{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:NOK35E26TRQUG2UMCHMV3WQBHN","short_pith_number":"pith:NOK35E26","schema_version":"1.0","canonical_sha256":"6b95be935e9c61436a8c11d95dda013b6552dadc044969cd920043e309be67a8","source":{"kind":"arxiv","id":"2608.05945","version":1},"attestation_state":"computed","paper":{"title":"Respect Your Zero-Shot Uncertainty: Conservative Calibration for Test-Time-Adapted Vision-Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Caiting Li, Jiazhen Huang, Jingyan Jiang, Pingting Hao, Xiao Chen, Yaru Sun, Yin Chen, Zhijian He","submitted_at":"2026-08-06T12:11:53Z","abstract_excerpt":"Test-time adaptation (TTA) can improve the recognition accuracy of vision-language models under distribution shift, but often degrades calibration, making predictive confidence unreliable for downstream decision-making. Many existing label-free calibration approaches are either coupled to prompt optimization or rely on logit-range statistics that provide only a coarse characterization of the predictive distribution. We show that TTA can increase confidence and reduce entropy even when the top-1 prediction and its correctness remain unchanged, a failure mode we term prediction-preserving sharpe"},"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":"2608.05945","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-08-06T12:11:53Z","cross_cats_sorted":[],"title_canon_sha256":"5a9373c8a2bad67c8af072538d0f3b985fcbe387abf53ebd31d9479f2ca4ed07","abstract_canon_sha256":"b87dd710e71bdd71c448a91021dc8ad8fe31427fa5f8dc4b8c3833377928a029"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-07T00:53:53.567373Z","signature_b64":"HmxFpxknRSnJZo1lR1vCAq56KWKxhuVw2Vs0H9FdaYra9tBnx+/Nh1G6CRaRSBmgwg/FXY1Jxoj7CUEnirdjBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6b95be935e9c61436a8c11d95dda013b6552dadc044969cd920043e309be67a8","last_reissued_at":"2026-08-07T00:53:53.565976Z","signature_status":"signed_v1","first_computed_at":"2026-08-07T00:53:53.565976Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Respect Your Zero-Shot Uncertainty: Conservative Calibration for Test-Time-Adapted Vision-Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Caiting Li, Jiazhen Huang, Jingyan Jiang, Pingting Hao, Xiao Chen, Yaru Sun, Yin Chen, Zhijian He","submitted_at":"2026-08-06T12:11:53Z","abstract_excerpt":"Test-time adaptation (TTA) can improve the recognition accuracy of vision-language models under distribution shift, but often degrades calibration, making predictive confidence unreliable for downstream decision-making. Many existing label-free calibration approaches are either coupled to prompt optimization or rely on logit-range statistics that provide only a coarse characterization of the predictive distribution. We show that TTA can increase confidence and reduce entropy even when the top-1 prediction and its correctness remain unchanged, a failure mode we term prediction-preserving sharpe"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.05945","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/2608.05945/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":"2608.05945","created_at":"2026-08-07T00:53:53.567528+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.05945v1","created_at":"2026-08-07T00:53:53.567528+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.05945","created_at":"2026-08-07T00:53:53.567528+00:00"},{"alias_kind":"pith_short_12","alias_value":"NOK35E26TRQU","created_at":"2026-08-07T00:53:53.567528+00:00"},{"alias_kind":"pith_short_16","alias_value":"NOK35E26TRQUG2UM","created_at":"2026-08-07T00:53:53.567528+00:00"},{"alias_kind":"pith_short_8","alias_value":"NOK35E26","created_at":"2026-08-07T00:53:53.567528+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/NOK35E26TRQUG2UMCHMV3WQBHN","json":"https://pith.science/pith/NOK35E26TRQUG2UMCHMV3WQBHN.json","graph_json":"https://pith.science/api/pith-number/NOK35E26TRQUG2UMCHMV3WQBHN/graph.json","events_json":"https://pith.science/api/pith-number/NOK35E26TRQUG2UMCHMV3WQBHN/events.json","paper":"https://pith.science/paper/NOK35E26"},"agent_actions":{"view_html":"https://pith.science/pith/NOK35E26TRQUG2UMCHMV3WQBHN","download_json":"https://pith.science/pith/NOK35E26TRQUG2UMCHMV3WQBHN.json","view_paper":"https://pith.science/paper/NOK35E26","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.05945&json=true","fetch_graph":"https://pith.science/api/pith-number/NOK35E26TRQUG2UMCHMV3WQBHN/graph.json","fetch_events":"https://pith.science/api/pith-number/NOK35E26TRQUG2UMCHMV3WQBHN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NOK35E26TRQUG2UMCHMV3WQBHN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NOK35E26TRQUG2UMCHMV3WQBHN/action/storage_attestation","attest_author":"https://pith.science/pith/NOK35E26TRQUG2UMCHMV3WQBHN/action/author_attestation","sign_citation":"https://pith.science/pith/NOK35E26TRQUG2UMCHMV3WQBHN/action/citation_signature","submit_replication":"https://pith.science/pith/NOK35E26TRQUG2UMCHMV3WQBHN/action/replication_record"}},"created_at":"2026-08-07T00:53:53.567528+00:00","updated_at":"2026-08-07T00:53:53.567528+00:00"}