-
Notifications
You must be signed in to change notification settings - Fork 25.3k
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
interpolation added for TVP. #30863
interpolation added for TVP. #30863
Conversation
…nmdev/transformers into interpolate-pos-encode-tvp
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.
Thanks for working on adding this!
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.
Thanks for the continued work on this!
Please make sure to either comment on suggestions why you're not implementing them or implement the suggestion before marking as resolved. Comments making suggestions to make e.g. docstrings in-line with the library standard will have to be implemented
expected_shape = torch.Size((1, 2)) | ||
assert outputs.logits.shape == expected_shape |
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.
Option 2 :)
@bhuvanmdev For the quality checks could you:
|
Done. Did I miss any changes to make? @amyeroberts |
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.
Thanks for the continued work on this!
Just a few final things to address before merge
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.
Thanks for adding this!
What does this PR do?
Adds interpolation to the tvp model for the position and pad embeddings.
Addresses #30579
Who can review?
@amyeroberts