{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:VEXSW2S4Y7QL3JPC77KRPZ5XKZ","short_pith_number":"pith:VEXSW2S4","schema_version":"1.0","canonical_sha256":"a92f2b6a5cc7e0bda5e2ffd517e7b75645f20da01206e55c6eee41ea01c6e5d3","source":{"kind":"arxiv","id":"2606.14782","version":2},"attestation_state":"computed","paper":{"title":"Last But Not Least: Boundary Attention CalibratiON for Multimodal KV Cache Compression","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Dongman Lee, Kelu Yao, Tianhao Chen, Xiaobin Hu, Xiaogang Xu, Yuheng Wu","submitted_at":"2026-06-10T10:09:59Z","abstract_excerpt":"Multimodal Large Language Models (MLLMs) achieve strong vision-language reasoning, but long visual contexts enlarge the KV cache and increase decoding latency. Existing compression methods rely on observation window attention for stable token-importance estimation, yet this aggregation can dilute sparse visual evidence and discard answer-critical tokens under aggressive compression. Therefore, we identify last-query attention as a complementary source for recovering such evidence, but its answer-irrelevant signals can mislead retention. We propose BACON, a plug-and-play method that calibrates "},"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":"2606.14782","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CV","submitted_at":"2026-06-10T10:09:59Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"737c002225f092b0088a403f5d74cd6f50f9f7c5da120464ec8d35d631da516a","abstract_canon_sha256":"672f6535e0a21848dac2851da37f1ca45cd75c7e0cdc8563113557e6b711b563"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-19T16:10:01.031866Z","signature_b64":"Ec6wYiegOU4vh6FpDdTKDFly4ehacOFgX0BOz67g21C5Y5h9yDzLIw6vUdej4kWY9j89TM1WoviHZ2WaoC93Cw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a92f2b6a5cc7e0bda5e2ffd517e7b75645f20da01206e55c6eee41ea01c6e5d3","last_reissued_at":"2026-06-19T16:10:01.031468Z","signature_status":"signed_v1","first_computed_at":"2026-06-19T16:10:01.031468Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Last But Not Least: Boundary Attention CalibratiON for Multimodal KV Cache Compression","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Dongman Lee, Kelu Yao, Tianhao Chen, Xiaobin Hu, Xiaogang Xu, Yuheng Wu","submitted_at":"2026-06-10T10:09:59Z","abstract_excerpt":"Multimodal Large Language Models (MLLMs) achieve strong vision-language reasoning, but long visual contexts enlarge the KV cache and increase decoding latency. Existing compression methods rely on observation window attention for stable token-importance estimation, yet this aggregation can dilute sparse visual evidence and discard answer-critical tokens under aggressive compression. Therefore, we identify last-query attention as a complementary source for recovering such evidence, but its answer-irrelevant signals can mislead retention. We propose BACON, a plug-and-play method that calibrates "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.14782","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/2606.14782/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":"2606.14782","created_at":"2026-06-19T16:10:01.031525+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.14782v2","created_at":"2026-06-19T16:10:01.031525+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.14782","created_at":"2026-06-19T16:10:01.031525+00:00"},{"alias_kind":"pith_short_12","alias_value":"VEXSW2S4Y7QL","created_at":"2026-06-19T16:10:01.031525+00:00"},{"alias_kind":"pith_short_16","alias_value":"VEXSW2S4Y7QL3JPC","created_at":"2026-06-19T16:10:01.031525+00:00"},{"alias_kind":"pith_short_8","alias_value":"VEXSW2S4","created_at":"2026-06-19T16:10:01.031525+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/VEXSW2S4Y7QL3JPC77KRPZ5XKZ","json":"https://pith.science/pith/VEXSW2S4Y7QL3JPC77KRPZ5XKZ.json","graph_json":"https://pith.science/api/pith-number/VEXSW2S4Y7QL3JPC77KRPZ5XKZ/graph.json","events_json":"https://pith.science/api/pith-number/VEXSW2S4Y7QL3JPC77KRPZ5XKZ/events.json","paper":"https://pith.science/paper/VEXSW2S4"},"agent_actions":{"view_html":"https://pith.science/pith/VEXSW2S4Y7QL3JPC77KRPZ5XKZ","download_json":"https://pith.science/pith/VEXSW2S4Y7QL3JPC77KRPZ5XKZ.json","view_paper":"https://pith.science/paper/VEXSW2S4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.14782&json=true","fetch_graph":"https://pith.science/api/pith-number/VEXSW2S4Y7QL3JPC77KRPZ5XKZ/graph.json","fetch_events":"https://pith.science/api/pith-number/VEXSW2S4Y7QL3JPC77KRPZ5XKZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VEXSW2S4Y7QL3JPC77KRPZ5XKZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VEXSW2S4Y7QL3JPC77KRPZ5XKZ/action/storage_attestation","attest_author":"https://pith.science/pith/VEXSW2S4Y7QL3JPC77KRPZ5XKZ/action/author_attestation","sign_citation":"https://pith.science/pith/VEXSW2S4Y7QL3JPC77KRPZ5XKZ/action/citation_signature","submit_replication":"https://pith.science/pith/VEXSW2S4Y7QL3JPC77KRPZ5XKZ/action/replication_record"}},"created_at":"2026-06-19T16:10:01.031525+00:00","updated_at":"2026-06-19T16:10:01.031525+00:00"}