{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:E6TGNSWX3VETNV3LFMRHDCOCOU","short_pith_number":"pith:E6TGNSWX","schema_version":"1.0","canonical_sha256":"27a666cad7dd4936d76b2b227189c27539346411accfe010f4e9eacde2dd707c","source":{"kind":"arxiv","id":"2411.07176","version":3},"attestation_state":"computed","paper":{"title":"More Expressive Attention with Negative Weights","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Ang Lv, Di Wang, Jiayi Liao, Rui Yan, Ruobing Xie, Shuaipeng Li, Xingwu Sun, Zhanhui Kang","submitted_at":"2024-11-11T17:56:28Z","abstract_excerpt":"We propose a novel attention mechanism, named Cog Attention, that enables attention weights to be negative for enhanced expressiveness, which stems from two key factors: (1) Cog Attention enhances parameter flexibility. For example, unlike traditional softmax attention heads that use a static output-value (OV) matrix to delete or copy inputs that the heads attend to, Cog Attention naturally learns to use the sign of dynamic query-key (QK) inner products to represent these operations. This enables Cog Attention to perform multiple operations simultaneously within a single head. Meanwhile, Cog A"},"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":"2411.07176","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2024-11-11T17:56:28Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"e27864037fcc628927d0433779d8194ff8793923e4776122cbce483954a6c9a6","abstract_canon_sha256":"68bedb1c8b55b6593504613dc05773426899213f71cb74bdbdd1c935b6acec9e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:07:28.463681Z","signature_b64":"dzW66uxS67NtyApy+votj8us1tsFbwJ2Iucl5X/k2g6MmOgiAn5Zj7q3DpW9DF/TdmhwBWs5x34vp+0bD3m4Bw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"27a666cad7dd4936d76b2b227189c27539346411accfe010f4e9eacde2dd707c","last_reissued_at":"2026-07-05T10:07:28.463211Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:07:28.463211Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"More Expressive Attention with Negative Weights","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Ang Lv, Di Wang, Jiayi Liao, Rui Yan, Ruobing Xie, Shuaipeng Li, Xingwu Sun, Zhanhui Kang","submitted_at":"2024-11-11T17:56:28Z","abstract_excerpt":"We propose a novel attention mechanism, named Cog Attention, that enables attention weights to be negative for enhanced expressiveness, which stems from two key factors: (1) Cog Attention enhances parameter flexibility. For example, unlike traditional softmax attention heads that use a static output-value (OV) matrix to delete or copy inputs that the heads attend to, Cog Attention naturally learns to use the sign of dynamic query-key (QK) inner products to represent these operations. This enables Cog Attention to perform multiple operations simultaneously within a single head. Meanwhile, Cog A"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2411.07176","kind":"arxiv","version":3},"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/2411.07176/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":"2411.07176","created_at":"2026-07-05T10:07:28.463270+00:00"},{"alias_kind":"arxiv_version","alias_value":"2411.07176v3","created_at":"2026-07-05T10:07:28.463270+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2411.07176","created_at":"2026-07-05T10:07:28.463270+00:00"},{"alias_kind":"pith_short_12","alias_value":"E6TGNSWX3VET","created_at":"2026-07-05T10:07:28.463270+00:00"},{"alias_kind":"pith_short_16","alias_value":"E6TGNSWX3VETNV3L","created_at":"2026-07-05T10:07:28.463270+00:00"},{"alias_kind":"pith_short_8","alias_value":"E6TGNSWX","created_at":"2026-07-05T10:07:28.463270+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.27372","citing_title":"DnA: Denoising Attention for Visual Tasks","ref_index":50,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/E6TGNSWX3VETNV3LFMRHDCOCOU","json":"https://pith.science/pith/E6TGNSWX3VETNV3LFMRHDCOCOU.json","graph_json":"https://pith.science/api/pith-number/E6TGNSWX3VETNV3LFMRHDCOCOU/graph.json","events_json":"https://pith.science/api/pith-number/E6TGNSWX3VETNV3LFMRHDCOCOU/events.json","paper":"https://pith.science/paper/E6TGNSWX"},"agent_actions":{"view_html":"https://pith.science/pith/E6TGNSWX3VETNV3LFMRHDCOCOU","download_json":"https://pith.science/pith/E6TGNSWX3VETNV3LFMRHDCOCOU.json","view_paper":"https://pith.science/paper/E6TGNSWX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2411.07176&json=true","fetch_graph":"https://pith.science/api/pith-number/E6TGNSWX3VETNV3LFMRHDCOCOU/graph.json","fetch_events":"https://pith.science/api/pith-number/E6TGNSWX3VETNV3LFMRHDCOCOU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/E6TGNSWX3VETNV3LFMRHDCOCOU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/E6TGNSWX3VETNV3LFMRHDCOCOU/action/storage_attestation","attest_author":"https://pith.science/pith/E6TGNSWX3VETNV3LFMRHDCOCOU/action/author_attestation","sign_citation":"https://pith.science/pith/E6TGNSWX3VETNV3LFMRHDCOCOU/action/citation_signature","submit_replication":"https://pith.science/pith/E6TGNSWX3VETNV3LFMRHDCOCOU/action/replication_record"}},"created_at":"2026-07-05T10:07:28.463270+00:00","updated_at":"2026-07-05T10:07:28.463270+00:00"}