{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:VP5AKV6WFR2APK5Z76WTYHJAA6","short_pith_number":"pith:VP5AKV6W","schema_version":"1.0","canonical_sha256":"abfa0557d62c7407abb9ffad3c1d200796bbc0a21a63721114b1998b73d7ed5c","source":{"kind":"arxiv","id":"2605.29714","version":1},"attestation_state":"computed","paper":{"title":"Leveraging Routing Dynamics in Mixture-of-Experts Models for Efficient Language Adaptation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Aditi Khandelwal, Golnoosh Farnadi, Marius Mosbach, Siva Reddy, Verna Dankers","submitted_at":"2026-05-28T10:12:41Z","abstract_excerpt":"Mixture-of-Experts (MoE) models are widely used to scale language models, yet their expert routing behavior and adaptation in a multilingual setting remain underexplored. In this work, we study multilingual routing dynamics during continual pre-training of an English-centric MoE model on a multilingual corpus, analyzing how expert usage varies across languages. We find that continual multilingual pre-training leads to diffused, language-agnostic routing in early and middle layers, with language specialization primarily emerging in the final layers. We also show that token-level vocabulary over"},"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":"2605.29714","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-05-28T10:12:41Z","cross_cats_sorted":[],"title_canon_sha256":"74de163f0a67db1fb8a3e5ffd02c5268b7bd8776663485f41d3a7da3d99f3746","abstract_canon_sha256":"02ef5049572a74725e9c3932fdff62cd4bcd3769bb93580c0eccfc9e5ff9b58c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-29T01:05:56.693309Z","signature_b64":"hVTl+tnjGUj8HK0W7PEKgcOPHkeA0J2qhxHpQwTxLpAq3ifVq+UEYbVIcifMRmSHtCa7rv+h+8vOff4J/4oZAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"abfa0557d62c7407abb9ffad3c1d200796bbc0a21a63721114b1998b73d7ed5c","last_reissued_at":"2026-05-29T01:05:56.692907Z","signature_status":"signed_v1","first_computed_at":"2026-05-29T01:05:56.692907Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Leveraging Routing Dynamics in Mixture-of-Experts Models for Efficient Language Adaptation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Aditi Khandelwal, Golnoosh Farnadi, Marius Mosbach, Siva Reddy, Verna Dankers","submitted_at":"2026-05-28T10:12:41Z","abstract_excerpt":"Mixture-of-Experts (MoE) models are widely used to scale language models, yet their expert routing behavior and adaptation in a multilingual setting remain underexplored. In this work, we study multilingual routing dynamics during continual pre-training of an English-centric MoE model on a multilingual corpus, analyzing how expert usage varies across languages. We find that continual multilingual pre-training leads to diffused, language-agnostic routing in early and middle layers, with language specialization primarily emerging in the final layers. We also show that token-level vocabulary over"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.29714","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/2605.29714/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":"2605.29714","created_at":"2026-05-29T01:05:56.692968+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.29714v1","created_at":"2026-05-29T01:05:56.692968+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.29714","created_at":"2026-05-29T01:05:56.692968+00:00"},{"alias_kind":"pith_short_12","alias_value":"VP5AKV6WFR2A","created_at":"2026-05-29T01:05:56.692968+00:00"},{"alias_kind":"pith_short_16","alias_value":"VP5AKV6WFR2APK5Z","created_at":"2026-05-29T01:05:56.692968+00:00"},{"alias_kind":"pith_short_8","alias_value":"VP5AKV6W","created_at":"2026-05-29T01:05:56.692968+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VP5AKV6WFR2APK5Z76WTYHJAA6","json":"https://pith.science/pith/VP5AKV6WFR2APK5Z76WTYHJAA6.json","graph_json":"https://pith.science/api/pith-number/VP5AKV6WFR2APK5Z76WTYHJAA6/graph.json","events_json":"https://pith.science/api/pith-number/VP5AKV6WFR2APK5Z76WTYHJAA6/events.json","paper":"https://pith.science/paper/VP5AKV6W"},"agent_actions":{"view_html":"https://pith.science/pith/VP5AKV6WFR2APK5Z76WTYHJAA6","download_json":"https://pith.science/pith/VP5AKV6WFR2APK5Z76WTYHJAA6.json","view_paper":"https://pith.science/paper/VP5AKV6W","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.29714&json=true","fetch_graph":"https://pith.science/api/pith-number/VP5AKV6WFR2APK5Z76WTYHJAA6/graph.json","fetch_events":"https://pith.science/api/pith-number/VP5AKV6WFR2APK5Z76WTYHJAA6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VP5AKV6WFR2APK5Z76WTYHJAA6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VP5AKV6WFR2APK5Z76WTYHJAA6/action/storage_attestation","attest_author":"https://pith.science/pith/VP5AKV6WFR2APK5Z76WTYHJAA6/action/author_attestation","sign_citation":"https://pith.science/pith/VP5AKV6WFR2APK5Z76WTYHJAA6/action/citation_signature","submit_replication":"https://pith.science/pith/VP5AKV6WFR2APK5Z76WTYHJAA6/action/replication_record"}},"created_at":"2026-05-29T01:05:56.692968+00:00","updated_at":"2026-05-29T01:05:56.692968+00:00"}