Train.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import torch\n",
    "import torchvision\n",
    "import matplotlib.pyplot as plt\n",
    "from time import time\n",
    "from torchvision import datasets, transforms\n",
    "from torch import nn, optim\n",
    "import torch.nn.functional as F\n",
    "\n",
    "from Network.simpleCNN import simpleCNN\n",
    "from Network.simpleFF import simpleFF"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Load data and models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "transform = transforms.Compose([transforms.ToTensor(),\n",
    "                              transforms.Normalize((0.5,), (0.5,)),\n",
    "                              ])\n",
    "\n",
    "# Prep the datasets\n",
    "trainset = datasets.MNIST('MnistData', download=True, train=True, transform=transform)\n",
    "valset = datasets.MNIST('MnistData', download=True, train=False, transform=transform)\n",
    "\n",
    "# Feeding for a NN\n",
    "trainloader = torch.utils.data.DataLoader(trainset, batch_size=64, shuffle=True)\n",
    "valloader = torch.utils.data.DataLoader(valset, batch_size=64, shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "criterion = nn.NLLLoss()\n",
    "\n",
    "ff_network = simpleFF(28*28, 64, 32, 10)\n",
    "ff_optimizer = optim.SGD(ff_network.parameters(), lr=0.003, momentum=0.9)\n",
    "\n",
    "cnn_network = simpleCNN(9 * 4 * 4, 64, 32, 10)\n",
    "cnn_optimizer = optim.SGD(cnn_network.parameters(), lr=0.003, momentum=0.9)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train a FF network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 10 train loss 94.37896613031626 validation loss 17.598387390375137"
     ]
    }
   ],
   "source": [
    "epochs = 10\n",
    "\n",
    "train_error = []\n",
    "valid_error = []\n",
    "\n",
    "for epoch in range(epochs):\n",
    "    training_loss= 0\n",
    "    validation_loss = 0\n",
    "    for images, labels in trainloader:\n",
    "        ff_network.train()\n",
    "        ff_optimizer.zero_grad()\n",
    "        \n",
