{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:KZCNBLRT7VOSMD7OTB43RLPFPQ","short_pith_number":"pith:KZCNBLRT","schema_version":"1.0","canonical_sha256":"5644d0ae33fd5d260fee9879b8ade57c2964ab504a546b9036a19fbf14b01671","source":{"kind":"arxiv","id":"2605.18813","version":1},"attestation_state":"computed","paper":{"title":"Composition of Memory Experts for Diffusion World Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Aram Davtyan, Pablo Acuaviva Huertos, Paolo Favaro, Sebastian Stapf","submitted_at":"2026-05-12T09:43:10Z","abstract_excerpt":"World models aim to predict plausible futures consistent with past observations, a capability central to planning and decision-making in reinforcement learning. Yet, existing architectures face a fundamental memory trade-off: transformers preserve local detail but are bottlenecked by quadratic attention, while recurrent and state-space models scale more efficiently but compress history at the cost of fidelity. To overcome this trade-off, we suggest decoupling future-past consistency from any single architecture and instead leveraging a set of specialized experts. We introduce a diffusion-based"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2605.18813","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-12T09:43:10Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"e28a9fe27af13d69b36aeb391df93f389211e03f6d2af0792a49f6b50bafd009","abstract_canon_sha256":"4d06aedd7b3ddc4581829ee271234a6169ec173a1ca125928ac27b2eef15e1f6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:06:23.753860Z","signature_b64":"w9zwLfhEoakt9xNcNREwJ8UU32mHVJnqO6KxijABzHvuGLNA1qDWvX/sf8p7YyPOaU2kA2L6AC0TdDxIGHSQBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5644d0ae33fd5d260fee9879b8ade57c2964ab504a546b9036a19fbf14b01671","last_reissued_at":"2026-05-20T00:06:23.753005Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:06:23.753005Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Composition of Memory Experts for Diffusion World Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Aram Davtyan, Pablo Acuaviva Huertos, Paolo Favaro, Sebastian Stapf","submitted_at":"2026-05-12T09:43:10Z","abstract_excerpt":"World models aim to predict plausible futures consistent with past observations, a capability central to planning and decision-making in reinforcement learning. Yet, existing architectures face a fundamental memory trade-off: transformers preserve local detail but are bottlenecked by quadratic attention, while recurrent and state-space models scale more efficiently but compress history at the cost of fidelity. To overcome this trade-off, we suggest decoupling future-past consistency from any single architecture and instead leveraging a set of specialized experts. We introduce a diffusion-based"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.18813","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.18813/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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2605.18813","created_at":"2026-05-20T00:06:23.753160+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.18813v1","created_at":"2026-05-20T00:06:23.753160+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.18813","created_at":"2026-05-20T00:06:23.753160+00:00"},{"alias_kind":"pith_short_12","alias_value":"KZCNBLRT7VOS","created_at":"2026-05-20T00:06:23.753160+00:00"},{"alias_kind":"pith_short_16","alias_value":"KZCNBLRT7VOSMD7O","created_at":"2026-05-20T00:06:23.753160+00:00"},{"alias_kind":"pith_short_8","alias_value":"KZCNBLRT","created_at":"2026-05-20T00:06:23.753160+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KZCNBLRT7VOSMD7OTB43RLPFPQ","json":"https://pith.science/pith/KZCNBLRT7VOSMD7OTB43RLPFPQ.json","graph_json":"https://pith.science/api/pith-number/KZCNBLRT7VOSMD7OTB43RLPFPQ/graph.json","events_json":"https://pith.science/api/pith-number/KZCNBLRT7VOSMD7OTB43RLPFPQ/events.json","paper":"https://pith.science/paper/KZCNBLRT"},"agent_actions":{"view_html":"https://pith.science/pith/KZCNBLRT7VOSMD7OTB43RLPFPQ","download_json":"https://pith.science/pith/KZCNBLRT7VOSMD7OTB43RLPFPQ.json","view_paper":"https://pith.science/paper/KZCNBLRT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.18813&json=true","fetch_graph":"https://pith.science/api/pith-number/KZCNBLRT7VOSMD7OTB43RLPFPQ/graph.json","fetch_events":"https://pith.science/api/pith-number/KZCNBLRT7VOSMD7OTB43RLPFPQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KZCNBLRT7VOSMD7OTB43RLPFPQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KZCNBLRT7VOSMD7OTB43RLPFPQ/action/storage_attestation","attest_author":"https://pith.science/pith/KZCNBLRT7VOSMD7OTB43RLPFPQ/action/author_attestation","sign_citation":"https://pith.science/pith/KZCNBLRT7VOSMD7OTB43RLPFPQ/action/citation_signature","submit_replication":"https://pith.science/pith/KZCNBLRT7VOSMD7OTB43RLPFPQ/action/replication_record"}},"created_at":"2026-05-20T00:06:23.753160+00:00","updated_at":"2026-05-20T00:06:23.753160+00:00"}