Reflecting on Reflexion – by Noah Shinn and Ashwin Gopinath

Here’s a discussion on our recent paper with an additional experiment that uses GPT-4 to beat past GPT-4 standards. In our recent paper, “Reflexion: An Autonomous Agent with Dynamic Memory and Self-Reflection,” we introduce a framework that allows AI agents to emulate human-like self-reflection and evaluate its performance on the
— Read on nanothoughts.substack.com/p/reflecting-on-reflexion