Cnn on Cifr.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "name": "26.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.7.3"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "7AMl8vxrgnex"
      },
      "source": [
        "import keras\n",
        "from keras.datasets import cifar10\n",
        "from keras.models import Model, Sequential\n",
        "from keras.layers import Dense, Dropout, Flatten, Input, AveragePooling2D, merge, Activation\n",
        "from keras.layers import Conv2D, MaxPooling2D, BatchNormalization\n",
        "from keras.layers import Concatenate\n",
        "from keras.optimizers import Adam\n",
        "from tensorflow.keras import models, layers\n",
        "from tensorflow.keras.models import Model\n",
        "from tensorflow.keras.layers import BatchNormalization, Activation, Flatten\n",
        "from tensorflow.keras.optimizers import Adam\n"
      ],
      "execution_count": 16,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PM67ZMKuhes5"
      },
      "source": [
        "# Hyperparameters\n",
        "batch_size = 128\n",
        "num_classes = 10\n",
        "epochs = 45\n",
        "l = 40\n",
        "num_filter = 12\n",
        "compression = 0.5\n",
        "dropout_rate = 0.2"
      ],
      "execution_count": 17,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NxiVqpjGhVvc"
      },
      "source": [
        "import tensorflow as tf"
      ],
      "execution_count": 18,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FR5HDHrjiV6N"
      },
      "source": [
        "num_classes = 10"
      ],
      "execution_count": 19,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EzfWa7syhZgU"
      },
      "source": [
        "# Load CIFAR10 Data\n",
        "(X_train, y_train), (X_test, y_test) = tf.keras.datasets.cifar10.load_data()\n",
        "img_height, img_width, channel = X_train.shape[1],X_train.shape[2],X_train.shape[3]\n",
        "\n",
        "# convert to one hot encoing \n",
        "y_train = tf.keras.utils.to_categorical(y_train, num_classes)\n",
        "y_test = tf.keras.utils.to_categorical(y_test, num_classes) "
      ],
      "execution_count": 20,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lx2hEjNCJa6z"
      },
      "source": [
        "X_train = X_train / 255.0\n",
        "X_test = X_test / 255.0"
      ],
      "execution_count": 21,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "QgmPzEJ4iRdn",
        "outputId": "3982586e-13fc-4901-f84b-f9762f52723b"
      },
      "source": [
        "print(X_test.shape)\n",
        "X_train.shape"
      ],
      "execution_count": 22,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(10000, 32, 32, 3)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(50000, 32, 32, 3)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 22
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ADieaCPRi02f"
      },
      "source": [
        "# Dense Block\n",
        "def denseblock(input, num_filter = 12, dropout_rate = 0.2):\n",
        "  global compression\n",
        "  temp = input\n",
        "  for _ in range(l):\n",
        "    BatchNorm = layers.BatchNormalization()(temp)\n",
        "    relu = layers.Activation('relu')(BatchNorm)\n",
        "    Conv2D_3_3 = layers.Conv2D(int(num_filter*compression), (3,3), use_bias=False ,padding='same')(relu)\n",
        "    if (dropout_rate>0):\n",
        "      Conv2D_3_3 = layers.Dropout(dropout_rate)(Conv2D_3_3)\n",
        "      concat = layers.Concatenate(axis=-1)([temp,Conv2D_3_3])\n",
        "        \n",
        "    temp = concat\n",
        "        \n",
        "  return temp\n",
        "\n",
        "## transition Blosck\n",
        "def transition(input, num_filter = 12, dropout_rate = 0.2):\n",
        "  global compression\n",
        "  BatchNorm = layers.BatchNormalization()(input)\n",
        "  relu = layers.Activation('relu')(BatchNorm)\n",
        "  Conv2D_BottleNeck = layers.Conv2D(int(num_filter*compression), (1,1), use_bias=False ,padding='same')(relu)\n",
        "  if dropout_rate>0:\n",
        "    Conv2D_BottleNeck = layers.Dropout(dropout_rate)(Conv2D_BottleNeck)\n",
        "  avg = layers.AveragePooling2D(pool_size=(2,2))(Conv2D_BottleNeck)\n",
        "  return avg\n",
        "\n",
        "#output layer\n",
        "def output_layer(input):\n",
        "  global compression\n",
        "  BatchNorm = layers.BatchNormalization()(input)\n",
        "  relu = layers.Activation('relu')(BatchNorm)\n",
        "  AvgPooling = layers.AveragePooling2D(pool_size=(2,2))(relu)\n",
        "  flat = layers.Flatten()(AvgPooling)\n",
        "  output = layers.Dense(num_classes, activation='softmax')(flat)\n",
        "  return output"
      ],
      "execution_count": 23,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "TL0jYPqvjWK-",
        "outputId": "8c3fd28e-193e-4342-e3f6-854627391ad3"
      },
      "source": [
        "X_train[0].shape"
      ],
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(32, 32, 3)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 24
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HUJNaKuCgNr_"
      },
      "source": [
        "num_filter = 12\n",
        "dropout_rate = 0.2\n",
        "l = 12\n",
        "input = layers.Input(shape=(img_height, img_width, channel,))\n",
        "First_Conv2D = layers.Conv2D(num_filter, (3,3), use_bias=False ,padding='same')(input)\n",
        "\n",
        "First_Block = denseblock(First_Conv2D, num_filter, dropout_rate)\n",
        "First_Transition = transition(First_Block, num_filter, dropout_rate)\n",
        "\n",
        "Second_Block = denseblock(First_Transition, num_filter, dropout_rate)\n",
        "Second_Transition = transition(Second_Block, num_filter, dropout_rate)\n",
        "\n",
        "Third_Block = denseblock(Second_Transition, num_filter, dropout_rate)\n",
        "Third_Transition = transition(Third_Block, num_filter, dropout_rate)\n",
        "\n",
        "Last_Block = denseblock(Third_Transition,  num_filter, dropout_rate)\n",
        "output = output_layer(Last_Block)"
      ],
      "execution_count": 25,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rms3Pkj0hCoK",
        "outputId": "ce85d00f-0b58-4fb9-b2c1-8e7f92f5378a"
      },
      "source": [
        "model = Model(inputs=[input], outputs=[output])\n",
        "model.summary()"
      ],
      "execution_count": 26,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"model_1\"\n",
            "__________________________________________________________________________________________________\n",
            "Layer (type)                    Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            "input_2 (InputLayer)            [(None, 32, 32, 3)]  0                                            \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_52 (Conv2D)              (None, 32, 32, 12)   324         input_2[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_52 (BatchNo (None, 32, 32, 12)   48          conv2d_52[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "activation_52 (Activation)      (None, 32, 32, 12)   0           batch_normalization_52[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_53 (Conv2D)              (None, 32, 32, 6)    648         activation_52[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_51 (Dropout)            (None, 32, 32, 6)    0           conv2d_53[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_48 (Concatenate)    (None, 32, 32, 18)   0           conv2d_52[0][0]                  \n",
            "                                                                 dropout_51[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_53 (BatchNo (None, 32, 32, 18)   72          concatenate_48[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_53 (Activation)      (None, 32, 32, 18)   0           batch_normalization_53[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_54 (Conv2D)              (None, 32, 32, 6)    972         activation_53[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_52 (Dropout)            (None, 32, 32, 6)    0           conv2d_54[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_49 (Concatenate)    (None, 32, 32, 24)   0           concatenate_48[0][0]             \n",
            "                                                                 dropout_52[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_54 (BatchNo (None, 32, 32, 24)   96          concatenate_49[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_54 (Activation)      (None, 32, 32, 24)   0           batch_normalization_54[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_55 (Conv2D)              (None, 32, 32, 6)    1296        activation_54[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_53 (Dropout)            (None, 32, 32, 6)    0           conv2d_55[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_50 (Concatenate)    (None, 32, 32, 30)   0           concatenate_49[0][0]             \n",
            "                                                                 dropout_53[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_55 (BatchNo (None, 32, 32, 30)   120         concatenate_50[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_55 (Activation)      (None, 32, 32, 30)   0           batch_normalization_55[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_56 (Conv2D)              (None, 32, 32, 6)    1620        activation_55[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_54 (Dropout)            (None, 32, 32, 6)    0           conv2d_56[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_51 (Concatenate)    (None, 32, 32, 36)   0           concatenate_50[0][0]             \n",
            "                                                                 dropout_54[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_56 (BatchNo (None, 32, 32, 36)   144         concatenate_51[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_56 (Activation)      (None, 32, 32, 36)   0           batch_normalization_56[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_57 (Conv2D)              (None, 32, 32, 6)    1944        activation_56[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_55 (Dropout)            (None, 32, 32, 6)    0           conv2d_57[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_52 (Concatenate)    (None, 32, 32, 42)   0           concatenate_51[0][0]             \n",
            "                                                                 dropout_55[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_57 (BatchNo (None, 32, 32, 42)   168         concatenate_52[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_57 (Activation)      (None, 32, 32, 42)   0           batch_normalization_57[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_58 (Conv2D)              (None, 32, 32, 6)    2268        activation_57[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_56 (Dropout)            (None, 32, 32, 6)    0           conv2d_58[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_53 (Concatenate)    (None, 32, 32, 48)   0           concatenate_52[0][0]             \n",
            "                                                                 dropout_56[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_58 (BatchNo (None, 32, 32, 48)   192         concatenate_53[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_58 (Activation)      (None, 32, 32, 48)   0           batch_normalization_58[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_59 (Conv2D)              (None, 32, 32, 6)    2592        activation_58[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_57 (Dropout)            (None, 32, 32, 6)    0           conv2d_59[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_54 (Concatenate)    (None, 32, 32, 54)   0           concatenate_53[0][0]             \n",
            "                                                                 dropout_57[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_59 (BatchNo (None, 32, 32, 54)   216         concatenate_54[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_59 (Activation)      (None, 32, 32, 54)   0           batch_normalization_59[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_60 (Conv2D)              (None, 32, 32, 6)    2916        activation_59[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_58 (Dropout)            (None, 32, 32, 6)    0           conv2d_60[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_55 (Concatenate)    (None, 32, 32, 60)   0           concatenate_54[0][0]             \n",
            "                                                                 dropout_58[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_60 (BatchNo (None, 32, 32, 60)   240         concatenate_55[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_60 (Activation)      (None, 32, 32, 60)   0           batch_normalization_60[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_61 (Conv2D)              (None, 32, 32, 6)    3240        activation_60[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_59 (Dropout)            (None, 32, 32, 6)    0           conv2d_61[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_56 (Concatenate)    (None, 32, 32, 66)   0           concatenate_55[0][0]             \n",
            "                                                                 dropout_59[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_61 (BatchNo (None, 32, 32, 66)   264         concatenate_56[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_61 (Activation)      (None, 32, 32, 66)   0           batch_normalization_61[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_62 (Conv2D)              (None, 32, 32, 6)    3564        activation_61[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_60 (Dropout)            (None, 32, 32, 6)    0           conv2d_62[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_57 (Concatenate)    (None, 32, 32, 72)   0           concatenate_56[0][0]             \n",
            "                                                                 dropout_60[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_62 (BatchNo (None, 32, 32, 72)   288         concatenate_57[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_62 (Activation)      (None, 32, 32, 72)   0           batch_normalization_62[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_63 (Conv2D)              (None, 32, 32, 6)    3888        activation_62[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_61 (Dropout)            (None, 32, 32, 6)    0           conv2d_63[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_58 (Concatenate)    (None, 32, 32, 78)   0           concatenate_57[0][0]             \n",
            "                                                                 dropout_61[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_63 (BatchNo (None, 32, 32, 78)   312         concatenate_58[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_63 (Activation)      (None, 32, 32, 78)   0           batch_normalization_63[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_64 (Conv2D)              (None, 32, 32, 6)    4212        activation_63[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_62 (Dropout)            (None, 32, 32, 6)    0           conv2d_64[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_59 (Concatenate)    (None, 32, 32, 84)   0           concatenate_58[0][0]             \n",
            "                                                                 dropout_62[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_64 (BatchNo (None, 32, 32, 84)   336         concatenate_59[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_64 (Activation)      (None, 32, 32, 84)   0           batch_normalization_64[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_65 (Conv2D)              (None, 32, 32, 6)    504         activation_64[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_63 (Dropout)            (None, 32, 32, 6)    0           conv2d_65[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "average_pooling2d_4 (AveragePoo (None, 16, 16, 6)    0           dropout_63[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_65 (BatchNo (None, 16, 16, 6)    24          average_pooling2d_4[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "activation_65 (Activation)      (None, 16, 16, 6)    0           batch_normalization_65[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_66 (Conv2D)              (None, 16, 16, 6)    324         activation_65[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_64 (Dropout)            (None, 16, 16, 6)    0           conv2d_66[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_60 (Concatenate)    (None, 16, 16, 12)   0           average_pooling2d_4[0][0]        \n",
            "                                                                 dropout_64[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_66 (BatchNo (None, 16, 16, 12)   48          concatenate_60[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_66 (Activation)      (None, 16, 16, 12)   0           batch_normalization_66[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_67 (Conv2D)              (None, 16, 16, 6)    648         activation_66[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_65 (Dropout)            (None, 16, 16, 6)    0           conv2d_67[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_61 (Concatenate)    (None, 16, 16, 18)   0           concatenate_60[0][0]             \n",
            "                                                                 dropout_65[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_67 (BatchNo (None, 16, 16, 18)   72          concatenate_61[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_67 (Activation)      (None, 16, 16, 18)   0           batch_normalization_67[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_68 (Conv2D)              (None, 16, 16, 6)    972         activation_67[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_66 (Dropout)            (None, 16, 16, 6)    0           conv2d_68[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_62 (Concatenate)    (None, 16, 16, 24)   0           concatenate_61[0][0]             \n",
            "                                                                 dropout_66[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_68 (BatchNo (None, 16, 16, 24)   96          concatenate_62[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_68 (Activation)      (None, 16, 16, 24)   0           batch_normalization_68[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_69 (Conv2D)              (None, 16, 16, 6)    1296        activation_68[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_67 (Dropout)            (None, 16, 16, 6)    0           conv2d_69[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_63 (Concatenate)    (None, 16, 16, 30)   0           concatenate_62[0][0]             \n",
            "                                                                 dropout_67[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_69 (BatchNo (None, 16, 16, 30)   120         concatenate_63[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_69 (Activation)      (None, 16, 16, 30)   0           batch_normalization_69[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_70 (Conv2D)              (None, 16, 16, 6)    1620        activation_69[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_68 (Dropout)            (None, 16, 16, 6)    0           conv2d_70[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_64 (Concatenate)    (None, 16, 16, 36)   0           concatenate_63[0][0]             \n",
            "                                                                 dropout_68[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_70 (BatchNo (None, 16, 16, 36)   144         concatenate_64[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_70 (Activation)      (None, 16, 16, 36)   0           batch_normalization_70[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_71 (Conv2D)              (None, 16, 16, 6)    1944        activation_70[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_69 (Dropout)            (None, 16, 16, 6)    0           conv2d_71[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_65 (Concatenate)    (None, 16, 16, 42)   0           concatenate_64[0][0]             \n",
            "                                                                 dropout_69[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_71 (BatchNo (None, 16, 16, 42)   168         concatenate_65[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_71 (Activation)      (None, 16, 16, 42)   0           batch_normalization_71[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_72 (Conv2D)              (None, 16, 16, 6)    2268        activation_71[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_70 (Dropout)            (None, 16, 16, 6)    0           conv2d_72[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_66 (Concatenate)    (None, 16, 16, 48)   0           concatenate_65[0][0]             \n",
            "                                                                 dropout_70[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_72 (BatchNo (None, 16, 16, 48)   192         concatenate_66[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_72 (Activation)      (None, 16, 16, 48)   0           batch_normalization_72[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_73 (Conv2D)              (None, 16, 16, 6)    2592        activation_72[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_71 (Dropout)            (None, 16, 16, 6)    0           conv2d_73[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_67 (Concatenate)    (None, 16, 16, 54)   0           concatenate_66[0][0]             \n",
            "                                                                 dropout_71[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_73 (BatchNo (None, 16, 16, 54)   216         concatenate_67[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_73 (Activation)      (None, 16, 16, 54)   0           batch_normalization_73[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_74 (Conv2D)              (None, 16, 16, 6)    2916        activation_73[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_72 (Dropout)            (None, 16, 16, 6)    0           conv2d_74[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_68 (Concatenate)    (None, 16, 16, 60)   0           concatenate_67[0][0]             \n",
            "                                                                 dropout_72[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_74 (BatchNo (None, 16, 16, 60)   240         concatenate_68[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_74 (Activation)      (None, 16, 16, 60)   0           batch_normalization_74[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_75 (Conv2D)              (None, 16, 16, 6)    3240        activation_74[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_73 (Dropout)            (None, 16, 16, 6)    0           conv2d_75[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_69 (Concatenate)    (None, 16, 16, 66)   0           concatenate_68[0][0]             \n",
            "                                                                 dropout_73[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_75 (BatchNo (None, 16, 16, 66)   264         concatenate_69[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_75 (Activation)      (None, 16, 16, 66)   0           batch_normalization_75[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_76 (Conv2D)              (None, 16, 16, 6)    3564        activation_75[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_74 (Dropout)            (None, 16, 16, 6)    0           conv2d_76[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_70 (Concatenate)    (None, 16, 16, 72)   0           concatenate_69[0][0]             \n",
            "                                                                 dropout_74[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_76 (BatchNo (None, 16, 16, 72)   288         concatenate_70[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_76 (Activation)      (None, 16, 16, 72)   0           batch_normalization_76[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_77 (Conv2D)              (None, 16, 16, 6)    3888        activation_76[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_75 (Dropout)            (None, 16, 16, 6)    0           conv2d_77[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_71 (Concatenate)    (None, 16, 16, 78)   0           concatenate_70[0][0]             \n",
            "                                                                 dropout_75[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_77 (BatchNo (None, 16, 16, 78)   312         concatenate_71[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_77 (Activation)      (None, 16, 16, 78)   0           batch_normalization_77[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_78 (Conv2D)              (None, 16, 16, 6)    468         activation_77[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_76 (Dropout)            (None, 16, 16, 6)    0           conv2d_78[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "average_pooling2d_5 (AveragePoo (None, 8, 8, 6)      0           dropout_76[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_78 (BatchNo (None, 8, 8, 6)      24          average_pooling2d_5[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "activation_78 (Activation)      (None, 8, 8, 6)      0           batch_normalization_78[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_79 (Conv2D)              (None, 8, 8, 6)      324         activation_78[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_77 (Dropout)            (None, 8, 8, 6)      0           conv2d_79[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_72 (Concatenate)    (None, 8, 8, 12)     0           average_pooling2d_5[0][0]        \n",
            "                                                                 dropout_77[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_79 (BatchNo (None, 8, 8, 12)     48          concatenate_72[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_79 (Activation)      (None, 8, 8, 12)     0           batch_normalization_79[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_80 (Conv2D)              (None, 8, 8, 6)      648         activation_79[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_78 (Dropout)            (None, 8, 8, 6)      0           conv2d_80[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_73 (Concatenate)    (None, 8, 8, 18)     0           concatenate_72[0][0]             \n",
            "                                                                 dropout_78[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_80 (BatchNo (None, 8, 8, 18)     72          concatenate_73[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_80 (Activation)      (None, 8, 8, 18)     0           batch_normalization_80[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_81 (Conv2D)              (None, 8, 8, 6)      972         activation_80[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_79 (Dropout)            (None, 8, 8, 6)      0           conv2d_81[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_74 (Concatenate)    (None, 8, 8, 24)     0           concatenate_73[0][0]             \n",
            "                                                                 dropout_79[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_81 (BatchNo (None, 8, 8, 24)     96          concatenate_74[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_81 (Activation)      (None, 8, 8, 24)     0           batch_normalization_81[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_82 (Conv2D)              (None, 8, 8, 6)      1296        activation_81[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_80 (Dropout)            (None, 8, 8, 6)      0           conv2d_82[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_75 (Concatenate)    (None, 8, 8, 30)     0           concatenate_74[0][0]             \n",
            "                                                                 dropout_80[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_82 (BatchNo (None, 8, 8, 30)     120         concatenate_75[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_82 (Activation)      (None, 8, 8, 30)     0           batch_normalization_82[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_83 (Conv2D)              (None, 8, 8, 6)      1620        activation_82[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_81 (Dropout)            (None, 8, 8, 6)      0           conv2d_83[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_76 (Concatenate)    (None, 8, 8, 36)     0           concatenate_75[0][0]             \n",
            "                                                                 dropout_81[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_83 (BatchNo (None, 8, 8, 36)     144         concatenate_76[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_83 (Activation)      (None, 8, 8, 36)     0           batch_normalization_83[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_84 (Conv2D)              (None, 8, 8, 6)      1944        activation_83[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_82 (Dropout)            (None, 8, 8, 6)      0           conv2d_84[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_77 (Concatenate)    (None, 8, 8, 42)     0           concatenate_76[0][0]             \n",
            "                                                                 dropout_82[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_84 (BatchNo (None, 8, 8, 42)     168         concatenate_77[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_84 (Activation)      (None, 8, 8, 42)     0           batch_normalization_84[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_85 (Conv2D)              (None, 8, 8, 6)      2268        activation_84[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_83 (Dropout)            (None, 8, 8, 6)      0           conv2d_85[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_78 (Concatenate)    (None, 8, 8, 48)     0           concatenate_77[0][0]             \n",
            "                                                                 dropout_83[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_85 (BatchNo (None, 8, 8, 48)     192         concatenate_78[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_85 (Activation)      (None, 8, 8, 48)     0           batch_normalization_85[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_86 (Conv2D)              (None, 8, 8, 6)      2592        activation_85[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_84 (Dropout)            (None, 8, 8, 6)      0           conv2d_86[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_79 (Concatenate)    (None, 8, 8, 54)     0           concatenate_78[0][0]             \n",
            "                                                                 dropout_84[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_86 (BatchNo (None, 8, 8, 54)     216         concatenate_79[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_86 (Activation)      (None, 8, 8, 54)     0           batch_normalization_86[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_87 (Conv2D)              (None, 8, 8, 6)      2916        activation_86[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_85 (Dropout)            (None, 8, 8, 6)      0           conv2d_87[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_80 (Concatenate)    (None, 8, 8, 60)     0           concatenate_79[0][0]             \n",
            "                                                                 dropout_85[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_87 (BatchNo (None, 8, 8, 60)     240         concatenate_80[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_87 (Activation)      (None, 8, 8, 60)     0           batch_normalization_87[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_88 (Conv2D)              (None, 8, 8, 6)      3240        activation_87[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_86 (Dropout)            (None, 8, 8, 6)      0           conv2d_88[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_81 (Concatenate)    (None, 8, 8, 66)     0           concatenate_80[0][0]             \n",
            "                                                                 dropout_86[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_88 (BatchNo (None, 8, 8, 66)     264         concatenate_81[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_88 (Activation)      (None, 8, 8, 66)     0           batch_normalization_88[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_89 (Conv2D)              (None, 8, 8, 6)      3564        activation_88[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_87 (Dropout)            (None, 8, 8, 6)      0           conv2d_89[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_82 (Concatenate)    (None, 8, 8, 72)     0           concatenate_81[0][0]             \n",
            "                                                                 dropout_87[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_89 (BatchNo (None, 8, 8, 72)     288         concatenate_82[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_89 (Activation)      (None, 8, 8, 72)     0           batch_normalization_89[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_90 (Conv2D)              (None, 8, 8, 6)      3888        activation_89[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_88 (Dropout)            (None, 8, 8, 6)      0           conv2d_90[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_83 (Concatenate)    (None, 8, 8, 78)     0           concatenate_82[0][0]             \n",
            "                                                                 dropout_88[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_90 (BatchNo (None, 8, 8, 78)     312         concatenate_83[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_90 (Activation)      (None, 8, 8, 78)     0           batch_normalization_90[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_91 (Conv2D)              (None, 8, 8, 6)      468         activation_90[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_89 (Dropout)            (None, 8, 8, 6)      0           conv2d_91[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "average_pooling2d_6 (AveragePoo (None, 4, 4, 6)      0           dropout_89[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_91 (BatchNo (None, 4, 4, 6)      24          average_pooling2d_6[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "activation_91 (Activation)      (None, 4, 4, 6)      0           batch_normalization_91[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_92 (Conv2D)              (None, 4, 4, 6)      324         activation_91[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_90 (Dropout)            (None, 4, 4, 6)      0           conv2d_92[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_84 (Concatenate)    (None, 4, 4, 12)     0           average_pooling2d_6[0][0]        \n",
            "                                                                 dropout_90[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_92 (BatchNo (None, 4, 4, 12)     48          concatenate_84[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_92 (Activation)      (None, 4, 4, 12)     0           batch_normalization_92[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_93 (Conv2D)              (None, 4, 4, 6)      648         activation_92[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_91 (Dropout)            (None, 4, 4, 6)      0           conv2d_93[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_85 (Concatenate)    (None, 4, 4, 18)     0           concatenate_84[0][0]             \n",
            "                                                                 dropout_91[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_93 (BatchNo (None, 4, 4, 18)     72          concatenate_85[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_93 (Activation)      (None, 4, 4, 18)     0           batch_normalization_93[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_94 (Conv2D)              (None, 4, 4, 6)      972         activation_93[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_92 (Dropout)            (None, 4, 4, 6)      0           conv2d_94[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_86 (Concatenate)    (None, 4, 4, 24)     0           concatenate_85[0][0]             \n",
            "                                                                 dropout_92[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_94 (BatchNo (None, 4, 4, 24)     96          concatenate_86[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_94 (Activation)      (None, 4, 4, 24)     0           batch_normalization_94[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_95 (Conv2D)              (None, 4, 4, 6)      1296        activation_94[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_93 (Dropout)            (None, 4, 4, 6)      0           conv2d_95[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_87 (Concatenate)    (None, 4, 4, 30)     0           concatenate_86[0][0]             \n",
            "                                                                 dropout_93[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_95 (BatchNo (None, 4, 4, 30)     120         concatenate_87[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_95 (Activation)      (None, 4, 4, 30)     0           batch_normalization_95[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_96 (Conv2D)              (None, 4, 4, 6)      1620        activation_95[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_94 (Dropout)            (None, 4, 4, 6)      0           conv2d_96[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_88 (Concatenate)    (None, 4, 4, 36)     0           concatenate_87[0][0]             \n",
            "                                                                 dropout_94[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_96 (BatchNo (None, 4, 4, 36)     144         concatenate_88[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_96 (Activation)      (None, 4, 4, 36)     0           batch_normalization_96[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_97 (Conv2D)              (None, 4, 4, 6)      1944        activation_96[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_95 (Dropout)            (None, 4, 4, 6)      0           conv2d_97[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_89 (Concatenate)    (None, 4, 4, 42)     0           concatenate_88[0][0]             \n",
            "                                                                 dropout_95[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_97 (BatchNo (None, 4, 4, 42)     168         concatenate_89[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_97 (Activation)      (None, 4, 4, 42)     0           batch_normalization_97[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_98 (Conv2D)              (None, 4, 4, 6)      2268        activation_97[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_96 (Dropout)            (None, 4, 4, 6)      0           conv2d_98[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_90 (Concatenate)    (None, 4, 4, 48)     0           concatenate_89[0][0]             \n",
            "                                                                 dropout_96[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_98 (BatchNo (None, 4, 4, 48)     192         concatenate_90[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_98 (Activation)      (None, 4, 4, 48)     0           batch_normalization_98[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_99 (Conv2D)              (None, 4, 4, 6)      2592        activation_98[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_97 (Dropout)            (None, 4, 4, 6)      0           conv2d_99[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_91 (Concatenate)    (None, 4, 4, 54)     0           concatenate_90[0][0]             \n",
            "                                                                 dropout_97[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_99 (BatchNo (None, 4, 4, 54)     216         concatenate_91[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_99 (Activation)      (None, 4, 4, 54)     0           batch_normalization_99[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_100 (Conv2D)             (None, 4, 4, 6)      2916        activation_99[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "dropout_98 (Dropout)            (None, 4, 4, 6)      0           conv2d_100[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_92 (Concatenate)    (None, 4, 4, 60)     0           concatenate_91[0][0]             \n",
            "                                                                 dropout_98[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_100 (BatchN (None, 4, 4, 60)     240         concatenate_92[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_100 (Activation)     (None, 4, 4, 60)     0           batch_normalization_100[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_101 (Conv2D)             (None, 4, 4, 6)      3240        activation_100[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_99 (Dropout)            (None, 4, 4, 6)      0           conv2d_101[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_93 (Concatenate)    (None, 4, 4, 66)     0           concatenate_92[0][0]             \n",
            "                                                                 dropout_99[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_101 (BatchN (None, 4, 4, 66)     264         concatenate_93[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_101 (Activation)     (None, 4, 4, 66)     0           batch_normalization_101[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_102 (Conv2D)             (None, 4, 4, 6)      3564        activation_101[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_100 (Dropout)           (None, 4, 4, 6)      0           conv2d_102[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_94 (Concatenate)    (None, 4, 4, 72)     0           concatenate_93[0][0]             \n",
            "                                                                 dropout_100[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_102 (BatchN (None, 4, 4, 72)     288         concatenate_94[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_102 (Activation)     (None, 4, 4, 72)     0           batch_normalization_102[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_103 (Conv2D)             (None, 4, 4, 6)      3888        activation_102[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_101 (Dropout)           (None, 4, 4, 6)      0           conv2d_103[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_95 (Concatenate)    (None, 4, 4, 78)     0           concatenate_94[0][0]             \n",
            "                                                                 dropout_101[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_103 (BatchN (None, 4, 4, 78)     312         concatenate_95[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_103 (Activation)     (None, 4, 4, 78)     0           batch_normalization_103[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "average_pooling2d_7 (AveragePoo (None, 2, 2, 78)     0           activation_103[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "flatten_1 (Flatten)             (None, 312)          0           average_pooling2d_7[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "dense_1 (Dense)                 (None, 10)           3130        flatten_1[0][0]                  \n",
            "==================================================================================================\n",
            "Total params: 118,918\n",
            "Trainable params: 114,394\n",
            "Non-trainable params: 4,524\n",
            "__________________________________________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GcwkN1dOzA0r"
      },
      "source": [
        "from tensorflow.keras.callbacks import ReduceLROnPlateau\n",
        "DynamicLR =ReduceLROnPlateau(monitor='val_accuracy', factor=0.9, patience=2, verbose=1, mode='max',min_delta=0, cooldown=0, min_lr=0.0001)"
      ],
      "execution_count": 27,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RC5Pm5flkDcj"
      },
      "source": [
        "model.compile(loss='categorical_crossentropy',\n",
        "              optimizer=Adam(0.001),\n",
        "              metrics=['accuracy'])"
      ],
      "execution_count": 28,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "avGtVVFskQgb",
        "outputId": "f56dad04-3e7e-43cc-f0e2-fabc12b5844c"
      },
      "source": [
        "history=model.fit(X_train, y_train,\n",
        "                    batch_size=batch_size,\n",
        "                    epochs=epochs,\n",
        "                    verbose=1, \n",
        "                    validation_data=(X_test, y_test),callbacks=[DynamicLR])"
      ],
      "execution_count": 29,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/45\n",
            "391/391 [==============================] - 45s 101ms/step - loss: 1.9460 - accuracy: 0.2754 - val_loss: 2.8759 - val_accuracy: 0.2032\n",
            "Epoch 2/45\n",
            "391/391 [==============================] - 38s 98ms/step - loss: 1.4348 - accuracy: 0.4694 - val_loss: 1.4630 - val_accuracy: 0.4842\n",
            "Epoch 3/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 1.2537 - accuracy: 0.5391 - val_loss: 1.3320 - val_accuracy: 0.5442\n",
            "Epoch 4/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 1.1374 - accuracy: 0.5859 - val_loss: 1.5929 - val_accuracy: 0.4779\n",
            "Epoch 5/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 1.0457 - accuracy: 0.6234 - val_loss: 1.3661 - val_accuracy: 0.5521\n",
            "Epoch 6/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.9913 - accuracy: 0.6419 - val_loss: 1.0956 - val_accuracy: 0.6354\n",
            "Epoch 7/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.9377 - accuracy: 0.6637 - val_loss: 1.1892 - val_accuracy: 0.6325\n",
            "Epoch 8/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.9116 - accuracy: 0.6713 - val_loss: 1.0511 - val_accuracy: 0.6457\n",
            "Epoch 9/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.8891 - accuracy: 0.6771 - val_loss: 1.2356 - val_accuracy: 0.6049\n",
            "Epoch 10/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.8654 - accuracy: 0.6895 - val_loss: 1.9061 - val_accuracy: 0.4936\n",
            "\n",
            "Epoch 00010: ReduceLROnPlateau reducing learning rate to 0.0009000000427477062.\n",
            "Epoch 11/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.8310 - accuracy: 0.7022 - val_loss: 1.2118 - val_accuracy: 0.6017\n",
            "Epoch 12/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.8076 - accuracy: 0.7132 - val_loss: 1.0136 - val_accuracy: 0.6726\n",
            "Epoch 13/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.8038 - accuracy: 0.7103 - val_loss: 0.9712 - val_accuracy: 0.6824\n",
            "Epoch 14/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.7796 - accuracy: 0.7217 - val_loss: 1.0015 - val_accuracy: 0.6639\n",
            "Epoch 15/45\n",
            "391/391 [==============================] - 38s 98ms/step - loss: 0.7540 - accuracy: 0.7288 - val_loss: 1.2778 - val_accuracy: 0.6114\n",
            "\n",
            "Epoch 00015: ReduceLROnPlateau reducing learning rate to 0.0008100000384729356.\n",
            "Epoch 16/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.7563 - accuracy: 0.7259 - val_loss: 0.9598 - val_accuracy: 0.6841\n",
            "Epoch 17/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.7267 - accuracy: 0.7388 - val_loss: 0.8799 - val_accuracy: 0.7073\n",
            "Epoch 18/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.7188 - accuracy: 0.7414 - val_loss: 0.9697 - val_accuracy: 0.6960\n",
            "Epoch 19/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.7124 - accuracy: 0.7440 - val_loss: 0.9215 - val_accuracy: 0.6918\n",
            "\n",
            "Epoch 00019: ReduceLROnPlateau reducing learning rate to 0.0007290000503417104.\n",
            "Epoch 20/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.6922 - accuracy: 0.7531 - val_loss: 1.0867 - val_accuracy: 0.6703\n",
            "Epoch 21/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.6849 - accuracy: 0.7552 - val_loss: 0.9519 - val_accuracy: 0.6974\n",
            "\n",
            "Epoch 00021: ReduceLROnPlateau reducing learning rate to 0.0006561000715009868.\n",
            "Epoch 22/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.6688 - accuracy: 0.7602 - val_loss: 0.8120 - val_accuracy: 0.7335\n",
            "Epoch 23/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.6631 - accuracy: 0.7616 - val_loss: 1.0492 - val_accuracy: 0.6675\n",
            "Epoch 24/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.6527 - accuracy: 0.7688 - val_loss: 0.8016 - val_accuracy: 0.7316\n",
            "\n",
            "Epoch 00024: ReduceLROnPlateau reducing learning rate to 0.0005904900433961303.\n",
            "Epoch 25/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.6497 - accuracy: 0.7692 - val_loss: 0.9092 - val_accuracy: 0.7100\n",
            "Epoch 26/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.6300 - accuracy: 0.7749 - val_loss: 0.9295 - val_accuracy: 0.7025\n",
            "\n",
            "Epoch 00026: ReduceLROnPlateau reducing learning rate to 0.0005314410547725857.\n",
            "Epoch 27/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.6194 - accuracy: 0.7782 - val_loss: 0.9499 - val_accuracy: 0.7072\n",
            "Epoch 28/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.6047 - accuracy: 0.7847 - val_loss: 0.9627 - val_accuracy: 0.7047\n",
            "\n",
            "Epoch 00028: ReduceLROnPlateau reducing learning rate to 0.00047829695977270604.\n",
            "Epoch 29/45\n",
            "391/391 [==============================] - 38s 98ms/step - loss: 0.6011 - accuracy: 0.7865 - val_loss: 0.8033 - val_accuracy: 0.7452\n",
            "Epoch 30/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.6027 - accuracy: 0.7866 - val_loss: 0.7644 - val_accuracy: 0.7493\n",
            "Epoch 31/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.5957 - accuracy: 0.7877 - val_loss: 0.8148 - val_accuracy: 0.7399\n",
            "Epoch 32/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.5906 - accuracy: 0.7878 - val_loss: 0.7719 - val_accuracy: 0.7504\n",
            "Epoch 33/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.5901 - accuracy: 0.7915 - val_loss: 0.7919 - val_accuracy: 0.7438\n",
            "Epoch 34/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.5862 - accuracy: 0.7923 - val_loss: 0.8051 - val_accuracy: 0.7456\n",
            "\n",
            "Epoch 00034: ReduceLROnPlateau reducing learning rate to 0.0004304672533180565.\n",
            "Epoch 35/45\n",
            "391/391 [==============================] - 38s 98ms/step - loss: 0.5766 - accuracy: 0.7949 - val_loss: 0.8335 - val_accuracy: 0.7370\n",
            "Epoch 36/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.5676 - accuracy: 0.7968 - val_loss: 0.7146 - val_accuracy: 0.7663\n",
            "Epoch 37/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.5640 - accuracy: 0.7975 - val_loss: 0.7965 - val_accuracy: 0.7527\n",
            "Epoch 38/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.5676 - accuracy: 0.7948 - val_loss: 0.8537 - val_accuracy: 0.7417\n",
            "\n",
            "Epoch 00038: ReduceLROnPlateau reducing learning rate to 0.00038742052274756136.\n",
            "Epoch 39/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.5571 - accuracy: 0.8023 - val_loss: 0.9187 - val_accuracy: 0.7206\n",
            "Epoch 40/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.5561 - accuracy: 0.8014 - val_loss: 0.7385 - val_accuracy: 0.7638\n",
            "\n",
            "Epoch 00040: ReduceLROnPlateau reducing learning rate to 0.0003486784757114947.\n",
            "Epoch 41/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.5499 - accuracy: 0.8057 - val_loss: 0.8549 - val_accuracy: 0.7392\n",
            "Epoch 42/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.5454 - accuracy: 0.8069 - val_loss: 0.8229 - val_accuracy: 0.7448\n",
            "\n",
            "Epoch 00042: ReduceLROnPlateau reducing learning rate to 0.00031381062290165574.\n",
            "Epoch 43/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.5337 - accuracy: 0.8098 - val_loss: 0.8122 - val_accuracy: 0.7462\n",
            "Epoch 44/45\n",
            "391/391 [==============================] - 39s 99ms/step - loss: 0.5359 - accuracy: 0.8093 - val_loss: 0.7985 - val_accuracy: 0.7503\n",
            "\n",
            "Epoch 00044: ReduceLROnPlateau reducing learning rate to 0.0002824295632308349.\n",
            "Epoch 45/45\n",
            "391/391 [==============================] - 38s 98ms/step - loss: 0.5285 - accuracy: 0.8097 - val_loss: 0.8871 - val_accuracy: 0.7337\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "XflDZq78Wu_Y",
        "outputId": "4cc528f5-d413-4f32-d412-5912a1237379"
      },
      "source": [
        "import keras\n",
        "from matplotlib import pyplot as plt\n",
        "plt.plot(history.history['accuracy'])\n",
        "plt.plot(history.history['val_accuracy'])\n",
        "plt.title('model accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'val'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 30,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEWCAYAAABrDZDcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3dd3yV5fn48c+VvRNIQggJEJbsoSDi3i0OwLpn68TW2jq67Pja1o5f11dbW61Vv7bWbZ1Ytwi4UAFZMgQCCAmBDMgeZFy/P+4ncBIScgI5OUnO9X69zuuc88w7j/Jc57nHdYuqYowxJnSFBbsAxhhjgssCgTHGhDgLBMYYE+IsEBhjTIizQGCMMSHOAoExxoQ4CwQmpIjIv0Tk135uu1VEzgh0mYwJNgsExhgT4iwQGNMLiUhEsMtg+g4LBKbH8apkfiAiq0SkSkT+T0QyROR1EakQkXdEpJ/P9rNFZI2IlIrIQhEZ67PuSBH5zNvvGSCm1bnOFZEV3r4ficgkP8t4jogsF5FyEdkuIr9otf4E73il3vqrveWxIvK/IvKliJSJyAfeslNEJK+N63CG9/kXIvKciDwuIuXA1SIyXUQWe+coEJG/iUiUz/7jReRtEdktIrtE5CciMlBEqkUk1We7o0SkSEQi/fnbTd9jgcD0VBcAZwJHALOA14GfAOm4/2+/CyAiRwBPAbd6614DXhGRKO+m+BLwGNAf+I93XLx9jwQeAW4EUoF/APNEJNqP8lUBXwdSgHOAb4nIed5xh3rl/atXpinACm+/PwFTgeO8Mv0QaPLzmswBnvPO+QTQCNwGpAHHAqcDN3llSATeAd4ABgEjgfmquhNYCFzsc9yrgKdVtd7Pcpg+xgKB6an+qqq7VDUfeB/4RFWXq2ot8CJwpLfdJcCrqvq2dyP7ExCLu9HOACKBP6tqvao+ByzxOcdc4B+q+omqNqrqo0Cdt99BqepCVV2tqk2qugoXjE72Vl8OvKOqT3nnLVHVFSISBlwL3KKq+d45P1LVOj+vyWJVfck7Z42qLlPVj1W1QVW34gJZcxnOBXaq6v+qaq2qVqjqJ966R4ErAUQkHLgMFyxNiLJAYHqqXT6fa9r4nuB9HgR82bxCVZuA7UCWty5fW2ZW/NLn81Dge17VSqmIlAKDvf0OSkSOEZEFXpVKGfBN3C9zvGPktrFbGq5qqq11/tjeqgxHiMh/RWSnV130Wz/KAPAyME5EhuGeuspU9dNDLJPpAywQmN5uB+6GDoCICO4mmA8UAFnesmZDfD5vB36jqik+rzhVfcqP8z4JzAMGq2oy8ADQfJ7twIg29ikGattZVwXE+fwd4bhqJV+tUwX/HVgPjFLVJFzVmW8ZhrdVcO+p6lncU8FV2NNAyLNAYHq7Z4FzROR0r7Hze7jqnY+AxUAD8F0RiRSR84HpPvs+BHzT+3UvIhLvNQIn+nHeRGC3qtaKyHRcdVCzJ4AzRORiEYkQkVQRmeI9rTwC3C0ig0QkXESO9dokNgAx3vkjgZ8BHbVVJALlQKWIjAG+5bPuv0CmiNwqItEikigix/is/zdwNTAbCwQhzwKB6dVU9QvcL9u/4n5xzwJmqepeVd0LnI+74e3GtSe84LPvUuAG4G/AHmCTt60/bgLuEpEK4E5cQGo+7jbgbFxQ2o1rKJ7srf4+sBrXVrEb+D0Qpqpl3jEfxj3NVAEtehG14fu4AFSBC2rP+JShAlftMwvYCWwETvVZ/yGukfozVfWtLjMhSGxiGmNCk4i8Czypqg8HuywmuCwQGBOCRORo4G1cG0dFsMtjgsuqhowJMSLyKG6Mwa0WBAzYE4ExxoQ8eyIwxpgQ1+sSV6WlpWlOTk6wi2GMMb3KsmXLilW19dgUoBcGgpycHJYuXRrsYhhjTK8iIu12E7aqIWOMCXEWCIwxJsRZIDDGmBDX69oI2lJfX09eXh61tbXBLkpAxcTEkJ2dTWSkzR9ijOk6fSIQ5OXlkZiYSE5ODi0TTfYdqkpJSQl5eXkMGzYs2MUxxvQhfaJqqLa2ltTU1D4bBABEhNTU1D7/1GOM6X59IhAAfToINAuFv9EY0/36RNWQMcb0dKrKjrJaNhVWUlq9l/pGZW9DE3sbGtnb2LTve0xkOEmxESTFRJIYE0FSbCRJMZEkxUSQEhdFVETX/34PaCAQkZnAX4Bw4GFV/V2r9UNw86emeNvcoaqvBbJMgVBaWsqTTz7JTTfd1Kn9zj77bJ588klSUlICVDJjTCCU1dTzeX4Ze6r3EhkeRmS4EBkeRkRYGFERQkRYGIUVdWwqrGRjYQWbCivZVFhJ9d7GwzrvXXPG8/Vjc7rmj/ARsEDgTbV3H25yjDxgiYjMU9W1Ppv9DHhWVf8uIuOA14CcQJUpUEpLS7n//vsPCAQNDQ1ERLR/iV97rdfFPGNCTlVdA5/nl7E6v4xVee59S3GV3/sPTIphVEYCF08bzKiMBEamJ5CaEEVUeDhRES6IREWEuc9hYdQ2NFJR20B5TT3ltfWUe58rahuYPqx/QP7GQD4RTAc2qepmABF5GpgD+AYCBZK8z8m4+Wd7nTvuuIPc3FymTJlCZGQkMTEx9OvXj/Xr17NhwwbOO+88tm/fTm1tLbfccgtz584F9qfLqKys5KyzzuKEE07go48+Iisri5dffpnY2Ngg/2XG9B17G5ooqaqjuGIvxZV1FFXWUVzpvpfW7KW6rpGqvQ1U7230Xg1U1TVSUlVHc5LmQckxTMpO4cKp2UzKTiYjKYb6xiYaGpV6r3rHvTeREhfFqIwEkmI61907LiqCuKgIMpJiAnAV2hbIQJCFm0C7WR5wTKttfgG8JSLfAeKBM9o6kIjMBeYCDBkypK1N9vnlK2tYu6P80ErcjnGDkvj5rPHtrv/d737H559/zooVK1i4cCHnnHMOn3/++b5uno888gj9+/enpqaGo48+mgsuuIDU1NQWx9i4cSNPPfUUDz30EBdffDHPP/88V155ZZf+Hcb0NqpKkVfFsqnIVa/sLKtlQFI0g1JiyUqJZZD3ykiMJjxM3PZFleQWVpJbVEWut19BWds97uKiwukXF0V8dDhxURHER7f8npEUzaTsZCZmpZCe2NE00r1TsBuLLwP+par/KyLHAo+JyARvku99VPVB4EGAadOm9fgJFKZPn96ir/+9997Liy++CMD27dvZuHHjAYFg2LBhTJkyBYCpU6eydevWbiuvMcFW39jEtt3VbPZu3JuLKtlY6G7m5bUN+7ZLiI4gMzmGJVt3s6e6vsUxwgRiIsNb1MMnREcwIj2eY4enMjQ1nvTEaNISokhLjCYtPpq0xCjiooJ9Gwy+QF6BfGCwz/dsb5mv64CZAKq6WERigDSg8FBPerBf7t0lPj5+3+eFCxfyzjvvsHjxYuLi4jjllFPaHAsQHb3/l0Z4eDg1NTXdUlZjAkFVKa9toLC8lvLaeirrGqmqa6CyroHK2gaq6hoor61nS3E1m4sr2VZSTUPT/t94aQnRjBwQz+wpgxiZnsDIAYmMHJBARlL0vm7U1Xsb2FFay47Smn2v8toGhqXFM3JAAiPSW25v2hfIQLAEGCUiw3AB4FLg8lbbbANOB/4lImOBGKAogGUKiMTERCoq2p7xr6ysjH79+hEXF8f69ev5+OOPu7l0xnReU5NSWFHH1pIqviyp4suSauoamogIFyLDwtx7eBjhYUJEmLC7ai87y2vZWea9yms77CETHRHG0NQ4jhiQyMzxAxmRnsDw9HiGpyeQHNtxvXpcVAQjByQwckBCV/3ZIStggUBVG0TkZuBNXNfQR1R1jYjcBSxV1XnA94CHROQ2XMPx1doL585MTU3l+OOPZ8KECcTGxpKRkbFv3cyZM3nggQcYO3Yso0ePZsaMGUEsqQllDY1N5JfWUFRRR0VdAxW1DVTUut4old7nHWW1LW78zSLChOiIMOqblIbGJppa/SuNCBMykmLISIpmbGYSp4weQGZyDAOSokmJiyIhOpz46AjioyJIiI4gPjoiIP3hzaHpdXMWT5s2TVtPTLNu3TrGjh0bpBJ1r1D6W03nNde1by2uYmtJNV+WuPdtJVXk7alpUf3iK0xcffrA5BiGpsaTkxrnvcczNDWOQSmxhIftr2JpalIampSGpiYampSEqAjCwqwKpicTkWWqOq2tddZKYkwvVF5bz4adFfsaV3OLqtqsa0+MjmBoWhzjs5I5Z1ImQ1PjGZAYTaI3UjXRG70aFxXeqbr0sDAhKkyI6jtZakKaBQJjeriymnrWeAOaVueX8Xl+GVtLqvetjwoPIyetZV17Tlo8w9Li6RcXaY2lpkMWCIwJsuq9DRSU1VJQWktBWY377L1vKXb19c2yUmKZmJXMRdMGMzYzkRHpCWT3i2tRbWNMZ1kgMKYbNf+6X5Vfxuq8Mlbll7J994FdhdMSoshMjmX8oCQunjaYiVnJTMhKpn98VBBKbfo6CwTGBIiqsqW4ig9zS/h0y25W55W2qNIZ3D+WSVkpXDJtMFn9YslMjmVQciwZydFER4QHseQ9WGUhrHgSjrwK4lM73t74xQKBMV2ooKyGjzaV8GFuMYtzS/alNRiYFMOUwSlc5P26n5iVTD/7dd85TY3w3LWw9X346F6Y+XuYeCEEsg1k9XOwewuc9P3AnifILBAEQUJCApWVlcEuhumkmr2NrNtZTmF5HSVVdeyu3EtJlfeqrCO/tGZffX6/uEiOG5HGcSNTOW5EGjmpcdZoe7g+/LMLAif9EHLfhReuh1XPwLl3Q8rBc5Adkpo98MqtsLcCBDjpB11/jh7CAoEx7dhRWsOyL/fw2bY9fPblHtbsKD+gH35iTASp8VGkJkQzdmASV80YynEj0hgzMNH61Xel7Uvg3d/A+K/BqT+BU+6ATx+E+b+C+2bA6XfC9BsgrAur1D75hwsCw0+Bd38N/Ya5J5A+yAJBF7jjjjsYPHgw3/72twH4xS9+QUREBAsWLGDPnj3U19fz61//mjlz5gS5pKa1piZlV0UtX3qDr74sqWZLcRUrtpfuq9aJiQxjcnYKc08azpTBKQxKiSUtIZp+8ZFWl98dasvg+esgKQvO/bOropFwmPEtGH02vHo7vPEj+Pw5mHUvZIzrmnN+fD+MORcufAQe+xq89C1IGgRDjzv84/cwfW9k8et3wM7VXXvSgRPhrN+1u3r58uXceuutLFq0CIBx48bx5ptvkpycTFJSEsXFxcyYMYONGzciIodVNWQjiw+dqpJbVMWHm4r5eHMJmwor2bb7wFQKg/vHMSErmalDUpg6tD9jMhOJDLeBU21qaoQt70FiJqSP7vp6dFV44Qb4/AW45nUY0jqTvbfN6v/A6z9yN/Ajr4CTfwTJ2Yd+3vf+6J4C5i6CQVOgejf835lQXQLXz4fUEYd+7LaoQnk+FG+AQUdBbNfPWmgjiwPsyCOPpLCwkB07dlBUVES/fv0YOHAgt912G++99x5hYWHk5+eza9cuBg4cGOzihpSdZbV8uKnYvXKL2VVeB7j++OMHJXHK6HSGemkUclLjyUyOISIUbvqNDVD8BRSsgoKV7rXrc+g3FGbcBBMugIiD5N5vaoTPn3c3zOINbln/ETDmbPcrOvvorqmmWfm0u8mf+tO2gwC44DPpYhhxGiz6Ayz7p9tv2rVw4vcgYUDnzllXAYvvgyNmuiAAENcfrvgPPHwGPHEhXPfOofdaamqCPVugYMX+679zlQsy4No7Ln5s/7m7Qd97IgiSO++8k7S0NHbu3MnAgQNJSkri9ddf5/HHHycyMpKcnBwWLlxITk6OPREESEVtPWt2lLM6r3lawf3dNfvFRXLcyDSOH5HGCSPTGJIaF+TSBsmKJ+HTh6BwLTR46dAjYt1Tb8Z42PYxFK2DhAw4+gZ3M/W94TU2uCqY9/4IJZtgwDg44XaoK4f1r7qng6Z6iE93N9IjZkJ4FFQVuVd1MVQVu891FTDyTDjqKkhs4wdSSS784yTInAzfeMX/wFK6HRb93v2tEdFwzI1w3HfdzdwfH/wZ3vk5XP8uZE9tuW7bJ/DoLBh0JHz9ZYhsNYtYU6O7qW95D/Z8CbWlUFPqnlT2fS6FJm+OhbBIGDDW/Y2Zk13QeuMn7vqcezcc2XWTUx3sicACQRdZs2YNN9xwA8XFxSxatIhnn32WTZs28de//pUFCxZw2mmnsWXLFgsEXaCxScnbU71vQvC1Be7mv9lnHtnM5BgmZCUzPac/x41MZezAJGu8Ld0Gf50KqSPdr+fmm0/qyP03WVXXI+fj+2HTOxARA5Mvhek3ul+w7/0Rdm+GjAlw8g9hzCwI83mCqi1z+61/FTa85RpbfUXEuiARnwYSBvlLISzC1fVPuxaGneyO17DXVcXs2Qrf+vDQqnlKcmHh/3NdQKMT4cTb4fhbD159tbcK/jzJXZerXmh7m89fgOeucU9N5z/sft1vXghbFrkAULPHbRfb31XxxKQc+N5/OGROgvSxENGqG3FVsesmu2URTL0azvrDwZ/O/GRVQ91g/PjxVFRUkJWVRWZmJldccQWzZs1i4sSJTJs2jTFjxgS7iL1SQ2MTH2wq5rNtpd7Ug5VsLq5ir0+9fvNN/7wjs5iY7fropyX0zSkFD8ui3wPiqjjau7GKwMjT3atwvQsIK5+GZf9y6wdOhEseh9HntAwAzWKS3Q1ywgXQUAf5n0F4pLvxx6dDVHzL7UtyXVXO8idg3Tx3g5x6DZRtd4HnkscPva4/dQRc8DCccJvrXfTOL6AsH87+Y/vBYNm/3FPLyT9q/7gTzncBav4vXQBortJJynbXZfjJMOyktp9y/BGfBle+AAt+DR/c46qPLnns8No8OmBPBL1MqPyt6wrKeeGzPF5asYOiijrCBAb3j2NEupuIZGR6AiMGxDMyPZHkuM5NDh6SijfCfdPhmG/BzN92bt+qYldP3y/HVfUEYjxEfS2sewWWPgLbPnLLpl4Ds/7cNcdXhbfvdAPRjr4ezv7TgX9HfQ38ZbJr9P7GKx0f791fues6/GQYdooLPF19bda9Ai9+yz01XPiI68p6iOyJwPQKRRV1vLwinxc+y2dtQTmR4cKpowdw/lHZnDI6nZjIIHfVLN0OL97oujCmH+H/fqpQUeC6HgbLgt9AZJyrHums+DTXVTOQImNg0kXuVbgOtrzv2g66igiceZd7//Av7r/J2X9q+VTz2b+hcpe74fpzvNPv7LrytWfsLEgfA89c6bqwzv6b6xXVxSwQmG6nqhRV1LFhVyVf7Kpg464KvthVwaq8MhqblMnZyfxy9nhmTR7Us5KsvfkT+PJDWP4YfOVX/u+36hkXQM76g2u47G4FK2HNi25Ebnxa95+/swaMda+uJgJn/BIQN0oZhbP/12uTqHONxEOPh5wTuv7chyNtlOuy+saPYEhgZjjsM4FAVfv8EP7eVo3n64udFby6agcfb97NhsIKSqvr963rHx/FERkJ3HjScL52ZBajMhKDWNJ2bF7o6rDDo93jevOvS3+sfNq9v/5D1yh79PUBK2ab3v21a6A87ubuPW9PJAJn/MK9f3CPezI4525Y/jhU7IDz7g92CdsWnQBz7gvY4ftEIIiJiaGkpITU1NQ+GwxUlZKSEmJiYjreuIfILarkvysL+O+qHWwsrCRMYPLgFM6akMnojASOyEjkiIGJPb9ht7HeDVRM8frYv/Ej2LUGBk7oeN+qEteT5NibXcPoq99zvWSmXh3wYgPw5WLY+Jb7JRyT3D3n7OlE4PSfAwIf3A3aCLkLIXv6YdXB92Z9IhBkZ2eTl5dHUVFRsIsSUDExMWRnB67nQFfYUlzFa6sL+O+qAtYVlCMCR+f051dzxjNzQibpiT38pt+WJQ+7vvWXPuluFm/c4Z4O/AkEX7zqbjQTL3LVHc9cCa/c4oJBF/YRb5MqzL/LjQmYPjew5+ptmuv4JQze/5Nbdu49fTrD6MH0iUAQGRnJsGHDgl2MkNTUpKzOL+OttTt5a80uNha68RFHDknhznPHcfbETAYm956nmANUFsGC/+f63Y8+290ohh7nqodO/UnH+695yfW2yZzs9r34MXj6Mnj5ZhcMJl8auLLnznc9cM7+E0SF6AC6gxGB037mql2KN7kusyGqTwQC072q6hpY9uUe3lq7k3fWFrKzvJbwMOGYYf25/JghnDkug+x+feTG8+5dUF/lct83/1ocO8s9FRRvgrSR7e9bvdsNCjr22/v3jYxxTxZPXuKSmIVFBCajZfPTQMoQOOobXX/8vkLEjTMIcRYIzEEVVtSyZkc5a3eUs7agnHU7ytlSUoUqxEaGc/IR6XxlfAanjRlASlwP6uHTFfI/g88eczdy3+6iY851gWD9Kwe/iXzxmkslMO68lssjY+Gyp+GJi1xCNRE3AKsrrZvnegud98CBI1eNaaVPDCgzXaewopb3NxSzcEMRi3NLKK6s27ducP9YxmUmMS4zmUnZyRw7IrV7+vbv+dK99xsa+HM1a2qCR77izv2dZRCT1HL9g6e6G/gN77Z/jMcvdIndblnVdt1zXaVLYLZtMWRMdE8GEy/seASpqsvzU13iUjMnZkK4z2+6pka4fwYgcNPirs3Rb3qtoA0oE5GZwF+AcOBhVf1dq/X3AKd6X+OAAara9flXTbvqG5tYvq2UhV8UsmhDEWt2lAOQlhDNCSNTmZSdwrhBSYzNTCI5NggjeHethX/OdDfNo77uhv4nZQb+vKuegbwlcN7fDwwC4KqH5v8SyvLavnHX7HFdTmd8s/0GyOgEuPJ513Vx1bMu0dk7P3d92SdeBOPmuERpteWQv8yVp/nVnM8GXINn4iBXjuRsQF1G0IsfsyBg/BKwJwIRCQc2AGcCecAS4DJVXdvO9t8BjlTVaw92XHsi6BoNjU3866Ot3Dt/I+W1DYSHCVOH9OPk0emcfEQ64zJ7QJK20m3wf19xv4DHnO2qacIi3M31+Fsgtl9gzltb7pKzpQyB695uO6dO8Sb421TXdjDjmweuX/GkawNoK4Nle3ZvdgnSVj0LJRtdZsp+Oe7XP96/0/QxLsVz9tHuaaA83+XlKcvzXttdPp3saS5/f4j2gjEHCtYTwXRgk6pu9grxNDAHaDMQAJcBPw9geYxn5fZSfvzCatYWlHPq6HQuOXowx41MIymmB+XsqSp2Q+rrq90NLWO8SyW84LduBOjSf7r6+WNudHXuXem9P7g0wJc/3XYQANdInD7W9R5qKxCseQmSh0DWUf6ft/9wl9HzpB+4+v3V/3FBYOKF7safNdW/CUuamrxZvCwIGP8EMhBkAdt9vucBbc4sISJDgWFAmxWuIjIXmAswZEgAJqkOERW19fzvWxt4dPFWBiRG88CVR/HV8QN73iC8ugpXd16W53K+Z4x3y/sPgwseguO/63rEvPNz+OQBmHK5yw+fOcVVjRzO35O3DD7+u+vjn9XBL/lxs11a5soiSEjfv7ym1KVyPubGQyuLiJuU5FAnJmkveBnTjp7Sa+hS4DlVbWxrpao+CDwIrmqoOwvWF6gqb67Zyc/nraGwoo5vHJvD975yBIk96QmgWUMdPH2FS7176ZNt51YZONGlUt76oUum9sGf3aAtgLhUL8/+FPc+6swDUx+3p6oYnv26q28/866Otx87y6V2/uI1mOrTRXPDG25ylta9hYzpoQIZCPKBwT7fs71lbbkU+HYAyxKyNhdV8ptX1zF/fSHjMpP4x1XTmDK4h7bHNzXCC3Nd3/vz/g6jZx58+5zj4ZrXXPrgXWtgx3Jv2sUVLt1wU4MLCF9/ueMqlaZGNxlIVRFc95Z/s1llTHB1+OvmtQwEa15yuemz26yONabHCWQgWAKMEpFhuABwKXB5641EZAzQD1gcwLKEnF3ltfz5nY08u3Q70RFh/PTssVxzfE7PnY9X1SVlW/sSnPkrV93jr8hYd9P1vfHW17pf6i/MhcfPh6tePHiunXd/5QLQnPv8r5IRcU8FHz/gqoNiU9wMXbnz3TSPPa3KzZh2BOyuoKoNwM3Am8A64FlVXSMid4nIbJ9NLwWe1t42oKGHKqup5/dvrOfkPy7guWXbuWrGUN774anccNLwrgsCjQ1QsatrjtVs8d9cTp/jvuvaAA5XZIybSeqSx1w10+MXuN5AbVn3istEOfXqzuf/GTvHVQNtfMt93/AmNO51XT+N6SVsQFkfUVvfyKMfbeX+hbmU19YzZ/Igbj9zdNdO0r63ynXhXHyf67Y46y9dM3mIqpsntn8OfH1eYGZ5+s/VrvH3yufd/LXNijfBg6e4nO/XvtH5uWGbmuCece5p5JLH4anLXRXVbWus0db0KDZDWR+2qbCCFz7L5/nP8thVXsfJR6Tzw5mjGT+oC1MOVxXDpw+6V80eGHKsG+U772bX1//UnxzezbvoCyjbBifeFpjqlLGz3KxT/7kGnrgYrnzONSDXVcIzV7gUDBf/+9AmCA8Lcyknlj8OlYVu4vZp11oQML2KBYJeqLiyjnkrdvDi8nxW55cRHiacOCqNP18ygmNHpHbdiXZvcb/+lz8ODTVuYu7jb4Ehx7gc/f+9zfW5L90Gs/966DltNr7p3kd9pevK3tq4OW4i8+evcwnfLn8G5n3HjcC96kVIGdzxMdozdhYseQhe+z401lm1kOl1LBD0Eg2NTby5ZhfPLdvOexuLaWxSJmQl8T/njmP25EFdn+d/y/vw7zkufcHkS13dvW/itfBId/NPGQoLfu2qii553L8BT61teAsGjO84x87hmnC+6x304lyXi6d0m5ugZPgph3fcocdDbH9Y+7LL+zO4zeEyxvRYFgh6uLqGRp5fls8Di3LZtruazOQY5p40nPMDPaXjyqdcXfpNH7ef20cETv6BS8Xw8rfhkZlwxbPuu79qy1zSta5oIPbHpItAm9wcwmPO7ZoUxOERLgXG8sdh7GyrFjK9jgWCHqqqroEnP9nGQ+9vprCijsnZyfz0nKmcMTaD8EDnAGpqdL1fRp3pX4K3yZe47Z6+Eh4+Ay5/1v8umLnvusFgo756eGXujMmXuMbdlCFd1yYx8SJY/oR7N6aXsUDQw5RW7+VfH23lXx9tpbS6nuNGpHLPJVM4bkQ3zsec/xlUFzLI7RAAAB51SURBVMMRHQzo8jXsJLjuTZdj/7HzXK8Zf0b0bnzbTayeffShl/dQpI7o2uMNPwW+v7FlqgljegkLBD3IJ5tLuPHxZZRW13PG2AxuOnUERw0JUIbNg9nwOkh456fuGzDWNcg+8lX4/HmXNvpgmppc//uRp7fMp99bWRAwvVQf+NfXN7y4PI8fPreKwf3iePL6GYwb1EYO/O6y4U3XRfRQ0jwPPsZl5Vz6SMeBoGC5S+nQndVCxpgDWKtWkKkq97y9gdueWcnUof144abjghsESrfBrs87zvPTHhGYdo0bVLVj+cG33fg2ICE9abgxPYEFgiCqa2jk9mdX8pf5G7ngqGz+fe0xwZ/3d4PXp78z7QOtTboEImLdnAEdnSt7GsSnHfq5jDGHzQJBkOyp2stVD3/Ki8vz+f5XjuBPF00iKqKD/xzlBa5ePZA2vAH9R7iUC4cqNsVNxr76ufbz+1QWwo7PrFrImB7AAkEQbC6q5Gv3f8iKvFLuvexIbj5t1MF7BNWUulG8d491WTIDpa4Strx3eE8DzaZdC/VVbpattmx6x72POvPwz2WMOSwWCLrZR5uK+dr9H1Fe28BTNxzD7MmD2t9Y1f2q/tvRsOxfbirDj/8OFTsDU7jNC13mzENtH/CVdZSbQGbpP93f0dqGNyFhoJs8xhgTVBYIutFji7dy1SOfkpEUzUs3Hc/UoQeZ/GT3ZpdH//nrIDkLbljgZuVqqof3/hSYAm54HaKTXY+hwyUCU6+BXashf1nLdY31kLvAPQ1Yzn5jgs66j3aD+sYmfvnKGh7/eBunjxnA3/s9TtSLv3Zz8PYb5vM+3E2e8tG9bi7csEg46w9w9PUQFu4OduSV7unguO+4DKBdpanJ5fwZebrLI9QVJl0Mb9/pupL6Thqz/ROoKwtskjljjN8sEATYnqq93PTEZyzeXMI3Tx7BD07OIPyP/4bkwVC+A1Y9C/hUnYRFuCkWx86Gs34PSa2qjk76Iax4Chb9Ac67r+sKumM5VBXC6LO67pjRiTDxQlj5DHz1t/sT0m140wW5Ead23bmMMYfMAkEAbdxVwfX/XkpBaS13XzyZ84/KhvWvuaRn590POSe4KRVLt8GeLS7tc9l2GHYyHNHOr+XkLDj6OvjkATjh1sPr3eNrwxsu0+jIM7rmeM2mXeueYFY9A8fc6JZtfBuGHtdyghhjTNBYIAiQRRuK+PYTnxETGc7TN87Ynypi6/sQEbM/t05kjEvv7JviuSMn3A7LHoUFv4WLOuir768Nr8PgGf5N2t4ZmZNh0FGuemj6XBfoitZ1zcxmxpguYY3FAbCtpJqbHl/G4P5xzLv5+Jb5gra8D4OnH9psWM0S0mHGN2HNC7Bz9eEXuCzfHeeIAPXpn3YNFK2HbR/vn9vX2geM6TEsEHSxxible/9ZQZgID39jGoNSYvevrN7tetHknHT4JzruO66Hz7u/OfxjbXjDvXdl+4CvCRdAdJJ7KtjwlmsYTx0ZmHMZYzrNAkEXe+j9zSzZuodfzhlPlm8QAFctBDDsxMM/UWw/OP47rkpn+5LDO9aGN6FfDqR1onqqM6LiXdqJtS/DlkXuycO6jRrTY1gg6ELrCsq5+60NzBw/kK8dmXXgBlveh8g4V2feFY75FsSlHd5o473V3s35rMDenKdd4+bzbai10cTG9DAWCLpIXUMjtz2zgqTYSH57/sS2U0Zsfd8N1jrUSd5bi06AE293N/It7x3aMbYscjfnQLUPNMsY71JUR8bB0BMCey5jTKcENBCIyEwR+UJENonIHe1sc7GIrBWRNSLyZCDLE0j3vL2R9Tsr+MOFE+kf38aNvrLQNZh2RbWQr2nXQeIgmP+rtlM5dOSL1yEq0U3AHmhz7oNLn3Q9pYwxPUbAAoGIhAP3AWcB44DLRGRcq21GAT8GjlfV8cCtgSpPIC3Zupt/vJfLZdMHc9qYjLY3am4f6IqGYl+RMW4C+bxPXa6gzlB17QMjT+u6p5SDSRtlg8iM6YEC+UQwHdikqptVdS/wNDCn1TY3APep6h4AVS0MYHkCorKugdufXUF2v1h+es649jfc8r775R2IJGuTL3dTS279oHP7FayEyp2ufcAYE7ICGQiygO0+3/O8Zb6OAI4QkQ9F5GMRaTPtpYjMFZGlIrK0qKgoQMU9NL95dS15e2q4++IpJEQfZHze1vfdaNpAzM0bGQPpo2Hnqs7tl+f1Nurq6ipjTK8S7MbiCGAUcApwGfCQiKS03khVH1TVaao6LT2950wQ/u76XTz16XZuPGkER+ccZERu+Q4o2RTYG+7ASVDQyUBQsML1Okpqo4eTMSZkBDIQ5AODfb5ne8t85QHzVLVeVbcAG3CBocfb29DEnS+vYXRGIred2UGRtzS3DwQwEGROctU8lZ2oXdux0lVVWZ9+Y0JaIAPBEmCUiAwTkSjgUmBeq21ewj0NICJpuKqizQEsU5d5Zsk28vbU8JNzxhIdEX7wjbe+BzEpbqKWQBk4yb37Wz1UX+ty/tjEMMaEvIAFAlVtAG4G3gTWAc+q6hoRuUtEZnubvQmUiMhaYAHwA1UtCVSZukr13gbufXcT04f156RRfky8vuV9l2k0rIOAcTgGTnDv/lYPFa516a4tEBgT8gKafVRVXwNea7XsTp/PCtzuvXqNRz/6kqKKOv5+xVEHn2sYXIrp0i9hxk2BLVRsP0gZ4v8TQcFK9z5oSuDKZIzpFYLdWNzrlNXU88CiXE4bM4BpB2sgbralC/MLdaQzDcYFK91saCldOMuZMaZXskDQSQ+9t5mymnq+9xU/E7RteQ/iUiF9bGALBq6aZ/dmqKvoeNuCFdZQbIwBLBB0SlFFHY98uIVzJ2UyflByxzuouvEDOSdAWDdc6oGTAIVdaw6+XWO928baB4wxWCDolPsWbKKuoYnbz/TzaWD3ZijPh2FdnFaiPc29kjqqHipaD417IdPaB4wxFgj8lrenmic/2cZFU7MZnp7g306Byi/UnqRBrhpq58qDb9fcUGyBwBiDn4FARF4QkXNEJGQDx73zN4LAd0/vxHi3Le9DQkbXTTDfERFXPdTR9JU7VkBUAvQf3j3lMsb0aP7e2O8HLgc2isjvRGR0AMvU42wqrOS5ZXlcNWNoy6knD2Zf+8CJ3dsgmzkJCte5doD2FKx0AaM72i2MMT2eX3cCVX1HVa8AjgK2Au+IyEcico2IRAaygD3BPW9vIDYynJtOGeH/TsUboXJX9yd0GzjJ1f8XrW97fVOje2KwhmJjjMfvn4QikgpcDVwPLAf+ggsMbwekZD3E5/llvLq6gOtOHE5qQrT/O25Z5N4DmV+oLc2pJtprMC7eCA01FgiMMfv4NbJYRF4ERgOPAbNUtcBb9YyILA1U4XqCf7y3meTYSK4/cVjndtz6vsvq2d318Kkj3HSQ7bUT2IhiY0wr/qaYuFdVF7S1QlWndWF5epSavY28s3YX5x+VRVJMJ2rAGhtg8yIYfXb3D9gKC4eMCe2nmihYARGxkNorkrwaY7qBv1VD43znCRCRfiIS4OQ5wbfwi0Jq6+s5Z3wn50DIXwa1pTDqzMAUrCMDJ7ongqamA9cVrHQJ6gIxQY4xplfyNxDcoKqlzV+8qSVvCEyReo5XVxfwm9inOHb+BZ2bGH7T2yBhwZufN3MS1JVD6daWy5uaXNuBtQ8YY3z4GwjCxSfNpjcxfTfMdh48NXsbeX9dHl+TRciuz2HX5/7vvPFtyJ7uMoIGQ3sNxnu2wN4KG0hmjGnB30DwBq5h+HQROR14ylvWZy38opAZjcuIbap0C9a94t+OlYWuHn7UGYErXEcGjHOT2bduMN6x3L3bE4Exxoe/geBHuIljvuW95gM/DFSheoJXVxdwUdTHaHw6DJ7hfyDY9I57Hxmk9gHwJrMfc2CDccFKCI9y64wxxuPvgLImVf27ql7ovf6hqo2BLlyw1NY38un6rZwsnyETLoDx57kZvUpyO95549sQP2B/9UywDJx4YNVQwUr3tBDRp2v1jDGd5G+uoVEi8pyIrBWRzc2vQBcuWBZ+UcjJjYuJ1L0w8SIYc65b0dFTQWMD5L7regsFO31D68nsVV0gsGohY0wr/t6t/gn8HWgATgX+DTweqEIF26urd3Jh5GK03zDImgopg2HQkR0HguZuoyOD2D7QrPVk9qXbXNlsIJkxphV/A0Gsqs4HRFW/VNVfAOcErljBU1vfyMp16zmaz5GJF+0fEDZ2FuQvhbL89ncOdrdRX63nJihY4d7ticAY04q/gaDOS0G9UURuFpGvAX4m5e9dFn5RyJmNHxCGwqSL968YO9u9r3+1/Z2D3W3UV2xKy8nsC1a6nkQDxge3XMaYHsffQHALEAd8F5gKXAl8I1CFCqZXV+/k/MjFaOaUlvMIpI1yvW3WzWt7x4pdwe822prvZPYFK2HAWNejyBhjfHQYCLzBY5eoaqWq5qnqNap6gap+3A3l61a19Y1sWrec8eS6aqHWxpwLX34IVSUHrsud796D2W20Nd/J7HessGohY0ybOgwEXjfRE7qhLEG38Isivtr0PorAhAsO3GDsLNAm2PD6ges2vu1mIwt2t1FfzZPZb3wbqottRLExpk3+Vg0tF5F5InKViJzf/OpoJxGZKSJfiMgmEbmjjfVXi0iRiKzwXtd3+i/oQq+t2sH5ER+hOSdCUuaBG2ROhuQhB/Yeau42OvKM4Hcb9ZXpBaXlj3nf7YnAGHMgf1NQxgAlwGk+yxR4ob0dvCql+4AzgTxgiYjMU9W1rTZ9RlVv9r/IgVFb38iu9R8xJGwnTPpZ2xuJuKeCJQ+56pboRLe8J3Ub9ZWY6Sazz10AiMs6aowxrfgVCFT1mkM49nRgk6puBhCRp4E5QOtA0CM0Vws1RUQSNnZW+xuOnQUf3wcb39pffdSTuo36ap7MfvMCSBsNUfHBLpExpgfyd4ayf+KeAFpQ1WsPslsWsN3nex5wTBvbXSAiJwEbgNtUdXvrDURkLjAXYMiQIf4UudNeX5XH/0QshiNmuq6X7Rk83aWQWPfK/kDQk7qNtpbpBQKrFjLGtMPfCu3/Aq96r/lAElDZBed/BchR1Um4uY8fbWsjVX1QVaep6rT09E5OEuOH2vpGKte/SxplhE1qo7eQr7BwGHOOu/nX1/bMbqO+mhuvbUSxMaYd/lYNPe/7XUSeAj7oYLd8YLDP92xvme9xffthPgz8wZ/ydLVFG4qY2fQ+DdEJRIz6asc7jJ0Fy/4JmxdCzW63bNRXAlrGQ5Zzonsa6KnlM8YE3aHOVzgKGNDBNkuAUSIyDBcALgUu991ARDJVtcD7OhtYd4jlOSyfbMjn9vAlhI07378BVzknQnSyqx6qr+553UZ9JWbAje8FuxTGmB7M3zaCClq2EezEzVHQLlVtEJGbgTeBcOARVV0jIncBS1V1HvBdEZmNS2a3G7i683/C4Uve/i4JUgOTL+54Y3BpnEfPhC9edVk9x5zT/ZPUG2NMF/G3aijxUA6uqq8Br7VadqfP5x8DPz6UY3eljNLl1Eks0Tkn+r/T2Fmw6hn3uad1GzXGmE7wdz6Cr4lIss/3FBE5L3DF6j6VdQ1E15dSG93fNQT7a8TpEBHbM7uNGmNMJ/jba+jnqlrW/EVVS4GfB6ZI3WtzUSX9qUBj+3dux6g4l5109Nk9s9uoMcb4yd/G4rYCxqE2NPcom4uqGCaVRCYcwviE2fd2fYGMMaab+ftEsFRE7haREd7rbmBZIAvWXXKLKuknlcQkd9QJyhhj+iZ/A8F3gL3AM8DTQC3w7UAVqjvlFlWSKhWEx6cGuyjGGBMU/vYaqgIOyB7aF2wrLCWeGojrZBuBMcb0Ef72GnpbRFJ8vvcTkTcDV6zu0dik7C4pdF+swdcYE6L8rRpK83oKAaCqe+h4ZHGPl7+nhoRGrzNUnFUNGWNCk7+BoElE9nWrEZEc2shG2tvkFlfSrzl3nlUNGWNClL9dQH8KfCAiiwABTsRLC92b5RZWkiIV7ktnxxEYY0wf4W9j8RsiMg13818OvATUBLJg3SG3qIqs6Br3bGNVQ8aYEOVv0rnrgVtwqaRXADOAxbScurLXyS2qZE58nZtZwaqGjDEhyt82gluAo4EvVfVU4Eig9OC79Hybi6oYHF3rcgZFxga7OMYYExT+BoJaVa0FEJFoVV0PjA5csQKvrLqe4so6MiKrrFrIGBPS/G0szvPGEbwEvC0ie4AvA1eswMstdr2F+kslxNkYAmNM6PK3sfhr3sdfiMgCIBl4I2Cl6gabi6oASNRy6zFkjAlpnc4gqqqLAlGQ7pZbVElkuBC9twxShwa7OMYYEzT+thH0ObmFlQxNjUdqSqzHkDEmpIVuICiqZGRaDNSUWtWQMSakhWQgqG9sYtvuasb1A1B7IjDGhLSQDATbd1dT36gckbTXLbDuo8aYEBaSgSDX6zE0PK7OLbCqIWNMCAvJQLC5yI0hyIrx0iXZOAJjTAgLyUCQW1RJWkI08Q02F4ExxgQ0EIjITBH5QkQ2iUi7U12KyAUiol6G04DLLapiRHo8VO92C6xqyBgTwgIWCEQkHLgPOAsYB1wmIuPa2C4Rl9Tuk0CVpbXNRZWMGJAANbshLAKiE7vr1MYY0+ME8olgOrBJVTer6l7gaWBOG9v9Cvg9UBvAsuyzu2ove6rrGZGe4J4I4lJBpDtObYwxPVIgA0EWsN3ne563bB8ROQoYrKqvHuxAIjJXRJaKyNKioqLDKlSu11A8PD0eqkusWsgYE/KC1lgsImHA3cD3OtpWVR9U1WmqOi09Pf2wzptb6ALByPQEqNljg8mMMSEvkIEgHxjs8z3bW9YsEZgALBSRrbhZz+YFusF4c3EV0RFhDEqJ9aqGLBAYY0JbIAPBEmCUiAwTkSjgUmBe80pVLVPVNFXNUdUc4GNgtqouDWCZyC2sZFhaPOFhYlVDxhhDAAOBqjYANwNvAuuAZ1V1jYjcJSKzA3XejuQWVbqGYlXXa8ieCIwxIa7T8xF0hqq+BrzWatmd7Wx7SiDLAlDX0Mj2PTXMnjwI6iqgqcEGkxljQl5IjSzeVlJNY5O6MQTVJW6hVQ0ZY0JcSAWCfV1H07zBZGBVQ8aYkBdigcDLOpoeD9V73EJ7IjDGhLgQCwSVZCbHEB8dsb9qyNoIjDEhLsQCQZV7GgCrGjLGGE/IBAJVZXOh13UUvMyjAjHJQS2XMcYEW8gEgqLKOirqGnwCQQnE9oOw8OAWzBhjgixkAkFuoWso3hcIbDCZMcYAoRQIfLOOgqsash5DxhgTOoEgIymGmeMHMjApxi2o2W09howxhgCnmOhJzhyXwZnjMvYvqN4NAycFr0DGGNNDhMwTwQGqd7vGYmOMCXGhGQj2VkNDjVUNGWMMoRoIbDCZMcbsE5qBoNoLBNZryBhjQjQQ7HsisKohY4wJzUCwL+GcPREYY0yIBgKrGjLGmGahGQhqmucisO6jxhgTmoGgugSikyAiKtglMcaYoAvRQGCDyYwxplloBgLLPGqMMfuEZiCoLrGuo8YY4wloIBCRmSLyhYhsEpE72lj/TRFZLSIrROQDERkXyPLsYymojTFmn4AFAhEJB+4DzgLGAZe1caN/UlUnquoU4A/A3YEqTws1e6xqyBhjPIF8IpgObFLVzaq6F3gamOO7gaqW+3yNBzSA5XEa66Gu3KqGjDHGE8j5CLKA7T7f84BjWm8kIt8GbgeigNPaOpCIzAXmAgwZMuTwSrVvMJn1GjLGGOgBjcWqep+qjgB+BPysnW0eVNVpqjotPT398E5omUeNMaaFQAaCfGCwz/dsb1l7ngbOC2B5nGpLOGeMMb4CGQiWAKNEZJiIRAGXAvN8NxCRUT5fzwE2BrA8TnPCOes1ZIwxQADbCFS1QURuBt4EwoFHVHWNiNwFLFXVecDNInIGUA/sAb4RqPLsY1VDxhjTQkAnr1fV14DXWi270+fzLYE8f5ss86gxxrQQ9MbiblddAhGxEBUX7JIYY0yPEHqBwAaTGWNMC6EXCCy9hDHGtBCCgaDEngiMMcZH6AUCS0FtjDEthF4gsKohY4xpIbQCQVMj1JbaqGJjjPERWoGgtgy0yaqGjDHGR2gFAhtMZowxBwitQFBjCeeMMaa10AoEzQnn4mwuAmOMaRZigcCqhowxprXQCgRWNWSMMQcIrUBQXQJhERCdGOySGGNMjxFigcAbTCYS7JIYY0yPEVqBoGa3VQsZY0wroRUIqi3PkDHGtBZ6gSDWuo4aY4yv0AoElnnUGGMOEDqBQNWbi8DaCIwxxlfoBIK6CmhqsMFkxhjTSugEgn2DySwQGGOMr9AJBNU2qtgYY9oSeoHAqoaMMaaFgAYCEZkpIl+IyCYRuaON9beLyFoRWSUi80VkaMAKY1VDxhjTpoAFAhEJB+4DzgLGAZeJyLhWmy0HpqnqJOA54A+BKo9VDRljTNsC+UQwHdikqptVdS/wNDDHdwNVXaCq1d7Xj4HsgJUmZTCMORdikgN2CmOM6Y0iAnjsLGC7z/c84JiDbH8d8HpbK0RkLjAXYMiQIYdWmjHnuJcxxpgWekRjsYhcCUwD/tjWelV9UFWnqeq09PT07i2cMcb0cYF8IsgHBvt8z/aWtSAiZwA/BU5W1boAlscYY0wbAvlEsAQYJSLDRCQKuBSY57uBiBwJ/AOYraqFASyLMcaYdgQsEKhqA3Az8CawDnhWVdeIyF0iMtvb7I9AAvAfEVkhIvPaOZwxxpgACWTVEKr6GvBaq2V3+nw+I5DnN8YY07Ee0VhsjDEmeCwQGGNMiLNAYIwxIU5UNdhl6BQRKQK+PMTd04DiLixOX2DXpG12XQ5k1+RAvemaDFXVNgdi9bpAcDhEZKmqTgt2OXoSuyZts+tyILsmB+or18SqhowxJsRZIDDGmBAXaoHgwWAXoAeya9I2uy4HsmtyoD5xTUKqjcAYY8yBQu2JwBhjTCsWCIwxJsSFTCDoaP7kUCAij4hIoYh87rOsv4i8LSIbvfd+wSxjdxORwSKywJs7e42I3OItD9nrIiIxIvKpiKz0rskvveXDROQT79/QM15W4ZAiIuEislxE/ut97xPXJCQCgZ/zJ4eCfwEzWy27A5ivqqOA+d73UNIAfE9VxwEzgG97/2+E8nWpA05T1cnAFGCmiMwAfg/co6ojgT24WQVDzS24bMrN+sQ1CYlAgB/zJ4cCVX0P2N1q8RzgUe/zo8B53VqoIFPVAlX9zPtcgftHnkUIXxd1Kr2vkd5LgdOA57zlIXVNAEQkGzgHeNj7LvSRaxIqgaCt+ZOzglSWniZDVQu8zzuBjGAWJphEJAc4EviEEL8uXhXICqAQeBvIBUq9eUYgNP8N/Rn4IdDkfU+lj1yTUAkExg/q+hKHZH9iEUkAngduVdVy33WheF1UtVFVp+CmmJ0OjAlykYJKRM4FClV1WbDLEggBnZimB/Fr/uQQtUtEMlW1QEQycb8AQ4qIROKCwBOq+oK3OOSvC4CqlorIAuBYIEVEIrxfwKH2b+h4YLaInA3EAEnAX+gj1yRUngg6nD85hM0DvuF9/gbwchDL0u28et7/A9ap6t0+q0L2uohIuoikeJ9jgTNxbScLgAu9zULqmqjqj1U1W1VzcPePd1X1CvrINQmZkcVeJP8zEA48oqq/CXKRup2IPAWcgkuduwv4OfAS8CwwBJfe+2JVbd2g3GeJyAnA+8Bq9tf9/gTXThCS10VEJuEaPsNxPxafVdW7RGQ4rqNFf2A5cKWq1gWvpMEhIqcA31fVc/vKNQmZQGCMMaZtoVI1ZIwxph0WCIwxJsRZIDDGmBBngcAYY0KcBQJjjAlxFgiM6UYickpz5kpjegoLBMYYE+IsEBjTBhG50svJv0JE/uElYasUkXu8HP3zRSTd23aKiHwsIqtE5MXmuQtEZKSIvOPl9f9MREZ4h08QkedEZL2IPOGNbjYmaCwQGNOKiIwFLgGO9xKvNQJXAPHAUlUdDyzCjcwG+DfwI1WdhBuh3Lz8CeA+L6//cUBzNtMjgVtxc2MMx+WxMSZoQiXpnDGdcTowFVji/ViPxSWdawKe8bZ5HHhBRJKBFFVd5C1/FPiPiCQCWar6IoCq1gJ4x/tUVfO87yuAHOCDwP9ZxrTNAoExBxLgUVX9cYuFIv/TartDzc/im4umEft3aILMqoaMOdB84EIRGQD75i8eivv30pxp8nLgA1UtA/aIyIne8quARd5sZ3kicp53jGgRievWv8IYP9kvEWNaUdW1IvIz4C0RCQPqgW8DVcB0b10hrh0BXPrhB7wb/WbgGm/5VcA/ROQu7xgXdeOfYYzfLPuoMX4SkUpVTQh2OYzpalY1ZIwxIc6eCIwxJsTZE4ExxoQ4CwTGGBPiLBAYY0yIs0BgjDEhzgKBMcaEuP8P9UVcHlbbKXIAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "X-OboZ4gW4_7",
        "outputId": "ebce57fa-e62d-48cf-ef7b-3f3cbe2d3fc2"
      },
      "source": [
        "import keras\n",
        "from matplotlib import pyplot as plt\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title('model loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'val'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 31,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEWCAYAAABrDZDcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3dd3hc1Zn48e+r3iWruchFAtywDXaQjWnGIUBMZ+mhhBACIWFDSVuS32ZTluwmm4QQAgm9BgymhE4cijEmGGy527hj2ZKbitX7aM7vj3NHdSSNymgk3ffzPPPMzL137hxd8H3ntPeIMQallFLuFRbqAiillAotDQRKKeVyGgiUUsrlNBAopZTLaSBQSimX00CglFIup4FAqQCJyJMicneAx+aLyJn9PY9Sg0EDgVJKuZwGAqWUcjkNBGpEcZpkfiQiG0WkRkQeE5HRIvKOiFSJyHsiMqrN8ReKyBYRKReRD0Vkept9c0RkrfO5F4CYDt91voisdz77iYgc18cy3yQiu0TkiIi8LiLjnO0iIn8UkSIRqRSRTSIy09l3roh87pRtv4j8sE8XTCk0EKiR6VLgLGAKcAHwDvBTIAP7//xtACIyBVgM3OHsext4Q0SiRCQKeBV4BkgFXnTOi/PZOcDjwLeBNOAh4HURie5NQUXkDOB/gSuAscBe4Hln99nAAufvSHaOKXX2PQZ82xiTCMwEPujN9yrVlgYCNRL92Rhz2BizH1gBfGaMWWeMqQf+DsxxjrsSeMsY864xpgn4PRALnAzMByKBe40xTcaYl4DVbb7jZuAhY8xnxphmY8xTQIPzud64BnjcGLPWGNMA/AQ4SUSygSYgEZgGiDFmqzHmoPO5JuBYEUkyxpQZY9b28nuVaqGBQI1Eh9u8rvPzPsF5PQ77CxwAY4wXKACynH37TfusjHvbvJ4E/MBpFioXkXJggvO53uhYhmrsr/4sY8wHwP3AA0CRiDwsIknOoZcC5wJ7RWS5iJzUy+9VqoUGAuVmB7A3dMC2yWNv5vuBg0CWs81nYpvXBcCvjTEpbR5xxpjF/SxDPLapaT+AMeY+Y8wJwLHYJqIfOdtXG2MuAjKxTVhLevm9SrXQQKDcbAlwnoh8RUQigR9gm3c+AVYCHuA2EYkUkUuAeW0++whwi4ic6HTqxovIeSKS2MsyLAZuEJHZTv/C/2CbsvJFZK5z/kigBqgHvE4fxjUikuw0aVUC3n5cB+VyGgiUaxljtgPXAn8GSrAdyxcYYxqNMY3AJcA3gCPY/oRX2nw2D7gJ23RTBuxyju1tGd4Dfga8jK2FHA1c5exOwgacMmzzUSnwO2ffdUC+iFQCt2D7GpTqE9GFaZRSyt20RqCUUi6ngUAppVxOA4FSSrmcBgKllHK5iFAXoLfS09NNdnZ2qIuhlFLDypo1a0qMMRn+9g27QJCdnU1eXl6oi6GUUsOKiOztap82DSmllMtpIFBKKZfTQKCUUi437PoI/GlqaqKwsJD6+vpQFyXoYmJiGD9+PJGRkaEuilJqhBgRgaCwsJDExESys7NpnyxyZDHGUFpaSmFhITk5OaEujlJqhBgRTUP19fWkpaWN6CAAICKkpaW5ouajlBo8IyIQACM+CPi45e9USg2eERMIetRUB5UHoNkT6pIopdSQ4p5A4GmA6sPgbRzwU5eXl/OXv/yl158799xzKS8vH/DyKKVUb7gnEIjzp3oHfiGnrgKBx9N97ePtt98mJSVlwMujlFK9MSJGDQUkLNw+m+YBP/Vdd93F7t27mT17NpGRkcTExDBq1Ci2bdvGjh07uPjiiykoKKC+vp7bb7+dm2++GWhNl1FdXc0555zDqaeeyieffEJWVhavvfYasbGxA15WpZTqaMQFgl++sYXPD1R23mG80FQLEdUQ1rs/+9hxSfz8ghld7v/Nb37D5s2bWb9+PR9++CHnnXcemzdvbhni+fjjj5OamkpdXR1z587l0ksvJS0trd05du7cyeLFi3nkkUe44oorePnll7n22mt7VU6llOqLERcIutQy2ib4S3POmzev3Tj/++67j7///e8AFBQUsHPnzk6BICcnh9mzZwNwwgknkJ+fH/RyKqUUjMBA0OUvd68XDm2AxLGQOCaoZYiPj295/eGHH/Lee++xcuVK4uLiWLhwod95ANHR0S2vw8PDqaurC2oZlVLKx0WdxQJIUPoIEhMTqaqq8ruvoqKCUaNGERcXx7Zt2/j0008H/PuVUqo/RlyNoEsitsM4CKOG0tLSOOWUU5g5cyaxsbGMHj26Zd+iRYt48MEHmT59OlOnTmX+/PkD/v1KKdUfYkzw28wHUm5urum4MM3WrVuZPn16zx8+vAUi4yE1OziFGyQB/71KKeUQkTXGmFx/+9zTNAS2RhCEpiGllBrO3BUIJBy8GgiUUqotdwUCrREopVQn7goEEm4nlimllGrhrkAQpk1DSinVkbsCgYTZpqFhNlJKKaWCyV2BoCXxXGibhxISEkL6/Uop1Za7AoE4gUCbh5RSqoV7ZhZD0FJR33XXXUyYMIFbb70VgF/84hdERESwbNkyysrKaGpq4u677+aiiy4a0O9VSqmBMPICwTt3waFN/vd5PeCpg8i41tpBIMbMgnN+0+XuK6+8kjvuuKMlECxZsoSlS5dy2223kZSURElJCfPnz+fCCy/UNYeVUkPOyAsE3fHdhI2BAbwfz5kzh6KiIg4cOEBxcTGjRo1izJgx3HnnnXz00UeEhYWxf/9+Dh8+zJgxwc18qpRSvTXyAkE3v9xpqoPibZAyCeJSB/RrL7/8cl566SUOHTrElVdeybPPPktxcTFr1qwhMjKS7Oxsv+mnlVIq1EZeIOhOEEcNXXnlldx0002UlJSwfPlylixZQmZmJpGRkSxbtoy9e/cO+HcqpdRAcFcgCOKooRkzZlBVVUVWVhZjx47lmmuu4YILLmDWrFnk5uYybdq0Af9OpZQaCC4LBM5o2SDlG9q0qbWTOj09nZUrV/o9rrq6Oijfr5RSfeGyeQTi5BvSeQRKKeXjrkAAQVulTCmlhqsREwgCXmltmK9JMNxWlFNKDX0jIhDExMRQWloa2E0yLGzYNg0ZYygtLSUmJibURVFKjSAjorN4/PjxFBYWUlxc3PPBNcW2RlDsCX7BgiAmJobx48eHuhhKqREkaIFARCYATwOjAQM8bIz5U4djFgKvAXucTa8YY37V2++KjIwkJycnsINfvgcKPoM7Nvb2a5RSakQKZo3AA/zAGLNWRBKBNSLyrjHm8w7HrTDGnB/EcrQXnQgNlYP2dUopNdQFrY/AGHPQGLPWeV0FbAWygvV9AYtJgoYqXZxGKaUcg9JZLCLZwBzgMz+7TxKRDSLyjojM6OLzN4tInojkBdQP0J3oJJuFtKmuf+dRSqkRIuiBQEQSgJeBO4wxHdtk1gKTjDHHA38GXvV3DmPMw8aYXGNMbkZGRv8KFJNkn7V5SCmlgCAHAhGJxAaBZ40xr3Tcb4ypNMZUO6/fBiJFJD2YZSI62T7XayBQSikIYiAQuwLLY8BWY8w9XRwzxjkOEZnnlKc0WGUCtEaglFIdBHPU0CnAdcAmEVnvbPspMBHAGPMgcBnwHRHxAHXAVSbYU2ejE+1zfUVQv0YppYaLoAUCY8zH9LAOmDHmfuD+YJXBr2hfjaBqUL9WKaWGqhGRYqJXtGlIKaXacV8g8NUItLNYKaUAVwYCp49AawRKKQW4MRCEhUNUotYIlFLK4b5AAJpvSCml2nBnIIhJ0kCglFIOdwaC6CRtGlJKKYc7A4HWCJRSqoU7A4HWCJRSqoU7A4HWCJRSqoU7A0G0Dh9VSikflwaCZGhuAE9DqEuilFIh585AEKOJ55RSysedgaAl35CmolZKKXcGAs1AqpRSLdwZCDQDqVJKtXBpINAMpEop5ePOQKCdxUop1cKdgSA62T5r05BSSrk0EGhnsVJKtXBnIAiPhIhYHT6qlFK4NRCA5htSSimHewOB5htSSinA1YEgSUcNKaUUbg4E2jSklFKAmwOBLk6jlFKAmwOB1giUUgpwcyCITtYagVJK4eZAEJMETTXQ7Al1SZRSKqTcGwh8iecadeSQUsrdXBwINBW1UkqBmwOB5htSSinAzYFAawRKKQUEMRCIyAQRWSYin4vIFhG53c8xIiL3icguEdkoIl8KVnk60RqBUkoBEBHEc3uAHxhj1opIIrBGRN41xnze5phzgMnO40Tgr85z8OmaBEopBQSxRmCMOWiMWeu8rgK2AlkdDrsIeNpYnwIpIjI2WGVqR5erVEopYJD6CEQkG5gDfNZhVxZQ0OZ9IZ2DBSJys4jkiUhecXHxwBRKm4aUUgoYhEAgIgnAy8Adxpg+3XWNMQ8bY3KNMbkZGRkDU7CIGAiL1KYhpZTrBTUQiEgkNgg8a4x5xc8h+4EJbd6Pd7YFn4jmG1JKKYI7akiAx4Ctxph7ujjsdeDrzuih+UCFMeZgsMrUiWYgVUqpoI4aOgW4DtgkIuudbT8FJgIYYx4E3gbOBXYBtcANQSxPZ1ojUEqp4AUCY8zHgPRwjAFuDVYZeqSrlCmllItnFoM2DSmlFG4PBG5oGjIGVj0CVYdDXRKl1BDl7kDghhpBRQG8/UPY8FyoS6KUGqLcHQh8NQKvN9QlCZ7yffa5ojC05VBKDVnuDgTRSYCBxupQlyR4yp2J2xoIlFJdcHkg8OUbGsEjh3w1gvKC7o9TSrmWuwOBG/INVWjTkFKqe+4OBG5IRe2rCTRUQH1FaMuilBqS3B0IXFEjKIDwaOf14KRxUkoNL+4OBC3LVY7QX8per20SGp9r32vzkFLKD3cHgpFeI6g+DM2NMOlk+97XX6CUUm24OxCM9FFDFU7/QFauXXtBawRKKT/cHQiiEkDCRm5nsW/o6KhJkDROA4FSyq+AAoGI3C4iSc66AY+JyFoROTvYhQs6EVsrGKlNQ75AkDzBPjQQKKX8CLRG8E1nmcmzgVHYdQZ+E7RSDabo5JFbI6gogNhUiE6AFA0ESin/Ag0EvnUFzgWeMcZsoYe1BoaNkZyBtLzABgCA5PFQeQCaPaEtk1JqyAk0EKwRkX9iA8FSEUkEhlWmtnX7yrj12bVU1Te13zGSM5CW77NNQmADgWmGqsFbCVQpNTwEGghuBO4C5hpjaoFIBntZyX6qaWjmrU0HWbevvP2OkdpHYIxtGkqZaN8nj7fP2jyklOog0EBwErDdGFMuItcC/wkMq1lYcyamEB4mrM4/0n7HSG0aqj0CTbVtAoFTM9BAoJTqINBA8FegVkSOB34A7AaeDlqpgiA+OoIZ45JYtadDIBipTUMVbUYMQZsagWYhVUq1F2gg8DgLzV8E3G+MeQBIDF6xgmNudirrC8pp8DS3bvTVCIwJXcGCwTd01NdZHBVvRxBpIFBKdRBoIKgSkZ9gh42+JSJh2H6CYWVudioNHi+b97dp1YpOAq8HmupCV7Bg8GUd9TUNga0VaNOQUqqDQAPBlUADdj7BIWA88LuglSpI5maPAmB1flnrxpGab6iiAKISISaldZtOKlNK+RFQIHBu/s8CySJyPlBvjBlWfQQAaQnRHJURz+q2/QS+DKQjLd+Qbw6BtJnuoTUCpZQfgaaYuAJYBVwOXAF8JiKXBbNgwTIvO5W8vWV4vU6fQEsq6hFWI2g7h8AnZYKt+YzUtNtKqT4JtGno/2HnEFxvjPk6MA/4WfCKFTxzs1OpqGtiR5FTA2hpGhphN8eKfe37B6B15JCuX6yUaiPQQBBmjClq8760F58dUublpAK0Ng+NxBpBvfOrP6VDjUDnEiil/Aj0Zv4PEVkqIt8QkW8AbwFvB69YwTN+VCxjkmJY5esw7k9nsbcZPA0DV7iB4hsi2rFpSOcSKKX8CLSz+EfAw8BxzuNhY8x/BLNgwSIi5GaPYvWeIxhj+lcjeP9X8MC8oZfIrWUOQYemofhMCI/SGoFSqp2IQA80xrwMvBzEsgyaeTmpvLnxIIVldUxI6eMqZcbAxiVQdQD2fAjHnDng5ewzf3MIAMLCIClLA4FSqp1uawQiUiUilX4eVSIybBvV52Y7/QT5RyAs3K5U1tumoQNrbRAA2PDCAJewnyr2QUQMxGd03qdDSJVSHXQbCIwxicaYJD+PRGNM0mAVcqBNHZ1IUkxEawK6vuQb2vYWSDgcexFsexMaqge+oH1VXmBv+OJnyYjkCdpHoJRqZ1iO/OmvsDAhNzu1NQFdTFLvh49uewsmnQwnfsdm+dz25sAXtK/8zSHwSR5v1yRobvK/XynlOkELBCLyuIgUicjmLvYvFJEKEVnvPP4rWGXxJzd7FLuLayitbuh9jaBkFxRvg+kXwIQTbVv8xiHUPNR2HYKOkseD8eoCNUqpFsGsETwJLOrhmBXGmNnO41dBLEsn85x+gry9Zb1fk8D363/qubYD9rgr4YsPoerQwBe0t5rqoKa48xwCnxSdS6CUai9ogcAY8xFwpMcDQ2TW+GSiIsLsxLLoxN6NGtr2Fow9vvWmOusK+yt700vBKWxv+G7wyV3VCDQQKKXaC3UfwUkiskFE3hGRGV0dJCI3i0ieiOQVFxcPyBdHR4Qze0KK7TDuTdNQ1SEoXAXTLmjdljEFxs0ZGs1D5Xvtc1dNQ0lZznH7Bqc8SqkhL5SBYC0wyRhzPPBn4NWuDjTGPGyMyTXG5GZk+BkS2UfzslPZfKCSxsherFu83ZlQPe289tuPuwoObYSirQNWvj5pmUPQRdNQVBzEpWmNQCnVImSBwBhTaYypdl6/DUSKSPpgliE3exTNXsPB+kjw1IOnsecPbXsLRuVA5vT222deaoeThrpWUFEAYRGQOLbrY3QugVKqjZAFAhEZI2IHuovIPKcspYNZhhMmjSJMYFdVlN1Q9Hn3H6ivgC+Ww/TzO4/RT8iAY74CG18Erzc4BQ5EeQEkjbMT5bqiC9QopdoI5vDRxcBKYKqIFIrIjSJyi4jc4hxyGbBZRDYA9wFXOesiD5rEmEimj03ihdpc21zy9o+6v4nvfBe8TTDtfP/7j7sSKgth77+CU+BAlO+DlEndH+ObVDbS1mlWSvVJMEcNfc0YM9YYE2mMGW+MecwY86Ax5kFn//3GmBnGmOONMfONMZ8EqyzdmZudyor9Xjxn3m07gdc+2fXB296yaRvGz/W/f+q5Nl3FxueDUtaAVBR0PZnMJ3k8NFZDffnglEkpNaSFetRQyM3LSaWuqZlNaYsg+zR49xdQdbjzgZ4GWyOYem7XzS5RcTD9Qvj8dTuef7A1N9mJYl11FPu0pKPW5iGllAYCcn0L2u8tg/PvBU8dLP1J5wP3fASNVV03C/kcd4UdgbT9nSCUtgeV++18hh5rBDqXQCnVyvWBIDMxhuy0OFbnl0H6MXDaD2Hzy7DzvfYHbn3DNvvkLOj+hDkL7IidjUuCV+iudLUOQUdaI1BKteH6QAC2nyAv/4hd0P7UOyBtMrz1fWistQd4m+38gclnQWRM9ycLC4dZl8Gud6FmUAdB9TyHwCc+A8KjNQupUgrQQADAaVMyKKtt4v1tRRARDRfca2fofvR/9oDCPJu/p6dmIZ/jrgKvB7a8ErxC+1NRAAgkje/+uLAwSM7SReyVUoAGAgDOnTmG7LQ47nl3h60VZJ8Ks6+FT/4Mhz+HbW9AWKStEQRizEzInDH4zUPlBZA4BiKiej5WJ5UppRwaCICI8DBuP3MyWw9WsnSLk0H07P+GmGR48w47bDRngX0fqBn/Zoej+huBFCzle3vuH/DRSWVKKYcGAseFx2dxdEY8f3xvB81eA3GpcPavoeAzOPJF59xCPZnqZODeuXTgC9uVQOYQ+OgCNUophwYCR3iYcMeZU9hxuJq3NjmLthx/lZ1bgNj5A70xeqa9KQ/WMFKvFyr299xR7JM8ATBQeSCoxVJKDX0aCNo4b9ZYpo1J5N73duBp9tp8Qpc9Ade+DEndJHHzRwSmLILdywZncln1IZv+ojc1AtCRQ0opDQRthTm1gi+Ka3htvfNL2ZdMri+mnmMnqH2xfOAK2ZWWOQQ95Bny0UllSimHBoIOvjpjNDPGJfGn93fS1NzPLKLZp9pJaDsGoXko0DkEPsnOAjVaI1DK9TQQdCAifP+sKew7Ussra/v5azki2tYmtv8j+KmpK5waQXIPcwh8ImMhLn1wawSVB+Hz1wbv+5RSAdFA4McZ0zI5fkIK972/i0ZPP2/gU86x7fcH1w9M4bpSXmBv7FHxgX8mZZCHkC7/LSz5ug0ISqkhQwOBH75awf7yOl7I62fTyeSzQcJgxz8GpnBdKd8XeLOQz2BOKjMGdjhDafNXDM53KqUCooGgCwsmp5M7aRQPfLCL+qbmvp8oPg0mnNi61nGw9GYOgU/yBFuTGIwFag5tgiqnA37PIHSeK6UCpoGgCyLC98+ewqHKehav2te/k009x94Ig/Xr29Nob+iBzir2SR4PTTVQVxaccrXlqw1MONGm9FZKDRkaCLpx8tHpzD8qlQeW7eJITQAL23dlyjn2OViTyza/bIepHn1G7z43mOmod/wDsk6AmZfZZqyy/OB/p1IqIBoIevCz84+lss7DD5astwnp+iJ9MqQeHZx+AmNg5QOQMb0PgcBpSjryxcCXq63qIti/xk6w863nsEf7CZQaKjQQ9GDGuGR+dv50lm0v5uEVfbxhitjmoT0fQUNV98f2tr0+fwUc3gQnfdd+T2+MngGxqbD5pd59rrd2vgsYmPJVyJgK8ZnaPKTUEKKBIADXzp/EebPG8rul28nLP9K3k0w9B5obbcqJrtRXwhPnwJt3Bn7elQ/YYaOzruh9mSKi4UvXwba3bZ6iYNm51K7aNuY4G6xyFthAMBid1EqpHmkgCICI8L+XzmL8qFi+t3hd3/oLJsyHmJSu+wmam+DF62HfSsh7HPI/7vmcJTttc9O8m3peOa0rJ9xg1zle+1TfPt8TTyPs+sDWBnw1lpwFdm5Fyc7gfKdSqlc0EAQoKSaSB67+EqXVjX3rLwiPsHMKdi61S1+2ZYxd92D3B3Du7yF5Irz9Y2j2dH/OT/9ql5zMvbF3ZWkrNccuuLPmqeCkpN73CTRW2f4Bn5Z+Ah1GqtRQoIGgF2Zm9bO/YOoiqC2FwtXtt3/0O1j3Nzj9P+yv+0X/A0VbIO+xrs9VewTWPwfHXWET4/XH3G/ZX+jb3urfefzZsRQiYiDn9NZto7JtsNN+AqWGBA0EvdSv/oJjzoSwiPbNQ+sXw7Jfw/Ffg4U/sdumnW9HAH3wa6gu9n+uvMftkNGTbu3bH9KxXCkTYfWj/T9XW8bYvzX7NIiKa90uAjmn2Y7uYOdgUkr1SANBL/WrvyAmGSad0hoIvvgQXv93+2v5gvta29BF4Jz/g6ZaeP8Xnc/jaYRVj9hgkTm9v38ShIXbvoL8FVC8vf/n8yndBWV7bP9ARzkL7ES2w5sH7vuUUn2igaAP2vYX3P78ut6loJh6LpRst80wL1wH6VPgymc6LzifPtkOCV33NyjMa79vyyu2KWcgagM+c66D8ChY3U1zVG/55k34CwTZp9lnbR5SKuQ0EPTRzKxk7r54Jit2lvCNJ1ZRWR9gR6tvLePnr7GZQq950dYU/FnwIzvs8u0ftjahGAMr74eMaXB0HxfM8SchA469GDYshsaagTnnjqWQOcN/6ovkLEg7RgOBUkOABoJ+uGLuBP501Wzy8su46qFPKa5q6PlDo7LtzTEqHq5e0v36AdGJcNZ/w4F1sO4Zuy3/Y5u3aH4fJpD1ZO63oKESNr3Y/3PVlcPeT/zXBnxyFthjehodFQwlO20ZlVIaCPrrotlZPHp9LntKarjswU/YV1rb84eu+hvcvBzGHtfzsbMug4knw3u/sCOFfBPIjuvDBLKeTJgHo2fZTuP+Tvba/QGY5vbDRjvKWWCHlgZ7rYaOqovhoQXwt0u1s1opNBAMiIVTM3nuphOpqGvikr9+wpYDFd1/IPUoSD8msJOLwLm/g/pyePU7dtnLuTfaFcYGmgjM/aatcXQc4tpbO5ba9BXjc7s+pqWfYJDnE3xyn+2I35/XWtNSysU0EAyQORNH8dItJxEZLlz10Kd8+kXpwJ18zEyYe5PtfA2Psk04wTLrCohK7N9QUm8z7PynnUAXFt71cfHpMHrm4PYT1JTYv23mZXYE13s/h5oB+m9VXaxpM9SwpIFgAB2TmcjL3zmZ0ckxfP3xVby1cQCXZPzyTyFhDMy+BhIyB+68HUUnwOyvwZa/9/0GWZgHdUdgytk9H5uzAPZ9Cp4A+lcGwsr7oakOTv+xncXdUOV/iG5vFW+HP0y1102pYUYDwQAblxLLi98+iRnjkrj1ubV8+5k89pfX9f/EsSnwvTzbTBRsuTfaBHl9bTbZ8Q+Q8MBGNeUsAE99/5uiAlF7xM6/mHmJzYI6+liY/x1Y+zQUrOrfuTe+YPtENi4ZmLIqNYiCFghE5HERKRIRvzOGxLpPRHaJyEYR+VKwyjLYRsVH8cLNJ/HjRVNZvqOYM/+wnAeX76bR08+OyehECI8cmEJ2J3Oabb/Pe7xzXqRA7FgKk062wasnk062azoPRvPQyvvt0NgFP27ddvpdkDgO3vp+30cvGWMXBwLY/T7U99BHpNQQE8wawZNAN0NGOAeY7DxuBv4axLIMuqiIML678Bje+/7pnDY5nd+8s43z7lsxsH0HwTT3W1C+F548D/L/FfjnyvfZPEndDRttKyYZxs0JfiCoPQKfPQwzLraBzic6ARb9r+0g72u/yIG1dsW1L33d1qS2B2EBIqWCKGiBwBjzEdBdMp6LgKeN9SmQIiJjg1WeUBk/Ko6Hv57LY9fnUtfUzFUPf8qdL6ynqKo+1EXr3rEXwXn3wJE98OS58My/QeGaro8v3gHv/RIed2J/d8NGO8pZYJuGBmoimz+f/sUOVW1bG/A59iKbrmPZr6HqUO/PvellCIuEM39paxefv9b/8io1iELZR5AFFLR5X+hs60REbhaRPBHJKy7uIgnbEPeV6aN5987T+d4Zx/DWxoMs/N2H/OGf26moC0Lq54EgYoep3r4ezr4bDm6AR8+AxV+DQ05rn6/N/ZEz4IG58K8/2VXPrhRzC4IAABpSSURBVH7RpsgIVM4C8HrsWgzBUFcGnz1kb/ijj+28X8R2HHvq4Z//2btze7025cfksyAu1X7HrvfsIkNKDRPDorPYGPOwMSbXGJObkdHPlMshFBsVzg/OnsrSOxdwxrRM/vzBLhb83zL++uFu6hr70BY/GCJj4eTvwe0b4Mv/aZuJHjwVHj0Tfj/Fpr/wNMDZv4bvb7UpMwIZLdTWhPn2F3Vf1jEuWAW73u9+2OanD9oZ0/5qAz5pR8Mpd9hZ1b1pptq3EqoOwsxL7fsZF0Nzg+0nUWqYEBPEcc8ikg28aYyZ6WffQ8CHxpjFzvvtwEJjTLdjLnNzc01eXl53hwwbWw5U8Pul21m2vZiMxGhuO+MYrpw7kaiIIRyf68rgkz/b5S2P/rJNnx3IDOmePH6OHcr57Y8gLMC/P/9f8LdL7C/5Y86ERb/tPFGvrhzuPc6mvb7q2e7P11QHD5xol/C85V+dEwH68+adsOF5+NEumzbE64U/HgtZJ/T8fQOhuck235XscB477XN8Olz1XPfzOJSriMgaY4zfGZ6hvOO8DnzdGT00H6joKQiMNDPGJfPEDfN48ZaTyEmL52evbeGMP3zIc5/to7YxBPl3AhE7Cr7yX3Drp7aTdSCCAMDxV8LhTfDadwMbqXRos22mSp5g2+YLVsFf5sO7/2UDis9nD0FDhV30pyeRsXZ4bskO+PSBno9vbrL9AVMW2SAANohNv9A2DzVU93yO3jDG3vTX/Q3+fgv8ORd+PcY2y71wDbz/S5va3HjtEN7Nrwzs96sRK2g1AhFZDCwE0oHDwM+BSABjzIMiIsD92JFFtcANxpgef+qPpBpBW8YYlu8o5g//3MGm/RUkxkRw+QkTuHb+RI7KSAh18QbH8t/BsrvtrN9/e8gu7+lPWT48dradq3DjPyFlAlQdtjfC9c/aiXdn/7cduXTvLJh0KnztucDLsfhrtnno3/MgqZvxC7ves/mKrnoOpp3Xun3vJ/DEOXDpYzZXVF8ZY9dzyP+XTTaY/zFUFtp9cWm2SS1zmk1lnj4Z0iZDTJKtlTx0mq3h3Lqq6+uoXKW7GkFQm4aCYaQGAh9jDHl7y3h65V7+sfkgTc2G0yanc938SZwxLZOI8CHcbDQQPr7Xpn2YfiFc9njneRPVxfD42baj+ptL2w8FBShYDe/8yGZsTRhj1224eTmMmx14GY58YZuIZlwClzzU9XF//45dV+JHO21zko+3Ge6ZDhNOtGtNBMoY+935H8Ne5+Zfud/ui0uH7FNbHxnTus8+u+0teP5quOgBmHNt4GVwI2NsLTImKdQlCaruAoH+VBhiRIS52anMzU6lqGo6L6wq4NnP9nHzM2vISonlwtnj+PLUTL40MWVkBoVT77D5lJb+BJZcD5c/0XqTra+EZy+FyoNw/eudgwDAhLnwrQ/srOj3f2nXWOhNEACbFPDk78GKP0DuN2HiiZ2PaaqHbW/C9AvaBwGw7fLTL7RNOI01rc1G/nga7azkPcvtjb/KaR2Nz7A3/Emn2Ml9GVN7l3Z86rl2fsaHv7X5owLp73CjmhJ46QY4sN7WABNHh7pEIaE1gmHA0+zlva2HefazfazcXYrHa0iKiWDBlAy+PDWT06dmkJ4Q3fOJhpNVj9gRSZPPhiuesTfBZy+zzSRfez6wkUm+mcJ9aRppqIb759q8Tjd90LnTdeubtl3+2lfgGD+pNPI/tpPxLnvCprToyht3wJonIGF0+xt/+uT+rzex8z0bOM+7xw4FVu0dWGdXCawushMBF/wQzujl8OFhRGsEw1xEeBiLZo5l0cyxVNY38fHOEpZtK+LDHcW8ufEgInDc+BS+Mi2TM6ePZvrYRGSgF60ZbPNusjWDN26HxVfZavuej2zfQaDDU/vTNh6dAGf9Cl75lv1lf8L17fdvfsk21+Sc7v/zE0+C+EzbmdxVINi4xAaBk2+z3zXQ/82O+Yptnvro9zZZYWTMwJ5/OFv3rB3xlZAJNy6112j1o3Dqnd3X4EJpxR9gyjn+58L0kwaCYSYpJpJzZ43l3Flj8XoNnx+sZNm2Ij7YXsQf39vBPe/uYFxyDGceO5ozp4/mxKNSiY4YpkMIT7je9hG8+l3A2LkKx181eN8/6zJ7c3j/l3aimC93UkO1TSMx55qug01YuG022rAYGmshKq79/qJtNshNPBm+8vOBDwJgz3nGf8JTF9iAM/87A/8dw42nEZb+FFY/YicyXvaEHWp78m22qW/ds3DizaEuZWcf3wvv/8rmsTrrVwN+em0aGkGKqxpYtq2Id7ceZsXOYuqbvMRHhXP61Awu/dJ4Fk7NJDxsGNYUtv/Dtp3n3jD4331wAzx0Opx4C5zzG7tt00vw8o1wwzs2aV5XvlgOT18IVzxtA4lPY42djV1TArd83P3IpIHw5PlQvM1OChyqv3YHQ9VhWPJ1KPjU9gF95RftA/mjZ0H1Ybht3dCaf7HmKXjjNjua7pJHAp9n08FQnUegBlhGYjRXzJ3AI1/PZf1/nc1j1+dy4ewsVu0p48an8vjy7z/kkY++oKJ2iKa16MrURaEJAgBjj4cTvgGrHoairXbbppcgKcsO3+zOpFNs89GWV1u3GWObJIq3w6WPBj8IgK0V1BTbfhe3yv/YLk96aKMd1nv23Z1rc6fcZhMtbn09sHOufQZeutH2xfQlS28gPn8N3rwDjjkLLv5rn4NAT7RG4AJNzV6WbjnEU5/kszq/jNjIcC6ek8X1J09i2piRPWRuQNSUwp/nwNjZcMVT8LvJcOK34au/7vmzb9xh+wJ+vNtOWMt7wv7DXvhTWBjAJLeB8rdLYf8auH3jiB8m2Y6nAT64286GT82xAw/GdEp0YHmb4f5ciEmxAwS6a647sN6mWTFeuw5F0ng7THfONZAycWDKvnsZPHeFHf113audmxd7SWsELhcZHsb5x43jxVtO5s3vncqFx4/jlbWFLLp3BVc8tJKHP9rN+oJympp1IXe/4tNsnqU9y21/hbepNbdQT469CJpq7OSzgxvgnf+wmU4X/Ci4Ze7oy//PSb734OB+b280e2zeqFdvhXuOhe3v9O98h7fYJrhP7rO1uls+7joIgG0OOulWm1a8uwSIjbXwyk22b+H7W+HyJyFjCiz/rU1n8swlthbY1I8Mw4Vr4Plr7CTBq1/odxDoidYIXKqsppEleQW8kFfAF8U2/XNsZDhfmpTC3OxU5uWkMmfCKGKjhlBbaSg1e+xs3aLP7TyD760NrIO32QN/mAJZuTZ1hacBbllhbyKDbfHVtonkjg02VUhvHNoEYRGQOX1gy+T1QsFndhTWllehtgSik+wiTJ56+O6nvV+a1eu1acff/6Vd7+LC+23zYiAaa+HemTB+Hlz9vP9j3vqh7Wy+7lWbb8unfJ/tbF73N2cGuNgmxFHZMGqSfU5xntOO7vr/geLtNp17TJKdNJk4JvC/vRs6s1h1q6iyntX5ZazaU8qq/DK2HarEGAgTyEyMYUxyDGOSnOfkGMYmxzAuJZaZ45LdFSj2fGRH4Cz4MZzx/wL/3Ou3wdqnbEqMG96GiT30LQTLoc3w4Ckw/1bbRh5Ie3NNCfzzZ7DBSdEx499s7aI3acbbqiu3AbF4OxzebOdjVBZCRIzN2TTrMtseXpZv2/SPPgO+tjjwUVUVhTYPU/4KO6nugvsgoZcZi5f9Lyz/Ddy62v7Sb2vHP+G5y+01XPQ//j/vbbbNOoWroGyv/VvK97ZOFvSJS7eBNWOanRyZMc0GrueutGnZv7nUNmcNEA0Eqlcq6ppYu7eMdQXlHCiv41BFPYcq6zlUUU91Q2syvIgwYca4JHKzU8mdNIoTskeRmTjCx6rv+xTGzOrd6BtfADn7bjtaJZRe+TZsfB4yj7Vj5mdc4n8IrNcL656Gd39uRzmd/D3bdLLyL+Cpg9lX20R+XbWHG2Nv+Hs/sbWo4u32Ud1m4Z/waDhqob35Tz3H1gLaWvmAHep54f3wpet6/tv2rLDNKV6PHeE157q+DcutLra1guOuhAvva91eUwJ/Ocn+kr9pWe/nZTTVQXmBDQylO+3gg+JtdihxY5tEiTHJ8I23u2/G6gMNBGrAVNU3cbiynr2ltazZW0be3jI2FJTT4KzHPCktjtxJqczNHsW8nFRy0uOH/+S2gVC62zYphfpaNHvsQjor7oHirbaZ4pTb4firW29shzbbkU2Fq2zCvvPvsSkuwN4kP76ndVnP3G/CaT+wv26LttiZ33v/ZQNAbYk9JirBJsbLmGZ/YWc4ifJGZXc/TNPrtcNvD6yD7/zLHt+VPStsx2rKRFuDSD2qf9fpjTtg/XNw52bbNGWMDTK73rVBYCBv0sbYnFJF22yAOGrhwDfBoYFABVmjx8vmAxXk5R8hL98GhyM1jQCkJ0S15E6al5PK9LFJw3Muw0jj9cKOd+xs1f1rbIK+k2614+g//avtQzj7bjuBz1/wKi+Aj/7PtolHRNuJf/UVdl/yRMg+xc6xmHiybQ/vawAs3wd/OdmmO7/+Df+Bo20QuP7N3jcF+VOyy44g8qWd8I3l/+r/2Os0DGkgUIPKGMPu4hpW7TnC6vwjrNpzhP3ldYDtkD46M57JmYkck5nQ8piUGjcyk+gNdcbY0VAr/tC6MtsJ37CzneNSe/58yS745E/29aRTYdJJAzd80mfds3adCn9Na8EIAj7PX2NrN9e/YdOej59rO4iDNJY/2DQQqJA7UF7H6vwjrC8oZ1dRNbuLqjlQ0Tq8LjJcODojgePGJzNrfArHj09m6pjE4ZseYzg6uMGODBo9I9Qlaa9ts8zNy1tz7QQzCIDtD3r8qxCVaGsi3/kEkv0uqz4saCBQQ1J1g4fdRdXsKqpmV3E1Ww9WsrGwoqVZKSo8jGljE5mVlcyx45LISY8nJz2e0YkxhGnzkrtUF9sV6JLG2jTjBZ+1CQJv9H6IaaAePcv2lVz+pB0xNYxpIFDDhjGG/eV1bCyscB7lbCqsoKrNaKWYyDCy0+LtIz2eSWlxjE2OISsllrEpsSREay7FEcm32I5vKdBgBwGwa0DvXzO4yQ6DRAOBGta8XsPBynryS2rY4zzyS2rYU1pDwZFamprb/z+cGBNhg0JyDMdkJnD8hBRmT0ghKyVWRzANd69+1y5HmjEt+EFghNFAoEYsT7OXw1UNHCyvY395HQcr6jlYXseBinr2l9Wxq7iaRmdoa3pCNLMnpDBnog0Mk9LiSIuPdtekuOGuvtIOXZ1zrQaBXtJAoFyr0eNl26FK1heUs35fOesLyvmipKbdMbGR4aTGR5GWEEVqvH0cnZHA8eNTmJWVTHJcZBdnV2r40ECgVBsVtU1sKLSzpktrGimraeRITSOlznNJdQMH24xoyk6LaxnJNDMrmbHJMSTHRpIYE6lzItSwoUtVKtVGclwkC6Z0P9SworaJjfvLWzqs1+Qf4Y0NBzodlxAd4QQF+zw6yeZisjmZbD/F2OQY0hKiNWioIUsDgVJ+JMdFctrkDE6b3Bowiqsa2HKggtLqRirrm6ioa6KyztPy2lfT+MeW+pZ+CZ/IcGFiahxHZSRwdEYCR2XEc3RGAkdnxJMSFzXYf55S7WggUCpAGYnRLJzacwelMYYjNY0crLCJ+g5W1LG/vJ49JdXsLq7hw+1F7UY6pcRFkpkYTXpCm0diFOkJ0YxOsiOfxiXH6IgnFTQaCJQaYCJCWkI0aQnRzMxK7rTf0+ylsKyOL0qq2V1UQ35pDcVVDZRUN7C+oJyS6gZqG9svfRgfFc4xoxOZkpnA5NEJTB6dSE5aPClxkSRER2h6DtUvGgiUGmQR4WFkp9vJcGdM839MbaOHkqpGDlbUsdOZfb3jcBUf7ijmxTWFnY6PjwonKTaSpJhIkmIjSI6NIjMpmszEaDITY+xzkn2dnhClgUO1o4FAqSEoLiqCiWkRTEyL48Sj0trtK6tpZFdxNXtLa6msa6Kyvomqek/L68o6D4VltazbV0apk66jo6SYCEbFRzEqzg6XTYmLJDUuitSEKNLjo1uG06YnRJOWEEVclN4qRjL9r6vUMDMqPoq58Ta1d0+amr2UVDdQVNlAUVUDhyvrKa1upKzWPo7UNFJc1cD2Q1WU1TZ2apLyiY0MJy4qHBFBxK5eFyaCYJvCYqPCSYyJIDHGjqBKiokkKSaCxJgIEqIjiI9ufW59bedvaJAJPf0voNQIFhke5gxjjQ3o+PqmZkprGimttn0WJdWNlFbb93VNzRhsZ7gxNimo1xi8xn7ON3qqsKyWyjoPVfVNLQsWdScuKpy0hCjS4n2d5a21kfSEaDKcjvSMhGiSYiO00zwINBAopVrERIaTlRJLVkpggaMnDZ5mahqaqWnwUN3gafPcTHVDE0dqmlqCTmlNI4VltWwoLOdITSPN3s6TXaPCw0iNjyIyQogICyNMsM9hQkSYEBEuJETbmkhitO0v8dVSfDWSuKhw4qMjiI20z/FR4cQ5z24NMhoIlFJBEx0RTnSEbQLqDa/XUFbbSEm1nentG1VVXN3AkepGPF6Dx2vweg0er5dmLzR7vXi8hqp6DwfK66iq91BV76GuyX9zV0cidoJgUpvAkRgTQUpcFBmJtkaSkWg74DOcR3Js5IgIHhoIlFJDTlhY6xDcqST2/IFuNDV7qa53aiKNHmobm6ltaHZee9rVWKrq7QTBaieIlFQ3srOomuKqBr/NXCIQExFObFQ4MRFhxESFt7xPiolgTHIMmYkxjE6KYXSSnReSmWQDSFR42JAJIhoIlFIjWmR4mB0h1ctaSVvGGKoaPBRX2dpJkfNcXttIfVMz9U1e6pqaqWtqpqGpmdrGZoqqGti0v5LSmga6SukWHRFGVESYU3MKIzoirN2iSx1zwV01dyI3LTiqz39HV4IaCERkEfAnIBx41Bjzmw77vwH8DtjvbLrfGPNoMMuklFK9JSLOSKhIjs5I6NVnm5q9FDsjtg5X2ufqBg8NTc00eLwtj0aPlwZPc+eg0abSkJkU3f8/xo+gBQIRCQceAM4CCoHVIvK6MebzDoe+YIz592CVQymlQikyPIxxKbGMG6AO+GAI5vTCecAuY8wXxphG4HngoiB+n1JKqT4IZiDIAgravC90tnV0qYhsFJGXRGSCvxOJyM0ikiciecXFxcEoq1JKuVaoE468AWQbY44D3gWe8neQMeZhY0yuMSY3I6P7PPJKKaV6J5iBYD/Q9hf+eFo7hQEwxpQaYxqct48CJwSxPEoppfwIZiBYDUwWkRwRiQKuAl5ve4CIjG3z9kJgaxDLo5RSyo+gjRoyxnhE5N+Bpdjho48bY7aIyK+APGPM68BtInIh4AGOAN8IVnmUUkr5p4vXK6WUC3S3eH2oO4uVUkqF2LCrEYhIMbC3jx9PB0oGsDgjhV6XzvSadKbXpLPhdE0mGWP8DrscdoGgP0Qkr6uqkZvpdelMr0lnek06GynXRJuGlFLK5TQQKKWUy7ktEDwc6gIMUXpdOtNr0plek85GxDVxVR+BUkqpztxWI1BKKdWBBgKllHI51wQCEVkkIttFZJeI3BXq8oSCiDwuIkUisrnNtlQReVdEdjrPo0JZxsEmIhNEZJmIfC4iW0Tkdme7a6+LiMSIyCoR2eBck18623NE5DPn39ALTg4xVxGRcBFZJyJvOu9HxDVxRSBos1raOcCxwNdE5NjQliokngQWddh2F/C+MWYy8L7z3k08wA+MMccC84Fbnf833HxdGoAzjDHHA7OBRSIyH/gt8EdjzDFAGXBjCMsYKrfTPjnmiLgmrggE6GppABhjPsIm92vrIlrXgXgKuHhQCxVixpiDxpi1zusq7D/yLFx8XYxV7byNdB4GOAN4ydnuqmsCICLjgfOwKfMREWGEXBO3BIJAV0tzo9HGmIPO60PA6FAWJpREJBuYA3yGy6+L0wSyHijCLhq1Gyg3xnicQ9z4b+he4MeA13mfxgi5Jm4JBCoAxo4lduV4YhFJAF4G7jDGVLbd58brYoxpNsbMxi4oNQ+YFuIihZSInA8UGWPWhLoswRC09QiGmB5XS3OxwyIy1hhz0FkoqCjUBRpsIhKJDQLPGmNecTa7/roAGGPKRWQZcBKQIiIRzi9gt/0bOgW4UETOBWKAJOBPjJBr4pYaQY+rpbnY68D1zuvrgddCWJZB57TzPgZsNcbc02aXa6+LiGSISIrzOhY4C9t3sgy4zDnMVdfEGPMTY8x4Y0w29v7xgTHmGkbINXHNzGInkt9L62ppvw5xkQadiCwGFmJT5x4Gfg68CiwBJmLTe19hjOnYoTxiicipwApgE61tvz/F9hO48rqIyHHYjs9w7I/FJcaYX4nIUdiBFqnAOuDaNmuOu4aILAR+aIw5f6RcE9cEAqWUUv65pWlIKaVUFzQQKKWUy2kgUEopl9NAoJRSLqeBQCmlXE4DgVKDSEQW+jJXKjVUaCBQSimX00CglB8icq2Tk3+9iDzkJGGrFpE/Ojn63xeRDOfY2SLyqYhsFJG/+9YuEJFjROQ9J6//WhE52jl9goi8JCLbRORZZ3azUiGjgUCpDkRkOnAlcIqTeK0ZuAaIB/KMMTOA5diZ2QBPA/9hjDkOO0PZt/1Z4AEnr//JgC+b6RzgDuzaGEdh89goFTJuSTqnVG98BTgBWO38WI/FJp3zAi84x/wNeEVEkoEUY8xyZ/tTwIsikghkGWP+DmCMqQdwzrfKGFPovF8PZAMfB//PUso/DQRKdSbAU8aYn7TbKPKzDsf1NT9L21w0zei/QxVi2jSkVGfvA5eJSCa0rF88CfvvxZdp8mrgY2NMBVAmIqc5268DljurnRWKyMXOOaJFJG5Q/wqlAqS/RJTqwBjzuYj8J/BPEQkDmoBbgRpgnrOvCNuPADb98IPOjf4L4AZn+3XAQyLyK+cclw/in6FUwDT7qFIBEpFqY0xCqMuh1EDTpiGllHI5rREopZTLaY1AKaVcTgOBUkq5nAYCpZRyOQ0ESinlchoIlFLK5f4/l6n6eKtOiEYAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ixnytC4InCTq"
      },
      "source": [
        "DATA AUGMENTATION"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HWNeLghmnGV4"
      },
      "source": [
        "from tensorflow.keras.preprocessing.image import ImageDataGenerator"
      ],
      "execution_count": 32,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DVbYPpIsoWTl"
      },
      "source": [
        "datagen = ImageDataGenerator(width_shift_range=0.1, height_shift_range=0.1, horizontal_flip=True)"
      ],
      "execution_count": 33,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uyiQL9gZs2QL"
      },
      "source": [
        "input = layers.Input(shape=(img_height, img_width, channel,))\n",
        "First_Conv2D = layers.Conv2D(num_filter, (3,3), use_bias=False ,padding='same')(input)\n",
        "\n",
        "First_Block = denseblock(First_Conv2D, num_filter, dropout_rate)\n",
        "First_Transition = transition(First_Block, num_filter, dropout_rate)\n",
        "\n",
        "Second_Block = denseblock(First_Transition, num_filter, dropout_rate)\n",
        "Second_Transition = transition(Second_Block, num_filter, dropout_rate)\n",
        "\n",
        "Third_Block = denseblock(Second_Transition, num_filter, dropout_rate)\n",
        "Third_Transition = transition(Third_Block, num_filter, dropout_rate)\n",
        "\n",
        "Last_Block = denseblock(Third_Transition,  num_filter, dropout_rate)\n",
        "output = output_layer(Last_Block)"
      ],
      "execution_count": 34,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_Ohxn0hws-84",
        "outputId": "e35388f9-5a22-4c3d-8713-87fd3d549d87"
      },
      "source": [
        "model1 = Model(inputs=[input], outputs=[output])\n",
        "model1.summary()"
      ],
      "execution_count": 35,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"model_2\"\n",
            "__________________________________________________________________________________________________\n",
            "Layer (type)                    Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            "input_3 (InputLayer)            [(None, 32, 32, 3)]  0                                            \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_104 (Conv2D)             (None, 32, 32, 12)   324         input_3[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_104 (BatchN (None, 32, 32, 12)   48          conv2d_104[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "activation_104 (Activation)     (None, 32, 32, 12)   0           batch_normalization_104[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_105 (Conv2D)             (None, 32, 32, 6)    648         activation_104[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_102 (Dropout)           (None, 32, 32, 6)    0           conv2d_105[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_96 (Concatenate)    (None, 32, 32, 18)   0           conv2d_104[0][0]                 \n",
            "                                                                 dropout_102[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_105 (BatchN (None, 32, 32, 18)   72          concatenate_96[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_105 (Activation)     (None, 32, 32, 18)   0           batch_normalization_105[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_106 (Conv2D)             (None, 32, 32, 6)    972         activation_105[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_103 (Dropout)           (None, 32, 32, 6)    0           conv2d_106[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_97 (Concatenate)    (None, 32, 32, 24)   0           concatenate_96[0][0]             \n",
            "                                                                 dropout_103[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_106 (BatchN (None, 32, 32, 24)   96          concatenate_97[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_106 (Activation)     (None, 32, 32, 24)   0           batch_normalization_106[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_107 (Conv2D)             (None, 32, 32, 6)    1296        activation_106[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_104 (Dropout)           (None, 32, 32, 6)    0           conv2d_107[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_98 (Concatenate)    (None, 32, 32, 30)   0           concatenate_97[0][0]             \n",
            "                                                                 dropout_104[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_107 (BatchN (None, 32, 32, 30)   120         concatenate_98[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_107 (Activation)     (None, 32, 32, 30)   0           batch_normalization_107[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_108 (Conv2D)             (None, 32, 32, 6)    1620        activation_107[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_105 (Dropout)           (None, 32, 32, 6)    0           conv2d_108[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_99 (Concatenate)    (None, 32, 32, 36)   0           concatenate_98[0][0]             \n",
            "                                                                 dropout_105[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_108 (BatchN (None, 32, 32, 36)   144         concatenate_99[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "activation_108 (Activation)     (None, 32, 32, 36)   0           batch_normalization_108[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_109 (Conv2D)             (None, 32, 32, 6)    1944        activation_108[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_106 (Dropout)           (None, 32, 32, 6)    0           conv2d_109[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_100 (Concatenate)   (None, 32, 32, 42)   0           concatenate_99[0][0]             \n",
            "                                                                 dropout_106[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_109 (BatchN (None, 32, 32, 42)   168         concatenate_100[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_109 (Activation)     (None, 32, 32, 42)   0           batch_normalization_109[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_110 (Conv2D)             (None, 32, 32, 6)    2268        activation_109[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_107 (Dropout)           (None, 32, 32, 6)    0           conv2d_110[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_101 (Concatenate)   (None, 32, 32, 48)   0           concatenate_100[0][0]            \n",
            "                                                                 dropout_107[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_110 (BatchN (None, 32, 32, 48)   192         concatenate_101[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_110 (Activation)     (None, 32, 32, 48)   0           batch_normalization_110[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_111 (Conv2D)             (None, 32, 32, 6)    2592        activation_110[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_108 (Dropout)           (None, 32, 32, 6)    0           conv2d_111[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_102 (Concatenate)   (None, 32, 32, 54)   0           concatenate_101[0][0]            \n",
            "                                                                 dropout_108[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_111 (BatchN (None, 32, 32, 54)   216         concatenate_102[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_111 (Activation)     (None, 32, 32, 54)   0           batch_normalization_111[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_112 (Conv2D)             (None, 32, 32, 6)    2916        activation_111[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_109 (Dropout)           (None, 32, 32, 6)    0           conv2d_112[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_103 (Concatenate)   (None, 32, 32, 60)   0           concatenate_102[0][0]            \n",
            "                                                                 dropout_109[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_112 (BatchN (None, 32, 32, 60)   240         concatenate_103[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_112 (Activation)     (None, 32, 32, 60)   0           batch_normalization_112[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_113 (Conv2D)             (None, 32, 32, 6)    3240        activation_112[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_110 (Dropout)           (None, 32, 32, 6)    0           conv2d_113[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_104 (Concatenate)   (None, 32, 32, 66)   0           concatenate_103[0][0]            \n",
            "                                                                 dropout_110[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_113 (BatchN (None, 32, 32, 66)   264         concatenate_104[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_113 (Activation)     (None, 32, 32, 66)   0           batch_normalization_113[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_114 (Conv2D)             (None, 32, 32, 6)    3564        activation_113[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_111 (Dropout)           (None, 32, 32, 6)    0           conv2d_114[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_105 (Concatenate)   (None, 32, 32, 72)   0           concatenate_104[0][0]            \n",
            "                                                                 dropout_111[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_114 (BatchN (None, 32, 32, 72)   288         concatenate_105[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_114 (Activation)     (None, 32, 32, 72)   0           batch_normalization_114[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_115 (Conv2D)             (None, 32, 32, 6)    3888        activation_114[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_112 (Dropout)           (None, 32, 32, 6)    0           conv2d_115[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_106 (Concatenate)   (None, 32, 32, 78)   0           concatenate_105[0][0]            \n",
            "                                                                 dropout_112[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_115 (BatchN (None, 32, 32, 78)   312         concatenate_106[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_115 (Activation)     (None, 32, 32, 78)   0           batch_normalization_115[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_116 (Conv2D)             (None, 32, 32, 6)    4212        activation_115[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_113 (Dropout)           (None, 32, 32, 6)    0           conv2d_116[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_107 (Concatenate)   (None, 32, 32, 84)   0           concatenate_106[0][0]            \n",
            "                                                                 dropout_113[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_116 (BatchN (None, 32, 32, 84)   336         concatenate_107[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_116 (Activation)     (None, 32, 32, 84)   0           batch_normalization_116[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_117 (Conv2D)             (None, 32, 32, 6)    504         activation_116[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_114 (Dropout)           (None, 32, 32, 6)    0           conv2d_117[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "average_pooling2d_8 (AveragePoo (None, 16, 16, 6)    0           dropout_114[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_117 (BatchN (None, 16, 16, 6)    24          average_pooling2d_8[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "activation_117 (Activation)     (None, 16, 16, 6)    0           batch_normalization_117[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_118 (Conv2D)             (None, 16, 16, 6)    324         activation_117[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_115 (Dropout)           (None, 16, 16, 6)    0           conv2d_118[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_108 (Concatenate)   (None, 16, 16, 12)   0           average_pooling2d_8[0][0]        \n",
            "                                                                 dropout_115[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_118 (BatchN (None, 16, 16, 12)   48          concatenate_108[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_118 (Activation)     (None, 16, 16, 12)   0           batch_normalization_118[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_119 (Conv2D)             (None, 16, 16, 6)    648         activation_118[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_116 (Dropout)           (None, 16, 16, 6)    0           conv2d_119[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_109 (Concatenate)   (None, 16, 16, 18)   0           concatenate_108[0][0]            \n",
            "                                                                 dropout_116[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_119 (BatchN (None, 16, 16, 18)   72          concatenate_109[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_119 (Activation)     (None, 16, 16, 18)   0           batch_normalization_119[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_120 (Conv2D)             (None, 16, 16, 6)    972         activation_119[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_117 (Dropout)           (None, 16, 16, 6)    0           conv2d_120[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_110 (Concatenate)   (None, 16, 16, 24)   0           concatenate_109[0][0]            \n",
            "                                                                 dropout_117[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_120 (BatchN (None, 16, 16, 24)   96          concatenate_110[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_120 (Activation)     (None, 16, 16, 24)   0           batch_normalization_120[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_121 (Conv2D)             (None, 16, 16, 6)    1296        activation_120[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_118 (Dropout)           (None, 16, 16, 6)    0           conv2d_121[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_111 (Concatenate)   (None, 16, 16, 30)   0           concatenate_110[0][0]            \n",
            "                                                                 dropout_118[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_121 (BatchN (None, 16, 16, 30)   120         concatenate_111[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_121 (Activation)     (None, 16, 16, 30)   0           batch_normalization_121[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_122 (Conv2D)             (None, 16, 16, 6)    1620        activation_121[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_119 (Dropout)           (None, 16, 16, 6)    0           conv2d_122[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_112 (Concatenate)   (None, 16, 16, 36)   0           concatenate_111[0][0]            \n",
            "                                                                 dropout_119[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_122 (BatchN (None, 16, 16, 36)   144         concatenate_112[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_122 (Activation)     (None, 16, 16, 36)   0           batch_normalization_122[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_123 (Conv2D)             (None, 16, 16, 6)    1944        activation_122[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_120 (Dropout)           (None, 16, 16, 6)    0           conv2d_123[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_113 (Concatenate)   (None, 16, 16, 42)   0           concatenate_112[0][0]            \n",
            "                                                                 dropout_120[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_123 (BatchN (None, 16, 16, 42)   168         concatenate_113[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_123 (Activation)     (None, 16, 16, 42)   0           batch_normalization_123[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_124 (Conv2D)             (None, 16, 16, 6)    2268        activation_123[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_121 (Dropout)           (None, 16, 16, 6)    0           conv2d_124[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_114 (Concatenate)   (None, 16, 16, 48)   0           concatenate_113[0][0]            \n",
            "                                                                 dropout_121[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_124 (BatchN (None, 16, 16, 48)   192         concatenate_114[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_124 (Activation)     (None, 16, 16, 48)   0           batch_normalization_124[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_125 (Conv2D)             (None, 16, 16, 6)    2592        activation_124[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_122 (Dropout)           (None, 16, 16, 6)    0           conv2d_125[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_115 (Concatenate)   (None, 16, 16, 54)   0           concatenate_114[0][0]            \n",
            "                                                                 dropout_122[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_125 (BatchN (None, 16, 16, 54)   216         concatenate_115[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_125 (Activation)     (None, 16, 16, 54)   0           batch_normalization_125[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_126 (Conv2D)             (None, 16, 16, 6)    2916        activation_125[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_123 (Dropout)           (None, 16, 16, 6)    0           conv2d_126[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_116 (Concatenate)   (None, 16, 16, 60)   0           concatenate_115[0][0]            \n",
            "                                                                 dropout_123[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_126 (BatchN (None, 16, 16, 60)   240         concatenate_116[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_126 (Activation)     (None, 16, 16, 60)   0           batch_normalization_126[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_127 (Conv2D)             (None, 16, 16, 6)    3240        activation_126[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_124 (Dropout)           (None, 16, 16, 6)    0           conv2d_127[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_117 (Concatenate)   (None, 16, 16, 66)   0           concatenate_116[0][0]            \n",
            "                                                                 dropout_124[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_127 (BatchN (None, 16, 16, 66)   264         concatenate_117[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_127 (Activation)     (None, 16, 16, 66)   0           batch_normalization_127[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_128 (Conv2D)             (None, 16, 16, 6)    3564        activation_127[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_125 (Dropout)           (None, 16, 16, 6)    0           conv2d_128[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_118 (Concatenate)   (None, 16, 16, 72)   0           concatenate_117[0][0]            \n",
            "                                                                 dropout_125[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_128 (BatchN (None, 16, 16, 72)   288         concatenate_118[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_128 (Activation)     (None, 16, 16, 72)   0           batch_normalization_128[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_129 (Conv2D)             (None, 16, 16, 6)    3888        activation_128[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_126 (Dropout)           (None, 16, 16, 6)    0           conv2d_129[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_119 (Concatenate)   (None, 16, 16, 78)   0           concatenate_118[0][0]            \n",
            "                                                                 dropout_126[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_129 (BatchN (None, 16, 16, 78)   312         concatenate_119[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_129 (Activation)     (None, 16, 16, 78)   0           batch_normalization_129[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_130 (Conv2D)             (None, 16, 16, 6)    468         activation_129[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_127 (Dropout)           (None, 16, 16, 6)    0           conv2d_130[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "average_pooling2d_9 (AveragePoo (None, 8, 8, 6)      0           dropout_127[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_130 (BatchN (None, 8, 8, 6)      24          average_pooling2d_9[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "activation_130 (Activation)     (None, 8, 8, 6)      0           batch_normalization_130[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_131 (Conv2D)             (None, 8, 8, 6)      324         activation_130[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_128 (Dropout)           (None, 8, 8, 6)      0           conv2d_131[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_120 (Concatenate)   (None, 8, 8, 12)     0           average_pooling2d_9[0][0]        \n",
            "                                                                 dropout_128[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_131 (BatchN (None, 8, 8, 12)     48          concatenate_120[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_131 (Activation)     (None, 8, 8, 12)     0           batch_normalization_131[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_132 (Conv2D)             (None, 8, 8, 6)      648         activation_131[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_129 (Dropout)           (None, 8, 8, 6)      0           conv2d_132[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_121 (Concatenate)   (None, 8, 8, 18)     0           concatenate_120[0][0]            \n",
            "                                                                 dropout_129[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_132 (BatchN (None, 8, 8, 18)     72          concatenate_121[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_132 (Activation)     (None, 8, 8, 18)     0           batch_normalization_132[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_133 (Conv2D)             (None, 8, 8, 6)      972         activation_132[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_130 (Dropout)           (None, 8, 8, 6)      0           conv2d_133[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_122 (Concatenate)   (None, 8, 8, 24)     0           concatenate_121[0][0]            \n",
            "                                                                 dropout_130[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_133 (BatchN (None, 8, 8, 24)     96          concatenate_122[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_133 (Activation)     (None, 8, 8, 24)     0           batch_normalization_133[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_134 (Conv2D)             (None, 8, 8, 6)      1296        activation_133[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_131 (Dropout)           (None, 8, 8, 6)      0           conv2d_134[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_123 (Concatenate)   (None, 8, 8, 30)     0           concatenate_122[0][0]            \n",
            "                                                                 dropout_131[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_134 (BatchN (None, 8, 8, 30)     120         concatenate_123[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_134 (Activation)     (None, 8, 8, 30)     0           batch_normalization_134[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_135 (Conv2D)             (None, 8, 8, 6)      1620        activation_134[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_132 (Dropout)           (None, 8, 8, 6)      0           conv2d_135[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_124 (Concatenate)   (None, 8, 8, 36)     0           concatenate_123[0][0]            \n",
            "                                                                 dropout_132[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_135 (BatchN (None, 8, 8, 36)     144         concatenate_124[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_135 (Activation)     (None, 8, 8, 36)     0           batch_normalization_135[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_136 (Conv2D)             (None, 8, 8, 6)      1944        activation_135[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_133 (Dropout)           (None, 8, 8, 6)      0           conv2d_136[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_125 (Concatenate)   (None, 8, 8, 42)     0           concatenate_124[0][0]            \n",
            "                                                                 dropout_133[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_136 (BatchN (None, 8, 8, 42)     168         concatenate_125[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_136 (Activation)     (None, 8, 8, 42)     0           batch_normalization_136[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_137 (Conv2D)             (None, 8, 8, 6)      2268        activation_136[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_134 (Dropout)           (None, 8, 8, 6)      0           conv2d_137[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_126 (Concatenate)   (None, 8, 8, 48)     0           concatenate_125[0][0]            \n",
            "                                                                 dropout_134[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_137 (BatchN (None, 8, 8, 48)     192         concatenate_126[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_137 (Activation)     (None, 8, 8, 48)     0           batch_normalization_137[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_138 (Conv2D)             (None, 8, 8, 6)      2592        activation_137[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_135 (Dropout)           (None, 8, 8, 6)      0           conv2d_138[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_127 (Concatenate)   (None, 8, 8, 54)     0           concatenate_126[0][0]            \n",
            "                                                                 dropout_135[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_138 (BatchN (None, 8, 8, 54)     216         concatenate_127[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_138 (Activation)     (None, 8, 8, 54)     0           batch_normalization_138[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_139 (Conv2D)             (None, 8, 8, 6)      2916        activation_138[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_136 (Dropout)           (None, 8, 8, 6)      0           conv2d_139[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_128 (Concatenate)   (None, 8, 8, 60)     0           concatenate_127[0][0]            \n",
            "                                                                 dropout_136[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_139 (BatchN (None, 8, 8, 60)     240         concatenate_128[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_139 (Activation)     (None, 8, 8, 60)     0           batch_normalization_139[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_140 (Conv2D)             (None, 8, 8, 6)      3240        activation_139[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_137 (Dropout)           (None, 8, 8, 6)      0           conv2d_140[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_129 (Concatenate)   (None, 8, 8, 66)     0           concatenate_128[0][0]            \n",
            "                                                                 dropout_137[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_140 (BatchN (None, 8, 8, 66)     264         concatenate_129[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_140 (Activation)     (None, 8, 8, 66)     0           batch_normalization_140[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_141 (Conv2D)             (None, 8, 8, 6)      3564        activation_140[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_138 (Dropout)           (None, 8, 8, 6)      0           conv2d_141[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_130 (Concatenate)   (None, 8, 8, 72)     0           concatenate_129[0][0]            \n",
            "                                                                 dropout_138[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_141 (BatchN (None, 8, 8, 72)     288         concatenate_130[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_141 (Activation)     (None, 8, 8, 72)     0           batch_normalization_141[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_142 (Conv2D)             (None, 8, 8, 6)      3888        activation_141[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_139 (Dropout)           (None, 8, 8, 6)      0           conv2d_142[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_131 (Concatenate)   (None, 8, 8, 78)     0           concatenate_130[0][0]            \n",
            "                                                                 dropout_139[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_142 (BatchN (None, 8, 8, 78)     312         concatenate_131[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_142 (Activation)     (None, 8, 8, 78)     0           batch_normalization_142[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_143 (Conv2D)             (None, 8, 8, 6)      468         activation_142[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_140 (Dropout)           (None, 8, 8, 6)      0           conv2d_143[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "average_pooling2d_10 (AveragePo (None, 4, 4, 6)      0           dropout_140[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_143 (BatchN (None, 4, 4, 6)      24          average_pooling2d_10[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "activation_143 (Activation)     (None, 4, 4, 6)      0           batch_normalization_143[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_144 (Conv2D)             (None, 4, 4, 6)      324         activation_143[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_141 (Dropout)           (None, 4, 4, 6)      0           conv2d_144[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_132 (Concatenate)   (None, 4, 4, 12)     0           average_pooling2d_10[0][0]       \n",
            "                                                                 dropout_141[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_144 (BatchN (None, 4, 4, 12)     48          concatenate_132[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_144 (Activation)     (None, 4, 4, 12)     0           batch_normalization_144[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_145 (Conv2D)             (None, 4, 4, 6)      648         activation_144[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_142 (Dropout)           (None, 4, 4, 6)      0           conv2d_145[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_133 (Concatenate)   (None, 4, 4, 18)     0           concatenate_132[0][0]            \n",
            "                                                                 dropout_142[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_145 (BatchN (None, 4, 4, 18)     72          concatenate_133[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_145 (Activation)     (None, 4, 4, 18)     0           batch_normalization_145[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_146 (Conv2D)             (None, 4, 4, 6)      972         activation_145[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_143 (Dropout)           (None, 4, 4, 6)      0           conv2d_146[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_134 (Concatenate)   (None, 4, 4, 24)     0           concatenate_133[0][0]            \n",
            "                                                                 dropout_143[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_146 (BatchN (None, 4, 4, 24)     96          concatenate_134[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_146 (Activation)     (None, 4, 4, 24)     0           batch_normalization_146[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_147 (Conv2D)             (None, 4, 4, 6)      1296        activation_146[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_144 (Dropout)           (None, 4, 4, 6)      0           conv2d_147[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_135 (Concatenate)   (None, 4, 4, 30)     0           concatenate_134[0][0]            \n",
