{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:EVAF4SNQR6L2N5D5NOLICNV6YP","short_pith_number":"pith:EVAF4SNQ","schema_version":"1.0","canonical_sha256":"25405e49b08f97a6f47d6b968136bec3fbcb69ffe4ea0bc6a1ccd37a75e33a4e","source":{"kind":"arxiv","id":"2607.22034","version":1},"attestation_state":"computed","paper":{"title":"Small Vision-Language Models Know When They Are Wrong But Cannot Say So: A Two-Model Study of Stated versus Internal Confidence Under Realistic Image Degradation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.CV","authors_text":"M M Asif Ferdous","submitted_at":"2026-07-24T07:00:32Z","abstract_excerpt":"Vision-language models (VLMs) are increasingly deployed on consumer hardware where input images are degraded by compression, camera shake, and poor lighting. In such settings, a reliable uncertainty signal matters more than raw accuracy, because it determines when a system should defer rather than answer. We evaluate two small open-weight VLMs -- Qwen2-VL-2B-Instruct and SmolVLM-Instruct -- across six realistic photographic degradations at three severity levels, comparing two confidence signals: the confidence the model states in natural language, and the model's own mean token probability ove"},"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":"2607.22034","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-07-24T07:00:32Z","cross_cats_sorted":["cs.CL","cs.LG"],"title_canon_sha256":"cf049364426aafdb60439312e45bb6c1afa183339cb108dc3282345daff7cd8d","abstract_canon_sha256":"62491fb2f20f27c24b2aa094f00ce271e6f5347f6fec91efbb63ba4f8162816a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-27T01:20:44.390333Z","signature_b64":"4qDdNvYnx1CC/Lp+ZDCgAajCXiCM7uDmzZDE15oprWYqJflK/sfmQC/HXhGiA28eqZJncI2YzfUR0BjRWbiXAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"25405e49b08f97a6f47d6b968136bec3fbcb69ffe4ea0bc6a1ccd37a75e33a4e","last_reissued_at":"2026-07-27T01:20:44.389491Z","signature_status":"signed_v1","first_computed_at":"2026-07-27T01:20:44.389491Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Small Vision-Language Models Know When They Are Wrong But Cannot Say So: A Two-Model Study of Stated versus Internal Confidence Under Realistic Image Degradation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.CV","authors_text":"M M Asif Ferdous","submitted_at":"2026-07-24T07:00:32Z","abstract_excerpt":"Vision-language models (VLMs) are increasingly deployed on consumer hardware where input images are degraded by compression, camera shake, and poor lighting. In such settings, a reliable uncertainty signal matters more than raw accuracy, because it determines when a system should defer rather than answer. We evaluate two small open-weight VLMs -- Qwen2-VL-2B-Instruct and SmolVLM-Instruct -- across six realistic photographic degradations at three severity levels, comparing two confidence signals: the confidence the model states in natural language, and the model's own mean token probability ove"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.22034","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/2607.22034/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":"2607.22034","created_at":"2026-07-27T01:20:44.389924+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.22034v1","created_at":"2026-07-27T01:20:44.389924+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.22034","created_at":"2026-07-27T01:20:44.389924+00:00"},{"alias_kind":"pith_short_12","alias_value":"EVAF4SNQR6L2","created_at":"2026-07-27T01:20:44.389924+00:00"},{"alias_kind":"pith_short_16","alias_value":"EVAF4SNQR6L2N5D5","created_at":"2026-07-27T01:20:44.389924+00:00"},{"alias_kind":"pith_short_8","alias_value":"EVAF4SNQ","created_at":"2026-07-27T01:20:44.389924+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/EVAF4SNQR6L2N5D5NOLICNV6YP","json":"https://pith.science/pith/EVAF4SNQR6L2N5D5NOLICNV6YP.json","graph_json":"https://pith.science/api/pith-number/EVAF4SNQR6L2N5D5NOLICNV6YP/graph.json","events_json":"https://pith.science/api/pith-number/EVAF4SNQR6L2N5D5NOLICNV6YP/events.json","paper":"https://pith.science/paper/EVAF4SNQ"},"agent_actions":{"view_html":"https://pith.science/pith/EVAF4SNQR6L2N5D5NOLICNV6YP","download_json":"https://pith.science/pith/EVAF4SNQR6L2N5D5NOLICNV6YP.json","view_paper":"https://pith.science/paper/EVAF4SNQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.22034&json=true","fetch_graph":"https://pith.science/api/pith-number/EVAF4SNQR6L2N5D5NOLICNV6YP/graph.json","fetch_events":"https://pith.science/api/pith-number/EVAF4SNQR6L2N5D5NOLICNV6YP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/EVAF4SNQR6L2N5D5NOLICNV6YP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/EVAF4SNQR6L2N5D5NOLICNV6YP/action/storage_attestation","attest_author":"https://pith.science/pith/EVAF4SNQR6L2N5D5NOLICNV6YP/action/author_attestation","sign_citation":"https://pith.science/pith/EVAF4SNQR6L2N5D5NOLICNV6YP/action/citation_signature","submit_replication":"https://pith.science/pith/EVAF4SNQR6L2N5D5NOLICNV6YP/action/replication_record"}},"created_at":"2026-07-27T01:20:44.389924+00:00","updated_at":"2026-07-27T01:20:44.389924+00:00"}