models/AE_VAEs/notebooks/denoising_AE.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Denosing Autoencoder\n",
    "```\n",
    "Denosing autoencoder is a type of autoencoder that is trained to reconstruct the original input from a corrupted version of it by first corrupting the input and then learn to recover the original input.\n",
    "```"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# necessary imports\n",
    "# add ../src to sys.path\n",
    "import sys\n",
    "sys.path.append('../src')\n",
    "\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F\n",
    "import torch.optim as optim\n",
    "import torchvision\n",
    "import torchvision.transforms as transforms\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "\n",
    "from tqdm import tqdm, trange\n",
    "\n",
    "import AE as ae"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# load MNIST data\n",
    "transform = transforms.Compose([transforms.ToTensor()])\n",
    "\n",
    "# datasets\n",
    "trainset = torchvision.datasets.MNIST(root='../../data', train=True, download=True, transform=transform)\n",
    "testset = torchvision.datasets.MNIST(root='../../data', train=False, download=True, transform=transform)\n",
    "\n",
    "\n",
    "# data loaders\n",
    "batch_size = 128\n",
    "trainloader = torch.utils.data.DataLoader(trainset, batch_size=batch_size, shuffle=True)\n",
    "testloader = torch.utils.data.DataLoader(testset, batch_size=batch_size, shuffle=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 500x500 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# visualize some data\n",
    "def visualize_data(data, labels):\n",
    "    fig, axs = plt.subplots(2,2, figsize=(5,5))\n",
    "    axs = axs.flatten()\n",
    "    for i in range(4):\n",
    "        axs[i].imshow(data[i].squeeze(), cmap='gray')\n",
    "        axs[i].set_title(labels[i].item())\n",
    "        axs[i].axis('off')\n",
    "    plt.tight_layout()\n",
    "    plt.show()\n",
    "\n",
    "# visualize some data\n",
    "imgs, labels = next(iter(trainloader))\n",
    "visualize_data(imgs, labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# corrupt the data with gaussian noise\n",
    "def gaussian_noise(data, mean=0, std=0.1):\n",
    "    noise = torch.randn(data.size()) * std + mean\n",
    "    return data + noise"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAdMAAAHqCAYAAABfi6TIAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy81sbWrAAAACXBIWXMAAA9hAAAPYQGoP6dpAAA640lEQVR4nO3deXBe5Xn+8UteJMuWZVu25X3f8YINwWIxGBKDaSjLUMAxadpCUwKhndCEDmNmSGegkCnJtJnQFEIIJtM0HRrSJqXtTCiLKY6JF4wXvFtetXlfJC+SZfn3F0xIcl/Pi57+LHnm+/lT95xHR+c959y8yXP5Ljp37tw5AQCAduvS0ScAAMCFjmYKAEAmmikAAJlopgAAZKKZAgCQiWYKAEAmmikAAJlopgAAZKKZAgCQiWYKAEAmmukFqKmpSX/913+tG2+8URUVFSoqKtJLL73U0acF4BNavXq1brnlFlVUVKhnz56aNm2avvOd73T0aaEdunX0CeCTO3jwoB5//HGNHDlSF198sZYsWdLRpwTgE3rttdd08803a9asWXrsscdUVlam6upq1dTUdPSpoR1ophegIUOGqL6+XoMHD9aqVat02WWXdfQpAfgEjh8/rj/6oz/STTfdpFdeeUVduvA/El7o+AQvQCUlJRo8eHBHnwaAdvrxj3+sffv26cknn1SXLl104sQJtbW1dfRpIQPNFADOs9dff13l5eWqra3VpEmTVFZWpvLycj3wwAM6ffp0R58e2oFmCgDn2bZt29Ta2qpbb71V8+fP109/+lPde++9eu6553TPPfd09OmhHfj/TAHgPGtqatLJkyd1//33f7R79/bbb1dLS4u+973v6fHHH9eECRM6+CzxSfDNFADOs9LSUknSwoULP/bzu+++W5L07rvvnvdzQh6aKQCcZ0OHDpUkDRo06GM/r6yslCQdOXLkvJ8T8tBMAeA8u/TSSyVJtbW1H/t5XV2dJGngwIHn/ZyQh2YKAOfZXXfdJUn6wQ9+8LGfv/DCC+rWrZuuvfbaDjgr5GAD0gXqH/7hH3T06NGP/kv21Vdf/ehfTvmLv/gL9enTpyNPD4Axa9Ys3XvvvXrxxRfV2tqquXPnasmSJfrJT36iRYsWffQ/A+PCUXTu3LlzHX0S+ORGjx6t3bt3/87azp07NXr06PN7QgA+kTNnzuipp57S4sWLVVdXp1GjRunBBx/UQw891NGnhnagmQIAkIn/zxQAgEw0UwAAMtFMAQDIRDMFACATzRQAgEw0UwAAMtFMAQDIVPC/gDR58uSw9pv/WPNvOnz4cFhzMdfevXvbdT+cvPBJufORpO7du4e1Ll38f3+4emtra1hL/Ysn7pyPHz8e1o4ePWrXHTx4cFhraWkJayUlJXbdbdu2hbWLLroorDU1Ndl1Dxw4ENZOnTplj62oqAhrxcXFYa2srMyuO3bs2LD2L//yL/bY8+2yyy4La6nnrVevXmGtvr4+rBUVFdl1u3WLX0PuH3wfPny4XbexsTGspZ7jgwcPtuvYAQMG2HXdM+XOd9++fXZd937u27dvWEt9Nu45du+d6667zq77m/8m8a9z71/J3xPufZZ6P7jP9c0337THSnwzBQAgG80UAIBMNFMAADLRTAEAyEQzBQAgE80UAIBMBUdjXPzlxIkT9tiBAweGNbdF2m1Pl/w26DNnzoS16dOn23VffvnlsHbllVfaY3v06BHWNm7cGNZSf+uYMWNsPZKaa/rhcPHfxV3D1JZ6tw3dRYSGDBli112/fn1Yu/jii+2xDQ0NYa2trS2suc9U8p9rZ9O1a9ewdvLkSXusi7G52IyLV0jSyJEjw5qLzWzYsMGu694PqWhXz549w9qoUaPCWirC4p4p955M3YOnT58Oay5y42qSv06XXHJJWNu/f79dd9y4cWFt8+bN9lgXn3MRrVT0KxUjTOGbKQAAmWimAABkopkCAJCJZgoAQCaaKQAAmWimAABkopkCAJCp4Jypy6e5fJTk8zsuz5XKQLkM2o4dO8Jac3OzXXfYsGFhbdeuXfbYSZMmhTWXoXQ5PUk6e/ZsWHNZ0b1799p1Z8+eHdZcftiNkpKk8vLysNavX7+wlpNPS43VmjhxYlhzn2sqRz1jxgxb70zcOK7USEOX4XOZTpcjlfzz6PKVU6ZMses6qXeL+73unZUaIejGxrl7331ukrR169aw1qdPn7A2fvx4u+6f//mfh7V33nknrL366qt2XTe+LXUfuuvvRlmmctSpd3AK30wBAMhEMwUAIBPNFACATDRTAAAy0UwBAMhEMwUAIFPB0ZgjR460qyZJ3bt3D2tu9FVqZE5NTU1Yc9vxDxw4YNetqKgIa+58JWnPnj1hzUU3Dh8+bNd10QO3jT81gq26ujqsuW3zH3zwgV3XnZO7X1wES5IqKyvDmrv2kv/cXXwgtaXexRI6GzciLBW/cHU3Aiz1vI0dOzasuc80NZZw+/btYW3EiBH2WBfdcHGz1HPsxre591lZWZld18W+3LP4d3/3d3Zdd++70YPuOU1x/ULyz5v7vW4spJR+t6fwzRQAgEw0UwAAMtFMAQDIRDMFACATzRQAgEw0UwAAMhUcjTlz5kxYu/jii+2xbqvz7t27w1pOXMRFAKZOnWrXdefkJtWkfq+bLpKaZOO46RmpiR07d+4MawcPHgxrxcXFdl03eaekpCSsrV692q7r7sOqqip7rLsP165dG9ZS00lS90Rn4uJZbrqT5D83F39J3SsuhuImUqUmOLlYR2trqz3WPY/u3j506JBd1/1eF0VLTVNyz+rDDz8c1tw7VPIROHdObsqV5KNU7j6TpJkzZ4Y1Fy9K3YcuQlQIvpkCAJCJZgoAQCaaKQAAmWimAABkopkCAJCJZgoAQCaaKQAAmQoOyLnxS6lsVa9evcJaKvvj1NXVhTWXRUqNdhs6dGhYS435+upXvxrW7rjjjrDW0NBg13388cfDWlFRUVhbvny5XffTn/50WHPZS5dXlHze1t0PqayXuw9d1k7yY9bcZ57KALuRXJ2NG+U1aNAge6zLBrqsbWqk1pIlS8La5MmTw1rqfK+++uqw9vOf/9weO3z48LC2adOmsJYa7dbY2BjWXM63vLzcruvy5NOmTQtr7jmVpG984xth7dixY2GtZ8+edt2mpqaw5vK2krRq1aqw5kbc5WRfC8E3UwAAMtFMAQDIRDMFACATzRQAgEw0UwAAMtFMAQDIVHA0xo0O2rZtmz3WbW9345dS28FdJOHIkSNhbfPmzXbdnG3bf/AHfxDWXCzBxTYk6bbbbgtrX//61+2xjhtZ5GIoqW3mLuLiYkCpyI27TufOnbPHupFc7nxTI6FSEY3OxD1vrib5cV0uHpeKO7U3/pKKQrnxYRUVFfZYF1MZN25cWHOj6CTp1KlTYc3Fi1J/68KFC8Oae2e9/fbbdl3397j7wY3Ok3x0JnUfunvCvQNSEa3UOafwzRQAgEw0UwAAMtFMAQDIRDMFACATzRQAgEw0UwAAMhUcjXFbul2EQvJbnd10l+3bt9t13fQXt6U7FXVw53v48GF77DPPPBPW/uZv/iasTZgwwa47YMCAsPb3f//3Yc3FcSS/ldzFX7Zu3WrXdZNh3GSHtrY2u66b7ODuUcnHN9zUDRepkdL3U2fi4gypqRnus+nfv39Ymzhxol13//79Yc1FKFIxFHfvnzlzxh7rrtOKFSvC2pQpU+y6GzZsCGtjxowJa/Pnz7fruuicm/707LPP2nXdOfXr1y+srVu3zq7rooupd5Z7Vl1M072TJP+OLQTfTAEAyEQzBQAgE80UAIBMNFMAADLRTAEAyEQzBQAgE80UAIBMBedM3Zgkl7uSpPr6+rDmRmrt2rXLrusyktOmTbPHOi432NLSYo9dtmxZu2pXXXWVXXfEiBFhbcGCBWHt5z//uV3XjUNzea7Zs2fbdd3IKJfjdSPhJP+5pnJkbgRTaWlpWDtx4oRdN3d00/nksqRu7Jjkn4vRo0eHtU2bNtl13Ttg586dYW3o0KF2XZdfraystMe6c3ajIV2WWZKGDx8e1tx1ePDBB+267nn77ne/G9ZS4/GWL18e1iZNmhTWUuMzXYbdXSPJv7OKiorCmstCSz77Wgi+mQIAkIlmCgBAJpopAACZaKYAAGSimQIAkIlmCgBApoKjMe+8805YS0US3FgiN8orNZbMbeVvamoKa3V1dXbd4uLisJbaPt21a9ewtmTJkrDmtplLPiZ06aWXhrXUqLQhQ4aENRctSG2pb21tDWtuxJKL40h+LF9qO/6wYcPCmos0uFF/krR3715b70zc5+biFZKPxriog4vNSNLKlSvDmrv2Luok+fFtqfvXRaXc+LbUGDv3XDzyyCNhLTVu7tFHHw1r3brFr/nUu9t95u0dnSf5d/v69evtse6d1djYGNZSo0Ld+LZC8M0UAIBMNFMAADLRTAEAyEQzBQAgE80UAIBMNFMAADLRTAEAyFRwznTMmDFhzeWNJJ+9mjp1alhzo9sknxVz55QaqeWybansq+MyZm50kOT/Vjee7a233rLruqyeG3GVOl+nuro6rLl8n+THM7mMmSRt3LgxrO3bty+suXtUSo8Y60w++OCDsJbK4bkRVi5/nRpLduWVV4Y1l5NO5cUHDx4c1lweXPLPqstmpvKrV199dVibN29eWPv+979v13Xvyra2trCW+szHjRvXrnXdeEzJ3y9jx461x7r3t8uSp+4XNwKvEHwzBQAgE80UAIBMNFMAADLRTAEAyEQzBQAgE80UAIBMBUdj3LZiN2JN8uODtmzZEtZSW6RPnToV1tyW70GDBtl1XcQiteW7S5f4v0/cdnC3zVySWlpawpobN/eXf/mXdt1/+qd/Cmup+IDjxj65EUru+kk+tuRG8kk+3uUiN25knCTNmDHD1jsTF+3auXOnPdZdIxdXSEVjXKzD3UepWIcbjefGLEo+GuNqd999t133jjvuCGvf+ta3wtrPfvYzu25FRUVYc+87N05O8s+U6wmpEWzHjh0La1VVVfbYHTt2hLWBAweGtebmZrtu6lqk8M0UAIBMNFMAADLRTAEAyEQzBQAgE80UAIBMNFMAADIVHI1xW69d5EPyW8ld/CUVdTh+/HhYc9MO3LZsSSovLw9rLuYj+e3XtbW1Yc3FfCQ/3aVfv35hLRV3cNEDFx9IbX13ExhcjGLAgAF23ffff7/dx7r7ycVfUvehi+t0Ni56lPpM3XPu4k6pqTouurFr166w5t4rkp8ak5owVFlZGdbmzJkT1mbOnGnXdc+ji7+4qUaSNGnSpLDW0NAQ1ty7Tmr/1JhU1M9NIEp9Ni4S5d6xrodJ/hoWgm+mAABkopkCAJCJZgoAQCaaKQAAmWimAABkopkCAJCJZgoAQKaCc6aHDx8Oa6n8jtOrV6+wlsrvuWymyyoVFRXZdU+fPh3WDh48aI8dP358WHP51hUrVth1XbbNZVBTmS2X+XQ5PpfDk/zfumfPnnb9TslnSVOj9dx1cmPsUvdLKsvbmbh839mzZ+2x7t53Y9SmT59u13Wfizun7du323XdvVJaWmqPddnMG264Iaylnotnn302rLkcb2ocpbtHXQ7d5cwlacOGDWHN5T3dMy7592TqeerRo0dYcyMGU6MAXaa5EHwzBQAgE80UAIBMNFMAADLRTAEAyEQzBQAgE80UAIBMBUdjWlpawlpqm7kb8+PiLalxZ24buotYpEZquTFqbtSU5KMmbsRV6pzcOCn3t37lK1+x6y5cuDCsudjS/v377brunFxUIrVV33FxBslfw5MnT4Y1d+9LfkxVZ+NiX6nr5z5TN3IvJzo0YsSIsJZ6Fl10w0X9JGnBggVhzcWLFi9ebNdds2ZNWPva174W1lzkQ5L27t0b1tz4ts2bN9t1XRzKRW7cKDTJR56GDRtmjy0pKQlrLuLm3mdSuo+l8M0UAIBMNFMAADLRTAEAyEQzBQAgE80UAIBMNFMAADIVnENwMRQXJZH8v9bvtq+ntuq7LdQuYlFcXGzXnThxYlhz01Akv+W7ra0trO3evduu6+IZbopCamqMm97gPtfU9JxZs2aFNReHSk1ucLGE1JShM2fOhDUXwdiyZYtdd+vWrbbemdTV1YU19yxKfhKI+1xSU3fcc+6iDiluEtOMGTPssddcc01Yc9chNTXmscceC2vDhw8Pa8ePH7frlpWVhbUbb7wxrKUm7zz99NNhzT0X7p0kSdXV1WHNRQglH8lxE4hS76z169fbegrfTAEAyEQzBQAgE80UAIBMNFMAADLRTAEAyEQzBQAgE80UAIBMBedMXd4rNTKnvr4+rLn8ZCoL5jJzLneVyjG5cWip0WPu97pRXe+++65d12UZ3XimQYMG2XVd9sp9rl26+P8OcyPN2jvKS/I51GuvvdYe63JxLufr8n+SH2vW2VRUVIS11N8xevTosOZGDx45csSu6zK+AwcODGupfK9b94orrrDHuus0ZsyYsJbKmbp76dlnnw1rqffDtGnTwlrXrl3D2t13323XdSMcv/Wtb4U1l6mXfB7f5Ugl/+8EuM88J6tbCL6ZAgCQiWYKAEAmmikAAJlopgAAZKKZAgCQiWYKAECmgqMxLiaR2sp87ty5sOa2badiKG47vhvBlooAuO3VqS3fLnbjxhKloiaDBw8Oay625GI+kv9s2jtyS5I2btwY1kaNGhXWUiPu3H24dOlSe+yQIUPCmovrpBw+fLjdx55vqXiA40a0bd68OawNGDDAruuiM3v37g1r5eXldl33mU6ZMsUee+DAgbDmnmM3XlCSvvzlL4c1Fydz4+QkHxN0z/GnPvUpu657t7tnNTV2z72zUnG+2tradp1TagRbaoRjCt9MAQDIRDMFACATzRQAgEw0UwAAMtFMAQDIRDMFACBTwdEYt624ubnZHuu2QbuYRK9evey67f1X/t30Fkl6++23w5qL8kh+2/z69evbdZwkrV69Oqy5a9i9e3e7rostuO3tJSUldl23vd3dD26KkOSjSamIllvbRX3cBBEpHf3oTFw0xkUzJB+jcpNUUnGnM2fOhDX3jA8dOtSu66I8q1atssded911Yc1NwXLvDkk6dOhQWHOfzcyZM+26Ll7k4nqTJk2y67opNy625KbYSD6mlroPnZqamrDmph5JPk5ZCL6ZAgCQiWYKAEAmmikAAJlopgAAZKKZAgCQiWYKAEAmmikAAJkKDtaUlpaGNZcblKTGxsaw5sYkuRFrqXNyuc0NGzbYdV1ucMeOHfbYL3zhC2HNjQhzI9YkP6LNXcOBAwfadd2YO5dPc3m51LFuBF7qOrgsXmo83tmzZ8Pavn37wlpqZF/qPu1Mrr766rC2a9cue6y7z1wOfeLEiXZdl9t02dbt27fbdV32NTUq7f333w9r8+bNC2v/+7//a9d19697jxYXF9t13Ui573//+2EtdW9//vOfD2tTp04Na24knOSf1VQ23qmqqgprqXdWzhhGiW+mAABko5kCAJCJZgoAQCaaKQAAmWimAABkopkCAJCp6Jzbe/5r3NiyVPzCjWhzW6RTW5Xdtu5Zs2aFtdSIn8OHD9u64y6nG/GT2kruxqw999xzYc2Nt5KkF198May5bf6pLfXTp08Pa++9915YGzZsmF3XxVBS5+TiEG5MmBsnl/q9qfjG+XbppZeGtdTz5iILLtKUis652JcbeXj06NF2r+tqko/HPfTQQ2EtNRbSjR5bt25dWNuzZ49d132ubvTYU089Zdddvnx5WHMjGlOfjbsOffr0scc2NDSENdeLUmMu3TX+1a9+ZY+V+GYKAEA2mikAAJlopgAAZKKZAgCQiWYKAEAmmikAAJkKnhrjtnxv27bNHjt+/Piw1rdv37DmtkBLfmu2i7fs3bvXruu2UKfiF05lZWVYS008KSkpCWsvvfRSWFu0aJFd93Of+1xYe/3118Oai/lIPhJSUVER1ty0Dslfh9SW+hEjRoQ1F/tIxR3cxJnOxl17N9FE8pEyN9UkNaHFTS5y01CGDx9u13X3aGrST01NTVhz0Zjrr7/ervvAAw+ENXcP3nzzzXbdLVu2hDU3ySr12bj3nXt3u0iT5Cc4pfqJ4+4l9/ynzqkQfDMFACATzRQAgEw0UwAAMtFMAQDIRDMFACATzRQAgEw0UwAAMhWcM3VZup49e9pje/XqFdbceCuXu5KkXbt2hTWXMx07dqxdd+3atWHNZWYln5FyOb1Uxs/9Pa62c+dOu64bNfVXf/VXYe2ZZ56x67rPvLq6Oqy50W2Szwi7UVOSdPDgwbDmRkKlPht3DTsbl6Xr3bu3Pba0tDSsudxm6v3g3gHu2i5btsyu655zl1GXfEbV5