Junyi42 commited on
Commit
89f213b
·
verified ·
1 Parent(s): 24ceec0

Upload checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins

Browse files
checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins/wandb/offline-run-20260127_015413-checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins-run0/files/output.log CHANGED
@@ -1008,6 +1008,53 @@ wandb: For more information, check out the docs at: https://weave-docs.wandb.ai/
1008
  [2026-01-27 03:48:48] (step=0000997) Train Loss mse: 0.0083, Train Loss ce: 0.1699, Train Steps/Sec: 0.15,
1009
  [2026-01-27 03:48:54] (step=0000998) Train Loss mse: 0.0078, Train Loss ce: 0.1487, Train Steps/Sec: 0.15,
1010
  [2026-01-27 03:49:01] (step=0000999) Train Loss mse: 0.0096, Train Loss ce: 0.1769, Train Steps/Sec: 0.15,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1011
  FullyShardedDataParallel(
1012
  (_fsdp_wrapped_module): Bagel(
1013
  (language_model): Qwen2ForCausalLM(
@@ -1194,13 +1241,6 @@ Preparing Dataset vlm_gym_match_move_fix3_unit_celoss_evalonce/vlm_gym_match_mov
1194
  fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
1195
  fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
1196
  ce_avg: 0.20799879729747772, mse_avg: 0.007969205267727375
1197
- base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins/eval_used_rows, step_tag is checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins_step1000
1198
- Preparing Dataset vlm_gym_match_move_fix3_unit_celoss_evalonce/vlm_gym_match_move_fix3_unit_val
1199
- [eval debug] first 3 batch fingerprints:
1200
- fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
1201
- fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
1202
- fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
1203
- ce_avg: 0.17712201178073883, mse_avg: 0.00836949609220028
1204
  base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins/eval_used_rows, step_tag is checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins_step1500
1205
  Preparing Dataset vlm_gym_match_move_fix3_unit_celoss_evalonce/vlm_gym_match_move_fix3_unit_val
1206
  [eval debug] first 3 batch fingerprints:
@@ -1215,53 +1255,6 @@ Preparing Dataset vlm_gym_match_move_fix3_unit_celoss_evalonce/vlm_gym_match_mov
1215
  fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
1216
  fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
1217
  ce_avg: 0.1749173104763031, mse_avg: 0.009318988770246506
1218
- [2026-01-27 03:49:39] (step=0001000) Train Loss mse: 0.0100, Train Loss ce: 0.1701, Train Steps/Sec: 0.03,
1219
- [2026-01-27 03:49:45] (step=0001001) Train Loss mse: 0.0078, Train Loss ce: 0.1978, Train Steps/Sec: 0.16,
1220
- [2026-01-27 03:49:51] (step=0001002) Train Loss mse: 0.0093, Train Loss ce: 0.1890, Train Steps/Sec: 0.16,
1221
- [2026-01-27 03:49:58] (step=0001003) Train Loss mse: 0.0089, Train Loss ce: 0.1751, Train Steps/Sec: 0.16,
1222
- [2026-01-27 03:50:04] (step=0001004) Train Loss mse: 0.0100, Train Loss ce: 0.1660, Train Steps/Sec: 0.16,
1223
- [2026-01-27 03:50:10] (step=0001005) Train Loss mse: 0.0092, Train Loss ce: 0.1881, Train Steps/Sec: 0.16,
1224
- [2026-01-27 03:50:17] (step=0001006) Train Loss mse: 0.0088, Train Loss ce: 0.1926, Train Steps/Sec: 0.15,
1225
- [2026-01-27 03:50:24] (step=0001007) Train Loss mse: 0.0083, Train Loss ce: 0.1839, Train Steps/Sec: 0.15,
1226
- [2026-01-27 03:50:30] (step=0001008) Train Loss mse: 0.0097, Train Loss ce: 0.2083, Train Steps/Sec: 0.16,
1227
- [2026-01-27 03:50:37] (step=0001009) Train Loss mse: 0.0083, Train Loss ce: 0.1949, Train Steps/Sec: 0.16,
1228
- [2026-01-27 03:50:43] (step=0001010) Train Loss mse: 0.0074, Train Loss ce: 0.1642, Train Steps/Sec: 0.16,
1229
- [2026-01-27 03:50:49] (step=0001011) Train Loss mse: 0.0095, Train Loss ce: 0.2152, Train Steps/Sec: 0.16,
1230
- [2026-01-27 03:50:56] (step=0001012) Train Loss mse: 0.0085, Train Loss ce: 0.1922, Train Steps/Sec: 0.16,
1231
- [2026-01-27 03:51:02] (step=0001013) Train Loss mse: 0.0084, Train Loss ce: 0.1974, Train Steps/Sec: 0.16,
