spaceml / galaxygan_python Goto Github PK

Hello, GPU test is failing when running GPU test:
print "TEST if GPU is available: " + str(tf.test.is_gpu_available()) in 'train.py' :

I've followed the instructions and launched the given EC2 machine with the provided image. I did the test since training was running very slow, and it turned out that the GPU is not being used.
Does anyone know how to make it use the GPU? Thank you in advance.

It seems like folder GalaxyGAN_Python/figures/checkpoint is empty on the AMI provided. "./datasets/facades/checkpoint/model_100.ckpt" is not found either.

Can you release your pretrained model onto ami?

Would it be possible to share the dataset as a simple download, protobuf, npy or whatever else?
I guess I could pull it out of the ec2 instance, but that seems overly complicated.