val_set = None
else:
train_set = MF(train=True, real=real, **kwargs)
val_set = MF(train=False, real=real, **kwargs)
else:
raise NotImplementedError
# trainer
config_name = args.config_file.split('/')[-1]
config_name = config_name.split('.')[0]
experiment_name = '{:s}_{:s}_{:s}_{:s}'.format(args.dataset, args.scene,
args.model, config_name)
if args.learn_beta:
experiment_name = '{:s}_learn_beta'.format(experiment_name)
if args.learn_gamma:
experiment_name = '{:s}_learn_gamma'.format(experiment_name)
experiment_name += args.suffix
trainer = Trainer(model,
optimizer,
train_criterion,
args.config_file,
experiment_name,
train_set,
val_set,
device=args.device,
checkpoint_file=args.checkpoint,
resume_optim=args.resume_optim,
val_criterion=val_criterion)
lstm = args.model == 'vidloc'
trainer.train_val(lstm=lstm)
train_set = Env(train=True, **kwargs)
val_set = Env(train=False, **kwargs)
elif args.model.find('mapnet') >= 0:
kwargs = dict(kwargs, skip=skip, steps=steps)
train_set = MF(train=True, **kwargs)
val_set = MF(train=False, **kwargs)
else:
raise NotImplementedError
# trainer
config_name = args.config_file.split('/')[-1]
config_name = config_name.split('.')[0]
if args.reduce is None:
experiment_name = '{:s}_{:s}_{:s}_{:s}'.format(args.dataset, args.scene,
args.model, config_name)
else:
experiment_name = '{:s}_{:s}_{:s}_{:s}_reduce'.format(
args.dataset, args.scene, args.model, config_name)
trainer = Trainer(model,
optimizer,
train_criterion,
args.config_file,
experiment_name,
train_set,
val_set,
device='0',
checkpoint_file=args.checkpoint,
resume_optim=False,
val_criterion=val_criterion)
trainer.train_val()
experiment_name = '{:s}_augmented'.format(experiment_name)
elif args.use_augmentation == 'only':
experiment_name = '{:s}_only_augmented'.format(experiment_name)
if args.use_stylization:
experiment_name = '{:s}_stylized'.format(experiment_name)
if args.use_stylization > 1:
experiment_name = '{:s}_{:d}_styles'.format(experiment_name,
args.use_stylization)
if args.use_synthetic:
experiment_name = '{:s}_synthetic'.format(experiment_name)
if det_seed >= 0:
experiment_name = '{:s}_seed{}'.format(experiment_name, det_seed)
#if args.styles > 0:
# experiment_name = '{:s}_{}_styles'.format(experiment_name, args.styles)
experiment_name += args.suffix
trainer = Trainer(model,
optimizer,
train_criterion,
args.config_file,
experiment_name,
train_set,
val_set,
device=args.device,
checkpoint_file=args.checkpoint,
visdom_server=args.server,
visdom_port=args.port,
resume_optim=args.resume_optim,
val_criterion=val_criterion)
lstm = args.model == 'vidloc'
trainer.train_val(lstm=lstm, dual_target='multitask' in args.model)