racheltong commited on
Commit
2153ec0
·
verified ·
1 Parent(s): 614c40a

Training in progress, step 3000, checkpoint

Browse files
last-checkpoint/model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2029391cdc6b46f4bce12fa986e429f088e4ca217b00c752919f09b25d1aa4bb
3
  size 430972780
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c93dea5672a94e182e4397e46e957256dd9b846289709ace4f8f52f5cb1f6261
3
  size 430972780
last-checkpoint/optimizer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:30a6e4c391f19b7ce4a9247653997956b7791e634e3e7d6e4b83d1aaf8b9d735
3
  size 862068555
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0a52b46121dff2f3989693cb1f9371359b95edfcc49e5b52a31f52e221461dca
3
  size 862068555
last-checkpoint/rng_state.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e58b7b99958a5f50469af7b1d06bb70e5e8e003683247f5979cf608a8c999b4b
3
  size 14645
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f52cdf95e913ec93635df4b53ed0a27c36d33386e0ffa4a498c6b323ccd609a1
3
  size 14645
last-checkpoint/scaler.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:51009c0139f68650982f878b1fe583aaec3addccf6a83da1e1088ec01a295e18
3
  size 1383
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:226d41de2cc700f20b8b0f77d5bec21bf9dbf120fa3828c7b922257be8a4302f
3
  size 1383
last-checkpoint/scheduler.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:358d68264cb066817a0cd8c719c28d96b7ac27147c3224998b457ffec6474d88
3
  size 1465
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3d646e14b9059ed374c0eac5b08f32a92dbd61dc812669511b3d88fa44a1ae0c
3
  size 1465
last-checkpoint/trainer_state.json CHANGED
@@ -2,9 +2,9 @@
2
  "best_global_step": 2700,
3
  "best_metric": 0.9454253611556982,
4
  "best_model_checkpoint": "/content/drive/MyDrive/my_works/Work/CSA/models/comparison/ClinicalBERT/checkpoints/checkpoint-2700",
5
- "epoch": 1.5959207803724504,
6
  "eval_steps": 300,
7
- "global_step": 2700,
8
  "is_hyper_param_search": false,
9
  "is_local_process_zero": true,
10
  "is_world_process_zero": true,
@@ -593,6 +593,71 @@
593
  "eval_samples_per_second": 529.051,
594
  "eval_steps_per_second": 16.548,
595
  "step": 2700
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
596
  }
597
  ],
598
  "logging_steps": 50,
@@ -612,7 +677,7 @@
612
  "attributes": {}
613
  }
614
  },
615
- "total_flos": 5643840120381696.0,
616
  "train_batch_size": 16,
617
  "trial_name": null,
618
  "trial_params": null
 
2
  "best_global_step": 2700,
3
  "best_metric": 0.9454253611556982,
4
  "best_model_checkpoint": "/content/drive/MyDrive/my_works/Work/CSA/models/comparison/ClinicalBERT/checkpoints/checkpoint-2700",
5
+ "epoch": 1.7732781554832988,
6
  "eval_steps": 300,
7
+ "global_step": 3000,
8
  "is_hyper_param_search": false,
9
  "is_local_process_zero": true,
10
  "is_world_process_zero": true,
 
593
  "eval_samples_per_second": 529.051,
594
  "eval_steps_per_second": 16.548,
595
  "step": 2700
596
+ },
597
+ {
598
+ "epoch": 1.6254803428909252,
599
+ "grad_norm": 0.04367101565003395,
600
+ "learning_rate": 2.070393698648836e-06,
601
+ "loss": 0.005545063018798828,
602
+ "step": 2750
603
+ },
604
+ {
605
+ "epoch": 1.6550399054093998,
606
+ "grad_norm": 0.1608746349811554,
607
+ "learning_rate": 1.7667815410260181e-06,
608
+ "loss": 0.00613764226436615,
609
+ "step": 2800
610
+ },
611
+ {
612
+ "epoch": 1.6845994679278746,
613
+ "grad_norm": 0.037590060383081436,
614
+ "learning_rate": 1.4850741428268244e-06,
615
+ "loss": 0.0038146468997001647,
616
+ "step": 2850
617
+ },
618
+ {
619
+ "epoch": 1.7141590304463494,
620
+ "grad_norm": 0.06141559034585953,
621
+ "learning_rate": 1.2260209962232628e-06,
622
+ "loss": 0.005997980237007141,
623
+ "step": 2900
624
+ },
625
+ {
626
+ "epoch": 1.7437185929648242,
627
+ "grad_norm": 0.04523707181215286,
628
+ "learning_rate": 9.903113209758098e-07,
629
+ "loss": 0.00510674774646759,
630
+ "step": 2950
631
+ },
632
+ {
633
+ "epoch": 1.7732781554832988,
634
+ "grad_norm": 8.224442481994629,
635
+ "learning_rate": 7.785722307406685e-07,
636
+ "loss": 0.004304350912570953,
637
+ "step": 3000
638
+ },
639
+ {
640
+ "epoch": 1.7732781554832988,
641
+ "eval_f1_ACCOUNT": 0.97,
642
+ "eval_f1_AGE_OVER_89": 0.99,
643
+ "eval_f1_DATE": 0.82,
644
+ "eval_f1_DEVICE_ID": 0.99,
645
+ "eval_f1_EMAIL": 0.99,
646
+ "eval_f1_LOCATION": 0.9,
647
+ "eval_f1_MRN": 0.96,
648
+ "eval_f1_NAME": 0.94,
649
+ "eval_f1_PHONE": 0.93,
650
+ "eval_f1_SSN": 1.0,
651
+ "eval_f1_URL": 1.0,
652
+ "eval_f1_macro": 0.943275291101378,
653
+ "eval_f1_weighted": 0.943275291101378,
654
+ "eval_loss": 0.001783758751116693,
655
+ "eval_precision": 0.9123785906178963,
656
+ "eval_recall": 0.9763379035824856,
657
+ "eval_runtime": 11.95,
658
+ "eval_samples_per_second": 551.129,
659
+ "eval_steps_per_second": 17.238,
660
+ "step": 3000
661
  }
662
  ],
663
  "logging_steps": 50,
 
677
  "attributes": {}
678
  }
679
  },
680
+ "total_flos": 6271071399895296.0,
681
  "train_batch_size": 16,
682
  "trial_name": null,
683
  "trial_params": null