ViT-Explainer: An Interactive Walkthrough of the Vision Transformer Pipeline

Juan Manuel Hernandez; Mariana Fernandez-Espinosa; Denis Parra; Diego Gomez-Zara

cs.CVApr 2, 2026

ViT-Explainer: An Interactive Walkthrough of the Vision Transformer Pipeline

Authors: Juan Manuel Hernandez, Mariana Fernandez-Espinosa, Denis Parra, Diego Gomez-Zara

Organizations: Pontificia Universidad Católica de Chile · University of Notre Dame

Abstract

Transformer-based architectures have become the shared backbone of natural language processing and computer vision. However, understanding how these models operate remains challenging, particularly in vision settings, where images are processed as sequences of patch tokens. Existing interpretability tools often focus on isolated components or expert-oriented analysis, leaving a gap in guided, end-to-end understanding of the full inference pipeline. To bridge this gap, we present ViT-Explainer, a web-based interactive system that provides an integrated visualization of Vision Transformer inference, from patch tokenization to final classification. The system combines animated walkthroughs, patch-level attention overlays, and a vision-adapted Logit Lens within both guided and free exploration modes. A user study with six participants suggests that ViT-Explainer is easy to learn and use, helping users interpret and understand Vision Transformer behavior.

Figures & tables

Figure 1: ViT-Explainer interface. Top: The input image is divided into non-overlapping patches, decomposed into RGB channels, flattened into vectors, and projected into an embedding space via a learned linear transformation. Bottom: A Transformer encoder block is visualized step by step, including layer normalization, Multi-Head Self-Attention, residual connections, and the MLP sublayer. The classification head produces logits from the class token, and users can navigate the processing stages through guided controls (top-left).

Figure 2: Multi-Head Self-Attention: ViT-Explainer animates the full attention computation to show how each head redistributes token information.

Figure 3: Attention Mapping: Visualizes how much attention a selected patch (or the [CLS] token) pays to other patches.

Figure 4: Logit-Lens: Layer-by-layer logit chart (colored curves for the top predicted classes) and a side panel to select and reshuffle classes for comparison.

Explore similar work

Inside the Visual Mind: Neuroscience-Motivated Concept Circuits for Interpreting and Steering Vision Transformers
Jun 4, 2026Tang Li, Yanlin Chen, Mengmeng Ma +1Transformer Interpretability Vision Transformer
Department of Computer & Information Science, University of Delaware, Newark, DE, USA · Department of Computer Science, University of Virginia, Charlottesville, VA, USA
Decision-Aware Attention Propagation for Vision Transformer Explainability
Apr 20, 2026Sehyeong Jo, Gangjae Jang, Haesol ParkTransformer Interpretability Vision Transformer
Aarhus University · University of Colorado Boulder · KIST
What Helps---and What Hurts: Bidirectional Explanations for Vision Transformers
Mar 2, 2026Qin Su, Tie LuoTransformer Interpretability Explainable Image Classification
Department of Electrical and Computer Engineering University of Kentucky, Lexington, KY, USA

Jun 4, 2026J/K move · Enter open · S save

Inside the Visual Mind: Neuroscience-Motivated Concept Circuits for Interpreting and Steering Vision Transformers

Tang Li, Yanlin Chen, Mengmeng Ma, Xi Peng

Department of Computer & Information Science, University of Delaware, Newark, DE, USA · Department of Computer Science, University of Virginia, Charlottesville, VA, USA

Transformer Interpretability Vision Transformer Sparse Autoencoders Mechanistic Interpretability Concept-Based Explanations

Open paper AskPDF arXiv

Despite high accuracy, Vision Transformer (ViT) predictions can be driven by spurious cues, raising the need to understand their inner workings before safe deployment. Sparse autoencoders (SAEs) provide a promising lens for decomposing model representations into human-interpretable concepts, yet adapting SAE-based interpretation to ViTs remains challenging due to limited control over concept coverage and subjective, non-scalable feature interpretation. To fill the gaps, motivated by neuroscience-inspired principles, we propose ViSAE, a mechanistic interpretability toolbox for understanding ViT inner workings through concept circuits. ViSAE consists of three components: (1) A probing suite with 64K images and a 16K visually grounded concept vocabulary, improving concept coverage efficiency by 20x over ImageNet and interpretation accuracy by 28.7% over existing concept sets. (2) Top-down concept reading and Bottom-up circuit tracing algorithms that automatically recover ViT inner workings via concept circuits. (3) Applications for auditing and steering ViT behavior. Through concept editing, ViSAE improves the worst-group accuracy on WaterBirds by 48.2%, outperforming existing methods by 23.8%. Our data and code: https://github.com/deep-real/ViSAE.

Abstract

Explore similar work

Inside the Visual Mind: Neuroscience-Motivated Concept Circuits for Interpreting and Steering Vision Transformers

Decision-Aware Attention Propagation for Vision Transformer Explainability

What Helps---and What Hurts: Bidirectional Explanations for Vision Transformers