Interpreting Transformer Architectures as Implicit Multinomial Regression

• URL: http://arxiv.org/abs/2509.04653v1
• Date: Thu, 04 Sep 2025 20:40:37 GMT
• Title: Interpreting Transformer Architectures as Implicit Multinomial Regression
• Authors: Jonas A. Actor, Anthony Gruber, Eric C. Cyr,
• Abstract summary: We show that in a fixed multinomial regression setting, optimizing over latent features yields optimal solutions that align with the dynamics induced by attention blocks.<n>In other words, the evolution of representations through a transformer can be interpreted as a trajectory that recovers the optimal features for classification.
• Score: 3.2371089062298317
• License: http://creativecommons.org/licenses/by-nc-nd/4.0/
• Abstract: Mechanistic interpretability aims to understand how internal components of modern machine learning models, such as weights, activations, and layers, give rise to the model's overall behavior. One particularly opaque mechanism is attention: despite its central role in transformer models, its mathematical underpinnings and relationship to concepts like feature polysemanticity, superposition, and model performance remain poorly understood. This paper establishes a novel connection between attention mechanisms and multinomial regression. Specifically, we show that in a fixed multinomial regression setting, optimizing over latent features yields optimal solutions that align with the dynamics induced by attention blocks. In other words, the evolution of representations through a transformer can be interpreted as a trajectory that recovers the optimal features for classification.

