Hugging Face's logo

yoshitomo-matsubara
/
bert-large-uncased-rte

Text Classification Transformers PyTorch Safetensors
English bert rte glue torchdistill Inference Endpoints
Model card Files Community
1
yoshitomo-matsubara commited on
Commit
527a6a3
•
1 Parent(s): 9ea3db7

tuned hyperparameters

Browse files
Files changed (3) hide show
  1. pytorch_model.bin +1 -1
  2. tokenizer.json +0 -0
  3. training.log +43 -43
pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:1cc0cfbe790c20bbc00bf2cba58aa3d0afa500ad9c4e971953082e53de345df7
3
  size 1340746825
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0420c6d7099dd98a7abc6589eecf192a0c63b4fc661bd591c9bf00b4836b74af
3
  size 1340746825
tokenizer.json CHANGED
The diff for this file is too large to render. See raw diff
training.log CHANGED
@@ -1,49 +1,49 @@
1
- 2021-05-22 19:52:56,786 INFO __main__ Namespace(adjust_lr=False, config='torchdistill/configs/sample/glue/rte/ce/bert_large_uncased.yaml', log='log/glue/rte/ce/bert_large_uncased.txt', private_output='leaderboard/glue/standard/bert_large_uncased/', seed=None, student_only=False, task_name='rte', test_only=False, world_size=1)
2
- 2021-05-22 19:52:56,832 INFO __main__ Distributed environment: NO
3
  Num processes: 1
4
  Process index: 0
5
  Local process index: 0
6
  Device: cuda
7
  Use FP16 precision: True
8
 
9
- 2021-05-22 19:53:31,901 INFO __main__ Start training
10
- 2021-05-22 19:53:31,901 INFO torchdistill.models.util [student model]
11
- 2021-05-22 19:53:31,901 INFO torchdistill.models.util Using the original student model
12
- 2021-05-22 19:53:31,901 INFO torchdistill.core.training Loss = 1.0 * OrgLoss
13
- 2021-05-22 19:53:36,610 INFO torchdistill.misc.log Epoch: [0] [ 0/312] eta: 0:01:00 lr: 4.99465811965812e-05 sample/s: 20.909627502592325 loss: 1.1803 (1.1803) time: 0.1948 data: 0.0035 max mem: 5355
14
- 2021-05-22 19:53:43,676 INFO torchdistill.misc.log Epoch: [0] [ 50/312] eta: 0:00:37 lr: 4.7275641025641026e-05 sample/s: 28.0015087940663 loss: 0.6993 (0.7507) time: 0.1425 data: 0.0017 max mem: 7364
15
- 2021-05-22 19:53:50,696 INFO torchdistill.misc.log Epoch: [0] [100/312] eta: 0:00:29 lr: 4.460470085470086e-05 sample/s: 28.603703072254238 loss: 0.6589 (0.7230) time: 0.1386 data: 0.0016 max mem: 7364
16
- 2021-05-22 19:53:57,721 INFO torchdistill.misc.log Epoch: [0] [150/312] eta: 0:00:22 lr: 4.1933760683760684e-05 sample/s: 28.46394082318211 loss: 0.6511 (0.7048) time: 0.1420 data: 0.0017 max mem: 7364
17
- 2021-05-22 19:54:04,797 INFO torchdistill.misc.log Epoch: [0] [200/312] eta: 0:00:15 lr: 3.9262820512820513e-05 sample/s: 30.642365442534203 loss: 0.6556 (0.6884) time: 0.1418 data: 0.0017 max mem: 7364
18
- 2021-05-22 19:54:11,917 INFO torchdistill.misc.log Epoch: [0] [250/312] eta: 0:00:08 lr: 3.659188034188034e-05 sample/s: 28.549188306163426 loss: 0.5260 (0.6722) time: 0.1422 data: 0.0017 max mem: 7364
19
