Explore high-quality datasets for your AI and machine learning projects.
The dataset consists of chess‑board states and associated move sequences extracted from games downloaded from lichess.org. Each game is parsed into multiple records; each record starts with a FEN string followed by 1‑10 SAN moves. The data are intended for training the ChessRoberta model and have not been filtered, so they may not be optimal for high‑performance chess modelling.