1232
- [2026-01-27 03:51:09] (step=0001014) Train Loss mse: 0.0080, Train Loss ce: 0.1732, Train Steps/Sec: 0.15,
1233
- [2026-01-27 03:51:15] (step=0001015) Train Loss mse: 0.0084, Train Loss ce: 0.2082, Train Steps/Sec: 0.16,
1234
- [2026-01-27 03:51:22] (step=0001016) Train Loss mse: 0.0088, Train Loss ce: 0.1730, Train Steps/Sec: 0.15,
1235
- [2026-01-27 03:51:28] (step=0001017) Train Loss mse: 0.0093, Train Loss ce: 0.2090, Train Steps/Sec: 0.16,
1236
- [2026-01-27 03:51:34] (step=0001018) Train Loss mse: 0.0091, Train Loss ce: 0.2026, Train Steps/Sec: 0.16,
1237
- [2026-01-27 03:51:41] (step=0001019) Train Loss mse: 0.0095, Train Loss ce: 0.1736, Train Steps/Sec: 0.16,
1238
- [2026-01-27 03:51:47] (step=0001020) Train Loss mse: 0.0092, Train Loss ce: 0.2094, Train Steps/Sec: 0.16,
1239
- [2026-01-27 03:51:53] (step=0001021) Train Loss mse: 0.0091, Train Loss ce: 0.1846, Train Steps/Sec: 0.16,
1240
- [2026-01-27 03:52:00] (step=0001022) Train Loss mse: 0.0093, Train Loss ce: 0.1848, Train Steps/Sec: 0.16,
1241
- [2026-01-27 03:52:06] (step=0001023) Train Loss mse: 0.0078, Train Loss ce: 0.1752, Train Steps/Sec: 0.15,
1242
- [2026-01-27 03:52:13] (step=0001024) Train Loss mse: 0.0087, Train Loss ce: 0.1691, Train Steps/Sec: 0.14,
1243
- [2026-01-27 03:52:20] (step=0001025) Train Loss mse: 0.0084, Train Loss ce: 0.1881, Train Steps/Sec: 0.16,
1244
- [2026-01-27 03:52:26] (step=0001026) Train Loss mse: 0.0088, Train Loss ce: 0.2076, Train Steps/Sec: 0.16,
1245
- [2026-01-27 03:52:32] (step=0001027) Train Loss mse: 0.0108, Train Loss ce: 0.1895, Train Steps/Sec: 0.16,
1246
- [2026-01-27 03:52:39] (step=0001028) Train Loss mse: 0.0085, Train Loss ce: 0.1782, Train Steps/Sec: 0.16,
1247
- [2026-01-27 03:52:45] (step=0001029) Train Loss mse: 0.0085, Train Loss ce: 0.1808, Train Steps/Sec: 0.16,
1248
- [2026-01-27 03:52:51] (step=0001030) Train Loss mse: 0.0106, Train Loss ce: 0.1796, Train Steps/Sec: 0.18,
1249
- [2026-01-27 03:52:57] (step=0001031) Train Loss mse: 0.0087, Train Loss ce: 0.1868, Train Steps/Sec: 0.16,
1250
- [2026-01-27 03:53:04] (step=0001032) Train Loss mse: 0.0089, Train Loss ce: 0.2020, Train Steps/Sec: 0.15,
1251
- [2026-01-27 03:53:11] (step=0001033) Train Loss mse: 0.0086, Train Loss ce: 0.1978, Train Steps/Sec: 0.14,
1252
- [2026-01-27 03:53:17] (step=0001034) Train Loss mse: 0.0092, Train Loss ce: 0.1917, Train Steps/Sec: 0.15,
1253
- [2026-01-27 03:53:24] (step=0001035) Train Loss mse: 0.0099, Train Loss ce: 0.1958, Train Steps/Sec: 0.16,
1254
- [2026-01-27 03:53:30] (step=0001036) Train Loss mse: 0.0081, Train Loss ce: 0.1909, Train Steps/Sec: 0.16,
1255
- [2026-01-27 03:53:36] (step=0001037) Train Loss mse: 0.0079, Train Loss ce: 0.1868, Train Steps/Sec: 0.16,
1256
- [2026-01-27 03:53:43] (step=0001038) Train Loss mse: 0.0103, Train Loss ce: 0.1744, Train Steps/Sec: 0.16,
1257
- [2026-01-27 03:53:49] (step=0001039) Train Loss mse: 0.0090, Train Loss ce: 0.2039, Train Steps/Sec: 0.16,
1258
- [2026-01-27 03:53:55] (step=0001040) Train Loss mse: 0.0081, Train Loss ce: 0.1573, Train Steps/Sec: 0.16,
1259
- [2026-01-27 03:54:02] (step=0001041) Train Loss mse: 0.0091, Train Loss ce: 0.1810, Train Steps/Sec: 0.15,
1260
- [2026-01-27 03:54:09] (step=0001042) Train Loss mse: 0.0087, Train Loss ce: 0.2339, Train Steps/Sec: 0.15,
1261
- [2026-01-27 03:54:15] (step=0001043) Train Loss mse: 0.0082, Train Loss ce: 0.1785, Train Steps/Sec: 0.15,
1262
- [2026-01-27 03:54:22] (step=0001044) Train Loss mse: 0.0112, Train Loss ce: 0.2073, Train Steps/Sec: 0.16,
1263
- [2026-01-27 03:54:28] (step=0001045) Train Loss mse: 0.0096, Train Loss ce: 0.1835, Train Steps/Sec: 0.16,
1264
- [2026-01-27 03:54:34] (step=0001046) Train Loss mse: 0.0086, Train Loss ce: 0.1964, Train Steps/Sec: 0.16,
1265
  [2026-01-27 03:54:41] (step=0001047) Train Loss mse: 0.0085, Train Loss ce: 0.1884, Train Steps/Sec: 0.16,
1266
