{"id":28653867,"url":"https://github.com/tiger-ai-lab/verl-tool","last_synced_at":"2025-12-24T10:25:14.614Z","repository":{"id":283897613,"uuid":"952356097","full_name":"TIGER-AI-Lab/verl-tool","owner":"TIGER-AI-Lab","description":"A version of verl to support tool use","archived":false,"fork":false,"pushed_at":"2025-06-09T14:18:01.000Z","size":11815,"stargazers_count":196,"open_issues_count":4,"forks_count":10,"subscribers_count":3,"default_branch":"main","last_synced_at":"2025-06-09T15:29:36.914Z","etag":null,"topics":["learning","llm","reinforcement"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/TIGER-AI-Lab.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null}},"created_at":"2025-03-21T06:32:01.000Z","updated_at":"2025-06-09T08:52:20.000Z","dependencies_parsed_at":null,"dependency_job_id":"989722cc-d395-4eff-8c62-2f91a1d0375f","html_url":"https://github.com/TIGER-AI-Lab/verl-tool","commit_stats":null,"previous_names":["tiger-ai-lab/verl-tool"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/TIGER-AI-Lab/verl-tool","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/TIGER-AI-Lab%2Fverl-tool","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/TIGER-AI-Lab%2Fverl-tool/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/TIGER-AI-Lab%2Fverl-tool/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/TIGER-AI-Lab%2Fverl-tool/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/TIGER-AI-Lab","download_url":"https://codeload.github.com/TIGER-AI-Lab/verl-tool/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/TIGER-AI-Lab%2Fverl-tool/sbom","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":259599331,"owners_count":22882357,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["learning","llm","reinforcement"],"created_at":"2025-06-13T07:07:54.450Z","updated_at":"2025-12-24T10:25:14.601Z","avatar_url":"https://github.com/TIGER-AI-Lab.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Verl-Tool\n\n\u003cp align=\"center\"\u003e\n  \u003cpicture\u003e\n    \u003csource media=\"(prefers-color-scheme: dark)\" srcset=\"assets/imgs/logo.png\"\u003e\n    \u003cimg alt=\"VerlTool\" src=\"assets/imgs/logo.png\" width=20%\u003e\n  \u003c/picture\u003e\n\u003c/p\u003e\n\n\u003ch3 align=\"center\"\u003e\nVerlTool: A unified and easy-to-extend tool-agent training framework based on verl.\n\u003c/h3\u003e\n\n\u003cp align=\"center\"\u003e\n| \n\u003ca href=\"https://arxiv.org/abs/2509.01055\"\u003e\u003cb\u003ePaper\u003c/b\u003e\u003c/a\u003e |\n\u003ca href=\"https://github.com/TIGER-AI-Lab/verl-tool/blob/main/assets/docs/install.md\"\u003e\u003cb\u003eQuick Start\u003c/b\u003e\u003c/a\u003e |\n  \u003ca href=\"https://github.com/TIGER-AI-Lab/verl-tool/tree/main/examples/train\"\u003e\u003cb\u003eTraining Recipes\u003c/b\u003e\u003c/a\u003e |\n  \u003ca href=\"https://deepwiki.com/TIGER-AI-Lab/verl-tool\"\u003e\u003cb\u003eDeepWiki\u003c/b\u003e\u003c/a\u003e |\n  \u003ca href=\"https://github.com/TIGER-AI-Lab/verl-tool/tree/main/assets/imgs/wechat_group.jpg\"\u003e\u003cb\u003eWeChat Group\u003c/b\u003e\u003c/a\u003e |\n  \u003ca href=\"https://discord.gg/kZggJmaz\"\u003e\u003cb\u003eDiscord\u003c/b\u003e\u003c/a\u003e\n|\n\u003c/p\u003e\n\n---\n\n\n\n## News\n+ [2025/11/10] VerlTool has re-organized its codebase to improve modularity and maintainability, supporting to the latest verl (`0.6.0`) and vllm (`0.11.0`) versions. Please refer to the [verl-tool v0.6.0.dev Upgrade Notes](/assets/docs/updates/verltool_v0.6.0_upgrade.md) for more details.\n+ [2025/09/02] VerlTool's tech report is out! See on [Hugging Face Daily Paper](https://huggingface.co/papers/2509.01055)!\n+ [2025/06/30] We reproduce Search-R1 with even higher performance on the same benchmarks! See [PR](https://github.com/TIGER-AI-Lab/verl-tool/pull/71) and training [README](examples/train/search_r1/README.md) for more details.