File size: 6,420 Bytes
89dd8c5 |
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 |
---
license: apache-2.0
pipeline_tag: image-text-to-text
base_model:
- Qwen/Qwen2.5-VL-72B-Instruct
language:
- multilingual
---
# SafeWork-R1
[📂 GitHub](https://github.com/AI45Lab/SafeWork-R1) · [📜Technical Report](https://arxiv.org/abs/2507.18576) · [💬Online Chat](https://safework-r1.ai45.shlab.org.cn/)
<div align="center">
<img alt="image" src="/static-proxy?url=https%3A%2F%2Fcdn-uploads.huggingface.co%2Fproduction%2Fuploads%2F666fe1a5b07525f0bde69c27%2F9VqjAkK1Lshl3TVpMFV9-.png%26quot%3B%3C%2Fspan%3E%26gt%3B%3C%2Fspan%3E%3C%2Fspan%3E
</div>
## Overview
We introduce SafeWork-R1, a cutting-edge multimodal reasoning model demonstrating the coevolution of safety and general intelligence under the guiding principle of the AI-45° Law.
SafeWork-R1 is built upon the SafeLadder framework, which integrates large-scale, progressive, safety-oriented reinforcement learning post-training supported by multi-principled verifiers. Unlike conventional RLHF that simply learns human preferences, SafeLadder enables SafeWork-R1 to develop intrinsic safety reasoning and self-reflection abilities, leading to emergent safety “aha” moments.
<div align="center">
</div>
## Model Zoo
<table>
<tr>
<th>Model Variant</th>
<th>Parameters</th>
<th>Base Model</th>
<th>Link</th>
</tr>
<tr>
<td>SafeWork-R1</td>
<td>72B</td>
<td>Qwen2.5-VL-72B</td>
<td><a href="https://huggingface.co/AI45Research/SafeWork-R1">🤗 link</a></td>
</tr>
<tr>
<td>SafeWork-R1-InternVL3-78B</td>
<td>78B</td>
<td>InternVL3-78B</td>
<td><a href="https://huggingface.co/AI45Research/SafeWork-R1-InternVL3-78B">🤗 link</a></td>
</tr>
<tr>
<td>SafeWork-R1-DeepSeek-70B</td>
<td>70B</td>
<td>Deepseek-R1-DistillLlama-70B</td>
<td><a href="https://huggingface.co/AI45Research/SafeWork-R1-DeepSeek-70B">🤗 link</a></td>
</tr>
<tr>
<td>SafeWork-R1-Qwen2.5VL-7B</td>
<td>7B</td>
<td>Qwen2.5-VL-7B</td>
<td><a href="https://huggingface.co/AI45Research/SafeWork-R1-Qwen2.5VL-7B">🤗 link</a></td>
</tr>
</table>
## Performance
### Safety Benchmarks
| Model | MM-SafetyBench | MSSBench | XSTest-Safe | SIUO | Avg. |
|----------------|----------------|-----------|--------------|-------|-------|
| Gemini 2.5 pro | 79.3 | 70.5 | **100.0** | 76.7 | 81.6 |
| Claude Opus 4 | 82.1 | 59.6 | 96.8 | 62.8 | 75.3 |
| GPT-4.1 | 78.2 | 69.1 | 96.4 | **92.9** | 84.1 |
| GPT-4o | 70.2 | 58.8 | 94.0 | 51.8 | 68.7 |
| Qwen2.5-VL-72B | 70.4 | 53.8 | 91.2 | 38.2 | 63.4 |
| **SafeWork-R1**| **92.0**<sup>↑21.6</sup> | **74.8**<sup>↑21.0</sup> | **99.2**<sup>↑8.0</sup> | **90.5**<sup>↑52.3</sup> | **89.2**<sup>↑25.8</sup> |
### Value Benchmarks
| Model | FLAMES | M³oralBench (Judge) | M³oralBench (Classification) | M³oralBench (Response) | Avg. |
|----------------|---------|---------------------|------------------------------|------------------------|-------|
| Gemini 2.5 Pro | 16.8 | 70.0 | 66.2 | **86.8** | 44.7 |
| Claude Opus 4 | 38.1 | 70.7 | **74.7** | 72.5 | 52.2 |
| GPT-4.1 | 33.3 | **74.4** | 62.7 | 61.7 | 53.0 |
| GPT-4o | 36.6 | 72.4 | 65.9 | 79.7 | 55.5 |
| Qwen2.5-VL-72B | 39.1 | 58.4 | 48.1 | 75.7 | 49.9 |
| **SafeWork-R1**| **65.3**<sup>↑26.2</sup> | **68.1**<sup>↑9.7</sup> | **54.6**<sup>↑6.5</sup> | 70.9<sup>↓4.8</sup> | **64.9**<sup>↑15.0</sup> |
### General Benchmarks
| Model | MMMU | MathVista | Olympiad | GPQA Diamond | GAOKAO-MM | Avg. |
|----------------|------|------------|-----------|---------------|------------|-------|
| Gemini 2.5 Pro | **82.0** | **83.0** | **81.8** | **86.9** | **87.2** | **84.2** |
| Claude Opus 4 | 73.0 | 73.0 | 68.5 | 74.7 | 73.7 | 72.6 |
| GPT-4.1 | 72.4 | 72.0 | 49.0 | 69.2 | 60.2 | 64.6 |
| GPT-4o | 70.6 | 61.6 | 33.7 | 46.9 | 33.8 | 49.3 |
| Qwen2.5-VL-72B | 67.2 | 74.8 | 40.4 | 50.5 | 73.1 | 61.2 |
| **SafeWork-R1**| **70.9**<sup>↑3.7</sup> | **76.1**<sup>↑1.3</sup> | **59.9**<sup>↑19.5</sup> | **59.6**<sup>↑9.1</sup> | **78.2**<sup>↑5.1</sup> | **68.9**<sup>↑7.7</sup> |
## Quick Start
```python
from transformers import AutoProcessor, AutoModelForCausalLM
import torch
model_name = "AI45Research/SafeWork-R1"
processor = AutoProcessor.from_pretrained(model_name, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(model_name, device_map="auto", torch_dtype="auto", trust_remote_code=True)
messages = [
{
"role": "user",
"content": [
{
"type": "image",
"image": "file:///path/to/image",
},
{"type": "text", "text": "Prompt containing harmful content."},
],
}
]
# Preparation for inference
text = processor.apply_chat_template(
messages, tokenize=False, add_generation_prompt=True
)
image_inputs, video_inputs = process_vision_info(messages)
inputs = processor(
text=[text],
images=image_inputs,
videos=video_inputs,
padding=True,
return_tensors="pt",
)
inputs = inputs.to("cuda")
# Inference: Generation of the output
generated_ids = model.generate(**inputs, max_new_tokens=8192)
generated_ids_trimmed = [
out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)
]
output_text = processor.batch_decode(
generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False
)
print(output_text)
```
## License
This project is released under the Apache 2.0 license.
## Citation
If you find this work useful, feel free to give us a cite.
```
@misc{lab2025safework,
title={SafeWork-R1: Coevolving Safety and Intelligence under the AI-45 Law},
author={Lab, Shanghai AI and Bao, Yicheng and Chen, Guanxu and Chen, Mingkang and Chen, Yunhao and Chen, Chiyu and Chen, Lingjie and Chen, Sirui and Chen, Xinquan and Cheng, Jie and others},
journal={arXiv preprint arXiv:2507.18576},
year={2025}
}
``` |