    "        ff_out = ff_network(images.view(images.shape[0], -1))\n",
    "        loss = criterion(ff_out, labels)\n",
    "        loss.backward()\n",
    "        ff_optimizer.step()\n",
    "        training_loss += loss.item()\n",
    "        \n",
    "    for images, labels in valloader:\n",
    "        ff_network.eval()\n",
    "        ff_out = ff_network(images.view(images.shape[0], -1))\n",
    "        loss = criterion(ff_out, labels)\n",
    "        validation_loss += loss.item()\n",
    "    \n",
    "    train_error.append(training_loss)\n",
    "    valid_error.append(validation_loss)\n",
    "    print(\"\\rEpoch {} train loss {} validation loss {}\".format(epoch+1, training_loss, validation_loss), end=\"\")\n",
    "    \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure()\n",
    "plt.plot(train_error)\n",
    "plt.plot(valid_error)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "torch.save(ff_network.state_dict(), \"Network/ff_network\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train a CNN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 10 train loss 38.338099632412195 validation loss 6.529189169406891"
     ]
    }
   ],
   "source": [
    "epochs = 10\n",
    "\n",
    "train_error = []\n",
    "valid_error = []\n",
    "\n",
    "for epoch in range(epochs):\n",
    "    training_loss= 0\n",
    "    validation_loss = 0\n",
    "    for images, labels in trainloader:\n",
    "        cnn_network.train()\n",
    "        cnn_optimizer.zero_grad()\n",
    "        \n",
    "        cnn_out = cnn_network(images)\n",
    "        loss = criterion(cnn_out, labels)\n",
    "        loss.backward()\n",
    "        cnn_optimizer.step()\n",
    "        training_loss += loss.item()\n",
    "        \n",
    "    for images, labels in valloader:\n",
    "        cnn_network.eval()\n",
    "        cnn_out = cnn_network(images)\n",
    "        loss = criterion(cnn_out, labels)\n",
    "        validation_loss += loss.item()\n",
    "    \n",
    "    train_error.append(training_loss)\n",
    "    valid_error.append(validation_loss)\n",
