在pytorch中如何查看模型model参数parameters

pytorch查看模型model参数parameters

示例1：pytorch自带的faster r-cnn模型

import torch
import torchvision

model = torchvision.models.detection.fasterrcnn_resnet50_fpn(pretrained=True)

for name, p in model.named_parameters():
   print(name)
   print(p.requires_grad)
   print(...)

#或者

for p in model.parameters():
   print(p)
   print(...)

示例2：自定义网络模型

class Net(nn.Module):
   def __init__(self):
       super(Net, self).__init__()

cfg = [64, 64, 'M', 128, 128, 'M', 256, 256, 256, 'M', 512, 512, 512, 'M', 512, 512, 512]
       self.features = self._vgg_layers(cfg)

def _vgg_layers(self, cfg):
       layers = []
       in_channels = 3
       for x in cfg:
           if x == 'M':
               layers += [nn.MaxPool2d(kernel_size=2, stride=2)]
           else:
               layers += [nn.Conv2d(in_channels, x ,kernel_size=3, padding=1),
                       nn.BatchNorm2d(x),
                       nn.ReLU(inplace=True)
                       ]
               in_channels = x

return nn.Sequential(*layers)

def forward(self, data):
       out_map = self.features(data)
       return out_map

Model = Net()

for name, p in model.named_parameters():
   print(name)
   print(p.requires_grad)
   print(...)

#或者

for p in model.parameters():
   print(p)
   print(...)

在自定义网络中，model.parameters()方法继承自nn.Module

pytorch查看模型参数总结

1：DNN_printer

其中（3, 32, 32）是输入的大小，其他方法中的参数同理

from DNN_printer import DNN_printer

batch_size = 512
def train(epoch):
   print('\nEpoch: ％d' ％ epoch)
   net.train()
   train_loss = 0
   correct = 0
   total = 0
   // put the code here and you can get the result
   DNN_printer(net, (3, 32, 32),batch_size)

结果

2：parameters

def cnn_paras_count(net):
   """cnn参数量统计, 使用方式cnn_paras_count(net)"""
   # Find total parameters and trainable parameters
   total_params = sum(p.numel() for p in net.parameters())
   print(f'{total_params:,} total parameters.')
   total_trainable_params = sum(p.numel() for p in net.parameters() if p.requires_grad)
   print(f'{total_trainable_params:,} training parameters.')
   return total_params, total_trainable_params

cnn_paras_count(net)

直接输出参数量，然后自己计算

需要注意的是，一般模型中参数是以float32保存的，也就是一个参数由4个bytes表示，那么就可以将参数量转化为存储大小。

例如：

• 44426个参数*4 / 1024 ≈ 174KB

3：get_model_complexity_info()

from ptflops import get_model_complexity_info
from torchvision import models

net = models.mobilenet_v2()
ops, params = get_model_complexity_info(net, (3, 224, 224), as_strings=True,
print_per_layer_stat=True, verbose=True)

4：torchstat

from torchstat import stat
import torchvision.models as models
model = models.resnet152()
stat(model, (3, 224, 224))

输出

来源：https://blog.csdn.net/qq_38600065/article/details/105552816