File size: 16,522 Bytes
cc4ce6b
f70f39d
 
 
 
 
 
cc4ce6b
f70f39d
 
 
 
 
cc2acf9
f70f39d
 
 
 
 
 
cc2acf9
 
cc4ce6b
cc2acf9
f70f39d
 
 
 
 
 
 
 
 
 
36cd788
 
 
f70f39d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
36cd788
f70f39d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
cc2acf9
 
f70f39d
 
 
 
 
 
36cd788
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
f70f39d
 
 
cc2acf9
 
f70f39d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
---
annotations_creators:
- shibing624
language_creators:
- shibing624
language:
- zh
license: cc-by-4.0
multilinguality:
- monolingual
size_categories:
- 1M<n<10M
source_datasets:
- https://github.com/shibing624/text2vec
task_categories:
- text-classification
task_ids:
- natural-language-inference
- semantic-similarity-scoring
- text-scoring
paperswithcode_id: nli
pretty_name: Chinese Natural Language Inference
---
# Dataset Card for nli-zh-all

## Dataset Description
- **Repository:** [Chinese NLI dataset](https://github.com/shibing624/text2vec)
- **Dataset:** [zh NLI](https://huggingface.co/datasets/shibing624/nli-zh-all)
- **Size of downloaded dataset files:** 4.7 GB
- **Total amount of disk used:** 4.7 GB
### Dataset Summary

中文自然语言推理(NLI)数据合集(nli-zh-all)

整合了文本推理,相似,摘要,问答,指令微调等任务的820万高质量数据,并转化为匹配格式数据集。


### Supported Tasks and Leaderboards

Supported Tasks: 支持中文文本匹配任务,文本相似度计算等相关任务。

中文匹配任务的结果目前在顶会paper上出现较少,我罗列一个我自己训练的结果:

**Leaderboard:** [NLI_zh leaderboard](https://github.com/shibing624/text2vec) 

### Languages

数据集均是简体中文文本。

## Dataset Structure
### Data Instances
An example of 'train' looks as follows.
```
{"text1":"借款后多长时间给打电话","text2":"借款后多久打电话啊","label":1}
{"text1":"没看到微粒贷","text2":"我借那么久也没有提升啊","label":0}
```

- label 有2个标签,1表示相似,0表示不相似。
### Data Fields
The data fields are the same among all splits.

- `text1`: a `string` feature.
- `text2`: a `string` feature.
- `label`: a classification label, with possible values including entailment(1), contradiction(0)。

### Data Splits

after remove None and len(text) < 1 data:
```shell
$ wc -l nli-zh-all/*
   48818 nli-zh-all/alpaca_gpt4-train.jsonl
    5000 nli-zh-all/amazon_reviews-train.jsonl
  519255 nli-zh-all/belle-train.jsonl
   16000 nli-zh-all/cblue_chip_sts-train.jsonl
  549326 nli-zh-all/chatmed_consult-train.jsonl
   10142 nli-zh-all/cmrc2018-train.jsonl
  395927 nli-zh-all/csl-train.jsonl
   50000 nli-zh-all/dureader_robust-train.jsonl
  709761 nli-zh-all/firefly-train.jsonl
    9568 nli-zh-all/mlqa-train.jsonl
  455875 nli-zh-all/nli_zh-train.jsonl
   50486 nli-zh-all/ocnli-train.jsonl
 2678694 nli-zh-all/simclue-train.jsonl
  419402 nli-zh-all/snli_zh-train.jsonl
    3024 nli-zh-all/webqa-train.jsonl
 1213780 nli-zh-all/wiki_atomic_edits-train.jsonl
   93404 nli-zh-all/xlsum-train.jsonl
 1006218 nli-zh-all/zhihu_kol-train.jsonl
 8234680 total
```

### Data Length

![len](https://huggingface.co/datasets/shibing624/nli-zh-all/resolve/main/nli-zh-all-len.png)

count text length script: https://github.com/shibing624/text2vec/blob/master/examples/data/count_text_length.py

## Dataset Creation
### Curation Rationale
受[m3e-base](https://huggingface.co/moka-ai/m3e-base#M3E%E6%95%B0%E6%8D%AE%E9%9B%86)启发,合并了中文高质量NLI(natural langauge inference)数据集,
这里把这个数据集上传到huggingface的datasets,方便大家使用。
### Source Data
#### Initial Data Collection and Normalization

