2020-10-19 21:08:55

docs/da/127.md

@@ -2,7 +2,7 @@
 
 In [1]:
 
-```
+```py
 import theano
 import theano.tensor as T
 import numpy as np
@@ -16,7 +16,7 @@ def floatX(X):
 
 ```
 
-```
+```py
 Using gpu device 1: Tesla C2075 (CNMeM is disabled)
 
 ```
@@ -27,7 +27,7 @@ Using gpu device 1: Tesla C2075 (CNMeM is disabled)
 
 In [2]:
 
-```
+```py
 def dropout(X, prob=0.):
     if prob > 0:
         X *= srng.binomial(X.shape, p=1-prob, dtype = theano.config.floatX)
@@ -50,7 +50,7 @@ $$ \begin{aligned} & h_1 = \text{rectify}(W_{h_1} \ x) \\ & h_2 = \text{rectify}
 
 In [3]:
 
-```
+```py
 def softmax(X):
     e_x = T.exp(X - X.max(axis=1).dimshuffle(0, 'x'))
     return e_x / e_x.sum(axis=1).dimshuffle(0, 'x')
@@ -85,7 +85,7 @@ def model(X, w_h1, w_h2, w_o, p_drop_input, p_drop_hidden):
 
 In [4]:
 
-```
+```py
 def init_weights(shape):
     return theano.shared(floatX(np.random.randn(*shape) * 0.01))
 
@@ -99,7 +99,7 @@ w_o = init_weights((625, 10))
 
 In [5]:
 
-```
+```py
 X = T.matrix()
 Y = T.matrix()
 
@@ -109,7 +109,7 @@ Y = T.matrix()
 
 In [6]:
 
-```
+```py
 def RMSprop(cost, params, accs, lr=0.001, rho=0.9, epsilon=1e-6):
     grads = T.grad(cost=cost, wrt=params)
     updates = []
@@ -127,7 +127,7 @@ def RMSprop(cost, params, accs, lr=0.001, rho=0.9, epsilon=1e-6):
 
 In [7]:
 
-```
+```py
 # 有 dropout，用来训练
 noise_h1, noise_h2, noise_py_x = model(X, w_h1, w_h2, w_o, 0.2, 0.5)
 cost = T.mean(T.nnet.categorical_crossentropy(noise_py_x, Y))
@@ -143,7 +143,7 @@ train = theano.function(inputs=[X, Y], outputs=cost, updates=updates, allow_inpu
 
 In [8]:
 
-```
+```py
 # 没有 dropout，用来预测
 h1, h2, py_x = model(X, w_h1, w_h2, w_o, 0., 0.)
 # 预测的结果
@@ -156,7 +156,7 @@ predict = theano.function(inputs=[X], outputs=y_x, allow_input_downcast=True)
 
 In [9]:
 
-```
+```py
 trX, teX, trY, teY = mnist(onehot=True)
 
 for i in range(50):
@@ -166,7 +166,7 @@ for i in range(50):
 
 ```
 
-```
+```py
 iter 001 accuracy: 0.943
 iter 002 accuracy: 0.9665
 iter 003 accuracy: 0.9732