[Text Generation] Optimize the slow update method in the KVCacheDecoder
#1190
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
bfineran
reviewed
Aug 21, 2023
rahul-tuli
reviewed
Aug 22, 2023
rahul-tuli
reviewed
Aug 22, 2023
rahul-tuli
approved these changes
Aug 23, 2023
bfineran
approved these changes
Aug 24, 2023
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Outdated suggestions cannot be applied.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.
As reported by @mgoin and investigated by myself, the
updatemethod in the KVCacheDecoder is very slow.Profiling has shown that this is due to the repeated use of the
numpy.deletefunction:This discussion:
https://stackoverflow.com/questions/30399534/shift-elements-in-a-numpy-array
hints that the most elegant and quite efficient replacement for
numpy.deletewould be slicing the arrays. This is the change that this PR introduces.Short benchmarking numbers: