TP_pytorch.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "TP_pytorch.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": [
        "YAz-fhRRdFaR"
      ],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/LMaxence/Cifar10_Classification/blob/master/TP_pytorch.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "4zsDWUk5cavn",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "# Deep Learning for Image Classification\n",
        "\n",
        "Welcome to deep learning for image classification tutorial!\n",
        "**In this notebook, you will**:\n",
        "- Learn the basics of PyTorch, a powerful but easy to use package for scientific computing (and deep learning)\n",
        "- Learn how to build and train a convolutional neural network for image classification.\n",
        "\n",
        "If you have never used jupyter notebooks, nor Colab notebooks, [here](https://colab.research.google.com/notebooks/welcome.ipynb) is a short intro.\n",
        "\n",
        "\n",
        "## I. PyTorch Tutorial\n",
        "\n",
        "We will briefly go through the basics of the PyTorch package, playing with toy examples.\n",
        "\n",
        "If you know already how to use PyTorch, then you can directly go to the second part of this tutorial\n",
        "\n",
        "## II. Training a classifier\n",
        "\n",
        "In this part, we will train a Convolutional Neural Network to classify images of 10 different classes (dogs, cats, car, ...) and see how our model performs on the test set.  \n",
        "\n",
        "\n",
        "## III. Exploring CNN Architectures\n",
        "\n",
        "This is the part where you get your hands dirty ;). Your mission is to experiment different CNN architectures and set hyperparameters in order to obtain the best accuracy on the test set!\n"
      ]
    },
    {
      "metadata": {
        "id": "JgRltjas9PpN",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "The following command sets the backend of matplotlib to the 'inline' backend so that the output of plotting commands is displayed inline within frontends like the Jupyter notebook, directly below the code cell that produced it:"
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "deletable": true,
        "editable": true,
        "id": "GkjN23FKt2D-",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "%matplotlib inline"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "YAz-fhRRdFaR",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "Plotting functions and useful imports\n",
        "\n",
        "You can skip this part"
      ]
    },
    {
      "metadata": {
        "id": "nnee2WPudA9K",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# Python 2/3 compatibility\n",
        "from __future__ import print_function, division\n",
        "\n",
        "import itertools\n",
        "import time\n",
        "\n",
        "\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "# Colors from Colorbrewer Paired_12\n",
        "colors = [[31, 120, 180], [51, 160, 44]]\n",
        "colors = [(r / 255, g / 255, b / 255) for (r, g, b) in colors]\n",
        "\n",
        "# functions to show an image\n",
        "def imshow(img):\n",
        "    \"\"\"\n",
        "    :param img: (PyTorch Tensor)\n",
        "    \"\"\"\n",
        "    # unnormalize\n",
        "    img = img / 2 + 0.5     \n",
        "    # Convert tensor to numpy array\n",
        "    npimg = img.numpy()\n",
        "    # Color channel first -> color channel last\n",
        "    plt.imshow(np.transpose(npimg, (1, 2, 0)))\n",
        "\n",
        "\n",
        "\n",
        "def plot_losses(train_history, val_history):\n",
        "    x = np.arange(1, len(train_history) + 1)\n",
        "\n",
        "    plt.figure(figsize=(8, 6))\n",
        "    plt.plot(x, train_history, color=colors[0], label=\"Training loss\", linewidth=2)\n",
        "    plt.plot(x, val_history, color=colors[1], label=\"Validation loss\", linewidth=2)\n",
        "    plt.xlabel('Epoch')\n",
        "    plt.ylabel('Loss')\n",
        "    plt.legend(loc='upper right')\n",
        "    plt.title(\"Evolution of the training and validation loss\")\n",
        "    plt.show()\n",
        "\n",
        "def plot_confusion_matrix(cm, classes,\n",
        "                          normalize=False,\n",
        "                          title='Confusion matrix',\n",
        "                          cmap=plt.cm.Blues):\n",
        "    \"\"\"\n",
        "    This function prints and plots the confusion matrix.\n",
        "    Normalization can be applied by setting `normalize=True`.\n",
        "    from http://scikit-learn.org/stable/auto_examples/model_selection/plot_confusion_matrix.html\n",
        "    :param cm: (numpy matrix) confusion matrix\n",
        "    :param classes: [str]\n",
        "    :param normalize: (bool)\n",
        "    :param title: (str)\n",
        "    :param cmap: (matplotlib color map)\n",
        "    \"\"\"\n",
        "    if normalize:\n",
        "        cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
        "        \n",
        "    plt.figure(figsize=(8, 8))   \n",
        "    plt.imshow(cm, interpolation='nearest', cmap=cmap)\n",
        "    plt.title(title)\n",
        "    plt.colorbar()\n",
        "    tick_marks = np.arange(len(classes))\n",
        "    plt.xticks(tick_marks, classes, rotation=45)\n",
        "    plt.yticks(tick_marks, classes)\n",
        "\n",
        "    fmt = '.2f' if normalize else 'd'\n",
        "    thresh = cm.max() / 2.\n",
        "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
        "        plt.text(j, i, format(cm[i, j], fmt),\n",
        "                 horizontalalignment=\"center\",\n",
        "                 color=\"white\" if cm[i, j] > thresh else \"black\")\n",
        "\n",
        "    plt.tight_layout()\n",
        "    plt.ylabel('True label')\n",
        "    plt.xlabel('Predicted label')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "aH_K9V7icav6",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "# I. What is PyTorch ?\n",
        "\n",
        "It’s a Python based scientific computing package targeted at two sets of audiences:\n",
        "\n",
        "- A replacement for numpy to use the power of GPUs\n",
        "- A deep learning research platform that provides maximum flexibility and speed\n",
        "\n",
        "\n",
        "## PyTorch Basics\n",
        "\n",
        "In the next steps, we will briefly see how to use PyTorch and exploit its power:\n",
        "\n",
        "1. PyTorch Installation\n",
        "2. PyTorch Tensors\n",
        "3. Numpy Bridge\n",
        "4. Automatic differentiation\n",
        "5. PyTorch and GPU (CUDA support)\n",
        "\n",
        "\n",
        "### 1. Install PyTorch and Torchvision\n",
        "\n"
      ]
    },
    {
      "metadata": {
        "id": "e0y5PLM6ciB5",
        "colab_type": "code",
        "outputId": "f5ba5e5d-2604-4787-e97f-d102c2bc9c50",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 462
        }
      },
      "cell_type": "code",
      "source": [
        "!pip install http://download.pytorch.org/whl/cu92/torch-0.4.1-cp36-cp36m-linux_x86_64.whl\n",
        "!pip install torchvision\n",
        "!pip install pytorch-ignite"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Collecting torch==0.4.1 from http://download.pytorch.org/whl/cu92/torch-0.4.1-cp36-cp36m-linux_x86_64.whl\n",
            "\u001b[?25l  Downloading http://download.pytorch.org/whl/cu92/torch-0.4.1-cp36-cp36m-linux_x86_64.whl (512.6MB)\n",
            "\u001b[K    100% |████████████████████████████████| 512.6MB 48.1MB/s \n",
            "tcmalloc: large alloc 1073750016 bytes == 0x58734000 @  0x7f7bf8d602a4 0x591a07 0x5b5d56 0x502e9a 0x506859 0x502209 0x502f3d 0x506859 0x504c28 0x502540 0x502f3d 0x506859 0x504c28 0x502540 0x502f3d 0x506859 0x504c28 0x502540 0x502f3d 0x507641 0x502209 0x502f3d 0x506859 0x504c28 0x502540 0x502f3d 0x507641 0x504c28 0x502540 0x502f3d 0x507641\n",
            "\u001b[?25hInstalling collected packages: torch\n",
            "Successfully installed torch-0.4.1\n",
            "Collecting torchvision\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/ca/0d/f00b2885711e08bd71242ebe7b96561e6f6d01fdb4b9dcf4d37e2e13c5e1/torchvision-0.2.1-py2.py3-none-any.whl (54kB)\n",
            "\u001b[K    100% |████████████████████████████████| 61kB 4.0MB/s \n",
            "\u001b[?25hCollecting pillow>=4.1.1 (from torchvision)\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/62/94/5430ebaa83f91cc7a9f687ff5238e26164a779cca2ef9903232268b0a318/Pillow-5.3.0-cp36-cp36m-manylinux1_x86_64.whl (2.0MB)\n",
            "\u001b[K    100% |████████████████████████████████| 2.0MB 9.7MB/s \n",
            "\u001b[?25hRequirement already satisfied: numpy in /usr/local/lib/python3.6/dist-packages (from torchvision) (1.14.6)\n",
            "Requirement already satisfied: torch in /usr/local/lib/python3.6/dist-packages (from torchvision) (0.4.1)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.6/dist-packages (from torchvision) (1.11.0)\n",
            "Installing collected packages: pillow, torchvision\n",
            "  Found existing installation: Pillow 4.0.0\n",
            "    Uninstalling Pillow-4.0.0:\n",
            "      Successfully uninstalled Pillow-4.0.0\n",
            "Successfully installed pillow-5.3.0 torchvision-0.2.1\n",
            "Collecting pytorch-ignite\n",
            "  Downloading https://files.pythonhosted.org/packages/f1/94/ffc413d8fdbdf94b612ee7873dab5376e6c147c1da3783bb31231d071eef/pytorch_ignite-0.1.1-py2.py3-none-any.whl\n",
            "Requirement already satisfied: torch in /usr/local/lib/python3.6/dist-packages (from pytorch-ignite) (0.4.1)\n",
            "Installing collected packages: pytorch-ignite\n",
            "Successfully installed pytorch-ignite-0.1.1\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "LcGVaagRcav8",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "import torch as th"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "g2-brDDHcawE",
        "colab_type": "code",
        "outputId": "cee6d48f-3318-43b5-ba61-560806b10589",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        }
      },
      "cell_type": "code",
      "source": [
        "# Create tensor of ones (FloatTensor by default)\n",
        "ones = th.ones(3, 2)\n",
        "print(ones)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "tensor([[1., 1.],\n",
            "        [1., 1.],\n",
            "        [1., 1.]])\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "6RvPibnScawC",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "### 2. PyTorch Tensors\n",
        "\n",
        "A `torch.Tensor` is a multi-dimensional matrix containing elements of a single data type.\n",
        "\n",
        "Tensors are similar to numpy’s ndarrays, but they have a super-power: Tensors can also be used on a GPU to accelerate computing."
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "QcJgJQERcawQ",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "#### Tensor Shape\n",
        "To know the shape of a given tensor, you can use the `.size()` method (the numpy equivalent is `.shape`)"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "SI96-W9acawS",
        "colab_type": "code",
        "outputId": "30e16d76-eaf7-4c01-a9bf-508e72328acd",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "cell_type": "code",
      "source": [
        "# Display the shape of a tensor\n",
        "# it can be used as a tuple\n",
        "print(\"Tensor Shape: {}\".format(ones.size()))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Tensor Shape: torch.Size([3, 2])\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "pUPWrNarcawZ",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "#### Reshape tensors\n",
        "\n",
        "To reshape tensors (e.g. flatten a 3D tensor to a 1D array), you can use the `.view()` method:\n",
        "\n",
        "- **x.view(new_shape)**: Returns a new tensor with the same data but different size. It is the equivalent of numpy function *reshape* (Gives a new shape to an array without changing its data.). You can read the full documentation [here.](http://pytorch.org/docs/master/tensors.html#torch.Tensor.view)\n",
        "\n",
        "[WARNING] when precising a new shape, you have to make sure that the number of elements is constant.\n",
        "For example, a 2D matrix of size 3x3 can only be viewed as a 1D array of size $3 \\cdot 3 = 9$"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "vX-oxI6Vcawb",
        "colab_type": "code",
        "outputId": "a8268770-b407-48f5-da81-c230be82e9a9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        }
      },
      "cell_type": "code",
      "source": [
        "# Create a 3D tensor of size 3x2x2\n",
        "zeros_3d_tensor = th.zeros(3, 2, 2)\n",
        "print(\"Original size:\", zeros_3d_tensor.size())\n",
        "\n",
        "# Reshape it to a 1D array of size 3*2*2 = 12\n",
        "zeros_1d_array = zeros_3d_tensor.view(3 * 2 * 2)\n",
        "print(\"Reshaped tensor:\", zeros_1d_array.size())\n",
        "\n",
        "\n",
        "# Let's view our original tensor as a 2D matrix\n",
        "# If you want PyTorch to guess one remaining dimension,\n",
        "# you specify '-1' instead of the actual size\n",
        "zeros_2d_matrix = zeros_3d_tensor.view(-1, 2 * 2)\n",
        "\n",
        "print(\"Matrix shape:\", zeros_2d_matrix.size())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Original size: torch.Size([3, 2, 2])\n",
            "Reshaped tensor: torch.Size([12])\n",
            "Matrix shape: torch.Size([3, 4])\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "kTO_FFswcawj",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "#### Basic Operations on tensors\n",
        "\n",
        "Tensor support all basic linear algebra operations. You can read the full documentation [here](http://pytorch.org/docs/master/tensors.html)"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "Ay7LvYeVcawl",
        "colab_type": "code",
        "outputId": "dde5fba6-b5fe-4274-dbff-65d9f2886f90",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        }
      },
      "cell_type": "code",
      "source": [
        "2 * ones + 1"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "tensor([[3., 3.],\n",
              "        [3., 3.],\n",
              "        [3., 3.]])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 10
        }
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "OD7ZOT4jcaws",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "PyTorch tensors also supports numpy indexing:"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "srzDzj_ocawu",
        "colab_type": "code",
        "outputId": "6aaa96cd-e6ec-4ca9-da0e-66ca72ecc4a9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        }
      },
      "cell_type": "code",
      "source": [
        "print(\"\\n Indexing Demo:\")\n",
        "print(ones[:, 1])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n",
            " Indexing Demo:\n",
            "tensor([1., 1., 1.])\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "xrjqKguqcaw0",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "### 3. Numpy Bridge\n",
        "WARNING: PyTorch Tensors are different from numpy arrays\n",
        "even if they have a lot in common\n",
        "\n",
        "Though, it is **easy with PyTorch to tranform Tensors to Numpy arrays and vice versa**"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "gVAntrTVcaw3",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "#### Numpy <-> PyTorch\n",
        "\n",
        "Creating PyTorch tensors from numpy array is done via the `torch.from_numpy()` function  \n",
        "(here `th.from_numpy()` because we renamed *torch* as *th*)\n",
        "\n",
        "To transform a PyTorch tensor to a numpy array, you can simply call `.numpy()` method."
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "t2ENcAKOcaw5",
        "colab_type": "code",
        "outputId": "31784d4f-b231-4160-ab05-35bc47090142",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        }
      },
      "cell_type": "code",
      "source": [
        "# np.float32 -> th.FloatTensor\n",
        "ones_matrix = np.ones((2, 2), dtype=np.float32)\n",
        "\n",
        "# the matrix is passed by reference:\n",
        "# if we modify the original numpy array, the tensor is also edited\n",
        "ones_tensor = th.from_numpy(ones_matrix)\n",
        "# Convert back to a numpy matrix\n",
        "numpy_matrix = ones_tensor.numpy()\n",
        "\n",
        "print(\"PyTorch Tensor:\")\n",
        "print(ones_tensor)\n",
        "\n",
        "print(\"Numpy Matrix:\")\n",
        "print(numpy_matrix)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "PyTorch Tensor:\n",
            "tensor([[1., 1.],\n",
            "        [1., 1.]])\n",
            "Numpy Matrix:\n",
            "[[1. 1.]\n",
            " [1. 1.]]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "Y0Itjyg-caxD",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "### 4. Automatic Differentiation\n",
        "\n",
        "Pytorch tensors allow to **automatically compute gradients**. That is particulary useful for backpropagation.\n",
        "\n",
        "Once you finish your computation you can call `.backward()` and have all the gradients computed automatically.\n",
        "\n",
        "You can access the gradient w.r.t. this variable using `.grad`.\n"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "WrPNcIpYcaxK",
        "colab_type": "code",
        "outputId": "14685ca4-70e7-4ee2-dad4-96259e434fb4",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "cell_type": "code",
      "source": [
        "# We need to specify that we want to compute the gradient\n",
        "# as it requires extra memory and computation\n",
        "ones_tensor = th.ones(2,2, requires_grad=True)\n",
        "\n",
        "print(ones_tensor)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "tensor([[1., 1.],\n",
            "        [1., 1.]], requires_grad=True)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "IEZDUibxcaxj",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "To demonstrate the use of PyTorch Variable,\n",
        "let's define a simple linear transformation of a variable $x$ :\n",
        "\n",
        "$$y = a \\cdot x + b$$\n",
        "\n",
        "PyTorch will allows us to automatically compute $$\\frac{dy}{dx} $$"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "A4j85JjZcaxl",
        "colab_type": "code",
        "outputId": "cd2dc9a9-0467-403b-9a8b-ab3278acb656",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "cell_type": "code",
      "source": [
        "# Create a tensor and tell PyTorch\n",
        "# that we want to compute the gradient\n",
        "x = th.ones(1, requires_grad=True)\n",
        "\n",
        "# Transformation constants\n",
        "a = 2\n",
        "b = 5\n",
        "\n",
        "# Define the tranformation and store the result\n",
        "# in a new variable\n",
        "y = a * x + b\n",
        "\n",
        "print(y)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "tensor([7.], grad_fn=<AddBackward>)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "_mxnlvwxcaxq",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "Let's backprop!"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "X1i-pN-Fcaxs",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "y.backward()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "skgIGZdmcaxw",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "`x.grad` prints the gradient:\n",
        "\n",
        "$$\\frac{dy}{dx} = a$$\n",
        "\n",
        "because:\n",
        "\n",
        "$$y = a \\cdot x + b$$"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "_TYbuwsXcaxx",
        "colab_type": "code",
        "outputId": "9e0b9365-0b25-4ae0-83b8-d5d42b61214f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "cell_type": "code",
      "source": [
        "x.grad"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "tensor([2.])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 16
        }
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "ggu-PBGvcax3",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "You can now change the values of $a$ and $b$ see their effects on the gradient\n",
        "(HINT: `x.grad` only depends on the value of `a`)"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "8iPn0C59cax5",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "### 5. PyTorch and GPU (CUDA support)\n",
        "\n",
        "Google colab provides a CUDA enabled GPU, so we are going to use its power. \n",
        "You can move tensor to the GPU by simply using the `to()` method.\n",
        "Otherwise, PyTorch will use the CPU.\n",
        "\n",
        "Here, we will demonstrate the usefulness of the GPU on a simple matrix multiplication:"
      ]
    },
    {
      "metadata": {
        "id": "EwF6ePTpeefQ",
        "colab_type": "code",
        "outputId": "9faa6cc8-1546-4867-bec9-099e4c47dfce",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "cell_type": "code",
      "source": [
        "if th.cuda.is_available():\n",
        "  # Create tensors\n",
        "  x = th.ones(1000, 1000)\n",
        "  y = 2 * x + 3\n",
        "  # Do the calculation on cpu (default)\n",
        "  start_time = time.time()\n",
        "  # Matrix multiplication (for benchmark purpose)\n",
        "  results = th.mm(x, y)\n",
        "  time_cpu = time.time() - start_time\n",
        "  \n",
        "  # Do the same calculation but on the gpu\n",
        "  # First move tensors to gpu\n",
        "  x = x.to(\"cuda\")\n",
        "  y = y.to(\"cuda\")\n",
        "  start_time = time.time()\n",
        "  # Matrix multiplication (for benchmark purpose)\n",
        "  results = th.mm(x, y)\n",
        "  time_gpu = time.time() - start_time\n",
        "  \n",
        "  print(\"Time on CPU: {:.5f}s \\t Time on GPU: {:.5f}s\".format(time_cpu, time_gpu))\n",
        "  print(\"Speed up: Computation was {:.0f}X faster on GPU!\".format(time_cpu / time_gpu))\n",
        "  \n",
        "else:\n",
        "  print(\"You need to enable GPU accelaration in colab (runtime->change runtime type)\")"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Time on CPU: 0.03031s \t Time on GPU: 0.00124s\n",
            "Speed up: Computation was 24X faster on GPU!\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "E-AOzDy9lFwi",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "As expected, matrix multiplication is way faster on a GPU, so we'd better use it."
