Update to Pytorch 2.3.0 and transformers v4.40.2 #41
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Now EosTokenCriteria in transformers requires inputs to have at least 2 dimensions, but the TGI generator was giving only one. Unsqueezing those seem to fix the issue.
Setting it to current nightly, because earlier versions do not see the libtpu and fail. Also the basic python test that checks if XLA device is removed, since it was flawed beccause it could fall back to CPU and not detecting anything, and also because the `make tgi_test` command in the workflow was re-installing a different torch xla and libtpu versions anyway.
tengomucho
force-pushed
the
alvaro/update-pt23-transformers4.40
branch
from
1626217
to
213991d
Compare
|
I will update to transformers 4.41 later, because there are some incompatibilities due to changes in static KV cache handling. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What does this PR do?
A rather uninteresting PR, that updates Pytorch and transformers dependency versions, and makes sure that everything still works as before.
This will however allow to integrate some features that have been integrated on PT2.3 and newer transformers vesions.
Before submitting