{"paper":{"title":"From Next Token Prediction to (STRIPS) World Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Next-token prediction on action traces yields STRIPS world models accurate enough for planning on unseen states and goals.","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Carlos N\\'u\\~nez-Molina, Hector Geffner, Vicen\\c{c} G\\'omez","submitted_at":"2025-09-16T14:03:58Z","abstract_excerpt":"We study whether next-token prediction can yield world models that truly support planning, in a controlled symbolic setting where propositional STRIPS action models are learned from action traces alone and correctness can be evaluated exactly. We introduce two architectures. The first is the STRIPS Transformer, a symbolically aligned model grounded in theoretical results linking transformers and the formal language structure of STRIPS domains. The second is a standard transformer architecture without explicit symbolic structure built in, for which we study different positional encoding schemes"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Both the STRIPS Transformer and a standard transformer with stick-breaking attention can be used to produce models that support planning with off-the-shelf STRIPS planners over exponentially many unseen initial states and goals.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The learned next-token models are sufficiently accurate and complete to serve as drop-in STRIPS action models for arbitrary unseen states and goals, with correctness evaluated exactly in the symbolic setting (abstract, evaluation section implied by results on generalization and planning performance).","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Transformers trained via next-token prediction on action traces can learn STRIPS action models that support planning over exponentially many unseen initial states and goals.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Next-token prediction on action traces yields STRIPS world models accurate enough for planning on unseen states and goals.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"cda8e7471192580d7771e5564c42e9ff1cc2359f7c6a8563565fc6171b18c921"},"source":{"id":"2509.13389","kind":"arxiv","version":7},"verdict":{"id":"4bac0768-2c65-4d4a-8ffc-84b0840420c2","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-18T16:17:00.156670Z","strongest_claim":"Both the STRIPS Transformer and a standard transformer with stick-breaking attention can be used to produce models that support planning with off-the-shelf STRIPS planners over exponentially many unseen initial states and goals.","one_line_summary":"Transformers trained via next-token prediction on action traces can learn STRIPS action models that support planning over exponentially many unseen initial states and goals.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The learned next-token models are sufficiently accurate and complete to serve as drop-in STRIPS action models for arbitrary unseen states and goals, with correctness evaluated exactly in the symbolic setting (abstract, evaluation section implied by results on generalization and planning performance).","pith_extraction_headline":"Next-token prediction on action traces yields STRIPS world models accurate enough for planning on unseen states and goals."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2509.13389/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"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"}