{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:WF7VBPRV6KQ4N7FLAT73KCJERL","short_pith_number":"pith:WF7VBPRV","schema_version":"1.0","canonical_sha256":"b17f50be35f2a1c6fcab04ffb509248ac23aedbc73caa76e8911ed969c1e4cf1","source":{"kind":"arxiv","id":"2404.15778","version":2},"attestation_state":"computed","paper":{"title":"BASS: Batched Attention-optimized Speculative Sampling","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Anoop Deoras, Haifeng Qian, Mingyue Shang, Ramesh Nallapati, Sanjay Krishna Gouda, Sudipta Sengupta, Sujan Kumar Gonugondla, Sungsoo Ha, Xiaofei Ma","submitted_at":"2024-04-24T09:57:11Z","abstract_excerpt":"Speculative decoding has emerged as a powerful method to improve latency and throughput in hosting large language models. However, most existing implementations focus on generating a single sequence. Real-world generative AI applications often require multiple responses and how to perform speculative decoding in a batched setting while preserving its latency benefits poses non-trivial challenges. This paper describes a system of batched speculative decoding that sets a new state of the art in multi-sequence generation latency and that demonstrates superior GPU utilization as well as quality of"},"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":"2404.15778","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-04-24T09:57:11Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"a565980bd3b0b44afde7d7ea45f99e6de2b7e3860ccc2fa210f9991e0e6bf68c","abstract_canon_sha256":"1d94efda89e460e2c8484ab7e57551c5eeefc29ade7ef8994ddf90f123a6b52a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:37:06.877584Z","signature_b64":"aAWX49LMumI2nnHzwmroJUXt97yGLCIZJ8zA1gXzFwOIHDaj3slhWFTMKb9gv6AmGhgMKxrOlB+M+U5qIa+pDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b17f50be35f2a1c6fcab04ffb509248ac23aedbc73caa76e8911ed969c1e4cf1","last_reissued_at":"2026-07-05T08:37:06.877094Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:37:06.877094Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"BASS: Batched Attention-optimized Speculative Sampling","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Anoop Deoras, Haifeng Qian, Mingyue Shang, Ramesh Nallapati, Sanjay Krishna Gouda, Sudipta Sengupta, Sujan Kumar Gonugondla, Sungsoo Ha, Xiaofei Ma","submitted_at":"2024-04-24T09:57:11Z","abstract_excerpt":"Speculative decoding has emerged as a powerful method to improve latency and throughput in hosting large language models. However, most existing implementations focus on generating a single sequence. Real-world generative AI applications often require multiple responses and how to perform speculative decoding in a batched setting while preserving its latency benefits poses non-trivial challenges. This paper describes a system of batched speculative decoding that sets a new state of the art in multi-sequence generation latency and that demonstrates superior GPU utilization as well as quality of"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2404.15778","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/2404.15778/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":"2404.15778","created_at":"2026-07-05T08:37:06.877150+00:00"},{"alias_kind":"arxiv_version","alias_value":"2404.15778v2","created_at":"2026-07-05T08:37:06.877150+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2404.15778","created_at":"2026-07-05T08:37:06.877150+00:00"},{"alias_kind":"pith_short_12","alias_value":"WF7VBPRV6KQ4","created_at":"2026-07-05T08:37:06.877150+00:00"},{"alias_kind":"pith_short_16","alias_value":"WF7VBPRV6KQ4N7FL","created_at":"2026-07-05T08:37:06.877150+00:00"},{"alias_kind":"pith_short_8","alias_value":"WF7VBPRV","created_at":"2026-07-05T08:37:06.877150+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2411.13157","citing_title":"Closer Look at Efficient Inference Methods: A Survey of Speculative Decoding","ref_index":27,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/WF7VBPRV6KQ4N7FLAT73KCJERL","json":"https://pith.science/pith/WF7VBPRV6KQ4N7FLAT73KCJERL.json","graph_json":"https://pith.science/api/pith-number/WF7VBPRV6KQ4N7FLAT73KCJERL/graph.json","events_json":"https://pith.science/api/pith-number/WF7VBPRV6KQ4N7FLAT73KCJERL/events.json","paper":"https://pith.science/paper/WF7VBPRV"},"agent_actions":{"view_html":"https://pith.science/pith/WF7VBPRV6KQ4N7FLAT73KCJERL","download_json":"https://pith.science/pith/WF7VBPRV6KQ4N7FLAT73KCJERL.json","view_paper":"https://pith.science/paper/WF7VBPRV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2404.15778&json=true","fetch_graph":"https://pith.science/api/pith-number/WF7VBPRV6KQ4N7FLAT73KCJERL/graph.json","fetch_events":"https://pith.science/api/pith-number/WF7VBPRV6KQ4N7FLAT73KCJERL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WF7VBPRV6KQ4N7FLAT73KCJERL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WF7VBPRV6KQ4N7FLAT73KCJERL/action/storage_attestation","attest_author":"https://pith.science/pith/WF7VBPRV6KQ4N7FLAT73KCJERL/action/author_attestation","sign_citation":"https://pith.science/pith/WF7VBPRV6KQ4N7FLAT73KCJERL/action/citation_signature","submit_replication":"https://pith.science/pith/WF7VBPRV6KQ4N7FLAT73KCJERL/action/replication_record"}},"created_at":"2026-07-05T08:37:06.877150+00:00","updated_at":"2026-07-05T08:37:06.877150+00:00"}