XBMU-MC: A Multilingual Parallel Corpus

Each data sample contains text pairs in both source and target languages, where the source language is Chinese and the target languages include Tibetan, Uyghur and Mongolian. Each sample has a uniform structure, including two main fields: instruction and input, and the corresponding output. the instruction field is used to describe the type or requirement of the translation task, the input field contains the original text in the source language, and the output field is the translated text in the target language.To ensure the quality and consistency of the data, each translation pair is manually reviewed and automatically evaluated to ensure alignment accuracy between source and target languages as well as translation accuracy.

The dataset covers a wide range of fields such as culture, science and technology, and society, and each translated text contains different expressions and language structures, which helps to enhance the robustness and generalization ability of the model. The dataset is stored in standard JSON format, which is convenient for subsequent task processing, analysis and model training.