{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:F5X7JSFI5VTT3HFJVPYXSKR33R","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":"1c84029d03a85e273726e1ff54bc6df01590e0d1f25403f03d84ad349082b78c","cross_cats_sorted":["cs.LG","cs.RO"],"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-12T19:25:17Z","title_canon_sha256":"f08dd3004bd6c564804105a7d1f54d3e55105fa59590bbcacd6d209e926929a9"},"schema_version":"1.0","source":{"id":"2605.12674","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.12674","created_at":"2026-05-18T03:09:50Z"},{"alias_kind":"arxiv_version","alias_value":"2605.12674v1","created_at":"2026-05-18T03:09:50Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.12674","created_at":"2026-05-18T03:09:50Z"},{"alias_kind":"pith_short_12","alias_value":"F5X7JSFI5VTT","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"F5X7JSFI5VTT3HFJ","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"F5X7JSFI","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:db0d62c5d72b0420eafe9d5439786c50c3ea8e97f69a0755e638482f12be4d60","target":"graph","created_at":"2026-05-18T03:09:50Z","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":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"We introduce REVELIO, a framework for systematically uncovering interpretable failure modes in VLMs... uncovering previously unreported vulnerabilities in state-of-the-art VLMs."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That the searched concept compositions correspond to genuine, consistent real-world failure modes rather than artifacts of the simulation or search heuristics."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"REVELIO uncovers interpretable failure modes in VLMs by searching combinatorial concept spaces with diversity-aware beam search and Gaussian-process Thompson sampling, revealing vulnerabilities in autonomous driving and indoor robotics."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"REVELIO uncovers interpretable concept compositions that cause consistent failures in vision-language models."}],"snapshot_sha256":"5ee097f560c6dabd233b8134cd5222b6336b1b5393f3603f4a39ca5d910995b8"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"72f7a87453b88f76e969da74fd45d0ed54029c0c5b87df5101475dcdfa1bd8c9"},"paper":{"abstract_excerpt":"Vision-Language Models (VLMs) are increasingly used in safety-critical applications because of their broad reasoning capabilities and ability to generalize with minimal task-specific engineering. Despite these advantages, they can exhibit catastrophic failures in specific real-world situations, constituting failure modes.\n  We introduce REVELIO, a framework for systematically uncovering interpretable failure modes in VLMs. We define a failure mode as a composition of interpretable, domain-relevant concepts-such as pedestrian proximity or adverse weather conditions-under which a target VLM cons","authors_text":"Gagandeep Singh, Isha Chaudhary, Kavya Sachdeva, Sayan Ranu, Vedaant V Jain","cross_cats":["cs.LG","cs.RO"],"headline":"REVELIO uncovers interpretable concept compositions that cause consistent failures in vision-language models.","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-12T19:25:17Z","title":"Revealing Interpretable Failure Modes of VLMs"},"references":{"count":222,"internal_anchors":3,"resolved_work":222,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Claude Sonnet","work_id":"f55b445a-ab6e-4e4d-b496-c3a74994d3ed","year":2025},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Large language model-assisted autonomous vehicle recovery from immobilization, 2025","work_id":"0ffe6562-93d4-4fef-8d21-e5d1e61a4cba","year":2025},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"The use of mmr, diversity-based reranking for reordering documents and producing summaries","work_id":"98357628-52da-45c9-8f07-1e7de5d019ca","year":1998},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Gonzalez, and Ion Stoica","work_id":"7b049cb6-6a65-4610-bb33-d5863d8ae9eb","year":2025},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"Kevin Kai-Chun Chang, Ekin Beyazit, Alberto Sangiovanni-Vincentelli, Tichakorn Wongpirom- sarn, and Sanjit A. Seshia. Scenicrules: An autonomous driving benchmark with multi-objective specifications a","work_id":"34756ee5-9047-48e6-94a2-0bdd3a67767f","year":2026}],"snapshot_sha256":"7612c06159f671a97d092a461973de24bd8ddc6736d7d64bca477774a45c3ecc"},"source":{"id":"2605.12674","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-14T20:24:46.550617Z","id":"963a4c7c-278c-48e5-8987-37b50bf4cf64","model_set":{"reader":"grok-4.3"},"one_line_summary":"REVELIO uncovers interpretable failure modes in VLMs by searching combinatorial concept spaces with diversity-aware beam search and Gaussian-process Thompson sampling, revealing vulnerabilities in autonomous driving and indoor robotics.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"REVELIO uncovers interpretable concept compositions that cause consistent failures in vision-language models.","strongest_claim":"We introduce REVELIO, a framework for systematically uncovering interpretable failure modes in VLMs... uncovering previously unreported vulnerabilities in state-of-the-art VLMs.","weakest_assumption":"That the searched concept compositions correspond to genuine, consistent real-world failure modes rather than artifacts of the simulation or search heuristics."}},"verdict_id":"963a4c7c-278c-48e5-8987-37b50bf4cf64"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:3b878126428f1398c1e6d054200ad1305789cf4ab91870dc16c03acbf195e1d3","target":"record","created_at":"2026-05-18T03:09:50Z","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":"1c84029d03a85e273726e1ff54bc6df01590e0d1f25403f03d84ad349082b78c","cross_cats_sorted":["cs.LG","cs.RO"],"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-12T19:25:17Z","title_canon_sha256":"f08dd3004bd6c564804105a7d1f54d3e55105fa59590bbcacd6d209e926929a9"},"schema_version":"1.0","source":{"id":"2605.12674","kind":"arxiv","version":1}},"canonical_sha256":"2f6ff4c8a8ed673d9ca9abf1792a3bdc73bd12c42f53f645a94850b3fec3b427","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"2f6ff4c8a8ed673d9ca9abf1792a3bdc73bd12c42f53f645a94850b3fec3b427","first_computed_at":"2026-05-18T03:09:50.100529Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-18T03:09:50.100529Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"i6/n2hOEjLFSMxqsIrWQfMjx8QSSBK5WH60NCWEzQJbU1s7ywtXYmft0XEC79kHbXi73QNZSqIo4OtjTIK8CCg==","signature_status":"signed_v1","signed_at":"2026-05-18T03:09:50.101504Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.12674","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:3b878126428f1398c1e6d054200ad1305789cf4ab91870dc16c03acbf195e1d3","sha256:db0d62c5d72b0420eafe9d5439786c50c3ea8e97f69a0755e638482f12be4d60"],"state_sha256":"9a1aaeb858b7198f8ac8cfee702a0623fb9a9c2ea136ef7806c6e69c69ef2228"}