{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:QK2Y7X4SECJRRJLAJSWD6RXGNB","short_pith_number":"pith:QK2Y7X4S","schema_version":"1.0","canonical_sha256":"82b58fdf92209318a5604cac3f46e6684df9f200f0177ae8f627e6ebcd64a4e7","source":{"kind":"arxiv","id":"2502.05172","version":2},"attestation_state":"computed","paper":{"title":"Joint MoE Scaling Laws: Mixture of Experts Can Be Memory Efficient","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Jakub Krajewski, Jan Ludziejewski, Jan Ma{\\l}a\\'snicki, Kamil Adamczewski, Maciej Pi\\'oro, Maciej Stefaniak, Marek Cygan, Micha{\\l} Krutul, Piotr Mi{\\l}o\\'s, Piotr Sankowski, Sebastian Jaszczur","submitted_at":"2025-02-07T18:55:38Z","abstract_excerpt":"Mixture of Experts (MoE) architectures have significantly increased computational efficiency in both research and real-world applications of large-scale machine learning models. However, their scalability and efficiency under memory constraints remain relatively underexplored. In this work, we present joint scaling laws for dense and MoE models, incorporating key factors such as the number of active parameters, dataset size, and the number of experts. Our findings provide a principled framework for selecting the optimal MoE configuration under fixed memory and compute budgets. Surprisingly, we"},"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":"2502.05172","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-02-07T18:55:38Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"d665987ef3912e682d183469e976ac98d239151feec13f2d8bfcab4ea82cad02","abstract_canon_sha256":"f7d4731bb8ab1e12aacadb6b01738c4b074163bfd1ed48ec3bc6fc241b3fab58"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:16:59.913875Z","signature_b64":"zcDELPFp9mbKTYclt8bufEV0MbFhPTsAsJvNVO4PD5TH9YlRO1Fl3pSOwuioYo1N59Nz7K2a8EUK8U0wOxh3Bw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"82b58fdf92209318a5604cac3f46e6684df9f200f0177ae8f627e6ebcd64a4e7","last_reissued_at":"2026-07-05T10:16:59.913393Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:16:59.913393Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Joint MoE Scaling Laws: Mixture of Experts Can Be Memory Efficient","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Jakub Krajewski, Jan Ludziejewski, Jan Ma{\\l}a\\'snicki, Kamil Adamczewski, Maciej Pi\\'oro, Maciej Stefaniak, Marek Cygan, Micha{\\l} Krutul, Piotr Mi{\\l}o\\'s, Piotr Sankowski, Sebastian Jaszczur","submitted_at":"2025-02-07T18:55:38Z","abstract_excerpt":"Mixture of Experts (MoE) architectures have significantly increased computational efficiency in both research and real-world applications of large-scale machine learning models. However, their scalability and efficiency under memory constraints remain relatively underexplored. In this work, we present joint scaling laws for dense and MoE models, incorporating key factors such as the number of active parameters, dataset size, and the number of experts. Our findings provide a principled framework for selecting the optimal MoE configuration under fixed memory and compute budgets. Surprisingly, we"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.05172","kind":"arxiv","version":2},"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/2502.05172/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":"2502.05172","created_at":"2026-07-05T10:16:59.913452+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.05172v2","created_at":"2026-07-05T10:16:59.913452+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.05172","created_at":"2026-07-05T10:16:59.913452+00:00"},{"alias_kind":"pith_short_12","alias_value":"QK2Y7X4SECJR","created_at":"2026-07-05T10:16:59.913452+00:00"},{"alias_kind":"pith_short_16","alias_value":"QK2Y7X4SECJRRJLA","created_at":"2026-07-05T10:16:59.913452+00:00"},{"alias_kind":"pith_short_8","alias_value":"QK2Y7X4S","created_at":"2026-07-05T10:16:59.913452+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.15079","citing_title":"Ling and Ring 2.6 Technical Report: Efficient and Instant Agentic Intelligence at Trillion-Parameter Scale","ref_index":83,"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":25,"is_internal_anchor":false},{"citing_arxiv_id":"2605.27358","citing_title":"MobileMoE: Scaling On-Device Mixture of Experts","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2506.12119","citing_title":"Mixture-of-Experts Can Surpass Dense LLMs Under Strictly Equal Resource","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2605.13769","citing_title":"Dense vs Sparse Pretraining at Tiny Scale: Active-Parameter vs Total-Parameter Matching","ref_index":10,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/QK2Y7X4SECJRRJLAJSWD6RXGNB","json":"https://pith.science/pith/QK2Y7X4SECJRRJLAJSWD6RXGNB.json","graph_json":"https://pith.science/api/pith-number/QK2Y7X4SECJRRJLAJSWD6RXGNB/graph.json","events_json":"https://pith.science/api/pith-number/QK2Y7X4SECJRRJLAJSWD6RXGNB/events.json","paper":"https://pith.science/paper/QK2Y7X4S"},"agent_actions":{"view_html":"https://pith.science/pith/QK2Y7X4SECJRRJLAJSWD6RXGNB","download_json":"https://pith.science/pith/QK2Y7X4SECJRRJLAJSWD6RXGNB.json","view_paper":"https://pith.science/paper/QK2Y7X4S","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.05172&json=true","fetch_graph":"https://pith.science/api/pith-number/QK2Y7X4SECJRRJLAJSWD6RXGNB/graph.json","fetch_events":"https://pith.science/api/pith-number/QK2Y7X4SECJRRJLAJSWD6RXGNB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QK2Y7X4SECJRRJLAJSWD6RXGNB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QK2Y7X4SECJRRJLAJSWD6RXGNB/action/storage_attestation","attest_author":"https://pith.science/pith/QK2Y7X4SECJRRJLAJSWD6RXGNB/action/author_attestation","sign_citation":"https://pith.science/pith/QK2Y7X4SECJRRJLAJSWD6RXGNB/action/citation_signature","submit_replication":"https://pith.science/pith/QK2Y7X4SECJRRJLAJSWD6RXGNB/action/replication_record"}},"created_at":"2026-07-05T10:16:59.913452+00:00","updated_at":"2026-07-05T10:16:59.913452+00:00"}