Hugging Face's logo

mlfoundations-dev
/
a1_math_numina_math

Text Generation
Transformers
Safetensors
qwen2
llama-factory
full
Generated from Trainer
conversational
text-generation-inference
Model card Files
xet
Community
1
ryanmarten commited on
Commit
d2e6e2f
·
verified ·
1 Parent(s): 0a717c1

Training in progress, epoch 3

Browse files
Files changed (6) hide show
  1. model-00001-of-00004.safetensors +1 -1
  2. model-00002-of-00004.safetensors +1 -1
  3. model-00003-of-00004.safetensors +1 -1
  4. model-00004-of-00004.safetensors +1 -1
  5. trainer_log.jsonl +310 -0
  6. training_args.bin +1 -1
model-00001-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:853f61fc073e49c4c073c1b9e87d6340ba590bd5b5feddf32ed45ec61a944c37
3
  size 4877660776
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7084ec7d9ce330d83f5a4e6297e747ef4b00f4eb012aa0bdc2ff09eeefd3ec8e
3
  size 4877660776
model-00002-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:449169069bb788ad2cbb9d324f439e79020566dea446183e7a88be7e292ff4bd
3
  size 4932751008
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:85569831145490912ed3c505b0c065606aa72c679894d9d93581137a06f194b4
3
  size 4932751008
model-00003-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:baa1ca31da61db8871b04506856ce0b1165a8888c509e621a8b108ee51929924
3
  size 4330865200
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8230b170f1e59b604bd34712f8706cc677c930004b39e83f8d6bc20a5f6489e5
3
  size 4330865200
model-00004-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c6f647a11e02f4dd03cb611b546715122cde930b6c78df28fa1f8f54bd5efee1
3
  size 1089994880
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:942140ed438a19bc1cbccce4d0dce776bdb108dfacf40f9884e48f93155f8749
3
  size 1089994880
trainer_log.jsonl CHANGED
@@ -1576,3 +1576,313 @@
1576
  {"current_steps": 834, "total_steps": 1230, "loss": 0.1774, "lr": 1.135521394010506e-05, "epoch": 3.3827848101265823, "percentage": 67.8, "elapsed_time": "19:37:24", "remaining_time": "9:19:03"}
1577
  {"current_steps": 743, "total_steps": 1230, "loss": 0.1758, "lr": 1.624791353909428e-05, "epoch": 3.0141772151898736, "percentage": 60.41, "elapsed_time": "17:27:31", "remaining_time": "11:26:36"}
1578
  {"current_steps": 835, "total_steps": 1230, "loss": 0.1816, "lr": 1.1304066178317367e-05, "epoch": 3.386835443037975, "percentage": 67.89, "elapsed_time": "19:38:48", "remaining_time": "9:17:38"}
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1576
  {"current_steps": 834, "total_steps": 1230, "loss": 0.1774, "lr": 1.135521394010506e-05, "epoch": 3.3827848101265823, "percentage": 67.8, "elapsed_time": "19:37:24", "remaining_time": "9:19:03"}
1577
  {"current_steps": 743, "total_steps": 1230, "loss": 0.1758, "lr": 1.624791353909428e-05, "epoch": 3.0141772151898736, "percentage": 60.41, "elapsed_time": "17:27:31", "remaining_time": "11:26:36"}
1578
  {"current_steps": 835, "total_steps": 1230, "loss": 0.1816, "lr": 1.1304066178317367e-05, "epoch": 3.386835443037975, "percentage": 67.89, "elapsed_time": "19:38:48", "remaining_time": "9:17:38"}
1579
+ {"current_steps": 744, "total_steps": 1230, "loss": 0.1671, "lr": 1.619217781670663e-05, "epoch": 3.0182278481012657, "percentage": 60.49, "elapsed_time": "17:28:50", "remaining_time": "11:25:08"}
1580
+ {"current_steps": 836, "total_steps": 1230, "loss": 0.1838, "lr": 1.1252988452385199e-05, "epoch": 3.390886075949367, "percentage": 67.97, "elapsed_time": "19:40:14", "remaining_time": "9:16:14"}
1581
+ {"current_steps": 745, "total_steps": 1230, "loss": 0.1892, "lr": 1.6136472761995373e-05, "epoch": 3.022278481012658, "percentage": 60.57, "elapsed_time": "17:30:24", "remaining_time": "11:23:49"}
1582
+ {"current_steps": 837, "total_steps": 1230, "loss": 0.1733, "lr": 1.1201981173681536e-05, "epoch": 3.3949367088607594, "percentage": 68.05, "elapsed_time": "19:41:43", "remaining_time": "9:14:51"}
1583
+ {"current_steps": 746, "total_steps": 1230, "loss": 0.16, "lr": 1.608079882360134e-05, "epoch": 3.0263291139240507, "percentage": 60.65, "elapsed_time": "17:31:46", "remaining_time": "11:22:23"}
1584
+ {"current_steps": 838, "total_steps": 1230, "loss": 0.18, "lr": 1.1151044753011991e-05, "epoch": 3.398987341772152, "percentage": 68.13, "elapsed_time": "19:43:07", "remaining_time": "9:13:26"}
1585
+ {"current_steps": 747, "total_steps": 1230, "loss": 0.176, "lr": 1.60251564499148e-05, "epoch": 3.030379746835443, "percentage": 60.73, "elapsed_time": "17:33:05", "remaining_time": "11:20:54"}
1586
+ {"current_steps": 839, "total_steps": 1230, "loss": 0.1952, "lr": 1.1100179600611491e-05, "epoch": 3.4030379746835444, "percentage": 68.21, "elapsed_time": "19:44:44", "remaining_time": "9:12:07"}
1587
+ {"current_steps": 748, "total_steps": 1230, "loss": 0.171, "lr": 1.596954608907176e-05, "epoch": 3.0344303797468353, "percentage": 60.81, "elapsed_time": "17:34:20", "remaining_time": "11:19:23"}
1588
+ {"current_steps": 840, "total_steps": 1230, "loss": 0.1738, "lr": 1.1049386126140985e-05, "epoch": 3.407088607594937, "percentage": 68.29, "elapsed_time": "19:46:07", "remaining_time": "9:10:42"}
1589
+ {"current_steps": 749, "total_steps": 1230, "loss": 0.1766, "lr": 1.591396818895043e-05, "epoch": 3.038481012658228, "percentage": 60.89, "elapsed_time": "17:35:41", "remaining_time": "11:17:57"}
1590
+ {"current_steps": 841, "total_steps": 1230, "loss": 0.1568, "lr": 1.0998664738684128e-05, "epoch": 3.411139240506329, "percentage": 68.37, "elapsed_time": "19:47:38", "remaining_time": "9:09:20"}
1591
+ {"current_steps": 750, "total_steps": 1230, "loss": 0.194, "lr": 1.585842319716759e-05, "epoch": 3.0425316455696203, "percentage": 60.98, "elapsed_time": "17:37:14", "remaining_time": "11:16:38"}
1592
+ {"current_steps": 842, "total_steps": 1230, "loss": 0.1772, "lr": 1.0948015846744e-05, "epoch": 3.4151898734177215, "percentage": 68.46, "elapsed_time": "19:48:58", "remaining_time": "9:07:53"}
1593
+ {"current_steps": 751, "total_steps": 1230, "loss": 0.1912, "lr": 1.5802911561074944e-05, "epoch": 3.046582278481013, "percentage": 61.06, "elapsed_time": "17:38:36", "remaining_time": "11:15:11"}
1594
+ {"current_steps": 843, "total_steps": 1230, "loss": 0.1857, "lr": 1.0897439858239832e-05, "epoch": 3.419240506329114, "percentage": 68.54, "elapsed_time": "19:50:22", "remaining_time": "9:06:28"}
1595
+ {"current_steps": 752, "total_steps": 1230, "loss": 0.1747, "lr": 1.5747433727755595e-05, "epoch": 3.050632911392405, "percentage": 61.14, "elapsed_time": "17:40:03", "remaining_time": "11:13:48"}
1596
+ {"current_steps": 844, "total_steps": 1230, "loss": 0.189, "lr": 1.0846937180503652e-05, "epoch": 3.423291139240506, "percentage": 68.62, "elapsed_time": "19:52:01", "remaining_time": "9:05:10"}
1597
+ {"current_steps": 753, "total_steps": 1230, "loss": 0.1875, "lr": 1.5691990144020376e-05, "epoch": 3.0546835443037974, "percentage": 61.22, "elapsed_time": "17:41:19", "remaining_time": "11:12:18"}
1598
+ {"current_steps": 845, "total_steps": 1230, "loss": 0.1695, "lr": 1.0796508220277117e-05, "epoch": 3.4273417721518986, "percentage": 68.7, "elapsed_time": "19:53:32", "remaining_time": "9:03:48"}
1599
+ {"current_steps": 754, "total_steps": 1230, "loss": 0.1475, "lr": 1.5636581256404297e-05, "epoch": 3.05873417721519, "percentage": 61.3, "elapsed_time": "17:42:39", "remaining_time": "11:10:51"}
1600
+ {"current_steps": 846, "total_steps": 1230, "loss": 0.1714, "lr": 1.0746153383708107e-05, "epoch": 3.431392405063291, "percentage": 68.78, "elapsed_time": "19:54:51", "remaining_time": "9:02:20"}
1601
+ {"current_steps": 755, "total_steps": 1230, "loss": 0.1776, "lr": 1.558120751116291e-05, "epoch": 3.0627848101265824, "percentage": 61.38, "elapsed_time": "17:44:04", "remaining_time": "11:09:27"}
1602
+ {"current_steps": 847, "total_steps": 1230, "loss": 0.1709, "lr": 1.0695873076347579e-05, "epoch": 3.4354430379746836, "percentage": 68.86, "elapsed_time": "19:56:15", "remaining_time": "9:00:55"}
1603
+ {"current_steps": 756, "total_steps": 1230, "loss": 0.1598, "lr": 1.552586935426876e-05, "epoch": 3.0668354430379745, "percentage": 61.46, "elapsed_time": "17:45:15", "remaining_time": "11:07:53"}
1604
+ {"current_steps": 848, "total_steps": 1230, "loss": 0.1796, "lr": 1.0645667703146205e-05, "epoch": 3.439493670886076, "percentage": 68.94, "elapsed_time": "19:57:45", "remaining_time": "8:59:33"}
