Building a RCI Chain for Agents with LangChain Expression Language

Explore the implementation of a Recursive Criticism and Improvements (RCI) chain for agents using LangChain Expression Language (LCEL) in this informative video tutorial. Dive into the concepts behind RCI, inspired by recent research papers on language models solving computer tasks and verbal reinforcement learning. Learn about the three key prompts involved in the RCI process and visualize their interaction through a helpful diagram. Follow along with hands-on coding examples, starting with multi-chains and progressing to a full RCI implementation using LCEL. Discover how to combine these chains for improved agent performance. Access the provided Colab notebook and additional resources to further enhance your understanding of this advanced language model application.

Intro
Paper: Language Models Can Solve Computer Task
Paper: Reflexion: Language Agents with Verbal Reinforcement Learning
RCI Recursive Criticism and Improvements
Three Prompts
Three Prompts Diagram
Code Time
Multi Chains
RCI with LCEL
Combined Chain