DCGAN.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/bfortuner/anaconda3/lib/python3.6/site-packages/matplotlib/__init__.py:1401: UserWarning:  This call to matplotlib.use() has no effect\n",
      "because the backend has already been chosen;\n",
      "matplotlib.use() must be called *before* pylab, matplotlib.pyplot,\n",
      "or matplotlib.backends is imported for the first time.\n",
      "\n",
      "  warnings.warn(_use_error_msg)\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import os\n",
    "import random\n",
    "import torch\n",
    "from PIL import Image\n",
    "import scipy.misc\n",
    "from torchvision import datasets, transforms\n",
    "from torch import nn\n",
    "from torch.autograd import Variable\n",
    "import torch.nn.init as init\n",
    "import torch.optim as optim\n",
    "import torch.backends.cudnn as cudnn\n",
    "import torch.autograd as autograd\n",
    "import torchvision\n",
    "import matplotlib as mpl\n",
    "from tqdm import tqdm\n",
    "mpl.use('Agg')\n",
    "import matplotlib.pyplot as plt\n",
    "plt.style.use('bmh')\n",
    "import torchvision.utils as vutils\n",
    "\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "random.seed(1)\n",
    "torch.manual_seed(1)\n",
    "DATA_PATH='/media/bfortuner/bigguy/data/'\n",
    "CUB_PATH=DATA_PATH+'cub/'\n",
    "CUB_IMAGES_PATH=CUB_PATH+'images/'\n",
    "MNIST_PATH=DATA_PATH+'mnist/'\n",
    "MNIST_WEIGHTS_PATH=MNIST_PATH+'weights/'\n",
    "MNIST_RESULTS_PATH='results/'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "MNIST_BATCH_SIZE = 128\n",
    "MNIST_MEAN = np.array([0.1307,])\n",
    "MNIST_STD = np.array([0.3081,])\n",
    "train_loader = torch.utils.data.DataLoader(\n",
    "    datasets.MNIST(MNIST_PATH, train=True, download=True,\n",
    "                   transform=transforms.Compose([\n",
    "                       transforms.ToTensor(),\n",
    "                       transforms.Normalize(MNIST_MEAN, MNIST_STD)\n",
    "                   ])),\n",
    "    batch_size=MNIST_BATCH_SIZE, shuffle=True)\n",
    "test_loader = torch.utils.data.DataLoader(\n",
    "    datasets.MNIST(MNIST_PATH, train=False, transform=transforms.Compose([\n",
    "                       transforms.ToTensor(),\n",
    "                       transforms.Normalize(MNIST_MEAN, MNIST_STD)\n",
    "                   ])),\n",
    "    batch_size=MNIST_BATCH_SIZE, shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def plot_mnist(arr, title=None):\n",
    "    \"\"\"If normalized with mean/std\"\"\"\n",
    "    if type(arr) is not np.ndarray:\n",
    "        arr = arr.numpy().transpose((1, 2, 0))\n",
    "    arr = denorm_meanstd(arr, MNIST_MEAN, MNIST_STD)\n",
    "    arr = arr.reshape((28, 28))\n",
    "    plt.imshow(arr, cmap='gray')\n",
    "    if title is not None:\n",
    "        plt.title(title)\n",
    "    plt.pause(0.001)  # pause a bit so that plots are updated\n",
    "    plt.axis('off')\n",
    "    plt.show()\n",
    "\n",
    "def plot_samples(arr, dim=(4,4), figsize=(6,6)):\n",
    "    if type(arr) is not np.ndarray:\n",
    "        arr = arr.numpy()\n",
    "    bs = arr.shape[0]\n",
    "    arr = arr.reshape(bs, 28, 28)\n",
    "    plt.figure(figsize=figsize)\n",
    "    for i,img in enumerate(arr):\n",
    "        plt.subplot(*dim, i+1)\n",
    "        plt.imshow(img, cmap='gray')\n",
    "        plt.axis('off')\n",
    "    plt.tight_layout()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Use the 1,28,28 MNIST images \n",
    "\n",
    "def get_noise(batch_size, vec_size):\n",
    "    return Variable(torch.FloatTensor(\n",
    "        batch_size, vec_size, 1, 1).normal_(0,1).cuda())\n",
    "    \n",
    "def weights_init(m):\n",
    "    if isinstance(m, nn.ConvTranspose2d) or isinstance(m, nn.Conv2d):\n",
    "        init.normal(m.weight.data, mean=0, std=0.2)\n",
    "    elif isinstance(m, nn.BatchNorm2d):\n",
    "        init.normal(m.weight.data, mean=1, std=0.2)\n",
    "        m.bias.data.zero_()\n",