- 2021-05-22 19:54:18,879 INFO torchdistill.misc.log Epoch: [0] [300/312] eta: 0:00:01 lr: 3.392094017094017e-05 sample/s: 32.85520747452711 loss: 0.5626 (0.6617) time: 0.1405 data: 0.0017 max mem: 7364
20
- 2021-05-22 19:54:20,395 INFO torchdistill.misc.log Epoch: [0] Total time: 0:00:43
21
- 2021-05-22 19:54:21,113 INFO /usr/local/lib/python3.7/dist-packages/datasets/metric.py Removing /root/.cache/huggingface/metrics/glue/rte/default_experiment-1-0.arrow
22
- 2021-05-22 19:54:21,114 INFO __main__ Validation: accuracy = 0.7111913357400722
23
- 2021-05-22 19:54:21,114 INFO __main__ Updating ckpt
24
- 2021-05-22 19:54:26,919 INFO torchdistill.misc.log Epoch: [1] [ 0/312] eta: 0:00:42 lr: 3.327991452991453e-05 sample/s: 30.052638461221544 loss: 0.9158 (0.9158) time: 0.1359 data: 0.0028 max mem: 7364
25
- 2021-05-22 19:54:33,925 INFO torchdistill.misc.log Epoch: [1] [ 50/312] eta: 0:00:36 lr: 3.0608974358974366e-05 sample/s: 27.751990353011625 loss: 0.4395 (0.5118) time: 0.1401 data: 0.0017 max mem: 7364
26
- 2021-05-22 19:54:41,004 INFO torchdistill.misc.log Epoch: [1] [100/312] eta: 0:00:29 lr: 2.793803418803419e-05 sample/s: 27.844522780532817 loss: 0.5357 (0.5354) time: 0.1406 data: 0.0017 max mem: 7364
27
- 2021-05-22 19:54:48,026 INFO torchdistill.misc.log Epoch: [1] [150/312] eta: 0:00:22 lr: 2.5267094017094017e-05 sample/s: 28.286567888912117 loss: 0.4564 (0.5309) time: 0.1389 data: 0.0017 max mem: 7364
28
- 2021-05-22 19:54:55,136 INFO torchdistill.misc.log Epoch: [1] [200/312] eta: 0:00:15 lr: 2.2596153846153846e-05 sample/s: 28.280893743626052 loss: 0.5141 (0.5348) time: 0.1406 data: 0.0017 max mem: 7364
29
- 2021-05-22 19:55:02,247 INFO torchdistill.misc.log Epoch: [1] [250/312] eta: 0:00:08 lr: 1.992521367521368e-05 sample/s: 27.716275688891827 loss: 0.5522 (0.5412) time: 0.1416 data: 0.0017 max mem: 7364
30
- 2021-05-22 19:55:09,289 INFO torchdistill.misc.log Epoch: [1] [300/312] eta: 0:00:01 lr: 1.7254273504273504e-05 sample/s: 32.32847169633171 loss: 0.5199 (0.5340) time: 0.1396 data: 0.0016 max mem: 7364
31
- 2021-05-22 19:55:10,829 INFO torchdistill.misc.log Epoch: [1] Total time: 0:00:44
32
- 2021-05-22 19:55:11,553 INFO /usr/local/lib/python3.7/dist-packages/datasets/metric.py Removing /root/.cache/huggingface/metrics/glue/rte/default_experiment-1-0.arrow
33
- 2021-05-22 19:55:11,554 INFO __main__ Validation: accuracy = 0.7075812274368231
34
- 2021-05-22 19:55:11,698 INFO torchdistill.misc.log Epoch: [2] [ 0/312] eta: 0:00:44 lr: 1.6613247863247862e-05 sample/s: 28.265599202097192 loss: 0.3159 (0.3159) time: 0.1434 data: 0.0019 max mem: 7364
35
- 2021-05-22 19:55:18,779 INFO torchdistill.misc.log Epoch: [2] [ 50/312] eta: 0:00:37 lr: 1.3942307692307693e-05 sample/s: 28.61760837860658 loss: 0.0392 (0.2136) time: 0.1403 data: 0.0017 max mem: 7364
36
