README.md

metadata

license: mit
size_categories:
  - 10M<n<100M
task_categories:
  - question-answering
  - token-classification
pretty_name: Chess Evaluations
dataset_info:
  - config_name: evals_large
    features:
      - name: FEN
        dtype: string
      - name: Evaluation
        dtype: string
    splits:
      - name: train
        num_bytes: 872492457
        num_examples: 12954834
    download_size: 334299450
    dataset_size: 872492457
  - config_name: pretrain_conv
    features:
      - name: id
        dtype: string
      - name: state
        dtype: string
      - name: conversations
        list:
          - name: from
            dtype: string
          - name: value
            dtype: string
    splits:
      - name: train
        num_bytes: 3870520164
        num_examples: 10000000
    download_size: 636940823
    dataset_size: 3870520164
  - config_name: randoms
    features:
      - name: FEN
        dtype: string
      - name: Evaluation
        dtype: string
    splits:
      - name: train
        num_bytes: 71226739
        num_examples: 1000273
    download_size: 18919700
    dataset_size: 71226739
  - config_name: tactics
    features:
      - name: FEN
        dtype: string
      - name: Evaluation
        dtype: string
      - name: Move
        dtype: string
    splits:
      - name: train
        num_bytes: 192267899
        num_examples: 2628219
    download_size: 92596702
    dataset_size: 192267899
configs:
  - config_name: evals_large
    data_files:
      - split: train
        path: evals_large/train-*
  - config_name: pretrain_conv
    data_files:
      - split: train
        path: pretrain_conv/train-*
  - config_name: randoms
    data_files:
      - split: train
        path: randoms/train-*
  - config_name: tactics
    data_files:
      - split: train
        path: tactics/train-*
tags:
  - rl
  - chess
  - reinforcement learning

Chess Evaluations Dataset

This dataset contains chess positions represented in FEN (Forsyth-Edwards Notation) along with their evaluations and next moves for tactical evals. The dataset is divided into three configurations:

1. tactics: Includes chess positions, their evaluations, and the best move in the position.
2. randoms: Contains random chess positions and their evaluations.
3. chess_data: General chess positions with evaluations.

This is an in progress dataset which contains millions of positions with stockfish 11 (depth 22) evaluations. Please help contribute evaluations of the positions to the repo, the original owner of the dataset is r2dev2.

❗❗❗ Updates to the original dataset will be on the version hosted on kaggle.

Dataset Structure

Each configuration can be loaded separately:

• tactics: Columns - FEN, Evaluation, Move
• randoms: Columns - FEN, Evaluation
• chess_data: Columns - FEN, Evaluation

Usage

You can load each configuration using the datasets library:

from datasets import load_dataset

# Load the tactics dataset
tactics_dataset = load_dataset("someshsingh22/chess-evaluations", "tactics")

# Load the randoms dataset
randoms_dataset = load_dataset("someshsingh22/chess-evaluations", "randoms")

Contributing

To get started download a pre-built executable from the releases of chess contributor and run it.

The evaluation should go in eval folder under same name