arXiv cs.AI· Zipeng Qiu, Shusheng Xu, Jiale Li, Yukun Zhou, Chenyue Li, Wenshuang Li, Tongkai Yang, Binhang Yuan·· 1 小时前AI 评分51
arXiv 论文提出代码即策略,让视觉游戏智能体离线思考实时行动
Think Offline, Act in Real Time: Self-Improving Visual Game Agents through Code
AI 导读
arXiv 论文(arXiv:2611.24590)研究实时视觉游戏中智能体如何从少量尝试中自我改进,对比模型即策略与代码即策略两种方案。在 170 个 GameWorld 任务上,每次运行最多八次一分钟执行,代码即策略达到 47.80% 平均进度,高于训练 2B 控制器的 30.90%,并在全部五个游戏类别领先;仅改进单个游戏控制器权重收益有限,结果支持让模型离线思考并把所学存入实时执行的程序。
正文 · 原文
Abstract:Pre-trained models can read screens and reason about games, yet on an unfamiliar game, an agent still has to learn to play. We ask how an agent can self-improve on such a game from a few of its own attempts, in real-time visual games where the agent sees only the screen and the game keeps running while the agent thinks. We study two kinds of policy that can carry this improvement. In model as policy, a model selects actions during play; we adapt real-time chunking so that committed actions keep executing while the model reasons, and train a 2B controller for this interface. In code as policy, a model writes, tests, and revises a program between attempts, and the program acts during play. Across 170 GameWorld tasks, with at most eight one-minute executions per run, code as policy reaches 47.80% mean progress, against 30.90% for our trained controller, and leads in all five game categories. Improving the controller's weights on a single game, by contrast, yields only modest gains. These results favor letting a model think offline and storing what it learns in a program that acts in real time. Project repository: this https URL
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2611.24590 [cs.AI] |
| (or arXiv:2611.24590v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2611.24590 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Zipeng Qiu [view email]
[v1]
Wed, 18 Nov 2026 15:18:22 UTC (1,266 KB)
来源:arXiv cs.AI · arxiv.org