, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008159477217122912, + "rkd_loss_weight": 0.4591653048992157, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3840829849243164, + "rkd_teacher_entropy": 3.7425554275512694, + "rkd_teacher_similarity_mean": 0.7823412716388702, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16230 + }, + { + "epoch": 4.725050916496945, + "flow_matching_loss": 0.012323667900636793, + "grad_norm": 0.14918406307697296, + "learning_rate": 1.41918447472238e-05, + "loss": 0.0168, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009812349407002329, + "rkd_loss_weight": 0.4588319659233093, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39028979539871217, + "rkd_teacher_entropy": 3.762683129310608, + "rkd_teacher_similarity_mean": 0.7876702845096588, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16240 + }, + { + "epoch": 4.727960430608088, + "flow_matching_loss": 0.016340806661173702, + "grad_norm": 0.1251564472913742, + "learning_rate": 1.4175334558729869e-05, + "loss": 0.0202, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008491035690531135, + "rkd_loss_weight": 0.45849861204624176, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3816111385822296, + "rkd_teacher_entropy": 3.7279937982559206, + "rkd_teacher_similarity_mean": 0.7790988683700562, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16250 + }, + { + "epoch": 4.730869944719232, + "flow_matching_loss": 0.015879856841638684, + "grad_norm": 0.10575098544359207, + "learning_rate": 1.4158825372282211e-05, + "loss": 0.0205, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010004031099379062, + "rkd_loss_weight": 0.4581652730703354, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3913699984550476, + "rkd_teacher_entropy": 3.768678832054138, + "rkd_teacher_similarity_mean": 0.7886599838733673, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16260 + }, + { + "epoch": 4.733779458830376, + "flow_matching_loss": 0.016410147305577993, + "grad_norm": 0.12538853287696838, + "learning_rate": 1.4142317207941042e-05, + "loss": 0.02, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00790631934069097, + "rkd_loss_weight": 0.4578319311141968, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3861207187175751, + "rkd_teacher_entropy": 3.749780559539795, + "rkd_teacher_similarity_mean": 0.7822514832019806, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16270 + }, + { + "epoch": 4.7366889729415185, + "flow_matching_loss": 0.01597724054008722, + "grad_norm": 0.10508973896503448, + "learning_rate": 1.4125810085765339e-05, + "loss": 0.0203, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00937026720494032, + "rkd_loss_weight": 0.45749858021736145, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3905534207820892, + "rkd_teacher_entropy": 3.7560924768447874, + "rkd_teacher_similarity_mean": 0.7856193125247956, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16280 + }, + { + "epoch": 4.739598487052662, + "flow_matching_loss": 0.01948058055713773, + "grad_norm": 0.14520952105522156, + "learning_rate": 1.410930402581281e-05, + "loss": 0.0238, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009515751898288728, + "rkd_loss_weight": 0.4571652412414551, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3903353989124298, + "rkd_teacher_entropy": 3.7572060585021974, + "rkd_teacher_similarity_mean": 0.7865484535694123, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16290 + }, + { + "epoch": 4.742508001163806, + "flow_matching_loss": 0.016636367328464984, + "grad_norm": 0.11760727316141129, + "learning_rate": 1.4092799048139876e-05, + "loss": 0.0201, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007652377057820558, + "rkd_loss_weight": 0.45683189332485197, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39102972447872164, + "rkd_teacher_entropy": 3.761765718460083, + "rkd_teacher_similarity_mean": 0.7879501938819885, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16300 + }, + { + "epoch": 4.745417515274949, + "flow_matching_loss": 0.014291183277964591, + "grad_norm": 0.10875878483057022, + "learning_rate": 1.4076295172801647e-05, + "loss": 0.0187, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009759938297793268, + "rkd_loss_weight": 0.45649854838848114, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3899804353713989, + "rkd_teacher_entropy": 3.7508010387420656, + "rkd_teacher_similarity_mean": 0.7865015029907226, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16310 + }, + { + "epoch": 4.748327029386092, + "flow_matching_loss": 0.014864725805819035, + "grad_norm": 0.1475476175546646, + "learning_rate": 1.4059792419851884e-05, + "loss": 0.0194, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00987413888797164, + "rkd_loss_weight": 0.45616520941257477, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37693079710006716, + "rkd_teacher_entropy": 3.719870400428772, + "rkd_teacher_similarity_mean": 0.7737183988094329, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16320 + }, + { + "epoch": 4.751236543497236, + "flow_matching_loss": 0.013837451301515102, + "grad_norm": 0.20112408697605133, + "learning_rate": 1.4043290809342984e-05, + "loss": 0.0181, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009440224152058363, + "rkd_loss_weight": 0.45583185851573943, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39523318111896516, + "rkd_teacher_entropy": 3.7811736106872558, + "rkd_teacher_similarity_mean": 0.7933356285095214, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16330 + }, + { + "epoch": 4.754146057608379, + "flow_matching_loss": 0.016299525368958713, + "grad_norm": 0.1478998064994812, + "learning_rate": 1.4026790361325958e-05, + "loss": 0.0206, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009443998150527478, + "rkd_loss_weight": 0.45549851655960083, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3840632438659668, + "rkd_teacher_entropy": 3.743922305107117, + "rkd_teacher_similarity_mean": 0.7813034534454346, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16340 + }, + { + "epoch": 4.757055571719523, + "flow_matching_loss": 0.012710078712552786, + "grad_norm": 0.11518093198537827, + "learning_rate": 1.4010291095850415e-05, + "loss": 0.0165, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008280854020267726, + "rkd_loss_weight": 0.45516517460346223, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39048852622509, + "rkd_teacher_entropy": 3.758874225616455, + "rkd_teacher_similarity_mean": 0.7875153064727783, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16350 + }, + { + "epoch": 4.759965085830666, + "flow_matching_loss": 0.016076556593179702, + "grad_norm": 0.19343321025371552, + "learning_rate": 1.3993793032964515e-05, + "loss": 0.02, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008727744920179248, + "rkd_loss_weight": 0.4548318237066269, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38329629600048065, + "rkd_teacher_entropy": 3.73451886177063, + "rkd_teacher_similarity_mean": 0.7801028609275817, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16360 + }, + { + "epoch": 4.76287459994181, + "flow_matching_loss": 0.0129342470318079, + "grad_norm": 0.09596683084964752, + "learning_rate": 1.3977296192714956e-05, + "loss": 0.0162, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007235393719747662, + "rkd_loss_weight": 0.4544984847307205, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38157893121242525, + "rkd_teacher_entropy": 3.731699800491333, + "rkd_teacher_similarity_mean": 0.7784791648387909, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16370 + }, + { + "epoch": 4.765784114052953, + "flow_matching_loss": 0.014287403319031, + "grad_norm": 0.12154729664325714, + "learning_rate": 1.396080059514696e-05, + "loss": 0.0187, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009644236089661718, + "rkd_loss_weight": 0.4541651397943497, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38458074629306793, + "rkd_teacher_entropy": 3.753644251823425, + "rkd_teacher_similarity_mean": 0.7819950819015503, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16380 + }, + { + "epoch": 4.768693628164097, + "flow_matching_loss": 0.016327610053122044, + "grad_norm": 0.1319584995508194, + "learning_rate": 1.3944306260304232e-05, + "loss": 0.0203, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008661342039704322, + "rkd_loss_weight": 0.4538317918777466, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3896676987409592, + "rkd_teacher_entropy": 3.7655247688293456, + "rkd_teacher_similarity_mean": 0.7863612473011017, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16390 + }, + { + "epoch": 4.77160314227524, + "flow_matching_loss": 0.013357165642082692, + "grad_norm": 0.12740103900432587, + "learning_rate": 1.3927813208228936e-05, + "loss": 0.0167, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00743536651134491, + "rkd_loss_weight": 0.4534984529018402, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3877872467041016, + "rkd_teacher_entropy": 3.765086531639099, + "rkd_teacher_similarity_mean": 0.7860346376895905, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16400 + }, + { + "epoch": 4.774512656386383, + "flow_matching_loss": 0.014670388400554657, + "grad_norm": 0.1360052078962326, + "learning_rate": 1.39113214589617e-05, + "loss": 0.0193, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01016053995117545, + "rkd_loss_weight": 0.45316510200500487, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3859191030263901, + "rkd_teacher_entropy": 3.738920259475708, + "rkd_teacher_similarity_mean": 0.7817574858665466, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16410 + }, + { + "epoch": 4.777422170497527, + "flow_matching_loss": 0.014995530154556036, + "grad_norm": 0.09582250565290451, + "learning_rate": 1.3894831032541545e-05, + "loss": 0.0194, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009695097664371133, + "rkd_loss_weight": 0.45283176004886627, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3971705287694931, + "rkd_teacher_entropy": 3.7783612966537476, + "rkd_teacher_similarity_mean": 0.7944601833820343, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16420 + }, + { + "epoch": 4.7803316846086705, + "flow_matching_loss": 0.014541258569806814, + "grad_norm": 0.11889338493347168, + "learning_rate": 1.3878341949005895e-05, + "loss": 0.0185, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008742892276495695, + "rkd_loss_weight": 0.4524984210729599, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3889907896518707, + "rkd_teacher_entropy": 3.7592159271240235, + "rkd_teacher_similarity_mean": 0.7872393131256104, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16430 + }, + { + "epoch": 4.783241198719814, + "flow_matching_loss": 0.013389364443719387, + "grad_norm": 0.09916619211435318, + "learning_rate": 1.3861854228390546e-05, + "loss": 0.0167, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007376239402219653, + "rkd_loss_weight": 0.45216506719589233, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38898330330848696, + "rkd_teacher_entropy": 3.7455806970596313, + "rkd_teacher_similarity_mean": 0.7848682045936585, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16440 + }, + { + "epoch": 4.786150712830957, + "flow_matching_loss": 0.015714614558964966, + "grad_norm": 0.146769717335701, + "learning_rate": 1.3845367890729621e-05, + "loss": 0.0204, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010471797827631236, + "rkd_loss_weight": 0.45183172821998596, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3897237956523895, + "rkd_teacher_entropy": 3.766534352302551, + "rkd_teacher_similarity_mean": 0.7861898243427277, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16450 + }, + { + "epoch": 4.789060226942101, + "flow_matching_loss": 0.012736823689192534, + "grad_norm": 0.09822241961956024, + "learning_rate": 1.3828882956055591e-05, + "loss": 0.0164, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008091743476688862, + "rkd_loss_weight": 0.45149838626384736, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3871641010046005, + "rkd_teacher_entropy": 3.7482367753982544, + "rkd_teacher_similarity_mean": 0.7850597500801086, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16460 + }, + { + "epoch": 4.7919697410532445, + "flow_matching_loss": 0.014125462248921394, + "grad_norm": 0.12119465321302414, + "learning_rate": 1.38123994443992e-05, + "loss": 0.0173, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007115744473412633, + "rkd_loss_weight": 0.451165035367012, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37860511541366576, + "rkd_teacher_entropy": 3.7212722063064576, + "rkd_teacher_similarity_mean": 0.7747478067874909, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16470 + }, + { + "epoch": 4.794879255164387, + "flow_matching_loss": 0.017271231301128863, + "grad_norm": 0.1468982696533203, + "learning_rate": 1.3795917375789465e-05, + "loss": 0.0211, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00844237576238811, + "rkd_loss_weight": 0.45083169639110565, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3949080765247345, + "rkd_teacher_entropy": 3.7793610811233522, + "rkd_teacher_similarity_mean": 0.7914578974246979, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16480 + }, + { + "epoch": 4.797788769275531, + "flow_matching_loss": 0.01588433180004358, + "grad_norm": 0.13991719484329224, + "learning_rate": 1.3779436770253658e-05, + "loss": 0.0197, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008462911238893867, + "rkd_loss_weight": 0.4504983514547348, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3949684053659439, + "rkd_teacher_entropy": 3.7633102416992186, + "rkd_teacher_similarity_mean": 0.7916171610355377, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16490 + }, + { + "epoch": 4.800698283386675, + "flow_matching_loss": 0.012174562830477954, + "grad_norm": 0.12717734277248383, + "learning_rate": 1.3762957647817258e-05, + "loss": 0.0159, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008172076288610696, + "rkd_loss_weight": 0.4501650035381317, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39028871059417725, + "rkd_teacher_entropy": 3.7506423234939574, + "rkd_teacher_similarity_mean": 0.7855149805545807, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16500 + }, + { + "epoch": 4.8036077974978175, + "flow_matching_loss": 0.014738254388794303, + "grad_norm": 0.16773822903633118, + "learning_rate": 1.374648002850397e-05, + "loss": 0.0192, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00998387518338859, + "rkd_loss_weight": 0.44983166456222534, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3865087300539017, + "rkd_teacher_entropy": 3.7482754945755006, + "rkd_teacher_similarity_mean": 0.7811686277389527, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16510 + }, + { + "epoch": 4.806517311608961, + "flow_matching_loss": 0.013973601162433624, + "grad_norm": 0.14009231328964233, + "learning_rate": 1.3730003932335642e-05, + "loss": 0.0173, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007346889888867736, + "rkd_loss_weight": 0.44949831366539, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3849385201931, + "rkd_teacher_entropy": 3.733507513999939, + "rkd_teacher_similarity_mean": 0.7828215539455414, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16520 + }, + { + "epoch": 4.809426825720105, + "flow_matching_loss": 0.014202863164246082, + "grad_norm": 0.1384524255990982, + "learning_rate": 1.3713529379332292e-05, + "loss": 0.0184, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009285189630463719, + "rkd_loss_weight": 0.4491649717092514, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3885956406593323, + "rkd_teacher_entropy": 3.7516103982925415, + "rkd_teacher_similarity_mean": 0.7848836421966553, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16530 + }, + { + "epoch": 4.812336339831248, + "flow_matching_loss": 0.013685667049139738, + "grad_norm": 0.11449544131755829, + "learning_rate": 1.3697056389512054e-05, + "loss": 0.0191, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.012167370272800326, + "rkd_loss_weight": 0.44883163273334503, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38627262115478517, + "rkd_teacher_entropy": 3.7410914421081545, + "rkd_teacher_similarity_mean": 0.7803571581840515, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16540 + }, + { + "epoch": 4.8152458539423915, + "flow_matching_loss": 0.01464482992887497, + "grad_norm": 0.11837508529424667, + "learning_rate": 1.3680584982891162e-05, + "loss": 0.0186, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008743258658796549, + "rkd_loss_weight": 0.44849827885627747, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3914063721895218, + "rkd_teacher_entropy": 3.766029453277588, + "rkd_teacher_similarity_mean": 0.7878912210464477, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16550 + }, + { + "epoch": 4.818155368053535, + "flow_matching_loss": 0.014734086487442255, + "grad_norm": 0.10220171511173248, + "learning_rate": 1.3664115179483921e-05, + "loss": 0.0186, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008718914678320288, + "rkd_loss_weight": 0.4481649398803711, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3914694875478745, + "rkd_teacher_entropy": 3.777697134017944, + "rkd_teacher_similarity_mean": 0.7893478214740753, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16560 + }, + { + "epoch": 4.821064882164679, + "flow_matching_loss": 0.014729698747396469, + "grad_norm": 0.11475520581007004, + "learning_rate": 1.364764699930271e-05, + "loss": 0.0182, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007816631346940994, + "rkd_loss_weight": 0.44783159494400027, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3839704513549805, + "rkd_teacher_entropy": 3.7485999584198, + "rkd_teacher_similarity_mean": 0.7824200212955474, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16570 + }, + { + "epoch": 4.823974396275822, + "flow_matching_loss": 0.015947508905082942, + "grad_norm": 0.13398949801921844, + "learning_rate": 1.363118046235791e-05, + "loss": 0.0198, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008719906397163867, + "rkd_loss_weight": 0.44749824702739716, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3920761436223984, + "rkd_teacher_entropy": 3.767094039916992, + "rkd_teacher_similarity_mean": 0.7901588678359985, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16580 + }, + { + "epoch": 4.826883910386965, + "flow_matching_loss": 0.016161994263529777, + "grad_norm": 0.11242518573999405, + "learning_rate": 1.3614715588657921e-05, + "loss": 0.0205, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009730301471427083, + "rkd_loss_weight": 0.4471649080514908, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39094132781028745, + "rkd_teacher_entropy": 3.7620081424713137, + "rkd_teacher_similarity_mean": 0.787828117609024, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16590 + }, + { + "epoch": 4.829793424498109, + "flow_matching_loss": 0.018357780389487745, + "grad_norm": 0.10891557484865189, + "learning_rate": 1.3598252398209118e-05, + "loss": 0.0224, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009005799936130644, + "rkd_loss_weight": 0.4468315601348877, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38110066056251524, + "rkd_teacher_entropy": 3.738075852394104, + "rkd_teacher_similarity_mean": 0.778520530462265, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16600 + }, + { + "epoch": 4.832702938609252, + "flow_matching_loss": 0.014077302487567066, + "grad_norm": 0.13655202090740204, + "learning_rate": 1.358179091101582e-05, + "loss": 0.0185, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009797657420858742, + "rkd_loss_weight": 0.44649821519851685, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3885165125131607, + "rkd_teacher_entropy": 3.753691029548645, + "rkd_teacher_similarity_mean": 0.7862050414085389, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16610 + }, + { + "epoch": 4.835612452720396, + "flow_matching_loss": 0.014898120146244764, + "grad_norm": 0.12215575575828552, + "learning_rate": 1.3565331147080299e-05, + "loss": 0.0186, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008266754494979978, + "rkd_loss_weight": 0.4461648762226105, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37928909063339233, + "rkd_teacher_entropy": 3.743750476837158, + "rkd_teacher_similarity_mean": 0.7777012884616852, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16620 + }, + { + "epoch": 4.838521966831539, + "flow_matching_loss": 0.01427306206896901, + "grad_norm": 0.11696159839630127, + "learning_rate": 1.3548873126402716e-05, + "loss": 0.018, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008279418060556053, + "rkd_loss_weight": 0.4458315223455429, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3872001737356186, + "rkd_teacher_entropy": 3.753273868560791, + "rkd_teacher_similarity_mean": 0.7851507246494294, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16630 + }, + { + "epoch": 4.841431480942683, + "flow_matching_loss": 0.016779970936477186, + "grad_norm": 0.13912853598594666, + "learning_rate": 1.3532416868981117e-05, + "loss": 0.0202, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007643310073763132, + "rkd_loss_weight": 0.44549818336963654, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.381782266497612, + "rkd_teacher_entropy": 3.739902400970459, + "rkd_teacher_similarity_mean": 0.7783509194850922, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16640 + }, + { + "epoch": 4.844340995053826, + "flow_matching_loss": 0.015033865068107843, + "grad_norm": 0.1236300989985466, + "learning_rate": 1.3515962394811401e-05, + "loss": 0.0193, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009491972625255585, + "rkd_loss_weight": 0.44516484141349794, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38001956343650817, + "rkd_teacher_entropy": 3.733230400085449, + "rkd_teacher_similarity_mean": 0.7759272456169128, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16650 + }, + { + "epoch": 4.84725050916497, + "flow_matching_loss": 0.015269838552922011, + "grad_norm": 0.1404808759689331, + "learning_rate": 1.3499509723887313e-05, + "loss": 0.0186, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007593165384605527, + "rkd_loss_weight": 0.4448314905166626, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38660902678966524, + "rkd_teacher_entropy": 3.7490856409072877, + "rkd_teacher_similarity_mean": 0.783628499507904, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16660 + }, + { + "epoch": 4.850160023276112, + "flow_matching_loss": 0.01736106164753437, + "grad_norm": 0.11510219424962997, + "learning_rate": 1.3483058876200396e-05, + "loss": 0.0212, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008637631358578802, + "rkd_loss_weight": 0.4444981515407562, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38258245289325715, + "rkd_teacher_entropy": 3.7471537828445434, + "rkd_teacher_similarity_mean": 0.7798565924167633, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16670 + }, + { + "epoch": 4.853069537387256, + "flow_matching_loss": 0.015622807387262582, + "grad_norm": 0.15200237929821014, + "learning_rate": 1.3466609871739987e-05, + "loss": 0.0195, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008798568788915873, + "rkd_loss_weight": 0.4441648066043854, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38822206258773806, + "rkd_teacher_entropy": 3.746212697029114, + "rkd_teacher_similarity_mean": 0.7835648834705353, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16680 + }, + { + "epoch": 4.8559790514984, + "flow_matching_loss": 0.017686393950134514, + "grad_norm": 0.1326553374528885, + "learning_rate": 1.3450162730493172e-05, + "loss": 0.0212, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007913734903559089, + "rkd_loss_weight": 0.4438314586877823, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3885331451892853, + "rkd_teacher_entropy": 3.7508071422576905, + "rkd_teacher_similarity_mean": 0.7840459883213043, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16690 + }, + { + "epoch": 4.8588885656095435, + "flow_matching_loss": 0.014568758942186833, + "grad_norm": 0.1132860779762268, + "learning_rate": 1.3433717472444781e-05, + "loss": 0.0187, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00926488833501935, + "rkd_loss_weight": 0.4434981197118759, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38800759315490724, + "rkd_teacher_entropy": 3.755470132827759, + "rkd_teacher_similarity_mean": 0.7850421309471131, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16700 + }, + { + "epoch": 4.861798079720686, + "flow_matching_loss": 0.013346158340573311, + "grad_norm": 0.13566990196704865, + "learning_rate": 1.3417274117577343e-05, + "loss": 0.0177, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009820146067067981, + "rkd_loss_weight": 0.4431647717952728, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3885887682437897, + "rkd_teacher_entropy": 3.76058144569397, + "rkd_teacher_similarity_mean": 0.7842707633972168, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16710 + }, + { + "epoch": 4.86470759383183, + "flow_matching_loss": 0.013783876970410346, + "grad_norm": 0.10240406543016434, + "learning_rate": 1.3400832685871102e-05, + "loss": 0.0178, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009068318409845233, + "rkd_loss_weight": 0.442831426858902, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.388415601849556, + "rkd_teacher_entropy": 3.741869878768921, + "rkd_teacher_similarity_mean": 0.7833279252052308, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16720 + }, + { + "epoch": 4.867617107942974, + "flow_matching_loss": 0.014191335253417491, + "grad_norm": 0.16537585854530334, + "learning_rate": 1.3384393197303936e-05, + "loss": 0.0185, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009719647839665413, + "rkd_loss_weight": 0.4424980878829956, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3974566698074341, + "rkd_teacher_entropy": 3.7867506742477417, + "rkd_teacher_similarity_mean": 0.7943146586418152, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16730 + }, + { + "epoch": 4.870526622054117, + "flow_matching_loss": 0.013799358531832695, + "grad_norm": 0.13640426099300385, + "learning_rate": 1.336795567185138e-05, + "loss": 0.0174, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00814897627569735, + "rkd_loss_weight": 0.44216473400592804, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3904894798994064, + "rkd_teacher_entropy": 3.7507652282714843, + "rkd_teacher_similarity_mean": 0.7865561425685883, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16740 + }, + { + "epoch": 4.87343613616526, + "flow_matching_loss": 0.01591430902481079, + "grad_norm": 0.10797498375177383, + "learning_rate": 1.3351520129486573e-05, + "loss": 0.0193, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0076416367664933205, + "rkd_loss_weight": 0.44183139503002167, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3975854843854904, + "rkd_teacher_entropy": 3.7755707502365112, + "rkd_teacher_similarity_mean": 0.7940646648406983, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16750 + }, + { + "epoch": 4.876345650276404, + "flow_matching_loss": 0.01418664138764143, + "grad_norm": 0.11862216889858246, + "learning_rate": 1.3335086590180252e-05, + "loss": 0.0185, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009755283081904053, + "rkd_loss_weight": 0.44149805307388307, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38583631813526154, + "rkd_teacher_entropy": 3.7417086362838745, + "rkd_teacher_similarity_mean": 0.7819791376590729, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16760 + }, + { + "epoch": 4.879255164387548, + "flow_matching_loss": 0.01481955461204052, + "grad_norm": 0.12529711425304413, + "learning_rate": 1.3318655073900708e-05, + "loss": 0.0184, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008091191947460174, + "rkd_loss_weight": 0.44116470217704773, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.382436403632164, + "rkd_teacher_entropy": 3.7345171928405763, + "rkd_teacher_similarity_mean": 0.7785740256309509, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16770 + }, + { + "epoch": 4.8821646784986905, + "flow_matching_loss": 0.014891039952635765, + "grad_norm": 0.14907346665859222, + "learning_rate": 1.3302225600613796e-05, + "loss": 0.0188, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008832384459674358, + "rkd_loss_weight": 0.44083136320114136, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39178000688552855, + "rkd_teacher_entropy": 3.750535583496094, + "rkd_teacher_similarity_mean": 0.7873482763767242, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16780 + }, + { + "epoch": 4.885074192609834, + "flow_matching_loss": 0.013717815652489663, + "grad_norm": 0.09618689119815826, + "learning_rate": 1.3285798190282869e-05, + "loss": 0.0171, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0076053245458751915, + "rkd_loss_weight": 0.44049801528453825, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.382876792550087, + "rkd_teacher_entropy": 3.736726689338684, + "rkd_teacher_similarity_mean": 0.7799051821231842, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16790 + }, + { + "epoch": 4.887983706720978, + "flow_matching_loss": 0.015313946548849345, + "grad_norm": 0.18608510494232178, + "learning_rate": 1.3269372862868776e-05, + "loss": 0.019, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008423927519470454, + "rkd_loss_weight": 0.4401646703481674, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39008342325687406, + "rkd_teacher_entropy": 3.7491461992263795, + "rkd_teacher_similarity_mean": 0.785614800453186, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16800 + }, + { + "epoch": 4.890893220832121, + "flow_matching_loss": 0.014875093661248683, + "grad_norm": 0.13977333903312683, + "learning_rate": 1.3252949638329845e-05, + "loss": 0.0183, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007787360576912761, + "rkd_loss_weight": 0.43983133137226105, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39047546982765197, + "rkd_teacher_entropy": 3.747819519042969, + "rkd_teacher_similarity_mean": 0.7867685496807099, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16810 + }, + { + "epoch": 4.8938027349432645, + "flow_matching_loss": 0.015210855938494205, + "grad_norm": 0.12717556953430176, + "learning_rate": 1.3236528536621834e-05, + "loss": 0.0198, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010404212214052678, + "rkd_loss_weight": 0.4394979804754257, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37993282079696655, + "rkd_teacher_entropy": 3.7225666284561156, + "rkd_teacher_similarity_mean": 0.7749350905418396, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16820 + }, + { + "epoch": 4.896712249054408, + "flow_matching_loss": 0.01616974249482155, + "grad_norm": 0.1498907208442688, + "learning_rate": 1.3220109577697943e-05, + "loss": 0.0207, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010266018472611904, + "rkd_loss_weight": 0.4391646385192871, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3863260567188263, + "rkd_teacher_entropy": 3.751391577720642, + "rkd_teacher_similarity_mean": 0.7812528371810913, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16830 + }, + { + "epoch": 4.899621763165551, + "flow_matching_loss": 0.014083653036504985, + "grad_norm": 0.11548786610364914, + "learning_rate": 1.3203692781508745e-05, + "loss": 0.018, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008858457999303937, + "rkd_loss_weight": 0.4388312965631485, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38729516565799715, + "rkd_teacher_entropy": 3.7447236299514772, + "rkd_teacher_similarity_mean": 0.7829338073730469, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16840 + }, + { + "epoch": 4.902531277276695, + "flow_matching_loss": 0.015452941227704287, + "grad_norm": 0.13128319382667542, + "learning_rate": 1.3187278168002198e-05, + "loss": 0.0194, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00908687929622829, + "rkd_loss_weight": 0.43849794566631317, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39094823598861694, + "rkd_teacher_entropy": 3.7660931587219237, + "rkd_teacher_similarity_mean": 0.7872149407863617, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16850 + }, + { + "epoch": 4.905440791387838, + "flow_matching_loss": 0.014242711197584867, + "grad_norm": 0.14328056573867798, + "learning_rate": 1.3170865757123606e-05, + "loss": 0.0184, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009422347042709589, + "rkd_loss_weight": 0.4381646066904068, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3918573886156082, + "rkd_teacher_entropy": 3.771160435676575, + "rkd_teacher_similarity_mean": 0.7881142258644104, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16860 + }, + { + "epoch": 4.908350305498981, + "flow_matching_loss": 0.014384433766826988, + "grad_norm": 0.14890757203102112, + "learning_rate": 1.3154455568815592e-05, + "loss": 0.018, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008204107591882347, + "rkd_loss_weight": 0.43783126175403597, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39021443426609037, + "rkd_teacher_entropy": 3.7588591814041137, + "rkd_teacher_similarity_mean": 0.787041038274765, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16870 + }, + { + "epoch": 4.911259819610125, + "flow_matching_loss": 0.013548502139747143, + "grad_norm": 0.15170732140541077, + "learning_rate": 1.3138047623018087e-05, + "loss": 0.0181, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010346473986282944, + "rkd_loss_weight": 0.43749791383743286, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3928408056497574, + "rkd_teacher_entropy": 3.7613388538360595, + "rkd_teacher_similarity_mean": 0.790706580877304, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16880 + }, + { + "epoch": 4.914169333721269, + "flow_matching_loss": 0.013075528526678681, + "grad_norm": 0.09789878875017166, + "learning_rate": 1.3121641939668293e-05, + "loss": 0.0162, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007156109344214201, + "rkd_loss_weight": 0.4371645748615265, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39477090537548065, + "rkd_teacher_entropy": 3.766971707344055, + "rkd_teacher_similarity_mean": 0.7915410339832306, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16890 + }, + { + "epoch": 4.917078847832412, + "flow_matching_loss": 0.016757176723331213, + "grad_norm": 0.12440808117389679, + "learning_rate": 1.3105238538700659e-05, + "loss": 0.0206, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008873674506321549, + "rkd_loss_weight": 0.4368312269449234, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3925904154777527, + "rkd_teacher_entropy": 3.7586645126342773, + "rkd_teacher_similarity_mean": 0.7879855275154114, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16900 + }, + { + "epoch": 4.919988361943555, + "flow_matching_loss": 0.013164891675114632, + "grad_norm": 0.09759111702442169, + "learning_rate": 1.3088837440046866e-05, + "loss": 0.0166, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007934551872313022, + "rkd_loss_weight": 0.43649788200855255, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39293172359466555, + "rkd_teacher_entropy": 3.760398530960083, + "rkd_teacher_similarity_mean": 0.7894979596138001, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16910 + }, + { + "epoch": 4.922897876054699, + "flow_matching_loss": 0.01410817876458168, + "grad_norm": 0.12422437220811844, + "learning_rate": 1.3072438663635787e-05, + "loss": 0.0183, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009634113265201449, + "rkd_loss_weight": 0.4361645430326462, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3868144780397415, + "rkd_teacher_entropy": 3.7486738204956054, + "rkd_teacher_similarity_mean": 0.7848834276199341, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16920 + }, + { + "epoch": 4.925807390165843, + "flow_matching_loss": 0.014751943573355675, + "grad_norm": 0.12497567385435104, + "learning_rate": 1.3056042229393497e-05, + "loss": 0.0186, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008831624733284115, + "rkd_loss_weight": 0.43583119213581084, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3837065815925598, + "rkd_teacher_entropy": 3.7426422834396362, + "rkd_teacher_similarity_mean": 0.7806344866752625, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16930 + }, + { + "epoch": 4.928716904276985, + "flow_matching_loss": 0.014149879384785891, + "grad_norm": 0.12179902195930481, + "learning_rate": 1.30396481572432e-05, + "loss": 0.0179, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00867424295283854, + "rkd_loss_weight": 0.43549785017967224, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3842959225177765, + "rkd_teacher_entropy": 3.7414050102233887, + "rkd_teacher_similarity_mean": 0.7803046345710755, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16940 + }, + { + "epoch": 4.931626418388129, + "flow_matching_loss": 0.013471247535198927, + "grad_norm": 0.14147208631038666, + "learning_rate": 1.3023256467105237e-05, + "loss": 0.0177, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009789348486810923, + "rkd_loss_weight": 0.43516450822353364, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3860211312770844, + "rkd_teacher_entropy": 3.74424147605896, + "rkd_teacher_similarity_mean": 0.7808491230010987, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16950 + }, + { + "epoch": 4.934535932499273, + "flow_matching_loss": 0.0128095593303442, + "grad_norm": 0.11100373417139053, + "learning_rate": 1.3006867178897067e-05, + "loss": 0.0172, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010183390695601702, + "rkd_loss_weight": 0.4348311573266983, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39818887412548065, + "rkd_teacher_entropy": 3.778689455986023, + "rkd_teacher_similarity_mean": 0.7950714468955994, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16960 + }, + { + "epoch": 4.9374454466104165, + "flow_matching_loss": 0.012725412845611572, + "grad_norm": 0.11567201465368271, + "learning_rate": 1.2990480312533213e-05, + "loss": 0.0168, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00936968312598765, + "rkd_loss_weight": 0.43449781835079193, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38918392062187196, + "rkd_teacher_entropy": 3.755135250091553, + "rkd_teacher_similarity_mean": 0.7840233564376831, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16970 + }, + { + "epoch": 4.940354960721559, + "flow_matching_loss": 0.015397272072732449, + "grad_norm": 0.12759414315223694, + "learning_rate": 1.2974095887925258e-05, + "loss": 0.0193, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00891390866599977, + "rkd_loss_weight": 0.4341644734144211, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3809702455997467, + "rkd_teacher_entropy": 3.73373486995697, + "rkd_teacher_similarity_mean": 0.7768326997756958, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16980 + }, + { + "epoch": 4.943264474832703, + "flow_matching_loss": 0.015868757292628287, + "grad_norm": 0.10656990855932236, + "learning_rate": 1.2957713924981837e-05, + "loss": 0.0196, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00865980377420783, + "rkd_loss_weight": 0.433831125497818, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3860973477363586, + "rkd_teacher_entropy": 3.739105749130249, + "rkd_teacher_similarity_mean": 0.7825042843818665, + "rkd_vlm_temp": 0.10000000149011612, + "step": 16990 + }, + { + "epoch": 4.946173988943847, + "flow_matching_loss": 0.015541715268045663, + "grad_norm": 0.14518649876117706, + "learning_rate": 1.2941334443608572e-05, + "loss": 0.019, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00803334629163146, + "rkd_loss_weight": 0.4334977865219116, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38136833608150483, + "rkd_teacher_entropy": 3.728788375854492, + "rkd_teacher_similarity_mean": 0.7782284796237946, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17000 + }, + { + "epoch": 4.94908350305499, + "flow_matching_loss": 0.015886332001537086, + "grad_norm": 0.10793568193912506, + "learning_rate": 1.2924957463708073e-05, + "loss": 0.0206, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010955089796334505, + "rkd_loss_weight": 0.4331644356250763, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38379005193710325, + "rkd_teacher_entropy": 3.7307998657226564, + "rkd_teacher_similarity_mean": 0.7798720717430114, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17010 + }, + { + "epoch": 4.951993017166133, + "flow_matching_loss": 0.01490375567227602, + "grad_norm": 0.15061984956264496, + "learning_rate": 1.2908583005179924e-05, + "loss": 0.0195, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01054331725463271, + "rkd_loss_weight": 0.4328310936689377, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3778848171234131, + "rkd_teacher_entropy": 3.710316705703735, + "rkd_teacher_similarity_mean": 0.773385763168335, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17020 + }, + { + "epoch": 4.954902531277277, + "flow_matching_loss": 0.01329686543904245, + "grad_norm": 0.11886380612850189, + "learning_rate": 1.2892211087920624e-05, + "loss": 0.0166, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007541483454406262, + "rkd_loss_weight": 0.4324977546930313, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38595998883247373, + "rkd_teacher_entropy": 3.7345031976699827, + "rkd_teacher_similarity_mean": 0.7822970688343048, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17030 + }, + { + "epoch": 4.95781204538842, + "flow_matching_loss": 0.015002910420298577, + "grad_norm": 0.12184422463178635, + "learning_rate": 1.287584173182361e-05, + "loss": 0.0189, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009108011284843088, + "rkd_loss_weight": 0.43216440081596375, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38994148671627044, + "rkd_teacher_entropy": 3.7617838621139525, + "rkd_teacher_similarity_mean": 0.7855618953704834, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17040 + }, + { + "epoch": 4.9607215594995635, + "flow_matching_loss": 0.014107976295053958, + "grad_norm": 0.1888691931962967, + "learning_rate": 1.2859474956779188e-05, + "loss": 0.0181, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00912986770272255, + "rkd_loss_weight": 0.4318310618400574, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3855409562587738, + "rkd_teacher_entropy": 3.7357999324798583, + "rkd_teacher_similarity_mean": 0.7799339830875397, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17050 + }, + { + "epoch": 4.963631073610707, + "flow_matching_loss": 0.014885675814002752, + "grad_norm": 0.08573607355356216, + "learning_rate": 1.2843110782674532e-05, + "loss": 0.0195, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010709481919184327, + "rkd_loss_weight": 0.43149771690368655, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3976457864046097, + "rkd_teacher_entropy": 3.7685885429382324, + "rkd_teacher_similarity_mean": 0.7935177087783813, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17060 + }, + { + "epoch": 4.96654058772185, + "flow_matching_loss": 0.013366680033504963, + "grad_norm": 0.09121432155370712, + "learning_rate": 1.2826749229393655e-05, + "loss": 0.0178, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010316961957141758, + "rkd_loss_weight": 0.43116436898708344, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37851929366588594, + "rkd_teacher_entropy": 3.7329603910446165, + "rkd_teacher_similarity_mean": 0.7743109047412873, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17070 + }, + { + "epoch": 4.969450101832994, + "flow_matching_loss": 0.01578598441556096, + "grad_norm": 0.10671060532331467, + "learning_rate": 1.2810390316817381e-05, + "loss": 0.0198, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009346334217116236, + "rkd_loss_weight": 0.43083103001117706, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39208600521087644, + "rkd_teacher_entropy": 3.755829429626465, + "rkd_teacher_similarity_mean": 0.7879933297634125, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17080 + }, + { + "epoch": 4.9723596159441374, + "flow_matching_loss": 0.015337451361119747, + "grad_norm": 0.15531691908836365, + "learning_rate": 1.2794034064823346e-05, + "loss": 0.019, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008541983459144831, + "rkd_loss_weight": 0.43049768209457395, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3869166404008865, + "rkd_teacher_entropy": 3.755582356452942, + "rkd_teacher_similarity_mean": 0.7847764670848847, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17090 + }, + { + "epoch": 4.975269130055281, + "flow_matching_loss": 0.014943023864179849, + "grad_norm": 0.153790682554245, + "learning_rate": 1.277768049328593e-05, + "loss": 0.0189, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009223786415532231, + "rkd_loss_weight": 0.4301643371582031, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3911420822143555, + "rkd_teacher_entropy": 3.757761812210083, + "rkd_teacher_similarity_mean": 0.7886817216873169, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17100 + }, + { + "epoch": 4.978178644166424, + "flow_matching_loss": 0.016446484718471766, + "grad_norm": 0.12795911729335785, + "learning_rate": 1.2761329622076269e-05, + "loss": 0.0207, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009991586301475763, + "rkd_loss_weight": 0.42983099818229675, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38323180079460145, + "rkd_teacher_entropy": 3.7383511304855346, + "rkd_teacher_similarity_mean": 0.7798739194869995, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17110 + }, + { + "epoch": 4.981088158277568, + "flow_matching_loss": 0.01362457051873207, + "grad_norm": 0.1275118887424469, + "learning_rate": 1.274498147106221e-05, + "loss": 0.0171, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00815085405483842, + "rkd_loss_weight": 0.4294976472854614, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3900855451822281, + "rkd_teacher_entropy": 3.7651114225387574, + "rkd_teacher_similarity_mean": 0.7874786496162415, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17120 + }, + { + "epoch": 4.983997672388711, + "flow_matching_loss": 0.015694119688123464, + "grad_norm": 0.1099030151963234, + "learning_rate": 1.272863606010829e-05, + "loss": 0.0199, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009694493049755693, + "rkd_loss_weight": 0.4291643053293228, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.380391052365303, + "rkd_teacher_entropy": 3.7305815696716307, + "rkd_teacher_similarity_mean": 0.7771995723247528, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17130 + }, + { + "epoch": 4.986907186499854, + "flow_matching_loss": 0.015712808445096017, + "grad_norm": 0.10580971837043762, + "learning_rate": 1.2712293409075739e-05, + "loss": 0.0193, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008276013983413577, + "rkd_loss_weight": 0.42883096635341644, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38886924386024474, + "rkd_teacher_entropy": 3.7539252281188964, + "rkd_teacher_similarity_mean": 0.7862893342971802, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17140 + }, + { + "epoch": 4.989816700610998, + "flow_matching_loss": 0.014024265948683024, + "grad_norm": 0.1130358874797821, + "learning_rate": 1.2695953537822413e-05, + "loss": 0.0178, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008796522952616214, + "rkd_loss_weight": 0.4284976124763489, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38521506190299987, + "rkd_teacher_entropy": 3.746518683433533, + "rkd_teacher_similarity_mean": 0.7833248615264893, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17150 + }, + { + "epoch": 4.992726214722142, + "flow_matching_loss": 0.015107601415365934, + "grad_norm": 0.08749102056026459, + "learning_rate": 1.2679616466202792e-05, + "loss": 0.0183, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007390630710870027, + "rkd_loss_weight": 0.4281642735004425, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39578603506088256, + "rkd_teacher_entropy": 3.779216480255127, + "rkd_teacher_similarity_mean": 0.7934728682041168, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17160 + }, + { + "epoch": 4.995635728833285, + "flow_matching_loss": 0.014174924325197935, + "grad_norm": 0.1513959914445877, + "learning_rate": 1.2663282214067967e-05, + "loss": 0.0178, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008515589032322168, + "rkd_loss_weight": 0.4278309285640717, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.388134640455246, + "rkd_teacher_entropy": 3.7635093688964845, + "rkd_teacher_similarity_mean": 0.7853020787239074, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17170 + }, + { + "epoch": 4.998545242944428, + "flow_matching_loss": 0.01625337889418006, + "grad_norm": 0.13241255283355713, + "learning_rate": 1.264695080126559e-05, + "loss": 0.0199, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008558918116614222, + "rkd_loss_weight": 0.42749758064746857, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3858380883932114, + "rkd_teacher_entropy": 3.7480547428131104, + "rkd_teacher_similarity_mean": 0.783887243270874, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17180 + }, + { + "epoch": 5.001454757055572, + "flow_matching_loss": 0.01379312016069889, + "grad_norm": 0.1279592365026474, + "learning_rate": 1.263062224763986e-05, + "loss": 0.0176, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008858784008771181, + "rkd_loss_weight": 0.4271642416715622, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3933221399784088, + "rkd_teacher_entropy": 3.7762542009353637, + "rkd_teacher_similarity_mean": 0.7931945741176605, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17190 + }, + { + "epoch": 5.004364271166716, + "flow_matching_loss": 0.014495661947876215, + "grad_norm": 0.12226243317127228, + "learning_rate": 1.2614296573031522e-05, + "loss": 0.0181, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008544148039072752, + "rkd_loss_weight": 0.4268308937549591, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37906753122806547, + "rkd_teacher_entropy": 3.7318467140197753, + "rkd_teacher_similarity_mean": 0.7765849888324737, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17200 + }, + { + "epoch": 5.007273785277858, + "flow_matching_loss": 0.014507812727242709, + "grad_norm": 0.12246362864971161, + "learning_rate": 1.259797379727781e-05, + "loss": 0.0184, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009197370568290352, + "rkd_loss_weight": 0.42649754881858826, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3862753391265869, + "rkd_teacher_entropy": 3.751355791091919, + "rkd_teacher_similarity_mean": 0.7832093656063079, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17210 + }, + { + "epoch": 5.010183299389002, + "flow_matching_loss": 0.01685039009898901, + "grad_norm": 0.14690840244293213, + "learning_rate": 1.2581653940212423e-05, + "loss": 0.0204, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008374272566288709, + "rkd_loss_weight": 0.4261642098426819, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38190343976020813, + "rkd_teacher_entropy": 3.7312166690826416, + "rkd_teacher_similarity_mean": 0.7776849806308747, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17220 + }, + { + "epoch": 5.013092813500146, + "flow_matching_loss": 0.01656571524217725, + "grad_norm": 0.11015764623880386, + "learning_rate": 1.2565337021665537e-05, + "loss": 0.0206, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009364139381796121, + "rkd_loss_weight": 0.4258308559656143, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3789647430181503, + "rkd_teacher_entropy": 3.7264919757843016, + "rkd_teacher_similarity_mean": 0.7758941769599914, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17230 + }, + { + "epoch": 5.016002327611289, + "flow_matching_loss": 0.013817184185609222, + "grad_norm": 0.1603635847568512, + "learning_rate": 1.2549023061463739e-05, + "loss": 0.0171, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007723746821284294, + "rkd_loss_weight": 0.42549751698970795, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3764418810606003, + "rkd_teacher_entropy": 3.725432300567627, + "rkd_teacher_similarity_mean": 0.7750512838363648, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17240 + }, + { + "epoch": 5.018911841722432, + "flow_matching_loss": 0.01492272661998868, + "grad_norm": 0.12486115097999573, + "learning_rate": 1.2532712079430038e-05, + "loss": 0.0186, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00853400882333517, + "rkd_loss_weight": 0.42516417503356935, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39445820152759553, + "rkd_teacher_entropy": 3.7707746267318725, + "rkd_teacher_similarity_mean": 0.7922846257686615, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17250 + }, + { + "epoch": 5.021821355833576, + "flow_matching_loss": 0.01579422326758504, + "grad_norm": 0.1459885984659195, + "learning_rate": 1.2516404095383813e-05, + "loss": 0.0192, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008029559627175331, + "rkd_loss_weight": 0.424830824136734, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3828062742948532, + "rkd_teacher_entropy": 3.7402369737625123, + "rkd_teacher_similarity_mean": 0.7800046384334565, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17260 + }, + { + "epoch": 5.024730869944719, + "flow_matching_loss": 0.014039627648890019, + "grad_norm": 0.13262486457824707, + "learning_rate": 1.2500099129140796e-05, + "loss": 0.0174, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007801411719992757, + "rkd_loss_weight": 0.42449748516082764, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3859206110239029, + "rkd_teacher_entropy": 3.755650520324707, + "rkd_teacher_similarity_mean": 0.7832360982894897, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17270 + }, + { + "epoch": 5.0276403840558626, + "flow_matching_loss": 0.014738634787499904, + "grad_norm": 0.09379161149263382, + "learning_rate": 1.2483797200513065e-05, + "loss": 0.0179, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00735880401916802, + "rkd_loss_weight": 0.4241641372442245, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3932476550340652, + "rkd_teacher_entropy": 3.7585498571395872, + "rkd_teacher_similarity_mean": 0.7907406985759735, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17280 + }, + { + "epoch": 5.030549898167006, + "flow_matching_loss": 0.013626905530691147, + "grad_norm": 0.1264609396457672, + "learning_rate": 1.2467498329308992e-05, + "loss": 0.0177, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009529888397082686, + "rkd_loss_weight": 0.4238307923078537, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38672392070293427, + "rkd_teacher_entropy": 3.754014182090759, + "rkd_teacher_similarity_mean": 0.7821463525295258, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17290 + }, + { + "epoch": 5.03345941227815, + "flow_matching_loss": 0.013921229820698499, + "grad_norm": 0.10216625779867172, + "learning_rate": 1.2451202535333253e-05, + "loss": 0.0176, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008796367282047867, + "rkd_loss_weight": 0.4234974533319473, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3890970438718796, + "rkd_teacher_entropy": 3.75610466003418, + "rkd_teacher_similarity_mean": 0.7869192123413086, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17300 + }, + { + "epoch": 5.036368926389293, + "flow_matching_loss": 0.015115504339337348, + "grad_norm": 0.14165419340133667, + "learning_rate": 1.2434909838386768e-05, + "loss": 0.0193, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009794124122709035, + "rkd_loss_weight": 0.423164102435112, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38488803803920746, + "rkd_teacher_entropy": 3.743880105018616, + "rkd_teacher_similarity_mean": 0.7826468825340271, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17310 + }, + { + "epoch": 5.0392784405004365, + "flow_matching_loss": 0.01583687076345086, + "grad_norm": 0.11704482138156891, + "learning_rate": 1.2418620258266703e-05, + "loss": 0.0199, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009601313527673483, + "rkd_loss_weight": 0.4228307604789734, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38937437534332275, + "rkd_teacher_entropy": 3.7572198152542113, + "rkd_teacher_similarity_mean": 0.7876894891262054, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17320 + }, + { + "epoch": 5.04218795461158, + "flow_matching_loss": 0.016147022228688003, + "grad_norm": 0.13013793528079987, + "learning_rate": 1.2402333814766432e-05, + "loss": 0.02, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009175658784806728, + "rkd_loss_weight": 0.422497421503067, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38594362437725066, + "rkd_teacher_entropy": 3.759230923652649, + "rkd_teacher_similarity_mean": 0.7852511763572693, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17330 + }, + { + "epoch": 5.045097468722723, + "flow_matching_loss": 0.01195036363787949, + "grad_norm": 0.10328301042318344, + "learning_rate": 1.2386050527675515e-05, + "loss": 0.0156, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008644063444808125, + "rkd_loss_weight": 0.42216406762599945, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38444823026657104, + "rkd_teacher_entropy": 3.7448694944381713, + "rkd_teacher_similarity_mean": 0.780383950471878, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17340 + }, + { + "epoch": 5.048006982833867, + "flow_matching_loss": 0.01443180302157998, + "grad_norm": 0.1296122819185257, + "learning_rate": 1.2369770416779693e-05, + "loss": 0.0182, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008858826756477357, + "rkd_loss_weight": 0.4218307286500931, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38360033333301546, + "rkd_teacher_entropy": 3.736821722984314, + "rkd_teacher_similarity_mean": 0.7811167180538178, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17350 + }, + { + "epoch": 5.05091649694501, + "flow_matching_loss": 0.01400065291672945, + "grad_norm": 0.14199911057949066, + "learning_rate": 1.2353493501860828e-05, + "loss": 0.0174, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008168195839971304, + "rkd_loss_weight": 0.4214973866939545, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39088245332241056, + "rkd_teacher_entropy": 3.7524676084518434, + "rkd_teacher_similarity_mean": 0.7882379353046417, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17360 + }, + { + "epoch": 5.053826011056153, + "flow_matching_loss": 0.015021313354372979, + "grad_norm": 0.11321830004453659, + "learning_rate": 1.2337219802696905e-05, + "loss": 0.0182, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007441269280388951, + "rkd_loss_weight": 0.42116403579711914, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3859957307577133, + "rkd_teacher_entropy": 3.7487094402313232, + "rkd_teacher_similarity_mean": 0.7823050498962403, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17370 + }, + { + "epoch": 5.056735525167297, + "flow_matching_loss": 0.014059239346534014, + "grad_norm": 0.2811999022960663, + "learning_rate": 1.2320949339062008e-05, + "loss": 0.0186, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01080891564488411, + "rkd_loss_weight": 0.42083069682121277, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3881700962781906, + "rkd_teacher_entropy": 3.740795850753784, + "rkd_teacher_similarity_mean": 0.7852764844894409, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17380 + }, + { + "epoch": 5.059645039278441, + "flow_matching_loss": 0.014638255257159472, + "grad_norm": 0.1269218474626541, + "learning_rate": 1.2304682130726281e-05, + "loss": 0.0179, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007721100654453039, + "rkd_loss_weight": 0.42049734890460966, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3794092535972595, + "rkd_teacher_entropy": 3.7266505479812624, + "rkd_teacher_similarity_mean": 0.7767669856548309, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17390 + }, + { + "epoch": 5.0625545533895835, + "flow_matching_loss": 0.016691856365650893, + "grad_norm": 0.13155913352966309, + "learning_rate": 1.2288418197455913e-05, + "loss": 0.0201, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008118091197684408, + "rkd_loss_weight": 0.42016400396823883, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3861532360315323, + "rkd_teacher_entropy": 3.7421367883682253, + "rkd_teacher_similarity_mean": 0.7817550122737884, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17400 + }, + { + "epoch": 5.065464067500727, + "flow_matching_loss": 0.013849705923348665, + "grad_norm": 0.12047506123781204, + "learning_rate": 1.2272157559013126e-05, + "loss": 0.0174, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008571353973820806, + "rkd_loss_weight": 0.41983066499233246, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3829998433589935, + "rkd_teacher_entropy": 3.7400038719177244, + "rkd_teacher_similarity_mean": 0.779197096824646, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17410 + }, + { + "epoch": 5.068373581611871, + "flow_matching_loss": 0.013246566988527775, + "grad_norm": 0.13492846488952637, + "learning_rate": 1.2255900235156123e-05, + "loss": 0.0181, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011578266136348247, + "rkd_loss_weight": 0.4194973140954971, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37338861227035525, + "rkd_teacher_entropy": 3.6999720335006714, + "rkd_teacher_similarity_mean": 0.769416481256485, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17420 + }, + { + "epoch": 5.071283095723015, + "flow_matching_loss": 0.014013312105089427, + "grad_norm": 0.12786994874477386, + "learning_rate": 1.2239646245639087e-05, + "loss": 0.0175, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008334123622626066, + "rkd_loss_weight": 0.4191639721393585, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38062675297260284, + "rkd_teacher_entropy": 3.724316930770874, + "rkd_teacher_similarity_mean": 0.7764016807079315, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17430 + }, + { + "epoch": 5.074192609834157, + "flow_matching_loss": 0.012173824105411768, + "grad_norm": 0.13217394053936005, + "learning_rate": 1.2223395610212142e-05, + "loss": 0.0157, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008412343496456743, + "rkd_loss_weight": 0.4188306301832199, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3800476551055908, + "rkd_teacher_entropy": 3.7333793878555297, + "rkd_teacher_similarity_mean": 0.7765242040157319, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17440 + }, + { + "epoch": 5.077102123945301, + "flow_matching_loss": 0.012199910450726747, + "grad_norm": 0.11835221201181412, + "learning_rate": 1.2207148348621348e-05, + "loss": 0.0161, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009247649274766446, + "rkd_loss_weight": 0.4184972792863846, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3894006431102753, + "rkd_teacher_entropy": 3.748264217376709, + "rkd_teacher_similarity_mean": 0.7876887619495392, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17450 + }, + { + "epoch": 5.080011638056445, + "flow_matching_loss": 0.01589820422232151, + "grad_norm": 0.16846513748168945, + "learning_rate": 1.219090448060866e-05, + "loss": 0.0196, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008875024504959583, + "rkd_loss_weight": 0.4181639403104782, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3899712026119232, + "rkd_teacher_entropy": 3.7674320697784425, + "rkd_teacher_similarity_mean": 0.7880248844623565, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17460 + }, + { + "epoch": 5.082921152167588, + "flow_matching_loss": 0.01588259506970644, + "grad_norm": 0.1761631965637207, + "learning_rate": 1.217466402591191e-05, + "loss": 0.0203, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010538640664890409, + "rkd_loss_weight": 0.4178305953741074, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38330833315849305, + "rkd_teacher_entropy": 3.740097999572754, + "rkd_teacher_similarity_mean": 0.7799115896224975, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17470 + }, + { + "epoch": 5.085830666278731, + "flow_matching_loss": 0.014694028533995152, + "grad_norm": 0.1162189394235611, + "learning_rate": 1.2158427004264785e-05, + "loss": 0.0179, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007615864928811789, + "rkd_loss_weight": 0.4174972474575043, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38550362884998324, + "rkd_teacher_entropy": 3.7471046924591063, + "rkd_teacher_similarity_mean": 0.7819080352783203, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17480 + }, + { + "epoch": 5.088740180389875, + "flow_matching_loss": 0.01625860622152686, + "grad_norm": 0.12437520921230316, + "learning_rate": 1.2142193435396788e-05, + "loss": 0.0198, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008482585102319718, + "rkd_loss_weight": 0.4171639084815979, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37875014543533325, + "rkd_teacher_entropy": 3.730122375488281, + "rkd_teacher_similarity_mean": 0.7755092442035675, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17490 + }, + { + "epoch": 5.091649694501018, + "flow_matching_loss": 0.013029033225029707, + "grad_norm": 0.12421374768018723, + "learning_rate": 1.2125963339033234e-05, + "loss": 0.0164, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008039010455831885, + "rkd_loss_weight": 0.41683055758476256, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3858205765485764, + "rkd_teacher_entropy": 3.73201060295105, + "rkd_teacher_similarity_mean": 0.7824061989784241, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17500 + }, + { + "epoch": 5.094559208612162, + "flow_matching_loss": 0.012789891753345727, + "grad_norm": 0.13156233727931976, + "learning_rate": 1.2109736734895237e-05, + "loss": 0.0158, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007311440305784345, + "rkd_loss_weight": 0.41649721562862396, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3904715210199356, + "rkd_teacher_entropy": 3.7521790266036987, + "rkd_teacher_similarity_mean": 0.785764068365097, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17510 + }, + { + "epoch": 5.097468722723305, + "flow_matching_loss": 0.01623479016125202, + "grad_norm": 0.12560881674289703, + "learning_rate": 1.2093513642699634e-05, + "loss": 0.0205, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010334222204983234, + "rkd_loss_weight": 0.4161638766527176, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3854055941104889, + "rkd_teacher_entropy": 3.745263195037842, + "rkd_teacher_similarity_mean": 0.7820225477218627, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17520 + }, + { + "epoch": 5.100378236834449, + "flow_matching_loss": 0.016766019444912673, + "grad_norm": 0.14023569226264954, + "learning_rate": 1.2077294082159024e-05, + "loss": 0.0205, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008867472922429443, + "rkd_loss_weight": 0.41583052277565, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38718315958976746, + "rkd_teacher_entropy": 3.7516492366790772, + "rkd_teacher_similarity_mean": 0.7843135058879852, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17530 + }, + { + "epoch": 5.103287750945592, + "flow_matching_loss": 0.01618796344846487, + "grad_norm": 0.15353047847747803, + "learning_rate": 1.2061078072981697e-05, + "loss": 0.02, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009156864508986473, + "rkd_loss_weight": 0.41549718379974365, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38151835799217226, + "rkd_teacher_entropy": 3.734715151786804, + "rkd_teacher_similarity_mean": 0.7780771970748901, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17540 + }, + { + "epoch": 5.1061972650567355, + "flow_matching_loss": 0.01446445593610406, + "grad_norm": 0.129802405834198, + "learning_rate": 1.2044865634871626e-05, + "loss": 0.0186, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00994926723651588, + "rkd_loss_weight": 0.41516384184360505, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3933195024728775, + "rkd_teacher_entropy": 3.761670970916748, + "rkd_teacher_similarity_mean": 0.789610105752945, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17550 + }, + { + "epoch": 5.109106779167879, + "flow_matching_loss": 0.014778683148324489, + "grad_norm": 0.11220529675483704, + "learning_rate": 1.2028656787528468e-05, + "loss": 0.0185, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0090569369494915, + "rkd_loss_weight": 0.4148304909467697, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3822144061326981, + "rkd_teacher_entropy": 3.731535863876343, + "rkd_teacher_similarity_mean": 0.7792213141918183, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17560 + }, + { + "epoch": 5.112016293279022, + "flow_matching_loss": 0.015182885527610778, + "grad_norm": 0.09833871573209763, + "learning_rate": 1.2012451550647491e-05, + "loss": 0.0185, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007889809506013989, + "rkd_loss_weight": 0.41449715197086334, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38612076044082644, + "rkd_teacher_entropy": 3.7383012533187867, + "rkd_teacher_similarity_mean": 0.783705198764801, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17570 + }, + { + "epoch": 5.114925807390166, + "flow_matching_loss": 0.015676661301404236, + "grad_norm": 0.10872539132833481, + "learning_rate": 1.1996249943919589e-05, + "loss": 0.0198, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009835963137447834, + "rkd_loss_weight": 0.4141638070344925, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3858494758605957, + "rkd_teacher_entropy": 3.7524195194244383, + "rkd_teacher_similarity_mean": 0.7841075479984283, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17580 + }, + { + "epoch": 5.1178353215013095, + "flow_matching_loss": 0.016243503242731095, + "grad_norm": 0.14591649174690247, + "learning_rate": 1.1980051987031237e-05, + "loss": 0.0201, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009240622678771616, + "rkd_loss_weight": 0.4138304591178894, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.391249942779541, + "rkd_teacher_entropy": 3.7750272274017336, + "rkd_teacher_similarity_mean": 0.7902799248695374, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17590 + }, + { + "epoch": 5.120744835612452, + "flow_matching_loss": 0.015248441603034735, + "grad_norm": 0.12762320041656494, + "learning_rate": 1.1963857699664483e-05, + "loss": 0.019, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009015782503411174, + "rkd_loss_weight": 0.41349712014198303, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3841536968946457, + "rkd_teacher_entropy": 3.736326289176941, + "rkd_teacher_similarity_mean": 0.7804123640060425, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17600 + }, + { + "epoch": 5.123654349723596, + "flow_matching_loss": 0.014017033483833074, + "grad_norm": 0.12815892696380615, + "learning_rate": 1.1947667101496911e-05, + "loss": 0.0177, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008981423918157816, + "rkd_loss_weight": 0.4131637692451477, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3910353511571884, + "rkd_teacher_entropy": 3.7590286254882814, + "rkd_teacher_similarity_mean": 0.7885815739631653, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17610 + }, + { + "epoch": 5.12656386383474, + "flow_matching_loss": 0.013495597802102566, + "grad_norm": 0.12525685131549835, + "learning_rate": 1.1931480212201633e-05, + "loss": 0.0176, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009896274842321872, + "rkd_loss_weight": 0.4128304272890091, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3818435430526733, + "rkd_teacher_entropy": 3.738660144805908, + "rkd_teacher_similarity_mean": 0.7793823421001435, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17620 + }, + { + "epoch": 5.129473377945883, + "flow_matching_loss": 0.017727692052721976, + "grad_norm": 0.12222641706466675, + "learning_rate": 1.1915297051447237e-05, + "loss": 0.0218, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009975833399221302, + "rkd_loss_weight": 0.4124970883131027, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38586616814136504, + "rkd_teacher_entropy": 3.7466893434524535, + "rkd_teacher_similarity_mean": 0.7845059156417846, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17630 + }, + { + "epoch": 5.132382892057026, + "flow_matching_loss": 0.015807269141077995, + "grad_norm": 0.18082274496555328, + "learning_rate": 1.1899117638897793e-05, + "loss": 0.0196, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00914086913689971, + "rkd_loss_weight": 0.41216373443603516, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38642749190330505, + "rkd_teacher_entropy": 3.7535400867462156, + "rkd_teacher_similarity_mean": 0.7846439778804779, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17640 + }, + { + "epoch": 5.13529240616817, + "flow_matching_loss": 0.015082413796335459, + "grad_norm": 0.1129683405160904, + "learning_rate": 1.1882941994212811e-05, + "loss": 0.0183, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00787343797273934, + "rkd_loss_weight": 0.4118303954601288, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38412727415561676, + "rkd_teacher_entropy": 3.7441303968429565, + "rkd_teacher_similarity_mean": 0.7814971089363099, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17650 + }, + { + "epoch": 5.138201920279314, + "flow_matching_loss": 0.016144743794575332, + "grad_norm": 0.12544189393520355, + "learning_rate": 1.1866770137047221e-05, + "loss": 0.0206, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010847700713202357, + "rkd_loss_weight": 0.41149705052375796, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3912490963935852, + "rkd_teacher_entropy": 3.762008261680603, + "rkd_teacher_similarity_mean": 0.7869123935699462, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17660 + }, + { + "epoch": 5.1411114343904565, + "flow_matching_loss": 0.012646517995744944, + "grad_norm": 0.14063799381256104, + "learning_rate": 1.1850602087051363e-05, + "loss": 0.0163, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008849041815847159, + "rkd_loss_weight": 0.41116370260715485, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3881258308887482, + "rkd_teacher_entropy": 3.761419749259949, + "rkd_teacher_similarity_mean": 0.7853536128997802, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17670 + }, + { + "epoch": 5.1440209485016, + "flow_matching_loss": 0.014281210023909807, + "grad_norm": 0.11463971436023712, + "learning_rate": 1.183443786387094e-05, + "loss": 0.0178, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008478621253743768, + "rkd_loss_weight": 0.4108303636312485, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3899088501930237, + "rkd_teacher_entropy": 3.7588746309280396, + "rkd_teacher_similarity_mean": 0.7873956799507141, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17680 + }, + { + "epoch": 5.146930462612744, + "flow_matching_loss": 0.01612028554081917, + "grad_norm": 0.11692272126674652, + "learning_rate": 1.1818277487147005e-05, + "loss": 0.0195, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008115250989794731, + "rkd_loss_weight": 0.41049701571464536, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3894421488046646, + "rkd_teacher_entropy": 3.769529342651367, + "rkd_teacher_similarity_mean": 0.7870561122894287, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17690 + }, + { + "epoch": 5.149839976723887, + "flow_matching_loss": 0.013796504866331816, + "grad_norm": 0.13093732297420502, + "learning_rate": 1.180212097651594e-05, + "loss": 0.0174, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008814111724495888, + "rkd_loss_weight": 0.41016367077827454, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38069077134132384, + "rkd_teacher_entropy": 3.732079339027405, + "rkd_teacher_similarity_mean": 0.7771007061004639, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17700 + }, + { + "epoch": 5.15274949083503, + "flow_matching_loss": 0.012824838608503341, + "grad_norm": 0.08920325338840485, + "learning_rate": 1.1785968351609422e-05, + "loss": 0.0172, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010645185643807054, + "rkd_loss_weight": 0.40983033180236816, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38364529609680176, + "rkd_teacher_entropy": 3.7381603479385377, + "rkd_teacher_similarity_mean": 0.7797192990779876, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17710 + }, + { + "epoch": 5.155659004946174, + "flow_matching_loss": 0.014091711584478616, + "grad_norm": 0.13106080889701843, + "learning_rate": 1.1769819632054427e-05, + "loss": 0.0176, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008514885604381562, + "rkd_loss_weight": 0.4094969779253006, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.378363361954689, + "rkd_teacher_entropy": 3.7238003492355345, + "rkd_teacher_similarity_mean": 0.7767682373523712, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17720 + }, + { + "epoch": 5.158568519057318, + "flow_matching_loss": 0.014076944440603256, + "grad_norm": 0.10286714881658554, + "learning_rate": 1.1753674837473165e-05, + "loss": 0.0177, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00882105971686542, + "rkd_loss_weight": 0.4091636389493942, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3843758523464203, + "rkd_teacher_entropy": 3.7444303750991823, + "rkd_teacher_similarity_mean": 0.7812453806400299, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17730 + }, + { + "epoch": 5.161478033168461, + "flow_matching_loss": 0.012476716842502355, + "grad_norm": 0.12353744357824326, + "learning_rate": 1.1737533987483081e-05, + "loss": 0.0158, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008203988149762153, + "rkd_loss_weight": 0.4088302969932556, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3912886410951614, + "rkd_teacher_entropy": 3.766464614868164, + "rkd_teacher_similarity_mean": 0.7885408878326416, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17740 + }, + { + "epoch": 5.164387547279604, + "flow_matching_loss": 0.01445584073662758, + "grad_norm": 0.12799310684204102, + "learning_rate": 1.1721397101696835e-05, + "loss": 0.0182, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009096112102270126, + "rkd_loss_weight": 0.4084969460964203, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37772339284420015, + "rkd_teacher_entropy": 3.7190322399139406, + "rkd_teacher_similarity_mean": 0.7752059936523438, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17750 + }, + { + "epoch": 5.167297061390748, + "flow_matching_loss": 0.012186534702777863, + "grad_norm": 0.1237703263759613, + "learning_rate": 1.1705264199722257e-05, + "loss": 0.0158, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008794427243992686, + "rkd_loss_weight": 0.4081636071205139, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38410237729549407, + "rkd_teacher_entropy": 3.727741813659668, + "rkd_teacher_similarity_mean": 0.7801803946495056, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17760 + }, + { + "epoch": 5.170206575501891, + "flow_matching_loss": 0.01424002218991518, + "grad_norm": 0.19041547179222107, + "learning_rate": 1.1689135301162356e-05, + "loss": 0.0182, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00968166794627905, + "rkd_loss_weight": 0.4078302621841431, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38210111260414126, + "rkd_teacher_entropy": 3.7377527952194214, + "rkd_teacher_similarity_mean": 0.7801364421844482, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17770 + }, + { + "epoch": 5.173116089613035, + "flow_matching_loss": 0.014420908130705356, + "grad_norm": 0.13360317051410675, + "learning_rate": 1.1673010425615258e-05, + "loss": 0.0177, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008113703690469265, + "rkd_loss_weight": 0.40749691426754, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3940142184495926, + "rkd_teacher_entropy": 3.781206798553467, + "rkd_teacher_similarity_mean": 0.7911659598350524, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17780 + }, + { + "epoch": 5.176025603724178, + "flow_matching_loss": 0.01444921623915434, + "grad_norm": 0.10990741103887558, + "learning_rate": 1.1656889592674208e-05, + "loss": 0.0181, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008993617119267583, + "rkd_loss_weight": 0.4071635752916336, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3929114431142807, + "rkd_teacher_entropy": 3.7716962575912474, + "rkd_teacher_similarity_mean": 0.7921916902065277, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17790 + }, + { + "epoch": 5.178935117835321, + "flow_matching_loss": 0.013647576421499252, + "grad_norm": 0.11953029781579971, + "learning_rate": 1.1640772821927533e-05, + "loss": 0.0172, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008683840045705438, + "rkd_loss_weight": 0.4068302273750305, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3903841197490692, + "rkd_teacher_entropy": 3.772855281829834, + "rkd_teacher_similarity_mean": 0.788249671459198, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17800 + }, + { + "epoch": 5.181844631946465, + "flow_matching_loss": 0.014635348413139581, + "grad_norm": 0.10358544439077377, + "learning_rate": 1.162466013295864e-05, + "loss": 0.0177, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00741795445792377, + "rkd_loss_weight": 0.40649688243865967, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38616629540920255, + "rkd_teacher_entropy": 3.740492248535156, + "rkd_teacher_similarity_mean": 0.7840955436229706, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17810 + }, + { + "epoch": 5.1847541460576085, + "flow_matching_loss": 0.016615706030279398, + "grad_norm": 0.12329193204641342, + "learning_rate": 1.160855154534595e-05, + "loss": 0.0203, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009025845862925052, + "rkd_loss_weight": 0.4061635434627533, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38851761221885683, + "rkd_teacher_entropy": 3.75823175907135, + "rkd_teacher_similarity_mean": 0.7845571994781494, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17820 + }, + { + "epoch": 5.187663660168752, + "flow_matching_loss": 0.014202978555113077, + "grad_norm": 0.14760424196720123, + "learning_rate": 1.1592447078662931e-05, + "loss": 0.0173, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00753236492164433, + "rkd_loss_weight": 0.40583018958568573, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3834084630012512, + "rkd_teacher_entropy": 3.746839714050293, + "rkd_teacher_similarity_mean": 0.7804798364639283, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17830 + }, + { + "epoch": 5.190573174279895, + "flow_matching_loss": 0.01532916584983468, + "grad_norm": 0.12894168496131897, + "learning_rate": 1.1576346752478025e-05, + "loss": 0.0195, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010201913630589842, + "rkd_loss_weight": 0.40549685060977936, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37472793459892273, + "rkd_teacher_entropy": 3.718621850013733, + "rkd_teacher_similarity_mean": 0.7709916412830353, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17840 + }, + { + "epoch": 5.193482688391039, + "flow_matching_loss": 0.015099346544593573, + "grad_norm": 0.14057199656963348, + "learning_rate": 1.1560250586354643e-05, + "loss": 0.0192, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010186909791082143, + "rkd_loss_weight": 0.40516350865364076, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3896093964576721, + "rkd_teacher_entropy": 3.7541695117950438, + "rkd_teacher_similarity_mean": 0.7862587213516236, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17850 + }, + { + "epoch": 5.1963922025021825, + "flow_matching_loss": 0.013181378040462733, + "grad_norm": 0.09243664145469666, + "learning_rate": 1.1544158599851146e-05, + "loss": 0.0165, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008233369886875152, + "rkd_loss_weight": 0.4048301577568054, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38478220999240875, + "rkd_teacher_entropy": 3.734083342552185, + "rkd_teacher_similarity_mean": 0.7801999926567078, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17860 + }, + { + "epoch": 5.199301716613325, + "flow_matching_loss": 0.014853213261812926, + "grad_norm": 0.1260611116886139, + "learning_rate": 1.1528070812520809e-05, + "loss": 0.0186, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009334013611078263, + "rkd_loss_weight": 0.40449681878089905, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38506863713264466, + "rkd_teacher_entropy": 3.744528388977051, + "rkd_teacher_similarity_mean": 0.7817022800445557, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17870 + }, + { + "epoch": 5.202211230724469, + "flow_matching_loss": 0.013723299745470286, + "grad_norm": 0.13019110262393951, + "learning_rate": 1.1511987243911821e-05, + "loss": 0.0173, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008750924654304982, + "rkd_loss_weight": 0.40416347086429594, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3860905349254608, + "rkd_teacher_entropy": 3.73631637096405, + "rkd_teacher_similarity_mean": 0.7821680724620819, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17880 + }, + { + "epoch": 5.205120744835613, + "flow_matching_loss": 0.013936192449182271, + "grad_norm": 0.11064374446868896, + "learning_rate": 1.1495907913567228e-05, + "loss": 0.0173, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008292898302897811, + "rkd_loss_weight": 0.4038301259279251, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38871370553970336, + "rkd_teacher_entropy": 3.755351209640503, + "rkd_teacher_similarity_mean": 0.7849652230739593, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17890 + }, + { + "epoch": 5.2080302589467555, + "flow_matching_loss": 0.012593671306967735, + "grad_norm": 0.10447721928358078, + "learning_rate": 1.1479832841024936e-05, + "loss": 0.0158, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008063534554094077, + "rkd_loss_weight": 0.40349678695201874, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3809956908226013, + "rkd_teacher_entropy": 3.732384753227234, + "rkd_teacher_similarity_mean": 0.7773865282535553, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17900 + }, + { + "epoch": 5.210939773057899, + "flow_matching_loss": 0.012942450679838657, + "grad_norm": 0.13617515563964844, + "learning_rate": 1.1463762045817675e-05, + "loss": 0.0169, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009713672008365393, + "rkd_loss_weight": 0.4031634360551834, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3865212440490723, + "rkd_teacher_entropy": 3.736460280418396, + "rkd_teacher_similarity_mean": 0.7822563111782074, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17910 + }, + { + "epoch": 5.213849287169043, + "flow_matching_loss": 0.015596907772123813, + "grad_norm": 0.13335861265659332, + "learning_rate": 1.1447695547472964e-05, + "loss": 0.0194, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009348096698522568, + "rkd_loss_weight": 0.4028300940990448, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3864863902330399, + "rkd_teacher_entropy": 3.73529531955719, + "rkd_teacher_similarity_mean": 0.7821124851703644, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17920 + }, + { + "epoch": 5.216758801280186, + "flow_matching_loss": 0.013281596777960657, + "grad_norm": 0.0992865189909935, + "learning_rate": 1.1431633365513132e-05, + "loss": 0.0167, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008614868391305208, + "rkd_loss_weight": 0.4024967521429062, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3917661041021347, + "rkd_teacher_entropy": 3.76445951461792, + "rkd_teacher_similarity_mean": 0.7882758677005768, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17930 + }, + { + "epoch": 5.2196683153913295, + "flow_matching_loss": 0.014543695095926523, + "grad_norm": 0.11190516501665115, + "learning_rate": 1.1415575519455235e-05, + "loss": 0.0184, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009623039094731212, + "rkd_loss_weight": 0.40216340124607086, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.381576007604599, + "rkd_teacher_entropy": 3.7267783164978026, + "rkd_teacher_similarity_mean": 0.7762974381446839, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17940 + }, + { + "epoch": 5.222577829502473, + "flow_matching_loss": 0.01389854117296636, + "grad_norm": 0.1095363050699234, + "learning_rate": 1.1399522028811069e-05, + "loss": 0.0175, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008967379666864872, + "rkd_loss_weight": 0.4018300622701645, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39050759971141813, + "rkd_teacher_entropy": 