File size: 1,654 Bytes
e923d3b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
Writing logs to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/roberta-base-glue:cola-2020-06-29-14:54/log.txt.
Loading nlp dataset glue, subset cola, split train.
Loading nlp dataset glue, subset cola, split validation.
Loaded dataset. Found: 2 labels: ([0, 1])
Loading transformers AutoModelForSequenceClassification: roberta-base
Tokenizing training data. (len: 8551)
Tokenizing eval data (len: 1043)
Loaded data and tokenized in 20.26492166519165s
Training model across 4 GPUs
***** Running training *****
	Num examples = 8551
	Batch size = 32
	Max sequence length = 128
	Num steps = 1335
	Num epochs = 5
	Learning rate = 2e-05
Eval accuracy: 81.87919463087249%
Best acc found. Saved model to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/roberta-base-glue:cola-2020-06-29-14:54/.
Eval accuracy: 85.0431447746884%
Best acc found. Saved model to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/roberta-base-glue:cola-2020-06-29-14:54/.
Eval accuracy: 84.18024928092042%
Eval accuracy: 84.0843720038351%
Eval accuracy: 84.75551294343241%
Saved tokenizer <textattack.models.tokenizers.auto_tokenizer.AutoTokenizer object at 0x7f94dc097dc0> to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/roberta-base-glue:cola-2020-06-29-14:54/.
Wrote README to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/roberta-base-glue:cola-2020-06-29-14:54/README.md.
Wrote training args to /p/qdata/jm8wx/research/text_attacks/textattack/outputs/training/roberta-base-glue:cola-2020-06-29-14:54/train_args.json.