[Pytorch系列-48]：如何查看和修改预定义神经网络的网络架构、网络参数属性

#环境准备
import numpy as np              # numpy数组库
import math                     # 数学运算库
import matplotlib.pyplot as plt # 画图库
import time as time

import torch             # torch基础库
import torch.nn as nn    # torch神经网络库
import torch.nn.functional as F
import torchvision.datasets as dataset  #公开数据集的下载和管理
import torchvision.transforms as transforms  #公开数据集的预处理库,格式转换
import torchvision.utils as utils 
import torch.utils.data as data_utils  #对数据集进行分批加载的工具集
from PIL import Image #图片显示
from collections import OrderedDict
import torchvision.models as models

print("Hello World")
print(torch.__version__)
print(torch.cuda.is_available())
print(torch.version.cuda)
print(torch.backends.cudnn.version())

2.2 生成预定义网络实例

# 定义升级网络
model = models.resnet101()

2.3 显示网络结构

# 显示网络的全部架构
print(model)

ResNet(
  (conv1): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)
  (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
  (relu): ReLU(inplace=True)
  (maxpool): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)
  (layer1): Sequential(
    (0): Bottleneck(
      (conv1): Conv2d(64, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
      (downsample): Sequential(
        (0): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
        (1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      )
    )
    (1): Bottleneck(
      (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
    (2): Bottleneck(
      (conv1): Conv2d(256, 64, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(64, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
  )
  (layer2): Sequential(
    (0): Bottleneck(
      (conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
      (downsample): Sequential(
        (0): Conv2d(256, 512, kernel_size=(1, 1), stride=(2, 2), bias=False)
        (1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      )
    )
    (1): Bottleneck(
      (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
    (2): Bottleneck(
      (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
    (3): Bottleneck(
      (conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(128, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
  )
  (layer3): Sequential(
    (0): Bottleneck(
      (conv1): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
      (downsample): Sequential(
        (0): Conv2d(512, 1024, kernel_size=(1, 1), stride=(2, 2), bias=False)
        (1): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      )
    )
    (1): Bottleneck(
      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
    (2): Bottleneck(
      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
    (3): Bottleneck(
      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
    (4): Bottleneck(
      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
    (5): Bottleneck(
      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
    (6): Bottleneck(
      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
    (7): Bottleneck(
      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
    (8): Bottleneck(
      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
    (9): Bottleneck(
      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
    (10): Bottleneck(
      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
    (11): Bottleneck(
      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
    (12): Bottleneck(
      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
    (13): Bottleneck(
      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
    (14): Bottleneck(
      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
    (15): Bottleneck(
      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
    (16): Bottleneck(
      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
    (17): Bottleneck(
      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
    (18): Bottleneck(
      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
    (19): Bottleneck(
      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
    (20): Bottleneck(
      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
    (21): Bottleneck(
      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
    (22): Bottleneck(
      (conv1): Conv2d(1024, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(256, 1024, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
  )
  (layer4): Sequential(
    (0): Bottleneck(
      (conv1): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(2, 2), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
      (downsample): Sequential(
        (0): Conv2d(1024, 2048, kernel_size=(1, 1), stride=(2, 2), bias=False)
        (1): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      )
    )
    (1): Bottleneck(
      (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
    (2): Bottleneck(
      (conv1): Conv2d(2048, 512, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
      (bn2): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (conv3): Conv2d(512, 2048, kernel_size=(1, 1), stride=(1, 1), bias=False)
      (bn3): BatchNorm2d(2048, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
      (relu): ReLU(inplace=True)
    )
  )
  (avgpool): AdaptiveAvgPool2d(output_size=(1, 1))
  (fc): Linear(in_features=2048, out_features=1000, bias=True)
)

2.4 查看网络的内部特定结构以及对应的名称

# 查看网络的内部特定结构以及对应的名称，以后续替换相应的层
print(model.conv1)   #显示conv1层的信息
print(model.bn1)     #显示bn1层的信息
print(model.relu)
print(model.maxpool) #显示maxpool层的信息
print(model.avgpool) #显示avgpool层的信息
print(model.fc)      #显示fc层的信息
print(model.fc.in_features)   #显示fc层的输入特征的个数
print(model.fc.out_features)  #显示fc层的输出特征的个数

Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)
BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
ReLU(inplace=True)
MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)
AdaptiveAvgPool2d(output_size=(1, 1))
Linear(in_features=2048, out_features=1000, bias=True)
2048
1000

备注：

默认输出是1000分类

第3章查看预定义神经网络的参数

3.1 查看模型参数的名称以及结构

for name, parameters in model.named_parameters():
    print(name, ':', parameters.size())

