This is a more detailed tutorial for loading a custom dataset based off of the Quickstart Tutorial from Pytorch. It loads Pytorch's MNIST dataset to train and my provided dataset to test. Code derived from the Quickstart Tutorial is slimmed down to focus more on the loading a custom dataset with test values like "epochs" minimized to debug for compatibility faster.
MyDatasetTutorial.zip - Jupyter notebook option. Includes all the dataset files in the directory stucture needed to run locally.
dataloadingtutorial_2025.ipynb - Google Colab option
MyImages.zip - The image dataset of personally handwritten 0's and 1's
- Resolution: 32x32
- Format: 8-bit greyscale JPEG
- Items: 1000 images
MyLabels.csv - References index to the image name and its (0 or 1) label