3.751810097694397, + "rkd_teacher_similarity_mean": 0.7852224946022034, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17950 + }, + { + "epoch": 5.225487343613617, + "flow_matching_loss": 0.014300686586648226, + "grad_norm": 0.1392774134874344, + "learning_rate": 1.1383472913087143e-05, + "loss": 0.018, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009092561900615692, + "rkd_loss_weight": 0.40149671733379366, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38650211989879607, + "rkd_teacher_entropy": 3.7403067111968995, + "rkd_teacher_similarity_mean": 0.7837467908859252, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17960 + }, + { + "epoch": 5.22839685772476, + "flow_matching_loss": 0.01633510389365256, + "grad_norm": 0.15184669196605682, + "learning_rate": 1.1367428191784643e-05, + "loss": 0.0205, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01039212835021317, + "rkd_loss_weight": 0.40116336941719055, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38253907561302186, + "rkd_teacher_entropy": 3.7405802488327025, + "rkd_teacher_similarity_mean": 0.7800827980041504, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17970 + }, + { + "epoch": 5.231306371835903, + "flow_matching_loss": 0.012101490143686533, + "grad_norm": 0.10520178079605103, + "learning_rate": 1.1351387884399413e-05, + "loss": 0.0151, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007601034967228771, + "rkd_loss_weight": 0.4008300304412842, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3803724110126495, + "rkd_teacher_entropy": 3.7395374536514283, + "rkd_teacher_similarity_mean": 0.7778365135192871, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17980 + }, + { + "epoch": 5.234215885947047, + "flow_matching_loss": 0.012828211626037955, + "grad_norm": 0.13591499626636505, + "learning_rate": 1.1335352010421953e-05, + "loss": 0.0162, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008398582274094225, + "rkd_loss_weight": 0.40049668252468107, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3828947633504868, + "rkd_teacher_entropy": 3.727626371383667, + "rkd_teacher_similarity_mean": 0.7808921694755554, + "rkd_vlm_temp": 0.10000000149011612, + "step": 17990 + }, + { + "epoch": 5.23712540005819, + "flow_matching_loss": 0.013621988426893949, + "grad_norm": 0.13921679556369781, + "learning_rate": 1.1319320589337352e-05, + "loss": 0.0172, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009019609028473496, + "rkd_loss_weight": 0.40016333758831024, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38441730439662936, + "rkd_teacher_entropy": 3.755030655860901, + "rkd_teacher_similarity_mean": 0.7805925846099854, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18000 + }, + { + "epoch": 5.240034914169334, + "flow_matching_loss": 0.012635687086731195, + "grad_norm": 0.12224158644676208, + "learning_rate": 1.1303293640625299e-05, + "loss": 0.0157, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007771981600672007, + "rkd_loss_weight": 0.39982999861240387, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3854727536439896, + "rkd_teacher_entropy": 3.7410734415054323, + "rkd_teacher_similarity_mean": 0.780816376209259, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18010 + }, + { + "epoch": 5.242944428280477, + "flow_matching_loss": 0.014497330598533153, + "grad_norm": 0.1239977777004242, + "learning_rate": 1.1287271183760052e-05, + "loss": 0.0174, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007259277114644647, + "rkd_loss_weight": 0.39949664771556853, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3816336661577225, + "rkd_teacher_entropy": 3.7378718852996826, + "rkd_teacher_similarity_mean": 0.7780173301696778, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18020 + }, + { + "epoch": 5.24585394239162, + "flow_matching_loss": 0.014330321922898293, + "grad_norm": 0.11668892204761505, + "learning_rate": 1.1271253238210398e-05, + "loss": 0.0174, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007713177939876914, + "rkd_loss_weight": 0.39916330575942993, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3888562172651291, + "rkd_teacher_entropy": 3.7543063402175902, + "rkd_teacher_similarity_mean": 0.7854340255260468, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18030 + }, + { + "epoch": 5.248763456502764, + "flow_matching_loss": 0.013845199812203645, + "grad_norm": 0.1150316447019577, + "learning_rate": 1.1255239823439663e-05, + "loss": 0.0168, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00747577422298491, + "rkd_loss_weight": 0.39882996380329133, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3797029972076416, + "rkd_teacher_entropy": 3.7155457735061646, + "rkd_teacher_similarity_mean": 0.7754449367523193, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18040 + }, + { + "epoch": 5.251672970613908, + "flow_matching_loss": 0.013672403525561094, + "grad_norm": 0.11740221083164215, + "learning_rate": 1.1239230958905652e-05, + "loss": 0.0165, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007060019392520189, + "rkd_loss_weight": 0.398496612906456, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3751417249441147, + "rkd_teacher_entropy": 3.7010055065155028, + "rkd_teacher_similarity_mean": 0.7712095081806183, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18050 + }, + { + "epoch": 5.254582484725051, + "flow_matching_loss": 0.013085001148283482, + "grad_norm": 0.11255808919668198, + "learning_rate": 1.1223226664060641e-05, + "loss": 0.0164, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008220805181190371, + "rkd_loss_weight": 0.3981632739305496, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3817143142223358, + "rkd_teacher_entropy": 3.7356921195983888, + "rkd_teacher_similarity_mean": 0.7775457084178925, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18060 + }, + { + "epoch": 5.257491998836194, + "flow_matching_loss": 0.013648435566574335, + "grad_norm": 0.11292597651481628, + "learning_rate": 1.1207226958351362e-05, + "loss": 0.0169, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008212279248982667, + "rkd_loss_weight": 0.3978299289941788, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38479911386966703, + "rkd_teacher_entropy": 3.7339542388916014, + "rkd_teacher_similarity_mean": 0.7812948286533355, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18070 + }, + { + "epoch": 5.260401512947338, + "flow_matching_loss": 0.014909514039754868, + "grad_norm": 0.1604927033185959, + "learning_rate": 1.1191231861218955e-05, + "loss": 0.0189, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010073728393763304, + "rkd_loss_weight": 0.3974965810775757, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38403829634189607, + "rkd_teacher_entropy": 3.743010663986206, + "rkd_teacher_similarity_mean": 0.7795518040657043, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18080 + }, + { + "epoch": 5.2633110270584815, + "flow_matching_loss": 0.015306134894490241, + "grad_norm": 0.1287616342306137, + "learning_rate": 1.1175241392098988e-05, + "loss": 0.0184, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007862168597057462, + "rkd_loss_weight": 0.3971632421016693, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38575621545314787, + "rkd_teacher_entropy": 3.742937421798706, + "rkd_teacher_similarity_mean": 0.7814976394176483, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18090 + }, + { + "epoch": 5.266220541169624, + "flow_matching_loss": 0.013855281006544828, + "grad_norm": 0.0998476892709732, + "learning_rate": 1.1159255570421376e-05, + "loss": 0.018, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01035881363786757, + "rkd_loss_weight": 0.396829891204834, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3910078853368759, + "rkd_teacher_entropy": 3.7572781801223756, + "rkd_teacher_similarity_mean": 0.7872822165489197, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18100 + }, + { + "epoch": 5.269130055280768, + "flow_matching_loss": 0.014042248949408531, + "grad_norm": 0.12007700651884079, + "learning_rate": 1.1143274415610408e-05, + "loss": 0.0175, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00883440924808383, + "rkd_loss_weight": 0.3964965492486954, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3872728407382965, + "rkd_teacher_entropy": 3.7407527446746824, + "rkd_teacher_similarity_mean": 0.7824478387832642, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18110 + }, + { + "epoch": 5.272039569391912, + "flow_matching_loss": 0.01329301348887384, + "grad_norm": 0.13405142724514008, + "learning_rate": 1.1127297947084692e-05, + "loss": 0.0163, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0074992711190134285, + "rkd_loss_weight": 0.396163210272789, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39025316536426546, + "rkd_teacher_entropy": 3.74773964881897, + "rkd_teacher_similarity_mean": 0.7860959589481353, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18120 + }, + { + "epoch": 5.274949083503055, + "flow_matching_loss": 0.013556070998311043, + "grad_norm": 0.1099054217338562, + "learning_rate": 1.1111326184257135e-05, + "loss": 0.0174, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0095908026676625, + "rkd_loss_weight": 0.39582985639572144, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38519872426986695, + "rkd_teacher_entropy": 3.7397322177886965, + "rkd_teacher_similarity_mean": 0.7807887017726898, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18130 + }, + { + "epoch": 5.277858597614198, + "flow_matching_loss": 0.01733652139082551, + "grad_norm": 0.11333293467760086, + "learning_rate": 1.109535914653495e-05, + "loss": 0.0209, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00904659335501492, + "rkd_loss_weight": 0.39549651741981506, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3906672537326813, + "rkd_teacher_entropy": 3.754715347290039, + "rkd_teacher_similarity_mean": 0.7871730506420136, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18140 + }, + { + "epoch": 5.280768111725342, + "flow_matching_loss": 0.012719581928104162, + "grad_norm": 0.14346849918365479, + "learning_rate": 1.1079396853319582e-05, + "loss": 0.0159, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008138358360156417, + "rkd_loss_weight": 0.39516317248344424, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3846787601709366, + "rkd_teacher_entropy": 3.7387081384658813, + "rkd_teacher_similarity_mean": 0.7820696234703064, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18150 + }, + { + "epoch": 5.283677625836486, + "flow_matching_loss": 0.013913521636277437, + "grad_norm": 0.09584322571754456, + "learning_rate": 1.1063439324006724e-05, + "loss": 0.0174, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008841276541352272, + "rkd_loss_weight": 0.3948298245668411, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3825412720441818, + "rkd_teacher_entropy": 3.736071991920471, + "rkd_teacher_similarity_mean": 0.7780450642108917, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18160 + }, + { + "epoch": 5.2865871399476285, + "flow_matching_loss": 0.01596480282023549, + "grad_norm": 0.10097100585699081, + "learning_rate": 1.1047486577986285e-05, + "loss": 0.0191, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007913587521761656, + "rkd_loss_weight": 0.39449648559093475, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38574838936328887, + "rkd_teacher_entropy": 3.7364279508590696, + "rkd_teacher_similarity_mean": 0.7805839776992798, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18170 + }, + { + "epoch": 5.289496654058772, + "flow_matching_loss": 0.01458544172346592, + "grad_norm": 0.12325957417488098, + "learning_rate": 1.1031538634642353e-05, + "loss": 0.0188, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010720117297023535, + "rkd_loss_weight": 0.39416313767433164, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3921475440263748, + "rkd_teacher_entropy": 3.755645251274109, + "rkd_teacher_similarity_mean": 0.7865599989891052, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18180 + }, + { + "epoch": 5.292406168169916, + "flow_matching_loss": 0.014529503416270018, + "grad_norm": 0.2120215892791748, + "learning_rate": 1.1015595513353172e-05, + "loss": 0.0176, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007906173495575786, + "rkd_loss_weight": 0.3938297927379608, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38556484282016756, + "rkd_teacher_entropy": 3.732545566558838, + "rkd_teacher_similarity_mean": 0.7791014909744263, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18190 + }, + { + "epoch": 5.295315682281059, + "flow_matching_loss": 0.013149645738303661, + "grad_norm": 0.10966597497463226, + "learning_rate": 1.0999657233491156e-05, + "loss": 0.0165, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008590393606573344, + "rkd_loss_weight": 0.39349645376205444, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38959750831127166, + "rkd_teacher_entropy": 3.749889850616455, + "rkd_teacher_similarity_mean": 0.7845771372318268, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18200 + }, + { + "epoch": 5.2982251963922025, + "flow_matching_loss": 0.013782541174441576, + "grad_norm": 0.1127820611000061, + "learning_rate": 1.0983723814422808e-05, + "loss": 0.0176, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00968498201109469, + "rkd_loss_weight": 0.3931631028652191, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38202885389328, + "rkd_teacher_entropy": 3.7366586208343504, + "rkd_teacher_similarity_mean": 0.7772453427314758, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18210 + }, + { + "epoch": 5.301134710503346, + "flow_matching_loss": 0.01640688618645072, + "grad_norm": 0.13297690451145172, + "learning_rate": 1.0967795275508735e-05, + "loss": 0.0201, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00946344891563058, + "rkd_loss_weight": 0.3928297609090805, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38446696996688845, + "rkd_teacher_entropy": 3.736145544052124, + "rkd_teacher_similarity_mean": 0.7789798378944397, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18220 + }, + { + "epoch": 5.304044224614489, + "flow_matching_loss": 0.01410022946074605, + "grad_norm": 0.10558122396469116, + "learning_rate": 1.095187163610361e-05, + "loss": 0.0178, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009439004585146904, + "rkd_loss_weight": 0.39249642193317413, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38513694405555726, + "rkd_teacher_entropy": 3.737053322792053, + "rkd_teacher_similarity_mean": 0.7796743512153625, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18230 + }, + { + "epoch": 5.306953738725633, + "flow_matching_loss": 0.01316250218078494, + "grad_norm": 0.10626554489135742, + "learning_rate": 1.0935952915556163e-05, + "loss": 0.0168, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009252638183534145, + "rkd_loss_weight": 0.39216306805610657, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38109139204025266, + "rkd_teacher_entropy": 3.7198425054550173, + "rkd_teacher_similarity_mean": 0.7759782135486603, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18240 + }, + { + "epoch": 5.309863252836776, + "flow_matching_loss": 0.013779789954423905, + "grad_norm": 0.12984886765480042, + "learning_rate": 1.0920039133209137e-05, + "loss": 0.0169, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008009268855676054, + "rkd_loss_weight": 0.3918297290802002, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3868537127971649, + "rkd_teacher_entropy": 3.7362393379211425, + "rkd_teacher_similarity_mean": 0.7816109836101532, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18250 + }, + { + "epoch": 5.31277276694792, + "flow_matching_loss": 0.014767464250326157, + "grad_norm": 0.11736045032739639, + "learning_rate": 1.0904130308399283e-05, + "loss": 0.0182, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008740299521014095, + "rkd_loss_weight": 0.39149638414382937, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3821833699941635, + "rkd_teacher_entropy": 3.7334394216537476, + "rkd_teacher_similarity_mean": 0.7780345499515533, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18260 + }, + { + "epoch": 5.315682281059063, + "flow_matching_loss": 0.015293083898723126, + "grad_norm": 0.13030138611793518, + "learning_rate": 1.0888226460457318e-05, + "loss": 0.0193, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010297625372186304, + "rkd_loss_weight": 0.39116303622722626, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3784483790397644, + "rkd_teacher_entropy": 3.7245192766189574, + "rkd_teacher_similarity_mean": 0.7761903643608093, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18270 + }, + { + "epoch": 5.318591795170207, + "flow_matching_loss": 0.014250316191464663, + "grad_norm": 0.1215590164065361, + "learning_rate": 1.087232760870792e-05, + "loss": 0.0177, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008913521142676472, + "rkd_loss_weight": 0.3908296972513199, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3827434927225113, + "rkd_teacher_entropy": 3.736507534980774, + "rkd_teacher_similarity_mean": 0.7800016760826111, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18280 + }, + { + "epoch": 5.32150130928135, + "flow_matching_loss": 0.0151600303594023, + "grad_norm": 0.11407576501369476, + "learning_rate": 1.0856433772469686e-05, + "loss": 0.0183, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00815622853115201, + "rkd_loss_weight": 0.3904963493347168, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38879933059215543, + "rkd_teacher_entropy": 3.758385682106018, + "rkd_teacher_similarity_mean": 0.7866605639457702, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18290 + }, + { + "epoch": 5.324410823392493, + "flow_matching_loss": 0.013168953079730272, + "grad_norm": 0.12355196475982666, + "learning_rate": 1.0840544971055135e-05, + "loss": 0.017, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009818439604714513, + "rkd_loss_weight": 0.39016300439834595, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3828652143478394, + "rkd_teacher_entropy": 3.7406888008117676, + "rkd_teacher_similarity_mean": 0.7800738334655761, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18300 + }, + { + "epoch": 5.327320337503637, + "flow_matching_loss": 0.015289750695228577, + "grad_norm": 0.10840165615081787, + "learning_rate": 1.0824661223770655e-05, + "loss": 0.0188, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008972523035481572, + "rkd_loss_weight": 0.3898296654224396, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3887000858783722, + "rkd_teacher_entropy": 3.75548939704895, + "rkd_teacher_similarity_mean": 0.7860162734985352, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18310 + }, + { + "epoch": 5.330229851614781, + "flow_matching_loss": 0.01371519649401307, + "grad_norm": 0.09005517512559891, + "learning_rate": 1.0808782549916498e-05, + "loss": 0.017, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00834338366985321, + "rkd_loss_weight": 0.389496311545372, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3781516581773758, + "rkd_teacher_entropy": 3.7171386957168577, + "rkd_teacher_similarity_mean": 0.7754336595535278, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18320 + }, + { + "epoch": 5.333139365725923, + "flow_matching_loss": 0.017297673597931863, + "grad_norm": 0.1262844204902649, + "learning_rate": 1.079290896878675e-05, + "loss": 0.0208, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009095894312486053, + "rkd_loss_weight": 0.38916297256946564, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38274829983711245, + "rkd_teacher_entropy": 3.7465555667877197, + "rkd_teacher_similarity_mean": 0.7812252819538117, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18330 + }, + { + "epoch": 5.336048879837067, + "flow_matching_loss": 0.014908191841095686, + "grad_norm": 0.1226666048169136, + "learning_rate": 1.07770404996693e-05, + "loss": 0.0187, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009718608064576983, + "rkd_loss_weight": 0.38882963061332704, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3740021467208862, + "rkd_teacher_entropy": 3.720041346549988, + "rkd_teacher_similarity_mean": 0.7707926452159881, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18340 + }, + { + "epoch": 5.338958393948211, + "flow_matching_loss": 0.012814925611019134, + "grad_norm": 0.10689310729503632, + "learning_rate": 1.0761177161845848e-05, + "loss": 0.0159, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008000634936615824, + "rkd_loss_weight": 0.3884962797164917, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37954002916812896, + "rkd_teacher_entropy": 3.7226922512054443, + "rkd_teacher_similarity_mean": 0.7762757956981658, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18350 + }, + { + "epoch": 5.3418679080593545, + "flow_matching_loss": 0.013953308295458555, + "grad_norm": 0.12612877786159515, + "learning_rate": 1.0745318974591841e-05, + "loss": 0.0177, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00967709645628929, + "rkd_loss_weight": 0.3881629407405853, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3885745137929916, + "rkd_teacher_entropy": 3.753244733810425, + "rkd_teacher_similarity_mean": 0.7848390460014343, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18360 + }, + { + "epoch": 5.344777422170497, + "flow_matching_loss": 0.014935180358588696, + "grad_norm": 0.13585993647575378, + "learning_rate": 1.0729465957176469e-05, + "loss": 0.0182, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008469291822984814, + "rkd_loss_weight": 0.3878295928239822, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3795093774795532, + "rkd_teacher_entropy": 3.734135317802429, + "rkd_teacher_similarity_mean": 0.7780311822891235, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18370 + }, + { + "epoch": 5.347686936281641, + "flow_matching_loss": 0.016013426892459392, + "grad_norm": 0.12291660904884338, + "learning_rate": 1.0713618128862642e-05, + "loss": 0.0189, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007515594316646457, + "rkd_loss_weight": 0.3874962478876114, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3963284820318222, + "rkd_teacher_entropy": 3.783415102958679, + "rkd_teacher_similarity_mean": 0.7947528958320618, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18380 + }, + { + "epoch": 5.350596450392785, + "flow_matching_loss": 0.013528777565807105, + "grad_norm": 0.14912594854831696, + "learning_rate": 1.0697775508906969e-05, + "loss": 0.0167, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008072787104174495, + "rkd_loss_weight": 0.387162908911705, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38085779547691345, + "rkd_teacher_entropy": 3.7339072704315184, + "rkd_teacher_similarity_mean": 0.7778485000133515, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18390 + }, + { + "epoch": 5.353505964503928, + "flow_matching_loss": 0.015149947162717581, + "grad_norm": 0.15151375532150269, + "learning_rate": 1.068193811655972e-05, + "loss": 0.0188, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009379574749618768, + "rkd_loss_weight": 0.3868295580148697, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37664671540260314, + "rkd_teacher_entropy": 3.714749050140381, + "rkd_teacher_similarity_mean": 0.7733345866203308, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18400 + }, + { + "epoch": 5.356415478615071, + "flow_matching_loss": 0.015605917200446129, + "grad_norm": 0.1402735859155655, + "learning_rate": 1.0666105971064835e-05, + "loss": 0.0195, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010181832127273083, + "rkd_loss_weight": 0.3864962160587311, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39170620739459994, + "rkd_teacher_entropy": 3.76808922290802, + "rkd_teacher_similarity_mean": 0.788997358083725, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18410 + }, + { + "epoch": 5.359324992726215, + "flow_matching_loss": 0.012553248275071382, + "grad_norm": 0.10257551074028015, + "learning_rate": 1.0650279091659852e-05, + "loss": 0.0153, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00704936203546822, + "rkd_loss_weight": 0.3861628770828247, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3915017664432526, + "rkd_teacher_entropy": 3.7663677453994753, + "rkd_teacher_similarity_mean": 0.789738142490387, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18420 + }, + { + "epoch": 5.362234506837358, + "flow_matching_loss": 0.015766152739524843, + "grad_norm": 0.1294785737991333, + "learning_rate": 1.0634457497575926e-05, + "loss": 0.019, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008371856482699513, + "rkd_loss_weight": 0.38582952320575714, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3845030635595322, + "rkd_teacher_entropy": 3.748825526237488, + "rkd_teacher_similarity_mean": 0.7831713557243347, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18430 + }, + { + "epoch": 5.3651440209485015, + "flow_matching_loss": 0.012707448191940784, + "grad_norm": 0.09511593729257584, + "learning_rate": 1.061864120803778e-05, + "loss": 0.0164, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00970515264198184, + "rkd_loss_weight": 0.38549618422985077, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.381957420706749, + "rkd_teacher_entropy": 3.7452966928482057, + "rkd_teacher_similarity_mean": 0.7793116688728332, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18440 + }, + { + "epoch": 5.368053535059645, + "flow_matching_loss": 0.013935290835797786, + "grad_norm": 0.13356059789657593, + "learning_rate": 1.0602830242263701e-05, + "loss": 0.017, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00783483386039734, + "rkd_loss_weight": 0.38516284227371217, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.389267286658287, + "rkd_teacher_entropy": 3.760854625701904, + "rkd_teacher_similarity_mean": 0.785817664861679, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18450 + }, + { + "epoch": 5.370963049170788, + "flow_matching_loss": 0.0148350662086159, + "grad_norm": 0.10729048401117325, + "learning_rate": 1.0587024619465506e-05, + "loss": 0.0178, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007713806536048651, + "rkd_loss_weight": 0.38482949137687683, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38819935023784635, + "rkd_teacher_entropy": 3.764089894294739, + "rkd_teacher_similarity_mean": 0.7849335551261902, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18460 + }, + { + "epoch": 5.373872563281932, + "flow_matching_loss": 0.012875851662829518, + "grad_norm": 0.08940937370061874, + "learning_rate": 1.0571224358848517e-05, + "loss": 0.016, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008170055411756038, + "rkd_loss_weight": 0.38449615240097046, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38292282819747925, + "rkd_teacher_entropy": 3.7347058773040773, + "rkd_teacher_similarity_mean": 0.7810585737228394, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18470 + }, + { + "epoch": 5.3767820773930755, + "flow_matching_loss": 0.01406774129718542, + "grad_norm": 0.10450098663568497, + "learning_rate": 1.0555429479611537e-05, + "loss": 0.0177, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009439568547531963, + "rkd_loss_weight": 0.38416280448436735, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39102146327495574, + "rkd_teacher_entropy": 3.7604936361312866, + "rkd_teacher_similarity_mean": 0.7889373362064361, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18480 + }, + { + "epoch": 5.379691591504219, + "flow_matching_loss": 0.013921454921364785, + "grad_norm": 0.1034068763256073, + "learning_rate": 1.0539640000946835e-05, + "loss": 0.017, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008096683025360107, + "rkd_loss_weight": 0.3838294595479965, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38414928019046785, + "rkd_teacher_entropy": 3.74664511680603, + "rkd_teacher_similarity_mean": 0.7810984969139099, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18490 + }, + { + "epoch": 5.382601105615362, + "flow_matching_loss": 0.013245486188679934, + "grad_norm": 0.11866546422243118, + "learning_rate": 1.0523855942040111e-05, + "loss": 0.0161, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007417175639420748, + "rkd_loss_weight": 0.38349612057209015, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39729043245315554, + "rkd_teacher_entropy": 3.775868844985962, + "rkd_teacher_similarity_mean": 0.7939732193946838, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18500 + }, + { + "epoch": 5.385510619726506, + "flow_matching_loss": 0.01560020875185728, + "grad_norm": 0.10294470936059952, + "learning_rate": 1.0508077322070496e-05, + "loss": 0.0187, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008049558941274881, + "rkd_loss_weight": 0.3831627696752548, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3851623237133026, + "rkd_teacher_entropy": 3.740228295326233, + "rkd_teacher_similarity_mean": 0.7804933249950409, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18510 + }, + { + "epoch": 5.388420133837649, + "flow_matching_loss": 0.0154572119936347, + "grad_norm": 0.13334181904792786, + "learning_rate": 1.0492304160210495e-05, + "loss": 0.0187, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008564070099964738, + "rkd_loss_weight": 0.3828294277191162, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3854646235704422, + "rkd_teacher_entropy": 3.7398815631866453, + "rkd_teacher_similarity_mean": 0.7819018065929413, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18520 + }, + { + "epoch": 5.391329647948792, + "flow_matching_loss": 0.012692270148545504, + "grad_norm": 0.12342040985822678, + "learning_rate": 1.0476536475625985e-05, + "loss": 0.0156, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007549444865435362, + "rkd_loss_weight": 0.3824960857629776, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3873851150274277, + "rkd_teacher_entropy": 3.7569405317306517, + "rkd_teacher_similarity_mean": 0.7840625822544098, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18530 + }, + { + "epoch": 5.394239162059936, + "flow_matching_loss": 0.01452653417363763, + "grad_norm": 0.14733056724071503, + "learning_rate": 1.0460774287476195e-05, + "loss": 0.0176, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008078892575576901, + "rkd_loss_weight": 0.3821627348661423, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.395072677731514, + "rkd_teacher_entropy": 3.7743804931640623, + "rkd_teacher_similarity_mean": 0.7926190793514252, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18540 + }, + { + "epoch": 5.39714867617108, + "flow_matching_loss": 0.013781472388654948, + "grad_norm": 0.15561695396900177, + "learning_rate": 1.044501761491366e-05, + "loss": 0.017, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008386164670810103, + "rkd_loss_weight": 0.3818293958902359, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38433613777160647, + "rkd_teacher_entropy": 3.746696400642395, + "rkd_teacher_similarity_mean": 0.7818399786949157, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18550 + }, + { + "epoch": 5.400058190282223, + "flow_matching_loss": 0.013561790343374014, + "grad_norm": 0.11215226352214813, + "learning_rate": 1.0429266477084235e-05, + "loss": 0.0166, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007969857566058636, + "rkd_loss_weight": 0.3814960509538651, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38550729751586915, + "rkd_teacher_entropy": 3.7405553817749024, + "rkd_teacher_similarity_mean": 0.7817425191402435, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18560 + }, + { + "epoch": 5.402967704393366, + "flow_matching_loss": 0.015051846392452717, + "grad_norm": 0.1313440501689911, + "learning_rate": 1.0413520893127034e-05, + "loss": 0.0185, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009105160692706704, + "rkd_loss_weight": 0.38116270303726196, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37717447280883787, + "rkd_teacher_entropy": 3.720692276954651, + "rkd_teacher_similarity_mean": 0.7733867347240448, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18570 + }, + { + "epoch": 5.40587721850451, + "flow_matching_loss": 0.012187412846833468, + "grad_norm": 0.12355964630842209, + "learning_rate": 1.0397780882174424e-05, + "loss": 0.016, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009890725137665868, + "rkd_loss_weight": 0.3808293640613556, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3868157953023911, + "rkd_teacher_entropy": 3.7417689323425294, + "rkd_teacher_similarity_mean": 0.7827053248882294, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18580 + }, + { + "epoch": 5.408786732615654, + "flow_matching_loss": 0.015096977539360523, + "grad_norm": 0.10670827329158783, + "learning_rate": 1.0382046463351998e-05, + "loss": 0.0186, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009269351325929164, + "rkd_loss_weight": 0.38049601316452025, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3813810974359512, + "rkd_teacher_entropy": 3.7352916955947877, + "rkd_teacher_similarity_mean": 0.7794749975204468, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18590 + }, + { + "epoch": 5.411696246726796, + "flow_matching_loss": 0.014310559909790754, + "grad_norm": 0.15410004556179047, + "learning_rate": 1.0366317655778568e-05, + "loss": 0.0173, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007783546578139067, + "rkd_loss_weight": 0.38016267120838165, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3917637437582016, + "rkd_teacher_entropy": 3.7655731439590454, + "rkd_teacher_similarity_mean": 0.7886327981948853, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18600 + }, + { + "epoch": 5.41460576083794, + "flow_matching_loss": 0.015372804738581181, + "grad_norm": 0.09440214931964874, + "learning_rate": 1.035059447856611e-05, + "loss": 0.018, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0068853470962494615, + "rkd_loss_weight": 0.3798293322324753, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3873211830854416, + "rkd_teacher_entropy": 3.750714898109436, + "rkd_teacher_similarity_mean": 0.7846387922763824, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18610 + }, + { + "epoch": 5.417515274949084, + "flow_matching_loss": 0.013417213596403599, + "grad_norm": 0.09744429588317871, + "learning_rate": 1.0334876950819777e-05, + "loss": 0.0163, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007628565095365048, + "rkd_loss_weight": 0.3794959783554077, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38156236708164215, + "rkd_teacher_entropy": 3.7266119718551636, + "rkd_teacher_similarity_mean": 0.7778575122356415, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18620 + }, + { + "epoch": 5.420424789060227, + "flow_matching_loss": 0.012946874741464854, + "grad_norm": 0.10120096057653427, + "learning_rate": 1.0319165091637837e-05, + "loss": 0.0161, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00820364011451602, + "rkd_loss_weight": 0.37916263937950134, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38876845240592955, + "rkd_teacher_entropy": 3.75244677066803, + "rkd_teacher_similarity_mean": 0.784942764043808, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18630 + }, + { + "epoch": 5.42333430317137, + "flow_matching_loss": 0.013374126888811589, + "grad_norm": 0.09979993104934692, + "learning_rate": 1.0303458920111686e-05, + "loss": 0.0176, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011128830490633845, + "rkd_loss_weight": 0.3788292944431305, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39183357954025266, + "rkd_teacher_entropy": 3.7667824983596803, + "rkd_teacher_similarity_mean": 0.7891761720180511, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18640 + }, + { + "epoch": 5.426243817282514, + "flow_matching_loss": 0.012601640075445175, + "grad_norm": 0.16650819778442383, + "learning_rate": 1.0287758455325803e-05, + "loss": 0.0158, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008442805940285325, + "rkd_loss_weight": 0.3784959465265274, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3801669180393219, + "rkd_teacher_entropy": 3.7340648412704467, + "rkd_teacher_similarity_mean": 0.7763871908187866, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18650 + }, + { + "epoch": 5.429153331393657, + "flow_matching_loss": 0.013320984691381455, + "grad_norm": 0.11168943345546722, + "learning_rate": 1.0272063716357729e-05, + "loss": 0.0165, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008321406040340661, + "rkd_loss_weight": 0.37816260755062103, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37796256840229037, + "rkd_teacher_entropy": 3.713219904899597, + "rkd_teacher_similarity_mean": 0.7739410698413849, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18660 + }, + { + "epoch": 5.432062845504801, + "flow_matching_loss": 0.011236143345013262, + "grad_norm": 0.18166440725326538, + "learning_rate": 1.0256374722278057e-05, + "loss": 0.0144, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008337573613971472, + "rkd_loss_weight": 0.3778292626142502, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38736557960510254, + "rkd_teacher_entropy": 3.75637469291687, + "rkd_teacher_similarity_mean": 0.7845349907875061, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18670 + }, + { + "epoch": 5.434972359615944, + "flow_matching_loss": 0.014840792305767537, + "grad_norm": 0.14435233175754547, + "learning_rate": 1.0240691492150388e-05, + "loss": 0.0177, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007621364109218121, + "rkd_loss_weight": 0.3774959146976471, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3850031137466431, + "rkd_teacher_entropy": 3.7473708152770997, + "rkd_teacher_similarity_mean": 0.7829352855682373, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18680 + }, + { + "epoch": 5.437881873727088, + "flow_matching_loss": 0.014539999049156904, + "grad_norm": 0.10690046101808548, + "learning_rate": 1.0225014045031334e-05, + "loss": 0.0174, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007565008709207177, + "rkd_loss_weight": 0.3771625757217407, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3875956565141678, + "rkd_teacher_entropy": 3.750115704536438, + "rkd_teacher_similarity_mean": 0.7844063937664032, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18690 + }, + { + "epoch": 5.440791387838231, + "flow_matching_loss": 0.014517427515238523, + "grad_norm": 0.13418319821357727, + "learning_rate": 1.0209342399970468e-05, + "loss": 0.0176, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008241490600630641, + "rkd_loss_weight": 0.3768292248249054, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38846769332885744, + "rkd_teacher_entropy": 3.752187204360962, + "rkd_teacher_similarity_mean": 0.7849414467811584, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18700 + }, + { + "epoch": 5.4437009019493745, + "flow_matching_loss": 0.01350990803912282, + "grad_norm": 0.21613755822181702, + "learning_rate": 1.019367657601031e-05, + "loss": 0.017, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009398403530940413, + "rkd_loss_weight": 0.3764958828687668, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.392218092083931, + "rkd_teacher_entropy": 3.766426610946655, + "rkd_teacher_similarity_mean": 0.7877313077449799, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18710 + }, + { + "epoch": 5.446610416060518, + "flow_matching_loss": 0.014932684972882271, + "grad_norm": 0.12085069715976715, + "learning_rate": 1.0178016592186323e-05, + "loss": 0.018, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008197857206687332, + "rkd_loss_weight": 0.3761625438928604, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38809109926223756, + "rkd_teacher_entropy": 3.7656555652618406, + "rkd_teacher_similarity_mean": 0.7846464097499848, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18720 + }, + { + "epoch": 5.449519930171661, + "flow_matching_loss": 0.012820542929694057, + "grad_norm": 0.11214613914489746, + "learning_rate": 1.0162362467526861e-05, + "loss": 0.0166, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010140908090397716, + "rkd_loss_weight": 0.37582919001579285, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3805496126413345, + "rkd_teacher_entropy": 3.733042526245117, + "rkd_teacher_similarity_mean": 0.778374719619751, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18730 + }, + { + "epoch": 5.452429444282805, + "flow_matching_loss": 0.01641819002106786, + "grad_norm": 0.13540467619895935, + "learning_rate": 1.0146714221053159e-05, + "loss": 0.0194, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007944308640435337, + "rkd_loss_weight": 0.3754958510398865, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38079687356948855, + "rkd_teacher_entropy": 3.7319682836532593, + "rkd_teacher_similarity_mean": 0.7766648232936859, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18740 + }, + { + "epoch": 5.4553389583939484, + "flow_matching_loss": 0.014728565607219935, + "grad_norm": 0.1119328960776329, + "learning_rate": 1.0131071871779314e-05, + "loss": 0.0171, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006393078062683344, + "rkd_loss_weight": 0.37516250610351565, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38813226521015165, + "rkd_teacher_entropy": 3.7458861827850343, + "rkd_teacher_similarity_mean": 0.7838988721370697, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18750 + }, + { + "epoch": 5.458248472505091, + "flow_matching_loss": 0.01391244437545538, + "grad_norm": 0.1381942331790924, + "learning_rate": 1.0115435438712249e-05, + "loss": 0.017, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008327389927580952, + "rkd_loss_weight": 0.37482915818691254, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3870064228773117, + "rkd_teacher_entropy": 3.7565358877182007, + "rkd_teacher_similarity_mean": 0.7847128748893738, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18760 + }, + { + "epoch": 5.461157986616235, + "flow_matching_loss": 0.014579894207417964, + "grad_norm": 0.12866440415382385, + "learning_rate": 1.0099804940851716e-05, + "loss": 0.0175, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007876731641590595, + "rkd_loss_weight": 0.37449581921100616, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3850386530160904, + "rkd_teacher_entropy": 3.733672094345093, + "rkd_teacher_similarity_mean": 0.7814016044139862, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18770 + }, + { + "epoch": 5.464067500727379, + "flow_matching_loss": 0.013757744245231152, + "grad_norm": 0.10265292972326279, + "learning_rate": 1.0084180397190238e-05, + "loss": 0.0171, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00905072158202529, + "rkd_loss_weight": 0.37416247129440305, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3857186585664749, + "rkd_teacher_entropy": 3.751433825492859, + "rkd_teacher_similarity_mean": 0.7818217217922211, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18780 + }, + { + "epoch": 5.4669770148385215, + "flow_matching_loss": 0.013886014837771654, + "grad_norm": 0.12230198830366135, + "learning_rate": 1.006856182671311e-05, + "loss": 0.0177, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010324618313461542, + "rkd_loss_weight": 0.3738291263580322, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37344678640365603, + "rkd_teacher_entropy": 3.708458495140076, + "rkd_teacher_similarity_mean": 0.7685448229312897, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18790 + }, + { + "epoch": 5.469886528949665, + "flow_matching_loss": 0.015793214086443186, + "grad_norm": 0.12442748248577118, + "learning_rate": 1.0052949248398367e-05, + "loss": 0.0195, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.01004016175866127, + "rkd_loss_weight": 0.37349578738212585, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38230122029781344, + "rkd_teacher_entropy": 3.7518893241882325, + "rkd_teacher_similarity_mean": 0.7798469126224518, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18800 + }, + { + "epoch": 5.472796043060809, + "flow_matching_loss": 0.012626920687034726, + "grad_norm": 0.14809884130954742, + "learning_rate": 1.003734268121676e-05, + "loss": 0.0167, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010835978342220188, + "rkd_loss_weight": 0.3731624335050583, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38697092831134794, + "rkd_teacher_entropy": 3.755920100212097, + "rkd_teacher_similarity_mean": 0.7835931479930878, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18810 + }, + { + "epoch": 5.475705557171953, + "flow_matching_loss": 0.014629080798476935, + "grad_norm": 0.12744945287704468, + "learning_rate": 1.0021742144131744e-05, + "loss": 0.0178, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008403012994676828, + "rkd_loss_weight": 0.3728290945291519, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3819166600704193, + "rkd_teacher_entropy": 3.726690483093262, + "rkd_teacher_similarity_mean": 0.7791210889816285, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18820 + }, + { + "epoch": 5.478615071283095, + "flow_matching_loss": 0.013269651774317026, + "grad_norm": 0.1333579570055008, + "learning_rate": 1.000614765609944e-05, + "loss": 0.0161, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007524949032813311, + "rkd_loss_weight": 0.3724957525730133, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3852552086114883, + "rkd_teacher_entropy": 3.745557451248169, + "rkd_teacher_similarity_mean": 0.7826406419277191, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18830 + }, + { + "epoch": 5.481524585394239, + "flow_matching_loss": 0.014183822087943554, + "grad_norm": 0.08936091512441635, + "learning_rate": 9.990559236068628e-06, + "loss": 0.017, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007433829084038735, + "rkd_loss_weight": 0.372162401676178, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3868942826986313, + "rkd_teacher_entropy": 3.7448565006256103, + "rkd_teacher_similarity_mean": 0.7835212171077728, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18840 + }, + { + "epoch": 5.484434099505383, + "flow_matching_loss": 0.015226296801120042, + "grad_norm": 0.14582142233848572, + "learning_rate": 9.974976902980703e-06, + "loss": 0.0185, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008906079921871424, + "rkd_loss_weight": 0.3718290627002716, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3857954055070877, + "rkd_teacher_entropy": 3.7432409048080446, + "rkd_teacher_similarity_mean": 0.7829107403755188, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18850 + }, + { + "epoch": 5.487343613616526, + "flow_matching_loss": 0.012639936432242394, + "grad_norm": 0.11368010193109512, + "learning_rate": 9.959400675769668e-06, + "loss": 0.0165, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010271093528717756, + "rkd_loss_weight": 0.3714957147836685, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3895038962364197, + "rkd_teacher_entropy": 3.7565399169921876, + "rkd_teacher_similarity_mean": 0.7876225292682648, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18860 + }, + { + "epoch": 5.490253127727669, + "flow_matching_loss": 0.013625963032245636, + "grad_norm": 0.12396068871021271, + "learning_rate": 9.943830573362106e-06, + "loss": 0.0164, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007479551527649164, + "rkd_loss_weight": 0.37116236984729767, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3788144916296005, + "rkd_teacher_entropy": 3.719012451171875, + "rkd_teacher_similarity_mean": 0.7753188133239746, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18870 + }, + { + "epoch": 5.493162641838813, + "flow_matching_loss": 0.012942242436110973, + "grad_norm": 0.10645463317632675, + "learning_rate": 9.928266614677168e-06, + "loss": 0.0161, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008502928726375102, + "rkd_loss_weight": 0.3708290308713913, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.389976492524147, + "rkd_teacher_entropy": 3.7629366874694825, + "rkd_teacher_similarity_mean": 0.7866086840629578, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18880 + }, + { + "epoch": 5.496072155949957, + "flow_matching_loss": 0.013496694061905146, + "grad_norm": 0.15358808636665344, + "learning_rate": 9.912708818626524e-06, + "loss": 0.0162, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007318219589069486, + "rkd_loss_weight": 0.3704956829547882, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38507344424724577, + "rkd_teacher_entropy": 3.7377546072006225, + "rkd_teacher_similarity_mean": 0.7811358153820038, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18890 + }, + { + "epoch": 5.4989816700611, + "flow_matching_loss": 0.01302338419482112, + "grad_norm": 0.09241962432861328, + "learning_rate": 9.897157204114369e-06, + "loss": 0.0158, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007555444026365876, + "rkd_loss_weight": 0.37016233801841736, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38281813859939573, + "rkd_teacher_entropy": 3.7304664850234985, + "rkd_teacher_similarity_mean": 0.7801417708396912, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18900 + }, + { + "epoch": 5.501891184172243, + "flow_matching_loss": 0.01249561351723969, + "grad_norm": 0.08922839909791946, + "learning_rate": 9.881611790037375e-06, + "loss": 0.0153, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007702673459425569, + "rkd_loss_weight": 0.369828999042511, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3948790580034256, + "rkd_teacher_entropy": 3.769077682495117, + "rkd_teacher_similarity_mean": 0.791460371017456, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18910 + }, + { + "epoch": 5.504800698283387, + "flow_matching_loss": 0.015112499985843898, + "grad_norm": 0.1317487359046936, + "learning_rate": 9.866072595284684e-06, + "loss": 0.0179, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0075802597217261795, + "rkd_loss_weight": 0.3694956451654434, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3959277838468552, + "rkd_teacher_entropy": 3.775772047042847, + "rkd_teacher_similarity_mean": 0.7938357949256897, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18920 + }, + { + "epoch": 5.50771021239453, + "flow_matching_loss": 0.013876820169389247, + "grad_norm": 0.12146107852458954, + "learning_rate": 9.850539638737894e-06, + "loss": 0.0175, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009852054016664624, + "rkd_loss_weight": 0.36916230618953705, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38747820258140564, + "rkd_teacher_entropy": 3.7647981882095336, + "rkd_teacher_similarity_mean": 0.786593359708786, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18930 + }, + { + "epoch": 5.5106197265056736, + "flow_matching_loss": 0.01218960015103221, + "grad_norm": 0.12456505000591278, + "learning_rate": 9.835012939271006e-06, + "loss": 0.0156, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009302908927202225, + "rkd_loss_weight": 0.36882896423339845, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3896078705787659, + "rkd_teacher_entropy": 3.7622538566589356, + "rkd_teacher_similarity_mean": 0.7864825904369355, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18940 + }, + { + "epoch": 5.513529240616817, + "flow_matching_loss": 0.014188263984397053, + "grad_norm": 0.1442098766565323, + "learning_rate": 9.819492515750418e-06, + "loss": 0.0179, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010197174781933428, + "rkd_loss_weight": 0.3684956133365631, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3855878621339798, + "rkd_teacher_entropy": 3.7525303602218627, + "rkd_teacher_similarity_mean": 0.7839270293712616, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18950 + }, + { + "epoch": 5.51643875472796, + "flow_matching_loss": 0.01368238814175129, + "grad_norm": 0.1188138872385025, + "learning_rate": 9.803978387034918e-06, + "loss": 0.0171, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009387949015945196, + "rkd_loss_weight": 0.36816227436065674, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3861252009868622, + "rkd_teacher_entropy": 3.754753661155701, + "rkd_teacher_similarity_mean": 0.7826702296733856, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18960 + }, + { + "epoch": 5.519348268839104, + "flow_matching_loss": 0.015169345494359731, + "grad_norm": 0.09026435017585754, + "learning_rate": 9.788470571975624e-06, + "loss": 0.0181, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008051464008167386, + "rkd_loss_weight": 0.3678289264440536, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37728654146194457, + "rkd_teacher_entropy": 3.7193347215652466, + "rkd_teacher_similarity_mean": 0.7741157531738281, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18970 + }, + { + "epoch": 5.5222577829502475, + "flow_matching_loss": 0.01551768514327705, + "grad_norm": 0.13621175289154053, + "learning_rate": 9.772969089416008e-06, + "loss": 0.0184, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007744950475171208, + "rkd_loss_weight": 0.3674955815076828, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.393805605173111, + "rkd_teacher_entropy": 3.7682251214981077, + "rkd_teacher_similarity_mean": 0.7922203540802002, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18980 + }, + { + "epoch": 5.52516729706139, + "flow_matching_loss": 0.014452950563281775, + "grad_norm": 0.19155940413475037, + "learning_rate": 9.757473958191822e-06, + "loss": 0.0177, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008932931441813707, + "rkd_loss_weight": 0.36716224253177643, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3843419313430786, + "rkd_teacher_entropy": 3.747865414619446, + "rkd_teacher_similarity_mean": 0.7829476952552795, + "rkd_vlm_temp": 0.10000000149011612, + "step": 18990 + }, + { + "epoch": 5.528076811172534, + "flow_matching_loss": 0.014607917051762342, + "grad_norm": 0.2604758143424988, + "learning_rate": 9.741985197131118e-06, + "loss": 0.0188, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011369865480810403, + "rkd_loss_weight": 0.3668288916349411, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3823282033205032, + "rkd_teacher_entropy": 3.7424532175064087, + "rkd_teacher_similarity_mean": 0.7805102109909058, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19000 + }, + { + "epoch": 5.530986325283678, + "flow_matching_loss": 0.013498740270733833, + "grad_norm": 0.09033159911632538, + "learning_rate": 9.726502825054195e-06, + "loss": 0.0166, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008369132271036505, + "rkd_loss_weight": 0.3664955496788025, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3798125684261322, + "rkd_teacher_entropy": 3.726536178588867, + "rkd_teacher_similarity_mean": 0.7763506889343261, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19010 + }, + { + "epoch": 5.533895839394821, + "flow_matching_loss": 0.012858963292092084, + "grad_norm": 0.1281682550907135, + "learning_rate": 9.711026860773595e-06, + "loss": 0.0164, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009578770166262985, + "rkd_loss_weight": 0.3661622077226639, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38171955943107605, + "rkd_teacher_entropy": 3.7327091455459596, + "rkd_teacher_similarity_mean": 0.7793770670890808, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19020 + }, + { + "epoch": 5.536805353505964, + "flow_matching_loss": 0.015652081184089183, + "grad_norm": 0.11253971606492996, + "learning_rate": 9.69555732309408e-06, + "loss": 0.0181, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0066982688382267955, + "rkd_loss_weight": 0.36582885682582855, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3858668446540833, + "rkd_teacher_entropy": 3.7514936447143556, + "rkd_teacher_similarity_mean": 0.7846804678440094, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19030 + }, + { + "epoch": 5.539714867617108, + "flow_matching_loss": 0.015696996916085483, + "grad_norm": 0.15667793154716492, + "learning_rate": 9.68009423081259e-06, + "loss": 0.0187, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008337355544790625, + "rkd_loss_weight": 0.3654955178499222, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38414581418037413, + "rkd_teacher_entropy": 3.7406391382217405, + "rkd_teacher_similarity_mean": 0.781990784406662, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19040 + }, + { + "epoch": 5.542624381728252, + "flow_matching_loss": 0.014514983305707574, + "grad_norm": 0.1291312873363495, + "learning_rate": 9.664637602718249e-06, + "loss": 0.0177, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008836918883025647, + "rkd_loss_weight": 0.36516217291355135, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37427602112293246, + "rkd_teacher_entropy": 3.718691921234131, + "rkd_teacher_similarity_mean": 0.7728536605834961, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19050 + }, + { + "epoch": 5.5455338958393945, + "flow_matching_loss": 0.016550262179225683, + "grad_norm": 0.13520050048828125, + "learning_rate": 9.649187457592313e-06, + "loss": 0.0197, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008703551767393947, + "rkd_loss_weight": 0.36482882499694824, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38355956971645355, + "rkd_teacher_entropy": 3.7362881183624266, + "rkd_teacher_similarity_mean": 0.7801513433456421, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19060 + }, + { + "epoch": 5.548443409950538, + "flow_matching_loss": 0.014674823312088848, + "grad_norm": 0.12014176696538925, + "learning_rate": 9.63374381420816e-06, + "loss": 0.0178, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008632830902934074, + "rkd_loss_weight": 0.36449548602104187, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3809558779001236, + "rkd_teacher_entropy": 3.73399441242218, + "rkd_teacher_similarity_mean": 0.7775800824165344, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19070 + }, + { + "epoch": 5.551352924061682, + "flow_matching_loss": 0.015197000186890363, + "grad_norm": 0.12158652395009995, + "learning_rate": 9.618306691331279e-06, + "loss": 0.0187, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009615510329604149, + "rkd_loss_weight": 0.36416213512420653, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38764487504959105, + "rkd_teacher_entropy": 3.7523582220077514, + "rkd_teacher_similarity_mean": 0.7840466439723969, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19080 + }, + { + "epoch": 5.554262438172826, + "flow_matching_loss": 0.01327782180160284, + "grad_norm": 0.11530046910047531, + "learning_rate": 9.602876107719228e-06, + "loss": 0.0162, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008166473032906652, + "rkd_loss_weight": 0.36382879316806793, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38733904957771303, + "rkd_teacher_entropy": 3.746823215484619, + "rkd_teacher_similarity_mean": 0.7834905683994293, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19090 + }, + { + "epoch": 5.557171952283968, + "flow_matching_loss": 0.014723822847008705, + "grad_norm": 0.15876959264278412, + "learning_rate": 9.587452082121621e-06, + "loss": 0.0176, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008030050480738283, + "rkd_loss_weight": 0.36349545419216156, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39641195833683013, + "rkd_teacher_entropy": 3.7703731775283815, + "rkd_teacher_similarity_mean": 0.7938444912433624, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19100 + }, + { + "epoch": 5.560081466395112, + "flow_matching_loss": 0.015657761506736277, + "grad_norm": 0.10069136321544647, + "learning_rate": 9.57203463328011e-06, + "loss": 0.0182, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00706097730435431, + "rkd_loss_weight": 0.3631621032953262, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3870761752128601, + "rkd_teacher_entropy": 3.7524784088134764, + "rkd_teacher_similarity_mean": 0.7844051063060761, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19110 + }, + { + "epoch": 5.562990980506255, + "flow_matching_loss": 0.014929930586367846, + "grad_norm": 0.14155767858028412, + "learning_rate": 9.556623779928341e-06, + "loss": 0.0181, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008634003950282932, + "rkd_loss_weight": 0.3628287613391876, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38396011888980863, + "rkd_teacher_entropy": 3.745244336128235, + "rkd_teacher_similarity_mean": 0.7797833621501923, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19120 + }, + { + "epoch": 5.565900494617399, + "flow_matching_loss": 0.014901773072779179, + "grad_norm": 0.11051974445581436, + "learning_rate": 9.541219540791949e-06, + "loss": 0.0179, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00839628973044455, + "rkd_loss_weight": 0.362495419383049, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3807242840528488, + "rkd_teacher_entropy": 3.7362214088439942, + "rkd_teacher_similarity_mean": 0.7758479118347168, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19130 + }, + { + "epoch": 5.568810008728542, + "flow_matching_loss": 0.014873104821890592, + "grad_norm": 0.1279737949371338, + "learning_rate": 9.525821934588551e-06, + "loss": 0.0181, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008949492778629064, + "rkd_loss_weight": 0.3621620684862137, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3896672874689102, + "rkd_teacher_entropy": 3.756749391555786, + "rkd_teacher_similarity_mean": 0.7855954468250275, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19140 + }, + { + "epoch": 5.571719522839686, + "flow_matching_loss": 0.014430228620767593, + "grad_norm": 0.11545747518539429, + "learning_rate": 9.510430980027684e-06, + "loss": 0.0174, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008236369071528315, + "rkd_loss_weight": 0.3618287295103073, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3817524701356888, + "rkd_teacher_entropy": 3.7403303384780884, + "rkd_teacher_similarity_mean": 0.7797148108482361, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19150 + }, + { + "epoch": 5.574629036950829, + "flow_matching_loss": 0.015565591398626566, + "grad_norm": 0.11853135377168655, + "learning_rate": 9.495046695810809e-06, + "loss": 0.0188, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008977377321571112, + "rkd_loss_weight": 0.3614953845739365, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3761819988489151, + "rkd_teacher_entropy": 3.7196951866149903, + "rkd_teacher_similarity_mean": 0.7739126801490783, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19160 + }, + { + "epoch": 5.577538551061973, + "flow_matching_loss": 0.011751068755984306, + "grad_norm": 0.10419289022684097, + "learning_rate": 9.47966910063128e-06, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007615610025823116, + "rkd_loss_weight": 0.3611620366573334, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38844835162162783, + "rkd_teacher_entropy": 3.7589377641677855, + "rkd_teacher_similarity_mean": 0.7864994943141937, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19170 + }, + { + "epoch": 5.580448065173116, + "flow_matching_loss": 0.014921037340536713, + "grad_norm": 0.10896266251802444, + "learning_rate": 9.464298213174325e-06, + "loss": 0.0175, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007218017568811774, + "rkd_loss_weight": 0.360828697681427, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.389155176281929, + "rkd_teacher_entropy": 3.7640448331832888, + "rkd_teacher_similarity_mean": 0.7878100275993347, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19180 + }, + { + "epoch": 5.583357579284259, + "flow_matching_loss": 0.014797660615295171, + "grad_norm": 0.11878225207328796, + "learning_rate": 9.448934052117024e-06, + "loss": 0.0178, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008349504647776484, + "rkd_loss_weight": 0.36049534678459166, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3867450714111328, + "rkd_teacher_entropy": 3.7490558385849, + "rkd_teacher_similarity_mean": 0.7849937617778778, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19190 + }, + { + "epoch": 5.586267093395403, + "flow_matching_loss": 0.015020151436328889, + "grad_norm": 0.129739910364151, + "learning_rate": 9.433576636128284e-06, + "loss": 0.0179, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00797702819108963, + "rkd_loss_weight": 0.36016200482845306, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3849485754966736, + "rkd_teacher_entropy": 3.741564726829529, + "rkd_teacher_similarity_mean": 0.7822735249996186, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19200 + }, + { + "epoch": 5.5891766075065465, + "flow_matching_loss": 0.01282137082889676, + "grad_norm": 0.11655597388744354, + "learning_rate": 9.41822598386881e-06, + "loss": 0.0158, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00822284510359168, + "rkd_loss_weight": 0.3598286658525467, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.378058859705925, + "rkd_teacher_entropy": 3.733080840110779, + "rkd_teacher_similarity_mean": 0.7762285232543945, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19210 + }, + { + "epoch": 5.59208612161769, + "flow_matching_loss": 0.014823104534298182, + "grad_norm": 0.1408599317073822, + "learning_rate": 9.402882113991094e-06, + "loss": 0.018, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008778395783156156, + "rkd_loss_weight": 0.3594953119754791, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39174685180187224, + "rkd_teacher_entropy": 3.7685508012771605, + "rkd_teacher_similarity_mean": 0.7917436838150025, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19220 + }, + { + "epoch": 5.594995635728833, + "flow_matching_loss": 0.0151581808924675, + "grad_norm": 0.15187564492225647, + "learning_rate": 9.387545045139381e-06, + "loss": 0.0183, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008679694356396794, + "rkd_loss_weight": 0.35916197299957275, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38396369814872744, + "rkd_teacher_entropy": 3.747377109527588, + "rkd_teacher_similarity_mean": 0.7809351801872253, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19230 + }, + { + "epoch": 5.597905149839977, + "flow_matching_loss": 0.01418868750333786, + "grad_norm": 0.0933331772685051, + "learning_rate": 9.372214795949655e-06, + "loss": 0.0172, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008459172490984202, + "rkd_loss_weight": 0.3588286280632019, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3851057171821594, + "rkd_teacher_entropy": 3.7346983671188356, + "rkd_teacher_similarity_mean": 0.7825183928012848, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19240 + }, + { + "epoch": 5.6008146639511205, + "flow_matching_loss": 0.01624138131737709, + "grad_norm": 0.13550132513046265, + "learning_rate": 9.356891385049618e-06, + "loss": 0.0195, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009122904948890209, + "rkd_loss_weight": 0.3584952801465988, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38481564819812775, + "rkd_teacher_entropy": 3.7376336336135862, + "rkd_teacher_similarity_mean": 0.7820718288421631, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19250 + }, + { + "epoch": 5.603724178062263, + "flow_matching_loss": 0.013711008802056312, + "grad_norm": 0.1313135027885437, + "learning_rate": 9.341574831058664e-06, + "loss": 0.0165, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00785947684198618, + "rkd_loss_weight": 0.35816194117069244, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3855238288640976, + "rkd_teacher_entropy": 3.7518937587738037, + "rkd_teacher_similarity_mean": 0.7837838709354401, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19260 + }, + { + "epoch": 5.606633692173407, + "flow_matching_loss": 0.011779371416196227, + "grad_norm": 0.13460153341293335, + "learning_rate": 9.326265152587844e-06, + "loss": 0.0143, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006951420241966844, + "rkd_loss_weight": 0.35782859325408933, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3835281729698181, + "rkd_teacher_entropy": 3.738928747177124, + "rkd_teacher_similarity_mean": 0.7808833062648773, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19270 + }, + { + "epoch": 5.609543206284551, + "flow_matching_loss": 0.016814441606402396, + "grad_norm": 0.1475302278995514, + "learning_rate": 9.310962368239861e-06, + "loss": 0.0201, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009054882591590286, + "rkd_loss_weight": 0.3574952483177185, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3768090546131134, + "rkd_teacher_entropy": 3.714671087265015, + "rkd_teacher_similarity_mean": 0.7733791410923004, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19280 + }, + { + "epoch": 5.6124527203956935, + "flow_matching_loss": 0.013322852831333876, + "grad_norm": 0.12336583435535431, + "learning_rate": 9.295666496609036e-06, + "loss": 0.0166, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009255923004820943, + "rkd_loss_weight": 0.35716190934181213, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37730841934680937, + "rkd_teacher_entropy": 3.711124300956726, + "rkd_teacher_similarity_mean": 0.7737025380134582, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19290 + }, + { + "epoch": 5.615362234506837, + "flow_matching_loss": 0.017030993942171335, + "grad_norm": 0.12689010798931122, + "learning_rate": 9.280377556281306e-06, + "loss": 0.0197, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007604671223089099, + "rkd_loss_weight": 0.35682855546474457, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3834738671779633, + "rkd_teacher_entropy": 3.7540758848190308, + "rkd_teacher_similarity_mean": 0.7813916504383087, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19300 + }, + { + "epoch": 5.618271748617981, + "flow_matching_loss": 0.013876956608146428, + "grad_norm": 0.10663988441228867, + "learning_rate": 9.265095565834172e-06, + "loss": 0.0166, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007666701963171363, + "rkd_loss_weight": 0.3564952164888382, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.381978902220726, + "rkd_teacher_entropy": 3.7431122064590454, + "rkd_teacher_similarity_mean": 0.7804988265037537, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19310 + }, + { + "epoch": 5.621181262729124, + "flow_matching_loss": 0.016436332557350396, + "grad_norm": 0.1293450891971588, + "learning_rate": 9.249820543836688e-06, + "loss": 0.0192, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00789463329128921, + "rkd_loss_weight": 0.3561618775129318, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3939783364534378, + "rkd_teacher_entropy": 3.77505099773407, + "rkd_teacher_similarity_mean": 0.7916499137878418, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19320 + }, + { + "epoch": 5.6240907768402675, + "flow_matching_loss": 0.012899624276906252, + "grad_norm": 0.10007073730230331, + "learning_rate": 9.234552508849454e-06, + "loss": 0.0152, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006504094647243619, + "rkd_loss_weight": 0.35582852363586426, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38644673526287077, + "rkd_teacher_entropy": 3.7374465465545654, + "rkd_teacher_similarity_mean": 0.7837383151054382, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19330 + }, + { + "epoch": 5.627000290951411, + "flow_matching_loss": 0.013983975164592266, + "grad_norm": 0.1304953694343567, + "learning_rate": 9.21929147942456e-06, + "loss": 0.0168, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00797174433246255, + "rkd_loss_weight": 0.3554951846599579, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39024725258350373, + "rkd_teacher_entropy": 3.7573428392410277, + "rkd_teacher_similarity_mean": 0.7866397440433502, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19340 + }, + { + "epoch": 5.629909805062555, + "flow_matching_loss": 0.01186087871901691, + "grad_norm": 0.08301686495542526, + "learning_rate": 9.204037474105611e-06, + "loss": 0.0152, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009392812941223383, + "rkd_loss_weight": 0.35516183972358706, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38906869292259216, + "rkd_teacher_entropy": 3.7500463485717774, + "rkd_teacher_similarity_mean": 0.785338532924652, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19350 + }, + { + "epoch": 5.632819319173698, + "flow_matching_loss": 0.012418809998780488, + "grad_norm": 0.08667871356010437, + "learning_rate": 9.188790511427653e-06, + "loss": 0.015, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007307721907272935, + "rkd_loss_weight": 0.35482849180698395, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3838766187429428, + "rkd_teacher_entropy": 3.746929097175598, + "rkd_teacher_similarity_mean": 0.7820014953613281, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19360 + }, + { + "epoch": 5.635728833284841, + "flow_matching_loss": 0.0164956821128726, + "grad_norm": 0.13562582433223724, + "learning_rate": 9.173550609917186e-06, + "loss": 0.0192, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0077240998856723305, + "rkd_loss_weight": 0.3544951528310776, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38052459359169005, + "rkd_teacher_entropy": 3.729495477676392, + "rkd_teacher_similarity_mean": 0.778154706954956, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19370 + }, + { + "epoch": 5.638638347395985, + "flow_matching_loss": 0.013140602968633175, + "grad_norm": 0.0953436866402626, + "learning_rate": 9.158317788092124e-06, + "loss": 0.0162, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008570336643606425, + "rkd_loss_weight": 0.35416180491447447, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3803606301546097, + "rkd_teacher_entropy": 3.734503483772278, + "rkd_teacher_similarity_mean": 0.7759492814540863, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19380 + }, + { + "epoch": 5.641547861507128, + "flow_matching_loss": 0.015168675035238267, + "grad_norm": 0.14919032156467438, + "learning_rate": 9.143092064461783e-06, + "loss": 0.0179, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007853888161480426, + "rkd_loss_weight": 0.35382845997810364, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3861041754484177, + "rkd_teacher_entropy": 3.7481813192367555, + "rkd_teacher_similarity_mean": 0.7837308764457702, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19390 + }, + { + "epoch": 5.644457375618272, + "flow_matching_loss": 0.015338162099942565, + "grad_norm": 0.08924771845340729, + "learning_rate": 9.127873457526855e-06, + "loss": 0.0179, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007279699761420488, + "rkd_loss_weight": 0.35349512100219727, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37859163284301756, + "rkd_teacher_entropy": 3.7222436904907226, + "rkd_teacher_similarity_mean": 0.7756715655326843, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19400 + }, + { + "epoch": 5.647366889729415, + "flow_matching_loss": 0.013261904893442988, + "grad_norm": 0.1063886433839798, + "learning_rate": 9.112661985779382e-06, + "loss": 0.0157, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006945141684263944, + "rkd_loss_weight": 0.3531617671251297, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3893887221813202, + "rkd_teacher_entropy": 3.752264308929443, + "rkd_teacher_similarity_mean": 0.786723917722702, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19410 + }, + { + "epoch": 5.650276403840559, + "flow_matching_loss": 0.014471643604338169, + "grad_norm": 0.15998727083206177, + "learning_rate": 9.09745766770274e-06, + "loss": 0.0174, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008292549289762974, + "rkd_loss_weight": 0.3528284281492233, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3838358461856842, + "rkd_teacher_entropy": 3.739435243606567, + "rkd_teacher_similarity_mean": 0.7813596367835999, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19420 + }, + { + "epoch": 5.653185917951702, + "flow_matching_loss": 0.013037430681288242, + "grad_norm": 0.11210382729768753, + "learning_rate": 9.082260521771601e-06, + "loss": 0.0157, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007462653750553727, + "rkd_loss_weight": 0.35249508619308473, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3861883103847504, + "rkd_teacher_entropy": 3.739874005317688, + "rkd_teacher_similarity_mean": 0.7829312145709991, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19430 + }, + { + "epoch": 5.656095432062846, + "flow_matching_loss": 0.016463536955416202, + "grad_norm": 0.12380987405776978, + "learning_rate": 9.067070566451936e-06, + "loss": 0.0195, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008723744144663214, + "rkd_loss_weight": 0.3521617352962494, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38329939544200897, + "rkd_teacher_entropy": 3.7459678649902344, + "rkd_teacher_similarity_mean": 0.7809336602687835, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19440 + }, + { + "epoch": 5.659004946173989, + "flow_matching_loss": 0.011134470999240875, + "grad_norm": 0.3846559226512909, + "learning_rate": 9.051887820200966e-06, + "loss": 0.0143, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00887125451117754, + "rkd_loss_weight": 0.351828396320343, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38472372889518736, + "rkd_teacher_entropy": 3.740630269050598, + "rkd_teacher_similarity_mean": 0.7819532632827759, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19450 + }, + { + "epoch": 5.661914460285132, + "flow_matching_loss": 0.013502406515181064, + "grad_norm": 0.10619660466909409, + "learning_rate": 9.03671230146717e-06, + "loss": 0.0163, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007931527635082602, + "rkd_loss_weight": 0.3514950484037399, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3786702811717987, + "rkd_teacher_entropy": 3.730018663406372, + "rkd_teacher_similarity_mean": 0.7774804711341858, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19460 + }, + { + "epoch": 5.664823974396276, + "flow_matching_loss": 0.01299853646196425, + "grad_norm": 0.13903072476387024, + "learning_rate": 9.021544028690228e-06, + "loss": 0.0157, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007587975542992354, + "rkd_loss_weight": 0.3511617034673691, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3829739630222321, + "rkd_teacher_entropy": 3.742351698875427, + "rkd_teacher_similarity_mean": 0.7798621237277985, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19470 + }, + { + "epoch": 5.6677334885074195, + "flow_matching_loss": 0.012732130102813244, + "grad_norm": 0.10766018182039261, + "learning_rate": 9.006383020301024e-06, + "loss": 0.0154, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007519184704869985, + "rkd_loss_weight": 0.3508283644914627, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38221116065979005, + "rkd_teacher_entropy": 3.7276865243911743, + "rkd_teacher_similarity_mean": 0.7794120192527771, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19480 + }, + { + "epoch": 5.670643002618562, + "flow_matching_loss": 0.013354691024869681, + "grad_norm": 0.10122286528348923, + "learning_rate": 8.991229294721613e-06, + "loss": 0.0162, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008149747038260102, + "rkd_loss_weight": 0.35049501359462737, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3771376371383667, + "rkd_teacher_entropy": 3.7265792369842528, + "rkd_teacher_similarity_mean": 0.7763448238372803, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19490 + }, + { + "epoch": 5.673552516729706, + "flow_matching_loss": 0.012849786691367627, + "grad_norm": 0.09305652230978012, + "learning_rate": 8.976082870365191e-06, + "loss": 0.016, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009120300970971584, + "rkd_loss_weight": 0.35016167163848877, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3766151309013367, + "rkd_teacher_entropy": 3.729241728782654, + "rkd_teacher_similarity_mean": 0.773969829082489, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19500 + }, + { + "epoch": 5.67646203084085, + "flow_matching_loss": 0.01667993487790227, + "grad_norm": 0.13275711238384247, + "learning_rate": 8.960943765636105e-06, + "loss": 0.0193, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007546270871534944, + "rkd_loss_weight": 0.34982832968235017, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3906465470790863, + "rkd_teacher_entropy": 3.7599359273910524, + "rkd_teacher_similarity_mean": 0.7892180323600769, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19510 + }, + { + "epoch": 5.679371544951993, + "flow_matching_loss": 0.013438183628022671, + "grad_norm": 0.10837037116289139, + "learning_rate": 8.945811998929792e-06, + "loss": 0.0161, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007486040843650699, + "rkd_loss_weight": 0.34949497878551483, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38216531872749326, + "rkd_teacher_entropy": 3.744595241546631, + "rkd_teacher_similarity_mean": 0.7793874144554138, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19520 + }, + { + "epoch": 5.682281059063136, + "flow_matching_loss": 0.016362584568560122, + "grad_norm": 0.12436442077159882, + "learning_rate": 8.93068758863277e-06, + "loss": 0.0199, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010031354473903775, + "rkd_loss_weight": 0.34916163980960846, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3799345701932907, + "rkd_teacher_entropy": 3.7365590572357177, + "rkd_teacher_similarity_mean": 0.7764274716377259, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19530 + }, + { + "epoch": 5.68519057317428, + "flow_matching_loss": 0.013002124894410372, + "grad_norm": 0.12348423898220062, + "learning_rate": 8.915570553122628e-06, + "loss": 0.0161, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008996240887790918, + "rkd_loss_weight": 0.34882829785346986, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38521662056446077, + "rkd_teacher_entropy": 3.7567309856414797, + "rkd_teacher_similarity_mean": 0.782341480255127, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19540 + }, + { + "epoch": 5.688100087285424, + "flow_matching_loss": 0.01460687890648842, + "grad_norm": 0.12978701293468475, + "learning_rate": 8.900460910767982e-06, + "loss": 0.0176, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008634819509461522, + "rkd_loss_weight": 0.3484949469566345, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38302065432071686, + "rkd_teacher_entropy": 3.742527174949646, + "rkd_teacher_similarity_mean": 0.783010071516037, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19550 + }, + { + "epoch": 5.6910096013965665, + "flow_matching_loss": 0.0138827552087605, + "grad_norm": 0.0993485227227211, + "learning_rate": 8.885358679928486e-06, + "loss": 0.0169, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008785452228039503, + "rkd_loss_weight": 0.34816160798072815, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3806841552257538, + "rkd_teacher_entropy": 3.737105917930603, + "rkd_teacher_similarity_mean": 0.7790379703044892, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19560 + }, + { + "epoch": 5.69391911550771, + "flow_matching_loss": 0.014810833800584077, + "grad_norm": 0.10287746042013168, + "learning_rate": 8.870263878954767e-06, + "loss": 0.0177, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008395984675735236, + "rkd_loss_weight": 0.34782826006412504, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.387217116355896, + "rkd_teacher_entropy": 3.7562502145767214, + "rkd_teacher_similarity_mean": 0.7851491153240204, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19570 + }, + { + "epoch": 5.696828629618854, + "flow_matching_loss": 0.013707290031015873, + "grad_norm": 0.11132865399122238, + "learning_rate": 8.855176526188428e-06, + "loss": 0.0162, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007044498017057777, + "rkd_loss_weight": 0.3474949151277542, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37936480939388273, + "rkd_teacher_entropy": 3.735624885559082, + "rkd_teacher_similarity_mean": 0.7779896676540374, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19580 + }, + { + "epoch": 5.699738143729997, + "flow_matching_loss": 0.011611045431345701, + "grad_norm": 0.1031002402305603, + "learning_rate": 8.840096639962033e-06, + "loss": 0.014, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006749139539897442, + "rkd_loss_weight": 0.34716157615184784, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3846980452537537, + "rkd_teacher_entropy": 3.7525333881378176, + "rkd_teacher_similarity_mean": 0.7841349601745605, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19590 + }, + { + "epoch": 5.7026476578411405, + "flow_matching_loss": 0.012843337096273898, + "grad_norm": 0.11478544771671295, + "learning_rate": 8.825024238599056e-06, + "loss": 0.0153, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006977029331028462, + "rkd_loss_weight": 0.3468282252550125, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3851486384868622, + "rkd_teacher_entropy": 3.7515658855438234, + "rkd_teacher_similarity_mean": 0.7837251663208008, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19600 + }, + { + "epoch": 5.705557171952284, + "flow_matching_loss": 0.013307405915111304, + "grad_norm": 0.14032365381717682, + "learning_rate": 8.809959340413891e-06, + "loss": 0.0159, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007511771749705076, + "rkd_loss_weight": 0.3464948832988739, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3774134159088135, + "rkd_teacher_entropy": 3.7151094675064087, + "rkd_teacher_similarity_mean": 0.7745754778385162, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19610 + }, + { + "epoch": 5.708466686063428, + "flow_matching_loss": 0.01315897647291422, + "grad_norm": 0.10041221231222153, + "learning_rate": 8.79490196371181e-06, + "loss": 0.016, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00809963415376842, + "rkd_loss_weight": 0.3461615413427353, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38530039191246035, + "rkd_teacher_entropy": 3.7467652559280396, + "rkd_teacher_similarity_mean": 0.7832872331142425, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19620 + }, + { + "epoch": 5.711376200174571, + "flow_matching_loss": 0.014437284460291267, + "grad_norm": 0.1032831147313118, + "learning_rate": 8.779852126788949e-06, + "loss": 0.0173, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00840882738120854, + "rkd_loss_weight": 0.34582819044589996, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37996619641780854, + "rkd_teacher_entropy": 3.737192749977112, + "rkd_teacher_similarity_mean": 0.7788737297058106, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19630 + }, + { + "epoch": 5.714285714285714, + "flow_matching_loss": 0.013080263137817382, + "grad_norm": 0.11804749816656113, + "learning_rate": 8.764809847932277e-06, + "loss": 0.0165, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00976946959272027, + "rkd_loss_weight": 0.3454948514699936, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3818225085735321, + "rkd_teacher_entropy": 3.729133868217468, + "rkd_teacher_similarity_mean": 0.7796737790107727, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19640 + }, + { + "epoch": 5.717195228396857, + "flow_matching_loss": 0.014248036686331034, + "grad_norm": 0.110244981944561, + "learning_rate": 8.749775145419577e-06, + "loss": 0.0171, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00830284496769309, + "rkd_loss_weight": 0.34516150653362276, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3751851350069046, + "rkd_teacher_entropy": 3.714650821685791, + "rkd_teacher_similarity_mean": 0.7731361091136932, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19650 + }, + { + "epoch": 5.720104742508001, + "flow_matching_loss": 0.015708611439913513, + "grad_norm": 0.10019414126873016, + "learning_rate": 8.734748037519427e-06, + "loss": 0.0189, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009330838825553656, + "rkd_loss_weight": 0.34482815861701965, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37447769939899445, + "rkd_teacher_entropy": 3.7148788928985597, + "rkd_teacher_similarity_mean": 0.7717259407043457, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19660 + }, + { + "epoch": 5.723014256619145, + "flow_matching_loss": 0.01638174904510379, + "grad_norm": 0.11382316797971725, + "learning_rate": 8.719728542491195e-06, + "loss": 0.0191, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007889989903196693, + "rkd_loss_weight": 0.3444948196411133, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3869311988353729, + "rkd_teacher_entropy": 3.7476162195205687, + "rkd_teacher_similarity_mean": 0.7848954916000366, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19670 + }, + { + "epoch": 5.725923770730288, + "flow_matching_loss": 0.013479450903832912, + "grad_norm": 0.10790976881980896, + "learning_rate": 