Non-record: 11L GEPA + 25k Steps + Pure Int6 + Legal TTT (val_bpb=1.0944) - unlimited compute category#644
Open
Christopher-Lee-McClendon wants to merge 1 commit intoopenai:mainfrom
Conversation
- Non-record unlimited-compute submission: val_bpb=1.0944 - 25000-step training (12000 peak-LR + 13000 warmdown) on 4xA100-40GB - Pure int6 per-row quantization with 15-candidate GPTQ-lite + zstd-22 - Legal score-first TTT (SGD, 10 epochs, momentum 0.9): -0.014 BPP gain - Float base 1.1088, artifact 13.75 MiB (14,496,936 bytes total) - Includes model artifact (final_model.int6.ptz) for reproducibility
Christopher-Lee-McClendon
changed the title
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
final_model.int6.ptz) for reproducibilityKey Result
Scaling Law (5 data points, warmdown is the dominant lever)
All three metrics improve monotonically: float base, TTT BPP, and artifact size.
Key Insight: Warmdown Acceleration
The BPP improvement accelerates in the final warmdown steps despite the cosine LR schedule decelerating:
This suggests fine-grained optimization at low LR is disproportionately effective.
Prior Submissions in This Series
Acknowledgments
Builds on techniques from: @signalrush (PR #414, GPTQ-lite/EMA), @jfprincz (PRs #287/#315, XSA/Partial RoPE/LN Scale), @unnir (PR #265, Efficient XSA), raahilshah (PR #162, SmearGate/BigramHash), @aruniyer (PR #86, Int6 QAT), samacqua (LoRA TTT), @abaybektursun (PR #549, LeakyReLU²), and the OpenAI baseline.