from templates import * | |
from templates_latent import * | |
if __name__ == '__main__': | |
# 256 requires 8x v100s, in our case, on two nodes. | |
# do not run this directly, use `sbatch run_ffhq256.sh` to spawn the srun properly. | |
gpus = [0, 1, 2, 3] | |
nodes = 2 | |
conf = ffhq256_autoenc() | |
train(conf, gpus=gpus, nodes=nodes) |