08_Autoencoder/2_Convolutional_Denoising_Autoencoder.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Convolutional Denoising Autoencoder\n",
    "- MNIST dataset\n",
    "- Convolutional Neural Network\n",
    "- 2 hidden layers"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. Settings\n",
    "### 1) Import required libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.optim as optim\n",
    "import torch.nn.init as init\n",
    "import torchvision.datasets as dset\n",
    "import torchvision.transforms as transforms\n",
    "from torch.utils.data import DataLoader\n",
    "from torch.autograd import Variable\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2) Set hyperparameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "batch_size = 100\n",
    "learning_rate = 0.0002\n",
    "num_epoch = 1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. Data\n",
    "\n",
    "### 1) Download Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "mnist_train = dset.MNIST(\"./\", train=True, transform=transforms.ToTensor(), target_transform=None, download=True)\n",
    "mnist_test = dset.MNIST(\"./\", train=False, transform=transforms.ToTensor(), target_transform=None, download=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2) Set DataLoader"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_loader = torch.utils.data.DataLoader(mnist_train,batch_size=batch_size, shuffle=True,num_workers=2,drop_last=True)\n",
    "test_loader = torch.utils.data.DataLoader(mnist_test,batch_size=batch_size, shuffle=False,num_workers=2,drop_last=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4. Model & Optimizer\n",
    "### 1) Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "class Encoder(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(Encoder,self).__init__()\n",
    "        self.layer1 = nn.Sequential(\n",
    "                        nn.Conv2d(1,16,3,padding=1),   # batch x 16 x 28 x 28\n",
    "                        nn.ReLU(),\n",
    "                        nn.BatchNorm2d(16),\n",
    "                        nn.Conv2d(16,32,3,padding=1),  # batch x 32 x 28 x 28\n",
    "                        nn.ReLU(),\n",
    "                        nn.BatchNorm2d(32),\n",
    "                        nn.Conv2d(32,64,3,padding=1),  # batch x 32 x 28 x 28\n",
    "                        nn.ReLU(),\n",
    "                        nn.BatchNorm2d(64),\n",
    "                        nn.MaxPool2d(2,2)   # batch x 64 x 14 x 14\n",
    "        )\n",
    "        self.layer2 = nn.Sequential(\n",
    "                        nn.Conv2d(64,128,3,padding=1),  # batch x 64 x 14 x 14\n",
    "                        nn.ReLU(),\n",
    "                        nn.BatchNorm2d(128),\n",
    "                        nn.MaxPool2d(2,2),\n",
    "                        nn.Conv2d(128,256,3,padding=1),  # batch x 64 x 7 x 7\n",
    "                        nn.ReLU()\n",
    "        )\n",
    "        \n",
    "                \n",
    "    def forward(self,x):\n",
    "        out = self.layer1(x)\n",
    "        out = self.layer2(out)\n",
    "        out = out.view(batch_size, -1)\n",
    "        return out\n",
    "    \n",
    "encoder = Encoder().cuda()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "class Decoder(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(Decoder,self).__init__()\n",
    "        self.layer1 = nn.Sequential(\n",
    "                        nn.ConvTranspose2d(256,128,3,2,1,1),\n",
    "                        nn.ReLU(),\n",
    "                        nn.BatchNorm2d(128),\n",
    "                        nn.ConvTranspose2d(128,64,3,1,1),\n",
    "                        nn.ReLU(),\n",
    "                        nn.BatchNorm2d(64)\n",
    "        )\n",
    "        self.layer2 = nn.Sequential(\n",
    "                        nn.ConvTranspose2d(64,16,3,1,1),\n",
    "                        nn.ReLU(),\n",
    "                        nn.BatchNorm2d(16),\n",
    "                        nn.ConvTranspose2d(16,1,3,2,1,1),\n",
    "                        nn.ReLU()\n",
    "        )\n",
    "        \n",
    "    def forward(self,x):\n",
    "        out = x.view(batch_size,256,7,7)\n",
    "        out = self.layer1(out)\n",
    "        out = self.layer2(out)\n",
    "        return out\n",
    "\n",
    "decoder = Decoder().cuda()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2) Loss func & Optimizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "parameters = list(encoder.parameters())+ list(decoder.parameters())\n",
    "loss_func = nn.MSELoss()\n",
    "optimizer = torch.optim.Adam(parameters, lr=learning_rate)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5. Train "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "--------model restored--------\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2172\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.5/dist-packages/torch/serialization.py:147: UserWarning: Couldn't retrieve source code for container of type Encoder. It won't be checked for correctness upon loading.\n",
