Upload checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_mse_only_ins/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_mse_only_ins

checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_mse_only_ins/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_mse_only_ins/wandb/offline-run-20260128_052914-vlm_gym_match_move_fix3_unit_one_img_lr2e_5_mse_only_ins-run0/files/output.log

@@ -184,13 +184,6 @@ Preparing Dataset vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce/vlm_gym_ma
   fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
   fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
 ce_avg: 0.0, mse_avg: 0.007629983127117157
-base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_mse_only_ins/eval_used_rows, step_tag is vlm_gym_match_move_fix3_unit_one_img_lr2e_5_mse_only_ins_step1000
-Preparing Dataset vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce/vlm_gym_match_move_fix3_unit_val
-[eval debug] first 3 batch fingerprints:
-  fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-  fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-  fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-ce_avg: 0.0, mse_avg: 0.007755194790661335
 wandb: Detected [huggingface_hub.inference] in use.
 wandb: Use W&B Weave for improved LLM call tracing. Install Weave with `pip install weave` then add `import weave` to the top of your script.
 wandb: For more information, check out the docs at: https://weave-docs.wandb.ai/
@@ -1235,20 +1228,6 @@ wandb: For more information, check out the docs at: https://weave-docs.wandb.ai/
 [2026-01-28 07:27:28] (step=0001031) Train Loss mse: 0.0077, Train Loss ce: 0.0000, Train Steps/Sec: 0.16,
 [2026-01-28 07:27:34] (step=0001032) Train Loss mse: 0.0074, Train Loss ce: 0.0000, Train Steps/Sec: 0.16,
 [2026-01-28 07:27:40] (step=0001033) Train Loss mse: 0.0070, Train Loss ce: 0.0000, Train Steps/Sec: 0.15,
-base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_mse_only_ins/eval_used_rows, step_tag is vlm_gym_match_move_fix3_unit_one_img_lr2e_5_mse_only_ins_step1500
-Preparing Dataset vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce/vlm_gym_match_move_fix3_unit_val
-[eval debug] first 3 batch fingerprints:
-  fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-  fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-  fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-ce_avg: 0.0, mse_avg: 0.00788091029971838
-base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_mse_only_ins/eval_used_rows, step_tag is vlm_gym_match_move_fix3_unit_one_img_lr2e_5_mse_only_ins_step2000
-Preparing Dataset vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce/vlm_gym_match_move_fix3_unit_val
-[eval debug] first 3 batch fingerprints:
-  fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-  fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-  fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-ce_avg: 0.0, mse_avg: 0.008042296394705772
 [2026-01-28 07:27:47] (step=0001034) Train Loss mse: 0.0078, Train Loss ce: 0.0000, Train Steps/Sec: 0.16,
 [2026-01-28 07:27:53] (step=0001035) Train Loss mse: 0.0087, Train Loss ce: 0.0000, Train Steps/Sec: 0.16,
 [2026-01-28 07:28:00] (step=0001036) Train Loss mse: 0.0074, Train Loss ce: 0.0000, Train Steps/Sec: 0.16,
@@ -1268,6 +1247,27 @@ ce_avg: 0.0, mse_avg: 0.008042296394705772
 [2026-01-28 07:29:29] (step=0001050) Train Loss mse: 0.0083, Train Loss ce: 0.0000, Train Steps/Sec: 0.15,
 [2026-01-28 07:29:36] (step=0001051) Train Loss mse: 0.0079, Train Loss ce: 0.0000, Train Steps/Sec: 0.15,
 [2026-01-28 07:29:42] (step=0001052) Train Loss mse: 0.0070, Train Loss ce: 0.0000, Train Steps/Sec: 0.16,
+base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_mse_only_ins/eval_used_rows, step_tag is vlm_gym_match_move_fix3_unit_one_img_lr2e_5_mse_only_ins_step1000
+Preparing Dataset vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce/vlm_gym_match_move_fix3_unit_val
+[eval debug] first 3 batch fingerprints:
