Datasets:
Tasks:
Text Classification
Modalities:
Text
Formats:
json
Languages:
Chinese
Size:
10K - 100K
ArXiv:
License:
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,42 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: cc-by-nc-4.0
|
| 3 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: cc-by-nc-4.0
|
| 3 |
+
---
|
| 4 |
+
<h1 align="center"> ChineseHarm-bench</h1>
|
| 5 |
+
<h3 align="center"> A Chinese Harmful Content Detection Benchmark </h3>
|
| 6 |
+
|
| 7 |
+
> ⚠️ **WARNING**: This project and associated data contain content that may be toxic, offensive, or disturbing. Use responsibly and with discretion.
|
| 8 |
+
|
| 9 |
+
<p align="center">
|
| 10 |
+
<a href="">Project</a> •
|
| 11 |
+
<a href="">Paper</a> •
|
| 12 |
+
<a href="https://huggingface.co/collections/zjunlp/chineseharm-bench-683b452c5dcd1d6831c3316c">Hugging Face</a>
|
| 13 |
+
</p>
|
| 14 |
+
|
| 15 |
+
|
| 16 |
+
|
| 17 |
+
<div>
|
| 18 |
+
</div>
|
| 19 |
+
<div align="center">
|
| 20 |
+
<p align="center">
|
| 21 |
+
<img src="figs/chineseharm_case.png" width="80%"/></p>
|
| 22 |
+
</div>
|
| 23 |
+
|
| 24 |
+
## 🌟Benchmark
|
| 25 |
+
|
| 26 |
+
This folder contains the ChineseHarm-Bench.
|
| 27 |
+
|
| 28 |
+
* `bench.json` is the full benchmark combining all categories.
|
| 29 |
+
* The other files (e.g., `低俗色情.json`, `欺诈.json`) are category-specific subsets.
|
| 30 |
+
|
| 31 |
+
Each file is a list of examples with:
|
| 32 |
+
|
| 33 |
+
* `"文本"`: the input Chinese text
|
| 34 |
+
* `"标签"`: the ground-truth label
|
| 35 |
+
|
| 36 |
+
## 🚩Citation
|
| 37 |
+
|
| 38 |
+
Please cite our repository if you use ChineseHarm-bench in your work. Thanks!
|
| 39 |
+
|
| 40 |
+
```bibtex
|
| 41 |
+
|
| 42 |
+
```
|