{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:BWQXTQUKLJXKI2G5AXG7XACYSV","short_pith_number":"pith:BWQXTQUK","schema_version":"1.0","canonical_sha256":"0da179c28a5a6ea468dd05cdfb805895673b4721aaafb9e1f52e1e14bf317033","source":{"kind":"arxiv","id":"2402.12374","version":3},"attestation_state":"computed","paper":{"title":"Sequoia: Scalable, Robust, and Hardware-aware Speculative Decoding","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Avner May, Beidi Chen, Max Ryabinin, Ruslan Svirschevski, Yuhsun Huang, Zhihao Jia, Zhuoming Chen","submitted_at":"2024-02-19T18:58:32Z","abstract_excerpt":"As the usage of large language models (LLMs) grows, performing efficient inference with these models becomes increasingly important. While speculative decoding has recently emerged as a promising direction for speeding up inference, existing methods are limited in their ability to scale to larger speculation budgets, and adapt to different hyperparameters and hardware. This paper introduces Sequoia, a scalable, robust, and hardware-aware algorithm for speculative decoding. To attain better scalability, Sequoia introduces a dynamic programming algorithm to find the optimal tree structure for th"},"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":"2402.12374","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-02-19T18:58:32Z","cross_cats_sorted":[],"title_canon_sha256":"d3be89ed34b75284dde176390ff4f403336b0d71a224e541c8ef412c1e48fb5f","abstract_canon_sha256":"4ed9a79f2cc190efc397146329cbe39b7301c2acf20dfa9aaf2aaf1f37cb6563"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:32:14.504712Z","signature_b64":"eeaj23zPkjFlmFmllLhJN4/p+KJhEZnd9pVag8PRvR5b136I0ZqEvfKxB1c+alH5KOBLXo54dTKzp4NFYbOaDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0da179c28a5a6ea468dd05cdfb805895673b4721aaafb9e1f52e1e14bf317033","last_reissued_at":"2026-07-05T11:32:14.504214Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:32:14.504214Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Sequoia: Scalable, Robust, and Hardware-aware Speculative Decoding","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Avner May, Beidi Chen, Max Ryabinin, Ruslan Svirschevski, Yuhsun Huang, Zhihao Jia, Zhuoming Chen","submitted_at":"2024-02-19T18:58:32Z","abstract_excerpt":"As the usage of large language models (LLMs) grows, performing efficient inference with these models becomes increasingly important. While speculative decoding has recently emerged as a promising direction for speeding up inference, existing methods are limited in their ability to scale to larger speculation budgets, and adapt to different hyperparameters and hardware. This paper introduces Sequoia, a scalable, robust, and hardware-aware algorithm for speculative decoding. To attain better scalability, Sequoia introduces a dynamic programming algorithm to find the optimal tree structure for th"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.12374","kind":"arxiv","version":3},"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/2402.12374/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":"2402.12374","created_at":"2026-07-05T11:32:14.504268+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.12374v3","created_at":"2026-07-05T11:32:14.504268+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.12374","created_at":"2026-07-05T11:32:14.504268+00:00"},{"alias_kind":"pith_short_12","alias_value":"BWQXTQUKLJXK","created_at":"2026-07-05T11:32:14.504268+00:00"},{"alias_kind":"pith_short_16","alias_value":"BWQXTQUKLJXKI2G5","created_at":"2026-07-05T11:32:14.504268+00:00"},{"alias_kind":"pith_short_8","alias_value":"BWQXTQUK","created_at":"2026-07-05T11:32:14.504268+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":11,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.06763","citing_title":"Trees from Marginals: Autoregressive drafting with factorized priors","ref_index":28,"is_internal_anchor":true},{"citing_arxiv_id":"2606.25097","citing_title":"Speculative Decoding at Temperature Zero: A Scoped Safety-Invariance Screen with a 48,072-Sample Expansion","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2606.24957","citing_title":"Dustin: Draft-Augmented Sparse Verification for Efficient Long-Context Generation with Speculative Decoding","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2606.12303","citing_title":"From 2D Grids to 1D Tokens: Reforming Shared Representations for Multimodal Image Fusion","ref_index":48,"is_internal_anchor":false},{"citing_arxiv_id":"2606.04446","citing_title":"D^2SD: Accelerating Speculative Decoding with Dual Diffusion Draft Models","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2605.29727","citing_title":"Bastion: Budget-Aware Speculative Decoding with Tree-structured Block Diffusion Drafting","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00535","citing_title":"DREAM-S: Speculative Decoding with Searchable Drafting and Target-Aware Refinement for Multimodal Generation","ref_index":98,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14978","citing_title":"Performance-Driven Policy Optimization for Speculative Decoding with Adaptive Windowing","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12001","citing_title":"CR^2: Cost-Aware Risk-Controlled Routing for Wireless Device-Edge LLM Inference","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2604.26412","citing_title":"When Hidden States Drift: Can KV Caches Rescue Long-Range Speculative Decoding?","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2604.26412","citing_title":"When Hidden States Drift: Can KV Caches Rescue Long-Range Speculative Decoding?","ref_index":3,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/BWQXTQUKLJXKI2G5AXG7XACYSV","json":"https://pith.science/pith/BWQXTQUKLJXKI2G5AXG7XACYSV.json","graph_json":"https://pith.science/api/pith-number/BWQXTQUKLJXKI2G5AXG7XACYSV/graph.json","events_json":"https://pith.science/api/pith-number/BWQXTQUKLJXKI2G5AXG7XACYSV/events.json","paper":"https://pith.science/paper/BWQXTQUK"},"agent_actions":{"view_html":"https://pith.science/pith/BWQXTQUKLJXKI2G5AXG7XACYSV","download_json":"https://pith.science/pith/BWQXTQUKLJXKI2G5AXG7XACYSV.json","view_paper":"https://pith.science/paper/BWQXTQUK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.12374&json=true","fetch_graph":"https://pith.science/api/pith-number/BWQXTQUKLJXKI2G5AXG7XACYSV/graph.json","fetch_events":"https://pith.science/api/pith-number/BWQXTQUKLJXKI2G5AXG7XACYSV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BWQXTQUKLJXKI2G5AXG7XACYSV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BWQXTQUKLJXKI2G5AXG7XACYSV/action/storage_attestation","attest_author":"https://pith.science/pith/BWQXTQUKLJXKI2G5AXG7XACYSV/action/author_attestation","sign_citation":"https://pith.science/pith/BWQXTQUKLJXKI2G5AXG7XACYSV/action/citation_signature","submit_replication":"https://pith.science/pith/BWQXTQUKLJXKI2G5AXG7XACYSV/action/replication_record"}},"created_at":"2026-07-05T11:32:14.504268+00:00","updated_at":"2026-07-05T11:32:14.504268+00:00"}