Democratizing Reinforcement Learning for LLMs

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

Official implementation of GDPO: Group reward-Decoupled Normalization Policy Optimization for Multi-reward RL Optimization

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

Revisiting Mid-training in the Era of Reinforcement Learning Scaling

Official repository of DARE: Diffusion Large Language Models Alignment and Reinforcement Executor

[arxiv: 2512.19673] Bottom-up Policy Optimization: Your Language Model Policy Secretly Contains Internal Policies

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

qwen3-base family of models RL on gsm8k using verl, is there an RL power law on downstream tasks?

PersonaMem-v2: Towards Personalized Intelligence via Learning Implicit User Personas and Agentic Memory

Long COT RFT and Reinforcement Learning Creates Generalize

Using automated curriculum learning to enhance LLM's RL training process.

[ACL 2026 main] DGPO: Distillation-Guided Policy Optimization for Preserving Agentic RAG Capabilities

A list of uv environments templates for LLM development.

AVR: Learning Adaptive Reasoning Paths for Efficient Visual Reasoning

Sample for Fine-Tuning LLMs & VLMs

Training Script using VeRL for multi-turn GPRO w/ MCP tool-calling

RLVεR: Reinforcement Learning with Verifiable Noisy Rewards

[ACL 2026 Findings] SWE-AGILE: A Software Agent Framework for Efficiently Managing Dynamic Reasoning Context

Fork of VeRL to support Turing Family of GPUs