通过知识分解实现高效形态控制与跨任务迁移，大幅降低模型体积和训练成本。
Knowledge Diversion for Efficient Morphology Control and Policy Transfer
- 将Transformer权重分解为共享与专用组件，动态选择激活
- 跨任务迁移样本效率提升3倍，单智能体部署模型缩小17倍
- 适合需要快速适配新任务和轻量化部署的机器人控制场景
通用形态控制旨在学习可泛化于异构智能体形态的通用策略，基于Transformer的控制器虽受欢迎，但计算开销大且跨任务泛化能力有限，新任务需从头训练。为此，我们提出DivMorph，一种模块化训练范式，通过奇异值分解（SVD）在训练前将随机初始化的Transformer权重分解为因子单元，并利用动态软门控机制根据任务和形态嵌入调节这些单元，将其分离为共享的“learngenes”和形态/任务特定的“tailors”，实现知识解耦。通过选择性激活相关组件，DivMorph支持可扩展、高效的策略部署，并能有效迁移到新任务。大量实验表明，DivMorph在跨任务迁移中样本效率比直接微调提升3倍，在单智能体部署中模型规模减少17倍，达到当前最优性能。
原文摘要 · Abstract (English)
Universal morphology control aims to learn a universal policy that generalizes across heterogeneous agent morphologies, with Transformer-based controllers emerging as a popular choice. However, such architectures incur substantial computational costs, resulting in high deployment overhead, and existing methods exhibit limited cross-task generalization, necessitating training from scratch for each new task. To this end, we propose \textbf{DivMorph}, a modular training paradigm that leverages knowledge diversion to learn decomposable controllers. DivMorph factorizes randomly initialized Transformer weights into factor units via SVD prior to training and employs dynamic soft gating to modulate these units based on task and morphology embeddings, separating them into shared \textit{learngenes} and morphology- and task-specific \textit{tailors}, thereby achieving knowledge disentanglement. By selectively activating relevant components, DivMorph enables scalable and efficient policy deployment while supporting effective policy transfer to novel tasks. Extensive experiments demonstrate that DivMorph achieves state-of-the-art performance, achieving a 3$\times$ improvement in sample efficiency over direct finetuning for cross-task transfer and a 17$\times$ reduction in model size for single-agent deployment.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。