    "\n",
    "class G_Block(nn.Sequential):\n",
    "    def __init__(self, in_chans, out_chans, f_size, stride, pad):\n",
    "        super(G_Block, self).__init__()\n",
    "        self.convTrans = nn.ConvTranspose2d(\n",
    "            in_chans, out_chans, f_size, stride, pad, bias=False)\n",
    "        self.batchNorm = nn.BatchNorm2d(out_chans)\n",
    "        self.ReLU = nn.ReLU(True)\n",
    "\n",
    "class netG(nn.Module):\n",
    "    def __init__(self, in_chans, out_chans):\n",
    "        super(netG, self).__init__()\n",
    "        self.block1 = G_Block(in_chans, 256, 7, stride=1, pad=0)\n",
    "        self.block2 = G_Block(256, 128, 4, stride=2, pad=1)\n",
    "#         self.block3 = G_Block(256, 128, 4, stride=2, pad=1)\n",
    "#         self.block4 = G_Block(128, 64, 4, stride=2, pad=1)\n",
    "        self.convTrans = nn.ConvTranspose2d(\n",
    "            128, out_chans, 4, stride=2, padding=1)\n",
    "        self.tanh = nn.Tanh()\n",
    "\n",
    "    def forward(self, x):\n",
    "        #print (x.size())\n",
    "        out = self.block1(x)\n",
    "        #print(out.size())\n",
    "        out = self.block2(out)\n",
    "        #print(out.size())\n",
    "#         out = self.block3(out)\n",
    "#         print(out.size())\n",
    "#         out = self.block4(out)\n",
    "#         print(out.size())\n",
    "        out = self.convTrans(out)\n",
    "        #print(out.size())\n",
    "        out = self.tanh(out)\n",
    "        return out\n",
    "\n",
    "class D_Block(nn.Sequential):\n",
    "    def __init__(self, in_chans, out_chans, f_size, stride, pad, bn=True):\n",
    "        super(D_Block, self).__init__()\n",
    "        self.conv = nn.Conv2d(\n",
    "            in_chans, out_chans, f_size, stride, pad, bias=False)\n",
    "        if bn:\n",
    "            self.bn = nn.BatchNorm2d(out_chans)\n",
    "        self.relu = nn.LeakyReLU(negative_slope=0.2, inplace=True)\n",
    "\n",
    "class netD(nn.Module):\n",
    "    def __init__(self, in_chans):\n",
    "        super(netD, self).__init__()\n",
    "        self.block1 = D_Block(in_chans, 64, 4, stride=2, pad=1, bn=False)\n",
    "        self.block2 = D_Block(64, 128, 4, stride=2, pad=1)\n",
    "        self.block3 = D_Block(128, 256, 4, stride=2, pad=1)\n",
    "        #self.block4 = D_Block(256, 512, 4, stride=2, pad=1)\n",
    "        self.conv = nn.Conv2d(256, 1, 3, stride=1, padding=0)\n",
    "        self.sigmoid = nn.Sigmoid()\n",
    "\n",
    "    def forward(self, x):\n",
    "        #print(x.size())\n",
    "        out = self.block1(x)\n",
    "        #print(out.size())\n",
    "        out = self.block2(out)\n",
    "        #print(out.size())\n",
    "        out = self.block3(out)\n",
    "        #print(out.size())\n",
    "        #out = self.block4(out)\n",
    "        #print(out.size())\n",
    "        out = self.conv(out)\n",
    "        #print(out.size())\n",
    "        out = self.sigmoid(out)\n",
    "        return out.view(-1,1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "cudnn.benchmark = True\n",
    "\n",
    "noise_vec_size = 100 #input to generator\n",
    "\n",
    "D = netD(1).cuda()\n",
    "G = netG(noise_vec_size, 1).cuda()\n",
    "D.apply(weights_init)\n",
    "G.apply(weights_init)\n",
    "\n",
    "# both are evaluated with a 1/0 answer from discriminator\n",
    "criterion = nn.BCELoss().cuda() \n",
    "D_optimizer = optim.Adam(D.parameters(), lr=2e-4, betas=(0.5, 0.999))\n",
    "G_optimizer = optim.Adam(G.parameters(), lr=2e-4, betas=(0.5, 0.999))\n",
    "\n",
    "fixed_noise = Variable(torch.FloatTensor(\n",
    "    MNIST_BATCH_SIZE, noise_vec_size, 1, 1).normal_(0,1).cuda())\n",
    "real_labels = Variable(torch.ones(MNIST_BATCH_SIZE, 1).cuda())\n",
    "fake_labels = Variable(torch.zeros(MNIST_BATCH_SIZE, 1).cuda())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def train(G, D, criterion, G_optim, D_optim, dataloader, epochs):\n",
    "    n_batches = len(dataloader)\n",
