How can I load rotated mnist dataset (https://github.com/ChaitanyaBaweja/RotNIST/tree/master/data) using pytorch ?
The main.py seems to show how to load the data. Probably you could adapt the code to load the data first and then wrap it in a Dataset
. The Data Loading tutorial might help you to implement an own Dataset
.
Note that in a vanilla classification use case PyTorch expects a target filled with class indices while the code seems to load target in a one-hot encoded format.
To get the indices you can just call torch.argmax(targets, dim=1)
.
Thank you sir.