8.704716678584976e-06, + "loss": 0.0164, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008436189638450742, + "rkd_loss_weight": 0.34416146874427794, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3777635425329208, + "rkd_teacher_entropy": 3.7264212369918823, + "rkd_teacher_similarity_mean": 0.7740024209022522, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19680 + }, + { + "epoch": 5.728833284841431, + "flow_matching_loss": 0.013404826633632183, + "grad_norm": 0.13631868362426758, + "learning_rate": 8.689712464041596e-06, + "loss": 0.0164, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008747150748968124, + "rkd_loss_weight": 0.34382812678813934, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37071567475795747, + "rkd_teacher_entropy": 3.6977361917495726, + "rkd_teacher_similarity_mean": 0.7662915050983429, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19690 + }, + { + "epoch": 5.731742798952575, + "flow_matching_loss": 0.01326607894152403, + "grad_norm": 0.09242387115955353, + "learning_rate": 8.674715917092592e-06, + "loss": 0.0157, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007202721619978547, + "rkd_loss_weight": 0.34349478781223297, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38508527874946596, + "rkd_teacher_entropy": 3.741080641746521, + "rkd_teacher_similarity_mean": 0.7814291954040528, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19700 + }, + { + "epoch": 5.734652313063719, + "flow_matching_loss": 0.011567497905343772, + "grad_norm": 0.09056466817855835, + "learning_rate": 8.65972705596018e-06, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008415537979453802, + "rkd_loss_weight": 0.3431614339351654, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3793465793132782, + "rkd_teacher_entropy": 3.733012294769287, + "rkd_teacher_similarity_mean": 0.775648319721222, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19710 + }, + { + "epoch": 5.737561827174861, + "flow_matching_loss": 0.013619311898946763, + "grad_norm": 0.10197701305150986, + "learning_rate": 8.64474589885725e-06, + "loss": 0.0163, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007873890874907374, + "rkd_loss_weight": 0.34282809495925903, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38204345703125, + "rkd_teacher_entropy": 3.7275833606719972, + "rkd_teacher_similarity_mean": 0.7800173163414001, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19720 + }, + { + "epoch": 5.740471341286005, + "flow_matching_loss": 0.013593338895589113, + "grad_norm": 0.10612992197275162, + "learning_rate": 8.629772463987315e-06, + "loss": 0.0162, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007713951077312231, + "rkd_loss_weight": 0.3424947500228882, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38510420024394987, + "rkd_teacher_entropy": 3.743719220161438, + "rkd_teacher_similarity_mean": 0.782535296678543, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19730 + }, + { + "epoch": 5.743380855397149, + "flow_matching_loss": 0.01380244456231594, + "grad_norm": 0.14116838574409485, + "learning_rate": 8.614806769544507e-06, + "loss": 0.0166, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00806769672781229, + "rkd_loss_weight": 0.3421614021062851, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3793699413537979, + "rkd_teacher_entropy": 3.717989182472229, + "rkd_teacher_similarity_mean": 0.7765833139419556, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19740 + }, + { + "epoch": 5.7462903695082925, + "flow_matching_loss": 0.013542938698083162, + "grad_norm": 0.09683000296354294, + "learning_rate": 8.59984883371356e-06, + "loss": 0.0163, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008121533878147602, + "rkd_loss_weight": 0.3418280631303787, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3973685562610626, + "rkd_teacher_entropy": 3.7869679927825928, + "rkd_teacher_similarity_mean": 0.7939172208309173, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19750 + }, + { + "epoch": 5.749199883619435, + "flow_matching_loss": 0.015712739480659366, + "grad_norm": 0.14155898988246918, + "learning_rate": 8.584898674669772e-06, + "loss": 0.0193, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010441062320023776, + "rkd_loss_weight": 0.3414947181940079, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38372411727905276, + "rkd_teacher_entropy": 3.7485784769058226, + "rkd_teacher_similarity_mean": 0.7811584234237671, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19760 + }, + { + "epoch": 5.752109397730579, + "flow_matching_loss": 0.013022738322615624, + "grad_norm": 0.11957558989524841, + "learning_rate": 8.569956310579001e-06, + "loss": 0.0153, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006781922001391649, + "rkd_loss_weight": 0.3411613702774048, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3889711171388626, + "rkd_teacher_entropy": 3.7609837532043455, + "rkd_teacher_similarity_mean": 0.7876717627048493, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19770 + }, + { + "epoch": 5.755018911841723, + "flow_matching_loss": 0.012121230969205499, + "grad_norm": 0.10471464693546295, + "learning_rate": 8.555021759597627e-06, + "loss": 0.0148, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0077777613885700704, + "rkd_loss_weight": 0.3408280313014984, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3757386177778244, + "rkd_teacher_entropy": 3.717937731742859, + "rkd_teacher_similarity_mean": 0.7737741827964782, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19780 + }, + { + "epoch": 5.757928425952866, + "flow_matching_loss": 0.015252547338604927, + "grad_norm": 0.09738875925540924, + "learning_rate": 8.54009503987253e-06, + "loss": 0.0179, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007680649496614933, + "rkd_loss_weight": 0.3404946804046631, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39044795632362367, + "rkd_teacher_entropy": 3.7752005815505982, + "rkd_teacher_similarity_mean": 0.7891905248165131, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19790 + }, + { + "epoch": 5.760837940064009, + "flow_matching_loss": 0.01476107994094491, + "grad_norm": 0.09471528232097626, + "learning_rate": 8.525176169541095e-06, + "loss": 0.0172, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007278221985325217, + "rkd_loss_weight": 0.3401613384485245, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3862131297588348, + "rkd_teacher_entropy": 3.7519704341888427, + "rkd_teacher_similarity_mean": 0.7848526000976562, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19800 + }, + { + "epoch": 5.763747454175153, + "flow_matching_loss": 0.01293002488091588, + "grad_norm": 0.08510476350784302, + "learning_rate": 8.510265166731151e-06, + "loss": 0.0154, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007340373797342181, + "rkd_loss_weight": 0.3398279994726181, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38446577787399294, + "rkd_teacher_entropy": 3.7394585371017457, + "rkd_teacher_similarity_mean": 0.7824827492237091, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19810 + }, + { + "epoch": 5.766656968286296, + "flow_matching_loss": 0.012003493681550025, + "grad_norm": 0.0810059905052185, + "learning_rate": 8.495362049560964e-06, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007593092322349548, + "rkd_loss_weight": 0.33949464559555054, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3838979214429855, + "rkd_teacher_entropy": 3.738375449180603, + "rkd_teacher_similarity_mean": 0.7801841616630554, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19820 + }, + { + "epoch": 5.7695664823974395, + "flow_matching_loss": 0.014849353302270173, + "grad_norm": 0.11130782961845398, + "learning_rate": 8.480466836139241e-06, + "loss": 0.0176, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008060963777825236, + "rkd_loss_weight": 0.33916130661964417, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.379948815703392, + "rkd_teacher_entropy": 3.737638258934021, + "rkd_teacher_similarity_mean": 0.7773240625858306, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19830 + }, + { + "epoch": 5.772475996508583, + "flow_matching_loss": 0.013785842061042785, + "grad_norm": 0.1446077972650528, + "learning_rate": 8.465579544565066e-06, + "loss": 0.0169, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009137603919953108, + "rkd_loss_weight": 0.33882796168327334, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38323813378810884, + "rkd_teacher_entropy": 3.7424068212509156, + "rkd_teacher_similarity_mean": 0.7804829835891723, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19840 + }, + { + "epoch": 5.775385510619726, + "flow_matching_loss": 0.013488675560802221, + "grad_norm": 0.09023953229188919, + "learning_rate": 8.450700192927897e-06, + "loss": 0.0161, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007633145153522492, + "rkd_loss_weight": 0.3384946137666702, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3830310940742493, + "rkd_teacher_entropy": 3.746961998939514, + "rkd_teacher_similarity_mean": 0.780554986000061, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19850 + }, + { + "epoch": 5.77829502473087, + "flow_matching_loss": 0.01424835491925478, + "grad_norm": 0.12603344023227692, + "learning_rate": 8.435828799307551e-06, + "loss": 0.0168, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007505543669685722, + "rkd_loss_weight": 0.33816127479076385, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3842509359121323, + "rkd_teacher_entropy": 3.7380527019500733, + "rkd_teacher_similarity_mean": 0.7813432335853576, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19860 + }, + { + "epoch": 5.7812045388420135, + "flow_matching_loss": 0.013745275977998971, + "grad_norm": 0.0919518768787384, + "learning_rate": 8.420965381774172e-06, + "loss": 0.0162, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007389549631625414, + "rkd_loss_weight": 0.33782792687416074, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38528808653354646, + "rkd_teacher_entropy": 3.747422432899475, + "rkd_teacher_similarity_mean": 0.7818024933338166, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19870 + }, + { + "epoch": 5.784114052953157, + "flow_matching_loss": 0.014212350081652402, + "grad_norm": 0.13052210211753845, + "learning_rate": 8.406109958388222e-06, + "loss": 0.017, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008160969382151962, + "rkd_loss_weight": 0.3374945819377899, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38706876933574674, + "rkd_teacher_entropy": 3.7578572273254394, + "rkd_teacher_similarity_mean": 0.7842554867267608, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19880 + }, + { + "epoch": 5.7870235670643, + "flow_matching_loss": 0.01333358846604824, + "grad_norm": 0.09965483099222183, + "learning_rate": 8.391262547200434e-06, + "loss": 0.0156, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0066603526007384065, + "rkd_loss_weight": 0.33716124296188354, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3857252687215805, + "rkd_teacher_entropy": 3.745685410499573, + "rkd_teacher_similarity_mean": 0.7830296039581299, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19890 + }, + { + "epoch": 5.789933081175444, + "flow_matching_loss": 0.011837691720575094, + "grad_norm": 0.09671928733587265, + "learning_rate": 8.376423166251816e-06, + "loss": 0.0144, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007615684159100055, + "rkd_loss_weight": 0.336827889084816, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38431658446788786, + "rkd_teacher_entropy": 3.747498893737793, + "rkd_teacher_similarity_mean": 0.7822267651557923, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19900 + }, + { + "epoch": 5.792842595286587, + "flow_matching_loss": 0.016244534216821194, + "grad_norm": 0.10232214629650116, + "learning_rate": 8.361591833573612e-06, + "loss": 0.0195, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009750600811094046, + "rkd_loss_weight": 0.3364945501089096, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3801311880350113, + "rkd_teacher_entropy": 3.742525053024292, + "rkd_teacher_similarity_mean": 0.7780135750770569, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19910 + }, + { + "epoch": 5.79575210939773, + "flow_matching_loss": 0.013556306157261134, + "grad_norm": 0.11667943000793457, + "learning_rate": 8.346768567187287e-06, + "loss": 0.0168, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009661378338932991, + "rkd_loss_weight": 0.336161208152771, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3735381573438644, + "rkd_teacher_entropy": 3.710241675376892, + "rkd_teacher_similarity_mean": 0.7702315986156464, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19920 + }, + { + "epoch": 5.798661623508874, + "flow_matching_loss": 0.0181644294410944, + "grad_norm": 0.121161088347435, + "learning_rate": 8.331953385104516e-06, + "loss": 0.0212, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008965113619342446, + "rkd_loss_weight": 0.33582785725593567, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39264667928218844, + "rkd_teacher_entropy": 3.764517903327942, + "rkd_teacher_similarity_mean": 0.7892713069915771, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19930 + }, + { + "epoch": 5.801571137620018, + "flow_matching_loss": 0.012839359929785132, + "grad_norm": 0.09355441480875015, + "learning_rate": 8.317146305327134e-06, + "loss": 0.0155, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007980672316625715, + "rkd_loss_weight": 0.3354945182800293, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38095383644104003, + "rkd_teacher_entropy": 3.744923210144043, + "rkd_teacher_similarity_mean": 0.7773473262786865, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19940 + }, + { + "epoch": 5.804480651731161, + "flow_matching_loss": 0.01388341560959816, + "grad_norm": 0.10084467381238937, + "learning_rate": 8.302347345847146e-06, + "loss": 0.0162, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0069422383792698385, + "rkd_loss_weight": 0.3351611703634262, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37615762948989867, + "rkd_teacher_entropy": 3.712851142883301, + "rkd_teacher_similarity_mean": 0.7733774065971375, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19950 + }, + { + "epoch": 5.807390165842304, + "flow_matching_loss": 0.01690417220816016, + "grad_norm": 0.13595399260520935, + "learning_rate": 8.287556524646683e-06, + "loss": 0.02, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009127618232741951, + "rkd_loss_weight": 0.33482782542705536, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38082205057144164, + "rkd_teacher_entropy": 3.7316985607147215, + "rkd_teacher_similarity_mean": 0.7782938003540039, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19960 + }, + { + "epoch": 5.810299679953448, + "flow_matching_loss": 0.011667487118393183, + "grad_norm": 0.11781518161296844, + "learning_rate": 8.272773859697978e-06, + "loss": 0.0143, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0077731927623972295, + "rkd_loss_weight": 0.334494486451149, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39291303157806395, + "rkd_teacher_entropy": 3.7661935329437255, + "rkd_teacher_similarity_mean": 0.7909363746643067, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19970 + }, + { + "epoch": 5.813209194064592, + "flow_matching_loss": 0.013130722288042306, + "grad_norm": 0.09992530196905136, + "learning_rate": 8.257999368963374e-06, + "loss": 0.0164, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009829345997422934, + "rkd_loss_weight": 0.3341611385345459, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37376088500022886, + "rkd_teacher_entropy": 3.716088342666626, + "rkd_teacher_similarity_mean": 0.7709300577640533, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19980 + }, + { + "epoch": 5.816118708175734, + "flow_matching_loss": 0.012047610757872463, + "grad_norm": 0.11075232923030853, + "learning_rate": 8.243233070395266e-06, + "loss": 0.0152, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009390482073649764, + "rkd_loss_weight": 0.33382779359817505, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3810868948698044, + "rkd_teacher_entropy": 3.733223390579224, + "rkd_teacher_similarity_mean": 0.7775182366371155, + "rkd_vlm_temp": 0.10000000149011612, + "step": 19990 + }, + { + "epoch": 5.819028222286878, + "flow_matching_loss": 0.013843486551195383, + "grad_norm": 0.10559090226888657, + "learning_rate": 8.2284749819361e-06, + "loss": 0.0165, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00794531824067235, + "rkd_loss_weight": 0.3334944546222687, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38195971250534055, + "rkd_teacher_entropy": 3.7350760459899903, + "rkd_teacher_similarity_mean": 0.7785433053970336, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20000 + }, + { + "epoch": 5.821937736398022, + "flow_matching_loss": 0.01147370357066393, + "grad_norm": 0.1107826679944992, + "learning_rate": 8.213725121518346e-06, + "loss": 0.0142, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008205021405592561, + "rkd_loss_weight": 0.3331611007452011, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3900675743818283, + "rkd_teacher_entropy": 3.7673099040985107, + "rkd_teacher_similarity_mean": 0.7884146749973298, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20010 + }, + { + "epoch": 5.824847250509165, + "flow_matching_loss": 0.01362992124632001, + "grad_norm": 0.1424456536769867, + "learning_rate": 8.198983507064469e-06, + "loss": 0.0164, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008398503763601184, + "rkd_loss_weight": 0.33282776176929474, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3879469335079193, + "rkd_teacher_entropy": 3.764182472229004, + "rkd_teacher_similarity_mean": 0.7860942661762238, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20020 + }, + { + "epoch": 5.827756764620308, + "flow_matching_loss": 0.014386871922761202, + "grad_norm": 0.09240309149026871, + "learning_rate": 8.18425015648692e-06, + "loss": 0.0168, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007390068750828505, + "rkd_loss_weight": 0.33249441981315614, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38588736951351166, + "rkd_teacher_entropy": 3.746461534500122, + "rkd_teacher_similarity_mean": 0.782226836681366, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20030 + }, + { + "epoch": 5.830666278731452, + "flow_matching_loss": 0.014206555765122175, + "grad_norm": 0.1044064611196518, + "learning_rate": 8.169525087688122e-06, + "loss": 0.0167, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007657087175175548, + "rkd_loss_weight": 0.3321610689163208, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39298332929611207, + "rkd_teacher_entropy": 3.772724747657776, + "rkd_teacher_similarity_mean": 0.7908666133880615, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20040 + }, + { + "epoch": 5.833575792842595, + "flow_matching_loss": 0.0111418966203928, + "grad_norm": 0.10422074794769287, + "learning_rate": 8.154808318560413e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007426170259714126, + "rkd_loss_weight": 0.33182772994041443, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38905459046363833, + "rkd_teacher_entropy": 3.7593995571136474, + "rkd_teacher_similarity_mean": 0.7876255095005036, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20050 + }, + { + "epoch": 5.836485306953739, + "flow_matching_loss": 0.012317559821531177, + "grad_norm": 0.08649667352437973, + "learning_rate": 8.140099866986058e-06, + "loss": 0.0148, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007559852581471205, + "rkd_loss_weight": 0.3314943820238113, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37864546477794647, + "rkd_teacher_entropy": 3.7255903244018556, + "rkd_teacher_similarity_mean": 0.7757059335708618, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20060 + }, + { + "epoch": 5.839394821064882, + "flow_matching_loss": 0.013832767214626073, + "grad_norm": 0.11929824948310852, + "learning_rate": 8.125399750837202e-06, + "loss": 0.0162, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007094353856518865, + "rkd_loss_weight": 0.3311610370874405, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.387790185213089, + "rkd_teacher_entropy": 3.758297300338745, + "rkd_teacher_similarity_mean": 0.786455899477005, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20070 + }, + { + "epoch": 5.842304335176026, + "flow_matching_loss": 0.013926652539521456, + "grad_norm": 0.1323174238204956, + "learning_rate": 8.110707987975888e-06, + "loss": 0.0166, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008039374579675495, + "rkd_loss_weight": 0.3308276981115341, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3858775645494461, + "rkd_teacher_entropy": 3.7538610696792603, + "rkd_teacher_similarity_mean": 0.7843511819839477, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20080 + }, + { + "epoch": 5.845213849287169, + "flow_matching_loss": 0.012044458417221904, + "grad_norm": 0.1271970421075821, + "learning_rate": 8.096024596253982e-06, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0076944510452449325, + "rkd_loss_weight": 0.3304943472146988, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.380757674574852, + "rkd_teacher_entropy": 3.74743537902832, + "rkd_teacher_similarity_mean": 0.7806860387325287, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20090 + }, + { + "epoch": 5.8481233633983125, + "flow_matching_loss": 0.012491018697619437, + "grad_norm": 0.09321044385433197, + "learning_rate": 8.081349593513197e-06, + "loss": 0.015, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007457330450415611, + "rkd_loss_weight": 0.3301610052585602, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3811021834611893, + "rkd_teacher_entropy": 3.738813304901123, + "rkd_teacher_similarity_mean": 0.7811416029930115, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20100 + }, + { + "epoch": 5.851032877509456, + "flow_matching_loss": 0.014641230832785368, + "grad_norm": 0.10965746641159058, + "learning_rate": 8.066682997585045e-06, + "loss": 0.0183, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011090755043551326, + "rkd_loss_weight": 0.3298276633024216, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3862557739019394, + "rkd_teacher_entropy": 3.7495600461959837, + "rkd_teacher_similarity_mean": 0.7842636168003082, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20110 + }, + { + "epoch": 5.853942391620599, + "flow_matching_loss": 0.013644531928002834, + "grad_norm": 0.12176405638456345, + "learning_rate": 8.052024826290819e-06, + "loss": 0.0165, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008632116997614504, + "rkd_loss_weight": 0.32949431240558624, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3862050801515579, + "rkd_teacher_entropy": 3.756082797050476, + "rkd_teacher_similarity_mean": 0.7856173872947693, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20120 + }, + { + "epoch": 5.856851905731743, + "flow_matching_loss": 0.013498800247907639, + "grad_norm": 0.11775030195713043, + "learning_rate": 8.037375097441574e-06, + "loss": 0.016, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007678472390398383, + "rkd_loss_weight": 0.32916097342967987, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38009604513645173, + "rkd_teacher_entropy": 3.7307750940322877, + "rkd_teacher_similarity_mean": 0.7791218101978302, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20130 + }, + { + "epoch": 5.8597614198428865, + "flow_matching_loss": 0.012965480517596006, + "grad_norm": 0.1264566034078598, + "learning_rate": 8.022733828838124e-06, + "loss": 0.0156, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007875895453616976, + "rkd_loss_weight": 0.32882762849330904, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38177129328250886, + "rkd_teacher_entropy": 3.7471258640289307, + "rkd_teacher_similarity_mean": 0.7807296931743621, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20140 + }, + { + "epoch": 5.86267093395403, + "flow_matching_loss": 0.013980075903236865, + "grad_norm": 0.08422046899795532, + "learning_rate": 8.008101038270988e-06, + "loss": 0.0171, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00960221686400473, + "rkd_loss_weight": 0.32849428057670593, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3870015531778336, + "rkd_teacher_entropy": 3.7698755264282227, + "rkd_teacher_similarity_mean": 0.7865849733352661, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20150 + }, + { + "epoch": 5.865580448065173, + "flow_matching_loss": 0.014491154346615076, + "grad_norm": 0.09549972414970398, + "learning_rate": 7.99347674352038e-06, + "loss": 0.0171, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007927181990817189, + "rkd_loss_weight": 0.32816094160079956, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3833238810300827, + "rkd_teacher_entropy": 3.744373607635498, + "rkd_teacher_similarity_mean": 0.7831584870815277, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20160 + }, + { + "epoch": 5.868489962176317, + "flow_matching_loss": 0.014817127026617527, + "grad_norm": 0.11915026605129242, + "learning_rate": 7.978860962356202e-06, + "loss": 0.0175, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008188080321997405, + "rkd_loss_weight": 0.3278275907039642, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3835840284824371, + "rkd_teacher_entropy": 3.749585199356079, + "rkd_teacher_similarity_mean": 0.782893979549408, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20170 + }, + { + "epoch": 5.8713994762874595, + "flow_matching_loss": 0.012283265590667725, + "grad_norm": 0.10837852209806442, + "learning_rate": 7.964253712537997e-06, + "loss": 0.0148, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0076436650473624464, + "rkd_loss_weight": 0.3274942487478256, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3909968614578247, + "rkd_teacher_entropy": 3.7706077337265014, + "rkd_teacher_similarity_mean": 0.7914880454540253, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20180 + }, + { + "epoch": 5.874308990398603, + "flow_matching_loss": 0.013721537962555885, + "grad_norm": 0.10481642931699753, + "learning_rate": 7.949655011814968e-06, + "loss": 0.0166, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008854038501158356, + "rkd_loss_weight": 0.32716090977191925, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38568430244922636, + "rkd_teacher_entropy": 3.7661895751953125, + "rkd_teacher_similarity_mean": 0.7855430662631988, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20190 + }, + { + "epoch": 5.877218504509747, + "flow_matching_loss": 0.012210262566804886, + "grad_norm": 0.09629273414611816, + "learning_rate": 7.9350648779259e-06, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007294376892969012, + "rkd_loss_weight": 0.3268275588750839, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3833862543106079, + "rkd_teacher_entropy": 3.7450302124023436, + "rkd_teacher_similarity_mean": 0.7820578455924988, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20200 + }, + { + "epoch": 5.880128018620891, + "flow_matching_loss": 0.01613626955077052, + "grad_norm": 0.10968171805143356, + "learning_rate": 7.92048332859919e-06, + "loss": 0.019, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008751895651221276, + "rkd_loss_weight": 0.3264942169189453, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3806027382612228, + "rkd_teacher_entropy": 3.7349985361099245, + "rkd_teacher_similarity_mean": 0.7796731889247894, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20210 + }, + { + "epoch": 5.8830375327320334, + "flow_matching_loss": 0.015316270664334298, + "grad_norm": 0.14019419252872467, + "learning_rate": 7.90591038155278e-06, + "loss": 0.0181, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008598554413765669, + "rkd_loss_weight": 0.3261608749628067, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.386236360669136, + "rkd_teacher_entropy": 3.7510588645935057, + "rkd_teacher_similarity_mean": 0.7841537773609162, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20220 + }, + { + "epoch": 5.885947046843177, + "flow_matching_loss": 0.014593241363763809, + "grad_norm": 0.13330772519111633, + "learning_rate": 7.891346054494194e-06, + "loss": 0.017, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007495972141623497, + "rkd_loss_weight": 0.3258275240659714, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3794673174619675, + "rkd_teacher_entropy": 3.7329254865646364, + "rkd_teacher_similarity_mean": 0.7779119133949279, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20230 + }, + { + "epoch": 5.888856560954321, + "flow_matching_loss": 0.014092010166496038, + "grad_norm": 0.1290627270936966, + "learning_rate": 7.876790365120447e-06, + "loss": 0.0163, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006808806629851461, + "rkd_loss_weight": 0.325494185090065, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3868826121091843, + "rkd_teacher_entropy": 3.7437666177749636, + "rkd_teacher_similarity_mean": 0.7857817769050598, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20240 + }, + { + "epoch": 5.891766075065464, + "flow_matching_loss": 0.01281377924606204, + "grad_norm": 0.09940583258867264, + "learning_rate": 7.862243331118088e-06, + "loss": 0.0152, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007431898033246398, + "rkd_loss_weight": 0.3251608401536942, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38268173635005953, + "rkd_teacher_entropy": 3.740181255340576, + "rkd_teacher_similarity_mean": 0.7811793804168701, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20250 + }, + { + "epoch": 5.894675589176607, + "flow_matching_loss": 0.013594514410942792, + "grad_norm": 0.11332206428050995, + "learning_rate": 7.847704970163127e-06, + "loss": 0.016, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007457792991772294, + "rkd_loss_weight": 0.32482749223709106, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37388226985931394, + "rkd_teacher_entropy": 3.715754580497742, + "rkd_teacher_similarity_mean": 0.7718526124954224, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20260 + }, + { + "epoch": 5.897585103287751, + "flow_matching_loss": 0.013639661390334368, + "grad_norm": 0.10229828953742981, + "learning_rate": 7.833175299921046e-06, + "loss": 0.0166, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009118820680305362, + "rkd_loss_weight": 0.3244941532611847, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38950142860412595, + "rkd_teacher_entropy": 3.756943702697754, + "rkd_teacher_similarity_mean": 0.7869029521942139, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20270 + }, + { + "epoch": 5.900494617398895, + "flow_matching_loss": 0.0130123739130795, + "grad_norm": 0.09448205679655075, + "learning_rate": 7.81865433804676e-06, + "loss": 0.0157, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008270405093207955, + "rkd_loss_weight": 0.32416080236434935, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39244953989982606, + "rkd_teacher_entropy": 3.77154483795166, + "rkd_teacher_similarity_mean": 0.7907474160194397, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20280 + }, + { + "epoch": 5.903404131510038, + "flow_matching_loss": 0.013515841495245696, + "grad_norm": 0.10021449625492096, + "learning_rate": 7.80414210218461e-06, + "loss": 0.0162, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00825424208305776, + "rkd_loss_weight": 0.32382746040821075, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39500330090522767, + "rkd_teacher_entropy": 3.777628016471863, + "rkd_teacher_similarity_mean": 0.7931114196777344, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20290 + }, + { + "epoch": 5.906313645621181, + "flow_matching_loss": 0.014785094372928142, + "grad_norm": 0.0815265104174614, + "learning_rate": 7.789638609968332e-06, + "loss": 0.0179, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009477605810388923, + "rkd_loss_weight": 0.3234941214323044, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3842967003583908, + "rkd_teacher_entropy": 3.7480144023895265, + "rkd_teacher_similarity_mean": 0.7830333888530732, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20300 + }, + { + "epoch": 5.909223159732325, + "flow_matching_loss": 0.013853608909994364, + "grad_norm": 0.12662968039512634, + "learning_rate": 7.775143879021037e-06, + "loss": 0.0164, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007838240871205927, + "rkd_loss_weight": 0.3231607675552368, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37881200313568114, + "rkd_teacher_entropy": 3.7318273544311524, + "rkd_teacher_similarity_mean": 0.7769440233707428, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20310 + }, + { + "epoch": 5.912132673843468, + "flow_matching_loss": 0.011920716241002082, + "grad_norm": 0.0999918058514595, + "learning_rate": 7.760657926955187e-06, + "loss": 0.0142, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0071638910565525295, + "rkd_loss_weight": 0.32282742857933044, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3855977714061737, + "rkd_teacher_entropy": 3.7652007579803466, + "rkd_teacher_similarity_mean": 0.7833076238632202, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20320 + }, + { + "epoch": 5.915042187954612, + "flow_matching_loss": 0.012537400145083666, + "grad_norm": 0.10322363674640656, + "learning_rate": 7.746180771372584e-06, + "loss": 0.0149, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007416965160518885, + "rkd_loss_weight": 0.3224940836429596, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3888218730688095, + "rkd_teacher_entropy": 3.7508896827697753, + "rkd_teacher_similarity_mean": 0.7869063675403595, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20330 + }, + { + "epoch": 5.917951702065755, + "flow_matching_loss": 0.013842113688588143, + "grad_norm": 0.09711740911006927, + "learning_rate": 7.731712429864325e-06, + "loss": 0.0164, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007962431013584136, + "rkd_loss_weight": 0.3221607357263565, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3817759156227112, + "rkd_teacher_entropy": 3.741001510620117, + "rkd_teacher_similarity_mean": 0.7797507822513581, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20340 + }, + { + "epoch": 5.920861216176898, + "flow_matching_loss": 0.014104951824992894, + "grad_norm": 0.10309296101331711, + "learning_rate": 7.717252920010828e-06, + "loss": 0.0168, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008486546389758588, + "rkd_loss_weight": 0.32182739675045013, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38420974612236025, + "rkd_teacher_entropy": 3.749264287948608, + "rkd_teacher_similarity_mean": 0.7821714818477631, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20350 + }, + { + "epoch": 5.923770730288042, + "flow_matching_loss": 0.013206715695559978, + "grad_norm": 0.0958198606967926, + "learning_rate": 7.702802259381742e-06, + "loss": 0.0154, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006803092220798135, + "rkd_loss_weight": 0.321494048833847, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39437201917171477, + "rkd_teacher_entropy": 3.771923637390137, + "rkd_teacher_similarity_mean": 0.792813116312027, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20360 + }, + { + "epoch": 5.9266802443991855, + "flow_matching_loss": 0.013917553052306176, + "grad_norm": 0.10475688427686691, + "learning_rate": 7.688360465535999e-06, + "loss": 0.0162, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007111676363274455, + "rkd_loss_weight": 0.3211607038974762, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38375788927078247, + "rkd_teacher_entropy": 3.7422229051589966, + "rkd_teacher_similarity_mean": 0.7813681840896607, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20370 + }, + { + "epoch": 5.929589758510328, + "flow_matching_loss": 0.017149273492395877, + "grad_norm": 0.09826965630054474, + "learning_rate": 7.673927556021734e-06, + "loss": 0.02, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008836914505809545, + "rkd_loss_weight": 0.3208273649215698, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3862613648176193, + "rkd_teacher_entropy": 3.750595784187317, + "rkd_teacher_similarity_mean": 0.7849772334098816, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20380 + }, + { + "epoch": 5.932499272621472, + "flow_matching_loss": 0.011946355272084475, + "grad_norm": 0.10462342947721481, + "learning_rate": 7.659503548376284e-06, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007991840736940503, + "rkd_loss_weight": 0.32049401104450226, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38757044076919556, + "rkd_teacher_entropy": 3.747017192840576, + "rkd_teacher_similarity_mean": 0.7843822419643403, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20390 + }, + { + "epoch": 5.935408786732616, + "flow_matching_loss": 0.014319493528455495, + "grad_norm": 0.10393837839365005, + "learning_rate": 7.645088460126193e-06, + "loss": 0.0168, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007788811950013042, + "rkd_loss_weight": 0.3201606720685959, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38195734322071073, + "rkd_teacher_entropy": 3.729153299331665, + "rkd_teacher_similarity_mean": 0.7796867489814758, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20400 + }, + { + "epoch": 5.9383183008437594, + "flow_matching_loss": 0.011724268179386854, + "grad_norm": 0.10544048994779587, + "learning_rate": 7.630682308787145e-06, + "loss": 0.0142, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007642602827399969, + "rkd_loss_weight": 0.3198273330926895, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.379104220867157, + "rkd_teacher_entropy": 3.7272726774215696, + "rkd_teacher_similarity_mean": 0.7777094364166259, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20410 + }, + { + "epoch": 5.941227814954902, + "flow_matching_loss": 0.012727691140025855, + "grad_norm": 0.1444564014673233, + "learning_rate": 7.6162851118639754e-06, + "loss": 0.0151, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007369440421462059, + "rkd_loss_weight": 0.31949397921562195, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3743880450725555, + "rkd_teacher_entropy": 3.7122196197509765, + "rkd_teacher_similarity_mean": 0.7715960919857026, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20420 + }, + { + "epoch": 5.944137329066046, + "flow_matching_loss": 0.011489389557391405, + "grad_norm": 0.12207003682851791, + "learning_rate": 7.601896886850632e-06, + "loss": 0.0137, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007054987968876958, + "rkd_loss_weight": 0.3191606402397156, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38608144223690033, + "rkd_teacher_entropy": 3.741344690322876, + "rkd_teacher_similarity_mean": 0.78350830078125, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20430 + }, + { + "epoch": 5.94704684317719, + "flow_matching_loss": 0.013687317445874213, + "grad_norm": 0.10180289298295975, + "learning_rate": 7.587517651230168e-06, + "loss": 0.0162, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007818809617310763, + "rkd_loss_weight": 0.31882729530334475, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3852980673313141, + "rkd_teacher_entropy": 3.745560073852539, + "rkd_teacher_similarity_mean": 0.7832649230957032, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20440 + }, + { + "epoch": 5.9499563572883325, + "flow_matching_loss": 0.013367178663611413, + "grad_norm": 0.10019715875387192, + "learning_rate": 7.573147422474702e-06, + "loss": 0.0161, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008657039795070886, + "rkd_loss_weight": 0.31849394738674164, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3893663138151169, + "rkd_teacher_entropy": 3.7647923946380617, + "rkd_teacher_similarity_mean": 0.7873373925685883, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20450 + }, + { + "epoch": 5.952865871399476, + "flow_matching_loss": 0.012027015909552574, + "grad_norm": 0.0965118557214737, + "learning_rate": 7.55878621804543e-06, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008026785729452968, + "rkd_loss_weight": 0.31816060841083527, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3832399815320969, + "rkd_teacher_entropy": 3.7386399030685427, + "rkd_teacher_similarity_mean": 0.7812838733196259, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20460 + }, + { + "epoch": 5.95577538551062, + "flow_matching_loss": 0.012648438382893801, + "grad_norm": 0.1050739735364914, + "learning_rate": 7.544434055392564e-06, + "loss": 0.0156, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009280629409477114, + "rkd_loss_weight": 0.31782726049423216, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3780208259820938, + "rkd_teacher_entropy": 3.719313311576843, + "rkd_teacher_similarity_mean": 0.7746911585330963, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20470 + }, + { + "epoch": 5.958684899621764, + "flow_matching_loss": 0.01677647102624178, + "grad_norm": 0.11648432165384293, + "learning_rate": 7.530090951955334e-06, + "loss": 0.0193, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007879411149770022, + "rkd_loss_weight": 0.31749391555786133, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38076406717300415, + "rkd_teacher_entropy": 3.7395266771316527, + "rkd_teacher_similarity_mean": 0.7788000583648682, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20480 + }, + { + "epoch": 5.961594413732906, + "flow_matching_loss": 0.013621110375970602, + "grad_norm": 0.11927922815084457, + "learning_rate": 7.515756925161959e-06, + "loss": 0.0167, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009619313525035977, + "rkd_loss_weight": 0.31716057658195496, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37761918008327483, + "rkd_teacher_entropy": 3.725801157951355, + "rkd_teacher_similarity_mean": 0.775311428308487, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20490 + }, + { + "epoch": 5.96450392784405, + "flow_matching_loss": 0.01125129689462483, + "grad_norm": 0.08567246049642563, + "learning_rate": 7.501431992429643e-06, + "loss": 0.0134, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006903417780995369, + "rkd_loss_weight": 0.3168272227048874, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3876025527715683, + "rkd_teacher_entropy": 3.75613329410553, + "rkd_teacher_similarity_mean": 0.7852023601531982, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20500 + }, + { + "epoch": 5.967413441955194, + "flow_matching_loss": 0.013874069135636092, + "grad_norm": 0.13230131566524506, + "learning_rate": 7.487116171164519e-06, + "loss": 0.0167, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008822679053992033, + "rkd_loss_weight": 0.316493883728981, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3878200322389603, + "rkd_teacher_entropy": 3.744412899017334, + "rkd_teacher_similarity_mean": 0.7854157030582428, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20510 + }, + { + "epoch": 5.970322956066337, + "flow_matching_loss": 0.011626134160906076, + "grad_norm": 0.09769465774297714, + "learning_rate": 7.472809478761672e-06, + "loss": 0.014, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0075474671553820375, + "rkd_loss_weight": 0.3161605417728424, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3814563602209091, + "rkd_teacher_entropy": 3.7218593835830687, + "rkd_teacher_similarity_mean": 0.7788723886013031, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20520 + }, + { + "epoch": 5.97323247017748, + "flow_matching_loss": 0.013302081264555454, + "grad_norm": 0.1204606294631958, + "learning_rate": 7.4585119326050775e-06, + "loss": 0.0166, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010584429511800409, + "rkd_loss_weight": 0.3158271908760071, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38052864372730255, + "rkd_teacher_entropy": 3.7322181463241577, + "rkd_teacher_similarity_mean": 0.7785805225372314, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20530 + }, + { + "epoch": 5.976141984288624, + "flow_matching_loss": 0.01525504169985652, + "grad_norm": 0.11755624413490295, + "learning_rate": 7.444223550067599e-06, + "loss": 0.0179, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008225462678819895, + "rkd_loss_weight": 0.3154938519001007, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3871865779161453, + "rkd_teacher_entropy": 3.7572670459747313, + "rkd_teacher_similarity_mean": 0.7860036015510559, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20540 + }, + { + "epoch": 5.979051498399767, + "flow_matching_loss": 0.012902149045839906, + "grad_norm": 0.11305952817201614, + "learning_rate": 7.429944348510964e-06, + "loss": 0.0154, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007826633052900434, + "rkd_loss_weight": 0.3151605039834976, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38347187936306, + "rkd_teacher_entropy": 3.745866298675537, + "rkd_teacher_similarity_mean": 0.7839801192283631, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20550 + }, + { + "epoch": 5.981961012510911, + "flow_matching_loss": 0.01218975828960538, + "grad_norm": 0.10787167400121689, + "learning_rate": 7.415674345285763e-06, + "loss": 0.0156, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010767971770837903, + "rkd_loss_weight": 0.31482715904712677, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38388667106628416, + "rkd_teacher_entropy": 3.7439910411834716, + "rkd_teacher_similarity_mean": 0.7825083613395691, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20560 + }, + { + "epoch": 5.984870526622054, + "flow_matching_loss": 0.01455173809081316, + "grad_norm": 0.16070443391799927, + "learning_rate": 7.401413557731387e-06, + "loss": 0.0172, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008455066476017237, + "rkd_loss_weight": 0.3144938200712204, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3901717871427536, + "rkd_teacher_entropy": 3.761036252975464, + "rkd_teacher_similarity_mean": 0.7894716680049896, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20570 + }, + { + "epoch": 5.987780040733197, + "flow_matching_loss": 0.013972824439406395, + "grad_norm": 0.14997714757919312, + "learning_rate": 7.387162003176037e-06, + "loss": 0.0168, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008868581941351294, + "rkd_loss_weight": 0.31416046917438506, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38114420771598817, + "rkd_teacher_entropy": 3.7328904628753663, + "rkd_teacher_similarity_mean": 0.7789261996746063, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20580 + }, + { + "epoch": 5.990689554844341, + "flow_matching_loss": 0.012964910687878728, + "grad_norm": 0.09450636804103851, + "learning_rate": 7.372919698936697e-06, + "loss": 0.0153, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007497245213016868, + "rkd_loss_weight": 0.31382712721824646, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3850893348455429, + "rkd_teacher_entropy": 3.74357328414917, + "rkd_teacher_similarity_mean": 0.7835992574691772, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20590 + }, + { + "epoch": 5.9935990689554846, + "flow_matching_loss": 0.013662948552519084, + "grad_norm": 0.08886890858411789, + "learning_rate": 7.358686662319104e-06, + "loss": 0.0161, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007931938441470266, + "rkd_loss_weight": 0.31349378526210786, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.382804799079895, + "rkd_teacher_entropy": 3.7446666717529298, + "rkd_teacher_similarity_mean": 0.7825183928012848, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20600 + }, + { + "epoch": 5.996508583066628, + "flow_matching_loss": 0.013925186824053526, + "grad_norm": 0.10125941038131714, + "learning_rate": 7.344462910617752e-06, + "loss": 0.017, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00978501937352121, + "rkd_loss_weight": 0.3131604343652725, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3765764534473419, + "rkd_teacher_entropy": 3.7292219400405884, + "rkd_teacher_similarity_mean": 0.775879293680191, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20610 + }, + { + "epoch": 5.999418097177771, + "flow_matching_loss": 0.012030711397528648, + "grad_norm": 0.1153179258108139, + "learning_rate": 7.330248461115832e-06, + "loss": 0.0144, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007514382246881723, + "rkd_loss_weight": 0.31282709538936615, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3773365169763565, + "rkd_teacher_entropy": 3.7223270893096925, + "rkd_teacher_similarity_mean": 0.777654379606247, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20620 + }, + { + "epoch": 6.002327611288915, + "flow_matching_loss": 0.015048464015126228, + "grad_norm": 0.11984255909919739, + "learning_rate": 7.316043331085242e-06, + "loss": 0.0172, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0069484028965234755, + "rkd_loss_weight": 0.31249375343322755, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3815376490354538, + "rkd_teacher_entropy": 3.7405873775482177, + "rkd_teacher_similarity_mean": 0.7808752238750458, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20630 + }, + { + "epoch": 6.0052371254000585, + "flow_matching_loss": 0.012891081580892205, + "grad_norm": 0.0804128646850586, + "learning_rate": 7.30184753778655e-06, + "loss": 0.0152, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0072722805198282, + "rkd_loss_weight": 0.3121604025363922, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37801794409751893, + "rkd_teacher_entropy": 3.7253939390182493, + "rkd_teacher_similarity_mean": 0.7766592562198639, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20640 + }, + { + "epoch": 6.008146639511201, + "flow_matching_loss": 0.012538491003215312, + "grad_norm": 0.10987523198127747, + "learning_rate": 7.2876610984689935e-06, + "loss": 0.015, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0077995934523642065, + "rkd_loss_weight": 0.31182706356048584, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3888234615325928, + "rkd_teacher_entropy": 3.7632940769195558, + "rkd_teacher_similarity_mean": 0.7884720981121063, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20650 + }, + { + "epoch": 6.011056153622345, + "flow_matching_loss": 0.013908687140792608, + "grad_norm": 0.13753792643547058, + "learning_rate": 7.273484030370423e-06, + "loss": 0.017, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009897637972608208, + "rkd_loss_weight": 0.31149371564388273, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3822051525115967, + "rkd_teacher_entropy": 3.7508787393569945, + "rkd_teacher_similarity_mean": 0.781114149093628, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20660 + }, + { + "epoch": 6.013965667733489, + "flow_matching_loss": 0.010911365179345012, + "grad_norm": 0.09948351979255676, + "learning_rate": 7.259316350717324e-06, + "loss": 0.0135, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008410366531461478, + "rkd_loss_weight": 0.3111603707075119, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3815594553947449, + "rkd_teacher_entropy": 3.7362868785858154, + "rkd_teacher_similarity_mean": 0.7797999858856202, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20670 + }, + { + "epoch": 6.0168751818446315, + "flow_matching_loss": 0.014510389510542154, + "grad_norm": 0.10436635464429855, + "learning_rate": 7.245158076724757e-06, + "loss": 0.0171, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008326612785458565, + "rkd_loss_weight": 0.31082703173160553, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3768758326768875, + "rkd_teacher_entropy": 3.718636417388916, + "rkd_teacher_similarity_mean": 0.7750454843044281, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20680 + }, + { + "epoch": 6.019784695955775, + "flow_matching_loss": 0.014004976768046618, + "grad_norm": 0.10263691842556, + "learning_rate": 7.23100922559636e-06, + "loss": 0.016, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0065515111200511456, + "rkd_loss_weight": 0.3104936808347702, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3809226661920547, + "rkd_teacher_entropy": 3.7450552225112914, + "rkd_teacher_similarity_mean": 0.7805846154689788, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20690 + }, + { + "epoch": 6.022694210066919, + "flow_matching_loss": 0.014674247521907092, + "grad_norm": 0.08210599422454834, + "learning_rate": 7.216869814524323e-06, + "loss": 0.0175, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009186919778585434, + "rkd_loss_weight": 0.3101603388786316, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3812595307826996, + "rkd_teacher_entropy": 3.7468892335891724, + "rkd_teacher_similarity_mean": 0.7799776017665863, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20700 + }, + { + "epoch": 6.025603724178063, + "flow_matching_loss": 0.013050070032477379, + "grad_norm": 0.1172787994146347, + "learning_rate": 7.202739860689357e-06, + "loss": 0.0156, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008354520332068205, + "rkd_loss_weight": 0.309826996922493, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3854564934968948, + "rkd_teacher_entropy": 3.75559139251709, + "rkd_teacher_similarity_mean": 0.7836867511272431, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20710 + }, + { + "epoch": 6.0285132382892055, + "flow_matching_loss": 0.015308145992457866, + "grad_norm": 0.1264730840921402, + "learning_rate": 7.188619381260698e-06, + "loss": 0.0179, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008290981501340866, + "rkd_loss_weight": 0.30949364602565765, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3790269732475281, + "rkd_teacher_entropy": 3.7256844282150268, + "rkd_teacher_similarity_mean": 0.7762627243995667, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20720 + }, + { + "epoch": 6.031422752400349, + "flow_matching_loss": 0.013857660815119743, + "grad_norm": 0.0914141982793808, + "learning_rate": 7.174508393396057e-06, + "loss": 0.0159, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006502507114782929, + "rkd_loss_weight": 0.3091603070497513, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3868199348449707, + "rkd_teacher_entropy": 3.747248315811157, + "rkd_teacher_similarity_mean": 0.7850839674472809, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20730 + }, + { + "epoch": 6.034332266511493, + "flow_matching_loss": 0.011828565737232566, + "grad_norm": 0.09551240503787994, + "learning_rate": 7.160406914241615e-06, + "loss": 0.014, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007096123229712248, + "rkd_loss_weight": 0.30882696211338045, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38391115367412565, + "rkd_teacher_entropy": 3.7386985778808595, + "rkd_teacher_similarity_mean": 0.7829119503498078, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20740 + }, + { + "epoch": 6.037241780622636, + "flow_matching_loss": 0.012784900283440948, + "grad_norm": 0.10715952515602112, + "learning_rate": 7.146314960931996e-06, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.005924890749156475, + "rkd_loss_weight": 0.30849361419677734, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38357536792755126, + "rkd_teacher_entropy": 3.744080352783203, + "rkd_teacher_similarity_mean": 0.7838508248329162, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20750 + }, + { + "epoch": 6.040151294733779, + "flow_matching_loss": 0.011155224591493606, + "grad_norm": 0.12346649914979935, + "learning_rate": 7.132232550590247e-06, + "loss": 0.014, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009147377125918866, + "rkd_loss_weight": 0.30816027522087097, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3830151230096817, + "rkd_teacher_entropy": 3.739885354042053, + "rkd_teacher_similarity_mean": 0.7812703251838684, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20760 + }, + { + "epoch": 6.043060808844923, + "flow_matching_loss": 0.012755773216485976, + "grad_norm": 0.1320912092924118, + "learning_rate": 7.118159700327839e-06, + "loss": 0.0153, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00820133457891643, + "rkd_loss_weight": 0.30782692432403563, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38134744465351106, + "rkd_teacher_entropy": 3.7427887678146363, + "rkd_teacher_similarity_mean": 0.7798984706401825, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20770 + }, + { + "epoch": 6.045970322956066, + "flow_matching_loss": 0.013632971607148648, + "grad_norm": 0.11980052292346954, + "learning_rate": 7.104096427244603e-06, + "loss": 0.0163, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008547650137916207, + "rkd_loss_weight": 0.30749358236789703, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3876504570245743, + "rkd_teacher_entropy": 3.757816123962402, + "rkd_teacher_similarity_mean": 0.786648690700531, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20780 + }, + { + "epoch": 6.04887983706721, + "flow_matching_loss": 0.011667978391051293, + "grad_norm": 0.0892900824546814, + "learning_rate": 7.090042748428737e-06, + "loss": 0.0139, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007138275727629662, + "rkd_loss_weight": 0.30716024339199066, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38101787865161896, + "rkd_teacher_entropy": 3.7370598554611205, + "rkd_teacher_similarity_mean": 0.7795981407165528, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20790 + }, + { + "epoch": 6.051789351178353, + "flow_matching_loss": 0.011483307834714651, + "grad_norm": 0.10807269811630249, + "learning_rate": 7.075998680956797e-06, + "loss": 0.0135, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006451604794710875, + "rkd_loss_weight": 0.3068268895149231, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3848274737596512, + "rkd_teacher_entropy": 3.7437115907669067, + "rkd_teacher_similarity_mean": 0.7837573707103729, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20800 + }, + { + "epoch": 6.054698865289497, + "flow_matching_loss": 0.01446629473939538, + "grad_norm": 0.12483026087284088, + "learning_rate": 7.061964241893639e-06, + "loss": 0.0174, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009674386586993933, + "rkd_loss_weight": 0.3064935505390167, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38239030838012694, + "rkd_teacher_entropy": 3.7437485218048097, + "rkd_teacher_similarity_mean": 0.7808739125728608, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20810 + }, + { + "epoch": 6.05760837940064, + "flow_matching_loss": 0.012738514505326749, + "grad_norm": 0.08796081691980362, + "learning_rate": 7.047939448292442e-06, + "loss": 0.0149, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007085125846788287, + "rkd_loss_weight": 0.3061602056026459, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38438527286052704, + "rkd_teacher_entropy": 3.748482823371887, + "rkd_teacher_similarity_mean": 0.783588582277298, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20820 + }, + { + "epoch": 6.060517893511784, + "flow_matching_loss": 0.014282485563308, + "grad_norm": 0.11559157818555832, + "learning_rate": 7.033924317194644e-06, + "loss": 0.0169, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008411075966432691, + "rkd_loss_weight": 0.3058268576860428, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3784505784511566, + "rkd_teacher_entropy": 3.7296930313110352, + "rkd_teacher_similarity_mean": 0.7772214949131012, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20830 + }, + { + "epoch": 6.063427407622927, + "flow_matching_loss": 0.011614344734698533, + "grad_norm": 0.1087723895907402, + "learning_rate": 7.0199188656299524e-06, + "loss": 0.0139, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007325645629316568, + "rkd_loss_weight": 0.3054935187101364, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37612953782081604, + "rkd_teacher_entropy": 3.7165258646011354, + "rkd_teacher_similarity_mean": 0.7729065299034119, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20840 + }, + { + "epoch": 6.06633692173407, + "flow_matching_loss": 0.013034964073449373, + "grad_norm": 0.12348949164152145, + "learning_rate": 7.005923110616313e-06, + "loss": 0.0156, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008444891776889563, + "rkd_loss_weight": 0.3051601707935333, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38534447848796843, + "rkd_teacher_entropy": 3.7479437351226808, + "rkd_teacher_similarity_mean": 0.784741735458374, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20850 + }, + { + "epoch": 6.069246435845214, + "flow_matching_loss": 0.013669714611023665, + "grad_norm": 0.10350044816732407, + "learning_rate": 6.9919370691598855e-06, + "loss": 0.0158, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007138367835432291, + "rkd_loss_weight": 0.3048268258571625, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3828985184431076, + "rkd_teacher_entropy": 3.746484112739563, + "rkd_teacher_similarity_mean": 0.7823854804039001, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20860 + }, + { + "epoch": 6.0721559499563575, + "flow_matching_loss": 0.013565769325941801, + "grad_norm": 0.10709867626428604, + "learning_rate": 6.977960758255024e-06, + "loss": 0.0161, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00821955492720008, + "rkd_loss_weight": 0.3044934868812561, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3874219238758087, + "rkd_teacher_entropy": 3.7521692276000977, + "rkd_teacher_similarity_mean": 0.7866232216358184, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20870 + }, + { + "epoch": 6.0750654640675, + "flow_matching_loss": 0.0114660678897053, + "grad_norm": 0.09076465666294098, + "learning_rate": 6.963994194884276e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007137960847467184, + "rkd_loss_weight": 0.30416013598442077, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3840854912996292, + "rkd_teacher_entropy": 3.7386425495147706, + "rkd_teacher_similarity_mean": 0.7826256275177002, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20880 + }, + { + "epoch": 6.077974978178644, + "flow_matching_loss": 0.013501224853098393, + "grad_norm": 0.10571800917387009, + "learning_rate": 6.950037396018328e-06, + "loss": 0.016, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008261865098029375, + "rkd_loss_weight": 0.30382679402828217, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.383340510725975, + "rkd_teacher_entropy": 3.7549261808395387, + "rkd_teacher_similarity_mean": 0.7830353081226349, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20890 + }, + { + "epoch": 6.080884492289788, + "flow_matching_loss": 0.013901414349675179, + "grad_norm": 0.09270743280649185, + "learning_rate": 6.9360903786160056e-06, + "loss": 0.0162, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007728572282940149, + "rkd_loss_weight": 0.3034934550523758, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38726274371147157, + "rkd_teacher_entropy": 3.7656489849090575, + "rkd_teacher_similarity_mean": 0.7867876708507537, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20900 + }, + { + "epoch": 6.083794006400931, + "flow_matching_loss": 0.014142685476690531, + "grad_norm": 0.10137009620666504, + "learning_rate": 6.9221531596242485e-06, + "loss": 0.0164, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007295955205336213, + "rkd_loss_weight": 0.3031601011753082, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.388133230805397, + "rkd_teacher_entropy": 3.7638437509536744, + "rkd_teacher_similarity_mean": 0.7884635210037232, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20910 + }, + { + "epoch": 6.086703520512074, + "flow_matching_loss": 0.014488889649510384, + "grad_norm": 0.10425275564193726, + "learning_rate": 6.908225755978089e-06, + "loss": 0.0166, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006976422155275941, + "rkd_loss_weight": 0.30282676219940186, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3775335609912872, + "rkd_teacher_entropy": 3.7213034868240356, + "rkd_teacher_similarity_mean": 0.7762757062911987, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20920 + }, + { + "epoch": 6.089613034623218, + "flow_matching_loss": 0.012646771501749754, + "grad_norm": 0.11475187540054321, + "learning_rate": 6.8943081846006466e-06, + "loss": 0.0148, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007167579559609294, + "rkd_loss_weight": 0.30249341726303103, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3869589179754257, + "rkd_teacher_entropy": 3.748911499977112, + "rkd_teacher_similarity_mean": 0.7864787638187408, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20930 + }, + { + "epoch": 6.092522548734362, + "flow_matching_loss": 0.014739264268428087, + "grad_norm": 0.14025866985321045, + "learning_rate": 6.88040046240307e-06, + "loss": 0.0178, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010227359691634774, + "rkd_loss_weight": 0.3021600693464279, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.378938701748848, + "rkd_teacher_entropy": 3.7322240829467774, + "rkd_teacher_similarity_mean": 0.7791915535926819, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20940 + }, + { + "epoch": 6.0954320628455045, + "flow_matching_loss": 0.013818423636257649, + "grad_norm": 0.10313891619443893, + "learning_rate": 6.866502606284566e-06, + "loss": 0.0164, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008534390386193991, + "rkd_loss_weight": 0.30182673037052155, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38243833780288694, + "rkd_teacher_entropy": 3.7276857137680053, + "rkd_teacher_similarity_mean": 0.7802651345729827, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20950 + }, + { + "epoch": 6.098341576956648, + "flow_matching_loss": 0.014646969828754663, + "grad_norm": 0.10490883886814117, + "learning_rate": 6.852614633132335e-06, + "loss": 0.0169, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007625608239322901, + "rkd_loss_weight": 0.30149338245391843, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37844547629356384, + "rkd_teacher_entropy": 3.7311031341552736, + "rkd_teacher_similarity_mean": 0.7758755445480346, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20960 + }, + { + "epoch": 6.101251091067792, + "flow_matching_loss": 0.0144320884719491, + "grad_norm": 0.11768890917301178, + "learning_rate": 6.8387365598215665e-06, + "loss": 0.0169, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008314570039510727, + "rkd_loss_weight": 0.3011600375175476, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3806682288646698, + "rkd_teacher_entropy": 3.7281460046768187, + "rkd_teacher_similarity_mean": 0.7783378660678864, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20970 + }, + { + "epoch": 6.104160605178935, + "flow_matching_loss": 0.014335998240858317, + "grad_norm": 0.1328372061252594, + "learning_rate": 6.82486840321544e-06, + "loss": 0.0169, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008491386892274022, + "rkd_loss_weight": 0.30082669854164124, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38115206062793733, + "rkd_teacher_entropy": 3.7488773107528686, + "rkd_teacher_similarity_mean": 0.780630373954773, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20980 + }, + { + "epoch": 6.1070701192900785, + "flow_matching_loss": 0.013195353839546443, + "grad_norm": 0.12437675148248672, + "learning_rate": 6.811010180165069e-06, + "loss": 0.0153, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0070384024176746605, + "rkd_loss_weight": 0.30049334466457367, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3884857565164566, + "rkd_teacher_entropy": 3.756098508834839, + "rkd_teacher_similarity_mean": 0.7870683789253234, + "rkd_vlm_temp": 0.10000000149011612, + "step": 20990 + }, + { + "epoch": 6.109979633401222, + "flow_matching_loss": 0.012102872878313065, + "grad_norm": 0.07480885833501816, + "learning_rate": 6.797161907509499e-06, + "loss": 0.0142, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006917969509959221, + "rkd_loss_weight": 0.3001600056886673, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3838708817958832, + "rkd_teacher_entropy": 3.7447019815444946, + "rkd_teacher_similarity_mean": 0.7825215041637421, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21000 + }, + { + "epoch": 6.112889147512365, + "flow_matching_loss": 0.015529763232916594, + "grad_norm": 0.11315333098173141, + "learning_rate": 6.783323602075688e-06, + "loss": 0.0176, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007020529266446829, + "rkd_loss_weight": 0.2998266637325287, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3855596423149109, + "rkd_teacher_entropy": 3.751180076599121, + "rkd_teacher_similarity_mean": 0.7856233775615692, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21010 + }, + { + "epoch": 6.115798661623509, + "flow_matching_loss": 0.014818832557648421, + "grad_norm": 0.09594959765672684, + "learning_rate": 6.7694952806784784e-06, + "loss": 0.0178, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010089949239045382, + "rkd_loss_weight": 0.29949331283569336, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38089596331119535, + "rkd_teacher_entropy": 3.7466994762420653, + "rkd_teacher_similarity_mean": 0.780489730834961, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21020 + }, + { + "epoch": 6.118708175734652, + "flow_matching_loss": 0.012728230934590102, + "grad_norm": 0.1291360706090927, + "learning_rate": 6.7556769601205944e-06, + "loss": 0.0151, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008031494962051512, + "rkd_loss_weight": 0.299159973859787, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.387647220492363, + "rkd_teacher_entropy": 3.757465052604675, + "rkd_teacher_similarity_mean": 0.7876400887966156, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21030 + }, + { + "epoch": 6.121617689845796, + "flow_matching_loss": 0.014913064427673817, + "grad_norm": 0.11567318439483643, + "learning_rate": 6.741868657192594e-06, + "loss": 0.0173, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00785582079552114, + "rkd_loss_weight": 0.2988266259431839, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37810713350772857, + "rkd_teacher_entropy": 3.7271780490875246, + "rkd_teacher_similarity_mean": 0.7782279193401337, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21040 + }, + { + "epoch": 6.124527203956939, + "flow_matching_loss": 0.013993367925286293, + "grad_norm": 0.11757612973451614, + "learning_rate": 6.728070388672864e-06, + "loss": 0.0167, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008926253579556943, + "rkd_loss_weight": 0.29849328100681305, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38470159769058226, + "rkd_teacher_entropy": 3.753795099258423, + "rkd_teacher_similarity_mean": 0.7843352913856506, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21050 + }, + { + "epoch": 6.127436718068083, + "flow_matching_loss": 0.012629915773868561, + "grad_norm": 0.1419638991355896, + "learning_rate": 6.714282171327604e-06, + "loss": 0.0151, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008279273100197316, + "rkd_loss_weight": 0.2981599420309067, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3819705218076706, + "rkd_teacher_entropy": 3.727568507194519, + "rkd_teacher_similarity_mean": 0.7812910854816437, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21060 + }, + { + "epoch": 6.130346232179226, + "flow_matching_loss": 0.011968784732744098, + "grad_norm": 0.11494634300470352, + "learning_rate": 6.700504021910791e-06, + "loss": 0.0143, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0077095803339034315, + "rkd_loss_weight": 0.29782659113407134, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3804876536130905, + "rkd_teacher_entropy": 3.7408402442932127, + "rkd_teacher_similarity_mean": 0.7801079988479614, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21070 + }, + { + "epoch": 6.133255746290369, + "flow_matching_loss": 0.015353717934340238, + "grad_norm": 0.1306244283914566, + "learning_rate": 6.68673595716419e-06, + "loss": 0.0182, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009595428127795458, + "rkd_loss_weight": 0.29749324917793274, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37680227160453794, + "rkd_teacher_entropy": 3.7270928382873536, + "rkd_teacher_similarity_mean": 0.7760280966758728, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21080 + }, + { + "epoch": 6.136165260401513, + "flow_matching_loss": 0.012753668706864119, + "grad_norm": 0.12351679056882858, + "learning_rate": 6.6729779938172856e-06, + "loss": 0.0151, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00793175441212952, + "rkd_loss_weight": 0.29715991020202637, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38231031894683837, + "rkd_teacher_entropy": 3.739804410934448, + "rkd_teacher_similarity_mean": 0.7810729920864106, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21090 + }, + { + "epoch": 6.139074774512657, + "flow_matching_loss": 0.013326582824811339, + "grad_norm": 0.15421009063720703, + "learning_rate": 6.659230148587311e-06, + "loss": 0.0157, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007920297468081117, + "rkd_loss_weight": 0.2968265563249588, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3828830987215042, + "rkd_teacher_entropy": 3.7473968982696535, + "rkd_teacher_similarity_mean": 0.7816892445087433, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21100 + }, + { + "epoch": 6.141984288623799, + "flow_matching_loss": 0.012372752279043197, + "grad_norm": 0.1062193214893341, + "learning_rate": 6.645492438179188e-06, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0076329958159476515, + "rkd_loss_weight": 0.29649321734905243, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3747123837471008, + "rkd_teacher_entropy": 3.718140459060669, + "rkd_teacher_similarity_mean": 0.7726657688617706, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21110 + }, + { + "epoch": 6.144893802734943, + "flow_matching_loss": 0.012837481079623103, + "grad_norm": 0.12328284978866577, + "learning_rate": 6.631764879285531e-06, + "loss": 0.0152, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00806269682943821, + "rkd_loss_weight": 0.29615987539291383, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38578113317489626, + "rkd_teacher_entropy": 3.7580461740493774, + "rkd_teacher_similarity_mean": 0.7846780359745026, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21120 + }, + { + "epoch": 6.147803316846087, + "flow_matching_loss": 0.014468206372112036, + "grad_norm": 0.1329139918088913, + "learning_rate": 6.618047488586614e-06, + "loss": 0.0168, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007846780633553863, + "rkd_loss_weight": 0.2958265244960785, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37283054888248446, + "rkd_teacher_entropy": 3.7194245576858522, + "rkd_teacher_similarity_mean": 0.7717519760131836, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21130 + }, + { + "epoch": 6.1507128309572305, + "flow_matching_loss": 0.012799212243407965, + "grad_norm": 0.10084279626607895, + "learning_rate": 6.604340282750371e-06, + "loss": 0.0153, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008532117074355482, + "rkd_loss_weight": 0.2954931855201721, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39087899327278136, + "rkd_teacher_entropy": 3.7607017278671266, + "rkd_teacher_similarity_mean": 0.7896879136562347, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21140 + }, + { + "epoch": 6.153622345068373, + "flow_matching_loss": 0.012851734738796949, + "grad_norm": 0.09288040548563004, + "learning_rate": 6.590643278432342e-06, + "loss": 0.0149, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006888229120522738, + "rkd_loss_weight": 0.295159837603569, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38323408663272857, + "rkd_teacher_entropy": 3.7399495124816893, + "rkd_teacher_similarity_mean": 0.7813428699970245, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21150 + }, + { + "epoch": 6.156531859179517, + "flow_matching_loss": 0.011392732383683324, + "grad_norm": 0.09655419737100601, + "learning_rate": 6.576956492275682e-06, + "loss": 0.0137, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007975998939946294, + "rkd_loss_weight": 0.2948264926671982, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.383913192152977, + "rkd_teacher_entropy": 3.7374286890029906, + "rkd_teacher_similarity_mean": 0.7825051546096802, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21160 + }, + { + "epoch": 6.159441373290661, + "flow_matching_loss": 0.013735253270715475, + "grad_norm": 0.11560092866420746, + "learning_rate": 6.563279940911124e-06, + "loss": 0.0159, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007480406621471048, + "rkd_loss_weight": 0.2944931536912918, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3884110331535339, + "rkd_teacher_entropy": 3.771073317527771, + "rkd_teacher_similarity_mean": 0.7889213919639587, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21170 + }, + { + "epoch": 6.162350887401804, + "flow_matching_loss": 0.012614363059401511, + "grad_norm": 0.17440634965896606, + "learning_rate": 6.549613640956963e-06, + "loss": 0.0155, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009835981577634812, + "rkd_loss_weight": 0.29415980279445647, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3785723477602005, + "rkd_teacher_entropy": 3.718935966491699, + "rkd_teacher_similarity_mean": 0.7763738870620728, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21180 + }, + { + "epoch": 6.165260401512947, + "flow_matching_loss": 0.014256300125271081, + "grad_norm": 0.12116437405347824, + "learning_rate": 6.535957609019053e-06, + "loss": 0.0164, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007411521533504129, + "rkd_loss_weight": 0.29382646083831787, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3899273216724396, + "rkd_teacher_entropy": 3.7714722633361815, + "rkd_teacher_similarity_mean": 0.7891839802265167, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21190 + }, + { + "epoch": 6.168169915624091, + "flow_matching_loss": 0.012100922735407948, + "grad_norm": 0.09705296903848648, + "learning_rate": 6.5223118616907545e-06, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008580709341913462, + "rkd_loss_weight": 0.29349311888217927, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38025195598602296, + "rkd_teacher_entropy": 3.73023099899292, + "rkd_teacher_similarity_mean": 0.7786204993724823, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21200 + }, + { + "epoch": 6.171079429735234, + "flow_matching_loss": 0.013702771533280611, + "grad_norm": 0.09961173683404922, + "learning_rate": 6.508676415552939e-06, + "loss": 0.0163, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00901141120120883, + "rkd_loss_weight": 0.29315976798534393, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37333630621433256, + "rkd_teacher_entropy": 3.720479679107666, + "rkd_teacher_similarity_mean": 0.771304589509964, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21210 + }, + { + "epoch": 6.1739889438463775, + "flow_matching_loss": 0.01136711980216205, + "grad_norm": 0.10291881114244461, + "learning_rate": 6.495051287173952e-06, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010906112380325794, + "rkd_loss_weight": 0.29282642900943756, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37405534088611603, + "rkd_teacher_entropy": 3.7137231826782227, + "rkd_teacher_similarity_mean": 0.7734290301799774, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21220 + }, + { + "epoch": 6.176898457957521, + "flow_matching_loss": 0.014060010481625796, + "grad_norm": 0.11743607372045517, + "learning_rate": 6.481436493109621e-06, + "loss": 0.0166, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008606346463784576, + "rkd_loss_weight": 0.29249308407306673, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38585024774074556, + "rkd_teacher_entropy": 3.753564715385437, + "rkd_teacher_similarity_mean": 0.7853853642940521, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21230 + }, + { + "epoch": 6.179807972068665, + "flow_matching_loss": 0.01381339831277728, + "grad_norm": 0.13544917106628418, + "learning_rate": 6.467832049903194e-06, + "loss": 0.0161, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007729422766715288, + "rkd_loss_weight": 0.2921597361564636, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38732328116893766, + "rkd_teacher_entropy": 3.75159637928009, + "rkd_teacher_similarity_mean": 0.7860674917697906, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21240 + }, + { + "epoch": 6.182717486179808, + "flow_matching_loss": 0.011254294542595744, + "grad_norm": 0.107029028236866, + "learning_rate": 6.454237974085361e-06, + "loss": 0.0139, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00902167004533112, + "rkd_loss_weight": 0.29182639718055725, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38340129554271696, + "rkd_teacher_entropy": 3.7474010944366456, + "rkd_teacher_similarity_mean": 0.7829734742641449, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21250 + }, + { + "epoch": 6.1856270002909515, + "flow_matching_loss": 0.011921621672809125, + "grad_norm": 0.10289028286933899, + "learning_rate": 6.440654282174203e-06, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008732069795951248, + "rkd_loss_weight": 0.2914930462837219, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37788433134555816, + "rkd_teacher_entropy": 3.730106782913208, + "rkd_teacher_similarity_mean": 0.7777033686637879, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21260 + }, + { + "epoch": 6.188536514402095, + "flow_matching_loss": 0.012605037167668342, + "grad_norm": 0.1084321066737175, + "learning_rate": 6.427080990675187e-06, + "loss": 0.0148, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007709334278479219, + "rkd_loss_weight": 0.2911597043275833, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38253356218338014, + "rkd_teacher_entropy": 3.7433481454849242, + "rkd_teacher_similarity_mean": 0.783216267824173, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21270 + }, + { + "epoch": 6.191446028513238, + "flow_matching_loss": 0.01210402250289917, + "grad_norm": 0.09267369657754898, + "learning_rate": 6.41351811608114e-06, + "loss": 0.0141, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006761487107723951, + "rkd_loss_weight": 0.29082636535167694, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3807113140821457, + "rkd_teacher_entropy": 3.731534218788147, + "rkd_teacher_similarity_mean": 0.7805165469646453, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21280 + }, + { + "epoch": 6.194355542624382, + "flow_matching_loss": 0.012798432353883981, + "grad_norm": 0.0866793841123581, + "learning_rate": 6.399965674872228e-06, + "loss": 0.015, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007528997864574194, + "rkd_loss_weight": 0.2904930114746094, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38059123158454894, + "rkd_teacher_entropy": 3.7408227205276487, + "rkd_teacher_similarity_mean": 0.7792174816131592, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21290 + }, + { + "epoch": 6.197265056735525, + "flow_matching_loss": 0.013021770259365439, + "grad_norm": 0.08614625036716461, + "learning_rate": 6.386423683515957e-06, + "loss": 0.0151, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007142478460446, + "rkd_loss_weight": 0.290159672498703, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3891389787197113, + "rkd_teacher_entropy": 3.7614867448806764, + "rkd_teacher_similarity_mean": 0.7882800340652466, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21300 + }, + { + "epoch": 6.200174570846668, + "flow_matching_loss": 0.015264936303719879, + "grad_norm": 0.10052722692489624, + "learning_rate": 6.372892158467119e-06, + "loss": 0.0178, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008669673232361674, + "rkd_loss_weight": 0.2898263305425644, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3906680554151535, + "rkd_teacher_entropy": 3.7742948055267336, + "rkd_teacher_similarity_mean": 0.791988629102707, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21310 + }, + { + "epoch": 6.203084084957812, + "flow_matching_loss": 0.011400031112134456, + "grad_norm": 0.10042117536067963, + "learning_rate": 6.359371116167791e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0076091892551630735, + "rkd_loss_weight": 0.28949297964572906, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3859324276447296, + "rkd_teacher_entropy": 3.7470072507858276, + "rkd_teacher_similarity_mean": 0.7841217756271363, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21320 + }, + { + "epoch": 6.205993599068956, + "flow_matching_loss": 0.012149263452738524, + "grad_norm": 0.11944536864757538, + "learning_rate": 6.345860573047314e-06, + "loss": 0.0144, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007751456182450056, + "rkd_loss_weight": 0.2891596406698227, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.381196528673172, + "rkd_teacher_entropy": 3.7304707527160645, + "rkd_teacher_similarity_mean": 0.780468899011612, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21330 + }, + { + "epoch": 6.208903113180099, + "flow_matching_loss": 0.01335556348785758, + "grad_norm": 0.11038617044687271, + "learning_rate": 6.332360545522268e-06, + "loss": 0.0157, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008028032258152962, + "rkd_loss_weight": 0.28882629573345187, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37495545446872713, + "rkd_teacher_entropy": 3.7171921491622926, + "rkd_teacher_similarity_mean": 0.7736901760101318, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21340 + }, + { + "epoch": 6.211812627291242, + "flow_matching_loss": 0.014979666937142611, + "grad_norm": 0.1358538269996643, + "learning_rate": 6.31887104999647e-06, + "loss": 0.0173, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0080699750687927, + "rkd_loss_weight": 0.28849294781684875, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37422044277191163, + "rkd_teacher_entropy": 3.7195093870162963, + "rkd_teacher_similarity_mean": 0.7733404159545898, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21350 + }, + { + "epoch": 6.214722141402386, + "flow_matching_loss": 0.013883230928331613, + "grad_norm": 0.11328251659870148, + "learning_rate": 6.305392102860924e-06, + "loss": 0.016, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007228639908134937, + "rkd_loss_weight": 0.2881596088409424, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3832240641117096, + "rkd_teacher_entropy": 3.7445489883422853, + "rkd_teacher_similarity_mean": 0.7833566427230835, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21360 + }, + { + "epoch": 6.21763165551353, + "flow_matching_loss": 0.012062927521765233, + "grad_norm": 0.1543009877204895, + "learning_rate": 6.291923720493814e-06, + "loss": 0.015, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010225739888846874, + "rkd_loss_weight": 0.28782625794410704, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3831558793783188, + "rkd_teacher_entropy": 3.750443124771118, + "rkd_teacher_similarity_mean": 0.7823782801628113, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21370 + }, + { + "epoch": 6.220541169624672, + "flow_matching_loss": 0.0119944391772151, + "grad_norm": 0.11114174872636795, + "learning_rate": 6.278465919260507e-06, + "loss": 0.0144, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008503923844546079, + "rkd_loss_weight": 0.28749291598796844, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3794282078742981, + "rkd_teacher_entropy": 3.7383015632629393, + "rkd_teacher_similarity_mean": 0.7782214701175689, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21380 + }, + { + "epoch": 6.223450683735816, + "flow_matching_loss": 0.01215395936742425, + "grad_norm": 0.11852195113897324, + "learning_rate": 6.26501871551349e-06, + "loss": 0.0144, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00790541204623878, + "rkd_loss_weight": 0.2871595770120621, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38136587142944334, + "rkd_teacher_entropy": 3.7431785583496096, + "rkd_teacher_similarity_mean": 0.780180811882019, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21390 + }, + { + "epoch": 6.22636019784696, + "flow_matching_loss": 0.013783875573426484, + "grad_norm": 0.0917251706123352, + "learning_rate": 6.251582125592394e-06, + "loss": 0.0163, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008943019667640328, + "rkd_loss_weight": 0.2868262231349945, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3774545431137085, + "rkd_teacher_entropy": 3.7432562112808228, + "rkd_teacher_similarity_mean": 0.7755466878414154, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21400 + }, + { + "epoch": 6.229269711958103, + "flow_matching_loss": 0.012585729733109475, + "grad_norm": 0.08434329926967621, + "learning_rate": 6.2381561658239386e-06, + "loss": 