BatchNormalizationのモデルサンプル１

CIFAR10BN_model1.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "CIFAR10BN_model1.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "3mgvuAy0flDD",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 391
        },
        "outputId": "11f4171c-152f-47f7-8c00-74f9cbfc9f1f"
      },
      "source": [
        "import torch\n",
        "from torchsummary import summary\n",
        "import torch.nn as nn\n",
        "import torch.nn.functional as F\n",
        "\n",
        "class Net(nn.Module):\n",
        "    def __init__(self):\n",
        "        super(Net, self).__init__()\n",
        "        self.conv1 = nn.Conv2d(3, 64, 5)\n",
        "        self.pool = nn.MaxPool2d(2, 2)\n",
        "        self.conv2 = nn.Conv2d(64, 128, 5)\n",
        "        self.fc1 = nn.Linear(128 * 5 * 5, 120)\n",
        "        self.fc2 = nn.Linear(120, 10)\n",
        "        self.bn1 = nn.BatchNorm2d(64)\n",
        "        self.bn2 = nn.BatchNorm2d(128)\n",
        "        self.bn3 = nn.BatchNorm1d(120)\n",
        "\n",
        "    def forward(self, x):\n",
        "        x = self.conv1(x)\n",
        "        x = self.bn1(x)\n",
        "        x = self.pool(F.relu(x))\n",
        "        x = self.conv2(x)\n",
        "        x = self.bn2(x)\n",
        "        x = self.pool(F.relu(x))\n",
        "        x = x.view(-1, 128 * 5 * 5)\n",
        "        x = F.relu(self.fc1(x))\n",
        "        x = self.bn3(x)\n",
        "        x = self.fc2(x)\n",
        "        return x\n",
        "\n",
        "summary(Net(), (3, 32, 32))"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "----------------------------------------------------------------\n",
            "        Layer (type)               Output Shape         Param #\n",
            "================================================================\n",
            "            Conv2d-1           [-1, 64, 28, 28]           4,864\n",
            "       BatchNorm2d-2           [-1, 64, 28, 28]             128\n",
            "         MaxPool2d-3           [-1, 64, 14, 14]               0\n",
            "            Conv2d-4          [-1, 128, 10, 10]         204,928\n",
            "       BatchNorm2d-5          [-1, 128, 10, 10]             256\n",
            "         MaxPool2d-6            [-1, 128, 5, 5]               0\n",
            "            Linear-7                  [-1, 120]         384,120\n",
            "       BatchNorm1d-8                  [-1, 120]             240\n",
            "            Linear-9                   [-1, 10]           1,210\n",
            "================================================================\n",
            "Total params: 595,746\n",
            "Trainable params: 595,746\n",
            "Non-trainable params: 0\n",
            "----------------------------------------------------------------\n",
            "Input size (MB): 0.01\n",
            "Forward/backward pass size (MB): 1.08\n",
            "Params size (MB): 2.27\n",
            "Estimated Total Size (MB): 3.37\n",
            "----------------------------------------------------------------\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}