ChessLab
A chess learner that starts from random weights and is never told anything about chess.
The learner gets the rules of chess and nothing else: no openings, no piece values, no engine scores as targets. It plays itself, trains a candidate on its own games, and the candidate only becomes the new champion if it beats the old one over enough games to prove it.
Everything is measured against something the learner cannot author. Fixed opponents that never drift, frozen mate positions, and its own past generations, archived so they can be replayed.
The interesting findings are the ones that contradicted the plan. A network sixteen times bigger learned nothing more. Half the time went to Python’s garbage collector. And the search was the ceiling, not the model: the same network solves a third more mates when it is allowed to think longer.
- Generation 27 measured against chess-gpt v4
- Board cost cut 2.2x, mate probe hits 87
- Learner starts sparring against outside opponents
- Gap to chess-gpt closes from 382 to 108 Elo
- Fixed reference ladder gives first absolute score
- Gumbel AlphaZero added as search mode
- Optional Rust move generator speeds up play
- Transformer brain scaffolded for searchless research
No posts from this project yet — they'll gather here as they publish.