lama/configs/eval2.yaml

evaluator_kwargs:
  batch_size: 8
  device: cuda

dataset_kwargs:
  img_suffix: .png
  inpainted_suffix: .png