CNN_mfcc_latest.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "CNN-mfcc-latest.ipynb",
      "provenance": [],
      "machine_shape": "hm",
      "mount_file_id": "1HTEPwkxW4PkX2rZ8rxNMD0wsoAurfYzM",
      "authorship_tag": "ABX9TyOKpZ4ePV2HmIL7aVM/LyN+",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "TPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/DarshanGowda0/MusicalInstrumentsClassification/blob/main/CNN_mfcc_latest.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "S398m7sh87kI"
      },
      "source": [
        "# Connect to TPU\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "7DRa5YeI2Wg7",
        "outputId": "2bf7706d-7090-4ab0-efda-6855b4a94e82"
      },
      "source": [
        "%tensorflow_version 2.x\n",
        "import tensorflow as tf\n",
        "print(\"Tensorflow version \" + tf.__version__)\n",
        "\n",
        "try:\n",
        "  tpu = tf.distribute.cluster_resolver.TPUClusterResolver()  # TPU detection\n",
        "  print('Running on TPU ', tpu.cluster_spec().as_dict()['worker'])\n",
        "except ValueError:\n",
        "  raise BaseException('ERROR: Not connected to a TPU runtime; please see the previous cell in this notebook for instructions!')\n",
        "\n",
        "tf.config.experimental_connect_to_cluster(tpu)\n",
        "tf.tpu.experimental.initialize_tpu_system(tpu)\n",
        "tpu_strategy = tf.distribute.experimental.TPUStrategy(tpu)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Tensorflow version 2.4.1\n",
            "Running on TPU  ['10.114.90.50:8470']\n",
            "INFO:tensorflow:Initializing the TPU system: grpc://10.114.90.50:8470\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:Initializing the TPU system: grpc://10.114.90.50:8470\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:Clearing out eager caches\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:Clearing out eager caches\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:Finished initializing TPU system.\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:Finished initializing TPU system.\n",
            "WARNING:absl:`tf.distribute.experimental.TPUStrategy` is deprecated, please use  the non experimental symbol `tf.distribute.TPUStrategy` instead.\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:Found TPU system:\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:Found TPU system:\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Num TPU Cores: 8\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Num TPU Cores: 8\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Num TPU Workers: 1\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Num TPU Workers: 1\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Num TPU Cores Per Worker: 8\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Num TPU Cores Per Worker: 8\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Available Device: _DeviceAttributes(/job:localhost/replica:0/task:0/device:CPU:0, CPU, 0, 0)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Available Device: _DeviceAttributes(/job:localhost/replica:0/task:0/device:CPU:0, CPU, 0, 0)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Available Device: _DeviceAttributes(/job:worker/replica:0/task:0/device:CPU:0, CPU, 0, 0)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Available Device: _DeviceAttributes(/job:worker/replica:0/task:0/device:CPU:0, CPU, 0, 0)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Available Device: _DeviceAttributes(/job:worker/replica:0/task:0/device:TPU:0, TPU, 0, 0)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Available Device: _DeviceAttributes(/job:worker/replica:0/task:0/device:TPU:0, TPU, 0, 0)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Available Device: _DeviceAttributes(/job:worker/replica:0/task:0/device:TPU:1, TPU, 0, 0)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Available Device: _DeviceAttributes(/job:worker/replica:0/task:0/device:TPU:1, TPU, 0, 0)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Available Device: _DeviceAttributes(/job:worker/replica:0/task:0/device:TPU:2, TPU, 0, 0)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Available Device: _DeviceAttributes(/job:worker/replica:0/task:0/device:TPU:2, TPU, 0, 0)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Available Device: _DeviceAttributes(/job:worker/replica:0/task:0/device:TPU:3, TPU, 0, 0)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Available Device: _DeviceAttributes(/job:worker/replica:0/task:0/device:TPU:3, TPU, 0, 0)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Available Device: _DeviceAttributes(/job:worker/replica:0/task:0/device:TPU:4, TPU, 0, 0)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Available Device: _DeviceAttributes(/job:worker/replica:0/task:0/device:TPU:4, TPU, 0, 0)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Available Device: _DeviceAttributes(/job:worker/replica:0/task:0/device:TPU:5, TPU, 0, 0)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Available Device: _DeviceAttributes(/job:worker/replica:0/task:0/device:TPU:5, TPU, 0, 0)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Available Device: _DeviceAttributes(/job:worker/replica:0/task:0/device:TPU:6, TPU, 0, 0)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Available Device: _DeviceAttributes(/job:worker/replica:0/task:0/device:TPU:6, TPU, 0, 0)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Available Device: _DeviceAttributes(/job:worker/replica:0/task:0/device:TPU:7, TPU, 0, 0)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Available Device: _DeviceAttributes(/job:worker/replica:0/task:0/device:TPU:7, TPU, 0, 0)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Available Device: _DeviceAttributes(/job:worker/replica:0/task:0/device:TPU_SYSTEM:0, TPU_SYSTEM, 0, 0)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Available Device: _DeviceAttributes(/job:worker/replica:0/task:0/device:TPU_SYSTEM:0, TPU_SYSTEM, 0, 0)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Available Device: _DeviceAttributes(/job:worker/replica:0/task:0/device:XLA_CPU:0, XLA_CPU, 0, 0)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:*** Available Device: _DeviceAttributes(/job:worker/replica:0/task:0/device:XLA_CPU:0, XLA_CPU, 0, 0)\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "YWjKld859CuR"
      },
      "source": [
        "# Load preprocessed data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "tcShzxg82mMZ",
        "outputId": "90a46cca-1f98-49af-8acb-b7f1922325bf"
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "2ypW1aUd2IYE",
        "outputId": "167f8141-89fc-4daf-8dd4-766f56d8585c"
      },
      "source": [
        "import pandas as pd\n",
        "import numpy as np\n",
        "import librosa, librosa.display\n",
        "import os\n",
        "from tqdm.notebook import tqdm\n",
        "from sklearn.preprocessing import LabelEncoder\n",
        "le = LabelEncoder()\n",
        "\n",
        "!pip install --upgrade tables\n",
        "\n",
        "df = pd.read_hdf('/content/drive/MyDrive/mfcc-16-128-and-other-features.h5', key='s', mode='r')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Collecting tables\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/0f/cb/4097be890a773af95343389faa8c283b0d9ff606f144227a548461dcbdd5/tables-3.6.1-cp37-cp37m-manylinux1_x86_64.whl (4.3MB)\n",
            "\u001b[K     |████████████████████████████████| 4.3MB 6.2MB/s \n",
            "\u001b[?25hRequirement already satisfied, skipping upgrade: numpy>=1.9.3 in /usr/local/lib/python3.7/dist-packages (from tables) (1.19.5)\n",
            "Requirement already satisfied, skipping upgrade: numexpr>=2.6.2 in /usr/local/lib/python3.7/dist-packages (from tables) (2.7.3)\n",
            "Installing collected packages: tables\n",
            "  Found existing installation: tables 3.4.4\n",
            "    Uninstalling tables-3.4.4:\n",
            "      Successfully uninstalled tables-3.4.4\n",
            "Successfully installed tables-3.6.1\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pjpntmC72tcb",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "1728b0bc-f4b6-4c70-adf9-cdbbf385d511"
      },
      "source": [
        "from sklearn.utils import shuffle\n",
        "\n",
        "INSTRUMENTS = ['flu', 'pia', 'tru', 'gac', 'voi', 'org']\n",
        "df = df.loc[df['label'].isin(INSTRUMENTS)]\n",
        "\n",
        "df = shuffle(df)\n",
        "print(df.columns)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Index(['mfcc', 'chroma_stft', 'rms', 'spec_centroid', 'spec_bandwidth',\n",
            "       'rolloff', 'zero_crossing_rate', 'label'],\n",
            "      dtype='object')\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3BepSqzh9LY3"
      },
      "source": [
        "# Normalize the data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VB4LIRBw27Z6"
      },
      "source": [
        "X, y = df['mfcc'], df['label']\n",
        "y_cat = le.fit_transform(y)\n",
        "X = np.stack([x for x in X])\n",
        "X = np.expand_dims(X, axis=-1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "e6BzOGBg3L40",
        "outputId": "b3e1c5e8-9684-4775-9050-8219d61388b8"
      },
      "source": [
        "from collections import Counter\n",
        "print(X.shape, Counter(y))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(3846, 128, 16, 1) Counter({'voi': 778, 'pia': 721, 'org': 682, 'gac': 637, 'tru': 577, 'flu': 451})\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NkEqGKIE3YAt"
      },
      "source": [
        "mean = np.mean(X, axis=(1,2), keepdims=True)\n",
        "std = np.std(X, axis=(1,2), keepdims=True)\n",
        "X_std = (X - mean) / std"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "x1gTySB69PaM"
      },
      "source": [
        "## Split train and test"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Yj3OOJjy6Yuh",
        "outputId": "f17552bf-282a-42c0-83b6-4d4af729c86e"
      },
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "import tensorflow.keras as keras\n",
        "\n",
        "X_train, X_test, y_train, y_test = train_test_split(X_std, y_cat, test_size=0.15)\n",
        "X_train.shape, y_train.shape, X_test.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "((3269, 128, 16, 1), (3269,), (577, 128, 16, 1))"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 11
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "C2pS9tl49S59"
      },
      "source": [
        "# CNN Model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7HLGambD6qmJ"
      },
      "source": [
        "def create_model():\n",
        "\n",
        "  model = keras.Sequential([\n",
        "                            keras.layers.Conv2D(filters=32, kernel_size=(3,3),\n",
        "                                                padding='same', activation='relu',\n",
        "                                                input_shape=(X.shape[1], X.shape[2], 1),\n",
        "                                                # activity_regularizer=keras.regularizers.l2(0.005)\n",
        "                                                ),\n",
        "\n",
        "                            keras.layers.MaxPool2D(pool_size=(2,2)),\n",
        "\n",
        "                            keras.layers.Dropout(0.25),\n",
        "\n",
        "                            keras.layers.Conv2D(filters=64, kernel_size=(3,3),\n",
        "                                                padding='same', activation='relu',\n",
        "                                                # activity_regularizer=keras.regularizers.l2(0.005),\n",
        "                                                ),\n",
        "\n",
        "                            keras.layers.MaxPool2D(pool_size=(2,2)),\n",
        "\n",
        "                            keras.layers.Dropout(0.25),\n",
        "\n",
        "                            # keras.layers.Conv2D(filters=128, kernel_size=(3,3),\n",
        "                            #                     padding='same', activation='relu',\n",
        "                            #                     # activity_regularizer=keras.regularizers.l2(0.005)\n",
        "                            #                     ),\n",
        "\n",
        "                            # keras.layers.MaxPool2D(pool_size=(2,2)),\n",
        "\n",
        "                            # keras.layers.Dropout(0.25),\n",
        "\n",
        "                            # keras.layers.GlobalMaxPool2D(),\n",
        "\n",
        "                            # keras.layers.Dropout(0.25),\n",
        "\n",
        "                            keras.layers.Flatten(),\n",
        "\n",
        "                            keras.layers.Dense(256, activation='relu', \n",
        "                                              #  activity_regularizer=keras.regularizers.l2(0.001)\n",
        "                                               ),\n",
        "\n",
        "                            keras.layers.Dropout(0.5),\n",
        "\n",
        "                            keras.layers.Dense(11, activation='softmax')\n",
        "  ])\n",
        "\n",
        "  # optimizer = keras.optimizers.Adam(learning_rate=0.0001)\n",
        "  model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])\n",
        "  \n",
        "  return model"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4_maNEh73aM5",
        "outputId": "cc5077bd-6827-4609-e643-856d20fd2932"
      },
      "source": [
        "y_test"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([2, 4, 0, 3, 4, 5, 1, 5, 2, 4, 5, 0, 4, 0, 0, 5, 2, 5, 2, 4, 5, 3,\n",
              "       4, 5, 1, 2, 2, 3, 3, 2, 2, 5, 2, 5, 3, 1, 2, 2, 4, 1, 3, 2, 0, 3,\n",
              "       5, 4, 2, 4, 2, 2, 2, 3, 5, 2, 1, 2, 2, 1, 1, 2, 2, 0, 2, 3, 1, 5,\n",
              "       2, 5, 4, 2, 0, 3, 0, 2, 0, 3, 0, 2, 5, 5, 5, 5, 3, 4, 3, 0, 4, 1,\n",
              "       0, 4, 3, 2, 4, 4, 3, 4, 5, 1, 2, 4, 3, 5, 5, 3, 1, 5, 1, 1, 1, 1,\n",
              "       0, 2, 1, 1, 4, 5, 2, 2, 5, 5, 5, 1, 0, 5, 4, 4, 5, 4, 0, 2, 4, 1,\n",
              "       5, 4, 2, 2, 3, 5, 1, 1, 1, 5, 3, 4, 5, 0, 3, 2, 0, 4, 5, 3, 5, 3,\n",
              "       3, 5, 5, 2, 5, 4, 1, 5, 1, 3, 5, 5, 3, 2, 2, 2, 1, 1, 5, 2, 5, 2,\n",
              "       4, 5, 0, 0, 4, 4, 1, 0, 4, 5, 5, 5, 3, 3, 1, 3, 4, 0, 4, 5, 3, 3,\n",
              "       2, 2, 2, 4, 0, 3, 5, 2, 3, 1, 5, 5, 5, 5, 1, 4, 3, 2, 5, 4, 4, 0,\n",
              "       5, 5, 2, 2, 1, 2, 1, 1, 3, 5, 0, 1, 3, 5, 1, 5, 0, 1, 3, 1, 0, 2,\n",
              "       5, 0, 0, 3, 2, 1, 3, 5, 5, 1, 1, 1, 3, 4, 5, 5, 3, 1, 2, 4, 5, 1,\n",
              "       4, 3, 1, 4, 3, 2, 5, 5, 1, 4, 3, 4, 3, 5, 3, 5, 1, 2, 5, 1, 2, 3,\n",
              "       0, 3, 1, 4, 1, 1, 3, 2, 1, 4, 3, 3, 0, 1, 4, 3, 3, 3, 2, 1, 3, 3,\n",
              "       3, 5, 5, 4, 3, 4, 3, 1, 2, 1, 5, 2, 4, 5, 2, 5, 3, 0, 5, 5, 0, 1,\n",
              "       2, 3, 0, 4, 4, 5, 2, 1, 5, 5, 1, 5, 4, 3, 5, 0, 1, 2, 0, 3, 5, 5,\n",
              "       4, 2, 3, 4, 4, 1, 1, 3, 0, 0, 4, 5, 3, 1, 0, 4, 2, 4, 3, 1, 0, 4,\n",
              "       5, 3, 5, 5, 1, 0, 2, 0, 3, 0, 1, 4, 2, 2, 1, 2, 5, 1, 4, 3, 1, 2,\n",
              "       2, 2, 1, 4, 4, 4, 1, 1, 0, 2, 3, 0, 3, 5, 5, 3, 2, 3, 0, 4, 5, 5,\n",
              "       5, 0, 2, 3, 0, 2, 5, 2, 3, 4, 3, 4, 3, 2, 0, 2, 2, 0, 5, 1, 3, 4,\n",
              "       2, 3, 2, 2, 5, 4, 2, 2, 5, 1, 4, 1, 5, 1, 5, 2, 1, 5, 0, 3, 2, 4,\n",
              "       4, 2, 0, 1, 5, 3, 1, 5, 3, 0, 5, 1, 1, 3, 1, 5, 4, 5, 4, 3, 4, 5,\n",
              "       3, 3, 2, 1, 0, 4, 4, 1, 2, 4, 4, 5, 5, 0, 5, 1, 5, 1, 2, 0, 1, 3,\n",
              "       3, 5, 5, 2, 3, 0, 4, 0, 1, 5, 1, 4, 2, 0, 0, 0, 1, 5, 1, 3, 4, 5,\n",
              "       4, 4, 0, 1, 4, 1, 1, 4, 1, 5, 3, 4, 0, 5, 4, 4, 1, 1, 4, 3, 2, 3,\n",
              "       5, 3, 2, 1, 3, 3, 5, 4, 4, 3, 5, 0, 4, 4, 0, 0, 1, 1, 5, 3, 1, 1,\n",
              "       1, 1, 2, 2, 1])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 13
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eNeliP6Z7f6D"
      },
      "source": [
        "with tpu_strategy.scope():\n",
        "  model = create_model()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "MfrlDs_k7RBu",
        "outputId": "d926c34f-bce3-4b5a-cfc3-ab915f9c55a4"
      },
      "source": [
        "model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d (Conv2D)              (None, 128, 16, 32)       320       \n",
            "_________________________________________________________________\n",
            "max_pooling2d (MaxPooling2D) (None, 64, 8, 32)         0         \n",
            "_________________________________________________________________\n",
            "dropout (Dropout)            (None, 64, 8, 32)         0         \n",
            "_________________________________________________________________\n",
            "conv2d_1 (Conv2D)            (None, 64, 8, 64)         18496     \n",
            "_________________________________________________________________\n",
            "max_pooling2d_1 (MaxPooling2 (None, 32, 4, 64)         0         \n",
            "_________________________________________________________________\n",
            "dropout_1 (Dropout)          (None, 32, 4, 64)         0         \n",
            "_________________________________________________________________\n",
            "flatten (Flatten)            (None, 8192)              0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 256)               2097408   \n",
