English
Related papers

Related papers: Variance Is Not Importance: Structural Analysis of…

200 papers

Transformers underpin modern large language models (LLMs) and are commonly assumed to be behaviorally unstructured at random initialization, with all meaningful preferences emerging only through large-scale training. We challenge this…

Machine Learning · Statistics 2026-02-06 Siquan Li , Yao Tong , Haonan Wang , Tianyang Hu

A simple design recipe for deep Transformers is to compose identical building blocks. But standard transformer blocks are far from simple, interweaving attention and MLP sub-blocks with skip connections & normalisation layers in precise…

Machine Learning · Computer Science 2024-06-03 Bobby He , Thomas Hofmann

Why do language models trained on contradictory data prefer correct answers? In controlled experiments with small transformers (3.5M--86M parameters), we show that this preference tracks the compressibility structure of errors rather than…

Computation and Language · Computer Science 2026-04-07 Konstantin Krestnikov

In recent years, there has been a growing emphasis on compressing large pre-trained transformer models for resource-constrained devices. However, traditional pruning methods often leave the embedding layer untouched, leading to model…

Machine Learning · Computer Science 2024-06-13 Oshin Dutta , Ritvik Gupta , Sumeet Agarwal

The computational overhead of Vision Transformers in practice stems fundamentally from their deep architectures, yet existing acceleration strategies have primarily targeted algorithmic-level optimizations such as token pruning and…

Computer Vision and Pattern Recognition · Computer Science 2025-11-26 Chengwei Zhou , Vipin Chaudhary , Gourav Datta

Owing to success in the data-rich domain of natural images, Transformers have recently become popular in medical image segmentation. However, the pairing of Transformers with convolutional blocks in varying architectural permutations leaves…

Computer Vision and Pattern Recognition · Computer Science 2023-04-11 Saikat Roy , Gregor Koehler , Michael Baumgartner , Constantin Ulrich , Jens Petersen , Fabian Isensee , Klaus Maier-Hein

Compressibility transformations are used to relate hypersonic zero-pressure-gradient (ZPG) turbulent boundary layers (TBLs) to incompressible reference states, but their assessment has largely focused on the collapse of transformed mean…

Fluid Dynamics · Physics 2025-12-11 Engin Danis

We propose a non-perturbative definition for refined topological strings. This can be used to compute the partition function of superconformal theories in 5 dimensions on squashed S^5 and the superconformal index of a large number of 6…

High Energy Physics - Theory · Physics 2013-04-24 Guglielmo Lockhart , Cumrun Vafa

While confinement is understood to increase the power and thrust coefficients of cross-flow turbines, how the optimal rotor geometry changes with the blockage ratio -- defined as the ratio between the array projected area and the channel…

Fluid Dynamics · Physics 2025-06-17 Aidan Hunt , Gregory Talpey , Gemma Calandra , Brian Polagye

Deeper Vision Transformers often perform worse than shallower ones, which challenges common scaling assumptions. Through a systematic empirical analysis of ViT-S, ViT-B, and ViT-L on ImageNet, we identify a consistent three-phase…

Machine Learning · Computer Science 2025-11-27 Anantha Padmanaban Krishna Kumar

Transformers have demonstrated remarkable success across various applications. However, the success of transformers have not been understood in theory. In this work, we give a case study of how transformers can be trained to learn a classic…

Machine Learning · Statistics 2025-04-14 Chenyang Zhang , Xuran Meng , Yuan Cao

The attention operator is arguably the key distinguishing factor of transformer architectures, which have demonstrated state-of-the-art performance on a variety of tasks. However, transformer attention operators often impose a significant…

Machine Learning · Computer Science 2024-12-24 Ziyang Wu , Tianjiao Ding , Yifu Lu , Druv Pai , Jingyuan Zhang , Weida Wang , Yaodong Yu , Yi Ma , Benjamin D. Haeffele

We study how a one-layer attention-only transformer develops relevant structures while learning to sort lists of numbers. At the end of training, the model organizes its attention heads in two main modes that we refer to as…

Machine Learning · Computer Science 2025-02-03 Einar Urdshals , Jasmina Urdshals

In this paper we introduce and study the algebraic generalization of non commutative convolutional neural networks. We leverage the theory of algebraic signal processing to model convolutional non commutative architectures, and we derive…

Machine Learning · Computer Science 2023-07-07 Alejandro Parada-Mayorga , Landon Butler , Alejandro Ribeiro

Mechanistic interpretability improves the safety, reliability, and robustness of large AI models. This study examined individual attention heads in vision transformers (ViTs) fine tuned on distorted 2D spectrogram images containing non…

Machine Learning · Computer Science 2025-03-25 Nooshin Bahador

Counting properties (e.g. determining whether certain tokens occur more than other tokens in a given input text) have played a significant role in the study of expressiveness of transformers. In this paper, we provide a formal framework for…

Computation and Language · Computer Science 2026-03-03 Marco Sälzer , Chris Köcher , Alexander Kozachinskiy , Georg Zetzsche , Anthony Widjaja Lin

Rate-distortion optimization through neural networks has accomplished competitive results in compression efficiency and image quality. This learning-based approach seeks to minimize the compromise between compression rate and reconstructed…

Computer Vision and Pattern Recognition · Computer Science 2024-06-11 Raül Pérez-Gonzalo , Andreas Espersen , Antonio Agudo

Statistical inference for stochastic block models typically relies on the spectrum of the normalized adjacency matrix $\A^*$. In practice, the true probability matrix $\mathbf{B}$ is unknown and must be replaced by a plug-in estimator…

Methodology · Statistics 2026-04-09 Jianwei Hu , Ding Chen , Ji Zhu

In this paper, we present a novel transformer-based architecture for end-to-end image compression. Our architecture incorporates blocks that effectively capture local dependencies between tokens, eliminating the need for positional encoding…

Image and Video Processing · Electrical Eng. & Systems 2024-09-09 Bouzid Arezki , Fangchen Feng , Anissa Mokraoui

Continuous-control reinforcement learning (RL) often exhibits large closed-loop variance, high-frequency control jitter, and sensitivity to disturbance injection. Existing explanations usually emphasize disturbance sources such as action…

Systems and Control · Electrical Eng. & Systems 2026-04-21 Wu Yue