{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:QU7AWHWVZPO5EW2K3DU5NWPAWM","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"957b5bcaa2413bde8720e2436a6eb52fab7af7cfc93504b97b2059b3661f4fbc","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-08-05T14:09:55Z","title_canon_sha256":"0c5a500c87ed7962d54bda389ff60e29fe23d4c88746739c4ad115cefd5235a2"},"schema_version":"1.0","source":{"id":"2608.04885","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2608.04885","created_at":"2026-08-06T01:47:27Z"},{"alias_kind":"arxiv_version","alias_value":"2608.04885v1","created_at":"2026-08-06T01:47:27Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.04885","created_at":"2026-08-06T01:47:27Z"},{"alias_kind":"pith_short_12","alias_value":"QU7AWHWVZPO5","created_at":"2026-08-06T01:47:27Z"},{"alias_kind":"pith_short_16","alias_value":"QU7AWHWVZPO5EW2K","created_at":"2026-08-06T01:47:27Z"},{"alias_kind":"pith_short_8","alias_value":"QU7AWHWV","created_at":"2026-08-06T01:47:27Z"}],"graph_snapshots":[{"event_id":"sha256:e141f71b41037761f8e3d9e5a4090f5e3c71bbb86a7b56bd390629ecf2502449","target":"graph","created_at":"2026-08-06T01:47:27Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2608.04885/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Standard accuracy metrics for VLMs often mask significant reliability failures in sensitive domains. In this work, we utilize a histopathology-validated brain MRI dataset to systematically assess the diagnostic robustness of four VLM families under evidence-preserving perturbations. By reordering anatomical slices and swapping target label positions, we evaluate whether models maintain consistent predictions when clinical evidence remains invariant. Our results reveal significant vulnerabilities in presentation-order stability, with models exhibiting prediction flips in up to 48.9% of cases un","authors_text":"Ali Khoramfar, Alireza Mohamadian, Heshaam Faili, Mohammad Javad Dousti","cross_cats":["cs.CL"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-08-05T14:09:55Z","title":"Evaluating the Diagnostic Robustness of Vision-Language Models Under Visual and Textual Perturbations"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.04885","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:c4bc1718d514739e5709bc02dcacaeeb4f26e34cb676f0c8ae1cf7e1df6f028d","target":"record","created_at":"2026-08-06T01:47:27Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"957b5bcaa2413bde8720e2436a6eb52fab7af7cfc93504b97b2059b3661f4fbc","cross_cats_sorted":["cs.CL"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-08-05T14:09:55Z","title_canon_sha256":"0c5a500c87ed7962d54bda389ff60e29fe23d4c88746739c4ad115cefd5235a2"},"schema_version":"1.0","source":{"id":"2608.04885","kind":"arxiv","version":1}},"canonical_sha256":"853e0b1ed5cbddd25b4ad8e9d6d9e0b3042c443b5667ab98b7038aabaef5f034","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"853e0b1ed5cbddd25b4ad8e9d6d9e0b3042c443b5667ab98b7038aabaef5f034","first_computed_at":"2026-08-06T01:47:27.443652Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-08-06T01:47:27.443652Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"/xP6qjksoWumbLPv5TkN4+OIPeRMSwb5Q8TQW0b8w8zupwldoDGfc66v1bjY7cqFwo1UWRmsFVoQKQxRKZdMDg==","signature_status":"signed_v1","signed_at":"2026-08-06T01:47:27.445083Z","signed_message":"canonical_sha256_bytes"},"source_id":"2608.04885","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:c4bc1718d514739e5709bc02dcacaeeb4f26e34cb676f0c8ae1cf7e1df6f028d","sha256:e141f71b41037761f8e3d9e5a4090f5e3c71bbb86a7b56bd390629ecf2502449"],"state_sha256":"a15099beab3fbf241779dea9def39da1e37ec515c071e444ecc88de79fe6cce0"}