conv1.weight : torch.Size([64, 3, 7, 7])
bn1.weight : torch.Size([64])
bn1.bias : torch.Size([64])
layer1.0.conv1.weight : torch.Size([64, 64, 1, 1])
layer1.0.bn1.weight : torch.Size([64])
layer1.0.bn1.bias : torch.Size([64])
layer1.0.conv2.weight : torch.Size([64, 64, 3, 3])
layer1.0.bn2.weight : torch.Size([64])
layer1.0.bn2.bias : torch.Size([64])
layer1.0.conv3.weight : torch.Size([256, 64, 1, 1])
layer1.0.bn3.weight : torch.Size([256])
layer1.0.bn3.bias : torch.Size([256])
layer1.0.downsample.0.weight : torch.Size([256, 64, 1, 1])
layer1.0.downsample.1.weight : torch.Size([256])
layer1.0.downsample.1.bias : torch.Size([256])
layer1.1.conv1.weight : torch.Size([64, 256, 1, 1])
layer1.1.bn1.weight : torch.Size([64])
layer1.1.bn1.bias : torch.Size([64])
layer1.1.conv2.weight : torch.Size([64, 64, 3, 3])
layer1.1.bn2.weight : torch.Size([64])
layer1.1.bn2.bias : torch.Size([64])
layer1.1.conv3.weight : torch.Size([256, 64, 1, 1])
layer1.1.bn3.weight : torch.Size([256])
layer1.1.bn3.bias : torch.Size([256])
layer1.2.conv1.weight : torch.Size([64, 256, 1, 1])
layer1.2.bn1.weight : torch.Size([64])
layer1.2.bn1.bias : torch.Size([64])
layer1.2.conv2.weight : torch.Size([64, 64, 3, 3])
layer1.2.bn2.weight : torch.Size([64])
layer1.2.bn2.bias : torch.Size([64])
layer1.2.conv3.weight : torch.Size([256, 64, 1, 1])
layer1.2.bn3.weight : torch.Size([256])
layer1.2.bn3.bias : torch.Size([256])
layer2.0.conv1.weight : torch.Size([128, 256, 1, 1])
layer2.0.bn1.weight : torch.Size([128])
layer2.0.bn1.bias : torch.Size([128])
layer2.0.conv2.weight : torch.Size([128, 128, 3, 3])
layer2.0.bn2.weight : torch.Size([128])
layer2.0.bn2.bias : torch.Size([128])
layer2.0.conv3.weight : torch.Size([512, 128, 1, 1])
layer2.0.bn3.weight : torch.Size([512])
layer2.0.bn3.bias : torch.Size([512])
layer2.0.downsample.0.weight : torch.Size([512, 256, 1, 1])
layer2.0.downsample.1.weight : torch.Size([512])
layer2.0.downsample.1.bias : torch.Size([512])
layer2.1.conv1.weight : torch.Size([128, 512, 1, 1])
layer2.1.bn1.weight : torch.Size([128])
layer2.1.bn1.bias : torch.Size([128])
layer2.1.conv2.weight : torch.Size([128, 128, 3, 3])
layer2.1.bn2.weight : torch.Size([128])
layer2.1.bn2.bias : torch.Size([128])
layer2.1.conv3.weight : torch.Size([512, 128, 1, 1])
layer2.1.bn3.weight : torch.Size([512])
layer2.1.bn3.bias : torch.Size([512])
layer2.2.conv1.weight : torch.Size([128, 512, 1, 1])
layer2.2.bn1.weight : torch.Size([128])
layer2.2.bn1.bias : torch.Size([128])
layer2.2.conv2.weight : torch.Size([128, 128, 3, 3])
layer2.2.bn2.weight : torch.Size([128])
layer2.2.bn2.bias : torch.Size([128])
layer2.2.conv3.weight : torch.Size([512, 128, 1, 1])
layer2.2.bn3.weight : torch.Size([512])
layer2.2.bn3.bias : torch.Size([512])
layer2.3.conv1.weight : torch.Size([128, 512, 1, 1])
layer2.3.bn1.weight : torch.Size([128])
layer2.3.bn1.bias : torch.Size([128])
layer2.3.conv2.weight : torch.Size([128, 128, 3, 3])
layer2.3.bn2.weight : torch.Size([128])
layer2.3.bn2.bias : torch.Size([128])
layer2.3.conv3.weight : torch.Size([512, 128, 1, 1])
layer2.3.bn3.weight : torch.Size([512])
layer2.3.bn3.bias : torch.Size([512])
layer3.0.conv1.weight : torch.Size([256, 512, 1, 1])
layer3.0.bn1.weight : torch.Size([256])
layer3.0.bn1.bias : torch.Size([256])
layer3.0.conv2.weight : torch.Size([256, 256, 3, 3])
layer3.0.bn2.weight : torch.Size([256])
layer3.0.bn2.bias : torch.Size([256])
layer3.0.conv3.weight : torch.Size([1024, 256, 1, 1])
layer3.0.bn3.weight : torch.Size([1024])
layer3.0.bn3.bias : torch.Size([1024])
layer3.0.downsample.0.weight : torch.Size([1024, 512, 1, 1])
layer3.0.downsample.1.weight : torch.Size([1024])
layer3.0.downsample.1.bias : torch.Size([1024])
layer3.1.conv1.weight : torch.Size([256, 1024, 1, 1])
layer3.1.bn1.weight : torch.Size([256])
layer3.1.bn1.bias : torch.Size([256])
layer3.1.conv2.weight : torch.Size([256, 256, 3, 3])
layer3.1.bn2.weight : torch.Size([256])
layer3.1.bn2.bias : torch.Size([256])
layer3.1.conv3.weight : torch.Size([1024, 256, 1, 1])
layer3.1.bn3.weight : torch.Size([1024])
layer3.1.bn3.bias : torch.Size([1024])
layer3.2.conv1.weight : torch.Size([256, 1024, 1, 1])
layer3.2.bn1.weight : torch.Size([256])
layer3.2.bn1.bias : torch.Size([256])
layer3.2.conv2.weight : torch.Size([256, 256, 3, 3])
layer3.2.bn2.weight : torch.Size([256])
layer3.2.bn2.bias : torch.Size([256])
layer3.2.conv3.weight : torch.Size([1024, 256, 1, 1])
layer3.2.bn3.weight : torch.Size([1024])
layer3.2.bn3.bias : torch.Size([1024])
layer3.3.conv1.weight : torch.Size([256, 1024, 1, 1])
layer3.3.bn1.weight : torch.Size([256])
layer3.3.bn1.bias : torch.Size([256])
layer3.3.conv2.weight : torch.Size([256, 256, 3, 3])
