Skip to content

fix sglang_patch. fix apptainers action execution#29

Open
skzhang1 wants to merge 2 commits into
stablefrom
fix_sglang_patch
Open

fix sglang_patch. fix apptainers action execution#29
skzhang1 wants to merge 2 commits into
stablefrom
fix_sglang_patch

Conversation

@skzhang1
Copy link
Copy Markdown
Contributor

  1. fix sglang patch
  2. fix apptainer in cluster

Copy link
Copy Markdown

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 adjusts Apptainer runtime execution to support direct apptainer exec mode for cluster environments and updates the SGLang patch script to persist canonical prompt token IDs after tokenization.

Changes:

  • Adds optional direct-exec mode for Apptainer, including shared overlay setup and command prefix reuse.
  • Updates the SGLang patch to store input_token_ids on request state and emit them through meta_info.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File Description
src/polar/runtime/apptainer.py Adds direct-exec support and centralizes Apptainer exec argument construction.
scripts/patch/patch_sglang.sh Updates tokenizer manager patching to persist and expose prompt token IDs from request state.

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

Comment on lines +29 to +31
self._direct_exec = bool(os.environ.get("POLAR_APPTAINER_DIRECT_EXEC")) or bool(
spec.kwargs.get("direct_exec", False)
)
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