{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:YDATY2SOFXLOMSRWUDIEQ255IL","short_pith_number":"pith:YDATY2SO","schema_version":"1.0","canonical_sha256":"c0c13c6a4e2dd6e64a36a0d0486bbd42da1477affc2216b03a36f25cb2dc46ba","source":{"kind":"arxiv","id":"2305.15287","version":2},"attestation_state":"computed","paper":{"title":"The Crucial Role of Normalization in Sharpness-Aware Minimization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Kwangjun Ahn, Suvrit Sra, Yan Dai","submitted_at":"2023-05-24T16:09:41Z","abstract_excerpt":"Sharpness-Aware Minimization (SAM) is a recently proposed gradient-based optimizer (Foret et al., ICLR 2021) that greatly improves the prediction performance of deep neural networks. Consequently, there has been a surge of interest in explaining its empirical success. We focus, in particular, on understanding the role played by normalization, a key component of the SAM updates. We theoretically and empirically study the effect of normalization in SAM for both convex and non-convex functions, revealing two key roles played by normalization: i) it helps in stabilizing the algorithm; and ii) it e"},"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":"2305.15287","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-05-24T16:09:41Z","cross_cats_sorted":["cs.AI","stat.ML"],"title_canon_sha256":"965f3234889712145ef55f4011a173f981be2889d64084ba74fd39272cad6807","abstract_canon_sha256":"b247b1ff6e6fbe89b69fe270710114aab26893190bf47d5a502cab138b254554"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:03:59.362329Z","signature_b64":"TzzTKm8QvCCQjx2CzNMwiXZOokE3NuK9aOlZsrVv+Or/jY6Hv9/YqJA093jhDl5kWeeKnhnQOOw86WvKyDUwCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c0c13c6a4e2dd6e64a36a0d0486bbd42da1477affc2216b03a36f25cb2dc46ba","last_reissued_at":"2026-07-05T07:03:59.361853Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:03:59.361853Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The Crucial Role of Normalization in Sharpness-Aware Minimization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Kwangjun Ahn, Suvrit Sra, Yan Dai","submitted_at":"2023-05-24T16:09:41Z","abstract_excerpt":"Sharpness-Aware Minimization (SAM) is a recently proposed gradient-based optimizer (Foret et al., ICLR 2021) that greatly improves the prediction performance of deep neural networks. Consequently, there has been a surge of interest in explaining its empirical success. We focus, in particular, on understanding the role played by normalization, a key component of the SAM updates. We theoretically and empirically study the effect of normalization in SAM for both convex and non-convex functions, revealing two key roles played by normalization: i) it helps in stabilizing the algorithm; and ii) it e"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2305.15287","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/2305.15287/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":"2305.15287","created_at":"2026-07-05T07:03:59.361908+00:00"},{"alias_kind":"arxiv_version","alias_value":"2305.15287v2","created_at":"2026-07-05T07:03:59.361908+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2305.15287","created_at":"2026-07-05T07:03:59.361908+00:00"},{"alias_kind":"pith_short_12","alias_value":"YDATY2SOFXLO","created_at":"2026-07-05T07:03:59.361908+00:00"},{"alias_kind":"pith_short_16","alias_value":"YDATY2SOFXLOMSRW","created_at":"2026-07-05T07:03:59.361908+00:00"},{"alias_kind":"pith_short_8","alias_value":"YDATY2SO","created_at":"2026-07-05T07:03:59.361908+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.24399","citing_title":"LightSAM: Parameter-Agnostic Sharpness-Aware Minimization","ref_index":8,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YDATY2SOFXLOMSRWUDIEQ255IL","json":"https://pith.science/pith/YDATY2SOFXLOMSRWUDIEQ255IL.json","graph_json":"https://pith.science/api/pith-number/YDATY2SOFXLOMSRWUDIEQ255IL/graph.json","events_json":"https://pith.science/api/pith-number/YDATY2SOFXLOMSRWUDIEQ255IL/events.json","paper":"https://pith.science/paper/YDATY2SO"},"agent_actions":{"view_html":"https://pith.science/pith/YDATY2SOFXLOMSRWUDIEQ255IL","download_json":"https://pith.science/pith/YDATY2SOFXLOMSRWUDIEQ255IL.json","view_paper":"https://pith.science/paper/YDATY2SO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2305.15287&json=true","fetch_graph":"https://pith.science/api/pith-number/YDATY2SOFXLOMSRWUDIEQ255IL/graph.json","fetch_events":"https://pith.science/api/pith-number/YDATY2SOFXLOMSRWUDIEQ255IL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YDATY2SOFXLOMSRWUDIEQ255IL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YDATY2SOFXLOMSRWUDIEQ255IL/action/storage_attestation","attest_author":"https://pith.science/pith/YDATY2SOFXLOMSRWUDIEQ255IL/action/author_attestation","sign_citation":"https://pith.science/pith/YDATY2SOFXLOMSRWUDIEQ255IL/action/citation_signature","submit_replication":"https://pith.science/pith/YDATY2SOFXLOMSRWUDIEQ255IL/action/replication_record"}},"created_at":"2026-07-05T07:03:59.361908+00:00","updated_at":"2026-07-05T07:03:59.361908+00:00"}