Deep reinforcement learning for complex environment 1

Abstract

This project delves into creating an Artificial Intelligence (AI) model tailored for the game Werewolf, focusing on strategic gameplay and realistic interaction. It aims to craft an AI by understanding and adapting to the intricate dynamics of the game and the nuances of player behaviour using deep reinforcement learning and large language models (LLMs). Through this, the project also aims to find out weaknesses in logic of LLMs and how these weaknesses can be alleviated with the aid of reinforcement learning and other means.