[CI / core] Fix CI with GC + pytorch 2.2#29026
Closed
younesbelkada wants to merge 2 commits intohuggingface:mainfrom
Closed
[CI / core] Fix CI with GC + pytorch 2.2#29026younesbelkada wants to merge 2 commits intohuggingface:mainfrom
CI / core] Fix CI with GC + pytorch 2.2#29026younesbelkada wants to merge 2 commits intohuggingface:mainfrom
Conversation
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
Contributor
Author
|
It appears the rootcause was slightlly different, see: #29027 |
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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?
Fixes the current failing CI for Mistral, Mixtral and Qwen2 for gradient checkpointing. For some reason, since pytorch 2.2, gradient checkpointing raises an error when going through in-place operations such as
tensor.mul_(xxx)which was not the case in earlier versions.Simply replacing
causal_mask.mul_(~torch.eq(causal_mask, causal_mask.min()).all(dim=-1)[..., None])bycausal_mask = causal_mask * (~torch.eq(causal_mask, causal_mask.min()).all(dim=-1)[..., None])This makes me think we should maybe have a job that runs the CI on torch nightly to catch these early bugs, do we have that already? If not, happy to have a look
cc @ArthurZucker @amyeroberts