  [2026-01-27 03:54:47] (step=0001048) Train Loss mse: 0.0083, Train Loss ce: 0.2145, Train Steps/Sec: 0.16,
1267
  [2026-01-27 03:54:53] (step=0001049) Train Loss mse: 0.0093, Train Loss ce: 0.1888, Train Steps/Sec: 0.16,
@@ -2610,20 +2603,6 @@ ce_avg: 0.1749173104763031, mse_avg: 0.009318988770246506
2610
  [2026-01-27 06:21:07] (step=0002392) Train Loss mse: 0.0086, Train Loss ce: 0.1506, Train Steps/Sec: 0.16,
2611
  [2026-01-27 06:21:14] (step=0002393) Train Loss mse: 0.0084, Train Loss ce: 0.1657, Train Steps/Sec: 0.15,
2612
  [2026-01-27 06:21:20] (step=0002394) Train Loss mse: 0.0074, Train Loss ce: 0.1535, Train Steps/Sec: 0.15,
2613
- base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins/eval_used_rows, step_tag is checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins_step2500
2614
- Preparing Dataset vlm_gym_match_move_fix3_unit_celoss_evalonce/vlm_gym_match_move_fix3_unit_val
2615
- [eval debug] first 3 batch fingerprints:
2616
- fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
2617
- fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
2618
- fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
2619
- ce_avg: 0.16938790678977966, mse_avg: 0.009705973789095879
2620
- base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins/eval_used_rows, step_tag is checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins_step3000
2621
- Preparing Dataset vlm_gym_match_move_fix3_unit_celoss_evalonce/vlm_gym_match_move_fix3_unit_val
2622
- [eval debug] first 3 batch fingerprints:
2623
- fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
2624
- fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
2625
- fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
2626
- ce_avg: 0.1402648687362671, mse_avg: 0.007266872562468052
2627
  [2026-01-27 06:21:27] (step=0002395) Train Loss mse: 0.0083, Train Loss ce: 0.1900, Train Steps/Sec: 0.16,
2628
  [2026-01-27 06:21:34] (step=0002396) Train Loss mse: 0.0074, Train Loss ce: 0.1865, Train Steps/Sec: 0.15,
2629
  [2026-01-27 06:21:40] (step=0002397) Train Loss mse: 0.0096, Train Loss ce: 0.1719, Train Steps/Sec: 0.15,
@@ -2723,6 +2702,27 @@ ce_avg: 0.1402648687362671, mse_avg: 0.007266872562468052
2723
  [2026-01-27 06:31:50] (step=0002491) Train Loss mse: 0.0079, Train Loss ce: 0.1600, Train Steps/Sec: 0.16,
2724
  [2026-01-27 06:31:57] (step=0002492) Train Loss mse: 0.0075, Train Loss ce: 0.1430, Train Steps/Sec: 0.15,
2725
  [2026-01-27 06:32:03] (step=0002493) Train Loss mse: 0.0081, Train Loss ce: 0.1674, Train Steps/Sec: 0.15,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2726
  [2026-01-27 06:32:10] (step=0002494) Train Loss mse: 0.0080, Train Loss ce: 0.1591, Train Steps/Sec: 0.15,
2727
  [2026-01-27 06:32:17] (step=0002495) Train Loss mse: 0.0089, Train Loss ce: 0.1406, Train Steps/Sec: 0.15,
2728
  [2026-01-27 06:32:23] (step=0002496) Train Loss mse: 0.0077, Train Loss ce: 0.1890, Train Steps/Sec: 0.15,
@@ -3751,6 +3751,20 @@ ce_avg: 0.1402648687362671, mse_avg: 0.007266872562468052
3751
  [2026-01-27 08:26:36] (step=0003516) Train Loss mse: 0.0078, Train Loss ce: 0.1360, Train Steps/Sec: 0.16,
3752
  [2026-01-27 08:26:43] (step=0003517) Train Loss mse: 0.0089, Train Loss ce: 0.1218, Train Steps/Sec: 0.16,
3753
  [2026-01-27 08:26:49] (step=0003518) Train Loss mse: 0.0083, Train Loss ce: 0.1212, Train Steps/Sec: 0.16,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3754
  [2026-01-27 08:26:56] (step=0003519) Train Loss mse: 0.0099, Train Loss ce: 0.1251, Train Steps/Sec: 0.16,
3755
  [2026-01-27 08:27:02] (step=0003520) Train Loss mse: 0.0078, Train Loss ce: 0.1416, Train Steps/Sec: 0.16,
3756
  [2026-01-27 08:27:08] (step=0003521) Train Loss mse: 0.0084, Train Loss ce: 0.1459, Train Steps/Sec: 0.16,
@@ -3807,27 +3821,6 @@ ce_avg: 0.1402648687362671, mse_avg: 0.007266872562468052
3807
  [2026-01-27 08:32:34] (step=0003572) Train Loss mse: 0.0101, Train Loss ce: 0.1471, Train Steps/Sec: 0.18,
3808
  [2026-01-27 08:32:40] (step=0003573) Train Loss mse: 0.0079, Train Loss ce: 0.1128, Train Steps/Sec: 0.16,
