{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2018:TFXYSFKZSNUSNPR5FDE2W3OK5X","short_pith_number":"pith:TFXYSFKZ","schema_version":"1.0","canonical_sha256":"996f891559936926be3d28c9ab6dcaedceca439e31c7bc0840e9b8fd09ec5eeb","source":{"kind":"arxiv","id":"1809.02839","version":4},"attestation_state":"computed","paper":{"title":"Efficient and Robust Parallel DNN Training through Model Parallelism on Multi-GPU Platform","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.DC","authors_text":"Chia-Lin Yang, Chi-Chung Chen, Hsiang-Yun Cheng","submitted_at":"2018-09-08T17:21:58Z","abstract_excerpt":"The training process of Deep Neural Network (DNN) is compute-intensive, often taking days to weeks to train a DNN model. Therefore, parallel execution of DNN training on GPUs is a widely adopted approach to speed up the process nowadays. Due to the implementation simplicity, data parallelism is currently the most commonly used parallelization method. Nonetheless, data parallelism suffers from excessive inter-GPU communication overhead due to frequent weight synchronization among GPUs. Another approach is pipelined model parallelism, which partitions a DNN model among GPUs, and processes multip"},"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":"1809.02839","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2018-09-08T17:21:58Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"bf443d06fd02c670210be19ff5fb63071217a28637a268588753fc72aaa1ae29","abstract_canon_sha256":"856b9279fe2bc0da330dcb763ee808bcbf8cd733778915b93b053d62aee9066d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T00:15:00.946550Z","signature_b64":"UPdDGbzTppPBtC5rG/QhB0b9YHqsMfUJNl0KMkdtCyok+jNbLMTXuBGbAqGzprxWE0XXU+W4uqlJvxPLqsixAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"996f891559936926be3d28c9ab6dcaedceca439e31c7bc0840e9b8fd09ec5eeb","last_reissued_at":"2026-07-05T00:15:00.946114Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T00:15:00.946114Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Efficient and Robust Parallel DNN Training through Model Parallelism on Multi-GPU Platform","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.DC","authors_text":"Chia-Lin Yang, Chi-Chung Chen, Hsiang-Yun Cheng","submitted_at":"2018-09-08T17:21:58Z","abstract_excerpt":"The training process of Deep Neural Network (DNN) is compute-intensive, often taking days to weeks to train a DNN model. Therefore, parallel execution of DNN training on GPUs is a widely adopted approach to speed up the process nowadays. Due to the implementation simplicity, data parallelism is currently the most commonly used parallelization method. Nonetheless, data parallelism suffers from excessive inter-GPU communication overhead due to frequent weight synchronization among GPUs. Another approach is pipelined model parallelism, which partitions a DNN model among GPUs, and processes multip"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1809.02839","kind":"arxiv","version":4},"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/1809.02839/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":"1809.02839","created_at":"2026-07-05T00:15:00.946183+00:00"},{"alias_kind":"arxiv_version","alias_value":"1809.02839v4","created_at":"2026-07-05T00:15:00.946183+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1809.02839","created_at":"2026-07-05T00:15:00.946183+00:00"},{"alias_kind":"pith_short_12","alias_value":"TFXYSFKZSNUS","created_at":"2026-07-05T00:15:00.946183+00:00"},{"alias_kind":"pith_short_16","alias_value":"TFXYSFKZSNUSNPR5","created_at":"2026-07-05T00:15:00.946183+00:00"},{"alias_kind":"pith_short_8","alias_value":"TFXYSFKZ","created_at":"2026-07-05T00:15:00.946183+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.07881","citing_title":"Breaking the Bubble: Asynchronous Pipeline Parallel Training with Bounded Weight Inconsistency","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2604.27085","citing_title":"Efficient Training on Multiple Consumer GPUs with RoundPipe","ref_index":3,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/TFXYSFKZSNUSNPR5FDE2W3OK5X","json":"https://pith.science/pith/TFXYSFKZSNUSNPR5FDE2W3OK5X.json","graph_json":"https://pith.science/api/pith-number/TFXYSFKZSNUSNPR5FDE2W3OK5X/graph.json","events_json":"https://pith.science/api/pith-number/TFXYSFKZSNUSNPR5FDE2W3OK5X/events.json","paper":"https://pith.science/paper/TFXYSFKZ"},"agent_actions":{"view_html":"https://pith.science/pith/TFXYSFKZSNUSNPR5FDE2W3OK5X","download_json":"https://pith.science/pith/TFXYSFKZSNUSNPR5FDE2W3OK5X.json","view_paper":"https://pith.science/paper/TFXYSFKZ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1809.02839&json=true","fetch_graph":"https://pith.science/api/pith-number/TFXYSFKZSNUSNPR5FDE2W3OK5X/graph.json","fetch_events":"https://pith.science/api/pith-number/TFXYSFKZSNUSNPR5FDE2W3OK5X/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TFXYSFKZSNUSNPR5FDE2W3OK5X/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TFXYSFKZSNUSNPR5FDE2W3OK5X/action/storage_attestation","attest_author":"https://pith.science/pith/TFXYSFKZSNUSNPR5FDE2W3OK5X/action/author_attestation","sign_citation":"https://pith.science/pith/TFXYSFKZSNUSNPR5FDE2W3OK5X/action/citation_signature","submit_replication":"https://pith.science/pith/TFXYSFKZSNUSNPR5FDE2W3OK5X/action/replication_record"}},"created_at":"2026-07-05T00:15:00.946183+00:00","updated_at":"2026-07-05T00:15:00.946183+00:00"}