Datasets:
ArXiv:
License:
| license: mit | |
| task_categories: | |
| - text-classification | |
| - question-answering | |
| - translation | |
| - summarization | |
| language: | |
| - bo | |
| - ug | |
| - kk | |
| - mn | |
| # MiLiC-Eval | |
| MiLiC-Eval is an NLP evaluation suite for **Mi**nority **L**anguages **i**n **C**hina, covering Tibetan (bo), Uyghur (ug), Kazakh (kk, in the Kazakh Arabic script), and Mongolian (mn, in the traditional Mongolian script). | |
| **Paper:** [MiLiC-Eval: Benchmarking Multilingual LLMs for China's Minority Languages](https://arxiv.org/abs/2503.01150) | |
| ## Statistics | |
| ### Tasks | |
| Currently, MiLiC-Eval consists of 9 tasks and 4 languages, with a total of 24K instances. The statistics of each task are shown in the following table. | |
| | Task | Size | Metric | Languages | | |
| | --- | --- | --- | --- | | |
| | Vocabulary Understanding | 1,000/lang | Accuracy | bo, ug, kk, mn | | |
| | Topic Classification (Sentence) | 492/lang | Accuracy | bo, ug, kk, mn, zh, en | | |
| | Topic Classification (Passage) | 600/lang | Accuracy | bo, ug, kk, mn | | |
| | Reading Comprehension | 250/lang | Accuracy | bo, ug, kk, mn, zh, en | | |
| | Response Selection | 507/lang | Accuracy | bo, ug, kk, mn, zh, en | | |
| | Title Generation | 1,000/lang | ROUGE-L | bo, ug, kk, mn | | |
| | Machine Translation (Article) | 1,012/lang | chrF++ | bo, ug, kk, mn, zh, en | | |
| | Machine Translation (Dialogue) | 773/lang | chrF++ | bo, ug, kk, mn, zh, en | | |
| | Math Reasoning | 250/lang | Accuracy | bo, ug, kk, mn, zh, en | | |
| ### Data Splits | |
| For each task, we provide a data split, including training, development, and test sets. | |
| The training sets are small and used for in-context learning. For each task, we provide three training sets sampled with different seeds, to reduce the impact of randomness during prompting. | |
| The development sets are used for hyperparameter tuning. The test sets are used for evaluation. | |
| For each language, the data split is shown in the following table. | |
| | Task | Train | Dev | Test | | |
| | --- | --- | --- | --- | | |
| | Vocabulary Understanding | 20 * 3 | 40 | 900 | | |
| | Topic Classification (Sentence) | 10 * 3 | 30 | 432 | | |
| | Topic Classification (Passage) | 16 * 3 | 48 | 504 | | |
| | Reading Comprehension | 10 * 3 | 20 | 200 | | |
| | Response Selection | 20 * 3 | 40 | 407 | | |
| | Title Generation | 20 * 3 | 40 | 900 | | |
| | Machine Translation (Article) | 20 * 3 | 40 | 912 | | |
| | Machine Translation (Dialogue) | 20 * 3 | 40 | 673 | | |
| | Math Reasoning | 10 * 3 | 20 | 200 | | |
| ## Usage | |
| See steps of using the data on [GitHub](https://github.com/luciusssss/MiLiC-Eval) | |
| ### Pretraining Corpus | |
| Current LLMs have limited performance on minority languages due to the lack of pretraining data. | |
| We provide a pretraining corpus, MC^2 for the four languages in MiLiC-Eval. | |
| The corpus can be downloaded from [Hugging Face](https://huggingface.co/datasets/pkupie/mc2_corpus). | |
| You can read the details of the corpus in our paper [MC^2: Towards Transparent and Culturally-Aware NLP for Minority Languages in China](https://aclanthology.org/2024.acl-long.479.pdf) (ACL 2024). | |
| ## Citation | |
| If you use MiLiC-Eval in your research, please cite our GitHub repository: | |
| ```bibtex | |
| @article{zhang2025milic, | |
| title={MiLiC-Eval: Benchmarking Multilingual LLMs for China's Minority Languages}, | |
| author={Zhang, Chen and Tao, Mingxu and Liao, Zhiyuan and Feng, Yansong }, | |
| journal={arXiv preprint arXiv:2503.01150}, | |
| year={2025}, | |
| url={https://arxiv.org/abs/2503.01150}, | |
| } | |
| ``` | |