超越单纯的环境扩展：为多模态智能体学习设计有效的环境分布

摘要

近期的工作通过构建大规模多模态环境池来训练智能体。然而，我们发现单纯增加多模态环境的数量并不总是有益。我们通过一系列实验进一步分析了当前多模态环境分布中存在的局限。基于这些发现，我们从两个维度研究如何构建更有效的训练环境分布：**多样性**和**难度结构**。在多样性方面，我们提出**能力感知环境选择（AES）**以获取多样化的环境集合。在难度结构方面，我们提出**层次化难度课程（HDC）**，通过两个难度层级组织课程学习：约束减弱和状态规模递进。实验表明，AES和HDC能有效提升多模态智能体的训练效果。

English

Recent works train agents by constructing large-scale multimodal environment pools. However, we find that simply increasing the number of multimodal environments does not always benefit. We further analyze the limitations in current multimodal environment distributions through a series of experiments. Based on these findings, we study how to build more effective training environment distributions from two dimensions: **diversity** and **difficulty structure**. For diversity, we propose **Ability-aware Environment Selection (AES)** to obtain diverse environment sets. For difficulty structure, we propose **Hierarchical Difficulty Curriculum (HDC)**, which organizes curriculum learning through two difficulty levels: harness weakening and state-scale progression. Experiments show that AES and HDC effectively improve multimodal agent training.