AttentionMetadata Preparation for Encoder-only Models #145
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
Signed-off-by: slokesha <[email protected]>
Signed-off-by: slokesha <[email protected]>
Signed-off-by: slokesha <[email protected]>
Signed-off-by: slokesha <[email protected]>
Signed-off-by: slokesha <[email protected]>
Signed-off-by: slokesha <[email protected]>
Signed-off-by: slokesha <[email protected]>
slokesha
requested review from
kzawora-intel,
xuechendi,
mswiniarsk and
adobrzyn
as code owners
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Outdated suggestions cannot be applied.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.
Upstream-Vllm implements an encoder-only attention layer that initializes its own AttentionMetadataBuilder
On GPU, the AttentionMetadataBuilder is the standard way to create attention metadata.
On Gaudi, however, attention metadata is created through the make_prefill_metadata function. This causes encoder-only models to bypass the builder logic and fail in scenarios where no KV cache is used.
This PR introduces encoder-only attention metadata support for Gaudi by aligning with the upstream behavior, while handling Gaudi-specific paths. And also Handles edge cases where no KV cache is present
Dependency on Upstream PR - vllm-project/vllm#24612