Team Ai
Datasetpublic

groloch/lc0_games

[!WARNING] Due to a parsing error, the castling moves in the chess960 games are encoded as (king starting square) -> (king target square) and are not compatible with python-chess. A workaround is shown in the Use section. LC0 Games This dataset contains 170m games played by Leela Chess Zero against itself. The moves are written in uci format. The games were sourced directly from the lc0 official data. This dataset contain some chess960 games, for which the start FEN is… See the full description on the dataset page: https://huggingface.co/datasets/groloch/lc0_games.

sourceHugging Faceodblupdated 4d agoView on Hugging Face
0likes601downloads
Dataset Card
[!WARNING] Due to a parsing error, the castling moves in the chess960 games are encoded as (king starting square) -> (king target square) and are not compatible with python-chess. A workaround is shown in the Use section.

LC0 Games

This dataset contains 170m games played by Leela Chess Zero against itself. The moves are written in uci format.

The games were sourced directly from the lc0 official data.

This dataset contain some chess960 games, for which the start FEN is specified.

For each name, the lc0 version (run id) playing are specified. You can find more details on the specific runs there.

[!NOTE] Some noise was added to the move selection during selfplay to ensure that every game is different, therefore not every move played is the best move found by lc0. The level of the games is still far stronger than any human.

Use

To use this dataset, copy the following snippet:

python
from datasets import load_dataset

dataset = load_dataset(
  'groloch/lc0_games',
  streaming=True,
  split='train'
)
dataset = dataset.filter(
  lambda x: x['variant'] == 'normal'
) # workaround castling parsing issue