            "                                                                 dropout_144[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_147 (BatchN (None, 4, 4, 30)     120         concatenate_135[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_147 (Activation)     (None, 4, 4, 30)     0           batch_normalization_147[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_148 (Conv2D)             (None, 4, 4, 6)      1620        activation_147[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_145 (Dropout)           (None, 4, 4, 6)      0           conv2d_148[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_136 (Concatenate)   (None, 4, 4, 36)     0           concatenate_135[0][0]            \n",
            "                                                                 dropout_145[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_148 (BatchN (None, 4, 4, 36)     144         concatenate_136[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_148 (Activation)     (None, 4, 4, 36)     0           batch_normalization_148[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_149 (Conv2D)             (None, 4, 4, 6)      1944        activation_148[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_146 (Dropout)           (None, 4, 4, 6)      0           conv2d_149[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_137 (Concatenate)   (None, 4, 4, 42)     0           concatenate_136[0][0]            \n",
            "                                                                 dropout_146[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_149 (BatchN (None, 4, 4, 42)     168         concatenate_137[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_149 (Activation)     (None, 4, 4, 42)     0           batch_normalization_149[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_150 (Conv2D)             (None, 4, 4, 6)      2268        activation_149[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_147 (Dropout)           (None, 4, 4, 6)      0           conv2d_150[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_138 (Concatenate)   (None, 4, 4, 48)     0           concatenate_137[0][0]            \n",
            "                                                                 dropout_147[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_150 (BatchN (None, 4, 4, 48)     192         concatenate_138[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_150 (Activation)     (None, 4, 4, 48)     0           batch_normalization_150[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_151 (Conv2D)             (None, 4, 4, 6)      2592        activation_150[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_148 (Dropout)           (None, 4, 4, 6)      0           conv2d_151[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_139 (Concatenate)   (None, 4, 4, 54)     0           concatenate_138[0][0]            \n",
            "                                                                 dropout_148[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_151 (BatchN (None, 4, 4, 54)     216         concatenate_139[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_151 (Activation)     (None, 4, 4, 54)     0           batch_normalization_151[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_152 (Conv2D)             (None, 4, 4, 6)      2916        activation_151[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_149 (Dropout)           (None, 4, 4, 6)      0           conv2d_152[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_140 (Concatenate)   (None, 4, 4, 60)     0           concatenate_139[0][0]            \n",
            "                                                                 dropout_149[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_152 (BatchN (None, 4, 4, 60)     240         concatenate_140[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_152 (Activation)     (None, 4, 4, 60)     0           batch_normalization_152[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_153 (Conv2D)             (None, 4, 4, 6)      3240        activation_152[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_150 (Dropout)           (None, 4, 4, 6)      0           conv2d_153[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_141 (Concatenate)   (None, 4, 4, 66)     0           concatenate_140[0][0]            \n",
            "                                                                 dropout_150[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_153 (BatchN (None, 4, 4, 66)     264         concatenate_141[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_153 (Activation)     (None, 4, 4, 66)     0           batch_normalization_153[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_154 (Conv2D)             (None, 4, 4, 6)      3564        activation_153[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_151 (Dropout)           (None, 4, 4, 6)      0           conv2d_154[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_142 (Concatenate)   (None, 4, 4, 72)     0           concatenate_141[0][0]            \n",
            "                                                                 dropout_151[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_154 (BatchN (None, 4, 4, 72)     288         concatenate_142[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_154 (Activation)     (None, 4, 4, 72)     0           batch_normalization_154[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "conv2d_155 (Conv2D)             (None, 4, 4, 6)      3888        activation_154[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "dropout_152 (Dropout)           (None, 4, 4, 6)      0           conv2d_155[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "concatenate_143 (Concatenate)   (None, 4, 4, 78)     0           concatenate_142[0][0]            \n",
            "                                                                 dropout_152[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "batch_normalization_155 (BatchN (None, 4, 4, 78)     312         concatenate_143[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "activation_155 (Activation)     (None, 4, 4, 78)     0           batch_normalization_155[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "average_pooling2d_11 (AveragePo (None, 2, 2, 78)     0           activation_155[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "flatten_2 (Flatten)             (None, 312)          0           average_pooling2d_11[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "dense_2 (Dense)                 (None, 10)           3130        flatten_2[0][0]                  \n",
            "==================================================================================================\n",
            "Total params: 118,918\n",
            "Trainable params: 114,394\n",
            "Non-trainable params: 4,524\n",
            "__________________________________________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NQNgitiBzptf"
      },
      "source": [
        "DynamicLR =ReduceLROnPlateau(monitor='val_accuracy', factor=0.9, patience=2, verbose=1, mode='max',min_delta=0, cooldown=0, min_lr=0.0001)"
      ],
      "execution_count": 36,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-Jxr1W_3tA1P"
      },
      "source": [
        "model1.compile(loss='categorical_crossentropy',\n",
        "              optimizer=Adam(0.001),\n",
        "              metrics=['accuracy'])"
      ],
      "execution_count": 37,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "X6MUJyO7Jtqd",
        "outputId": "e2133087-cd18-45a4-c2b4-20885f2a8edf"
      },
      "source": [
        "history2=model1.fit_generator(datagen.flow(X_train,y_train,batch_size=128),validation_data = (X_test,y_test), steps_per_epoch = 391,epochs=45,callbacks=[DynamicLR])"
      ],
      "execution_count": 38,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/training.py:1844: UserWarning: `Model.fit_generator` is deprecated and will be removed in a future version. Please use `Model.fit`, which supports generators.\n",
            "  warnings.warn('`Model.fit_generator` is deprecated and '\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/45\n",
            "391/391 [==============================] - 47s 106ms/step - loss: 1.9747 - accuracy: 0.2577 - val_loss: 1.6432 - val_accuracy: 0.4016\n",
            "Epoch 2/45\n",
            "391/391 [==============================] - 40s 102ms/step - loss: 1.5056 - accuracy: 0.4402 - val_loss: 1.5382 - val_accuracy: 0.4801\n",
            "Epoch 3/45\n",
            "391/391 [==============================] - 40s 103ms/step - loss: 1.3355 - accuracy: 0.5132 - val_loss: 2.3519 - val_accuracy: 0.3519\n",
            "Epoch 4/45\n",
            "391/391 [==============================] - 40s 102ms/step - loss: 1.2472 - accuracy: 0.5510 - val_loss: 1.2285 - val_accuracy: 0.5668\n",
            "Epoch 5/45\n",
            "391/391 [==============================] - 40s 102ms/step - loss: 1.1888 - accuracy: 0.5658 - val_loss: 1.2924 - val_accuracy: 0.5652\n",
            "Epoch 6/45\n",
            "391/391 [==============================] - 40s 102ms/step - loss: 1.1329 - accuracy: 0.5909 - val_loss: 1.1845 - val_accuracy: 0.5982\n",
            "Epoch 7/45\n",
            "391/391 [==============================] - 40s 103ms/step - loss: 1.0696 - accuracy: 0.6145 - val_loss: 1.3405 - val_accuracy: 0.5740\n",
            "Epoch 8/45\n",
            "391/391 [==============================] - 40s 102ms/step - loss: 1.0578 - accuracy: 0.6194 - val_loss: 1.2276 - val_accuracy: 0.6086\n",
            "Epoch 9/45\n",
            "391/391 [==============================] - 40s 103ms/step - loss: 1.0119 - accuracy: 0.6367 - val_loss: 1.5363 - val_accuracy: 0.5554\n",
            "Epoch 10/45\n",
            "391/391 [==============================] - 40s 103ms/step - loss: 0.9917 - accuracy: 0.6427 - val_loss: 1.0776 - val_accuracy: 0.6398\n",
            "Epoch 11/45\n",
            "391/391 [==============================] - 40s 103ms/step - loss: 0.9691 - accuracy: 0.6527 - val_loss: 1.7840 - val_accuracy: 0.5007\n",
            "Epoch 12/45\n",
            "391/391 [==============================] - 40s 103ms/step - loss: 0.9405 - accuracy: 0.6642 - val_loss: 1.3595 - val_accuracy: 0.6092\n",
            "\n",
            "Epoch 00012: ReduceLROnPlateau reducing learning rate to 0.0009000000427477062.\n",
            "Epoch 13/45\n",
            "391/391 [==============================] - 40s 103ms/step - loss: 0.9167 - accuracy: 0.6702 - val_loss: 1.1870 - val_accuracy: 0.6374\n",
            "Epoch 14/45\n",
            "391/391 [==============================] - 40s 103ms/step - loss: 0.9017 - accuracy: 0.6772 - val_loss: 0.9557 - val_accuracy: 0.6735\n",
            "Epoch 15/45\n",
            "391/391 [==============================] - 40s 103ms/step - loss: 0.8899 - accuracy: 0.6859 - val_loss: 1.0948 - val_accuracy: 0.6485\n",
            "Epoch 16/45\n",
            "391/391 [==============================] - 40s 102ms/step - loss: 0.8643 - accuracy: 0.6904 - val_loss: 1.4840 - val_accuracy: 0.5903\n",
            "\n",
            "Epoch 00016: ReduceLROnPlateau reducing learning rate to 0.0008100000384729356.\n",
            "Epoch 17/45\n",
            "391/391 [==============================] - 40s 103ms/step - loss: 0.8463 - accuracy: 0.6995 - val_loss: 0.9817 - val_accuracy: 0.6801\n",
            "Epoch 18/45\n",
            "391/391 [==============================] - 40s 102ms/step - loss: 0.8377 - accuracy: 0.7021 - val_loss: 0.8720 - val_accuracy: 0.7099\n",
            "Epoch 19/45\n",
            "391/391 [==============================] - 40s 103ms/step - loss: 0.8228 - accuracy: 0.7059 - val_loss: 0.9512 - val_accuracy: 0.6862\n",
            "Epoch 20/45\n",
            "391/391 [==============================] - 40s 103ms/step - loss: 0.8054 - accuracy: 0.7117 - val_loss: 1.1271 - val_accuracy: 0.6664\n",
            "\n",
            "Epoch 00020: ReduceLROnPlateau reducing learning rate to 0.0007290000503417104.\n",
            "Epoch 21/45\n",
            "391/391 [==============================] - 40s 103ms/step - loss: 0.8046 - accuracy: 0.7154 - val_loss: 0.8546 - val_accuracy: 0.7133\n",
            "Epoch 22/45\n",
            "391/391 [==============================] - 40s 103ms/step - loss: 0.7858 - accuracy: 0.7239 - val_loss: 1.1095 - val_accuracy: 0.6678\n",
            "Epoch 23/45\n",
            "391/391 [==============================] - 40s 102ms/step - loss: 0.7807 - accuracy: 0.7227 - val_loss: 1.0355 - val_accuracy: 0.6845\n",
            "\n",
            "Epoch 00023: ReduceLROnPlateau reducing learning rate to 0.0006561000715009868.\n",
            "Epoch 24/45\n",
            "391/391 [==============================] - 40s 103ms/step - loss: 0.7595 - accuracy: 0.7310 - val_loss: 1.0318 - val_accuracy: 0.6872\n",
            "Epoch 25/45\n",
            "391/391 [==============================] - 40s 103ms/step - loss: 0.7526 - accuracy: 0.7347 - val_loss: 1.1684 - val_accuracy: 0.6671\n",
            "\n",
            "Epoch 00025: ReduceLROnPlateau reducing learning rate to 0.0005904900433961303.\n",
            "Epoch 26/45\n",
            "391/391 [==============================] - 40s 103ms/step - loss: 0.7436 - accuracy: 0.7376 - val_loss: 1.1874 - val_accuracy: 0.6698\n",
            "Epoch 27/45\n",
            "391/391 [==============================] - 40s 102ms/step - loss: 0.7253 - accuracy: 0.7452 - val_loss: 0.7713 - val_accuracy: 0.7536\n",
            "Epoch 28/45\n",
            "391/391 [==============================] - 40s 102ms/step - loss: 0.7232 - accuracy: 0.7430 - val_loss: 1.0463 - val_accuracy: 0.6801\n",
            "Epoch 29/45\n",
            "391/391 [==============================] - 40s 102ms/step - loss: 0.7296 - accuracy: 0.7411 - val_loss: 0.8238 - val_accuracy: 0.7401\n",
            "\n",
            "Epoch 00029: ReduceLROnPlateau reducing learning rate to 0.0005314410547725857.\n",
            "Epoch 30/45\n",
            "391/391 [==============================] - 40s 102ms/step - loss: 0.7117 - accuracy: 0.7470 - val_loss: 0.8708 - val_accuracy: 0.7355\n",
            "Epoch 31/45\n",
            "391/391 [==============================] - 40s 103ms/step - loss: 0.7005 - accuracy: 0.7527 - val_loss: 0.9428 - val_accuracy: 0.7170\n",
            "\n",
            "Epoch 00031: ReduceLROnPlateau reducing learning rate to 0.00047829695977270604.\n",
            "Epoch 32/45\n",
            "391/391 [==============================] - 40s 103ms/step - loss: 0.6856 - accuracy: 0.7568 - val_loss: 0.7896 - val_accuracy: 0.7480\n",
            "Epoch 33/45\n",
            "391/391 [==============================] - 40s 102ms/step - loss: 0.6937 - accuracy: 0.7574 - val_loss: 0.8226 - val_accuracy: 0.7441\n",
            "\n",
            "Epoch 00033: ReduceLROnPlateau reducing learning rate to 0.0004304672533180565.\n",
            "Epoch 34/45\n",
            "391/391 [==============================] - 40s 102ms/step - loss: 0.6770 - accuracy: 0.7611 - val_loss: 0.8990 - val_accuracy: 0.7217\n",
            "Epoch 35/45\n",
            "391/391 [==============================] - 40s 103ms/step - loss: 0.6769 - accuracy: 0.7606 - val_loss: 0.7793 - val_accuracy: 0.7602\n",
            "Epoch 36/45\n",
            "391/391 [==============================] - 40s 103ms/step - loss: 0.6846 - accuracy: 0.7587 - val_loss: 0.8666 - val_accuracy: 0.7380\n",
            "Epoch 37/45\n",
            "391/391 [==============================] - 40s 102ms/step - loss: 0.6697 - accuracy: 0.7630 - val_loss: 0.7352 - val_accuracy: 0.7631\n",
            "Epoch 38/45\n",
            "391/391 [==============================] - 40s 102ms/step - loss: 0.6581 - accuracy: 0.7672 - val_loss: 0.8582 - val_accuracy: 0.7379\n",
            "Epoch 39/45\n",
            "391/391 [==============================] - 40s 102ms/step - loss: 0.6608 - accuracy: 0.7683 - val_loss: 0.7176 - val_accuracy: 0.7735\n",
            "Epoch 40/45\n",
            "391/391 [==============================] - 40s 103ms/step - loss: 0.6611 - accuracy: 0.7625 - val_loss: 0.9865 - val_accuracy: 0.7174\n",
            "Epoch 41/45\n",
            "391/391 [==============================] - 40s 103ms/step - loss: 0.6665 - accuracy: 0.7633 - val_loss: 0.7472 - val_accuracy: 0.7627\n",
            "\n",
            "Epoch 00041: ReduceLROnPlateau reducing learning rate to 0.00038742052274756136.\n",
            "Epoch 42/45\n",
            "391/391 [==============================] - 40s 102ms/step - loss: 0.6449 - accuracy: 0.7727 - val_loss: 0.9286 - val_accuracy: 0.7230\n",
            "Epoch 43/45\n",
            "391/391 [==============================] - 40s 102ms/step - loss: 0.6377 - accuracy: 0.7752 - val_loss: 0.8403 - val_accuracy: 0.7397\n",
            "\n",
            "Epoch 00043: ReduceLROnPlateau reducing learning rate to 0.0003486784757114947.\n",
            "Epoch 44/45\n",
            "391/391 [==============================] - 40s 102ms/step - loss: 0.6441 - accuracy: 0.7742 - val_loss: 0.7630 - val_accuracy: 0.7687\n",
            "Epoch 45/45\n",
            "391/391 [==============================] - 40s 103ms/step - loss: 0.6467 - accuracy: 0.7745 - val_loss: 0.8227 - val_accuracy: 0.7503\n",
            "\n",
            "Epoch 00045: ReduceLROnPlateau reducing learning rate to 0.00031381062290165574.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "qanEl3wgeXmc",
        "outputId": "0088d065-7604-4920-e4ed-07663ed9f197"
      },
      "source": [
        "import keras\n",
        "from matplotlib import pyplot as plt\n",
        "plt.plot(history2.history['accuracy'])\n",
        "plt.plot(history2.history['val_accuracy'])\n",
        "plt.title('model accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'val'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 39,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "JooKqawBebeA",
        "outputId": "8fbf9235-bdbf-40f4-8db7-6d5c9adf208e"
      },
      "source": [
        "import keras\n",
        "from matplotlib import pyplot as plt\n",
        "plt.plot(history2.history['loss'])\n",
        "plt.plot(history2.history['val_loss'])\n",
        "plt.title('model loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'val'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": 40,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    }
  ]
}