{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:OMLEIK4VZFYNX7P2WONTZZPMSF","short_pith_number":"pith:OMLEIK4V","schema_version":"1.0","canonical_sha256":"7316442b95c970dbfdfab39b3ce5ec916cf6b8d569708d975fb9233081c743b4","source":{"kind":"arxiv","id":"2408.10433","version":1},"attestation_state":"computed","paper":{"title":"CLIP-DPO: Vision-Language Models as a Source of Preference for Fixing Hallucinations in LVLMs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Adrian Bulat, Brais Martinez, Georgios Tzimiropoulos, Yassine Ouali","submitted_at":"2024-08-19T21:56:20Z","abstract_excerpt":"Despite recent successes, LVLMs or Large Vision Language Models are prone to hallucinating details like objects and their properties or relations, limiting their real-world deployment. To address this and improve their robustness, we present CLIP-DPO, a preference optimization method that leverages contrastively pre-trained Vision-Language (VL) embedding models, such as CLIP, for DPO-based optimization of LVLMs. Unlike prior works tackling LVLM hallucinations, our method does not rely on paid-for APIs, and does not require additional training data or the deployment of other external LVLMs. Ins"},"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":"2408.10433","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2024-08-19T21:56:20Z","cross_cats_sorted":[],"title_canon_sha256":"b4a87e32097f7725f6121f3be04ff3600dbab873146fae794a5e6f128c2fca0f","abstract_canon_sha256":"f50a35db572b894b33f3ac94b5c4cf62a1ab82a83beb0fb975c42a2be1122cc9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:57:11.782770Z","signature_b64":"fyAwu1dMtNXpae3bM8jjeXHhH3HVLJbADya9hNLKuJ8cudbggi+KNo1+aBYZjfMxJv719hBg/d7GoZURSrjtBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7316442b95c970dbfdfab39b3ce5ec916cf6b8d569708d975fb9233081c743b4","last_reissued_at":"2026-07-05T08:57:11.782248Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:57:11.782248Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"CLIP-DPO: Vision-Language Models as a Source of Preference for Fixing Hallucinations in LVLMs","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Adrian Bulat, Brais Martinez, Georgios Tzimiropoulos, Yassine Ouali","submitted_at":"2024-08-19T21:56:20Z","abstract_excerpt":"Despite recent successes, LVLMs or Large Vision Language Models are prone to hallucinating details like objects and their properties or relations, limiting their real-world deployment. To address this and improve their robustness, we present CLIP-DPO, a preference optimization method that leverages contrastively pre-trained Vision-Language (VL) embedding models, such as CLIP, for DPO-based optimization of LVLMs. Unlike prior works tackling LVLM hallucinations, our method does not rely on paid-for APIs, and does not require additional training data or the deployment of other external LVLMs. Ins"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2408.10433","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/2408.10433/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":"2408.10433","created_at":"2026-07-05T08:57:11.782306+00:00"},{"alias_kind":"arxiv_version","alias_value":"2408.10433v1","created_at":"2026-07-05T08:57:11.782306+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2408.10433","created_at":"2026-07-05T08:57:11.782306+00:00"},{"alias_kind":"pith_short_12","alias_value":"OMLEIK4VZFYN","created_at":"2026-07-05T08:57:11.782306+00:00"},{"alias_kind":"pith_short_16","alias_value":"OMLEIK4VZFYNX7P2","created_at":"2026-07-05T08:57:11.782306+00:00"},{"alias_kind":"pith_short_8","alias_value":"OMLEIK4V","created_at":"2026-07-05T08:57:11.782306+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2510.21122","citing_title":"NoisyGRPO: Incentivizing Multimodal CoT Reasoning via Noise Injection and Bayesian Estimation","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18512","citing_title":"S2H-DPO: Hardness-Aware Preference Optimization for Vision-Language Models","ref_index":66,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/OMLEIK4VZFYNX7P2WONTZZPMSF","json":"https://pith.science/pith/OMLEIK4VZFYNX7P2WONTZZPMSF.json","graph_json":"https://pith.science/api/pith-number/OMLEIK4VZFYNX7P2WONTZZPMSF/graph.json","events_json":"https://pith.science/api/pith-number/OMLEIK4VZFYNX7P2WONTZZPMSF/events.json","paper":"https://pith.science/paper/OMLEIK4V"},"agent_actions":{"view_html":"https://pith.science/pith/OMLEIK4VZFYNX7P2WONTZZPMSF","download_json":"https://pith.science/pith/OMLEIK4VZFYNX7P2WONTZZPMSF.json","view_paper":"https://pith.science/paper/OMLEIK4V","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2408.10433&json=true","fetch_graph":"https://pith.science/api/pith-number/OMLEIK4VZFYNX7P2WONTZZPMSF/graph.json","fetch_events":"https://pith.science/api/pith-number/OMLEIK4VZFYNX7P2WONTZZPMSF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/OMLEIK4VZFYNX7P2WONTZZPMSF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/OMLEIK4VZFYNX7P2WONTZZPMSF/action/storage_attestation","attest_author":"https://pith.science/pith/OMLEIK4VZFYNX7P2WONTZZPMSF/action/author_attestation","sign_citation":"https://pith.science/pith/OMLEIK4VZFYNX7P2WONTZZPMSF/action/citation_signature","submit_replication":"https://pith.science/pith/OMLEIK4VZFYNX7P2WONTZZPMSF/action/replication_record"}},"created_at":"2026-07-05T08:57:11.782306+00:00","updated_at":"2026-07-05T08:57:11.782306+00:00"}