{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:Z2SWVFT5IPSA5EHO6DCULKZNP4","short_pith_number":"pith:Z2SWVFT5","schema_version":"1.0","canonical_sha256":"cea56a967d43e40e90eef0c545ab2d7f049e82ff23453362caee163a5ab32c96","source":{"kind":"arxiv","id":"2212.08044","version":3},"attestation_state":"computed","paper":{"title":"Benchmarking Robustness of Multimodal Image-Text Models under Distribution Shift","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Bo Li, Ding Zhao, Florian Wenzel, Jielin Qiu, Mu Li, Xingjian Shi, Yi Zhu, Zhiqiang Tang","submitted_at":"2022-12-15T18:52:03Z","abstract_excerpt":"Multimodal image-text models have shown remarkable performance in the past few years. However, evaluating robustness against distribution shifts is crucial before adopting them in real-world applications. In this work, we investigate the robustness of 12 popular open-sourced image-text models under common perturbations on five tasks (image-text retrieval, visual reasoning, visual entailment, image captioning, and text-to-image generation). In particular, we propose several new multimodal robustness benchmarks by applying 17 image perturbation and 16 text perturbation techniques on top of exist"},"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":"2212.08044","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2022-12-15T18:52:03Z","cross_cats_sorted":[],"title_canon_sha256":"96c0e86b9eca1e8e0c7d790f7d8440f8ef3815c5173ef813a82e5b8fb156c983","abstract_canon_sha256":"c7f9fe813715de8429a2968995a6445f58fdc6e9f60292c689707e1a511e335d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:35:17.403503Z","signature_b64":"0LAkHWP5pyM6UncnnERnr1zK8v0NMgCfbcuKPBUFUAMEpqWtrs0qqEso0E1ECWgZNo3mAir6HOLz7KtsXfgeAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"cea56a967d43e40e90eef0c545ab2d7f049e82ff23453362caee163a5ab32c96","last_reissued_at":"2026-07-05T07:35:17.403006Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:35:17.403006Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Benchmarking Robustness of Multimodal Image-Text Models under Distribution Shift","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Bo Li, Ding Zhao, Florian Wenzel, Jielin Qiu, Mu Li, Xingjian Shi, Yi Zhu, Zhiqiang Tang","submitted_at":"2022-12-15T18:52:03Z","abstract_excerpt":"Multimodal image-text models have shown remarkable performance in the past few years. However, evaluating robustness against distribution shifts is crucial before adopting them in real-world applications. In this work, we investigate the robustness of 12 popular open-sourced image-text models under common perturbations on five tasks (image-text retrieval, visual reasoning, visual entailment, image captioning, and text-to-image generation). In particular, we propose several new multimodal robustness benchmarks by applying 17 image perturbation and 16 text perturbation techniques on top of exist"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2212.08044","kind":"arxiv","version":3},"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/2212.08044/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":"2212.08044","created_at":"2026-07-05T07:35:17.403068+00:00"},{"alias_kind":"arxiv_version","alias_value":"2212.08044v3","created_at":"2026-07-05T07:35:17.403068+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2212.08044","created_at":"2026-07-05T07:35:17.403068+00:00"},{"alias_kind":"pith_short_12","alias_value":"Z2SWVFT5IPSA","created_at":"2026-07-05T07:35:17.403068+00:00"},{"alias_kind":"pith_short_16","alias_value":"Z2SWVFT5IPSA5EHO","created_at":"2026-07-05T07:35:17.403068+00:00"},{"alias_kind":"pith_short_8","alias_value":"Z2SWVFT5","created_at":"2026-07-05T07:35:17.403068+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.08894","citing_title":"Are Reasoning Vision-Language Models Robust to Semantic Visual Distractions?","ref_index":43,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10009","citing_title":"Hystar: Hypernetwork-driven Style-adaptive Retrieval via Dynamic SVD Modulation","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09836","citing_title":"ReCoVR: Closing the Loop in Interactive Composed Video Retrieval","ref_index":67,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/Z2SWVFT5IPSA5EHO6DCULKZNP4","json":"https://pith.science/pith/Z2SWVFT5IPSA5EHO6DCULKZNP4.json","graph_json":"https://pith.science/api/pith-number/Z2SWVFT5IPSA5EHO6DCULKZNP4/graph.json","events_json":"https://pith.science/api/pith-number/Z2SWVFT5IPSA5EHO6DCULKZNP4/events.json","paper":"https://pith.science/paper/Z2SWVFT5"},"agent_actions":{"view_html":"https://pith.science/pith/Z2SWVFT5IPSA5EHO6DCULKZNP4","download_json":"https://pith.science/pith/Z2SWVFT5IPSA5EHO6DCULKZNP4.json","view_paper":"https://pith.science/paper/Z2SWVFT5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2212.08044&json=true","fetch_graph":"https://pith.science/api/pith-number/Z2SWVFT5IPSA5EHO6DCULKZNP4/graph.json","fetch_events":"https://pith.science/api/pith-number/Z2SWVFT5IPSA5EHO6DCULKZNP4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/Z2SWVFT5IPSA5EHO6DCULKZNP4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/Z2SWVFT5IPSA5EHO6DCULKZNP4/action/storage_attestation","attest_author":"https://pith.science/pith/Z2SWVFT5IPSA5EHO6DCULKZNP4/action/author_attestation","sign_citation":"https://pith.science/pith/Z2SWVFT5IPSA5EHO6DCULKZNP4/action/citation_signature","submit_replication":"https://pith.science/pith/Z2SWVFT5IPSA5EHO6DCULKZNP4/action/replication_record"}},"created_at":"2026-07-05T07:35:17.403068+00:00","updated_at":"2026-07-05T07:35:17.403068+00:00"}