PyTorch_23_NeuralNetwork1D.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "PyTorch_23_NeuralNetwork1D.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyO7McePsImibJc2rMddLYse",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/codeplinth/pytorchbootcamp/blob/master/PyTorch_23_NeuralNetwork1D.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JoQAZdEsDpvG",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import torch\n",
        "from torch import nn,optim\n",
        "from torch.utils.data import Dataset,DataLoader\n",
        "import matplotlib.pyplot as plt\n",
        "%matplotlib inline"
      ],
      "execution_count": 59,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gUFg9-3MALUF",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "58849c7c-3f39-40d5-a90a-42fd9095e87d"
      },
      "source": [
        "torch.manual_seed(0)"
      ],
      "execution_count": 60,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<torch._C.Generator at 0x7f60e99fd0c0>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 60
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Z16q9B1l2O9n",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#make some sample data\n",
        "X = torch.arange(-20,20,1).view(-1,1).type(torch.FloatTensor)\n",
        "Y = torch.zeros(X.shape[0])\n",
        "Y[(X[:, 0] > -4) & (X[:, 0] < 4)] = 1.0"
      ],
      "execution_count": 61,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1QfjIlvVATCx",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# The function for plotting the model\n",
        "def PlotStuff(X, Y, model, epoch, leg=True):\n",
        "    plt.plot(X.numpy(), model(X).detach().numpy(), label=('epoch ' + str(epoch)))\n",
        "    plt.plot(X.numpy(), Y.numpy(), 'r')\n",
        "    plt.xlabel('x')\n",
        "    if leg == True:\n",
        "        plt.legend()\n",
        "    else:\n",
        "        pass"
      ],
      "execution_count": 62,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2PW7LElTCTVj",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#Define the network\n",
        "class NeuralNet(nn.Module):\n",
        "  def __init__(self,D_in,H,D_out):\n",
        "    super(NeuralNet,self).__init__()\n",
        "    self.l1 = nn.Linear(in_features = D_in,out_features = H)\n",
        "    self.l2 = nn.Linear(in_features = H,out_features = D_out)\n",
        "    self.sigmoid = nn.Sigmoid()\n",
        "  def forward(self,x):\n",
        "    out1 = self.sigmoid(self.l1(x))\n",
        "    y_pred = self.sigmoid(self.l2(out1))\n",
        "    return y_pred\n"
      ],
      "execution_count": 63,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-rCFJ_J51p1c",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 104
        },
        "outputId": "735b7cba-a2de-4d54-c243-5d8e85cad202"
      },
      "source": [
        "#create the model\n",
        "model = NeuralNet(1,2,1)\n",
        "model.state_dict()"
      ],
      "execution_count": 64,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "OrderedDict([('l1.weight', tensor([[-0.0075],\n",
              "                      [ 0.5364]])),\n",
              "             ('l1.bias', tensor([-0.8230, -0.7359])),\n",
              "             ('l2.weight', tensor([[-0.2723,  0.1896]])),\n",
              "             ('l2.bias', tensor([-0.0140]))])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 64
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NM5vp9Ed5pw0",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#create optimizer,criterion\n",
        "optimizer = optim.SGD(model.parameters(),lr = 0.01)\n",
        "criterion = nn.BCELoss()"
      ],
      "execution_count": 65,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sp92Y9ww6vKy",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "LOSS = []\n",
        "EPOCHS = 3000\n",
        "def train_model(EPOCHS):\n",
        "  for epoch in range(EPOCHS):\n",
        "    for x,y in zip(X,Y):\n",
        "      y_pred = model(x)\n",
        "      loss = criterion(y_pred,y)\n",
        "      optimizer.zero_grad()\n",
        "      loss.backward()\n",
        "      optimizer.step()\n",
        "    LOSS.append(loss.item())\n",
        "    if ( epoch % 300 == 0 ):\n",
        "      PlotStuff(X, Y, model, epoch, leg=True)\n",
        "      plt.show()"
      ],
      "execution_count": 66,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aJE4bqGKAj-B",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "fa56be41-6c4d-4366-b88d-ec51ea611018"
      },
      "source": [
        "train_model(EPOCHS)"
      ],
      "execution_count": 67,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/loss.py:529: UserWarning: Using a target size (torch.Size([])) that is different to the input size (torch.Size([1])) is deprecated. Please ensure they have the same size.\n",
            "  return F.binary_cross_entropy(input, target, weight=self.weight, reduction=self.reduction)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/loss.py:529: UserWarning: Using a target size (torch.Size([])) that is different to the input size (torch.Size([1])) is deprecated. Please ensure they have the same size.\n",
            "  return F.binary_cross_entropy(input, target, weight=self.weight, reduction=self.reduction)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/loss.py:529: UserWarning: Using a target size (torch.Size([])) that is different to the input size (torch.Size([1])) is deprecated. Please ensure they have the same size.\n",
            "  return F.binary_cross_entropy(input, target, weight=self.weight, reduction=self.reduction)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/loss.py:529: UserWarning: Using a target size (torch.Size([])) that is different to the input size (torch.Size([1])) is deprecated. Please ensure they have the same size.\n",
