{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:2Z6G4CWZOAXI5WVDRKLUEGGRYR","short_pith_number":"pith:2Z6G4CWZ","schema_version":"1.0","canonical_sha256":"d67c6e0ad9702e8edaa38a974218d1c448634851fede47a7402ae42b328e8fb9","source":{"kind":"arxiv","id":"2512.04632","version":2},"attestation_state":"computed","paper":{"title":"Turbo-Muon: Almost-Orthogonal Pre-Conditioning for Fast Muon Updates","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Franck Mamalet, IRIT-MISFIT), Mathieu Serrurier (IRIT-MISFIT), Thibaut Boissin (IRIT-MISFIT), Thomas Massena (DTIPG - SNCF","submitted_at":"2025-12-04T10:06:22Z","abstract_excerpt":"Orthogonality-based optimizers, such as Muon, have recently shown strong performance across large-scale training and community-driven efficiency challenges. However, these methods rely on a costly gradient orthogonalization step. Even efficient iterative approximations such as Newton-Schulz remain expensive, typically requiring dozens of matrix multiplications to converge. We introduce a pre-conditioning procedure that improves the initialization of the Newton--Schulz iterations while incurring negligible overhead. Furthermore, our pre-conditioning reduces the initial polar error and enables t"},"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":"2512.04632","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-12-04T10:06:22Z","cross_cats_sorted":[],"title_canon_sha256":"a5e63098d3dce9713cf0a916b863c002214156e613e6a7b271915cec2ebd25b8","abstract_canon_sha256":"b9a46d02ebf3d5bfcb4e56bb65f19eaad1108a65b9689f3192e5c31e871c1919"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-07T02:19:43.382566Z","signature_b64":"I8ha7rNHsHL6bUekKthd5HyvV3F1bqckoyFCUKQnb/mhtf4d4BLdIY1fAxrlnmpyhl4TPDXztErSe0CPZ3RkCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d67c6e0ad9702e8edaa38a974218d1c448634851fede47a7402ae42b328e8fb9","last_reissued_at":"2026-07-07T02:19:43.381697Z","signature_status":"signed_v1","first_computed_at":"2026-07-07T02:19:43.381697Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Turbo-Muon: Almost-Orthogonal Pre-Conditioning for Fast Muon Updates","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Franck Mamalet, IRIT-MISFIT), Mathieu Serrurier (IRIT-MISFIT), Thibaut Boissin (IRIT-MISFIT), Thomas Massena (DTIPG - SNCF","submitted_at":"2025-12-04T10:06:22Z","abstract_excerpt":"Orthogonality-based optimizers, such as Muon, have recently shown strong performance across large-scale training and community-driven efficiency challenges. However, these methods rely on a costly gradient orthogonalization step. Even efficient iterative approximations such as Newton-Schulz remain expensive, typically requiring dozens of matrix multiplications to converge. We introduce a pre-conditioning procedure that improves the initialization of the Newton--Schulz iterations while incurring negligible overhead. Furthermore, our pre-conditioning reduces the initial polar error and enables t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2512.04632","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/2512.04632/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":"2512.04632","created_at":"2026-07-07T02:19:43.381810+00:00"},{"alias_kind":"arxiv_version","alias_value":"2512.04632v2","created_at":"2026-07-07T02:19:43.381810+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2512.04632","created_at":"2026-07-07T02:19:43.381810+00:00"},{"alias_kind":"pith_short_12","alias_value":"2Z6G4CWZOAXI","created_at":"2026-07-07T02:19:43.381810+00:00"},{"alias_kind":"pith_short_16","alias_value":"2Z6G4CWZOAXI5WVD","created_at":"2026-07-07T02:19:43.381810+00:00"},{"alias_kind":"pith_short_8","alias_value":"2Z6G4CWZ","created_at":"2026-07-07T02:19:43.381810+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":4,"sample":[{"citing_arxiv_id":"2606.04662","citing_title":"Why Muon Outperforms Adam: A Curvature Perspective","ref_index":127,"is_internal_anchor":true},{"citing_arxiv_id":"2605.18106","citing_title":"Symmetry-Compatible Principle for Optimizer Design: Embeddings, LM Heads, SwiGLU MLPs, and MoE Routers","ref_index":17,"is_internal_anchor":true},{"citing_arxiv_id":"2605.18106","citing_title":"Symmetry-Compatible Principle for Optimizer Design: Embeddings, LM Heads, SwiGLU MLPs, and MoE Routers","ref_index":17,"is_internal_anchor":true},{"citing_arxiv_id":"2605.11396","citing_title":"MuonQ: Enhancing Low-Bit Muon Quantization via Directional Fidelity Optimization","ref_index":3,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2Z6G4CWZOAXI5WVDRKLUEGGRYR","json":"https://pith.science/pith/2Z6G4CWZOAXI5WVDRKLUEGGRYR.json","graph_json":"https://pith.science/api/pith-number/2Z6G4CWZOAXI5WVDRKLUEGGRYR/graph.json","events_json":"https://pith.science/api/pith-number/2Z6G4CWZOAXI5WVDRKLUEGGRYR/events.json","paper":"https://pith.science/paper/2Z6G4CWZ"},"agent_actions":{"view_html":"https://pith.science/pith/2Z6G4CWZOAXI5WVDRKLUEGGRYR","download_json":"https://pith.science/pith/2Z6G4CWZOAXI5WVDRKLUEGGRYR.json","view_paper":"https://pith.science/paper/2Z6G4CWZ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2512.04632&json=true","fetch_graph":"https://pith.science/api/pith-number/2Z6G4CWZOAXI5WVDRKLUEGGRYR/graph.json","fetch_events":"https://pith.science/api/pith-number/2Z6G4CWZOAXI5WVDRKLUEGGRYR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2Z6G4CWZOAXI5WVDRKLUEGGRYR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2Z6G4CWZOAXI5WVDRKLUEGGRYR/action/storage_attestation","attest_author":"https://pith.science/pith/2Z6G4CWZOAXI5WVDRKLUEGGRYR/action/author_attestation","sign_citation":"https://pith.science/pith/2Z6G4CWZOAXI5WVDRKLUEGGRYR/action/citation_signature","submit_replication":"https://pith.science/pith/2Z6G4CWZOAXI5WVDRKLUEGGRYR/action/replication_record"}},"created_at":"2026-07-07T02:19:43.381810+00:00","updated_at":"2026-07-07T02:19:43.381810+00:00"}