test_CAT_car_q10.yml

# general settings
name: test_CAR_CAT_q10
model_type: CATModel
scale: 1
num_gpu: 1
manual_seed: 10

datasets:
  test_1:  # the 1st test dataset
    task: CAR
    name: Classic5
    type: PairedImageDataset
    dataroot_gt: datasets/CAR/classic5/Classic5_HQ
    dataroot_lq: datasets/CAR/classic5/Classic5_LQ/10
    filename_tmpl: '{}'
    io_backend:
      type: disk

  test_2:  # the 2st test dataset
    task: CAR
    name: LIVE1
    type: PairedImageDataset
    dataroot_gt: datasets/CAR/LIVE1/LIVE1_HQ
    dataroot_lq: datasets/CAR/LIVE1/LIVE1_LQ/10
    filename_tmpl: '{}'
    io_backend:
      type: disk

  test_3:  # the 3st test dataset
    task: CAR
    name: Urban100
    type: PairedImageDataset
    dataroot_gt: datasets/CAR/Urban100/Urban100_HQ
    dataroot_lq: datasets/CAR/Urban100/Urban100_LQ/10
    filename_tmpl: '{}'
    io_backend:
      type: disk


# network structures
network_g:
  type: CAT
  upscale: 1
  in_chans: 1
  img_size: 128
  split_size_0: [2,2,2,4,4,4]
  split_size_1: [0,0,0,0,0,0] # The other side is H or W
  img_range: 255.
  depth: [6,6,6,6,6,6]
  embed_dim: 180
  num_heads: [6,6,6,6,6,6]
  mlp_ratio: 4
  resi_connection: '1conv'
  block_name: 'CATB_axial'

# path
path:
  pretrain_network_g: experiments/pretrained_models/CAR/CAR_CAT_q10.pth
  strict_load_g: True

# validation settings
val:
  save_img: False
  suffix: ~  # add suffix to saved images, if None, use exp name
  use_chop: True

  metrics:
    psnr: # metric name, can be arbitrary
      type: calculate_psnr
      crop_border: 0
      test_y_channel: True
    ssim:
      type: calculate_ssim
      crop_border: 0
      test_y_channel: True