{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:GM2RMMLVYD6GTKX24DQ5AEYZM5","short_pith_number":"pith:GM2RMMLV","schema_version":"1.0","canonical_sha256":"3335163175c0fc69aafae0e1d01319675b8bc1fd9210e536ebfba77204c25bfc","source":{"kind":"arxiv","id":"2506.16691","version":1},"attestation_state":"computed","paper":{"title":"LaVi: Efficient Large Vision-Language Models via Internal Feature Modulation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Hua Huang, Jing Liu, Longteng Guo, Tongtian Yue, Xinxin Zhu, Yepeng Tang, Zijia Zhao","submitted_at":"2025-06-20T02:25:33Z","abstract_excerpt":"Despite the impressive advancements of Large Vision-Language Models (LVLMs), existing approaches suffer from a fundamental bottleneck: inefficient visual-language integration. Current methods either disrupt the model's inherent structure or introduce severe long-context computational burden, severely limiting scalability and efficiency. In this paper, we rethink multimodal integration and present LaVi, a novel LVLM that enables seamless and efficient vision-language fusion through internal feature modulation within the Large Language Models (LLMs). Unlike dominant LVLMs that rely on visual tok"},"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":"2506.16691","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-06-20T02:25:33Z","cross_cats_sorted":[],"title_canon_sha256":"341a173feebcc7a0da1224f357a0f568b9351345a39169ccd999b26336beef5f","abstract_canon_sha256":"6bead6354b3f355a80bc90e12a4a600ab727720bbc93e67777bc92ec539d13ae"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:24:40.960401Z","signature_b64":"z2FQcFXyHWprJyF5uwJFUzUFwQvi7gybZLfQCI9pO5G5bvFpvx2vlpCXT3iBjJCmbFJDf+P5dsefOdCtjgvsBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3335163175c0fc69aafae0e1d01319675b8bc1fd9210e536ebfba77204c25bfc","last_reissued_at":"2026-07-05T11:24:40.959895Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:24:40.959895Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LaVi: Efficient Large Vision-Language Models via Internal Feature Modulation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Hua Huang, Jing Liu, Longteng Guo, Tongtian Yue, Xinxin Zhu, Yepeng Tang, Zijia Zhao","submitted_at":"2025-06-20T02:25:33Z","abstract_excerpt":"Despite the impressive advancements of Large Vision-Language Models (LVLMs), existing approaches suffer from a fundamental bottleneck: inefficient visual-language integration. Current methods either disrupt the model's inherent structure or introduce severe long-context computational burden, severely limiting scalability and efficiency. In this paper, we rethink multimodal integration and present LaVi, a novel LVLM that enables seamless and efficient vision-language fusion through internal feature modulation within the Large Language Models (LLMs). Unlike dominant LVLMs that rely on visual tok"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.16691","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/2506.16691/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":"2506.16691","created_at":"2026-07-05T11:24:40.959951+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.16691v1","created_at":"2026-07-05T11:24:40.959951+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.16691","created_at":"2026-07-05T11:24:40.959951+00:00"},{"alias_kind":"pith_short_12","alias_value":"GM2RMMLVYD6G","created_at":"2026-07-05T11:24:40.959951+00:00"},{"alias_kind":"pith_short_16","alias_value":"GM2RMMLVYD6GTKX2","created_at":"2026-07-05T11:24:40.959951+00:00"},{"alias_kind":"pith_short_8","alias_value":"GM2RMMLV","created_at":"2026-07-05T11:24:40.959951+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.20077","citing_title":"The Hidden Evolution of Disguised Visual Context inside the VLM","ref_index":76,"is_internal_anchor":false},{"citing_arxiv_id":"2507.21420","citing_title":"ReGATE: Learning Faster and Better with Fewer Tokens in MLLMs","ref_index":43,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GM2RMMLVYD6GTKX24DQ5AEYZM5","json":"https://pith.science/pith/GM2RMMLVYD6GTKX24DQ5AEYZM5.json","graph_json":"https://pith.science/api/pith-number/GM2RMMLVYD6GTKX24DQ5AEYZM5/graph.json","events_json":"https://pith.science/api/pith-number/GM2RMMLVYD6GTKX24DQ5AEYZM5/events.json","paper":"https://pith.science/paper/GM2RMMLV"},"agent_actions":{"view_html":"https://pith.science/pith/GM2RMMLVYD6GTKX24DQ5AEYZM5","download_json":"https://pith.science/pith/GM2RMMLVYD6GTKX24DQ5AEYZM5.json","view_paper":"https://pith.science/paper/GM2RMMLV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.16691&json=true","fetch_graph":"https://pith.science/api/pith-number/GM2RMMLVYD6GTKX24DQ5AEYZM5/graph.json","fetch_events":"https://pith.science/api/pith-number/GM2RMMLVYD6GTKX24DQ5AEYZM5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GM2RMMLVYD6GTKX24DQ5AEYZM5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GM2RMMLVYD6GTKX24DQ5AEYZM5/action/storage_attestation","attest_author":"https://pith.science/pith/GM2RMMLVYD6GTKX24DQ5AEYZM5/action/author_attestation","sign_citation":"https://pith.science/pith/GM2RMMLVYD6GTKX24DQ5AEYZM5/action/citation_signature","submit_replication":"https://pith.science/pith/GM2RMMLVYD6GTKX24DQ5AEYZM5/action/replication_record"}},"created_at":"2026-07-05T11:24:40.959951+00:00","updated_at":"2026-07-05T11:24:40.959951+00:00"}