avg_attention_resampling_noise.ipynb

{
  "cells": [
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "va-QmMMoqrL7",
        "outputId": "d53a9bce-53da-4cf5-f471-a80b04d1f220"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "2.9.2\n"
          ]
        }
      ],
      "source": [
        "from google.colab import drive\n",
        "import os \n",
        "import numpy as np \n",
        "import pandas as pd \n",
        "import seaborn as sns \n",
        "import matplotlib.pyplot as plt\n",
        "import tensorflow as tf \n",
        "import tensorflow.keras as tfk \n",
        "import tensorflow.keras.layers as tfkl \n",
        "plt.rc('font', size=16) \n",
        "import random\n",
        "from sklearn.metrics import accuracy_score, f1_score, precision_score, recall_score\n",
        "from sklearn import model_selection\n",
        "from sklearn.metrics import confusion_matrix\n",
        "from sklearn.utils import class_weight\n",
        "from sklearn.preprocessing import MinMaxScaler\n",
        "import warnings\n",
        "import logging\n",
        "from sklearn import preprocessing \n",
        "\n",
        "\n",
        "tfk = tf.keras\n",
        "tfkl = tf.keras.layers\n",
        "print(tf.__version__)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "id": "AtNSmv7Yu6j_"
      },
      "outputs": [],
      "source": [
        "# Random seed for reproducibility\n",
        "seed = 89\n",
        "random.seed(seed)\n",
        "os.environ['PYTHONHASHSEED'] = str(seed)\n",
        "np.random.seed(seed)\n",
        "tf.random.set_seed(seed)\n",
        "tf.compat.v1.set_random_seed(seed)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "id": "yA8mp8WzX5sj"
      },
      "outputs": [],
      "source": [
        "from google.colab import drive"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_7sucj6TrC-v",
        "outputId": "73595929-18c3-43f3-ec43-70efdbc4cb0f"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Mounted at /gdrive\n"
          ]
        }
      ],
      "source": [
        "drive.mount('/gdrive')\n",
        "os.chdir(r'/gdrive/MyDrive/annHomework2/')\n",
        "x_train = np.load(r'x_train.npy')\n",
        "y_train = np.load(r'y_train.npy')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "0qxzUpvArntW",
        "outputId": "d82f9ea6-636b-4c63-d763-33a9949470fc"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "((2429, 36, 6), (2429,))"
            ]
          },
          "execution_count": 5,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "x_train.shape, y_train.shape "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 291
        },
        "id": "rAprzP2bkm4S",
        "outputId": "feea4c10-d09f-4206-a54c-54ed537843bf"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "# plot labels distribution \n",
        "sns.countplot( x = y_train) \n",
        "plt.title('Dataset Labels Distribution')\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "id": "QKDkjdJ7754A"
      },
      "outputs": [],
      "source": [
        "labels = np.unique(np.fromiter([i for i in y_train], np.int32) )\n",
        "class_weights = class_weight.compute_class_weight(class_weight='balanced', classes=labels, y=y_train) "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ulb_XXpY-qDz",
        "outputId": "ea3444eb-698a-4a5b-b3ba-6fd4037f74ea"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "(2429, 117, 6)\n"
          ]
        }
      ],
      "source": [
        "# interpolation \n",
        "from scipy.interpolate import interp1d\n",
        "x_new = [] \n",
        "for el in x_train: \n",
        "    interp = interp1d(x=np.arange(36), y=el, axis=0, kind='cubic')\n",
        "    el = interp(np.arange(0, 35, .3))\n",
        "    x_new.append(el)\n",
        "x_new = np.array(x_new) \n",
        "print(x_new.shape) \n",
        "x_train = x_new\n",
        "y_train = tfk.utils.to_categorical(y_train) "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "metadata": {
        "id": "3DE3PNbC-dmt"
      },
      "outputs": [],
      "source": [
        "# \n",
        "train_data = np.concatenate([x for x in x_train]) \n",
        "scaler = preprocessing.StandardScaler().fit(train_data)\n",
        "\n",
        "x_train_trans = [] \n",
        "for el in x_train: \n",
        "  x_train_trans.append(scaler.transform(el))\n",
        "del x_train \n",
        "\n",
        "x_train = np.asarray(x_train_trans) \n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 10,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 345
        },
        "id": "e2oMV37Cytys",
        "outputId": "d7409795-fb0b-402a-d954-00e499a2e29a"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "(9324, 702) (9324, 12)\n",
            "(9324, 117, 6) (9324, 12)\n"
          ]
        }
      ],
      "source": [
        "OLD_SHAPE = x_train.shape[1]\n",
        "x_train_reshaped = tf.reshape(x_train, shape = (x_train.shape[0], x_train.shape[1]*6) )\n",
        "\n",
        "from imblearn.over_sampling import RandomOverSampler, SVMSMOTE, SMOTE\n",
        "from imblearn.combine import SMOTEENN\n",
        "\n",
        "ros = SMOTE(sampling_strategy='not majority' ,random_state=seed)\n",
        "train_data, labels = ros.fit_resample(x_train_reshaped, y_train)\n",
        "\n",
        "plotting_labels = np.argmax(labels, axis = 1)\n",
        "sns.countplot( y = plotting_labels) \n",
        "plt.title('Dataset Labels Distribution')\n",
        "plt.show()\n",
        "\n",
        "print(train_data.shape, labels.shape) \n",
        "x_res = tf.reshape(train_data, shape = (train_data.shape[0], OLD_SHAPE, 6) ) \n",
        "y_res = labels \n",
        "print(x_res.shape, y_res.shape)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 11,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "BjouEKJoABlN",
        "outputId": "2265f286-339c-4e8c-8e07-78a25fbfc70d"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "(2429, 117, 6) (2429, 12)\n"
          ]
        },
        {
          "data": {
            "text/plain": [
              "array([1., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.], dtype=float32)"
            ]
          },
          "execution_count": 11,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "# check \n",
        "print(x_train.shape, y_train.shape) \n",
        "x_train[0][1]\n",
        "y_train[0]"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VSP-6T9Vu8BO"
      },
      "source": [
        "**bagging training sets**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 12,
      "metadata": {
        "id": "gaghP_Qxu_id"
      },
      "outputs": [],
      "source": [
        "x_train1, x_test1, y_train1, y_test1 = model_selection.train_test_split(x_train, y_train, test_size=0.2, random_state=1, stratify=y_train)\n",
        "x_train2, x_test2, y_train2, y_test2 = model_selection.train_test_split(x_train, y_train, test_size=0.2, random_state=2, stratify=y_train)\n",
        "x_train3, x_test3, y_train3, y_test3 = model_selection.train_test_split(x_train, y_train, test_size=0.2, random_state=3, stratify=y_train)\n",
        "x_train4, x_test4, y_train4, y_test4 = model_selection.train_test_split(x_train, y_train, test_size=0.2, random_state=4, stratify=y_train)\n",
        "x_train5, x_test5, y_train5, y_test5 = model_selection.train_test_split(x_train, y_train, test_size=0.2, random_state=5, stratify=y_train)\n",
        "x_train6, x_test6, y_train6, y_test6 = model_selection.train_test_split(x_train, y_train, test_size=0.2, random_state=6, stratify=y_train)\n",
        "x_train7, x_test7, y_train7, y_test7 = model_selection.train_test_split(x_train, y_train, test_size=0.2, random_state=7, stratify=y_train)\n",
        "x_train8, x_test8, y_train8, y_test8 = model_selection.train_test_split(x_train, y_train, test_size=0.2, random_state=8, stratify=y_train)\n",
        "x_train9, x_test9, y_train9, y_test9 = model_selection.train_test_split(x_train, y_train, test_size=0.2, random_state=9, stratify=y_train)\n",
        "x_train10, x_test10, y_train10, y_test10 = model_selection.train_test_split(x_train, y_train, test_size=0.2, random_state=10, stratify=y_train)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 13,
      "metadata": {
        "id": "KRI6J1GZuQHw"
      },
      "outputs": [],
      "source": [
        "input_shape = (117,6) "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 14,
      "metadata": {
        "id": "7nxJrPoojt3z"
      },
      "outputs": [],
      "source": [
        "# ubernet : norm + skip + attention \n",
        "def build_ffnn_classifier(input_shape, classes):\n",
        "    input_layer = tfkl.Input(shape=input_shape, name='Input')\n",
        "    x = input_layer\n",
        "    x = tfkl.GaussianNoise(seed=seed, stddev=0.001) (x)\n",
        "    cnn_1 = tfkl.Conv1D(256,3,padding='same',activation='relu')(x) \n",
        "    cnn = tfkl.MaxPooling1D()(cnn_1)\n",
        "    cnn = tfkl.Conv1D(256,3,padding='same',activation='relu')(cnn)\n",
        "    cnn = tfkl.MaxPooling1D()(cnn)\n",
        "    cnn = tfkl.Conv1D(256,3,padding='same',activation='relu')(cnn)\n",
        "    x = tfkl.MultiHeadAttention( key_dim=6, num_heads=5, dropout=0.3 )(cnn,tfkl.MaxPool1D(strides = 2)(input_layer)) \n",
        "    x = tfkl.Add()([cnn, tfkl.MaxPool1D(strides = 4)(cnn_1) ])\n",
        "    \n",
        "    norm = tfkl.BatchNormalization()(x)\n",
        "    \n",
        "    cnn_1 = tfkl.Conv1D(256,3,padding='same',activation='relu')(norm) \n",
        "    cnn = tfkl.MaxPooling1D()(cnn_1)\n",
        "    cnn = tfkl.Conv1D(256,3,padding='same',activation='relu')(cnn)\n",
        "    cnn = tfkl.MaxPooling1D()(cnn)\n",
        "    cnn = tfkl.Conv1D(256,3,padding='same',activation='relu')(cnn)\n",
        "    \n",
        "    x = tfkl.MultiHeadAttention( key_dim=6, num_heads=5, dropout=0.3 )(cnn,tfkl.MaxPool1D(strides = 2)(cnn) ) \n",
        "    x = tfkl.Add()([cnn, tfkl.MaxPool1D(strides = 4)(cnn_1)  ])\n",
        "    norm = tfkl.BatchNormalization()(x)\n",
        "    \n",
        "    gap = tfkl.GlobalAveragePooling1D()(norm)\n",
        "    dropout = tfkl.Dropout(.3, seed=seed)(gap)\n",
        "\n",
        "    hidden_layer1 = tfkl.Dense(units=256, activation ='relu', name = 'Hidden1', kernel_regularizer = tfk.regularizers.L2(l2 = 0.003))(dropout) \n",
        "    dropout = tfkl.Dropout(.3, seed=seed)(hidden_layer1)\n",
        "    hidden_layer2 = tfkl.Dense(units=256, activation ='relu', name = 'Hidden2', kernel_regularizer = tfk.regularizers.L2(l2 = 0.003))(dropout) \n",
        "    \n",
        "    output_layer = tfkl.Dense(classes, activation='softmax', \n",
        "                              kernel_initializer=tfk.initializers.GlorotUniform(seed))(hidden_layer2)\n",
        "      \n",
        "    model = tfk.Model(inputs=input_layer, outputs=output_layer, name='model')\n",
        "    model.compile(loss=tfk.losses.CategoricalCrossentropy(), optimizer=tfk.optimizers.Adam(), metrics='accuracy')\n",
        "    return model"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 15,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "d1zEtb3Hjt1O",
        "outputId": "c8d2ce46-d101-4ba5-b623-cf48361173d4"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Model: \"model\"\n",
            "__________________________________________________________________________________________________\n",
            " Layer (type)                   Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            " Input (InputLayer)             [(None, 117, 6)]     0           []                               \n",
            "                                                                                                  \n",
            " gaussian_noise_6 (GaussianNois  (None, 117, 6)      0           ['Input[0][0]']                  \n",
            " e)                                                                                               \n",
            "                                                                                                  \n",
            " conv1d_36 (Conv1D)             (None, 117, 256)     4864        ['gaussian_noise_6[0][0]']       \n",
            "                                                                                                  \n",
            " max_pooling1d_48 (MaxPooling1D  (None, 58, 256)     0           ['conv1d_36[0][0]']              \n",
            " )                                                                                                \n",
            "                                                                                                  \n",
            " conv1d_37 (Conv1D)             (None, 58, 256)      196864      ['max_pooling1d_48[0][0]']       \n",
            "                                                                                                  \n",
            " max_pooling1d_49 (MaxPooling1D  (None, 29, 256)     0           ['conv1d_37[0][0]']              \n",
            " )                                                                                                \n",
            "                                                                                                  \n",
            " conv1d_38 (Conv1D)             (None, 29, 256)      196864      ['max_pooling1d_49[0][0]']       \n",
            "                                                                                                  \n",
            " max_pooling1d_51 (MaxPooling1D  (None, 29, 256)     0           ['conv1d_36[0][0]']              \n",
            " )                                                                                                \n",
            "                                                                                                  \n",
            " add_12 (Add)                   (None, 29, 256)      0           ['conv1d_38[0][0]',              \n",
            "                                                                  'max_pooling1d_51[0][0]']       \n",
            "                                                                                                  \n",
            " batch_normalization_12 (BatchN  (None, 29, 256)     1024        ['add_12[0][0]']                 \n",
            " ormalization)                                                                                    \n",
            "                                                                                                  \n",
            " conv1d_39 (Conv1D)             (None, 29, 256)      196864      ['batch_normalization_12[0][0]'] \n",
            "                                                                                                  \n",
            " max_pooling1d_52 (MaxPooling1D  (None, 14, 256)     0           ['conv1d_39[0][0]']              \n",
            " )                                                                                                \n",
            "                                                                                                  \n",
            " conv1d_40 (Conv1D)             (None, 14, 256)      196864      ['max_pooling1d_52[0][0]']       \n",
            "                                                                                                  \n",
            " max_pooling1d_53 (MaxPooling1D  (None, 7, 256)      0           ['conv1d_40[0][0]']              \n",
            " )                                                                                                \n",
            "                                                                                                  \n",
            " conv1d_41 (Conv1D)             (None, 7, 256)       196864      ['max_pooling1d_53[0][0]']       \n",
            "                                                                                                  \n",
            " max_pooling1d_55 (MaxPooling1D  (None, 7, 256)      0           ['conv1d_39[0][0]']              \n",
            " )                                                                                                \n",
            "                                                                                                  \n",
            " add_13 (Add)                   (None, 7, 256)       0           ['conv1d_41[0][0]',              \n",
            "                                                                  'max_pooling1d_55[0][0]']       \n",
            "                                                                                                  \n",
            " batch_normalization_13 (BatchN  (None, 7, 256)      1024        ['add_13[0][0]']                 \n",
            " ormalization)                                                                                    \n",
            "                                                                                                  \n",
            " global_average_pooling1d_6 (Gl  (None, 256)         0           ['batch_normalization_13[0][0]'] \n",
            " obalAveragePooling1D)                                                                            \n",
            "                                                                                                  \n",
            " dropout_12 (Dropout)           (None, 256)          0           ['global_average_pooling1d_6[0][0\n",
            "                                                                 ]']                              \n",
            "                                                                                                  \n",
            " Hidden1 (Dense)                (None, 256)          65792       ['dropout_12[0][0]']             \n",
            "                                                                                                  \n",
            " dropout_13 (Dropout)           (None, 256)          0           ['Hidden1[0][0]']                \n",
            "                                                                                                  \n",
            " Hidden2 (Dense)                (None, 256)          65792       ['dropout_13[0][0]']             \n",
            "                                                                                                  \n",
            " dense_6 (Dense)                (None, 12)           3084        ['Hidden2[0][0]']                \n",
            "                                                                                                  \n",
            "==================================================================================================\n",
            "Total params: 1,125,900\n",
            "Trainable params: 1,124,876\n",
            "Non-trainable params: 1,024\n",
            "__________________________________________________________________________________________________\n"
          ]
        }
      ],
      "source": [
        "model1 = build_ffnn_classifier(input_shape, 12)\n",
        "model2 = build_ffnn_classifier(input_shape, 12)\n",
        "model3 = build_ffnn_classifier(input_shape, 12)\n",
        "model4 = build_ffnn_classifier(input_shape, 12)\n",
        "model5 = build_ffnn_classifier(input_shape, 12)\n",
        "model6 = build_ffnn_classifier(input_shape, 12)\n",
        "model7 = build_ffnn_classifier(input_shape, 12)\n",
        "model8 = build_ffnn_classifier(input_shape, 12)\n",
        "model9 = build_ffnn_classifier(input_shape, 12)\n",
        "model10 = build_ffnn_classifier(input_shape, 12)\n",
        "\n",
        "model7.summary()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "G4OjjxFVvwd9"
      },
      "source": [
        "# training"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 16,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "LQOYU2X0jty5",
        "outputId": "af66fe2f-6401-48cb-fcb0-11168adacf7d"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1/200\n",
            "31/31 [==============================] - 12s 45ms/step - loss: 3.5211 - accuracy: 0.3546 - val_loss: 3.7017 - val_accuracy: 0.0926 - lr: 0.0010\n",
            "Epoch 2/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 3.0421 - accuracy: 0.4328 - val_loss: 3.8583 - val_accuracy: 0.1214 - lr: 0.0010\n",
            "Epoch 3/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 2.7269 - accuracy: 0.4725 - val_loss: 3.9496 - val_accuracy: 0.1235 - lr: 0.0010\n",
