tensorflow_mnist_fashion_data.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "tensorflow_mnist_fashion_data.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "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/gb08/Deep_learning_with_Colab/blob/master/tensorflow_mnist_fashion_data.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gzLOFhO3w8t3",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# !pip install -U -q PyDrive\n",
        "from google.colab import drive"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZLJLBuVonkKn",
        "colab_type": "code",
        "outputId": "a419e388-c5b0-4ecf-c246-3b45a4d8c3f9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 63
        }
      },
      "source": [
        "import matplotlib as mpl\n",
        "import matplotlib.pyplot as plt\n",
        "import numpy as np\n",
        "import os\n",
        "import pandas as pd\n",
        "import sklearn\n",
        "import sys\n",
        "import tensorflow as tf\n",
        "from tensorflow import keras  # tf.keras\n",
        "import time"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<p style=\"color: red;\">\n",
              "The default version of TensorFlow in Colab will soon switch to TensorFlow 2.x.<br>\n",
              "We recommend you <a href=\"https://www.tensorflow.org/guide/migrate\" target=\"_blank\">upgrade</a> now \n",
              "or ensure your notebook will continue to use TensorFlow 1.x via the <code>%tensorflow_version 1.x</code> magic:\n",
              "<a href=\"https://colab.research.google.com/notebooks/tensorflow_version.ipynb\" target=\"_blank\">more info</a>.</p>\n"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Yy5Hv24nxT_I",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "drive.mount('/content/gdrive')\n",
        "#!ls gdrive/My\\ Drive\n",
        "\n",
        "model_path = '/content/gdrive/My Drive/tensorflow_mnist_model/'"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "us1Vv77TxR26",
        "colab_type": "text"
      },
      "source": [
        ""
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "P8lceGo-no8F",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "\n",
        "%matplotlib inline\n",
        "%load_ext tensorboard"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-LDQNlaTR1v4",
        "colab_type": "text"
      },
      "source": [
        "Check Python and keras version"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "x8aEW2choEqT",
        "colab_type": "code",
        "outputId": "20615ab1-b820-4ad4-82fc-b7421575ee77",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 153
        }
      },
      "source": [
        "print(\"python\", sys.version)\n",
        "for module in mpl, np, pd, sklearn, tf, keras:\n",
        "    print(module.__name__, module.__version__)"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "python 3.6.8 (default, Oct  7 2019, 12:59:55) \n",
            "[GCC 8.3.0]\n",
            "matplotlib 3.1.1\n",
            "numpy 1.17.4\n",
            "pandas 0.25.3\n",
            "sklearn 0.21.3\n",
            "tensorflow 1.15.0\n",
            "tensorflow.python.keras.api._v1.keras 2.2.4-tf\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Chi1B-MmSJto",
        "colab_type": "text"
      },
      "source": [
        "Load mnist fashion dataset"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cZv2C5JKoJjb",
        "colab_type": "code",
        "outputId": "36ddca85-2a55-4218-9b1f-aa7353c928fa",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 153
        }
      },
      "source": [
        "fashion_mnist = keras.datasets.fashion_mnist\n",
        "(X_train_full, y_train_full), (X_test, y_test) = (\n",
        "    fashion_mnist.load_data())\n"
      ],
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/train-labels-idx1-ubyte.gz\n",
            "32768/29515 [=================================] - 0s 0us/step\n",
            "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/train-images-idx3-ubyte.gz\n",
            "26427392/26421880 [==============================] - 0s 0us/step\n",
            "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/t10k-labels-idx1-ubyte.gz\n",
            "8192/5148 [===============================================] - 0s 0us/step\n",
            "Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/t10k-images-idx3-ubyte.gz\n",
            "4423680/4422102 [==============================] - 0s 0us/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "z8MKnRMLSgrY",
        "colab_type": "text"
      },
      "source": [
        "Devide dataset into trai and test set"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YQXxhMTduTuE",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "X_valid, X_train = X_train_full[:5000], X_train_full[5000:]\n",
        "y_valid, y_train = y_train_full[:5000], y_train_full[5000:]"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-gf_S-Rjuqqt",
        "colab_type": "code",
        "outputId": "57133e2f-51f1-4546-f58d-e5fa1d99ae8d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "\n",
        "X_train.shape"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(55000, 28, 28)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 7
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VrKt9MS7SyGz",
        "colab_type": "text"
      },
      "source": [
        "Plot first image from training dataset"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bfRYfYbiwKuq",
        "colab_type": "code",
        "outputId": "57cab89e-8537-485b-8df1-fa6b5d562a23",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        }
      },
      "source": [
