{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:HWOPJDCHOSCE7MQFLERLPCS4AE","short_pith_number":"pith:HWOPJDCH","schema_version":"1.0","canonical_sha256":"3d9cf48c4774844fb2055922b78a5c013c0c115204cdab2cf0e9b15ea6acb181","source":{"kind":"arxiv","id":"2305.15013","version":2},"attestation_state":"computed","paper":{"title":"Local SGD Accelerates Convergence by Exploiting Second Order Information of the Loss Function","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Linxuan Pan, Shenghui Song","submitted_at":"2023-05-24T10:54:45Z","abstract_excerpt":"With multiple iterations of updates, local statistical gradient descent (L-SGD) has been proven to be very effective in distributed machine learning schemes such as federated learning. In fact, many innovative works have shown that L-SGD with independent and identically distributed (IID) data can even outperform SGD. As a result, extensive efforts have been made to unveil the power of L-SGD. However, existing analysis failed to explain why the multiple local updates with small mini-batches of data (L-SGD) can not be replaced by the update with one big batch of data and a larger learning rate ("},"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":"2305.15013","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2023-05-24T10:54:45Z","cross_cats_sorted":[],"title_canon_sha256":"145a3269461a6d7d2c84f7923e14208a788c0e9216508a0a76f26d1cda656f16","abstract_canon_sha256":"48cddd6ba45b35cc8051014e95283cfb15df5d72f4d363b2f6931beb1e607651"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:14:11.228501Z","signature_b64":"bwJ0lmwnxVA4InnZQdkb+EZyQIHv7HLau2lTOFts01NgJ/3+dzG0WBaUwsQT7syZKWOYuuxavbp7FJCo12bQAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3d9cf48c4774844fb2055922b78a5c013c0c115204cdab2cf0e9b15ea6acb181","last_reissued_at":"2026-07-05T06:14:11.228097Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:14:11.228097Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Local SGD Accelerates Convergence by Exploiting Second Order Information of the Loss Function","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Linxuan Pan, Shenghui Song","submitted_at":"2023-05-24T10:54:45Z","abstract_excerpt":"With multiple iterations of updates, local statistical gradient descent (L-SGD) has been proven to be very effective in distributed machine learning schemes such as federated learning. In fact, many innovative works have shown that L-SGD with independent and identically distributed (IID) data can even outperform SGD. As a result, extensive efforts have been made to unveil the power of L-SGD. However, existing analysis failed to explain why the multiple local updates with small mini-batches of data (L-SGD) can not be replaced by the update with one big batch of data and a larger learning rate ("},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2305.15013","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/2305.15013/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":"2305.15013","created_at":"2026-07-05T06:14:11.228149+00:00"},{"alias_kind":"arxiv_version","alias_value":"2305.15013v2","created_at":"2026-07-05T06:14:11.228149+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2305.15013","created_at":"2026-07-05T06:14:11.228149+00:00"},{"alias_kind":"pith_short_12","alias_value":"HWOPJDCHOSCE","created_at":"2026-07-05T06:14:11.228149+00:00"},{"alias_kind":"pith_short_16","alias_value":"HWOPJDCHOSCE7MQF","created_at":"2026-07-05T06:14:11.228149+00:00"},{"alias_kind":"pith_short_8","alias_value":"HWOPJDCH","created_at":"2026-07-05T06:14:11.228149+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2412.07210","citing_title":"EDiT: A Local-SGD-Based Efficient Distributed Training Method for Large Language Models","ref_index":23,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/HWOPJDCHOSCE7MQFLERLPCS4AE","json":"https://pith.science/pith/HWOPJDCHOSCE7MQFLERLPCS4AE.json","graph_json":"https://pith.science/api/pith-number/HWOPJDCHOSCE7MQFLERLPCS4AE/graph.json","events_json":"https://pith.science/api/pith-number/HWOPJDCHOSCE7MQFLERLPCS4AE/events.json","paper":"https://pith.science/paper/HWOPJDCH"},"agent_actions":{"view_html":"https://pith.science/pith/HWOPJDCHOSCE7MQFLERLPCS4AE","download_json":"https://pith.science/pith/HWOPJDCHOSCE7MQFLERLPCS4AE.json","view_paper":"https://pith.science/paper/HWOPJDCH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2305.15013&json=true","fetch_graph":"https://pith.science/api/pith-number/HWOPJDCHOSCE7MQFLERLPCS4AE/graph.json","fetch_events":"https://pith.science/api/pith-number/HWOPJDCHOSCE7MQFLERLPCS4AE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HWOPJDCHOSCE7MQFLERLPCS4AE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HWOPJDCHOSCE7MQFLERLPCS4AE/action/storage_attestation","attest_author":"https://pith.science/pith/HWOPJDCHOSCE7MQFLERLPCS4AE/action/author_attestation","sign_citation":"https://pith.science/pith/HWOPJDCHOSCE7MQFLERLPCS4AE/action/citation_signature","submit_replication":"https://pith.science/pith/HWOPJDCHOSCE7MQFLERLPCS4AE/action/replication_record"}},"created_at":"2026-07-05T06:14:11.228149+00:00","updated_at":"2026-07-05T06:14:11.228149+00:00"}