{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:XBFJ4D62JH627W7MPLMZBYTZQW","short_pith_number":"pith:XBFJ4D62","schema_version":"1.0","canonical_sha256":"b84a9e0fda49fdafdbec7ad990e27985a068718e72c97f0408256d41dc0ea48c","source":{"kind":"arxiv","id":"2607.10240","version":1},"attestation_state":"computed","paper":{"title":"What Does Your Short-Answer VQA Score Actually Measure? Evaluator-Dependent Instability in Multimodal Short-Answer Benchmarks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.MM"],"primary_cat":"cs.CV","authors_text":"Guanhua Ye, Meiyu Liang, Niu Jingbin, Yan Li, Yawen Li, Yingxia Shao, Zhe Xue","submitted_at":"2026-07-11T10:01:29Z","abstract_excerpt":"Short-answer VQA benchmarks conflate two distinct quantities: whether a model's answer is semantically correct, and whether that answer matches the surface form expected by the automatic evaluator. We study this conflation across six vision--language models and six benchmarks, using a human-validated semantic judge (97.6% precision) to audit over 37k official errors. A second text-only judge reproduces the same benchmark-level false-negative pattern, showing that the effect is not an artifact of a single audit model. On text-rich benchmarks, up to half of these errors are semantically acceptab"},"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":"2607.10240","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-07-11T10:01:29Z","cross_cats_sorted":["cs.MM"],"title_canon_sha256":"fc380f1b56fcf5cf2d6a967018edb097f839271905739b7a1b66cfc98ace85d8","abstract_canon_sha256":"60a53b69b3f69bace15e45bb201d587c9a0d3a9bcf1fe6a1e3b876e9cd7c9f65"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-14T01:20:32.356001Z","signature_b64":"yKyAndGt9fhfQKgq/REfzCxAABTRIJrjyBVH1fvH7V80Y69bxeHWU604oAWm/jkiGVHGd6zeR6ZHqukSLeH2BA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b84a9e0fda49fdafdbec7ad990e27985a068718e72c97f0408256d41dc0ea48c","last_reissued_at":"2026-07-14T01:20:32.355054Z","signature_status":"signed_v1","first_computed_at":"2026-07-14T01:20:32.355054Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"What Does Your Short-Answer VQA Score Actually Measure? Evaluator-Dependent Instability in Multimodal Short-Answer Benchmarks","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.MM"],"primary_cat":"cs.CV","authors_text":"Guanhua Ye, Meiyu Liang, Niu Jingbin, Yan Li, Yawen Li, Yingxia Shao, Zhe Xue","submitted_at":"2026-07-11T10:01:29Z","abstract_excerpt":"Short-answer VQA benchmarks conflate two distinct quantities: whether a model's answer is semantically correct, and whether that answer matches the surface form expected by the automatic evaluator. We study this conflation across six vision--language models and six benchmarks, using a human-validated semantic judge (97.6% precision) to audit over 37k official errors. A second text-only judge reproduces the same benchmark-level false-negative pattern, showing that the effect is not an artifact of a single audit model. On text-rich benchmarks, up to half of these errors are semantically acceptab"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.10240","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/2607.10240/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":"2607.10240","created_at":"2026-07-14T01:20:32.355545+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.10240v1","created_at":"2026-07-14T01:20:32.355545+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.10240","created_at":"2026-07-14T01:20:32.355545+00:00"},{"alias_kind":"pith_short_12","alias_value":"XBFJ4D62JH62","created_at":"2026-07-14T01:20:32.355545+00:00"},{"alias_kind":"pith_short_16","alias_value":"XBFJ4D62JH627W7M","created_at":"2026-07-14T01:20:32.355545+00:00"},{"alias_kind":"pith_short_8","alias_value":"XBFJ4D62","created_at":"2026-07-14T01:20:32.355545+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/XBFJ4D62JH627W7MPLMZBYTZQW","json":"https://pith.science/pith/XBFJ4D62JH627W7MPLMZBYTZQW.json","graph_json":"https://pith.science/api/pith-number/XBFJ4D62JH627W7MPLMZBYTZQW/graph.json","events_json":"https://pith.science/api/pith-number/XBFJ4D62JH627W7MPLMZBYTZQW/events.json","paper":"https://pith.science/paper/XBFJ4D62"},"agent_actions":{"view_html":"https://pith.science/pith/XBFJ4D62JH627W7MPLMZBYTZQW","download_json":"https://pith.science/pith/XBFJ4D62JH627W7MPLMZBYTZQW.json","view_paper":"https://pith.science/paper/XBFJ4D62","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.10240&json=true","fetch_graph":"https://pith.science/api/pith-number/XBFJ4D62JH627W7MPLMZBYTZQW/graph.json","fetch_events":"https://pith.science/api/pith-number/XBFJ4D62JH627W7MPLMZBYTZQW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XBFJ4D62JH627W7MPLMZBYTZQW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XBFJ4D62JH627W7MPLMZBYTZQW/action/storage_attestation","attest_author":"https://pith.science/pith/XBFJ4D62JH627W7MPLMZBYTZQW/action/author_attestation","sign_citation":"https://pith.science/pith/XBFJ4D62JH627W7MPLMZBYTZQW/action/citation_signature","submit_replication":"https://pith.science/pith/XBFJ4D62JH627W7MPLMZBYTZQW/action/replication_record"}},"created_at":"2026-07-14T01:20:32.355545+00:00","updated_at":"2026-07-14T01:20:32.355545+00:00"}