ChatPaper.aiChatPaper

PokéLLMon: A Human-Parity Agent for Pokémon Battles with Large Language Models

February 2, 2024
Authors: Sihao Hu, Tiansheng Huang, Ling Liu
cs.AI

Abstract

We introduce Pok\'eLLMon, the first LLM-embodied agent that achieves human-parity performance in tactical battle games, as demonstrated in Pok\'emon battles. The design of Pok\'eLLMon incorporates three key strategies: (i) In-context reinforcement learning that instantly consumes text-based feedback derived from battles to iteratively refine the policy; (ii) Knowledge-augmented generation that retrieves external knowledge to counteract hallucination and enables the agent to act timely and properly; (iii) Consistent action generation to mitigate the panic switching phenomenon when the agent faces a powerful opponent and wants to elude the battle. We show that online battles against human demonstrates Pok\'eLLMon's human-like battle strategies and just-in-time decision making, achieving 49\% of win rate in the Ladder competitions and 56\% of win rate in the invited battles. Our implementation and playable battle logs are available at: https://github.com/git-disl/PokeLLMon.

PDF323December 15, 2024