            "Epoch 4/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 2.5099 - accuracy: 0.5126 - val_loss: 3.6057 - val_accuracy: 0.1255 - lr: 0.0010\n",
            "Epoch 5/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 2.3546 - accuracy: 0.5301 - val_loss: 3.1772 - val_accuracy: 0.1255 - lr: 0.0010\n",
            "Epoch 6/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 2.1597 - accuracy: 0.5630 - val_loss: 3.6010 - val_accuracy: 0.1317 - lr: 0.0010\n",
            "Epoch 7/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 2.0421 - accuracy: 0.5703 - val_loss: 3.3713 - val_accuracy: 0.1379 - lr: 0.0010\n",
            "Epoch 8/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.9021 - accuracy: 0.6089 - val_loss: 3.2361 - val_accuracy: 0.1584 - lr: 0.0010\n",
            "Epoch 9/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.7815 - accuracy: 0.6269 - val_loss: 2.6827 - val_accuracy: 0.2716 - lr: 0.0010\n",
            "Epoch 10/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.6799 - accuracy: 0.6346 - val_loss: 2.1045 - val_accuracy: 0.5514 - lr: 0.0010\n",
            "Epoch 11/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.5912 - accuracy: 0.6454 - val_loss: 2.2714 - val_accuracy: 0.3868 - lr: 0.0010\n",
            "Epoch 12/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.5991 - accuracy: 0.6336 - val_loss: 2.2985 - val_accuracy: 0.4486 - lr: 0.0010\n",
            "Epoch 13/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.4858 - accuracy: 0.6557 - val_loss: 2.3933 - val_accuracy: 0.3560 - lr: 0.0010\n",
            "Epoch 14/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.4451 - accuracy: 0.6541 - val_loss: 1.8062 - val_accuracy: 0.5638 - lr: 0.0010\n",
            "Epoch 15/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.3255 - accuracy: 0.6701 - val_loss: 1.6290 - val_accuracy: 0.5741 - lr: 0.0010\n",
            "Epoch 16/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.3020 - accuracy: 0.6788 - val_loss: 1.6891 - val_accuracy: 0.5556 - lr: 0.0010\n",
            "Epoch 17/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.2485 - accuracy: 0.6871 - val_loss: 1.5520 - val_accuracy: 0.6070 - lr: 0.0010\n",
            "Epoch 18/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.1774 - accuracy: 0.7046 - val_loss: 1.8941 - val_accuracy: 0.4712 - lr: 0.0010\n",
            "Epoch 19/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.1498 - accuracy: 0.7015 - val_loss: 1.5895 - val_accuracy: 0.5761 - lr: 0.0010\n",
            "Epoch 20/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.1037 - accuracy: 0.7108 - val_loss: 1.3352 - val_accuracy: 0.6440 - lr: 0.0010\n",
            "Epoch 21/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.0364 - accuracy: 0.7406 - val_loss: 1.4259 - val_accuracy: 0.6029 - lr: 0.0010\n",
            "Epoch 22/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.0408 - accuracy: 0.7257 - val_loss: 1.4277 - val_accuracy: 0.5700 - lr: 0.0010\n",
            "Epoch 23/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.0229 - accuracy: 0.7262 - val_loss: 1.2543 - val_accuracy: 0.6399 - lr: 0.0010\n",
            "Epoch 24/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.9812 - accuracy: 0.7277 - val_loss: 1.2173 - val_accuracy: 0.6584 - lr: 0.0010\n",
            "Epoch 25/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.9272 - accuracy: 0.7396 - val_loss: 1.3007 - val_accuracy: 0.6317 - lr: 0.0010\n",
            "Epoch 26/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.9061 - accuracy: 0.7576 - val_loss: 1.2553 - val_accuracy: 0.6420 - lr: 0.0010\n",
            "Epoch 27/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8627 - accuracy: 0.7705 - val_loss: 1.2651 - val_accuracy: 0.6214 - lr: 0.0010\n",
            "Epoch 28/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8162 - accuracy: 0.7874 - val_loss: 1.1942 - val_accuracy: 0.6317 - lr: 0.0010\n",
            "Epoch 29/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8199 - accuracy: 0.7735 - val_loss: 1.7389 - val_accuracy: 0.5123 - lr: 0.0010\n",
            "Epoch 30/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.7458 - accuracy: 0.8049 - val_loss: 1.3096 - val_accuracy: 0.6399 - lr: 0.0010\n",
            "Epoch 31/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.7491 - accuracy: 0.7910 - val_loss: 1.4949 - val_accuracy: 0.5700 - lr: 0.0010\n",
            "Epoch 32/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.7174 - accuracy: 0.8029 - val_loss: 1.3589 - val_accuracy: 0.6173 - lr: 0.0010\n",
            "Epoch 33/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.7110 - accuracy: 0.7988 - val_loss: 1.1456 - val_accuracy: 0.6379 - lr: 0.0010\n",
            "Epoch 34/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.6671 - accuracy: 0.8183 - val_loss: 1.2171 - val_accuracy: 0.6420 - lr: 0.0010\n",
            "Epoch 35/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6221 - accuracy: 0.8363 - val_loss: 1.1689 - val_accuracy: 0.6626 - lr: 0.0010\n",
            "Epoch 36/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6050 - accuracy: 0.8379 - val_loss: 1.4003 - val_accuracy: 0.5988 - lr: 0.0010\n",
            "Epoch 37/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.5863 - accuracy: 0.8461 - val_loss: 1.4088 - val_accuracy: 0.5885 - lr: 0.0010\n",
            "Epoch 38/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.5756 - accuracy: 0.8538 - val_loss: 1.2368 - val_accuracy: 0.6708 - lr: 0.0010\n",
            "Epoch 39/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.5858 - accuracy: 0.8446 - val_loss: 1.1747 - val_accuracy: 0.6770 - lr: 0.0010\n",
            "Epoch 40/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.5287 - accuracy: 0.8595 - val_loss: 1.5423 - val_accuracy: 0.5741 - lr: 0.0010\n",
            "Epoch 41/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.5715 - accuracy: 0.8482 - val_loss: 1.1745 - val_accuracy: 0.6770 - lr: 0.0010\n",
            "Epoch 42/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.4743 - accuracy: 0.8883 - val_loss: 1.4030 - val_accuracy: 0.6091 - lr: 0.0010\n",
            "Epoch 43/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.4431 - accuracy: 0.8950 - val_loss: 1.3693 - val_accuracy: 0.6461 - lr: 0.0010\n",
            "Epoch 44/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.4062 - accuracy: 0.9094 - val_loss: 1.3235 - val_accuracy: 0.6749 - lr: 0.0010\n",
            "Epoch 45/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.3738 - accuracy: 0.9218 - val_loss: 1.3872 - val_accuracy: 0.6626 - lr: 0.0010\n",
            "Epoch 46/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3854 - accuracy: 0.9104 - val_loss: 1.7361 - val_accuracy: 0.6008 - lr: 0.0010\n",
            "Epoch 47/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.4226 - accuracy: 0.8966 - val_loss: 1.3905 - val_accuracy: 0.6667 - lr: 0.0010\n",
            "Epoch 48/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.4100 - accuracy: 0.9022 - val_loss: 1.3530 - val_accuracy: 0.6646 - lr: 0.0010\n",
            "Epoch 49/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3736 - accuracy: 0.9125 - val_loss: 1.3639 - val_accuracy: 0.6605 - lr: 0.0010\n",
            "Epoch 50/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3733 - accuracy: 0.9182 - val_loss: 1.3280 - val_accuracy: 0.6667 - lr: 0.0010\n",
            "Epoch 51/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.3252 - accuracy: 0.9346 - val_loss: 1.6115 - val_accuracy: 0.6687 - lr: 0.0010\n",
            "Epoch 52/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.3259 - accuracy: 0.9346 - val_loss: 1.5510 - val_accuracy: 0.6687 - lr: 0.0010\n",
            "Epoch 53/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2853 - accuracy: 0.9403 - val_loss: 1.3069 - val_accuracy: 0.6749 - lr: 0.0010\n",
            "Epoch 54/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.2482 - accuracy: 0.9511 - val_loss: 1.3565 - val_accuracy: 0.6831 - lr: 0.0010\n",
            "Epoch 55/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2375 - accuracy: 0.9568 - val_loss: 1.3351 - val_accuracy: 0.7037 - lr: 0.0010\n",
            "Epoch 56/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.3121 - accuracy: 0.9362 - val_loss: 1.4988 - val_accuracy: 0.6193 - lr: 0.0010\n",
            "Epoch 57/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3133 - accuracy: 0.9285 - val_loss: 1.3505 - val_accuracy: 0.6564 - lr: 0.0010\n",
            "Epoch 58/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.2625 - accuracy: 0.9537 - val_loss: 1.4568 - val_accuracy: 0.6811 - lr: 0.0010\n",
            "Epoch 59/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2882 - accuracy: 0.9388 - val_loss: 1.2856 - val_accuracy: 0.6852 - lr: 0.0010\n",
            "Epoch 60/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2326 - accuracy: 0.9583 - val_loss: 1.2684 - val_accuracy: 0.6934 - lr: 0.0010\n",
            "Epoch 61/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2304 - accuracy: 0.9593 - val_loss: 1.4501 - val_accuracy: 0.6358 - lr: 0.0010\n",
            "Epoch 62/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2467 - accuracy: 0.9547 - val_loss: 1.5303 - val_accuracy: 0.6831 - lr: 0.0010\n",
            "Epoch 63/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2234 - accuracy: 0.9593 - val_loss: 1.4109 - val_accuracy: 0.6543 - lr: 0.0010\n",
            "Epoch 64/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2069 - accuracy: 0.9655 - val_loss: 1.3317 - val_accuracy: 0.6955 - lr: 0.0010\n",
            "Epoch 65/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1723 - accuracy: 0.9810 - val_loss: 1.4557 - val_accuracy: 0.6626 - lr: 0.0010\n",
            "Epoch 66/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2061 - accuracy: 0.9665 - val_loss: 1.5312 - val_accuracy: 0.6111 - lr: 0.0010\n",
            "Epoch 67/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2108 - accuracy: 0.9578 - val_loss: 1.5896 - val_accuracy: 0.6173 - lr: 0.0010\n",
            "Epoch 68/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.2183 - accuracy: 0.9629 - val_loss: 1.4591 - val_accuracy: 0.6605 - lr: 0.0010\n",
            "Epoch 69/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2156 - accuracy: 0.9573 - val_loss: 1.3651 - val_accuracy: 0.6955 - lr: 0.0010\n",
            "Epoch 70/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1913 - accuracy: 0.9686 - val_loss: 1.7019 - val_accuracy: 0.6708 - lr: 0.0010\n",
            "Epoch 71/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1415 - accuracy: 0.9887 - val_loss: 1.3573 - val_accuracy: 0.6893 - lr: 3.0000e-04\n",
            "Epoch 72/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1181 - accuracy: 0.9974 - val_loss: 1.3002 - val_accuracy: 0.7078 - lr: 3.0000e-04\n",
            "Epoch 73/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1134 - accuracy: 0.9959 - val_loss: 1.3471 - val_accuracy: 0.6934 - lr: 3.0000e-04\n",
            "Epoch 74/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1136 - accuracy: 0.9923 - val_loss: 1.3405 - val_accuracy: 0.6996 - lr: 3.0000e-04\n",
            "Epoch 75/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1059 - accuracy: 0.9979 - val_loss: 1.3536 - val_accuracy: 0.7037 - lr: 3.0000e-04\n",
            "Epoch 76/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1013 - accuracy: 0.9974 - val_loss: 1.3430 - val_accuracy: 0.6975 - lr: 3.0000e-04\n",
            "Epoch 77/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0991 - accuracy: 0.9974 - val_loss: 1.2819 - val_accuracy: 0.6975 - lr: 3.0000e-04\n",
            "Epoch 78/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1056 - accuracy: 0.9949 - val_loss: 1.3857 - val_accuracy: 0.6975 - lr: 3.0000e-04\n",
            "Epoch 79/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0983 - accuracy: 0.9974 - val_loss: 1.3787 - val_accuracy: 0.7016 - lr: 3.0000e-04\n",
            "Epoch 80/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0913 - accuracy: 0.9979 - val_loss: 1.3461 - val_accuracy: 0.7016 - lr: 3.0000e-04\n",
            "Epoch 81/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0908 - accuracy: 0.9985 - val_loss: 1.3735 - val_accuracy: 0.6996 - lr: 3.0000e-04\n",
            "Epoch 82/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0897 - accuracy: 0.9990 - val_loss: 1.3445 - val_accuracy: 0.7016 - lr: 3.0000e-04\n",
            "Epoch 83/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0859 - accuracy: 0.9990 - val_loss: 1.3393 - val_accuracy: 0.7119 - lr: 3.0000e-04\n",
            "Epoch 84/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0846 - accuracy: 0.9985 - val_loss: 1.3226 - val_accuracy: 0.7160 - lr: 3.0000e-04\n",
            "Epoch 85/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0817 - accuracy: 1.0000 - val_loss: 1.2979 - val_accuracy: 0.7140 - lr: 3.0000e-04\n",
            "Epoch 86/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0809 - accuracy: 0.9985 - val_loss: 1.2946 - val_accuracy: 0.7037 - lr: 3.0000e-04\n",
            "Epoch 87/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0800 - accuracy: 0.9990 - val_loss: 1.3017 - val_accuracy: 0.7140 - lr: 3.0000e-04\n",
            "Epoch 88/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0789 - accuracy: 0.9990 - val_loss: 1.2743 - val_accuracy: 0.7243 - lr: 3.0000e-04\n",
            "Epoch 89/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0781 - accuracy: 0.9995 - val_loss: 1.2584 - val_accuracy: 0.7058 - lr: 3.0000e-04\n",
            "Epoch 90/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0757 - accuracy: 0.9995 - val_loss: 1.3003 - val_accuracy: 0.6996 - lr: 3.0000e-04\n",
            "Epoch 91/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0752 - accuracy: 0.9990 - val_loss: 1.2615 - val_accuracy: 0.7058 - lr: 3.0000e-04\n",
            "Epoch 92/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0735 - accuracy: 0.9995 - val_loss: 1.2384 - val_accuracy: 0.6975 - lr: 3.0000e-04\n",
            "Epoch 93/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0727 - accuracy: 1.0000 - val_loss: 1.2749 - val_accuracy: 0.7099 - lr: 3.0000e-04\n",
            "Epoch 94/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0729 - accuracy: 0.9990 - val_loss: 1.2579 - val_accuracy: 0.7058 - lr: 3.0000e-04\n",
            "Epoch 95/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0702 - accuracy: 0.9990 - val_loss: 1.2832 - val_accuracy: 0.7099 - lr: 3.0000e-04\n",
            "Epoch 96/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0689 - accuracy: 1.0000 - val_loss: 1.3132 - val_accuracy: 0.6996 - lr: 3.0000e-04\n",
            "Epoch 97/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0701 - accuracy: 1.0000 - val_loss: 1.2757 - val_accuracy: 0.7222 - lr: 3.0000e-04\n",
            "Epoch 98/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0707 - accuracy: 0.9979 - val_loss: 1.2397 - val_accuracy: 0.7181 - lr: 3.0000e-04\n",
            "Epoch 99/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0678 - accuracy: 0.9990 - val_loss: 1.2570 - val_accuracy: 0.7037 - lr: 3.0000e-04\n",
            "Epoch 100/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0655 - accuracy: 0.9995 - val_loss: 1.2407 - val_accuracy: 0.7181 - lr: 3.0000e-04\n",
            "Epoch 101/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0656 - accuracy: 1.0000 - val_loss: 1.2288 - val_accuracy: 0.7058 - lr: 3.0000e-04\n",
            "Epoch 102/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0634 - accuracy: 0.9995 - val_loss: 1.2353 - val_accuracy: 0.7140 - lr: 3.0000e-04\n",
            "Epoch 103/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0631 - accuracy: 0.9990 - val_loss: 1.2584 - val_accuracy: 0.7140 - lr: 3.0000e-04\n",
            "Epoch 104/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0631 - accuracy: 1.0000 - val_loss: 1.2462 - val_accuracy: 0.7140 - lr: 9.0000e-05\n",
            "Epoch 105/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0610 - accuracy: 1.0000 - val_loss: 1.2412 - val_accuracy: 0.7119 - lr: 9.0000e-05\n",
            "Epoch 106/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0607 - accuracy: 1.0000 - val_loss: 1.2548 - val_accuracy: 0.7078 - lr: 9.0000e-05\n",
            "Epoch 107/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0609 - accuracy: 1.0000 - val_loss: 1.2526 - val_accuracy: 0.7037 - lr: 9.0000e-05\n",
            "Epoch 108/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0609 - accuracy: 1.0000 - val_loss: 1.2454 - val_accuracy: 0.7058 - lr: 9.0000e-05\n",
            "Epoch 109/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0604 - accuracy: 1.0000 - val_loss: 1.2557 - val_accuracy: 0.7037 - lr: 9.0000e-05\n",
            "Epoch 110/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0612 - accuracy: 0.9990 - val_loss: 1.2370 - val_accuracy: 0.7037 - lr: 9.0000e-05\n",
            "Epoch 111/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0618 - accuracy: 0.9995 - val_loss: 1.2263 - val_accuracy: 0.7058 - lr: 9.0000e-05\n",
            "Epoch 112/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0604 - accuracy: 0.9985 - val_loss: 1.2322 - val_accuracy: 0.7160 - lr: 9.0000e-05\n",
            "Epoch 113/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0606 - accuracy: 1.0000 - val_loss: 1.2437 - val_accuracy: 0.7016 - lr: 9.0000e-05\n",
            "Epoch 114/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0589 - accuracy: 1.0000 - val_loss: 1.2794 - val_accuracy: 0.7016 - lr: 9.0000e-05\n",
            "Epoch 115/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0597 - accuracy: 0.9990 - val_loss: 1.2618 - val_accuracy: 0.7160 - lr: 9.0000e-05\n",
            "Epoch 116/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0596 - accuracy: 0.9985 - val_loss: 1.2302 - val_accuracy: 0.7099 - lr: 9.0000e-05\n",
            "Epoch 117/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0597 - accuracy: 0.9995 - val_loss: 1.2470 - val_accuracy: 0.7099 - lr: 9.0000e-05\n",
            "Epoch 118/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0585 - accuracy: 1.0000 - val_loss: 1.2523 - val_accuracy: 0.7099 - lr: 9.0000e-05\n"
          ]
        }
      ],
      "source": [
        "history1 = model1.fit(\n",
        "    x = x_train1,\n",
        "    y = y_train1,\n",
        "    batch_size = 64,\n",
        "    epochs = 200,\n",
        "    validation_data = (x_test1, y_test1),\n",
        "    callbacks = [\n",
        "        tfk.callbacks.EarlyStopping(monitor='val_accuracy', mode='max', patience=30, restore_best_weights=True),\n",
        "        tfk.callbacks.ReduceLROnPlateau(monitor='val_accuracy', mode='max', patience=15, factor=0.3, min_lr=1e-5)\n",
        "    ]\n",
        ").history"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 17,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "fDWPFIo6wO1x",
        "outputId": "af8de63a-9569-4cad-d3b7-a55a2ba9801a"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1/200\n",
            "31/31 [==============================] - 2s 25ms/step - loss: 3.4685 - accuracy: 0.3659 - val_loss: 3.6254 - val_accuracy: 0.1584 - lr: 0.0010\n",
            "Epoch 2/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 3.0177 - accuracy: 0.4467 - val_loss: 3.6707 - val_accuracy: 0.1337 - lr: 0.0010\n",
            "Epoch 3/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 2.7518 - accuracy: 0.4725 - val_loss: 3.5011 - val_accuracy: 0.1461 - lr: 0.0010\n",
            "Epoch 4/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 2.5430 - accuracy: 0.5080 - val_loss: 3.0963 - val_accuracy: 0.2819 - lr: 0.0010\n",
            "Epoch 5/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.3219 - accuracy: 0.5564 - val_loss: 3.1574 - val_accuracy: 0.2695 - lr: 0.0010\n",
            "Epoch 6/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.1677 - accuracy: 0.5780 - val_loss: 2.7649 - val_accuracy: 0.4198 - lr: 0.0010\n",
            "Epoch 7/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 2.0313 - accuracy: 0.5991 - val_loss: 2.9085 - val_accuracy: 0.3374 - lr: 0.0010\n",
            "Epoch 8/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.8980 - accuracy: 0.6217 - val_loss: 2.8284 - val_accuracy: 0.2695 - lr: 0.0010\n",
            "Epoch 9/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.8187 - accuracy: 0.6145 - val_loss: 2.6660 - val_accuracy: 0.3045 - lr: 0.0010\n",