layer3.3.bn2.weight : torch.Size([256])
layer3.3.bn2.bias : torch.Size([256])
layer3.3.conv3.weight : torch.Size([1024, 256, 1, 1])
layer3.3.bn3.weight : torch.Size([1024])
layer3.3.bn3.bias : torch.Size([1024])
layer3.4.conv1.weight : torch.Size([256, 1024, 1, 1])
layer3.4.bn1.weight : torch.Size([256])
layer3.4.bn1.bias : torch.Size([256])
layer3.4.conv2.weight : torch.Size([256, 256, 3, 3])
layer3.4.bn2.weight : torch.Size([256])
layer3.4.bn2.bias : torch.Size([256])
layer3.4.conv3.weight : torch.Size([1024, 256, 1, 1])
layer3.4.bn3.weight : torch.Size([1024])
layer3.4.bn3.bias : torch.Size([1024])
layer3.5.conv1.weight : torch.Size([256, 1024, 1, 1])
layer3.5.bn1.weight : torch.Size([256])
layer3.5.bn1.bias : torch.Size([256])
layer3.5.conv2.weight : torch.Size([256, 256, 3, 3])
layer3.5.bn2.weight : torch.Size([256])
layer3.5.bn2.bias : torch.Size([256])
layer3.5.conv3.weight : torch.Size([1024, 256, 1, 1])
layer3.5.bn3.weight : torch.Size([1024])
layer3.5.bn3.bias : torch.Size([1024])
layer3.6.conv1.weight : torch.Size([256, 1024, 1, 1])
layer3.6.bn1.weight : torch.Size([256])
layer3.6.bn1.bias : torch.Size([256])
layer3.6.conv2.weight : torch.Size([256, 256, 3, 3])
layer3.6.bn2.weight : torch.Size([256])
layer3.6.bn2.bias : torch.Size([256])
layer3.6.conv3.weight : torch.Size([1024, 256, 1, 1])
layer3.6.bn3.weight : torch.Size([1024])
layer3.6.bn3.bias : torch.Size([1024])
layer3.7.conv1.weight : torch.Size([256, 1024, 1, 1])
layer3.7.bn1.weight : torch.Size([256])
layer3.7.bn1.bias : torch.Size([256])
layer3.7.conv2.weight : torch.Size([256, 256, 3, 3])
layer3.7.bn2.weight : torch.Size([256])
layer3.7.bn2.bias : torch.Size([256])
layer3.7.conv3.weight : torch.Size([1024, 256, 1, 1])
layer3.7.bn3.weight : torch.Size([1024])
layer3.7.bn3.bias : torch.Size([1024])
layer3.8.conv1.weight : torch.Size([256, 1024, 1, 1])
layer3.8.bn1.weight : torch.Size([256])
layer3.8.bn1.bias : torch.Size([256])
layer3.8.conv2.weight : torch.Size([256, 256, 3, 3])
layer3.8.bn2.weight : torch.Size([256])
layer3.8.bn2.bias : torch.Size([256])
layer3.8.conv3.weight : torch.Size([1024, 256, 1, 1])
layer3.8.bn3.weight : torch.Size([1024])
layer3.8.bn3.bias : torch.Size([1024])
layer3.9.conv1.weight : torch.Size([256, 1024, 1, 1])
layer3.9.bn1.weight : torch.Size([256])
layer3.9.bn1.bias : torch.Size([256])
layer3.9.conv2.weight : torch.Size([256, 256, 3, 3])
layer3.9.bn2.weight : torch.Size([256])
layer3.9.bn2.bias : torch.Size([256])
layer3.9.conv3.weight : torch.Size([1024, 256, 1, 1])
layer3.9.bn3.weight : torch.Size([1024])
layer3.9.bn3.bias : torch.Size([1024])
layer3.10.conv1.weight : torch.Size([256, 1024, 1, 1])
layer3.10.bn1.weight : torch.Size([256])
layer3.10.bn1.bias : torch.Size([256])
layer3.10.conv2.weight : torch.Size([256, 256, 3, 3])
layer3.10.bn2.weight : torch.Size([256])
layer3.10.bn2.bias : torch.Size([256])
layer3.10.conv3.weight : torch.Size([1024, 256, 1, 1])
layer3.10.bn3.weight : torch.Size([1024])
layer3.10.bn3.bias : torch.Size([1024])
layer3.11.conv1.weight : torch.Size([256, 1024, 1, 1])
layer3.11.bn1.weight : torch.Size([256])
layer3.11.bn1.bias : torch.Size([256])
layer3.11.conv2.weight : torch.Size([256, 256, 3, 3])
layer3.11.bn2.weight : torch.Size([256])
layer3.11.bn2.bias : torch.Size([256])
layer3.11.conv3.weight : torch.Size([1024, 256, 1, 1])
layer3.11.bn3.weight : torch.Size([1024])
layer3.11.bn3.bias : torch.Size([1024])
layer3.12.conv1.weight : torch.Size([256, 1024, 1, 1])
layer3.12.bn1.weight : torch.Size([256])
layer3.12.bn1.bias : torch.Size([256])
layer3.12.conv2.weight : torch.Size([256, 256, 3, 3])
layer3.12.bn2.weight : torch.Size([256])
layer3.12.bn2.bias : torch.Size([256])
layer3.12.conv3.weight : torch.Size([1024, 256, 1, 1])
layer3.12.bn3.weight : torch.Size([1024])
layer3.12.bn3.bias : torch.Size([1024])
layer3.13.conv1.weight : torch.Size([256, 1024, 1, 1])
layer3.13.bn1.weight : torch.Size([256])
layer3.13.bn1.bias : torch.Size([256])
layer3.13.conv2.weight : torch.Size([256, 256, 3, 3])
layer3.13.bn2.weight : torch.Size([256])
layer3.13.bn2.bias : torch.Size([256])
layer3.13.conv3.weight : torch.Size([1024, 256, 1, 1])
layer3.13.bn3.weight : torch.Size([1024])
layer3.13.bn3.bias : torch.Size([1024])
layer3.14.conv1.weight : torch.Size([256, 1024, 1, 1])
layer3.14.bn1.weight : torch.Size([256])
layer3.14.bn1.bias : torch.Size([256])
layer3.14.conv2.weight : torch.Size([256, 256, 3, 3])
layer3.14.bn2.weight : torch.Size([256])
layer3.14.bn2.bias : torch.Size([256])
layer3.14.conv3.weight : torch.Size([1024, 256, 1, 1])
layer3.14.bn3.weight : torch.Size([1024])
layer3.14.bn3.bias : torch.Size([1024])
layer3.15.conv1.weight : torch.Size([256, 1024, 1, 1])
layer3.15.bn1.weight : torch.Size([256])
layer3.15.bn1.bias : torch.Size([256])
layer3.15.conv2.weight : torch.Size([256, 256, 3, 3])
layer3.15.bn2.weight : torch.Size([256])
layer3.15.bn2.bias : torch.Size([256])
layer3.15.conv3.weight : torch.Size([1024, 256, 1, 1])
layer3.15.bn3.weight : torch.Size([1024])
layer3.15.bn3.bias : torch.Size([1024])
layer3.16.conv1.weight : torch.Size([256, 1024, 1, 1])
layer3.16.bn1.weight : torch.Size([256])
layer3.16.bn1.bias : torch.Size([256])
layer3.16.conv2.weight : torch.Size([256, 256, 3, 3])
layer3.16.bn2.weight : torch.Size([256])
layer3.16.bn2.bias : torch.Size([256])
layer3.16.conv3.weight : torch.Size([1024, 256, 1, 1])
layer3.16.bn3.weight : torch.Size([1024])
layer3.16.bn3.bias : torch.Size([1024])
layer3.17.conv1.weight : torch.Size([256, 1024, 1, 1])
layer3.17.bn1.weight : torch.Size([256])
layer3.17.bn1.bias : torch.Size([256])
layer3.17.conv2.weight : torch.Size([256, 256, 3, 3])
layer3.17.bn2.weight : torch.Size([256])
layer3.17.bn2.bias : torch.Size([256])
layer3.17.conv3.weight : torch.Size([1024, 256, 1, 1])
layer3.17.bn3.weight : torch.Size([1024])
layer3.17.bn3.bias : torch.Size([1024])
layer3.18.conv1.weight : torch.Size([256, 1024, 1, 1])
layer3.18.bn1.weight : torch.Size([256])
layer3.18.bn1.bias : torch.Size([256])
layer3.18.conv2.weight : torch.Size([256, 256, 3, 3])
layer3.18.bn2.weight : torch.Size([256])
layer3.18.bn2.bias : torch.Size([256])
layer3.18.conv3.weight : torch.Size([1024, 256, 1, 1])
layer3.18.bn3.weight : torch.Size([1024])
layer3.18.bn3.bias : torch.Size([1024])
layer3.19.conv1.weight : torch.Size([256, 1024, 1, 1])
layer3.19.bn1.weight : torch.Size([256])
layer3.19.bn1.bias : torch.Size([256])
layer3.19.conv2.weight : torch.Size([256, 256, 3, 3])
layer3.19.bn2.weight : torch.Size([256])
layer3.19.bn2.bias : torch.Size([256])
layer3.19.conv3.weight : torch.Size([1024, 256, 1, 1])
layer3.19.bn3.weight : torch.Size([1024])
layer3.19.bn3.bias : torch.Size([1024])
layer3.20.conv1.weight : torch.Size([256, 1024, 1, 1])
layer3.20.bn1.weight : torch.Size([256])
layer3.20.bn1.bias : torch.Size([256])
layer3.20.conv2.weight : torch.Size([256, 256, 3, 3])
layer3.20.bn2.weight : torch.Size([256])
layer3.20.bn2.bias : torch.Size([256])
layer3.20.conv3.weight : torch.Size([1024, 256, 1, 1])
layer3.20.bn3.weight : torch.Size([1024])
layer3.20.bn3.bias : torch.Size([1024])
layer3.21.conv1.weight : torch.Size([256, 1024, 1, 1])
layer3.21.bn1.weight : torch.Size([256])
layer3.21.bn1.bias : torch.Size([256])
layer3.21.conv2.weight : torch.Size([256, 256, 3, 3])
layer3.21.bn2.weight : torch.Size([256])
layer3.21.bn2.bias : torch.Size([256])
layer3.21.conv3.weight : torch.Size([1024, 256, 1, 1])
layer3.21.bn3.weight : torch.Size([1024])
layer3.21.bn3.bias : torch.Size([1024])
layer3.22.conv1.weight : torch.Size([256, 1024, 1, 1])
layer3.22.bn1.weight : torch.Size([256])
layer3.22.bn1.bias : torch.Size([256])
layer3.22.conv2.weight : torch.Size([256, 256, 3, 3])
layer3.22.bn2.weight : torch.Size([256])
layer3.22.bn2.bias : torch.Size([256])
layer3.22.conv3.weight : torch.Size([1024, 256, 1, 1])
layer3.22.bn3.weight : torch.Size([1024])
layer3.22.bn3.bias : torch.Size([1024])
layer4.0.conv1.weight : torch.Size([512, 1024, 1, 1])
layer4.0.bn1.weight : torch.Size([512])
layer4.0.bn1.bias : torch.Size([512])
layer4.0.conv2.weight : torch.Size([512, 512, 3, 3])
layer4.0.bn2.weight : torch.Size([512])
layer4.0.bn2.bias : torch.Size([512])
layer4.0.conv3.weight : torch.Size([2048, 512, 1, 1])
layer4.0.bn3.weight : torch.Size([2048])
layer4.0.bn3.bias : torch.Size([2048])
layer4.0.downsample.0.weight : torch.Size([2048, 1024, 1, 1])
layer4.0.downsample.1.weight : torch.Size([2048])
layer4.0.downsample.1.bias : torch.Size([2048])
layer4.1.conv1.weight : torch.Size([512, 2048, 1, 1])
layer4.1.bn1.weight : torch.Size([512])
layer4.1.bn1.bias : torch.Size([512])
layer4.1.conv2.weight : torch.Size([512, 512, 3, 3])
layer4.1.bn2.weight : torch.Size([512])
layer4.1.bn2.bias : torch.Size([512])
layer4.1.conv3.weight : torch.Size([2048, 512, 1, 1])
layer4.1.bn3.weight : torch.Size([2048])
layer4.1.bn3.bias : torch.Size([2048])
layer4.2.conv1.weight : torch.Size([512, 2048, 1, 1])
layer4.2.bn1.weight : torch.Size([512])
layer4.2.bn1.bias : torch.Size([512])
layer4.2.conv2.weight : torch.Size([512, 512, 3, 3])
layer4.2.bn2.weight : torch.Size([512])
layer4.2.bn2.bias : torch.Size([512])
layer4.2.conv3.weight : torch.Size([2048, 512, 1, 1])
layer4.2.bn3.weight : torch.Size([2048])
layer4.2.bn3.bias : torch.Size([2048])
fc.weight : torch.Size([1000, 2048])
fc.bias : torch.Size([1000])

