Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Li, Xiang, Hao, Yiyang, Fulop, Doug
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2505.03947
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866909602983968768
author Li, Xiang
Hao, Yiyang
Fulop, Doug
author_facet Li, Xiang
Hao, Yiyang
Fulop, Doug
contents One of the primary aspirations in reinforcement learning research is developing general-purpose agents capable of rapidly adapting to and mastering novel tasks. While RL gaming agents have mastered many Atari games, they remain slow and costly to train for each game. In this work, we demonstrate that latest reasoning LLMs with out-of-domain RL post-training can play a challenging Atari game called Frogger under a zero-shot setting. We then investigate the effect of in-context learning and the amount of reasoning effort on LLM performance. Lastly, we demonstrate a way to bootstrap traditional RL method with LLM demonstrations, which significantly improves their performance and sample efficiency. Our implementation is open sourced at https://github.com/AlienKevin/frogger.
format Preprint
id arxiv_https___arxiv_org_abs_2505_03947
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Frog Soup: Zero-Shot, In-Context, and Sample-Efficient Frogger Agents
Li, Xiang
Hao, Yiyang
Fulop, Doug
Artificial Intelligence
One of the primary aspirations in reinforcement learning research is developing general-purpose agents capable of rapidly adapting to and mastering novel tasks. While RL gaming agents have mastered many Atari games, they remain slow and costly to train for each game. In this work, we demonstrate that latest reasoning LLMs with out-of-domain RL post-training can play a challenging Atari game called Frogger under a zero-shot setting. We then investigate the effect of in-context learning and the amount of reasoning effort on LLM performance. Lastly, we demonstrate a way to bootstrap traditional RL method with LLM demonstrations, which significantly improves their performance and sample efficiency. Our implementation is open sourced at https://github.com/AlienKevin/frogger.
title Frog Soup: Zero-Shot, In-Context, and Sample-Efficient Frogger Agents
topic Artificial Intelligence
url https://arxiv.org/abs/2505.03947