A fast, efficient, and lightweight model for image segmentation — Hello There!! Today we’ll see how to implement SegFormer in PyTorch proposed in SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers. Code is here, an interactive version of this article can be downloaded from here. Let’s get started! The paper proposes a new transformer-based model to tackle image…

A new convnet that beats transformers — Hey, I am on LinkedIn come and say hi 👋 Hello There!! Today we are going to implement the famous ConvNext in PyTorch proposed in A ConvNet for the 2020s . Code is here, an interactive version of this article can be downloaded from here. Let’s get started! The paper…

transformers go brum brum Hi guys! Today we are going to implement Training data-efficient image transformers & distillation through attention a new method to perform knowledge distillation on Vision Transformers called DeiT. You will soon see how elegant and simple this new approach is. Code is here, an interactive version…

I am on LinkedIn, come and say hi 👋 Hi guys, happy new year! Today we are going to implement the famous Vi(sion) T(ransformer) proposed in AN IMAGE IS WORTH 16X16 WORDS: TRANSFORMERS FOR IMAGE RECOGNITION AT SCALE. Code is here, an interactive version of this article can be downloaded…

Benchmark the new PyTorch 2.0 model.compile Hey, I am on LinkedIn come and say hi 👋 code is here TL;DR Yes, it makes inference faster by around 22%. I still need to test it during training, but I think it will be similar. Introduction PyTorch 2.0 is out, let's see how fast…

Hey, I am on LinkedIn come and say hi 👋 Code is here OpenAI ChatGPT APIs are out! They open a new endless possibilities. Today we’re going to see how to use them in Python Setup Get an API key Log into OpenAI website and, if you are not on the free tier, add a…

I am on LinkedIn, come and say hi 👋 Today we’ll see how to implement non max suppression in PyTorch The code is here, and an interactive version of this article can be found here Preambula If you are doing computer vision (especially object detection), you know what non max suppression…

Hello There! Code is here Today we will see how to load a Pytorch model with linear memory consumption. Loading a model takes 2x memory space. Let’s see why: First, we need a model: Upon creation, the model takes 1x memory, where x is its size At some…

DropPath is available in glasses my computer vision library — Code is here an interactive version of this article can be downloaded from here. Introduction Today we are going to implement Stochastic Depth also known as Drop Path in PyTorch! Stochastic Depth introduced by Gao Huang et al is a technique to “deactivate” some layers during training. We’ll stick with DropPath. …