torchinfo:查看网络各层输出大小，参数大小

 torchinfo，可用来可视化我们的网络模型，查看每层的输出大小和参数大小。

(formerly torch-summary)

Torchinfo provides information complementary to what is provided by print(your_model) in PyTorch, similar to Tensorflow's model.summary() API to view the visualization of the model, which is helpful while debugging your network. In this project, we implement a similar functionality in PyTorch and create a clean, simple interface to use in your projects.

本版本torchinfo以替代以前旧的torchsummary和torchsummaryX ，命令也发生了改变。

This is a completely rewritten version of the original torchsummary and torchsummaryX projects by @sksq96 and @nmhkahn. This project addresses all of the issues and pull requests left on the original projects by introducing a completely new API.

Supports PyTorch versions 1.4.0+.

命令：

pip install torchinfo

Alternatively, via conda:

conda install -c conda-forge torchinfo

How To Use

代码：

from torchinfo import summary

model = ConvNet()
batch_size = 16
summary(model, input_size=(batch_size, 1, 28, 28))

 输出结果：

================================================================================================================
Layer (type:depth-idx)          Input Shape          Output Shape         Param #            Mult-Adds
================================================================================================================
SingleInputNet                  [7, 1, 28, 28]       [7, 10]              --                 --
├─Conv2d: 1-1                   [7, 1, 28, 28]       [7, 10, 24, 24]      260                1,048,320
├─Conv2d: 1-2                   [7, 10, 12, 12]      [7, 20, 8, 8]        5,020              2,248,960
├─Dropout2d: 1-3                [7, 20, 8, 8]        [7, 20, 8, 8]        --                 --
├─Linear: 1-4                   [7, 320]             [7, 50]              16,050             112,350
├─Linear: 1-5                   [7, 50]              [7, 10]              510                3,570
================================================================================================================
Total params: 21,840
Trainable params: 21,840
Non-trainable params: 0
Total mult-adds (M): 3.41
================================================================================================================
Input size (MB): 0.02
Forward/backward pass size (MB): 0.40
Params size (MB): 0.09
Estimated Total Size (MB): 0.51
================================================================================================================