{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:OHA27JS5FPYGPE4GG2L6VQX77I","short_pith_number":"pith:OHA27JS5","schema_version":"1.0","canonical_sha256":"71c1afa65d2bf06793863697eac2fffa1d1e1cafa26d4b2a3ea10f4f07eabe0b","source":{"kind":"arxiv","id":"2507.14204","version":1},"attestation_state":"computed","paper":{"title":"LaCache: Ladder-Shaped KV Caching for Efficient Long-Context Modeling of Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Dachuan Shi, Haoran You, Jan Kautz, Pavlo Molchanov, Sixu Li, Xiangchi Yuan, Xin Dong, Yingyan (Celine) Lin, Yonggan Fu, Zhongzhi Yu","submitted_at":"2025-07-14T19:09:57Z","abstract_excerpt":"Recent advancements in Large Language Models (LLMs) have spurred interest in numerous applications requiring robust long-range capabilities, essential for processing extensive input contexts and continuously generating extended outputs. As sequence lengths increase, the number of Key-Value (KV) pairs in LLMs escalates, creating a significant efficiency bottleneck. In this paper, we propose a new KV cache optimization paradigm called LaCache, a training-free method for efficient and accurate generative inference of LLMs. LaCache enables LLMs to simultaneously address both of the critical challe"},"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":"2507.14204","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-07-14T19:09:57Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"5e56d39c01f49451687628b2b5ec1352b9d9458cd9463cbbcf42239ff5306d83","abstract_canon_sha256":"ce33ee135e3d1ec9fa4ab7187a5dfca6f4f211cd5f042061d9c5fe9a3dbf40bc"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:39:30.745428Z","signature_b64":"dvqsBOKbni5Rdm5CM5IDWXR+hMuSy+fpRWTPZtmmpjJJQsP3hD9AT4hqPG4AqimhZwuQQdQiLjFAHtq4I61wCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"71c1afa65d2bf06793863697eac2fffa1d1e1cafa26d4b2a3ea10f4f07eabe0b","last_reissued_at":"2026-07-05T11:39:30.744889Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:39:30.744889Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LaCache: Ladder-Shaped KV Caching for Efficient Long-Context Modeling of Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Dachuan Shi, Haoran You, Jan Kautz, Pavlo Molchanov, Sixu Li, Xiangchi Yuan, Xin Dong, Yingyan (Celine) Lin, Yonggan Fu, Zhongzhi Yu","submitted_at":"2025-07-14T19:09:57Z","abstract_excerpt":"Recent advancements in Large Language Models (LLMs) have spurred interest in numerous applications requiring robust long-range capabilities, essential for processing extensive input contexts and continuously generating extended outputs. As sequence lengths increase, the number of Key-Value (KV) pairs in LLMs escalates, creating a significant efficiency bottleneck. In this paper, we propose a new KV cache optimization paradigm called LaCache, a training-free method for efficient and accurate generative inference of LLMs. LaCache enables LLMs to simultaneously address both of the critical challe"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.14204","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/2507.14204/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":"2507.14204","created_at":"2026-07-05T11:39:30.744963+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.14204v1","created_at":"2026-07-05T11:39:30.744963+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.14204","created_at":"2026-07-05T11:39:30.744963+00:00"},{"alias_kind":"pith_short_12","alias_value":"OHA27JS5FPYG","created_at":"2026-07-05T11:39:30.744963+00:00"},{"alias_kind":"pith_short_16","alias_value":"OHA27JS5FPYGPE4G","created_at":"2026-07-05T11:39:30.744963+00:00"},{"alias_kind":"pith_short_8","alias_value":"OHA27JS5","created_at":"2026-07-05T11:39:30.744963+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.08840","citing_title":"ReST-KV: Robust KV Cache Eviction with Layer-wise Output Reconstruction and Spatial-Temporal Smoothing","ref_index":20,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/OHA27JS5FPYGPE4GG2L6VQX77I","json":"https://pith.science/pith/OHA27JS5FPYGPE4GG2L6VQX77I.json","graph_json":"https://pith.science/api/pith-number/OHA27JS5FPYGPE4GG2L6VQX77I/graph.json","events_json":"https://pith.science/api/pith-number/OHA27JS5FPYGPE4GG2L6VQX77I/events.json","paper":"https://pith.science/paper/OHA27JS5"},"agent_actions":{"view_html":"https://pith.science/pith/OHA27JS5FPYGPE4GG2L6VQX77I","download_json":"https://pith.science/pith/OHA27JS5FPYGPE4GG2L6VQX77I.json","view_paper":"https://pith.science/paper/OHA27JS5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.14204&json=true","fetch_graph":"https://pith.science/api/pith-number/OHA27JS5FPYGPE4GG2L6VQX77I/graph.json","fetch_events":"https://pith.science/api/pith-number/OHA27JS5FPYGPE4GG2L6VQX77I/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/OHA27JS5FPYGPE4GG2L6VQX77I/action/timestamp_anchor","attest_storage":"https://pith.science/pith/OHA27JS5FPYGPE4GG2L6VQX77I/action/storage_attestation","attest_author":"https://pith.science/pith/OHA27JS5FPYGPE4GG2L6VQX77I/action/author_attestation","sign_citation":"https://pith.science/pith/OHA27JS5FPYGPE4GG2L6VQX77I/action/citation_signature","submit_replication":"https://pith.science/pith/OHA27JS5FPYGPE4GG2L6VQX77I/action/replication_record"}},"created_at":"2026-07-05T11:39:30.744963+00:00","updated_at":"2026-07-05T11:39:30.744963+00:00"}