verl topic

超简单复现Deepseek-R1-Zero和Deepseek-R1，以「24点游戏」为例。通过zero-RL、SFT以及SFT+RL，以激发LLM的自主验证反思能力。 About Clean, minimal, accessible reproduction of DeepSeek R1-Zero, DeepSeek R1

sylvain-wei

A Framework for LLM-based Multi-Agent Reinforced Training and Inference

TsinghuaC3I

Official repository for "RLVR-World: Training World Models with Reinforcement Learning" (NeurIPS 2025), https://arxiv.org/abs/2505.13934

thuml

Democratizing Reinforcement Learning for LLMs

rllm-org