Add gemma3-opus-distill post#3377
Open
vicArc wants to merge 1 commit intohuggingface:mainfrom
Open
Conversation
A negative-results writeup on QLoRA-distilling Claude Opus 4.6 reasoning traces into Gemma 3 4B on a single RTX 3050 — fine-tune trains cleanly but regresses on MATH-500 (-5pp) and GSM8K (-15pp), with postmortem on evaluation format mismatch and capability narrowing.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
A community/guest post: a negative-results postmortem on QLoRA-distilling Claude Opus 4.6 reasoning traces into Gemma 3 4B on a single RTX 3050. The fine-tune trains cleanly (loss 2.67 → 1.01) but regresses against the base on MATH-500 (-5.0pp) and GSM8K (-15.0pp). The post explains the two contributing causes (eval format mismatch, capability narrowing on a narrow 1.9k-example dataset), what I'd do differently, and why publishing measured negative results is more useful than the usual celebratory fine-tune writeup.
Preparing the Article
_blog.ymlassets/gemma3-opus-distill/thumbnail.jpg)gemma3-opus-distill)Viesar, guest)Notes