When More Planning Hurts: Learned Planners for World Models

We also conduct another matched experiment where we only replay the first ten commands in each context and then allow the next solve 4, 8 or 12 updates (subsequent solves all get 8 updates). It comes from a paper posted to arXiv on 30 October 2026. On PushT, our fixed action updater (1,505 parameters) solves 8 of 16 scenes reserved from our configuration selection. The comparison planners perform as follows: Selected Adam (7 tasks solved, total planning and encoding time 218.440s), adapted iCEM (2 tasks solved, total planning and encoding time 49.309s), released LePlanner (16 tasks solved, total planning and encoding time 5.652s). The total planning and encoding time for our action updater is 59.079s. Note that our action updater does not satisfy the task-cost criterion we fixed before evaluation. Among all 16 contexts (8 scene clusters), the 3 budgets get 9, 8 and 7 successes, respectively. Interestingly, the extra updates did not rescue any failing task in this experiment. All the changes in success/failure happen in one scene cluster, where we observe that the extra updates decrease the costs of the selected latent world-model trajectories for our updater, while the tasks that used to be successes become failures. The code, weights, evaluators, execution traces and cost accounting accompany the paper.

Why it counts

We also conduct another matched experiment where we only replay the first ten commands in each context and then allow the next solve 4, 8 or 12 updates (subsequent solves all get 8 updates).

Sources

Primary source: primary source

What is not known

This brief does not claim independent replication. Claims that appear only on X and not in the primary source stay unknown.