0.0151, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008869079360738396, + "rkd_loss_weight": 0.28649288415908813, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3871888518333435, + "rkd_teacher_entropy": 3.7479228496551515, + "rkd_teacher_similarity_mean": 0.7849154710769654, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21410 + }, + { + "epoch": 6.232179226069246, + "flow_matching_loss": 0.013480461109429598, + "grad_norm": 0.08318453282117844, + "learning_rate": 6.2247408525219276e-06, + "loss": 0.0155, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007186410389840603, + "rkd_loss_weight": 0.2861595392227173, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38767299652099607, + "rkd_teacher_entropy": 3.77186598777771, + "rkd_teacher_similarity_mean": 0.7858823478221894, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21420 + }, + { + "epoch": 6.23508874018039, + "flow_matching_loss": 0.013429636228829622, + "grad_norm": 0.11856591701507568, + "learning_rate": 6.2113362019872335e-06, + "loss": 0.0158, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00837855776771903, + "rkd_loss_weight": 0.2858261913061142, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38469446301460264, + "rkd_teacher_entropy": 3.750714659690857, + "rkd_teacher_similarity_mean": 0.7837720334529876, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21430 + }, + { + "epoch": 6.237998254291533, + "flow_matching_loss": 0.01369821298867464, + "grad_norm": 0.1268233060836792, + "learning_rate": 6.197942230507772e-06, + "loss": 0.0159, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0076956568285822865, + "rkd_loss_weight": 0.2854928523302078, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38405721485614774, + "rkd_teacher_entropy": 3.7517180681228637, + "rkd_teacher_similarity_mean": 0.7827770352363587, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21440 + }, + { + "epoch": 6.240907768402677, + "flow_matching_loss": 0.012313305120915174, + "grad_norm": 0.09034231305122375, + "learning_rate": 6.184558954358475e-06, + "loss": 0.0142, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006499326741322875, + "rkd_loss_weight": 0.2851595044136047, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3900153964757919, + "rkd_teacher_entropy": 3.7643157482147216, + "rkd_teacher_similarity_mean": 0.7893395960330963, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21450 + }, + { + "epoch": 6.24381728251382, + "flow_matching_loss": 0.013235576823353768, + "grad_norm": 0.11855398118495941, + "learning_rate": 6.1711863898012965e-06, + "loss": 0.0152, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006871280912309885, + "rkd_loss_weight": 0.2848261594772339, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39175636470317843, + "rkd_teacher_entropy": 3.776726722717285, + "rkd_teacher_similarity_mean": 0.7918421804904938, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21460 + }, + { + "epoch": 6.246726796624964, + "flow_matching_loss": 0.014340079110115766, + "grad_norm": 0.08543150871992111, + "learning_rate": 6.157824553085155e-06, + "loss": 0.0165, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007763898791745305, + "rkd_loss_weight": 0.2844928205013275, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38058953881263735, + "rkd_teacher_entropy": 3.7470120906829836, + "rkd_teacher_similarity_mean": 0.780327844619751, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21470 + }, + { + "epoch": 6.249636310736107, + "flow_matching_loss": 0.013752542063593864, + "grad_norm": 0.10453467071056366, + "learning_rate": 6.144473460445945e-06, + "loss": 0.0158, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007134099211543799, + "rkd_loss_weight": 0.28415946662425995, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38167097568511965, + "rkd_teacher_entropy": 3.748147749900818, + "rkd_teacher_similarity_mean": 0.7815089523792267, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21480 + }, + { + "epoch": 6.2525458248472505, + "flow_matching_loss": 0.01370814060792327, + "grad_norm": 0.11450114846229553, + "learning_rate": 6.131133128106503e-06, + "loss": 0.0156, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006789977476000786, + "rkd_loss_weight": 0.2838261276483536, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38067152798175813, + "rkd_teacher_entropy": 3.7440834522247313, + "rkd_teacher_similarity_mean": 0.7806635677814484, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21490 + }, + { + "epoch": 6.255455338958394, + "flow_matching_loss": 0.011517224740236998, + "grad_norm": 0.08124151825904846, + "learning_rate": 6.117803572276584e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007196090975776314, + "rkd_loss_weight": 0.283492785692215, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3830836683511734, + "rkd_teacher_entropy": 3.755461406707764, + "rkd_teacher_similarity_mean": 0.783110111951828, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21500 + }, + { + "epoch": 6.258364853069537, + "flow_matching_loss": 0.014376628026366235, + "grad_norm": 0.14207486808300018, + "learning_rate": 6.104484809152866e-06, + "loss": 0.0162, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006539870006963611, + "rkd_loss_weight": 0.28315943479537964, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37594465017318723, + "rkd_teacher_entropy": 3.7339295387268066, + "rkd_teacher_similarity_mean": 0.7756898760795593, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21510 + }, + { + "epoch": 6.261274367180681, + "flow_matching_loss": 0.013917411211878061, + "grad_norm": 0.14424382150173187, + "learning_rate": 6.091176854918892e-06, + "loss": 0.0162, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007984069222584367, + "rkd_loss_weight": 0.28282609581947327, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3828570395708084, + "rkd_teacher_entropy": 3.7585158824920653, + "rkd_teacher_similarity_mean": 0.7832929491996765, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21520 + }, + { + "epoch": 6.2641838812918245, + "flow_matching_loss": 0.011980564845725895, + "grad_norm": 0.15929636359214783, + "learning_rate": 6.077879725745092e-06, + "loss": 0.014, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007325841207057237, + "rkd_loss_weight": 0.28249275088310244, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38325375616550444, + "rkd_teacher_entropy": 3.7348204612731934, + "rkd_teacher_similarity_mean": 0.7831856906414032, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21530 + }, + { + "epoch": 6.267093395402968, + "flow_matching_loss": 0.012341897748410702, + "grad_norm": 0.12049716711044312, + "learning_rate": 6.064593437788727e-06, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007810813374817371, + "rkd_loss_weight": 0.28215940296649933, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3857292771339417, + "rkd_teacher_entropy": 3.7635390043258665, + "rkd_teacher_similarity_mean": 0.7859474360942841, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21540 + }, + { + "epoch": 6.270002909514111, + "flow_matching_loss": 0.014186625694856047, + "grad_norm": 0.12140262126922607, + "learning_rate": 6.051318007193881e-06, + "loss": 0.0166, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008606077684089542, + "rkd_loss_weight": 0.28182606399059296, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3764761030673981, + "rkd_teacher_entropy": 3.7211941719055175, + "rkd_teacher_similarity_mean": 0.7753873467445374, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21550 + }, + { + "epoch": 6.272912423625255, + "flow_matching_loss": 0.012306543299928308, + "grad_norm": 0.1050451248884201, + "learning_rate": 6.038053450091469e-06, + "loss": 0.0144, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007363239955157042, + "rkd_loss_weight": 0.28149271607398985, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37865925431251524, + "rkd_teacher_entropy": 3.7274477243423463, + "rkd_teacher_similarity_mean": 0.7776244342327118, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21560 + }, + { + "epoch": 6.275821937736398, + "flow_matching_loss": 0.012852516258135438, + "grad_norm": 0.10507518798112869, + "learning_rate": 6.0247997825991705e-06, + "loss": 0.0149, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007378141721710562, + "rkd_loss_weight": 0.281159371137619, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38068507313728334, + "rkd_teacher_entropy": 3.7511919021606444, + "rkd_teacher_similarity_mean": 0.7813495695590973, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21570 + }, + { + "epoch": 6.278731451847541, + "flow_matching_loss": 0.013155852816998959, + "grad_norm": 0.11177048832178116, + "learning_rate": 6.011557020821444e-06, + "loss": 0.015, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0066666063386946915, + "rkd_loss_weight": 0.28082603216171265, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38038588166236875, + "rkd_teacher_entropy": 3.7382375478744505, + "rkd_teacher_similarity_mean": 0.7799275994300843, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21580 + }, + { + "epoch": 6.281640965958685, + "flow_matching_loss": 0.012692561140283942, + "grad_norm": 0.17218062281608582, + "learning_rate": 5.99832518084949e-06, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0065746643813326955, + "rkd_loss_weight": 0.2804926782846451, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38584989309310913, + "rkd_teacher_entropy": 3.754214334487915, + "rkd_teacher_similarity_mean": 0.7854091823101044, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21590 + }, + { + "epoch": 6.284550480069829, + "flow_matching_loss": 0.012418211624026298, + "grad_norm": 0.11375381052494049, + "learning_rate": 5.98510427876124e-06, + "loss": 0.0144, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007128327619284391, + "rkd_loss_weight": 0.2801593393087387, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3735949009656906, + "rkd_teacher_entropy": 3.7166704654693605, + "rkd_teacher_similarity_mean": 0.773271507024765, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21600 + }, + { + "epoch": 6.2874599941809715, + "flow_matching_loss": 0.014143276214599609, + "grad_norm": 0.09657038748264313, + "learning_rate": 5.971894330621344e-06, + "loss": 0.0167, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009010922769084573, + "rkd_loss_weight": 0.2798259973526001, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37363220751285553, + "rkd_teacher_entropy": 3.718584966659546, + "rkd_teacher_similarity_mean": 0.7739151775836944, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21610 + }, + { + "epoch": 6.290369508292115, + "flow_matching_loss": 0.012673120200634002, + "grad_norm": 0.09707143902778625, + "learning_rate": 5.958695352481134e-06, + "loss": 0.0144, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006322079012170434, + "rkd_loss_weight": 0.27949264645576477, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38426002860069275, + "rkd_teacher_entropy": 3.7481465339660645, + "rkd_teacher_similarity_mean": 0.7843905031681061, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21620 + }, + { + "epoch": 6.293279022403259, + "flow_matching_loss": 0.012304824916645885, + "grad_norm": 0.11285651475191116, + "learning_rate": 5.945507360378608e-06, + "loss": 0.0142, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006933488417416811, + "rkd_loss_weight": 0.2791593074798584, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38374751806259155, + "rkd_teacher_entropy": 3.7450515031814575, + "rkd_teacher_similarity_mean": 0.7836446523666382, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21630 + }, + { + "epoch": 6.296188536514402, + "flow_matching_loss": 0.012310709059238433, + "grad_norm": 0.0942019671201706, + "learning_rate": 5.9323303703384225e-06, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007687607919797302, + "rkd_loss_weight": 0.2788259595632553, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3810804903507233, + "rkd_teacher_entropy": 3.734437584877014, + "rkd_teacher_similarity_mean": 0.7810730934143066, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21640 + }, + { + "epoch": 6.299098050625545, + "flow_matching_loss": 0.013754615280777215, + "grad_norm": 0.07846884429454803, + "learning_rate": 5.919164398371857e-06, + "loss": 0.0161, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008479242399334907, + "rkd_loss_weight": 0.27849261462688446, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3790287435054779, + "rkd_teacher_entropy": 3.7309563636779783, + "rkd_teacher_similarity_mean": 0.778371798992157, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21650 + }, + { + "epoch": 6.302007564736689, + "flow_matching_loss": 0.011529747815802693, + "grad_norm": 0.11504285782575607, + "learning_rate": 5.906009460476818e-06, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006418594066053629, + "rkd_loss_weight": 0.2781592756509781, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37898936569690705, + "rkd_teacher_entropy": 3.7375118494033814, + "rkd_teacher_similarity_mean": 0.7792660057544708, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21660 + }, + { + "epoch": 6.304917078847833, + "flow_matching_loss": 0.013615939486771821, + "grad_norm": 0.10372284799814224, + "learning_rate": 5.892865572637789e-06, + "loss": 0.0154, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006477637542411685, + "rkd_loss_weight": 0.27782592475414275, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3780125617980957, + "rkd_teacher_entropy": 3.7266109228134154, + "rkd_teacher_similarity_mean": 0.7776168048381805, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21670 + }, + { + "epoch": 6.307826592958976, + "flow_matching_loss": 0.013846303336322308, + "grad_norm": 0.13258424401283264, + "learning_rate": 5.8797327508258396e-06, + "loss": 0.0162, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008539180411025882, + "rkd_loss_weight": 0.27749258279800415, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.370265331864357, + "rkd_teacher_entropy": 3.7171345472335817, + "rkd_teacher_similarity_mean": 0.7701380789279938, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21680 + }, + { + "epoch": 6.310736107070119, + "flow_matching_loss": 0.01113377120345831, + "grad_norm": 0.09577906131744385, + "learning_rate": 5.866611010998584e-06, + "loss": 0.0132, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007333650207147002, + "rkd_loss_weight": 0.27715924084186555, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3828715831041336, + "rkd_teacher_entropy": 3.7402453422546387, + "rkd_teacher_similarity_mean": 0.7831986904144287, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21690 + }, + { + "epoch": 6.313645621181263, + "flow_matching_loss": 0.013474972127005458, + "grad_norm": 0.08860200643539429, + "learning_rate": 5.853500369100172e-06, + "loss": 0.0158, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008251250628381968, + "rkd_loss_weight": 0.2768258899450302, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37935810089111327, + "rkd_teacher_entropy": 3.746433901786804, + "rkd_teacher_similarity_mean": 0.7798234581947326, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21700 + }, + { + "epoch": 6.316555135292406, + "flow_matching_loss": 0.013594843447208405, + "grad_norm": 0.09532913565635681, + "learning_rate": 5.840400841061265e-06, + "loss": 0.016, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008861090615391731, + "rkd_loss_weight": 0.27649255096912384, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3752537578344345, + "rkd_teacher_entropy": 3.7333172082901003, + "rkd_teacher_similarity_mean": 0.7758554875850677, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21710 + }, + { + "epoch": 6.31946464940355, + "flow_matching_loss": 0.01322223450988531, + "grad_norm": 0.13314157724380493, + "learning_rate": 5.827312442799034e-06, + "loss": 0.0153, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0076156556606292725, + "rkd_loss_weight": 0.276159206032753, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3906512796878815, + "rkd_teacher_entropy": 3.765079975128174, + "rkd_teacher_similarity_mean": 0.7913875043392181, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21720 + }, + { + "epoch": 6.322374163514693, + "flow_matching_loss": 0.012735649570822715, + "grad_norm": 0.12478158622980118, + "learning_rate": 5.814235190217114e-06, + "loss": 0.0148, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007410240592435002, + "rkd_loss_weight": 0.2758258581161499, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3827408879995346, + "rkd_teacher_entropy": 3.733053278923035, + "rkd_teacher_similarity_mean": 0.7818647563457489, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21730 + }, + { + "epoch": 6.325283677625836, + "flow_matching_loss": 0.011414681514725088, + "grad_norm": 0.08490439504384995, + "learning_rate": 5.801169099205599e-06, + "loss": 0.0137, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008260603528469801, + "rkd_loss_weight": 0.27549251914024353, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3820325195789337, + "rkd_teacher_entropy": 3.7321767807006836, + "rkd_teacher_similarity_mean": 0.7809220790863037, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21740 + }, + { + "epoch": 6.32819319173698, + "flow_matching_loss": 0.01361813247203827, + "grad_norm": 0.13060584664344788, + "learning_rate": 5.788114185641022e-06, + "loss": 0.0156, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007264970382675528, + "rkd_loss_weight": 0.2751591712236404, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39035335183143616, + "rkd_teacher_entropy": 3.754322719573975, + "rkd_teacher_similarity_mean": 0.7902082622051239, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21750 + }, + { + "epoch": 6.3311027058481235, + "flow_matching_loss": 0.012544028274714947, + "grad_norm": 0.09120646864175797, + "learning_rate": 5.775070465386328e-06, + "loss": 0.0149, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008586584078148007, + "rkd_loss_weight": 0.2748258262872696, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38614324629306795, + "rkd_teacher_entropy": 3.7541924953460692, + "rkd_teacher_similarity_mean": 0.7862500309944153, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21760 + }, + { + "epoch": 6.334012219959266, + "flow_matching_loss": 0.01219741078093648, + "grad_norm": 0.13531678915023804, + "learning_rate": 5.7620379542908765e-06, + "loss": 0.0141, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006981030246242881, + "rkd_loss_weight": 0.2744924873113632, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38250536322593687, + "rkd_teacher_entropy": 3.7384802103042603, + "rkd_teacher_similarity_mean": 0.782401192188263, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21770 + }, + { + "epoch": 6.33692173407041, + "flow_matching_loss": 0.013176437653601169, + "grad_norm": 0.1044936552643776, + "learning_rate": 5.749016668190395e-06, + "loss": 0.0156, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00896386792883277, + "rkd_loss_weight": 0.2741591364145279, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37903415262699125, + "rkd_teacher_entropy": 3.726941466331482, + "rkd_teacher_similarity_mean": 0.777740079164505, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21780 + }, + { + "epoch": 6.339831248181554, + "flow_matching_loss": 0.01089831991121173, + "grad_norm": 0.08265693485736847, + "learning_rate": 5.736006622906973e-06, + "loss": 0.0127, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006432105600833893, + "rkd_loss_weight": 0.2738257944583893, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3863248020410538, + "rkd_teacher_entropy": 3.753118634223938, + "rkd_teacher_similarity_mean": 0.7861849248409272, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21790 + }, + { + "epoch": 6.3427407622926975, + "flow_matching_loss": 0.013295525126159191, + "grad_norm": 0.13222506642341614, + "learning_rate": 5.723007834249034e-06, + "loss": 0.0155, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008164799120277167, + "rkd_loss_weight": 0.2734924525022507, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3826671689748764, + "rkd_teacher_entropy": 3.7402043104171754, + "rkd_teacher_similarity_mean": 0.7821196496486664, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21800 + }, + { + "epoch": 6.34565027640384, + "flow_matching_loss": 0.013330667512491345, + "grad_norm": 0.1055857241153717, + "learning_rate": 5.710020318011346e-06, + "loss": 0.0153, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007232021540403366, + "rkd_loss_weight": 0.27315910160541534, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3858011782169342, + "rkd_teacher_entropy": 3.7491193056106566, + "rkd_teacher_similarity_mean": 0.785664439201355, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21810 + }, + { + "epoch": 6.348559790514984, + "flow_matching_loss": 0.013503204099833965, + "grad_norm": 0.141961932182312, + "learning_rate": 5.697044089974955e-06, + "loss": 0.0165, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.011024458426982164, + "rkd_loss_weight": 0.27282576262950897, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38133966028690336, + "rkd_teacher_entropy": 3.7440494775772093, + "rkd_teacher_similarity_mean": 0.7797088205814362, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21820 + }, + { + "epoch": 6.351469304626128, + "flow_matching_loss": 0.013807710399851203, + "grad_norm": 0.11200295388698578, + "learning_rate": 5.68407916590721e-06, + "loss": 0.0158, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007297349441796541, + "rkd_loss_weight": 0.27249241769313814, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38103753328323364, + "rkd_teacher_entropy": 3.73595871925354, + "rkd_teacher_similarity_mean": 0.7802226781845093, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21830 + }, + { + "epoch": 6.3543788187372705, + "flow_matching_loss": 0.012923381198197604, + "grad_norm": 0.1397387534379959, + "learning_rate": 5.671125561561713e-06, + "loss": 0.0148, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00680011734366417, + "rkd_loss_weight": 0.27215906977653503, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37356662154197695, + "rkd_teacher_entropy": 3.721289849281311, + "rkd_teacher_similarity_mean": 0.7724689960479736, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21840 + }, + { + "epoch": 6.357288332848414, + "flow_matching_loss": 0.014121169783174991, + "grad_norm": 0.10038493573665619, + "learning_rate": 5.658183292678315e-06, + "loss": 0.0159, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006682040262967348, + "rkd_loss_weight": 0.27182573080062866, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.376901188492775, + "rkd_teacher_entropy": 3.7209784030914306, + "rkd_teacher_similarity_mean": 0.7754959881305694, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21850 + }, + { + "epoch": 6.360197846959558, + "flow_matching_loss": 0.013677221909165382, + "grad_norm": 0.11973843723535538, + "learning_rate": 5.645252374983091e-06, + "loss": 0.0157, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007483878079801798, + "rkd_loss_weight": 0.2714923799037933, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38796072602272036, + "rkd_teacher_entropy": 3.7590242862701415, + "rkd_teacher_similarity_mean": 0.7859143912792206, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21860 + }, + { + "epoch": 6.363107361070702, + "flow_matching_loss": 0.013695664703845978, + "grad_norm": 0.10480210185050964, + "learning_rate": 5.6323328241883224e-06, + "loss": 0.0157, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007312144665047526, + "rkd_loss_weight": 0.2711590379476547, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3845749765634537, + "rkd_teacher_entropy": 3.744803762435913, + "rkd_teacher_similarity_mean": 0.7825128912925721, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21870 + }, + { + "epoch": 6.3660168751818444, + "flow_matching_loss": 0.013710768707096577, + "grad_norm": 0.10732293128967285, + "learning_rate": 5.61942465599249e-06, + "loss": 0.016, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008557942509651185, + "rkd_loss_weight": 0.27082569897174835, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.388265585899353, + "rkd_teacher_entropy": 3.761547493934631, + "rkd_teacher_similarity_mean": 0.7882125318050385, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21880 + }, + { + "epoch": 6.368926389292988, + "flow_matching_loss": 0.013623124547302723, + "grad_norm": 0.12043724954128265, + "learning_rate": 5.60652788608023e-06, + "loss": 0.0162, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0095841811504215, + "rkd_loss_weight": 0.2704923450946808, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.380126616358757, + "rkd_teacher_entropy": 3.728399229049683, + "rkd_teacher_similarity_mean": 0.7789280652999878, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21890 + }, + { + "epoch": 6.371835903404132, + "flow_matching_loss": 0.012519299192354083, + "grad_norm": 0.08258281648159027, + "learning_rate": 5.593642530122336e-06, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00753250289708376, + "rkd_loss_weight": 0.2701590061187744, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3831793636083603, + "rkd_teacher_entropy": 3.742618942260742, + "rkd_teacher_similarity_mean": 0.782850843667984, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21900 + }, + { + "epoch": 6.374745417515275, + "flow_matching_loss": 0.012689227424561977, + "grad_norm": 0.12735828757286072, + "learning_rate": 5.58076860377573e-06, + "loss": 0.0141, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0053905485663563015, + "rkd_loss_weight": 0.2698256611824036, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39844112694263456, + "rkd_teacher_entropy": 3.7909165382385255, + "rkd_teacher_similarity_mean": 0.7989763557910919, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21910 + }, + { + "epoch": 6.377654931626418, + "flow_matching_loss": 0.012466795463114976, + "grad_norm": 0.13129504024982452, + "learning_rate": 5.567906122683439e-06, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007656256202608347, + "rkd_loss_weight": 0.2694923132658005, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.389781191945076, + "rkd_teacher_entropy": 3.762184762954712, + "rkd_teacher_similarity_mean": 0.7893721103668213, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21920 + }, + { + "epoch": 6.380564445737562, + "flow_matching_loss": 0.01557477079331875, + "grad_norm": 0.10795193165540695, + "learning_rate": 5.555055102474604e-06, + "loss": 0.018, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008925121882930398, + "rkd_loss_weight": 0.2691589742898941, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3836907982826233, + "rkd_teacher_entropy": 3.7417718172073364, + "rkd_teacher_similarity_mean": 0.7841211318969726, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21930 + }, + { + "epoch": 6.383473959848705, + "flow_matching_loss": 0.013114710990339517, + "grad_norm": 0.10041624307632446, + "learning_rate": 5.542215558764421e-06, + "loss": 0.0153, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00827423669397831, + "rkd_loss_weight": 0.268825626373291, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38231500089168546, + "rkd_teacher_entropy": 3.7495840787887573, + "rkd_teacher_similarity_mean": 0.7826592683792114, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21940 + }, + { + "epoch": 6.386383473959849, + "flow_matching_loss": 0.0129871373064816, + "grad_norm": 0.13119642436504364, + "learning_rate": 5.529387507154138e-06, + "loss": 0.0149, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007046547718346119, + "rkd_loss_weight": 0.26849228143692017, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3775122225284576, + "rkd_teacher_entropy": 3.7273245811462403, + "rkd_teacher_similarity_mean": 0.7779600024223328, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21950 + }, + { + "epoch": 6.389292988070992, + "flow_matching_loss": 0.011748011782765388, + "grad_norm": 0.11489468812942505, + "learning_rate": 5.516570963231058e-06, + "loss": 0.0135, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006424958584830165, + "rkd_loss_weight": 0.2681589424610138, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3783364534378052, + "rkd_teacher_entropy": 3.733150506019592, + "rkd_teacher_similarity_mean": 0.7789305984973908, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21960 + }, + { + "epoch": 6.392202502182135, + "flow_matching_loss": 0.012785492092370987, + "grad_norm": 0.09791462868452072, + "learning_rate": 5.503765942568484e-06, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006819490622729063, + "rkd_loss_weight": 0.26782559156417846, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38662211000919344, + "rkd_teacher_entropy": 3.7581047296524046, + "rkd_teacher_similarity_mean": 0.7855065405368805, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21970 + }, + { + "epoch": 6.395112016293279, + "flow_matching_loss": 0.0134774137288332, + "grad_norm": 0.10273843258619308, + "learning_rate": 5.49097246072573e-06, + "loss": 0.0157, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008266536239534616, + "rkd_loss_weight": 0.26749224960803986, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3850802659988403, + "rkd_teacher_entropy": 3.7637826442718505, + "rkd_teacher_similarity_mean": 0.7844417631626129, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21980 + }, + { + "epoch": 6.398021530404423, + "flow_matching_loss": 0.012667105440050364, + "grad_norm": 0.10581400990486145, + "learning_rate": 5.4781905332480805e-06, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00720638046041131, + "rkd_loss_weight": 0.2671589106321335, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3787087917327881, + "rkd_teacher_entropy": 3.7372145652770996, + "rkd_teacher_similarity_mean": 0.7791398525238037, + "rkd_vlm_temp": 0.10000000149011612, + "step": 21990 + }, + { + "epoch": 6.400931044515566, + "flow_matching_loss": 0.013173624034970998, + "grad_norm": 0.13025900721549988, + "learning_rate": 5.4654201756667795e-06, + "loss": 0.0154, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00820303983055055, + "rkd_loss_weight": 0.2668255567550659, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38036414682865144, + "rkd_teacher_entropy": 3.7380271434783934, + "rkd_teacher_similarity_mean": 0.7798047363758087, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22000 + }, + { + "epoch": 6.403840558626709, + "flow_matching_loss": 0.013095996575430036, + "grad_norm": 0.14386218786239624, + "learning_rate": 5.452661403499017e-06, + "loss": 0.0151, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00739912879653275, + "rkd_loss_weight": 0.26649221777915955, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3767847716808319, + "rkd_teacher_entropy": 3.7342288732528686, + "rkd_teacher_similarity_mean": 0.7767343640327453, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22010 + }, + { + "epoch": 6.406750072737853, + "flow_matching_loss": 0.012637541349977254, + "grad_norm": 0.11752122640609741, + "learning_rate": 5.4399142322479005e-06, + "loss": 0.0147, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0075980515219271185, + "rkd_loss_weight": 0.2661588728427887, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38628549575805665, + "rkd_teacher_entropy": 3.746689772605896, + "rkd_teacher_similarity_mean": 0.7864336013793946, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22020 + }, + { + "epoch": 6.4096595868489965, + "flow_matching_loss": 0.012940611969679595, + "grad_norm": 0.12103645503520966, + "learning_rate": 5.427178677402452e-06, + "loss": 0.015, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007929950207471847, + "rkd_loss_weight": 0.2658255249261856, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3719021052122116, + "rkd_teacher_entropy": 3.720836067199707, + "rkd_teacher_similarity_mean": 0.7726060152053833, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22030 + }, + { + "epoch": 6.412569100960139, + "flow_matching_loss": 0.012434200197458268, + "grad_norm": 0.13119274377822876, + "learning_rate": 5.41445475443757e-06, + "loss": 0.0147, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008359749615192414, + "rkd_loss_weight": 0.26549218595027924, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38531077206134795, + "rkd_teacher_entropy": 3.7545902252197267, + "rkd_teacher_similarity_mean": 0.7852225482463837, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22040 + }, + { + "epoch": 6.415478615071283, + "flow_matching_loss": 0.013572407327592373, + "grad_norm": 0.10892996937036514, + "learning_rate": 5.4017424788140194e-06, + "loss": 0.0156, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007698015728965402, + "rkd_loss_weight": 0.2651588380336761, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.381572625041008, + "rkd_teacher_entropy": 3.7443453073501587, + "rkd_teacher_similarity_mean": 0.7820142686367035, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22050 + }, + { + "epoch": 6.418388129182427, + "flow_matching_loss": 0.009449582500383257, + "grad_norm": 0.09687745571136475, + "learning_rate": 5.389041865978414e-06, + "loss": 0.0114, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007479532435536384, + "rkd_loss_weight": 0.2648254930973053, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38574435114860534, + "rkd_teacher_entropy": 3.7534762620925903, + "rkd_teacher_similarity_mean": 0.7852429687976837, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22060 + }, + { + "epoch": 6.4212976432935704, + "flow_matching_loss": 0.012835238222032785, + "grad_norm": 0.09355805814266205, + "learning_rate": 5.37635293136319e-06, + "loss": 0.0149, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00771111948415637, + "rkd_loss_weight": 0.2644921541213989, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3819135308265686, + "rkd_teacher_entropy": 3.7430436849594115, + "rkd_teacher_similarity_mean": 0.7815626561641693, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22070 + }, + { + "epoch": 6.424207157404713, + "flow_matching_loss": 0.012804587883874775, + "grad_norm": 0.09951098263263702, + "learning_rate": 5.363675690386609e-06, + "loss": 0.0151, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008694839663803577, + "rkd_loss_weight": 0.26415880024433136, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3799386531114578, + "rkd_teacher_entropy": 3.7324751138687136, + "rkd_teacher_similarity_mean": 0.780401611328125, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22080 + }, + { + "epoch": 6.427116671515857, + "flow_matching_loss": 0.011508603394031525, + "grad_norm": 0.13507069647312164, + "learning_rate": 5.351010158452708e-06, + "loss": 0.0131, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006070348527282476, + "rkd_loss_weight": 0.263825461268425, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3866204977035522, + "rkd_teacher_entropy": 3.757246398925781, + "rkd_teacher_similarity_mean": 0.7870341658592224, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22090 + }, + { + "epoch": 6.430026185627001, + "flow_matching_loss": 0.011845802422612906, + "grad_norm": 0.07543909549713135, + "learning_rate": 5.338356350951308e-06, + "loss": 0.0139, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007666609156876803, + "rkd_loss_weight": 0.2634921193122864, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3905374020338058, + "rkd_teacher_entropy": 3.772280287742615, + "rkd_teacher_similarity_mean": 0.7907771110534668, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22100 + }, + { + "epoch": 6.4329356997381435, + "flow_matching_loss": 0.013284196145832539, + "grad_norm": 0.11306269466876984, + "learning_rate": 5.325714283257974e-06, + "loss": 0.0156, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008640807028859854, + "rkd_loss_weight": 0.26315876841545105, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3837485760450363, + "rkd_teacher_entropy": 3.753568983078003, + "rkd_teacher_similarity_mean": 0.7841917097568512, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22110 + }, + { + "epoch": 6.435845213849287, + "flow_matching_loss": 0.011124091688543558, + "grad_norm": 0.10997922718524933, + "learning_rate": 5.313083970734013e-06, + "loss": 0.0134, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008746403688564896, + "rkd_loss_weight": 0.2628254294395447, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3773739457130432, + "rkd_teacher_entropy": 3.728630614280701, + "rkd_teacher_similarity_mean": 0.7770538628101349, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22120 + }, + { + "epoch": 6.438754727960431, + "flow_matching_loss": 0.014170865062624216, + "grad_norm": 0.10639003664255142, + "learning_rate": 5.300465428726439e-06, + "loss": 0.0159, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006757671060040593, + "rkd_loss_weight": 0.26249208152294157, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3883766561746597, + "rkd_teacher_entropy": 3.7662378311157227, + "rkd_teacher_similarity_mean": 0.7889964878559113, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22130 + }, + { + "epoch": 6.441664242071574, + "flow_matching_loss": 0.012079961597919464, + "grad_norm": 0.12209311127662659, + "learning_rate": 5.287858672567979e-06, + "loss": 0.014, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007278136583045125, + "rkd_loss_weight": 0.26215873658657074, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37822422981262205, + "rkd_teacher_entropy": 3.727664852142334, + "rkd_teacher_similarity_mean": 0.7773135602474213, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22140 + }, + { + "epoch": 6.444573756182717, + "flow_matching_loss": 0.013118639960885048, + "grad_norm": 0.094544917345047, + "learning_rate": 5.275263717577028e-06, + "loss": 0.015, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007336131855845451, + "rkd_loss_weight": 0.26182539761066437, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3874555766582489, + "rkd_teacher_entropy": 3.7641700506210327, + "rkd_teacher_similarity_mean": 0.7878447413444519, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22150 + }, + { + "epoch": 6.447483270293861, + "flow_matching_loss": 0.015829056967049837, + "grad_norm": 0.09558893740177155, + "learning_rate": 5.2626805790576435e-06, + "loss": 0.0176, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006588989775627852, + "rkd_loss_weight": 0.26149204671382903, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37837542593479156, + "rkd_teacher_entropy": 3.7419316291809084, + "rkd_teacher_similarity_mean": 0.7785996496677399, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22160 + }, + { + "epoch": 6.450392784405004, + "flow_matching_loss": 0.01261992584913969, + "grad_norm": 0.1199973076581955, + "learning_rate": 5.250109272299523e-06, + "loss": 0.0143, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006603428209200501, + "rkd_loss_weight": 0.26115870475769043, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38938207030296323, + "rkd_teacher_entropy": 3.7648151874542237, + "rkd_teacher_similarity_mean": 0.7898073732852936, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22170 + }, + { + "epoch": 6.453302298516148, + "flow_matching_loss": 0.012560274126008153, + "grad_norm": 0.11736756563186646, + "learning_rate": 5.237549812577987e-06, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007680546026676893, + "rkd_loss_weight": 0.26082536578178406, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37642196714878084, + "rkd_teacher_entropy": 3.732498550415039, + "rkd_teacher_similarity_mean": 0.7753084301948547, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22180 + }, + { + "epoch": 6.456211812627291, + "flow_matching_loss": 0.012194771319627762, + "grad_norm": 0.09164272993803024, + "learning_rate": 5.225002215153973e-06, + "loss": 0.0143, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00820860881358385, + "rkd_loss_weight": 0.2604920119047165, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3769872307777405, + "rkd_teacher_entropy": 3.735915923118591, + "rkd_teacher_similarity_mean": 0.777320945262909, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22190 + }, + { + "epoch": 6.459121326738435, + "flow_matching_loss": 0.012644085008651018, + "grad_norm": 0.1230141818523407, + "learning_rate": 5.21246649527399e-06, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0071601904928684235, + "rkd_loss_weight": 0.2601586729288101, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3806758612394333, + "rkd_teacher_entropy": 3.751622724533081, + "rkd_teacher_similarity_mean": 0.7813938856124878, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22200 + }, + { + "epoch": 6.462030840849578, + "flow_matching_loss": 0.015029383916407823, + "grad_norm": 0.10272075980901718, + "learning_rate": 5.199942668170118e-06, + "loss": 0.0168, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006724086590111256, + "rkd_loss_weight": 0.2598253309726715, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38151854276657104, + "rkd_teacher_entropy": 3.751729393005371, + "rkd_teacher_similarity_mean": 0.7825010597705842, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22210 + }, + { + "epoch": 6.464940354960722, + "flow_matching_loss": 0.011364368768408895, + "grad_norm": 0.10279938578605652, + "learning_rate": 5.187430749059986e-06, + "loss": 0.0131, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0066184498835355045, + "rkd_loss_weight": 0.2594919800758362, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38522942662239074, + "rkd_teacher_entropy": 3.7581095695495605, + "rkd_teacher_similarity_mean": 0.7865125417709351, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22220 + }, + { + "epoch": 6.467849869071865, + "flow_matching_loss": 0.01188847040757537, + "grad_norm": 0.09369553625583649, + "learning_rate": 5.174930753146763e-06, + "loss": 0.0135, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00603679814375937, + "rkd_loss_weight": 0.2591586410999298, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3895007610321045, + "rkd_teacher_entropy": 3.765169358253479, + "rkd_teacher_similarity_mean": 0.7901349425315857, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22230 + }, + { + "epoch": 6.470759383183008, + "flow_matching_loss": 0.01114125563763082, + "grad_norm": 0.09092098474502563, + "learning_rate": 5.162442695619114e-06, + "loss": 0.0131, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007675937889143824, + "rkd_loss_weight": 0.2588252931833267, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3871292918920517, + "rkd_teacher_entropy": 3.7667427539825438, + "rkd_teacher_similarity_mean": 0.7877265989780426, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22240 + }, + { + "epoch": 6.473668897294152, + "flow_matching_loss": 0.01213226211257279, + "grad_norm": 0.09807062149047852, + "learning_rate": 5.149966591651215e-06, + "loss": 0.0138, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006504086311906576, + "rkd_loss_weight": 0.25849194824695587, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3781827986240387, + "rkd_teacher_entropy": 3.7330597400665284, + "rkd_teacher_similarity_mean": 0.7787578403949738, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22250 + }, + { + "epoch": 6.4765784114052956, + "flow_matching_loss": 0.01189876366406679, + "grad_norm": 0.1127638891339302, + "learning_rate": 5.137502456402705e-06, + "loss": 0.0139, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0076589444652199745, + "rkd_loss_weight": 0.2581586092710495, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38184831440448763, + "rkd_teacher_entropy": 3.7424323320388795, + "rkd_teacher_similarity_mean": 0.7829585313796997, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22260 + }, + { + "epoch": 6.479487925516438, + "flow_matching_loss": 0.013691139593720435, + "grad_norm": 0.11294307559728622, + "learning_rate": 5.125050305018683e-06, + "loss": 0.0155, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00707923537120223, + "rkd_loss_weight": 0.25782525837421416, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3785692274570465, + "rkd_teacher_entropy": 3.7355223417282106, + "rkd_teacher_similarity_mean": 0.779225480556488, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22270 + }, + { + "epoch": 6.482397439627582, + "flow_matching_loss": 0.014505922747775912, + "grad_norm": 0.09766807407140732, + "learning_rate": 5.112610152629687e-06, + "loss": 0.0167, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008486737310886384, + "rkd_loss_weight": 0.25749191641807556, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3807820737361908, + "rkd_teacher_entropy": 3.7461132526397707, + "rkd_teacher_similarity_mean": 0.7811914086341858, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22280 + }, + { + "epoch": 6.485306953738726, + "flow_matching_loss": 0.011633002944290639, + "grad_norm": 0.09194035083055496, + "learning_rate": 5.10018201435167e-06, + "loss": 0.0137, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008125982154160738, + "rkd_loss_weight": 0.25715857446193696, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3808481693267822, + "rkd_teacher_entropy": 3.741414260864258, + "rkd_teacher_similarity_mean": 0.7803412854671479, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22290 + }, + { + "epoch": 6.488216467849869, + "flow_matching_loss": 0.014422412402927876, + "grad_norm": 0.12540708482265472, + "learning_rate": 5.087765905286004e-06, + "loss": 0.0166, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008413319336250424, + "rkd_loss_weight": 0.2568252235651016, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37973212003707885, + "rkd_teacher_entropy": 3.728524613380432, + "rkd_teacher_similarity_mean": 0.7788430511951446, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22300 + }, + { + "epoch": 6.491125981961012, + "flow_matching_loss": 0.013379154168069363, + "grad_norm": 0.09478532522916794, + "learning_rate": 5.0753618405194245e-06, + "loss": 0.0154, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007956836465746164, + "rkd_loss_weight": 0.25649188458919525, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38065767884254453, + "rkd_teacher_entropy": 3.740811347961426, + "rkd_teacher_similarity_mean": 0.7800867021083832, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22310 + }, + { + "epoch": 6.494035496072156, + "flow_matching_loss": 0.012612348608672618, + "grad_norm": 0.10322656482458115, + "learning_rate": 5.062969835124042e-06, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0076171484310179945, + "rkd_loss_weight": 0.2561585396528244, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3790205180644989, + "rkd_teacher_entropy": 3.7443438291549684, + "rkd_teacher_similarity_mean": 0.7798996031284332, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22320 + }, + { + "epoch": 6.4969450101833, + "flow_matching_loss": 0.013402412505820393, + "grad_norm": 0.07594092935323715, + "learning_rate": 5.05058990415731e-06, + "loss": 0.0156, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008647770388051867, + "rkd_loss_weight": 0.2558251917362213, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37924252450466156, + "rkd_teacher_entropy": 3.747185754776001, + "rkd_teacher_similarity_mean": 0.7785724401473999, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22330 + }, + { + "epoch": 6.4998545242944425, + "flow_matching_loss": 0.013557683769613504, + "grad_norm": 0.10533206164836884, + "learning_rate": 5.038222062662007e-06, + "loss": 0.0157, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008206399390473962, + "rkd_loss_weight": 0.25549185276031494, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38833406269550325, + "rkd_teacher_entropy": 3.746119999885559, + "rkd_teacher_similarity_mean": 0.7880587637424469, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22340 + }, + { + "epoch": 6.502764038405586, + "flow_matching_loss": 0.01415517800487578, + "grad_norm": 0.10990821570158005, + "learning_rate": 5.025866325666237e-06, + "loss": 0.0162, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007865264732390641, + "rkd_loss_weight": 0.2551585018634796, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.384262353181839, + "rkd_teacher_entropy": 3.748931956291199, + "rkd_teacher_similarity_mean": 0.7832795143127441, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22350 + }, + { + "epoch": 6.50567355251673, + "flow_matching_loss": 0.010504519520327448, + "grad_norm": 0.1001206636428833, + "learning_rate": 5.013522708183383e-06, + "loss": 0.0124, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007606381457298994, + "rkd_loss_weight": 0.254825159907341, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38390191197395324, + "rkd_teacher_entropy": 3.7492154359817507, + "rkd_teacher_similarity_mean": 0.7829306900501252, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22360 + }, + { + "epoch": 6.508583066627873, + "flow_matching_loss": 0.014220924396067858, + "grad_norm": 0.1159403920173645, + "learning_rate": 5.001191225212096e-06, + "loss": 0.0161, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007364943623542786, + "rkd_loss_weight": 0.25449182093143463, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3809934020042419, + "rkd_teacher_entropy": 3.744218277931213, + "rkd_teacher_similarity_mean": 0.7811046361923217, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22370 + }, + { + "epoch": 6.5114925807390165, + "flow_matching_loss": 0.012735582049936056, + "grad_norm": 0.12492982298135757, + "learning_rate": 4.9888718917363e-06, + "loss": 0.015, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00895763784646988, + "rkd_loss_weight": 0.25415846705436707, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3798820227384567, + "rkd_teacher_entropy": 3.7444851398468018, + "rkd_teacher_similarity_mean": 0.7805333197116852, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22380 + }, + { + "epoch": 6.51440209485016, + "flow_matching_loss": 0.01318327160552144, + "grad_norm": 0.08280167728662491, + "learning_rate": 4.9765647227251395e-06, + "loss": 0.0149, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00661499765701592, + "rkd_loss_weight": 0.2538251280784607, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3784418284893036, + "rkd_teacher_entropy": 3.72482545375824, + "rkd_teacher_similarity_mean": 0.7785675346851348, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22390 + }, + { + "epoch": 6.517311608961304, + "flow_matching_loss": 0.013027964532375336, + "grad_norm": 0.11051181703805923, + "learning_rate": 4.964269733132992e-06, + "loss": 0.0155, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009810257935896516, + "rkd_loss_weight": 0.2534917861223221, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.383895942568779, + "rkd_teacher_entropy": 3.7597798347473144, + "rkd_teacher_similarity_mean": 0.7838768064975739, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22400 + }, + { + "epoch": 6.520221123072447, + "flow_matching_loss": 0.014039894193410873, + "grad_norm": 0.11355586349964142, + "learning_rate": 4.951986937899428e-06, + "loss": 0.0158, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00679206452332437, + "rkd_loss_weight": 0.25315843522548676, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38043967485427854, + "rkd_teacher_entropy": 3.7383635520935057, + "rkd_teacher_similarity_mean": 0.7792766630649567, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22410 + }, + { + "epoch": 6.52313063718359, + "flow_matching_loss": 0.014011414162814618, + "grad_norm": 0.10934744030237198, + "learning_rate": 4.9397163519492e-06, + "loss": 0.0158, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007056075986474753, + "rkd_loss_weight": 0.2528250962495804, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37782483398914335, + "rkd_teacher_entropy": 3.735480809211731, + "rkd_teacher_similarity_mean": 0.7772524237632752, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22420 + }, + { + "epoch": 6.526040151294734, + "flow_matching_loss": 0.012143846414983273, + "grad_norm": 0.1352817863225937, + "learning_rate": 4.9274579901922275e-06, + "loss": 0.014, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007422450091689825, + "rkd_loss_weight": 0.25249175131320956, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3888721942901611, + "rkd_teacher_entropy": 3.7690848827362062, + "rkd_teacher_similarity_mean": 0.7895104348659515, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22430 + }, + { + "epoch": 6.528949665405877, + "flow_matching_loss": 0.013304830249398946, + "grad_norm": 0.10289844125509262, + "learning_rate": 4.9152118675235705e-06, + "loss": 0.0153, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007828517071902751, + "rkd_loss_weight": 0.25215840339660645, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3824045419692993, + "rkd_teacher_entropy": 3.737667417526245, + "rkd_teacher_similarity_mean": 0.7835635066032409, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22440 + }, + { + "epoch": 6.531859179517021, + "flow_matching_loss": 0.013649118551984429, + "grad_norm": 0.07228781282901764, + "learning_rate": 4.902977998823432e-06, + "loss": 0.0155, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007208827883005142, + "rkd_loss_weight": 0.2518250644207001, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38550983667373656, + "rkd_teacher_entropy": 3.753341627120972, + "rkd_teacher_similarity_mean": 0.786211347579956, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22450 + }, + { + "epoch": 6.534768693628164, + "flow_matching_loss": 0.013975539477542043, + "grad_norm": 0.07703651487827301, + "learning_rate": 4.890756398957114e-06, + "loss": 0.0156, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006589022884145379, + "rkd_loss_weight": 0.25149171352386473, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38196481168270113, + "rkd_teacher_entropy": 3.742888021469116, + "rkd_teacher_similarity_mean": 0.7824344873428345, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22460 + }, + { + "epoch": 6.537678207739307, + "flow_matching_loss": 0.010848545702174306, + "grad_norm": 0.11495005339384079, + "learning_rate": 4.8785470827750075e-06, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008372307196259499, + "rkd_loss_weight": 0.25115837156772614, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3748028039932251, + "rkd_teacher_entropy": 3.7169750213623045, + "rkd_teacher_similarity_mean": 0.7746335566043854, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22470 + }, + { + "epoch": 6.540587721850451, + "flow_matching_loss": 0.011868776008486747, + "grad_norm": 0.10202347487211227, + "learning_rate": 4.866350065112588e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007041819533333183, + "rkd_loss_weight": 0.25082503259181976, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38105118870735166, + "rkd_teacher_entropy": 3.7424760103225707, + "rkd_teacher_similarity_mean": 0.7814694166183471, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22480 + }, + { + "epoch": 6.543497235961595, + "flow_matching_loss": 0.010604529827833175, + "grad_norm": 0.12406713515520096, + "learning_rate": 4.854165360790381e-06, + "loss": 0.0125, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007399747287854553, + "rkd_loss_weight": 0.2504916787147522, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38364812433719636, + "rkd_teacher_entropy": 3.7507951259613037, + "rkd_teacher_similarity_mean": 0.7838803708553315, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22490 + }, + { + "epoch": 6.546406750072737, + "flow_matching_loss": 0.012302719242870807, + "grad_norm": 0.11279428750276566, + "learning_rate": 4.8419929846139425e-06, + "loss": 0.0142, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007573759788647294, + "rkd_loss_weight": 0.2501583397388458, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3791841059923172, + "rkd_teacher_entropy": 3.7313671827316286, + "rkd_teacher_similarity_mean": 0.7786276042461395, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22500 + }, + { + "epoch": 6.549316264183881, + "flow_matching_loss": 0.013683950854465365, + "grad_norm": 0.12345770746469498, + "learning_rate": 4.829832951373872e-06, + "loss": 0.0159, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008848720137029887, + "rkd_loss_weight": 0.24982499331235886, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3732611119747162, + "rkd_teacher_entropy": 3.7286669492721556, + "rkd_teacher_similarity_mean": 0.7731414616107941, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22510 + }, + { + "epoch": 6.552225778295025, + "flow_matching_loss": 0.013708867970854044, + "grad_norm": 0.09919732809066772, + "learning_rate": 4.817685275845746e-06, + "loss": 0.0155, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007035345444455743, + "rkd_loss_weight": 0.24949165433645248, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.383197158575058, + "rkd_teacher_entropy": 3.7533050775527954, + "rkd_teacher_similarity_mean": 0.783841997385025, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22520 + }, + { + "epoch": 6.5551352924061685, + "flow_matching_loss": 0.010527995694428683, + "grad_norm": 0.09595748782157898, + "learning_rate": 4.805549972790143e-06, + "loss": 0.0126, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008121636928990483, + "rkd_loss_weight": 0.24915830045938492, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37612672746181486, + "rkd_teacher_entropy": 3.737882375717163, + "rkd_teacher_similarity_mean": 0.7762819111347199, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22530 + }, + { + "epoch": 6.558044806517311, + "flow_matching_loss": 0.014078356605023146, + "grad_norm": 0.12228520214557648, + "learning_rate": 4.7934270569525966e-06, + "loss": 0.0158, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007008654391393065, + "rkd_loss_weight": 0.24882496148347855, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38020255863666536, + "rkd_teacher_entropy": 3.7314455270767213, + "rkd_teacher_similarity_mean": 0.7803729176521301, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22540 + }, + { + "epoch": 6.560954320628455, + "flow_matching_loss": 0.013495683483779431, + "grad_norm": 0.09292861074209213, + "learning_rate": 4.7813165430635874e-06, + "loss": 0.0154, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007559434650465846, + "rkd_loss_weight": 0.2484916180372238, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38432934880256653, + "rkd_teacher_entropy": 3.7509095907211303, + "rkd_teacher_similarity_mean": 0.7844566524028778, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22550 + }, + { + "epoch": 6.563863834739599, + "flow_matching_loss": 0.011561538605019451, + "grad_norm": 0.09151482582092285, + "learning_rate": 4.769218445838544e-06, + "loss": 0.0134, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007518402021378279, + "rkd_loss_weight": 0.2481582686305046, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3830284088850021, + "rkd_teacher_entropy": 3.7494733572006225, + "rkd_teacher_similarity_mean": 0.7843711137771606, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22560 + }, + { + "epoch": 6.566773348850742, + "flow_matching_loss": 0.012140765879303217, + "grad_norm": 0.09335058927536011, + "learning_rate": 4.757132779977787e-06, + "loss": 0.014, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007437807228416205, + "rkd_loss_weight": 0.24782492965459824, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3783578336238861, + "rkd_teacher_entropy": 3.740155339241028, + "rkd_teacher_similarity_mean": 0.7793946623802185, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22570 + }, + { + "epoch": 6.569682862961885, + "flow_matching_loss": 0.013661486096680164, + "grad_norm": 0.1014314591884613, + "learning_rate": 4.745059560166542e-06, + "loss": 0.0157, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008119518868625164, + "rkd_loss_weight": 0.24749158322811127, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37828730046749115, + "rkd_teacher_entropy": 3.7438149452209473, + "rkd_teacher_similarity_mean": 0.7801688075065613, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22580 + }, + { + "epoch": 6.572592377073029, + "flow_matching_loss": 0.012295627780258656, + "grad_norm": 0.09959813952445984, + "learning_rate": 4.732998801074911e-06, + "loss": 0.0144, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008371378388255834, + "rkd_loss_weight": 0.2471582368016243, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38695188164711, + "rkd_teacher_entropy": 3.7631392002105715, + "rkd_teacher_similarity_mean": 0.787601113319397, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22590 + }, + { + "epoch": 6.575501891184173, + "flow_matching_loss": 0.011903251707553863, + "grad_norm": 0.10966150462627411, + "learning_rate": 4.7209505173578445e-06, + "loss": 0.0135, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006475010188296437, + "rkd_loss_weight": 0.24682489782571793, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37994368374347687, + "rkd_teacher_entropy": 3.7436580896377563, + "rkd_teacher_similarity_mean": 0.7803733348846436, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22600 + }, + { + "epoch": 6.5784114052953155, + "flow_matching_loss": 0.014090213738381863, + "grad_norm": 0.09220422059297562, + "learning_rate": 4.7089147236551555e-06, + "loss": 0.0164, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00949807963334024, + "rkd_loss_weight": 0.2464915469288826, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.384131532907486, + "rkd_teacher_entropy": 3.755584716796875, + "rkd_teacher_similarity_mean": 0.7844083726406097, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22610 + }, + { + "epoch": 6.581320919406459, + "flow_matching_loss": 0.014433907903730869, + "grad_norm": 0.10763092339038849, + "learning_rate": 4.696891434591464e-06, + "loss": 0.0168, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009692423371598124, + "rkd_loss_weight": 0.246158204972744, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38463796973228453, + "rkd_teacher_entropy": 3.7576294422149656, + "rkd_teacher_similarity_mean": 0.7850043296813964, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22620 + }, + { + "epoch": 6.584230433517602, + "flow_matching_loss": 0.013139937026426195, + "grad_norm": 0.0926392674446106, + "learning_rate": 4.684880664776199e-06, + "loss": 0.0154, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009155562566593289, + "rkd_loss_weight": 0.2458248645067215, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37663582861423495, + "rkd_teacher_entropy": 3.736901593208313, + "rkd_teacher_similarity_mean": 0.7766493558883667, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22630 + }, + { + "epoch": 6.587139947628746, + "flow_matching_loss": 0.011335056042298675, + "grad_norm": 0.08914307504892349, + "learning_rate": 4.6728824288035784e-06, + "loss": 0.0138, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010158546222373842, + "rkd_loss_weight": 0.24549151211977005, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38939182460308075, + "rkd_teacher_entropy": 3.7724687814712525, + "rkd_teacher_similarity_mean": 0.7898285984992981, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22640 + }, + { + "epoch": 6.5900494617398895, + "flow_matching_loss": 0.013352912571281194, + "grad_norm": 0.09971798211336136, + "learning_rate": 4.660896741252586e-06, + "loss": 0.0155, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00869281766936183, + "rkd_loss_weight": 0.24515817314386368, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38778471648693086, + "rkd_teacher_entropy": 3.7710410356521606, + "rkd_teacher_similarity_mean": 0.7887621104717255, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22650 + }, + { + "epoch": 6.592958975851033, + "flow_matching_loss": 0.013927439274266363, + "grad_norm": 0.11554619669914246, + "learning_rate": 4.6489236166869714e-06, + "loss": 0.0155, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0065733951982110735, + "rkd_loss_weight": 0.24482482820749282, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3775597184896469, + "rkd_teacher_entropy": 3.7352564573287963, + "rkd_teacher_similarity_mean": 0.7777588129043579, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22660 + }, + { + "epoch": 6.595868489962176, + "flow_matching_loss": 0.014418090786784887, + "grad_norm": 0.10057772696018219, + "learning_rate": 4.6369630696552e-06, + "loss": 0.0165, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008701793802902102, + "rkd_loss_weight": 0.24449148029088974, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3773873895406723, + "rkd_teacher_entropy": 3.728636884689331, + "rkd_teacher_similarity_mean": 0.7783690094947815, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22670 + }, + { + "epoch": 6.59877800407332, + "flow_matching_loss": 0.0131136579439044, + "grad_norm": 0.09481587260961533, + "learning_rate": 4.625015114690474e-06, + "loss": 0.0152, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00834138048812747, + "rkd_loss_weight": 0.24415814131498337, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38126092255115507, + "rkd_teacher_entropy": 3.749670457839966, + "rkd_teacher_similarity_mean": 0.7817369163036346, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22680 + }, + { + "epoch": 6.601687518184463, + "flow_matching_loss": 0.01427104091271758, + "grad_norm": 0.11026648432016373, + "learning_rate": 4.613079766310682e-06, + "loss": 0.0163, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008382280124351382, + "rkd_loss_weight": 0.24382479339838029, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38013516664505004, + "rkd_teacher_entropy": 3.739501214027405, + "rkd_teacher_similarity_mean": 0.7797503054141999, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22690 + }, + { + "epoch": 6.604597032295606, + "flow_matching_loss": 0.012798397662118077, + "grad_norm": 0.09016290307044983, + "learning_rate": 4.601157039018396e-06, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007065955828875303, + "rkd_loss_weight": 0.24349144846200943, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38023283183574674, + "rkd_teacher_entropy": 3.7511146068573, + "rkd_teacher_similarity_mean": 0.7815090358257294, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22700 + }, + { + "epoch": 6.60750654640675, + "flow_matching_loss": 0.014488753955811262, + "grad_norm": 0.10638568550348282, + "learning_rate": 4.589246947300849e-06, + "loss": 0.0162, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007034706231206655, + "rkd_loss_weight": 0.24315810948610306, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38555861115455625, + "rkd_teacher_entropy": 3.754341387748718, + "rkd_teacher_similarity_mean": 0.7872154295444489, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22710 + }, + { + "epoch": 6.610416060517894, + "flow_matching_loss": 0.014039021730422974, + "grad_norm": 0.13622871041297913, + "learning_rate": 4.577349505629936e-06, + "loss": 0.016, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008051762310788035, + "rkd_loss_weight": 0.2428247570991516, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38459133207798, + "rkd_teacher_entropy": 3.761341166496277, + "rkd_teacher_similarity_mean": 0.786364209651947, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22720 + }, + { + "epoch": 6.613325574629037, + "flow_matching_loss": 0.011625660629943014, + "grad_norm": 0.1089872345328331, + "learning_rate": 4.565464728462165e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008038247376680374, + "rkd_loss_weight": 0.24249141663312912, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38327919244766234, + "rkd_teacher_entropy": 3.7640344858169557, + "rkd_teacher_similarity_mean": 0.785412746667862, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22730 + }, + { + "epoch": 6.61623508874018, + "flow_matching_loss": 0.01370899211615324, + "grad_norm": 0.08745058625936508, + "learning_rate": 4.553592630238663e-06, + "loss": 0.0156, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007721410086378455, + "rkd_loss_weight": 0.24215807467699052, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3766379624605179, + "rkd_teacher_entropy": 3.7186053514480593, + "rkd_teacher_similarity_mean": 0.7772859632968903, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22740 + }, + { + "epoch": 6.619144602851324, + "flow_matching_loss": 0.012859628815203906, + "grad_norm": 0.11519481986761093, + "learning_rate": 4.541733225385146e-06, + "loss": 0.0151, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00925979888997972, + "rkd_loss_weight": 0.24182472378015518, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38457308113574984, + "rkd_teacher_entropy": 3.758592057228088, + "rkd_teacher_similarity_mean": 0.7853247880935669, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22750 + }, + { + "epoch": 6.622054116962468, + "flow_matching_loss": 0.012543501425534486, + "grad_norm": 0.10983368754386902, + "learning_rate": 4.529886528311905e-06, + "loss": 0.0144, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007581485016271472, + "rkd_loss_weight": 0.2414913848042488, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3763586789369583, + "rkd_teacher_entropy": 3.7314807176589966, + "rkd_teacher_similarity_mean": 0.7767627120018006, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22760 + }, + { + "epoch": 6.62496363107361, + "flow_matching_loss": 0.012732807081192732, + "grad_norm": 0.09616762399673462, + "learning_rate": 4.5180525534138045e-06, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007512559834867716, + "rkd_loss_weight": 0.24115803837776184, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37547399699687956, + "rkd_teacher_entropy": 3.715413451194763, + "rkd_teacher_similarity_mean": 0.7766775608062744, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22770 + }, + { + "epoch": 6.627873145184754, + "flow_matching_loss": 0.012771383812651037, + "grad_norm": 0.13737896084785461, + "learning_rate": 4.506231315070235e-06, + "loss": 0.015, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009102489799261093, + "rkd_loss_weight": 0.24082469195127487, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3795696884393692, + "rkd_teacher_entropy": 3.7428233861923217, + "rkd_teacher_similarity_mean": 0.7795490443706512, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22780 + }, + { + "epoch": 6.630782659295898, + "flow_matching_loss": 0.012946492806077004, + "grad_norm": 0.0917893648147583, + "learning_rate": 4.494422827645115e-06, + "loss": 0.0144, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.005989156384021044, + "rkd_loss_weight": 0.2404913529753685, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.389931258559227, + "rkd_teacher_entropy": 3.766772246360779, + "rkd_teacher_similarity_mean": 0.7903881371021271, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22790 + }, + { + "epoch": 6.633692173407041, + "flow_matching_loss": 0.010533162811771035, + "grad_norm": 0.09522905945777893, + "learning_rate": 4.482627105486865e-06, + "loss": 0.0123, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007191645447164774, + "rkd_loss_weight": 0.2401580035686493, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39015727043151854, + "rkd_teacher_entropy": 3.7681796550750732, + "rkd_teacher_similarity_mean": 0.7913016259670258, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22800 + }, + { + "epoch": 6.636601687518184, + "flow_matching_loss": 0.012439016625285149, + "grad_norm": 0.12290854752063751, + "learning_rate": 4.470844162928412e-06, + "loss": 0.0141, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00697288396768272, + "rkd_loss_weight": 0.23982466012239456, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3771596819162369, + "rkd_teacher_entropy": 3.7311580419540404, + "rkd_teacher_similarity_mean": 0.7785924553871155, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22810 + }, + { + "epoch": 6.639511201629328, + "flow_matching_loss": 0.011014192178845406, + "grad_norm": 0.06372597813606262, + "learning_rate": 4.459074014287129e-06, + "loss": 0.0128, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007562747318297624, + "rkd_loss_weight": 0.2394913211464882, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38951043486595155, + "rkd_teacher_entropy": 3.7594120264053346, + "rkd_teacher_similarity_mean": 0.7910479128360748, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22820 + }, + { + "epoch": 6.642420715740471, + "flow_matching_loss": 0.0098007143009454, + "grad_norm": 0.07467318326234818, + "learning_rate": 4.447316673864869e-06, + "loss": 0.0113, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006228005327284336, + "rkd_loss_weight": 0.23915796726942062, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3901080757379532, + "rkd_teacher_entropy": 3.769511318206787, + "rkd_teacher_similarity_mean": 0.792135339975357, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22830 + }, + { + "epoch": 6.645330229851615, + "flow_matching_loss": 0.013076422363519668, + "grad_norm": 0.13853660225868225, + "learning_rate": 4.435572155947903e-06, + "loss": 0.0152, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008758363034576177, + "rkd_loss_weight": 0.23882462829351425, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3711283653974533, + "rkd_teacher_entropy": 3.7234554052352906, + "rkd_teacher_similarity_mean": 0.7722449958324432, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22840 + }, + { + "epoch": 6.648239743962758, + "flow_matching_loss": 0.01235520290210843, + "grad_norm": 0.1320371925830841, + "learning_rate": 4.423840474806929e-06, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009445455949753523, + "rkd_loss_weight": 0.2384912848472595, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38436618745326995, + "rkd_teacher_entropy": 3.755094313621521, + "rkd_teacher_similarity_mean": 0.7845781981945038, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22850 + }, + { + "epoch": 6.651149258073902, + "flow_matching_loss": 0.011964441183954478, + "grad_norm": 0.07747308909893036, + "learning_rate": 4.412121644697043e-06, + "loss": 0.0137, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00730693363584578, + "rkd_loss_weight": 0.2381579354405403, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39194883704185485, + "rkd_teacher_entropy": 3.77952241897583, + "rkd_teacher_similarity_mean": 0.7926759541034698, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22860 + }, + { + "epoch": 6.654058772185045, + "flow_matching_loss": 0.013405908457934856, + "grad_norm": 0.10173448920249939, + "learning_rate": 4.400415679857727e-06, + "loss": 0.0153, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007790099084377289, + "rkd_loss_weight": 0.23782459646463394, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37916160821914674, + "rkd_teacher_entropy": 3.7415202856063843, + "rkd_teacher_similarity_mean": 0.7786977171897889, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22870 + }, + { + "epoch": 6.6569682862961885, + "flow_matching_loss": 0.013560700789093971, + "grad_norm": 0.06778272986412048, + "learning_rate": 4.388722594512838e-06, + "loss": 0.0151, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006686075311154127, + "rkd_loss_weight": 0.23749124854803086, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38100893795490265, + "rkd_teacher_entropy": 3.734580564498901, + "rkd_teacher_similarity_mean": 0.7812192618846894, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22880 + }, + { + "epoch": 6.659877800407332, + "flow_matching_loss": 0.012961510848253966, + "grad_norm": 0.09926754236221313, + "learning_rate": 4.377042402870577e-06, + "loss": 0.0148, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00786311929114163, + "rkd_loss_weight": 0.23715790361166, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38701657950878143, + "rkd_teacher_entropy": 3.759049820899963, + "rkd_teacher_similarity_mean": 0.7866376459598541, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22890 + }, + { + "epoch": 6.662787314518475, + "flow_matching_loss": 0.012586156185716391, + "grad_norm": 0.06751061975955963, + "learning_rate": 4.365375119123476e-06, + "loss": 0.0142, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007000424014404416, + "rkd_loss_weight": 0.23682456463575363, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3768974095582962, + "rkd_teacher_entropy": 3.7293856382369994, + "rkd_teacher_similarity_mean": 0.7765302896499634, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22900 + }, + { + "epoch": 6.665696828629619, + "flow_matching_loss": 0.011905561108142138, + "grad_norm": 0.07341573387384415, + "learning_rate": 4.353720757448383e-06, + "loss": 0.0138, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007844524551182985, + "rkd_loss_weight": 0.2364912137389183, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3840048938989639, + "rkd_teacher_entropy": 3.7446248054504396, + "rkd_teacher_similarity_mean": 0.7841666996479034, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22910 + }, + { + "epoch": 6.6686063427407625, + "flow_matching_loss": 0.013878956902772188, + "grad_norm": 0.13464069366455078, + "learning_rate": 4.342079332006445e-06, + "loss": 0.0154, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006552795181050897, + "rkd_loss_weight": 0.2361578717827797, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3794423371553421, + "rkd_teacher_entropy": 3.744407868385315, + "rkd_teacher_similarity_mean": 0.7793568730354309, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22920 + }, + { + "epoch": 6.671515856851906, + "flow_matching_loss": 0.013280384335666895, + "grad_norm": 0.09655208885669708, + "learning_rate": 4.330450856943101e-06, + "loss": 0.0149, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0069960882421582936, + "rkd_loss_weight": 0.2358245313167572, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37517748177051546, + "rkd_teacher_entropy": 3.722513937950134, + "rkd_teacher_similarity_mean": 0.775333046913147, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22930 + }, + { + "epoch": 6.674425370963049, + "flow_matching_loss": 0.01282816929742694, + "grad_norm": 0.14059630036354065, + "learning_rate": 4.318835346388036e-06, + "loss": 0.0148, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008573314268141985, + "rkd_loss_weight": 0.23549117892980576, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3840047687292099, + "rkd_teacher_entropy": 3.75018150806427, + "rkd_teacher_similarity_mean": 0.7861378610134124, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22940 + }, + { + "epoch": 6.677334885074193, + "flow_matching_loss": 0.011081099836155772, + "grad_norm": 0.1166229397058487, + "learning_rate": 4.307232814455189e-06, + "loss": 0.0128, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0071312270127236845, + "rkd_loss_weight": 0.23515783995389938, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3795600414276123, + "rkd_teacher_entropy": 3.7472482442855837, + "rkd_teacher_similarity_mean": 0.7811908602714539, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22950 + }, + { + "epoch": 6.680244399185336, + "flow_matching_loss": 0.011548895621672273, + "grad_norm": 0.08618544787168503, + "learning_rate": 4.29564327524274e-06, + "loss": 0.0132, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007200027583166957, + "rkd_loss_weight": 0.23482449501752853, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37918283641338346, + "rkd_teacher_entropy": 3.7325855493545532, + "rkd_teacher_similarity_mean": 0.7801104962825776, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22960 + }, + { + "epoch": 6.683153913296479, + "flow_matching_loss": 0.012872222904115915, + "grad_norm": 0.10519473999738693, + "learning_rate": 4.284066742833061e-06, + "loss": 0.0147, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007607488520443439, + "rkd_loss_weight": 0.23449114710092545, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37812291383743285, + "rkd_teacher_entropy": 3.7431411027908323, + "rkd_teacher_similarity_mean": 0.7782415509223938, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22970 + }, + { + "epoch": 6.686063427407623, + "flow_matching_loss": 0.013442297233268619, + "grad_norm": 0.10325375944375992, + "learning_rate": 4.272503231292741e-06, + "loss": 0.0154, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008182231429964304, + "rkd_loss_weight": 0.23415780812501907, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38235801458358765, + "rkd_teacher_entropy": 3.755338358879089, + "rkd_teacher_similarity_mean": 0.7824240624904633, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22980 + }, + { + "epoch": 6.688972941518767, + "flow_matching_loss": 0.013305389322340488, + "grad_norm": 0.1492932289838791, + "learning_rate": 4.260952754672532e-06, + "loss": 0.0153, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00838477499783039, + "rkd_loss_weight": 0.23382445871829988, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3737627983093262, + "rkd_teacher_entropy": 3.728377103805542, + "rkd_teacher_similarity_mean": 0.77476966381073, + "rkd_vlm_temp": 0.10000000149011612, + "step": 22990 + }, + { + "epoch": 6.6918824556299095, + "flow_matching_loss": 0.011749878013506531, + "grad_norm": 0.07564643770456314, + "learning_rate": 4.249415327007353e-06, + "loss": 0.0135, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007560363551601767, + "rkd_loss_weight": 0.23349111527204514, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.380230313539505, + "rkd_teacher_entropy": 3.7454477548599243, + "rkd_teacher_similarity_mean": 0.7813251435756683, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23000 + }, + { + "epoch": 6.694791969741053, + "flow_matching_loss": 0.012619482586160302, + "grad_norm": 0.12082517892122269, + "learning_rate": 4.237890962316264e-06, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007999917143024504, + "rkd_loss_weight": 0.23315777629613876, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3853632271289825, + "rkd_teacher_entropy": 3.7646910905838014, + "rkd_teacher_similarity_mean": 0.787107127904892, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23010 + }, + { + "epoch": 6.697701483852197, + "flow_matching_loss": 0.011561255995184183, + "grad_norm": 0.07331380993127823, + "learning_rate": 4.226379674602452e-06, + "loss": 0.0134, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007978458236902953, + "rkd_loss_weight": 0.2328244239091873, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.381771057844162, + "rkd_teacher_entropy": 3.7477914810180666, + "rkd_teacher_similarity_mean": 0.7833493530750275, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23020 + }, + { + "epoch": 6.70061099796334, + "flow_matching_loss": 0.01378315594047308, + "grad_norm": 0.15026795864105225, + "learning_rate": 4.214881477853227e-06, + "loss": 0.0153, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006548639945685863, + "rkd_loss_weight": 0.23249108344316483, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38337883055210115, + "rkd_teacher_entropy": 3.7604062795639037, + "rkd_teacher_similarity_mean": 0.784507155418396, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23030 + }, + { + "epoch": 6.703520512074483, + "flow_matching_loss": 0.01306982603855431, + "grad_norm": 0.09064676612615585, + "learning_rate": 4.2033963860399745e-06, + "loss": 0.0149, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0077549435663968325, + "rkd_loss_weight": 0.23215774148702623, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3794834017753601, + "rkd_teacher_entropy": 3.7296758890151978, + "rkd_teacher_similarity_mean": 0.7800310730934144, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23040 + }, + { + "epoch": 6.706430026185627, + "flow_matching_loss": 0.014757163263857365, + "grad_norm": 0.12693417072296143, + "learning_rate": 4.1919244131181696e-06, + "loss": 0.0166, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008043133514001965, + "rkd_loss_weight": 0.2318243905901909, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3741744339466095, + "rkd_teacher_entropy": 3.72819721698761, + "rkd_teacher_similarity_mean": 0.7752136766910553, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23050 + }, + { + "epoch": 6.709339540296771, + "flow_matching_loss": 0.011819002497941256, + "grad_norm": 0.09696738421916962, + "learning_rate": 4.180465573027335e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007646825350821018, + "rkd_loss_weight": 0.23149105161428452, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37536950409412384, + "rkd_teacher_entropy": 3.7322296857833863, + "rkd_teacher_similarity_mean": 0.7757737696170807, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23060 + }, + { + "epoch": 6.712249054407914, + "flow_matching_loss": 0.012578786257654429, + "grad_norm": 0.12293358147144318, + "learning_rate": 4.169019879691046e-06, + "loss": 0.0147, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009347415622323751, + "rkd_loss_weight": 0.23115770518779755, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.390140500664711, + "rkd_teacher_entropy": 3.7785464763641357, + "rkd_teacher_similarity_mean": 0.791283518075943, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23070 + }, + { + "epoch": 6.715158568519057, + "flow_matching_loss": 0.012802533246576786, + "grad_norm": 0.11007905751466751, + "learning_rate": 4.157587347016891e-06, + "loss": 0.0148, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008607877604663371, + "rkd_loss_weight": 0.23082435876131058, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3758583128452301, + "rkd_teacher_entropy": 3.7376428127288817, + "rkd_teacher_similarity_mean": 0.7760608732700348, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23080 + }, + { + "epoch": 6.718068082630201, + "flow_matching_loss": 0.011780734360218047, + "grad_norm": 0.08483777940273285, + "learning_rate": 4.14616798889649e-06, + "loss": 0.0135, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0074135139118880035, + "rkd_loss_weight": 0.2304910197854042, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38653615713119505, + "rkd_teacher_entropy": 3.762662148475647, + "rkd_teacher_similarity_mean": 0.7873778641223907, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23090 + }, + { + "epoch": 6.720977596741344, + "flow_matching_loss": 0.013948152400553226, + "grad_norm": 0.1160115897655487, + "learning_rate": 4.134761819205425e-06, + "loss": 0.0155, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006927925581112504, + "rkd_loss_weight": 0.23015766888856887, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38230018615722655, + "rkd_teacher_entropy": 3.7437740325927735, + "rkd_teacher_similarity_mean": 0.783982515335083, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23100 + }, + { + "epoch": 6.723887110852488, + "flow_matching_loss": 0.013725476060062646, + "grad_norm": 0.0807652622461319, + "learning_rate": 4.1233688518032794e-06, + "loss": 0.0155, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007757029868662358, + "rkd_loss_weight": 0.22982432693243027, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39164497554302213, + "rkd_teacher_entropy": 3.7781848430633547, + "rkd_teacher_similarity_mean": 0.7934822857379913, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23110 + }, + { + "epoch": 6.726796624963631, + "flow_matching_loss": 0.012979158712550998, + "grad_norm": 0.10350925475358963, + "learning_rate": 4.1119891005335795e-06, + "loss": 0.0147, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007371560996398329, + "rkd_loss_weight": 0.2294909879565239, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37805190682411194, + "rkd_teacher_entropy": 3.738327145576477, + "rkd_teacher_similarity_mean": 0.7802258372306824, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23120 + }, + { + "epoch": 6.729706139074775, + "flow_matching_loss": 0.012170683033764363, + "grad_norm": 0.09920371323823929, + "learning_rate": 4.1006225792237895e-06, + "loss": 0.0137, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006548859877511859, + "rkd_loss_weight": 0.22915763407945633, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3887647479772568, + "rkd_teacher_entropy": 3.7712897062301636, + "rkd_teacher_similarity_mean": 0.7905812919139862, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23130 + }, + { + "epoch": 6.732615653185918, + "flow_matching_loss": 0.011101084342226387, + "grad_norm": 0.10260821133852005, + "learning_rate": 4.089269301685314e-06, + "loss": 0.0128, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007436524983495474, + "rkd_loss_weight": 