1605
+ {"current_steps": 757, "total_steps": 1230, "loss": 0.1702, "lr": 1.547056723140774e-05, "epoch": 3.070886075949367, "percentage": 61.54, "elapsed_time": "17:46:48", "remaining_time": "11:06:34"}
1606
+ {"current_steps": 849, "total_steps": 1230, "loss": 0.159, "lr": 1.0595537668451161e-05, "epoch": 3.4435443037974682, "percentage": 69.02, "elapsed_time": "19:58:58", "remaining_time": "8:58:03"}
1607
+ {"current_steps": 758, "total_steps": 1230, "loss": 0.1725, "lr": 1.5415301587975565e-05, "epoch": 3.0749367088607595, "percentage": 61.63, "elapsed_time": "17:48:02", "remaining_time": "11:05:03"}
1608
+ {"current_steps": 850, "total_steps": 1230, "loss": 0.1584, "lr": 1.0545483376002854e-05, "epoch": 3.4475949367088607, "percentage": 69.11, "elapsed_time": "20:00:02", "remaining_time": "8:56:29"}
1609
+ {"current_steps": 759, "total_steps": 1230, "loss": 0.1727, "lr": 1.536007286907411e-05, "epoch": 3.078987341772152, "percentage": 61.71, "elapsed_time": "17:49:17", "remaining_time": "11:03:33"}
1610
+ {"current_steps": 851, "total_steps": 1230, "loss": 0.1743, "lr": 1.0495505228931676e-05, "epoch": 3.4516455696202533, "percentage": 69.19, "elapsed_time": "20:01:20", "remaining_time": "8:55:01"}
1611
+ {"current_steps": 760, "total_steps": 1230, "loss": 0.1583, "lr": 1.5304881519507896e-05, "epoch": 3.083037974683544, "percentage": 61.79, "elapsed_time": "17:50:19", "remaining_time": "11:01:54"}
1612
+ {"current_steps": 761, "total_steps": 1230, "loss": 0.1546, "lr": 1.5249727983780453e-05, "epoch": 3.0870886075949366, "percentage": 61.87, "elapsed_time": "17:51:31", "remaining_time": "11:00:22"}
1613
+ {"current_steps": 852, "total_steps": 1230, "loss": 0.1679, "lr": 1.044560362975474e-05, "epoch": 3.4556962025316453, "percentage": 69.27, "elapsed_time": "20:02:46", "remaining_time": "8:53:37"}
1614
+ {"current_steps": 853, "total_steps": 1230, "loss": 0.1587, "lr": 1.0395778980372695e-05, "epoch": 3.459746835443038, "percentage": 69.35, "elapsed_time": "20:03:55", "remaining_time": "8:52:06"}
1615
+ {"current_steps": 762, "total_steps": 1230, "loss": 0.1728, "lr": 1.5194612706090786e-05, "epoch": 3.091139240506329, "percentage": 61.95, "elapsed_time": "17:52:49", "remaining_time": "10:58:53"}
1616
+ {"current_steps": 854, "total_steps": 1230, "loss": 0.1763, "lr": 1.0346031682066381e-05, "epoch": 3.4637974683544304, "percentage": 69.43, "elapsed_time": "20:05:08", "remaining_time": "8:50:36"}
1617
+ {"current_steps": 763, "total_steps": 1230, "loss": 0.1561, "lr": 1.5139536130329771e-05, "epoch": 3.0951898734177217, "percentage": 62.03, "elapsed_time": "17:54:01", "remaining_time": "10:57:22"}
1618
+ {"current_steps": 764, "total_steps": 1230, "loss": 0.1716, "lr": 1.508449870007656e-05, "epoch": 3.0992405063291137, "percentage": 62.11, "elapsed_time": "17:55:19", "remaining_time": "10:55:53"}
1619
+ {"current_steps": 855, "total_steps": 1230, "loss": 0.1744, "lr": 1.0296362135493724e-05, "epoch": 3.467848101265823, "percentage": 69.51, "elapsed_time": "20:06:34", "remaining_time": "8:49:11"}
1620
+ {"current_steps": 856, "total_steps": 1230, "loss": 0.161, "lr": 1.0246770740686422e-05, "epoch": 3.4718987341772154, "percentage": 69.59, "elapsed_time": "20:07:46", "remaining_time": "8:47:41"}
1621
+ {"current_steps": 765, "total_steps": 1230, "loss": 0.1746, "lr": 1.5029500858595056e-05, "epoch": 3.1032911392405063, "percentage": 62.2, "elapsed_time": "17:56:42", "remaining_time": "10:54:28"}
1622
+ {"current_steps": 766, "total_steps": 1230, "loss": 0.1787, "lr": 1.4974543048830328e-05, "epoch": 3.1073417721518988, "percentage": 62.28, "elapsed_time": "17:58:03", "remaining_time": "10:53:01"}
1623
+ {"current_steps": 857, "total_steps": 1230, "loss": 0.1745, "lr": 1.0197257897046743e-05, "epoch": 3.4759493670886075, "percentage": 69.67, "elapsed_time": "20:09:18", "remaining_time": "8:46:20"}
1624
+ {"current_steps": 858, "total_steps": 1230, "loss": 0.1859, "lr": 1.014782400334433e-05, "epoch": 3.48, "percentage": 69.76, "elapsed_time": "20:10:29", "remaining_time": "8:44:49"}
1625
+ {"current_steps": 767, "total_steps": 1230, "loss": 0.1754, "lr": 1.4919625713405e-05, "epoch": 3.1113924050632913, "percentage": 62.36, "elapsed_time": "17:59:41", "remaining_time": "10:51:45"}
1626
+ {"current_steps": 859, "total_steps": 1230, "loss": 0.1598, "lr": 1.009846945771296e-05, "epoch": 3.4840506329113925, "percentage": 69.84, "elapsed_time": "20:11:38", "remaining_time": "8:43:18"}
1627
+ {"current_steps": 768, "total_steps": 1230, "loss": 0.1776, "lr": 1.4864749294615756e-05, "epoch": 3.1154430379746834, "percentage": 62.44, "elapsed_time": "18:00:58", "remaining_time": "10:50:16"}
1628
+ {"current_steps": 860, "total_steps": 1230, "loss": 0.1548, "lr": 1.0049194657647363e-05, "epoch": 3.4881012658227846, "percentage": 69.92, "elapsed_time": "20:12:47", "remaining_time": "8:41:47"}
1629
+ {"current_steps": 769, "total_steps": 1230, "loss": 0.1631, "lr": 1.4809914234429716e-05, "epoch": 3.119493670886076, "percentage": 62.52, "elapsed_time": "18:02:14", "remaining_time": "10:48:46"}
1630
+ {"current_steps": 861, "total_steps": 1230, "loss": 0.1757, "lr": 1.0000000000000006e-05, "epoch": 3.492151898734177, "percentage": 70.0, "elapsed_time": "20:14:05", "remaining_time": "8:40:19"}
1631
+ {"current_steps": 770, "total_steps": 1230, "loss": 0.1802, "lr": 1.4755120974480923e-05, "epoch": 3.1235443037974684, "percentage": 62.6, "elapsed_time": "18:03:38", "remaining_time": "10:47:21"}
1632
+ {"current_steps": 862, "total_steps": 1230, "loss": 0.155, "lr": 9.950885880977891e-06, "epoch": 3.4962025316455696, "percentage": 70.08, "elapsed_time": "20:15:20", "remaining_time": "8:38:50"}
1633
+ {"current_steps": 771, "total_steps": 1230, "loss": 0.1621, "lr": 1.4700369956066771e-05, "epoch": 3.127594936708861, "percentage": 62.68, "elapsed_time": "18:04:59", "remaining_time": "10:45:55"}
1634
+ {"current_steps": 863, "total_steps": 1230, "loss": 0.1951, "lr": 9.901852696139382e-06, "epoch": 3.500253164556962, "percentage": 70.16, "elapsed_time": "20:16:55", "remaining_time": "8:37:30"}
1635
+ {"current_steps": 772, "total_steps": 1230, "loss": 0.196, "lr": 1.4645661620144413e-05, "epoch": 3.131645569620253, "percentage": 62.76, "elapsed_time": "18:06:37", "remaining_time": "10:44:39"}
1636
+ {"current_steps": 864, "total_steps": 1230, "loss": 0.1998, "lr": 9.852900840391027e-06, "epoch": 3.5043037974683546, "percentage": 70.24, "elapsed_time": "20:18:29", "remaining_time": "8:36:09"}
1637
+ {"current_steps": 773, "total_steps": 1230, "loss": 0.17, "lr": 1.4590996407327284e-05, "epoch": 3.1356962025316455, "percentage": 62.85, "elapsed_time": "18:07:54", "remaining_time": "10:43:10"}
1638
+ {"current_steps": 865, "total_steps": 1230, "loss": 0.1786, "lr": 9.804030707984313e-06, "epoch": 3.5083544303797467, "percentage": 70.33, "elapsed_time": "20:20:04", "remaining_time": "8:34:49"}
1639
+ {"current_steps": 774, "total_steps": 1230, "loss": 0.166, "lr": 1.4536374757881487e-05, "epoch": 3.139746835443038, "percentage": 62.93, "elapsed_time": "18:09:06", "remaining_time": "10:41:38"}
1640
+ {"current_steps": 866, "total_steps": 1230, "loss": 0.1701, "lr": 9.755242692512599e-06, "epoch": 3.512405063291139, "percentage": 70.41, "elapsed_time": "20:21:18", "remaining_time": "8:33:20"}
1641
+ {"current_steps": 775, "total_steps": 1230, "loss": 0.1809, "lr": 1.4481797111722271e-05, "epoch": 3.1437974683544305, "percentage": 63.01, "elapsed_time": "18:10:36", "remaining_time": "10:40:17"}
1642
+ {"current_steps": 867, "total_steps": 1230, "loss": 0.1656, "lr": 9.70653718690782e-06, "epoch": 3.5164556962025317, "percentage": 70.49, "elapsed_time": "20:22:35", "remaining_time": "8:31:53"}
1643
+ {"current_steps": 776, "total_steps": 1230, "loss": 0.1781, "lr": 1.4427263908410507e-05, "epoch": 3.1478481012658226, "percentage": 63.09, "elapsed_time": "18:11:50", "remaining_time": "10:38:46"}
1644
+ {"current_steps": 868, "total_steps": 1230, "loss": 0.1607, "lr": 9.657914583437454e-06, "epoch": 3.520506329113924, "percentage": 70.57, "elapsed_time": "20:23:58", "remaining_time": "8:30:27"}
1645
