MetaVLA通过元学习实现高效多任务训练，显著降低机器人模型的调优成本。
MetaVLA: Unified Meta Co-training For Efficient Embodied Adaption
- 引入上下文感知的元协同训练，统一多任务微调流程。
- 在LIBERO上减少75%训练步数，GPU时间节省76%，长程任务提升8.0%。
- 适合追求低资源、强泛化的具身智能系统开发者使用。
视觉-语言-动作（VLA）模型在具身推理中展现出潜力，但仍远未成为真正的通用模型——通常需要特定任务微调，计算成本高，且对未见任务泛化能力差。本文提出MetaVLA，一种统一的、骨干网络无关的后训练框架，实现高效可扩展的对齐。MetaVLA引入上下文感知的元协同训练，将多样目标任务整合至单一微调阶段，并利用结构多样的辅助任务提升域内泛化能力。相比简单的多任务SFT，MetaVLA采用轻量级元学习机制（源自注意力神经过程），可在极少架构改动和推理开销下快速适应多种场景。在LIBERO基准测试中，使用六个辅助任务的MetaVLA相较于OpenVLA，在长程任务上性能提升最高达8.0%，训练步数从24万降至7.5万，GPU耗时减少约76%。结果表明，可扩展的低资源后训练是可行的，为通用具身智能体铺平道路。代码将公开。
原文摘要 · Abstract (English)
Vision-Language-Action (VLA) models show promise in embodied reasoning, yet remain far from true generalists-they often require task-specific fine-tuning, incur high compute costs, and generalize poorly to unseen tasks. We propose MetaVLA, a unified, backbone-agnostic post-training framework for efficient and scalable alignment. MetaVLA introduces Context-Aware Meta Co-Training, which consolidates diverse target tasks into a single fine-tuning stage while leveraging structurally diverse auxiliary tasks to improve in-domain generalization. Unlike naive multi-task SFT, MetaVLA integrates a lightweight meta-learning mechanism-derived from Attentive Neural Processes-to enable rapid adaptation from diverse contexts with minimal architectural change or inference overhead. On the LIBERO benchmark, MetaVLA with six auxiliary tasks outperforms OpenVLA by up to 8.0% on long-horizon tasks, reduces training steps from 240K to 75K, and cuts GPU time by ~76%. These results show that scalable, low-resource post-training is achievable-paving the way toward general-purpose embodied agents. Code will be available.
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。