0.22882429510354996, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3790803194046021, + "rkd_teacher_entropy": 3.740744185447693, + "rkd_teacher_similarity_mean": 0.7808093130588531, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23140 + }, + { + "epoch": 6.7355251672970615, + "flow_matching_loss": 0.014302877709269524, + "grad_norm": 0.13572318851947784, + "learning_rate": 4.077929281713453e-06, + "loss": 0.0158, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0067685666028410195, + "rkd_loss_weight": 0.22849095165729522, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38971195816993714, + "rkd_teacher_entropy": 3.7770960092544557, + "rkd_teacher_similarity_mean": 0.7908454418182373, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23150 + }, + { + "epoch": 6.738434681408204, + "flow_matching_loss": 0.01363942101597786, + "grad_norm": 0.13167113065719604, + "learning_rate": 4.066602533087398e-06, + "loss": 0.0154, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007836391357704997, + "rkd_loss_weight": 0.22815760225057602, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3803690791130066, + "rkd_teacher_entropy": 3.7418336153030394, + "rkd_teacher_similarity_mean": 0.7820827841758728, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23160 + }, + { + "epoch": 6.741344195519348, + "flow_matching_loss": 0.011316744051873684, + "grad_norm": 0.10076659172773361, + "learning_rate": 4.0552890695702175e-06, + "loss": 0.0129, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007059033447876573, + "rkd_loss_weight": 0.22782426327466965, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3734854221343994, + "rkd_teacher_entropy": 3.719628405570984, + "rkd_teacher_similarity_mean": 0.7744933128356933, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23170 + }, + { + "epoch": 6.744253709630492, + "flow_matching_loss": 0.01267685303464532, + "grad_norm": 0.08092125505208969, + "learning_rate": 4.04398890490883e-06, + "loss": 0.0142, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0065850299317389725, + "rkd_loss_weight": 0.22749091535806656, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3814384967088699, + "rkd_teacher_entropy": 3.7397215127944947, + "rkd_teacher_similarity_mean": 0.7822676539421082, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23180 + }, + { + "epoch": 6.7471632237416355, + "flow_matching_loss": 0.013740817923098803, + "grad_norm": 0.10775212943553925, + "learning_rate": 4.032702052834013e-06, + "loss": 0.0154, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007085090968757868, + "rkd_loss_weight": 0.2271575704216957, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3756613850593567, + "rkd_teacher_entropy": 3.7347797393798827, + "rkd_teacher_similarity_mean": 0.777236258983612, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23190 + }, + { + "epoch": 6.750072737852778, + "flow_matching_loss": 0.011815234599635006, + "grad_norm": 0.09627853333950043, + "learning_rate": 4.021428527060352e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00767274466343224, + "rkd_loss_weight": 0.22682423144578934, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.378879252076149, + "rkd_teacher_entropy": 3.7349647521972655, + "rkd_teacher_similarity_mean": 0.7804480731487274, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23200 + }, + { + "epoch": 6.752982251963922, + "flow_matching_loss": 0.01103606536053121, + "grad_norm": 0.09216948598623276, + "learning_rate": 4.010168341286241e-06, + "loss": 0.0126, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006697496073320508, + "rkd_loss_weight": 0.22649087905883789, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3765504240989685, + "rkd_teacher_entropy": 3.738286781311035, + "rkd_teacher_similarity_mean": 0.777872908115387, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23210 + }, + { + "epoch": 6.755891766075066, + "flow_matching_loss": 0.012766593601554632, + "grad_norm": 0.08945710211992264, + "learning_rate": 3.998921509193868e-06, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007649594824761153, + "rkd_loss_weight": 0.2261575385928154, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3769763112068176, + "rkd_teacher_entropy": 3.733157014846802, + "rkd_teacher_similarity_mean": 0.7779051542282105, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23220 + }, + { + "epoch": 6.7588012801862085, + "flow_matching_loss": 0.011971529200673103, + "grad_norm": 0.10252240300178528, + "learning_rate": 3.987688044449193e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007293166639283299, + "rkd_loss_weight": 0.2258241981267929, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3835733890533447, + "rkd_teacher_entropy": 3.7516312122344972, + "rkd_teacher_similarity_mean": 0.7845924973487854, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23230 + }, + { + "epoch": 6.761710794297352, + "flow_matching_loss": 0.01401589959859848, + "grad_norm": 0.12159574031829834, + "learning_rate": 3.9764679607019385e-06, + "loss": 0.0159, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00849879179149866, + "rkd_loss_weight": 0.22549084573984146, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3796185225248337, + "rkd_teacher_entropy": 3.7458085775375367, + "rkd_teacher_similarity_mean": 0.7811271965503692, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23240 + }, + { + "epoch": 6.764620308408496, + "flow_matching_loss": 0.012014202028512954, + "grad_norm": 0.12246780842542648, + "learning_rate": 3.965261271585561e-06, + "loss": 0.0139, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008590208273380995, + "rkd_loss_weight": 0.2251575067639351, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38277030885219576, + "rkd_teacher_entropy": 3.7572347640991213, + "rkd_teacher_similarity_mean": 0.7833294987678527, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23250 + }, + { + "epoch": 6.76752982251964, + "flow_matching_loss": 0.014788002800196409, + "grad_norm": 0.0793648213148117, + "learning_rate": 3.954067990717251e-06, + "loss": 0.0166, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008059094101190567, + "rkd_loss_weight": 0.22482416182756423, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3736671179533005, + "rkd_teacher_entropy": 3.724301314353943, + "rkd_teacher_similarity_mean": 0.7759536445140839, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23260 + }, + { + "epoch": 6.7704393366307825, + "flow_matching_loss": 0.011174322664737701, + "grad_norm": 0.09008661657571793, + "learning_rate": 3.9428881316978935e-06, + "loss": 0.0128, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007047816552221775, + "rkd_loss_weight": 0.22449081391096115, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3782269686460495, + "rkd_teacher_entropy": 3.746478486061096, + "rkd_teacher_similarity_mean": 0.7796037375926972, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23270 + }, + { + "epoch": 6.773348850741926, + "flow_matching_loss": 0.012832442298531532, + "grad_norm": 0.09571956098079681, + "learning_rate": 3.931721708112075e-06, + "loss": 0.0147, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00818809736520052, + "rkd_loss_weight": 0.22415747493505478, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38127149641513824, + "rkd_teacher_entropy": 3.7398379325866697, + "rkd_teacher_similarity_mean": 0.7822768568992615, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23280 + }, + { + "epoch": 6.77625836485307, + "flow_matching_loss": 0.012080466188490391, + "grad_norm": 0.12863454222679138, + "learning_rate": 3.920568733528047e-06, + "loss": 0.0139, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008087919605895878, + "rkd_loss_weight": 0.22382412552833558, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38263540267944335, + "rkd_teacher_entropy": 3.755079507827759, + "rkd_teacher_similarity_mean": 0.7840518832206727, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23290 + }, + { + "epoch": 6.779167878964213, + "flow_matching_loss": 0.012053468730300665, + "grad_norm": 0.09818295389413834, + "learning_rate": 3.909429221497734e-06, + "loss": 0.0134, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0061103444080799815, + "rkd_loss_weight": 0.22349078208208084, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3846116214990616, + "rkd_teacher_entropy": 3.7494224309921265, + "rkd_teacher_similarity_mean": 0.7853468537330628, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23300 + }, + { + "epoch": 6.782077393075356, + "flow_matching_loss": 0.01332550561055541, + "grad_norm": 0.0975281372666359, + "learning_rate": 3.898303185556689e-06, + "loss": 0.0148, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006418082909658551, + "rkd_loss_weight": 0.22315744310617447, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3835855484008789, + "rkd_teacher_entropy": 3.756677746772766, + "rkd_teacher_similarity_mean": 0.7847942590713501, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23310 + }, + { + "epoch": 6.7849869071865, + "flow_matching_loss": 0.010893031721934676, + "grad_norm": 0.09849993884563446, + "learning_rate": 3.887190639224097e-06, + "loss": 0.0124, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006620741030201316, + "rkd_loss_weight": 0.2228240892291069, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38081252872943877, + "rkd_teacher_entropy": 3.7347397089004515, + "rkd_teacher_similarity_mean": 0.7813574433326721, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23320 + }, + { + "epoch": 6.787896421297643, + "flow_matching_loss": 0.01101532750762999, + "grad_norm": 0.09992172569036484, + "learning_rate": 3.876091596002746e-06, + "loss": 0.0129, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008432715944945813, + "rkd_loss_weight": 0.22249075025320053, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37902618050575254, + "rkd_teacher_entropy": 3.7398956298828123, + "rkd_teacher_similarity_mean": 0.7794402301311493, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23330 + }, + { + "epoch": 6.790805935408787, + "flow_matching_loss": 0.014624278480187058, + "grad_norm": 0.163798987865448, + "learning_rate": 3.865006069379019e-06, + "loss": 0.0167, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009285771939903497, + "rkd_loss_weight": 0.22215740829706193, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.379381051659584, + "rkd_teacher_entropy": 3.7371700048446654, + "rkd_teacher_similarity_mean": 0.7795135855674744, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23340 + }, + { + "epoch": 6.79371544951993, + "flow_matching_loss": 0.011425732774659992, + "grad_norm": 0.13720297813415527, + "learning_rate": 3.853934072822881e-06, + "loss": 0.0131, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007598633505403995, + "rkd_loss_weight": 0.2218240574002266, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3793476462364197, + "rkd_teacher_entropy": 3.743931293487549, + "rkd_teacher_similarity_mean": 0.7801170468330383, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23350 + }, + { + "epoch": 6.796624963631073, + "flow_matching_loss": 0.01255854992195964, + "grad_norm": 0.07753849774599075, + "learning_rate": 3.8428756197878514e-06, + "loss": 0.0143, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007782393647357821, + "rkd_loss_weight": 0.22149071842432022, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3776449054479599, + "rkd_teacher_entropy": 3.7440842151641847, + "rkd_teacher_similarity_mean": 0.7794034600257873, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23360 + }, + { + "epoch": 6.799534477742217, + "flow_matching_loss": 0.013954900298267603, + "grad_norm": 0.11261771619319916, + "learning_rate": 3.831830723710992e-06, + "loss": 0.0156, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007465180521830916, + "rkd_loss_weight": 0.22115737199783325, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38045124113559725, + "rkd_teacher_entropy": 3.7572343587875365, + "rkd_teacher_similarity_mean": 0.7813105642795563, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23370 + }, + { + "epoch": 6.802443991853361, + "flow_matching_loss": 0.012896849913522601, + "grad_norm": 0.07125963270664215, + "learning_rate": 3.820799398012889e-06, + "loss": 0.0143, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006234421581029892, + "rkd_loss_weight": 0.22082402557134628, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37795027196407316, + "rkd_teacher_entropy": 3.728108811378479, + "rkd_teacher_similarity_mean": 0.7788526117801666, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23380 + }, + { + "epoch": 6.805353505964504, + "flow_matching_loss": 0.01266826712526381, + "grad_norm": 0.09105566889047623, + "learning_rate": 3.80978165609765e-06, + "loss": 0.0147, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009067863132804633, + "rkd_loss_weight": 0.2204906865954399, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37308527529239655, + "rkd_teacher_entropy": 3.7102717638015745, + "rkd_teacher_similarity_mean": 0.7730321109294891, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23390 + }, + { + "epoch": 6.808263020075647, + "flow_matching_loss": 0.011132755409926176, + "grad_norm": 0.09256598353385925, + "learning_rate": 3.798777511352864e-06, + "loss": 0.0126, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006675373436883092, + "rkd_loss_weight": 0.22015733569860457, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37392482459545134, + "rkd_teacher_entropy": 3.718214178085327, + "rkd_teacher_similarity_mean": 0.7753673136234284, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23400 + }, + { + "epoch": 6.811172534186791, + "flow_matching_loss": 0.013129530008882284, + "grad_norm": 0.08559148758649826, + "learning_rate": 3.787786977149613e-06, + "loss": 0.0147, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0073548770975321535, + "rkd_loss_weight": 0.21982399374246597, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38169043958187104, + "rkd_teacher_entropy": 3.7422420024871825, + "rkd_teacher_similarity_mean": 0.7837016880512238, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23410 + }, + { + "epoch": 6.8140820482979345, + "flow_matching_loss": 0.013424085592851043, + "grad_norm": 0.08735772222280502, + "learning_rate": 3.776810066842427e-06, + "loss": 0.0152, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007903511170297861, + "rkd_loss_weight": 0.2194906532764435, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3831388890743256, + "rkd_teacher_entropy": 3.755003499984741, + "rkd_teacher_similarity_mean": 0.7847031772136688, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23420 + }, + { + "epoch": 6.816991562409077, + "flow_matching_loss": 0.013524357415735721, + "grad_norm": 0.07941961288452148, + "learning_rate": 3.7658467937692866e-06, + "loss": 0.0153, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008302081981673837, + "rkd_loss_weight": 0.21915730088949203, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38181292712688447, + "rkd_teacher_entropy": 3.7478076696395872, + "rkd_teacher_similarity_mean": 0.7832413971424103, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23430 + }, + { + "epoch": 6.819901076520221, + "flow_matching_loss": 0.011735449638217687, + "grad_norm": 0.08897367119789124, + "learning_rate": 3.7548971712516e-06, + "loss": 0.0134, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007453949889168143, + "rkd_loss_weight": 0.21882396191358566, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3791778087615967, + "rkd_teacher_entropy": 3.736636734008789, + "rkd_teacher_similarity_mean": 0.781164002418518, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23440 + }, + { + "epoch": 6.822810590631365, + "flow_matching_loss": 0.010948620876297355, + "grad_norm": 0.1188398152589798, + "learning_rate": 3.743961212594197e-06, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009243539161980152, + "rkd_loss_weight": 0.21849061846733092, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38178700804710386, + "rkd_teacher_entropy": 3.753226327896118, + "rkd_teacher_similarity_mean": 0.7833451986312866, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23450 + }, + { + "epoch": 6.8257201047425085, + "flow_matching_loss": 0.01386297745630145, + "grad_norm": 0.09310923516750336, + "learning_rate": 3.7330389310852975e-06, + "loss": 0.0153, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006711321184411645, + "rkd_loss_weight": 0.21815726906061172, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38180347383022306, + "rkd_teacher_entropy": 3.7417645692825316, + "rkd_teacher_similarity_mean": 0.7820663928985596, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23460 + }, + { + "epoch": 6.828629618853651, + "flow_matching_loss": 0.014112170692533255, + "grad_norm": 0.12563681602478027, + "learning_rate": 3.7221303399965008e-06, + "loss": 0.0158, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007812835369259119, + "rkd_loss_weight": 0.21782393008470535, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3830070674419403, + "rkd_teacher_entropy": 3.741135501861572, + "rkd_teacher_similarity_mean": 0.7842270314693451, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23470 + }, + { + "epoch": 6.831539132964795, + "flow_matching_loss": 0.013583587296307088, + "grad_norm": 0.1012815311551094, + "learning_rate": 3.711235452582775e-06, + "loss": 0.015, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006508781807497144, + "rkd_loss_weight": 0.21749058216810227, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37611958384513855, + "rkd_teacher_entropy": 3.7326800346374513, + "rkd_teacher_similarity_mean": 0.7778143465518952, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23480 + }, + { + "epoch": 6.834448647075939, + "flow_matching_loss": 0.011288755852729083, + "grad_norm": 0.09343720972537994, + "learning_rate": 3.7003542820824347e-06, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007702915323898196, + "rkd_loss_weight": 0.21715723723173141, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3843393623828888, + "rkd_teacher_entropy": 3.749368691444397, + "rkd_teacher_similarity_mean": 0.7854797065258026, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23490 + }, + { + "epoch": 6.8373581611870815, + "flow_matching_loss": 0.013586571533232927, + "grad_norm": 0.09822569787502289, + "learning_rate": 3.6894868417171252e-06, + "loss": 0.0149, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006173741584643721, + "rkd_loss_weight": 0.21682389825582504, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38056626319885256, + "rkd_teacher_entropy": 3.741936540603638, + "rkd_teacher_similarity_mean": 0.7826789796352387, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23500 + }, + { + "epoch": 6.840267675298225, + "flow_matching_loss": 0.011458507692441344, + "grad_norm": 0.0880507379770279, + "learning_rate": 3.678633144691818e-06, + "loss": 0.0129, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0066522478591650724, + "rkd_loss_weight": 0.2164905458688736, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3818870335817337, + "rkd_teacher_entropy": 3.7458300828933715, + "rkd_teacher_similarity_mean": 0.7835050761699677, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23510 + }, + { + "epoch": 6.843177189409369, + "flow_matching_loss": 0.013948464207351208, + "grad_norm": 0.10363747924566269, + "learning_rate": 3.667793204194776e-06, + "loss": 0.0155, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006954206340014935, + "rkd_loss_weight": 0.2161572054028511, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3787635385990143, + "rkd_teacher_entropy": 3.737066411972046, + "rkd_teacher_similarity_mean": 0.7799793243408203, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23520 + }, + { + "epoch": 6.846086703520512, + "flow_matching_loss": 0.013118347059935332, + "grad_norm": 0.07978163659572601, + "learning_rate": 3.656967033397544e-06, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00695153484120965, + "rkd_loss_weight": 0.2158238634467125, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3834875047206879, + "rkd_teacher_entropy": 3.7477794885635376, + "rkd_teacher_similarity_mean": 0.7845905303955079, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23530 + }, + { + "epoch": 6.8489962176316554, + "flow_matching_loss": 0.014158975984901189, + "grad_norm": 0.09480612725019455, + "learning_rate": 3.646154645454948e-06, + "loss": 0.0156, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006790025066584349, + "rkd_loss_weight": 0.21549051254987717, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.376969701051712, + "rkd_teacher_entropy": 3.7379016876220703, + "rkd_teacher_similarity_mean": 0.7782981514930725, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23540 + }, + { + "epoch": 6.851905731742799, + "flow_matching_loss": 0.013316289009526371, + "grad_norm": 0.06930617988109589, + "learning_rate": 3.6353560535050524e-06, + "loss": 0.015, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007738776737824082, + "rkd_loss_weight": 0.2151571735739708, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3842946648597717, + "rkd_teacher_entropy": 3.7443347454071043, + "rkd_teacher_similarity_mean": 0.7851429939270019, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23550 + }, + { + "epoch": 6.854815245853942, + "flow_matching_loss": 0.013619643449783326, + "grad_norm": 0.10887596011161804, + "learning_rate": 3.624571270669172e-06, + "loss": 0.0151, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006857546372339129, + "rkd_loss_weight": 0.21482382863759994, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38263104259967806, + "rkd_teacher_entropy": 3.7480687141418456, + "rkd_teacher_similarity_mean": 0.7842005968093873, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23560 + }, + { + "epoch": 6.857724759965086, + "flow_matching_loss": 0.013917484879493713, + "grad_norm": 0.10717153549194336, + "learning_rate": 3.6138003100518296e-06, + "loss": 0.0154, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006954974634572864, + "rkd_loss_weight": 0.21449048072099686, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3786046177148819, + "rkd_teacher_entropy": 3.732834076881409, + "rkd_teacher_similarity_mean": 0.7795029580593109, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23570 + }, + { + "epoch": 6.860634274076229, + "flow_matching_loss": 0.012097044661641122, + "grad_norm": 0.11725668609142303, + "learning_rate": 3.6030431847407595e-06, + "loss": 0.0138, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007897907681763173, + "rkd_loss_weight": 0.21415714174509048, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37941636741161344, + "rkd_teacher_entropy": 3.745238208770752, + "rkd_teacher_similarity_mean": 0.7813012897968292, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23580 + }, + { + "epoch": 6.863543788187373, + "flow_matching_loss": 0.012794473208487034, + "grad_norm": 0.11450792104005814, + "learning_rate": 3.592299907806881e-06, + "loss": 0.0143, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007011684635654092, + "rkd_loss_weight": 0.2138237923383713, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.377201572060585, + "rkd_teacher_entropy": 3.739117479324341, + "rkd_teacher_similarity_mean": 0.7790502429008483, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23590 + }, + { + "epoch": 6.866453302298516, + "flow_matching_loss": 0.012691220548003913, + "grad_norm": 0.10177553445100784, + "learning_rate": 3.5815704923042892e-06, + "loss": 0.0141, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006453384691849351, + "rkd_loss_weight": 0.21349044889211655, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38780907094478606, + "rkd_teacher_entropy": 3.7626456737518312, + "rkd_teacher_similarity_mean": 0.7891537606716156, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23600 + }, + { + "epoch": 6.86936281640966, + "flow_matching_loss": 0.01223007244989276, + "grad_norm": 0.11069784313440323, + "learning_rate": 3.5708549512702392e-06, + "loss": 0.0137, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006869550654664635, + "rkd_loss_weight": 0.21315710991621017, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38116687834262847, + "rkd_teacher_entropy": 3.739595055580139, + "rkd_teacher_similarity_mean": 0.7824538290500641, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23610 + }, + { + "epoch": 6.872272330520803, + "flow_matching_loss": 0.013031279761344194, + "grad_norm": 0.1233118548989296, + "learning_rate": 3.5601532977251217e-06, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00726284165866673, + "rkd_loss_weight": 0.2128237560391426, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3880386620759964, + "rkd_teacher_entropy": 3.772171235084534, + "rkd_teacher_similarity_mean": 0.7898020207881927, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23620 + }, + { + "epoch": 6.875181844631946, + "flow_matching_loss": 0.011461562570184469, + "grad_norm": 0.10780677944421768, + "learning_rate": 3.5494655446724544e-06, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00732838329859078, + "rkd_loss_weight": 0.21249041706323624, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38562028110027313, + "rkd_teacher_entropy": 3.753942060470581, + "rkd_teacher_similarity_mean": 0.7867147564888001, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23630 + }, + { + "epoch": 6.87809135874309, + "flow_matching_loss": 0.011611443804576993, + "grad_norm": 0.18159174919128418, + "learning_rate": 3.538791705098866e-06, + "loss": 0.0132, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0075681642163544895, + "rkd_loss_weight": 0.2121570736169815, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38179111778736113, + "rkd_teacher_entropy": 3.7389869928359984, + "rkd_teacher_similarity_mean": 0.7826611697673798, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23640 + }, + { + "epoch": 6.881000872854234, + "flow_matching_loss": 0.01235193470492959, + "grad_norm": 0.11140963435173035, + "learning_rate": 3.5281317919740718e-06, + "loss": 0.0137, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0062403635587543246, + "rkd_loss_weight": 0.2118237242102623, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38806926310062406, + "rkd_teacher_entropy": 3.76102135181427, + "rkd_teacher_similarity_mean": 0.7903694570064544, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23650 + }, + { + "epoch": 6.883910386965377, + "flow_matching_loss": 0.015521541610360146, + "grad_norm": 0.09128298610448837, + "learning_rate": 3.5174858182508825e-06, + "loss": 0.0168, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00607926188968122, + "rkd_loss_weight": 0.21149038523435593, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3871419817209244, + "rkd_teacher_entropy": 3.761278176307678, + "rkd_teacher_similarity_mean": 0.7896155595779419, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23660 + }, + { + "epoch": 6.88681990107652, + "flow_matching_loss": 0.01238162936642766, + "grad_norm": 0.10018537193536758, + "learning_rate": 3.506853796865153e-06, + "loss": 0.0139, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007220873888581991, + "rkd_loss_weight": 0.21115703880786896, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38707602620124815, + "rkd_teacher_entropy": 3.762303113937378, + "rkd_teacher_similarity_mean": 0.7884172558784485, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23670 + }, + { + "epoch": 6.889729415187664, + "flow_matching_loss": 0.011445346847176551, + "grad_norm": 0.11203636974096298, + "learning_rate": 3.496235740735789e-06, + "loss": 0.0131, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00799516486003995, + "rkd_loss_weight": 0.210823692381382, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37826670706272125, + "rkd_teacher_entropy": 3.7395495414733886, + "rkd_teacher_similarity_mean": 0.7798990130424499, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23680 + }, + { + "epoch": 6.892638929298807, + "flow_matching_loss": 0.01201472757384181, + "grad_norm": 0.09766878932714462, + "learning_rate": 3.485631662764738e-06, + "loss": 0.0134, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006637014960870147, + "rkd_loss_weight": 0.21049035340547562, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38788416385650637, + "rkd_teacher_entropy": 3.767730236053467, + "rkd_teacher_similarity_mean": 0.7887253940105439, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23690 + }, + { + "epoch": 6.89554844340995, + "flow_matching_loss": 0.011523917317390442, + "grad_norm": 0.10732553899288177, + "learning_rate": 3.475041575836952e-06, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007138367975130677, + "rkd_loss_weight": 0.21015700250864028, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37929115891456605, + "rkd_teacher_entropy": 3.7450058460235596, + "rkd_teacher_similarity_mean": 0.7800599694252014, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23700 + }, + { + "epoch": 6.898457957521094, + "flow_matching_loss": 0.012042757915332913, + "grad_norm": 0.0930151417851448, + "learning_rate": 3.4644654928203796e-06, + "loss": 0.0134, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0062703189440071585, + "rkd_loss_weight": 0.20982366055250168, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38115242421627044, + "rkd_teacher_entropy": 3.749306559562683, + "rkd_teacher_similarity_mean": 0.7820179045200348, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23710 + }, + { + "epoch": 6.901367471632238, + "flow_matching_loss": 0.013455954706296324, + "grad_norm": 0.09120018035173416, + "learning_rate": 3.4539034265659663e-06, + "loss": 0.0156, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010087090078741312, + "rkd_loss_weight": 0.2094903200864792, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3740623205900192, + "rkd_teacher_entropy": 3.72916784286499, + "rkd_teacher_similarity_mean": 0.7748982071876526, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23720 + }, + { + "epoch": 6.904276985743381, + "flow_matching_loss": 0.012441083136945964, + "grad_norm": 0.102339006960392, + "learning_rate": 3.4433553899076163e-06, + "loss": 0.014, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007351391529664397, + "rkd_loss_weight": 0.20915696769952774, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3804057419300079, + "rkd_teacher_entropy": 3.7398097038269045, + "rkd_teacher_similarity_mean": 0.7819024384021759, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23730 + }, + { + "epoch": 6.907186499854524, + "flow_matching_loss": 0.012569505069404841, + "grad_norm": 0.07637594640254974, + "learning_rate": 3.432821395662189e-06, + "loss": 0.0143, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008152215043082833, + "rkd_loss_weight": 0.20882362872362137, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38255648612976073, + "rkd_teacher_entropy": 3.7425499677658083, + "rkd_teacher_similarity_mean": 0.784493750333786, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23740 + }, + { + "epoch": 6.910096013965668, + "flow_matching_loss": 0.014048501104116439, + "grad_norm": 0.09278383105993271, + "learning_rate": 3.4223014566294785e-06, + "loss": 0.0156, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007423314545303583, + "rkd_loss_weight": 0.2084902837872505, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3812211722135544, + "rkd_teacher_entropy": 3.750631260871887, + "rkd_teacher_similarity_mean": 0.7837764501571656, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23750 + }, + { + "epoch": 6.913005528076811, + "flow_matching_loss": 0.013585527893155813, + "grad_norm": 0.11407074332237244, + "learning_rate": 3.4117955855922e-06, + "loss": 0.0149, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006386577989906073, + "rkd_loss_weight": 0.20815693587064743, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38367114663124086, + "rkd_teacher_entropy": 3.7470186233520506, + "rkd_teacher_similarity_mean": 0.7856448471546174, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23760 + }, + { + "epoch": 6.9159150421879545, + "flow_matching_loss": 0.012351514212787152, + "grad_norm": 0.0957583338022232, + "learning_rate": 3.4013037953159855e-06, + "loss": 0.014, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008136109542101621, + "rkd_loss_weight": 0.20782359689474106, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3831821411848068, + "rkd_teacher_entropy": 3.762167525291443, + "rkd_teacher_similarity_mean": 0.7854302585124969, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23770 + }, + { + "epoch": 6.918824556299098, + "flow_matching_loss": 0.012091860361397266, + "grad_norm": 0.08560098707675934, + "learning_rate": 3.390826098549345e-06, + "loss": 0.0135, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0068278172053396705, + "rkd_loss_weight": 0.20749024897813798, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.378206393122673, + "rkd_teacher_entropy": 3.7431463479995726, + "rkd_teacher_similarity_mean": 0.7802953541278839, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23780 + }, + { + "epoch": 6.921734070410242, + "flow_matching_loss": 0.011827969737350941, + "grad_norm": 0.11087825149297714, + "learning_rate": 3.3803625080236684e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008542798133566976, + "rkd_loss_weight": 0.20715690404176712, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38214340806007385, + "rkd_teacher_entropy": 3.7544870376586914, + "rkd_teacher_similarity_mean": 0.7839450776576996, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23790 + }, + { + "epoch": 6.924643584521385, + "flow_matching_loss": 0.012422195682302118, + "grad_norm": 0.07997659593820572, + "learning_rate": 3.369913036453205e-06, + "loss": 0.0139, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007295929966494441, + "rkd_loss_weight": 0.20682356506586075, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37713133692741396, + "rkd_teacher_entropy": 3.731781077384949, + "rkd_teacher_similarity_mean": 0.7790745079517365, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23800 + }, + { + "epoch": 6.927553098632528, + "flow_matching_loss": 0.013495744206011295, + "grad_norm": 0.12664760649204254, + "learning_rate": 3.3594776965350434e-06, + "loss": 0.0151, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007855021767318249, + "rkd_loss_weight": 0.2064902126789093, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3813059449195862, + "rkd_teacher_entropy": 3.754289436340332, + "rkd_teacher_similarity_mean": 0.7834789395332337, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23810 + }, + { + "epoch": 6.930462612743672, + "flow_matching_loss": 0.013042822107672691, + "grad_norm": 0.11514409631490707, + "learning_rate": 3.34905650094911e-06, + "loss": 0.0144, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006652460293844342, + "rkd_loss_weight": 0.2061568722128868, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38563867211341857, + "rkd_teacher_entropy": 3.757410931587219, + "rkd_teacher_similarity_mean": 0.7876181244850159, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23820 + }, + { + "epoch": 6.933372126854815, + "flow_matching_loss": 0.014748659403994679, + "grad_norm": 0.09517209231853485, + "learning_rate": 3.338649462358145e-06, + "loss": 0.0162, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00698957284912467, + "rkd_loss_weight": 0.2058235302567482, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38000325560569764, + "rkd_teacher_entropy": 3.7402981519699097, + "rkd_teacher_similarity_mean": 0.7829568266868592, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23830 + }, + { + "epoch": 6.936281640965959, + "flow_matching_loss": 0.01225650981068611, + "grad_norm": 0.0895058661699295, + "learning_rate": 3.3282565934076765e-06, + "loss": 0.0139, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008041245769709349, + "rkd_loss_weight": 0.20549017935991287, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38166499733924864, + "rkd_teacher_entropy": 3.759355640411377, + "rkd_teacher_similarity_mean": 0.7845830500125885, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23840 + }, + { + "epoch": 6.939191155077102, + "flow_matching_loss": 0.013680597301572561, + "grad_norm": 0.10661032050848007, + "learning_rate": 3.31787790672602e-06, + "loss": 0.0153, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007688658637925982, + "rkd_loss_weight": 0.2051568403840065, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36732088029384613, + "rkd_teacher_entropy": 3.719183611869812, + "rkd_teacher_similarity_mean": 0.7694434762001038, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23850 + }, + { + "epoch": 6.942100669188245, + "flow_matching_loss": 0.012961806962266564, + "grad_norm": 0.08785957843065262, + "learning_rate": 3.307513414924255e-06, + "loss": 0.0144, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007090135989710688, + "rkd_loss_weight": 0.20482349395751953, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38338465094566343, + "rkd_teacher_entropy": 3.7534162998199463, + "rkd_teacher_similarity_mean": 0.785918003320694, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23860 + }, + { + "epoch": 6.945010183299389, + "flow_matching_loss": 0.011415663454681636, + "grad_norm": 0.10401216149330139, + "learning_rate": 3.2971631305962246e-06, + "loss": 0.0129, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007121427683159709, + "rkd_loss_weight": 0.20449014753103256, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3779472917318344, + "rkd_teacher_entropy": 3.741560626029968, + "rkd_teacher_similarity_mean": 0.7799657583236694, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23870 + }, + { + "epoch": 6.947919697410533, + "flow_matching_loss": 0.010686937533318997, + "grad_norm": 0.08737578988075256, + "learning_rate": 3.2868270663184964e-06, + "loss": 0.0122, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007397160399705172, + "rkd_loss_weight": 0.2041568085551262, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3841910272836685, + "rkd_teacher_entropy": 3.7589470863342287, + "rkd_teacher_similarity_mean": 0.7863159120082855, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23880 + }, + { + "epoch": 6.950829211521675, + "flow_matching_loss": 0.011687392136082053, + "grad_norm": 0.07174022495746613, + "learning_rate": 3.276505234650362e-06, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006562671298161149, + "rkd_loss_weight": 0.203823459148407, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3766987532377243, + "rkd_teacher_entropy": 3.7445033311843874, + "rkd_teacher_similarity_mean": 0.7793184578418731, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23890 + }, + { + "epoch": 6.953738725632819, + "flow_matching_loss": 0.012887128535658121, + "grad_norm": 0.09851603209972382, + "learning_rate": 3.266197648133821e-06, + "loss": 0.0143, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006753779714927077, + "rkd_loss_weight": 0.20349011570215225, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3746067315340042, + "rkd_teacher_entropy": 3.7245826005935667, + "rkd_teacher_similarity_mean": 0.7763565063476563, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23900 + }, + { + "epoch": 6.956648239743963, + "flow_matching_loss": 0.011598311737179757, + "grad_norm": 0.07607049494981766, + "learning_rate": 3.255904319293562e-06, + "loss": 0.0127, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0055507483892142774, + "rkd_loss_weight": 0.20315677672624588, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38778649270534515, + "rkd_teacher_entropy": 3.7617443561553956, + "rkd_teacher_similarity_mean": 0.7901895821094513, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23910 + }, + { + "epoch": 6.9595577538551066, + "flow_matching_loss": 0.010772425122559071, + "grad_norm": 0.10443125665187836, + "learning_rate": 3.2456252606369485e-06, + "loss": 0.0122, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006889711460098624, + "rkd_loss_weight": 0.20282342284917831, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38300151824951173, + "rkd_teacher_entropy": 3.7490406036376953, + "rkd_teacher_similarity_mean": 0.784889942407608, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23920 + }, + { + "epoch": 6.962467267966249, + "flow_matching_loss": 0.01266639712266624, + "grad_norm": 0.09718842804431915, + "learning_rate": 3.235360484654011e-06, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009685920272022485, + "rkd_loss_weight": 0.20249008387327194, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3709290325641632, + "rkd_teacher_entropy": 3.7195124864578246, + "rkd_teacher_similarity_mean": 0.7730159223079681, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23930 + }, + { + "epoch": 6.965376782077393, + "flow_matching_loss": 0.012139493599534035, + "grad_norm": 0.10174751281738281, + "learning_rate": 3.225110003817418e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007022657804191113, + "rkd_loss_weight": 0.2021567404270172, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38126669228076937, + "rkd_teacher_entropy": 3.7493746757507322, + "rkd_teacher_similarity_mean": 0.7829598724842072, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23940 + }, + { + "epoch": 6.968286296188537, + "flow_matching_loss": 0.013680159859359265, + "grad_norm": 0.13525554537773132, + "learning_rate": 3.214873830582467e-06, + "loss": 0.015, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0065231202635914086, + "rkd_loss_weight": 0.201823391020298, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.383462256193161, + "rkd_teacher_entropy": 3.7698866367340087, + "rkd_teacher_similarity_mean": 0.7865497410297394, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23950 + }, + { + "epoch": 6.97119581029968, + "flow_matching_loss": 0.014463658025488258, + "grad_norm": 0.09719022363424301, + "learning_rate": 3.2046519773870803e-06, + "loss": 0.016, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007479506684467197, + "rkd_loss_weight": 0.20149005204439163, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38092707097530365, + "rkd_teacher_entropy": 3.7584715843200684, + "rkd_teacher_similarity_mean": 0.7832080841064453, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23960 + }, + { + "epoch": 6.974105324410823, + "flow_matching_loss": 0.013663134071975946, + "grad_norm": 0.10389577597379684, + "learning_rate": 3.1944444566517654e-06, + "loss": 0.0151, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007379324221983552, + "rkd_loss_weight": 0.20115670412778855, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3779841661453247, + "rkd_teacher_entropy": 3.7454139471054075, + "rkd_teacher_similarity_mean": 0.7802463948726654, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23970 + }, + { + "epoch": 6.977014838521967, + "flow_matching_loss": 0.011598690599203109, + "grad_norm": 0.08153069764375687, + "learning_rate": 3.18425128077963e-06, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007089219847694039, + "rkd_loss_weight": 0.2008233591914177, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3830148994922638, + "rkd_teacher_entropy": 3.748057985305786, + "rkd_teacher_similarity_mean": 0.7851170301437378, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23980 + }, + { + "epoch": 6.979924352633111, + "flow_matching_loss": 0.012108326377347111, + "grad_norm": 0.1011844277381897, + "learning_rate": 3.174072462156342e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007256355974823237, + "rkd_loss_weight": 0.20049002021551132, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3881136178970337, + "rkd_teacher_entropy": 3.770579528808594, + "rkd_teacher_similarity_mean": 0.7903448104858398, + "rkd_vlm_temp": 0.10000000149011612, + "step": 23990 + }, + { + "epoch": 6.9828338667442535, + "flow_matching_loss": 0.011622015433385969, + "grad_norm": 0.09391707181930542, + "learning_rate": 3.163908013150124e-06, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008224899508059024, + "rkd_loss_weight": 0.20015666931867598, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3886081039905548, + "rkd_teacher_entropy": 3.7755069971084594, + "rkd_teacher_similarity_mean": 0.7912979066371918, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24000 + }, + { + "epoch": 6.985743380855397, + "flow_matching_loss": 0.01388529916293919, + "grad_norm": 0.09073863178491592, + "learning_rate": 3.153757946111741e-06, + "loss": 0.0153, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007202300056815147, + "rkd_loss_weight": 0.19982332736253738, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3730397939682007, + "rkd_teacher_entropy": 3.729470944404602, + "rkd_teacher_similarity_mean": 0.7750666558742523, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24010 + }, + { + "epoch": 6.988652894966541, + "flow_matching_loss": 0.012420962750911712, + "grad_norm": 0.07880554348230362, + "learning_rate": 3.1436222733744776e-06, + "loss": 0.0138, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007025262946262956, + "rkd_loss_weight": 0.1994899868965149, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37704954445362093, + "rkd_teacher_entropy": 3.7405802488327025, + "rkd_teacher_similarity_mean": 0.7792537987232209, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24020 + }, + { + "epoch": 6.991562409077684, + "flow_matching_loss": 0.011448951903730631, + "grad_norm": 0.0937567949295044, + "learning_rate": 3.133501007254138e-06, + "loss": 0.0129, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00741081889718771, + "rkd_loss_weight": 0.19915663450956345, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3753785640001297, + "rkd_teacher_entropy": 3.7331709623336793, + "rkd_teacher_similarity_mean": 0.7773591697216033, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24030 + }, + { + "epoch": 6.9944719231888275, + "flow_matching_loss": 0.009918189654126764, + "grad_norm": 0.10198641568422318, + "learning_rate": 3.1233941600490123e-06, + "loss": 0.0113, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006879179179668427, + "rkd_loss_weight": 0.19882329553365707, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38725231885910033, + "rkd_teacher_entropy": 3.7604126214981077, + "rkd_teacher_similarity_mean": 0.7901251554489136, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24040 + }, + { + "epoch": 6.997381437299971, + "flow_matching_loss": 0.012436175672337412, + "grad_norm": 0.15286406874656677, + "learning_rate": 3.113301744039875e-06, + "loss": 0.014, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007708895299583673, + "rkd_loss_weight": 0.19848995059728622, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37920781672000886, + "rkd_teacher_entropy": 3.760540580749512, + "rkd_teacher_similarity_mean": 0.7825496912002563, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24050 + }, + { + "epoch": 7.000290951411114, + "flow_matching_loss": 0.012353783752769232, + "grad_norm": 0.12222149223089218, + "learning_rate": 3.103223771489959e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006094420049339533, + "rkd_loss_weight": 0.19815660268068314, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38099702894687654, + "rkd_teacher_entropy": 3.750609803199768, + "rkd_teacher_similarity_mean": 0.7830971002578735, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24060 + }, + { + "epoch": 7.003200465522258, + "flow_matching_loss": 0.013083796622231603, + "grad_norm": 0.12215430289506912, + "learning_rate": 3.09316025464495e-06, + "loss": 0.0143, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006079593859612942, + "rkd_loss_weight": 0.19782326370477676, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38358030915260316, + "rkd_teacher_entropy": 3.7569415092468263, + "rkd_teacher_similarity_mean": 0.7863927006721496, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24070 + }, + { + "epoch": 7.006109979633401, + "flow_matching_loss": 0.012499267794191837, + "grad_norm": 0.08956348896026611, + "learning_rate": 3.0831112057329775e-06, + "loss": 0.0139, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007049805345013737, + "rkd_loss_weight": 0.19748991429805757, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3828989028930664, + "rkd_teacher_entropy": 3.7549780130386354, + "rkd_teacher_similarity_mean": 0.7857079148292542, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24080 + }, + { + "epoch": 7.009019493744544, + "flow_matching_loss": 0.011426019109785556, + "grad_norm": 0.08794720470905304, + "learning_rate": 3.073076636964579e-06, + "loss": 0.0131, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008441937528550625, + "rkd_loss_weight": 0.19715657085180283, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37597274482250215, + "rkd_teacher_entropy": 3.728035259246826, + "rkd_teacher_similarity_mean": 0.778240954875946, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24090 + }, + { + "epoch": 7.011929007855688, + "flow_matching_loss": 0.013165349513292313, + "grad_norm": 0.13646413385868073, + "learning_rate": 3.063056560532698e-06, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007252021040767431, + "rkd_loss_weight": 0.19682323187589645, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3866230219602585, + "rkd_teacher_entropy": 3.763349437713623, + "rkd_teacher_similarity_mean": 0.7897724270820617, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24100 + }, + { + "epoch": 7.014838521966832, + "flow_matching_loss": 0.01114709279499948, + "grad_norm": 0.08258352428674698, + "learning_rate": 3.0530509886126807e-06, + "loss": 0.0127, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007742922427132726, + "rkd_loss_weight": 0.196489879488945, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3846994012594223, + "rkd_teacher_entropy": 3.7563939809799196, + "rkd_teacher_similarity_mean": 0.7868459641933441, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24110 + }, + { + "epoch": 7.017748036077975, + "flow_matching_loss": 0.01197499018162489, + "grad_norm": 0.1110636442899704, + "learning_rate": 3.043059933362236e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008197730779647827, + "rkd_loss_weight": 0.19615653902292252, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38529573380947113, + "rkd_teacher_entropy": 3.7609210729599, + "rkd_teacher_similarity_mean": 0.7864724397659302, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24120 + }, + { + "epoch": 7.020657550189118, + "flow_matching_loss": 0.013031517807394265, + "grad_norm": 0.10242190212011337, + "learning_rate": 3.0330834069214337e-06, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007276195008307696, + "rkd_loss_weight": 0.19582319706678392, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3761366277933121, + "rkd_teacher_entropy": 3.7402481555938722, + "rkd_teacher_similarity_mean": 0.7780576527118683, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24130 + }, + { + "epoch": 7.023567064300262, + "flow_matching_loss": 0.011110958131030201, + "grad_norm": 0.10411585867404938, + "learning_rate": 3.023121421412703e-06, + "loss": 0.0129, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009250898100435733, + "rkd_loss_weight": 0.19548984616994858, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38202362060546874, + "rkd_teacher_entropy": 3.744604229927063, + "rkd_teacher_similarity_mean": 0.7848250508308411, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24140 + }, + { + "epoch": 7.026476578411406, + "flow_matching_loss": 0.01118717696517706, + "grad_norm": 0.0976228192448616, + "learning_rate": 3.0131739889407896e-06, + "loss": 0.0125, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006781862024217844, + "rkd_loss_weight": 0.1951565071940422, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.373107048869133, + "rkd_teacher_entropy": 3.738135313987732, + "rkd_teacher_similarity_mean": 0.7761294603347778, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24150 + }, + { + "epoch": 7.029386092522548, + "flow_matching_loss": 0.013963990518823266, + "grad_norm": 0.06579551845788956, + "learning_rate": 3.0032411215927645e-06, + "loss": 0.0157, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009084810363128781, + "rkd_loss_weight": 0.19482316076755524, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3761926978826523, + "rkd_teacher_entropy": 3.726281523704529, + "rkd_teacher_similarity_mean": 0.7782343208789826, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24160 + }, + { + "epoch": 7.032295606633692, + "flow_matching_loss": 0.012757793348282576, + "grad_norm": 0.12696240842342377, + "learning_rate": 2.9933228314379975e-06, + "loss": 0.014, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00645528226159513, + "rkd_loss_weight": 0.19448981434106827, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38398362398147584, + "rkd_teacher_entropy": 3.7687795162200928, + "rkd_teacher_similarity_mean": 0.7870869994163513, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24170 + }, + { + "epoch": 7.035205120744836, + "flow_matching_loss": 0.014015696197748184, + "grad_norm": 0.112371526658535, + "learning_rate": 2.9834191305281405e-06, + "loss": 0.0154, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007296952744945883, + "rkd_loss_weight": 0.1941564753651619, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38411096632480624, + "rkd_teacher_entropy": 3.7618808507919312, + "rkd_teacher_similarity_mean": 0.786459106206894, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24180 + }, + { + "epoch": 7.038114634855979, + "flow_matching_loss": 0.010498097725212574, + "grad_norm": 0.09574353694915771, + "learning_rate": 2.973530030897135e-06, + "loss": 0.0119, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007021745480597019, + "rkd_loss_weight": 0.19382312446832656, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37147842049598695, + "rkd_teacher_entropy": 3.729020929336548, + "rkd_teacher_similarity_mean": 0.7736204981803894, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24190 + }, + { + "epoch": 7.041024148967122, + "flow_matching_loss": 0.01398667781613767, + "grad_norm": 0.0957377701997757, + "learning_rate": 2.9636555445611646e-06, + "loss": 0.0158, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009182768315076828, + "rkd_loss_weight": 0.19348978251218796, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3832815855741501, + "rkd_teacher_entropy": 3.750755214691162, + "rkd_teacher_similarity_mean": 0.7854157745838165, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24200 + }, + { + "epoch": 7.043933663078266, + "flow_matching_loss": 0.013278771378099918, + "grad_norm": 0.15074291825294495, + "learning_rate": 2.953795683518659e-06, + "loss": 0.0149, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008417942933738232, + "rkd_loss_weight": 0.19315644353628159, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3796223163604736, + "rkd_teacher_entropy": 3.74172101020813, + "rkd_teacher_similarity_mean": 0.7816882908344269, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24210 + }, + { + "epoch": 7.04684317718941, + "flow_matching_loss": 0.011557407630607485, + "grad_norm": 0.12067998945713043, + "learning_rate": 2.9439504597502815e-06, + "loss": 0.0131, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008196311490610241, + "rkd_loss_weight": 0.19282308965921402, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3775753349065781, + "rkd_teacher_entropy": 3.735390543937683, + "rkd_teacher_similarity_mean": 0.7787594497203827, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24220 + }, + { + "epoch": 7.049752691300553, + "flow_matching_loss": 0.011173372250050306, + "grad_norm": 0.09119869023561478, + "learning_rate": 2.934119885218902e-06, + "loss": 0.0127, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007897358806803822, + "rkd_loss_weight": 0.19248975068330765, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3764561742544174, + "rkd_teacher_entropy": 3.7372962474822997, + "rkd_teacher_similarity_mean": 0.778447276353836, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24230 + }, + { + "epoch": 7.052662205411696, + "flow_matching_loss": 0.010938878124579787, + "grad_norm": 0.12385358661413193, + "learning_rate": 2.9243039718696045e-06, + "loss": 0.0122, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006574618071317673, + "rkd_loss_weight": 0.1921564072370529, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37952726185321806, + "rkd_teacher_entropy": 3.7489739894866942, + "rkd_teacher_similarity_mean": 0.7826929211616516, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24240 + }, + { + "epoch": 7.05557171952284, + "flow_matching_loss": 0.012056991737335921, + "grad_norm": 0.09991125762462616, + "learning_rate": 2.914502731629641e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008170268451794981, + "rkd_loss_weight": 0.1918230578303337, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.382707291841507, + "rkd_teacher_entropy": 3.756605792045593, + "rkd_teacher_similarity_mean": 0.7855329155921936, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24250 + }, + { + "epoch": 7.058481233633983, + "flow_matching_loss": 0.011394788417965174, + "grad_norm": 0.06727956980466843, + "learning_rate": 2.904716176408451e-06, + "loss": 0.0126, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006351246498525143, + "rkd_loss_weight": 0.19148971885442734, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38393411338329314, + "rkd_teacher_entropy": 3.7589589595794677, + "rkd_teacher_similarity_mean": 0.7867552757263183, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24260 + }, + { + "epoch": 7.0613907477451265, + "flow_matching_loss": 0.01191969057545066, + "grad_norm": 0.08428480476140976, + "learning_rate": 2.8949443180976172e-06, + "loss": 0.0132, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006581934774294496, + "rkd_loss_weight": 0.19115637093782425, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38614304959774015, + "rkd_teacher_entropy": 3.7705126285552977, + "rkd_teacher_similarity_mean": 0.7881952166557312, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24270 + }, + { + "epoch": 7.06430026185627, + "flow_matching_loss": 0.01130530545488, + "grad_norm": 0.09397133439779282, + "learning_rate": 2.885187168570864e-06, + "loss": 0.0127, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007167172897607088, + "rkd_loss_weight": 0.1908230260014534, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37877035737037656, + "rkd_teacher_entropy": 3.7262209177017214, + "rkd_teacher_similarity_mean": 0.7805983781814575, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24280 + }, + { + "epoch": 7.067209775967413, + "flow_matching_loss": 0.013548035733401775, + "grad_norm": 0.09709274768829346, + "learning_rate": 2.875444739684058e-06, + "loss": 0.0151, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00809950614348054, + "rkd_loss_weight": 0.19048968702554703, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37489968836307525, + "rkd_teacher_entropy": 3.73705940246582, + "rkd_teacher_similarity_mean": 0.7764600098133088, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24290 + }, + { + "epoch": 7.070119290078557, + "flow_matching_loss": 0.012827383633702994, + "grad_norm": 0.0841253325343132, + "learning_rate": 2.865717043275165e-06, + "loss": 0.014, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006132028484717012, + "rkd_loss_weight": 0.19015633463859558, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.39003941118717195, + "rkd_teacher_entropy": 3.769544792175293, + "rkd_teacher_similarity_mean": 0.7928165197372437, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24300 + }, + { + "epoch": 7.0730288041897005, + "flow_matching_loss": 0.01329714059829712, + "grad_norm": 0.1136249452829361, + "learning_rate": 2.8560040911642528e-06, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0063583249226212505, + "rkd_loss_weight": 0.1898229941725731, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3806077629327774, + "rkd_teacher_entropy": 3.738208055496216, + "rkd_teacher_similarity_mean": 0.782726240158081, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24310 + }, + { + "epoch": 7.075938318300844, + "flow_matching_loss": 0.012250790931284427, + "grad_norm": 0.09186644107103348, + "learning_rate": 2.846305895153472e-06, + "loss": 0.0138, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008020242024213076, + "rkd_loss_weight": 0.1894896522164345, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.373862025141716, + "rkd_teacher_entropy": 3.716590714454651, + "rkd_teacher_similarity_mean": 0.7759554088115692, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24320 + }, + { + "epoch": 7.078847832411987, + "flow_matching_loss": 0.011466969829052686, + "grad_norm": 0.10052520036697388, + "learning_rate": 2.8366224670270487e-06, + "loss": 0.0126, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.005895882425829768, + "rkd_loss_weight": 0.18915630131959915, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.373380184173584, + "rkd_teacher_entropy": 3.7225843191146852, + "rkd_teacher_similarity_mean": 0.7764353692531586, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24330 + }, + { + "epoch": 7.081757346523131, + "flow_matching_loss": 0.012906362442299724, + "grad_norm": 0.11032227426767349, + "learning_rate": 2.8269538185512535e-06, + "loss": 0.0143, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0072261821012943985, + "rkd_loss_weight": 0.18882296234369278, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37769356071949006, + "rkd_teacher_entropy": 3.7305437564849853, + "rkd_teacher_similarity_mean": 0.7797700881958007, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24340 + }, + { + "epoch": 7.084666860634274, + "flow_matching_loss": 0.012980031361803413, + "grad_norm": 0.1008971557021141, + "learning_rate": 2.817299961474417e-06, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008056209748610854, + "rkd_loss_weight": 0.18848961740732192, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38347192406654357, + "rkd_teacher_entropy": 3.7559412002563475, + "rkd_teacher_similarity_mean": 0.7863874733448029, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24350 + }, + { + "epoch": 7.087576374745417, + "flow_matching_loss": 0.012736615817993879, + "grad_norm": 0.10726355761289597, + "learning_rate": 2.8076609075268816e-06, + "loss": 0.014, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006857710983604192, + "rkd_loss_weight": 0.18815626949071884, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3753533959388733, + "rkd_teacher_entropy": 3.727437448501587, + "rkd_teacher_similarity_mean": 0.7784372687339782, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24360 + }, + { + "epoch": 7.090485888856561, + "flow_matching_loss": 0.012237892486155033, + "grad_norm": 0.08076462149620056, + "learning_rate": 2.798036668421006e-06, + "loss": 0.0137, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007641266006976366, + "rkd_loss_weight": 0.18782293051481247, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38417730033397673, + "rkd_teacher_entropy": 3.7491862058639525, + "rkd_teacher_similarity_mean": 0.7866010010242462, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24370 + }, + { + "epoch": 7.093395402967705, + "flow_matching_loss": 0.011764521431177854, + "grad_norm": 0.11649276316165924, + "learning_rate": 2.7884272558511433e-06, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006482640327885747, + "rkd_loss_weight": 0.18748958110809327, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37643325626850127, + "rkd_teacher_entropy": 3.741234564781189, + "rkd_teacher_similarity_mean": 0.7792174220085144, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24380 + }, + { + "epoch": 7.0963049170788475, + "flow_matching_loss": 0.012418563710525632, + "grad_norm": 0.11024600267410278, + "learning_rate": 2.7788326814936456e-06, + "loss": 0.0139, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008028216753154993, + "rkd_loss_weight": 0.18715623766183853, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3769201338291168, + "rkd_teacher_entropy": 3.736274743080139, + "rkd_teacher_similarity_mean": 0.7796147167682648, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24390 + }, + { + "epoch": 7.099214431189991, + "flow_matching_loss": 0.013284822553396225, + "grad_norm": 0.13980980217456818, + "learning_rate": 2.7692529570068176e-06, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006704284623265266, + "rkd_loss_weight": 0.18682289868593216, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38329027593135834, + "rkd_teacher_entropy": 3.7588900089263917, + "rkd_teacher_similarity_mean": 0.7863677382469177, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24400 + }, + { + "epoch": 7.102123945301135, + "flow_matching_loss": 0.015000894572585821, + "grad_norm": 0.10553234815597534, + "learning_rate": 2.759688094030933e-06, + "loss": 0.0166, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008649412635713815, + "rkd_loss_weight": 0.1864895448088646, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37308749556541443, + "rkd_teacher_entropy": 3.7244170188903807, + "rkd_teacher_similarity_mean": 0.7752682864665985, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24410 + }, + { + "epoch": 7.105033459412278, + "flow_matching_loss": 0.012409939244389535, + "grad_norm": 0.1040610820055008, + "learning_rate": 2.750138104188201e-06, + "loss": 0.014, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008375042350962758, + "rkd_loss_weight": 0.18615620583295822, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3809733778238297, + "rkd_teacher_entropy": 3.7384239435195923, + "rkd_teacher_similarity_mean": 0.7828615427017211, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24420 + }, + { + "epoch": 7.107942973523421, + "flow_matching_loss": 0.013391517428681254, + "grad_norm": 0.12214700877666473, + "learning_rate": 2.740602999082759e-06, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006546367984265089, + "rkd_loss_weight": 0.18582286238670348, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3756397426128387, + "rkd_teacher_entropy": 3.7389634370803835, + "rkd_teacher_similarity_mean": 0.7782905340194702, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24430 + }, + { + "epoch": 7.110852487634565, + "flow_matching_loss": 0.013782945089042186, + "grad_norm": 0.1239011362195015, + "learning_rate": 2.731082790300654e-06, + "loss": 0.015, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0062997804023325445, + "rkd_loss_weight": 0.18548951297998428, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37534553110599517, + "rkd_teacher_entropy": 3.739330196380615, + "rkd_teacher_similarity_mean": 0.7779644310474396, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24440 + }, + { + "epoch": 7.113762001745709, + "flow_matching_loss": 0.016444796323776247, + "grad_norm": 0.08916416019201279, + "learning_rate": 2.7215774894098474e-06, + "loss": 0.0178, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007336745131760835, + "rkd_loss_weight": 0.1851561740040779, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37594208121299744, + "rkd_teacher_entropy": 3.738461899757385, + "rkd_teacher_similarity_mean": 0.7784265518188477, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24450 + }, + { + "epoch": 7.116671515856852, + "flow_matching_loss": 0.010296064335852862, + "grad_norm": 0.10536254942417145, + "learning_rate": 2.7120871079601696e-06, + "loss": 0.0115, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006446058675646782, + "rkd_loss_weight": 0.18482282757759094, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3748438239097595, + "rkd_teacher_entropy": 3.7223411083221434, + "rkd_teacher_similarity_mean": 0.7764786958694458, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24460 + }, + { + "epoch": 7.119581029967995, + "flow_matching_loss": 0.010346818342804908, + "grad_norm": 0.08432812988758087, + "learning_rate": 2.7026116574833314e-06, + "loss": 0.0114, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.005943568516522646, + "rkd_loss_weight": 0.18448948115110397, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3759351968765259, + "rkd_teacher_entropy": 3.7249749660491944, + "rkd_teacher_similarity_mean": 0.7774170935153961, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24470 + }, + { + "epoch": 7.122490544079139, + "flow_matching_loss": 0.013336391607299446, + "grad_norm": 0.10152863711118698, + "learning_rate": 2.6931511494928953e-06, + "loss": 0.0147, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007155357161536813, + "rkd_loss_weight": 0.1841561421751976, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37370431423187256, + "rkd_teacher_entropy": 3.723104190826416, + "rkd_teacher_similarity_mean": 0.7751308262348175, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24480 + }, + { + "epoch": 7.125400058190282, + "flow_matching_loss": 0.010223017027601599, + "grad_norm": 0.07600191980600357, + "learning_rate": 2.683705595484275e-06, + "loss": 0.0115, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0071643897332251075, + "rkd_loss_weight": 0.18382279127836226, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3838941305875778, + "rkd_teacher_entropy": 3.7643729209899903, + "rkd_teacher_similarity_mean": 0.786830598115921, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24490 + }, + { + "epoch": 7.128309572301426, + "flow_matching_loss": 0.012733341567218304, + "grad_norm": 0.0846569612622261, + "learning_rate": 2.6742750069347014e-06, + "loss": 0.0143, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00834404956549406, + "rkd_loss_weight": 0.18348944932222366, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37799469828605653, + "rkd_teacher_entropy": 3.7456097841262816, + "rkd_teacher_similarity_mean": 0.7796947240829468, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24500 + }, + { + "epoch": 7.131219086412569, + "flow_matching_loss": 0.012908111745491623, + "grad_norm": 0.12343193590641022, + "learning_rate": 2.664859395303239e-06, + "loss": 0.0141, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006338148657232523, + "rkd_loss_weight": 0.18315610885620118, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3825916200876236, + "rkd_teacher_entropy": 3.7637795209884644, + "rkd_teacher_similarity_mean": 0.7852071166038513, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24510 + }, + { + "epoch": 7.134128600523712, + "flow_matching_loss": 0.012092644674703478, + "grad_norm": 0.0850968062877655, + "learning_rate": 2.655458772030739e-06, + "loss": 0.0135, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007822236837819218, + "rkd_loss_weight": 0.18282275646924973, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38288667500019075, + "rkd_teacher_entropy": 3.7560988664627075, + "rkd_teacher_similarity_mean": 0.7851081490516663, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24520 + }, + { + "epoch": 7.137038114634856, + "flow_matching_loss": 0.011339478334411979, + "grad_norm": 0.0900174230337143, + "learning_rate": 2.6460731485398403e-06, + "loss": 0.0128, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007818849710747599, + "rkd_loss_weight": 0.18248941749334335, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3774486243724823, + "rkd_teacher_entropy": 3.7199081659317015, + "rkd_teacher_similarity_mean": 0.779291695356369, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24530 + }, + { + "epoch": 7.1399476287459995, + "flow_matching_loss": 0.012007723934948445, + "grad_norm": 0.10101868957281113, + "learning_rate": 2.6367025362349695e-06, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006982216984033585, + "rkd_loss_weight": 0.1821560725569725, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3804424852132797, + "rkd_teacher_entropy": 3.7515140771865845, + "rkd_teacher_similarity_mean": 0.7831335306167603, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24540 + }, + { + "epoch": 7.142857142857143, + "flow_matching_loss": 0.012034629425033926, + "grad_norm": 0.0713689774274826, + "learning_rate": 2.6273469465022965e-06, + "loss": 0.0134, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007683690311387182, + "rkd_loss_weight": 0.18182272464036942, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37632564902305604, + "rkd_teacher_entropy": 3.7420266389846804, + "rkd_teacher_similarity_mean": 0.779486232995987, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24550 + }, + { + "epoch": 7.145766656968286, + "flow_matching_loss": 0.01202224539592862, + "grad_norm": 0.13164648413658142, + "learning_rate": 2.618006390709753e-06, + "loss": 0.0134, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007586974976584315, + "rkd_loss_weight": 0.18148938566446304, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3754670351743698, + "rkd_teacher_entropy": 3.734321188926697, + "rkd_teacher_similarity_mean": 0.7778929948806763, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24560 + }, + { + "epoch": 7.14867617107943, + "flow_matching_loss": 0.01178633184172213, + "grad_norm": 0.0775759220123291, + "learning_rate": 2.60868088020699e-06, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006820347439497709, + "rkd_loss_weight": 0.18115603774785996, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38159301578998567, + "rkd_teacher_entropy": 3.751025915145874, + "rkd_teacher_similarity_mean": 0.785516369342804, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24570 + }, + { + "epoch": 7.1515856851905735, + "flow_matching_loss": 0.01167334928177297, + "grad_norm": 0.11160510778427124, + "learning_rate": 2.599370426325382e-06, + "loss": 0.0131, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008065495174378157, + "rkd_loss_weight": 0.1808226928114891, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37681123316287995, + "rkd_teacher_entropy": 3.743682885169983, + "rkd_teacher_similarity_mean": 0.7808907806873322, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24580 + }, + { + "epoch": 7.154495199301716, + "flow_matching_loss": 0.012621591985225677, + "grad_norm": 0.15415135025978088, + "learning_rate": 2.5900750403780093e-06, + "loss": 0.0141, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007969624875113369, + "rkd_loss_weight": 0.18048935383558273, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38265064656734465, + "rkd_teacher_entropy": 3.752289962768555, + "rkd_teacher_similarity_mean": 0.7864420294761658, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24590 + }, + { + "epoch": 7.15740471341286, + "flow_matching_loss": 0.012808417528867721, + "grad_norm": 0.09469406306743622, + "learning_rate": 2.580794733659639e-06, + "loss": 0.014, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006618401221930981, + "rkd_loss_weight": 0.18015600144863128, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3800832390785217, + "rkd_teacher_entropy": 3.7459866046905517, + "rkd_teacher_similarity_mean": 0.7832932353019715, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24600 + }, + { + "epoch": 7.160314227524004, + "flow_matching_loss": 0.01153176063671708, + "grad_norm": 0.08596181869506836, + "learning_rate": 2.5715295174467236e-06, + "loss": 0.0128, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007039882661774755, + "rkd_loss_weight": 0.1798226609826088, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38277179300785064, + "rkd_teacher_entropy": 3.754349374771118, + "rkd_teacher_similarity_mean": 0.7860901892185211, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24610 + }, + { + "epoch": 7.1632237416351465, + "flow_matching_loss": 0.012090206658467651, + "grad_norm": 0.09137695282697678, + "learning_rate": 2.5622794029973746e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008202634146437049, + "rkd_loss_weight": 0.1794893190264702, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3801551222801208, + "rkd_teacher_entropy": 3.748258924484253, + "rkd_teacher_similarity_mean": 0.7836535453796387, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24620 + }, + { + "epoch": 7.16613325574629, + "flow_matching_loss": 0.011607001908123494, + "grad_norm": 0.09264186024665833, + "learning_rate": 2.553044401551351e-06, + "loss": 0.0128, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006657923012971878, + "rkd_loss_weight": 0.17915596812963486, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3792854368686676, + "rkd_teacher_entropy": 3.7429922342300417, + "rkd_teacher_similarity_mean": 0.7824876308441162, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24630 + }, + { + "epoch": 7.169042769857434, + "flow_matching_loss": 0.010899344692006707, + "grad_norm": 0.07696487754583359, + "learning_rate": 2.54382452433005e-06, + "loss": 0.0121, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006645560497418046, + "rkd_loss_weight": 0.17882262915372849, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3912817448377609, + "rkd_teacher_entropy": 3.7823041677474976, + "rkd_teacher_similarity_mean": 0.7946557343006134, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24640 + }, + { + "epoch": 7.171952283968578, + "flow_matching_loss": 0.009819107968360186, + "grad_norm": 0.08043856173753738, + "learning_rate": 2.53461978253649e-06, + "loss": 0.011, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006366408849135041, + "rkd_loss_weight": 0.17848928272724152, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3902051866054535, + "rkd_teacher_entropy": 3.783569097518921, + "rkd_teacher_similarity_mean": 0.7930052936077118, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24650 + }, + { + "epoch": 7.1748617980797205, + "flow_matching_loss": 0.012385064596310258, + "grad_norm": 0.09478919208049774, + "learning_rate": 2.5254301873553065e-06, + "loss": 0.0137, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007227604603394866, + "rkd_loss_weight": 0.17815593630075455, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3765837997198105, + "rkd_teacher_entropy": 3.7359710454940798, + "rkd_teacher_similarity_mean": 0.7784701108932495, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24660 + }, + { + "epoch": 7.177771312190864, + "flow_matching_loss": 0.010813013790175319, + "grad_norm": 0.12865008413791656, + "learning_rate": 2.5162557499527184e-06, + "loss": 0.0119, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006339678494259715, + "rkd_loss_weight": 0.17782259732484818, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38017805218696593, + "rkd_teacher_entropy": 3.7504098176956178, + "rkd_teacher_similarity_mean": 0.7838734865188599, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24670 + }, + { + "epoch": 7.180680826302008, + "flow_matching_loss": 0.010893255192786454, + "grad_norm": 0.10854727774858475, + "learning_rate": 2.5070964814765335e-06, + "loss": 0.0124, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008268548967316746, + "rkd_loss_weight": 0.17748924791812898, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37839062213897706, + "rkd_teacher_entropy": 3.7424361944198608, + "rkd_teacher_similarity_mean": 0.7808538675308228, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24680 + }, + { + "epoch": 7.183590340413151, + "flow_matching_loss": 0.01274982513859868, + "grad_norm": 0.11096342653036118, + "learning_rate": 2.4979523930561295e-06, + "loss": 0.0139, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006503686169162392, + "rkd_loss_weight": 0.17715590447187424, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37507728636264803, + "rkd_teacher_entropy": 3.7404135704040526, + "rkd_teacher_similarity_mean": 0.7780845642089844, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24690 + }, + { + "epoch": 7.186499854524294, + "flow_matching_loss": 0.012158163636922837, + "grad_norm": 0.13427159190177917, + "learning_rate": 2.4888234958024333e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008058407204225659, + "rkd_loss_weight": 0.17682256549596786, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3780748188495636, + "rkd_teacher_entropy": 3.7346734523773195, + "rkd_teacher_similarity_mean": 0.7813831150531769, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24700 + }, + { + "epoch": 7.189409368635438, + "flow_matching_loss": 0.013745489064604044, + "grad_norm": 0.11682642996311188, + "learning_rate": 2.479709800807915e-06, + "loss": 0.0153, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008612199453637004, + "rkd_loss_weight": 0.1764892116189003, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36906901001930237, + "rkd_teacher_entropy": 3.719767427444458, + "rkd_teacher_similarity_mean": 0.7718216359615326, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24710 + }, + { + "epoch": 7.192318882746581, + "flow_matching_loss": 0.010304634971544147, + "grad_norm": 0.08053049445152283, + "learning_rate": 2.4706113191465767e-06, + "loss": 0.0117, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007666313089430332, + "rkd_loss_weight": 0.17615587264299393, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37667115926742556, + "rkd_teacher_entropy": 3.7377652645111086, + "rkd_teacher_similarity_mean": 0.7788771092891693, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24720 + }, + { + "epoch": 7.195228396857725, + "flow_matching_loss": 0.011192236328497529, + "grad_norm": 0.1200626865029335, + "learning_rate": 2.46152806187393e-06, + "loss": 0.0123, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006160815851762891, + "rkd_loss_weight": 0.17582252919673919, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38324780762195587, + "rkd_teacher_entropy": 3.7558791399002076, + "rkd_teacher_similarity_mean": 0.7865875899791718, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24730 + }, + { + "epoch": 7.198137910968868, + "flow_matching_loss": 0.013142428360879421, + "grad_norm": 0.18045109510421753, + "learning_rate": 2.452460040026989e-06, + "loss": 0.0143, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006554967490956187, + "rkd_loss_weight": 0.17548917979002, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37311461567878723, + "rkd_teacher_entropy": 3.7253948926925657, + "rkd_teacher_similarity_mean": 0.7759824514389038, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24740 + }, + { + "epoch": 7.201047425080012, + "flow_matching_loss": 0.012103831442072988, + "grad_norm": 0.08727128803730011, + "learning_rate": 2.4434072646242555e-06, + "loss": 0.0134, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007498540636152029, + "rkd_loss_weight": 0.17515584081411362, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3828128546476364, + "rkd_teacher_entropy": 3.7526267051696776, + "rkd_teacher_similarity_mean": 0.7854439914226532, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24750 + }, + { + "epoch": 7.203956939191155, + "flow_matching_loss": 0.011247678892686963, + "grad_norm": 0.07341539859771729, + "learning_rate": 2.4343697466657015e-06, + "loss": 0.0127, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00832334952428937, + "rkd_loss_weight": 0.17482249289751053, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38066175282001496, + "rkd_teacher_entropy": 3.7425191164016725, + "rkd_teacher_similarity_mean": 0.7822576224803924, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24760 + }, + { + "epoch": 7.206866453302299, + "flow_matching_loss": 0.014066072041168809, + "grad_norm": 0.10852877795696259, + "learning_rate": 2.42534749713277e-06, + "loss": 0.0155, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00837583770044148, + "rkd_loss_weight": 0.17448914796113968, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3757855653762817, + "rkd_teacher_entropy": 3.7350014209747315, + "rkd_teacher_similarity_mean": 0.7779429674148559, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24770 + }, + { + "epoch": 7.209775967413442, + "flow_matching_loss": 0.011740692425519228, + "grad_norm": 0.06935690343379974, + "learning_rate": 2.4163405269883425e-06, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00698430729098618, + "rkd_loss_weight": 0.1741558089852333, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3844777226448059, + "rkd_teacher_entropy": 3.7625970602035523, + "rkd_teacher_similarity_mean": 0.7874682962894439, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24780 + }, + { + "epoch": 7.212685481524585, + "flow_matching_loss": 0.012982643861323594, + "grad_norm": 0.09231150895357132, + "learning_rate": 2.407348847176735e-06, + "loss": 0.0142, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007234722143039107, + "rkd_loss_weight": 0.17382245808839797, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3820429116487503, + "rkd_teacher_entropy": 3.757726311683655, + "rkd_teacher_similarity_mean": 0.7849872231483459, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24790 + }, + { + "epoch": 7.215594995635729, + "flow_matching_loss": 0.012520241597667337, + "grad_norm": 0.090769924223423, + "learning_rate": 2.398372468623688e-06, + "loss": 0.014, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008316804794594646, + "rkd_loss_weight": 0.17348911613225937, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3818795204162598, + "rkd_teacher_entropy": 3.7460500955581666, + "rkd_teacher_similarity_mean": 0.7845864593982697, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24800 + }, + { + "epoch": 7.2185045097468725, + "flow_matching_loss": 0.013075197022408247, + "grad_norm": 0.13608050346374512, + "learning_rate": 2.3894114022363446e-06, + "loss": 0.0142, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006581641407683492, + "rkd_loss_weight": 0.17315577566623688, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3786025494337082, + "rkd_teacher_entropy": 3.7498111009597777, + "rkd_teacher_similarity_mean": 0.7820965826511384, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24810 + }, + { + "epoch": 7.221414023858015, + "flow_matching_loss": 0.012933731637895107, + "grad_norm": 0.11268986016511917, + "learning_rate": 2.3804656589032514e-06, + "loss": 0.0141, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006902591465041042, + "rkd_loss_weight": 0.17282242327928543, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3843523621559143, + "rkd_teacher_entropy": 3.761560082435608, + "rkd_teacher_similarity_mean": 0.7873814880847931, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24820 + }, + { + "epoch": 7.224323537969159, + "flow_matching_loss": 0.013263117242604493, + "grad_norm": 0.0853184387087822, + "learning_rate": 2.3715352494943237e-06, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007409990532323718, + "rkd_loss_weight": 0.17248908430337906, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37442862391471865, + "rkd_teacher_entropy": 3.7209873676300047, + "rkd_teacher_similarity_mean": 0.7770199775695801, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24830 + }, + { + "epoch": 7.227233052080303, + "flow_matching_loss": 0.011568348808214068, + "grad_norm": 0.10046704113483429, + "learning_rate": 2.3626201848608582e-06, + "loss": 0.0128, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0069217098411172625, + "rkd_loss_weight": 0.1721557393670082, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3815658688545227, + "rkd_teacher_entropy": 3.749808430671692, + "rkd_teacher_similarity_mean": 0.7847361445426941, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24840 + }, + { + "epoch": 7.2301425661914465, + "flow_matching_loss": 0.011908556148409844, + "grad_norm": 0.10092496871948242, + "learning_rate": 2.353720475835494e-06, + "loss": 0.0132, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00762426913715899, + "rkd_loss_weight": 0.17182239145040512, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37872556447982786, + "rkd_teacher_entropy": 3.751067852973938, + "rkd_teacher_similarity_mean": 0.7816336512565613, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24850 + }, + { + "epoch": 7.233052080302589, + "flow_matching_loss": 0.016689881216734648, + "grad_norm": 0.11225739866495132, + "learning_rate": 2.344836133232215e-06, + "loss": 0.0179, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0068885980173945425, + "rkd_loss_weight": 0.17148905247449875, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3780813068151474, + "rkd_teacher_entropy": 3.753322887420654, + "rkd_teacher_similarity_mean": 0.7816628634929657, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24860 + }, + { + "epoch": 7.235961594413733, + "flow_matching_loss": 0.011797564756125211, + "grad_norm": 0.10876041650772095, + "learning_rate": 2.335967167846342e-06, + "loss": 0.0131, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007770991045981645, + "rkd_loss_weight": 0.17115570306777955, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3779893845319748, + "rkd_teacher_entropy": 3.74771363735199, + "rkd_teacher_similarity_mean": 0.7815308570861816, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24870 + }, + { + "epoch": 7.238871108524877, + "flow_matching_loss": 0.013096835650503635, + "grad_norm": 0.13943076133728027, + "learning_rate": 2.3271135904544993e-06, + "loss": 0.0144, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007607578858733177, + "rkd_loss_weight": 0.1708223596215248, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37812983989715576, + "rkd_teacher_entropy": 3.737583589553833, + "rkd_teacher_similarity_mean": 0.7805487811565399, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24880 + }, + { + "epoch": 7.2417806226360195, + "flow_matching_loss": 0.01267428919672966, + "grad_norm": 0.1033770889043808, + "learning_rate": 2.3182754118146206e-06, + "loss": 0.014, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007841321453452111, + "rkd_loss_weight": 0.17048902064561844, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3772297501564026, + "rkd_teacher_entropy": 3.7416380882263183, + "rkd_teacher_similarity_mean": 0.779664158821106, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24890 + }, + { + "epoch": 7.244690136747163, + "flow_matching_loss": 0.01129131531342864, + "grad_norm": 0.11133464425802231, + "learning_rate": 2.309452642665923e-06, + "loss": 0.0126, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007451932039111852, + "rkd_loss_weight": 0.170155668258667, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38020209074020384, + "rkd_teacher_entropy": 3.7501025199890137, + "rkd_teacher_similarity_mean": 0.7838683247566223, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24900 + }, + { + "epoch": 7.247599650858307, + "flow_matching_loss": 0.013509171223267913, + "grad_norm": 0.0878431424498558, + "learning_rate": 2.3006452937289065e-06, + "loss": 0.0147, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007088551577180624, + "rkd_loss_weight": 0.1698223277926445, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.389209109544754, + "rkd_teacher_entropy": 3.757659935951233, + "rkd_teacher_similarity_mean": 0.7919379353523255, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24910 + }, + { + "epoch": 7.25050916496945, + "flow_matching_loss": 0.013327189674600958, + "grad_norm": 0.09343649446964264, + "learning_rate": 2.291853375705327e-06, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007752062100917101, + "rkd_loss_weight": 0.1694889858365059, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38136955797672273, + "rkd_teacher_entropy": 3.7567452907562258, + "rkd_teacher_similarity_mean": 0.7843735873699188, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24920 + }, + { + "epoch": 7.2534186790805935, + "flow_matching_loss": 0.013624595012515783, + "grad_norm": 0.1156170442700386, + "learning_rate": 2.283076899278197e-06, + "loss": 0.0148, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007008328661322593, + "rkd_loss_weight": 0.16915563493967056, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.377497261762619, + "rkd_teacher_entropy": 3.7359915733337403, + "rkd_teacher_similarity_mean": 0.7800198912620544, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24930 + }, + { + "epoch": 7.256328193191737, + "flow_matching_loss": 0.013301348872482776, + "grad_norm": 0.1284758448600769, + "learning_rate": 2.274315875111763e-06, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007529431208968163, + "rkd_loss_weight": 0.1688222959637642, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37073449194431307, + "rkd_teacher_entropy": 3.7234962940216065, + "rkd_teacher_similarity_mean": 0.7733130753040314, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24940 + }, + { + "epoch": 7.25923770730288, + "flow_matching_loss": 0.010657722083851694, + "grad_norm": 0.08908815681934357, + "learning_rate": 2.265570313851493e-06, + "loss": 0.0119, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007648871699348092, + "rkd_loss_weight": 0.16848894953727722, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37388087809085846, + "rkd_teacher_entropy": 3.7201011896133425, + "rkd_teacher_similarity_mean": 0.7757645010948181, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24950 + }, + { + "epoch": 7.262147221414024, + "flow_matching_loss": 0.013896826980635523, + "grad_norm": 0.08776012063026428, + "learning_rate": 2.2568402261240685e-06, + "loss": 0.0153, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008339145313948393, + "rkd_loss_weight": 0.16815560311079025, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38042648136615753, + "rkd_teacher_entropy": 3.746688151359558, + "rkd_teacher_similarity_mean": 0.7823402881622314, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24960 + }, + { + "epoch": 7.265056735525167, + "flow_matching_loss": 0.013491247314959764, + "grad_norm": 0.09378748387098312, + "learning_rate": 2.248125622537373e-06, + "loss": 0.0149, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00862513529136777, + "rkd_loss_weight": 0.16782226413488388, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37483988106250765, + "rkd_teacher_entropy": 3.7353569507598876, + "rkd_teacher_similarity_mean": 0.7782086253166198, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24970 + }, + { + "epoch": 7.267966249636311, + "flow_matching_loss": 0.011038178484886885, + "grad_norm": 0.08342298120260239, + "learning_rate": 2.239426513680467e-06, + "loss": 0.0122, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006719258334487677, + "rkd_loss_weight": 0.16748891323804854, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38147229254245757, + "rkd_teacher_entropy": 3.743691635131836, + "rkd_teacher_similarity_mean": 0.7849116325378418, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24980 + }, + { + "epoch": 7.270875763747454, + "flow_matching_loss": 0.011313790176063775, + "grad_norm": 0.09163515269756317, + "learning_rate": 2.2307429101235915e-06, + "loss": 0.0125, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006829454563558102, + "rkd_loss_weight": 0.16715557128190994, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38953508734703063, + "rkd_teacher_entropy": 3.767976927757263, + "rkd_teacher_similarity_mean": 0.7936582744121552, + "rkd_vlm_temp": 0.10000000149011612, + "step": 24990 + }, + { + "epoch": 7.273785277858598, + "flow_matching_loss": 0.014635430742055177, + "grad_norm": 0.093799889087677, + "learning_rate": 2.222074822418145e-06, + "loss": 0.0159, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007517919642850757, + "rkd_loss_weight": 0.16682223230600357, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38040450513362883, + "rkd_teacher_entropy": 3.748318099975586, + "rkd_teacher_similarity_mean": 0.7843678712844848, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25000 + }, + { + "epoch": 7.276694791969741, + "flow_matching_loss": 0.011687023285776376, + "grad_norm": 0.12777671217918396, + "learning_rate": 2.213422261096668e-06, + "loss": 0.0132, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009072733530774712, + "rkd_loss_weight": 0.166488878428936, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3775327205657959, + "rkd_teacher_entropy": 3.7439834594726564, + "rkd_teacher_similarity_mean": 0.7807684123516083, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25010 + }, + { + "epoch": 7.279604306080884, + "flow_matching_loss": 0.011213647527620196, + "grad_norm": 0.11952345818281174, + "learning_rate": 2.204785236672835e-06, + "loss": 0.0124, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0073745307512581345, + "rkd_loss_weight": 0.16615553945302963, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37843462228775027, + "rkd_teacher_entropy": 3.748834562301636, + "rkd_teacher_similarity_mean": 0.7806372940540314, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25020 + }, + { + "epoch": 7.282513820192028, + "flow_matching_loss": 0.01309609622694552, + "grad_norm": 0.06670437008142471, + "learning_rate": 2.1961637596414536e-06, + "loss": 0.0144, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008020079974085093, + "rkd_loss_weight": 0.1658221960067749, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3736039996147156, + "rkd_teacher_entropy": 3.7401276588439942, + "rkd_teacher_similarity_mean": 0.7760354816913605, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25030 + }, + { + "epoch": 7.285423334303172, + "flow_matching_loss": 0.01341675124131143, + "grad_norm": 0.06966359168291092, + "learning_rate": 2.1875578404784275e-06, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007108672708272934, + "rkd_loss_weight": 0.1654888466000557, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3794110924005508, + "rkd_teacher_entropy": 3.73509840965271, + "rkd_teacher_similarity_mean": 0.7826692461967468, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25040 + }, + { + "epoch": 7.288332848414315, + "flow_matching_loss": 0.011267459858208895, + "grad_norm": 0.1116958037018776, + "learning_rate": 2.1789674896407583e-06, + "loss": 0.0126, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007963131787255407, + "rkd_loss_weight": 0.16515550762414932, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3805832743644714, + "rkd_teacher_entropy": 3.7448548555374144, + "rkd_teacher_similarity_mean": 0.7843149542808533, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25050 + }, + { + "epoch": 7.291242362525458, + "flow_matching_loss": 0.01407982110977173, + "grad_norm": 0.11267498135566711, + "learning_rate": 2.170392717566534e-06, + "loss": 0.0152, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00654614195227623, + "rkd_loss_weight": 0.16482215970754624, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3806066304445267, + "rkd_teacher_entropy": 3.750339722633362, + "rkd_teacher_similarity_mean": 0.7840961217880249, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25060 + }, + { + "epoch": 7.294151876636602, + "flow_matching_loss": 0.011343804839998483, + "grad_norm": 0.15230028331279755, + "learning_rate": 2.161833534674909e-06, + "loss": 0.0127, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008464428177103401, + "rkd_loss_weight": 0.16448881477117538, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3766348272562027, + "rkd_teacher_entropy": 3.753529453277588, + "rkd_teacher_similarity_mean": 0.7793148636817933, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25070 + }, + { + "epoch": 7.2970613907477455, + "flow_matching_loss": 0.014959265943616629, + "grad_norm": 0.12772764265537262, + "learning_rate": 2.1532899513661015e-06, + "loss": 0.0161, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006848287675529719, + "rkd_loss_weight": 0.164155475795269, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38121290504932404, + "rkd_teacher_entropy": 3.7523136377334594, + "rkd_teacher_similarity_mean": 0.7847594618797302, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25080 + }, + { + "epoch": 7.299970904858888, + "flow_matching_loss": 0.010807662084698678, + "grad_norm": 0.10951800644397736, + "learning_rate": 2.1447619780213682e-06, + "loss": 0.0119, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00644530295394361, + "rkd_loss_weight": 0.16382212340831756, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3773960679769516, + "rkd_teacher_entropy": 3.7351301193237303, + "rkd_teacher_similarity_mean": 0.7806457042694092, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25090 + }, + { + "epoch": 7.302880418970032, + "flow_matching_loss": 0.011961040226742626, + "grad_norm": 0.13837750256061554, + "learning_rate": 2.136249625003001e-06, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008477855054661632, + "rkd_loss_weight": 0.16348878294229507, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3837982416152954, + "rkd_teacher_entropy": 3.7580647706985473, + "rkd_teacher_similarity_mean": 0.7878703892230987, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25100 + }, + { + "epoch": 7.305789933081176, + "flow_matching_loss": 0.013078694324940443, + "grad_norm": 0.0987599715590477, + "learning_rate": 2.1277529026543107e-06, + "loss": 0.0142, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007142589101567865, + "rkd_loss_weight": 0.1631554424762726, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3814489871263504, + "rkd_teacher_entropy": 3.7420599937438963, + "rkd_teacher_similarity_mean": 0.7851105570793152, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25110 + }, + { + "epoch": 7.308699447192319, + "flow_matching_loss": 0.011533572059124707, + "grad_norm": 0.1107928603887558, + "learning_rate": 2.1192718212996188e-06, + "loss": 0.0128, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0076404337771236895, + "rkd_loss_weight": 0.16282209008932114, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3868133336305618, + "rkd_teacher_entropy": 3.7660142183303833, + "rkd_teacher_similarity_mean": 0.7899029850959778, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25120 + }, + { + "epoch": 7.311608961303462, + "flow_matching_loss": 0.012864224053919316, + "grad_norm": 0.12089799344539642, + "learning_rate": 2.110806391244235e-06, + "loss": 0.0144, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009383043367415666, + "rkd_loss_weight": 0.16248875111341476, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3792435139417648, + "rkd_teacher_entropy": 3.74988112449646, + "rkd_teacher_similarity_mean": 0.7815149605274201, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25130 + }, + { + "epoch": 7.314518475414606, + "flow_matching_loss": 0.013614563224837183, + "grad_norm": 0.11616456508636475, + "learning_rate": 2.1023566227744596e-06, + "loss": 0.0149, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007743428228422999, + "rkd_loss_weight": 0.1621554061770439, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36997610330581665, + "rkd_teacher_entropy": 3.7260599613189695, + "rkd_teacher_similarity_mean": 0.7729720950126648, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25140 + }, + { + "epoch": 7.317427989525749, + "flow_matching_loss": 0.011711859935894608, + "grad_norm": 0.0718507319688797, + "learning_rate": 2.0939225261575563e-06, + "loss": 0.0128, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006691029435023666, + "rkd_loss_weight": 0.16182205826044083, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3837968945503235, + "rkd_teacher_entropy": 3.760466289520264, + "rkd_teacher_similarity_mean": 0.7875659346580506, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25150 + }, + { + "epoch": 7.3203375036368925, + "flow_matching_loss": 0.012179236020892858, + "grad_norm": 0.1511986404657364, + "learning_rate": 2.0855041116417473e-06, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0072357970289886, + "rkd_loss_weight": 0.16148871928453445, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3837065726518631, + "rkd_teacher_entropy": 3.754588794708252, + "rkd_teacher_similarity_mean": 0.7876157104969025, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25160 + }, + { + "epoch": 7.323247017748036, + "flow_matching_loss": 0.01235636817291379, + "grad_norm": 0.10597396641969681, + "learning_rate": 2.0771013894562003e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007925776578485966, + "rkd_loss_weight": 0.16115536987781526, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3814107716083527, + "rkd_teacher_entropy": 3.75331506729126, + "rkd_teacher_similarity_mean": 0.7853796184062958, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25170 + }, + { + "epoch": 7.32615653185918, + "flow_matching_loss": 0.012302409298717976, + "grad_norm": 0.12660261988639832, + "learning_rate": 2.0687143698110113e-06, + "loss": 0.0135, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007357263844460249, + "rkd_loss_weight": 0.16082202643156052, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3828469455242157, + "rkd_teacher_entropy": 3.7555347442626954, + "rkd_teacher_similarity_mean": 0.7865870237350464, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25180 + }, + { + "epoch": 7.329066045970323, + "flow_matching_loss": 0.012491988297551871, + "grad_norm": 0.09607406705617905, + "learning_rate": 2.060343062897208e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006962588708847761, + "rkd_loss_weight": 0.16048868745565414, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3864427775144577, + "rkd_teacher_entropy": 3.768745756149292, + "rkd_teacher_similarity_mean": 0.790697306394577, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25190 + }, + { + "epoch": 7.3319755600814664, + "flow_matching_loss": 0.012319572316482664, + "grad_norm": 0.07388830929994583, + "learning_rate": 2.0519874788867127e-06, + "loss": 0.0135, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007120903069153428, + "rkd_loss_weight": 0.16015533357858658, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38266211152076723, + "rkd_teacher_entropy": 3.7655791521072386, + "rkd_teacher_similarity_mean": 0.7868941962718964, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25200 + }, + { + "epoch": 7.33488507419261, + "flow_matching_loss": 0.012580497376620769, + "grad_norm": 0.11175258457660675, + "learning_rate": 2.043647627932347e-06, + "loss": 0.0139, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008260458847507835, + "rkd_loss_weight": 0.1598219946026802, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37753707766532896, + "rkd_teacher_entropy": 3.7517721891403197, + "rkd_teacher_similarity_mean": 0.7802675426006317, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25210 + }, + { + "epoch": 7.337794588303753, + "flow_matching_loss": 0.012052945280447602, + "grad_norm": 0.09252263605594635, + "learning_rate": 2.0353235201678193e-06, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006120958365499973, + "rkd_loss_weight": 0.1594886526465416, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3881575435400009, + "rkd_teacher_entropy": 3.7723977088928224, + "rkd_teacher_similarity_mean": 0.7924451947212219, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25220 + }, + { + "epoch": 7.340704102414897, + "flow_matching_loss": 0.012380624935030938, + "grad_norm": 0.0822908878326416, + "learning_rate": 2.0270151657076995e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007426821580156684, + "rkd_loss_weight": 0.15915530174970627, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3927491694688797, + "rkd_teacher_entropy": 3.786937403678894, + "rkd_teacher_similarity_mean": 0.7963875949382782, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25230 + }, + { + "epoch": 7.34361361652604, + "flow_matching_loss": 0.010964497365057469, + "grad_norm": 0.09280093759298325, + "learning_rate": 2.0187225746474286e-06, + "loss": 0.0118, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.005532406968995929, + "rkd_loss_weight": 0.1588219627737999, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3821416527032852, + "rkd_teacher_entropy": 3.7564475297927857, + "rkd_teacher_similarity_mean": 0.7864486336708069, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25240 + }, + { + "epoch": 7.346523130637183, + "flow_matching_loss": 0.012764419382438063, + "grad_norm": 0.14786449074745178, + "learning_rate": 2.0104457570632845e-06, + "loss": 0.0138, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006703991768881678, + "rkd_loss_weight": 0.15848861634731293, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38092761039733886, + "rkd_teacher_entropy": 3.7461313009262085, + "rkd_teacher_similarity_mean": 0.7847744226455688, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25250 + }, + { + "epoch": 7.349432644748327, + "flow_matching_loss": 0.012365560559555889, + "grad_norm": 0.07542014122009277, + "learning_rate": 2.0021847230123775e-06, + "loss": 0.0137, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008721322938799857, + "rkd_loss_weight": 0.15815526992082596, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3772660672664642, + "rkd_teacher_entropy": 3.745271325111389, + "rkd_teacher_similarity_mean": 0.7791560888290405, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25260 + }, + { + "epoch": 7.352342158859471, + "flow_matching_loss": 0.01325711589306593, + "grad_norm": 0.08405331522226334, + "learning_rate": 1.9939394825326494e-06, + "loss": 0.0143, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006310522649437189, + "rkd_loss_weight": 0.1578219309449196, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3807956397533417, + "rkd_teacher_entropy": 3.750863480567932, + "rkd_teacher_similarity_mean": 0.7849681138992309, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25270 + }, + { + "epoch": 7.355251672970613, + "flow_matching_loss": 0.01123445937409997, + "grad_norm": 0.15920263528823853, + "learning_rate": 1.9857100456428385e-06, + "loss": 0.0126, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008881018124520778, + "rkd_loss_weight": 0.15748858004808425, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37557098269462585, + "rkd_teacher_entropy": 3.73310067653656, + "rkd_teacher_similarity_mean": 0.7788725435733795, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25280 + }, + { + "epoch": 7.358161187081757, + "flow_matching_loss": 0.011807563854381442, + "grad_norm": 0.06638739258050919, + "learning_rate": 1.977496422342492e-06, + "loss": 0.0131, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008044533943757416, + "rkd_loss_weight": 0.15715523809194565, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37909586131572726, + "rkd_teacher_entropy": 3.7457546949386598, + "rkd_teacher_similarity_mean": 0.7821943819522857, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25290 + }, + { + "epoch": 7.361070701192901, + "flow_matching_loss": 0.011000914825126528, + "grad_norm": 0.08483266830444336, + "learning_rate": 1.9692986226119345e-06, + "loss": 0.0121, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007219727383926511, + "rkd_loss_weight": 0.15682189762592316, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3786453068256378, + "rkd_teacher_entropy": 3.7425981044769285, + "rkd_teacher_similarity_mean": 0.7825079143047333, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25300 + }, + { + "epoch": 7.363980215304045, + "flow_matching_loss": 0.013018511049449445, + "grad_norm": 0.13022376596927643, + "learning_rate": 1.9611166564122658e-06, + "loss": 0.0143, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007885040808469057, + "rkd_loss_weight": 0.1564885452389717, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3817223161458969, + "rkd_teacher_entropy": 3.7542054653167725, + "rkd_teacher_similarity_mean": 0.7857665300369263, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25310 + }, + { + "epoch": 7.366889729415187, + "flow_matching_loss": 0.010685022687539458, + "grad_norm": 0.08739830553531647, + "learning_rate": 1.9529505336853456e-06, + "loss": 0.0117, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006488185189664364, + "rkd_loss_weight": 0.15615520626306534, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3777788817882538, + "rkd_teacher_entropy": 3.7365393161773683, + "rkd_teacher_similarity_mean": 0.7815722405910492, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25320 + }, + { + "epoch": 7.369799243526331, + "flow_matching_loss": 0.012690361589193344, + "grad_norm": 0.12949833273887634, + "learning_rate": 1.9448002643537833e-06, + "loss": 0.0137, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006564751360565424, + "rkd_loss_weight": 0.1558218628168106, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3697139769792557, + "rkd_teacher_entropy": 3.7240753889083864, + "rkd_teacher_similarity_mean": 0.7735828042030335, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25330 + }, + { + "epoch": 7.372708757637475, + "flow_matching_loss": 0.01226863544434309, + "grad_norm": 0.09286848455667496, + "learning_rate": 1.9366658583209205e-06, + "loss": 0.0134, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007116226013749838, + "rkd_loss_weight": 0.1554885134100914, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3818546622991562, + "rkd_teacher_entropy": 3.752535820007324, + "rkd_teacher_similarity_mean": 0.7845243453979492, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25340 + }, + { + "epoch": 7.375618271748618, + "flow_matching_loss": 0.012843988044187427, + "grad_norm": 0.08864776045084, + "learning_rate": 1.9285473254708363e-06, + "loss": 0.0138, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0064755786210298535, + "rkd_loss_weight": 0.15515517443418503, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37812248468399046, + "rkd_teacher_entropy": 3.7285268306732178, + "rkd_teacher_similarity_mean": 0.7806644678115845, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25350 + }, + { + "epoch": 7.378527785859761, + "flow_matching_loss": 0.011580925015732647, + "grad_norm": 0.12908248603343964, + "learning_rate": 1.9204446756683092e-06, + "loss": 0.0128, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0078044144436717035, + "rkd_loss_weight": 0.15482182651758195, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37779817581176756, + "rkd_teacher_entropy": 3.7449259757995605, + "rkd_teacher_similarity_mean": 0.7815929234027863, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25360 + }, + { + "epoch": 7.381437299970905, + "flow_matching_loss": 0.011589470086619258, + "grad_norm": 0.11623083055019379, + "learning_rate": 1.912357918758822e-06, + "loss": 0.0128, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008004846004769205, + "rkd_loss_weight": 0.1544884815812111, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3792956113815308, + "rkd_teacher_entropy": 3.7458529710769652, + "rkd_teacher_similarity_mean": 0.7826601684093475, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25370 + }, + { + "epoch": 7.384346814082049, + "flow_matching_loss": 0.011681467108428479, + "grad_norm": 0.09387537091970444, + "learning_rate": 1.9042870645685478e-06, + "loss": 0.0128, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007364504784345627, + "rkd_loss_weight": 0.15415514260530472, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38269294500350953, + "rkd_teacher_entropy": 3.7587775230407714, + "rkd_teacher_similarity_mean": 0.7861493587493896, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25380 + }, + { + "epoch": 7.387256328193192, + "flow_matching_loss": 0.01188346385024488, + "grad_norm": 0.11973694711923599, + "learning_rate": 1.8962321229043322e-06, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008997277077287435, + "rkd_loss_weight": 0.15382179021835327, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37805465459823606, + "rkd_teacher_entropy": 3.7421393394470215, + "rkd_teacher_similarity_mean": 0.7812664210796356, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25390 + }, + { + "epoch": 7.390165842304335, + "flow_matching_loss": 0.011087476694956421, + "grad_norm": 0.08620588481426239, + "learning_rate": 1.8881931035536943e-06, + "loss": 0.0121, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006500582443550229, + "rkd_loss_weight": 0.15348844975233078, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3747721463441849, + "rkd_teacher_entropy": 3.7330174922943113, + "rkd_teacher_similarity_mean": 0.778230619430542, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25400 + }, + { + "epoch": 7.393075356415479, + "flow_matching_loss": 0.013432449474930764, + "grad_norm": 0.11208293586969376, + "learning_rate": 1.8801700162847952e-06, + "loss": 0.0147, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007968564797192813, + "rkd_loss_weight": 0.15315510779619218, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3824624359607697, + "rkd_teacher_entropy": 3.761274313926697, + "rkd_teacher_similarity_mean": 0.7852746665477752, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25410 + }, + { + "epoch": 7.395984870526622, + "flow_matching_loss": 0.013427660800516605, + "grad_norm": 0.0864509642124176, + "learning_rate": 1.87216287084645e-06, + "loss": 0.0144, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0063555799424648285, + "rkd_loss_weight": 0.15282175689935684, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37195714712142947, + "rkd_teacher_entropy": 3.731191611289978, + "rkd_teacher_similarity_mean": 0.7757470011711121, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25420 + }, + { + "epoch": 7.3988943846377655, + "flow_matching_loss": 0.011895452486351132, + "grad_norm": 0.09075820446014404, + "learning_rate": 1.8641716769680895e-06, + "loss": 0.0131, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007780543528497219, + "rkd_loss_weight": 0.15248841792345047, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3743669271469116, + "rkd_teacher_entropy": 3.729157304763794, + "rkd_teacher_similarity_mean": 0.7777301788330078, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25430 + }, + { + "epoch": 7.401803898748909, + "flow_matching_loss": 0.011181478016078472, + "grad_norm": 0.10958155244588852, + "learning_rate": 1.8561964443597662e-06, + "loss": 0.0126, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009331724047660828, + "rkd_loss_weight": 0.15215507298707961, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3791724979877472, + "rkd_teacher_entropy": 3.7461560487747194, + "rkd_teacher_similarity_mean": 0.7817777991294861, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25440 + }, + { + "epoch": 7.404713412860052, + "flow_matching_loss": 0.013228462543338537, + "grad_norm": 0.09551651030778885, + "learning_rate": 1.848237182712147e-06, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00900837597437203, + "rkd_loss_weight": 0.15182172507047653, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.384621849656105, + "rkd_teacher_entropy": 3.7547317266464235, + "rkd_teacher_similarity_mean": 0.7882687151432037, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25450 + }, + { + "epoch": 7.407622926971196, + "flow_matching_loss": 0.010478701209649444, + "grad_norm": 0.12001924216747284, + "learning_rate": 1.8402939016964788e-06, + "loss": 0.0117, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007867951411753893, + "rkd_loss_weight": 0.15148838609457016, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37282989621162416, + "rkd_teacher_entropy": 3.7194848537445067, + "rkd_teacher_similarity_mean": 0.776693993806839, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25460 + }, + { + "epoch": 7.410532441082339, + "flow_matching_loss": 0.013662335649132728, + "grad_norm": 0.08724372088909149, + "learning_rate": 1.832366610964601e-06, + "loss": 0.0148, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007769674714654684, + "rkd_loss_weight": 0.15115503668785096, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3796834617853165, + "rkd_teacher_entropy": 3.752483677864075, + "rkd_teacher_similarity_mean": 0.78327516913414, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25470 + }, + { + "epoch": 7.413441955193482, + "flow_matching_loss": 0.013364926213398576, + "grad_norm": 0.07939090579748154, + "learning_rate": 1.8244553201489166e-06, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007537377486005426, + "rkd_loss_weight": 0.15082169324159622, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3833269327878952, + "rkd_teacher_entropy": 3.7624451875686646, + "rkd_teacher_similarity_mean": 0.7863671123981476, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25480 + }, + { + "epoch": 7.416351469304626, + "flow_matching_loss": 0.01235917117446661, + "grad_norm": 0.0773521214723587, + "learning_rate": 1.8165600388623882e-06, + "loss": 0.0134, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006602603290230036, + "rkd_loss_weight": 0.15048835426568985, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38277175426483157, + "rkd_teacher_entropy": 3.752322030067444, + "rkd_teacher_similarity_mean": 0.7857486844062805, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25490 + }, + { + "epoch": 7.41926098341577, + "flow_matching_loss": 0.013910516072064638, + "grad_norm": 0.11705436557531357, + "learning_rate": 1.8086807766985331e-06, + "loss": 0.015, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006946576526388526, + "rkd_loss_weight": 0.15015500038862228, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38094034492969514, + "rkd_teacher_entropy": 3.7549222469329835, + "rkd_teacher_similarity_mean": 0.7839981198310852, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25500 + }, + { + "epoch": 7.422170497526913, + "flow_matching_loss": 0.012500174203887581, + "grad_norm": 0.13467459380626678, + "learning_rate": 1.8008175432313946e-06, + "loss": 0.0135, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0068691069260239605, + "rkd_loss_weight": 0.1498216614127159, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3821470081806183, + "rkd_teacher_entropy": 3.7601824522018434, + "rkd_teacher_similarity_mean": 0.7858833312988281, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25510 + }, + { + "epoch": 7.425080011638056, + "flow_matching_loss": 0.012263915222138166, + "grad_norm": 0.08526118099689484, + "learning_rate": 1.7929703480155429e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009041221439838409, + "rkd_loss_weight": 0.14948831796646117, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38244504630565646, + "rkd_teacher_entropy": 3.752846336364746, + "rkd_teacher_similarity_mean": 0.7860873103141784, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25520 + }, + { + "epoch": 7.4279895257492, + "flow_matching_loss": 0.012800006475299597, + "grad_norm": 0.09206122159957886, + "learning_rate": 1.7851392005860595e-06, + "loss": 0.0141, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008428818685933948, + "rkd_loss_weight": 0.14915496855974197, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3796811908483505, + "rkd_teacher_entropy": 3.7457157373428345, + "rkd_teacher_similarity_mean": 0.7831104338169098, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25530 + }, + { + "epoch": 7.430899039860344, + "flow_matching_loss": 0.012610492156818509, + "grad_norm": 0.11647413671016693, + "learning_rate": 1.7773241104585243e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006370233837515116, + "rkd_loss_weight": 0.1488216295838356, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37359611690044403, + "rkd_teacher_entropy": 3.7425769805908202, + "rkd_teacher_similarity_mean": 0.7773543536663056, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25540 + }, + { + "epoch": 7.433808553971486, + "flow_matching_loss": 0.012917431211099029, + "grad_norm": 0.10938823968172073, + "learning_rate": 1.7695250871290108e-06, + "loss": 0.014, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007268603285774588, + "rkd_loss_weight": 0.14848828315734863, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3844142913818359, + "rkd_teacher_entropy": 3.7659033060073854, + "rkd_teacher_similarity_mean": 0.7879562616348267, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25550 + }, + { + "epoch": 7.43671806808263, + "flow_matching_loss": 0.013232497079297901, + "grad_norm": 0.14010140299797058, + "learning_rate": 1.761742140074074e-06, + "loss": 0.0144, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00799658647738397, + "rkd_loss_weight": 0.14815493673086166, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.387367969751358, + "rkd_teacher_entropy": 3.765701508522034, + "rkd_teacher_similarity_mean": 0.7901209712028503, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25560 + }, + { + "epoch": 7.439627582193774, + "flow_matching_loss": 0.013397769629955291, + "grad_norm": 0.10159342736005783, + "learning_rate": 1.7539752787507208e-06, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007489047013223171, + "rkd_loss_weight": 0.1478215977549553, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3740694999694824, + "rkd_teacher_entropy": 3.733632755279541, + "rkd_teacher_similarity_mean": 0.7783754408359528, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25570 + }, + { + "epoch": 7.442537096304917, + "flow_matching_loss": 0.012642479129135608, + "grad_norm": 0.13238072395324707, + "learning_rate": 1.7462245125964233e-06, + "loss": 0.0138, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007747506676241756, + "rkd_loss_weight": 0.14748824685811995, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3818220168352127, + "rkd_teacher_entropy": 3.755189800262451, + "rkd_teacher_similarity_mean": 0.7864359855651856, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25580 + }, + { + "epoch": 7.44544661041606, + "flow_matching_loss": 0.0145421227440238, + "grad_norm": 0.08704285323619843, + "learning_rate": 1.7384898510290931e-06, + "loss": 0.0159, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009132801555097103, + "rkd_loss_weight": 0.14715490490198135, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3710764735937119, + "rkd_teacher_entropy": 3.7322441816329954, + "rkd_teacher_similarity_mean": 0.7741999447345733, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25590 + }, + { + "epoch": 7.448356124527204, + "flow_matching_loss": 0.013456479646265507, + "grad_norm": 0.09847104549407959, + "learning_rate": 1.730771303447069e-06, + "loss": 0.0144, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006122482428327203, + "rkd_loss_weight": 0.14682156443595887, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3779860734939575, + "rkd_teacher_entropy": 3.7520700216293337, + "rkd_teacher_similarity_mean": 0.7821817934513092, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25600 + }, + { + "epoch": 7.451265638638348, + "flow_matching_loss": 0.012204780662432312, + "grad_norm": 0.11445295065641403, + "learning_rate": 1.7230688792291204e-06, + "loss": 0.0134, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008064930979162454, + "rkd_loss_weight": 0.14648821204900742, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3714314877986908, + "rkd_teacher_entropy": 3.7271178722381593, + "rkd_teacher_similarity_mean": 0.7751021444797516, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25610 + }, + { + "epoch": 7.454175152749491, + "flow_matching_loss": 0.012061313167214394, + "grad_norm": 0.08257123082876205, + "learning_rate": 1.7153825877344175e-06, + "loss": 0.0131, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007364188972860575, + "rkd_loss_weight": 0.14615487307310104, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3784128427505493, + "rkd_teacher_entropy": 3.7504598617553713, + "rkd_teacher_similarity_mean": 0.7824440181255341, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25620 + }, + { + "epoch": 7.457084666860634, + "flow_matching_loss": 0.012534487433731556, + "grad_norm": 0.13042445480823517, + "learning_rate": 1.7077124383025278e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00738243842497468, + "rkd_loss_weight": 0.1458215281367302, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3772858649492264, + "rkd_teacher_entropy": 3.741162967681885, + "rkd_teacher_similarity_mean": 0.7810740292072296, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25630 + }, + { + "epoch": 7.459994180971778, + "flow_matching_loss": 0.010434039402753115, + "grad_norm": 0.09263988584280014, + "learning_rate": 1.7000584402534058e-06, + "loss": 0.0115, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007056739274412393, + "rkd_loss_weight": 0.1454881802201271, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3777722388505936, + "rkd_teacher_entropy": 3.7427240133285524, + "rkd_teacher_similarity_mean": 0.7826183676719666, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25640 + }, + { + "epoch": 7.462903695082921, + "flow_matching_loss": 0.012734034331515431, + "grad_norm": 0.09760919213294983, + "learning_rate": 1.6924206028873774e-06, + "loss": 0.014, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008558039320632815, + "rkd_loss_weight": 0.14515484124422073, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3697606444358826, + "rkd_teacher_entropy": 3.727484369277954, + "rkd_teacher_similarity_mean": 0.7744230031967163, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25650 + }, + { + "epoch": 7.4658132091940645, + "flow_matching_loss": 0.01121033849194646, + "grad_norm": 0.1127580925822258, + "learning_rate": 1.6847989354851434e-06, + "loss": 0.0122, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00687768766656518, + "rkd_loss_weight": 0.14482149332761765, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3809154510498047, + "rkd_teacher_entropy": 3.7593483209609984, + "rkd_teacher_similarity_mean": 0.7852942645549774, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25660 + }, + { + "epoch": 7.468722723305208, + "flow_matching_loss": 0.012239850033074618, + "grad_norm": 0.13419604301452637, + "learning_rate": 1.6771934473077443e-06, + "loss": 0.0132, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006974299997091293, + "rkd_loss_weight": 0.1444881483912468, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37910529375076296, + "rkd_teacher_entropy": 3.7443341255187987, + "rkd_teacher_similarity_mean": 0.7837815523147583, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25670 + }, + { + "epoch": 7.471632237416351, + "flow_matching_loss": 0.01308520045131445, + "grad_norm": 0.08959221094846725, + "learning_rate": 1.669604147596563e-06, + "loss": 0.0141, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007108447374776006, + "rkd_loss_weight": 0.14415480941534042, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37491527795791624, + "rkd_teacher_entropy": 3.736230492591858, + "rkd_teacher_similarity_mean": 0.7785612642765045, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25680 + }, + { + "epoch": 7.474541751527495, + "flow_matching_loss": 0.011962279211729765, + "grad_norm": 0.10095176845788956, + "learning_rate": 1.662031045573319e-06, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007060797791928053, + "rkd_loss_weight": 0.14382145702838897, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38871941566467283, + "rkd_teacher_entropy": 3.779118800163269, + "rkd_teacher_similarity_mean": 0.7940221726894379, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25690 + }, + { + "epoch": 7.4774512656386385, + "flow_matching_loss": 0.01250752918422222, + "grad_norm": 0.10309316962957382, + "learning_rate": 1.6544741504400402e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007678029593080282, + "rkd_loss_weight": 0.14348811656236649, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3822268545627594, + "rkd_teacher_entropy": 3.7615596532821653, + "rkd_teacher_similarity_mean": 0.7865983843803406, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25700 + }, + { + "epoch": 7.480360779749782, + "flow_matching_loss": 0.01272489856928587, + "grad_norm": 0.11078282445669174, + "learning_rate": 1.6469334713790712e-06, + "loss": 0.0137, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007087932992726564, + "rkd_loss_weight": 0.14315477460622789, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3887431979179382, + "rkd_teacher_entropy": 3.7803752422332764, + "rkd_teacher_similarity_mean": 0.7920523107051849, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25710 + }, + { + "epoch": 7.483270293860925, + "flow_matching_loss": 0.012118465034291148, + "grad_norm": 0.10460951179265976, + "learning_rate": 1.639409017553048e-06, + "loss": 0.0131, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006910900166258216, + "rkd_loss_weight": 0.14282142370939255, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3768858939409256, + "rkd_teacher_entropy": 3.7398307085037232, + "rkd_teacher_similarity_mean": 0.7803358137607574, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25720 + }, + { + "epoch": 7.486179807972069, + "flow_matching_loss": 0.01050061616115272, + "grad_norm": 0.09815014898777008, + "learning_rate": 1.6319007981048884e-06, + "loss": 0.0116, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007621697196736932, + "rkd_loss_weight": 0.14248808473348618, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3792613685131073, + "rkd_teacher_entropy": 3.74576199054718, + "rkd_teacher_similarity_mean": 0.7832050025463104, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25730 + }, + { + "epoch": 7.489089322083212, + "flow_matching_loss": 0.013268234953284264, + "grad_norm": 0.09528958052396774, + "learning_rate": 1.6244088221577896e-06, + "loss": 0.0142, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006843577045947314, + "rkd_loss_weight": 0.1421547383069992, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38231375217437746, + "rkd_teacher_entropy": 3.750823903083801, + "rkd_teacher_similarity_mean": 0.786252623796463, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25740 + }, + { + "epoch": 7.491998836194355, + "flow_matching_loss": 0.011101347813382744, + "grad_norm": 0.09208687394857407, + "learning_rate": 1.6169330988152081e-06, + "loss": 0.0121, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00725034112110734, + "rkd_loss_weight": 0.14182139188051224, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3814693599939346, + "rkd_teacher_entropy": 3.7545852184295656, + "rkd_teacher_similarity_mean": 0.7851521849632264, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25750 + }, + { + "epoch": 7.494908350305499, + "flow_matching_loss": 0.011224688263610005, + "grad_norm": 0.09473301470279694, + "learning_rate": 1.6094736371608475e-06, + "loss": 0.0122, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00691928849555552, + "rkd_loss_weight": 0.14148805290460587, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38324550688266756, + "rkd_teacher_entropy": 3.7664973020553587, + "rkd_teacher_similarity_mean": 0.7878724932670593, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25760 + }, + { + "epoch": 7.497817864416643, + "flow_matching_loss": 0.012284513330087066, + "grad_norm": 0.09423227608203888, + "learning_rate": 1.602030446258666e-06, + "loss": 0.0132, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006578296469524503, + "rkd_loss_weight": 0.14115470349788667, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3816424340009689, + "rkd_teacher_entropy": 3.755685067176819, + "rkd_teacher_similarity_mean": 0.7857130110263825, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25770 + }, + { + "epoch": 7.5007273785277855, + "flow_matching_loss": 0.011914146784693002, + "grad_norm": 0.0973547101020813, + "learning_rate": 1.5946035351528388e-06, + "loss": 0.0129, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0071005330886691805, + "rkd_loss_weight": 0.14082136005163193, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37691947519779206, + "rkd_teacher_entropy": 3.7332266330718995, + "rkd_teacher_similarity_mean": 0.7807425081729888, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25780 + }, + { + "epoch": 7.503636892638929, + "flow_matching_loss": 0.014154416695237159, + "grad_norm": 0.09367144107818604, + "learning_rate": 1.5871929128677621e-06, + "loss": 0.0153, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007845992967486382, + "rkd_loss_weight": 0.14048802107572556, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3854264259338379, + "rkd_teacher_entropy": 3.7592517852783205, + "rkd_teacher_similarity_mean": 0.7899465799331665, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25790 + }, + { + "epoch": 7.506546406750073, + "flow_matching_loss": 0.011676508467644453, + "grad_norm": 0.08385199308395386, + "learning_rate": 1.5797985884080412e-06, + "loss": 0.0127, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00728757930919528, + "rkd_loss_weight": 0.140154667198658, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.383277440071106, + "rkd_teacher_entropy": 3.764460492134094, + "rkd_teacher_similarity_mean": 0.7878405511379242, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25800 + }, + { + "epoch": 7.509455920861216, + "flow_matching_loss": 0.011889638612046838, + "grad_norm": 0.10314717143774033, + "learning_rate": 1.572420570758476e-06, + "loss": 0.0129, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007157112937420607, + "rkd_loss_weight": 0.13982132822275162, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37648005187511446, + "rkd_teacher_entropy": 3.744246315956116, + "rkd_teacher_similarity_mean": 0.7812500059604645, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25810 + }, + { + "epoch": 7.512365434972359, + "flow_matching_loss": 0.012260918226093053, + "grad_norm": 0.09524758905172348, + "learning_rate": 1.5650588688840567e-06, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0073963765986263755, + "rkd_loss_weight": 0.13948798477649688, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37268355786800383, + "rkd_teacher_entropy": 3.7319231033325195, + "rkd_teacher_similarity_mean": 0.7778992474079132, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25820 + }, + { + "epoch": 7.515274949083503, + "flow_matching_loss": 0.01105263871140778, + "grad_norm": 0.08447790890932083, + "learning_rate": 1.5577134917299413e-06, + "loss": 0.012, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006961054960265755, + "rkd_loss_weight": 0.13915463536977768, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.374344402551651, + "rkd_teacher_entropy": 3.740295958518982, + "rkd_teacher_similarity_mean": 0.7780421495437622, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25830 + }, + { + "epoch": 7.518184463194647, + "flow_matching_loss": 0.013574021030217409, + "grad_norm": 0.15028999745845795, + "learning_rate": 1.5503844482214618e-06, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007215849170461297, + "rkd_loss_weight": 0.1388212963938713, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38128367960453036, + "rkd_teacher_entropy": 3.7655558586120605, + "rkd_teacher_similarity_mean": 0.7868770122528076, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25840 + }, + { + "epoch": 7.52109397730579, + "flow_matching_loss": 0.011400113487616182, + "grad_norm": 0.08463289588689804, + "learning_rate": 1.5430717472640932e-06, + "loss": 0.0124, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007412422681227326, + "rkd_loss_weight": 0.13848794847726822, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3834575295448303, + "rkd_teacher_entropy": 3.770273947715759, + "rkd_teacher_similarity_mean": 0.7880859851837159, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25850 + }, + { + "epoch": 7.524003491416933, + "flow_matching_loss": 0.012592968856915831, + "grad_norm": 0.1016404777765274, + "learning_rate": 1.5357753977434541e-06, + "loss": 0.0137, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007796131679788232, + "rkd_loss_weight": 0.13815460354089737, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3806977242231369, + "rkd_teacher_entropy": 3.7597585678100587, + "rkd_teacher_similarity_mean": 0.7854149281978607, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25860 + }, + { + "epoch": 7.526913005528077, + "flow_matching_loss": 0.013377506192773581, + "grad_norm": 0.11784348636865616, + "learning_rate": 1.5284954085253043e-06, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007857538014650344, + "rkd_loss_weight": 0.137821264564991, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38584615886211393, + "rkd_teacher_entropy": 3.7832756996154786, + "rkd_teacher_similarity_mean": 0.7908591091632843, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25870 + }, + { + "epoch": 7.52982251963922, + "flow_matching_loss": 0.011147712264209986, + "grad_norm": 0.09560080617666245, + "learning_rate": 1.5212317884555115e-06, + "loss": 0.0124, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008858954766765237, + "rkd_loss_weight": 0.13748791366815566, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38186357617378236, + "rkd_teacher_entropy": 3.7567675828933718, + "rkd_teacher_similarity_mean": 0.7863526821136475, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25880 + }, + { + "epoch": 7.532732033750364, + "flow_matching_loss": 0.013294386630877852, + "grad_norm": 0.11188846081495285, + "learning_rate": 1.5139845463600605e-06, + "loss": 0.0143, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007385265082120895, + "rkd_loss_weight": 0.13715457171201706, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37727595269680025, + "rkd_teacher_entropy": 3.747028851509094, + "rkd_teacher_similarity_mean": 0.7822724521160126, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25890 + }, + { + "epoch": 7.535641547861507, + "flow_matching_loss": 0.013466296344995498, + "grad_norm": 0.10006058216094971, + "learning_rate": 1.5067536910450352e-06, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007254385109990835, + "rkd_loss_weight": 0.13682123124599457, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3788990467786789, + "rkd_teacher_entropy": 3.756138491630554, + "rkd_teacher_similarity_mean": 0.7829174280166626, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25900 + }, + { + "epoch": 7.538551061972651, + "flow_matching_loss": 0.011814645491540432, + "grad_norm": 0.09147275984287262, + "learning_rate": 1.4995392312966044e-06, + "loss": 0.0129, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007612568698823452, + "rkd_loss_weight": 0.13648787885904312, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3728588312864304, + "rkd_teacher_entropy": 3.7292138814926146, + "rkd_teacher_similarity_mean": 0.7770254492759705, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25910 + }, + { + "epoch": 7.541460576083794, + "flow_matching_loss": 0.011179479071870446, + "grad_norm": 0.12883396446704865, + "learning_rate": 1.492341175881014e-06, + "loss": 0.0124, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008867026166990399, + "rkd_loss_weight": 0.13615453988313675, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3804960519075394, + "rkd_teacher_entropy": 3.7479474782943725, + "rkd_teacher_similarity_mean": 0.7846249103546142, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25920 + }, + { + "epoch": 7.5443700901949375, + "flow_matching_loss": 0.01082310387864709, + "grad_norm": 0.07996859401464462, + "learning_rate": 1.4851595335445895e-06, + "loss": 0.0118, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006838376354426146, + "rkd_loss_weight": 0.1358211949467659, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3764369755983353, + "rkd_teacher_entropy": 3.7414693355560305, + "rkd_teacher_similarity_mean": 0.7801186203956604, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25930 + }, + { + "epoch": 7.547279604306081, + "flow_matching_loss": 0.01250547613017261, + "grad_norm": 0.10321784764528275, + "learning_rate": 1.4779943130136953e-06, + "loss": 0.0135, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0076716090086847545, + "rkd_loss_weight": 0.1354878470301628, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.375046843290329, + "rkd_teacher_entropy": 3.728399062156677, + "rkd_teacher_similarity_mean": 0.7794605731964112, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25940 + }, + { + "epoch": 7.550189118417224, + "flow_matching_loss": 0.012341415509581566, + "grad_norm": 0.1009804755449295, + "learning_rate": 1.4708455229947537e-06, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0074534572660923, + "rkd_loss_weight": 0.13515450805425644, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37703992426395416, + "rkd_teacher_entropy": 3.7416229009628297, + "rkd_teacher_similarity_mean": 0.7821932435035706, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25950 + }, + { + "epoch": 7.553098632528368, + "flow_matching_loss": 0.011243250267580152, + "grad_norm": 0.12169624865055084, + "learning_rate": 1.4637131721742148e-06, + "loss": 0.0122, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007161002187058329, + "rkd_loss_weight": 0.13482115864753724, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3770558714866638, + "rkd_teacher_entropy": 3.7601297378540037, + "rkd_teacher_similarity_mean": 0.7814187824726104, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25960 + }, + { + "epoch": 7.5560081466395115, + "flow_matching_loss": 0.012575871823355556, + "grad_norm": 0.09360231459140778, + "learning_rate": 1.4565972692185626e-06, + "loss": 0.0135, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007029810640960932, + "rkd_loss_weight": 0.1344878152012825, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37311155200004575, + "rkd_teacher_entropy": 3.7287065029144286, + "rkd_teacher_similarity_mean": 0.7779333472251893, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25970 + }, + { + "epoch": 7.558917660750654, + "flow_matching_loss": 0.014093488175421953, + "grad_norm": 0.10129669308662415, + "learning_rate": 1.4494978227742855e-06, + "loss": 0.0151, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007220259495079517, + "rkd_loss_weight": 0.13415447622537613, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37223329544067385, + "rkd_teacher_entropy": 3.729373073577881, + "rkd_teacher_similarity_mean": 0.7760354042053222, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25980 + }, + { + "epoch": 7.561827174861798, + "flow_matching_loss": 0.013882152177393437, + "grad_norm": 0.10219737887382507, + "learning_rate": 1.442414841467885e-06, + "loss": 0.015, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008249407820403575, + "rkd_loss_weight": 0.13382112383842468, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.376346093416214, + "rkd_teacher_entropy": 3.746593165397644, + "rkd_teacher_similarity_mean": 0.7804905116558075, + "rkd_vlm_temp": 0.10000000149011612, + "step": 25990 + }, + { + "epoch": 7.564736688972942, + "flow_matching_loss": 0.01178991855122149, + "grad_norm": 0.10390707105398178, + "learning_rate": 1.4353483339058498e-06, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008967003086581826, + "rkd_loss_weight": 0.1334877833724022, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3767400473356247, + "rkd_teacher_entropy": 3.7281382083892822, + "rkd_teacher_similarity_mean": 0.7809982240200043, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26000 + }, + { + "epoch": 7.5676462030840845, + "flow_matching_loss": 0.014207605738192797, + "grad_norm": 0.09739471226930618, + "learning_rate": 1.4282983086746498e-06, + "loss": 0.0155, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009595928620547057, + "rkd_loss_weight": 0.1331544414162636, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36956012845039365, + "rkd_teacher_entropy": 3.719816780090332, + "rkd_teacher_similarity_mean": 0.7737009465694428, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26010 + }, + { + "epoch": 7.570555717195228, + "flow_matching_loss": 0.012386573012918234, + "grad_norm": 0.14914482831954956, + "learning_rate": 1.4212647743407325e-06, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006841737916693092, + "rkd_loss_weight": 0.13282109051942825, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.371686127781868, + "rkd_teacher_entropy": 3.741011047363281, + "rkd_teacher_similarity_mean": 0.7768801569938659, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26020 + }, + { + "epoch": 7.573465231306372, + "flow_matching_loss": 0.011912118550390005, + "grad_norm": 0.07198430597782135, + "learning_rate": 1.4142477394505065e-06, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008389981370419264, + "rkd_loss_weight": 0.13248775154352188, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38076316714286806, + "rkd_teacher_entropy": 3.761901926994324, + "rkd_teacher_similarity_mean": 0.7852799952030182, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26030 + }, + { + "epoch": 7.576374745417516, + "flow_matching_loss": 0.01333710146136582, + "grad_norm": 0.09798389673233032, + "learning_rate": 1.4072472125303299e-06, + "loss": 0.0143, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007563729025423527, + "rkd_loss_weight": 0.1321544051170349, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37878274619579316, + "rkd_teacher_entropy": 3.757675790786743, + "rkd_teacher_similarity_mean": 0.7827875137329101, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26040 + }, + { + "epoch": 7.5792842595286585, + "flow_matching_loss": 0.01115707727149129, + "grad_norm": 0.09182567149400711, + "learning_rate": 1.4002632020865031e-06, + "loss": 0.012, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006750396825373173, + "rkd_loss_weight": 0.13182105869054794, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3799226492643356, + "rkd_teacher_entropy": 3.75181884765625, + "rkd_teacher_similarity_mean": 0.784122633934021, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26050 + }, + { + "epoch": 7.582193773639802, + "flow_matching_loss": 0.014533271174877882, + "grad_norm": 0.10036831349134445, + "learning_rate": 1.393295716605258e-06, + "loss": 0.0154, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006401869421824813, + "rkd_loss_weight": 0.13148771971464157, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3824644833803177, + "rkd_teacher_entropy": 3.75860116481781, + "rkd_teacher_similarity_mean": 0.7869633376598358, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26060 + }, + { + "epoch": 7.585103287750946, + "flow_matching_loss": 0.011071550147607922, + "grad_norm": 0.09810465574264526, + "learning_rate": 1.386344764552742e-06, + "loss": 0.012, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006804362684488296, + "rkd_loss_weight": 0.13115436881780623, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37665065228939054, + "rkd_teacher_entropy": 3.7440810680389403, + "rkd_teacher_similarity_mean": 0.7812131226062775, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26070 + }, + { + "epoch": 7.588012801862089, + "flow_matching_loss": 0.012350849993526935, + "grad_norm": 0.09749455004930496, + "learning_rate": 1.3794103543750246e-06, + "loss": 0.0132, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006509248772636056, + "rkd_loss_weight": 0.13082102686166763, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3837202399969101, + "rkd_teacher_entropy": 3.7651509761810305, + "rkd_teacher_similarity_mean": 0.7875137865543366, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26080 + }, + { + "epoch": 7.590922315973232, + "flow_matching_loss": 0.01087768548168242, + "grad_norm": 0.11738992482423782, + "learning_rate": 1.3724924944980637e-06, + "loss": 0.0118, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007252942584455013, + "rkd_loss_weight": 0.13048768788576126, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3728910058736801, + "rkd_teacher_entropy": 3.7282613277435304, + "rkd_teacher_similarity_mean": 0.7767372727394104, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26090 + }, + { + "epoch": 7.593831830084376, + "flow_matching_loss": 0.010866340156644583, + "grad_norm": 0.0983157753944397, + "learning_rate": 1.365591193327712e-06, + "loss": 0.0119, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0076490580569952725, + "rkd_loss_weight": 0.1301543340086937, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3753913938999176, + "rkd_teacher_entropy": 3.734896922111511, + "rkd_teacher_similarity_mean": 0.7796958923339844, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26100 + }, + { + "epoch": 7.59674134419552, + "flow_matching_loss": 0.013722170516848564, + "grad_norm": 0.0906657949090004, + "learning_rate": 1.3587064592496984e-06, + "loss": 0.0147, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0073559688404202465, + "rkd_loss_weight": 0.12982099503278732, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.380139496922493, + "rkd_teacher_entropy": 3.747922897338867, + "rkd_teacher_similarity_mean": 0.7841612040996552, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26110 + }, + { + "epoch": 7.599650858306663, + "flow_matching_loss": 0.014484168868511916, + "grad_norm": 0.10319297760725021, + "learning_rate": 1.3518383006296292e-06, + "loss": 0.0154, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0067311437800526615, + "rkd_loss_weight": 0.12948765158653258, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3752014249563217, + "rkd_teacher_entropy": 3.7436032056808473, + "rkd_teacher_similarity_mean": 0.7798107147216797, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26120 + }, + { + "epoch": 7.602560372417806, + "flow_matching_loss": 0.010782784409821033, + "grad_norm": 0.09290505945682526, + "learning_rate": 1.344986725812959e-06, + "loss": 0.0117, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007244752999395132, + "rkd_loss_weight": 0.12915430217981339, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3682955950498581, + "rkd_teacher_entropy": 3.714538335800171, + "rkd_teacher_similarity_mean": 0.7723274290561676, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26130 + }, + { + "epoch": 7.605469886528949, + "flow_matching_loss": 0.011477630957961083, + "grad_norm": 0.12739285826683044, + "learning_rate": 1.3381517431250046e-06, + "loss": 0.0125, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007941483333706856, + "rkd_loss_weight": 0.128820963203907, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3682777225971222, + "rkd_teacher_entropy": 3.72472198009491, + "rkd_teacher_similarity_mean": 0.7724101781845093, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26140 + }, + { + "epoch": 7.608379400640093, + "flow_matching_loss": 0.012455994123592972, + "grad_norm": 0.09769244492053986, + "learning_rate": 1.3313333608709095e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008645965298637748, + "rkd_loss_weight": 0.12848761528730393, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37350199520587923, + "rkd_teacher_entropy": 3.7298937320709227, + "rkd_teacher_similarity_mean": 0.7771114945411682, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26150 + }, + { + "epoch": 7.611288914751237, + "flow_matching_loss": 0.011663148505613207, + "grad_norm": 0.09571979194879532, + "learning_rate": 1.3245315873356528e-06, + "loss": 0.0129, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009893083758652211, + "rkd_loss_weight": 0.12815427035093307, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3822471916675568, + "rkd_teacher_entropy": 3.7637198686599733, + "rkd_teacher_similarity_mean": 0.7873669981956481, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26160 + }, + { + "epoch": 7.61419842886238, + "flow_matching_loss": 0.012074717599898577, + "grad_norm": 0.08467935770750046, + "learning_rate": 1.3177464307840304e-06, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006862969230860471, + "rkd_loss_weight": 0.1278209313750267, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3866754025220871, + "rkd_teacher_entropy": 3.773643207550049, + "rkd_teacher_similarity_mean": 0.7920791745185852, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26170 + }, + { + "epoch": 7.617107942973523, + "flow_matching_loss": 0.010788119258359074, + "grad_norm": 0.08584409952163696, + "learning_rate": 1.310977899460642e-06, + "loss": 0.0118, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007589446473866701, + "rkd_loss_weight": 0.12748757898807525, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37753931879997255, + "rkd_teacher_entropy": 3.7495483636856077, + "rkd_teacher_similarity_mean": 0.782368278503418, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26180 + }, + { + "epoch": 7.620017457084667, + "flow_matching_loss": 0.012093659397214651, + "grad_norm": 0.1020212322473526, + "learning_rate": 1.3042260015899006e-06, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007064470602199435, + "rkd_loss_weight": 0.12715423852205276, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3709303945302963, + "rkd_teacher_entropy": 3.7329805850982667, + "rkd_teacher_similarity_mean": 0.7759361863136292, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26190 + }, + { + "epoch": 7.6229269711958105, + "flow_matching_loss": 0.012552515603601933, + "grad_norm": 0.12994857132434845, + "learning_rate": 1.2974907453759948e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00812778645195067, + "rkd_loss_weight": 0.12682089805603028, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3784648537635803, + "rkd_teacher_entropy": 3.7570501804351806, + "rkd_teacher_similarity_mean": 0.782561582326889, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26200 + }, + { + "epoch": 7.625836485306953, + "flow_matching_loss": 0.011236599693074823, + "grad_norm": 0.08078854531049728, + "learning_rate": 1.2907721390028937e-06, + "loss": 0.0121, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006898174853995442, + "rkd_loss_weight": 0.12648754566907883, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3780629485845566, + "rkd_teacher_entropy": 3.746833825111389, + "rkd_teacher_similarity_mean": 0.7831549167633056, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26210 + }, + { + "epoch": 7.628745999418097, + "flow_matching_loss": 0.01247716681100428, + "grad_norm": 0.10522489249706268, + "learning_rate": 1.2840701906343417e-06, + "loss": 0.0137, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009750197269022464, + "rkd_loss_weight": 0.12615420669317245, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37957742512226106, + "rkd_teacher_entropy": 3.744958591461182, + "rkd_teacher_similarity_mean": 0.7833422362804413, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26220 + }, + { + "epoch": 7.631655513529241, + "flow_matching_loss": 0.01270741131156683, + "grad_norm": 0.08661258220672607, + "learning_rate": 1.2773849084138312e-06, + "loss": 0.0137, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007690655020996928, + "rkd_loss_weight": 0.1258208617568016, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3712980419397354, + "rkd_teacher_entropy": 3.741428804397583, + "rkd_teacher_similarity_mean": 0.7750083684921265, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26230 + }, + { + "epoch": 7.6345650276403845, + "flow_matching_loss": 0.011723544029518963, + "grad_norm": 0.1078261137008667, + "learning_rate": 1.2707163004646177e-06, + "loss": 0.0125, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006247518304735422, + "rkd_loss_weight": 0.12548751384019852, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3883759111166, + "rkd_teacher_entropy": 3.766649770736694, + "rkd_teacher_similarity_mean": 0.7925559997558593, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26240 + }, + { + "epoch": 7.637474541751527, + "flow_matching_loss": 0.010939778946340083, + "grad_norm": 0.09417291730642319, + "learning_rate": 1.2640643748896879e-06, + "loss": 0.012, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008082098606973887, + "rkd_loss_weight": 0.12515417486429214, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3854550004005432, + "rkd_teacher_entropy": 3.7724554538726807, + "rkd_teacher_similarity_mean": 0.7905585587024688, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26250 + }, + { + "epoch": 7.640384055862671, + "flow_matching_loss": 0.01463551064953208, + "grad_norm": 0.07874279469251633, + "learning_rate": 1.257429139771753e-06, + "loss": 0.0155, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007206315267831087, + "rkd_loss_weight": 0.12482082843780518, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3770459443330765, + "rkd_teacher_entropy": 3.7420338869094847, + "rkd_teacher_similarity_mean": 0.781254518032074, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26260 + }, + { + "epoch": 7.643293569973815, + "flow_matching_loss": 0.010885364469140768, + "grad_norm": 0.08194015175104141, + "learning_rate": 1.2508106031732553e-06, + "loss": 0.0118, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007152989786118269, + "rkd_loss_weight": 0.1244874820113182, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3842093914747238, + "rkd_teacher_entropy": 3.7634684324264525, + "rkd_teacher_similarity_mean": 0.7882121980190278, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26270 + }, + { + "epoch": 7.6462030840849575, + "flow_matching_loss": 0.012710588099434973, + "grad_norm": 0.09668288379907608, + "learning_rate": 1.2442087731363366e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007362530753016472, + "rkd_loss_weight": 0.12415413782000542, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38482044339179994, + "rkd_teacher_entropy": 3.7788442373275757, + "rkd_teacher_similarity_mean": 0.7887599647045136, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26280 + }, + { + "epoch": 7.649112598196101, + "flow_matching_loss": 0.011585604026913643, + "grad_norm": 0.09938699007034302, + "learning_rate": 1.237623657682847e-06, + "loss": 0.0125, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007606254983693361, + "rkd_loss_weight": 0.12382079437375068, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36924932301044466, + "rkd_teacher_entropy": 3.7252680540084837, + "rkd_teacher_similarity_mean": 0.7735438704490661, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26290 + }, + { + "epoch": 7.652022112307245, + "flow_matching_loss": 0.010284106712788343, + "grad_norm": 0.09104360640048981, + "learning_rate": 1.2310552648143208e-06, + "loss": 0.0112, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007617164030671119, + "rkd_loss_weight": 0.1234874501824379, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38881228268146517, + "rkd_teacher_entropy": 3.7783234834671022, + "rkd_teacher_similarity_mean": 0.7937063097953796, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26300 + }, + { + "epoch": 7.654931626418388, + "flow_matching_loss": 0.014756133127957583, + "grad_norm": 0.07728949934244156, + "learning_rate": 1.2245036025119738e-06, + "loss": 0.0155, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00624233614653349, + "rkd_loss_weight": 0.12315410450100898, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3791664391756058, + "rkd_teacher_entropy": 3.745740008354187, + "rkd_teacher_similarity_mean": 0.7838302433490754, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26310 + }, + { + "epoch": 7.6578411405295315, + "flow_matching_loss": 0.0114593047183007, + "grad_norm": 0.10468946397304535, + "learning_rate": 1.2179686787366962e-06, + "loss": 0.0124, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007937559578567743, + "rkd_loss_weight": 0.1228207603096962, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38113785088062285, + "rkd_teacher_entropy": 3.7602964639663696, + "rkd_teacher_similarity_mean": 0.7854891240596771, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26320 + }, + { + "epoch": 7.660750654640675, + "flow_matching_loss": 0.013313118601217866, + "grad_norm": 0.08179773390293121, + "learning_rate": 1.2114505014290329e-06, + "loss": 0.0141, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006306689372286201, + "rkd_loss_weight": 0.12248741686344147, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3731807589530945, + "rkd_teacher_entropy": 3.7241615056991577, + "rkd_teacher_similarity_mean": 0.7779548466205597, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26330 + }, + { + "epoch": 7.663660168751818, + "flow_matching_loss": 0.015692265052348376, + "grad_norm": 0.11439964175224304, + "learning_rate": 1.2049490785091833e-06, + "loss": 0.0166, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00714306216686964, + "rkd_loss_weight": 0.12215407192707062, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3732238829135895, + "rkd_teacher_entropy": 3.7339576482772827, + "rkd_teacher_similarity_mean": 0.7771491289138794, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26340 + }, + { + "epoch": 7.666569682862962, + "flow_matching_loss": 0.011590288020670414, + "grad_norm": 0.08619778603315353, + "learning_rate": 1.1984644178769949e-06, + "loss": 0.0125, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007471108436584472, + "rkd_loss_weight": 0.12182072699069976, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3779499024152756, + "rkd_teacher_entropy": 3.744958686828613, + "rkd_teacher_similarity_mean": 0.7825733840465545, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26350 + }, + { + "epoch": 7.669479196974105, + "flow_matching_loss": 0.010147521225735544, + "grad_norm": 0.1198737621307373, + "learning_rate": 1.1919965274119381e-06, + "loss": 0.0111, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007818743772804736, + "rkd_loss_weight": 0.12148738279938698, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36995474696159364, + "rkd_teacher_entropy": 3.7214395999908447, + "rkd_teacher_similarity_mean": 0.774198979139328, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26360 + }, + { + "epoch": 7.672388711085249, + "flow_matching_loss": 0.012761392956599593, + "grad_norm": 0.10736460983753204, + "learning_rate": 1.18554541497311e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0068248988594859835, + "rkd_loss_weight": 0.12115403935313225, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3762444734573364, + "rkd_teacher_entropy": 3.7461599588394163, + "rkd_teacher_similarity_mean": 0.7815120577812195, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26370 + }, + { + "epoch": 7.675298225196392, + "flow_matching_loss": 0.013406555680558086, + "grad_norm": 0.12939946353435516, + "learning_rate": 1.1791110883992213e-06, + "loss": 0.0144, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007907785940915346, + "rkd_loss_weight": 0.12082069367170334, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.374562206864357, + "rkd_teacher_entropy": 3.739597773551941, + "rkd_teacher_similarity_mean": 0.7792029321193695, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26380 + }, + { + "epoch": 7.678207739307536, + "flow_matching_loss": 0.014385159965604544, + "grad_norm": 0.12453721463680267, + "learning_rate": 1.1726935555085817e-06, + "loss": 0.0152, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0071141320746392015, + "rkd_loss_weight": 0.12048734873533248, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37037305235862733, + "rkd_teacher_entropy": 3.721959042549133, + "rkd_teacher_similarity_mean": 0.7750711560249328, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26390 + }, + { + "epoch": 7.681117253418679, + "flow_matching_loss": 0.01310357041656971, + "grad_norm": 0.13531503081321716, + "learning_rate": 1.166292824099105e-06, + "loss": 0.0141, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008245040290057659, + "rkd_loss_weight": 0.12015400528907776, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38096262514591217, + "rkd_teacher_entropy": 3.7557332515716553, + "rkd_teacher_similarity_mean": 0.7853150606155396, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26400 + }, + { + "epoch": 7.684026767529822, + "flow_matching_loss": 0.010905154794454575, + "grad_norm": 0.07402972877025604, + "learning_rate": 1.1599089019482768e-06, + "loss": 0.0118, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007482893764972687, + "rkd_loss_weight": 0.11982066184282303, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3746795982122421, + "rkd_teacher_entropy": 3.7333268880844117, + "rkd_teacher_similarity_mean": 0.7797871887683868, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26410 + }, + { + "epoch": 7.686936281640966, + "flow_matching_loss": 0.01067523555830121, + "grad_norm": 0.12483689934015274, + "learning_rate": 1.1535417968131718e-06, + "loss": 0.0116, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007560924952849745, + "rkd_loss_weight": 0.11948731541633606, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3728205054998398, + "rkd_teacher_entropy": 3.7347640991210938, + "rkd_teacher_similarity_mean": 0.7773735463619232, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26420 + }, + { + "epoch": 7.68984579575211, + "flow_matching_loss": 0.013089272659271955, + "grad_norm": 0.12172364443540573, + "learning_rate": 1.147191516430418e-06, + "loss": 0.0139, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006983358319848776, + "rkd_loss_weight": 0.11915397122502328, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3682065963745117, + "rkd_teacher_entropy": 3.726590085029602, + "rkd_teacher_similarity_mean": 0.7741218388080597, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26430 + }, + { + "epoch": 7.692755309863253, + "flow_matching_loss": 0.012637545773759484, + "grad_norm": 0.09517254680395126, + "learning_rate": 1.140858068516205e-06, + "loss": 0.0134, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006805389560759068, + "rkd_loss_weight": 0.11882062777876853, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37766703963279724, + "rkd_teacher_entropy": 3.7507100343704223, + "rkd_teacher_similarity_mean": 0.7831737041473389, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26440 + }, + { + "epoch": 7.695664823974396, + "flow_matching_loss": 0.010972346179187298, + "grad_norm": 0.08728571236133575, + "learning_rate": 1.1345414607662736e-06, + "loss": 0.0117, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006234329659491777, + "rkd_loss_weight": 0.11848728358745575, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3756466507911682, + "rkd_teacher_entropy": 3.7354292392730715, + "rkd_teacher_similarity_mean": 0.7805611550807953, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26450 + }, + { + "epoch": 7.69857433808554, + "flow_matching_loss": 0.011186461569741369, + "grad_norm": 0.085120290517807, + "learning_rate": 1.1282417008558937e-06, + "loss": 0.0121, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007666925014927983, + "rkd_loss_weight": 0.11815393790602684, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37833278477191923, + "rkd_teacher_entropy": 3.7589509010314943, + "rkd_teacher_similarity_mean": 0.7836698770523072, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26460 + }, + { + "epoch": 7.7014838521966835, + "flow_matching_loss": 0.012701825192198157, + "grad_norm": 0.13347609341144562, + "learning_rate": 1.1219587964398674e-06, + "loss": 0.0137, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008466065488755703, + "rkd_loss_weight": 0.11782059371471405, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3828764379024506, + "rkd_teacher_entropy": 3.759234666824341, + "rkd_teacher_similarity_mean": 0.7886174917221069, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26470 + }, + { + "epoch": 7.704393366307826, + "flow_matching_loss": 0.01242824923247099, + "grad_norm": 0.09849438816308975, + "learning_rate": 1.1156927551525198e-06, + "loss": 0.0134, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008213393902406097, + "rkd_loss_weight": 0.11748725026845933, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3771266877651215, + "rkd_teacher_entropy": 3.745274066925049, + "rkd_teacher_similarity_mean": 0.7830470323562622, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26480 + }, + { + "epoch": 7.70730288041897, + "flow_matching_loss": 0.012461978010833263, + "grad_norm": 0.10185939818620682, + "learning_rate": 1.109443584607675e-06, + "loss": 0.0134, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007664639805443585, + "rkd_loss_weight": 0.11715390533208847, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37311350405216215, + "rkd_teacher_entropy": 3.7299612283706667, + "rkd_teacher_similarity_mean": 0.7775557339191437, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26490 + }, + { + "epoch": 7.710212394530114, + "flow_matching_loss": 0.012589430855587124, + "grad_norm": 0.10563139617443085, + "learning_rate": 1.1032112923986708e-06, + "loss": 0.0135, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00790561931207776, + "rkd_loss_weight": 0.11682055965065956, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3795893728733063, + "rkd_teacher_entropy": 3.751855897903442, + "rkd_teacher_similarity_mean": 0.784920734167099, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26500 + }, + { + "epoch": 7.713121908641257, + "flow_matching_loss": 0.013422481436282396, + "grad_norm": 0.10162055492401123, + "learning_rate": 1.0969958860983299e-06, + "loss": 0.0143, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007368072168901563, + "rkd_loss_weight": 0.11648721620440483, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37514000236988065, + "rkd_teacher_entropy": 3.7399257898330687, + "rkd_teacher_similarity_mean": 0.7797565758228302, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26510 + }, + { + "epoch": 7.7160314227524, + "flow_matching_loss": 0.011157978186383843, + "grad_norm": 0.1242537647485733, + "learning_rate": 1.0907973732589537e-06, + "loss": 0.0121, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007723146909847855, + "rkd_loss_weight": 0.1161538727581501, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3812152802944183, + "rkd_teacher_entropy": 3.7528372287750242, + "rkd_teacher_similarity_mean": 0.7863085567951202, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26520 + }, + { + "epoch": 7.718940936863544, + "flow_matching_loss": 0.013363456958904863, + "grad_norm": 0.08279789984226227, + "learning_rate": 1.0846157614123215e-06, + "loss": 0.0141, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006137346802279353, + "rkd_loss_weight": 0.11582052707672119, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3779588520526886, + "rkd_teacher_entropy": 3.75495285987854, + "rkd_teacher_similarity_mean": 0.7824867069721222, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26530 + }, + { + "epoch": 7.721850450974687, + "flow_matching_loss": 0.011487497296184302, + "grad_norm": 0.0908002033829689, + "learning_rate": 1.078451058069671e-06, + "loss": 0.0124, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007672825269401074, + "rkd_loss_weight": 0.11548718214035034, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37519254386425016, + "rkd_teacher_entropy": 3.7392019510269163, + "rkd_teacher_similarity_mean": 0.7797906160354614, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26540 + }, + { + "epoch": 7.7247599650858305, + "flow_matching_loss": 0.012467002402991057, + "grad_norm": 0.10766805708408356, + "learning_rate": 1.0723032707217056e-06, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007126267440617085, + "rkd_loss_weight": 0.11515383869409561, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38035663962364197, + "rkd_teacher_entropy": 3.7424172163009644, + "rkd_teacher_similarity_mean": 0.7850790798664093, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26550 + }, + { + "epoch": 7.727669479196974, + "flow_matching_loss": 0.011563682416453957, + "grad_norm": 0.12125833332538605, + "learning_rate": 1.066172406838558e-06, + "loss": 0.0122, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0057835773099213835, + "rkd_loss_weight": 0.11482049524784088, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3815727710723877, + "rkd_teacher_entropy": 3.7623626470565794, + "rkd_teacher_similarity_mean": 0.7870263695716858, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26560 + }, + { + "epoch": 7.730578993308118, + "flow_matching_loss": 0.016595673095434905, + "grad_norm": 0.09301720559597015, + "learning_rate": 1.0600584738698127e-06, + "loss": 0.0174, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007398491306230426, + "rkd_loss_weight": 0.11448714882135391, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36533705294132235, + "rkd_teacher_entropy": 3.7143701314926147, + "rkd_teacher_similarity_mean": 0.7711060583591461, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26570 + }, + { + "epoch": 7.733488507419261, + "flow_matching_loss": 0.012665232690051199, + "grad_norm": 0.1035815179347992, + "learning_rate": 1.0539614792444735e-06, + "loss": 0.0134, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006700356677174568, + "rkd_loss_weight": 0.11415380463004113, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3851964294910431, + "rkd_teacher_entropy": 3.763132166862488, + "rkd_teacher_similarity_mean": 0.7901616394519806, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26580 + }, + { + "epoch": 7.7363980215304045, + "flow_matching_loss": 0.010767572512850165, + "grad_norm": 0.07091835141181946, + "learning_rate": 1.0478814303709627e-06, + "loss": 0.0115, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006407215399667621, + "rkd_loss_weight": 0.11382046118378639, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3669286996126175, + "rkd_teacher_entropy": 3.718942427635193, + "rkd_teacher_similarity_mean": 0.7719950020313263, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26590 + }, + { + "epoch": 7.739307535641548, + "flow_matching_loss": 0.014508095849305392, + "grad_norm": 0.12527678906917572, + "learning_rate": 1.041818334637109e-06, + "loss": 0.0156, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009514106810092926, + "rkd_loss_weight": 0.1134871169924736, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3763846129179001, + "rkd_teacher_entropy": 3.7439221143722534, + "rkd_teacher_similarity_mean": 0.781148898601532, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26600 + }, + { + "epoch": 7.742217049752691, + "flow_matching_loss": 0.011759638646617531, + "grad_norm": 0.10401691496372223, + "learning_rate": 1.0357721994101526e-06, + "loss": 