3.2 查看模型全部参数的结构以及当前的数值

# 查看模型全部参数的结构以及当前的数值
for parameters in model.parameters():
    print(parameters)

Parameter containing:
tensor([[[[ 0.0129,  0.0295,  0.0005,  ...,  0.0436, -0.0144,  0.0082],
          [ 0.0123,  0.0027, -0.0260,  ..., -0.0539, -0.0083, -0.0259],
          [ 0.0027, -0.0140,  0.0041,  ..., -0.0145,  0.0109, -0.0182],
          ...,
          [ 0.0128, -0.0022,  0.0388,  ..., -0.0116,  0.0571, -0.0283],
          [-0.0015, -0.0179, -0.0010,  ..., -0.0110,  0.0009,  0.0310],
          [ 0.0100, -0.0215,  0.0241,  ..., -0.0019, -0.0834, -0.0293]],

         [[ 0.0086,  0.0038,  0.0213,  ...,  0.0403,  0.0004, -0.0281],
          [-0.0243,  0.0175, -0.0021,  ..., -0.0457, -0.0118, -0.0098],
          [-0.0215,  0.0212,  0.0349,  ..., -0.0090, -0.0021, -0.0105],

.....................

3.3 无名查看模型参数的是否可训练的属性

# 无名查看模型参数的是否可训练的属性
for param in model.parameters():
    print(param.name, param.requires_grad)

None True
None True
None True
None True
None True
........

