Decoupled supervision: equal-weight move prediction, outcomes only supervise value
andreisavu · x · 2026-11-08
Parts 9-10 of the chess transformer thread explain the supervision split:
- Ordinary next-move targets are trained with equal weights, independent of game outcomes, keeping result signals out of policy learning.
- Game outcomes supervise only the value head, decoupling "how to move" from "how good is the position."
More from Research
- The meta-question as agent swarms scale: can models beat gradient descent at its own game? — JensHonack · 2026-11-08
- Why can LLMs solve PhD-level math but still lose at chess? Reddit asks — MesloubFC · 2026-11-08
- WebHarbor launches: 153 mirrored real websites and 3,518 tasks for GUI web agents — HuaxiuYaoML · 2026-11-08
- METR Progress per R&D Dollar Fell 63% Even as Yearly Gains Doubled — sebkrier · 2026-11-08
- GenrobotAI's DAS Dex 2.0 captures 21-DoF hand motion plus 980 tactile points for robot learning — CyberRobooo · 2026-11-08
- fordycelab proposes benchmarking binder-design algorithms with Kd measurements on 24,000-VHH libraries — anshulkundaje · 2026-11-08