0.0125, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006572941364720464, + "rkd_loss_weight": 0.11315377056598663, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.383938854932785, + "rkd_teacher_entropy": 3.7586118936538697, + "rkd_teacher_similarity_mean": 0.7894130885601044, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26610 + }, + { + "epoch": 7.745126563863835, + "flow_matching_loss": 0.010996896773576736, + "grad_norm": 0.12046606093645096, + "learning_rate": 1.0297430320367135e-06, + "loss": 0.012, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008506613038480281, + "rkd_loss_weight": 0.1128204271197319, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3815420001745224, + "rkd_teacher_entropy": 3.7568989992141724, + "rkd_teacher_similarity_mean": 0.7872408986091614, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26620 + }, + { + "epoch": 7.748036077974978, + "flow_matching_loss": 0.012814051890745758, + "grad_norm": 0.10232411324977875, + "learning_rate": 1.0237308398428014e-06, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007286545820534229, + "rkd_loss_weight": 0.11248708367347718, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3763263404369354, + "rkd_teacher_entropy": 3.74998459815979, + "rkd_teacher_similarity_mean": 0.7810013473033905, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26630 + }, + { + "epoch": 7.750945592086122, + "flow_matching_loss": 0.012862328113988042, + "grad_norm": 0.06952159851789474, + "learning_rate": 1.0177356301337925e-06, + "loss": 0.0137, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007346498826518655, + "rkd_loss_weight": 0.11215373873710632, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37193041741847993, + "rkd_teacher_entropy": 3.7309985637664793, + "rkd_teacher_similarity_mean": 0.7772140502929688, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26640 + }, + { + "epoch": 7.753855106197265, + "flow_matching_loss": 0.01210819948464632, + "grad_norm": 0.13904865086078644, + "learning_rate": 1.0117574101944348e-06, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00820866208523512, + "rkd_loss_weight": 0.11182039305567741, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3838785350322723, + "rkd_teacher_entropy": 3.764944314956665, + "rkd_teacher_similarity_mean": 0.788957166671753, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26650 + }, + { + "epoch": 7.756764620308409, + "flow_matching_loss": 0.011691773217171431, + "grad_norm": 0.08332493156194687, + "learning_rate": 1.0057961872888293e-06, + "loss": 0.0125, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007493679178878665, + "rkd_loss_weight": 0.11148704960942268, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3764202266931534, + "rkd_teacher_entropy": 3.734555196762085, + "rkd_teacher_similarity_mean": 0.7815695226192474, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26660 + }, + { + "epoch": 7.7596741344195515, + "flow_matching_loss": 0.011432364117354155, + "grad_norm": 0.07245858758687973, + "learning_rate": 9.998519686604235e-07, + "loss": 0.0122, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006857591029256582, + "rkd_loss_weight": 0.11115370616316796, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3789626657962799, + "rkd_teacher_entropy": 3.7589675188064575, + "rkd_teacher_similarity_mean": 0.7849681138992309, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26670 + }, + { + "epoch": 7.762583648530695, + "flow_matching_loss": 0.011388254910707473, + "grad_norm": 0.09057165682315826, + "learning_rate": 9.939247615320047e-07, + "loss": 0.0122, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006958759156987071, + "rkd_loss_weight": 0.11082036048173904, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37957180142402647, + "rkd_teacher_entropy": 3.7484827280044555, + "rkd_teacher_similarity_mean": 0.7847976565361023, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26680 + }, + { + "epoch": 7.765493162641839, + "flow_matching_loss": 0.01203977195546031, + "grad_norm": 0.10388442128896713, + "learning_rate": 9.880145731056834e-07, + "loss": 0.0129, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007378859561868012, + "rkd_loss_weight": 0.11048701554536819, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37829920649528503, + "rkd_teacher_entropy": 3.7538580179214476, + "rkd_teacher_similarity_mean": 0.7832732200622559, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26690 + }, + { + "epoch": 7.768402676752983, + "flow_matching_loss": 0.012871231324970723, + "grad_norm": 0.08631335943937302, + "learning_rate": 9.821214105629034e-07, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006832461524754763, + "rkd_loss_weight": 0.11015367209911346, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3822422742843628, + "rkd_teacher_entropy": 3.759543037414551, + "rkd_teacher_similarity_mean": 0.7866835594177246, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26700 + }, + { + "epoch": 7.771312190864125, + "flow_matching_loss": 0.012247852515429258, + "grad_norm": 0.08815015852451324, + "learning_rate": 9.762452810644068e-07, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00701467520557344, + "rkd_loss_weight": 0.10982032790780068, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3753958225250244, + "rkd_teacher_entropy": 3.7304423332214354, + "rkd_teacher_similarity_mean": 0.7809902250766754, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26710 + }, + { + "epoch": 7.774221704975269, + "flow_matching_loss": 0.012654901621863246, + "grad_norm": 0.0813853070139885, + "learning_rate": 9.703861917502487e-07, + "loss": 0.0134, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007108154008165002, + "rkd_loss_weight": 0.10948698222637177, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3884408801794052, + "rkd_teacher_entropy": 3.782763123512268, + "rkd_teacher_similarity_mean": 0.7942464888095856, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26720 + }, + { + "epoch": 7.777131219086413, + "flow_matching_loss": 0.014989019185304642, + "grad_norm": 0.12701529264450073, + "learning_rate": 9.645441497397744e-07, + "loss": 0.0158, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00715579972602427, + "rkd_loss_weight": 0.10915363803505898, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3805153280496597, + "rkd_teacher_entropy": 3.7605531454086303, + "rkd_teacher_similarity_mean": 0.785873007774353, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26730 + }, + { + "epoch": 7.780040733197556, + "flow_matching_loss": 0.012035532388836145, + "grad_norm": 0.09302866458892822, + "learning_rate": 9.587191621316154e-07, + "loss": 0.0129, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007761475117877126, + "rkd_loss_weight": 0.10882029458880424, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3721580356359482, + "rkd_teacher_entropy": 3.7294304609298705, + "rkd_teacher_similarity_mean": 0.7767806708812713, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26740 + }, + { + "epoch": 7.782950247308699, + "flow_matching_loss": 0.010998431546613573, + "grad_norm": 0.0779944658279419, + "learning_rate": 9.529112360036802e-07, + "loss": 0.0117, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006508209044113755, + "rkd_loss_weight": 0.10848695039749146, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3795007854700089, + "rkd_teacher_entropy": 3.7507683753967287, + "rkd_teacher_similarity_mean": 0.7850961029529572, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26750 + }, + { + "epoch": 7.785859761419843, + "flow_matching_loss": 0.013165998505428433, + "grad_norm": 0.11305279284715652, + "learning_rate": 9.471203784131449e-07, + "loss": 0.0141, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008730724779888987, + "rkd_loss_weight": 0.10815360397100449, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.371595898270607, + "rkd_teacher_entropy": 3.7298595190048216, + "rkd_teacher_similarity_mean": 0.7769080460071563, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26760 + }, + { + "epoch": 7.788769275530987, + "flow_matching_loss": 0.01206478481180966, + "grad_norm": 0.05607437342405319, + "learning_rate": 9.41346596396453e-07, + "loss": 0.0128, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00726152379065752, + "rkd_loss_weight": 0.10782026052474976, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3695975124835968, + "rkd_teacher_entropy": 3.721114420890808, + "rkd_teacher_similarity_mean": 0.7747730493545533, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26770 + }, + { + "epoch": 7.79167878964213, + "flow_matching_loss": 0.012212370242923497, + "grad_norm": 0.07345113158226013, + "learning_rate": 9.355898969692889e-07, + "loss": 0.0129, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006450751796364784, + "rkd_loss_weight": 0.10748691707849503, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37451801300048826, + "rkd_teacher_entropy": 3.7344220161437987, + "rkd_teacher_similarity_mean": 0.7792039215564728, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26780 + }, + { + "epoch": 7.794588303753273, + "flow_matching_loss": 0.014394150301814079, + "grad_norm": 0.1432756632566452, + "learning_rate": 9.29850287126589e-07, + "loss": 0.0152, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00724761220626533, + "rkd_loss_weight": 0.10715357214212418, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37541561424732206, + "rkd_teacher_entropy": 3.7448647737503054, + "rkd_teacher_similarity_mean": 0.7816770970821381, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26790 + }, + { + "epoch": 7.797497817864417, + "flow_matching_loss": 0.013010657904669643, + "grad_norm": 0.10869129747152328, + "learning_rate": 9.24127773842518e-07, + "loss": 0.0138, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007205146504566074, + "rkd_loss_weight": 0.10682022646069526, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37642520368099214, + "rkd_teacher_entropy": 3.7468257665634157, + "rkd_teacher_similarity_mean": 0.7817409574985504, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26800 + }, + { + "epoch": 7.80040733197556, + "flow_matching_loss": 0.010915984585881233, + "grad_norm": 0.08899399638175964, + "learning_rate": 9.18422364070467e-07, + "loss": 0.0116, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006845020083710551, + "rkd_loss_weight": 0.10648688301444054, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.370593786239624, + "rkd_teacher_entropy": 3.722456216812134, + "rkd_teacher_similarity_mean": 0.7745956599712371, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26810 + }, + { + "epoch": 7.8033168460867035, + "flow_matching_loss": 0.011597826704382896, + "grad_norm": 0.1183694452047348, + "learning_rate": 9.127340647430538e-07, + "loss": 0.0123, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006761012505739928, + "rkd_loss_weight": 0.10615353882312775, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3730690449476242, + "rkd_teacher_entropy": 3.7364239931106566, + "rkd_teacher_similarity_mean": 0.7782242596149445, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26820 + }, + { + "epoch": 7.806226360197847, + "flow_matching_loss": 0.012512046424672008, + "grad_norm": 0.06671930849552155, + "learning_rate": 9.070628827720929e-07, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007695631450042128, + "rkd_loss_weight": 0.1058201938867569, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38737807273864744, + "rkd_teacher_entropy": 3.77441828250885, + "rkd_teacher_similarity_mean": 0.7927246570587159, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26830 + }, + { + "epoch": 7.80913587430899, + "flow_matching_loss": 0.010228851065039634, + "grad_norm": 0.0744984820485115, + "learning_rate": 9.014088250486052e-07, + "loss": 0.0109, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00669039455242455, + "rkd_loss_weight": 0.10548684895038604, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37827691435813904, + "rkd_teacher_entropy": 3.737491250038147, + "rkd_teacher_similarity_mean": 0.7842587411403656, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26840 + }, + { + "epoch": 7.812045388420134, + "flow_matching_loss": 0.012225481634959578, + "grad_norm": 0.1093188002705574, + "learning_rate": 8.95771898442807e-07, + "loss": 0.0132, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009402819769456982, + "rkd_loss_weight": 0.10515350550413131, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36774087250232695, + "rkd_teacher_entropy": 3.7281627893447875, + "rkd_teacher_similarity_mean": 0.7736045360565186, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26850 + }, + { + "epoch": 7.8149549025312774, + "flow_matching_loss": 0.012515736371278762, + "grad_norm": 0.10037433356046677, + "learning_rate": 8.90152109804091e-07, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007528354041278362, + "rkd_loss_weight": 0.10482016131281853, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37504996359348297, + "rkd_teacher_entropy": 3.7412851333618162, + "rkd_teacher_similarity_mean": 0.7805840194225311, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26860 + }, + { + "epoch": 7.81786441664242, + "flow_matching_loss": 0.011016866052523255, + "grad_norm": 0.08719870448112488, + "learning_rate": 8.845494659610365e-07, + "loss": 0.0119, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008233181620016693, + "rkd_loss_weight": 0.10448681563138962, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3827264666557312, + "rkd_teacher_entropy": 3.7705522537231446, + "rkd_teacher_similarity_mean": 0.7887242555618286, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26870 + }, + { + "epoch": 7.820773930753564, + "flow_matching_loss": 0.011136137507855892, + "grad_norm": 0.09007089585065842, + "learning_rate": 8.789639737213812e-07, + "loss": 0.0118, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006168798264116049, + "rkd_loss_weight": 0.10415347144007683, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3817711055278778, + "rkd_teacher_entropy": 3.75501229763031, + "rkd_teacher_similarity_mean": 0.7876939713954926, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26880 + }, + { + "epoch": 7.823683444864708, + "flow_matching_loss": 0.010925171244889498, + "grad_norm": 0.07602719962596893, + "learning_rate": 8.733956398720244e-07, + "loss": 0.0117, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007203950127586722, + "rkd_loss_weight": 0.10382012799382209, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3735785722732544, + "rkd_teacher_entropy": 3.7371170997619627, + "rkd_teacher_similarity_mean": 0.7784948945045471, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26890 + }, + { + "epoch": 7.826592958975851, + "flow_matching_loss": 0.011485560797154903, + "grad_norm": 0.062324512749910355, + "learning_rate": 8.678444711790157e-07, + "loss": 0.0122, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006712861405685544, + "rkd_loss_weight": 0.10348678380250931, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38474429249763487, + "rkd_teacher_entropy": 3.7612410306930544, + "rkd_teacher_similarity_mean": 0.7899091422557831, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26900 + }, + { + "epoch": 7.829502473086994, + "flow_matching_loss": 0.011807667277753352, + "grad_norm": 0.0955718457698822, + "learning_rate": 8.623104743875493e-07, + "loss": 0.0127, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008690646756440401, + "rkd_loss_weight": 0.10315343737602234, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38330215215682983, + "rkd_teacher_entropy": 3.7661356687545777, + "rkd_teacher_similarity_mean": 0.7883273541927338, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26910 + }, + { + "epoch": 7.832411987198138, + "flow_matching_loss": 0.01090981881134212, + "grad_norm": 0.10048281401395798, + "learning_rate": 8.567936562219547e-07, + "loss": 0.0117, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007486637122929097, + "rkd_loss_weight": 0.10282009392976761, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3714721202850342, + "rkd_teacher_entropy": 3.7315879821777345, + "rkd_teacher_similarity_mean": 0.7761390209197998, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26920 + }, + { + "epoch": 7.835321501309282, + "flow_matching_loss": 0.012063642079010607, + "grad_norm": 0.09179902821779251, + "learning_rate": 8.512940233856847e-07, + "loss": 0.0128, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007663340168073773, + "rkd_loss_weight": 0.10248674973845481, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37393843531608584, + "rkd_teacher_entropy": 3.738185262680054, + "rkd_teacher_similarity_mean": 0.7792378544807435, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26930 + }, + { + "epoch": 7.838231015420424, + "flow_matching_loss": 0.010218395851552486, + "grad_norm": 0.0861823558807373, + "learning_rate": 8.458115825613122e-07, + "loss": 0.011, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007187048019841313, + "rkd_loss_weight": 0.10215340554714203, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37136208415031435, + "rkd_teacher_entropy": 3.743108534812927, + "rkd_teacher_similarity_mean": 0.7773667812347412, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26940 + }, + { + "epoch": 7.841140529531568, + "flow_matching_loss": 0.011453305976465345, + "grad_norm": 0.08937013149261475, + "learning_rate": 8.403463404105199e-07, + "loss": 0.0122, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007394282845780253, + "rkd_loss_weight": 0.10182005986571312, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37926667332649233, + "rkd_teacher_entropy": 3.758007788658142, + "rkd_teacher_similarity_mean": 0.7853772640228271, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26950 + }, + { + "epoch": 7.844050043642712, + "flow_matching_loss": 0.012697810912504793, + "grad_norm": 0.09403014183044434, + "learning_rate": 8.348983035740926e-07, + "loss": 0.0134, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0073624579701572655, + "rkd_loss_weight": 0.10148671641945839, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3672031044960022, + "rkd_teacher_entropy": 3.7194044828414916, + "rkd_teacher_similarity_mean": 0.7733334720134735, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26960 + }, + { + "epoch": 7.846959557753856, + "flow_matching_loss": 0.009746303921565414, + "grad_norm": 0.07166633754968643, + "learning_rate": 8.294674786719064e-07, + "loss": 0.0104, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0061467030551284555, + "rkd_loss_weight": 0.1011533722281456, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36960749328136444, + "rkd_teacher_entropy": 3.718385338783264, + "rkd_teacher_similarity_mean": 0.7754179537296295, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26970 + }, + { + "epoch": 7.849869071864998, + "flow_matching_loss": 0.012987887067720295, + "grad_norm": 0.09198352694511414, + "learning_rate": 8.240538723029295e-07, + "loss": 0.0138, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008218751288950443, + "rkd_loss_weight": 0.10082002729177475, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3784124583005905, + "rkd_teacher_entropy": 3.75211341381073, + "rkd_teacher_similarity_mean": 0.7840400159358978, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26980 + }, + { + "epoch": 7.852778585976142, + "flow_matching_loss": 0.012515618558973074, + "grad_norm": 0.1143496111035347, + "learning_rate": 8.186574910451983e-07, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0076066955458372835, + "rkd_loss_weight": 0.1004866823554039, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38272927701473236, + "rkd_teacher_entropy": 3.7679339170455934, + "rkd_teacher_similarity_mean": 0.7879935383796692, + "rkd_vlm_temp": 0.10000000149011612, + "step": 26990 + }, + { + "epoch": 7.855688100087286, + "flow_matching_loss": 0.012699880683794618, + "grad_norm": 0.08723198622465134, + "learning_rate": 8.132783414558326e-07, + "loss": 0.0135, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008162695588544011, + "rkd_loss_weight": 0.10015333890914917, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37418310940265653, + "rkd_teacher_entropy": 3.7398863077163695, + "rkd_teacher_similarity_mean": 0.7798891842365265, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27000 + }, + { + "epoch": 7.858597614198429, + "flow_matching_loss": 0.011372127011418343, + "grad_norm": 0.08332361280918121, + "learning_rate": 8.079164300710023e-07, + "loss": 0.012, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006240847520530224, + "rkd_loss_weight": 0.09981999471783638, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37283516824245455, + "rkd_teacher_entropy": 3.7236113786697387, + "rkd_teacher_similarity_mean": 0.7787882030010224, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27010 + }, + { + "epoch": 7.861507128309572, + "flow_matching_loss": 0.011962102539837361, + "grad_norm": 0.06899730861186981, + "learning_rate": 8.025717634059326e-07, + "loss": 0.0125, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.005763150379061699, + "rkd_loss_weight": 0.09948664903640747, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3829425543546677, + "rkd_teacher_entropy": 3.7629391670227053, + "rkd_teacher_similarity_mean": 0.7892882525920868, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27020 + }, + { + "epoch": 7.864416642420716, + "flow_matching_loss": 0.012928071524947881, + "grad_norm": 0.09073855727910995, + "learning_rate": 7.972443479549035e-07, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006728393398225308, + "rkd_loss_weight": 0.09915330484509469, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3794745445251465, + "rkd_teacher_entropy": 3.753195071220398, + "rkd_teacher_similarity_mean": 0.7851791083812714, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27030 + }, + { + "epoch": 7.867326156531859, + "flow_matching_loss": 0.012456639437004924, + "grad_norm": 0.09275709092617035, + "learning_rate": 7.919341901912236e-07, + "loss": 0.0132, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007312366971746087, + "rkd_loss_weight": 0.09881996065378189, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3766300916671753, + "rkd_teacher_entropy": 3.7427829027175905, + "rkd_teacher_similarity_mean": 0.7822638332843781, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27040 + }, + { + "epoch": 7.870235670643003, + "flow_matching_loss": 0.012629497237503529, + "grad_norm": 0.14414680004119873, + "learning_rate": 7.866412965672365e-07, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007095928443595767, + "rkd_loss_weight": 0.09848661720752716, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3775900065898895, + "rkd_teacher_entropy": 3.746611499786377, + "rkd_teacher_similarity_mean": 0.7840627253055572, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27050 + }, + { + "epoch": 7.873145184754146, + "flow_matching_loss": 0.011626400006935, + "grad_norm": 0.10018373280763626, + "learning_rate": 7.813656735143044e-07, + "loss": 0.0123, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006733789388090372, + "rkd_loss_weight": 0.09815327078104019, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38373264074325564, + "rkd_teacher_entropy": 3.7671413898468016, + "rkd_teacher_similarity_mean": 0.7893864214420319, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27060 + }, + { + "epoch": 7.876054698865289, + "flow_matching_loss": 0.014087874069809913, + "grad_norm": 0.16060557961463928, + "learning_rate": 7.761073274428076e-07, + "loss": 0.0148, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007417841628193855, + "rkd_loss_weight": 0.09781992733478546, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37842308878898623, + "rkd_teacher_entropy": 3.7585165739059447, + "rkd_teacher_similarity_mean": 0.7848203599452972, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27070 + }, + { + "epoch": 7.878964212976433, + "flow_matching_loss": 0.013844816852360964, + "grad_norm": 0.09411334246397018, + "learning_rate": 7.708662647421355e-07, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0072551760356873276, + "rkd_loss_weight": 0.09748658314347267, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3674804300069809, + "rkd_teacher_entropy": 3.7203120470046995, + "rkd_teacher_similarity_mean": 0.7734599471092224, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27080 + }, + { + "epoch": 7.8818737270875765, + "flow_matching_loss": 0.01136146909557283, + "grad_norm": 0.08124435693025589, + "learning_rate": 7.65642491780672e-07, + "loss": 0.0122, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008189172903075814, + "rkd_loss_weight": 0.09715323895215988, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3738311767578125, + "rkd_teacher_entropy": 3.7185943126678467, + "rkd_teacher_similarity_mean": 0.7785942673683166, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27090 + }, + { + "epoch": 7.88478324119872, + "flow_matching_loss": 0.011925000930204988, + "grad_norm": 0.08721104264259338, + "learning_rate": 7.604360149057916e-07, + "loss": 0.0127, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007560216821730137, + "rkd_loss_weight": 0.09681989327073097, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3829262375831604, + "rkd_teacher_entropy": 3.767310380935669, + "rkd_teacher_similarity_mean": 0.7888040959835052, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27100 + }, + { + "epoch": 7.887692755309863, + "flow_matching_loss": 0.011615842627361418, + "grad_norm": 0.13705456256866455, + "learning_rate": 7.552468404438568e-07, + "loss": 0.0123, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0068718456663191315, + "rkd_loss_weight": 0.09648654982447624, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3730247765779495, + "rkd_teacher_entropy": 3.730087971687317, + "rkd_teacher_similarity_mean": 0.7787325084209442, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27110 + }, + { + "epoch": 7.890602269421007, + "flow_matching_loss": 0.011465528095141054, + "grad_norm": 0.08909647166728973, + "learning_rate": 7.500749747002022e-07, + "loss": 0.0122, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008158110035583376, + "rkd_loss_weight": 0.09615320563316346, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.373677533864975, + "rkd_teacher_entropy": 3.7299543142318727, + "rkd_teacher_similarity_mean": 0.7793206214904785, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27120 + }, + { + "epoch": 7.89351178353215, + "flow_matching_loss": 0.012559197004884481, + "grad_norm": 0.08112391084432602, + "learning_rate": 7.449204239591345e-07, + "loss": 0.0132, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006959378067404032, + "rkd_loss_weight": 0.0958198606967926, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3795293360948563, + "rkd_teacher_entropy": 3.7574335098266602, + "rkd_teacher_similarity_mean": 0.7853854238986969, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27130 + }, + { + "epoch": 7.896421297643293, + "flow_matching_loss": 0.011492186784744262, + "grad_norm": 0.10081139206886292, + "learning_rate": 7.397831944839167e-07, + "loss": 0.0122, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007920686528086663, + "rkd_loss_weight": 0.09548651576042175, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37707874476909636, + "rkd_teacher_entropy": 3.748281717300415, + "rkd_teacher_similarity_mean": 0.7826813995838166, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27140 + }, + { + "epoch": 7.899330811754437, + "flow_matching_loss": 0.010356543771922588, + "grad_norm": 0.06305339187383652, + "learning_rate": 7.346632925167706e-07, + "loss": 0.011, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006905270973220468, + "rkd_loss_weight": 0.09515317156910896, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37762540876865386, + "rkd_teacher_entropy": 3.744967007637024, + "rkd_teacher_similarity_mean": 0.7827304840087891, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27150 + }, + { + "epoch": 7.902240325865581, + "flow_matching_loss": 0.010559307131916284, + "grad_norm": 0.10366988182067871, + "learning_rate": 7.295607242788566e-07, + "loss": 0.0113, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007748984359204769, + "rkd_loss_weight": 0.09481982812285424, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3713778585195541, + "rkd_teacher_entropy": 3.723267126083374, + "rkd_teacher_similarity_mean": 0.7766313433647156, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27160 + }, + { + "epoch": 7.905149839976724, + "flow_matching_loss": 0.011365539254620671, + "grad_norm": 0.08557479083538055, + "learning_rate": 7.244754959702765e-07, + "loss": 0.012, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006461550295352936, + "rkd_loss_weight": 0.09448648244142532, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37617592215538026, + "rkd_teacher_entropy": 3.740343165397644, + "rkd_teacher_similarity_mean": 0.7824517965316773, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27170 + }, + { + "epoch": 7.908059354087867, + "flow_matching_loss": 0.013802455179393292, + "grad_norm": 0.15374019742012024, + "learning_rate": 7.194076137700628e-07, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007838268205523491, + "rkd_loss_weight": 0.09415313825011254, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38013227581977843, + "rkd_teacher_entropy": 3.756909728050232, + "rkd_teacher_similarity_mean": 0.7858552038669586, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27180 + }, + { + "epoch": 7.910968868199011, + "flow_matching_loss": 0.01230579623952508, + "grad_norm": 0.09914897382259369, + "learning_rate": 7.143570838361696e-07, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007787945540621876, + "rkd_loss_weight": 0.09381979405879974, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3795024245977402, + "rkd_teacher_entropy": 3.7560014724731445, + "rkd_teacher_similarity_mean": 0.7850481331348419, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27190 + }, + { + "epoch": 7.913878382310154, + "flow_matching_loss": 0.012173656281083823, + "grad_norm": 0.09816098213195801, + "learning_rate": 7.093239123054696e-07, + "loss": 0.0129, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007477201474830508, + "rkd_loss_weight": 0.09348645061254501, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3771630525588989, + "rkd_teacher_entropy": 3.7496586561203005, + "rkd_teacher_similarity_mean": 0.782274317741394, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27200 + }, + { + "epoch": 7.916787896421297, + "flow_matching_loss": 0.01301913196220994, + "grad_norm": 0.09743443876504898, + "learning_rate": 7.043081052937389e-07, + "loss": 0.0138, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008255389519035816, + "rkd_loss_weight": 0.09315310418605804, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3813410609960556, + "rkd_teacher_entropy": 3.7593887090682983, + "rkd_teacher_similarity_mean": 0.7868609249591827, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27210 + }, + { + "epoch": 7.919697410532441, + "flow_matching_loss": 0.011328407609835268, + "grad_norm": 0.08672679960727692, + "learning_rate": 6.993096688956558e-07, + "loss": 0.012, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007518420275300741, + "rkd_loss_weight": 0.09281975999474526, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38592447638511657, + "rkd_teacher_entropy": 3.764736866950989, + "rkd_teacher_similarity_mean": 0.7924607872962952, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27220 + }, + { + "epoch": 7.922606924643585, + "flow_matching_loss": 0.01114275106228888, + "grad_norm": 0.11616432666778564, + "learning_rate": 6.943286091847884e-07, + "loss": 0.0119, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007878144271671772, + "rkd_loss_weight": 0.09248641654849052, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3701056808233261, + "rkd_teacher_entropy": 3.7223669290542603, + "rkd_teacher_similarity_mean": 0.775535935163498, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27230 + }, + { + "epoch": 7.925516438754728, + "flow_matching_loss": 0.011889069434255361, + "grad_norm": 0.09128065407276154, + "learning_rate": 6.89364932213602e-07, + "loss": 0.0125, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007118348451331258, + "rkd_loss_weight": 0.09215307235717773, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37432610988616943, + "rkd_teacher_entropy": 3.737179231643677, + "rkd_teacher_similarity_mean": 0.7788900256156921, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27240 + }, + { + "epoch": 7.928425952865871, + "flow_matching_loss": 0.012903844565153122, + "grad_norm": 0.10495377331972122, + "learning_rate": 6.844186440134253e-07, + "loss": 0.0135, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.005975837260484696, + "rkd_loss_weight": 0.09181972667574882, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37696962654590604, + "rkd_teacher_entropy": 3.7445042610168455, + "rkd_teacher_similarity_mean": 0.7834719121456146, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27250 + }, + { + "epoch": 7.931335466977015, + "flow_matching_loss": 0.013004620326682925, + "grad_norm": 0.07694404572248459, + "learning_rate": 6.794897505944692e-07, + "loss": 0.0137, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008057945733889937, + "rkd_loss_weight": 0.09148638248443604, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37114075720310213, + "rkd_teacher_entropy": 3.7247594594955444, + "rkd_teacher_similarity_mean": 0.7765421032905578, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27260 + }, + { + "epoch": 7.934244981088158, + "flow_matching_loss": 0.010887104459106922, + "grad_norm": 0.08903307467699051, + "learning_rate": 6.745782579457993e-07, + "loss": 0.0115, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006175591517239809, + "rkd_loss_weight": 0.09115303903818131, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38009274303913115, + "rkd_teacher_entropy": 3.7511270523071287, + "rkd_teacher_similarity_mean": 0.7853686332702636, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27270 + }, + { + "epoch": 7.937154495199302, + "flow_matching_loss": 0.012627649120986462, + "grad_norm": 0.06568225473165512, + "learning_rate": 6.696841720353453e-07, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007564012054353952, + "rkd_loss_weight": 0.09081969410181046, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3737136423587799, + "rkd_teacher_entropy": 3.736636233329773, + "rkd_teacher_similarity_mean": 0.7793459296226501, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27280 + }, + { + "epoch": 7.940064009310445, + "flow_matching_loss": 0.012260021548718214, + "grad_norm": 0.09087016433477402, + "learning_rate": 6.648074988098834e-07, + "loss": 0.0129, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006714490102604031, + "rkd_loss_weight": 0.0904863491654396, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3820734560489655, + "rkd_teacher_entropy": 3.7687615156173706, + "rkd_teacher_similarity_mean": 0.7880788028240204, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27290 + }, + { + "epoch": 7.942973523421589, + "flow_matching_loss": 0.012259822292253374, + "grad_norm": 0.08204719424247742, + "learning_rate": 6.599482441950322e-07, + "loss": 0.0129, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007293884456157684, + "rkd_loss_weight": 0.09015300497412682, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3808486610651016, + "rkd_teacher_entropy": 3.7521082162857056, + "rkd_teacher_similarity_mean": 0.7864771068096161, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27300 + }, + { + "epoch": 7.945883037532732, + "flow_matching_loss": 0.012185277044773101, + "grad_norm": 0.08639559149742126, + "learning_rate": 6.551064140952417e-07, + "loss": 0.0128, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007206818787381053, + "rkd_loss_weight": 0.08981966152787209, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3679603099822998, + "rkd_teacher_entropy": 3.7165259599685667, + "rkd_teacher_similarity_mean": 0.7725911855697631, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27310 + }, + { + "epoch": 7.9487925516438755, + "flow_matching_loss": 0.012542311148718, + "grad_norm": 0.11868185549974442, + "learning_rate": 6.502820143937943e-07, + "loss": 0.0131, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0059214777313172815, + "rkd_loss_weight": 0.08948631584644318, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37209259867668154, + "rkd_teacher_entropy": 3.7264490365982055, + "rkd_teacher_similarity_mean": 0.7775718569755554, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27320 + }, + { + "epoch": 7.951702065755019, + "flow_matching_loss": 0.011859321454539894, + "grad_norm": 0.08960586786270142, + "learning_rate": 6.454750509527874e-07, + "loss": 0.0125, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0070394123904407024, + "rkd_loss_weight": 0.08915297091007232, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37411411106586456, + "rkd_teacher_entropy": 3.7306084871292113, + "rkd_teacher_similarity_mean": 0.7800465643405914, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27330 + }, + { + "epoch": 7.954611579866162, + "flow_matching_loss": 0.012266021734103561, + "grad_norm": 0.0553118959069252, + "learning_rate": 6.406855296131387e-07, + "loss": 0.0131, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00915422597900033, + "rkd_loss_weight": 0.0888196274638176, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3746532052755356, + "rkd_teacher_entropy": 3.7477948904037475, + "rkd_teacher_similarity_mean": 0.7799176454544068, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27340 + }, + { + "epoch": 7.957521093977306, + "flow_matching_loss": 0.011902372259646654, + "grad_norm": 0.09814677387475967, + "learning_rate": 6.359134561945662e-07, + "loss": 0.0125, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00685414420440793, + "rkd_loss_weight": 0.08848628401756287, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37928272783756256, + "rkd_teacher_entropy": 3.74308705329895, + "rkd_teacher_similarity_mean": 0.7851034700870514, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27350 + }, + { + "epoch": 7.9604306080884495, + "flow_matching_loss": 0.01145589337684214, + "grad_norm": 0.10289360582828522, + "learning_rate": 6.311588364955883e-07, + "loss": 0.0121, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006856763921678066, + "rkd_loss_weight": 0.0881529375910759, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3809296727180481, + "rkd_teacher_entropy": 3.760587239265442, + "rkd_teacher_similarity_mean": 0.7864610314369201, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27360 + }, + { + "epoch": 7.963340122199592, + "flow_matching_loss": 0.010560317290946841, + "grad_norm": 0.09212930500507355, + "learning_rate": 6.264216762935187e-07, + "loss": 0.0113, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008051863964647054, + "rkd_loss_weight": 0.08781959339976311, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37817070782184603, + "rkd_teacher_entropy": 3.748538279533386, + "rkd_teacher_similarity_mean": 0.7833175539970398, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27370 + }, + { + "epoch": 7.966249636310736, + "flow_matching_loss": 0.011940168449655175, + "grad_norm": 0.1211598813533783, + "learning_rate": 6.21701981344453e-07, + "loss": 0.0126, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008019275963306427, + "rkd_loss_weight": 0.08748624995350837, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38011670112609863, + "rkd_teacher_entropy": 3.7659173250198363, + "rkd_teacher_similarity_mean": 0.7850517749786377, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27380 + }, + { + "epoch": 7.96915915042188, + "flow_matching_loss": 0.010082961106672884, + "grad_norm": 0.07243558764457703, + "learning_rate": 6.16999757383262e-07, + "loss": 0.0109, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009123369213193655, + "rkd_loss_weight": 0.08715290576219559, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37012394666671755, + "rkd_teacher_entropy": 3.7283925771713258, + "rkd_teacher_similarity_mean": 0.775076562166214, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27390 + }, + { + "epoch": 7.9720686645330225, + "flow_matching_loss": 0.01222692159935832, + "grad_norm": 0.09121600538492203, + "learning_rate": 6.123150101235952e-07, + "loss": 0.0128, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00663352245464921, + "rkd_loss_weight": 0.08681956008076668, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38091742992401123, + "rkd_teacher_entropy": 3.7560139179229735, + "rkd_teacher_similarity_mean": 0.7864520013332367, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27400 + }, + { + "epoch": 7.974978178644166, + "flow_matching_loss": 0.013112795632332563, + "grad_norm": 0.11717384308576584, + "learning_rate": 6.076477452578594e-07, + "loss": 0.0137, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007099655130878091, + "rkd_loss_weight": 0.08648621588945389, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37766543924808504, + "rkd_teacher_entropy": 3.7517886638641356, + "rkd_teacher_similarity_mean": 0.7838190913200378, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27410 + }, + { + "epoch": 7.97788769275531, + "flow_matching_loss": 0.010656890086829662, + "grad_norm": 0.09173481911420822, + "learning_rate": 6.029979684572262e-07, + "loss": 0.0114, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008092395635321737, + "rkd_loss_weight": 0.08615287244319916, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37807331085205076, + "rkd_teacher_entropy": 3.749593901634216, + "rkd_teacher_similarity_mean": 0.7841456532478333, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27420 + }, + { + "epoch": 7.980797206866454, + "flow_matching_loss": 0.012850819807499646, + "grad_norm": 0.10756661742925644, + "learning_rate": 5.983656853716097e-07, + "loss": 0.0135, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00743251102976501, + "rkd_loss_weight": 0.08581952750682831, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37463376820087435, + "rkd_teacher_entropy": 3.734402084350586, + "rkd_teacher_similarity_mean": 0.7794490694999695, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27430 + }, + { + "epoch": 7.9837067209775965, + "flow_matching_loss": 0.013066686876118182, + "grad_norm": 0.1034800112247467, + "learning_rate": 5.937509016296689e-07, + "loss": 0.0137, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006925565889105201, + "rkd_loss_weight": 0.0854861818253994, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37795881628990174, + "rkd_teacher_entropy": 3.7485029220581056, + "rkd_teacher_similarity_mean": 0.7837418556213379, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27440 + }, + { + "epoch": 7.98661623508874, + "flow_matching_loss": 0.014173048594966531, + "grad_norm": 0.09954072535037994, + "learning_rate": 5.89153622838805e-07, + "loss": 0.0149, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007975727086886764, + "rkd_loss_weight": 0.08515283837914467, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3748771846294403, + "rkd_teacher_entropy": 3.744842267036438, + "rkd_teacher_similarity_mean": 0.7806681394577026, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27450 + }, + { + "epoch": 7.989525749199884, + "flow_matching_loss": 0.013093519490212202, + "grad_norm": 0.0900537520647049, + "learning_rate": 5.845738545851431e-07, + "loss": 0.0138, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008817015634849667, + "rkd_loss_weight": 0.08481949493288994, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3799325615167618, + "rkd_teacher_entropy": 3.7508874893188477, + "rkd_teacher_similarity_mean": 0.7862448573112488, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27460 + }, + { + "epoch": 7.992435263311027, + "flow_matching_loss": 0.010756112588569521, + "grad_norm": 0.07526960968971252, + "learning_rate": 5.800116024335334e-07, + "loss": 0.0114, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008185861352831125, + "rkd_loss_weight": 0.08448614925146103, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37945695221424103, + "rkd_teacher_entropy": 3.7516884803771973, + "rkd_teacher_similarity_mean": 0.7850192785263062, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27470 + }, + { + "epoch": 7.99534477742217, + "flow_matching_loss": 0.011610667733475567, + "grad_norm": 0.12448565661907196, + "learning_rate": 5.75466871927543e-07, + "loss": 0.0122, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00686421561986208, + "rkd_loss_weight": 0.08415280431509017, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37050587832927706, + "rkd_teacher_entropy": 3.7235272407531737, + "rkd_teacher_similarity_mean": 0.7756999850273132, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27480 + }, + { + "epoch": 7.998254291533314, + "flow_matching_loss": 0.010289576696231961, + "grad_norm": 0.06077221408486366, + "learning_rate": 5.709396685894463e-07, + "loss": 0.0108, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.005515392869710922, + "rkd_loss_weight": 0.08381946086883545, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37783063054084776, + "rkd_teacher_entropy": 3.738024353981018, + "rkd_teacher_similarity_mean": 0.7830184698104858, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27490 + }, + { + "epoch": 8.001163805644458, + "flow_matching_loss": 0.010452114604413509, + "grad_norm": 0.09635799378156662, + "learning_rate": 5.664299979202264e-07, + "loss": 0.011, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0065967872273176905, + "rkd_loss_weight": 0.08348611742258072, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3800352215766907, + "rkd_teacher_entropy": 3.7579216718673707, + "rkd_teacher_similarity_mean": 0.7858185708522797, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27500 + }, + { + "epoch": 8.004073319755602, + "flow_matching_loss": 0.01216883142478764, + "grad_norm": 0.0855664387345314, + "learning_rate": 5.61937865399555e-07, + "loss": 0.0128, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007455602986738086, + "rkd_loss_weight": 0.08315277099609375, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37960188686847685, + "rkd_teacher_entropy": 3.7587877750396728, + "rkd_teacher_similarity_mean": 0.7850881040096283, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27510 + }, + { + "epoch": 8.006982833866743, + "flow_matching_loss": 0.013732689898461103, + "grad_norm": 0.1377425342798233, + "learning_rate": 5.574632764857974e-07, + "loss": 0.0143, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007133695064112544, + "rkd_loss_weight": 0.08281942680478097, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38595737516880035, + "rkd_teacher_entropy": 3.765540528297424, + "rkd_teacher_similarity_mean": 0.792149031162262, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27520 + }, + { + "epoch": 8.009892347977887, + "flow_matching_loss": 0.010836285864934325, + "grad_norm": 0.09555933624505997, + "learning_rate": 5.53006236616001e-07, + "loss": 0.0114, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007382035069167614, + "rkd_loss_weight": 0.08248608335852622, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38436340987682344, + "rkd_teacher_entropy": 3.771575379371643, + "rkd_teacher_similarity_mean": 0.7904638707637787, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27530 + }, + { + "epoch": 8.012801862089031, + "flow_matching_loss": 0.011887491401284932, + "grad_norm": 0.08922610431909561, + "learning_rate": 5.485667512058873e-07, + "loss": 0.0125, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007613953808322549, + "rkd_loss_weight": 0.08215273916721344, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3753383219242096, + "rkd_teacher_entropy": 3.7327155590057375, + "rkd_teacher_similarity_mean": 0.7809820234775543, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27540 + }, + { + "epoch": 8.015711376200175, + "flow_matching_loss": 0.01094581801444292, + "grad_norm": 0.07235146313905716, + "learning_rate": 5.441448256498544e-07, + "loss": 0.0115, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006352403713390231, + "rkd_loss_weight": 0.08181939274072647, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37540377378463746, + "rkd_teacher_entropy": 3.746286416053772, + "rkd_teacher_similarity_mean": 0.7818063795566559, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27550 + }, + { + "epoch": 8.018620890311318, + "flow_matching_loss": 0.013496697600930928, + "grad_norm": 0.10697628557682037, + "learning_rate": 5.397404653209559e-07, + "loss": 0.0141, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007283720886334777, + "rkd_loss_weight": 0.08148604929447174, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3808874309062958, + "rkd_teacher_entropy": 3.76474506855011, + "rkd_teacher_similarity_mean": 0.7871821761131287, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27560 + }, + { + "epoch": 8.021530404422462, + "flow_matching_loss": 0.012428773753345012, + "grad_norm": 0.10577893257141113, + "learning_rate": 5.35353675570906e-07, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006837353156879544, + "rkd_loss_weight": 0.08115270584821702, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37299837470054625, + "rkd_teacher_entropy": 3.7262189388275146, + "rkd_teacher_similarity_mean": 0.7782887876033783, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27570 + }, + { + "epoch": 8.024439918533606, + "flow_matching_loss": 0.009878268279135228, + "grad_norm": 0.08125673234462738, + "learning_rate": 5.309844617300674e-07, + "loss": 0.0104, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006883008545264602, + "rkd_loss_weight": 0.08081936091184616, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3768969506025314, + "rkd_teacher_entropy": 3.753406596183777, + "rkd_teacher_similarity_mean": 0.7831932842731476, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27580 + }, + { + "epoch": 8.027349432644748, + "flow_matching_loss": 0.012049136869609355, + "grad_norm": 0.09235331416130066, + "learning_rate": 5.266328291074468e-07, + "loss": 0.0126, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007453192630782724, + "rkd_loss_weight": 0.08048601523041725, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38220539689064026, + "rkd_teacher_entropy": 3.760468339920044, + "rkd_teacher_similarity_mean": 0.7890452802181244, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27590 + }, + { + "epoch": 8.030258946755891, + "flow_matching_loss": 0.012199094891548157, + "grad_norm": 0.1085914745926857, + "learning_rate": 5.222987829906861e-07, + "loss": 0.0127, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006644988153129816, + "rkd_loss_weight": 0.08015267178416252, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38830490708351134, + "rkd_teacher_entropy": 3.7702593564987184, + "rkd_teacher_similarity_mean": 0.7949404895305634, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27600 + }, + { + "epoch": 8.033168460867035, + "flow_matching_loss": 0.010449245572090149, + "grad_norm": 0.10482793301343918, + "learning_rate": 5.179823286460644e-07, + "loss": 0.0111, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007535770162940025, + "rkd_loss_weight": 0.0798193283379078, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37576407194137573, + "rkd_teacher_entropy": 3.740163469314575, + "rkd_teacher_similarity_mean": 0.7814809143543243, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27610 + }, + { + "epoch": 8.036077974978179, + "flow_matching_loss": 0.013426597556099295, + "grad_norm": 0.08245353400707245, + "learning_rate": 5.13683471318478e-07, + "loss": 0.0139, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0059948799200356005, + "rkd_loss_weight": 0.07948598265647888, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.378817218542099, + "rkd_teacher_entropy": 3.7572251319885255, + "rkd_teacher_similarity_mean": 0.7845935106277466, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27620 + }, + { + "epoch": 8.038987489089322, + "flow_matching_loss": 0.012822559755295514, + "grad_norm": 0.10051332414150238, + "learning_rate": 5.094022162314449e-07, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009404677012935281, + "rkd_loss_weight": 0.07915263772010803, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3723642349243164, + "rkd_teacher_entropy": 3.726223278045654, + "rkd_teacher_similarity_mean": 0.7774175703525543, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27630 + }, + { + "epoch": 8.041897003200466, + "flow_matching_loss": 0.013821684569120408, + "grad_norm": 0.10122719407081604, + "learning_rate": 5.051385685870924e-07, + "loss": 0.0143, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006587777426466346, + "rkd_loss_weight": 0.0788192942738533, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3869370847940445, + "rkd_teacher_entropy": 3.7804487466812136, + "rkd_teacher_similarity_mean": 0.7930172085762024, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27640 + }, + { + "epoch": 8.044806517311608, + "flow_matching_loss": 0.011746452702209354, + "grad_norm": 0.09864344447851181, + "learning_rate": 5.008925335661546e-07, + "loss": 0.0123, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007029896229505539, + "rkd_loss_weight": 0.07848595008254051, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3800327628850937, + "rkd_teacher_entropy": 3.7585609436035154, + "rkd_teacher_similarity_mean": 0.7858092069625855, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27650 + }, + { + "epoch": 8.047716031422752, + "flow_matching_loss": 0.011631215829402209, + "grad_norm": 0.0937078669667244, + "learning_rate": 4.966641163279662e-07, + "loss": 0.0123, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008483629627153277, + "rkd_loss_weight": 0.0781526044011116, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37643593549728394, + "rkd_teacher_entropy": 3.7420650720596313, + "rkd_teacher_similarity_mean": 0.7820482194423676, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27660 + }, + { + "epoch": 8.050625545533896, + "flow_matching_loss": 0.011317350016906858, + "grad_norm": 0.0799282044172287, + "learning_rate": 4.924533220104521e-07, + "loss": 0.0119, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007672242587432265, + "rkd_loss_weight": 0.07781926020979882, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37688024938106535, + "rkd_teacher_entropy": 3.743864893913269, + "rkd_teacher_similarity_mean": 0.7822041988372803, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27670 + }, + { + "epoch": 8.05353505964504, + "flow_matching_loss": 0.012150411540642381, + "grad_norm": 0.09227612614631653, + "learning_rate": 4.882601557301231e-07, + "loss": 0.0126, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006397382402792573, + "rkd_loss_weight": 0.07748591676354408, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3689422130584717, + "rkd_teacher_entropy": 3.7245369434356688, + "rkd_teacher_similarity_mean": 0.7754164099693298, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27680 + }, + { + "epoch": 8.056444573756183, + "flow_matching_loss": 0.014365149009972811, + "grad_norm": 0.09438900649547577, + "learning_rate": 4.840846225820717e-07, + "loss": 0.015, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00784529666416347, + "rkd_loss_weight": 0.07715257257223129, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3777476102113724, + "rkd_teacher_entropy": 3.7570874452590943, + "rkd_teacher_similarity_mean": 0.7839945614337921, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27690 + }, + { + "epoch": 8.059354087867327, + "flow_matching_loss": 0.010280849365517496, + "grad_norm": 0.07099520415067673, + "learning_rate": 4.79926727639966e-07, + "loss": 0.0108, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007290233112871647, + "rkd_loss_weight": 0.07681922614574432, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37809066474437714, + "rkd_teacher_entropy": 3.74787974357605, + "rkd_teacher_similarity_mean": 0.7845157980918884, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27700 + }, + { + "epoch": 8.06226360197847, + "flow_matching_loss": 0.012717403285205365, + "grad_norm": 0.0672406479716301, + "learning_rate": 4.7578647595603817e-07, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007775878114625811, + "rkd_loss_weight": 0.0764858826994896, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3750449985265732, + "rkd_teacher_entropy": 3.7399729013442995, + "rkd_teacher_similarity_mean": 0.7818020284175873, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27710 + }, + { + "epoch": 8.065173116089612, + "flow_matching_loss": 0.01600937657058239, + "grad_norm": 0.11646077781915665, + "learning_rate": 4.716638725610889e-07, + "loss": 0.0167, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00904202377423644, + "rkd_loss_weight": 0.07615253925323487, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37252111434936525, + "rkd_teacher_entropy": 3.738773775100708, + "rkd_teacher_similarity_mean": 0.7781358540058136, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27720 + }, + { + "epoch": 8.068082630200756, + "flow_matching_loss": 0.013151324400678278, + "grad_norm": 0.07489293813705444, + "learning_rate": 4.6755892246446606e-07, + "loss": 0.0138, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008477012068033219, + "rkd_loss_weight": 0.07581919431686401, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3634876191616058, + "rkd_teacher_entropy": 3.703768420219421, + "rkd_teacher_similarity_mean": 0.7689342260360718, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27730 + }, + { + "epoch": 8.0709921443119, + "flow_matching_loss": 0.011224389914423228, + "grad_norm": 0.09170624613761902, + "learning_rate": 4.634716306540715e-07, + "loss": 0.0118, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007506012171506882, + "rkd_loss_weight": 0.0754858486354351, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3844873487949371, + "rkd_teacher_entropy": 3.772533583641052, + "rkd_teacher_similarity_mean": 0.7907561480998992, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27740 + }, + { + "epoch": 8.073901658423043, + "flow_matching_loss": 0.012798356637358666, + "grad_norm": 0.09853336960077286, + "learning_rate": 4.5940200209635244e-07, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006552938558161259, + "rkd_loss_weight": 0.07515250518918037, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38324114978313445, + "rkd_teacher_entropy": 3.7665277481079102, + "rkd_teacher_similarity_mean": 0.7897216558456421, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27750 + }, + { + "epoch": 8.076811172534187, + "flow_matching_loss": 0.012429110892117023, + "grad_norm": 0.12593317031860352, + "learning_rate": 4.5535004173628657e-07, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0072724585421383384, + "rkd_loss_weight": 0.07481916099786759, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38182617723941803, + "rkd_teacher_entropy": 3.770838975906372, + "rkd_teacher_similarity_mean": 0.7885780334472656, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27760 + }, + { + "epoch": 8.07972068664533, + "flow_matching_loss": 0.010885826451703906, + "grad_norm": 0.09162618964910507, + "learning_rate": 4.5131575449739225e-07, + "loss": 0.0115, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007657867297530175, + "rkd_loss_weight": 0.07448581606149673, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3829422652721405, + "rkd_teacher_entropy": 3.769458055496216, + "rkd_teacher_similarity_mean": 0.7888712704181671, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27770 + }, + { + "epoch": 8.082630200756475, + "flow_matching_loss": 0.012704721745103597, + "grad_norm": 0.06321751326322556, + "learning_rate": 4.472991452817099e-07, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007744645141065121, + "rkd_loss_weight": 0.07415247112512588, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37434032559394836, + "rkd_teacher_entropy": 3.7365039825439452, + "rkd_teacher_similarity_mean": 0.7808421313762665, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27780 + }, + { + "epoch": 8.085539714867616, + "flow_matching_loss": 0.0124601892195642, + "grad_norm": 0.07760682702064514, + "learning_rate": 4.433002189697938e-07, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007384664379060268, + "rkd_loss_weight": 0.07381912767887115, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3787922441959381, + "rkd_teacher_entropy": 3.7553478956222532, + "rkd_teacher_similarity_mean": 0.7850870072841645, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27790 + }, + { + "epoch": 8.08844922897876, + "flow_matching_loss": 0.013399427058175206, + "grad_norm": 0.11772481352090836, + "learning_rate": 4.393189804207187e-07, + "loss": 0.014, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008644142094999552, + "rkd_loss_weight": 0.07348578348755837, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3753668308258057, + "rkd_teacher_entropy": 3.737457275390625, + "rkd_teacher_similarity_mean": 0.7812995374202728, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27800 + }, + { + "epoch": 8.091358743089904, + "flow_matching_loss": 0.012296863086521625, + "grad_norm": 0.09585131704807281, + "learning_rate": 4.353554344720651e-07, + "loss": 0.0129, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007947205007076264, + "rkd_loss_weight": 0.07315243780612946, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36670062839984896, + "rkd_teacher_entropy": 3.722439932823181, + "rkd_teacher_similarity_mean": 0.7737090528011322, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27810 + }, + { + "epoch": 8.094268257201048, + "flow_matching_loss": 0.013239311147481203, + "grad_norm": 0.07760980725288391, + "learning_rate": 4.314095859399153e-07, + "loss": 0.0138, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008321277797222137, + "rkd_loss_weight": 0.07281909361481667, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37797881960868834, + "rkd_teacher_entropy": 3.744307208061218, + "rkd_teacher_similarity_mean": 0.7846023499965668, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27820 + }, + { + "epoch": 8.097177771312191, + "flow_matching_loss": 0.014001683983951806, + "grad_norm": 0.08051418513059616, + "learning_rate": 4.274814396188459e-07, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007537041837349534, + "rkd_loss_weight": 0.07248575016856193, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3747942179441452, + "rkd_teacher_entropy": 3.7562562227249146, + "rkd_teacher_similarity_mean": 0.7814399063587188, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27830 + }, + { + "epoch": 8.100087285423335, + "flow_matching_loss": 0.010434210253879427, + "grad_norm": 0.10228188335895538, + "learning_rate": 4.235710002819271e-07, + "loss": 0.011, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008336171368137002, + "rkd_loss_weight": 0.07215240597724915, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37563187479972837, + "rkd_teacher_entropy": 3.7598838806152344, + "rkd_teacher_similarity_mean": 0.7817158222198486, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27840 + }, + { + "epoch": 8.102996799534477, + "flow_matching_loss": 0.011679781042039395, + "grad_norm": 0.0877297967672348, + "learning_rate": 4.196782726807097e-07, + "loss": 0.0123, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007966931303963065, + "rkd_loss_weight": 0.07181905955076218, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36525385081768036, + "rkd_teacher_entropy": 3.7074559688568116, + "rkd_teacher_similarity_mean": 0.771511310338974, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27850 + }, + { + "epoch": 8.10590631364562, + "flow_matching_loss": 0.011412024917080998, + "grad_norm": 0.15578900277614594, + "learning_rate": 4.158032615452251e-07, + "loss": 0.0119, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006387532781809568, + "rkd_loss_weight": 0.07148571610450745, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38222997188568114, + "rkd_teacher_entropy": 3.7695095539093018, + "rkd_teacher_similarity_mean": 0.7887200653553009, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27860 + }, + { + "epoch": 8.108815827756764, + "flow_matching_loss": 0.012526459991931915, + "grad_norm": 0.08390897512435913, + "learning_rate": 4.1194597158397853e-07, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00722263976931572, + "rkd_loss_weight": 0.07115237191319465, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3802606761455536, + "rkd_teacher_entropy": 3.7649447441101076, + "rkd_teacher_similarity_mean": 0.7869053483009338, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27870 + }, + { + "epoch": 8.111725341867908, + "flow_matching_loss": 0.012978680431842804, + "grad_norm": 0.09235383570194244, + "learning_rate": 4.081064074839408e-07, + "loss": 0.0135, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00732490848749876, + "rkd_loss_weight": 0.07081902772188187, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3747428089380264, + "rkd_teacher_entropy": 3.735640287399292, + "rkd_teacher_similarity_mean": 0.7808182954788208, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27880 + }, + { + "epoch": 8.114634855979052, + "flow_matching_loss": 0.013328887615352869, + "grad_norm": 0.14005883038043976, + "learning_rate": 4.042845739105433e-07, + "loss": 0.0139, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00762084499001503, + "rkd_loss_weight": 0.07048568204045295, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37925155758857726, + "rkd_teacher_entropy": 3.7588191747665407, + "rkd_teacher_similarity_mean": 0.7861069202423095, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27890 + }, + { + "epoch": 8.117544370090195, + "flow_matching_loss": 0.013157534506171941, + "grad_norm": 0.13935759663581848, + "learning_rate": 4.004804755076746e-07, + "loss": 0.0137, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008057761983945966, + "rkd_loss_weight": 0.07015233859419823, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37364334166049956, + "rkd_teacher_entropy": 3.7409740447998048, + "rkd_teacher_similarity_mean": 0.7796074688434601, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27900 + }, + { + "epoch": 8.12045388420134, + "flow_matching_loss": 0.013983591366559267, + "grad_norm": 0.10168354213237762, + "learning_rate": 3.9669411689767063e-07, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0073812420014292, + "rkd_loss_weight": 0.06981899440288544, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37964175939559935, + "rkd_teacher_entropy": 3.759015989303589, + "rkd_teacher_similarity_mean": 0.7858595788478852, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27910 + }, + { + "epoch": 8.123363398312481, + "flow_matching_loss": 0.014005259657278658, + "grad_norm": 0.09605920314788818, + "learning_rate": 3.929255026813194e-07, + "loss": 0.0145, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007547018630430102, + "rkd_loss_weight": 0.06948564946651459, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37238913476467134, + "rkd_teacher_entropy": 3.7298325538635253, + "rkd_teacher_similarity_mean": 0.778256094455719, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27920 + }, + { + "epoch": 8.126272912423625, + "flow_matching_loss": 0.010557828517630697, + "grad_norm": 0.10818608105182648, + "learning_rate": 3.8917463743783797e-07, + "loss": 0.0112, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008846822381019592, + "rkd_loss_weight": 0.06915230453014373, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3683426588773727, + "rkd_teacher_entropy": 3.724115991592407, + "rkd_teacher_similarity_mean": 0.7746507942676544, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27930 + }, + { + "epoch": 8.129182426534769, + "flow_matching_loss": 0.012170963548123836, + "grad_norm": 0.1419907659292221, + "learning_rate": 3.854415257248839e-07, + "loss": 0.0127, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0076496005989611145, + "rkd_loss_weight": 0.068818961083889, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37718028128147124, + "rkd_teacher_entropy": 3.7490233421325683, + "rkd_teacher_similarity_mean": 0.7845263600349426, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27940 + }, + { + "epoch": 8.132091940645912, + "flow_matching_loss": 0.011380449123680591, + "grad_norm": 0.11039046198129654, + "learning_rate": 3.8172617207853864e-07, + "loss": 0.0119, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007056768983602524, + "rkd_loss_weight": 0.06848561689257622, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3756575047969818, + "rkd_teacher_entropy": 3.7454076290130613, + "rkd_teacher_similarity_mean": 0.7818271040916442, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27950 + }, + { + "epoch": 8.135001454757056, + "flow_matching_loss": 0.012944484129548073, + "grad_norm": 0.12254052609205246, + "learning_rate": 3.7802858101330773e-07, + "loss": 0.0135, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007710828958079219, + "rkd_loss_weight": 0.06815227121114731, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37263765931129456, + "rkd_teacher_entropy": 3.7331079244613647, + "rkd_teacher_similarity_mean": 0.7784252643585206, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27960 + }, + { + "epoch": 8.1379109688682, + "flow_matching_loss": 0.011015136074274779, + "grad_norm": 0.1089547872543335, + "learning_rate": 3.743487570221121e-07, + "loss": 0.0115, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00732269873842597, + "rkd_loss_weight": 0.06781892701983452, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37693763971328736, + "rkd_teacher_entropy": 3.747089076042175, + "rkd_teacher_similarity_mean": 0.7822026491165162, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27970 + }, + { + "epoch": 8.140820482979342, + "flow_matching_loss": 0.011884321365505457, + "grad_norm": 0.09298048168420792, + "learning_rate": 3.7068670457628517e-07, + "loss": 0.0124, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007346872938796878, + "rkd_loss_weight": 0.06748558282852173, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37676735520362853, + "rkd_teacher_entropy": 3.749192500114441, + "rkd_teacher_similarity_mean": 0.7830807566642761, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27980 + }, + { + "epoch": 8.143729997090485, + "flow_matching_loss": 0.012850030465051532, + "grad_norm": 0.08013428002595901, + "learning_rate": 3.6704242812556576e-07, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00697497483342886, + "rkd_loss_weight": 0.067152239382267, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3770668625831604, + "rkd_teacher_entropy": 3.7585943698883058, + "rkd_teacher_similarity_mean": 0.782889449596405, + "rkd_vlm_temp": 0.10000000149011612, + "step": 27990 + }, + { + "epoch": 8.146639511201629, + "flow_matching_loss": 0.013049266673624516, + "grad_norm": 0.09153623133897781, + "learning_rate": 3.634159320980951e-07, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008296709833666682, + "rkd_loss_weight": 0.06681889295578003, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38636381924152374, + "rkd_teacher_entropy": 3.7711677312850953, + "rkd_teacher_similarity_mean": 0.7913193047046662, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28000 + }, + { + "epoch": 8.149549025312773, + "flow_matching_loss": 0.014199863467365504, + "grad_norm": 0.08771390467882156, + "learning_rate": 3.5980722090040663e-07, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006742555648088455, + "rkd_loss_weight": 0.0664855495095253, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3787985950708389, + "rkd_teacher_entropy": 3.7542370557785034, + "rkd_teacher_similarity_mean": 0.7851855039596558, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28010 + }, + { + "epoch": 8.152458539423916, + "flow_matching_loss": 0.011401322577148676, + "grad_norm": 0.08655243366956711, + "learning_rate": 3.5621629891742447e-07, + "loss": 0.0119, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007298422511667013, + "rkd_loss_weight": 0.0661522053182125, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3742612272500992, + "rkd_teacher_entropy": 3.7383099794387817, + "rkd_teacher_similarity_mean": 0.7805112957954407, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28020 + }, + { + "epoch": 8.15536805353506, + "flow_matching_loss": 0.011120184883475303, + "grad_norm": 0.12045027315616608, + "learning_rate": 3.526431705124583e-07, + "loss": 0.0116, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007035684864968061, + "rkd_loss_weight": 0.06581886112689972, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3723204404115677, + "rkd_teacher_entropy": 3.738250803947449, + "rkd_teacher_similarity_mean": 0.7781973302364349, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28030 + }, + { + "epoch": 8.158277567646204, + "flow_matching_loss": 0.01288265697658062, + "grad_norm": 0.11778733879327774, + "learning_rate": 3.4908784002719517e-07, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.005915430188179016, + "rkd_loss_weight": 0.06548551544547081, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37768196165561674, + "rkd_teacher_entropy": 3.742539572715759, + "rkd_teacher_similarity_mean": 0.7841030240058899, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28040 + }, + { + "epoch": 8.161187081757346, + "flow_matching_loss": 0.012076907558366656, + "grad_norm": 0.08262386918067932, + "learning_rate": 3.4555031178169773e-07, + "loss": 0.0126, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007556052366271615, + "rkd_loss_weight": 0.06515217199921608, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3764186531305313, + "rkd_teacher_entropy": 3.739666247367859, + "rkd_teacher_similarity_mean": 0.7831603944301605, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28050 + }, + { + "epoch": 8.16409659586849, + "flow_matching_loss": 0.012825012067332863, + "grad_norm": 0.12113137543201447, + "learning_rate": 3.420305900743942e-07, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007100804289802909, + "rkd_loss_weight": 0.0648188278079033, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37996485233306887, + "rkd_teacher_entropy": 3.759515881538391, + "rkd_teacher_similarity_mean": 0.7864659249782562, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28060 + }, + { + "epoch": 8.167006109979633, + "flow_matching_loss": 0.01056315889582038, + "grad_norm": 0.09020458906888962, + "learning_rate": 3.385286791820802e-07, + "loss": 0.011, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006757224048487842, + "rkd_loss_weight": 0.06448548287153244, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.380883526802063, + "rkd_teacher_entropy": 3.7477195262908936, + "rkd_teacher_similarity_mean": 0.7872355818748474, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28070 + }, + { + "epoch": 8.169915624090777, + "flow_matching_loss": 0.014100848417729139, + "grad_norm": 0.10124682635068893, + "learning_rate": 3.350445833599086e-07, + "loss": 0.0146, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0071668052580207585, + "rkd_loss_weight": 0.06415213793516159, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36915277540683744, + "rkd_teacher_entropy": 3.725750541687012, + "rkd_teacher_similarity_mean": 0.7742494642734528, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28080 + }, + { + "epoch": 8.17282513820192, + "flow_matching_loss": 0.012845045048743486, + "grad_norm": 0.13097211718559265, + "learning_rate": 3.3157830684138303e-07, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006652799854055047, + "rkd_loss_weight": 0.0638187937438488, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3753725826740265, + "rkd_teacher_entropy": 3.7412886381149293, + "rkd_teacher_similarity_mean": 0.7817744433879852, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28090 + }, + { + "epoch": 8.175734652313064, + "flow_matching_loss": 0.011201358865946531, + "grad_norm": 0.07503959536552429, + "learning_rate": 3.281298538383559e-07, + "loss": 0.0116, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006984984921291471, + "rkd_loss_weight": 0.06348545029759407, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3788801670074463, + "rkd_teacher_entropy": 3.7497955799102782, + "rkd_teacher_similarity_mean": 0.7853199660778045, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28100 + }, + { + "epoch": 8.178644166424208, + "flow_matching_loss": 0.014749286510050297, + "grad_norm": 0.08936253190040588, + "learning_rate": 3.2469922854102387e-07, + "loss": 0.0152, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007710709515959024, + "rkd_loss_weight": 0.06315210461616516, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37436450719833375, + "rkd_teacher_entropy": 3.745198702812195, + "rkd_teacher_similarity_mean": 0.7811429321765899, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28110 + }, + { + "epoch": 8.18155368053535, + "flow_matching_loss": 0.01200278652831912, + "grad_norm": 0.11608058214187622, + "learning_rate": 3.21286435117919e-07, + "loss": 0.0124, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006663683149963617, + "rkd_loss_weight": 0.06281876042485238, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38206765055656433, + "rkd_teacher_entropy": 3.7565555810928344, + "rkd_teacher_similarity_mean": 0.7875523746013642, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28120 + }, + { + "epoch": 8.184463194646494, + "flow_matching_loss": 0.013156946329399944, + "grad_norm": 0.10412732511758804, + "learning_rate": 3.1789147771590754e-07, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007094581890851259, + "rkd_loss_weight": 0.06248541735112667, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3716495096683502, + "rkd_teacher_entropy": 3.7397642612457274, + "rkd_teacher_similarity_mean": 0.7780215501785278, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28130 + }, + { + "epoch": 8.187372708757637, + "flow_matching_loss": 0.01273508914746344, + "grad_norm": 0.06330526620149612, + "learning_rate": 3.14514360460183e-07, + "loss": 0.0132, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007889949204400181, + "rkd_loss_weight": 0.06215207129716873, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38106344938278197, + "rkd_teacher_entropy": 3.753018689155579, + "rkd_teacher_similarity_mean": 0.7870179235935211, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28140 + }, + { + "epoch": 8.190282222868781, + "flow_matching_loss": 0.012453560205176472, + "grad_norm": 0.07817783951759338, + "learning_rate": 3.1115508745426293e-07, + "loss": 0.0129, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008022556686773897, + "rkd_loss_weight": 0.061818727478384974, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3811339229345322, + "rkd_teacher_entropy": 3.758006238937378, + "rkd_teacher_similarity_mean": 0.7878718912601471, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28150 + }, + { + "epoch": 8.193191736979925, + "flow_matching_loss": 0.01254758844152093, + "grad_norm": 0.07463248074054718, + "learning_rate": 3.0781366277997716e-07, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007909533800557256, + "rkd_loss_weight": 0.06148538254201412, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3712663918733597, + "rkd_teacher_entropy": 3.720069694519043, + "rkd_teacher_similarity_mean": 0.7772905349731445, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28160 + }, + { + "epoch": 8.196101251091068, + "flow_matching_loss": 0.01284175287000835, + "grad_norm": 0.07620598375797272, + "learning_rate": 3.044900904974762e-07, + "loss": 0.0134, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009168790373951196, + "rkd_loss_weight": 0.06115203872323036, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3804387301206589, + "rkd_teacher_entropy": 3.7700170278549194, + "rkd_teacher_similarity_mean": 0.7866118788719177, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28170 + }, + { + "epoch": 8.19901076520221, + "flow_matching_loss": 0.011994805932044984, + "grad_norm": 0.11846686154603958, + "learning_rate": 3.011843746452081e-07, + "loss": 0.0124, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007009560149163008, + "rkd_loss_weight": 0.060818693786859515, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.376235693693161, + "rkd_teacher_entropy": 3.7410850286483766, + "rkd_teacher_similarity_mean": 0.7825799405574798, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28180 + }, + { + "epoch": 8.201920279313354, + "flow_matching_loss": 0.012218388449400663, + "grad_norm": 0.10856499522924423, + "learning_rate": 2.9789651923993464e-07, + "loss": 0.0126, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006613591453060508, + "rkd_loss_weight": 0.060485349595546724, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3663996309041977, + "rkd_teacher_entropy": 3.7067148447036744, + "rkd_teacher_similarity_mean": 0.7716993808746337, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28190 + }, + { + "epoch": 8.204829793424498, + "flow_matching_loss": 0.009534493694081903, + "grad_norm": 0.0910472422838211, + "learning_rate": 2.9462652827670855e-07, + "loss": 0.01, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007131696445867419, + "rkd_loss_weight": 0.060152005031704904, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3700011968612671, + "rkd_teacher_entropy": 3.7202988862991333, + "rkd_teacher_similarity_mean": 0.7772412419319152, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28200 + }, + { + "epoch": 8.207739307535642, + "flow_matching_loss": 0.01096996576525271, + "grad_norm": 0.06607803702354431, + "learning_rate": 2.913744057288764e-07, + "loss": 0.0114, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007333104545250535, + "rkd_loss_weight": 0.05981866084039211, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37606361508369446, + "rkd_teacher_entropy": 3.7485987901687623, + "rkd_teacher_similarity_mean": 0.7818496108055115, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28210 + }, + { + "epoch": 8.210648821646785, + "flow_matching_loss": 0.01106551494449377, + "grad_norm": 0.07378891110420227, + "learning_rate": 2.8814015554807396e-07, + "loss": 0.0115, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007178425835445523, + "rkd_loss_weight": 0.059485315904021265, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38258591294288635, + "rkd_teacher_entropy": 3.7580637454986574, + "rkd_teacher_similarity_mean": 0.788899838924408, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28220 + }, + { + "epoch": 8.213558335757929, + "flow_matching_loss": 0.010140216443687677, + "grad_norm": 0.0921780914068222, + "learning_rate": 2.849237816642192e-07, + "loss": 0.0105, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006039319559931755, + "rkd_loss_weight": 0.05915197171270847, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3777262717485428, + "rkd_teacher_entropy": 3.738680696487427, + "rkd_teacher_similarity_mean": 0.7840020298957825, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28230 + }, + { + "epoch": 8.216467849869073, + "flow_matching_loss": 0.012028932385146619, + "grad_norm": 0.07859889417886734, + "learning_rate": 2.817252879855109e-07, + "loss": 0.0124, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006420651823282242, + "rkd_loss_weight": 0.05881862752139568, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3756225109100342, + "rkd_teacher_entropy": 3.7436981201171875, + "rkd_teacher_similarity_mean": 0.7818060517311096, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28240 + }, + { + "epoch": 8.219377363980215, + "flow_matching_loss": 0.01270127878524363, + "grad_norm": 0.10167451947927475, + "learning_rate": 2.785446783984219e-07, + "loss": 0.0131, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007180586969479918, + "rkd_loss_weight": 0.058485282585024834, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3737993687391281, + "rkd_teacher_entropy": 3.7317977428436278, + "rkd_teacher_similarity_mean": 0.7795741498470307, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28250 + }, + { + "epoch": 8.222286878091358, + "flow_matching_loss": 0.013503658957779408, + "grad_norm": 0.0979238823056221, + "learning_rate": 2.7538195676769076e-07, + "loss": 0.0139, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0072084711864590645, + "rkd_loss_weight": 0.05815193839371204, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3801179796457291, + "rkd_teacher_entropy": 3.7550862312316893, + "rkd_teacher_similarity_mean": 0.7868933260440827, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28260 + }, + { + "epoch": 8.225196392202502, + "flow_matching_loss": 0.013368216343224049, + "grad_norm": 0.12468547374010086, + "learning_rate": 2.722371269363216e-07, + "loss": 0.0138, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00785008752718568, + "rkd_loss_weight": 0.05781859382987022, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37824691236019137, + "rkd_teacher_entropy": 3.7511904716491697, + "rkd_teacher_similarity_mean": 0.7843807339668274, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28270 + }, + { + "epoch": 8.228105906313646, + "flow_matching_loss": 0.01198377450928092, + "grad_norm": 0.07474357634782791, + "learning_rate": 2.6911019272558125e-07, + "loss": 0.0123, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006299699330702424, + "rkd_loss_weight": 0.05748525001108647, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37973593175411224, + "rkd_teacher_entropy": 3.7582555532455446, + "rkd_teacher_similarity_mean": 0.7859557688236236, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28280 + }, + { + "epoch": 8.23101542042479, + "flow_matching_loss": 0.011875295592471958, + "grad_norm": 0.09533700346946716, + "learning_rate": 2.6600115793498857e-07, + "loss": 0.0122, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0058754303492605684, + "rkd_loss_weight": 0.05715190470218658, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3876366585493088, + "rkd_teacher_entropy": 3.775562810897827, + "rkd_teacher_similarity_mean": 0.7944239854812623, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28290 + }, + { + "epoch": 8.233924934535933, + "flow_matching_loss": 0.011803085776045918, + "grad_norm": 0.10510502755641937, + "learning_rate": 2.629100263423151e-07, + "loss": 0.0123, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008446428366005421, + "rkd_loss_weight": 0.05681856088340283, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3700870871543884, + "rkd_teacher_entropy": 3.727730417251587, + "rkd_teacher_similarity_mean": 0.7771398901939393, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28300 + }, + { + "epoch": 8.236834448647077, + "flow_matching_loss": 0.014885173551738263, + "grad_norm": 0.09859117865562439, + "learning_rate": 2.598368017035763e-07, + "loss": 0.0153, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0076689396519213915, + "rkd_loss_weight": 0.05648521594703197, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3732260078191757, + "rkd_teacher_entropy": 3.734542179107666, + "rkd_teacher_similarity_mean": 0.7792669117450715, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28310 + }, + { + "epoch": 8.239743962758219, + "flow_matching_loss": 0.01075512864626944, + "grad_norm": 0.058684125542640686, + "learning_rate": 2.5678148775303e-07, + "loss": 0.0112, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007945879315957426, + "rkd_loss_weight": 0.056151872128248216, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37609375417232516, + "rkd_teacher_entropy": 3.7531072378158568, + "rkd_teacher_similarity_mean": 0.7830525279045105, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28320 + }, + { + "epoch": 8.242653476869362, + "flow_matching_loss": 0.011042852653190494, + "grad_norm": 0.08635207265615463, + "learning_rate": 2.537440882031683e-07, + "loss": 0.0115, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007791991997510195, + "rkd_loss_weight": 0.05581852681934833, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3787707030773163, + "rkd_teacher_entropy": 3.7592215538024902, + "rkd_teacher_similarity_mean": 0.7856899380683899, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28330 + }, + { + "epoch": 8.245562990980506, + "flow_matching_loss": 0.013252040557563304, + "grad_norm": 0.12013935297727585, + "learning_rate": 2.5072460674472385e-07, + "loss": 0.0137, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007281319517642259, + "rkd_loss_weight": 0.05548518300056458, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37998964786529543, + "rkd_teacher_entropy": 