如果您想要查看数据集的构建方法,你可以在 [https://github.com/shibing624/text2vec/blob/master/examples/data/build_zh_nli_dataset.py](https://github.com/shibing624/text2vec/blob/master/examples/data/build_zh_nli_dataset.py) 中找到生成 nli-zh-all 数据集的脚本,所有数据均上传到 huggingface datasets。

| 数据集名称                | 领域 | 数量        | 任务类型          | Prompt | 质量 | 数据提供者                                                                                         | 说明                                                                                                                                                         | 是否开源/研究使用 | 是否商用 | 脚本 | Done | URL                                                                                         | 是否同质 |
|:---------------------| :---- |:-----------|:---------------- |:------ |:----|:-----------------------------------------------------------------------------------------------|:------------------------------------------------------------------------------------------------------------------------------------------------------------|:----------------- |:------|:---- |:---- |:---------------------------------------------------------------------------------------------|:------|
| cmrc2018             | 百科 | 14,363    | 问答              | 问答   | 优  | Yiming Cui, Ting Liu, Wanxiang Che, Li Xiao, Zhipeng Chen, Wentao Ma, Shijin Wang, Guoping Hu | https://github.com/ymcui/cmrc2018/blob/master/README_CN.md 专家标注的基于维基百科的中文阅读理解数据集,将问题和上下文视为正例                                                               | 是                | 否    | 是   | 是   | https://huggingface.co/datasets/cmrc2018                                                    | 否    |
| belle_0.5m           | 百科 | 500,000   | 指令微调          | 无     | 优  | LianjiaTech/BELLE                                                                             | belle 的指令微调数据集,使用 self instruct 方法基于 gpt3.5 生成                                                                                                             | 是                | 否    | 是   | 是   | https://huggingface.co/datasets/BelleGroup/                                                 | 否    |
| firefily             | 百科 | 1,649,399 | 指令微调          | 无     | 优  | YeungNLP                                                                                      | Firefly(流萤) 是一个开源的中文对话式大语言模型,使用指令微调(Instruction Tuning)在中文数据集上进行调优。使用了词表裁剪、ZeRO等技术,有效降低显存消耗和提高训练效率。 在训练中,我们使用了更小的模型参数量,以及更少的计算资源。                          | 未说明            | 未说明  | 是   | 是   | https://huggingface.co/datasets/YeungNLP/firefly-train-1.1M                                 | 否    |
| alpaca_gpt4          | 百科 | 48,818    | 指令微调          | 无     | 优  | Baolin Peng, Chunyuan Li, Pengcheng He, Michel Galley, Jianfeng Gao                           | 本数据集是参考Alpaca方法基于GPT4得到的self-instruct数据,约5万条。                                                                                                              | 是                | 否    | 是   | 是   | https://huggingface.co/datasets/shibing624/alpaca-zh                                        | 否    |
| zhihu_kol            | 百科 | 1,006,218 | 问答              | 问答   | 优  | wangrui6                                                                                      | 知乎问答                                                                                                                                                       | 未说明            | 未说明  | 是   | 是   | https://huggingface.co/datasets/wangrui6/Zhihu-KOL                                          | 否    |
| amazon_reviews_multi | 电商 | 210,000   | 问答 文本分类     | 摘要   | 优  | 亚马逊                                                                                           | 亚马逊产品评论数据集                                                                                                                                                 | 是                | 否    | 是   | 是   | https://huggingface.co/datasets/amazon_reviews_multi/viewer/zh/train?row=8                  | 否    |
| mlqa                 | 百科 | 85,853    | 问答              | 问答   | 良  | patrickvonplaten                                                                              | 一个用于评估跨语言问答性能的基准数据集                                                                                                                                        | 是                | 未说明  | 是   | 是   | https://huggingface.co/datasets/mlqa/viewer/mlqa-translate-train.zh/train?p=2               | 否    |
| xlsum                | 新闻 | 93,404    | 摘要              | 摘要   | 良  | BUET CSE NLP Group                                                                            | BBC的专业注释文章摘要对                                                                                                                                              | 是                | 否    | 是   | 是   | https://huggingface.co/datasets/csebuetnlp/xlsum/viewer/chinese_simplified/train?row=259    | 否    |
| ocnli                | 口语 | 17,726    | 自然语言推理      | 推理   | 良  | Thomas Wolf                                                                                   | 自然语言推理数据集                                                                                                                                                  | 是                | 否    | 是   | 是   | https://huggingface.co/datasets/clue/viewer/ocnli                                           | 是    |
