lstm-chinanews-chinese/train_log.txt

Writing logs to ./outputs/2024-03-04-17-14-45-472632/train_log.txt.
Wrote original training args to ./outputs/2024-03-04-17-14-45-472632/training_args.json.
***** Running training *****
  Num examples = 50000
  Num epochs = 10
  Num clean epochs = 10
  Instantaneous batch size per device = 16
  Total train batch size (w. parallel, distributed & accumulation) = 16
  Gradient accumulation steps = 1
  Total optimization steps = 31250
==========================================================
Epoch 1
Running clean epoch 1/10
Train accuracy: 73.86%
Eval accuracy: 85.25%
Best score found. Saved model to ./outputs/2024-03-04-17-14-45-472632/best_model/
==========================================================
Epoch 2
Running clean epoch 2/10
Train accuracy: 88.42%
Eval accuracy: 87.05%
Best score found. Saved model to ./outputs/2024-03-04-17-14-45-472632/best_model/
==========================================================
Epoch 3
Running clean epoch 3/10
Train accuracy: 92.21%
Eval accuracy: 87.67%
Best score found. Saved model to ./outputs/2024-03-04-17-14-45-472632/best_model/
==========================================================
Epoch 4
Running clean epoch 4/10
Train accuracy: 94.91%
Eval accuracy: 87.34%
==========================================================
Epoch 5
Running clean epoch 5/10
Train accuracy: 97.15%
Eval accuracy: 87.03%
==========================================================
Epoch 6
Running clean epoch 6/10
Train accuracy: 98.45%
Eval accuracy: 86.12%
==========================================================
Epoch 7
Running clean epoch 7/10
Train accuracy: 99.25%
Eval accuracy: 85.90%
==========================================================
Epoch 8
Running clean epoch 8/10
Train accuracy: 99.60%
Eval accuracy: 85.51%
==========================================================
Epoch 9
Running clean epoch 9/10
Train accuracy: 99.82%
Eval accuracy: 85.36%
==========================================================
Epoch 10
Running clean epoch 10/10
Train accuracy: 99.91%
Eval accuracy: 84.99%
Wrote README to ./outputs/2024-03-04-17-14-45-472632/README.md.