+  fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+  fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+  fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+ce_avg: 0.0, mse_avg: 0.007755194790661335
+base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_mse_only_ins/eval_used_rows, step_tag is vlm_gym_match_move_fix3_unit_one_img_lr2e_5_mse_only_ins_step1500
+Preparing Dataset vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce/vlm_gym_match_move_fix3_unit_val
+[eval debug] first 3 batch fingerprints:
+  fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+  fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+  fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+ce_avg: 0.0, mse_avg: 0.00788091029971838
+base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_mse_only_ins/eval_used_rows, step_tag is vlm_gym_match_move_fix3_unit_one_img_lr2e_5_mse_only_ins_step2000
+Preparing Dataset vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce/vlm_gym_match_move_fix3_unit_val
+[eval debug] first 3 batch fingerprints:
+  fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+  fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+  fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+ce_avg: 0.0, mse_avg: 0.008042296394705772
 [2026-01-28 07:29:49] (step=0001053) Train Loss mse: 0.0081, Train Loss ce: 0.0000, Train Steps/Sec: 0.16,
 [2026-01-28 07:29:55] (step=0001054) Train Loss mse: 0.0078, Train Loss ce: 0.0000, Train Steps/Sec: 0.16,
 [2026-01-28 07:30:01] (step=0001055) Train Loss mse: 0.0073, Train Loss ce: 0.0000, Train Steps/Sec: 0.16,
@@ -2645,6 +2645,20 @@ ce_avg: 0.0, mse_avg: 0.008042296394705772
 [2026-01-28 09:58:23] (step=0002427) Train Loss mse: 0.0073, Train Loss ce: 0.0000, Train Steps/Sec: 0.16,
 [2026-01-28 09:58:30] (step=0002428) Train Loss mse: 0.0070, Train Loss ce: 0.0000, Train Steps/Sec: 0.15,
 [2026-01-28 09:58:36] (step=0002429) Train Loss mse: 0.0078, Train Loss ce: 0.0000, Train Steps/Sec: 0.16,
+base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_mse_only_ins/eval_used_rows, step_tag is vlm_gym_match_move_fix3_unit_one_img_lr2e_5_mse_only_ins_step2500
+Preparing Dataset vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce/vlm_gym_match_move_fix3_unit_val
+[eval debug] first 3 batch fingerprints:
+  fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+  fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+  fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+ce_avg: 0.0, mse_avg: 0.008206211030483246
+base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_mse_only_ins/eval_used_rows, step_tag is vlm_gym_match_move_fix3_unit_one_img_lr2e_5_mse_only_ins_step3000
+Preparing Dataset vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce/vlm_gym_match_move_fix3_unit_val
+[eval debug] first 3 batch fingerprints:
+  fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+  fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+  fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+ce_avg: 0.0, mse_avg: 0.008431533351540565
 [2026-01-28 09:58:42] (step=0002430) Train Loss mse: 0.0063, Train Loss ce: 0.0000, Train Steps/Sec: 0.16,
 [2026-01-28 09:58:49] (step=0002431) Train Loss mse: 0.0073, Train Loss ce: 0.0000, Train Steps/Sec: 0.15,
 [2026-01-28 09:58:56] (step=0002432) Train Loss mse: 0.0072, Train Loss ce: 0.0000, Train Steps/Sec: 0.14,
@@ -2681,27 +2695,6 @@ ce_avg: 0.0, mse_avg: 0.008042296394705772
 [2026-01-28 10:02:15] (step=0002463) Train Loss mse: 0.0075, Train Loss ce: 0.0000, Train Steps/Sec: 0.16,
 [2026-01-28 10:02:22] (step=0002464) Train Loss mse: 0.0084, Train Loss ce: 0.0000, Train Steps/Sec: 0.16,
 [2026-01-28 10:02:28] (step=0002465) Train Loss mse: 0.0072, Train Loss ce: 0.0000, Train Steps/Sec: 0.15,
-base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_mse_only_ins/eval_used_rows, step_tag is vlm_gym_match_move_fix3_unit_one_img_lr2e_5_mse_only_ins_step2500
-Preparing Dataset vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce/vlm_gym_match_move_fix3_unit_val
-[eval debug] first 3 batch fingerprints:
-  fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-  fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-  fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-ce_avg: 0.0, mse_avg: 0.008206211030483246
-base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_mse_only_ins/eval_used_rows, step_tag is vlm_gym_match_move_fix3_unit_one_img_lr2e_5_mse_only_ins_step3000
-Preparing Dataset vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce/vlm_gym_match_move_fix3_unit_val
-[eval debug] first 3 batch fingerprints:
-  fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-  fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-  fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-ce_avg: 0.0, mse_avg: 0.008431533351540565
-base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_mse_only_ins/eval_used_rows, step_tag is vlm_gym_match_move_fix3_unit_one_img_lr2e_5_mse_only_ins_step3500
-Preparing Dataset vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce/vlm_gym_match_move_fix3_unit_val
-[eval debug] first 3 batch fingerprints:
-  fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-  fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-  fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-ce_avg: 0.0, mse_avg: 0.0084471320733428
 [2026-01-28 10:02:35] (step=0002466) Train Loss mse: 0.0070, Train Loss ce: 0.0000, Train Steps/Sec: 0.15,
 [2026-01-28 10:02:42] (step=0002467) Train Loss mse: 0.0078, Train Loss ce: 0.0000, Train Steps/Sec: 0.15,
 [2026-01-28 10:02:48] (step=0002468) Train Loss mse: 0.0063, Train Loss ce: 0.0000, Train Steps/Sec: 0.16,
@@ -3615,6 +3608,27 @@ ce_avg: 0.0, mse_avg: 0.0084471320733428
 [2026-01-28 11:41:24] (step=0003376) Train Loss mse: 0.0065, Train Loss ce: 0.0000, Train Steps/Sec: 0.15,
 [2026-01-28 11:41:31] (step=0003377) Train Loss mse: 0.0067, Train Loss ce: 0.0000, Train Steps/Sec: 0.16,
 [2026-01-28 11:41:37] (step=0003378) Train Loss mse: 0.0068, Train Loss ce: 0.0000, Train Steps/Sec: 0.16,
+base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_mse_only_ins/eval_used_rows, step_tag is vlm_gym_match_move_fix3_unit_one_img_lr2e_5_mse_only_ins_step3500
+Preparing Dataset vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce/vlm_gym_match_move_fix3_unit_val
+[eval debug] first 3 batch fingerprints:
+  fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+  fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+  fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+ce_avg: 0.0, mse_avg: 0.0084471320733428
+base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_mse_only_ins/eval_used_rows, step_tag is vlm_gym_match_move_fix3_unit_one_img_lr2e_5_mse_only_ins_step4000
+Preparing Dataset vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce/vlm_gym_match_move_fix3_unit_val
+[eval debug] first 3 batch fingerprints:
+  fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+  fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+  fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+ce_avg: 0.0, mse_avg: 0.008800854906439781
+base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_mse_only_ins/eval_used_rows, step_tag is vlm_gym_match_move_fix3_unit_one_img_lr2e_5_mse_only_ins_step4500
+Preparing Dataset vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce/vlm_gym_match_move_fix3_unit_val
+[eval debug] first 3 batch fingerprints:
+  fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+  fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+  fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+ce_avg: 0.0, mse_avg: 0.009047330357134342
 [2026-01-28 11:41:44] (step=0003379) Train Loss mse: 0.0061, Train Loss ce: 0.0000, Train Steps/Sec: 0.15,
 [2026-01-28 11:41:50] (step=0003380) Train Loss mse: 0.0074, Train Loss ce: 0.0000, Train Steps/Sec: 0.16,
 [2026-01-28 11:41:57] (step=0003381) Train Loss mse: 0.0073, Train Loss ce: 0.0000, Train Steps/Sec: 0.15,
@@ -3669,20 +3683,6 @@ ce_avg: 0.0, mse_avg: 0.0084471320733428
 [2026-01-28 11:47:14] (step=0003430) Train Loss mse: 0.0073, Train Loss ce: 0.0000, Train Steps/Sec: 0.15,
