---
license: mit
library_name: transformers
pipeline_tag: text-generation
base_model: Qwen/Qwen2.5-3B-Instruct
tags:
- agent
- tool-use
- reinforcement-learning
- travel-planner
---

# Agent-STAR-RL-3B

This repository contains the **Agent-STAR-RL-3B** model, a 3B parameter Large Language Model fine-tuned for long-horizon tool orchestration tasks. It was introduced in the paper [Demystifying Reinforcement Learning for Long-Horizon Tool-Using Agents: A Comprehensive Recipe](https://huggingface.co/papers/2603.21972).

## Model Description

Agent-STAR is a unified post-training pipeline consisting of **[Data Synthesis → SFT → RL]**. This specific checkpoint is the RL-tuned version based on the **Qwen2.5-3B-Instruct** backbone, optimized for the [TravelPlanner](https://github.com/OSU-NLP-Group/TravelPlanner/) benchmark. 

The model was developed to handle complex, multi-turn agentic environments where it must call various tools to satisfy multifaceted constraints. According to the research findings, smaller models like this 3B variant benefit from staged rewards and enhanced exploration during the RL phase to achieve high performance.

## Resources

- **Paper:** [Demystifying Reinforcement Learning for Long-Horizon Tool-Using Agents: A Comprehensive Recipe](https://huggingface.co/papers/2603.21972)
- **GitHub Repository:** [WxxShirley/Agent-STAR](https://github.com/WxxShirley/Agent-STAR)
- **Dataset:** [Agent-STAR-TravelDataset](https://huggingface.co/datasets/xxwu/Agent-STAR-TravelDataset)

## Inference

To run inference with this model, please refer to the instructions and ReAct-based inference pipeline provided in the [official GitHub repository](https://github.com/WxxShirley/Agent-STAR).

## Citation

If you find Agent-STAR helpful to your work, please consider citing:

```bibtex
@misc{wu2026agentstar,
      title={Demystifying Reinforcement Learning for Long-Horizon Tool-Using Agents: A Comprehensive Recipe}, 
      author={Xixi Wu and Qianguo Sun and Ruiyang Zhang and Chao Song and Junlong Wu and Yiyan Qi and Hong Cheng},
      year={2026},
      eprint={2603.21972},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/2603.21972}, 
}
```

## Acknowledgements

We appreciate the open-sourced [rLLM](https://github.com/rllm-org/rllm/) framework and the authors of [TravelPlanner](https://github.com/OSU-NLP-Group/TravelPlanner) for providing the benchmark and resources that supported this research.