Open DavidRees87 opened 2 years ago
https://github.com/autonomousvision/projected_gan/issues/87#issuecomment-1180581191 someone else facing the same issue , they just deleted act1 from projector.py but this is not recommended
Should wait for @xl-sr 's reply on this issue in my opinion, or perhaps @woctezuma 's since he is also very active here.
This error occurs because of the python package "timm". It was updated yesterday and that broke it. Just install version 0.5.4 of timm.
replace pip install timm with pip install timm==0.5.4
I'm getting this error message when running in colab, didn't happen before today. Any ideas?
Constructing networks... Downloading: "https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-weights/tf_efficientnet_lite0-0aa007d2.pth" to /root/.cache/torch/hub/checkpoints/tf_efficientnet_lite0-0aa007d2.pth
AttributeError Traceback (most recent call last) in ()
21 seed=0,
22 workers=0,
---> 23 restart_every=999999,
24 )
9 frames in train(**kwargs)
78
79 # Launch.
---> 80 launch_training(c=c, desc=desc, outdir=opts.outdir)
/content/projected_gan/training/training_loop.py in training_loop(run_dir, training_set_kwargs, data_loader_kwargs, G_kwargs, D_kwargs, G_opt_kwargs, D_opt_kwargs, loss_kwargs, metrics, random_seed, num_gpus, rank, batch_size, batch_gpu, ema_kimg, ema_rampup, G_reg_interval, D_reg_interval, total_kimg, kimg_per_tick, image_snapshot_ticks, network_snapshot_ticks, resume_pkl, resume_kimg, cudnn_benchmark, abort_fn, progress_fn, restart_every) 159 common_kwargs = dict(c_dim=training_set.label_dim, img_resolution=training_set.resolution, img_channels=training_set.num_channels) 160 G = dnnlib.util.construct_class_by_name(G_kwargs, common_kwargs).train().requiresgrad(False).to(device) # subclass of torch.nn.Module --> 161 D = dnnlib.util.construct_class_by_name(D_kwargs, common_kwargs).train().requiresgrad(False).to(device) # subclass of torch.nn.Module 162 G_ema = copy.deepcopy(G).eval() 163
/content/projected_gan/dnnlib/util.py in construct_class_by_name(class_name, *args, kwargs) 301 def construct_class_by_name(*args, class_name: str = None, *kwargs) -> Any: 302 """Finds the python class with the given name and constructs it with the given arguments.""" --> 303 return call_func_by_name(args, func_name=class_name, kwargs) 304 305
/content/projected_gan/dnnlib/util.py in call_func_by_name(func_name, *args, *kwargs) 296 func_obj = get_obj_by_name(func_name) 297 assert callable(func_obj) --> 298 return func_obj(args, **kwargs) 299 300
/content/projected_gan/pg_modules/discriminator.py in init(self, diffaug, interp224, backbone_kwargs, kwargs) 159 self.diffaug = diffaug 160 self.interp224 = interp224 --> 161 self.feature_network = F_RandomProj(backbone_kwargs) 162 self.discriminator = MultiScaleD( 163 channels=self.feature_network.CHANNELS,
/content/projected_gan/pg_modules/projector.py in init(self, im_res, cout, expand, proj_type, **kwargs) 106 107 # build pretrained feature network and random decoder (scratch) --> 108 self.pretrained, self.scratch = _make_projector(im_res=im_res, cout=self.cout, proj_type=self.proj_type, expand=self.expand) 109 self.CHANNELS = self.pretrained.CHANNELS 110 self.RESOLUTIONS = self.pretrained.RESOLUTIONS
/content/projected_gan/pg_modules/projector.py in _make_projector(im_res, cout, proj_type, expand) 62 ### Build pretrained feature network 63 model = timm.create_model('tf_efficientnet_lite0', pretrained=True) ---> 64 pretrained = _make_efficientnet(model) 65 66 # determine resolution of feature maps, this is later used to calculate the number
/content/projected_gan/pg_modules/projector.py in _make_efficientnet(model) 33 def _make_efficientnet(model): 34 pretrained = nn.Module() ---> 35 pretrained.layer0 = nn.Sequential(model.conv_stem, model.bn1, model.act1, model.blocks[0:2]) 36 pretrained.layer1 = nn.Sequential(model.blocks[2:3]) 37 pretrained.layer2 = nn.Sequential(*model.blocks[3:5])
/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py in getattr(self, name) 1184 return modules[name] 1185 raise AttributeError("'{}' object has no attribute '{}'".format( -> 1186 type(self).name, name)) 1187 1188 def setattr(self, name: str, value: Union[Tensor, 'Module']) -> None:
AttributeError: 'EfficientNet' object has no attribute 'act1'