BLOG ILLUSTRATION / EXECUTION STORY

Joint WAM prediction targets

Observation and task context are inputs to a shared world-action model. Future-world trajectories and action trajectories are different output branches. This diagram has no mandatory runtime edge from future world to action.

Joint WAM prediction targetsObservation and task context are inputs to a shared world-action model. Future-world trajectories and action trajectories are different output branches. This diagram has no mandatory runtime edge from future world to action. Two targets, shared learning Observation + task Input o_t + instruction Joint WAM Shared learning Future world Observation / latent path Future actions Embodiment commands No universal deployment path
A joint WAM learns future-world trajectories and action trajectories from observation and task context. The joint training distribution does not specify a universal inference path. Some systems use imagined futures, but others deploy a faster action path shaped by prediction supervision. Author-created explanatory diagram. Playback time is illustrative and is not measured runtime. Open full-size SVG