Tag
A chess coaching tool that uses an LLM to explain moves in natural language like a grandmaster, replacing raw engine evaluations with contextual coaching narratives. It analyzes user games from Chess.com and Lichess with local Stockfish, detects recurring mistakes, and offers personalized chat and spaced repetition for blunders.
The article models chess as a concurrent system and derives state and transition invariants, demonstrating formal verification techniques using TLA+.
Nicholas Carlini's project implements a 2-ply minimax chess engine using 84,688 regular expressions, executed sequentially to play valid chess moves. The post explains the design of a regular expression computer that interprets instructions.
Trained transformer-based chess models for rating buckets from 800 to 2500+, predicting moves, thinking time, and outcome. Achieves strong accuracy with only 9M parameters, and includes a novel thinking-time prediction component.