Vanilla_GAN.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Vanilla_GAN.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": []
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.7.1"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "PIpHAZ9IqyTO",
        "colab": {}
      },
      "source": [
        "import torch\n",
        "import torch.nn as nn\n",
        "import torch.optim as optim\n",
        "from torch.autograd.variable import Variable\n",
        "from torchvision import transforms\n",
        "from torchvision.datasets import MNIST\n",
        "from torchvision.utils import make_grid\n",
        "from torch.utils.data import DataLoader\n",
        "import imageio"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "_CQc_0i8qyTR",
        "colab": {}
      },
      "source": [
        "transform = transforms.Compose([\n",
        "                transforms.ToTensor(),\n",
        "                transforms.Normalize((0.5,),(0.5,))\n",
        "                ])\n",
        "to_image = transforms.ToPILImage()\n",
        "trainset = MNIST(root='./data/', train=True, download=True, transform=transform)\n",
        "trainloader = DataLoader(trainset, batch_size=100, shuffle=True)\n",
        "\n",
        "device = 'cuda'"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "0cJKODM2qyTT",
        "colab": {}
      },
      "source": [
        "class Generator(nn.Module):\n",
        "    def __init__(self):\n",
        "        super(Generator, self).__init__()\n",
        "        self.n_features = 128\n",
        "        self.n_out = 784\n",
        "        self.fc0 = nn.Sequential(\n",
        "                    nn.Linear(self.n_features, 256),\n",
        "                    nn.LeakyReLU(0.2)\n",
        "                    )\n",
        "        self.fc1 = nn.Sequential(\n",
        "                    nn.Linear(256, 512),\n",
        "                    nn.LeakyReLU(0.2)\n",
        "                    )\n",
        "        self.fc2 = nn.Sequential(\n",
        "                    nn.Linear(512, 1024),\n",
        "                    nn.LeakyReLU(0.2)\n",
        "                    )\n",
        "        self.fc3 = nn.Sequential(\n",
        "                    nn.Linear(1024, self.n_out),\n",
        "                    nn.Tanh()\n",
        "                    )\n",
        "    def forward(self, x):\n",
        "        x = self.fc0(x)\n",
        "        x = self.fc1(x)\n",
        "        x = self.fc2(x)\n",
        "        x = self.fc3(x)\n",
        "        x = x.view(-1, 1, 28, 28)\n",
        "        return x\n",
        "\n",
        "class Discriminator(nn.Module):\n",
        "    def __init__(self):\n",
        "        super(Discriminator, self).__init__()\n",
        "        self.n_in = 784\n",
        "        self.n_out = 1\n",
        "        self.fc0 = nn.Sequential(\n",
        "                    nn.Linear(self.n_in, 1024),\n",
        "                    nn.LeakyReLU(0.2),\n",
        "                    nn.Dropout(0.3)\n",
        "                    )\n",
        "        self.fc1 = nn.Sequential(\n",
        "                    nn.Linear(1024, 512),\n",
        "                    nn.LeakyReLU(0.2),\n",
        "                    nn.Dropout(0.3)\n",
        "                    )\n",
        "        self.fc2 = nn.Sequential(\n",
        "                    nn.Linear(512, 256),\n",
        "                    nn.LeakyReLU(0.2),\n",
        "                    nn.Dropout(0.3)\n",
        "                    )\n",
        "        self.fc3 = nn.Sequential(\n",
        "                    nn.Linear(256, self.n_out),\n",
        "                    nn.Sigmoid()\n",
        "                    )\n",
        "    def forward(self, x):\n",
        "        x = x.view(-1, 784)\n",
        "        x = self.fc0(x)\n",
        "        x = self.fc1(x)\n",
        "        x = self.fc2(x)\n",
        "        x = self.fc3(x)\n",
        "        return x"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "5cuePtAyqyTV",
        "colab": {}
      },
      "source": [
        "generator = Generator()\n",
        "discriminator = Discriminator()\n",
        "\n",
        "generator.to(device)\n",
        "discriminator.to(device)\n",
