{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:MQ2C7ZZ4NDXSGND3VF5H7NDF45","short_pith_number":"pith:MQ2C7ZZ4","schema_version":"1.0","canonical_sha256":"64342fe73c68ef23347ba97a7fb465e7775deec22d953e2b850489a75cbe36de","source":{"kind":"arxiv","id":"2407.02466","version":3},"attestation_state":"computed","paper":{"title":"PWM: Policy Learning with Multi-Task World Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.RO"],"primary_cat":"cs.LG","authors_text":"Animesh Garg, Ignat Georgiev, Nicklas Hansen, Varun Giridhar","submitted_at":"2024-07-02T17:47:03Z","abstract_excerpt":"Reinforcement Learning (RL) has made significant strides in complex tasks but struggles in multi-task settings with different embodiments. World model methods offer scalability by learning a simulation of the environment but often rely on inefficient gradient-free optimization methods for policy extraction. In contrast, gradient-based methods exhibit lower variance but fail to handle discontinuities. Our work reveals that well-regularized world models can generate smoother optimization landscapes than the actual dynamics, facilitating more effective first-order optimization. We introduce Polic"},"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":"2407.02466","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-07-02T17:47:03Z","cross_cats_sorted":["cs.AI","cs.RO"],"title_canon_sha256":"84ce245183c15cac7c9b2aa9bbdcee60ab42ba0b508f0190028cb443ccdb79bf","abstract_canon_sha256":"9630fcad6779e3e7e1012ed97992e8052e1c036d0aa3067d5c6514b298766221"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:18:39.165705Z","signature_b64":"Wh8DEW1elOoRRVs/en+lNz5ayCHpaJcMNoSfLNaRnot9zn/PPymvbqXeVzMrMfZ8osgqkyLIWL3nPTJQ/iu8AQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"64342fe73c68ef23347ba97a7fb465e7775deec22d953e2b850489a75cbe36de","last_reissued_at":"2026-07-05T10:18:39.165149Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:18:39.165149Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"PWM: Policy Learning with Multi-Task World Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.RO"],"primary_cat":"cs.LG","authors_text":"Animesh Garg, Ignat Georgiev, Nicklas Hansen, Varun Giridhar","submitted_at":"2024-07-02T17:47:03Z","abstract_excerpt":"Reinforcement Learning (RL) has made significant strides in complex tasks but struggles in multi-task settings with different embodiments. World model methods offer scalability by learning a simulation of the environment but often rely on inefficient gradient-free optimization methods for policy extraction. In contrast, gradient-based methods exhibit lower variance but fail to handle discontinuities. Our work reveals that well-regularized world models can generate smoother optimization landscapes than the actual dynamics, facilitating more effective first-order optimization. We introduce Polic"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.02466","kind":"arxiv","version":3},"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/2407.02466/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":"2407.02466","created_at":"2026-07-05T10:18:39.165208+00:00"},{"alias_kind":"arxiv_version","alias_value":"2407.02466v3","created_at":"2026-07-05T10:18:39.165208+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.02466","created_at":"2026-07-05T10:18:39.165208+00:00"},{"alias_kind":"pith_short_12","alias_value":"MQ2C7ZZ4NDXS","created_at":"2026-07-05T10:18:39.165208+00:00"},{"alias_kind":"pith_short_16","alias_value":"MQ2C7ZZ4NDXSGND3","created_at":"2026-07-05T10:18:39.165208+00:00"},{"alias_kind":"pith_short_8","alias_value":"MQ2C7ZZ4","created_at":"2026-07-05T10:18:39.165208+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.22913","citing_title":"Intend, Reflect, Refine: An Adaptive Multimodal Reflection Framework for Autonomous Driving","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2604.20151","citing_title":"Toward Safe Autonomous Robotic Endovascular Interventions using World Models","ref_index":13,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MQ2C7ZZ4NDXSGND3VF5H7NDF45","json":"https://pith.science/pith/MQ2C7ZZ4NDXSGND3VF5H7NDF45.json","graph_json":"https://pith.science/api/pith-number/MQ2C7ZZ4NDXSGND3VF5H7NDF45/graph.json","events_json":"https://pith.science/api/pith-number/MQ2C7ZZ4NDXSGND3VF5H7NDF45/events.json","paper":"https://pith.science/paper/MQ2C7ZZ4"},"agent_actions":{"view_html":"https://pith.science/pith/MQ2C7ZZ4NDXSGND3VF5H7NDF45","download_json":"https://pith.science/pith/MQ2C7ZZ4NDXSGND3VF5H7NDF45.json","view_paper":"https://pith.science/paper/MQ2C7ZZ4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2407.02466&json=true","fetch_graph":"https://pith.science/api/pith-number/MQ2C7ZZ4NDXSGND3VF5H7NDF45/graph.json","fetch_events":"https://pith.science/api/pith-number/MQ2C7ZZ4NDXSGND3VF5H7NDF45/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MQ2C7ZZ4NDXSGND3VF5H7NDF45/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MQ2C7ZZ4NDXSGND3VF5H7NDF45/action/storage_attestation","attest_author":"https://pith.science/pith/MQ2C7ZZ4NDXSGND3VF5H7NDF45/action/author_attestation","sign_citation":"https://pith.science/pith/MQ2C7ZZ4NDXSGND3VF5H7NDF45/action/citation_signature","submit_replication":"https://pith.science/pith/MQ2C7ZZ4NDXSGND3VF5H7NDF45/action/replication_record"}},"created_at":"2026-07-05T10:18:39.165208+00:00","updated_at":"2026-07-05T10:18:39.165208+00:00"}