{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:Q6CEY27D3C6377PY2LWDTNTXWL","short_pith_number":"pith:Q6CEY27D","schema_version":"1.0","canonical_sha256":"87844c6be3d8bdbffdf8d2ec39b677b2c57e34de2364d3f9ae05f95fddd002ac","source":{"kind":"arxiv","id":"2301.13703","version":2},"attestation_state":"computed","paper":{"title":"Dissecting the Effects of SGD Noise in Distinct Regimes of Deep Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cond-mat.dis-nn"],"primary_cat":"cs.LG","authors_text":"Antonio Sclocchi, Mario Geiger, Matthieu Wyart","submitted_at":"2023-01-31T15:22:24Z","abstract_excerpt":"Understanding when the noise in stochastic gradient descent (SGD) affects generalization of deep neural networks remains a challenge, complicated by the fact that networks can operate in distinct training regimes. Here we study how the magnitude of this noise $T$ affects performance as the size of the training set $P$ and the scale of initialization $\\alpha$ are varied. For gradient descent, $\\alpha$ is a key parameter that controls if the network is `lazy'($\\alpha\\gg1$) or instead learns features ($\\alpha\\ll1$). For classification of MNIST and CIFAR10 images, our central results are: (i) obta"},"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":"2301.13703","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-01-31T15:22:24Z","cross_cats_sorted":["cond-mat.dis-nn"],"title_canon_sha256":"7123ed565dc3d032214198b99fd970012d4b9e8121f76dedb02543029f3ff203","abstract_canon_sha256":"24f89acb595ebf7baa1945fc176f1018bf62e8478942adaaf120b162932ef48a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:15:23.132979Z","signature_b64":"Ug43vkNpZbGdGXwrBr+2AFlHp4HSZb3vkcW2V7NeSytnYqUSFNEB9+jAqNt9i292QxlKTybNnqP/ACGEmssxAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"87844c6be3d8bdbffdf8d2ec39b677b2c57e34de2364d3f9ae05f95fddd002ac","last_reissued_at":"2026-07-05T06:15:23.132491Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:15:23.132491Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Dissecting the Effects of SGD Noise in Distinct Regimes of Deep Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cond-mat.dis-nn"],"primary_cat":"cs.LG","authors_text":"Antonio Sclocchi, Mario Geiger, Matthieu Wyart","submitted_at":"2023-01-31T15:22:24Z","abstract_excerpt":"Understanding when the noise in stochastic gradient descent (SGD) affects generalization of deep neural networks remains a challenge, complicated by the fact that networks can operate in distinct training regimes. Here we study how the magnitude of this noise $T$ affects performance as the size of the training set $P$ and the scale of initialization $\\alpha$ are varied. For gradient descent, $\\alpha$ is a key parameter that controls if the network is `lazy'($\\alpha\\gg1$) or instead learns features ($\\alpha\\ll1$). For classification of MNIST and CIFAR10 images, our central results are: (i) obta"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2301.13703","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/2301.13703/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":"2301.13703","created_at":"2026-07-05T06:15:23.132553+00:00"},{"alias_kind":"arxiv_version","alias_value":"2301.13703v2","created_at":"2026-07-05T06:15:23.132553+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2301.13703","created_at":"2026-07-05T06:15:23.132553+00:00"},{"alias_kind":"pith_short_12","alias_value":"Q6CEY27D3C63","created_at":"2026-07-05T06:15:23.132553+00:00"},{"alias_kind":"pith_short_16","alias_value":"Q6CEY27D3C6377PY","created_at":"2026-07-05T06:15:23.132553+00:00"},{"alias_kind":"pith_short_8","alias_value":"Q6CEY27D","created_at":"2026-07-05T06:15:23.132553+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.28486","citing_title":"Spectral phase transitions and trainability in neural network learning dynamics","ref_index":13,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/Q6CEY27D3C6377PY2LWDTNTXWL","json":"https://pith.science/pith/Q6CEY27D3C6377PY2LWDTNTXWL.json","graph_json":"https://pith.science/api/pith-number/Q6CEY27D3C6377PY2LWDTNTXWL/graph.json","events_json":"https://pith.science/api/pith-number/Q6CEY27D3C6377PY2LWDTNTXWL/events.json","paper":"https://pith.science/paper/Q6CEY27D"},"agent_actions":{"view_html":"https://pith.science/pith/Q6CEY27D3C6377PY2LWDTNTXWL","download_json":"https://pith.science/pith/Q6CEY27D3C6377PY2LWDTNTXWL.json","view_paper":"https://pith.science/paper/Q6CEY27D","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2301.13703&json=true","fetch_graph":"https://pith.science/api/pith-number/Q6CEY27D3C6377PY2LWDTNTXWL/graph.json","fetch_events":"https://pith.science/api/pith-number/Q6CEY27D3C6377PY2LWDTNTXWL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/Q6CEY27D3C6377PY2LWDTNTXWL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/Q6CEY27D3C6377PY2LWDTNTXWL/action/storage_attestation","attest_author":"https://pith.science/pith/Q6CEY27D3C6377PY2LWDTNTXWL/action/author_attestation","sign_citation":"https://pith.science/pith/Q6CEY27D3C6377PY2LWDTNTXWL/action/citation_signature","submit_replication":"https://pith.science/pith/Q6CEY27D3C6377PY2LWDTNTXWL/action/replication_record"}},"created_at":"2026-07-05T06:15:23.132553+00:00","updated_at":"2026-07-05T06:15:23.132553+00:00"}