        "\n",
        "g_optim = optim.Adam(generator.parameters(), lr=2e-4)\n",
        "d_optim = optim.Adam(discriminator.parameters(), lr=2e-4)\n",
        "\n",
        "g_losses = []\n",
        "d_losses = []\n",
        "images = []\n",
        "\n",
        "criterion = nn.BCELoss()\n",
        "\n",
        "def noise(n, n_features=128):\n",
        "    return Variable(torch.randn(n, n_features)).to(device)\n",
        "\n",
        "def make_ones(size):\n",
        "    data = Variable(torch.ones(size, 1))\n",
        "    return data.to(device)\n",
        "\n",
        "def make_zeros(size):\n",
        "    data = Variable(torch.zeros(size, 1))\n",
        "    return data.to(device)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "Xe8tdY9tqyTX",
        "colab": {}
      },
      "source": [
        "def train_discriminator(optimizer, real_data, fake_data):\n",
        "    n = real_data.size(0)\n",
        "\n",
        "    optimizer.zero_grad()\n",
        "    \n",
        "    prediction_real = discriminator(real_data)\n",
        "    error_real = criterion(prediction_real, make_ones(n))\n",
        "    error_real.backward()\n",
        "\n",
        "    prediction_fake = discriminator(fake_data)\n",
        "    error_fake = criterion(prediction_fake, make_zeros(n))\n",
        "    \n",
        "    error_fake.backward()\n",
        "    optimizer.step()\n",
        "    \n",
        "    return error_real + error_fake\n",
        "\n",
        "def train_generator(optimizer, fake_data):\n",
        "    n = fake_data.size(0)\n",
        "    optimizer.zero_grad()\n",
        "    \n",
        "    prediction = discriminator(fake_data)\n",
        "    error = criterion(prediction, make_ones(n))\n",
        "    \n",
        "    error.backward()\n",
        "    optimizer.step()\n",
        "    \n",
        "    return error"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "MfmlwbTHqyTZ",
        "scrolled": true,
        "colab": {}
      },
      "source": [
        "num_epochs = 250\n",
        "k = 1\n",
        "test_noise = noise(64)\n",
        "\n",
        "generator.train()\n",
        "discriminator.train()\n",
        "for epoch in range(num_epochs):\n",
        "    g_error = 0.0\n",
        "    d_error = 0.0\n",
        "    for i, data in enumerate(trainloader):\n",
        "        imgs, _ = data\n",
        "        n = len(imgs)\n",
        "        for j in range(k):\n",
        "            fake_data = generator(noise(n)).detach()\n",
        "            real_data = imgs.to(device)\n",
        "            d_error += train_discriminator(d_optim, real_data, fake_data)\n",
        "        fake_data = generator(noise(n))\n",
        "        g_error += train_generator(g_optim, fake_data)\n",
        "\n",
        "    img = generator(test_noise).cpu().detach()\n",
        "    img = make_grid(img)\n",
        "    images.append(img)\n",
        "    g_losses.append(g_error/i)\n",
        "    d_losses.append(d_error/i)\n",
        "    print('Epoch {}: g_loss: {:.8f} d_loss: {:.8f}\\r'.format(epoch, g_error/i, d_error/i))\n",
        "    \n",
        "print('Training Finished')\n",
        "torch.save(generator.state_dict(), 'mnist_generator.pth')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cxt9KxwZ39l8",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import numpy as np\n",
        "from matplotlib import pyplot as plt\n",
        "imgs = [np.array(to_image(i)) for i in images]\n",
        "imageio.mimsave('progress.gif', imgs)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UEy9K5BDRMkw",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 269
        },
        "outputId": "aac5bca5-344f-4ce9-8932-a48fa95a91dc"
      },
      "source": [
        "plt.plot(g_losses, label='Generator_Losses')\n",
        "plt.plot(d_losses, label='Discriminator Losses')\n",
        "plt.legend()\n",
        "plt.savefig('loss.png')"
      ],
      "execution_count": 30,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAW4AAAD8CAYAAABXe05zAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSHALIUSEkeAWQogI8/8BDm0GUhhj87oAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HC2wPPzVSQeM",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}