{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:ZORZOCXUB2XY3WQI65CMRRBNSF","short_pith_number":"pith:ZORZOCXU","schema_version":"1.0","canonical_sha256":"cba3970af40eaf8dda08f744c8c42d914be8c5645c4ee368ffb15ab96e127619","source":{"kind":"arxiv","id":"2310.07699","version":3},"attestation_state":"computed","paper":{"title":"VeCLIP: Improving CLIP Training via Visual-enriched Captions","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Aleksei Timofeev, Bowen Zhang, Chen-Nee Chuah, Haoping Bai, Haotian Zhang, Jiulong Shan, Meng Cao, Wentao Wu, Xianzhi Du, Yinfei Yang, Zhe Gan, Zhengfeng Lai","submitted_at":"2023-10-11T17:49:13Z","abstract_excerpt":"Large-scale web-crawled datasets are fundamental for the success of pre-training vision-language models, such as CLIP. However, the inherent noise and potential irrelevance of web-crawled AltTexts pose challenges in achieving precise image-text alignment. Existing methods utilizing large language models (LLMs) for caption rewriting have shown promise on small, curated datasets like CC3M and CC12M. This study introduces a scalable pipeline for noisy caption rewriting. Unlike recent LLM rewriting techniques, we emphasize the incorporation of visual concepts into captions, termed as Visual-enrich"},"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":"2310.07699","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2023-10-11T17:49:13Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"cc299c3e8c08591848418bc2dc55c21a98224ca7a9a2b25d0771522e5136fcd1","abstract_canon_sha256":"c58a5808ec519d4d91c1f2e4779b2e73686833183a8abbe25d5e702938a2943a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:55:45.416129Z","signature_b64":"S1F8wUX4QFh4o5YDm9BgYjMhYKE5vriyHjG14RX3+eMixbgauxRdqx3uRHe1NeGknRh3U0mtulL2yu+XXNe4Dw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"cba3970af40eaf8dda08f744c8c42d914be8c5645c4ee368ffb15ab96e127619","last_reissued_at":"2026-07-05T07:55:45.415408Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:55:45.415408Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"VeCLIP: Improving CLIP Training via Visual-enriched Captions","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Aleksei Timofeev, Bowen Zhang, Chen-Nee Chuah, Haoping Bai, Haotian Zhang, Jiulong Shan, Meng Cao, Wentao Wu, Xianzhi Du, Yinfei Yang, Zhe Gan, Zhengfeng Lai","submitted_at":"2023-10-11T17:49:13Z","abstract_excerpt":"Large-scale web-crawled datasets are fundamental for the success of pre-training vision-language models, such as CLIP. However, the inherent noise and potential irrelevance of web-crawled AltTexts pose challenges in achieving precise image-text alignment. Existing methods utilizing large language models (LLMs) for caption rewriting have shown promise on small, curated datasets like CC3M and CC12M. This study introduces a scalable pipeline for noisy caption rewriting. Unlike recent LLM rewriting techniques, we emphasize the incorporation of visual concepts into captions, termed as Visual-enrich"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.07699","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/2310.07699/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":"2310.07699","created_at":"2026-07-05T07:55:45.415486+00:00"},{"alias_kind":"arxiv_version","alias_value":"2310.07699v3","created_at":"2026-07-05T07:55:45.415486+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.07699","created_at":"2026-07-05T07:55:45.415486+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZORZOCXUB2XY","created_at":"2026-07-05T07:55:45.415486+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZORZOCXUB2XY3WQI","created_at":"2026-07-05T07:55:45.415486+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZORZOCXU","created_at":"2026-07-05T07:55:45.415486+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":9,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.21734","citing_title":"HPP: Hierarchical Programmatic Probing for Long Video Understanding by Decoupling Perception and Reasoning","ref_index":251,"is_internal_anchor":false},{"citing_arxiv_id":"2606.13275","citing_title":"Zero-Shot Captioning for Cultural Heritage: Automated Image Analysis of Traditional Indonesian Clothing","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2403.09611","citing_title":"MM1: Methods, Analysis & Insights from Multimodal LLM Pre-training","ref_index":57,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11405","citing_title":"20/20 Vision Language Models: A Prescription for Better VLMs through Data Curation Alone","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2311.12793","citing_title":"ShareGPT4V: Improving Large Multi-Modal Models with Better Captions","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11405","citing_title":"20/20 Vision Language Models: A Prescription for Better VLMs through Data Curation Alone","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2502.14786","citing_title":"SigLIP 2: Multilingual Vision-Language Encoders with Improved Semantic Understanding, Localization, and Dense Features","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2604.12498","citing_title":"Lit2Vec: A Reproducible Workflow for Building a Legally Screened Chemistry Corpus from S2ORC for Downstream Retrieval and Text Mining","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01048","citing_title":"Compared to What? Baselines and Metrics for Counterfactual Prompting","ref_index":90,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZORZOCXUB2XY3WQI65CMRRBNSF","json":"https://pith.science/pith/ZORZOCXUB2XY3WQI65CMRRBNSF.json","graph_json":"https://pith.science/api/pith-number/ZORZOCXUB2XY3WQI65CMRRBNSF/graph.json","events_json":"https://pith.science/api/pith-number/ZORZOCXUB2XY3WQI65CMRRBNSF/events.json","paper":"https://pith.science/paper/ZORZOCXU"},"agent_actions":{"view_html":"https://pith.science/pith/ZORZOCXUB2XY3WQI65CMRRBNSF","download_json":"https://pith.science/pith/ZORZOCXUB2XY3WQI65CMRRBNSF.json","view_paper":"https://pith.science/paper/ZORZOCXU","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2310.07699&json=true","fetch_graph":"https://pith.science/api/pith-number/ZORZOCXUB2XY3WQI65CMRRBNSF/graph.json","fetch_events":"https://pith.science/api/pith-number/ZORZOCXUB2XY3WQI65CMRRBNSF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZORZOCXUB2XY3WQI65CMRRBNSF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZORZOCXUB2XY3WQI65CMRRBNSF/action/storage_attestation","attest_author":"https://pith.science/pith/ZORZOCXUB2XY3WQI65CMRRBNSF/action/author_attestation","sign_citation":"https://pith.science/pith/ZORZOCXUB2XY3WQI65CMRRBNSF/action/citation_signature","submit_replication":"https://pith.science/pith/ZORZOCXUB2XY3WQI65CMRRBNSF/action/replication_record"}},"created_at":"2026-07-05T07:55:45.415486+00:00","updated_at":"2026-07-05T07:55:45.415486+00:00"}