{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:JFJXXYF4HR23XNMYPHJM7KFJZY","short_pith_number":"pith:JFJXXYF4","schema_version":"1.0","canonical_sha256":"49537be0bc3c75bbb59879d2cfa8a9ce021dbe8a444ef107c7a2f0ab5b454fc7","source":{"kind":"arxiv","id":"2501.12206","version":3},"attestation_state":"computed","paper":{"title":"PAINT: Paying Attention to INformed Tokens to Mitigate Hallucination in Large Vision-Language Model","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Chris Thomas, Kazi Hasan Ibn Arif, Khizar Hussain, Lang Zhang, Sajib Acharjee Dip","submitted_at":"2025-01-21T15:22:31Z","abstract_excerpt":"Large Vision Language Models (LVLMs) have demonstrated remarkable capabilities in understanding and describing visual content, achieving state-of-the-art performance across various vision-language tasks. However, these models often generate descriptions containing objects or details that are absent in the input image, a phenomenon commonly known as hallucination. Our work investigates the key reasons behind this issue by analyzing the pattern of self-attention in transformer layers. We find that hallucinations often arise from the progressive weakening of attention weight to visual tokens in t"},"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":"2501.12206","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-01-21T15:22:31Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"fea2943891a75015659500c23b32e7b7cccdd39db7e44ffaca3a4e437d70ad3e","abstract_canon_sha256":"fff964e039c245a24e37a124fc77f2afc71238ea23693b243e50961255f9df5b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:39:25.362151Z","signature_b64":"l4JrV4gAaOn7I9bFSEFDsnPZ2zgL6GrVtr8FyF38hX2KjzCE6QA7Dq3J9ZsnRMRouwzm4CiCjKkXKn/IzLsPAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"49537be0bc3c75bbb59879d2cfa8a9ce021dbe8a444ef107c7a2f0ab5b454fc7","last_reissued_at":"2026-07-05T10:39:25.361679Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:39:25.361679Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"PAINT: Paying Attention to INformed Tokens to Mitigate Hallucination in Large Vision-Language Model","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Chris Thomas, Kazi Hasan Ibn Arif, Khizar Hussain, Lang Zhang, Sajib Acharjee Dip","submitted_at":"2025-01-21T15:22:31Z","abstract_excerpt":"Large Vision Language Models (LVLMs) have demonstrated remarkable capabilities in understanding and describing visual content, achieving state-of-the-art performance across various vision-language tasks. However, these models often generate descriptions containing objects or details that are absent in the input image, a phenomenon commonly known as hallucination. Our work investigates the key reasons behind this issue by analyzing the pattern of self-attention in transformer layers. We find that hallucinations often arise from the progressive weakening of attention weight to visual tokens in t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.12206","kind":"arxiv","version":3},"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/2501.12206/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":"2501.12206","created_at":"2026-07-05T10:39:25.361741+00:00"},{"alias_kind":"arxiv_version","alias_value":"2501.12206v3","created_at":"2026-07-05T10:39:25.361741+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.12206","created_at":"2026-07-05T10:39:25.361741+00:00"},{"alias_kind":"pith_short_12","alias_value":"JFJXXYF4HR23","created_at":"2026-07-05T10:39:25.361741+00:00"},{"alias_kind":"pith_short_16","alias_value":"JFJXXYF4HR23XNMY","created_at":"2026-07-05T10:39:25.361741+00:00"},{"alias_kind":"pith_short_8","alias_value":"JFJXXYF4","created_at":"2026-07-05T10:39:25.361741+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.25799","citing_title":"Addressing Exacerbated Attention Sink for Source-Free Cross-Domain Few-Shot Learning","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04641","citing_title":"CAST: Mitigating Object Hallucination in Large Vision-Language Models via Caption-Guided Visual Attention Steering","ref_index":74,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/JFJXXYF4HR23XNMYPHJM7KFJZY","json":"https://pith.science/pith/JFJXXYF4HR23XNMYPHJM7KFJZY.json","graph_json":"https://pith.science/api/pith-number/JFJXXYF4HR23XNMYPHJM7KFJZY/graph.json","events_json":"https://pith.science/api/pith-number/JFJXXYF4HR23XNMYPHJM7KFJZY/events.json","paper":"https://pith.science/paper/JFJXXYF4"},"agent_actions":{"view_html":"https://pith.science/pith/JFJXXYF4HR23XNMYPHJM7KFJZY","download_json":"https://pith.science/pith/JFJXXYF4HR23XNMYPHJM7KFJZY.json","view_paper":"https://pith.science/paper/JFJXXYF4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2501.12206&json=true","fetch_graph":"https://pith.science/api/pith-number/JFJXXYF4HR23XNMYPHJM7KFJZY/graph.json","fetch_events":"https://pith.science/api/pith-number/JFJXXYF4HR23XNMYPHJM7KFJZY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JFJXXYF4HR23XNMYPHJM7KFJZY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JFJXXYF4HR23XNMYPHJM7KFJZY/action/storage_attestation","attest_author":"https://pith.science/pith/JFJXXYF4HR23XNMYPHJM7KFJZY/action/author_attestation","sign_citation":"https://pith.science/pith/JFJXXYF4HR23XNMYPHJM7KFJZY/action/citation_signature","submit_replication":"https://pith.science/pith/JFJXXYF4HR23XNMYPHJM7KFJZY/action/replication_record"}},"created_at":"2026-07-05T10:39:25.361741+00:00","updated_at":"2026-07-05T10:39:25.361741+00:00"}