{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:AXLJAR7OEMHBQHN7NSKW4A3NNG","short_pith_number":"pith:AXLJAR7O","schema_version":"1.0","canonical_sha256":"05d69047ee230e181dbf6c956e036d69893c9ea1c73ee6f69f9d21b2ebff9e32","source":{"kind":"arxiv","id":"2403.15146","version":1},"attestation_state":"computed","paper":{"title":"On the Convergence of Adam under Non-uniform Smoothness: Separability from SGDM and Beyond","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["math.OC"],"primary_cat":"cs.LG","authors_text":"Bohan Wang, Huishuai Zhang, Qi Meng, Ruoyu Sun, Wei Chen, Zhi-Ming Ma","submitted_at":"2024-03-22T11:57:51Z","abstract_excerpt":"This paper aims to clearly distinguish between Stochastic Gradient Descent with Momentum (SGDM) and Adam in terms of their convergence rates. We demonstrate that Adam achieves a faster convergence compared to SGDM under the condition of non-uniformly bounded smoothness. Our findings reveal that: (1) in deterministic environments, Adam can attain the known lower bound for the convergence rate of deterministic first-order optimizers, whereas the convergence rate of Gradient Descent with Momentum (GDM) has higher order dependence on the initial function value; (2) in stochastic setting, Adam's co"},"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":"2403.15146","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-03-22T11:57:51Z","cross_cats_sorted":["math.OC"],"title_canon_sha256":"bfae599b7529bcd8700aeaa8e08836c8ef2ad193246a57c9536c191b30946563","abstract_canon_sha256":"f02197f0f3d48808a91b37daadc39bdfe7cd163806cb8c052f5f183dddbe883b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:59:30.945882Z","signature_b64":"ShQy5I7nFVXtYW+FTr3wOwtiHLofduYDx2vNY0zRcZMIvDeHyabd1wEb2r9+Mt4SZLW4QaHtCbe57ZbpxA5aAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"05d69047ee230e181dbf6c956e036d69893c9ea1c73ee6f69f9d21b2ebff9e32","last_reissued_at":"2026-07-05T07:59:30.945433Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:59:30.945433Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"On the Convergence of Adam under Non-uniform Smoothness: Separability from SGDM and Beyond","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["math.OC"],"primary_cat":"cs.LG","authors_text":"Bohan Wang, Huishuai Zhang, Qi Meng, Ruoyu Sun, Wei Chen, Zhi-Ming Ma","submitted_at":"2024-03-22T11:57:51Z","abstract_excerpt":"This paper aims to clearly distinguish between Stochastic Gradient Descent with Momentum (SGDM) and Adam in terms of their convergence rates. We demonstrate that Adam achieves a faster convergence compared to SGDM under the condition of non-uniformly bounded smoothness. Our findings reveal that: (1) in deterministic environments, Adam can attain the known lower bound for the convergence rate of deterministic first-order optimizers, whereas the convergence rate of Gradient Descent with Momentum (GDM) has higher order dependence on the initial function value; (2) in stochastic setting, Adam's co"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2403.15146","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/2403.15146/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":"2403.15146","created_at":"2026-07-05T07:59:30.945508+00:00"},{"alias_kind":"arxiv_version","alias_value":"2403.15146v1","created_at":"2026-07-05T07:59:30.945508+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2403.15146","created_at":"2026-07-05T07:59:30.945508+00:00"},{"alias_kind":"pith_short_12","alias_value":"AXLJAR7OEMHB","created_at":"2026-07-05T07:59:30.945508+00:00"},{"alias_kind":"pith_short_16","alias_value":"AXLJAR7OEMHBQHN7","created_at":"2026-07-05T07:59:30.945508+00:00"},{"alias_kind":"pith_short_8","alias_value":"AXLJAR7O","created_at":"2026-07-05T07:59:30.945508+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2603.03099","citing_title":"Why Adam Can Beat SGD: Second-Moment Normalization Yields Sharper Tails","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2603.03099","citing_title":"Why Adam Can Beat SGD: Second-Moment Normalization Yields Sharper Tails","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14345","citing_title":"Convergence of difference inclusions via a diameter criterion","ref_index":117,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/AXLJAR7OEMHBQHN7NSKW4A3NNG","json":"https://pith.science/pith/AXLJAR7OEMHBQHN7NSKW4A3NNG.json","graph_json":"https://pith.science/api/pith-number/AXLJAR7OEMHBQHN7NSKW4A3NNG/graph.json","events_json":"https://pith.science/api/pith-number/AXLJAR7OEMHBQHN7NSKW4A3NNG/events.json","paper":"https://pith.science/paper/AXLJAR7O"},"agent_actions":{"view_html":"https://pith.science/pith/AXLJAR7OEMHBQHN7NSKW4A3NNG","download_json":"https://pith.science/pith/AXLJAR7OEMHBQHN7NSKW4A3NNG.json","view_paper":"https://pith.science/paper/AXLJAR7O","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2403.15146&json=true","fetch_graph":"https://pith.science/api/pith-number/AXLJAR7OEMHBQHN7NSKW4A3NNG/graph.json","fetch_events":"https://pith.science/api/pith-number/AXLJAR7OEMHBQHN7NSKW4A3NNG/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AXLJAR7OEMHBQHN7NSKW4A3NNG/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AXLJAR7OEMHBQHN7NSKW4A3NNG/action/storage_attestation","attest_author":"https://pith.science/pith/AXLJAR7OEMHBQHN7NSKW4A3NNG/action/author_attestation","sign_citation":"https://pith.science/pith/AXLJAR7OEMHBQHN7NSKW4A3NNG/action/citation_signature","submit_replication":"https://pith.science/pith/AXLJAR7OEMHBQHN7NSKW4A3NNG/action/replication_record"}},"created_at":"2026-07-05T07:59:30.945508+00:00","updated_at":"2026-07-05T07:59:30.945508+00:00"}