        "plt.imshow(X_train[0], cmap=\"binary\")\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPsAAAD4CAYAAAAq5pAIAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Y-zY1zlLS68B",
        "colab_type": "text"
      },
      "source": [
        "Define Class names."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3EI2xMVDwL09",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "class_names = [\"T-shirt/top\", \"Trouser\", \"Pullover\", \"Dress\", \"Coat\",\n",
        "               \"Sandal\", \"Shirt\", \"Sneaker\", \"Bag\", \"Ankle boot\"]"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PwiVe4UaxBuu",
        "colab_type": "code",
        "outputId": "ad8ca638-3bb7-4328-a593-a4003cda1549",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "class_names[y_train[0]]"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'Coat'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 11
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QppXMviATShr",
        "colab_type": "text"
      },
      "source": [
        "Glimpse of dataset images"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lbVNCv4ZxIXF",
        "colab_type": "code",
        "outputId": "1c237bfa-3a84-4185-a743-720425056980",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 231
        }
      },
      "source": [
        "n_rows = 5\n",
        "n_cols = 10\n",
        "plt.figure(figsize=(n_cols*1.4, n_rows * 1.6))\n",
        "for row in range(n_rows):\n",
        "    for col in range(n_cols):\n",
        "        index = n_cols * row + col\n",
        "        plt.subplot(n_rows, n_cols, index + 1)\n",
        "        plt.imshow(X_train[index], interpolation=\"nearest\")\n",
        "        plt.axis('off')\n",
        "        plt.title(class_names[y_train[index]])\n",
        "plt.show()"
      ],
      "execution_count": 1,
      "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-1-0c6549eb8b11>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0mn_rows\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;36m5\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      2\u001b[0m \u001b[0mn_cols\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;36m10\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 3\u001b[0;31m \u001b[0mplt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfigure\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfigsize\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mn_cols\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0;36m1.4\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mn_rows\u001b[0m \u001b[0;34m*\u001b[0m \u001b[0;36m1.6\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      4\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mrow\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mrange\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mn_rows\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[1;32m      5\u001b[0m     \u001b[0;32mfor\u001b[0m \u001b[0mcol\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mrange\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mn_cols\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[0;31mNameError\u001b[0m: name 'plt' is not defined"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "z_lPxgA3TdTv",
        "colab_type": "text"
      },
      "source": [
        "**Intialize Keras Model**\n",
        "**Add flatten input layer**\n",
        "**Add First hidden layer with 300 neurons, activation fnc ReLu**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8am4KJ3Ix-mD",
        "colab_type": "code",
        "outputId": "77bc654e-f57b-4d96-9601-e72fdb1986f2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 88
        }
      },
      "source": [
        "model = keras.models.Sequential()\n",
        "model.add(keras.layers.Flatten(input_shape=[28,28]))\n",
        "model.add(keras.layers.Dense(300, activation=\"relu\"))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow_core/python/ops/resource_variable_ops.py:1630: calling BaseResourceVariable.__init__ (from tensorflow.python.ops.resource_variable_ops) with constraint is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "If using Keras pass *_constraint arguments to layers.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mh7WBdqnURiz",
        "colab_type": "text"
      },
      "source": [
        "**Add Output Layer with 10 neurons(one for each class) with activation fnc Softmax**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wPSs7GClzfXw",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model.add(keras.layers.Dense(10, activation=\"softmax\"))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "_utaHyyhUlXz",
        "colab_type": "text"
      },
      "source": [
        "Check Model Summary"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dEO2xLtEzs5q",
        "colab_type": "code",
        "outputId": "082ee9ed-fdf4-4558-bf2f-dcbba2314f0a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 255
        }
      },
      "source": [
        "model.summary()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "flatten (Flatten)            (None, 784)               0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 300)               235500    \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 10)                3010      \n",
            "=================================================================\n",
            "Total params: 238,510\n",
            "Trainable params: 238,510\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "NQDqNoOSUn2f",
        "colab_type": "text"
      },
      "source": [
        "Plot Model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4StjyOMHz02n",
