pytorch快速搭建卷积神经网络【优化器_第3课_获取模型参数_state_dict】

理论

state_dict()
功能：获取模型当前的参数，以一个有序字典形式返回。 这个有序字典中，key 是各层参数名，value 就是参数。

代码

# coding: utf-8

import torch.nn as nn
import torch.nn.functional as F


# ----------------------------------- state_dict
class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        self.conv1 = nn.Conv2d(3, 1, 3)
        self.pool = nn.MaxPool2d(2, 2)
        self.fc1 = nn.Linear(1 * 3 * 3, 2)

    def forward(self, x):
        x = self.pool(F.relu(self.conv1(x)))
        x = x.view(-1, 1 * 3 * 3)
        x = F.relu(self.fc1(x))
        return x

net = Net()

# 获取网络当前参数
net_state_dict = net.state_dict()

print('net_state_dict类型：', type(net_state_dict))
print('net_state_dict管理的参数: ', net_state_dict.keys())
print('net_state_dict:', net_state_dict)
for key, value in net_state_dict.items():
    print('参数名: ', key, '\t大小: ',  value.shape)

结果

net_state_dict类型： <class 'collections.OrderedDict'>
net_state_dict管理的参数:  odict_keys(['conv1.weight', 'conv1.bias', 'fc1.weight', 'fc1.bias'])
net_state_dict: OrderedDict([
          ('conv1.weight', tensor([[[[ 0.1847,  0.1356,  0.0460],
							          [-0.0917, -0.1756, -0.0554],
							          [-0.1254, -0.1861, -0.0111]],
          
							         [[ 0.0801,  0.0135, -0.1842],
							          [-0.1620,  0.0984,  0.0845],
							          [ 0.0077,  0.0937, -0.0286]],
							
							         [[ 0.0141,  0.0263,  0.1229],
							          [ 0.1565,  0.0061,  0.0611],
							          [-0.0304,  0.1011,  0.1769]]]])), 
          ('conv1.bias', tensor([0.0328])), 
          ('fc1.weight', tensor([[-0.0093, -0.0534, -0.2707, -0.2470,  0.2448,  0.2793, -0.2237,  0.1177, 0.1881],
                                 [-0.1974, -0.2257,  0.0995, -0.2708,  0.3137, -0.0334,  0.2609, -0.0878,-0.1505]])), 
          ('fc1.bias',   tensor([0.2143, 0.0708]))])
          
参数名:  conv1.weight 	大小:  torch.Size([1, 3, 3, 3])
参数名:  conv1.bias 	大小:  torch.Size([1])
参数名:  fc1.weight 	大小:  torch.Size([2, 9])
参数名:  fc1.bias 	    大小:  torch.Size([2])