    "    print(\"\\rEpoch {} train loss {} validation loss {}\".format(epoch+1, training_loss, validation_loss), end=\"\")\n",
    "    \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYAAAAD8CAYAAAB+UHOxAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi40LCBodHRwOi8vbWF0cGxvdGxpYi5vcmcv7US4rQAAHJNJREFUeJzt3XtsXOd55/HvMzOcoXiZ0Y26cCRHsq04FqdJ7WUdOwGKNm7aOE3rFGiAdHcbIzBgLDbbpk2B1t1/AuxigRQImgu6COCN0zptkLZwA9gogiSG42CxWNsbOU5jyUoiWY4t6krdeBUvM/PsH+cdckiNSIpD8nDm/D7AYM55z3tmHg4k/njOO+c95u6IiEjypOIuQERE4qEAEBFJKAWAiEhCKQBERBJKASAiklAKABGRhFIAiIgklAJARCShFAAiIgmVibuApezcudMPHDgQdxkiIi3llVdeueTufcv129QBcODAAY4cORJ3GSIiLcXM3lpJP50CEhFJKAWAiEhCKQBERBJKASAiklAKABGRhFIAiIgklAJARCSh2jIAzly7zue/+zNOX5mMuxQRkU1r2QAws6+Z2UUzO1rXtt3MnjOzE+F5W2g3M/uymZ00s5+Y2b11+zwS+p8ws0fW58eJjE+V+ZsXTvLKW1fX821ERFraSo4A/g740KK2x4Hn3f0Q8HxYB3gIOBQejwFfgSgwgM8C7wXuAz5bC431cEdfN7lMitfOjKzXW4iItLxlA8Dd/zdwZVHzw8BTYfkp4KN17V/3yEvAVjPbC/wW8Jy7X3H3q8Bz3BgqayaTTnH33jxHFQAiIje12jGA3e5+DiA87wrtReB0Xb+h0Haz9nVTKuZ5/ewo1aqv59uIiLSstR4EtgZtvkT7jS9g9piZHTGzI8PDw6supNRfYGy6zNsaCBYRaWi1AXAhnNohPF8M7UPA/rp++4CzS7TfwN2fcPdBdx/s61t2NtObKhULABw9q9NAIiKNrDYAngVq3+R5BHimrv0T4dtA9wMj4RTRd4HfNLNtYfD3N0Pbujm0u4eOtHH0zOh6vo2ISMta9n4AZvZN4NeAnWY2RPRtns8B/2xmjwJvAx8L3b8NfBg4CUwCnwRw9ytm9t+BH4Z+/83dFw8sr6lcJs07d/dyTEcAIiINLRsA7v4HN9n0YIO+DnzqJq/zNeBrt1Rdk0r9Bb73+nncHbNGwxAiIsnVllcC15SKea5OznJ2ZCruUkRENp02D4AwEKzrAUREbtDWAXD33jzplHFMASAicoO2DoDOjjR39vVw9Ky+CSQislhbBwDAQFFTQoiINNL2AVDqL3BxbJqLoxoIFhGp1/4BEAaCj+k0kIjIAm0fAIf784C+CSQisljbB0BPLsPtO7t1bwARkUXaPgAABooFnQISEVkkEQFQ6s9z5tp1rkzMxF2KiMimkYwAmBsI1mkgEZGaRATAwNxAsE4DiYjUJCIAtnZl2bdti24OIyJSJxEBANEFYZoTSERkXnICoJjnF5cnGZ2ajbsUEZFNIUEBEA0Ev66vg4qIAAkKgIF+3RtARKReYgKgrzfHnnynLggTEQkSEwAQjQPoCEBEJJKoABjoL/DG8DiTM+W4SxERiV2iAqBULFB1OH5uLO5SRERil7AAiK4I1pQQIiIJC4A9+U52dGd5bUgBICKSqAAwMwaKBd0kXkSEhAUARFNDn7gwxtRsJe5SRERilbwAKBYoV52fX9BAsIgkW/ICYO6KYJ0GEpFkS1wA7N++hd7OjKaGFpHES1wAmJmmhhYRockAMLM/NbNjZnbUzL5pZp1mdtDMXjazE2b2T2aWDX1zYf1k2H5gLX6A1SgV8xw/P8ZspRpXCSIisVt1AJhZEfhjYNDdS0Aa+DjwV8AX3P0QcBV4NOzyKHDV3e8EvhD6xaJULDBTrnLy4nhcJYiIxK7ZU0AZYIuZZYAu4BzwAeDpsP0p4KNh+eGwTtj+oJlZk++/KrV7A2hiOBFJslUHgLufAT4PvE30i38EeAW45u612daGgGJYLgKnw77l0H/Hat+/GQd3dNOdTWtqaBFJtGZOAW0j+qv+INAPdAMPNejqtV2W2Fb/uo+Z2REzOzI8PLza8paUShmH+zU1tIgkWzOngH4DeNPdh919FvgW8D5gazglBLAPOBuWh4D9AGF7Abiy+EXd/Ql3H3T3wb6+vibKW9pAf4HXz41Sqd6QQSIiidBMALwN3G9mXeFc/oPA68ALwO+HPo8Az4TlZ8M6Yfv33T22376lYoHJmQpvXpqIqwQRkVg1MwbwMtFg7o+A18JrPQH8BfAZMztJdI7/ybDLk8CO0P4Z4PEm6m6apoYWkaTLLN/l5tz9s8BnFzWfAu5r0HcK+Fgz77eW7uzrIZdJcfTMCA//cnH5HURE2kzirgSuyaRTvGtvntc0ECwiCZXYAIBoauhjZ0apaiBYRBIo2QFQLDA2Xeb01cm4SxER2XDJDgBNDS0iCZboAHjnnh4yKdPU0CKSSIkOgFwmzTt39+qKYBFJpEQHAETXAxw7O0qM16SJiMRCAVAscGVihnMjU3GXIiKyoRQAmhpaRBIq8QFw9548KYOjmhpaRBIm8QGwJZvmzl09ukewiCRO4gMAousB9FVQEUkaBQAwUCxwYXSai2MaCBaR5FAAEM0JBOgWkSKSKAoA4HAtADQOICIJogAAejs7OLizW1NDi0iiKACCgf68JoUTkURRAASlYoEz165zdWIm7lJERDaEAiCoTQ2tgWARSQoFQDAQBoJ1PYCIJIUCINjWnaW4dYvmBBKRxFAA1KlNDS0ikgQKgDql/gJvXppgbGo27lJERNadAqBObWro13UUICIJoACoM3dvAAWAiCSAAqBOX2+O3fmcpoQQkURQACyiqaFFJCkUAIsMFAucvDjO9ZlK3KWIiKwrBcAipf48VYfj5zUOICLtTQGwSG0gWOMAItLumgoAM9tqZk+b2U/N7LiZPWBm283sOTM7EZ63hb5mZl82s5Nm9hMzu3dtfoS1tbfQyfburKaGFpG21+wRwJeA77j7u4D3AMeBx4Hn3f0Q8HxYB3gIOBQejwFfafK914WZaWpoEUmEVQeAmeWBXwWeBHD3GXe/BjwMPBW6PQV8NCw/DHzdIy8BW81s76orX0elYoGfXxhjuqyBYBFpX80cAdwODAN/a2avmtlXzawb2O3u5wDC867Qvwicrtt/KLRtOqX+AuWq8/Pz43GXIiKybpoJgAxwL/AVd78HmGD+dE8j1qDNb+hk9piZHTGzI8PDw02Ut3qloqaGFpH210wADAFD7v5yWH+aKBAu1E7thOeLdf331+2/Dzi7+EXd/Ql3H3T3wb6+vibKW73btnfR25nR1NAi0tZWHQDufh44bWZ3haYHgdeBZ4FHQtsjwDNh+VngE+HbQPcDI7VTRZvN3ECw5gQSkTaWaXL/PwK+YWZZ4BTwSaJQ+WczexR4G/hY6Ptt4MPASWAy9N20Sv0Fvv7SW8xWqnSkdbmEiLSfpgLA3X8MDDbY9GCDvg58qpn320ilYoGZcpU3hsd515583OWIiKw5/Wl7E3NTQ+t6ABFpUwqAmzi4s5uubFoDwSLSthQAN5FOGYf35jmmr4KKSJtSACyhVCxw7Owo1eoNlyuIiLQ8BcASBvrzTM5UePPyRNyliIisOQXAEuYHgnUaSETajwJgCXfu6iGbSXFMF4SJSBtSACyhI53i7j29vDakIwARaT8KgGUMFKObxEfXsYmItA8FwDJK/QXGpsqcvnI97lJERNaUAmAZmhpaRNqVAmAZ79zdSyZl+iaQiLQdBcAyOjvSHNrdq6mhRaTtKABWoNSf59gZDQSLSHtRAKxAqVjg8sQM50en4i5FRGTNKABWYG4gWFNDi0gbUQCswN1786RMU0KISHtRAKxAVzbDHX09mhpaRNqKAmCFSsWCTgGJSFtRAKzQQH+e86NTDI9Nx12KiMiaUACsUG1qaJ0GEpF2oQBYocP90TeBNDW0iLQLBcAK5Ts7OLCjS1NDi0jbUADcgtrU0CIi7UABcAtK/QWGrl7n