3.7652213096618654, + "rkd_teacher_similarity_mean": 0.7868000864982605, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28340 + }, + { + "epoch": 8.24847250509165, + "flow_matching_loss": 0.011186406249180436, + "grad_norm": 0.07380098104476929, + "learning_rate": 2.4772304704664683e-07, + "loss": 0.0116, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007858423655852676, + "rkd_loss_weight": 0.05515183843672276, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37601693868637087, + "rkd_teacher_entropy": 3.748474621772766, + "rkd_teacher_similarity_mean": 0.7817144215106964, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28350 + }, + { + "epoch": 8.251382019202794, + "flow_matching_loss": 0.012890602182596921, + "grad_norm": 0.10341385006904602, + "learning_rate": 2.447394127561181e-07, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007219045702368021, + "rkd_loss_weight": 0.05481849387288094, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3784365624189377, + "rkd_teacher_entropy": 3.7570704936981203, + "rkd_teacher_similarity_mean": 0.785050243139267, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28360 + }, + { + "epoch": 8.254291533313937, + "flow_matching_loss": 0.011273408168926835, + "grad_norm": 0.0881502777338028, + "learning_rate": 2.41773707498531e-07, + "loss": 0.0118, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00928212613798678, + "rkd_loss_weight": 0.05448514930903912, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3662202090024948, + "rkd_teacher_entropy": 3.7117252111434937, + "rkd_teacher_similarity_mean": 0.7719428956508636, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28370 + }, + { + "epoch": 8.25720104742508, + "flow_matching_loss": 0.012603648426011204, + "grad_norm": 0.08969157934188843, + "learning_rate": 2.388259348774996e-07, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0069840842392295596, + "rkd_loss_weight": 0.054151805117726326, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.373081237077713, + "rkd_teacher_entropy": 3.734459948539734, + "rkd_teacher_similarity_mean": 0.7796282887458801, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28380 + }, + { + "epoch": 8.260110561536223, + "flow_matching_loss": 0.011772137461230158, + "grad_norm": 0.08750943839550018, + "learning_rate": 2.358960984748454e-07, + "loss": 0.0121, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.005875503411516547, + "rkd_loss_weight": 0.053818460926413535, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38172579407691953, + "rkd_teacher_entropy": 3.7593195915222166, + "rkd_teacher_similarity_mean": 0.7886581122875214, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28390 + }, + { + "epoch": 8.263020075647367, + "flow_matching_loss": 0.01075462973676622, + "grad_norm": 0.10952269285917282, + "learning_rate": 2.3298420185059565e-07, + "loss": 0.0111, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006969721475616098, + "rkd_loss_weight": 0.053485115990042686, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3824943989515305, + "rkd_teacher_entropy": 3.7670379877090454, + "rkd_teacher_similarity_mean": 0.7882997155189514, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28400 + }, + { + "epoch": 8.26592958975851, + "flow_matching_loss": 0.01186542771756649, + "grad_norm": 0.07867731153964996, + "learning_rate": 2.3009024854297844e-07, + "loss": 0.0123, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007284539472311735, + "rkd_loss_weight": 0.053151771798729895, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36533392071723936, + "rkd_teacher_entropy": 3.7124507904052733, + "rkd_teacher_similarity_mean": 0.771656584739685, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28410 + }, + { + "epoch": 8.268839103869654, + "flow_matching_loss": 0.011183063313364983, + "grad_norm": 0.07150617241859436, + "learning_rate": 2.272142420684209e-07, + "loss": 0.0116, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007410629047080874, + "rkd_loss_weight": 0.052818427234888075, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3744475096464157, + "rkd_teacher_entropy": 3.746083378791809, + "rkd_teacher_similarity_mean": 0.7803472995758056, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28420 + }, + { + "epoch": 8.271748617980798, + "flow_matching_loss": 0.011806100560352206, + "grad_norm": 0.07561058551073074, + "learning_rate": 2.243561859215426e-07, + "loss": 0.0122, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006683843489736318, + "rkd_loss_weight": 0.052485083043575284, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3776226550340652, + "rkd_teacher_entropy": 3.7495290756225588, + "rkd_teacher_similarity_mean": 0.7841911017894745, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28430 + }, + { + "epoch": 8.274658132091941, + "flow_matching_loss": 0.01082800216972828, + "grad_norm": 0.11130575090646744, + "learning_rate": 2.2151608357514897e-07, + "loss": 0.0112, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007914782874286175, + "rkd_loss_weight": 0.052151738107204436, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37802245318889616, + "rkd_teacher_entropy": 3.7457903385162354, + "rkd_teacher_similarity_mean": 0.783665144443512, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28440 + }, + { + "epoch": 8.277567646203083, + "flow_matching_loss": 0.011210382916033268, + "grad_norm": 0.10557731240987778, + "learning_rate": 2.186939384802361e-07, + "loss": 0.0116, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006904402188956738, + "rkd_loss_weight": 0.05181839428842068, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3720190405845642, + "rkd_teacher_entropy": 3.737221026420593, + "rkd_teacher_similarity_mean": 0.778816419839859, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28450 + }, + { + "epoch": 8.280477160314227, + "flow_matching_loss": 0.01093646944500506, + "grad_norm": 0.09114987403154373, + "learning_rate": 2.1588975406597266e-07, + "loss": 0.0113, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007027358887717128, + "rkd_loss_weight": 0.051485049352049825, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3801101863384247, + "rkd_teacher_entropy": 3.7642085313797, + "rkd_teacher_similarity_mean": 0.7863068580627441, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28460 + }, + { + "epoch": 8.28338667442537, + "flow_matching_loss": 0.012624576780945062, + "grad_norm": 0.14740096032619476, + "learning_rate": 2.131035337397097e-07, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007119165500625968, + "rkd_loss_weight": 0.05115170516073704, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3724604040384293, + "rkd_teacher_entropy": 3.7360626220703126, + "rkd_teacher_similarity_mean": 0.7785394251346588, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28470 + }, + { + "epoch": 8.286296188536515, + "flow_matching_loss": 0.012294847285374999, + "grad_norm": 0.09495094418525696, + "learning_rate": 2.1033528088696575e-07, + "loss": 0.0126, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006504607619717717, + "rkd_loss_weight": 0.050818360224366185, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3752150028944016, + "rkd_teacher_entropy": 3.7486710786819457, + "rkd_teacher_similarity_mean": 0.7818016588687897, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28480 + }, + { + "epoch": 8.289205702647658, + "flow_matching_loss": 0.010395960975438356, + "grad_norm": 0.07735604792833328, + "learning_rate": 2.0758499887143012e-07, + "loss": 0.0108, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007449566386640072, + "rkd_loss_weight": 0.05048501640558243, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3726762056350708, + "rkd_teacher_entropy": 3.7445439100265503, + "rkd_teacher_similarity_mean": 0.7794829130172729, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28490 + }, + { + "epoch": 8.292115216758802, + "flow_matching_loss": 0.01512325182557106, + "grad_norm": 0.13756662607192993, + "learning_rate": 2.0485269103495462e-07, + "loss": 0.0156, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008587599731981754, + "rkd_loss_weight": 0.05015167184174061, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3774911820888519, + "rkd_teacher_entropy": 3.7548531532287597, + "rkd_teacher_similarity_mean": 0.7830513417720795, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28500 + }, + { + "epoch": 8.295024730869944, + "flow_matching_loss": 0.012023056019097566, + "grad_norm": 0.0754738450050354, + "learning_rate": 2.021383606975502e-07, + "loss": 0.0124, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007382175466045737, + "rkd_loss_weight": 0.04981832727789879, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3705852836370468, + "rkd_teacher_entropy": 3.740774893760681, + "rkd_teacher_similarity_mean": 0.7773898363113403, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28510 + }, + { + "epoch": 8.297934244981088, + "flow_matching_loss": 0.010572841390967369, + "grad_norm": 0.0578286312520504, + "learning_rate": 1.9944201115738524e-07, + "loss": 0.0109, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007453139498829841, + "rkd_loss_weight": 0.04948498271405697, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37860383093357086, + "rkd_teacher_entropy": 3.7520604372024535, + "rkd_teacher_similarity_mean": 0.7852066576480865, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28520 + }, + { + "epoch": 8.300843759092231, + "flow_matching_loss": 0.01035374402999878, + "grad_norm": 0.05935170501470566, + "learning_rate": 1.9676364569077732e-07, + "loss": 0.0107, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007160162925720215, + "rkd_loss_weight": 0.04915163852274418, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37984890043735503, + "rkd_teacher_entropy": 3.7464025259017943, + "rkd_teacher_similarity_mean": 0.7860241889953613, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28530 + }, + { + "epoch": 8.303753273203375, + "flow_matching_loss": 0.011937887826934456, + "grad_norm": 0.08071254193782806, + "learning_rate": 1.9410326755219477e-07, + "loss": 0.0123, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0074414990842342375, + "rkd_loss_weight": 0.04881829395890236, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37188871800899503, + "rkd_teacher_entropy": 3.7268539905548095, + "rkd_teacher_similarity_mean": 0.7778106451034545, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28540 + }, + { + "epoch": 8.306662787314519, + "flow_matching_loss": 0.011407050024718045, + "grad_norm": 0.08447825163602829, + "learning_rate": 1.9146087997424677e-07, + "loss": 0.0118, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008936416683718562, + "rkd_loss_weight": 0.04848494939506054, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3774318158626556, + "rkd_teacher_entropy": 3.7536117315292357, + "rkd_teacher_similarity_mean": 0.7841530919075013, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28550 + }, + { + "epoch": 8.309572301425662, + "flow_matching_loss": 0.010811090096831321, + "grad_norm": 0.10577377676963806, + "learning_rate": 1.8883648616768335e-07, + "loss": 0.0111, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006875270651653409, + "rkd_loss_weight": 0.04815160520374775, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38439969122409823, + "rkd_teacher_entropy": 3.758378839492798, + "rkd_teacher_similarity_mean": 0.7910309255123138, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28560 + }, + { + "epoch": 8.312481815536806, + "flow_matching_loss": 0.012713760975748301, + "grad_norm": 0.08552154153585434, + "learning_rate": 1.862300893213903e-07, + "loss": 0.0131, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00902555943466723, + "rkd_loss_weight": 0.04781826063990593, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37844640016555786, + "rkd_teacher_entropy": 3.7504715442657472, + "rkd_teacher_similarity_mean": 0.785857903957367, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28570 + }, + { + "epoch": 8.315391329647948, + "flow_matching_loss": 0.014326689718291163, + "grad_norm": 0.10852139443159103, + "learning_rate": 1.8364169260238595e-07, + "loss": 0.0147, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007422591699287296, + "rkd_loss_weight": 0.04748491607606411, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38050938546657564, + "rkd_teacher_entropy": 3.7585819244384764, + "rkd_teacher_similarity_mean": 0.7869416296482086, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28580 + }, + { + "epoch": 8.318300843759092, + "flow_matching_loss": 0.01193695580586791, + "grad_norm": 0.10270370543003082, + "learning_rate": 1.8107129915581777e-07, + "loss": 0.0123, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00679620117880404, + "rkd_loss_weight": 0.04715157151222229, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37606351971626284, + "rkd_teacher_entropy": 3.7487306118011476, + "rkd_teacher_similarity_mean": 0.7823517084121704, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28590 + }, + { + "epoch": 8.321210357870235, + "flow_matching_loss": 0.012910350691527128, + "grad_norm": 0.08784125745296478, + "learning_rate": 1.785189121049541e-07, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007344735134392977, + "rkd_loss_weight": 0.04681822769343853, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37617608308792116, + "rkd_teacher_entropy": 3.74461886882782, + "rkd_teacher_similarity_mean": 0.782601398229599, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28600 + }, + { + "epoch": 8.32411987198138, + "flow_matching_loss": 0.011030289763584732, + "grad_norm": 0.11254139989614487, + "learning_rate": 1.7598453455118567e-07, + "loss": 0.0114, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008766485145315528, + "rkd_loss_weight": 0.04648488238453865, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3728226125240326, + "rkd_teacher_entropy": 3.734792876243591, + "rkd_teacher_similarity_mean": 0.7789451360702515, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28610 + }, + { + "epoch": 8.327029386092523, + "flow_matching_loss": 0.01227530143223703, + "grad_norm": 0.07396378368139267, + "learning_rate": 1.7346816957402256e-07, + "loss": 0.0126, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006608425267040729, + "rkd_loss_weight": 0.04615153856575489, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3750007480382919, + "rkd_teacher_entropy": 3.7460850715637206, + "rkd_teacher_similarity_mean": 0.7816771507263184, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28620 + }, + { + "epoch": 8.329938900203667, + "flow_matching_loss": 0.012373499991372227, + "grad_norm": 0.08031122386455536, + "learning_rate": 1.7096982023108389e-07, + "loss": 0.0127, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006239244458265602, + "rkd_loss_weight": 0.04581819362938404, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3756211847066879, + "rkd_teacher_entropy": 3.747545313835144, + "rkd_teacher_similarity_mean": 0.7825293958187103, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28630 + }, + { + "epoch": 8.33284841431481, + "flow_matching_loss": 0.012712229182943702, + "grad_norm": 0.10036567598581314, + "learning_rate": 1.6848948955809963e-07, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006621928233653307, + "rkd_loss_weight": 0.04548484981060028, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38098135590553284, + "rkd_teacher_entropy": 3.765601110458374, + "rkd_teacher_similarity_mean": 0.787171459197998, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28640 + }, + { + "epoch": 8.335757928425952, + "flow_matching_loss": 0.010532679501920938, + "grad_norm": 0.07805584371089935, + "learning_rate": 1.66027180568904e-07, + "loss": 0.0109, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007095995312556625, + "rkd_loss_weight": 0.045151504874229434, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3731178879737854, + "rkd_teacher_entropy": 3.737197518348694, + "rkd_teacher_similarity_mean": 0.7790033400058747, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28650 + }, + { + "epoch": 8.338667442537096, + "flow_matching_loss": 0.011749699432402849, + "grad_norm": 0.11150839924812317, + "learning_rate": 1.6358289625543865e-07, + "loss": 0.0121, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008402782818302512, + "rkd_loss_weight": 0.04481816068291664, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36915053725242614, + "rkd_teacher_entropy": 3.726594018936157, + "rkd_teacher_similarity_mean": 0.7752720236778259, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28660 + }, + { + "epoch": 8.34157695664824, + "flow_matching_loss": 0.012803867924958467, + "grad_norm": 0.10486782342195511, + "learning_rate": 1.6115663958773618e-07, + "loss": 0.0131, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006799344252794981, + "rkd_loss_weight": 0.04448481611907482, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3733085870742798, + "rkd_teacher_entropy": 3.7334715127944946, + "rkd_teacher_similarity_mean": 0.7794034838676452, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28670 + }, + { + "epoch": 8.344486470759383, + "flow_matching_loss": 0.010051743779331445, + "grad_norm": 0.0851665511727333, + "learning_rate": 1.5874841351392832e-07, + "loss": 0.0103, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006071117566898465, + "rkd_loss_weight": 0.04415147192776203, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37753111124038696, + "rkd_teacher_entropy": 3.7488247632980345, + "rkd_teacher_similarity_mean": 0.7838576912879944, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28680 + }, + { + "epoch": 8.347395984870527, + "flow_matching_loss": 0.012514741439372302, + "grad_norm": 0.09908782690763474, + "learning_rate": 1.56358220960236e-07, + "loss": 0.0129, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008503400627523661, + "rkd_loss_weight": 0.04381812699139118, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37658863961696626, + "rkd_teacher_entropy": 3.746868896484375, + "rkd_teacher_similarity_mean": 0.7827234148979187, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28690 + }, + { + "epoch": 8.35030549898167, + "flow_matching_loss": 0.01187280435115099, + "grad_norm": 0.088261179625988, + "learning_rate": 1.5398606483096932e-07, + "loss": 0.0121, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006202051974833011, + "rkd_loss_weight": 0.04348478280007839, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3711105644702911, + "rkd_teacher_entropy": 3.733528900146484, + "rkd_teacher_similarity_mean": 0.7774519085884094, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28700 + }, + { + "epoch": 8.353215013092813, + "flow_matching_loss": 0.011428262991830706, + "grad_norm": 0.08372823148965836, + "learning_rate": 1.5163194800852265e-07, + "loss": 0.0117, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007316195499151945, + "rkd_loss_weight": 0.0431514386087656, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37545333206653597, + "rkd_teacher_entropy": 3.745968556404114, + "rkd_teacher_similarity_mean": 0.7811247646808624, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28710 + }, + { + "epoch": 8.356124527203956, + "flow_matching_loss": 0.012291114404797554, + "grad_norm": 0.08473807573318481, + "learning_rate": 1.4929587335336614e-07, + "loss": 0.0126, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007493793731555343, + "rkd_loss_weight": 0.04281809367239475, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3739624410867691, + "rkd_teacher_entropy": 3.7457905054092406, + "rkd_teacher_similarity_mean": 0.780257111787796, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28720 + }, + { + "epoch": 8.3590340413151, + "flow_matching_loss": 0.01045197369530797, + "grad_norm": 0.07685208320617676, + "learning_rate": 1.4697784370405587e-07, + "loss": 0.0107, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006627793028019369, + "rkd_loss_weight": 0.04248474948108196, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3754414230585098, + "rkd_teacher_entropy": 3.7434186220169066, + "rkd_teacher_similarity_mean": 0.7815900266170501, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28730 + }, + { + "epoch": 8.361943555426244, + "flow_matching_loss": 0.01174515075981617, + "grad_norm": 0.07816748321056366, + "learning_rate": 1.4467786187721376e-07, + "loss": 0.012, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00665253410115838, + "rkd_loss_weight": 0.04215140491724014, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38137098848819734, + "rkd_teacher_entropy": 3.7592029571533203, + "rkd_teacher_similarity_mean": 0.7874636948108673, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28740 + }, + { + "epoch": 8.364853069537387, + "flow_matching_loss": 0.012631174828857183, + "grad_norm": 0.10347677022218704, + "learning_rate": 1.4239593066753265e-07, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008285713894292713, + "rkd_loss_weight": 0.041818061098456386, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3795513242483139, + "rkd_teacher_entropy": 3.757315158843994, + "rkd_teacher_similarity_mean": 0.7857907950878144, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28750 + }, + { + "epoch": 8.367762583648531, + "flow_matching_loss": 0.013482477795332671, + "grad_norm": 0.07915369421243668, + "learning_rate": 1.4013205284777953e-07, + "loss": 0.0138, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007053913129493594, + "rkd_loss_weight": 0.0414847157895565, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3727234244346619, + "rkd_teacher_entropy": 3.7399656534194947, + "rkd_teacher_similarity_mean": 0.7785406231880188, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28760 + }, + { + "epoch": 8.370672097759675, + "flow_matching_loss": 0.011342685297131538, + "grad_norm": 0.1038823127746582, + "learning_rate": 1.3788623116877397e-07, + "loss": 0.0117, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00777679686434567, + "rkd_loss_weight": 0.041151371970772746, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3771396964788437, + "rkd_teacher_entropy": 3.745057201385498, + "rkd_teacher_similarity_mean": 0.7837725758552552, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28770 + }, + { + "epoch": 8.373581611870817, + "flow_matching_loss": 0.012921563209965825, + "grad_norm": 0.08515824377536774, + "learning_rate": 1.3565846835940477e-07, + "loss": 0.0132, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006476889504119754, + "rkd_loss_weight": 0.04081802703440189, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37758620381355285, + "rkd_teacher_entropy": 3.7563134908676146, + "rkd_teacher_similarity_mean": 0.7844019949436187, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28780 + }, + { + "epoch": 8.37649112598196, + "flow_matching_loss": 0.010949917370453477, + "grad_norm": 0.10397990792989731, + "learning_rate": 1.334487671266116e-07, + "loss": 0.0112, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007124114921316505, + "rkd_loss_weight": 0.040484683215618135, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37428476512432096, + "rkd_teacher_entropy": 3.750127983093262, + "rkd_teacher_similarity_mean": 0.7807576656341553, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28790 + }, + { + "epoch": 8.379400640093104, + "flow_matching_loss": 0.011604399513453245, + "grad_norm": 0.07367413491010666, + "learning_rate": 1.3125713015538998e-07, + "loss": 0.012, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00860914564691484, + "rkd_loss_weight": 0.04015133790671825, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3699904352426529, + "rkd_teacher_entropy": 3.7363729000091555, + "rkd_teacher_similarity_mean": 0.7764835774898529, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28800 + }, + { + "epoch": 8.382310154204248, + "flow_matching_loss": 0.012970435665920377, + "grad_norm": 0.08547332137823105, + "learning_rate": 1.290835601087864e-07, + "loss": 0.0132, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006833630474284291, + "rkd_loss_weight": 0.039817994087934495, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37643773257732394, + "rkd_teacher_entropy": 3.7499097108840944, + "rkd_teacher_similarity_mean": 0.782694810628891, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28810 + }, + { + "epoch": 8.385219668315392, + "flow_matching_loss": 0.013271929137408734, + "grad_norm": 0.11335690319538116, + "learning_rate": 1.2692805962789323e-07, + "loss": 0.0135, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006817546999081969, + "rkd_loss_weight": 0.039484649524092676, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3836976170539856, + "rkd_teacher_entropy": 3.7707191944122314, + "rkd_teacher_similarity_mean": 0.7915717542171479, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28820 + }, + { + "epoch": 8.388129182426535, + "flow_matching_loss": 0.012569156009703875, + "grad_norm": 0.08889774978160858, + "learning_rate": 1.2479063133184698e-07, + "loss": 0.0129, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007448453130200506, + "rkd_loss_weight": 0.039151304960250856, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37103723287582396, + "rkd_teacher_entropy": 3.7368372678756714, + "rkd_teacher_similarity_mean": 0.7770523071289063, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28830 + }, + { + "epoch": 8.391038696537677, + "flow_matching_loss": 0.011987391952425241, + "grad_norm": 0.1583925187587738, + "learning_rate": 1.2267127781782516e-07, + "loss": 0.0122, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006464928202331066, + "rkd_loss_weight": 0.038817960396409036, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38114712834358216, + "rkd_teacher_entropy": 3.762006735801697, + "rkd_teacher_similarity_mean": 0.7882611155509949, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28840 + }, + { + "epoch": 8.393948210648821, + "flow_matching_loss": 0.011577660543844103, + "grad_norm": 0.06700204312801361, + "learning_rate": 1.205700016610395e-07, + "loss": 0.0118, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006862412625923753, + "rkd_loss_weight": 0.038484616205096245, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3794972747564316, + "rkd_teacher_entropy": 3.7488112688064574, + "rkd_teacher_similarity_mean": 0.7863519668579102, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28850 + }, + { + "epoch": 8.396857724759965, + "flow_matching_loss": 0.012524808384478092, + "grad_norm": 0.07884331047534943, + "learning_rate": 1.1848680541474088e-07, + "loss": 0.0128, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007498934771865607, + "rkd_loss_weight": 0.038151272013783454, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38040803372859955, + "rkd_teacher_entropy": 3.759183073043823, + "rkd_teacher_similarity_mean": 0.7871242940425873, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28860 + }, + { + "epoch": 8.399767238871108, + "flow_matching_loss": 0.0159206117503345, + "grad_norm": 0.1380692720413208, + "learning_rate": 1.1642169161020955e-07, + "loss": 0.0162, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008218559995293617, + "rkd_loss_weight": 0.037817927077412605, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3830246150493622, + "rkd_teacher_entropy": 3.765042853355408, + "rkd_teacher_similarity_mean": 0.7903499901294708, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28870 + }, + { + "epoch": 8.402676752982252, + "flow_matching_loss": 0.012081322446465492, + "grad_norm": 0.09108998626470566, + "learning_rate": 1.1437466275675156e-07, + "loss": 0.0124, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0077435866929590706, + "rkd_loss_weight": 0.037484582886099814, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3805667906999588, + "rkd_teacher_entropy": 3.767560529708862, + "rkd_teacher_similarity_mean": 0.7867416739463806, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28880 + }, + { + "epoch": 8.405586267093396, + "flow_matching_loss": 0.010636518010869622, + "grad_norm": 0.0988449826836586, + "learning_rate": 1.1234572134170229e-07, + "loss": 0.0109, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007223815796896816, + "rkd_loss_weight": 0.037151238322257994, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37442317306995393, + "rkd_teacher_entropy": 3.7418654918670655, + "rkd_teacher_similarity_mean": 0.7807559967041016, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28890 + }, + { + "epoch": 8.40849578120454, + "flow_matching_loss": 0.01295430543832481, + "grad_norm": 0.09788165986537933, + "learning_rate": 1.1033486983041463e-07, + "loss": 0.0132, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007106482423841954, + "rkd_loss_weight": 0.0368178941309452, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36809725761413575, + "rkd_teacher_entropy": 3.7303255081176756, + "rkd_teacher_similarity_mean": 0.7747187435626983, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28900 + }, + { + "epoch": 8.411405295315681, + "flow_matching_loss": 0.014481448661535978, + "grad_norm": 0.10887172073125839, + "learning_rate": 1.0834211066626409e-07, + "loss": 0.0147, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006607078015804291, + "rkd_loss_weight": 0.036484549194574355, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.377535417675972, + "rkd_teacher_entropy": 3.740679717063904, + "rkd_teacher_similarity_mean": 0.7849459409713745, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28910 + }, + { + "epoch": 8.414314809426825, + "flow_matching_loss": 0.01084895241074264, + "grad_norm": 0.07343753427267075, + "learning_rate": 1.0636744627064044e-07, + "loss": 0.0111, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007357808970846235, + "rkd_loss_weight": 0.0361512053757906, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3776579976081848, + "rkd_teacher_entropy": 3.7438712120056152, + "rkd_teacher_similarity_mean": 0.7839939892292023, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28920 + }, + { + "epoch": 8.417224323537969, + "flow_matching_loss": 0.013059089053422212, + "grad_norm": 0.14846891164779663, + "learning_rate": 1.0441087904294766e-07, + "loss": 0.0134, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008575310092419386, + "rkd_loss_weight": 0.035817860439419744, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37553825676441194, + "rkd_teacher_entropy": 3.7363736629486084, + "rkd_teacher_similarity_mean": 0.7817663192749024, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28930 + }, + { + "epoch": 8.420133837649113, + "flow_matching_loss": 0.011372422007843852, + "grad_norm": 0.09558317810297012, + "learning_rate": 1.0247241136059904e-07, + "loss": 0.0116, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006621258519589901, + "rkd_loss_weight": 0.03548451624810696, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37765021324157716, + "rkd_teacher_entropy": 3.7475512504577635, + "rkd_teacher_similarity_mean": 0.7831174910068512, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28940 + }, + { + "epoch": 8.423043351760256, + "flow_matching_loss": 0.014741379721090197, + "grad_norm": 0.06860436499118805, + "learning_rate": 1.0055204557901709e-07, + "loss": 0.015, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008030550088733435, + "rkd_loss_weight": 0.035151171311736104, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37205937802791594, + "rkd_teacher_entropy": 3.7376014232635497, + "rkd_teacher_similarity_mean": 0.7779418766498566, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28950 + }, + { + "epoch": 8.4259528658714, + "flow_matching_loss": 0.013642414566129446, + "grad_norm": 0.08791674673557281, + "learning_rate": 9.864978403162361e-08, + "loss": 0.0139, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007162611931562424, + "rkd_loss_weight": 0.03481782749295235, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36773775815963744, + "rkd_teacher_entropy": 3.7239280939102173, + "rkd_teacher_similarity_mean": 0.7739505887031555, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28960 + }, + { + "epoch": 8.428862379982544, + "flow_matching_loss": 0.011839107330888509, + "grad_norm": 0.08354280889034271, + "learning_rate": 9.676562902984965e-08, + "loss": 0.0121, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007635414646938443, + "rkd_loss_weight": 0.03448448292911053, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37500707507133485, + "rkd_teacher_entropy": 3.746589016914368, + "rkd_teacher_similarity_mean": 0.7808877170085907, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28970 + }, + { + "epoch": 8.431771894093686, + "flow_matching_loss": 0.012439923733472824, + "grad_norm": 0.20441757142543793, + "learning_rate": 9.489958286311883e-08, + "loss": 0.0127, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006801312230527401, + "rkd_loss_weight": 0.03415113836526871, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37479018568992617, + "rkd_teacher_entropy": 3.739380669593811, + "rkd_teacher_similarity_mean": 0.7815635025501251, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28980 + }, + { + "epoch": 8.43468140820483, + "flow_matching_loss": 0.011240958468988538, + "grad_norm": 0.09474562108516693, + "learning_rate": 9.30516477988541e-08, + "loss": 0.0116, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009435731731355191, + "rkd_loss_weight": 0.03381779380142689, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3765105754137039, + "rkd_teacher_entropy": 3.743551778793335, + "rkd_teacher_similarity_mean": 0.7833987236022949, + "rkd_vlm_temp": 0.10000000149011612, + "step": 28990 + }, + { + "epoch": 8.437590922315973, + "flow_matching_loss": 0.011154254851862788, + "grad_norm": 0.08066828548908234, + "learning_rate": 9.122182608247098e-08, + "loss": 0.0114, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006657700473442674, + "rkd_loss_weight": 0.0334844496101141, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37983545660972595, + "rkd_teacher_entropy": 3.758946967124939, + "rkd_teacher_similarity_mean": 0.7862072169780732, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29000 + }, + { + "epoch": 8.440500436427117, + "flow_matching_loss": 0.011990392208099365, + "grad_norm": 0.10021498799324036, + "learning_rate": 8.94101199373759e-08, + "loss": 0.0123, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008494700910523534, + "rkd_loss_weight": 0.03315110504627228, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3714089274406433, + "rkd_teacher_entropy": 3.7314857006073, + "rkd_teacher_similarity_mean": 0.7776061177253724, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29010 + }, + { + "epoch": 8.44340995053826, + "flow_matching_loss": 0.01255634855479002, + "grad_norm": 0.1018059179186821, + "learning_rate": 8.761653156495964e-08, + "loss": 0.0128, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008621286973357201, + "rkd_loss_weight": 0.03281776048243046, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3744639277458191, + "rkd_teacher_entropy": 3.7424638271331787, + "rkd_teacher_similarity_mean": 0.781042605638504, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29020 + }, + { + "epoch": 8.446319464649404, + "flow_matching_loss": 0.012139985337853432, + "grad_norm": 0.10978969186544418, + "learning_rate": 8.584106314460549e-08, + "loss": 0.0124, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009086591564118862, + "rkd_loss_weight": 0.03248441629111767, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3720009088516235, + "rkd_teacher_entropy": 3.7295296669006346, + "rkd_teacher_similarity_mean": 0.7774313926696778, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29030 + }, + { + "epoch": 8.449228978760548, + "flow_matching_loss": 0.01229683361016214, + "grad_norm": 0.10857610404491425, + "learning_rate": 8.408371683367111e-08, + "loss": 0.0125, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0073825536295771595, + "rkd_loss_weight": 0.03215107172727585, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37627798318862915, + "rkd_teacher_entropy": 3.7515492916107176, + "rkd_teacher_similarity_mean": 0.7832549810409546, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29040 + }, + { + "epoch": 8.45213849287169, + "flow_matching_loss": 0.0128620860632509, + "grad_norm": 0.07115709036588669, + "learning_rate": 8.234449476750172e-08, + "loss": 0.0131, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006823187414556742, + "rkd_loss_weight": 0.03181772716343403, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3790363252162933, + "rkd_teacher_entropy": 3.7685832500457765, + "rkd_teacher_similarity_mean": 0.7858721196651459, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29050 + }, + { + "epoch": 8.455048006982834, + "flow_matching_loss": 0.013169905636459589, + "grad_norm": 0.08547176420688629, + "learning_rate": 8.062339905941351e-08, + "loss": 0.0134, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007112169358879328, + "rkd_loss_weight": 0.03148438259959221, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3791402906179428, + "rkd_teacher_entropy": 3.7521740198135376, + "rkd_teacher_similarity_mean": 0.7857940673828125, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29060 + }, + { + "epoch": 8.457957521093977, + "flow_matching_loss": 0.012964517902582883, + "grad_norm": 0.12184694409370422, + "learning_rate": 7.89204318007003e-08, + "loss": 0.0132, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00768358837813139, + "rkd_loss_weight": 0.031151038594543934, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3683477133512497, + "rkd_teacher_entropy": 3.732417941093445, + "rkd_teacher_similarity_mean": 0.7750196754932404, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29070 + }, + { + "epoch": 8.460867035205121, + "flow_matching_loss": 0.011715444317087532, + "grad_norm": 0.10965193063020706, + "learning_rate": 7.723559506062849e-08, + "loss": 0.012, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007964389165863395, + "rkd_loss_weight": 0.0308176938444376, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37261787354946135, + "rkd_teacher_entropy": 3.7485490083694457, + "rkd_teacher_similarity_mean": 0.778808331489563, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29080 + }, + { + "epoch": 8.463776549316265, + "flow_matching_loss": 0.010667387349531054, + "grad_norm": 0.08980146795511246, + "learning_rate": 7.556889088643381e-08, + "loss": 0.0109, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008776743477210403, + "rkd_loss_weight": 0.030484349466860294, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37236449122428894, + "rkd_teacher_entropy": 3.7404786109924317, + "rkd_teacher_similarity_mean": 0.779188472032547, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29090 + }, + { + "epoch": 8.466686063427408, + "flow_matching_loss": 0.012268220586702228, + "grad_norm": 0.07053288072347641, + "learning_rate": 7.392032130331793e-08, + "loss": 0.0125, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0070623218081891535, + "rkd_loss_weight": 0.03015100508928299, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3789093345403671, + "rkd_teacher_entropy": 3.760386276245117, + "rkd_teacher_similarity_mean": 0.7854063749313355, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29100 + }, + { + "epoch": 8.46959557753855, + "flow_matching_loss": 0.01268162433989346, + "grad_norm": 0.12004179507493973, + "learning_rate": 7.228988831445015e-08, + "loss": 0.0129, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007024823827669025, + "rkd_loss_weight": 0.02981766052544117, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3737578630447388, + "rkd_teacher_entropy": 3.7429000616073607, + "rkd_teacher_similarity_mean": 0.7805483043193817, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29110 + }, + { + "epoch": 8.472505091649694, + "flow_matching_loss": 0.014051408600062132, + "grad_norm": 0.1092677116394043, + "learning_rate": 7.06775939009574e-08, + "loss": 0.0143, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007650304026901722, + "rkd_loss_weight": 0.029484316147863864, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37703927755355837, + "rkd_teacher_entropy": 3.746528077125549, + "rkd_teacher_similarity_mean": 0.7833948612213135, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29120 + }, + { + "epoch": 8.475414605760838, + "flow_matching_loss": 0.0130795709323138, + "grad_norm": 0.08387481421232224, + "learning_rate": 6.908344002193256e-08, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00718750711530447, + "rkd_loss_weight": 0.029150971584022044, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.385457706451416, + "rkd_teacher_entropy": 3.7904392957687376, + "rkd_teacher_similarity_mean": 0.7922213196754455, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29130 + }, + { + "epoch": 8.478324119871981, + "flow_matching_loss": 0.01272636572830379, + "grad_norm": 0.06990223377943039, + "learning_rate": 6.750742861442116e-08, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007995266187936067, + "rkd_loss_weight": 0.028817627392709256, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38246020674705505, + "rkd_teacher_entropy": 3.7656827688217165, + "rkd_teacher_similarity_mean": 0.7894096553325654, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29140 + }, + { + "epoch": 8.481233633983125, + "flow_matching_loss": 0.01289745126850903, + "grad_norm": 0.0805157721042633, + "learning_rate": 6.594956159342636e-08, + "loss": 0.0131, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008566609537228941, + "rkd_loss_weight": 0.028484282828867436, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38305422067642214, + "rkd_teacher_entropy": 3.7588401556015016, + "rkd_teacher_similarity_mean": 0.7882326006889343, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29150 + }, + { + "epoch": 8.484143148094269, + "flow_matching_loss": 0.01191853084601462, + "grad_norm": 0.0976552739739418, + "learning_rate": 6.440984085190394e-08, + "loss": 0.0121, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006460720393806696, + "rkd_loss_weight": 0.02815093845129013, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3755894273519516, + "rkd_teacher_entropy": 3.74271183013916, + "rkd_teacher_similarity_mean": 0.7824760198593139, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29160 + }, + { + "epoch": 8.487052662205413, + "flow_matching_loss": 0.012031323183327914, + "grad_norm": 0.0973566398024559, + "learning_rate": 6.288826826075733e-08, + "loss": 0.0122, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007169649424031377, + "rkd_loss_weight": 0.02781759388744831, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3823927491903305, + "rkd_teacher_entropy": 3.7681154727935793, + "rkd_teacher_similarity_mean": 0.7894194543361663, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29170 + }, + { + "epoch": 8.489962176316554, + "flow_matching_loss": 0.013774420274421573, + "grad_norm": 0.07174689322710037, + "learning_rate": 6.138484566884261e-08, + "loss": 0.014, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007770699169486761, + "rkd_loss_weight": 0.027484249509871005, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37600449919700624, + "rkd_teacher_entropy": 3.7502832412719727, + "rkd_teacher_similarity_mean": 0.7818129479885101, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29180 + }, + { + "epoch": 8.492871690427698, + "flow_matching_loss": 0.011412661662325263, + "grad_norm": 0.11227567493915558, + "learning_rate": 5.989957490296183e-08, + "loss": 0.0116, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007575931213796139, + "rkd_loss_weight": 0.027150904946029186, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3649615854024887, + "rkd_teacher_entropy": 3.71903510093689, + "rkd_teacher_similarity_mean": 0.7714911103248596, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29190 + }, + { + "epoch": 8.495781204538842, + "flow_matching_loss": 0.010234741074964405, + "grad_norm": 0.0824415311217308, + "learning_rate": 5.843245776785466e-08, + "loss": 0.0104, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007645393861457706, + "rkd_loss_weight": 0.02681756056845188, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38044010698795316, + "rkd_teacher_entropy": 3.755080795288086, + "rkd_teacher_similarity_mean": 0.7866172552108764, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29200 + }, + { + "epoch": 8.498690718649986, + "flow_matching_loss": 0.011666385550051927, + "grad_norm": 0.14012175798416138, + "learning_rate": 5.6983496046210115e-08, + "loss": 0.0119, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008724505919963122, + "rkd_loss_weight": 0.026484216190874578, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3773469477891922, + "rkd_teacher_entropy": 3.744269847869873, + "rkd_teacher_similarity_mean": 0.7834981918334961, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29210 + }, + { + "epoch": 8.50160023276113, + "flow_matching_loss": 0.011973490007221699, + "grad_norm": 0.0923711359500885, + "learning_rate": 5.5552691498653165e-08, + "loss": 0.0122, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006807443732395768, + "rkd_loss_weight": 0.026150871627032758, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37108132243156433, + "rkd_teacher_entropy": 3.726953935623169, + "rkd_teacher_similarity_mean": 0.7774857401847839, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29220 + }, + { + "epoch": 8.504509746872273, + "flow_matching_loss": 0.012121305149048568, + "grad_norm": 0.09886476397514343, + "learning_rate": 5.4140045863743106e-08, + "loss": 0.0123, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006626608176156878, + "rkd_loss_weight": 0.025817527249455453, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3767861396074295, + "rkd_teacher_entropy": 3.7613128900527952, + "rkd_teacher_similarity_mean": 0.7834313213825226, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29230 + }, + { + "epoch": 8.507419260983415, + "flow_matching_loss": 0.012924473453313112, + "grad_norm": 0.0678325742483139, + "learning_rate": 5.274556085798188e-08, + "loss": 0.0131, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0067622634582221505, + "rkd_loss_weight": 0.025484182685613633, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3760396718978882, + "rkd_teacher_entropy": 3.7488744258880615, + "rkd_teacher_similarity_mean": 0.7834916651248932, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29240 + }, + { + "epoch": 8.510328775094559, + "flow_matching_loss": 0.010863593593239785, + "grad_norm": 0.1067485511302948, + "learning_rate": 5.136923817579742e-08, + "loss": 0.011, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007195999706164002, + "rkd_loss_weight": 0.02515083849430084, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3746611922979355, + "rkd_teacher_entropy": 3.7390209197998048, + "rkd_teacher_similarity_mean": 0.7810252606868744, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29250 + }, + { + "epoch": 8.513238289205702, + "flow_matching_loss": 0.01295489757321775, + "grad_norm": 0.06920285522937775, + "learning_rate": 5.001107948955197e-08, + "loss": 0.0131, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007028642576187849, + "rkd_loss_weight": 0.024817493930459022, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37485035359859464, + "rkd_teacher_entropy": 3.7412801027297973, + "rkd_teacher_similarity_mean": 0.7811437964439392, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29260 + }, + { + "epoch": 8.516147803316846, + "flow_matching_loss": 0.012443613819777965, + "grad_norm": 0.12139276415109634, + "learning_rate": 4.867108644953711e-08, + "loss": 0.0126, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006529054790735245, + "rkd_loss_weight": 0.024484149552881716, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3742437273263931, + "rkd_teacher_entropy": 3.748113989830017, + "rkd_teacher_similarity_mean": 0.7805661976337432, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29270 + }, + { + "epoch": 8.51905731742799, + "flow_matching_loss": 0.012427015695720911, + "grad_norm": 0.09857305139303207, + "learning_rate": 4.734926068397038e-08, + "loss": 0.0126, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007855452643707395, + "rkd_loss_weight": 0.024150804989039897, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37562273144721986, + "rkd_teacher_entropy": 3.745976424217224, + "rkd_teacher_similarity_mean": 0.7827023684978485, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29280 + }, + { + "epoch": 8.521966831539133, + "flow_matching_loss": 0.01294707292690873, + "grad_norm": 0.11276695132255554, + "learning_rate": 4.604560379899536e-08, + "loss": 0.0131, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007504062075167894, + "rkd_loss_weight": 0.023817460611462594, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3715743362903595, + "rkd_teacher_entropy": 3.7360175132751463, + "rkd_teacher_similarity_mean": 0.7779066085815429, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29290 + }, + { + "epoch": 8.524876345650277, + "flow_matching_loss": 0.011548599787056446, + "grad_norm": 0.102206751704216, + "learning_rate": 4.476011737867658e-08, + "loss": 0.0117, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006613968312740326, + "rkd_loss_weight": 0.02348411623388529, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.374597892165184, + "rkd_teacher_entropy": 3.743112087249756, + "rkd_teacher_similarity_mean": 0.7814869523048401, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29300 + }, + { + "epoch": 8.52778585976142, + "flow_matching_loss": 0.011482014786452055, + "grad_norm": 0.07578218728303909, + "learning_rate": 4.349280298500291e-08, + "loss": 0.0117, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007877640891820192, + "rkd_loss_weight": 0.02315077167004347, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3729608356952667, + "rkd_teacher_entropy": 3.7452751636505126, + "rkd_teacher_similarity_mean": 0.7788821339607239, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29310 + }, + { + "epoch": 8.530695373872563, + "flow_matching_loss": 0.014203146193176508, + "grad_norm": 0.12072637677192688, + "learning_rate": 4.224366215787923e-08, + "loss": 0.0144, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007738466979935766, + "rkd_loss_weight": 0.022817427292466164, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3768103688955307, + "rkd_teacher_entropy": 3.7516277551651003, + "rkd_teacher_similarity_mean": 0.7827991783618927, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29320 + }, + { + "epoch": 8.533604887983707, + "flow_matching_loss": 0.010823405161499978, + "grad_norm": 0.07358384132385254, + "learning_rate": 4.101269641513139e-08, + "loss": 0.011, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007188036059960723, + "rkd_loss_weight": 0.022484082728624344, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3799049824476242, + "rkd_teacher_entropy": 3.7583439111709596, + "rkd_teacher_similarity_mean": 0.7868063986301422, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29330 + }, + { + "epoch": 8.53651440209485, + "flow_matching_loss": 0.012374258320778608, + "grad_norm": 0.0951160192489624, + "learning_rate": 3.979990725249627e-08, + "loss": 0.0125, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007159401336684823, + "rkd_loss_weight": 0.02215073835104704, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3792489230632782, + "rkd_teacher_entropy": 3.7602333784103394, + "rkd_teacher_similarity_mean": 0.7861295700073242, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29340 + }, + { + "epoch": 8.539423916205994, + "flow_matching_loss": 0.013800981175154447, + "grad_norm": 0.1047191470861435, + "learning_rate": 3.860529614363006e-08, + "loss": 0.0139, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006409036228433251, + "rkd_loss_weight": 0.02181739378720522, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3697450190782547, + "rkd_teacher_entropy": 3.7264370679855348, + "rkd_teacher_similarity_mean": 0.776678878068924, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29350 + }, + { + "epoch": 8.542333430317138, + "flow_matching_loss": 0.012329838844016195, + "grad_norm": 0.08367092907428741, + "learning_rate": 3.742886454009664e-08, + "loss": 0.0125, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006979258917272091, + "rkd_loss_weight": 0.021484049409627913, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3753657191991806, + "rkd_teacher_entropy": 3.756095600128174, + "rkd_teacher_similarity_mean": 0.7821346998214722, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29360 + }, + { + "epoch": 8.545242944428281, + "flow_matching_loss": 0.011886262241750956, + "grad_norm": 0.10800106078386307, + "learning_rate": 3.627061387137087e-08, + "loss": 0.012, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00676258415915072, + "rkd_loss_weight": 0.02115070503205061, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37444421350955964, + "rkd_teacher_entropy": 3.7477715492248533, + "rkd_teacher_similarity_mean": 0.7815203189849853, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29370 + }, + { + "epoch": 8.548152458539423, + "flow_matching_loss": 0.011349557107314468, + "grad_norm": 0.1144319549202919, + "learning_rate": 3.5130545544835305e-08, + "loss": 0.0115, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0066752336919307705, + "rkd_loss_weight": 0.020817360654473305, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37052170634269715, + "rkd_teacher_entropy": 3.737827920913696, + "rkd_teacher_similarity_mean": 0.7777197778224945, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29380 + }, + { + "epoch": 8.551061972650567, + "flow_matching_loss": 0.012362240161746741, + "grad_norm": 0.09205601364374161, + "learning_rate": 3.400866094578348e-08, + "loss": 0.0126, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009503833344206214, + "rkd_loss_weight": 0.020484016090631486, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3778069257736206, + "rkd_teacher_entropy": 3.745040774345398, + "rkd_teacher_similarity_mean": 0.7839364469051361, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29390 + }, + { + "epoch": 8.55397148676171, + "flow_matching_loss": 0.012236251402646304, + "grad_norm": 0.08608333766460419, + "learning_rate": 3.290496143740829e-08, + "loss": 0.0124, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007027673395350575, + "rkd_loss_weight": 0.02015067171305418, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37865504920482634, + "rkd_teacher_entropy": 3.754538369178772, + "rkd_teacher_similarity_mean": 0.7854077756404877, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29400 + }, + { + "epoch": 8.556881000872854, + "flow_matching_loss": 0.011299362359568476, + "grad_norm": 0.07822097837924957, + "learning_rate": 3.18194483608103e-08, + "loss": 0.0114, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006237228214740753, + "rkd_loss_weight": 0.019817327335476875, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38271642923355104, + "rkd_teacher_entropy": 3.7668341398239136, + "rkd_teacher_similarity_mean": 0.7898830652236939, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29410 + }, + { + "epoch": 8.559790514983998, + "flow_matching_loss": 0.012336275912821293, + "grad_norm": 0.11535027623176575, + "learning_rate": 3.075212303498942e-08, + "loss": 0.0125, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0072369127534329895, + "rkd_loss_weight": 0.019483982771635055, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3790559858083725, + "rkd_teacher_entropy": 3.759717583656311, + "rkd_teacher_similarity_mean": 0.7859177529811859, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29420 + }, + { + "epoch": 8.562700029095142, + "flow_matching_loss": 0.015760408714413643, + "grad_norm": 0.1227986067533493, + "learning_rate": 2.9702986756846574e-08, + "loss": 0.0159, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006911476980894804, + "rkd_loss_weight": 0.01915063839405775, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3805494517087936, + "rkd_teacher_entropy": 3.763691282272339, + "rkd_teacher_similarity_mean": 0.7879332423210144, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29430 + }, + { + "epoch": 8.565609543206284, + "flow_matching_loss": 0.008698153868317604, + "grad_norm": 0.05505513772368431, + "learning_rate": 2.867204080118202e-08, + "loss": 0.0088, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007393289590254426, + "rkd_loss_weight": 0.01881729383021593, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38391519486904147, + "rkd_teacher_entropy": 3.76878936290741, + "rkd_teacher_similarity_mean": 0.7905469715595246, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29440 + }, + { + "epoch": 8.568519057317427, + "flow_matching_loss": 0.010844323644414544, + "grad_norm": 0.13055424392223358, + "learning_rate": 2.7659286420693708e-08, + "loss": 0.011, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006612886302173137, + "rkd_loss_weight": 0.018483949452638627, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36979403495788576, + "rkd_teacher_entropy": 3.722042608261108, + "rkd_teacher_similarity_mean": 0.7761749684810638, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29450 + }, + { + "epoch": 8.571428571428571, + "flow_matching_loss": 0.014166299253702164, + "grad_norm": 0.1288248747587204, + "learning_rate": 2.6664724845968934e-08, + "loss": 0.0143, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007780517637729645, + "rkd_loss_weight": 0.018150605075061322, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37035674750804903, + "rkd_teacher_entropy": 3.7396707057952883, + "rkd_teacher_similarity_mean": 0.7770422041416168, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29460 + }, + { + "epoch": 8.574338085539715, + "flow_matching_loss": 0.011656587570905685, + "grad_norm": 0.06758463382720947, + "learning_rate": 2.5688357285499343e-08, + "loss": 0.0118, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007544752769172191, + "rkd_loss_weight": 0.017817260511219502, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36660427451133726, + "rkd_teacher_entropy": 3.7172695875167845, + "rkd_teacher_similarity_mean": 0.773418378829956, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29470 + }, + { + "epoch": 8.577247599650859, + "flow_matching_loss": 0.011809022631496191, + "grad_norm": 0.08170061558485031, + "learning_rate": 2.473018492565926e-08, + "loss": 0.012, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008544079959392548, + "rkd_loss_weight": 0.017483916133642197, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37169910669326783, + "rkd_teacher_entropy": 3.7366882085800173, + "rkd_teacher_similarity_mean": 0.7783778548240662, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29480 + }, + { + "epoch": 8.580157113762002, + "flow_matching_loss": 0.012699697259813547, + "grad_norm": 0.0898415595293045, + "learning_rate": 2.3790208930720702e-08, + "loss": 0.0128, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007264252193272114, + "rkd_loss_weight": 0.01715057175606489, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3720902442932129, + "rkd_teacher_entropy": 3.7299802541732787, + "rkd_teacher_similarity_mean": 0.7784135699272156, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29490 + }, + { + "epoch": 8.583066627873146, + "flow_matching_loss": 0.01302870325744152, + "grad_norm": 0.10757849365472794, + "learning_rate": 2.2868430442843368e-08, + "loss": 0.0132, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.009364664182066917, + "rkd_loss_weight": 0.01681722719222307, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3737709105014801, + "rkd_teacher_entropy": 3.7400810956954955, + "rkd_teacher_similarity_mean": 0.7795722126960755, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29500 + }, + { + "epoch": 8.585976141984288, + "flow_matching_loss": 0.011702162958681584, + "grad_norm": 0.08888490498065948, + "learning_rate": 2.196485058207298e-08, + "loss": 0.0118, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006617501890286803, + "rkd_loss_weight": 0.016483882814645766, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37167355716228484, + "rkd_teacher_entropy": 3.727788043022156, + "rkd_teacher_similarity_mean": 0.7785680413246154, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29510 + }, + { + "epoch": 8.588885656095432, + "flow_matching_loss": 0.012443404458463193, + "grad_norm": 0.09273087233304977, + "learning_rate": 2.1079470446346282e-08, + "loss": 0.0126, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008142957510426641, + "rkd_loss_weight": 0.016150538437068464, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36968905925750734, + "rkd_teacher_entropy": 3.7335739850997927, + "rkd_teacher_similarity_mean": 0.7765285968780518, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29520 + }, + { + "epoch": 8.591795170206575, + "flow_matching_loss": 0.011094392789527775, + "grad_norm": 0.07275591790676117, + "learning_rate": 2.0212291111482707e-08, + "loss": 0.0112, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006385285500437021, + "rkd_loss_weight": 0.015817193873226644, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38043591678142546, + "rkd_teacher_entropy": 3.765472912788391, + "rkd_teacher_similarity_mean": 0.7874364197254181, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29530 + }, + { + "epoch": 8.594704684317719, + "flow_matching_loss": 0.012451403494924306, + "grad_norm": 0.09342821687459946, + "learning_rate": 1.9363313631187707e-08, + "loss": 0.0126, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007672385359182954, + "rkd_loss_weight": 0.015483849495649338, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3735052138566971, + "rkd_teacher_entropy": 3.7371692657470703, + "rkd_teacher_similarity_mean": 0.7797183930873871, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29540 + }, + { + "epoch": 8.597614198428863, + "flow_matching_loss": 0.010056595085188746, + "grad_norm": 0.08712940663099289, + "learning_rate": 1.853253903704777e-08, + "loss": 0.0102, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007296178303658962, + "rkd_loss_weight": 0.015150505024939776, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.371564581990242, + "rkd_teacher_entropy": 3.7391933679580687, + "rkd_teacher_similarity_mean": 0.7784183144569397, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29550 + }, + { + "epoch": 8.600523712540006, + "flow_matching_loss": 0.01244295109063387, + "grad_norm": 0.07596316933631897, + "learning_rate": 1.7719968338533733e-08, + "loss": 0.0125, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006693375715985894, + "rkd_loss_weight": 0.014817160554230213, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37259431183338165, + "rkd_teacher_entropy": 3.7437497854232786, + "rkd_teacher_similarity_mean": 0.779662299156189, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29560 + }, + { + "epoch": 8.603433226651148, + "flow_matching_loss": 0.011036696005612611, + "grad_norm": 0.09829017519950867, + "learning_rate": 1.6925602522992467e-08, + "loss": 0.0111, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007047889428213239, + "rkd_loss_weight": 0.014483816176652908, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37492158710956575, + "rkd_teacher_entropy": 3.7417265653610228, + "rkd_teacher_similarity_mean": 0.7818075895309449, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29570 + }, + { + "epoch": 8.606342740762292, + "flow_matching_loss": 0.01430424265563488, + "grad_norm": 0.09747501462697983, + "learning_rate": 1.6149442555658533e-08, + "loss": 0.0144, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006789289088919759, + "rkd_loss_weight": 0.014150471705943347, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37941196858882903, + "rkd_teacher_entropy": 3.758961224555969, + "rkd_teacher_similarity_mean": 0.7856990993022919, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29580 + }, + { + "epoch": 8.609252254873436, + "flow_matching_loss": 0.011627163924276828, + "grad_norm": 0.06721224635839462, + "learning_rate": 1.5391489379635858e-08, + "loss": 0.0117, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007787434430792928, + "rkd_loss_weight": 0.013817127235233784, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3727602630853653, + "rkd_teacher_entropy": 3.736543869972229, + "rkd_teacher_similarity_mean": 0.7788591086864471, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29590 + }, + { + "epoch": 8.61216176898458, + "flow_matching_loss": 0.012167699262499809, + "grad_norm": 0.07855939120054245, + "learning_rate": 1.4651743915909399e-08, + "loss": 0.0123, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007194629916921258, + "rkd_loss_weight": 0.013483782764524221, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36869427263736726, + "rkd_teacher_entropy": 3.726617360115051, + "rkd_teacher_similarity_mean": 0.775288051366806, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29600 + }, + { + "epoch": 8.615071283095723, + "flow_matching_loss": 0.011239948589354754, + "grad_norm": 0.13411098718643188, + "learning_rate": 1.3930207063340139e-08, + "loss": 0.0114, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.010021083150058984, + "rkd_loss_weight": 0.013150438386946916, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3667733758687973, + "rkd_teacher_entropy": 3.7077182292938233, + "rkd_teacher_similarity_mean": 0.7726501524448395, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29610 + }, + { + "epoch": 8.617980797206867, + "flow_matching_loss": 0.012247869558632375, + "grad_norm": 0.08544180542230606, + "learning_rate": 1.322687969866343e-08, + "loss": 0.0124, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00914307269267738, + "rkd_loss_weight": 0.012817093823105096, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3723228842020035, + "rkd_teacher_entropy": 3.737977957725525, + "rkd_teacher_similarity_mean": 0.7786461055278778, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29620 + }, + { + "epoch": 8.62089031131801, + "flow_matching_loss": 0.012760330829769373, + "grad_norm": 0.11078662425279617, + "learning_rate": 1.2541762676488989e-08, + "loss": 0.0128, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006842782394960523, + "rkd_loss_weight": 0.012483749445527792, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3776023954153061, + "rkd_teacher_entropy": 3.751891851425171, + "rkd_teacher_similarity_mean": 0.7844398438930511, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29630 + }, + { + "epoch": 8.623799825429153, + "flow_matching_loss": 0.013278418313711881, + "grad_norm": 0.0920582041144371, + "learning_rate": 1.1874856829297564e-08, + "loss": 0.0134, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007033931324258447, + "rkd_loss_weight": 0.012150405067950487, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37613825798034667, + "rkd_teacher_entropy": 3.7414815425872803, + "rkd_teacher_similarity_mean": 0.7822713017463684, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29640 + }, + { + "epoch": 8.626709339540296, + "flow_matching_loss": 0.012060327315703034, + "grad_norm": 0.1330152302980423, + "learning_rate": 1.1226162967442611e-08, + "loss": 0.0121, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007173639861866832, + "rkd_loss_weight": 0.011817060597240924, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3820945739746094, + "rkd_teacher_entropy": 3.7761820793151855, + "rkd_teacher_similarity_mean": 0.7892866849899292, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29650 + }, + { + "epoch": 8.62961885365144, + "flow_matching_loss": 0.011371010821312666, + "grad_norm": 0.06938835233449936, + "learning_rate": 1.059568187914861e-08, + "loss": 0.0114, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006726047536358237, + "rkd_loss_weight": 0.011483716126531363, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3733928740024567, + "rkd_teacher_entropy": 3.7301162481307983, + "rkd_teacher_similarity_mean": 0.7802997529506683, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29660 + }, + { + "epoch": 8.632528367762584, + "flow_matching_loss": 0.010072336066514254, + "grad_norm": 0.09394606947898865, + "learning_rate": 9.983414330511087e-09, + "loss": 0.0102, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00724882329814136, + "rkd_loss_weight": 0.0111503716558218, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37719191908836364, + "rkd_teacher_entropy": 3.741689920425415, + "rkd_teacher_similarity_mean": 0.7847223460674286, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29670 + }, + { + "epoch": 8.635437881873727, + "flow_matching_loss": 0.012445029336959124, + "grad_norm": 0.1277114301919937, + "learning_rate": 9.389361065489932e-09, + "loss": 0.0125, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007102689333260059, + "rkd_loss_weight": 0.010817027185112238, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3774881988763809, + "rkd_teacher_entropy": 3.7575833559036256, + "rkd_teacher_similarity_mean": 0.7843544185161591, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29680 + }, + { + "epoch": 8.638347395984871, + "flow_matching_loss": 0.012777579110115766, + "grad_norm": 0.08059908449649811, + "learning_rate": 8.81352280591774e-09, + "loss": 0.0129, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007228064071387053, + "rkd_loss_weight": 0.010483682807534932, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3784407079219818, + "rkd_teacher_entropy": 3.7548810243606567, + "rkd_teacher_similarity_mean": 0.7852893173694611, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29690 + }, + { + "epoch": 8.641256910096015, + "flow_matching_loss": 0.01353395483456552, + "grad_norm": 0.1066020280122757, + "learning_rate": 8.255900251491478e-09, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007528751576319337, + "rkd_loss_weight": 0.010150338336825371, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37733919024467466, + "rkd_teacher_entropy": 3.7597418785095216, + "rkd_teacher_similarity_mean": 0.7843454360961915, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29700 + }, + { + "epoch": 8.644166424207157, + "flow_matching_loss": 0.012607080396264791, + "grad_norm": 0.10149990767240524, + "learning_rate": 7.716494079775816e-09, + "loss": 0.0127, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00842201355844736, + "rkd_loss_weight": 0.009816993866115809, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3755749762058258, + "rkd_teacher_entropy": 3.7457210063934325, + "rkd_teacher_similarity_mean": 0.7810161411762238, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29710 + }, + { + "epoch": 8.6470759383183, + "flow_matching_loss": 0.01122334967367351, + "grad_norm": 0.09172727912664413, + "learning_rate": 7.195304946199799e-09, + "loss": 0.0113, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006569216866046191, + "rkd_loss_weight": 0.009483649488538503, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3795273661613464, + "rkd_teacher_entropy": 3.7591407775878904, + "rkd_teacher_similarity_mean": 0.786422598361969, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29720 + }, + { + "epoch": 8.649985452429444, + "flow_matching_loss": 0.01186260748654604, + "grad_norm": 0.09842565655708313, + "learning_rate": 6.692333484056845e-09, + "loss": 0.0119, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007186351437121629, + "rkd_loss_weight": 0.00915030501782894, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3712995797395706, + "rkd_teacher_entropy": 3.73626229763031, + "rkd_teacher_similarity_mean": 0.7784512102603912, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29730 + }, + { + "epoch": 8.652894966540588, + "flow_matching_loss": 0.010634208424016833, + "grad_norm": 0.1022212877869606, + "learning_rate": 6.207580304506411e-09, + "loss": 0.0107, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006667405366897583, + "rkd_loss_weight": 0.00881696054711938, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38268300592899324, + "rkd_teacher_entropy": 3.7720162868499756, + "rkd_teacher_similarity_mean": 0.7897070825099946, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29740 + }, + { + "epoch": 8.655804480651732, + "flow_matching_loss": 0.013255337718874216, + "grad_norm": 0.1106652244925499, + "learning_rate": 5.741045996567329e-09, + "loss": 0.0133, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007919418718665838, + "rkd_loss_weight": 0.008483616169542074, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3833591341972351, + "rkd_teacher_entropy": 3.771095299720764, + "rkd_teacher_similarity_mean": 0.7905888140201569, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29750 + }, + { + "epoch": 8.658713994762875, + "flow_matching_loss": 0.012530013686046004, + "grad_norm": 0.0918504074215889, + "learning_rate": 5.2927311271244725e-09, + "loss": 0.0126, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006505814986303449, + "rkd_loss_weight": 0.008150271698832512, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38044586181640627, + "rkd_teacher_entropy": 3.754545521736145, + "rkd_teacher_similarity_mean": 0.7867181658744812, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29760 + }, + { + "epoch": 8.661623508874017, + "flow_matching_loss": 0.011853586044162511, + "grad_norm": 0.07364419847726822, + "learning_rate": 4.8626362409204265e-09, + "loss": 0.0119, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007297363318502903, + "rkd_loss_weight": 0.007816927228122949, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38293265998363496, + "rkd_teacher_entropy": 3.7756296157836915, + "rkd_teacher_similarity_mean": 0.7895228207111359, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29770 + }, + { + "epoch": 8.664533022985161, + "flow_matching_loss": 0.012898790510371327, + "grad_norm": 0.09598074108362198, + "learning_rate": 4.4507618605638126e-09, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007752323662862182, + "rkd_loss_weight": 0.007483582803979516, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37444346249103544, + "rkd_teacher_entropy": 3.745263671875, + "rkd_teacher_similarity_mean": 0.7813834190368653, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29780 + }, + { + "epoch": 8.667442537096305, + "flow_matching_loss": 0.011939821625128388, + "grad_norm": 0.08007226139307022, + "learning_rate": 4.057108486519301e-09, + "loss": 0.012, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006194282788783312, + "rkd_loss_weight": 0.0071502383798360825, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3836446344852448, + "rkd_teacher_entropy": 3.774274396896362, + "rkd_teacher_similarity_mean": 0.7912096619606018, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29790 + }, + { + "epoch": 8.670352051207448, + "flow_matching_loss": 0.012144080409780145, + "grad_norm": 0.09623890370130539, + "learning_rate": 3.6816765971126043e-09, + "loss": 0.0122, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007558552781119943, + "rkd_loss_weight": 0.006816893862560391, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3775080442428589, + "rkd_teacher_entropy": 3.748309922218323, + "rkd_teacher_similarity_mean": 0.7838084101676941, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29800 + }, + { + "epoch": 8.673261565318592, + "flow_matching_loss": 0.01305982070043683, + "grad_norm": 0.10508178174495697, + "learning_rate": 3.324466648528812e-09, + "loss": 0.0131, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007728176703676581, + "rkd_loss_weight": 0.006483549438416958, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38427873849868777, + "rkd_teacher_entropy": 3.7648199081420897, + "rkd_teacher_similarity_mean": 0.7917699635028839, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29810 + }, + { + "epoch": 8.676171079429736, + "flow_matching_loss": 0.011673098383471369, + "grad_norm": 0.11501164734363556, + "learning_rate": 2.9854790748123917e-09, + "loss": 0.0117, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006769280275329947, + "rkd_loss_weight": 0.006150205014273524, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37351225316524506, + "rkd_teacher_entropy": 3.731976938247681, + "rkd_teacher_similarity_mean": 0.7799599587917327, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29820 + }, + { + "epoch": 8.67908059354088, + "flow_matching_loss": 0.012123622559010982, + "grad_norm": 0.08274076879024506, + "learning_rate": 2.664714287865522e-09, + "loss": 0.0122, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00656345491297543, + "rkd_loss_weight": 0.005816860590130091, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3703494131565094, + "rkd_teacher_entropy": 3.7324267864227294, + "rkd_teacher_similarity_mean": 0.7775156915187835, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29830 + }, + { + "epoch": 8.681990107652021, + "flow_matching_loss": 0.011966384947299957, + "grad_norm": 0.0935392901301384, + "learning_rate": 2.362172677446428e-09, + "loss": 0.012, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006541478540748357, + "rkd_loss_weight": 0.0054835160728543995, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37222449481487274, + "rkd_teacher_entropy": 3.737829852104187, + "rkd_teacher_similarity_mean": 0.7783968389034271, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29840 + }, + { + "epoch": 8.684899621763165, + "flow_matching_loss": 0.01299084834754467, + "grad_norm": 0.08913692086935043, + "learning_rate": 2.0778546111727137e-09, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007979734102264047, + "rkd_loss_weight": 0.005150171695277095, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36353912353515627, + "rkd_teacher_entropy": 3.7044272661209106, + "rkd_teacher_similarity_mean": 0.7694265305995941, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29850 + }, + { + "epoch": 8.687809135874309, + "flow_matching_loss": 0.013530783262103796, + "grad_norm": 0.063436359167099, + "learning_rate": 1.811760434516363e-09, + "loss": 0.0136, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0065337237901985645, + "rkd_loss_weight": 0.004816827224567532, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3819133698940277, + "rkd_teacher_entropy": 3.767391538619995, + "rkd_teacher_similarity_mean": 0.7882768332958221, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29860 + }, + { + "epoch": 8.690718649985453, + "flow_matching_loss": 0.01196725657209754, + "grad_norm": 0.07928457111120224, + "learning_rate": 1.5638904708087377e-09, + "loss": 0.012, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007159266388043761, + "rkd_loss_weight": 0.004483482800424099, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37463535368442535, + "rkd_teacher_entropy": 3.728318524360657, + "rkd_teacher_similarity_mean": 0.7808918118476867, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29870 + }, + { + "epoch": 8.693628164096596, + "flow_matching_loss": 0.012454949831590057, + "grad_norm": 0.10780801624059677, + "learning_rate": 1.3342450212355805e-09, + "loss": 0.0125, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00803315315861255, + "rkd_loss_weight": 0.004150138329714536, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3840633541345596, + "rkd_teacher_entropy": 3.7619615316390993, + "rkd_teacher_similarity_mean": 0.7896003723144531, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29880 + }, + { + "epoch": 8.69653767820774, + "flow_matching_loss": 0.01300233043730259, + "grad_norm": 0.0835670754313469, + "learning_rate": 1.122824364835351e-09, + "loss": 0.013, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.0072227213997393845, + "rkd_loss_weight": 0.003816793905571103, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37244221568107605, + "rkd_teacher_entropy": 3.7304155349731447, + "rkd_teacher_similarity_mean": 0.7795318543910981, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29890 + }, + { + "epoch": 8.699447192318882, + "flow_matching_loss": 0.012088582338765263, + "grad_norm": 0.07044794410467148, + "learning_rate": 9.296287585042196e-10, + "loss": 0.0121, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007254937011748552, + "rkd_loss_weight": 0.003483449458144605, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3734091013669968, + "rkd_teacher_entropy": 3.7433118343353273, + "rkd_teacher_similarity_mean": 0.7802669942378998, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29900 + }, + { + "epoch": 8.702356706430026, + "flow_matching_loss": 0.011578117962926626, + "grad_norm": 0.07756044715642929, + "learning_rate": 7.546584369960697e-10, + "loss": 0.0116, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007003492210060358, + "rkd_loss_weight": 0.003150104987435043, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37970939576625823, + "rkd_teacher_entropy": 3.75718412399292, + "rkd_teacher_similarity_mean": 0.7867781758308411, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29910 + }, + { + "epoch": 8.70526622054117, + "flow_matching_loss": 0.01267648539505899, + "grad_norm": 0.09335065633058548, + "learning_rate": 5.97913612912504e-10, + "loss": 0.0127, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.00705662383697927, + "rkd_loss_weight": 0.0028167605632916093, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.38234373927116394, + "rkd_teacher_entropy": 3.766027808189392, + "rkd_teacher_similarity_mean": 0.7884738266468048, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29920 + }, + { + "epoch": 8.708175734652313, + "flow_matching_loss": 0.014336996199563145, + "grad_norm": 0.10541074723005295, + "learning_rate": 4.593944767161684e-10, + "loss": 0.0144, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007551595335826278, + "rkd_loss_weight": 0.0024834161158651114, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3855361342430115, + "rkd_teacher_entropy": 3.7840657472610473, + "rkd_teacher_similarity_mean": 0.7920531034469604, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29930 + }, + { + "epoch": 8.711085248763457, + "flow_matching_loss": 0.012305569415912033, + "grad_norm": 0.06760773807764053, + "learning_rate": 3.391011967207591e-10, + "loss": 0.0123, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007728941785171628, + "rkd_loss_weight": 0.0021500716684386134, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37676357924938203, + "rkd_teacher_entropy": 3.74802405834198, + "rkd_teacher_similarity_mean": 0.7837477147579193, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29940 + }, + { + "epoch": 8.7139947628746, + "flow_matching_loss": 0.013422913989052177, + "grad_norm": 0.10112393647432327, + "learning_rate": 2.370339190910231e-10, + "loss": 0.0134, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008292284980416297, + "rkd_loss_weight": 0.0018167272326536476, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37067805230617523, + "rkd_teacher_entropy": 3.7324981927871703, + "rkd_teacher_similarity_mean": 0.7777444064617157, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29950 + }, + { + "epoch": 8.716904276985744, + "flow_matching_loss": 0.010679526627063752, + "grad_norm": 0.11599507182836533, + "learning_rate": 1.531927678510847e-10, + "loss": 0.0107, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007128407154232264, + "rkd_loss_weight": 0.0014833827735856175, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3736895978450775, + "rkd_teacher_entropy": 3.733028268814087, + "rkd_teacher_similarity_mean": 0.7796029508113861, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29960 + }, + { + "epoch": 8.719813791096886, + "flow_matching_loss": 0.011792660038918257, + "grad_norm": 0.09906724840402603, + "learning_rate": 8.757784487611886e-11, + "loss": 0.0118, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.008093094686046243, + "rkd_loss_weight": 0.0011500383378006517, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.3726320922374725, + "rkd_teacher_entropy": 3.738711404800415, + "rkd_teacher_similarity_mean": 0.7793073415756225, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29970 + }, + { + "epoch": 8.72272330520803, + "flow_matching_loss": 0.011459966469556094, + "grad_norm": 0.08702240139245987, + "learning_rate": 4.018922989235119e-11, + "loss": 0.0115, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.007659811992198229, + "rkd_loss_weight": 0.0008166938845533877, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37010207772254944, + "rkd_teacher_entropy": 3.7227819204330443, + "rkd_teacher_similarity_mean": 0.7764164268970489, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29980 + }, + { + "epoch": 8.725632819319173, + "flow_matching_loss": 0.010886103380471468, + "grad_norm": 0.0896395593881607, + "learning_rate": 1.1026980482053972e-11, + "loss": 0.0109, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006465431628748774, + "rkd_loss_weight": 0.00048334944585803896, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.37704059183597566, + "rkd_teacher_entropy": 3.7451969385147095, + "rkd_teacher_similarity_mean": 0.7839693486690521, + "rkd_vlm_temp": 0.10000000149011612, + "step": 29990 + }, + { + "epoch": 8.728542333430317, + "flow_matching_loss": 0.010409412160515785, + "grad_norm": 0.09349575638771057, + "learning_rate": 9.113208188082922e-14, + "loss": 0.0104, + "rkd_action_temp": 0.10000000149011612, + "rkd_fm_loss_weight": 1.0, + "rkd_loss": 0.006920031271874905, + "rkd_loss_weight": 0.0001500049969763495, + "rkd_relation_mode_id": 0.0, + "rkd_student_similarity_mean": 0.36756075024604795, + "rkd_teacher_entropy": 3.7220465421676634, + "rkd_teacher_similarity_mean": 0.7749577164649963, + "rkd_vlm_temp": 0.10000000149011612, + "step": 30000 + }, + { + "epoch": 8.728542333430317, + "step": 30000, + "total_flos": 0.0, + "train_loss": 0.022862061719596385, + "train_runtime": 33407.8458, + "train_samples_per_second": 114.943, + "train_steps_per_second": 0.898 + } + ], + "logging_steps": 10, + "max_steps": 30000, + "num_input_tokens_seen": 0, + "num_train_epochs": 9, + "save_steps": 1, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": false, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 0.0, + "train_batch_size": 64, + "trial_name": null, + "trial_params": null +} diff --git a/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/resolved_config.yaml b/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/resolved_config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..812ed4d211b88f92fcac88357664576d472f408f --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/resolved_config.yaml @@ -0,0 +1,36 @@ +experiment_name: rkd_phase3_aux_linear_decay +policy_type: groot_rkd +base_model_path: ${HOME}/groot_robocasa/robocasa_v2/RKD_Processing_line_Only/rkd_temp_0.1/phase2 +output_root: ${HOME}/groot_robocasa/robocasa_v2/processing_line_only/loss_strategy_ablation/rkd +sweep: + loss_strategy: + - aux_linear_decay + rkd_temp: + - 0.1 +dataset: + dataset_soup: my_atomic26_human +training: + num_gpus: 2 + batch_size: 64 +model: null +phases: +- name: phase3_fm_rkd_linear_decay + max_steps: 30000 + save_steps: 0 + trainable: + tune_llm: false + tune_visual: false + tune_projector: true + tune_diffusion_model: true + losses: + rkd_enabled: true + rkd_fm_loss_weight: 1.0 + rkd_loss_weight_schedule: linear_decay + rkd_loss_weight_start: 1.0 + rkd_loss_weight_end: 0.0 + rkd_action_temp: 0.1 + rkd_vlm_temp: 0.1 + rkd_relation_mode: token_pair_mean +resolved_sweep: + loss_strategy: aux_linear_decay + rkd_temp: 0.1 diff --git a/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/rkd_phase3_aux_linear_decay.yaml b/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/rkd_phase3_aux_linear_decay.yaml new file mode 100644 index 0000000000000000000000000000000000000000..f0923a1d5b60e3704ab28c1e6ee49cacd77cdde7 --- /dev/null +++ b/processing_line_only/loss_strategy_ablation/rkd/loss_strategy_aux_linear_decay__rkd_temp_0.1/rkd_phase3_aux_linear_decay.yaml @@ -0,0 +1,33 @@ +experiment_name: rkd_phase3_aux_linear_decay +policy_type: groot_rkd +base_model_path: ${HOME}/groot_robocasa/robocasa_v2/RKD_Processing_line_Only/rkd_temp_${sweep.rkd_temp}/phase2 +output_root: ${HOME}/groot_robocasa/robocasa_v2/processing_line_only/loss_strategy_ablation/rkd +sweep: + loss_strategy: + - aux_linear_decay + rkd_temp: + - 0.1 +dataset: + dataset_soup: my_atomic26_human +training: + num_gpus: 2 + batch_size: 64 +model: null +phases: +- name: phase3_fm_rkd_linear_decay + max_steps: 30000 + save_steps: 0 + trainable: + tune_llm: false + tune_visual: false + tune_projector: true + tune_diffusion_model: true + losses: + rkd_enabled: true + rkd_fm_loss_weight: 1.0 + rkd_loss_weight_schedule: linear_decay + rkd_loss_weight_start: 1.0 + rkd_loss_weight_end: 0.0 + rkd_action_temp: ${sweep.rkd_temp} + rkd_vlm_temp: ${sweep.rkd_temp} + rkd_relation_mode: token_pair_mean