Top AI Repos β open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
π§Tool-Star: Empowering LLM-brained Multi-Tool Reasoner via Reinforcement Learning
| Date | Stars |
|---|---|
| 2026-07-31 | 405 |
| 2026-08-01 | 407 |
| 2026-08-02 | 407 |
| 2026-08-04 | 406 |
| 2026-08-05 | 407 |
| 2026-08-06 | 407 |
Today
β stars today
This week
β stars this week
This month
β stars this month
Momentum
0.0
growth rate 0.00%/day
<div align="center"> <img src="https://github.com/dongguanting/Tool-Star/blob/main/img/image.png" width="120px"> </div> <h1 align="center"> π§β¨Tool-Star: Empowering Multi-Tool Collaborative Web Agent via Reinforcement Learning</a></h1> <div align="center"> [](https://arxiv.org/abs/2505.16410) [](https://huggingface.co/papers/2505.16410) [](https://opensource.org/licenses/MIT) [](https://www.python.org/downloads/release/python-390/) [](https://x.com/_akhaliq/status/1925924431676821698) </div> <p align="center"> π€ <a href="https://huggingface.co/dongguanting/Tool-Star-Qwen-0.5B" target="_blank">Tool-Star-Qwen-0.5B</a> ο½ π€ <a href="https://huggingface.co/dongguanting/Tool-Star-Qwen-1.5B" target="_blank">Tool-Star-Qwen-1.5B</a> ο½ π€ <a href="https://huggingface.co/dongguanting/Tool-Star-Qwen-3B" target="_blank">Tool-Star-Qwen-3B</a> ο½ π€ <a href="https://huggingface.co/dongguanting/Tool-Star-Qwen-7B" target="_blank">Tool-Star-Qwen-7B</a> ο½ </p> <p align="center"> π€ <a href="https://huggingface.co/datasets/dongguanting/Tool-Star-SFT-54K" target="_blank">Tool-Star-SFT-54K</a> ο½ π€ <a href="https://huggingface.co/datasets/dongguanting/Multi-Tool-RL-10K" target="_blank">Multi-Tool-RL-10K</a> </p> <h5 align="center"> If you like our project, please give us a star β on GitHub for the latest update.</h5> ## π£ Latest News - **[Apr 02, 2026]**: π Our paper **[Tool-Star: Empowering Multi-Tool Collaborative Web Agent via Reinforcement Learning](https://arxiv.org/abs/2507.19849)** has been accepted at SIGIR 2026! - **[Oct 16, 2025]**: πππ We propose a new algorithm [**AEPO**](https://www.arxiv.org/abs/2510.14545), which focused on entropy-balanced agentic RL and consistently outperforms ARPO on datasets like GAIA, HLE, and AIME. Full [codebase](https://github.com/RUC-NLPIR/ARPO/tree/main/AEPO) and [π€ HF-Models](https://huggingface.co/collections/dongguanting/aepo-68ef6832c99697ee03d5e1c7) of AEPO released. - **[July 25, 2025]**: πππ We have released a new project **[ARPO](https://github.com/dongguanting/ARPO)** , which significantly accelerates the training process for Tool-star (**~4 times faster** ) and supports training for the Qwen2.5, Qwen3, and Llama3 series models! We welcome everyone to try and star it!! - **[June 30, 2025]**: π₯ We have updated our **[π€Tool-Star-Qwen-7B](https://huggingface.co/dongguanting/Tool-Star-Qwen-7B)** and refreshed the **[Performance of Tool-Star Series Models](#-performance-of-tool-star-models)** in the README. We welcome everyone to reproduce and cite it! - **[June 6, 2025]**: We released more lightweight checkpoints of Tool-Star . Checkout **[π€Tool-Star-Qwen-0.5B](https://huggingface.co/dongguanting/Tool-Star-Qwen-0.5B)** & **[π€Tool-Star-Qwen-1.5B](https://huggingface.co/dongguanting/Tool-Star-Qwen-1.5B)** here. - **[May 21, 2025]**: The brief introduction of Tool-Star can be found on platforms like **[X](https://x.com/_akhaliq/status/1925924431676821698), [Zhihu](https://zhuanlan.zhihu.com/p/1911573573602115645) and [Wechat](https://mp.weixin.qq.com/s/UNP3P2GEtIuYhT7Z8wIV1g?scene=1)**. - **[May 21, 2025]**: **[π€ Tool-Star Collection](https://huggingface.co/collections/dongguanting/tool-star-682fd73dfa508bf3f40da032)** is now available on Hugging Face. We will keep update it! - **[May 21, 2025]**: π₯ We released an our cold-star SFT and RL dataset for tool-integrated reasoning. Checkout **[π€Tool-Star-SFT-54K](https://huggingface.co/datasets/dongguanting/Tool-Star-SFT-54K)** and **[Multi-Tool-RL-10K](https://huggingface.co/datasets/dongguanting/Multi-Tool-RL-10
Excerpt of 30,290 characters
Read on GitHubWould you bet a product on this? Bounded 0β100 and slow moving.
matched fp:b13b67a97df8688a, llm:Repository description: 'Tool-Star: Empowering LLM-brained Multi-Tool Reasoner via Reinforcement Learning' (Python). Indicates work on LLM-based multi-tool reasoning and reinforcement learning to coordinate tools.
matched fp:b13b67a97df8688a, llm:Repository description: 'Tool-Star: Empowering LLM-brained Multi-Tool Reasoner via Reinforcement Learning' (Python). Indicates work on LLM-based multi-tool reasoning and reinforcement learning to coordinate tools.
matched fp:b13b67a97df8688a, llm:Repository description: 'Tool-Star: Empowering LLM-brained Multi-Tool Reasoner via Reinforcement Learning' (Python). Indicates work on LLM-based multi-tool reasoning and reinforcement learning to coordinate tools.