{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:RZYYJYXXU6ZRUPJAITFKHMHYDI","short_pith_number":"pith:RZYYJYXX","schema_version":"1.0","canonical_sha256":"8e7184e2f7a7b31a3d2044caa3b0f81a1401d5a88ea6bfbef9e322ea875b96ae","source":{"kind":"arxiv","id":"2605.16360","version":1},"attestation_state":"computed","paper":{"title":"ProxyKV: Cross-Model Proxy Pruning for Efficient Long-Context LLM Inference","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Jie Li, Jiong Lou, Junjie Li","submitted_at":"2026-05-09T13:18:01Z","abstract_excerpt":"Efficient long-context inference in Large Language Models (LLMs) is severely constrained by the Key-Value (KV) cache memory wall, yet existing pruning methods force a choice between\n  low-latency heuristics that sacrifice precision and high-precision reconstruction methods that incur prohibitive prefilling overhead. To bridge this scoring-cost--accuracy gap, we propose\n  ProxyKV, a cross-model proxy pruning framework that offloads importance scoring to a lightweight intra-family Small-Model Proxy executed asynchronously to the Large-Model Target. To bridge\n  the architectural gap between heter"},"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":"2605.16360","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-09T13:18:01Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"cd042a377ff4d9ca6abefbea6e1276ee7152e0076b733f8586adc0660e5edf30","abstract_canon_sha256":"e45879c55e7b04a710bfb79f46f3cb03f1512859b5c277ebe08ce32b5e4a23b5"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:02:18.324954Z","signature_b64":"2RF1rtiQm+/dfEtO5TgW5UR1Dh6jCIiRVj+wJfw0AmTGTtKz4iLqJqD/SMJ9glHncA7RyYBQWEgFNdXlUSKCCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8e7184e2f7a7b31a3d2044caa3b0f81a1401d5a88ea6bfbef9e322ea875b96ae","last_reissued_at":"2026-05-20T00:02:18.324271Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:02:18.324271Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ProxyKV: Cross-Model Proxy Pruning for Efficient Long-Context LLM Inference","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Jie Li, Jiong Lou, Junjie Li","submitted_at":"2026-05-09T13:18:01Z","abstract_excerpt":"Efficient long-context inference in Large Language Models (LLMs) is severely constrained by the Key-Value (KV) cache memory wall, yet existing pruning methods force a choice between\n  low-latency heuristics that sacrifice precision and high-precision reconstruction methods that incur prohibitive prefilling overhead. To bridge this scoring-cost--accuracy gap, we propose\n  ProxyKV, a cross-model proxy pruning framework that offloads importance scoring to a lightweight intra-family Small-Model Proxy executed asynchronously to the Large-Model Target. To bridge\n  the architectural gap between heter"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.16360","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/2605.16360/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-19T20:40:54.669726Z","status":"skipped","version":"1.0.0","findings_count":0}],"snapshot_sha256":"5887a91db80f443a5f62cb627d064732253760541d8982fa2784804ce2915128"},"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":"2605.16360","created_at":"2026-05-20T00:02:18.324394+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.16360v1","created_at":"2026-05-20T00:02:18.324394+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.16360","created_at":"2026-05-20T00:02:18.324394+00:00"},{"alias_kind":"pith_short_12","alias_value":"RZYYJYXXU6ZR","created_at":"2026-05-20T00:02:18.324394+00:00"},{"alias_kind":"pith_short_16","alias_value":"RZYYJYXXU6ZRUPJA","created_at":"2026-05-20T00:02:18.324394+00:00"},{"alias_kind":"pith_short_8","alias_value":"RZYYJYXX","created_at":"2026-05-20T00:02:18.324394+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/RZYYJYXXU6ZRUPJAITFKHMHYDI","json":"https://pith.science/pith/RZYYJYXXU6ZRUPJAITFKHMHYDI.json","graph_json":"https://pith.science/api/pith-number/RZYYJYXXU6ZRUPJAITFKHMHYDI/graph.json","events_json":"https://pith.science/api/pith-number/RZYYJYXXU6ZRUPJAITFKHMHYDI/events.json","paper":"https://pith.science/paper/RZYYJYXX"},"agent_actions":{"view_html":"https://pith.science/pith/RZYYJYXXU6ZRUPJAITFKHMHYDI","download_json":"https://pith.science/pith/RZYYJYXXU6ZRUPJAITFKHMHYDI.json","view_paper":"https://pith.science/paper/RZYYJYXX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.16360&json=true","fetch_graph":"https://pith.science/api/pith-number/RZYYJYXXU6ZRUPJAITFKHMHYDI/graph.json","fetch_events":"https://pith.science/api/pith-number/RZYYJYXXU6ZRUPJAITFKHMHYDI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RZYYJYXXU6ZRUPJAITFKHMHYDI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RZYYJYXXU6ZRUPJAITFKHMHYDI/action/storage_attestation","attest_author":"https://pith.science/pith/RZYYJYXXU6ZRUPJAITFKHMHYDI/action/author_attestation","sign_citation":"https://pith.science/pith/RZYYJYXXU6ZRUPJAITFKHMHYDI/action/citation_signature","submit_replication":"https://pith.science/pith/RZYYJYXXU6ZRUPJAITFKHMHYDI/action/replication_record"}},"created_at":"2026-05-20T00:02:18.324394+00:00","updated_at":"2026-05-20T00:02:18.324394+00:00"}