{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:KJ4MW7BA7CZCKWSQ5HNYC3A77T","short_pith_number":"pith:KJ4MW7BA","schema_version":"1.0","canonical_sha256":"5278cb7c20f8b2255a50e9db816c1ffcce22e5428478424c922327f8934a9833","source":{"kind":"arxiv","id":"2407.01601","version":2},"attestation_state":"computed","paper":{"title":"Unveiling and Controlling Anomalous Attention Distribution in Transformers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Haoyu Deng, Jifang Hu, Jinrong Jiang, Lian Zhao, Linghan Zheng, Penghao Jiang, Qiuzhuang Sun, Ruiqing Yan, Xingbo Du, Yuhang Shao","submitted_at":"2024-06-26T11:53:35Z","abstract_excerpt":"With the advent of large models based on the Transformer architecture, researchers have observed an anomalous phenomenon in the Attention mechanism--there is a very high attention on the first element, which is prevalent across Transformer-based models. It is crucial to understand it for the development of techniques focusing on attention distribution, such as Key-Value (KV) Cache compression and infinite extrapolation; however, the latent cause leaves to be unknown. In this paper, we analyze such a phenomenon from the perspective of waiver phenomenon, which involves reducing the internal valu"},"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":"2407.01601","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-06-26T11:53:35Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"efd891f5da99d56810a9ec26297334f0e8698368ebd7b07d44af29adb04bc74b","abstract_canon_sha256":"d01755ad5d664ed5b59d846e1609f4cc8706734d2904e8b518cde1443a799e2e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:39:36.695809Z","signature_b64":"rqPLujI0z8nmvt7GTTOsIAOmD9e/zJIX0jU0U0JogmwmGRJCyoY/Dfu8MatjiphbNUbIJ9F3sk9aWCu757JjCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5278cb7c20f8b2255a50e9db816c1ffcce22e5428478424c922327f8934a9833","last_reissued_at":"2026-07-05T08:39:36.695371Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:39:36.695371Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Unveiling and Controlling Anomalous Attention Distribution in Transformers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Haoyu Deng, Jifang Hu, Jinrong Jiang, Lian Zhao, Linghan Zheng, Penghao Jiang, Qiuzhuang Sun, Ruiqing Yan, Xingbo Du, Yuhang Shao","submitted_at":"2024-06-26T11:53:35Z","abstract_excerpt":"With the advent of large models based on the Transformer architecture, researchers have observed an anomalous phenomenon in the Attention mechanism--there is a very high attention on the first element, which is prevalent across Transformer-based models. It is crucial to understand it for the development of techniques focusing on attention distribution, such as Key-Value (KV) Cache compression and infinite extrapolation; however, the latent cause leaves to be unknown. In this paper, we analyze such a phenomenon from the perspective of waiver phenomenon, which involves reducing the internal valu"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.01601","kind":"arxiv","version":2},"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/2407.01601/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":"2407.01601","created_at":"2026-07-05T08:39:36.695428+00:00"},{"alias_kind":"arxiv_version","alias_value":"2407.01601v2","created_at":"2026-07-05T08:39:36.695428+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.01601","created_at":"2026-07-05T08:39:36.695428+00:00"},{"alias_kind":"pith_short_12","alias_value":"KJ4MW7BA7CZC","created_at":"2026-07-05T08:39:36.695428+00:00"},{"alias_kind":"pith_short_16","alias_value":"KJ4MW7BA7CZCKWSQ","created_at":"2026-07-05T08:39:36.695428+00:00"},{"alias_kind":"pith_short_8","alias_value":"KJ4MW7BA","created_at":"2026-07-05T08:39:36.695428+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.06611","citing_title":"The Structural Origin of Attention Sink: Variance Discrepancy, Super Neurons, and Dimension Disparity","ref_index":30,"is_internal_anchor":false},{"citing_arxiv_id":"2604.10098","citing_title":"Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation","ref_index":168,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KJ4MW7BA7CZCKWSQ5HNYC3A77T","json":"https://pith.science/pith/KJ4MW7BA7CZCKWSQ5HNYC3A77T.json","graph_json":"https://pith.science/api/pith-number/KJ4MW7BA7CZCKWSQ5HNYC3A77T/graph.json","events_json":"https://pith.science/api/pith-number/KJ4MW7BA7CZCKWSQ5HNYC3A77T/events.json","paper":"https://pith.science/paper/KJ4MW7BA"},"agent_actions":{"view_html":"https://pith.science/pith/KJ4MW7BA7CZCKWSQ5HNYC3A77T","download_json":"https://pith.science/pith/KJ4MW7BA7CZCKWSQ5HNYC3A77T.json","view_paper":"https://pith.science/paper/KJ4MW7BA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2407.01601&json=true","fetch_graph":"https://pith.science/api/pith-number/KJ4MW7BA7CZCKWSQ5HNYC3A77T/graph.json","fetch_events":"https://pith.science/api/pith-number/KJ4MW7BA7CZCKWSQ5HNYC3A77T/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KJ4MW7BA7CZCKWSQ5HNYC3A77T/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KJ4MW7BA7CZCKWSQ5HNYC3A77T/action/storage_attestation","attest_author":"https://pith.science/pith/KJ4MW7BA7CZCKWSQ5HNYC3A77T/action/author_attestation","sign_citation":"https://pith.science/pith/KJ4MW7BA7CZCKWSQ5HNYC3A77T/action/citation_signature","submit_replication":"https://pith.science/pith/KJ4MW7BA7CZCKWSQ5HNYC3A77T/action/replication_record"}},"created_at":"2026-07-05T08:39:36.695428+00:00","updated_at":"2026-07-05T08:39:36.695428+00:00"}