      "  \"type \" + obj.__name__ + \". It won't be checked \"\n",
      "/usr/local/lib/python3.5/dist-packages/torch/serialization.py:147: UserWarning: Couldn't retrieve source code for container of type Decoder. It won't be checked for correctness upon loading.\n",
      "  \"type \" + obj.__name__ + \". It won't be checked \"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2208\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.1896\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2246\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2355\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2066\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2280\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2372\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2576\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2069\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2072\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.1843\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2058\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.1890\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2383\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.1834\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2250\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2464\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2286\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.1563\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2494\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2058\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.1611\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.1717\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.1957\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2393\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.1510\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2164\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.1798\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2558\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2407\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.1861\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2196\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.1836\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2082\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2651\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2998\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.1876\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2809\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2475\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2245\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2675\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2876\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2214\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2641\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.1919\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2419\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2163\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.1859\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2498\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2360\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.3222\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2190\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2477\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2207\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.3220\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.1909\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2633\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2624\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.1817\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n"
     ]
    }
   ],
   "source": [
    "try:\n",
    "    encoder, decoder = torch.load('./model/conv_deno_autoencoder.pkl')\n",
    "    print(\"\\n--------model restored--------\\n\")\n",
    "except:\n",
    "    print(\"\\n--------model not restored--------\\n\")\n",
    "    pass\n",
    "\n",
    "for i in range(num_epoch):\n",
    "    for j,[image,label] in enumerate(train_loader):\n",
    "        noise = init.normal(torch.FloatTensor(batch_size,1,28,28),0,0.1)\n",
    "        noise = Variable(noise.cuda())\n",
    "        \n",
    "        optimizer.zero_grad()\n",
    "        \n",
    "        image = Variable(image).cuda()\n",
    "        noise_image = image + noise\n",
    "        output = encoder(noise_image)\n",
    "        output = decoder(output)\n",
    "        loss = loss_func(output,image)\n",
    "        \n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "        \n",
    "        if j % 10 ==0:\n",
    "            print(loss)        \n",
    "            torch.save([encoder,decoder],'./model/conv_deno_autoencoder.pkl')\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 6. Check with Train Image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([100, 28, 28])\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nBwT5fAAAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fae4c71dbe0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAADkpJREFUeJzt3X+I1XW+x/HXW1M0NfIHjZZyZzO7sQq5tykuKOGt29IV\nwSSINTCD2llqixUk7mBI/hEosesi/bE0omixt90bu1tCP7QrQSxcNicxzbprrYysMjpKyWZZNvq+\nf8zXZbbm+/mezq/vmXk/HzDMOd/3+Z7vm8O85vs95/M934+5uwDEM6bsBgCUg/ADQRF+ICjCDwRF\n+IGgCD8QFOEHgiL8QFCEHwjqimZuzMw4nRBoMHe3Sh5X057fzO42sz+b2cdm1lXLcwFoLqv23H4z\nGyvpiKS7JB2XtE/SSnf/ILEOe36gwZqx579N0sfuftTdL0j6jaTlNTwfgCaqJfzXSfrrkPvHs2X/\nwMw6zazHzHpq2BaAOmv4B37u3i2pW+KwH2gltez5T0iaM+T+7GwZgBGglvDvkzTPzL5nZuMl/UjS\nrvq0BaDRqj7sd/cBM3tM0m5JYyVtd/fDdesMQENVPdRX1cZ4zw80XFNO8gEwchF+ICjCDwRF+IGg\nCD8QFOEHgiL8QFCEHwiK8ANBEX4gKMIPBEX4gaAIPxAU4QeCIvxAUIQfCIrwA0ERfiAowg8ERfiB\noAg/EBThB4Ii/EBQhB8IivADQRF+ICjCDwRF+IGgCD8QVNVTdEuSmfVK+kzSRUkD7t5Rj6YANF5N\n4c/8m7ufqcPzAGgiDvuBoGoNv0vaY2bvmllnPRoC0By1HvYvdvcTZnaNpDfN7P/c/e2hD8j+KfCP\nAWgx5u71eSKzDZLOufvPE4+pz8YA5HJ3q+RxVR/2m9kkM5ty+bakH0p6v9rnA9BctRz2t0n6g5ld\nfp7/cvc36tIVgIar22F/RRvjsH/Uyf7555owYUJu7fz58/VuB2rCYT+AkY3wA0ERfiAowg8ERfiB\noAg/EFQ9vtWHFnbTTTcl61u2bEnWb7jhhmR9zpw5yfq4ceNyawcOHEiuu27dumS9r68vWS96/ujY\n8wNBEX4gKMIPBEX4gaAIPxAU4QeCIvxAUHyltwW0t7cn6xs3bkzWJ02alFubP39+ct1rr702WR8Y\nGEjWz507l6ynxuJnz56dXHf8+PHJ+smTJ5P1tWvX5tZeffXV5LojGV/pBZBE+IGgCD8QFOEHgiL8\nQFCEHwiK8ANB8X3+Jpg4cWKy/sgjjyTry5YtS9a/+OKL3NqVV16ZXPf06dPJ+ltvvZWsd3V1Jeup\ncf59+/Yl1y26FsHMmTOT9SVLluTWRvM4f6XY8wNBEX4gKMIPBEX4gaAIPxAU4QeCIvxAUIXj/Ga2\nXdIySf3uviBbNk3SbyW1S+qVdJ+7f9q4NlvbzTffnKw//fTTyfqtt96arI8dOzZZ7+/vz6298MIL\nyXV3796drL/33nvJepFU70XXErh06VLVzy1JFy5cSNajq2TPv0PS3d9Y1iVpr7vPk7Q3uw9gBCkM\nv7u/LemTbyxeLmlndnunpHvq3BeABqv2PX+bu18+b/OkpLY69QOgSWo+t9/dPXVtPjPrlNRZ63YA\n1Fe1e/5TZjZLkrLfuZ84uXu3u3e4e0eV2wLQANWGf5ek1dnt1ZJeqU87AJqlMPxm9qKk/5X0z2Z2\n3MwekrRJ0l1m9pGkf8/uAxhBCt/zu/vKnNKdde6lpS1atCi3VjSWPmXKlGT966+/Tta3bt2arK9Z\nsya31sx5GYbz4IMP5tZS8w1I0pgxtZ2DVnSeQHSc4QcERfiBoAg/EBThB4Ii/EBQhB8Iikt3V+jJ\nJ5/MrU2fPj25btFXS7dt25asr1+/PlkvU9Flx5944omGbbtoiPSdd95p2LZHA/b8QFCEHwiK8ANB\nEX4gKMIPBEX4gaAIPxAU4/yZxYsXJ+t33HFHbq1ovPngwYPJ+lNPPZWsN9KCBQuS9ccffzxZX7p0\nabJ+9dVX59YGBgaS69b6ldxTp07VtP5ox54fCIrwA0ERfiAowg8ERfiBoAg/EBThB4JinD8zb968\nZN3McmtF49HPPvtsst7IS0wXjdOnrlMgFV9ee+LEicl6avrwl156KbnuAw88kKwXTdF95syZZD06\n9vxAUIQfCIrwA0ERfiAowg8ERfiBoAg/EFThOL+ZbZe0TFK/uy/Ilm2Q9GNJp7OHrXP31xrVZDPM\nnDmz6nW/+uqrZP3ixYvJ+vXXX5+sX3PNNcn6vffem1t7+OGHk+tecUX6T6Boiu+9e/cm65s2bcqt\nTZs2Lbnuo48+mqx/+eWXyfqMGTNya0ePHk2uG0Ele/4dku4eZvkv3X1h9jOigw9EVBh+d39b0idN\n6AVAE9Xynv8xMztoZtvNbGrdOgLQFNWG/1eS5kpaKKlP0i/yHmhmnWbWY2Y9VW4LQANUFX53P+Xu\nF939kqStkm5LPLbb3TvcvaPaJgHUX1XhN7NZQ+6ukPR+fdoB0CyVDPW9KGmJpBlmdlzSU5KWmNlC\nSS6pV9JPGtgjgAYoDL+7rxxmcXpC+RFo//79yXrqO/eTJ09Orrtjx45kPXWtAEkaN25csj5mTP4B\nXNG1As6fP5+sP/fcc8n6+vXrk/WUFStWJOtF1/Uv+j7/9OnTv3NPkXCGHxAU4QeCIvxAUIQfCIrw\nA0ERfiAoLt2d2b17d7K+ZcuW3Nr999+fXLdouO2qq66qaf1Dhw7l1t54443kuq+9lv5CZk9P487K\n7uvrq2n9CRMmJOup6cdff/31mrY9GrDnB4Ii/EBQhB8IivADQRF+ICjCDwRF+IGgGOevUFdXV1W1\neij6ym/R5bVb1dmzZ5P1okuiF32lt+jS4NGx5weCIvxAUIQfCIrwA0ERfiAowg8ERfiBoBjnHwFG\n6jh+kaLrFBTVi6Y+T13SHOz5gbAIPxAU4QeCIvxAUIQfCIrwA0ERfiCownF+M5sj6XlJbZJcUre7\nbzGzaZJ+K6ldUq+k+9z908a1itHmyJEjyXp/f3+y3tbWlqx/+il/jimV7PkHJK119+9L+ldJPzWz\n70vqkrTX3edJ2pvdBzBCFIbf3fvcfX92+zNJH0q6TtJySTuzh+2UdE+jmgRQf9/pPb+ZtUv6gaQ/\nSWpz98vzLZ3U4NsCACNExef2m9lkSb+TtMbd/zb0unLu7mY27AnoZtYpqbPWRgHUV0V7fjMbp8Hg\n/9rdf58tPmVms7L6LEnDfjrj7t3u3uHuHfVoGEB9FIbfBnfx2yR96O6bh5R2SVqd3V4t6ZX6tweg\nUSo57F8kaZWkQ2Z2IFu2TtImSf9tZg9JOibpvsa0iKiKvrJbdOnuokueR1cYfnf/o6S8V/HO+rYD\noFk4ww8IivADQRF+ICjCDwRF+IGgCD8QFJfuRss6duxYsj537txk/fbbb8+tbdy4saqeRhP2/EBQ\nhB8IivADQRF+ICjCDwRF+IGgCD8QFOP8aFkvv/xysn7nnelvlM+fPz+31t7enly3t7c3WR8N2PMD\nQRF+ICjCDwRF+IGgCD8QFOEHgiL8QFCM86Nl7dmzJ1n//PPPk/WpU6fm1m688cbkuozzAxi1CD8Q\nFOEHgiL8QFCEHwiK8ANBEX4gqMJxfjObI+l5SW2SXFK3u28xsw2SfizpdPbQde7+WqMaRTxFY+2H\nDx9O1m+55Zbc2qpVq5LrFp1jMBpUcpLPgKS17r7fzKZIetfM3sxqv3T3nzeuPQCNUhh+d++T1Jfd\n/szMPpR0XaMbA9BY3+k9v5m1S/qBpD9lix4zs4Nmtt3Mhj2X0sw6zazHzHpq6hRAXVUcfjObLOl3\nkta4+98k/UrSXEkLNXhk8Ivh1nP3bnfvcPeOOvQLoE4qCr+ZjdNg8H/t7r+XJHc/5e4X3f2SpK2S\nbmtcmwDqrTD8ZmaStkn60N03D1k+a8jDVkh6v/7tAWiUSj7tXyRplaRDZnYgW7ZO0kozW6jB4b9e\nST9pSIdAjmeeeSZZ37x5c27t7Nmz9W5nxKnk0/4/SrJhSozpAyMYZ/gBQRF+ICjCDwRF+IGgCD8Q\nFOEHgjJ3b97GzJq3MSAodx9uaP5b2PMDQRF+ICjCDwRF+IGgCD8QFOEHgiL8QFDNnqL7jKRjQ+7P\nyJa1olbtrVX7kuitWvXs7Z8qfWBTT/L51sbNelr12n6t2lur9iXRW7XK6o3DfiAowg8EVXb4u0ve\nfkqr9taqfUn0Vq1Seiv1PT+A8pS95wdQklLCb2Z3m9mfzexjM+sqo4c8ZtZrZofM7EDZU4xl06D1\nm9n7Q5ZNM7M3zeyj7Pew06SV1NsGMzuRvXYHzGxpSb3NMbO3zOwDMztsZj/Llpf62iX6KuV1a/ph\nv5mNlXRE0l2SjkvaJ2mlu3/Q1EZymFmvpA53L31M2Mxul3RO0vPuviBb9oykT9x9U/aPc6q7/2eL\n9LZB0rmyZ27OJpSZNXRmaUn3SHpQJb52ib7uUwmvWxl7/tskfezuR939gqTfSFpeQh8tz93flvTJ\nNxYvl7Qzu71Tg388TZfTW0tw9z5335/d/kzS5ZmlS33tEn2VoozwXyfpr0PuH1drTfntkvaY2btm\n1ll2M8Noy6ZNl6STktrKbGYYhTM3N9M3ZpZumdeumhmv640P/L5tsbv/i6T/kPTT7PC2Jfnge7ZW\nGq6paObmZhlmZum/K/O1q3bG63orI/wnJM0Zcn92tqwluPuJ7He/pD+o9WYfPnV5ktTsd3/J/fxd\nK83cPNzM0mqB166VZrwuI/z7JM0zs++Z2XhJP5K0q4Q+vsXMJmUfxMjMJkn6oVpv9uFdklZnt1dL\neqXEXv5Bq8zcnDeztEp+7Vpuxmt3b/qPpKUa/MT/L5KeLKOHnL6ul/Re9nO47N4kvajBw8CvNfjZ\nyEOSpkvaK+kjSf8jaVoL9faCpEOSDmowaLNK6m2xBg/pD0o6kP0sLfu1S/RVyuvGGX5AUHzgBwRF\n+IGgCD8QFOEHgiL8QFCEHwiK8ANBEX4gqP8HkMZ8nCpYCqsAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fadf25c69e8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fadf185ee80>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngvovGGmJ1bnt6sIAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fadf1411f98>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nwCoAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fadf185e7f0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fadf13c7588>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nzC9Eosj8QiSKzC9Eovw/RQqxvAKbffAAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fadf185ec50>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fadf13c73c8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fadf0e85cf8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAADgJJREFUeJzt3XGIXOW5x/HfE00UbdV4y12DyW3SGsQqXFsXqRik0puQ\nSiSJiERQI8ZulSgtVKha5IoXIV5spH9IcJvGbqSaFlSMpTRJg9SqtbjGqFFvmlU2ZpeY1ERpEogx\n7nP/mJN7t7rnndmZM3PO5vl+YNmZ88yZeTjsb8+ceeec19xdAOKZVHYDAMpB+IGgCD8QFOEHgiL8\nQFCEHwiK8ANBEX4gKMIPBHViJ1/MzPg6IdBm7m6NPK6lPb+ZzTez7WY2YGZ3tvJcADrLmv1uv5md\nIOlvkuZKGpL0iqRr3f3txDrs+YE268Se/2JJA+7+nrsfkbRO0sIWng9AB7US/rMl7Rp1fyhb9k/M\nrMfM+s2sv4XXAlCwtn/g5+69knol3vYDVdLKnn9Y0oxR96dnywBMAK2E/xVJs81slplNkbRE0vpi\n2gLQbk2/7Xf3o2Z2m6QNkk6QtMbd3yqsMwBt1fRQX1MvxjE/0HYd+ZIPgImL8ANBEX4gKMIPBEX4\ngaAIPxAU4QeCIvxAUIQfCIrwA0ERfiAowg8ERfiBoAg/EBThB4Ii/EBQhB8IivADQRF+ICjCDwRF\n+IGgOjpFN1AVkya1tt8bGRkpqJPysOcHgiL8QFCEHwiK8ANBEX4gKMIPBEX4gaBaGuc3s0FJByR9\nJumou3cX0RRQhKlTp+bWHn/88eS6hw8fTtYXL17cVE9VUsSXfC539w8LeB4AHcTbfiCoVsPvkjaa\n2atm1lNEQwA6o9W3/XPcfdjM/lXSJjP7H3d/fvQDsn8K/GMAKqalPb+7D2e/90p6WtLFYzym1927\n+TAQqJamw29mp5rZl4/dljRP0raiGgPQXq287e+S9LSZHXuex939D4V0BaDtmg6/u78n6d8L7AUo\n1KpVq3Jrl112WXLd1atXF91O5TDUBwRF+IGgCD8QFOEHgiL8QFCEHwjK3L1zL2bWuRfDhDd58uRk\n/bHHHkvWFy1alFt79913k+vWGwrct29fsl4md7dGHseeHwiK8ANBEX4gKMIPBEX4gaAIPxAU4QeC\nYpwflXXuuecm66+99lqyfuTIkdxad3f6wlIDAwPJepUxzg8gifADQRF+ICjCDwRF+IGgCD8QFOEH\ngipill5U2JVXXpms15uKetOmTUW2My533HFHsp4ax5eku+66K7c2kcfxi8KeHwiK8ANBEX4gKMIP\nBEX4gaAIPxAU4QeCqjvOb2ZrJC2QtNfdL8iWnSnpN5JmShqUdI27f9S+NpEyd+7c3NoDDzyQXHfl\nypVFt9Ownp6eZH3ZsmXJ+vvvv5+sr1u3btw9RdLInv9XkuZ/btmdkja7+2xJm7P7ACaQuuF39+cl\n7f/c4oWS+rLbfZLyp0YBUEnNHvN3ufvu7PYHkroK6gdAh7T83X5399S1+cysR1L64A5AxzW7599j\nZtMkKfu9N++B7t7r7t3unr5iIoCOajb86yUtzW4vlfRMMe0A6JS64TezJyT9RdK5ZjZkZsskrZA0\n18x2SPqP7D6ACaTuMb+7X5tT+m7BvSDHWWedlaw/+OCDubW+vr7cmiStXr26qZ6KcMkllyTrhw4d\nStavu+66ZP2jj/jqSQrf8AOCIvxAUIQfCIrwA0ERfiAowg8ExaW7K2DevHnJ+j333JOsn3LKKbm1\nhx9+uKmeinLLLbfk1hYtSp8P9tBDDyXrL7zwQlM9oYY9PxAU4QeCIvxAUIQfCIrwA0ERfiAowg8E\nZe65V+Aq/sUSl/s6ni1cuDBZX7t2bbL+8ccfJ+s33nhjbu25555Lrtuqk046KVnfsWNHbu3AgQPJ\ndc8///ymeorO3a2Rx7HnB4Ii/EBQhB8IivADQRF+ICjCDwRF+IGgOJ+/APfff3+yfvvttyfrAwMD\nyfoNN9yQrG/bti1Zb6dbb701WZ8xY0Zu7b777iu6HYwDe34gKMIPBEX4gaAIPxAU4QeCIvxAUIQf\nCKruOL+ZrZG0QNJed78gW3avpO9L+nv2sLvd/fftarIKLr300txa6tr0knTw4MFkffHixcn6zp07\nk/WUeufbp8bhJWnfvn3J+oIFC8bd0zGzZ89O1q+++upk/ejRo8n6J598klvbtWtXct0yvzvRKY3s\n+X8laf4Yyx9y9wuzn+M6+MDxqG743f15Sfs70AuADmrlmP82M3vDzNaY2dTCOgLQEc2Gf5Wkr0u6\nUNJuST/Le6CZ9ZhZv5n1N/laANqgqfC7+x53/8zdRyT9QtLFicf2unu3u3c32ySA4jUVfjObNuru\nYknH/0ejwHGmkaG+JyR9R9JXzGxI0n9K+o6ZXSjJJQ1K+kEbewTQBly3v0FLlizJrT3yyCPJdSdP\nnpysv/jii8n6GWeckayfeGL+//ApU6Yk1506Nf1Z7dDQULJ+3nnnJesp9cbpDx8+nKzX266p7zh8\n+umnyXU3bdqUrF9//fXJ+pEjR5L1duK6/QCSCD8QFOEHgiL8QFCEHwiK8ANBcenuBm3ZsiW3tnXr\n1uS655xzTrI+a9asZH14eDhZ378//7yrekN1qWFCqf7pxvWG45599tncWr2pyesNU5588snJ+mmn\nnZZbO/3005Pr1jvlt8yhvKKw5weCIvxAUIQfCIrwA0ERfiAowg8ERfiBoDiltwMmTUr/jx0ZGelQ\nJ19Ubyx9cHAwWd++fXuyfvnll4+3JbSIU3oBJBF+ICjCDwRF+IGgCD8QFOEHgiL8QFCcz98BZY7j\n17N8+fJkvd5lwx999NEi20EHsecHgiL8QFCEHwiK8ANBEX4gKMIPBEX4gaDqns9vZjMkrZXUJckl\n9br7z83sTEm/kTRT0qCka9z9ozrPFfJ8/ip7/fXXk/VDhw4l6/Wu679nz55x94TWFHk+/1FJP3b3\nb0j6tqTlZvYNSXdK2uzusyVtzu4DmCDqht/dd7v7luz2AUnvSDpb0kJJfdnD+iQtaleTAIo3rmN+\nM5sp6ZuS/iqpy913Z6UPVDssADBBNPzdfjP7kqQnJf3I3f9h9v+HFe7uecfzZtYjqafVRgEUq6E9\nv5lNVi34v3b3p7LFe8xsWlafJmnvWOu6e6+7d7t7dxENAyhG3fBbbRf/S0nvuPvKUaX1kpZmt5dK\neqb49gC0SyNDfXMk/VnSm5KOnZt6t2rH/b+V9G+Sdqo21Jc/V7QY6ivDRRddlKz39/cn6/Pnz0/W\nN2zYMO6e0F6NDvXVPeZ39xck5T3Zd8fTFIDq4Bt+QFCEHwiK8ANBEX4gKMIPBEX4gaCYovs4t3Hj\nxmR9aGgoWb/pppuKbAcdwBTdAJIIPxAU4QeCIvxAUIQfCIrwA0ERfiAopug+DqxYsSK3Nn369OS6\nN998c9HtYIJgzw8ERfiBoAg/EBThB4Ii/EBQhB8IivADQXE+/wQwc+bMZP2ll17KrV111VXJdV9+\n+eVmWkKFcT4/gCTCDwRF+IGgCD8QFOEHgiL8QFCEHwiq7ji/mc2QtFZSlySX1OvuPzezeyV9X9Lf\ns4fe7e6/r/NcjPMDbdboOH8j4Z8maZq7bzGzL0t6VdIiSddIOujuDzbaFOEH2q/R8Ne9ko+775a0\nO7t9wMzekXR2a+0BKNu4jvnNbKakb0r6a7boNjN7w8zWmNnUnHV6zKzfzPpb6hRAoRr+br+ZfUnS\nnyTd7+5PmVmXpA9V+xzgv1Q7NEhO7MbbfqD9CjvmlyQzmyzpd5I2uPvKMeozJf3O3S+o8zyEH2iz\nwk7sMTOT9EtJ74wOfvZB4DGLJW0bb5MAytPIp/1zJP1Z0puSRrLFd0u6VtKFqr3tH5T0g+zDwdRz\nsecH2qzQt/1FIfxA+3E+P4Akwg8ERfiBoAg/EBThB4Ii/EBQhB8IivADQRF+ICjCDwRF+IGgCD8Q\nFOEHgiL8QFB1L+BZsA8l7Rx1/yvZsiqqam9V7Uuit2YV2dtXG31gR8/n/8KLm/W7e3dpDSRUtbeq\n9iXRW7PK6o23/UBQhB8Iquzw95b8+ilV7a2qfUn01qxSeiv1mB9Aecre8wMoSSnhN7P5ZrbdzAbM\n7M4yeshjZoNm9qaZbS17irFsGrS9ZrZt1LIzzWyTme3Ifo85TVpJvd1rZsPZtttqZleU1NsMM3vO\nzN42s7fM7IfZ8lK3XaKvUrZbx9/2m9kJkv4maa6kIUmvSLrW3d/uaCM5zGxQUre7lz4mbGaXSToo\nae2x2ZDM7L8l7Xf3Fdk/zqnu/pOK9Havxjlzc5t6y5tZ+kaVuO2KnPG6CGXs+S+WNODu77n7EUnr\nJC0soY/Kc/fnJe3/3OKFkvqy232q/fF0XE5vleDuu919S3b7gKRjM0uXuu0SfZWijPCfLWnXqPtD\nqtaU3y5po5m9amY9ZTczhq5RMyN9IKmrzGbGUHfm5k763MzSldl2zcx4XTQ+8PuiOe7+LUnfk7Q8\ne3tbSV47ZqvScM0qSV9XbRq33ZJ+VmYz2czST0r6kbv/Y3StzG03Rl+lbLcywj8sacao+9OzZZXg\n7sPZ772SnlbtMKVK9hybJDX7vbfkfv6Pu+9x98/cfUTSL1Titstmln5S0q/d/alscenbbqy+ytpu\nZYT/FUmzzWyWmU2RtETS+hL6+AIzOzX7IEZmdqqkeare7MPrJS3Nbi+V9EyJvfyTqszcnDeztEre\ndpWb8drdO/4j6QrVPvF/V9JPy+ghp6+vSXo9+3mr7N4kPaHa28BPVftsZJmkf5G0WdIOSX+UdGaF\nentMtdmc31AtaNNK6m2Oam/p35C0Nfu5ouxtl+irlO3GN/yAoPjADwiK8ANBEX4gKMIPBEX4gaAI\nPxAU4QeCIvxAUP8L2w2V4cTX6RsAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fadf0ed9b70>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "out_img = torch.squeeze(output.cpu().data)\n",
    "print(out_img.size())\n",
    "\n",
    "for i in range(5):\n",
    "    plt.imshow(torch.squeeze(noise_image.cpu().data[i]).numpy(),cmap='gray')\n",
    "    plt.show()\n",
    "    plt.imshow(out_img[i].numpy(),cmap='gray')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.3486\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.4568\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.4580\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.5828\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.2822\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.3508\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  2.9815\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  1.3384\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  5.1656\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n",
      "Variable containing:\n",
      "1.00000e-03 *\n",
      "  2.7513\n",
      "[torch.cuda.FloatTensor of size 1 (GPU 0)]\n",
      "\n"
     ]
    }
   ],
   "source": [
    "noise = init.normal(torch.FloatTensor(batch_size,1,28,28),0,0.1)\n",
    "noise = Variable(noise.cuda())\n",
    "\n",
    "for i in range(1):\n",
    "    for j,[image,label] in enumerate(test_loader):\n",
    "       \n",
    "        image = Variable(image,volatile=True).cuda()\n",
    "        noise_image = image + noise\n",
    "        output = encoder(noise_image)\n",
    "        output = decoder(output)\n",
    "        loss = loss_func(output,image)\n",
    "        \n",
    "        if j % 10 ==0:\n",
    "            print(loss)        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([100, 28, 28])\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fadf13eb6d8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nt/3FAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fadf0e42cc0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nUWR+IRJF5hciUf4PPWlS0FgIdnsAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fadf0e3d7f0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRfiBoAg/ENT/AWEAyMYJz2FnAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fadf0f06390>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAFytJREFUeJztnWuM1GWWxp/TTYPcpbk0DXITAUVUxBYXxVWjo47iLSZk\nSNywUQdjZhInErPG/bD6YROzYWbUxJjgSgZH1/EyY/SD0XHxtpMgCioXQeRiIzTNHbtbbJpGzn7o\nYrZE/s9purqryn2fX0K6up46VW//qx7+VXXec465O4QQ6VFR6gUIIUqDzC9Eosj8QiSKzC9Eosj8\nQiSKzC9Eosj8QiSKzC9Eosj8QiRKr2I+mJl5RUX2/zd9+/al8Sz2yJEj0WMXpDOOHTtWkM7+LiA+\nLt9//z3VGd999x3V+/TpQ/VevfhLqK2tLVPr3bt3l2MBoLKykups92r//v1p7KFDh6geHfPouLH4\n6Li0t7dnam1tbTh69GinXswFmd/MrgfwOIBKAP/p7o+y21dUVNAX8vTp0+njsYOyfft2Ghs9GVVV\nVVRntLS0UL21tZXq/fr1o/p5551H9W+++YbqjE8++YTqkyZNonp1dTXVt27dmqmNGzeOxm7evJnq\nQ4YMofrRo0cztRkzZtDY6LgcPHiQ6hMmTKD6t99+m6mNGTOGxjY2NmZq69evp7H5dPltv5lVAngS\nwM8BTAUwz8ymdvX+hBDFpZDP/DMBbHb3re5+BMCfANzSPcsSQvQ0hZh/NID899o7ctf9ADNbYGYr\nzWylKgiFKB96/As/d18MYDEAVFZWyv1ClAmFnPkbAOR/M3FG7johxE+AQsz/MYBJZjbBzHoD+AWA\n17tnWUKInsYK+RxuZjcAeAwdqb4l7v7v7Pb9+vXzyZMnZ+osfwnwtNLhw4dpbPR3Ruk4lpZas2YN\njWV/MwA0NTUVpF966aWZ2saNG2lstL/hggsuoPpHH31EdZZijZ6zKP1aU1NDdba/Inpslk4DgNNP\nP53qhew7idbG2LJlC1pbW3s+z+/ubwB4o5D7EEKUBm3vFSJRZH4hEkXmFyJRZH4hEkXmFyJRZH4h\nEqWo9fwAz2+OGDGCxu7atStTi8pDBwwYQPWorn3Dhg2Z2vnnn09j9+zZQ/WoXj8q+V27dm2mFtW8\n19bWUn3ZsmUFxX/99deZWvR8R7nyaP9DtDZGtDb2dwFxqTPbdxI9Z83NzZla1DsiH535hUgUmV+I\nRJH5hUgUmV+IRJH5hUgUmV+IRCl2627agZd1NAV46iYqq41aNUepQlbCGaWkou6+UYvqQtJGURqR\npU8B4JxzzqF61DmYHbeo4/LAgQOpHpVKs/TsaaedRmOjVN4VV1xB9b1791J9586dmdqwYcNoLEvn\nRW3gf3DbTt9SCPH/CplfiESR+YVIFJlfiESR+YVIFJlfiESR+YVIlKLm+aMpvVFOmrWBnjhxIo2N\n8vxbtmyhOpuUG+1PiMqJo5LdqMSTlUlHsWySLcDz0UC8R4Hl6qNJttHrYd26dVRnE4aj5yQimoYb\nvd7YPoNoRPf+/fsztVMZ164zvxCJIvMLkSgyvxCJIvMLkSgyvxCJIvMLkSgyvxCJUlCe38zqAbQA\n+B7AUXevY7eP8vxfffUVfTxWGz548GAau2/fPqpHeV82wjsa733gwAGqR23Do34BrBdB1DY8ytP3\n6lXYVpCZM2dmalGfg1mzZlF9+PDhVD/33HMztUOHDtHYVatWUX3RokVUj44by8dH+xtGjRqVqUXt\nzPPpjk0+V7k7d5YQouzQ234hEqVQ8zuAv5rZKjNb0B0LEkIUh0Lf9s929wYzGwHgbTP7wt0/yL9B\n7j+FBUDcN00IUTwKOvO7e0Pu5x4ArwL40bc77r7Y3evcvS4qWBBCFI8um9/M+pvZwOOXAVwLgJdZ\nCSHKhkLe9tcAeDVXTtoLwH+5+5vdsiohRI/TZfO7+1YAF5xKzJEjR2iv9qh3PsvlR/X40X1HuXpW\ns19VVUVjo+86xo4dS/Wohzzr4x71vo/y0SNHjqT6NddcQ/U5c+ZkatHeisbGRqpHY7RZn4T29nYa\ny/YnAMDcuXOp/s4771CdPS8fffQRjb3ooosytVPZl6FUnxCJIvMLkSgyvxCJIvMLkSgyvxCJIvML\nkShFbd1dWVlJ03UNDQ00npXGRqWrLB0GxOXErHw0SocdPnyY6lHJbgRraR6VrkbpsieeeILqUato\ntrao1Jm1SweA+++/n+psTDYriwWA++67j+oRUZk2a7d+4YUX0thBgwZlalGr9nx05hciUWR+IRJF\n5hciUWR+IRJF5hciUWR+IRJF5hciUYqa5z969CjNx7PW3AAvg4xac0d61AZ62LBhmVpUTjxlyhSq\nR2uLyjRZyW+0B+Gqq66iejTCu6KCnz+2bt2aqUXjv6Nce7S26DllNDc3Uz1qrx3tK2Frj/aFbNq0\nqcux+ejML0SiyPxCJIrML0SiyPxCJIrML0SiyPxCJIrML0SiFDXPb2a0jfX48eNp/P79+zO1aI9A\nNC0oyod//vnnmVpUd/7FF19QPaotd3eqs9bhUXvsm266iepRzX11dTXVWU39Y489RmO/+eYbqg8Z\nMoTqbAT4hx9+SGPvvfdeqkf7G1jb8EiP9hhEr4fOojO/EIki8wuRKDK/EIki8wuRKDK/EIki8wuR\nKDK/EIkS5vnNbAmAOQD2uPu03HXVAF4EMB5APYC57n4wuq+qqipaY83y+AAfdR3lXaP66lWrVlGd\n9Z9fu3YtjT377LOpznLhAHDxxRdTfcmSJZkay3UDwObNm6ke7X+Ixo+zUdXRc7Z7926qX3ABnxDP\nZhbccccdNDZ6Tt577z2qR/X+rO4+6vnPxsVHcxTy6cyZ/w8Arj/hugcBLHP3SQCW5X4XQvyECM3v\n7h8AOHGb1y0AluYuLwVwazevSwjRw3T1M3+NuzfmLu8CUNNN6xFCFImC9/a7u5tZ5mZjM1sAYAHA\nPzcLIYpLV8/8u82sFgByPzO7crr7Ynevc/e6qLhGCFE8umr+1wHMz12eD+C17lmOEKJYhOY3sxcA\nLAcwxcx2mNldAB4F8DMz2wTgmtzvQoifEOFnfneflyFdfaoPZmb0c39Uv82I8tVXXHEF1aMe8Owj\nS7SH4OBBvgXiuuuuo/qtt/JkyrvvvpupRX0OHn/8carffffdVI+O24YNGzK1KI9/9dX8JbZs2TKq\ns14F48aNo7HRrITGxkaqR99vsT4LlZWVNHbo0KGZGpuLcSLa4SdEosj8QiSKzC9Eosj8QiSKzC9E\nosj8QiRKUVt3AzwtFpWHspLfKLUSld1GKTFWRhm1kL7xxhupfumll1I9gqXbXn31VRrLxnsDwMKF\nC7u0puOwdF40Qru+vp7qkydPpjpLoY4dO5bGHjlyhOoTJ06k+rp166jOUo3Ra3Hjxo2ZWpR2zkdn\nfiESReYXIlFkfiESReYXIlFkfiESReYXIlFkfiESpah5/oqKCjqaOGpZPGjQoEwtyhk3NDRQPYpn\nufSoDfSkSZOobmZUHzhwINUfeeSRTC1q5Ry1S29qaqL6hAkTqM72bkTHPMq1R/lw1to7GqG9aNEi\nqm/bto3qbGw6wFvFRyO4mQ+i11I+OvMLkSgyvxCJIvMLkSgyvxCJIvMLkSgyvxCJIvMLkShFzfO3\ntrbSOueonXL//v0ztagufdiwYVSPRnTfe++9mVo0QjtqST5+/HiqP/HEE1RnvQamTJlCY6N8d5RL\nj9pMs30AO3fupLE1NXwE5Lx5WV3lO2hra8vUvvzySxq7Y8cOqketvaPeFGPGjMnUoj0E7LGV5xdC\nhMj8QiSKzC9Eosj8QiSKzC9Eosj8QiSKzC9EooR5fjNbAmAOgD3uPi133cMAfglgb+5mD7n7G+GD\n9epF8+1RDfS+ffsytWhUdHNzM9VHjRpFdZbLZ+O7AT6OGQBeeeUVqr/xBj+0LGf8/vvv09hC89VR\nnp8d1+i43XnnnVS/+eabqb5p06ZM7cUXX6Sx0fjw6DmNRniz1+OZZ55JY5kPToXOnPn/AOD6k1z/\ne3efnvsXGl8IUV6E5nf3DwAcKMJahBBFpJDP/L82szVmtsTM+LwqIUTZ0VXzPwVgIoDpABoB/Dbr\nhma2wMxWmtnK9vb2Lj6cEKK76ZL53X23u3/v7scAPA1gJrntYnevc/e66As9IUTx6JL5zaw279fb\nAPCRpEKIsqMzqb4XAFwJYJiZ7QDwbwCuNLPpABxAPYB7enCNQogeIDS/u5+saPqZrjxYW1sbtmzZ\nkqlHPeZnzZqVqUV9+aPe91deeSXVWY/5aN0vv/wy1devX0/1lpYWqrPvUqI8/eHDh6ke1dRv376d\n6tdee22mxvrqA8DUqVOp/tRTT1F9165dmRrrKwHEfQwKnSnAnpeo/8PIkSMzNba34US0w0+IRJH5\nhUgUmV+IRJH5hUgUmV+IRJH5hUiUorbu7t27N23lzNKAAC9ljEoso5TXJZdcQvU+ffpkalF6JSqr\nZa23AeCyyy6jOktpRWWzUapvyBBetnH77bdTnR3XaHT5ihUrqL506VKqszLviRMn0tjotRi1go9S\noLW1tZnanj17aCwjKm3PR2d+IRJF5hciUWR+IRJF5hciUWR+IRJF5hciUWR+IRKlqHl+M6Otoo8d\nO0bjWWnr5MmTaey0adOoPmjQIKqzcc9PP/00jT1wgPc/jXLG0fjxvXv3ZmozZsygsXfffTfVo9LU\nKCfN/rbVq1fT2AceeIDqI0aMoPp3332XqUWttaPnJHpOBw8eTPXRo0dnamwUPQAcPHgwU4s8lI/O\n/EIkiswvRKLI/EIkiswvRKLI/EIkiswvRKLI/EIkSlHz/EBHrj+L6dOn09hDhw5lal999RWNvece\nPlogqi1/7rnnMjVWTw8AY8eOpXqUm41qtG+66aZM7frrTzZg+f8444wzqN7U1ET1+vp6qj/55JOZ\nGsvDA3F77KgPAjuu0cj2aN8HG4sOxH0S2D6BKJb1rohGpuejM78QiSLzC5EoMr8QiSLzC5EoMr8Q\niSLzC5EoMr8QiRLm+c1sDIBnAdQAcACL3f1xM6sG8CKA8QDqAcx19+xCY3T0kB81alSmHvU6Z7nR\nqC//xx9/TPW+fftS/fLLL8/UotrwOXPmUJ31CgCA2267jer79+/P1LZt20Zj33zzTaq3trZSffny\n5VRnvQaisenR/oZonwCb88D6SgDxHoKorz8bmw4A55xzTqY2ZcoUGvvZZ59latG4+Hw6c+Y/CmCh\nu08F8A8AfmVmUwE8CGCZu08CsCz3uxDiJ0JofndvdPdPcpdbAGwAMBrALQCOj0xZCuDWnlqkEKL7\nOaXP/GY2HsCFAFYAqHH34+93d6HjY4EQ4idCp81vZgMA/BnAb9z9Bxuj3d3R8X3AyeIWmNlKM1sZ\nfbYVQhSPTpnfzKrQYfzn3f0vuat3m1ltTq8FcNJOju6+2N3r3L2ODbsUQhSX0PzWUYb3DIAN7v67\nPOl1APNzl+cDeK37lyeE6Ck6U9J7GYB/ArDWzI7nGB4C8CiAl8zsLgDbAMyN7qi9vT1MizFGjhxJ\n75sRjfAeOnRol9YEAAsXLqR6lLKK3hFFI8DXr1+fqb311ls0NmpBHT02a6cO8BRqlJ6Nymo7Pm1m\nw0p6WcoZiFt3Ry3Lo9cb80FU4s3GrrOS+RMJze/ufwOQdY9Xd/qRhBBlhXb4CZEoMr8QiSLzC5Eo\nMr8QiSLzC5EoMr8QiVL0Ed1VVVWZepRzZrFMA4Dnn3++y/cNALNnz87UWNkqEOfxd+/eTfWXXnqJ\n6qx1OMsJd0avqODnh9raWqqzst2o3Tpr1Q7Ex3XChAmZWlSy269fv4L0aA8CaxUf7a1gaES3ECJE\n5hciUWR+IRJF5hciUWR+IRJF5hciUWR+IRLFonxkd9KnTx9nI6GjGmuWU47ytjt37uSLC2Cjj1kb\nZgDYsWMH1aN2y1E/ADbKOmpvHRGNF4/WztqKR6OoZ86cSXXWxwAAqqurM7XJkyfT2BdeeIHq559/\nPtWjfDur54/2ELBeAytWrEBzc3Onivp15hciUWR+IRJF5hciUWR+IRJF5hciUWR+IRJF5hciUYpa\nz9+nTx9aYx3Vb7OxyM3NzZkawHO+AM+VA3wfQX19PY2NeqlHufhonPSIESMytWiPQZRTjvZHRHsQ\n2B6IqG6d7REA4j4ILNce/V2zZs2ievRabWpqojo77tE8A/Z6UT2/ECJE5hciUWR+IRJF5hciUWR+\nIRJF5hciUWR+IRIlzPOb2RgAzwKoAeAAFrv742b2MIBfAjjetP4hd3+DPlivXhgyZEimHuV1Wb77\n9NNPp7Gst31nYOtua2ujsVEunPU4AOK699bW1kwt2iOwfft2qp933nlUj44r23/B6tIBoKWlherR\nzAD2tw8YMIDGRn0Kolx8tG9k+PDhmVo0Q4K9HqI9Jfl0ZpPPUQAL3f0TMxsIYJWZvZ3Tfu/uizr9\naEKIsiE0v7s3AmjMXW4xsw0ARvf0woQQPcspfeY3s/EALgSwInfVr81sjZktMbOTvi82swVmttLM\nVkZvj4UQxaPT5jezAQD+DOA37t4M4CkAEwFMR8c7g9+eLM7dF7t7nbvXRbPVhBDFo1PmN7MqdBj/\neXf/CwC4+253/97djwF4GgDvtiiEKCtC81vH14fPANjg7r/Luz7/q9bbAKzr/uUJIXqKsHW3mc0G\n8D8A1gI4Xi/4EIB56HjL7wDqAdyT+3Iwk759+/pZZ52Vqbe3t9O1TJ06NVM7ePAgjS10TDYrlYzK\nYgcPHkz1qCS4kJRWlJKKUnXRCO8ojcmIxn/379+f6g0NDVRnZbXRc8Zep0CcKozKtFk5c1R+ztLO\ny5cvR1NTU6fyfZ35tv9vAE52ZzSnL4Qob7TDT4hEkfmFSBSZX4hEkfmFSBSZX4hEkfmFSJSitu4G\neF44Km1lI5mjUsbRo3ktUtT6e9CgQZlaoXsIovHie/fupTrLh0d5/CjXHpUER8ed5dpHjhxJY6N9\nHxdddBHVWav36O/69NNPqR7tvVi9ejXV2T6CaG2sbbhadwshQmR+IRJF5hciUWR+IRJF5hciUWR+\nIRJF5hciUcJ6/m59MLO9ALblXTUMwL6iLeDUKNe1leu6AK2tq3Tn2sa5e3Zf8DyKav4fPbjZSnev\nK9kCCOW6tnJdF6C1dZVSrU1v+4VIFJlfiEQptfkXl/jxGeW6tnJdF6C1dZWSrK2kn/mFEKWj1Gd+\nIUSJKIn5zex6M9toZpvN7MFSrCELM6s3s7Vm9pmZrSzxWpaY2R4zW5d3XbWZvW1mm3I/s/s4F39t\nD5tZQ+7YfWZmN5RobWPM7F0zW29mn5vZfbnrS3rsyLpKctyK/rbfzCoBfAngZwB2APgYwDx3zy7W\nLyJmVg+gzt1LnhM2s38E8C2AZ919Wu66/wBwwN0fzf3HOcTd/6VM1vYwgG9LPbk5N1CmNn+yNIBb\nAfwzSnjsyLrmogTHrRRn/pkANrv7Vnc/AuBPAG4pwTrKHnf/AMCBE66+BcDS3OWl6HjxFJ2MtZUF\n7t7o7p/kLrcAOD5ZuqTHjqyrJJTC/KMBbM/7fQfKa+S3A/irma0yswWlXsxJqMmbjLQLQE0pF3MS\nwsnNxeSEydJlc+y6MvG6u9EXfj9mtrvPAPBzAL/Kvb0tS7zjM1s5pWs6Nbm5WJxksvTfKeWx6+rE\n6+6mFOZvADAm7/czcteVBe7ekPu5B8CrKL/pw7uPD0nN/dxT4vX8nXKa3HyyydIog2NXThOvS2H+\njwFMMrMJZtYbwC8AvF6CdfwIM+uf+yIGZtYfwLUov+nDrwOYn7s8H8BrJVzLDyiXyc1Zk6VR4mNX\ndhOv3b3o/wDcgI5v/LcA+NdSrCFjXWcCWJ3793mp1wbgBXS8DWxHx3cjdwEYCmAZgE0A/htAdRmt\n7Y/omOa8Bh1Gqy3R2maj4y39GgCf5f7dUOpjR9ZVkuOmHX5CJIq+8BMiUWR+IRJF5hciUWR+IRJF\n5hciUWR+IRJF5hciUWR+IRLlfwHCoDzFDxe74gAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fadf188c4e0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nDyO+p0G4zF4AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fadf0e70fd0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nlP8D5oOSmWBZOGkAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fadf187df60>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nMXoAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fadf137bef0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAF1NJREFUeJztnW2M1Ge5xq9bXpdlyzsLLu+UtFAQJFusqZ5IejSt0WA/\n2MgHwzFGWq3m2JhY0xM9/aCmaY41GhsNHpu2Jx5rE2vbmAa0DQZNqrLb8g6Hty6F7S67hRZ2YWEX\n9j4fdjDTus91LzvLzNTn+iVkl7nmmXnmP3Ptf2buN3N3CCHy432V3oAQojLI/EJkiswvRKbI/EJk\niswvRKbI/EJkiswvRKbI/EJkiswvRKaMLuudjR7tY8aMSepjx46l61k2Yl9fX3TfVDczql+8eDGp\nRftmawFg1KhRVH/f+/jf6MuXLye1/v7+ku470qMMUaZHa2tqaqje09NDdUZ039HjZscciF9P7Hkp\n5Tm5cOEC+vr6+J0XKMn8ZnY7gB8BGAXgv939IXb9MWPGYNGiRUl9/vz59P56e3uTWltbG107c+ZM\nqkdP1tGjR5Pa3Llz6dojR45Qfdq0aVQfP3481d9+++2kdu7cuZLue+LEiVSP/ugynT2fALB8+XKq\n79mzh+qM6I/i5MmTqX7mzBmqRycb9oerrq6OrmV7a25upmuLGfbbfjMbBeBRAHcAWAZgvZktG+7t\nCSHKSymf+dcAOOzuR929F8BTANaNzLaEENeaUszfAOB40f9PFC57B2a20cyazKwp+pwkhCgf1/zb\nfnff5O6N7t4YfZEhhCgfpZi/FUDxN11zCpcJId4DlGL+7QCWmNlCMxsL4HMAnh+ZbQkhrjXDDvW5\n+yUz+yqALRgI9T3m7nujdSxmHYV+jh07ltSWLFlC1+7evZvqUUhrypQpSe3EiRN0bX19PdW7u7up\nHj02dtxmzZpF1x4+fJjqM2bMoPq+ffuovmLFiqTW2dlJ17IQJgBMnz6d6iyW39XVRddG4dtob2vW\nrKH6m2++mdR27txJ17Lvzi5dukTXFlNSnN/dXwDwQim3IYSoDErvFSJTZH4hMkXmFyJTZH4hMkXm\nFyJTZH4hMsXKObGntrbWly1LF/5F6b+svjuKu0Zx3XHjxlGdxeqjuvKoLn3evHlUf+ONN6jO+gVM\nmjSppNuOSoKjWDt7Tk+ePEnXRs9JVPrKckqi11pURh0dFxbHB/jeoryQ9vb2pHbkyBH09PQMqZ5f\nZ34hMkXmFyJTZH4hMkXmFyJTZH4hMkXmFyJTytq6u7+/H+fPn0/qUXhl//79Se39738/XRu1zz51\n6hTVGVF7srNnz1I9av0dlc2ykNiFCxfo2qh7LwsrAXHHZVaGHZXNRiGvqByZdeg9ffo0XRuFGZcu\nXUr1qFs0e71Fr6fFixcntai8vBid+YXIFJlfiEyR+YXIFJlfiEyR+YXIFJlfiEyR+YXIlLKW9E6a\nNMlvvfXWpM5yAAA+8TWahBvFlKNps+w43XTTTXTt8ePHqd7R0UH1aO8slh+1cmYj04dCS0sL1VkO\nQ5TfUKo+YcKEpBZN2Y146623qB5N+WUlwVG79NbW9Gyc9vZ29Pb2qqRXCJFG5hciU2R+ITJF5hci\nU2R+ITJF5hciU2R+ITKlpHp+M2sB0AXgMoBL7t7Irn/x4kUcOnQoqUc10Cym3NDQQNdGrb2jfgCv\nv/56UotivqXG2g8cOEB1FlOO+hhMnTqV6rW1tSXpN998c1JjxxSI24qz9tcAsHbt2qQWtdbesmUL\n1VevXk31qI9CW1tbUotawbO9l21Ed4G17s6PpBCi6tDbfiEypVTzO4Dfm1mzmW0ciQ0JIcpDqW/7\nP+LurWY2E8AfzOyAu28rvkLhj8JGABg9uqwtA4UQhJLO/O7eWvjZAeC3ANYMcp1N7t7o7o3RfDQh\nRPkYtvnNrNbM6q78DuATAPaM1MaEENeWUt6H1wP4rZlduZ3/dffNI7IrIcQ1Z9jmd/ejAFYOY11S\ni3qlL1q0KKlF3ydEtd9Rb31W7x/lEESx171791J93bp1w16/YsUKupb1tgfiHIbouPf29ia1KVOm\n0LX33HMP1ZcvX071G2+8Mande++9dO2sWbOoHvUxYI8bAAonzUGJRpevXJm23e7du+naYhTqEyJT\nZH4hMkXmFyJTZH4hMkXmFyJTZH4hMqXs+bYsNNTZ2TnstVHr7Yiamhqqf+ADH0hqbHQ4EJcbR2W1\nUfkpC9dF7dCjYx6Ni47KsNlji8JtUTly1B77zjvvTGpdXV10bVQuHJXdLliwgOqsjDs65qwsPiol\nLkZnfiEyReYXIlNkfiEyReYXIlNkfiEyReYXIlNkfiEypawjumtra33ZsmVJPYqNslh+e3s7Xfva\na69RnZULAzwPIIqVjx8/nuoRUZx/2rRpSS1qCx7ddpSjcN9991GdldWyEdpA3BY8KqVmeSHRSPeo\nJPfZZ5+l+tatW6nOjjs7ZgAfL3748GH09PRoRLcQIo3ML0SmyPxCZIrML0SmyPxCZIrML0SmyPxC\nZEpZ6/nNjLbnPnfuHF3PRjZHa6M4fpQnMGfOnKQWjfeOYu3d3d1Uj1pcs3HPUW14dFx+8pOfUJ3l\nGABxS3RG1Db88ccfpzrLA4jyWx555BGqRzkI27dvpzrL/Th48CBdO3/+/KQW9SF4x3WHfE0hxD8V\nMr8QmSLzC5EpMr8QmSLzC5EpMr8QmSLzC5EpYZzfzB4D8CkAHe6+vHDZVAC/BrAAQAuAu9ydB2UH\n1mHUqFFJPYrVX3/99Ukt6lcexfEj2JjtaN9RrD0aHx7FbqdPn57UouPyta99jepRX/5onPTmzZuT\nWlRTv3PnTqq3trZSnc0MuO222+jaKPeiqamJ6tFzfurUqaQW9SmYNGnSsO+3mKGc+R8HcPu7LvsW\ngJfcfQmAlwr/F0K8hwjN7+7bAJx+18XrADxR+P0JAJ8Z4X0JIa4xw/3MX+/uV3JK2wHUj9B+hBBl\nouQv/HwgSTqZKG1mG82sycyaSp2nJ4QYOYZr/pNmNhsACj87Uld0903u3ujujVGBixCifAzX/M8D\n2FD4fQOA50ZmO0KIchGa38x+BeBlADeY2Qkz+yKAhwB83MwOAfjXwv+FEO8hwji/u69PSDxQOgi9\nvb04duxYUo9qy5ubm5ParFmz6FqWXwDE9dks7nv+/Hm6lsXhAd6nYCjU16e/b2XzBgDg1VdfpXpH\nR/ITHQDge9/7HtXnzp2b1KKYdPQxMeqjwPrbb9iwIakBwOTJk6ke5SCwvBCAP2fRnAc23yLKuyhG\nGX5CZIrML0SmyPxCZIrML0SmyPxCZIrML0SmlLV1NzBQ1ptix44ddO2HPvShpHb48GG6Ngr1lZJ9\nGIXyotbb/f39VI9KellYKQqnRaOkozDm7NmzqX7x4sWkFj3uqLQ1au396U9/OqlFqeZRyW4Unu3p\n6aE6CyVGt/36668ntWi0eDE68wuRKTK/EJki8wuRKTK/EJki8wuRKTK/EJki8wuRKRaNKh5JJkyY\n4DfccENSZxrAY/lRi+mohXUUz2bx8qgEM4pHR7H4++67j+osRyGKpS9cuJDqL774ItWjPIGopJgR\nxaxZDgEAPPPMM0ktiqV/85vfpDqLtQNx7gfLYYjKiY8ePZrUzpw5g0uXLqWTaYrQmV+ITJH5hcgU\nmV+ITJH5hcgUmV+ITJH5hcgUmV+ITClrPX9/fz9tgb19+3a6ntXFRy2LT5w4QfUoLrty5cqkFvUK\nuOOOO6h+9913Uz3qNcByNVhMGACmTZtG9RkzZlA9yp947rn0PJeo9XZdXR3Vv/vd71Kd9TlgI7KH\nct9Rbkf0mmD5F1EvgGXLliW1qCdGMTrzC5EpMr8QmSLzC5EpMr8QmSLzC5EpMr8QmSLzC5EpYZzf\nzB4D8CkAHe6+vHDZgwC+BKCzcLUH3P2FIdwWxo4dm9S7urroehZ7jeLZH/7wh6m+e/duqk+dOjWp\nffvb36Zro9rxZ599luqbN2+mOhujHdW8RzHlp59+murf+c53qM7i4aNH85ffPffcQ/WoD8KPf/zj\npLZr1y66NqrXv+6666gevZ7YDIooh4ARzXh4x3WHcJ3HAdw+yOU/dPdVhX+h8YUQ1UVofnffBuB0\nGfYihCgjpXzm/6qZ7TKzx8yMz6MSQlQdwzX/TwEsBrAKQBuAH6SuaGYbzazJzJqiz2hCiPIxLPO7\n+0l3v+zu/QB+DmANue4md29098ao2EEIUT6GZX4zKx7NeieAPSOzHSFEuRhKqO9XAD4GYLqZnQDw\nnwA+ZmarADiAFgC8JlUIUXWUtW//lClT/Lbbbkvqr732Gl1/7NixpNbQ0EDXRnHZKA/gK1/5SlKL\natq/8IUvUD2aQx/d/oIFC6jOmDVrFtXXr19PdVZbDgDz5s1Laiw/AQBaWlqo/uijj1Kd1cwfOHBg\n2GsBYMmSJVSPclaOHz+e1FgfAoA/Zzt27EB3d7f69gsh0sj8QmSKzC9Epsj8QmSKzC9Epsj8QmRK\nWVt39/X10RbaZ86coevZOGlWKgzEoZn777+f6n/605+S2pNPPknXjhs3jupR++yIaJQ1I2phzdql\nA8CiRYuovm/fvqve0xU2bdpE9ShcxzJKo/BoVNIbhSmj1zJrW97X10fXstDv1YTudeYXIlNkfiEy\nReYXIlNkfiEyReYXIlNkfiEyReYXIlPKPqKbtZKO2nyxGGYU57/llluo3tnZSXXWPpuVZwLx+O/o\ncZvxCk1W4rl8+XK69stf/jLVo+MSxfHPnz+f1L7//e/TtSy3AuDt1AFg8uTJSe3gwYN07dq1a6ne\n3NxM9agtOWuZ3tbWRteynJWr6ZalM78QmSLzC5EpMr8QmSLzC5EpMr8QmSLzC5EpMr8QmVLWOH/E\n4sWLqc7aIc+YMYOuLSUmDPBeAtFY5O7ubqp/9rOfpfrSpUupznokrFq1iq7dtm3bsG8bAD760Y9S\nfc2a5DAnmgMAxLkb9fX1VD937lxSi3oovPrqq1SfMGEC1aM8ghtvvDGprV69mq5l48WvpreDzvxC\nZIrML0SmyPxCZIrML0SmyPxCZIrML0SmyPxCZEoY5zezuQCeBFAPwAFscvcfmdlUAL8GsABAC4C7\n3P0tdlvuTuOQdXV1dC9sZHM03juqz2b11QBw1113JbUozh/FhFmPAwD4y1/+QnXWL+Dhhx+ma7du\n3Ur16LhEMenf/e53SS2K80+cOJHqUW981ucgyl+InrNoBPdNN91Eddabf+/evXRtTU1NUot6PxQz\nlDP/JQDfcPdlAG4BcK+ZLQPwLQAvufsSAC8V/i+EeI8Qmt/d29z9lcLvXQD2A2gAsA7AE4WrPQHg\nM9dqk0KIkeeqPvOb2QIAHwTwVwD17n6l31A7Bj4WCCHeIwzZ/GY2EcBvAHzd3c8Waz7QXG/QBntm\nttHMmsysKepVJ4QoH0Myv5mNwYDxf+nuzxQuPmlmswv6bACDTi50903u3ujujVfTXFAIcW0JzW8D\nXx/+AsB+d3+kSHoewIbC7xsAPDfy2xNCXCuGUtJ7K4DPA9htZjsKlz0A4CEAT5vZFwEcA5COhRXo\n7++n4Z2oTTQr0WxoaKBrn3rqKar/8Y9/pDrbW3Tf0ced/fv3U338+PFUf/vtt5NaVPZ63XXXUT0i\nag3OQmJRqfKYMWOoHr2T3LlzZ1KbP38+XTtnzhyqt7e3Uz06ri+//HJSi0rbWTlyNFq8mND87v5n\nAKng4W1DvichRFWhDD8hMkXmFyJTZH4hMkXmFyJTZH4hMkXmFyJTyt66m5W/njx5kq4tJSZdW1tL\n9QMHDlD97NmzSS0aqXzp0iWql1oSzHInottmjwuI9/6zn/2M6izOf+jQIbqWlb0C8Wj0mTNnJrXo\ncUWjx1lZLcBzLwCewxCNdGd5IRcuXKBri9GZX4hMkfmFyBSZX4hMkfmFyBSZX4hMkfmFyBSZX4hM\nsYEOXOWhpqbGFy1alNSjVs5szPbo0Txl4ciRI1SP6ruZHsWbo3p81pIciMeHs7r2qCY+yn+IRk1H\nI8BZDsSpU6fo2mhvUT0/06P8hqidejQSvpS25FGOAHs9tbS0oKenZ0j9u3XmFyJTZH4hMkXmFyJT\nZH4hMkXmFyJTZH4hMkXmFyJTyhrnnzhxoq9YsSKpR7HX/v7+pMbGMQNAR8egA4X+zty5c6nORh+X\nWtsdjZqO8gTYY4uOS0Q0ojvirbfSU9vnzZtH10ZjsFtbW6k+adKkpBbF6aPnLOoHsHDhQqpv2bIl\nqUXj5Nl9Nzc3o6urS3F+IUQamV+ITJH5hcgUmV+ITJH5hcgUmV+ITJH5hciUMM5vZnMBPAmgHoAD\n2OTuPzKzBwF8CcCVwfUPuPsL7LZqamr8+uuvT+qdnZ1JDeD9zFmPdgC4fPky1RsaGqjO5rFH8WgW\nbwbi/vRRbTjrzR/Neo9657/55ptUj/II2PMSxemj52zOnDlUZ6+n7u5uujbqoXDu3DmqR7DnPLpt\ndlza29vR29s7pDj/UIZ2XALwDXd/xczqADSb2R8K2g/d/b+GckdCiOoiNL+7twFoK/zeZWb7AfDT\npBCi6rmqz/xmtgDABwH8tXDRV81sl5k9ZmZTEms2mlmTmTVFb+OEEOVjyOY3s4kAfgPg6+5+FsBP\nASwGsAoD7wx+MNg6d9/k7o3u3hj1XBNClI8hmd/MxmDA+L9092cAwN1Puvtld+8H8HMAa67dNoUQ\nI01ofhsoZ/sFgP3u/kjR5bOLrnYngD0jvz0hxLViKN/23wrg8wB2m9mOwmUPAFhvZqswEP5rAXB3\ndEN9fX144403knoUumEhLVbuC8ShnQj2kSX6OBOFbnp7e4e1pyuwEs+//e1vdG0Uylu9ejXVozDk\nuHHjqM6IQqh1dXVUZ4+NlWgDcUkuC/0C8d7ZKO3omK5cuTKpRaXIxQzl2/4/AxjsSNGYvhCiulGG\nnxCZIvMLkSkyvxCZIvMLkSkyvxCZIvMLkSlDifOPGOPGjcOCBQuSehTn37MnnUcUjXOOxkG/8sor\nVGe3H5WmLl26lOoRLCYM8LbjUcnuzTffTPUoPyIqCWf5F1GOQRRr37VrF9XZGOwo/yAaTR6NZY9K\nxFlL9AkTJtC1LN8lyl94x+0M+ZpCiH8qZH4hMkXmFyJTZH4hMkXmFyJTZH4hMkXmFyJTyjqi28w6\nARwrumg6AB7srRzVurdq3RegvQ2XkdzbfHfn88cLlNX8/3DnZk3u3lixDRCqdW/Vui9Aexsuldqb\n3vYLkSkyvxCZUmnzb6rw/TOqdW/Vui9AexsuFdlbRT/zCyEqR6XP/EKIClER85vZ7Wb2f2Z22My+\nVYk9pDCzFjPbbWY7zKypwnt5zMw6zGxP0WVTzewPZnao8HPQMWkV2tuDZtZaOHY7zOyTFdrbXDPb\namb7zGyvmf174fKKHjuyr4oct7K/7TezUQAOAvg4gBMAtgNY7+77yrqRBGbWAqDR3SseEzazfwHQ\nDeBJd19euOxhAKfd/aHCH84p7n5/leztQQDdlZ7cXBgoM7t4sjSAzwD4N1Tw2JF93YUKHLdKnPnX\nADjs7kfdvRfAUwDWVWAfVY+7bwNw+l0XrwPwROH3JzDw4ik7ib1VBe7e5u6vFH7vAnBlsnRFjx3Z\nV0WohPkbABS3QTmB6hr57QB+b2bNZrax0psZhPrC2HQAaAdQX8nNDEI4ubmcvGuydNUcu+FMvB5p\n9IXfP/IRd18N4A4A9xbe3lYlPvCZrZrCNUOa3FwuBpks/XcqeeyGO/F6pKmE+VsBFDedm1O4rCpw\n99bCzw4Av0X1TR8+eWVIauFnR4X383eqaXLzYJOlUQXHrpomXlfC/NsBLDGzhWY2FsDnADxfgX38\nA2ZWW/giBmZWC+ATqL7pw88D2FD4fQOA5yq4l3dQLZObU5OlUeFjV3UTr9297P8AfBID3/gfAfAf\nldhDYl+LAOws/Ntb6b0B+BUG3gb2YeC7kS8CmAbgJQCHALwIYGoV7e1/AOwGsAsDRptdob19BANv\n6XcB2FH498lKHzuyr4ocN2X4CZEp+sJPiEyR+YXIFJlfiEyR+YXIFJlfiEyR+YXIFJlfiEyR+YXI\nlP8HM8vKmdGgMhgAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fadf0cdcda0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fadf0d09b00>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "out_img = torch.squeeze(output.cpu().data)\n",
    "print(out_img.size())\n",
    "\n",
    "for i in range(5):\n",
    "    plt.imshow(torch.squeeze(noise_image.cpu().data[i]).numpy(),cmap='gray')\n",
    "    plt.show()\n",
    "    plt.imshow(out_img[i].numpy(),cmap='gray')\n",
    "    plt.show()"
   ]
  }
 ],
 "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": "3.6.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}