- 2021-05-22 19:55:25,865 INFO torchdistill.misc.log Epoch: [2] [100/312] eta: 0:00:30 lr: 1.1271367521367522e-05 sample/s: 27.91173901687127 loss: 0.1969 (0.2786) time: 0.1434 data: 0.0017 max mem: 7364
37
- 2021-05-22 19:55:32,947 INFO torchdistill.misc.log Epoch: [2] [150/312] eta: 0:00:22 lr: 8.600427350427351e-06 sample/s: 28.14766048758227 loss: 0.1217 (0.2782) time: 0.1414 data: 0.0017 max mem: 7364
38
- 2021-05-22 19:55:40,055 INFO torchdistill.misc.log Epoch: [2] [200/312] eta: 0:00:15 lr: 5.929487179487179e-06 sample/s: 28.59614620884787 loss: 0.1386 (0.2659) time: 0.1419 data: 0.0017 max mem: 7364
39
- 2021-05-22 19:55:47,234 INFO torchdistill.misc.log Epoch: [2] [250/312] eta: 0:00:08 lr: 3.258547008547009e-06 sample/s: 28.29715935732188 loss: 0.0420 (0.2671) time: 0.1457 data: 0.0018 max mem: 7364
40
- 2021-05-22 19:55:54,275 INFO torchdistill.misc.log Epoch: [2] [300/312] eta: 0:00:01 lr: 5.876068376068376e-07 sample/s: 28.389669724956935 loss: 0.1216 (0.2705) time: 0.1419 data: 0.0017 max mem: 7364
41
- 2021-05-22 19:55:55,812 INFO torchdistill.misc.log Epoch: [2] Total time: 0:00:44
42
- 2021-05-22 19:55:56,531 INFO /usr/local/lib/python3.7/dist-packages/datasets/metric.py Removing /root/.cache/huggingface/metrics/glue/rte/default_experiment-1-0.arrow
43
- 2021-05-22 19:55:56,531 INFO __main__ Validation: accuracy = 0.7256317689530686
44
- 2021-05-22 19:55:56,532 INFO __main__ Updating ckpt
45
- 2021-05-22 19:56:10,518 INFO __main__ [Student: bert-large-uncased]
46
- 2021-05-22 19:56:11,255 INFO /usr/local/lib/python3.7/dist-packages/datasets/metric.py Removing /root/.cache/huggingface/metrics/glue/rte/default_experiment-1-0.arrow
47
- 2021-05-22 19:56:11,255 INFO __main__ Test: accuracy = 0.7256317689530686
48
- 2021-05-22 19:56:11,255 INFO __main__ Start prediction for private dataset(s)
49
- 2021-05-22 19:56:11,257 INFO __main__ rte/test: 3000 samples
1
+ 2021-05-27 03:06:57,518 INFO __main__ Namespace(adjust_lr=False, config='torchdistill/configs/sample/glue/rte/ce/bert_large_uncased.yaml', log='log/glue/rte/ce/bert_large_uncased.txt', private_output='leaderboard/glue/standard/bert_large_uncased/', seed=None, student_only=False, task_name='rte', test_only=False, world_size=1)
2
+ 2021-05-27 03:06:57,550 INFO __main__ Distributed environment: NO
3
  Num processes: 1
4
  Process index: 0
5
  Local process index: 0
6
  Device: cuda
7
  Use FP16 precision: True
8
 
9
+ 2021-05-27 03:07:10,458 INFO __main__ Start training
10
+ 2021-05-27 03:07:10,459 INFO torchdistill.models.util [student model]
11
+ 2021-05-27 03:07:10,459 INFO torchdistill.models.util Using the original student model
12
+ 2021-05-27 03:07:10,459 INFO torchdistill.core.training Loss = 1.0 * OrgLoss
13
+ 2021-05-27 03:07:13,632 INFO torchdistill.misc.log Epoch: [0] [ 0/312] eta: 0:02:13 lr: 1.997863247863248e-05 sample/s: 9.410467095760941 loss: 0.6985 (0.6985) time: 0.4283 data: 0.0032 max mem: 5355
14
