Writing logs to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/facebook/bart-base-glue^rte-2020-08-20-11-41-21-898210/log.txt.
Loading nlp dataset glue, subset rte, split train.
Loading nlp dataset glue, subset rte, split validation.
Loaded dataset. Found: 2 labels: [0, 1]
Loading transformers AutoModelForSequenceClassification: facebook/bart-base
Training model across 1 GPUs
Wrote original training args to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/facebook/bart-base-glue^rte-2020-08-20-11-41-21-898210/train_args.json.
***** Running training *****
	Num examples = 2490
	Batch size = 64
	Max sequence length = 128
	Num steps = 190
	Num epochs = 5
	Learning rate = 2e-05
Train accuracy: 50.6425702811245%
Eval accuracy: 58.48375451263538%
Best acc found. Saved model to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/facebook/bart-base-glue^rte-2020-08-20-11-41-21-898210/.
Saved updated args to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/facebook/bart-base-glue^rte-2020-08-20-11-41-21-898210/train_args.json
Train accuracy: 65.3012048192771%
Eval accuracy: 70.3971119133574%
Best acc found. Saved model to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/facebook/bart-base-glue^rte-2020-08-20-11-41-21-898210/.
Saved updated args to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/facebook/bart-base-glue^rte-2020-08-20-11-41-21-898210/train_args.json
Train accuracy: 77.75100401606426%
Eval accuracy: 70.3971119133574%
Train accuracy: 84.17670682730923%
Eval accuracy: 71.84115523465704%
Best acc found. Saved model to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/facebook/bart-base-glue^rte-2020-08-20-11-41-21-898210/.
Saved updated args to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/facebook/bart-base-glue^rte-2020-08-20-11-41-21-898210/train_args.json
Train accuracy: 89.87951807228916%
Eval accuracy: 72.56317689530685%
Best acc found. Saved model to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/facebook/bart-base-glue^rte-2020-08-20-11-41-21-898210/.
Saved updated args to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/facebook/bart-base-glue^rte-2020-08-20-11-41-21-898210/train_args.json
Finished training. Re-loading and evaluating model from disk.
Loading transformers AutoModelForSequenceClassification: facebook/bart-base
Saved model accuracy: 72.56317689530685%
Saved tokenizer <textattack.models.tokenizers.auto_tokenizer.AutoTokenizer object at 0x7f056ac36910> to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/facebook/bart-base-glue^rte-2020-08-20-11-41-21-898210/.
Wrote README to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/facebook/bart-base-glue^rte-2020-08-20-11-41-21-898210/README.md.
Wrote final training args to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/facebook/bart-base-glue^rte-2020-08-20-11-41-21-898210/train_args.json.