{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:FEIDO233DGX4TYELJ7CRIHF3Z6","short_pith_number":"pith:FEIDO233","schema_version":"1.0","canonical_sha256":"2910376b7b19afc9e08b4fc5141cbbcfbe747baa2f8ffb24d46951264a65f6fb","source":{"kind":"arxiv","id":"2509.09001","version":1},"attestation_state":"computed","paper":{"title":"Fast attention mechanisms: a tale of parallelism","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Alexandr Andoni, Clayton Sanford, Daniel Hsu, Hantao Yu, Jingwen Liu","submitted_at":"2025-09-10T20:59:44Z","abstract_excerpt":"Transformers have the representational capacity to simulate Massively Parallel Computation (MPC) algorithms, but they suffer from quadratic time complexity, which severely limits their scalability. We introduce an efficient attention mechanism called Approximate Nearest Neighbor Attention (ANNA) with sub-quadratic time complexity. We prove that ANNA-transformers (1) retain the expressive power previously established for standard attention in terms of matching the capabilities of MPC algorithms, and (2) can solve key reasoning tasks such as Match2 and $k$-hop with near-optimal depth. Using the "},"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":"2509.09001","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-09-10T20:59:44Z","cross_cats_sorted":[],"title_canon_sha256":"533aca3bc1ac5594d6fedf5ca32ce9641fa3d50484de75675b4c2ea7ff124053","abstract_canon_sha256":"2be338ba7df92afcdadafcd574a91716a5053fd2a627ab8d5b663cad56693e88"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:09:27.384594Z","signature_b64":"2HB6vCvVIgw/LoLT5epuK9XCOS9GUUFKlxlVA5TbJILElibteRBBy5WnW1FKtkbjW6XZH3dcjsHMmTYT3s0+Dg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2910376b7b19afc9e08b4fc5141cbbcfbe747baa2f8ffb24d46951264a65f6fb","last_reissued_at":"2026-07-05T12:09:27.383972Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:09:27.383972Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Fast attention mechanisms: a tale of parallelism","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Alexandr Andoni, Clayton Sanford, Daniel Hsu, Hantao Yu, Jingwen Liu","submitted_at":"2025-09-10T20:59:44Z","abstract_excerpt":"Transformers have the representational capacity to simulate Massively Parallel Computation (MPC) algorithms, but they suffer from quadratic time complexity, which severely limits their scalability. We introduce an efficient attention mechanism called Approximate Nearest Neighbor Attention (ANNA) with sub-quadratic time complexity. We prove that ANNA-transformers (1) retain the expressive power previously established for standard attention in terms of matching the capabilities of MPC algorithms, and (2) can solve key reasoning tasks such as Match2 and $k$-hop with near-optimal depth. Using the "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2509.09001","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/2509.09001/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":"2509.09001","created_at":"2026-07-05T12:09:27.384049+00:00"},{"alias_kind":"arxiv_version","alias_value":"2509.09001v1","created_at":"2026-07-05T12:09:27.384049+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.09001","created_at":"2026-07-05T12:09:27.384049+00:00"},{"alias_kind":"pith_short_12","alias_value":"FEIDO233DGX4","created_at":"2026-07-05T12:09:27.384049+00:00"},{"alias_kind":"pith_short_16","alias_value":"FEIDO233DGX4TYEL","created_at":"2026-07-05T12:09:27.384049+00:00"},{"alias_kind":"pith_short_8","alias_value":"FEIDO233","created_at":"2026-07-05T12:09:27.384049+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2607.01283","citing_title":"Scaling Laws for Grid-Based Approximate Nearest Neighbor Search in High Dimensions","ref_index":1,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FEIDO233DGX4TYELJ7CRIHF3Z6","json":"https://pith.science/pith/FEIDO233DGX4TYELJ7CRIHF3Z6.json","graph_json":"https://pith.science/api/pith-number/FEIDO233DGX4TYELJ7CRIHF3Z6/graph.json","events_json":"https://pith.science/api/pith-number/FEIDO233DGX4TYELJ7CRIHF3Z6/events.json","paper":"https://pith.science/paper/FEIDO233"},"agent_actions":{"view_html":"https://pith.science/pith/FEIDO233DGX4TYELJ7CRIHF3Z6","download_json":"https://pith.science/pith/FEIDO233DGX4TYELJ7CRIHF3Z6.json","view_paper":"https://pith.science/paper/FEIDO233","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2509.09001&json=true","fetch_graph":"https://pith.science/api/pith-number/FEIDO233DGX4TYELJ7CRIHF3Z6/graph.json","fetch_events":"https://pith.science/api/pith-number/FEIDO233DGX4TYELJ7CRIHF3Z6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FEIDO233DGX4TYELJ7CRIHF3Z6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FEIDO233DGX4TYELJ7CRIHF3Z6/action/storage_attestation","attest_author":"https://pith.science/pith/FEIDO233DGX4TYELJ7CRIHF3Z6/action/author_attestation","sign_citation":"https://pith.science/pith/FEIDO233DGX4TYELJ7CRIHF3Z6/action/citation_signature","submit_replication":"https://pith.science/pith/FEIDO233DGX4TYELJ7CRIHF3Z6/action/replication_record"}},"created_at":"2026-07-05T12:09:27.384049+00:00","updated_at":"2026-07-05T12:09:27.384049+00:00"}