+ 2021-05-27 03:07:34,141 INFO torchdistill.misc.log Epoch: [0] [ 50/312] eta: 0:01:47 lr: 1.891025641025641e-05 sample/s: 9.260763560189773 loss: 0.6855 (0.7316) time: 0.4141 data: 0.0017 max mem: 7365
15
+ 2021-05-27 03:07:54,769 INFO torchdistill.misc.log Epoch: [0] [100/312] eta: 0:01:27 lr: 1.7841880341880344e-05 sample/s: 9.300060421620962 loss: 0.6784 (0.7141) time: 0.4156 data: 0.0018 max mem: 7365
16
+ 2021-05-27 03:08:15,006 INFO torchdistill.misc.log Epoch: [0] [150/312] eta: 0:01:06 lr: 1.6773504273504274e-05 sample/s: 9.304485750349947 loss: 0.6591 (0.7030) time: 0.4016 data: 0.0017 max mem: 7365
17
+ 2021-05-27 03:08:35,380 INFO torchdistill.misc.log Epoch: [0] [200/312] eta: 0:00:45 lr: 1.5705128205128205e-05 sample/s: 9.304217428726076 loss: 0.6689 (0.6939) time: 0.4085 data: 0.0017 max mem: 7365
18
+ 2021-05-27 03:08:55,663 INFO torchdistill.misc.log Epoch: [0] [250/312] eta: 0:00:25 lr: 1.4636752136752137e-05 sample/s: 9.296092555242803 loss: 0.6506 (0.6908) time: 0.4118 data: 0.0018 max mem: 7365
19
+ 2021-05-27 03:09:16,350 INFO torchdistill.misc.log Epoch: [0] [300/312] eta: 0:00:04 lr: 1.356837606837607e-05 sample/s: 9.244190153358904 loss: 0.6405 (0.6823) time: 0.4173 data: 0.0017 max mem: 7365
20
+ 2021-05-27 03:09:20,776 INFO torchdistill.misc.log Epoch: [0] Total time: 0:02:07
21
+ 2021-05-27 03:09:24,609 INFO /usr/local/lib/python3.7/dist-packages/datasets/metric.py Removing /root/.cache/huggingface/metrics/glue/rte/default_experiment-1-0.arrow
22
+ 2021-05-27 03:09:24,610 INFO __main__ Validation: accuracy = 0.6173285198555957
23
+ 2021-05-27 03:09:24,610 INFO __main__ Updating ckpt at ./resource/ckpt/glue/rte/ce/rte-bert-large-uncased
24
+ 2021-05-27 03:09:29,657 INFO torchdistill.misc.log Epoch: [1] [ 0/312] eta: 0:02:19 lr: 1.3311965811965812e-05 sample/s: 9.012096369155467 loss: 0.6355 (0.6355) time: 0.4460 data: 0.0021 max mem: 7365
25
+ 2021-05-27 03:09:50,469 INFO torchdistill.misc.log Epoch: [1] [ 50/312] eta: 0:01:49 lr: 1.2243589743589746e-05 sample/s: 9.16842369299592 loss: 0.5310 (0.5850) time: 0.4242 data: 0.0017 max mem: 7365
26
+ 2021-05-27 03:10:10,381 INFO torchdistill.misc.log Epoch: [1] [100/312] eta: 0:01:26 lr: 1.1175213675213676e-05 sample/s: 11.36395200609068 loss: 0.4777 (0.5638) time: 0.4081 data: 0.0017 max mem: 7365
27
+ 2021-05-27 03:10:30,677 INFO torchdistill.misc.log Epoch: [1] [150/312] eta: 0:01:05 lr: 1.0106837606837608e-05 sample/s: 9.29443941850809 loss: 0.5052 (0.5646) time: 0.4139 data: 0.0017 max mem: 7365
28
+ 2021-05-27 03:10:51,658 INFO torchdistill.misc.log Epoch: [1] [200/312] eta: 0:00:45 lr: 9.03846153846154e-06 sample/s: 9.300792527973533 loss: 0.4903 (0.5516) time: 0.4198 data: 0.0017 max mem: 7365
29
+ 2021-05-27 03:11:12,170 INFO torchdistill.misc.log Epoch: [1] [250/312] eta: 0:00:25 lr: 7.970085470085472e-06 sample/s: 9.282976357127444 loss: 0.4948 (0.5433) time: 0.4087 data: 0.0017 max mem: 7365
30
