Draft
Conversation
Contributor
There was a problem hiding this comment.
Code Review
This pull request introduces a shared_weights attribute to the vLLM engine across Llama and Vision models to manage weight synchronization and reloading more granularly. Instead of unconditionally removing these calls, the code now uses conditional guards to skip them only when weights are shared. Additionally, it ensures lora_request is correctly passed during generation when weights are shared. The review feedback identifies several instances of redundant logic where multiple checks are used to verify the same state, suggesting simplifications to improve code clarity and consistency.
for more information, see https://pre-commit.ci
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
When using shared weights, we were removing sync_weights or reload_weights calls.
We now want to make it conditional so that people can use GRPOTrainer for non weight shared models as well.
When we detect weight sharing, we don't sync/reload. We use LoRARequest