{"id":"977870e452262b0b696a9a5ac2bd633e0b6f5da4972ebe2899041979af8ae6b8","title":"MuonGS: Pretraining LLMs with Relaxed Weight-Norm Control","argument":"arXiv cs.AI lists a paper by Athanasios Glentis, Dawei Li and Mingyi Hong proposing MuonGS, an optimizer that uses Gram--Schmidt (GS) projection to remove the radial component of each Muon update without fixing the weight norm. The authors analyze accumulated effective learning rate, which measures the sum of the nominal learning rate relative to the weight norm over training. They report that MuonGS guarantees accumulation of at least order square-root in the number of training steps under stated conditions, and establish a stationarity guarantee for smooth scale-invariant objectives. In pretraining experiments, MuonGS achieves lower validation perplexity than MuonH at long training horizons. For a dense model, the authors report an estimated token-equivalent speedup of up to about 15% at matched validation loss. The abstract does not establish whether these results extend to other models or training settings.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The supplied evidence is the paper’s abstract, not its full methods or results; it does not provide experiment details needed to assess the reported comparisons. The abstract reports results for specific pretraining experiments but does not establish whether the gains extend to other models or training settings.","speaker":{"name":"Glentis, Athanasios, Li, Dawei, Hong, Mingyi","roleAtSource":"Article author"},"source":{"title":"MuonGS: Pretraining LLMs with Relaxed Weight-Norm Control","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2610.37647","publishedAt":"2026-10-27","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors propose MuonGS, which uses Gram--Schmidt projection to remove each Muon update’s radial component without fixing the weight norm.","excerpt":"MuonGS, which uses Gram--Schmidt (GS) projection to remove the radial component of each Muon update without fixing the weight norm.","url":"https://arxiv.org/abs/2610.37647","relationship":"Source-stated"},{"text":"Under the stated conditions, the authors show MuonGS guarantees accumulated effective learning rate of at least order square-root in the number of training steps.","excerpt":"MuonGS guarantees $C_T=\\Omega(\\sqrt{T})$, with accumulation sustained even in later training.","url":"https://arxiv.org/abs/2610.37647","relationship":"Source-stated"},{"text":"For smooth scale-invariant objectives, the authors establish an $O(T^{-1/2})$ stationarity guarantee.","excerpt":"for smooth scale-invariant objectives we establish its $O(T^{-1/2})$ stationarity guarantee.","url":"https://arxiv.org/abs/2610.37647","relationship":"Source-stated"},{"text":"In the reported pretraining experiments, MuonGS achieved lower validation perplexity than MuonH at long training horizons.","excerpt":"MuonGS achieves lower validation perplexity than MuonH at long training horizons.","url":"https://arxiv.org/abs/2610.37647","relationship":"Source-stated"},{"text":"For a dense model, the authors report an estimated token-equivalent speedup of up to about 15% at matched validation loss to MuonH.","excerpt":"for the dense model it achieves an estimated token-equivalent speedup up to about $15\\%$ at matched validation loss to MuonH.","url":"https://arxiv.org/abs/2610.37647","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2610.37647","checkedAt":"2026-11-14T01:50:11.945Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"Do MuonGS’s reported validation and token-equivalent speedup results hold across other model sizes and training settings?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"MuonGS: Pretraining LLMs with Relaxed Weight-Norm Control","summary":"arXiv cs.AI lists a paper by Athanasios Glentis, Dawei Li and Mingyi Hong proposing MuonGS, an optimizer that uses Gram--Schmidt (GS) projection to remove the radial component of each Muon update without fixing the weight norm. The authors analyze accumulated effective learning rate, which measures the sum of the nominal learning rate relative to the weight norm over training. They report that MuonGS guarantees accumulation of at least order square-root in the number of training steps under stated conditions, and establish a stationarity guarantee for smooth scale-invariant objectives. In pretraining experiments, MuonGS achieves lower validation perplexity than MuonH at long training horizons. For a dense model, the authors report an estimated token-equivalent speedup of up to about 15% at matched validation loss. The abstract does not establish whether these results extend to other models or training settings.","question":"Do MuonGS’s reported validation and token-equivalent speedup results hold across other model sizes and training settings?","claims":[{"text":"The authors propose MuonGS, which uses Gram--Schmidt projection to remove each Muon update’s radial component without fixing the weight norm.","excerpt":"MuonGS, which uses Gram--Schmidt (GS) projection to remove the radial component of each Muon update without fixing the weight norm.","support":"source-stated"},{"text":"Under the stated conditions, the authors show MuonGS guarantees accumulated effective learning rate of at least order square-root in the number of training steps.","excerpt":"MuonGS guarantees $C_T=\\Omega(\\sqrt{T})$, with accumulation sustained even in later training.","support":"source-stated"},{"text":"For smooth scale-invariant objectives, the authors establish an $O(T^{-1/2})$ stationarity guarantee.","excerpt":"for smooth scale-invariant objectives we establish its $O(T^{-1/2})$ stationarity guarantee.","support":"source-stated"},{"text":"In the reported pretraining experiments, MuonGS achieved lower validation perplexity than MuonH at long training horizons.","excerpt":"MuonGS achieves lower validation perplexity than MuonH at long training horizons.","support":"source-stated"},{"text":"For a dense model, the authors report an estimated token-equivalent speedup of up to about 15% at matched validation loss to MuonH.","excerpt":"for the dense model it achieves an estimated token-equivalent speedup up to about $15\\%$ at matched validation loss to MuonH.","support":"source-stated"}],"limitations":["The supplied evidence is the paper’s abstract, not its full methods or results; it does not provide experiment details needed to assess the reported comparisons.","The abstract reports results for specific pretraining experiments but does not establish whether the gains extend to other models or training settings."],"mediaId":"ddacaec9701998e5417dda825f6e1100f3be0a974bbda7405b9ff732d9daa6d5","correctionReason":null,"id":"977870e452262b0b696a9a5ac2bd633e0b6f5da4972ebe2899041979af8ae6b8","publicationId":"2d2a6a9b-7361-429d-80e9-69648887d1bf","version":1,"sourceUrl":"https://arxiv.org/abs/2610.37647","sourceTitle":"MuonGS: Pretraining LLMs with Relaxed Weight-Norm Control","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-10-27","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-10-28T07:54:35.012Z","createdAt":"2026-11-14T01:34:14.753Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"ddacaec9701998e5417dda825f6e1100f3be0a974bbda7405b9ff732d9daa6d5","kind":"article","url":"https://arxiv.org/abs/2610.37647","sourceUrl":"https://arxiv.org/abs/2610.37647","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-10-30T04:16:59.601Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2610.37647","checkedAt":"2026-11-14T01:50:11.945Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"ddacaec9701998e5417dda825f6e1100f3be0a974bbda7405b9ff732d9daa6d5","confidence":0.87,"provider":"typesafe","model":"jev-1.13.0","inputHash":"35e645af0acf5a0620dd8876d0d739b650ebc2e5e7e09053d03002ea364f3a78","decidedAt":"2026-10-30T04:17:01.338Z"},"provenance":{"textHash":"ad07f1b580396795f9691600dbeb9fb99b569550d71bc92b6707ea07616f9248","observationId":"1a0428bfc2de251788cf0dd66de1f92e5389eac694f6e9e5d43bb50af68a576b","retrievedAt":"2026-10-30T04:16:59.336Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"f14c6718-f424-4b46-883c-c57167a8c05d"},"reviewer":{"model":"gpt-6-luna","runId":"a2a46298-fccc-45f4-884b-4ce0b9c0d8c7"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-14T01:34:14.753Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Glentis, Athanasios","url":null,"kind":"unknown"},{"name":"Li, Dawei","url":null,"kind":"unknown"},{"name":"Hong, Mingyi","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Glentis, Athanasios","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Glentis, Athanasios","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.37647","locator":"head/meta[name=author|citation_author]"},"assertionId":"54605e96a434037538e02569cfdba33bee96eb99d840ffe684a43e6bc3ca9d05:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-10-30T04:16:59.336Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"54605e96a434037538e02569cfdba33bee96eb99d840ffe684a43e6bc3ca9d