    "    D_total_loss = 0\n",
    "    G_total_loss = 0\n",
    "    for inputs,targets in train_loader:\n",
    "        batch_size = len(inputs)\n",
    "        real_imgs = Variable(inputs.cuda())\n",
    "\n",
    "        # Clear gradients\n",
    "        D.zero_grad()\n",
    "        G.zero_grad()\n",
    "        \n",
    "        ## Discriminator - Real\n",
    "        D_real_out = D(real_imgs)\n",
    "        D_real_loss = criterion(D_real_out, real_labels[:batch_size])\n",
    "        D_real_loss.backward()\n",
    "        \n",
    "        ## Discriminator - Fake\n",
    "        noise = get_noise(batch_size, noise_vec_size)\n",
    "        D_fake_out = D(G(noise).detach())\n",
    "        D_fake_loss = criterion(D_fake_out, fake_labels[:batch_size])\n",
    "\n",
    "        D_fake_loss.backward()\n",
    "        D_optim.step()\n",
    "        \n",
    "        D_loss = D_real_loss + D_fake_loss\n",
    "        \n",
    "        ## Generator\n",
    "        G_out = D(G(noise))\n",
    "        G_loss = criterion(G_out, real_labels[:batch_size])\n",
    "        \n",
    "        G_loss.backward()\n",
    "        G_optim.step()\n",
    "        \n",
    "        D_total_loss += D_loss.data[0] / n_batches\n",
    "        G_total_loss += G_loss.data[0] / n_batches\n",
    "        \n",
    "    # do checkpointing\n",
    "    torch.save(D.state_dict(), '%sD_mnist_%03d.pth' % (MNIST_WEIGHTS_PATH, epoch))\n",
    "    torch.save(G.state_dict(), '%sG_mnist_%03d.pth' % (MNIST_WEIGHTS_PATH, epoch))\n",
    "    return D_total_loss, G_total_loss\n",
    "\n",
    "def test(G, out_path, n_samples, epoch=1):\n",
    "    fake = G(fixed_noise[:n_samples])\n",
    "    vutils.save_image(fake.data.cpu(),\n",
    "            '%s/fake_samples_epoch_%03d.png' % (out_path, epoch),\n",
    "            normalize=True)\n",
    "    return fake.data.cpu()\n",
    "\n",
    "def plot_fake_samples(G, n_samples=12):\n",
    "    fake = G(fixed_noise[:n_samples]).data.cpu()\n",
    "    plot_samples(fake)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 2/2 [01:12<00:00, 36.30s/it]\n"
     ]
    }
   ],
   "source": [
    "EPOCHS=300\n",
    "\n",
    "loss_history = []\n",
    "for epoch in tqdm(range(1,EPOCHS+1)):\n",
    "    D_loss, G_loss = train(G, D, criterion, G_optimizer, D_optimizer, train_loader, epoch)\n",
    "    loss_history.append([D_loss, G_loss])\n",
    "    fake_samples = test(G, MNIST_RESULTS_PATH, 16, epoch)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAW0AAAD1CAYAAACbQ6S4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nLSKSIyraIiI5oqItIpIjKtoiIjny/wGnJYdO4a5UnAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f8bb6784048>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAakAAAFHCAYAAAAbYrOpAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f8bb6716358>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "loss_history = np.array(loss_history)\n",
    "plt.plot(loss_history[:,0]) #D_loss\n",
    "plt.plot(loss_history[:,1]) #G_loss\n",
    "plot_fake_samples(G)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.0"
  },
  "latex_envs": {
   "LaTeX_envs_menu_present": true,
   "autocomplete": true,
   "bibliofile": "biblio.bib",
   "cite_by": "apalike",
   "current_citInitial": 1,
   "eqLabelWithNumbers": true,
   "eqNumInitial": 1,
   "hotkeys": {
    "equation": "Ctrl-E",
    "itemize": "Ctrl-I"
   },
   "labels_anchors": false,
   "latex_user_defs": false,
   "report_style_numbering": false,
   "user_envs_cfg": false
  },
  "toc": {
   "colors": {
    "hover_highlight": "#DAA520",
    "running_highlight": "#FF0000",
    "selected_highlight": "#FFD700"
   },
   "moveMenuLeft": true,
   "nav_menu": {
    "height": "4px",
    "width": "254px"
   },
   "navigate_menu": true,
   "number_sections": true,
   "sideBar": true,
   "threshold": 4,
   "toc_cell": false,
   "toc_section_display": "block",
   "toc_window_display": false,
   "widenNotebook": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}