Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,42 @@
|
|
1 |
-
---
|
2 |
-
license: apache-2.0
|
3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: apache-2.0
|
3 |
+
datasets:
|
4 |
+
- CMLI-NLP/CUTE-Datasets
|
5 |
+
language:
|
6 |
+
- bo
|
7 |
+
- ug
|
8 |
+
- zh
|
9 |
+
- en
|
10 |
+
base_model:
|
11 |
+
- meta-llama/Llama-2-7b-hf
|
12 |
+
---
|
13 |
+
|
14 |
+
# CUTE-Llama
|
15 |
+
|
16 |
+
CUTE-Llama 是一个基于 Llama2-7B 的多语言模型,专门针对中文、维吾尔语和藏语进行了优化。该模型是 [CUTE (Chinese, Uyghur, Tibetan, English) Dataset](https://huggingface.co/datasets/CMLI-NLP/CUTE-Datasets) 项目的一部分,旨在增强低资源语言的跨语言知识迁移能力。
|
17 |
+
|
18 |
+
## 模型版本
|
19 |
+
|
20 |
+
我们提供两个版本的模型:
|
21 |
+
- **CUTE-Llama-Parallel**: 使用平行语料训练的版本
|
22 |
+
- **CUTE-Llama-Non-Parallel**: 使用非平行语料训练的版本
|
23 |
+
|
24 |
+
## 相关链接
|
25 |
+
|
26 |
+
- 📝 [论文](https://aclanthology.org/2025.coling-main.670/)
|
27 |
+
- 📊 [数据集](https://huggingface.co/datasets/CMLI-NLP/CUTE-Datasets)
|
28 |
+
- 📂 [GitHub](https://github.com/CMLI-NLP/CUTE)
|
29 |
+
|
30 |
+
## 引用
|
31 |
+
|
32 |
+
如果您使用了我们的模型,请引用我们的论文:
|
33 |
+
|
34 |
+
```bibtex
|
35 |
+
@inproceedings{zhuang2025cute,
|
36 |
+
title={CUTE: A Multilingual Dataset for Enhancing Cross-Lingual Knowledge Transfer in Low-Resource Languages},
|
37 |
+
author={Zhuang, Wenhao and Sun, Yuan},
|
38 |
+
booktitle={Proceedings of the 31st International Conference on Computational Linguistics},
|
39 |
+
pages={10037--10046},
|
40 |
+
year={2025}
|
41 |
+
}
|
42 |
+
```
|