AI brief
The research presents REFACTOR-VLA, a method for unsupervised library learning of typed motor programs, noting that most current vision-language-action models such as OpenVLA, π0, RT-2, and RDT-1B are monolithic and generate raw motor commands or short action sequences without reusable abstractions.
Written by AI from Apple Machine Learning Research's published text.
Source
“Most current vision-language-action (VLA) models—such as OpenVLA, π0, RT-2, and RDT-1B—are “monolithic.” This means they generate raw motor commands or very short sequences of actions, without organizing behaviors into reusable, well-defined abstractions.”
A paper, laboratory publication or research report.
Read original story ↗