---
library_name: transformers
license: other
base_model: meta-llama/Llama-3.2-1B
tags:
- llama-factory
- full
- generated_from_trainer
model-index:
- name: GuardReasoner 1B
  results: []
pipeline_tag: text-classification
language:
- en
metrics:
- f1
---

# GuardReasoner 1B

This model is a fine-tuned version of [meta-llama/Llama-3.2-1B](https://huggingface.co./meta-llama/Llama-3.2-1B) via R-SFT and HS-DPO. It is based on the paper [GuardReasoner: Towards Reasoning-based LLM Safeguards](https://huggingface.co./papers/2501.18492).

The training data of R-SFT can be found in [GuardReasonerTrain](https://huggingface.co./datasets/yueliu1999/GuardReasonerTrain).


```
@article{GuardReasoner,
  title={GuardReasoner: Towards Reasoning-based LLM Safeguards},
  author={Liu, Yue and Gao, Hongcheng and Zhai, Shengfang and Jun, Xia and Wu, Tianyi and Xue, Zhiwei and Chen, Yulin and Kawaguchi, Kenji and Zhang, Jiaheng and Hooi, Bryan},
  journal={arXiv preprint arXiv:2501.18492},
  year={2025}
}
```