[serve.llm][bugfix] Fix the wrong device_capability issue in vllm on quantized models. #51007
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
…text for compatibility of the devices Signed-off-by: Kourosh Hakhamaneshi <kourosh@anyscale.com>
comaniac
approved these changes
Mar 1, 2025
…kha/ray into kh/fix-serve-quantized Signed-off-by: Kourosh Hakhamaneshi <kourosh@anyscale.com>
|
The doc issue seems unrelated. Merging ... |
Michaelhess17
pushed a commit
to Michaelhess17/ray
that referenced
this pull request
Mar 3, 2025
…quantized models. (ray-project#51007) Fixes a bug where get_device_capability() was getting cached based on the wrong environment variables. `vllm` assumes that upon import the `CUDA_AVAILABLE_DEVICES` will have their final values and will cache some attributes like cuda device compatibility. If later we use ray actors to serialize the modules over the cache apparently also gets serialized with the wrong values. This PR clears the cache before creating engine so the values will be recomputed based on the right env variables. --------- Signed-off-by: Kourosh Hakhamaneshi <kourosh@anyscale.com>
xsuler
pushed a commit
to antgroup/ant-ray
that referenced
this pull request
Mar 4, 2025
…quantized models. (ray-project#51007) Fixes a bug where get_device_capability() was getting cached based on the wrong environment variables. `vllm` assumes that upon import the `CUDA_AVAILABLE_DEVICES` will have their final values and will cache some attributes like cuda device compatibility. If later we use ray actors to serialize the modules over the cache apparently also gets serialized with the wrong values. This PR clears the cache before creating engine so the values will be recomputed based on the right env variables. --------- Signed-off-by: Kourosh Hakhamaneshi <kourosh@anyscale.com>
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.
Fixes a bug where get_device_capability() was getting cached based on the wrong environment variables.
vllmassumes that upon import theCUDA_AVAILABLE_DEVICESwill have their final values and will cache some attributes like cuda device compatibility.If later we use ray actors to serialize the modules over the cache apparently also gets serialized with the wrong values. This PR clears the cache before creating engine so the values will be recomputed based on the right env variables.