{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:BCQOY6CENZRFM2WPLCGG4VCDV5","short_pith_number":"pith:BCQOY6CE","schema_version":"1.0","canonical_sha256":"08a0ec78446e62566acf588c6e5443af5b77bdd859734ae7cf327ed60d3bd366","source":{"kind":"arxiv","id":"2606.29563","version":1},"attestation_state":"computed","paper":{"title":"Coverage-Driven KV Cache Eviction for Efficient and Improved Inference of LLM","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Golnoosh Samei, Hossein Hajimirsadeghi, Mengyao Zhai, Shuvendu Roy","submitted_at":"2026-06-28T19:04:47Z","abstract_excerpt":"Large language models (LLMs) excel at complex tasks like question answering and summarization, thanks to their ability to handle long-context inputs. However, deploying LLMs is costly, not only due to the high computational demands of quadratic complexity of self-attention and auto-regressive generation, but also because of the significant memory overhead required for storing the key-value (KV) cache during inference. To reduce the memory cost, existing KV-cache eviction strategies leverage the sparsity in attention to selectively store a subset of tokens. While reducing the memory footprint, "},"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":"2606.29563","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2026-06-28T19:04:47Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"038fa53d8fb55faa8fe3c00b9cdaa10b953279548c94d31e5bb26cb4e831ed30","abstract_canon_sha256":"70f2ba5e1b4ebbe9ee667164f0aa89f466d56a07a99eac6aaac8549d4108bbbe"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-30T01:18:12.110844Z","signature_b64":"8Mj4hHDZDJnQoL5xwwYVU/b7KuKKLNrLINI7U3EnmsNWcIZo7bk5InQCYihCPsLVJK1f9Arl/UO1NzyxZtRJDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"08a0ec78446e62566acf588c6e5443af5b77bdd859734ae7cf327ed60d3bd366","last_reissued_at":"2026-06-30T01:18:12.110253Z","signature_status":"signed_v1","first_computed_at":"2026-06-30T01:18:12.110253Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Coverage-Driven KV Cache Eviction for Efficient and Improved Inference of LLM","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Golnoosh Samei, Hossein Hajimirsadeghi, Mengyao Zhai, Shuvendu Roy","submitted_at":"2026-06-28T19:04:47Z","abstract_excerpt":"Large language models (LLMs) excel at complex tasks like question answering and summarization, thanks to their ability to handle long-context inputs. However, deploying LLMs is costly, not only due to the high computational demands of quadratic complexity of self-attention and auto-regressive generation, but also because of the significant memory overhead required for storing the key-value (KV) cache during inference. To reduce the memory cost, existing KV-cache eviction strategies leverage the sparsity in attention to selectively store a subset of tokens. While reducing the memory footprint, "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.29563","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/2606.29563/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":"2606.29563","created_at":"2026-06-30T01:18:12.110330+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.29563v1","created_at":"2026-06-30T01:18:12.110330+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.29563","created_at":"2026-06-30T01:18:12.110330+00:00"},{"alias_kind":"pith_short_12","alias_value":"BCQOY6CENZRF","created_at":"2026-06-30T01:18:12.110330+00:00"},{"alias_kind":"pith_short_16","alias_value":"BCQOY6CENZRFM2WP","created_at":"2026-06-30T01:18:12.110330+00:00"},{"alias_kind":"pith_short_8","alias_value":"BCQOY6CE","created_at":"2026-06-30T01:18:12.110330+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/BCQOY6CENZRFM2WPLCGG4VCDV5","json":"https://pith.science/pith/BCQOY6CENZRFM2WPLCGG4VCDV5.json","graph_json":"https://pith.science/api/pith-number/BCQOY6CENZRFM2WPLCGG4VCDV5/graph.json","events_json":"https://pith.science/api/pith-number/BCQOY6CENZRFM2WPLCGG4VCDV5/events.json","paper":"https://pith.science/paper/BCQOY6CE"},"agent_actions":{"view_html":"https://pith.science/pith/BCQOY6CENZRFM2WPLCGG4VCDV5","download_json":"https://pith.science/pith/BCQOY6CENZRFM2WPLCGG4VCDV5.json","view_paper":"https://pith.science/paper/BCQOY6CE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.29563&json=true","fetch_graph":"https://pith.science/api/pith-number/BCQOY6CENZRFM2WPLCGG4VCDV5/graph.json","fetch_events":"https://pith.science/api/pith-number/BCQOY6CENZRFM2WPLCGG4VCDV5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BCQOY6CENZRFM2WPLCGG4VCDV5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BCQOY6CENZRFM2WPLCGG4VCDV5/action/storage_attestation","attest_author":"https://pith.science/pith/BCQOY6CENZRFM2WPLCGG4VCDV5/action/author_attestation","sign_citation":"https://pith.science/pith/BCQOY6CENZRFM2WPLCGG4VCDV5/action/citation_signature","submit_replication":"https://pith.science/pith/BCQOY6CENZRFM2WPLCGG4VCDV5/action/replication_record"}},"created_at":"2026-06-30T01:18:12.110330+00:00","updated_at":"2026-06-30T01:18:12.110330+00:00"}