{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:HFLLX4QDAWWUNZBEZNDIQQT46K","short_pith_number":"pith:HFLLX4QD","schema_version":"1.0","canonical_sha256":"3956bbf20305ad46e424cb4688427cf2b2bba96aab71190b9e47fa2efcdc0730","source":{"kind":"arxiv","id":"2606.30168","version":1},"attestation_state":"computed","paper":{"title":"Latent Noise Mask for Reducing Visual Redundancy in Multimodal Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Hongyuan Zhang, Kai Jiang, Ruishu Zhu, Siqi Huang, Xuelong Li","submitted_at":"2026-06-29T11:44:09Z","abstract_excerpt":"Multimodal large language models (MLLMs) often fail in fine-grained visual reasoning, as question-relevant visual cues are diluted by dense and redundant image tokens. Recent multimodal reasoning methods usually extend chain-of-thought from language models into visual or latent spaces, seeking to add intermediate reasoning states while overlooking the negative impact of redundant visual tokens. We propose LatEnt Noise maSk (Lens), a question-conditioned visual evidence purification framework that empowers MLLMs to reason with cleaner visual cues in latent space. Lens introduces a lightweight L"},"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":"2606.30168","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-06-29T11:44:09Z","cross_cats_sorted":[],"title_canon_sha256":"a8b91d19810ed29a7a07a9023a6ee5e37293cf510a7656cb3e218aabdf17b6a3","abstract_canon_sha256":"84d925be7ad73514c5b0082ff785a9d96ac6535c4f8ffc71b97b394d8a70cf93"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-30T02:17:52.358176Z","signature_b64":"yJ2i5jlAy+LRqLEAII7UELomnz219YKefFNUhMqmmytO+phjDcnOrczUa779+/6RxTrvzCjyAHU/GRhx3GCbDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3956bbf20305ad46e424cb4688427cf2b2bba96aab71190b9e47fa2efcdc0730","last_reissued_at":"2026-06-30T02:17:52.357630Z","signature_status":"signed_v1","first_computed_at":"2026-06-30T02:17:52.357630Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Latent Noise Mask for Reducing Visual Redundancy in Multimodal Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Hongyuan Zhang, Kai Jiang, Ruishu Zhu, Siqi Huang, Xuelong Li","submitted_at":"2026-06-29T11:44:09Z","abstract_excerpt":"Multimodal large language models (MLLMs) often fail in fine-grained visual reasoning, as question-relevant visual cues are diluted by dense and redundant image tokens. Recent multimodal reasoning methods usually extend chain-of-thought from language models into visual or latent spaces, seeking to add intermediate reasoning states while overlooking the negative impact of redundant visual tokens. We propose LatEnt Noise maSk (Lens), a question-conditioned visual evidence purification framework that empowers MLLMs to reason with cleaner visual cues in latent space. Lens introduces a lightweight L"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.30168","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/2606.30168/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":"2606.30168","created_at":"2026-06-30T02:17:52.357695+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.30168v1","created_at":"2026-06-30T02:17:52.357695+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.30168","created_at":"2026-06-30T02:17:52.357695+00:00"},{"alias_kind":"pith_short_12","alias_value":"HFLLX4QDAWWU","created_at":"2026-06-30T02:17:52.357695+00:00"},{"alias_kind":"pith_short_16","alias_value":"HFLLX4QDAWWUNZBE","created_at":"2026-06-30T02:17:52.357695+00:00"},{"alias_kind":"pith_short_8","alias_value":"HFLLX4QD","created_at":"2026-06-30T02:17:52.357695+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/HFLLX4QDAWWUNZBEZNDIQQT46K","json":"https://pith.science/pith/HFLLX4QDAWWUNZBEZNDIQQT46K.json","graph_json":"https://pith.science/api/pith-number/HFLLX4QDAWWUNZBEZNDIQQT46K/graph.json","events_json":"https://pith.science/api/pith-number/HFLLX4QDAWWUNZBEZNDIQQT46K/events.json","paper":"https://pith.science/paper/HFLLX4QD"},"agent_actions":{"view_html":"https://pith.science/pith/HFLLX4QDAWWUNZBEZNDIQQT46K","download_json":"https://pith.science/pith/HFLLX4QDAWWUNZBEZNDIQQT46K.json","view_paper":"https://pith.science/paper/HFLLX4QD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.30168&json=true","fetch_graph":"https://pith.science/api/pith-number/HFLLX4QDAWWUNZBEZNDIQQT46K/graph.json","fetch_events":"https://pith.science/api/pith-number/HFLLX4QDAWWUNZBEZNDIQQT46K/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HFLLX4QDAWWUNZBEZNDIQQT46K/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HFLLX4QDAWWUNZBEZNDIQQT46K/action/storage_attestation","attest_author":"https://pith.science/pith/HFLLX4QDAWWUNZBEZNDIQQT46K/action/author_attestation","sign_citation":"https://pith.science/pith/HFLLX4QDAWWUNZBEZNDIQQT46K/action/citation_signature","submit_replication":"https://pith.science/pith/HFLLX4QDAWWUNZBEZNDIQQT46K/action/replication_record"}},"created_at":"2026-06-30T02:17:52.357695+00:00","updated_at":"2026-06-30T02:17:52.357695+00:00"}