pmichel31415/are-16-heads-really-better-than-1

Issues

• RuntimeError: can't retain_grad on Tensor that has requires_grad=False

  #7 opened 4 years ago by YJiangcm
  2

• Is the code still able to run?

  #9 opened 3 years ago by bing0037
  3

• Why do we need different normalization for all the layers compared to the last layer in BERT during importance score calculation?

  #11 opened 2 years ago by Hritikbansal
  1

• Is BERT finetuned after pruning?

  #10 opened 2 years ago by Huan80805
  2

• about the params: --raw-text and --transformer-mask-heads

  #8 opened 4 years ago by LiangQiqi677
  0

• Systematic Pruning Experiments Problem

  #6 opened 4 years ago by ChuanyangZheng
  0

• a question about run_classifier.py

  #5 opened 4 years ago by Ixuanzhang
  1

• Not able to obtain pretrained WMT model

  #4 opened 5 years ago by marwash25
  2

• Not able to prune the BERT model

  #2 opened 5 years ago by ishita1995
  7

• BERT actually_prune option not working

  #3 opened 5 years ago by pglock
  1

• No code on master?

  #1 opened 5 years ago by aninrusimha
  1