Add GQA attention in experimental/gen_ai #2504
Closed
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
|
This pull request was exported from Phabricator. Differential Revision: D56110657 |
✅ Deploy Preview for pytorch-fbgemm-docs ready!
To edit notification comments on pull requests, go to your Netlify site configuration. |
sryap
added a commit
to sryap/FBGEMM
that referenced
this pull request
Apr 16, 2024
Summary: This diff open sources `mqa_attn_splitk`, which is a multi-query attention operator for LLM inference for the decoding phase. The operator supports the following: - BF16 input query and output types - BF16/INT4 KV cache types - 16,384 max context length - Fixed head dimension of 128 - Arbitrary query head size The INT4 KV `mqa_attn_splitk` is ~1.7x faster than the BF16 KV counterpart. Differential Revision: D56110657
|
This pull request was exported from Phabricator. Differential Revision: D56110657 |
sryap
added a commit
to sryap/FBGEMM
that referenced
this pull request
Apr 16, 2024
Summary: This diff open sources `mqa_attn_splitk`, which is a multi-query attention operator for LLM inference for the decoding phase. The operator supports the following: - BF16 input query and output types - BF16/INT4 KV cache types - 16,384 max context length - Fixed head dimension of 128 - Arbitrary query head size The INT4 KV `mqa_attn_splitk` is ~1.7x faster than the BF16 KV counterpart. Differential Revision: D56110657
|
This pull request was exported from Phabricator. Differential Revision: D56110657 |
sryap
added a commit
to sryap/FBGEMM
that referenced
this pull request
Apr 16, 2024
Summary: This diff open sources `mqa_attn_splitk`, which is a multi-query attention operator for LLM inference for the decoding phase. The operator supports the following: - BF16 input query and output types - BF16/INT4 KV cache types - 16,384 max context length - Fixed head dimension of 128 - Arbitrary query head size The INT4 KV `mqa_attn_splitk` is ~1.7x faster than the BF16 KV counterpart. Differential Revision: D56110657
|
This pull request was exported from Phabricator. Differential Revision: D56110657 |
sryap
added a commit
to sryap/FBGEMM
that referenced
this pull request
Apr 16, 2024
Summary: This diff open sources `mqa_attn_splitk`, which is a multi-query attention operator for LLM inference for the decoding phase. The operator supports the following: - BF16 input query and output types - BF16/INT4 KV cache types - 16,384 max context length - Fixed head dimension of 128 - Arbitrary query head size The INT4 KV `mqa_attn_splitk` is ~1.7x faster than the BF16 KV counterpart. Differential Revision: D56110657
|
This pull request was exported from Phabricator. Differential Revision: D56110657 |
sryap
added a commit
to sryap/FBGEMM
that referenced
this pull request
Apr 16, 2024
Summary: This diff open sources `mqa_attn_splitk`, which is a multi-query attention operator for LLM inference for the decoding phase. The operator supports the following: - BF16 input query and output types - BF16/INT4 KV cache types - 16,384 max context length - Fixed head dimension of 128 - Arbitrary query head size The INT4 KV `mqa_attn_splitk` is ~1.7x faster than the BF16 KV counterpart. Differential Revision: D56110657
|
This pull request was exported from Phabricator. Differential Revision: D56110657 |
Summary: This diff open sources `gqa_attn_splitk`, which is a grouped-query attention operator for LLM inference for the decoding phase. The operator supports the following: - BF16 input query and output types - BF16/INT4 KV cache types - 16,384 max context length - Fixed head dimension of 128 - Arbitrary number of query heads The INT4 KV `mqa_attn_splitk` is ~1.7x faster than the BF16 KV counterpart. Reviewed By: jianyuh, amylittleyang Differential Revision: D56110657
sryap
added a commit
to sryap/FBGEMM
that referenced
this pull request
Apr 17, 2024
Summary: This diff open sources `gqa_attn_splitk`, which is a grouped-query attention operator for LLM inference for the decoding phase. The operator supports the following: - BF16 input query and output types - BF16/INT4 KV cache types - 16,384 max context length - Fixed head dimension of 128 - Arbitrary number of query heads The INT4 KV `mqa_attn_splitk` is ~1.7x faster than the BF16 KV counterpart. Reviewed By: jianyuh, amylittleyang Differential Revision: D56110657
|
This pull request was exported from Phabricator. Differential Revision: D56110657 |
|
This pull request was exported from Phabricator. Differential Revision: D56110657 |
sryap
changed the title
|
This pull request has been merged in e839246. |
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.
Summary:
This diff open sources
gqa_attn_splitk, which is a grouped-queryattention operator for LLM inference for the decoding phase. The
operator supports the following:
The INT4 KV
gqa_attn_splitkis ~1.7x faster than the BF16 KVcounterpart.
Differential Revision: D56110657