Skip to content

Fixes for #33559 and #34103#1019

Open
iboiko-habana wants to merge 2 commits intovllm-project:releases/v0.16.0from
iboiko-habana:fix_34103
Open

Fixes for #33559 and #34103#1019
iboiko-habana wants to merge 2 commits intovllm-project:releases/v0.16.0from
iboiko-habana:fix_34103

Conversation

@iboiko-habana
Copy link
Collaborator

  1. Port of Qwen3_vl WarmUp Fix #994 - #33559
  2. change: "from vllm.multimodal.budget import MultiModalBudget" to "from vllm.multimodal.encoder_budget import MultiModalBudget" - #34103

Signed-off-by: Iryna Boiko <iboiko@habana.ai>
Copy link
Contributor

Copilot AI left a comment

Choose a reason for hiding this comment

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

Pull request overview

This PR ports a multimodal warmup fix (from vllm-gaudi PR #994) and updates the MultiModalBudget import path to match upstream vLLM changes, addressing issues #33559 and #34103.

Changes:

  • Fix _get_mm_dummy_batch to pass (modality, item) pairs into group_mm_kwargs_by_modality, aligning with how other call sites batch multimodal inputs.
  • Update MultiModalBudget import to vllm.multimodal.encoder_budget to reflect the new upstream module location.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

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.

2 participants