{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:KN3WL7MYIKH35QMD5EOATROK2P","short_pith_number":"pith:KN3WL7MY","schema_version":"1.0","canonical_sha256":"537765fd98428fbec183e91c09c5cad3fddb7848cf54b3593c8f1a3aba4c3fc8","source":{"kind":"arxiv","id":"2410.03065","version":2},"attestation_state":"computed","paper":{"title":"Compute Or Load KV Cache? Why Not Both?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Qingzhao Zhang, Shuowei Jin, Xueshen Liu, Z. Morley Mao","submitted_at":"2024-10-04T01:11:09Z","abstract_excerpt":"Large Language Models (LLMs) are increasingly deployed in large-scale online services, enabling sophisticated applications. However, the computational overhead of generating key-value (KV) caches in the prefill stage presents a major bottleneck, particularly for long-context inputs. Prefix caching mitigates this issue by storing KV caches for reuse, reducing redundant computation. Despite its advantages, prefix caching suffers from high latency due to the limited I/O bandwidth of storage devices, constraining inference efficiency. To address this challenge, we introduce Cake, a novel KV cache "},"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":"2410.03065","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-10-04T01:11:09Z","cross_cats_sorted":[],"title_canon_sha256":"688eb095aec270cf0b413bb24a032bcca4c066110063080f2614295a13f502bb","abstract_canon_sha256":"dcb8332bf234a20baa64e5d53d75210536ae00ecc6d0967b61670d67130f1594"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:17:50.276221Z","signature_b64":"yFqrL4rDSQV0UinqPxYgdcDoXp1jB3e45Af2fzH/Wmq1ztbVc1to30gE5YlR527+Bhr63BHWWEpbV/KrKQBsDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"537765fd98428fbec183e91c09c5cad3fddb7848cf54b3593c8f1a3aba4c3fc8","last_reissued_at":"2026-07-05T10:17:50.275743Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:17:50.275743Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Compute Or Load KV Cache? Why Not Both?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Qingzhao Zhang, Shuowei Jin, Xueshen Liu, Z. Morley Mao","submitted_at":"2024-10-04T01:11:09Z","abstract_excerpt":"Large Language Models (LLMs) are increasingly deployed in large-scale online services, enabling sophisticated applications. However, the computational overhead of generating key-value (KV) caches in the prefill stage presents a major bottleneck, particularly for long-context inputs. Prefix caching mitigates this issue by storing KV caches for reuse, reducing redundant computation. Despite its advantages, prefix caching suffers from high latency due to the limited I/O bandwidth of storage devices, constraining inference efficiency. To address this challenge, we introduce Cake, a novel KV cache "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.03065","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/2410.03065/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":"2410.03065","created_at":"2026-07-05T10:17:50.275803+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.03065v2","created_at":"2026-07-05T10:17:50.275803+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.03065","created_at":"2026-07-05T10:17:50.275803+00:00"},{"alias_kind":"pith_short_12","alias_value":"KN3WL7MYIKH3","created_at":"2026-07-05T10:17:50.275803+00:00"},{"alias_kind":"pith_short_16","alias_value":"KN3WL7MYIKH35QMD","created_at":"2026-07-05T10:17:50.275803+00:00"},{"alias_kind":"pith_short_8","alias_value":"KN3WL7MY","created_at":"2026-07-05T10:17:50.275803+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.02964","citing_title":"Multi-Segment Attention: Enabling Efficient KV-Cache Management for Faster Large Language Model Serving","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01563","citing_title":"MomentKV: Closing the Directional Gap in KV Cache Eviction for Long-Context Inference","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01065","citing_title":"Leyline: KV Cache Directives for Agentic Inference","ref_index":40,"is_internal_anchor":false},{"citing_arxiv_id":"2605.23640","citing_title":"CachePrune: Privacy-Aware and Fine-Grained KV Cache Sharing for Efficient LLM Inference","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2504.15965","citing_title":"From Human Memory to AI Memory: A Survey on Memory Mechanisms in the Era of LLMs","ref_index":135,"is_internal_anchor":false},{"citing_arxiv_id":"2604.21231","citing_title":"SparKV: Overhead-Aware KV Cache Loading for Efficient On-Device LLM Inference","ref_index":25,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KN3WL7MYIKH35QMD5EOATROK2P","json":"https://pith.science/pith/KN3WL7MYIKH35QMD5EOATROK2P.json","graph_json":"https://pith.science/api/pith-number/KN3WL7MYIKH35QMD5EOATROK2P/graph.json","events_json":"https://pith.science/api/pith-number/KN3WL7MYIKH35QMD5EOATROK2P/events.json","paper":"https://pith.science/paper/KN3WL7MY"},"agent_actions":{"view_html":"https://pith.science/pith/KN3WL7MYIKH35QMD5EOATROK2P","download_json":"https://pith.science/pith/KN3WL7MYIKH35QMD5EOATROK2P.json","view_paper":"https://pith.science/paper/KN3WL7MY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.03065&json=true","fetch_graph":"https://pith.science/api/pith-number/KN3WL7MYIKH35QMD5EOATROK2P/graph.json","fetch_events":"https://pith.science/api/pith-number/KN3WL7MYIKH35QMD5EOATROK2P/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KN3WL7MYIKH35QMD5EOATROK2P/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KN3WL7MYIKH35QMD5EOATROK2P/action/storage_attestation","attest_author":"https://pith.science/pith/KN3WL7MYIKH35QMD5EOATROK2P/action/author_attestation","sign_citation":"https://pith.science/pith/KN3WL7MYIKH35QMD5EOATROK2P/action/citation_signature","submit_replication":"https://pith.science/pith/KN3WL7MYIKH35QMD5EOATROK2P/action/replication_record"}},"created_at":"2026-07-05T10:17:50.275803+00:00","updated_at":"2026-07-05T10:17:50.275803+00:00"}