3809
  [2026-01-27 08:32:46] (step=0003574) Train Loss mse: 0.0077, Train Loss ce: 0.1507, Train Steps/Sec: 0.16,
3810
- [
3811
- base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins/eval_used_rows, step_tag is checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins_step3500
3812
- Preparing Dataset vlm_gym_match_move_fix3_unit_celoss_evalonce/vlm_gym_match_move_fix3_unit_val
3813
- [eval debug] first 3 batch fingerprints:
3814
- fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
3815
- fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
3816
- fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
3817
- ce_avg: 0.13325323164463043, mse_avg: 0.007202878128737211
3818
- base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins/eval_used_rows, step_tag is checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins_step4000
3819
- Preparing Dataset vlm_gym_match_move_fix3_unit_celoss_evalonce/vlm_gym_match_move_fix3_unit_val
3820
- [eval debug] first 3 batch fingerprints:
3821
- fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
3822
- fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
3823
- fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
3824
- ce_avg: 0.12750361859798431, mse_avg: 0.007511829491704702
3825
- base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins/eval_used_rows, step_tag is checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins_step4500
3826
- Preparing Dataset vlm_gym_match_move_fix3_unit_celoss_evalonce/vlm_gym_match_move_fix3_unit_val
3827
- [eval debug] first 3 batch fingerprints:
3828
- fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
3829
- fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
3830
- fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
3831
  [2026-01-27 08:32:53] (step=0003575) Train Loss mse: 0.0074, Train Loss ce: 0.1406, Train Steps/Sec: 0.15,
3832
  [2026-01-27 08:33:00] (step=0003576) Train Loss mse: 0.0093, Train Loss ce: 0.1464, Train Steps/Sec: 0.15,
3833
  [2026-01-27 08:33:06] (step=0003577) Train Loss mse: 0.0077, Train Loss ce: 0.1396, Train Steps/Sec: 0.15,
@@ -4953,13 +4946,6 @@ Preparing Dataset vlm_gym_match_move_fix3_unit_celoss_evalonce/vlm_gym_match_mov
4953
  [2026-01-27 10:34:38] (step=0004697) Train Loss mse: 0.0079, Train Loss ce: 0.1542, Train Steps/Sec: 0.15,
4954
  [2026-01-27 10:34:44] (step=0004698) Train Loss mse: 0.0085, Train Loss ce: 0.1140, Train Steps/Sec: 0.16,
4955
  [2026-01-27 10:34:51] (step=0004699) Train Loss mse: 0.0074, Train Loss ce: 0.1132, Train Steps/Sec: 0.15,
4956
- base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins/eval_used_rows, step_tag is checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins_step5000
4957
- Preparing Dataset vlm_gym_match_move_fix3_unit_celoss_evalonce/vlm_gym_match_move_fix3_unit_val
4958
- [eval debug] first 3 batch fingerprints:
4959
- fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
4960
- fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
4961
- fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
4962
- ce_avg: 0.1215158998966217, mse_avg: 0.0074405064806342125
4963
  [2026-01-27 10:34:58] (step=0004700) Train Loss mse: 0.0072, Train Loss ce: 0.1064, Train Steps/Sec: 0.14,
4964
  [2026-01-27 10:35:04] (step=0004701) Train Loss mse: 0.0087, Train Loss ce: 0.1294, Train Steps/Sec: 0.16,
4965
  [2026-01-27 10:35:11] (step=0004702) Train Loss mse: 0.0079, Train Loss ce: 0.1206, Train Steps/Sec: 0.16,
@@ -5262,4 +5248,11 @@ ce_avg: 0.1215158998966217, mse_avg: 0.0074405064806342125
5262
  [2026-01-27 11:07:18] (step=0004999) Train Loss mse: 0.0092, Train Loss ce: 0.1221, Train Steps/Sec: 0.16,
5263
  [2026-01-27 11:07:56] (step=0005000) Train Loss mse: 0.0077, Train Loss ce: 0.1160, Train Steps/Sec: 0.03,
5264
  [2026-01-27 11:07:56] Saving checkpoint to /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins/0005000.
5265
- [2026-01-27 11:10:32] Done!
 
