Trelis/yourbench-ai-tinkerers

该数据集包含多个配置：chunked（分块处理文档）、ingested（文档摄入）、lighteval（轻量级评估）、multi_hop_questions（多跳问题）、single_shot_questions（单次提问）和summarized（文档摘要）。每个配置都有不同的特征，如文档ID、文本、文件名、元数据、问题、答案等。数据集适用于文档处理、问题回答和摘要生成等NLP任务。

The dataset consists of multiple configurations: chunked (document chunking), ingested (document ingestion), lighteval (lightweight evaluation), multi_hop_questions (multi-hop questions), single_shot_questions (single-shot questions), and summarized (document summarization). Each configuration has different features such as document ID, text, filename, metadata, questions, answers, etc. The dataset is suitable for NLP tasks such as document processing, question answering, and summary generation.