[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-turbo-muon-cuts-ai-training-time-with-smarter-math":10,"sections":41},{"siteName":4,"siteTagline":5,"publisherName":4,"contactEmail":6},"The Revision","Tech news, decoded.","editor@therevision.news",{"gaMeasurementId":8,"adsenseClientId":9},"G-ZW2MV82GYR","ca-pub-8533917693782264",{"article":11},{"id":12,"slug":13,"title":14,"dek":15,"body_md":16,"tags_json":17,"published_at":18,"created_at":19,"updated_at":20,"status":21,"review_note":22,"review_notes":23,"image_url":22,"persona_id":22,"persona_name":22,"section":30,"tags":31,"sources":36,"feedback":40,"feedback_at":22,"cost_usd":40,"total_tokens":40},4053,"turbo-muon-cuts-ai-training-time-with-smarter-math","Turbo-Muon Cuts AI Training Time With Smarter Math","A new pre-conditioning trick for the Muon optimizer trims training time by roughly 3% with no hyperparameter tuning required.","A refinement to the Muon optimizer trims training overhead without touching hyperparameters or model architecture.\n\nMuon and similar orthogonality-based optimizers have gained traction in large-scale training benchmarks, but they carry a notable cost: a gradient orthogonalization step that relies on Newton-Schulz iterations, which typically require dozens of matrix multiplications to converge. Researchers behind Turbo-Muon introduce a pre-conditioning procedure that improves how those iterations are initialized, reducing the initial polar error enough to eliminate one of the five Newton-Schulz iterations normally run in practice. The result is a measurable drop in overhead with negligible cost added by the pre-conditioning itself. The code is live on GitHub and available in optax and Hugging Face kernels, as the paper describes.\n\nThe 3% training-time reduction may look modest on a slide deck, but at the scale where frontier labs run thousands of GPU-hours per run, that margin compounds fast. More practically, it drops in as a replacement with no retuning — the kind of tradeoff that actually gets adopted in production pipelines rather than shelved after a benchmark. The authors also offer a theoretical account of why the update resists feature collapse, which gives practitioners a reason to trust it beyond the numbers.\n\nOrthogonality-based optimizers are still a relatively recent entrant competing against entrenched adaptive methods like AdamW; a frictionless speed gain is exactly the kind of evidence that moves adoption — though whether these benchmark improvements hold across every architecture and data regime remains an open question.","[\"machine learning\",\"optimizers\",\"training efficiency\",\"ai research\"]","2026-07-07T04:00:00.000Z","2026-07-07T15:56:59.700Z","2026-07-07T15:57:02.585Z","published",null,[24],{"id":25,"reviewer":26,"round":27,"reason":28,"status":29},"editor-r1","editor",1,"The body states Newton-Schulz 'typically runs five times per update and involves enough matrix multiplications to become a real overhead cost,' but the source says it 'typically requiring dozens of matrix multiplications to converge' — the article understates the cost and misrepresents the source; additionally, the code is described as live 'in optax and Hugging Face kernels' without attributing this to the paper or GitHub, and the claim that 'dismissing it as a benchmark artifact would be prema","resolved","ai",[32,33,34,35],"machine learning","optimizers","training efficiency","ai research",[37],{"name":38,"url":39},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2512.04632",0,{"sections":42},[43,47,52,57,62,67,72,77,82,86,91,95,100,105],{"name":44,"slug":30,"count":45,"latest_published_at":46},"AI",2590,"2026-07-16T04:00:00.000Z",{"name":48,"slug":49,"count":50,"latest_published_at":51},"Security","security",294,"2026-07-15T19:59:48.000Z",{"name":53,"slug":54,"count":55,"latest_published_at":56},"Deals","deals",179,"2026-06-29T20:02:07.000Z",{"name":58,"slug":59,"count":60,"latest_published_at":61},"Policy","policy",158,"2026-07-16T00:02:48.000Z",{"name":63,"slug":64,"count":65,"latest_published_at":66},"Hardware","hardware",122,"2026-07-14T19:46:26.000Z",{"name":68,"slug":69,"count":70,"latest_published_at":71},"Consumer Tech","consumer-tech",93,"2026-07-13T13:20:48.000Z",{"name":73,"slug":74,"count":75,"latest_published_at":76},"Software","software",70,"2026-07-13T19:52:25.000Z",{"name":78,"slug":79,"count":80,"latest_published_at":81},"Science","science",66,"2026-07-10T10:29:37.000Z",{"name":83,"slug":84,"count":85,"latest_published_at":18},"Dev Tools","dev-tools",59,{"name":87,"slug":88,"count":89,"latest_published_at":90},"Gaming","gaming",41,"2026-07-09T04:00:00.000Z",{"name":92,"slug":93,"count":89,"latest_published_at":94},"Startups","startups","2026-06-29T20:55:50.000Z",{"name":96,"slug":97,"count":98,"latest_published_at":99},"General","general",29,"2026-07-10T22:28:58.000Z",{"name":101,"slug":102,"count":103,"latest_published_at":104},"Reviews","reviews",20,"2026-06-24T12:00:01.000Z",{"name":106,"slug":107,"count":108,"latest_published_at":109},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]