{"paper":{"title":"Latent Geometry Beyond Search: Amortizing Planning in World Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"In a pretrained world model whose latent space is regularized for smoothness and uniformity, a goal-conditioned inverse dynamics model can replace online search while matching its performance at far lower cost.","cross_cats":["cs.LG"],"primary_cat":"cs.RO","authors_text":"Hoang Nguyen, Xiaohao Xu, Xiaonan Huang","submitted_at":"2026-05-09T06:36:23Z","abstract_excerpt":"Modern vision-based world models can represent observations as compact yet expressive latent manifolds, but fast goal-oriented planning in these spaces remains challenging. This raises a central question: when does a learned representation simplify control, rather than merely enabling prediction? We study this question in a pretrained LeWorldModel, whose latent geometry is regularized for smoothness and uniformity. Our key insight is that, under such geometry, planning can be amortized into a latent inverse-dynamics mapping instead of requiring online search. We therefore replace iterative pla"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Under such geometry, planning can be amortized into a latent inverse-dynamics mapping instead of requiring online search. Empirically, the GC-IDM matches or exceeds CEM in seven of eight environment-protocol settings while reducing per-decision cost by 100-130x.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the smoothness and uniformity regularization already present in the pretrained LeWorldModel is sufficient for a learned inverse-dynamics map to capture the planning structure that would otherwise require online search.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"In regularized latent spaces of world models, planning can be amortized into a goal-conditioned inverse dynamics model that matches CEM performance at 100-130x lower per-decision cost.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"In a pretrained world model whose latent space is regularized for smoothness and uniformity, a goal-conditioned inverse dynamics model can replace online search while matching its performance at far lower cost.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"2202f9d84f78fcc559c8dba1494380c6ea42d63b587c5a393101dc099ac46d06"},"source":{"id":"2605.08732","kind":"arxiv","version":2},"verdict":{"id":"dc78a3c9-de34-4366-9a46-5359115c9a10","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-12T01:48:27.746074Z","strongest_claim":"Under such geometry, planning can be amortized into a latent inverse-dynamics mapping instead of requiring online search. Empirically, the GC-IDM matches or exceeds CEM in seven of eight environment-protocol settings while reducing per-decision cost by 100-130x.","one_line_summary":"In regularized latent spaces of world models, planning can be amortized into a goal-conditioned inverse dynamics model that matches CEM performance at 100-130x lower per-decision cost.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the smoothness and uniformity regularization already present in the pretrained LeWorldModel is sufficient for a learned inverse-dynamics map to capture the planning structure that would otherwise require online search.","pith_extraction_headline":"In a pretrained world model whose latent space is regularized for smoothness and uniformity, a goal-conditioned inverse dynamics model can replace online search while matching its performance at far lower cost."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.08732/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"claim_evidence","ran_at":"2026-05-20T09:02:01.944203Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"ai_meta_artifact","ran_at":"2026-05-19T22:35:37.212714Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_title_agreement","ran_at":"2026-05-19T14:31:17.596485Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T10:50:48.989575Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"368724d88641112b646f152e62760005e93a14096d22eedd07f290468511c103"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}