15
18

Delete article

Deleted articles cannot be recovered.

Draft of this article would be also deleted.

Are you sure you want to delete this article?

More than 3 years have passed since last update.

torchvizで計算グラフを可視化

Posted at

Torchvizのインストール

$ pip install torchviz

ネットワークの定義

  • 可視化したいネットワークを定義する
  • 本稿では,ニューラルネットワークを可視化する
NeuralNet.py
import torch
import torch.nn as nn
import torch.nn.functional as F

INPUT_SIZE = 10

class NeuralNet(nn.Module):
    def __init__(self):
        super(NeuralNet, self).__init__()
        self.fc1 = nn.Linear(INPUT, 256)
        self.fc2 = nn.Linear(256, 128)
        self.fc3 = nn.Linear(128, 10)

    def forward(self, x):
        x = self.fc1(x)
        x = F.relu(x)
        x = self.fc2(x)
        x = F.relu(x)
        x = self.fc3(x)
        x = F.softamx(x)

        return x

torchvizによる計算グラフの可視化

  • torchviz.make_dotにより可視化
  • 入力サイズに合わせたデータを用意し,モデルに流す
  • その出力と,モデルのパラメータを指定して画像を出力する
visualize_NeuralNet.py
import torch

from torchviz import make_dot
from NeuralNet import NeuralNet

INPUT_SIZE = 28*28

model = NeuralNet()
data = torch.randn(1, INPUT_SIZE)

y = model(data)

image = make_dot(y, params=dict(model.named_parameters()))
image.format = "png"
image.render("NeuralNet")

NeuralNet_784.png

15
18
0

Register as a new user and use Qiita more conveniently

  1. You get articles that match your needs
  2. You can efficiently read back useful information
  3. You can use dark theme
What you can do with signing up
15
18

Delete article

Deleted articles cannot be recovered.

Draft of this article would be also deleted.

Are you sure you want to delete this article?