x = F.relu(self.fc2(x))
nn.Dropout()
x = F.relu(self.fc3(x))
x = self.softmax(x)
return x
# Define a transform to normalize the data
transform = transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.5, ), (0.5, )),
])
# Download and load the training data
trainset = datasets.MNIST('./MNIST_data/',
download=True,
train=True,
transform=transform)
trainloader = torch.utils.data.DataLoader(trainset,
batch_size=128,
shuffle=True)
# Download and load the test data
testset = datasets.MNIST('./MNIST_data/',
download=True,
train=False,
transform=transform)
testloader = torch.utils.data.DataLoader(testset, batch_size=128, shuffle=True)
model = myModel()
#import numpy as np
def main():
# Training settings
parser = argparse.ArgumentParser(description='ONNX Runtime MNIST Example')
parser.add_argument(
'--train-steps',
type=int,
default=-1,
metavar='N',
help=
'number of steps to train. Set -1 to run through whole dataset (default: -1)'
)
parser.add_argument('--batch-size',
type=int,
default=20,
metavar='N',
help='input batch size for training (default: 20)')
parser.add_argument('--test-batch-size',
type=int,
default=1000,
metavar='N',
help='input batch size for testing (default: 1000)')
parser.add_argument('--epochs',
type=int,
default=1,
metavar='N',
help='number of epochs to train (default: 1)')
parser.add_argument('--lr',
type=float,
default=0.01,
metavar='LR',
help='learning rate (default: 0.01)')
parser.add_argument('--no-cuda',
action='store_true',
default=False,
help='disables CUDA training')
parser.add_argument('--seed',
type=int,
default=1,
metavar='S',
help='random seed (default: 1)')
parser.add_argument(
'--log-interval',
type=int,
default=10,
metavar='N',
help='how many batches to wait before logging training status')
parser.add_argument('--save-path',
type=str,
default='',
help='Path for Saving the current Model state')
# Basic setup
args = parser.parse_args()
if not args.no_cuda and torch.cuda.is_available():
device = "cuda"
else:
device = "cpu"
torch.manual_seed(args.seed)
onnxruntime.set_seed(args.seed)
# Data loader
train_loader = torch.utils.data.DataLoader(datasets.MNIST(
'./data',
train=True,
download=True,
transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.1307, ), (0.3081, ))
])),
batch_size=args.batch_size,
shuffle=True)
if args.test_batch_size > 0:
test_loader = torch.utils.data.DataLoader(
datasets.MNIST('./data',
train=False,
transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.1307, ), (0.3081, ))
])),
batch_size=args.test_batch_size,
shuffle=True)
# Modeling
model = NeuralNet(784, 500, 10)
model_desc = mnist_model_description()
optim_config = optim.SGDConfig(lr=args.lr)
opts = {'device': {'id': device}}
opts = ORTTrainerOptions(opts)
trainer = ORTTrainer(model,
model_desc,
optim_config,
loss_fn=my_loss,
options=opts)
# Train loop
for epoch in range(1, args.epochs + 1):
train(args.log_interval, trainer, device, train_loader, epoch,
args.train_steps)
if args.test_batch_size > 0:
test(trainer, device, test_loader)
# Save model
if args.save_path:
torch.save(model.state_dict(),
os.path.join(args.save_path, "mnist_cnn.pt"))
import torch
from torch import nn, optim
from torch.utils.data import DataLoader
from torchvision import transforms
from torchvision import datasets
from torchsummary import summary
# hyper parameters
batch_size = 100
learning_rate = 1e-3
num_epochs = 1
# download MNIST digit data set
train_set = datasets.MNIST(root='../data',
train=True,
transform=transforms.ToTensor(),
download=True)
test_set = datasets.MNIST(root='../data',
train=False,
transform=transforms.ToTensor())
train_loader = DataLoader(train_set, batch_size=batch_size, shuffle=True)
test_loader = DataLoader(test_set, batch_size=batch_size, shuffle=False)
# define a rnn model
class rnn(nn.Module):
def __init__(self, in_dim, hidden_dim, n_layer, n_class):
super(rnn, self).__init__()
self.lstm = nn.LSTM(in_dim, hidden_dim, n_layer, batch_first=True)
self.classifier = nn.Linear(hidden_dim, n_class)
import net
from visdom import Visdom
# (Hyper parameters)
batch_size = 64
learning_rate = 1e-2
