Merged
Conversation
Fix MTF performance regression by limiting alphabet to top 4096 frequent repo names. Infrequent repos use fallback encoding (raw indices). - Runtime: 2+ minutes → ~12 seconds (10x faster) - Size: 5,723,601 → 5,784,824 bytes (+61KB, +1.1%) The full MTF with 261K unique repos was O(n*m) = billions of operations. Limited alphabet keeps MTF benefits for frequent repos while avoiding the quadratic blowup for the long tail.
Co-authored-by: Codex <[email protected]> Co-authored-by: OpenCode <[email protected]>
Training data had max delta 251 (fits in u8), but test data has deltas up to 2689+ causing silent truncation and decode failures. Switch to LEB128 varint encoding which handles arbitrary delta sizes. Minimal size impact on training data (+3 bytes: 5,784,824 -> 5,784,827). Tested on 3 random GitHub Archive hours - xinyuzeng ranks agavra#1 on all.
Owner
|
Confirmed over CI/CD, very nicely done @XinyuZeng! |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Thanks @agavra for the golf :)! I learned a lot on how to vibe code on a performance problem through this. Especially after the first closed PR, in this new iteration I fixed some mediocre vibe settings and let the agents self-evolve with observable performance and trackable optimization history.