{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:XB7G7DWP4JCZVUBBEGFGVC6X3O","short_pith_number":"pith:XB7G7DWP","schema_version":"1.0","canonical_sha256":"b87e6f8ecfe2459ad021218a6a8bd7db9e7cd7ded1d18e8fe63a41b87036a0fb","source":{"kind":"arxiv","id":"2411.11767","version":2},"attestation_state":"computed","paper":{"title":"Drowning in Documents: Consequences of Scaling Reranker Inference","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.IR","authors_text":"Andrew Drozdov, Erik Lindgren, Matei Zaharia, Mathew Jacob, Michael Carbin, Omar Khattab","submitted_at":"2024-11-18T17:46:32Z","abstract_excerpt":"Rerankers, typically cross-encoders, are computationally intensive but are frequently used because they are widely assumed to outperform cheaper initial IR systems. We challenge this assumption by measuring reranker performance for full retrieval, not just re-scoring first-stage retrieval. To provide a more robust evaluation, we prioritize strong first-stage retrieval using modern dense embeddings and test rerankers on a variety of carefully chosen, challenging tasks, including internally curated datasets to avoid contamination, and out-of-domain ones. Our empirical results reveal a surprising"},"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":"2411.11767","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.IR","submitted_at":"2024-11-18T17:46:32Z","cross_cats_sorted":["cs.CL","cs.LG"],"title_canon_sha256":"3339a28902021ea675698619bf2db2f8c90b69107f4cd0a6da54bc7f7219f0c7","abstract_canon_sha256":"265119d0b107b15a5013dbef264a08eaac6380fdd2b6533a71695e2741e60c06"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:35:30.858875Z","signature_b64":"dB181c4uQTd9jwmy7qM9X+qZaS0TMVs5wKfgmiESNU59lD0JQNY8HYAi0S1wI1M/WWtT6PvndIGXIkv79IoSBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b87e6f8ecfe2459ad021218a6a8bd7db9e7cd7ded1d18e8fe63a41b87036a0fb","last_reissued_at":"2026-07-05T11:35:30.858413Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:35:30.858413Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Drowning in Documents: Consequences of Scaling Reranker Inference","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.IR","authors_text":"Andrew Drozdov, Erik Lindgren, Matei Zaharia, Mathew Jacob, Michael Carbin, Omar Khattab","submitted_at":"2024-11-18T17:46:32Z","abstract_excerpt":"Rerankers, typically cross-encoders, are computationally intensive but are frequently used because they are widely assumed to outperform cheaper initial IR systems. We challenge this assumption by measuring reranker performance for full retrieval, not just re-scoring first-stage retrieval. To provide a more robust evaluation, we prioritize strong first-stage retrieval using modern dense embeddings and test rerankers on a variety of carefully chosen, challenging tasks, including internally curated datasets to avoid contamination, and out-of-domain ones. Our empirical results reveal a surprising"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2411.11767","kind":"arxiv","version":2},"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/2411.11767/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":"2411.11767","created_at":"2026-07-05T11:35:30.858467+00:00"},{"alias_kind":"arxiv_version","alias_value":"2411.11767v2","created_at":"2026-07-05T11:35:30.858467+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2411.11767","created_at":"2026-07-05T11:35:30.858467+00:00"},{"alias_kind":"pith_short_12","alias_value":"XB7G7DWP4JCZ","created_at":"2026-07-05T11:35:30.858467+00:00"},{"alias_kind":"pith_short_16","alias_value":"XB7G7DWP4JCZVUBB","created_at":"2026-07-05T11:35:30.858467+00:00"},{"alias_kind":"pith_short_8","alias_value":"XB7G7DWP","created_at":"2026-07-05T11:35:30.858467+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2607.01538","citing_title":"Can Language Models Actually Retrieve In-Context? Drowning in Documents at Million Token Scale","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2604.27577","citing_title":"Reproducing Adaptive Reranking for Reasoning-Intensive IR","ref_index":13,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/XB7G7DWP4JCZVUBBEGFGVC6X3O","json":"https://pith.science/pith/XB7G7DWP4JCZVUBBEGFGVC6X3O.json","graph_json":"https://pith.science/api/pith-number/XB7G7DWP4JCZVUBBEGFGVC6X3O/graph.json","events_json":"https://pith.science/api/pith-number/XB7G7DWP4JCZVUBBEGFGVC6X3O/events.json","paper":"https://pith.science/paper/XB7G7DWP"},"agent_actions":{"view_html":"https://pith.science/pith/XB7G7DWP4JCZVUBBEGFGVC6X3O","download_json":"https://pith.science/pith/XB7G7DWP4JCZVUBBEGFGVC6X3O.json","view_paper":"https://pith.science/paper/XB7G7DWP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2411.11767&json=true","fetch_graph":"https://pith.science/api/pith-number/XB7G7DWP4JCZVUBBEGFGVC6X3O/graph.json","fetch_events":"https://pith.science/api/pith-number/XB7G7DWP4JCZVUBBEGFGVC6X3O/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XB7G7DWP4JCZVUBBEGFGVC6X3O/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XB7G7DWP4JCZVUBBEGFGVC6X3O/action/storage_attestation","attest_author":"https://pith.science/pith/XB7G7DWP4JCZVUBBEGFGVC6X3O/action/author_attestation","sign_citation":"https://pith.science/pith/XB7G7DWP4JCZVUBBEGFGVC6X3O/action/citation_signature","submit_replication":"https://pith.science/pith/XB7G7DWP4JCZVUBBEGFGVC6X3O/action/replication_record"}},"created_at":"2026-07-05T11:35:30.858467+00:00","updated_at":"2026-07-05T11:35:30.858467+00:00"}