{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2018:6VARSBRX6E3HDBTBZP3WZOXG6B","short_pith_number":"pith:6VARSBRX","schema_version":"1.0","canonical_sha256":"f541190637f136718661cbf76cbae6f0739e46cf4ee80f156d42ce5b21a0d6c1","source":{"kind":"arxiv","id":"1811.05233","version":2},"attestation_state":"computed","paper":{"title":"Massively Distributed SGD: ImageNet/ResNet-50 Training in a Flash","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Hiroaki Mikami, Hisahiro Suganuma, Pongsakorn U-Chupala, Yoshiki Tanaka, Yuichi Kageyama","submitted_at":"2018-11-13T11:52:04Z","abstract_excerpt":"Scaling the distributed deep learning to a massive GPU cluster level is challenging due to the instability of the large mini-batch training and the overhead of the gradient synchronization. We address the instability of the large mini-batch training with batch-size control and label smoothing. We address the overhead of the gradient synchronization with 2D-Torus all-reduce. Specifically, 2D-Torus all-reduce arranges GPUs in a logical 2D grid and performs a series of collective operation in different orientations. These two techniques are implemented with Neural Network Libraries (NNL). We have"},"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":"1811.05233","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2018-11-13T11:52:04Z","cross_cats_sorted":["cs.CV"],"title_canon_sha256":"d6cd81252e89069e2d685a8db0e14d63ceaabc34f2ea6bef568baa2887ba52e9","abstract_canon_sha256":"64a9eeb5c319b65bc3c68598ef29104918cd63cfecf9a846b0e74e3fb43d040f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:52:04.414826Z","signature_b64":"qVacuSf+RVHKpILiCUhTrivj8gHRe+kvKEhIUqCjrP17nIyjT+tp6SDaft9zouuDg08rqHx2OQZDLUPgpd41Cw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f541190637f136718661cbf76cbae6f0739e46cf4ee80f156d42ce5b21a0d6c1","last_reissued_at":"2026-05-17T23:52:04.414293Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:52:04.414293Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Massively Distributed SGD: ImageNet/ResNet-50 Training in a Flash","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.LG","authors_text":"Hiroaki Mikami, Hisahiro Suganuma, Pongsakorn U-Chupala, Yoshiki Tanaka, Yuichi Kageyama","submitted_at":"2018-11-13T11:52:04Z","abstract_excerpt":"Scaling the distributed deep learning to a massive GPU cluster level is challenging due to the instability of the large mini-batch training and the overhead of the gradient synchronization. We address the instability of the large mini-batch training with batch-size control and label smoothing. We address the overhead of the gradient synchronization with 2D-Torus all-reduce. Specifically, 2D-Torus all-reduce arranges GPUs in a logical 2D grid and performs a series of collective operation in different orientations. These two techniques are implemented with Neural Network Libraries (NNL). We have"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1811.05233","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":""},"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":"1811.05233","created_at":"2026-05-17T23:52:04.414386+00:00"},{"alias_kind":"arxiv_version","alias_value":"1811.05233v2","created_at":"2026-05-17T23:52:04.414386+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1811.05233","created_at":"2026-05-17T23:52:04.414386+00:00"},{"alias_kind":"pith_short_12","alias_value":"6VARSBRX6E3H","created_at":"2026-05-18T12:32:11.075285+00:00"},{"alias_kind":"pith_short_16","alias_value":"6VARSBRX6E3HDBTB","created_at":"2026-05-18T12:32:11.075285+00:00"},{"alias_kind":"pith_short_8","alias_value":"6VARSBRX","created_at":"2026-05-18T12:32:11.075285+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"1904.00962","citing_title":"Large Batch Optimization for Deep Learning: Training BERT in 76 minutes","ref_index":12,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6VARSBRX6E3HDBTBZP3WZOXG6B","json":"https://pith.science/pith/6VARSBRX6E3HDBTBZP3WZOXG6B.json","graph_json":"https://pith.science/api/pith-number/6VARSBRX6E3HDBTBZP3WZOXG6B/graph.json","events_json":"https://pith.science/api/pith-number/6VARSBRX6E3HDBTBZP3WZOXG6B/events.json","paper":"https://pith.science/paper/6VARSBRX"},"agent_actions":{"view_html":"https://pith.science/pith/6VARSBRX6E3HDBTBZP3WZOXG6B","download_json":"https://pith.science/pith/6VARSBRX6E3HDBTBZP3WZOXG6B.json","view_paper":"https://pith.science/paper/6VARSBRX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1811.05233&json=true","fetch_graph":"https://pith.science/api/pith-number/6VARSBRX6E3HDBTBZP3WZOXG6B/graph.json","fetch_events":"https://pith.science/api/pith-number/6VARSBRX6E3HDBTBZP3WZOXG6B/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6VARSBRX6E3HDBTBZP3WZOXG6B/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6VARSBRX6E3HDBTBZP3WZOXG6B/action/storage_attestation","attest_author":"https://pith.science/pith/6VARSBRX6E3HDBTBZP3WZOXG6B/action/author_attestation","sign_citation":"https://pith.science/pith/6VARSBRX6E3HDBTBZP3WZOXG6B/action/citation_signature","submit_replication":"https://pith.science/pith/6VARSBRX6E3HDBTBZP3WZOXG6B/action/replication_record"}},"created_at":"2026-05-17T23:52:04.414386+00:00","updated_at":"2026-05-17T23:52:04.414386+00:00"}