Related papers

• A Free Probabilistic Framework for Analyzing the Transformer-based Language Models [19.78896931593813]
  We present a formal operator-theoretic framework for analyzing Transformer-based language models using free probability theory.<n>This work offers a principled, though theoretical, perspective on structural dynamics in large language models.
  arXiv  Detail & Related papers  (2025-06-19T19:13:02Z)
• Self-Boost via Optimal Retraining: An Analysis via Approximate Message Passing [58.52119063742121]
  Retraining a model using its own predictions together with the original, potentially noisy labels is a well-known strategy for improving the model performance.<n>This paper addresses the question of how to optimally combine the model's predictions and the provided labels.<n>Our main contribution is the derivation of the Bayes optimal aggregator function to combine the current model's predictions and the given labels.
  arXiv  Detail & Related papers  (2025-05-21T07:16:44Z)
• Exact Learning Dynamics of In-Context Learning in Linear Transformers and Its Application to Non-Linear Transformers [1.7034813545878589]
  Transformer models exhibit remarkable in-context learning (ICL)<n>Our work offers an exact dynamical model for ICL and theoretically grounded tools for analyzing complex transformer training.
  arXiv  Detail & Related papers  (2025-04-17T13:05:33Z)
• Neural ODE Transformers: Analyzing Internal Dynamics and Adaptive Fine-tuning [30.781578037476347]
  We introduce a novel approach to modeling transformer architectures using highly flexible non-autonomous neural ordinary differential equations (ODEs)<n>Our proposed model parameterizes all weights of attention and feed-forward blocks through neural networks, expressing these weights as functions of a continuous layer index.<n>Our neural ODE transformer demonstrates performance comparable to or better than vanilla transformers across various configurations and datasets.
  arXiv  Detail & Related papers  (2025-03-03T09:12:14Z)
• Variational empirical Bayes variable selection in high-dimensional logistic regression [2.4032899110671955]
  We develop a novel and computationally efficient variational approximation thereof.<n>One such novelty is that we develop this approximation directly for the marginal distribution on the model space, rather than on the regression coefficients themselves.<n>We demonstrate the method's strong performance in simulations, and prove that our variational approximation inherits the strong selection consistency property satisfied by the posterior distribution that it is approximating.
  arXiv  Detail & Related papers  (2025-02-14T19:57:13Z)
• Explainable and Class-Revealing Signal Feature Extraction via Scattering Transform and Constrained Zeroth-Order Optimization [0.5893124686141783]
  We propose a new method to extract discriminant and explainable features from a machine learning model.<n>We adopt zeroth-order optimization algorithms to search an input pattern that maximizes the class probability of a class of interest.<n>We demonstrate the effectiveness of our proposed method using a couple of synthetic time-series classification problems.
  arXiv  Detail & Related papers  (2025-02-08T23:35:37Z)
• Interpreting Affine Recurrence Learning in GPT-style Transformers [54.01174470722201]
  In-context learning allows GPT-style transformers to generalize during inference without modifying their weights. This paper focuses specifically on their ability to learn and predict affine recurrences as an ICL task. We analyze the model's internal operations using both empirical and theoretical approaches.
  arXiv  Detail & Related papers  (2024-10-22T21:30:01Z)
• Dynamical Mean-Field Theory of Self-Attention Neural Networks [0.0]
  Transformer-based models have demonstrated exceptional performance across diverse domains. Little is known about how they operate or what are their expected dynamics. We use methods for the study of asymmetric Hopfield networks in nonequilibrium regimes.
  arXiv  Detail & Related papers  (2024-06-11T13:29:34Z)
• The Buffer Mechanism for Multi-Step Information Reasoning in Language Models [52.77133661679439]
  Investigating internal reasoning mechanisms of large language models can help us design better model architectures and training strategies. In this study, we constructed a symbolic dataset to investigate the mechanisms by which Transformer models employ vertical thinking strategy. We proposed a random matrix-based algorithm to enhance the model's reasoning ability, resulting in a 75% reduction in the training time required for the GPT-2 model.
  arXiv  Detail & Related papers  (2024-05-24T07:41:26Z)
• Scaling and renormalization in high-dimensional regression [72.59731158970894]
  We present a unifying perspective on recent results on ridge regression.<n>We use the basic tools of random matrix theory and free probability, aimed at readers with backgrounds in physics and deep learning.<n>Our results extend and provide a unifying perspective on earlier models of scaling laws.
  arXiv  Detail & Related papers  (2024-05-01T15:59:00Z)
• EulerFormer: Sequential User Behavior Modeling with Complex Vector Attention [88.45459681677369]
  We propose a novel transformer variant with complex vector attention, named EulerFormer. It provides a unified theoretical framework to formulate both semantic difference and positional difference. It is more robust to semantic variations and possesses moresuperior theoretical properties in principle.
  arXiv  Detail & Related papers  (2024-03-26T14:18:43Z)
• Uncovering mesa-optimization algorithms in Transformers [61.06055590704677]
  Some autoregressive models can learn as an input sequence is processed, without undergoing any parameter changes, and without being explicitly trained to do so. We show that standard next-token prediction error minimization gives rise to a subsidiary learning algorithm that adjusts the model as new inputs are revealed. Our findings explain in-context learning as a product of autoregressive loss minimization and inform the design of new optimization-based Transformer layers.
  arXiv  Detail & Related papers  (2023-09-11T22:42:50Z)
• Multi-Response Heteroscedastic Gaussian Process Models and Their Inference [1.52292571922932]
  We propose a novel framework for the modeling of heteroscedastic covariance functions. We employ variational inference to approximate the posterior and facilitate posterior predictive modeling. We show that our proposed framework offers a robust and versatile tool for a wide array of applications.
  arXiv  Detail & Related papers  (2023-08-29T15:06:47Z)
• Convexifying Transformers: Improving optimization and understanding of transformer networks [56.69983975369641]
  We study the training problem of attention/transformer networks and introduce a novel convex analytic approach. We first introduce a convex alternative to the self-attention mechanism and reformulate the regularized training problem of transformer networks. As a byproduct of our convex analysis, we reveal an implicit regularization mechanism, which promotes sparsity across tokens.
  arXiv  Detail & Related papers  (2022-11-20T18:17:47Z)
• Nonparametric Trace Regression in High Dimensions via Sign Series Representation [13.37650464374017]
  We develop a framework for nonparametric trace regression models via structured sign series representations of high dimensional functions. In the context of matrix completion, our framework leads to a substantially richer model based on what we coin as the "sign rank" of a matrix.
  arXiv  Detail & Related papers  (2021-05-04T22:20:00Z)
• Transformers with Competitive Ensembles of Independent Mechanisms [97.93090139318294]
  We propose a new Transformer layer which divides the hidden representation and parameters into multiple mechanisms, which only exchange information through attention. We study TIM on a large-scale BERT model, on the Image Transformer, and on speech enhancement and find evidence for semantically meaningful specialization as well as improved performance.
  arXiv  Detail & Related papers  (2021-02-27T21:48:46Z)
• Attention that does not Explain Away [54.42960937271612]
  Models based on the Transformer architecture have achieved better accuracy than the ones based on competing architectures for a large set of tasks. A unique feature of the Transformer is its universal application of a self-attention mechanism, which allows for free information flow at arbitrary distances. We propose a doubly-normalized attention scheme that is simple to implement and provides theoretical guarantees for avoiding the "explaining away" effect.
  arXiv  Detail & Related papers  (2020-09-29T21:05:39Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.