{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:JTEVEPWHOAFN5TRRVEXHEIPZ22","short_pith_number":"pith:JTEVEPWH","schema_version":"1.0","canonical_sha256":"4cc9523ec7700adece31a92e7221f9d6ab049515b1e3c1c929874de42e60e49a","source":{"kind":"arxiv","id":"2606.07684","version":1},"attestation_state":"computed","paper":{"title":"Semantic Cache Distillation: Efficient State Transfer via Reuse and Selective Patching","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Hanshuai Cui, Qianli Ma, Weijia Jia, Zhiqing Tang, Zhi Yao","submitted_at":"2026-06-05T02:46:04Z","abstract_excerpt":"Disaggregated serving alleviates memory bottlenecks in Large Language Model (LLM) inference but creates a severe communication bottleneck: transmitting high-dimensional Key-Value (KV) caches often dominates time-to-first-token (TTFT). Moreover, reusing caches across heterogeneous models (e.g., base and fine-tuned variants) causes semantic misalignment that accumulates over layers, degrading generation quality. We propose Semantic Cache Distillation (SCD), a loss-constrained framework that replaces raw KV transmission with compact semantic codes. SCD addresses these challenges via two mechanism"},"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":"2606.07684","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-05T02:46:04Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"c417a3b6792738e7207c72c609b1c923a28e4cf854421d669e0aff1af770f323","abstract_canon_sha256":"6805bc9f7b0ae8e7c10f073866b977e6bd2a8c027a6e2b755525692d734c3dbc"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-09T00:04:46.768779Z","signature_b64":"26FoUeuGo+Ws8ETi9V6AIJebW9tEfjKpUY+nEIam/SDmORXy96pFI87DveqOycpCGAITFOOZmrwcWAVX34z8BA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4cc9523ec7700adece31a92e7221f9d6ab049515b1e3c1c929874de42e60e49a","last_reissued_at":"2026-06-09T00:04:46.768405Z","signature_status":"signed_v1","first_computed_at":"2026-06-09T00:04:46.768405Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Semantic Cache Distillation: Efficient State Transfer via Reuse and Selective Patching","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Hanshuai Cui, Qianli Ma, Weijia Jia, Zhiqing Tang, Zhi Yao","submitted_at":"2026-06-05T02:46:04Z","abstract_excerpt":"Disaggregated serving alleviates memory bottlenecks in Large Language Model (LLM) inference but creates a severe communication bottleneck: transmitting high-dimensional Key-Value (KV) caches often dominates time-to-first-token (TTFT). Moreover, reusing caches across heterogeneous models (e.g., base and fine-tuned variants) causes semantic misalignment that accumulates over layers, degrading generation quality. We propose Semantic Cache Distillation (SCD), a loss-constrained framework that replaces raw KV transmission with compact semantic codes. SCD addresses these challenges via two mechanism"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.07684","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/2606.07684/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":"2606.07684","created_at":"2026-06-09T00:04:46.768455+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.07684v1","created_at":"2026-06-09T00:04:46.768455+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.07684","created_at":"2026-06-09T00:04:46.768455+00:00"},{"alias_kind":"pith_short_12","alias_value":"JTEVEPWHOAFN","created_at":"2026-06-09T00:04:46.768455+00:00"},{"alias_kind":"pith_short_16","alias_value":"JTEVEPWHOAFN5TRR","created_at":"2026-06-09T00:04:46.768455+00:00"},{"alias_kind":"pith_short_8","alias_value":"JTEVEPWH","created_at":"2026-06-09T00:04:46.768455+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/JTEVEPWHOAFN5TRRVEXHEIPZ22","json":"https://pith.science/pith/JTEVEPWHOAFN5TRRVEXHEIPZ22.json","graph_json":"https://pith.science/api/pith-number/JTEVEPWHOAFN5TRRVEXHEIPZ22/graph.json","events_json":"https://pith.science/api/pith-number/JTEVEPWHOAFN5TRRVEXHEIPZ22/events.json","paper":"https://pith.science/paper/JTEVEPWH"},"agent_actions":{"view_html":"https://pith.science/pith/JTEVEPWHOAFN5TRRVEXHEIPZ22","download_json":"https://pith.science/pith/JTEVEPWHOAFN5TRRVEXHEIPZ22.json","view_paper":"https://pith.science/paper/JTEVEPWH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.07684&json=true","fetch_graph":"https://pith.science/api/pith-number/JTEVEPWHOAFN5TRRVEXHEIPZ22/graph.json","fetch_events":"https://pith.science/api/pith-number/JTEVEPWHOAFN5TRRVEXHEIPZ22/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JTEVEPWHOAFN5TRRVEXHEIPZ22/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JTEVEPWHOAFN5TRRVEXHEIPZ22/action/storage_attestation","attest_author":"https://pith.science/pith/JTEVEPWHOAFN5TRRVEXHEIPZ22/action/author_attestation","sign_citation":"https://pith.science/pith/JTEVEPWHOAFN5TRRVEXHEIPZ22/action/citation_signature","submit_replication":"https://pith.science/pith/JTEVEPWHOAFN5TRRVEXHEIPZ22/action/replication_record"}},"created_at":"2026-06-09T00:04:46.768455+00:00","updated_at":"2026-06-09T00:04:46.768455+00:00"}