{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:HAHOFQKPVLN52IURQMDILAZSQQ","short_pith_number":"pith:HAHOFQKP","schema_version":"1.0","canonical_sha256":"380ee2c14faadbdd22918306858332841935ae213a16bbdfce94f40dccd1606b","source":{"kind":"arxiv","id":"2607.00466","version":1},"attestation_state":"computed","paper":{"title":"ELDR: Expert-Locality-Aware Decode Routing for PD-Disaggregated MoE Serving","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.DC","authors_text":"Peng Cheng, Sangjin Choi, Sukmin Cho, Yifan Xiong, Youngjin Kwon, Ziyue Yang","submitted_at":"2026-07-01T05:34:38Z","abstract_excerpt":"In prefill-decode (PD) disaggregated LLM serving, each request is assigned to a decode worker after prefill. Existing decode routers balance only load; for mixture-of-experts (MoE) models this is incomplete: equally loaded workers can differ in latency, since each decode step loads the weights of every distinct expert its batch activates. We present ELDR, an expert-locality-aware decode router for PD-disaggregated MoE serving. From a request's prefill expert activations, ELDR builds an expert signature predicting the experts it will activate during generation. Offline, balanced K-means partiti"},"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":"2607.00466","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2026-07-01T05:34:38Z","cross_cats_sorted":[],"title_canon_sha256":"bbf1b73e7a954f028ce19ec931fceadb8206e5fa6975edb31de71269aa63b6e6","abstract_canon_sha256":"1f262a7e0f3d26442de3ef76d394e62510971f95d47693704c604b082b668d9c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-02T01:17:44.246854Z","signature_b64":"DgEUMJZv1oSOCovDTOSYBQaK6MSRqY5T6l5tE7I7vYc10OJ8LLGGPUdGpDzdiVF2VcCAetgI46gcUWwOnoF/Cw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"380ee2c14faadbdd22918306858332841935ae213a16bbdfce94f40dccd1606b","last_reissued_at":"2026-07-02T01:17:44.246477Z","signature_status":"signed_v1","first_computed_at":"2026-07-02T01:17:44.246477Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ELDR: Expert-Locality-Aware Decode Routing for PD-Disaggregated MoE Serving","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.DC","authors_text":"Peng Cheng, Sangjin Choi, Sukmin Cho, Yifan Xiong, Youngjin Kwon, Ziyue Yang","submitted_at":"2026-07-01T05:34:38Z","abstract_excerpt":"In prefill-decode (PD) disaggregated LLM serving, each request is assigned to a decode worker after prefill. Existing decode routers balance only load; for mixture-of-experts (MoE) models this is incomplete: equally loaded workers can differ in latency, since each decode step loads the weights of every distinct expert its batch activates. We present ELDR, an expert-locality-aware decode router for PD-disaggregated MoE serving. From a request's prefill expert activations, ELDR builds an expert signature predicting the experts it will activate during generation. Offline, balanced K-means partiti"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.00466","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/2607.00466/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":"2607.00466","created_at":"2026-07-02T01:17:44.246531+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.00466v1","created_at":"2026-07-02T01:17:44.246531+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.00466","created_at":"2026-07-02T01:17:44.246531+00:00"},{"alias_kind":"pith_short_12","alias_value":"HAHOFQKPVLN5","created_at":"2026-07-02T01:17:44.246531+00:00"},{"alias_kind":"pith_short_16","alias_value":"HAHOFQKPVLN52IUR","created_at":"2026-07-02T01:17:44.246531+00:00"},{"alias_kind":"pith_short_8","alias_value":"HAHOFQKP","created_at":"2026-07-02T01:17:44.246531+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/HAHOFQKPVLN52IURQMDILAZSQQ","json":"https://pith.science/pith/HAHOFQKPVLN52IURQMDILAZSQQ.json","graph_json":"https://pith.science/api/pith-number/HAHOFQKPVLN52IURQMDILAZSQQ/graph.json","events_json":"https://pith.science/api/pith-number/HAHOFQKPVLN52IURQMDILAZSQQ/events.json","paper":"https://pith.science/paper/HAHOFQKP"},"agent_actions":{"view_html":"https://pith.science/pith/HAHOFQKPVLN52IURQMDILAZSQQ","download_json":"https://pith.science/pith/HAHOFQKPVLN52IURQMDILAZSQQ.json","view_paper":"https://pith.science/paper/HAHOFQKP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.00466&json=true","fetch_graph":"https://pith.science/api/pith-number/HAHOFQKPVLN52IURQMDILAZSQQ/graph.json","fetch_events":"https://pith.science/api/pith-number/HAHOFQKPVLN52IURQMDILAZSQQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HAHOFQKPVLN52IURQMDILAZSQQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HAHOFQKPVLN52IURQMDILAZSQQ/action/storage_attestation","attest_author":"https://pith.science/pith/HAHOFQKPVLN52IURQMDILAZSQQ/action/author_attestation","sign_citation":"https://pith.science/pith/HAHOFQKPVLN52IURQMDILAZSQQ/action/citation_signature","submit_replication":"https://pith.science/pith/HAHOFQKPVLN52IURQMDILAZSQQ/action/replication_record"}},"created_at":"2026-07-02T01:17:44.246531+00:00","updated_at":"2026-07-02T01:17:44.246531+00:00"}