pszemraj/infinity-instruct-7m-T2T_en

该数据集名为infinity-instruct 7m，主要用于文本到文本的转换任务，所有内容均为英文。数据集包含多个配置，每个配置都有id、source、instruction和response等特征。数据集经过了多种过滤和处理步骤，包括对话长度过滤、语言检测过滤、指令-响应提取、拒绝响应过滤、语言重新检查和词数过滤等。数据来源包括OpenHermes-2.5、flan、MetaMath等多个来源，每个来源的数据量在文件中有所统计。

The dataset named infinity-instruct 7m is primarily used for text-to-text conversion tasks, with all content in English. The dataset includes multiple configurations, each with features such as id, source, instruction, and response. The dataset has undergone various filtering and processing steps, including conversation length filtering, language detection filtering, instruction-response extraction, refusal response filtering, language re-checking, and word count filtering. Data sources include OpenHermes-2.5, flan, MetaMath, and others, with the data volume from each source statistically detailed in the file.