AttributeError: 'ACE' object has no attribute 'fc_mu105'

[infinityrgb]

Traceback (most recent call last): File "train.py", line 47, in trainer.run_generator_one_step(data_i) File "/d6295745ef534beab3ce2490bedcd8ab/lxy/inkpaint/SEAN/trainers/pix2pix_trainer.py", line 35, in run_generator_one_step g_losses, generated = self.pix2pix_model(data, mode='generator') File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py", line 541, in call result = self.forward(*input, kwargs) File "/usr/local/lib/python3.6/dist-packages/torch/nn/parallel/data_parallel.py", line 150, in forward return self.module(*inputs[0], *kwargs[0]) File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py", line 541, in call result = self.forward(input, kwargs) File "/d6295745ef534beab3ce2490bedcd8ab/lxy/inkpaint/SEAN/models/pix2pix_model.py", line 45, in forward input_semantics, real_image) File "/d6295745ef534beab3ce2490bedcd8ab/lxy/inkpaint/SEAN/models/pix2pix_model.py", line 144, in compute_generator_loss input_semantics, real_image, compute_kld_loss=self.opt.use_vae) File "/d6295745ef534beab3ce2490bedcd8ab/lxy/inkpaint/SEAN/models/pix2pix_model.py", line 196, in generate_fake fake_image = self.netG(input_semantics, real_image) File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py", line 541, in call result = self.forward(*input, kwargs) File "/d6295745ef534beab3ce2490bedcd8ab/lxy/inkpaint/SEAN/models/networks/generator.py", line 84, in forward x = self.head_0(x, seg, style_codes, obj_dic=obj_dic) File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py", line 541, in call result = self.forward(*input, *kwargs) File "/d6295745ef534beab3ce2490bedcd8ab/lxy/inkpaint/SEAN/models/networks/architecture.py", line 75, in forward dx = self.ace_0(x, seg, style_codes, obj_dic) File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py", line 541, in call result = self.forward(input, kwargs) File "/d6295745ef534beab3ce2490bedcd8ab/lxy/inkpaint/SEAN/models/networks/normalization.py", line 157, in forward middle_mu = F.relu(self.getattr('fc_mu' + str(j))(style_codes[i][j])) File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py", line 585, in getattr type(self).name, name)) AttributeError: 'ACE' object has no attribute 'fc_mu105'

I have meet this question, someone can help me?

[zmehdiz97]

You probably have less than 105 labels in you datset. Try --label_nc 104

[omaralvarez]

I have the same issue, when changing the number of labels from the default 19 to 42 it errors out similarly, looks like label_nc changes are not properly supported:

middle_mu = F.relu(self.__getattr__('fc_mu' + str(j))(style_codes[i][j])) File "/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1130, in __getattr__ raise AttributeError("'{}' object has no attribute '{}'".format( AttributeError: 'ACE' object has no attribute 'fc_mu19'

I have double and triple checked that my dataset is properly formatted according to the author guidelines.

[omaralvarez]

Yes, exactly, you need to modify the code in normalization.py for it to work with more labels. I'll try to make a pull request to fix it.

[fido20160817]

Yes, exactly, you need to modify the code in normalization.py for it to work with more labels. I'll try to make a pull request to fix it.

How to change? can you give me some tips?

[omaralvarez]

As you can see in:

https://github.com/ZPdesu/SEAN/blob/04c7536ff3fecd2d1a09c9ae046a1144636033a5/models/networks/normalization.py#L209-L227

There are only 19 fc_muXX, if you want more classes you need to add more, if not it will error out. There are more hardcoded things throughout the repo, so be on the lookout, I wish I could enumerate every single one, but it has been a while since I dealt with this codebase.

[fido20160817]

As you can see in:

https://github.com/ZPdesu/SEAN/blob/04c7536ff3fecd2d1a09c9ae046a1144636033a5/models/networks/normalization.py#L209-L227

There are only 19 fc_muXX, if you want more classes you need to add more, if not it will error out. There are more hardcoded things throughout the repo, so be on the lookout, I wish I could enumerate every single one, but it has been a while since I dealt with this codebase.

I see. I used following codes to help generate these sentences and paste them to 'normalization.py' and set the num of classes ahead.

generated_list_txt=open('more_fc_mu.txt', mode='a+')
for i in range(151):
  generated_list_txt.writelines(f'self.fc_mu{i} = nn.Linear(style_length, style_length)' + '\n')