{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:2SLRRC7RIPPUYRYW3VAAKC6QG3","short_pith_number":"pith:2SLRRC7R","schema_version":"1.0","canonical_sha256":"d497188bf143df4c4716dd40050bd036c1cb8e313b1173a4cd69e118cfdb2c2a","source":{"kind":"arxiv","id":"2203.07190","version":1},"attestation_state":"computed","paper":{"title":"CLIP Models are Few-shot Learners: Empirical Studies on VQA and Visual Entailment","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Furu Wei, Haoyu Song, Li Dong, Ting Liu, Wei-Nan Zhang","submitted_at":"2022-03-14T15:29:27Z","abstract_excerpt":"CLIP has shown a remarkable zero-shot capability on a wide range of vision tasks. Previously, CLIP is only regarded as a powerful visual encoder. However, after being pre-trained by language supervision from a large amount of image-caption pairs, CLIP itself should also have acquired some few-shot abilities for vision-language tasks. In this work, we empirically show that CLIP can be a strong vision-language few-shot learner by leveraging the power of language. We first evaluate CLIP's zero-shot performance on a typical visual question answering task and demonstrate a zero-shot cross-modality "},"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":"2203.07190","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CV","submitted_at":"2022-03-14T15:29:27Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"00b8a4ed8b2dd97ff3328ee6dc516e39208f1f8ab76cf0d32c8d58dcdd1c8bf3","abstract_canon_sha256":"4990404bbcc9be1bdf53b4686ce7cee11639e8076f67b87d2f92727825742b3b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:05:00.500287Z","signature_b64":"G7uh7W+6FMYvwnLqDy0+sfpwWcIEY/zBst61hVmoO8PZmASt3KrAMp5NvEayLwY5JDTg5VeoRGwEGrMElru8Cg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d497188bf143df4c4716dd40050bd036c1cb8e313b1173a4cd69e118cfdb2c2a","last_reissued_at":"2026-07-05T04:05:00.499800Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:05:00.499800Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"CLIP Models are Few-shot Learners: Empirical Studies on VQA and Visual Entailment","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Furu Wei, Haoyu Song, Li Dong, Ting Liu, Wei-Nan Zhang","submitted_at":"2022-03-14T15:29:27Z","abstract_excerpt":"CLIP has shown a remarkable zero-shot capability on a wide range of vision tasks. Previously, CLIP is only regarded as a powerful visual encoder. However, after being pre-trained by language supervision from a large amount of image-caption pairs, CLIP itself should also have acquired some few-shot abilities for vision-language tasks. In this work, we empirically show that CLIP can be a strong vision-language few-shot learner by leveraging the power of language. We first evaluate CLIP's zero-shot performance on a typical visual question answering task and demonstrate a zero-shot cross-modality "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2203.07190","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/2203.07190/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":"2203.07190","created_at":"2026-07-05T04:05:00.499866+00:00"},{"alias_kind":"arxiv_version","alias_value":"2203.07190v1","created_at":"2026-07-05T04:05:00.499866+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2203.07190","created_at":"2026-07-05T04:05:00.499866+00:00"},{"alias_kind":"pith_short_12","alias_value":"2SLRRC7RIPPU","created_at":"2026-07-05T04:05:00.499866+00:00"},{"alias_kind":"pith_short_16","alias_value":"2SLRRC7RIPPUYRYW","created_at":"2026-07-05T04:05:00.499866+00:00"},{"alias_kind":"pith_short_8","alias_value":"2SLRRC7R","created_at":"2026-07-05T04:05:00.499866+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2204.00598","citing_title":"Socratic Models: Composing Zero-Shot Multimodal Reasoning with Language","ref_index":47,"is_internal_anchor":false},{"citing_arxiv_id":"2604.05583","citing_title":"WRF4CIR: Weight-Regularized Fine-Tuning Network for Composed Image Retrieval","ref_index":53,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2SLRRC7RIPPUYRYW3VAAKC6QG3","json":"https://pith.science/pith/2SLRRC7RIPPUYRYW3VAAKC6QG3.json","graph_json":"https://pith.science/api/pith-number/2SLRRC7RIPPUYRYW3VAAKC6QG3/graph.json","events_json":"https://pith.science/api/pith-number/2SLRRC7RIPPUYRYW3VAAKC6QG3/events.json","paper":"https://pith.science/paper/2SLRRC7R"},"agent_actions":{"view_html":"https://pith.science/pith/2SLRRC7RIPPUYRYW3VAAKC6QG3","download_json":"https://pith.science/pith/2SLRRC7RIPPUYRYW3VAAKC6QG3.json","view_paper":"https://pith.science/paper/2SLRRC7R","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2203.07190&json=true","fetch_graph":"https://pith.science/api/pith-number/2SLRRC7RIPPUYRYW3VAAKC6QG3/graph.json","fetch_events":"https://pith.science/api/pith-number/2SLRRC7RIPPUYRYW3VAAKC6QG3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2SLRRC7RIPPUYRYW3VAAKC6QG3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2SLRRC7RIPPUYRYW3VAAKC6QG3/action/storage_attestation","attest_author":"https://pith.science/pith/2SLRRC7RIPPUYRYW3VAAKC6QG3/action/author_attestation","sign_citation":"https://pith.science/pith/2SLRRC7RIPPUYRYW3VAAKC6QG3/action/citation_signature","submit_replication":"https://pith.science/pith/2SLRRC7RIPPUYRYW3VAAKC6QG3/action/replication_record"}},"created_at":"2026-07-05T04:05:00.499866+00:00","updated_at":"2026-07-05T04:05:00.499866+00:00"}