{"id":26069,"date":"2026-04-29T00:17:23","date_gmt":"2026-04-28T15:17:23","guid":{"rendered":"https:\/\/aireviewirush.com\/?p=26069"},"modified":"2026-04-29T00:17:23","modified_gmt":"2026-04-28T15:17:23","slug":"gradient-based-planning-for-world-fashions-at-longer-horizons","status":"publish","type":"post","link":"https:\/\/aireviewirush.com\/?p=26069","title":{"rendered":"Gradient-based planning for world fashions at longer horizons"},"content":{"rendered":"


\n<\/p>\n

\n
\n \"BallNav
\n \"Push-T\n<\/div>\n

By Michael Psenka<\/a>, Mike Rabbat<\/a>, Aditi Krishnapriyan<\/a>, Yann LeCun<\/a>, Amir Bar<\/a><\/strong><\/p>\n

GRASP<\/strong> is a brand new gradient-based planner for discovered dynamics (a \u201cworld mannequin\u201d) that makes long-horizon planning sensible by (1) lifting the trajectory into digital states so optimization is parallel throughout time, (2) including stochasticity on to the state iterates for exploration, and (3) reshaping gradients so actions get clear indicators whereas we keep away from brittle \u201cstate-input\u201d gradients by means of high-dimensional imaginative and prescient fashions.<\/p>\n

Giant, discovered world fashions have gotten more and more succesful. They’ll predict lengthy sequences of future observations in high-dimensional visible areas and generalize throughout duties in ways in which had been tough to think about a couple of years in the past. As these fashions scale, they begin to look much less like task-specific predictors and extra like general-purpose simulators.<\/p>\n

However having a strong predictive mannequin isn’t the identical as with the ability to use it successfully for management\/studying\/planning. In follow, long-horizon planning with fashionable world fashions stays fragile: optimization turns into ill-conditioned, non-greedy construction creates unhealthy native minima, and high-dimensional latent areas introduce refined failure modes.<\/p>\n

On this weblog submit, I describe the issues that motivated this mission and our method to deal with them: why planning with fashionable world fashions might be surprisingly fragile, why lengthy horizons are the actual stress take a look at, and what we modified to make gradient-based planning far more strong.<\/p>\n\n

\n

This weblog submit discusses work achieved with Mike Rabbat, Aditi Krishnapriyan, Yann LeCun, and Amir Bar (* denotes equal advisorship), the place we suggest GRASP.<\/p>\n<\/blockquote>\n\n

\n
\n

Table of Contents<\/p>\n