{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:QO7HYVEZZPTBJIDU4UT7B4T5C6","short_pith_number":"pith:QO7HYVEZ","schema_version":"1.0","canonical_sha256":"83be7c5499cbe614a074e527f0f27d17aa9266c0fc31219cf90a072022f83a8a","source":{"kind":"arxiv","id":"2605.27358","version":1},"attestation_state":"computed","paper":{"title":"MobileMoE: Scaling On-Device Mixture of Experts","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Digant Desai, Ernie Chang, Hanxian Huang, Jacob Szwejbka, Raghuraman Krishnamoorthi, Vikas Chandra, Yanbei Chen, Zechun Liu","submitted_at":"2026-05-26T17:58:24Z","abstract_excerpt":"Mixture-of-Experts (MoE) has become the de facto architecture for hundred-billion-parameter language models, yet its advantages at sub-billion scales for on-device deployment remain largely unexplored. To close this gap, we present MobileMoE, a family of on-device MoE language models with sub-billion active parameters (0.3-0.9B active and 1.3-5.3B total) that establish a new Pareto frontier for on-device LLMs. We first formulate an on-device MoE scaling law that jointly optimizes MoE architecture under mobile memory and compute constraints, identifying an on-device sweet spot - moderate sparsi"},"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":"2605.27358","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-26T17:58:24Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"fe98159ccac2c5876bf9d759c48e544a34cb366ce9c67efbca428a557caac8a2","abstract_canon_sha256":"58d2507f2338fa70678d5aa4d15a28564d78d82c73bbc41bf1a29303f916b103"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-27T02:06:19.749857Z","signature_b64":"H8k3Evb5pquoZMDFq6/UPmUQe2diruC1RH+II/rWw2Dmg8TXqbDXLksS+CEsu7tO+T/VM/H07jKGAXgISP+XAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"83be7c5499cbe614a074e527f0f27d17aa9266c0fc31219cf90a072022f83a8a","last_reissued_at":"2026-05-27T02:06:19.749202Z","signature_status":"signed_v1","first_computed_at":"2026-05-27T02:06:19.749202Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MobileMoE: Scaling On-Device Mixture of Experts","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Digant Desai, Ernie Chang, Hanxian Huang, Jacob Szwejbka, Raghuraman Krishnamoorthi, Vikas Chandra, Yanbei Chen, Zechun Liu","submitted_at":"2026-05-26T17:58:24Z","abstract_excerpt":"Mixture-of-Experts (MoE) has become the de facto architecture for hundred-billion-parameter language models, yet its advantages at sub-billion scales for on-device deployment remain largely unexplored. To close this gap, we present MobileMoE, a family of on-device MoE language models with sub-billion active parameters (0.3-0.9B active and 1.3-5.3B total) that establish a new Pareto frontier for on-device LLMs. We first formulate an on-device MoE scaling law that jointly optimizes MoE architecture under mobile memory and compute constraints, identifying an on-device sweet spot - moderate sparsi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.27358","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/2605.27358/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":"2605.27358","created_at":"2026-05-27T02:06:19.749330+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.27358v1","created_at":"2026-05-27T02:06:19.749330+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.27358","created_at":"2026-05-27T02:06:19.749330+00:00"},{"alias_kind":"pith_short_12","alias_value":"QO7HYVEZZPTB","created_at":"2026-05-27T02:06:19.749330+00:00"},{"alias_kind":"pith_short_16","alias_value":"QO7HYVEZZPTBJIDU","created_at":"2026-05-27T02:06:19.749330+00:00"},{"alias_kind":"pith_short_8","alias_value":"QO7HYVEZ","created_at":"2026-05-27T02:06:19.749330+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/QO7HYVEZZPTBJIDU4UT7B4T5C6","json":"https://pith.science/pith/QO7HYVEZZPTBJIDU4UT7B4T5C6.json","graph_json":"https://pith.science/api/pith-number/QO7HYVEZZPTBJIDU4UT7B4T5C6/graph.json","events_json":"https://pith.science/api/pith-number/QO7HYVEZZPTBJIDU4UT7B4T5C6/events.json","paper":"https://pith.science/paper/QO7HYVEZ"},"agent_actions":{"view_html":"https://pith.science/pith/QO7HYVEZZPTBJIDU4UT7B4T5C6","download_json":"https://pith.science/pith/QO7HYVEZZPTBJIDU4UT7B4T5C6.json","view_paper":"https://pith.science/paper/QO7HYVEZ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.27358&json=true","fetch_graph":"https://pith.science/api/pith-number/QO7HYVEZZPTBJIDU4UT7B4T5C6/graph.json","fetch_events":"https://pith.science/api/pith-number/QO7HYVEZZPTBJIDU4UT7B4T5C6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QO7HYVEZZPTBJIDU4UT7B4T5C6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QO7HYVEZZPTBJIDU4UT7B4T5C6/action/storage_attestation","attest_author":"https://pith.science/pith/QO7HYVEZZPTBJIDU4UT7B4T5C6/action/author_attestation","sign_citation":"https://pith.science/pith/QO7HYVEZZPTBJIDU4UT7B4T5C6/action/citation_signature","submit_replication":"https://pith.science/pith/QO7HYVEZZPTBJIDU4UT7B4T5C6/action/replication_record"}},"created_at":"2026-05-27T02:06:19.749330+00:00","updated_at":"2026-05-27T02:06:19.749330+00:00"}