Manually duplicate both tool and non-tool requests. #978
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This reverts commit fa7e608.
hamishivi
reviewed
Sep 3, 2025
Collaborator
hamishivi
left a comment
hamishivi
left a comment
There was a problem hiding this comment.
Still seems to have some error:
File "/weka/oe-adapt-default/hamishi/pr_review/open-instruct/open_instruct/vllm_utils3.py", line 206, in <lambda>
outputs.sort(key=lambda x: int(x.request_id.split("_")[-1]))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ValueError: invalid literal for int() with base 10: '4-2'
hamishivi
approved these changes
Sep 3, 2025
Collaborator
hamishivi
left a comment
hamishivi
left a comment
There was a problem hiding this comment.
Debug run seems to work!
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
As part of #859, I made a number of major changes. To be a better software engineer, I'm breaking them out into separate PRs (and also because there's a bug in #859 that I can't identify).
This refactors the way we structure our main LLM engine loop to make it ready for inserting individual prompts. This unifies the way we handle requests for tool use and non-tool use. It shouldn't be observable to a caller of LLMRayActor.
Runs:
Single GPU: Beaker
Multi-node: Beaker
Single GPU with tool use: Beaker