{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:RPK6NES4K3U4EJVSPPTW6MWZEQ","short_pith_number":"pith:RPK6NES4","schema_version":"1.0","canonical_sha256":"8bd5e6925c56e9c226b27be76f32d9243fbf9464358357376743a8452a98eff8","source":{"kind":"arxiv","id":"2407.04153","version":1},"attestation_state":"computed","paper":{"title":"Mixture of A Million Experts","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Xu Owen He","submitted_at":"2024-07-04T20:59:20Z","abstract_excerpt":"The feedforward (FFW) layers in standard transformer architectures incur a linear increase in computational costs and activation memory as the hidden layer width grows. Sparse mixture-of-experts (MoE) architectures have emerged as a viable approach to address this issue by decoupling model size from computational cost. The recent discovery of the fine-grained MoE scaling law shows that higher granularity leads to better performance. However, existing MoE models are limited to a small number of experts due to computational and optimization challenges. This paper introduces PEER (parameter effic"},"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":"2407.04153","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-07-04T20:59:20Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"ccde7b14345b0145dc4727401939c884df1d155c62f9154a4f93473acee50a20","abstract_canon_sha256":"4302638db42088109bb267acb950ae0401978fc4dd5e5f457e8a8ff02f5a31d6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:40:29.888991Z","signature_b64":"lV6Ra0qXOm9AALxPaVDgCiKJuHG8QqY1MIs/C9q1P9kskV0rx2U1T5mAocawD5VgbqNF/6/5YhfZMjZ5+/d1AA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8bd5e6925c56e9c226b27be76f32d9243fbf9464358357376743a8452a98eff8","last_reissued_at":"2026-07-05T08:40:29.888535Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:40:29.888535Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Mixture of A Million Experts","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Xu Owen He","submitted_at":"2024-07-04T20:59:20Z","abstract_excerpt":"The feedforward (FFW) layers in standard transformer architectures incur a linear increase in computational costs and activation memory as the hidden layer width grows. Sparse mixture-of-experts (MoE) architectures have emerged as a viable approach to address this issue by decoupling model size from computational cost. The recent discovery of the fine-grained MoE scaling law shows that higher granularity leads to better performance. However, existing MoE models are limited to a small number of experts due to computational and optimization challenges. This paper introduces PEER (parameter effic"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.04153","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/2407.04153/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":"2407.04153","created_at":"2026-07-05T08:40:29.888590+00:00"},{"alias_kind":"arxiv_version","alias_value":"2407.04153v1","created_at":"2026-07-05T08:40:29.888590+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.04153","created_at":"2026-07-05T08:40:29.888590+00:00"},{"alias_kind":"pith_short_12","alias_value":"RPK6NES4K3U4","created_at":"2026-07-05T08:40:29.888590+00:00"},{"alias_kind":"pith_short_16","alias_value":"RPK6NES4K3U4EJVS","created_at":"2026-07-05T08:40:29.888590+00:00"},{"alias_kind":"pith_short_8","alias_value":"RPK6NES4","created_at":"2026-07-05T08:40:29.888590+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":14,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.12113","citing_title":"Augmenting Molecular Language Models with Local $n$-gram Memory","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2606.07414","citing_title":"Sparsely gated tiny linear experts","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01062","citing_title":"DAG-MoE: From Simple Mixture to Structural Aggregation in Mixture-of-Experts","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2606.28533","citing_title":"CMSL: Constructive Multi-Sequence Learning for Recommendation Systems","ref_index":99,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10933","citing_title":"DECO: Sparse Mixture-of-Experts with Dense-Comparable Performance on End-Side Devices","ref_index":138,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14200","citing_title":"How to Scale Mixture-of-Experts: From muP to the Maximally Scale-Stable Parameterization","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10933","citing_title":"DECO: Sparse Mixture-of-Experts with Dense-Comparable Performance on End-Side Devices","ref_index":138,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11537","citing_title":"Fast MoE Inference via Predictive Prefetching and Expert Replication","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10933","citing_title":"DECO: Sparse Mixture-of-Experts with Dense-Comparable Performance on End-Side Devices","ref_index":138,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06216","citing_title":"TIDE: Every Layer Knows the Token Beneath the Context","ref_index":84,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06665","citing_title":"UniPool: A Globally Shared Expert Pool for Mixture-of-Experts","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2604.14419","citing_title":"Equifinality in Mixture of Experts: Routing Topology Does Not Determine Language Modeling Quality","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2604.20156","citing_title":"Temporally Extended Mixture-of-Experts Models","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04952","citing_title":"Adaptive Inverted-Index Routing for Granular Mixtures-of-Experts","ref_index":20,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RPK6NES4K3U4EJVSPPTW6MWZEQ","json":"https://pith.science/pith/RPK6NES4K3U4EJVSPPTW6MWZEQ.json","graph_json":"https://pith.science/api/pith-number/RPK6NES4K3U4EJVSPPTW6MWZEQ/graph.json","events_json":"https://pith.science/api/pith-number/RPK6NES4K3U4EJVSPPTW6MWZEQ/events.json","paper":"https://pith.science/paper/RPK6NES4"},"agent_actions":{"view_html":"https://pith.science/pith/RPK6NES4K3U4EJVSPPTW6MWZEQ","download_json":"https://pith.science/pith/RPK6NES4K3U4EJVSPPTW6MWZEQ.json","view_paper":"https://pith.science/paper/RPK6NES4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2407.04153&json=true","fetch_graph":"https://pith.science/api/pith-number/RPK6NES4K3U4EJVSPPTW6MWZEQ/graph.json","fetch_events":"https://pith.science/api/pith-number/RPK6NES4K3U4EJVSPPTW6MWZEQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RPK6NES4K3U4EJVSPPTW6MWZEQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RPK6NES4K3U4EJVSPPTW6MWZEQ/action/storage_attestation","attest_author":"https://pith.science/pith/RPK6NES4K3U4EJVSPPTW6MWZEQ/action/author_attestation","sign_citation":"https://pith.science/pith/RPK6NES4K3U4EJVSPPTW6MWZEQ/action/citation_signature","submit_replication":"https://pith.science/pith/RPK6NES4K3U4EJVSPPTW6MWZEQ/action/replication_record"}},"created_at":"2026-07-05T08:40:29.888590+00:00","updated_at":"2026-07-05T08:40:29.888590+00:00"}