            "  return F.binary_cross_entropy(input, target, weight=self.weight, reduction=self.reduction)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXQAAAEGCAYAAAB1iW6ZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3deXScd33v8fd3RqPRLtuSHDvet8QbkEU2IUkJvSEkcIuDKQ3OgSZlaZImueVCCyct9+RCmjZQynIhpm0KKRBCDAmFGMhCGxIoYbGd1VsWx3ZiyXYsWba2kUbS6Hf/eJ6Rx7KW0WismWfm8zpHR7M8I331ePzRT7/nt5hzDhERCb5QrgsQEZHsUKCLiBQIBbqISIFQoIuIFAgFuohIgSjJ1Teur693CxcuzNW3FxEJpKeeeqrVOdcw0nM5C/SFCxeybdu2XH17EZFAMrNXR3tOXS4iIgVCgS4iUiAU6CIiBSJnfegiEnz9/f00NTXR29ub61IKTllZGXPnziUSiaT9GgW6iGSsqamJ6upqFi5ciJnlupyC4Zzj6NGjNDU1sWjRorRfN26Xi5ndbWZHzGzHKM+bmX3VzPaY2fNmdt4E6haRAOvt7aWurk5hnmVmRl1d3YT/8kmnD/1bwBVjPP9OYJn/cR3wzxOqQEQCTWF+emRyXsftcnHO/crMFo5xyJXAd5y3Du/vzGyamc12zh2acDUimXjiCfjFL3JdRWZCIbjmGli8ONeVSAHIRh/6HOBAyv0m/7FTAt3MrsNrxTN//vwsfGsR4FOfgq1bIYgtReeguxu+8IVcVyIjeOKJJ/inf/onfvrTn4553HPPPccNN9xAV1cXCxcu5N5776WmpgaAO+64g29+85uEw2G++tWvcvnllwPwyCOP8LGPfYxEIsFHP/pRbrnllknXO6XDFp1zdznnGp1zjQ0NI85cFZm4zk646ioYHAzex8yZ0NWV6zMok/TRj36Uz33uc2zfvp3169fzBf8X9K5du9i0aRM7d+7kkUce4cYbbySRSJBIJLjpppt4+OGH2bVrF/fddx+7du2adB3ZCPRmYF7K/bn+YyJTIxaDiopcV5GZigqvfsnYd7/7XdauXcs555zD9ddfTyKRAKCqqoqPf/zjrFq1iksvvZSWlhYAnn32WS644ALe+MY3sn79eo4dOwbAnj17ePvb386b3vQmzjvvPF555RUAurq6eN/73sfy5cv5wAc+wEi7vL300ku89a1vBeCyyy7jhz/8IQAPPvggGzZsIBqNsmjRIpYuXcqWLVvYsmULS5cuZfHixZSWlrJhwwYefPDBSZ+LbHS5bAZuNrNNwJuBdvWfy5RSoOeFz/5kJ7sOdmT1a648s4b/++5Voz6/e/duvv/97/Pkk08SiUS48cYbuffee7nmmmvo7u6msbGRL3/5y9x222189rOf5c477+Saa67ha1/7Gpdccgm33norn/3sZ/nKV77CBz7wAW655RbWr19Pb28vg4ODHDhwgGeeeYadO3dy5plnctFFF/Hkk09y8cUXn1THqlWrePDBB3nPe97D/fffz4EDXi90c3MzF1xwwdBxc+fOpbnZa+/OmzfvpMd///vfT/p8pTNs8T7gt8DZZtZkZh8xsxvM7Ab/kIeAvcAe4N+AGyddlchEKNCL1mOPPcZTTz3FmjVrOOecc3jsscfYu3cvAKFQiPe///0AfPCDH+TXv/417e3tHD9+nEsuuQSAa6+9ll/96ld0dnbS3NzM+vXrAW9ST4X/nlq7di1z584lFApxzjnnsH///lPquPvuu/n617/O+eefT2dnJ6WlpVPw058qnVEuV4/zvANuylpFIhPhnAI9T4zVkj5dnHNce+213HHHHeMem+nwymg0OnQ7HA4zMDBwyjHLly/n5z//OeB1v/zsZz8DYM6cOUOtdfAmYs2ZMwdg1McnQ2u5SLAlJ14o0IvSpZdeygMPPMCRI0cAaGtr49VXvdVlBwcHeeCBBwD43ve+x8UXX0xtbS3Tp0/nv//7vwG45557uOSSS6iurmbu3Ln8+Mc/BiAejxObwL9L8vsPDg5y++23c8MNXgfGunXr2LRpE/F4nH379vHyyy+zdu1a1qxZw8svv8y+ffvo6+tj06ZNrFu3btLnQ1P/JdiS/+mCHOhNTbmuIrBWrlzJ7bffzjve8Q4GBweJRCJs3LiRBQsWUFlZyZYtW7j99tuZOXMm3//+9wH49re/zQ033EAsFmPx4sX8+7//O+CF+/XXX8+tt95KJBLh/vvvT7uO++67j40bNwLw3ve+lw996EOA17d+1VVXsXLlSkpKSti4cSPhcBiAO++8k8svv5xEIsGHP/xhVq2a/F84NtIV26nQ2NjotMGFTNqBAzB/PnzjG/CRj+S6mon70z+F3/wG/BEVQbN7925WrFiR6zJGVFVVRVfAh4SOdH7N7CnnXONIx6vLRYKtEFro6nKRLFGgS7B1d3ufKytzW0emKitP/AySVUFvnWdCgS7BVigt9Bx1fWZDrrptC10m51WBLsFWCIGeSEB/f64ryUhZWRlHjx5VqGdZcj30srKyCb1Oo1wk2Aoh0MH7OXI0GWUy5s6dS1NT09C0esme5I5FE6FAl2ArpECfNi23tWQgEolMaEcdOb3U5SLBVkiBLjJJaqFLsJ3GQG/v6WfXwQ4Otfew8swals2sJhzK8prrCnTJIgW6BFuWAv1oV5ydBzvYcbCdnc3e51ePnhyyVdESzpk3jfPmT+Pc+dM5d/40plVMst9bgS5ZpECXYIvFvJ2KUhZQmogXDndw43efZm/ribHg82dUsHpODVc1zmP1nFpm15axo7mdp187xtOvHufOx/cw6A/qWNxQyTtXz+KvLjubUCatdwW6ZJECXYKtu9sLxQxW0ms6FuPau7cA8LfvWs7qObWsml1LbUXklGPPOqOa957njTjojg/wXNNxnnntOL/be5SNj79COBTiE5edNfH6k4GuyUWSBQp0CbZYLKNZom3dfVxz9xZ6+hLcf8OFnD2rOu3XVkZLuHBJPRcuqefGty3hUw88z1cfe5llM6t495vOnFghydrVQpcs0CgXCbYM1kKP9Q3woW9tpflYD9+4ds2Ewnw4M+P29atZs3A6f33/czzfdHxiX0BdLpJFCnQJtgkGen9ikL/47tNsbzrO164+l7WLZky6hGhJmH/+4PnUV0X58+9s43B7b/ovVqBLFinQJdgmEOiDg45PPfA8v3yphX9Y/wbesWpW1sqor4ryzT9rpKt3gOvu2UZvfyK9FyrQJYsU6BJsEwj0zz3yAj96ppm/fsdZbFg7P+ulLJ9Vw1c2nMv25nY++cDz6a1vUl7ufVagSxYo0CXY0gz0u371Cnf9ai/XvmUBN/3h0tNWzmUrz+BTly/nJ88d5M5f7Bn/BaEQlJUp0CUrFOgSbGkE+n883cQ/PPQC//ONs7n13asy3iw4XTdcspj3njuHL/7nSzy8/dD4L9AmF5IlCnQJtnECPT6Q4G9/tJ03L5rBl656U/an7o/AzPiH976Bc+dP4xM/eI4dze1jv0CBLlmiQJdgGyfQXzrcRW//INe8ZSHRkvCUlVUWCfOvf3o+NeUl3PHw7rEPVqBLlijQJdiSM0VHseOg1zpePadmqioaMrO6jEtXnMH2pvaxL5BWVGimqGSFAl2Cy7lxZ4ruaG6nuqyE+TNys7zuilnVdPQOcGissemVlWqhS1Yo0CW4+vu97dvGaqE3t7P6zNrTfiF0NCtme38Z7D7UMfpB6nKRLFGgS3CNs3Ruf2KQ3Yc7c9LdkpRcVuCFw52jH6RAlyxRoEtwjRPoe4500TcwyOo5tVNY1MmqyyLMm1HOLrXQZQoo0CW4xgn05HDBVWfmLtDBm0H6ggJdpoACXYIrjUCvLA2zuH7iy+tm04rZNexr7R59fRcFumSJAl2Ca7xAP9jByjNrMttJKItWzKpm0MFLr4/Sj65AlyxJK9DN7Aoze9HM9pjZLSM8P9/MHjezZ8zseTN7V/ZLFRlmjEBPDDp2HezIeXcLpDHSpaIC4nFvxI7IJIwb6GYWBjYC7wRWAleb2cphh/0f4AfOuXOBDcDXs12oyCmSk3FGCPR9rV309CdyekE0af6MCipKw+w+NEYLHdRKl0lLp4W+FtjjnNvrnOsDNgFXDjvGAcmxYbXAweyVKDKKZACOMLFoR7PXGn5DHgR6KGScPat69Ba6tqGTLEkn0OcAB1LuN/mPpfoM8EEzawIeAv7XSF/IzK4zs21mtq2lpSWDckVSjNHlsr25nWhJiCUNub0gmrR8Vg0vHO4ceQkAtdAlS7J1UfRq4FvOubnAu4B7zOyUr+2cu8s51+ica2xoaMjSt5aiNUag72huZ8XsGkrC+XHdf+Xsatp7+kdeAkCBLlmSzru9GZiXcn+u/1iqjwA/AHDO/RYoA+qzUaDIqEYJ9EH/gmguZ4gOt9y/MPrC4RG6XRTokiXpBPpWYJmZLTKzUryLnpuHHfMacCmAma3AC3T1qcjplQzA5DZuvtfaYnTGB/Ki/zwpuQTAiBdGFeiSJeMGunNuALgZeBTYjTeaZaeZ3WZm6/zD/gr4czN7DrgP+DOX1oaKIpMQi3nbt4VOfhsnl8zNhyGLSTVlEeZOLx/5wqgCXbKkJJ2DnHMP4V3sTH3s1pTbu4CLsluayDhG2dxie3M7kbBx1hnVOShqdCtm1yjQ5bTKjytGIpkYJdB3Nndw9qxqSkvy6+29Ylb1yEsAKNAlS/LrHS8yESPsVuScY8fB9rzqP09aMbtm5CUAkj+Ddi2SSVKgS3CN0EJvPt7D8Vh/XvWfJw2NdBl+YVQtdMkSBboE1wjbzyVniObDlP/hFvhLAJyyNroCXbJEgS7BNUILfUdzO+GQsXxWfl0QhRNLAJwyFj0S8T4U6DJJCnQJrpEC/WA7y2ZWURYJ56iosS2fVcPuQyMsAaAldCULFOgSXMMC3TnnbQqdh90tScklAA53DFsCQIEuWaBAl+AaFuhHOuO0dvWx+sz8mfI/3PLR1kZXoEsWKNAluIYFenIP0XxuoY+6BIACXbJAgS7BNSzQtze3YwYr87iFPuoSAAp0yQIFugRTIuFt23ZSC72DJQ1VVJSmtaJFzqyY7a2NfpKKCk0skklToEswjbB07s6D7Xndf560YlY1e1u6Tl4CQC10yQIFugTTsO3nWrviHGrvzev+86TkEgAvv9514sHKSgW6TJoCXYJpWAs9eUE0H6f8DzfiSBe10CULFOgSTMMCfedBLxxX5dEuRaNZMKOC8kiY3YcV6JJdCnQJphFa6AvrKqgpi+SwqPQklwBQC12yTYEuwTQ80A+2syoA/edJyZEuQ0sAJANdG33JJCjQJZhSAr091s+Bth5WB6D/PGnF7GqOx1KWAEiO1untHf1FIuNQoEswpQR6cg/RfNzUYjQrhq+NriV0JQsU6BJMyUk4FRUpI1zy/4JoUnIJgKG10bVrkWSBAl2C6aQWegdzppUzvbI0tzVNQHIJgKEZo2qhSxYo0CWYUgL9lSNdQy3eIPHWRh/WQlegyyQo0CWYUgK9pSvOzOpobuvJwMrZKUsAJLfSU6DLJCjQJZhiMYhEGAyX0NbdR31V8AJ9eeoSAGqhSxYo0CWY/KVzj8X6SAw66quC03+elBzpsvtwhwJdskKBLsHkB/rR7j4A6gPY5TJ/RgXRkhB7jqiFLtmhQJdg8gO9tTMOQF1l8AI9HDIaqqO0dMYV6JIVCnQJJj/QW7q8QG+oDl6XC0B9VZTWLgW6ZIcCXYKpu9troXf5XS4BvCgKXt0ntdA1sUgmQYEuwZTscumKEwkbteX5v8riSBqqS71fStEomKmFLpOSVqCb2RVm9qKZ7TGzW0Y55ioz22VmO83se9ktU2SY5EXRrjh1lVHMLNcVZaShKkpbd5yEQ0voyqSNu5uumYWBjcBlQBOw1cw2O+d2pRyzDPgb4CLn3DEzm3m6ChYBUlrofdQFcMhiUn11lEEHx2J91CvQZZLSaaGvBfY45/Y65/qATcCVw475c2Cjc+4YgHPuSHbLFBkmFoPKSlq74oHtP4cTff8tnXHtKyqTlk6gzwEOpNxv8h9LdRZwlpk9aWa/M7MrRvpCZnadmW0zs20tLS2ZVSwCJw1bLIRAHxrpokCXScjWRdESYBnwNuBq4N/MbNrwg5xzdznnGp1zjQ0NDVn61lKUYjFceTmtXX3UB3TIIjA0w1WBLtmQTqA3A/NS7s/1H0vVBGx2zvU75/YBL+EFvEj2OQexGPHSMvoSgzQEuYXuz3Bt7exToMukpRPoW4FlZrbIzEqBDcDmYcf8GK91jpnV43XB7M1inSIn+Nu0dYe91m2Qu1yqoyWUloTUQpesGDfQnXMDwM3Ao8Bu4AfOuZ1mdpuZrfMPexQ4ama7gMeBTzrnjp6uoqXI+aHX5Qd6kEe5mBkNVVFvxqsCXSZp3GGLAM65h4CHhj12a8ptB3zC/xA5vfzZlO3mTSYKcgsdvH701i6/y0UzRWUSNFNUgsdvxR6nUAI96i0ypha6TJICXYLHD71jroSQwYwA7SU6kpMW6FKgyyQo0CV4/NBrJcKMylLCoWBO+0+qry7laHcfLhnozuW6JAkoBboEjx/oLYlQINdBH66hKkpi0NETKYNEAvr7c12SBJQCXYLHD/QjA+FATypKSo5F7wr5K0aq20UypECX4PED7/BAKPAXROHERd2OkP/LSYEuGVKgS/D4gXeo3woq0NvNH0WsQJcMKdAlePzAa3ORggj05NIFx5y6XGRyFOgSPP7km95IdGhxqyCrKS+hNBziqPNb6JpcJBlSoEvwxGI4M+Lhwmihmxl1VaUcHQx7D6iFLhlSoEvwxGIkysrBCqMPHbx+9CMJBbpMjgJdgicWY6CsHKAghi2Ct57L6wP+f0cFumRIgS7BE4sRj3qBXggTi8BroR/qVwtdJkeBLsETixGPRKktj1BaUhhv4frqKIf6/SUMFOiSocL43yDFJRajpyQa6HXQh6uvimpikUyaAl2CJxaju6S0YC6IgteH3htRoMvkKNAleGIxOkOlgd5LdLiG6ijOQiSiZQp0yZgCXYKnu5vOUKQgJhUlJX85JcrKNLFIMqZAl8BxsRgdoULrcvF+lr5ouVrokjEFugTOYHc3vSXRoWVnC0FteYSSkNEXiSrQJWMKdAmeWA89kSh1Ad96LlUo5E3/71GgyyQo0CVYnMN6YvRECquFDl63S6xEgS6ZU6BLsPT3E0okiEXKCmqUC3iB3qVAl0lQoEuw+GHXWxItqIui4AV6p0UU6JIxBboEix92g+VllJeGc1xMdtVXl9JuEZwCXTKkQJdg8cMuXFWV40Kyr6EqSndJFNetQJfMKNAlWPxJN6U1hRfo9VVReiJRnCYWSYYU6BIsfgu9rEADvbckivWohS6ZUaBLsPiBXj6tOseFZF9DtddCD/X1wcBArsuRAFKgS6AkOrsAqJxek+NKsq++qpSeEn/kTk9PbouRQFKgS6B0H+8EoLoAA316RSnxUj/QNdJFMpBWoJvZFWb2opntMbNbxjjuj83MmVlj9koUOaHrWAcAVXXTclxJ9oVCRig5ekeBLhkYN9DNLAxsBN4JrASuNrOVIxxXDXwM+H22ixRJih33An1aXW2OKzk9IjUKdMlcOi30tcAe59xe51wfsAm4coTj/g74PNCbxfpETtLb7vWhT68vzECPKtBlEtIJ9DnAgZT7Tf5jQ8zsPGCec+5nY30hM7vOzLaZ2baWlpYJFysS7/ACvW5m4XW5AFTU+qN3FOiSgUlfFDWzEPAl4K/GO9Y5d5dzrtE519jQ0DDZby1FqL+jk96SUqrLC2fp3FQV/nBM19WV40okiNIJ9GZgXsr9uf5jSdXAauAJM9sPXABs1oVROR0GurqJR6KYWa5LOS2SwzF72ztzXIkEUTqBvhVYZmaLzKwU2ABsTj7pnGt3ztU75xY65xYCvwPWOee2nZaKpagNdnV727QVqJoZXqB3HVOgy8SNG+jOuQHgZuBRYDfwA+fcTjO7zczWne4CRVK5WMzbSLlA1frDMZPDM0UmoiSdg5xzDwEPDXvs1lGOfdvkyxIZmfXESJQVbgt9mj96R10ukgnNFJXAGBx0hHt7cBUVuS7ltJnhj95RoEsmFOgSGB29/ZT1xbHKylyXctrUTauiL1RCX4dGucjEKdAlMFq74pQPxAkVcKCHQ0a8NMpAl9ZEl4lToEtgtHT2UdYfp6SqcAMdoK+0jIQCXTKgQJfAaO2KU94fL8jdilINRMu1r6hkRIEugdHaFaeiv/fEeicFKlFWhsXUQpeJS2vYokg+aO2KUzZQ+C30wYoKQtrgQjKgFroExrH2GNHEQEGPcgGwigpK+3rpjmsbOpkYBboERnubP3uygMehA4QqKynvj9PaFc91KRIwCnQJjM7k+iYFHujhKi/QWzoV6DIxCnQJjFhyfZMC73KJ1FRRNqAWukycLopKIDjn6GkvjhZ6tKaawf44LV19uS5FAkaBLoHQ3Zcg3OuP/Cj4QK+CgTit6nKRCVKXiwRCa6c3qQgo+EAPV1VS0R+ntVPb88rEKNAlELxZon7AFXigJ3++oVE9ImlSoEsgJKf9A0UT6NrkQiZKgS6B0NLVR/lAcQV6twJdJkiBLoFQTH3oyZ+v57g2uZCJUaBLIBztjjPD/KnwRRLoLhajpy+R42IkSBToEgitnX1FF+jlmlwkE6RAl0Bo7YoznQGIRLyPQubPhC3vj3NEY9FlAhToEgitXXFq6S/81jmcaKFrgS6ZIAW6BEJrVx81g8UV6FrPRSZKgS55r7c/QVd8gOpEvKgCvaKvl9ZOreci6VOgS95LtlIrBvqKKtCn24Ba6DIhCnTJe63+qoPlA8XVQq9ToMsEKdAl7yVXHYz29RZHoEejYMYMBbpMkAJd8l4y1EqLJdDNoKKCWtc/9NeJSDoU6JL3koFe0lskgQ5QUUGN69ea6DIhCnTJey2dcaqjJYR6YkUV6NWD/XTGB+jt1/R/SU9agW5mV5jZi2a2x8xuGeH5T5jZLjN73sweM7MF2S9VitXzze2cNasaYrGC3090SGUllf7qkupHl3SNG+hmFgY2Au8EVgJXm9nKYYc9AzQ6594IPAD8Y7YLleLU05dge1M7axbO8AK9iFro5UOBrn50SU86LfS1wB7n3F7nXB+wCbgy9QDn3OPOuZh/93fA3OyWKcXq2QPHGRh0rF1QW3SBXtbn7dDUon50SVM6gT4HOJByv8l/bDQfAR4e6Qkzu87MtpnZtpaWlvSrlKK1dX8bZnD+TD/IiynQ++OEDHY0t+e6GgmIrF4UNbMPAo3AF0Z63jl3l3Ou0TnX2NDQkM1vLQVq6/42zj6j2luYC4oq0Et6e1g9p5bfvNKa62okINIJ9GZgXsr9uf5jJzGztwOfBtY55/Q3okzaQGKQp189dqL/HIoq0InFuHBJPc+8dpzu+ECuK5IASCfQtwLLzGyRmZUCG4DNqQeY2bnAv+KF+ZHslynFaNehDrr7EqxZVLyBftHSOgYGHVv2t+W6IgmAcQPdOTcA3Aw8CuwGfuCc22lmt5nZOv+wLwBVwP1m9qyZbR7ly4mkbev+YwCsLeIWeuOCGZSGQ/xmj7pdZHwl6RzknHsIeGjYY7em3H57lusSYeu+NubNKGdWbVnRBnp5JMS586fxm1eO5roiCQDNFJW85Jxj6/421iyY4T2QDPQimlhEIgH9/Vy0tJ5dhzo41q3x6DI2Bbrkpb2t3Rzt7vP6z6E4W+gw1I/uHPx2r1rpMjYFuuSlrfu8i4BrFvqB3t3tfS62QO/u5o1zp1FZGuZJ9aPLOBTokpe27j9GXWUpSxr8LpYibqFHwiHevLiO36ofXcahQJe8tHV/G40Lp2Nm3gNFHOgAFy6pY29rN4fae3JYlOQ7Bbrkndc7enmtLXaiuwUU6EvqAXhyj1rpMjoFuuSdLcP7z8ELNjNve7ZiMCzQl8+qZkZlqcajy5gU6JJ3tu5vo6I0zKoza048mFxpMdkFU+iGBXooZLxlSR1PvtKKcy6HhUk+U6BL3tm6/xjnzZ9OSTjl7VlMS+fCKYEOcNGSel7viLO3tTtHRUm+U6BLXmnv6eeFwx0nd7eAAh3vwiigbhcZlQJd8srTrx7DOVizcPrJTxTT9nNw4mdNCfQFdRXMmVauC6MyKgW65JUt+9soCRnnzh8W6N3dxdlC7z7RvWJmXLikjt/uPUpiUP3ocioFuuSVbfvbWD2nlvLS8MlPFFuXS3m59zmlhQ5w0dJ62nv62XWwIwdFSb5ToEve6O1P8NyBdtYumnHqk8UW6KEQlJWdEuhD/ejaxUhGoECXvPF8Uzt9icFTL4hC8QU6DC2hm2pmTRlLZ1bxpJYBkBEo0CVvbPV35WlcMP3UJxXoQy5aUsfWfW30DQzmoCjJZwp0yRtb9rWxbGYV0ytLT31SgT7kwqX19PQneOa1YzkoSvKZAl3yQmLQeRtCj9R/Dgr0FBcsriNkaBcjOYUCXfLCC4c76IwPePuHDuecAj1FbXmEN8yp1YVROYUCXfLC0IYWI7XQ+/u97dgU6EPesqSeZ147Tnd8YIqLknymQJe8sHX/MeZMK2fOtPJTnyy2/USTKitHDfSLltYxMOjY4l9IFgEFuuQB57xgahw+3T+p2LafS6qoOGmmaKrGBTMoDYe0roucRIEuOfdaW4yWzvjI48+h+Da3SBqjy6W8NMx5C6bpwqicRIEuOdXW3cff/2w3AG8ea4QLKNCHuWhJPbsOdfBfu16fwqIknynQJWce3XmYd3z5lzz+4hFueedylp1RPfKBCvQR/UnjPJY2VPHR72zjhnue4nB77xQWJ/lIgS5Trj3Wz8e//yzX3/MUM6vL2HzzxdxwyZLRX1DMgd7XBwMjj2SZVVvGz/7yD/jk5Wfz+ItHePuXfsm3ntynlRiLmAJdptQvXnidy778S37y3EE+dukyHrz5IlbMrhn7RcUc6AA9PaMeUloS4qY/XMrPP/5Wzlswnc/8ZBfrv/4kO5rbp6hIyScKdJkSHb39fPL+5/jwt7YxvaKUH990ER+/7Cwi4TTegsUe6AFmuvkAAAisSURBVGN0uyQtqKvk2x9aw1evPpeDx3tZd+evue0nu+jSOPWiUpLrAqSwDCQGOXi8l/1Hu3m1LcZrR7t59WiMp187Tlt3nJv+cAl/eekyoiXh8b9YkgI9rcPNjHVvOpNLzmrgHx95gbuf3Md9W15j/owK5s2oYN6Mcu/29Arm13mfT1l3XgItcIH+4uFOtuvPyREN3w3+lJ5UB85/NHmoS7k96BzOOQadt7bKoEt+ePf7E4P09g/S258gPpBIuT1IT1+Cg+09NB/rYSClDzdaEmL+jArOXzCNv3jbUs6ZN23iP1gxTyyCtAM9qbY8wt+vfwN/fP5cfvLcQQ609XCgLcZvXmkl1pc4+VuUholGwpSVhIhGwkT9z8n7kZBhZoRDEDIjFDJCZoTNu495X8cwbOg2KbdtzFpt7Kcn5XR+7cm68pw5XLC4LutfN3CB/sSLR7jj4RdyXUbRKg2HiEZClEXClEVClJWEh26vnlPLH71xNgtmVDK/roKFdZXMrI4SCk3yf1YxTyyCUScXjee8+dM5L2UrP+ccR7v7ONAW47W2GAfaYhyL9Q/9Uo4PnPgF3dufoD3WR8I5EoPea0/8kvd++Scvvqa2I5xzQw0JN861WXdqkyNrxvveuda4YJQhupOUVqCb2RXA/wPCwDecc58b9nwU+A5wPnAUeL9zbn92S/VsWDufd71h9un40gVpeCvF7ESbyYa1rgyGWmAhI6U15j0fCYcITzacM5FsoZaPsCxAIZtgl8t4zIz6qij1VdFT92yVgjBuoJtZGNgIXAY0AVvNbLNzblfKYR8BjjnnlprZBuDzwPtPR8G15RFqyyOn40tLvorFvO3YQkV2DT/LgS6FL50W+lpgj3NuL4CZbQKuBFID/UrgM/7tB4A7zczc8E7dbLj7bvjiF7P+ZSWPHTpUfK1zOBHo118PNeMM7ZRgufVWeH/227zpBPoc4EDK/SbgzaMd45