Basic_CNN_Implementation.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "collapsed_sections": [
        "E5kpkIntHAX4"
      ],
      "authorship_tag": "ABX9TyM7wZmhts9fWHJSZqp8mhFt",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU",
    "gpuClass": "standard"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/SamaSamrin/Basic-CNN-Implementation/blob/main/Basic_CNN_Implementation.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "id": "tEoE6GBFWPNe"
      },
      "outputs": [],
      "source": [
        "import tensorflow as tf\n",
        "from tensorflow.keras import datasets, layers, models, utils\n",
        "\n",
        "import matplotlib.pyplot as plt\n",
        "import numpy as np"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# **Importing & Exploring the dataset**\n",
        "\n",
        "The CIFAR10 dataset contains has **100 classes** containing **600 images each**. There are **500 training images and 100 testing images per class**. The 100 classes in the CIFAR-100 are grouped into **20 superclasses**. Each image comes with a \"fine\" label (the class to which it belongs) and a \"coarse\" label (the superclass to which it belongs)."
      ],
      "metadata": {
        "id": "WgmTto-Vegs7"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "(train_images, train_labels), (test_images, test_labels) = datasets.cifar100.load_data()"
      ],
      "metadata": {
        "id": "b2rHy24Vezr5",
        "outputId": "12971b88-6780-4044-c58f-c9c6ebe9f34b",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Downloading data from https://www.cs.toronto.edu/~kriz/cifar-100-python.tar.gz\n",
            "169001437/169001437 [==============================] - 2s 0us/step\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "type(train_images)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ald6umlKgxET",
        "outputId": "f2106299-67e5-48ce-a81e-1091c0ba03a2"
      },
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "numpy.ndarray"
            ]
          },
          "metadata": {},
          "execution_count": 4
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "np.shape(train_images)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "chqFvyL-hC-J",
        "outputId": "770ec572-f465-4877-fc54-2725ba235cf8"
      },
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(50000, 32, 32, 3)"
            ]
          },
          "metadata": {},
          "execution_count": 5
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "train_images[0,1,1]"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "xHzmu7kShQya",
        "outputId": "8d78930e-453a-48f5-fed4-3b3c257f8b06"
      },
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([254, 254, 254], dtype=uint8)"
            ]
          },
          "metadata": {},
          "execution_count": 6
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "test_images[0,1,1]"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4QKUa8J5igRF",
        "outputId": "8bf92a50-a868-4e5b-9148-c3210e8d1a0f"
      },
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([195, 208, 238], dtype=uint8)"
            ]
          },
          "metadata": {},
          "execution_count": 7
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "np.shape(test_labels)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "lxntsPumjo7B",
        "outputId": "29ca42b2-918d-42c7-bdb8-8d95be79d4c0"
      },
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(10000, 1)"
            ]
          },
          "metadata": {},
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "test_labels[1,0]"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "43ZrHJNdjt5o",
        "outputId": "6486d69c-e5ee-4e9c-93f6-504a25c5d37f"
      },
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "33"
            ]
          },
          "metadata": {},
          "execution_count": 9
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.figure(figsize=(10, 10))\n",
        "for i in range (9):\n",
        "  rand_num = np.random.randint(0, 50000)\n",
        "  img = plt.subplot(3,3, i+1)\n",
        "  plt.imshow(train_images[rand_num])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 0
        },
        "id": "Y5CSc998jUNA",
        "outputId": "25ac21f0-ec41-4f49-ad38-8647e7351291"
      },
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 720x720 with 9 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "train_images[0,0].max()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "7W098j4RkX4Y",
        "outputId": "eca83a01-75b2-42c6-d224-dae8911f58ae"
      },
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "255"
            ]
          },
          "metadata": {},
          "execution_count": 11
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "test_images[0].max()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "wzQOTo2sk3h_",
        "outputId": "6ec692dc-6f83-448e-e571-c4b933c14d3a"
      },
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "255"
            ]
          },
          "metadata": {},
          "execution_count": 12
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#Normalization\n",
        "train_images = train_images/255.0\n",
        "test_images = test_images/255.0"
      ],
      "metadata": {
        "id": "BlhkOr1Pk-c5"
      },
      "execution_count": 13,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#One Hot Encoding\n",
        "train_labels = utils.to_categorical(train_labels)\n",
        "test_labels = utils.to_categorical(test_labels)"
      ],
      "metadata": {
        "id": "V7-EWk01lWT_"
      },
      "execution_count": 14,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "#**CNN Model**"
      ],
      "metadata": {
        "id": "BKhLsJY7ov__"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from keras.models import Sequential\n",
        "from keras.layers import MaxPooling2D, Conv2D, Flatten, Dense, Activation, Dropout\n",
        "\n",
        "# CNN for feature extraction\n",
        "model = Sequential()\n",
        "model.add(Conv2D(32, (3,3), activation='relu', input_shape=(32,32,3)))\n",
        "model.add(Conv2D(32, (3,3), activation='relu'))\n",
        "model.add(MaxPooling2D(2,2))\n",
        "model.add(Conv2D(32, (3,3), activation='relu'))\n",
        "model.add(Conv2D(32, (3,3), activation='relu'))\n",
        "model.add(MaxPooling2D(2,2))\n",
        "model.add(Conv2D(32, (3,3), activation='relu'))\n",
        "model.add(MaxPooling2D(2,2))\n",
        "\n",
        "# FC for classifying\n",
        "model.add(Flatten())\n",
        "model.add(Dense(64, activation='relu'))\n",
        "model.add(Dense(100, activation='softmax'))\n",
        "\n",
        "model.summary()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "wt0stJ7_o22o",
        "outputId": "5451a25e-a8e2-4966-cd08-7f3089d8e608"
      },
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " conv2d (Conv2D)             (None, 30, 30, 32)        896       \n",
            "                                                                 \n",
            " conv2d_1 (Conv2D)           (None, 28, 28, 32)        9248      \n",
            "                                                                 \n",
            " max_pooling2d (MaxPooling2D  (None, 14, 14, 32)       0         \n",
            " )                                                               \n",
            "                                                                 \n",
            " conv2d_2 (Conv2D)           (None, 12, 12, 32)        9248      \n",
            "                                                                 \n",
            " conv2d_3 (Conv2D)           (None, 10, 10, 32)        9248      \n",
            "                                                                 \n",
            " max_pooling2d_1 (MaxPooling  (None, 5, 5, 32)         0         \n",
            " 2D)                                                             \n",
            "                                                                 \n",
            " conv2d_4 (Conv2D)           (None, 3, 3, 32)          9248      \n",
            "                                                                 \n",
            " max_pooling2d_2 (MaxPooling  (None, 1, 1, 32)         0         \n",
            " 2D)                                                             \n",
            "                                                                 \n",
            " flatten (Flatten)           (None, 32)                0         \n",
            "                                                                 \n",
            " dense (Dense)               (None, 64)                2112      \n",
            "                                                                 \n",
            " dense_1 (Dense)             (None, 100)               6500      \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 46,500\n",
            "Trainable params: 46,500\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### The input matrix of **30x30x32** is converted into the **1xN matrix (1x1x32)** after processing it with **5 convolution layers and 3 max pooling** operations. So it is ready to be flattened to a single column and fed to the fully connected layer which will classify the image. "
      ],
      "metadata": {
        "id": "84QWJIr_OfAR"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# **Training & Testing**"
      ],
      "metadata": {
        "id": "kAFSjWO-q444"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from keras import optimizers\n",
        "from keras import backend as K\n",
        "import time"
      ],
      "metadata": {
        "id": "8uMCthZzq-Fu"
      },
      "execution_count": 16,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "##1 - Learning Rate: 0.001, Epochs/Steps: 50/10, Training time: 1 hr 45 min (appx), Accuracy: 19.7%\n",
        "\n"
      ],
      "metadata": {
        "id": "E5kpkIntHAX4"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model.compile(optimizer=optimizers.Adam(learning_rate=1e-3),\n",
        "              loss='categorical_crossentropy',\n",
        "              metrics=['acc'])\n",
        "print(\"Learning rate before first fit:\", model.optimizer.learning_rate.numpy())"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "K4U4-z88q8Ld",
        "outputId": "0a2313f9-95d9-45b5-a92b-a4d1f6867f21"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Learning rate before first fit: 0.001\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "training_start = time.time()\n",
        "history = model.fit(train_images, train_labels,\n",
        "                    steps_per_epoch=10,\n",
        "                    epochs=50,\n",
        "                    validation_split=0.2) \n",
        "training_stop = time.time()\n",
        "training_time = training_stop - training_start\n",
        "print(f\"Training time: {training_time}\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "7VwbyXZDtBp8",
        "outputId": "b20ce5bb-06de-4f37-ce4c-083a3d075ad1"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/50\n",
            "10/10 [==============================] - 134s 13s/step - loss: 4.5992 - acc: 0.0105 - val_loss: 4.5835 - val_acc: 0.0164\n",
            "Epoch 2/50\n",
            "10/10 [==============================] - 123s 12s/step - loss: 4.5503 - acc: 0.0158 - val_loss: 4.5079 - val_acc: 0.0206\n",
            "Epoch 3/50\n",
            "10/10 [==============================] - 127s 13s/step - loss: 4.4734 - acc: 0.0203 - val_loss: 4.4365 - val_acc: 0.0287\n",
            "Epoch 4/50\n",
            "10/10 [==============================] - 123s 12s/step - loss: 4.3952 - acc: 0.0269 - val_loss: 4.3597 - val_acc: 0.0328\n",
            "Epoch 5/50\n",
            "10/10 [==============================] - 129s 13s/step - loss: 4.3096 - acc: 0.0357 - val_loss: 4.2775 - val_acc: 0.0442\n",
            "Epoch 6/50\n",
            "10/10 [==============================] - 133s 13s/step - loss: 4.2344 - acc: 0.0478 - val_loss: 4.2168 - val_acc: 0.0533\n",
            "Epoch 7/50\n",
            "10/10 [==============================] - 124s 13s/step - loss: 4.1726 - acc: 0.0635 - val_loss: 4.1477 - val_acc: 0.0661\n",
            "Epoch 8/50\n",
            "10/10 [==============================] - 126s 13s/step - loss: 4.1133 - acc: 0.0735 - val_loss: 4.0955 - val_acc: 0.0706\n",
            "Epoch 9/50\n",
            "10/10 [==============================] - 123s 12s/step - loss: 4.0694 - acc: 0.0795 - val_loss: 4.0706 - val_acc: 0.0780\n",
            "Epoch 10/50\n",
            "10/10 [==============================] - 127s 13s/step - loss: 4.0317 - acc: 0.0825 - val_loss: 4.0246 - val_acc: 0.0872\n",
            "Epoch 11/50\n",
            "10/10 [==============================] - 122s 12s/step - loss: 3.9828 - acc: 0.0946 - val_loss: 3.9769 - val_acc: 0.0956\n",
            "Epoch 12/50\n",
            "10/10 [==============================] - 128s 13s/step - loss: 3.9326 - acc: 0.1024 - val_loss: 3.9394 - val_acc: 0.1039\n",
            "Epoch 13/50\n",
            "10/10 [==============================] - 128s 13s/step - loss: 3.8954 - acc: 0.1084 - val_loss: 3.9074 - val_acc: 0.1065\n",
            "Epoch 14/50\n",
            "10/10 [==============================] - 126s 13s/step - loss: 3.8633 - acc: 0.1123 - val_loss: 3.8825 - val_acc: 0.1135\n",
            "Epoch 15/50\n",
            "10/10 [==============================] - 127s 13s/step - loss: 3.8239 - acc: 0.1214 - val_loss: 3.8405 - val_acc: 0.1205\n",
            "Epoch 16/50\n",
            "10/10 [==============================] - 124s 12s/step - loss: 3.7813 - acc: 0.1285 - val_loss: 3.7998 - val_acc: 0.1245\n",
            "Epoch 17/50\n",
            "10/10 [==============================] - 123s 12s/step - loss: 3.7480 - acc: 0.1314 - val_loss: 3.7870 - val_acc: 0.1286\n",
            "Epoch 18/50\n",
            "10/10 [==============================] - 129s 13s/step - loss: 3.7359 - acc: 0.1332 - val_loss: 3.7756 - val_acc: 0.1275\n",
            "Epoch 19/50\n",
            "10/10 [==============================] - 128s 13s/step - loss: 3.7029 - acc: 0.1393 - val_loss: 3.7323 - val_acc: 0.1333\n",
            "Epoch 20/50\n",
            "10/10 [==============================] - 126s 13s/step - loss: 3.6719 - acc: 0.1452 - val_loss: 3.7302 - val_acc: 0.1388\n",
            "Epoch 21/50\n",
            "10/10 [==============================] - 125s 13s/step - loss: 3.6403 - acc: 0.1478 - val_loss: 3.6743 - val_acc: 0.1409\n",
            "Epoch 22/50\n",
            "10/10 [==============================] - 126s 13s/step - loss: 3.6088 - acc: 0.1522 - val_loss: 3.6482 - val_acc: 0.1445\n",
            "Epoch 23/50\n",
            "10/10 [==============================] - 123s 12s/step - loss: 3.6359 - acc: 0.1481 - val_loss: 3.7170 - val_acc: 0.1340\n",
            "Epoch 24/50\n",
            "10/10 [==============================] - 128s 13s/step - loss: 3.6070 - acc: 0.1533 - val_loss: 3.6353 - val_acc: 0.1490\n",
            "Epoch 25/50\n",
            "10/10 [==============================] - 128s 13s/step - loss: 3.5577 - acc: 0.1609 - val_loss: 3.6055 - val_acc: 0.1526\n",
            "Epoch 26/50\n",
            "10/10 [==============================] - 130s 13s/step - loss: 3.5440 - acc: 0.1634 - val_loss: 3.6007 - val_acc: 0.1515\n",
            "Epoch 27/50\n",
            "10/10 [==============================] - 124s 12s/step - loss: 3.5232 - acc: 0.1657 - val_loss: 3.5780 - val_acc: 0.1587\n",
            "Epoch 28/50\n",
            "10/10 [==============================] - 126s 13s/step - loss: 3.5074 - acc: 0.1691 - val_loss: 3.5600 - val_acc: 0.1603\n",
            "Epoch 29/50\n",
            "10/10 [==============================] - 124s 12s/step - loss: 3.4894 - acc: 0.1714 - val_loss: 3.5601 - val_acc: 0.1589\n",
            "Epoch 30/50\n",
            "10/10 [==============================] - 127s 13s/step - loss: 3.4803 - acc: 0.1719 - val_loss: 3.5269 - val_acc: 0.1632\n",
            "Epoch 31/50\n",
            "10/10 [==============================] - 127s 13s/step - loss: 3.4603 - acc: 0.1763 - val_loss: 3.5225 - val_acc: 0.1650\n",
            "Epoch 32/50\n",
            "10/10 [==============================] - 126s 13s/step - loss: 3.4622 - acc: 0.1748 - val_loss: 3.5157 - val_acc: 0.1666\n",
            "Epoch 33/50\n",
            "10/10 [==============================] - 124s 13s/step - loss: 3.4401 - acc: 0.1780 - val_loss: 3.4939 - val_acc: 0.1709\n",
            "Epoch 34/50\n",
            "10/10 [==============================] - 124s 12s/step - loss: 3.4186 - acc: 0.1823 - val_loss: 3.4907 - val_acc: 0.1694\n",
            "Epoch 35/50\n",
            "10/10 [==============================] - 128s 13s/step - loss: 3.4151 - acc: 0.1848 - val_loss: 3.4727 - val_acc: 0.1725\n",
            "Epoch 36/50\n",
            "10/10 [==============================] - 122s 12s/step - loss: 3.3960 - acc: 0.1878 - val_loss: 3.4625 - val_acc: 0.1721\n",
            "Epoch 37/50\n",
            "10/10 [==============================] - 126s 13s/step - loss: 3.3792 - acc: 0.1928 - val_loss: 3.4730 - val_acc: 0.1752\n",
            "Epoch 38/50\n",
            "10/10 [==============================] - 128s 13s/step - loss: 3.3909 - acc: 0.1886 - val_loss: 3.5112 - val_acc: 0.1682\n",
            "Epoch 39/50\n",
            "10/10 [==============================] - 127s 13s/step - loss: 3.3815 - acc: 0.1905 - val_loss: 3.4529 - val_acc: 0.1760\n",
            "Epoch 40/50\n",
            "10/10 [==============================] - 127s 13s/step - loss: 3.3587 - acc: 0.1951 - val_loss: 3.4507 - val_acc: 0.1802\n",
            "Epoch 41/50\n",
            "10/10 [==============================] - 129s 13s/step - loss: 3.3420 - acc: 0.1981 - val_loss: 3.4399 - val_acc: 0.1799\n",
            "Epoch 42/50\n",
            "10/10 [==============================] - 126s 13s/step - loss: 3.3273 - acc: 0.1987 - val_loss: 3.4201 - val_acc: 0.1822\n",
            "Epoch 43/50\n",
            "10/10 [==============================] - 127s 13s/step - loss: 3.3236 - acc: 0.2001 - val_loss: 3.4046 - val_acc: 0.1870\n",
            "Epoch 44/50\n",
            "10/10 [==============================] - 128s 13s/step - loss: 3.3115 - acc: 0.2029 - val_loss: 3.4040 - val_acc: 0.1852\n",
            "Epoch 45/50\n",
            "10/10 [==============================] - 123s 12s/step - loss: 3.3160 - acc: 0.2013 - val_loss: 3.4094 - val_acc: 0.1863\n",
            "Epoch 46/50\n",
            "10/10 [==============================] - 127s 13s/step - loss: 3.3110 - acc: 0.2023 - val_loss: 3.4155 - val_acc: 0.1829\n",
            "Epoch 47/50\n",
            "10/10 [==============================] - 125s 13s/step - loss: 3.2991 - acc: 0.2064 - val_loss: 3.3890 - val_acc: 0.1917\n",
            "Epoch 48/50\n",
            "10/10 [==============================] - 127s 13s/step - loss: 3.2991 - acc: 0.2060 - val_loss: 3.3799 - val_acc: 0.1940\n",
            "Epoch 49/50\n",
            "10/10 [==============================] - 122s 12s/step - loss: 3.2964 - acc: 0.2063 - val_loss: 3.3956 - val_acc: 0.1862\n",
            "Epoch 50/50\n",
            "10/10 [==============================] - 128s 13s/step - loss: 3.2733 - acc: 0.2078 - val_loss: 3.3995 - val_acc: 0.1885\n",
            "Training time: 6324.650381088257\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.plot(history.history['acc'], label='accuracy')\n",
        "plt.plot(history.history['val_acc'], label = 'val_accuracy')\n",
        "plt.xlabel('Epoch')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.legend(loc='lower right')\n",
        "\n",
        "test_loss, test_acc = model.evaluate(test_images,  test_labels, verbose=2)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 296
        },
        "id": "TRsnQcNZvgSz",
        "outputId": "f165e768-840d-47be-f897-a08a4056f787"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 - 6s - loss: 3.3687 - acc: 0.1968 - 6s/epoch - 19ms/step\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(test_acc)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "qJQGXUdEwGQU",
        "outputId": "582ed50d-25cb-477c-a13b-522f43f9096a"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "0.19679999351501465\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "##2 - Learning Rate: 0.0001, Epochs/Steps: 100/100, GPU selected as Hardware Accelerator, Training time: 5.4 minutes, Accuracy: 22%"
      ],
      "metadata": {
        "id": "Ifdz_AJ4HFe0"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model.compile(optimizer=optimizers.Adam(learning_rate=0.0001),\n",
        "              loss='categorical_crossentropy',\n",
        "              metrics=['acc'])\n",
        "print(\"Learning rate before second fit:\", model.optimizer.learning_rate.numpy())"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "eYvo6-wUHGnn",
        "outputId": "077d022d-4bda-4cc7-b060-c23b7b1ad92e"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Learning rate before second fit: 1e-04\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import time\n",
        "training_start = time.time()\n",
        "history = model.fit(train_images, train_labels,\n",
        "                    steps_per_epoch=100,\n",
        "                    epochs=100,\n",
        "                    validation_split=0.2) \n",
        "training_stop = time.time()\n",
        "training_time = training_stop - training_start\n",
        "print(f\"Training time: {training_time}\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "bYblU0fCJvWj",
        "outputId": "2c61f173-50b1-41cc-9b75-d897c6bcad93"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/100\n",
            "100/100 [==============================] - 12s 37ms/step - loss: 4.6008 - acc: 0.0104 - val_loss: 4.5827 - val_acc: 0.0121\n",
            "Epoch 2/100\n",
            "100/100 [==============================] - 3s 29ms/step - loss: 4.5260 - acc: 0.0198 - val_loss: 4.4709 - val_acc: 0.0225\n",
            "Epoch 3/100\n",
            "100/100 [==============================] - 3s 29ms/step - loss: 4.4211 - acc: 0.0249 - val_loss: 4.3839 - val_acc: 0.0358\n",
            "Epoch 4/100\n",
            "100/100 [==============================] - 3s 29ms/step - loss: 4.3482 - acc: 0.0331 - val_loss: 4.3255 - val_acc: 0.0366\n",
            "Epoch 5/100\n",
            "100/100 [==============================] - 3s 28ms/step - loss: 4.2920 - acc: 0.0395 - val_loss: 4.2775 - val_acc: 0.0436\n",
            "Epoch 6/100\n",
            "100/100 [==============================] - 2s 24ms/step - loss: 4.2449 - acc: 0.0450 - val_loss: 4.2433 - val_acc: 0.0467\n",
            "Epoch 7/100\n",
            "100/100 [==============================] - 2s 24ms/step - loss: 4.2067 - acc: 0.0494 - val_loss: 4.2046 - val_acc: 0.0550\n",
            "Epoch 8/100\n",
            "100/100 [==============================] - 2s 24ms/step - loss: 4.1731 - acc: 0.0560 - val_loss: 4.1861 - val_acc: 0.0573\n",
            "Epoch 9/100\n",
            "100/100 [==============================] - 2s 25ms/step - loss: 4.1425 - acc: 0.0615 - val_loss: 4.1473 - val_acc: 0.0594\n",
            "Epoch 10/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 4.1113 - acc: 0.0685 - val_loss: 4.1126 - val_acc: 0.0721\n",
            "Epoch 11/100\n",
            "100/100 [==============================] - 3s 29ms/step - loss: 4.0808 - acc: 0.0758 - val_loss: 4.0822 - val_acc: 0.0765\n",
            "Epoch 12/100\n",
            "100/100 [==============================] - 2s 24ms/step - loss: 4.0505 - acc: 0.0811 - val_loss: 4.0577 - val_acc: 0.0826\n",
            "Epoch 13/100\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 4.0213 - acc: 0.0871 - val_loss: 4.0304 - val_acc: 0.0919\n",
            "Epoch 14/100\n",
            "100/100 [==============================] - 4s 39ms/step - loss: 3.9929 - acc: 0.0934 - val_loss: 4.0112 - val_acc: 0.0908\n",
            "Epoch 15/100\n",
            "100/100 [==============================] - 3s 34ms/step - loss: 3.9660 - acc: 0.0973 - val_loss: 3.9756 - val_acc: 0.0964\n",
            "Epoch 16/100\n",
            "100/100 [==============================] - 3s 28ms/step - loss: 3.9375 - acc: 0.1022 - val_loss: 3.9570 - val_acc: 0.0999\n",
            "Epoch 17/100\n",
            "100/100 [==============================] - 3s 28ms/step - loss: 3.9133 - acc: 0.1075 - val_loss: 3.9280 - val_acc: 0.1065\n",
            "Epoch 18/100\n",
            "100/100 [==============================] - 3s 33ms/step - loss: 3.8865 - acc: 0.1122 - val_loss: 3.9085 - val_acc: 0.1092\n",
            "Epoch 19/100\n",
            "100/100 [==============================] - 3s 33ms/step - loss: 3.8613 - acc: 0.1162 - val_loss: 3.8887 - val_acc: 0.1121\n",
            "Epoch 20/100\n",
