Repository navigation
Conversation
… scoring Score categorical candidates in groups of four. Read each observed category index once per group and use it to update four separate candidate scores. This reduces repeated scans of the model's observation buffers while preserving each candidate's addition order and mixture normalization.
y0z
marked this pull request as ready for review
October 15, 2026 07:48
Member
Author
|
I was unable to reproduce the performance improvement in my environment (nor did I observe any significant performance degradation). Since the implementation would become somewhat more complex, I am not sure whether this change should be adopted. @c-bata Are you able to reproduce this performance improvement in your environment? |
Contributor
|
I also confirmed that commit 2e0ce80 does not improve performance. While I do not think the concern raised in issue #271 is entirely off the mark, the sample implementation lacked thorough verification and fell short of the quality standard expected for submission. I sincerely apologize for this. I would like to withdraw issue #271. |
This branch has not been deployed
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
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Score categorical TPE candidates in batches of four, reusing each observed category index across candidate scores while preserving per-candidate addition order and mixture normalization. Models containing numerical parameters continue to use per-candidate evaluation.
A local macOS aarch64 benchmark showed essentially unchanged runtime (10.095 s vs. 10.140 s median); the reported speedup remains unconfirmed on this platform.
Closes #271.
Original implementation by @tkng, cherry-picked from 2e0ce80 without code changes, preserving the original commit author.