 [2026-01-28 11:47:21] (step=0003431) Train Loss mse: 0.0062, Train Loss ce: 0.0000, Train Steps/Sec: 0.15,
 [2026-01-28 11:47:27] (step=0003432) Train Loss mse: 0.0065, Train Loss ce: 0.0000, Train Steps/Sec: 0.16,
-base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_mse_only_ins/eval_used_rows, step_tag is vlm_gym_match_move_fix3_unit_one_img_lr2e_5_mse_only_ins_step4000
-Preparing Dataset vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce/vlm_gym_match_move_fix3_unit_val
-[eval debug] first 3 batch fingerprints:
-  fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-  fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-  fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-ce_avg: 0.0, mse_avg: 0.008800854906439781
-base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_mse_only_ins/eval_used_rows, step_tag is vlm_gym_match_move_fix3_unit_one_img_lr2e_5_mse_only_ins_step4500
-Preparing Dataset vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce/vlm_gym_match_move_fix3_unit_val
-[eval debug] first 3 batch fingerprints:
-  fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-  fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-  fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-ce_avg: 0.0, mse_avg: 0.009047330357134342
 [2026-01-28 11:47:34] (step=0003433) Train Loss mse: 0.0069, Train Loss ce: 0.0000, Train Steps/Sec: 0.15,
 [2026-01-28 11:47:40] (step=0003434) Train Loss mse: 0.0067, Train Loss ce: 0.0000, Train Steps/Sec: 0.16,
 [2026-01-28 11:47:47] (step=0003435) Train Loss mse: 0.0068, Train Loss ce: 0.0000, Train Steps/Sec: 0.16,
@@ -5084,6 +5084,13 @@ ce_avg: 0.0, mse_avg: 0.009047330357134342
 [2026-01-28 14:19:22] (step=0004831) Train Loss mse: 0.0069, Train Loss ce: 0.0000, Train Steps/Sec: 0.14,
 [2026-01-28 14:19:28] (step=0004832) Train Loss mse: 0.0069, Train Loss ce: 0.0000, Train Steps/Sec: 0.16,
 [2026-01-28 14:19:35] (step=0004833) Train Loss mse: 0.0065, Train Loss ce: 0.0000, Train Steps/Sec: 0.15,
+base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_mse_only_ins/eval_used_rows, step_tag is vlm_gym_match_move_fix3_unit_one_img_lr2e_5_mse_only_ins_step5000
+Preparing Dataset vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce/vlm_gym_match_move_fix3_unit_val
+[eval debug] first 3 batch fingerprints:
+  fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+  fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+  fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
+ce_avg: 0.0, mse_avg: 0.008728216402232647
 [2026-01-28 14:19:41] (step=0004834) Train Loss mse: 0.0078, Train Loss ce: 0.0000, Train Steps/Sec: 0.15,
 [2026-01-28 14:19:48] (step=0004835) Train Loss mse: 0.0075, Train Loss ce: 0.0000, Train Steps/Sec: 0.15,
 [2026-01-28 14:19:54] (step=0004836) Train Loss mse: 0.0072, Train Loss ce: 0.0000, Train Steps/Sec: 0.15,
@@ -5192,13 +5199,6 @@ ce_avg: 0.0, mse_avg: 0.009047330357134342
 [2026-01-28 14:31:03] (step=0004939) Train Loss mse: 0.0076, Train Loss ce: 0.0000, Train Steps/Sec: 0.16,
 [2026-01-28 14:31:09] (step=0004940) Train Loss mse: 0.0101, Train Loss ce: 0.0000, Train Steps/Sec: 0.16,
 [2026-01-28 14:31:16] (step=0004941) Train Loss mse: 0.0061, Train Loss ce: 0.0000, Train Steps/Sec: 0.15,
-base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_mse_only_ins/eval_used_rows, step_tag is vlm_gym_match_move_fix3_unit_one_img_lr2e_5_mse_only_ins_step5000
-Preparing Dataset vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce/vlm_gym_match_move_fix3_unit_val
-[eval debug] first 3 batch fingerprints:
-  fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-  fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-  fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_mse_loss_only_evalonce'}]
-ce_avg: 0.0, mse_avg: 0.008728216402232647
 [2026-01-28 14:31:22] (step=0004942) Train Loss mse: 0.0067, Train Loss ce: 0.0000, Train Steps/Sec: 0.15,
 [2026-01-28 14:31:29] (step=0004943) Train Loss mse: 0.0067, Train Loss ce: 0.0000, Train Steps/Sec: 0.15,
 [2026-01-28 14:31:35] (step=0004944) Train Loss mse: 0.0067, Train Loss ce: 0.0000, Train Steps/Sec: 0.16,