            "100/100 [==============================] - 2s 24ms/step - loss: 3.8416 - acc: 0.1184 - val_loss: 3.8677 - val_acc: 0.1155\n",
            "Epoch 21/100\n",
            "100/100 [==============================] - 3s 29ms/step - loss: 3.8164 - acc: 0.1222 - val_loss: 3.8539 - val_acc: 0.1203\n",
            "Epoch 22/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.7976 - acc: 0.1258 - val_loss: 3.8279 - val_acc: 0.1231\n",
            "Epoch 23/100\n",
            "100/100 [==============================] - 3s 28ms/step - loss: 3.7742 - acc: 0.1292 - val_loss: 3.8071 - val_acc: 0.1242\n",
            "Epoch 24/100\n",
            "100/100 [==============================] - 3s 28ms/step - loss: 3.7557 - acc: 0.1328 - val_loss: 3.7906 - val_acc: 0.1277\n",
            "Epoch 25/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.7383 - acc: 0.1329 - val_loss: 3.7732 - val_acc: 0.1315\n",
            "Epoch 26/100\n",
            "100/100 [==============================] - 3s 33ms/step - loss: 3.7194 - acc: 0.1373 - val_loss: 3.7603 - val_acc: 0.1306\n",
            "Epoch 27/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.6997 - acc: 0.1397 - val_loss: 3.7417 - val_acc: 0.1343\n",
            "Epoch 28/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.6858 - acc: 0.1436 - val_loss: 3.7277 - val_acc: 0.1378\n",
            "Epoch 29/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.6698 - acc: 0.1471 - val_loss: 3.7241 - val_acc: 0.1373\n",
            "Epoch 30/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.6548 - acc: 0.1488 - val_loss: 3.7024 - val_acc: 0.1408\n",
            "Epoch 31/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.6414 - acc: 0.1505 - val_loss: 3.6922 - val_acc: 0.1409\n",
            "Epoch 32/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.6282 - acc: 0.1530 - val_loss: 3.6813 - val_acc: 0.1419\n",
            "Epoch 33/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.6146 - acc: 0.1548 - val_loss: 3.6625 - val_acc: 0.1452\n",
            "Epoch 34/100\n",
            "100/100 [==============================] - 2s 25ms/step - loss: 3.6021 - acc: 0.1563 - val_loss: 3.6651 - val_acc: 0.1471\n",
            "Epoch 35/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.5919 - acc: 0.1588 - val_loss: 3.6451 - val_acc: 0.1499\n",
            "Epoch 36/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.5810 - acc: 0.1610 - val_loss: 3.6372 - val_acc: 0.1479\n",
            "Epoch 37/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.5683 - acc: 0.1626 - val_loss: 3.6292 - val_acc: 0.1497\n",
            "Epoch 38/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.5632 - acc: 0.1632 - val_loss: 3.6315 - val_acc: 0.1510\n",
            "Epoch 39/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.5470 - acc: 0.1663 - val_loss: 3.6134 - val_acc: 0.1514\n",
            "Epoch 40/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.5386 - acc: 0.1677 - val_loss: 3.6038 - val_acc: 0.1527\n",
            "Epoch 41/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.5268 - acc: 0.1693 - val_loss: 3.5965 - val_acc: 0.1544\n",
            "Epoch 42/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.5181 - acc: 0.1693 - val_loss: 3.5890 - val_acc: 0.1578\n",
            "Epoch 43/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.5077 - acc: 0.1717 - val_loss: 3.5825 - val_acc: 0.1585\n",
            "Epoch 44/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.4994 - acc: 0.1732 - val_loss: 3.5787 - val_acc: 0.1585\n",
            "Epoch 45/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.4922 - acc: 0.1743 - val_loss: 3.5591 - val_acc: 0.1616\n",
            "Epoch 46/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.4827 - acc: 0.1762 - val_loss: 3.5612 - val_acc: 0.1630\n",
            "Epoch 47/100\n",
            "100/100 [==============================] - 3s 29ms/step - loss: 3.4729 - acc: 0.1764 - val_loss: 3.5511 - val_acc: 0.1637\n",
            "Epoch 48/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.4648 - acc: 0.1786 - val_loss: 3.5446 - val_acc: 0.1669\n",
            "Epoch 49/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.4572 - acc: 0.1783 - val_loss: 3.5307 - val_acc: 0.1683\n",
            "Epoch 50/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.4501 - acc: 0.1810 - val_loss: 3.5332 - val_acc: 0.1650\n",
            "Epoch 51/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.4420 - acc: 0.1817 - val_loss: 3.5282 - val_acc: 0.1697\n",
            "Epoch 52/100\n",
            "100/100 [==============================] - 3s 29ms/step - loss: 3.4321 - acc: 0.1840 - val_loss: 3.5110 - val_acc: 0.1724\n",
            "Epoch 53/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.4245 - acc: 0.1852 - val_loss: 3.5061 - val_acc: 0.1747\n",
            "Epoch 54/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.4203 - acc: 0.1852 - val_loss: 3.4977 - val_acc: 0.1735\n",
            "Epoch 55/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.4110 - acc: 0.1869 - val_loss: 3.4997 - val_acc: 0.1730\n",
            "Epoch 56/100\n",
            "100/100 [==============================] - 4s 37ms/step - loss: 3.4045 - acc: 0.1883 - val_loss: 3.4896 - val_acc: 0.1756\n",
            "Epoch 57/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.3979 - acc: 0.1905 - val_loss: 3.4992 - val_acc: 0.1754\n",
            "Epoch 58/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.3930 - acc: 0.1902 - val_loss: 3.4758 - val_acc: 0.1793\n",
            "Epoch 59/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.3850 - acc: 0.1917 - val_loss: 3.4737 - val_acc: 0.1793\n",
            "Epoch 60/100\n",
            "100/100 [==============================] - 3s 29ms/step - loss: 3.3798 - acc: 0.1933 - val_loss: 3.4871 - val_acc: 0.1780\n",
            "Epoch 61/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.3730 - acc: 0.1940 - val_loss: 3.4620 - val_acc: 0.1784\n",
            "Epoch 62/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.3709 - acc: 0.1942 - val_loss: 3.4623 - val_acc: 0.1804\n",
            "Epoch 63/100\n",
            "100/100 [==============================] - 2s 25ms/step - loss: 3.3614 - acc: 0.1963 - val_loss: 3.4689 - val_acc: 0.1812\n",
            "Epoch 64/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.3534 - acc: 0.1970 - val_loss: 3.4640 - val_acc: 0.1816\n",
            "Epoch 65/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.3478 - acc: 0.1993 - val_loss: 3.4491 - val_acc: 0.1837\n",
            "Epoch 66/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.3437 - acc: 0.1985 - val_loss: 3.4356 - val_acc: 0.1881\n",
            "Epoch 67/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.3394 - acc: 0.2001 - val_loss: 3.4401 - val_acc: 0.1849\n",
            "Epoch 68/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.3320 - acc: 0.2019 - val_loss: 3.4311 - val_acc: 0.1884\n",
            "Epoch 69/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.3284 - acc: 0.2006 - val_loss: 3.4341 - val_acc: 0.1865\n",
            "Epoch 70/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.3217 - acc: 0.2034 - val_loss: 3.4209 - val_acc: 0.1872\n",
            "Epoch 71/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.3182 - acc: 0.2050 - val_loss: 3.4235 - val_acc: 0.1860\n",
            "Epoch 72/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.3142 - acc: 0.2054 - val_loss: 3.4074 - val_acc: 0.1916\n",
            "Epoch 73/100\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 3.3085 - acc: 0.2063 - val_loss: 3.4028 - val_acc: 0.1921\n",
            "Epoch 74/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.3003 - acc: 0.2068 - val_loss: 3.4033 - val_acc: 0.1932\n",
            "Epoch 75/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.2968 - acc: 0.2095 - val_loss: 3.3964 - val_acc: 0.1946\n",
            "Epoch 76/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.2898 - acc: 0.2098 - val_loss: 3.3971 - val_acc: 0.1967\n",
            "Epoch 77/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.2838 - acc: 0.2110 - val_loss: 3.3976 - val_acc: 0.1951\n",
            "Epoch 78/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.2866 - acc: 0.2093 - val_loss: 3.3848 - val_acc: 0.1967\n",
            "Epoch 79/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.2793 - acc: 0.2125 - val_loss: 3.3824 - val_acc: 0.1995\n",
            "Epoch 80/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.2756 - acc: 0.2114 - val_loss: 3.3758 - val_acc: 0.1999\n",
            "Epoch 81/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.2684 - acc: 0.2139 - val_loss: 3.3748 - val_acc: 0.1974\n",
            "Epoch 82/100\n",
            "100/100 [==============================] - 3s 29ms/step - loss: 3.2612 - acc: 0.2134 - val_loss: 3.3713 - val_acc: 0.1992\n",
            "Epoch 83/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.2577 - acc: 0.2132 - val_loss: 3.3661 - val_acc: 0.2018\n",
            "Epoch 84/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.2557 - acc: 0.2156 - val_loss: 3.3664 - val_acc: 0.2007\n",
            "Epoch 85/100\n",
            "100/100 [==============================] - 2s 25ms/step - loss: 3.2480 - acc: 0.2173 - val_loss: 3.3619 - val_acc: 0.2025\n",
            "Epoch 86/100\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 3.2455 - acc: 0.2171 - val_loss: 3.3531 - val_acc: 0.2041\n",
            "Epoch 87/100\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 3.2431 - acc: 0.2186 - val_loss: 3.3498 - val_acc: 0.2026\n",
            "Epoch 88/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.2383 - acc: 0.2184 - val_loss: 3.3467 - val_acc: 0.2059\n",
            "Epoch 89/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.2314 - acc: 0.2194 - val_loss: 3.3589 - val_acc: 0.2040\n",
            "Epoch 90/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.2285 - acc: 0.2209 - val_loss: 3.3427 - val_acc: 0.2037\n",
            "Epoch 91/100\n",
            "100/100 [==============================] - 3s 29ms/step - loss: 3.2236 - acc: 0.2205 - val_loss: 3.3398 - val_acc: 0.2026\n",
            "Epoch 92/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.2196 - acc: 0.2228 - val_loss: 3.3485 - val_acc: 0.2016\n",
            "Epoch 93/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.2153 - acc: 0.2227 - val_loss: 3.3440 - val_acc: 0.2051\n",
            "Epoch 94/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.2150 - acc: 0.2224 - val_loss: 3.3297 - val_acc: 0.2073\n",
            "Epoch 95/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.2096 - acc: 0.2245 - val_loss: 3.3242 - val_acc: 0.2074\n",
            "Epoch 96/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.2023 - acc: 0.2245 - val_loss: 3.3237 - val_acc: 0.2083\n",
            "Epoch 97/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.2007 - acc: 0.2275 - val_loss: 3.3299 - val_acc: 0.2099\n",
            "Epoch 98/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.1986 - acc: 0.2256 - val_loss: 3.3161 - val_acc: 0.2110\n",
            "Epoch 99/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.1931 - acc: 0.2250 - val_loss: 3.3226 - val_acc: 0.2074\n",
            "Epoch 100/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.1909 - acc: 0.2271 - val_loss: 3.3054 - val_acc: 0.2111\n",
            "Training time: 323.9760437011719\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.plot(history.history['acc'], label='accuracy')\n",
        "plt.plot(history.history['val_acc'], label = 'val_accuracy')\n",
        "plt.xlabel('Epoch')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.legend(loc='lower right')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 296
        },
        "id": "4oK8BAA9pOmi",
        "outputId": "6a1673f1-9c49-439d-f950-58e45c6748c4"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7f472006ac70>"
            ]
          },
          "metadata": {},
          "execution_count": 43
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "test_loss, test_acc = model.evaluate(test_images,  test_labels, verbose=2)\n",
        "print(test_acc)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "E5gkagCUwqYB",
        "outputId": "ad43cb01-b41f-49b5-9e4e-231dc0d4c73c"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 - 1s - loss: 3.2837 - acc: 0.2177 - 873ms/epoch - 3ms/step\n",
            "0.21770000457763672\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "##3 - Learning Rate: 1e-05, Epochs/Steps: 100/100, GPU as Hardware Accelerator, Training time: 5.33 minutes, Accuracy: 23%"
      ],
      "metadata": {
        "id": "BLuMBnfMxBL3"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model.compile(optimizer=optimizers.Adam(learning_rate=1e-05),\n",
        "              loss='categorical_crossentropy',\n",
        "              metrics=['acc'])\n",
        "print(\"Learning rate before third fit:\", model.optimizer.learning_rate.numpy())"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "fTnPOdsXxSvw",
        "outputId": "31ef5073-3f79-4300-b306-bed9b296322a"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Learning rate before third fit: 1e-05\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "training_start = time.time()\n",
        "history = model.fit(train_images, train_labels,\n",
        "                    steps_per_epoch=100,\n",
        "                    epochs=100,\n",
        "                    validation_split=0.2) \n",
        "training_stop = time.time()\n",
        "training_time = training_stop - training_start\n",
        "print(f\"Training time: {training_time}\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "GVPpyCfFxamp",
        "outputId": "fc56bd0c-d20d-42c0-9856-6e042718e558"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/100\n",
            "100/100 [==============================] - 4s 36ms/step - loss: 3.1278 - acc: 0.2394 - val_loss: 3.2241 - val_acc: 0.2261\n",
            "Epoch 2/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.1271 - acc: 0.2396 - val_loss: 3.2245 - val_acc: 0.2258\n",
            "Epoch 3/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.1266 - acc: 0.2403 - val_loss: 3.2241 - val_acc: 0.2260\n",
            "Epoch 4/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.1261 - acc: 0.2404 - val_loss: 3.2251 - val_acc: 0.2266\n",
            "Epoch 5/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.1253 - acc: 0.2406 - val_loss: 3.2253 - val_acc: 0.2238\n",
            "Epoch 6/100\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 3.1247 - acc: 0.2405 - val_loss: 3.2258 - val_acc: 0.2251\n",
            "Epoch 7/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.1240 - acc: 0.2404 - val_loss: 3.2261 - val_acc: 0.2270\n",
            "Epoch 8/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.1239 - acc: 0.2396 - val_loss: 3.2263 - val_acc: 0.2263\n",
            "Epoch 9/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.1233 - acc: 0.2398 - val_loss: 3.2262 - val_acc: 0.2258\n",
            "Epoch 10/100\n",
            "100/100 [==============================] - 3s 33ms/step - loss: 3.1234 - acc: 0.2401 - val_loss: 3.2365 - val_acc: 0.2233\n",
            "Epoch 11/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.1235 - acc: 0.2405 - val_loss: 3.2260 - val_acc: 0.2267\n",
            "Epoch 12/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.1223 - acc: 0.2394 - val_loss: 3.2264 - val_acc: 0.2253\n",
            "Epoch 13/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.1217 - acc: 0.2402 - val_loss: 3.2249 - val_acc: 0.2252\n",
            "Epoch 14/100\n",
            "100/100 [==============================] - 5s 51ms/step - loss: 3.1212 - acc: 0.2399 - val_loss: 3.2283 - val_acc: 0.2261\n",
            "Epoch 15/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.1208 - acc: 0.2410 - val_loss: 3.2259 - val_acc: 0.2272\n",
            "Epoch 16/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.1202 - acc: 0.2403 - val_loss: 3.2254 - val_acc: 0.2261\n",
            "Epoch 17/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.1201 - acc: 0.2409 - val_loss: 3.2266 - val_acc: 0.2244\n",
            "Epoch 18/100\n",
            "100/100 [==============================] - 3s 35ms/step - loss: 3.1196 - acc: 0.2402 - val_loss: 3.2256 - val_acc: 0.2268\n",
            "Epoch 19/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.1187 - acc: 0.2407 - val_loss: 3.2258 - val_acc: 0.2264\n",
            "Epoch 20/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.1190 - acc: 0.2405 - val_loss: 3.2296 - val_acc: 0.2260\n",
            "Epoch 21/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.1185 - acc: 0.2409 - val_loss: 3.2254 - val_acc: 0.2267\n",
            "Epoch 22/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.1177 - acc: 0.2410 - val_loss: 3.2270 - val_acc: 0.2261\n",
            "Epoch 23/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.1174 - acc: 0.2413 - val_loss: 3.2267 - val_acc: 0.2272\n",
            "Epoch 24/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.1171 - acc: 0.2411 - val_loss: 3.2282 - val_acc: 0.2268\n",
            "Epoch 25/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.1168 - acc: 0.2412 - val_loss: 3.2263 - val_acc: 0.2267\n",
            "Epoch 26/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.1165 - acc: 0.2411 - val_loss: 3.2258 - val_acc: 0.2272\n",
            "Epoch 27/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.1160 - acc: 0.2412 - val_loss: 3.2252 - val_acc: 0.2267\n",
            "Epoch 28/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.1154 - acc: 0.2407 - val_loss: 3.2255 - val_acc: 0.2270\n",
            "Epoch 29/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.1152 - acc: 0.2421 - val_loss: 3.2261 - val_acc: 0.2266\n",
            "Epoch 30/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.1148 - acc: 0.2409 - val_loss: 3.2251 - val_acc: 0.2272\n",
            "Epoch 31/100\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 3.1144 - acc: 0.2411 - val_loss: 3.2257 - val_acc: 0.2263\n",
            "Epoch 32/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.1141 - acc: 0.2418 - val_loss: 3.2243 - val_acc: 0.2283\n",
            "Epoch 33/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.1136 - acc: 0.2417 - val_loss: 3.2237 - val_acc: 0.2254\n",
            "Epoch 34/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.1129 - acc: 0.2417 - val_loss: 3.2252 - val_acc: 0.2261\n",
            "Epoch 35/100\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 3.1128 - acc: 0.2422 - val_loss: 3.2250 - val_acc: 0.2268\n",
            "Epoch 36/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.1124 - acc: 0.2421 - val_loss: 3.2232 - val_acc: 0.2261\n",
            "Epoch 37/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.1117 - acc: 0.2420 - val_loss: 3.2247 - val_acc: 0.2265\n",
            "Epoch 38/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.1115 - acc: 0.2411 - val_loss: 3.2240 - val_acc: 0.2280\n",
            "Epoch 39/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.1112 - acc: 0.2416 - val_loss: 3.2233 - val_acc: 0.2266\n",
            "Epoch 40/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.1106 - acc: 0.2419 - val_loss: 3.2237 - val_acc: 0.2263\n",
            "Epoch 41/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.1110 - acc: 0.2421 - val_loss: 3.2236 - val_acc: 0.2275\n",
            "Epoch 42/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.1099 - acc: 0.2425 - val_loss: 3.2230 - val_acc: 0.2273\n",
            "Epoch 43/100\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 3.1095 - acc: 0.2422 - val_loss: 3.2224 - val_acc: 0.2254\n",
            "Epoch 44/100\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 3.1102 - acc: 0.2421 - val_loss: 3.2228 - val_acc: 0.2271\n",
            "Epoch 45/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.1087 - acc: 0.2423 - val_loss: 3.2223 - val_acc: 0.2251\n",
            "Epoch 46/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.1090 - acc: 0.2431 - val_loss: 3.2244 - val_acc: 0.2264\n",
            "Epoch 47/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.1075 - acc: 0.2424 - val_loss: 3.2235 - val_acc: 0.2260\n",
            "Epoch 48/100\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 3.1076 - acc: 0.2422 - val_loss: 3.2213 - val_acc: 0.2276\n",
            "Epoch 49/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.1075 - acc: 0.2424 - val_loss: 3.2221 - val_acc: 0.2268\n",
            "Epoch 50/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.1069 - acc: 0.2427 - val_loss: 3.2213 - val_acc: 0.2268\n",
            "Epoch 51/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.1066 - acc: 0.2422 - val_loss: 3.2232 - val_acc: 0.2271\n",
            "Epoch 52/100\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 3.1068 - acc: 0.2427 - val_loss: 3.2210 - val_acc: 0.2261\n",
            "Epoch 53/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.1063 - acc: 0.2430 - val_loss: 3.2216 - val_acc: 0.2274\n",
            "Epoch 54/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.1055 - acc: 0.2430 - val_loss: 3.2205 - val_acc: 0.2269\n",
            "Epoch 55/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.1047 - acc: 0.2430 - val_loss: 3.2220 - val_acc: 0.2275\n",
            "Epoch 56/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.1048 - acc: 0.2435 - val_loss: 3.2197 - val_acc: 0.2284\n",
            "Epoch 57/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.1039 - acc: 0.2429 - val_loss: 3.2217 - val_acc: 0.2262\n",
            "Epoch 58/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.1039 - acc: 0.2433 - val_loss: 3.2210 - val_acc: 0.2280\n",
            "Epoch 59/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.1034 - acc: 0.2438 - val_loss: 3.2214 - val_acc: 0.2271\n",
            "Epoch 60/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.1034 - acc: 0.2432 - val_loss: 3.2207 - val_acc: 0.2280\n",
            "Epoch 61/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.1029 - acc: 0.2439 - val_loss: 3.2203 - val_acc: 0.2276\n",
            "Epoch 62/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.1022 - acc: 0.2438 - val_loss: 3.2200 - val_acc: 0.2281\n",
            "Epoch 63/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.1020 - acc: 0.2424 - val_loss: 3.2188 - val_acc: 0.2276\n",
            "Epoch 64/100\n",
            "100/100 [==============================] - 3s 28ms/step - loss: 3.1013 - acc: 0.2443 - val_loss: 3.2202 - val_acc: 0.2272\n",
            "Epoch 65/100\n",
            "100/100 [==============================] - 3s 28ms/step - loss: 3.1008 - acc: 0.2437 - val_loss: 3.2201 - val_acc: 0.2278\n",
            "Epoch 66/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.1018 - acc: 0.2433 - val_loss: 3.2196 - val_acc: 0.2274\n",
            "Epoch 67/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.1007 - acc: 0.2440 - val_loss: 3.2182 - val_acc: 0.2280\n",
            "Epoch 68/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.1003 - acc: 0.2444 - val_loss: 3.2197 - val_acc: 0.2264\n",
            "Epoch 69/100\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 3.0998 - acc: 0.2445 - val_loss: 3.2182 - val_acc: 0.2284\n",
            "Epoch 70/100\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 3.0994 - acc: 0.2440 - val_loss: 3.2184 - val_acc: 0.2286\n",
            "Epoch 71/100\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 3.0988 - acc: 0.2445 - val_loss: 3.2178 - val_acc: 0.2286\n",
            "Epoch 72/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.0990 - acc: 0.2441 - val_loss: 3.2176 - val_acc: 0.2272\n",
            "Epoch 73/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.0985 - acc: 0.2442 - val_loss: 3.2176 - val_acc: 0.2278\n",
            "Epoch 74/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.0981 - acc: 0.2441 - val_loss: 3.2167 - val_acc: 0.2279\n",
            "Epoch 75/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.0977 - acc: 0.2446 - val_loss: 3.2171 - val_acc: 0.2275\n",
            "Epoch 76/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.0973 - acc: 0.2448 - val_loss: 3.2162 - val_acc: 0.2283\n",
            "Epoch 77/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.0965 - acc: 0.2453 - val_loss: 3.2157 - val_acc: 0.2284\n",
            "Epoch 78/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.0962 - acc: 0.2450 - val_loss: 3.2159 - val_acc: 0.2274\n",
            "Epoch 79/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.0961 - acc: 0.2449 - val_loss: 3.2173 - val_acc: 0.2272\n",
            "Epoch 80/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.0953 - acc: 0.2442 - val_loss: 3.2157 - val_acc: 0.2285\n",
            "Epoch 81/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.0952 - acc: 0.2446 - val_loss: 3.2152 - val_acc: 0.2291\n",
