{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:UAPDLFDL4QCR4EXNCYC3CF4NUL","short_pith_number":"pith:UAPDLFDL","schema_version":"1.0","canonical_sha256":"a01e35946be4051e12ed1605b1178da2ee3cee02db8073a1a12212493406f621","source":{"kind":"arxiv","id":"2411.04625","version":2},"attestation_state":"computed","paper":{"title":"Sharp Analysis for KL-Regularized Contextual Bandits and RLHF","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Chenlu Ye, Heyang Zhao, Quanquan Gu, Tong Zhang","submitted_at":"2024-11-07T11:22:46Z","abstract_excerpt":"Reverse-Kullback-Leibler (KL) regularization has emerged to be a predominant technique used to enhance policy optimization in reinforcement learning (RL) and reinforcement learning from human feedback (RLHF), which forces the learned policy to stay close to a reference policy. While the effectiveness and necessity of KL-regularization have been empirically demonstrated in various practical scenarios, current theoretical analysis of KL-regularized RLHF still obtains the same $\\mathcal{O}(1 / \\epsilon^2)$ sample complexity as problems without KL-regularization. To understand the fundamental dist"},"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":"2411.04625","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-11-07T11:22:46Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"84d09a0bdbd8484d9c1c6fe800673a9dc9992a2b9684449c41cd6ee8eb5805ed","abstract_canon_sha256":"56b79cdfa0be4add40f3fd7866573db00aa1c8962c7c1d132252550327ad1770"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:12:23.228267Z","signature_b64":"XCwkm158Oll7Avuudmk77DY+C+F9o/pH6XhO+Lf6Hj+j8V5i/rcnphDexU0DUX0aLoQoUlZ/Mtv5qF+b88ITAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a01e35946be4051e12ed1605b1178da2ee3cee02db8073a1a12212493406f621","last_reissued_at":"2026-07-05T10:12:23.227785Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:12:23.227785Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Sharp Analysis for KL-Regularized Contextual Bandits and RLHF","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Chenlu Ye, Heyang Zhao, Quanquan Gu, Tong Zhang","submitted_at":"2024-11-07T11:22:46Z","abstract_excerpt":"Reverse-Kullback-Leibler (KL) regularization has emerged to be a predominant technique used to enhance policy optimization in reinforcement learning (RL) and reinforcement learning from human feedback (RLHF), which forces the learned policy to stay close to a reference policy. While the effectiveness and necessity of KL-regularization have been empirically demonstrated in various practical scenarios, current theoretical analysis of KL-regularized RLHF still obtains the same $\\mathcal{O}(1 / \\epsilon^2)$ sample complexity as problems without KL-regularization. To understand the fundamental dist"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2411.04625","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/2411.04625/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":"2411.04625","created_at":"2026-07-05T10:12:23.227846+00:00"},{"alias_kind":"arxiv_version","alias_value":"2411.04625v2","created_at":"2026-07-05T10:12:23.227846+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2411.04625","created_at":"2026-07-05T10:12:23.227846+00:00"},{"alias_kind":"pith_short_12","alias_value":"UAPDLFDL4QCR","created_at":"2026-07-05T10:12:23.227846+00:00"},{"alias_kind":"pith_short_16","alias_value":"UAPDLFDL4QCR4EXN","created_at":"2026-07-05T10:12:23.227846+00:00"},{"alias_kind":"pith_short_8","alias_value":"UAPDLFDL","created_at":"2026-07-05T10:12:23.227846+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.06053","citing_title":"Online KL-Regularized Reinforcement Learning with Function Approximation under Misspecification","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2606.27917","citing_title":"Graph Dimensionality Reduction for Contextual Bandits: Structure-Specific Regret Bounds under Approximate Smoothness and Noisy Eigenspaces","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06977","citing_title":"$f$-Divergence Regularized RLHF: Two Tales of Sampling and Unified Analyses","ref_index":17,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/UAPDLFDL4QCR4EXNCYC3CF4NUL","json":"https://pith.science/pith/UAPDLFDL4QCR4EXNCYC3CF4NUL.json","graph_json":"https://pith.science/api/pith-number/UAPDLFDL4QCR4EXNCYC3CF4NUL/graph.json","events_json":"https://pith.science/api/pith-number/UAPDLFDL4QCR4EXNCYC3CF4NUL/events.json","paper":"https://pith.science/paper/UAPDLFDL"},"agent_actions":{"view_html":"https://pith.science/pith/UAPDLFDL4QCR4EXNCYC3CF4NUL","download_json":"https://pith.science/pith/UAPDLFDL4QCR4EXNCYC3CF4NUL.json","view_paper":"https://pith.science/paper/UAPDLFDL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2411.04625&json=true","fetch_graph":"https://pith.science/api/pith-number/UAPDLFDL4QCR4EXNCYC3CF4NUL/graph.json","fetch_events":"https://pith.science/api/pith-number/UAPDLFDL4QCR4EXNCYC3CF4NUL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UAPDLFDL4QCR4EXNCYC3CF4NUL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UAPDLFDL4QCR4EXNCYC3CF4NUL/action/storage_attestation","attest_author":"https://pith.science/pith/UAPDLFDL4QCR4EXNCYC3CF4NUL/action/author_attestation","sign_citation":"https://pith.science/pith/UAPDLFDL4QCR4EXNCYC3CF4NUL/action/citation_signature","submit_replication":"https://pith.science/pith/UAPDLFDL4QCR4EXNCYC3CF4NUL/action/replication_record"}},"created_at":"2026-07-05T10:12:23.227846+00:00","updated_at":"2026-07-05T10:12:23.227846+00:00"}