{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:4E2CGBVKLCZ7WUJCHBHEP3L5HT","short_pith_number":"pith:4E2CGBVK","schema_version":"1.0","canonical_sha256":"e1342306aa58b3fb5122384e47ed7d3cfb632305a5fcafb0ef0ef6c9f81278e2","source":{"kind":"arxiv","id":"2102.08363","version":2},"attestation_state":"computed","paper":{"title":"COMBO: Conservative Offline Model-Based Policy Optimization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.RO"],"primary_cat":"cs.LG","authors_text":"Aravind Rajeswaran, Aviral Kumar, Chelsea Finn, Rafael Rafailov, Sergey Levine, Tianhe Yu","submitted_at":"2021-02-16T18:50:32Z","abstract_excerpt":"Model-based algorithms, which learn a dynamics model from logged experience and perform some sort of pessimistic planning under the learned model, have emerged as a promising paradigm for offline reinforcement learning (offline RL). However, practical variants of such model-based algorithms rely on explicit uncertainty quantification for incorporating pessimism. Uncertainty estimation with complex models, such as deep neural networks, can be difficult and unreliable. We overcome this limitation by developing a new model-based offline RL algorithm, COMBO, that regularizes the value function on "},"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":"2102.08363","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2021-02-16T18:50:32Z","cross_cats_sorted":["cs.AI","cs.RO"],"title_canon_sha256":"1fa2dfc1df486123f932840268a3b3e65db3093ba84e590d9b811848c709a6f2","abstract_canon_sha256":"53c5997f3bb61481bd551984332b64a789fd7c1dc3310a682a2d62d0fb37b427"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:51:54.559880Z","signature_b64":"hEApTY8u6wbrI9o7PSgb0GGBZ74W50wFKPwZr7PI0JwpagsxLyFgtwQnjkljACaqEr1IGRzZSak9musjRXg9Dg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e1342306aa58b3fb5122384e47ed7d3cfb632305a5fcafb0ef0ef6c9f81278e2","last_reissued_at":"2026-07-05T03:51:54.559393Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:51:54.559393Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"COMBO: Conservative Offline Model-Based Policy Optimization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.RO"],"primary_cat":"cs.LG","authors_text":"Aravind Rajeswaran, Aviral Kumar, Chelsea Finn, Rafael Rafailov, Sergey Levine, Tianhe Yu","submitted_at":"2021-02-16T18:50:32Z","abstract_excerpt":"Model-based algorithms, which learn a dynamics model from logged experience and perform some sort of pessimistic planning under the learned model, have emerged as a promising paradigm for offline reinforcement learning (offline RL). However, practical variants of such model-based algorithms rely on explicit uncertainty quantification for incorporating pessimism. Uncertainty estimation with complex models, such as deep neural networks, can be difficult and unreliable. We overcome this limitation by developing a new model-based offline RL algorithm, COMBO, that regularizes the value function on "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2102.08363","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/2102.08363/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":"2102.08363","created_at":"2026-07-05T03:51:54.559453+00:00"},{"alias_kind":"arxiv_version","alias_value":"2102.08363v2","created_at":"2026-07-05T03:51:54.559453+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2102.08363","created_at":"2026-07-05T03:51:54.559453+00:00"},{"alias_kind":"pith_short_12","alias_value":"4E2CGBVKLCZ7","created_at":"2026-07-05T03:51:54.559453+00:00"},{"alias_kind":"pith_short_16","alias_value":"4E2CGBVKLCZ7WUJC","created_at":"2026-07-05T03:51:54.559453+00:00"},{"alias_kind":"pith_short_8","alias_value":"4E2CGBVK","created_at":"2026-07-05T03:51:54.559453+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2310.10639","citing_title":"Zero-Shot Robotic Manipulation with Pretrained Image-Editing Diffusion Models","ref_index":63,"is_internal_anchor":false},{"citing_arxiv_id":"2604.01346","citing_title":"Safety, Security, and Cognitive Risks in World Models","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2108.03298","citing_title":"What Matters in Learning from Offline Human Demonstrations for Robot Manipulation","ref_index":43,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4E2CGBVKLCZ7WUJCHBHEP3L5HT","json":"https://pith.science/pith/4E2CGBVKLCZ7WUJCHBHEP3L5HT.json","graph_json":"https://pith.science/api/pith-number/4E2CGBVKLCZ7WUJCHBHEP3L5HT/graph.json","events_json":"https://pith.science/api/pith-number/4E2CGBVKLCZ7WUJCHBHEP3L5HT/events.json","paper":"https://pith.science/paper/4E2CGBVK"},"agent_actions":{"view_html":"https://pith.science/pith/4E2CGBVKLCZ7WUJCHBHEP3L5HT","download_json":"https://pith.science/pith/4E2CGBVKLCZ7WUJCHBHEP3L5HT.json","view_paper":"https://pith.science/paper/4E2CGBVK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2102.08363&json=true","fetch_graph":"https://pith.science/api/pith-number/4E2CGBVKLCZ7WUJCHBHEP3L5HT/graph.json","fetch_events":"https://pith.science/api/pith-number/4E2CGBVKLCZ7WUJCHBHEP3L5HT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4E2CGBVKLCZ7WUJCHBHEP3L5HT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4E2CGBVKLCZ7WUJCHBHEP3L5HT/action/storage_attestation","attest_author":"https://pith.science/pith/4E2CGBVKLCZ7WUJCHBHEP3L5HT/action/author_attestation","sign_citation":"https://pith.science/pith/4E2CGBVKLCZ7WUJCHBHEP3L5HT/action/citation_signature","submit_replication":"https://pith.science/pith/4E2CGBVKLCZ7WUJCHBHEP3L5HT/action/replication_record"}},"created_at":"2026-07-05T03:51:54.559453+00:00","updated_at":"2026-07-05T03:51:54.559453+00:00"}