Training/evaluation parameters Namespace(data_dir='bias_toxic_truthful/toxic/', model_dir='/data/ryanyip/Text-Classifier', model_name='bert-base-cased', output_dir='checkpoints/ToxicClassifier', do_train=True, do_predict=False, result_output_dir='checkpoints/ToxicClassifier/result', max_length=128, train_batch_size=16, eval_batch_size=16, learning_rate=5e-05, weight_decay=0.01, adam_epsilon=1e-08, max_grad_norm=1.0, epochs=10, warmup_proportion=0.1, earlystop_patience=2, logging_steps=10, save_steps=5000, seed=2021, device=device(type='cuda'), model_type='bert', task_name='qic')
***** Running training *****
Num samples 79212
Num epochs 10
Num training steps 49510
Num warmup steps 4951
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.2161971119862668
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-10
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.21791376350600827
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-20
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.22003433303039482
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-30
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.2237705745733616
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-40
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.24012925376148642
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-50
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.27819852569928305
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-60
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.3636271836817126
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-70
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.5175199434514793
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-80
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.6735332727456327
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-90
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.7494698576189034
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-100
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.7759264869231546
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-110
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.7842068060183782
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-120
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.7872361910532162
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-130
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.7874381500555387
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-140
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.7877410885590225
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-150
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.7877410885590225
Earlystopper counter: 1 out of 2
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.7875391295567
Earlystopper counter: 2 out of 2
Training Stop! The best step 150: 0.7877410885590225
Saving models checkpoint to checkpoints/ToxicClassifier
Training/evaluation parameters Namespace(data_dir='bias_toxic_truthful/toxic/', model_dir='/data/ryanyip/Text-Classifier', model_name='bert-base-cased', output_dir='checkpoints/ToxicClassifier', do_train=True, do_predict=False, result_output_dir='checkpoints/ToxicClassifier/result', max_length=128, train_batch_size=32, eval_batch_size=16, learning_rate=5e-05, weight_decay=0.01, adam_epsilon=1e-08, max_grad_norm=1.0, epochs=20, warmup_proportion=0.1, earlystop_patience=2, logging_steps=100, save_steps=5000, seed=2021, device=device(type='cuda'), model_type='bert', task_name='qic')
***** Running training *****
Num samples 79212
Num epochs 20
Num training steps 49520
Num warmup steps 4952
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.7775421589417348
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-100
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.7878420680601838
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-200
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.8174290619004342
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-300
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.8627688579218419
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-400
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.8869029586993841
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-500
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.8974048268201555
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-600
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.9020498838735737
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-700
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.906392002423508
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-800
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.9112390184792487
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-900
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.9129556699989902
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-1000
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.9129556699989902
Earlystopper counter: 1 out of 2
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.9181056245582147
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-1200
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.9213369685953752
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-1300
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.9223467636069878
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-1400
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.9253761486418257
Saving models checkpoint to checkpoints/ToxicClassifier/checkpoint-1500
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.9202261940826012
Earlystopper counter: 1 out of 2
***** Running evaluation *****
Num samples 9903
qic-bert-base-cased acc: 0.9241643946278906
Earlystopper counter: 2 out of 2
Training Stop! The best step 1500: 0.9253761486418257
Saving models checkpoint to checkpoints/ToxicClassifier