Datasets:
shibing624
commited on
Commit
•
cc2acf9
1
Parent(s):
36cd788
Update README.md
Browse files
README.md
CHANGED
@@ -11,17 +11,17 @@ multilinguality:
|
|
11 |
size_categories:
|
12 |
- 1M<n<10M
|
13 |
source_datasets:
|
14 |
-
- https://
|
15 |
task_categories:
|
16 |
- text-classification
|
17 |
task_ids:
|
18 |
- natural-language-inference
|
19 |
- semantic-similarity-scoring
|
20 |
- text-scoring
|
21 |
-
paperswithcode_id:
|
22 |
-
pretty_name:
|
23 |
---
|
24 |
-
# Dataset Card for
|
25 |
|
26 |
## Dataset Description
|
27 |
- **Repository:** [Chinese NLI dataset](https://github.com/shibing624/text2vec)
|
@@ -35,9 +35,6 @@ pretty_name: Stanford Natural Language Inference
|
|
35 |
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。
|
36 |
|
37 |
|
38 |
-
|
39 |
-
|
40 |
-
|
41 |
### Supported Tasks and Leaderboards
|
42 |
|
43 |
Supported Tasks: 支持中文文本匹配任务,文本相似度计算等相关任务。
|
@@ -70,7 +67,6 @@ The data fields are the same among all splits.
|
|
70 |
|
71 |
after remove None and len(text) < 1 data:
|
72 |
```shell
|
73 |
-
|
74 |
$ wc -l nli-zh-all/*
|
75 |
48818 nli-zh-all/alpaca_gpt4-train.jsonl
|
76 |
5000 nli-zh-all/amazon_reviews-train.jsonl
|
@@ -91,13 +87,14 @@ $ wc -l nli-zh-all/*
|
|
91 |
93404 nli-zh-all/xlsum-train.jsonl
|
92 |
1006218 nli-zh-all/zhihu_kol-train.jsonl
|
93 |
8234680 total
|
94 |
-
|
95 |
```
|
96 |
|
97 |
### Data Length
|
98 |
|
99 |
![len](https://huggingface.co/datasets/shibing624/nli-zh-all/resolve/main/nli-zh-all-len.png)
|
100 |
|
|
|
|
|
101 |
## Dataset Creation
|
102 |
### Curation Rationale
|
103 |
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,
|
@@ -132,7 +129,8 @@ $ wc -l nli-zh-all/*
|
|
132 |
#### Who are the source language producers?
|
133 |
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。
|
134 |
|
135 |
-
|
|
|
136 |
@inproceedings{snli:emnlp2015,
|
137 |
Author = {Bowman, Samuel R. and Angeli, Gabor and Potts, Christopher, and Manning, Christopher D.},
|
138 |
Booktitle = {Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing (EMNLP)},
|
|
|
11 |
size_categories:
|
12 |
- 1M<n<10M
|
13 |
source_datasets:
|
14 |
+
- https://github.com/shibing624/text2vec
|
15 |
task_categories:
|
16 |
- text-classification
|
17 |
task_ids:
|
18 |
- natural-language-inference
|
19 |
- semantic-similarity-scoring
|
20 |
- text-scoring
|
21 |
+
paperswithcode_id: nli
|
22 |
+
pretty_name: Chinese Natural Language Inference
|
23 |
---
|
24 |
+
# Dataset Card for nli-zh-all
|
25 |
|
26 |
## Dataset Description
|
27 |
- **Repository:** [Chinese NLI dataset](https://github.com/shibing624/text2vec)
|
|
|
35 |
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。
|
36 |
|
37 |
|
|
|
|
|
|
|
38 |
### Supported Tasks and Leaderboards
|
39 |
|
40 |
Supported Tasks: 支持中文文本匹配任务,文本相似度计算等相关任务。
|
|
|
67 |
|
68 |
after remove None and len(text) < 1 data:
|
69 |
```shell
|
|
|
70 |
$ wc -l nli-zh-all/*
|
71 |
48818 nli-zh-all/alpaca_gpt4-train.jsonl
|
72 |
5000 nli-zh-all/amazon_reviews-train.jsonl
|
|
|
87 |
93404 nli-zh-all/xlsum-train.jsonl
|
88 |
1006218 nli-zh-all/zhihu_kol-train.jsonl
|
89 |
8234680 total
|
|
|
90 |
```
|
91 |
|
92 |
### Data Length
|
93 |
|
94 |
![len](https://huggingface.co/datasets/shibing624/nli-zh-all/resolve/main/nli-zh-all-len.png)
|
95 |
|
96 |
+
count text length script: https://github.com/shibing624/text2vec/blob/master/examples/data/count_text_length.py
|
97 |
+
|
98 |
## Dataset Creation
|
99 |
### Curation Rationale
|
100 |
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,
|
|
|
129 |
#### Who are the source language producers?
|
130 |
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。
|
131 |
|
132 |
+
SNLI:
|
133 |
+
|
134 |
@inproceedings{snli:emnlp2015,
|
135 |
Author = {Bowman, Samuel R. and Angeli, Gabor and Potts, Christopher, and Manning, Christopher D.},
|
136 |
Booktitle = {Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing (EMNLP)},
|