            "_________________________________________________________________\n",
            "dropout_2 (Dropout)          (None, 256)               0         \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 11)                2827      \n",
            "=================================================================\n",
            "Total params: 2,119,051\n",
            "Trainable params: 2,119,051\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "pWQoBaBx7pHZ",
        "outputId": "2cbf29b8-daa0-493c-9366-7308890a6df4"
      },
      "source": [
        "history = model.fit(X_train, y_train, validation_data=(X_test, y_test), batch_size=32, epochs=500)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/500\n",
            "103/103 [==============================] - 9s 58ms/step - loss: 1.9840 - accuracy: 0.2058 - val_loss: 1.7596 - val_accuracy: 0.3102\n",
            "Epoch 2/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 1.6539 - accuracy: 0.3229 - val_loss: 1.6735 - val_accuracy: 0.3761\n",
            "Epoch 3/500\n",
            "103/103 [==============================] - 2s 20ms/step - loss: 1.5339 - accuracy: 0.3977 - val_loss: 1.5913 - val_accuracy: 0.4107\n",
            "Epoch 4/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 1.4386 - accuracy: 0.4386 - val_loss: 1.5361 - val_accuracy: 0.4402\n",
            "Epoch 5/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 1.3831 - accuracy: 0.4629 - val_loss: 1.4394 - val_accuracy: 0.4783\n",
            "Epoch 6/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 1.3660 - accuracy: 0.4861 - val_loss: 1.4192 - val_accuracy: 0.5026\n",
            "Epoch 7/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 1.3059 - accuracy: 0.5096 - val_loss: 1.3924 - val_accuracy: 0.5754\n",
            "Epoch 8/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 1.2655 - accuracy: 0.5254 - val_loss: 1.3362 - val_accuracy: 0.5667\n",
            "Epoch 9/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 1.2279 - accuracy: 0.5516 - val_loss: 1.3085 - val_accuracy: 0.5893\n",
            "Epoch 10/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 1.2144 - accuracy: 0.5408 - val_loss: 1.2464 - val_accuracy: 0.6118\n",
            "Epoch 11/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 1.1915 - accuracy: 0.5436 - val_loss: 1.2852 - val_accuracy: 0.6031\n",
            "Epoch 12/500\n",
            "103/103 [==============================] - 2s 20ms/step - loss: 1.1714 - accuracy: 0.5616 - val_loss: 1.2235 - val_accuracy: 0.6014\n",
            "Epoch 13/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 1.1663 - accuracy: 0.5632 - val_loss: 1.1899 - val_accuracy: 0.6222\n",
            "Epoch 14/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 1.1410 - accuracy: 0.5696 - val_loss: 1.1600 - val_accuracy: 0.6014\n",
            "Epoch 15/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 1.1035 - accuracy: 0.5893 - val_loss: 1.1607 - val_accuracy: 0.6308\n",
            "Epoch 16/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 1.1291 - accuracy: 0.5711 - val_loss: 1.1350 - val_accuracy: 0.6222\n",
            "Epoch 17/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 1.0875 - accuracy: 0.5930 - val_loss: 1.1242 - val_accuracy: 0.6326\n",
            "Epoch 18/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 1.0560 - accuracy: 0.6000 - val_loss: 1.1607 - val_accuracy: 0.6222\n",
            "Epoch 19/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 1.0620 - accuracy: 0.6024 - val_loss: 1.0883 - val_accuracy: 0.6378\n",
            "Epoch 20/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 1.0389 - accuracy: 0.6162 - val_loss: 1.1045 - val_accuracy: 0.6430\n",
            "Epoch 21/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 1.0197 - accuracy: 0.6362 - val_loss: 1.0935 - val_accuracy: 0.6412\n",
            "Epoch 22/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 1.0286 - accuracy: 0.6171 - val_loss: 1.0735 - val_accuracy: 0.6412\n",
            "Epoch 23/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 1.0273 - accuracy: 0.6277 - val_loss: 1.0385 - val_accuracy: 0.6638\n",
            "Epoch 24/500\n",
            "103/103 [==============================] - 2s 23ms/step - loss: 1.0087 - accuracy: 0.6277 - val_loss: 1.0449 - val_accuracy: 0.6880\n",
            "Epoch 25/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.9925 - accuracy: 0.6492 - val_loss: 1.0668 - val_accuracy: 0.6482\n",
            "Epoch 26/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.9895 - accuracy: 0.6407 - val_loss: 1.0104 - val_accuracy: 0.6551\n",
            "Epoch 27/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.9509 - accuracy: 0.6426 - val_loss: 1.0538 - val_accuracy: 0.6776\n",
            "Epoch 28/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 1.0038 - accuracy: 0.6350 - val_loss: 0.9851 - val_accuracy: 0.6690\n",
            "Epoch 29/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.9826 - accuracy: 0.6468 - val_loss: 1.0097 - val_accuracy: 0.6828\n",
            "Epoch 30/500\n",
            "103/103 [==============================] - 2s 20ms/step - loss: 0.9430 - accuracy: 0.6588 - val_loss: 0.9728 - val_accuracy: 0.6655\n",
            "Epoch 31/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.9632 - accuracy: 0.6543 - val_loss: 1.0266 - val_accuracy: 0.6412\n",
            "Epoch 32/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.9836 - accuracy: 0.6349 - val_loss: 0.9681 - val_accuracy: 0.6776\n",
            "Epoch 33/500\n",
            "103/103 [==============================] - 2s 20ms/step - loss: 0.9204 - accuracy: 0.6567 - val_loss: 0.9907 - val_accuracy: 0.6620\n",
            "Epoch 34/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.9392 - accuracy: 0.6559 - val_loss: 0.9464 - val_accuracy: 0.6846\n",
            "Epoch 35/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.9257 - accuracy: 0.6680 - val_loss: 0.9570 - val_accuracy: 0.6863\n",
            "Epoch 36/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.9309 - accuracy: 0.6652 - val_loss: 0.9718 - val_accuracy: 0.6742\n",
            "Epoch 37/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.9233 - accuracy: 0.6646 - val_loss: 1.0154 - val_accuracy: 0.6534\n",
            "Epoch 38/500\n",
            "103/103 [==============================] - 2s 20ms/step - loss: 0.9269 - accuracy: 0.6564 - val_loss: 0.9186 - val_accuracy: 0.7123\n",
            "Epoch 39/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.8990 - accuracy: 0.6592 - val_loss: 0.9128 - val_accuracy: 0.7158\n",
            "Epoch 40/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.8903 - accuracy: 0.6665 - val_loss: 0.9024 - val_accuracy: 0.6898\n",
            "Epoch 41/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.9076 - accuracy: 0.6700 - val_loss: 0.9100 - val_accuracy: 0.6950\n",
            "Epoch 42/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.9247 - accuracy: 0.6528 - val_loss: 0.9326 - val_accuracy: 0.6932\n",
            "Epoch 43/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.8810 - accuracy: 0.6805 - val_loss: 0.8990 - val_accuracy: 0.7036\n",
            "Epoch 44/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.8521 - accuracy: 0.6840 - val_loss: 0.9292 - val_accuracy: 0.6932\n",
            "Epoch 45/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.8484 - accuracy: 0.6962 - val_loss: 0.8956 - val_accuracy: 0.7210\n",
            "Epoch 46/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.8556 - accuracy: 0.6810 - val_loss: 0.9089 - val_accuracy: 0.7106\n",
            "Epoch 47/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.8314 - accuracy: 0.6946 - val_loss: 0.8940 - val_accuracy: 0.7106\n",
            "Epoch 48/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.8503 - accuracy: 0.7047 - val_loss: 0.8776 - val_accuracy: 0.7314\n",
            "Epoch 49/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.8594 - accuracy: 0.6751 - val_loss: 0.9110 - val_accuracy: 0.7088\n",
            "Epoch 50/500\n",
            "103/103 [==============================] - 2s 24ms/step - loss: 0.8428 - accuracy: 0.6927 - val_loss: 0.8663 - val_accuracy: 0.7036\n",
            "Epoch 51/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.8354 - accuracy: 0.6917 - val_loss: 0.8845 - val_accuracy: 0.7140\n",
            "Epoch 52/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.8281 - accuracy: 0.7011 - val_loss: 0.8429 - val_accuracy: 0.7106\n",
            "Epoch 53/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.8533 - accuracy: 0.6870 - val_loss: 0.8518 - val_accuracy: 0.7140\n",
            "Epoch 54/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.8554 - accuracy: 0.6875 - val_loss: 0.8605 - val_accuracy: 0.7244\n",
            "Epoch 55/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.8486 - accuracy: 0.6975 - val_loss: 0.8439 - val_accuracy: 0.7262\n",
            "Epoch 56/500\n",
            "103/103 [==============================] - 2s 20ms/step - loss: 0.8283 - accuracy: 0.7027 - val_loss: 0.9096 - val_accuracy: 0.7088\n",
            "Epoch 57/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.8025 - accuracy: 0.7110 - val_loss: 0.8718 - val_accuracy: 0.6950\n",
            "Epoch 58/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.8014 - accuracy: 0.6944 - val_loss: 0.8536 - val_accuracy: 0.7158\n",
            "Epoch 59/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.8019 - accuracy: 0.7085 - val_loss: 0.8245 - val_accuracy: 0.7418\n",
            "Epoch 60/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7914 - accuracy: 0.7063 - val_loss: 0.8668 - val_accuracy: 0.7210\n",
            "Epoch 61/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.8123 - accuracy: 0.6987 - val_loss: 0.8489 - val_accuracy: 0.7262\n",
            "Epoch 62/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.8051 - accuracy: 0.7045 - val_loss: 0.8574 - val_accuracy: 0.7279\n",
            "Epoch 63/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.7651 - accuracy: 0.7212 - val_loss: 0.8106 - val_accuracy: 0.7244\n",
            "Epoch 64/500\n",
            "103/103 [==============================] - 2s 23ms/step - loss: 0.7993 - accuracy: 0.7000 - val_loss: 0.8537 - val_accuracy: 0.7418\n",
            "Epoch 65/500\n",
            "103/103 [==============================] - 2s 23ms/step - loss: 0.7814 - accuracy: 0.7152 - val_loss: 0.8190 - val_accuracy: 0.7366\n",
            "Epoch 66/500\n",
            "103/103 [==============================] - 2s 23ms/step - loss: 0.8047 - accuracy: 0.7127 - val_loss: 0.8240 - val_accuracy: 0.7383\n",
            "Epoch 67/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.7880 - accuracy: 0.7165 - val_loss: 0.8226 - val_accuracy: 0.7314\n",
            "Epoch 68/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7567 - accuracy: 0.7332 - val_loss: 0.8364 - val_accuracy: 0.7348\n",
            "Epoch 69/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7598 - accuracy: 0.7247 - val_loss: 0.7876 - val_accuracy: 0.7556\n",
            "Epoch 70/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7563 - accuracy: 0.7378 - val_loss: 0.8263 - val_accuracy: 0.7158\n",
            "Epoch 71/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.8059 - accuracy: 0.7003 - val_loss: 0.8376 - val_accuracy: 0.7210\n",
            "Epoch 72/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.8079 - accuracy: 0.6999 - val_loss: 0.8397 - val_accuracy: 0.7210\n",
            "Epoch 73/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7678 - accuracy: 0.7158 - val_loss: 0.8196 - val_accuracy: 0.7296\n",
            "Epoch 74/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7604 - accuracy: 0.7153 - val_loss: 0.8109 - val_accuracy: 0.7452\n",
            "Epoch 75/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7505 - accuracy: 0.7301 - val_loss: 0.8201 - val_accuracy: 0.7331\n",
            "Epoch 76/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7948 - accuracy: 0.7203 - val_loss: 0.7914 - val_accuracy: 0.7452\n",
            "Epoch 77/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7812 - accuracy: 0.7236 - val_loss: 0.8229 - val_accuracy: 0.7574\n",
            "Epoch 78/500\n",
            "103/103 [==============================] - 2s 24ms/step - loss: 0.7536 - accuracy: 0.7295 - val_loss: 0.7990 - val_accuracy: 0.7400\n",
            "Epoch 79/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7499 - accuracy: 0.7346 - val_loss: 0.8274 - val_accuracy: 0.7348\n",
            "Epoch 80/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7544 - accuracy: 0.7238 - val_loss: 0.7866 - val_accuracy: 0.7678\n",
            "Epoch 81/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7280 - accuracy: 0.7431 - val_loss: 0.7987 - val_accuracy: 0.7435\n",
            "Epoch 82/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7808 - accuracy: 0.7168 - val_loss: 0.8102 - val_accuracy: 0.7348\n",
            "Epoch 83/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7496 - accuracy: 0.7307 - val_loss: 0.7765 - val_accuracy: 0.7556\n",
            "Epoch 84/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7252 - accuracy: 0.7271 - val_loss: 0.7853 - val_accuracy: 0.7539\n",
            "Epoch 85/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7375 - accuracy: 0.7241 - val_loss: 0.7825 - val_accuracy: 0.7383\n",
            "Epoch 86/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.7038 - accuracy: 0.7486 - val_loss: 0.7532 - val_accuracy: 0.7747\n",
            "Epoch 87/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6902 - accuracy: 0.7454 - val_loss: 0.7846 - val_accuracy: 0.7556\n",
            "Epoch 88/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.7112 - accuracy: 0.7451 - val_loss: 0.7733 - val_accuracy: 0.7539\n",
            "Epoch 89/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7141 - accuracy: 0.7491 - val_loss: 0.7611 - val_accuracy: 0.7747\n",
            "Epoch 90/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7103 - accuracy: 0.7409 - val_loss: 0.7690 - val_accuracy: 0.7608\n",
            "Epoch 91/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7332 - accuracy: 0.7294 - val_loss: 0.7680 - val_accuracy: 0.7556\n",
            "Epoch 92/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7527 - accuracy: 0.7338 - val_loss: 0.8054 - val_accuracy: 0.7227\n",
            "Epoch 93/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6878 - accuracy: 0.7566 - val_loss: 0.7539 - val_accuracy: 0.7574\n",
            "Epoch 94/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7218 - accuracy: 0.7415 - val_loss: 0.7520 - val_accuracy: 0.7470\n",
            "Epoch 95/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7237 - accuracy: 0.7307 - val_loss: 0.7779 - val_accuracy: 0.7452\n",
            "Epoch 96/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7022 - accuracy: 0.7432 - val_loss: 0.7446 - val_accuracy: 0.7504\n",
            "Epoch 97/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7179 - accuracy: 0.7424 - val_loss: 0.7750 - val_accuracy: 0.7574\n",
            "Epoch 98/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6776 - accuracy: 0.7608 - val_loss: 0.7582 - val_accuracy: 0.7366\n",
            "Epoch 99/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.6992 - accuracy: 0.7464 - val_loss: 0.7436 - val_accuracy: 0.7660\n",
            "Epoch 100/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6724 - accuracy: 0.7495 - val_loss: 0.7462 - val_accuracy: 0.7626\n",
            "Epoch 101/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7029 - accuracy: 0.7412 - val_loss: 0.7628 - val_accuracy: 0.7470\n",
            "Epoch 102/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6613 - accuracy: 0.7697 - val_loss: 0.7366 - val_accuracy: 0.7591\n",
            "Epoch 103/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6797 - accuracy: 0.7542 - val_loss: 0.7523 - val_accuracy: 0.7608\n",
            "Epoch 104/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7305 - accuracy: 0.7314 - val_loss: 0.7374 - val_accuracy: 0.7487\n",
            "Epoch 105/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6899 - accuracy: 0.7369 - val_loss: 0.7826 - val_accuracy: 0.7556\n",
            "Epoch 106/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7164 - accuracy: 0.7363 - val_loss: 0.7377 - val_accuracy: 0.7504\n",
            "Epoch 107/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7060 - accuracy: 0.7442 - val_loss: 0.7271 - val_accuracy: 0.7626\n",
            "Epoch 108/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6860 - accuracy: 0.7429 - val_loss: 0.7835 - val_accuracy: 0.7314\n",
            "Epoch 109/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6566 - accuracy: 0.7658 - val_loss: 0.7378 - val_accuracy: 0.7591\n",
            "Epoch 110/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.6451 - accuracy: 0.7649 - val_loss: 0.7065 - val_accuracy: 0.7591\n",
            "Epoch 111/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.6789 - accuracy: 0.7561 - val_loss: 0.7416 - val_accuracy: 0.7643\n",
            "Epoch 112/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.6853 - accuracy: 0.7557 - val_loss: 0.7236 - val_accuracy: 0.7556\n",
            "Epoch 113/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6788 - accuracy: 0.7489 - val_loss: 0.7253 - val_accuracy: 0.7643\n",
            "Epoch 114/500\n",
