Skip to content

core: bump vllm to 0.13.0#298

Merged
rebel-eunji merged 68 commits intodev-0.13from
version/vllm13
Feb 9, 2026
Merged

core: bump vllm to 0.13.0#298
rebel-eunji merged 68 commits intodev-0.13from
version/vllm13

Conversation

@rebel-eunji
Copy link
Copy Markdown
Collaborator

@rebel-eunji rebel-eunji commented Jan 26, 2026

🚀 Summary of Changes

  • This PR temporarily turns off the pre-commit hook.
  • Check whether LoRA works well.

📌 Related Issues / Tickets

  • Resolves #
  • Related to #

✅ Type of Change

  • ✨ Feature (feature)
  • 🧠 Model support (model)
  • 🧬 Core engine changes (core)
  • 🛠 Bug fix (bug-fix)
  • ⚙️ Performance improvement (perf)
  • 🔁 Refactor or code cleanup (refactor)
  • 📄 Documentation (docs)
  • ❓ Other (other): please describe

🧪 How to Test

  1. Run ...
  2. Verify output: ...
  3. Edge case tested: ...

📸 Screenshots / Logs (if applicable)


📋 Checklist

  • PR title follows Conventional Commits format
  • This PR is linked to an existing issue
  • The test method is described, and the expected result is clearly stated
  • Relevant documentation has been updated (if applicable)

💬 Notes


@rebel-eunji rebel-eunji marked this pull request as draft January 26, 2026 04:36
Comment thread .github/workflows/rbln_optimum_ci.yaml Outdated
Comment thread pyproject.toml Outdated
@rebel-eunji rebel-eunji marked this pull request as ready for review February 3, 2026 08:03
@rebel-eunji rebel-eunji changed the title (WIP) core: bump the version of vllm to 0.13.0 core: bump the version of vllm to 0.13.0 Feb 3, 2026
@rebel-eunji rebel-eunji changed the title core: bump the version of vllm to 0.13.0 core: bump vllm to 0.13.0 Feb 3, 2026
Comment thread vllm_rbln/v1/core/optimum_scheduler.py Outdated
Comment thread vllm_rbln/v1/core/optimum_scheduler.py Outdated
Comment thread vllm_rbln/v1/worker/optimum_model_runner.py Outdated
Comment thread vllm_rbln/v1/core/optimum_block_pool.py Outdated
@rebel-eunji rebel-eunji changed the base branch from dev to dev-0.13 February 4, 2026 05:59
@rebel-eunji rebel-eunji changed the base branch from dev-0.13 to dev February 4, 2026 06:05
def _build_output(
all_data: Union[torch.Tensor, list[torch.Tensor]], ) -> PoolerOutput:
"""Wrap tensor data into vLLM's PoolerOutput format."""
all_outputs = [PoolingSequenceGroupOutput(data) for data in all_data]
Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

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

Is PoolingSequenceGroupOutput deleted from vllm?

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

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

The PoolingSequenceGroupOutput function has been removed.
Each class now handles the hidden_states tensor directly, eliminating the need for the build_output function.
v0.10.2) https://github.com/vllm-project/vllm/blob/v0.10.2/vllm/model_executor/layers/pooler.py
v0.13.0) https://github.com/vllm-project/vllm/blob/v0.13.0/vllm/model_executor/layers/pooler.py


if image_embeds is not None:
if not isinstance(image_embeds, (torch.Tensor, list)):
if not isinstance(image_embeds, torch.Tensor | list):
Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

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

Are we deprecating py3.9?

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

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

vLLM has not supported Python 3.9 since version 0.11.1.
https://github.com/vllm-project/vllm/blob/v0.11.1/pyproject.toml
We are dropping support for Python 3.9, and the other SDKs are being updated accordingly.

Comment thread vllm_rbln/model_executor/models/optimum/qwen_vl.py Outdated
@rebel-eunji rebel-eunji self-assigned this Feb 4, 2026
@rebel-eunji rebel-eunji changed the base branch from dev to dev-0.13 February 7, 2026 05:16
Comment thread vllm_rbln/v1/worker/rbln_model_runner.py
Comment thread vllm_rbln/v1/worker/rbln_worker.py
Comment thread vllm_rbln/v1/worker/optimum_model_runner.py
Comment thread tests/torch_compile/v1/lora/test_layers.py
Comment thread tests/lora/test_lora.py
Comment thread tests/v1/core/utils.py
block_size: int = 16,
max_model_len: Optional[int] = None,
async_scheduling: bool = False,
is_torch_compile: bool = False,
Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

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

Great. We'll manage the tests for the torch.compile path separately, without relying on the optimum path codes. So you don’t need to worry about the torch.compile side in tests!

Copy link
Copy Markdown
Contributor

@rebel-jaehwang rebel-jaehwang left a comment

Choose a reason for hiding this comment

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

I have reviewed files under v1/.

@rebel-eunji rebel-eunji merged commit 1c2be7f into dev-0.13 Feb 9, 2026
13 of 14 checks passed
@rebel-eunji rebel-eunji deleted the version/vllm13 branch February 9, 2026 06:20
rebel-jindol21 pushed a commit that referenced this pull request Feb 11, 2026
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.

4 participants