An overview of reinforcement learning systems for training LLM agents, covering core operations like rollouts and optimizer steps, the split between inference engines and trainer engines, and why specialized frameworks exist.

Save to open this file