补充了模型的描述

Author: ChibisukeDragon

chapter3/3.2-mnist.ipynb

@@ -133,23 +133,25 @@
     "class ConvNet(nn.Module):\n",
     "    def __init__(self):\n",
     "        super().__init__()\n",
-    "        # 1,28x28\n",
-    "        self.conv1=nn.Conv2d(1,10,5) # 10, 24x24\n",
-    "        self.conv2=nn.Conv2d(10,20,3) # 128, 10x10\n",
-    "        self.fc1 = nn.Linear(20*10*10,500)\n",
-    "        self.fc2 = nn.Linear(500,10)\n",
+    "        # batch*1*28*28（每次会送入batch个样本，输入通道数1（黑白图像），图像分辨率是28x28）\n",
+    "        # 下面的卷积层Conv2d的第一个参数指输入通道数，第二个参数指输出通道数，第三个参数指卷积核的大小\n",
+    "        self.conv1 = nn.Conv2d(1, 10, 5) # 输入通道数1，输出通道数10，核的大小5\n",
+    "        self.conv2 = nn.Conv2d(10, 20, 3) # 输入通道数10，输出通道数20，核的大小3\n",
+    "        # 下面的全连接层Linear的第一个参数指输入通道数，第二个参数指输出通道数\n",
+    "        self.fc1 = nn.Linear(20*10*10, 500) # 输入通道数是2000，输出通道数是500\n",
+    "        self.fc2 = nn.Linear(500, 10) # 输入通道数是500，输出通道数是10，即10分类\n",
     "    def forward(self,x):\n",
-    "        in_size = x.size(0)\n",
-    "        out = self.conv1(x) #24\n",
-    "        out = F.relu(out)\n",
-    "        out = F.max_pool2d(out, 2, 2)  #12\n",
-    "        out = self.conv2(out) #10\n",
-    "        out = F.relu(out)\n",
-    "        out = out.view(in_size,-1)\n",
-    "        out = self.fc1(out)\n",
-    "        out = F.relu(out)\n",
-    "        out = self.fc2(out)\n",
-    "        out = F.log_softmax(out,dim=1)\n",
+    "        in_size = x.size(0) # 在本例中in_size=512，也就是BATCH_SIZE的值。输入的x可以看成是512*1*28*28的张量。\n",
+    "        out = self.conv1(x) # batch*1*28*28 -> batch*10*24*24（28x28的图像经过一次核为5x5的卷积，输出变为24x24）\n",
+    "        out = F.relu(out) # batch*10*24*24（激活函数ReLU不改变形状））\n",
+    "        out = F.max_pool2d(out, 2, 2) # batch*10*24*24 -> batch*10*12*12（2*2的池化层会减半）\n",
+    "        out = self.conv2(out) # batch*10*12*12 -> batch*20*10*10（再卷积一次，核的大小是3）\n",
+    "        out = F.relu(out) # batch*20*10*10\n",
+    "        out = out.view(in_size, -1) # batch*20*10*10 -> batch*2000（out的第二维是-1，说明是自动推算，本例中第二维是20*10*10）\n",
+    "        out = self.fc1(out) # batch*2000 -> batch*500\n",
+    "        out = F.relu(out) # batch*500\n",
+    "        out = self.fc2(out) # batch*500 -> batch*10\n",
+    "        out = F.log_softmax(out, dim=1) # 计算log(softmax(x))\n",
     "        return out"
    ]
   },