{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:VUD7OM6QMRD67TAVPKPACSIKX4","short_pith_number":"pith:VUD7OM6Q","schema_version":"1.0","canonical_sha256":"ad07f733d06447efcc157a9e01490abf2291155d5c6ae93dfddf985166d4501c","source":{"kind":"arxiv","id":"2403.04643","version":2},"attestation_state":"computed","paper":{"title":"QAQ: Quality Adaptive Quantization for LLM KV Cache","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Jiayu Qin, Shichen Dong, Wei Wang, Wen Cheng","submitted_at":"2024-03-07T16:42:37Z","abstract_excerpt":"The emergence of LLMs has ignited a fresh surge of breakthroughs in NLP applications, particularly in domains such as question-answering systems and text generation. As the need for longer context grows, a significant bottleneck in model deployment emerges due to the linear expansion of the Key-Value (KV) cache with the context length. Existing methods primarily rely on various hypotheses, such as sorting the KV cache based on attention scores for replacement or eviction, to compress the KV cache and improve model throughput. However, heuristics used by these strategies may wrongly evict essen"},"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":"2403.04643","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2024-03-07T16:42:37Z","cross_cats_sorted":[],"title_canon_sha256":"4dbc0865d79fe30f5c843c5b43015fa52aafd6c0f15d6bd686c966cdb964c717","abstract_canon_sha256":"6ce304b989dc4189fc3b2487a2074d40576379814a603d1da1e92faec8fa1103"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:07:14.982142Z","signature_b64":"FBbfAA2bDTR7fIxOhOSkDXnuGFtqAjTwgaRm53eh2BEi60t398UOuRgXMeoj91qap8E3Q48hlF037FpOhcJPDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ad07f733d06447efcc157a9e01490abf2291155d5c6ae93dfddf985166d4501c","last_reissued_at":"2026-07-05T08:07:14.981725Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:07:14.981725Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"QAQ: Quality Adaptive Quantization for LLM KV Cache","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Jiayu Qin, Shichen Dong, Wei Wang, Wen Cheng","submitted_at":"2024-03-07T16:42:37Z","abstract_excerpt":"The emergence of LLMs has ignited a fresh surge of breakthroughs in NLP applications, particularly in domains such as question-answering systems and text generation. As the need for longer context grows, a significant bottleneck in model deployment emerges due to the linear expansion of the Key-Value (KV) cache with the context length. Existing methods primarily rely on various hypotheses, such as sorting the KV cache based on attention scores for replacement or eviction, to compress the KV cache and improve model throughput. However, heuristics used by these strategies may wrongly evict essen"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2403.04643","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/2403.04643/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":"2403.04643","created_at":"2026-07-05T08:07:14.981782+00:00"},{"alias_kind":"arxiv_version","alias_value":"2403.04643v2","created_at":"2026-07-05T08:07:14.981782+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2403.04643","created_at":"2026-07-05T08:07:14.981782+00:00"},{"alias_kind":"pith_short_12","alias_value":"VUD7OM6QMRD6","created_at":"2026-07-05T08:07:14.981782+00:00"},{"alias_kind":"pith_short_16","alias_value":"VUD7OM6QMRD67TAV","created_at":"2026-07-05T08:07:14.981782+00:00"},{"alias_kind":"pith_short_8","alias_value":"VUD7OM6Q","created_at":"2026-07-05T08:07:14.981782+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":8,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.08057","citing_title":"Towards Efficient Large Language Model Serving: A Survey on System-Aware KV Cache Optimization","ref_index":2,"is_internal_anchor":true},{"citing_arxiv_id":"2503.19950","citing_title":"LogQuant: Log-Distributed 2-Bit Quantization of KV Cache with Superior Accuracy Preservation","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14292","citing_title":"Minimal-Intervention KV Retention via Set-Conditioned Diversity","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2504.19874","citing_title":"TurboQuant: Online Vector Quantization with Near-optimal Distortion Rate","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2407.11550","citing_title":"Ada-KV: Optimizing KV Cache Eviction by Adaptive Budget Allocation for Efficient LLM Inference","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2603.22910","citing_title":"EchoKV: Efficient KV Cache Compression via Similarity-Based Reconstruction","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08317","citing_title":"RDKV: Rate-Distortion Bit Allocation for Joint Eviction and Quantization of the KV Cache","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2605.02262","citing_title":"WindowQuant: Mixed-Precision KV Cache Quantization based on Window-Level Similarity for VLMs Inference Optimization","ref_index":12,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VUD7OM6QMRD67TAVPKPACSIKX4","json":"https://pith.science/pith/VUD7OM6QMRD67TAVPKPACSIKX4.json","graph_json":"https://pith.science/api/pith-number/VUD7OM6QMRD67TAVPKPACSIKX4/graph.json","events_json":"https://pith.science/api/pith-number/VUD7OM6QMRD67TAVPKPACSIKX4/events.json","paper":"https://pith.science/paper/VUD7OM6Q"},"agent_actions":{"view_html":"https://pith.science/pith/VUD7OM6QMRD67TAVPKPACSIKX4","download_json":"https://pith.science/pith/VUD7OM6QMRD67TAVPKPACSIKX4.json","view_paper":"https://pith.science/paper/VUD7OM6Q","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2403.04643&json=true","fetch_graph":"https://pith.science/api/pith-number/VUD7OM6QMRD67TAVPKPACSIKX4/graph.json","fetch_events":"https://pith.science/api/pith-number/VUD7OM6QMRD67TAVPKPACSIKX4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VUD7OM6QMRD67TAVPKPACSIKX4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VUD7OM6QMRD67TAVPKPACSIKX4/action/storage_attestation","attest_author":"https://pith.science/pith/VUD7OM6QMRD67TAVPKPACSIKX4/action/author_attestation","sign_citation":"https://pith.science/pith/VUD7OM6QMRD67TAVPKPACSIKX4/action/citation_signature","submit_replication":"https://pith.science/pith/VUD7OM6QMRD67TAVPKPACSIKX4/action/replication_record"}},"created_at":"2026-07-05T08:07:14.981782+00:00","updated_at":"2026-07-05T08:07:14.981782+00:00"}