AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

Compositional Planning with Jumpy World Models

Jesse Farebrother, Matteo Pirotta, Andrea Tirinzoni, Marc G Bellemare, Alessandro Lazaric, Ahmed Touati

ICML 2026 · Reinforcement learning › Visual and representation RL · approach: In-context and adaptive RL · 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 compositional planning over GHMs to a base policyGC-BC · OGBench cube-2success rate: 0.15 -> 0.97Improved
Add compositional planning over GHMs to a base policyCRL · OGBench cube-4success rate: 0.00 -> 0.39Improved
Add compositional planning over GHMs to a base policyHFBC · OGBench antmaze-giantsuccess rate: 0.42 -> 0.79Improved
Add compositional planning over GHMs to a base policyGC-TD3 · OGBench antmaze-mediumsuccess rate: 0.65 -> 0.65No clear effect
Add compositional planning over GHMs to a base policyHFBC · OGBench antmaze-mediumsuccess rate: 0.94 -> 0.94No clear effect
Plan over policy sequences (CompPlan) instead of generalized policy improvement (ablation)CompPlan · OGBench long-horizon domainsrelative success rate improvement: 89% relative improvement over GPIImproved
Plan over policy sequences with jumpy models instead of action-level planning with a one-step world model (ablation)CompPlan · OGBench long-horizon domainsrelative success rate improvement: 201% gain over ActionPlanImproved
Use test-time CompPlan with HFBC instead of learned hierarchiesCompPlan · OGBench cube-4success rate: 0.09 -> 0.67 vs SHARSA, 0.00 -> 0.67 vs HIQLImproved
Use CompPlan with HFBC instead of HIQLCompPlan · OGBench antmaze-mediumsuccess rate: 0.96 -> 0.94Hurt
Add horizon consistency loss (TD-HC) to GHM training (ablation)GHM with GC-1S · OGBench antmaze-giantEMD: 7.29 -> 5.25 (28% reduction)Improved
Add horizon consistency loss (TD-HC) to GHM training (ablation)GHM with CRL · OGBench antmaze-giantEMD: 6.77 -> 5.74Improved
Add horizon consistency loss (TD-HC) to GHM training (ablation)GHM with CRL · OGBench cube-1EMD: 1.60 -> 1.57Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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