05","sourceUrl":"https://arxiv.org/abs/2610.37647","contentSha256":"ebdfa0d194b18bc46214d00fcaaf79edb1f6930115f8eeb527a962a5be8509ee","textHash":"ad07f1b580396795f9691600dbeb9fb99b569550d71bc92b6707ea07616f9248","observedAt":"2026-10-30T04:16:59.336Z"}},{"actor":{"name":"Li, Dawei","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Li, Dawei","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.37647","locator":"head/meta[name=author|citation_author]"},"assertionId":"54605e96a434037538e02569cfdba33bee96eb99d840ffe684a43e6bc3ca9d05:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-10-30T04:16:59.336Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"54605e96a434037538e02569cfdba33bee96eb99d840ffe684a43e6bc3ca9d05","sourceUrl":"https://arxiv.org/abs/2610.37647","contentSha256":"ebdfa0d194b18bc46214d00fcaaf79edb1f6930115f8eeb527a962a5be8509ee","textHash":"ad07f1b580396795f9691600dbeb9fb99b569550d71bc92b6707ea07616f9248","observedAt":"2026-10-30T04:16:59.336Z"}},{"actor":{"name":"Hong, Mingyi","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Hong, Mingyi","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.37647","locator":"head/meta[name=author|citation_author]"},"assertionId":"54605e96a434037538e02569cfdba33bee96eb99d840ffe684a43e6bc3ca9d05:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-10-30T04:16:59.336Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"54605e96a434037538e02569cfdba33bee96eb99d840ffe684a43e6bc3ca9d05","sourceUrl":"https://arxiv.org/abs/2610.37647","contentSha256":"ebdfa0d194b18bc46214d00fcaaf79edb1f6930115f8eeb527a962a5be8509ee","textHash":"ad07f1b580396795f9691600dbeb9fb99b569550d71bc92b6707ea07616f9248","observedAt":"2026-10-30T04:16:59.336Z"}}],"participationCoverage":{"total":3,"returned":3,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2610.37647","snapshotId":"54605e96a434037538e02569cfdba33bee96eb99d840ffe684a43e6bc3ca9d05","contentSha256":"ebdfa0d194b18bc46214d00fcaaf79edb1f6930115f8eeb527a962a5be8509ee","textHash":"ad07f1b580396795f9691600dbeb9fb99b569550d71bc92b6707ea07616f9248","observedAt":"2026-10-30T04:16:59.336Z","basis":"publisher-page-metadata","sourceLinks":45,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/977870e452262b0b696a9a5ac2bd633e0b6f5da4972ebe2899041979af8ae6b8/source"}},"canonicalUrl":"https://escapemeridian.com/readings/muongs-pretraining-llms-with-relaxed-weight-norm-control-977870e4","formats":{"html":"/readings/muongs-pretraining-llms-with-relaxed-weight-norm-control-977870e4","markdown":"/readings/muongs-pretraining-llms-with-relaxed-weight-norm-control-977870e4.md","json":"/readings/muongs-pretraining-llms-with-relaxed-weight-norm-control-977870e4.json"},"attribution":"Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.","maintainedGuideLinks":[],"conditions":[],"readerContext":{"version":"reader-context-v1","attribution":"Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.","terms":[{"id":"term-pre-training","name":"Pre-training","explanation":"A model's initial training on a large dataset, before further adaptation for particular uses.","referenceUrl":"https://developers.google.com/machine-learning/glossary#pre-training"},{"id":"term-llm","name":"Large language model","explanation":"A language model with many learned numerical settings, used to process and generate language.","referenceUrl":"https://developers.google.com/machine-learning/glossary#large-language-model"}],"entities":[{"id":"arxiv","name":"arXiv","kind":"organization","context":"arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.","referenceUrls":["https://arxiv.org/","https://info.arxiv.org/help/api/index.html"],"path":"/entities/arxiv","json":"/entities/arxiv.json","markdown":"/entities/arxiv.md"}]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"977870e452262b0b696a9a5ac2bd633e0b6f5da4972ebe2899041979af8ae6b8","revision":"2d2a6a9b-7361-429d-80e9-69648887d1bf","revisionKind":"publication-id","title":"MuonGS: Pretraining LLMs with Relaxed Weight-Norm Control","url":"https://escapemeridian.com/readings/muongs-pretraining-llms-with-relaxed-weight-norm-control-977870e4"},"source":{"type":"source","id":"https://arxiv.org/abs/2610.37647","revision":"ad07f1b580396795f9691600dbeb9fb99b569550d71bc92b6707ea07616f9248","revisionKind":"sha256-source-text","title":"MuonGS: Pretraining LLMs with Relaxed Weight-Norm Control","url":"https://arxiv.org/abs/2610.37647"},"status":"available","items":[],"next":null,"scope":"Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter."