AI Paper Atlas

Research map / Optimization / Optimizers

Parallel Stochastic Gradient-Based Planning for World Models

Michael Psenka, Michael Rabbat, Aditi S. Krishnapriyan, Yann LeCun, Amir Bar

ICML 2026 · Optimization › Optimizers · approach: Differentiating through embedded optimization · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add periodic full-rollout GD sync steps (ablation)GRASP · Push-T H=40accuracy (%): 1.6 -> 59.0Improved
use 25 sync steps rather than 10 (ablation)GRASP · Push-T H=40accuracy (%): 48.0 -> 59.0Improved
increase sync steps from 25 to 50 (ablation)GRASP · Push-T H=40accuracy (%): 59.0 -> 58.0No clear effect
add Langevin state noise (sigma 0.5 vs none) (ablation)GRASP · Push-T H=40accuracy (%): 54.8 -> 59.0Improved
raise noise from sigma 0.5 to 1.0 (ablation)GRASP · Push-T H=40accuracy (%): 59.0 -> 50.6Hurt
stop gradients through world model state inputs (ablation)GRASP · Push-T H=40accuracy (%): 46.6 -> 59.0Improved
GRASP versus CEM and GD at long horizonGRASP · Push-T H=60success rate (%): 7.2 (CEM), 16.4 (GD) -> 26.2Improved
GRASP versus CEM at short horizon H=40GRASP · Push-T H=40success rate (%): 61.4 (CEM) vs 59.0No clear effect
GRASP compared to baselines at short horizonsGRASP · Push-T, PointMaze, WallSingle H=10-30success rate (%): Push-T H=20: 96.0 (CEM) vs 89.2No clear effect
replace GD sync with small CEM syncGRASP on le-wm PushT · le-wm PushT H=75success rate: 26% (GD) -> 46%Improved
apply GRASP to le-wm ReacherGRASP · le-wm Reacher H=75success rate: 42% (GD) -> 70%Improved
linearly decay state noise and goal coefficient between syncsGRASP · jepa-wm Metaworld H=30success rate: 12% (CEM) -> 19%Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy