Övül Arslan·Aug 13Reading More by Looking at Less: How Native Sparse Attention (NSA) WorksDeepSeek’s Native Sparse Attention
Övül Arslan·Nov 14, 2025Position by Rotation: The Intuition That Makes RoPE So Powerful (1D & 2D)Explanation of the paper “RoFormer: Enhanced Transformer with Rotary Position Embedding” RoPEA response icon2A response icon2
Övül Arslan·Aug 9, 2025MobileViT Explained: Global Context with Mobile Efficiency & PyTorch ImplementationMobileViT Paper Explained and PyTorch Implementation
Övül Arslan·Jul 25, 2025PVTv2: Small Tweaks, Big Gains — Evolved for Efficiency and Precision & PyTorch ImplementationPVT v2 Paper Explained and PyTorch Implementation
Övül Arslan·Jul 5, 2025Masked Autoencoders (MAE): The Art of Seeing More by Masking Most & PyTorch ImplementationMAE Paper Explained and PyTorch Implementation
Övül Arslan·Jun 15, 2025Making Transformers See: BEiT’s Visual Tokens and Masking MagicExploring how BEiT teaches Vision Transformers to understand images, all without labels.
Övül Arslan·Jun 1, 2025No Labels, No Problem: How DINO Learns to Understand Images From ScratchExploring how DINO teaches Vision Transformers to understand images all on their own.
Övül Arslan·May 17, 2025CrossViT: Vision Transformers That See the Forest AND the Squirrels & PyTorch ImplementationCrossViT Paper Explained and PyTorch Implementation
Övül Arslan·May 4, 2025CvT: The Best of CNNs and Transformers for Visual Recognition & PyTorch ImplementationCvT: Introducing Convolutions to Vision Transformers Paper Explained and PyTorch Implementation
Övül Arslan·Apr 19, 2025A Closer Look at Transformer in Transformer: Guide and PyTorch ImplementationTransformer in Transformer Paper Explained and PyTorch Implementation