},"revisions":[{"id":"977870e452262b0b696a9a5ac2bd633e0b6f5da4972ebe2899041979af8ae6b8","title":"MuonGS: Pretraining LLMs with Relaxed Weight-Norm Control","argument":"arXiv cs.AI lists a paper by Athanasios Glentis, Dawei Li and Mingyi Hong proposing MuonGS, an optimizer that uses Gram--Schmidt (GS) projection to remove the radial component of each Muon update without fixing the weight norm. The authors analyze accumulated effective learning rate, which measures the sum of the nominal learning rate relative to the weight norm over training. They report that MuonGS guarantees accumulation of at least order square-root in the number of training steps under stated conditions, and establish a stationarity guarantee for smooth scale-invariant objectives. In pretraining experiments, MuonGS achieves lower validation perplexity than MuonH at long training horizons. For a dense model, the authors report an estimated token-equivalent speedup of up to about 15% at matched validation loss. The abstract does not establish whether these results extend to other models or training settings.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The supplied evidence is the paper’s abstract, not its full methods or results; it does not provide experiment details needed to assess the reported comparisons. The abstract reports results for specific pretraining experiments but does not establish whether the gains extend to other models or training settings.","speaker":{"name":"Glentis, Athanasios, Li, Dawei, Hong, Mingyi","roleAtSource":"Article author"},"source":{"title":"MuonGS: Pretraining LLMs with Relaxed Weight-Norm Control","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2610.37647","publishedAt":"2026-10-27","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors propose MuonGS, which uses Gram--Schmidt projection to remove each Muon update’s radial component without fixing the weight norm.","excerpt":"MuonGS, which uses Gram--Schmidt (GS) projection to remove the radial component of each Muon update without fixing the weight norm.","url":"https://arxiv.org/abs/2610.37647","relationship":"Source-stated"},{"text":"Under the stated conditions, the authors show MuonGS guarantees accumulated effective learning rate of at least order square-root in the number of training steps.","excerpt":"MuonGS guarantees $C_T=\\Omega(\\sqrt{T})$, with accumulation sustained even in later training.","url":"https://arxiv.org/abs/2610.37647","relationship":"Source-stated"},{"text":"For smooth scale-invariant objectives, the authors establish an $O(T^{-1/2})$ stationarity guarantee.","excerpt":"for smooth scale-invariant objectives we establish its $O(T^{-1/2})$ stationarity guarantee.","url":"https://arxiv.org/abs/2610.37647","relationship":"Source-stated"},{"text":"In the reported pretraining experiments, MuonGS achieved lower validation perplexity than MuonH at long training horizons.","excerpt":"MuonGS achieves lower validation perplexity than MuonH at long training horizons.","url":"https://arxiv.org/abs/2610.37647","relationship":"Source-stated"},{"text":"For a dense model, the authors report an estimated token-equivalent speedup of up to about 15% at matched validation loss to MuonH.","excerpt":"for the dense model it achieves an estimated token-equivalent speedup up to about $15\\%$ at matched validation loss to MuonH.","url":"https://arxiv.org/abs/2610.37647","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2610.37647","checkedAt":"2026-11-14T01:50:11.945Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"Do MuonGS’s reported validation and token-equivalent speedup results hold across other model sizes and training settings?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"MuonGS: Pretraining LLMs with Relaxed Weight-Norm Control","summary":"arXiv cs.AI lists a paper by Athanasios Glentis, Dawei Li and Mingyi Hong proposing MuonGS, an optimizer that uses Gram--Schmidt (GS) projection to remove the radial component of each Muon update without fixing the weight norm. The authors analyze accumulated effective learning rate, which measures the sum of the nominal learning rate relative to the weight norm over training. They report that MuonGS guarantees accumulation of at least order square-root in the number of training steps under stated conditions, and establish a stationarity guarantee for smooth scale-invariant objectives. In pretraining experiments, MuonGS achieves lower validation perplexity than MuonH at long training horizons. For