            "Epoch 10/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.7198 - accuracy: 0.6269 - val_loss: 2.3751 - val_accuracy: 0.4650 - lr: 0.0010\n",
            "Epoch 11/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.7251 - accuracy: 0.6145 - val_loss: 2.6184 - val_accuracy: 0.2716 - lr: 0.0010\n",
            "Epoch 12/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.6165 - accuracy: 0.6202 - val_loss: 1.9042 - val_accuracy: 0.5679 - lr: 0.0010\n",
            "Epoch 13/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.4823 - accuracy: 0.6639 - val_loss: 1.9426 - val_accuracy: 0.5453 - lr: 0.0010\n",
            "Epoch 14/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.3939 - accuracy: 0.6732 - val_loss: 1.7545 - val_accuracy: 0.5658 - lr: 0.0010\n",
            "Epoch 15/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.3594 - accuracy: 0.6711 - val_loss: 1.7857 - val_accuracy: 0.5370 - lr: 0.0010\n",
            "Epoch 16/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.2974 - accuracy: 0.6933 - val_loss: 1.5802 - val_accuracy: 0.6132 - lr: 0.0010\n",
            "Epoch 17/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.2363 - accuracy: 0.6974 - val_loss: 1.4383 - val_accuracy: 0.6523 - lr: 0.0010\n",
            "Epoch 18/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.1672 - accuracy: 0.7144 - val_loss: 1.5365 - val_accuracy: 0.6255 - lr: 0.0010\n",
            "Epoch 19/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.1387 - accuracy: 0.7210 - val_loss: 1.4115 - val_accuracy: 0.6296 - lr: 0.0010\n",
            "Epoch 20/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.0817 - accuracy: 0.7236 - val_loss: 1.3311 - val_accuracy: 0.6379 - lr: 0.0010\n",
            "Epoch 21/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.0348 - accuracy: 0.7447 - val_loss: 1.3417 - val_accuracy: 0.6358 - lr: 0.0010\n",
            "Epoch 22/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.0205 - accuracy: 0.7313 - val_loss: 1.4086 - val_accuracy: 0.6255 - lr: 0.0010\n",
            "Epoch 23/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.9655 - accuracy: 0.7560 - val_loss: 1.5324 - val_accuracy: 0.5947 - lr: 0.0010\n",
            "Epoch 24/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.9524 - accuracy: 0.7545 - val_loss: 1.3766 - val_accuracy: 0.6193 - lr: 0.0010\n",
            "Epoch 25/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8771 - accuracy: 0.7710 - val_loss: 1.2986 - val_accuracy: 0.6770 - lr: 0.0010\n",
            "Epoch 26/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8261 - accuracy: 0.7823 - val_loss: 1.2517 - val_accuracy: 0.6481 - lr: 0.0010\n",
            "Epoch 27/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.7827 - accuracy: 0.8034 - val_loss: 1.6977 - val_accuracy: 0.5638 - lr: 0.0010\n",
            "Epoch 28/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8395 - accuracy: 0.7787 - val_loss: 1.3124 - val_accuracy: 0.6317 - lr: 0.0010\n",
            "Epoch 29/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.7716 - accuracy: 0.7900 - val_loss: 1.2269 - val_accuracy: 0.6605 - lr: 0.0010\n",
            "Epoch 30/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6867 - accuracy: 0.8183 - val_loss: 1.2235 - val_accuracy: 0.6543 - lr: 0.0010\n",
            "Epoch 31/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.6536 - accuracy: 0.8343 - val_loss: 1.4041 - val_accuracy: 0.6440 - lr: 0.0010\n",
            "Epoch 32/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.7145 - accuracy: 0.8163 - val_loss: 1.1912 - val_accuracy: 0.6749 - lr: 0.0010\n",
            "Epoch 33/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.6439 - accuracy: 0.8266 - val_loss: 1.3823 - val_accuracy: 0.6296 - lr: 0.0010\n",
            "Epoch 34/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6197 - accuracy: 0.8384 - val_loss: 1.4163 - val_accuracy: 0.6296 - lr: 0.0010\n",
            "Epoch 35/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.5955 - accuracy: 0.8451 - val_loss: 1.4516 - val_accuracy: 0.6296 - lr: 0.0010\n",
            "Epoch 36/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.5977 - accuracy: 0.8528 - val_loss: 1.3636 - val_accuracy: 0.6626 - lr: 0.0010\n",
            "Epoch 37/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.5332 - accuracy: 0.8652 - val_loss: 1.2533 - val_accuracy: 0.6687 - lr: 0.0010\n",
            "Epoch 38/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.5381 - accuracy: 0.8724 - val_loss: 1.2839 - val_accuracy: 0.6564 - lr: 0.0010\n",
            "Epoch 39/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.4861 - accuracy: 0.8796 - val_loss: 1.2596 - val_accuracy: 0.6708 - lr: 0.0010\n",
            "Epoch 40/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.4894 - accuracy: 0.8873 - val_loss: 1.3532 - val_accuracy: 0.6523 - lr: 0.0010\n",
            "Epoch 41/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.4026 - accuracy: 0.9032 - val_loss: 1.1617 - val_accuracy: 0.6934 - lr: 5.0000e-04\n",
            "Epoch 42/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3277 - accuracy: 0.9362 - val_loss: 1.2185 - val_accuracy: 0.6728 - lr: 5.0000e-04\n",
            "Epoch 43/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.3004 - accuracy: 0.9475 - val_loss: 1.3341 - val_accuracy: 0.6728 - lr: 5.0000e-04\n",
            "Epoch 44/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2873 - accuracy: 0.9527 - val_loss: 1.4337 - val_accuracy: 0.6626 - lr: 5.0000e-04\n",
            "Epoch 45/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2719 - accuracy: 0.9532 - val_loss: 1.2884 - val_accuracy: 0.6626 - lr: 5.0000e-04\n",
            "Epoch 46/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2643 - accuracy: 0.9557 - val_loss: 1.2245 - val_accuracy: 0.6708 - lr: 5.0000e-04\n",
            "Epoch 47/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2563 - accuracy: 0.9557 - val_loss: 1.3838 - val_accuracy: 0.6646 - lr: 5.0000e-04\n",
            "Epoch 48/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2463 - accuracy: 0.9588 - val_loss: 1.2917 - val_accuracy: 0.6728 - lr: 5.0000e-04\n",
            "Epoch 49/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2485 - accuracy: 0.9583 - val_loss: 1.3084 - val_accuracy: 0.6646 - lr: 5.0000e-04\n",
            "Epoch 50/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2292 - accuracy: 0.9635 - val_loss: 1.3305 - val_accuracy: 0.6646 - lr: 5.0000e-04\n",
            "Epoch 51/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2114 - accuracy: 0.9712 - val_loss: 1.6034 - val_accuracy: 0.6749 - lr: 5.0000e-04\n",
            "Epoch 52/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2050 - accuracy: 0.9738 - val_loss: 1.4671 - val_accuracy: 0.6667 - lr: 5.0000e-04\n",
            "Epoch 53/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2058 - accuracy: 0.9701 - val_loss: 1.7840 - val_accuracy: 0.6358 - lr: 5.0000e-04\n",
            "Epoch 54/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.2080 - accuracy: 0.9681 - val_loss: 1.3935 - val_accuracy: 0.6749 - lr: 5.0000e-04\n",
            "Epoch 55/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2055 - accuracy: 0.9681 - val_loss: 1.3552 - val_accuracy: 0.6687 - lr: 5.0000e-04\n",
            "Epoch 56/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1851 - accuracy: 0.9738 - val_loss: 1.3538 - val_accuracy: 0.6667 - lr: 5.0000e-04\n",
            "Epoch 57/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1676 - accuracy: 0.9810 - val_loss: 1.3940 - val_accuracy: 0.6852 - lr: 2.5000e-04\n",
            "Epoch 58/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1452 - accuracy: 0.9887 - val_loss: 1.3154 - val_accuracy: 0.6811 - lr: 2.5000e-04\n",
            "Epoch 59/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1398 - accuracy: 0.9907 - val_loss: 1.3491 - val_accuracy: 0.6811 - lr: 2.5000e-04\n",
            "Epoch 60/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1472 - accuracy: 0.9861 - val_loss: 1.4868 - val_accuracy: 0.6811 - lr: 2.5000e-04\n",
            "Epoch 61/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1349 - accuracy: 0.9876 - val_loss: 1.3765 - val_accuracy: 0.6852 - lr: 2.5000e-04\n",
            "Epoch 62/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1349 - accuracy: 0.9918 - val_loss: 1.4144 - val_accuracy: 0.6728 - lr: 2.5000e-04\n",
            "Epoch 63/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1405 - accuracy: 0.9887 - val_loss: 1.3429 - val_accuracy: 0.6831 - lr: 2.5000e-04\n",
            "Epoch 64/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1383 - accuracy: 0.9882 - val_loss: 1.3954 - val_accuracy: 0.6893 - lr: 2.5000e-04\n",
            "Epoch 65/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1259 - accuracy: 0.9928 - val_loss: 1.4113 - val_accuracy: 0.6811 - lr: 2.5000e-04\n",
            "Epoch 66/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1184 - accuracy: 0.9969 - val_loss: 1.4041 - val_accuracy: 0.6831 - lr: 2.5000e-04\n",
            "Epoch 67/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1204 - accuracy: 0.9928 - val_loss: 1.4409 - val_accuracy: 0.6728 - lr: 2.5000e-04\n",
            "Epoch 68/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1234 - accuracy: 0.9918 - val_loss: 1.4708 - val_accuracy: 0.6749 - lr: 2.5000e-04\n",
            "Epoch 69/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1125 - accuracy: 0.9974 - val_loss: 1.4393 - val_accuracy: 0.6831 - lr: 2.5000e-04\n",
            "Epoch 70/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1072 - accuracy: 0.9964 - val_loss: 1.4129 - val_accuracy: 0.6934 - lr: 2.5000e-04\n",
            "Epoch 71/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1125 - accuracy: 0.9959 - val_loss: 1.4299 - val_accuracy: 0.6893 - lr: 2.5000e-04\n"
          ]
        }
      ],
      "source": [
        "history2 = model2.fit(\n",
        "    x = x_train2,\n",
        "    y = y_train2,\n",
        "    batch_size = 64,\n",
        "    epochs = 200,\n",
        "    validation_data = (x_test2, y_test2),\n",
        "    callbacks = [\n",
        "        tfk.callbacks.EarlyStopping(monitor='val_accuracy', mode='max', patience=30, restore_best_weights=True),\n",
        "        tfk.callbacks.ReduceLROnPlateau(monitor='val_accuracy', mode='max', patience=15, factor=0.5, min_lr=1e-5)\n",
        "    ]\n",
        ").history"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 18,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "iICwi6igwOxQ",
        "outputId": "311e6c97-8b77-44db-e730-5aaca1c66c4b"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1/200\n",
            "31/31 [==============================] - 2s 27ms/step - loss: 3.4969 - accuracy: 0.3474 - val_loss: 3.6631 - val_accuracy: 0.1708 - lr: 0.0010\n",
            "Epoch 2/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 3.0687 - accuracy: 0.4303 - val_loss: 3.5690 - val_accuracy: 0.2016 - lr: 0.0010\n",
            "Epoch 3/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.7542 - accuracy: 0.4838 - val_loss: 3.4947 - val_accuracy: 0.1975 - lr: 0.0010\n",
            "Epoch 4/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.5219 - accuracy: 0.5162 - val_loss: 3.5620 - val_accuracy: 0.1852 - lr: 0.0010\n",
            "Epoch 5/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 2.3469 - accuracy: 0.5502 - val_loss: 3.4874 - val_accuracy: 0.1872 - lr: 0.0010\n",
            "Epoch 6/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 2.2065 - accuracy: 0.5636 - val_loss: 3.1039 - val_accuracy: 0.2222 - lr: 0.0010\n",
            "Epoch 7/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.0746 - accuracy: 0.5841 - val_loss: 2.6534 - val_accuracy: 0.2284 - lr: 0.0010\n",
            "Epoch 8/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.9215 - accuracy: 0.5991 - val_loss: 2.9307 - val_accuracy: 0.2366 - lr: 0.0010\n",
            "Epoch 9/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.8320 - accuracy: 0.6197 - val_loss: 2.5825 - val_accuracy: 0.2819 - lr: 0.0010\n",
            "Epoch 10/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.7193 - accuracy: 0.6310 - val_loss: 3.1897 - val_accuracy: 0.2387 - lr: 0.0010\n",
            "Epoch 11/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.5977 - accuracy: 0.6608 - val_loss: 2.1437 - val_accuracy: 0.4362 - lr: 0.0010\n",
            "Epoch 12/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.5517 - accuracy: 0.6495 - val_loss: 2.0303 - val_accuracy: 0.4733 - lr: 0.0010\n",
            "Epoch 13/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.4741 - accuracy: 0.6629 - val_loss: 1.8404 - val_accuracy: 0.5453 - lr: 0.0010\n",
            "Epoch 14/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.4444 - accuracy: 0.6613 - val_loss: 2.0101 - val_accuracy: 0.4259 - lr: 0.0010\n",
            "Epoch 15/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.3169 - accuracy: 0.6927 - val_loss: 1.9006 - val_accuracy: 0.4733 - lr: 0.0010\n",
            "Epoch 16/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.2699 - accuracy: 0.7046 - val_loss: 1.6943 - val_accuracy: 0.5412 - lr: 0.0010\n",
            "Epoch 17/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.2421 - accuracy: 0.6994 - val_loss: 1.6861 - val_accuracy: 0.5391 - lr: 0.0010\n",
            "Epoch 18/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.2572 - accuracy: 0.6773 - val_loss: 1.5818 - val_accuracy: 0.5432 - lr: 0.0010\n",
            "Epoch 19/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.1409 - accuracy: 0.7247 - val_loss: 1.4879 - val_accuracy: 0.5782 - lr: 0.0010\n",
            "Epoch 20/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.0781 - accuracy: 0.7293 - val_loss: 1.5674 - val_accuracy: 0.5679 - lr: 0.0010\n",
            "Epoch 21/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.0216 - accuracy: 0.7447 - val_loss: 1.4664 - val_accuracy: 0.5905 - lr: 0.0010\n",
            "Epoch 22/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.0098 - accuracy: 0.7355 - val_loss: 1.4721 - val_accuracy: 0.6008 - lr: 0.0010\n",
            "Epoch 23/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.9926 - accuracy: 0.7401 - val_loss: 1.2898 - val_accuracy: 0.6317 - lr: 0.0010\n",
            "Epoch 24/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.9598 - accuracy: 0.7488 - val_loss: 1.3638 - val_accuracy: 0.6173 - lr: 0.0010\n",
            "Epoch 25/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.9189 - accuracy: 0.7627 - val_loss: 1.3587 - val_accuracy: 0.5967 - lr: 0.0010\n",
            "Epoch 26/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.8606 - accuracy: 0.7741 - val_loss: 1.4679 - val_accuracy: 0.5823 - lr: 0.0010\n",
            "Epoch 27/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8353 - accuracy: 0.7771 - val_loss: 1.2688 - val_accuracy: 0.6481 - lr: 0.0010\n",
            "Epoch 28/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.7994 - accuracy: 0.7895 - val_loss: 1.4807 - val_accuracy: 0.6091 - lr: 0.0010\n",
            "Epoch 29/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.7611 - accuracy: 0.8065 - val_loss: 1.2581 - val_accuracy: 0.6502 - lr: 0.0010\n",
            "Epoch 30/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.7440 - accuracy: 0.8034 - val_loss: 1.2234 - val_accuracy: 0.6420 - lr: 0.0010\n",
            "Epoch 31/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.7233 - accuracy: 0.8111 - val_loss: 1.2602 - val_accuracy: 0.6502 - lr: 0.0010\n",
            "Epoch 32/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.7247 - accuracy: 0.8127 - val_loss: 1.3486 - val_accuracy: 0.5720 - lr: 0.0010\n",
            "Epoch 33/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6371 - accuracy: 0.8374 - val_loss: 1.4447 - val_accuracy: 0.6008 - lr: 0.0010\n",
            "Epoch 34/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.7000 - accuracy: 0.8147 - val_loss: 1.2429 - val_accuracy: 0.6584 - lr: 0.0010\n",
            "Epoch 35/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6310 - accuracy: 0.8369 - val_loss: 1.2522 - val_accuracy: 0.6440 - lr: 0.0010\n",
            "Epoch 36/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.5705 - accuracy: 0.8600 - val_loss: 1.3334 - val_accuracy: 0.6214 - lr: 0.0010\n",
            "Epoch 37/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.5460 - accuracy: 0.8646 - val_loss: 1.2976 - val_accuracy: 0.6255 - lr: 0.0010\n",
            "Epoch 38/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.5194 - accuracy: 0.8729 - val_loss: 1.3372 - val_accuracy: 0.6337 - lr: 0.0010\n",
            "Epoch 39/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.4872 - accuracy: 0.8832 - val_loss: 1.3048 - val_accuracy: 0.6420 - lr: 0.0010\n",
            "Epoch 40/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.4487 - accuracy: 0.8955 - val_loss: 1.4433 - val_accuracy: 0.6111 - lr: 0.0010\n",
            "Epoch 41/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.4784 - accuracy: 0.8811 - val_loss: 1.3920 - val_accuracy: 0.6173 - lr: 0.0010\n",
            "Epoch 42/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.4815 - accuracy: 0.8821 - val_loss: 1.5089 - val_accuracy: 0.6214 - lr: 0.0010\n",
            "Epoch 43/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.4459 - accuracy: 0.8976 - val_loss: 1.5561 - val_accuracy: 0.6379 - lr: 0.0010\n",
            "Epoch 44/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.4546 - accuracy: 0.8919 - val_loss: 1.3865 - val_accuracy: 0.6440 - lr: 0.0010\n",
            "Epoch 45/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.4384 - accuracy: 0.8971 - val_loss: 1.5551 - val_accuracy: 0.5885 - lr: 0.0010\n",
            "Epoch 46/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.4292 - accuracy: 0.9017 - val_loss: 1.4443 - val_accuracy: 0.6564 - lr: 0.0010\n",
            "Epoch 47/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3752 - accuracy: 0.9115 - val_loss: 1.4182 - val_accuracy: 0.6358 - lr: 0.0010\n",
            "Epoch 48/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3291 - accuracy: 0.9372 - val_loss: 1.3899 - val_accuracy: 0.6605 - lr: 0.0010\n",
            "Epoch 49/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3298 - accuracy: 0.9357 - val_loss: 1.4428 - val_accuracy: 0.6276 - lr: 0.0010\n",
            "Epoch 50/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.3224 - accuracy: 0.9393 - val_loss: 1.4591 - val_accuracy: 0.6296 - lr: 0.0010\n",
            "Epoch 51/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.3501 - accuracy: 0.9300 - val_loss: 1.5319 - val_accuracy: 0.6420 - lr: 0.0010\n",
            "Epoch 52/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.2883 - accuracy: 0.9475 - val_loss: 1.7990 - val_accuracy: 0.6337 - lr: 0.0010\n",
            "Epoch 53/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2743 - accuracy: 0.9501 - val_loss: 1.5492 - val_accuracy: 0.6358 - lr: 0.0010\n",
            "Epoch 54/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2691 - accuracy: 0.9485 - val_loss: 1.3817 - val_accuracy: 0.6543 - lr: 0.0010\n",
            "Epoch 55/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2518 - accuracy: 0.9532 - val_loss: 1.6497 - val_accuracy: 0.6173 - lr: 0.0010\n",
            "Epoch 56/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2306 - accuracy: 0.9619 - val_loss: 1.7091 - val_accuracy: 0.6029 - lr: 0.0010\n",
            "Epoch 57/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2456 - accuracy: 0.9537 - val_loss: 1.6473 - val_accuracy: 0.6276 - lr: 0.0010\n",
            "Epoch 58/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2216 - accuracy: 0.9609 - val_loss: 1.5586 - val_accuracy: 0.6502 - lr: 0.0010\n",
            "Epoch 59/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2111 - accuracy: 0.9635 - val_loss: 1.4724 - val_accuracy: 0.6605 - lr: 0.0010\n",
            "Epoch 60/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1925 - accuracy: 0.9727 - val_loss: 1.5598 - val_accuracy: 0.6626 - lr: 0.0010\n",
            "Epoch 61/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1951 - accuracy: 0.9691 - val_loss: 1.6954 - val_accuracy: 0.6543 - lr: 0.0010\n",
