Distilling Game Code World Model Generation into Lightweight Large Language Models
The paper discusses the generation of Game Code World Models (GameCWMs) using Large Language Models (LLMs). It presents a method to distill the capabilities of generating game environments into smaller models, enhancing accessibility and scalability. The authors introduce a dataset and a verification framework to improve the generation process, demonstrating increased correctness and adherence to game rules.
- ▪Game Code World Models (GameCWMs) are designed to translate game rules into executable code.
- ▪The authors propose a post-training pipeline that combines Supervised Fine-Tuning with Reinforcement Learning to enhance model capabilities.
- ▪A curated dataset of 30 games is introduced to support the generation of GameCWMs.
arXiv cs.AI files mainly under ai research. We currently carry 1,128 of its stories.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Artificial Intelligence arXiv:2605.24375 (cs) [Submitted on 23 May 2026] Title:Distilling Game Code World Model Generation into Lightweight Large Language Models Authors:Tyrone Serapio, Arjun Prakash, Haoyang Xu, Kevin Wang, Amy Greenwald View a PDF of the paper titled Distilling Game Code World Model Generation into Lightweight Large Language Models, by Tyrone Serapio and 4 other authors View PDF HTML (experimental) Abstract:Large Language Models (LLMs) have shown great ability in generating executable code from natural language, opening the possibility of automatically constructing environments for AI agents.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