      ]
    },
    {
      "metadata": {
        "colab_type": "text",
        "deletable": true,
        "editable": true,
        "id": "0kqEBjG6t2Eh"
      },
      "cell_type": "markdown",
      "source": [
        "\n",
        "# II. Training a classifier\n",
        "\n",
        "\n",
        "For this tutorial, we will use the CIFAR10 dataset.\n",
        "There are 10 classes: ‘airplane’, ‘automobile’, ‘bird’, ‘cat’, ‘deer’,\n",
        "‘dog’, ‘frog’, ‘horse’, ‘ship’, ‘truck’. The images in CIFAR-10 are of\n",
        "size 3x32x32, i.e. 3-channel color images of 32x32 pixels in size.\n",
        "\n",
        "\n",
        "![CIFAR10](http://pytorch.org/tutorials/_images/cifar10.png)\n",
        "\n",
        "\n",
        "Training an image classifier\n",
        "----------------------------\n",
        "\n",
        "We will do the following steps in order:\n",
        "\n",
        "1. Load and normalize the CIFAR10 training and test datasets using\n",
        "   ``torchvision``\n",
        "2. Define a Convolution Neural Network\n",
        "3. Define a loss function\n",
        "4. Train the network on the training data\n",
        "5. Test the network on the test data"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "UWTdj2uYcax7",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "### 1. Loading and normalizing CIFAR10 Dataset\n",
        "\n",
        "Using ``torchvision``, it’s extremely easy to load CIFAR10."
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "deletable": true,
        "editable": true,
        "id": "KRrvrIi0t2Em",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "import torch\n",
        "import torchvision\n",
        "import torchvision.transforms as transforms"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "iX2ltR_zcayA",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "Seed the random generator to have reproducible results:"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "335xvR6acayB",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "seed = 42\n",
        "np.random.seed(seed)\n",
        "torch.manual_seed(seed)\n",
        "\n",
        "if th.cuda.is_available():\n",
        "  # Make CuDNN Determinist\n",
        "  th.backends.cudnn.deterministic = True\n",
        "  th.cuda.manual_seed(seed)\n",
        "\n",
        "# Define default device, we should use the GPU (cuda) if available\n",
        "device = th.device(\"cuda\" if th.cuda.is_available() else \"cpu\")\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "7EzIeyD4cayG",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "### Define subset of the dataset (so it is faster to train)"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "Nwu-wWh3cayI",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "from torch.utils.data.sampler import SubsetRandomSampler\n",
        "\n",
        "n_training_samples = 49999 # Max: 50 000 - n_val_samples\n",
        "n_val_samples = 1\n",
        "n_test_samples = 10000\n",
        "\n",
        "train_sampler = SubsetRandomSampler(np.arange(n_training_samples, dtype=np.int64))\n",
        "val_sampler = SubsetRandomSampler(np.arange(n_training_samples, n_training_samples + n_val_samples, dtype=np.int64))\n",
        "test_sampler = SubsetRandomSampler(np.arange(n_test_samples, dtype=np.int64))\n",
        "# (In the last case, indexes do not need to account for training ones because the train=False parameter in datasets.CIFAR will select from the test set)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "colab_type": "text",
        "deletable": true,
        "editable": true,
        "id": "evFXNmbst2Ez"
      },
      "cell_type": "markdown",
      "source": [
        "The output of torchvision datasets are PILImage images of range [0, 1].\n",
        "We transform them to Tensors of normalized range [-1, 1]\n",
        "\n"
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "deletable": true,
        "editable": true,
        "id": "ZJ-hYN00t2E2",
        "outputId": "7a3e1a41-0db0-4b5d-b062-37239814a55f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "cell_type": "code",
      "source": [
        "num_workers = 2\n",
        "test_batch_size = 4\n",
        "\n",
        "\n",
        "transform_augmentation = transforms.Compose(\n",
        "    [transforms.RandomCrop(32, padding=4),\n",
        "     transforms.RandomHorizontalFlip(),\n",
        "     transforms.RandomRotation(15),\n",
        "     transforms.ToTensor(),\n",
        "     transforms.Normalize(mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5))\n",
        "    ])\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "transform = transforms.Compose(\n",
        "    [transforms.ToTensor(),\n",
        "     transforms.Normalize(mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5))\n",
        "    ])\n",
        "\n",
        "\n",
        "#train_set = torchvision.datasets.CIFAR10(root='./data', train=True,\n",
        "#                                        download=True, transform=transform)\n",
        "#train_loader = torch.utils.data.DataLoader(train_set, batch_size=test_batch_size, sampler=train_sampler,\n",
        "#                                          num_workers=num_workers)\n",
        "\n",
        "#test_set = torchvision.datasets.CIFAR10(root='./data', train=False,\n",
        "#                                       download=True, transform=transform)\n",
        "#test_loader = torch.utils.data.DataLoader(test_set, batch_size=test_batch_size, sampler=test_sampler,\n",
        "#                                         num_workers=num_workers)\n",
        "\n",
        "train_set_augmented = torchvision.datasets.CIFAR10(root='./data', train=True,\n",
        "                                        download=True, transform=transform_augmentation)\n",
        "train_loader_augmented = torch.utils.data.DataLoader(train_set_augmented, batch_size=test_batch_size, sampler=train_sampler,\n",
        "                                          num_workers=num_workers)\n",
        "\n",
        "classes = ('plane', 'car', 'bird', 'cat',\n",
        "           'deer', 'dog', 'frog', 'horse', 'ship', 'truck')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Files already downloaded and verified\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "text",
        "deletable": true,
        "editable": true,
        "id": "cGWVnBOft2FI"
      },
      "cell_type": "markdown",
      "source": [
        "Let us show some of the training images, for fun.\n",
        "\n"
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "deletable": true,
        "editable": true,
        "id": "68OfC35ut2FM",
        "outputId": "9624cc2c-ada3-4c79-8ecf-1e4ff402fceb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 184
        }
      },
      "cell_type": "code",
      "source": [
        "# get some random training images\n",
        "dataiter = iter(train_loader)\n",
        "images, labels = dataiter.next()\n",
        "\n",
        "# show images\n",
        "imshow(torchvision.utils.make_grid(images))\n",
        "# print labels\n",
        "print(' '.join('{:>10}'.format(classes[labels[j]]) for j in range(test_batch_size)))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "     horse        cat        cat      horse\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAd8AAACWCAYAAACfIIJIAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nvpZP+gAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7f4c7342df28>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "text",
        "deletable": true,
        "editable": true,
        "id": "8ULHEu5Zt2Fa"
      },
      "cell_type": "markdown",
      "source": [
        "### 2. Define a Convolution Neural Network\n"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "6k6rJyTTcayi",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# Useful imports\n",
        "import torch.nn as nn\n",
        "import torch.nn.functional as F"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "colab_type": "text",
        "deletable": true,
        "editable": true,
        "id": "0JcmlEe8t2Fe"
      },
      "cell_type": "markdown",
      "source": [
        "####  Forward propagation\n",
        "\n",
        "In PyTorch, there are built-in functions that carry out the convolution steps for you.\n",
        "\n",
        "- **nn.Conv2d(in_channels, out_channels, kernel_size, stride=1, padding=0):** Convolution layer. You can read the full documentation [here](http://pytorch.org/docs/master/nn.html#conv2d)\n",
        "\n",
        "- **nn.MaxPool2d(kernel_size, stride=None, padding=0):** Max pooling layer. You can read the full documentation [here](http://pytorch.org/docs/master/nn.html#maxpool2d)\n",
        "\n",
        "- **F.relu(Z1):** computes the elementwise ReLU of Z1 (which can be any shape). You can read the full documentation [here.](http://pytorch.org/docs/master/nn.html#torch.nn.ReLU)\n",
        "\n",
        "- **x.view(new_shape)**: Returns a new tensor with the same data but different size. It is the equivalent of numpy function *reshape* (Gives a new shape to an array without changing its data). You can read the full documentation [here.](http://pytorch.org/docs/master/tensors.html#torch.Tensor.view)\n",
        "\n",
        "- **nn.Linear(in_features, out_features):** Applies a linear transformation to the incoming data: $y = Ax + b$, it is also called a fully connected layer. You can read the full documentation [here.](http://pytorch.org/docs/master/nn.html#linear-layers)"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "rbykSRDTcaym",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "#### Simple Convolutional Neural Network\n",
        "\n",
        "ConvNet with one convolution layer followed by a max pooling operation,\n",
        "one fully connected layer and an output layer"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "X4pljAWycayn",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "def conv2d(in_planes, out_planes, kernel_size=3, stride=1):\n",
        "    \"Convolution with padding, default kernel size is 3x3, default stride is 1\"\n",
        "    return nn.Conv2d(in_planes, out_planes, kernel_size=kernel_size, stride=stride,\n",
        "                     padding=1, bias=False)\n",
        "\n",
        "class SimpleConvolutionalNetwork(nn.Module):\n",
        "    def __init__(self):\n",
        "        # Because inheritance does not automatically launch the initialisation of the extended class\n",
        "        super(SimpleConvolutionalNetwork, self).__init__()\n",
        "        \n",
        "        self.conv1 = nn.Conv2d(3, 18, kernel_size=3, stride=1, padding=1)\n",
        "        self.pool = nn.MaxPool2d(kernel_size=2, stride=2, padding=0)\n",
        "        \n",
        "        \n",
        "        \n",
        "        # cf comments in forward() to have step by step comments\n",
        "        # on the shape (how we pass from a 3x32x32 input image to a 18x16x16 volume)\n",
        "        self.fc1 = nn.Linear(18 * 16 * 16, 64) \n",
        "        self.fc2 = nn.Linear(64, 10)\n",
        "\n",
        "    def forward(self, x):\n",
        "        \"\"\"\n",
        "        Forward pass,\n",
        "        x shape is (batch_size, 3, 32, 32)\n",
        "        (color channel first)\n",
        "        in the comments, we omit the batch_size in the shape\n",
        "        \"\"\"\n",
        "        # shape : 3x32x32 -> 18x32x32\n",
        "        x = F.relu(self.conv1(x))\n",
        "        # 18x32x32 -> 18x16x16\n",
        "        x = self.pool(x)\n",
        "        # 18x16x16 -> 4608\n",
        "        x = x.view(-1, 18 * 16 * 16)\n",
        "        # 4608 -> 64\n",
        "        x = F.relu(self.fc1(x))\n",
        "        # 64 -> 10\n",
        "        # The softmax non-linearity is applied later (cf createLossAndOptimizer() fn)\n",
        "        x = self.fc2(x)\n",
        "        return x"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "4m-VHCtRcayr",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "#### Linear Classifier"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "Rj-togN6cays",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "class LinearClassifier(nn.Module):\n",
        "    \"\"\"\n",
        "    Linear Classifier\n",
        "    \"\"\"\n",
        "    def __init__(self):\n",
        "        super(LinearClassifier, self).__init__()\n",
        "        self.linear = nn.Linear(32 * 32 * 3, 10)\n",
        "\n",
        "    def forward(self, x):\n",
        "        # Flatten input 3x32x32 -> 3072\n",
        "        x = x.view(x.size(0), -1)\n",
        "        return self.linear(x)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "colab_type": "text",
        "deletable": true,
        "editable": true,
        "id": "2SQi9Xf-t2Fu"
      },
      "cell_type": "markdown",
      "source": [
        "### 3. Define a loss function and optimizer\n",
        "\n",
        "Let's use a Classification Cross-Entropy loss and ADAM (optionally, SGD with momentum). You can read more about  [optimization methods](https://pytorch.org/docs/stable/optim.html).\n",
        "\n"
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "deletable": true,
        "editable": true,
        "id": "DOUiPtZQt2Fx",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "import torch.optim as optim\n",
        "\n",
        "def createLossAndOptimizer(net, learning_rate=0.001):\n",
        "    # it combines softmax with negative log likelihood loss\n",
        "    criterion = nn.CrossEntropyLoss()  \n",
        "    #optimizer = optim.SGD(net.parameters(), lr=learning_rate, momentum=0.9)\n",
