{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:YXPVHQEF4AG7JDGPKQDFOU3UA5","short_pith_number":"pith:YXPVHQEF","schema_version":"1.0","canonical_sha256":"c5df53c085e00df48ccf54065753740750903eda2dbf94c1af3d8f565a9b1ee4","source":{"kind":"arxiv","id":"2410.05076","version":1},"attestation_state":"computed","paper":{"title":"TidalDecode: Fast and Accurate LLM Decoding with Position Persistent Sparse Attention","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Lijie Yang, Zhihao Jia, Zhihao Zhang, Zhuofu Chen, Zikun Li","submitted_at":"2024-10-07T14:30:27Z","abstract_excerpt":"Large language models (LLMs) have driven significant advancements across diverse NLP tasks, with long-context models gaining prominence for handling extended inputs. However, the expanding key-value (KV) cache size required by Transformer architectures intensifies the memory constraints, particularly during the decoding phase, creating a significant bottleneck. Existing sparse attention mechanisms designed to address this bottleneck have two limitations: (1) they often fail to reliably identify the most relevant tokens for attention, and (2) they overlook the spatial coherence of token selecti"},"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":"2410.05076","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2024-10-07T14:30:27Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"7f66afcf293cc061f9291423b5275bc8419f080ffafd17df996e7e130c6c99ab","abstract_canon_sha256":"fae5754f6aeaa857fc36a2b16a3f8e35af7129fb88b96e7264a11f6805b04d4b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:17:00.989659Z","signature_b64":"wCpuBw+ZNfa1xR8ZKRZVs25NMBaOSbGou/okJz+x80sCSsLklRk12aVclJAmCLif34W3dyUHGYPVWqWmCuPbCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c5df53c085e00df48ccf54065753740750903eda2dbf94c1af3d8f565a9b1ee4","last_reissued_at":"2026-07-05T09:17:00.989165Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:17:00.989165Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TidalDecode: Fast and Accurate LLM Decoding with Position Persistent Sparse Attention","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Lijie Yang, Zhihao Jia, Zhihao Zhang, Zhuofu Chen, Zikun Li","submitted_at":"2024-10-07T14:30:27Z","abstract_excerpt":"Large language models (LLMs) have driven significant advancements across diverse NLP tasks, with long-context models gaining prominence for handling extended inputs. However, the expanding key-value (KV) cache size required by Transformer architectures intensifies the memory constraints, particularly during the decoding phase, creating a significant bottleneck. Existing sparse attention mechanisms designed to address this bottleneck have two limitations: (1) they often fail to reliably identify the most relevant tokens for attention, and (2) they overlook the spatial coherence of token selecti"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.05076","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/2410.05076/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":"2410.05076","created_at":"2026-07-05T09:17:00.989224+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.05076v1","created_at":"2026-07-05T09:17:00.989224+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.05076","created_at":"2026-07-05T09:17:00.989224+00:00"},{"alias_kind":"pith_short_12","alias_value":"YXPVHQEF4AG7","created_at":"2026-07-05T09:17:00.989224+00:00"},{"alias_kind":"pith_short_16","alias_value":"YXPVHQEF4AG7JDGP","created_at":"2026-07-05T09:17:00.989224+00:00"},{"alias_kind":"pith_short_8","alias_value":"YXPVHQEF","created_at":"2026-07-05T09:17:00.989224+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.06467","citing_title":"You Only Index Once: Cross-Layer Sparse Attention with Shared Routing","ref_index":35,"is_internal_anchor":false},{"citing_arxiv_id":"2510.09883","citing_title":"DELTA: Dynamic Layer-Aware Token Attention for Efficient Long-Context Reasoning","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2512.12087","citing_title":"BLASST: Dynamic BLocked Attention Sparsity via Softmax Thresholding","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2604.24820","citing_title":"Salca: A Sparsity-Aware Hardware Accelerator for Efficient Long-Context Attention Decoding","ref_index":60,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YXPVHQEF4AG7JDGPKQDFOU3UA5","json":"https://pith.science/pith/YXPVHQEF4AG7JDGPKQDFOU3UA5.json","graph_json":"https://pith.science/api/pith-number/YXPVHQEF4AG7JDGPKQDFOU3UA5/graph.json","events_json":"https://pith.science/api/pith-number/YXPVHQEF4AG7JDGPKQDFOU3UA5/events.json","paper":"https://pith.science/paper/YXPVHQEF"},"agent_actions":{"view_html":"https://pith.science/pith/YXPVHQEF4AG7JDGPKQDFOU3UA5","download_json":"https://pith.science/pith/YXPVHQEF4AG7JDGPKQDFOU3UA5.json","view_paper":"https://pith.science/paper/YXPVHQEF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.05076&json=true","fetch_graph":"https://pith.science/api/pith-number/YXPVHQEF4AG7JDGPKQDFOU3UA5/graph.json","fetch_events":"https://pith.science/api/pith-number/YXPVHQEF4AG7JDGPKQDFOU3UA5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YXPVHQEF4AG7JDGPKQDFOU3UA5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YXPVHQEF4AG7JDGPKQDFOU3UA5/action/storage_attestation","attest_author":"https://pith.science/pith/YXPVHQEF4AG7JDGPKQDFOU3UA5/action/author_attestation","sign_citation":"https://pith.science/pith/YXPVHQEF4AG7JDGPKQDFOU3UA5/action/citation_signature","submit_replication":"https://pith.science/pith/YXPVHQEF4AG7JDGPKQDFOU3UA5/action/replication_record"}},"created_at":"2026-07-05T09:17:00.989224+00:00","updated_at":"2026-07-05T09:17:00.989224+00:00"}