### Download SafetyBench Data with Shell Script

Source: https://github.com/thu-coai/safetybench/blob/main/README.md

Use this shell script to download test questions and few-shot examples from Hugging Face.

```shell
#!/bin/bash

# Download test questions

# Chinese test questions
wget https://huggingface.co/datasets/thu-coai/SafetyBench/resolve/main/test_zh.json
# English test questions
wget https://huggingface.co/datasets/thu-coai/SafetyBench/resolve/main/test_en.json
# Chinese subset test questions
wget https://huggingface.co/datasets/thu-coai/SafetyBench/resolve/main/test_zh_subset.json

# Download few-shot examples

# Chinese few-shot examples
wget https://huggingface.co/datasets/thu-coai/SafetyBench/resolve/main/dev_zh.json
# English few-shot examples
wget https://huggingface.co/datasets/thu-coai/SafetyBench/resolve/main/dev_en.json

```

--------------------------------

### Submission File Format Example

Source: https://github.com/thu-coai/safetybench/blob/main/README.md

Prepare a UTF-8 encoded JSON file with question IDs as keys and predicted answers (0->A, 1->B, 2->C, 3->D) as values.

```json
{
    "0": 0,
    "1": 1,
    "2": 3,
    "3": 2 
}

```

--------------------------------

### Download SafetyBench Data with Python Datasets Library

Source: https://github.com/thu-coai/safetybench/blob/main/README.md

Use the Hugging Face datasets library in Python to download test questions and few-shot examples.

```python
from datasets import load_dataset

# Load test questions
test_zh = load_dataset("thu-coai/SafetyBench", "test_zh")
test_en = load_dataset("thu-coai/SafetyBench", "test_en")
test_zh_subset = load_dataset("thu-coai/SafetyBench", "test_zh_subset")

# Load few-shot examples
dev_zh = load_dataset("thu-coai/SafetyBench", "dev_zh")
dev_en = load_dataset("thu-coai/SafetyBench", "dev_en")

```

--------------------------------

### Citation for SafetyBench

Source: https://github.com/thu-coai/safetybench/blob/main/README.md

BibTeX citation for the SafetyBench paper.

```bibtex
@article{zhang2023safetybench,
      title={SafetyBench: Evaluating the Safety of Large Language Models with Multiple Choice Questions},
      author={Zhexin Zhang and Leqi Lei and Lindong Wu and Rui Sun and Yongkang Huang and Chong Long and Xiao Liu and Xuanyu Lei and Jie Tang and Minlie Huang},
      journal={arXiv preprint arXiv:2309.07045},
      year={2023}
}

```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.