{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:YBI4VZK4VF2MJ3DMDAISNJH4C4","short_pith_number":"pith:YBI4VZK4","schema_version":"1.0","canonical_sha256":"c051cae55ca974c4ec6c181126a4fc1710103b15acc2ba70754d0de8e2d7e55d","source":{"kind":"arxiv","id":"2505.13416","version":1},"attestation_state":"computed","paper":{"title":"Gluon: Making Muon & Scion Great Again! (Bridging Theory and Practice of LMO-based Optimizers for LLMs)","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["math.OC","stat.ML"],"primary_cat":"cs.LG","authors_text":"Artem Riabinin, Egor Shulgin, Kaja Gruntkowska, Peter Richt\\'arik","submitted_at":"2025-05-19T17:50:45Z","abstract_excerpt":"Recent developments in deep learning optimization have brought about radically new algorithms based on the Linear Minimization Oracle (LMO) framework, such as $\\sf Muon$ and $\\sf Scion$. After over a decade of $\\sf Adam$'s dominance, these LMO-based methods are emerging as viable replacements, offering several practical advantages such as improved memory efficiency, better hyperparameter transferability, and most importantly, superior empirical performance on large-scale tasks, including LLM training. However, a significant gap remains between their practical use and our current theoretical un"},"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":"2505.13416","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-05-19T17:50:45Z","cross_cats_sorted":["math.OC","stat.ML"],"title_canon_sha256":"ebda74d26a41b945e76556a0046d3aadf42c198d01a5236ce40ab85acefcb4b7","abstract_canon_sha256":"d1d75d05345497d17e3f1c63bd5308928ae6d733b43fc4a250e7851abd2b2bb2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:05:29.709487Z","signature_b64":"dBhN2E/r3GD65yGg23aXnlYeEoaYESyooNp8GKGjz4e+lIwJWCm74k2cBB0WqDJlSclFK3O5KjkBfaPHbz1DCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c051cae55ca974c4ec6c181126a4fc1710103b15acc2ba70754d0de8e2d7e55d","last_reissued_at":"2026-07-05T11:05:29.709022Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:05:29.709022Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Gluon: Making Muon & Scion Great Again! (Bridging Theory and Practice of LMO-based Optimizers for LLMs)","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["math.OC","stat.ML"],"primary_cat":"cs.LG","authors_text":"Artem Riabinin, Egor Shulgin, Kaja Gruntkowska, Peter Richt\\'arik","submitted_at":"2025-05-19T17:50:45Z","abstract_excerpt":"Recent developments in deep learning optimization have brought about radically new algorithms based on the Linear Minimization Oracle (LMO) framework, such as $\\sf Muon$ and $\\sf Scion$. After over a decade of $\\sf Adam$'s dominance, these LMO-based methods are emerging as viable replacements, offering several practical advantages such as improved memory efficiency, better hyperparameter transferability, and most importantly, superior empirical performance on large-scale tasks, including LLM training. However, a significant gap remains between their practical use and our current theoretical un"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.13416","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/2505.13416/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":"2505.13416","created_at":"2026-07-05T11:05:29.709078+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.13416v1","created_at":"2026-07-05T11:05:29.709078+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.13416","created_at":"2026-07-05T11:05:29.709078+00:00"},{"alias_kind":"pith_short_12","alias_value":"YBI4VZK4VF2M","created_at":"2026-07-05T11:05:29.709078+00:00"},{"alias_kind":"pith_short_16","alias_value":"YBI4VZK4VF2MJ3DM","created_at":"2026-07-05T11:05:29.709078+00:00"},{"alias_kind":"pith_short_8","alias_value":"YBI4VZK4","created_at":"2026-07-05T11:05:29.709078+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":22,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.21581","citing_title":"Convergence Analysis of Muon-type Methods with Inexact LMO in the Degenerate Case","ref_index":64,"is_internal_anchor":false},{"citing_arxiv_id":"2606.21354","citing_title":"Restart and Adaptive Acceleration in Stochastic Gradient Methods","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2606.14970","citing_title":"Zero-order Parameter-free Optimization for LMO-based Methods: Novel Approach for Efficient Fine-tuning","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2606.08783","citing_title":"OptMuon: Closed-Loop Orthogonalized Momentum Methods for Stochastic Optimization with Zero-Noise