+ {"current_steps": 777, "total_steps": 1230, "loss": 0.1685, "lr": 1.4372775587149108e-05, "epoch": 3.151898734177215, "percentage": 63.17, "elapsed_time": "18:13:14", "remaining_time": "10:37:22"}
1646
+ {"current_steps": 869, "total_steps": 1230, "loss": 0.1714, "lr": 9.609375273701246e-06, "epoch": 3.5245569620253163, "percentage": 70.65, "elapsed_time": "20:25:12", "remaining_time": "8:28:58"}
1647
+ {"current_steps": 778, "total_steps": 1230, "loss": 0.1984, "lr": 1.4318332586779522e-05, "epoch": 3.1559493670886076, "percentage": 63.25, "elapsed_time": "18:14:46", "remaining_time": "10:36:02"}
1648
+ {"current_steps": 870, "total_steps": 1230, "loss": 0.1927, "lr": 9.560919648628133e-06, "epoch": 3.528607594936709, "percentage": 70.73, "elapsed_time": "20:26:41", "remaining_time": "8:27:35"}
1649
+ {"current_steps": 779, "total_steps": 1230, "loss": 0.1737, "lr": 1.4263935345778202e-05, "epoch": 3.16, "percentage": 63.33, "elapsed_time": "18:16:12", "remaining_time": "10:34:38"}
1650
+ {"current_steps": 871, "total_steps": 1230, "loss": 0.1676, "lr": 9.512548098473047e-06, "epoch": 3.5326582278481014, "percentage": 70.81, "elapsed_time": "20:27:59", "remaining_time": "8:26:08"}
1651
+ {"current_steps": 780, "total_steps": 1230, "loss": 0.163, "lr": 1.420958430225303e-05, "epoch": 3.164050632911392, "percentage": 63.41, "elapsed_time": "18:17:26", "remaining_time": "10:33:08"}
1652
+ {"current_steps": 872, "total_steps": 1230, "loss": 0.1788, "lr": 9.464261012813825e-06, "epoch": 3.536708860759494, "percentage": 70.89, "elapsed_time": "20:29:25", "remaining_time": "8:24:44"}
1653
+ {"current_steps": 781, "total_steps": 1230, "loss": 0.1738, "lr": 1.415527989393985e-05, "epoch": 3.1681012658227847, "percentage": 63.5, "elapsed_time": "18:18:51", "remaining_time": "10:31:44"}
1654
+ {"current_steps": 782, "total_steps": 1230, "loss": 0.1582, "lr": 1.410102255819891e-05, "epoch": 3.1721518987341772, "percentage": 63.58, "elapsed_time": "18:19:53", "remaining_time": "10:30:07"}
1655
+ {"current_steps": 873, "total_steps": 1230, "loss": 0.1885, "lr": 9.416058780547987e-06, "epoch": 3.540759493670886, "percentage": 70.98, "elapsed_time": "20:31:05", "remaining_time": "8:23:26"}
1656
+ {"current_steps": 874, "total_steps": 1230, "loss": 0.149, "lr": 9.367941789889714e-06, "epoch": 3.5448101265822785, "percentage": 71.06, "elapsed_time": "20:32:14", "remaining_time": "8:21:55"}
1657
+ {"current_steps": 783, "total_steps": 1230, "loss": 0.1877, "lr": 1.404681273201131e-05, "epoch": 3.1762025316455698, "percentage": 63.66, "elapsed_time": "18:21:15", "remaining_time": "10:28:41"}
1658
+ {"current_steps": 875, "total_steps": 1230, "loss": 0.1744, "lr": 9.319910428366607e-06, "epoch": 3.548860759493671, "percentage": 71.14, "elapsed_time": "20:33:44", "remaining_time": "8:20:32"}
1659
+ {"current_steps": 784, "total_steps": 1230, "loss": 0.1794, "lr": 1.399265085197556e-05, "epoch": 3.180253164556962, "percentage": 63.74, "elapsed_time": "18:22:37", "remaining_time": "10:27:15"}
1660
+ {"current_steps": 876, "total_steps": 1230, "loss": 0.1604, "lr": 9.271965082816667e-06, "epoch": 3.552911392405063, "percentage": 71.22, "elapsed_time": "20:35:02", "remaining_time": "8:19:05"}
1661
+ {"current_steps": 785, "total_steps": 1230, "loss": 0.1871, "lr": 1.393853735430398e-05, "epoch": 3.1843037974683543, "percentage": 63.82, "elapsed_time": "18:24:12", "remaining_time": "10:25:57"}
1662
+ {"current_steps": 877, "total_steps": 1230, "loss": 0.1755, "lr": 9.224106139385111e-06, "epoch": 3.5569620253164556, "percentage": 71.3, "elapsed_time": "20:36:29", "remaining_time": "8:17:41"}
1663
+ {"current_steps": 786, "total_steps": 1230, "loss": 0.1667, "lr": 1.3884472674819246e-05, "epoch": 3.188354430379747, "percentage": 63.9, "elapsed_time": "18:25:29", "remaining_time": "10:24:28"}
1664
+ {"current_steps": 787, "total_steps": 1230, "loss": 0.1688, "lr": 1.3830457248950864e-05, "epoch": 3.1924050632911394, "percentage": 63.98, "elapsed_time": "18:26:43", "remaining_time": "10:22:58"}
1665
+ {"current_steps": 878, "total_steps": 1230, "loss": 0.1766, "lr": 9.176333983521291e-06, "epoch": 3.561012658227848, "percentage": 71.38, "elapsed_time": "20:38:01", "remaining_time": "8:16:20"}
1666
+ {"current_steps": 788, "total_steps": 1230, "loss": 0.1879, "lr": 1.377649151173163e-05, "epoch": 3.1964556962025314, "percentage": 64.07, "elapsed_time": "18:28:14", "remaining_time": "10:21:37"}
1667
+ {"current_steps": 879, "total_steps": 1230, "loss": 0.1719, "lr": 9.12864899997558e-06, "epoch": 3.5650632911392406, "percentage": 71.46, "elapsed_time": "20:39:26", "remaining_time": "8:14:55"}
1668
+ {"current_steps": 880, "total_steps": 1230, "loss": 0.1788, "lr": 9.08105157279628e-06, "epoch": 3.569113924050633, "percentage": 71.54, "elapsed_time": "20:40:54", "remaining_time": "8:13:32"}
1669
+ {"current_steps": 789, "total_steps": 1230, "loss": 0.1934, "lr": 1.3722575897794181e-05, "epoch": 3.200506329113924, "percentage": 64.15, "elapsed_time": "18:29:50", "remaining_time": "10:20:19"}
1670
+ {"current_steps": 881, "total_steps": 1230, "loss": 0.1871, "lr": 9.03354208532653e-06, "epoch": 3.573164556962025, "percentage": 71.63, "elapsed_time": "20:42:12", "remaining_time": "8:12:05"}
1671
+ {"current_steps": 790, "total_steps": 1230, "loss": 0.1804, "lr": 1.3668710841367472e-05, "epoch": 3.2045569620253165, "percentage": 64.23, "elapsed_time": "18:31:23", "remaining_time": "10:19:00"}
1672
+ {"current_steps": 882, "total_steps": 1230, "loss": 0.1777, "lr": 8.986120920201205e-06, "epoch": 3.5772151898734177, "percentage": 71.71, "elapsed_time": "20:43:32", "remaining_time": "8:10:38"}
1673
+ {"current_steps": 791, "total_steps": 1230, "loss": 0.1615, "lr": 1.361489677627324e-05, "epoch": 3.208607594936709, "percentage": 64.31, "elapsed_time": "18:32:34", "remaining_time": "10:17:28"}
1674
+ {"current_steps": 883, "total_steps": 1230, "loss": 0.1474, "lr": 8.938788459343852e-06, "epoch": 3.58126582278481, "percentage": 71.79, "elapsed_time": "20:44:41", "remaining_time": "8:09:08"}
1675
+ {"current_steps": 792, "total_steps": 1230, "loss": 0.1696, "lr": 1.3561134135922585e-05, "epoch": 3.212658227848101, "percentage": 64.39, "elapsed_time": "18:33:50", "remaining_time": "10:15:59"}
1676
+ {"current_steps": 884, "total_steps": 1230, "loss": 0.1881, "lr": 8.8915450839636e-06, "epoch": 3.5853164556962023, "percentage": 71.87, "elapsed_time": "20:46:16", "remaining_time": "8:07:47"}
1677
+ {"current_steps": 793, "total_steps": 1230, "loss": 0.1862, "lr": 1.350742335331241e-05, "epoch": 3.2167088607594936, "percentage": 64.47, "elapsed_time": "18:35:16", "remaining_time": "10:14:35"}
1678
+ {"current_steps": 885, "total_steps": 1230, "loss": 0.1664, "lr": 8.844391174552116e-06, "epoch": 3.589367088607595, "percentage": 71.95, "elapsed_time": "20:47:34", "remaining_time": "8:06:20"}
1679
+ {"current_steps": 794, "total_steps": 1230, "loss": 0.1787, "lr": 1.345376486102198e-05, "epoch": 3.220759493670886, "percentage": 64.55, "elapsed_time": "18:36:29", "remaining_time": "10:13:05"}
1680
+ {"current_steps": 795, "total_steps": 1230, "loss": 0.1762, "lr": 1.3400159091209414e-05, "epoch": 3.2248101265822786, "percentage": 64.63, "elapsed_time": "18:37:48", "remaining_time": "10:11:37"}
1681
+ {"current_steps": 886, "total_steps": 1230, "loss": 0.169, "lr": 8.797327110880479e-06, "epoch": 3.5934177215189873, "percentage": 72.03, "elapsed_time": "20:48:58", "remaining_time": "8:04:55"}
1682
+ {"current_steps": 887, "total_steps": 1230, "loss": 0.1641, "lr": 8.750353271996206e-06, "epoch": 3.59746835443038, "percentage": 72.11, "elapsed_time": "20:50:13", "remaining_time": "8:03:27"}
1683
+ {"current_steps": 796, "total_steps": 1230, "loss": 0.1786, "lr": 1.3346606475608216e-05, "epoch": 3.2288607594936707, "percentage": 64.72, "elapsed_time": "18:39:11", "remaining_time": "10:10:13"}
1684
+ {"current_steps": 888, "total_steps": 1230, "loss": 0.1551, "lr": 8.703470036220132e-06, "epoch": 3.6015189873417723, "percentage": 72.2, "elapsed_time": "20:51:31", "remaining_time": "8:02:00"}
1685
+ {"current_steps": 797, "total_steps": 1230, "loss": 0.1699, "lr": 1.3293107445523781e-05, "epoch": 3.232911392405063, "percentage": 64.8, "elapsed_time": "18:40:26", "remaining_time": "10:08:43"}
1686
