mrdbourke/pytorch-deep-learning

This repository contains the course materials for "Learn PyTorch for Deep Learning: Zero to Mastery," a structured curriculum for learning PyTorch from scratch. PyTorch is a popular open-source framework (a pre-built toolkit) used to build and train machine learning models, software systems that learn patterns from data rather than following hand-written rules. The course is organized into ten sections, starting with the fundamentals of PyTorch's core building blocks, then progressing through increasingly advanced topics: building neural networks (software inspired loosely by the human brain), classification problems (teaching a model to sort things into categories), computer vision (teaching a model to understand images), working with custom datasets, organizing code into reusable modules, transfer learning (adapting an already-trained model to a new task), experiment tracking, replicating published research, and finally deploying a trained model to the internet so others can use it. The materials are available as Jupyter Notebooks (interactive documents that mix explanatory text with runnable code), with a companion online book at learnpytorch.io and video content on YouTube. Each section includes exercises for practice. The full course contains 321 videos and covers the complete workflow from writing your first PyTorch code to getting a working model in front of users. It is aimed at people who are new to deep learning and machine learning, not just experienced engineers. The primary language is Python, and the materials work with both PyTorch 1.x and PyTorch 2.0.