num_epochs = 20
if __name__ == '__main__':
data_tf = transforms.Compose(
[transforms.ToTensor(),
transforms.Normalize([0.5], [0.5])])
train_dataset = datasets.MNIST(
root=r'C:\Users\beach\Desktop\Study\深度学习技术与应用\作业1\data',
train=True,
transform=data_tf,
download=True)
test_dataset = datasets.MNIST(
root=r'C:\Users\beach\Desktop\Study\深度学习技术与应用\作业1\data',
train=False,
transform=data_tf)
train_loader = DataLoader(train_dataset,
batch_size=batch_size,
shuffle=True) #batch批量梯度下降
test_loader = DataLoader(test_dataset,
batch_size=batch_size,
shuffle=False)
model = net.SimpleNet(28 * 28, 300, 100, 10)
x = self.pool(x)
x = x.reshape(x.shape[0], -1)
return x
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
learning_rate = 0.001
in_channels = 1
num_classes = 10
batch_size = 64
epochs = 5
# load dataset
train_dataset = datasets.MNIST(root='dataset/',
train=True,
transform=transforms.ToTensor(),
download=True)
train_loader = DataLoader(dataset=train_dataset,
shuffle=True,
batch_size=batch_size)
model = CNN(in_channels=in_channels, num_classes=num_classes)
model.to(device)
# Loss and Optimier function
criterion = nn.CrossEntropyLoss()
optimizer = optim.Adam(model.parameters(), lr=learning_rate, weight_decay=0.0)
writer = SummaryWriter(f'runs/MNIST/traingout_tensorboard')
step = 0
for epoch in epochs:
def main():
args = parse_args()
if args.name is None:
args.name = 'mnist_%s_%s_%dd' %(args.arch, args.metric, args.num_features)
if not os.path.exists('models/%s' %args.name):
os.makedirs('models/%s' %args.name)
print('Config -----')
for arg in vars(args):
print('%s: %s' %(arg, getattr(args, arg)))
print('------------')
with open('models/%s/args.txt' %args.name, 'w') as f:
for arg in vars(args):
print('%s: %s' %(arg, getattr(args, arg)), file=f)
joblib.dump(args, 'models/%s/args.pkl' %args.name)
criterion = nn.CrossEntropyLoss().cuda()
cudnn.benchmark = True
transform_train = transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.1307,), (0.3081,))
])
transform_test = transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.1307,), (0.3081,))
])
train_set = datasets.MNIST(
root='~/data',
train=True,
download=True,
transform=transform_train)
test_set = datasets.MNIST(
root='~/data',
train=False,
download=True,
transform=transform_train)
train_loader = torch.utils.data.DataLoader(
train_set,
batch_size=args.batch_size,
shuffle=True,
num_workers=8)
test_loader = torch.utils.data.DataLoader(
test_set,
batch_size=args.batch_size,
shuffle=False,
num_workers=8)
# create model
model = archs.__dict__[args.arch](args)
model = model.cuda()
if args.metric == 'adacos':
metric_fc = metrics.AdaCos(num_features=args.num_features, num_classes=10)
elif args.metric == 'arcface':
metric_fc = metrics.ArcFace(num_features=args.num_features, num_classes=10)
elif args.metric == 'sphereface':
metric_fc = metrics.SphereFace(num_features=args.num_features, num_classes=10)
elif args.metric == 'cosface':
metric_fc = metrics.CosFace(num_features=args.num_features, num_classes=10)
else:
metric_fc = nn.Linear(args.num_features, 10)
metric_fc = metric_fc.cuda()
optimizer = optim.SGD(filter(lambda p: p.requires_grad, model.parameters()), lr=args.lr,
momentum=args.momentum, weight_decay=args.weight_decay)
scheduler = lr_scheduler.CosineAnnealingLR(optimizer,
T_max=args.epochs, eta_min=args.min_lr)
log = pd.DataFrame(index=[], columns=[
'epoch', 'lr', 'loss', 'acc1', 'val_loss', 'val_acc1'
])
best_loss = float('inf')
for epoch in range(args.epochs):
print('Epoch [%d/%d]' %(epoch+1, args.epochs))
scheduler.step()
# train for one epoch
train_log = train(args, train_loader, model, metric_fc, criterion, optimizer)
# evaluate on validation set
val_log = validate(args, test_loader, model, metric_fc, criterion)
print('loss %.4f - acc1 %.4f - val_loss %.4f - val_acc %.4f'
%(train_log['loss'], train_log['acc1'], val_log['loss'], val_log['acc1']))
tmp = pd.Series([
epoch,
scheduler.get_lr()[0],
train_log['loss'],
train_log['acc1'],
val_log['loss'],
val_log['acc1'],
], index=['epoch', 'lr', 'loss', 'acc1', 'val_loss', 'val_acc1'])
log = log.append(tmp, ignore_index=True)
log.to_csv('models/%s/log.csv' %args.name, index=False)
if val_log['loss'] < best_loss:
torch.save(model.state_dict(), 'models/%s/model.pth' %args.name)
best_loss = val_log['loss']
print("=> saved best model")
import torch
import torchvision
from torchvision import datasets, transforms
from torch import nn
import torch.nn.functional as F
import time
import model
if __name__ == "__main__":
#读取训练集和测试集并转换成pytorch所需要的格式
transform = transforms.Compose(
[transforms.ToTensor(),
transforms.Normalize((0.5, ), (0.5, ))])
data_train = datasets.MNIST(root='./data/',
transform=transform,
train=True,
download=True)
data_test = datasets.MNIST(root='./data/',
transform=transform,
train=False)
data_loader_train = torch.utils.data.DataLoader(dataset=data_train,
batch_size=128,
shuffle=True)
data_loader_test = torch.utils.data.DataLoader(dataset=data_test,
batch_size=128,
shuffle=True)
#定义训练所使用的设备、模型、损失函数和优化器（梯度下降算法）
device = torch.device("cuda:0")
#model = LeNet().to(device)
model = model.Model().to(device)
img_side = 28
my_net = Net(img_side).double().cuda()
print('Loading model')
my_net.load_state_dict(torch.load(CKPT_PATH))
batch_size = 64
max_iter = 1000
save_every = 10
mu = 0.1307
sigma = 0.3801
origin = -1 * mu / sigma
lr = 1e-2
dl = torch.utils.data.DataLoader(datasets.MNIST('../data',
train=False,
download=False,
transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize(
(mu, ), (sigma, ))
])),
batch_size=batch_size,
shuffle=True)
def Energy1(out, target=1):
out_norm = torch.sqrt(out[:10]**2 + out[10:]**2)
return -out_norm[target] + 1
def Energy2(out, target=1):
out_norm = torch.sqrt(out[:10]**2 + out[10:]**2)
target_vector = np.zeros(10)
from torch.nn import functional as F
import matplotlib.pyplot as plt
from statistics import mean
# Hyper Parameters
input_size = 784
inter = 56
num_classes = 10
num_epochs = 25
batch_size = 100
learning_rate = 0.01
momentum = 0.5
# MNIST Dataset (Images and Labels)
train_dataset = dsets.MNIST(root='./files',
train=True,
transform=transforms.ToTensor(),
download=True)
test_dataset = dsets.MNIST(root='./files',
train=False,
transform=transforms.ToTensor(),
download=True)
# Dataset Loader (Input Pipline)
train_loader = torch.utils.data.DataLoader(dataset=train_dataset,
batch_size=batch_size,
shuffle=True)
test_loader = torch.utils.data.DataLoader(dataset=test_dataset,
batch_size=batch_size,
shuffle=False)
def train_supernet_mnist(nnet, training_settings, subnet=None, test_all=False):
# Training settings
seed = training_settings['seed']
batch_size = training_settings['batch_size']
test_batch_size = training_settings['test_batch_size']
epochs = training_settings['epochs']
lr = training_settings['learning_rate']
gamma = training_settings['gamma']
no_cuda = training_settings['no_cuda']
log_interval = training_settings['log_interval']
save_model = training_settings['save_model']
use_cuda = not no_cuda and torch.cuda.is_available()
torch.manual_seed(seed)
device = torch.device("cuda" if use_cuda else "cpu")
kwargs = {'batch_size': batch_size}
if use_cuda:
kwargs.update({
'num_workers': 1,
'pin_memory': True,
'shuffle': True
}, )
transform = transforms.Compose(
[transforms.ToTensor(),
transforms.Normalize((0.1307, ), (0.3081, ))])
dataset1 = datasets.MNIST('../data',
train=True,
download=True,
transform=transform)
dataset2 = datasets.MNIST('../data', train=False, transform=transform)
train_loader = torch.utils.data.DataLoader(dataset1, **kwargs)
test_loader = torch.utils.data.DataLoader(dataset2,
batch_size=test_batch_size)
model = nnet.to(device)
if subnet is not None:
model.set_subnet(subnet)
optimizer = optim.Adadelta(model.parameters(), lr=lr)
scheduler = StepLR(optimizer, step_size=1, gamma=gamma)
test_acc = []
if subnet is None:
print("\nTraining SuperNet\n")
else:
print("\nTraining subnet {}\n".format(subnet))
for epoch in range(1, epochs + 1):
train(model, device, train_loader, optimizer, epoch, log_interval,
F.nll_loss, subnet)
test_acc_ = []
if subnet is None or test_all:
for choice in [[0, 0], [1, 0], [0, 1], [1, 1]]:
model.set_subnet(choice)
test_acc_.append(test(model, device, test_loader, F.nll_loss))
test_acc.append(test_acc_)
else:
test_acc.append(test(model, device, test_loader, F.nll_loss))
scheduler.step()
if save_model:
torch.save(model.state_dict(), "mnist_supernet.pt")
return test_acc
def activation(x):
return 1 / (1 + torch.exp(-x))
#Defining Softmax
def softmax(x):
return torch.exp(x) / torch.sum(torch.exp(x), dim=1).reshape(-1, 1)
#Downloading MNIST data
transform = transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
])
trainset = datasets.MNIST('MNIST_data/',
download=True,
train=True,
transform=transform)
trainloader = torch.utils.data.DataLoader(trainset,
batch_size=64,
shuffle=True)
dataiter = iter(trainloader)
images, labels = dataiter.next()
print(type(images))
print(images.shape)
print(labels.shape)
#Showing one of the image
plt.imshow(images[1].numpy().squeeze(), cmap='Greys_r')
#Defining input, hidden and output units for network
number_input = images.view(images.shape[0], -1)
sample_batch_size = 25
obs = (1, 28, 28) if 'mnist' in args.dataset else (3, 32, 32)
input_channels = obs[0]
rescaling = lambda x: (x - .5) * 2.
rescaling_inv = lambda x: .5 * x + .5
kwargs = {'num_workers': 1, 'pin_memory': True, 'drop_last': True}
ds_transforms = transforms.Compose([transforms.ToTensor(), rescaling])
m_transforms = transforms.Compose(
[transforms.RandomCrop(size=(14, 30)),
transforms.ToTensor()])
#padding
if 'mnist' in args.dataset:
train_loader = torch.utils.data.DataLoader(datasets.MNIST(
args.data_dir, download=True, train=True, transform=ds_transforms),
batch_size=args.batch_size,
shuffle=True,
**kwargs)
test_loader = torch.utils.data.DataLoader(datasets.MNIST(
args.data_dir, train=False, transform=ds_transforms),
batch_size=args.batch_size,
shuffle=True,
**kwargs)
loss_op = lambda real, fake: discretized_mix_logistic_loss_1d(real, fake)
sample_op = lambda x: sample_from_discretized_mix_logistic_1d(
x, args.nr_logistic_mix)
elif 'mario' in args.dataset:
train_loader = torch.utils.data.DataLoader(MyDataset(
def forward(self, x):
return self.gen(x)
lr = 3e-4
z_dim = 64
img_dim = 28 * 28
epochs = 25
disc = Discriminator(img_dim).cuda()
gen = Generator(z_dim, img_dim).cuda()
noise = torch.randn((32, z_dim)).cuda()
transforms = transforms.Compose(
[transforms.ToTensor(),
transforms.Normalize((0.5), (0.5))])
dataset = datasets.MNIST("data/", transform=transforms, download=True)
loader = DataLoader(dataset, batch_size=32, shuffle=True)
opt_disc = optim.Adam(disc.parameters(), lr=lr)
opt_gen = optim.Adam(gen.parameters(), lr=lr)
crit = nn.BCELoss()
writer_fake = SummaryWriter(f"runs/Gan/fake")
writer_real = SummaryWriter(f"runs/Gan/real")
step = 0
for epoch in range(epochs):
for batch_idx, (imges, _) in enumerate(loader):
real = imges.view(-1, 784).cuda()
batch_size = real.shape[0]
#train discr
from matplotlib import pyplot as plt
batch_size = 32
n_epoch = 25
latent_dim = 100
n_classes = 10
img_shape = 28 * 28
lr = .0001
dataloader = torch.utils.data.DataLoader(
datasets.MNIST(
"../../data/mnist",
train=True,
download=True,
transform=transforms.Compose([
transforms.Resize(28),
transforms.ToTensor(),
transforms.Normalize([0.5], [0.5])
]),
),
batch_size=batch_size,
shuffle=True,
)
feat, label = next(iter(dataloader))
class Discriminator(nn.Module):
def __init__(self, n_classes, img_shape):
super(Discriminator, self).__init__()
type=int,
default=100,
help="latent dimension")
parser.add_argument("--out_layer",
type=int,
default=256,
help="number of out layer")
args = parser.parse_args()
print("device", device)
print(args)
os.makedirs('generated-images', exist_ok=True)
# setting data
train_loader = torch.utils.data.DataLoader(datasets.MNIST(
'../../data', train=True, download=True, transform=transforms.ToTensor()),
batch_size=args.batch_size,
shuffle=True)
test_loader = torch.utils.data.DataLoader(datasets.MNIST(
'../../data', train=False, transform=transforms.ToTensor()),
shuffle=True)
class Generator(nn.Module):
def __init__(self, out_layer):
super(Generator, self).__init__()
self.layer1 = nn.Linear(args.latent_vector + 10, out_layer)
