{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:356G454FNQLFMI73ULH6DP2LL3","short_pith_number":"pith:356G454F","schema_version":"1.0","canonical_sha256":"df7c6e77856c165623fba2cfe1bf4b5eef66536cb765741757b7fe1d684249f9","source":{"kind":"arxiv","id":"2503.18556","version":1},"attestation_state":"computed","paper":{"title":"Instruction-Aligned Visual Attention for Mitigating Hallucinations in Large Vision-Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Bin Li, Dehong Gao, Libin Yang, Linbo Jin, Shanqing Yu, Xiaoyan Cai, Yeyuan Wang","submitted_at":"2025-03-24T11:09:06Z","abstract_excerpt":"Despite the significant success of Large Vision-Language models(LVLMs), these models still suffer hallucinations when describing images, generating answers that include non-existent objects. It is reported that these models tend to over-focus on certain irrelevant image tokens that do not contain critical information for answering the question and distort the output. To address this, we propose an Instruction-Aligned Visual Attention(IAVA) approach, which identifies irrelevant tokens by comparing changes in attention weights under two different instructions. By applying contrastive decoding, w"},"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":"2503.18556","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-03-24T11:09:06Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"4cd90318feee367115b81654adb454e888f47703ebeaa8ff6e83f26c4c516fb2","abstract_canon_sha256":"4205cb637565991b8edba0ad594c1d04d05df3e68abb0f652cb67bd3f7bacdd2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:38:13.211330Z","signature_b64":"0go363b08XVQdysjusbn8gAuv5PypeNU/V+bb55juP3DW0n2gQ5y3ugDztNY64fWQfn37STMmTwhO2iTnUNCAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"df7c6e77856c165623fba2cfe1bf4b5eef66536cb765741757b7fe1d684249f9","last_reissued_at":"2026-07-05T10:38:13.210489Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:38:13.210489Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Instruction-Aligned Visual Attention for Mitigating Hallucinations in Large Vision-Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Bin Li, Dehong Gao, Libin Yang, Linbo Jin, Shanqing Yu, Xiaoyan Cai, Yeyuan Wang","submitted_at":"2025-03-24T11:09:06Z","abstract_excerpt":"Despite the significant success of Large Vision-Language models(LVLMs), these models still suffer hallucinations when describing images, generating answers that include non-existent objects. It is reported that these models tend to over-focus on certain irrelevant image tokens that do not contain critical information for answering the question and distort the output. To address this, we propose an Instruction-Aligned Visual Attention(IAVA) approach, which identifies irrelevant tokens by comparing changes in attention weights under two different instructions. By applying contrastive decoding, w"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.18556","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/2503.18556/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":"2503.18556","created_at":"2026-07-05T10:38:13.210598+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.18556v1","created_at":"2026-07-05T10:38:13.210598+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.18556","created_at":"2026-07-05T10:38:13.210598+00:00"},{"alias_kind":"pith_short_12","alias_value":"356G454FNQLF","created_at":"2026-07-05T10:38:13.210598+00:00"},{"alias_kind":"pith_short_16","alias_value":"356G454FNQLFMI73","created_at":"2026-07-05T10:38:13.210598+00:00"},{"alias_kind":"pith_short_8","alias_value":"356G454F","created_at":"2026-07-05T10:38:13.210598+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/356G454FNQLFMI73ULH6DP2LL3","json":"https://pith.science/pith/356G454FNQLFMI73ULH6DP2LL3.json","graph_json":"https://pith.science/api/pith-number/356G454FNQLFMI73ULH6DP2LL3/graph.json","events_json":"https://pith.science/api/pith-number/356G454FNQLFMI73ULH6DP2LL3/events.json","paper":"https://pith.science/paper/356G454F"},"agent_actions":{"view_html":"https://pith.science/pith/356G454FNQLFMI73ULH6DP2LL3","download_json":"https://pith.science/pith/356G454FNQLFMI73ULH6DP2LL3.json","view_paper":"https://pith.science/paper/356G454F","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.18556&json=true","fetch_graph":"https://pith.science/api/pith-number/356G454FNQLFMI73ULH6DP2LL3/graph.json","fetch_events":"https://pith.science/api/pith-number/356G454FNQLFMI73ULH6DP2LL3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/356G454FNQLFMI73ULH6DP2LL3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/356G454FNQLFMI73ULH6DP2LL3/action/storage_attestation","attest_author":"https://pith.science/pith/356G454FNQLFMI73ULH6DP2LL3/action/author_attestation","sign_citation":"https://pith.science/pith/356G454FNQLFMI73ULH6DP2LL3/action/citation_signature","submit_replication":"https://pith.science/pith/356G454FNQLFMI73ULH6DP2LL3/action/replication_record"}},"created_at":"2026-07-05T10:38:13.210598+00:00","updated_at":"2026-07-05T10:38:13.210598+00:00"}