| BQ                   | 金融 | 60,000    | 文本分类          | 相似   | 优  | Intelligent Computing Research Center, Harbin Institute of Technology(Shenzhen)               | http://icrc.hitsz.edu.cn/info/1037/1162.htm BQ 语料库包含来自网上银行自定义服务日志的 120,000 个问题对。它分为三部分:100,000 对用于训练,10,000 对用于验证,10,000 对用于测试。 数据提供者: 哈尔滨工业大学(深圳)智能计算研究中心 | 是                | 否    | 是   | 是   | https://huggingface.co/datasets/shibing624/nli_zh/viewer/BQ                                 | 是    |
| lcqmc                | 口语 | 149,226   | 文本分类          | 相似   | 优  | Ming Xu                                                                                       | 哈工大文本匹配数据集,LCQMC 是哈尔滨工业大学在自然语言处理国际顶会 COLING2018 构建的问题语义匹配数据集,其目标是判断两个问题的语义是否相同                                                                             | 是                | 否    | 是   | 是   | https://huggingface.co/datasets/shibing624/nli_zh/viewer/LCQMC/train                        | 是    |
| paws-x               | 百科 | 23,576    | 文本分类          | 相似   | 优  | Bhavitvya Malik                                                                               | PAWS Wiki中的示例                                                                                                                                              | 是                | 是    | 是   | 是   | https://huggingface.co/datasets/paws-x/viewer/zh/train                                      | 是    |
| wiki_atomic_edit     | 百科 | 1,213,780 | 平行语义          | 相似   | 优  | abhishek thakur                                                                               | 基于中文维基百科的编辑记录收集的数据集                                                                                                                                        | 未说明            | 未说明  | 是   | 是   | https://huggingface.co/datasets/wiki_atomic_edits                                           | 是    |
| chatmed_consult      | 医药 | 549,326   | 问答              | 问答   | 优  | Wei Zhu                                                                                       | 真实世界的医学相关的问题,使用 gpt3.5 进行回答                                                                                                                                | 是                | 否    | 是   | 是   | https://huggingface.co/datasets/michaelwzhu/ChatMed_Consult_Dataset                         | 否    |
| webqa                | 百科 | 42,216    | 问答              | 问答   | 优  | suolyer                                                                                       | 百度于2016年开源的数据集,数据来自于百度知道;格式为一个问题多篇意思基本一致的文章,分为人为标注以及浏览器检索;数据整体质量中,因为混合了很多检索而来的文章                                                                           | 是                | 未说明  | 是   | 是   | https://huggingface.co/datasets/suolyer/webqa/viewer/suolyer--webqa/train?p=3               | 否    |
| dureader_robust      | 百科 | 65,937    | 机器阅读理解 问答 | 问答   | 优  | 百度                                                                                            | DuReader robust旨在利用真实应用中的数据样本来衡量阅读理解模型的鲁棒性,评测模型的过敏感性、过稳定性以及泛化能力,是首个中文阅读理解鲁棒性数据集。                                                                           | 是                | 是    | 是   | 是   | https://huggingface.co/datasets/PaddlePaddle/dureader_robust/viewer/plain_text/train?row=96 | 否    |
| csl                  | 学术 | 395,927   | 语料              | 摘要   | 优  | Yudong Li, Yuqing Zhang, Zhe Zhao, Linlin Shen, Weijie Liu, Weiquan Mao and Hui Zhang         | 提供首个中文科学文献数据集(CSL),包含 396,209 篇中文核心期刊论文元信息 (标题、摘要、关键词、学科、门类)。CSL 数据集可以作为预训练语料,也可以构建许多NLP任务,例如文本摘要(标题预测)、 关键词生成和文本分类等。                                      | 是                | 是    | 是   | 是   | https://huggingface.co/datasets/neuclir/csl                                                 | 否    |
| snli-zh              | 口语 | 419,402   | 文本分类             | 推理   | 优  | liuhuanyong                                                                                   | 中文SNLI数据集,翻译自英文SNLI                                                                                                                                        | 是                | 否    | 是   | 是   | https://github.com/liuhuanyong/ChineseTextualInference/                                     | 是    |
| SimCLUE              | 百科 | 2,678,694 | 平行语义          | 相似   | 优  | 数据集合,请在 simCLUE 中查看                                                                           | 整合了中文领域绝大多数可用的开源的语义相似度和自然语言推理的数据集,并重新做了数据拆分和整理。                                                                                                            | 是                | 否    | 否   | 是   | https://github.com/CLUEbenchmark/SimCLUE                                                    | 是    |

#### Who are the source language producers?
数据集的版权归原作者所有,使用各数据集时请尊重原数据集的版权。

SNLI:

@inproceedings{snli:emnlp2015,
    Author = {Bowman, Samuel R. and Angeli, Gabor and Potts, Christopher, and Manning, Christopher D.},
    Booktitle = {Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing (EMNLP)},
    Publisher = {Association for Computational Linguistics},
    Title = {A large annotated corpus for learning natural language inference},
    Year = {2015}
}

#### Who are the annotators?
原作者。


### Social Impact of Dataset
This dataset was developed as a benchmark for evaluating representational systems for text, especially including those induced by representation learning methods, in the task of predicting truth conditions in a given context. 

Systems that are successful at such a task may be more successful in modeling semantic representations.


### Licensing Information

for reasearch

用于学术研究



### Contributions

[shibing624](https://github.com/shibing624) add this dataset.