            "Epoch 82/100\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 3.0955 - acc: 0.2449 - val_loss: 3.2159 - val_acc: 0.2290\n",
            "Epoch 83/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.0945 - acc: 0.2452 - val_loss: 3.2154 - val_acc: 0.2283\n",
            "Epoch 84/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.0951 - acc: 0.2442 - val_loss: 3.2176 - val_acc: 0.2297\n",
            "Epoch 85/100\n",
            "100/100 [==============================] - 3s 29ms/step - loss: 3.0941 - acc: 0.2447 - val_loss: 3.2152 - val_acc: 0.2301\n",
            "Epoch 86/100\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 3.0938 - acc: 0.2453 - val_loss: 3.2151 - val_acc: 0.2273\n",
            "Epoch 87/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.0930 - acc: 0.2454 - val_loss: 3.2140 - val_acc: 0.2281\n",
            "Epoch 88/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.0927 - acc: 0.2454 - val_loss: 3.2148 - val_acc: 0.2275\n",
            "Epoch 89/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.0925 - acc: 0.2455 - val_loss: 3.2157 - val_acc: 0.2282\n",
            "Epoch 90/100\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 3.0924 - acc: 0.2465 - val_loss: 3.2137 - val_acc: 0.2286\n",
            "Epoch 91/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.0917 - acc: 0.2452 - val_loss: 3.2128 - val_acc: 0.2272\n",
            "Epoch 92/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.0914 - acc: 0.2460 - val_loss: 3.2155 - val_acc: 0.2279\n",
            "Epoch 93/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.0912 - acc: 0.2454 - val_loss: 3.2140 - val_acc: 0.2287\n",
            "Epoch 94/100\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 3.0908 - acc: 0.2455 - val_loss: 3.2156 - val_acc: 0.2280\n",
            "Epoch 95/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.0903 - acc: 0.2459 - val_loss: 3.2117 - val_acc: 0.2294\n",
            "Epoch 96/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.0900 - acc: 0.2464 - val_loss: 3.2132 - val_acc: 0.2306\n",
            "Epoch 97/100\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 3.0896 - acc: 0.2460 - val_loss: 3.2131 - val_acc: 0.2294\n",
            "Epoch 98/100\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 3.0888 - acc: 0.2457 - val_loss: 3.2132 - val_acc: 0.2291\n",
            "Epoch 99/100\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 3.0887 - acc: 0.2459 - val_loss: 3.2112 - val_acc: 0.2293\n",
            "Epoch 100/100\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 3.0885 - acc: 0.2465 - val_loss: 3.2117 - val_acc: 0.2288\n",
            "Training time: 323.4463264942169\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.plot(history.history['acc'], label='accuracy')\n",
        "plt.plot(history.history['val_acc'], label = 'val_accuracy')\n",
        "plt.xlabel('Epoch')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.legend(loc='lower right')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 296
        },
        "id": "dgUV6q8DyotW",
        "outputId": "3e632486-fcc5-4da9-ff81-8d0c268981ca"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7f47201ff760>"
            ]
          },
          "metadata": {},
          "execution_count": 51
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "test_loss, test_acc = model.evaluate(test_images,  test_labels, verbose=2)\n",
        "print(test_acc)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "GDaa3dwa09uB",
        "outputId": "bbf209f1-eb23-4cd3-e878-c3bcd39ec7e2"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 - 1s - loss: 3.2063 - acc: 0.2302 - 1s/epoch - 4ms/step\n",
            "0.23019999265670776\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "##4 - Learning Rate: 1e-04, Epochs/Steps: 500/100, GPU as Hardware Accelerator, Training time: 24.22 minutes, Accuracy: 27.4%"
      ],
      "metadata": {
        "id": "aSHBfwOY1Lmt"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model.compile(optimizer=optimizers.Adam(learning_rate=1e-05),\n",
        "              loss='categorical_crossentropy',\n",
        "              metrics=['acc'])\n",
        "print(\"Learning rate before fourth fit:\", model.optimizer.learning_rate.numpy())"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "MnbGRiir1Tzs",
        "outputId": "58e418de-1bcd-4d3a-dbf9-6151a5f216a0"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Learning rate before fourth fit: 1e-05\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "training_start = time.time()\n",
        "history = model.fit(train_images, train_labels,\n",
        "                    steps_per_epoch=100,\n",
        "                    epochs=500,\n",
        "                    validation_split=0.2) \n",
        "training_stop = time.time()\n",
        "training_time = training_stop - training_start\n",
        "print(f\"Training time: {training_time}\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ImyDRzvy1cWk",
        "outputId": "ae9b1076-fe25-4ba9-f8e0-49c18db9659f"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/500\n",
            "100/100 [==============================] - 3s 35ms/step - loss: 2.9340 - acc: 0.2765 - val_loss: 3.0999 - val_acc: 0.2457\n",
            "Epoch 2/500\n",
            "100/100 [==============================] - 5s 45ms/step - loss: 2.9341 - acc: 0.2761 - val_loss: 3.0991 - val_acc: 0.2456\n",
            "Epoch 3/500\n",
            "100/100 [==============================] - 5s 48ms/step - loss: 2.9335 - acc: 0.2771 - val_loss: 3.0993 - val_acc: 0.2463\n",
            "Epoch 4/500\n",
            "100/100 [==============================] - 5s 52ms/step - loss: 2.9341 - acc: 0.2765 - val_loss: 3.1017 - val_acc: 0.2452\n",
            "Epoch 5/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.9339 - acc: 0.2762 - val_loss: 3.0989 - val_acc: 0.2461\n",
            "Epoch 6/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9328 - acc: 0.2772 - val_loss: 3.0988 - val_acc: 0.2451\n",
            "Epoch 7/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9321 - acc: 0.2768 - val_loss: 3.0998 - val_acc: 0.2454\n",
            "Epoch 8/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9324 - acc: 0.2764 - val_loss: 3.0983 - val_acc: 0.2459\n",
            "Epoch 9/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9315 - acc: 0.2768 - val_loss: 3.0980 - val_acc: 0.2464\n",
            "Epoch 10/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9311 - acc: 0.2778 - val_loss: 3.0980 - val_acc: 0.2470\n",
            "Epoch 11/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.9318 - acc: 0.2768 - val_loss: 3.1000 - val_acc: 0.2460\n",
            "Epoch 12/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.9313 - acc: 0.2772 - val_loss: 3.0970 - val_acc: 0.2463\n",
            "Epoch 13/500\n",
            "100/100 [==============================] - 2s 25ms/step - loss: 2.9307 - acc: 0.2772 - val_loss: 3.0979 - val_acc: 0.2452\n",
            "Epoch 14/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9309 - acc: 0.2784 - val_loss: 3.0979 - val_acc: 0.2465\n",
            "Epoch 15/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9302 - acc: 0.2770 - val_loss: 3.0982 - val_acc: 0.2462\n",
            "Epoch 16/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.9298 - acc: 0.2779 - val_loss: 3.0967 - val_acc: 0.2472\n",
            "Epoch 17/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9295 - acc: 0.2772 - val_loss: 3.0967 - val_acc: 0.2472\n",
            "Epoch 18/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9292 - acc: 0.2771 - val_loss: 3.0968 - val_acc: 0.2459\n",
            "Epoch 19/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9296 - acc: 0.2776 - val_loss: 3.0965 - val_acc: 0.2474\n",
            "Epoch 20/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.9293 - acc: 0.2774 - val_loss: 3.0967 - val_acc: 0.2469\n",
            "Epoch 21/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.9291 - acc: 0.2777 - val_loss: 3.0963 - val_acc: 0.2458\n",
            "Epoch 22/500\n",
            "100/100 [==============================] - 3s 34ms/step - loss: 2.9281 - acc: 0.2779 - val_loss: 3.0949 - val_acc: 0.2474\n",
            "Epoch 23/500\n",
            "100/100 [==============================] - 2s 25ms/step - loss: 2.9282 - acc: 0.2771 - val_loss: 3.0951 - val_acc: 0.2455\n",
            "Epoch 24/500\n",
            "100/100 [==============================] - 3s 29ms/step - loss: 2.9285 - acc: 0.2782 - val_loss: 3.0984 - val_acc: 0.2455\n",
            "Epoch 25/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.9281 - acc: 0.2776 - val_loss: 3.0948 - val_acc: 0.2468\n",
            "Epoch 26/500\n",
            "100/100 [==============================] - 2s 25ms/step - loss: 2.9271 - acc: 0.2775 - val_loss: 3.0957 - val_acc: 0.2474\n",
            "Epoch 27/500\n",
            "100/100 [==============================] - 2s 25ms/step - loss: 2.9275 - acc: 0.2774 - val_loss: 3.0948 - val_acc: 0.2474\n",
            "Epoch 28/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9271 - acc: 0.2768 - val_loss: 3.0984 - val_acc: 0.2462\n",
            "Epoch 29/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9271 - acc: 0.2779 - val_loss: 3.0973 - val_acc: 0.2458\n",
            "Epoch 30/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9266 - acc: 0.2782 - val_loss: 3.0950 - val_acc: 0.2474\n",
            "Epoch 31/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9262 - acc: 0.2779 - val_loss: 3.0969 - val_acc: 0.2473\n",
            "Epoch 32/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9263 - acc: 0.2788 - val_loss: 3.0940 - val_acc: 0.2462\n",
            "Epoch 33/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.9257 - acc: 0.2773 - val_loss: 3.0935 - val_acc: 0.2474\n",
            "Epoch 34/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.9256 - acc: 0.2772 - val_loss: 3.0935 - val_acc: 0.2461\n",
            "Epoch 35/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9250 - acc: 0.2785 - val_loss: 3.0926 - val_acc: 0.2473\n",
            "Epoch 36/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9249 - acc: 0.2783 - val_loss: 3.0932 - val_acc: 0.2465\n",
            "Epoch 37/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9254 - acc: 0.2780 - val_loss: 3.0933 - val_acc: 0.2470\n",
            "Epoch 38/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.9244 - acc: 0.2777 - val_loss: 3.0926 - val_acc: 0.2472\n",
            "Epoch 39/500\n",
            "100/100 [==============================] - 2s 25ms/step - loss: 2.9245 - acc: 0.2781 - val_loss: 3.0926 - val_acc: 0.2476\n",
            "Epoch 40/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.9241 - acc: 0.2773 - val_loss: 3.0925 - val_acc: 0.2462\n",
            "Epoch 41/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9232 - acc: 0.2780 - val_loss: 3.0933 - val_acc: 0.2472\n",
            "Epoch 42/500\n",
            "100/100 [==============================] - 3s 28ms/step - loss: 2.9230 - acc: 0.2783 - val_loss: 3.0927 - val_acc: 0.2466\n",
            "Epoch 43/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9234 - acc: 0.2787 - val_loss: 3.0918 - val_acc: 0.2478\n",
            "Epoch 44/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9230 - acc: 0.2789 - val_loss: 3.0917 - val_acc: 0.2472\n",
            "Epoch 45/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9224 - acc: 0.2786 - val_loss: 3.0934 - val_acc: 0.2463\n",
            "Epoch 46/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9221 - acc: 0.2783 - val_loss: 3.0933 - val_acc: 0.2475\n",
            "Epoch 47/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9221 - acc: 0.2783 - val_loss: 3.0910 - val_acc: 0.2460\n",
            "Epoch 48/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9218 - acc: 0.2790 - val_loss: 3.0944 - val_acc: 0.2462\n",
            "Epoch 49/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9216 - acc: 0.2785 - val_loss: 3.0916 - val_acc: 0.2468\n",
            "Epoch 50/500\n",
            "100/100 [==============================] - 3s 29ms/step - loss: 2.9213 - acc: 0.2784 - val_loss: 3.0915 - val_acc: 0.2469\n",
            "Epoch 51/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.9216 - acc: 0.2788 - val_loss: 3.0948 - val_acc: 0.2453\n",
            "Epoch 52/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9213 - acc: 0.2786 - val_loss: 3.0899 - val_acc: 0.2484\n",
            "Epoch 53/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9200 - acc: 0.2784 - val_loss: 3.0905 - val_acc: 0.2458\n",
            "Epoch 54/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9200 - acc: 0.2792 - val_loss: 3.0892 - val_acc: 0.2479\n",
            "Epoch 55/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9197 - acc: 0.2790 - val_loss: 3.0905 - val_acc: 0.2472\n",
            "Epoch 56/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9199 - acc: 0.2792 - val_loss: 3.0901 - val_acc: 0.2470\n",
            "Epoch 57/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9187 - acc: 0.2793 - val_loss: 3.0893 - val_acc: 0.2475\n",
            "Epoch 58/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.9192 - acc: 0.2789 - val_loss: 3.0897 - val_acc: 0.2479\n",
            "Epoch 59/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.9187 - acc: 0.2783 - val_loss: 3.0910 - val_acc: 0.2461\n",
            "Epoch 60/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9191 - acc: 0.2793 - val_loss: 3.0898 - val_acc: 0.2470\n",
            "Epoch 61/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9192 - acc: 0.2791 - val_loss: 3.0906 - val_acc: 0.2480\n",
            "Epoch 62/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9187 - acc: 0.2794 - val_loss: 3.0884 - val_acc: 0.2496\n",
            "Epoch 63/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.9176 - acc: 0.2796 - val_loss: 3.0886 - val_acc: 0.2474\n",
            "Epoch 64/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9178 - acc: 0.2795 - val_loss: 3.0874 - val_acc: 0.2486\n",
            "Epoch 65/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9173 - acc: 0.2788 - val_loss: 3.0886 - val_acc: 0.2469\n",
            "Epoch 66/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.9173 - acc: 0.2791 - val_loss: 3.0872 - val_acc: 0.2487\n",
            "Epoch 67/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9171 - acc: 0.2799 - val_loss: 3.0869 - val_acc: 0.2485\n",
            "Epoch 68/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.9171 - acc: 0.2799 - val_loss: 3.0878 - val_acc: 0.2492\n",
            "Epoch 69/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9168 - acc: 0.2793 - val_loss: 3.0877 - val_acc: 0.2471\n",
            "Epoch 70/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9157 - acc: 0.2793 - val_loss: 3.0861 - val_acc: 0.2480\n",
            "Epoch 71/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9161 - acc: 0.2791 - val_loss: 3.0869 - val_acc: 0.2466\n",
            "Epoch 72/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.9157 - acc: 0.2800 - val_loss: 3.0868 - val_acc: 0.2484\n",
            "Epoch 73/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9159 - acc: 0.2801 - val_loss: 3.0865 - val_acc: 0.2481\n",
            "Epoch 74/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9156 - acc: 0.2796 - val_loss: 3.0868 - val_acc: 0.2481\n",
            "Epoch 75/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9153 - acc: 0.2799 - val_loss: 3.0866 - val_acc: 0.2484\n",
            "Epoch 76/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.9148 - acc: 0.2797 - val_loss: 3.0858 - val_acc: 0.2477\n",
            "Epoch 77/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9145 - acc: 0.2796 - val_loss: 3.0845 - val_acc: 0.2479\n",
            "Epoch 78/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.9144 - acc: 0.2790 - val_loss: 3.0863 - val_acc: 0.2487\n",
            "Epoch 79/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9136 - acc: 0.2790 - val_loss: 3.0853 - val_acc: 0.2484\n",
            "Epoch 80/500\n",
            "100/100 [==============================] - 3s 29ms/step - loss: 2.9147 - acc: 0.2800 - val_loss: 3.0856 - val_acc: 0.2480\n",
            "Epoch 81/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9137 - acc: 0.2802 - val_loss: 3.0844 - val_acc: 0.2480\n",
            "Epoch 82/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9132 - acc: 0.2803 - val_loss: 3.0858 - val_acc: 0.2493\n",
            "Epoch 83/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9128 - acc: 0.2817 - val_loss: 3.0844 - val_acc: 0.2491\n",
            "Epoch 84/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9123 - acc: 0.2806 - val_loss: 3.0850 - val_acc: 0.2492\n",
            "Epoch 85/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.9128 - acc: 0.2807 - val_loss: 3.0849 - val_acc: 0.2482\n",
            "Epoch 86/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9120 - acc: 0.2803 - val_loss: 3.0844 - val_acc: 0.2489\n",
            "Epoch 87/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9117 - acc: 0.2793 - val_loss: 3.0850 - val_acc: 0.2481\n",
            "Epoch 88/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.9119 - acc: 0.2808 - val_loss: 3.0873 - val_acc: 0.2474\n",
            "Epoch 89/500\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 2.9124 - acc: 0.2803 - val_loss: 3.0834 - val_acc: 0.2473\n",
            "Epoch 90/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9112 - acc: 0.2824 - val_loss: 3.0823 - val_acc: 0.2493\n",
            "Epoch 91/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9104 - acc: 0.2814 - val_loss: 3.0844 - val_acc: 0.2474\n",
            "Epoch 92/500\n",
            "100/100 [==============================] - 2s 25ms/step - loss: 2.9106 - acc: 0.2810 - val_loss: 3.0835 - val_acc: 0.2478\n",
            "Epoch 93/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9108 - acc: 0.2807 - val_loss: 3.0864 - val_acc: 0.2481\n",
            "Epoch 94/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9102 - acc: 0.2807 - val_loss: 3.0835 - val_acc: 0.2497\n",
            "Epoch 95/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9102 - acc: 0.2808 - val_loss: 3.0823 - val_acc: 0.2487\n",
            "Epoch 96/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9101 - acc: 0.2808 - val_loss: 3.0818 - val_acc: 0.2500\n",
            "Epoch 97/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9096 - acc: 0.2810 - val_loss: 3.0822 - val_acc: 0.2491\n",
            "Epoch 98/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9088 - acc: 0.2814 - val_loss: 3.0814 - val_acc: 0.2496\n",
            "Epoch 99/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.9089 - acc: 0.2807 - val_loss: 3.0812 - val_acc: 0.2475\n",
            "Epoch 100/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9086 - acc: 0.2810 - val_loss: 3.0826 - val_acc: 0.2484\n",
            "Epoch 101/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9082 - acc: 0.2809 - val_loss: 3.0818 - val_acc: 0.2499\n",
            "Epoch 102/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.9080 - acc: 0.2818 - val_loss: 3.0805 - val_acc: 0.2514\n",
            "Epoch 103/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9082 - acc: 0.2797 - val_loss: 3.0822 - val_acc: 0.2487\n",
            "Epoch 104/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9073 - acc: 0.2808 - val_loss: 3.0806 - val_acc: 0.2513\n",
            "Epoch 105/500\n",
            "100/100 [==============================] - 3s 29ms/step - loss: 2.9081 - acc: 0.2808 - val_loss: 3.0864 - val_acc: 0.2448\n",
            "Epoch 106/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.9071 - acc: 0.2815 - val_loss: 3.0807 - val_acc: 0.2500\n",
            "Epoch 107/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.9066 - acc: 0.2819 - val_loss: 3.0818 - val_acc: 0.2487\n",
            "Epoch 108/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9068 - acc: 0.2816 - val_loss: 3.0820 - val_acc: 0.2494\n",
            "Epoch 109/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9065 - acc: 0.2812 - val_loss: 3.0791 - val_acc: 0.2504\n",
            "Epoch 110/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.9062 - acc: 0.2816 - val_loss: 3.0817 - val_acc: 0.2485\n",
            "Epoch 111/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9063 - acc: 0.2821 - val_loss: 3.0803 - val_acc: 0.2497\n",
            "Epoch 112/500\n",
            "100/100 [==============================] - 2s 25ms/step - loss: 2.9055 - acc: 0.2809 - val_loss: 3.0808 - val_acc: 0.2488\n",
            "Epoch 113/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9051 - acc: 0.2821 - val_loss: 3.0789 - val_acc: 0.2499\n",
            "Epoch 114/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9049 - acc: 0.2819 - val_loss: 3.0783 - val_acc: 0.2493\n",
            "Epoch 115/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9046 - acc: 0.2815 - val_loss: 3.0784 - val_acc: 0.2490\n",
            "Epoch 116/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.9053 - acc: 0.2823 - val_loss: 3.0798 - val_acc: 0.2495\n",
            "Epoch 117/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9042 - acc: 0.2817 - val_loss: 3.0799 - val_acc: 0.2486\n",
            "Epoch 118/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9044 - acc: 0.2820 - val_loss: 3.0780 - val_acc: 0.2502\n",
            "Epoch 119/500\n",
            "100/100 [==============================] - 3s 28ms/step - loss: 2.9042 - acc: 0.2824 - val_loss: 3.0799 - val_acc: 0.2477\n",
            "Epoch 120/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9038 - acc: 0.2819 - val_loss: 3.0786 - val_acc: 0.2506\n",
            "Epoch 121/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9032 - acc: 0.2822 - val_loss: 3.0770 - val_acc: 0.2512\n",
            "Epoch 122/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9035 - acc: 0.2824 - val_loss: 3.0770 - val_acc: 0.2489\n",
            "Epoch 123/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9028 - acc: 0.2818 - val_loss: 3.0775 - val_acc: 0.2495\n",
            "Epoch 124/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9024 - acc: 0.2823 - val_loss: 3.0769 - val_acc: 0.2482\n",
            "Epoch 125/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9021 - acc: 0.2820 - val_loss: 3.0777 - val_acc: 0.2507\n",
            "Epoch 126/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9022 - acc: 0.2818 - val_loss: 3.0767 - val_acc: 0.2505\n",
            "Epoch 127/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.9021 - acc: 0.2825 - val_loss: 3.0784 - val_acc: 0.2496\n",
            "Epoch 128/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.9020 - acc: 0.2820 - val_loss: 3.0778 - val_acc: 0.2513\n",
            "Epoch 129/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.9012 - acc: 0.2831 - val_loss: 3.0754 - val_acc: 0.2507\n",
            "Epoch 130/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9013 - acc: 0.2819 - val_loss: 3.0781 - val_acc: 0.2501\n",
            "Epoch 131/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9012 - acc: 0.2828 - val_loss: 3.0767 - val_acc: 0.2486\n",
            "Epoch 132/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.9007 - acc: 0.2822 - val_loss: 3.0763 - val_acc: 0.2485\n",
            "Epoch 133/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9009 - acc: 0.2827 - val_loss: 3.0782 - val_acc: 0.2490\n",
            "Epoch 134/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.9008 - acc: 0.2819 - val_loss: 3.0762 - val_acc: 0.2497\n",
            "Epoch 135/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.9002 - acc: 0.2828 - val_loss: 3.0754 - val_acc: 0.2501\n",
            "Epoch 136/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.9000 - acc: 0.2829 - val_loss: 3.0763 - val_acc: 0.2494\n",
            "Epoch 137/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.9006 - acc: 0.2827 - val_loss: 3.0750 - val_acc: 0.2497\n",
            "Epoch 138/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.8998 - acc: 0.2826 - val_loss: 3.0757 - val_acc: 0.2501\n",
            "Epoch 139/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.8994 - acc: 0.2829 - val_loss: 3.0771 - val_acc: 0.2497\n",
            "Epoch 140/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8994 - acc: 0.2828 - val_loss: 3.0738 - val_acc: 0.2508\n",
            "Epoch 141/500\n",
            "100/100 [==============================] - 3s 29ms/step - loss: 2.8985 - acc: 0.2828 - val_loss: 3.0745 - val_acc: 0.2510\n",
            "Epoch 142/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.8993 - acc: 0.2824 - val_loss: 3.0748 - val_acc: 0.2497\n",
            "Epoch 143/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8993 - acc: 0.2828 - val_loss: 3.0750 - val_acc: 0.2501\n",
            "Epoch 144/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8978 - acc: 0.2826 - val_loss: 3.0744 - val_acc: 0.2507\n",
            "Epoch 145/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8978 - acc: 0.2837 - val_loss: 3.0772 - val_acc: 0.2489\n",
            "Epoch 146/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.8980 - acc: 0.2831 - val_loss: 3.0743 - val_acc: 0.2496\n",
            "Epoch 147/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.8976 - acc: 0.2829 - val_loss: 3.0731 - val_acc: 0.2508\n",
            "Epoch 148/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8968 - acc: 0.2840 - val_loss: 3.0735 - val_acc: 0.2493\n",
            "Epoch 149/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8970 - acc: 0.2829 - val_loss: 3.0753 - val_acc: 0.2482\n",
            "Epoch 150/500\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 2.8972 - acc: 0.2829 - val_loss: 3.0746 - val_acc: 0.2509\n",
            "Epoch 151/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8960 - acc: 0.2831 - val_loss: 3.0742 - val_acc: 0.2496\n",
            "Epoch 152/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.8956 - acc: 0.2832 - val_loss: 3.0720 - val_acc: 0.2512\n",
            "Epoch 153/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8956 - acc: 0.2838 - val_loss: 3.0723 - val_acc: 0.2519\n",
            "Epoch 154/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8956 - acc: 0.2840 - val_loss: 3.0722 - val_acc: 0.2514\n",
            "Epoch 155/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.8955 - acc: 0.2837 - val_loss: 3.0726 - val_acc: 0.2491\n",
            "Epoch 156/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8954 - acc: 0.2837 - val_loss: 3.0740 - val_acc: 0.2489\n",
            "Epoch 157/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8950 - acc: 0.2835 - val_loss: 3.0719 - val_acc: 0.2522\n",
            "Epoch 158/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8949 - acc: 0.2827 - val_loss: 3.0725 - val_acc: 0.2513\n",
            "Epoch 159/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8950 - acc: 0.2840 - val_loss: 3.0719 - val_acc: 0.2513\n",
            "Epoch 160/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.8945 - acc: 0.2837 - val_loss: 3.0713 - val_acc: 0.2513\n",
            "Epoch 161/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8941 - acc: 0.2842 - val_loss: 3.0706 - val_acc: 0.2496\n",
            "Epoch 162/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8933 - acc: 0.2837 - val_loss: 3.0719 - val_acc: 0.2502\n",
            "Epoch 163/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8938 - acc: 0.2841 - val_loss: 3.0707 - val_acc: 0.2506\n",
            "Epoch 164/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8933 - acc: 0.2833 - val_loss: 3.0720 - val_acc: 0.2518\n",
            "Epoch 165/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.8928 - acc: 0.2834 - val_loss: 3.0710 - val_acc: 0.2502\n",
            "Epoch 166/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8928 - acc: 0.2846 - val_loss: 3.0700 - val_acc: 0.2530\n",
            "Epoch 167/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8927 - acc: 0.2844 - val_loss: 3.0711 - val_acc: 0.2495\n",
            "Epoch 168/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8930 - acc: 0.2838 - val_loss: 3.0712 - val_acc: 0.2519\n",
            "Epoch 169/500\n",
            "100/100 [==============================] - 2s 25ms/step - loss: 2.8919 - acc: 0.2842 - val_loss: 3.0726 - val_acc: 0.2482\n",
            "Epoch 170/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8919 - acc: 0.2844 - val_loss: 3.0695 - val_acc: 0.2509\n",
            "Epoch 171/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8917 - acc: 0.2844 - val_loss: 3.0699 - val_acc: 0.2524\n",
            "Epoch 172/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8913 - acc: 0.2837 - val_loss: 3.0702 - val_acc: 0.2502\n",
            "Epoch 173/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.8910 - acc: 0.2843 - val_loss: 3.0696 - val_acc: 0.2513\n",
            "Epoch 174/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8904 - acc: 0.2842 - val_loss: 3.0689 - val_acc: 0.2512\n",
            "Epoch 175/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8906 - acc: 0.2844 - val_loss: 3.0687 - val_acc: 0.2512\n",
            "Epoch 176/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8901 - acc: 0.2850 - val_loss: 3.0685 - val_acc: 0.2514\n",
            "Epoch 177/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8901 - acc: 0.2849 - val_loss: 3.0691 - val_acc: 0.2516\n",
            "Epoch 178/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.8903 - acc: 0.2831 - val_loss: 3.0678 - val_acc: 0.2520\n",
            "Epoch 179/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8893 - acc: 0.2848 - val_loss: 3.0691 - val_acc: 0.2510\n",
            "Epoch 180/500\n",
            "100/100 [==============================] - 3s 28ms/step - loss: 2.8899 - acc: 0.2842 - val_loss: 3.0670 - val_acc: 0.2516\n",
            "Epoch 181/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.8891 - acc: 0.2858 - val_loss: 3.0682 - val_acc: 0.2503\n",
            "Epoch 182/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.8889 - acc: 0.2847 - val_loss: 3.0680 - val_acc: 0.2516\n",
            "Epoch 183/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8886 - acc: 0.2850 - val_loss: 3.0694 - val_acc: 0.2509\n",
            "Epoch 184/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8885 - acc: 0.2845 - val_loss: 3.0682 - val_acc: 0.2514\n",
            "Epoch 185/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.8882 - acc: 0.2846 - val_loss: 3.0668 - val_acc: 0.2529\n",
            "Epoch 186/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8884 - acc: 0.2846 - val_loss: 3.0675 - val_acc: 0.2528\n",
            "Epoch 187/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8880 - acc: 0.2841 - val_loss: 3.0665 - val_acc: 0.2507\n",
            "Epoch 188/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8879 - acc: 0.2853 - val_loss: 3.0683 - val_acc: 0.2520\n",
            "Epoch 189/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8879 - acc: 0.2850 - val_loss: 3.0696 - val_acc: 0.2498\n",
            "Epoch 190/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8876 - acc: 0.2851 - val_loss: 3.0685 - val_acc: 0.2508\n",
            "Epoch 191/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8873 - acc: 0.2848 - val_loss: 3.0650 - val_acc: 0.2530\n",
            "Epoch 192/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8861 - acc: 0.2847 - val_loss: 3.0675 - val_acc: 0.2522\n",
            "Epoch 193/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8860 - acc: 0.2865 - val_loss: 3.0652 - val_acc: 0.2513\n",
            "Epoch 194/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8861 - acc: 0.2851 - val_loss: 3.0661 - val_acc: 0.2520\n",
            "Epoch 195/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8857 - acc: 0.2855 - val_loss: 3.0661 - val_acc: 0.2522\n",
            "Epoch 196/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.8859 - acc: 0.2855 - val_loss: 3.0644 - val_acc: 0.2520\n",
            "Epoch 197/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8857 - acc: 0.2853 - val_loss: 3.0651 - val_acc: 0.2525\n",
            "Epoch 198/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.8849 - acc: 0.2853 - val_loss: 3.0658 - val_acc: 0.2530\n",
            "Epoch 199/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.8845 - acc: 0.2855 - val_loss: 3.0654 - val_acc: 0.2522\n",
            "Epoch 200/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8844 - acc: 0.2858 - val_loss: 3.0655 - val_acc: 0.2511\n",
            "Epoch 201/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8851 - acc: 0.2858 - val_loss: 3.0649 - val_acc: 0.2519\n",
            "Epoch 202/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8846 - acc: 0.2856 - val_loss: 3.0645 - val_acc: 0.2513\n",
            "Epoch 203/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8842 - acc: 0.2859 - val_loss: 3.0653 - val_acc: 0.2519\n",
            "Epoch 204/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8834 - acc: 0.2852 - val_loss: 3.0635 - val_acc: 0.2535\n",
            "Epoch 205/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8836 - acc: 0.2856 - val_loss: 3.0659 - val_acc: 0.2517\n",
            "Epoch 206/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8831 - acc: 0.2862 - val_loss: 3.0637 - val_acc: 0.2525\n",
            "Epoch 207/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8833 - acc: 0.2859 - val_loss: 3.0663 - val_acc: 0.2498\n",
            "Epoch 208/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.8824 - acc: 0.2855 - val_loss: 3.0641 - val_acc: 0.2518\n",
            "Epoch 209/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8821 - acc: 0.2858 - val_loss: 3.0626 - val_acc: 0.2530\n",
            "Epoch 210/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8825 - acc: 0.2863 - val_loss: 3.0633 - val_acc: 0.2518\n",
            "Epoch 211/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.8822 - acc: 0.2870 - val_loss: 3.0625 - val_acc: 0.2523\n",
            "Epoch 212/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8822 - acc: 0.2867 - val_loss: 3.0631 - val_acc: 0.2527\n",
            "Epoch 213/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8818 - acc: 0.2865 - val_loss: 3.0629 - val_acc: 0.2531\n",
            "Epoch 214/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8815 - acc: 0.2866 - val_loss: 3.0629 - val_acc: 0.2530\n",
            "Epoch 215/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8816 - acc: 0.2866 - val_loss: 3.0620 - val_acc: 0.2530\n",
            "Epoch 216/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.8818 - acc: 0.2869 - val_loss: 3.0621 - val_acc: 0.2533\n",
            "Epoch 217/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8807 - acc: 0.2867 - val_loss: 3.0636 - val_acc: 0.2511\n",
            "Epoch 218/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8805 - acc: 0.2870 - val_loss: 3.0617 - val_acc: 0.2531\n",
            "Epoch 219/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.8813 - acc: 0.2862 - val_loss: 3.0618 - val_acc: 0.2520\n",
            "Epoch 220/500\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 2.8801 - acc: 0.2866 - val_loss: 3.0615 - val_acc: 0.2529\n",
            "Epoch 221/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8799 - acc: 0.2871 - val_loss: 3.0605 - val_acc: 0.2527\n",
            "Epoch 222/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8791 - acc: 0.2865 - val_loss: 3.0647 - val_acc: 0.2506\n",
            "Epoch 223/500\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 2.8799 - acc: 0.2867 - val_loss: 3.0605 - val_acc: 0.2526\n",
            "Epoch 224/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8794 - acc: 0.2865 - val_loss: 3.0617 - val_acc: 0.2515\n",
            "Epoch 225/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.8788 - acc: 0.2864 - val_loss: 3.0598 - val_acc: 0.2535\n",
            "Epoch 226/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8786 - acc: 0.2877 - val_loss: 3.0605 - val_acc: 0.2540\n",
            "Epoch 227/500\n",
            "100/100 [==============================] - 3s 29ms/step - loss: 2.8784 - acc: 0.2873 - val_loss: 3.0610 - val_acc: 0.2539\n",
            "Epoch 228/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8782 - acc: 0.2867 - val_loss: 3.0598 - val_acc: 0.2530\n",
            "Epoch 229/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.8778 - acc: 0.2875 - val_loss: 3.0607 - val_acc: 0.2535\n",
            "Epoch 230/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8776 - acc: 0.2865 - val_loss: 3.0605 - val_acc: 0.2529\n",
            "Epoch 231/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8771 - acc: 0.2864 - val_loss: 3.0609 - val_acc: 0.2529\n",
            "Epoch 232/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8776 - acc: 0.2878 - val_loss: 3.0618 - val_acc: 0.2530\n",
            "Epoch 233/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.8771 - acc: 0.2874 - val_loss: 3.0580 - val_acc: 0.2541\n",
            "Epoch 234/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8762 - acc: 0.2875 - val_loss: 3.0583 - val_acc: 0.2540\n",
            "Epoch 235/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.8767 - acc: 0.2873 - val_loss: 3.0605 - val_acc: 0.2530\n",
            "Epoch 236/500\n",
            "100/100 [==============================] - 3s 29ms/step - loss: 2.8767 - acc: 0.2876 - val_loss: 3.0589 - val_acc: 0.2529\n",
            "Epoch 237/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8762 - acc: 0.2871 - val_loss: 3.0613 - val_acc: 0.2516\n",
            "Epoch 238/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8770 - acc: 0.2864 - val_loss: 3.0599 - val_acc: 0.2529\n",
            "Epoch 239/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8760 - acc: 0.2869 - val_loss: 3.0592 - val_acc: 0.2530\n",
            "Epoch 240/500\n",
            "100/100 [==============================] - 3s 29ms/step - loss: 2.8758 - acc: 0.2870 - val_loss: 3.0590 - val_acc: 0.2526\n",
            "Epoch 241/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8761 - acc: 0.2867 - val_loss: 3.0571 - val_acc: 0.2558\n",
            "Epoch 242/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8749 - acc: 0.2882 - val_loss: 3.0577 - val_acc: 0.2541\n",
            "Epoch 243/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8748 - acc: 0.2878 - val_loss: 3.0571 - val_acc: 0.2524\n",
            "Epoch 244/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8746 - acc: 0.2883 - val_loss: 3.0570 - val_acc: 0.2532\n",
            "Epoch 245/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8745 - acc: 0.2875 - val_loss: 3.0566 - val_acc: 0.2527\n",
            "Epoch 246/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8739 - acc: 0.2877 - val_loss: 3.0574 - val_acc: 0.2514\n",
            "Epoch 247/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8745 - acc: 0.2879 - val_loss: 3.0572 - val_acc: 0.2521\n",
            "Epoch 248/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8742 - acc: 0.2883 - val_loss: 3.0576 - val_acc: 0.2531\n",
            "Epoch 249/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.8731 - acc: 0.2873 - val_loss: 3.0568 - val_acc: 0.2545\n",
            "Epoch 250/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.8733 - acc: 0.2882 - val_loss: 3.0572 - val_acc: 0.2536\n",
            "Epoch 251/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8734 - acc: 0.2883 - val_loss: 3.0559 - val_acc: 0.2547\n",
            "Epoch 252/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8730 - acc: 0.2880 - val_loss: 3.0554 - val_acc: 0.2527\n",
            "Epoch 253/500\n",
            "100/100 [==============================] - 3s 29ms/step - loss: 2.8730 - acc: 0.2883 - val_loss: 3.0555 - val_acc: 0.2545\n",
            "Epoch 254/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8724 - acc: 0.2868 - val_loss: 3.0551 - val_acc: 0.2556\n",
            "Epoch 255/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.8725 - acc: 0.2880 - val_loss: 3.0550 - val_acc: 0.2521\n",
            "Epoch 256/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8717 - acc: 0.2882 - val_loss: 3.0593 - val_acc: 0.2524\n",
            "Epoch 257/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8719 - acc: 0.2885 - val_loss: 3.0549 - val_acc: 0.2545\n",
            "Epoch 258/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8715 - acc: 0.2882 - val_loss: 3.0561 - val_acc: 0.2546\n",
            "Epoch 259/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8714 - acc: 0.2877 - val_loss: 3.0567 - val_acc: 0.2537\n",
            "Epoch 260/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8706 - acc: 0.2890 - val_loss: 3.0540 - val_acc: 0.2541\n",
            "Epoch 261/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8716 - acc: 0.2888 - val_loss: 3.0546 - val_acc: 0.2538\n",
            "Epoch 262/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8704 - acc: 0.2885 - val_loss: 3.0553 - val_acc: 0.2528\n",
            "Epoch 263/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8702 - acc: 0.2885 - val_loss: 3.0548 - val_acc: 0.2540\n",
            "Epoch 264/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8706 - acc: 0.2880 - val_loss: 3.0532 - val_acc: 0.2541\n",
            "Epoch 265/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8694 - acc: 0.2887 - val_loss: 3.0535 - val_acc: 0.2547\n",
            "Epoch 266/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8694 - acc: 0.2882 - val_loss: 3.0565 - val_acc: 0.2536\n",
            "Epoch 267/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8691 - acc: 0.2888 - val_loss: 3.0545 - val_acc: 0.2528\n",
            "Epoch 268/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8695 - acc: 0.2882 - val_loss: 3.0529 - val_acc: 0.2541\n",
            "Epoch 269/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8693 - acc: 0.2883 - val_loss: 3.0537 - val_acc: 0.2540\n",
            "Epoch 270/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8685 - acc: 0.2885 - val_loss: 3.0531 - val_acc: 0.2534\n",
            "Epoch 271/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8684 - acc: 0.2886 - val_loss: 3.0534 - val_acc: 0.2530\n",
            "Epoch 272/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8679 - acc: 0.2892 - val_loss: 3.0527 - val_acc: 0.2543\n",
            "Epoch 273/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8682 - acc: 0.2881 - val_loss: 3.0518 - val_acc: 0.2550\n",
            "Epoch 274/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8681 - acc: 0.2883 - val_loss: 3.0521 - val_acc: 0.2536\n",
            "Epoch 275/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.8673 - acc: 0.2891 - val_loss: 3.0550 - val_acc: 0.2525\n",
            "Epoch 276/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8674 - acc: 0.2892 - val_loss: 3.0538 - val_acc: 0.2529\n",
            "Epoch 277/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8679 - acc: 0.2893 - val_loss: 3.0531 - val_acc: 0.2530\n",
            "Epoch 278/500\n",
            "100/100 [==============================] - 3s 34ms/step - loss: 2.8684 - acc: 0.2894 - val_loss: 3.0521 - val_acc: 0.2544\n",
            "Epoch 279/500\n",
            "100/100 [==============================] - 3s 34ms/step - loss: 2.8665 - acc: 0.2894 - val_loss: 3.0527 - val_acc: 0.2550\n",
            "Epoch 280/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8665 - acc: 0.2891 - val_loss: 3.0522 - val_acc: 0.2537\n",
            "Epoch 281/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8659 - acc: 0.2887 - val_loss: 3.0525 - val_acc: 0.2536\n",
            "Epoch 282/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8665 - acc: 0.2890 - val_loss: 3.0518 - val_acc: 0.2537\n",
            "Epoch 283/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8664 - acc: 0.2895 - val_loss: 3.0518 - val_acc: 0.2541\n",
            "Epoch 284/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8653 - acc: 0.2896 - val_loss: 3.0503 - val_acc: 0.2548\n",
            "Epoch 285/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8656 - acc: 0.2885 - val_loss: 3.0509 - val_acc: 0.2539\n",
            "Epoch 286/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8650 - acc: 0.2898 - val_loss: 3.0526 - val_acc: 0.2552\n",
            "Epoch 287/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8651 - acc: 0.2891 - val_loss: 3.0563 - val_acc: 0.2525\n",
            "Epoch 288/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.8652 - acc: 0.2895 - val_loss: 3.0488 - val_acc: 0.2555\n",
            "Epoch 289/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8640 - acc: 0.2900 - val_loss: 3.0525 - val_acc: 0.2530\n",
            "Epoch 290/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8645 - acc: 0.2894 - val_loss: 3.0506 - val_acc: 0.2552\n",
            "Epoch 291/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8644 - acc: 0.2897 - val_loss: 3.0486 - val_acc: 0.2556\n",
            "Epoch 292/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8632 - acc: 0.2896 - val_loss: 3.0503 - val_acc: 0.2534\n",
            "Epoch 293/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8641 - acc: 0.2893 - val_loss: 3.0498 - val_acc: 0.2559\n",
            "Epoch 294/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8637 - acc: 0.2903 - val_loss: 3.0487 - val_acc: 0.2541\n",
            "Epoch 295/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8640 - acc: 0.2904 - val_loss: 3.0488 - val_acc: 0.2531\n",
            "Epoch 296/500\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 2.8630 - acc: 0.2898 - val_loss: 3.0489 - val_acc: 0.2550\n",
            "Epoch 297/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8626 - acc: 0.2898 - val_loss: 3.0517 - val_acc: 0.2541\n",
            "Epoch 298/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8628 - acc: 0.2904 - val_loss: 3.0479 - val_acc: 0.2539\n",
            "Epoch 299/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8621 - acc: 0.2893 - val_loss: 3.0488 - val_acc: 0.2534\n",
            "Epoch 300/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8625 - acc: 0.2901 - val_loss: 3.0490 - val_acc: 0.2539\n",
            "Epoch 301/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8615 - acc: 0.2899 - val_loss: 3.0490 - val_acc: 0.2556\n",
            "Epoch 302/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8617 - acc: 0.2903 - val_loss: 3.0476 - val_acc: 0.2547\n",
            "Epoch 303/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8619 - acc: 0.2900 - val_loss: 3.0486 - val_acc: 0.2544\n",
            "Epoch 304/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8608 - acc: 0.2904 - val_loss: 3.0471 - val_acc: 0.2548\n",
            "Epoch 305/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8610 - acc: 0.2903 - val_loss: 3.0501 - val_acc: 0.2542\n",
            "Epoch 306/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8614 - acc: 0.2905 - val_loss: 3.0480 - val_acc: 0.2546\n",
            "Epoch 307/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8607 - acc: 0.2898 - val_loss: 3.0471 - val_acc: 0.2563\n",
            "Epoch 308/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8605 - acc: 0.2903 - val_loss: 3.0469 - val_acc: 0.2560\n",
            "Epoch 309/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8601 - acc: 0.2910 - val_loss: 3.0491 - val_acc: 0.2546\n",
            "Epoch 310/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8606 - acc: 0.2903 - val_loss: 3.0471 - val_acc: 0.2557\n",
            "Epoch 311/500\n",
            "100/100 [==============================] - 3s 25ms/step - loss: 2.8593 - acc: 0.2912 - val_loss: 3.0469 - val_acc: 0.2550\n",
            "Epoch 312/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8596 - acc: 0.2903 - val_loss: 3.0475 - val_acc: 0.2551\n",
            "Epoch 313/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8592 - acc: 0.2899 - val_loss: 3.0456 - val_acc: 0.2542\n",
            "Epoch 314/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8589 - acc: 0.2914 - val_loss: 3.0478 - val_acc: 0.2550\n",
            "Epoch 315/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8589 - acc: 0.2901 - val_loss: 3.0465 - val_acc: 0.2538\n",
            "Epoch 316/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8585 - acc: 0.2907 - val_loss: 3.0457 - val_acc: 0.2566\n",
            "Epoch 317/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8585 - acc: 0.2908 - val_loss: 3.0456 - val_acc: 0.2563\n",
            "Epoch 318/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.8582 - acc: 0.2900 - val_loss: 3.0488 - val_acc: 0.2542\n",
            "Epoch 319/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8580 - acc: 0.2912 - val_loss: 3.0458 - val_acc: 0.2555\n",
            "Epoch 320/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8575 - acc: 0.2910 - val_loss: 3.0469 - val_acc: 0.2545\n",
            "Epoch 321/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8573 - acc: 0.2916 - val_loss: 3.0451 - val_acc: 0.2542\n",
            "Epoch 322/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.8574 - acc: 0.2912 - val_loss: 3.0477 - val_acc: 0.2548\n",
            "Epoch 323/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8567 - acc: 0.2905 - val_loss: 3.0486 - val_acc: 0.2566\n",
            "Epoch 324/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8575 - acc: 0.2906 - val_loss: 3.0491 - val_acc: 0.2528\n",
