Skip to content

Conversation

@yhtang
Copy link
Contributor

@yhtang yhtang commented Nov 25, 2025

This PR makes the following changes:

  • Bumps vLLM from 0.10.2 to 0.11.2.
  • Patches the vLLM weight loader (post model loading) as a workaround to ensure we can send exact weight shards to each vLLM GPU.
  • Addresses the known CVEs affecting vLLM versions prior to 0.11.1.

@yhtang yhtang requested a review from jreiffers November 25, 2025 08:22
@yhtang yhtang requested a review from jreiffers November 26, 2025 07:00
@yhtang
Copy link
Contributor Author

yhtang commented Nov 28, 2025

I've removed all non-essential changes and added a few more comments to explain the remaining changeset. Please let me know your thoughts. Thanks! @jreiffers

@yhtang yhtang merged commit 34cbc5f into main Nov 28, 2025
91 of 104 checks passed
@yhtang yhtang deleted the yhtang/vllm-0.11-bump branch November 28, 2025 21:26
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.

3 participants