2uRM3KWIiDRNAXALalcEaxxARNqBAuAWDPTrJvEi0j4UALdge3eW4tYtmhpaRNqCAuAWDYSpoUVEWp0C4BaVigVOXZpgbGo27lJERJrSdACYWdrMXjWzfw3rB83sZTM7YWb/ZGbZ0J4L6yfD9gPNvnccagPBx8+NxVyJiEhz1uII4NPA8br1vwK+4O6HgKvAo6H9UeCqu98JfCH0azm1KSE0ECwira6pADCzfcBvA18N6wZ8AHg6dHkK+GhYfjisE7Y/GPq3lF29nezqzemCMBFpec0eAXwR+HOgGtZ3ANfcvRzWh4BiWC4CpwHC9pHQv+WUigWOaWpoEWlxqw4AM/sIcNHdX6lvbtDVV7Ct/nUfM7MjZnZkeHh4teWtq1J/nhMXx7g+U4m7FBGRVWvmCOD9wO+a2S+AfyQ69fNFYKuZZUKffcDZsDwE7AcI2wvAlcUv6u5PuPuguw/29fU1Ud76GSgWqDr89LyOAkSkda06ANz9L919n7sfAD4OfN/d/wPwAvD7odsjwDNh+dmwTtj+fXe/4QigFcwNBOuCMBFpYetxHcBfAJ8xs5NE5/ifDO1PAjtC+2eAx9fhvTdEf6GTbV0duiBMRFpaZvkuy3P3HwA/CMungPsa9JkCPrYW7xc3M6NULPCaAkBEWpiuBF6lgf4CP78wxnRZA8Ei0poUAKtUKuaZrTgnLozHXYqIyKooAFappKmhRaTFKQBW6bbtXfTmMroiWERalgJglVIp43B/nqO6IlhEWpQCoAmlYoHj50YpV6rLdxYR2WQUAE0oFfNMl6u8MTwRdykiIrdMAdAEDQSLSCtTADTh9r4etnSkNRAsIi1JAdCEdBgI1tTQItKKFABNKvXnOXZ2hGq1Jee1E5EEUwA0aaBYYGKmwi8uayBYRFqLAqBJcwPBmhpaRFqMAqBJh3b3kE2nNDW0iLQcBUCTOtIp3rW3V1NDi0jLUQCsgYH+AkfPjNCiNzgTkYRSAKyBUjHP6FSZoavX4y5FRGTFFABrQFcEi0grUgCsgbv29JJOma4IFpGWogBYA50daQ7t6tHU0CLSUhQAa6RU1ECwiLQWBcAaKfXnuTwxw4XR6bhLERFZEQXAGikVNRAsIq1FAbBGDvfnMUMDwSLSMhQAa6Qrm+GOPg0Ei0jrUACsodrU0CIirUABsIZKxQLnRqa4NK6BYBHZ/BQAa2ggXBF8TFNDi0gLUACsocP9eUDfBBKR1qAAWEOFLR28Y0eXxgFEpCWsOgDMbL+ZvWBmx83smJl9OrRvN7PnzOxEeN4W2s3MvmxmJ83sJ2Z271r9EJtJqb+gewOISEto5gigDPyZu98N3A98yswOA48Dz7v7IeD5sA7wEHAoPB4DvtLEe29aA8U8p69cZ2RyNu5SRESWtOoAcPdz7v6jsDwGHAeKwMPAU6HbU8BHw/LDwNc98hKw1cz2rrryTao0NxCsowAR2dzWZAzAzA4A9wAvA7vd/RxEIQHsCt2KwOm63YZCW1sZqA0EKwBEZJNrOgDMrAf4F+BP3H2p7z9ag7Ybps40s8fM7IiZHRkeHm62vA23oydHf6FTVwSLyKbXVACYWQfRL/9vuPu3QvOF2qmd8HwxtA8B++t23wecXfya7v6Euw+6+2BfX18z5cVmoFjQEYCIbHrNfAvIgCeB4+7+13WbngUeCcuPAM/UtX8ifBvofmCkdqqo3ZT6C7x5aYLx6XLcpYiI3FQzRwDvB/4Q+ICZ/Tg8Pgx8DvigmZ0APhjWAb4NnAJOAv8L+M9NvPemVirmcYfj53QaSEQ2r8xqd3T3/0Pj8/oADzbo78CnVvt+raT+3gC/cmB7zNWIiDSmK4HXwe58J329OQ0Ei8impgBYJ5oaWkQ2OwXAOikVC5y4OM7UbCXuUkREGlIArJOB/gKVqvPT82NxlyIi0pACYJ2UipoaWkQ2NwXAOilu3cLWrg6NA4jIpqUAWCdmpqmhRWRTUwCso4Finp+dH+M7R89xdWIm7nJERBZY9YVgsrwP3LWLv3/xLf7TP/wIM7h7T54H7tjBA7fv4L7bt5Pv7Ii7RBFJMIsu0N2cBgcH/ciRI3GX0ZSZcpWfDF3jxTcu83/fuMwrb19lplwlZdFXRWuB8CsHttOdUx6LSPPM7BV3H1y2nwJgY03NVnj17Wu8eOoyL71xmVdPX2W24mRSxrv3RYHwvjt28u/esY3OjnTc5YpIC1IAtIjrMxWOvHWFF9+4zIunLvOToREqVSebTvHLt23lgdt38MAdO7jntq3kMgoEEVmeAqBFjU+X+eEvrvBSOGV07OwIVYdcJsXggW1zgfDufVvpSGsMX0RutNIA0EnnTaYnl+HX79rFr98V3Ulz5Pos/+/N+SOEz3/v5wB0ZdMMHtjO+8IYwkB/nowCQURugY4AWsyViRlePhWFwYtvXObExXEAenMZ7ju4nQfu2MH9t+/g8N48qdTNZusWkXamI4A2tb07y0O/tJeHfmkvAMNj07x0Kjpd9NKpyzz/0+gOnFu7Onjvwe3hlNFO3rm7h+gmbiIiEQVAi+vrzfE77+nnd97TD8D5kSlePHVp7pTRd49dAKA7m2ZPoZPd+U725DvZle9kTz7H7nwnu0P7rt6cxhVEEkQB0Gb2FDr5vXv28Xv37APg9JVJXjx1mdfPjnJxbIoLo9O8/OYVLo5NMVtZePrPDHZ0ZxeFRCe787koJHo72VPoZFtXh44mRNqAAqDN7d/exf7tXTe0V6vO1ckZzo9OcXF0mvOjU1yYe0xzbmSKfxu6xqXxG6ewyKZT7ApHD1FQ5EJQ1B459hQ66crqn5fIZqb/oQmVShk7enLs6Mkx0H/zfjPlKsPj05wfmQ+IudAYmeL4+VF+8LMpJmZuvPFNb2fmhpDY3p2lJ5ehO5ehpzNDT27hozuXIZvRaSiRjaAAkCVlMymKW7dQ3LplyX7j02XOj0xxMQREfUhcGJvipTfGuTg2Tbm6/LfOspkUvbWQqD0669fT9OQ66M6l6e1s0C+bmWvXmIbIzSkAZE305DLcuauHO3f13LRPteqMz5QZnyozMV1mbDp6Hp9auFzfZzw8Lo5NMXGpwlhov77CW23mMqkFwVA76ujOZejqSLMlm6Y7l6Yrm6Erm6Yrm2ZLNkN3NtrWVbfcnc2wJZsml0lpDETaggJANkwqZeQ7O9ZkFtRypcrETCUKiKn5oJhosD62qP3C6BSTMxUmZ8pMTleYnK1QWcGRydzPYcyFQXcuw5aOKDi6Qqh05cJ6g1Cpb4/CJAqUbCZFNh2eMykyKVPIyLpTAEhLyqRTFLakKGxpPkzcnelyleszURhMTpdDQISQWPBctxzC4/pMmYnpCiPXZzk/cp2J6QrXZytMTJeZLldXVZMZc4GQWxQOtbDIZdIL2nLphdtv7F+/Pr9vru65Fkhzyx3RvrqosD0pACTxzIzOjjSdHWm2rfFrV6rO9SVCZbpcZaZcZaYSnuvWa9sW9qks6D85GV6jsnDf2vJKxlxWoiNtC8IhWxcQC9YXb8+kQojM960F11zQdKTnAqsjbXSko+dMKkUmbWTTKTLpaLkjFW1L6whpTSgARNZROmVzA9RxqFSd2criEKkPiwrTs1WmK1WmZ0PwzFaYLtcHUFifresf2msBNTVbZeT67Nx61Gd++1oFUb1aSNRCI3PDei1IaqEy3yebmQ+YjtpzrX86RdqikMmkjFR4Tqfm29KpFOkUpFOpBX1SFranjbQt3C/aN0UqBZlUasHrLXgPi/bPplPrPiW8AkCkjUW/ZNKx31uiXJkPn/qAqAVNLSzKlSiwZqtOuRKtz1Sq0XLVmQ3by3V95tuc2Wrda1SccnX+NSZnynOvEe13Y59yWK9UnXXIrFvykXfv5W/+/b3r+h7tGQDT43Du36B7J3TthC3bIKWvA4rEJRP+Iu/Kxl3Jyrk7lapTrjpVj54rFadS3x6eK9UqlSpz4bH4Ua6G/SqLXi/sV6lWF72ec3Bn97r/jO0ZAMM/g7/78Py6paFrexQG3Tvng6G7D7p31C0rMEQkYmbRKaM2vg9TewbAzkPwiWdg4lL0mAzPE8MweRnOH42Wp6413t9S0LWjQWAsXu5TYIhIy2rPAOjMw+2/tny/yixMXgnBUAuLy9F6fXCcPxotX7/a+HUWB0bXjvkjii3bIN0BqQykwnM6E9ZrbenQXuuXnu+70