            "Epoch 325/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8567 - acc: 0.2910 - val_loss: 3.0449 - val_acc: 0.2551\n",
            "Epoch 326/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8559 - acc: 0.2911 - val_loss: 3.0458 - val_acc: 0.2557\n",
            "Epoch 327/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8561 - acc: 0.2914 - val_loss: 3.0436 - val_acc: 0.2558\n",
            "Epoch 328/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8557 - acc: 0.2921 - val_loss: 3.0481 - val_acc: 0.2551\n",
            "Epoch 329/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8555 - acc: 0.2916 - val_loss: 3.0447 - val_acc: 0.2549\n",
            "Epoch 330/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8558 - acc: 0.2903 - val_loss: 3.0438 - val_acc: 0.2554\n",
            "Epoch 331/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8549 - acc: 0.2917 - val_loss: 3.0465 - val_acc: 0.2549\n",
            "Epoch 332/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8551 - acc: 0.2916 - val_loss: 3.0442 - val_acc: 0.2555\n",
            "Epoch 333/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8549 - acc: 0.2921 - val_loss: 3.0431 - val_acc: 0.2558\n",
            "Epoch 334/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8545 - acc: 0.2922 - val_loss: 3.0440 - val_acc: 0.2550\n",
            "Epoch 335/500\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 2.8541 - acc: 0.2913 - val_loss: 3.0444 - val_acc: 0.2542\n",
            "Epoch 336/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8546 - acc: 0.2918 - val_loss: 3.0419 - val_acc: 0.2548\n",
            "Epoch 337/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8537 - acc: 0.2925 - val_loss: 3.0433 - val_acc: 0.2559\n",
            "Epoch 338/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8543 - acc: 0.2918 - val_loss: 3.0423 - val_acc: 0.2561\n",
            "Epoch 339/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.8538 - acc: 0.2918 - val_loss: 3.0415 - val_acc: 0.2559\n",
            "Epoch 340/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8536 - acc: 0.2913 - val_loss: 3.0420 - val_acc: 0.2567\n",
            "Epoch 341/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8532 - acc: 0.2913 - val_loss: 3.0442 - val_acc: 0.2553\n",
            "Epoch 342/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8528 - acc: 0.2914 - val_loss: 3.0416 - val_acc: 0.2562\n",
            "Epoch 343/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.8528 - acc: 0.2923 - val_loss: 3.0448 - val_acc: 0.2571\n",
            "Epoch 344/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8527 - acc: 0.2923 - val_loss: 3.0425 - val_acc: 0.2550\n",
            "Epoch 345/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8522 - acc: 0.2921 - val_loss: 3.0439 - val_acc: 0.2562\n",
            "Epoch 346/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8525 - acc: 0.2928 - val_loss: 3.0429 - val_acc: 0.2557\n",
            "Epoch 347/500\n",
            "100/100 [==============================] - 3s 28ms/step - loss: 2.8517 - acc: 0.2922 - val_loss: 3.0406 - val_acc: 0.2561\n",
            "Epoch 348/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8521 - acc: 0.2918 - val_loss: 3.0426 - val_acc: 0.2557\n",
            "Epoch 349/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8519 - acc: 0.2919 - val_loss: 3.0417 - val_acc: 0.2559\n",
            "Epoch 350/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8508 - acc: 0.2916 - val_loss: 3.0410 - val_acc: 0.2557\n",
            "Epoch 351/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8509 - acc: 0.2928 - val_loss: 3.0405 - val_acc: 0.2565\n",
            "Epoch 352/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8519 - acc: 0.2926 - val_loss: 3.0407 - val_acc: 0.2559\n",
            "Epoch 353/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8510 - acc: 0.2923 - val_loss: 3.0411 - val_acc: 0.2577\n",
            "Epoch 354/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8509 - acc: 0.2927 - val_loss: 3.0429 - val_acc: 0.2566\n",
            "Epoch 355/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8504 - acc: 0.2920 - val_loss: 3.0422 - val_acc: 0.2565\n",
            "Epoch 356/500\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 2.8500 - acc: 0.2928 - val_loss: 3.0406 - val_acc: 0.2569\n",
            "Epoch 357/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8503 - acc: 0.2923 - val_loss: 3.0405 - val_acc: 0.2558\n",
            "Epoch 358/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8503 - acc: 0.2921 - val_loss: 3.0397 - val_acc: 0.2554\n",
            "Epoch 359/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8496 - acc: 0.2926 - val_loss: 3.0405 - val_acc: 0.2563\n",
            "Epoch 360/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8490 - acc: 0.2930 - val_loss: 3.0379 - val_acc: 0.2588\n",
            "Epoch 361/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8491 - acc: 0.2924 - val_loss: 3.0431 - val_acc: 0.2558\n",
            "Epoch 362/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8484 - acc: 0.2931 - val_loss: 3.0380 - val_acc: 0.2561\n",
            "Epoch 363/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8482 - acc: 0.2929 - val_loss: 3.0403 - val_acc: 0.2565\n",
            "Epoch 364/500\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 2.8482 - acc: 0.2935 - val_loss: 3.0397 - val_acc: 0.2553\n",
            "Epoch 365/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8483 - acc: 0.2937 - val_loss: 3.0396 - val_acc: 0.2558\n",
            "Epoch 366/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8487 - acc: 0.2923 - val_loss: 3.0411 - val_acc: 0.2566\n",
            "Epoch 367/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8469 - acc: 0.2944 - val_loss: 3.0373 - val_acc: 0.2572\n",
            "Epoch 368/500\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 2.8480 - acc: 0.2921 - val_loss: 3.0379 - val_acc: 0.2576\n",
            "Epoch 369/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8466 - acc: 0.2935 - val_loss: 3.0393 - val_acc: 0.2547\n",
            "Epoch 370/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8470 - acc: 0.2934 - val_loss: 3.0386 - val_acc: 0.2561\n",
            "Epoch 371/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8467 - acc: 0.2933 - val_loss: 3.0389 - val_acc: 0.2570\n",
            "Epoch 372/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8469 - acc: 0.2930 - val_loss: 3.0402 - val_acc: 0.2563\n",
            "Epoch 373/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8471 - acc: 0.2930 - val_loss: 3.0375 - val_acc: 0.2558\n",
            "Epoch 374/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8461 - acc: 0.2946 - val_loss: 3.0389 - val_acc: 0.2569\n",
            "Epoch 375/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8456 - acc: 0.2943 - val_loss: 3.0371 - val_acc: 0.2586\n",
            "Epoch 376/500\n",
            "100/100 [==============================] - 3s 29ms/step - loss: 2.8454 - acc: 0.2932 - val_loss: 3.0363 - val_acc: 0.2580\n",
            "Epoch 377/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8467 - acc: 0.2929 - val_loss: 3.0372 - val_acc: 0.2577\n",
            "Epoch 378/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8450 - acc: 0.2930 - val_loss: 3.0375 - val_acc: 0.2562\n",
            "Epoch 379/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8454 - acc: 0.2936 - val_loss: 3.0368 - val_acc: 0.2572\n",
            "Epoch 380/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8455 - acc: 0.2932 - val_loss: 3.0379 - val_acc: 0.2563\n",
            "Epoch 381/500\n",
            "100/100 [==============================] - 3s 33ms/step - loss: 2.8449 - acc: 0.2935 - val_loss: 3.0367 - val_acc: 0.2569\n",
            "Epoch 382/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8452 - acc: 0.2928 - val_loss: 3.0356 - val_acc: 0.2572\n",
            "Epoch 383/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8441 - acc: 0.2941 - val_loss: 3.0368 - val_acc: 0.2589\n",
            "Epoch 384/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.8441 - acc: 0.2932 - val_loss: 3.0375 - val_acc: 0.2574\n",
            "Epoch 385/500\n",
            "100/100 [==============================] - 3s 29ms/step - loss: 2.8444 - acc: 0.2940 - val_loss: 3.0353 - val_acc: 0.2577\n",
            "Epoch 386/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8438 - acc: 0.2942 - val_loss: 3.0362 - val_acc: 0.2583\n",
            "Epoch 387/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8434 - acc: 0.2934 - val_loss: 3.0381 - val_acc: 0.2566\n",
            "Epoch 388/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.8432 - acc: 0.2937 - val_loss: 3.0353 - val_acc: 0.2565\n",
            "Epoch 389/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8432 - acc: 0.2943 - val_loss: 3.0363 - val_acc: 0.2560\n",
            "Epoch 390/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8428 - acc: 0.2942 - val_loss: 3.0373 - val_acc: 0.2575\n",
            "Epoch 391/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8429 - acc: 0.2937 - val_loss: 3.0353 - val_acc: 0.2574\n",
            "Epoch 392/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8424 - acc: 0.2938 - val_loss: 3.0362 - val_acc: 0.2567\n",
            "Epoch 393/500\n",
            "100/100 [==============================] - 3s 34ms/step - loss: 2.8424 - acc: 0.2937 - val_loss: 3.0336 - val_acc: 0.2586\n",
            "Epoch 394/500\n",
            "100/100 [==============================] - 3s 28ms/step - loss: 2.8418 - acc: 0.2946 - val_loss: 3.0349 - val_acc: 0.2580\n",
            "Epoch 395/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8417 - acc: 0.2938 - val_loss: 3.0353 - val_acc: 0.2591\n",
            "Epoch 396/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8420 - acc: 0.2934 - val_loss: 3.0350 - val_acc: 0.2579\n",
            "Epoch 397/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8415 - acc: 0.2946 - val_loss: 3.0336 - val_acc: 0.2580\n",
            "Epoch 398/500\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 2.8418 - acc: 0.2943 - val_loss: 3.0334 - val_acc: 0.2575\n",
            "Epoch 399/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8412 - acc: 0.2946 - val_loss: 3.0333 - val_acc: 0.2585\n",
            "Epoch 400/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.8410 - acc: 0.2946 - val_loss: 3.0344 - val_acc: 0.2574\n",
            "Epoch 401/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8405 - acc: 0.2940 - val_loss: 3.0341 - val_acc: 0.2575\n",
            "Epoch 402/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8410 - acc: 0.2944 - val_loss: 3.0337 - val_acc: 0.2582\n",
            "Epoch 403/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8401 - acc: 0.2938 - val_loss: 3.0338 - val_acc: 0.2586\n",
            "Epoch 404/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8399 - acc: 0.2955 - val_loss: 3.0332 - val_acc: 0.2576\n",
            "Epoch 405/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8392 - acc: 0.2952 - val_loss: 3.0340 - val_acc: 0.2583\n",
            "Epoch 406/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8403 - acc: 0.2940 - val_loss: 3.0418 - val_acc: 0.2551\n",
            "Epoch 407/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8397 - acc: 0.2952 - val_loss: 3.0325 - val_acc: 0.2587\n",
            "Epoch 408/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8390 - acc: 0.2954 - val_loss: 3.0314 - val_acc: 0.2591\n",
            "Epoch 409/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8388 - acc: 0.2938 - val_loss: 3.0329 - val_acc: 0.2563\n",
            "Epoch 410/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8389 - acc: 0.2943 - val_loss: 3.0335 - val_acc: 0.2578\n",
            "Epoch 411/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8384 - acc: 0.2951 - val_loss: 3.0317 - val_acc: 0.2586\n",
            "Epoch 412/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8383 - acc: 0.2951 - val_loss: 3.0340 - val_acc: 0.2578\n",
            "Epoch 413/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8381 - acc: 0.2947 - val_loss: 3.0340 - val_acc: 0.2586\n",
            "Epoch 414/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8380 - acc: 0.2950 - val_loss: 3.0331 - val_acc: 0.2570\n",
            "Epoch 415/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8376 - acc: 0.2949 - val_loss: 3.0327 - val_acc: 0.2583\n",
            "Epoch 416/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8374 - acc: 0.2943 - val_loss: 3.0307 - val_acc: 0.2595\n",
            "Epoch 417/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8379 - acc: 0.2948 - val_loss: 3.0328 - val_acc: 0.2585\n",
            "Epoch 418/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8377 - acc: 0.2950 - val_loss: 3.0309 - val_acc: 0.2580\n",
            "Epoch 419/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8367 - acc: 0.2948 - val_loss: 3.0311 - val_acc: 0.2598\n",
            "Epoch 420/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.8362 - acc: 0.2959 - val_loss: 3.0312 - val_acc: 0.2574\n",
            "Epoch 421/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8365 - acc: 0.2947 - val_loss: 3.0299 - val_acc: 0.2588\n",
            "Epoch 422/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8364 - acc: 0.2952 - val_loss: 3.0338 - val_acc: 0.2565\n",
            "Epoch 423/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8361 - acc: 0.2949 - val_loss: 3.0330 - val_acc: 0.2578\n",
            "Epoch 424/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8359 - acc: 0.2947 - val_loss: 3.0298 - val_acc: 0.2580\n",
            "Epoch 425/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.8356 - acc: 0.2954 - val_loss: 3.0325 - val_acc: 0.2574\n",
            "Epoch 426/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8357 - acc: 0.2950 - val_loss: 3.0314 - val_acc: 0.2574\n",
            "Epoch 427/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8350 - acc: 0.2964 - val_loss: 3.0304 - val_acc: 0.2588\n",
            "Epoch 428/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.8347 - acc: 0.2961 - val_loss: 3.0296 - val_acc: 0.2598\n",
            "Epoch 429/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8356 - acc: 0.2948 - val_loss: 3.0303 - val_acc: 0.2575\n",
            "Epoch 430/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8349 - acc: 0.2957 - val_loss: 3.0293 - val_acc: 0.2599\n",
            "Epoch 431/500\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 2.8348 - acc: 0.2959 - val_loss: 3.0285 - val_acc: 0.2592\n",
            "Epoch 432/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8337 - acc: 0.2959 - val_loss: 3.0297 - val_acc: 0.2597\n",
            "Epoch 433/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.8334 - acc: 0.2962 - val_loss: 3.0299 - val_acc: 0.2596\n",
            "Epoch 434/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8343 - acc: 0.2965 - val_loss: 3.0296 - val_acc: 0.2592\n",
            "Epoch 435/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8336 - acc: 0.2955 - val_loss: 3.0290 - val_acc: 0.2588\n",
            "Epoch 436/500\n",
            "100/100 [==============================] - 3s 28ms/step - loss: 2.8339 - acc: 0.2957 - val_loss: 3.0281 - val_acc: 0.2575\n",
            "Epoch 437/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8335 - acc: 0.2957 - val_loss: 3.0294 - val_acc: 0.2597\n",
            "Epoch 438/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8331 - acc: 0.2962 - val_loss: 3.0294 - val_acc: 0.2583\n",
            "Epoch 439/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8330 - acc: 0.2968 - val_loss: 3.0280 - val_acc: 0.2605\n",
            "Epoch 440/500\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 2.8321 - acc: 0.2960 - val_loss: 3.0290 - val_acc: 0.2595\n",
            "Epoch 441/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.8328 - acc: 0.2955 - val_loss: 3.0272 - val_acc: 0.2604\n",
            "Epoch 442/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8320 - acc: 0.2960 - val_loss: 3.0294 - val_acc: 0.2591\n",
            "Epoch 443/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8318 - acc: 0.2962 - val_loss: 3.0267 - val_acc: 0.2601\n",
            "Epoch 444/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8316 - acc: 0.2961 - val_loss: 3.0297 - val_acc: 0.2580\n",
            "Epoch 445/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8314 - acc: 0.2965 - val_loss: 3.0271 - val_acc: 0.2595\n",
            "Epoch 446/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8311 - acc: 0.2956 - val_loss: 3.0286 - val_acc: 0.2579\n",
            "Epoch 447/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8313 - acc: 0.2961 - val_loss: 3.0270 - val_acc: 0.2598\n",
            "Epoch 448/500\n",
            "100/100 [==============================] - 3s 29ms/step - loss: 2.8310 - acc: 0.2967 - val_loss: 3.0263 - val_acc: 0.2595\n",
            "Epoch 449/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8308 - acc: 0.2959 - val_loss: 3.0269 - val_acc: 0.2588\n",
            "Epoch 450/500\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 2.8304 - acc: 0.2960 - val_loss: 3.0277 - val_acc: 0.2585\n",
            "Epoch 451/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8301 - acc: 0.2968 - val_loss: 3.0278 - val_acc: 0.2588\n",
            "Epoch 452/500\n",
            "100/100 [==============================] - 3s 28ms/step - loss: 2.8305 - acc: 0.2958 - val_loss: 3.0269 - val_acc: 0.2579\n",
            "Epoch 453/500\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 2.8301 - acc: 0.2967 - val_loss: 3.0275 - val_acc: 0.2579\n",
            "Epoch 454/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8296 - acc: 0.2973 - val_loss: 3.0259 - val_acc: 0.2595\n",
            "Epoch 455/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8293 - acc: 0.2965 - val_loss: 3.0253 - val_acc: 0.2606\n",
            "Epoch 456/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8293 - acc: 0.2962 - val_loss: 3.0250 - val_acc: 0.2610\n",
            "Epoch 457/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8286 - acc: 0.2970 - val_loss: 3.0252 - val_acc: 0.2590\n",
            "Epoch 458/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8291 - acc: 0.2967 - val_loss: 3.0280 - val_acc: 0.2573\n",
            "Epoch 459/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8292 - acc: 0.2962 - val_loss: 3.0265 - val_acc: 0.2604\n",
            "Epoch 460/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8279 - acc: 0.2970 - val_loss: 3.0245 - val_acc: 0.2608\n",
            "Epoch 461/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8283 - acc: 0.2983 - val_loss: 3.0251 - val_acc: 0.2601\n",
            "Epoch 462/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8281 - acc: 0.2973 - val_loss: 3.0254 - val_acc: 0.2592\n",
            "Epoch 463/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8284 - acc: 0.2966 - val_loss: 3.0242 - val_acc: 0.2582\n",
            "Epoch 464/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8281 - acc: 0.2974 - val_loss: 3.0256 - val_acc: 0.2605\n",
            "Epoch 465/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8295 - acc: 0.2964 - val_loss: 3.0279 - val_acc: 0.2584\n",
            "Epoch 466/500\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 2.8271 - acc: 0.2967 - val_loss: 3.0274 - val_acc: 0.2595\n",
            "Epoch 467/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8269 - acc: 0.2968 - val_loss: 3.0265 - val_acc: 0.2577\n",
            "Epoch 468/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8277 - acc: 0.2968 - val_loss: 3.0246 - val_acc: 0.2593\n",
            "Epoch 469/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8274 - acc: 0.2964 - val_loss: 3.0262 - val_acc: 0.2565\n",
            "Epoch 470/500\n",
            "100/100 [==============================] - 3s 29ms/step - loss: 2.8264 - acc: 0.2970 - val_loss: 3.0249 - val_acc: 0.2591\n",
            "Epoch 471/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8259 - acc: 0.2972 - val_loss: 3.0253 - val_acc: 0.2584\n",
            "Epoch 472/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8260 - acc: 0.2968 - val_loss: 3.0250 - val_acc: 0.2603\n",
            "Epoch 473/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8257 - acc: 0.2979 - val_loss: 3.0227 - val_acc: 0.2616\n",
            "Epoch 474/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8257 - acc: 0.2974 - val_loss: 3.0249 - val_acc: 0.2597\n",
            "Epoch 475/500\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 2.8254 - acc: 0.2971 - val_loss: 3.0237 - val_acc: 0.2594\n",
            "Epoch 476/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8250 - acc: 0.2972 - val_loss: 3.0236 - val_acc: 0.2589\n",
            "Epoch 477/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.8248 - acc: 0.2985 - val_loss: 3.0244 - val_acc: 0.2576\n",
            "Epoch 478/500\n",
            "100/100 [==============================] - 3s 28ms/step - loss: 2.8249 - acc: 0.2982 - val_loss: 3.0237 - val_acc: 0.2604\n",
            "Epoch 479/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.8249 - acc: 0.2973 - val_loss: 3.0230 - val_acc: 0.2609\n",
            "Epoch 480/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8246 - acc: 0.2977 - val_loss: 3.0220 - val_acc: 0.2599\n",
            "Epoch 481/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8246 - acc: 0.2982 - val_loss: 3.0218 - val_acc: 0.2596\n",
            "Epoch 482/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.8237 - acc: 0.2980 - val_loss: 3.0213 - val_acc: 0.2620\n",
            "Epoch 483/500\n",
            "100/100 [==============================] - 3s 30ms/step - loss: 2.8243 - acc: 0.2980 - val_loss: 3.0228 - val_acc: 0.2597\n",
            "Epoch 484/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8233 - acc: 0.2977 - val_loss: 3.0229 - val_acc: 0.2598\n",
            "Epoch 485/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8245 - acc: 0.2977 - val_loss: 3.0218 - val_acc: 0.2600\n",
            "Epoch 486/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8231 - acc: 0.2977 - val_loss: 3.0239 - val_acc: 0.2591\n",
            "Epoch 487/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8229 - acc: 0.2975 - val_loss: 3.0222 - val_acc: 0.2600\n",
            "Epoch 488/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.8226 - acc: 0.2982 - val_loss: 3.0221 - val_acc: 0.2606\n",
            "Epoch 489/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8231 - acc: 0.2987 - val_loss: 3.0204 - val_acc: 0.2610\n",
            "Epoch 490/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8227 - acc: 0.2982 - val_loss: 3.0201 - val_acc: 0.2592\n",
            "Epoch 491/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8223 - acc: 0.2967 - val_loss: 3.0218 - val_acc: 0.2609\n",
            "Epoch 492/500\n",
            "100/100 [==============================] - 3s 32ms/step - loss: 2.8223 - acc: 0.2981 - val_loss: 3.0206 - val_acc: 0.2595\n",
            "Epoch 493/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8218 - acc: 0.2976 - val_loss: 3.0234 - val_acc: 0.2602\n",
            "Epoch 494/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8219 - acc: 0.2979 - val_loss: 3.0218 - val_acc: 0.2599\n",
            "Epoch 495/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8217 - acc: 0.2977 - val_loss: 3.0204 - val_acc: 0.2601\n",
            "Epoch 496/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8213 - acc: 0.2980 - val_loss: 3.0196 - val_acc: 0.2606\n",
            "Epoch 497/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8212 - acc: 0.2984 - val_loss: 3.0215 - val_acc: 0.2618\n",
            "Epoch 498/500\n",
            "100/100 [==============================] - 3s 26ms/step - loss: 2.8210 - acc: 0.2988 - val_loss: 3.0208 - val_acc: 0.2612\n",
            "Epoch 499/500\n",
            "100/100 [==============================] - 3s 31ms/step - loss: 2.8207 - acc: 0.2985 - val_loss: 3.0199 - val_acc: 0.2611\n",
            "Epoch 500/500\n",
            "100/100 [==============================] - 3s 27ms/step - loss: 2.8207 - acc: 0.2986 - val_loss: 3.0223 - val_acc: 0.2599\n",
            "Training time: 1463.2082512378693\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.plot(history.history['acc'], label='accuracy')\n",
        "plt.plot(history.history['val_acc'], label = 'val_accuracy')\n",
        "plt.xlabel('Epoch')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.legend(loc='lower right')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 296
        },
        "id": "I3glCnqyA0Jv",
        "outputId": "d6391c26-1e05-4f3a-e275-f477e7348f41"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7f471b676520>"