        "colab_type": "code",
        "outputId": "0e55b83c-90d5-4e3b-a4ba-da2fa01a5ebf",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 369
        }
      },
      "source": [
        "keras.utils.plot_model(model)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAOEAAAFgCAIAAADCSonkAAAABmJLR0QA/wD/AP+gvaeTAAAdUElE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\n",
            "text/plain": [
              "<IPython.core.display.Image object>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 16
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Pm5uGtcYUtOx",
        "colab_type": "text"
      },
      "source": [
        "Compile Model, Set loss fnc avaible in keras, optimizer(SGD with learning rate 1e-3) and metrics as accuracy "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AMFlI8YS0FyT",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model.compile(loss=\"sparse_categorical_crossentropy\", optimizer=keras.optimizers.SGD(lr=1e-3), metrics=[\"accuracy\"])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JE2z2hMJVISe",
        "colab_type": "text"
      },
      "source": [
        "Fit Model, Set values for Number of epochs also provide validation set, model will return history object containing statistics"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BviMurQr2dSr",
        "colab_type": "code",
        "outputId": "2bf27b35-dc3a-40bc-c735-ed86c28238c5",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 374
        }
      },
      "source": [
        "history = model.fit(epochs=10, x=X_train, y=y_train, validation_data=(X_valid, y_valid))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 55000 samples, validate on 5000 samples\n",
            "Epoch 1/10\n",
            "55000/55000 [==============================] - 6s 108us/sample - loss: 3.5390 - acc: 0.6738 - val_loss: 1.0418 - val_acc: 0.7180\n",
            "Epoch 2/10\n",
            "55000/55000 [==============================] - 5s 95us/sample - loss: 0.6742 - acc: 0.7529 - val_loss: 0.6732 - val_acc: 0.7610\n",
            "Epoch 3/10\n",
            "55000/55000 [==============================] - 5s 95us/sample - loss: 0.6032 - acc: 0.7763 - val_loss: 0.6798 - val_acc: 0.7864\n",
            "Epoch 4/10\n",
            "55000/55000 [==============================] - 5s 98us/sample - loss: 0.5652 - acc: 0.7910 - val_loss: 0.6236 - val_acc: 0.7696\n",
            "Epoch 5/10\n",
            "55000/55000 [==============================] - 5s 97us/sample - loss: 0.5397 - acc: 0.7967 - val_loss: 0.5722 - val_acc: 0.8074\n",
            "Epoch 6/10\n",
            "55000/55000 [==============================] - 5s 94us/sample - loss: 0.5214 - acc: 0.8027 - val_loss: 0.6041 - val_acc: 0.7780\n",
            "Epoch 7/10\n",
            "55000/55000 [==============================] - 5s 94us/sample - loss: 0.5025 - acc: 0.8097 - val_loss: 0.5772 - val_acc: 0.8150\n",
            "Epoch 8/10\n",
            "55000/55000 [==============================] - 5s 95us/sample - loss: 0.4853 - acc: 0.8151 - val_loss: 0.5498 - val_acc: 0.8142\n",
            "Epoch 9/10\n",
            "55000/55000 [==============================] - 5s 94us/sample - loss: 0.4758 - acc: 0.8188 - val_loss: 0.5308 - val_acc: 0.8202\n",
            "Epoch 10/10\n",
            "55000/55000 [==============================] - 5s 93us/sample - loss: 0.4537 - acc: 0.8332 - val_loss: 0.5277 - val_acc: 0.8366\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cizvSGLUVdvR",
        "colab_type": "text"
      },
      "source": [
        "Plot learning Curves"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GALrWZhC3NSi",
        "colab_type": "code",
        "outputId": "c2e2d563-75ae-4fc3-f303-36e36f76480f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        }
      },
      "source": [
        "pd.DataFrame(history.history).plot()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7fc267f135f8>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 19
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD4CAYAAAAXUaZHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nMVXIkrsxxlQhS+7GGFOFLLkbY0wV+v8B/DrhkDH9JsUAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UW2FlNqpCNVc",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kLXCbzZO3iHh",
        "colab_type": "code",
        "outputId": "de90fd8f-27d2-4b8f-8d04-281d0878a6eb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 324
        }
      },
      "source": [
        "plt.figure(figsize=(8,5)); plt.grid(True); plt.gca().set_ylim(0, 1); plt.plot(pd.DataFrame(history.history));"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAeMAAAEzCAYAAAACSWsXAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngg==\n",
            "text/plain": [
              "<Figure size 576x360 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tIrRvAlRC2VB",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "Train model for few more epochs, it will start where left last time"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1G98aP1OC8Kx",
        "colab_type": "code",
        "outputId": "2f72b50d-5a7d-4e5c-d2a7-132b84d55491",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 374
        }
      },
      "source": [
        "history = model.fit(epochs=10, x=X_train, y=y_train, validation_data=(X_valid, y_valid))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 55000 samples, validate on 5000 samples\n",
            "Epoch 1/10\n",
            "55000/55000 [==============================] - 5s 96us/sample - loss: 0.4315 - acc: 0.8443 - val_loss: 0.5093 - val_acc: 0.8416\n",
            "Epoch 2/10\n",