+ {"current_steps": 889, "total_steps": 1230, "loss": 0.1622, "lr": 8.656677781143394e-06, "epoch": 3.6055696202531644, "percentage": 72.28, "elapsed_time": "20:52:58", "remaining_time": "8:00:36"}
1687
+ {"current_steps": 798, "total_steps": 1230, "loss": 0.1659, "lr": 1.3239662431829949e-05, "epoch": 3.2369620253164557, "percentage": 64.88, "elapsed_time": "18:41:57", "remaining_time": "10:07:22"}
1688
+ {"current_steps": 890, "total_steps": 1230, "loss": 0.1882, "lr": 8.609976883624377e-06, "epoch": 3.609620253164557, "percentage": 72.36, "elapsed_time": "20:54:21", "remaining_time": "7:59:11"}
1689
+ {"current_steps": 799, "total_steps": 1230, "loss": 0.1926, "lr": 1.3186271864965509e-05, "epoch": 3.2410126582278482, "percentage": 64.96, "elapsed_time": "18:43:32", "remaining_time": "10:06:03"}
1690
+ {"current_steps": 891, "total_steps": 1230, "loss": 0.1496, "lr": 8.563367719785698e-06, "epoch": 3.6136708860759494, "percentage": 72.44, "elapsed_time": "20:55:39", "remaining_time": "7:57:44"}
1691
+ {"current_steps": 800, "total_steps": 1230, "loss": 0.1883, "lr": 1.3132936174930756e-05, "epoch": 3.2450632911392403, "percentage": 65.04, "elapsed_time": "18:45:01", "remaining_time": "10:04:42"}
1692
+ {"current_steps": 892, "total_steps": 1230, "loss": 0.1776, "lr": 8.516850665011138e-06, "epoch": 3.6177215189873415, "percentage": 72.52, "elapsed_time": "20:56:51", "remaining_time": "7:56:15"}
1693
+ {"current_steps": 801, "total_steps": 1230, "loss": 0.178, "lr": 1.3079655791283995e-05, "epoch": 3.249113924050633, "percentage": 65.12, "elapsed_time": "18:46:27", "remaining_time": "10:03:18"}
1694
+ {"current_steps": 893, "total_steps": 1230, "loss": 0.1634, "lr": 8.47042609394269e-06, "epoch": 3.621772151898734, "percentage": 72.6, "elapsed_time": "20:58:21", "remaining_time": "7:54:52"}
1695
+ {"current_steps": 802, "total_steps": 1230, "loss": 0.1684, "lr": 1.3026431143138108e-05, "epoch": 3.2531645569620253, "percentage": 65.2, "elapsed_time": "18:47:56", "remaining_time": "10:01:56"}
1696
+ {"current_steps": 894, "total_steps": 1230, "loss": 0.1752, "lr": 8.424094380477432e-06, "epoch": 3.6258227848101265, "percentage": 72.68, "elapsed_time": "20:59:43", "remaining_time": "7:53:27"}
1697
+ {"current_steps": 803, "total_steps": 1230, "loss": 0.1947, "lr": 1.2973262659157114e-05, "epoch": 3.257215189873418, "percentage": 65.28, "elapsed_time": "18:49:26", "remaining_time": "10:00:35"}
1698
+ {"current_steps": 895, "total_steps": 1230, "loss": 0.1848, "lr": 8.37785589776465e-06, "epoch": 3.629873417721519, "percentage": 72.76, "elapsed_time": "21:01:08", "remaining_time": "7:52:02"}
1699
+ {"current_steps": 804, "total_steps": 1230, "loss": 0.1816, "lr": 1.2920150767552651e-05, "epoch": 3.26126582278481, "percentage": 65.37, "elapsed_time": "18:50:38", "remaining_time": "9:59:04"}
1700
+ {"current_steps": 896, "total_steps": 1230, "loss": 0.1951, "lr": 8.331711018202694e-06, "epoch": 3.6339240506329116, "percentage": 72.85, "elapsed_time": "21:02:30", "remaining_time": "7:50:37"}
1701
+ {"current_steps": 805, "total_steps": 1230, "loss": 0.16, "lr": 1.2867095896080607e-05, "epoch": 3.2653164556962024, "percentage": 65.45, "elapsed_time": "18:51:52", "remaining_time": "9:57:34"}
1702
+ {"current_steps": 897, "total_steps": 1230, "loss": 0.1563, "lr": 8.285660113436104e-06, "epoch": 3.6379746835443036, "percentage": 72.93, "elapsed_time": "21:03:42", "remaining_time": "7:49:07"}
1703
+ {"current_steps": 806, "total_steps": 1230, "loss": 0.1692, "lr": 1.2814098472037612e-05, "epoch": 3.269367088607595, "percentage": 65.53, "elapsed_time": "18:53:18", "remaining_time": "9:56:11"}
1704
+ {"current_steps": 898, "total_steps": 1230, "loss": 0.1749, "lr": 8.239703554352527e-06, "epoch": 3.642025316455696, "percentage": 73.01, "elapsed_time": "21:05:09", "remaining_time": "7:47:44"}
1705
+ {"current_steps": 807, "total_steps": 1230, "loss": 0.1694, "lr": 1.276115892225764e-05, "epoch": 3.2734177215189875, "percentage": 65.61, "elapsed_time": "18:54:31", "remaining_time": "9:54:40"}
1706
+ {"current_steps": 899, "total_steps": 1230, "loss": 0.1636, "lr": 8.193841711079775e-06, "epoch": 3.6460759493670887, "percentage": 73.09, "elapsed_time": "21:06:25", "remaining_time": "7:46:16"}
1707
+ {"current_steps": 808, "total_steps": 1230, "loss": 0.1739, "lr": 1.2708277673108555e-05, "epoch": 3.27746835443038, "percentage": 65.69, "elapsed_time": "18:55:52", "remaining_time": "9:53:14"}
1708
+ {"current_steps": 900, "total_steps": 1230, "loss": 0.1701, "lr": 8.148074952982828e-06, "epoch": 3.6501265822784807, "percentage": 73.17, "elapsed_time": "21:07:48", "remaining_time": "7:44:51"}
1709
+ {"current_steps": 809, "total_steps": 1230, "loss": 0.1858, "lr": 1.2655455150488649e-05, "epoch": 3.281518987341772, "percentage": 65.77, "elapsed_time": "18:57:24", "remaining_time": "9:51:53"}
1710
+ {"current_steps": 901, "total_steps": 1230, "loss": 0.1709, "lr": 8.102403648660859e-06, "epoch": 3.6541772151898733, "percentage": 73.25, "elapsed_time": "21:09:08", "remaining_time": "7:43:25"}
1711
+ {"current_steps": 810, "total_steps": 1230, "loss": 0.169, "lr": 1.2602691779823272e-05, "epoch": 3.2855696202531646, "percentage": 65.85, "elapsed_time": "18:58:42", "remaining_time": "9:50:26"}
1712
+ {"current_steps": 902, "total_steps": 1230, "loss": 0.1599, "lr": 8.056828165944282e-06, "epoch": 3.6582278481012658, "percentage": 73.33, "elapsed_time": "21:10:27", "remaining_time": "7:41:59"}
1713
+ {"current_steps": 811, "total_steps": 1230, "loss": 0.1734, "lr": 1.2549987986061355e-05, "epoch": 3.289620253164557, "percentage": 65.93, "elapsed_time": "19:00:07", "remaining_time": "9:49:02"}
1714
+ {"current_steps": 903, "total_steps": 1230, "loss": 0.1786, "lr": 8.011348871891762e-06, "epoch": 3.6622784810126583, "percentage": 73.41, "elapsed_time": "21:11:51", "remaining_time": "7:40:34"}
1715
+ {"current_steps": 812, "total_steps": 1230, "loss": 0.1819, "lr": 1.2497344193672005e-05, "epoch": 3.293670886075949, "percentage": 66.02, "elapsed_time": "19:01:35", "remaining_time": "9:47:40"}
1716
+ {"current_steps": 904, "total_steps": 1230, "loss": 0.1706, "lr": 7.965966132787287e-06, "epoch": 3.666329113924051, "percentage": 73.5, "elapsed_time": "21:13:19", "remaining_time": "7:39:11"}
1717
+ {"current_steps": 813, "total_steps": 1230, "loss": 0.1598, "lr": 1.2444760826641092e-05, "epoch": 3.2977215189873417, "percentage": 66.1, "elapsed_time": "19:02:55", "remaining_time": "9:46:13"}
1718
+ {"current_steps": 905, "total_steps": 1230, "loss": 0.1606, "lr": 7.920680314137189e-06, "epoch": 3.670379746835443, "percentage": 73.58, "elapsed_time": "21:14:24", "remaining_time": "7:37:39"}
1719
+ {"current_steps": 814, "total_steps": 1230, "loss": 0.1586, "lr": 1.2392238308467817e-05, "epoch": 3.301772151898734, "percentage": 66.18, "elapsed_time": "19:04:14", "remaining_time": "9:44:46"}
1720
+ {"current_steps": 906, "total_steps": 1230, "loss": 0.1481, "lr": 7.875491780667246e-06, "epoch": 3.6744303797468354, "percentage": 73.66, "elapsed_time": "21:15:39", "remaining_time": "7:36:11"}
1721
+ {"current_steps": 815, "total_steps": 1230, "loss": 0.1889, "lr": 1.2339777062161326e-05, "epoch": 3.3058227848101267, "percentage": 66.26, "elapsed_time": "19:05:37", "remaining_time": "9:43:21"}
1722
+ {"current_steps": 907, "total_steps": 1230, "loss": 0.169, "lr": 7.830400896319667e-06, "epoch": 3.678481012658228, "percentage": 73.74, "elapsed_time": "21:16:55", "remaining_time": "7:34:44"}
1723
+ {"current_steps": 908, "total_steps": 1230, "loss": 0.1594, "lr": 7.785408024250259e-06, "epoch": 3.68253164556962, "percentage": 73.82, "elapsed_time": "21:18:16", "remaining_time": "7:33:18"}
1724
+ {"current_steps": 816, "total_steps": 1230, "loss": 0.1888, "lr": 1.2287377510237293e-05, "epoch": 3.309873417721519, "percentage": 66.34, "elapsed_time": "19:07:10", "remaining_time": "9:42:01"}
1725
+ {"current_steps": 817, "total_steps": 1230, "loss": 0.1786, "lr": 1.2235040074714488e-05, "epoch": 3.3139240506329113, "percentage": 66.42, "elapsed_time": "19:08:34", "remaining_time": "9:40:36"}
1726
+ {"current_steps": 909, "total_steps": 1230, "loss": 0.1807, "lr": 7.74051352682542e-06, "epoch": 3.6865822784810125, "percentage": 73.9, "elapsed_time": "21:19:43", "remaining_time": "7:31:54"}
