diff --git a/4.人工智能/4.2机器学习（AI）快速入门（quick start）.md b/4.人工智能/4.2机器学习（AI）快速入门（quick start）.md
index ea935c9..5376fd1 100644
--- a/4.人工智能/4.2机器学习（AI）快速入门（quick start）.md	
+++ b/4.人工智能/4.2机器学习（AI）快速入门（quick start）.md	
@@ -84,19 +84,19 @@
 
 ```python
 def estimate_house_sales_price(num_of_bedrooms, sqft, neighborhood):
-  price = 0<em>  # In my area, the average house costs $200 per sqft</em>
+  price = 0  # In my area, the average house costs $200 per sqft
   price_per_sqft = 200  if neighborhood == "hipsterton":
-   <em> # but some areas cost a bit more</em>
+    # but some areas cost a bit more
     price_per_sqft = 400  elif neighborhood == "skid row":
-   <em> # and some areas cost less</em>
-    price_per_sqft = 100  <em># start with a base price estimate based on how big the place is</em>
-  price = price_per_sqft * sqft  <em># now adjust our estimate based on the number of bedrooms</em>
+    # and some areas cost less
+    price_per_sqft = 100  # start with a base price estimate based on how big the place is
+  price = price_per_sqft * sqft  # now adjust our estimate based on the number of bedrooms
   if num_of_bedrooms == 0:
-    <em># Studio apartments are cheap</em>
+    # Studio apartments are cheap
     price = price — 20000
   else:
-   <em> # places with more bedrooms are usually</em>
-<em>    # more valuable</em>
+    # places with more bedrooms are usually
+    # more valuable
     price = price + (num_of_bedrooms * 1000) return price
 ```
 
@@ -212,9 +212,9 @@ def estimate_house_sales_price(num_of_bedrooms, sqft, neighborhood):
 ```python
 def estimate_house_sales_price(num_of_bedrooms, sqft, neighborhood):
  price = 0# a little pinch of this
- price += num_of_bedrooms * <strong>0.123</strong># and a big pinch of that
- price += sqft * <strong>0.41</strong># maybe a handful of this
- price += neighborhood * <strong>0.57</strong>
+ price += num_of_bedrooms * 0.123# and a big pinch of that
+ price += sqft * 0.41# maybe a handful of this
+ price += neighborhood * 0.57
  return price
 ```
 
@@ -328,10 +328,10 @@ print('y_pred=',y_test.data)
 请注意，我们的神经网络现在有了两个输出（而不仅仅是一个房子的价格）。第一个输出会预测图片是「8」的概率，而第二个则输出不是「8」的概率。概括地说，我们就可以依靠多种不同的输出，利用神经网络把要识别的物品进行分组。
 
 ```python
-<strong>model = Sequential([Dense(32, input_shape=(784,)), </strong>
-<strong>  Activation('relu'),Dense(10),Activation('softmax')])</strong># 你也可以通过 .add() 方法简单地添加层：<strong>  model = Sequential() </strong>
-<strong>  model.add(Dense(32, input_dim=784)) </strong>
-<strong>  model.add(Activation('relu'))# 激活函数，你可以理解为加上这个东西可以让他效果更好</strong>
+model = Sequential([Dense(32, input_shape=(784,)), 
+  Activation('relu'),Dense(10),Activation('softmax')])# 你也可以通过 .add() 方法简单地添加层：  model = Sequential() 
+  model.add(Dense(32, input_dim=784)) 
+  model.add(Activation('relu'))# 激活函数，你可以理解为加上这个东西可以让他效果更好
 ```
 
 虽然我们的神经网络要比上次大得多（这次有 324 个输入，上次只有 3 个！），但是现在的计算机一眨眼的功夫就能够对这几百个节点进行运算。当然，你的手机也可以做到。