{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:3HKWMK26YKDT4FMRD4XOUJZCXM","short_pith_number":"pith:3HKWMK26","schema_version":"1.0","canonical_sha256":"d9d5662b5ec2873e15911f2eea2722bb393670722e8320c8a06915680e9846ff","source":{"kind":"arxiv","id":"2112.02399","version":3},"attestation_state":"computed","paper":{"title":"VT-CLIP: Enhancing Vision-Language Models with Visual-guided Texts","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Guangnan Zhang, Longtian Qiu, Renrui Zhang, Yafeng Li, Zilu Guo, Ziyao Zeng, Ziyu Guo","submitted_at":"2021-12-04T18:34:24Z","abstract_excerpt":"Contrastive Language-Image Pre-training (CLIP) has drawn increasing attention recently for its transferable visual representation learning. However, due to the semantic gap within datasets, CLIP's pre-trained image-text alignment becomes sub-optimal on downstream tasks, which severely harms its transferring performance. To better adapt the cross-modality embedding space, we propose to enhance CLIP via Visual-guided Texts, named VT-CLIP. Specifically, we guide textual features of different categories to adaptively explore informative regions on the image and aggregate visual features by attenti"},"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":"2112.02399","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2021-12-04T18:34:24Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"02bacf07a5869f71181df563ce1af7430f4387bcb8e3295d4df938e1d53d7ed4","abstract_canon_sha256":"f749b03cbe15f96b7a18b83faa1857749fcbc7b6c3d11a96144738452b3f80b6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:39:48.211007Z","signature_b64":"e7X3GBMA+rWPQEesffBARsQ+PKYSmoUTxXSKURzJleEMSaNSduvtvbFoO46Kouv8u4CgJ08Gf00RqIynzp0LDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d9d5662b5ec2873e15911f2eea2722bb393670722e8320c8a06915680e9846ff","last_reissued_at":"2026-07-05T06:39:48.210576Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:39:48.210576Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"VT-CLIP: Enhancing Vision-Language Models with Visual-guided Texts","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Guangnan Zhang, Longtian Qiu, Renrui Zhang, Yafeng Li, Zilu Guo, Ziyao Zeng, Ziyu Guo","submitted_at":"2021-12-04T18:34:24Z","abstract_excerpt":"Contrastive Language-Image Pre-training (CLIP) has drawn increasing attention recently for its transferable visual representation learning. However, due to the semantic gap within datasets, CLIP's pre-trained image-text alignment becomes sub-optimal on downstream tasks, which severely harms its transferring performance. To better adapt the cross-modality embedding space, we propose to enhance CLIP via Visual-guided Texts, named VT-CLIP. Specifically, we guide textual features of different categories to adaptively explore informative regions on the image and aggregate visual features by attenti"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2112.02399","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/2112.02399/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":"2112.02399","created_at":"2026-07-05T06:39:48.210634+00:00"},{"alias_kind":"arxiv_version","alias_value":"2112.02399v3","created_at":"2026-07-05T06:39:48.210634+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2112.02399","created_at":"2026-07-05T06:39:48.210634+00:00"},{"alias_kind":"pith_short_12","alias_value":"3HKWMK26YKDT","created_at":"2026-07-05T06:39:48.210634+00:00"},{"alias_kind":"pith_short_16","alias_value":"3HKWMK26YKDT4FMR","created_at":"2026-07-05T06:39:48.210634+00:00"},{"alias_kind":"pith_short_8","alias_value":"3HKWMK26","created_at":"2026-07-05T06:39:48.210634+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2412.11375","citing_title":"Text and Image Are Mutually Beneficial: Enhancing Training-Free Few-Shot Classification with CLIP","ref_index":32,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/3HKWMK26YKDT4FMRD4XOUJZCXM","json":"https://pith.science/pith/3HKWMK26YKDT4FMRD4XOUJZCXM.json","graph_json":"https://pith.science/api/pith-number/3HKWMK26YKDT4FMRD4XOUJZCXM/graph.json","events_json":"https://pith.science/api/pith-number/3HKWMK26YKDT4FMRD4XOUJZCXM/events.json","paper":"https://pith.science/paper/3HKWMK26"},"agent_actions":{"view_html":"https://pith.science/pith/3HKWMK26YKDT4FMRD4XOUJZCXM","download_json":"https://pith.science/pith/3HKWMK26YKDT4FMRD4XOUJZCXM.json","view_paper":"https://pith.science/paper/3HKWMK26","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2112.02399&json=true","fetch_graph":"https://pith.science/api/pith-number/3HKWMK26YKDT4FMRD4XOUJZCXM/graph.json","fetch_events":"https://pith.science/api/pith-number/3HKWMK26YKDT4FMRD4XOUJZCXM/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/3HKWMK26YKDT4FMRD4XOUJZCXM/action/timestamp_anchor","attest_storage":"https://pith.science/pith/3HKWMK26YKDT4FMRD4XOUJZCXM/action/storage_attestation","attest_author":"https://pith.science/pith/3HKWMK26YKDT4FMRD4XOUJZCXM/action/author_attestation","sign_citation":"https://pith.science/pith/3HKWMK26YKDT4FMRD4XOUJZCXM/action/citation_signature","submit_replication":"https://pith.science/pith/3HKWMK26YKDT4FMRD4XOUJZCXM/action/replication_record"}},"created_at":"2026-07-05T06:39:48.210634+00:00","updated_at":"2026-07-05T06:39:48.210634+00:00"}