English
Related papers

Related papers: AdamHD: Decoupled Huber Decay Regularization for L…

200 papers

Adaptive first-order optimizers are fundamental tools in deep learning, although they may suffer from poor generalization due to the nonuniform gradient scaling. In this work, we propose AdamL, a novel variant of the Adam optimizer, that…

Machine Learning · Statistics 2023-12-27 Lu Xia , Stefano Massei

Much of the existing theory on first-order non-smooth optimization is built on a restrictive assumption that the gradients of the objective function are uniformly bounded. We introduce a much more realistic class of generalized Lipschitz…

Optimization and Control · Mathematics 2026-05-27 Dmitry Kovalev

Regularization is a critical component in deep learning. The most commonly used approach, weight decay, applies a constant penalty coefficient uniformly across all parameters. This may be overly restrictive for some parameters, while…

Machine Learning · Computer Science 2024-12-10 Jörg K. H. Franke , Michael Hefenbrock , Gregor Koehler , Frank Hutter

AdamW has been the default optimizer for transformer pretraining. For many years, our community searched for faster and more stable optimizers with only constrained positive outcomes. In this work, we propose a \textbf{one-line modification…

Machine Learning · Computer Science 2026-02-17 Kaizhao Liang , Lizhang Chen , Bo Liu , Qiang Liu

Quantization emerges as one of the most promising compression technologies for deploying efficient large models for various real time application in recent years. Considering that the storage and IO of weights take up the vast majority of…

Machine Learning · Computer Science 2024-04-22 Yi Guo , Fanliu Kong , Xiaoyang Li , Hui Li , Wei Chen , Xiaogang Tian , Jinping Cai , Yang Zhang , Shouda Liu

Adaptive gradient methods such as Adam have gained extreme popularity due to their success in training complex neural networks and less sensitivity to hyperparameter tuning compared to SGD. However, it has been recently shown that Adam can…

Machine Learning · Computer Science 2019-12-11 Pedro Savarese

We analyze cumulative parameter trajectories of transformer training under AdamW and identify a dominant low-dimensional drift direction ("backbone") that captures 60--80% of long-horizon displacement from initialization. This direction is…

Machine Learning · Computer Science 2026-03-20 Yongzhong Xu

We present an adaptive regularization scheme for optimizing composite energy functionals arising in image analysis problems. The scheme automatically trades off data fidelity and regularization depending on the current data fit during the…

Computer Vision and Pattern Recognition · Computer Science 2017-05-10 Byung-Woo Hong , Ja-Keoung Koo , Martin Burger , Stefano Soatto

To address the enormous size of Large Language Models (LLMs), model compression methods, such as quantization and pruning, are often deployed, especially on edge devices. In this work, we focus on layer-wise post-training quantization and…

Machine Learning · Computer Science 2025-12-02 Jing Liu , Toshiaki Koike-Akino , Ye Wang , Hassan Mansour , Matthew Brand

Adaptive optimization algorithms such as Adam are widely used in deep learning. The stability of such algorithms is often improved with a warmup schedule for the learning rate. Motivated by the difficulty of choosing and tuning warmup…

Machine Learning · Computer Science 2021-03-23 Jerry Ma , Denis Yarats

Weight decay is one of the standard tricks in the neural network toolbox, but the reasons for its regularization effect are poorly understood, and recent results have cast doubt on the traditional interpretation in terms of $L_2$…

Machine Learning · Computer Science 2018-10-30 Guodong Zhang , Chaoqi Wang , Bowen Xu , Roger Grosse

Adam has proven remarkable successful in training deep neural networks, but the mechanisms underlying its empirical successes and limitations remain underexplored. In this study, we demonstrate that the effectiveness of Adam stems largely…

Machine Learning · Computer Science 2025-07-10 Hanyang Peng , Shuang Qin , Yue Yu , Fangqing Jiang , Hui Wang , Wen Gao

Large-batch training has become a cornerstone in accelerating the training of deep neural networks, yet it poses challenges in optimization and generalization. Existing optimizers like AdamW present performance degradation during language…

Machine Learning · Computer Science 2025-08-29 Yang Luo , Zangwei Zheng , Ziheng Qin , Zirui Zhu , Yong Liu , Yang You

We present a class of novel optimisers for training neural networks that makes use of the Riemannian metric naturally induced when the loss landscape is embedded in higher-dimensional space. This is the same metric that underlies common…

Machine Learning · Computer Science 2025-09-05 Thomas R. Harvey

It is widely believed that stochastic gradient descent (SGD) performs significantly worse than adaptive optimizers such as Adam in pre-training Large Language Models (LLMs). Yet the underlying reason for this gap remains unclear. In this…

Machine Learning · Computer Science 2026-05-19 Athanasios Glentis , Dawei Li , Chung-Yiu Yau , Mingyi Hong

Adam-type optimizers, as a class of adaptive moment estimation methods with the exponential moving average scheme, have been successfully used in many applications of deep learning. Such methods are appealing due to the capability on…

Machine Learning · Computer Science 2020-12-17 Bingxin Zhou , Xuebin Zheng , Junbin Gao

Optimizers like Adam and AdaGrad have been very successful in training large-scale neural networks. Yet, the performance of these methods is heavily dependent on a carefully tuned learning rate schedule. We show that in many large-scale…

Machine Learning · Computer Science 2022-02-02 Ehsan Amid , Rohan Anil , Christopher Fifty , Manfred K. Warmuth

Alternating Direction Method of Multipliers (ADMM) has been used successfully in many conventional machine learning applications and is considered to be a useful alternative to Stochastic Gradient Descent (SGD) as a deep learning optimizer.…

Optimization and Control · Mathematics 2021-07-07 Junxiang Wang , Fuxun Yu , Xiang Chen , Liang Zhao

Dropout and other feature noising schemes control overfitting by artificially corrupting the training data. For generalized linear models, dropout performs a form of adaptive regularization. Using this viewpoint, we show that the dropout…

Machine Learning · Statistics 2013-11-04 Stefan Wager , Sida Wang , Percy Liang

Adaptive gradient methods are typically used for training over-parameterized models. To better understand their behaviour, we study a simplistic setting -- smooth, convex losses with models over-parameterized enough to interpolate the data.…

Machine Learning · Computer Science 2021-02-22 Sharan Vaswani , Issam Laradji , Frederik Kunstner , Si Yi Meng , Mark Schmidt , Simon Lacoste-Julien