N

Nat Mini In1k 224

Developed by shi-labs
NAT-Mini is a lightweight vision Transformer model based on neighborhood attention mechanism, designed for ImageNet image classification tasks
Downloads 109
Release Time : 11/15/2022

Model Overview

NAT is a hierarchical vision Transformer based on Neighborhood Attention, achieving efficient image classification through constrained self-attention patterns

Model Features

Neighborhood Attention Mechanism
Uses constrained self-attention patterns where each token's receptive field is limited to nearest neighboring pixels, preserving translation equivariance
Efficient Architecture
Hierarchical vision Transformer design that reduces computational complexity while maintaining performance
Flexible Implementation
Implemented in PyTorch through the NATTEN extension library, supporting sliding window attention patterns

Model Capabilities

Image Classification
Visual Feature Extraction

Use Cases

Computer Vision
ImageNet Image Classification
Classifies images into 1000 ImageNet categories
Accuracy metrics not provided
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
Š 2025AIbase