            "55000/55000 [==============================] - 5s 93us/sample - loss: 0.4200 - acc: 0.8495 - val_loss: 0.5028 - val_acc: 0.8436\n",
            "Epoch 3/10\n",
            "55000/55000 [==============================] - 5s 94us/sample - loss: 0.4101 - acc: 0.8517 - val_loss: 0.5287 - val_acc: 0.8384\n",
            "Epoch 4/10\n",
            "55000/55000 [==============================] - 5s 94us/sample - loss: 0.4036 - acc: 0.8540 - val_loss: 0.4881 - val_acc: 0.8484\n",
            "Epoch 5/10\n",
            "55000/55000 [==============================] - 5s 95us/sample - loss: 0.3961 - acc: 0.8579 - val_loss: 0.5172 - val_acc: 0.8456\n",
            "Epoch 6/10\n",
            "55000/55000 [==============================] - 5s 93us/sample - loss: 0.3885 - acc: 0.8598 - val_loss: 0.5809 - val_acc: 0.8380\n",
            "Epoch 7/10\n",
            "55000/55000 [==============================] - 5s 93us/sample - loss: 0.3806 - acc: 0.8614 - val_loss: 0.5993 - val_acc: 0.8228\n",
            "Epoch 8/10\n",
            "55000/55000 [==============================] - 5s 94us/sample - loss: 0.3785 - acc: 0.8625 - val_loss: 0.5117 - val_acc: 0.8422\n",
            "Epoch 9/10\n",
            "55000/55000 [==============================] - 5s 94us/sample - loss: 0.3736 - acc: 0.8644 - val_loss: 0.5287 - val_acc: 0.8364\n",
            "Epoch 10/10\n",
            "55000/55000 [==============================] - 5s 95us/sample - loss: 0.3686 - acc: 0.8655 - val_loss: 0.5701 - val_acc: 0.8166\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lpZmUYU1WSuS",
        "colab_type": "text"
      },
      "source": [
        "Evaluate model on test set"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wTPddtIeEzcP",
        "colab_type": "code",
        "outputId": "57ed2c05-f530-4542-f3bc-300c6ca28452",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "evall = model.evaluate(x=X_test, y=y_test)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "10000/10000 [==============================] - 1s 66us/sample - loss: 0.6139 - acc: 0.8083\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "W0fotWbKFwn6",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "X_new = X_test[0:10]"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9sUqZUyzWY1E",
        "colab_type": "text"
      },
      "source": [
        "Predict values for few of the test samples"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "G_EWLA3pF-zm",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "predicted_out = model.predict(X_new)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tNGYWo3vGEYG",
        "colab_type": "code",
        "outputId": "01fcd4be-401c-4eaa-8f54-292e604ac41b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 163
        }
      },
      "source": [
        "predicted_out.round(2)"
      ],
      "execution_count": 2,
      "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-2-4fa4b942935e>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mpredicted_out\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mround\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m2\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
            "\u001b[0;31mNameError\u001b[0m: name 'predicted_out' is not defined"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "FOHkoft0WjK3",
        "colab_type": "text"
      },
      "source": [
        "Get class Ids of most likely class"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "51Ztwn0OGGgz",
        "colab_type": "code",
        "outputId": "572124c1-28e3-4cc9-d3dd-858e276cae71",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "np.argmax(predicted_out, axis=1)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([9, 2, 1, 1, 6, 1, 4, 4, 5, 7])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 26
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Zait-eBoWuNz",
        "colab_type": "text"
      },
      "source": [
        "Use predict class method directly"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4LnauQ1BGRbh",
        "colab_type": "code",
        "outputId": "ff32fc40-4254-4cf4-c0cc-ee43596ceff9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "model.predict_classes(X_new)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([9, 2, 1, 1, 6, 1, 4, 4, 5, 7])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 27
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vZSa7zQDWzDb",
        "colab_type": "text"
      },
      "source": [
        "Get Probabilities of predicted class"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ejJj2jQUIUFH",
        "colab_type": "code",
        "outputId": "2c9008e2-fd41-4b2b-9cc4-008b024b3bc1",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "source": [
        "predicted_out.max(axis=1).round(2)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([0.98, 0.66, 1.  , 1.  , 0.65, 1.  , 0.95, 0.55, 1.  , 1.  ],\n",
              "      dtype=float32)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 28
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WpMvZa3PW7C4",
        "colab_type": "text"
      },
      "source": [
        "Get Top three predicted classes for image"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LhDYmsQhJpjg",
        "colab_type": "code",
        "outputId": "da946605-7dcd-453a-b6e2-9e5f9d6c45ee",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 187
        }
      },
      "source": [
        "\n",
        "k = 3\n",
        "top_k = np.argsort(-predicted_out, axis=1)[:, :k]\n",
        "top_k"
      ],
      "execution_count": 46,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[9, 7, 5],\n",
              "       [2, 4, 6],\n",
