pggan_8xb4-12Mimg_celeba-hq-1024x1024.py

_base_ = ['../_base_/gen_default_runtime.py']

# define GAN model
model = dict(
    type='ProgressiveGrowingGAN',
    data_preprocessor=dict(type='GenDataPreprocessor'),
    noise_size=512,
    generator=dict(type='PGGANGenerator', out_scale=1024, noise_size=512),
    discriminator=dict(type='PGGANDiscriminator', in_scale=1024),
    nkimgs_per_scale={
        '4': 600,
        '8': 1200,
        '16': 1200,
        '32': 1200,
        '64': 1200,
        '128': 1200,
        '256': 1200,
        '512': 1200,
        '1024': 12000,
    },
    transition_kimgs=600,
    ema_config=dict(interval=1))

# MODEL
model_wrapper_cfg = dict(find_unused_parameters=True)

# TRAIN
train_cfg = dict(max_iters=280000)

optim_wrapper = dict(
    constructor='PGGANOptimWrapperConstructor',
    generator=dict(optimizer=dict(type='Adam', lr=0.001, betas=(0., 0.99))),
    discriminator=dict(
        optimizer=dict(type='Adam', lr=0.001, betas=(0., 0.99))),
    lr_schedule=dict(
        generator={
            '128': 0.0015,
            '256': 0.002,
            '512': 0.003,
            '1024': 0.003
        },
        discriminator={
            '128': 0.0015,
            '256': 0.002,
            '512': 0.003,
            '1024': 0.003
        }))

# DATA
dataset_type = 'GrowScaleImgDataset'

pipeline = [
    dict(type='LoadImageFromFile', key='img'),
    dict(type='Flip', keys=['img'], direction='horizontal'),
    dict(type='PackEditInputs')
]

train_dataloader = dict(
    num_workers=4,
    batch_size=64,
    dataset=dict(
        type=dataset_type,
        data_roots={
            '64': './data/celebahq/imgs_64',
            '256': './data/celebahq/imgs_256',
            '512': './data/celebahq/imgs_512',
            '1024': './data/celebahq/imgs_1024'
        },
        gpu_samples_base=4,
        # note that this should be changed with total gpu number
        gpu_samples_per_scale={
            '4': 64,
            '8': 32,
            '16': 16,
            '32': 8,
            '64': 4
        },
        len_per_stage=300000,
        pipeline=pipeline),
    sampler=dict(type='InfiniteSampler', shuffle=True))

test_dataloader = dict(
    num_workers=4,
    batch_size=64,
    dataset=dict(
        type='BasicImageDataset',
        pipeline=pipeline,
        data_root='./data/celebahq/imgs_1024'),
    sampler=dict(type='DefaultSampler', shuffle=False))

# VIS_HOOK + DATAFETCH
custom_hooks = [
    dict(
        type='GenVisualizationHook',
        interval=5000,
        fixed_input=True,
        # vis ema and orig at the same time
        vis_kwargs_list=dict(
            type='Noise',
            name='fake_img',
            sample_model='ema/orig',
            target_keys=['ema', 'orig'])),
    dict(type='PGGANFetchDataHook')
]
default_hooks = dict(
    checkpoint=dict(
        max_keep_ckpts=20,
        save_best=['swd/avg', 'ms-ssim/avg'],
        rule=['less', 'greater']))

# METRICS
metrics = [
    dict(
        type='SWD', fake_nums=16384, image_shape=(3, 1024, 1024),
        prefix='SWD'),
    dict(type='MS_SSIM', fake_nums=10000, prefix='MS-SSIM')
]

# do not evaluate in training
val_cfg = val_evaluator = val_dataloader = None
test_evaluator = dict(metrics=metrics)