cX//d//3a77xhtvhDV3DVO5TTei0f0tqXesG2lWX18f1lLjzlpaWsKau88kP6bR5WZTY+xS2fkUvpkCAJCJZgoAQCaaKQAAmWimAABkopkCAJCJZgoAQKaC9wJPmDAhrKVGmrkt1EeOHAlrqe3Vbgu1O6fm5ma7bmtra1g7d+6cPfaKK64Ia5s3bw5rqViCi3Vs3LgxrKVGxrlr7K5vXV2dXdeN+po8eXJY27Fjh13XjfpyUSnJRxpqa2vb9Tsladq0abbembhoTM7YrKamprCWiqG4sXr/+Z//Gdb69+9v1z1w4EBYc8+45J8LN+YrxcUv5syZE9Z2795t13XxuJEjR4a1xsZGu66L1rnxeG4knOTHoaWur4taufdS6v52I+UKwTdTAAAy0UwBAMhEMwUAIBPNFACATDRTAAAy0UwBAMhEMwUAIFPBOVM3qieVkXRZx7KysrDmRjNJ0qpVq8KayzKmcqZHjx4Na6nMnBtV53J6LkcqSVOnTg1rW7duDWtf+tKX7Lo33XRTWHPjr1Kj6Fxmy+XI2tra7Lrufjl27Jg91t1Pbjyeux8kn/O9kKRGpbn719XcOD5J2rBhQ1hz47ZcTZK6dIm/K6xZs8Ye67Kv7h5MZV9d1tm9z2bMmGHXdWMC3cjD1PPmcvUuo9qvXz+7rntWU2Ph3LN65ZVXhrWGhga7bk5+WOKbKQAA2WimAABkopkCAJCJZgoAQCaaKQAAmWimAABkKjga47Yju3FbknTixImw5kbmpMZ8uTE+bnRQKgIwevTosOYiAJLfoj5mzJiwNnz4cLtuTU1NWLvkkkvCWktLi133+eefD2suSpLaUu/qLgqRGpPkPptULMHdwy7usGfPHrtu6rPrTFxkwY2hk/y9NHDgwHav6+IiLlbnRqyl6qnYnbsfXOQm5dSpU2EtZ+ybe97cZ5OKfY0dOzasuTFr7nemfm/qOXb3oYsmrl692q5bVVVl6yl8MwUAIBPNFACATDRTAAAy0UwBAMhEMwUAIBPNFACATP8nU2NS/xq/izO46QHFxcV23TNnzoS1yy+/PKy5KSuSn+DS2tpqj3Vb6t22bRepkfy2+aamprBWWlpq13VxHRcJScWL3N8zf/78sJa6l/bu3RvWUpMqunWLb3cXm3FxBsnHNzob97kcPnzYHuviDu7eTq3rIm7u/nT3giT16dMnrLk4juQ/U3evpN4PLvZRXl4e1ly8UPLvB/c73XGSjzWNGjUqrLl3syTNmjUrrK1du9Ye62JN7j5MRV9SsbwUvpkCAJCJZgoAQCaaKQAAmWimAABkopkCAJCJZgoAQCaaKQAAmQrOmboRP6nMlstIuSxYKjfo8n/r1q2zxzouK+oynZLPQBUVFbWrJkklJSVh7b333gtrN998s13XjZRz45dS469cXnTXrl1hbfbs2XZdlxV195Lkc8suj5saN7dt2zZb70xcXjwn1+3GAKbyoC7ruGbNmrCWGtXl/p5U9tWdkxs5ee7cObuue2e5sWSpsW/uuXB58VRG2mWA6+vrw9qRI0fsuu49mRob6bK87u9x/36AlM7GpvDNFACATDRTAAAy0UwBAMhEMwUAIBPNFACATDRTAAAyFZ1L7eUGAAAW30wBAMhEMwUAIBPNFACATDRTAAAy0UwBAMhEMwUAIBPNFACATDRTAAAy0UwBAMhEMwUAIBPNFACATDRTAAAy0UwBAMhEMwUAIBPN9ALU3NysRx55REOHDlVpaamqqqr0P//zPx19WgA+odWrV+uWW25RRUWFevbsqWnTpuk73/lOR58W2qFbR58APrk/+ZM/0SuvvKKHHnpIEyZM0EsvvaTPfvazeuuttzRnzpyOPj0ABXjttdd08803a9asWXrsscdUVlam6upq1dTUdPSpoR0YDn6BWbFihaqqqvTNb35TDz/8sCTp9OnTmjZtmiorK7Vs2bIOPkMAKcePH9fEiRN15ZVX6pVXXlGXLvyPhBc6PsELzCuvvKKuXbvqvvvu++hnPXr00J/+6Z/q3Xff1d69ezvw7AAU4sc//rH27dunJ598Ul26dNGJEyfU1tbW0aeFDDTTC8z777+viRMnqry8/GM/nz17tiRpzZo1HXBWAD6J119/XeXl5aqtrdWkSZNUVlam8vJyPfDAAzp9+nRHnx7agWZ6gamvr9eQIUN+6+cf/qyuru58nxKAT2jbtm1qbW3Vrbfeqvnz5+unP/2p7r33Xj333HO65557Ovr00A5sQLrAnDp1SiUlJb/18x49enxUB9C5NTU16eTJk7r//vs/2r17++23q6WlRd/73vf0+OOPa8KECR18lvgk+GZ6gSktLVVzc/Nv/fzD/2motLT0fJ8SgE/ow+d04cKFH/v53XffLUl69913z/s5IQ/N9AIzZMgQ1dfX/9bPP/zZ0KFDz/cpAfiEPnxOBw0a9LGfV1ZWSpKOHDly3s8JeWimF5iZM2dq69atOn78+Md+vnz58o/qADq3Sy+9VJJUW1v7sZ9/uOdh4MCB5/2ckIdmeoG54447dPbsWT3//PMf/ay5uVmLFy9WVVWVRowY0YFnB6AQd911lyTpBz/4wcd+/sILL6hbt2669tprO+CskIMNSBeYqqoq3XnnnVq0aJH279+v8ePH64c//KF27dr1Ww8mgM5p1qxZuvfee/Xiiy+qtbVVc+fO1ZIlS/STn/xEixYt4v+uuQDxLyBdgE6fPq3HHntMP/rRj3TkyBHNmDFDTzzxhObPn9/RpwagQGfOnNFTTz2lxYsXq66uTqNGjdKDDz6ohx56qKNPDe1AMwUAIBP/nykAAJlopgAAZKKZAgCQiWYKAEAmmikAAJlopgAAZKKZAgCQiWYKAECmgv85wTlz5oS1AwcO2GPdjM3fNZvzQ6mJ8+Xl5WGtuLg4rHXr5v/sYcOGhbVt27bZY8+ePRvW3Hi01D9svXXr1rD2u4aFf2jy5Ml23U2bNoW13zXq7UNlZWV2XTf1wtWKiorsusOHDw9r7tpLUv/+/cPaypUrw1rv3r3tuq6+fv16e+z5Nm7cuLBWUVFhj21tbQ1r7t9+2bdvn1138ODBYe03p6r8uoaGBrvuhxNYfpc1a9bYY3v27BnWzpw5E9bceyd1Ti0tLWHNXaMUd76p+ce/OVDj11188cVhLTVCrq2tLaz16dPHHnvw4MF2Hdu9e3e7rnt/FPIc880UAIBMNFMAADLRTAEAyEQzBQAgE80UAIBMBe/m3bNnT1gbOXKkPbapqSmsud28qZ2dbsed22FZV1dn192yZUtYS+1MdDtr3e5Dt1NS8jtRBwwYENZqa2vtum5nrbu+bhewJM2cOTOsud3Ubsez5O/D1E5Kx+2mPnz4sD3W3d+djXsu3K7P/5/r9ujRI6y5ZzG1+3j//v1hrVevXvZY9047ceJEWHMJA8nvKHW75xsbG+267v1x9OjRdp2P5HcRr127Nqy5z1Tyz2oqKeB23brrkLpfUrv2U/hmCgBAJpopAACZaKYAAGSimQIAkIlmCgBAJpopAACZaKYAAGQqOGfqpkK43JXkM1IuW+UmhEh+eoCbZJPKQLnspct7Sj635XKzqYxkdXV1WHN50BSX63T5NJdtlaRDhw6FNTfZwX2mkjR06NCwlroOu3fvDmtjxowJa6m8s1u3s3HPW+ozdfegu49SWeeTJ0+GNZdDTz2LmzdvDmujRo2yx27fvj2sTZ06NaytW7fOruveae76XnfddXZdl/l057tixQq7rntXuik3qayo6xmpY7t27RrW+vbtG9ZSefG9e/faegrfTAEAyEQzBQAgE80UAIBMNFMAADLRTAEAyEQzBQAgU8HRGDf2JrVF3UUW3DboHTt22HXdSB03Cmnr1q12XTfGJzVOyo3xcdvMU+u6LfVuq3hqtJuLAbmohIszSFJbW1tYc1vUUyOs3O9NRZ5OnToV1lx8KzWua+zYsbbembhr5EaWSdLEiRPDmouSXH755XZdF7Fwz0UqhjJ69OiwVlNTY4917yV3P1x55ZV23Z07d4Y1F+tIjY10v3f9+vVhzY1Yk6TTp0+HNTdKMRWHcvG41EhD14vcveSiUlI6ipnCN1MAADLRTAEAyEQzBQAgE80UAIBMNFMAADLRTAEAyFRwNGbYsGFhLbXN3EUS3JQVN7VE8tENF1FJTQEZP358WNuyZYs91k3PcNv8UxM7XNTExTq6dPH/veRiS3Pnzm3XcZL0yiuvhLU9e/aEtVSMwm2bd1EpyU/PeO+998JaKvqS2srfmbj4Syoa4GJLLi7i4iCSnwTk3g9Dhgyx67rJJFdddZU91k1hqaqqavc5LV68OKy5CTmpe8x9ru76umlgko88uWiMq0n+/eHiepKPAblpVSNGjLDruvdDIfhmCgBAJpopAACZaKYAAGSimQIAkIlmCgBAJpopAACZaKYAAGQqOGfqRvGMHDnSHnvgwIGwNmHChLC2YcMGu64bt+NyWW78j+RH/KRGFrkRVy4P6vKpks9PDRo0KKy5jJkkPfjgg2HNZfxSeVA3Ps+tu3v3bruuy80uX77cHutyvjmZuQspZ3rxxReHNZcHl/zz9pnPfCaspZ7j6urqsOae1dR4QZdldL9Tkh5//PGwdtlll4W11L3gzvnpp58Oa+7aS/6dtXLlyrA2c+ZMu657P7vfmXqOXW7e9RrJvwtT4zWdgwcPtvtYiW+mAABko5kCAJCJZgoAQCaaKQAAmWimAABkopkCAJCp4GiM2/Kd2lLvtqi7CEW/fv3suvv27Qtrbnt137597br19fVhzY1Jknz8ZdSoUWEtNRausrIyrN1///1h7YorrrDruvN129fdyDLJb+V3o9JSY5LcCLzU5+piQs3NzWEtFcFIxaU6E/fMFBcX22PdSEQ3Zm3y5Ml2XRd/cXEFNwpN8vfKnXfeaY91ESI3qs69zyQfRXHj21Jj7NzISfe3pHTt2jWs1dXVhbVU1M/9ramxkW5s3LFjx8KaGyMqpftYCt9MAQDIRDMFACATzRQAgEw0UwAAMtFMAQDIRDMFACBTwdEYt20+9a/tuziJ2xafiqG4Ld9r1qwJa6tWrbLrzpgxI6ydPHnSHusiLG7rtZv8Ikn33ntvWHORkF69etl1XcRlypQpYS0Vo/j6178e1srKysJaalv8ihUrwtrPfvYze+yyZcvCmou3uG38klRbW2vrnYmb9JGa1uH+TneN9uzZY9d194Orpa67i4WlImP9+/cPa6tXrw5re/futeuOGzcurLl4lvvcJGndunVh7aqrrgprNTU1dl33znIRFRebkaRt27aFtTFjxthj3fSnnH7C1BgAADoYzRQAgEw0UwAAMtFMAQDIRDMFACATzRQAgEw0UwAAMhWcM506dWpYW79+vT3W5SDdmK8DBw7YdV1WyeUGU3kjNz7MZZwkP67r85//fFi7/vrr7brl5eVh7atf/WpYmzBhgl13wYIFYe3QoUNh7fjx43Zdl2/NyTq6sXyXX365Pdbdp+5vTY1mSl2LzuSDDz4Ia6lRXe4d4EaEuTyi1P5xiS7TLUl33HFHWHNZckk6cuRIWLvvvvvC2g033GDX/cxnPhPWbrvttrD2i1/8wq67f//+sObu+4EDB9p1f/nLX4a1sWPHhjWX05WkiRMnhrXU/eLG3LnxeA0NDXbdVL9J4ZspAACZaKYAAGSimQIAkIlmCgBAJpopAACZaKYAAGQqOBqze/fusNajRw97rBuVVFVVFdZ27dpl13XxFxfNcPEVyUcdSktL7bFf/OIXw9r8+fPDWmpMlRsvtm/fvrDmxjpJUn19fVgrKioKa6lRUydOnAhrLnYwYMAAu+6f/dmfhbVLLrnEHrto0aKw9vDDD4c1FwGQpJ49e9p6Z+JGgLl7QfIxNhexaGpqsuuOGDEirLl728W6JGn27NlhLTW+bcmSJWGte/fuYS0VYXHxi02bNoU19yxK0ujRo8Pali1bwtr48ePtum4cmosQuiif5ONxqfezG8vnRmSm3t2p5zyFb6YAAGSimQIAkIlmCgBAJpopAACZaKYAAGSimQIAkIlmCgBApoJzpn369AlrqdE1bmzZsWPHwlpFRYVd12VJ161bF9aGDBli1+3du3dYc2OoJJ9tc7nZBx980K5bXFwc1tyYtWXLltl1Xa7TZQcPHjxo13Wf66xZs8La1q1b7brf/va3w9oTTzxhj3XZN3dPdOni/5szldXrTNx91NjYaI91188dmxqV5q59t27xK8pl1CU/2u29996zx7q8qMvNuusr+fvbZbenT59u13XvFpfHd2MsU8d27do1rKWug8uKpt4tLt/qxre597rk77VC8M0UAIBMNFMAADLRTAEAyEQzBQAgE80UAIBMNFMAADIVvBfYbdseNmyYPdZtqXdjn1w0Q/Jb312UJ7X1un///mHNjfGS/LijN998M6z17dvXrnv48OGwtnPnzrCWGt3U3ghRasRSe0ewpWJLDQ0NYW3QoEH2WBelcmP3UtfQHdvZuEhCagSVu0fdSC33nEpSTU1NWLvuuuvCWirq4CJ5S5cutcdu3749rLnrkHq3uHF9rnbo0CG7rovruHdhavSju19cJO9rX/uaXdeNsUv1E/eOffTRR8Oae3dIfsRgIfhmCgBAJpopAACZaKYAAGSimQIAkIlmCgBAJpopAACZCo7GNDc3hzUXdZCkKVOmhLX9+/eHNRebkfy27dbW1rA2evRou+5NN90U1lxERfJb7jdv3hzWUtu23QQdd06pyM3JkyfD2uTJk8PaqlWr7LouhuLiA5MmTbLrHj16NKy56FFq7R49ethjHRft6GzcuboJQpJUXV0d1kpLS9t9Tm5ah4tKpSIUa9asCWupCTluUpCblpJ6t6xYsSKsXXPNNWHNxXwk/75zE5xScahvfvObYc1Fedy7Q5I2btwY1twzLvlYzRVXXBHWXn31Vbuui0MVgm+mAABkopkCAJCJZgoAQCaaKQAAmWimAABkopkCAJCJZgoAQKaCc6Yux7Rjxw577L59+8Kay2W5PJckHThwIKy53Null15q173tttvC2pIlS+yxLmf6wQcfhLXUSDM3xsrlvVLc9Xefm8u9ptZ1ueQNGzbYdceMGRPWUuOv5s2bF9bcCKtRo0bZdbdt22brnYnL06ZGULl8n/tMXUZd8jlTN4bRZRUl6T/+4z/C2rJly+yxVVVVYW3lypVh7aKLLrLrXnXVVWGtrq4urLW0tNh13bvy5ptvDmt33XWXXddl+d0z/sMf/tCu+4tf/CKsffrTn7bHzp49O6x96lOfCmsvvPCCXfeyyy6z9RS+mQIAkIlmCgBAJpopAACZaKYAAGSimQIAkIlmCgBApoKjMS4e4GIzkh+p47Z0l5SU2HVd/KW4uDisffGLX7TrHj9+PKy9/PLL9tja2tqwVllZGdZSW9/PnTvXrlpqjF3//v1tPeJGrEnSqVOnwpr7XFevXm3XdX/r888/b491a584cSKsHTp0yK7rxoR1Nm4MXWr0lYvOuHvbRawkfz+42N3QoUPtusOHDw9rgwYNssfu3LkzrLkoz969e+26Lk7i7kEXB5F8nMTFRVJxsu9+97thbffu3WEtNVLSRdFS0Zh+/fqFtbVr14a1wYMH23Vz8c0UAIBMNFMAADLRTAEAyEQzBQAgE80UAIBMNFMAADIVHI1x0yZSW+rdRBQXzXAxk9Tv/dGPftSu4yTpX//1X8NaKrrhtl+7aIHbKi5JI0eODGsuyjNs2DC7rvtcXaymT58+dl0XaXDHuvORpCeffDKsvf/++/bYb3zjG+36valrmPrsOpNNmzaFtb59+9pjXYTFTS5KxVDctBT3uaQmFy1YsCCspaY/7dq1K6y5d1Yq4lZTUxPWfv/3fz+s3XfffXZdd/33798f1t555x27rpuI5CJPqc986tSpYe26666zx7r78O233w5rRUVFdt1UX0jhmykAAJlopgAAZKKZAgCQiWYKAEAmmikAAJlopgAAZKKZAgCQqeCcqcsyrl+/3h7rRiG5bE+XLr7XX3/99WHN5a6am5vtuv/93/8d1lJjn9xIo9OnT4c1N8JK8llGt24q9+bW7d69e1jbs2ePXdeNSXLrPvzww3bdgQMHhrVvf/vb9lh3ndwowGPHjtl1L6QRbBdddFFYa2xstMe68Xcuy+jGN0p+nN+//du/hbU//MM/tOu65/xLX/qSPfbpp5+29UgqJ+3ylV/+8pfD2smTJ+26EydODGuLFi0Ka6lstsuSuuy7+zsl/9m50ZqSH3P3X//1X2EtlaN2PaMQfDMFACATzRQAgEw0UwAAMtFMAQDIRDMFACATzRQAgEwFR2PcSJ0RI0bYY912cRdJcDEISbr88svDmttevXTpUrvuiRMnwtqpU6fsse73DhgwIKyltmV36xZ/VG7dhoYGu6471o1fctEXSdq5c2dYc6Ox7rnnHrvu1q1bw1pqFJobf+Xu71WrVtl1U/GjzsSNEHSjEiV/r7hrsHHjRrvuhAkTwtqhQ4fC2q9+9Su7rosBpd4tzz77bFhz8a3Pfvazdl0X53Pxl+LiYrvuo48+GtbcWDI3zkzyIyXnzJkT1h566CG7rnsvbd++3R77yCOPhLXJkyeHtdTfWl1dbespfDMFACATzRQAgEw0UwAAMtFMAQDIRDMFACATzRQAgEwFR2Nc1MHFCiRp06ZNYc1NYWltbbXr3n777WHNxTqef/55u67jJlxIPjLiIjepaRNuUoKb5pHaUr9jx46w5qJHqQkMLgLwe7/3e2HNRV8k6R//8R/D2u7du+2xkyZNCmvuc03Foc6ePWvrnYl73lIRny1btoS1YcOGhbUDBw7Ydd1n3rt377D2xBNP2HX/+Z//Oaz179/fHuuiaO7enjdvnl3XPecuuuHue8nH/W655Zaw5qb9SH7ijHvXpaJ+7l5zk4Ik6cyZM2HNRXlSEa2ysjJbT+GbKQAAmWimAABkopkCAJCJZgoAQCaaKQAAmWimAABkopkCAJCp6JwLKf4alyVNjW5y+R2X4Uud2htvvBHWDh48GNa+8IUv2HVdRqpLF//fH6dPnw5rbkRYRUWFXbeoqCisuSxeapyR+70LFy4MazfeeKNd12XBamtrw9qLL75o192wYUNYa2trs8e6+9CN5Erd3+4ar1mzxh57vl122WVhLZXrXrt2bVgbNWpUWBs/frxdt76+PqxVVlaGNZd7laT58+eHNZeflKQhQ4aENZebdc+/5DPJLmeayjK7XKzL1DY2Ntp1XZbUjcd77rnn7Lovv/xyWHOj/iTpyJEjYc1dw6NHj9p13SjAJUuW2GMlvpkCAJCNZgoAQCaaKQAAmWimAABkopkCAJCJZgoAQKaCozFTpkwJa25btuS33LuRRGPGjLHrfuUrXwlrbjt4apvzSy+9FNaOHTtmj3WxGhdv2bNnj1133LhxYc3FW+bMmWPXveKKK8KaO98ZM2bYdd0YpSeffDKsuUiT5CM577//vj3WRWOGDx8e1pqbm+26Lq6zb98+e+z5Nnny5LDm4kGSj7G56EZqtFv37t3Dmnsu3Di51LqpiNuCBQvC2q233hrWUveKi4VdcsklYc2NlJT8SMRU5MlxI9qeeeaZsLZ582a7rms7I0eOtMdWV1eHNRdpSl0H975buXKlPVbimykAANlopgAAZKKZAgCQiWYKAEAmmikAAJlopgAAZCo4GuOmTaS2mbtt3dOnTw9rqWkdblrK3/7t34a11J/soj6pqSZuMsHJkyfD2h//8R/bdV0MoK6uLqy5qRuStHz58rC2dOnSsLZjxw677u7du8OamzDiJkJIUkNDQ1hz0ZeUmpqasOZiM5KPfqQmm5xvLgrlYhCSvwfdvZ2apOLiZO5ZTK3rJhel/lYX9Zk3b15Yu/POO+267nl055R6Ltw7zU01WrVqlV3XXf+ce9t95qNHj7bHus/dnZObgCP5eNEvf/lLe6zEN1MAALLRTAEAyEQzBQAgE80UAIBMNFMAADLRTAEAyEQzBQAgU8E50xtuuCGsvffee/ZYl606fvx4WHPj2STpxIkTYe32228Pa5/73Ofsum4UUipfWVpaGtYGDRoU1lKZWpcHc6PHVq9ebdd9/fXXbT2SGjXlsqQuY+bGmUk+0+xGhEn+s3GjyVxmVvLZ4s42gm3u3LlhzT2LklRSUtKuWlNTk13X3Q/ueUuNaHTjEouLi+2xrt6nT5+w5vKpks86unFzbjyY5K+xe8W7jLoktbW1hTX3bwS4+0HyefFUS3Kj1FyvSY13dMe+/fbb9liJb6YAAGSjmQIAkIlmCgBAJpopAACZaKYAAGSimQIAkKngaIzb0n3RRRfZY90288OHD4c1t2Ve8mOJBg8eHNZSo5tcNCa15dtFQtw2/7Vr19p13bVw29tTEQA3YsnFUFJxB3dsfX19u46T/PV3I7ckP9rJRVjcvST5LfebNm2yx55v7j6aM2eOPdZFN3r16hXWunfvbtd1MSu3biqe5cbCpaJo7v5ubGwMay7+Jvn4kXsvufFgKe55S73PrrnmmrDm4nqpcYju73E9QfKfjYtLunezJNXW1oa1Xbt22WMlvpkCAJCNZgoAQCaaKQAAmWimAABkopkCAJCJZgoAQCaaKQAAmeKQ4W+YNm1aWDtw4IA91mWKXB4pNe6soqIirLnsXyqf9tprr4W1VEbV/a1uPJPLR0nSsGHDwpobSdS/f3+7rst01dTUhLXUSCh3HVz21eWZJWns2LFh7Y033rDHusyi40asSelsYWfisnapEXaOu49c3lOSBg4cGNZc9m/o0KF2Xff+GD58uD3WPeezZs0Ka0uXLrXrulF/Lj+Zeu+4z86Nsky9H1y22GXU3TtJ8qMfU5+rG73nfm/qPkzl1FP4ZgoAQCaaKQAAmWimAABkopkCAJCJZgoAQCaaKQAAmQoewea21LuIiuS3fLtozPbt2+26l19+ebt+Z0rXrl3DmtuqL/lRXiNGjAhrpaWldt2GhoawNmDAgLDmxtRJfiSXi7Ckojy9e/cOay4KsXr1arvulClTwloq+nLo0KGwtnnz5rA2d+5cu667J9avX2+PPd9uv/32sJaKDrjYgYsVuEiY5KNHLiqViiw5qZFm7h3g/p6Wlha7rnvOXRRt4sSJdt233norrI0bNy6sufFsklRZWRnWpk6dGtZSY9RcXMc9p5IfiejGwrlYneTPecOGDfZYiW+mAABko5kCAJCJZgoAQCaaKQAAmWimAABkopkCAJCp4GiM+5f6XbxC8tEBt/W9Sxff67t37x7Wjh07FtZclESSSkpKwlpqO76Lfbjt9m4CgyRt2rQprKW2fLeX+1tT5+siAqn7xXGRBhctkCR3q7t1U5+5iwmlogfnm5uW4p4nyX+mQ4YMCWsuBiH5iSjuPktNf3KTVCZNmmSPra6uDmsuEpKKdrlzctc/dQ3b2trCmosupiKE7ve693Pqs3HxotTEmT59+th6pLGx0dbdJKwPPvgguT7fTAEAyEQzBQAgE80UAIBMNFMAADLRTAEAyEQzBQAgE80UAIBMPiz4a1xuKJUHdZlPN6pr6NChdl037syNkzp+/Lhd1+UgXd5W8qPS3Lg5lz+TpPLy8rC2bt26sDZ58mS77sGDB8OaG92UOl83CsmNzkuNjHO5uAkTJthjt23bFtZcxix1v+SMAjvfXIbajVmU2n/93HMq+VF/bmxhKi/+5ptvhrXUuDm3thvH5fKekr/G7pxSOVP3btm6dWtYcyPWJP+8uQy1y9tL0vTp08Na6nlz/cRlSVPjM901LATfTAEAyEQzBQAgE80UAIBMNFMAADLRTAEAyEQzBQAgU8Ej2AAAwO/GN1MAADLRTAEAyEQzBQAgE80UAIBMNFMAADLRTAEAyEQzBQAgE80UAIBMNFMAADL9P4tH5LE+/nU9AAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 500x500 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# visualize some data\n",
    "imgs, labels = next(iter(trainloader))\n",
    "imgs = gaussian_noise(imgs)\n",
    "visualize_data(imgs, labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cuda:0\n"
     ]
    }
   ],
   "source": [
    "# device\n",
    "device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\n",
    "print(device)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# image size\n",
    "img_size = imgs[0].shape[1]**2 # 28*28 = 784\n",
    "# latent size\n",
    "latent_size = 20"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Reconstructed images shape:  torch.Size([128, 784])\n",
      "Encoded images shape:  torch.Size([128, 20])\n"
     ]
    }
   ],
   "source": [
    "# define model\n",
    "model = ae.AE(img_size, latent_size).to(device)\n",
    "\n",
    "# pass some data through the model\n",
    "imgs, labels = next(iter(trainloader)) #imgs.shape = [batch_size, 1, 28, 28], labels.shape = [128]\n",
    "\n",
    "# flatten images\n",
    "imgs = imgs.view(imgs.shape[0], -1).to(device) #imgs.shape = [128, 784]\n",
    "recon_imgs, encoded_imgs = model(imgs)\n",
    "\n",
    "print('Reconstructed images shape: ', recon_imgs.shape)\n",
    "print('Encoded images shape: ', encoded_imgs.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# optimizer\n",
    "optimizer = optim.Adam(model.parameters(), lr=1e-3)\n",
    "\n",
    "# loss function\n",
    "criterion = nn.MSELoss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  1%|          | 1/100 [00:03<06:32,  3.96s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 0, Loss: 0.0331\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 11%|█         | 11/100 [00:43<05:53,  3.97s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 10, Loss: 0.0170\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 21%|██        | 21/100 [01:23<05:18,  4.03s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 20, Loss: 0.0168\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 31%|███       | 31/100 [02:02<04:29,  3.90s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 30, Loss: 0.0172\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 41%|████      | 41/100 [02:41<03:47,  3.86s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 