Open Problem: Is AdamW Effective Under Heavy-Tailed Noise?
Dingzhi Yu, Hongyi Tao, Yuanyu Wan, Luo Luo, Lijun Zhang
cs.LG, cs.AI, math.OC, stat.ML
Submitted: 2026-06-22
Code: https://github.com/karpathy/nanoGPT
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- The Geometry of Sign Gradient Descent
- Better Convergence Guarantees for Sign-Based Momentum Methods
- Kimi K2: Open Agentic Intelligence
- Muon is Scalable for LLM Training
- Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less
- Can Adaptive Gradient Methods Converge under Heavy-Tailed Noise? A Case Study of AdaGrad
- Simple Convergence Proof of Adam From a Sign-like Descent Perspective
- Benchmarking Optimizers for Large Language Model Pretraining
- Practical Efficiency of Muon for Pretraining
- When and Why SignSGD Outperforms SGD: A Theoretical Study Based on $\ell_1$-norm Lower Bounds
- Fantastic Pretraining Optimizers and Where to Find Them
- Mirror Descent Under Generalized Smoothness
- StoSignSGD: Unbiased Structural Stochasticity Fixes SignSGD for Training Large Language Models
- Sign-Based Optimizers Are Effective Under Heavy-Tailed Noise
- GLM-5: from Vibe Coding to Agentic Engineering
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks