{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:ALDHNP255HTGPUF4XWFUN3UNA3","short_pith_number":"pith:ALDHNP25","schema_version":"1.0","canonical_sha256":"02c676bf5de9e667d0bcbd8b46ee8d06f7aabe7c92005909b6f863562bfc1cab","source":{"kind":"arxiv","id":"2403.00376","version":3},"attestation_state":"computed","paper":{"title":"Spurious Feature Eraser: Stabilizing Test-Time Adaptation for Vision-Language Foundation Model","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Baoyuan Wu, Bingzhe Wu, Changqing Zhang, Huan Ma, Long-Kai Huang, Peilin Zhao, Qinghua Hu, Yan Zhu","submitted_at":"2024-03-01T09:01:53Z","abstract_excerpt":"Vision-language foundation models have exhibited remarkable success across a multitude of downstream tasks due to their scalability on extensive image-text paired data. However, these models also display significant limitations when applied to downstream tasks, such as fine-grained image classification, as a result of ``decision shortcuts'' that hinder their generalization capabilities. In this work, we find that the CLIP model possesses a rich set of features, encompassing both \\textit{desired invariant causal features} and \\textit{undesired decision shortcuts}. Moreover, the underperformance"},"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":"2403.00376","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2024-03-01T09:01:53Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"08fd6da57b226ed3409776ed7a539ed3b9d3fe9e759669bc8599a4527afc5586","abstract_canon_sha256":"276c45e4ad5b423840753156ab668dc8e3a8f4ebed3ab29732a7fb54edcda19c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:00:35.029680Z","signature_b64":"jYhM4I8qY5NW0aevE20H6sxdCKIs9ASZ0AD+JhrqVRw7EMx1nPfWWKX9ubHmvVtf+FGPG6WHP/113gbCnaWQAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"02c676bf5de9e667d0bcbd8b46ee8d06f7aabe7c92005909b6f863562bfc1cab","last_reissued_at":"2026-07-05T10:00:35.028814Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:00:35.028814Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Spurious Feature Eraser: Stabilizing Test-Time Adaptation for Vision-Language Foundation Model","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Baoyuan Wu, Bingzhe Wu, Changqing Zhang, Huan Ma, Long-Kai Huang, Peilin Zhao, Qinghua Hu, Yan Zhu","submitted_at":"2024-03-01T09:01:53Z","abstract_excerpt":"Vision-language foundation models have exhibited remarkable success across a multitude of downstream tasks due to their scalability on extensive image-text paired data. However, these models also display significant limitations when applied to downstream tasks, such as fine-grained image classification, as a result of ``decision shortcuts'' that hinder their generalization capabilities. In this work, we find that the CLIP model possesses a rich set of features, encompassing both \\textit{desired invariant causal features} and \\textit{undesired decision shortcuts}. Moreover, the underperformance"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2403.00376","kind":"arxiv","version":3},"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/2403.00376/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":"2403.00376","created_at":"2026-07-05T10:00:35.028869+00:00"},{"alias_kind":"arxiv_version","alias_value":"2403.00376v3","created_at":"2026-07-05T10:00:35.028869+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2403.00376","created_at":"2026-07-05T10:00:35.028869+00:00"},{"alias_kind":"pith_short_12","alias_value":"ALDHNP255HTG","created_at":"2026-07-05T10:00:35.028869+00:00"},{"alias_kind":"pith_short_16","alias_value":"ALDHNP255HTGPUF4","created_at":"2026-07-05T10:00:35.028869+00:00"},{"alias_kind":"pith_short_8","alias_value":"ALDHNP25","created_at":"2026-07-05T10:00:35.028869+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/ALDHNP255HTGPUF4XWFUN3UNA3","json":"https://pith.science/pith/ALDHNP255HTGPUF4XWFUN3UNA3.json","graph_json":"https://pith.science/api/pith-number/ALDHNP255HTGPUF4XWFUN3UNA3/graph.json","events_json":"https://pith.science/api/pith-number/ALDHNP255HTGPUF4XWFUN3UNA3/events.json","paper":"https://pith.science/paper/ALDHNP25"},"agent_actions":{"view_html":"https://pith.science/pith/ALDHNP255HTGPUF4XWFUN3UNA3","download_json":"https://pith.science/pith/ALDHNP255HTGPUF4XWFUN3UNA3.json","view_paper":"https://pith.science/paper/ALDHNP25","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2403.00376&json=true","fetch_graph":"https://pith.science/api/pith-number/ALDHNP255HTGPUF4XWFUN3UNA3/graph.json","fetch_events":"https://pith.science/api/pith-number/ALDHNP255HTGPUF4XWFUN3UNA3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ALDHNP255HTGPUF4XWFUN3UNA3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ALDHNP255HTGPUF4XWFUN3UNA3/action/storage_attestation","attest_author":"https://pith.science/pith/ALDHNP255HTGPUF4XWFUN3UNA3/action/author_attestation","sign_citation":"https://pith.science/pith/ALDHNP255HTGPUF4XWFUN3UNA3/action/citation_signature","submit_replication":"https://pith.science/pith/ALDHNP255HTGPUF4XWFUN3UNA3/action/replication_record"}},"created_at":"2026-07-05T10:00:35.028869+00:00","updated_at":"2026-07-05T10:00:35.028869+00:00"}