{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:SWTIWW5WFQOZTCKCLPI6J5P5GW","short_pith_number":"pith:SWTIWW5W","schema_version":"1.0","canonical_sha256":"95a68b5bb62c1d9989425bd1e4f5fd35b892e9ac04da0d0235c5d34208b523fe","source":{"kind":"arxiv","id":"2606.03075","version":1},"attestation_state":"computed","paper":{"title":"TGV-KV: Text-Grounded KV Eviction for Vision-Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Jizhihui Liu, Miao Zhang, Rui Shao, Ruizi Han, Weili Guan, Xuebo Liu, Yaowei Wang","submitted_at":"2026-06-02T03:06:17Z","abstract_excerpt":"Vision-Language Models (VLMs) inherit the auto-regressive generation paradigm and cache the keys and values (KV) of all previous tokens to accelerate inference, resulting in memory consumption that scales linearly with context length. This issue is particularly pronounced in VLMs due to substantial redundancy in the visual modality. Although KV cache eviction approaches can effectively reduce inference memory, they often incur significant performance degradation in VLMs, as most are designed for language models and overlook the inherent gap between text and vision. By systematically analyzing "},"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":"2606.03075","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-06-02T03:06:17Z","cross_cats_sorted":[],"title_canon_sha256":"e66e15e724615b653293447a9cb5262a1e1060a48620e927a6fd86daf2c312b0","abstract_canon_sha256":"c49cc5519838ddaa82c00540498f4374b14f8a39889da9fe66f993e90783507c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-03T01:05:30.963458Z","signature_b64":"5R5t5tlZvYtk7VLPwHOVPTYpodlJwJBO1BFROTvVcSyG55UA+KE9inLvElyqXKmN3v+ebASR6rGL6lrZ3t/aCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"95a68b5bb62c1d9989425bd1e4f5fd35b892e9ac04da0d0235c5d34208b523fe","last_reissued_at":"2026-06-03T01:05:30.963030Z","signature_status":"signed_v1","first_computed_at":"2026-06-03T01:05:30.963030Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TGV-KV: Text-Grounded KV Eviction for Vision-Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Jizhihui Liu, Miao Zhang, Rui Shao, Ruizi Han, Weili Guan, Xuebo Liu, Yaowei Wang","submitted_at":"2026-06-02T03:06:17Z","abstract_excerpt":"Vision-Language Models (VLMs) inherit the auto-regressive generation paradigm and cache the keys and values (KV) of all previous tokens to accelerate inference, resulting in memory consumption that scales linearly with context length. This issue is particularly pronounced in VLMs due to substantial redundancy in the visual modality. Although KV cache eviction approaches can effectively reduce inference memory, they often incur significant performance degradation in VLMs, as most are designed for language models and overlook the inherent gap between text and vision. By systematically analyzing "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.03075","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/2606.03075/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":"2606.03075","created_at":"2026-06-03T01:05:30.963093+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.03075v1","created_at":"2026-06-03T01:05:30.963093+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.03075","created_at":"2026-06-03T01:05:30.963093+00:00"},{"alias_kind":"pith_short_12","alias_value":"SWTIWW5WFQOZ","created_at":"2026-06-03T01:05:30.963093+00:00"},{"alias_kind":"pith_short_16","alias_value":"SWTIWW5WFQOZTCKC","created_at":"2026-06-03T01:05:30.963093+00:00"},{"alias_kind":"pith_short_8","alias_value":"SWTIWW5W","created_at":"2026-06-03T01:05:30.963093+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/SWTIWW5WFQOZTCKCLPI6J5P5GW","json":"https://pith.science/pith/SWTIWW5WFQOZTCKCLPI6J5P5GW.json","graph_json":"https://pith.science/api/pith-number/SWTIWW5WFQOZTCKCLPI6J5P5GW/graph.json","events_json":"https://pith.science/api/pith-number/SWTIWW5WFQOZTCKCLPI6J5P5GW/events.json","paper":"https://pith.science/paper/SWTIWW5W"},"agent_actions":{"view_html":"https://pith.science/pith/SWTIWW5WFQOZTCKCLPI6J5P5GW","download_json":"https://pith.science/pith/SWTIWW5WFQOZTCKCLPI6J5P5GW.json","view_paper":"https://pith.science/paper/SWTIWW5W","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.03075&json=true","fetch_graph":"https://pith.science/api/pith-number/SWTIWW5WFQOZTCKCLPI6J5P5GW/graph.json","fetch_events":"https://pith.science/api/pith-number/SWTIWW5WFQOZTCKCLPI6J5P5GW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/SWTIWW5WFQOZTCKCLPI6J5P5GW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/SWTIWW5WFQOZTCKCLPI6J5P5GW/action/storage_attestation","attest_author":"https://pith.science/pith/SWTIWW5WFQOZTCKCLPI6J5P5GW/action/author_attestation","sign_citation":"https://pith.science/pith/SWTIWW5WFQOZTCKCLPI6J5P5GW/action/citation_signature","submit_replication":"https://pith.science/pith/SWTIWW5WFQOZTCKCLPI6J5P5GW/action/replication_record"}},"created_at":"2026-06-03T01:05:30.963093+00:00","updated_at":"2026-06-03T01:05:30.963093+00:00"}