[Bugfix] Mamba cache Cuda Graph padding #6214
Merged
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
…ape and not by number of sequences. This is so we pad the mamba cache to the captured CG batch sizes
simon-mo
approved these changes
Jul 8, 2024
tlrmchlsmth
added a commit
to neuralmagic/nm-vllm
that referenced
this pull request
Jul 16, 2024
dtrifiro
pushed a commit
to opendatahub-io/vllm
that referenced
this pull request
Jul 17, 2024
xjpang
pushed a commit
to xjpang/vllm
that referenced
this pull request
Jul 24, 2024
Alvant
pushed a commit
to compressa-ai/vllm
that referenced
this pull request
Oct 26, 2024
Signed-off-by: Alvant <alvasian@yandex.ru>
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Outdated suggestions cannot be applied.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.
The current Jamba implementation doesn't allow use of Cuda Graph with batch sizes that weren't captured. When trying to do that, a RuntimeError is raised with incompatible tensor shapes. This happens because the mamba cache wasn't padded to the Cuda Graph batch size.
This PR fixes this issue, and adds a test to assert it.