{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:PDJU25XPUL6AJKFURRAMXZK4OV","short_pith_number":"pith:PDJU25XP","schema_version":"1.0","canonical_sha256":"78d34d76efa2fc04a8b48c40cbe55c7573fbec8a032cc852720dd063585e665e","source":{"kind":"arxiv","id":"2504.05457","version":1},"attestation_state":"computed","paper":{"title":"Taxonomy-Aware Evaluation of Vision-Language Models","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Kevin Du, Nico Lang, Niklas Stoehr, Ryan Cotterell, Serge Belongie, Stella Frank, V\\'esteinn Sn{\\ae}bjarnarson","submitted_at":"2025-04-07T19:46:59Z","abstract_excerpt":"When a vision-language model (VLM) is prompted to identify an entity depicted in an image, it may answer 'I see a conifer,' rather than the specific label 'norway spruce'. This raises two issues for evaluation: First, the unconstrained generated text needs to be mapped to the evaluation label space (i.e., 'conifer'). Second, a useful classification measure should give partial credit to less-specific, but not incorrect, answers ('norway spruce' being a type of 'conifer'). To meet these requirements, we propose a framework for evaluating unconstrained text predictions, such as those generated fr"},"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":"2504.05457","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CV","submitted_at":"2025-04-07T19:46:59Z","cross_cats_sorted":[],"title_canon_sha256":"12312232b89e943988e422a792aa54bc92a5cca004febf330e29ce45207817ea","abstract_canon_sha256":"6600b08e937b808acbd6082e0eb2183ff30be2f6c957358219eda295b69df44f"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:45:42.692963Z","signature_b64":"m2yDHIYbP+QgL+iTIb5VlGO9whVh24yTqQPz3DFRxg+f6XlY1kw7m+DiGJRRCOjL/sb7IEcJQKQ9yy5ISTaTBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"78d34d76efa2fc04a8b48c40cbe55c7573fbec8a032cc852720dd063585e665e","last_reissued_at":"2026-07-05T10:45:42.692486Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:45:42.692486Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Taxonomy-Aware Evaluation of Vision-Language Models","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Kevin Du, Nico Lang, Niklas Stoehr, Ryan Cotterell, Serge Belongie, Stella Frank, V\\'esteinn Sn{\\ae}bjarnarson","submitted_at":"2025-04-07T19:46:59Z","abstract_excerpt":"When a vision-language model (VLM) is prompted to identify an entity depicted in an image, it may answer 'I see a conifer,' rather than the specific label 'norway spruce'. This raises two issues for evaluation: First, the unconstrained generated text needs to be mapped to the evaluation label space (i.e., 'conifer'). Second, a useful classification measure should give partial credit to less-specific, but not incorrect, answers ('norway spruce' being a type of 'conifer'). To meet these requirements, we propose a framework for evaluating unconstrained text predictions, such as those generated fr"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.05457","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/2504.05457/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":"2504.05457","created_at":"2026-07-05T10:45:42.692546+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.05457v1","created_at":"2026-07-05T10:45:42.692546+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.05457","created_at":"2026-07-05T10:45:42.692546+00:00"},{"alias_kind":"pith_short_12","alias_value":"PDJU25XPUL6A","created_at":"2026-07-05T10:45:42.692546+00:00"},{"alias_kind":"pith_short_16","alias_value":"PDJU25XPUL6AJKFU","created_at":"2026-07-05T10:45:42.692546+00:00"},{"alias_kind":"pith_short_8","alias_value":"PDJU25XP","created_at":"2026-07-05T10:45:42.692546+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.24340","citing_title":"GeoVision Labeler: Zero-Shot Geospatial Classification with Vision and Language Models","ref_index":2025,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/PDJU25XPUL6AJKFURRAMXZK4OV","json":"https://pith.science/pith/PDJU25XPUL6AJKFURRAMXZK4OV.json","graph_json":"https://pith.science/api/pith-number/PDJU25XPUL6AJKFURRAMXZK4OV/graph.json","events_json":"https://pith.science/api/pith-number/PDJU25XPUL6AJKFURRAMXZK4OV/events.json","paper":"https://pith.science/paper/PDJU25XP"},"agent_actions":{"view_html":"https://pith.science/pith/PDJU25XPUL6AJKFURRAMXZK4OV","download_json":"https://pith.science/pith/PDJU25XPUL6AJKFURRAMXZK4OV.json","view_paper":"https://pith.science/paper/PDJU25XP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.05457&json=true","fetch_graph":"https://pith.science/api/pith-number/PDJU25XPUL6AJKFURRAMXZK4OV/graph.json","fetch_events":"https://pith.science/api/pith-number/PDJU25XPUL6AJKFURRAMXZK4OV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PDJU25XPUL6AJKFURRAMXZK4OV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PDJU25XPUL6AJKFURRAMXZK4OV/action/storage_attestation","attest_author":"https://pith.science/pith/PDJU25XPUL6AJKFURRAMXZK4OV/action/author_attestation","sign_citation":"https://pith.science/pith/PDJU25XPUL6AJKFURRAMXZK4OV/action/citation_signature","submit_replication":"https://pith.science/pith/PDJU25XPUL6AJKFURRAMXZK4OV/action/replication_record"}},"created_at":"2026-07-05T10:45:42.692546+00:00","updated_at":"2026-07-05T10:45:42.692546+00:00"}