 
 
 
 
 
 
 
1008
  [2026-01-27 03:48:48] (step=0000997) Train Loss mse: 0.0083, Train Loss ce: 0.1699, Train Steps/Sec: 0.15,
1009
  [2026-01-27 03:48:54] (step=0000998) Train Loss mse: 0.0078, Train Loss ce: 0.1487, Train Steps/Sec: 0.15,
1010
  [2026-01-27 03:49:01] (step=0000999) Train Loss mse: 0.0096, Train Loss ce: 0.1769, Train Steps/Sec: 0.15,
1011
+ [2026-01-27 03:49:39] (step=0001000) Train Loss mse: 0.0100, Train Loss ce: 0.1701, Train Steps/Sec: 0.03,
1012
+ [2026-01-27 03:49:45] (step=0001001) Train Loss mse: 0.0078, Train Loss ce: 0.1978, Train Steps/Sec: 0.16,
1013
+ [2026-01-27 03:49:51] (step=0001002) Train Loss mse: 0.0093, Train Loss ce: 0.1890, Train Steps/Sec: 0.16,
1014
+ [2026-01-27 03:49:58] (step=0001003) Train Loss mse: 0.0089, Train Loss ce: 0.1751, Train Steps/Sec: 0.16,
1015
+ [2026-01-27 03:50:04] (step=0001004) Train Loss mse: 0.0100, Train Loss ce: 0.1660, Train Steps/Sec: 0.16,
1016
+ [2026-01-27 03:50:10] (step=0001005) Train Loss mse: 0.0092, Train Loss ce: 0.1881, Train Steps/Sec: 0.16,
1017
+ [2026-01-27 03:50:17] (step=0001006) Train Loss mse: 0.0088, Train Loss ce: 0.1926, Train Steps/Sec: 0.15,
1018
+ [2026-01-27 03:50:24] (step=0001007) Train Loss mse: 0.0083, Train Loss ce: 0.1839, Train Steps/Sec: 0.15,
1019
+ [2026-01-27 03:50:30] (step=0001008) Train Loss mse: 0.0097, Train Loss ce: 0.2083, Train Steps/Sec: 0.16,
1020
+ [2026-01-27 03:50:37] (step=0001009) Train Loss mse: 0.0083, Train Loss ce: 0.1949, Train Steps/Sec: 0.16,
1021
+ [2026-01-27 03:50:43] (step=0001010) Train Loss mse: 0.0074, Train Loss ce: 0.1642, Train Steps/Sec: 0.16,
1022
+ [2026-01-27 03:50:49] (step=0001011) Train Loss mse: 0.0095, Train Loss ce: 0.2152, Train Steps/Sec: 0.16,
1023
+ [2026-01-27 03:50:56] (step=0001012) Train Loss mse: 0.0085, Train Loss ce: 0.1922, Train Steps/Sec: 0.16,
1024
+ [2026-01-27 03:51:02] (step=0001013) Train Loss mse: 0.0084, Train Loss ce: 0.1974, Train Steps/Sec: 0.16,
1025
+ [2026-01-27 03:51:09] (step=0001014) Train Loss mse: 0.0080, Train Loss ce: 0.1732, Train Steps/Sec: 0.15,
1026
+ [2026-01-27 03:51:15] (step=0001015) Train Loss mse: 0.0084, Train Loss ce: 0.2082, Train Steps/Sec: 0.16,
1027
+ [2026-01-27 03:51:22] (step=0001016) Train Loss mse: 0.0088, Train Loss ce: 0.1730, Train Steps/Sec: 0.15,
1028
+ [2026-01-27 03:51:28] (step=0001017) Train Loss mse: 0.0093, Train Loss ce: 0.2090, Train Steps/Sec: 0.16,
1029
+ [2026-01-27 03:51:34] (step=0001018) Train Loss mse: 0.0091, Train Loss ce: 0.2026, Train Steps/Sec: 0.16,
1030
+ [2026-01-27 03:51:41] (step=0001019) Train Loss mse: 0.0095, Train Loss ce: 0.1736, Train Steps/Sec: 0.16,
1031
+ [2026-01-27 03:51:47] (step=0001020) Train Loss mse: 0.0092, Train Loss ce: 0.2094, Train Steps/Sec: 0.16,
1032
+ [2026-01-27 03:51:53] (step=0001021) Train Loss mse: 0.0091, Train Loss ce: 0.1846, Train Steps/Sec: 0.16,
1033
+ [2026-01-27 03:52:00] (step=0001022) Train Loss mse: 0.0093, Train Loss ce: 0.1848, Train Steps/Sec: 0.16,
1034
+ [2026-01-27 03:52:06] (step=0001023) Train Loss mse: 0.0078, Train Loss ce: 0.1752, Train Steps/Sec: 0.15,
1035
+ [2026-01-27 03:52:13] (step=0001024) Train Loss mse: 0.0087, Train Loss ce: 0.1691, Train Steps/Sec: 0.14,
1036
+ [2026-01-27 03:52:20] (step=0001025) Train Loss mse: 0.0084, Train Loss ce: 0.1881, Train Steps/Sec: 0.16,
1037
+ [2026-01-27 03:52:26] (step=0001026) Train Loss mse: 0.0088, Train Loss ce: 0.2076, Train Steps/Sec: 0.16,
1038
+ [2026-01-27 03:52:32] (step=0001027) Train Loss mse: 0.0108, Train Loss ce: 0.1895, Train Steps/Sec: 0.16,
1039
+ [2026-01-27 03:52:39] (step=0001028) Train Loss mse: 0.0085, Train Loss ce: 0.1782, Train Steps/Sec: 0.16,
1040
+ [2026-01-27 03:52:45] (step=0001029) Train Loss mse: 0.0085, Train Loss ce: 0.1808, Train Steps/Sec: 0.16,
1041
+ [2026-01-27 03:52:51] (step=0001030) Train Loss mse: 0.0106, Train Loss ce: 0.1796, Train Steps/Sec: 0.18,
1042
+ [2026-01-27 03:52:57] (step=0001031) Train Loss mse: 0.0087, Train Loss ce: 0.1868, Train Steps/Sec: 0.16,
1043
+ [2026-01-27 03:53:04] (step=0001032) Train Loss mse: 0.0089, Train Loss ce: 0.2020, Train Steps/Sec: 0.15,
1044
+ [2026-01-27 03:53:11] (step=0001033) Train Loss mse: 0.0086, Train Loss ce: 0.1978, Train Steps/Sec: 0.14,
1045
+ [2026-01-27 03:53:17] (step=0001034) Train Loss mse: 0.0092, Train Loss ce: 0.1917, Train Steps/Sec: 0.15,
1046
+ [2026-01-27 03:53:24] (step=0001035) Train Loss mse: 0.0099, Train Loss ce: 0.1958, Train Steps/Sec: 0.16,
1047
+ [2026-01-27 03:53:30] (step=0001036) Train Loss mse: 0.0081, Train Loss ce: 0.1909, Train Steps/Sec: 0.16,
1048
+ [2026-01-27 03:53:36] (step=0001037) Train Loss mse: 0.0079, Train Loss ce: 0.1868, Train Steps/Sec: 0.16,
1049
+ [2026-01-27 03:53:43] (step=0001038) Train Loss mse: 0.0103, Train Loss ce: 0.1744, Train Steps/Sec: 0.16,
1050
+ [2026-01-27 03:53:49] (step=0001039) Train Loss mse: 0.0090, Train Loss ce: 0.2039, Train Steps/Sec: 0.16,
1051
+ [2026-01-27 03:53:55] (step=0001040) Train Loss mse: 0.0081, Train Loss ce: 0.1573, Train Steps/Sec: 0.16,
1052
+ [2026-01-27 03:54:02] (step=0001041) Train Loss mse: 0.0091, Train Loss ce: 0.1810, Train Steps/Sec: 0.15,
1053
+ [2026-01-27 03:54:09] (step=0001042) Train Loss mse: 0.0087, Train Loss ce: 0.2339, Train Steps/Sec: 0.15,
1054
+ [2026-01-27 03:54:15] (step=0001043) Train Loss mse: 0.0082, Train Loss ce: 0.1785, Train Steps/Sec: 0.15,
1055
+ [2026-01-27 03:54:22] (step=0001044) Train Loss mse: 0.0112, Train Loss ce: 0.2073, Train Steps/Sec: 0.16,
1056
+ [2026-01-27 03:54:28] (step=0001045) Train Loss mse: 0.0096, Train Loss ce: 0.1835, Train Steps/Sec: 0.16,
1057
+ [2026-01-27 03:54:34] (step=0001046) Train Loss mse: 0.0086, Train Loss ce: 0.1964, Train Steps/Sec: 0.16,
1058
  FullyShardedDataParallel(
1059
  (_fsdp_wrapped_module): Bagel(
1060
  (language_model): Qwen2ForCausalLM(
 
1241
  fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
1242
  fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
1243
  ce_avg: 0.20799879729747772, mse_avg: 0.007969205267727375
 
 
 
 
 
 
 