            "103/103 [==============================] - 2s 24ms/step - loss: 0.6640 - accuracy: 0.7690 - val_loss: 0.7367 - val_accuracy: 0.7747\n",
            "Epoch 115/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6604 - accuracy: 0.7615 - val_loss: 0.7651 - val_accuracy: 0.7522\n",
            "Epoch 116/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6603 - accuracy: 0.7653 - val_loss: 0.7824 - val_accuracy: 0.7279\n",
            "Epoch 117/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6931 - accuracy: 0.7502 - val_loss: 0.7337 - val_accuracy: 0.7504\n",
            "Epoch 118/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6986 - accuracy: 0.7514 - val_loss: 0.7348 - val_accuracy: 0.7574\n",
            "Epoch 119/500\n",
            "103/103 [==============================] - 2s 20ms/step - loss: 0.6869 - accuracy: 0.7557 - val_loss: 0.7052 - val_accuracy: 0.7834\n",
            "Epoch 120/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6893 - accuracy: 0.7432 - val_loss: 0.7372 - val_accuracy: 0.7608\n",
            "Epoch 121/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6832 - accuracy: 0.7481 - val_loss: 0.7591 - val_accuracy: 0.7695\n",
            "Epoch 122/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6453 - accuracy: 0.7598 - val_loss: 0.7520 - val_accuracy: 0.7626\n",
            "Epoch 123/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6345 - accuracy: 0.7720 - val_loss: 0.7340 - val_accuracy: 0.7591\n",
            "Epoch 124/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7034 - accuracy: 0.7424 - val_loss: 0.7894 - val_accuracy: 0.7452\n",
            "Epoch 125/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6587 - accuracy: 0.7560 - val_loss: 0.7060 - val_accuracy: 0.7938\n",
            "Epoch 126/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6885 - accuracy: 0.7557 - val_loss: 0.7117 - val_accuracy: 0.7695\n",
            "Epoch 127/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6446 - accuracy: 0.7649 - val_loss: 0.7012 - val_accuracy: 0.7712\n",
            "Epoch 128/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6570 - accuracy: 0.7706 - val_loss: 0.7377 - val_accuracy: 0.7522\n",
            "Epoch 129/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6230 - accuracy: 0.7681 - val_loss: 0.7124 - val_accuracy: 0.7782\n",
            "Epoch 130/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6709 - accuracy: 0.7508 - val_loss: 0.7031 - val_accuracy: 0.7886\n",
            "Epoch 131/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6414 - accuracy: 0.7807 - val_loss: 0.6881 - val_accuracy: 0.7851\n",
            "Epoch 132/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6803 - accuracy: 0.7635 - val_loss: 0.7350 - val_accuracy: 0.7608\n",
            "Epoch 133/500\n",
            "103/103 [==============================] - 2s 24ms/step - loss: 0.6811 - accuracy: 0.7544 - val_loss: 0.7848 - val_accuracy: 0.7314\n",
            "Epoch 134/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.7022 - accuracy: 0.7572 - val_loss: 0.7048 - val_accuracy: 0.7660\n",
            "Epoch 135/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6718 - accuracy: 0.7600 - val_loss: 0.7283 - val_accuracy: 0.7574\n",
            "Epoch 136/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6837 - accuracy: 0.7418 - val_loss: 0.7307 - val_accuracy: 0.7591\n",
            "Epoch 137/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6330 - accuracy: 0.7672 - val_loss: 0.7440 - val_accuracy: 0.7418\n",
            "Epoch 138/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.6558 - accuracy: 0.7636 - val_loss: 0.7563 - val_accuracy: 0.7608\n",
            "Epoch 139/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6568 - accuracy: 0.7438 - val_loss: 0.7025 - val_accuracy: 0.7816\n",
            "Epoch 140/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6421 - accuracy: 0.7612 - val_loss: 0.6926 - val_accuracy: 0.7834\n",
            "Epoch 141/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6403 - accuracy: 0.7612 - val_loss: 0.7015 - val_accuracy: 0.7764\n",
            "Epoch 142/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6467 - accuracy: 0.7523 - val_loss: 0.6876 - val_accuracy: 0.7868\n",
            "Epoch 143/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6543 - accuracy: 0.7586 - val_loss: 0.7026 - val_accuracy: 0.7799\n",
            "Epoch 144/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6181 - accuracy: 0.7817 - val_loss: 0.7025 - val_accuracy: 0.7730\n",
            "Epoch 145/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6465 - accuracy: 0.7678 - val_loss: 0.6970 - val_accuracy: 0.7851\n",
            "Epoch 146/500\n",
            "103/103 [==============================] - 2s 23ms/step - loss: 0.6773 - accuracy: 0.7580 - val_loss: 0.7336 - val_accuracy: 0.7574\n",
            "Epoch 147/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6146 - accuracy: 0.7774 - val_loss: 0.7074 - val_accuracy: 0.7695\n",
            "Epoch 148/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6420 - accuracy: 0.7675 - val_loss: 0.7212 - val_accuracy: 0.7712\n",
            "Epoch 149/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6171 - accuracy: 0.7772 - val_loss: 0.7318 - val_accuracy: 0.7660\n",
            "Epoch 150/500\n",
            "103/103 [==============================] - 2s 23ms/step - loss: 0.6353 - accuracy: 0.7656 - val_loss: 0.7105 - val_accuracy: 0.7591\n",
            "Epoch 151/500\n",
            "103/103 [==============================] - 2s 20ms/step - loss: 0.6201 - accuracy: 0.7914 - val_loss: 0.6717 - val_accuracy: 0.7886\n",
            "Epoch 152/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6473 - accuracy: 0.7554 - val_loss: 0.7294 - val_accuracy: 0.7522\n",
            "Epoch 153/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6456 - accuracy: 0.7677 - val_loss: 0.6994 - val_accuracy: 0.7643\n",
            "Epoch 154/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6136 - accuracy: 0.7867 - val_loss: 0.7099 - val_accuracy: 0.7695\n",
            "Epoch 155/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6331 - accuracy: 0.7656 - val_loss: 0.6947 - val_accuracy: 0.7712\n",
            "Epoch 156/500\n",
            "103/103 [==============================] - 2s 20ms/step - loss: 0.6101 - accuracy: 0.7826 - val_loss: 0.7080 - val_accuracy: 0.7608\n",
            "Epoch 157/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6331 - accuracy: 0.7682 - val_loss: 0.7188 - val_accuracy: 0.7764\n",
            "Epoch 158/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6149 - accuracy: 0.7877 - val_loss: 0.6864 - val_accuracy: 0.7886\n",
            "Epoch 159/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6113 - accuracy: 0.7752 - val_loss: 0.7226 - val_accuracy: 0.7591\n",
            "Epoch 160/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6048 - accuracy: 0.7788 - val_loss: 0.7676 - val_accuracy: 0.7210\n",
            "Epoch 161/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6283 - accuracy: 0.7763 - val_loss: 0.7266 - val_accuracy: 0.7556\n",
            "Epoch 162/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6231 - accuracy: 0.7689 - val_loss: 0.7016 - val_accuracy: 0.7730\n",
            "Epoch 163/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6396 - accuracy: 0.7639 - val_loss: 0.6987 - val_accuracy: 0.7851\n",
            "Epoch 164/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6188 - accuracy: 0.7782 - val_loss: 0.6940 - val_accuracy: 0.7695\n",
            "Epoch 165/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6196 - accuracy: 0.7797 - val_loss: 0.7494 - val_accuracy: 0.7314\n",
            "Epoch 166/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6118 - accuracy: 0.7793 - val_loss: 0.7010 - val_accuracy: 0.7591\n",
            "Epoch 167/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6052 - accuracy: 0.7801 - val_loss: 0.6628 - val_accuracy: 0.7990\n",
            "Epoch 168/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.6096 - accuracy: 0.7812 - val_loss: 0.6704 - val_accuracy: 0.7799\n",
            "Epoch 169/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6050 - accuracy: 0.7937 - val_loss: 0.6826 - val_accuracy: 0.7764\n",
            "Epoch 170/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5925 - accuracy: 0.7821 - val_loss: 0.7007 - val_accuracy: 0.7695\n",
            "Epoch 171/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6334 - accuracy: 0.7792 - val_loss: 0.7361 - val_accuracy: 0.7522\n",
            "Epoch 172/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6458 - accuracy: 0.7613 - val_loss: 0.7260 - val_accuracy: 0.7730\n",
            "Epoch 173/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6382 - accuracy: 0.7717 - val_loss: 0.7437 - val_accuracy: 0.7730\n",
            "Epoch 174/500\n",
            "103/103 [==============================] - 2s 20ms/step - loss: 0.6382 - accuracy: 0.7708 - val_loss: 0.6728 - val_accuracy: 0.7851\n",
            "Epoch 175/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6151 - accuracy: 0.7735 - val_loss: 0.7354 - val_accuracy: 0.7712\n",
            "Epoch 176/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5816 - accuracy: 0.7799 - val_loss: 0.6942 - val_accuracy: 0.7834\n",
            "Epoch 177/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6160 - accuracy: 0.7811 - val_loss: 0.6962 - val_accuracy: 0.7712\n",
            "Epoch 178/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6065 - accuracy: 0.7819 - val_loss: 0.6941 - val_accuracy: 0.7712\n",
            "Epoch 179/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6489 - accuracy: 0.7597 - val_loss: 0.6714 - val_accuracy: 0.7799\n",
            "Epoch 180/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.6397 - accuracy: 0.7690 - val_loss: 0.7006 - val_accuracy: 0.7764\n",
            "Epoch 181/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5958 - accuracy: 0.7866 - val_loss: 0.6740 - val_accuracy: 0.7851\n",
            "Epoch 182/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5813 - accuracy: 0.7885 - val_loss: 0.6840 - val_accuracy: 0.7868\n",
            "Epoch 183/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6036 - accuracy: 0.7893 - val_loss: 0.6822 - val_accuracy: 0.8076\n",
            "Epoch 184/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6240 - accuracy: 0.7618 - val_loss: 0.6787 - val_accuracy: 0.7886\n",
            "Epoch 185/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5731 - accuracy: 0.7858 - val_loss: 0.6516 - val_accuracy: 0.7816\n",
            "Epoch 186/500\n",
            "103/103 [==============================] - 2s 24ms/step - loss: 0.6182 - accuracy: 0.7726 - val_loss: 0.7347 - val_accuracy: 0.7730\n",
            "Epoch 187/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6077 - accuracy: 0.7842 - val_loss: 0.7143 - val_accuracy: 0.7782\n",
            "Epoch 188/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6313 - accuracy: 0.7761 - val_loss: 0.6857 - val_accuracy: 0.7834\n",
            "Epoch 189/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6044 - accuracy: 0.7814 - val_loss: 0.6742 - val_accuracy: 0.7799\n",
            "Epoch 190/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6008 - accuracy: 0.7646 - val_loss: 0.7036 - val_accuracy: 0.7678\n",
            "Epoch 191/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5896 - accuracy: 0.7913 - val_loss: 0.6913 - val_accuracy: 0.7868\n",
            "Epoch 192/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6345 - accuracy: 0.7714 - val_loss: 0.6933 - val_accuracy: 0.7764\n",
            "Epoch 193/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5830 - accuracy: 0.7910 - val_loss: 0.6639 - val_accuracy: 0.7886\n",
            "Epoch 194/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5913 - accuracy: 0.7821 - val_loss: 0.6783 - val_accuracy: 0.7868\n",
            "Epoch 195/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5836 - accuracy: 0.7787 - val_loss: 0.6536 - val_accuracy: 0.7816\n",
            "Epoch 196/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5981 - accuracy: 0.7819 - val_loss: 0.6713 - val_accuracy: 0.7799\n",
            "Epoch 197/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5820 - accuracy: 0.7879 - val_loss: 0.7029 - val_accuracy: 0.7730\n",
            "Epoch 198/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5896 - accuracy: 0.7863 - val_loss: 0.7174 - val_accuracy: 0.7782\n",
            "Epoch 199/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6272 - accuracy: 0.7730 - val_loss: 0.7140 - val_accuracy: 0.7816\n",
            "Epoch 200/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6211 - accuracy: 0.7846 - val_loss: 0.6750 - val_accuracy: 0.7695\n",
            "Epoch 201/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5602 - accuracy: 0.7939 - val_loss: 0.6699 - val_accuracy: 0.7816\n",
            "Epoch 202/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5673 - accuracy: 0.7970 - val_loss: 0.7263 - val_accuracy: 0.7470\n",
            "Epoch 203/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5707 - accuracy: 0.7930 - val_loss: 0.7071 - val_accuracy: 0.7695\n",
            "Epoch 204/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5940 - accuracy: 0.7901 - val_loss: 0.6719 - val_accuracy: 0.7955\n",
            "Epoch 205/500\n",
            "103/103 [==============================] - 2s 23ms/step - loss: 0.6137 - accuracy: 0.7787 - val_loss: 0.6724 - val_accuracy: 0.7834\n",
            "Epoch 206/500\n",
            "103/103 [==============================] - 2s 23ms/step - loss: 0.5640 - accuracy: 0.8002 - val_loss: 0.6965 - val_accuracy: 0.7799\n",
            "Epoch 207/500\n",
            "103/103 [==============================] - 3s 25ms/step - loss: 0.6014 - accuracy: 0.7759 - val_loss: 0.6671 - val_accuracy: 0.7938\n",
            "Epoch 208/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5852 - accuracy: 0.7972 - val_loss: 0.6931 - val_accuracy: 0.7834\n",
            "Epoch 209/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5808 - accuracy: 0.7966 - val_loss: 0.6978 - val_accuracy: 0.7816\n",
            "Epoch 210/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5744 - accuracy: 0.7911 - val_loss: 0.6578 - val_accuracy: 0.7938\n",
            "Epoch 211/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6080 - accuracy: 0.7918 - val_loss: 0.6982 - val_accuracy: 0.7660\n",
            "Epoch 212/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5672 - accuracy: 0.7855 - val_loss: 0.6702 - val_accuracy: 0.7834\n",
            "Epoch 213/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5856 - accuracy: 0.7972 - val_loss: 0.7131 - val_accuracy: 0.7591\n",
            "Epoch 214/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5870 - accuracy: 0.7957 - val_loss: 0.6849 - val_accuracy: 0.7730\n",
            "Epoch 215/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5796 - accuracy: 0.7897 - val_loss: 0.7249 - val_accuracy: 0.7556\n",
            "Epoch 216/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5955 - accuracy: 0.7871 - val_loss: 0.6785 - val_accuracy: 0.7868\n",
            "Epoch 217/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5593 - accuracy: 0.7938 - val_loss: 0.6993 - val_accuracy: 0.7730\n",
            "Epoch 218/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5873 - accuracy: 0.7862 - val_loss: 0.7135 - val_accuracy: 0.7747\n",
            "Epoch 219/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6154 - accuracy: 0.7787 - val_loss: 0.6868 - val_accuracy: 0.7695\n",
            "Epoch 220/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6157 - accuracy: 0.7777 - val_loss: 0.7017 - val_accuracy: 0.7747\n",
            "Epoch 221/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5835 - accuracy: 0.7818 - val_loss: 0.7041 - val_accuracy: 0.7695\n",
            "Epoch 222/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5610 - accuracy: 0.8054 - val_loss: 0.6989 - val_accuracy: 0.7868\n",
            "Epoch 223/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5730 - accuracy: 0.7971 - val_loss: 0.6669 - val_accuracy: 0.7920\n",
            "Epoch 224/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5174 - accuracy: 0.8140 - val_loss: 0.7509 - val_accuracy: 0.7452\n",
            "Epoch 225/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5767 - accuracy: 0.7860 - val_loss: 0.6503 - val_accuracy: 0.7799\n",
            "Epoch 226/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5859 - accuracy: 0.7916 - val_loss: 0.6901 - val_accuracy: 0.7712\n",
            "Epoch 227/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6259 - accuracy: 0.7760 - val_loss: 0.6793 - val_accuracy: 0.7782\n",
            "Epoch 228/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6017 - accuracy: 0.7844 - val_loss: 0.6726 - val_accuracy: 0.7747\n",
            "Epoch 229/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5846 - accuracy: 0.7926 - val_loss: 0.7270 - val_accuracy: 0.7626\n",
            "Epoch 230/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5766 - accuracy: 0.7958 - val_loss: 0.6632 - val_accuracy: 0.7868\n",
            "Epoch 231/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5560 - accuracy: 0.8023 - val_loss: 0.6614 - val_accuracy: 0.7764\n",
            "Epoch 232/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5806 - accuracy: 0.7956 - val_loss: 0.7087 - val_accuracy: 0.7556\n",
            "Epoch 233/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5785 - accuracy: 0.7873 - val_loss: 0.6652 - val_accuracy: 0.7834\n",
            "Epoch 234/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5614 - accuracy: 0.8036 - val_loss: 0.6854 - val_accuracy: 0.7764\n",
            "Epoch 235/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5418 - accuracy: 0.8103 - val_loss: 0.6805 - val_accuracy: 0.7643\n",
            "Epoch 236/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5589 - accuracy: 0.8101 - val_loss: 0.6891 - val_accuracy: 0.7730\n",
            "Epoch 237/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5783 - accuracy: 0.7935 - val_loss: 0.6762 - val_accuracy: 0.7695\n",
            "Epoch 238/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5963 - accuracy: 0.7800 - val_loss: 0.6837 - val_accuracy: 0.7660\n",
            "Epoch 239/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5623 - accuracy: 0.8028 - val_loss: 0.6350 - val_accuracy: 0.7903\n",
            "Epoch 240/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5564 - accuracy: 0.8004 - val_loss: 0.6744 - val_accuracy: 0.7764\n",
            "Epoch 241/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5550 - accuracy: 0.8067 - val_loss: 0.6807 - val_accuracy: 0.7626\n",
            "Epoch 242/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5737 - accuracy: 0.7902 - val_loss: 0.6817 - val_accuracy: 0.7626\n",
            "Epoch 243/500\n",
            "103/103 [==============================] - 2s 24ms/step - loss: 0.5477 - accuracy: 0.8081 - val_loss: 0.7125 - val_accuracy: 0.7747\n",
            "Epoch 244/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5690 - accuracy: 0.8049 - val_loss: 0.6931 - val_accuracy: 0.7730\n",
            "Epoch 245/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5453 - accuracy: 0.8066 - val_loss: 0.7111 - val_accuracy: 0.7712\n",
            "Epoch 246/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5285 - accuracy: 0.8165 - val_loss: 0.6685 - val_accuracy: 0.7782\n",
            "Epoch 247/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5426 - accuracy: 0.8029 - val_loss: 0.6894 - val_accuracy: 0.7764\n",
            "Epoch 248/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5648 - accuracy: 0.7986 - val_loss: 0.6282 - val_accuracy: 0.8094\n",
            "Epoch 249/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5880 - accuracy: 0.7838 - val_loss: 0.6511 - val_accuracy: 0.7851\n",
            "Epoch 250/500\n",
            "103/103 [==============================] - 2s 23ms/step - loss: 0.5751 - accuracy: 0.7963 - val_loss: 0.6532 - val_accuracy: 0.7990\n",
            "Epoch 251/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5800 - accuracy: 0.8013 - val_loss: 0.7153 - val_accuracy: 0.7730\n",
            "Epoch 252/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5717 - accuracy: 0.7927 - val_loss: 0.6962 - val_accuracy: 0.7782\n",
            "Epoch 253/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5676 - accuracy: 0.7989 - val_loss: 0.6877 - val_accuracy: 0.7660\n",
            "Epoch 254/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5524 - accuracy: 0.7906 - val_loss: 0.6826 - val_accuracy: 0.7764\n",
            "Epoch 255/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5420 - accuracy: 0.7993 - val_loss: 0.6668 - val_accuracy: 0.7730\n",
            "Epoch 256/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5681 - accuracy: 0.7924 - val_loss: 0.6769 - val_accuracy: 0.7764\n",
            "Epoch 257/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5334 - accuracy: 0.8106 - val_loss: 0.6652 - val_accuracy: 0.7764\n",
            "Epoch 258/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5847 - accuracy: 0.7933 - val_loss: 0.6445 - val_accuracy: 0.7903\n",
            "Epoch 259/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5326 - accuracy: 0.8149 - val_loss: 0.6542 - val_accuracy: 0.7886\n",
            "Epoch 260/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5699 - accuracy: 0.7971 - val_loss: 0.6491 - val_accuracy: 0.7938\n",
            "Epoch 261/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5626 - accuracy: 0.7946 - val_loss: 0.6473 - val_accuracy: 0.7660\n",
            "Epoch 262/500\n",
            "103/103 [==============================] - 3s 30ms/step - loss: 0.5672 - accuracy: 0.8172 - val_loss: 0.6873 - val_accuracy: 0.7626\n",
            "Epoch 263/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5448 - accuracy: 0.8149 - val_loss: 0.6895 - val_accuracy: 0.7799\n",
            "Epoch 264/500\n",
            "103/103 [==============================] - 2s 24ms/step - loss: 0.5270 - accuracy: 0.8171 - val_loss: 0.6385 - val_accuracy: 0.7851\n",
            "Epoch 265/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5919 - accuracy: 0.7839 - val_loss: 0.6456 - val_accuracy: 0.7955\n",
            "Epoch 266/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5776 - accuracy: 0.7997 - val_loss: 0.6496 - val_accuracy: 0.7834\n",
            "Epoch 267/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5535 - accuracy: 0.8005 - val_loss: 0.6853 - val_accuracy: 0.7903\n",
            "Epoch 268/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5713 - accuracy: 0.7905 - val_loss: 0.6726 - val_accuracy: 0.7747\n",
            "Epoch 269/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5480 - accuracy: 0.8053 - val_loss: 0.6462 - val_accuracy: 0.8076\n",
            "Epoch 270/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5772 - accuracy: 0.7897 - val_loss: 0.6613 - val_accuracy: 0.7868\n",
            "Epoch 271/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5488 - accuracy: 0.7948 - val_loss: 0.6301 - val_accuracy: 0.7938\n",
            "Epoch 272/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5256 - accuracy: 0.8146 - val_loss: 0.6636 - val_accuracy: 0.7972\n",
            "Epoch 273/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5650 - accuracy: 0.7939 - val_loss: 0.6882 - val_accuracy: 0.7574\n",
            "Epoch 274/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5267 - accuracy: 0.8040 - val_loss: 0.6591 - val_accuracy: 0.7920\n",
            "Epoch 275/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5332 - accuracy: 0.8042 - val_loss: 0.6402 - val_accuracy: 0.7990\n",
            "Epoch 276/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5834 - accuracy: 0.7912 - val_loss: 0.6748 - val_accuracy: 0.7747\n",
            "Epoch 277/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5252 - accuracy: 0.8068 - val_loss: 0.6633 - val_accuracy: 0.7851\n",
            "Epoch 278/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5313 - accuracy: 0.8083 - val_loss: 0.6826 - val_accuracy: 0.7695\n",
            "Epoch 279/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5432 - accuracy: 0.8052 - val_loss: 0.7404 - val_accuracy: 0.7383\n",
            "Epoch 280/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5428 - accuracy: 0.8033 - val_loss: 0.6934 - val_accuracy: 0.7747\n",
            "Epoch 281/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5355 - accuracy: 0.8002 - val_loss: 0.6502 - val_accuracy: 0.7972\n",
            "Epoch 282/500\n",
            "103/103 [==============================] - 2s 24ms/step - loss: 0.5426 - accuracy: 0.7998 - val_loss: 0.6157 - val_accuracy: 0.8059\n",
            "Epoch 283/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4989 - accuracy: 0.8301 - val_loss: 0.6603 - val_accuracy: 0.7834\n",
            "Epoch 284/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5375 - accuracy: 0.8035 - val_loss: 0.6355 - val_accuracy: 0.7955\n",
            "Epoch 285/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5791 - accuracy: 0.7995 - val_loss: 0.6717 - val_accuracy: 0.7764\n",
            "Epoch 286/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5269 - accuracy: 0.8198 - val_loss: 0.6727 - val_accuracy: 0.7643\n",
            "Epoch 287/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5560 - accuracy: 0.8116 - val_loss: 0.6955 - val_accuracy: 0.7608\n",
            "Epoch 288/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5416 - accuracy: 0.8015 - val_loss: 0.6916 - val_accuracy: 0.7695\n",
            "Epoch 289/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5142 - accuracy: 0.8162 - val_loss: 0.6811 - val_accuracy: 0.7678\n",
            "Epoch 290/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5266 - accuracy: 0.8164 - val_loss: 0.7103 - val_accuracy: 0.7574\n",
            "Epoch 291/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5625 - accuracy: 0.7931 - val_loss: 0.6673 - val_accuracy: 0.7920\n",
            "Epoch 292/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5840 - accuracy: 0.7907 - val_loss: 0.6529 - val_accuracy: 0.8059\n",
            "Epoch 293/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5491 - accuracy: 0.8029 - val_loss: 0.6556 - val_accuracy: 0.7834\n",
            "Epoch 294/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5210 - accuracy: 0.8091 - val_loss: 0.6430 - val_accuracy: 0.7920\n",
            "Epoch 295/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5572 - accuracy: 0.7988 - val_loss: 0.6731 - val_accuracy: 0.7851\n",
            "Epoch 296/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5364 - accuracy: 0.8090 - val_loss: 0.6458 - val_accuracy: 0.7920\n",
            "Epoch 297/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5396 - accuracy: 0.8027 - val_loss: 0.7129 - val_accuracy: 0.7764\n",
            "Epoch 298/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5873 - accuracy: 0.8009 - val_loss: 0.6514 - val_accuracy: 0.7903\n",
            "Epoch 299/500\n",
            "103/103 [==============================] - 2s 24ms/step - loss: 0.5574 - accuracy: 0.7947 - val_loss: 0.6620 - val_accuracy: 0.7799\n",
            "Epoch 300/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5524 - accuracy: 0.7938 - val_loss: 0.6513 - val_accuracy: 0.7903\n",
            "Epoch 301/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5843 - accuracy: 0.7937 - val_loss: 0.6680 - val_accuracy: 0.7920\n",
            "Epoch 302/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5528 - accuracy: 0.8041 - val_loss: 0.6519 - val_accuracy: 0.7938\n",
            "Epoch 303/500\n",
            "103/103 [==============================] - 2s 23ms/step - loss: 0.5143 - accuracy: 0.8050 - val_loss: 0.6334 - val_accuracy: 0.7990\n",
            "Epoch 304/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5299 - accuracy: 0.8122 - val_loss: 0.6816 - val_accuracy: 0.7834\n",
            "Epoch 305/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5342 - accuracy: 0.8059 - val_loss: 0.6651 - val_accuracy: 0.7886\n",
            "Epoch 306/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5578 - accuracy: 0.8039 - val_loss: 0.6469 - val_accuracy: 0.7920\n",
            "Epoch 307/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5286 - accuracy: 0.8076 - val_loss: 0.6476 - val_accuracy: 0.7868\n",
            "Epoch 308/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5150 - accuracy: 0.8027 - val_loss: 0.6845 - val_accuracy: 0.7574\n",
            "Epoch 309/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5029 - accuracy: 0.8275 - val_loss: 0.6434 - val_accuracy: 0.7851\n",
            "Epoch 310/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5078 - accuracy: 0.8196 - val_loss: 0.6563 - val_accuracy: 0.7972\n",
            "Epoch 311/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5432 - accuracy: 0.8138 - val_loss: 0.6412 - val_accuracy: 0.7903\n",
            "Epoch 312/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5272 - accuracy: 0.8082 - val_loss: 0.6729 - val_accuracy: 0.7938\n",
            "Epoch 313/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5252 - accuracy: 0.8138 - val_loss: 0.6808 - val_accuracy: 0.7851\n",
            "Epoch 314/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5353 - accuracy: 0.8049 - val_loss: 0.6318 - val_accuracy: 0.8111\n",
            "Epoch 315/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5390 - accuracy: 0.7987 - val_loss: 0.6843 - val_accuracy: 0.7799\n",
            "Epoch 316/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5493 - accuracy: 0.7962 - val_loss: 0.6817 - val_accuracy: 0.7712\n",
            "Epoch 317/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5114 - accuracy: 0.8100 - val_loss: 0.6034 - val_accuracy: 0.8007\n",
            "Epoch 318/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5665 - accuracy: 0.7973 - val_loss: 0.6198 - val_accuracy: 0.7938\n",
            "Epoch 319/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5375 - accuracy: 0.8061 - val_loss: 0.6738 - val_accuracy: 0.7868\n",
            "Epoch 320/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.6035 - accuracy: 0.7775 - val_loss: 0.6400 - val_accuracy: 0.7903\n",
            "Epoch 321/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5369 - accuracy: 0.8092 - val_loss: 0.7007 - val_accuracy: 0.7678\n",
            "Epoch 322/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5195 - accuracy: 0.8089 - val_loss: 0.6702 - val_accuracy: 0.7782\n",
            "Epoch 323/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5113 - accuracy: 0.8179 - val_loss: 0.6342 - val_accuracy: 0.8042\n",
            "Epoch 324/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5243 - accuracy: 0.8088 - val_loss: 0.6933 - val_accuracy: 0.7608\n",
            "Epoch 325/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4881 - accuracy: 0.8265 - val_loss: 0.6725 - val_accuracy: 0.7678\n",
            "Epoch 326/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5457 - accuracy: 0.7977 - val_loss: 0.6602 - val_accuracy: 0.7816\n",
            "Epoch 327/500\n",
            "103/103 [==============================] - 2s 24ms/step - loss: 0.5412 - accuracy: 0.8139 - val_loss: 0.6395 - val_accuracy: 0.7886\n",
            "Epoch 328/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5296 - accuracy: 0.8088 - val_loss: 0.6346 - val_accuracy: 0.7920\n",
            "Epoch 329/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.4907 - accuracy: 0.8211 - val_loss: 0.6559 - val_accuracy: 0.7834\n",
            "Epoch 330/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5478 - accuracy: 0.8074 - val_loss: 0.6222 - val_accuracy: 0.8111\n",
            "Epoch 331/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5320 - accuracy: 0.8152 - val_loss: 0.6461 - val_accuracy: 0.7955\n",
            "Epoch 332/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5735 - accuracy: 0.7829 - val_loss: 0.6438 - val_accuracy: 0.7938\n",
            "Epoch 333/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4959 - accuracy: 0.8269 - val_loss: 0.6301 - val_accuracy: 0.8059\n",
            "Epoch 334/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5112 - accuracy: 0.8127 - val_loss: 0.6270 - val_accuracy: 0.7868\n",
            "Epoch 335/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5382 - accuracy: 0.8058 - val_loss: 0.6413 - val_accuracy: 0.7712\n",
            "Epoch 336/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5009 - accuracy: 0.8249 - val_loss: 0.6707 - val_accuracy: 0.7903\n",
            "Epoch 337/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5413 - accuracy: 0.8115 - val_loss: 0.6438 - val_accuracy: 0.7868\n",
            "Epoch 338/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5643 - accuracy: 0.8071 - val_loss: 0.6134 - val_accuracy: 0.8042\n",
            "Epoch 339/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5565 - accuracy: 0.8020 - val_loss: 0.6710 - val_accuracy: 0.7903\n",
            "Epoch 340/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4908 - accuracy: 0.8314 - val_loss: 0.6117 - val_accuracy: 0.8024\n",
            "Epoch 341/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5076 - accuracy: 0.8110 - val_loss: 0.6416 - val_accuracy: 0.7990\n",
            "Epoch 342/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5257 - accuracy: 0.8091 - val_loss: 0.6596 - val_accuracy: 0.7851\n",
            "Epoch 343/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5515 - accuracy: 0.8026 - val_loss: 0.6630 - val_accuracy: 0.7938\n",
            "Epoch 344/500\n",
            "103/103 [==============================] - 3s 26ms/step - loss: 0.5164 - accuracy: 0.8179 - val_loss: 0.6635 - val_accuracy: 0.7938\n",
            "Epoch 345/500\n",
            "103/103 [==============================] - 2s 23ms/step - loss: 0.5236 - accuracy: 0.8084 - val_loss: 0.6676 - val_accuracy: 0.7886\n",
            "Epoch 346/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5289 - accuracy: 0.8051 - val_loss: 0.6686 - val_accuracy: 0.7851\n",
            "Epoch 347/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5565 - accuracy: 0.8133 - val_loss: 0.6833 - val_accuracy: 0.7903\n",
            "Epoch 348/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5319 - accuracy: 0.8111 - val_loss: 0.6679 - val_accuracy: 0.8024\n",
            "Epoch 349/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5293 - accuracy: 0.8089 - val_loss: 0.6964 - val_accuracy: 0.7747\n",
            "Epoch 350/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5384 - accuracy: 0.8068 - val_loss: 0.6240 - val_accuracy: 0.8198\n",
