Record: 11L Muon TTT + Entropy-Adaptive Epochs (8×H100) — val_bpb 1.1179 (3-seed mean)#999
Open
aamodbhatt wants to merge 1 commit intoopenai:mainfrom
Open
Conversation
…179 (3-seed mean) Two novel TTT innovations: (1) Muon-style Newton-Schulz orthogonalized updates replace SGD in the TTT loop; (2) entropy-adaptive 2/3/4 epochs per chunk based on globally-synced chunk NLL. 3-seed mean 1.1179, std 0.0002. All under 16MB/600s.
aamodbhatt
force-pushed
the
record-2026-03-28-muon-ttt-entropy-adaptive
branch
from
bffd426 to
f219235
Compare
aryanbhosale
added a commit
to aryanbhosale/parameter-golf
that referenced
this pull request
Mar 28, 2026
slope 0.75 + LR 0.027 + warmdown 3700 (PR openai#977) No SWA with QAT (PR openai#989) QAT from 50% + range fix [-31,31] mHC 22-param residual mixing (PR openai#928) VE128 + no gated_attn + no value_residual (PR openai#549) LZMA preset 7 compression (PR openai#999) Muon TTT with NS3 (PR openai#999) Entropy-adaptive TTT epochs 2/3/4 (PR openai#999) Per-layer TTT LR (PR openai#995) TTT momentum 0.95 (PR openai#995)
eamon831
added a commit
to eamon831/parameter-golf
that referenced
this pull request
Mar 28, 2026
CLAUDE.md: Complete project state for cross-session continuity - Leaderboard intel (verified SOTA + unverified PRs openai#1006, openai#999, openai#831) - 8192 vocab analysis (doesn't fit — only 9,994 bytes headroom) - Three planned improvements with code status - Environment setup instructions (Mac MLX + RunPod H100) - Codebase layout and git remotes experiments.md: 4 planned experiments with commands + success criteria Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
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
Two novel TTT innovations on the SOTA base stack (PR #399 + PR #414 + PR #461): Muon-style Newton-Schulz orthogonalized updates replace SGD in the TTT loop, and entropy-adaptive epoch selection concentrates adaptation budget on harder content. Beats current SOTA (1.1194) with a 3-seed mean of 1.1179.
Run Results (3 seeds)
legal_ttt_exact val_bpblegal_ttt_exact val_lossval_bpb1.117650301.887100721.1366599.1s477.9s15,944,410 bytes1.118129291.887909471.1371599.1s485.3s15,873,826 bytes1.117899341.887521211.1367599.1s479.2s15,879,042 bytes1.11789Method Notes
NUM_LAYERS=11,BIGRAM_VOCAB_SIZE=1536,XSA_LAST_N=4TTT_ENABLED=1, score-first pathTTT_MUON=1— Newton-Schulz orthogonalized updates in TTT loop (NS steps=3)TTT_ENTROPY_ADAPT=1— entropy-adaptive 2/3/4 epochs per chunk (H_HIGH=2.1, H_LOW=1.75)TTT_LR=0.002,TTT_EPOCHS=3,TTT_CHUNK_TOKENS=32768NGRAM_EVAL_ENABLED=0NGRAM_TWO_PASS_ENABLED=0NGRAM_FULL_RESCORE=0EMA_ENABLED=1,SWA_ENABLED=1,LATE_QAT=1,VE_ENABLED=1WARMDOWN_ITERS=3500,MAX_WALLCLOCK_SECONDS=599Submission Checklist
records/track_10min_16mb/README.md,submission.json,train_gpt.py, and train logs (3 seeds)