            "Epoch 62/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1968 - accuracy: 0.9691 - val_loss: 1.5165 - val_accuracy: 0.6523 - lr: 0.0010\n",
            "Epoch 63/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1778 - accuracy: 0.9722 - val_loss: 1.6591 - val_accuracy: 0.6687 - lr: 0.0010\n",
            "Epoch 64/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1848 - accuracy: 0.9727 - val_loss: 1.6678 - val_accuracy: 0.6626 - lr: 0.0010\n",
            "Epoch 65/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2155 - accuracy: 0.9573 - val_loss: 1.4297 - val_accuracy: 0.6770 - lr: 0.0010\n",
            "Epoch 66/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2208 - accuracy: 0.9609 - val_loss: 1.5308 - val_accuracy: 0.6399 - lr: 0.0010\n",
            "Epoch 67/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.2141 - accuracy: 0.9614 - val_loss: 1.4478 - val_accuracy: 0.6461 - lr: 0.0010\n",
            "Epoch 68/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2012 - accuracy: 0.9686 - val_loss: 1.6772 - val_accuracy: 0.6584 - lr: 0.0010\n",
            "Epoch 69/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1723 - accuracy: 0.9768 - val_loss: 1.8555 - val_accuracy: 0.6420 - lr: 0.0010\n",
            "Epoch 70/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1688 - accuracy: 0.9748 - val_loss: 1.7307 - val_accuracy: 0.6255 - lr: 0.0010\n",
            "Epoch 71/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1641 - accuracy: 0.9763 - val_loss: 1.6528 - val_accuracy: 0.6399 - lr: 0.0010\n",
            "Epoch 72/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1391 - accuracy: 0.9861 - val_loss: 1.6277 - val_accuracy: 0.6152 - lr: 0.0010\n",
            "Epoch 73/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1299 - accuracy: 0.9892 - val_loss: 1.6307 - val_accuracy: 0.6399 - lr: 0.0010\n",
            "Epoch 74/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1407 - accuracy: 0.9804 - val_loss: 2.0412 - val_accuracy: 0.5947 - lr: 0.0010\n",
            "Epoch 75/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1585 - accuracy: 0.9763 - val_loss: 1.7679 - val_accuracy: 0.6564 - lr: 0.0010\n",
            "Epoch 76/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1497 - accuracy: 0.9799 - val_loss: 1.8887 - val_accuracy: 0.6296 - lr: 0.0010\n",
            "Epoch 77/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1627 - accuracy: 0.9732 - val_loss: 1.8956 - val_accuracy: 0.6379 - lr: 0.0010\n",
            "Epoch 78/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1239 - accuracy: 0.9871 - val_loss: 1.6987 - val_accuracy: 0.6276 - lr: 0.0010\n",
            "Epoch 79/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1109 - accuracy: 0.9907 - val_loss: 1.6993 - val_accuracy: 0.6337 - lr: 0.0010\n",
            "Epoch 80/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1165 - accuracy: 0.9876 - val_loss: 1.6491 - val_accuracy: 0.6317 - lr: 0.0010\n",
            "Epoch 81/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1010 - accuracy: 0.9928 - val_loss: 1.6595 - val_accuracy: 0.6502 - lr: 5.0000e-04\n",
            "Epoch 82/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0851 - accuracy: 0.9985 - val_loss: 1.6573 - val_accuracy: 0.6543 - lr: 5.0000e-04\n",
            "Epoch 83/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0852 - accuracy: 0.9964 - val_loss: 1.6646 - val_accuracy: 0.6626 - lr: 5.0000e-04\n",
            "Epoch 84/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0769 - accuracy: 0.9995 - val_loss: 1.6261 - val_accuracy: 0.6687 - lr: 5.0000e-04\n",
            "Epoch 85/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0746 - accuracy: 0.9995 - val_loss: 1.5778 - val_accuracy: 0.6749 - lr: 5.0000e-04\n",
            "Epoch 86/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0715 - accuracy: 0.9990 - val_loss: 1.6130 - val_accuracy: 0.6728 - lr: 5.0000e-04\n",
            "Epoch 87/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0713 - accuracy: 0.9985 - val_loss: 1.5558 - val_accuracy: 0.6605 - lr: 5.0000e-04\n",
            "Epoch 88/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0671 - accuracy: 0.9995 - val_loss: 1.5910 - val_accuracy: 0.6605 - lr: 5.0000e-04\n",
            "Epoch 89/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0676 - accuracy: 0.9985 - val_loss: 1.5153 - val_accuracy: 0.6626 - lr: 5.0000e-04\n",
            "Epoch 90/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0656 - accuracy: 0.9985 - val_loss: 1.6202 - val_accuracy: 0.6523 - lr: 5.0000e-04\n",
            "Epoch 91/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0650 - accuracy: 0.9995 - val_loss: 1.5166 - val_accuracy: 0.6708 - lr: 5.0000e-04\n",
            "Epoch 92/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0638 - accuracy: 0.9995 - val_loss: 1.5583 - val_accuracy: 0.6605 - lr: 5.0000e-04\n",
            "Epoch 93/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0608 - accuracy: 1.0000 - val_loss: 1.5100 - val_accuracy: 0.6584 - lr: 5.0000e-04\n",
            "Epoch 94/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0595 - accuracy: 1.0000 - val_loss: 1.5410 - val_accuracy: 0.6584 - lr: 5.0000e-04\n",
            "Epoch 95/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0582 - accuracy: 1.0000 - val_loss: 1.5419 - val_accuracy: 0.6605 - lr: 5.0000e-04\n"
          ]
        }
      ],
      "source": [
        "history3 = model3.fit(\n",
        "    x = x_train3,\n",
        "    y = y_train3,\n",
        "    batch_size = 64,\n",
        "    epochs = 200,\n",
        "    validation_data = (x_test3, y_test3),\n",
        "    callbacks = [\n",
        "        tfk.callbacks.EarlyStopping(monitor='val_accuracy', mode='max', patience=30, restore_best_weights=True),\n",
        "        tfk.callbacks.ReduceLROnPlateau(monitor='val_accuracy', mode='max', patience=15, factor=0.5, min_lr=1e-5)\n",
        "    ]\n",
        ").history"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 19,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4B69wzwBwOtE",
        "outputId": "84d8919d-aa67-4f24-d11e-ce0140d205bb"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1/200\n",
            "31/31 [==============================] - 2s 22ms/step - loss: 3.4763 - accuracy: 0.3582 - val_loss: 3.6675 - val_accuracy: 0.1296 - lr: 0.0010\n",
            "Epoch 2/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 3.0005 - accuracy: 0.4545 - val_loss: 3.7625 - val_accuracy: 0.1276 - lr: 0.0010\n",
            "Epoch 3/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.7242 - accuracy: 0.4853 - val_loss: 3.5526 - val_accuracy: 0.1770 - lr: 0.0010\n",
            "Epoch 4/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.5890 - accuracy: 0.4936 - val_loss: 3.4248 - val_accuracy: 0.2325 - lr: 0.0010\n",
            "Epoch 5/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.4015 - accuracy: 0.5306 - val_loss: 3.1723 - val_accuracy: 0.2449 - lr: 0.0010\n",
            "Epoch 6/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 2.2217 - accuracy: 0.5574 - val_loss: 2.7949 - val_accuracy: 0.3313 - lr: 0.0010\n",
            "Epoch 7/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.0720 - accuracy: 0.5733 - val_loss: 3.2125 - val_accuracy: 0.2469 - lr: 0.0010\n",
            "Epoch 8/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.9672 - accuracy: 0.5934 - val_loss: 2.7808 - val_accuracy: 0.2510 - lr: 0.0010\n",
            "Epoch 9/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.8466 - accuracy: 0.6068 - val_loss: 2.4079 - val_accuracy: 0.5412 - lr: 0.0010\n",
            "Epoch 10/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.7533 - accuracy: 0.6166 - val_loss: 2.7804 - val_accuracy: 0.2840 - lr: 0.0010\n",
            "Epoch 11/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.6493 - accuracy: 0.6320 - val_loss: 2.5363 - val_accuracy: 0.3827 - lr: 0.0010\n",
            "Epoch 12/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.5505 - accuracy: 0.6541 - val_loss: 2.4136 - val_accuracy: 0.3621 - lr: 0.0010\n",
            "Epoch 13/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.4911 - accuracy: 0.6603 - val_loss: 2.0820 - val_accuracy: 0.4753 - lr: 0.0010\n",
            "Epoch 14/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.4076 - accuracy: 0.6716 - val_loss: 2.0477 - val_accuracy: 0.4959 - lr: 0.0010\n",
            "Epoch 15/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.3537 - accuracy: 0.6747 - val_loss: 1.8500 - val_accuracy: 0.5309 - lr: 0.0010\n",
            "Epoch 16/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.2922 - accuracy: 0.6850 - val_loss: 1.8457 - val_accuracy: 0.5206 - lr: 0.0010\n",
            "Epoch 17/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.2425 - accuracy: 0.6891 - val_loss: 1.4683 - val_accuracy: 0.6235 - lr: 0.0010\n",
            "Epoch 18/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.1775 - accuracy: 0.7113 - val_loss: 1.5460 - val_accuracy: 0.6029 - lr: 0.0010\n",
            "Epoch 19/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.1551 - accuracy: 0.7041 - val_loss: 1.5075 - val_accuracy: 0.5741 - lr: 0.0010\n",
            "Epoch 20/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.1054 - accuracy: 0.7226 - val_loss: 1.6061 - val_accuracy: 0.6070 - lr: 0.0010\n",
            "Epoch 21/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.0684 - accuracy: 0.7288 - val_loss: 1.6229 - val_accuracy: 0.5576 - lr: 0.0010\n",
            "Epoch 22/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.0114 - accuracy: 0.7452 - val_loss: 1.5068 - val_accuracy: 0.6132 - lr: 0.0010\n",
            "Epoch 23/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.9457 - accuracy: 0.7735 - val_loss: 1.5875 - val_accuracy: 0.5905 - lr: 0.0010\n",
            "Epoch 24/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8950 - accuracy: 0.7694 - val_loss: 1.5248 - val_accuracy: 0.5905 - lr: 0.0010\n",
            "Epoch 25/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.8808 - accuracy: 0.7653 - val_loss: 1.4249 - val_accuracy: 0.6235 - lr: 0.0010\n",
            "Epoch 26/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.8757 - accuracy: 0.7653 - val_loss: 1.3954 - val_accuracy: 0.5905 - lr: 0.0010\n",
            "Epoch 27/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8547 - accuracy: 0.7751 - val_loss: 1.3738 - val_accuracy: 0.6132 - lr: 0.0010\n",
            "Epoch 28/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8380 - accuracy: 0.7864 - val_loss: 1.3583 - val_accuracy: 0.6605 - lr: 0.0010\n",
            "Epoch 29/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.7679 - accuracy: 0.8029 - val_loss: 1.4902 - val_accuracy: 0.6111 - lr: 0.0010\n",
            "Epoch 30/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8065 - accuracy: 0.7844 - val_loss: 1.5560 - val_accuracy: 0.6070 - lr: 0.0010\n",
            "Epoch 31/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6888 - accuracy: 0.8296 - val_loss: 1.2280 - val_accuracy: 0.6626 - lr: 0.0010\n",
            "Epoch 32/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6540 - accuracy: 0.8312 - val_loss: 1.2809 - val_accuracy: 0.6584 - lr: 0.0010\n",
            "Epoch 33/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6392 - accuracy: 0.8338 - val_loss: 1.3336 - val_accuracy: 0.6749 - lr: 0.0010\n",
            "Epoch 34/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6202 - accuracy: 0.8420 - val_loss: 1.4381 - val_accuracy: 0.6564 - lr: 0.0010\n",
            "Epoch 35/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.5926 - accuracy: 0.8466 - val_loss: 1.3603 - val_accuracy: 0.6337 - lr: 0.0010\n",
            "Epoch 36/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.5431 - accuracy: 0.8729 - val_loss: 1.5986 - val_accuracy: 0.6461 - lr: 0.0010\n",
            "Epoch 37/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.5371 - accuracy: 0.8718 - val_loss: 1.6411 - val_accuracy: 0.6420 - lr: 0.0010\n",
            "Epoch 38/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.5486 - accuracy: 0.8636 - val_loss: 1.4769 - val_accuracy: 0.6790 - lr: 0.0010\n",
            "Epoch 39/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.4807 - accuracy: 0.8785 - val_loss: 1.5842 - val_accuracy: 0.6399 - lr: 0.0010\n",
            "Epoch 40/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.5142 - accuracy: 0.8677 - val_loss: 1.5418 - val_accuracy: 0.6358 - lr: 0.0010\n",
            "Epoch 41/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.4768 - accuracy: 0.8868 - val_loss: 1.6414 - val_accuracy: 0.6358 - lr: 0.0010\n",
            "Epoch 42/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.4226 - accuracy: 0.9043 - val_loss: 1.3938 - val_accuracy: 0.6790 - lr: 0.0010\n",
            "Epoch 43/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.3688 - accuracy: 0.9213 - val_loss: 1.4151 - val_accuracy: 0.6893 - lr: 0.0010\n",
            "Epoch 44/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3821 - accuracy: 0.9197 - val_loss: 1.9817 - val_accuracy: 0.5309 - lr: 0.0010\n",
            "Epoch 45/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.4371 - accuracy: 0.8935 - val_loss: 1.4934 - val_accuracy: 0.6317 - lr: 0.0010\n",
            "Epoch 46/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.4316 - accuracy: 0.8986 - val_loss: 1.4666 - val_accuracy: 0.6543 - lr: 0.0010\n",
            "Epoch 47/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3727 - accuracy: 0.9171 - val_loss: 1.3337 - val_accuracy: 0.6708 - lr: 0.0010\n",
            "Epoch 48/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.3377 - accuracy: 0.9326 - val_loss: 1.5123 - val_accuracy: 0.6687 - lr: 0.0010\n",
            "Epoch 49/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3056 - accuracy: 0.9372 - val_loss: 1.5904 - val_accuracy: 0.6749 - lr: 0.0010\n",
            "Epoch 50/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.3041 - accuracy: 0.9341 - val_loss: 1.4948 - val_accuracy: 0.6749 - lr: 0.0010\n",
            "Epoch 51/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2881 - accuracy: 0.9460 - val_loss: 1.6628 - val_accuracy: 0.6975 - lr: 0.0010\n",
            "Epoch 52/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3178 - accuracy: 0.9362 - val_loss: 1.4149 - val_accuracy: 0.6749 - lr: 0.0010\n",
            "Epoch 53/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2795 - accuracy: 0.9444 - val_loss: 1.5790 - val_accuracy: 0.6708 - lr: 0.0010\n",
            "Epoch 54/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.2584 - accuracy: 0.9511 - val_loss: 1.5567 - val_accuracy: 0.6770 - lr: 0.0010\n",
            "Epoch 55/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3067 - accuracy: 0.9341 - val_loss: 1.5001 - val_accuracy: 0.6811 - lr: 0.0010\n",
            "Epoch 56/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.2657 - accuracy: 0.9490 - val_loss: 1.3871 - val_accuracy: 0.7037 - lr: 0.0010\n",
            "Epoch 57/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.2014 - accuracy: 0.9727 - val_loss: 1.3781 - val_accuracy: 0.6852 - lr: 0.0010\n",
            "Epoch 58/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.2031 - accuracy: 0.9671 - val_loss: 1.4045 - val_accuracy: 0.6996 - lr: 0.0010\n",
            "Epoch 59/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2709 - accuracy: 0.9382 - val_loss: 1.4921 - val_accuracy: 0.6358 - lr: 0.0010\n",
            "Epoch 60/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2437 - accuracy: 0.9557 - val_loss: 1.6666 - val_accuracy: 0.6543 - lr: 0.0010\n",
            "Epoch 61/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.2428 - accuracy: 0.9542 - val_loss: 1.7855 - val_accuracy: 0.6523 - lr: 0.0010\n",
            "Epoch 62/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2300 - accuracy: 0.9563 - val_loss: 1.6765 - val_accuracy: 0.6399 - lr: 0.0010\n",
            "Epoch 63/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2066 - accuracy: 0.9681 - val_loss: 1.8477 - val_accuracy: 0.6461 - lr: 0.0010\n",
            "Epoch 64/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1684 - accuracy: 0.9779 - val_loss: 1.8984 - val_accuracy: 0.6687 - lr: 0.0010\n",
            "Epoch 65/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1606 - accuracy: 0.9784 - val_loss: 1.7502 - val_accuracy: 0.6626 - lr: 0.0010\n",
            "Epoch 66/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.1731 - accuracy: 0.9743 - val_loss: 1.6514 - val_accuracy: 0.6276 - lr: 0.0010\n",
            "Epoch 67/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1904 - accuracy: 0.9696 - val_loss: 2.0156 - val_accuracy: 0.5700 - lr: 0.0010\n",
            "Epoch 68/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1935 - accuracy: 0.9650 - val_loss: 1.6010 - val_accuracy: 0.6523 - lr: 0.0010\n",
            "Epoch 69/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1387 - accuracy: 0.9856 - val_loss: 1.7506 - val_accuracy: 0.6626 - lr: 0.0010\n",
            "Epoch 70/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1314 - accuracy: 0.9856 - val_loss: 1.5848 - val_accuracy: 0.6749 - lr: 0.0010\n",
            "Epoch 71/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2164 - accuracy: 0.9619 - val_loss: 2.0906 - val_accuracy: 0.5535 - lr: 0.0010\n",
            "Epoch 72/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1875 - accuracy: 0.9614 - val_loss: 1.5050 - val_accuracy: 0.6667 - lr: 5.0000e-04\n",
            "Epoch 73/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1274 - accuracy: 0.9892 - val_loss: 1.4964 - val_accuracy: 0.7016 - lr: 5.0000e-04\n",
            "Epoch 74/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1174 - accuracy: 0.9913 - val_loss: 1.4460 - val_accuracy: 0.7037 - lr: 5.0000e-04\n",
            "Epoch 75/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1020 - accuracy: 0.9959 - val_loss: 1.5330 - val_accuracy: 0.7078 - lr: 5.0000e-04\n",
            "Epoch 76/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0979 - accuracy: 0.9979 - val_loss: 1.4821 - val_accuracy: 0.6955 - lr: 5.0000e-04\n",
            "Epoch 77/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0934 - accuracy: 0.9964 - val_loss: 1.5045 - val_accuracy: 0.7119 - lr: 5.0000e-04\n",
            "Epoch 78/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0923 - accuracy: 0.9969 - val_loss: 1.4703 - val_accuracy: 0.7078 - lr: 5.0000e-04\n",
            "Epoch 79/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0859 - accuracy: 0.9979 - val_loss: 1.4588 - val_accuracy: 0.7099 - lr: 5.0000e-04\n",
            "Epoch 80/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0809 - accuracy: 0.9985 - val_loss: 1.5507 - val_accuracy: 0.6955 - lr: 5.0000e-04\n",
            "Epoch 81/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0786 - accuracy: 0.9995 - val_loss: 1.5206 - val_accuracy: 0.7181 - lr: 5.0000e-04\n",
            "Epoch 82/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0768 - accuracy: 0.9990 - val_loss: 1.4583 - val_accuracy: 0.7037 - lr: 5.0000e-04\n",
            "Epoch 83/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0758 - accuracy: 0.9995 - val_loss: 1.4689 - val_accuracy: 0.7099 - lr: 5.0000e-04\n",
            "Epoch 84/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0712 - accuracy: 0.9995 - val_loss: 1.4904 - val_accuracy: 0.7160 - lr: 5.0000e-04\n",
            "Epoch 85/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0708 - accuracy: 0.9995 - val_loss: 1.4913 - val_accuracy: 0.7202 - lr: 5.0000e-04\n",
            "Epoch 86/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0712 - accuracy: 0.9995 - val_loss: 1.4585 - val_accuracy: 0.7037 - lr: 5.0000e-04\n",
            "Epoch 87/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0705 - accuracy: 0.9990 - val_loss: 1.4114 - val_accuracy: 0.7099 - lr: 5.0000e-04\n",
            "Epoch 88/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0671 - accuracy: 1.0000 - val_loss: 1.4227 - val_accuracy: 0.7099 - lr: 5.0000e-04\n",
            "Epoch 89/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0737 - accuracy: 0.9969 - val_loss: 1.4535 - val_accuracy: 0.7058 - lr: 5.0000e-04\n",
            "Epoch 90/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0666 - accuracy: 1.0000 - val_loss: 1.4570 - val_accuracy: 0.7181 - lr: 5.0000e-04\n",
