{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:LUCUYP67UQRLER75NHMUFDI4WM","short_pith_number":"pith:LUCUYP67","schema_version":"1.0","canonical_sha256":"5d054c3fdfa422b247fd69d9428d1cb332366556500c3d30743fbec59141182c","source":{"kind":"arxiv","id":"2607.08780","version":1},"attestation_state":"computed","paper":{"title":"Sticky Routing: Training MoE Models for Memory-Efficient Inference","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Ali Kayyam","submitted_at":"2026-06-12T23:47:19Z","abstract_excerpt":"Mixture-of-Experts (MoE) models activate only a sparse subset of experts per token, yet consecutive tokens frequently activate different experts -- causing constant weight swapping between slow storage and fast memory on edge devices. Existing remedies are either system-level (caching heuristics) or post-hoc (router fine-tuning), leaving the root cause unchanged during pretraining. We propose StickyMoE, a differentiable routing consistency loss that penalises abrupt expert switches between adjacent tokens, encouraging the router to maintain the same expert assignment across semantically cohere"},"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":"2607.08780","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-06-12T23:47:19Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"9c726cf1d3c0ca9e3c2743b110cf0a1971d18eddb29e06aa384655bc4ad6bc4a","abstract_canon_sha256":"be416e6b0eaf3cd2b2936ad947acbbd6fe2be1364efef5cb9d030c94737a5e82"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-13T00:17:16.014853Z","signature_b64":"wkHT2VBR+VuXF29K+d4BYiuuPVRQ+MedWanhPh1YwzHiWBontOVTqzW6R1ECqSUkGRBxyfKkbITchKCmUs/+DA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5d054c3fdfa422b247fd69d9428d1cb332366556500c3d30743fbec59141182c","last_reissued_at":"2026-07-13T00:17:16.013715Z","signature_status":"signed_v1","first_computed_at":"2026-07-13T00:17:16.013715Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Sticky Routing: Training MoE Models for Memory-Efficient Inference","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Ali Kayyam","submitted_at":"2026-06-12T23:47:19Z","abstract_excerpt":"Mixture-of-Experts (MoE) models activate only a sparse subset of experts per token, yet consecutive tokens frequently activate different experts -- causing constant weight swapping between slow storage and fast memory on edge devices. Existing remedies are either system-level (caching heuristics) or post-hoc (router fine-tuning), leaving the root cause unchanged during pretraining. We propose StickyMoE, a differentiable routing consistency loss that penalises abrupt expert switches between adjacent tokens, encouraging the router to maintain the same expert assignment across semantically cohere"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.08780","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/2607.08780/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":"2607.08780","created_at":"2026-07-13T00:17:16.014226+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.08780v1","created_at":"2026-07-13T00:17:16.014226+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.08780","created_at":"2026-07-13T00:17:16.014226+00:00"},{"alias_kind":"pith_short_12","alias_value":"LUCUYP67UQRL","created_at":"2026-07-13T00:17:16.014226+00:00"},{"alias_kind":"pith_short_16","alias_value":"LUCUYP67UQRLER75","created_at":"2026-07-13T00:17:16.014226+00:00"},{"alias_kind":"pith_short_8","alias_value":"LUCUYP67","created_at":"2026-07-13T00:17:16.014226+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/LUCUYP67UQRLER75NHMUFDI4WM","json":"https://pith.science/pith/LUCUYP67UQRLER75NHMUFDI4WM.json","graph_json":"https://pith.science/api/pith-number/LUCUYP67UQRLER75NHMUFDI4WM/graph.json","events_json":"https://pith.science/api/pith-number/LUCUYP67UQRLER75NHMUFDI4WM/events.json","paper":"https://pith.science/paper/LUCUYP67"},"agent_actions":{"view_html":"https://pith.science/pith/LUCUYP67UQRLER75NHMUFDI4WM","download_json":"https://pith.science/pith/LUCUYP67UQRLER75NHMUFDI4WM.json","view_paper":"https://pith.science/paper/LUCUYP67","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.08780&json=true","fetch_graph":"https://pith.science/api/pith-number/LUCUYP67UQRLER75NHMUFDI4WM/graph.json","fetch_events":"https://pith.science/api/pith-number/LUCUYP67UQRLER75NHMUFDI4WM/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LUCUYP67UQRLER75NHMUFDI4WM/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LUCUYP67UQRLER75NHMUFDI4WM/action/storage_attestation","attest_author":"https://pith.science/pith/LUCUYP67UQRLER75NHMUFDI4WM/action/author_attestation","sign_citation":"https://pith.science/pith/LUCUYP67UQRLER75NHMUFDI4WM/action/citation_signature","submit_replication":"https://pith.science/pith/LUCUYP67UQRLER75NHMUFDI4WM/action/replication_record"}},"created_at":"2026-07-13T00:17:16.014226+00:00","updated_at":"2026-07-13T00:17:16.014226+00:00"}