Check activated Layers in torch backend

[Pacman1984]

Hi,

is there a way to check the activated layers in the torch backend?

If i build a model from the ground with transformers and torch i usually do somethinglike this:

for name, param in model.named_parameters():
    print(f'{name}: {param.requires_grad}')

and it will show me the activated (trainable) layers.

Best

[ThilinaRajapakse]

You can do the same thing by accessing the model attribute of any Simple Transformers model. Alternatively, you can just use the get_named_parameters() method of ClassificationModel, NERModel, etc.