+ 2021-05-27 03:11:32,902 INFO torchdistill.misc.log Epoch: [1] [300/312] eta: 0:00:04 lr: 6.901709401709402e-06 sample/s: 9.302071412730095 loss: 0.4903 (0.5442) time: 0.4143 data: 0.0017 max mem: 7365
31
+ 2021-05-27 03:11:36,971 INFO torchdistill.misc.log Epoch: [1] Total time: 0:02:07
32
+ 2021-05-27 03:11:40,801 INFO /usr/local/lib/python3.7/dist-packages/datasets/metric.py Removing /root/.cache/huggingface/metrics/glue/rte/default_experiment-1-0.arrow
33
+ 2021-05-27 03:11:40,801 INFO __main__ Validation: accuracy = 0.740072202166065
34
+ 2021-05-27 03:11:40,802 INFO __main__ Updating ckpt at ./resource/ckpt/glue/rte/ce/rte-bert-large-uncased
35
+ 2021-05-27 03:11:46,120 INFO torchdistill.misc.log Epoch: [2] [ 0/312] eta: 0:02:19 lr: 6.645299145299145e-06 sample/s: 9.060380879964056 loss: 0.7969 (0.7969) time: 0.4475 data: 0.0060 max mem: 7365
36
+ 2021-05-27 03:12:06,781 INFO torchdistill.misc.log Epoch: [2] [ 50/312] eta: 0:01:48 lr: 5.576923076923077e-06 sample/s: 10.741735558355614 loss: 0.3631 (0.3967) time: 0.4164 data: 0.0017 max mem: 7365
37
+ 2021-05-27 03:12:27,317 INFO torchdistill.misc.log Epoch: [2] [100/312] eta: 0:01:27 lr: 4.508547008547009e-06 sample/s: 9.16765717108441 loss: 0.3599 (0.3882) time: 0.4054 data: 0.0017 max mem: 7365
38
+ 2021-05-27 03:12:47,698 INFO torchdistill.misc.log Epoch: [2] [150/312] eta: 0:01:06 lr: 3.4401709401709403e-06 sample/s: 13.347942659561435 loss: 0.2746 (0.3688) time: 0.3992 data: 0.0017 max mem: 7365
39
+ 2021-05-27 03:13:08,144 INFO torchdistill.misc.log Epoch: [2] [200/312] eta: 0:00:45 lr: 2.371794871794872e-06 sample/s: 9.294048107726917 loss: 0.2737 (0.3552) time: 0.4062 data: 0.0017 max mem: 7365
40
+ 2021-05-27 03:13:29,127 INFO torchdistill.misc.log Epoch: [2] [250/312] eta: 0:00:25 lr: 1.3034188034188036e-06 sample/s: 9.276037203192194 loss: 0.2493 (0.3468) time: 0.4134 data: 0.0017 max mem: 7365
41
+ 2021-05-27 03:13:49,840 INFO torchdistill.misc.log Epoch: [2] [300/312] eta: 0:00:04 lr: 2.3504273504273505e-07 sample/s: 9.257023235231964 loss: 0.1451 (0.3412) time: 0.4061 data: 0.0018 max mem: 7365
42
+ 2021-05-27 03:13:54,183 INFO torchdistill.misc.log Epoch: [2] Total time: 0:02:08
43
+ 2021-05-27 03:13:58,014 INFO /usr/local/lib/python3.7/dist-packages/datasets/metric.py Removing /root/.cache/huggingface/metrics/glue/rte/default_experiment-1-0.arrow
44
+ 2021-05-27 03:13:58,014 INFO __main__ Validation: accuracy = 0.7220216606498195
45
+ 2021-05-27 03:14:04,515 INFO __main__ [Student: bert-large-uncased]
46
+ 2021-05-27 03:14:08,368 INFO /usr/local/lib/python3.7/dist-packages/datasets/metric.py Removing /root/.cache/huggingface/metrics/glue/rte/default_experiment-1-0.arrow
47
+ 2021-05-27 03:14:08,368 INFO __main__ Test: accuracy = 0.740072202166065
48
+ 2021-05-27 03:14:08,368 INFO __main__ Start prediction for private dataset(s)
49
+ 2021-05-27 03:14:08,369 INFO __main__ rte/test: 3000 samples