CerebralSeed

What is considered optimal for the MIA score? Obviously, it should be lower than the initial model. But just wanted to clarify, are we aiming to have the last chart...

I suggest changing In [41] to ``` # download pre-trained weights import os path="weights_resnet18_cifar10.pth" if not os.path.exists(path): response = requests.get( "https://unlearning-challenge.s3.eu-west-1.amazonaws.com/weights_resnet18_cifar10.pth" ) open(path, "wb").write(response.content) weights_pretrained = torch.load("weights_resnet18_cifar10.pth", map_location=DEVICE) # load...

xformers installs overwrites torch with 2.2.1

1

The current instructions may need to be updated. This particular set of code results in xformers replacing torch with the latest version: ``` cd .. git clone https://github.com/facebookresearch/xformers.git cd xformers...

This is a feature request. Support for alternative dtypes would be quite helpful, such as float16 or bfloat16.

Handle corrupt images

4

This code will inform the user of a corrupt image in their set and continue training.