            ]
          },
          "metadata": {},
          "execution_count": 57
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "test_loss, test_acc = model.evaluate(test_images,  test_labels, verbose=2)\n",
        "print(test_acc)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "3qh1lFkLA6uw",
        "outputId": "e0b4b9f4-4ce0-4ff5-c1c2-0d407f469419"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 - 1s - loss: 3.0036 - acc: 0.2741 - 893ms/epoch - 3ms/step\n",
            "0.27410000562667847\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "##5 - Learning Rate: 1e-04, Epochs/Steps: 100/500, GPU as Hardware Accelerator, Training time: 7.23 minutes, Accuracy: 31%"
      ],
      "metadata": {
        "id": "m2AZ2zuyBELv"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model.compile(optimizer=optimizers.Adam(learning_rate=1e-04),\n",
        "              loss='categorical_crossentropy',\n",
        "              metrics=['acc'])\n",
        "print(\"Learning rate before second fit:\", model.optimizer.learning_rate.numpy())"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4Rs7mGatBKeA",
        "outputId": "48d2bf53-cb03-44f2-ff70-d95a90509d1b"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Learning rate before second fit: 1e-04\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "training_start = time.time()\n",
        "history = model.fit(train_images, train_labels,\n",
        "                    steps_per_epoch=500,\n",
        "                    epochs=100,\n",
        "                    validation_split=0.2) \n",
        "training_stop = time.time()\n",
        "training_time = training_stop - training_start\n",
        "print(f\"Training time: {training_time}\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "fUXRs0wKBf4P",
        "outputId": "1ebacef0-3827-4604-a503-d76d0abfb548"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/100\n",
            "500/500 [==============================] - 6s 10ms/step - loss: 2.8792 - acc: 0.2849 - val_loss: 3.0681 - val_acc: 0.2522\n",
            "Epoch 2/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.8707 - acc: 0.2858 - val_loss: 3.0333 - val_acc: 0.2574\n",
            "Epoch 3/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.8697 - acc: 0.2865 - val_loss: 3.0369 - val_acc: 0.2568\n",
            "Epoch 4/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.8688 - acc: 0.2844 - val_loss: 3.0443 - val_acc: 0.2560\n",
            "Epoch 5/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.8647 - acc: 0.2868 - val_loss: 3.0260 - val_acc: 0.2557\n",
            "Epoch 6/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.8588 - acc: 0.2888 - val_loss: 3.0233 - val_acc: 0.2616\n",
            "Epoch 7/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.8523 - acc: 0.2906 - val_loss: 3.0886 - val_acc: 0.2496\n",
            "Epoch 8/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.8545 - acc: 0.2928 - val_loss: 3.0394 - val_acc: 0.2557\n",
            "Epoch 9/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.8453 - acc: 0.2928 - val_loss: 3.0586 - val_acc: 0.2552\n",
            "Epoch 10/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.8439 - acc: 0.2930 - val_loss: 3.0009 - val_acc: 0.2638\n",
            "Epoch 11/100\n",
            "500/500 [==============================] - 5s 9ms/step - loss: 2.8334 - acc: 0.2945 - val_loss: 3.0408 - val_acc: 0.2569\n",
            "Epoch 12/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.8348 - acc: 0.2924 - val_loss: 3.0204 - val_acc: 0.2634\n",
            "Epoch 13/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.8338 - acc: 0.2930 - val_loss: 3.0280 - val_acc: 0.2626\n",
            "Epoch 14/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.8309 - acc: 0.2948 - val_loss: 3.0078 - val_acc: 0.2611\n",
            "Epoch 15/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.8269 - acc: 0.2947 - val_loss: 3.0222 - val_acc: 0.2608\n",
            "Epoch 16/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.8200 - acc: 0.2970 - val_loss: 3.0076 - val_acc: 0.2620\n",
            "Epoch 17/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.8165 - acc: 0.2970 - val_loss: 3.0009 - val_acc: 0.2621\n",
            "Epoch 18/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.8113 - acc: 0.2972 - val_loss: 3.0391 - val_acc: 0.2604\n",
            "Epoch 19/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.8105 - acc: 0.2970 - val_loss: 3.0102 - val_acc: 0.2576\n",
            "Epoch 20/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.8042 - acc: 0.2991 - val_loss: 2.9979 - val_acc: 0.2647\n",
            "Epoch 21/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.8014 - acc: 0.2992 - val_loss: 3.0086 - val_acc: 0.2628\n",
            "Epoch 22/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.8001 - acc: 0.2984 - val_loss: 3.0133 - val_acc: 0.2606\n",
            "Epoch 23/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.7879 - acc: 0.3010 - val_loss: 2.9785 - val_acc: 0.2683\n",
            "Epoch 24/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.7921 - acc: 0.3016 - val_loss: 2.9892 - val_acc: 0.2640\n",
            "Epoch 25/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.7842 - acc: 0.3035 - val_loss: 2.9664 - val_acc: 0.2711\n",
            "Epoch 26/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.7795 - acc: 0.3031 - val_loss: 2.9713 - val_acc: 0.2662\n",
            "Epoch 27/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.7771 - acc: 0.3062 - val_loss: 2.9795 - val_acc: 0.2639\n",
            "Epoch 28/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.7726 - acc: 0.3044 - val_loss: 2.9801 - val_acc: 0.2681\n",
            "Epoch 29/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.7692 - acc: 0.3072 - val_loss: 2.9787 - val_acc: 0.2665\n",
            "Epoch 30/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.7679 - acc: 0.3054 - val_loss: 2.9643 - val_acc: 0.2704\n",
            "Epoch 31/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.7626 - acc: 0.3074 - val_loss: 2.9721 - val_acc: 0.2687\n",
            "Epoch 32/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.7576 - acc: 0.3092 - val_loss: 2.9632 - val_acc: 0.2706\n",
            "Epoch 33/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.7506 - acc: 0.3086 - val_loss: 2.9583 - val_acc: 0.2715\n",
            "Epoch 34/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.7498 - acc: 0.3112 - val_loss: 2.9639 - val_acc: 0.2694\n",
            "Epoch 35/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.7507 - acc: 0.3102 - val_loss: 2.9618 - val_acc: 0.2692\n",
            "Epoch 36/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.7474 - acc: 0.3109 - val_loss: 2.9590 - val_acc: 0.2689\n",
            "Epoch 37/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.7383 - acc: 0.3132 - val_loss: 2.9545 - val_acc: 0.2741\n",
            "Epoch 38/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.7380 - acc: 0.3122 - val_loss: 2.9340 - val_acc: 0.2729\n",
            "Epoch 39/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.7309 - acc: 0.3133 - val_loss: 2.9347 - val_acc: 0.2746\n",
            "Epoch 40/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.7283 - acc: 0.3185 - val_loss: 2.9437 - val_acc: 0.2727\n",
            "Epoch 41/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.7219 - acc: 0.3163 - val_loss: 2.9372 - val_acc: 0.2745\n",
            "Epoch 42/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.7163 - acc: 0.3168 - val_loss: 2.9270 - val_acc: 0.2766\n",
            "Epoch 43/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.7159 - acc: 0.3176 - val_loss: 2.9458 - val_acc: 0.2717\n",
            "Epoch 44/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.7087 - acc: 0.3189 - val_loss: 2.9397 - val_acc: 0.2728\n",
            "Epoch 45/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.7086 - acc: 0.3180 - val_loss: 2.9590 - val_acc: 0.2769\n",
            "Epoch 46/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.7063 - acc: 0.3195 - val_loss: 2.9380 - val_acc: 0.2745\n",
            "Epoch 47/100\n",
            "500/500 [==============================] - 5s 9ms/step - loss: 2.7022 - acc: 0.3195 - val_loss: 2.9343 - val_acc: 0.2774\n",
            "Epoch 48/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.6993 - acc: 0.3189 - val_loss: 2.9370 - val_acc: 0.2791\n",
            "Epoch 49/100\n",
            "500/500 [==============================] - 3s 7ms/step - loss: 2.6963 - acc: 0.3220 - val_loss: 2.9260 - val_acc: 0.2740\n",
            "Epoch 50/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.6878 - acc: 0.3249 - val_loss: 2.9316 - val_acc: 0.2791\n",
            "Epoch 51/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.6906 - acc: 0.3213 - val_loss: 2.9254 - val_acc: 0.2783\n",
            "Epoch 52/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.6821 - acc: 0.3232 - val_loss: 2.9111 - val_acc: 0.2796\n",
            "Epoch 53/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.6763 - acc: 0.3253 - val_loss: 2.9369 - val_acc: 0.2777\n",
            "Epoch 54/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.6718 - acc: 0.3259 - val_loss: 2.8938 - val_acc: 0.2801\n",
            "Epoch 55/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.6710 - acc: 0.3244 - val_loss: 2.9032 - val_acc: 0.2806\n",
            "Epoch 56/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.6642 - acc: 0.3259 - val_loss: 2.9112 - val_acc: 0.2860\n",
            "Epoch 57/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.6641 - acc: 0.3286 - val_loss: 2.8818 - val_acc: 0.2890\n",
            "Epoch 58/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.6583 - acc: 0.3301 - val_loss: 2.8788 - val_acc: 0.2877\n",
            "Epoch 59/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.6530 - acc: 0.3291 - val_loss: 2.8779 - val_acc: 0.2892\n",
            "Epoch 60/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.6489 - acc: 0.3311 - val_loss: 2.8877 - val_acc: 0.2849\n",
            "Epoch 61/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.6470 - acc: 0.3309 - val_loss: 2.9050 - val_acc: 0.2817\n",
            "Epoch 62/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.6478 - acc: 0.3293 - val_loss: 2.9094 - val_acc: 0.2821\n",
            "Epoch 63/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.6409 - acc: 0.3335 - val_loss: 2.9077 - val_acc: 0.2783\n",
            "Epoch 64/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.6380 - acc: 0.3319 - val_loss: 2.8635 - val_acc: 0.2901\n",
            "Epoch 65/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.6346 - acc: 0.3334 - val_loss: 2.8702 - val_acc: 0.2925\n",
            "Epoch 66/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.6303 - acc: 0.3354 - val_loss: 2.8867 - val_acc: 0.2880\n",
            "Epoch 67/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.6259 - acc: 0.3358 - val_loss: 2.8700 - val_acc: 0.2876\n",
            "Epoch 68/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.6213 - acc: 0.3373 - val_loss: 2.8773 - val_acc: 0.2873\n",
            "Epoch 69/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.6226 - acc: 0.3377 - val_loss: 2.8672 - val_acc: 0.2922\n",
            "Epoch 70/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.6130 - acc: 0.3380 - val_loss: 2.8698 - val_acc: 0.2911\n",
            "Epoch 71/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.6119 - acc: 0.3385 - val_loss: 2.8840 - val_acc: 0.2867\n",
            "Epoch 72/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.6100 - acc: 0.3381 - val_loss: 2.8749 - val_acc: 0.2893\n",
            "Epoch 73/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.6018 - acc: 0.3383 - val_loss: 2.8653 - val_acc: 0.2929\n",
            "Epoch 74/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.5997 - acc: 0.3410 - val_loss: 2.8474 - val_acc: 0.2936\n",
            "Epoch 75/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.6009 - acc: 0.3396 - val_loss: 2.8434 - val_acc: 0.2934\n",
            "Epoch 76/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.5954 - acc: 0.3421 - val_loss: 2.8723 - val_acc: 0.2893\n",
            "Epoch 77/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.5924 - acc: 0.3432 - val_loss: 2.8526 - val_acc: 0.2947\n",
            "Epoch 78/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.5855 - acc: 0.3427 - val_loss: 2.8593 - val_acc: 0.2912\n",
            "Epoch 79/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.5871 - acc: 0.3413 - val_loss: 2.9083 - val_acc: 0.2857\n",
            "Epoch 80/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.5854 - acc: 0.3438 - val_loss: 2.8379 - val_acc: 0.2979\n",
            "Epoch 81/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.5789 - acc: 0.3444 - val_loss: 2.8249 - val_acc: 0.3018\n",
            "Epoch 82/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.5781 - acc: 0.3434 - val_loss: 2.8956 - val_acc: 0.2843\n",
            "Epoch 83/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.5735 - acc: 0.3444 - val_loss: 2.8345 - val_acc: 0.2959\n",
            "Epoch 84/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.5655 - acc: 0.3492 - val_loss: 2.8304 - val_acc: 0.2986\n",
            "Epoch 85/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.5631 - acc: 0.3484 - val_loss: 2.8337 - val_acc: 0.2959\n",
            "Epoch 86/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.5601 - acc: 0.3487 - val_loss: 2.8318 - val_acc: 0.2960\n",
            "Epoch 87/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.5574 - acc: 0.3512 - val_loss: 2.8267 - val_acc: 0.2975\n",
            "Epoch 88/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.5570 - acc: 0.3480 - val_loss: 2.8565 - val_acc: 0.2948\n",
            "Epoch 89/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.5517 - acc: 0.3517 - val_loss: 2.8344 - val_acc: 0.3008\n",
            "Epoch 90/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.5503 - acc: 0.3505 - val_loss: 2.8126 - val_acc: 0.3004\n",
            "Epoch 91/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.5474 - acc: 0.3517 - val_loss: 2.8218 - val_acc: 0.3006\n",
            "Epoch 92/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.5429 - acc: 0.3511 - val_loss: 2.8686 - val_acc: 0.2922\n",
            "Epoch 93/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.5378 - acc: 0.3524 - val_loss: 2.8306 - val_acc: 0.2985\n",
            "Epoch 94/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.5391 - acc: 0.3524 - val_loss: 2.8142 - val_acc: 0.3009\n",
            "Epoch 95/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.5332 - acc: 0.3530 - val_loss: 2.8199 - val_acc: 0.2998\n",
            "Epoch 96/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.5292 - acc: 0.3553 - val_loss: 2.8322 - val_acc: 0.2944\n",
            "Epoch 97/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.5283 - acc: 0.3548 - val_loss: 2.8435 - val_acc: 0.2974\n",
            "Epoch 98/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.5246 - acc: 0.3562 - val_loss: 2.8070 - val_acc: 0.3019\n",
            "Epoch 99/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.5202 - acc: 0.3554 - val_loss: 2.8294 - val_acc: 0.2962\n",
            "Epoch 100/100\n",
            "500/500 [==============================] - 5s 9ms/step - loss: 2.5173 - acc: 0.3585 - val_loss: 2.7964 - val_acc: 0.3015\n",
            "Training time: 443.7711670398712\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.plot(history.history['acc'], label='accuracy')\n",
        "plt.plot(history.history['val_acc'], label = 'val_accuracy')\n",
        "plt.xlabel('Epoch')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.legend(loc='lower right')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 296
        },
        "id": "wZqWUZTADXH3",
        "outputId": "b8bc224b-d17f-4e58-bd2a-082f3a230e73"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7f478c7d8fd0>"
            ]
          },
          "metadata": {},
          "execution_count": 61
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEGCAYAAAB/+QKOAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/NK7nSAAAACXBIWXMAAAsTAAALEwEAmpwYAABLH0lEQVR4nO3dd3hUVfrA8e+bRiChBBJCJ/TQW+goSlEsgA1BURBF7I11Fcva3d+uurquYkFFQZqKIkXFgijSCZ3QO6GGJISEkH5+f5xJMoEJCZDJpLyf58kzuXfuvXNuBuad094jxhiUUkqps3l5ugBKKaVKJg0QSimlXNIAoZRSyiUNEEoppVzSAKGUUsolH08XoKgEBwebsLAwTxdDKaVKlTVr1pwwxoS4eq7MBIiwsDAiIyM9XQyllCpVRGR/fs9pE5NSSimXNEAopZRyya0BQkQGish2EdklIuNdPH+/iGwSkfUiskREWjk9105ElotIlOMYf3eWVSmlVF5uCxAi4g1MAK4BWgG3OQcAh+nGmLbGmA7AG8DbjnN9gKnA/caY1sAVQLq7yqqUUupc7qxBdAV2GWP2GGPSgJnAEOcDjDGnnDYDgOzEUFcBG40xGxzHxRpjMt1YVqWUUmdxZ4CoCxx02o527MtDRB4Skd3YGsSjjt3NASMiP4vIWhF5yo3lVEop5YLHO6mNMROMMU2Ap4HnHbt9gN7ACMfjjSLS7+xzRWSsiESKSGRMTEyxlVkppcoDdwaIQ0B9p+16jn35mQnc4Pg9GlhsjDlhjEkGfgQ6nX2CMWaiMSbCGBMREuJynodSSpVp3687xOx10bhj6QZ3BojVQDMRaSQifsBwYK7zASLSzGnzOmCn4/efgbYiUsnRYd0H2OLGsiqlVKmTmJLOq/O38PXqaLdc320zqY0xGSLyMPbD3huYZIyJEpFXgEhjzFzgYRHpjx2hFA+McpwbLyJvY4OMAX40xvzgrrIqpVRp9Mlfe4k9ncb4a8IRkSK/vltTbRhjfsQ2Dznve8Hp98fOc+5U7FBXpZRSZzmemMKnf+3huna1aV+/mltew+Od1EoppQp2JOEMJ5PTcrbf/W0naRlZ/P2qFm57zTKTrE8ppcqquRsO89SsDQDcGlGffi1Dmbn6ICO6NSAsOMBtr6sBQimlSoj0zCxe/2ErVSv6MqBVKC1rV+GtX7bz4R+7iWgYROOQAGasOsCU5fsJ8PPm0X7NCr7oJdAAoZRSJcR7C3fyxbJ9iMC7C3dS2d+HxJQMbu/WgJcGtcbPx4u/XdWCaSv206JWFYIDK7i1PBoglFKqBFi1N473F+3ils71GH9NOL9vPc6SXSfo3TSYW7vkTikLreLPODf2OzjTAKGUUh6WcCadJ75aT/3qlXhpcGsCK/hwa5f6eQKDJ2iAUEopD4pNSuX57zdz9FQKs+7vQWCFkvOxXHJKopRSZVhqRiYLNh8lMSWDzCxDYko6i3ecIHJ/HFkGnhrYgo4NgjxdzDw0QCil1EVIz8ziZHI6IZXP7SjOyjJ4eeXObI6OT+ahaWvZEJ2Q57jwWpV5uG8zrmoVSpu6Vd1e5gulAUIppc4jK8uQkWXw88k7r/jV+VuYvvIAf7+6Bfde1hgvL+FkchovzInix01H6Na4Ote3q0PVir48890msrIME27vRJdGQfh4eeHrLVT29/XQXRWOBgillMpHVpZh1OerOJmczuwHe+LjbYPEiaRUZq4+SLVKvvzfT9tYsusEN3eqx+s/biX+dBqDO9Rh3YGTPPPdJgBa1q7ChyM6uXVSmztogFBKqXxMWrqXv3aeAGDWmmiGd20AwJTl+0nPzOKr+3qwck8cr8yP4q+dJwivVZnP7+pCm7pVMcYQdfgU248mcl272vj7envyVi6KBgillHJh57FE3vh5O/1bhhKfnMY7v+1gSAe7KOaXy/fRv2UoTUICaRISSNdG1VmxJ5ahEfWo4GMDgYjQpm7VEtm3UFgaIJRS5dZPm47g6+1F3/CaeTqV0zOzGPf1BgIr+PB/N7VlX+xphn60nElL91LF34f45HTGXt445/imNQNpWjPQE7fgVhoglFLlUtzpNB6ZsY6MLEPz0EDu79OEJiGBbD+ayKLtx9l0KIEPR3QipHIFQipXoH/LUD76YzdVK/nSsUE1IhqWrCGp7qABQilVLv2w8TAZWYanBrZg7vrDjPt6Q85z/r5ejOndiGva1s7Z9/TAFlz938Ukpmbw3LUt3bJAT0mjAUIpVS7NXneI8FqVefCKpjzQpwlLdp3gdGoGLWpVoUH1Snh75Q0AzUIrM7JHGCv2xHJV61oeKnXx0gChlCp3DsQms/bASZ4eGA7YDuXLmoUUeN6Lg1rlHF8eaIBQSpU7368/BMCQDnUu6LzyEhiy6ZKjSqlyxRjD9+sP0a1RdepUq+jp4pRoGiCUUuXK5kOn2BNzmhs61vV0UUo8DRBKqXJl9rpD+Hl7cW2b2gUfXM5pH4RSqszIyjJsOpTAL1uOErkvnoFtanFn94Y5OZT+2hnDd+uiuTI8hKqVSnaivJJAA4RSqkzYe+I0d362kuj4M3h7CQ2rV+LleVuYueog465qzg8bjzB3w2EaBQcwbkDxLNlZ2mmAUEqVeD9HHeW3Lcfo1DCIy5uHUPeszuXMLMPfvl5PYkoGb9/anr7hNala0Zefo47x2g9buO/LNfh5e/FYv2Y8cEWTUpk4zxM0QCilSqyjCSm8MGczv2w5RkVfb75ZEw3Y9Nlv3tIuJxHeZ0v2sPbASf47rEOezueBbWpxRYsQZq+zo5Yah5S9fEnupAFCKVUi/bH9OA9PX0d6Zhbjrwnnnt6N2HfiNH/uiOGzJXsZ+tFy3r61Pc1CK/PWLzu4qlWoy3kN/r7e3OZI060ujAYIpVSJE5uUyt++3kC9oIp8fGdnGtawC+00C61Ms9DKDO5Qh/u+XMMD09ZSq4o/lfy8ee3GNuVuIpu7uXWYq4gMFJHtIrJLRMa7eP5+EdkkIutFZImItDrr+QYikiQiT7qznEopz5mx6gDPzd5EclpGzr4X5kZxKiWdd4d3zAkOzmpW9mfGvd25qVNdjp5K4ZUhbahZ2b84i10uuK0GISLewARgABANrBaRucaYLU6HTTfGfOQ4fjDwNjDQ6fm3gZ/cVUallGftPXGaF+dEkZaZxaZDCXw6MoJV++L4YeMR/n51C1rUqpzvuf6+3vxnaHueHhhOaBUNDu7gziamrsAuY8weABGZCQwBcgKEMeaU0/EBgMneEJEbgL3AaTeWUSnlIcYYXpizmQo+Xrx6Q2tenreFGyYsJSUji3b1qnKf04I8+RERDQ5u5M4mprrAQaftaMe+PETkIRHZDbwBPOrYFwg8Dbx8vhcQkbEiEikikTExMUVWcKWU+/2w6Qh/7TzBk1e3YFiXBnx9Xw8yjSEpJYP/DG2fM7lNeY7H3wFjzARjTBNsQHjesfsl4B1jTFIB5040xkQYYyJCQgpO1auUKl6xSamcSkk/Z39iSjqvzNtCm7pVuKN7QwDa1K3Kj49exvxHe9MsNP+mJVV83NnEdAio77Rdz7EvPzOBDx2/dwNuEZE3gGpAloikGGPed0dBlVKXzhhDfHI6u2OSWLLzBL9vs8t2ikCzmoF0ahBE1Yq+xCSlsuNYIjFJqUwcGZFnYZ4agRWoEVjBg3ehnLkzQKwGmolII2xgGA7c7nyAiDQzxux0bF4H7AQwxlzmdMxLQJIGB6VKlkxH3qM/t8ewZFcMO44lkXDG1hZEoGP9ajx5VXMys2DdwXh+2nyUM+mZhARWIDjQj39c14oO9at59ibUebktQBhjMkTkYeBnwBuYZIyJEpFXgEhjzFzgYRHpD6QD8cAod5VHKVV01h6I58Gpazl6KgURaFe3KoPa1yasRgCNggPo2CCI6gF+ec4xxo5B0bkKpYdkv2mlXUREhImMjPR0MZQq837dcoxHZqwltIo/4wY057JmIecEA1V6iMgaY0yEq+d0JrVSqlBS0jP5avVBXp4XRdu6Vfnsri4Ea39BmaYBQimVr5PJafznlx1E7o9nx7FEMrMM/cJr8t7tHankpx8fZZ2+w0opl1IzMhn75RrWHYine+Ma9A1vTIf6QVzZIkTnKJQTGiCUUucwxvDMt5tYtTeOd4d3YEgHXb+5PNKvAUqVY/kNUnnv9118t+4Q4wY01+BQjmmAUKqc+nr1Qbr+cyGR++Ly7J+8bB9v/7qDmzrV5ZG+TT1UOlUSaIBQqhzKyjJ8+OduYhJTueOzlfyx/TjGGN7+dQcvzo1iQKtQ/nVTO52zUM5pH4RS5dDyPbHsPXGa569ryex1h7h3SiS9mwazaHsMwyLq8/qNbbQjWmkNQqnyaNrK/VSr5Msd3RsyY2x3OtYPYtH2GB68ogn/urmtBgcFaA1CqTJpweaj/LkjhjpV/akbVJGODYJoFGxXZjt+KoVfoo4xulcY/r7e+Pt6M+Weruw4lki7etU8W3BVomiAUKqM+SXqKA9OW4O/rzfJaZkA+Pl48d5tHbm6dS2+jjxIRpbhtq4Ncs7x9/XW4KDOoQFCqVIq7nQaj81cR0Vfb8ZfE07jkEDWHYjn0ZnraFu3KjPGdsdLhINxyTz17UYemLqG129sy4xVB+nVtAaNQwI9fQuqhNNkfUqVQtHxyYyctIro+DP4eXuRmpHJ7V0bMG/jEQIr+PDdgz3z5ElKTsvg/qlrWbzDrrz4wYhOXNu2tqeKr0oQTdanVBmy9cgpRk1aRUp6JtPGdKNhjUr85+cdTFmxn2oVffli9LlJ9Cr5+fDpyAienb2JzYcSGNAq1EOlV6WJ1iCUKkXiTqcx4O0/8fX2YvLdXWlRK3dpzp3HEvHz8aJhjYDzXsMYo/MbVA6tQShVRvzzx60knEln/qO98wQHoNDrOGtwUIWlg52VKiWW7T7BrDXRjL28MeG1qni6OKoc0AChVCmQkp7Jc7M306B6JR7t18zTxVHlhAYIpUqIM2mZ/LkjhpT0zHOe++CP3ew9cZrXb2yDv6+3B0qnyiPtg1DKwzZGn2Tm6oPMW3+YxNQMOjaoxmejuuSs8zxl+T7e/30nN3Sow2XNQjxcWlWeaA1CKTeJSUxl9rrofNdcAJi9LprB7y/lu7XRDGgVyvPXtSTq8Clu+XAZ+06c5qW5UbwwJ4q+4TV5/ca2xVh6pbQGoZTbPPPdRn7bepzQKv70bBJ8zvObDyUw/ttNdGtUnU9GRVDF3xeADvWrcc/kSPq//ScZWYa7ezXiueta4u2lo49U8dIahFJu8Mf24/y29Tgi8OEfu895Pu50Gvd9uYYaAX5MGNEpJzgARIRV59sHetKxQTVevaENLwxqpcFBeYTWIJQqYmkZWbwyfwuNggO4oUNd3vltB5uiE2hbryoAGZlZPDx9LTFJqcy6v8c5s54BmtYM5Jv7exZ30ZXKQ2sQShWxycv2sSfmNP+4viWje4dRuYIPH/1paxHGGJ6bvZllu2N5/YY2mkFVlWhag1CqCB1PTOHdhTu5skUIfcNtvqM7ejTkoz/tMNWvIw/yVeRBHu3blKER9T1cWqXOT2sQShWhd37dQWpGJv+4vlXOvrt7NcLX24u7v1jNh3/s5vZuDXhiQHMPllKpwnFrgBCRgSKyXUR2ich4F8/fLyKbRGS9iCwRkVaO/QNEZI3juTUi0ted5VSqKOw8lshXqw9yZ/ewPGsthFSuwK0R9dh74jTXtq3Fq0PaaD4kVSq4rYlJRLyBCcAAIBpYLSJzjTFbnA6bboz5yHH8YOBtYCBwAhhkjDksIm2An4G67iqrUkXh3wu2EVDBh0f6Nj3nuSevakHz0MoM61JfRySpUsOdNYiuwC5jzB5jTBowExjifIAx5pTTZgBgHPvXGWMOO/ZHARVF5NyhHkqVECv2xPLb1uM8eEVTghwzoJ1Vq+THyB5hVPDRNBmq9HBnJ3Vd4KDTdjTQ7eyDROQhYBzgB7hqSroZWGuMSXVx7lhgLECDBg3OflqpImOM4aM/97Bs9wm8vQQfL6FOtYoMal+Hzg2C+L8ft1K7qj+je4V5uqhKFRmPj2IyxkwAJojI7cDzwKjs50SkNfBv4Kp8zp0ITAS7YJD7S6vKuozMLD5evIcaAX7c0rkePt5eZGRm8ezsTXwdGU14rcpU8PEiPdOwZNcJpizfT3CgHyeS0nhraHtNpKfKFHcGiEOA8zi+eo59+ZkJfJi9ISL1gNnASGPMuVNRlSqEIwlnGPbxCl4e0porW9Q877Hxp9N4aPpalu2OBWDS0r08PTCcr1Yf5Jctx3isXzMe798sp4P5dGoGv2w5yvfrDuPjJdzYUbvJVNnizgCxGmgmIo2wgWE4cLvzASLSzBiz07F5HbDTsb8a8AMw3hiz1I1lVGXcvA2HORCXzOMz1zP/kd7Ur17J5XHbjp7i3imRHEtI5c1b2lHZ34f/+2kb90yORAReHtyaUT3D8pwTUMGHGzvW48aO9YrhTpQqfm4LEMaYDBF5GDsCyRuYZIyJEpFXgEhjzFzgYRHpD6QD8eQ2Lz0MNAVeEJEXHPuuMsYcd1d5Vdn00+ajhNWoROzpNB6evpav7++R01FsjCFyfzxfLNvHgs1HqRHgx1f3dadjgyAArgyvydeR0dSu4k//VqGevA2lPELOl4q4NImIiDCRkZGeLoYqQY4knKHH//3O369uQZOQQO6fuoaRPRpyU6d6/L71GL9sOca2o4lU8fdheNcGjLmsETUr+3u62EoVKxFZY4yJcPWcxzuplXKXnzcfBWBgm1o0CQlkTO9GfLpkL1OW78dLoFODIF6/sQ03dqxLJT/9r6DU2fR/hSqzftp8lOahgTRxzGp++ppwqlXypV5QJfo0D3E5X0EplUsDhCqTTiSlsnpfHA/3bZazz9fbK8+2Uur8CpxJLSKDREST+qlS5ZeoY2QZuKZNLU8XRalSqzAf/MOAnSLyhoiEu7tAShWFnzYfIaxGJcJrVfZ0UZQqtQoMEMaYO4COwG7gCxFZLiJjRUT/5ymPSs/MYlN0AmePxItJTGX57lgGtqmtWVOVugSFajpyJNWbhZ3tXBu4EVgrIo+4sWxKndcz321i0PtLuGdyJIdOngHgr50xDHpvCQA3dKzjyeIpVeoV2EntSMM9GjtxbQrQ1RhzXEQqAVuA99xbRKXONWf9IWatiaZP8xBW7IllwNt/cnmzEBZEHaVJSACfjOxFeK0qni6mUqVaYUYx3Qy8Y4xZ7LzTGJMsIve4p1hK5W9/7Gmem72Zzg2D+GxUBEdPpfCP7zezIOood/UMY/w14Zo0T6kiUJgA8RJwJHtDRCoCocaYfcaYhe4qmCrflu46wfqDJxnVM4zACrn/TNMysnh0xjq8BN4d3gEfby/qBVVi0l1diE9Op7rObVCqyBQmQHwD9HTaznTs6+KWEqly7/ipFB6ctpaEM+lMXraPpwaG0y+8JrPXHWLGqgPsPJ7EByM6US8oN/GeiGhwUKqIFSZA+DhWhAPAGJMmIvo/UbmFMYbnv99MSnom7w7vwOdL9/HkNxsQAWOgff1qvDu8A9e2re3poipV5hUmQMSIyGBH9lVEZAh2zWilLln2ENXs4ag/bDrCL1uO8cw14QzpUJdB7eowb+Nhth5JZFD72rSuU9WTxVWqXClMgLgfmCYi7wOCXUZ0pFtLpcqFUynp3PnZKo4lpDCofW2uDK/JC3OiaF+vKvf0bgSAl5cwpENdhnTwbFmVKo8KDBCO1dy6i0igYzvJ7aVSZd6ZtEzu+WI1Ww4n0KNJMJ8v3ccnf+3F11t4c2h3fLw1u4tSnlaoZH0ich3QGvDPbgowxrzixnKpMiw1I5P7pq5hzf543rutE9e1q0386TQWRB0lOLACzUN1kr5SJUFhJsp9BFQCrgQ+BW4BVrm5XKoMWLT9OC1rVaFW1dxFeLKyDOO+2sDiHTG8cXM7rmtnO5uDAvy4rWsDTxVVKeVCYerxPY0xI4F4Y8zLQA+guXuLpUq7FXtiGf35am75aBlHE1Jy9v97wTZ+2HSE565tya1d6nuwhEqpghQmQGT/704WkTrY9aN1jKECIDPL8O2aaOJO54yEJiMzi5fmRhFapQInk9O547OVxCalMn3lAT5evIc7uzdkzGWNPFhqpVRhFKYPYp6IVAPeBNYCBvjEnYVSpcfXkQd55rtNtKxdhZn3dqdqJV+mrTzAtqOJfDiiE9UD/Bg5aRVDP17O/thkrmgRwouDWmmWVaVKgfPWIBwLBS00xpw0xnwLNATCjTEvFEvpVIl2Ji2Td37dQaPgAHYdT2T0F6s4GJfMf37ZTq+mNRjYphbdGtfgozs6cyA2mWY1A3n/9k46QkmpUuK8NQhjTJaITMCuB4ExJhVILY6CqZJv0tK9HE9M5ZsRnTiRmMpD09dyzbt/kZKeyUuDWufUEq4Mr8mCxy+nZpUKefIqKaVKtsJ8lVsoIjeLtgkoJ3Gn0/joj930bxlKl7DqXNO2Nv++uR1JqRnc1TOMZmcNVW1aM5Aq/r4eKq1S6mIU5uvcfcA4IENEUrCzqY0xRpPtlyOHTp7hg0W7aF2nKr2bBvP5sr2cTsvg6YEtco4ZGlGfro2qU98piZ5SqvQqzExqnbVUziUkpzNq0ip2Hc87iX54l/rn1BQa1ggozqIppdyoMBPlLne1/+wFhFTZlJqRyb1fRnIgNpkZ93anZpUKLN11gq1HEhk3QKfDKFWWFaaJ6e9Ov/sDXYE1QF+3lEiVGJlZhie/2ciqvXG8O7wDPZrUAKBJSKCHS6ZUATIzYNVEaD8cKlX3dGmKxp9vQEgLaDUk7/7kOPCvBl5