New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add more datasets and some fixes #1455
Conversation
Thanks!
|
|
||
# Add a way for the model to terminate generation | ||
# When we predict the start of a new expected question, we want to be able to stop generation | ||
messages.append(self.tokenizer.eos_token) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
For this collator to work, we need to replace this eos token with the tag. Let me know if I am wrong.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I would change it anyway, for both this collator and the default collator
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We can merge for now, and I can make some changes now
Added collate function for mixing different tasks during in batch to make sure the attention learns to attention relevant previous context and question
Fix bugs from other PR