board-grounding
收藏资源简介:
该数据集是一个包含约104.3万个样本的大规模结构化数据集,仅提供训练集。每个样本包含五个字段:fen(字符串类型,可能表示棋局状态,如国际象棋的FEN格式)、move(字符串类型,可能表示棋步)、question(字符串类型)、answer(字符串类型)和task_type(字符串类型)。字段设计表明数据集可能用于棋类游戏相关的问答或任务执行场景,例如基于棋局状态生成问题、预测棋步或进行多任务学习,但具体任务定义和内容需进一步通过数据样本确认。数据以纯文本形式存储,总大小约为136.8 MB。
This dataset is a large-scale structured dataset containing approximately 1.043 million samples, providing only a training set. Each sample includes five fields: fen (string type, possibly representing board state, such as FEN format in chess), move (string type, possibly representing a move), question (string type), answer (string type), and task_type (string type). The field design suggests that the dataset may be used for chess-related question-answering or task execution scenarios, such as generating questions based on board state, predicting moves, or performing multi-task learning, but the specific task definitions and content require further confirmation through data samples. The data is stored in plain text format, with a total size of approximately 136.8 MB.
数据集概述:board-grounding
- 数据集名称:board-grounding
- 数据集地址:https://huggingface.co/datasets/michaelc0des/board-grounding
数据集特征
数据集包含以下五个字段,均为字符串类型:
- fen:国际象棋棋盘FEN表示法
- move:走棋动作
- question:提问内容
- answer:回答内容
- task_type:任务类型
数据集划分
- 训练集(train):包含 1,042,933 个样本,占用约 130.5 MB 存储空间
数据集大小
- 下载大小:约 45.76 MB
- 数据集总大小:约 130.5 MB
配置文件
- 配置名称:default
- 数据文件路径:
data/train-*(仅包含训练集划分)




