{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:54ILOMC2WCXVKAB2A3C2E52PCE","short_pith_number":"pith:54ILOMC2","schema_version":"1.0","canonical_sha256":"ef10b7305ab0af55003a06c5a2774f1138f34100c0a40e30bdbf82ca3193011e","source":{"kind":"arxiv","id":"2605.19049","version":1},"attestation_state":"computed","paper":{"title":"KVBuffer: IO-aware Serving for Linear Attention","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Lin Zhong, Longwei Zou","submitted_at":"2026-05-18T19:14:52Z","abstract_excerpt":"Linear attention has recently gained significant attention for long-context inference due to its constant decoding cost with respect to context length. However, existing serving systems typically serve linear attention by recurrently computing and updating a large linear attention state in every decoding step. Since the state is much larger than the per-token key and value, recurrent decoding incurs substantial memory access and becomes inefficient for serving linear attention. In this paper, we propose KVBuffer, an IO-aware serving mechanism for linear attention. By buffering recent keys and "},"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":"2605.19049","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-05-18T19:14:52Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"1e03518078c447808dfcbf153f4abdc5cd70d527271244f6d735e4fb8b167433","abstract_canon_sha256":"110238b62d1e097b59c7b0aaa0a6db96536584bcfbb6acec79d06093d72bdb4e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T01:05:24.740044Z","signature_b64":"rGY8BuEmiPj0cz1SVvrCbUnfMUwXoPN7tTs706NPidhsHuYSWoyJWXsO4+k4rCIaajLniCRoIieBnLly757gDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ef10b7305ab0af55003a06c5a2774f1138f34100c0a40e30bdbf82ca3193011e","last_reissued_at":"2026-05-20T01:05:24.739502Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T01:05:24.739502Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"KVBuffer: IO-aware Serving for Linear Attention","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Lin Zhong, Longwei Zou","submitted_at":"2026-05-18T19:14:52Z","abstract_excerpt":"Linear attention has recently gained significant attention for long-context inference due to its constant decoding cost with respect to context length. However, existing serving systems typically serve linear attention by recurrently computing and updating a large linear attention state in every decoding step. Since the state is much larger than the per-token key and value, recurrent decoding incurs substantial memory access and becomes inefficient for serving linear attention. In this paper, we propose KVBuffer, an IO-aware serving mechanism for linear attention. By buffering recent keys and "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.19049","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/2605.19049/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":"2605.19049","created_at":"2026-05-20T01:05:24.739590+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.19049v1","created_at":"2026-05-20T01:05:24.739590+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.19049","created_at":"2026-05-20T01:05:24.739590+00:00"},{"alias_kind":"pith_short_12","alias_value":"54ILOMC2WCXV","created_at":"2026-05-20T01:05:24.739590+00:00"},{"alias_kind":"pith_short_16","alias_value":"54ILOMC2WCXVKAB2","created_at":"2026-05-20T01:05:24.739590+00:00"},{"alias_kind":"pith_short_8","alias_value":"54ILOMC2","created_at":"2026-05-20T01:05:24.739590+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/54ILOMC2WCXVKAB2A3C2E52PCE","json":"https://pith.science/pith/54ILOMC2WCXVKAB2A3C2E52PCE.json","graph_json":"https://pith.science/api/pith-number/54ILOMC2WCXVKAB2A3C2E52PCE/graph.json","events_json":"https://pith.science/api/pith-number/54ILOMC2WCXVKAB2A3C2E52PCE/events.json","paper":"https://pith.science/paper/54ILOMC2"},"agent_actions":{"view_html":"https://pith.science/pith/54ILOMC2WCXVKAB2A3C2E52PCE","download_json":"https://pith.science/pith/54ILOMC2WCXVKAB2A3C2E52PCE.json","view_paper":"https://pith.science/paper/54ILOMC2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.19049&json=true","fetch_graph":"https://pith.science/api/pith-number/54ILOMC2WCXVKAB2A3C2E52PCE/graph.json","fetch_events":"https://pith.science/api/pith-number/54ILOMC2WCXVKAB2A3C2E52PCE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/54ILOMC2WCXVKAB2A3C2E52PCE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/54ILOMC2WCXVKAB2A3C2E52PCE/action/storage_attestation","attest_author":"https://pith.science/pith/54ILOMC2WCXVKAB2A3C2E52PCE/action/author_attestation","sign_citation":"https://pith.science/pith/54ILOMC2WCXVKAB2A3C2E52PCE/action/citation_signature","submit_replication":"https://pith.science/pith/54ILOMC2WCXVKAB2A3C2E52PCE/action/replication_record"}},"created_at":"2026-05-20T01:05:24.739590+00:00","updated_at":"2026-05-20T01:05:24.739590+00:00"}