{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:EBOAK5KBXERD6FTWDYAJED3K5P","short_pith_number":"pith:EBOAK5KB","schema_version":"1.0","canonical_sha256":"205c057541b9223f16761e00920f6aebea877188030afe02d7d964290829a3c5","source":{"kind":"arxiv","id":"2403.09346","version":2},"attestation_state":"computed","paper":{"title":"B-AVIBench: Towards Evaluating the Robustness of Large Vision-Language Model on Black-box Adversarial Visual-Instructions","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Hao Zhang, Hong Liu, Kaipeng Zhang, Nanning Zheng, Ping Luo, Wenqi Shao, Yongqiang Ma, Yu Qiao","submitted_at":"2024-03-14T12:51:07Z","abstract_excerpt":"Large Vision-Language Models (LVLMs) have shown significant progress in responding well to visual-instructions from users. However, these instructions, encompassing images and text, are susceptible to both intentional and inadvertent attacks. Despite the critical importance of LVLMs' robustness against such threats, current research in this area remains limited. To bridge this gap, we introduce B-AVIBench, a framework designed to analyze the robustness of LVLMs when facing various Black-box Adversarial Visual-Instructions (B-AVIs), including four types of image-based B-AVIs, ten types of text-"},"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":"2403.09346","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2024-03-14T12:51:07Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"be8c21c2e87a0d7abfebef1d434e418c80e971209307cb230b1fb9d809c0b783","abstract_canon_sha256":"1ee99c16f01449fbcfc4647c63ebc497879fba4ef5b9df508ff7137f4a736e9a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:54:07.869606Z","signature_b64":"YAD4DTas3SDdvj0yTNB0FIBmmR6+DtWbJE3fY3ihQMb2lbta2aFETpDHzNOTdeYpvy0LyAk/u7s0MtnCp3c2Dg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"205c057541b9223f16761e00920f6aebea877188030afe02d7d964290829a3c5","last_reissued_at":"2026-07-05T11:54:07.869159Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:54:07.869159Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"B-AVIBench: Towards Evaluating the Robustness of Large Vision-Language Model on Black-box Adversarial Visual-Instructions","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Hao Zhang, Hong Liu, Kaipeng Zhang, Nanning Zheng, Ping Luo, Wenqi Shao, Yongqiang Ma, Yu Qiao","submitted_at":"2024-03-14T12:51:07Z","abstract_excerpt":"Large Vision-Language Models (LVLMs) have shown significant progress in responding well to visual-instructions from users. However, these instructions, encompassing images and text, are susceptible to both intentional and inadvertent attacks. Despite the critical importance of LVLMs' robustness against such threats, current research in this area remains limited. To bridge this gap, we introduce B-AVIBench, a framework designed to analyze the robustness of LVLMs when facing various Black-box Adversarial Visual-Instructions (B-AVIs), including four types of image-based B-AVIs, ten types of text-"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2403.09346","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/2403.09346/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":"2403.09346","created_at":"2026-07-05T11:54:07.869214+00:00"},{"alias_kind":"arxiv_version","alias_value":"2403.09346v2","created_at":"2026-07-05T11:54:07.869214+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2403.09346","created_at":"2026-07-05T11:54:07.869214+00:00"},{"alias_kind":"pith_short_12","alias_value":"EBOAK5KBXERD","created_at":"2026-07-05T11:54:07.869214+00:00"},{"alias_kind":"pith_short_16","alias_value":"EBOAK5KBXERD6FTW","created_at":"2026-07-05T11:54:07.869214+00:00"},{"alias_kind":"pith_short_8","alias_value":"EBOAK5KB","created_at":"2026-07-05T11:54:07.869214+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2406.14194","citing_title":"VLBiasBench: A Comprehensive Benchmark for Evaluating Bias in Large Vision-Language Model","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2512.21815","citing_title":"High-Entropy Tokens as Multimodal Failure Points in Vision-Language Models","ref_index":51,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/EBOAK5KBXERD6FTWDYAJED3K5P","json":"https://pith.science/pith/EBOAK5KBXERD6FTWDYAJED3K5P.json","graph_json":"https://pith.science/api/pith-number/EBOAK5KBXERD6FTWDYAJED3K5P/graph.json","events_json":"https://pith.science/api/pith-number/EBOAK5KBXERD6FTWDYAJED3K5P/events.json","paper":"https://pith.science/paper/EBOAK5KB"},"agent_actions":{"view_html":"https://pith.science/pith/EBOAK5KBXERD6FTWDYAJED3K5P","download_json":"https://pith.science/pith/EBOAK5KBXERD6FTWDYAJED3K5P.json","view_paper":"https://pith.science/paper/EBOAK5KB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2403.09346&json=true","fetch_graph":"https://pith.science/api/pith-number/EBOAK5KBXERD6FTWDYAJED3K5P/graph.json","fetch_events":"https://pith.science/api/pith-number/EBOAK5KBXERD6FTWDYAJED3K5P/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/EBOAK5KBXERD6FTWDYAJED3K5P/action/timestamp_anchor","attest_storage":"https://pith.science/pith/EBOAK5KBXERD6FTWDYAJED3K5P/action/storage_attestation","attest_author":"https://pith.science/pith/EBOAK5KBXERD6FTWDYAJED3K5P/action/author_attestation","sign_citation":"https://pith.science/pith/EBOAK5KBXERD6FTWDYAJED3K5P/action/citation_signature","submit_replication":"https://pith.science/pith/EBOAK5KBXERD6FTWDYAJED3K5P/action/replication_record"}},"created_at":"2026-07-05T11:54:07.869214+00:00","updated_at":"2026-07-05T11:54:07.869214+00:00"}