wbMLN2oA44aeUgM7sOuA5g/vz5mVVcVwcrV2b2WgmmlSvhggtyXcXUW77cC/OjWq+l4Ew/PV1eU3pR1Dl3F3AXQGNjY2at9yuv9D5ECl0kAv/yL7muQgIknU7JZmBeyv25/mMjHmNmJUAt3sVRERGZIukE+lZgmZktMrNSYAOwedgxm4Fr/dvvA35xWvrPRURkVON2ufh94jcDj+INW7zbObfTzG4DtjnnNgPfBO4xsz1AG17oi4jIFEqrD9059xDw0LDHbk253Qv8SXZLExGRiSiygb0iIoVLgS4iUiAU6CIiBUKBLiJSICxXowvNrAV4NcOX1zNsFmoeUW2ZUW2ZUW2ZCXJtC5xzDSM9kbNAnwwz2+aca8x1HSNRbZlRbZlRbZkp1NrU5SIiUiAU6CIiBSKogX5XrgsYg2rLjGrLjGrLTEHWFsg+dBEROVVQW+giIjKMAl1EpEAEKtDN7Atm9oKZPW9mPzKzaSnP/Y2Z7TGzF83s8hzU9idmttPMBs2sMeXxhWbWY2bP+h9TvmPBaLX5z+X0vA2r5TNm1pxyrt6Vy3r8mq7wz80eM7sl1/WkMrP9ZrbdP1fbclzL3WZ2xMx2pDw2w8z+08xe9j/nZGfqUWrLi/eamc0zs8fNbJf/f/Rj/uOZnTvnXGA+gHcAJf7tzwOf92+vBJ4DosAi4BUgPMW1rQDOBp4AGlMeXwjsyPF5G622nJ+3YXV+BvjrXL/PUuoJ++dkMVDqn6uVua4rpb79QH2u6/BreStwXup7HfhH4Bb/9i3J/695UltevNeA2cB5/u1q4CX//2VG5y5QLXTn3M+dcwP+3d/h7Z4E3ibVm5xzcefcPmAP3ubWU1nbbufci1P5PdM1Rm05P295bmiDdOdcH5DcIF2Gcc79Cm8vhFRXAt/2b38beM+UFuUbpba84Jw75Jx72r/dCezG26M5o3MXqEAf5sPAw/7tkTaynjPlFY1ukZk9Y2a/NLM/yHUxKfLxvN3sd6ndnas/0VPk4/lJ5YCfm9lT/gbs+eYM59wh//Zh4IxcFjOCfHqvYWYLgXOB35PhuZvSTaLTYWb/Bcwa4alPO+ce9I/5NDAA3JtvtY3gEDDfOXfUzM4Hfmxmq5xzHXlQ25Qbq07gn4G/wwuqvwO+iPeLW0Z2sXOu2cxmAv9pZi/4rdG845xzZpZPY6Tz6r1mZlXAD4H/7ZzrMLOh5yZy7vIu0J1zbx/reTP7M+CPgEud38FEehtZn/baRnlNHIj7t58ys1eAs4CsXsTKpDam6LylSrdOM/s34Kens5Y0TPn5mQjnXLP/+YiZ/QiviyifAv11M5vtnDtkZrOBI7kuKMk593rydq7fa2YWwQvze51z/+E/nNG5C1SXi5ldAXwKWOeci6U8tRnYYGZRM1sELAO25KLG4cyswczC/u3FeLXtzW1VQ/LqvPlv3KT1wI7Rjp0i6WyQnhNmVmlm1cnbeAMGcn2+hkvdPP5aIJ/+UsyL95p5TfFvArudc19KeSqzc5frq7wTvCK8B69P81n/419Snvs03oiEF4F35qC29Xh9rHHgdeBR//E/Bnb69T4NvDtfasuH8zasznuA7cDz/ht6dh68596FN/LgFbzuq5zWk1LXYrxRN8/576+c1gbch9e92O+/1z4C1AGPAS8D/wXMyKPa8uK9BlyM1+3zfEquvSvTc6ep/yIiBSJQXS4iIjI6BbqISIFQoIuIFAgFuohIgVCgi4gUCAW6iEiBUKCLiBQIBbqIz8zW+Is1lfkzMXea2epc1yWSLk0sEklhZrcDZUA50OScuyPHJYmkTYEuksJfs2Ur0Atc6JxL5LgkkbSpy0XkZHVAFd7uMWU5rkVkQtRCF0lhZpvxdiZahLdg0805LkkkbXm3HrpIrpjZNUC/c+57/pLHvzGz/+Gc+0WuaxNJh1roIiIFQn3oIiIFQoEuIlIgFOgiIgVCgS4iUiAU6CIiBUKBLiJSIBToIiIF4v8DaVE5INuh+nwAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/loss.py:529: UserWarning: Using a target size (torch.Size([])) that is different to the input size (torch.Size([1])) is deprecated. Please ensure they have the same size.\n",
            "  return F.binary_cross_entropy(input, target, weight=self.weight, reduction=self.reduction)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/loss.py:529: UserWarning: Using a target size (torch.Size([])) that is different to the input size (torch.Size([1])) is deprecated. Please ensure they have the same size.\n",
            "  return F.binary_cross_entropy(input, target, weight=self.weight, reduction=self.reduction)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/loss.py:529: UserWarning: Using a target size (torch.Size([])) that is different to the input size (torch.Size([1])) is deprecated. Please ensure they have the same size.\n",
