kvcache: skip multi-turn cache reads in decode-only mode#284
Open
LouisDDN wants to merge 1 commit intomlcommons:mainfrom
Open
kvcache: skip multi-turn cache reads in decode-only mode#284LouisDDN wants to merge 1 commit intomlcommons:mainfrom
LouisDDN wants to merge 1 commit intomlcommons:mainfrom
Conversation
|
MLCommons CLA bot All contributors have signed the MLCommons CLA ✍️ ✅ |
Skip multi-turn conversation cache reads when running in decode-only mode, since previous turn cache entries are never written in this mode. This change: - Prevents wasteful cache lookups that always miss - Cleans up multi_turn_cache_misses metrics (no longer polluted) - Improves code correctness by not checking cache that was never written The multi-turn cache read (Step 2) is now guarded by the same `if not self.decode_only` check as the prefill write (Step 3), since both operations are meaningless in decode-only mode. Performance impact: negligible (<0.01%), but improves code clarity.
LouisDDN
force-pushed
the
ld/skip-multiturn-decode-only
branch
from
984ee49 to
f45de66
Compare
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Skip multi-turn conversation cache reads when running in decode-only mode, since previous turn cache entries are never written in this mode.
This change:
The multi-turn cache read (Step 2) is now guarded by the same
if not self.decode_onlycheck as the prefill write (Step 3), since both operations are meaningless in decode-only mode.Performance impact: negligible (<0.01%), but improves code clarity.