TOME-Math: 380K proof-supervision records from 2,118 textbooks lift math benchmarks 16.7 points
_akhaliq · x · 2026-11-13
TOME-Math is a structured textbook corpus for math reasoning, curated from 2,118 university textbooks into 380K proof-supervision records that preserve chapter order and proposition-proof links. Fine-tuning on it lifts average math benchmark performance by 16.7 percentage points.
More from Research
- Amplify deep dive: how model training breeds slop, and what Taste does to fix it — sarahcat21 · 2026-11-13
- Oxford/DeepMind paper surveys 8,234 workers: why basic income can't replace meaning of work — lawhsw · 2026-11-13
- 3rd CoRL Workshop on Dexterous Manipulation Set for Nov 2026, Focuses on Full-Stack Robotics — HaozhiQ · 2026-11-13
- ostris trains H3 image VAE overnight with face-identity and OCR losses, benchmark vs MiniMax — ostrisai · 2026-11-13
- Weight-space editing playbook: build your own harness before trusting LLM edits — viksit · 2026-11-13
- SteerDistill: closed-loop teacher steering keeps sub-4-bit LLM agents within 1 point of BF16 — _akhaliq · 2026-11-13