1244
  base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins/eval_used_rows, step_tag is checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins_step1500
1245
  Preparing Dataset vlm_gym_match_move_fix3_unit_celoss_evalonce/vlm_gym_match_move_fix3_unit_val
1246
  [eval debug] first 3 batch fingerprints:
 
1255
  fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
1256
  fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
1257
  ce_avg: 0.1749173104763031, mse_avg: 0.009318988770246506
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1258
  [2026-01-27 03:54:41] (step=0001047) Train Loss mse: 0.0085, Train Loss ce: 0.1884, Train Steps/Sec: 0.16,
1259
  [2026-01-27 03:54:47] (step=0001048) Train Loss mse: 0.0083, Train Loss ce: 0.2145, Train Steps/Sec: 0.16,
1260
  [2026-01-27 03:54:53] (step=0001049) Train Loss mse: 0.0093, Train Loss ce: 0.1888, Train Steps/Sec: 0.16,
 
2603
  [2026-01-27 06:21:07] (step=0002392) Train Loss mse: 0.0086, Train Loss ce: 0.1506, Train Steps/Sec: 0.16,
2604
  [2026-01-27 06:21:14] (step=0002393) Train Loss mse: 0.0084, Train Loss ce: 0.1657, Train Steps/Sec: 0.15,
2605
  [2026-01-27 06:21:20] (step=0002394) Train Loss mse: 0.0074, Train Loss ce: 0.1535, Train Steps/Sec: 0.15,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2606
  [2026-01-27 06:21:27] (step=0002395) Train Loss mse: 0.0083, Train Loss ce: 0.1900, Train Steps/Sec: 0.16,
2607
  [2026-01-27 06:21:34] (step=0002396) Train Loss mse: 0.0074, Train Loss ce: 0.1865, Train Steps/Sec: 0.15,
2608
  [2026-01-27 06:21:40] (step=0002397) Train Loss mse: 0.0096, Train Loss ce: 0.1719, Train Steps/Sec: 0.15,
 
2702
  [2026-01-27 06:31:50] (step=0002491) Train Loss mse: 0.0079, Train Loss ce: 0.1600, Train Steps/Sec: 0.16,
2703
  [2026-01-27 06:31:57] (step=0002492) Train Loss mse: 0.0075, Train Loss ce: 0.1430, Train Steps/Sec: 0.15,
2704
  [2026-01-27 06:32:03] (step=0002493) Train Loss mse: 0.0081, Train Loss ce: 0.1674, Train Steps/Sec: 0.15,
2705
+ base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins/eval_used_rows, step_tag is checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins_step2500
2706
+ Preparing Dataset vlm_gym_match_move_fix3_unit_celoss_evalonce/vlm_gym_match_move_fix3_unit_val
2707
+ [eval debug] first 3 batch fingerprints:
2708
+ fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
2709
+ fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
2710
+ fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
2711
+ ce_avg: 0.16938790678977966, mse_avg: 0.009705973789095879
2712
+ base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins/eval_used_rows, step_tag is checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins_step3000
2713
+ Preparing Dataset vlm_gym_match_move_fix3_unit_celoss_evalonce/vlm_gym_match_move_fix3_unit_val
2714
+ [eval debug] first 3 batch fingerprints:
2715
+ fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
2716
+ fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
2717
+ fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
2718
+ ce_avg: 0.1402648687362671, mse_avg: 0.007266872562468052
2719
+ base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins/eval_used_rows, step_tag is checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins_step3500
2720
+ Preparing Dataset vlm_gym_match_move_fix3_unit_celoss_evalonce/vlm_gym_match_move_fix3_unit_val
2721
+ [eval debug] first 3 batch fingerprints:
2722
+ fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
2723
+ fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
2724
+ fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
2725
+ ce_avg: 0.13325323164463043, mse_avg: 0.007202878128737211
2726
  [2026-01-27 06:32:10] (step=0002494) Train Loss mse: 0.0080, Train Loss ce: 0.1591, Train Steps/Sec: 0.15,
2727
  [2026-01-27 06:32:17] (step=0002495) Train Loss mse: 0.0089, Train Loss ce: 0.1406, Train Steps/Sec: 0.15,
2728
  [2026-01-27 06:32:23] (step=0002496) Train Loss mse: 0.0077, Train Loss ce: 0.1890, Train Steps/Sec: 0.15,
 
