{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:XTQKRVTPTN7332XNVDHO2I7TFW","short_pith_number":"pith:XTQKRVTP","schema_version":"1.0","canonical_sha256":"bce0a8d66f9b7fbdeaeda8ceed23f32da2877c0c55bda6746e31d9447dceff63","source":{"kind":"arxiv","id":"2502.01770","version":1},"attestation_state":"computed","paper":{"title":"Hamming Attention Distillation: Binarizing Keys and Queries for Efficient Long-Context Transformers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","eess.IV"],"primary_cat":"cs.LG","authors_text":"Bhavna Gopal, Brady Taylor, Chiyue Wei, Cong Guo, Deliang Fan, Hai Li, Mark Horton, Peter Liu, Shan X. Wang, Tergel Molom-Ochir, Yiran Chen","submitted_at":"2025-02-03T19:24:01Z","abstract_excerpt":"Pre-trained transformer models with extended context windows are notoriously expensive to run at scale, often limiting real-world deployment due to their high computational and memory requirements. In this paper, we introduce Hamming Attention Distillation (HAD), a novel framework that binarizes keys and queries in the attention mechanism to achieve significant efficiency gains. By converting keys and queries into {-1, +1} vectors and replacing dot-product operations with efficient Hamming distance computations, our method drastically reduces computational overhead. Additionally, we incorporat"},"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":"2502.01770","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-03T19:24:01Z","cross_cats_sorted":["cs.AI","eess.IV"],"title_canon_sha256":"85a63360d6180b62a6e7d84724d282566eb8d624a909f9961fb66db45071c05a","abstract_canon_sha256":"14d22d7adde672952f62f15b2fada3b5fcd76cf1669694e5fd151e903503fc88"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:09:18.782640Z","signature_b64":"UD9qa2l6uXowWKUcCUdNIrku+3Q4lVSUkgDnDMd5nA42JCl6H/U8Y+cbeE2o6xZbAB9CA0qDj3f3nyxEr8VcBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"bce0a8d66f9b7fbdeaeda8ceed23f32da2877c0c55bda6746e31d9447dceff63","last_reissued_at":"2026-07-05T10:09:18.782148Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:09:18.782148Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Hamming Attention Distillation: Binarizing Keys and Queries for Efficient Long-Context Transformers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","eess.IV"],"primary_cat":"cs.LG","authors_text":"Bhavna Gopal, Brady Taylor, Chiyue Wei, Cong Guo, Deliang Fan, Hai Li, Mark Horton, Peter Liu, Shan X. Wang, Tergel Molom-Ochir, Yiran Chen","submitted_at":"2025-02-03T19:24:01Z","abstract_excerpt":"Pre-trained transformer models with extended context windows are notoriously expensive to run at scale, often limiting real-world deployment due to their high computational and memory requirements. In this paper, we introduce Hamming Attention Distillation (HAD), a novel framework that binarizes keys and queries in the attention mechanism to achieve significant efficiency gains. By converting keys and queries into {-1, +1} vectors and replacing dot-product operations with efficient Hamming distance computations, our method drastically reduces computational overhead. Additionally, we incorporat"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.01770","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/2502.01770/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":"2502.01770","created_at":"2026-07-05T10:09:18.782206+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.01770v1","created_at":"2026-07-05T10:09:18.782206+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.01770","created_at":"2026-07-05T10:09:18.782206+00:00"},{"alias_kind":"pith_short_12","alias_value":"XTQKRVTPTN73","created_at":"2026-07-05T10:09:18.782206+00:00"},{"alias_kind":"pith_short_16","alias_value":"XTQKRVTPTN7332XN","created_at":"2026-07-05T10:09:18.782206+00:00"},{"alias_kind":"pith_short_8","alias_value":"XTQKRVTP","created_at":"2026-07-05T10:09:18.782206+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/XTQKRVTPTN7332XNVDHO2I7TFW","json":"https://pith.science/pith/XTQKRVTPTN7332XNVDHO2I7TFW.json","graph_json":"https://pith.science/api/pith-number/XTQKRVTPTN7332XNVDHO2I7TFW/graph.json","events_json":"https://pith.science/api/pith-number/XTQKRVTPTN7332XNVDHO2I7TFW/events.json","paper":"https://pith.science/paper/XTQKRVTP"},"agent_actions":{"view_html":"https://pith.science/pith/XTQKRVTPTN7332XNVDHO2I7TFW","download_json":"https://pith.science/pith/XTQKRVTPTN7332XNVDHO2I7TFW.json","view_paper":"https://pith.science/paper/XTQKRVTP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.01770&json=true","fetch_graph":"https://pith.science/api/pith-number/XTQKRVTPTN7332XNVDHO2I7TFW/graph.json","fetch_events":"https://pith.science/api/pith-number/XTQKRVTPTN7332XNVDHO2I7TFW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XTQKRVTPTN7332XNVDHO2I7TFW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XTQKRVTPTN7332XNVDHO2I7TFW/action/storage_attestation","attest_author":"https://pith.science/pith/XTQKRVTPTN7332XNVDHO2I7TFW/action/author_attestation","sign_citation":"https://pith.science/pith/XTQKRVTPTN7332XNVDHO2I7TFW/action/citation_signature","submit_replication":"https://pith.science/pith/XTQKRVTPTN7332XNVDHO2I7TFW/action/replication_record"}},"created_at":"2026-07-05T10:09:18.782206+00:00","updated_at":"2026-07-05T10:09:18.782206+00:00"}