{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:VPHTPNDZPICECTS2JLFSL7J4MF","short_pith_number":"pith:VPHTPNDZ","schema_version":"1.0","canonical_sha256":"abcf37b4797a04414e5a4acb25fd3c616b249f4604a45659af84e3443e716534","source":{"kind":"arxiv","id":"2502.09647","version":2},"attestation_state":"computed","paper":{"title":"Unveiling Simplicities of Attention: Adaptive Long-Context Head Identification","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Charles Arnal, David Lopez-Paz, Kartik Ahuja, Konstantin Donhauser, Mohammad Pezeshki, Vivien Cabannes","submitted_at":"2025-02-11T00:04:32Z","abstract_excerpt":"The ability to process long contexts is crucial for many natural language processing tasks, yet it remains a significant challenge. While substantial progress has been made in enhancing the efficiency of attention mechanisms, there is still a gap in understanding how attention heads function in long-context settings. In this paper, we observe that while certain heads consistently attend to local information only, others swing between attending to local and long-context information depending on the query. This raises the question: can we identify which heads require long-context information to "},"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":"2502.09647","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-02-11T00:04:32Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"b24090efc3a79ee9fb13ac9fa04deb1809e779c40d8e8d625eca48d862bca7a3","abstract_canon_sha256":"95d973cb763c0cb259de985efad977cd77199a4ea78bd6c75cad42c2325e6c07"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:24:55.923939Z","signature_b64":"hTxcrZQoiSm8cg8RQtbZrcICSxwTwnt/ZnLvvU8ZvmbT8+++CU7qXkThEPN/YcOF4uQu/NtXjjtEFFVKFufFAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"abcf37b4797a04414e5a4acb25fd3c616b249f4604a45659af84e3443e716534","last_reissued_at":"2026-07-05T10:24:55.923292Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:24:55.923292Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Unveiling Simplicities of Attention: Adaptive Long-Context Head Identification","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Charles Arnal, David Lopez-Paz, Kartik Ahuja, Konstantin Donhauser, Mohammad Pezeshki, Vivien Cabannes","submitted_at":"2025-02-11T00:04:32Z","abstract_excerpt":"The ability to process long contexts is crucial for many natural language processing tasks, yet it remains a significant challenge. While substantial progress has been made in enhancing the efficiency of attention mechanisms, there is still a gap in understanding how attention heads function in long-context settings. In this paper, we observe that while certain heads consistently attend to local information only, others swing between attending to local and long-context information depending on the query. This raises the question: can we identify which heads require long-context information to "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.09647","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/2502.09647/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":"2502.09647","created_at":"2026-07-05T10:24:55.923372+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.09647v2","created_at":"2026-07-05T10:24:55.923372+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.09647","created_at":"2026-07-05T10:24:55.923372+00:00"},{"alias_kind":"pith_short_12","alias_value":"VPHTPNDZPICE","created_at":"2026-07-05T10:24:55.923372+00:00"},{"alias_kind":"pith_short_16","alias_value":"VPHTPNDZPICECTS2","created_at":"2026-07-05T10:24:55.923372+00:00"},{"alias_kind":"pith_short_8","alias_value":"VPHTPNDZ","created_at":"2026-07-05T10:24:55.923372+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2509.24328","citing_title":"Speculative Verification: Exploiting Information Gain to Refine Speculative Decoding","ref_index":16,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VPHTPNDZPICECTS2JLFSL7J4MF","json":"https://pith.science/pith/VPHTPNDZPICECTS2JLFSL7J4MF.json","graph_json":"https://pith.science/api/pith-number/VPHTPNDZPICECTS2JLFSL7J4MF/graph.json","events_json":"https://pith.science/api/pith-number/VPHTPNDZPICECTS2JLFSL7J4MF/events.json","paper":"https://pith.science/paper/VPHTPNDZ"},"agent_actions":{"view_html":"https://pith.science/pith/VPHTPNDZPICECTS2JLFSL7J4MF","download_json":"https://pith.science/pith/VPHTPNDZPICECTS2JLFSL7J4MF.json","view_paper":"https://pith.science/paper/VPHTPNDZ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.09647&json=true","fetch_graph":"https://pith.science/api/pith-number/VPHTPNDZPICECTS2JLFSL7J4MF/graph.json","fetch_events":"https://pith.science/api/pith-number/VPHTPNDZPICECTS2JLFSL7J4MF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VPHTPNDZPICECTS2JLFSL7J4MF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VPHTPNDZPICECTS2JLFSL7J4MF/action/storage_attestation","attest_author":"https://pith.science/pith/VPHTPNDZPICECTS2JLFSL7J4MF/action/author_attestation","sign_citation":"https://pith.science/pith/VPHTPNDZPICECTS2JLFSL7J4MF/action/citation_signature","submit_replication":"https://pith.science/pith/VPHTPNDZPICECTS2JLFSL7J4MF/action/replication_record"}},"created_at":"2026-07-05T10:24:55.923372+00:00","updated_at":"2026-07-05T10:24:55.923372+00:00"}