arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2602.10085cs.AI
收藏

CODE-SHARP: 连续开放发现和演化的技能作为层次奖励程序

CODE-SHARP: Continuous Open-ended Discovery and Evolution of Skills as Hierarchical Reward Programs

Imperial College London(帝国理工学院伦敦分校);Sony Interactive Entertainment(索尼互动娱乐)

Richard Bornemann, Pierluigi Vito Amadori, Antoine Cully

AI总结:

该研究提出CODE-SHARP框架,通过基础模型自主发现和演化技能作为层次奖励程序,实现通用智能体政策的从零开始强化学习,无需预定义奖励,有效学习长周期技能。

摘要:

一般智能的核心特征是能够自主扩展和演化其掌握的技能集。尽管最近基于基础模型(FM)的方法在这一目标上显示出有希望的结果,但它们通常依赖于显著的人工工程,限制了其在新环境中的可转移性。为了解决这个问题,我们引入了连续开放发现和演化技能作为层次奖励程序(CODE-SHARP)框架,该框架利用基础模型来自主增长和演化一个编码技能的Python程序档案,通过强化学习训练通用智能体策略。这些程序被称为技能作为层次奖励程序(SHARPs),每个程序编码一个局部成功条件和一组被委托给先前发现的SHARPs的先决条件。在运行时,SHARPs根据当前状态动态路由智能体通过其先决条件链,奖励沿途的每个完成,要求智能体仅学习每个新SHARP引入的边际行为,从而在无需预定义奖励的情况下高效学习长周期技能。在Craftax-Classic和XLand上,由CODE-SHARP完全自主训练的智能体在中位性能上比先前工作高出6倍和2.6倍,并且是唯一能够制作铁工具和开采钻石的智能体。在扩展的Craftax上,CODE-SHARP在超过90个发现的SHARPs上训练通用智能体,使其能够零样本解决具有挑战性的长周期任务,与基于真实奖励训练的智能体表现相当。

英文摘要:

A core quality of general intelligence is the ability to open-endedly expand and evolve its set of mastered skills autonomously. While recent Foundation Model (FM) driven approaches have shown promising results towards this goal, they typically rely on significant human-in-the-loop engineering, limiting their transferability to novel environments. To address this, we introduce Continuous Open-ended Discovery and Evolution of Skills as Hierarchical Reward Programs (CODE-SHARP), a framework that leverages FMs to open-endedly grow and evolve an archive of Python programs encoding skills to train a generalist agent policy entirely from scratch via reinforcement learning, directly from source code. These programs, termed Skills as Hierarchical Reward Programs (SHARPs), each encode a local success condition and a set of prerequisites delegated to previously discovered SHARPs. At runtime, SHARPs dynamically route the agent through their prerequisite chain based on the current state, rewarding each completion along the way, requiring the agent to learn only the marginal behaviour each new SHARP introduces, enabling efficient learning of long-horizon skills without any pre-defined rewards. On Craftax-Classic and XLand, agents trained fully autonomously by CODE-SHARP outperform previous works by 6x and 2.6x in median performance and are the only agents capable of crafting iron tools and mining diamonds. Scaled to Craftax-Extended, CODE-SHARP trains a generalist agent on over 90 discovered SHARPs, enabling the agent to solve challenging long-horizon tasks zero-shot, matching agents trained on ground-truth rewards.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