{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:OHB5BAZ2XAX3O2ULLGWX3H7GPA","short_pith_number":"pith:OHB5BAZ2","schema_version":"1.0","canonical_sha256":"71c3d0833ab82fb76a8b59ad7d9fe6780149b3824bce705f9648a0d41c0b50df","source":{"kind":"arxiv","id":"2607.02494","version":1},"attestation_state":"computed","paper":{"title":"Towards Robustness against Typographic Attack with Training-free Concept Localization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Aidong Zhang, Bohan Liu, Guangzhi Xiong, Sanchit Sinha, Wenqian Ye, Zhenghao He","submitted_at":"2026-07-02T17:55:24Z","abstract_excerpt":"Models trained via Contrastive Language-Image Pretraining (CLIP) serve as the foundational vision encoders for most modern Large Vision Language Models (LVLMs). Despite their widespread adoption, CLIP models exhibit a critical yet underexplored failure mode: irrelevant text appearing within images confounds visual representations, biasing them toward lexical meaning rather than true visual semantics. This robustness issue, commonly described as a Typographic Attack (TA), exposes a vulnerability that poses a significant risk to safety-critical applications such as autonomous driving. To achieve"},"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":"2607.02494","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-07-02T17:55:24Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"302962ea1b3fda239348b8e101b2a266c0e2a9553d3a6a1555e9eecda1d445ac","abstract_canon_sha256":"6c0df7d9fff60c4e583434a93b2ef0e101dc5894b4f7ed4576f1c23ea810d937"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-03T01:18:00.863912Z","signature_b64":"8ajjRnmyTvDaKlz5/kTT7kN4EzF9qtGFEE1cscrsfY5kBrr91ycx0dKIhhgKS60nUFW9Jin/IbbBM8gnJYEmCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"71c3d0833ab82fb76a8b59ad7d9fe6780149b3824bce705f9648a0d41c0b50df","last_reissued_at":"2026-07-03T01:18:00.863519Z","signature_status":"signed_v1","first_computed_at":"2026-07-03T01:18:00.863519Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Towards Robustness against Typographic Attack with Training-free Concept Localization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Aidong Zhang, Bohan Liu, Guangzhi Xiong, Sanchit Sinha, Wenqian Ye, Zhenghao He","submitted_at":"2026-07-02T17:55:24Z","abstract_excerpt":"Models trained via Contrastive Language-Image Pretraining (CLIP) serve as the foundational vision encoders for most modern Large Vision Language Models (LVLMs). Despite their widespread adoption, CLIP models exhibit a critical yet underexplored failure mode: irrelevant text appearing within images confounds visual representations, biasing them toward lexical meaning rather than true visual semantics. This robustness issue, commonly described as a Typographic Attack (TA), exposes a vulnerability that poses a significant risk to safety-critical applications such as autonomous driving. To achieve"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.02494","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/2607.02494/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":"2607.02494","created_at":"2026-07-03T01:18:00.863578+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.02494v1","created_at":"2026-07-03T01:18:00.863578+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.02494","created_at":"2026-07-03T01:18:00.863578+00:00"},{"alias_kind":"pith_short_12","alias_value":"OHB5BAZ2XAX3","created_at":"2026-07-03T01:18:00.863578+00:00"},{"alias_kind":"pith_short_16","alias_value":"OHB5BAZ2XAX3O2UL","created_at":"2026-07-03T01:18:00.863578+00:00"},{"alias_kind":"pith_short_8","alias_value":"OHB5BAZ2","created_at":"2026-07-03T01:18:00.863578+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/OHB5BAZ2XAX3O2ULLGWX3H7GPA","json":"https://pith.science/pith/OHB5BAZ2XAX3O2ULLGWX3H7GPA.json","graph_json":"https://pith.science/api/pith-number/OHB5BAZ2XAX3O2ULLGWX3H7GPA/graph.json","events_json":"https://pith.science/api/pith-number/OHB5BAZ2XAX3O2ULLGWX3H7GPA/events.json","paper":"https://pith.science/paper/OHB5BAZ2"},"agent_actions":{"view_html":"https://pith.science/pith/OHB5BAZ2XAX3O2ULLGWX3H7GPA","download_json":"https://pith.science/pith/OHB5BAZ2XAX3O2ULLGWX3H7GPA.json","view_paper":"https://pith.science/paper/OHB5BAZ2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.02494&json=true","fetch_graph":"https://pith.science/api/pith-number/OHB5BAZ2XAX3O2ULLGWX3H7GPA/graph.json","fetch_events":"https://pith.science/api/pith-number/OHB5BAZ2XAX3O2ULLGWX3H7GPA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/OHB5BAZ2XAX3O2ULLGWX3H7GPA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/OHB5BAZ2XAX3O2ULLGWX3H7GPA/action/storage_attestation","attest_author":"https://pith.science/pith/OHB5BAZ2XAX3O2ULLGWX3H7GPA/action/author_attestation","sign_citation":"https://pith.science/pith/OHB5BAZ2XAX3O2ULLGWX3H7GPA/action/citation_signature","submit_replication":"https://pith.science/pith/OHB5BAZ2XAX3O2ULLGWX3H7GPA/action/replication_record"}},"created_at":"2026-07-03T01:18:00.863578+00:00","updated_at":"2026-07-03T01:18:00.863578+00:00"}