Practical PCG Through Large Language Models

TL;DR

This paper demonstrates how large language models like GPT-3 can be practically used for procedural content generation in 2D games, specifically generating playable game rooms with limited data through human-in-the-loop fine-tuning.

Contribution

It introduces a method leveraging GPT-3 and human-in-the-loop fine-tuning to generate playable game levels from scarce data in complex constrained scenarios.

Findings

Achieved 37% playable and novel levels from only 60 seed rooms.

Utilized human-in-the-loop fine-tuning for effective level generation.

Applied to a non-trivial game with local and global constraints.

Abstract

Large Language Models (LLMs) have proven to be useful tools in various domains outside of the field of their inception, which was natural language processing. In this study, we provide practical directions on how to use LLMs to generate 2D-game rooms for an under-development game, named Metavoidal. Our technique can harness the power of GPT-3 by Human-in-the-loop fine-tuning which allows our method to create 37% Playable-Novel levels from as scarce data as only 60 hand-designed rooms under a scenario of the non-trivial game, with respect to (Procedural Content Generation) PCG, that has a good amount of local and global constraints.