{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2019:CA3YCGUJLHX4PQQDBMEJOYLEL5","short_pith_number":"pith:CA3YCGUJ","canonical_record":{"source":{"id":"1902.06855","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2019-02-19T01:18:56Z","cross_cats_sorted":[],"title_canon_sha256":"dbfed68a024ab8842aa74c06dafa7cbb171eee86a571b40fe6c6bd913a30bae4","abstract_canon_sha256":"95e5bb278978ef34459591f01996abc0074899f5751bd6d3f0e9d7e62585d57f"},"schema_version":"1.0"},"canonical_sha256":"1037811a8959efc7c2030b089761645f688e3efee87ce7437b051eae453e15bd","source":{"kind":"arxiv","id":"1902.06855","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1902.06855","created_at":"2026-07-05T00:13:41Z"},{"alias_kind":"arxiv_version","alias_value":"1902.06855v3","created_at":"2026-07-05T00:13:41Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1902.06855","created_at":"2026-07-05T00:13:41Z"},{"alias_kind":"pith_short_12","alias_value":"CA3YCGUJLHX4","created_at":"2026-07-05T00:13:41Z"},{"alias_kind":"pith_short_16","alias_value":"CA3YCGUJLHX4PQQD","created_at":"2026-07-05T00:13:41Z"},{"alias_kind":"pith_short_8","alias_value":"CA3YCGUJ","created_at":"2026-07-05T00:13:41Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2019:CA3YCGUJLHX4PQQDBMEJOYLEL5","target":"record","payload":{"canonical_record":{"source":{"id":"1902.06855","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2019-02-19T01:18:56Z","cross_cats_sorted":[],"title_canon_sha256":"dbfed68a024ab8842aa74c06dafa7cbb171eee86a571b40fe6c6bd913a30bae4","abstract_canon_sha256":"95e5bb278978ef34459591f01996abc0074899f5751bd6d3f0e9d7e62585d57f"},"schema_version":"1.0"},"canonical_sha256":"1037811a8959efc7c2030b089761645f688e3efee87ce7437b051eae453e15bd","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T00:13:41.883022Z","signature_b64":"3Hk/tKYNg9LmvIQtzM5I7sazsO81BktM8hL0Lec5Y4y+MAOddwE3vIKgSjlFf8yaeAgp3SLMGzNuWBpYkuZ9Cg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1037811a8959efc7c2030b089761645f688e3efee87ce7437b051eae453e15bd","last_reissued_at":"2026-07-05T00:13:41.882682Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T00:13:41.882682Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"1902.06855","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T00:13:41Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"iirZAIrYWdpEgEHe3hI52Yh9gG/6YIZv3E3GyKv5gIQwABe/Q+KgVo+RTGjdqOb3sfF+vt6edQH3lzlFyAL1BQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-18T16:14:45.859388Z"},"content_sha256":"4c56009f20f3bd2865f8683e0bf0352e879220357baaaf29be92813ce000216a","schema_version":"1.0","event_id":"sha256:4c56009f20f3bd2865f8683e0bf0352e879220357baaaf29be92813ce000216a"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2019:CA3YCGUJLHX4PQQDBMEJOYLEL5","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Optimizing Network Performance for Distributed DNN Training on GPU Clusters: ImageNet/AlexNet Training in 1.5 Minutes","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.DC","authors_text":"Peng Sun, Ruobing Han, Shengen Yan, Wansen Feng, Yonggang Wen","submitted_at":"2019-02-19T01:18:56Z","abstract_excerpt":"It is important to scale out deep neural network (DNN) training for reducing model training time. The high communication overhead is one of the major performance bottlenecks for distributed DNN training across multiple GPUs. Our investigations have shown that popular open-source DNN systems could only achieve 2.5 speedup ratio on 64 GPUs connected by 56 Gbps network. To address this problem, we propose a communication backend named GradientFlow for distributed DNN training, and employ a set of network optimization techniques. First, we integrate ring-based allreduce, mixed-precision training, "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1902.06855","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/1902.06855/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T00:13:41Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"8ZCLUTxe9xNxK5FGaqS5N57Axp4twKMdTt1IO000EjATKFW2nZD+ZcD445UCHssiblTLI196b/lEuzMT4mYDAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-18T16:14:45.859763Z"},"content_sha256":"ca1e44f5c90be8c40374c5ca39fd51b2182c40bb5664bd3f062a725da067ae2f","schema_version":"1.0","event_id":"sha256:ca1e44f5c90be8c40374c5ca39fd51b2182c40bb5664bd3f062a725da067ae2f"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/CA3YCGUJLHX4PQQDBMEJOYLEL5/bundle.json","state_url":"https://pith.science/pith/CA3YCGUJLHX4PQQDBMEJOYLEL5/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/CA3YCGUJLHX4PQQDBMEJOYLEL5/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-07-18T16:14:45Z","links":{"resolver":"https://pith.science/pith/CA3YCGUJLHX4PQQDBMEJOYLEL5","bundle":"https://pith.science/pith/CA3YCGUJLHX4PQQDBMEJOYLEL5/bundle.json","state":"https://pith.science/pith/CA3YCGUJLHX4PQQDBMEJOYLEL5/state.json","well_known_bundle":"https://pith.science/.well-known/pith/CA3YCGUJLHX4PQQDBMEJOYLEL5/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2019:CA3YCGUJLHX4PQQDBMEJOYLEL5","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"95e5bb278978ef34459591f01996abc0074899f5751bd6d3f0e9d7e62585d57f","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2019-02-19T01:18:56Z","title_canon_sha256":"dbfed68a024ab8842aa74c06dafa7cbb171eee86a571b40fe6c6bd913a30bae4"},"schema_version":"1.0","source":{"id":"1902.06855","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1902.06855","created_at":"2026-07-05T00:13:41Z"},{"alias_kind":"arxiv_version","alias_value":"1902.06855v3","created_at":"2026-07-05T00:13:41Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1902.06855","created_at":"2026-07-05T00:13:41Z"},{"alias_kind":"pith_short_12","alias_value":"CA3YCGUJLHX4","created_at":"2026-07-05T00:13:41Z"},{"alias_kind":"pith_short_16","alias_value":"CA3YCGUJLHX4PQQD","created_at":"2026-07-05T00:13:41Z"},{"alias_kind":"pith_short_8","alias_value":"CA3YCGUJ","created_at":"2026-07-05T00:13:41Z"}],"graph_snapshots":[{"event_id":"sha256:ca1e44f5c90be8c40374c5ca39fd51b2182c40bb5664bd3f062a725da067ae2f","target":"graph","created_at":"2026-07-05T00:13:41Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/1902.06855/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"It is important to scale out deep neural network (DNN) training for reducing model training time. The high communication overhead is one of the major performance bottlenecks for distributed DNN training across multiple GPUs. Our investigations have shown that popular open-source DNN systems could only achieve 2.5 speedup ratio on 64 GPUs connected by 56 Gbps network. To address this problem, we propose a communication backend named GradientFlow for distributed DNN training, and employ a set of network optimization techniques. First, we integrate ring-based allreduce, mixed-precision training, ","authors_text":"Peng Sun, Ruobing Han, Shengen Yan, Wansen Feng, Yonggang Wen","cross_cats":[],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2019-02-19T01:18:56Z","title":"Optimizing Network Performance for Distributed DNN Training on GPU Clusters: ImageNet/AlexNet Training in 1.5 Minutes"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1902.06855","kind":"arxiv","version":3},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:4c56009f20f3bd2865f8683e0bf0352e879220357baaaf29be92813ce000216a","target":"record","created_at":"2026-07-05T00:13:41Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"95e5bb278978ef34459591f01996abc0074899f5751bd6d3f0e9d7e62585d57f","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2019-02-19T01:18:56Z","title_canon_sha256":"dbfed68a024ab8842aa74c06dafa7cbb171eee86a571b40fe6c6bd913a30bae4"},"schema_version":"1.0","source":{"id":"1902.06855","kind":"arxiv","version":3}},"canonical_sha256":"1037811a8959efc7c2030b089761645f688e3efee87ce7437b051eae453e15bd","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"1037811a8959efc7c2030b089761645f688e3efee87ce7437b051eae453e15bd","first_computed_at":"2026-07-05T00:13:41.882682Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T00:13:41.882682Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"3Hk/tKYNg9LmvIQtzM5I7sazsO81BktM8hL0Lec5Y4y+MAOddwE3vIKgSjlFf8yaeAgp3SLMGzNuWBpYkuZ9Cg==","signature_status":"signed_v1","signed_at":"2026-07-05T00:13:41.883022Z","signed_message":"canonical_sha256_bytes"},"source_id":"1902.06855","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:4c56009f20f3bd2865f8683e0bf0352e879220357baaaf29be92813ce000216a","sha256:ca1e44f5c90be8c40374c5ca39fd51b2182c40bb5664bd3f062a725da067ae2f"],"state_sha256":"3d363fc6e9039e212046440ace5960f98d14c9ba77ca3c424f08747d6ba229f0"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"tOby1VEOCa1nJamV4gZ/O+TnHtCLnU/WMkjV/ib/Lxtr1bgGq+KNYehPepq5d/DnlHteMpjZl0Wj4v5KpY8GBw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-18T16:14:45.862358Z","bundle_sha256":"5465eb6f4030c16eb8bf047ef063931f904a386c4277f061a8e4d15593d13cbc"}}