{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:7JREJF2EI6J4SBSIMWF4MJAY6Z","short_pith_number":"pith:7JREJF2E","schema_version":"1.0","canonical_sha256":"fa624497444793c90648658bc62418f641e2cf4d3b1819973063cc313f6c6eb8","source":{"kind":"arxiv","id":"2606.01790","version":1},"attestation_state":"computed","paper":{"title":"STaR-KV: Spatio-Temporal Adaptive Re-weighting for KV Cache Compression in GUI Vision-Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Linfeng Zhang, Siteng Huang, Wenzheng Yang, Xiangqi Jin, Yaojie Zhang, Yuhang Han, Yujie Chen","submitted_at":"2026-06-01T07:10:00Z","abstract_excerpt":"Vision-language-model-based graphical user interface (GUI) agents have shown broad automation capabilities, yet deployment is bottlenecked by a key-value (KV) cache that grows linearly with interaction steps. For instance, UI-TARS-1.5-7B consumes 76 GB of GPU memory on merely five screenshots, approaching the capacity of mainstream 80 GB accelerators. Existing KV compression methods share two structural assumptions: aggregating visual-token importance into a single shared saliency map, and applying a fixed top-B cutoff to the fused score distribution. Pilot measurements refute both: spatial sp"},"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":"2606.01790","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-06-01T07:10:00Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"58c0a9b86b29e5fa63dec1a6f940300e4b6466e6bdee41546a2a80255413e660","abstract_canon_sha256":"17344f5eec8a57797bb1bf0352ae61ef627ea27968f10bc753bd5fc4fec927e4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T02:04:56.928724Z","signature_b64":"u29DWHNsxAQoWENt5PClrsR+z4S1hKbooEbZ2PVoQJ34CwT740YZNbjv/S1LPt3VaxO7fDnZN6oJ0gvDS1JOBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fa624497444793c90648658bc62418f641e2cf4d3b1819973063cc313f6c6eb8","last_reissued_at":"2026-06-02T02:04:56.928259Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T02:04:56.928259Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"STaR-KV: Spatio-Temporal Adaptive Re-weighting for KV Cache Compression in GUI Vision-Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Linfeng Zhang, Siteng Huang, Wenzheng Yang, Xiangqi Jin, Yaojie Zhang, Yuhang Han, Yujie Chen","submitted_at":"2026-06-01T07:10:00Z","abstract_excerpt":"Vision-language-model-based graphical user interface (GUI) agents have shown broad automation capabilities, yet deployment is bottlenecked by a key-value (KV) cache that grows linearly with interaction steps. For instance, UI-TARS-1.5-7B consumes 76 GB of GPU memory on merely five screenshots, approaching the capacity of mainstream 80 GB accelerators. Existing KV compression methods share two structural assumptions: aggregating visual-token importance into a single shared saliency map, and applying a fixed top-B cutoff to the fused score distribution. Pilot measurements refute both: spatial sp"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.01790","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/2606.01790/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":"2606.01790","created_at":"2026-06-02T02:04:56.928333+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.01790v1","created_at":"2026-06-02T02:04:56.928333+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.01790","created_at":"2026-06-02T02:04:56.928333+00:00"},{"alias_kind":"pith_short_12","alias_value":"7JREJF2EI6J4","created_at":"2026-06-02T02:04:56.928333+00:00"},{"alias_kind":"pith_short_16","alias_value":"7JREJF2EI6J4SBSI","created_at":"2026-06-02T02:04:56.928333+00:00"},{"alias_kind":"pith_short_8","alias_value":"7JREJF2E","created_at":"2026-06-02T02:04:56.928333+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/7JREJF2EI6J4SBSIMWF4MJAY6Z","json":"https://pith.science/pith/7JREJF2EI6J4SBSIMWF4MJAY6Z.json","graph_json":"https://pith.science/api/pith-number/7JREJF2EI6J4SBSIMWF4MJAY6Z/graph.json","events_json":"https://pith.science/api/pith-number/7JREJF2EI6J4SBSIMWF4MJAY6Z/events.json","paper":"https://pith.science/paper/7JREJF2E"},"agent_actions":{"view_html":"https://pith.science/pith/7JREJF2EI6J4SBSIMWF4MJAY6Z","download_json":"https://pith.science/pith/7JREJF2EI6J4SBSIMWF4MJAY6Z.json","view_paper":"https://pith.science/paper/7JREJF2E","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.01790&json=true","fetch_graph":"https://pith.science/api/pith-number/7JREJF2EI6J4SBSIMWF4MJAY6Z/graph.json","fetch_events":"https://pith.science/api/pith-number/7JREJF2EI6J4SBSIMWF4MJAY6Z/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7JREJF2EI6J4SBSIMWF4MJAY6Z/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7JREJF2EI6J4SBSIMWF4MJAY6Z/action/storage_attestation","attest_author":"https://pith.science/pith/7JREJF2EI6J4SBSIMWF4MJAY6Z/action/author_attestation","sign_citation":"https://pith.science/pith/7JREJF2EI6J4SBSIMWF4MJAY6Z/action/citation_signature","submit_replication":"https://pith.science/pith/7JREJF2EI6J4SBSIMWF4MJAY6Z/action/replication_record"}},"created_at":"2026-06-02T02:04:56.928333+00:00","updated_at":"2026-06-02T02:04:56.928333+00:00"}