Issue in changing mobilenetv3 width_mult parameter

vision
1

I’m trying to change the width_mult=1.9 in Mobilenet_v3_large but getting the following error:

model = torchvision.models.mobilenet_v3_large(pretrained=True, width_mult=1.9, reduced_tail=False, dilated=False)

RuntimeError: Error(s) in loading state_dict for MobileNetV3:
size mismatch for features.0.0.weight: copying a param with shape torch.Size([16, 3, 3, 3]) from checkpoint, the shape in current model is torch.Size([32, 3, 3, 3]).
size mismatch for features.0.1.weight: copying a param with shape torch.Size([16]) from checkpoint, the shape in current model is torch.Size([32]).

2

You won’t be able to load a pretrained state_dict into the model if you change width_mult=1.9. Remove it and it should work. In case you need this change you might need to train the model from scratch.

3

Is their any other possible way I can get the weights without traning the model for width_mul=1.9