【[20星]Compose-RL:一个用于强化学习与人工反馈(RLHF)的框架,旨在简化不同 RLHF 技术的集成,提供模块化和组合式的实验能力,适用于研究人员和实践者】'Compose RL is a framework for Reinforcement Learning with Human Feedback (RLHF), designed to streamline the integration of various RLHF techniques.' GitHub: github.com/databricks/Compose-RL