        "    optimizer = optim.Adam(net.parameters(), lr=learning_rate, weight_decay = 1e-4)\n",
        "    return criterion, optimizer"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "colab_type": "text",
        "deletable": true,
        "editable": true,
        "id": "saJW5bKRt2F9"
      },
      "cell_type": "markdown",
      "source": [
        "### 4. Train the network\n",
        "\n",
        "\n",
        "This is when things start to get interesting.\n",
        "We simply have to loop over our data iterator, feed the inputs to the network, and optimize\n",
        "\n"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "mNf1e8QZcay1",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "#### Data loader"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "EqDD8_z8cay2",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "def get_train_loader(batch_size):\n",
        "    return torch.utils.data.DataLoader(train_set, batch_size=batch_size, sampler=train_sampler,\n",
        "                                              num_workers=num_workers)\n",
        "\n",
        "# Use larger batch size for validation to speed up computation\n",
        "val_loader = torch.utils.data.DataLoader(train_set, batch_size=128, sampler=val_sampler,\n",
        "                                          num_workers=num_workers)\n",
        "\n",
        "def get_train_loader_augmented(batch_size):\n",
        "    return torch.utils.data.DataLoader(train_set_augmented, batch_size=batch_size, sampler=train_sampler,\n",
        "                                              num_workers=num_workers)\n",
        "\n",
        "# Use larger batch size for validation to speed up computation\n",
        "val_loader_augmented = torch.utils.data.DataLoader(train_set_augmented, batch_size=128, sampler=val_sampler,\n",
        "                                          num_workers=num_workers)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "yTDHHbLpcay5",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "#### Training loop\n",
        "The training script: it takes ~10s per epoch with batch_size = 32"
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "deletable": true,
        "editable": true,
        "id": "dATbDR5pt2GE",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "def train(net, batch_size, n_epochs, learning_rate, early_stopping = False, patience_early_stopping = 1000):\n",
        "    \"\"\"\n",
        "    Train a neural network and print statistics of the training\n",
        "    \n",
        "    :param net: (PyTorch Neural Network)\n",
        "    :param batch_size: (int)\n",
        "    :param n_epochs: (int)  Number of iterations on the training set\n",
        "    :param learning_rate: (float) learning rate used by the optimizer\n",
        "    \"\"\"\n",
        "    print(\"===== HYPERPARAMETERS =====\")\n",
        "    print(\"batch_size=\", batch_size)\n",
        "    print(\"n_epochs=\", n_epochs)\n",
        "    print(\"learning_rate=\", learning_rate)\n",
        "    print(\"=\" * 30)\n",
        "    \n",
        "    train_loader = get_train_loader_augmented(batch_size)\n",
        "    n_minibatches = len(train_loader)\n",
        "\n",
        "    criterion, optimizer = createLossAndOptimizer(net, learning_rate)\n",
        "    lr_scheduler = torch.optim.lr_scheduler.MultiStepLR(optimizer, milestones=[75,100], gamma=0.5)\n",
        "    # Init variables used for plotting the loss\n",
        "    train_history = []\n",
        "    val_history = []\n",
        "\n",
        "    training_start_time = time.time()\n",
        "    best_error = np.inf\n",
        "    best_model_path = \"best_model.pth\"\n",
        "    \n",
        "    # Move model to gpu if possible\n",
        "    net = net.to(device)\n",
        "\n",
        "    for epoch in range(n_epochs):  # loop over the dataset multiple times\n",
        "\n",
        "        running_loss = 0.0\n",
        "        print_every = n_minibatches // 10\n",
        "        start_time = time.time()\n",
        "        total_train_loss = 0\n",
        "        \n",
        "        lr_scheduler.step()\n",
        "        \n",
        "        for i, (inputs, labels) in enumerate(train_loader):\n",
        "\n",
        "            # Move tensors to correct device\n",
        "            inputs, labels = inputs.to(device), labels.to(device)\n",
        "\n",
        "            # zero the parameter gradients\n",
        "            optimizer.zero_grad()\n",
        "\n",
        "            # forward + backward + optimize\n",
        "            outputs = net(inputs)\n",
        "            loss = criterion(outputs, labels)\n",
        "            loss.backward()\n",
        "            optimizer.step()\n",
        "\n",
        "            # print statistics\n",
        "            running_loss += loss.item()\n",
        "            total_train_loss += loss.item()\n",
        "\n",
        "            # print every 10th of epoch\n",
        "            if (i + 1) % (print_every + 1) == 0:    \n",
        "                print(\"Epoch {}, {:d}% \\t train_loss: {:.2f} took: {:.2f}s\".format(\n",
        "                      epoch + 1, int(100 * (i + 1) / n_minibatches), running_loss / print_every,\n",
        "                      time.time() - start_time))\n",
        "                running_loss = 0.0\n",
        "                start_time = time.time()\n",
        "\n",
        "        train_history.append(total_train_loss / len(train_loader))\n",
        "\n",
        "        total_val_loss = 0\n",
        "        # Do a pass on the validation set\n",
        "        # We don't need to compute gradient,\n",
        "        # we save memory and computation using th.no_grad()\n",
        "        with th.no_grad():\n",
        "          for inputs, labels in val_loader_augmented:\n",
        "              # Move tensors to correct device\n",
        "              inputs, labels = inputs.to(device), labels.to(device)\n",
        "              # Forward pass\n",
        "              predictions = net(inputs)\n",
        "              val_loss = criterion(predictions, labels)\n",
        "              total_val_loss += val_loss.item()\n",
        "            \n",
        "        val_history.append(total_val_loss / len(val_loader_augmented))\n",
        "        # Save model that performs best on validation set\n",
        "        if total_val_loss < best_error:\n",
        "            best_error = total_val_loss\n",
        "            th.save(net.state_dict(), best_model_path)\n",
        "\n",
        "        print(\"Validation loss = {:.2f}\".format(total_val_loss / len(val_loader_augmented)))\n",
        "\n",
        "    print(\"Training Finished, took {:.2f}s\".format(time.time() - training_start_time))\n",
        "    \n",
        "    # Load best model\n",
        "    net.load_state_dict(th.load(best_model_path))\n",
        "    \n",
        "    return train_history, val_history"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "cJX2anB5cay_",
        "colab_type": "code",
        "outputId": "b5aba135-a28e-4a59-f51f-cc1298b43e06",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 198
        }
      },
      "cell_type": "code",
      "source": [
        "net = SimpleConvolutionalNetwork()\n",
        "\n",
        "train_history, val_history = train(net, batch_size=16, n_epochs=10, learning_rate=0.001)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "error",
          "ename": "NameError",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-30-50d75a0389e1>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mnet\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mSimpleConvolutionalNetwork\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      2\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      3\u001b[0m \u001b[0mtrain_history\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mval_history\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtrain\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnet\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mbatch_size\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m16\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mn_epochs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m10\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlearning_rate\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m0.001\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mNameError\u001b[0m: name 'SimpleConvolutionalNetwork' is not defined"
          ]
        }
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "UkVKNPtccazC",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "Now, let's look at the evolution of the losses"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "4CUQt-HJcazF",
        "colab_type": "code",
        "outputId": "be7673e6-674a-4182-e898-720609a9f6d1",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 403
        }
      },
      "cell_type": "code",
      "source": [
        "plot_losses(train_history, val_history)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAe8AAAGCCAYAAADJ40tJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4yLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvNQv5yAAAIABJREFUeJzs3Xd8VFX+//HXlEx6JYVQhCCdCNJF\nEJAaENzFAqiAiiuKsi6631XXhgoq6Oq6YF1EdwEXCwL+3AURECwgSMfQOwklvfdk7u+PkIGQBEIZ\nZpK8n/vw8TC3nPv5zGT95J577jkmwzAMREREpMYwuzoAERERuTgq3iIiIjWMireIiEgNo+ItIiJS\nw6h4i4iI1DAq3iIiIjWMirc4VatWrRg4cCAxMTHl/tmxY8cltff000/z3nvvXfC4L774wvHvMTEx\nJCcnX9L1Lsann35Kz549ef/9988bz9ixY/n6668v61pLly4lOzv7os+799572blz53mPmT9/Pm+/\n/falhuYUzz77LLNmzboibfXr149NmzaxY8cOHnjggcu63tnfa3U+2+qaNWsWzz777BVpS2onq6sD\nkNpv3rx51K9f/6pdLykpiY8++oiRI0cC8O23316V63733XdMnjyZO++8s9z2kpISXn/9dUc8V8LM\nmTPp1KkTfn5+F3Xev//97wseM2bMmEsNq0Zp3749c+bMueTzz/1eq/PZilwpuvMWl7njjjtYvny5\n4+eVK1c6/kO4bNkyhg0bRkxMDOPGjePYsWMVzm/VqhWnTp2q8PPo0aM5ceIEMTExFBYWljtu7ty5\nDB06lJiYGCZOnEhqaipQekc/c+ZM7r//fm6++Wbuv/9+8vLyKlyzoKCAF154gcGDBzNkyBCmT5/u\n+I/4tm3b+Mc//lHhju3+++8nKyuLmJgY4uLiAIiPj2fs2LHcdNNNPPHEE9jtdgA2b97M7bffzsCB\nAxk5cqTj+LP99a9/5fDhw4wdO5ZNmzbx9NNP89prrzF8+HCWLVtGXl4ekydPZvDgwfTr148ZM2Y4\nzi2764yPj6dXr17MnTuX4cOHc9NNN7F06VKg/F3f2LFj+eSTT7jrrrscsZbN67Ro0SJ69uzJrbfe\nyqJFi2jVqlWl3/OqVasYPnw4gwcP5rbbbmP37t0AbNiwgVGjRvHmm28yZMgQ+vXrx6+//gpAWloa\n48ePp1+/fkyYMIGsrKwK7R44cIBu3bpRXFzs2PbII4+wYMGC834GZTZs2MDAgQMveL2tW7dy2223\nERMTw9ChQ1m3bl2l32vZZwtV//7OmjWLl19+mUcffZT+/ftzxx13kJiYWOnnVubEiRM88MADDB48\nmGHDhrFkyRIAiouLefbZZxk8eDADBw5k0qRJZGdnV7ldahlDxIlatmxpnDx5stJ9//znP40nn3zS\n8fOTTz5pfPzxx8bx48eNzp07G0eOHDEMwzDmzJlj3HvvvYZhGMZTTz1lvPvuu5W2Xfbz+vXrjQED\nBlTYvnXrVqN3795GcnKyYRiG8fLLLxvPPPOMo90hQ4YYaWlpRlFRkXHrrbcaX3/9dYWYP/zwQ+PB\nBx80ioqKjLy8POP22283lixZYhiGYYwZM8bx72eLi4sz2rRp4/h5zJgxxrhx44y8vDwjOzvbuPHG\nG42NGzcaWVlZRteuXY2ff/7ZMAzD+Oabb4wRI0Zc8HN96qmnjOHDhxv5+fmOz+sPf/iDYbfbjfT0\ndKNbt27Gxo0bDcMwjJtvvtnYuHGjERcXZ7Rt29aYN2+eYRiGsXTpUmPgwIGGYRjGzJkzHZ/LmDFj\njDFjxhh5eXlGTk6O0aNHD2PTpk1GWlqa0b59e2Pv3r1GSUmJ8fjjjxstW7asEGdRUZHRpUsXY+vW\nrYZhGMasWbMc3+X69euN6OhoY8WKFYZhGMbs2bON++67zzAMw5gxY4bxxBNPOD6/jh07GjNnzqzQ\n/pAhQ4xffvnFMAzDyM3NNTp27GikpKRU6zM4+/fkfNcbNmyY8d///tcwDMNYvHix45xzv9eyds/3\n+ztz5kyjR48eRnx8vGG3240JEyYY7733XoW8zv4Oxo8fb3zwwQeGYRhGfHy80blzZyMuLs5YvXq1\nMW7cOMNutxt2u934+9//bvz4449VbpfaRXfe4nRjx44t97z77rvvBkqfRf/www+UlJRQXFzMmjVr\niImJYe3atXTv3p0mTZoAcOedd7Jhw4Zyd1iXYs2aNQwePJh69eo52l27dq1jf58+fQgKCsJqtdKy\nZUtOnjxZaRsjR47EarXi5eXF8OHDy7VRXYMGDcLLywtfX1+aNGnCqVOn2Lx5MxEREfTs2ROAYcOG\ncezYMU6cOHHB9nr06IGnpycA48eP57333sNkMhEYGEiLFi2Ij4+vcE5xcTG33XYbAO3atavyOjEx\nMXh5eeHj40PTpk05efIk27dvp2nTprRs2RKz2cxdd91V6blWq5V169Zx/fXXA9ClS5dyvQm+vr4M\nGDCgQgybNm1iyJAhADRq1Ihu3bpV2v7gwYP5/vvvAfjpp59o3749ISEh1f4MypzvekuWLHHs69y5\nc6W9IWe70O9vly5daNiwISaTiTZt2lT6e1amqKiIdevWOf4/07BhQ7p378769esJCQnh4MGDrFix\nwtHTcNNNN1W5XWoXPfMWp6vqmXfjxo2JjIxk69atFBUVERUVRWRkJGlpaQQEBDiO8/f3xzAM0tLS\nLiuO1NRUwsPDHT8HBASQkpJS7jplLBYLJSUllbYRGBjo+DkwMLBcG9V19rPqsmtlZmYSFxdHTEyM\nY5/NZiM1NZUGDRqct72zYzpy5AjTp0/n0KFDmM1mTp065SjSZ7NYLPj4+ABgNpsdXffVjfXsa0ZE\nRFQZ27x581i8eDGFhYUUFhZiMpkc+87+zM+OISMjo9y+s38fzjZ48GAmTZrEM888w8qVKxk6dOhF\nfQZlzne9b775hrlz55KTk4Pdbnc8NqjKhX5/q/N7ViY9PR3DMCrElpqaSvv27XnuueeYN28eTz31\nFP369WPKlClVbq/qM5SaSXfe4lKDBw9m1apVrFq1ynF3U69ePdLT0x3HZGRkYDabCQ4OLneu2Wx2\n/IcvIyPjgtcKDQ0t1256ejqhoaEXFe+VaKMq4eHhNGvWjG+//dbxz7p164iOjr6odl5++WVatGjB\nsmXL+Pbbb2nduvUVie9sfn5+5ObmOn6u6rntli1bmD17Nu+//z7Lly9n2rRp1Wo/ICCg3HPnsrEJ\n52rdujUWi4U9e/bw888/O55hX+xnUNX1EhISeO6553jllVdYvnw5s2fPvmDs1f39rY7g4GDMZnO5\n3+/09HRH71FMTAzz5s1j9erV5OXlOQbgVbVdag8Vb3GpwYMH88svv7B69WrHHWfPnj3ZtGmTo3vy\ns88+o2fPnlit5TuKwsLC2LNnDwBfffUVZnPpr7PVaiU3N7dCN3vfvn1ZsWKF4w7os88+o0+fPhcV\nb9++fVm4cCElJSXk5uby9ddfX7ANDw8P7Hb7BQcNdejQgaSkJLZv3w5AXFwcf/nLXyq907NarWRm\nZlbaTkpKCm3atMFisbB27VqOHj1artBeCe3atWPv3r0cPXoUu93OwoULKz0uNTWVevXq0aBBA/Ly\n8li8eDG5ubkXvHu9/vrrWblyJQDHjh1j8+bNVR47ePBgZs2aRZs2bRwF8mI/g6qul5qaio+PD82a\nNaO4uJjPP/8cgJycnCq/1+r+/laH1WqlV69ejuseO3aMTZs2ceONN/LVV1/x7rvvAhAUFESzZs0A\nqtwutYuKtzjduc+8Y2JimD9/PgBRUVHY7XYiIiIcXa/169dn2rRpPPLII8TExLBx40ZefvnlCu0+\n/vjjvPjii/zud7/D29vb0b3bqlUrAgMD6dmzZ7nnuO3bt2fChAncc889xMTEkJWVxeOPP37RudSv\nX59bbrmF22+/nb59+zp6DKoSFhZG586dufnmm9myZUuVx3l5eTFz5kymTp3KkCFDePTRR4mJiSnX\nzVwmJiaG0aNHO0aIn23ixInMmDGDYcOG8euvvzJp0iRmzZp13gJ4scLDw3niiScYN24cd955J507\nd670uJtuuonw8HAGDBjA+PHjuffee/H39+exxx47b/sPPfQQx48fp1+/fkydOpVBgwZVeezgwYNZ\nuXJluccNF/sZVHW91q1b07t3bwYPHsyoUaPo168f119/PWPHjq3ye63u7291vfTSS2zYsIGYmBge\nffRRpk2bRmRkJP3792fnzp0MGjSIIUOGcODAAe6///4qt0vtYjIu9CewiEglDMNw/GGxf/9+7r77\nbjZu3OjiqETqBt15i8hFKy4u5qabbnJ08S9dutQxolxEnM+pd96vv/46mzdvpri4mIceeqhc19f6\n9et56623MJvNREVF8corrzieWYqI+1uxYgVvvvkmhmEQFhbGK6+84ng9SkScy2nFe/369cyZM4fZ\ns2eTlpbGiBEjWLNmjWP/oEGDmDt3LvXr1+exxx7j9ttvv+jBQyIiInWR097z7tq1K+3btwdKX8PI\ny8ujpKQEi8UClE6tWDbAKCQk5LLf4RUREakrnNZPffYEEAsXLqR3796Owg1nJn5ITExk7dq1uusW\nERGpJqfPsLZy5UoWLlzIxx9/XGFfSkoKDz/8MFOmTLngBAbFxSVYrZbzHiMiIlIXOLV4//TTT3zw\nwQd89NFH5ab3A8jOzubBBx9k8uTJ9OrV64JtpaVVb5KJsDB/kpIqrkBUUykf96Z83JvycW/Kp3pt\nVsZpxTsrK4vXX3+df/3rXwQFBVXYP336dO6991569+7trBBERERqJacV76VLl5KWlsbkyZMd27p3\n706rVq3o1asXS5Ys4ejRo45pFYcNG8aoUaOcFY6IiEit4bTiPWrUqPMW49jYWGddWkREpFbTrCgi\nIiI1jIq3iIhIDaPiLSIiUsOoeIuIiNQwTp+kRURE5FyzZv2dw4f3c+pUAvn5+TRo0JCAgEBeffWN\nC567dOk3+Pr60afPzZXu/8c/3uTOO0fToEHDS4pt0qQJPPHEkzRr1vySzr8aVLxFROSq++MfHycs\nzJ9///s/HDp0kEmTJl/4pNOGDh1+3v1/+tOfLzc8t6fiLSIibmPLlk189tl8cnNzmTTpcbZu3cya\nNauw2+306NGT8eMnMGfOhwQFBREVdS2LFn2ByWTm6NHD9O3bn/HjJzjunFevXkVOTjbHjh3l+PF4\nHnvsz/To0ZP58//FypXf0aBBQ4qLixk9+h46depSIZbs7GxeeeVFsrOzKC4uZvLkv9CqVWvefvsN\n9uzZTUlJCSNG3MHQocN5++03OHhwH/n5hY5tzqTiLSJSx01eHMvaw6lXtM2eUSG8PSL6ks49ePAA\nCxYswmazsXXrZt577yPMZjMjR/6OUaPuLnfsrl07+c9/vsJut3PnncMZP35Cuf2JiQn87W8zWb9+\nHV9//RXt2kWzaNGXLFjwFTk5OYwefRujR99TaRxffrmAdu2iGTPmPvbs2cWsWW/x6qtvsG7dz3zx\nxdcUFxezdOk3ZGZmsG7dz6xe/T0nT6axdOk3l5T3xaiTxTs9r4hluxO5pW04AV4erg5HRETO0rx5\nC2w2GwBeXl5MmjQBi8VCeno6mZmZ5Y5t1ao1Xl5eVbbVvv31AISHh5OdnU18fBzNml2Lp6cXnp5e\ntGnTrspz9+zZxbhxDwDQunVb4uPjCAgIpHHjJjz99BPcfPMAYmJuwWaz0bhxEyZOnEjPnn2Jibnl\ncj+CC6qTxfu/OxP4x4+HyC8u4f5u17g6HBERl7rUO2Rn8fAovak6deokn3/+KR9//Ck+Pj6MHTuy\nwrFnLzVdmbP3G4aBYYDZfOZFK5Op6nNNJhOGYTh+ttvtALz55kz27t3DihXf8u23/+Pvf3+XN9+c\nSWLiMb74YpFjmzPVyVfFfGylacen5bk4EhERqUp6ejrBwcH4+Piwd+8eTp06RVFR0WW1GRkZyaFD\nBykuLiYtLY09e3ZXeWzr1m3ZunUTALGxvxEVdS0nT57gyy8/o1Wr1kyaNJmMjAzHtnbt2jm2OVud\nvPMO9/MEIDG70MWRiIhIVVq0aIm3tw8TJ47nuuuu53e/u40335xB+/YdLrnNkJB6DBwYw4MPjqNJ\nkyjatm1X5d37yJF38eqrL/HYYw9jt9t54omnCA0NIzZ2O6tWfYeHhwe33HKrY9vo0aMBM7fccusl\nx1ddJuPsPgE3Vt01Uquznur+pGzunreFqBAfvriv4ghDd6L1bt2b8nFvyse9uSqfpUu/YeDAGCwW\nC+PGjeatt2YRHh5x2e3WivW83VmEf+mdd0JWAYZhYDrfQw8REalVUlJSmDDhXjw8bAwaFHNFCvfV\nVieLt7+nFS+rmdyiEnIKS/DzrJMfg4hInTR27H2MHXufq8O4LHVywJrJZHLcfZ/KKnBxNCIiIhen\nThZvKN91LiIiUpPU2eIdfrp4J6p4i4hIDVNni3eEn+68RUSkZqq7xbus2zxbxVtE5Gp76KH7iY2N\nLbftgw/eYcGC+ZUev2XLJp577kkAnn76iQr7v/rqc+bM+bDK6x04sJ9jx44CMGXKXykoyL/U0Lnj\njuHk5uZe8vlXQp0t3uo2FxFxnYEDB7Ns2bJy29as+Z4BAwZd8Nzp09+66Ov98MP3xMUdA+Cll17D\n07Pq+dBrgjr7jpQGrImIuE7//oOYNOlB7rvvYQD27NlNWFgYYWHhbNy4gY8++gAPDw/8/f15+eXp\n5c695Zb+/O9/q9i06VdmznyTkJB61KsX6lji85VXXiQpKZG8vDzGj59A/fqRfP31In744XuCg4N5\n4YW/Mnfu52RnZ/Haay9TVFSE2Wzm6aefx2Qy8corL9KgQUMOHNhPy5atePrp5yvNITExodz5r78+\nHavVj5dffp6UlGQKCwt54IGH6NKlW4VtN9xw42V9fnW2eNfXRC0iIgC8sPHPbEz65Yq22TWsBy93\nfbPK/cHBITRu3Jhdu2Jp2zaa779fwcCBMQBkZWUxZco0GjRoyNSpL7Bhwy/4+PhUaOPDD9/h+een\n0qJFS/7v/x6jQYOGZGVl0q3bDQwZMozjx+N5/vmn+fjj+XTv3oO+ffvTtu2ZRVg++ugDhg37Hf37\nD2L16pV8/PE/eeCBh9i7dzcvvfQqwcEhjBgxlKysLPz9K850du7577zzDsOH30FGRjrvvjubrKws\nfvllLQcPHqiw7XLV2W5zX5sFHw8L+cV2sgqKXR2OiEidM2zYMFatWgHA2rU/0rdvfwCCgoKYMWMa\nkyZNYOvWzWRmVr7Qx8mTJ2nRoiUA11/fCQB//wB2797JxInjeeWVF6s8F2Dv3t107NgZgE6durB/\n/14AGjZsTL16oZjNZkJDw8jJya7W+bt27aJJk6bk5uYwderzbNmykQEDBlW67XLV2TvvsolaDqfm\nkpBVoHW9RaTOOt8dsjMNHDiQd999j4EDB9O48TUEBAQA8NprU3njjbdp2jSKt96aUeX5Zy/tWbZM\nx4oV35KZmcm7735EZmYmf/jD2PNEcGbJz6KiYkym0vbOXaik6iVAyp9vNpvx8vLiww//xW+/7WDZ\nsm9Yu/YnnnlmSqXbLkedvfMGPfcWEXElPz8/rr22BXPnfuLoMgfIyckmIqI+WVlZbNmyucplQEND\nwzh27AiGYbB162agdBnRyMgGmM1mfvjhe8e5JpOJkpKScue3adOWLVtKl/zctm0zrVu3uaj4zz0/\nOjrasc53hw7X83//91eOHDlc6bbLVWfvvOFM8U7U62IiIi4xcGAM06ZNYcqUqY5tt912JxMnPkDj\nxtdwzz3j+PjjfzJhwiMVzp0w4RGee+4p6tePdCwu0rdvP55++gl27YrllltuJTw8nE8+mU2HDh15\n++03yj07/8MfHua116byzTdLsFo9+Otfn6e4uPqPUc89/29/m0F2djEffvguX3+9CLPZzN13jyUy\nskGFbZerTi4JWuaf644we/0xxndvzMSeUZcTntNoCUD3pnzcm/Jxb8qnem1WRt3mqNtcRERqFhVv\nICGr0MWRiIiIVJ9Tn3m//vrrbN68meLiYh566CEGDTozPH7dunW89dZbWCwWevfuzaOPPurMUCql\nWdZERKQmclrxXr9+Pfv37+fzzz8nLS2NESNGlCve06ZNY86cOURERDBmzBgGDx5M8+bNnRVOpSI0\nUYuIiNRATiveXbt2pX379gAEBASQl5dHSUkJFouFuLg4AgMDiYyMBKBPnz788ssvV714+9qs+Hla\nyC4oISO/mCBvvestIiLuz2nF22KxOIbkL1y4kN69eztefE9KSiIkJMRxbEhICHFxcedtLzjYB6vV\nct5jylQ1Oq8yDYK82ZeQTaHVclHnXU3uGtelUj7uTfm4N+Xj3q5WPk5/z3vlypUsXLiQjz/++LLa\nSUur3vJrFztUv97pu+09x9II93C/8Xt6lcK9KR/3pnzcm/KpXpuVcWrx/umnn/jggw/46KOPyk3q\nHh4eTnJysuPnhIQEwsPDnRlKlfS6mIiI1DROu9XMysri9ddf58MPPyQoKKjcvkaNGpGdnU18fDzF\nxcWsXr2anj17OiuU8wr30yxrIiJSszjtznvp0qWkpaUxefJkx7bu3bvTqlUrBg4cyIsvvsif//xn\nAIYOHUpUlGtmONOdt4iI1DROK96jRo1i1KhRVe7v2rUrn3/+ubMuX20q3iIiUtO43witqyxCE7WI\niEgNo+J91spiNWSNFhERqePqfPH29rAQ4GmlsMQgLa/yNWNFRETcSZ0v3qA5zkVEpGZR8UaD1kRE\npGZR8UbFW0REahYVbyDczwZAQrbW9RYREfen4s3Zd975Lo5ERETkwlS80bveIiJSs6h4c2a0ubrN\nRUSkJqiTxTutIJUFB/5FekEaABF+Z+687ZqoRURE3FydLN6rjn/L3H3/ZOmxJQB4eVgI9LJSbDdI\nzdVELSIi4t7qZPH2tfoCcCr3uGObnnuLiEhNUSeLd5h3BACJ+QmObeF611tERGqIulm8vUqLd3Je\nomObJmoREZGaom4Wb+9wAJLyExwriZ29upiIiIg7q5PF28fqi5/Vn0J7IRmF6cCZEee68xYREXdX\nJ4s3nHnunXT6ube6zUVEpKaou8Xbq7TrPDGvfPFWt7mIiLi7ulu8T995J5++8w4rm6glu5ASuyZq\nERER91Vni3d42etip++8Pa1mgr09KLEbpOZqmlQREXFfdbZ4h5a9LpZf8XUxTdQiIiLurM4W7zN3\n3qfObNOgNRERqQHqbPEum6glKe/MLGtlr4ud0qA1ERFxY3W2eNfzCsOEidSCFIrtxcDZ3eZ65i0i\nIu6rzhZvq9lKiGc9DAxS8pMACPe3Aeo2FxER91ZnizdoohYREamZ6nTxDveuD2iiFhERqVnqdPEO\nPT3LmmOiFt/S4p2cXaCJWkRExG05tXjv27ePAQMGMH/+/Ar7Pv30U0aNGsVdd93FK6+84swwqnTu\nRC02q5kQHw9KDEjO0aA1ERFxT04r3rm5uUydOpUePXpU2Jednc2cOXP49NNPWbBgAQcPHmTbtm3O\nCqVKoZW9Lqbn3iIi4uacVrxtNhuzZ88mPDy8wj4PDw88PDzIzc2luLiYvLw8AgMDnRVKlcIdA9bO\nmmXNT8+9RUTEvVmd1rDVitVaefOenp48+uijDBgwAE9PT2655RaioqKcFUqVKp2oRXfeIiLi5pxW\nvM8nOzubDz/8kG+//RY/Pz/uvfde9uzZQ+vWras8JzjYB6vVUq32w8L8q3VcqOGHp8WT7OIsfILM\n+Hr40iwyALadILPYqHY7zuYucVwpyse9KR/3pnzc29XKxyXF++DBgzRu3JiQkBAAunTpQmxs7HmL\nd1pabrXaDgvzJykpq9qxhHqGczw3jt1xB2niH4WfqXT70cSsi2rHWS42H3enfNyb8nFvyse9OSOf\nqv4YcMmrYg0bNuTgwYPk5+cDEBsbS9OmTV0RCqHepc/kyyZq0eIkIiLi7px25x0bG8uMGTM4fvw4\nVquV5cuX069fPxo1asTAgQN54IEHGDduHBaLhY4dO9KlSxdnhXJe4ec899YzbxERcXdOK97R0dHM\nmzevyv2jR49m9OjRzrp8tYWeM0VqmK8NE6XveReX2LFa6vQ8NiIi4obqfGVyvC6WV/q6mNVipp6v\nDQNN1CIiIu6pzhdvx+ti+acc29R1LiIi7kzF+5wpUuGs4q2JWkRExA2peDsWJ0nCbtgBCPfTnbeI\niLivOl+8va0++Hn4U2QvJKMwHVC3uYiIuLc6X7wBwr1K1/U+93WxxCwNWBMREfej4s2Z595nJmqx\nAbrzFhER96TizZnn3hUmatGANRERcUMq3lS88w719cRsgtScQopK7K4MTUREpAIVb8686132upjV\nbCL09EQtSdl67i0iIu5FxZuz77wTHds04lxERNyVijcQ5l3+mTecNeJcz71FRMTNqHgD9TzDMGMm\nrSCFInsRoIlaRETEfal4A1azlRCvUAwMUvKTAHWbi4iI+1LxPq3sdbHEChO1qHiLiIh7UfE+rWzQ\nWnLZRC1+etdbRETck4r3aee+LqZucxERcVcq3qed+7pYPV8bFrOJ1NwiCos1UYuIiLgPFe/TwsuK\nd94pACxmE2G+pXOc63UxERFxJyrep5V1m589UUu4us5FRMQNqXif5ug210QtIiLi5lS8TwvwCMRm\ntpFTnE1OUQ4AEZqoRURE3JCK92kmk4kw7/rAWa+LqdtcRETckIr3WaqcqEUri4mIiBtR8T7Luet6\n611vERFxRyreZwn3Kj9oLTKgtHjHp+dhNwyXxSUiInI2Fe+znHvnHeJjI8Lfk5zCEo6k5royNBER\nEQcV77M43vXOO/Oud3R9fwBiT2a5JCYREZFzqXifJcz79IC1/FOObe0iy4p3pktiEhEROZeK91nK\n7ryT85OwG6XzmUfXDwAg9pTuvEVExD04tXjv27ePAQMGMH/+/Ar7Tp48yV133cUdd9zBCy+84Mww\nqs3L6k2ARyDF9iLSC9MAaBPhh8UEB5NzyCsqcXGEIiIiTizeubm5TJ06lR49elS6f/r06YwfP56F\nCxdisVg4ceKEs0K5KKGnu841F0HNAAAgAElEQVTLRpx7eVhoHuaH3YDdCbr7FhER13Na8bbZbMye\nPZvw8PAK++x2O5s3b6Zfv34ATJkyhQYNGjgrlIty7uticGbQ2k4NWhMRETdgdVrDVitWa+XNp6am\n4uvry2uvvcbOnTvp0qULf/7zn8/bXnCwD1arpVrXDgvzv+h4yzQObsT6RMi1pDvauaFlGF/tOMn+\ntLzLavtSueKazqR83JvycW/Kx71drXycVrzPxzAMEhISGDduHA0bNmTChAmsWbOGvn37VnlOWlr1\n3rMOC/MnKenS75D9CQHgSPIxRztNTq/rvflI6mW1fSkuNx93o3zcm/Jxb8rHvTkjn6r+GHDJaPPg\n4GAaNGjANddcg8VioUePHuzfv98VoVRw5nWxM93mTUK88fO0kJhdSKKmShURERdzSfG2Wq00btyY\nI0eOALBz506ioqJcEUoFYV6nVxY7a6IWs8lEu7LJWvTKmIiIuJjTus1jY2OZMWMGx48fx2q1snz5\ncvr160ejRo0YOHAgzzzzDE8//TSGYdCyZUvH4DVXCz89RerZd94A7eoHsOFoOjtPZtKvRagrQhMR\nEQGcWLyjo6OZN29elfubNGnCggULnHX5SxbiWQ8zZtIKUigsKcRmKX3eHR2pO28REXEPmmHtHBaz\nlXpeYQCkFCQ5tpe9LrY7IYtiu1YYExER11HxrkTYORO1AAT72GgQ6EVekZ1DKTmuCk1ERETFuzKO\n1cXOee6tFcZERMQdqHhXomxd78S8cwatRZbNtKYVxkRExHVUvCsRVskUqQDXaYUxERFxAyrelSh7\nXSwpP7Hc9pbhfljNJg6n5JJdUOyK0ERERFS8KxPqVTZg7VS57Z5WMy3D/DDQCmMiIuI6Kt6VCPcu\nnWXt3DtvOOt9bw1aExERF1HxroS/RwCeFi9yi3PIKcout0/TpIqIiKupeFfCZDIRdrrr/NwR59GR\npwetnczEMDRZi4iIXH0q3lUI8678Xe/GQV4EellJzS3ilFYYExERF1DxrkJVr4uZzl5hTM+9RUTE\nBVS8qxBexZ03nNV1fkqTtYiIyNWn4l2F0CruvAHdeYuIiEupeFehqola4Ezx3puQTXGJ/arGJSIi\nouJdhaqeeQMEentwTZA3BSV29idrhTEREbm6VLyrEHp6WdDk/ETsRsW763aarEVERFxExbsKXhYv\nAmxBFBvFpBekVthfNmhtpwatiYjIVabifR6OiVoqG3GuQWsiIuIiKt7n4Ri0Vslz7xZhvtgsJo6m\n5ZGZX3S1QxMRkTpMxfs8zjdozcNiplV46d33Ts1zLiIiV5GK93mUTZFaWbc5aIUxERFxDRXv8yi7\n807Oq/iuN5x57q07bxERuZpUvM/jzJ33qUr3a4UxERFxBRXv86jvHQlAfPYxCksqriAWGeBJiI8H\nGfnFHM/Iv9rhiYhIHaXifR4hXqFcG9CSvJJcNiWtr7BfK4yJiIgrqHhfQN8GAwH44eTKSvef3XUu\nIiJyNah4X0DvyP4AbEj4mbzi3Ar7HXfeGrQmIiJXiYr3BYR716dtcHsK7AWsT/i5wv529f0xAfuS\nsiks1gpjIiLifCre1dA3srTrfM3JFRX2+XlaaRriQ1GJwb6k7KsdmoiI1EFOLd779u1jwIABzJ8/\nv8pj3nzzTcaOHevMMC5br8ibMWNmS9IGsgorPtvWZC0iInI1Oa145+bmMnXqVHr06FHlMQcOHGDj\nxo3OCuGKCfYMoUNoZ4qNYn4+tbrCfkfx1gpjIiJyFTiteNtsNmbPnk14eHiVx0yfPp3HH3/cWSFc\nUWVd55WNOm9Xv2zEue68RUTE+azVOSg2NpakpCRuvvlm/v73v7Nt2zb++Mc/0qVLl6obtlqxWqtu\nftGiRXTr1o2GDRtWK9DgYB+sVku1jg0L86/WcRfj1sBbeGfnG+xI2YLJN59Qn7AzsYX44u1h4XhG\nPmZvG/X8PK/otZ2RjyspH/emfNyb8nFvVyufahXvadOmMX36dDZt2sRvv/3G888/z8svv8zcuXMv\n6aLp6eksWrSITz75hISEyhf9OFdaWsXXtCoTFuZPUpIz7oBNdAm7gV8SfmJx7Nf8PmpUub2tw/3Y\nejyD5duOM7BVWBVtXDzn5eMayse9KR/3pnzcmzPyqeqPgWp1m3t6etK0aVNWrVrFyJEjad68OWbz\npfe4r1+/ntTUVO655x4mTZrEzp07efXVVy+5vaulz3m6zm9uEQrA17Enr2pMIiJS91SrAufl5bFs\n2TJWrlxJr169SE9PJzPz0gdnxcTEsHTpUr744gveeecd2rVrxzPPPHPJ7V0t3cN74mnxYk/6Tk7l\nnii375a24XhazGw4mk5cWp6LIhQRkbqgWsX7iSee4JtvvuHxxx/Hz8+PefPmcd999533nNjYWMaO\nHcvixYuZO3cuY8eO5ZNPPmHFiorvStcUXlZveoTfBMAPJ8rffQd4eTCwdWl3+eLfdPctIiLOU61n\n3jfccAPR0dH4+fmRnJxMjx496NSp03nPiY6OZt68eRdsu1GjRtU6zl30aTCQNSdX8MPJlYxqPq7c\nvtvbR/LfnQn8v9hTPHxjU2xWzYEjIiJXXrWqy9SpU1m2bBnp6emMHj2a+fPn8+KLLzo5NPfUKbQb\nflZ/Dmcd4GjW4XL72tX3p2WYLxn5xXy/P9lFEYqISG1XreK9a9cu7rzzTpYtW8aIESN4++23OXr0\nqLNjc0s2i42e9fsC8MM506WaTCZu71C6BvhXO06ce6qIiMgVUa3ibRgGAGvWrKFfv34AFBYWOi8q\nN9enbJnQEysdn02Zwa3D8fGwsO14JgeSc1wRnoiI1HLVKt5RUVEMHTqUnJwc2rRpw5IlSwgMDHR2\nbG6rfb2OBNtCOJEbz/6MPeX2+dqsDGlbOqvc4h0auCYiIldetYr3tGnTePPNN/n4448BaN68Oa+/\n/rpTA3NnFpOFmyJLeyDO7ToHuK19adf5/3YlkFdUclVjExGR2q9axTs/P5/vv/+exx57jIkTJ7J2\n7VpsNpuzY3Nrfcu6zk+uwm6UX8e7ZZgf10UGkFNYwnd7El0RnoiI1GLVKt7PP/882dnZjB49mpEj\nR5KcnMxzzz3n7NjcWuugaMK965OSn8TOtO0V9p8ZuKaucxERubKqVbyTk5N56qmn6Nu3LzfffDPP\nPvtsteckr61MJhN9IgcAsOZExelS+7cIJcDTyu6EbHadqj1z94qIiOtVe3rUvLwzU37m5uZSUFDg\ntKBqirKu859Pfk+xvbjcPi8PC8PaRQCwSHffIiJyBVVrhrVRo0YxZMgQoqOjAdi5cyd/+tOfnBpY\nTRDl35zGfk2Jyz7CtpRNdAm7odz+Ee0j+c+W4yzfk8ifejfD36taH7eIiMh5VevO+4477mDBggX8\n/ve/Z8SIEXz22WccOHDA2bG5vfJd5xVHnTcN8aFL4yDyi+0s2123HzOIiMiVU+3JtyMjIxkwYAD9\n+/cnIiKCHTt2ODOuGqOs63xdwg8UlFR8lHD2wLVzJ3QRERG5FJe8coYKUamGvo1pEdCavOJcNiX9\nUmF/n2vrEeLjwaGUXLafuPRlVEVERMpccvE2mUxXMo4arXeDqrvOPSxmfhddH4CvtmvgmoiIXL7z\njqDq06dPpUXaMAzS0tKcFlRN0yeyP3P2vMOviWvJLsrCz8O/3P7fXxfJv36NY9X+JP6cdy1B3h4u\nilRERGqD8xbv//znP1crjhotzDuC60I68lvqVh5f9yBPXf8SzQNbOfY3CPTixqgQ1h5O5ZvYU4zt\n2tiF0YqISE133m7zhg0bnvcfOeOx6Cdp4hdFfM4xHl/3IF8enE+JcWZe87L5zhftOIld4wVEROQy\nXPIzbymvkV8T/tHzY25tcgfFRjEf732PZzb8iaS80lfEekaFEOHvSXxGPhuPpbs4WhERqclUvK8g\nT4snE9s9wUtd/kaQLZgdqVt45Kdx/HTyeyxmEyOu08A1ERG5fCreTtAt/Ebev2ke3cJuJLs4i1e3\nPsdb26cxqHUAFhP8eDCZpGxNLysiIpdGxdtJgjxDeLHLGzza7v+wmW2sOL6UF7Y9SJfmGZQYsFjz\nnYuIyCVS8XYik8nEsCa3MbPnJzTzb8Gp3BMctszAu95K5m06Snx63oUbEREROYeK91XQxD+Kv984\nm9uj7saOHd+wlXiEL2Dqd3s18lxERC6aivdVYrPY+EObSbza7R94W3zwDNjBrpylGrwmIiIXTcX7\nKusY2pUn2j8LgG/4Ut79dTXHM9R9LiIi1afi7QK9Im9mRNNRmEx2bBHzeem7zeo+FxGRalPxdpHx\nrR+lZUA0Fo8MDhqz+Wp7vKtDEhGRGkLF20WsZivPdZ6GtzkAm99+Poj9iBMZ+a4OS0REagAVbxcK\n8w7nuc4vg2HCI3glz6xaonXSRUTkgpxavPft28eAAQOYP39+hX3r169n5MiRjB49mr/+9a/Y7XZn\nhuK2OoV14/ao+zCZDE5aP2Le1lhXhyQiIm7OacU7NzeXqVOn0qNHj0r3v/DCC8ycOZPPPvuMnJwc\nfvrpJ2eF4vbGt3mAJt4dMVtzmX/kVeLSs1wdkoiIuDGnFW+bzcbs2bMJDw+vdP+iRYuoX790oY6Q\nkBDS0tKcFYrbM5vMTL9xGh5GCBavo/zlxxnqPhcRkSo5rXhbrVa8vLyq3O/n5wdAYmIia9eupU+f\nPs4KpUYI8gzm2U5TMQwLGdbv+fuvi10dkoiIuCmrKy+ekpLCww8/zJQpUwgODj7vscHBPlitlmq1\nGxbmfyXCu+qGhfXml4QJLD/xPt8lzWJEQXfCaF1j86mK8nFvyse9KR/3drXycVnxzs7O5sEHH2Ty\n5Mn06tXrgsenpeVWq92wMH+SkmruM+PH2t/DL/EbyTRvYvKKyawas4js9BJXh3XF1PTv51zKx70p\nH/emfKrXZmVc9qrY9OnTuffee+ndu7erQnBLZrOZv930MkZhGIXm40xc+qyrQxIRETfjtDvv2NhY\nZsyYwfHjx7FarSxfvpx+/frRqFEjevXqxZIlSzh69CgLFy4EYNiwYYwaNcpZ4dQojQODGBP1LPPj\n/kJs+kq+PTyImKgL906IiEjd4LTiHR0dzbx586rcHxur95nP5572XVkVN4xTLGbm9nfoHtGNYB+b\nq8MSERE3oBnW3JTJZOKtARMx2wMwbMeYtOxTCovr5kQ2IiJSnoq3Gwv28uPhjhMBSDQvYep3u/X+\nt4iIqHi7u3HtRxHq2QCrZxKrTy5jzoZjrg5JRERcTMXbzVnNHjzQ5iEAfEJX8uEvB1i+J9HFUYmI\niCupeNcAvSP7c21ASywemXgHr+Pl5XvZfiLD1WGJiIiLqHjXAGaTmftbPQxAYPiPFBm5/OXrXcSn\n57k4MhERcQUV7xqiU2h32od0opgcmkdtIC2viCeW7CQrv9jVoYmIyFWm4l1DmEwm7m9dOvI8x3M1\nTcKKOJyay1//t4viEr1CJiJSl6h41yCtg9pxY0QfCu0FdG33KyE+Hmw4ms4bqw/qFTIRkTpExbuG\nubfVQ5gx82PCUp4aHITNYmLRjpP8Z8txV4cmIiJXiYp3DXONX1MGNBqK3ShhQ/p/mBLTCoB//HCI\ndYdTXRydiIhcDSreNdCYFg/gYbbx48lVRNVPZUKPJhjA9FX7yS+qPcuHiohI5VS8a6Aw7wiGN7kd\ngH/t/YD7u19DizBfTmYW8MmvmoFNRKS2U/GuoUZdOw4fqy9bkn8lNnUzT/VvDsC8TfEcTct1cXQi\nIuJMKt41VIAtkDua3QPAJ3vfp31kAMPbRVBUYvC37zX6XESkNlPxrsFGNB1FsC2EfRm7WXtqDZNu\nisLf08r6o2l8vz/Z1eGJiIiTqHjXYF5Wb+5ucT8A/973IYFeZh7p1RSAt9YcJLdQg9dERGojFe8a\nLqbx74j0aUh8zjFWHF/KiOsiaRPhR2J2IXPWH3V1eCIi4gQq3jWc1WxlXMsHAZi/fw7FRiFP9W+B\nCfh0y3EOpeS4NkAREbniVLxrgd6RA7g2oCUp+Ul8c3Qh7er7M6J9JCV2g9dXHdDgNRGRWkbFuxY4\ne8nQzw/OJbsoi4k9mxLoZWVzfAbL9yS5OEIREbmSVLxribIlQ7OLsvjy4HyCvD34401RALz94yGy\nC7R0qIhIbaHiXUucvWTo10e+ICU/ieHR9bkuMoCUnEL++YsGr4mI1BYq3rVI66B29IzoS4G9gP/s\n/wSzycRT/ZtjNsEXW4+zPynb1SGKiMgVoOJdy9zbagJmzHwb/w3x2cdoFe7HnR0aUGLADA1eExGp\nFVS8a5nGfk0Z1PgW7EYJc/f9E4CHbmxKiI8H209k8r9dCS6OUERELpeKdy10d/MHsJlt/HTqe/al\n78bfy8qfejcDYOaPh8nML3JxhCIicjlUvGuhMO9wbm16J1C6aAnAkDbhdGwYSFpeEVO+3cuvx9Io\nKLa7MkwREblEKt611MhmY/Gz+rMtZRNbkzdiMpl4sn9zrGYTPx9K5dGFv9H/vXX88avf+HRzPAeS\nc/Q8XESkhrC6OgBxDn9bAHdeew+f7P2AT/a8T4eenWke6su/7u7I8j2JrD+axv6kHNYfTWP90TQA\nQn1tdG8SRPcmwXS7Jph6vjYXZyEiIpVxavHet28fjzzyCPfddx9jxowpt2/dunW89dZbWCwWevfu\nzaOPPurMUOqkW5uO5OsjC9mfuYefT62md2R/WoX70Srcj8eAlJxCfj2WxoajaWw4mk5yTiH/25XI\n/3YlAtA2wp8pMS1