[Fix] Fixes speculative decode indexing and unpad condition for attention metadata #5626

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.

Sign up for GitHub

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

Merged

yiz-liu merged 2 commits into vllm-project:main from yiz-liu:unpad

Jan 8, 2026

Merged

[Fix] Fixes speculative decode indexing and unpad condition for attention metadata #5626

Preserves full token mappings to avoid shape mismatch

DCO / DCO succeeded Jan 8, 2026 in 1s

DCO

All commits are signed off!

View more details on DCO