Exception: It's not possible to apply those transforms to your dataset: expected scalar type Float but found Double

[curryJ]

Description

Thanks for this great project.
I run the 02_state_of_the_art.ipynb,but I get a Exception which is the title;
My computer is cuda10.1 and GPU version is rtx2080ti.I update the version of pytorch and torchvision to 1.7.0 and 0.8.0 respectively

[PatrickBue]

Looks like a version conflict. You could try this conda environment, that should work:
https://github.com/microsoft/computervision-recipes/blob/master/environment.yml