Unlikelihood loss about nag-bert HOT 3 OPEN

@yxuansu Thank you for your response. Looking forward for the code of the unlikelihood loss.

from nag-bert.

Hi, thanks for a great paper and code repository!

When looking at your code inside the main.py file, I see that you used the regular negative log-likelihood loss and I haven't seen in the code any reference to the unlikelihood loss of the context-aware (with a context window of size c) term that is mentioned on the paper.

Can you please point out to where in the code this loss is configured? Thanks.

Hey, thank you for your question. We found only using NLL loss + CRF loss would make the training most stable. So we only release the base version of our model. We will update the unlikelihood loss soon.

from nag-bert.

@yxuansu hi，When will you update the unlikelihood loss ?

from nag-bert.