1727
+ {"current_steps": 910, "total_steps": 1230, "loss": 0.1567, "lr": 7.695717765619257e-06, "epoch": 3.690632911392405, "percentage": 73.98, "elapsed_time": "21:20:49", "remaining_time": "7:30:24"}
1728
+ {"current_steps": 818, "total_steps": 1230, "loss": 0.1875, "lr": 1.2182765177111434e-05, "epoch": 3.317974683544304, "percentage": 66.5, "elapsed_time": "19:10:12", "remaining_time": "9:39:19"}
1729
+ {"current_steps": 911, "total_steps": 1230, "loss": 0.1591, "lr": 7.651021101410673e-06, "epoch": 3.6946835443037975, "percentage": 74.07, "elapsed_time": "21:22:00", "remaining_time": "7:28:55"}
1730
+ {"current_steps": 819, "total_steps": 1230, "loss": 0.1594, "lr": 1.213055323844297e-05, "epoch": 3.3220253164556963, "percentage": 66.59, "elapsed_time": "19:11:30", "remaining_time": "9:37:51"}
1731
+ {"current_steps": 912, "total_steps": 1230, "loss": 0.1714, "lr": 7.606423894180464e-06, "epoch": 3.69873417721519, "percentage": 74.15, "elapsed_time": "21:23:18", "remaining_time": "7:27:28"}
1732
+ {"current_steps": 820, "total_steps": 1230, "loss": 0.1597, "lr": 1.2078404679216864e-05, "epoch": 3.3260759493670884, "percentage": 66.67, "elapsed_time": "19:12:46", "remaining_time": "9:36:23"}
1733
+ {"current_steps": 913, "total_steps": 1230, "loss": 0.1832, "lr": 7.56192650310839e-06, "epoch": 3.702784810126582, "percentage": 74.23, "elapsed_time": "21:24:41", "remaining_time": "7:26:03"}
1734
+ {"current_steps": 821, "total_steps": 1230, "loss": 0.1737, "lr": 1.2026319919430458e-05, "epoch": 3.330126582278481, "percentage": 66.75, "elapsed_time": "19:14:14", "remaining_time": "9:35:00"}
1735
+ {"current_steps": 914, "total_steps": 1230, "loss": 0.1815, "lr": 7.517529286570349e-06, "epoch": 3.7068354430379746, "percentage": 74.31, "elapsed_time": "21:26:04", "remaining_time": "7:24:38"}
1736
+ {"current_steps": 822, "total_steps": 1230, "loss": 0.1667, "lr": 1.1974299378567227e-05, "epoch": 3.3341772151898734, "percentage": 66.83, "elapsed_time": "19:15:42", "remaining_time": "9:33:38"}
1737
+ {"current_steps": 915, "total_steps": 1230, "loss": 0.1905, "lr": 7.473232602135387e-06, "epoch": 3.710886075949367, "percentage": 74.39, "elapsed_time": "21:27:42", "remaining_time": "7:23:18"}
1738
+ {"current_steps": 823, "total_steps": 1230, "loss": 0.1797, "lr": 1.1922343475593462e-05, "epoch": 3.338227848101266, "percentage": 66.91, "elapsed_time": "19:17:09", "remaining_time": "9:32:15"}
1739
+ {"current_steps": 916, "total_steps": 1230, "loss": 0.17, "lr": 7.429036806562935e-06, "epoch": 3.714936708860759, "percentage": 74.47, "elapsed_time": "21:29:16", "remaining_time": "7:21:57"}
1740
+ {"current_steps": 824, "total_steps": 1230, "loss": 0.2137, "lr": 1.187045262895488e-05, "epoch": 3.3422784810126585, "percentage": 66.99, "elapsed_time": "19:18:46", "remaining_time": "9:30:56"}
1741
+ {"current_steps": 917, "total_steps": 1230, "loss": 0.1688, "lr": 7.3849422557998455e-06, "epoch": 3.7189873417721517, "percentage": 74.55, "elapsed_time": "21:30:27", "remaining_time": "7:20:28"}
1742
+ {"current_steps": 825, "total_steps": 1230, "loss": 0.159, "lr": 1.1818627256573203e-05, "epoch": 3.3463291139240505, "percentage": 67.07, "elapsed_time": "19:19:51", "remaining_time": "9:29:22"}
1743
+ {"current_steps": 918, "total_steps": 1230, "loss": 0.1775, "lr": 7.340949304977567e-06, "epoch": 3.7230379746835442, "percentage": 74.63, "elapsed_time": "21:31:45", "remaining_time": "7:19:01"}
1744
+ {"current_steps": 826, "total_steps": 1230, "loss": 0.1886, "lr": 1.1766867775842864e-05, "epoch": 3.350379746835443, "percentage": 67.15, "elapsed_time": "19:21:23", "remaining_time": "9:28:02"}
1745
+ {"current_steps": 919, "total_steps": 1230, "loss": 0.1628, "lr": 7.297058308409282e-06, "epoch": 3.7270886075949368, "percentage": 74.72, "elapsed_time": "21:33:09", "remaining_time": "7:17:37"}
1746
+ {"current_steps": 827, "total_steps": 1230, "loss": 0.1743, "lr": 1.1715174603627615e-05, "epoch": 3.3544303797468356, "percentage": 67.24, "elapsed_time": "19:22:35", "remaining_time": "9:26:32"}
1747
+ {"current_steps": 920, "total_steps": 1230, "loss": 0.1914, "lr": 7.25326961958704e-06, "epoch": 3.7311392405063293, "percentage": 74.8, "elapsed_time": "21:34:34", "remaining_time": "7:16:13"}
1748
+ {"current_steps": 828, "total_steps": 1230, "loss": 0.1767, "lr": 1.1663548156257147e-05, "epoch": 3.3584810126582276, "percentage": 67.32, "elapsed_time": "19:23:53", "remaining_time": "9:25:04"}
1749
+ {"current_steps": 921, "total_steps": 1230, "loss": 0.166, "lr": 7.209583591178921e-06, "epoch": 3.7351898734177214, "percentage": 74.88, "elapsed_time": "21:36:00", "remaining_time": "7:14:49"}
1750
+ {"current_steps": 829, "total_steps": 1230, "loss": 0.1601, "lr": 1.161198884952377e-05, "epoch": 3.36253164556962, "percentage": 67.4, "elapsed_time": "19:25:07", "remaining_time": "9:23:35"}
1751
+ {"current_steps": 830, "total_steps": 1230, "loss": 0.1752, "lr": 1.1560497098679056e-05, "epoch": 3.3665822784810127, "percentage": 67.48, "elapsed_time": "19:26:26", "remaining_time": "9:22:08"}
1752
+ {"current_steps": 922, "total_steps": 1230, "loss": 0.1784, "lr": 7.1660005750261925e-06, "epoch": 3.739240506329114, "percentage": 74.96, "elapsed_time": "21:37:41", "remaining_time": "7:13:30"}
1753
+ {"current_steps": 923, "total_steps": 1230, "loss": 0.1682, "lr": 7.1225209221404765e-06, "epoch": 3.7432911392405064, "percentage": 75.04, "elapsed_time": "21:38:55", "remaining_time": "7:12:02"}
1754
+ {"current_steps": 831, "total_steps": 1230, "loss": 0.189, "lr": 1.1509073318430479e-05, "epoch": 3.370632911392405, "percentage": 67.56, "elapsed_time": "19:27:55", "remaining_time": "9:20:46"}
1755
+ {"current_steps": 832, "total_steps": 1230, "loss": 0.1541, "lr": 1.1457717922938116e-05, "epoch": 3.3746835443037977, "percentage": 67.64, "elapsed_time": "19:29:07", "remaining_time": "9:19:16"}
1756
+ {"current_steps": 924, "total_steps": 1230, "loss": 0.1762, "lr": 7.079144982700909e-06, "epoch": 3.747341772151899, "percentage": 75.12, "elapsed_time": "21:40:19", "remaining_time": "7:10:37"}
1757
+ {"current_steps": 833, "total_steps": 1230, "loss": 0.164, "lr": 1.1406431325811233e-05, "epoch": 3.3787341772151898, "percentage": 67.72, "elapsed_time": "19:30:34", "remaining_time": "9:17:52"}
1758
+ {"current_steps": 925, "total_steps": 1230, "loss": 0.2018, "lr": 7.0358731060513695e-06, "epoch": 3.7513924050632914, "percentage": 75.2, "elapsed_time": "21:41:53", "remaining_time": "7:09:16"}
1759
+ {"current_steps": 834, "total_steps": 1230, "loss": 0.1777, "lr": 1.135521394010506e-05, "epoch": 3.3827848101265823, "percentage": 67.8, "elapsed_time": "19:31:53", "remaining_time": "9:16:26"}
1760
+ {"current_steps": 926, "total_steps": 1230, "loss": 0.1639, "lr": 6.99270564069757e-06, "epoch": 3.7554430379746835, "percentage": 75.28, "elapsed_time": "21:43:19", "remaining_time": "7:07:52"}
1761
+ {"current_steps": 835, "total_steps": 1230, "loss": 0.1822, "lr": 1.1304066178317367e-05, "epoch": 3.386835443037975, "percentage": 67.89, "elapsed_time": "19:33:17", "remaining_time": "9:15:01"}
1762
+ {"current_steps": 927, "total_steps": 1230, "loss": 0.1666, "lr": 6.949642934304375e-06, "epoch": 3.759493670886076, "percentage": 75.37, "elapsed_time": "21:44:36", "remaining_time": "7:06:25"}
1763
+ {"current_steps": 836, "total_steps": 1230, "loss": 0.1836, "lr": 1.1252988452385199e-05, "epoch": 3.390886075949367, "percentage": 67.97, "elapsed_time": "19:34:43", "remaining_time": "9:13:38"}
1764
+ {"current_steps": 928, "total_steps": 1230, "loss": 0.1902, "lr": 6.906685333692871e-06, "epoch": 3.7635443037974685, "percentage": 75.45, "elapsed_time": "21:46:08", "remaining_time": "7:05:03"}
1765
+ {"current_steps": 837, "total_steps": 1230, "loss": 0.1734, "lr": 1.1201981173681536e-05, "epoch": 3.3949367088607594, "percentage": 68.05, "elapsed_time": "19:36:12", "remaining_time": "9:12:15"}
1766
+ {"current_steps": 929, "total_steps": 1230, "loss": 0.1668, "lr": 6.86383318483769e-06, "epoch": 3.7675949367088606, "percentage": 75.53, "elapsed_time": "21:47:26", "remaining_time": "7:03:36"}
1767
+ {"current_steps": 838, "total_steps": 1230, "loss": 0.1803, "lr": 1.1151044753011991e-05, "epoch": 3.398987341772152, "percentage": 68.13, "elapsed_time": "19:37:35", "remaining_time": "9:10:51"}
1768
+ {"current_steps": 930, "total_steps": 1230, "loss": 0.1742, "lr": 6.821086832864139e-06, "epoch": 3.771645569620253, "percentage": 75.61, "elapsed_time": "21:48:48", "remaining_time": "7:02:11"}
1769
+ {"current_steps": 931, "total_steps": 1230, "loss": 0.1736, "lr": 6.77844662204546e-06, "epoch": 3.7756962025316456, "percentage": 75.69, "elapsed_time": "21:50:03", "remaining_time": "7:00:44"}
1770
+ {"current_steps": 839, "total_steps": 1230, "loss": 0.1956, "lr": 1.1100179600611491e-05, "epoch": 3.4030379746835444, "percentage": 68.21, "elapsed_time": "19:39:12", "remaining_time": "9:09:32"}
1771
+ {"current_steps": 840, "total_steps": 1230, "loss": 0.1736, "lr": 1.1049386126140985e-05, "epoch": 3.407088607594937, "percentage": 68.29, "elapsed_time": "19:40:35", "remaining_time": "9:08:08"}
1772
+ {"current_steps": 932, "total_steps": 1230, "loss": 0.1908, "lr": 6.7359128958000455e-06, "epoch": 3.779746835443038, "percentage": 75.77, "elapsed_time": "21:52:01", "remaining_time": "6:59:30"}
1773
+ {"current_steps": 933, "total_steps": 1230, "loss": 0.1592, "lr": 6.693485996688695e-06, "epoch": 3.7837974683544306, "percentage": 75.85, "elapsed_time": "21:53:14", "remaining_time": "6:58:02"}
1774
+ {"current_steps": 841, "total_steps": 1230, "loss": 0.1567, "lr": 1.0998664738684128e-05, "epoch": 3.411139240506329, "percentage": 68.37, "elapsed_time": "19:42:06", "remaining_time": "9:06:46"}
1775
+ {"current_steps": 934, "total_steps": 1230, "loss": 0.1642, "lr": 6.651166266411801e-06, "epoch": 3.7878481012658227, "percentage": 75.93, "elapsed_time": "21:54:33", "remaining_time": "6:56:36"}
1776
+ {"current_steps": 842, "total_steps": 1230, "loss": 0.1769, "lr": 1.0948015846744e-05, "epoch": 3.4151898734177215, "percentage": 68.46, "elapsed_time": "19:43:26", "remaining_time": "9:05:20"}
1777
+ {"current_steps": 843, "total_steps": 1230, "loss": 0.1858, "lr": 1.0897439858239832e-05, "epoch": 3.419240506329114, "percentage": 68.54, "elapsed_time": "19:44:50", "remaining_time": "9:03:55"}
1778
+ {"current_steps": 935, "total_steps": 1230, "loss": 0.1781, "lr": 6.6089540458066725e-06, "epoch": 3.7918987341772152, "percentage": 76.02, "elapsed_time": "21:56:02", "remaining_time": "6:55:13"}
1779
+ {"current_steps": 936, "total_steps": 1230, "loss": 0.1713, "lr": 6.566849674844711e-06, "epoch": 3.7959493670886078, "percentage": 76.1, "elapsed_time": "21:57:19", "remaining_time": "6:53:46"}
1780
+ {"current_steps": 844, "total_steps": 1230, "loss": 0.1891, "lr": 1.0846937180503652e-05, "epoch": 3.423291139240506, "percentage": 68.62, "elapsed_time": "19:46:28", "remaining_time": "9:02:37"}
1781
+ {"current_steps": 937, "total_steps": 1230, "loss": 0.1837, "lr": 6.524853492628747e-06, "epoch": 3.8, "percentage": 76.18, "elapsed_time": "21:58:54", "remaining_time": "6:52:25"}
1782
+ {"current_steps": 845, "total_steps": 1230, "loss": 0.169, "lr": 1.0796508220277117e-05, "epoch": 3.4273417721518986, "percentage": 68.7, "elapsed_time": "19:47:59", "remaining_time": "9:01:16"}
1783
+ {"current_steps": 938, "total_steps": 1230, "loss": 0.1556, "lr": 6.4829658373902536e-06, "epoch": 3.8040506329113923, "percentage": 76.26, "elapsed_time": "22:00:21", "remaining_time": "6:51:01"}
1784
+ {"current_steps": 846, "total_steps": 1230, "loss": 0.1707, "lr": 1.0746153383708107e-05, "epoch": 3.431392405063291, "percentage": 68.78, "elapsed_time": "19:49:17", "remaining_time": "8:59:49"}
1785
+ {"current_steps": 939, "total_steps": 1230, "loss": 0.1708, "lr": 6.441187046486648e-06, "epoch": 3.808101265822785, "percentage": 76.34, "elapsed_time": "22:01:41", "remaining_time": "6:49:35"}
1786
+ {"current_steps": 847, "total_steps": 1230, "loss": 0.1704, "lr": 1.0695873076347579e-05, "epoch": 3.4354430379746836, "percentage": 68.86, "elapsed_time": "19:50:42", "remaining_time": "8:58:25"}
1787
+ {"current_steps": 940, "total_steps": 1230, "loss": 0.1632, "lr": 6.399517456398567e-06, "epoch": 3.8121518987341774, "percentage": 76.42, "elapsed_time": "22:02:50", "remaining_time": "6:48:06"}
1788
+ {"current_steps": 848, "total_steps": 1230, "loss": 0.1795, "lr": 1.0645667703146205e-05, "epoch": 3.439493670886076, "percentage": 68.94, "elapsed_time": "19:52:11", "remaining_time": "8:57:03"}
1789
+ {"current_steps": 941, "total_steps": 1230, "loss": 0.1646, "lr": 6.357957402727164e-06, "epoch": 3.81620253164557, "percentage": 76.5, "elapsed_time": "22:04:15", "remaining_time": "6:46:42"}
1790
+ {"current_steps": 849, "total_steps": 1230, "loss": 0.159, "lr": 1.0595537668451161e-05, "epoch": 3.4435443037974682, "percentage": 69.02, "elapsed_time": "19:53:24", "remaining_time": "8:55:33"}
1791
+ {"current_steps": 942, "total_steps": 1230, "loss": 0.159, "lr": 6.316507220191395e-06, "epoch": 3.820253164556962, "percentage": 76.59, "elapsed_time": "22:05:32", "remaining_time": "6:45:15"}
1792
+ {"current_steps": 850, "total_steps": 1230, "loss": 0.1581, "lr": 1.0545483376002854e-05, "epoch": 3.4475949367088607, "percentage": 69.11, "elapsed_time": "19:54:28", "remaining_time": "8:54:00"}
1793
+ {"current_steps": 943, "total_steps": 1230, "loss": 0.1611, "lr": 6.275167242625331e-06, "epoch": 3.8243037974683545, "percentage": 76.67, "elapsed_time": "22:06:42", "remaining_time": "6:43:46"}
1794
+ {"current_steps": 851, "total_steps": 1230, "loss": 0.1741, "lr": 1.0495505228931676e-05, "epoch": 3.4516455696202533, "percentage": 69.19, "elapsed_time": "19:55:46", "remaining_time": "8:52:32"}
1795
+ {"current_steps": 944, "total_steps": 1230, "loss": 0.1592, "lr": 6.233937802975471e-06, "epoch": 3.828354430379747, "percentage": 76.75, "elapsed_time": "22:07:58", "remaining_time": "6:42:19"}
1796
+ {"current_steps": 852, "total_steps": 1230, "loss": 0.1674, "lr": 1.044560362975474e-05, "epoch": 3.4556962025316453, "percentage": 69.27, "elapsed_time": "19:57:12", "remaining_time": "8:51:09"}
1797
+ {"current_steps": 945, "total_steps": 1230, "loss": 0.1785, "lr": 6.192819233298046e-06, "epoch": 3.832405063291139, "percentage": 76.83, "elapsed_time": "22:09:19", "remaining_time": "6:40:54"}
1798
+ {"current_steps": 853, "total_steps": 1230, "loss": 0.1591, "lr": 1.0395778980372695e-05, "epoch": 3.459746835443038, "percentage": 69.35, "elapsed_time": "19:58:21", "remaining_time": "8:49:38"}
1799
+ {"current_steps": 854, "total_steps": 1230, "loss": 0.1766, "lr": 1.0346031682066381e-05, "epoch": 3.4637974683544304, "percentage": 69.43, "elapsed_time": "19:59:33", "remaining_time": "8:48:08"}
1800
+ {"current_steps": 946, "total_steps": 1230, "loss": 0.1678, "lr": 6.151811864756383e-06, "epoch": 3.8364556962025316, "percentage": 76.91, "elapsed_time": "22:10:44", "remaining_time": "6:39:30"}
1801
+ {"current_steps": 947, "total_steps": 1230, "loss": 0.1626, "lr": 6.1109160276181655e-06, "epoch": 3.840506329113924, "percentage": 76.99, "elapsed_time": "22:12:08", "remaining_time": "6:38:05"}
1802
+ {"current_steps": 855, "total_steps": 1230, "loss": 0.1738, "lr": 1.0296362135493724e-05, "epoch": 3.467848101265823, "percentage": 69.51, "elapsed_time": "20:00:59", "remaining_time": "8:46:45"}
1803
+ {"current_steps": 856, "total_steps": 1230, "loss": 0.1611, "lr": 1.0246770740686422e-05, "epoch": 3.4718987341772154, "percentage": 69.59, "elapsed_time": "20:02:11", "remaining_time": "8:45:15"}
1804
+ {"current_steps": 948, "total_steps": 1230, "loss": 0.1746, "lr": 6.070132051252868e-06, "epoch": 3.8445569620253166, "percentage": 77.07, "elapsed_time": "22:13:39", "remaining_time": "6:36:43"}
1805
+ {"current_steps": 857, "total_steps": 1230, "loss": 0.1738, "lr": 1.0197257897046743e-05, "epoch": 3.4759493670886075, "percentage": 69.67, "elapsed_time": "20:03:43", "remaining_time": "8:43:54"}
1806
+ {"current_steps": 949, "total_steps": 1230, "loss": 0.173, "lr": 6.0294602641290034e-06, "epoch": 3.848607594936709, "percentage": 77.15, "elapsed_time": "22:15:04", "remaining_time": "6:35:19"}
1807
+ {"current_steps": 858, "total_steps": 1230, "loss": 0.1863, "lr": 1.014782400334433e-05, "epoch": 3.48, "percentage": 69.76, "elapsed_time": "20:04:54", "remaining_time": "8:42:24"}
1808
+ {"current_steps": 950, "total_steps": 1230, "loss": 0.1637, "lr": 5.988900993811575e-06, "epoch": 3.852658227848101, "percentage": 77.24, "elapsed_time": "22:16:23", "remaining_time": "6:33:53"}
