Conversation
|
MLCommons CLA bot All contributors have signed the MLCommons CLA ✍️ ✅ |
hvagadia
commented
Sep 14, 2026
| | Metric | Kimi K3 | Qwen3.6-35B-A3B | DSV4 | | ||
| | ------------------ | -----------------------: | -----------------------: | ---: | | ||
| | Inline accuracy | `>= 58.32%` (`58.9%`) | `>= 55.86%` (`56.43%`) | TBD | | ||
| | OSL per-turn mean | `425-520` tokens (`472`) | `344-422` tokens (`383`) | TBD | |
Collaborator
There was a problem hiding this comment.
Please give me some time to verify. ETA 2 days.
Collaborator
Author
There was a problem hiding this comment.
Sure, we will keep this open. Its only a 10 point change for Q3, so most likely your runs will fall in the range.
Collaborator
There was a problem hiding this comment.
We see ~380 from our end and are okay with the Qwen3.6 range.
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Due to a bug in mean OSL counting described in #500, we need to redo the mean OSL computation. This PR updates the Kimi K3 reference mean to 472 tokens with a range of 425–520 tokens, and the Qwen3.6-35B-A3B reference mean to 383 tokens with a range of 344–422 tokens.