bert-large-uncased-stsb / training.log
yoshitomo-matsubara's picture
tuned hyperparameters
808baa0
2021-05-25 20:17:27,706 INFO __main__ Namespace(adjust_lr=False, config='torchdistill/configs/sample/glue/stsb/mse/bert_large_uncased.yaml', log='log/glue/stsb/mse/bert_large_uncased.txt', private_output='leaderboard/glue/standard/bert_large_uncased/', seed=None, student_only=False, task_name='stsb', test_only=False, world_size=1)
2021-05-25 20:17:27,744 INFO __main__ Distributed environment: NO
Num processes: 1
Process index: 0
Local process index: 0
Device: cuda
Use FP16 precision: True
2021-05-25 20:17:37,444 INFO __main__ Start training
2021-05-25 20:17:37,444 INFO torchdistill.models.util [student model]
2021-05-25 20:17:37,445 INFO torchdistill.models.util Using the original student model
2021-05-25 20:17:37,445 INFO torchdistill.core.training Loss = 1.0 * OrgLoss
2021-05-25 20:17:44,656 INFO torchdistill.misc.log Epoch: [0] [ 0/180] eta: 0:02:54 lr: 2.9944444444444443e-05 sample/s: 4.157108273720882 loss: 12.1362 (12.1362) time: 0.9706 data: 0.0084 max mem: 6148
2021-05-25 20:18:24,842 INFO torchdistill.misc.log Epoch: [0] [ 50/180] eta: 0:01:44 lr: 2.716666666666667e-05 sample/s: 5.079776319530379 loss: 7.2100 (10.7919) time: 0.7976 data: 0.0035 max mem: 10858
2021-05-25 20:19:03,758 INFO torchdistill.misc.log Epoch: [0] [100/180] eta: 0:01:03 lr: 2.438888888888889e-05 sample/s: 5.071751190230392 loss: 0.7149 (6.2958) time: 0.7898 data: 0.0039 max mem: 10858
2021-05-25 20:19:42,944 INFO torchdistill.misc.log Epoch: [0] [150/180] eta: 0:00:23 lr: 2.161111111111111e-05 sample/s: 4.658769325438214 loss: 0.5977 (4.4207) time: 0.7743 data: 0.0036 max mem: 10861
2021-05-25 20:20:07,011 INFO torchdistill.misc.log Epoch: [0] Total time: 0:02:23
2021-05-25 20:20:15,900 INFO /usr/local/lib/python3.7/dist-packages/datasets/metric.py Removing /root/.cache/huggingface/metrics/glue/stsb/default_experiment-1-0.arrow
2021-05-25 20:20:15,900 INFO __main__ Validation: pearson = 0.8737826052062094, spearmanr = 0.8735916113223366
2021-05-25 20:20:15,900 INFO __main__ Updating ckpt
2021-05-25 20:20:21,438 INFO torchdistill.misc.log Epoch: [1] [ 0/180] eta: 0:01:58 lr: 1.9944444444444447e-05 sample/s: 6.128770842374588 loss: 0.4325 (0.4325) time: 0.6593 data: 0.0067 max mem: 10861
2021-05-25 20:21:00,988 INFO torchdistill.misc.log Epoch: [1] [ 50/180] eta: 0:01:42 lr: 1.7166666666666666e-05 sample/s: 4.297354562406508 loss: 0.3274 (0.3845) time: 0.8074 data: 0.0038 max mem: 10861
2021-05-25 20:21:41,487 INFO torchdistill.misc.log Epoch: [1] [100/180] eta: 0:01:03 lr: 1.438888888888889e-05 sample/s: 7.083348708467611 loss: 0.3129 (0.3857) time: 0.8153 data: 0.0037 max mem: 12382
2021-05-25 20:22:21,324 INFO torchdistill.misc.log Epoch: [1] [150/180] eta: 0:00:23 lr: 1.161111111111111e-05 sample/s: 6.354031294441572 loss: 0.2832 (0.3657) time: 0.8022 data: 0.0038 max mem: 12382
2021-05-25 20:22:43,475 INFO torchdistill.misc.log Epoch: [1] Total time: 0:02:22
2021-05-25 20:22:52,330 INFO /usr/local/lib/python3.7/dist-packages/datasets/metric.py Removing /root/.cache/huggingface/metrics/glue/stsb/default_experiment-1-0.arrow
2021-05-25 20:22:52,330 INFO __main__ Validation: pearson = 0.8993742147980403, spearmanr = 0.8971063152009764
2021-05-25 20:22:52,331 INFO __main__ Updating ckpt
2021-05-25 20:22:58,258 INFO torchdistill.misc.log Epoch: [2] [ 0/180] eta: 0:02:25 lr: 9.944444444444445e-06 sample/s: 4.98979598816883 loss: 0.1504 (0.1504) time: 0.8066 data: 0.0050 max mem: 12382
2021-05-25 20:23:37,330 INFO torchdistill.misc.log Epoch: [2] [ 50/180] eta: 0:01:41 lr: 7.166666666666667e-06 sample/s: 5.076305784711369 loss: 0.1316 (0.1620) time: 0.7998 data: 0.0037 max mem: 12382
2021-05-25 20:24:18,428 INFO torchdistill.misc.log Epoch: [2] [100/180] eta: 0:01:04 lr: 4.388888888888889e-06 sample/s: 4.601912661611403 loss: 0.1340 (0.1553) time: 0.8041 data: 0.0037 max mem: 12382
2021-05-25 20:24:57,662 INFO torchdistill.misc.log Epoch: [2] [150/180] eta: 0:00:23 lr: 1.6111111111111111e-06 sample/s: 5.072415148418462 loss: 0.1245 (0.1507) time: 0.7877 data: 0.0038 max mem: 12382
2021-05-25 20:25:20,536 INFO torchdistill.misc.log Epoch: [2] Total time: 0:02:23
2021-05-25 20:25:29,396 INFO /usr/local/lib/python3.7/dist-packages/datasets/metric.py Removing /root/.cache/huggingface/metrics/glue/stsb/default_experiment-1-0.arrow
2021-05-25 20:25:29,397 INFO __main__ Validation: pearson = 0.9034122016001204, spearmanr = 0.9010440275420903
2021-05-25 20:25:29,397 INFO __main__ Updating ckpt
2021-05-25 20:25:40,164 INFO __main__ [Student: bert-large-uncased]
2021-05-25 20:25:49,037 INFO /usr/local/lib/python3.7/dist-packages/datasets/metric.py Removing /root/.cache/huggingface/metrics/glue/stsb/default_experiment-1-0.arrow
2021-05-25 20:25:49,038 INFO __main__ Test: pearson = 0.9034122016001204, spearmanr = 0.9010440275420903
2021-05-25 20:25:49,038 INFO __main__ Start prediction for private dataset(s)
2021-05-25 20:25:49,039 INFO __main__ stsb/test: 1379 samples