{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:GA2G4BRIMYLHIWRPQU4EIJSMP2","short_pith_number":"pith:GA2G4BRI","schema_version":"1.0","canonical_sha256":"30346e06286616745a2f853844264c7e960358296767fcf8921478c7242d3ab2","source":{"kind":"arxiv","id":"2011.02999","version":1},"attestation_state":"computed","paper":{"title":"CPR: Understanding and Improving Failure Tolerant Training for Deep Learning Recommendation with Partial Recovery","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.DC"],"primary_cat":"cs.LG","authors_text":"Bor-Yiing Su, Brandon Lucia, Carole-Jean Wu, Caroline Trippel, Isabel Gao, Jiyan Yang, Kiwan Maeng, Mark C. Jeffrey, Mike Rabbat, Shivam Bharuka, Vikram Saraph","submitted_at":"2020-11-05T17:54:35Z","abstract_excerpt":"The paper proposes and optimizes a partial recovery training system, CPR, for recommendation models. CPR relaxes the consistency requirement by enabling non-failed nodes to proceed without loading checkpoints when a node fails during training, improving failure-related overheads. The paper is the first to the extent of our knowledge to perform a data-driven, in-depth analysis of applying partial recovery to recommendation models and identified a trade-off between accuracy and performance. Motivated by the analysis, we present CPR, a partial recovery training system that can reduce the training"},"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":"2011.02999","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2020-11-05T17:54:35Z","cross_cats_sorted":["cs.DC"],"title_canon_sha256":"6d4da3a68b649870199e4461d7e7b64364f818becd2b486a937ca784a603e8df","abstract_canon_sha256":"6bc362f3d4d43eee9610e8ea5f8983387a56c4789eb3a24cac8e05449ec53fe2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:49:33.062965Z","signature_b64":"S8IUUTgZW10peou45alRk/7wEsf8Bx2tL4deZ9gR6FAhEGPolTxGSV5nTFGi5H0swF9JfblHmvglUJftgWwcAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"30346e06286616745a2f853844264c7e960358296767fcf8921478c7242d3ab2","last_reissued_at":"2026-07-05T01:49:33.062511Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:49:33.062511Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"CPR: Understanding and Improving Failure Tolerant Training for Deep Learning Recommendation with Partial Recovery","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.DC"],"primary_cat":"cs.LG","authors_text":"Bor-Yiing Su, Brandon Lucia, Carole-Jean Wu, Caroline Trippel, Isabel Gao, Jiyan Yang, Kiwan Maeng, Mark C. Jeffrey, Mike Rabbat, Shivam Bharuka, Vikram Saraph","submitted_at":"2020-11-05T17:54:35Z","abstract_excerpt":"The paper proposes and optimizes a partial recovery training system, CPR, for recommendation models. CPR relaxes the consistency requirement by enabling non-failed nodes to proceed without loading checkpoints when a node fails during training, improving failure-related overheads. The paper is the first to the extent of our knowledge to perform a data-driven, in-depth analysis of applying partial recovery to recommendation models and identified a trade-off between accuracy and performance. Motivated by the analysis, we present CPR, a partial recovery training system that can reduce the training"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2011.02999","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":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2011.02999/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":"2011.02999","created_at":"2026-07-05T01:49:33.062569+00:00"},{"alias_kind":"arxiv_version","alias_value":"2011.02999v1","created_at":"2026-07-05T01:49:33.062569+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2011.02999","created_at":"2026-07-05T01:49:33.062569+00:00"},{"alias_kind":"pith_short_12","alias_value":"GA2G4BRIMYLH","created_at":"2026-07-05T01:49:33.062569+00:00"},{"alias_kind":"pith_short_16","alias_value":"GA2G4BRIMYLHIWRP","created_at":"2026-07-05T01:49:33.062569+00:00"},{"alias_kind":"pith_short_8","alias_value":"GA2G4BRI","created_at":"2026-07-05T01:49:33.062569+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2607.01646","citing_title":"DeadPool: Resilient LLM Training with Hot-Swapping via Zero-Overhead Checkpoint","ref_index":45,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GA2G4BRIMYLHIWRPQU4EIJSMP2","json":"https://pith.science/pith/GA2G4BRIMYLHIWRPQU4EIJSMP2.json","graph_json":"https://pith.science/api/pith-number/GA2G4BRIMYLHIWRPQU4EIJSMP2/graph.json","events_json":"https://pith.science/api/pith-number/GA2G4BRIMYLHIWRPQU4EIJSMP2/events.json","paper":"https://pith.science/paper/GA2G4BRI"},"agent_actions":{"view_html":"https://pith.science/pith/GA2G4BRIMYLHIWRPQU4EIJSMP2","download_json":"https://pith.science/pith/GA2G4BRIMYLHIWRPQU4EIJSMP2.json","view_paper":"https://pith.science/paper/GA2G4BRI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2011.02999&json=true","fetch_graph":"https://pith.science/api/pith-number/GA2G4BRIMYLHIWRPQU4EIJSMP2/graph.json","fetch_events":"https://pith.science/api/pith-number/GA2G4BRIMYLHIWRPQU4EIJSMP2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GA2G4BRIMYLHIWRPQU4EIJSMP2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GA2G4BRIMYLHIWRPQU4EIJSMP2/action/storage_attestation","attest_author":"https://pith.science/pith/GA2G4BRIMYLHIWRPQU4EIJSMP2/action/author_attestation","sign_citation":"https://pith.science/pith/GA2G4BRIMYLHIWRPQU4EIJSMP2/action/citation_signature","submit_replication":"https://pith.science/pith/GA2G4BRIMYLHIWRPQU4EIJSMP2/action/replication_record"}},"created_at":"2026-07-05T01:49:33.062569+00:00","updated_at":"2026-07-05T01:49:33.062569+00:00"}