pVs/XtYmIiEg5Tus2z83NZerUqfTo0aPS/dOmTWPWrFksWLCAtWvXcuDAAWeF\nUmd5Wby4p8V4AP6990OK7eVnWavna2NImwhejGnN0gndWTCuM5P7NKNH02A8rWZ2JWTx8Bc7OJCk\nxU1ERNyJ04q3zWZj9uzZhIeHV9gXFxdHYGAgkZGRmM1m+vTpwy+//OKsUOq0QY2G0dCnMSdy4/ku\n/r9VHmcymWge6ss9nRsx87brWDGxBzc0CSYtr4iHv9zO3kRN8CIi4i6c1m1utVqxWitvPikpiZCQ\nEMfPISEhxMXFnbe94GAfrFZLta4dFuZf/UBrgMvN549d/8TTP/wfCw5+wqgOd+Bl9b7gOcX2Yt4d\n144nv9zHqj2JPLJwB/PGd6dD46DLigX0/bg75ePelI97u1r51JgBa2lpudU6LizMn6SkLCdHc/Vc\niXza+/SgRWBr9mfs4aONHzOq+bgKx5TYizmQuY8dKVv4LXUrsWnbySvO5caIvtxQ3If1B7y5+6P1\n/OO2aDo0CLzkWPT9uDfl496Uj3tzRj5V/THgkuIdHh5OcnKy4+eEhIRKu9flyjCZTNzf6hGe+fUx\nvjw0n6HX/B4fq4+jWO9I3crO08X6bGaThXUJa8C6hmtbdeLIkV489hX8fUQ7OjW6/DtwERG5NC4p\n3o0aNSI7O5v4+Hjq16/P6tWr+dvf/uaKUOqMjqFd6FivK1tTNvLELxNIyU8mr6R8sW7o05jr6nWk\nfUgnrqvXERPw5aFPWXZsCRn2LQRHbaEgqy2P/28AbwwZTLdrgl2TjIhIHee04h0bG8uMGTM4fvw4\nVquV5cuX069fPxo1asTAgQN58cUX+fOf/wzA0KFDiYqKclYoctr9rSeyde1G4nOOARWLdahXWIVz\nHm47mZHNxrDw0KcsPbYE/HeB/y6eXb+Ch3MmcHub7lc8zqNZh1h8aAlN/BvQMawrTfyaYTKZrvh1\nRERqKpNRQ6bVqu5zBD1DOb/NSRvILMyoslifT1pBKl8e+pSvD3+FnUIArvXtwh87PEyroLbVauN8\n+RiGwZLDC/lozzvYOTP/erAthA71OnN9aFeuD+1MhHfkRcXtTPp9c2/Kx70pn+q1WRkVbzfnjvmk\n5afy9A/vcbRoBSZzaZHtFt6TB1o/yjV+Tc97blX5pBek8sa2V9iSUvrKYH7G9YAJL7+DYMksd2yk\nT0Our9eFjqFdaV+vE4E21z1/d8fv53IoH/emfNxbrR+wJjVbsFcIHwx6ltd/uIWlcV/iHbyOXxPX\nsilpPTGNb2VMiwcI9gy5cEOnbUpazxvbppJZlIa9xAtSR/LCjXfwya/H2L0/Cy+vJGI6ZlBi28eO\nlK2czD3OydzjLIv7GoCm/tfSJiia1kHtaBMcTUPfazCbNG2/iNReKt5ySUwmE0/2uR7fn4OYu6Un\nPqEr8Q76laXHFrP6+HLuvPYeRkTdhZfFq8o2CksK+Hjv+3x95IvSn3OaUS//Xv4x4iYiA7y4MSqE\nN1cfYPFvJpb8Es7vom9ibt+mHMvdz7bkTWxL2cSutN84knWQI1kHHcXcz8Of1mXFPCiaVkFt8fXw\nuyqfi4jI1aBuczfn7vkYhsGy3Ym8/cMhMkvi8Q3/FpvfbgDqeYYyruUE+jcagsVUOsFOWT5Hsg4y\nY9uLHMk6iGGYyU0aRHv/3zF9WDv8PMv/TfnNzlPMWHmAghI7bSL8mD6sLQ0CS/8oKCgp4EDGXnan\nx7InPZY9abGkFCSXO9+EiWv8mnJdSEfuaTGeoIvoFbgQd/9+LpbycW/Kx73pmXclVLzdW2Z+Ee/9\nfIRFO05i9TlIYP1lYIsHIMq/OX9oPYlOYd0IDfXj403/Ys6edyi0F1JSGErWiVH8rmU3/q9f6ZKl\nldmbmM2T3+ziREY+gV5WXh7SmhujKhZhwzBIyk9gT/pOdqfFsjs9loMZeyk2Sud1D/euz4udXycq\noPkVybumfD/VpXzcm/JxbyrelVDxrhliT2YyfdUB9iZm4hmwneDI7yg2lS452jm0O16eNtYe/wmA\n/PQu5CQM50+923B3p4YXfB0sM7+IKcv28vPhVEzAgz2a8MAN12C+wHmFJQUcyNzHP3fPZG/6Trws\n3jx5/RR6RPS+7Hxr2vdzIcrHvSkf93Y1i7flxRdffPGKXslJcnMLq3Wcr69ntY+tCWpaPuH+ntwa\nXZ8gLxtbDvuQltQVC154+x4nPvcIcVnHMBneZJ4YhZHZn9eGXcet0fWr9R63p9XCoNZhWM0mNsdl\nsDk+g12nsrixaQheHlXPe28xWwnzjqBfg8Ek5J3kQOZefjy5Cg+zB22D21/WO+Q17fu5EOXj3pSP\ne3NGPr6+npVuV/F2czUxH7PJRHRkALe0jSAxu5jdcWHkpHUh0NuM2R5C8pF7CLK04J07rqPrRc7S\nZjKZ6NQoiOsi/Vl3OJUDybl8szOBwhI7zUN98TzP4jUWs5UbI/pgs9jYlrKZrSmbOJV3gi5hN2Ax\nX9rYzZr4/ZyP8nFvyse9qXhXQsW75vG1WenfMowODQL47Xg+xxOakp3ejuYh9Xh/ZHuiQnwvue1G\nQd4MbBXGrlNZHEnLY1NcBl9uO0F6XhFR9XwqDHorYzKZaBfSgWYBLdiQuJb9GXvYlrKJbuE98bb6\nXHyONfj7qYzycW/Kx71dzeKtZ95urrbkU1BsZ8GWeHJK4L7ODfC1XZm3FA3DYFNcOv/eGMeGo+kA\nWMwmYlqHM65rI5rVq/oPhEOZ+3lp05Mk5icQ5hXBlC4zuDag5UVdv7Z8P2WUj3tTPu5Nz7wroTvv\nms1qNnF9w0Bu6dSIooLiK9auyWSiYaA3Q9tG0LtZCNmFJRxKyWFfUg4Lt59kb2I29QM8qe9f8X3z\nYM969G0wiN3pv3E0+xCrjn9LY78mF5wl7my15fspo3zcm/Jxb+o2r4SKd+3gzHxC/Tzp3zKMIW3C\nKbYbHErO5VBKLv8vNoGNx9II9vHgmiDvcgPUvK3e3NxgMEn5CezP2MOPJ1dhMVmIDu5QrYFs+n7c\nm/Jxb8qnem1WRsXbzSmfixfg5UGvZvX4ffv62CwmDiTnciwtj+V7kvj5cCoR/p40CvJyFGeL2UKP\niN54Wb3ZlryJ7Smb2Z+xh1ZBbfG3Bbg8n6tJ+bg35ePermbx1gTQUmuF+NiY2DOKbx7sxuQ+zajn\na2N3QjZ/WhzLHz7fzqZj6Y5jTSYTdzS7hxc6z8Db6sOvSet46Me7mbPnXXKKclyYhYhIRSreUuv5\n2qzc07kRS8Z35bHeUQR6WdlxIpOJC3cw8csdbD+R4Tj2hohefNT7MwY2uoUSo4SFhz7lDz+MZHnc\nN5QYJS7MQkTkDHWbuznlc+VYLWY6NAjk9g6ReFnN7EnM5khqHv8vNoGdJzNpGuJDmJ8n3lYfekT0\npmvYjRzLPkxczlHWJ/7MhsS1XOPXlHDv+m6RjzMoH/emfNybnnlXQsW7dnCHfGwWM50aBXFb+0is\nZjN7E7M5lJrL4t9OsT8pm2YhvoT42qjnFcagRsNo5NeEvem7OJ5zjBXx/yMu+wgtg9rg5+HvFvlc\nScrHvSkf96biXQkV79rBnfLxtFroek0Qv7+uPoZRuvjJwZRcFu04SUJWAZ0aBeJptdDU/1qGXvN7\nPMxW9qbv4lDWfpYeW0KRvYiOkR0ozK8RUyVUizt9P1eC8nFvyqd6bVZGxdvNKR/n8/awcEPTYG6N\njqC4xGBvUja7ErL5bm8iLcP8aBDohdVspX29TvRvOIS0glQOZu0nNnUbi/d9RVz2UeyGnTCvcDzM\nHq5O57K44/dzOZSPe1M+1WuzMpphzc0pn6vvUEoOL367l90J2QCM7tiQR3s1Lbf4yc7UHXy4+232\nZ+xxbPMw2+hQrxPdwnrSLeJGIrwjr3rsl6smfD8XQ/m4N+VTvTYro+Lt5pSPaxSX2Pnk1zjmbDhG\nid2gSbA3L8W0ol3kmfe+DcMg0+MUy/Z8x4bEn9mbvguDM/93aup/Ld3De9I9vBctg9pgMVW9aIq7\nqCnfT3UpH/emfKrXZmVUvN2c8nGt3QlZTPl2L4dTcrGY4N5u1/CHG67Bw1L6luXZ+aQXpLIx6Rc2\nJKxlS/Kv5JXkOtoJsgVze7O7+X3TUVgvcQWzq6GmfT8Xonzcm/KpXpuV0TNvN6d8XCvMr3R98oJi\nOztOZLL1eAY/H06lfUZ2NdkAACAASURBVIMAQnxs5fLxsnpzbUBLejfoz4io0USHdMDXw5/0wlRS\nCpLZmryRdafW0NT/2nKvm7mTmvb9XIjycW/Kp3ptVkaTtIhcgKfVzOQ+zfhgZHsaBHqxNzGbcZ9u\nYe7GOErslXdc2Sw2Ood155F2T/Cvvl8xtetbRPo05Gj2Yf6y/hHe3D6N9IK0q5yJiNQWKt4i1dSp\nURD/GduJEdfVp6jEYNZPhxnx3loW7ThJel5RleeZTCa6hN3A+zfN5+7m47GaPVh5fCkP/jiapceW\nYDfsTo3bbthx96djBSUFHMjY6/TPQqS20DNvN6d83NPaw6lM+24fyTmlXWQWE3S9JpiBrcLo27we\nAV5VvzJ2PCeOd3f+ja3JGwFoFdSOSe3+j+aBra5IbLnFOexO+42daTuITd3O3vSdlBgl+HsElP5j\nCySg3L8H4m8r/fmasEjsuTYCbYH42wKd/upbWkEq/z26iP8eW0RmYTrRwR2YFP0Xmvg3uyLt15bf\ntzLKx71pwFolVLxrh9qUT3ZBMRtPZbN4cxy/Hkt3dKFbzSZuaFJayHtfWw8/z4oD1AzD4KdT3/Ph\nrn+QWpCMGTPDm97B2BYP4uvhe1FxpBekEpu6ndi07exM3c6hzP3YuTJ3sD5WXwJsgQR6BBFgCyTA\nFkSgLZBAWxCtg6JpHRSNzWK76HaPZB1k8eHP+f7Ecortpb0WHmYbRfZCLCYLt0Xdxd3N78fL6n1Z\n8dem3zdQPu5OxbsSKt61Q23NJz2viNX7k1mxL4nNcemUPQq3WUz0aBrCwFZh9Lm2Xrl3xQFyinKY\nt3823xxZiB07IZ6hDG48DIvJgmEY2DEAw9H17fifYZBVlMmutB3E5xwr16bFZKF5YGvaBbcnOqQD\n7YI74GXxJrsok6yiTDKLMsgszCS7KJPMwgyyijLJKsogqzCLPLJIyUklszCDjKIM7BdYjMXT4kX7\nkI50Cu1Gx9BuXOPXtMp10A3DYEvyBhYd/owtyb8CYMLEDRG9GBE1mqb+1/LvvR+y9NgSDAzCvev/\n//buPTyq6lz8+HfP/ZZkMpkkkATInQSQAiIVEREKVsVTa60W/Vnbam37o9qeWrUYi/TKRX08Vnra\nqkfPeQ7esGr7s7VVVNSi5SJogYRALuSekGSSzCSTmUnm9vsjMBIZ7iSZGd7P8+TZM3tmNuvlTfJm\nr7X3Wiyf8mM+nznvzJJD4n6/JQqJ59SOGY0U7xgn8cS2aPF0ewbZXO3grQOdfNLsitz5bdap+WJJ\nBl+aNo4pmZZhRa7GdYDfVjzCAWfFabdBrzZQap0WKdQl1qlnfMZ6dDyhcAhPoB/XoJPeQSeuQRe9\ng056/S46ve3s6f6E+r7aYZ9PM6Qzyz6HWfaLmJF2EVZ9KoPBAd5t3cSf6l6kwV0XafMVOUu5NvdG\nss0Thh1jv7OC9eUPcbC3GoBLMhfwvSn/Trox86ziSQQST2xLmOK9evVqdu/ejaIolJWVMX369Mhr\nzz33HK+99hoqlYpp06bxwAMPnPBYUrwTw/kWT6d7gM3VDt6o7KD80KfvK7Sb+dK0cVxVmoHVODSu\nHAqH+Efb2zS661FQoQCKokKlKICCChWKoqCgoCgq9Co9xdZSCpKLz9m946ebny5fJ584dvKJYwcf\nO3bgHBx+BX1BcjEOXweuwaG109P0dr6UewNXTbiWJF1ytEMCEAwFeK3hFTZUPYU36MGgNnJL0e1c\nm3vjacV6vn2/xRuJ59SOGc2IFe8dO3bw9NNP88QTT1BbW0tZWRkbN24EwO1286UvfYlNmzah0Wi4\n7bbb+MEPfsCMGTOOezwp3onhfI6n1tHPa+WH+FtlR+TqdK1aYUGBnWunZXLRxFTUquhdzqPlbPIT\nCoeo76vlY8dHfOzYTnn3bvyhoQv6CpMnc13eMuaPX3RaF8E5fJ08se8xPjj0LgB5SYXcNe1eSlMv\nOKXPn8/fb/FA4jm1Y0YzYlM9bd26lcWLFwNQUFCAy+XC7XZjsVjQarVotVo8Hg8mkwmv10tKSspI\nNUWImFBgN/Ojywu4c34eW2q7+H/lh9jW0MPbVZ28XdXJuCQ910zN5NI8GzazjlSj9pgx8limUlTk\nJxeRn1zEV/NvZiA4wL6evRjUBkqsU487Fn4idkM6D8z6NTs6/snvKx6lrq+Gu7d+l4vS5zI7fS6z\n7HPINk84o2MLEc9GrHg7HA6mTp0aeW6z2ejs7MRisaDX6/n+97/P4sWL0ev1LF26lLy8vJFqihAx\nRatWsag4nUXF6bT3DfDXikO8VtFOq8vHf21r5L+2fXoBmkGjItWkJdWoI9WkxWrUkmrUYjNpsVt0\nzMxJYVySYQyjOT69Ws9M++xzcqw5GZcwPW0WL9b8D68cfJ6POrfyUedWADKM4w6Ps89hRtrsE3bH\nC5EoRqzbfOXKlSxYsCBy9n3TTTexevVq8vLycLvdfO1rX2PDhg1YLBa+8Y1vsGrVKkpKSo57vEAg\niEYTP2chQpyOUCjMtrouXt7VzP5DfXT3D9LdP8hg8OS3fBVmWLisKJ3Liux8Pi8Noy6xf046PR1s\nbfmQba1b2dG2bdhMdSpFRWnaVC7OmsvFWZdwQfp0NGdxr7o/6Mft76Nv8MhXLwaNkaLU4tO+pU+I\nc2nEzrwzMjJwOByR5x0dHaSnpwNQW1vLhAkTsNlsAMyePZvy8vITFu+eHs9xXzuajKHENonn+IqS\n9dy/sCDyPBwO0z8YxOn10+Px0+314/QM0uP10+P109Tj5eNmFzUdbmo63DzzYR06tcKM7BQuzk3l\n4kmpFNrNx3Qp+/xBWnt9NDt9NDu9Q1uXlxanD18wRKZFR1aygawUA9kpQ9usFAOZSQY0YzwmP8TI\nXOti5loXEyoNcbC3mo8d2/nYsYOK7j1UOPZS4djL03ueREFBp9KhVevQq/To1Hp0Kt3hrR7d4f0a\nlRZf0Ee/v4/+QD/9/j7cATcDQV/UFigoZJlyKEgppiC5iPzkYgqSi0nV20Y0cvn5iW0JMeY9b948\n1q9fz7Jly6ioqCAjIwOLxQJAdnY2tbW1+Hw+DAYD5eXlLFiwYKSaIkRcUhQFi16DRa8hxxr91q9A\nMMSetl62N/Swtb6H/e1udjQ62dHo5HHqsJt1fH6SFbWi0OT00eLy0uE+8cIJHX0D7G079heQWoHM\nJP3hom7kkjwb8/NtkRXWxoJKUVGYMpnClMncWHArvoCXPd2f8PHhq9+b3PUMhAYYCA3g5vR/qaoU\nNRaNBbPWgkWThElrps/fS2NfHS2eJlo8Tfyj7Z3I+216O/nJRRQkF1GQXEyxtZQMwzgZkxfn3Ije\nKvbII4+wc+dOFEVh1apV7Nu3j6SkJJYsWcKLL77Iq6++ilqtZubMmdx3330nPJZcbZ4YJJ6R5fT6\n2dHQw9aGHrbX99DZf2yhVqsUspL1ZKcYybEayLEayUk5vB2XTEV9F60uH60uHy0uH629Q4873YN8\n9pdFikHDlSUZXDM1k8kZlpgrUrY0E60d3QwGBxgMDTIYHGAgOMBg6PBXZP8gRo0RsyYJi9aCWZuE\nRWNBrzZEjckf8tPorqPWVcXBvmpqXVXU9lXjDRzbQ5iiszI5ZQrF1lKKD29TdNYziifWvt/OlsRz\naseMRiZpiXEST2yL5XjC4TC1XR52NTnRqJSh4mw9cff3ieIZDIRo6xsq5NWd/fy9soMaR3/k9UK7\nmWumZnJlSQZp5tOfMnUkjGZ+QuEQhzyt1PZWRb6qnJX0+l3HvDfTOJ7J1ikUpZQy2VrK5JQp6NTR\nl348Wix/v50JiefUjhmNFO8YJ/HEtvM5nnA4TFVnP3+pOMQblR24fAFgqHv9kjwb10wdN+bd6mOd\nn3A4TLu3jQPOfVS5Kqly7qO698AxY+lmjYUFWUu4ImcpxSmlx+3BGOt4zjWJ59SOGY0U7xgn8cQ2\niWeIPxjig4Pd/HVfOx8e7CJ4+LdKikHDpfk2Cuxm8m1m8u0mMpP0h2eNG3mxmJ9gKECjuyFSzPc5\n9w6bZnaSJY8lOUtZlH3lMRfAnSyecDhMh+8QjX315CYVkG7MGLE4oml2N/BG01/IsUzkizn/dtJh\nlFjMz9mQ4h2FFO/EIPHEtnMRT1f/IG/s7+Av5Yeo7Tp2DNioVZFnM5OXZqIgzURemon8NDPjks99\nUY+X/NT31fJW8+u80/JGZCpZtaJmTsYlXJFzDbPT56JRaY6Jp2egmypnJVWuSqpdQ9sjn1dQ+Fza\nhSzKvpJ54xZg0ozMrW3hcJh/de3kT3UvRu69B1iYdQU/uGAFBvXx5yGIl/ycKineUUjxTgwST2w7\nl/Ec6Vbf29ZLXZeHg139HOzy0O3xR32/Wadmfn4aXyxJ5+JJqWjOQXd7vOXHH/LzUcc/2dT8Oh91\nbo2s6paqs7Eo+0rm589jd/Neqpz7qXJV0ulrP+YYydoUciyTqHJVRpZb1av0zB23gC9kX8nMtNmo\nz8Fc+EcWnPlz3Ubq3QcB0Kl0zM28jG0dHzAQ9FGQXMzKC9eQaRwf9Rjxlp+TkeIdhRTvxCDxxLbR\niMfp9Q8r5gcPPz66qKcYNCwqsvPFkgxm5qSc8Rl5POene6CLzS1vsKn5dZrc9VHfY1SbKEyZTHFK\n6eGr2UvJNI5HURT6/L1sadvM5pY3qejZHflMqj6Ny8cvZlH2lRQkF5/2HQLdA138teFV/tb4p8hZ\nvk1v55pJX+HqiV8mRWelrreWX368gjZPC8naFO6f+UtmRJltL57zE40U7yikeCcGiSe2jWU8zU4v\nmw508ub+Dg4e1d2ebtaxZHI6V5RkHLOU6skkQn7C4TD7nRW81fw6hwaayDHmUZQyVKhzLBNRKyef\nUe+Qp5XNLW+yueUNWjxNkf2TLHlcmP55TBoLRrURg8aIQW3EqDFiUJsObw0Y1Sacgz38teFV3mt7\nK3JGP7TgzNeYP/4Lxyw40+fvZd0nq9jl2I5KUfPtku/z5dyvDctfIuTnaFK8o5DinRgkntgWK/HU\ndPbz5oEONu3voLV3ILJ/gtXAkskZTM6woFMr6DQqdOqjvjSqYfttaRZqmnro8Q7i9Pgjs9NFHnv8\nOL1+3AMBSjKTWFiUxrw8G2bdiM1fdVbONj/hcJgDrn1sbnmD91vfjnob28koKMzNvIwv593ItNQZ\nJ/xjKhgOsqHqKTbW/i9w7Dj4ufh+C4VDeANePIF+vEEPNn0aFm30gjfSpHhHIcU7MUg8sS3W4gmH\nw+xt62PT/g7equo87nj5uaRTK8yZlMqiIjvz89Mi663HgnOZH3/Iz67O7TS6D+IL+vAGvPiCXnwB\nL96gF1/Qgy/gG3oc8BAizLxxl3Nt7g2MN2Wf1r+1pW0zj+75Nb6gd9g4+IniCYaDNPbVUeksp9pV\niWvQhSfQP1SkAx48AQ/egAdvcPhFkXqVnh9ecD8Ls6844/+bowVCAdSK+pR6fKR4RyHFOzFIPLEt\nluMJhMLsanLybrWDrsOLtgx9hRkMDD32B0MMBEL4g2EGgyFUikKyQUOqUYvVNLQi2/DHOqwmLXq1\nih2NPbxb42B3S29kJjm1ArMmWFlYaOfywjTSLSefSGUkxXJ+Tqa+r5Zf7Bo+Dr6kdGEkHtegk/09\nFex3llPpLKfKWXlMYT4eo8aESW1Cq9JxyNsKwHW5X+P2ku+f1cV5W9o2858Vj2BQG1lW+A0WZ1+N\n5gTHk+IdhRTvxCDxxDaJBxz9g/yj1sHmagc7m1wEQ5/+irxgfDILi9JYWGg/7nzzIyne89Pn7+Wh\nf/2MnZ3bUClqlpXeTLurk8qeclo9zce8f5wxixLrVEpSp2I3ZGLSmDBpzEPFWmPGeHhcXqUM3ZkQ\nDod5vfFP/GHffxAMB5lum8X9M3+B9TQXjOnz9/L7ikd5t3XTMe25qfCbfCH7yqh/FEjxjkKKd2KQ\neGKbxDOcy+vng7pu3q12sK2+h4GjlmgttJu5vDCNywvtFKcfu3rbSEiE/Hx2HPwIvUpPkbWUUus0\nSlOnMdk6FZs+7Yz+jYruPfz6kwfoGejCbsjgp7NWM9k65ZQ+u6tzO/+xdzVdvk70Kj23l9xJki6J\n56qfobm/EYDxpmxuKvwWi7KuGFbEpXhHIcU7MUg8sU3iOT7PYJB/1nfzXrWDD+q66R8MRl7LStaz\noNDOgoI0ZmSnoB6hpVMTKT87Oj5kb99OMtQ5lKZOIzep4IRd0qery9fJrz9+gEpnOVqVju9PvYcv\nTrjmuO/3Bbw8vf8/+WvjqwCUWKdyz+ceJNs8ARj6o+P91rd4vvq/I1fsZ5smcHPRt1iQtQS1opbi\nHY0U78Qg8cQ2iefU+IMhdjY5ea+mi/dru+g6avU2q1HLZfk2Liu0M3VcEvZzuEiL5Of0+EN+/rDv\nMf7W+CcAlk68ju9O+fdjbmur7NnLI7t/SaunGY2i4Zaib/PV/Jujdo0HQwHebX2L52ueoc3TAkCO\neSI3F97G9dO/THeUWQXPhhTvOCXxxDaJJ7aNRjyhcJi9bb28X9PFezUOmpzDFx2xmbQUpZspsluG\ntulmcm2mM1qwRfJzZt5s+iu/rXiYQMjPlNQLKJv5K9IM6fhDfp6rfpo/1j5LiBC5lnzumfEgBcnF\nJz1mMBTgnZY3eKHmfyIXyRWlFvOrC39Dsi7lnLVdinecknhim8QT20Y7niPLsL5f42BbQw/Vnf3D\nuteP0KgU8tJMFKebKbRbuHhSKoXpJ597XPJz5g449/Grj8tw+DpI1adxe8lyXj34Igf7qlFQuD7/\nZr5edAc69en1lARCAd5p+TvP1/w3XT4Hv5v/v0y05J6zdkvxjlMST2yTeGLbWMcTDodp6x2gutNN\nVWc/1Z391Djcx5ydA8zMTuaGGdksLEw77rzuYx3PuTba8TgHuln9yUr2dn8S2TfOlMWPp69kmu1z\nZ3XsQCiAMQX8fed2gp/jFe/YnEZICCESgKIoZKUYyEoxsKDQHtnvGQxS4xgq5PsOuXm7qpNPWnr5\npKUXu1nHV6aP57oLxmEf4/vKE41Vb2P1nN/wzP7/5C8Nr7AkZyl3lN6FUWM662NrVBqshiQ6+0bn\njxE5845xEk9sk3hiW7zE0z8Y4O+VHbz0r1bqDl/wpFYpLCqyc+OMLD6XlYyiKHETz6kay3iCocA5\nWV3taKN5tbmceQshxBgz6zR89XNZXD99PLuaXPxxdyvv1zh460Anbx3opMhu5oYZWdwyf/QnhklU\n57pwj7b4br0QQiQQRVGYPdHK7IlWDvX5+NOeNv689xDVjn5Wv13Nb/5xkJnZKcyeaOWiiVYK7eYz\nXi5VxDcp3kIIEYPGJRn4v/PyuP3zk3inupM//quVvW19fFDXzQd13cDQPeWzJ6Qwe4KViyamMsFq\nGJWZ3sTYk+IthBAxTKdRcVVpJleVZhLQaXjzXy181Ojko8YeOtyDvF3l4O0qBwAZFh0XTUzloolW\nPj8p9ZxOECNiixRvIYSIE+NTjCydksnSKZmEw2GanD4+auxhZ5OTjxqddLgHeX1fO6/vawdg6rgk\nLs23MT8/bdTmXxejQ4q3EELEIUVRmJhqZGKqkes/l0UoHKbG0c9HjU52NAwV9IpDfVQc6uOJfzaQ\nYdExPz+NS/NtzJ5gxaBVj3UI4ixI8RZCiASgUhSK0y0Up1v4Pxfm4PUH2dHoZEttFx/UddPhHuSV\nPW28sqcNvUbFnIlW5uenMXuClfEpBjQjtJiKGBlSvIUQIgEZtWoWFKSxoCCNUDjMgQ53pJBXtrvZ\ncrCbLQeHLnxTqxTGJ+nJsRrJsRqGtikGsg9v5Sw99kjxFkKIBKdSFEozkyjNTOI7l+TS0TfAh3Xd\nbDnYxYEONx3uQZpdPppdPmg49vPpZh05VgOZSQbGJevJTNIzLunI1oBFr5bx9FEmxVsIIc4zGUl6\nrps+nuumjwfA5w/S2uuj2emj2ekd2rq8tDh9tPb66OwfpLN/EOiNejyzTk3GUQU9J8VIrs3IJJuJ\nnBTDcedqF2dOircQQpznDFo1+Wlm8tOOXdksGArT3jdAi8vLob4B2vsGONQ7tG3vG+BQn4/+wSB1\nXZ7I1K5HU6sUclIMTLKZyE0dKuiTbEZyU02kj0ZwCUqKtxBCiONSqz5dXCWacDhM30AgUtjbegdo\ndnpp6PZQ3+OlzeWjocdLQ4+Xf3zms2lmHbmpRvLTzBTYTYf/gDCRYtSOfGBxbkSL9+rVq9m9ezeK\nolBWVsb06dMjr7W1tXH33Xfj9/uZMmUKv/jFL0ayKUIIIUaAoigkG7QkG7QUp1uOed3nD9LkHCre\n9d0e6rs9NB5+3NU/SFf/ILuaXcM+YzfryE/7tJgX2Ie2Fr2cbx4xYv8TO3bsoKGhgY0bN1JbW0tZ\nWRkbN26MvL527Vpuu+02lixZws9//nNaW1vJysoaqeYIIYQYAwatmqJ0C0WfKezhcBi/TstHVR0c\n7OrnYJeHWkc/dV0eHP2DOPoH2dHoHPaZzCQ9BYeLeoF9qKjn2Uzn5dXwI1a8t27dyuLFiwEoKCjA\n5XLhdruxWCyEQiF27drFo48+CsCqVatGqhlCCCFikKIoZFuN6PJszMuzRfaHwmFaXT4Odnk42NVP\nrcNDbVc/Dd2eyDj7P+t7Pj0OkJ1iiJyd56eZybebmJRqQq9J3AvlRqx4OxwOpk6dGnlus9no7OzE\nYrHQ3d2N2WxmzZo1VFRUMHv2bH784x+f8HipqSY0mlP76+p465/GK4kntkk8sU3iiW3R4snMSGZm\n0fB9gWCIxm4PVe19VLW7OdDeR1V7H3WO/shtbu/XdkXer1Jgos1EYUYSRRmWoa/MJArSzZh0I9f9\nPlr5GbUBhHA4POxxe3s7t956K9nZ2XznO9/hvffe4/LLLz/u53t6jr2KMRpZrD62STyxTeKJbed7\nPEnAhZkWLsy0AOMA8AdDNPR4Oejop/Zw93tdl4dmp5f6Lg/1XR7ermwfdpysZD0TUo0oKARCIfzB\n8NBXKEQgGMYfDBEIhfGHwgSCIfQaFTaTjlSTFptJh82kHfbYZtKRatRSPMmGs7v/HP4PHf+PgREr\n3hkZGTgcjsjzjo4O0tOHbgxITU0lKyuLiRMnAjB37lyqq6tPWLyFEEKIz9KqVRTazRTah9/mNhgI\n0ej0Hr6FrZ+6bg8Hu4YulmvtHaC1d+C0/p0O9+BJ32Mz6/ifm2YwPjn6lfnn0ogV73nz5rF+/XqW\nLVtGRUUFGRkZWCxDFyxoNBomTJhAfX09ubm5VFRUsHTp0pFqihBCiPOMTnN0Uf/0jvJAMESzy0eL\n04eigFatoFWr0KoUNGrV0HOVCo1aiezz+oN0e/x0ewbpObztPmp7ZJ9Rq0Y1SjPNjVjxnjVrFlOn\nTmXZsmUoisKqVat49dVXSUpKYsmSJZSVlbFixQrC4TDFxcUsWrRopJoihBBCAKBRq8i1mci1mU75\nM1aj9pTOpkdzWGNEx7zvueeeYc9LSkoijydNmsQLL7wwkv+8EEIIkZAS9zp6IYQQIkFJ8RZCCCHi\njBRvIYQQIs5I8RZCCCHijBRvIYQQIs5I8RZCCCHijBRvIYQQIs5I8RZCCCHijBRvIYQQIs5I8RZC\nCCHijBRvIYQQIs5I8RZCCCHijBIOh8Nj3QghhBBCnDo58xZCCCHijBRvIYQQIs5I8RZCCCHijBRv\nIYQQIs5I8RZCCCHijBRvIYQQIs5oxroB59Lq1avZvXs3iqJQVlbG9OnTx7pJZ2z79u388Ic/pKio\nCIDi4mJWrlw5xq06fVVVVSxfvpxvfvOb3HLLLbS1tXHfffcRDAZJT0/n4YcfRqfTjXUzT9ln41mx\nYgUVFRVYrVYAbr/9di6//PKxbeRpeOihh9i1axeBQIDvfve7XHDBBXGdn8/Gs3nz5rjNj9frZcWK\nFXR1dTEwMMDy5cspKSmJ2/xEi+fNN9+M2/wc4fP5uOaaa1i+fDlz584dtfwkTPHesWMHDQ0NbNy4\nkdraWsrKyti4ceNYN+uszJkzh8cff3ysm3HGPB4Pv/zlL5k7d25k3+OPP87NN9/MVVddxaOPPsrL\nL7/MzTffPIatPHXR4gG4++67Wbhw4Ri16sxt27aN6upqNm7cSE9PD9dddx1z586N2/xEi+fiiy+O\n2/y8++67TJs2jTvuuIOWlhZuu+02Zs2aFbf5iRbPzJkz4zY/R/z+978nJSUFGN3fbwnTbb5161YW\nL14MQEFBAS6XC7fbPcatOr/pdDqeeuopMjIyIvu2b9/OF77wBQAWLlzI1q1bx6p5py1aPPHsoosu\n4je/+Q0AycnJeL3euM5PtHiCweAYt+rMXX311dxxxx0AtLW1kZmZGdf5iRZPvKutraWmpibSWzCa\n+UmY4u1wOEhNTY08t9lsdHZ2jmGLzl5NTQ3f+973uOmmm/jwww/HujmnTaPRYDAYhu3zer2RbqS0\ntLS4ylG0eACeffZZbr31Vn70ox/R3d09Bi07M2q1GpPJBMDLL7/MZZddFtf5iRaPWq2O2/wcsWzZ\nMu655x7KysriOj9HHB0PxO/PD8C6detYsWJF5Plo5idhus0/K95nfc3NzeXOO+/kqquuoqmpiVtv\nvZVNmzbFzfjWqYj3HAFce+21WK1WSktLefLJJ/ntb3/Lgw8+ONbNOi1vv/02L7/8Ms888wxXXHFF\nZH+85ufoeMrLy+M+Py+++CKVlZXce++9w3ISr/k5Op6ysrK4zc+f//xnZsyYwYQJE6K+PtL5SZgz\n74yMDBwOR+R5R0cH6enpY9iis5OZmcnVV1+NoihMnDgRu91Oe3v7WDfrrJlMJnw+HwDt7e1x3wU9\nd+5cSktLAVi0aBFVVVVj3KLTs2XLFv7whz/w1FNPkZSUFPf5+Ww88Zyf8vJy2traACgtLSUYDGI2\nm+M2P9HiKS4ujtv8vPfee7zzzjvceOON/PGPf+R3v/vdqP78JEzxnjdvHm+++SYAFRUVZGRkYLFY\nxrhVZ+61117jpYsuCwAAA7lJREFU6aefBqCzs5Ourq6EGCO65JJLInnatGkT8+fPH+MWnZ277rqL\npqYmYGi868jdAfGgr6+Phx56iCeeeCJytW885ydaPPGcn507d/LMM88AQ8OCHo8nrvMTLZ4HH3ww\nbvPz2GOP8corr/DSSy9xww03sHz58lHNT0KtKvbII4+wc+dOFEVh1apVlJSUjHWTzpjb7eaee+6h\nt7cXv9/PnXfeyYIFC8a6WaelvLycdevW0dLSgkajITMzk0ceeYQVK1YwMDBAVlYWa9asQavVjnVT\nT0m0eG655RaefPJJjEYjJpOJNWvWkJaWNtZNPSUbN25k/fr15OXlRfatXbuWn/70p3GZn2jxfOUr\nX+HZZ5+Ny/z4fD4eeOAB2tra8Pl83HnnnUybNo2f/OQncZmfaPGYTCYefvjhuMzP0davX092djaX\nXnrpqOUnoYq3EEIIcT5ImG5zIYQQ4nwhxVsIIYSIM1K8hRBCiDgjxVsIIYSIM1K8hRBCiDiTsDOs\nCSGGa25u5sorr2TmzJnD9i9YsIBvf/vbZ3387du389hjj/HCCy+c9bGEECcmxVuI84jNZmPDhg1j\n3QwhxFmS4i2EYMqUKSxfvpzt27fT39/P2rVrKS4uZvfu3axduxaNRoOiKDz44IMUFhZSX1/PypUr\nCYVC6PV61qxZA0AoFGLVqlVUVlai0+l44oknMJvNYxydEIlHxryFEASDQYqKitiwYQM33XRTZB35\n++67j/vvv58NGzbwrW99i5///OcArFq1ittvv53nnnuO66+/nr///e/A0BKJd911Fy+99BIajYYP\nPvhgzGISIpHJmbcQ55Hu7m6+/vWvD9t37733AnDppZcCMGvWLJ5++ml6e3vp6upi+vTpAMyZM4e7\n774bgD179jBnzhwAli5dCgyNeefn52O32wEYN24cvb29Ix+UEOchKd5CnEdONOZ99EzJiqKgKMpx\nX4ehLvLPUqvV56CVQoiTkW5zIQQA27ZtA2DXrl1MnjyZpKQk0tPT2b17NwBbt25lxowZwNDZ+ZYt\nWwD429/+xqOPPjo2jRbiPCVn3kKcR6J1m+fk5ACwb98+XnjhBVwuF+vWrQNg3bp1rF27FrVajUql\n4mc/+xkAK1euZOXKlTz//PNoNBpWr15NY2PjqMYixPlMVhUTQjB58mQqKirQaOTveSHigXSbCyGE\nEHFGzryFEEKIOCNn3kIIIUSckeIthBBCxBkp3kIIIUSckeIthBBCxBkp3kIIIUSckeIthBBCxJn/\nD9J55JgYqUORAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7f7560d25e80>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "text",
        "deletable": true,
        "editable": true,
        "id": "O90WcUTwt2GU"
      },
      "cell_type": "markdown",
      "source": [
        "### 5. Test the network on the test data\n",
        "\n",
        "\n",
        "We have trained the network for 2 passes over the training dataset.\n",
        "But we need to check if the network has learnt anything at all.\n",
        "\n",
        "We will check this by predicting the class label that the neural network\n",
        "outputs, and checking it against the ground-truth. If the prediction is\n",
        "correct, we add the sample to the list of correct predictions.\n",
        "\n",
        "Okay, first step. Let us display an image from the test set to get familiar.\n",
        "\n"
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "deletable": true,
        "editable": true,
        "id": "V4vljwBlt2GX",
        "outputId": "87b82e2b-6e22-4773-8fda-e1b701d22a16",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 614
        }
      },
      "cell_type": "code",
      "source": [
        "try:\n",
        "  images, labels = next(iter(test_loader))\n",
        "except EOFError:\n",
        "  pass\n",
        "\n",
        "# print images\n",
        "imshow(torchvision.utils.make_grid(images))\n",
        "print(\"Ground truth:\\n\")\n",
        "\n",
        "print(' '.join('{:>10}'.format(classes[labels[j]]) for j in range(test_batch_size)))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Ground truth:\n",
            "\n",
            "       cat      plane        car       frog\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Exception ignored in: <bound method _DataLoaderIter.__del__ of <torch.utils.data.dataloader._DataLoaderIter object at 0x7f35237ecb38>>\n",
            "Traceback (most recent call last):\n",
            "  File \"/usr/local/lib/python3.6/dist-packages/torch/utils/data/dataloader.py\", line 399, in __del__\n",
            "    self._shutdown_workers()\n",
            "  File \"/usr/local/lib/python3.6/dist-packages/torch/utils/data/dataloader.py\", line 378, in _shutdown_workers\n",
            "    self.worker_result_queue.get()\n",
            "  File \"/usr/lib/python3.6/multiprocessing/queues.py\", line 337, in get\n",
            "    return _ForkingPickler.loads(res)\n",
            "  File \"/usr/local/lib/python3.6/dist-packages/torch/multiprocessing/reductions.py\", line 151, in rebuild_storage_fd\n",
            "    fd = df.detach()\n",
            "  File \"/usr/lib/python3.6/multiprocessing/resource_sharer.py\", line 57, in detach\n",
            "    with _resource_sharer.get_connection(self._id) as conn:\n",
            "  File \"/usr/lib/python3.6/multiprocessing/resource_sharer.py\", line 87, in get_connection\n",
            "    c = Client(address, authkey=process.current_process().authkey)\n",
            "  File \"/usr/lib/python3.6/multiprocessing/connection.py\", line 493, in Client\n",
            "    answer_challenge(c, authkey)\n",
            "  File \"/usr/lib/python3.6/multiprocessing/connection.py\", line 732, in answer_challenge\n",
            "    message = connection.recv_bytes(256)         # reject large message\n",
            "  File \"/usr/lib/python3.6/multiprocessing/connection.py\", line 216, in recv_bytes\n",
            "    buf = self._recv_bytes(maxlength)\n",
            "  File \"/usr/lib/python3.6/multiprocessing/connection.py\", line 407, in _recv_bytes\n",
            "    buf = self._recv(4)\n",
            "  File \"/usr/lib/python3.6/multiprocessing/connection.py\", line 383, in _recv\n",
            "    raise EOFError\n",
            "EOFError: \n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAW8AAAB5CAYAAAD2xKAmAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYV/eFhYWFj2I/x9OHiUJCXw95wAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7f3525737908>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "text",
        "deletable": true,
        "editable": true,
        "id": "KpmaQT4Zt2Gn"
      },
      "cell_type": "markdown",
      "source": [
        "Okay, now let us see what the neural network thinks these examples above are:\n",
        "\n"
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "deletable": true,
        "editable": true,
        "id": "utIfocFrt2Gs",
        "outputId": "ef1da9a3-a271-449e-cd29-685046905f59",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "cell_type": "code",
      "source": [
        "outputs = net(images.to(device))\n",
        "print(outputs.size())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "torch.Size([4, 10])\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "text",
        "deletable": true,
        "editable": true,
        "id": "6mU42O0Gt2G2"
      },
      "cell_type": "markdown",
      "source": [
        "The outputs are energies for the 10 classes.\n",
        "The higher the energy for a class, the more the network\n",
        "thinks that the image is from that particular class.\n",
        "So, let's get the index of the highest energy:\n",
        "\n"
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "deletable": true,
        "editable": true,
        "id": "IWTWHHs9t2G5",
        "outputId": "8ccbbafc-69ac-49af-8e07-d728bbeef53c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 189
        }
      },
      "cell_type": "code",
      "source": [
        "_, predicted = torch.max(outputs, 1)\n",
        "\n",
        "print(\"Predicted:\\n\")\n",
        "imshow(torchvision.utils.make_grid(images))\n",
        "\n",
        "print(' '.join('{:>10}'.format(classes[predicted[j]]) for j in range(test_batch_size)))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Predicted:\n",
            "\n",
            "      ship      plane        car       frog\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAW8AAAB5CAYAAAD2xKAmAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYV/eFhYWFj2I/x9OHiUJCXw95wAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7f3523803a20>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "text",
        "deletable": true,
        "editable": true,
        "id": "AUpCEAOTt2HK"
      },
      "cell_type": "markdown",
      "source": [
        "The results seem pretty good.\n",
        "\n",
        "Let us look at how the network performs on the whole test set.\n",
        "\n"
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "deletable": true,
        "editable": true,
        "id": "LI6JtYwTt2HM",
        "outputId": "9ee120b6-ec71-473c-c0e9-ddc0ae2fb9d6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "cell_type": "code",
      "source": [
        "def dataset_accuracy(net, data_loader, name=\"\"):\n",
        "    net = net.to(device)\n",
        "    correct = 0\n",
        "    total = 0\n",
        "    for images, labels in data_loader:\n",
        "        images, labels = images.to(device), labels.to(device)\n",
        "        outputs = net(images)\n",
        "        _, predicted = torch.max(outputs, 1)\n",
        "        total += labels.size(0)\n",
        "        correct += (predicted == labels).sum()\n",
        "    accuracy = 100 * float(correct) / total\n",
        "    print('Accuracy of the network on the {} {} images: {:.2f} %'.format(total, name, accuracy))\n",
        "\n",
        "def train_set_accuracy(net):\n",
        "    dataset_accuracy(net, train_loader, \"train\")\n",
        "\n",
        "def val_set_accuracy(net):\n",
        "    dataset_accuracy(net, val_loader, \"validation\")  \n",
        "    \n",
        "def test_set_accuracy(net):\n",
        "    dataset_accuracy(net, test_loader, \"test\")\n",
        "\n",
        "def compute_accuracy(net):\n",
        "    train_set_accuracy(net)\n",
        "    val_set_accuracy(net)\n",
        "    test_set_accuracy(net)\n",
        "    \n",
        "print(\"Computing accuracy...\")\n",
        "# compute_accuracy(net)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Computing accuracy...\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "text",
        "deletable": true,
        "editable": true,
        "id": "iGGyra-4t2HW"
      },
      "cell_type": "markdown",
      "source": [
        "That initial 59.78 % on the test set of images looks waaay better than chance, which is 10% accuracy (randomly picking\n",
        "a class out of 10 classes).\n",
        "Seems like the network learnt something.\n",
        "As a baseline, a linear model achieves around 30% accuracy.\n",
        "\n",
        "What are the classes that performed well, and the classes that did not perform well?\n",
        "\n"
      ]
    },
    {
      "metadata": {
        "colab_type": "code",
        "deletable": true,
        "editable": true,
        "id": "rkim9_INt2HY",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "def accuracy_per_class(net):\n",
        "    net = net.to(device)\n",
        "    n_classes = 10\n",
        "    # (real, predicted)\n",
        "    confusion_matrix = np.zeros((n_classes, n_classes), dtype=np.int64)\n",
        "\n",
        "    for images, labels in test_loader:\n",
        "        images, labels = images, labels = images.to(device), labels.to(device)\n",
        "        outputs = net(images)\n",
        "        _, predicted = torch.max(outputs.data, 1)\n",
        "        for i in range(test_batch_size):\n",
        "            confusion_matrix[labels[i], predicted[i]] += 1\n",
        "            label = labels[i]\n",
        "\n",
        "    print(\"{:<10} {:^10}\".format(\"Class\", \"Accuracy (%)\"))\n",
        "    for i in range(n_classes):\n",
        "        class_total = confusion_matrix[i, :].sum()\n",
        "        class_correct = confusion_matrix[i, i]\n",
        "        percentage_correct = 100.0 * float(class_correct) / class_total\n",
        "        \n",
        "        print('{:<10} {:^10.2f}'.format(classes[i], percentage_correct))\n",
        "    return confusion_matrix\n",
        "\n",
        "# confusion_matrix = accuracy_per_class(net)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "AZKLymOacazg",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "### Confusion Matrix"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "ekJHz3vpcazg",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "Let's look at what type of error our networks makes... \n",
        "It seems that our network is pretty good at classifying ships,\n",
        "but has some difficulties to differentiate cats and dogs.\n",
