{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:2BBY2EFX4U3RO6KU2DZYNG53S7","short_pith_number":"pith:2BBY2EFX","schema_version":"1.0","canonical_sha256":"d0438d10b7e537177954d0f3869bbb97ee2e1678f742df0e28f9852cad35c31a","source":{"kind":"arxiv","id":"2305.10912","version":2},"attestation_state":"computed","paper":{"title":"A Generalist Dynamics Model for Control","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.RO"],"primary_cat":"cs.AI","authors_text":"Arunkumar Byravan, Emilio Parisotto, Ingmar Schubert, Jake Bruce, Jingwei Zhang, Jost Tobias Springenberg, Leonard Hasenclever, Martin Riedmiller, Nicolas Heess, Sarah Bechtle","submitted_at":"2023-05-18T12:14:49Z","abstract_excerpt":"We investigate the use of transformer sequence models as dynamics models (TDMs) for control. We find that TDMs exhibit strong generalization capabilities to unseen environments, both in a few-shot setting, where a generalist TDM is fine-tuned with small amounts of data from the target environment, and in a zero-shot setting, where a generalist TDM is applied to an unseen environment without any further training. Here, we demonstrate that generalizing system dynamics can work much better than generalizing optimal behavior directly as a policy. Additional results show that TDMs also perform well"},"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":"2305.10912","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2023-05-18T12:14:49Z","cross_cats_sorted":["cs.RO"],"title_canon_sha256":"68a52d99921ed9a75d15cf4cc183631d86f7a37f31b3bdb82c84acdbec0ee77b","abstract_canon_sha256":"8a366b1d482a55e577ef91aac532f9c6861567d8018b0d1a22657eab514d301d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:53:37.783502Z","signature_b64":"B8eG/a8DyFGUAriN0oYn6aodFmsjgim88iHBKmp+bA18hfjTGg4TjUIHBbImHcrlirSbRocG65SHGIqARQWSAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d0438d10b7e537177954d0f3869bbb97ee2e1678f742df0e28f9852cad35c31a","last_reissued_at":"2026-07-05T06:53:37.783025Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:53:37.783025Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Generalist Dynamics Model for Control","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.RO"],"primary_cat":"cs.AI","authors_text":"Arunkumar Byravan, Emilio Parisotto, Ingmar Schubert, Jake Bruce, Jingwei Zhang, Jost Tobias Springenberg, Leonard Hasenclever, Martin Riedmiller, Nicolas Heess, Sarah Bechtle","submitted_at":"2023-05-18T12:14:49Z","abstract_excerpt":"We investigate the use of transformer sequence models as dynamics models (TDMs) for control. We find that TDMs exhibit strong generalization capabilities to unseen environments, both in a few-shot setting, where a generalist TDM is fine-tuned with small amounts of data from the target environment, and in a zero-shot setting, where a generalist TDM is applied to an unseen environment without any further training. Here, we demonstrate that generalizing system dynamics can work much better than generalizing optimal behavior directly as a policy. Additional results show that TDMs also perform well"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2305.10912","kind":"arxiv","version":2},"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/2305.10912/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":"2305.10912","created_at":"2026-07-05T06:53:37.783081+00:00"},{"alias_kind":"arxiv_version","alias_value":"2305.10912v2","created_at":"2026-07-05T06:53:37.783081+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2305.10912","created_at":"2026-07-05T06:53:37.783081+00:00"},{"alias_kind":"pith_short_12","alias_value":"2BBY2EFX4U3R","created_at":"2026-07-05T06:53:37.783081+00:00"},{"alias_kind":"pith_short_16","alias_value":"2BBY2EFX4U3RO6KU","created_at":"2026-07-05T06:53:37.783081+00:00"},{"alias_kind":"pith_short_8","alias_value":"2BBY2EFX","created_at":"2026-07-05T06:53:37.783081+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.29874","citing_title":"Evolutionary Dynamics of Cooperation in Next-Generation LLM Agent Systems: A Cross-Provider Empirical Extension","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2502.11537","citing_title":"Simulus: Combining Improvements in Sample-Efficient World Model Agents","ref_index":47,"is_internal_anchor":false},{"citing_arxiv_id":"2603.14392","citing_title":"WestWorld: A Knowledge-Encoded Scalable Trajectory World Model for Diverse Robotic Systems","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2310.16828","citing_title":"TD-MPC2: Scalable, Robust World Models for Continuous Control","ref_index":163,"is_internal_anchor":false},{"citing_arxiv_id":"2604.09035","citing_title":"Advantage-Guided Diffusion for Model-Based Reinforcement Learning","ref_index":9,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2BBY2EFX4U3RO6KU2DZYNG53S7","json":"https://pith.science/pith/2BBY2EFX4U3RO6KU2DZYNG53S7.json","graph_json":"https://pith.science/api/pith-number/2BBY2EFX4U3RO6KU2DZYNG53S7/graph.json","events_json":"https://pith.science/api/pith-number/2BBY2EFX4U3RO6KU2DZYNG53S7/events.json","paper":"https://pith.science/paper/2BBY2EFX"},"agent_actions":{"view_html":"https://pith.science/pith/2BBY2EFX4U3RO6KU2DZYNG53S7","download_json":"https://pith.science/pith/2BBY2EFX4U3RO6KU2DZYNG53S7.json","view_paper":"https://pith.science/paper/2BBY2EFX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2305.10912&json=true","fetch_graph":"https://pith.science/api/pith-number/2BBY2EFX4U3RO6KU2DZYNG53S7/graph.json","fetch_events":"https://pith.science/api/pith-number/2BBY2EFX4U3RO6KU2DZYNG53S7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2BBY2EFX4U3RO6KU2DZYNG53S7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2BBY2EFX4U3RO6KU2DZYNG53S7/action/storage_attestation","attest_author":"https://pith.science/pith/2BBY2EFX4U3RO6KU2DZYNG53S7/action/author_attestation","sign_citation":"https://pith.science/pith/2BBY2EFX4U3RO6KU2DZYNG53S7/action/citation_signature","submit_replication":"https://pith.science/pith/2BBY2EFX4U3RO6KU2DZYNG53S7/action/replication_record"}},"created_at":"2026-07-05T06:53:37.783081+00:00","updated_at":"2026-07-05T06:53:37.783081+00:00"}