{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:WWKO5UNTEJTFJWDBQEIGMMXU4L","short_pith_number":"pith:WWKO5UNT","schema_version":"1.0","canonical_sha256":"b594eed1b3226654d86181106632f4e2c57bc0dd465458bdce5dd25f32ab5ca1","source":{"kind":"arxiv","id":"2402.02103","version":2},"attestation_state":"computed","paper":{"title":"D\\'ej\\`a Vu Memorization in Vision-Language Models","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Bargav Jayaraman, Chuan Guo, Kamalika Chaudhuri","submitted_at":"2024-02-03T09:55:35Z","abstract_excerpt":"Vision-Language Models (VLMs) have emerged as the state-of-the-art representation learning solution, with myriads of downstream applications such as image classification, retrieval and generation. A natural question is whether these models memorize their training data, which also has implications for generalization. We propose a new method for measuring memorization in VLMs, which we call d\\'ej\\`a vu memorization. For VLMs trained on image-caption pairs, we show that the model indeed retains information about individual objects in the training images beyond what can be inferred from correlatio"},"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":"2402.02103","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CV","submitted_at":"2024-02-03T09:55:35Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"b6f57e16f7ecf977cf1ede4fa1c5c34b28865e039509c03c5fd3ce6dd089b890","abstract_canon_sha256":"ee5d383685faf15d7dba24e0bba449989d27a1f8dc2017fa1ca1d6b8ddb630c2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:27:20.389732Z","signature_b64":"JRZo6e4C9vu0+A+1Yb+soDoLUzZ7Qr67LsIWsU62beC1RN5FdFTAlW4R/x5BDk25vUsPchH/hQP5l1B+3RLeAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b594eed1b3226654d86181106632f4e2c57bc0dd465458bdce5dd25f32ab5ca1","last_reissued_at":"2026-07-05T09:27:20.389340Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:27:20.389340Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"D\\'ej\\`a Vu Memorization in Vision-Language Models","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Bargav Jayaraman, Chuan Guo, Kamalika Chaudhuri","submitted_at":"2024-02-03T09:55:35Z","abstract_excerpt":"Vision-Language Models (VLMs) have emerged as the state-of-the-art representation learning solution, with myriads of downstream applications such as image classification, retrieval and generation. A natural question is whether these models memorize their training data, which also has implications for generalization. We propose a new method for measuring memorization in VLMs, which we call d\\'ej\\`a vu memorization. For VLMs trained on image-caption pairs, we show that the model indeed retains information about individual objects in the training images beyond what can be inferred from correlatio"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.02103","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/2402.02103/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":"2402.02103","created_at":"2026-07-05T09:27:20.389398+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.02103v2","created_at":"2026-07-05T09:27:20.389398+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.02103","created_at":"2026-07-05T09:27:20.389398+00:00"},{"alias_kind":"pith_short_12","alias_value":"WWKO5UNTEJTF","created_at":"2026-07-05T09:27:20.389398+00:00"},{"alias_kind":"pith_short_16","alias_value":"WWKO5UNTEJTFJWDB","created_at":"2026-07-05T09:27:20.389398+00:00"},{"alias_kind":"pith_short_8","alias_value":"WWKO5UNT","created_at":"2026-07-05T09:27:20.389398+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.09132","citing_title":"Vision Language Model Helps Private Information De-Identification in Vision Data","ref_index":57,"is_internal_anchor":false},{"citing_arxiv_id":"2508.00756","citing_title":"LeakyCLIP: Extracting Training Data from CLIP","ref_index":17,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/WWKO5UNTEJTFJWDBQEIGMMXU4L","json":"https://pith.science/pith/WWKO5UNTEJTFJWDBQEIGMMXU4L.json","graph_json":"https://pith.science/api/pith-number/WWKO5UNTEJTFJWDBQEIGMMXU4L/graph.json","events_json":"https://pith.science/api/pith-number/WWKO5UNTEJTFJWDBQEIGMMXU4L/events.json","paper":"https://pith.science/paper/WWKO5UNT"},"agent_actions":{"view_html":"https://pith.science/pith/WWKO5UNTEJTFJWDBQEIGMMXU4L","download_json":"https://pith.science/pith/WWKO5UNTEJTFJWDBQEIGMMXU4L.json","view_paper":"https://pith.science/paper/WWKO5UNT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.02103&json=true","fetch_graph":"https://pith.science/api/pith-number/WWKO5UNTEJTFJWDBQEIGMMXU4L/graph.json","fetch_events":"https://pith.science/api/pith-number/WWKO5UNTEJTFJWDBQEIGMMXU4L/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WWKO5UNTEJTFJWDBQEIGMMXU4L/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WWKO5UNTEJTFJWDBQEIGMMXU4L/action/storage_attestation","attest_author":"https://pith.science/pith/WWKO5UNTEJTFJWDBQEIGMMXU4L/action/author_attestation","sign_citation":"https://pith.science/pith/WWKO5UNTEJTFJWDBQEIGMMXU4L/action/citation_signature","submit_replication":"https://pith.science/pith/WWKO5UNTEJTFJWDBQEIGMMXU4L/action/replication_record"}},"created_at":"2026-07-05T09:27:20.389398+00:00","updated_at":"2026-07-05T09:27:20.389398+00:00"}