{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:U5KSI7L7FAUCGPCADRUKFFXYQP","short_pith_number":"pith:U5KSI7L7","schema_version":"1.0","canonical_sha256":"a755247d7f2828233c401c68a296f883eb9fd31445bbc650a2b6f8af9baf7139","source":{"kind":"arxiv","id":"2301.02280","version":2},"attestation_state":"computed","paper":{"title":"Filtering, Distillation, and Hard Negatives for Vision-Language Pre-Training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Abhimanyu Dubey, Abhishek Kadian, Dhruv Mahajan, Filip Radenovic, Simon Vandenhende, Todor Mihaylov, Vignesh Ramanathan, Yash Patel, Yi Wen","submitted_at":"2023-01-05T19:48:01Z","abstract_excerpt":"Vision-language models trained with contrastive learning on large-scale noisy data are becoming increasingly popular for zero-shot recognition problems. In this paper we improve the following three aspects of the contrastive pre-training pipeline: dataset noise, model initialization and the training objective. First, we propose a straightforward filtering strategy titled Complexity, Action, and Text-spotting (CAT) that significantly reduces dataset size, while achieving improved performance across zero-shot vision-language tasks. Next, we propose an approach titled Concept Distillation to leve"},"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":"2301.02280","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2023-01-05T19:48:01Z","cross_cats_sorted":[],"title_canon_sha256":"0a9d5058595e8736e7613d85f59315d83f0e2b1dc705eefb65b098ac6045f70e","abstract_canon_sha256":"60fe8a84ced99dfb4a94f06a5dcebd8a007ce1ad70d68b4fbec4dd5253473224"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:56:15.594462Z","signature_b64":"4Q4ADgWemK+05Hhq90I6KBEbgZkz/i9GgV8vS+JMAn4VZLRwF85QV5s8/0x6XXi9HlG56aamyesRk236BAJyDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a755247d7f2828233c401c68a296f883eb9fd31445bbc650a2b6f8af9baf7139","last_reissued_at":"2026-07-05T05:56:15.593948Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:56:15.593948Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Filtering, Distillation, and Hard Negatives for Vision-Language Pre-Training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Abhimanyu Dubey, Abhishek Kadian, Dhruv Mahajan, Filip Radenovic, Simon Vandenhende, Todor Mihaylov, Vignesh Ramanathan, Yash Patel, Yi Wen","submitted_at":"2023-01-05T19:48:01Z","abstract_excerpt":"Vision-language models trained with contrastive learning on large-scale noisy data are becoming increasingly popular for zero-shot recognition problems. In this paper we improve the following three aspects of the contrastive pre-training pipeline: dataset noise, model initialization and the training objective. First, we propose a straightforward filtering strategy titled Complexity, Action, and Text-spotting (CAT) that significantly reduces dataset size, while achieving improved performance across zero-shot vision-language tasks. Next, we propose an approach titled Concept Distillation to leve"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2301.02280","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/2301.02280/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":"2301.02280","created_at":"2026-07-05T05:56:15.594008+00:00"},{"alias_kind":"arxiv_version","alias_value":"2301.02280v2","created_at":"2026-07-05T05:56:15.594008+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2301.02280","created_at":"2026-07-05T05:56:15.594008+00:00"},{"alias_kind":"pith_short_12","alias_value":"U5KSI7L7FAUC","created_at":"2026-07-05T05:56:15.594008+00:00"},{"alias_kind":"pith_short_16","alias_value":"U5KSI7L7FAUCGPCA","created_at":"2026-07-05T05:56:15.594008+00:00"},{"alias_kind":"pith_short_8","alias_value":"U5KSI7L7","created_at":"2026-07-05T05:56:15.594008+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2303.09540","citing_title":"SemDeDup: Data-efficient learning at web-scale through semantic deduplication","ref_index":14,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/U5KSI7L7FAUCGPCADRUKFFXYQP","json":"https://pith.science/pith/U5KSI7L7FAUCGPCADRUKFFXYQP.json","graph_json":"https://pith.science/api/pith-number/U5KSI7L7FAUCGPCADRUKFFXYQP/graph.json","events_json":"https://pith.science/api/pith-number/U5KSI7L7FAUCGPCADRUKFFXYQP/events.json","paper":"https://pith.science/paper/U5KSI7L7"},"agent_actions":{"view_html":"https://pith.science/pith/U5KSI7L7FAUCGPCADRUKFFXYQP","download_json":"https://pith.science/pith/U5KSI7L7FAUCGPCADRUKFFXYQP.json","view_paper":"https://pith.science/paper/U5KSI7L7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2301.02280&json=true","fetch_graph":"https://pith.science/api/pith-number/U5KSI7L7FAUCGPCADRUKFFXYQP/graph.json","fetch_events":"https://pith.science/api/pith-number/U5KSI7L7FAUCGPCADRUKFFXYQP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/U5KSI7L7FAUCGPCADRUKFFXYQP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/U5KSI7L7FAUCGPCADRUKFFXYQP/action/storage_attestation","attest_author":"https://pith.science/pith/U5KSI7L7FAUCGPCADRUKFFXYQP/action/author_attestation","sign_citation":"https://pith.science/pith/U5KSI7L7FAUCGPCADRUKFFXYQP/action/citation_signature","submit_replication":"https://pith.science/pith/U5KSI7L7FAUCGPCADRUKFFXYQP/action/replication_record"}},"created_at":"2026-07-05T05:56:15.594008+00:00","updated_at":"2026-07-05T05:56:15.594008+00:00"}