{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:32XD3NHYL3PZEMFHWT55H6BYCS","short_pith_number":"pith:32XD3NHY","schema_version":"1.0","canonical_sha256":"deae3db4f85edf9230a7b4fbd3f83814bdc1e85fec30bcd4ded720bb8fcd0fd1","source":{"kind":"arxiv","id":"2406.11193","version":2},"attestation_state":"computed","paper":{"title":"MMNeuron: Discovering Neuron-Level Domain-Specific Interpretation in Multimodal Large Language Model","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Boren Hu, Jiahao Huo, Xuming Hu, Yibo Yan, Yutao Yue","submitted_at":"2024-06-17T03:59:44Z","abstract_excerpt":"Projecting visual features into word embedding space has become a significant fusion strategy adopted by Multimodal Large Language Models (MLLMs). However, its internal mechanisms have yet to be explored. Inspired by multilingual research, we identify domain-specific neurons in multimodal large language models. Specifically, we investigate the distribution of domain-specific neurons and the mechanism of how MLLMs process features from diverse domains. Furthermore, we propose a three-stage mechanism for language model modules in MLLMs when handling projected image features, and verify this hypo"},"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":"2406.11193","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-06-17T03:59:44Z","cross_cats_sorted":[],"title_canon_sha256":"8b223cfd46c9dd108ccbd5db1f057dda9e54f682b4643f21a95c9afec8d42b24","abstract_canon_sha256":"45c99d2ff2b97a70807ccc0fb773e2aaba84f7b57ff84bffdf5c800fdcf82dbb"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:06:08.192045Z","signature_b64":"BCX+cJt1eOgnWdVTOhVzZPgB9OBWGm8ZfjwFNQkaxuU7LGQ0/IYqP9hhDEsK/2wg8Sb/RiIDtFQMOtWT0hdZCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"deae3db4f85edf9230a7b4fbd3f83814bdc1e85fec30bcd4ded720bb8fcd0fd1","last_reissued_at":"2026-07-05T11:06:08.191556Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:06:08.191556Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MMNeuron: Discovering Neuron-Level Domain-Specific Interpretation in Multimodal Large Language Model","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Boren Hu, Jiahao Huo, Xuming Hu, Yibo Yan, Yutao Yue","submitted_at":"2024-06-17T03:59:44Z","abstract_excerpt":"Projecting visual features into word embedding space has become a significant fusion strategy adopted by Multimodal Large Language Models (MLLMs). However, its internal mechanisms have yet to be explored. Inspired by multilingual research, we identify domain-specific neurons in multimodal large language models. Specifically, we investigate the distribution of domain-specific neurons and the mechanism of how MLLMs process features from diverse domains. Furthermore, we propose a three-stage mechanism for language model modules in MLLMs when handling projected image features, and verify this hypo"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.11193","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/2406.11193/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":"2406.11193","created_at":"2026-07-05T11:06:08.191614+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.11193v2","created_at":"2026-07-05T11:06:08.191614+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.11193","created_at":"2026-07-05T11:06:08.191614+00:00"},{"alias_kind":"pith_short_12","alias_value":"32XD3NHYL3PZ","created_at":"2026-07-05T11:06:08.191614+00:00"},{"alias_kind":"pith_short_16","alias_value":"32XD3NHYL3PZEMFH","created_at":"2026-07-05T11:06:08.191614+00:00"},{"alias_kind":"pith_short_8","alias_value":"32XD3NHY","created_at":"2026-07-05T11:06:08.191614+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2410.04509","citing_title":"ErrorRadar: Benchmarking Complex Mathematical Reasoning of Multimodal Large Language Models Via Error Detection","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2502.02871","citing_title":"Position: Multimodal Large Language Models Can Significantly Advance Scientific Reasoning","ref_index":73,"is_internal_anchor":false},{"citing_arxiv_id":"2509.14837","citing_title":"V-SEAM: Visual Semantic Editing and Attention Modulating for Causal Interpretability of Vision-Language Models","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05909","citing_title":"Null Space Constrained Contrastive Visual Forgetting for MLLM Unlearning","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04641","citing_title":"CAST: Mitigating Object Hallucination in Large Vision-Language Models via Caption-Guided Visual Attention Steering","ref_index":27,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/32XD3NHYL3PZEMFHWT55H6BYCS","json":"https://pith.science/pith/32XD3NHYL3PZEMFHWT55H6BYCS.json","graph_json":"https://pith.science/api/pith-number/32XD3NHYL3PZEMFHWT55H6BYCS/graph.json","events_json":"https://pith.science/api/pith-number/32XD3NHYL3PZEMFHWT55H6BYCS/events.json","paper":"https://pith.science/paper/32XD3NHY"},"agent_actions":{"view_html":"https://pith.science/pith/32XD3NHYL3PZEMFHWT55H6BYCS","download_json":"https://pith.science/pith/32XD3NHYL3PZEMFHWT55H6BYCS.json","view_paper":"https://pith.science/paper/32XD3NHY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.11193&json=true","fetch_graph":"https://pith.science/api/pith-number/32XD3NHYL3PZEMFHWT55H6BYCS/graph.json","fetch_events":"https://pith.science/api/pith-number/32XD3NHYL3PZEMFHWT55H6BYCS/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/32XD3NHYL3PZEMFHWT55H6BYCS/action/timestamp_anchor","attest_storage":"https://pith.science/pith/32XD3NHYL3PZEMFHWT55H6BYCS/action/storage_attestation","attest_author":"https://pith.science/pith/32XD3NHYL3PZEMFHWT55H6BYCS/action/author_attestation","sign_citation":"https://pith.science/pith/32XD3NHYL3PZEMFHWT55H6BYCS/action/citation_signature","submit_replication":"https://pith.science/pith/32XD3NHYL3PZEMFHWT55H6BYCS/action/replication_record"}},"created_at":"2026-07-05T11:06:08.191614+00:00","updated_at":"2026-07-05T11:06:08.191614+00:00"}