self.layer2 = nn.Linear(out_layer, out_layer * 2)
self.layer3 = nn.Linear(out_layer * 2, out_layer * 4)
self.layer4 = nn.Linear(out_layer * 4, 784)
help='how many batches to wait before logging training status')
parser.add_argument('--save-model', action='store_true', default=False,
help='For Saving the current Model')
args = parser.parse_args()
use_cuda = not args.no_cuda and torch.cuda.is_available()
torch.manual_seed(args.seed)
device = torch.device("cuda" if use_cuda else "cpu")
kwargs = {'num_workers': 1, 'pin_memory': True} if use_cuda else {}
train_loader = torch.utils.data.DataLoader(
datasets.MNIST('../data', train=True, download=True,
transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.1307,), (0.3081,))
])),
batch_size=args.batch_size, shuffle=True, **kwargs)
test_loader = torch.utils.data.DataLoader(
datasets.MNIST('../data', train=False, transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.1307,), (0.3081,))
])),
batch_size=args.test_batch_size, shuffle=True, **kwargs)
dt = [ex for ex in test_loader][:1]
dt = [[ex[0][:n_test], ex[1][:n_test]] for ex in dt]
for s in range(n_samples):
print (s, n_samples)
model = Net()
model.eval()
#未归一化，正确率不高
#基于MNIST数据集
import torch
from torch.utils.data import DataLoader
import torchvision.datasets as dsets
import torchvision.transforms as transforms
import matplotlib.pyplot as plt
import numpy as np
import KNN_classify_algorithm as KNN
batch_size = 100
#MNIST dataset
train_dataset = dsets.MNIST(root='/dataset/MNIST',
train=True,
transform=None,
download=True)
test_dataset = dsets.MNIST(root='/dataset/MNIST',
train=False,
transform=None,
download=True)
train_loader = DataLoader(dataset=train_dataset,
batch_size=batch_size,
shuffle=True)
test_loader = DataLoader(dataset=test_dataset,
batch_size=batch_size,
shuffle=True)
'''
parser.add_argument('--no-cuda', action='store_true', default=False,
help='disables CUDA training')
parser.add_argument('--seed', type=int, default=1, metavar='S',
help='random seed (default: 1)')
parser.add_argument('--log-interval', type=int, default=10, metavar='N',
help='how many batches to wait before logging training status')
args = parser.parse_args()
args.cuda = not args.no_cuda and torch.cuda.is_available()
torch.manual_seed(args.seed)
device = torch.device("cuda" if args.cuda else "cpu")
kwargs = {'num_workers': 1, 'pin_memory': True} if args.cuda else {}
train_loader = torch.utils.data.DataLoader(
datasets.MNIST('../data', train=True, download=True,
transform=transforms.ToTensor()),
batch_size=args.batch_size, shuffle=True, **kwargs)
test_loader = torch.utils.data.DataLoader(
datasets.MNIST('../data', train=False, transform=transforms.ToTensor()),
batch_size=args.batch_size, shuffle=True, **kwargs)
class VAE(nn.Module):
def __init__(self):
super(VAE, self).__init__()
self.fc1 = nn.Linear(784, 400)
self.fc21 = nn.Linear(400, 20)
self.fc22 = nn.Linear(400, 20)
self.fc3 = nn.Linear(20, 400)
self.fc4 = nn.Linear(400, 784)
type=str,
help='distributed backend')
args = parser.parse_args()
args.cuda = not args.no_cuda and torch.cuda.is_available()
# initialize distributed group
dist.init_process_group(backend=args.dist_backend,
init_method=args.dist_url,
world_size=args.world_size)
kwargs = {'num_workers': 1, 'pin_memory': True} if args.cuda else {}
train_loader = torch.utils.data.DataLoader(datasets.MNIST(
'../data',
train=True,
download=True,
transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.1307, ), (0.3081, ))
])),
batch_size=args.batch_size,
shuffle=True,
**kwargs)
test_loader = torch.utils.data.DataLoader(datasets.MNIST(
'../data',
train=False,
transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.1307, ), (0.3081, ))
])),
batch_size=args.batch_size,
shuffle=True,
def main():
# Training settings
parser = argparse.ArgumentParser(description='PyTorch MNIST Example')
parser.add_argument('--batch-size',
type=int,
default=64,
metavar='N',
help='input batch size for training (default: 64)')
parser.add_argument('--test-batch-size',
type=int,
default=1,
metavar='N',
help='input batch size for testing (default: 1000)')
