{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:AMX3BAS5LGA2HVZ44QMORDTSYS","short_pith_number":"pith:AMX3BAS5","schema_version":"1.0","canonical_sha256":"032fb0825d5981a3d73ce418e88e72c483b14a3a957f9724a9a74a06b17297cc","source":{"kind":"arxiv","id":"2508.02419","version":1},"attestation_state":"computed","paper":{"title":"Modality Bias in LVLMs: Analyzing and Mitigating Object Hallucination via Attention Lens","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Haohan Zheng, Zhenguo Zhang","submitted_at":"2025-08-04T13:40:59Z","abstract_excerpt":"Large vision-language models (LVLMs) have demonstrated remarkable multimodal comprehension and reasoning capabilities, but they still suffer from severe object hallucination. Previous studies primarily attribute the flaw to linguistic prior caused by the scale mismatch between visual encoders and large language models (LLMs) in LVLMs. Specifically, as current LVLMs are built upon LLMs, they tend to over-rely on textual prompts and internal knowledge of LLMs, generating descriptions inconsistent with visual cues. However, through an in-depth investigation of the hallucinated mechanisms, we empi"},"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":"2508.02419","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-08-04T13:40:59Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"c354049c25ba581e455754cf686e24f0a91ebdad6cfe0d6fa7ed7f6768da6bb1","abstract_canon_sha256":"28065b647909bf471342963ccec1a7eb827a113936b45cbeeb23517c90c8d34d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:48:14.043018Z","signature_b64":"fg//DE3+jwSsabrtXis85zpfxIvMxcglj+TVr2aLN2S3jYYjlfI0SnUxgnIPw3DnbmhWc2dM2JsznVFcHJjMDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"032fb0825d5981a3d73ce418e88e72c483b14a3a957f9724a9a74a06b17297cc","last_reissued_at":"2026-07-05T11:48:14.042476Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:48:14.042476Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Modality Bias in LVLMs: Analyzing and Mitigating Object Hallucination via Attention Lens","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Haohan Zheng, Zhenguo Zhang","submitted_at":"2025-08-04T13:40:59Z","abstract_excerpt":"Large vision-language models (LVLMs) have demonstrated remarkable multimodal comprehension and reasoning capabilities, but they still suffer from severe object hallucination. Previous studies primarily attribute the flaw to linguistic prior caused by the scale mismatch between visual encoders and large language models (LLMs) in LVLMs. Specifically, as current LVLMs are built upon LLMs, they tend to over-rely on textual prompts and internal knowledge of LLMs, generating descriptions inconsistent with visual cues. However, through an in-depth investigation of the hallucinated mechanisms, we empi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.02419","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/2508.02419/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":"2508.02419","created_at":"2026-07-05T11:48:14.042540+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.02419v1","created_at":"2026-07-05T11:48:14.042540+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.02419","created_at":"2026-07-05T11:48:14.042540+00:00"},{"alias_kind":"pith_short_12","alias_value":"AMX3BAS5LGA2","created_at":"2026-07-05T11:48:14.042540+00:00"},{"alias_kind":"pith_short_16","alias_value":"AMX3BAS5LGA2HVZ4","created_at":"2026-07-05T11:48:14.042540+00:00"},{"alias_kind":"pith_short_8","alias_value":"AMX3BAS5","created_at":"2026-07-05T11:48:14.042540+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.05874","citing_title":"Evaluating Stochastic Collapse and Implicit Bias in Multimodal Large Language Models","ref_index":86,"is_internal_anchor":false},{"citing_arxiv_id":"2606.28845","citing_title":"Personalizing MLLMs via Reinforced Multimodal Reference Game","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2512.14044","citing_title":"OmniDrive-R1: Reinforcement-driven Interleaved Multi-modal Chain-of-Thought for Trustworthy Vision-Language Autonomous Driving","ref_index":52,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/AMX3BAS5LGA2HVZ44QMORDTSYS","json":"https://pith.science/pith/AMX3BAS5LGA2HVZ44QMORDTSYS.json","graph_json":"https://pith.science/api/pith-number/AMX3BAS5LGA2HVZ44QMORDTSYS/graph.json","events_json":"https://pith.science/api/pith-number/AMX3BAS5LGA2HVZ44QMORDTSYS/events.json","paper":"https://pith.science/paper/AMX3BAS5"},"agent_actions":{"view_html":"https://pith.science/pith/AMX3BAS5LGA2HVZ44QMORDTSYS","download_json":"https://pith.science/pith/AMX3BAS5LGA2HVZ44QMORDTSYS.json","view_paper":"https://pith.science/paper/AMX3BAS5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.02419&json=true","fetch_graph":"https://pith.science/api/pith-number/AMX3BAS5LGA2HVZ44QMORDTSYS/graph.json","fetch_events":"https://pith.science/api/pith-number/AMX3BAS5LGA2HVZ44QMORDTSYS/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AMX3BAS5LGA2HVZ44QMORDTSYS/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AMX3BAS5LGA2HVZ44QMORDTSYS/action/storage_attestation","attest_author":"https://pith.science/pith/AMX3BAS5LGA2HVZ44QMORDTSYS/action/author_attestation","sign_citation":"https://pith.science/pith/AMX3BAS5LGA2HVZ44QMORDTSYS/action/citation_signature","submit_replication":"https://pith.science/pith/AMX3BAS5LGA2HVZ44QMORDTSYS/action/replication_record"}},"created_at":"2026-07-05T11:48:14.042540+00:00","updated_at":"2026-07-05T11:48:14.042540+00:00"}