Abstract
Transformers have revolutionized machine learning by making attention a central mechanism for modeling interactions within a context. Despite the central role of attention, the theoretical capabilities of Transformers for representing contextual relations remain unclear. In this work, we address this question by developing a mathematical framework based on probability and optimal transport. We view a text as a distribution of its representations and attention as a probabilistic relation between them. This perspective reveals a connection between attention normalization and optimal transport: standard softmax normalization produces conditional relations, while Sinkhorn normalization produces joint relations with prescribed marginals. Thus, both mechanisms provide structured probabilistic relations from attention scores. Under mild conditions, we establish universal approximation results for both settings. We show that Transformer architectures with Sinkhorn normalization can approximate arbitrary contextual relations represented as joint probabilities, while standard softmax Transformers can approximate arbitrary contextual relations represented as conditional probabilities. These results provide a mathematical characterization of the expressive power of Transformers for contextual relations and show how the choice of normalization determines the probabilistic structure of the relations represented by attention.
Explore similar work
May 7, 2026
Transformers have demonstrated remarkable in-context learning (ICL) capabilities. The strong ICL performance of transformers is commonly believed to arise from their ability to implicitly execute certain algorithms on the context, thereby enhancing prediction and generation. In this work, we investigate how transformers with softmax attention perform in-context learning on linear classification data. We first construct a class of multi-layer transformers that can perform in-context logistic regression, with each layer exactly performing one step of normalized gradient descent on an in-context loss. Then, we show that our constructed transformer can be obtained through (i) training a single self-attention layer supervised by one-step gradient descent, and (ii) recurrently applying the trained layer to obtain a looped model. Training convergence guarantees of the self-attention layer and out-of-distribution generalization guarantees of the looped model are provided. Our results advance the theoretical understanding of ICL mechanism by showcasing how softmax transformers can effectively act as in-context learners.
Chenyang Zhang, Yuan Cao
School of Computing & Data Science, The University of Hong Kong
Apr 12, 2026
Sequence modelling requires determining which past tokens are causally relevant from the context and their importance: a process inherent to the attention layers in transformers, yet whose underlying learned mechanisms remain poorly understood. In this work, we formalize the task of estimating token importance as an in-context learning problem by introducing a framework based on Mixture of Transition Distributions, where a latent variable determines the influence of past tokens on the next. The distribution over this latent variable is parameterized by unobserved mixture weights that transformers must learn in-context. We demonstrate that transformers can implement Mirror Descent to learn these weights from the context. Specifically, we give an explicit construction of a three-layer transformer that exactly implements one step of Mirror Descent and prove that the resulting estimator is a first-order approximation of the Bayes-optimal predictor. Corroborating our construction and its learnability via gradient descent, we empirically show that transformers trained from scratch learn solutions consistent with our theory: their predictive distributions, attention patterns, and learned transition matrix closely match the construction, while deeper models achieve performance comparable to multi-step Mirror Descent.
Francesco D'Angelo, Nicolas Flammarion
TML Lab, EPFL, Switzerland
Mar 3, 2026
Transformer networks have achieved remarkable empirical success across a wide range of applications, yet their theoretical expressive power remains insufficiently understood. In this paper, we study the expressive capabilities of Transformer architectures. We first establish an explicit approximation of maxout networks by Transformer networks while preserving comparable model complexity. As a consequence, Transformers inherit the universal approximation capability of ReLU networks under similar complexity constraints. Building on this connection, we develop a framework to analyze the approximation of continuous piecewise linear functions by Transformers and quantitatively characterize their expressivity via the number of linear regions, which grows exponentially with depth. Our analysis establishes a theoretical bridge between approximation theory for standard feedforward neural networks and Transformer architectures. It also yields structural insights into Transformers: self-attention layers implement max-type operations, while feedforward layers realize token-wise affine transformations.
Linyan Gu, Lihua Yang, Feng Zhou
School of Mathematics, Sun Yat-sen University · School of Information Science, Guangdong University of Finance and Economics