How to solve this error- Size Mismatch error

1

I am new to pyTorch and getting following Size Mismatch error when i try to run the DA-GAN:

size mismatch for module.conv1.filter.weight: copying a param of torch.Size([96, 3, 5, 5]) from checkpoint, where the shape is torch.Size([96, 1, 5, 5]) in current model.

2

I think this is what is happening: Your model expects 3-channel images, and some input instance has only one channel.

3

The error seems to point to a shape mismatch while loading a state_dict, so my guess would be that the model architecture might have been changed after storing the state_dict.