{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:I5DNKIGPUA2CK2XIKTF46DHYIL","short_pith_number":"pith:I5DNKIGP","schema_version":"1.0","canonical_sha256":"4746d520cfa034256ae854cbcf0cf842d80cf951b63d14bc1d9f02968280240f","source":{"kind":"arxiv","id":"2401.15969","version":2},"attestation_state":"computed","paper":{"title":"Routers in Vision Mixture of Experts: An Empirical Study","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Carlos Riquelme, Joan Puigcerver, Mathieu Blondel, Tianlin Liu","submitted_at":"2024-01-29T08:58:07Z","abstract_excerpt":"Mixture-of-Experts (MoE) models are a promising way to scale up model capacity without significantly increasing computational cost. A key component of MoEs is the router, which decides which subset of parameters (experts) process which feature embeddings (tokens). In this paper, we present a comprehensive study of routers in MoEs for computer vision tasks. We introduce a unified MoE formulation that subsumes different MoEs with two parametric routing tensors. This formulation covers both sparse MoE, which uses a binary or hard assignment between experts and tokens, and soft MoE, which uses a s"},"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":"2401.15969","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2024-01-29T08:58:07Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"0615ff91657f7d992213fbada0cb78067001d68220ccad4a18c3c9fb688611ff","abstract_canon_sha256":"d06b74403d234911d169d359db43733c4852dc87d0152aad00146e8f83835ec5"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:09:48.096277Z","signature_b64":"7ChBHPSaYgUdA91Uc0GH2iwIkMdawra87oW6yDa8Jfbu2M8JW4+ScinlRuZjL+jLt5Qmj6X15D04T+Oa4U4ZBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4746d520cfa034256ae854cbcf0cf842d80cf951b63d14bc1d9f02968280240f","last_reissued_at":"2026-07-05T08:09:48.095739Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:09:48.095739Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Routers in Vision Mixture of Experts: An Empirical Study","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CV","authors_text":"Carlos Riquelme, Joan Puigcerver, Mathieu Blondel, Tianlin Liu","submitted_at":"2024-01-29T08:58:07Z","abstract_excerpt":"Mixture-of-Experts (MoE) models are a promising way to scale up model capacity without significantly increasing computational cost. A key component of MoEs is the router, which decides which subset of parameters (experts) process which feature embeddings (tokens). In this paper, we present a comprehensive study of routers in MoEs for computer vision tasks. We introduce a unified MoE formulation that subsumes different MoEs with two parametric routing tensors. This formulation covers both sparse MoE, which uses a binary or hard assignment between experts and tokens, and soft MoE, which uses a s"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2401.15969","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/2401.15969/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":"2401.15969","created_at":"2026-07-05T08:09:48.095816+00:00"},{"alias_kind":"arxiv_version","alias_value":"2401.15969v2","created_at":"2026-07-05T08:09:48.095816+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2401.15969","created_at":"2026-07-05T08:09:48.095816+00:00"},{"alias_kind":"pith_short_12","alias_value":"I5DNKIGPUA2C","created_at":"2026-07-05T08:09:48.095816+00:00"},{"alias_kind":"pith_short_16","alias_value":"I5DNKIGPUA2CK2XI","created_at":"2026-07-05T08:09:48.095816+00:00"},{"alias_kind":"pith_short_8","alias_value":"I5DNKIGP","created_at":"2026-07-05T08:09:48.095816+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.31397","citing_title":"Mixture-of-Control: State-Aware Fine-Tuning for Transformer-based Models","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04058","citing_title":"MP-ISMoE: Mixed-Precision Interactive Side Mixture-of-Experts for Efficient Transfer Learning","ref_index":146,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06708","citing_title":"Visual Text Compression as Measure Transport","ref_index":26,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/I5DNKIGPUA2CK2XIKTF46DHYIL","json":"https://pith.science/pith/I5DNKIGPUA2CK2XIKTF46DHYIL.json","graph_json":"https://pith.science/api/pith-number/I5DNKIGPUA2CK2XIKTF46DHYIL/graph.json","events_json":"https://pith.science/api/pith-number/I5DNKIGPUA2CK2XIKTF46DHYIL/events.json","paper":"https://pith.science/paper/I5DNKIGP"},"agent_actions":{"view_html":"https://pith.science/pith/I5DNKIGPUA2CK2XIKTF46DHYIL","download_json":"https://pith.science/pith/I5DNKIGPUA2CK2XIKTF46DHYIL.json","view_paper":"https://pith.science/paper/I5DNKIGP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2401.15969&json=true","fetch_graph":"https://pith.science/api/pith-number/I5DNKIGPUA2CK2XIKTF46DHYIL/graph.json","fetch_events":"https://pith.science/api/pith-number/I5DNKIGPUA2CK2XIKTF46DHYIL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/I5DNKIGPUA2CK2XIKTF46DHYIL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/I5DNKIGPUA2CK2XIKTF46DHYIL/action/storage_attestation","attest_author":"https://pith.science/pith/I5DNKIGPUA2CK2XIKTF46DHYIL/action/author_attestation","sign_citation":"https://pith.science/pith/I5DNKIGPUA2CK2XIKTF46DHYIL/action/citation_signature","submit_replication":"https://pith.science/pith/I5DNKIGPUA2CK2XIKTF46DHYIL/action/replication_record"}},"created_at":"2026-07-05T08:09:48.095816+00:00","updated_at":"2026-07-05T08:09:48.095816+00:00"}