本文整理汇总了Python中Network.train方法的典型用法代码示例。如果您正苦于以下问题：Python Network.train方法的具体用法？Python Network.train怎么用？Python Network.train使用的例子？那么恭喜您, 这里精选的方法代码示例或许可以为您提供帮助。您也可以进一步了解该方法所在类Network的用法示例。
在下文中一共展示了Network.train方法的2个代码示例，这些例子默认根据受欢迎程度排序。您可以为喜欢或者感觉有用的代码点赞，您的评价将有助于系统推荐出更棒的Python代码示例。
示例1: Network
# 需要导入模块: import Network [as 别名]
# 或者: from Network import train [as 别名]
mnist = data.load("mnist", flatvecs = True)
test_data_xs = mnist["test"]["xs"]
train_data = mnist["train"]
net = Network(
layers = [InputLayer([784]),
FullyConnectedLayer([20], activation = sigmoid),
FullyConnectedLayer([784], activation = sigmoid)],
cost=mse)
for i in range(30):
for m in range(150):
batch_xs, _ = train_data.get_batch(30)
net.train(batch_xs, batch_xs, learning_rate = 0.01 )
print i, "Test Cost:", net.test(test_data_xs, test_data_xs )
# Let's look at how it reproduces a few examples...
examples = net(test_data_xs[0:3])
examples = examples.reshape(-1, 28, 28)
vis.vis_matrix(examples, s2x2 = True)
print "----"
# Example Results:
#
# 0 Test Cost: 0.058131233367
# | |
# | |
# ░░░░ | ░░░░░░░░░ | ░░░░░░░
# ░░░░░░░░░░░░ | ░░░░▒▒▒▒▒▒░░░ | ░░░░░▒▒░░░░░ 示例2: Network
# 需要导入模块: import Network [as 别名]
# 或者: from Network import train [as 别名]
mnist = data.load("mnist")
test_data = mnist["test"].values()
train_data = mnist["train"]
net = Network(
layers = [InputLayer([28,28]),
ConvolutionLayer([6,6], 8, activation = ReLU),
PoolLayer([2,2]),
FullyConnectedLayer(100, activation = ReLU),
SoftMaxLayer(10)],
cost=log_likelihood)
for m in range(100):
net.train(*train_data.get_batch(30), learning_rate = 0.01 )
xs, ys = train_data.get_batch(5)
info = net.introspect(xs, ys)
info.keys()
for i in range(20):
for m in range(1000):
net.train(*train_data.get_batch(30), learning_rate = 0.01 )
print "Training Step", i
print "Test Accuracy:", net.accuracy(*test_data)
print "Test Cost:", net.test(*test_data)
print ""