{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:2KIHN6SGUFV3CSYMIXWMXUMOOE","short_pith_number":"pith:2KIHN6SG","schema_version":"1.0","canonical_sha256":"d29076fa46a16bb14b0c45eccbd18e711322a00328c3d0cccabe58d558dd9ac8","source":{"kind":"arxiv","id":"2507.20030","version":1},"attestation_state":"computed","paper":{"title":"FAEDKV: Infinite-Window Fourier Transform for Unbiased KV Cache Compression","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Haotian Yu, Mu Sheng, Pan Li, Runchao Li, Xianxuan Long, Yao Fu","submitted_at":"2025-07-26T18:20:25Z","abstract_excerpt":"The efficacy of Large Language Models (LLMs) in long-context tasks is often hampered by the substantial memory footprint and computational demands of the Key-Value (KV) cache. Current compression strategies, including token eviction and learned projections, frequently lead to biased representations -- either by overemphasizing recent/high-attention tokens or by repeatedly degrading information from earlier context -- and may require costly model retraining. We present FAEDKV (Frequency-Adaptive Infinite-Window for KV cache), a novel, training-free KV cache compression framework that ensures un"},"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":"2507.20030","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-07-26T18:20:25Z","cross_cats_sorted":[],"title_canon_sha256":"1a8c2dd3eb585e062771348ae22489f97a24a9b919df6b015f5f9d3e33a7b999","abstract_canon_sha256":"bf908df3c7afa4be9d72de7102184cfcac5ce338e6b3917ce3e71d1a3ff0f1b7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:43:46.949892Z","signature_b64":"RsHLEeHPhrLysoJjqz/dxrZoLHrd1q4gv9bWQCXHJavZkGMJr39zWZ/Lk09orSBcoJzcclJwWijN7XWI652/AQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d29076fa46a16bb14b0c45eccbd18e711322a00328c3d0cccabe58d558dd9ac8","last_reissued_at":"2026-07-05T11:43:46.949364Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:43:46.949364Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"FAEDKV: Infinite-Window Fourier Transform for Unbiased KV Cache Compression","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Haotian Yu, Mu Sheng, Pan Li, Runchao Li, Xianxuan Long, Yao Fu","submitted_at":"2025-07-26T18:20:25Z","abstract_excerpt":"The efficacy of Large Language Models (LLMs) in long-context tasks is often hampered by the substantial memory footprint and computational demands of the Key-Value (KV) cache. Current compression strategies, including token eviction and learned projections, frequently lead to biased representations -- either by overemphasizing recent/high-attention tokens or by repeatedly degrading information from earlier context -- and may require costly model retraining. We present FAEDKV (Frequency-Adaptive Infinite-Window for KV cache), a novel, training-free KV cache compression framework that ensures un"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.20030","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/2507.20030/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":"2507.20030","created_at":"2026-07-05T11:43:46.949424+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.20030v1","created_at":"2026-07-05T11:43:46.949424+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.20030","created_at":"2026-07-05T11:43:46.949424+00:00"},{"alias_kind":"pith_short_12","alias_value":"2KIHN6SGUFV3","created_at":"2026-07-05T11:43:46.949424+00:00"},{"alias_kind":"pith_short_16","alias_value":"2KIHN6SGUFV3CSYM","created_at":"2026-07-05T11:43:46.949424+00:00"},{"alias_kind":"pith_short_8","alias_value":"2KIHN6SG","created_at":"2026-07-05T11:43:46.949424+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.09862","citing_title":"Blurry Window Attention","ref_index":39,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2KIHN6SGUFV3CSYMIXWMXUMOOE","json":"https://pith.science/pith/2KIHN6SGUFV3CSYMIXWMXUMOOE.json","graph_json":"https://pith.science/api/pith-number/2KIHN6SGUFV3CSYMIXWMXUMOOE/graph.json","events_json":"https://pith.science/api/pith-number/2KIHN6SGUFV3CSYMIXWMXUMOOE/events.json","paper":"https://pith.science/paper/2KIHN6SG"},"agent_actions":{"view_html":"https://pith.science/pith/2KIHN6SGUFV3CSYMIXWMXUMOOE","download_json":"https://pith.science/pith/2KIHN6SGUFV3CSYMIXWMXUMOOE.json","view_paper":"https://pith.science/paper/2KIHN6SG","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.20030&json=true","fetch_graph":"https://pith.science/api/pith-number/2KIHN6SGUFV3CSYMIXWMXUMOOE/graph.json","fetch_events":"https://pith.science/api/pith-number/2KIHN6SGUFV3CSYMIXWMXUMOOE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2KIHN6SGUFV3CSYMIXWMXUMOOE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2KIHN6SGUFV3CSYMIXWMXUMOOE/action/storage_attestation","attest_author":"https://pith.science/pith/2KIHN6SGUFV3CSYMIXWMXUMOOE/action/author_attestation","sign_citation":"https://pith.science/pith/2KIHN6SGUFV3CSYMIXWMXUMOOE/action/citation_signature","submit_replication":"https://pith.science/pith/2KIHN6SGUFV3CSYMIXWMXUMOOE/action/replication_record"}},"created_at":"2026-07-05T11:43:46.949424+00:00","updated_at":"2026-07-05T11:43:46.949424+00:00"}