{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:DJPV7COVSWDEII3L3SWWXT3GF2","short_pith_number":"pith:DJPV7COV","schema_version":"1.0","canonical_sha256":"1a5f5f89d5958644236bdcad6bcf662e8e10dbe915b42e6fbe5e9e42117f0216","source":{"kind":"arxiv","id":"2607.05895","version":1},"attestation_state":"computed","paper":{"title":"MatrixFSDP: communication-free matrix optimizers under ZeRO-3 parameter sharding","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.DC","authors_text":"Hao Zhang, Ming Gao, Yanwu Xu","submitted_at":"2026-07-07T06:45:39Z","abstract_excerpt":"Matrix optimizers such as Muon are attractive for large-scale training because they can improve convergence and token efficiency over coordinate-wise optimizers. Muon does this by orthogonalizing momentum-smoothed matrix updates with Newton-Schulz, producing spectrum-balanced updates that require the complete 2D matrix as input. This exposes a systems mismatch: FSDP/ZeRO-3 saves memory by making the optimizer see shards, not whole matrices. Existing systems therefore either reconstruct matrices at every optimizer step, paying weight-sized communication after backward, or make the update local "},"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":"2607.05895","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.DC","submitted_at":"2026-07-07T06:45:39Z","cross_cats_sorted":[],"title_canon_sha256":"32db0eb0d6939f3382a484289c024953e3e002cc61514e95dc18205473481d8c","abstract_canon_sha256":"ec4e4050e3702a8dc0a610a4a43b53fdbab81a07c30f2a8f241f4f878effae13"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-08T01:18:49.557464Z","signature_b64":"TE4r5uZZKYT6/oRbWLpRweOsSUucyyvPUdds7zkQIRlRP5qaYU6gj7/qDRjObEuNVlPUCssl7InxW4JUk431BA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1a5f5f89d5958644236bdcad6bcf662e8e10dbe915b42e6fbe5e9e42117f0216","last_reissued_at":"2026-07-08T01:18:49.556770Z","signature_status":"signed_v1","first_computed_at":"2026-07-08T01:18:49.556770Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MatrixFSDP: communication-free matrix optimizers under ZeRO-3 parameter sharding","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.DC","authors_text":"Hao Zhang, Ming Gao, Yanwu Xu","submitted_at":"2026-07-07T06:45:39Z","abstract_excerpt":"Matrix optimizers such as Muon are attractive for large-scale training because they can improve convergence and token efficiency over coordinate-wise optimizers. Muon does this by orthogonalizing momentum-smoothed matrix updates with Newton-Schulz, producing spectrum-balanced updates that require the complete 2D matrix as input. This exposes a systems mismatch: FSDP/ZeRO-3 saves memory by making the optimizer see shards, not whole matrices. Existing systems therefore either reconstruct matrices at every optimizer step, paying weight-sized communication after backward, or make the update local "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.05895","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/2607.05895/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":"2607.05895","created_at":"2026-07-08T01:18:49.556842+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.05895v1","created_at":"2026-07-08T01:18:49.556842+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.05895","created_at":"2026-07-08T01:18:49.556842+00:00"},{"alias_kind":"pith_short_12","alias_value":"DJPV7COVSWDE","created_at":"2026-07-08T01:18:49.556842+00:00"},{"alias_kind":"pith_short_16","alias_value":"DJPV7COVSWDEII3L","created_at":"2026-07-08T01:18:49.556842+00:00"},{"alias_kind":"pith_short_8","alias_value":"DJPV7COV","created_at":"2026-07-08T01:18:49.556842+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/DJPV7COVSWDEII3L3SWWXT3GF2","json":"https://pith.science/pith/DJPV7COVSWDEII3L3SWWXT3GF2.json","graph_json":"https://pith.science/api/pith-number/DJPV7COVSWDEII3L3SWWXT3GF2/graph.json","events_json":"https://pith.science/api/pith-number/DJPV7COVSWDEII3L3SWWXT3GF2/events.json","paper":"https://pith.science/paper/DJPV7COV"},"agent_actions":{"view_html":"https://pith.science/pith/DJPV7COVSWDEII3L3SWWXT3GF2","download_json":"https://pith.science/pith/DJPV7COVSWDEII3L3SWWXT3GF2.json","view_paper":"https://pith.science/paper/DJPV7COV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.05895&json=true","fetch_graph":"https://pith.science/api/pith-number/DJPV7COVSWDEII3L3SWWXT3GF2/graph.json","fetch_events":"https://pith.science/api/pith-number/DJPV7COVSWDEII3L3SWWXT3GF2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DJPV7COVSWDEII3L3SWWXT3GF2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DJPV7COVSWDEII3L3SWWXT3GF2/action/storage_attestation","attest_author":"https://pith.science/pith/DJPV7COVSWDEII3L3SWWXT3GF2/action/author_attestation","sign_citation":"https://pith.science/pith/DJPV7COVSWDEII3L3SWWXT3GF2/action/citation_signature","submit_replication":"https://pith.science/pith/DJPV7COVSWDEII3L3SWWXT3GF2/action/replication_record"}},"created_at":"2026-07-08T01:18:49.556842+00:00","updated_at":"2026-07-08T01:18:49.556842+00:00"}