Helsinki-NLP/tatoeba_mt_full

这是一个包含多种语言到英语翻译数据的数据集，每种语言都有对应的训练集和测试集。数据集中的每一条记录都包含源文本、目标文本、源语言和目标语言信息。不同语言的数据集大小和示例数量不同，其中英语到英语的训练集规模最大，包含超过4亿字节和4300万个示例。

This dataset contains translation data from multiple languages to English, with both training and test sets for each language (some languages may only have a training set). Each pair of sentences in the dataset includes the source text, target text, source language, and target language. The size of the dataset and the number of examples vary, with the English to English training set being the largest, containing over 4 billion bytes and 43 million examples.