{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:PFFT726S2JCH7EO3XXS2KKGGVP","short_pith_number":"pith:PFFT726S","schema_version":"1.0","canonical_sha256":"794b3febd2d2447f91dbbde5a528c6abc6a3fbdfa46aeb510f474f5819107d61","source":{"kind":"arxiv","id":"2212.07016","version":2},"attestation_state":"computed","paper":{"title":"Understanding Zero-Shot Adversarial Robustness for Large-Scale Models","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Carl Vondrick, Chengzhi Mao, Junfeng Yang, Scott Geng, Xin Wang","submitted_at":"2022-12-14T04:08:56Z","abstract_excerpt":"Pretrained large-scale vision-language models like CLIP have exhibited strong generalization over unseen tasks. Yet imperceptible adversarial perturbations can significantly reduce CLIP's performance on new tasks. In this work, we identify and explore the problem of \\emph{adapting large-scale models for zero-shot adversarial robustness}. We first identify two key factors during model adaption -- training losses and adaptation methods -- that affect the model's zero-shot adversarial robustness. We then propose a text-guided contrastive adversarial training loss, which aligns the text embeddings"},"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":"2212.07016","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CV","submitted_at":"2022-12-14T04:08:56Z","cross_cats_sorted":[],"title_canon_sha256":"64f04c2575bd958366ef0778e68cc36dc1133e83b0ec77a09a48554a8f09a78b","abstract_canon_sha256":"bcef528c763c5c144ca08e484e74336059503aca45c65e39dbfe5f9ac58f0390"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:03:10.768114Z","signature_b64":"DWvcXLQwZ37RJJVa5x9ft/ClVxNu0MRriV0m60gSm1rWlAaJgST8Ch3vtJVj3tMtVrrYsNmfSSzysCfo/apNAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"794b3febd2d2447f91dbbde5a528c6abc6a3fbdfa46aeb510f474f5819107d61","last_reissued_at":"2026-07-05T06:03:10.767587Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:03:10.767587Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Understanding Zero-Shot Adversarial Robustness for Large-Scale Models","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Carl Vondrick, Chengzhi Mao, Junfeng Yang, Scott Geng, Xin Wang","submitted_at":"2022-12-14T04:08:56Z","abstract_excerpt":"Pretrained large-scale vision-language models like CLIP have exhibited strong generalization over unseen tasks. Yet imperceptible adversarial perturbations can significantly reduce CLIP's performance on new tasks. In this work, we identify and explore the problem of \\emph{adapting large-scale models for zero-shot adversarial robustness}. We first identify two key factors during model adaption -- training losses and adaptation methods -- that affect the model's zero-shot adversarial robustness. We then propose a text-guided contrastive adversarial training loss, which aligns the text embeddings"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2212.07016","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/2212.07016/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":"2212.07016","created_at":"2026-07-05T06:03:10.767660+00:00"},{"alias_kind":"arxiv_version","alias_value":"2212.07016v2","created_at":"2026-07-05T06:03:10.767660+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2212.07016","created_at":"2026-07-05T06:03:10.767660+00:00"},{"alias_kind":"pith_short_12","alias_value":"PFFT726S2JCH","created_at":"2026-07-05T06:03:10.767660+00:00"},{"alias_kind":"pith_short_16","alias_value":"PFFT726S2JCH7EO3","created_at":"2026-07-05T06:03:10.767660+00:00"},{"alias_kind":"pith_short_8","alias_value":"PFFT726S","created_at":"2026-07-05T06:03:10.767660+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":8,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.19584","citing_title":"Language-Instructed Vision Embeddings for Controllable and Generalizable Perception","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03730","citing_title":"Beyond False Stability: High-Noise Drift Gating for Test-Time Adversarial Defenses in Vision-Language Models","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03713","citing_title":"Investigating Adversarial Robustness of Multi-modal Large Language Models","ref_index":38,"is_internal_anchor":false},{"citing_arxiv_id":"2605.25922","citing_title":"Closed-Loop Bidirectional Prompting for Adversarial Robustness of Vision Language Models","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15584","citing_title":"AGC: Adaptive Geodesic Correction for Adversarial Robustness on Vision-Language Models","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2512.07222","citing_title":"Pay Less Attention to Function Words for Free Robustness of Vision-Language Models","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2604.12833","citing_title":"Challenging Vision-Language Models with Physically Deployable Multimodal Semantic Lighting Attacks","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2604.06440","citing_title":"Visual prompting reimagined: The power of the Activation Prompts","ref_index":26,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/PFFT726S2JCH7EO3XXS2KKGGVP","json":"https://pith.science/pith/PFFT726S2JCH7EO3XXS2KKGGVP.json","graph_json":"https://pith.science/api/pith-number/PFFT726S2JCH7EO3XXS2KKGGVP/graph.json","events_json":"https://pith.science/api/pith-number/PFFT726S2JCH7EO3XXS2KKGGVP/events.json","paper":"https://pith.science/paper/PFFT726S"},"agent_actions":{"view_html":"https://pith.science/pith/PFFT726S2JCH7EO3XXS2KKGGVP","download_json":"https://pith.science/pith/PFFT726S2JCH7EO3XXS2KKGGVP.json","view_paper":"https://pith.science/paper/PFFT726S","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2212.07016&json=true","fetch_graph":"https://pith.science/api/pith-number/PFFT726S2JCH7EO3XXS2KKGGVP/graph.json","fetch_events":"https://pith.science/api/pith-number/PFFT726S2JCH7EO3XXS2KKGGVP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PFFT726S2JCH7EO3XXS2KKGGVP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PFFT726S2JCH7EO3XXS2KKGGVP/action/storage_attestation","attest_author":"https://pith.science/pith/PFFT726S2JCH7EO3XXS2KKGGVP/action/author_attestation","sign_citation":"https://pith.science/pith/PFFT726S2JCH7EO3XXS2KKGGVP/action/citation_signature","submit_replication":"https://pith.science/pith/PFFT726S2JCH7EO3XXS2KKGGVP/action/replication_record"}},"created_at":"2026-07-05T06:03:10.767660+00:00","updated_at":"2026-07-05T06:03:10.767660+00:00"}