English
Related papers

Related papers: Emergence of Minimal Circuits for Indirect Object …

200 papers

Mechanistic interpretability seeks to reverse engineer a trained neural network by identifying the minimal subset of internal components. We perform a mechanistic interpretability analysis of the Particle Transformer architecture, trained…

High Energy Physics - Phenomenology · Physics 2026-05-12 Saurabh Rai , Sanmay Ganguly

Chain-of-Thought (CoT) is an efficient prompting method that enables the reasoning ability of large language models by augmenting the query using multiple examples with multiple intermediate steps. Despite the empirical success, the…

Machine Learning · Computer Science 2025-05-27 Hongkang Li , Songtao Lu , Pin-Yu Chen , Xiaodong Cui , Meng Wang

Large language models (LLMs) are powerful models that can learn concepts at the inference stage via in-context learning (ICL). While theoretical studies, e.g., \cite{zhang2023trained}, attempt to explain the mechanism of ICL, they assume…

Machine Learning · Computer Science 2024-06-19 Yue Xing , Xiaofeng Lin , Chenheng Xu , Namjoon Suh , Qifan Song , Guang Cheng

Neural networks are growing more capable on their own, but we do not understand their neural mechanisms. Understanding these mechanisms' decision-making processes, or mechanistic interpretability, enables (1) accountability and control in…

Computation and Language · Computer Science 2026-03-02 Mason Kadem , Rong Zheng

The fields of explainable AI and mechanistic interpretability aim to uncover the internal structure of neural networks, with circuit discovery as a central tool for understanding model computations. Existing approaches, however, rely on…

Machine Learning · Computer Science 2026-03-05 Elena Golimblevskaia , Aakriti Jain , Bruno Puri , Ammar Ibrahim , Wojciech Samek , Sebastian Lapuschkin

Multi-head attention layers, as used in the Transformer neural sequence model, are a powerful alternative to RNNs for moving information across and between sequences. While training these layers is generally fast and simple, due to…

Neural and Evolutionary Computing · Computer Science 2019-11-07 Noam Shazeer

An attention matrix of a transformer self-attention sublayer can provably be decomposed into two components and only one of them (effective attention) contributes to the model output. This leads us to ask whether visualizing effective…

Computation and Language · Computer Science 2021-05-20 Kaiser Sun , Ana Marasović

"Induction heads" are attention heads that implement a simple algorithm to complete token sequences like [A][B] ... [A] -> [B]. In this work, we present preliminary and indirect evidence for a hypothesis that induction heads might…

We propose an adaptation to the training of Vision Transformers (ViTs) that allows for an explicit modeling of objects during the attention computation. This is achieved by adding a new branch to selected attention layers that computes an…

Computer Vision and Pattern Recognition · Computer Science 2025-04-14 Vivek Trivedy , Amani Almalki , Longin Jan Latecki

The attention mechanism lies at the core of the transformer architecture, providing an interpretable model-internal signal that has motivated a growing interest in attention-based model explanations. Although attention weights do not…

Machine Learning · Computer Science 2025-08-13 Marte Eggen , Jacob Lysnæs-Larsen , Inga Strümke

Attention-based transformers have been remarkably successful at modeling generative processes across various domains and modalities. In this paper, we study the behavior of transformers on data drawn from \kth Markov processes, where the…

Machine Learning · Computer Science 2024-07-26 Nived Rajaraman , Marco Bondaschi , Kannan Ramchandran , Michael Gastpar , Ashok Vardhan Makkuva

Language recognition tasks are fundamental in natural language processing (NLP) and have been widely used to benchmark the performance of large language models (LLMs). These tasks also play a crucial role in explaining the working…

Machine Learning · Computer Science 2025-05-30 Ruiquan Huang , Yingbin Liang , Jing Yang

With the recent successful adaptation of transformers to the vision domain, particularly when trained in a self-supervised fashion, it has been shown that vision transformers can learn impressive object-reasoning-like behaviour and features…

Computer Vision and Pattern Recognition · Computer Science 2022-10-26 Oscar Vikström , Alexander Ilin

The transformer architecture, which processes sequences of input tokens to produce outputs for query tokens, has revolutionized numerous areas of machine learning. A defining feature of transformers is their ability to perform previously…

Machine Learning · Computer Science 2025-10-02 Hongbo Li , Lingjie Duan , Yingbin Liang

The intriguing in-context learning (ICL) abilities of deep Transformer models have lately garnered significant attention. By studying in-context linear regression on unimodal Gaussian data, recent empirical and theoretical works have argued…

Machine Learning · Computer Science 2024-10-30 Khashayar Gatmiry , Nikunj Saunshi , Sashank J. Reddi , Stefanie Jegelka , Sanjiv Kumar

Transformer has become the dominant architecture for sequence modeling, yet a detailed understanding of how its structural parameters influence expressive power remains limited. In this work, we study the approximation properties of…

Machine Learning · Computer Science 2026-04-01 Penghao Yu , Haotian Jiang , Zeyu Bao , Ruoxi Yu , Qianxiao Li

Transformer training is notoriously difficult, requiring a careful design of optimizers and use of various heuristics. We make progress towards understanding the subtleties of training Transformers by carefully studying a simple yet…

Machine Learning · Computer Science 2024-03-14 Kwangjun Ahn , Xiang Cheng , Minhak Song , Chulhee Yun , Ali Jadbabaie , Suvrit Sra

The multi-head self-attention mechanism of the transformer model has been thoroughly investigated recently. In one vein of study, researchers are interested in understanding why and how transformers work. In another vein, researchers…

Computation and Language · Computer Science 2022-10-28 Raymond Li , Wen Xiao , Linzi Xing , Lanjun Wang , Gabriel Murray , Giuseppe Carenini

Vision Transformers have achieved state-of-the-art performance in many visual tasks. Due to the quadratic computational and memory complexities of self-attention, recent works either apply attention only to low-resolution inputs or restrict…

Computer Vision and Pattern Recognition · Computer Science 2022-07-05 Cheng Li , Yangxin Liu

Transformer-based LLMs demonstrate strong performance on graph reasoning tasks, yet their internal mechanisms remain underexplored. To uncover these reasoning process mechanisms in a fundamental and unified view, we set the basic…

Machine Learning · Computer Science 2025-09-25 Xinnan Dai , Chung-Hsiang Lo , Kai Guo , Shenglai Zeng , Dongsheng Luo , Jiliang Tang