Skip to content

12.03 — Section Resources

Overview

This lesson provides references to the academic paper, implementation resources, and tools used throughout the Reflexion Agent section.


Research Paper

Resource Details
Paper Reflexion: Language Agents with Verbal Reinforcement Learning
Authors Noah Shinn, Federico Cassano, Ashwin Gopinath, Karthik Narasimhan, Shunyu Yao
Institutions Northeastern University, MIT, Princeton University
Key Idea Agents that verbally reflect on task feedback signals, maintaining a self-reflective text in episodic memory to improve decision-making in subsequent trials

The Reflexion paper formalizes the pattern of iterative self-improvement through verbal self-reflection — exactly what this section implements.


Implementation References

Resource Description
LangChain Blog The LangChain team (particularly Lance Martin) published a blog post implementing the Reflexion architecture with LangGraph. This course's implementation is a refactored, simplified version of that work.
Course Repository All code for this section is available in the course repository under projects/reflexion-agent branch.

Tools & Services

Tool Purpose Link
OpenAI GPT-4 Turbo LLM for article generation, self-critique, and revision platform.openai.com
Tavily Search API Web search engine optimized for LLM applications tavily.com
LangSmith Tracing and observability for LangChain/LangGraph smith.langchain.com
LangGraph Graph-based workflow orchestration framework langchain-ai.github.io/langgraph