Fix for KeyError on Loading LLaMA #1978
Merged
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
allenhaozi
added a commit
to allenhaozi/vllm
that referenced
this pull request
Dec 8, 2023
hongxiayang
pushed a commit
to hongxiayang/vllm
that referenced
this pull request
Feb 13, 2024
This was referenced Apr 2, 2024
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Outdated suggestions cannot be applied.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.
This PR addresses the issue where loading LLaMA model parameters into vLLM results in a KeyError due to the presence of
rotary_emb.cos_cachedandrotary_emb.sin_cached. These cached values from LLaMA are not required or directly managed by vLLM, leading to compatibility issues.Related issue #1977
Changes Made:
rotary_emb.cos_cachedandrotary_emb.sin_cachedduring the loading process.