{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:5QOGUNYTWWDX5JA6PCUACD4YNX","short_pith_number":"pith:5QOGUNYT","schema_version":"1.0","canonical_sha256":"ec1c6a3713b5877ea41e78a8010f986de7c80b598376a5b272d05252895e70d5","source":{"kind":"arxiv","id":"2405.00263","version":1},"attestation_state":"computed","paper":{"title":"Clover: Regressive Lightweight Speculative Decoding with Sequential Knowledge","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Bin Cui, Bin Xiao, Chunan Shi, Fan Yang, Lei Su, Weipeng Chen, Xiangwei Deng, Xiaonan Nie","submitted_at":"2024-05-01T00:46:22Z","abstract_excerpt":"Large language models (LLMs) suffer from low efficiency as the mismatch between the requirement of auto-regressive decoding and the design of most contemporary GPUs. Specifically, billions to trillions of parameters must be loaded to the GPU cache through its limited memory bandwidth for computation, but only a small batch of tokens is actually computed. Consequently, the GPU spends most of its time on memory transfer instead of computation. Recently, parallel decoding, a type of speculative decoding algorithms, is becoming more popular and has demonstrated impressive efficiency improvement in"},"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":"2405.00263","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-05-01T00:46:22Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"cf477dc32ef2d412c3789b3221901d95307d7664b79bc97e35d2ccf42dd8344a","abstract_canon_sha256":"ad25c1939375e81cc632806a2f4eacaae72eee6b1dc165812502016985e88139"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:14:10.813136Z","signature_b64":"LGMLgSe9//hAg5Mf8GPacyCWgUInYfEO1VLyPep98qSdo+mSMwF0NyMwAEaFKSb5IWu+A2E/L/9LKjbtTmYtCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ec1c6a3713b5877ea41e78a8010f986de7c80b598376a5b272d05252895e70d5","last_reissued_at":"2026-07-05T08:14:10.812645Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:14:10.812645Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Clover: Regressive Lightweight Speculative Decoding with Sequential Knowledge","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Bin Cui, Bin Xiao, Chunan Shi, Fan Yang, Lei Su, Weipeng Chen, Xiangwei Deng, Xiaonan Nie","submitted_at":"2024-05-01T00:46:22Z","abstract_excerpt":"Large language models (LLMs) suffer from low efficiency as the mismatch between the requirement of auto-regressive decoding and the design of most contemporary GPUs. Specifically, billions to trillions of parameters must be loaded to the GPU cache through its limited memory bandwidth for computation, but only a small batch of tokens is actually computed. Consequently, the GPU spends most of its time on memory transfer instead of computation. Recently, parallel decoding, a type of speculative decoding algorithms, is becoming more popular and has demonstrated impressive efficiency improvement in"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.00263","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/2405.00263/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":"2405.00263","created_at":"2026-07-05T08:14:10.812699+00:00"},{"alias_kind":"arxiv_version","alias_value":"2405.00263v1","created_at":"2026-07-05T08:14:10.812699+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.00263","created_at":"2026-07-05T08:14:10.812699+00:00"},{"alias_kind":"pith_short_12","alias_value":"5QOGUNYTWWDX","created_at":"2026-07-05T08:14:10.812699+00:00"},{"alias_kind":"pith_short_16","alias_value":"5QOGUNYTWWDX5JA6","created_at":"2026-07-05T08:14:10.812699+00:00"},{"alias_kind":"pith_short_8","alias_value":"5QOGUNYT","created_at":"2026-07-05T08:14:10.812699+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2507.01449","citing_title":"LogitSpec: Accelerating Retrieval-based Speculative Decoding via Next Next Token Speculation","ref_index":17,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5QOGUNYTWWDX5JA6PCUACD4YNX","json":"https://pith.science/pith/5QOGUNYTWWDX5JA6PCUACD4YNX.json","graph_json":"https://pith.science/api/pith-number/5QOGUNYTWWDX5JA6PCUACD4YNX/graph.json","events_json":"https://pith.science/api/pith-number/5QOGUNYTWWDX5JA6PCUACD4YNX/events.json","paper":"https://pith.science/paper/5QOGUNYT"},"agent_actions":{"view_html":"https://pith.science/pith/5QOGUNYTWWDX5JA6PCUACD4YNX","download_json":"https://pith.science/pith/5QOGUNYTWWDX5JA6PCUACD4YNX.json","view_paper":"https://pith.science/paper/5QOGUNYT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2405.00263&json=true","fetch_graph":"https://pith.science/api/pith-number/5QOGUNYTWWDX5JA6PCUACD4YNX/graph.json","fetch_events":"https://pith.science/api/pith-number/5QOGUNYTWWDX5JA6PCUACD4YNX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5QOGUNYTWWDX5JA6PCUACD4YNX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5QOGUNYTWWDX5JA6PCUACD4YNX/action/storage_attestation","attest_author":"https://pith.science/pith/5QOGUNYTWWDX5JA6PCUACD4YNX/action/author_attestation","sign_citation":"https://pith.science/pith/5QOGUNYTWWDX5JA6PCUACD4YNX/action/citation_signature","submit_replication":"https://pith.science/pith/5QOGUNYTWWDX5JA6PCUACD4YNX/action/replication_record"}},"created_at":"2026-07-05T08:14:10.812699+00:00","updated_at":"2026-07-05T08:14:10.812699+00:00"}