              "       [1, 3, 0],\n",
              "       [1, 3, 0],\n",
              "       [6, 4, 2],\n",
              "       [1, 3, 0],\n",
              "       [4, 6, 2],\n",
              "       [4, 6, 2],\n",
              "       [5, 8, 7],\n",
              "       [7, 5, 9]])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 46
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BLRd9REIXGtO",
        "colab_type": "text"
      },
      "source": [
        "Get probabilities of top K classes"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XQE16JyfJ5ta",
        "colab_type": "code",
        "outputId": "cfc8c3c7-3d83-4360-80a5-32addc407ee8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 187
        }
      },
      "source": [
        "row_indices = np.tile(np.arange(len(top_k)), [k, 1]).T\n",
        "predicted_out[row_indices, top_k].round(2)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[0.98, 0.02, 0.  ],\n",
              "       [0.66, 0.27, 0.06],\n",
              "       [1.  , 0.  , 0.  ],\n",
              "       [1.  , 0.  , 0.  ],\n",
              "       [0.65, 0.31, 0.03],\n",
              "       [1.  , 0.  , 0.  ],\n",
              "       [0.95, 0.04, 0.01],\n",
              "       [0.55, 0.44, 0.01],\n",
              "       [1.  , 0.  , 0.  ],\n",
              "       [1.  , 0.  , 0.  ]], dtype=float32)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 31
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7WHqy5PiXNr8",
        "colab_type": "text"
      },
      "source": [
        "**Now, Try with Standerdized dataset**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RD6qN3tWKFHv",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "scaler = sklearn.preprocessing.StandardScaler()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "S4p5ogtiXXH9",
        "colab_type": "text"
      },
      "source": [
        "Fit Scalar object on Training data and transform test and validation set"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OHjiVlGislr1",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "X_train_scaled = scaler.fit_transform(X_train.astype(np.float32).reshape(-1, 28 * 28)).reshape(-1, 28, 28)\n",
        "X_valid_scaled = scaler.transform(X_valid.astype(np.float32).reshape(-1, 28 * 28)).reshape(-1, 28, 28)\n",
        "X_test_scaled = scaler.transform(X_test.astype(np.float32).reshape(-1, 28 * 28)).reshape(-1, 28, 28)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "K9DXrxavXiRl",
        "colab_type": "text"
      },
      "source": [
        "Build Model, With one more hidden Layer"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yciVgLl3t4Oh",
        "colab_type": "code",
        "outputId": "da716eef-7d18-466f-f870-aecddf935d27",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 374
        }
      },
      "source": [
        "model = keras.models.Sequential([\n",
        "    keras.layers.Flatten(input_shape=[28, 28]),\n",
        "    keras.layers.Dense(300, activation=\"relu\"),\n",
        "    keras.layers.Dense(100, activation=\"relu\"),\n",
        "    keras.layers.Dense(10, activation=\"softmax\")\n",
        "])\n",
        "model.compile(loss=\"sparse_categorical_crossentropy\",\n",
        "              optimizer=keras.optimizers.SGD(1e-3), metrics=[\"accuracy\"])\n",
        "history = model.fit(X_train_scaled, y_train, epochs=10,\n",
        "                    validation_data=(X_valid_scaled, y_valid))"
      ],
      "execution_count": 26,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 55000 samples, validate on 5000 samples\n",
            "Epoch 1/10\n",
            "55000/55000 [==============================] - 6s 113us/sample - loss: 0.9298 - acc: 0.6977 - val_loss: 0.6256 - val_acc: 0.7910\n",
            "Epoch 2/10\n",
            "55000/55000 [==============================] - 5s 98us/sample - loss: 0.5787 - acc: 0.7975 - val_loss: 0.5233 - val_acc: 0.8238\n",
            "Epoch 3/10\n",
            "55000/55000 [==============================] - 5s 98us/sample - loss: 0.5068 - acc: 0.8203 - val_loss: 0.4782 - val_acc: 0.8364\n",
            "Epoch 4/10\n",
            "55000/55000 [==============================] - 5s 98us/sample - loss: 0.4676 - acc: 0.8339 - val_loss: 0.4526 - val_acc: 0.8466\n",
            "Epoch 5/10\n",
            "55000/55000 [==============================] - 5s 98us/sample - loss: 0.4415 - acc: 0.8441 - val_loss: 0.4331 - val_acc: 0.8504\n",
            "Epoch 6/10\n",
            "55000/55000 [==============================] - 5s 99us/sample - loss: 0.4223 - acc: 0.8508 - val_loss: 0.4187 - val_acc: 0.8578\n",
            "Epoch 7/10\n",
            "55000/55000 [==============================] - 5s 98us/sample - loss: 0.4072 - acc: 0.8562 - val_loss: 0.4070 - val_acc: 0.8590\n",
            "Epoch 8/10\n",
            "55000/55000 [==============================] - 5s 99us/sample - loss: 0.3943 - acc: 0.8610 - val_loss: 0.3986 - val_acc: 0.8622\n",
            "Epoch 9/10\n",
            "55000/55000 [==============================] - 5s 98us/sample - loss: 0.3835 - acc: 0.8654 - val_loss: 0.3906 - val_acc: 0.8640\n",
            "Epoch 10/10\n",
            "55000/55000 [==============================] - 5s 96us/sample - loss: 0.3741 - acc: 0.8682 - val_loss: 0.3846 - val_acc: 0.8666\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hRBpQVRgu-e1",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def plot_learning_curves(history):\n",
        "    pd.DataFrame(history.history).plot(figsize=(8, 5))\n",
        "    plt.grid(True)\n",
        "    plt.gca().set_ylim(0, 1)\n",
        "    plt.show()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vfYxlY5oXuxU",
