{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:KFKCSGA4CCUKDJBMUHHKV5G77X","short_pith_number":"pith:KFKCSGA4","schema_version":"1.0","canonical_sha256":"515429181c10a8a1a42ca1ceaaf4dffdc8d07a643a13fecb49eb2d5e4cb0af66","source":{"kind":"arxiv","id":"2605.24022","version":1},"attestation_state":"computed","paper":{"title":"Adaptive KV Cache Reuse for Fast Long-Context LLM Serving","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.DC"],"primary_cat":"cs.AR","authors_text":"Fei Li, Jinhua Cui, Jinyu Wang, Shiqiang Nie, Song Liu, Weiguo Wu, Yan Liu","submitted_at":"2026-05-20T08:59:48Z","abstract_excerpt":"In long-context Large Language Model (LLM) inference, the Time-To-First-Token (TTFT) latency incurred by the prefill stage has become the foremost bottleneck limiting interactive performance and deployment cost. KV Cache reuse offers a direct path to reduce redundant prefill, yet traditional prefix caching applies only to strict-prefix scenarios; directly reusing KV Cache in non-prefix settings breaks the cross-chunk global attention relationships and causes significant degradation in generation quality. When reusable KV Cache is offloaded to GPU-external cache pools, I/O overheads across hete"},"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":"2605.24022","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AR","submitted_at":"2026-05-20T08:59:48Z","cross_cats_sorted":["cs.DC"],"title_canon_sha256":"a204636f5eae488c76ffcb54711778cdbdf236a949350b61a831b967c948a521","abstract_canon_sha256":"77309f591b6ec04f6074d8bca5a8673ba9698ff912507c4d3d06e712bf1c142e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-26T01:02:41.545630Z","signature_b64":"/Ah98liyz7O8DiHzjgeOkTMc3riQqh8JWXnoCZ3cBO2o2VbcOCRojmWGAQN7BKDRRddaiL9EemR961Jr6fDHBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"515429181c10a8a1a42ca1ceaaf4dffdc8d07a643a13fecb49eb2d5e4cb0af66","last_reissued_at":"2026-05-26T01:02:41.544572Z","signature_status":"signed_v1","first_computed_at":"2026-05-26T01:02:41.544572Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Adaptive KV Cache Reuse for Fast Long-Context LLM Serving","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.DC"],"primary_cat":"cs.AR","authors_text":"Fei Li, Jinhua Cui, Jinyu Wang, Shiqiang Nie, Song Liu, Weiguo Wu, Yan Liu","submitted_at":"2026-05-20T08:59:48Z","abstract_excerpt":"In long-context Large Language Model (LLM) inference, the Time-To-First-Token (TTFT) latency incurred by the prefill stage has become the foremost bottleneck limiting interactive performance and deployment cost. KV Cache reuse offers a direct path to reduce redundant prefill, yet traditional prefix caching applies only to strict-prefix scenarios; directly reusing KV Cache in non-prefix settings breaks the cross-chunk global attention relationships and causes significant degradation in generation quality. When reusable KV Cache is offloaded to GPU-external cache pools, I/O overheads across hete"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.24022","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/2605.24022/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":"2605.24022","created_at":"2026-05-26T01:02:41.544695+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.24022v1","created_at":"2026-05-26T01:02:41.544695+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.24022","created_at":"2026-05-26T01:02:41.544695+00:00"},{"alias_kind":"pith_short_12","alias_value":"KFKCSGA4CCUK","created_at":"2026-05-26T01:02:41.544695+00:00"},{"alias_kind":"pith_short_16","alias_value":"KFKCSGA4CCUKDJBM","created_at":"2026-05-26T01:02:41.544695+00:00"},{"alias_kind":"pith_short_8","alias_value":"KFKCSGA4","created_at":"2026-05-26T01:02:41.544695+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/KFKCSGA4CCUKDJBMUHHKV5G77X","json":"https://pith.science/pith/KFKCSGA4CCUKDJBMUHHKV5G77X.json","graph_json":"https://pith.science/api/pith-number/KFKCSGA4CCUKDJBMUHHKV5G77X/graph.json","events_json":"https://pith.science/api/pith-number/KFKCSGA4CCUKDJBMUHHKV5G77X/events.json","paper":"https://pith.science/paper/KFKCSGA4"},"agent_actions":{"view_html":"https://pith.science/pith/KFKCSGA4CCUKDJBMUHHKV5G77X","download_json":"https://pith.science/pith/KFKCSGA4CCUKDJBMUHHKV5G77X.json","view_paper":"https://pith.science/paper/KFKCSGA4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.24022&json=true","fetch_graph":"https://pith.science/api/pith-number/KFKCSGA4CCUKDJBMUHHKV5G77X/graph.json","fetch_events":"https://pith.science/api/pith-number/KFKCSGA4CCUKDJBMUHHKV5G77X/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KFKCSGA4CCUKDJBMUHHKV5G77X/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KFKCSGA4CCUKDJBMUHHKV5G77X/action/storage_attestation","attest_author":"https://pith.science/pith/KFKCSGA4CCUKDJBMUHHKV5G77X/action/author_attestation","sign_citation":"https://pith.science/pith/KFKCSGA4CCUKDJBMUHHKV5G77X/action/citation_signature","submit_replication":"https://pith.science/pith/KFKCSGA4CCUKDJBMUHHKV5G77X/action/replication_record"}},"created_at":"2026-05-26T01:02:41.544695+00:00","updated_at":"2026-05-26T01:02:41.544695+00:00"}