codertimo/BERT-pytorch

Question about the loss of Masked LM

Open

#49 opened on Dec 7, 2018

View on GitHub
 (5 comments) (13 reactions) (0 assignees)Python (5,757 stars) (1,252 forks)batch import
good first issue

Description

Thank you very much for this great contribution. I found the loss of masked LM didn't decrease when it reaches the value around 7. However, in the official tensorflow implementation, the loss of MLM decreases to 1 easily. I think something went wrong in your implementation. In additional, I found the code can not predict the next sentence correctly. I think the reason is: self.criterion = nn.NLLLoss(ignore_index=0). It can not be used as criterion for sentence prediction because the label of sentence is 1 or 0. We should remove ignore_index=0 for sentence prediction. I am looking forward to your reply~

Contributor guide

Tech stack
pythonpytorch
Domain
machine learningai
Issue type
bug
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
4
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
half day
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
stale
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
clear
Prerequisites
BERT modelPyTorch basicsNLLLossMasked Language Model
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
30
Research direction
Investigate the masked LM loss implementation in the BERT pytorch repository. Compare with the official TensorFlow BERT to understand expected loss behavior. Check the use of nn.NLLLoss with ignore index=0 for next sentence prediction, as the issue suggests it should not be used there. Look at the training loop and loss calculation in the code, and consider contributions from other users in the comments.