{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:XONKV7Z56EQ27SVBIVUAFPIRR4","short_pith_number":"pith:XONKV7Z5","schema_version":"1.0","canonical_sha256":"bb9aaaff3df121afcaa1456802bd118f30199f0de6bcb68f4fa47c724c49fb52","source":{"kind":"arxiv","id":"2401.06706","version":1},"attestation_state":"computed","paper":{"title":"Multi-Candidate Speculative Decoding","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Jiajun Chen, Sen Yang, Shujian Huang, Xinyu Dai","submitted_at":"2024-01-12T17:15:23Z","abstract_excerpt":"Large language models have shown impressive capabilities across a variety of NLP tasks, yet their generating text autoregressively is time-consuming. One way to speed them up is speculative decoding, which generates candidate segments (a sequence of tokens) from a fast draft model that is then verified in parallel by the target model. However, the acceptance rate of candidate tokens receives limitations from several factors, such as the model, the dataset, and the decoding setup. This paper proposes sampling multiple candidates from a draft model and then organising them in batches for verific"},"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":"2401.06706","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-01-12T17:15:23Z","cross_cats_sorted":[],"title_canon_sha256":"73311b65c0600d513aa371e2eecb063cad6136648f68632f72fcc31f77c2ddf6","abstract_canon_sha256":"b8614e77788cf594a647a35a3c2643d2aed3295d0d83e8628f4387fba135facf"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:33:04.268167Z","signature_b64":"qWXOD3PNgkUUfQvqs83mgK1R+UGr6Str4qbRH8mhsYVVrXXK7ULcn3JLh3UmX3l5xdQRffmREAiCrPtLpIKxBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bb9aaaff3df121afcaa1456802bd118f30199f0de6bcb68f4fa47c724c49fb52","last_reissued_at":"2026-07-05T07:33:04.267702Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:33:04.267702Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Multi-Candidate Speculative Decoding","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Jiajun Chen, Sen Yang, Shujian Huang, Xinyu Dai","submitted_at":"2024-01-12T17:15:23Z","abstract_excerpt":"Large language models have shown impressive capabilities across a variety of NLP tasks, yet their generating text autoregressively is time-consuming. One way to speed them up is speculative decoding, which generates candidate segments (a sequence of tokens) from a fast draft model that is then verified in parallel by the target model. However, the acceptance rate of candidate tokens receives limitations from several factors, such as the model, the dataset, and the decoding setup. This paper proposes sampling multiple candidates from a draft model and then organising them in batches for verific"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2401.06706","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/2401.06706/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":"2401.06706","created_at":"2026-07-05T07:33:04.267762+00:00"},{"alias_kind":"arxiv_version","alias_value":"2401.06706v1","created_at":"2026-07-05T07:33:04.267762+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2401.06706","created_at":"2026-07-05T07:33:04.267762+00:00"},{"alias_kind":"pith_short_12","alias_value":"XONKV7Z56EQ2","created_at":"2026-07-05T07:33:04.267762+00:00"},{"alias_kind":"pith_short_16","alias_value":"XONKV7Z56EQ27SVB","created_at":"2026-07-05T07:33:04.267762+00:00"},{"alias_kind":"pith_short_8","alias_value":"XONKV7Z5","created_at":"2026-07-05T07:33:04.267762+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.30265","citing_title":"When Is a Draft Accepted? A Theory of Acceptance in Speculative Decoding","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2604.25925","citing_title":"SpecTr-GBV: Multi-Draft Block Verification Accelerating Speculative Decoding","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04543","citing_title":"UniVer: A Unified Perspective for Multi-step and Multi-draft Speculative Decoding","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2604.05417","citing_title":"Multi-Drafter Speculative Decoding with Alignment Feedback","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18170","citing_title":"Copy-as-Decode: Grammar-Constrained Parallel Prefill for LLM Editing","ref_index":29,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/XONKV7Z56EQ27SVBIVUAFPIRR4","json":"https://pith.science/pith/XONKV7Z56EQ27SVBIVUAFPIRR4.json","graph_json":"https://pith.science/api/pith-number/XONKV7Z56EQ27SVBIVUAFPIRR4/graph.json","events_json":"https://pith.science/api/pith-number/XONKV7Z56EQ27SVBIVUAFPIRR4/events.json","paper":"https://pith.science/paper/XONKV7Z5"},"agent_actions":{"view_html":"https://pith.science/pith/XONKV7Z56EQ27SVBIVUAFPIRR4","download_json":"https://pith.science/pith/XONKV7Z56EQ27SVBIVUAFPIRR4.json","view_paper":"https://pith.science/paper/XONKV7Z5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2401.06706&json=true","fetch_graph":"https://pith.science/api/pith-number/XONKV7Z56EQ27SVBIVUAFPIRR4/graph.json","fetch_events":"https://pith.science/api/pith-number/XONKV7Z56EQ27SVBIVUAFPIRR4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XONKV7Z56EQ27SVBIVUAFPIRR4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XONKV7Z56EQ27SVBIVUAFPIRR4/action/storage_attestation","attest_author":"https://pith.science/pith/XONKV7Z56EQ27SVBIVUAFPIRR4/action/author_attestation","sign_citation":"https://pith.science/pith/XONKV7Z56EQ27SVBIVUAFPIRR4/action/citation_signature","submit_replication":"https://pith.science/pith/XONKV7Z56EQ27SVBIVUAFPIRR4/action/replication_record"}},"created_at":"2026-07-05T07:33:04.267762+00:00","updated_at":"2026-07-05T07:33:04.267762+00:00"}