Don't Panic! PyTorch Data Loaders and Datasets - DZone AI

When I first started using PyTorch, I was not happy with this new data loading paradigm that I had to learn. I was migrating to PyTorch from Keras, which wasn't as prescriptive. PyTorch had a specific way it wanted to access data, and I didn't know what it was, nor did I really want to spend time learning yet another way to load data into a deep learning framework. But I didn't really have any choice, so I got to work. I ended up looking through the code for the MNIST dataset to figure out what I needed to do.