search

songyingxin / Bert-TextClassification

Implemention some Baseline Model upon Bert for Text Classification
Other
682 stars 149 forks source link

gradient accumulation #14

Open Atomu2014 opened 4 years ago

Atomu2014 commented 4 years ago

Thanks for sharing the code. Your gradient accumulation implementation helps me a lot on my datasets (roughly >10% f1 improvements with very large batch size).

Please check line 87 of train_evaluate.py. I think it should be "train_steps" instead of "step".

Thanks