def update_argparser(parser):
models.update_argparser(parser)
args, _ = parser.parse_known_args()
parser.add_argument('--num_blocks',
help='Number of residual blocks in networks.',
default=16,
type=int)
parser.add_argument('--num_residual_units',
help='Number of residual units in networks.',
default=32,
type=int)
parser.add_argument('--width_multiplier',
help='Width multiplier inside residual blocks.',
default=4,
type=float)
parser.add_argument('--temporal_size',
help='Number of frames for burst input.',
default=None,
type=int)
if args.dataset.startswith('div2k'):
parser.set_defaults(
train_epochs=30,
learning_rate_milestones=(20, 25),
learning_rate_decay=0.2,
save_checkpoints_epochs=1,
lr_patch_size=48,
train_temporal_size=1,
eval_temporal_size=1,
)
else:
raise NotImplementedError(
'Needs to tune hyper parameters for new dataset.')
def update_argparser(parser):
models.update_argparser(parser)
args, _ = parser.parse_known_args()
if args.dataset.startswith('video'):
parser.add_argument('--num-blocks',
help='Number of residual blocks in networks',
default=16,
type=int)
parser.add_argument('--num-residual-units',
help='Number of residual units in networks',
default=32,
type=int)
parser.add_argument('--width_multiplier',
help='Width multiplier inside residual blocks',
default=4,
type=int)
parser.set_defaults(
train_epochs=20,
learning_rate_milestones=(15, 18),
save_checkpoints_epochs=1,
lr_patch_size=64,
train_temporal_size=1,
eval_temporal_size=1,
)
else:
raise NotImplementedError(
'Needs to tune hyper parameters for new dataset.')
def update_argparser(parser):
models.update_argparser(parser)
args, _ = parser.parse_known_args()
parser.add_argument('--num-steps',
help='Number of steps in recurrent networks',
default=12,
type=int)
parser.add_argument('--num-filters',
help='Number of filters in networks',
default=128,
type=int)
parser.add_argument('--non-local-field-size',
help='Size of receptive field in non-local blocks',
default=35,
type=int)
parser.add_argument(
'--init-ckpt',
help='Checkpoint path to initialize',
default=None,
type=str,
)
parser.set_defaults(
train_steps=500000,
learning_rate=((100000, 200000, 300000, 400000, 450000),
(1e-3, 5e-4, 2.5e-4, 1.25e-4, 6.25e-5, 3.125e-5)),
save_checkpoints_steps=20000,
save_summary_steps=1000,
)
def update_argparser(parser):
models.update_argparser(parser)
args, _ = parser.parse_known_args()
if args.dataset == 'cifar10':
parser.add_argument('--num-layers',
help='Number of layers in networks',
default=110,
type=int)
parser.add_argument('--mixup',
help='Hyper parameter for mixup training',
default=0.0,
type=float)
parser.set_defaults(
train_steps=150000,
learning_rate=((32000, 48000, 120000), (0.1, 0.01, 0.001, 0.0002)),
save_checkpoints_steps=5000,
)
else:
raise NotImplementedError(
'Needs to tune hyper parameters for new dataset.')
def update_argparser(parser):
models.update_argparser(parser)
args, _ = parser.parse_known_args()
if args.dataset == 'div2k':
parser.add_argument('--num-blocks',
help='Number of residual blocks in networks',
default=16,
type=int)
parser.add_argument('--num-residual-units',
help='Number of residual units in networks',
default=64,
type=int)
parser.set_defaults(
train_steps=1500000,
learning_rate=((1000000, ), (1e-4, 5e-5)),
save_checkpoints_steps=50000,
save_summary_steps=10000,
)
else:
raise NotImplementedError(
'Needs to tune hyper parameters for new dataset.')