训练单一模型通过强化学习掌握跨层级智能体动作
Training One Model to Master Cross-Level Agentic Actions via Reinforcement Learning
论文提出的多轮 GRPO 训练流程与自适应动作粒度切换机制，可迁移到视觉-语言-动作模型（VLA）或 GUI 智能体中，让模型在长程视觉任务里自主决定何时用高层 API、何时用低层像素级操作，从而提升长程推理的效率与泛化。
原文依据 we propose CrossHA, a unified agentic model that masters heterogeneous action spaces and autonomously selects the most effective interface for each step of a trajectory
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论智能体 AI 的范式正从工程化的复杂工作流转向后训练原生模型。然而，现有智能体通常局限于静态、预定义的动作空间——例如仅使用 API、GUI 事件或机器人指令。这种刚性限制了它们在动态环境中的适应性，因为在这些环境中，最优的交互粒度会随上下文而变化。为弥合这一差距，我们提出 CrossHA，一个统一的智能体模型，它掌握异构动作空间，并为轨迹的每一步自主选择最有效的接口。我们引入了一套完整的训练流程，将冷启动监督微调与多轮组相对策略优化（GRPO）算法相结合。该方法使智能体无需人工指定规则即可学习自适应动作切换——在高层次效率与低层次精度之间取得平衡。在开放世界 Minecraft 环境中超过 800 个任务上的大量实验表明，CrossHA 达到了最优性能。通过动态利用多样动作空间的优势，我们的模型显著优于固定动作基线，在长程推理中展现出更优的泛化性与效率。所有代码与模型均可在 https://github.com/CraftJarvis/OpenHA 获取。
英文原文摘要 1322 字符
The paradigm of agentic AI is shifting from engineered complex workflows to post-training native models. However, existing agents are typically confined to static, predefined action spaces--such as exclusively using APIs, GUI events, or robotic commands. This rigidity limits their adaptability in dynamic environments where the optimal granularity of interaction varies contextually. To bridge this gap, we propose CrossHA, a unified agentic model that masters heterogeneous action spaces and autonomously selects the most effective interface for each step of a trajectory. We introduce a comprehensive training pipeline that integrates cold-start supervised fine-tuning with a Multi-Turn Group Relative Policy Optimization (GRPO) algorithm. This approach enables the agent to learn adaptive action switching--balancing high-level efficiency with low-level precision--without human-specified rules. Extensive experiments on over 800 tasks in the open-world Minecraft environment demonstrate that CrossHA achieves state-of-the-art performance. By dynamically leveraging the strengths of diverse action spaces, our model significantly outperforms fixed-action baselines, exhibiting superior generalization and efficiency in long-horizon reasoning. All code and models are available at https://github.com/CraftJarvis/OpenHA.