Suryanshg/SUPER-NATURALINSTRUCTIONS-english-filtered-100x-augmented
收藏数据链接:
官方服务:
资源简介:
该数据集包含多个任务,每个样本有任务名称、指令、输入语言、输出语言、指令语言、类别、输入文本、输出文本、总token数以及精简指令列表。训练集中有2,790,891个样本,测试集中有458,488个样本。数据集覆盖多种语言和任务类型,适用于指令微调、多语言翻译等场景。
This dataset contains multiple tasks, each sample includes task name, instruction, input language, output language, instruction language, categories, input text, output text, total tokens, and reduced instructions list. The training set has 2,790,891 examples and the test set has 458,488 examples. It covers multiple languages and task types, suitable for instruction fine-tuning, multilingual translation, etc.
提供机构:
Suryanshg


