{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:OO3PRYZ4ANCE77YWHFNXT3JC6U","short_pith_number":"pith:OO3PRYZ4","schema_version":"1.0","canonical_sha256":"73b6f8e33c03444fff16395b79ed22f5370e5d35983dadde58f5fec87e28a3fc","source":{"kind":"arxiv","id":"2503.10742","version":2},"attestation_state":"computed","paper":{"title":"Keyframe-oriented Vision Token Pruning: Enhancing Efficiency of Large Vision Language Models on Long-Form Video Processing","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Hai Li, Jianyi Zhang, Jingwei Sun, Jingyang Zhang, Ming Yin, Qinsi Wang, Yiran Chen, Yudong Liu, Yueqian Lin","submitted_at":"2025-03-13T17:47:52Z","abstract_excerpt":"Vision language models (VLMs) demonstrate strong capabilities in jointly processing visual and textual data. However, they often incur substantial computational overhead due to redundant visual information, particularly in long-form video scenarios. Existing approaches predominantly focus on either vision token pruning, which may overlook spatio-temporal dependencies, or keyframe selection, which identifies informative frames but discards others, thus disrupting contextual continuity. In this work, we propose KVTP (Keyframe-oriented Vision Token Pruning), a novel framework that overcomes the d"},"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":"2503.10742","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-03-13T17:47:52Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"4c3164b4957ea558f9f7920a98ded82c6aa5a0a07fb015fb3df54608fc11a90e","abstract_canon_sha256":"8847bb6fbb79749909bcc4454a398a9073bb15bf2dfd51ec68fe976d6270624c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:53:16.411355Z","signature_b64":"Lx+QH9U5PbX6sWCEqjaWdJd3Pgbd/P0DJDvNHS67zvoKUZ+yDkjoHzIcW3fIz99jrWab8FP4CRVh9js8jOVtBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"73b6f8e33c03444fff16395b79ed22f5370e5d35983dadde58f5fec87e28a3fc","last_reissued_at":"2026-07-05T10:53:16.410847Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:53:16.410847Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Keyframe-oriented Vision Token Pruning: Enhancing Efficiency of Large Vision Language Models on Long-Form Video Processing","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Hai Li, Jianyi Zhang, Jingwei Sun, Jingyang Zhang, Ming Yin, Qinsi Wang, Yiran Chen, Yudong Liu, Yueqian Lin","submitted_at":"2025-03-13T17:47:52Z","abstract_excerpt":"Vision language models (VLMs) demonstrate strong capabilities in jointly processing visual and textual data. However, they often incur substantial computational overhead due to redundant visual information, particularly in long-form video scenarios. Existing approaches predominantly focus on either vision token pruning, which may overlook spatio-temporal dependencies, or keyframe selection, which identifies informative frames but discards others, thus disrupting contextual continuity. In this work, we propose KVTP (Keyframe-oriented Vision Token Pruning), a novel framework that overcomes the d"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.10742","kind":"arxiv","version":2},"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/2503.10742/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":"2503.10742","created_at":"2026-07-05T10:53:16.410908+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.10742v2","created_at":"2026-07-05T10:53:16.410908+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.10742","created_at":"2026-07-05T10:53:16.410908+00:00"},{"alias_kind":"pith_short_12","alias_value":"OO3PRYZ4ANCE","created_at":"2026-07-05T10:53:16.410908+00:00"},{"alias_kind":"pith_short_16","alias_value":"OO3PRYZ4ANCE77YW","created_at":"2026-07-05T10:53:16.410908+00:00"},{"alias_kind":"pith_short_8","alias_value":"OO3PRYZ4","created_at":"2026-07-05T10:53:16.410908+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.19235","citing_title":"CoreMatching: A Co-adaptive Sparse Inference Framework with Token and Neuron Pruning for Comprehensive Acceleration of Vision-Language Models","ref_index":8,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/OO3PRYZ4ANCE77YWHFNXT3JC6U","json":"https://pith.science/pith/OO3PRYZ4ANCE77YWHFNXT3JC6U.json","graph_json":"https://pith.science/api/pith-number/OO3PRYZ4ANCE77YWHFNXT3JC6U/graph.json","events_json":"https://pith.science/api/pith-number/OO3PRYZ4ANCE77YWHFNXT3JC6U/events.json","paper":"https://pith.science/paper/OO3PRYZ4"},"agent_actions":{"view_html":"https://pith.science/pith/OO3PRYZ4ANCE77YWHFNXT3JC6U","download_json":"https://pith.science/pith/OO3PRYZ4ANCE77YWHFNXT3JC6U.json","view_paper":"https://pith.science/paper/OO3PRYZ4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.10742&json=true","fetch_graph":"https://pith.science/api/pith-number/OO3PRYZ4ANCE77YWHFNXT3JC6U/graph.json","fetch_events":"https://pith.science/api/pith-number/OO3PRYZ4ANCE77YWHFNXT3JC6U/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/OO3PRYZ4ANCE77YWHFNXT3JC6U/action/timestamp_anchor","attest_storage":"https://pith.science/pith/OO3PRYZ4ANCE77YWHFNXT3JC6U/action/storage_attestation","attest_author":"https://pith.science/pith/OO3PRYZ4ANCE77YWHFNXT3JC6U/action/author_attestation","sign_citation":"https://pith.science/pith/OO3PRYZ4ANCE77YWHFNXT3JC6U/action/citation_signature","submit_replication":"https://pith.science/pith/OO3PRYZ4ANCE77YWHFNXT3JC6U/action/replication_record"}},"created_at":"2026-07-05T10:53:16.410908+00:00","updated_at":"2026-07-05T10:53:16.410908+00:00"}