Add CPU option (fix issue #23)

[kheimpel]

Added "if CUDA" to all cuda() calls in trainer.py
Makes it possible to evaluate & train on CPU

[bjente]

This fix still doesn't enable me to evaluate and train on CPU... I keep on getting this error: AttributeError: module 'torch._C' has no attribute '_cuda_getDevice
Do you know how to fix it?

[kheimpel]

This fix still doesn't enable me to evaluate and train on CPU... I keep on getting this error: AttributeError: module 'torch._C' has no attribute '_cuda_getDevice
Do you know how to fix it?

Could you please provide filename and line number?

[bjente]

Traceback (most recent call last):
File "main.py", line 136, in
algo = trainer(output_dir, dataloader, dataset.n_words, dataset.ixtoword)
File "/Users/bentedebruin/Documents/AttnGAN/code/trainer.py", line 36, in init
torch.cuda.set_device(cfg.GPU_ID)
File "/Users/bentedebruin/miniconda3/lib/python3.7/site-packages/torch/cuda/init.py", line 264, in set_device
torch._C._cuda_setDevice(device)
AttributeError: module 'torch._C' has no attribute '_cuda_setDevice'

This is the error I'm getting when I put this in command line:
python main.py --cfg cfg/eval_coco.yml

[kheimpel]

Did you set GPU_ID = 0 in your config file (eval_coco.yml)?

[bjente]

I just did, but I get the exact same error. Same filenames and line numbers...

[kheimpel]

Ok. Please try GPU_ID = -1 in your yml file.

Sorry - has been a while...

[bjente]

Thank you so much! I don't get that error anymore!!! :)