{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:TWB5BOGIVY3TV3YQ3SDH2VHYUU","short_pith_number":"pith:TWB5BOGI","schema_version":"1.0","canonical_sha256":"9d83d0b8c8ae373aef10dc867d54f8a51412f6a98502db720e7af5f8546cc909","source":{"kind":"arxiv","id":"2606.01563","version":1},"attestation_state":"computed","paper":{"title":"MomentKV: Closing the Directional Gap in KV Cache Eviction for Long-Context Inference","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Binxu Li, Tian Lan, Yu Li","submitted_at":"2026-06-01T02:08:40Z","abstract_excerpt":"Autoregressive decoding in Transformer-based language models relies on the KV cache, whose memory footprint grows linearly with sequence length and becomes the primary bottleneck for long-context inference. KV cache eviction addresses this by retaining a fixed-size subset of key-value pairs and discarding the rest. We identify that a primary source of output degradation is not the residual attention mass on evicted tokens, which existing methods already minimize, but a directional mismatch between the retained and evicted token sets. Specifically, the evicted tokens in practice are often near-"},"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":"2606.01563","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-01T02:08:40Z","cross_cats_sorted":[],"title_canon_sha256":"fb5ac4743532934749d830290e88bd1ba5d2ae4fe81b55d9486355df9e768f52","abstract_canon_sha256":"335a552bd1eda923c85640aaac2b804655ba279f1cbe93d7dab3344f9f56746d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T02:04:36.690097Z","signature_b64":"ReQ7oUNSOYayPnMOhMUQQL3HFbbhzPC2w0uByShuJEL7Mw08Ri8TQRJgF9T4gC4ECCHgEJR783XBoWd6Mv+kAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9d83d0b8c8ae373aef10dc867d54f8a51412f6a98502db720e7af5f8546cc909","last_reissued_at":"2026-06-02T02:04:36.689722Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T02:04:36.689722Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MomentKV: Closing the Directional Gap in KV Cache Eviction for Long-Context Inference","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Binxu Li, Tian Lan, Yu Li","submitted_at":"2026-06-01T02:08:40Z","abstract_excerpt":"Autoregressive decoding in Transformer-based language models relies on the KV cache, whose memory footprint grows linearly with sequence length and becomes the primary bottleneck for long-context inference. KV cache eviction addresses this by retaining a fixed-size subset of key-value pairs and discarding the rest. We identify that a primary source of output degradation is not the residual attention mass on evicted tokens, which existing methods already minimize, but a directional mismatch between the retained and evicted token sets. Specifically, the evicted tokens in practice are often near-"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.01563","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/2606.01563/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":"2606.01563","created_at":"2026-06-02T02:04:36.689787+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.01563v1","created_at":"2026-06-02T02:04:36.689787+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.01563","created_at":"2026-06-02T02:04:36.689787+00:00"},{"alias_kind":"pith_short_12","alias_value":"TWB5BOGIVY3T","created_at":"2026-06-02T02:04:36.689787+00:00"},{"alias_kind":"pith_short_16","alias_value":"TWB5BOGIVY3TV3YQ","created_at":"2026-06-02T02:04:36.689787+00:00"},{"alias_kind":"pith_short_8","alias_value":"TWB5BOGI","created_at":"2026-06-02T02:04:36.689787+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/TWB5BOGIVY3TV3YQ3SDH2VHYUU","json":"https://pith.science/pith/TWB5BOGIVY3TV3YQ3SDH2VHYUU.json","graph_json":"https://pith.science/api/pith-number/TWB5BOGIVY3TV3YQ3SDH2VHYUU/graph.json","events_json":"https://pith.science/api/pith-number/TWB5BOGIVY3TV3YQ3SDH2VHYUU/events.json","paper":"https://pith.science/paper/TWB5BOGI"},"agent_actions":{"view_html":"https://pith.science/pith/TWB5BOGIVY3TV3YQ3SDH2VHYUU","download_json":"https://pith.science/pith/TWB5BOGIVY3TV3YQ3SDH2VHYUU.json","view_paper":"https://pith.science/paper/TWB5BOGI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.01563&json=true","fetch_graph":"https://pith.science/api/pith-number/TWB5BOGIVY3TV3YQ3SDH2VHYUU/graph.json","fetch_events":"https://pith.science/api/pith-number/TWB5BOGIVY3TV3YQ3SDH2VHYUU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TWB5BOGIVY3TV3YQ3SDH2VHYUU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TWB5BOGIVY3TV3YQ3SDH2VHYUU/action/storage_attestation","attest_author":"https://pith.science/pith/TWB5BOGIVY3TV3YQ3SDH2VHYUU/action/author_attestation","sign_citation":"https://pith.science/pith/TWB5BOGIVY3TV3YQ3SDH2VHYUU/action/citation_signature","submit_replication":"https://pith.science/pith/TWB5BOGIVY3TV3YQ3SDH2VHYUU/action/replication_record"}},"created_at":"2026-06-02T02:04:36.689787+00:00","updated_at":"2026-06-02T02:04:36.689787+00:00"}