        "colab_type": "text"
      },
      "source": [
        "Plot Learning Curves for this model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2LKD-ijZvM8P",
        "colab_type": "code",
        "outputId": "02e47878-a92d-478e-beb3-b21e32182b96",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 324
        }
      },
      "source": [
        "plot_learning_curves(history)"
      ],
      "execution_count": 28,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAeYAAAEzCAYAAADkYKBTAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngg==\n",
            "text/plain": [
              "<Figure size 576x360 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "btKh8ThCvPMR",
        "colab_type": "code",
        "outputId": "7066e6b6-d756-441f-9a40-db28dfcfda67",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "# model_path"
      ],
      "execution_count": 29,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'/content/gdrive/My Drive/tensorflow_model_1/'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 29
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9BJyCx-6X4zz",
        "colab_type": "text"
      },
      "source": [
        "Train model with Callback. Provide Google drive path for log directory and model checkpoints, save best model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lI5xsvwnvwIU",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "40a81545-fa64-4309-bd16-f3018c5cd1d0"
      },
      "source": [
        "model = keras.models.Sequential([\n",
        "    keras.layers.Flatten(input_shape=[28, 28]),\n",
        "    keras.layers.Dense(300, activation=\"relu\"),\n",
        "    keras.layers.Dense(100, activation=\"relu\"),\n",
        "    keras.layers.Dense(10, activation=\"softmax\")\n",
        "])\n",
        "model.compile(loss=\"sparse_categorical_crossentropy\",\n",
        "              optimizer=keras.optimizers.SGD(1e-3), metrics=[\"accuracy\"])\n",
        "\n",
        "\n",
        "logdir = os.path.join(model_path, \"run_{}\".format(time.time()))\n",
        "\n",
        "callbacks = [\n",
        "    keras.callbacks.TensorBoard(logdir),\n",
        "    keras.callbacks.EarlyStopping(patience=5),\n",
        "    keras.callbacks.ModelCheckpoint(\"my_mnist_model.h5\", save_best_only=True),\n",
        "]\n",
        "history_c = model.fit(X_train_scaled, y_train, epochs=50,\n",
        "                    validation_data=(X_valid_scaled, y_valid),\n",
        "                    callbacks=callbacks)"
      ],
      "execution_count": 30,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 55000 samples, validate on 5000 samples\n",
            "Epoch 1/50\n",
            "WARNING:tensorflow:Method (on_train_batch_end) is slow compared to the batch update (0.197673). Check your callbacks.\n",
            "   32/55000 [..............................] - ETA: 7:08 - loss: 2.3254 - acc: 0.1875WARNING:tensorflow:Method (on_train_batch_end) is slow compared to the batch update (0.107156). Check your callbacks.\n",
            "55000/55000 [==============================] - 6s 113us/sample - loss: 0.9698 - acc: 0.6931 - val_loss: 0.6395 - val_acc: 0.7880\n",
            "Epoch 2/50\n",
            "55000/55000 [==============================] - 6s 105us/sample - loss: 0.5867 - acc: 0.8007 - val_loss: 0.5384 - val_acc: 0.8168\n",
            "Epoch 3/50\n",
            "55000/55000 [==============================] - 6s 109us/sample - loss: 0.5131 - acc: 0.8200 - val_loss: 0.4913 - val_acc: 0.8332\n",
            "Epoch 4/50\n",
            "55000/55000 [==============================] - 6s 104us/sample - loss: 0.4733 - acc: 0.8314 - val_loss: 0.4632 - val_acc: 0.8414\n",
            "Epoch 5/50\n",
            "55000/55000 [==============================] - 6s 104us/sample - loss: 0.4467 - acc: 0.8416 - val_loss: 0.4444 - val_acc: 0.8472\n",
            "Epoch 6/50\n",
            "55000/55000 [==============================] - 6s 105us/sample - loss: 0.4274 - acc: 0.8477 - val_loss: 0.4300 - val_acc: 0.8530\n",
            "Epoch 7/50\n",
            "55000/55000 [==============================] - 6s 102us/sample - loss: 0.4120 - acc: 0.8537 - val_loss: 0.4188 - val_acc: 0.8558\n",
            "Epoch 8/50\n",
            "55000/55000 [==============================] - 6s 107us/sample - loss: 0.3996 - acc: 0.8579 - val_loss: 0.4098 - val_acc: 0.8606\n",
            "Epoch 9/50\n",
            "55000/55000 [==============================] - 6s 106us/sample - loss: 0.3887 - acc: 0.8615 - val_loss: 0.4023 - val_acc: 0.8622\n",
            "Epoch 10/50\n",
            "55000/55000 [==============================] - 6s 101us/sample - loss: 0.3793 - acc: 0.8642 - val_loss: 0.3967 - val_acc: 0.8628\n",
            "Epoch 11/50\n",
            "55000/55000 [==============================] - 6s 103us/sample - loss: 0.3711 - acc: 0.8672 - val_loss: 0.3908 - val_acc: 0.8652\n",
            "Epoch 12/50\n",
            "55000/55000 [==============================] - 6s 104us/sample - loss: 0.3638 - acc: 0.8701 - val_loss: 0.3881 - val_acc: 0.8678\n",
            "Epoch 13/50\n",
            "55000/55000 [==============================] - 6s 109us/sample - loss: 0.3572 - acc: 0.8727 - val_loss: 0.3830 - val_acc: 0.8692\n",
            "Epoch 14/50\n",
            "55000/55000 [==============================] - 6s 104us/sample - loss: 0.3509 - acc: 0.8750 - val_loss: 0.3793 - val_acc: 0.8672\n",
            "Epoch 15/50\n",
            "55000/55000 [==============================] - 6s 113us/sample - loss: 0.3454 - acc: 0.8778 - val_loss: 0.3758 - val_acc: 0.8708\n",
            "Epoch 16/50\n",
