{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:VWQO3XCAMRJ5TE7EH7ALM6UEKI","short_pith_number":"pith:VWQO3XCA","schema_version":"1.0","canonical_sha256":"ada0eddc406453d993e43fc0b67a8452216f47b00fdffa4444f3360cfdeda32c","source":{"kind":"arxiv","id":"2411.16828","version":1},"attestation_state":"computed","paper":{"title":"CLIPS: An Enhanced CLIP Framework for Learning with Synthetic Captions","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Bingchen Zhao, Cihang Xie, Xianhang Li, Yanqing Liu, Zeyu Wang","submitted_at":"2024-11-25T18:49:02Z","abstract_excerpt":"Previous works show that noisy, web-crawled image-text pairs may limit vision-language pretraining like CLIP and propose learning with synthetic captions as a promising alternative. Our work continues this effort, introducing two simple yet effective designs to better leverage richly described synthetic captions. Firstly, by observing a strong inverse effect in learning with synthetic captions -- the short synthetic captions can generally lead to MUCH higher performance than full-length ones -- we therefore fed only partial synthetic captions to the text encoder. Secondly, we incorporate an au"},"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":"2411.16828","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2024-11-25T18:49:02Z","cross_cats_sorted":[],"title_canon_sha256":"12b2c7d3a7f1baba9675e4f0f296407e8b85ba3941382e493f171d3a709e637b","abstract_canon_sha256":"ec09165ffbb4b2457a84c02b88c4944a0c21c63bb3697406c8d84244cbf5da13"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:40:23.702321Z","signature_b64":"SEZSfwp31ZJjK5Gnsx760nYrC6scvfrCLVgao9Jdzf1WPMSrcHhlggbjxSs+hk+g+9sHPDuzX5NL/0sTKmqLCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ada0eddc406453d993e43fc0b67a8452216f47b00fdffa4444f3360cfdeda32c","last_reissued_at":"2026-07-05T09:40:23.701828Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:40:23.701828Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"CLIPS: An Enhanced CLIP Framework for Learning with Synthetic Captions","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Bingchen Zhao, Cihang Xie, Xianhang Li, Yanqing Liu, Zeyu Wang","submitted_at":"2024-11-25T18:49:02Z","abstract_excerpt":"Previous works show that noisy, web-crawled image-text pairs may limit vision-language pretraining like CLIP and propose learning with synthetic captions as a promising alternative. Our work continues this effort, introducing two simple yet effective designs to better leverage richly described synthetic captions. Firstly, by observing a strong inverse effect in learning with synthetic captions -- the short synthetic captions can generally lead to MUCH higher performance than full-length ones -- we therefore fed only partial synthetic captions to the text encoder. Secondly, we incorporate an au"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2411.16828","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/2411.16828/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":"2411.16828","created_at":"2026-07-05T09:40:23.701887+00:00"},{"alias_kind":"arxiv_version","alias_value":"2411.16828v1","created_at":"2026-07-05T09:40:23.701887+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2411.16828","created_at":"2026-07-05T09:40:23.701887+00:00"},{"alias_kind":"pith_short_12","alias_value":"VWQO3XCAMRJ5","created_at":"2026-07-05T09:40:23.701887+00:00"},{"alias_kind":"pith_short_16","alias_value":"VWQO3XCAMRJ5TE7E","created_at":"2026-07-05T09:40:23.701887+00:00"},{"alias_kind":"pith_short_8","alias_value":"VWQO3XCA","created_at":"2026-07-05T09:40:23.701887+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.26794","citing_title":"ReasonCLIP-58M: Visually Grounded Commonsense Reasoning Supervision for CLIP","ref_index":58,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03730","citing_title":"Beyond False Stability: High-Noise Drift Gating for Test-Time Adversarial Defenses in Vision-Language Models","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03713","citing_title":"Investigating Adversarial Robustness of Multi-modal Large Language Models","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2605.00809","citing_title":"Let ViT Speak: Generative Language-Image Pre-training","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2605.00809","citing_title":"Let ViT Speak: Generative Language-Image Pre-training","ref_index":45,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VWQO3XCAMRJ5TE7EH7ALM6UEKI","json":"https://pith.science/pith/VWQO3XCAMRJ5TE7EH7ALM6UEKI.json","graph_json":"https://pith.science/api/pith-number/VWQO3XCAMRJ5TE7EH7ALM6UEKI/graph.json","events_json":"https://pith.science/api/pith-number/VWQO3XCAMRJ5TE7EH7ALM6UEKI/events.json","paper":"https://pith.science/paper/VWQO3XCA"},"agent_actions":{"view_html":"https://pith.science/pith/VWQO3XCAMRJ5TE7EH7ALM6UEKI","download_json":"https://pith.science/pith/VWQO3XCAMRJ5TE7EH7ALM6UEKI.json","view_paper":"https://pith.science/paper/VWQO3XCA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2411.16828&json=true","fetch_graph":"https://pith.science/api/pith-number/VWQO3XCAMRJ5TE7EH7ALM6UEKI/graph.json","fetch_events":"https://pith.science/api/pith-number/VWQO3XCAMRJ5TE7EH7ALM6UEKI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VWQO3XCAMRJ5TE7EH7ALM6UEKI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VWQO3XCAMRJ5TE7EH7ALM6UEKI/action/storage_attestation","attest_author":"https://pith.science/pith/VWQO3XCAMRJ5TE7EH7ALM6UEKI/action/author_attestation","sign_citation":"https://pith.science/pith/VWQO3XCAMRJ5TE7EH7ALM6UEKI/action/citation_signature","submit_replication":"https://pith.science/pith/VWQO3XCAMRJ5TE7EH7ALM6UEKI/action/replication_record"}},"created_at":"2026-07-05T09:40:23.701887+00:00","updated_at":"2026-07-05T09:40:23.701887+00:00"}