Writing logs to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/albert-base-v2-yelp_polarity-2020-07-02-10:35/log.txt. Loading nlp dataset yelp_polarity, split train. Loading nlp dataset yelp_polarity, split test. Loaded dataset. Found: 2 labels: ([0, 1]) Loading transformers AutoModelForSequenceClassification: albert-base-v2 Tokenizing training data. (len: 560000) Tokenizing eval data (len: 38000) Loaded data and tokenized in 1196.1114358901978s Training model across 4 GPUs ***** Running training ***** Num examples = 560000 Batch size = 16 Max sequence length = 512 Num steps = 175000 Num epochs = 5 Learning rate = 3e-05 Eval accuracy: 96.92368421052632% Best acc found. Saved model to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/albert-base-v2-yelp_polarity-2020-07-02-10:35/. Eval accuracy: 97.23947368421054% Best acc found. Saved model to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/albert-base-v2-yelp_polarity-2020-07-02-10:35/. Eval accuracy: 97.18947368421053% Eval accuracy: 97.5078947368421% Best acc found. Saved model to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/albert-base-v2-yelp_polarity-2020-07-02-10:35/. Eval accuracy: 97.42894736842105% Saved tokenizer to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/albert-base-v2-yelp_polarity-2020-07-02-10:35/. Wrote README to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/albert-base-v2-yelp_polarity-2020-07-02-10:35/README.md. Wrote training args to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/albert-base-v2-yelp_polarity-2020-07-02-10:35/train_args.json.