parser.add_argument('--epochs',
type=int,
default=25,
metavar='N',
help='number of epochs to train (default: 14)')
parser.add_argument('--lr',
type=float,
default=1.0,
metavar='LR',
help='learning rate (default: 1.0)')
parser.add_argument('--gamma',
type=float,
default=0.7,
metavar='M',
help='Learning rate step gamma (default: 0.7)')
parser.add_argument('--no-cuda',
action='store_true',
default=False,
help='disables CUDA training')
parser.add_argument('--seed',
type=int,
default=1,
metavar='S',
help='random seed (default: 1)')
parser.add_argument(
'--log-interval',
type=int,
default=10,
metavar='N',
help='how many batches to wait before logging training status')
parser.add_argument('--save-model',
action='store_true',
default=True,
help='For Saving the current Model')
args = parser.parse_args()
use_cuda = not args.no_cuda and torch.cuda.is_available()
torch.manual_seed(args.seed)
device = torch.device("cuda" if use_cuda else "cpu")
kwargs = {'num_workers': 1, 'pin_memory': True} if use_cuda else {}
train_loader = torch.utils.data.DataLoader(datasets.MNIST(
'../data',
train=True,
download=True,
transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.1307, ), (0.3081, ))
])),
batch_size=args.batch_size,
shuffle=True,
**kwargs)
test_loader = torch.utils.data.DataLoader(datasets.MNIST(
'../data',
train=False,
transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.1307, ), (0.3081, ))
])),
batch_size=args.test_batch_size,
shuffle=True,
**kwargs)
model = Net().to(device)
optimizer = optim.Adadelta(model.parameters(), lr=args.lr)
scheduler = StepLR(optimizer, step_size=1, gamma=args.gamma)
for epoch in range(1, args.epochs + 1):
train(args, model, device, train_loader, optimizer, epoch)
scheduler.step()
test(model, device, test_loader)
if args.save_model:
torch.save(model, "mnist_cnn.pth")
c.apply(init_weights)
load_model(g, None, load_g_path)
g.eval()
for param in g.parameters():
param.requires_grad = False
# Configure data loader
transform = transforms.Compose([transforms.Resize(img_size), transforms.ToTensor(), transforms.Normalize([0.5], [0.5])])
os.makedirs("../data/mnist", exist_ok=True)
dataloader = torch.utils.data.DataLoader(
datasets.MNIST(
"../data/mnist",
train=True,
download=True,
transform=transform,
),
batch_size=batch_size,
sampler=torch.utils.data.SubsetRandomSampler(list(range(training_set_size)))
)
test_dataloader = torch.utils.data.DataLoader(
datasets.MNIST(
"../data/mnist",
train=False,
download=True,
transform=transform,
),
batch_size=batch_size,
shuffle=True
)
def run():
parser = argparse.ArgumentParser(description='PyTorch MNIST Example')
parser.add_argument('--batch-size',
type=int,
default=64,
metavar='N',
help='input batch size for training (default: 64)')
parser.add_argument('--test-batch-size',
type=int,
default=1,
metavar='N',
help='input batch size for testing (default: 1000)')
parser.add_argument('--epochs',
type=int,
default=25,
metavar='N',
help='number of epochs to train (default: 14)')
parser.add_argument('--lr',
type=float,
default=1.0,
metavar='LR',
help='learning rate (default: 1.0)')
parser.add_argument('--gamma',
type=float,
default=0.7,
metavar='M',
help='Learning rate step gamma (default: 0.7)')
parser.add_argument('--no-cuda',
action='store_true',
default=False,
help='disables CUDA training')
parser.add_argument('--seed',
type=int,
default=1,
metavar='S',
help='random seed (default: 1)')
parser.add_argument(
'--log-interval',
type=int,
default=10,
metavar='N',
help='how many batches to wait before logging training status')
parser.add_argument('--save-model',
action='store_true',
default=True,
help='For Saving the current Model')
args = parser.parse_args()
use_cuda = not args.no_cuda and torch.cuda.is_available()
device = torch.device("cuda" if use_cuda else "cpu")
kwargs = {'num_workers': 1, 'pin_memory': True} if use_cuda else {}
test_loader = torch.utils.data.DataLoader(datasets.MNIST(
'../data',
download=True,
train=False,
transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.1307, ), (0.3081, ))
])),
batch_size=args.test_batch_size,
