[core] gracefully error out when attn-backend x cp combo isn't supported. #12832
+31
−4
Conversation
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
|
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. |
DN6
reviewed
Dec 30, 2025
DN6
reviewed
Jan 2, 2026
| raise ValueError(f"`{backend=}` must be one of the following: " + ", ".join(available_backends)) | ||
|
|
||
| backend = AttentionBackendName(backend) | ||
| if not _AttentionBackendRegistry._is_context_parallel_available(backend): |
Collaborator
There was a problem hiding this comment.
This will raise an error if the backend doesn't support context parallel (regardless of whether you want to use CP or not)
We need to first check if the attention processor has a _parallel_config set and then check if the backend supports context parallel.
Member
Author
There was a problem hiding this comment.
That should be fixed now. Thanks for flagging.
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Outdated suggestions cannot be applied.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?
We should be able to error out when an attention backend isn't supported with CP.
Refer to #12829 (comment) and #12829 (comment).
Additionally, when we specify
parallel_configtofrom_pretrained()and then set an attention backend throughset_attention_backend()(which is CP-incompatible), it goes through silently.This PR eliminates those and gracefully errors out.