{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:3WVURSOOXAWTB3IBH2MOIRUL2V","short_pith_number":"pith:3WVURSOO","schema_version":"1.0","canonical_sha256":"ddab48c9ceb82d30ed013e98e4468bd5610d12c57bad8e635bb011ece99e6e5b","source":{"kind":"arxiv","id":"2212.06202","version":2},"attestation_state":"computed","paper":{"title":"Doubly Right Object Recognition: A Why Prompt for Visual Rationales","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Amrutha Sundar, Carl Vondrick, Chengzhi Mao, Junfeng Yang, Revant Teotia, Sachit Menon, Xin Wang","submitted_at":"2022-12-12T19:25:45Z","abstract_excerpt":"Many visual recognition models are evaluated only on their classification accuracy, a metric for which they obtain strong performance. In this paper, we investigate whether computer vision models can also provide correct rationales for their predictions. We propose a ``doubly right'' object recognition benchmark, where the metric requires the model to simultaneously produce both the right labels as well as the right rationales. We find that state-of-the-art visual models, such as CLIP, often provide incorrect rationales for their categorical predictions. However, by transferring the rationales"},"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":"2212.06202","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CV","submitted_at":"2022-12-12T19:25:45Z","cross_cats_sorted":[],"title_canon_sha256":"479a8992ac1c89eea0da6de5850f01dd0d1b6326d2b3b369026e5992aa9e1114","abstract_canon_sha256":"a5bac26efe0741ead398947a090760147cf3598ae510437929edea77e6a3a257"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:54:03.748353Z","signature_b64":"9rtCSJ0qKRlWlyiFnLqpcHQpZ8KwNFla2AuT16nr5KeVw1bJnOuiD81yndEOxgsBsnzh+Kkvj2WaN45hhMK5DQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ddab48c9ceb82d30ed013e98e4468bd5610d12c57bad8e635bb011ece99e6e5b","last_reissued_at":"2026-07-05T05:54:03.747761Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:54:03.747761Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Doubly Right Object Recognition: A Why Prompt for Visual Rationales","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Amrutha Sundar, Carl Vondrick, Chengzhi Mao, Junfeng Yang, Revant Teotia, Sachit Menon, Xin Wang","submitted_at":"2022-12-12T19:25:45Z","abstract_excerpt":"Many visual recognition models are evaluated only on their classification accuracy, a metric for which they obtain strong performance. In this paper, we investigate whether computer vision models can also provide correct rationales for their predictions. We propose a ``doubly right'' object recognition benchmark, where the metric requires the model to simultaneously produce both the right labels as well as the right rationales. We find that state-of-the-art visual models, such as CLIP, often provide incorrect rationales for their categorical predictions. However, by transferring the rationales"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2212.06202","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/2212.06202/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":"2212.06202","created_at":"2026-07-05T05:54:03.747831+00:00"},{"alias_kind":"arxiv_version","alias_value":"2212.06202v2","created_at":"2026-07-05T05:54:03.747831+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2212.06202","created_at":"2026-07-05T05:54:03.747831+00:00"},{"alias_kind":"pith_short_12","alias_value":"3WVURSOOXAWT","created_at":"2026-07-05T05:54:03.747831+00:00"},{"alias_kind":"pith_short_16","alias_value":"3WVURSOOXAWTB3IB","created_at":"2026-07-05T05:54:03.747831+00:00"},{"alias_kind":"pith_short_8","alias_value":"3WVURSOO","created_at":"2026-07-05T05:54:03.747831+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2303.08128","citing_title":"ViperGPT: Visual Inference via Python Execution for Reasoning","ref_index":36,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/3WVURSOOXAWTB3IBH2MOIRUL2V","json":"https://pith.science/pith/3WVURSOOXAWTB3IBH2MOIRUL2V.json","graph_json":"https://pith.science/api/pith-number/3WVURSOOXAWTB3IBH2MOIRUL2V/graph.json","events_json":"https://pith.science/api/pith-number/3WVURSOOXAWTB3IBH2MOIRUL2V/events.json","paper":"https://pith.science/paper/3WVURSOO"},"agent_actions":{"view_html":"https://pith.science/pith/3WVURSOOXAWTB3IBH2MOIRUL2V","download_json":"https://pith.science/pith/3WVURSOOXAWTB3IBH2MOIRUL2V.json","view_paper":"https://pith.science/paper/3WVURSOO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2212.06202&json=true","fetch_graph":"https://pith.science/api/pith-number/3WVURSOOXAWTB3IBH2MOIRUL2V/graph.json","fetch_events":"https://pith.science/api/pith-number/3WVURSOOXAWTB3IBH2MOIRUL2V/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/3WVURSOOXAWTB3IBH2MOIRUL2V/action/timestamp_anchor","attest_storage":"https://pith.science/pith/3WVURSOOXAWTB3IBH2MOIRUL2V/action/storage_attestation","attest_author":"https://pith.science/pith/3WVURSOOXAWTB3IBH2MOIRUL2V/action/author_attestation","sign_citation":"https://pith.science/pith/3WVURSOOXAWTB3IBH2MOIRUL2V/action/citation_signature","submit_replication":"https://pith.science/pith/3WVURSOOXAWTB3IBH2MOIRUL2V/action/replication_record"}},"created_at":"2026-07-05T05:54:03.747831+00:00","updated_at":"2026-07-05T05:54:03.747831+00:00"}