_, _, X_test, Y_test = get_data(dataset)
num_samples = np.shape(X_test)[0]
num_rand_samples = 1328
random_samples = np.random.randint(0, num_samples, num_rand_samples)
new_X_test = np.zeros((num_rand_samples, 1, 28, 28))
for i, sample_no in enumerate(random_samples):
new_X_test[i, 0, :, :] = (X_test[sample_no, :, :, 0])
new_Y_test = Y_test[random_samples, :]
f = open(os.path.join(args.log_dir, 'Random_Test_%s_.p' % (dataset)), 'w')
pickle.dump({"adv_input": new_X_test, "adv_labels": new_Y_test}, f)
f.close()
if (args.attack == 'cw-l2' or args.attack == 'all'):
#No softmax for Carlini attack
pytorch_network = Net()
pytorch_network.load_state_dict(torch.load(args_ckpt))
pytorch_network.eval()
model = Model(torch_model=pytorch_network, softmax=False)
keras_network = model.model
transfer.pytorch_to_keras(pytorch_network, model.model)
pytorch_network.eval()
model = model.model
batch_size = 16
craft_one_type(sess,
model,
new_X_test,
new_Y_test,
dataset,
'cw-l2',
batch_size,
list_noisy_names.append('noisy_' + advs)
attack_file = os.path.join(args.adv_ex_dir,
'Noisy_%s_%s.p' % (dataset, advs))
adv_loader = torch.utils.data.DataLoader(custom_datasets.Adv(
filename=attack_file, transp=True),
batch_size=args.batch_size,
shuffle=False,
**kwargs)
list_noisy_loader.append(adv_loader)
from model import CW_Net as Net
#from small_model import Very_Small_Net as Net
print("Eval using model", Net)
model = Net()
print("Loading ckpt", args.ckpt)
model.load_state_dict(torch.load(args.ckpt))
if args.cuda:
model.cuda()
model.eval()
print("Args:", args)
fixed_dxs = pickle.load(
open(os.path.join(args.fingerprint_dir, "fp_inputs_dx.pkl"), "rb"))
fixed_dys = pickle.load(
open(os.path.join(args.fingerprint_dir, "fp_outputs.pkl"), "rb"))
fp = Fingerprints()
fp_target[i, j, i] = 0.7
pickle.dump(fp_target, open(os.path.join(args.log_dir, "fp_outputs.pkl"),
"wb"))
fp_target = util.np2var(fp_target, args.cuda)
fp = Fingerprints()
fp.dxs = fp_dx
fp.dys = fp_target
#from model import Net
from model import CW_Net as Net
#from small_model import Very_Small_Net as Net
model = Net()
if args.cuda:
model.cuda()
optimizer = optim.Adam(model.parameters(), lr=args.lr)
print("Args:", args)
for epoch in range(1, args.epochs + 1):
if (epoch == 1):
fp_train.test(epoch, args, model, test_loader, fp.dxs, fp.dys)
fp_train.train(epoch, args, model, optimizer, train_loader, fp.dxs, fp.dys)
fp_train.test(epoch, args, model, test_loader, fp.dxs, fp.dys)
path = os.path.join(args.log_dir, "ckpt",
"state_dict-ep_{}.pth".format(epoch))