{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:X3ATP4FXYSB2FH6A76UB7FJUKD","short_pith_number":"pith:X3ATP4FX","schema_version":"1.0","canonical_sha256":"bec137f0b7c483a29fc0ffa81f953450c65a55eb11929e2bb13d31390fca8039","source":{"kind":"arxiv","id":"2004.14255","version":2},"attestation_state":"computed","paper":{"title":"Efficient Document Re-Ranking for Transformers by Precomputing Term Representations","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.IR","authors_text":"Franco Maria Nardini, Nazli Goharian, Nicola Tonellotto, Ophir Frieder, Raffaele Perego, Sean MacAvaney","submitted_at":"2020-04-29T15:04:22Z","abstract_excerpt":"Deep pretrained transformer networks are effective at various ranking tasks, such as question answering and ad-hoc document ranking. However, their computational expenses deem them cost-prohibitive in practice. Our proposed approach, called PreTTR (Precomputing Transformer Term Representations), considerably reduces the query-time latency of deep transformer networks (up to a 42x speedup on web document ranking) making these networks more practical to use in a real-time ranking scenario. Specifically, we precompute part of the document term representations at indexing time (without a query), a"},"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":"2004.14255","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.IR","submitted_at":"2020-04-29T15:04:22Z","cross_cats_sorted":[],"title_canon_sha256":"bfe96ce645a5ecb3e3838279b0a74f76fd041746eaf77ec7ac3daaf5b2d20a49","abstract_canon_sha256":"6372c250c0a264141e95054cec8f696ba94a09ab54977ee67a4922fb149aaade"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:05:49.504565Z","signature_b64":"vVGT2ljxKV/+a9V07XXlDzwKF0YVTZ40DTFPW/CsoCZVldA5yKVhpzpxHkN7Mh8GDGKxlvaHIDMRQJOfZ92kDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bec137f0b7c483a29fc0ffa81f953450c65a55eb11929e2bb13d31390fca8039","last_reissued_at":"2026-07-05T01:05:49.504100Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:05:49.504100Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Efficient Document Re-Ranking for Transformers by Precomputing Term Representations","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.IR","authors_text":"Franco Maria Nardini, Nazli Goharian, Nicola Tonellotto, Ophir Frieder, Raffaele Perego, Sean MacAvaney","submitted_at":"2020-04-29T15:04:22Z","abstract_excerpt":"Deep pretrained transformer networks are effective at various ranking tasks, such as question answering and ad-hoc document ranking. However, their computational expenses deem them cost-prohibitive in practice. Our proposed approach, called PreTTR (Precomputing Transformer Term Representations), considerably reduces the query-time latency of deep transformer networks (up to a 42x speedup on web document ranking) making these networks more practical to use in a real-time ranking scenario. Specifically, we precompute part of the document term representations at indexing time (without a query), a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2004.14255","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/2004.14255/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":"2004.14255","created_at":"2026-07-05T01:05:49.504162+00:00"},{"alias_kind":"arxiv_version","alias_value":"2004.14255v2","created_at":"2026-07-05T01:05:49.504162+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2004.14255","created_at":"2026-07-05T01:05:49.504162+00:00"},{"alias_kind":"pith_short_12","alias_value":"X3ATP4FXYSB2","created_at":"2026-07-05T01:05:49.504162+00:00"},{"alias_kind":"pith_short_16","alias_value":"X3ATP4FXYSB2FH6A","created_at":"2026-07-05T01:05:49.504162+00:00"},{"alias_kind":"pith_short_8","alias_value":"X3ATP4FX","created_at":"2026-07-05T01:05:49.504162+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2507.14946","citing_title":"FullRecall: A Semantic Search-Based Ranking Approach for Maximizing Recall in Patent Retrieval","ref_index":37,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/X3ATP4FXYSB2FH6A76UB7FJUKD","json":"https://pith.science/pith/X3ATP4FXYSB2FH6A76UB7FJUKD.json","graph_json":"https://pith.science/api/pith-number/X3ATP4FXYSB2FH6A76UB7FJUKD/graph.json","events_json":"https://pith.science/api/pith-number/X3ATP4FXYSB2FH6A76UB7FJUKD/events.json","paper":"https://pith.science/paper/X3ATP4FX"},"agent_actions":{"view_html":"https://pith.science/pith/X3ATP4FXYSB2FH6A76UB7FJUKD","download_json":"https://pith.science/pith/X3ATP4FXYSB2FH6A76UB7FJUKD.json","view_paper":"https://pith.science/paper/X3ATP4FX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2004.14255&json=true","fetch_graph":"https://pith.science/api/pith-number/X3ATP4FXYSB2FH6A76UB7FJUKD/graph.json","fetch_events":"https://pith.science/api/pith-number/X3ATP4FXYSB2FH6A76UB7FJUKD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/X3ATP4FXYSB2FH6A76UB7FJUKD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/X3ATP4FXYSB2FH6A76UB7FJUKD/action/storage_attestation","attest_author":"https://pith.science/pith/X3ATP4FXYSB2FH6A76UB7FJUKD/action/author_attestation","sign_citation":"https://pith.science/pith/X3ATP4FXYSB2FH6A76UB7FJUKD/action/citation_signature","submit_replication":"https://pith.science/pith/X3ATP4FXYSB2FH6A76UB7FJUKD/action/replication_record"}},"created_at":"2026-07-05T01:05:49.504162+00:00","updated_at":"2026-07-05T01:05:49.504162+00:00"}