{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:BDNCWLH7MR3OTJQMLPUHOLDCR5","short_pith_number":"pith:BDNCWLH7","schema_version":"1.0","canonical_sha256":"08da2b2cff6476e9a60c5be8772c628f7a3cabe76e63f5445996133a0c9594a9","source":{"kind":"arxiv","id":"2408.05148","version":3},"attestation_state":"computed","paper":{"title":"Impacts of floating-point non-associativity on reproducibility for HPC and deep learning applications","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.DC","authors_text":"Ada Sedova, Christopher Culver, Mark Coletti, Mathieu Taillefumier, Oscar Hernandez, Sanjif Shanmugavelu","submitted_at":"2024-08-09T16:07:37Z","abstract_excerpt":"Run to run variability in parallel programs caused by floating-point non-associativity has been known to significantly affect reproducibility in iterative algorithms, due to accumulating errors. Non-reproducibility can critically affect the efficiency and effectiveness of correctness testing for stochastic programs. Recently, the sensitivity of deep learning training and inference pipelines to floating-point non-associativity has been found to sometimes be extreme. It can prevent certification for commercial applications, accurate assessment of robustness and sensitivity, and bug detection. Ne"},"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":"2408.05148","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2024-08-09T16:07:37Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"cfebfdbf1f8335784faf57955a75695e99e4af57e8c55fbc6623272c771a7a2b","abstract_canon_sha256":"9d95d908d5e2157866ea4f60fa46c4b880212a6f779c55ce4fbabf17a09ba8d3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:28:33.824073Z","signature_b64":"JjfR6VJTalzxjw2o8atL8iDM1lnloZgsF3rw82Vx7lcch5cum39SGREs4ZNBCo+vf0oTBEsn7OUH93dC4Y3cDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"08da2b2cff6476e9a60c5be8772c628f7a3cabe76e63f5445996133a0c9594a9","last_reissued_at":"2026-07-05T09:28:33.823584Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:28:33.823584Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Impacts of floating-point non-associativity on reproducibility for HPC and deep learning applications","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.DC","authors_text":"Ada Sedova, Christopher Culver, Mark Coletti, Mathieu Taillefumier, Oscar Hernandez, Sanjif Shanmugavelu","submitted_at":"2024-08-09T16:07:37Z","abstract_excerpt":"Run to run variability in parallel programs caused by floating-point non-associativity has been known to significantly affect reproducibility in iterative algorithms, due to accumulating errors. Non-reproducibility can critically affect the efficiency and effectiveness of correctness testing for stochastic programs. Recently, the sensitivity of deep learning training and inference pipelines to floating-point non-associativity has been found to sometimes be extreme. It can prevent certification for commercial applications, accurate assessment of robustness and sensitivity, and bug detection. Ne"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2408.05148","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/2408.05148/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":"2408.05148","created_at":"2026-07-05T09:28:33.823639+00:00"},{"alias_kind":"arxiv_version","alias_value":"2408.05148v3","created_at":"2026-07-05T09:28:33.823639+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2408.05148","created_at":"2026-07-05T09:28:33.823639+00:00"},{"alias_kind":"pith_short_12","alias_value":"BDNCWLH7MR3O","created_at":"2026-07-05T09:28:33.823639+00:00"},{"alias_kind":"pith_short_16","alias_value":"BDNCWLH7MR3OTJQM","created_at":"2026-07-05T09:28:33.823639+00:00"},{"alias_kind":"pith_short_8","alias_value":"BDNCWLH7","created_at":"2026-07-05T09:28:33.823639+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.27934","citing_title":"Self-Verifying Measurement Records: Hash-Linked Evidence Graphs for Hardware Benchmarking","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2604.22411","citing_title":"Introducing Background Temperature to Characterise Hidden Randomness in Large Language Models","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2604.22032","citing_title":"Kernel Contracts: A Specification Language for ML Kernel Correctness Across Heterogeneous Silicon","ref_index":6,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/BDNCWLH7MR3OTJQMLPUHOLDCR5","json":"https://pith.science/pith/BDNCWLH7MR3OTJQMLPUHOLDCR5.json","graph_json":"https://pith.science/api/pith-number/BDNCWLH7MR3OTJQMLPUHOLDCR5/graph.json","events_json":"https://pith.science/api/pith-number/BDNCWLH7MR3OTJQMLPUHOLDCR5/events.json","paper":"https://pith.science/paper/BDNCWLH7"},"agent_actions":{"view_html":"https://pith.science/pith/BDNCWLH7MR3OTJQMLPUHOLDCR5","download_json":"https://pith.science/pith/BDNCWLH7MR3OTJQMLPUHOLDCR5.json","view_paper":"https://pith.science/paper/BDNCWLH7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2408.05148&json=true","fetch_graph":"https://pith.science/api/pith-number/BDNCWLH7MR3OTJQMLPUHOLDCR5/graph.json","fetch_events":"https://pith.science/api/pith-number/BDNCWLH7MR3OTJQMLPUHOLDCR5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BDNCWLH7MR3OTJQMLPUHOLDCR5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BDNCWLH7MR3OTJQMLPUHOLDCR5/action/storage_attestation","attest_author":"https://pith.science/pith/BDNCWLH7MR3OTJQMLPUHOLDCR5/action/author_attestation","sign_citation":"https://pith.science/pith/BDNCWLH7MR3OTJQMLPUHOLDCR5/action/citation_signature","submit_replication":"https://pith.science/pith/BDNCWLH7MR3OTJQMLPUHOLDCR5/action/replication_record"}},"created_at":"2026-07-05T09:28:33.823639+00:00","updated_at":"2026-07-05T09:28:33.823639+00:00"}