FPzqwME1Mg5y3RaQ+8N/CXFxEBgLvAt7Ap8aYf531/P3AQ9hFiJKAscaYLY7nngHucTz3qDHm58K8pro0xhiW7Y7lp81H+DnqGDGJqYx3pN5WqtTYOgd+fgZ8K0LEaE+X5tKdPACL/gmhbc4NEHMfgVOHYeyiIn/ZixlzGA20LOggEfEGJgADHOesFpG52QHAYbox5iPH8YOBt4GBItIKGI5NEFgH+E1EmhtjMi+ivOoCvPXLdiYs2k1FX2+uDA9hcPs6XN26lqeLpdSFWfGRfUw65r7X2PMHrP4U6nSCBj0guDkkHYWEaBAvaDYg/3NP7ISvR0Kbm6DX4+BdwAi/DV8BBo5tstevWs/uTzsNu36DTqOK6KbyKkwfxHu2ZIAdFtsBO6O6IF2BXcaYPY7rzASGADkBwhhzyun4AKfXGQLMdMy72CsiuxzXW16I11UXKepwAh/9uYcbOtTh/25qR0U/b08XSakLF70Goh35RBOPuu91lr0PexbB1nmun394DQQ3PXd/Rhp8OwZid8Hvr0HUHBjyHtTp6Po6xsD6aRDUCOL3wo6focs99rldCyEjBVoOcn3uJSpMDSLS6fcMYIYxZmkhzquLXVwoWzTQ7eyDROQh7DBaP3L7NeoCK84695w2DhEZC4wFaNBAM4FeiozMLMZ/u4mgSr68NLi1BgdVeq38EPwqQ0AN99Ug0s/AviUQcQ/0eRoOroC4vVClNmRlwXdj4PBa1wHij/+DI+th2FRb05g/Dj7pByO+gab9zj3+wHIbGG740PZDOAeIrfOgUg1bg3GDwgSIWUBKdvOOiHiLSCVjTHJRFMAYMwGYICK3A88Dha4rGWMmAhMBIiIiTAGHl3uLth9n9tpDJKakk5iSQVCAH3f3akT3xtX5fOk+Nh1K4P3bO1Ktki45rkqpU0cgajZ0uRdO7HBfDWL/Msg4A03720AUfl3uc5kZtl/g8Dpod+u55y15Bzremfutv2EveD8C1n3pOkCsnwZ+gbbv4chGWPM5pCWDlw/sWACtBoO3ezIUFGomNVDRabsi8FshzjsEOOdzrufYl5+ZwA0Xea4qwJr98dw3ZQ3LdsdyIikNPx8v1h2I57ZPVnDTh8t4+9cd9G9Zk+vaaqJeVYpFfgZZmdBtLASGQtLx/I81Bn59Abb9cOGvs+s38K4AYb3Pfc7bB2q1hcPr8+5PTYTv7oOgMBjoNF6nYjVocS3s/BUyzspklHYaor6HVjeAXwA0v9o2Ke1dbH9ST0HLwRde/kIqTNjxd15m1BiTJCKFmSq7GmgmIo2wH+7DgdudDxCRZsaYnY7N64Ds3+cC00XkbWwndTN0kaKLdjQhhfunrqF2NX/mPNQrp4aQkp7JN5EH+XjxHny9hVeGtNEsq6r0Sk+ByM+h+UCo3hgqh9omJmPA1b/rjV/B0ndtc9TDq6BKncK/1q7fIKwX+OXzUVinI6ybaoOVl6O5dtuPkHAARs2DCmcNFQ+/HtZOhj1/QvOrcvdvmQtpSdBxhN1u2MvWJnYsAJNly96oT+HLfYEKU4M4LSKdsjdEpDNwpqCTjDEZwMPAz8BW4GtjTJSIvOIYsQTwsIhEich6bD/EKMe5UcDX2A7tBcBDOoKpcBLOpLPh4Ek2H0og4Uw6KemZjP0ykuTUDD4ZGZGn+cjf15s7e4Txx5NXsHR8X+pUq3ieKytVwh2KhOQT0Gmk3Q6sBVnpdp7A2U7Hws/P2mGjWenw09Our2kMLPkv/Plm7r74/bb5qul5RinV6Qjpp+1opWx7/oCK1aGhi1pH4z72g3/b/Lz7szuns/sYfPygSV/bD7HtBxtMfP3PvV4RKUwN4nHgGxE5jM3DVAsYVpiLG2N+BH48a98LTr8/dp5zXwdeL8zrlEdn0jJ5bOY6dsUk4eMleIkQk5hKrNPCPQAVfLxIzchi4p2d813r2cfbi8qagluVdnF77GNoK/tYOdQ+Jh21/QTOfn0BUhJg1HzY8RMsfAW2/wQtrsk9JiMN5j0KG2bY7Xqd7YfzLkcLe9P++ZelTgf7eGQ91Ay3gWbPIhsIXE1o86lgh8Vu/xGy3rG1jn1LYN9f0P/lvDWg5gNh61z7e/j1Bf1VLklhJsqtFpFwIDsr23ZjTLpbS6UK9PK8KH7ZcoyBrWshAhlZhg71q9EoOICw4ACMMRyIS+ZAXDJdG9XgKp3LoMq6uD3g5QtVHHMEAh3/5hOPQmjr3OP2/gXrp0LvcTaY1GgKG7+BH56EsMts809KAnw9yn6o93kaNn8L85+AB5bboaVVG0Bws/zLEtwcfCvZjur2wyFmOyQegcZX5n9O+PW2gz16NdTrams4VepBt/vyHtdsACDg7Xf+uRZFoDDzIB4CphljNju2g0TkNmPMB24tmcrXnPWHmLn6IA9e0YSnBoZ7ujhKlQxxe6Fag9wRPYE17aNzR7Ux8MPfbEfx5Y4sQj5+MOhdmHQVvNMKMtMhPRnEG4ZMgI532M7oyYNg0euw9087Oul8/XVe3lC7vQ0QYJuXABpfkf85zQbYALdtvg12RzbATZ/a2eDOAmvamkhACFRwby7VwjQx3esYigqAMSZeRO4FNEB4wJ6YJJ79bhMRDYM0WZ5SzuL22M7pbJUdNYgkp6Gup0/Aie1w1et5O5gbdLPzDPYvA/+qUKGK/TBv4Ji61ehy6DAClr9vt8/XvJStdgfb8ZyZYWsi1RtDUMP8j/eval9nyxzYNAvqdoY2N7s+9o7vCn79IlCYAOEtImKMMZCTQkMHyheBjMws9sWepmnN/L8F7Dtxmr99s4HElHQq+flwNCEFXx8v/ndbR126U5UfMTvsh23/l12P+TfG1iAadM/d5xdgR/kkOk2Wi9ttH4NdfLnqcLv9yc+AV20/RWqi/SAvSJ2OdtLe8Sjbn9CuEF23La+3TVkAQyfnn4DPq3gmshbmE2YB8JWI9BORfsAM4Cf3FqvsS07LYOyXa+j/9mKmrdzv8pjo+GRGfLqSPTFJNA4OpLK/D41DAvhgRCcdcaTKl+Xv25/N37p+PjkW0hLz1iDAMdTVqQYR6wgQNZpceBkCasDQL+CafxeuaSc7dcaqT+xQ1Sbn6X/I1uJaO7u69Y25tRcPKkwN4mlsOov7HdsbsSOZ1EU6kZTK3V+sZvOhBMJrVebFOVE0Dg7MSacNcCThDLd/spLElHRmjO1O6zpVPVhipdzk1GFYO8V+cFeta5PQhV0OIU7f8LMy7egegL/egrZDz/1mnT2CKahR3v2BoXlrELG7bN9CtYtMzdO4j/0pjBpN7dDVDTPth37YZQWfU7kWjF4ANQvMh1osCjOKKUtEVgJNgFuBYCCfMK6ypaRnsvZAPBV8vKnk502WMRyMO0N0fDJTlu/neGIKE++MoGvj6tz0wTIemLaGuQ/1pmaVCvyy5Rjv/LqDuNNpTBvTTYODKnuORcEf/7Jj+U0WVKlrv+lnZdgP1XFbbJs82FE9p2PsKJ9t8+0Qz9Y35L1e3F77eHYNIjDUdvbmHLfbdlAXlD21KHh52Y7q/UuhXhc7Y7owSkDNIVu+AUJEmgO3OX5OAF8BGGMKUU8q3zIysxj52SpW7XMxQQeoU9WfGfd2p2ODIAA+HRnBkAlLufXj5ZxOzSAxNYO61SryxegutK9frRhLrhQQvw+W/g+uetW24xdG2mnwqVi4RWsyM2DGcEg5BT0fhs6joXojW1PYv9SOFtowM3d457b5dnTPkPchZhssfsvmJXIeRRS3B5BzO4Er14Kdv+Rux+65uOali1W7g72n8w1vLcHOV4PYBvwFXG+M2QUgIk8US6lKubd/3cGqfXE8f11LmtYM5ExaJgaoH1SJBtUrUbVS3m8vYcEBfHhHJ/7+zUaual2LmzvXpXujGnh5adoLVcwy02HW3XBojf3227kQuTPPnIR329uO3yHv21XPzifqO7sAzvAZEH5t7n4vb9v5W7czrP4Muo61+7fOt806FYPgsr/B9w/YmcQtBuaeG7/XNk/5VMj7WoGhtv0/NckGu7jd0KgQTT1FpV6EfXSVhK8UOF+AuAmbP2mRiCzAJtPTT6wCLNp+nA/+2M1tXesz5rLGBZ/g0LNJMEvH6yquysP+fMMGhwpVbZqHwgSIbfMh5SQc3wof9YbLn4Lej7tuxjHGZjMNaWlnBLvSZYwNAvuW2FTW8Xuh16P2ubZDbbrsxW/axHXZtYi4PbYWcrbA7NnUx+x8gvTkc5uh3KnVELjrx7yjq0qRfOuDxpjvjTHDgXBgETblRk0R+VBErsrvvPLs8MkzjPtqve14HtS64BOUKkkOrLCdwO1vg8v/BgdXwoldBZ+3aZZt1390nU1hveg1m7rClR0/w/EtNoDk1xzV+kZbW1j9qSPTqkALRzptb1/o9ZjNu5Q9CQ1sH8TZHdSQm24j8eiljWC6WF7eNqlfKVVgg6Ex5rQxZrpjbep6wDrsyCblJCE5nTGTI0nLyOKDEZ3w99UFd1QpknIKvhsLVevDNW/YMfviDRumn/+8pBg7s7jNzRAYArdMst+a1315bupqsLWHqg3ynwAG9pt+hxG2ZrJ+mu3gzf6gB2h1ox0VlD2yKeWUTdLnqmaQnW4j6ZgdwQR2dJEqlAuaaWWMiTfGTDTGlM4GNTdJSs3gri9Wset4Eh/c0ZnGIYEFn6RUSbLpGzi5384m9q9iO3eb9of1M2zncX62fG9HIbW5JXdfp5FwJj73Azzb/mV25bWejxQ8iijibjuiKX5v3sV4wM5HaNDDps8Gewy4bmKq7BQg4nbbNRyyczWpAulU3EuUkp7JmMmr2RidwHu3d6RP8xBPF0mVdce32W/uzoyB5RNsc8/FOLETfAOgYc/cfR1uh8TDuXmEXNn8re1PyM6gCnbETpV6sPbLvOX789+2T6HjHQWXp0YTaOL4HupqveUW19oZyvH7cudAuKpBVAyyI6Cym5iqNy7cSCsFaIC4JJlZhoemrWXl3jjevrU9V2vGVOVuJw/CxD7wQXfYvcjuy0iD2ffZ7J8/jLPrJV+o7DxGzkNHW1xjP2DXT8u/LAeWQ9uzmou8vG1w2f07JETbfRu/toHm8r/nv8jO2a56zeZMctVnkD36aduP+U+SA3s/gY6Fg2J3F2//QxmgAeISvP7DVhZuO87Lg1szpENdTxdHlQcLX7aPlWrAlzfCon/CtFvs6mitb7JpqrfMvfDrxu2GGmd9A/epYEcNbZ0PW+fZRXacRc22j61vOvd6HUcABtZPt9lUFzxtU1hnD10tjNBWdp6EK9Ub25rL9h9tB3VAzXNXactWORROHbJNURogLoh7VrouB6au2M+kpXu5q2cYI3uEebo4qjyIjrR9BZc9CZeNg3mP22YbLx/bd9BuuB3Zs3YKtC/Uml5WZoZtqnG1tnHEPbDhK/jK0SwU3BxqtbPrK2z8Cup0cv2hGxRml8Jc9yUc22wn0g15v2iTzIVfa1d7OxN//qGrgbXswjuZaVBdA8SF0ABxEf7aGcOLc6O4skUI/7i+VcEnKHWpjLFNSIGhdoioXwDcNNE2A1WuDQ0dS1J2GmlrGSd2QXAhR+skHLAdwq4+ZGuGw5M7bOA5sAwOrrLDYTc7+joG/jv/63a8E74bYyfF9X2+4Al0F6rFdfDXf2wAan9b/sdVDoXUU/Z3rUFcEA0QLmRkZvHDpiNMXraPa9vWzjPh7dipFB6dsY5mNQN57/ZOeOtsZ1UcombbeQmD38vNJCoCbc5q3ukwwi5qs3ayTZVRGNlt+Pl9ePr62wCUHYTAzp6O32fXdM5Py+vBv5odOtvr8cKV5ULU6WhrB0lHC6hBOA2R1SGuF0QDhJPUjExmrYnm4z/3cCAumcr+Prz+41YahwTQNzyUrCzDuK/Xk5KexYQRnQisoH8+VQyyMuG3lyC0rQ0A51M51M5QXj8d+v7DrpZWkNjsUUAX8O26YjWo2OH8x/hWhLt/hkrV3ZMcz8vLpttY84XrDups2QHCLzBvsFAF0k5qbGCYumI/V775B8/N3kxQgB8f39mZlc/2o1XtKjw2Yz17YpL4bMlelu6K5YVBrWiicx1UcYlebeco9H68cG34ne+yE8fOnoeQn7jdjg/PmpdSStdqhrvnutna3GIn9NVul/8x2XMhzh6lpQpU7r8C7489zW0TV3A4IYWODarxr5vbcVmzYMTxD+njOzsz6L0ljP5iNYdPnuHq1qEM71Lfw6VW5crWeXYsf7NCZrhp0tfOQ4j87Ny02K7E7raTzErjh2ejy+DpvbmpwV3JrjVo/8MFK/c1iHpBlejaqDqT7+7Kdw/05PLmITnBIfv5Cbd34mBcMjUCKvCvm9rleV4ptzLG5iNq3MfOcC4ML2/ofj/sXWwT3hUkbnfpHt1zvuAATgFC+x8uVLkPEN5ewn+H2xnQ+X3w92wazNQx3Zh2bzeCAnQ5blWMjm91pJu4/sLO6zIGKteB3162QSY/mel2lFFZ/nZdpQ70Hle4NaFVHuU+QBRWzybB2u+gil9ONtNrCzw0D9+KcMXTEL0Ktp9nCfmT5xniWlaIQP8XIbiZp0tS6miAUKok2zbv3GymhdXhDtus8vur+Sfci7uIEUyq3Cj3ndRKlVgnD9r1lAfks7ZCQbx94MrnYNZom2bbL8BmVPWtBDd+ZL9Ze2KNBFVqaIBQqqTKHqZ6of0PzlrdALX/a2sRAJWC7RDYdkNtOu+43eBXGQI0C7E6l1ubmERkoIhsF5FdIjLexfPjRGSLiGwUkYUi0tDpuTdEJEpEtorI/0SHDqnyZtt8CAm/tG/3Xl5w6xQY+gU8sQXGbbWzj5dPsM9nL9Wp/72UC24LECLiDUwArgFaAbeJyNmJi9YBEcaYdsAs4A3HuT2BXkA7oA3QBejjrrIqVWIc32bXW/60vx2meim1h2xBYXYZz6p17czqrvfaVNzHtmgKbHVe7qxBdAV2GWP2GGPSgJnAEOcDjDGLjDHJjs0V2CVNAQzgD/gBFQBf4Jgby6pU8cnKhKObzx1+uvt3u87D76/ZY654Fi77W9G/fsTdth9i6bt2FJN2UKt8uDNA1AUOOm1HO/bl5x7gJwBjzHJgEXDE8fOzMWbr2SeIyFgRiRSRyJiYmLOfVqp4JMXY+QpnS0u2H/qZ6bn7Tp+AqTfDR71sTSFbchzMfsCm0/7bdhi7yA5TLeziOheiUnW7oM/GmWAytQah8lUihrmKyB1ABPCmY7sp0BJbo6gL9BWRy84+z7E+doQxJiIkRDvZlAfE7YWJV9gmobTkvM8tedsu6vNeJ1j1iW0y+ugyO5KoYS9H1tUvbU1i3qOQHAs3f5qbO8iduj0AOPodyvIcCHVJ3DmK6RDgnLSonmNfHiLSH3gO6GOMSXXsvhFYYYxJchzzE9AD+MuN5VXqwsTthS+uh9MxkJkKu36FVo5WVGNg83dQs7UdXvrjk3Z/UBiM+RVqtoLpw2DeY3Yy29Z5djjr+ZLOFaXgpnYtie0/ahOTypc7axCrgWYi0khE/IDhQJ61EEWkI/AxMNgYc9zpqQNAHxHxERFfbAe1izq8Uh4Svw8mD4L003DPz3b4aNT3uc8f22yHkHa9F+75Be76EfqMh7F/Qu32Nv31rZOhVhu7AlzYZdDjkeK9h6v/Cde+BYFa+1auua0GYYzJEJGHgZ8Bb2CSMSZKRF4BIo0xc7FNSoHAN45RrAeMMYOxI5r6ApuwHdYLjDHz3FVWpfIVu9umrahSJ3dfVhZ8cxekJcHIufZbf8tBsPFrSD9jj4/63qahbjnIDiEN62V/nFWoDLd/YzuLez5sh6QWp+qNbABTKh9unShnjPkR+PGsfS84/d4/n/MygfvcWTZVzp05CXMfge4PQMOero/JSIPPr7FrPj+wFCoG2f3rp9olOG/6NLdJqPUNsOZz2PmrDQpbvoew3hAQfP5yVA6Fgf8soptSqmiViE5qpYpVViZ8Owa2zoW/3s7/uG3zIOkYnDoE88fZfoUzJ22G1Prdoe0tucc27A2VatjAcCwKYncVbi0GpUowTbWhyp/fX7UdyjVbw55FcDoWAmqce9zqz6BaQ+h4hx1x1HygzY2UHAt3fJt39rG3j6OZ6RuoXBvEC8IHFd89KeUGWoNQZVPiUdf7N39nE9d1vgtu+timut4659zjjm2B/Uuhyz12slqDHvDD32DVx9BpJNTpcO45rW6wndYrP7LNS9r5q0o5DRCq9MqekezMGFj4Kvynhe38dXZwFcx5COp3g2vehNA2dmLapm/PvXbkJPCuYFNme3nDjR/bGoNvAPR74dzjwY5EqlTDBp1WNxTJLSrlSRogVOm14kM7I3nWPbZvwBibpuKvt6Bqffj1BdvkA3B0E0y7xU5Cu/VLm5NIxC56v38pnDqce93URNgw0+Yvym56CmoIo+bByNn5dzx7+0DLwY7RS4PdeutKFQcNEKr0ivrOfmOPmg0f9bajkv56yzYBPbzafqP//gGI/NzOaPYLhJFz8i6+0+YmwNhrZNv4NaQl2mU7ndXpAHU7n79M/V6Auxdo85IqEzRAqNLp5EE4tAZ6PmInonn5wLovbXC4/l07F2HYVLvM5PzHbe1i5Byo1iDvdYKbQa12sNnRzHRoje2jqNUO6kVceLkqVYf6XS/59pQqCXQUkyqdtjrmTbYcbJPN3b8EDiyHJv1yJ5xVrAYjZsHCl20gyW9N4ra32OaoOQ/D+ml2vYRr39I1ElS5pzUIVTJkZsD66Xmbes5n61zbyZydibRCIDQbcO5s5Kp14aaJUKtt/tdqfZN9XPelHd300Epo0O2Cb0GpskZrEKpw4vbA3Efh5s/ytuFfKmPsh/3CVyF2J+AYKdT8qvzPSTwKB1bAlc8WTRmq1YcbJ9rH/GZVK1UOaQ1CFc6WObDvL7uGwMXY8BVER+bdl5EG04bC1yPtUNKhX9jkdd+OgRO78r/W1nmAKdqRQu2HaXBQ6iwaIFTh7F9uHze7mDNQkBO7YPZYmxp772K7LyvLjjDa9Stc/X/wwDI7rHTYNDtcdObtkHLK9fW2zLHzF2qGX9y9KKUKRZuYVMGysuDgCjtM9MgG+4Ef3LTw56/6GLx87QiiabfCiK9h5y+weRb0exF6PJh7bFBDGDoZpgyBSVdD3U5QtYFdR6FmSzusdf9S9yzFqZTKQwOEKtjxLZCSYMf4L3zV1iKueLpw56Yk2M7ntrfAgFftGgpf3gRZ6dDlXuj9xLnnNLrMzlxe+aHNjprkYjlynYimlNtpgFB5HVoLqz+FQe/aRW3ADh8FaHMz7Fpov/n3ecr1MNC9f9l+hOzU2Oum2nUTut1vJ4+NmmdnNAc3h2v+nf9Q0nZD7Q9ARqpdve14lM2R5OVz/lFJSqkioQFC5bV+mv1pcY3NTgq2SadKXZvZtM3N8MM4u2La2R/Sh9fB5OshJBzu/B4Ca9rEdQ165Ca3CwyBsX9c2BwDnwq2v6FmuH19pVSx0E5qlVf0avu4ZrJ9NMZ2UDfoYT/UW91gcw256qyO/Bx8/CEhGj4fCCs/hpMHbO3BmU5AU6pU0AChcqUl2+yoFarCrt9sOov4vZB0FBr2sMcE1IAmV9oAYUzuuSmnYNMs29cwcq5NnvfzMzZpXvj1HrkdpdSl0QChch1eByYT+j5vt9dPyx3e2sBpjkCbW2zNYNdvufs2fW3XQuh8N9TrDKN/hBpNoc/TdtiqUqrU0QChcmU3L7W5GRpfAWu/tP0P/tVsv0K2VkOgRjO7tkJSjK1JRH5h+yTqdrLHhLaGR9ZApzuL+SaUUkVFv9qVJymnICPFdh67Er0aqje2zUidR8E3d9nU1037581x5FfJznr+pC98fz/0GQ/HNsF1b2v/gsqRnp5OdHQ0KSkpni6KAvz9/alXrx6+vr6FPkcDRHky9xE7k/meX87NbGqMDRCNr7DbLa6DSsGQfCK3/8FZrTYw8J92Gc5jW+wkuna3uv0WVOkRHR1N5cqVCQsLQ/SLg0cZY4iNjSU6OppGjRoV+jxtYipupw7Dkv/a2cnFKSsL9v4JZ+Jg6k3nrtmccNBOSKvXxW77+EGH2+zvDfLJURRxj52wlnjYdk5XqOy+8qtSJyUlhRo1amhwKAFEhBo1alxwbU4DRHFbMxl+exGObsy7P34/fNIPFr9pRwAVtRPb4Uy8nb18OtZOVnPOdZTd/5AdIAB6j4Pr/pP/wjkiMPg96Hqfpr5QLmlwKDku5r3QAFHcjmywj/uX5t2/ZQ4cirRrKr/TBn75B6QmFd3r7l9mH7s/ALdOts1C34yCrEy7PzoSfCrazuVslarbZTfP9w+rYjW49o1zV2pTSpV6GiCK25H19nHfWQFi758Q3MKujNb8Klj+PnxxLSS6yEOU7eAqO9Jo+QT449+5wceVA8shMNR2QjcbANe/A7t/h0X/zL1W3U656TWUUuWedlJfqt2/Q3oKhF9b8LGJxyDxCHhXgAPLbL+Al5ddF2H/Mugwwg4VvWUStBtuv+F/1h/u+C5vp3LaaVjwDKydnPf6S9+FO7513ansPBsa7Cil6NXw11tQu71t8ur+wMX/HZQqxzIyMvDxKXsfp269IxEZCLwLeAOfGmP+ddbz44AxQAYQA9xtjNnveK4B8ClQHzDAtcaYfe4s70X58SnAFC5AZH/DbzfUJrE7vsWOBjoUCenJ0LhP7rHNr4K75tv02J8NgHbDoGYrCAi26yfH7oZej0OXe6BCFXv+5MG2b+GOb6FB99xrnTwAp6Kh4aN5y3Ptm7ZGM+tum13Vuf9BqSL08rwothzOZ32Pi9SqThVeHNS6wONuuOEGDh48SEpKCo899hhjx45lwYIFPPvss2RmZhIcHMzChQtJSkrikUceITIyEhHhxRdf5OabbyYwMJCkJNvcO2vWLObPn88XX3zBXXfdhb+/P+vWraNXr14MHz6cxx57jJSUFCpWrMjnn39OixYtyMzM5Omnn2bBggV4eXlx77330rp1a/73v//x/fffA/Drr7/ywQcfMHt2IZfcLSZuCxAi4g1MAAYA0cBqEZlrjNnidNg6IMIYkywiDwBvAMMcz00BXjfG/CoigUAxD/sphLg9dplML1+7pnJBM4azA0S3B2yA2L/MBoi9i0G8IKx33uPrdoYxv9qlPtdOsUEAbOK8UfNsWuxsFavZgPLFdTD1ZlvryF5XOWc29Fk1C9+KcOuXMLGPTcutAUKVQZMmTaJ69eqcOXOGLl26MGTIEO69914WL15Mo0aNiIuLA+DVV1+latWqbNq0CYD4+PgCrx0dHc2yZcvw9vbm1KlT/PXXX/j4+PDbb7/x7LPP8u233zJx4kT27dvH+vXr8fHxIS4ujqCgIB588EFiYmIICQnh888/5+6773br3+FiuLMG0RXYZYzZAyAiM4EhQE6AMMYscjp+BXCH49hWgI8x5lfHcUXYW1uEdjpSTWSl22/oQWHnP/7IeqjexAaFqvVh/xLoNhb2/GmbebJTZDur3th+8Gdlwcl9ELvHjiqqWO3cYyvXglHz4fNr4Nt74KGV4Bdgm7MqVMnbAZ1z/UYwbKptKqtc68LuX6lCKsw3fXf53//+l/PN/ODBg0ycOJHLL788Zz5A9erVAfjtt9+YOTN3Sd2gIBf/H88ydOhQvL29AUhISGDUqFHs3LkTESE9PT3nuvfff39OE1T26915551MnTqV0aNHs3z5cqZMmVJEd1x03NlJXRc46LQd7diXn3uAnxy/NwdOish3IrJORN501EjyEJGxIhIpIpExMTFFVvBC2/Wr/eYPtjZRkCMbctNeN+xlaxCpSbYvoNHl5z/Xy8vRwdzfdXDIVqU23PChndew+C27b/9yqN/NrvvsSqPLof9LBZdfqVLmjz/+4LfffmP58uVs2LCBjh070qFDhwu6hvPw0LPnEQQEBOT8/o9//IMrr7ySzZs3M2/evALnHIwePZqpU6cyY8YMhg4dWiL7MErEKCYRuQOIAN507PIBLgOeBLoAjYG7zj7PGDPRGBNhjIkICQkpptI6pJ+xTUPZmUoLChCnY+2Hdu32djusF5yOsU1HWenQqM/5z78QDXtA+9th2XtwYIWdA+Gq41qpMi4hIYGgoCAqVarEtm3bWLFiBSkpKSxevJi9e/cC5DQxDRgwgAkTJuScm93EFBoaytatW8nKyjpvH0FCQgJ169rvwF988UXO/gEDBvDxxx+TkZGR5/Xq1KlDnTp1eO211xg9enTR3XQRcmeAOITtYM5Wz7EvDxHpDzwHDDbGpDp2RwPrjTF7jDEZwPdAJzeW9cLtW2rzGnUaaecPxO09//HZw1trd7CPDXvZxyXvgLffuf0Dl2rAKzZn0szb7XZ+s6GVKsMGDhxIRkYGLVu2ZPz48XTv3p2QkBAmTpzITTfdRPv27Rk2zHZ7Pv/888THx9OmTRvat2/PokW2Bfxf//oX119/PT179qR27dr5vtZTTz3FM888Q8eOHXOCAcCYMWNo0KAB7dq1o3379kyfPj3nuREjRlC/fn1atmzppr/AJTLGuOUHWwvYAzQC/IANQOuzjukI7AaanbXf23F8iGP7c+Ch871e586dTbH64e/GvBpqTFqyMRN6GDN9+PmPX/wfY16sYkxynN3OyjLmzeZ236Rr3VPGlRPt9V8JMSY9xT2voVQ+tmzZ4ukilHgPPfSQ+fTTT4vt9Vy9J0Ckyedz1W01CGO/+T8M/AxsBb42xkSJyCsikr3i/JtAIPCNiKwXkbmOczOxzUsLRWQTIMAn7iprjoxU+KAHfHYVrPrENgvlZ+cvdhSRb0Xb0Ru7+/zXPrLBdmJnd0SL2GYmKLj/4WJF3A11I2w5fSq45zWUUhelc+fObNy4kTvuuMPTRcmXW3tFjDE/Aj+ete8Fp9/7n+fcX4F27iudC3v+tHMTqtSDH5+EBePtusy9x0Ftp6LE7rYrrfV4yG5Xb2wDRlZm/h3BR9bn9j9kC7vMrsyWnUG1qHl5w10/uOfaSqlLsmbNGk8XoUAlr9vck7bOscNBH10LJ3bChhm2Ezlqtl0TIXuy2u6F9vimjvhWvTFkptlMrdXqn3vdM/EQv8/2VzjrMAIq14b6Xd13T77+7ru2UqpM0wCRLTMdtv0AzQfa5phabaDW63D53yHyM1j+Qd4lNms0tU1LYAME2JFM2QFi/XR7TuM+uQv0ZHdQZ/PxgxYD3XpbSil1sTRAZNu3xH7TbzUk7/6K1Wwq656PQuwuOBZlm6EaOo0Kcg4Q2ekyIifZWsOq7bZ2Aec2MSmlVAmmASLbljngGwBN+7l+3tsXara0P2erUtcm4ItzdFSfiYdDa+CyJ6HXYzZTa0aKzaOklFKlhAYIsJ3L2+bbBHm+FS/8fC8v29yUPRdizx9gsmywqRAI4dcVaXGVUqo4lIiZ1B53YLmd1dxycMHH5qd649zZ1LsWQoWqdoipUqrUCAwM9HQRShStQQBsmQs+/tDsqou/RvXGsHuRTaq3+3dofHnB2V2VKi9+Gg9HNxXtNWu1hWv+VfBxpVBJWV9CaxBZWbB1nh2yWuESvj1UbwQZZ2DvH3DqUO4QWKWUx4wfPz5PfqWXXnqJ1157jX79+tGpUyfatm3LnDlzCnWtpKSkfM+bMmVKTiqNO++8E4Bjx45x44030r59e9q3b8+yZcvYt28fbdq0yTnvrbfe4qWXXgLgiiuu4PHHHyciIoJ3332XefPm0a1bNzp27Ej//v05duxYTjlGjx5N27ZtadeuHd9++y2TJk3i8ccfz7nuJ598whNPPHGxf7Zc+U2xLm0/F51qI26vMf8KM2b9zIs7P9uuhTatxfTb7GP8gUu7nlKlXElItbF27Vpz+eWX52y3bNnSHDhwwCQkJBhjjImJiTFNmjQxWVlZxhhjAgIC8r1Wenq6y/M2b95smjVrZmJiYowxxsTGxhpjjLn11lvNO++8Y4wxJiMjw5w8edLs3bvXtG7dOueab775pnnxxReNMcb06dPHPPDAAznPxcXF5ZTrk08+MePGjTPGGPPUU0+Zxx57LM9xiYmJpnHjxiYtLc0YY0yPHj3Mxo0bz7mHC0214fk6jKcFhcGTO22n8qXIHuq64ycIbu56wpxSqlh17NiR48ePc/jwYWJiYggKCqJWrVo88cQTLF68GC8vLw4dOsSxY8eoVev866EYY3j22WfPOe/3339n6NChBAfbUYrZ6z38/vvvOWs8eHt7U7Vq1QIXIcpOHAh2MaJhw4Zx5MgR0tLSctavyG/dir59+zJ//nxatmxJeno6bdu2vcC/1rk0QEDR9BVUqWdXlstKhyb5DJVVShW7oUOHMmvWLI4ePcqwYcOYNm0aMTExrFmzBl9fX8LCwgpcuwG46POc+fj4kJWV+2X0fOtLPPLII4wbN47Bgwfzxx9/5DRF5WfMmDH885//JDw8vMjSh2sfRFHx9sldUS6/uRRKqWI3bNgwZs6cyaxZsxg6dCgJCQnUrFkTX19fFi1axP79+wt1nfzO69u3L9988w2xsTa5Z/Z6D/369ePDDz8EIDMzk4SEBEJDQzl+/DixsbGkpqYyf/78875e9voSkydPztmf37oV3bp14+DBg0yfPp3bbrutsH+e89IAUZSqN7YT5rLXelBKeVzr1q1JTEykbt261K5dmxEjRhAZGUnbtm2ZMmUK4eHhhbpOfue1bt2a5557jj59+tC+fXvGjRsHwLvvvsuiRYto27YtnTt3ZsuWLfj6+vLCCy/QtWtXBgwYcN7Xfumllxg6dCidO3fOab6C/NetALj11lvp1atXoZZLLQyxfRSlX0REhImMjPRsIfb8aedCRJTM1aGUKk5bt24tuQvhlFHXX389TzzxBP36uW7FcPWeiMgaY4zLSVtagyhKjftocFBKFbuTJ0/SvHlzKlasmG9wuBjaSa2UUk42bdqUM5chW4UKFVi5cqWHSlSwatWqsWPHjiK/rgYIpZTbGGMQEU8X44K0bduW9evXe7oYRe5iuhO0iUkp5Rb+/v7ExsZe1AeTKlrGGGJjY/H3v7AFxLQGoZRyi3r16hEdHU1MTIyni6KwAbtevXoXdI4GCKWUW/j6+ubM/lWlkzYxKaWUckkDhFJKKZc0QCillHKpzMykFpEYoHBJVVwLBk4UUXFKi/J4z1A+77s83jOUz/u+0HtuaIwJcfVEmQkQl0pEIvObbl5Wlcd7hvJ53+XxnqF83ndR3rM2MSmllHJJA4RSSimXNEDkmujpAnhAebxnKJ/3XR7vGcrnfRfZPWsfhFJKKZe0BqGUUsolDRBKKaVcKvcBQkQGish2EdklIuM9XR53EZH6IrJIRLaISJSIPObYX11EfhWRnY7HolmrsAQREW8RWSci8x3bjURkpeM9/0pE/DxdxqImItVEZJaIbBORrSLSo6y/1yLyhOPf9mYRmSEi/mXxvRaRSSJyXEQ2O+1z+d6K9T/H/W8UkU4X8lrlOkCIiDcwAbgGaAXcJiKtPFsqt8kA/maMaQV0Bx5y3Ot4YKExphmw0LFd1jwGbHXa/jfwjjGmKRAP3OORUrnXu8ACY0w40B57/2X2vRaRusCjQIQxpg3gDQynbL7XXwADz9qX33t7DdDM8TMW+PBCXqhcBwigK7DLGLPHGJMGzASGeLhMbmGMOWKMWev4PRH7gVEXe7+THYdNBm7wSAHdRETqAdcBnzq2BegLzHIcUhbvuSpwOfAZgDEmzRhzkjL+XmOzU1cUER+gEnCEMvheG2MWA3Fn7c7vvR0CTDHWCqCaiNQu7GuV9wBRFzjotB3t2FemiUgY0BFYCYQaY444njoKhHqqXG7yX+ApIMuxXQM4aYzJcGyXxfe8ERADfO5oWvtURAIow++1MeYQ8BZwABsYEoA1lP33Olt+7+0lfcaV9wBR7ohIIPAt8Lgx5pTzc8aOeS4z455F5HrguDFmjafLUsx8gE7Ah8aYjsBpzmpOKoPvdRD223IjoA4QwLnNMOVCUb635T1AHALqO23Xc+wrk0TEFxscphljvnPsPpZd5XQ8HvdU+dygFzBYRPZhmw/7YtvmqzmaIaBsvufRQLQxZqVjexY2YJTl97o/sNcYE2OMSQe+w77/Zf29zpbfe3tJn3HlPUCsBpo5Rjr4YTu15nq4TG7haHv/DNhqjHnb6am5wCjH76OAOcVdNncxxjxjjKlnjAnDvre/G2NGAIuAWxyHlal7BjDGHAUOikgLx65+wBbK8HuNbVrqLiKVHP/Ws++5TL/XTvJ7b+cCIx2jmboDCU5NUQUq9zOpReRabDu1NzDJGPO6Z0vkHiLSG/gL2ERue/yz2H6Ir4EG2HTptxpjzu4AK/VE5ArgSWPM9SLSGFujqA6sA+4wxqR6sHhFTkQ6YDvm/YA9wGjsF8Iy+16LyMvAMOyIvXXAGGx7e5l6r0VkBnAFNq33MeBF4HtcvLeOYPk+trktGRhtjIks9GuV9wChlFLKtfLexKSUUiofGiCUUkq5pAFCKaWUSxoglFJKuaQBQimllEsaIJS6ACKSKSLrnX6KLOGdiIQ5Z+hUytN8Cj5EKeXkjDGmg6cLoVRx0BqEUkVARPaJyBsisklEVolIU8f+MBH53ZGLf6GINHDsDxWR2SKywfHT03EpbxH5xLGuwS8iUtFjN6XKPQ0QSl2Yimc1MQ1zei7BGNMWO3P1v4597wGTjTHtgGnA/xz7/wf8aYxpj82TFOXY3wyYYIxpDZwEbnbr3Sh1HjqTWqkLICJJxphAF/v3AX2NMXscSRGPGmNqiMgJoLYxJt2x/4gxJlhEYoB6zmkfHGnYf3Us+oKIPA34GmNeK4ZbU+ocWoNQquiYfH6/EM55gjLRfkLlQRoglCo6w5welzt+X4bNJAswApswEeyykA9AzprZVYurkEoVln47UerCVBSR9U7bC4wx2UNdg0RkI7YWcJtj3yPYld3+jl3lbbRj/2PARBG5B1tTeAC7EppSJYb2QShVBBx9EBHGmBOeLotSRUWbmJRSSrmkNQillFIuaQ1CKaWUSxoglFJKuaQBQimllEsaIJRSSrmkAUIppZRL/w/ux/LmNqlpHgAAAABJRU5ErkJggg==\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "test_loss, test_acc = model.evaluate(test_images,  test_labels, verbose=2)\n",
        "print(test_acc)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "b7O304UHDbln",
        "outputId": "747f88e6-976b-48de-e6d0-0d2859671e05"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 - 1s - loss: 2.7652 - acc: 0.3114 - 878ms/epoch - 3ms/step\n",
            "0.31139999628067017\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "##6 - Learning Rate: 1e-05, Epochs/Steps: 100/500, GPU as Hardware Accelerator, Training time: 6.40 minutes, Accuracy: 31.9%"
      ],
      "metadata": {
        "id": "9S_DEQlhDo03"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model.compile(optimizer=optimizers.Adam(learning_rate=1e-05),\n",
        "              loss='categorical_crossentropy',\n",
        "              metrics=['acc'])\n",
        "print(\"Learning rate before fifth fit:\", model.optimizer.learning_rate.numpy())"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "z4vExtYMDsmV",
        "outputId": "67261369-fab0-419d-f1a4-0f880d9db5af"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Learning rate before fifth fit: 1e-05\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "training_start = time.time()\n",
        "history = model.fit(train_images, train_labels,\n",
        "                    steps_per_epoch=500,\n",
        "                    epochs=100,\n",
        "                    validation_split=0.2) \n",
        "training_stop = time.time()\n",
        "training_time = training_stop - training_start\n",
        "print(f\"Training time: {training_time}\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "D4el-zZFDxmd",
        "outputId": "be509e25-bce9-4bff-dee8-cf621e458f1b"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/100\n",
            "500/500 [==============================] - 5s 9ms/step - loss: 2.4705 - acc: 0.3686 - val_loss: 2.7722 - val_acc: 0.3118\n",
            "Epoch 2/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4659 - acc: 0.3706 - val_loss: 2.7708 - val_acc: 0.3117\n",
            "Epoch 3/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.4648 - acc: 0.3695 - val_loss: 2.7733 - val_acc: 0.3117\n",
            "Epoch 4/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4636 - acc: 0.3709 - val_loss: 2.7733 - val_acc: 0.3112\n",
            "Epoch 5/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4638 - acc: 0.3706 - val_loss: 2.7712 - val_acc: 0.3131\n",
            "Epoch 6/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4622 - acc: 0.3717 - val_loss: 2.7730 - val_acc: 0.3123\n",
            "Epoch 7/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4635 - acc: 0.3716 - val_loss: 2.7705 - val_acc: 0.3113\n",
            "Epoch 8/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4620 - acc: 0.3712 - val_loss: 2.7754 - val_acc: 0.3114\n",
            "Epoch 9/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4615 - acc: 0.3711 - val_loss: 2.7746 - val_acc: 0.3106\n",
            "Epoch 10/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4611 - acc: 0.3711 - val_loss: 2.7731 - val_acc: 0.3131\n",
            "Epoch 11/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4603 - acc: 0.3711 - val_loss: 2.7717 - val_acc: 0.3141\n",
            "Epoch 12/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4601 - acc: 0.3712 - val_loss: 2.7740 - val_acc: 0.3125\n",
            "Epoch 13/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.4593 - acc: 0.3702 - val_loss: 2.7707 - val_acc: 0.3150\n",
            "Epoch 14/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4596 - acc: 0.3707 - val_loss: 2.7708 - val_acc: 0.3117\n",
            "Epoch 15/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.4585 - acc: 0.3713 - val_loss: 2.7712 - val_acc: 0.3139\n",
            "Epoch 16/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.4582 - acc: 0.3715 - val_loss: 2.7732 - val_acc: 0.3126\n",
            "Epoch 17/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4580 - acc: 0.3711 - val_loss: 2.7699 - val_acc: 0.3129\n",
            "Epoch 18/100\n",
            "500/500 [==============================] - 5s 10ms/step - loss: 2.4576 - acc: 0.3721 - val_loss: 2.7711 - val_acc: 0.3125\n",
            "Epoch 19/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.4570 - acc: 0.3709 - val_loss: 2.7714 - val_acc: 0.3120\n",
            "Epoch 20/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4578 - acc: 0.3715 - val_loss: 2.7732 - val_acc: 0.3119\n",
            "Epoch 21/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.4567 - acc: 0.3713 - val_loss: 2.7693 - val_acc: 0.3132\n",
            "Epoch 22/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4555 - acc: 0.3725 - val_loss: 2.7713 - val_acc: 0.3114\n",
            "Epoch 23/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4558 - acc: 0.3715 - val_loss: 2.7720 - val_acc: 0.3129\n",
            "Epoch 24/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.4558 - acc: 0.3712 - val_loss: 2.7769 - val_acc: 0.3101\n",
            "Epoch 25/100\n",
            "500/500 [==============================] - 3s 7ms/step - loss: 2.4546 - acc: 0.3719 - val_loss: 2.7679 - val_acc: 0.3145\n",
            "Epoch 26/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.4549 - acc: 0.3724 - val_loss: 2.7681 - val_acc: 0.3140\n",
            "Epoch 27/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.4543 - acc: 0.3726 - val_loss: 2.7687 - val_acc: 0.3132\n",
            "Epoch 28/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.4544 - acc: 0.3721 - val_loss: 2.7730 - val_acc: 0.3115\n",
            "Epoch 29/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.4541 - acc: 0.3722 - val_loss: 2.7703 - val_acc: 0.3139\n",
            "Epoch 30/100\n",
            "500/500 [==============================] - 5s 9ms/step - loss: 2.4531 - acc: 0.3720 - val_loss: 2.7752 - val_acc: 0.3116\n",
            "Epoch 31/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.4528 - acc: 0.3725 - val_loss: 2.7690 - val_acc: 0.3121\n",
            "Epoch 32/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.4521 - acc: 0.3729 - val_loss: 2.7690 - val_acc: 0.3142\n",
            "Epoch 33/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4519 - acc: 0.3728 - val_loss: 2.7711 - val_acc: 0.3133\n",
            "Epoch 34/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.4521 - acc: 0.3724 - val_loss: 2.7708 - val_acc: 0.3128\n",
            "Epoch 35/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.4510 - acc: 0.3726 - val_loss: 2.7697 - val_acc: 0.3130\n",
            "Epoch 36/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4515 - acc: 0.3738 - val_loss: 2.7724 - val_acc: 0.3110\n",
            "Epoch 37/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.4506 - acc: 0.3740 - val_loss: 2.7713 - val_acc: 0.3140\n",
            "Epoch 38/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.4501 - acc: 0.3732 - val_loss: 2.7707 - val_acc: 0.3128\n",
            "Epoch 39/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4496 - acc: 0.3726 - val_loss: 2.7691 - val_acc: 0.3118\n",
            "Epoch 40/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.4492 - acc: 0.3728 - val_loss: 2.7702 - val_acc: 0.3123\n",
            "Epoch 41/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4494 - acc: 0.3745 - val_loss: 2.7688 - val_acc: 0.3124\n",
            "Epoch 42/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.4496 - acc: 0.3726 - val_loss: 2.7666 - val_acc: 0.3132\n",
            "Epoch 43/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4492 - acc: 0.3731 - val_loss: 2.7686 - val_acc: 0.3139\n",
            "Epoch 44/100\n",
            "500/500 [==============================] - 3s 7ms/step - loss: 2.4481 - acc: 0.3738 - val_loss: 2.7678 - val_acc: 0.3123\n",
            "Epoch 45/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4477 - acc: 0.3728 - val_loss: 2.7684 - val_acc: 0.3142\n",
            "Epoch 46/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4479 - acc: 0.3744 - val_loss: 2.7683 - val_acc: 0.3136\n",
            "Epoch 47/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4470 - acc: 0.3741 - val_loss: 2.7666 - val_acc: 0.3137\n",
            "Epoch 48/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.4468 - acc: 0.3734 - val_loss: 2.7675 - val_acc: 0.3130\n",
            "Epoch 49/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4462 - acc: 0.3735 - val_loss: 2.7645 - val_acc: 0.3149\n",
            "Epoch 50/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.4463 - acc: 0.3745 - val_loss: 2.7661 - val_acc: 0.3113\n",
            "Epoch 51/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4459 - acc: 0.3728 - val_loss: 2.7662 - val_acc: 0.3145\n",
            "Epoch 52/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4460 - acc: 0.3732 - val_loss: 2.7681 - val_acc: 0.3144\n",
            "Epoch 53/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.4447 - acc: 0.3742 - val_loss: 2.7656 - val_acc: 0.3137\n",
            "Epoch 54/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.4448 - acc: 0.3740 - val_loss: 2.7635 - val_acc: 0.3152\n",
            "Epoch 55/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.4444 - acc: 0.3753 - val_loss: 2.7655 - val_acc: 0.3156\n",
            "Epoch 56/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4445 - acc: 0.3742 - val_loss: 2.7631 - val_acc: 0.3137\n",
            "Epoch 57/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4442 - acc: 0.3756 - val_loss: 2.7654 - val_acc: 0.3138\n",
            "Epoch 58/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.4434 - acc: 0.3745 - val_loss: 2.7644 - val_acc: 0.3143\n",
            "Epoch 59/100\n",
            "500/500 [==============================] - 3s 7ms/step - loss: 2.4431 - acc: 0.3751 - val_loss: 2.7658 - val_acc: 0.3128\n",
            "Epoch 60/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.4427 - acc: 0.3748 - val_loss: 2.7640 - val_acc: 0.3165\n",
            "Epoch 61/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4422 - acc: 0.3753 - val_loss: 2.7645 - val_acc: 0.3126\n",
            "Epoch 62/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.4425 - acc: 0.3746 - val_loss: 2.7647 - val_acc: 0.3144\n",
            "Epoch 63/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4416 - acc: 0.3746 - val_loss: 2.7630 - val_acc: 0.3144\n",
            "Epoch 64/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4416 - acc: 0.3758 - val_loss: 2.7661 - val_acc: 0.3146\n",
            "Epoch 65/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4409 - acc: 0.3752 - val_loss: 2.7687 - val_acc: 0.3130\n",
            "Epoch 66/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.4418 - acc: 0.3745 - val_loss: 2.7656 - val_acc: 0.3148\n",
            "Epoch 67/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4397 - acc: 0.3757 - val_loss: 2.7679 - val_acc: 0.3150\n",
            "Epoch 68/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.4400 - acc: 0.3753 - val_loss: 2.7670 - val_acc: 0.3118\n",
            "Epoch 69/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.4392 - acc: 0.3744 - val_loss: 2.7642 - val_acc: 0.3159\n",
            "Epoch 70/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4392 - acc: 0.3744 - val_loss: 2.7621 - val_acc: 0.3141\n",
            "Epoch 71/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4393 - acc: 0.3761 - val_loss: 2.7660 - val_acc: 0.3149\n",
            "Epoch 72/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4388 - acc: 0.3754 - val_loss: 2.7623 - val_acc: 0.3148\n",
            "Epoch 73/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4389 - acc: 0.3758 - val_loss: 2.7656 - val_acc: 0.3140\n",
            "Epoch 74/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4382 - acc: 0.3753 - val_loss: 2.7625 - val_acc: 0.3148\n",
            "Epoch 75/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4385 - acc: 0.3762 - val_loss: 2.7616 - val_acc: 0.3159\n",
            "Epoch 76/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4373 - acc: 0.3756 - val_loss: 2.7613 - val_acc: 0.3157\n",
            "Epoch 77/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4372 - acc: 0.3756 - val_loss: 2.7640 - val_acc: 0.3154\n",
            "Epoch 78/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.4366 - acc: 0.3767 - val_loss: 2.7610 - val_acc: 0.3136\n",
            "Epoch 79/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.4365 - acc: 0.3748 - val_loss: 2.7616 - val_acc: 0.3158\n",
            "Epoch 80/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4361 - acc: 0.3761 - val_loss: 2.7608 - val_acc: 0.3150\n",
            "Epoch 81/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.4352 - acc: 0.3755 - val_loss: 2.7630 - val_acc: 0.3151\n",
            "Epoch 82/100\n",
            "500/500 [==============================] - 3s 7ms/step - loss: 2.4354 - acc: 0.3753 - val_loss: 2.7616 - val_acc: 0.3156\n",
            "Epoch 83/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.4353 - acc: 0.3772 - val_loss: 2.7662 - val_acc: 0.3130\n",
            "Epoch 84/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4353 - acc: 0.3759 - val_loss: 2.7626 - val_acc: 0.3138\n",
            "Epoch 85/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.4342 - acc: 0.3759 - val_loss: 2.7612 - val_acc: 0.3159\n",
            "Epoch 86/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.4346 - acc: 0.3776 - val_loss: 2.7602 - val_acc: 0.3144\n",
            "Epoch 87/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4344 - acc: 0.3762 - val_loss: 2.7607 - val_acc: 0.3152\n",
            "Epoch 88/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.4337 - acc: 0.3771 - val_loss: 2.7578 - val_acc: 0.3165\n",
            "Epoch 89/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4337 - acc: 0.3760 - val_loss: 2.7601 - val_acc: 0.3170\n",
            "Epoch 90/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4327 - acc: 0.3772 - val_loss: 2.7632 - val_acc: 0.3143\n",
            "Epoch 91/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4329 - acc: 0.3755 - val_loss: 2.7618 - val_acc: 0.3158\n",
            "Epoch 92/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 2.4324 - acc: 0.3760 - val_loss: 2.7631 - val_acc: 0.3138\n",
            "Epoch 93/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4322 - acc: 0.3769 - val_loss: 2.7598 - val_acc: 0.3161\n",
            "Epoch 94/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4312 - acc: 0.3765 - val_loss: 2.7620 - val_acc: 0.3166\n",
            "Epoch 95/100\n",
            "500/500 [==============================] - 5s 9ms/step - loss: 2.4313 - acc: 0.3769 - val_loss: 2.7602 - val_acc: 0.3167\n",
            "Epoch 96/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4316 - acc: 0.3756 - val_loss: 2.7587 - val_acc: 0.3156\n",
            "Epoch 97/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4312 - acc: 0.3765 - val_loss: 2.7584 - val_acc: 0.3164\n",
            "Epoch 98/100\n",
            "500/500 [==============================] - 5s 10ms/step - loss: 2.4305 - acc: 0.3769 - val_loss: 2.7662 - val_acc: 0.3143\n",
            "Epoch 99/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 2.4304 - acc: 0.3762 - val_loss: 2.7574 - val_acc: 0.3163\n",
            "Epoch 100/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 2.4294 - acc: 0.3771 - val_loss: 2.7617 - val_acc: 0.3151\n",
            "Training time: 399.33709931373596\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.plot(history.history['acc'], label='accuracy')\n",
        "plt.plot(history.history['val_acc'], label = 'val_accuracy')\n",
        "plt.xlabel('Epoch')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.legend(loc='lower right')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 298
        },
        "id": "Ys0p0GPcFdqF",
        "outputId": "e91518cf-d3e2-43d5-9888-2d1fe8993f17"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7f478c8c5a00>"
            ]
          },
          "metadata": {},
          "execution_count": 66
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "test_loss, test_acc = model.evaluate(test_images,  test_labels, verbose=2)\n",
        "print(test_acc)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "jtBPfCsmFiFU",
        "outputId": "3f491bd9-ebd0-4e6a-f277-bcb3ea87ca8a"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 - 1s - loss: 2.7311 - acc: 0.3192 - 1s/epoch - 3ms/step\n",
            "0.3192000091075897\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [],
      "metadata": {
        "id": "lNo4CCVQ-SAJ"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "##7 - Learning Rate: 1e-06, Epochs/Steps: 100/500, GPU as Hardware Accelerator, Training time: 7.40 minutes, Accuracy: 23%"
      ],
      "metadata": {
        "id": "eIBd_W7k-Slb"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model.compile(optimizer=optimizers.Adam(learning_rate=1e-06),\n",
        "              loss='categorical_crossentropy',\n",
        "              metrics=['acc'])\n",
        "print(\"Learning rate before seventh fit:\", model.optimizer.learning_rate.numpy())"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "495c75e3-094b-47b1-961d-16d4bd3644de",
        "id": "U_-LdhFh-Slc"
      },
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Learning rate before seventh fit: 1e-06\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "training_start = time.time()\n",
        "history = model.fit(train_images, train_labels,\n",
        "                    steps_per_epoch=500,\n",
        "                    epochs=100,\n",
        "                    validation_split=0.2) \n",
        "training_stop = time.time()\n",
        "training_time = training_stop - training_start\n",
        "print(f\"Training time: {training_time}\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "a4a77bea-84ac-4792-dc50-e96f2da77215",
        "id": "XPZC4xLO-Slc"
      },
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/100\n",
            "500/500 [==============================] - 14s 9ms/step - loss: 4.6050 - acc: 0.0080 - val_loss: 4.6056 - val_acc: 0.0061\n",
            "Epoch 2/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 4.6049 - acc: 0.0080 - val_loss: 4.6055 - val_acc: 0.0062\n",
            "Epoch 3/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.6048 - acc: 0.0079 - val_loss: 4.6054 - val_acc: 0.0061\n",
            "Epoch 4/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.6046 - acc: 0.0079 - val_loss: 4.6053 - val_acc: 0.0065\n",
            "Epoch 5/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 4.6046 - acc: 0.0079 - val_loss: 4.6052 - val_acc: 0.0067\n",
            "Epoch 6/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.6045 - acc: 0.0081 - val_loss: 4.6052 - val_acc: 0.0066\n",
            "Epoch 7/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.6044 - acc: 0.0081 - val_loss: 4.6051 - val_acc: 0.0066\n",
            "Epoch 8/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 4.6043 - acc: 0.0082 - val_loss: 4.6051 - val_acc: 0.0066\n",
            "Epoch 9/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 4.6042 - acc: 0.0080 - val_loss: 4.6050 - val_acc: 0.0065\n",
            "Epoch 10/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.6041 - acc: 0.0080 - val_loss: 4.6049 - val_acc: 0.0066\n",
            "Epoch 11/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.6041 - acc: 0.0079 - val_loss: 4.6049 - val_acc: 0.0068\n",
            "Epoch 12/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.6040 - acc: 0.0078 - val_loss: 4.6048 - val_acc: 0.0068\n",
            "Epoch 13/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.6039 - acc: 0.0078 - val_loss: 4.6047 - val_acc: 0.0069\n",
            "Epoch 14/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.6038 - acc: 0.0077 - val_loss: 4.6047 - val_acc: 0.0071\n",
            "Epoch 15/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.6037 - acc: 0.0077 - val_loss: 4.6046 - val_acc: 0.0071\n",
            "Epoch 16/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.6037 - acc: 0.0076 - val_loss: 4.6046 - val_acc: 0.0071\n",
            "Epoch 17/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 4.6036 - acc: 0.0074 - val_loss: 4.6045 - val_acc: 0.0068\n",
            "Epoch 18/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.6035 - acc: 0.0072 - val_loss: 4.6044 - val_acc: 0.0066\n",
            "Epoch 19/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.6034 - acc: 0.0072 - val_loss: 4.6043 - val_acc: 0.0066\n",
            "Epoch 20/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 4.6033 - acc: 0.0073 - val_loss: 4.6043 - val_acc: 0.0065\n",
            "Epoch 21/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.6032 - acc: 0.0072 - val_loss: 4.6042 - val_acc: 0.0062\n",
            "Epoch 22/100\n",
            "500/500 [==============================] - 6s 11ms/step - loss: 4.6031 - acc: 0.0074 - val_loss: 4.6041 - val_acc: 0.0063\n",
            "Epoch 23/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 4.6030 - acc: 0.0074 - val_loss: 4.6040 - val_acc: 0.0062\n",
            "Epoch 24/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 4.6029 - acc: 0.0074 - val_loss: 4.6039 - val_acc: 0.0061\n",
            "Epoch 25/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.6028 - acc: 0.0074 - val_loss: 4.6038 - val_acc: 0.0060\n",
            "Epoch 26/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.6027 - acc: 0.0076 - val_loss: 4.6037 - val_acc: 0.0060\n",
            "Epoch 27/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.6025 - acc: 0.0076 - val_loss: 4.6036 - val_acc: 0.0061\n",
            "Epoch 28/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 4.6024 - acc: 0.0075 - val_loss: 4.6035 - val_acc: 0.0060\n",
            "Epoch 29/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.6023 - acc: 0.0076 - val_loss: 4.6034 - val_acc: 0.0060\n",
            "Epoch 30/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 4.6021 - acc: 0.0078 - val_loss: 4.6033 - val_acc: 0.0060\n",
            "Epoch 31/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.6020 - acc: 0.0077 - val_loss: 4.6031 - val_acc: 0.0061\n",
            "Epoch 32/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 4.6018 - acc: 0.0078 - val_loss: 4.6030 - val_acc: 0.0061\n",
            "Epoch 33/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.6017 - acc: 0.0078 - val_loss: 4.6029 - val_acc: 0.0061\n",
            "Epoch 34/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 4.6015 - acc: 0.0080 - val_loss: 4.6027 - val_acc: 0.0063\n",
            "Epoch 35/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 4.6013 - acc: 0.0081 - val_loss: 4.6026 - val_acc: 0.0066\n",
            "Epoch 36/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.6011 - acc: 0.0081 - val_loss: 4.6024 - val_acc: 0.0068\n",
            "Epoch 37/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.6009 - acc: 0.0082 - val_loss: 4.6022 - val_acc: 0.0072\n",
            "Epoch 38/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.6007 - acc: 0.0082 - val_loss: 4.6020 - val_acc: 0.0073\n",
            "Epoch 39/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.6005 - acc: 0.0085 - val_loss: 4.6018 - val_acc: 0.0078\n",
            "Epoch 40/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.6002 - acc: 0.0086 - val_loss: 4.6016 - val_acc: 0.0077\n",
            "Epoch 41/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 4.6000 - acc: 0.0090 - val_loss: 4.6014 - val_acc: 0.0078\n",
            "Epoch 42/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5997 - acc: 0.0094 - val_loss: 4.6011 - val_acc: 0.0079\n",
            "Epoch 43/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 4.5994 - acc: 0.0095 - val_loss: 4.6009 - val_acc: 0.0082\n",
            "Epoch 44/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5991 - acc: 0.0098 - val_loss: 4.6006 - val_acc: 0.0081\n",
            "Epoch 45/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5988 - acc: 0.0100 - val_loss: 4.6003 - val_acc: 0.0080\n",
            "Epoch 46/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5985 - acc: 0.0100 - val_loss: 4.6000 - val_acc: 0.0082\n",
            "Epoch 47/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5981 - acc: 0.0104 - val_loss: 4.5997 - val_acc: 0.0084\n",
            "Epoch 48/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5977 - acc: 0.0105 - val_loss: 4.5993 - val_acc: 0.0086\n",
            "Epoch 49/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5973 - acc: 0.0110 - val_loss: 4.5990 - val_acc: 0.0086\n",
            "Epoch 50/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 4.5969 - acc: 0.0115 - val_loss: 4.5986 - val_acc: 0.0089\n",
            "Epoch 51/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 4.5965 - acc: 0.0119 - val_loss: 4.5982 - val_acc: 0.0094\n",
            "Epoch 52/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5961 - acc: 0.0120 - val_loss: 4.5978 - val_acc: 0.0096\n",
            "Epoch 53/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5956 - acc: 0.0123 - val_loss: 4.5974 - val_acc: 0.0099\n",
            "Epoch 54/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5951 - acc: 0.0128 - val_loss: 4.5968 - val_acc: 0.0105\n",
            "Epoch 55/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5945 - acc: 0.0128 - val_loss: 4.5963 - val_acc: 0.0110\n",
            "Epoch 56/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5939 - acc: 0.0134 - val_loss: 4.5958 - val_acc: 0.0113\n",
            "Epoch 57/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5933 - acc: 0.0135 - val_loss: 4.5952 - val_acc: 0.0115\n",
            "Epoch 58/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5926 - acc: 0.0137 - val_loss: 4.5946 - val_acc: 0.0120\n",
            "Epoch 59/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5918 - acc: 0.0138 - val_loss: 4.5939 - val_acc: 0.0119\n",
            "Epoch 60/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5911 - acc: 0.0143 - val_loss: 4.5932 - val_acc: 0.0120\n",
            "Epoch 61/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 4.5903 - acc: 0.0145 - val_loss: 4.5925 - val_acc: 0.0128\n",
            "Epoch 62/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 4.5895 - acc: 0.0149 - val_loss: 4.5916 - val_acc: 0.0133\n",
            "Epoch 63/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5885 - acc: 0.0152 - val_loss: 4.5908 - val_acc: 0.0139\n",
            "Epoch 64/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5876 - acc: 0.0162 - val_loss: 4.5899 - val_acc: 0.0147\n",
            "Epoch 65/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5866 - acc: 0.0166 - val_loss: 4.5890 - val_acc: 0.0150\n",
            "Epoch 66/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5855 - acc: 0.0170 - val_loss: 4.5880 - val_acc: 0.0153\n",
            "Epoch 67/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 4.5845 - acc: 0.0174 - val_loss: 4.5870 - val_acc: 0.0157\n",
            "Epoch 68/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 4.5833 - acc: 0.0178 - val_loss: 4.5859 - val_acc: 0.0162\n",
            "Epoch 69/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 4.5822 - acc: 0.0180 - val_loss: 4.5849 - val_acc: 0.0169\n",
            "Epoch 70/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5809 - acc: 0.0183 - val_loss: 4.5838 - val_acc: 0.0174\n",
            "Epoch 71/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 4.5797 - acc: 0.0188 - val_loss: 4.5825 - val_acc: 0.0176\n",
            "Epoch 72/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5783 - acc: 0.0191 - val_loss: 4.5814 - val_acc: 0.0177\n",
            "Epoch 73/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 4.5770 - acc: 0.0194 - val_loss: 4.5801 - val_acc: 0.0180\n",
            "Epoch 74/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 4.5756 - acc: 0.0198 - val_loss: 4.5788 - val_acc: 0.0184\n",
            "Epoch 75/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5741 - acc: 0.0201 - val_loss: 4.5774 - val_acc: 0.0188\n",
            "Epoch 76/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5726 - acc: 0.0205 - val_loss: 4.5760 - val_acc: 0.0195\n",
            "Epoch 77/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5710 - acc: 0.0208 - val_loss: 4.5745 - val_acc: 0.0192\n",
            "Epoch 78/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5694 - acc: 0.0212 - val_loss: 4.5730 - val_acc: 0.0192\n",
            "Epoch 79/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5677 - acc: 0.0213 - val_loss: 4.5714 - val_acc: 0.0195\n",
            "Epoch 80/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 4.5660 - acc: 0.0219 - val_loss: 4.5698 - val_acc: 0.0199\n",
            "Epoch 81/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5642 - acc: 0.0221 - val_loss: 4.5681 - val_acc: 0.0202\n",
            "Epoch 82/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 4.5624 - acc: 0.0229 - val_loss: 4.5663 - val_acc: 0.0203\n",
            "Epoch 83/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 4.5605 - acc: 0.0230 - val_loss: 4.5647 - val_acc: 0.0206\n",
            "Epoch 84/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5586 - acc: 0.0235 - val_loss: 4.5629 - val_acc: 0.0214\n",
            "Epoch 85/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5567 - acc: 0.0237 - val_loss: 4.5611 - val_acc: 0.0219\n",
            "Epoch 86/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5547 - acc: 0.0236 - val_loss: 4.5592 - val_acc: 0.0218\n",
            "Epoch 87/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 4.5527 - acc: 0.0239 - val_loss: 4.5574 - val_acc: 0.0221\n",
            "Epoch 88/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5506 - acc: 0.0238 - val_loss: 4.5555 - val_acc: 0.0220\n",
            "Epoch 89/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5486 - acc: 0.0241 - val_loss: 4.5535 - val_acc: 0.0222\n",
            "Epoch 90/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 4.5464 - acc: 0.0246 - val_loss: 4.5515 - val_acc: 0.0227\n",
            "Epoch 91/100\n",
            "500/500 [==============================] - 4s 9ms/step - loss: 4.5443 - acc: 0.0247 - val_loss: 4.5495 - val_acc: 0.0228\n",
            "Epoch 92/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5421 - acc: 0.0251 - val_loss: 4.5474 - val_acc: 0.0229\n",
            "Epoch 93/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5398 - acc: 0.0253 - val_loss: 4.5453 - val_acc: 0.0226\n",
            "Epoch 94/100\n",
            "500/500 [==============================] - 5s 10ms/step - loss: 4.5376 - acc: 0.0253 - val_loss: 4.5433 - val_acc: 0.0226\n",
            "Epoch 95/100\n",
            "500/500 [==============================] - 4s 7ms/step - loss: 4.5353 - acc: 0.0257 - val_loss: 4.5410 - val_acc: 0.0233\n",
            "Epoch 96/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5330 - acc: 0.0256 - val_loss: 4.5389 - val_acc: 0.0230\n",
            "Epoch 97/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5306 - acc: 0.0255 - val_loss: 4.5366 - val_acc: 0.0237\n",
            "Epoch 98/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5283 - acc: 0.0255 - val_loss: 4.5344 - val_acc: 0.0232\n",
            "Epoch 99/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5258 - acc: 0.0257 - val_loss: 4.5322 - val_acc: 0.0230\n",
            "Epoch 100/100\n",
            "500/500 [==============================] - 4s 8ms/step - loss: 4.5234 - acc: 0.0255 - val_loss: 4.5300 - val_acc: 0.0232\n",
            "Training time: 445.8565990924835\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.plot(history.history['acc'], label='accuracy')\n",
        "plt.plot(history.history['val_acc'], label = 'val_accuracy')\n",
        "plt.xlabel('Epoch')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.legend(loc='lower right')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 296
        },
        "outputId": "d3df58f0-8725-45fb-a26a-e3785cb74f77",
        "id": "5NuwTkzU-Sld"
      },
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7f44480c4490>"
            ]
          },
          "metadata": {},
          "execution_count": 20
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "test_loss, test_acc = model.evaluate(test_images,  test_labels, verbose=2)\n",
        "print(test_acc)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "1b466824-f673-46da-db8c-67834a664141",
        "id": "PigDLxx2-Sld"
      },
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "313/313 - 1s - loss: 4.5255 - acc: 0.0230 - 918ms/epoch - 3ms/step\n",
            "0.023000000044703484\n"
          ]
        }
      ]
    }
  ]
}