anke01 commited on
Commit
bb9563b
·
verified ·
1 Parent(s): fd565e6

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +63 -0
README.md ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # 维吾尔语多语言词典数据集
2
+
3
+ 维吾尔语-汉语-英语多语言词典数据集,适用于大型语言模型(LLM)微调训练。
4
+
5
+ ## 数据集统计
6
+
7
+ | 数据集 | 条目数 | 大小 | 语言方向 |
8
+ |--------|--------|------|----------|
9
+ | `ug-cn.jsonl` | 3,079,016 | 456 MB | 维吾尔语 ⟷ 汉语 |
10
+ | `ug-en.jsonl` | 685,836 | 108 MB | 维吾尔语 ⟷ 英语 |
11
+ | `en-ug.jsonl` | 705,534 | 112 MB | 英语 ⟷ 维吾尔语 |
12
+ | `ug-ug.jsonl` | 139,920 | 46 MB | 维吾尔语释义 |
13
+ | `cn-cn.jsonl` | 127,646 | 39 MB | 汉语释义 |
14
+
15
+ **总计**: 4,737,952 条(双向)/ 约 236 万唯一对
16
+
17
+ ## 数据格式
18
+
19
+ ```json
20
+ {
21
+ "instruction": "请翻译以下维吾尔语词汇",
22
+ "input": "مەركىزى",
23
+ "output": "中心的,中央的"
24
+ }
25
+ ```
26
+
27
+ 字段说明:
28
+ - `instruction`: 任务指令
29
+ - `input`: 输入文本
30
+ - `output`: 输出文本
31
+
32
+ ## 使用示例
33
+
34
+ ```python
35
+ import json
36
+
37
+ with open('uyghur_data/output/ug-cn.jsonl', 'r', encoding='utf-8') as f:
38
+ for line in f:
39
+ data = json.loads(line)
40
+ print(f"输入: {data['input']}")
41
+ print(f"输出: {data['output']}")
42
+ ```
43
+
44
+ ## 数据来源
45
+
46
+ - Gheyret 词典 (~37万条)
47
+ - Tilkan 词典数据库 (150MB)
48
+ - Bilkan 词典 (5方向)
49
+ - Online Dictionary (~33万条)
50
+
51
+ ## 文件位置
52
+
53
+ ```
54
+ uyghur_data/output/
55
+ ├── ug-cn.jsonl # 维汉数据集
56
+ ├── ug-en.jsonl # 维英数据集
57
+ ├── en-ug.jsonl # 英维数据集
58
+ ├── ug-ug.jsonl # 维维释义
59
+ └── cn-cn.jsonl # 汉汉释义
60
+ ```
61
+
62
+ ---
63
+ **版本**: 1.0 | **更新**: 2025-03-01