{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:YQWZYTKMHX6OOANJOINTLDLM5O","short_pith_number":"pith:YQWZYTKM","schema_version":"1.0","canonical_sha256":"c42d9c4d4c3dfce701a9721b358d6cebb2f40aab2473d09237f8091eb496349f","source":{"kind":"arxiv","id":"2206.10654","version":3},"attestation_state":"computed","paper":{"title":"On the Maximum Hessian Eigenvalue and Generalization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Jeremy Cohen, Simran Kaur, Zachary C. Lipton","submitted_at":"2022-06-21T18:06:24Z","abstract_excerpt":"The mechanisms by which certain training interventions, such as increasing learning rates and applying batch normalization, improve the generalization of deep networks remains a mystery. Prior works have speculated that \"flatter\" solutions generalize better than \"sharper\" solutions to unseen data, motivating several metrics for measuring flatness (particularly $\\lambda_{max}$, the largest eigenvalue of the Hessian of the loss); and algorithms, such as Sharpness-Aware Minimization (SAM) [1], that directly optimize for flatness. Other works question the link between $\\lambda_{max}$ and generaliz"},"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":"2206.10654","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2022-06-21T18:06:24Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"cd8053f7380286a2be44df2a57be3c4455d010b896b44d9148e7256caa2ea458","abstract_canon_sha256":"7ab437731c7a5d6fdca1b74ac4b95d9a4c532a5f362bdc3926ba3dda90711c16"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:13:15.862854Z","signature_b64":"d1wPEtMiDoymTNqrIBSOd4zlPg3QQNrk3rzGXbeqtLmjAs94ROucaCKWpeaCq6OZYCgfR2BKi/TtZPzQKgdYAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c42d9c4d4c3dfce701a9721b358d6cebb2f40aab2473d09237f8091eb496349f","last_reissued_at":"2026-07-05T06:13:15.862437Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:13:15.862437Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"On the Maximum Hessian Eigenvalue and Generalization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Jeremy Cohen, Simran Kaur, Zachary C. Lipton","submitted_at":"2022-06-21T18:06:24Z","abstract_excerpt":"The mechanisms by which certain training interventions, such as increasing learning rates and applying batch normalization, improve the generalization of deep networks remains a mystery. Prior works have speculated that \"flatter\" solutions generalize better than \"sharper\" solutions to unseen data, motivating several metrics for measuring flatness (particularly $\\lambda_{max}$, the largest eigenvalue of the Hessian of the loss); and algorithms, such as Sharpness-Aware Minimization (SAM) [1], that directly optimize for flatness. Other works question the link between $\\lambda_{max}$ and generaliz"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2206.10654","kind":"arxiv","version":3},"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/2206.10654/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":"2206.10654","created_at":"2026-07-05T06:13:15.862493+00:00"},{"alias_kind":"arxiv_version","alias_value":"2206.10654v3","created_at":"2026-07-05T06:13:15.862493+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2206.10654","created_at":"2026-07-05T06:13:15.862493+00:00"},{"alias_kind":"pith_short_12","alias_value":"YQWZYTKMHX6O","created_at":"2026-07-05T06:13:15.862493+00:00"},{"alias_kind":"pith_short_16","alias_value":"YQWZYTKMHX6OOANJ","created_at":"2026-07-05T06:13:15.862493+00:00"},{"alias_kind":"pith_short_8","alias_value":"YQWZYTKM","created_at":"2026-07-05T06:13:15.862493+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.05436","citing_title":"Estimating Implicit Regularization in Deep Learning","ref_index":19,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YQWZYTKMHX6OOANJOINTLDLM5O","json":"https://pith.science/pith/YQWZYTKMHX6OOANJOINTLDLM5O.json","graph_json":"https://pith.science/api/pith-number/YQWZYTKMHX6OOANJOINTLDLM5O/graph.json","events_json":"https://pith.science/api/pith-number/YQWZYTKMHX6OOANJOINTLDLM5O/events.json","paper":"https://pith.science/paper/YQWZYTKM"},"agent_actions":{"view_html":"https://pith.science/pith/YQWZYTKMHX6OOANJOINTLDLM5O","download_json":"https://pith.science/pith/YQWZYTKMHX6OOANJOINTLDLM5O.json","view_paper":"https://pith.science/paper/YQWZYTKM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2206.10654&json=true","fetch_graph":"https://pith.science/api/pith-number/YQWZYTKMHX6OOANJOINTLDLM5O/graph.json","fetch_events":"https://pith.science/api/pith-number/YQWZYTKMHX6OOANJOINTLDLM5O/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YQWZYTKMHX6OOANJOINTLDLM5O/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YQWZYTKMHX6OOANJOINTLDLM5O/action/storage_attestation","attest_author":"https://pith.science/pith/YQWZYTKMHX6OOANJOINTLDLM5O/action/author_attestation","sign_citation":"https://pith.science/pith/YQWZYTKMHX6OOANJOINTLDLM5O/action/citation_signature","submit_replication":"https://pith.science/pith/YQWZYTKMHX6OOANJOINTLDLM5O/action/replication_record"}},"created_at":"2026-07-05T06:13:15.862493+00:00","updated_at":"2026-07-05T06:13:15.862493+00:00"}