video1.0<iframe src="https://www.loom.com/embed/98b1c9ca826546dea5a5206be9583e25" frameborder="0" width="1108" height="831" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>8311108Loomhttps://www.loom.com8311108https://cdn.loom.com/sessions/thumbnails/98b1c9ca826546dea5a5206be9583e25-2fade3805b184dbd.gif129.689In this video, I present our project at Week Half 3, where agents engage in real-time Mafia games while adapting their strategies. Initially, they play without strategies, then reflect on their decisions to create dynamic cheat sheets for improvement in subsequent games. We evaluate the effectiveness of these strategies using LLM as a judge, tracking the evolution of their cheat sheets over time. I encourage you to consider how these reflections and updates can enhance performance in similar scenarios. Please share your thoughts on the strategies discussed.