Recursive Language Models (RLMs) - Let's build the coolest agents ever! (Theory & Code)

In this video, we explain how Recursive Language Models or RLMs work, look at actual trajectories on real problems, see how to implement it from scratch using Deno and Pyodide, and discuss what their key features and benefits are. RLMs are an inference technique where a LLM interacts with arbitrarily long prompts through an external REPL. The LLM can write code to explore, decompose and transform the prompt. It can recursively invoke sub-agents to complete smaller subtasks too! Crucially, the subagent responses do not get automatically loaded into the parent agent's context, it gets returned …