[NVIDIA] Add config option to use cudnn flash attention #73
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Outdated suggestions cannot be applied.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.
This PR is to allow users to enable the cudnn flash attention. The PR depends on google/praxis#53.
The preliminary results for the GPT3-5B, we can observe ~30% perf improve on 8xH100 GPUs.
With this PR, users can simply set
USE_CUDNN_FLASH_ATTENTION=True
in their config and then the attention part will be replaced with the cudnn flash attention.cc. @nluehr @zhangqiaorjc