{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:YR2MJWX2ZZTNR34FQUJT4UBWSE","short_pith_number":"pith:YR2MJWX2","schema_version":"1.0","canonical_sha256":"c474c4daface66d8ef8585133e503691051c54155a8b7aa8e7afec03ba2d6945","source":{"kind":"arxiv","id":"2503.21483","version":1},"attestation_state":"computed","paper":{"title":"BOLT: Boost Large Vision-Language Model Without Training for Long-form Video Understanding","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Bernard Ghanem, Chen Zhao, Shuming Liu, Tianqi Xu","submitted_at":"2025-03-27T13:18:40Z","abstract_excerpt":"Large video-language models (VLMs) have demonstrated promising progress in various video understanding tasks. However, their effectiveness in long-form video analysis is constrained by limited context windows. Traditional approaches, such as uniform frame sampling, often inevitably allocate resources to irrelevant content, diminishing their effectiveness in real-world scenarios. In this paper, we introduce BOLT, a method to BOost Large VLMs without additional Training through a comprehensive study of frame selection strategies. First, to enable a more realistic evaluation of VLMs in long-form "},"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":"2503.21483","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-03-27T13:18:40Z","cross_cats_sorted":[],"title_canon_sha256":"5a51bd00b66161321df12a74a94a6534edd070053ca8c6a84298189141eb3d69","abstract_canon_sha256":"852cdaddb0c2045229397f5e14fb21cb458b992c098db2895f854af2117f0734"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:40:26.134108Z","signature_b64":"uae7atSYIY4wO+JfIAHCr5IzG/qZcS+wqSt5sTeAKvmep1oVhPwhnX20za1GKfppDsiMA0oXl+WlzUG6uD1VAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c474c4daface66d8ef8585133e503691051c54155a8b7aa8e7afec03ba2d6945","last_reissued_at":"2026-07-05T10:40:26.133624Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:40:26.133624Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"BOLT: Boost Large Vision-Language Model Without Training for Long-form Video Understanding","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Bernard Ghanem, Chen Zhao, Shuming Liu, Tianqi Xu","submitted_at":"2025-03-27T13:18:40Z","abstract_excerpt":"Large video-language models (VLMs) have demonstrated promising progress in various video understanding tasks. However, their effectiveness in long-form video analysis is constrained by limited context windows. Traditional approaches, such as uniform frame sampling, often inevitably allocate resources to irrelevant content, diminishing their effectiveness in real-world scenarios. In this paper, we introduce BOLT, a method to BOost Large VLMs without additional Training through a comprehensive study of frame selection strategies. First, to enable a more realistic evaluation of VLMs in long-form "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.21483","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/2503.21483/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":"2503.21483","created_at":"2026-07-05T10:40:26.133681+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.21483v1","created_at":"2026-07-05T10:40:26.133681+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.21483","created_at":"2026-07-05T10:40:26.133681+00:00"},{"alias_kind":"pith_short_12","alias_value":"YR2MJWX2ZZTN","created_at":"2026-07-05T10:40:26.133681+00:00"},{"alias_kind":"pith_short_16","alias_value":"YR2MJWX2ZZTNR34F","created_at":"2026-07-05T10:40:26.133681+00:00"},{"alias_kind":"pith_short_8","alias_value":"YR2MJWX2","created_at":"2026-07-05T10:40:26.133681+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.02371","citing_title":"Internalized Reasoning for Long-Context Visual Document Understanding","ref_index":31,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YR2MJWX2ZZTNR34FQUJT4UBWSE","json":"https://pith.science/pith/YR2MJWX2ZZTNR34FQUJT4UBWSE.json","graph_json":"https://pith.science/api/pith-number/YR2MJWX2ZZTNR34FQUJT4UBWSE/graph.json","events_json":"https://pith.science/api/pith-number/YR2MJWX2ZZTNR34FQUJT4UBWSE/events.json","paper":"https://pith.science/paper/YR2MJWX2"},"agent_actions":{"view_html":"https://pith.science/pith/YR2MJWX2ZZTNR34FQUJT4UBWSE","download_json":"https://pith.science/pith/YR2MJWX2ZZTNR34FQUJT4UBWSE.json","view_paper":"https://pith.science/paper/YR2MJWX2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.21483&json=true","fetch_graph":"https://pith.science/api/pith-number/YR2MJWX2ZZTNR34FQUJT4UBWSE/graph.json","fetch_events":"https://pith.science/api/pith-number/YR2MJWX2ZZTNR34FQUJT4UBWSE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YR2MJWX2ZZTNR34FQUJT4UBWSE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YR2MJWX2ZZTNR34FQUJT4UBWSE/action/storage_attestation","attest_author":"https://pith.science/pith/YR2MJWX2ZZTNR34FQUJT4UBWSE/action/author_attestation","sign_citation":"https://pith.science/pith/YR2MJWX2ZZTNR34FQUJT4UBWSE/action/citation_signature","submit_replication":"https://pith.science/pith/YR2MJWX2ZZTNR34FQUJT4UBWSE/action/replication_record"}},"created_at":"2026-07-05T10:40:26.133681+00:00","updated_at":"2026-07-05T10:40:26.133681+00:00"}