pytorch参数管理

import torch
from torch import nn

net=nn.Sequential(nn.Linear(4,8),nn.ReLU(),nn.Linear(8,1))
x=torch.rand(size=(2,4))
y=net(x)

print(y)

print(net[2].state_dict())

print(type(net[2].bias))
print(net[2].bias)
print(net[2].bias.data)

print(net[2].weight.grad==None)

print(*[(name,param.shape)for name,param in net[0].named_parameters()])
print(*[(name,param.shape)for name,param in net.named_parameters()])

a=net.state_dict()['2.bias'].data
print(a)

# 有嵌套情况下
def block1():
    return nn.Sequential(nn.Linear(4,8),nn.ReLU(),nn.Linear(8,4),nn.ReLU())

def block2():
    net=nn.Sequential()
    for i in range(4):
        net.add_module(f'block{i}',block1())
    return net

rgnet=nn.Sequential(block2(),nn.Linear(4,1))
rgnet(x)

print(rgnet)

# 修改默认的初始参数
def init_normal(m):#m就是一个moudle
    if type(m)==nn.Linear:
        nn.init.normal_(m.weight,mean=0,std=0.01)#替换函数
        nn.init.zeros_(m.bias)
#所有函数都在init里面
net.apply(init_normal)#对于所有net的层，调用函数，遍历一遍6
c,d=net[0].weight.data[0],net[0].bias.data[0]
print(c,d)

def init_constant(m):
    if type(m)==nn.Linear:
        nn.init.constant_(m.weight,1)
        nn.init.zeros_(m.bias)

net.apply(init_constant)

e,f=net[0].weight.data[0],net[0].bias.data[0]
print(e,f)

#简单的方法
net[0].weight.data[:]+=1
net[0].weight.data[0,0]=42
g=net[0].weight.data[0]


#层共享参数
shared=nn.Linear(8,8)
net=nn.Sequential(nn.Linear(4,8),nn.ReLU(),shared,nn.ReLU(),shared,nn.ReLU(),nn.Linear(8,1))
print(net)