            "Epoch 91/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0630 - accuracy: 0.9990 - val_loss: 1.4422 - val_accuracy: 0.7140 - lr: 5.0000e-04\n",
            "Epoch 92/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0639 - accuracy: 0.9985 - val_loss: 1.4551 - val_accuracy: 0.7243 - lr: 5.0000e-04\n",
            "Epoch 93/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0805 - accuracy: 0.9943 - val_loss: 1.4787 - val_accuracy: 0.7037 - lr: 5.0000e-04\n",
            "Epoch 94/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0673 - accuracy: 0.9990 - val_loss: 1.4303 - val_accuracy: 0.7058 - lr: 5.0000e-04\n",
            "Epoch 95/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0664 - accuracy: 0.9979 - val_loss: 1.4260 - val_accuracy: 0.7202 - lr: 5.0000e-04\n",
            "Epoch 96/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0657 - accuracy: 0.9985 - val_loss: 1.4844 - val_accuracy: 0.7140 - lr: 5.0000e-04\n",
            "Epoch 97/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0620 - accuracy: 1.0000 - val_loss: 1.4484 - val_accuracy: 0.7078 - lr: 5.0000e-04\n",
            "Epoch 98/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0615 - accuracy: 0.9990 - val_loss: 1.4528 - val_accuracy: 0.7058 - lr: 5.0000e-04\n",
            "Epoch 99/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0579 - accuracy: 0.9995 - val_loss: 1.4346 - val_accuracy: 0.7181 - lr: 5.0000e-04\n",
            "Epoch 100/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0657 - accuracy: 0.9969 - val_loss: 1.4671 - val_accuracy: 0.6934 - lr: 5.0000e-04\n",
            "Epoch 101/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0598 - accuracy: 0.9990 - val_loss: 1.5191 - val_accuracy: 0.6955 - lr: 5.0000e-04\n",
            "Epoch 102/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0559 - accuracy: 1.0000 - val_loss: 1.4795 - val_accuracy: 0.7160 - lr: 5.0000e-04\n",
            "Epoch 103/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0538 - accuracy: 0.9995 - val_loss: 1.4267 - val_accuracy: 0.7202 - lr: 5.0000e-04\n",
            "Epoch 104/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0538 - accuracy: 0.9995 - val_loss: 1.4210 - val_accuracy: 0.7160 - lr: 5.0000e-04\n",
            "Epoch 105/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0533 - accuracy: 1.0000 - val_loss: 1.3475 - val_accuracy: 0.7099 - lr: 5.0000e-04\n",
            "Epoch 106/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0507 - accuracy: 1.0000 - val_loss: 1.4104 - val_accuracy: 0.7058 - lr: 5.0000e-04\n",
            "Epoch 107/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0504 - accuracy: 1.0000 - val_loss: 1.3983 - val_accuracy: 0.7243 - lr: 5.0000e-04\n",
            "Epoch 108/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0496 - accuracy: 1.0000 - val_loss: 1.3783 - val_accuracy: 0.7160 - lr: 2.5000e-04\n",
            "Epoch 109/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0480 - accuracy: 1.0000 - val_loss: 1.4241 - val_accuracy: 0.7099 - lr: 2.5000e-04\n",
            "Epoch 110/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0495 - accuracy: 1.0000 - val_loss: 1.3903 - val_accuracy: 0.7181 - lr: 2.5000e-04\n",
            "Epoch 111/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0479 - accuracy: 1.0000 - val_loss: 1.4002 - val_accuracy: 0.7160 - lr: 2.5000e-04\n",
            "Epoch 112/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0473 - accuracy: 1.0000 - val_loss: 1.4054 - val_accuracy: 0.7140 - lr: 2.5000e-04\n",
            "Epoch 113/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0468 - accuracy: 1.0000 - val_loss: 1.3961 - val_accuracy: 0.7160 - lr: 2.5000e-04\n",
            "Epoch 114/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0472 - accuracy: 1.0000 - val_loss: 1.3874 - val_accuracy: 0.7140 - lr: 2.5000e-04\n",
            "Epoch 115/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0472 - accuracy: 0.9990 - val_loss: 1.4091 - val_accuracy: 0.7140 - lr: 2.5000e-04\n",
            "Epoch 116/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0491 - accuracy: 0.9990 - val_loss: 1.4583 - val_accuracy: 0.7140 - lr: 2.5000e-04\n",
            "Epoch 117/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0483 - accuracy: 0.9995 - val_loss: 1.3577 - val_accuracy: 0.7058 - lr: 2.5000e-04\n",
            "Epoch 118/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0472 - accuracy: 1.0000 - val_loss: 1.3810 - val_accuracy: 0.7160 - lr: 2.5000e-04\n",
            "Epoch 119/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0453 - accuracy: 1.0000 - val_loss: 1.3658 - val_accuracy: 0.7140 - lr: 2.5000e-04\n",
            "Epoch 120/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0461 - accuracy: 0.9995 - val_loss: 1.3687 - val_accuracy: 0.7263 - lr: 2.5000e-04\n",
            "Epoch 121/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0459 - accuracy: 1.0000 - val_loss: 1.4037 - val_accuracy: 0.7160 - lr: 2.5000e-04\n",
            "Epoch 122/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0464 - accuracy: 0.9995 - val_loss: 1.3416 - val_accuracy: 0.7202 - lr: 2.5000e-04\n",
            "Epoch 123/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0445 - accuracy: 1.0000 - val_loss: 1.3684 - val_accuracy: 0.7181 - lr: 2.5000e-04\n",
            "Epoch 124/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0437 - accuracy: 1.0000 - val_loss: 1.3868 - val_accuracy: 0.7160 - lr: 2.5000e-04\n",
            "Epoch 125/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0434 - accuracy: 1.0000 - val_loss: 1.3991 - val_accuracy: 0.7181 - lr: 2.5000e-04\n",
            "Epoch 126/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0430 - accuracy: 0.9995 - val_loss: 1.3835 - val_accuracy: 0.7140 - lr: 2.5000e-04\n",
            "Epoch 127/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0436 - accuracy: 1.0000 - val_loss: 1.3828 - val_accuracy: 0.7222 - lr: 2.5000e-04\n",
            "Epoch 128/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0424 - accuracy: 1.0000 - val_loss: 1.4206 - val_accuracy: 0.7160 - lr: 2.5000e-04\n",
            "Epoch 129/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0423 - accuracy: 1.0000 - val_loss: 1.3673 - val_accuracy: 0.7099 - lr: 2.5000e-04\n",
            "Epoch 130/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0429 - accuracy: 0.9995 - val_loss: 1.3882 - val_accuracy: 0.7119 - lr: 2.5000e-04\n",
            "Epoch 131/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0419 - accuracy: 1.0000 - val_loss: 1.3847 - val_accuracy: 0.7263 - lr: 2.5000e-04\n",
            "Epoch 132/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0414 - accuracy: 1.0000 - val_loss: 1.4041 - val_accuracy: 0.7181 - lr: 2.5000e-04\n",
            "Epoch 133/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0415 - accuracy: 1.0000 - val_loss: 1.3714 - val_accuracy: 0.7222 - lr: 2.5000e-04\n",
            "Epoch 134/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0407 - accuracy: 1.0000 - val_loss: 1.4145 - val_accuracy: 0.7099 - lr: 2.5000e-04\n",
            "Epoch 135/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0403 - accuracy: 1.0000 - val_loss: 1.3661 - val_accuracy: 0.7140 - lr: 2.5000e-04\n",
            "Epoch 136/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0411 - accuracy: 1.0000 - val_loss: 1.4003 - val_accuracy: 0.7140 - lr: 1.2500e-04\n",
            "Epoch 137/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0399 - accuracy: 1.0000 - val_loss: 1.4480 - val_accuracy: 0.7222 - lr: 1.2500e-04\n",
            "Epoch 138/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0400 - accuracy: 1.0000 - val_loss: 1.3914 - val_accuracy: 0.7181 - lr: 1.2500e-04\n",
            "Epoch 139/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0413 - accuracy: 0.9990 - val_loss: 1.3542 - val_accuracy: 0.7263 - lr: 1.2500e-04\n",
            "Epoch 140/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0405 - accuracy: 1.0000 - val_loss: 1.4098 - val_accuracy: 0.7078 - lr: 1.2500e-04\n",
            "Epoch 141/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0419 - accuracy: 0.9995 - val_loss: 1.3490 - val_accuracy: 0.7243 - lr: 1.2500e-04\n",
            "Epoch 142/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0429 - accuracy: 0.9985 - val_loss: 1.3933 - val_accuracy: 0.7119 - lr: 1.2500e-04\n",
            "Epoch 143/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0415 - accuracy: 0.9995 - val_loss: 1.4050 - val_accuracy: 0.7099 - lr: 1.2500e-04\n",
            "Epoch 144/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0391 - accuracy: 1.0000 - val_loss: 1.4255 - val_accuracy: 0.7181 - lr: 1.2500e-04\n",
            "Epoch 145/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0386 - accuracy: 1.0000 - val_loss: 1.4010 - val_accuracy: 0.7140 - lr: 1.2500e-04\n",
            "Epoch 146/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0391 - accuracy: 1.0000 - val_loss: 1.3905 - val_accuracy: 0.7037 - lr: 1.2500e-04\n",
            "Epoch 147/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0382 - accuracy: 1.0000 - val_loss: 1.3768 - val_accuracy: 0.7181 - lr: 1.2500e-04\n",
            "Epoch 148/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0388 - accuracy: 1.0000 - val_loss: 1.3696 - val_accuracy: 0.7181 - lr: 1.2500e-04\n",
            "Epoch 149/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0385 - accuracy: 1.0000 - val_loss: 1.3643 - val_accuracy: 0.7160 - lr: 1.2500e-04\n",
            "Epoch 150/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0378 - accuracy: 1.0000 - val_loss: 1.3786 - val_accuracy: 0.7222 - lr: 1.2500e-04\n"
          ]
        }
      ],
      "source": [
        "history4 = model4.fit(\n",
        "    x = x_train4,\n",
        "    y = y_train4,\n",
        "    batch_size = 64,\n",
        "    epochs = 200,\n",
        "    validation_data = (x_test4, y_test4),\n",
        "    callbacks = [\n",
        "        tfk.callbacks.EarlyStopping(monitor='val_accuracy', mode='max', patience=30, restore_best_weights=True),\n",
        "        tfk.callbacks.ReduceLROnPlateau(monitor='val_accuracy', mode='max', patience=15, factor=0.5, min_lr=1e-5)\n",
        "    ]\n",
        ").history"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 20,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "pdNG7rdVwOnw",
        "outputId": "268c66c8-957f-40b4-eec7-6ab138d633d6"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1/200\n",
            "31/31 [==============================] - 2s 22ms/step - loss: 3.4909 - accuracy: 0.3603 - val_loss: 3.7128 - val_accuracy: 0.1193 - lr: 0.0010\n",
            "Epoch 2/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 3.0439 - accuracy: 0.4406 - val_loss: 3.7038 - val_accuracy: 0.1399 - lr: 0.0010\n",
            "Epoch 3/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.7790 - accuracy: 0.4627 - val_loss: 3.5529 - val_accuracy: 0.1152 - lr: 0.0010\n",
            "Epoch 4/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 2.5156 - accuracy: 0.5322 - val_loss: 3.0460 - val_accuracy: 0.2243 - lr: 0.0010\n",
            "Epoch 5/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.3854 - accuracy: 0.5455 - val_loss: 2.9969 - val_accuracy: 0.2387 - lr: 0.0010\n",
            "Epoch 6/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.2327 - accuracy: 0.5481 - val_loss: 2.6470 - val_accuracy: 0.4918 - lr: 0.0010\n",
            "Epoch 7/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 2.0894 - accuracy: 0.5965 - val_loss: 2.5817 - val_accuracy: 0.3539 - lr: 0.0010\n",
            "Epoch 8/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.9619 - accuracy: 0.5914 - val_loss: 2.4586 - val_accuracy: 0.3786 - lr: 0.0010\n",
            "Epoch 9/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.8580 - accuracy: 0.6016 - val_loss: 2.5227 - val_accuracy: 0.3498 - lr: 0.0010\n",
            "Epoch 10/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.7335 - accuracy: 0.6233 - val_loss: 2.4858 - val_accuracy: 0.3704 - lr: 0.0010\n",
            "Epoch 11/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.6559 - accuracy: 0.6372 - val_loss: 2.0388 - val_accuracy: 0.5267 - lr: 0.0010\n",
            "Epoch 12/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.5738 - accuracy: 0.6500 - val_loss: 1.8711 - val_accuracy: 0.5267 - lr: 0.0010\n",
            "Epoch 13/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.4630 - accuracy: 0.6655 - val_loss: 1.7826 - val_accuracy: 0.5823 - lr: 0.0010\n",
            "Epoch 14/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.4052 - accuracy: 0.6794 - val_loss: 1.7946 - val_accuracy: 0.5700 - lr: 0.0010\n",
            "Epoch 15/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.3687 - accuracy: 0.6768 - val_loss: 1.7732 - val_accuracy: 0.5473 - lr: 0.0010\n",
            "Epoch 16/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.2888 - accuracy: 0.6917 - val_loss: 1.6287 - val_accuracy: 0.5864 - lr: 0.0010\n",
            "Epoch 17/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.3259 - accuracy: 0.6752 - val_loss: 1.6749 - val_accuracy: 0.5535 - lr: 0.0010\n",
            "Epoch 18/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.2362 - accuracy: 0.6958 - val_loss: 1.5984 - val_accuracy: 0.5823 - lr: 0.0010\n",
            "Epoch 19/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.1446 - accuracy: 0.7226 - val_loss: 1.4841 - val_accuracy: 0.6008 - lr: 0.0010\n",
            "Epoch 20/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.0991 - accuracy: 0.7216 - val_loss: 1.5450 - val_accuracy: 0.5741 - lr: 0.0010\n",
            "Epoch 21/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.0682 - accuracy: 0.7267 - val_loss: 1.4014 - val_accuracy: 0.6173 - lr: 0.0010\n",
            "Epoch 22/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.9939 - accuracy: 0.7519 - val_loss: 1.3109 - val_accuracy: 0.6399 - lr: 0.0010\n",
            "Epoch 23/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.9559 - accuracy: 0.7540 - val_loss: 1.4220 - val_accuracy: 0.6255 - lr: 0.0010\n",
            "Epoch 24/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.9590 - accuracy: 0.7519 - val_loss: 1.2981 - val_accuracy: 0.6461 - lr: 0.0010\n",
            "Epoch 25/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.9237 - accuracy: 0.7540 - val_loss: 1.4673 - val_accuracy: 0.5926 - lr: 0.0010\n",
            "Epoch 26/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.9090 - accuracy: 0.7560 - val_loss: 1.3448 - val_accuracy: 0.6255 - lr: 0.0010\n",
            "Epoch 27/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8488 - accuracy: 0.7684 - val_loss: 1.3529 - val_accuracy: 0.6132 - lr: 0.0010\n",
            "Epoch 28/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8326 - accuracy: 0.7756 - val_loss: 1.2632 - val_accuracy: 0.6564 - lr: 0.0010\n",
            "Epoch 29/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.7610 - accuracy: 0.8044 - val_loss: 1.2610 - val_accuracy: 0.6296 - lr: 0.0010\n",
            "Epoch 30/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.7348 - accuracy: 0.8106 - val_loss: 1.2823 - val_accuracy: 0.6337 - lr: 0.0010\n",
            "Epoch 31/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.7289 - accuracy: 0.8194 - val_loss: 1.3187 - val_accuracy: 0.6152 - lr: 0.0010\n",
            "Epoch 32/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6987 - accuracy: 0.8168 - val_loss: 1.3586 - val_accuracy: 0.6337 - lr: 0.0010\n",
            "Epoch 33/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6568 - accuracy: 0.8245 - val_loss: 1.3536 - val_accuracy: 0.6584 - lr: 0.0010\n",
            "Epoch 34/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6262 - accuracy: 0.8338 - val_loss: 1.2764 - val_accuracy: 0.6523 - lr: 0.0010\n",
            "Epoch 35/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.5915 - accuracy: 0.8513 - val_loss: 1.4313 - val_accuracy: 0.6276 - lr: 0.0010\n",
            "Epoch 36/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6327 - accuracy: 0.8332 - val_loss: 1.4206 - val_accuracy: 0.6029 - lr: 0.0010\n",
            "Epoch 37/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.5426 - accuracy: 0.8595 - val_loss: 1.2893 - val_accuracy: 0.6317 - lr: 0.0010\n",
            "Epoch 38/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.5237 - accuracy: 0.8636 - val_loss: 1.3643 - val_accuracy: 0.6029 - lr: 0.0010\n",
            "Epoch 39/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.5633 - accuracy: 0.8482 - val_loss: 1.3360 - val_accuracy: 0.6481 - lr: 0.0010\n",
            "Epoch 40/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.5264 - accuracy: 0.8677 - val_loss: 1.2883 - val_accuracy: 0.6358 - lr: 0.0010\n",
            "Epoch 41/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.5251 - accuracy: 0.8682 - val_loss: 1.2965 - val_accuracy: 0.6111 - lr: 0.0010\n",
            "Epoch 42/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.4591 - accuracy: 0.8966 - val_loss: 1.2691 - val_accuracy: 0.6770 - lr: 0.0010\n",
            "Epoch 43/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.4186 - accuracy: 0.9084 - val_loss: 1.3199 - val_accuracy: 0.6564 - lr: 0.0010\n",
            "Epoch 44/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.3855 - accuracy: 0.9151 - val_loss: 1.4531 - val_accuracy: 0.6564 - lr: 0.0010\n",
            "Epoch 45/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3468 - accuracy: 0.9274 - val_loss: 1.5726 - val_accuracy: 0.6276 - lr: 0.0010\n",
            "Epoch 46/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3930 - accuracy: 0.9099 - val_loss: 1.5498 - val_accuracy: 0.6235 - lr: 0.0010\n",
            "Epoch 47/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.3835 - accuracy: 0.9146 - val_loss: 1.6201 - val_accuracy: 0.5885 - lr: 0.0010\n",
            "Epoch 48/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.3695 - accuracy: 0.9238 - val_loss: 1.3194 - val_accuracy: 0.6235 - lr: 0.0010\n",
            "Epoch 49/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.3427 - accuracy: 0.9269 - val_loss: 1.3242 - val_accuracy: 0.6399 - lr: 0.0010\n",
            "Epoch 50/200\n",
            "31/31 [==============================] - 0s 15ms/step - loss: 0.3532 - accuracy: 0.9264 - val_loss: 1.3800 - val_accuracy: 0.6358 - lr: 0.0010\n",
            "Epoch 51/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.3321 - accuracy: 0.9274 - val_loss: 1.4398 - val_accuracy: 0.6831 - lr: 0.0010\n",
            "Epoch 52/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.2920 - accuracy: 0.9418 - val_loss: 1.3284 - val_accuracy: 0.6420 - lr: 0.0010\n",
            "Epoch 53/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2989 - accuracy: 0.9408 - val_loss: 1.4547 - val_accuracy: 0.6440 - lr: 0.0010\n",
            "Epoch 54/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.2818 - accuracy: 0.9398 - val_loss: 1.3584 - val_accuracy: 0.6646 - lr: 0.0010\n",
            "Epoch 55/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2431 - accuracy: 0.9578 - val_loss: 1.5502 - val_accuracy: 0.6276 - lr: 0.0010\n",
            "Epoch 56/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.3046 - accuracy: 0.9372 - val_loss: 1.4297 - val_accuracy: 0.6584 - lr: 0.0010\n",
            "Epoch 57/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.2569 - accuracy: 0.9470 - val_loss: 1.4843 - val_accuracy: 0.6543 - lr: 0.0010\n",
            "Epoch 58/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2233 - accuracy: 0.9645 - val_loss: 1.6652 - val_accuracy: 0.6687 - lr: 0.0010\n",
            "Epoch 59/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2068 - accuracy: 0.9722 - val_loss: 1.6754 - val_accuracy: 0.6543 - lr: 0.0010\n",
            "Epoch 60/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.2124 - accuracy: 0.9614 - val_loss: 1.5628 - val_accuracy: 0.6564 - lr: 0.0010\n",
            "Epoch 61/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1922 - accuracy: 0.9701 - val_loss: 1.8060 - val_accuracy: 0.6461 - lr: 0.0010\n",
            "Epoch 62/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.1991 - accuracy: 0.9686 - val_loss: 1.8417 - val_accuracy: 0.6235 - lr: 0.0010\n",
            "Epoch 63/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2018 - accuracy: 0.9645 - val_loss: 1.4724 - val_accuracy: 0.6317 - lr: 0.0010\n",
