Skip to content

delete ray remote function return value#4422

Open
grimoire wants to merge 4 commits intoInternLM:mainfrom
grimoire:del-ray-output
Open

delete ray remote function return value#4422
grimoire wants to merge 4 commits intoInternLM:mainfrom
grimoire:del-ray-output

Conversation

@grimoire
Copy link
Collaborator

Thanks for your contribution and we appreciate it a lot. The following instructions would make your pull request more healthy and more easily receiving feedbacks. If you do not understand some items, don't worry, just make the pull request and seek help from maintainers.

Motivation

Please describe the motivation of this PR and the goal you want to achieve through this PR.

Modification

Please briefly describe what modification is made in this PR.

BC-breaking (Optional)

Does the modification introduce changes that break the backward-compatibility of the downstream repositories?
If so, please describe how it breaks the compatibility and how the downstream projects should modify their code to keep compatibility with this PR.

Use cases (Optional)

If this PR introduces a new feature, it is better to list some use cases here, and update the documentation.

Checklist

  1. Pre-commit or other linting tools are used to fix the potential lint issues.
  2. The modification is covered by complete unit tests. If not, please add more unit tests to ensure the correctness.
  3. If the modification has a dependency on downstream projects of a newer version, this PR should be tested with all supported versions of downstream projects.
  4. The documentation has been modified accordingly, like docstring or example tutorials.

@grimoire grimoire marked this pull request as ready for review March 18, 2026 07:10
Copilot AI review requested due to automatic review settings March 18, 2026 07:11
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 updates the PyTorch Ray execution path to better manage memory/serialization by keeping outputs as detached CPU tensors (instead of NumPy conversions) and explicitly releasing shared-memory-backed outputs promptly in the engine loop.

Changes:

  • Detach tensors when converting model outputs to CPU to avoid holding autograd graphs/references.
  • Remove Ray-side output packing/unpacking that converted outputs to NumPy / back to tensors.
  • Explicitly del out after enqueuing step outputs to release shared-memory objects sooner.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.

File Description
lmdeploy/pytorch/envs.py Attempts to enable Ray zero-copy tensors via env var (currently ineffective due to no-op).
lmdeploy/pytorch/engine/model_agent/agent.py Detach tensors during CPU transfer for outputs/logprobs.
lmdeploy/pytorch/engine/executor/ray_executor.py Stops converting outputs to NumPy and back; returns tensor-based outputs directly.
lmdeploy/pytorch/engine/engine_loop.py Deletes output object after enqueuing to help timely shared-memory release.

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

You can also share your feedback on Copilot code review. Take the survey.

if out is not None:
step_outputs = self._make_infer_outputs(out, running=running, model_inputs=model_inputs, delta=delta)
self.resp_queue.put_nowait(step_outputs)
# out might comes from shared memory, need to explicitly delete to release memory in time
@grimoire grimoire marked this pull request as draft March 20, 2026 07:58
@grimoire grimoire marked this pull request as ready for review March 20, 2026 09:16
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants