OPEN-1B claims world's first fully auditable transformer training run
benfielding · x · 2026-09-17
benfielding announced OPEN-1B, billed as the world's first fully auditable transformer training run: logging and auditing every step of AI training and inference to provide proof of training data, recipe, biases, and weights. The author frames it as an answer to AI distrust that, he argues, is manufactured by the labs building AI rather than a technical limitation.
Related event: Gensyn releases open-1b, the first fully auditable open model(9 posts)→
More from Research
- Why mathematicians have a point about AI proofs: models hunt trophies, not exploration — _lewtun · 2026-09-17
- TMLR tightens desk rejections as submissions surge, quizzes authors on their own papers — thegautamkamath · 2026-09-17
- ModernBERT Contributor Confirms: Multi-Head Training With 1K Examples Works Now — alexisgallagher · 2026-09-17
- ModernBERT Contributor: Train Multiple Heads With ~1K Examples in an Afternoon — alexisgallagher · 2026-09-17
- CHOP uses AI to cut children's heart 3D modeling from 4 hours to seconds — kimmonismus · 2026-09-17
- Liouville Goldbach Conjecture Fully Proven and Lean-Verified, Code Released on GitHub — ctjlewis · 2026-09-17