3.4 有名查看模型参数的是否可训练的属性

# 有名查看模型参数的是否可训练的属性
for name, parameters in model.named_parameters():
    print(name, ':', parameters.requires_grad)

......

layer4.2.bn1.weight : True
layer4.2.bn1.bias : True
layer4.2.conv2.weight : True
layer4.2.bn2.weight : True
layer4.2.bn2.bias : True
layer4.2.conv3.weight : True
layer4.2.bn3.weight : True
layer4.2.bn3.bias : True
fc.weight : True
fc.bias : True

此时全连接参数是可训练的。

第4章修改网络的结构与参数

4.1 # 锁定网络参数的训练

# 锁定网络参数的训练
for param in model.parameters():
    param.requires_grad = False


# 有查看模型参数的是否可训练的属性
for name, parameters in model.named_parameters():
    print(name, ':', parameters.requires_grad)

layer4.2.conv1.weight : False
layer4.2.bn1.weight : False
layer4.2.bn1.bias : False
layer4.2.conv2.weight : False
layer4.2.bn2.weight : False
layer4.2.bn2.bias : False
layer4.2.conv3.weight : False
layer4.2.bn3.weight : False
layer4.2.bn3.bias : False
fc.weight : False
fc.bias : False

备注：所有参数不可训练

4.2 替换全连接网络

#替换升级网络的全连接层
model.fc = nn.Sequential(nn.Linear(in_features = 2048, out_features = 100))


# 有查看模型参数的是否可训练的属性
for name, parameters in model.named_parameters():
    print(name, ':', parameters.requires_grad)

layer4.2.conv1.weight : False
layer4.2.bn1.weight : False
layer4.2.bn1.bias : False
layer4.2.conv2.weight : False
layer4.2.bn2.weight : False
layer4.2.bn2.bias : False
layer4.2.conv3.weight : False
layer4.2.bn3.weight : False
layer4.2.bn3.bias : False
fc.0.weight : True
fc.0.bias : True

备注：只替换了全连接网络

4.3 显示全连接网络

print(model.fc)      #显示fc层的信息

Sequential(
  (0): Linear(in_features=2048, out_features=100, bias=True)
)

作者主页(文火冰糖的硅基工坊)：文火冰糖（王文兵）的博客_文火冰糖的硅基工坊_CSDN博客

本文网址：https://blog.csdn.net/HiWangWenBing/article/details/121342500

本文内容由网友自发贡献，版权归原作者所有，本站不承担相应法律责任。如您发现有涉嫌抄袭侵权的内容，请联系:hwhale#tublm.com(使用前将#替换为@)

人工智能PyTorch

人工智能深度学习

深度学习

神经网络

人工智能

[Pytorch系列-48]：如何查看和修改预定义神经网络的网络架构、网络参数属性的相关文章

用通俗易懂的方式讲解：内容讲解+代码案例，轻松掌握大模型应用框架 LangChain

本文介绍了 LangChain 框架它能够将大型语言模型与其他计算或知识来源相结合从而实现功能更加强大的应用接着对LangChain的关键概念进行了详细说明并基于该框架进行了一些案例尝试旨在帮助读者更轻松地理解 LangChai
用通俗易懂的方式讲解：如何用大语言模型构建一个知识问答系统

传统搜索系统基于关键字匹配在面向游戏攻略技术图谱知识库等业务场景时缺少对用户问题理解和答案二次处理能力本文探索使用大语言模型 Large Language Model LLM 通过其对自然语言理解和生成的能力揣摩用户意图并对
【路径规划】基于A*算法路径规划研究（Matlab代码实现）

欢迎来到本博客博主优势博客内容尽量做到思维缜密逻辑清晰为了方便读者座右铭行百里者半于九十本文目录如下目录 1 概述 2 运行结果 3 参考文献 4 Matlab代码实现
socket网络编程几大模型？看看CHAT是如何回复的？

CHAT回复网络编程中常见的有以下几种模型 1 阻塞I O模型 Blocking I O 传统的同步I O模型一次只处理一个请求 2 非阻塞I O模型 Non blocking I O 应用程序轮询调用socket相关函数检查请求不需
什么是充放电振子理论？

CHAT回复充放电振子模型 Charging Reversal Oscillator Model 是一种解释ENSO现象的理论模型这个模型把ENSO现象比喻成一个热力学振荡系统在这个模型中 ENSO现象由三个组成部分充电 Char
扬帆证券：三只松鼠去年扣非净利预增超1.4倍

在高端性价比战略驱动下三只松鼠 300783 重拾增势 1月15日晚间三只松鼠发布成绩预告预计2023年度净赢利为2亿元至2 2亿元同比增加54 97 至70 47 扣非后净赢利为1亿元至1 1亿元同比增速达146 9 至17
明日 15:00 | NeurIPS 2023 Spotlight 论文

点击蓝字关注我们 AI TIME欢迎每一位AI爱好者的加入哔哩哔哩直播通道扫码关注AITIME哔哩哔哩官方账号预约直播 1月17日 15 00 16 00 讲者介绍黄若孜腾讯AI LAB游戏AI研究员 2020年复旦大学硕士毕业后
毕业设计：基于卷积神经网络的图像分类系统 python人工智能

目录前言设计思路一课题背景与意义二算法理论原理 2 1 卷积神经网络 2 2 SVM算法三检测的实现最后前言大四是整个大学期间最忙碌的时光一边要忙着备考或实习为毕业后面临的就业升学做准备一边要为毕业设计耗费大量精力
作物叶片病害识别系统

介绍由于植物疾病的检测在农业领域中起着重要作用因为植物疾病是相当自然的现象如果在这个领域不采取适当的护理措施就会对植物产生严重影响进而影响相关产品的质量数量或产量植物疾病会引起疾病的周期性爆发导致大规模死亡这些问题需要在初
强烈推荐收藏！LlamaIndex 官方发布高清大图，纵览高级 RAG技术

