{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:ROW6UXUXSWJWJZ4JPTI4RGR3Y5","short_pith_number":"pith:ROW6UXUX","schema_version":"1.0","canonical_sha256":"8badea5e97959364e7897cd1c89a3bc754e79b2584a5c1f4c99938ff98b65029","source":{"kind":"arxiv","id":"2607.19749","version":1},"attestation_state":"computed","paper":{"title":"The World Model Remembers, the Actor Forgets: Dream Rehearsal for Continual Model-Based RL","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Gurp Nijjer","submitted_at":"2026-07-22T04:46:49Z","abstract_excerpt":"Model-based reinforcement-learning agents of the DreamerV3 family forget catastrophically when trained on task sequences, even when an unbounded replay buffer preserves every earlier experience. We ask a question the continual-RL literature has assumed an answer to but never measured: which component forgets? Under never-clear replay, pre-registered component-level probes (n=3 seeds throughout) show that the world model retains essentially everything measurable about old tasks -- reward discrimination (retention ratio ~1.0), value estimates, and termination structure -- while the actor's behav"},"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":"2607.19749","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-07-22T04:46:49Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"b273c54aede877b21a477901ad2cfb0cf77d5fb235258123b0dbf2b016387b31","abstract_canon_sha256":"1ea57b5b41ea9e58911d9cf5f3d324f47484b175c1b9927d470b16b5779ef31d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-23T00:24:07.152016Z","signature_b64":"Z/kkVGX9YX3dP5DjTkkekYhx6k5dyRi7FJCff0FbRhQl4Oh7OWeDD8kL9TjHbsU97fkPWh4bB372pi1NFZy+Bw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8badea5e97959364e7897cd1c89a3bc754e79b2584a5c1f4c99938ff98b65029","last_reissued_at":"2026-07-23T00:24:07.151173Z","signature_status":"signed_v1","first_computed_at":"2026-07-23T00:24:07.151173Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The World Model Remembers, the Actor Forgets: Dream Rehearsal for Continual Model-Based RL","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Gurp Nijjer","submitted_at":"2026-07-22T04:46:49Z","abstract_excerpt":"Model-based reinforcement-learning agents of the DreamerV3 family forget catastrophically when trained on task sequences, even when an unbounded replay buffer preserves every earlier experience. We ask a question the continual-RL literature has assumed an answer to but never measured: which component forgets? Under never-clear replay, pre-registered component-level probes (n=3 seeds throughout) show that the world model retains essentially everything measurable about old tasks -- reward discrimination (retention ratio ~1.0), value estimates, and termination structure -- while the actor's behav"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.19749","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/2607.19749/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":"2607.19749","created_at":"2026-07-23T00:24:07.151621+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.19749v1","created_at":"2026-07-23T00:24:07.151621+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.19749","created_at":"2026-07-23T00:24:07.151621+00:00"},{"alias_kind":"pith_short_12","alias_value":"ROW6UXUXSWJW","created_at":"2026-07-23T00:24:07.151621+00:00"},{"alias_kind":"pith_short_16","alias_value":"ROW6UXUXSWJWJZ4J","created_at":"2026-07-23T00:24:07.151621+00:00"},{"alias_kind":"pith_short_8","alias_value":"ROW6UXUX","created_at":"2026-07-23T00:24:07.151621+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/ROW6UXUXSWJWJZ4JPTI4RGR3Y5","json":"https://pith.science/pith/ROW6UXUXSWJWJZ4JPTI4RGR3Y5.json","graph_json":"https://pith.science/api/pith-number/ROW6UXUXSWJWJZ4JPTI4RGR3Y5/graph.json","events_json":"https://pith.science/api/pith-number/ROW6UXUXSWJWJZ4JPTI4RGR3Y5/events.json","paper":"https://pith.science/paper/ROW6UXUX"},"agent_actions":{"view_html":"https://pith.science/pith/ROW6UXUXSWJWJZ4JPTI4RGR3Y5","download_json":"https://pith.science/pith/ROW6UXUXSWJWJZ4JPTI4RGR3Y5.json","view_paper":"https://pith.science/paper/ROW6UXUX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.19749&json=true","fetch_graph":"https://pith.science/api/pith-number/ROW6UXUXSWJWJZ4JPTI4RGR3Y5/graph.json","fetch_events":"https://pith.science/api/pith-number/ROW6UXUXSWJWJZ4JPTI4RGR3Y5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ROW6UXUXSWJWJZ4JPTI4RGR3Y5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ROW6UXUXSWJWJZ4JPTI4RGR3Y5/action/storage_attestation","attest_author":"https://pith.science/pith/ROW6UXUXSWJWJZ4JPTI4RGR3Y5/action/author_attestation","sign_citation":"https://pith.science/pith/ROW6UXUXSWJWJZ4JPTI4RGR3Y5/action/citation_signature","submit_replication":"https://pith.science/pith/ROW6UXUXSWJWJZ4JPTI4RGR3Y5/action/replication_record"}},"created_at":"2026-07-23T00:24:07.151621+00:00","updated_at":"2026-07-23T00:24:07.151621+00:00"}