            "Epoch 64/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1988 - accuracy: 0.9696 - val_loss: 1.6410 - val_accuracy: 0.6584 - lr: 0.0010\n",
            "Epoch 65/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1687 - accuracy: 0.9732 - val_loss: 2.0564 - val_accuracy: 0.6564 - lr: 0.0010\n",
            "Epoch 66/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1736 - accuracy: 0.9738 - val_loss: 1.6383 - val_accuracy: 0.6502 - lr: 0.0010\n",
            "Epoch 67/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1460 - accuracy: 0.9846 - val_loss: 1.6223 - val_accuracy: 0.6626 - lr: 5.0000e-04\n",
            "Epoch 68/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1363 - accuracy: 0.9856 - val_loss: 1.5593 - val_accuracy: 0.6728 - lr: 5.0000e-04\n",
            "Epoch 69/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1278 - accuracy: 0.9871 - val_loss: 1.7290 - val_accuracy: 0.6399 - lr: 5.0000e-04\n",
            "Epoch 70/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1063 - accuracy: 0.9979 - val_loss: 1.7832 - val_accuracy: 0.6481 - lr: 5.0000e-04\n",
            "Epoch 71/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1043 - accuracy: 0.9943 - val_loss: 1.6008 - val_accuracy: 0.6646 - lr: 5.0000e-04\n",
            "Epoch 72/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0986 - accuracy: 0.9959 - val_loss: 1.5597 - val_accuracy: 0.6523 - lr: 5.0000e-04\n",
            "Epoch 73/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0936 - accuracy: 0.9985 - val_loss: 1.5882 - val_accuracy: 0.6770 - lr: 5.0000e-04\n",
            "Epoch 74/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0911 - accuracy: 0.9985 - val_loss: 1.6420 - val_accuracy: 0.6811 - lr: 5.0000e-04\n",
            "Epoch 75/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0915 - accuracy: 0.9964 - val_loss: 1.6697 - val_accuracy: 0.6708 - lr: 5.0000e-04\n",
            "Epoch 76/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0912 - accuracy: 0.9954 - val_loss: 1.5856 - val_accuracy: 0.6626 - lr: 5.0000e-04\n",
            "Epoch 77/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0861 - accuracy: 0.9979 - val_loss: 1.5259 - val_accuracy: 0.6605 - lr: 5.0000e-04\n",
            "Epoch 78/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0805 - accuracy: 0.9990 - val_loss: 1.6051 - val_accuracy: 0.6605 - lr: 5.0000e-04\n",
            "Epoch 79/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0790 - accuracy: 0.9985 - val_loss: 1.5248 - val_accuracy: 0.6605 - lr: 5.0000e-04\n",
            "Epoch 80/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0765 - accuracy: 1.0000 - val_loss: 1.6106 - val_accuracy: 0.6687 - lr: 5.0000e-04\n",
            "Epoch 81/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0805 - accuracy: 0.9964 - val_loss: 1.5941 - val_accuracy: 0.6831 - lr: 5.0000e-04\n"
          ]
        }
      ],
      "source": [
        "history5 = model5.fit(\n",
        "    x = x_train5,\n",
        "    y = y_train5,\n",
        "    batch_size = 64,\n",
        "    epochs = 200,\n",
        "    validation_data = (x_test5, y_test5),\n",
        "    callbacks = [\n",
        "        tfk.callbacks.EarlyStopping(monitor='val_accuracy', mode='max', patience=30, restore_best_weights=True),\n",
        "        tfk.callbacks.ReduceLROnPlateau(monitor='val_accuracy', mode='max', patience=15, factor=0.5, min_lr=1e-5)\n",
        "    ]\n",
        ").history"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 21,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "3dl3HD-_wOlJ",
        "outputId": "8af86184-e8b7-4229-efb2-f5af98c1a3b9"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1/200\n",
            "31/31 [==============================] - 2s 33ms/step - loss: 3.4979 - accuracy: 0.3556 - val_loss: 3.6796 - val_accuracy: 0.1214 - lr: 0.0010\n",
            "Epoch 2/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.9786 - accuracy: 0.4534 - val_loss: 3.5994 - val_accuracy: 0.1399 - lr: 0.0010\n",
            "Epoch 3/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.7162 - accuracy: 0.4843 - val_loss: 3.3028 - val_accuracy: 0.1852 - lr: 0.0010\n",
            "Epoch 4/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.4879 - accuracy: 0.5214 - val_loss: 3.2217 - val_accuracy: 0.1687 - lr: 0.0010\n",
            "Epoch 5/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.3362 - accuracy: 0.5404 - val_loss: 2.7636 - val_accuracy: 0.4280 - lr: 0.0010\n",
            "Epoch 6/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 2.2275 - accuracy: 0.5481 - val_loss: 2.8377 - val_accuracy: 0.2284 - lr: 0.0010\n",
            "Epoch 7/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.0149 - accuracy: 0.5872 - val_loss: 2.4699 - val_accuracy: 0.4115 - lr: 0.0010\n",
            "Epoch 8/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.9062 - accuracy: 0.6037 - val_loss: 2.4027 - val_accuracy: 0.4300 - lr: 0.0010\n",
            "Epoch 9/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.8419 - accuracy: 0.6058 - val_loss: 2.3988 - val_accuracy: 0.4115 - lr: 0.0010\n",
            "Epoch 10/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.7580 - accuracy: 0.6089 - val_loss: 2.2907 - val_accuracy: 0.4177 - lr: 0.0010\n",
            "Epoch 11/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.6179 - accuracy: 0.6346 - val_loss: 2.1262 - val_accuracy: 0.4506 - lr: 0.0010\n",
            "Epoch 12/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.5261 - accuracy: 0.6598 - val_loss: 2.0898 - val_accuracy: 0.5226 - lr: 0.0010\n",
            "Epoch 13/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.5018 - accuracy: 0.6475 - val_loss: 1.7028 - val_accuracy: 0.5864 - lr: 0.0010\n",
            "Epoch 14/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.4278 - accuracy: 0.6562 - val_loss: 1.7220 - val_accuracy: 0.5802 - lr: 0.0010\n",
            "Epoch 15/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.3581 - accuracy: 0.6655 - val_loss: 1.5190 - val_accuracy: 0.6070 - lr: 0.0010\n",
            "Epoch 16/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.2730 - accuracy: 0.6912 - val_loss: 1.7846 - val_accuracy: 0.5123 - lr: 0.0010\n",
            "Epoch 17/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.3240 - accuracy: 0.6557 - val_loss: 1.6696 - val_accuracy: 0.5535 - lr: 0.0010\n",
            "Epoch 18/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.2073 - accuracy: 0.6783 - val_loss: 1.4232 - val_accuracy: 0.6276 - lr: 0.0010\n",
            "Epoch 19/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.1545 - accuracy: 0.6943 - val_loss: 1.4469 - val_accuracy: 0.5823 - lr: 0.0010\n",
            "Epoch 20/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.1154 - accuracy: 0.7030 - val_loss: 1.3332 - val_accuracy: 0.6296 - lr: 0.0010\n",
            "Epoch 21/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.0702 - accuracy: 0.7077 - val_loss: 1.4694 - val_accuracy: 0.6317 - lr: 0.0010\n",
            "Epoch 22/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.0448 - accuracy: 0.7113 - val_loss: 1.3914 - val_accuracy: 0.6317 - lr: 0.0010\n",
            "Epoch 23/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.9816 - accuracy: 0.7349 - val_loss: 1.4145 - val_accuracy: 0.6049 - lr: 0.0010\n",
            "Epoch 24/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.9387 - accuracy: 0.7488 - val_loss: 1.4917 - val_accuracy: 0.5823 - lr: 0.0010\n",
            "Epoch 25/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.9273 - accuracy: 0.7422 - val_loss: 1.4315 - val_accuracy: 0.6399 - lr: 0.0010\n",
            "Epoch 26/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8708 - accuracy: 0.7514 - val_loss: 1.3347 - val_accuracy: 0.6440 - lr: 0.0010\n",
            "Epoch 27/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.8712 - accuracy: 0.7550 - val_loss: 1.3525 - val_accuracy: 0.6214 - lr: 0.0010\n",
            "Epoch 28/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8019 - accuracy: 0.7864 - val_loss: 1.3020 - val_accuracy: 0.6564 - lr: 0.0010\n",
            "Epoch 29/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.7720 - accuracy: 0.7854 - val_loss: 1.2608 - val_accuracy: 0.6523 - lr: 0.0010\n",
            "Epoch 30/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.7895 - accuracy: 0.7813 - val_loss: 1.5970 - val_accuracy: 0.5823 - lr: 0.0010\n",
            "Epoch 31/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.7373 - accuracy: 0.8029 - val_loss: 1.2836 - val_accuracy: 0.6708 - lr: 0.0010\n",
            "Epoch 32/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.7315 - accuracy: 0.7993 - val_loss: 1.2824 - val_accuracy: 0.6605 - lr: 0.0010\n",
            "Epoch 33/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6698 - accuracy: 0.8224 - val_loss: 1.2047 - val_accuracy: 0.6523 - lr: 0.0010\n",
            "Epoch 34/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6419 - accuracy: 0.8353 - val_loss: 1.6995 - val_accuracy: 0.6214 - lr: 0.0010\n",
            "Epoch 35/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.6589 - accuracy: 0.8245 - val_loss: 1.1729 - val_accuracy: 0.6955 - lr: 0.0010\n",
            "Epoch 36/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6021 - accuracy: 0.8430 - val_loss: 1.4400 - val_accuracy: 0.6502 - lr: 0.0010\n",
            "Epoch 37/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.5547 - accuracy: 0.8600 - val_loss: 1.2622 - val_accuracy: 0.6770 - lr: 0.0010\n",
            "Epoch 38/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.4964 - accuracy: 0.8785 - val_loss: 1.3884 - val_accuracy: 0.6687 - lr: 0.0010\n",
            "Epoch 39/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.4908 - accuracy: 0.8832 - val_loss: 1.4792 - val_accuracy: 0.6440 - lr: 0.0010\n",
            "Epoch 40/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.4801 - accuracy: 0.8837 - val_loss: 1.3549 - val_accuracy: 0.6502 - lr: 0.0010\n",
            "Epoch 41/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.4521 - accuracy: 0.8899 - val_loss: 1.2724 - val_accuracy: 0.6770 - lr: 0.0010\n",
            "Epoch 42/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.4128 - accuracy: 0.9027 - val_loss: 1.2302 - val_accuracy: 0.6893 - lr: 0.0010\n",
            "Epoch 43/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.4589 - accuracy: 0.8883 - val_loss: 1.3985 - val_accuracy: 0.6728 - lr: 0.0010\n",
            "Epoch 44/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.4433 - accuracy: 0.8924 - val_loss: 1.3651 - val_accuracy: 0.6502 - lr: 0.0010\n",
            "Epoch 45/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.3641 - accuracy: 0.9233 - val_loss: 1.3224 - val_accuracy: 0.6811 - lr: 0.0010\n",
            "Epoch 46/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3482 - accuracy: 0.9290 - val_loss: 1.3735 - val_accuracy: 0.6790 - lr: 0.0010\n",
            "Epoch 47/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3664 - accuracy: 0.9197 - val_loss: 1.9342 - val_accuracy: 0.6132 - lr: 0.0010\n",
            "Epoch 48/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.3263 - accuracy: 0.9285 - val_loss: 1.6672 - val_accuracy: 0.6605 - lr: 0.0010\n",
            "Epoch 49/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.3273 - accuracy: 0.9279 - val_loss: 1.2400 - val_accuracy: 0.6728 - lr: 0.0010\n",
            "Epoch 50/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.2777 - accuracy: 0.9552 - val_loss: 1.2977 - val_accuracy: 0.6543 - lr: 0.0010\n",
            "Epoch 51/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2406 - accuracy: 0.9604 - val_loss: 1.2625 - val_accuracy: 0.6996 - lr: 5.0000e-04\n",
            "Epoch 52/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.2256 - accuracy: 0.9624 - val_loss: 1.4201 - val_accuracy: 0.6790 - lr: 5.0000e-04\n",
            "Epoch 53/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1961 - accuracy: 0.9753 - val_loss: 1.4248 - val_accuracy: 0.6872 - lr: 5.0000e-04\n",
            "Epoch 54/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1941 - accuracy: 0.9722 - val_loss: 1.4516 - val_accuracy: 0.6955 - lr: 5.0000e-04\n",
            "Epoch 55/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.1722 - accuracy: 0.9794 - val_loss: 1.3773 - val_accuracy: 0.6996 - lr: 5.0000e-04\n",
            "Epoch 56/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1678 - accuracy: 0.9810 - val_loss: 1.5654 - val_accuracy: 0.7016 - lr: 5.0000e-04\n",
            "Epoch 57/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1705 - accuracy: 0.9799 - val_loss: 1.3817 - val_accuracy: 0.7058 - lr: 5.0000e-04\n",
            "Epoch 58/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.1771 - accuracy: 0.9748 - val_loss: 1.6362 - val_accuracy: 0.6564 - lr: 5.0000e-04\n",
            "Epoch 59/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1696 - accuracy: 0.9768 - val_loss: 1.3557 - val_accuracy: 0.6996 - lr: 5.0000e-04\n",
            "Epoch 60/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.1485 - accuracy: 0.9856 - val_loss: 1.2948 - val_accuracy: 0.6996 - lr: 5.0000e-04\n",
            "Epoch 61/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1381 - accuracy: 0.9861 - val_loss: 1.4662 - val_accuracy: 0.6893 - lr: 5.0000e-04\n",
            "Epoch 62/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1400 - accuracy: 0.9902 - val_loss: 1.5422 - val_accuracy: 0.6996 - lr: 5.0000e-04\n",
            "Epoch 63/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1310 - accuracy: 0.9892 - val_loss: 1.4352 - val_accuracy: 0.7160 - lr: 5.0000e-04\n",
            "Epoch 64/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1339 - accuracy: 0.9887 - val_loss: 1.4514 - val_accuracy: 0.7016 - lr: 5.0000e-04\n",
            "Epoch 65/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1306 - accuracy: 0.9856 - val_loss: 1.4368 - val_accuracy: 0.6975 - lr: 5.0000e-04\n",
            "Epoch 66/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1248 - accuracy: 0.9913 - val_loss: 1.5129 - val_accuracy: 0.6996 - lr: 5.0000e-04\n",
            "Epoch 67/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1188 - accuracy: 0.9913 - val_loss: 1.4448 - val_accuracy: 0.7016 - lr: 5.0000e-04\n",
            "Epoch 68/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1152 - accuracy: 0.9907 - val_loss: 1.5718 - val_accuracy: 0.6934 - lr: 5.0000e-04\n",
            "Epoch 69/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1227 - accuracy: 0.9897 - val_loss: 1.6906 - val_accuracy: 0.6728 - lr: 5.0000e-04\n",
            "Epoch 70/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.1177 - accuracy: 0.9913 - val_loss: 1.8796 - val_accuracy: 0.6770 - lr: 5.0000e-04\n",
            "Epoch 71/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1038 - accuracy: 0.9959 - val_loss: 1.5576 - val_accuracy: 0.6893 - lr: 5.0000e-04\n",
            "Epoch 72/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1050 - accuracy: 0.9938 - val_loss: 1.5028 - val_accuracy: 0.6914 - lr: 5.0000e-04\n",
            "Epoch 73/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1334 - accuracy: 0.9846 - val_loss: 1.6302 - val_accuracy: 0.6502 - lr: 5.0000e-04\n",
            "Epoch 74/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1133 - accuracy: 0.9913 - val_loss: 1.3215 - val_accuracy: 0.6872 - lr: 5.0000e-04\n",
            "Epoch 75/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0985 - accuracy: 0.9959 - val_loss: 1.3849 - val_accuracy: 0.7058 - lr: 5.0000e-04\n",
            "Epoch 76/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1013 - accuracy: 0.9928 - val_loss: 1.4389 - val_accuracy: 0.6934 - lr: 5.0000e-04\n",
            "Epoch 77/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0913 - accuracy: 0.9964 - val_loss: 1.4604 - val_accuracy: 0.6872 - lr: 5.0000e-04\n",
            "Epoch 78/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1103 - accuracy: 0.9887 - val_loss: 1.4194 - val_accuracy: 0.6811 - lr: 5.0000e-04\n",
            "Epoch 79/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0927 - accuracy: 0.9964 - val_loss: 1.4545 - val_accuracy: 0.6934 - lr: 2.5000e-04\n",
            "Epoch 80/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0835 - accuracy: 0.9990 - val_loss: 1.5646 - val_accuracy: 0.6996 - lr: 2.5000e-04\n",
            "Epoch 81/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0795 - accuracy: 0.9995 - val_loss: 1.5556 - val_accuracy: 0.6893 - lr: 2.5000e-04\n",
            "Epoch 82/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0807 - accuracy: 0.9990 - val_loss: 1.4356 - val_accuracy: 0.6996 - lr: 2.5000e-04\n",
            "Epoch 83/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0789 - accuracy: 0.9990 - val_loss: 1.4199 - val_accuracy: 0.7058 - lr: 2.5000e-04\n",
            "Epoch 84/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0752 - accuracy: 0.9995 - val_loss: 1.4080 - val_accuracy: 0.7037 - lr: 2.5000e-04\n",
            "Epoch 85/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0740 - accuracy: 0.9995 - val_loss: 1.4206 - val_accuracy: 0.7037 - lr: 2.5000e-04\n",
            "Epoch 86/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0748 - accuracy: 0.9990 - val_loss: 1.4107 - val_accuracy: 0.6955 - lr: 2.5000e-04\n",
            "Epoch 87/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0752 - accuracy: 0.9979 - val_loss: 1.3631 - val_accuracy: 0.7078 - lr: 2.5000e-04\n",
            "Epoch 88/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0728 - accuracy: 1.0000 - val_loss: 1.4467 - val_accuracy: 0.6955 - lr: 2.5000e-04\n",
            "Epoch 89/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0702 - accuracy: 0.9995 - val_loss: 1.4094 - val_accuracy: 0.7099 - lr: 2.5000e-04\n",
            "Epoch 90/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0710 - accuracy: 0.9990 - val_loss: 1.3962 - val_accuracy: 0.7119 - lr: 2.5000e-04\n",
            "Epoch 91/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0694 - accuracy: 1.0000 - val_loss: 1.3597 - val_accuracy: 0.7037 - lr: 2.5000e-04\n",
            "Epoch 92/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0678 - accuracy: 0.9995 - val_loss: 1.3501 - val_accuracy: 0.6934 - lr: 2.5000e-04\n",
            "Epoch 93/200\n",
            "31/31 [==============================] - 0s 16ms/step - loss: 0.0667 - accuracy: 0.9995 - val_loss: 1.3916 - val_accuracy: 0.7099 - lr: 2.5000e-04\n"
          ]
        }
      ],
      "source": [
        "history6 = model6.fit(\n",
        "    x = x_train6,\n",
        "    y = y_train6,\n",
        "    batch_size = 64,\n",
        "    epochs = 200,\n",
        "    validation_data = (x_test6, y_test6),\n",
        "    callbacks = [\n",
        "        tfk.callbacks.EarlyStopping(monitor='val_accuracy', mode='max', patience=30, restore_best_weights=True),\n",
        "        tfk.callbacks.ReduceLROnPlateau(monitor='val_accuracy', mode='max', patience=15, factor=0.5, min_lr=1e-5)\n",
        "    ]\n",
        ").history"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 22,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "8qRkDoU-wRL6",
        "outputId": "d9a76337-1517-4c0f-e077-d18aac75e45f"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1/200\n",
            "31/31 [==============================] - 2s 22ms/step - loss: 3.4542 - accuracy: 0.3628 - val_loss: 3.7172 - val_accuracy: 0.1111 - lr: 0.0010\n",
            "Epoch 2/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 3.0458 - accuracy: 0.4421 - val_loss: 3.7586 - val_accuracy: 0.1152 - lr: 0.0010\n",
            "Epoch 3/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.7627 - accuracy: 0.4797 - val_loss: 3.7097 - val_accuracy: 0.1399 - lr: 0.0010\n",
            "Epoch 4/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 2.5537 - accuracy: 0.5111 - val_loss: 3.5953 - val_accuracy: 0.2181 - lr: 0.0010\n",
            "Epoch 5/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.3710 - accuracy: 0.5383 - val_loss: 3.5137 - val_accuracy: 0.2449 - lr: 0.0010\n",
            "Epoch 6/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 2.2389 - accuracy: 0.5471 - val_loss: 2.6307 - val_accuracy: 0.4897 - lr: 0.0010\n",
            "Epoch 7/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 2.0787 - accuracy: 0.5785 - val_loss: 3.0615 - val_accuracy: 0.2160 - lr: 0.0010\n",
