{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:FJXTDZI5NAZ3NQGPI3KVVW2PPC","short_pith_number":"pith:FJXTDZI5","schema_version":"1.0","canonical_sha256":"2a6f31e51d6833b6c0cf46d55adb4f78b6e3aab5c5217ebd20e0d135179ea8b1","source":{"kind":"arxiv","id":"2412.10616","version":1},"attestation_state":"computed","paper":{"title":"Hybrid Preference Optimization for Alignment: Provably Faster Convergence Rates by Combining Offline Preferences with Online Exploration","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Aadirupa Saha, Avinandan Bose, Maryam Fazel, Simon Shaolei Du, Zhihan Xiong","submitted_at":"2024-12-13T23:42:24Z","abstract_excerpt":"Reinforcement Learning from Human Feedback (RLHF) is currently the leading approach for aligning large language models with human preferences. Typically, these models rely on extensive offline preference datasets for training. However, offline algorithms impose strict concentrability requirements, which are often difficult to satisfy. On the other hand, while online algorithms can avoid the concentrability issue, pure online exploration could be expensive due to the active preference query cost and real-time implementation overhead. In this paper, we propose a novel approach: Hybrid Preference"},"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":"2412.10616","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-12-13T23:42:24Z","cross_cats_sorted":[],"title_canon_sha256":"19f60ed04faa3ac59d2292dc8b9f077b0cf2d5cecfea74b78f701e96f0f90815","abstract_canon_sha256":"42e3b442ff58f6aac194c8d0527b7cab28c5c1fdf97b8f92a08673101b3ac470"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:49:18.390627Z","signature_b64":"SOwlbeTXwGZT3EXuVCkd5uJoW/dZFguMFqn83wmEjgrYEU8pUiJqZomXpRsQOhc3GqaD2MIczPC7jHvpA5KtBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2a6f31e51d6833b6c0cf46d55adb4f78b6e3aab5c5217ebd20e0d135179ea8b1","last_reissued_at":"2026-07-05T09:49:18.389702Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:49:18.389702Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Hybrid Preference Optimization for Alignment: Provably Faster Convergence Rates by Combining Offline Preferences with Online Exploration","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Aadirupa Saha, Avinandan Bose, Maryam Fazel, Simon Shaolei Du, Zhihan Xiong","submitted_at":"2024-12-13T23:42:24Z","abstract_excerpt":"Reinforcement Learning from Human Feedback (RLHF) is currently the leading approach for aligning large language models with human preferences. Typically, these models rely on extensive offline preference datasets for training. However, offline algorithms impose strict concentrability requirements, which are often difficult to satisfy. On the other hand, while online algorithms can avoid the concentrability issue, pure online exploration could be expensive due to the active preference query cost and real-time implementation overhead. In this paper, we propose a novel approach: Hybrid Preference"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.10616","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/2412.10616/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":"2412.10616","created_at":"2026-07-05T09:49:18.389758+00:00"},{"alias_kind":"arxiv_version","alias_value":"2412.10616v1","created_at":"2026-07-05T09:49:18.389758+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.10616","created_at":"2026-07-05T09:49:18.389758+00:00"},{"alias_kind":"pith_short_12","alias_value":"FJXTDZI5NAZ3","created_at":"2026-07-05T09:49:18.389758+00:00"},{"alias_kind":"pith_short_16","alias_value":"FJXTDZI5NAZ3NQGP","created_at":"2026-07-05T09:49:18.389758+00:00"},{"alias_kind":"pith_short_8","alias_value":"FJXTDZI5","created_at":"2026-07-05T09:49:18.389758+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FJXTDZI5NAZ3NQGPI3KVVW2PPC","json":"https://pith.science/pith/FJXTDZI5NAZ3NQGPI3KVVW2PPC.json","graph_json":"https://pith.science/api/pith-number/FJXTDZI5NAZ3NQGPI3KVVW2PPC/graph.json","events_json":"https://pith.science/api/pith-number/FJXTDZI5NAZ3NQGPI3KVVW2PPC/events.json","paper":"https://pith.science/paper/FJXTDZI5"},"agent_actions":{"view_html":"https://pith.science/pith/FJXTDZI5NAZ3NQGPI3KVVW2PPC","download_json":"https://pith.science/pith/FJXTDZI5NAZ3NQGPI3KVVW2PPC.json","view_paper":"https://pith.science/paper/FJXTDZI5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2412.10616&json=true","fetch_graph":"https://pith.science/api/pith-number/FJXTDZI5NAZ3NQGPI3KVVW2PPC/graph.json","fetch_events":"https://pith.science/api/pith-number/FJXTDZI5NAZ3NQGPI3KVVW2PPC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FJXTDZI5NAZ3NQGPI3KVVW2PPC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FJXTDZI5NAZ3NQGPI3KVVW2PPC/action/storage_attestation","attest_author":"https://pith.science/pith/FJXTDZI5NAZ3NQGPI3KVVW2PPC/action/author_attestation","sign_citation":"https://pith.science/pith/FJXTDZI5NAZ3NQGPI3KVVW2PPC/action/citation_signature","submit_replication":"https://pith.science/pith/FJXTDZI5NAZ3NQGPI3KVVW2PPC/action/replication_record"}},"created_at":"2026-07-05T09:49:18.389758+00:00","updated_at":"2026-07-05T09:49:18.389758+00:00"}