Gen-Verse/OpenClaw-RL

Registry 信息

项目简介

OpenClaw-RL 是一个基于 Python 的框架，让你只需通过对话就能训练任意智能体，利用强化学习和在线策略蒸馏来个性化你的 Claude Skill，无需手动编码。

要点

• 通过自然语言对话训练智能体
• 在线策略蒸馏实现高效技能学习
• 集成 GRPO 和 RLHF 支持
• 异步编码与记忆系统
• 图形界面应用便于交互
• 集成 SGLang 和 Slime 实现可扩展训练

使用场景

• 通过对话个性化 Claude Skill
• 快速原型开发基于强化学习的智能体
• 强化学习的教学演示
• 以最少代码构建自定义 AI 助手
• 在线策略蒸馏与技能学习研究

README 摘要

<div align="center"> <h1 align="center"> <img src="assets/spacer.png" alt="" width="23" height="40" align="absmiddle" /> OpenClaw-RL<!-- --><sup> <img src="assets/clawistool.png" alt="Claw-RL logo" width="23" height="40" align="absmiddle" /> <sup> </h1> <p><b>Empowering OpenClaw with RL — Train a personalized agent simply by talking to it.</b></p> <p><b>Scalable RL in real-world settings — Agentic RL for terminal, GUI, SWE, and tool-call settings.</b></p> </div> <p align="center"> <img src="https://img.shields.io/badge/⚡_Fully_Async-yellow?style=for-the-badge" alt="Fully Async" /> <img src="https://img.shields.io/badge/💰_Zero_API_or_Zero_GPU-blue?style=for-the-badge" alt="Zero API or Zero GPU" /> <img src="https://img.shields.io/badge/🤖_Personalized-success?style=for-the-badge" alt="Personalized" /> <img src="https://img.shields.io/badge/🛠️_Auto_Optimization-orange?style=for-the-badge" alt="Auto" /> <img src="https://img.shields.io/badge/💬_Language_Feedback-purple?style=for-the-badge" alt="Language Feedback" /> <img src="https://img.shields.io/badge/🧠_Hybrid_RL-red?style=for-the-badge" alt="Hybrid RL" /> <img src="https://img.shields.io/badge/🌍_Real_World_Agentic_RL-green?style=for-the-badge" alt="General Agentic RL" /> <br><br> <a href="https://arxiv.org/abs/2603.10165"><img src="https://img.shields.io/badge/📄_Tech_Report-red?style=flat-square" alt="Tech Report" /></a> <a href="https://yinjjiew.github.io/projects/openclawrl1"><img src="https://img.shields.io/badge/Blog-Page-blue?style=flat-square" alt="OpenClaw-RL Blog" /></a> <a href="https://openclaw.ai"><img src="https://img.shields.io/badge/OpenClaw-Plugin-orange?style=flat-square" alt="OpenClaw Plugin" /></a> <a href="https://github.com/THUDM/slime"><img src="http

话题