LianeMarilin/long-context-qa-curated-20

Dataset Card / 数据集卡 Dataset Description / 数据集简介 This public release contains 20 curated samples selected from a 10,000-record long-context QA collection. It targets retrieval over long documents, cross-section evidence synthesis, numerical reasoning, timeline reconstruction, and structured answer evaluation. The public subset contains 15 short-answer questions and 5 multiple-choice questions, balanced across Chinese and English. 本公开版本从 10,000 条长上下文问答数据中精选 20… See the full description on the dataset page: https://huggingface.co/datasets/LianeMarilin/long-context-qa-curated-20.