2020-10-19 21:08:55

docs/da/124.md

@@ -8,7 +8,7 @@
 
 In [1]:
 
-```
+```py
 %%file download_mnist.py
 import os
 import os.path
@@ -33,7 +33,7 @@ download_and_gzip('mnist/t10k-labels-idx1-ubyte')
 
 ```
 
-```
+```py
 Overwriting download_mnist.py
 
 ```
@@ -42,7 +42,7 @@ Overwriting download_mnist.py
 
 In [2]:
 
-```
+```py
 %run download_mnist.py
 
 ```
@@ -53,7 +53,7 @@ In [2]:
 
 In [3]:
 
-```
+```py
 %%file load.py
 import numpy as np
 import os
@@ -106,7 +106,7 @@ def mnist(ntrain=60000,ntest=10000,onehot=True):
 
 ```
 
-```
+```py
 Overwriting load.py
 
 ```
@@ -123,7 +123,7 @@ Overwriting load.py
 
 In [4]:
 
-```
+```py
 import theano
 from theano import tensor as T
 import numpy as np
@@ -131,7 +131,7 @@ from load import mnist
 
 ```
 
-```
+```py
 Using gpu device 1: Tesla C2075 (CNMeM is disabled)
 
 ```
@@ -142,7 +142,7 @@ Using gpu device 1: Tesla C2075 (CNMeM is disabled)
 
 In [5]:
 
-```
+```py
 def floatX(X):
     return np.asarray(X, dtype=theano.config.floatX)
 
@@ -155,7 +155,7 @@ def init_weights(shape):
 
 In [6]:
 
-```
+```py
 A = T.matrix()
 
 B = T.nnet.softmax(A)
@@ -173,7 +173,7 @@ print b.sum(1)
 
 ```
 
-```
+```py
 (3, 4)
 [ 1.00000012  1\.          1\.        ]
 
@@ -185,7 +185,7 @@ print b.sum(1)
 
 In [7]:
 
-```
+```py
 def model(X, w):
     return T.nnet.softmax(T.dot(X, w))
 
@@ -195,7 +195,7 @@ def model(X, w):
 
 In [8]:
 
-```
+```py
 trX, teX, trY, teY = mnist(onehot=True)
 
 ```
@@ -204,7 +204,7 @@ trX, teX, trY, teY = mnist(onehot=True)
 
 In [9]:
 
-```
+```py
 X = T.fmatrix()
 Y = T.fmatrix()
 
@@ -216,7 +216,7 @@ w = init_weights((784, 10))
 
 In [10]:
 
-```
+```py
 py_x = model(X, w)
 y_pred = T.argmax(py_x, axis=1)
 
@@ -226,7 +226,7 @@ y_pred = T.argmax(py_x, axis=1)
 
 In [11]:
 
-```
+```py
 cost = T.mean(T.nnet.categorical_crossentropy(py_x, Y))
 gradient = T.grad(cost=cost, wrt=w)
 update = [[w, w - gradient * 0.05]]
@@ -237,7 +237,7 @@ update = [[w, w - gradient * 0.05]]
 
 In [12]:
 
-```
+```py
 train = theano.function(inputs=[X, Y], outputs=cost, updates=update, allow_input_downcast=True)
 predict = theano.function(inputs=[X], outputs=y_pred, allow_input_downcast=True)
 
@@ -247,7 +247,7 @@ predict = theano.function(inputs=[X], outputs=y_pred, allow_input_downcast=True)
 
 In [13]:
 
-```
+```py
 for i in range(100):
     for start, end in zip(range(0, len(trX), 128), range(128, len(trX), 128)):
         cost = train(trX[start:end], trY[start:end])
@@ -255,7 +255,7 @@ for i in range(100):
 
 ```
 
-```
+```py
 000 0.8862
 001 0.8985
 002 0.9042