{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:GDV46QHWLB6AGJVNDS7TLFCWY2","short_pith_number":"pith:GDV46QHW","schema_version":"1.0","canonical_sha256":"30ebcf40f6587c0326ad1cbf359456c6a539c274676e43cc41abba0df6f8f6da","source":{"kind":"arxiv","id":"2311.16839","version":2},"attestation_state":"computed","paper":{"title":"Beyond Hallucinations: Enhancing LVLMs through Hallucination-Aware Direct Preference Optimization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"HA-DPO trains multimodal models to prefer accurate image descriptions over hallucinatory ones by optimizing on paired responses.","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Bin Wang, Conghui He, Jiaqi Wang, Linke Ouyang, Xiaoyi Dong, Zhiyuan Zhao","submitted_at":"2023-11-28T14:54:37Z","abstract_excerpt":"Multimodal large language models have made significant advancements in recent years, yet they still suffer from a common issue known as the \"hallucination problem\", in which the models generate textual descriptions that inaccurately depict or entirely fabricate content from associated images. This paper introduces a novel solution, Hallucination-Aware Direct Preference Optimization (HA-DPO), which reframes the hallucination problem as a preference selection task. The model is trained to favor the non-hallucinating response when presented with two responses of the same image (one accurate and o"},"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":true,"formal_links_present":false},"canonical_record":{"source":{"id":"2311.16839","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2023-11-28T14:54:37Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"cfd50767e98148c085f71f832e663a8ccb28a0b88fa4daa98d3430463eb4a27c","abstract_canon_sha256":"005fb17ca22012a217f56f74db2d7f620a1da22e2332a44702e9a4daa218764b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:38:14.313144Z","signature_b64":"qNsy8/9mRDSA3l9DFns72Xxn5Rzd5LIr0zuT9eMgKndUU9ZGvW1ht5sWh9lsh5ZGSZBwAvChHdswNQ38OxbeAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"30ebcf40f6587c0326ad1cbf359456c6a539c274676e43cc41abba0df6f8f6da","last_reissued_at":"2026-05-17T23:38:14.312573Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:38:14.312573Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Beyond Hallucinations: Enhancing LVLMs through Hallucination-Aware Direct Preference Optimization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"HA-DPO trains multimodal models to prefer accurate image descriptions over hallucinatory ones by optimizing on paired responses.","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Bin Wang, Conghui He, Jiaqi Wang, Linke Ouyang, Xiaoyi Dong, Zhiyuan Zhao","submitted_at":"2023-11-28T14:54:37Z","abstract_excerpt":"Multimodal large language models have made significant advancements in recent years, yet they still suffer from a common issue known as the \"hallucination problem\", in which the models generate textual descriptions that inaccurately depict or entirely fabricate content from associated images. This paper introduces a novel solution, Hallucination-Aware Direct Preference Optimization (HA-DPO), which reframes the hallucination problem as a preference selection task. The model is trained to favor the non-hallucinating response when presented with two responses of the same image (one accurate and o"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"When applied to three mainstream multimodal models, HA-DPO significantly reduced hallucination issues and amplified the models' generalization capabilities. Notably, the MiniGPT-4 model, when enhanced with HA-DPO, demonstrated a substantial improvement: POPE accuracy rose from 51.13% to 86.13% (an absolute improvement of 35%), and the MME score surged from 932.00 to 1326.46 (a relative improvement of 42.32%).","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The constructed positive and negative sample pairs are high-quality, style-consistent, and free of new biases that could undermine preference learning or generalization beyond the tested benchmarks.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"HA-DPO reframes hallucination reduction in LVLMs as direct preference optimization over style-consistent positive and negative response pairs, yielding large gains such as 35-point POPE accuracy jumps on MiniGPT-4.