RL-Align/RL-Kernel

PyTorch-native post-training at scale

verl/HybridFlow: A Flexible and Efficient RL Post-Training Framework

verl Zero-Mismatch Dense/MoE HuggingFace Rollout

The complete post-training stack for LLMs, plus RL for any problem. SFT, DPO and multi-turn agentic RL, with evolutionary hyperparameter optimization built in.

A high-performance RL training-inference weight synchronization framework, designed to enable second-level parameter updates from training to inference in RL workflows

Scalable toolkit for efficient model reinforcement

Run more RL experiments. Wait less for GPUs.

Stable and Efficient Reinforcement Learning for Trillion-Parameter LLMs

Arctic Training and Inference Platform