{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:BP2AHTRCDHRU6J6APUCAXDS2C3","short_pith_number":"pith:BP2AHTRC","schema_version":"1.0","canonical_sha256":"0bf403ce2219e34f27c07d040b8e5a16e3c6a47d599e55221b34fbdeadd90b3f","source":{"kind":"arxiv","id":"2301.06276","version":1},"attestation_state":"computed","paper":{"title":"The Role of Baselines in Policy Gradient Optimization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Bo Dai, Csaba Szepesvari, Dale Schuurmans, Jincheng Mei, Valentin Thomas, Wesley Chung","submitted_at":"2023-01-16T06:28:00Z","abstract_excerpt":"We study the effect of baselines in on-policy stochastic policy gradient optimization, and close the gap between the theory and practice of policy optimization methods. Our first contribution is to show that the \\emph{state value} baseline allows on-policy stochastic \\emph{natural} policy gradient (NPG) to converge to a globally optimal policy at an $O(1/t)$ rate, which was not previously known. The analysis relies on two novel findings: the expected progress of the NPG update satisfies a stochastic version of the non-uniform \\L{}ojasiewicz (N\\L{}) inequality, and with probability 1 the state "},"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":"2301.06276","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-01-16T06:28:00Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"ae48f9e0f137b92136dd59bfb10e7cb188652d79e4892fa7c409557c24893b9f","abstract_canon_sha256":"ba7ccee56f6525951b5e68dcd83b31f64fdd9a20e10bdce473907e97fad0cfc3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:33:29.888056Z","signature_b64":"Zpv+ZR/KKzJ3omY56GDK1uCNOMHV/7N5K6wtrzAVy8mEJnKdbQ+4JRE3UQLWGG+1kfyI/1fpAoC9+lpKCVOLBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0bf403ce2219e34f27c07d040b8e5a16e3c6a47d599e55221b34fbdeadd90b3f","last_reissued_at":"2026-07-05T05:33:29.887558Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:33:29.887558Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The Role of Baselines in Policy Gradient Optimization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Bo Dai, Csaba Szepesvari, Dale Schuurmans, Jincheng Mei, Valentin Thomas, Wesley Chung","submitted_at":"2023-01-16T06:28:00Z","abstract_excerpt":"We study the effect of baselines in on-policy stochastic policy gradient optimization, and close the gap between the theory and practice of policy optimization methods. Our first contribution is to show that the \\emph{state value} baseline allows on-policy stochastic \\emph{natural} policy gradient (NPG) to converge to a globally optimal policy at an $O(1/t)$ rate, which was not previously known. The analysis relies on two novel findings: the expected progress of the NPG update satisfies a stochastic version of the non-uniform \\L{}ojasiewicz (N\\L{}) inequality, and with probability 1 the state "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2301.06276","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/2301.06276/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":"2301.06276","created_at":"2026-07-05T05:33:29.887621+00:00"},{"alias_kind":"arxiv_version","alias_value":"2301.06276v1","created_at":"2026-07-05T05:33:29.887621+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2301.06276","created_at":"2026-07-05T05:33:29.887621+00:00"},{"alias_kind":"pith_short_12","alias_value":"BP2AHTRCDHRU","created_at":"2026-07-05T05:33:29.887621+00:00"},{"alias_kind":"pith_short_16","alias_value":"BP2AHTRCDHRU6J6A","created_at":"2026-07-05T05:33:29.887621+00:00"},{"alias_kind":"pith_short_8","alias_value":"BP2AHTRC","created_at":"2026-07-05T05:33:29.887621+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2309.16797","citing_title":"Promptbreeder: Self-Referential Self-Improvement Via Prompt Evolution","ref_index":224,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/BP2AHTRCDHRU6J6APUCAXDS2C3","json":"https://pith.science/pith/BP2AHTRCDHRU6J6APUCAXDS2C3.json","graph_json":"https://pith.science/api/pith-number/BP2AHTRCDHRU6J6APUCAXDS2C3/graph.json","events_json":"https://pith.science/api/pith-number/BP2AHTRCDHRU6J6APUCAXDS2C3/events.json","paper":"https://pith.science/paper/BP2AHTRC"},"agent_actions":{"view_html":"https://pith.science/pith/BP2AHTRCDHRU6J6APUCAXDS2C3","download_json":"https://pith.science/pith/BP2AHTRCDHRU6J6APUCAXDS2C3.json","view_paper":"https://pith.science/paper/BP2AHTRC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2301.06276&json=true","fetch_graph":"https://pith.science/api/pith-number/BP2AHTRCDHRU6J6APUCAXDS2C3/graph.json","fetch_events":"https://pith.science/api/pith-number/BP2AHTRCDHRU6J6APUCAXDS2C3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BP2AHTRCDHRU6J6APUCAXDS2C3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BP2AHTRCDHRU6J6APUCAXDS2C3/action/storage_attestation","attest_author":"https://pith.science/pith/BP2AHTRCDHRU6J6APUCAXDS2C3/action/author_attestation","sign_citation":"https://pith.science/pith/BP2AHTRCDHRU6J6APUCAXDS2C3/action/citation_signature","submit_replication":"https://pith.science/pith/BP2AHTRCDHRU6J6APUCAXDS2C3/action/replication_record"}},"created_at":"2026-07-05T05:33:29.887621+00:00","updated_at":"2026-07-05T05:33:29.887621+00:00"}