{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2019:AXXRXJITB6QIRXW3QXIOEOHX5N","short_pith_number":"pith:AXXRXJIT","canonical_record":{"source":{"id":"1905.03960","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2019-05-10T06:22:35Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"7e54a02ae94d00592d77a4b8b0444111c7921d2db1920d3126e6b5cb9f7a9b5b","abstract_canon_sha256":"c7d5491e248f3ed85ebfa14fd4ca41e9e1478e8f1ec12fb44f4de97049e3ea09"},"schema_version":"1.0"},"canonical_sha256":"05ef1ba5130fa088dedb85d0e238f7eb56ddee1a529a48d001592fc015e6a4b6","source":{"kind":"arxiv","id":"1905.03960","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1905.03960","created_at":"2026-05-17T23:46:35Z"},{"alias_kind":"arxiv_version","alias_value":"1905.03960v1","created_at":"2026-05-17T23:46:35Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1905.03960","created_at":"2026-05-17T23:46:35Z"},{"alias_kind":"pith_short_12","alias_value":"AXXRXJITB6QI","created_at":"2026-05-18T12:33:12Z"},{"alias_kind":"pith_short_16","alias_value":"AXXRXJITB6QIRXW3","created_at":"2026-05-18T12:33:12Z"},{"alias_kind":"pith_short_8","alias_value":"AXXRXJIT","created_at":"2026-05-18T12:33:12Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2019:AXXRXJITB6QIRXW3QXIOEOHX5N","target":"record","payload":{"canonical_record":{"source":{"id":"1905.03960","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2019-05-10T06:22:35Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"7e54a02ae94d00592d77a4b8b0444111c7921d2db1920d3126e6b5cb9f7a9b5b","abstract_canon_sha256":"c7d5491e248f3ed85ebfa14fd4ca41e9e1478e8f1ec12fb44f4de97049e3ea09"},"schema_version":"1.0"},"canonical_sha256":"05ef1ba5130fa088dedb85d0e238f7eb56ddee1a529a48d001592fc015e6a4b6","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:46:35.037847Z","signature_b64":"R+Nu/8m6xdZhkLlVe9KLBfArfJUt94COSi5eggMOWLFMl8TmWAjm94vjYLRnaHdOvXvtqr3t+EGzCCSudlUODw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"05ef1ba5130fa088dedb85d0e238f7eb56ddee1a529a48d001592fc015e6a4b6","last_reissued_at":"2026-05-17T23:46:35.037151Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:46:35.037151Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"1905.03960","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:46:35Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"kn60ox7isExZGaU9o/WJrX4kALLJgEd6ux3V8fCbzorwGFTTIroJChVdLARENf55IA6a5JTn3h1T9yvVH5TaBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-07T20:08:58.855519Z"},"content_sha256":"32ef08226366f92555945f2445da9eb0da1276efdccaee21ff42980854df13a1","schema_version":"1.0","event_id":"sha256:32ef08226366f92555945f2445da9eb0da1276efdccaee21ff42980854df13a1"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2019:AXXRXJITB6QIRXW3QXIOEOHX5N","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Priority-based Parameter Propagation for Distributed DNN Training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.DC","authors_text":"Alexandra Fedorova, Anand Jayarajan, Garth Gibson, Gennady Pekhimenko, Jinliang Wei","submitted_at":"2019-05-10T06:22:35Z","abstract_excerpt":"Data parallel training is widely used for scaling distributed deep neural network (DNN) training. However, the performance benefits are often limited by the communication-heavy parameter synchronization step. In this paper, we take advantage of the domain specific knowledge of DNN training and overlap parameter synchronization with computation in order to improve the training performance. We make two key observations: (1) the optimal data representation granularity for the communication may differ from that used by the underlying DNN model implementation and (2) different parameters can afford"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1905.03960","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":""},"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":"2026-05-17T23:46:35Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"TF5O7JMLxvW4BGp2NyIfaSDXwMLDRBUmbG3Oj/+uSODcMg/me0MvSIggm0rOBC3WWyFXvdC1IrPj+V2AbDtaAQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-07T20:08:58.856251Z"},"content_sha256":"0414ca3be7a4c80189c0ee8f8bd5937325aeacfe9efd6e439c854194bfe23176","schema_version":"1.0","event_id":"sha256:0414ca3be7a4c80189c0ee8f8bd5937325aeacfe9efd6e439c854194bfe23176"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/AXXRXJITB6QIRXW3QXIOEOHX5N/bundle.json","state_url":"https://pith.science/pith/AXXRXJITB6QIRXW3QXIOEOHX5N/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/AXXRXJITB6QIRXW3QXIOEOHX5N/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":"2026-06-07T20:08:58Z","links":{"resolver":"https://pith.science/pith/AXXRXJITB6QIRXW3QXIOEOHX5N","bundle":"https://pith.science/pith/AXXRXJITB6QIRXW3QXIOEOHX5N/bundle.json","state":"https://pith.science/pith/AXXRXJITB6QIRXW3QXIOEOHX5N/state.json","well_known_bundle":"https://pith.science/.well-known/pith/AXXRXJITB6QIRXW3QXIOEOHX5N/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2019:AXXRXJITB6QIRXW3QXIOEOHX5N","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":"c7d5491e248f3ed85ebfa14fd4ca41e9e1478e8f1ec12fb44f4de97049e3ea09","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2019-05-10T06:22:35Z","title_canon_sha256":"7e54a02ae94d00592d77a4b8b0444111c7921d2db1920d3126e6b5cb9f7a9b5b"},"schema_version":"1.0","source":{"id":"1905.03960","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"1905.03960","created_at":"2026-05-17T23:46:35Z"},{"alias_kind":"arxiv_version","alias_value":"1905.03960v1","created_at":"2026-05-17T23:46:35Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1905.03960","created_at":"2026-05-17T23:46:35Z"},{"alias_kind":"pith_short_12","alias_value":"AXXRXJITB6QI","created_at":"2026-05-18T12:33:12Z"},{"alias_kind":"pith_short_16","alias_value":"AXXRXJITB6QIRXW3","created_at":"2026-05-18T12:33:12Z"},{"alias_kind":"pith_short_8","alias_value":"AXXRXJIT","created_at":"2026-05-18T12:33:12Z"}],"graph_snapshots":[{"event_id":"sha256:0414ca3be7a4c80189c0ee8f8bd5937325aeacfe9efd6e439c854194bfe23176","target":"graph","created_at":"2026-05-17T23:46:35Z","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"},"paper":{"abstract_excerpt":"Data parallel training is widely used for scaling distributed deep neural network (DNN) training. However, the performance benefits are often limited by the communication-heavy parameter synchronization step. In this paper, we take advantage of the domain specific knowledge of DNN training and overlap parameter synchronization with computation in order to improve the training performance. We make two key observations: (1) the optimal data representation granularity for the communication may differ from that used by the underlying DNN model implementation and (2) different parameters can afford","authors_text":"Alexandra Fedorova, Anand Jayarajan, Garth Gibson, Gennady Pekhimenko, Jinliang Wei","cross_cats":["cs.LG"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2019-05-10T06:22:35Z","title":"Priority-based Parameter Propagation for Distributed DNN Training"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1905.03960","kind":"arxiv","version":1},"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":"sha256:32ef08226366f92555945f2445da9eb0da1276efdccaee21ff42980854df13a1","target":"record","created_at":"2026-05-17T23:46:35Z","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":"c7d5491e248f3ed85ebfa14fd4ca41e9e1478e8f1ec12fb44f4de97049e3ea09","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2019-05-10T06:22:35Z","title_canon_sha256":"7e54a02ae94d00592d77a4b8b0444111c7921d2db1920d3126e6b5cb9f7a9b5b"},"schema_version":"1.0","source":{"id":"1905.03960","kind":"arxiv","version":1}},"canonical_sha256":"05ef1ba5130fa088dedb85d0e238f7eb56ddee1a529a48d001592fc015e6a4b6","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"05ef1ba5130fa088dedb85d0e238f7eb56ddee1a529a48d001592fc015e6a4b6","first_computed_at":"2026-05-17T23:46:35.037151Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:46:35.037151Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"R+Nu/8m6xdZhkLlVe9KLBfArfJUt94COSi5eggMOWLFMl8TmWAjm94vjYLRnaHdOvXvtqr3t+EGzCCSudlUODw==","signature_status":"signed_v1","signed_at":"2026-05-17T23:46:35.037847Z","signed_message":"canonical_sha256_bytes"},"source_id":"1905.03960","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:32ef08226366f92555945f2445da9eb0da1276efdccaee21ff42980854df13a1","sha256:0414ca3be7a4c80189c0ee8f8bd5937325aeacfe9efd6e439c854194bfe23176"],"state_sha256":"e5cd77b755a1758f98d21057177d48dd9e2f5bf65969257ad7d948d6235223f2"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"QxRbnXv++SpcIGe2tW8makkf7TZVof15gLDaNQxz/nTQyPDMCHtCfa21TKQa7KJ2oTr3xpfHI3/ptWH7w5ecAw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-07T20:08:58.859499Z","bundle_sha256":"174a289ba56161ef5431ced0497dea08afd2dab2dba78ab97992be4bee55e800"}}