Optimality","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2605.18106","citing_title":"Symmetry-Compatible Principle for Optimizer Design: Embeddings, LM Heads, SwiGLU MLPs, and MoE Routers","ref_index":134,"is_internal_anchor":false},{"citing_arxiv_id":"2605.24770","citing_title":"Muon in Vision Transformers: Optimizer-Recipe Interactions and Gradient Spectra","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2606.08783","citing_title":"OptMuon: Closed-Loop Orthogonalized Momentum Methods for Stochastic Optimization with Zero-Noise Optimality","ref_index":191,"is_internal_anchor":false},{"citing_arxiv_id":"2605.31371","citing_title":"Softsign: Smooth Sign in Your Optimizer For Better Parameter Heterogeneity Handling","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2510.10777","citing_title":"Preconditioned Norms: A Unified Framework for Steepest Descent, Quasi-Newton and Adaptive Methods","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2605.20866","citing_title":"LOSCAR-SGD: Local SGD with Communication-Computation Overlap and Delay-Corrected Sparse Model Averaging","ref_index":300,"is_internal_anchor":false},{"citing_arxiv_id":"2605.18174","citing_title":"Ringmaster LMO: Asynchronous Linear Minimization Oracle Momentum Method","ref_index":299,"is_internal_anchor":false},{"citing_arxiv_id":"2605.18106","citing_title":"Symmetry-Compatible Principle for Optimizer Design: Embeddings, LM Heads, SwiGLU MLPs, and MoE Routers","ref_index":132,"is_internal_anchor":false},{"citing_arxiv_id":"2605.19619","citing_title":"MiMuon: Mixed Muon Optimizer with Improved Generalization for Large Models","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2509.11983","citing_title":"Low-rank Orthogonalization for Large-scale Matrix Optimization with Applications to Foundation Model Training","ref_index":43,"is_internal_anchor":false},{"citing_arxiv_id":"2605.13434","citing_title":"Rescaled Asynchronous SGD: Optimal Distributed Optimization under Data and System Heterogeneity","ref_index":59,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11850","citing_title":"Constrained Stochastic Spectral Preconditioning Converges for Nonconvex Objectives","ref_index":56,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08850","citing_title":"Local LMO: Constrained Gradient Optimization via a Local Linear Minimization Oracle","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09552","citing_title":"Phases of Muon: When Muon Eclipses SignSGD","ref_index":63,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09238","citing_title":"Intrinsic Muon: Spectral Optimization on Riemannian Matrix Manifolds","ref_index":51,"is_internal_anchor":false},{"citing_arxiv_id":"2604.23980","citing_title":"SUDA-Muon: Structural Design Principles and Boundaries for Fully Decentralized Muon","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2604.10689","citing_title":"Communication-Efficient Gluon in Federated Learning","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2605.06884","citing_title":"Muon with Nesterov Momentum: Heavy-Tailed Noise and (Randomized) Inexact Polar Decomposition","ref_index":43,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YBI4VZK4VF2MJ3DMDAISNJH4C4","json":"https://pith.science/pith/YBI4VZK4VF2MJ3DMDAISNJH4C4.json","graph_json":"https://pith.science/api/pith-number/YBI4VZK4VF2MJ3DMDAISNJH4C4/graph.json","events_json":"https://pith.science/api/pith-number/YBI4VZK4VF2MJ3DMDAISNJH4C4/events.json","paper":"https://pith.science/paper/YBI4VZK4"},"agent_actions":{"view_html":"https://pith.science/pith/YBI4VZK4VF2MJ3DMDAISNJH4C4","download_json":"https://pith.science/pith/YBI4VZK4VF2MJ3DMDAISNJH4C4.json","view_paper":"https://pith.science/paper/YBI4VZK4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.13416&json=true","fetch_graph":"https://pith.science/api/pith-number/YBI4VZK4VF2MJ3DMDAISNJH4C4/graph.json","fetch_events":"https://pith.science/api/pith-number/YBI4VZK4VF2MJ3DMDAISNJH4C4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YBI4VZK4VF2MJ3DMDAISNJH4C4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YBI4VZK4VF2MJ3DMDAISNJH4C4/action/storage_attestation","attest_author":"https://pith.science/pith/YBI4VZK4VF2MJ3DMDAISNJH4C4/action/author_attestation","sign_citation":"https://pith.science/pith/YBI4VZK4VF2MJ3DMDAISNJH4C4/action/citation_signature","submit_replication":"https://pith.science/pith/YBI4VZK4VF2MJ3DMDAISNJH4C4/action/replication_record"}},"created_at":"2026-07-05T11:05:29.709078+00:00","updated_at":"2026-07-05T11:05:29.709078+00:00"}