{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:47CY66C6VUNJKCYKINFI6GWLF3","short_pith_number":"pith:47CY66C6","schema_version":"1.0","canonical_sha256":"e7c58f785ead1a950b0a434a8f1acb2ec73810d021750a639525071aaa94a118","source":{"kind":"arxiv","id":"2401.15914","version":2},"attestation_state":"computed","paper":{"title":"Overcoming the Pitfalls of Vision-Language Model Finetuning for OOD Generalization","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Chen Huang, Hanlin Goh, Josh Susskind, Yuhang Zang","submitted_at":"2024-01-29T06:57:48Z","abstract_excerpt":"Existing vision-language models exhibit strong generalization on a variety of visual domains and tasks. However, such models mainly perform zero-shot recognition in a closed-set manner, and thus struggle to handle open-domain visual concepts by design. There are recent finetuning methods, such as prompt learning, that not only study the discrimination between in-distribution (ID) and out-of-distribution (OOD) samples, but also show some improvements in both ID and OOD accuracies. In this paper, we first demonstrate that vision-language models, after long enough finetuning but without proper re"},"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":"2401.15914","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CV","submitted_at":"2024-01-29T06:57:48Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"854c4f972cbd60d51e1969f138e214700644d21f78a79552fe65b70c0d595e3d","abstract_canon_sha256":"0b2c99506d6ecf47fedc8621efadc4415e235b4f56d35ff386772dd8e4fa6c67"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:08:20.484303Z","signature_b64":"Q8TmoOWrIzNXONLivxH5zDcMeYC9LSyVxrDXCsYWVO93jxVI1gK5ovCQ+gbemzRbdzUxtK3MvB52eOQL1SLQDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e7c58f785ead1a950b0a434a8f1acb2ec73810d021750a639525071aaa94a118","last_reissued_at":"2026-07-05T08:08:20.483823Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:08:20.483823Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Overcoming the Pitfalls of Vision-Language Model Finetuning for OOD Generalization","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Chen Huang, Hanlin Goh, Josh Susskind, Yuhang Zang","submitted_at":"2024-01-29T06:57:48Z","abstract_excerpt":"Existing vision-language models exhibit strong generalization on a variety of visual domains and tasks. However, such models mainly perform zero-shot recognition in a closed-set manner, and thus struggle to handle open-domain visual concepts by design. There are recent finetuning methods, such as prompt learning, that not only study the discrimination between in-distribution (ID) and out-of-distribution (OOD) samples, but also show some improvements in both ID and OOD accuracies. In this paper, we first demonstrate that vision-language models, after long enough finetuning but without proper re"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2401.15914","kind":"arxiv","version":2},"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/2401.15914/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":"2401.15914","created_at":"2026-07-05T08:08:20.483881+00:00"},{"alias_kind":"arxiv_version","alias_value":"2401.15914v2","created_at":"2026-07-05T08:08:20.483881+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2401.15914","created_at":"2026-07-05T08:08:20.483881+00:00"},{"alias_kind":"pith_short_12","alias_value":"47CY66C6VUNJ","created_at":"2026-07-05T08:08:20.483881+00:00"},{"alias_kind":"pith_short_16","alias_value":"47CY66C6VUNJKCYK","created_at":"2026-07-05T08:08:20.483881+00:00"},{"alias_kind":"pith_short_8","alias_value":"47CY66C6","created_at":"2026-07-05T08:08:20.483881+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.15951","citing_title":"From Failure to Feedback: Group Revision Unlocks Hard Cases in Object-Level Grounding","ref_index":95,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/47CY66C6VUNJKCYKINFI6GWLF3","json":"https://pith.science/pith/47CY66C6VUNJKCYKINFI6GWLF3.json","graph_json":"https://pith.science/api/pith-number/47CY66C6VUNJKCYKINFI6GWLF3/graph.json","events_json":"https://pith.science/api/pith-number/47CY66C6VUNJKCYKINFI6GWLF3/events.json","paper":"https://pith.science/paper/47CY66C6"},"agent_actions":{"view_html":"https://pith.science/pith/47CY66C6VUNJKCYKINFI6GWLF3","download_json":"https://pith.science/pith/47CY66C6VUNJKCYKINFI6GWLF3.json","view_paper":"https://pith.science/paper/47CY66C6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2401.15914&json=true","fetch_graph":"https://pith.science/api/pith-number/47CY66C6VUNJKCYKINFI6GWLF3/graph.json","fetch_events":"https://pith.science/api/pith-number/47CY66C6VUNJKCYKINFI6GWLF3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/47CY66C6VUNJKCYKINFI6GWLF3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/47CY66C6VUNJKCYKINFI6GWLF3/action/storage_attestation","attest_author":"https://pith.science/pith/47CY66C6VUNJKCYKINFI6GWLF3/action/author_attestation","sign_citation":"https://pith.science/pith/47CY66C6VUNJKCYKINFI6GWLF3/action/citation_signature","submit_replication":"https://pith.science/pith/47CY66C6VUNJKCYKINFI6GWLF3/action/replication_record"}},"created_at":"2026-07-05T08:08:20.483881+00:00","updated_at":"2026-07-05T08:08:20.483881+00:00"}