Skip to content
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

[Bugfix] Fix hard-coded value of x in context_attention_fwd #6373

Merged
merged 1 commit into from
Jul 13, 2024

Conversation

tdoublep
Copy link
Member

@tdoublep tdoublep commented Jul 12, 2024

Fixes #6332

The prefix prefill code assumes that x=8 but this is only the case for fp16 (e.g., see this line). In certain scenarios it is useful to be able to use the prefix prefill code with fp32 (e.g., to compare logprobs generated by chunked prefill against HF generate in CI tests). The correct value of x is very easy to extract from the key_cache.

cc @caoshiyi @zhuohan123 @DouHappy

Copy link
Collaborator

@WoosukKwon WoosukKwon left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM! Thanks for the fix!

@WoosukKwon WoosukKwon merged commit e1684a7 into vllm-project:main Jul 13, 2024
72 checks passed
dtrifiro pushed a commit to opendatahub-io/vllm that referenced this pull request Jul 17, 2024
xjpang pushed a commit to xjpang/vllm that referenced this pull request Jul 24, 2024
Alvant pushed a commit to compressa-ai/vllm that referenced this pull request Oct 26, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

[Bug]: tests/basic_correctness/test_chunked_prefill.py is failing on main in fp32
2 participants