\n+ [2025/06/28] We support NL2SQL tool RL training. See NL2SQL [README](https://github.com/TIGER-AI-Lab/verl-tool/tree/main/examples/train/skysql) for more details.\n+ [2025/06/26] We support DAPO recipe training. See [DAPO.md](./assets/docs/DAPO.md) for more details.\n+ [2025/06/18] VerlTool now officially supports Trajectory-Level asynchronous, speeding up the rollout generation with tool calling by at least 2x! see [asyncRL.md](./assets/docs/asyncRL.md) for more details.\n+ [2025/06/16] We have updated the verl submodule to the latest version (06/16) and modified some code to adapt to the new version.\n+ [2025/06/13] We integrated [DeepWiki](https://deepwiki.com/TIGER-AI-Lab/verl-tool) for Verl-Tool. Feel free to browse the AI-generated docs and chat with Verl-tool codes.\n+ [2025/06/06] We have updated a detailed design overview in the README, including how to add new tools, how to use the tool server, and how to train your own models with verl-tool.\n+ [2025/05/31] We released the Verl-tool training/evaluation code with ToRL training as an initial example (see [X post](https://x.com/DongfuJiang/status/1929198238017720379)). We are working on the paper and will release it very soon.\n\n## Features\n\n- 🔧 **Complete decoupling of actor rollout and environment interaction** - We use verl as a submodule to benefit from ongoing verl repository updates. All tool calling is integrated via a unified API, allowing you to easily add new tools by simply adding a Python file and testing independently.\n- 🌍 **Tool-as-environment paradigm** - Each tool interaction can modify the environment state. We store and reload environment states for each trajectory.\n- ⚡ **Native RL framework for tool-calling agents** - verl-tool natively supports multi-turn interactive loops between agents and their tool environments.\n- 📊 **User-friendly evaluation suite** - Launch your trained model with OpenAI API alongside the tool server. Simply send questions and get final outputs with all interactions handled internally. See [benchmarks](benchmarks).\n\n![Verl-Tool Architecture](assets/imgs/verl_tool_architecture.png)\n\n## 📚 Contents Link\n- 📖 [Installation Guide](./assets/docs/install.md)\n- ⚡ [Synchronous Rollout Design](./assets/docs/sync_design.md)\n- 🔄 [Asynchronous Rollout Design](./assets/docs/asyncRL.md)\n- 🛠️ [Tool Server Design](./assets/docs/tool_server.md)\n- 🎯 [Training Guide](./assets/docs/training_guide.md)\n- 📊 [Evaluation Guide](./assets/docs/evaluation.md)\n- 🔧 [Update Verl Submodule Version](./assets/docs/update_verl.md)\n- 📈 [Existing Training Results](./assets/docs/training_results.md)\n- 🤝 [Contributing Guide](./assets/docs/contributing.md)\n\n## Core Contributors\n\n\u003ctable\u003e\n\u003ctr\u003e\n    \u003ctd align=\"center\"\u003e\n        \u003ca href=\"https://github.com/jdf-prog\"\u003e\n            \u003cimg src=\"https://github.com/jdf-prog.png\" width=\"75px;\" alt=\"Dongfu Jiang\"/\u003e\n            \u003cbr /\u003e\n            \u003csub\u003e\u003cb\u003eDongfu Jiang\u003c/b\u003e\u003c/sub\u003e\n        \u003c/a\u003e\n    \u003c/td\u003e\n    \u003ctd align=\"center\"\u003e\n        \u003ca href=\"https://github.com/Zhuofeng-Li\"\u003e\n            \u003cimg src=\"https://github.com/Zhuofeng-Li.png\" width=\"75px;\" alt=\"Zhuofeng Li\"/\u003e\n            \u003cbr /\u003e\n            \u003csub\u003e\u003cb\u003eZhuofeng Li\u003c/b\u003e\u003c/sub\u003e\n        \u003c/a\u003e\n    \u003c/td\u003e\n    \u003ctd align=\"center\"\u003e\n        \u003ca href=\"https://github.com/EigenTom\"\u003e\n            \u003cimg src=\"https://github.com/EigenTom.png\" width=\"75px;\" alt=\"Yi Lu\"/\u003e\n            \u003cbr /\u003e\n            \u003csub\u003e\u003cb\u003eYi Lu\u003c/b\u003e\u003c/sub\u003e\n        \u003c/a\u003e\n    \u003c/td\u003e\n    \u003ctd align=\"center\"\u003e\n        \u003ca href=\"https://github.com/cogito233\"\u003e\n            \u003cimg src=\"https://github.com/cogito233.png\" width=\"75px;\" alt=\"Zhiheng Lvu\"/\u003e\n            \u003cbr /\u003e\n            \u003csub\u003e\u003cb\u003eZhiheng Lvu\u003c/b\u003e\u003c/sub\u003e\n        \u003c/a\u003e\n    \u003c/td\u003e\n    \u003ctd align=\"center\"\u003e\n        \u003ca href=\"https://github.com/erenup\"\u003e\n            \u003cimg src=\"https://github.com/erenup.png\" width=\"75px;\" alt=\"Ping Nie\"/\u003e\n            \u003cbr /\u003e\n            \u003csub\u003e\u003cb\u003ePing Nie\u003c/b\u003e\u003c/sub\u003e\n        \u003c/a\u003e\n    \u003c/td\u003e\n\u003c/tr\u003e\n\u003c/table\u003e\n\n## Advisors\n\n\u003ctable\u003e\n\u003ctr\u003e\n    \u003ctd align=\"center\"\u003e\n        \u003ca href=\"https://github.com/wenhuchen\"\u003e\n            \u003cimg src=\"https://github.com/wenhuchen.png\" width=\"75px;\" alt=\"Wenhu Chen\"/\u003e\n            \u003cbr /\u003e\n            \u003csub\u003e\u003cb\u003eWenhu Chen\u003c/b\u003e\u003c/sub\u003e\n        \u003c/a\u003e\n    \u003c/td\u003e\n    \u003ctd align=\"center\"\u003e\n        \u003ca href=\"https://github.com/P2333\"\u003e\n            \u003cimg src=\"https://github.com/P2333.png\" width=\"75px;\" alt=\"Tianyu Pang\"/\u003e\n            \u003cbr /\u003e\n            \u003csub\u003e\u003cb\u003eTianyu Pang\u003c/b\u003e\u003c/sub\u003e\n        \u003c/a\u003e\n    \u003c/td\u003e\n    \u003ctd align=\"center\"\u003e\n        \u003ca href=\"https://github.com/duchao0726\"\u003e\n            \u003cimg src=\"https://github.com/duchao0726.png\" width=\"75px;\" alt=\"Chao Du\"/\u003e\n            \u003cbr /\u003e\n            \u003csub\u003e\u003cb\u003eChao Du\u003c/b\u003e\u003c/sub\u003e\n        \u003c/a\u003e\n    \u003c/td\u003e\n\u003c/tr\u003e\n\u003c/table\u003e\n\n## Acknowledgements\n\nWe thank the following open-source projects for making verl-tool possible:\n- [VLLM](https://github.com/vllm-project/vllm) and [SGLang](https://github.com/sgl-project/sglang) for their fast LLM inference support!\n- [verl](https://github.com/volcengine/verl) for the excellent RL framework design.\n- [SearchR1](https://github.com/PeterGriffinJin/Search-R1), [RAGEN](https://github.com/RAGEN-AI/RAGEN), and [ToRL](https://github.com/GAIR-NLP/ToRL) for their early-stage exploration of tool-agent RL training.\n\nWe thank [Netmind.AI](https://www.netmind.ai/), [SeaAI Lab](https://sail.sea.com/), and [Map](https://huggingface.co/m-a-p) for GPU support!\n\n## Community Projects Inspired by Verl-Tool\n- [AgentFlow](https://github.com/lupantech/AgentFlow): In-the-Flow Agentic System Optimization\n\n## Star History\n\n[![Star History Chart](https://api.star-history.com/svg?repos=TIGER-AI-Lab/verl-tool\u0026type=Date)](https://www.star-history.com/#TIGER-AI-Lab/verl-tool\u0026Date)\n\n\n## Badge\n\n[![MseeP.ai Security Assessment Badge](https://mseep.net/pr/tiger-ai-lab-verl-tool-badge.png)](https://mseep.ai/app/tiger-ai-lab-verl-tool)\n\n## Citation\n```bibtex\n@article{jiang2025verltool,\n  title={VerlTool: Towards Holistic Agentic Reinforcement Learning with Tool Use},\n  author={Jiang, Dongfu and Lu, Yi and Li, Zhuofeng and Lyu, Zhiheng and Nie, Ping and Wang, Haozhe and Su, Alex and Chen, Hui and Zou, Kai and Du, Chao and others},\n  journal={arXiv preprint arXiv:2509.01055},\n  year={2025}\n}\n```\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Ftiger-ai-lab%2Fverl-tool","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Ftiger-ai-lab%2Fverl-tool","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Ftiger-ai-lab%2Fverl-tool/lists"}