Skip to content

Tiny bump: 11L TTT Burst + EMA + GPTQ-lite (val_bpb=1.1232)#445

Open
newjordan wants to merge 7 commits intoopenai:mainfrom
newjordan:submission/ttt-burst-gptq15
Open

Tiny bump: 11L TTT Burst + EMA + GPTQ-lite (val_bpb=1.1232)#445
newjordan wants to merge 7 commits intoopenai:mainfrom
newjordan:submission/ttt-burst-gptq15

Conversation

@newjordan
Copy link

Summary

Test plan

  • Seed 1337: 1.1232 BPB, 15.68 MB, 600s on 8xH100
  • Seed 42: 1.1240 BPB (validates approach, artifact over size)
  • Post-quant int6 roundtrip verified
  • train_gpt.py under 1500 lines (1443)
  • No TTT on validation data

🤖 Generated with Claude Code

Octavian and others added 7 commits March 18, 2026 18:06
Seeds 42 and 2024 in progress.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Seed 1337: 1.12319 BPB, 15.68 MB (qualifying)
Seed 42: 1.12397 BPB, 16.37 MB (over size, validates BPB)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant