bijode8658-kudimi/sft-1

Sft 1数据集包含多个配置，如chunked、ingested、multi_hop_questions、prepared_lighteval、single_shot_questions和summarized。每个配置具有不同的特征和分割，其中train是所有配置共有的分割。特征包括文档ID、文本、文件名、元数据、摘要、块、问题、答案等。该数据集适用于自然语言处理任务，特别是涉及文档摘要、问答和多跳推理的任务。The Sft 1 dataset includes multiple configurations such as chunked, ingested, multi_hop_questions, prepared_lighteval, single_shot_questions, and summarized. Each configuration has distinct features and splits, with train being the common split across all configurations. Features include document IDs, text, filenames, metadata, summaries, chunks, questions, answers, and more. The dataset is designed for NLP tasks, particularly those involving document summarization, question answering, and multi-hop reasoning.