v4iIrdowwPAzD4EfAlIA191989tdA1z0h3Quzt6rEQtMCYv1YXE4sC4DBeOLR0Ya87qwqIuIFL1gZKO+s19da62HNYbLRuL+q50v8V965YtVVdnqHVuuRZ2HXU/R0dd0GUaLC/a72Z90nUBmq77Z1+7IZI74Ms/L9mXBu2N2uq3Lf7clnrmJu2pFb7GMu9lqcbLc20r7ZtqsJ9sRhsaAGaWBv4n8EFgCPihmT3r7q9vZB2rtprAmBqF6ixUy9F6tRIt19qqldBea6vU9Q3b12T/Co1/KS1epvEvu/rlJV/jZvuF52oFylN1P8tsXa21umfrfq6wTVrcCsNiwR8TzK8vWFyqj61Pn4b7Lfr5FvRZg7ZDH4Tf+h+sp40+ArgPOOnupwDM7B+Bh4HWCIBble6IBpmlObXgWBCE5QbBEUKwUq5bbtCnMhu97lJHL8v9JX0rf7Ev9XpLHnWwzPa6Z6822MbKj2zmnquLXnep96jeZJlb6Ft73Vqt4WdesN6obSP60KBt0X7r2ZYvst42OgCKwOm69SHgvfUdzOwx4DGA2267beMqk83LLDptk97wM5YibW2jv7rS6GTgwpx1f8LdB919sK+vb4PKEhFJno0OgCFgf936PuDsBtcgIiJsfAD8EDhkZgfNLAt8HHh2g2sQERE2eAzA3ctm9l+A7xJ9DfRr7n5sI2sQEZHIho+qufu3gW9v9PuKiMhCmr9ARCShFAAiIgmlABARSSjzBVfAbS5mNgy81cRL7AQurVE5rU6fxUL6PObps1ioHT6Pd7j7shdSbeoAaJaZHXH3wbjr2Az0WSykz2OePouFkvR56BSQiEhCKQBERBKq3QPgibgL2ET0WSykz2OePouFEvN5tPUYgIiI3Fy7HwGIiMhNtGUAmNmHzOxnZnbSzB6Pu544mdl+M3vBzI6b2TEz+3TcNcXNzNJm9qqZ/WvctcTNzLaa2dNm9tPwb+SBuGuKk5n9afh/ctTMvmlmnXHXtJ7aLgDqbjv5EHAY+AMzOxxvVbEqA3/m7ncD9wOfSvjnAfBp4HjcRWwSXwK+4+7vAt5Dgj8XMysCfwwMunuJaMLKj8db1fpquwCg7raT7j4D1G47mUjufs7dfxSWx4j+g6//veY2KTPbB/w28NW4a4mbmeWBXwWeBHD3GXe/Fm9VscsAW8wsA3TR5vcraccAaHTbycT+wqtnZgeAe4CX460kVl8E/hyoxl3IJnA7MAz8bTgl9lUz6467qLi4+xng88DbwDlgxN2/F29V66sdA2DZ204mkZn1AP8C/Im7j8ZdTxzM7CPARXd/Je5aNokMcC/wFXe/B5gAEjtmZmbbiM4WHAT6gW4z+4/xVrW+2jEAdNvJRcysg+iX/zfc/Vtx1xOj9wO/a2a/IDo1+AEz+4d4S4rVEDDk7rUjwqeJAiGpfgN4092H3X0W+BbwvphrWlftGAC67WQdMzOic7zH3f2v464nTu7+l+6+z90PEP27+L67t/VfeEtx9/PAaTO7KzQ9CLweY0lxexu438y6wv+bB2nzQfENvyPYetNtJ2/wfuAPgdfM7Meh7b+GO7OJ/BHwjfDH0ingkzHXExt3f9nMngZ+RPTtuVdp86uCdSWwiEhCteMpIBERWQEFgIhIQikAREQSSgEgIpJQCgARkYRSAIiIJJQCQEQkoRQAIiIJ9f8B72qLIjXdl1AAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure()\n",
    "plt.plot(train_error)\n",
    "plt.plot(valid_error)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "torch.save(cnn_network.state_dict(), \"Network/cnn_network\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "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": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}