        "Also, it classifies a lot of trucks as cars."
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "1aYMqD1Ocazi",
        "colab_type": "code",
        "outputId": "99035d9b-49d4-4cb8-82e3-ba65ee26eb6a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1153
        }
      },
      "cell_type": "code",
      "source": [
        "# Plot normalized confusion matrix\n",
        "plot_confusion_matrix(confusion_matrix, classes, normalize=True,\n",
        "                      title='Normalized confusion matrix')\n",
        "\n",
        "# Plot non-normalized confusion matrix\n",
        "plot_confusion_matrix(confusion_matrix, classes,\n",
        "                      title='Confusion matrix, without normalization')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAikAAAI4CAYAAACvGXkgAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\npZRSfun/Aa0KgnXDidTHAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7f35236d3240>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAiwAAAI4CAYAAABJMLJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nCYuIiIjknhIWERERyb3/Bzv1ogp+IWuaAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7f35236d3d30>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "colab_type": "text",
        "deletable": true,
        "editable": true,
        "id": "MVv-mV8Pt2Hs"
      },
      "cell_type": "markdown",
      "source": [
        "# III. Exploring CNN Architectures\n",
        "\n",
        "Now, it is your turn to build a Convolutional Neural Network. The goal of this section is to explore different CNN architectures and set hyperparameters in order to obtain the best accuracy on the **test** set!\n",
        "\n",
        "The network that you have to tweak is called **MyConvolutionalNetwork**.\n",
        "\n",
        "You can start changing the batch_size, number of epochs and then try adding more convolutional layers."
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "h1blK9eicazo",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "### PyTorch functions to build the network\n",
        "- **nn.Conv2d(in_channels, out_channels, kernel_size, stride=1, padding=0):** Convolution layer. You can read the full documentation [here](http://pytorch.org/docs/master/nn.html#conv2d)\n",
        "\n",
        "- **nn.MaxPool2d(kernel_size, stride=None, padding=0):** Max pooling layer. You can read the full documentation [here](http://pytorch.org/docs/master/nn.html#maxpool2d)\n",
        "\n",
        "- **F.relu(Z1):** computes the element-wise ReLU of Z1 (which can be of any shape). You can read the full documentation [here.](http://pytorch.org/docs/master/nn.html#torch.nn.ReLU)\n",
        "\n",
        "- **x.view(new_shape)**: Returns a new tensor with the same data but different size. It is the equivalent of numpy function *reshape* (Gives a new shape to an array without changing its data.). You can read the full documentation [here.](http://pytorch.org/docs/master/tensors.html#torch.Tensor.view)\n",
        "\n",
        "- **nn.Linear(in_features, out_features):** Applies a linear transformation to the incoming data: $y = Ax + b$, it is also called a fully connected (fc) layer. You can read the full documentation [here.](http://pytorch.org/docs/master/nn.html#linear-layers)"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "a8-lKBaacazp",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "**Convolution Formulas**:\n",
        "\n",
        "The formulas relating the output shape $(C_2, H_2, W_2)$ of the convolution to the input shape $(C_1, H_1, W_1)$ are:\n",
        "\n",
        "\n",
        "$$ H_2 = \\lfloor \\frac{H_1 - kernel\\_size + 2 \\times padding}{stride} \\rfloor +1 $$\n",
        "\n",
        "$$ W_2 = \\lfloor \\frac{W_1 - kernel\\_size + 2 \\times padding}{stride} \\rfloor +1 $$\n",
        "\n",
        "$$ C_2 = \\text{number of filters used in the convolution}$$\n",
        "\n",
        "NOTE: $C_2 = C_1$ in the case of max pooling\n",
        "\n",
        "where:\n",
        "- $H_2$: height of the output volume  \n",
        "- $W_2$: width of the output volume  \n",
        "- $C_1$: in_channels, number of channels in the input volume\n",
        "- $C_2$: out_channels"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "acppf3nkcazr",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "def get_output_size(in_size, kernel_size, stride=1, padding=0):\n",
        "    \"\"\"\n",
        "    Get the output size given all the parameters of the convolution\n",
        "    :param in_size: (int) input size\n",
        "    :param kernel_size: (int)\n",
        "    :param stride: (int)\n",
        "    :param paddind: (int)\n",
        "    :return: (int)\n",
        "    \"\"\"\n",
        "    return int((in_size - kernel_size + 2 * padding) / stride) + 1"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "SEsbZoTOcazu",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "#### Example of use of helper method get_output_size() \n",
        "\n",
        "Let's assume you have an *input volume of size 3x32x32* (where 3 is the number of channels)\n",
        "and you use a 2D convolution with the following parameters:\n",
        "\n",
        "```python\n",
        "conv1 = nn.Conv2d(3, 18, kernel_size=7, stride=2, padding=1)\n",
        "```\n",
        "then, the size of the output volume is 18x?x? (because we have 18 filters) where ? is given by the convolution formulas (see above).\n",
        "\n",
        "**get_output_size()** function allows to compute that size:\n",
        "\n",
        "```\n",
        "out_size = get_output_size(in_size=32, kernel_size=7, stride=2, padding=1)\n",
        "print(out_size) # prints 14\n",
        "```\n",
        "\n",
        "That is to say, *the output volume is 18x14x14*"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "2JFQ1wgKcazv",
        "colab_type": "code",
        "outputId": "96571347-87e3-4085-df50-844a59c773bc",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "cell_type": "code",
      "source": [
        "out_size = get_output_size(in_size=32, kernel_size=3, stride=1, padding=1)\n",
        "print(out_size)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "32\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "wviV5iQIcazz",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "Below is the neural network you have to edit:"
      ]
    },
    {
      "metadata": {
        "id": "mlrHF_atnwiP",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "class MyConvolutionalNetwork_basic(nn.Module):\n",
        "    def __init__(self):\n",
        "        super(MyConvolutionalNetwork_basic, self).__init__()\n",
        "        \n",
        "        # 32 * 32 * 3 -> 32 * 32 * 32\n",
        "        self.conv1 = nn.Conv2d(3, 32, kernel_size=3, stride=1, padding=1)\n",
        "        #self.bn1 = nn.BatchNorm2d(32)\n",
        "        self.conv2 = nn.Conv2d(32, 32, kernel_size=3, stride=1, padding=1)\n",
        "        #self.bn2 = nn.BatchNorm2d(32)\n",
        "        \n",
        "        self.maxpool1 = nn.MaxPool2d(kernel_size=2, stride=2, padding=0)\n",
        "        #self.drop1 = nn.Dropout(0.2)\n",
        "        \n",
        "        self.conv3 = nn.Conv2d(32, 64, kernel_size=3, stride=1, padding=1)\n",
        "        #self.bn3 = nn.BatchNorm2d(64)\n",
        "        self.conv4 = nn.Conv2d(64, 64, kernel_size=3, stride=1, padding=1)\n",
        "        #self.bn4 = nn.BatchNorm2d(64)\n",
        "        \n",
        "        self.maxpool2 = nn.MaxPool2d(kernel_size=2, stride=2, padding=0)\n",
        "        #self.drop2 = nn.Dropout(0.3)\n",
        "        \n",
        "        self.conv5 = nn.Conv2d(64, 128, kernel_size=3, stride=1, padding=1)\n",
        "        #self.bn5 = nn.BatchNorm2d(128)\n",
        "        self.conv6 = nn.Conv2d(128, 128, kernel_size=3, stride=1, padding=1)\n",
        "        #self.bn6 = nn.BatchNorm2d(128)\n",
        "        \n",
        "        self.maxpool3 = nn.MaxPool2d(kernel_size=2, stride=2, padding=0)\n",
        "        #self.drop3 = nn.Dropout(0.4)\n",
        "        \n",
        "        # Size of the output of the last convolution:\n",
        "        self.flattened_size = 4 * 4 * 128\n",
        "        \n",
        "        self.fc1 = nn.Linear(self.flattened_size, 10)\n",
        "        \n",
        "\n",
        "    def forward(self, x):\n",
        "        \"\"\"\n",
        "        Forward pass,\n",
        "        x shape is (batch_size, 3, 32, 32)\n",
        "        (color channel first)\n",
        "        in the comments, we omit the batch_size in the shape\n",
        "        \"\"\"\n",
        "        # shape : 3x32x32 -> 32x32x32\n",
        "        x = self.conv1(x)\n",
        "        x = F.relu(x)\n",
        "        #x = self.bn1(x)\n",
        "        x = self.conv2(x)\n",
        "        x = F.relu(x)\n",
        "        #x = self.bn2(x)\n",
        "        \n",
        "        x = self.maxpool1(x)\n",
        "        #x = self.drop1(x)\n",
        "        \n",
        "        x = self.conv3(x)\n",
        "        x = F.relu(x)\n",
        "        #x = self.bn3(x)\n",
        "        x = self.conv4(x)\n",
        "        x = F.relu(x)\n",
        "        #x = self.bn4(x)\n",
        "        \n",
        "        x = self.maxpool2(x)\n",
        "        #x = self.drop2(x)\n",
        "        \n",
        "        x = self.conv5(x)\n",
        "        x = F.relu(x)\n",
        "        #x = self.bn5(x)\n",
        "        x = self.conv6(x)\n",
        "        x = F.relu(x)\n",
        "        #x = self.bn6(x)\n",
        "        \n",
        "        x = self.maxpool3(x)\n",
        "        #x = self.drop3(x)\n",
        "        \n",
        "        # Check the output size\n",
        "        output_size = np.prod(x.size()[1:])\n",
        "        assert output_size == self.flattened_size,\\\n",
        "                \"self.flattened_size is invalid {} != {}\".format(output_size, self.flattened_size)\n",
        "        \n",
        "        # 128x4x4 -> 2048\n",
        "        x = x.view(-1, self.flattened_size)\n",
        "        # 2048 -> 10\n",
        "        x = F.leaky_relu(self.fc1(x), negative_slope=0.001)\n",
        "        \n",
        "        return x"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "fnKUPUDTcaz1",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "class MyConvolutionalNetwork(nn.Module):\n",
        "    def __init__(self):\n",
        "        super(MyConvolutionalNetwork, self).__init__()\n",
        "        \n",
        "        # 32 * 32 * 3 -> 32 * 32 * 32\n",
        "        self.conv1 = nn.Conv2d(3, 32, kernel_size=3, stride=1, padding=1)\n",
        "        self.bn1 = nn.BatchNorm2d(32)\n",
        "        self.conv2 = nn.Conv2d(32, 32, kernel_size=3, stride=1, padding=1)\n",
        "        self.bn2 = nn.BatchNorm2d(32)\n",
        "        \n",
        "        self.maxpool1 = nn.MaxPool2d(kernel_size=2, stride=2, padding=0)\n",
        "        self.drop1 = nn.Dropout(0.2)\n",
        "        \n",
        "        self.conv3 = nn.Conv2d(32, 64, kernel_size=3, stride=1, padding=1)\n",
        "        self.bn3 = nn.BatchNorm2d(64)\n",
        "        self.conv4 = nn.Conv2d(64, 64, kernel_size=3, stride=1, padding=1)\n",
        "        self.bn4 = nn.BatchNorm2d(64)\n",
        "        \n",
        "        self.maxpool2 = nn.MaxPool2d(kernel_size=2, stride=2, padding=0)\n",
        "        self.drop2 = nn.Dropout(0.3)\n",
        "        \n",
        "        self.conv5 = nn.Conv2d(64, 128, kernel_size=3, stride=1, padding=1)\n",
        "        self.bn5 = nn.BatchNorm2d(128)\n",
        "        self.conv6 = nn.Conv2d(128, 128, kernel_size=3, stride=1, padding=1)\n",
        "        self.bn6 = nn.BatchNorm2d(128)\n",
        "        \n",
        "        self.maxpool3 = nn.MaxPool2d(kernel_size=2, stride=2, padding=0)\n",
        "        self.drop3 = nn.Dropout(0.4)\n",
        "        \n",
        "        # Size of the output of the last convolution:\n",
        "        self.flattened_size = 4 * 4 * 128\n",
        "        \n",
        "        self.fc1 = nn.Linear(self.flattened_size, 10)\n",
        "        \n",
        "\n",
        "    def forward(self, x):\n",
        "        \"\"\"\n",
        "        Forward pass,\n",
        "        x shape is (batch_size, 3, 32, 32)\n",
        "        (color channel first)\n",
        "        in the comments, we omit the batch_size in the shape\n",
        "        \"\"\"\n",
        "        # shape : 3x32x32 -> 32x32x32\n",
        "        x = self.conv1(x)\n",
        "        x = F.leaky_relu(x, negative_slope=0.001)\n",
        "        x = self.bn1(x)\n",
        "        x = self.conv2(x)\n",
        "        x = F.leaky_relu(x, negative_slope=0.001)\n",
        "        x = self.bn2(x)\n",
        "        \n",
        "        x = self.maxpool1(x)\n",
        "        x = self.drop1(x)\n",
        "        \n",
        "        x = self.conv3(x)\n",
        "        x = F.leaky_relu(x, negative_slope=0.001)\n",
        "        x = self.bn3(x)\n",
        "        x = self.conv4(x)\n",
        "        x = F.leaky_relu(x, negative_slope=0.001)\n",
        "        x = self.bn4(x)\n",
        "        \n",
        "        x = self.maxpool2(x)\n",
        "        x = self.drop2(x)\n",
        "        \n",
        "        x = self.conv5(x)\n",
        "        x = F.leaky_relu(x, negative_slope=0.001)\n",
        "        x = self.bn5(x)\n",
        "        x = self.conv6(x)\n",
        "        x = F.leaky_relu(x, negative_slope=0.001)\n",
        "        x = self.bn6(x)\n",
        "        \n",
