feat(vllm-tensorizer): Update vllm-tensorizer cloned repository, build with vllm-flash-attn, other optimizations
#72
Conversation
This file contains 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
vLLM replaced their usages of the regular `flash-attn` library with their own `vllm-flash-attn` fork, which, as of right now, is fairly easy to compile. This change compiles it from source for compatibility with the `ml-containers/torch` base images. [skip ci]
build(vllm-tensorizer): Compile `vllm-flash-attn` from source
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Outdated suggestions cannot be applied.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.
vllm-tensorizerhasn't had updates since vLLM's formal adoption oftensorizermodel loading. An update to build for the most recent commit to vLLM that includes sharded tensorizer support is presented, along with some fixes to successfully build vLLM with recent updates to the source code. These include:tensorizer)cmakexformersversion to0.0.26.post1flash-attn, which is built here from source