{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:GGKTXBSJILSLXQNT4SG5X5XJ4I","short_pith_number":"pith:GGKTXBSJ","schema_version":"1.0","canonical_sha256":"31953b864942e4bbc1b3e48ddbf6e9e20324b129bcd7f11772cee9869403676f","source":{"kind":"arxiv","id":"2507.08045","version":2},"attestation_state":"computed","paper":{"title":"Krul: Efficient State Restoration for Multi-turn Conversations with Dynamic Cross-layer KV Sharing","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Junyi Wen, Junyuan Liang, Ting Cai, Wuhui Chen, Zibin Zheng, Zicong Hong","submitted_at":"2025-07-10T01:51:17Z","abstract_excerpt":"Efficient state restoration in multi-turn conversations with large language models (LLMs) remains a critical challenge, primarily due to the overhead of recomputing or loading full key-value (KV) caches for all historical tokens. To address this, existing approaches compress KV caches across adjacent layers with highly similar attention patterns. However, these methods often apply a fixed compression scheme across all conversations, selecting the same layer pairs for compression without considering conversation-specific attention dynamics. This static strategy overlooks variability in attentio"},"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":"2507.08045","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2025-07-10T01:51:17Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"955392081962a4af484c3885998221eda9c20bc2f4d8557dd7383e3fa9377ade","abstract_canon_sha256":"e0445c99348d6a5bbcc2a093289491469b80057ee1dbc5788ae5a9ae13246986"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:59:17.575788Z","signature_b64":"YCuVEwQ8sJ1MFGd1MbVmQg+90TImBEXI3jt1fl0Rh4ZaG8Rx6LH03+B/CSv4H9CV5Ffr1gIasMZmzeF1vGldAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"31953b864942e4bbc1b3e48ddbf6e9e20324b129bcd7f11772cee9869403676f","last_reissued_at":"2026-07-05T11:59:17.575247Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:59:17.575247Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Krul: Efficient State Restoration for Multi-turn Conversations with Dynamic Cross-layer KV Sharing","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Junyi Wen, Junyuan Liang, Ting Cai, Wuhui Chen, Zibin Zheng, Zicong Hong","submitted_at":"2025-07-10T01:51:17Z","abstract_excerpt":"Efficient state restoration in multi-turn conversations with large language models (LLMs) remains a critical challenge, primarily due to the overhead of recomputing or loading full key-value (KV) caches for all historical tokens. To address this, existing approaches compress KV caches across adjacent layers with highly similar attention patterns. However, these methods often apply a fixed compression scheme across all conversations, selecting the same layer pairs for compression without considering conversation-specific attention dynamics. This static strategy overlooks variability in attentio"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.08045","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/2507.08045/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":"2507.08045","created_at":"2026-07-05T11:59:17.575323+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.08045v2","created_at":"2026-07-05T11:59:17.575323+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.08045","created_at":"2026-07-05T11:59:17.575323+00:00"},{"alias_kind":"pith_short_12","alias_value":"GGKTXBSJILSL","created_at":"2026-07-05T11:59:17.575323+00:00"},{"alias_kind":"pith_short_16","alias_value":"GGKTXBSJILSLXQNT","created_at":"2026-07-05T11:59:17.575323+00:00"},{"alias_kind":"pith_short_8","alias_value":"GGKTXBSJ","created_at":"2026-07-05T11:59:17.575323+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/GGKTXBSJILSLXQNT4SG5X5XJ4I","json":"https://pith.science/pith/GGKTXBSJILSLXQNT4SG5X5XJ4I.json","graph_json":"https://pith.science/api/pith-number/GGKTXBSJILSLXQNT4SG5X5XJ4I/graph.json","events_json":"https://pith.science/api/pith-number/GGKTXBSJILSLXQNT4SG5X5XJ4I/events.json","paper":"https://pith.science/paper/GGKTXBSJ"},"agent_actions":{"view_html":"https://pith.science/pith/GGKTXBSJILSLXQNT4SG5X5XJ4I","download_json":"https://pith.science/pith/GGKTXBSJILSLXQNT4SG5X5XJ4I.json","view_paper":"https://pith.science/paper/GGKTXBSJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.08045&json=true","fetch_graph":"https://pith.science/api/pith-number/GGKTXBSJILSLXQNT4SG5X5XJ4I/graph.json","fetch_events":"https://pith.science/api/pith-number/GGKTXBSJILSLXQNT4SG5X5XJ4I/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GGKTXBSJILSLXQNT4SG5X5XJ4I/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GGKTXBSJILSLXQNT4SG5X5XJ4I/action/storage_attestation","attest_author":"https://pith.science/pith/GGKTXBSJILSLXQNT4SG5X5XJ4I/action/author_attestation","sign_citation":"https://pith.science/pith/GGKTXBSJILSLXQNT4SG5X5XJ4I/action/citation_signature","submit_replication":"https://pith.science/pith/GGKTXBSJILSLXQNT4SG5X5XJ4I/action/replication_record"}},"created_at":"2026-07-05T11:59:17.575323+00:00","updated_at":"2026-07-05T11:59:17.575323+00:00"}