English
Related papers

Related papers: The Depth Delusion: Why Transformers Should Be Wid…

200 papers

Performance of deep learning models is strongly governed by architectural capacity, with width and depth as primary controls. However, in physical-science applications, models are often compared at a single fixed size or by separating…

Machine Learning · Computer Science 2026-05-07 Alexander I. Khrabry , Edward A. Startsev , Andrew T. Powis , Igor D. Kaganovich

Stacking non-linear layers allows deep neural networks to model complicated functions, and including residual connections in Transformer layers is beneficial for convergence and performance. However, residual connections may make the model…

Computation and Language · Computer Science 2024-04-05 Hongfei Xu , Yang Song , Qiuhui Liu , Josef van Genabith , Deyi Xiong

The population loss of trained deep neural networks often follows precise power-law scaling relations with either the size of the training dataset or the number of parameters in the network. We propose a theory that explains the origins of…

Machine Learning · Computer Science 2024-06-28 Yasaman Bahri , Ethan Dyer , Jared Kaplan , Jaehoon Lee , Utkarsh Sharma

Linear layers hold most of a transformer's parameters. We replace each linear layer with one that stores $K$ out of $mn$ two-dimensional DCT coefficients per weight matrix and reconstructs the full matrix through an inverse DCT at every…

Performance · Computer Science 2026-04-10 Mohamed Amine Bergach

Scaling laws have played a major role in the modern AI revolution, providing practitioners predictive power over how the model performance will improve with increasing data, compute, and number of model parameters. This has spurred an…

Machine Learning · Computer Science 2026-01-16 Maissam Barkeshli , Alberto Alfarano , Andrey Gromov

As neural networks continue to grow in size but datasets might not, it is vital to understand how much performance improvement can be expected: is it more important to scale network size or data volume? Thus, neural network scaling laws,…

Machine Learning · Computer Science 2024-09-10 Akhilan Boopathy , Ila Fiete

Deep neural networks are widely used prediction algorithms whose performance often improves as the number of weights increases, leading to over-parametrization. We consider a two-layered neural network whose first layer is frozen while the…

Machine Learning · Computer Science 2023-04-10 Roman Worschech , Bernd Rosenow

When optimizing convolutional neural networks (CNN) for a specific image-based task, specialists commonly overshoot the number of convolutional layers in their designs. By implication, these CNNs are unnecessarily resource intensive to…

Machine Learning · Computer Science 2022-06-23 Mats L. Richter , Julius Schöning , Anna Wiedenroth , Ulf Krumnack

While scaling Transformer-based large language models (LLMs) has demonstrated promising performance across various tasks, it also introduces redundant architectures, posing efficiency challenges for real-world deployment. Despite some…

Machine Learning · Computer Science 2024-10-18 Shwai He , Guoheng Sun , Zheyu Shen , Ang Li

Recent works have shown that gradient descent can find a global minimum for over-parameterized neural networks where the widths of all the hidden layers scale polynomially with $N$ ($N$ being the number of training samples). In this paper,…

Machine Learning · Computer Science 2020-12-21 Quynh Nguyen , Marco Mondelli

We propose a novel scaling law for general-purpose decoder-only language models (LMs) trained on multilingual data, tackling the problem of balancing languages during multilingual pretraining. A primary challenge in studying multilingual…

Computation and Language · Computer Science 2024-12-05 Yifei He , Alon Benhaim , Barun Patra , Praneetha Vaddamanu , Sanchit Ahuja , Parul Chopra , Vishrav Chaudhary , Han Zhao , Xia Song

We present a deep learning approach to estimation of the bead parameters in welding tasks. Our model is based on a four-hidden-layer neural network architecture. More specifically, the first three hidden layers of this architecture utilize…

Machine Learning · Computer Science 2015-07-09 Soheil Keshmiri , Xin Zheng , Chee Meng Chew , Chee Khiang Pang

In recent years, the state-of-the-art in deep learning has been dominated by very large models that have been pre-trained on vast amounts of data. The paradigm is very simple: investing more computational resources (optimally) leads to…

Machine Learning · Computer Science 2024-05-24 Sotiris Anagnostidis , Gregor Bachmann , Imanol Schlag , Thomas Hofmann

Transformer architecture has become the de-facto model for many machine learning tasks from natural language processing and computer vision. As such, improving its computational efficiency becomes paramount. One of the major computational…

Computation and Language · Computer Science 2022-05-17 Yue Guan , Zhengyi Li , Jingwen Leng , Zhouhan Lin , Minyi Guo

Large language models (LLMs) demonstrate strong performance across natural language processing tasks, yet undergo significant performance degradation when modified for deployment through quantization, pruning, or decoding strategy…

Computation and Language · Computer Science 2025-04-01 Ziyang Ma , Zuchao Li , Lefei Zhang , Gui-Song Xia , Bo Du , Liangpei Zhang , Dacheng Tao

Large language models are often viewed as high-dimensional nonlinear systems and treated as black boxes. Here, we show that transformer depth dynamics admit accurate low-order linear surrogates within context. Across tasks including…

Machine Learning · Computer Science 2026-03-16 Buddhika Nettasinghe , Geethu Joseph

LLM architecture research generally aims to maximize model quality subject to fixed compute/latency budgets. However, many applications of interest such as edge and on-device deployment are further constrained by the model's memory…

Machine Learning · Computer Science 2026-04-28 Abbas Zeitoun , Lucas Torroba-Hennigen , Yoon Kim

Recently, DeepNorm scales Transformers into extremely deep (i.e., 1000 layers) and reveals the promising potential of deep scaling. To stabilize the training of deep models, DeepNorm (Wang et al., 2022) attempts to constrain the model…

Machine Learning · Computer Science 2023-05-05 Yijin Liu , Xianfeng Zeng , Fandong Meng , Jie Zhou

We provide several new depth-based separation results for feed-forward neural networks, proving that various types of simple and natural functions can be better approximated using deeper networks than shallower ones, even if the shallower…

Machine Learning · Computer Science 2020-05-14 Itay Safran , Ohad Shamir

The Transformer model has achieved state-of-the-art performance in many sequence modeling tasks. However, how to leverage model capacity with large or variable depths is still an open challenge. We present a probabilistic framework to…

Computation and Language · Computer Science 2020-10-19 Xian Li , Asa Cooper Stickland , Yuqing Tang , Xiang Kong