qz267/maia-chess — explained in plain English
Analysis updated 2026-07-18 · repo last pushed 2021-01-14
Play against a chess engine that makes human-like mistakes at your own skill level instead of an unbeatable AI.
Study how an AI model learns to mimic human decision-making, as a case study in AI alignment research.
Download the pre-trained Maia models to power a teaching tool or chess trainer app.
Train your own human-like chess model from a different set of games using the included training code.
| qz267/maia-chess | 0verflowme/alarm-clock | 0verflowme/seclists | |
|---|---|---|---|
| Language | — | CSS | — |
| Last pushed | 2021-01-14 | 2022-10-03 | 2020-05-03 |
| Maintenance | Dormant | Dormant | Dormant |
| Setup difficulty | moderate | easy | easy |
| Complexity | 3/5 | 2/5 | 1/5 |
| Audience | researcher | vibe coder | ops devops |
Figures from each repo's GitHub metadata at analysis time.
Training your own models requires a dataset of games and machine learning setup, though pre-trained models are ready to download.
Maia is a chess engine trained on millions of real Lichess games to play like a human at a chosen skill level, instead of playing at inhuman perfection like traditional engines.
Dormant — no commits in 2+ years (last push 2021-01-14).
Setup difficulty is rated moderate, with roughly 1h+ to a first successful run.
Mainly researcher.
This repo across BitVibe Labs
double-check against the repo, no cap.