            "Epoch 8/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.9794 - accuracy: 0.5960 - val_loss: 2.6603 - val_accuracy: 0.2737 - lr: 0.0010\n",
            "Epoch 9/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.8989 - accuracy: 0.5924 - val_loss: 2.3850 - val_accuracy: 0.4198 - lr: 0.0010\n",
            "Epoch 10/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.7546 - accuracy: 0.6125 - val_loss: 2.5067 - val_accuracy: 0.3292 - lr: 0.0010\n",
            "Epoch 11/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.6573 - accuracy: 0.6356 - val_loss: 2.5389 - val_accuracy: 0.2778 - lr: 0.0010\n",
            "Epoch 12/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.5940 - accuracy: 0.6397 - val_loss: 2.0834 - val_accuracy: 0.4753 - lr: 0.0010\n",
            "Epoch 13/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.4987 - accuracy: 0.6624 - val_loss: 2.5094 - val_accuracy: 0.3169 - lr: 0.0010\n",
            "Epoch 14/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.4102 - accuracy: 0.6809 - val_loss: 1.8277 - val_accuracy: 0.5473 - lr: 0.0010\n",
            "Epoch 15/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.3651 - accuracy: 0.6783 - val_loss: 1.9165 - val_accuracy: 0.4918 - lr: 0.0010\n",
            "Epoch 16/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.3415 - accuracy: 0.6773 - val_loss: 2.5602 - val_accuracy: 0.2984 - lr: 0.0010\n",
            "Epoch 17/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.2839 - accuracy: 0.6891 - val_loss: 1.7590 - val_accuracy: 0.5288 - lr: 0.0010\n",
            "Epoch 18/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.2402 - accuracy: 0.6979 - val_loss: 1.5760 - val_accuracy: 0.5947 - lr: 0.0010\n",
            "Epoch 19/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.1565 - accuracy: 0.7087 - val_loss: 1.7514 - val_accuracy: 0.5432 - lr: 0.0010\n",
            "Epoch 20/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.1266 - accuracy: 0.7113 - val_loss: 1.5596 - val_accuracy: 0.5823 - lr: 0.0010\n",
            "Epoch 21/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.0786 - accuracy: 0.7231 - val_loss: 1.5089 - val_accuracy: 0.6111 - lr: 0.0010\n",
            "Epoch 22/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.0404 - accuracy: 0.7360 - val_loss: 1.5166 - val_accuracy: 0.6008 - lr: 0.0010\n",
            "Epoch 23/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.9758 - accuracy: 0.7422 - val_loss: 1.4657 - val_accuracy: 0.5926 - lr: 0.0010\n",
            "Epoch 24/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.9532 - accuracy: 0.7545 - val_loss: 1.2737 - val_accuracy: 0.6872 - lr: 0.0010\n",
            "Epoch 25/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.9204 - accuracy: 0.7581 - val_loss: 1.8141 - val_accuracy: 0.5123 - lr: 0.0010\n",
            "Epoch 26/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.9064 - accuracy: 0.7581 - val_loss: 1.3353 - val_accuracy: 0.6708 - lr: 0.0010\n",
            "Epoch 27/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8631 - accuracy: 0.7720 - val_loss: 1.3711 - val_accuracy: 0.6337 - lr: 0.0010\n",
            "Epoch 28/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.8247 - accuracy: 0.7730 - val_loss: 1.4084 - val_accuracy: 0.6235 - lr: 0.0010\n",
            "Epoch 29/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8008 - accuracy: 0.7962 - val_loss: 1.6759 - val_accuracy: 0.5802 - lr: 0.0010\n",
            "Epoch 30/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.7673 - accuracy: 0.7998 - val_loss: 1.2549 - val_accuracy: 0.6605 - lr: 0.0010\n",
            "Epoch 31/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6749 - accuracy: 0.8394 - val_loss: 1.3289 - val_accuracy: 0.6543 - lr: 0.0010\n",
            "Epoch 32/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.6825 - accuracy: 0.8209 - val_loss: 1.5917 - val_accuracy: 0.6132 - lr: 0.0010\n",
            "Epoch 33/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.6406 - accuracy: 0.8353 - val_loss: 1.3147 - val_accuracy: 0.6728 - lr: 0.0010\n",
            "Epoch 34/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6477 - accuracy: 0.8369 - val_loss: 1.5068 - val_accuracy: 0.5885 - lr: 0.0010\n",
            "Epoch 35/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6654 - accuracy: 0.8214 - val_loss: 1.5415 - val_accuracy: 0.6358 - lr: 0.0010\n",
            "Epoch 36/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6224 - accuracy: 0.8405 - val_loss: 1.4519 - val_accuracy: 0.6337 - lr: 0.0010\n",
            "Epoch 37/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.5615 - accuracy: 0.8641 - val_loss: 1.4220 - val_accuracy: 0.6337 - lr: 0.0010\n",
            "Epoch 38/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.5117 - accuracy: 0.8760 - val_loss: 1.3694 - val_accuracy: 0.6770 - lr: 0.0010\n",
            "Epoch 39/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.5163 - accuracy: 0.8755 - val_loss: 1.5283 - val_accuracy: 0.6461 - lr: 0.0010\n",
            "Epoch 40/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.4579 - accuracy: 0.8955 - val_loss: 1.3208 - val_accuracy: 0.6749 - lr: 5.0000e-04\n",
            "Epoch 41/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.3764 - accuracy: 0.9207 - val_loss: 1.3283 - val_accuracy: 0.6811 - lr: 5.0000e-04\n",
            "Epoch 42/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3705 - accuracy: 0.9218 - val_loss: 1.2606 - val_accuracy: 0.6749 - lr: 5.0000e-04\n",
            "Epoch 43/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.3324 - accuracy: 0.9418 - val_loss: 1.2937 - val_accuracy: 0.6872 - lr: 5.0000e-04\n",
            "Epoch 44/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.3201 - accuracy: 0.9465 - val_loss: 1.4286 - val_accuracy: 0.6646 - lr: 5.0000e-04\n",
            "Epoch 45/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.3042 - accuracy: 0.9454 - val_loss: 1.3676 - val_accuracy: 0.6852 - lr: 5.0000e-04\n",
            "Epoch 46/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.2779 - accuracy: 0.9588 - val_loss: 1.3499 - val_accuracy: 0.6934 - lr: 5.0000e-04\n",
            "Epoch 47/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.2954 - accuracy: 0.9490 - val_loss: 1.4527 - val_accuracy: 0.6728 - lr: 5.0000e-04\n",
            "Epoch 48/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.2891 - accuracy: 0.9449 - val_loss: 1.4306 - val_accuracy: 0.6811 - lr: 5.0000e-04\n",
            "Epoch 49/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2710 - accuracy: 0.9573 - val_loss: 1.3476 - val_accuracy: 0.6708 - lr: 5.0000e-04\n",
            "Epoch 50/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.2413 - accuracy: 0.9640 - val_loss: 1.4648 - val_accuracy: 0.6955 - lr: 5.0000e-04\n",
            "Epoch 51/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.2199 - accuracy: 0.9691 - val_loss: 1.3942 - val_accuracy: 0.6914 - lr: 5.0000e-04\n",
            "Epoch 52/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.2375 - accuracy: 0.9619 - val_loss: 1.4157 - val_accuracy: 0.6975 - lr: 5.0000e-04\n",
            "Epoch 53/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.2230 - accuracy: 0.9645 - val_loss: 1.4943 - val_accuracy: 0.7037 - lr: 5.0000e-04\n",
            "Epoch 54/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2111 - accuracy: 0.9712 - val_loss: 1.4900 - val_accuracy: 0.6790 - lr: 5.0000e-04\n",
            "Epoch 55/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2108 - accuracy: 0.9665 - val_loss: 1.4565 - val_accuracy: 0.6708 - lr: 5.0000e-04\n",
            "Epoch 56/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.2034 - accuracy: 0.9701 - val_loss: 1.5598 - val_accuracy: 0.6728 - lr: 5.0000e-04\n",
            "Epoch 57/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.2366 - accuracy: 0.9593 - val_loss: 1.4249 - val_accuracy: 0.6605 - lr: 5.0000e-04\n",
            "Epoch 58/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.2216 - accuracy: 0.9655 - val_loss: 1.4266 - val_accuracy: 0.6852 - lr: 5.0000e-04\n",
            "Epoch 59/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1825 - accuracy: 0.9732 - val_loss: 1.6559 - val_accuracy: 0.6872 - lr: 5.0000e-04\n",
            "Epoch 60/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.1866 - accuracy: 0.9763 - val_loss: 1.4863 - val_accuracy: 0.6749 - lr: 5.0000e-04\n",
            "Epoch 61/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.1790 - accuracy: 0.9743 - val_loss: 1.5534 - val_accuracy: 0.6584 - lr: 5.0000e-04\n",
            "Epoch 62/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1708 - accuracy: 0.9774 - val_loss: 1.4343 - val_accuracy: 0.6831 - lr: 5.0000e-04\n",
            "Epoch 63/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1628 - accuracy: 0.9840 - val_loss: 1.5294 - val_accuracy: 0.6975 - lr: 5.0000e-04\n",
            "Epoch 64/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1635 - accuracy: 0.9789 - val_loss: 1.5706 - val_accuracy: 0.6996 - lr: 5.0000e-04\n",
            "Epoch 65/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1708 - accuracy: 0.9779 - val_loss: 1.4733 - val_accuracy: 0.6914 - lr: 5.0000e-04\n",
            "Epoch 66/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1411 - accuracy: 0.9840 - val_loss: 1.4365 - val_accuracy: 0.6852 - lr: 5.0000e-04\n",
            "Epoch 67/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.1472 - accuracy: 0.9825 - val_loss: 1.5712 - val_accuracy: 0.6975 - lr: 5.0000e-04\n",
            "Epoch 68/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1592 - accuracy: 0.9799 - val_loss: 1.7015 - val_accuracy: 0.6893 - lr: 5.0000e-04\n",
            "Epoch 69/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1330 - accuracy: 0.9866 - val_loss: 1.4622 - val_accuracy: 0.6893 - lr: 2.5000e-04\n",
            "Epoch 70/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.1191 - accuracy: 0.9923 - val_loss: 1.4999 - val_accuracy: 0.6893 - lr: 2.5000e-04\n",
            "Epoch 71/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.1204 - accuracy: 0.9928 - val_loss: 1.5332 - val_accuracy: 0.6893 - lr: 2.5000e-04\n",
            "Epoch 72/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1071 - accuracy: 0.9979 - val_loss: 1.5556 - val_accuracy: 0.6811 - lr: 2.5000e-04\n",
            "Epoch 73/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1068 - accuracy: 0.9949 - val_loss: 1.5362 - val_accuracy: 0.6975 - lr: 2.5000e-04\n",
            "Epoch 74/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1019 - accuracy: 0.9979 - val_loss: 1.4942 - val_accuracy: 0.6955 - lr: 2.5000e-04\n",
            "Epoch 75/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.1033 - accuracy: 0.9954 - val_loss: 1.4850 - val_accuracy: 0.6893 - lr: 2.5000e-04\n",
            "Epoch 76/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.1091 - accuracy: 0.9938 - val_loss: 1.4796 - val_accuracy: 0.7016 - lr: 2.5000e-04\n",
            "Epoch 77/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1072 - accuracy: 0.9933 - val_loss: 1.5865 - val_accuracy: 0.6872 - lr: 2.5000e-04\n",
            "Epoch 78/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0957 - accuracy: 0.9974 - val_loss: 1.5486 - val_accuracy: 0.6955 - lr: 2.5000e-04\n",
            "Epoch 79/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0922 - accuracy: 0.9985 - val_loss: 1.5400 - val_accuracy: 0.6872 - lr: 2.5000e-04\n",
            "Epoch 80/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0921 - accuracy: 0.9969 - val_loss: 1.4971 - val_accuracy: 0.6831 - lr: 2.5000e-04\n",
            "Epoch 81/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0894 - accuracy: 0.9990 - val_loss: 1.5170 - val_accuracy: 0.7037 - lr: 2.5000e-04\n",
            "Epoch 82/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0895 - accuracy: 0.9974 - val_loss: 1.4906 - val_accuracy: 0.7119 - lr: 2.5000e-04\n",
            "Epoch 83/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0874 - accuracy: 0.9969 - val_loss: 1.5065 - val_accuracy: 0.6914 - lr: 2.5000e-04\n",
            "Epoch 84/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0855 - accuracy: 0.9995 - val_loss: 1.5193 - val_accuracy: 0.7016 - lr: 2.5000e-04\n",
            "Epoch 85/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0899 - accuracy: 0.9979 - val_loss: 1.6235 - val_accuracy: 0.6975 - lr: 2.5000e-04\n",
            "Epoch 86/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0861 - accuracy: 0.9985 - val_loss: 1.5240 - val_accuracy: 0.6996 - lr: 2.5000e-04\n",
            "Epoch 87/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0847 - accuracy: 0.9985 - val_loss: 1.4882 - val_accuracy: 0.7016 - lr: 2.5000e-04\n",
            "Epoch 88/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0824 - accuracy: 0.9990 - val_loss: 1.5799 - val_accuracy: 0.6996 - lr: 2.5000e-04\n",
            "Epoch 89/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0865 - accuracy: 0.9959 - val_loss: 1.5535 - val_accuracy: 0.6955 - lr: 2.5000e-04\n",
            "Epoch 90/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1037 - accuracy: 0.9923 - val_loss: 1.5428 - val_accuracy: 0.7140 - lr: 2.5000e-04\n",
            "Epoch 91/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0851 - accuracy: 0.9964 - val_loss: 1.5746 - val_accuracy: 0.7078 - lr: 2.5000e-04\n",
            "Epoch 92/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0792 - accuracy: 0.9990 - val_loss: 1.4868 - val_accuracy: 0.7037 - lr: 2.5000e-04\n",
            "Epoch 93/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0759 - accuracy: 0.9995 - val_loss: 1.5077 - val_accuracy: 0.6996 - lr: 2.5000e-04\n",
            "Epoch 94/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0738 - accuracy: 0.9990 - val_loss: 1.4969 - val_accuracy: 0.6996 - lr: 2.5000e-04\n",
            "Epoch 95/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0762 - accuracy: 0.9974 - val_loss: 1.4836 - val_accuracy: 0.6914 - lr: 2.5000e-04\n",
            "Epoch 96/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0738 - accuracy: 0.9990 - val_loss: 1.4901 - val_accuracy: 0.7099 - lr: 2.5000e-04\n",
            "Epoch 97/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0757 - accuracy: 0.9985 - val_loss: 1.4969 - val_accuracy: 0.6852 - lr: 2.5000e-04\n",
            "Epoch 98/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0769 - accuracy: 0.9959 - val_loss: 1.5046 - val_accuracy: 0.6955 - lr: 2.5000e-04\n",
            "Epoch 99/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0725 - accuracy: 0.9995 - val_loss: 1.5298 - val_accuracy: 0.6934 - lr: 2.5000e-04\n",
            "Epoch 100/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0700 - accuracy: 1.0000 - val_loss: 1.5159 - val_accuracy: 0.6975 - lr: 2.5000e-04\n",
            "Epoch 101/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0698 - accuracy: 0.9995 - val_loss: 1.5261 - val_accuracy: 0.6975 - lr: 2.5000e-04\n",
            "Epoch 102/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0676 - accuracy: 1.0000 - val_loss: 1.5608 - val_accuracy: 0.6955 - lr: 2.5000e-04\n",
            "Epoch 103/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0666 - accuracy: 0.9995 - val_loss: 1.5153 - val_accuracy: 0.6914 - lr: 2.5000e-04\n",
            "Epoch 104/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0668 - accuracy: 1.0000 - val_loss: 1.4899 - val_accuracy: 0.6934 - lr: 2.5000e-04\n",
            "Epoch 105/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0663 - accuracy: 1.0000 - val_loss: 1.5121 - val_accuracy: 0.6955 - lr: 2.5000e-04\n",
            "Epoch 106/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0636 - accuracy: 1.0000 - val_loss: 1.5476 - val_accuracy: 0.6934 - lr: 1.2500e-04\n",
            "Epoch 107/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0652 - accuracy: 0.9995 - val_loss: 1.5087 - val_accuracy: 0.7016 - lr: 1.2500e-04\n",
            "Epoch 108/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0628 - accuracy: 1.0000 - val_loss: 1.4920 - val_accuracy: 0.7037 - lr: 1.2500e-04\n",
            "Epoch 109/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0620 - accuracy: 1.0000 - val_loss: 1.4984 - val_accuracy: 0.7016 - lr: 1.2500e-04\n",
            "Epoch 110/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0629 - accuracy: 0.9995 - val_loss: 1.5012 - val_accuracy: 0.6975 - lr: 1.2500e-04\n",
            "Epoch 111/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0615 - accuracy: 1.0000 - val_loss: 1.5033 - val_accuracy: 0.6955 - lr: 1.2500e-04\n",
            "Epoch 112/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0607 - accuracy: 1.0000 - val_loss: 1.5089 - val_accuracy: 0.6893 - lr: 1.2500e-04\n",
            "Epoch 113/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0611 - accuracy: 1.0000 - val_loss: 1.5124 - val_accuracy: 0.6975 - lr: 1.2500e-04\n",
            "Epoch 114/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0593 - accuracy: 1.0000 - val_loss: 1.4890 - val_accuracy: 0.6996 - lr: 1.2500e-04\n",
            "Epoch 115/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0611 - accuracy: 0.9990 - val_loss: 1.4746 - val_accuracy: 0.6872 - lr: 1.2500e-04\n",
            "Epoch 116/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0585 - accuracy: 1.0000 - val_loss: 1.4888 - val_accuracy: 0.6975 - lr: 1.2500e-04\n",
            "Epoch 117/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0599 - accuracy: 0.9995 - val_loss: 1.4862 - val_accuracy: 0.6955 - lr: 1.2500e-04\n",
            "Epoch 118/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0619 - accuracy: 0.9990 - val_loss: 1.5267 - val_accuracy: 0.6955 - lr: 1.2500e-04\n",
            "Epoch 119/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0615 - accuracy: 0.9995 - val_loss: 1.4940 - val_accuracy: 0.6975 - lr: 1.2500e-04\n",
            "Epoch 120/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0623 - accuracy: 0.9995 - val_loss: 1.4901 - val_accuracy: 0.6872 - lr: 1.2500e-04\n"
          ]
        }
      ],
      "source": [
        "history7 = model7.fit(\n",
        "    x = x_train7,\n",
        "    y = y_train7,\n",
        "    batch_size = 64,\n",
        "    epochs = 200,\n",
        "    validation_data = (x_test7, y_test7),\n",
        "    callbacks = [\n",
        "        tfk.callbacks.EarlyStopping(monitor='val_accuracy', mode='max', patience=30, restore_best_weights=True),\n",
        "        tfk.callbacks.ReduceLROnPlateau(monitor='val_accuracy', mode='max', patience=15, factor=0.5, min_lr=1e-5)\n",
        "    ]\n",
        ").history"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 23,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "fnZgHNiTwRGs",
        "outputId": "3b604fd9-3547-44a6-ca59-220625358459"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1/200\n",
            "31/31 [==============================] - 2s 22ms/step - loss: 3.4810 - accuracy: 0.3649 - val_loss: 3.6783 - val_accuracy: 0.1255 - lr: 0.0010\n",
            "Epoch 2/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 3.0235 - accuracy: 0.4370 - val_loss: 3.8187 - val_accuracy: 0.1255 - lr: 0.0010\n",
            "Epoch 3/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.6952 - accuracy: 0.4941 - val_loss: 3.5623 - val_accuracy: 0.2016 - lr: 0.0010\n",
            "Epoch 4/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.4578 - accuracy: 0.5363 - val_loss: 3.3928 - val_accuracy: 0.1811 - lr: 0.0010\n",
            "Epoch 5/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.3293 - accuracy: 0.5291 - val_loss: 3.3275 - val_accuracy: 0.1934 - lr: 0.0010\n",
            "Epoch 6/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.2023 - accuracy: 0.5502 - val_loss: 3.2155 - val_accuracy: 0.2181 - lr: 0.0010\n",
            "Epoch 7/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.0370 - accuracy: 0.5769 - val_loss: 3.0717 - val_accuracy: 0.2346 - lr: 0.0010\n",
            "Epoch 8/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.9183 - accuracy: 0.5939 - val_loss: 2.7853 - val_accuracy: 0.2654 - lr: 0.0010\n",
            "Epoch 9/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.8001 - accuracy: 0.6300 - val_loss: 2.4822 - val_accuracy: 0.3663 - lr: 0.0010\n",
