return 2**-8
elif 8000 <= x:
return 2**-11
update_step_size_method = 'schedule'
else:
raise NotImplementedError
# Set up the data, parameters, model, optimizer, and results objects
data = opt_structures.Data(train_loader, valid_loader, test_loader)
params = opt_structures.Params(num_classes=num_classes,
ckn=True,
train_w_layers=None,
lambda_filters=0,
normalize=True,
w_last_init=None,
update_step_size_method=update_step_size_method,
step_size_init=args.step_size,
step_size_schedule=step_size_schedule,
tau=args.hessian_reg,
num_iters=args.num_iters,
save_path=save_file + '_params.pickle',
eval_test_every=args.eval_test_every,
save_every=500)
model = opt_structures.Model(model, save_path=save_file + '_model.pickle')
results = opt_structures.Results(save_path=save_file + '_results.pickle')
optimizer = train_network.TrainSupervised(data, model, params, results)
# Train the model
optimizer.train()
data = opt_structures.Data(None,
train_unlabeled_loader,
None,
None,
test_loader,
deepcluster_loader=train_loader)
params = opt_structures.Params(
nclasses=nclasses,
min_frac_points_class=min_frac_points_class,
max_frac_points_class=max_frac_points_class,
ckn=ckn,
project=True,
lambda_pix=args.lambda_pix,
lam=args.lam,
normalize=True,
balanced=balanced_version,
labeling_method=args.labeling_method,
deepcluster_k=args.num_clusters,
deepcluster_update_clusters_every=args.update_clusters_every,
labeling_burnin=args.labeling_burnin,
step_size_init_sup=args.lr_sup_init,
step_size_init_semisup=args.lr_semisup,
maxiter=args.num_iters,
eval_test_every=args.eval_test_every,
save_every=args.save_every,
save_path=save_file + '_params.pickle')
model = opt_structures.Model(model, save_path=save_file + '_model.pickle')
results = opt_structures.Results(save_path=save_file + '_results.pickle')
optimizer = train_xsdc.TrainSupervised(data, model, params, results)
# Train the model with XSDC
train_loader, train_labeled_loader, train_unlabeled_loader, valid_loader, test_loader = \
mnist.get_dataloaders(data_path=args.data_path, num_labeled=args.num_labeled, num_train=args.num_train,
num_workers=0, seed=args.seed)
print('Size of training set:', len(train_loader.dataset))
# Load and initialize the model
model_params = parse_config.load_config('../../cfg/lenet-5_ckn.cfg')
if args.num_filters > 0:
nlayers = len(model_params['num_filters'])
model_params['num_filters'] = [args.num_filters] * nlayers
model_params['patch_sigma'] = [bw] * nlayers
layers = parse_config.create_layers(model_params)
model = net.CKN(layers).to(defaults.device)
model.init(train_loader)
print('Done with initialization')
# Set up the data, parameters, model, results, and optimizer objects
data = opt_structures.Data(train_labeled_loader, train_unlabeled_loader, valid_loader, test_loader)
params = opt_structures.Params(batch_size=args.batch_size, ckn=True, epsilon=args.epsilon,
eval_test_every=args.eval_test_every, lambda_cov=args.lambda_cov, lambda_params=0,
lr=args.lr, lr_schedule=None, min_dist=args.min_dist, num_classes=10,
num_iters=args.num_iters, project=True, save_every=args.save_every,
save_path=save_file + '_params.pickle', train_w_layers=[0, 2, 4, 5])
model = opt_structures.Model(model, save_path=save_file + '_model.pickle')
results = opt_structures.Results(save_path=save_file + '_results.pickle')
optimizer = train_network.TrainSupervised(data, model, params, results)
# Train the model
optimizer.train()
# Load and initialize the model
model = sum_sigmoids.SumSigmoids(args.num_filters)
model.apply(sum_sigmoids.init_normal)
model.to(defaults.device)
print('Initial thetas:', torch.stack(list(model.state_dict()['thetas'])))
# Set up the data, parameters, model, results, and optimizer objects
data = opt_structures.Data(train_labeled_loader, train_unlabeled_loader,
valid_loader, test_loader)
params = opt_structures.Params(ckn=False,
epsilon=args.epsilon,
eval_test_every=args.eval_test_every,
lambda_cov=args.lambda_cov,
lambda_params=args.lambda_params,
lr=args.lr,
lr_schedule=None,
min_dist=args.min_dist,
num_classes=2,
num_iters=args.num_iters,
project=False,
save_every=args.save_every,
save_path=save_file + '_params.pickle',
train_w_layers=None)
model = opt_structures.Model(model, save_path=save_file + '_model.pickle')
results = opt_structures.Results(save_path=save_file + '_results.pickle')
optimizer = train_network.TrainSupervised(data, model, params, results)
# Train the model
optimizer.train()
test_loader,
deepcluster_loader=train_loader)
params = opt_structures.Params(
nclasses=nclasses,
min_frac_points_class=min_frac_points_class,
max_frac_points_class=max_frac_points_class,
ckn=ckn,
project=True,
train_w_layers=[0, 2, 4, 5],
lam=args.lam,
normalize=True,
augment=args.augment,
add_constraints=add_constraints,
add_constraints_method=add_constraints_method,
add_constraints_frac=add_constraints_frac,
add_constraints_classes=add_constraints_classes,
balanced=balanced_version,
labeling_method=args.labeling_method,
rounding=args.rounding,
deepcluster_k=args.num_clusters,
deepcluster_update_clusters_every=args.update_clusters_every,
labeling_burnin=args.labeling_burnin,
step_size_init_sup=args.lr_sup_init,
step_size_init_semisup=args.lr_semisup,
update_lambda=args.update_lambda,
maxiter=args.num_iters,
eval_test_every=args.eval_test_every,
save_every=args.save_every,
save_path=save_file + '_params.pickle',
)
model = opt_structures.Model(model, save_path=save_file + '_model.pickle')