RuntimeError: The expanded size of the tensor (3) must match the existing size (2) at non-singleton dimension 0. Target sizes: [3, -1, -1]. Tensor sizes: [2, 128, 128]

Alright, I’m completely stuck. I keep running into this error. I don’t know much about how GANs work but the error makes me think something is wrong with one of my data samples. I’ve been trying to refine my selection but when there’s well over 3000 samples I can’t really go through one-by-one to try to find some outlier.

I initially thought it was being picky against grayscale images, so I filtered them out, but to no avail.

I’ve tried with and without --transparency in case somehow an image containing some made it in, but output is not different.

Then I thought maybe it was trying to include sub-directories in the data folder, so I removed those, and nothing changed.

I adjusted batch size and related components in case this was a memory-related error but active memory use remains ~3GB and the error persists.

A data pool of ~100 images picked from a different source made it through to iteration 3000+ without encountering this breakage, so that furthers my assumption one of my samples is broken.

I’d appreciate some insight on what these dimensions are and how I can go about troubleshooting.

$ stylegan2_pytorch --data /media/LinuxDATA/nebula/x/generic_old/AAA-normal-stylegan --name illusfront --attn-layers [1,2] --aug-prob 0.2 --batch-size 1 --gradient-accumulate-every 15 --network-capacity 48 --new

illusfront</media/LinuxDATA/nebula/x/generic_old/AAA-normal-stylegan>:   0%|         | 0/150000 [00:00<?, ?it/s]/home/haruka/.local/lib/python3.6/site-packages/torch/nn/functional.py:3121: UserWarning: Default upsampling behavior when mode=bilinear is changed to align_corners=False since 0.4.0. Please specify align_corners=True if the old behavior is desired. See the documentation of nn.Upsample for details.
  "See the documentation of nn.Upsample for details.".format(mode))
/home/haruka/.local/lib/python3.6/site-packages/adamp/adamp.py:80: UserWarning: This overload of add_ is deprecated:
	add_(Number alpha, Tensor other)
Consider using one of the following signatures instead:
	add_(Tensor other, *, Number alpha) (Triggered internally at  /pytorch/torch/csrc/utils/python_arg_parser.cpp:766.)
  exp_avg.mul_(beta1).add_(1 - beta1, grad)
G: 104.41 | D: 9.24 | GP: 371.04 | PL: 0.00 | CR: 0.00 | Q: 0.00
illusfront</media/LinuxDATA/nebula/x/generic_old/AAA-normal-stylegan>:   0%| | 25/150000 [03:56<394:52:38,  9.48
Traceback (most recent call last):
  File "/home/haruka/.local/bin/stylegan2_pytorch", line 92, in <module>
    fire.Fire(train_from_folder)
  File "/home/haruka/.local/lib/python3.6/site-packages/fire/core.py", line 138, in Fire
    component_trace = _Fire(component, args, parsed_flag_args, context, name)
  File "/home/haruka/.local/lib/python3.6/site-packages/fire/core.py", line 468, in _Fire
    target=component.__name__)
  File "/home/haruka/.local/lib/python3.6/site-packages/fire/core.py", line 672, in _CallAndUpdateTrace
    component = fn(*varargs, **kwargs)
  File "/home/haruka/.local/bin/stylegan2_pytorch", line 87, in train_from_folder
    retry_call(model.train, tries=3, exceptions=NanException)
  File "/home/haruka/.local/lib/python3.6/site-packages/retry/api.py", line 101, in retry_call
    return __retry_internal(partial(f, *args, **kwargs), exceptions, tries, delay, max_delay, backoff, jitter, logger)
  File "/home/haruka/.local/lib/python3.6/site-packages/retry/api.py", line 33, in __retry_internal
    return f()
  File "/home/haruka/.local/lib/python3.6/site-packages/stylegan2_pytorch/stylegan2_pytorch.py", line 749, in train
    image_batch = next(self.loader).cuda()
  File "/home/haruka/.local/lib/python3.6/site-packages/stylegan2_pytorch/stylegan2_pytorch.py", line 113, in cycle
    for i in iterable:
  File "/home/haruka/.local/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 363, in __next__
    data = self._next_data()
  File "/home/haruka/.local/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 989, in _next_data
    return self._process_data(data)
  File "/home/haruka/.local/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 1014, in _process_data
    data.reraise()
  File "/home/haruka/.local/lib/python3.6/site-packages/torch/_utils.py", line 395, in reraise
    raise self.exc_type(msg)
RuntimeError: Caught RuntimeError in DataLoader worker process 5.
Original Traceback (most recent call last):
  File "/home/haruka/.local/lib/python3.6/site-packages/torch/utils/data/_utils/worker.py", line 185, in _worker_loop
    data = fetcher.fetch(index)
  File "/home/haruka/.local/lib/python3.6/site-packages/torch/utils/data/_utils/fetch.py", line 44, in fetch
    data = [self.dataset[idx] for idx in possibly_batched_index]
  File "/home/haruka/.local/lib/python3.6/site-packages/torch/utils/data/_utils/fetch.py", line 44, in <listcomp>
    data = [self.dataset[idx] for idx in possibly_batched_index]
  File "/home/haruka/.local/lib/python3.6/site-packages/stylegan2_pytorch/stylegan2_pytorch.py", line 244, in __getitem__
    return self.transform(img)
  File "/home/haruka/.local/lib/python3.6/site-packages/torchvision/transforms/transforms.py", line 61, in __call__
    img = t(img)
  File "/home/haruka/.local/lib/python3.6/site-packages/torchvision/transforms/transforms.py", line 359, in __call__
    return self.lambd(img)
  File "/home/haruka/.local/lib/python3.6/site-packages/stylegan2_pytorch/stylegan2_pytorch.py", line 212, in __call__
    return tensor.expand(self.num_channels, -1, -1)
RuntimeError: The expanded size of the tensor (3) must match the existing size (2) at non-singleton dimension 0.  Target sizes: [3, -1, -1].  Tensor sizes: [2, 128, 128]

[catthou]

Alright, I pulled out all the problematic images and set it loose on 'em. It seems to be spitting out models just fine, so thanks!

[lucidrains]

@catthou ok, done! https://github.com/lucidrains/stylegan2-pytorch/releases/tag/0.18.4 let me know if it works / doesn’t work