---
license: apache-2.0
language:
- en
base_model:
- Qwen/Qwen2.5-1.5B-Instruct
library_name: transformers
pipeline_tag: text-generation
---

# TARS-SFT-1.5B

## Overview

**TARS-SFT-1.5B** is a lightweight SFT-tuned reasoning model for safety used for **TARS**: *Training Adaptive Reasoners for Safety* introduced in the paper: [**Reasoning as an Adaptive Defense for Safety**](https://arxiv.org/abs/2507.00971). This model is \\(\pi_{SFT}\\), which is used as the base model for RL training, trained starting from [Qwen2.5-1.5B-Instruct](https://huggingface.co/Qwen/Qwen2.5-1.5B-Instruct).

For full details, please check out our [paper](https://arxiv.org/pdf/2507.00971) or [blogpost](https://training-adaptive-reasoners-safety.github.io).

---

## 📖 Citation

If you use **TARS-SFT-1.5B** in your work, please cite us:

```bibtex
@misc{kim2025reasoningadaptivedefensesafety,
  title        = {Reasoning as an Adaptive Defense for Safety},
  author       = {Taeyoun Kim and Fahim Tajwar and Aditi Raghunathan and Aviral Kumar},
  year         = {2025},
  eprint       = {2507.00971},
  archivePrefix= {arXiv},
  primaryClass = {cs.LG},
  url          = {https://arxiv.org/abs/2507.00971}
}