Can't run on CPU

[yig]

On my mac laptop (no CUDA), I installed all the dependencies and downloaded all the pre-trained models.

git clone https://github.com/arnabgho/iSketchNFill
cd iSketchNFill
pip3 install -r requirements.txt
bash scripts/download_pretrained_scribble_dataset.sh

I then ran the example line, with --gpu_ids -1 to use the CPU version:

python3 main_gui_shadow_draw_color.py --name wgangp_sparse_label_channel_pix2pix_autocomplete_multiscale_nz_256_nc_1_nf_32_gp_0_multigpu --model sparse_wgangp_pix2pix  --checkpoints_dir checkpoints_sparse --gpu_ids -1 --nz 256 --sparseSize 4 --fineSize 128 --ngf 32 --ndf 32 --num_interpolate 8 --input_nc 1 --output_nc 1

I then got AssertionError: Torch not compiled with CUDA enabled errors in models/sparse_wgangp_pix2pix_model.py lines 113 (z= z.cuda()) and 123 (self.label = self.label.cuda()), so I changed them to:

if torch.cuda.is_available(): z= z.cuda()

and

if torch.cuda.is_available(): self.label = self.label.cuda()

That gets me past those errors, but then line 76 calls self.netG.cuda(). A debugger shows that the if statement gating this line (if self.isTrain) is indeed False.

Any advice?