·
DataBubble
  • Home
  • Models
  • News
  • Compare
  • Boards
  • Pricing
  • About
  • Newsletter
  • Methodology
  • Contact
Latest
Google’s latest AI weather model gives you no excuse to forget your umbrella1h◆NeoMME: an efficient Multimodal-native and Multilingual Encoder3h◆Nvidia confirms it will buy Hugging Face for $12.9 billion4h◆Nvidia is buying Hugging Face for almost $13 billion4h◆Meta-ethics and AI: exploring the novel meta-ethical questions in the era of AI12h◆SSAKG 2.0: An Open-Source Package for Structural Associative Sequence Memory and Context-Based Retrieval12h◆Epistemic Sybil Resistance: Multiplying AI Agents Without Multiplying Evidence12h◆DocHop: Benchmarking Out-of-domain Multi-hop Reasoning in Information-Dense Documents12h◆MASkills: Continual Skills Optimization for Multi-Agent LLM Systems12h◆Train at Moving Edge: Online-Verified Prompt Selection for Efficient RL Training of Large Reasoning Model12h◆SpecMine: A Large-Scale Corpus of Spec-Driven Development Artifacts12h◆Elite political incivility is rising across democracies12h◆Learning Query-Specific Rubrics from Human Preferences for DeepResearch Report Generation12h◆Sim2Signal: Sim-to-Real Benchmarks for Traffic Signal Control12h◆Recursive Value Learning for Long-Horizon Offline Goal-Conditioned RL12h◆Similarity-Aware Personalized Federated Learning in Heterogeneous Environments12h◆Oracle, will I ever learn? A study of prediction convergence and complementarity across link prediction models12h◆UE5M3 FP4 Block Scaling for Stable Language Model Pretraining12h◆Network-Aware Forecasting on Wireless Access Points12h◆Monotonic anomaly detection12h◆Google’s latest AI weather model gives you no excuse to forget your umbrella1h◆NeoMME: an efficient Multimodal-native and Multilingual Encoder3h◆Nvidia confirms it will buy Hugging Face for $12.9 billion4h◆Nvidia is buying Hugging Face for almost $13 billion4h◆Meta-ethics and AI: exploring the novel meta-ethical questions in the era of AI12h◆SSAKG 2.0: An Open-Source Package for Structural Associative Sequence Memory and Context-Based Retrieval12h◆Epistemic Sybil Resistance: Multiplying AI Agents Without Multiplying Evidence12h◆DocHop: Benchmarking Out-of-domain Multi-hop Reasoning in Information-Dense Documents12h◆MASkills: Continual Skills Optimization for Multi-Agent LLM Systems12h◆Train at Moving Edge: Online-Verified Prompt Selection for Efficient RL Training of Large Reasoning Model12h◆SpecMine: A Large-Scale Corpus of Spec-Driven Development Artifacts12h◆Elite political incivility is rising across democracies12h◆Learning Query-Specific Rubrics from Human Preferences for DeepResearch Report Generation12h◆Sim2Signal: Sim-to-Real Benchmarks for Traffic Signal Control12h◆Recursive Value Learning for Long-Horizon Offline Goal-Conditioned RL12h◆Similarity-Aware Personalized Federated Learning in Heterogeneous Environments12h◆Oracle, will I ever learn? A study of prediction convergence and complementarity across link prediction models12h◆UE5M3 FP4 Block Scaling for Stable Language Model Pretraining12h◆Network-Aware Forecasting on Wireless Access Points12h◆Monotonic anomaly detection12h◆
News/AMUSE: Anytime Muon with Stable Gradient Evaluation
arxiv
PublishedJuly 15, 2026 at 4:00 AM

AMUSE: Anytime Muon with Stable Gradient Evaluation

Source
arxiv.orgfull article ↗
Read on arxiv→
Publisher summary· verbatim

arXiv:2605.22432v2 Announce Type: replace Abstract: Modern deep learning commonly relies on AdamW with prescribed learning rate schedules, but recent works challenge both components: Schedule-Free optimization removes explicit schedules via iterate averaging, and Muon improves the update geometry by

Stay posted· Newsletter

A 5-min weekly brief — top movers, price watch, story of the week.

// no spam · unsubscribe one-click · free forever

Discussion
Source
↗
arxiv
Read original ↗All from arxiv →

No replies yet. Be first.

Source
↗
arxiv
Read original ↗All from arxiv →

Related coverage

More from ARXIV
arxivMeta-ethics and AI: exploring the novel meta-ethical questions in the era of AI12harxivSSAKG 2.0: An Open-Source Package for Structural Associative Sequence Memory and Context-Based Retrieval12harxivEpistemic Sybil Resistance: Multiplying AI Agents Without Multiplying Evidence12harxivDocHop: Benchmarking Out-of-domain Multi-hop Reasoning in Information-Dense Documents12h
The Bubble Brief
WEEKLY

Read AI insights every Tuesday — top movers, new releases, story of the week.

// no spam · unsubscribe one-click · free forever

Originally published on arxiv ↗
HomeModelsNews