{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:AAM4JB7AVBBI4DI6WCXNJIULZX","short_pith_number":"pith:AAM4JB7A","schema_version":"1.0","canonical_sha256":"0019c487e0a8428e0d1eb0aed4a28bcde204e9ff407359f4803225780d5485da","source":{"kind":"arxiv","id":"2407.14093","version":3},"attestation_state":"computed","paper":{"title":"Routing Experts: Learning to Route Dynamic Experts in Multi-modal Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.MM","authors_text":"Qiong Wu, Rongrong Ji, Xiaoshuai Sun, Yiyi Zhou, Zhaoxi Ke","submitted_at":"2024-07-19T07:57:48Z","abstract_excerpt":"Recently, mixture of experts (MoE) has become a popular paradigm for achieving the trade-off between modal capacity and efficiency of multi-modal large language models (MLLMs). Different from previous efforts, we are dedicated to exploring the dynamic expert path in an already exist MLLM and show that a standard MLLM can be also a mixture of experts. To approach this target, we propose a novel dynamic expert scheme for MLLMs, termed Routing Experts (RoE), which can achieve example-dependent optimal path routing without obvious structure tweaks. Meanwhile, a new regularization of structure spar"},"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":"2407.14093","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.MM","submitted_at":"2024-07-19T07:57:48Z","cross_cats_sorted":[],"title_canon_sha256":"d0ce5ff47f516dd0b78d226a2f3e06feebb7f0268e3b9421866d2a77c041fbe6","abstract_canon_sha256":"787636bfb5228045371f5ca64db022732faa5c8be0350414ecc340fefb9224e3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:13:01.971279Z","signature_b64":"mSi2C9aNu73t2TWlAqCzwn8uux1CgLNIPZWJxidhK6mIBqxchgsAX0IWzCjqylHIwHM8ceKHyDGGfJVE/CvyCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0019c487e0a8428e0d1eb0aed4a28bcde204e9ff407359f4803225780d5485da","last_reissued_at":"2026-07-05T10:13:01.970775Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:13:01.970775Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Routing Experts: Learning to Route Dynamic Experts in Multi-modal Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.MM","authors_text":"Qiong Wu, Rongrong Ji, Xiaoshuai Sun, Yiyi Zhou, Zhaoxi Ke","submitted_at":"2024-07-19T07:57:48Z","abstract_excerpt":"Recently, mixture of experts (MoE) has become a popular paradigm for achieving the trade-off between modal capacity and efficiency of multi-modal large language models (MLLMs). Different from previous efforts, we are dedicated to exploring the dynamic expert path in an already exist MLLM and show that a standard MLLM can be also a mixture of experts. To approach this target, we propose a novel dynamic expert scheme for MLLMs, termed Routing Experts (RoE), which can achieve example-dependent optimal path routing without obvious structure tweaks. Meanwhile, a new regularization of structure spar"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.14093","kind":"arxiv","version":3},"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/2407.14093/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":"2407.14093","created_at":"2026-07-05T10:13:01.970842+00:00"},{"alias_kind":"arxiv_version","alias_value":"2407.14093v3","created_at":"2026-07-05T10:13:01.970842+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.14093","created_at":"2026-07-05T10:13:01.970842+00:00"},{"alias_kind":"pith_short_12","alias_value":"AAM4JB7AVBBI","created_at":"2026-07-05T10:13:01.970842+00:00"},{"alias_kind":"pith_short_16","alias_value":"AAM4JB7AVBBI4DI6","created_at":"2026-07-05T10:13:01.970842+00:00"},{"alias_kind":"pith_short_8","alias_value":"AAM4JB7A","created_at":"2026-07-05T10:13:01.970842+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.06574","citing_title":"Skip a Layer or Loop It? Learning Program-of-Layers in LLMs","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2604.19016","citing_title":"AlignCultura: Towards Culturally Aligned Large Language Models?","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2604.16930","citing_title":"CoGR-MoE: Concept-Guided Expert Routing with Consistent Selection and Flexible Reasoning for Visual Question Answering","ref_index":35,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/AAM4JB7AVBBI4DI6WCXNJIULZX","json":"https://pith.science/pith/AAM4JB7AVBBI4DI6WCXNJIULZX.json","graph_json":"https://pith.science/api/pith-number/AAM4JB7AVBBI4DI6WCXNJIULZX/graph.json","events_json":"https://pith.science/api/pith-number/AAM4JB7AVBBI4DI6WCXNJIULZX/events.json","paper":"https://pith.science/paper/AAM4JB7A"},"agent_actions":{"view_html":"https://pith.science/pith/AAM4JB7AVBBI4DI6WCXNJIULZX","download_json":"https://pith.science/pith/AAM4JB7AVBBI4DI6WCXNJIULZX.json","view_paper":"https://pith.science/paper/AAM4JB7A","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2407.14093&json=true","fetch_graph":"https://pith.science/api/pith-number/AAM4JB7AVBBI4DI6WCXNJIULZX/graph.json","fetch_events":"https://pith.science/api/pith-number/AAM4JB7AVBBI4DI6WCXNJIULZX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AAM4JB7AVBBI4DI6WCXNJIULZX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AAM4JB7AVBBI4DI6WCXNJIULZX/action/storage_attestation","attest_author":"https://pith.science/pith/AAM4JB7AVBBI4DI6WCXNJIULZX/action/author_attestation","sign_citation":"https://pith.science/pith/AAM4JB7AVBBI4DI6WCXNJIULZX/action/citation_signature","submit_replication":"https://pith.science/pith/AAM4JB7AVBBI4DI6WCXNJIULZX/action/replication_record"}},"created_at":"2026-07-05T10:13:01.970842+00:00","updated_at":"2026-07-05T10:13:01.970842+00:00"}