{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:7QRFGGDJEVP5QFSGJ5S7GF335K","short_pith_number":"pith:7QRFGGDJ","schema_version":"1.0","canonical_sha256":"fc22531869255fd816464f65f3177beaad7c9ed2df9ab5edb3b1e3437e0ecd21","source":{"kind":"arxiv","id":"2112.10751","version":2},"attestation_state":"computed","paper":{"title":"RvS: What is Essential for Offline RL via Supervised Learning?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Benjamin Eysenbach, Ilya Kostrikov, Scott Emmons, Sergey Levine","submitted_at":"2021-12-20T18:55:16Z","abstract_excerpt":"Recent work has shown that supervised learning alone, without temporal difference (TD) learning, can be remarkably effective for offline RL. When does this hold true, and which algorithmic components are necessary? Through extensive experiments, we boil supervised learning for offline RL down to its essential elements. In every environment suite we consider, simply maximizing likelihood with a two-layer feedforward MLP is competitive with state-of-the-art results of substantially more complex methods based on TD learning or sequence modeling with Transformers. Carefully choosing model capacity"},"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":"2112.10751","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-12-20T18:55:16Z","cross_cats_sorted":["cs.AI","stat.ML"],"title_canon_sha256":"70cc29ca3780cc34035a2665a310244768a319658feb75f8543d8a389fe07587","abstract_canon_sha256":"7eae074ab6a903443324aff9c5dfa7a47bad8c85a56760f6e51c182508965be4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:22:18.989277Z","signature_b64":"/kF1afTuS3OIidddj+oUzKkKx3gFxybjDj+oINQk0LDnEuCZZ+AIITH6V0HndOWJs1Pg0AqJ6eFGptZ3NH9tAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fc22531869255fd816464f65f3177beaad7c9ed2df9ab5edb3b1e3437e0ecd21","last_reissued_at":"2026-07-05T04:22:18.988792Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:22:18.988792Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"RvS: What is Essential for Offline RL via Supervised Learning?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Benjamin Eysenbach, Ilya Kostrikov, Scott Emmons, Sergey Levine","submitted_at":"2021-12-20T18:55:16Z","abstract_excerpt":"Recent work has shown that supervised learning alone, without temporal difference (TD) learning, can be remarkably effective for offline RL. When does this hold true, and which algorithmic components are necessary? Through extensive experiments, we boil supervised learning for offline RL down to its essential elements. In every environment suite we consider, simply maximizing likelihood with a two-layer feedforward MLP is competitive with state-of-the-art results of substantially more complex methods based on TD learning or sequence modeling with Transformers. Carefully choosing model capacity"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2112.10751","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/2112.10751/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":"2112.10751","created_at":"2026-07-05T04:22:18.988855+00:00"},{"alias_kind":"arxiv_version","alias_value":"2112.10751v2","created_at":"2026-07-05T04:22:18.988855+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2112.10751","created_at":"2026-07-05T04:22:18.988855+00:00"},{"alias_kind":"pith_short_12","alias_value":"7QRFGGDJEVP5","created_at":"2026-07-05T04:22:18.988855+00:00"},{"alias_kind":"pith_short_16","alias_value":"7QRFGGDJEVP5QFSG","created_at":"2026-07-05T04:22:18.988855+00:00"},{"alias_kind":"pith_short_8","alias_value":"7QRFGGDJ","created_at":"2026-07-05T04:22:18.988855+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.08312","citing_title":"Neuro-Symbolic Injection of LTLf Constraints in Autoregressive Reinforcement Learning Policies","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2606.07366","citing_title":"Dash2Sim: Closed-Loop Driving Simulation from in-the-wild Dashcam Videos","ref_index":67,"is_internal_anchor":false},{"citing_arxiv_id":"2606.32027","citing_title":"Freeform Preference Learning for Robotic Manipulation","ref_index":30,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01862","citing_title":"QHyer: Q-conditioned Hybrid Attention-mamba Transformer for Offline Goal-conditioned RL","ref_index":47,"is_internal_anchor":false},{"citing_arxiv_id":"2604.04528","citing_title":"Receding-Horizon Control via Drifting Models","ref_index":18,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/7QRFGGDJEVP5QFSGJ5S7GF335K","json":"https://pith.science/pith/7QRFGGDJEVP5QFSGJ5S7GF335K.json","graph_json":"https://pith.science/api/pith-number/7QRFGGDJEVP5QFSGJ5S7GF335K/graph.json","events_json":"https://pith.science/api/pith-number/7QRFGGDJEVP5QFSGJ5S7GF335K/events.json","paper":"https://pith.science/paper/7QRFGGDJ"},"agent_actions":{"view_html":"https://pith.science/pith/7QRFGGDJEVP5QFSGJ5S7GF335K","download_json":"https://pith.science/pith/7QRFGGDJEVP5QFSGJ5S7GF335K.json","view_paper":"https://pith.science/paper/7QRFGGDJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2112.10751&json=true","fetch_graph":"https://pith.science/api/pith-number/7QRFGGDJEVP5QFSGJ5S7GF335K/graph.json","fetch_events":"https://pith.science/api/pith-number/7QRFGGDJEVP5QFSGJ5S7GF335K/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7QRFGGDJEVP5QFSGJ5S7GF335K/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7QRFGGDJEVP5QFSGJ5S7GF335K/action/storage_attestation","attest_author":"https://pith.science/pith/7QRFGGDJEVP5QFSGJ5S7GF335K/action/author_attestation","sign_citation":"https://pith.science/pith/7QRFGGDJEVP5QFSGJ5S7GF335K/action/citation_signature","submit_replication":"https://pith.science/pith/7QRFGGDJEVP5QFSGJ5S7GF335K/action/replication_record"}},"created_at":"2026-07-05T04:22:18.988855+00:00","updated_at":"2026-07-05T04:22:18.988855+00:00"}