{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:IA7DKZOABHOFNQHWMXKLBXSZ3D","short_pith_number":"pith:IA7DKZOA","schema_version":"1.0","canonical_sha256":"403e3565c009dc56c0f665d4b0de59d8e33f7acfaf42c670d7c5c04ca87f9edc","source":{"kind":"arxiv","id":"2112.00987","version":1},"attestation_state":"computed","paper":{"title":"On Large Batch Training and Sharp Minima: A Fokker-Planck Perspective","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["math.ST","stat.TH"],"primary_cat":"cs.LG","authors_text":"Xiaowu Dai, Yuhua Zhu","submitted_at":"2021-12-02T05:24:05Z","abstract_excerpt":"We study the statistical properties of the dynamic trajectory of stochastic gradient descent (SGD). We approximate the mini-batch SGD and the momentum SGD as stochastic differential equations (SDEs). We exploit the continuous formulation of SDE and the theory of Fokker-Planck equations to develop new results on the escaping phenomenon and the relationship with large batch and sharp minima. In particular, we find that the stochastic process solution tends to converge to flatter minima regardless of the batch size in the asymptotic regime. However, the convergence rate is rigorously proven to de"},"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":"2112.00987","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2021-12-02T05:24:05Z","cross_cats_sorted":["math.ST","stat.TH"],"title_canon_sha256":"63234f58c094a1a23145c8f3fc551dbaede326b7314546933a09e3a372f80f35","abstract_canon_sha256":"b48f4b6e2522cc461f8b550afd07f9ae82703c75d161858c157e1bd152e0f215"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:37:02.273395Z","signature_b64":"ERuVckQtHYr4mvnnQBq8QjdLODIi9/b/BjWW6PBnkVRGfB2DF7sA/g0xOTV03Hy/f3HQUBXw/YSRHqHqTvZ9CA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"403e3565c009dc56c0f665d4b0de59d8e33f7acfaf42c670d7c5c04ca87f9edc","last_reissued_at":"2026-07-05T03:37:02.272919Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:37:02.272919Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"On Large Batch Training and Sharp Minima: A Fokker-Planck Perspective","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["math.ST","stat.TH"],"primary_cat":"cs.LG","authors_text":"Xiaowu Dai, Yuhua Zhu","submitted_at":"2021-12-02T05:24:05Z","abstract_excerpt":"We study the statistical properties of the dynamic trajectory of stochastic gradient descent (SGD). We approximate the mini-batch SGD and the momentum SGD as stochastic differential equations (SDEs). We exploit the continuous formulation of SDE and the theory of Fokker-Planck equations to develop new results on the escaping phenomenon and the relationship with large batch and sharp minima. In particular, we find that the stochastic process solution tends to converge to flatter minima regardless of the batch size in the asymptotic regime. However, the convergence rate is rigorously proven to de"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2112.00987","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/2112.00987/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":"2112.00987","created_at":"2026-07-05T03:37:02.272969+00:00"},{"alias_kind":"arxiv_version","alias_value":"2112.00987v1","created_at":"2026-07-05T03:37:02.272969+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2112.00987","created_at":"2026-07-05T03:37:02.272969+00:00"},{"alias_kind":"pith_short_12","alias_value":"IA7DKZOABHOF","created_at":"2026-07-05T03:37:02.272969+00:00"},{"alias_kind":"pith_short_16","alias_value":"IA7DKZOABHOFNQHW","created_at":"2026-07-05T03:37:02.272969+00:00"},{"alias_kind":"pith_short_8","alias_value":"IA7DKZOA","created_at":"2026-07-05T03:37:02.272969+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.06412","citing_title":"Nonreversible Gauge Fields in Fokker--Planck Dynamics: Supersymmetric Hamiltonians and Learned Finite Forces","ref_index":18,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IA7DKZOABHOFNQHWMXKLBXSZ3D","json":"https://pith.science/pith/IA7DKZOABHOFNQHWMXKLBXSZ3D.json","graph_json":"https://pith.science/api/pith-number/IA7DKZOABHOFNQHWMXKLBXSZ3D/graph.json","events_json":"https://pith.science/api/pith-number/IA7DKZOABHOFNQHWMXKLBXSZ3D/events.json","paper":"https://pith.science/paper/IA7DKZOA"},"agent_actions":{"view_html":"https://pith.science/pith/IA7DKZOABHOFNQHWMXKLBXSZ3D","download_json":"https://pith.science/pith/IA7DKZOABHOFNQHWMXKLBXSZ3D.json","view_paper":"https://pith.science/paper/IA7DKZOA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2112.00987&json=true","fetch_graph":"https://pith.science/api/pith-number/IA7DKZOABHOFNQHWMXKLBXSZ3D/graph.json","fetch_events":"https://pith.science/api/pith-number/IA7DKZOABHOFNQHWMXKLBXSZ3D/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IA7DKZOABHOFNQHWMXKLBXSZ3D/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IA7DKZOABHOFNQHWMXKLBXSZ3D/action/storage_attestation","attest_author":"https://pith.science/pith/IA7DKZOABHOFNQHWMXKLBXSZ3D/action/author_attestation","sign_citation":"https://pith.science/pith/IA7DKZOABHOFNQHWMXKLBXSZ3D/action/citation_signature","submit_replication":"https://pith.science/pith/IA7DKZOABHOFNQHWMXKLBXSZ3D/action/replication_record"}},"created_at":"2026-07-05T03:37:02.272969+00:00","updated_at":"2026-07-05T03:37:02.272969+00:00"}