            "Epoch 351/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5282 - accuracy: 0.8085 - val_loss: 0.6579 - val_accuracy: 0.7886\n",
            "Epoch 352/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5367 - accuracy: 0.7944 - val_loss: 0.6802 - val_accuracy: 0.7868\n",
            "Epoch 353/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5048 - accuracy: 0.8186 - val_loss: 0.6764 - val_accuracy: 0.7816\n",
            "Epoch 354/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5345 - accuracy: 0.8080 - val_loss: 0.7018 - val_accuracy: 0.7834\n",
            "Epoch 355/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5595 - accuracy: 0.7954 - val_loss: 0.6543 - val_accuracy: 0.7868\n",
            "Epoch 356/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5370 - accuracy: 0.8074 - val_loss: 0.6675 - val_accuracy: 0.7903\n",
            "Epoch 357/500\n",
            "103/103 [==============================] - 2s 20ms/step - loss: 0.5150 - accuracy: 0.8088 - val_loss: 0.6580 - val_accuracy: 0.7868\n",
            "Epoch 358/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5207 - accuracy: 0.8079 - val_loss: 0.6396 - val_accuracy: 0.7990\n",
            "Epoch 359/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5309 - accuracy: 0.8140 - val_loss: 0.6792 - val_accuracy: 0.7730\n",
            "Epoch 360/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5583 - accuracy: 0.8033 - val_loss: 0.6568 - val_accuracy: 0.7886\n",
            "Epoch 361/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5156 - accuracy: 0.8203 - val_loss: 0.7034 - val_accuracy: 0.7678\n",
            "Epoch 362/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5064 - accuracy: 0.8230 - val_loss: 0.6635 - val_accuracy: 0.7747\n",
            "Epoch 363/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4828 - accuracy: 0.8237 - val_loss: 0.6719 - val_accuracy: 0.7834\n",
            "Epoch 364/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5000 - accuracy: 0.8141 - val_loss: 0.6275 - val_accuracy: 0.7920\n",
            "Epoch 365/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5410 - accuracy: 0.8091 - val_loss: 0.6514 - val_accuracy: 0.7851\n",
            "Epoch 366/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5059 - accuracy: 0.8131 - val_loss: 0.6911 - val_accuracy: 0.7730\n",
            "Epoch 367/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5084 - accuracy: 0.8229 - val_loss: 0.6878 - val_accuracy: 0.7782\n",
            "Epoch 368/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5544 - accuracy: 0.8065 - val_loss: 0.6563 - val_accuracy: 0.7955\n",
            "Epoch 369/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5161 - accuracy: 0.8175 - val_loss: 0.6680 - val_accuracy: 0.7851\n",
            "Epoch 370/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5062 - accuracy: 0.8054 - val_loss: 0.6908 - val_accuracy: 0.7816\n",
            "Epoch 371/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5666 - accuracy: 0.7992 - val_loss: 0.6913 - val_accuracy: 0.7643\n",
            "Epoch 372/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5174 - accuracy: 0.7974 - val_loss: 0.6743 - val_accuracy: 0.7816\n",
            "Epoch 373/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5383 - accuracy: 0.8128 - val_loss: 0.6302 - val_accuracy: 0.8007\n",
            "Epoch 374/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5392 - accuracy: 0.8125 - val_loss: 0.6487 - val_accuracy: 0.7851\n",
            "Epoch 375/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5073 - accuracy: 0.8188 - val_loss: 0.6400 - val_accuracy: 0.8007\n",
            "Epoch 376/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5007 - accuracy: 0.8269 - val_loss: 0.6536 - val_accuracy: 0.7938\n",
            "Epoch 377/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5277 - accuracy: 0.8087 - val_loss: 0.6908 - val_accuracy: 0.7730\n",
            "Epoch 378/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5210 - accuracy: 0.8168 - val_loss: 0.7011 - val_accuracy: 0.7591\n",
            "Epoch 379/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5192 - accuracy: 0.8101 - val_loss: 0.6876 - val_accuracy: 0.7626\n",
            "Epoch 380/500\n",
            "103/103 [==============================] - 2s 24ms/step - loss: 0.5122 - accuracy: 0.8160 - val_loss: 0.6697 - val_accuracy: 0.7764\n",
            "Epoch 381/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5386 - accuracy: 0.8062 - val_loss: 0.6916 - val_accuracy: 0.7730\n",
            "Epoch 382/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5142 - accuracy: 0.8132 - val_loss: 0.6745 - val_accuracy: 0.7868\n",
            "Epoch 383/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5112 - accuracy: 0.8177 - val_loss: 0.6537 - val_accuracy: 0.7886\n",
            "Epoch 384/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.4745 - accuracy: 0.8319 - val_loss: 0.6732 - val_accuracy: 0.7730\n",
            "Epoch 385/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5184 - accuracy: 0.8155 - val_loss: 0.6330 - val_accuracy: 0.7972\n",
            "Epoch 386/500\n",
            "103/103 [==============================] - 3s 25ms/step - loss: 0.5500 - accuracy: 0.7981 - val_loss: 0.6493 - val_accuracy: 0.7903\n",
            "Epoch 387/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5328 - accuracy: 0.8153 - val_loss: 0.6459 - val_accuracy: 0.7938\n",
            "Epoch 388/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5074 - accuracy: 0.8240 - val_loss: 0.6558 - val_accuracy: 0.7955\n",
            "Epoch 389/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5197 - accuracy: 0.8201 - val_loss: 0.6618 - val_accuracy: 0.7903\n",
            "Epoch 390/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5475 - accuracy: 0.8060 - val_loss: 0.6386 - val_accuracy: 0.8007\n",
            "Epoch 391/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5118 - accuracy: 0.8153 - val_loss: 0.6298 - val_accuracy: 0.8163\n",
            "Epoch 392/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4979 - accuracy: 0.8196 - val_loss: 0.6672 - val_accuracy: 0.7730\n",
            "Epoch 393/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5263 - accuracy: 0.8140 - val_loss: 0.6435 - val_accuracy: 0.7903\n",
            "Epoch 394/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5239 - accuracy: 0.8144 - val_loss: 0.6839 - val_accuracy: 0.7730\n",
            "Epoch 395/500\n",
            "103/103 [==============================] - 2s 20ms/step - loss: 0.5100 - accuracy: 0.8210 - val_loss: 0.6562 - val_accuracy: 0.7834\n",
            "Epoch 396/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5026 - accuracy: 0.8187 - val_loss: 0.6507 - val_accuracy: 0.7920\n",
            "Epoch 397/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5232 - accuracy: 0.8160 - val_loss: 0.6479 - val_accuracy: 0.8007\n",
            "Epoch 398/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5129 - accuracy: 0.8200 - val_loss: 0.6904 - val_accuracy: 0.7695\n",
            "Epoch 399/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4983 - accuracy: 0.8193 - val_loss: 0.6552 - val_accuracy: 0.7868\n",
            "Epoch 400/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5339 - accuracy: 0.8102 - val_loss: 0.7009 - val_accuracy: 0.7903\n",
            "Epoch 401/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5211 - accuracy: 0.8100 - val_loss: 0.6598 - val_accuracy: 0.7903\n",
            "Epoch 402/500\n",
            "103/103 [==============================] - 3s 25ms/step - loss: 0.5241 - accuracy: 0.8254 - val_loss: 0.6567 - val_accuracy: 0.7955\n",
            "Epoch 403/500\n",
            "103/103 [==============================] - 2s 20ms/step - loss: 0.4960 - accuracy: 0.8180 - val_loss: 0.6423 - val_accuracy: 0.7920\n",
            "Epoch 404/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5229 - accuracy: 0.8114 - val_loss: 0.6399 - val_accuracy: 0.7955\n",
            "Epoch 405/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5293 - accuracy: 0.8089 - val_loss: 0.6433 - val_accuracy: 0.7955\n",
            "Epoch 406/500\n",
            "103/103 [==============================] - 2s 20ms/step - loss: 0.5127 - accuracy: 0.8115 - val_loss: 0.6440 - val_accuracy: 0.7938\n",
            "Epoch 407/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4821 - accuracy: 0.8294 - val_loss: 0.6615 - val_accuracy: 0.7799\n",
            "Epoch 408/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5076 - accuracy: 0.8137 - val_loss: 0.6945 - val_accuracy: 0.7816\n",
            "Epoch 409/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5055 - accuracy: 0.8185 - val_loss: 0.6749 - val_accuracy: 0.7816\n",
            "Epoch 410/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5609 - accuracy: 0.8048 - val_loss: 0.6903 - val_accuracy: 0.7764\n",
            "Epoch 411/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5511 - accuracy: 0.8119 - val_loss: 0.6533 - val_accuracy: 0.8024\n",
            "Epoch 412/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5240 - accuracy: 0.8079 - val_loss: 0.6664 - val_accuracy: 0.7938\n",
            "Epoch 413/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5105 - accuracy: 0.8149 - val_loss: 0.6612 - val_accuracy: 0.7938\n",
            "Epoch 414/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4926 - accuracy: 0.8219 - val_loss: 0.6496 - val_accuracy: 0.7903\n",
            "Epoch 415/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5471 - accuracy: 0.8012 - val_loss: 0.6386 - val_accuracy: 0.7990\n",
            "Epoch 416/500\n",
            "103/103 [==============================] - 2s 23ms/step - loss: 0.5161 - accuracy: 0.8093 - val_loss: 0.6382 - val_accuracy: 0.7886\n",
            "Epoch 417/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4664 - accuracy: 0.8367 - val_loss: 0.6419 - val_accuracy: 0.8042\n",
            "Epoch 418/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5200 - accuracy: 0.8129 - val_loss: 0.6309 - val_accuracy: 0.8111\n",
            "Epoch 419/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5408 - accuracy: 0.8002 - val_loss: 0.6659 - val_accuracy: 0.7868\n",
            "Epoch 420/500\n",
            "103/103 [==============================] - 2s 20ms/step - loss: 0.5323 - accuracy: 0.8099 - val_loss: 0.6438 - val_accuracy: 0.7886\n",
            "Epoch 421/500\n",
            "103/103 [==============================] - 2s 20ms/step - loss: 0.5306 - accuracy: 0.8082 - val_loss: 0.6665 - val_accuracy: 0.7920\n",
            "Epoch 422/500\n",
            "103/103 [==============================] - 2s 20ms/step - loss: 0.4906 - accuracy: 0.8338 - val_loss: 0.6775 - val_accuracy: 0.7764\n",
            "Epoch 423/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5098 - accuracy: 0.8150 - val_loss: 0.6776 - val_accuracy: 0.7764\n",
            "Epoch 424/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5310 - accuracy: 0.8080 - val_loss: 0.6628 - val_accuracy: 0.7747\n",
            "Epoch 425/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4999 - accuracy: 0.8179 - val_loss: 0.6459 - val_accuracy: 0.7816\n",
            "Epoch 426/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5411 - accuracy: 0.8116 - val_loss: 0.6312 - val_accuracy: 0.7955\n",
            "Epoch 427/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4962 - accuracy: 0.8300 - val_loss: 0.7113 - val_accuracy: 0.7730\n",
            "Epoch 428/500\n",
            "103/103 [==============================] - 2s 24ms/step - loss: 0.5161 - accuracy: 0.8138 - val_loss: 0.6564 - val_accuracy: 0.7990\n",
            "Epoch 429/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.4806 - accuracy: 0.8279 - val_loss: 0.6461 - val_accuracy: 0.7938\n",
            "Epoch 430/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5168 - accuracy: 0.8302 - val_loss: 0.6095 - val_accuracy: 0.8111\n",
            "Epoch 431/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4914 - accuracy: 0.8207 - val_loss: 0.6162 - val_accuracy: 0.7938\n",
            "Epoch 432/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5153 - accuracy: 0.8174 - val_loss: 0.6461 - val_accuracy: 0.7955\n",
            "Epoch 433/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4883 - accuracy: 0.8259 - val_loss: 0.6455 - val_accuracy: 0.7955\n",
            "Epoch 434/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5002 - accuracy: 0.8212 - val_loss: 0.6332 - val_accuracy: 0.7955\n",
            "Epoch 435/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5214 - accuracy: 0.8194 - val_loss: 0.6580 - val_accuracy: 0.7816\n",
            "Epoch 436/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5315 - accuracy: 0.8081 - val_loss: 0.6437 - val_accuracy: 0.8076\n",
            "Epoch 437/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4935 - accuracy: 0.8330 - val_loss: 0.6221 - val_accuracy: 0.8059\n",
            "Epoch 438/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5070 - accuracy: 0.8140 - val_loss: 0.6863 - val_accuracy: 0.7574\n",
            "Epoch 439/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5413 - accuracy: 0.8142 - val_loss: 0.6914 - val_accuracy: 0.7816\n",
            "Epoch 440/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5353 - accuracy: 0.8186 - val_loss: 0.6377 - val_accuracy: 0.8024\n",
            "Epoch 441/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4860 - accuracy: 0.8271 - val_loss: 0.6583 - val_accuracy: 0.7712\n",
            "Epoch 442/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5265 - accuracy: 0.8118 - val_loss: 0.6481 - val_accuracy: 0.7972\n",
            "Epoch 443/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.4836 - accuracy: 0.8307 - val_loss: 0.7262 - val_accuracy: 0.7522\n",
            "Epoch 444/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.4667 - accuracy: 0.8235 - val_loss: 0.6571 - val_accuracy: 0.7903\n",
            "Epoch 445/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4712 - accuracy: 0.8336 - val_loss: 0.6322 - val_accuracy: 0.7938\n",
            "Epoch 446/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4901 - accuracy: 0.8249 - val_loss: 0.7266 - val_accuracy: 0.7452\n",
            "Epoch 447/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4975 - accuracy: 0.8279 - val_loss: 0.6548 - val_accuracy: 0.7972\n",
            "Epoch 448/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4974 - accuracy: 0.8216 - val_loss: 0.6349 - val_accuracy: 0.8059\n",
            "Epoch 449/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4887 - accuracy: 0.8298 - val_loss: 0.6480 - val_accuracy: 0.7903\n",
            "Epoch 450/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5300 - accuracy: 0.8100 - val_loss: 0.6473 - val_accuracy: 0.7972\n",
            "Epoch 451/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5014 - accuracy: 0.8187 - val_loss: 0.6782 - val_accuracy: 0.7764\n",
            "Epoch 452/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4776 - accuracy: 0.8291 - val_loss: 0.6450 - val_accuracy: 0.7972\n",
            "Epoch 453/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5094 - accuracy: 0.8284 - val_loss: 0.6760 - val_accuracy: 0.7730\n",
            "Epoch 454/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4978 - accuracy: 0.8163 - val_loss: 0.6384 - val_accuracy: 0.7920\n",
            "Epoch 455/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4980 - accuracy: 0.8461 - val_loss: 0.6585 - val_accuracy: 0.7938\n",
            "Epoch 456/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5188 - accuracy: 0.8172 - val_loss: 0.6560 - val_accuracy: 0.8007\n",
            "Epoch 457/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5348 - accuracy: 0.8157 - val_loss: 0.6817 - val_accuracy: 0.7695\n",
            "Epoch 458/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4865 - accuracy: 0.8387 - val_loss: 0.6087 - val_accuracy: 0.8076\n",
            "Epoch 459/500\n",
            "103/103 [==============================] - 2s 24ms/step - loss: 0.5109 - accuracy: 0.8186 - val_loss: 0.6850 - val_accuracy: 0.7834\n",
            "Epoch 460/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5185 - accuracy: 0.8212 - val_loss: 0.6722 - val_accuracy: 0.7851\n",
            "Epoch 461/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4813 - accuracy: 0.8294 - val_loss: 0.6188 - val_accuracy: 0.8042\n",
            "Epoch 462/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.4876 - accuracy: 0.8224 - val_loss: 0.6400 - val_accuracy: 0.7938\n",
            "Epoch 463/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.4880 - accuracy: 0.8298 - val_loss: 0.6377 - val_accuracy: 0.8042\n",
            "Epoch 464/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4706 - accuracy: 0.8323 - val_loss: 0.6350 - val_accuracy: 0.7886\n",
            "Epoch 465/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5032 - accuracy: 0.8283 - val_loss: 0.6314 - val_accuracy: 0.7851\n",
            "Epoch 466/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5188 - accuracy: 0.8184 - val_loss: 0.6795 - val_accuracy: 0.7816\n",
            "Epoch 467/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4447 - accuracy: 0.8416 - val_loss: 0.6634 - val_accuracy: 0.7938\n",
            "Epoch 468/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4929 - accuracy: 0.8206 - val_loss: 0.6805 - val_accuracy: 0.7712\n",
