Out of memory when the resolution is more than 256x256

[malisit]

Hi, Thanks for the great implementations! I'm trying to train a PGAN to generate images from a custom dataset of 1024x1024 images. Up until generation of 512x512 there is no issues but when it starts training for 512x512 it immediately stops and raises an out of memory error. I was wondering what would be the right way to lower the batch_size so I wouldn't run out of memory.

Here's my exact config, it's the same as the celeaHQ

{
  "pathDB": "dataset",
  "config": {},
  "miniBatchScheduler": {
    "7": 12,
    "8": 8
  }
}

Thanks!

EDIT: I don't know what I was thinking when I opened up this issue, answer is in the config. smh Sorry for the inconvenience!