import torch
import torch.nn as nn
import torch.nn.functional as F
import inspect
import torch.optim as optim
'''
 自動求導機制： 對一個標量用backward() 會反向計算在計算圖中用到的葉節點的梯度， 如果設定了非葉節點.retain_grad
 則可以記錄中間節點的梯度。  節點的grad_fn 記錄了這個節點通過一個什麼樣的函式得來的
'''

class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        self.conv1 = nn.Conv2d(1, 6, 5)
        self.conv2 = nn.Conv2d(6, 16, 5)
        self.fc1 = nn.Linear(16 * 5 * 5, 120)
        self.fc2 = nn.Linear(120, 84)
        self.fc3 = nn.Linear(84, 10)


    # 定義好forward函式， 呼叫backward()會自動計算導數
    def forward(self, x):
        x = F.max_pool2d(F.relu(self.conv1(x)), (2, 2))
        x = F.max_pool2d(F.relu(self.conv2(x)), 2)
        x = x.view(x.size()[0], -1)
        x = F.relu(self.fc1(x))
        x = F.relu(self.fc2(x))
        x = self.fc3(x)
        return x


    def num_flat_features(self, x):
        # 取除了batchsize以外的維度
        pass

net = Net()
print(net)
input = torch.randn(1, 1, 32, 32)
output = net(input)
target = torch.randn(10)
# 新增一個批次的維度
target = target.view(1, -1)
criterion = nn.MSELoss()
loss = criterion(output, target)

net.zero_grad()
print(net.conv1.bias.grad)
loss.backward()
print(net.conv1.bias.grad)

params = list(net.parameters())
for i in range(len(params)):
    print(params[i].size(), '\n')

# 自定義的一個簡化的SGD演算法
learning_rate = 0.1
for f in net.parameters():
    # 利用sub in-place
    f.data.sub_(f.grad.data * learning_rate)

# 或者採用系統定義的優化方法
optimizer = optim.SGD(net.parameters(), lr=0.01)


 

net.parameters:

torch.Size([6, 1, 5, 5]) 

torch.Size([6]) 

torch.Size([16, 6, 5, 5]) 

torch.Size([16]) 

torch.Size([120, 400]) 

torch.Size([120]) 

torch.Size([84, 120]) 

torch.Size([84]) 

torch.Size([10, 84]) 

torch.Size([10])