configuration.yaml

exp:
  detector_attack: False
  lambda_cri: 1
  lambda_det: 0

wandb:
  wandb_project: "batch1_jw"
  log_freq: 10

generator:
  pretrained_generator: "scratch"
  gen_arch: "progan"
# progan, 

detector:
  det_arch: "CLIP:ViT-L/14"
  det_ckpt: "experiments/detector/fc_weights.pth"
# 'Imagenet:resnet18','Imagenet:resnet34','Imagenet:resnet50','Imagenet:resnet101','Imagenet:resnet152','Imagenet:vgg11','Imagenet:vgg19','Imagenet:swin-b','Imagenet:swin-s','Imagenet:swin-t','Imagenet:vit_b_16','Imagenet:vit_b_32','Imagenet:vit_l_16','Imagenet:vit_l_32','CLIP:RN50','CLIP:RN101','CLIP:RN50x4','CLIP:RN50x16','CLIP:RN50x64','CLIP:ViT-B/32','CLIP:ViT-B/16','CLIP:ViT-L/14','CLIP:ViT-L/14@336px',

hyperparameters:
  criterion: "MSE"
  num_epochs: 500
  batch_size: 1
  # shuffle_batch = False
  # traindata_size = 100

optimizer:
  learning_rate: 0.0002
  learning_rate_noise: 0.0002
  betas: [0.9, 0.999]
  betas_noise: [0.9, 0.999]

scheduler:
  T_max: 100
  T_max_noise: 100
  eta_min: 0.0001
  eta_min_noise: 0.0001