            "55000/55000 [==============================] - 6s 106us/sample - loss: 0.3400 - acc: 0.8790 - val_loss: 0.3720 - val_acc: 0.8706\n",
            "Epoch 17/50\n",
            "55000/55000 [==============================] - 6s 104us/sample - loss: 0.3350 - acc: 0.8807 - val_loss: 0.3693 - val_acc: 0.8720\n",
            "Epoch 18/50\n",
            "55000/55000 [==============================] - 6s 105us/sample - loss: 0.3303 - acc: 0.8826 - val_loss: 0.3678 - val_acc: 0.8722\n",
            "Epoch 19/50\n",
            "55000/55000 [==============================] - 6s 105us/sample - loss: 0.3259 - acc: 0.8839 - val_loss: 0.3643 - val_acc: 0.8712\n",
            "Epoch 20/50\n",
            "55000/55000 [==============================] - 6s 103us/sample - loss: 0.3218 - acc: 0.8859 - val_loss: 0.3631 - val_acc: 0.8738\n",
            "Epoch 21/50\n",
            "55000/55000 [==============================] - 6s 108us/sample - loss: 0.3178 - acc: 0.8875 - val_loss: 0.3619 - val_acc: 0.8750\n",
            "Epoch 22/50\n",
            "55000/55000 [==============================] - 6s 107us/sample - loss: 0.3139 - acc: 0.8887 - val_loss: 0.3586 - val_acc: 0.8762\n",
            "Epoch 23/50\n",
            "55000/55000 [==============================] - 6s 105us/sample - loss: 0.3104 - acc: 0.8905 - val_loss: 0.3575 - val_acc: 0.8768\n",
            "Epoch 24/50\n",
            "55000/55000 [==============================] - 6s 101us/sample - loss: 0.3068 - acc: 0.8916 - val_loss: 0.3551 - val_acc: 0.8772\n",
            "Epoch 25/50\n",
            "55000/55000 [==============================] - 6s 103us/sample - loss: 0.3034 - acc: 0.8931 - val_loss: 0.3541 - val_acc: 0.8764\n",
            "Epoch 26/50\n",
            "55000/55000 [==============================] - 6s 102us/sample - loss: 0.3003 - acc: 0.8941 - val_loss: 0.3522 - val_acc: 0.8780\n",
            "Epoch 27/50\n",
            "55000/55000 [==============================] - 6s 102us/sample - loss: 0.2969 - acc: 0.8949 - val_loss: 0.3527 - val_acc: 0.8802\n",
            "Epoch 28/50\n",
            "55000/55000 [==============================] - 6s 102us/sample - loss: 0.2938 - acc: 0.8967 - val_loss: 0.3510 - val_acc: 0.8786\n",
            "Epoch 29/50\n",
            "55000/55000 [==============================] - 6s 105us/sample - loss: 0.2910 - acc: 0.8979 - val_loss: 0.3501 - val_acc: 0.8806\n",
            "Epoch 30/50\n",
            "55000/55000 [==============================] - 6s 103us/sample - loss: 0.2880 - acc: 0.8983 - val_loss: 0.3473 - val_acc: 0.8804\n",
            "Epoch 31/50\n",
            "55000/55000 [==============================] - 6s 103us/sample - loss: 0.2853 - acc: 0.9000 - val_loss: 0.3464 - val_acc: 0.8816\n",
            "Epoch 32/50\n",
            "55000/55000 [==============================] - 6s 107us/sample - loss: 0.2823 - acc: 0.9009 - val_loss: 0.3463 - val_acc: 0.8786\n",
            "Epoch 33/50\n",
            "55000/55000 [==============================] - 6s 103us/sample - loss: 0.2796 - acc: 0.9020 - val_loss: 0.3450 - val_acc: 0.8808\n",
            "Epoch 34/50\n",
            "55000/55000 [==============================] - 6s 104us/sample - loss: 0.2771 - acc: 0.9025 - val_loss: 0.3432 - val_acc: 0.8808\n",
            "Epoch 35/50\n",
            "55000/55000 [==============================] - 6s 107us/sample - loss: 0.2745 - acc: 0.9036 - val_loss: 0.3424 - val_acc: 0.8814\n",
            "Epoch 36/50\n",
            "55000/55000 [==============================] - 6s 103us/sample - loss: 0.2719 - acc: 0.9047 - val_loss: 0.3419 - val_acc: 0.8810\n",
            "Epoch 37/50\n",
            "55000/55000 [==============================] - 6s 102us/sample - loss: 0.2695 - acc: 0.9057 - val_loss: 0.3409 - val_acc: 0.8832\n",
            "Epoch 38/50\n",
            "55000/55000 [==============================] - 6s 107us/sample - loss: 0.2671 - acc: 0.9065 - val_loss: 0.3409 - val_acc: 0.8828\n",
            "Epoch 39/50\n",
            "55000/55000 [==============================] - 6s 103us/sample - loss: 0.2648 - acc: 0.9066 - val_loss: 0.3395 - val_acc: 0.8822\n",
            "Epoch 40/50\n",
            "55000/55000 [==============================] - 6s 104us/sample - loss: 0.2624 - acc: 0.9078 - val_loss: 0.3378 - val_acc: 0.8838\n",
            "Epoch 41/50\n",
            "55000/55000 [==============================] - 6s 109us/sample - loss: 0.2600 - acc: 0.9085 - val_loss: 0.3373 - val_acc: 0.8852\n",
            "Epoch 42/50\n",
            "55000/55000 [==============================] - 6s 106us/sample - loss: 0.2580 - acc: 0.9099 - val_loss: 0.3375 - val_acc: 0.8844\n",
            "Epoch 43/50\n",
            "55000/55000 [==============================] - 6s 105us/sample - loss: 0.2557 - acc: 0.9093 - val_loss: 0.3375 - val_acc: 0.8854\n",
            "Epoch 44/50\n",
            "55000/55000 [==============================] - 6s 107us/sample - loss: 0.2535 - acc: 0.9108 - val_loss: 0.3379 - val_acc: 0.8850\n",
            "Epoch 45/50\n",
            "55000/55000 [==============================] - 6s 105us/sample - loss: 0.2513 - acc: 0.9116 - val_loss: 0.3352 - val_acc: 0.8858\n",
            "Epoch 46/50\n",
            "55000/55000 [==============================] - 6s 104us/sample - loss: 0.2493 - acc: 0.9121 - val_loss: 0.3346 - val_acc: 0.8872\n",
            "Epoch 47/50\n",
            "55000/55000 [==============================] - 6s 104us/sample - loss: 0.2472 - acc: 0.9128 - val_loss: 0.3351 - val_acc: 0.8864\n",
            "Epoch 48/50\n",
            "55000/55000 [==============================] - 6s 106us/sample - loss: 0.2452 - acc: 0.9138 - val_loss: 0.3333 - val_acc: 0.8858\n",
