Pretrained to Imagine, Fine-Tuned to Act: The Rise of World-Action Models
NVIDIA introduces World-Action Models (WAMs), robot policies fine-tuned from pretrained world/video models to act.
“WAM World-Action Model: a policy that starts from a pretrained world-model or video”
NVIDIA's developer blog frames an emerging class of robot policies called World-Action Models (WAMs) that build on pretrained world or video models, paralleling Vision-Language-Action models built on VLMs. This signals a shift in robot learning toward leveraging generative world models for action generation, relevant to the embodied AI and robotics foundation-model trend.