SKIML-ICL/hoh_10000_filtered
收藏资源简介:
该数据集包含3452个测试样本,每个样本具有多个特征,包括问题ID(qid)、问题文本(question)、冲突时间(conflict_time)、当前时间(current_time)、文档标题(document_title)、文档ID(document_id)、所有过时信息(all_outdated_infos,包括答案、证据和最后修改时间)、上下文(ctxs,包括是否有答案、NLI标签、段落ID、排名、分数、文本和标题)、可回答前缀(answerable_prefix)、冲突答案(conflict_answers)、答案列表(answers)、上下文文本(context)、冲突段落(conflict_passage)、是否为有效冲突段落(is_valid_conflict_passage)、是否有答案(hasanswer)和可回答性(answerable)。数据集主要用于测试,总大小为196489291字节,下载大小为90833789字节。
The dataset contains 3452 test samples, each with multiple features including question ID (qid), question text (question), conflict time (conflict_time), current time (current_time), document title (document_title), document ID (document_id), all outdated information (all_outdated_infos, including answer, evidence, and last modified time), contexts (ctxs, including hasanswer, NLI label, paragraph ID, rank, score, text, and title), answerable prefix (answerable_prefix), conflict answers (conflict_answers), list of answers (answers), context text (context), conflict passage (conflict_passage), whether it is a valid conflict passage (is_valid_conflict_passage), hasanswer (hasanswer), and answerability (answerable). The dataset is primarily for testing, with a total size of 196489291 bytes and a download size of 90833789 bytes.




