Add model card for SynPlanResearch-R1-8B

README.md

@@ -0,0 +1,41 @@
+---
+library_name: transformers
+pipeline_tag: text-generation
+tags:
+- tool-use
+- research-agent
+- synthetic-data
+- qwen
+---
+
+# SynPlanResearch-R1-8B
+
+This repository contains the weights for **SynPlanResearch-R1-8B**, a research agent model based on the Qwen3-8B backbone. It was introduced in the paper [SynPlanResearch-R1: Encouraging Tool Exploration for Deep Research with Synthetic Plans](https://huggingface.co/papers/2603.07853).
+
+## Model Description
+
+Research Agents enable models to gather information from the web using tools to answer user queries. However, these agents often exhibit poor exploration behaviors, such as premature termination and biased tool usage. 
+
+SynPlanResearch-R1 is a framework that addresses these issues by synthesizing tool-use trajectories that encourage deeper exploration. These synthetic plans are used during cold-start supervised fine-tuning (SFT) to provide a strong initialization for subsequent reinforcement learning with verifiable rewards (RLVR).
+
+Across seven multi-hop and open-web benchmarks, this framework improves performance by up to 6.0% on Qwen3-8B backbones compared to state-of-the-art baselines.
+
+## Resources
+
+- **Paper:** [SynPlanResearch-R1: Encouraging Tool Exploration for Deep Research with Synthetic Plans](https://huggingface.co/papers/2603.07853)
+- **GitHub Repository:** [HansiZeng/syn-plan-research](https://github.com/HansiZeng/syn-plan-research)
+
+## Training and Evaluation
+
+For detailed instructions on environment setup, data downloading, and running the SFT/RL training or evaluation scripts, please refer to the [official GitHub repository](https://github.com/HansiZeng/syn-plan-research).
+
+## Citation
+
+```bibtex
+@article{zeng2026synplanresearch,
+  title={SynPlanResearch-R1: Encouraging Tool Exploration for Deep Research with Synthetic Plans},
+  author={Zeng, Hansi and others},
+  journal={arXiv preprint arXiv:2603.07853},
+  year={2026}
+}
+```