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"HA-DPO trains multimodal models to prefer accurate image descriptions over hallucinatory ones by optimizing on paired responses.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"6bcaca4640cf7cd7cb48662118152e2fd33ec01a53e36f970b9ae61704a55d89"},"source":{"id":"2311.16839","kind":"arxiv","version":2},"verdict":{"id":"0a33879e-fd3f-4015-b277-eaf4122f1c90","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-17T10:48:14.256416Z","strongest_claim":"When applied to three mainstream multimodal models, HA-DPO significantly reduced hallucination issues and amplified the models' generalization capabilities. Notably, the MiniGPT-4 model, when enhanced with HA-DPO, demonstrated a substantial improvement: POPE accuracy rose from 51.13% to 86.13% (an absolute improvement of 35%), and the MME score surged from 932.00 to 1326.46 (a relative improvement of 42.32%).","one_line_summary":"HA-DPO reframes hallucination reduction in LVLMs as direct preference optimization over style-consistent positive and negative response pairs, yielding large gains such as 35-point POPE accuracy jumps on MiniGPT-4.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The constructed positive and negative sample pairs are high-quality, style-consistent, and free of new biases that could undermine preference learning or generalization beyond the tested benchmarks.","pith_extraction_headline":"HA-DPO trains multimodal models to prefer accurate image descriptions over hallucinatory ones by optimizing on paired responses."},"references":{"count":79,"sample":[{"doi":"","year":2022,"title":"Brown, Jack Clark, Sam McCandlish, Chris Olah, Benjamin Mann, and Jared Kaplan","work_id":"42c49e87-0d71-4159-82b0-335c1cdf982a","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2014,"title":"The exploration-exploitation dilemma: a multidisci- plinary framework","work_id":"1b85bd8c-2331-4169-adae-4e756e673fca","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2020,"title":"Tom B. Brown, Benjamin Mann, Nick Ryder, Melanie Sub- biah, Jared Kaplan, Prafulla Dhariwal, Arvind Neelakan- tan, Pranav Shyam, Girish Sastry, Amanda Askell, Sand- hini Agarwal, Ariel Herbert-V oss, ","work_id":"36ef20ad-b8cb-4b11-9b0f-2e2d164b0174","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2017,"title":"Christiano, Jan Leike, Tom B","work_id":"a2d819f8-2c2f-42a1-bd50-b96b93d59729","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Wenliang Dai, Junnan Li, Dongxu Li, Anthony Meng Huat Tiong, Junqi Zhao, Weisheng Wang, Boyang Li, Pascale Fung, and Steven C. H. Hoi. Instructblip: Towards general- purpose vision-language models wit","work_id":"ca0fe730-e9d7-436a-a761-ceba8bc2e1d3","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":79,"snapshot_sha256":"28bd0671d7c0b4edd136e93e111c48f18e82204dcaadae758ba18f98c8437de2","internal_anchors":1},"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":"2311.16839","created_at":"2026-05-17T23:38:14.312645+00:00"},{"alias_kind":"arxiv_version","alias_value":"2311.16839v2","created_at":"2026-05-17T23:38:14.312645+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2311.16839","created_at":"2026-05-17T23:38:14.312645+00:00"},{"alias_kind":"pith_short_12","alias_value":"GDV46QHWLB6A","created_at":"2026-05-18T12:33:33.725879+00:00"},{"alias_kind":"pith_short_16","alias_value":"GDV46QHWLB6AGJVN","created_at":"2026-05-18T12:33:33.725879+00:00"},{"alias_kind":"pith_short_8","alias_value":"GDV46QHW","created_at":"2026-05-18T12:33:33.725879+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":20,"internal_anchor_count":20,"sample":[{"citing_arxiv_id":"2605.15300","citing_title":"Deep Pre-Alignment for VLMs","ref_index":90,"is_internal_anchor":true},{"citing_arxiv_id":"2506.06856","citing_title":"Vision-EKIPL: External Knowledge-Infused Policy Learning for Visual Reasoning","ref_index":70,"is_internal_anchor":true},{"citing_arxiv_id":"2402.11411","citing_title":"Aligning Modalities in Vision Large Language Models via Preference Fine-tuning","ref_index":184,"is_internal_anchor":true},{"citing_arxiv_id":"2407.03320","citing_title":"InternLM-XComposer-2.5: A Versatile Large Vision Language Model Supporting Long-Contextual