            "Epoch 469/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4720 - accuracy: 0.8319 - val_loss: 0.6831 - val_accuracy: 0.7764\n",
            "Epoch 470/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4993 - accuracy: 0.8068 - val_loss: 0.6312 - val_accuracy: 0.7903\n",
            "Epoch 471/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5202 - accuracy: 0.8149 - val_loss: 0.6550 - val_accuracy: 0.7782\n",
            "Epoch 472/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.4982 - accuracy: 0.8244 - val_loss: 0.6232 - val_accuracy: 0.7868\n",
            "Epoch 473/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5511 - accuracy: 0.8115 - val_loss: 0.6609 - val_accuracy: 0.7990\n",
            "Epoch 474/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4875 - accuracy: 0.8213 - val_loss: 0.6587 - val_accuracy: 0.7886\n",
            "Epoch 475/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5561 - accuracy: 0.8062 - val_loss: 0.6174 - val_accuracy: 0.8146\n",
            "Epoch 476/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5108 - accuracy: 0.8148 - val_loss: 0.6458 - val_accuracy: 0.8007\n",
            "Epoch 477/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4935 - accuracy: 0.8141 - val_loss: 0.6367 - val_accuracy: 0.8094\n",
            "Epoch 478/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4919 - accuracy: 0.8203 - val_loss: 0.6809 - val_accuracy: 0.7920\n",
            "Epoch 479/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5109 - accuracy: 0.8203 - val_loss: 0.6391 - val_accuracy: 0.8007\n",
            "Epoch 480/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.4953 - accuracy: 0.8222 - val_loss: 0.6726 - val_accuracy: 0.7764\n",
            "Epoch 481/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4953 - accuracy: 0.8258 - val_loss: 0.7176 - val_accuracy: 0.7712\n",
            "Epoch 482/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5408 - accuracy: 0.8109 - val_loss: 0.6570 - val_accuracy: 0.7782\n",
            "Epoch 483/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4778 - accuracy: 0.8300 - val_loss: 0.6372 - val_accuracy: 0.8042\n",
            "Epoch 484/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.4891 - accuracy: 0.8244 - val_loss: 0.6622 - val_accuracy: 0.7886\n",
            "Epoch 485/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.4872 - accuracy: 0.8271 - val_loss: 0.6512 - val_accuracy: 0.7886\n",
            "Epoch 486/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.5107 - accuracy: 0.8153 - val_loss: 0.6666 - val_accuracy: 0.7730\n",
            "Epoch 487/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.4653 - accuracy: 0.8239 - val_loss: 0.6546 - val_accuracy: 0.7851\n",
            "Epoch 488/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4986 - accuracy: 0.8237 - val_loss: 0.6964 - val_accuracy: 0.7643\n",
            "Epoch 489/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5082 - accuracy: 0.8245 - val_loss: 0.6995 - val_accuracy: 0.7799\n",
            "Epoch 490/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4632 - accuracy: 0.8340 - val_loss: 0.6416 - val_accuracy: 0.7972\n",
            "Epoch 491/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4846 - accuracy: 0.8285 - val_loss: 0.6774 - val_accuracy: 0.7816\n",
            "Epoch 492/500\n",
            "103/103 [==============================] - 2s 24ms/step - loss: 0.4928 - accuracy: 0.8189 - val_loss: 0.6603 - val_accuracy: 0.7712\n",
            "Epoch 493/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4920 - accuracy: 0.8277 - val_loss: 0.6647 - val_accuracy: 0.7886\n",
            "Epoch 494/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4974 - accuracy: 0.8107 - val_loss: 0.6686 - val_accuracy: 0.8042\n",
            "Epoch 495/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5016 - accuracy: 0.8288 - val_loss: 0.6102 - val_accuracy: 0.8059\n",
            "Epoch 496/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5269 - accuracy: 0.8203 - val_loss: 0.6496 - val_accuracy: 0.8076\n",
            "Epoch 497/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.4990 - accuracy: 0.8124 - val_loss: 0.6299 - val_accuracy: 0.8354\n",
            "Epoch 498/500\n",
            "103/103 [==============================] - 2s 21ms/step - loss: 0.5086 - accuracy: 0.8196 - val_loss: 0.6750 - val_accuracy: 0.7938\n",
            "Epoch 499/500\n",
            "103/103 [==============================] - 2s 22ms/step - loss: 0.4864 - accuracy: 0.8342 - val_loss: 0.6536 - val_accuracy: 0.7955\n",
            "Epoch 500/500\n",
            "103/103 [==============================] - 2s 23ms/step - loss: 0.4868 - accuracy: 0.8254 - val_loss: 0.6498 - val_accuracy: 0.7834\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DHdEBImL7sLz",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 533
        },
        "outputId": "40993818-590d-458b-af7d-4fb6a5fc4868"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "plt.figure(figsize=(8,8))\n",
        "plt.title('Loss Value')\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.legend(['loss', 'val_loss'])\n",
        "print('loss:', history.history['loss'][-1])\n",
        "print('val_loss:', history.history['val_loss'][-1])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "loss: 0.49377259612083435\n",
            "val_loss: 0.6498277187347412\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x576 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9sEY-Ng288eM",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 533
        },
        "outputId": "e808679f-8a9c-4964-9714-6c1596dd4349"
      },
      "source": [
        "plt.figure(figsize=(8,8))\n",
        "plt.title('Accuracy')\n",
        "plt.plot(history.history['accuracy'])\n",
        "plt.plot(history.history['val_accuracy'])\n",
        "plt.legend(['acc', 'val_acc'])\n",
        "print('acc:', history.history['accuracy'][-1])\n",
        "print('val_acc:', history.history['val_accuracy'][-1])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "acc: 0.8213521242141724\n",
            "val_acc: 0.7833622097969055\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAeMAAAHiCAYAAADbK6SdAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOydeXhU1fnHP2e2TPY9JBBCwo4YEHBBRRFX3Fek7lqttlqt1dba1l9rW+ve1lqt1rUutai4a0VFQBRR2RfZdxJC9nWSzCSZ+/vj3Dt3mckCAol6Ps+TJzN37nLmzsz5nvc973lfoWkaCoVCoVAoeg9XbzdAoVAoFIrvO0qMFQqFQqHoZZQYKxQKhULRyygxVigUCoWil1FirFAoFApFL6PEWKFQKBSKXkaJsUKhUCgUvYwSY4WijyGEmCeEqBVCxPV2WxQKxYFBibFC0YcQQhQCxwAacNYBvK7nQF1LoVBEo8RYoehbXA58AfwbuMLYKIQYKIR4XQhRKYSoFkI8YnntR0KItUKIRiHEGiHEeH27JoQYatnv30KIu/THxwkhSoQQvxJC7AaeFUKkCyHe1a9Rqz/OtxyfIYR4VgixS3/9TX37aiHEmZb9vEKIKiHEuP12lxSK7xhKjBWKvsXlwH/0v1OEEP2EEG7gXWA7UAgMAGYACCGmAXfqx6UgrenqHl4rF8gABgHXIvuDZ/XnBUAL8Ihl/xeABGA0kAP8Td/+PHCpZb/TgDJN05b1sB0KxfceoXJTKxR9AyHEJGAukKdpWpUQYh3wL6Sl/La+vd1xzAfA/zRN+3uM82nAME3TNunP/w2UaJp2hxDiOOBDIEXTtNZO2nMIMFfTtHQhRB5QCmRqmlbr2K8/sB4YoGlagxBiJvCVpmn37/XNUCi+ZyjLWKHoO1wBfKhpWpX+/CV920Bgu1OIdQYCm/fyepVWIRZCJAgh/iWE2C6EaADmA2m6ZT4QqHEKMYCmabuABcD5Qog04FSkZa9QKHqICtpQKPoAQoh44ELArc/hAsQBaUA5UCCE8MQQ5J3AkE5O24x0KxvkAiWW50632K3ACOAITdN265bxMkDo18kQQqRpmlYX41rPAdcg+5SFmqaVdv5uFQqFE2UZKxR9g3OADuAg4BD9bxTwqf5aGXCvECJRCOEXQhytH/cU8AshxAQhGSqEGKS/thy4WAjhFkJMBSZ304Zk5DxxnRAiA/i98YKmaWXA+8A/9UAvrxDiWMuxbwLjgZ8h55AVCsUeoMRYoegbXAE8q2naDk3Tdht/yACqi4AzgaHADqR1Ox1A07RXgT8jXdqNSFHM0M/5M/24OuAS/bWueAiIB6qQ89SzHK9fBrQB64AK4GbjBU3TWoDXgCLg9T187wrF9x4VwKVQKPYJQojfAcM1Tbu0250VCoUNNWesUCi+Mbpb+2qk9axQKPYQ5aZWKBTfCCHEj5ABXu9rmja/t9ujUHwbUW5qhUKhUCh6GWUZKxQKhULRyygxVigUCoWil+m1AK6srCytsLCwty6vUCgUCsUBZcmSJVWapmXHeq3XxLiwsJDFixf31uUVCoVCoTigCCG2d/aaclMrFAqFQtHLKDFWKBQKhaKXUWKsUCgUCkUv06cycLW1tVFSUkJra8zyqt97/H4/+fn5eL3e3m6KQqFQKPYhfUqMS0pKSE5OprCwECFEbzenT6FpGtXV1ZSUlFBUVNTbzVEoFArFPqRPualbW1vJzMxUQhwDIQSZmZnKa6BQKBTfQfqUGANKiLtA3RuFQqH4btLnxFihUCgUiu8bSowVCoVCoehllBjH4JxzzmHChAmMHj2aJ554AoBZs2Yxfvx4xo4dywknnABAU1MTV111FcXFxYwZM4bXXnutN5utUCgUim8pfSqa2sof3vmaNbsa9uk5D+qfwu/PHN3tfs888wwZGRm0tLRw2GGHcfbZZ/OjH/2I+fPnU1RURE1NDQB/+tOfSE1NZdWqVQDU1tbu0/YqFAqF4vtBnxXj3uThhx/mjTfeAGDnzp088cQTHHvssZElRRkZGQDMnj2bGTNmRI5LT08/8I1VKBQKxbeePivGPbFg9wfz5s1j9uzZLFy4kISEBI477jgOOeQQ1q1b1yvtUSgUCsV3HzVn7KC+vp709HQSEhJYt24dX3zxBa2trcyfP5+tW7cCRNzUJ510Eo8++mjkWOWmVigUCsXeoMTYwdSpU2lvb2fUqFHcfvvtTJw4kezsbJ544gnOO+88xo4dy/Tp0wG44447qK2t5eCDD2bs2LHMnTu3l1uvUCgUim8jfdZN3VvExcXx/vvvx3zt1FNPtT1PSkriueeeOxDNUigUCsV3GGUZKxQKhULRyygxVigUCkWvUtkYZMydH7Bk+4GLu7n4yS84/i/zol9oroGOtgPWDgMlxgqFQqHoVTaUN9LQ2s6KnXUH7Jqfb65mS2XAvjHcAfcXwTs/O2DtMFBirFAoFIpepbSuBYBd+v/9jaZpMR/T1iz/r5p5QNphRYmxQqFQKHqVsjpZGnZXfQvbqwNsrmyKud+63Q389KWlhNrDe32tto4wb6/YFXneGGw3XwzpYuw68LHNSowVCoVC0auU1RuWcSuTH5jHCX/5JOZ+P31pGe+uLGNTRWyxjoWmadQ3m3PALyzczs9mLI88r2oMmjuH9PO63HvQ+n2DEmOFQqFQHBCW76xj7rqKyPOKxlbW7GrokZt62Y5aKnXhrGsJ9fiaT3+2lbF//DBy7voWe3BWpVWMDTe1EuNvF0lJSb3dBIVC8R2kOdTe/U69SFOwnZUlex5sdc6jC7jq34siz298aRmnPfwpS/Uo6gqLMIbD5lzu7vpWzv3n5xEhvfjJL/nNG6t6dM33V+8GYHu1FNqGVrsYVzWZwr5ss+6+Vm5qhUKh+H6zubKJg373AW8uK+3tpnTKFc98xVmPLKCisbVH+weC9sGFMdj4Wq/MFwh1kOK3C2CdLrzLdtTyy5kros750pc7enRtj0sAUB2QQr+7Xrb5vHEDALjhpaX8b1UZmqbx1Mcr5UG9IMZ9NwPX+7fD7p6NfHpMbjGcem+nL99+++0MHDiQG264AYA777wTj8fD3Llzqa2tpa2tjbvuuouzzz6720s1NTVx9tlnxzzu+eef58EHH0QIwZgxY3jhhRcoLy/nxz/+MVu2bAHgscce46ijjtoHb1qhUPR16ppDzF1fwbnj8lldWg/A7LXlnKMLxr6iKdjO1soAxfmpe32ObVWByHrgD1bv5rIjC7vcf3d9KxPv+Zg7Th9lOUczg7MTaWnrYGx+KocVZjAoM4H/e+vryD6LttWwfGcdzy7YSmvb3gds+TzS5jSCxMrqW5k0NIsHpo3ldX3Ac/1/lvLytRNpbw2ADzpw8drinVwwPh+XLub7m74rxr3A9OnTufnmmyNi/Morr/DBBx9w0003kZKSQlVVFRMnTuSss85CiK4/IL/fzxtvvBF13Jo1a7jrrrv4/PPPycrKihSduOmmm5g8eTJvvPEGHR0dNDX1PEBBoVD0PX77xipyU/zceMKwbvd9fWkpf3x3DUcUZdLa1gFAnEfOW/70paUcNSSLi48o+EbtqW4KctyD82hsbefFq49g0rCsvTrPjEU7cbsEiT43by3f1akYB4LtVDYGmb+xEoC73lsbee20hz/ll6eMoCOsce2xQzh9TB5bHBHU172wBIDCzASOG5HD8p11LO/hOuS65hCBUAf9kuMI6pHXu/QgsfKGVoYOzcLtENlH5m4iE916bgxx28yV+L1uzhrbv0fX/Kb0XTHuwoLdX4wbN46Kigp27dpFZWUl6enp5Obm8vOf/5z58+fjcrkoLS2lvLyc3NzcLs+laRq/+c1voo6bM2cO06ZNIytL/hCM2shz5szh+eefB8DtdpOauvcjV4VC0btsLG/kP7obtSdibMyVnv/Y55TpblSfx0VFQyvvrizj3ZVlXYpxOKxFLLiaQIiOsEZ2cpxtnxUldTS2SvfwfbPW8d9FCZwwMofzxud3el5N0yKGR2VjEE3TeGt5KccOy+LooVnc9d5avtpaw/iCNLZVB2hobSfe60YI+MWrK1hd2sDRQzNjnvuBD9YDcFD/FAD6p8VH7fPqj4/ksELZRwbbO3hy/hYe/HBD5PX2jjAet4t/fLyRxmA7vzlNWt8X/mshG8rt4v7Oil00Bzsoq28lL9UPwLs3TiLZ7+G4B+fx6cYqLnbLzyEUltb0wx9v5PTivCjh3h/0XTHuJaZNm8bMmTPZvXs306dP5z//+Q+VlZUsWbIEr9dLYWEhra3dz5Ps7XEKhaLvsHBzNX/5cD0v/WhixN3J2ndg3Xtw7uOdHjdj0c7I44rGVnKS/V1ex4joNYQYIM7j4outNZHn26oCFGYlRp5vrQpQ1RRk0bYa7p+1nltPGs4xw7M559EFjMxNZtbNxwLw+aYqSmpbCOjztBcdXsDLi3awqrSe91aW2cTYKr7B9g5G3DGLG6YMYVhOMje/bC4H+tXUkZwyOpd/zd/CL15dQV6qny+31pCZ6MPlErYI5QWbqjn5oH58uKY86n0n+NwMykgAwO91k5HooyZgBlSNG5hmuR9ufnr8MJsY17W0kRrv5S8fyW2/PGUEXrcrSohBBmq9vFh+LsZA5eAB0ugpyEhge3UzSULe/w7c/Ons0ZTUthBqDxPv2//R1SqAy8H06dOZMWMGM2fOZNq0adTX15OTk4PX62Xu3Lls3769R+fp7Ljjjz+eV199lerqasCsjXzCCSfw2GOPAdDR0UF9ff1+eHcKxbcfW8akHrBoWw07a5r36lq3vbaCxdtrKam1HP/ypbDiv10et8NyvWU7unetVjYFo7YF28N8saU68vzTTVWRx2t2NTDlwXlMe3wh98+SFuYTn25h5hIpNut2N0YE8ZkFW7n/g/XsqGkmweemeEAqlkDlSNRyRWMrh941m/dWlgGwdLts96NzN9uE2CXguBHZxPvcPHn5odQGQnypDxqqAyH7UiGdU0abnsSfnTCMcQVSZEfmJtvmZPun2QctHne0RD12yXiyknx6G2tZZBmwrCqtpzbQ+bKn168/iuuPG2JrD8CwHLkyZkSGFN0OXFw6cRC/Pm3UARFiUGIcxejRo2lsbGTAgAHk5eVxySWXsHjxYoqLi3n++ecZOXJkj87T2XGjR4/mt7/9LZMnT2bs2LHccsstAPz9739n7ty5FBcXM2HCBNasWbPf3qNC8W3llpeXM+y37/Px2mgrKxZNwXamPb6Qy57+cq+u53XJLrK2ObpwQGswxDXPLWKNHhGsaRoPzd7Aut0NVDUFOawwHZ/bxaKtNbS2dbC1KmBbrmNQVt/C7vro9bWNrW3MWVvBSQf1I9HnZrMl0cXG8noucn+MFzNKuTAzMbJ8B4gIeWldK9WBIFsqAxRkJJDnELyZS0t46tMtPPPZNqoDIT7bJOd4F1jEH6QIA4wrSCctQYrhIQPTeP7qwxlrsWANph86MPLYGjA2ZWQOxw3PAUwXtUFeqnRV//rUkbz2k9gBrKcW5/HQ9HEAXPvCEi5+yvxsF26uZvUuacicN24AQ3WRvejwAl760RGML0jntqkj6ZdivwfD+iXL9mRJ4fXQ0W1c0L5GualjsGqVGcWdlZXFwoULY+7XVZBVV8ddccUVXHHFFbZt/fr146233tqL1ioU3x/eXVVGe1hj6Y5aThjVr9P9dtY084d31jBJn6+sjmEtfbKhkiue+YqFvz6euesqGZOfGnFbGnjc5nypkw2llcxeW8H26mY+umUy63Y38tDsjTw0eyOp8V6mjMjG53HxyYZK4n1u/jFnE6ePyWNUbjIXTBhIbqqfHdXNHPvA3Jjv4dONVdS3tHF78UjKG1ptKSKT177CPd6nSSPAC57zcAkp6vUtbZxenMf8DZV8vrmaM8f2Z1ddC5oGS7bXctSQzMh8qcFtM+VyHsMNbyw3+nyzKcZHFGXw3x9N5OevLOf4kTm248cVpPPWDUdzzP1z2FkjBxVHD83kvgvGMHttOU3BdgZb3OuZiT6G5MjnB+XZ73d+ejw+j4trjx3cpRimJ3ptz+85r5iP5s5j67attLYVIAT8/szRLNtZy5XPLmLi4AyOGtJ5wNq0Cfn43C4GNn4IQJxQVZsUCkUvo2kaT8zfbJu76wmfb66KRALvDwLB9khOYqPT74yHZm9k9tpy7nxHepgKMhJobG1j4WbT7fuunp/49aWl/OaNVZzxj88AaV19+LVMFOHRLeNYbuRAUyMAtc3yPlktyfqWNrKS4pgyIoeNFU18+LW05N9bWcaDH27gxy/KSOFH527q9D3Ut7ThcQlOGJXD4KxEPt1YxTH3z6G1rYNwfQkA8aKVn50wjOsmD6GqKcSOmmaKshI5YnAGCzdX0RRsjyTKaAq2S8s41QyU+ucl4zlTjxYOtYc5akgm63Y30trWwepdDRGBHqG7k//+g3GcfUjs5Vaj81JJivOQ4HMzoSAdkJbv2Pw0PG4Xw/tJKzUzycfhRRlMGJTOscPtAnndsUN46vJDu7VK03XLHGD2LZO56PACnmm5kd9tv5LXl5YyaWgWqQlejhuRw+xbJncbET04O4mfnzSceD2aOsmz90up9hZlGX9DVq1axWWXXWbbFhcXx5df7p1bTKHobZbvrOPu/61j0bZanrz80B4ds7asgYuf/JIrjyrkzrNG75N2zFlXjkAwRbfErNbpDsccsKZp1DW3kZ4oO+mkOPs8X00gxB/fWcOrS0qYdfMxjMxNiViIM5eURPYLBNu56701VDeF+MuHG1hfLgW3siE6+LKmri5yboDPNlWRm+Jnt75vZlIck4dnc9d7ayPnMVi+sw5N05i7voKuyE31k+z3MjhbCtnOmhaufm4Rk0vLwQMBLZ4RmQm2QVBBZgIZiT5mr62IDDgMhvVLIsUvBdPrdnFacR6nFecxJDuR8oYghxel8/nmat5fXUaoPcy0Cfm8uqSEEbnJXbYT4LapI7isfhAD0uIjbuC/TT8EY4r/xWuOYNmOOhJ8HhJ8nphu6NxUP7mpXQe7gV2MDVc0QApNlNa1cPupI2O+3h3udvm9SvXuv0FlZygx/oYUFxezfPny7ndUKL4lGNZneQwBMthV18Lxf5nHy9ceydiBaWyvlnVht1aZ9WGX76xjeL8kEnx718388N+LAXjrhqMZOzCNKt06HZAWT0ltM4/N28zl8yeTOPgI3hr9MLe9tpLPbptCTooft8vu9KsOhFi7W7pf0x8rpjJnJPX5f4lq83sry1hT1oCmERFViG0Zr9tRDngIazKj1KKtNZw7fgAvfiGXNGUl+Riak0RSnIemYDtDshPZbKmfe/kzX9nSP8bCGDBYlykt2FTNaR7ZtgB+BmUm0BwyxWNQRgKpCdKNe/vr9sRJYwemIYQgN9WPzxIcdfOJwwEoXTGHbf6L+fn8R4AMrjq6CK/HxckHOZZyLv8vfHIf3LgU9Hs9ODspMmgwyEoy252T7I8KnNpbjKCqCw/VI8HDpiWb4vdw0kGdT2F0SUh+PqJ9z7xC+4I+56be00jJ7xPq3ij2mhUz4J9H9mhXIw1hV2Xq/reqjNa2MDMWSeHZogtasp7SsKKxlfP+uYCnP93Ko3M3Mf1fC2kOtfPr11dGCshP/9dCHpu3udv2vKFnSTIs40MK0qhqCvH+6jISw42waTZz1lUQag+zaJvMDFUTMEXu0EHphNrDkbnQfqKO7MovogoGgCwqEOtnVtEQ5PbXVnL5M19Fti3fWhZ5/O6KMgKhDiYMSo+sSc1KjkMIEYn8PVGf485JjmPCoHQ+3Sjd2r8/8yAuPDSfnxw3hMOLMtj451Mja4pzdZfyueMG8MezTY9DopBu+sOLMhiancRBeSmcMSaP04vzGJOfxoh+ybb9DYblSAv35INyI+2x0n/LywD4K5aRHOdhZG4yd59bHLVmmd2roHYrBBuib9YBYsNdp3LveWPkkzbTUzLfeyP+lp4F+EVhlFBsbyXmF2E/0qfE2O/3U11drUQnBpqmUV1djd/fvQtH8Q2p3QZv3QC9MDreb7xxHVSsgfauLTEgsjSkrcMhxhVr4d2fQ7gjkt832S8tMKOkXYtuoS3eVktYg6U7anngg/V8ubWGj9aU89+vdnL2owuoagry5dYa7pu1LmYbrLmMjfzHhmVsrD1dWWIu/zMih5fuqNX3NT+7I4fIIC5Ng3vPK45sr29pswUzDc5OjHInG8zfWMmMRTuZv6Eysi3Y3ESCbqE9/okcVIwvSCcpTg5IshLta1mPHirnR0f3T7G5UUfmpnD/BWP51dSRvHLdkXjdLpL1c+SmyHP4vW4uP7KQX54yglNG92NEuhT8s0Zn4XG78HvdPHLxeB69ZDzxPjdCCC4/spDXrz+Kiw43k4UYA4XbTx3JL04ZEfU+RZ1cGtWoJXBacV7nqSBbau3/Qa69/vCOby5iK16Gxc92u5vP4zLbFzKD29LaymVb9oZWYxmaBh0HNoirT7mp8/PzKSkpobKysvudv4f4/X7y8zvPlqPYR2yeA8tehCNvhJyeLWX71hBsAk9cl7vUNBti7OhUX74MqjfCxOtZUyYtIkMgjWU3hts1sGwm40QHK0rMub1XFpuJML7+7G0muzbySXisLXuUgTX5RWVjkJ01zTy3cDtCSGvQiXFdU4zl88xEX2RNK2CLqK2tKuMm/3x+XT8FEIzMTY4s/3HOSUfdCyBehDi4fyqVTUG2VAXITPRRkJHA0JwklmyvJU13Ff/5nGLGFaQzaWgWU0Zkc/yofpF1rSAjiJ0Y18tMsn9WN0wZKh/82wUNQEfXg6vxBemML0jn5NH9SIv3drkvAHXS05EoWs1rxcImxkXy8ew/QNV6yBgCh17V+bFfvwEp+TDwsNivv3Gt/N/VOZyEAvbncd3PcUex8lU5YE3MgUCFvLceX/fH7SN6JMZCiKnA3wE38JSmafc6Xi8AngPS9H1u1zTtf3vaGK/XS1FR0Z4eplDsW1p111vrtz/xiqZpvLGslHOFC6GFIdQIibHTExoYlnFdcyiSbhAAIf9r7cFIMQMj+b7hpjas2Glb7mBaHBQGXoqcd8Gmaob3S2JDeROTv7iGyT4obH2J7TXNfPD1btaWNfDQ9EMQQkQs734pcVQ0Bvn166si1veoPLk21YO9EtDU0bl8sGY3q0vrqQ6EuHJsIr8c18FWS/arggxT+H7W8BemuFcw8pwzaM4qZsmWcjTXl4wafyl/nb0xst/Vk4p4+rOtZCXF4XYButHtJ8imuhaKB6SytSrAFUcVIoTgsUvGM+vr3RGRTU3wcvUk2a89e9XhUfc7VsBSU1BaZcn+TrrooG7B98DTATBlRE73O2ka1MsB050nFxCfmdD5vrEs48whUozX/69rIX31Svn/zm5+Xx3tsPZtyBsLLXWQP6HzfUOOZaZBi4cj2ARb58PI0zo/vr4E3rsVBh4Bo8+FWbfLe7s3or6XdOumFkK4gUeBU4GDgIuEEAc5drsDeEXTtHHAD4B/7uuGKhQHDGMerHXP67XuK2Z8tSNmpG1HjKQRnfHWcrlk55ZXVtCBHl0cjL02PtQejgQy1QSkEDS0tjPi/2aZ13RLK6E50EiDnuN4d0Mr9S1tNLa24/O4qGoK0dJQE30BnemHFeD0fK4sqePe99fx1vJdPPbJZto7wpGk/mPz06hoCNJkcVt79cFBAqYQ+dwu7j6vmPQEHw/N3khNIMRN228kccZ5ZOgWakFGArSZS6KGueRc9Lj8ZI4emsVZDf/lMd/fmeaZz5gU8z4dPCCF/910DK/95EhGu83I63hCTBqaxS9OGcFvThvJT3VLMifFz+VHFvY4aYQ3RpapQFC6+w2XdxTGd7OHYhyTmi120QqYS7Pi6SZ1bywxNqzT9iCEO6A8RuKiPRngLnoSZl4F/xgPTx3f9b5Oy9jarvdvgxkXQfnXdMqqmRCslylOvfog5Jvc272gJ5bx4cAmTdO2AAghZgBnA9Y7rQFGKpVUwB5Pr1B8m+gDlrERBbvt3tMj26qbgky4azZ3nD6Ka44Z3O05fjbDjPIP6+Pu9tYGrnr6S248fhiHF2WwsqQOt0vw05eWsbUqwOxbJkfWzYIU/zW7GijKTiTJLbuL+tpqQJCTHMeuuhZKa6XAjRuYxpdba1i9ehmGA3JEv2TbPOypB+fy7IKtYFkm/N+vzLq0989aT2uoI2KNj8lP5cM15WyrDpDi9/DAtLGAdO2215prhodkJ5LhbeOEkTm8trSEsAYZrfK8eckefjV1JGcd0h/azE67P/rx+uBrkEcOIvLm3crbwK5flbGuvJEpI3KksG77jGdafhY5/s9nDMFz+GjiPO49Wj5j8Mp1R1JT3yjnJtuDEGee47apIwi2d8QMsgKgWR/wdDjiGkIB8CVG7x+Lh8fJsrI//kw/p3k/bSIdi67EuKMNPrkfPrkXrv/SPtVTs7Xr81oHi9b2dIdTjJsq5D31xMkYkO7O11wFHj9kDIadepBe+4GtJdCTAK4BwE7L8xJ9m5U7gUuFECXA/4Ab90nrFIrewBDhXhLjFssylWC7+bhEF7273ltLvZ6ecf3uRl5fWmKzmLdWBWhotQefGGJcUV3DpxuruOGlpQCc9cgCTn/4s4hVfON/lzFnXQWFFhflmY98xql/nx+xjBvrZac2Jj+NYHs4kn7QqK7z3LtmRqlXf3IkT19hrlXunxbP0AzTBuif6ueLLTV4XILVfziFwwrTeXjOJh7/ZDODMhMYoLt665rb+OGkosjSmBevPoKbjzUTORwW/AIeGMahOR04nQeiI8RPjhvCgLR4W6ftEvqOLdLKFI6C8v0bV3H8yH6mheuor54oQpEyh3vD4UUZTF16HTx3JtwzANbPirw2KDORp644jMRYlnF70PTeWK23LZ/AvYOgoSz6GCfGUiDre7KKldPta0XTLGJs8R5FxDgE2xfIxwGHd6dmS+fnXfyMvA8GXsdcelcBVc72LnoS7tJd80aMRFsXiWKaayE+w76/c6Czn9lX0dQXAf/WNC0fOA14QQgRdW4hxLVCiMVCiMUqSEtxQFj3P7gz1bQkeoLR0bXsWzd1WX0Lj87d1O1qAWtRAmM5DiBgIBkAACAASURBVGAT2C1VTeyqa+GUh+ZzyysrWLCpir9+uJ7bX1vJlAfncdETX9jO2a7/1Pu/cwl3eF6wCb6BS8jkHSADh967aVLktZ01LYRd0t377ByZPnHi4Axu9szkxFnShXjZkYO46uhCBgm5rERLyCTF7+WEUf1498ZJfPRzWUWoKN60OAozZMdnrMc1auP2T4vn/bQHOGr1nZF9iywpFQuzEvnBWDOQ64qDBLQFmOCz2g061k61LUbBCGPQ5XYEOK150/7caS06z9XWIr9ry1+ix1RthB162txVr/TsmIZS87E1gKt0CYTboLEHYuxs+xs/hn9b5lQ7mc6gtR7+kCavA51YxkHznrsc99Qqxs7fwReP2Z87f7Od