h4ng/necai

--- license: mit tags: - chess - ai - games - fen - reinforcement-learning - board-games dataset_info: - config_name: default features: - name: fen dtype: large_string - name: result dtype: large_string - name: white_elo dtype: int64 - name: black_elo dtype: int64 splits: - name: train num_bytes: 12460971 num_examples: 134843 download_size: 2688433 dataset_size: 12460971 - config_name: openings features: [] splits: - name: train num_bytes: 736648 num_examples: 3641 download_size: 199258 dataset_size: 736648 configs: - config_name: default data_files: - split: train path: data/train-* - config_name: openings data_files: - split: train path: openings/train-* - config_name: memoization_white data_files: - split: train path: memoization/white/train-* - config_name: memoization_black data_files: - split: train path: memoization/black/train-* --- # Chess Positions Dataset A large-scale chess dataset derived from other databases to train my ai/ml chess model "necai" ## Source ## Contents - FEN positions extracted from rated games - Player Elo ratings (filtered 2000+) - Game results (White wins / Black wins / Draw) - Opening names - Move evaluations (Stockfish) ## Format - Parquet files (columnar, ML-ready) ## Use Case Training chess neural networks and ML models.