Input and Output","ref_index":181,"is_internal_anchor":true},{"citing_arxiv_id":"2511.18740","citing_title":"Multimodal Large Language Models with Adaptive Preference Optimization for Sequential Recommendation","ref_index":73,"is_internal_anchor":true},{"citing_arxiv_id":"2401.16420","citing_title":"InternLM-XComposer2: Mastering Free-form Text-Image Composition and Comprehension in Vision-Language Large Model","ref_index":100,"is_internal_anchor":true},{"citing_arxiv_id":"2602.11824","citing_title":"Revis: Sparse Latent Steering to Mitigate Object Hallucination in Large Vision-Language Models","ref_index":12,"is_internal_anchor":true},{"citing_arxiv_id":"2503.05236","citing_title":"Unified Reward Model for Multimodal Understanding and Generation","ref_index":27,"is_internal_anchor":true},{"citing_arxiv_id":"2503.01785","citing_title":"Visual-RFT: Visual Reinforcement Fine-Tuning","ref_index":49,"is_internal_anchor":true},{"citing_arxiv_id":"2402.00253","citing_title":"A Survey on Hallucination in Large Vision-Language Models","ref_index":54,"is_internal_anchor":true},{"citing_arxiv_id":"2604.03556","citing_title":"Focus Matters: Phase-Aware Suppression for Hallucination in Vision-Language Models","ref_index":28,"is_internal_anchor":true},{"citing_arxiv_id":"2605.11808","citing_title":"Mitigating Action-Relation Hallucinations in LVLMs via Relation-aware Visual Enhancement","ref_index":18,"is_internal_anchor":true},{"citing_arxiv_id":"2605.10622","citing_title":"Vocabulary Hijacking in LVLMs: Unveiling Critical Attention Heads by Excluding Inert Tokens to Mitigate Hallucination","ref_index":40,"is_internal_anchor":true},{"citing_arxiv_id":"2604.24395","citing_title":"Aligning with Your Own Voice: Self-Corrected Preference Learning for Hallucination Mitigation in LVLMs","ref_index":3,"is_internal_anchor":true},{"citing_arxiv_id":"2605.04874","citing_title":"Uncertainty-Aware Exploratory Direct Preference Optimization for Multimodal Large Language Models","ref_index":46,"is_internal_anchor":true},{"citing_arxiv_id":"2404.18930","citing_title":"Hallucination of Multimodal Large Language Models: A Survey","ref_index":219,"is_internal_anchor":true},{"citing_arxiv_id":"2604.12424","citing_title":"Decoding by Perturbation: Mitigating MLLM Hallucinations via Dynamic Textual Perturbation","ref_index":64,"is_internal_anchor":true},{"citing_arxiv_id":"2605.08031","citing_title":"Object Hallucination-Free Reinforcement Unlearning for Vision-Language Models","ref_index":18,"is_internal_anchor":true},{"citing_arxiv_id":"2604.14129","citing_title":"Don't Let the Video Speak: Audio-Contrastive Preference Optimization for Audio-Visual Language Models","ref_index":56,"is_internal_anchor":true},{"citing_arxiv_id":"2604.18512","citing_title":"S2H-DPO: Hardness-Aware Preference Optimization for Vision-Language Models","ref_index":132,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GDV46QHWLB6AGJVNDS7TLFCWY2","json":"https://pith.science/pith/GDV46QHWLB6AGJVNDS7TLFCWY2.json","graph_json":"https://pith.science/api/pith-number/GDV46QHWLB6AGJVNDS7TLFCWY2/graph.json","events_json":"https://pith.science/api/pith-number/GDV46QHWLB6AGJVNDS7TLFCWY2/events.json","paper":"https://pith.science/paper/GDV46QHW"},"agent_actions":{"view_html":"https://pith.science/pith/GDV46QHWLB6AGJVNDS7TLFCWY2","download_json":"https://pith.science/pith/GDV46QHWLB6AGJVNDS7TLFCWY2.json","view_paper":"https://pith.science/paper/GDV46QHW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2311.16839&json=true","fetch_graph":"https://pith.science/api/pith-number/GDV46QHWLB6AGJVNDS7TLFCWY2/graph.json","fetch_events":"https://pith.science/api/pith-number/GDV46QHWLB6AGJVNDS7TLFCWY2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GDV46QHWLB6AGJVNDS7TLFCWY2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GDV46QHWLB6AGJVNDS7TLFCWY2/action/storage_attestation","attest_author":"https://pith.science/pith/GDV46QHWLB6AGJVNDS7TLFCWY2/action/author_attestation","sign_citation":"https://pith.science/pith/GDV46QHWLB6AGJVNDS7TLFCWY2/action/citation_signature","submit_replication":"https://pith.science/pith/GDV46QHWLB6AGJVNDS7TLFCWY2/action/replication_record"}},"created_at":"2026-05-17T23:38:14.312645+00:00","updated_at":"2026-05-17T23:38:14.312645+00:00"}