40, Loss: 0.0164\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 51%|█████     | 51/100 [03:20<03:10,  3.89s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 50, Loss: 0.0158\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 61%|██████    | 61/100 [03:59<02:31,  3.89s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 60, Loss: 0.0155\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 71%|███████   | 71/100 [04:37<01:50,  3.81s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 70, Loss: 0.0159\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 81%|████████  | 81/100 [05:15<01:12,  3.82s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 80, Loss: 0.0157\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      " 91%|█████████ | 91/100 [05:53<00:34,  3.82s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch: 90, Loss: 0.0149\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 100/100 [06:28<00:00,  3.89s/it]\n"
     ]
    }
   ],
   "source": [
    "# training loop\n",
    "epochs = 100\n",
    "\n",
    "for epoch in trange(epochs):\n",
    "    for i, (imgs, labels) in enumerate(trainloader):\n",
    "\n",
    "        # corrupt images\n",
    "        imgs = gaussian_noise(imgs)\n",
    "\n",
    "        # flatten images\n",
    "        imgs = imgs.view(imgs.shape[0], -1).to(device)\n",
    "\n",
    "\n",
    "        # zero the parameter gradients\n",
    "        optimizer.zero_grad()\n",
    "\n",
    "        # forward pass\n",
    "        recon_imgs, encoded_imgs = model(imgs.to(device))\n",
    "\n",
    "        # compute loss\n",
    "        loss = criterion(recon_imgs, imgs)\n",
    "\n",
    "        # backward pass\n",
    "        loss.backward()\n",
    "\n",
    "        # update parameters\n",
    "        optimizer.step()\n",
    "\n",
    "    # print loss\n",
    "    if epoch % 10 == 0:\n",
    "        print('Epoch: {}, Loss: {:.4f}'.format(epoch, loss.item()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "# test model\n",
    "original_imgs, labels = next(iter(testloader))\n",
    "# corrupt images\n",
    "noisy_imgs = gaussian_noise(original_imgs)\n",
    "noisy_imgs = noisy_imgs.view(noisy_imgs.shape[0], -1).to(device)\n",
    "\n",
    "# pass through model\n",
    "recon_imgs, encoded_imgs = model(noisy_imgs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 500x500 with 9 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# visualize the original, noisy and reconstructed images. 3x3 grid\n",
    "# first column: original images, second column: noisy images, third column: reconstructed images\n",
    "plt.subplots(3, 3, figsize=(5,5))\n",
    "for i in range(3):\n",
    "    plt.subplot(3,3,i+1)\n",
    "    plt.imshow(original_imgs[i].detach().numpy().squeeze(), cmap='gray')\n",
    "    plt.title(\"Original\")\n",
    "    plt.axis('off')\n",
    "    plt.subplot(3,3,i+4)\n",
    "    plt.imshow(noisy_imgs[i].reshape(28,28).cpu().detach().numpy(), cmap='gray')\n",
    "    plt.title(\"Noisy\")\n",
    "    plt.axis('off')\n",
    "    plt.subplot(3,3,i+7)\n",
    "    plt.imshow(recon_imgs[i].reshape(28,28).cpu().detach().numpy(), cmap='gray')\n",
    "    plt.title(\"Reconstructed\")\n",
    "    plt.axis('off')\n",
    "\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "```\n",
    "Worked nicely. Good stuff!\n",
    "```"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "torch",
   "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": "3.11.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}