a dense model, the authors report an estimated token-equivalent speedup of up to about 15% at matched validation loss. The abstract does not establish whether these results extend to other models or training settings.","question":"Do MuonGS’s reported validation and token-equivalent speedup results hold across other model sizes and training settings?","claims":[{"text":"The authors propose MuonGS, which uses Gram--Schmidt projection to remove each Muon update’s radial component without fixing the weight norm.","excerpt":"MuonGS, which uses Gram--Schmidt (GS) projection to remove the radial component of each Muon update without fixing the weight norm.","support":"source-stated"},{"text":"Under the stated conditions, the authors show MuonGS guarantees accumulated effective learning rate of at least order square-root in the number of training steps.","excerpt":"MuonGS guarantees $C_T=\\Omega(\\sqrt{T})$, with accumulation sustained even in later training.","support":"source-stated"},{"text":"For smooth scale-invariant objectives, the authors establish an $O(T^{-1/2})$ stationarity guarantee.","excerpt":"for smooth scale-invariant objectives we establish its $O(T^{-1/2})$ stationarity guarantee.","support":"source-stated"},{"text":"In the reported pretraining experiments, MuonGS achieved lower validation perplexity than MuonH at long training horizons.","excerpt":"MuonGS achieves lower validation perplexity than MuonH at long training horizons.","support":"source-stated"},{"text":"For a dense model, the authors report an estimated token-equivalent speedup of up to about 15% at matched validation loss to MuonH.","excerpt":"for the dense model it achieves an estimated token-equivalent speedup up to about $15\\%$ at matched validation loss to MuonH.","support":"source-stated"}],"limitations":["The supplied evidence is the paper’s abstract, not its full methods or results; it does not provide experiment details needed to assess the reported comparisons.","The abstract reports results for specific pretraining experiments but does not establish whether the gains extend to other models or training settings."],"mediaId":"ddacaec9701998e5417dda825f6e1100f3be0a974bbda7405b9ff732d9daa6d5","correctionReason":null,"id":"977870e452262b0b696a9a5ac2bd633e0b6f5da4972ebe2899041979af8ae6b8","publicationId":"2d2a6a9b-7361-429d-80e9-69648887d1bf","version":1,"sourceUrl":"https://arxiv.org/abs/2610.37647","sourceTitle":"MuonGS: Pretraining LLMs with Relaxed Weight-Norm Control","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-10-27","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-10-28T07:54:35.012Z","createdAt":"2026-11-14T01:34:14.753Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"ddacaec9701998e5417dda825f6e1100f3be0a974bbda7405b9ff732d9daa6d5","kind":"article","url":"https://arxiv.org/abs/2610.37647","sourceUrl":"https://arxiv.org/abs/2610.37647","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-10-30T04:16:59.601Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2610.37647","checkedAt":"2026-11-14T01:50:11.945Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"ddacaec9701998e5417dda825f6e1100f3be0a974bbda7405b9ff732d9daa6d5","confidence":0.87,"provider":"typesafe","model":"jev-1.13.0","inputHash":"35e645af0acf5a0620dd8876d0d739b650ebc2e5e7e09053d03002ea364f3a78","decidedAt":"2026-10-30T04:17:01.338Z"},"provenance":{"textHash":"ad07f1b580396795f9691600dbeb9fb99b569550d71bc92b6707ea07616f9248","observationId":"1a0428bfc2de251788cf0dd66de1f92e5389eac694f6e9e5d43bb50af68a576b","retrievedAt":"2026-10-30T04:16:59.336Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"f14c6718-f424-4b46-883c-c57167a8c05d"},"reviewer":{"model":"gpt-6-luna","runId":"a2a46298-fccc-45f4-884b-4ce0b9c0d8c7"}},"audit":{"status":"uncertain","reviewedAt":"2026-11-14T01:34:14.753Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Glentis, Athanasios","url":null,"kind":"unknown"},{"name":"Li, Dawei","url":null,"kind":"unknown"},{"name":"Hong, Mingyi","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Glentis, Athanasios","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Glentis, Athanasios","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.37647","locator":"head/meta[name=author|citation_author]"},"assertionId":"54605e96a434037538e02569cfdba33bee96eb99d840ffe684a43e6bc3ca9d05:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-10-30T04:16:59.336Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"54605e96a434037538e02569cfdba33bee96eb99d