Just Action Model: decoupling what separates VLA from world-action models
Yonghao Yu · hf · 2026-10-28
A University of Tokyo / NII paper, Vision-Language-Action or World-Action? Just Action Models, argues the VLA-vs-WAM debate bundles three separable choices.
- The confounded variables: backbone initialization (vision-language vs. visual generators), the objective applied to future observations, and whether the machinery for future-observation losses is kept at inference
- They propose a common framework, the Just Action Model (JAM): a pretrained visual backbone with an action expert trained with an action loss plus a foresight loss that trains current-observation features to predict frozen encodings of future observations, reducing computation at inference
- Conclusion: the gap between a VLA and a WAM cannot be attributed to any single choice
A substantive framework contribution in robot learning.
More from Embodied
- Houston Robotaxi called most 'fake' AV service map: 40 min to cross 2 blocks — reed · 2026-10-28
- Indie roboticist gets first anthropometry-based robot walking with agent-driven design loop — IanPritchard · 2026-10-28
- Reward AI's OM-1 Learns Manipulation Skills From Human Demonstrations for Cross-Body Transfer — CyberRobooo · 2026-10-28
- 4 Stanford freshmen build assistive walker robot in 36 hours, win grand prize at CalHacks 13.0 — Scobleizer · 2026-10-28
- NVIDIA's ThinkWorld learns robot skills from unlabeled human videos via chain-of-dynamics reasoning — nvidia · 2026-10-28
- NVIDIA Research releases Foundation4D: unified 4D perception model turns camera video into 3D scenes, no lidar needed — Scobleizer · 2026-10-28