avalonstrel/GatedConvolution_pytorch

Start training from given latest_ckpt.pth.tar

DongHwanJang opened this issue · 1 comments

Is it possible to start transfer learning from given latest_ckpt.pth.tar?
I've tried already but it makes an error in netG.load_state_dict & netD.load_state_dict
Is it ok to replace those line with load_consistent_state_dict(netX_state_dict, netX) ?
Thx

I think it is a possible way. And what is the error you met? If you make sure the main layers weight have been assigned correctly, use load_consistent_state_dict may also work.