nn.Conv2D(channels=384, kernel_size=3,
padding=1, activation='relu'),
nn.Conv2D(channels=384, kernel_size=3,
padding=1, activation='relu'),
nn.MaxPool2D(pool_size=3, strides=2),
nn.Flatten(),
nn.Dense(4096, activation='relu'),
nn.Dropout(.5),
nn.Dense(4096, activation='relu'),
nn.Dropout(.5),
nn.Dense(10)
)
return net
batch_size = 64
train_iter, test_iter = utils.loadMnistData(batch_size, resize=224)
# train_iter, test_iter = utils.load_data_fashion_mnist(batch_size, resize=224)
# for data, label in train_iter:
# print(data.shape)
# break
net = get_net()
ctx = utils.getCtx()
net.initialize(ctx=ctx, init=init.Xavier())
softmax_loss = loss.SoftmaxCrossEntropyLoss()
epochs = 5
trainer = Trainer(net.collect_params(), 'sgd', {'learning_rate': 0.5})
from mxnet import gluon
from mxnet import autograd
from mxnet import ndarray as nd
from utils import loadMnistData, accuracy, evaluate_accuracy
net = gluon.nn.Sequential()
with net.name_scope():
net.add(gluon.nn.Flatten())
net.add(gluon.nn.Dense(256, activation="relu"))
net.add(gluon.nn.Dense(10))
net.initialize()
batch_size = 256
train_data, test_data = loadMnistData(batch_size)
softmax_cross_entropy = gluon.loss.SoftmaxCrossEntropyLoss()
trainer = gluon.Trainer(net.collect_params(), 'sgd', {'learning_rate': 0.5})
for epoch in range(5):
train_loss = 0.
train_acc = 0.
for data, label in train_data:
with autograd.record():
output = net(data)
loss = softmax_cross_entropy(output, label)
loss.backward()
trainer.step(batch_size)
train_loss += nd.mean(loss).asscalar()
from mxnet import autograd
from utils import accuracy, evaluate_accuracy, loadMnistData
batch_size = 256
num_hidden = 256
net = gluon.nn.Sequential()
with net.name_scope():
net.add(gluon.nn.Flatten())
net.add(gluon.nn.Dense(256, activation="relu"))
net.add(gluon.nn.Dense(10))
net.initialize()
train_iter, test_iter = loadMnistData(batch_size)
trainer = gluon.Trainer(net.collect_params(), 'sgd', {'learning_rate': 0.5})
softmax_cross_entropy = gluon.loss.SoftmaxCrossEntropyLoss()
epochs = 5
for epoch in range(epochs):
total_loss = .0
total_acc = .0
for data, label in train_iter:
with autograd.record():
output = net(data)
loss = softmax_cross_entropy(output, label)
loss.backward()
trainer.step(batch_size)