{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:RFWFVCVXGRYABGTSV7NET3ROQZ","short_pith_number":"pith:RFWFVCVX","schema_version":"1.0","canonical_sha256":"896c5a8ab73470009a72afda49ee2e864912a0e9dae4eea49b13b0f036770156","source":{"kind":"arxiv","id":"2305.10282","version":1},"attestation_state":"computed","paper":{"title":"Reward-agnostic Fine-tuning: Provable Statistical Benefits of Hybrid Reinforcement Learning","license":"http://creativecommons.org/publicdomain/zero/1.0/","headline":"","cross_cats":["cs.IT","math.IT","math.ST","stat.ML","stat.TH"],"primary_cat":"cs.LG","authors_text":"Gen Li, Jason D. Lee, Wenhao Zhan, Yuejie Chi, Yuxin Chen","submitted_at":"2023-05-17T15:17:23Z","abstract_excerpt":"This paper studies tabular reinforcement learning (RL) in the hybrid setting, which assumes access to both an offline dataset and online interactions with the unknown environment. A central question boils down to how to efficiently utilize online data collection to strengthen and complement the offline dataset and enable effective policy fine-tuning. Leveraging recent advances in reward-agnostic exploration and model-based offline RL, we design a three-stage hybrid RL algorithm that beats the best of both worlds -- pure offline RL and pure online RL -- in terms of sample complexities. The prop"},"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":"2305.10282","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/publicdomain/zero/1.0/","primary_cat":"cs.LG","submitted_at":"2023-05-17T15:17:23Z","cross_cats_sorted":["cs.IT","math.IT","math.ST","stat.ML","stat.TH"],"title_canon_sha256":"cbb8b9bd88621290992337c6996f38349c9ca87cff1dd0e75b4e7bf33de2760c","abstract_canon_sha256":"a0326ec4631aa9b760bf1618ecb40406998f024ad23942862d253099734a3f79"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:11:12.099447Z","signature_b64":"avBi84H6x2PmIDmuX0nnW/MBym2pMuyfNo6uvTu/W+EwYz3kqlp7YCJSdk9tEOYcQlOt0TP6TGv0tj8ztajEDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"896c5a8ab73470009a72afda49ee2e864912a0e9dae4eea49b13b0f036770156","last_reissued_at":"2026-07-05T06:11:12.099049Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:11:12.099049Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Reward-agnostic Fine-tuning: Provable Statistical Benefits of Hybrid Reinforcement Learning","license":"http://creativecommons.org/publicdomain/zero/1.0/","headline":"","cross_cats":["cs.IT","math.IT","math.ST","stat.ML","stat.TH"],"primary_cat":"cs.LG","authors_text":"Gen Li, Jason D. Lee, Wenhao Zhan, Yuejie Chi, Yuxin Chen","submitted_at":"2023-05-17T15:17:23Z","abstract_excerpt":"This paper studies tabular reinforcement learning (RL) in the hybrid setting, which assumes access to both an offline dataset and online interactions with the unknown environment. A central question boils down to how to efficiently utilize online data collection to strengthen and complement the offline dataset and enable effective policy fine-tuning. Leveraging recent advances in reward-agnostic exploration and model-based offline RL, we design a three-stage hybrid RL algorithm that beats the best of both worlds -- pure offline RL and pure online RL -- in terms of sample complexities. The prop"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2305.10282","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/2305.10282/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":"2305.10282","created_at":"2026-07-05T06:11:12.099106+00:00"},{"alias_kind":"arxiv_version","alias_value":"2305.10282v1","created_at":"2026-07-05T06:11:12.099106+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2305.10282","created_at":"2026-07-05T06:11:12.099106+00:00"},{"alias_kind":"pith_short_12","alias_value":"RFWFVCVXGRYA","created_at":"2026-07-05T06:11:12.099106+00:00"},{"alias_kind":"pith_short_16","alias_value":"RFWFVCVXGRYABGTS","created_at":"2026-07-05T06:11:12.099106+00:00"},{"alias_kind":"pith_short_8","alias_value":"RFWFVCVX","created_at":"2026-07-05T06:11:12.099106+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2508.08413","citing_title":"Decentralized Relaxed Smooth Optimization with Gradient Descent Methods","ref_index":2013,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RFWFVCVXGRYABGTSV7NET3ROQZ","json":"https://pith.science/pith/RFWFVCVXGRYABGTSV7NET3ROQZ.json","graph_json":"https://pith.science/api/pith-number/RFWFVCVXGRYABGTSV7NET3ROQZ/graph.json","events_json":"https://pith.science/api/pith-number/RFWFVCVXGRYABGTSV7NET3ROQZ/events.json","paper":"https://pith.science/paper/RFWFVCVX"},"agent_actions":{"view_html":"https://pith.science/pith/RFWFVCVXGRYABGTSV7NET3ROQZ","download_json":"https://pith.science/pith/RFWFVCVXGRYABGTSV7NET3ROQZ.json","view_paper":"https://pith.science/paper/RFWFVCVX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2305.10282&json=true","fetch_graph":"https://pith.science/api/pith-number/RFWFVCVXGRYABGTSV7NET3ROQZ/graph.json","fetch_events":"https://pith.science/api/pith-number/RFWFVCVXGRYABGTSV7NET3ROQZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RFWFVCVXGRYABGTSV7NET3ROQZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RFWFVCVXGRYABGTSV7NET3ROQZ/action/storage_attestation","attest_author":"https://pith.science/pith/RFWFVCVXGRYABGTSV7NET3ROQZ/action/author_attestation","sign_citation":"https://pith.science/pith/RFWFVCVXGRYABGTSV7NET3ROQZ/action/citation_signature","submit_replication":"https://pith.science/pith/RFWFVCVXGRYABGTSV7NET3ROQZ/action/replication_record"}},"created_at":"2026-07-05T06:11:12.099106+00:00","updated_at":"2026-07-05T06:11:12.099106+00:00"}