{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:2QAJ4R7UIK5EQ37JN2GIMAWB3Q","short_pith_number":"pith:2QAJ4R7U","schema_version":"1.0","canonical_sha256":"d4009e47f442ba486fe96e8c8602c1dc149bbb12cfeb854be10abd1b0e52a0d7","source":{"kind":"arxiv","id":"2408.04678","version":1},"attestation_state":"computed","paper":{"title":"CREST: Effectively Compacting a Datastore For Retrieval-Based Speculative Decoding","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.DB"],"primary_cat":"cs.CL","authors_text":"Jinsol Park, Patrick Wang, Sophia Ho","submitted_at":"2024-08-08T03:38:49Z","abstract_excerpt":"We present CREST (Compact Retrieval-Based Speculative Decoding), a redesign of REST that allows it to be effectively \"compacted\". REST is a drafting technique for speculative decoding based on retrieving exact n-gram matches of the most recent n tokens generated by the target LLM from a datastore. The key idea of CREST is to only store a subset of the smallest and most common n-grams in the datastore with the hope of achieving comparable performance with less storage space. We found that storing a subset of n-grams both reduces storage space and improves performance. CREST matches REST's accep"},"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":"2408.04678","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-08-08T03:38:49Z","cross_cats_sorted":["cs.AI","cs.DB"],"title_canon_sha256":"118e4d8ff9c252db8556feeca1a86cb60dae7807855623c594062e76471bd0c7","abstract_canon_sha256":"18af9a1d86b36f38d413746e665edd3b536823d1159173e394c9caf06b9ada20"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:53:42.326188Z","signature_b64":"RcXAFj8Rhbhh3k0KXuNAp7JaCnpD+780O0zTEdy3fpFqsErCuREhL5a6PxH/SIqtVeOiZZvw7JKpjdCOR91xAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d4009e47f442ba486fe96e8c8602c1dc149bbb12cfeb854be10abd1b0e52a0d7","last_reissued_at":"2026-07-05T08:53:42.325736Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:53:42.325736Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"CREST: Effectively Compacting a Datastore For Retrieval-Based Speculative Decoding","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.DB"],"primary_cat":"cs.CL","authors_text":"Jinsol Park, Patrick Wang, Sophia Ho","submitted_at":"2024-08-08T03:38:49Z","abstract_excerpt":"We present CREST (Compact Retrieval-Based Speculative Decoding), a redesign of REST that allows it to be effectively \"compacted\". REST is a drafting technique for speculative decoding based on retrieving exact n-gram matches of the most recent n tokens generated by the target LLM from a datastore. The key idea of CREST is to only store a subset of the smallest and most common n-grams in the datastore with the hope of achieving comparable performance with less storage space. We found that storing a subset of n-grams both reduces storage space and improves performance. CREST matches REST's accep"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2408.04678","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/2408.04678/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":"2408.04678","created_at":"2026-07-05T08:53:42.325793+00:00"},{"alias_kind":"arxiv_version","alias_value":"2408.04678v1","created_at":"2026-07-05T08:53:42.325793+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2408.04678","created_at":"2026-07-05T08:53:42.325793+00:00"},{"alias_kind":"pith_short_12","alias_value":"2QAJ4R7UIK5E","created_at":"2026-07-05T08:53:42.325793+00:00"},{"alias_kind":"pith_short_16","alias_value":"2QAJ4R7UIK5EQ37J","created_at":"2026-07-05T08:53:42.325793+00:00"},{"alias_kind":"pith_short_8","alias_value":"2QAJ4R7U","created_at":"2026-07-05T08:53:42.325793+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2608.03839","citing_title":"Oilbird: Training-Free Speculative Decoding with Keys the Verifier Already Computes","ref_index":2025,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2QAJ4R7UIK5EQ37JN2GIMAWB3Q","json":"https://pith.science/pith/2QAJ4R7UIK5EQ37JN2GIMAWB3Q.json","graph_json":"https://pith.science/api/pith-number/2QAJ4R7UIK5EQ37JN2GIMAWB3Q/graph.json","events_json":"https://pith.science/api/pith-number/2QAJ4R7UIK5EQ37JN2GIMAWB3Q/events.json","paper":"https://pith.science/paper/2QAJ4R7U"},"agent_actions":{"view_html":"https://pith.science/pith/2QAJ4R7UIK5EQ37JN2GIMAWB3Q","download_json":"https://pith.science/pith/2QAJ4R7UIK5EQ37JN2GIMAWB3Q.json","view_paper":"https://pith.science/paper/2QAJ4R7U","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2408.04678&json=true","fetch_graph":"https://pith.science/api/pith-number/2QAJ4R7UIK5EQ37JN2GIMAWB3Q/graph.json","fetch_events":"https://pith.science/api/pith-number/2QAJ4R7UIK5EQ37JN2GIMAWB3Q/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2QAJ4R7UIK5EQ37JN2GIMAWB3Q/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2QAJ4R7UIK5EQ37JN2GIMAWB3Q/action/storage_attestation","attest_author":"https://pith.science/pith/2QAJ4R7UIK5EQ37JN2GIMAWB3Q/action/author_attestation","sign_citation":"https://pith.science/pith/2QAJ4R7UIK5EQ37JN2GIMAWB3Q/action/citation_signature","submit_replication":"https://pith.science/pith/2QAJ4R7UIK5EQ37JN2GIMAWB3Q/action/replication_record"}},"created_at":"2026-07-05T08:53:42.325793+00:00","updated_at":"2026-07-05T08:53:42.325793+00:00"}