近日 Llamaindex 官方博客重磅发布了一篇博文 A Cheat Sheet and Some Recipes For Building Advanced RAG 通过一张图给开发者总结了当下主流的高级RAG技术帮助应对复杂的生产场
如何用GPT进行论文润色与改写？

详情点击链接如何用GPT GPT4进行论文润色与改写一OpenAI 1 最新大模型GPT 4 Turbo 2 最新发布的高级数据分析 AI画图图像识别文档API 3 GPT Store 4 从0到1创建自己的GPT应用 5 模型Ge
机器学习算法实战案例：Informer实现多变量负荷预测

文章目录机器学习算法实战案例系列答疑技术交流 1 实验数据集 2 如何运行自己的数据集 3 报错分析机器学习算法实战案例系
不要再苦苦寻觅了！AI 大模型面试指南（含答案）的最全总结来了！

AI 大模型技术经过2023年的狂飙 2024年必将迎来应用的落地对 IT 同学来讲这里蕴含着大量的技术机会越来越多的企业开始招聘 AI 大模型岗位本文梳理了 AI 大模型开发技术的面试之道从 AI 大模型基础面 AI 大模型进阶
蒙特卡洛在发电系统中的应用（Matlab代码实现）

欢迎来到本博客博主优势博客内容尽量做到思维缜密逻辑清晰为了方便读者座右铭行百里者半于九十本文目录如下目录 1 概述 2 运行结果 3 参考文献 4 Matlab代码实现
史上最全自动驾驶岗位介绍

作者自动驾驶转型者编辑汽车人原文链接 https zhuanlan zhihu com p 353480028 点击下方卡片关注自动驾驶之心公众号 ADAS巨卷干货即可获取点击进入自动驾驶之心求职交流技术交流群本
【GRNN-RBFNN-ILC算法】【轨迹跟踪】基于神经网络的迭代学习控制用于未知SISO非线性系统的轨迹跟踪（Matlab代码实现）

欢迎来到本博客博主优势博客内容尽量做到思维缜密逻辑清晰为了方便读者座右铭行百里者半于九十本文目录如下目录 1 概述 2 运行结果 2 1 第1部分 2 2 第2部分
考虑光伏出力利用率的电动汽车充电站能量调度策略研究（Matlab代码实现）

欢迎来到本博客博主优势博客内容尽量做到思维缜密逻辑清晰为了方便读者座右铭行百里者半于九十本文目录如下目录 1 概述 2 运行结果 3 参考文献 4 Matlab代码数据
【GRNN-RBFNN-ILC算法】【轨迹跟踪】基于神经网络的迭代学习控制用于未知SISO非线性系统的轨迹跟踪（Matlab代码实现）

欢迎来到本博客博主优势博客内容尽量做到思维缜密逻辑清晰为了方便读者座右铭行百里者半于九十本文目录如下目录 1 概述 2 运行结果 2 1 第1部分 2 2 第2部分
深度学习(5)--Keras实战

一 Keras基础概念 Keras是深度学习中的一个神经网络框架是一个高级神经网络API 用Python编写可以在TensorFlow CNTK或Theano之上运行 Keras优点 1 允许简单快速的原型设计用户友好性模块化和可扩
实力认证！鼎捷软件荣膺“领军企业”和“创新产品”两大奖项

近日由中国科学院软件研究所中科软科技股份有限公司联合主办的 2023中国软件技术大会于北京成功举办本届大会以大模型驱动下的软件变革为主题数十位来自知名互联网公司和软件巨头企业的技术大咖不同领域行业专家畅销书作者等分享嘉宾

随机推荐

RabbitMQ应用之消息堆积、消息丢失、有序消费、重复消费

文章目录前言一消息堆积 1 消息堆积的产生与影响 2 消息堆积的解决方案二消息丢失 1 情景 2 解决方案三有序消费 1 情景 2 解决方案四重复消费 1 情景 2 解决方案前言最近接触了多线程和MQ等性能相关的内容
HTML 特殊符号编码对照表

特殊符号命名实体十进制编码特殊符号命名实体十进制编码特殊符号命名实体十进制编码 Alpha 913 Beta 914 Gamma 915 Delta 916 Epsilon 917 Zeta 918 Eta 919 Thet
工业互联网产业链全景图深度分析

工业互联网领域有哪些投资机会新基建是与传统基建相对应结合新一轮科技革命和产业变革特征面向国家战略需求为经济社会的创新协调绿色开放共享发展提供底层支撑的具有乘数效应的战略性网络型基础设施其中新基建包括5G基建特高压
openwrt 修改feeds.conf.default为GitHub源

lede和openwrt合并之后 lede官网挂了 git openwrt org 也访问不了只好去github上找最新源码 git clone https github com openwrt openwrt git 复制代码最新的l
嵌入式linux 配置usb otg,嵌入式系统设计中的USB OTG方案

速外设操作时最大为80mA TD1120整个芯片支持功率节省模式包括主机控制器以及外设控制器的延缓模式以使功率消耗最小化www cechina cn 延长系统电池寿命对于移动设备来说电池寿命是很关键的性能接口性能表现 USB数据传输
Bitmap之压缩方案