/YZXzDDnoS94FrIdQZcefU6+q99zSw0k6N8nty7GfdAyLgUGWp7n69usXA28AqBp2kLAD0QlAtU07QlN0w7VNO3Q7OzsvWuxQrEnLHxU/t+9sufHtO6fOeOfvLiUBz5Yb0syEYudtbEr/jS0mPOmdc1tfGQpSbemrIGH52yKlO4zRPzqSUUUD0ilTTN/6td43qcp2M5nG805woxEH3+98BBG64n7i7ISbWkTAXY3SWsqjSYe9T7EWUnruNnzOhntFcR5pNv692eO5qKhsqMWYVPwDx6QGknGf8nBZsDSsDRpdY7UMzydXpzHPy8Zz1s/PpyE0gX02/xqZN8h2Ulyudl/L5LCY7Fyh6TIjn1Qu3SDplgDn6ydqtOdCaYYOztfw73p3M/AaWnpKSr55P7oa8RC0+zfscZO1sZ2tMFL06Fksf06YLeMDaHrygI0sN6HjnZ7FaqEzM4tY+u1QYpY5JxNZnsNK3bePTD3HksbLW7qJ46TAjzrN9GvgX3QAZ27mQ0rHKBoMiQ5XPuGpes8ftHT5mfVXAPx6fb9D/DSxp7MGS8ChgkhipAi/APgYsc+O4ATgH8LIUYhxViZvorex6t3/HsSjBHcB3PGLbUw716WDr+ZqlbByaNzI1mqGlvbaesI88T8LVx6xKBItiQDa97ldWWmZWxNUlHbHOKjNeUMzk6kNdRh229kbjIp8V4qG4Pccfoonv5sK+2zBTjiiS61VDIampPEOeMGcM64AeysaSYj0UeCz83xrqVkiEZmdkxmS1Uz/QUc717G4a71aF/cFTl+cKo74s7N9+kdfWsd/O82OP634DcLAgxNMEUv2yc7vAK9pq3bJTitOA9KlkT2mXHtRHZUN8uBQsVaGa1btgJO/6v5ZvSO1lu1jmevulguHfq7/pr1s48pxrogOi2nJoeLtTsxNgKg4rqYP/7iMcg7BAYdKcUrbIkI78yirdsBG2bBrmVw6v3w2d/kdpfXIcZb7e1qroG5f4YBh8qo5LE/gIPOlq9Zxbbe4U1IGxR7zvjrN6B0qX3biv/K++JLNIW5PWh6GbZ+Iv+m/Fo+tw4+ypbLP38aTL3btLYj7XIIf7O9ilSExt3mY18iWAPnwmEQ+lSCU4y/fgMadsHk22Tbc0bJ7YYYv3szXPcpuA/MCuBur6JpWrsQ4qfAB8hlS89omva1EOKPwGJN094GbgWeFEL8HBnMdaWmMnd8v6nZCuWrYdSZvdsOw0W1Jy6nfWEZL/g7fPk4b37WwlYtj5NvOgNPsJYL3EuoDhzK55ureeCD9aTGe7l04iDboTtrmonzuJgwKJ3PNlXx1KdbuHpSkc1NXRMIsXRHLdMm5LOztoXlO822njiqH7eePJy2DlkoflBmohlN3QlGEn+2L2SgEJAxEYCr3e+TLyqZF38SnlAQBBzukvVzRfbISITqcUUWK9rakX/1L2lpHfcrc5tlHjHbK8U4P8OxjGbH55GHEwdnMnGwvhzL6JAbSuGrJyzn1LeXf82Ucx3LeHrqpra6XONSpEh9cr8cSBx+LQQc9oXzXE27zWM7Y9bt8v+d9dHi31Aqo5Bdjs/K+O42lcOrlmpvGUX29xaxjPV2LX0OFj0l/wA2zYbModBvtH1Q8rI9tz4JmbGFz3AFOylbCQ0W4exok9a2lVAAlv0nttUeajJ/c1bqY1jGzTVyMDbuUnO7VYw9cY557CbL4MQhxs015rbmGnPO2HBTV6yR6+oNkd7P9GjOWNO0/2maNlzTtCGapv1Z3/Y7XYjRNG2NpmlHa5o2VtO0QzRN+3B/NlrxLeCJybIIe29jiHFPXHcGwW8WwPXeyjKqW6RLN50mXvDdC49P4iHX33nQ+y9aKrbw2UbZsRtrZzeWN0YKDpTWtXBUSiWHZbRSVt8aKTRQ39KG2yVwCVk8vjnUwdCcJEblJbNNr2P7w6OLuGHKUIQQkYo7o/KScdN5asvcFD/TJugzUc9OhWdOibw2QFTRT9QyOi8FPw63XeWGyMNfHmcJvHFaVc4lPhZROz95FX85bQDnZ1sWYDTXmNMLIFMUbtaDwqwd7+aPLcfonWrl+mghMKzHUAA2fiQfn/O4tADB7LytYpxbLMVv7p/l0piqjdGWq/M7ZbTNuTZ1yzzZJqfb0znYa2+1vz+DzuZv45It760ZGnfZ22X5fDjhd9IKX/26vr9FjMvtgWnEJclrVq6PJAGJyaFXw6BJcOU79u0dwWgr983r4f1fwrZPTbEzCLfD3Lujz2+8n1Q9g1hzNbx2jcyOV21Jo2q9Z0I45rGbTC+AU4xbauRvvKNdHmPMGacXWs7dg/n3fUSfSoep+A4RmYc7sGv1otDd1B0tddHpHWOhaaaY7EUAV6g9zM9fXs7nO6U1kyTMDnu4S7oDGwKtfLZJdgwfryvnreWlnPS3+Zzx8KcAVDS08mzzjVy/7vLIsRvKm2hoaSMt3ktago9leqapoqwkirKkVZtKE0cVxEWS6BsMSIsnXtiF4N0bzeCsP51zMGMHJEs3qJVwmIGeGvyijcOTKknCIT6VayMPXW0WwQg2RBKEANEZvyydomfunzh/zmTcz54sLaFgE7zzM2npjtetwHdughfOkUtOYokVmJZcR1AGRLVZPCHGd/Cl6bD4afm4cBLcvBIGTJDf1cZye8eba6bNBGDnF9HXtlrGrQ3SEwT2QLBtC+D5s2H+A3aRgNjfr1i5nkOWwc0AS+ILd5xse6BK5om2tqu+FFbKPNP4U2HCVZCQZQ6EDDG+yozgjuBLlO149HD412S5zenovGERnPFXuOo9SM4zt3sTpLXujHw2guHC7ZAco1jEl49BQYzc6SNOgxv1KYvmGjkoAnmNpgr5fXF6LMabvxtCAfO9WgPANE1/rkHddtA6TMs4KVsWwIDO5/H3A0qMFfsX66g+fOBrhBqW8VMfLePY+2MXcbcRagJNb2dXlrHT+tLZUN5IqCMcSZyRJ8wOIFkXs9KKKtaWNeAScn7YWA+8S8865W6Q1khcm3n9DbulZZwS7yUtwRup/lOUnRjJ9PSU70HGrrYlxwNACBFl1RqlBkEPdlr9ugyosRKoxK1bODesvYShLkf6AOt8p9XSCjZBisVSbnNMEXTmpfjbQXJpy9q34bjbYcgUuX2VHsTVlRgHqsFYmvTcGTKbkkFHUH73tn1qbjPW4vpT5ef8l+GmWPqSIceS1yghU6ZIjXJTW97H69fKFKpgFhsAc96zZJFdjMMd0eIM0e7acIfd03C0vs45c6gc5JQuhqdPskcpBxvkZ6l1wNT74Nb10upLyrGIsf679KfA4Cn2a/qSpVcAzHlg5yDBupbZugQpPkN6AMKxfx9oYUjppLbwwCOit/lTZUrKuFQ5iDPaHWyEx46Cj/4POTNq4Zhb4cIXzP2M72bNFrMPams2I9ENgU8wo/MjAwZlGSu+Mxij+q+ehD+mx54b6ilLnpNLRwIOd9OTx8PfDo59jB7R6wo2UFbfypQH50XyF8fEaJ8nXnaWsUIfKjfAnzJlQnsHX+trboPN8n0XCfPH7BdS2BZtlB30Qf2j5xY1TaMwoEd+J/dnxe9OZmhOEut2y6xXKfFeMvRarnEeF3kpfrl+Fugvqklq2h7d3o42PJiRzZonnkyrGMd7ocbi9jOiSp0BNFZ8jiAlY9BleBas1lJU4FMP5u9Hni5F0MqOhbJzzBgMP15gf6252j63ZywXAmkZly52tN8Q4zT7/Ogxt8L1n9sFo+BIGeyjOQaTtVulBdhUARstM3ObP5bf09YGqNfdvG0tdvFt3G13U6fq0wRW0Qt3wCOHmXO1l8yUAVi3boBrPjY9DrXb7W7b9e/LefmjboJDf2iKZWIMy9iXCBfNgJv0BDHeBHsAmjGoCjjmkDtLLBKfHtsytuJPk9czvmcG1u+MdV+QKVzrS0xvRMMu+V62zo8+RgiI148LNZnvtXEXvPlj+djqsq7WxTjeIsa+RDkA6Gzwtx9QYqzYPxjrC41Oeunz8v/6blKWf3K/udTBieFirHMITumS6IhQA91FmYL8EW+tCvCXD9d3fn1dgLS8sbJjbHK4qV76AW0z9ICXN661W0HA6lLZmQo909MoET3nFhduITXeyx/OGs2F7rm87zODm2qb2xir6e3LGExqgpcR/ZJZt7uB9kANz1VfwqEuORdYlJWIyyUilYeSaMHXGiPwxmGJinAbfo/500/2e8xBSG5xZADT6T0FaZlZMQZd7fp8oVXMouZGWyCxm6WNKQPsYpx3COz8UnaOyf0h1zH46gjaO/fMIZbXQrDLUXNcT2CCP9U+LxrugLQCKVwgPSsjLOtvjUHIyDOgehN8+Th8/aa0Qp0EKs0I59qtdjGuL7G7qY15SuvAZelz9kGS4aJO7ifFxngPWodcuhefISOHdyyUInb8/9kLHSTmmBHiETFOllM5GUVw7Tz46SJ7UJhhiTojyzsV4zTZnq68Sm6vvJ5zsJUcI9uYEYVfdKyMGzCC2Yz7Ur1J/r/gGbjesrbe+JyC+pyx8RlW6lXCrC7rWJax0R5lGSu+9RijduPHbHTOq2Z2fdy6d2HlDDkSn/lD+4/acDFF55Oxs+ZtmP+gfKy7olKF6Ub1ul3SqrUGCYG06t6/jRZvGjds1hP6G/OABhvex1ttlv3r2LmIa55bxAdfyxH06l31uF2CRD23byTLk4V4ghw3IpsJgzK43/sko1w7idPdyCtK6sgT+qhddxGPK0ijpLYFb8Uq0sK1nF0to2NP1guoy9SRGom04mqOsaLQ6RYOt9s63JR4r7Sk0gth2Mmyo7YUDbAx5ATpAvQ6op/fv11G6xouVaubOpZl3JUY+1Ol+9TaWQ87SYpb467Yc44A3kS4VA9Qsg6i2luh4mvTygIzqCwhw7R4c4ulNQnmtZNyYNQZ5nHG9uIL5ABh/SyZrKPfwTDtOcgcZnmfzab7uLHM7kqu22EfpGToBXKslnHZCvv7cwqgdS6+dIkcgBifS25xdMWhpBzTwjV+l9Zz9h8HqfkycvyUe2Dyr/QApza7i94dF50gxRh8G4LWVTUp60DISnxG9L6GhVs8zZbKlGpHJq/BU+yeESOIzrCMs4bD2ItldrLXr7PHRxiC7rx+cm70YHw/osRY8c1p2AULHra7dI0fq/GjN5YpVG0gJvWlMOcuqNkmXUjv3gyrX4OVlqxERqdpdReGY1gkr1wGc/4kH0csY/OH7BZCWupfPQmf/wP0Gq4010DZcv7WfCqfh2Vhdq18jbRo5vw55lzntqUfM3ttBb96bSUdYY21ZQ0cNSSTBNG5K/ZM90JuH6FnqdIX/2YiO+Gl22tJE0byBCmYk4ZJKy3QLn+uyVojR7lWc3mGOVCIow2PCCOCDdFBc+16u4efCjl6wXnLHG+SzyMtn8Rs2ZlrHfLaZSujEygMPBwOOis6KCtQAS+eb4qJVTCdgUrtLXYxH3G6/XXDZWtzG+qWTkOZaQFf9DIc8RNzH288DD0BCo+xB960h2TRgn6jicIq+Mf9xqxolZIPE6+Hi1+VojH1Pjj/aXNe2h0Hg46G7Z/J+eDiC2D0OTIwzKC1QQqw0d4yi3W+eY78XsVnyAHAuMvMYwysrmHhir7n1qjk2m3SfW+4pOPTiCIxSwqaEdQk3LHLaabmw5HXmwOmllp7WstYVrHxeTpdz7EwxNi5BMwpzgCDjpL/C46SHhED68DGnxZt1Rrfl5Y6OTDwJcl9AhVysP/uzea+nVrGecoyVnzL+PpNGUhhHUUaPzjDTW1YWbEy+4TD8OgRMuJUX1YU2qKvM7XMTTa0SJFpDFjO4ZzLsqJpEbdWijDdydWBkPyR1W6FD+8gvGIGJbXNtFbIH+VGLZ86kqkgnTnz56J9/gjMvx++MIuRlWhZrA0XULpyDgBJIkjF5/9hUPs2Li9qYHRW5+t6z3B/Sd5b0wEQ+vvLEvJ9f7m1hlRj4KAvhRmhZ65K0gU+Ly7IS767yXr3h5Fznj/a0pF1Fmg05kLZyYLtc3C5hLyPiTlmRxsKwI4v5HzpoeZ1Ip290Yk7O1DDMk4fJI91eWNbxt54KL4QzngIpr9g93ak6jW7PT5Zzu78p812dQTNxyOmwiRLp+rTBSEuxVzzC7plvFYGZZ1ytz3a1irG1vficsHUe8wiBxN/LAXXGGS6fVAw0dy/+EK9DRahaiyTv4lBR8vnFWulAB5yCax9R7rc49Ph5D9B/mHyNatlbHUNO+erIVpI0wstYhxDFBP19dcf/xG+fl3+trqqLGWIU3O1/XfmjBcA87pdrbE2MO6hvwsxTh0IJ/9Zlk8E+XkUn2++bnXfx/KUGPPexvfAlxjtFjcIVAAi+rucXiTf6wFKmaHEWPHNMawsa9SnIcahJrndcMnFWjNZtcG+fAPwBXVXrcUdVhuQYlxZY+ncLSPXndUBLn/mK/O11vqIlZhqsYx319vXc36+/GvOuO9t6ld/AMBxE6WLel1HPlkt23hnngwW0laaqRkrtDQ+CY/haNdqrhy4m/8LPUTe7Bv4IO52Tpp/AcPSBBvCFlftxeaxNvRO46CUIP1S4vhqa43FMg7CruUI4NGLx3PNEdJScQct71+/53efXmhuc87vGWLsjbeJ7UBLbV8CFXJJh/F61QYZfFRwJJzxNzN5i8chxlGBXPrnGJ8OP5wlBSzWnLHHD+c/Keveutx28TDEGGDav+U5rBG7nUXyevXtznW+9Tvl9yt7BBx5A5z1D/M1awcdy5p0YljGHp+0ghMy5f1J1T9r6/0wXKGGdVexVl5j5BmyPVs/Na8phBQnm2VcSVTatFhtMUgd2I0Y65bul49Ly9KZXMSJcW+aq+1TFrEs4yl6nEesICwnnVnG1vs/7d9w1E/tr4+7zGyTdeDvnDIB87tgeEg6E+OUfPPazvsx5ddw/cKuByz7ECXGim+OIaTWUb3VMjaicrNH6knkHZGWXWW6srmG5Qi1tdUSNGUR1afmreGzDZYfaXM1mm4Z54i6SI3WQGOtbf6pqaqEF3z30G+JTK+YVyitoW1aLkViNxP0gClhWVdbrqXzSPs5dCT249a4tznF7YjWDQWo8/ZjRzgbzZcEw0+G38VIdK933scPFEwamo0gTJoxv129SSZP+eIxTh+RxKSBRgITy/uvWBu5XgSnt6BBX5Lkjbe4e0t5/2fH8tVvT5Cu/uZq000NsF33TOTrxRCN7YYIG8lUjP8GhhgbgmgsHbLS1mqmKTUwOkqXN7Y72drh2h4nEhGsiHXmFGP9++d0uYPdNRnLTerE6LDdcfLYX262ew6sQmXM+RZMlG3UOqRIGoFKoUb7PHZciv03FKjsOvuTMwNYaj6RexFLjDP0sptZI+T/lhjfRytWMS5fYz6PJcYTrpBZxfbETe2831ZxjuU+zx4Bt22J/hxjibHLJb/rTV2IcVyKGTgXa776AKPEWPHN6dIybjQTvmePiN4PojrrYKqlVq+lw3HprrpAwDw+bBHj9xZvIhPzXFpzNTsramnQEogTbdw1cgfTJuSTjX1pU46opdi1LfJ8QJbsJLZr/UgRzQwQ1ayl0HZMYuYA7rnoaLzDppBcMo8oQgHGDhnAhgtmI27Vo6NdbqIsHb1jO3mQi2OHZ5FEa3S2rA9+Dffkx04SYRRMt3ocrPN79aXw8iXyscdiGb94PknlS8hJ9svOVgvb3dRGhLEhHEaHZwR+GZ2l1TJ1eSxirHes/jQpMNa5/fYW08I2SMiUwvrLjWayDytWAbA+dntMN6WvM8tYF+NYQWOduak7wwhUMqwlp9XkswiDEcGdNdy8dnyGXbCs1qDVMg41S6+SswKRlSgxHmhGG8cSxayh8ItNcFU3KxoMIlZohRz0DdTd8l3VS3YGdnW1j9MytlqmzkGeFefnW3h07P18SWbfE5ccLcbx6aYXxjlf3AsoMVZ8c2KJsUGwyRJhq3/xnfPGDjGuH3Cs5dzmvkIXqZaA2Qk1VpruszgtyGC/2YYPF68h2NrCZ+GDKdUyOcf3FdMOHUg/YRfjPFcdjZopDtnJUmi2auZcVGORPcjoyHFjOHNsf0QsKw4gFCAuIYUTxxQ6Cgd0Mv8UqOSssf2ZeYXe+XpjdHjr3oveFrGMrWJsmTO2Ro36Eu0d6atXwMPj4aEx8nmSRYwNATM6dUOMDQGIZRmH283lO1bLGOyfcSzLODFbXj8+Pbb7tDM3NZiWUreWsSNnNdg76FhzoU4MIeksqYV17jvUKAc4ccmmGPc/xCHGlsdxqXLgEqiGu3V3b1diHHKK8QAzeK8zCzUp21y21R3GvfnfL+QAqkBPytHVfYpl0ToxBupdWdFdncf4fAuPgWvmyKjvmPslyfzSIJfiOcU4IcMU454MxPYzSoy/q9SXwuOTpIvyg9/C54/su3Nv+BBe+oEZ2GDNimNgWFDWRfcR95yjco7RUV/7CVz8CvVpFtectcPRLeOWFnm91aX1vLnAXP4xtp+PkwaaYvfR4jXEEaIFH1+GR+EuW874gjSK4uyDhjyqSBYtfNBxKKeG7o0k1diu6e31JdM++PjI/iUTfoX30CvlE2umJutIP9TUtQUB8v4Zbvgv/omY/wAj3tTXQ8YSjp1fRG8zRv7We99UKSPRX7nCdJWeer9sq7UjbSyTgTDtLTJQathJ5iCgoVR2moYIOy2HWJYxmBa1cR3jfVjnsWNZxsffAec46tlasQ5OnG5JY57S2G4VY5fXnO+MZRlb70dP5ga767SdkeyGa9gYIBVMlKJreEj8Tsu43hQQML1JsWhz/I688V1bxlZ+NEf+3rrCEyeD5wyMDFldWsa+zl9z7jPmQrkcLOa142NvB3MA2O9gyJ/Q+dx3pApTvHRHR1nGFjHu7bS99KyEouLbyNLnYPcqWPJvWKgLsTMgYm95aZr8HwrI0Wcsy9iYFw5axDgp19xmxRDjnFHQ/xBqq+aYr+mWWGtbhxRjAcEWKWAfr61gsDCv+YvjB1K21qz0kk4jftHOiQcXUObqD6s/w1O9nru1h2O+rZkdx1KZMFRftws7tRw0BGLg4aTlFplNGnupuQSmnyX5RHKeOecX7IEYv32jvWbu3D9bzpVrzzfcGRVr4Z2bzSQXbp90U6+fJfMBe/yyIzriOvl6rPm1xGw47ynp8jXcrPWlskM3BOrIG+TneNiP5HOrZXzxK3IJ2uqZuvAJU6QNF3LTbjMyOZZl3JXogN3963OKsT5oMqxVQ4xdHtm+UKOc43VazLDnwTlnPSKTzwycGPt1Z6eepa87NjI+DZwo5zPj0/SlTY4549YGe3KQrgKiYqUV7c4yNrDmuO6K4gvkvkufk6UY3XH7QIx170J8mlwOFiu2sSvL2FjWVNDJZ2BQcKSMWjeCs5wBegkZkKYvo4tVzesAo8T4u0qkQPYelA7cU0JNDjG2BJ8Yi/5XzjCzE1kDVyyEW+oQHj9Cb3NVwlDe6Diac90L5LnXvE2tZnYubcEAtIcYveNFcoU5Zzw4VZCWKq8b1gQZopE4QqQmJZE6bCKsRq4tBnYmjSG3ZTOecAtCt7i3WdzSI/ols768ESb+BIqOpV+uucYxNc0ywk7KgQlXSsvy079ClT4/HG7rXoyX6flzEUS5r51BKqkFZmpFKzWb5Z9RgShtkLTCjDna9lb7gCEpB8ZMN4sIAEz5rVmz1WhzsB5SrAUA4uHE35vPjTWuXj8MP0V+31bPlBa1x2+KnCEmxtx+OCy/G11ZPrGwuakdblLnNSIu8jRAg5D+vvdFVGxStsyb3RkTrpCpNzd+pN97fRrjkldlmkoj6jo+XYqu1TI2thlz/vmHyeQdk26R67udnPWIXA7Y/xDznvTUMt4TMorgxDvl4yOug6LJne/blRj7kvWBkWOfU+6JrgHd1ZyxEZTVEzEG00hwuWHiDfK731wlLWNjuZfT5d8LKDH+rhKp47sf3S/BJkjGFOOSJTL9n7PO6qpXpZViRCxaLGNN05jx6WpO9cVz2T8+5R8Xjae5w8Uv2m5gnNhE4eY5sPQ5Mr0p1Og1eduCLTDrV5y44xlwQTgxB1egAkLNZIhmGrV4mognnUZ8ok0KheFO1nMWD7z2Ffjod9IqXPUqIW8y23f3gxZpXc24diK7G1oReXKuOMOy1jA1yWKZCQFn6lXsl79kvz89mYMEaXUWTLSXnLSKsTcBRpwqawNbI26Nzg3MFKHphXJqoiMEQ0+UHdHwqea5XG447wmZz3jnFzD1Xrm8KHItywCiJ3N6hqgm6POQDbvsVo3xPhrLZIS2kdDDaRl3R1du6rE/kElnxuhrfa3z1Yb12FXGr+Jp0cuE9pb4dJj+osxNDaYYDz1B/ln3A7u1lpQtP08jCc1V70sr0joIspIzEi542r7NGPjuSzG2cvKfun7d+tmfcjd8oC95cvvka6HG6CAvY+27la4CwU59AJb/p/NMbAa5Y+Tv3jqnPPVuORhc9JS0jDOHyu9GZ/f4AKLE+LtC3Q4ZIOXSwwCMH4U1cKaj3bSA9gWGKBhzxhvel3931kfXbvUlmqNfS7BRfUsbKaKFmnY/q0sbuOKZr7h6knQJtxAXsQZ9bQ2kCvme3MF6wotnNL06LQAAIABJREFURAIeXGkF0ppoa4aWWtwJ6XiI52BvGH9DmxyYpPSX4mXMx/lT5DpXgKl34wMeWlUWKaCQnuizVTYSFqvK7erEwsoba7c4Y7mEY+FNkBmG4lJkOyvX2Tua27aYFYE8cWCMrzKKZE7iSMPi5PFlK6SFNPREODW6ihNgnt+6phfsLuAuxVgXU0NUjX0bd9vbHpck73vdTph9p+X4b2IZO+5reiH81lJRymVxgxri1JUYn//UnrVlT8jpJMDPuF9Wy9hoY8Ua+XpPIpOdHPETWY6wp9+9fY21zUfeAP3HyxrZHr/ZJ3VlPU+4Uk6tdeXFOOJa+ddtWzxynbATI7YjPkN+l365qftzHQBUANd3gaqN8FAxLHjI3GZYctbAmc7WFWpa7LSSsbC6cwxRdQZkaZq0zI6+mcZMPYOOT++UgY07yyJ5psvqW0khQAPS8tlR08x9s2Tu5wB26yleV6GD21bhsi7/MQRFF+OE1Cyy01M5OFMg0GQnIIQZhCTcMTur04rzOGJwJ1l6gDqtG7fzxOvhtAfN584MQ53hjZdz0Ldtkcnuf1djt6o9flPgrNmAbMUahBS+xGw5MAk2xE68b2C4dp1i7PGbEcF7YhlHBFKLdjEm58ryh1b22DK2zhl343EwooVHnGa2JVZA3P5kwKF6Wzr5PkUsY8s9Nlym5WvMx3vK1Hvgd7UHLFFFFE6hNfJjeyz5rLsS4zMegt/veR3xPcL4XfaB5UxWlBh/FzCiRbdY6vUac0fWNafOzEwGr14J9wzs2bUazACpyFyMU4wbd0O4Dc0Tx1cV8itWFfLw3xVyMLB12TxZTnHTx5TVt5AimmnQEvC5XRw1JJPmkBwYNGuyw5/fYS/0PsolreUSTe90rUEYLbWyg/PGm14Bo0M2fnz+lL3qrDb8YAEfn/F55zsIYReNWAFDsaJxjWPcXnkOl9ue5F8IUzytgSZG0FZSrswGFZ9uF52ugn/SBgJCzkU734M+aNojy9j6vmOJcfkqx/F7aBm7XOYx3Vl9WcNkib5jbjWnRLoLENvXXP6WLHXYGbHc1IZl3FDSfVWrzhDC9I71Bk6hNWILPH7zcVcWvxD7fyARsYz3kyt/L1Fi/F3AmO+yWrdGzdgmy5pTZ85igzVvymUSlZ2XFgy2d7DpvmMoe91S3jDYJK/pjETUE+K3ht00IDvO0mY3v35nI+2ai0M06SpuX/I8q0sbSKaZRhJITfCSn2520q3IH/YardC2DhhgfTif3ZouroaghJrNCFVPXLQYGz++nuTPjcHho4o44dBO3I4GVosv1nVu+ArOfcJxTAxhcnZqhmVsvddG0FbWMLks6IJn7J14V3Nq4y+HK9+NbbkZwUJd3SenZez2mRa10+qNNSjYU8sYzPvUExdsRpHs1Bv0NcbWJWgHgrikrj0TRvyE3zFnHOvxt4koyzjO/G+81pOI6/2JMQDqLFd1L6HE+LuELcuRIcaWhPmdiXG6vmxn9Wv27cEmaTXXl/Kvt+cztGUlebssRdRDjbGXBJQuBaAu6KJBkx1ns+YHBAH8hDT5tdu6ZSN//WhDxDJOi/eSlmD+ULMS5H4VWho7NLvbbrdvEIW5umWcYrEaDcvY4zfF2JlkoKfu473B00XUL0iBNHIVG8QSF2eHFctt2e8gmRf5gmelpZs31i7GSV2IcVyyvcKQleGnyP/WAuxOIuuMdVG1egWclrGRLMJ2/B5axiDjDrwJe2f5dZacpbcYfxmc87g9itj62e2tZdzbRFWWMtzU8aa3p7fFeNgpsgpX7pjebYcDJcbfBQwr2JoVKNaSps7c1IY7tWItVG6AeffKucnVM+HrN+CT+2jdvCD6uGBjlIt6tzefmsWyZnFtUKNenwtu0a3cZhFPqibd23Etsj0pNNNAImkJXtISpAsr2e9hfI6Mni7X0tnttltXBYVDyErXR7haWF9PGpDz4vHp3VjG+zHbjs0yjuGmhuhlHLEsY2en5vHJxBhXzza3+ZJkXmSrFdV/HIw+T9ZuNdzYe8q4y2RN21hRrs72WUU1UtHJIcYHnWvZRxfsvbKME/Y+MClWXureJDUfDrnIvs36PehjQtFjnC7omJbxXgSm7UvikmQVrt5058dARVN/FzDq1WpWN3WMhABVnbihdetWa9iFeFQvDDDuUlPMWmq4JvBGZPdw1ghc1RuprqkhXF1DNsCQ43ltQ4itbXn8witX8Ve3Chr0oKfUeC+0Ab4kkkPS4ipwVXKSazFxoo0GLYHUeB/pumUcbAtHorUbPJk0xueDxQgvKhwCw2+Q1v7g42QnHaiUA5L4dFmtxbgfRocQEeNORHJfYBWnzq7jc2zviZsa4NhfOo6LIUxxSTDt2a7b2B1eP5z2QNf7OOeMoXMxTsyEo38mrb0vn5AR8ntjGXvjoyOpu2PqfTKxSm8FNO0tRpWsbxvuzixjvyl+vW0Z91GUGH+baamTFXoM4bVZxo71xQVHyfq0xjFZQ6X1W7o0EiEtSi2VhxrLTTFe+w7WuMOmxEGkNO7m7a/WM2vpp7wMcNg13Po1jBA7+IWeUqeqJUzQI4UnRf/9+eKTZBIGnSd9slJSAwmkJXhJ1y3jUEc4kjR/3OgRtKeMhi9nmgcm58p1ltfolqI3wQwuMyxjA6dlLPbjiLgnlrHbI9tg1KiN1Tn1pMPqLqnI/iSmZdyF1XvSH+X/1a9JMdZ6GL1vxZe45+vmJ/54z6/Tm4w6U77HnpRz7It0NWcslBh3Rd+y0xWxaamLLAWy8cl98NwZFjEOyx9ysMm0lg2GnShr1P7rGHhkghTiz/4GTx0Pjbuiz12z2awFClRqKbwwVApnuXcAHb4kkmjBG5Kp+zq8UnjKNFO2dzVp+BKlAGYluDmiKIOExNgCFWvO2Cgmfsu5x3LBqSfJ9cuGVekMTvLEmRmYjDljA0O0DDHeGyHoKVZx6qpe7O9rpUUP9gQpkfP0IOF+b4pxUq4MHEyzRGNHLOMurN5D9ApSe+M2ThsE6YP2/LhvE9NflNm6vq0YeQzG6i54q2XcV9zUfRQlxn2dQBXcNwg+fTD6tdptUoCMedtwO7x2NdwzIDpYy8ilayTyb62DjR/SKa//CJa/GHnaqCXQb8ghhDXBTvcgmvGTKFoZLmS0an2SDAIz1gsDfF3ewtACuY413e/i5euOlJaxzhPtZiUk0zK2iPEJv4fbd9pFx8g45YzQ9fhNMfan2cXMcG0ay4p6uqZ6b9iTuVAjXV+sYJ2urAfjuN60MFIHyHXRg440tzlrHsfi8B/Br7bLaOc95cyHZNF5Rd/m1yUyVSfoKz2E/F14lBh3hRLjvo6xhnjN29GvNZYBmhkxrXXAOr1W6db5ZsUYiC6bVl9ipt2DbqvRtBBH/4KhXKjdzfWrhrCxXpBECyPETjris3hymRHIZc7NhfBy0ljdcgrLwhHuOFNYN2hmwolGLYHUBF/ETQ1Iy7KzyGenZez1W9JEJtgtYyOVoiFenZW/2xfsyVzosb+UlXPyYgTrdCW0l7wKP13c+/Ogzu9MZOlRN/dgb12wnrjuz63ofeKSTQtZCPm52Sxj5aaOhRLjvo6RVtIT4wtsWIINupu5I2SWDQNpPd7wFVz+dnQnVl9irsGEyFrdoBY7jCCAn34pfmpSDqI17KZJiydJtDDStYMvAv14bN7mqGMGZqWSlaZ32MYyH9160txx/GG6Wbe42ZVIYWaC3U3dFc41sFbx9cTbLVTDsjb2cUYz70v2xDJ2uWWS/1gY1mWs+e24ZLMaUF8i4qbugYtd8f3BE9e3oqn7KEqM+zrOYu4G4Q6zeklDmfwfCkSCniLHZI+AwZOjLLaW0tX28+lZrNqJPc/ZosWRmegjO1l2tJ74ZEalawwXJawLm/OGl04sICTkj25wbrpconHC7+Dcx+UOuhgLr5/ENNM9+9KNU5k0NAufp5uv5LWfwNmPRluFVgHw+mPPGQ86WlYoOuMh9ht7EyUcC6PD2p/BZvuaiJtaWa8KC6fcI3NOK8u4S1Q0dV/HSDnptDYCVWY0bqTAfJPdBeuNsQZUp23Vm9i26DmK2/BgVCLQDrkUoc8bDx6Qg8sl6JciRc7lT8HfvBwhQuzUTFG97tghtK5OwtdeQ78MPe3kMbea1/FZOmxLOrr09MyIwP7w6CIOK+wkVV3/Q2Jbk07L2CbO+jVdLph8W+zz7iv21ajfWCIiuggC62tE3NR7sYZY8d1lnB6015NCEd9jvkXD7u8pxjyoc/1eY1n0446gvUqTTZDsYpxSu5qwZlqXWoopxs8nXU246DieSL+F1eFCAAb2k4KbmSR/SN6EFIRutV9z0tjIeTISffiTZUT1pBFmDWCzHZblL5GlRm5bkNbvzjyIU4u7yKsci64s4wPpNt1X87hduan7Kp1l4FIowFIoQrmpY6Es476O4Xa2zhnX7TTr10L00hh3nBRmq4C73HJEatm3liQykZb3lo5shgD/aj+Dea6z+ajdz6fvr+OteL+se693tOGwrBrkssy75ueYqRoTfG5EYjrUQoJoi34/VlemEcjjT/3mImZ1jTot494IdCq+8Jsdb+Qb/1aJcSdJPxQKsBSK+P/27jzOsrK+8/j3d7faq3qrXuiVphuw2USaBhQySEQxqJjgKBKDC7FHExKNZtFkhjgk+kpiomYSTMYkJs4kiopbx2AY10hUtEFB7Iammxbpbnqp3mqvuz7zxzn33nOrqulbXcu956nP+/WqV51z7qlbTx2o+vbvOc95HirjycToN32eqlTG4f/AhZz0kQulz9x26q8pPzZSHDdBQhhYPy4Fr99fvLzy0qd3jGjd2Cf1hZZXa8+RIT2w+6h+87oNuvjssEINu5dTyeB/GYuOcm7p0tlLgsrWzKTzw0eWxo/glqoVcLo1+Bdypuu0I7nrUgnf8ujNBt63fF9/da3kM1UOtE03Tb89c4UwxnPpXBpMRTu+lw+SqIybX7kyLt8LHr8mcbp94mINi9YHC9TnRvTHX96pR/ef1Gff9sLgj2W2X0+U1uituXerXx26NfUNSdJ39wWThFxxzmLd99ghmUm3XrFWdl9YyYajoX/jug0qOacLlhyQyrNrtnRr2x0XaygbtvFF75QuvLl2QohKe8dNDNG2cGYWbigHQKq1+jhFnKVbpXfvarqVZZ7TdOadhv9e8Ebp/FdUH3tCDSrjRjvxM+l9PdKhcL3X0RPSBzdIe/8j2M+G94DL0wBGV9Lp6K1OGBGd7zhchcnlhvT3//lTbX/6hAbH8iqFgTWmjA5rkcZUDazRcPv85UEw/tKlq7S8p3XCsnUL2jP6w1deoHR7bWXc1ZrWip7wXLPJgzjyPpU/2O2LapeRO1PjVxHyoTrrWh6v+2v1zMCF+SuVqa6whgkI40bb9ZXg88OfCD4feTyYPeu+3w72y5VxeRWmaBh3La8OgoouEReOjM6OVB9zeuxAv4rJcCR0ZuIfy1EXhNktW1brT37pIv3Za8KJKMp/YMdP0B+dd3kqCy9UnvkN3/dl75eu+x/1f/2pjA8CH8I4buqZgQvApAjjRivPX1yeL7k8z/TRJ4O5o8v3jMuTf0TDuHN5dZat6L84w+2jx09UDj26r1+FRPBHsqVt4pzGI2rRDRcs19KuVt2yZY2SiXDQU7q2m7oiuj+VMB4/S9O6q6XVl5/6/HpNqIwJhDlX7wxcACYgjButPFq2PF/yaDVA9bX3BTNlSaeujMvzFA8frR5feZkOLNisdwy/WX9288Vau7hdj+47qZwFQdXeHoRxOlkdZfzgH75Sf33rpRPbN66buqISwDa1BQvSkQFcM6lyz5hAaJiVl0nrXywtObfRLQFihzButPJjN26SMJakY3uCz5V7xpEBXF0rqhP1H3y0crjQslCvGf19tax/oV57+WpdsmqBHt1/UtlwZqyOcOWkrta09NL3S5JaW9srI6VrVCrjU4RxS9fUHh3KzNIsTVTGjdezSrrti/Fd/g9oIMK40aLLH0rBcomS9Js/qj2vXBlHK+Cu5dLSTZIkd+HN6j//dZKkn/Y7Hewf0y+9ILh3fMnqBTrYP6anB4Lv0bsweJSoqzUlvfCO4FGcUwVq5Z7xKbqpp9JFHX2/2a6MuWcMIEYI40YrL38YrYwznZUR0ZKkrrMmHU19ILEiuOf8+8/qG+t/V5c+8krt3bpbTxwekiRtWhGMeL5kVRC++4eCwO3pDo53tdbxiMHpuqmnHMbjBnDNlHL4UhkDiCHCeDYNH5P+/FzpwMOnPic3FH4OQ3n0RDBCOlqpbvj5mnvGxxZeohuzH9AHHg8XaM90aO+xMZWU0Fd2Deie7c8omTCt7w2C74KzemQmpVuCAGwJ1xTuaqnjsZnyhBzjJ+bIdEiyqYdxpiO4Tz7VrzudVOvkn5ddOLPfBwBmAWE8m048HaysdHjnqc8ph3B51PTo8eo9t7d/T68rvV8P/GykpjIutizQDrdOfcPBCOsjg2N6+ljwPh+8f5e+s+eYOltSak0HI7XbMkk99Acv0Y2XnRPshwO46qqMz3+F9Lp/lhaurT1uFlTwUw7jdunWz0qXvmFqX3c6lXvGYcWdTEtv+Lx025dm9vsAwCwgjGfTWHj/t7zy0mTKYTzWL333r6Qn/z3y7PAmfT93tnYcyamQH9NYviidfEb9maAiPjaUlXNOW97/df3L95+pedtl3bXdtIs7W5QIg6oaxnVUxpl26XmvnPy1lq4zq3A3viSY7GMmja+IpaBHYbIpOQGgyTAv2WwqV7vjw/joHumpb0hXbK12Ux98tDoiOpxNK18MBlxllVaqlNUXH9yh14ydVF96pSTpqb5hve2fJ3aB/9y5vbrrVRdMOF6+75vMtGtZd1EreqY5yOmKrdLiDdN7j5lSuWfMI00A4ocwnk3l2bOy1ZmwlB+V/vqyYPuSW6qVcdTBRyRJA6PBqkdZF1SwXYPBY07PJqtLE96/4/CEL7/tyrVat2SSZ38jg5w+/2sXa0HbNKdavPq3pvf1MylNGAOIL7qpZ9NklfFT36xuj56oDePysnub3yJJGhgLFl7Ihv9m6hzYLUn6xBPBf7ZFHZMvRbZhaeekx6NTRq5c0KaOFo/+LTb+0SYAiBHCeDaUSlIxX62My13RktS/r7o9erw2jFdvke48If1cMC91f7kyVhC6Sw8/IEl6MrdYG5d26tNbr5z0269e1D7pcXWfJSXS8VoJqF7jJ/0AgBjxqDRqIve+Sdr5JemKtwX70cq4JoxP1Ab16iukRPXfR5VuagXdyRtPPKB9pV5lldGC9nRNV/QHfvEi3XjxCrWmE9V5pcc79+XSO38sdXgYxpnO4JGpmVgBCgDmGGE8G3aGj9OUp66sCeP91e1yN/Ulrw/W+lxxsT7ytSd1zcYlunBlT7UydtV7u2/Lv1OSlEyY0pHpK2+94hRLFkYlEkF17KO2BdJb7ue5YgCxRBjPpr4ngs/jw3jpJunIzmoYty+W1l6lQrGkj3xtt+75wT4dGhjTm164TpKUCyvjx7VeO1wwM9fJkSCo/+VXrzh1JTzfrN7S6BYAwBnhnvF0Hd4pffuDk792JJzsIzqaun9/tXobPiblRyrzPJ8IA/bQQDDb1pceOSBJ+qtXLJck/aiwrvI2x8MJP160YYmuXO9htzMAzCOE8XT9zVXSN/64diBW+b5lKRgNreygdOCH0t7/kAYPSYvPCeZoLt8/bgnCuBywZSdG8sokE0pd+Grt7d6ijxRurrz2sguWz9qPBACYW3RTT8dYf3U7N1Jd17dtYXX2LSkI4797cXV/ycbgnJM/C/ZbgoUbjg1nJ3yL7ra0rHuFvr75f+vIfY9Lku7ZeqUuW7twRn8UAEDjUBlPx74fVLejo6Llas8rV8hS8BzsxpdJbQuVPfpTSdJYslPffOLIhMpYko4OBQG96azuyrGNSztrBm8BAOKNv+jTUV6LWKrtpo5ut4+bG/l5r5BaOlVqXaCWwaCb+j/3ZfXmf9quxw5UK+1rz+vVhqWdetkFwTzUF62qrprUM92ZswAATYVu6uko5avbNWE8Im26SXr+G4L7wv/2ruD4i94pXRNsP5tr1arw9P2jQbg+8ky1a3tFT5s+/sbLK/vdkUUdUlTFAOAV/qpPRzHS/Vzupi6VghHSS86Tzn1p7eM2666RWnv0pUcO6Fv7SpXDzwwHQfvo/moYL+1qUSJhSkQeW7pm4xJ117PsIQAgVvjLPh3FyD3ecmVcGJXkgqUHpeCZ4rKeoBb+1q4+rVO123n7wbyktMby1YBeOm4JREn6xJu3yHikGAC8U1dlbGY3mNkuM9tjZu+Z5PUPm9kj4ceTZnZysvfxzvhu6kIueMxJCh5dkqREsnpOT7D04WMH+rV46crK4d0D1XMuWdWjW69Yo+vOXzrh2yUSJiONAcA7p62MzSwp6W5J10vaL2m7mW1zzu0sn+Oc+63I+b8h6dJZaGvzGd9N/fQD0oMfDfYzkSUMX/236n/4s3r8QE4XrSzoqb4h9Vy8SgpnyxxTdfWlLWcv0h/cGKmmAQDeq6cy3iJpj3Nur3MuJ+keSTc9x/mvl/SpmWhc0xvfTR1dBCLTLuecPr39GY1ueq0u2X27bvnYg9rx7ICck1actSryRtVq963XrJ/9dgMAmko994xXSoqkjPZLumKyE81sraSzJX1j+k2LgfHd1NFnjdMd+s6eY/q9zz1W88jSV35yUAmTzj3nHCmytPFdN12gy9Yu1NJulgAEgPlmpgdw3SLpXudccbIXzWyrpK2StGZNHasMNbtyN3WqLQjj0ePV1zLtGh0JLsP+E9Xnkf/xO0/rsrUL1b14ReXYP735cl2zsZcFHwBgnqqnm/qApNWR/VXhscncoufoonbOfcw5t9k5t7m3t7f+Vjarcjd1a3dQFUeXR8x0qPw48KH+sZove/F5vVJrdTT1tectJYgBYB6rpzLeLmmjmZ2tIIRvkXTr+JPM7HxJCyV9b0Zb2MxKeSmRDlZd+uEnal9Ld2gkF1TG+46PVA5fs3GJbnvhOvGMEgCg7LRh7JwrmNkdku6XlJT0cefcDjO7S9JDzrlt4am3SLrHOedO9V7eKealZEaySToYEkkNZ4Nu7OEwlL/8G1frwpXVilg3fVTqPmsuWgoAaGJ13TN2zt0n6b5xx+4ct/++mWtWTJQKUjIlHdsd7CfS0vV3Sc98T1qwVkPZfTWnr+gZNzjr0l+eo4YCAJoZM3BNRzEXBHDZ7+wOlka86tckSUNj1eeQM6mEFnVkxr8DAADMTT1Bbrh2neLnUsxLybR0znXBflvtGsPDuWoYn9PbyexZAIBJURmP95XflU78THrTl099TqkUDMAqFYIw/uV7pdLEp7mGstUw3ri0czZaCwDwAGE83sl90uDBU79eKkofvUpatikYuJVIB/NPR+egDg1Hwnh9b8eE1wEAkOimnig7KBWyp3794X+Uju6Sdnyh2k09iQd29+mB3Ucr+71dE1dhAgBAIownyg1JhcgkHU/eL337z6v7h8P1MdId1W7qcQ72j+pX/uEHOj5cnbv68nWLZqvFAICYo5t6vOxQbWX8ydcGn3/ut8PXB4PP+WFp5HjtaGpJo7mi3vXpRyv7Lz6vVx953aXqaZ+8ggYAgMp4vOxgbWVcVipVXy/r31dTGQ9nC/rFj35HD/70mC5bG4ysHhgrEMQAgOdEZRzlXLjykgvC98jO6mvZfmngWSk7UD02cEBaFCx5+Jnt+/T0sWE9cWhQH3zNxVq3pEP/9W+/pwORRSIAAJgMYRyVG5YUzub5xJelz/xK9bU9X5c+d3uwvegc6fhTwXYiJeec/ujLOzUYjp6+dM3CyoCt0jyaHRQAcGYI46hoF/TxvbWv9T1R3V6ysRrGybQOnBytBLEkLe7IqKctrTtfsUkv3LB4FhsMAPABYRyVG6puj1/8IXofuWtFMJo6Pywl0nriYDXEEyb1tAX3iN9y9dmz2VoAgCcYwBUVvR88crT2teFj1e2WrurUl8m0dh2uhvHC9owSrE0MAJgCwjgqG6mMh8eF8eCz1e2Wbqm1O9hOpvX4wWqIsxgEAGCqCOOo3HOF8aHqdktXEMiSlEhr16FBldeAIIwBAFNFGEdFB3CVu6nbFkmJlDQQma+6pUtq7ZEkFRMp7T06rItWBvuLOwljAMDUEMZR0TAe7pN61ki/91Op66zgOeOylq5KN3V/ViqWnK7esEQSlTEAYOoI46jx3dQtXcF224Ka0z7145P67E+C+8THx4KZuaphzIIQAICpIYyjopVxfkRqCdcgLody6Fu7j6svH4Turr4xLWhPa/O6Rbrl8tV6yfOWzlVrAQCe4DnjqOhoaqkawpnOmsPHRooaSQXrEz97YkS//tINyqQS+pObL56LVgIAPENlHJWbPIxPlqpdz49d9F495M7TpRtWS5IWZ4r6lavWzlkTAQD+IYyjsgNSa+T+cBjGD+4Pl1RsW6TPpV+ptnRKW85bJUnavLJVrenkXLcUAOARwjgqOyR1LKnuh88SnygElbFLt+mHz5zQxat61NUZBPWargnvAgDAlBDGUdlBqT0axkHSnigGjysVlNLOZwf0grULpXR7cE5+krWPAQCYAsI4KjeuMm7tUankdLIQhPHwWE6FktML1iyU0m3BOfmRBjQUAOATRlNHZYcqM2tJknrP01CuoCEXBO9YLidJumRVj5QP7hlrOSOoAQDTQxhHZQdrH2NaeoEGxwoacq2SpIQrakF7Wr1dLZKdI239D2nppgY1FgDgC7qpy5yTcoO1E3x0LtXAaF7DCsI4pYLOXdolK68KcdbzpRTTXwIApocwLsuPSq5UnXVLkszCMA66qVMqacOyzlO8AQAAZ4Zu6rLyVJiZTuna35eWbJCkmm7qpIo6dylhDACYWYRxWXn2rZZuactbK4cHxqqVccZK+vnnLWtE6wAAHqObWpJicm+iAAAZfUlEQVT69wdLJkq13dSqrYzTVtTqRe1z3ToAgOeojEtF6cMXVPfHLQoRvWcsV5rDhgEA5gvC+PBPavfbFlY2H/7ZCf3FV5+UFC4UseKSuWsXAGDeIIyfeTD4/At/HgTx8osqL/3l13eHWybd/jVp8Tlz3z4AgPcI42e+J/Wsrhm0JUn7jo/ou3uO6sXn9eotV58tre5tUAMBAL4jjE88LfWeV3Po8MCYXn33d9SSSuh/vGKT1vfyOBMAYPYwmnrkWO1KTZLet22HhnMFff7XXkQQAwBm3fytjIuFYPDWyHGpfXHl8Fi+qK89fli3XbVO5y1nsWIAwOybn2E81i898W/SF98e7Lcvqry049l+5YtOl69bdIovBgBgZs2/MHZO+pM1tcfCynjnswO6+W++J0l6wdoFc90yAMA8Nf/uGedHJx4Lw/jeh/dLknra0lra1TqXrQIAzGPzMIxHJh4Lw3gom5ckfe7tV81liwAA89z866aeJIyLbYv0F//+hB4/OKhNK7q1YSkDtwAAc2f+hXFuYhjvPJnSR7/1pCTp6g1LJrwOAMBsml/d1P0HpH0PTjg8nOyubC/syMxliwAAmGeV8Tc/ID3yz9X9ntVS/z4dH62uxrSYMAYAzLH5FcYjR6vbt39NWrVZMtOxB39WObywnTAGAMyt+dVNPTZQ3c50SGaSpONDucrhRZ2EMQBgbs2vMM5Gw7i9snl8OFvZ7m6dX50FAIDGm19hHK2M09UwPjpcrYzzRTeXLQIAYJ7dM872V7cjYXx8KKeWVEIJMx5tAgDMuflRGe/4QrA6U3aweizdVtk8PpzTfzm3V4//0Q1a3sM0mACAueV/ZTx8VPrsm6RlF0mu+giTEklJknNOfUNZFoYAADSM/5VxuRo+/NikL+8+MqTjwzlduLJnDhsFAEDV/AnjU/jqzsOSpJc8b9lctAYAgAnmfRh/+8k+XbSyR8u6uVcMAGiMusLYzG4ws11mtsfM3nOKc15rZjvNbIeZfXJmmzkNpwnjPUeGtGlF93OeAwDAbDrtAC4zS0q6W9L1kvZL2m5m25xzOyPnbJT0Xkkvcs6dMLOls9XgKXuOMO4fyevYcE7rezvmsEEAANSqpzLeImmPc26vcy4n6R5JN407562S7nbOnZAk59yRmW3mNERn3RrnqaNDkqRzejvnqjUAAExQz6NNKyXti+zvl3TFuHPOlSQz+46kpKT3Oef+fUZaOF3lyvjGD0mPfkrqXCYtWi9J2ts3LElUxgCAhpqp54xTkjZKulbSKknfNrOLnHMnoyeZ2VZJWyVpzZo1M/StTyM7KFlC2vwW6fLba17adWhAqYRp9aL2U3wxAACzr55u6gOSVkf2V4XHovZL2uacyzvnfirpSQXhXMM59zHn3Gbn3Obe3t4zbfPUZAellq7KCk1lhWJJ//roQb1owxKlk/4PKgcANK96Umi7pI1mdraZZSTdImnbuHO+qKAqlpktUdBtvXcG23nmsoNSy8TR0g/sPqpDA2O69Yo5qtABADiF04axc64g6Q5J90t6XNJnnHM7zOwuM3tVeNr9ko6Z2U5J35T0O865Y7PV6CnJDgSV8TgP7j2mTDKha8+bowodAIBTqOuesXPuPkn3jTt2Z2TbSXpX+NFcckOThvGPnjmpTWd1qyWVbECjAACo8v9mafmecUShWNJjB/r1/NUsDgEAaLx5Gca7Dg9qNF/UpWsIYwBA483LMH5kX/DE1aWrFzaiRQAA1JgfYZwZF8bPnNSijoxWL2prUKMAAKjyO4xLxUkHcP1o30ldunqBbNyzxwAANILfYZwL5p6OhnG2UNRTfUO6YGVPgxoFAEAtv8O4PC91JIyPDuXknHRWD+sXAwCaw7wL4yMDY5Kkpd0tjWgRAAATzJMwrk6HeWQwK0la2kVlDABoDp6HcbiWcbQyroQxlTEAoDn4G8a5EenQT4LtSBj3DYwpYdLiTsIYANAcZmo94+bzxbdLO78YbLd0Vg4fGcxqcWeLkgkeawIANAd/K+N9P6huj+umposaANBM/A3jBZF1iiMzcD17cpQwBgA0FY/DeHV1Oxn0xveP5LXr8KAuXsUCEQCA5uFvGLdODNzvPnVUzknXbFzSgAYBADA5f8PYFScc+u5Tx9SRSeoS1jEGADQRf8O4mA8+b7i+cmjnwQFdsLJH6aS/PzYAIH78TaVSUepZI73hXkmSc05PHhrUecu6TvOFAADMLY/DuCAlkpXdZ/vHNJgt6LzlhDEAoLl4HMZ5KVGd02TXoWBqTMIYANBsPA7jQk0Y7+0bliRt6O081VcAANAQHodxsfJ8sSQdG84plTAtaE83sFEAAEzkcRjXVsYnhnNa2JGRGXNSAwCai79hXKy9Z3xsOKdF7ZkGNggAgMn5G8algpSodkmfGM5pUQdhDABoPh6HcbHm0abjI4QxAKA5eRzG+UnuGTN4CwDQfDwO4+oArmLJ6eRonnvGAICm5HcYJ4NKuH80L+ekhXRTAwCaUOr0p8RUMZgO8/4dhzSWD1Zw4p4xAKAZ+RvGYTf1f/u/D1cOLaSbGgDQhPzupk7UDti6ZBXrGAMAmo/HYVxU0aqPNj1vRbd6mAoTANCEPA7jvHKlYOrL9Us69He3XdbgBgEAMDmPw7igbCmojH/r+nO1amF7gxsEAMDkPA/joDLubqN7GgDQvPwN42JB2VLw4/UQxgCAJuZvGJcKGisGlTFhDABoZl6H8ShhDACIAa/DeKwY/Hhdrf7ObQIAiD8/w7hUlOQ0WpA6Mkmlk37+mAAAP/iZUqWCJGm4mKCLGgDQ9LwO45E8jzUBAJqfn2FczEuSRovcLwYAND8/w7gULJk4UjB1thDGAIDm5mkYh93UBVNnK93UAIDm5mkYB93UwwVRGQMAmp6nYRyOps5zzxgA0Pw8DePgnvFoMaEuKmMAQJPzNIyDyriopDqpjAEATc7PMA4fbSooyT1jAEDT8zOMw8q4oCT3jAEATc/TMA7uGReUUGcLjzYBAJqbp2EcdFNzzxgAEAeehnG1m5p7xgCAZud3GDvuGQMAmp+fYVwsP9qUoDIGADS9usLYzG4ws11mtsfM3jPJ628ysz4zeyT8+NWZb+oUhJWxJdNqSycb2hQAAE7ntGWjmSUl3S3pekn7JW03s23OuZ3jTv20c+6OWWjj1IVhfO6KBUokrMGNAQDgudVTGW+RtMc5t9c5l5N0j6SbZrdZ0zM4NCBJunDd8ga3BACA06snjFdK2hfZ3x8eG+9mM/uxmd1rZqtnpHVnaP+hPknShWdP1kwAAJrLTA3g+ldJ65xzF0v6qqRPTHaSmW01s4fM7KG+vr4Z+tYTFceCyrirZ+GsfQ8AAGZKPWF8QFK00l0VHqtwzh1zzmXD3b+XdNlkb+Sc+5hzbrNzbnNvb++ZtLculhuSJGXaOmftewAAMFPqCePtkjaa2dlmlpF0i6Rt0RPMbEVk91WSHp+5Jk6d5YY15FqVSfNYEwCg+Z02rZxzBTO7Q9L9kpKSPu6c22Fmd0l6yDm3TdJvmtmrJBUkHZf0plls82kl80MaUataUjzWBABofnWVjs65+yTdN+7YnZHt90p678w27cwl8kFl3J3yc04TAIBfvEyrZGFYw2pVJunljwcA8IyXaZUqjGhEbUz4AQCIBS/DOF0Y1qi1NboZAADUxc8wLo5oLNHe6GYAAFAXL8O4pTSiMSpjAEBMeBnGmeKIssmORjcDAIC6+BfGpaJa3JiydFMDAGLCvzAOp8LMpwhjAEA8+BfG2TCMk4QxACAe/Avj/KgkqZhiABcAIB78C+PCmCTJJVsb3BAAAOrjYRgHKzkSxgCAuPAwjINuaqUJYwBAPHgYxkE3taVbGtwQAADq42EYB93UlqIyBgDEg4dhHFTGiTSjqQEA8eBfGOfL3dRUxgCAePAujEthGCcyhDEAIB68C+NiOOlHMs0MXACAePAvjLNBZZyiMgYAxIR/YVyujAljAEBM+BfGuVGNubQy6WSjmwIAQF1SjW7ATCtkR+WUVmeLdz8aAMBT3iVWITeqgjLqavXuRwMAeMq7xCrmRlVwaXW3phvdFAAA6uLdPeNSblRZpamMAQCx4V8Y58c0poy6qIwBADHhXRi7whiVMQAgVrwLYytklVNGrTzaBACICQ/DeEzFRKbRzQAAoG7ehXGimFUp2dLoZgAAUDfvwjhZyqqUZCpMAEB8eBjGObkUlTEAID68C+O0y8oIYwBAjHgYxnlZmm5qAEB8+BXGzqnVZaV0R6NbAgBA3bwKY5cfUcKcEpn2RjcFAIC6eRXG2dHhYIMwBgDEiFdhPDo0IElKtNBNDQCID6/CeGx0SJKUpDIGAMSIV2GcHQnCONXa2eCWAABQP6/CODc6KElKtdJNDQCID6/COB8O4Mq0dTW4JQAA1M+vMM4G3dSZdrqpAQDx4VUYF8eCMG5tI4wBAPHhVRgXsiOSpNYOuqkBAPHhVRi7XHDPuK2dMAYAxIdXYVwKK2PCGAAQJ16FsfIjyrq0EqlUo1sCAEDdvApjy49ozFjLGAAQL16FcaIwqixhDACIGQ/DuLXRzQAAYEq8CuNkcVT5BGEMAIgXr8I4XRxTjjAGAMSMV2GccWPKJ7hnDACIF8/COKsC94wBADHjVRinXY7KGAAQO3WFsZndYGa7zGyPmb3nOc672cycmW2euSbWL+NyKiQJYwBAvJw2jM0sKeluSS+XtEnS681s0yTndUl6h6Tvz3Qj69XisioygAsAEDP1VMZbJO1xzu11zuUk3SPppknO+yNJfyppbAbbNyUZ5VSgmxoAEDP1hPFKSfsi+/vDYxVm9gJJq51z/zaDbZsa59SinIp0UwMAYmbaA7jMLCHpQ5LeXce5W83sITN7qK+vb7rfulYxp4Scikm6qQEA8VJPGB+QtDqyvyo8VtYl6UJJ3zKzpyVdKWnbZIO4nHMfc85tds5t7u3tPfNWTyY/Kkkq0k0NAIiZesJ4u6SNZna2mWUk3SJpW/lF51y/c26Jc26dc26dpAclvco599CstPhUCsGt6lKKyhgAEC+nDWPnXEHSHZLul/S4pM8453aY2V1m9qrZbmDdwsq4xD1jAEDMpOo5yTl3n6T7xh278xTnXjv9Zp2BcmWcbGvItwcA4Ez5MwNXWBk7uqkBADHjTRiXcmE3dYpuagBAvHgTxsUwjB2PNgEAYsabMC6F3dRKc88YABAv3oRxMTcSbHDPGAAQM96EsQu7qQljAEDceBPG5W5ql25vcEsAAJgab8LY5amMAQDx5E8Yh93UliaMAQDx4k8Y58dUcAkl05lGNwUAgCnxJoxVGNWYMkom/PmRAADzgzfJ5QpZ5ZRSOmGNbgoAAFPiTxgX8yoopSRhDACIGY/CuKC8kkonvfmRAADzhD/JVcqr4JJUxgCA2PEmjF2xoIKSShHGAICY8SaMVcwHYUw3NQAgZvxJrlIQxnRTAwDixp8wLuaVp5saABBD/oRxqaiikkolCWMAQLx4E8YWdlOnmIELABAz/iRXqaC8ozIGAMSPN2EcVMYp7hkDAGLHozAuqqAEo6kBALHjURgHlTHTYQIA4sab5DJX4DljAEAs+RPGpQLPGQMAYsmrMC4yHSYAIIa8Sa6ECx9tojIGAMSMP2Fc4p4xACCevAnj8gAuKmMAQNx4E8YJV1TRkjIjjAEA8eJRGBdUslSjmwEAwJR5E8ZJwhgAEFN+hHGpKJNTkTAGAMSQJ2FckCQ5Sza4IQAATJ0fYVzMB5+ojAEAMeRHGJeCMOaeMQAgjvwI42LQTU1lDACIIz/CmHvGAIAY8ySM6aYGAMSXH2EcDuAqJQhjAED8+BHGYTc1lTEAII4IYwAAGsyPMA67qR1hDACIIT/CuFwZc88YABBDfoRxuTImjAEAMeRHGJefM06kG9wQAACmzpMwLt8zZtIPAED8+BHGxXJlTDc1ACB+/AjjEveMAQDx5UcYVwZwZRrcEAAAps6PMA4HcCnhx48DAJhf/EivShgzmhoAED9+hHF5oQgjjAEA8eNHGC85V/+avkH5VEejWwIAwJT5EcZrrtCHW96ubLq70S0BAGDK/AhjSUXnlExYo5sBAMCU1RXGZnaDme0ysz1m9p5JXn+bmT1mZo+Y2X+a2aaZb+pzK5ackkYYAwDi57RhbGZJSXdLermkTZJeP0nYftI5d5Fz7vmS/kzSh2a8padRKjklqIwBADFUT2W8RdIe59xe51xO0j2Sboqe4JwbiOx2SHIz18T6FB2VMQAgnuqZP3KlpH2R/f2Srhh/kpn9uqR3ScpIum5GWjcFxZKUTBLGAID4mbEBXM65u51z50j6PUn/fbJzzGyrmT1kZg/19fXN1LeWJBVLJSpjAEAs1RPGByStjuyvCo+dyj2SXj3ZC865jznnNjvnNvf29tbfyjoUS4ymBgDEUz1hvF3SRjM728wykm6RtC16gpltjOzeKGn3zDWxPiUnJaiMAQAxdNp7xs65gpndIel+SUlJH3fO7TCzuyQ95JzbJukOM3uJpLykE5LeOJuNnkxQGc/1dwUAYPrqWgDYOXefpPvGHbszsv2OGW7XlAWTfpDGAID48Sa9qIwBAHHlTXwxAxcAIK68CONSKZhjhBm4AABx5EUYF10QxlTGAIA48iOMw8qYGbgAAHHkVxhTGQMAYsiPMC53U3PPGAAQQ16EcWUAF5UxACCGvAjjSjc1lTEAIIb8CGO6qQEAMeZHGFMZAwBizK8w5p4xACCGvAjjUin4zAxcAIA48iKMq/eMG9wQAADOgBfxVeTRJgBAjHkVxinWMwYAxJAX6VUdTd3ghgAAcAa8iK+So5saABBfXoQxzxkDAOLMjzAuV8aEMQAghvwI48oALsIYABA/XoUxM3ABAOLIizCuLKFIZQwAiCEvwphVmwAAceZHGDMDFwAgxrwI4/JzxgzgAgDEkRdhXCjSTQ0AiC8vwpgZuAAAceZFGBfD9YypjAEAceRHGLOeMQAgxryIr1JlbmovfhwAwDzjRXoVmIELABBjXoRxdQauBjcEAIAz4EV8LezI6JLVC9SSSja6KQAATFmq0Q2YCddvWqbrNy1rdDMAADgjXlTGAADEGWEMAECDEcYAADQYYQwAQIMRxgAANBhhDABAgxHGAAA0GGEMAECDEcYAADQYYQwAQIMRxgAANBhhDABAgxHGAAA0GGEMAECDEcYAADQYYQwAQIMRxgAANBhhDABAg5lzrjHf2KxP0s9m8C2XSDo6g+83X3Edp49rOH1cw5nBdZy+mbyGa51zvZO90LAwnmlm9pBzbnOj2xF3XMfp4xpOH9dwZnAdp2+uriHd1AAANBhhDABAg/kUxh9rdAM8wXWcPq7h9HENZwbXcfrm5Bp6c88YAIC48qkyBgAglrwIYzO7wcx2mdkeM3tPo9vTrMzs42Z2xMx+Ejm2yMy+ama7w88Lw+NmZv8rvKY/NrMXNK7lzcPMVpvZN81sp5ntMLN3hMe5jlNgZq1m9gMzezS8jv8zPH62mX0/vF6fNrNMeLwl3N8Tvr6uke1vJmaWNLMfmdmXw32u4RSY2dNm9piZPWJmD4XH5vz3OfZhbGZJSXdLermkTZJeb2abGtuqpvVPkm4Yd+w9kr7unNso6evhvhRcz43hx1ZJfzNHbWx2BUnvds5tknSlpF8P/3/jOk5NVtJ1zrlLJD1f0g1mdqWkP5X0YefcBkknJN0enn+7pBPh8Q+H5yHwDkmPR/a5hlP3Yufc8yOPMM3573Psw1jSFkl7nHN7nXM5SfdIuqnBbWpKzrlvSzo+7vBNkj4Rbn9C0qsjx/+PCzwoaYGZrZibljYv59xB59wPw+1BBX8EV4rrOCXh9RgKd9Phh5N0naR7w+Pjr2P5+t4r6efNzOaouU3LzFZJulHS34f7Jq7hTJjz32cfwnilpH2R/f3hMdRnmXPuYLh9SNKycJvrehphN9+lkr4vruOUhd2rj0g6Iumrkp6SdNI5VwhPiV6rynUMX++XtHhuW9yUPiLpdyWVwv3F4hpOlZP0/8zsYTPbGh6b89/n1Ey8CfzgnHNmxvD6OphZp6TPSXqnc24gWmBwHevjnCtKer6ZLZD0BUnnN7hJsWJmr5B0xDn3sJld2+j2xNjVzrkDZrZU0lfN7Inoi3P1++xDZXxA0urI/qrwGOpzuNzNEn4+Eh7nup6CmaUVBPG/OOc+Hx7mOp4h59xJSd+UdJWCbr9ykRC9VpXrGL7eI+nYHDe12bxI0qvM7GkFt+euk/SX4hpOiXPuQPj5iIJ/FG5RA36ffQjj7ZI2hiMIM5JukbStwW2Kk22S3hhuv1HSlyLHbwtHD14pqT/SbTNvhffY/kHS4865D0Ve4jpOgZn1hhWxzKxN0vUK7r9/U9JrwtPGX8fy9X2NpG+4eT5JgnPuvc65Vc65dQr+7n3DOffL4hrWzcw6zKyrvC3ppZJ+okb8PjvnYv8h6RckPangntMfNLo9zfoh6VOSDkrKK7jXcbuCe0Zfl7Rb0tckLQrPNQWj1J+S9JikzY1ufzN8SLpawT2mH0t6JPz4Ba7jlK/jxZJ+FF7Hn0i6Mzy+XtIPJO2R9FlJLeHx1nB/T/j6+kb/DM30IelaSV/mGk75uq2X9Gj4saOcH434fWYGLgAAGsyHbmoAAGKNMAYAoMEIYwAAGowwBgCgwQhjAAAajDAGAKDBCGMAABqMMAYAoMH+P92/qaynQgbHAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 576x576 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ivg89WnDAv9s"
      },
      "source": [
        "model.save_weights('/content/drive/MyDrive/CNN-model/first')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LEfFGgH29fmQ"
      },
      "source": [
        "# Experimenting on evals"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "43F2aHtT9BOn"
      },
      "source": [
        "def plot_confusion_matrix(cm, classes,\n",
        "                          normalize=False,\n",
        "                          title='Confusion matrix',\n",
        "                          cmap=plt.cm.Blues):\n",
        "  \"\"\"\n",
        "  This function prints and plots the confusion matrix.\n",
        "  Normalization can be applied by setting `normalize=True`.\n",
        "  \n",
        "  if normalize:\n",
        "    cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
        "    print(\"Normalized confusion matrix\")\n",
        "  else:\n",
        "    print('Confusion matrix, without normalization')\n",
        "  \"\"\"\n",
        "  #print(cm)\n",
        "\n",
        "  plt.imshow(cm, interpolation='nearest', cmap=cmap)\n",
        "  plt.title(title)\n",
        "  plt.colorbar()\n",
        "  tick_marks = np.arange(len(classes))\n",
        "  plt.xticks(tick_marks, classes, rotation=45)\n",
        "  plt.yticks(tick_marks, classes)\n",
        "\n",
        "  fmt = '.2f' if normalize else 'd'\n",
        "  thresh = cm.max() / 2.\n",
        "  for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
        "    plt.text(j, i, format(cm[i, j], fmt),\n",
        "            horizontalalignment=\"center\",\n",
        "            color=\"white\" if cm[i, j] > thresh else \"black\")\n",
        "\n",
        "  plt.tight_layout()\n",
        "  plt.ylabel('True label')\n",
        "  plt.xlabel('Predicted label')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MfRsmX5p9btA"
      },
      "source": [
        "from sklearn.metrics import confusion_matrix, f1_score, classification_report\n",
        "from sklearn.metrics import recall_score, precision_score, accuracy_score\n",
        "import seaborn as sns #visualization library, must be imported before all other plotting libraries\n",
        "import matplotlib.pyplot as plt"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AEj-z5XN9h98",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "ed869c85-9a00-4898-ecb3-5b2fbde850c3"
      },
      "source": [
        "from collections import Counter\n",
        "Counter(y)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Counter({'flu': 451,\n",
              "         'gac': 637,\n",
              "         'org': 682,\n",
              "         'pia': 721,\n",
              "         'tru': 577,\n",
              "         'voi': 778})"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 27
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_ugCGtuG9vvP"
      },
      "source": [
        "def plot_confusion_matrix(cnf_matrix):\n",
        "  df_cm = pd.DataFrame(cnf_matrix, index=INSTRUMENTS, columns=INSTRUMENTS)\n",
        "  plt.figure(figsize = (10,7))\n",
        "  sns.set(font_scale=1.0) #for label size\n",
        "  sns.heatmap(df_cm, annot=True, annot_kws={'size': 18}, fmt='g') # font size\n",
        "  plt.show()\n",
        "\n",
        "def display_score(y_test, y_pred):\n",
        "  print('Recall: ', recall_score(y_test, y_pred, average=None))\n",
        "  print('Precision: ', precision_score(y_test, y_pred, average=None))\n",
        "  print('F1-Score: ', f1_score(y_test, y_pred, average=None))\n",
        "  print('Accuracy: %.2f ' % accuracy_score(y_test, y_pred, normalize=True))\n",
        "  print('Number of samples: ', y_test.shape[0])\n",
        "\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7GDRM6G2CWWv"
      },
      "source": [
        "y_pred = model.predict(X_test)\n",
        "y_pred_cat = np.argmax(y_pred, axis=1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "7thUi-xmCoVv",
        "outputId": "203509b2-f56a-491d-ff7e-19972043e0cc"
      },
      "source": [
        "display_score(y_test, y_pred_cat)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Recall:  [0.69230769 0.71681416 0.86725664 0.87128713 0.72093023 0.78787879]\n",
            "Precision:  [0.69230769 0.87096774 0.66216216 0.72727273 0.95384615 0.91764706]\n",
            "F1-Score:  [0.69230769 0.78640777 0.75095785 0.79279279 0.82119205 0.84782609]\n",
            "Accuracy: 0.78 \n",
            "Number of samples:  577\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 431
        },
        "id": "eSgnFg3mCtG5",
        "outputId": "739a4e9a-1f84-4d74-def3-67eb91697aff"
      },
      "source": [
        "cnf_matrix = confusion_matrix(y_test, y_pred_cat)\n",
        "plot_confusion_matrix(cnf_matrix)  "
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x504 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "orD1Me15DleY"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}