3751
  [2026-01-27 08:26:36] (step=0003516) Train Loss mse: 0.0078, Train Loss ce: 0.1360, Train Steps/Sec: 0.16,
3752
  [2026-01-27 08:26:43] (step=0003517) Train Loss mse: 0.0089, Train Loss ce: 0.1218, Train Steps/Sec: 0.16,
3753
  [2026-01-27 08:26:49] (step=0003518) Train Loss mse: 0.0083, Train Loss ce: 0.1212, Train Steps/Sec: 0.16,
3754
+ base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins/eval_used_rows, step_tag is checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins_step4000
3755
+ Preparing Dataset vlm_gym_match_move_fix3_unit_celoss_evalonce/vlm_gym_match_move_fix3_unit_val
3756
+ [eval debug] first 3 batch fingerprints:
3757
+ fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
3758
+ fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
3759
+ fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
3760
+ ce_avg: 0.12750361859798431, mse_avg: 0.007511829491704702
3761
+ base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins/eval_used_rows, step_tag is checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins_step4500
3762
+ Preparing Dataset vlm_gym_match_move_fix3_unit_celoss_evalonce/vlm_gym_match_move_fix3_unit_val
3763
+ [eval debug] first 3 batch fingerprints:
3764
+ fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
3765
+ fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
3766
+ fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
3767
+ ce_avg: 0.12324703484773636, mse_avg: 0.007881568744778633
3768
  [2026-01-27 08:26:56] (step=0003519) Train Loss mse: 0.0099, Train Loss ce: 0.1251, Train Steps/Sec: 0.16,
3769
  [2026-01-27 08:27:02] (step=0003520) Train Loss mse: 0.0078, Train Loss ce: 0.1416, Train Steps/Sec: 0.16,
3770
  [2026-01-27 08:27:08] (step=0003521) Train Loss mse: 0.0084, Train Loss ce: 0.1459, Train Steps/Sec: 0.16,
 
3821
  [2026-01-27 08:32:34] (step=0003572) Train Loss mse: 0.0101, Train Loss ce: 0.1471, Train Steps/Sec: 0.18,
3822
  [2026-01-27 08:32:40] (step=0003573) Train Loss mse: 0.0079, Train Loss ce: 0.1128, Train Steps/Sec: 0.16,
3823
  [2026-01-27 08:32:46] (step=0003574) Train Loss mse: 0.0077, Train Loss ce: 0.1507, Train Steps/Sec: 0.16,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3824
  [2026-01-27 08:32:53] (step=0003575) Train Loss mse: 0.0074, Train Loss ce: 0.1406, Train Steps/Sec: 0.15,
3825
  [2026-01-27 08:33:00] (step=0003576) Train Loss mse: 0.0093, Train Loss ce: 0.1464, Train Steps/Sec: 0.15,
3826
  [2026-01-27 08:33:06] (step=0003577) Train Loss mse: 0.0077, Train Loss ce: 0.1396, Train Steps/Sec: 0.15,
 
4946
  [2026-01-27 10:34:38] (step=0004697) Train Loss mse: 0.0079, Train Loss ce: 0.1542, Train Steps/Sec: 0.15,
4947
  [2026-01-27 10:34:44] (step=0004698) Train Loss mse: 0.0085, Train Loss ce: 0.1140, Train Steps/Sec: 0.16,
4948
  [2026-01-27 10:34:51] (step=0004699) Train Loss mse: 0.0074, Train Loss ce: 0.1132, Train Steps/Sec: 0.15,
 
 
 
 
 
 
 
4949
  [2026-01-27 10:34:58] (step=0004700) Train Loss mse: 0.0072, Train Loss ce: 0.1064, Train Steps/Sec: 0.14,
4950
  [2026-01-27 10:35:04] (step=0004701) Train Loss mse: 0.0087, Train Loss ce: 0.1294, Train Steps/Sec: 0.16,
4951
  [2026-01-27 10:35:11] (step=0004702) Train Loss mse: 0.0079, Train Loss ce: 0.1206, Train Steps/Sec: 0.16,
 
5248
  [2026-01-27 11:07:18] (step=0004999) Train Loss mse: 0.0092, Train Loss ce: 0.1221, Train Steps/Sec: 0.16,
5249
  [2026-01-27 11:07:56] (step=0005000) Train Loss mse: 0.0077, Train Loss ce: 0.1160, Train Steps/Sec: 0.03,
5250
  [2026-01-27 11:07:56] Saving checkpoint to /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins/0005000.
5251
+ [2026-01-27 11:10:32] Done!
5252
+ base_dir is /dev/shm/models/checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins/eval_used_rows, step_tag is checkpoints_vlm_gym_match_move_fix3_unit_one_image_lr2e_5_ce_ins_step5000
5253
+ Preparing Dataset vlm_gym_match_move_fix3_unit_celoss_evalonce/vlm_gym_match_move_fix3_unit_val
5254
+ [eval debug] first 3 batch fingerprints:
5255
+ fp[0]: [{'data_indexes': [0], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
5256
+ fp[1]: [{'data_indexes': [8], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
5257
+ fp[2]: [{'data_indexes': [16], 'worker_id': 0, 'dataset_name': 'vlm_gym_match_move_fix3_unit_celoss_evalonce'}]
5258
+ ce_avg: 0.1215158998966217, mse_avg: 0.0074405064806342125