        "        x = self.maxpool3(x)\n",
        "        x = self.drop3(x)\n",
        "        \n",
        "        # Check the output size\n",
        "        output_size = np.prod(x.size()[1:])\n",
        "        assert output_size == self.flattened_size,\\\n",
        "                \"self.flattened_size is invalid {} != {}\".format(output_size, self.flattened_size)\n",
        "        \n",
        "        # 128x4x4 -> 2048\n",
        "        x = x.view(-1, self.flattened_size)\n",
        "        # 2048 -> 10\n",
        "        x = F.leaky_relu(self.fc1(x), negative_slope=0.001)\n",
        "        \n",
        "        return x"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "scrolled": false,
        "id": "ruLWyTSocaz5",
        "colab_type": "code",
        "outputId": "74e5d000-504e-4de9-8404-bbaded6bbd82",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1819
        }
      },
      "cell_type": "code",
      "source": [
        "#net = MyConvolutionalNetwork()\n",
        "#print(net)\n",
        "#model_parameters = filter(lambda p: p.requires_grad, net.parameters())\n",
        "#print(\"Parameters :\", sum([np.prod(p.size()) for p in model_parameters]))\n",
        "train_history, val_history = train(net, batch_size=32, n_epochs=10, learning_rate=0.0001)\n",
        "\n"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "===== HYPERPARAMETERS =====\n",
            "batch_size= 32\n",
            "n_epochs= 10\n",
            "learning_rate= 0.0001\n",
            "==============================\n",
            "Epoch 1, 10% \t train_loss: 0.36 took: 3.48s\n",
            "Epoch 1, 20% \t train_loss: 0.36 took: 3.65s\n",
            "Epoch 1, 30% \t train_loss: 0.38 took: 3.70s\n",
            "Epoch 1, 40% \t train_loss: 0.37 took: 3.71s\n",
            "Epoch 1, 50% \t train_loss: 0.35 took: 3.75s\n",
            "Epoch 1, 60% \t train_loss: 0.35 took: 3.79s\n",
            "Epoch 1, 70% \t train_loss: 0.38 took: 3.76s\n",
            "Epoch 1, 80% \t train_loss: 0.36 took: 3.76s\n",
            "Epoch 1, 90% \t train_loss: 0.37 took: 3.77s\n",
            "Validation loss = 3.58\n",
            "Epoch 2, 10% \t train_loss: 0.36 took: 4.01s\n",
            "Epoch 2, 20% \t train_loss: 0.39 took: 3.84s\n",
            "Epoch 2, 30% \t train_loss: 0.35 took: 3.79s\n",
            "Epoch 2, 40% \t train_loss: 0.36 took: 3.88s\n",
            "Epoch 2, 50% \t train_loss: 0.36 took: 3.84s\n",
            "Epoch 2, 60% \t train_loss: 0.39 took: 3.83s\n",
            "Epoch 2, 70% \t train_loss: 0.34 took: 3.72s\n",
            "Epoch 2, 80% \t train_loss: 0.33 took: 3.68s\n",
            "Epoch 2, 90% \t train_loss: 0.37 took: 3.73s\n",
            "Validation loss = 6.19\n",
            "Epoch 3, 10% \t train_loss: 0.36 took: 3.84s\n",
            "Epoch 3, 20% \t train_loss: 0.35 took: 3.71s\n",
            "Epoch 3, 30% \t train_loss: 0.36 took: 3.73s\n",
            "Epoch 3, 40% \t train_loss: 0.35 took: 3.76s\n",
            "Epoch 3, 50% \t train_loss: 0.36 took: 3.88s\n",
            "Epoch 3, 60% \t train_loss: 0.36 took: 3.79s\n",
            "Epoch 3, 70% \t train_loss: 0.36 took: 3.80s\n",
            "Epoch 3, 80% \t train_loss: 0.36 took: 3.81s\n",
            "Epoch 3, 90% \t train_loss: 0.36 took: 3.80s\n",
            "Validation loss = 2.96\n",
            "Epoch 4, 10% \t train_loss: 0.36 took: 3.91s\n",
            "Epoch 4, 20% \t train_loss: 0.34 took: 3.78s\n",
            "Epoch 4, 30% \t train_loss: 0.35 took: 3.79s\n",
            "Epoch 4, 40% \t train_loss: 0.36 took: 3.75s\n",
            "Epoch 4, 50% \t train_loss: 0.36 took: 3.77s\n",
            "Epoch 4, 60% \t train_loss: 0.38 took: 3.72s\n",
            "Epoch 4, 70% \t train_loss: 0.35 took: 3.74s\n",
            "Epoch 4, 80% \t train_loss: 0.36 took: 3.74s\n",
            "Epoch 4, 90% \t train_loss: 0.37 took: 3.78s\n",
            "Validation loss = 1.47\n",
            "Epoch 5, 10% \t train_loss: 0.37 took: 3.97s\n",
            "Epoch 5, 20% \t train_loss: 0.35 took: 3.76s\n",
            "Epoch 5, 30% \t train_loss: 0.35 took: 3.78s\n",
            "Epoch 5, 40% \t train_loss: 0.36 took: 3.76s\n",
            "Epoch 5, 50% \t train_loss: 0.36 took: 3.80s\n",
            "Epoch 5, 60% \t train_loss: 0.35 took: 3.78s\n",
            "Epoch 5, 70% \t train_loss: 0.37 took: 3.72s\n",
            "Epoch 5, 80% \t train_loss: 0.37 took: 3.73s\n",
            "Epoch 5, 90% \t train_loss: 0.36 took: 3.68s\n",
            "Validation loss = 4.20\n",
            "Epoch 6, 10% \t train_loss: 0.36 took: 3.86s\n",
            "Epoch 6, 20% \t train_loss: 0.35 took: 3.72s\n",
            "Epoch 6, 30% \t train_loss: 0.34 took: 3.74s\n",
            "Epoch 6, 40% \t train_loss: 0.36 took: 3.83s\n",
            "Epoch 6, 50% \t train_loss: 0.35 took: 3.79s\n",
            "Epoch 6, 60% \t train_loss: 0.35 took: 3.83s\n",
            "Epoch 6, 70% \t train_loss: 0.34 took: 3.80s\n",
            "Epoch 6, 80% \t train_loss: 0.36 took: 3.78s\n",
            "Epoch 6, 90% \t train_loss: 0.36 took: 3.85s\n",
            "Validation loss = 1.64\n",
            "Epoch 7, 10% \t train_loss: 0.37 took: 3.95s\n",
            "Epoch 7, 20% \t train_loss: 0.36 took: 3.71s\n",
            "Epoch 7, 30% \t train_loss: 0.36 took: 3.69s\n",
            "Epoch 7, 40% \t train_loss: 0.36 took: 3.72s\n",
            "Epoch 7, 50% \t train_loss: 0.36 took: 3.74s\n",
            "Epoch 7, 60% \t train_loss: 0.35 took: 3.80s\n",
            "Epoch 7, 70% \t train_loss: 0.34 took: 3.75s\n",
            "Epoch 7, 80% \t train_loss: 0.34 took: 3.72s\n",
            "Epoch 7, 90% \t train_loss: 0.38 took: 3.75s\n",
            "Validation loss = 4.41\n",
            "Epoch 8, 10% \t train_loss: 0.36 took: 3.93s\n",
            "Epoch 8, 20% \t train_loss: 0.36 took: 3.70s\n",
            "Epoch 8, 30% \t train_loss: 0.35 took: 3.76s\n",
            "Epoch 8, 40% \t train_loss: 0.37 took: 3.70s\n",
            "Epoch 8, 50% \t train_loss: 0.37 took: 3.71s\n",
            "Epoch 8, 60% \t train_loss: 0.34 took: 3.70s\n",
            "Epoch 8, 70% \t train_loss: 0.36 took: 3.76s\n",
            "Epoch 8, 80% \t train_loss: 0.36 took: 3.84s\n",
            "Epoch 8, 90% \t train_loss: 0.34 took: 3.79s\n",
            "Validation loss = 5.13\n",
            "Epoch 9, 10% \t train_loss: 0.34 took: 3.89s\n",
            "Epoch 9, 20% \t train_loss: 0.35 took: 3.71s\n",
            "Epoch 9, 30% \t train_loss: 0.34 took: 3.68s\n",
            "Epoch 9, 40% \t train_loss: 0.38 took: 3.71s\n",
            "Epoch 9, 50% \t train_loss: 0.36 took: 3.69s\n",
            "Epoch 9, 60% \t train_loss: 0.36 took: 3.75s\n",
            "Epoch 9, 70% \t train_loss: 0.34 took: 3.71s\n",
            "Epoch 9, 80% \t train_loss: 0.34 took: 3.67s\n",
            "Epoch 9, 90% \t train_loss: 0.37 took: 3.71s\n",
            "Validation loss = 2.01\n",
            "Epoch 10, 10% \t train_loss: 0.34 took: 3.88s\n",
            "Epoch 10, 20% \t train_loss: 0.35 took: 3.75s\n",
            "Epoch 10, 30% \t train_loss: 0.36 took: 3.70s\n",
            "Epoch 10, 40% \t train_loss: 0.36 took: 3.70s\n",
            "Epoch 10, 50% \t train_loss: 0.35 took: 3.68s\n",
            "Epoch 10, 60% \t train_loss: 0.35 took: 3.79s\n",
            "Epoch 10, 70% \t train_loss: 0.34 took: 3.72s\n",
            "Epoch 10, 80% \t train_loss: 0.36 took: 3.74s\n",
            "Epoch 10, 90% \t train_loss: 0.37 took: 3.83s\n",
            "Validation loss = 0.99\n",
            "Training Finished, took 376.14s\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "u7cgVbkDcaz9",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "### Losses Plot"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "XtXu67qbcaz-",
        "colab_type": "code",
        "outputId": "87fb5420-88e1-4e08-ddce-087579be1a51",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 403
        }
      },
      "cell_type": "code",
      "source": [
        "plot_losses(train_history, val_history)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAe8AAAGCCAYAAADJ40tJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nCNHN/H97OJOtWjQdpAAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x7fb360db19e8>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "TuGKgAMWcaz_",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "### Accuracy of the trained model"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "TWowqQhYca0B",
        "colab_type": "code",
        "outputId": "11f0ff00-bb9a-4419-fd28-9277d3f58a96",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 425
        }
      },
      "cell_type": "code",
      "source": [
        "compute_accuracy(net)\n",
        "accuracy_per_class(net)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Accuracy of the network on the 40000 train images: 81.31 %\n",
            "Accuracy of the network on the 1 validation images: 0.00 %\n",
            "Accuracy of the network on the 10000 test images: 78.06 %\n",
            "Class      Accuracy (%)\n",
            "plane        75.00   \n",
            "car          87.70   \n",
            "bird         69.00   \n",
            "cat          63.00   \n",
            "deer         73.60   \n",
            "dog          74.60   \n",
            "frog         73.40   \n",
            "horse        86.30   \n",
            "ship         87.10   \n",
            "truck        87.20   \n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[750,  15,  65,  15,  13,   8,   9,  14,  63,  48],\n",
              "       [ 11, 877,   5,  10,   1,   4,   3,   1,  24,  64],\n",
              "       [ 40,  10, 690,  52,  44,  64,  28,  39,  10,  23],\n",
              "       [ 16,  12,  37, 630,  36, 152,  27,  37,  19,  34],\n",
              "       [ 18,   3,  43,  39, 736,  45,  11,  78,   9,  18],\n",
              "       [  8,   8,  34, 108,  13, 746,  17,  46,   6,  14],\n",
              "       [  9,  18,  35,  60,  34,  38, 734,  22,  14,  36],\n",
              "       [ 16,   5,  15,  20,  31,  30,   1, 863,   2,  17],\n",
              "       [ 31,  24,  16,   9,   1,   3,   9,   6, 871,  30],\n",
              "       [ 19,  51,   3,   5,   3,   3,   2,  12,  30, 872]])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 31
        }
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "st9f_4opca0F",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "**Baseline: Simple Convolutional Neural Network (form part II)**\n",
        "\n",
        "<table>\n",
        "<tr>\n",
        "    <td>Accuracy on the test set:</td>\n",
        "    <td>59.98 %</td>\n",
        "</tr>\n",
        "</table>"
      ]
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "OKvmb4p-ca0I",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "confusion_matrix = accuracy_per_class(net)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "deletable": true,
        "editable": true,
        "id": "ih5Pj0WBca0L",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "plot_confusion_matrix(confusion_matrix, classes,\n",
        "                      title='Confusion matrix, without normalization')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "colab_type": "text",
        "deletable": true,
        "editable": true,
        "id": "7xzIkZqit2IA"
      },
      "cell_type": "markdown",
      "source": [
        "### Going further\n",
        "\n",
        "- [Coursera Course on CNN](https://www.coursera.org/learn/convolutional-neural-networks)\n",
        "- [Standford Course](http://cs231n.stanford.edu/syllabus.html)\n",
        "- [PyTorch Tutorial](http://pytorch.org/tutorials/beginner/blitz/cifar10_tutorial.html)\n",
        "- [How backpropagation works](neuralnetworksanddeeplearning.com/chap2.html) (Michael )\n",
        "\n",
        "If you feel like this was too easy peasy:\n",
        "\n",
        "-Investigate further [optimization methods](https://pytorch.org/docs/stable/optim.html) beyond SGD, and Adam and their parameters.\n",
        "\n",
        "-Look at ways to improve your network using regularization techniques\n",
        "\n",
        "-Look at ways to visualize network activations for model interpretability\n",
        "\n",
        "-Use transfer learning, in order to use torchvision with pretrained=True with some pretrained models\n",
        "\n",
        "\n",
        "Note: This tutorial is based on the [original PyTorch tutorial](https://pytorch.org/tutorials/beginner/blitz/cifar10_tutorial.html) and was adapted by [Antonin Raffin](http://araffin.github.io/) for the ROB313 course at ENSTA ParisTech. Thanks to Clement Pinard for feedback!"
      ]
    },
    {
      "metadata": {
        "id": "XEFW6M6jZtWk",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "### More documentation/ questions to explore about Google Colab: \n",
        "\n",
        "-How to connect your Google Drive with Google Colab?\n",
        "\n",
        "-How to import a new notebook and save it to your GDrive?\n",
        "\n",
        "-How to use files which are contained in your GDrive?\n",
        "\n",
        "Some tips [here](https://medium.com/deep-learning-turkey/google-colab-free-gpu-tutorial-e113627b9f5d)\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "## Extras to read later\n",
        "### Visualizing Convolution parameters:\n",
        "[A guide to convolution arithmetic for deep learning](https://github.com/vdumoulin/conv_arithmetic/blob/master/README.md) \n",
        "by Vincent Dumoulin, Francesco Visin \n",
        "\n",
        "\n",
        "### Documentation of autograd and Function: \n",
        "[Autograd](http://pytorch.org/docs/autograd)\n"
      ]
    }
  ]
}