1809
+ {"current_steps": 859, "total_steps": 1230, "loss": 0.1594, "lr": 1.009846945771296e-05, "epoch": 3.4840506329113925, "percentage": 69.84, "elapsed_time": "20:06:03", "remaining_time": "8:40:53"}
1810
+ {"current_steps": 951, "total_steps": 1230, "loss": 0.1544, "lr": 5.948454566959363e-06, "epoch": 3.8567088607594937, "percentage": 77.32, "elapsed_time": "22:17:40", "remaining_time": "6:32:26"}
1811
+ {"current_steps": 860, "total_steps": 1230, "loss": 0.1543, "lr": 1.0049194657647363e-05, "epoch": 3.4881012658227846, "percentage": 69.92, "elapsed_time": "20:07:12", "remaining_time": "8:39:22"}
1812
+ {"current_steps": 952, "total_steps": 1230, "loss": 0.1743, "lr": 5.908121309322328e-06, "epoch": 3.8607594936708862, "percentage": 77.4, "elapsed_time": "22:19:11", "remaining_time": "6:31:04"}
1813
+ {"current_steps": 861, "total_steps": 1230, "loss": 0.1753, "lr": 1.0000000000000006e-05, "epoch": 3.492151898734177, "percentage": 70.0, "elapsed_time": "20:08:29", "remaining_time": "8:37:55"}
1814
+ {"current_steps": 953, "total_steps": 1230, "loss": 0.16, "lr": 5.867901545738976e-06, "epoch": 3.8648101265822783, "percentage": 77.48, "elapsed_time": "22:20:30", "remaining_time": "6:29:38"}
1815
+ {"current_steps": 862, "total_steps": 1230, "loss": 0.1548, "lr": 9.950885880977891e-06, "epoch": 3.4962025316455696, "percentage": 70.08, "elapsed_time": "20:09:44", "remaining_time": "8:36:27"}
1816
+ {"current_steps": 954, "total_steps": 1230, "loss": 0.161, "lr": 5.827795600133774e-06, "epoch": 3.868860759493671, "percentage": 77.56, "elapsed_time": "22:21:47", "remaining_time": "6:28:11"}
1817
+ {"current_steps": 863, "total_steps": 1230, "loss": 0.195, "lr": 9.901852696139382e-06, "epoch": 3.500253164556962, "percentage": 70.16, "elapsed_time": "20:11:19", "remaining_time": "8:35:07"}
1818
+ {"current_steps": 955, "total_steps": 1230, "loss": 0.1924, "lr": 5.787803795514466e-06, "epoch": 3.8729113924050633, "percentage": 77.64, "elapsed_time": "22:23:17", "remaining_time": "6:26:48"}
1819
+ {"current_steps": 864, "total_steps": 1230, "loss": 0.1997, "lr": 9.852900840391027e-06, "epoch": 3.5043037974683546, "percentage": 70.24, "elapsed_time": "20:12:53", "remaining_time": "8:33:47"}
1820
+ {"current_steps": 956, "total_steps": 1230, "loss": 0.1617, "lr": 5.747926453969576e-06, "epoch": 3.876962025316456, "percentage": 77.72, "elapsed_time": "22:24:38", "remaining_time": "6:25:23"}
1821
+ {"current_steps": 865, "total_steps": 1230, "loss": 0.178, "lr": 9.804030707984313e-06, "epoch": 3.5083544303797467, "percentage": 70.33, "elapsed_time": "20:14:28", "remaining_time": "8:32:28"}
1822
+ {"current_steps": 957, "total_steps": 1230, "loss": 0.172, "lr": 5.708163896665708e-06, "epoch": 3.8810126582278484, "percentage": 77.8, "elapsed_time": "22:26:02", "remaining_time": "6:23:58"}
1823
+ {"current_steps": 866, "total_steps": 1230, "loss": 0.1699, "lr": 9.755242692512599e-06, "epoch": 3.512405063291139, "percentage": 70.41, "elapsed_time": "20:15:41", "remaining_time": "8:30:59"}
1824
+ {"current_steps": 958, "total_steps": 1230, "loss": 0.1732, "lr": 5.668516443845047e-06, "epoch": 3.8850632911392404, "percentage": 77.89, "elapsed_time": "22:27:26", "remaining_time": "6:22:34"}
1825
+ {"current_steps": 867, "total_steps": 1230, "loss": 0.1659, "lr": 9.70653718690782e-06, "epoch": 3.5164556962025317, "percentage": 70.49, "elapsed_time": "20:16:59", "remaining_time": "8:29:32"}
1826
+ {"current_steps": 959, "total_steps": 1230, "loss": 0.1855, "lr": 5.6289844148227225e-06, "epoch": 3.889113924050633, "percentage": 77.97, "elapsed_time": "22:28:59", "remaining_time": "6:21:12"}
1827
+ {"current_steps": 868, "total_steps": 1230, "loss": 0.1612, "lr": 9.657914583437454e-06, "epoch": 3.520506329113924, "percentage": 70.57, "elapsed_time": "20:18:22", "remaining_time": "8:28:07"}
1828
+ {"current_steps": 960, "total_steps": 1230, "loss": 0.1918, "lr": 5.5895681279842615e-06, "epoch": 3.8931645569620255, "percentage": 78.05, "elapsed_time": "22:30:34", "remaining_time": "6:19:50"}
1829
+ {"current_steps": 869, "total_steps": 1230, "loss": 0.1713, "lr": 9.609375273701246e-06, "epoch": 3.5245569620253163, "percentage": 70.65, "elapsed_time": "20:19:35", "remaining_time": "8:26:38"}
1830
+ {"current_steps": 961, "total_steps": 1230, "loss": 0.166, "lr": 5.550267900783019e-06, "epoch": 3.8972151898734175, "percentage": 78.13, "elapsed_time": "22:31:45", "remaining_time": "6:18:22"}
1831
+ {"current_steps": 870, "total_steps": 1230, "loss": 0.192, "lr": 9.560919648628133e-06, "epoch": 3.528607594936709, "percentage": 70.73, "elapsed_time": "20:21:04", "remaining_time": "8:25:16"}
1832
+ {"current_steps": 962, "total_steps": 1230, "loss": 0.1804, "lr": 5.511084049737623e-06, "epoch": 3.90126582278481, "percentage": 78.21, "elapsed_time": "22:33:11", "remaining_time": "6:16:58"}
1833
+ {"current_steps": 871, "total_steps": 1230, "loss": 0.1677, "lr": 9.512548098473047e-06, "epoch": 3.5326582278481014, "percentage": 70.81, "elapsed_time": "20:22:22", "remaining_time": "8:23:49"}
1834
+ {"current_steps": 963, "total_steps": 1230, "loss": 0.1705, "lr": 5.4720168904294215e-06, "epoch": 3.9053164556962026, "percentage": 78.29, "elapsed_time": "22:34:32", "remaining_time": "6:15:33"}
1835
+ {"current_steps": 872, "total_steps": 1230, "loss": 0.1783, "lr": 9.464261012813825e-06, "epoch": 3.536708860759494, "percentage": 70.89, "elapsed_time": "20:23:47", "remaining_time": "8:22:25"}
1836
+ {"current_steps": 964, "total_steps": 1230, "loss": 0.1622, "lr": 5.433066737499948e-06, "epoch": 3.909367088607595, "percentage": 78.37, "elapsed_time": "22:35:30", "remaining_time": "6:14:01"}
1837
+ {"current_steps": 873, "total_steps": 1230, "loss": 0.1877, "lr": 9.416058780547987e-06, "epoch": 3.540759493670886, "percentage": 70.98, "elapsed_time": "20:25:27", "remaining_time": "8:21:07"}
1838
+ {"current_steps": 965, "total_steps": 1230, "loss": 0.1617, "lr": 5.394233904648376e-06, "epoch": 3.9134177215189876, "percentage": 78.46, "elapsed_time": "22:36:48", "remaining_time": "6:12:35"}
1839
+ {"current_steps": 874, "total_steps": 1230, "loss": 0.1483, "lr": 9.367941789889714e-06, "epoch": 3.5448101265822785, "percentage": 71.06, "elapsed_time": "20:26:36", "remaining_time": "8:19:37"}
1840
+ {"current_steps": 966, "total_steps": 1230, "loss": 0.164, "lr": 5.355518704628997e-06, "epoch": 3.9174683544303797, "percentage": 78.54, "elapsed_time": "22:38:04", "remaining_time": "6:11:09"}
1841
+ {"current_steps": 875, "total_steps": 1230, "loss": 0.1744, "lr": 9.319910428366607e-06, "epoch": 3.548860759493671, "percentage": 71.14, "elapsed_time": "20:28:06", "remaining_time": "8:18:15"}
1842
+ {"current_steps": 967, "total_steps": 1230, "loss": 0.1749, "lr": 5.316921449248731e-06, "epoch": 3.921518987341772, "percentage": 78.62, "elapsed_time": "22:39:32", "remaining_time": "6:09:45"}
1843
+ {"current_steps": 876, "total_steps": 1230, "loss": 0.1607, "lr": 9.271965082816667e-06, "epoch": 3.552911392405063, "percentage": 71.22, "elapsed_time": "20:29:24", "remaining_time": "8:16:48"}
1844
+ {"current_steps": 968, "total_steps": 1230, "loss": 0.1893, "lr": 5.278442449364538e-06, "epoch": 3.9255696202531647, "percentage": 78.7, "elapsed_time": "22:41:06", "remaining_time": "6:08:24"}
1845
+ {"current_steps": 877, "total_steps": 1230, "loss": 0.1754, "lr": 9.224106139385111e-06, "epoch": 3.5569620253164556, "percentage": 71.3, "elapsed_time": "20:30:50", "remaining_time": "8:15:25"}
1846
+ {"current_steps": 969, "total_steps": 1230, "loss": 0.1724, "lr": 5.240082014881016e-06, "epoch": 3.9296202531645568, "percentage": 78.78, "elapsed_time": "22:42:26", "remaining_time": "6:06:58"}
1847
+ {"current_steps": 878, "total_steps": 1230, "loss": 0.1762, "lr": 9.176333983521291e-06, "epoch": 3.561012658227848, "percentage": 71.38, "elapsed_time": "20:32:22", "remaining_time": "8:14:04"}
1848
+ {"current_steps": 970, "total_steps": 1230, "loss": 0.1746, "lr": 5.201840454747822e-06, "epoch": 3.9336708860759493, "percentage": 78.86, "elapsed_time": "22:43:56", "remaining_time": "6:05:35"}
1849
