{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:76KDTVCSOHUM7Y63HHCHDE27PK","short_pith_number":"pith:76KDTVCS","schema_version":"1.0","canonical_sha256":"ff9439d45271e8cfe3db39c471935f7a8720c1e6e979ac838e771e3e92571874","source":{"kind":"arxiv","id":"2501.16297","version":2},"attestation_state":"computed","paper":{"title":"FALCON: Resolving Visual Redundancy and Fragmentation in High-resolution Multimodal Large Language Models via Visual Registers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Gongwei Chen, Kaiwen Zhou, Liqiang Nie, Miao Zhang, Renshan Zhang, Rui Shao, Weili Guan","submitted_at":"2025-01-27T18:36:10Z","abstract_excerpt":"The incorporation of high-resolution visual input equips multimodal large language models (MLLMs) with enhanced visual perception capabilities for real-world tasks. However, most existing high-resolution MLLMs rely on a cropping-based approach to process images, which leads to fragmented visual encoding and a sharp increase in redundant tokens. To tackle these issues, we propose the FALCON model. FALCON introduces a novel visual register technique to simultaneously: 1) Eliminate redundant tokens at the stage of visual encoding. To directly address the visual redundancy present in the output of"},"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":"2501.16297","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-01-27T18:36:10Z","cross_cats_sorted":[],"title_canon_sha256":"04976555e0170af1f7fad4fb7b889f685d26c86a6b766b2246ef1399d2a21f43","abstract_canon_sha256":"6e255281523b0a8f2b752ee6584591349648e7e6951762431e130567dbdb7cb1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:29:06.081548Z","signature_b64":"1LX+7Fd8oaIJTrWfzNHPKDQS4h3EMH3TRkNl+SZSLo45rc6/3oiYzvtMHyir3npPFhqT4kv+462WZGtlhOxMCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ff9439d45271e8cfe3db39c471935f7a8720c1e6e979ac838e771e3e92571874","last_reissued_at":"2026-07-05T11:29:06.080997Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:29:06.080997Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"FALCON: Resolving Visual Redundancy and Fragmentation in High-resolution Multimodal Large Language Models via Visual Registers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Gongwei Chen, Kaiwen Zhou, Liqiang Nie, Miao Zhang, Renshan Zhang, Rui Shao, Weili Guan","submitted_at":"2025-01-27T18:36:10Z","abstract_excerpt":"The incorporation of high-resolution visual input equips multimodal large language models (MLLMs) with enhanced visual perception capabilities for real-world tasks. However, most existing high-resolution MLLMs rely on a cropping-based approach to process images, which leads to fragmented visual encoding and a sharp increase in redundant tokens. To tackle these issues, we propose the FALCON model. FALCON introduces a novel visual register technique to simultaneously: 1) Eliminate redundant tokens at the stage of visual encoding. To directly address the visual redundancy present in the output of"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.16297","kind":"arxiv","version":2},"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/2501.16297/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":"2501.16297","created_at":"2026-07-05T11:29:06.081058+00:00"},{"alias_kind":"arxiv_version","alias_value":"2501.16297v2","created_at":"2026-07-05T11:29:06.081058+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.16297","created_at":"2026-07-05T11:29:06.081058+00:00"},{"alias_kind":"pith_short_12","alias_value":"76KDTVCSOHUM","created_at":"2026-07-05T11:29:06.081058+00:00"},{"alias_kind":"pith_short_16","alias_value":"76KDTVCSOHUM7Y63","created_at":"2026-07-05T11:29:06.081058+00:00"},{"alias_kind":"pith_short_8","alias_value":"76KDTVCS","created_at":"2026-07-05T11:29:06.081058+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.25820","citing_title":"Visual-Redundancy-Controlled Parallel Decoding for Diffusion-Based Multimodal Large Language Models","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2507.23372","citing_title":"UniEmo: Unifying Emotional Understanding and Generation with Learnable Expert Queries","ref_index":54,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/76KDTVCSOHUM7Y63HHCHDE27PK","json":"https://pith.science/pith/76KDTVCSOHUM7Y63HHCHDE27PK.json","graph_json":"https://pith.science/api/pith-number/76KDTVCSOHUM7Y63HHCHDE27PK/graph.json","events_json":"https://pith.science/api/pith-number/76KDTVCSOHUM7Y63HHCHDE27PK/events.json","paper":"https://pith.science/paper/76KDTVCS"},"agent_actions":{"view_html":"https://pith.science/pith/76KDTVCSOHUM7Y63HHCHDE27PK","download_json":"https://pith.science/pith/76KDTVCSOHUM7Y63HHCHDE27PK.json","view_paper":"https://pith.science/paper/76KDTVCS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2501.16297&json=true","fetch_graph":"https://pith.science/api/pith-number/76KDTVCSOHUM7Y63HHCHDE27PK/graph.json","fetch_events":"https://pith.science/api/pith-number/76KDTVCSOHUM7Y63HHCHDE27PK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/76KDTVCSOHUM7Y63HHCHDE27PK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/76KDTVCSOHUM7Y63HHCHDE27PK/action/storage_attestation","attest_author":"https://pith.science/pith/76KDTVCSOHUM7Y63HHCHDE27PK/action/author_attestation","sign_citation":"https://pith.science/pith/76KDTVCSOHUM7Y63HHCHDE27PK/action/citation_signature","submit_replication":"https://pith.science/pith/76KDTVCSOHUM7Y63HHCHDE27PK/action/replication_record"}},"created_at":"2026-07-05T11:29:06.081058+00:00","updated_at":"2026-07-05T11:29:06.081058+00:00"}