            "Epoch 49/50\n",
            "55000/55000 [==============================] - 6s 103us/sample - loss: 0.2432 - acc: 0.9147 - val_loss: 0.3334 - val_acc: 0.8856\n",
            "Epoch 50/50\n",
            "55000/55000 [==============================] - 6s 108us/sample - loss: 0.2408 - acc: 0.9154 - val_loss: 0.3329 - val_acc: 0.8866\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jzS9Uy7ElKEH",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 17
        },
        "outputId": "1ebb6449-f0e4-482e-86c4-9ef9ec4e0867"
      },
      "source": [
        "\n",
        "%tensorboard --logdir=./my_logs"
      ],
      "execution_count": 31,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "\n",
              "    <div id=\"root\"></div>\n",
              "    <script>\n",
              "      (function() {\n",
              "        window.TENSORBOARD_ENV = window.TENSORBOARD_ENV || {};\n",
              "        window.TENSORBOARD_ENV[\"IN_COLAB\"] = true;\n",
              "        document.querySelector(\"base\").href = \"https://localhost:6006\";\n",
              "        function fixUpTensorboard(root) {\n",
              "          const tftb = root.querySelector(\"tf-tensorboard\");\n",
              "          // Disable the fragment manipulation behavior in Colab. Not\n",
              "          // only is the behavior not useful (as the iframe's location\n",
              "          // is not visible to the user), it causes TensorBoard's usage\n",
              "          // of `window.replace` to navigate away from the page and to\n",
              "          // the `localhost:<port>` URL specified by the base URI, which\n",
              "          // in turn causes the frame to (likely) crash.\n",
              "          tftb.removeAttribute(\"use-hash\");\n",
              "        }\n",
              "        function executeAllScripts(root) {\n",
              "          // When `script` elements are inserted into the DOM by\n",
              "          // assigning to an element's `innerHTML`, the scripts are not\n",
              "          // executed. Thus, we manually re-insert these scripts so that\n",
              "          // TensorBoard can initialize itself.\n",
              "          for (const script of root.querySelectorAll(\"script\")) {\n",
              "            const newScript = document.createElement(\"script\");\n",
              "            newScript.type = script.type;\n",
              "            newScript.textContent = script.textContent;\n",
              "            root.appendChild(newScript);\n",
              "            script.remove();\n",
              "          }\n",
              "        }\n",
              "        function setHeight(root, height) {\n",
              "          // We set the height dynamically after the TensorBoard UI has\n",
              "          // been initialized. This avoids an intermediate state in\n",
              "          // which the container plus the UI become taller than the\n",
              "          // final width and cause the Colab output frame to be\n",
              "          // permanently resized, eventually leading to an empty\n",
              "          // vertical gap below the TensorBoard UI. It's not clear\n",
              "          // exactly what causes this problematic intermediate state,\n",
              "          // but setting the height late seems to fix it.\n",
              "          root.style.height = `${height}px`;\n",
              "        }\n",
              "        const root = document.getElementById(\"root\");\n",
              "        fetch(\".\")\n",
              "          .then((x) => x.text())\n",
              "          .then((html) => void (root.innerHTML = html))\n",
              "          .then(() => fixUpTensorboard(root))\n",
              "          .then(() => executeAllScripts(root))\n",
              "          .then(() => setHeight(root, 800));\n",
              "      })();\n",
              "    </script>\n",
              "  "
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QjnDZmnYcMIC",
        "colab_type": "text"
      },
      "source": [
        "Load Saved model from google drive"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6X23PnwvmuEF",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 139
        },
        "outputId": "0335cfeb-f7ca-42e2-a7a5-e4b4c755fcb3"
      },
      "source": [
        "model_c =  keras.models.load_model(\"my_mnist_model.h5\")"
      ],
      "execution_count": 32,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow_core/python/ops/init_ops.py:97: calling GlorotUniform.__init__ (from tensorflow.python.ops.init_ops) with dtype is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Call initializer instance with the dtype argument instead of passing it to the constructor\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow_core/python/ops/init_ops.py:97: calling Zeros.__init__ (from tensorflow.python.ops.init_ops) with dtype is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Call initializer instance with the dtype argument instead of passing it to the constructor\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Zp8G-8QCcSMN",
        "colab_type": "text"
      },
      "source": [
        "Evaluate model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nTbYcAbUnNAu",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model_c.evaluate(X_valid_scaled, y_valid)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ze_uUto6cxPG",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}