{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:PKQLMHWKZNAJCQ54MSTUS5E65S","short_pith_number":"pith:PKQLMHWK","schema_version":"1.0","canonical_sha256":"7aa0b61ecacb409143bc64a749749eecabd8c5630651e1a4a7fd3e93280304cf","source":{"kind":"arxiv","id":"2606.06467","version":1},"attestation_state":"computed","paper":{"title":"You Only Index Once: Cross-Layer Sparse Attention with Shared Routing","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Furu Wei, Jianyong Wang, Li Dong, Yanqi Zhang, Yutao Sun","submitted_at":"2026-06-04T17:54:04Z","abstract_excerpt":"Long-context inference in modern LLMs is increasingly constrained by decoding efficiency, especially in reasoning-heavy settings where models generate long intermediate chains of thought. Existing sparse attention methods often face a practical efficiency-quality trade-off. Structured block sparse methods typically provide stronger acceleration but incur noticeable quality loss, while token sparse methods are usually more accurate yet deliver limited end-to-end speedup because top-k routing over the full cache remains expensive. In this work, we propose cross-layer sparse attention (CLSA), whi"},"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":"2606.06467","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-06-04T17:54:04Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"b80b9cc9fa35ed4c93455359025fc33faf0cdbda569f744ca7802ffbab52c066","abstract_canon_sha256":"136fc0c7944dbfabf5bd6ce9003374c691aa763648954e9c72b6f305ff757700"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-05T01:15:46.423239Z","signature_b64":"iVRsIlRcXtxCVrKiH03kb8zYjtdrddjUa7F64sbfNfeW0e04wNUrLfFwbSW/JOrK0vRjpfH85vJGEVmceXVNCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7aa0b61ecacb409143bc64a749749eecabd8c5630651e1a4a7fd3e93280304cf","last_reissued_at":"2026-06-05T01:15:46.422690Z","signature_status":"signed_v1","first_computed_at":"2026-06-05T01:15:46.422690Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"You Only Index Once: Cross-Layer Sparse Attention with Shared Routing","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Furu Wei, Jianyong Wang, Li Dong, Yanqi Zhang, Yutao Sun","submitted_at":"2026-06-04T17:54:04Z","abstract_excerpt":"Long-context inference in modern LLMs is increasingly constrained by decoding efficiency, especially in reasoning-heavy settings where models generate long intermediate chains of thought. Existing sparse attention methods often face a practical efficiency-quality trade-off. Structured block sparse methods typically provide stronger acceleration but incur noticeable quality loss, while token sparse methods are usually more accurate yet deliver limited end-to-end speedup because top-k routing over the full cache remains expensive. In this work, we propose cross-layer sparse attention (CLSA), whi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.06467","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/2606.06467/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":"2606.06467","created_at":"2026-06-05T01:15:46.422772+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.06467v1","created_at":"2026-06-05T01:15:46.422772+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.06467","created_at":"2026-06-05T01:15:46.422772+00:00"},{"alias_kind":"pith_short_12","alias_value":"PKQLMHWKZNAJ","created_at":"2026-06-05T01:15:46.422772+00:00"},{"alias_kind":"pith_short_16","alias_value":"PKQLMHWKZNAJCQ54","created_at":"2026-06-05T01:15:46.422772+00:00"},{"alias_kind":"pith_short_8","alias_value":"PKQLMHWK","created_at":"2026-06-05T01:15:46.422772+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/PKQLMHWKZNAJCQ54MSTUS5E65S","json":"https://pith.science/pith/PKQLMHWKZNAJCQ54MSTUS5E65S.json","graph_json":"https://pith.science/api/pith-number/PKQLMHWKZNAJCQ54MSTUS5E65S/graph.json","events_json":"https://pith.science/api/pith-number/PKQLMHWKZNAJCQ54MSTUS5E65S/events.json","paper":"https://pith.science/paper/PKQLMHWK"},"agent_actions":{"view_html":"https://pith.science/pith/PKQLMHWKZNAJCQ54MSTUS5E65S","download_json":"https://pith.science/pith/PKQLMHWKZNAJCQ54MSTUS5E65S.json","view_paper":"https://pith.science/paper/PKQLMHWK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.06467&json=true","fetch_graph":"https://pith.science/api/pith-number/PKQLMHWKZNAJCQ54MSTUS5E65S/graph.json","fetch_events":"https://pith.science/api/pith-number/PKQLMHWKZNAJCQ54MSTUS5E65S/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PKQLMHWKZNAJCQ54MSTUS5E65S/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PKQLMHWKZNAJCQ54MSTUS5E65S/action/storage_attestation","attest_author":"https://pith.science/pith/PKQLMHWKZNAJCQ54MSTUS5E65S/action/author_attestation","sign_citation":"https://pith.science/pith/PKQLMHWKZNAJCQ54MSTUS5E65S/action/citation_signature","submit_replication":"https://pith.science/pith/PKQLMHWKZNAJCQ54MSTUS5E65S/action/replication_record"}},"created_at":"2026-06-05T01:15:46.422772+00:00","updated_at":"2026-06-05T01:15:46.422772+00:00"}