Error when trying to get untrained ResNet-50 architecture

Hello everyone.

I want to train the CNN ResNet-50 from scratch. According to the manual (https://www.mathworks.com/help/deeplearning/ref/resnet50.html), I can get an untrained version using the following command:

lgraph = resnet50('Weights','none')

However, this command returns an error:

Error using resnet50

Too many input arguments.

I've already used the pretrained version of ResNet-50 with no problems through the following command:

net = resnet50

Am I missing something?

Thanks for helping