+ {"current_steps": 879, "total_steps": 1230, "loss": 0.1722, "lr": 9.12864899997558e-06, "epoch": 3.5650632911392406, "percentage": 71.46, "elapsed_time": "20:33:47", "remaining_time": "8:12:40"}
1850
+ {"current_steps": 971, "total_steps": 1230, "loss": 0.1784, "lr": 5.163718076957223e-06, "epoch": 3.937721518987342, "percentage": 78.94, "elapsed_time": "22:45:21", "remaining_time": "6:04:11"}
1851
+ {"current_steps": 880, "total_steps": 1230, "loss": 0.1785, "lr": 9.08105157279628e-06, "epoch": 3.569113924050633, "percentage": 71.54, "elapsed_time": "20:35:15", "remaining_time": "8:11:17"}
1852
+ {"current_steps": 972, "total_steps": 1230, "loss": 0.1891, "lr": 5.125715188541609e-06, "epoch": 3.9417721518987343, "percentage": 79.02, "elapsed_time": "22:46:55", "remaining_time": "6:02:49"}
1853
+ {"current_steps": 881, "total_steps": 1230, "loss": 0.187, "lr": 9.03354208532653e-06, "epoch": 3.573164556962025, "percentage": 71.63, "elapsed_time": "20:36:33", "remaining_time": "8:09:51"}
1854
+ {"current_steps": 973, "total_steps": 1230, "loss": 0.1992, "lr": 5.087832095571021e-06, "epoch": 3.945822784810127, "percentage": 79.11, "elapsed_time": "22:48:40", "remaining_time": "6:01:30"}
1855
+ {"current_steps": 882, "total_steps": 1230, "loss": 0.1775, "lr": 8.986120920201205e-06, "epoch": 3.5772151898734177, "percentage": 71.71, "elapsed_time": "20:37:52", "remaining_time": "8:08:24"}
1856
+ {"current_steps": 883, "total_steps": 1230, "loss": 0.1473, "lr": 8.938788459343852e-06, "epoch": 3.58126582278481, "percentage": 71.79, "elapsed_time": "20:39:01", "remaining_time": "8:06:54"}
1857
+ {"current_steps": 974, "total_steps": 1230, "loss": 0.1858, "lr": 5.0500691031506766e-06, "epoch": 3.949873417721519, "percentage": 79.19, "elapsed_time": "22:50:17", "remaining_time": "6:00:09"}
1858
+ {"current_steps": 884, "total_steps": 1230, "loss": 0.1881, "lr": 8.8915450839636e-06, "epoch": 3.5853164556962023, "percentage": 71.87, "elapsed_time": "20:40:36", "remaining_time": "8:05:34"}
1859
+ {"current_steps": 975, "total_steps": 1230, "loss": 0.1744, "lr": 5.01242651541854e-06, "epoch": 3.9539240506329114, "percentage": 79.27, "elapsed_time": "22:51:58", "remaining_time": "5:58:49"}
1860
+ {"current_steps": 885, "total_steps": 1230, "loss": 0.1663, "lr": 8.844391174552116e-06, "epoch": 3.589367088607595, "percentage": 71.95, "elapsed_time": "20:41:54", "remaining_time": "8:04:07"}
1861
+ {"current_steps": 976, "total_steps": 1230, "loss": 0.1798, "lr": 4.974904635542815e-06, "epoch": 3.957974683544304, "percentage": 79.35, "elapsed_time": "22:53:21", "remaining_time": "5:57:24"}
1862
+ {"current_steps": 886, "total_steps": 1230, "loss": 0.1685, "lr": 8.797327110880479e-06, "epoch": 3.5934177215189873, "percentage": 72.03, "elapsed_time": "20:43:18", "remaining_time": "8:02:43"}
1863
+ {"current_steps": 977, "total_steps": 1230, "loss": 0.1693, "lr": 4.937503765719582e-06, "epoch": 3.962025316455696, "percentage": 79.43, "elapsed_time": "22:54:49", "remaining_time": "5:56:01"}
1864
+ {"current_steps": 887, "total_steps": 1230, "loss": 0.1644, "lr": 8.750353271996206e-06, "epoch": 3.59746835443038, "percentage": 72.11, "elapsed_time": "20:44:32", "remaining_time": "8:01:15"}
1865
+ {"current_steps": 978, "total_steps": 1230, "loss": 0.1723, "lr": 4.900224207170299e-06, "epoch": 3.9660759493670885, "percentage": 79.51, "elapsed_time": "22:56:05", "remaining_time": "5:54:34"}
1866
+ {"current_steps": 888, "total_steps": 1230, "loss": 0.1549, "lr": 8.703470036220132e-06, "epoch": 3.6015189873417723, "percentage": 72.2, "elapsed_time": "20:45:51", "remaining_time": "7:59:49"}
1867
+ {"current_steps": 979, "total_steps": 1230, "loss": 0.1807, "lr": 4.8630662601394065e-06, "epoch": 3.970126582278481, "percentage": 79.59, "elapsed_time": "22:57:32", "remaining_time": "5:53:10"}
1868
+ {"current_steps": 889, "total_steps": 1230, "loss": 0.1613, "lr": 8.656677781143394e-06, "epoch": 3.6055696202531644, "percentage": 72.28, "elapsed_time": "20:47:17", "remaining_time": "7:58:25"}
1869
+ {"current_steps": 980, "total_steps": 1230, "loss": 0.1697, "lr": 4.8260302238918995e-06, "epoch": 3.9741772151898735, "percentage": 79.67, "elapsed_time": "22:58:54", "remaining_time": "5:51:45"}
1870
+ {"current_steps": 890, "total_steps": 1230, "loss": 0.1883, "lr": 8.609976883624377e-06, "epoch": 3.609620253164557, "percentage": 72.36, "elapsed_time": "20:48:40", "remaining_time": "7:57:01"}
1871
+ {"current_steps": 981, "total_steps": 1230, "loss": 0.1538, "lr": 4.789116396710924e-06, "epoch": 3.978227848101266, "percentage": 79.76, "elapsed_time": "23:00:01", "remaining_time": "5:50:16"}
1872
+ {"current_steps": 891, "total_steps": 1230, "loss": 0.1497, "lr": 8.563367719785698e-06, "epoch": 3.6136708860759494, "percentage": 72.44, "elapsed_time": "20:49:57", "remaining_time": "7:55:34"}
1873
+ {"current_steps": 982, "total_steps": 1230, "loss": 0.1717, "lr": 4.752325075895368e-06, "epoch": 3.982278481012658, "percentage": 79.84, "elapsed_time": "23:01:19", "remaining_time": "5:48:50"}
1874
+ {"current_steps": 892, "total_steps": 1230, "loss": 0.1771, "lr": 8.516850665011138e-06, "epoch": 3.6177215189873415, "percentage": 72.52, "elapsed_time": "20:51:09", "remaining_time": "7:54:05"}
1875
+ {"current_steps": 983, "total_steps": 1230, "loss": 0.1488, "lr": 4.715656557757473e-06, "epoch": 3.9863291139240506, "percentage": 79.92, "elapsed_time": "23:02:22", "remaining_time": "5:47:21"}
1876
+ {"current_steps": 893, "total_steps": 1230, "loss": 0.1636, "lr": 8.47042609394269e-06, "epoch": 3.621772151898734, "percentage": 72.6, "elapsed_time": "20:52:39", "remaining_time": "7:52:43"}
1877
+ {"current_steps": 984, "total_steps": 1230, "loss": 0.1797, "lr": 4.679111137620442e-06, "epoch": 3.990379746835443, "percentage": 80.0, "elapsed_time": "23:03:54", "remaining_time": "5:45:58"}
1878
+ {"current_steps": 894, "total_steps": 1230, "loss": 0.1755, "lr": 8.424094380477432e-06, "epoch": 3.6258227848101265, "percentage": 72.68, "elapsed_time": "20:54:01", "remaining_time": "7:51:18"}
1879
+ {"current_steps": 985, "total_steps": 1230, "loss": 0.1741, "lr": 4.6426891098160585e-06, "epoch": 3.9944303797468352, "percentage": 80.08, "elapsed_time": "23:05:18", "remaining_time": "5:44:34"}
1880
+ {"current_steps": 895, "total_steps": 1230, "loss": 0.1847, "lr": 8.37785589776465e-06, "epoch": 3.629873417721519, "percentage": 72.76, "elapsed_time": "20:55:26", "remaining_time": "7:49:54"}
1881
+ {"current_steps": 986, "total_steps": 1230, "loss": 0.1732, "lr": 4.6063907676823474e-06, "epoch": 3.9984810126582278, "percentage": 80.16, "elapsed_time": "23:06:36", "remaining_time": "5:43:08"}
1882
+ {"current_steps": 896, "total_steps": 1230, "loss": 0.1948, "lr": 8.331711018202694e-06, "epoch": 3.6339240506329116, "percentage": 72.85, "elapsed_time": "20:56:48", "remaining_time": "7:48:29"}
1883
+ {"current_steps": 897, "total_steps": 1230, "loss": 0.1564, "lr": 8.285660113436104e-06, "epoch": 3.6379746835443036, "percentage": 72.93, "elapsed_time": "20:57:59", "remaining_time": "7:47:00"}
1884
+ {"current_steps": 987, "total_steps": 1230, "loss": 0.1394, "lr": 4.570216403561141e-06, "epoch": 4.004050632911392, "percentage": 80.24, "elapsed_time": "23:09:15", "remaining_time": "5:42:02"}
1885
+ {"current_steps": 898, "total_steps": 1230, "loss": 0.1743, "lr": 8.239703554352527e-06, "epoch": 3.642025316455696, "percentage": 73.01, "elapsed_time": "20:59:26", "remaining_time": "7:45:37"}
1886
+ {"current_steps": 988, "total_steps": 1230, "loss": 0.1317, "lr": 4.534166308795815e-06, "epoch": 4.008101265822785, "percentage": 80.33, "elapsed_time": "23:10:38", "remaining_time": "5:40:37"}
1887
+ {"current_steps": 899, "total_steps": 1230, "loss": 0.1638, "lr": 8.193841711079775e-06, "epoch": 3.6460759493670887, "percentage": 73.09, "elapsed_time": "21:00:42", "remaining_time": "7:44:10"}
1888
+ {"current_steps": 989, "total_steps": 1230, "loss": 0.1346, "lr": 4.498240773728859e-06, "epoch": 4.012151898734177, "percentage": 80.41, "elapsed_time": "23:12:06", "remaining_time": "5:39:13"}
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:fbfe2753ce890402118e13912d3877e6acebdd7767919752be931aea1065b060
3
  size 7288
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2b109346f835356baa84e067fdb9a82cdcae7d60dd3739e1d9b7b4e6b1eca559
3
  size 7288