shuffle=True,
**kwargs)
model = torch.load("mnist_cnn.pth",
map_location=lambda storage, loc: storage)
model.to(device)
test(model, device, test_loader)
transforms.CenterCrop(args.img_size),
transforms.ToTensor(),
transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5)),
]))
# Create the dataloader
dataloader = torch.utils.data.DataLoader(dataset,
batch_size=args.batch_size,
shuffle=True,
num_workers=args.num_workers)
elif args.data == 'mnist':
dataloader = torch.utils.data.DataLoader(dset.MNIST(
'./data/mnist',
train=True,
download=True,
transform=transforms.Compose([
transforms.Grayscale(num_output_channels=1),
transforms.Resize(args.img_size),
transforms.ToTensor(),
transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
])),
batch_size=args.batch_size,
shuffle=True,
num_workers=args.num_workers)
elif args.data == 'cifar10':
dataloader = torch.utils.data.DataLoader(dset.CIFAR10(
'./data/cifar10',
train=True,
download=True,
transform=transforms.Compose([
transforms.Grayscale(num_output_channels=1),
if __name__ == "__main__":
# get arguments
args = parse_args()
# init and load model
classifier = Classifier(args)
classifier.load_state_dict(
torch.load('../pretrained_model/classifier_mnist.pt'))
classifier.eval()
classifier = classifier.cuda()
# init dataset
transform = transforms.Compose(
[transforms.CenterCrop(args.image_size),
transforms.ToTensor()])
dataset = datasets.MNIST('../data',
train=True,
download=True,
transform=transform)
dataloader = torch.utils.data.DataLoader(dataset,
batch_size=512,
shuffle=True,
num_workers=1)
# adversarial methods
adv_list = ['fgsm', 'r-fgsm', 'cw']
# test for accuracy
xs = list()
ys = list()
advs = list()
for image, label in dataloader:
image = image.cuda()
label = label.cuda()
batch += 1
if cuda:
generator.cuda()
discriminator.cuda()
adversarial_loss.cuda()
# Initialize weights
generator.apply(weights_init_normal)
discriminator.apply(weights_init_normal)
# Configure data loader
os.makedirs('../../data/mnist', exist_ok=True)
dataloader = torch.utils.data.DataLoader(
datasets.MNIST('../../data/mnist', train=True, download=True,
transform=transforms.Compose([
transforms.Resize(opt.img_size),
transforms.ToTensor(),
transforms.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5))
])),
batch_size=opt.batch_size, shuffle=True)
# Optimizers
optimizer_G = torch.optim.Adam(generator.parameters(), lr=opt.lr, betas=(opt.b1, opt.b2))
optimizer_D = torch.optim.Adam(discriminator.parameters(), lr=opt.lr, betas=(opt.b1, opt.b2))
Tensor = torch.cuda.FloatTensor if cuda else torch.FloatTensor
# ----------
# Training
# ----------
for epoch in range(opt.n_epochs):
elif opt.dataset == 'cifar10':
dataset = dset.CIFAR10(root=opt.dataroot,
download=True,
transform=transforms.Compose([
transforms.Resize(opt.imageSize),
transforms.ToTensor(),
transforms.Normalize((0.5, 0.5, 0.5),
(0.5, 0.5, 0.5)),
]))
nc = 3
elif opt.dataset == 'mnist':
dataset = dset.MNIST(root=opt.dataroot,
download=True,
transform=transforms.Compose([
transforms.Resize(opt.imageSize),
transforms.ToTensor(),
transforms.Normalize((0.5, ), (0.5, )),
]))
nc = 1
elif opt.dataset == 'fake':
dataset = dset.FakeData(image_size=(3, opt.imageSize, opt.imageSize),
transform=transforms.ToTensor())
nc = 3
assert dataset
dataloader = torch.utils.data.DataLoader(dataset,
batch_size=opt.batch_size,
shuffle=True,
num_workers=int(opt.workers))
def main():
global args
# Training settings
parser = argparse.ArgumentParser(description='PyTorch MNIST Example')
parser.add_argument('--batch-size',
type=int,
default=64,
metavar='N',
help='input batch size for training (default: 64)')
parser.add_argument('--test-batch-size',
type=int,
default=1000,
metavar='N',
help='input batch size for testing (default: 1000)')
parser.add_argument('--epochs',
type=int,
default=14,
metavar='N',
help='number of epochs to train (default: 14)')
parser.add_argument('--lr',
type=float,
default=0.01,
metavar='LR',
help='learning rate (default: 0.1)')
parser.add_argument('--momentum',
type=float,
default=0.9,
metavar='M',
help='Momentum (default: 0.9)')