文章目录前言 1 基础知识 1 1色彩模式 1 2四种模式的区别 1 3具体对比 1 4bitmap内存占用大小计算方式 1 5图片存在的形式 1 6BitampFactory加载Bitmap对象的方式 2 压缩方案 2 1采样率压缩 2
Bugku 计算器

首先打开题目链接发现一个式子但答案有三位数而只能输入一个数字直接F12查看原代码发现maxlenthen 1 maxlenthen意思是文件域可接受的字符数量的上限可输入字符串最大的长度容质所以把1改为3就好啦然后得到fl
Redisson源码-多线程之首个获取锁的线程加解锁流程

Redisson源码多线程之首个获取锁的线程加解锁流程简介当有多个线程同时去获取同一把锁时第一个获取到锁的线程会进行加解锁其他线程需订阅消息并等待锁释放以下源码分析基于redisson 3 17 6版本不同版本源码会有些许不同
openEuler 20.03 LTS SP2以及SP3安装完gnome后，gdm登陆进入不了桌面问题

一问题原因是由于CVE 2020 17489相关补丁引入的暂不清楚是何原因造成但除去该相关补丁之后该问题消失在网上查了下 CVE 2020 17489的问题是gnome shell的某些配置中会发现注销账户时登陆对话框中的密
SQLI-Labs(15-17)

目录 15关 16关 17关 15关看到这个那么我们可以首先尝试报错或者盲注 payload or length database 8 qwe 在这里我们发现and 会报错跟之前我们利用and爆错不一样那为什么这里我们在post传参时
Dubbo是什么

Dubbo是什么 Dubbo是一个分布式服务框架致力于提供高性能和透明化的RPC远程服务调用方案以及SOA服务治理方案简单的说 dubbo就是个服务框架如果没有分布式的需求其实是不需要用的只有在分布式的时候才有dubbo这样的
统计oracle 数据库 lawpeople表lawtype字段多个值只统计一次问题，按照地区分类

select temparea name case when lawtype like 501 then 501 when lawtype like 502 then 502 when lawtype like 503 then 503 w
CSR867x — 如何看懂一份psr文件

XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX XX 作者文化人 XX 联系方式 XX 版权声明原创文章欢迎评论和转载转载时能告诉我一声就最好了 XX 要说的话作者
Dubbo路由规则：静态标签的使用与扩展

一路由的流程路由是通过互联网把信息从源地址传输到目的地址的过程而决定路由目标地址的是路由规则在Dubbo里路由规则在发起一次RPC调用前起到过滤目标服务器地址的作用过滤后的地址列表将作为消费端最终发起RPC调用的备选地址它能
LeetCode 62. 不同路径

欢迎来到茶色岛独家岛屿本期将为大家揭晓LeetCode 62 不同路径做好准备了么那么开始吧一题目名称 LeetCode 62 不同路径二题目要求一个机器人位于一个 m x n 网格的左上角起始点在下图中标记为 Start
git 导出版本之间差异文件

查看 commit id 首先用 git log 查看版本库日志找出需要导出的 commit id git log pretty oneline 456bcbccd91278f7fdf6bf11bc73c4e3a6193c7f HEAD
基于深度神经网络的社交媒体用户级心理压力检测

User Level Psychological Stress Detection from Social Media Using Deep Neural Network 基于深度神经网络的社交媒体用户级心理压力检测 ABSTRACT It
软件anyconnec-win安装下载

anyconnec win介绍 1 安装下载地址 http www drv5 cn sfinfo 14287 html softdown 找到适合自己操作系统的版本下载并安装 2 直接安装下载点击next就ok了需要注意的是下载安装完
IDEA小技巧

IDEA小技巧常用快捷键 Alt Insert 可以自动生成get set toString方法 Alt Enter 可以帮助解决各种报错抛个异常啊导个包啊之类的常见行操作 Shift Enter 添加空行相比普通换行不管光标在
[Pytorch系列-48]：如何查看和修改预定义神经网络的网络架构、网络参数属性

作者主页文火冰糖的硅基工坊文火冰糖王文兵的博客文火冰糖的硅基工坊 CSDN博客本文网址 https blog csdn net HiWangWenBing article details 121342500 目录第1章 Fin

[Pytorch系列-48]：如何查看和修改预定义神经网络的网络架构、网络参数属性

第1章 FineTuning、Transfer Trainning的理论基础与深度解析

第2章 查看预定义神经网络的网络架构

2.1 前置条件

2.2 生成预定义网络实例

2.3 显示网络结构

2.4 查看网络的内部特定结构以及对应的名称

第3章 查看预定义神经网络的参数

3.1 查看模型参数的名称以及结构

3.2 查看模型全部参数的结构以及当前的数值

3.3 无名查看模型参数的是否可训练的属性

3.4 有名查看模型参数的是否可训练的属性

第4章 修改网络的结构与参数

4.1 # 锁定网络参数的训练

4.2 替换全连接网络

4.3 显示全连接网络

[Pytorch系列-48]：如何查看和修改预定义神经网络的网络架构、网络参数属性 的相关文章

随机推荐

热门标签

第2章查看预定义神经网络的网络架构

第3章查看预定义神经网络的参数

第4章修改网络的结构与参数

[Pytorch系列-48]：如何查看和修改预定义神经网络的网络架构、网络参数属性的相关文章