{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:ZERXMQ533RGLM54Q33GCDC322N","short_pith_number":"pith:ZERXMQ53","schema_version":"1.0","canonical_sha256":"c9237643bbdc4cb67790decc218b7ad3516b2eefa228ea647cccf4b038107476","source":{"kind":"arxiv","id":"2409.06669","version":1},"attestation_state":"computed","paper":{"title":"DA-MoE: Towards Dynamic Expert Allocation for Mixture-of-Experts Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Hongpeng Jin, Maryam Akhavan Aghdam, Yanzhao Wu","submitted_at":"2024-09-10T17:36:15Z","abstract_excerpt":"Transformer-based Mixture-of-Experts (MoE) models have been driving several recent technological advancements in Natural Language Processing (NLP). These MoE models adopt a router mechanism to determine which experts to activate for routing input tokens. However, existing router mechanisms allocate a fixed number of experts to each token, which neglects the varying importance of different input tokens. In this study, we propose a novel dynamic router mechanism that Dynamically Allocates a variable number of experts for Mixture-of-Experts (DA-MoE) models based on an effective token importance m"},"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":"2409.06669","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-09-10T17:36:15Z","cross_cats_sorted":[],"title_canon_sha256":"ed790cda36e95fb6dcb07cbc3ef084ac2ef9ea69bfd4c1f54e019affd9705c5a","abstract_canon_sha256":"d5fdfb9730f96f6090565917ed37e1a2567b0729ecaa7b4e46d2ad1847dbc0c0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:05:27.960833Z","signature_b64":"iggUOlXVMBQ9vU94KxlaJIFerrJI6Q7Mxr2MqjFDPA1NwOHtzWXPvnx6WRnmKgbLtTCErQ9wUk4tVMMSKXBxAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c9237643bbdc4cb67790decc218b7ad3516b2eefa228ea647cccf4b038107476","last_reissued_at":"2026-07-05T09:05:27.960367Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:05:27.960367Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"DA-MoE: Towards Dynamic Expert Allocation for Mixture-of-Experts Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Hongpeng Jin, Maryam Akhavan Aghdam, Yanzhao Wu","submitted_at":"2024-09-10T17:36:15Z","abstract_excerpt":"Transformer-based Mixture-of-Experts (MoE) models have been driving several recent technological advancements in Natural Language Processing (NLP). These MoE models adopt a router mechanism to determine which experts to activate for routing input tokens. However, existing router mechanisms allocate a fixed number of experts to each token, which neglects the varying importance of different input tokens. In this study, we propose a novel dynamic router mechanism that Dynamically Allocates a variable number of experts for Mixture-of-Experts (DA-MoE) models based on an effective token importance m"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2409.06669","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/2409.06669/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":"2409.06669","created_at":"2026-07-05T09:05:27.960437+00:00"},{"alias_kind":"arxiv_version","alias_value":"2409.06669v1","created_at":"2026-07-05T09:05:27.960437+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2409.06669","created_at":"2026-07-05T09:05:27.960437+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZERXMQ533RGL","created_at":"2026-07-05T09:05:27.960437+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZERXMQ533RGLM54Q","created_at":"2026-07-05T09:05:27.960437+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZERXMQ53","created_at":"2026-07-05T09:05:27.960437+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.01509","citing_title":"ProbMoE: Differentiable Probabilistic Routing for Mixture-of-Experts","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2606.29982","citing_title":"Beyond Uniform Experts: Cost-Aware Expert Execution for Efficient Multi-Device MoE Inference","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2605.30876","citing_title":"dMoE: dLLMs with Learnable Block Experts","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14438","citing_title":"BEAM: Binary Expert Activation Masking for Dynamic Routing in MoE","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2604.26340","citing_title":"Adaptive and Fine-grained Module-wise Expert Pruning for Efficient LoRA-MoE Fine-Tuning","ref_index":17,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ZERXMQ533RGLM54Q33GCDC322N","json":"https://pith.science/pith/ZERXMQ533RGLM54Q33GCDC322N.json","graph_json":"https://pith.science/api/pith-number/ZERXMQ533RGLM54Q33GCDC322N/graph.json","events_json":"https://pith.science/api/pith-number/ZERXMQ533RGLM54Q33GCDC322N/events.json","paper":"https://pith.science/paper/ZERXMQ53"},"agent_actions":{"view_html":"https://pith.science/pith/ZERXMQ533RGLM54Q33GCDC322N","download_json":"https://pith.science/pith/ZERXMQ533RGLM54Q33GCDC322N.json","view_paper":"https://pith.science/paper/ZERXMQ53","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2409.06669&json=true","fetch_graph":"https://pith.science/api/pith-number/ZERXMQ533RGLM54Q33GCDC322N/graph.json","fetch_events":"https://pith.science/api/pith-number/ZERXMQ533RGLM54Q33GCDC322N/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZERXMQ533RGLM54Q33GCDC322N/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZERXMQ533RGLM54Q33GCDC322N/action/storage_attestation","attest_author":"https://pith.science/pith/ZERXMQ533RGLM54Q33GCDC322N/action/author_attestation","sign_citation":"https://pith.science/pith/ZERXMQ533RGLM54Q33GCDC322N/action/citation_signature","submit_replication":"https://pith.science/pith/ZERXMQ533RGLM54Q33GCDC322N/action/replication_record"}},"created_at":"2026-07-05T09:05:27.960437+00:00","updated_at":"2026-07-05T09:05:27.960437+00:00"}