840ffe684a43e6bc3ca9d05","sourceUrl":"https://arxiv.org/abs/2610.37647","contentSha256":"ebdfa0d194b18bc46214d00fcaaf79edb1f6930115f8eeb527a962a5be8509ee","textHash":"ad07f1b580396795f9691600dbeb9fb99b569550d71bc92b6707ea07616f9248","observedAt":"2026-10-30T04:16:59.336Z"}},{"actor":{"name":"Li, Dawei","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Li, Dawei","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.37647","locator":"head/meta[name=author|citation_author]"},"assertionId":"54605e96a434037538e02569cfdba33bee96eb99d840ffe684a43e6bc3ca9d05:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-10-30T04:16:59.336Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"54605e96a434037538e02569cfdba33bee96eb99d840ffe684a43e6bc3ca9d05","sourceUrl":"https://arxiv.org/abs/2610.37647","contentSha256":"ebdfa0d194b18bc46214d00fcaaf79edb1f6930115f8eeb527a962a5be8509ee","textHash":"ad07f1b580396795f9691600dbeb9fb99b569550d71bc92b6707ea07616f9248","observedAt":"2026-10-30T04:16:59.336Z"}},{"actor":{"name":"Hong, Mingyi","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Hong, Mingyi","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2610.37647","locator":"head/meta[name=author|citation_author]"},"assertionId":"54605e96a434037538e02569cfdba33bee96eb99d840ffe684a43e6bc3ca9d05:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-10-30T04:16:59.336Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"54605e96a434037538e02569cfdba33bee96eb99d840ffe684a43e6bc3ca9d05","sourceUrl":"https://arxiv.org/abs/2610.37647","contentSha256":"ebdfa0d194b18bc46214d00fcaaf79edb1f6930115f8eeb527a962a5be8509ee","textHash":"ad07f1b580396795f9691600dbeb9fb99b569550d71bc92b6707ea07616f9248","observedAt":"2026-10-30T04:16:59.336Z"}}],"participationCoverage":{"total":3,"returned":3,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2610.37647","snapshotId":"54605e96a434037538e02569cfdba33bee96eb99d840ffe684a43e6bc3ca9d05","contentSha256":"ebdfa0d194b18bc46214d00fcaaf79edb1f6930115f8eeb527a962a5be8509ee","textHash":"ad07f1b580396795f9691600dbeb9fb99b569550d71bc92b6707ea07616f9248","observedAt":"2026-10-30T04:16:59.336Z","basis":"publisher-page-metadata","sourceLinks":45,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/977870e452262b0b696a9a5ac2bd633e0b6f5da4972ebe2899041979af8ae6b8/source"}}}],"audits":[{"publicationId":"2d2a6a9b-7361-429d-80e9-69648887d1bf","verdict":"uncertain","findings":[{"dimension":"attribution","status":"pass","explanation":"The paper is correctly attributed to Athanasios Glentis, Dawei Li and Mingyi Hong. Findings are framed as the authors' claims. arXiv lists the primary category as cs.LG with a cs.AI cross-list, so the cs.AI listing is acceptable.","excerpt":"Authors:Athanasios Glentis, Dawei Li, Mingyi Hong"},{"dimension":"summary","status":"uncertain","explanation":"Central terms are left unexplained for a general reader: optimizer, Muon, MuonH, weight norm, radial component, validation perplexity, stationarity and scale-invariant objectives. The reading also drops the paper's motivating argument: vanilla Muon accumulates only logarithmically, which may limit exploration and help explain its slower progress, while slow norm growth allows faster accumulation. Without this, readers cannot see why MuonGS matters. Smallest repair: add one sentence on that contrast and plain-language tooltips for Muon, MuonH and perplexity. Also note that tests used 160M and 400M dense models and a MoE model.","excerpt":"We show that vanilla Muon can accumulate only logarithmically, potentially limiting exploration and helping explain its slower empirical progress"},{"dimension":"date","status":"pass","explanation":"originalPublished 2026-10-27 matches the v1 submission date, not the later feed date.","excerpt":"[Submitted on 27 Oct 2026]"},{"dimension":"support","status":"pass","explanation":"Each claim matches the abstract and is qualified as stated conditions or reported results. Wording such as 'estimated', 'up to about 15%' and 'at long training horizons' is preserved. The limitations and question do not overstate generalization or real-world benefit.","excerpt":"MuonGS achieves lower validation perplexity than MuonH at long training horizons"},{"dimension":"media","status":"pass","explanation":"The selected media is the verified direct link to the original arXiv abstract page, which is the most direct access to the source.","excerpt":null}],"createdAt":"2026-11-14T01:34:14.753Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}