parser.add_argument('--wd',
type=float,
default=0.0001,
metavar='M',
help='Weight decay (default: 0.0005)')
parser.add_argument('--no-cuda',
action='store_true',
default=False,
help='disables CUDA training')
parser.add_argument('--seed',
type=int,
default=1,
metavar='S',
help='random seed (default: 1)')
parser.add_argument(
'--log-interval',
type=int,
default=10,
metavar='N',
help='how many batches to wait before logging training status')
parser.add_argument('--save-model',
action='store_true',
default=False,
help='For Saving the current Model')
parser.add_argument('--data',
type=str,
default='../data',
help='Location to store data')
parser.add_argument('--sparsity',
type=float,
default=0.1,
help='how sparse is each layer')
args = parser.parse_args()
use_cuda = not args.no_cuda and torch.cuda.is_available()
torch.manual_seed(args.seed)
device = torch.device("cuda" if use_cuda else "cpu")
kwargs = {'num_workers': 1, 'pin_memory': True} if use_cuda else {}
train_loader = torch.utils.data.DataLoader(datasets.MNIST(
os.path.join(args.data, 'mnist'),
train=True,
download=True,
transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.1307, ), (0.3081, ))
])),
batch_size=args.batch_size,
shuffle=True,
**kwargs)
test_loader = torch.utils.data.DataLoader(datasets.MNIST(
os.path.join(args.data, 'mnist'),
train=False,
transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.1307, ), (0.3081, ))
])),
batch_size=args.test_batch_size,
shuffle=True,
**kwargs)
model = Net().to(device)
# NOTE: only pass the parameters where p.requires_grad == True to the optimizer! Important!
optimizer = optim.SGD(
[p for p in model.parameters() if p.requires_grad],
lr=args.lr,
momentum=args.momentum,
weight_decay=args.wd,
)
criterion = nn.CrossEntropyLoss().to(device)
scheduler = CosineAnnealingLR(optimizer, T_max=args.epochs)
for epoch in range(1, args.epochs + 1):
train(model, device, train_loader, optimizer, criterion, epoch)
test(model, device, criterion, test_loader)
scheduler.step()
if args.save_model:
torch.save(model.state_dict(), "mnist_cnn.pt")
# At this point the data come in Python tuples, a 28x28 image and a label.
# while the label is a tensor, the image is not; it needs to be converted.
# So we need to transform PIL image to tensor and then normalize it.
# Normalization is quite a good practise to avoid numerical and convergence
# problems. For that we need the dataset's mean and std which fortunately
# can be computed!
# ******************
mean = 0.1307
std = 0.3081
# Bundle our transforms sequentially, one after another. This is important.
# Convert images to tensors + normalize
transform = tTrans.Compose(
[tTrans.ToTensor(), tTrans.Normalize((mean, ), (std, ))])
# Load data set
mnistTrainset = tdata.MNIST(root='../data',
train=True,
download=True,
transform=transform)
mnistTestset = tdata.MNIST(root='../data',
train=False,
download=True,
transform=transform)
# Once we have a dataset, torch.utils has a very nice lirary for iterating on that
# dataset, wit hshuffle AND batch logic. Very usefull in larger datasets.
trainLoader = torch.utils.data.DataLoader(mnistTrainset,
batch_size=batch,
**comArgs)
testLoader = torch.utils.data.DataLoader(mnistTestset,
batch_size=10 * batch,
**comArgs)
print(opt)
# the size of image
img_shape = (opt.channels, opt.img_size, opt.img_size)
print(*img_shape)
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
print(device)
# Configure data loader
os.makedirs("../../data/mnist", exist_ok=True)
dataloader = DataLoader(
dataset=datasets.MNIST(
root="../../data/mnist",
train=True,
download=True,
transform=transforms.Compose(
[transforms.Resize(opt.img_size), transforms.ToTensor(), transforms.Normalize([0.5], [0.5])]
)
),
batch_size=opt.batch_size,
shuffle=True
)
class Generator(nn.Module):
def __init__(self):
super(Generator, self).__init__()
self.label_embedding = nn.Embedding(opt.n_classes, opt.n_classes)
def generator_block(in_channels, out_channels, bn=True):
def download_test_data(self):
return datasets.MNIST('./data',
train=False,
download=True,
transform=test_transforms)