            "  return F.binary_cross_entropy(input, target, weight=self.weight, reduction=self.reduction)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/loss.py:529: UserWarning: Using a target size (torch.Size([])) that is different to the input size (torch.Size([1])) is deprecated. Please ensure they have the same size.\n",
            "  return F.binary_cross_entropy(input, target, weight=self.weight, reduction=self.reduction)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/loss.py:529: UserWarning: Using a target size (torch.Size([])) that is different to the input size (torch.Size([1])) is deprecated. Please ensure they have the same size.\n",
            "  return F.binary_cross_entropy(input, target, weight=self.weight, reduction=self.reduction)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/loss.py:529: UserWarning: Using a target size (torch.Size([])) that is different to the input size (torch.Size([1])) is deprecated. Please ensure they have the same size.\n",
            "  return F.binary_cross_entropy(input, target, weight=self.weight, reduction=self.reduction)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/loss.py:529: UserWarning: Using a target size (torch.Size([])) that is different to the input size (torch.Size([1])) is deprecated. Please ensure they have the same size.\n",
            "  return F.binary_cross_entropy(input, target, weight=self.weight, reduction=self.reduction)\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FDGWMddsA-DG",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "57eea8a4-1b62-4e74-b7d7-1d48a5370fe3"
      },
      "source": [
        "x=torch.tensor([0.0])\n",
        "yhat=model(x)\n",
        "yhat"
      ],
      "execution_count": 68,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "tensor([0.9861], grad_fn=<SigmoidBackward>)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 68
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_VVJCN8LB9Xj",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 69
        },
        "outputId": "ae566b5e-b7fb-4d8b-92f7-99d295e33c5f"
      },
      "source": [
        "X_=torch.tensor([[0.0],[2.0],[-30.0]])\n",
        "Yhat=model(X_)\n",
        "Yhat"
      ],
      "execution_count": 69,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "tensor([[0.9861],\n",
              "        [0.9788],\n",
              "        [0.0039]], grad_fn=<SigmoidBackward>)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 69
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "g6IYr4dnCBWN",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 313
        },
        "outputId": "229ecdeb-f183-4100-c0b7-fdac4e92288a"
      },
      "source": [
        "plt.plot(LOSS)\n",
        "plt.xlabel('Epochs')\n",
        "plt.title('Cross Entropy loss')"
      ],
      "execution_count": 70,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Text(0.5, 1.0, 'Cross Entropy loss')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 70
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NTguX8V9CU0t",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}