            "Epoch 10/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.7307 - accuracy: 0.6130 - val_loss: 2.7487 - val_accuracy: 0.2860 - lr: 0.0010\n",
            "Epoch 11/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.6730 - accuracy: 0.6253 - val_loss: 2.2468 - val_accuracy: 0.4691 - lr: 0.0010\n",
            "Epoch 12/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.5408 - accuracy: 0.6541 - val_loss: 2.0076 - val_accuracy: 0.5185 - lr: 0.0010\n",
            "Epoch 13/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.5037 - accuracy: 0.6454 - val_loss: 2.4094 - val_accuracy: 0.3560 - lr: 0.0010\n",
            "Epoch 14/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.4367 - accuracy: 0.6459 - val_loss: 1.8422 - val_accuracy: 0.5473 - lr: 0.0010\n",
            "Epoch 15/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.3683 - accuracy: 0.6634 - val_loss: 2.0613 - val_accuracy: 0.4877 - lr: 0.0010\n",
            "Epoch 16/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.2771 - accuracy: 0.6912 - val_loss: 1.6325 - val_accuracy: 0.5905 - lr: 0.0010\n",
            "Epoch 17/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.2412 - accuracy: 0.6948 - val_loss: 1.8252 - val_accuracy: 0.5453 - lr: 0.0010\n",
            "Epoch 18/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.2029 - accuracy: 0.6948 - val_loss: 1.5237 - val_accuracy: 0.6070 - lr: 0.0010\n",
            "Epoch 19/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.1292 - accuracy: 0.7205 - val_loss: 1.4564 - val_accuracy: 0.5988 - lr: 0.0010\n",
            "Epoch 20/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.0751 - accuracy: 0.7190 - val_loss: 1.6515 - val_accuracy: 0.5658 - lr: 0.0010\n",
            "Epoch 21/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.0355 - accuracy: 0.7452 - val_loss: 1.5236 - val_accuracy: 0.5823 - lr: 0.0010\n",
            "Epoch 22/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.0510 - accuracy: 0.7231 - val_loss: 1.9628 - val_accuracy: 0.4383 - lr: 0.0010\n",
            "Epoch 23/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.9541 - accuracy: 0.7524 - val_loss: 1.5668 - val_accuracy: 0.5926 - lr: 0.0010\n",
            "Epoch 24/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.9482 - accuracy: 0.7514 - val_loss: 1.3320 - val_accuracy: 0.6399 - lr: 0.0010\n",
            "Epoch 25/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8584 - accuracy: 0.7746 - val_loss: 1.4691 - val_accuracy: 0.5885 - lr: 0.0010\n",
            "Epoch 26/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8161 - accuracy: 0.7818 - val_loss: 1.3560 - val_accuracy: 0.6317 - lr: 0.0010\n",
            "Epoch 27/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.8712 - accuracy: 0.7689 - val_loss: 1.4312 - val_accuracy: 0.6132 - lr: 0.0010\n",
            "Epoch 28/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.8148 - accuracy: 0.7921 - val_loss: 1.4046 - val_accuracy: 0.6235 - lr: 0.0010\n",
            "Epoch 29/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.7298 - accuracy: 0.8137 - val_loss: 1.9844 - val_accuracy: 0.5658 - lr: 0.0010\n",
            "Epoch 30/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.7196 - accuracy: 0.8075 - val_loss: 1.6537 - val_accuracy: 0.5761 - lr: 0.0010\n",
            "Epoch 31/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6935 - accuracy: 0.8199 - val_loss: 1.5523 - val_accuracy: 0.6255 - lr: 0.0010\n",
            "Epoch 32/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6657 - accuracy: 0.8230 - val_loss: 2.1709 - val_accuracy: 0.5350 - lr: 0.0010\n",
            "Epoch 33/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6673 - accuracy: 0.8204 - val_loss: 1.3676 - val_accuracy: 0.6564 - lr: 0.0010\n",
            "Epoch 34/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6205 - accuracy: 0.8374 - val_loss: 1.3165 - val_accuracy: 0.6235 - lr: 0.0010\n",
            "Epoch 35/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.5873 - accuracy: 0.8487 - val_loss: 1.4159 - val_accuracy: 0.6193 - lr: 0.0010\n",
            "Epoch 36/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.5727 - accuracy: 0.8497 - val_loss: 1.3750 - val_accuracy: 0.6420 - lr: 0.0010\n",
            "Epoch 37/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.5364 - accuracy: 0.8636 - val_loss: 1.6138 - val_accuracy: 0.6296 - lr: 0.0010\n",
            "Epoch 38/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.5207 - accuracy: 0.8662 - val_loss: 1.3525 - val_accuracy: 0.6358 - lr: 0.0010\n",
            "Epoch 39/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.4864 - accuracy: 0.8811 - val_loss: 1.3770 - val_accuracy: 0.6358 - lr: 0.0010\n",
            "Epoch 40/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.4567 - accuracy: 0.8929 - val_loss: 1.6942 - val_accuracy: 0.6255 - lr: 0.0010\n",
            "Epoch 41/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.4581 - accuracy: 0.8796 - val_loss: 1.3761 - val_accuracy: 0.6543 - lr: 0.0010\n",
            "Epoch 42/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.5008 - accuracy: 0.8708 - val_loss: 1.6267 - val_accuracy: 0.5473 - lr: 0.0010\n",
            "Epoch 43/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.4432 - accuracy: 0.8909 - val_loss: 1.7065 - val_accuracy: 0.6420 - lr: 0.0010\n",
            "Epoch 44/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3981 - accuracy: 0.9104 - val_loss: 1.7399 - val_accuracy: 0.6214 - lr: 0.0010\n",
            "Epoch 45/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3744 - accuracy: 0.9110 - val_loss: 1.4981 - val_accuracy: 0.6337 - lr: 0.0010\n",
            "Epoch 46/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3714 - accuracy: 0.9213 - val_loss: 1.5527 - val_accuracy: 0.6399 - lr: 0.0010\n",
            "Epoch 47/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3141 - accuracy: 0.9388 - val_loss: 1.5081 - val_accuracy: 0.6667 - lr: 0.0010\n",
            "Epoch 48/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.3634 - accuracy: 0.9218 - val_loss: 1.4262 - val_accuracy: 0.6502 - lr: 0.0010\n",
            "Epoch 49/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3599 - accuracy: 0.9161 - val_loss: 1.4963 - val_accuracy: 0.6132 - lr: 0.0010\n",
            "Epoch 50/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3159 - accuracy: 0.9295 - val_loss: 1.4683 - val_accuracy: 0.6605 - lr: 0.0010\n",
            "Epoch 51/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.2813 - accuracy: 0.9480 - val_loss: 1.4679 - val_accuracy: 0.6296 - lr: 0.0010\n",
            "Epoch 52/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.3031 - accuracy: 0.9341 - val_loss: 1.6148 - val_accuracy: 0.6317 - lr: 0.0010\n",
            "Epoch 53/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.3099 - accuracy: 0.9331 - val_loss: 1.5506 - val_accuracy: 0.6605 - lr: 0.0010\n",
            "Epoch 54/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2884 - accuracy: 0.9424 - val_loss: 1.6174 - val_accuracy: 0.6667 - lr: 0.0010\n",
            "Epoch 55/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3260 - accuracy: 0.9254 - val_loss: 1.7172 - val_accuracy: 0.6399 - lr: 0.0010\n",
            "Epoch 56/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2608 - accuracy: 0.9527 - val_loss: 1.6025 - val_accuracy: 0.6626 - lr: 0.0010\n",
            "Epoch 57/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.2842 - accuracy: 0.9382 - val_loss: 1.6417 - val_accuracy: 0.6214 - lr: 0.0010\n",
            "Epoch 58/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.2507 - accuracy: 0.9496 - val_loss: 1.7214 - val_accuracy: 0.6420 - lr: 0.0010\n",
            "Epoch 59/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2187 - accuracy: 0.9676 - val_loss: 1.6744 - val_accuracy: 0.6481 - lr: 0.0010\n",
            "Epoch 60/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2050 - accuracy: 0.9707 - val_loss: 1.9438 - val_accuracy: 0.6399 - lr: 0.0010\n",
            "Epoch 61/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2121 - accuracy: 0.9650 - val_loss: 1.5843 - val_accuracy: 0.6523 - lr: 0.0010\n",
            "Epoch 62/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.2343 - accuracy: 0.9516 - val_loss: 1.7649 - val_accuracy: 0.6687 - lr: 0.0010\n",
            "Epoch 63/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.2089 - accuracy: 0.9635 - val_loss: 1.7266 - val_accuracy: 0.6255 - lr: 0.0010\n",
            "Epoch 64/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2100 - accuracy: 0.9599 - val_loss: 1.9582 - val_accuracy: 0.6317 - lr: 0.0010\n",
            "Epoch 65/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1802 - accuracy: 0.9738 - val_loss: 1.7728 - val_accuracy: 0.6708 - lr: 0.0010\n",
            "Epoch 66/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1767 - accuracy: 0.9738 - val_loss: 1.7826 - val_accuracy: 0.6811 - lr: 0.0010\n",
            "Epoch 67/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1892 - accuracy: 0.9676 - val_loss: 1.7360 - val_accuracy: 0.6626 - lr: 0.0010\n",
            "Epoch 68/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1547 - accuracy: 0.9815 - val_loss: 1.6851 - val_accuracy: 0.6564 - lr: 0.0010\n",
            "Epoch 69/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.1697 - accuracy: 0.9758 - val_loss: 1.8145 - val_accuracy: 0.6379 - lr: 0.0010\n",
            "Epoch 70/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.1547 - accuracy: 0.9794 - val_loss: 1.9204 - val_accuracy: 0.6687 - lr: 0.0010\n",
            "Epoch 71/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.1394 - accuracy: 0.9830 - val_loss: 1.8611 - val_accuracy: 0.6749 - lr: 0.0010\n",
            "Epoch 72/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1610 - accuracy: 0.9768 - val_loss: 1.8250 - val_accuracy: 0.6399 - lr: 0.0010\n",
            "Epoch 73/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.2171 - accuracy: 0.9537 - val_loss: 1.7578 - val_accuracy: 0.6502 - lr: 0.0010\n",
            "Epoch 74/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2103 - accuracy: 0.9568 - val_loss: 1.6116 - val_accuracy: 0.6605 - lr: 0.0010\n",
            "Epoch 75/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1472 - accuracy: 0.9799 - val_loss: 2.1723 - val_accuracy: 0.6584 - lr: 0.0010\n",
            "Epoch 76/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.1573 - accuracy: 0.9789 - val_loss: 1.8759 - val_accuracy: 0.6440 - lr: 0.0010\n",
            "Epoch 77/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1438 - accuracy: 0.9835 - val_loss: 1.8765 - val_accuracy: 0.6708 - lr: 0.0010\n",
            "Epoch 78/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.1333 - accuracy: 0.9815 - val_loss: 1.6435 - val_accuracy: 0.6523 - lr: 0.0010\n",
            "Epoch 79/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1223 - accuracy: 0.9856 - val_loss: 1.7229 - val_accuracy: 0.6667 - lr: 0.0010\n",
            "Epoch 80/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1239 - accuracy: 0.9876 - val_loss: 1.6207 - val_accuracy: 0.6543 - lr: 0.0010\n",
            "Epoch 81/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1250 - accuracy: 0.9840 - val_loss: 1.6515 - val_accuracy: 0.6605 - lr: 0.0010\n",
            "Epoch 82/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0886 - accuracy: 0.9979 - val_loss: 1.6066 - val_accuracy: 0.6687 - lr: 5.0000e-04\n",
            "Epoch 83/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0831 - accuracy: 0.9979 - val_loss: 1.6422 - val_accuracy: 0.6749 - lr: 5.0000e-04\n",
            "Epoch 84/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0792 - accuracy: 0.9979 - val_loss: 1.6510 - val_accuracy: 0.6872 - lr: 5.0000e-04\n",
            "Epoch 85/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0733 - accuracy: 0.9995 - val_loss: 1.5439 - val_accuracy: 0.6728 - lr: 5.0000e-04\n",
            "Epoch 86/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0747 - accuracy: 0.9990 - val_loss: 1.5460 - val_accuracy: 0.6687 - lr: 5.0000e-04\n",
            "Epoch 87/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0710 - accuracy: 0.9990 - val_loss: 1.5527 - val_accuracy: 0.6831 - lr: 5.0000e-04\n",
            "Epoch 88/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0686 - accuracy: 0.9995 - val_loss: 1.5801 - val_accuracy: 0.6708 - lr: 5.0000e-04\n",
            "Epoch 89/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0670 - accuracy: 1.0000 - val_loss: 1.5376 - val_accuracy: 0.6790 - lr: 5.0000e-04\n",
            "Epoch 90/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0707 - accuracy: 0.9979 - val_loss: 1.4909 - val_accuracy: 0.6708 - lr: 5.0000e-04\n",
            "Epoch 91/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0696 - accuracy: 0.9979 - val_loss: 1.5376 - val_accuracy: 0.6708 - lr: 5.0000e-04\n",
            "Epoch 92/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0807 - accuracy: 0.9933 - val_loss: 1.5174 - val_accuracy: 0.6728 - lr: 5.0000e-04\n",
            "Epoch 93/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0694 - accuracy: 0.9974 - val_loss: 1.5425 - val_accuracy: 0.6872 - lr: 5.0000e-04\n",
            "Epoch 94/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0682 - accuracy: 0.9985 - val_loss: 1.6162 - val_accuracy: 0.6708 - lr: 5.0000e-04\n",
            "Epoch 95/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0628 - accuracy: 0.9995 - val_loss: 1.6197 - val_accuracy: 0.6893 - lr: 5.0000e-04\n",
            "Epoch 96/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0605 - accuracy: 0.9995 - val_loss: 1.5460 - val_accuracy: 0.6502 - lr: 5.0000e-04\n",
            "Epoch 97/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0579 - accuracy: 0.9995 - val_loss: 1.5464 - val_accuracy: 0.6831 - lr: 5.0000e-04\n",
            "Epoch 98/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0550 - accuracy: 1.0000 - val_loss: 1.5171 - val_accuracy: 0.6790 - lr: 5.0000e-04\n",
            "Epoch 99/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0560 - accuracy: 0.9995 - val_loss: 1.5278 - val_accuracy: 0.6687 - lr: 5.0000e-04\n",
            "Epoch 100/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0540 - accuracy: 1.0000 - val_loss: 1.6277 - val_accuracy: 0.6831 - lr: 5.0000e-04\n",
            "Epoch 101/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0527 - accuracy: 1.0000 - val_loss: 1.5204 - val_accuracy: 0.6872 - lr: 5.0000e-04\n",
            "Epoch 102/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0507 - accuracy: 1.0000 - val_loss: 1.5133 - val_accuracy: 0.6770 - lr: 5.0000e-04\n",
            "Epoch 103/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0507 - accuracy: 1.0000 - val_loss: 1.4617 - val_accuracy: 0.6831 - lr: 5.0000e-04\n",
            "Epoch 104/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0506 - accuracy: 1.0000 - val_loss: 1.4957 - val_accuracy: 0.6790 - lr: 5.0000e-04\n",
            "Epoch 105/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0492 - accuracy: 0.9995 - val_loss: 1.4390 - val_accuracy: 0.6790 - lr: 5.0000e-04\n",
            "Epoch 106/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0490 - accuracy: 0.9995 - val_loss: 1.5001 - val_accuracy: 0.6831 - lr: 5.0000e-04\n",
            "Epoch 107/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0483 - accuracy: 0.9995 - val_loss: 1.5206 - val_accuracy: 0.6770 - lr: 5.0000e-04\n",
            "Epoch 108/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0568 - accuracy: 0.9969 - val_loss: 1.6516 - val_accuracy: 0.6543 - lr: 5.0000e-04\n",
            "Epoch 109/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0590 - accuracy: 0.9954 - val_loss: 1.5354 - val_accuracy: 0.6872 - lr: 5.0000e-04\n",
            "Epoch 110/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0569 - accuracy: 0.9985 - val_loss: 1.5947 - val_accuracy: 0.6770 - lr: 5.0000e-04\n",
            "Epoch 111/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0509 - accuracy: 1.0000 - val_loss: 1.6194 - val_accuracy: 0.6728 - lr: 2.5000e-04\n",
            "Epoch 112/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0499 - accuracy: 1.0000 - val_loss: 1.5620 - val_accuracy: 0.6770 - lr: 2.5000e-04\n",
            "Epoch 113/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0485 - accuracy: 0.9995 - val_loss: 1.5440 - val_accuracy: 0.6831 - lr: 2.5000e-04\n",
            "Epoch 114/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0467 - accuracy: 1.0000 - val_loss: 1.5531 - val_accuracy: 0.6852 - lr: 2.5000e-04\n",
            "Epoch 115/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0475 - accuracy: 0.9995 - val_loss: 1.5309 - val_accuracy: 0.6831 - lr: 2.5000e-04\n",
            "Epoch 116/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0454 - accuracy: 0.9995 - val_loss: 1.5515 - val_accuracy: 0.6934 - lr: 2.5000e-04\n",
            "Epoch 117/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0447 - accuracy: 1.0000 - val_loss: 1.5318 - val_accuracy: 0.6852 - lr: 2.5000e-04\n",
            "Epoch 118/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0439 - accuracy: 1.0000 - val_loss: 1.5230 - val_accuracy: 0.6811 - lr: 2.5000e-04\n",
            "Epoch 119/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0444 - accuracy: 0.9995 - val_loss: 1.5338 - val_accuracy: 0.6893 - lr: 2.5000e-04\n",
            "Epoch 120/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0446 - accuracy: 0.9995 - val_loss: 1.5325 - val_accuracy: 0.6893 - lr: 2.5000e-04\n",
            "Epoch 121/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0440 - accuracy: 1.0000 - val_loss: 1.5116 - val_accuracy: 0.6811 - lr: 2.5000e-04\n",
            "Epoch 122/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0419 - accuracy: 1.0000 - val_loss: 1.5134 - val_accuracy: 0.6749 - lr: 2.5000e-04\n",
            "Epoch 123/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0423 - accuracy: 1.0000 - val_loss: 1.5127 - val_accuracy: 0.6770 - lr: 2.5000e-04\n",
            "Epoch 124/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0425 - accuracy: 1.0000 - val_loss: 1.4990 - val_accuracy: 0.6831 - lr: 2.5000e-04\n",
            "Epoch 125/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0426 - accuracy: 0.9995 - val_loss: 1.4614 - val_accuracy: 0.6770 - lr: 2.5000e-04\n",
            "Epoch 126/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0418 - accuracy: 1.0000 - val_loss: 1.4722 - val_accuracy: 0.6749 - lr: 2.5000e-04\n",
            "Epoch 127/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0411 - accuracy: 1.0000 - val_loss: 1.4657 - val_accuracy: 0.6728 - lr: 2.5000e-04\n",
            "Epoch 128/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0410 - accuracy: 1.0000 - val_loss: 1.4471 - val_accuracy: 0.6811 - lr: 2.5000e-04\n",
            "Epoch 129/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0414 - accuracy: 1.0000 - val_loss: 1.4808 - val_accuracy: 0.6790 - lr: 2.5000e-04\n",
            "Epoch 130/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0408 - accuracy: 1.0000 - val_loss: 1.4877 - val_accuracy: 0.6893 - lr: 2.5000e-04\n",
            "Epoch 131/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0414 - accuracy: 1.0000 - val_loss: 1.5255 - val_accuracy: 0.6831 - lr: 2.5000e-04\n",
            "Epoch 132/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0416 - accuracy: 0.9995 - val_loss: 1.4925 - val_accuracy: 0.6770 - lr: 1.2500e-04\n",
            "Epoch 133/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0408 - accuracy: 1.0000 - val_loss: 1.4913 - val_accuracy: 0.6728 - lr: 1.2500e-04\n",
            "Epoch 134/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0409 - accuracy: 0.9995 - val_loss: 1.5212 - val_accuracy: 0.6770 - lr: 1.2500e-04\n",
            "Epoch 135/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0401 - accuracy: 1.0000 - val_loss: 1.5063 - val_accuracy: 0.6831 - lr: 1.2500e-04\n",
            "Epoch 136/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0399 - accuracy: 1.0000 - val_loss: 1.4968 - val_accuracy: 0.6790 - lr: 1.2500e-04\n",
            "Epoch 137/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0384 - accuracy: 1.0000 - val_loss: 1.4943 - val_accuracy: 0.6770 - lr: 1.2500e-04\n",
            "Epoch 138/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0396 - accuracy: 1.0000 - val_loss: 1.5331 - val_accuracy: 0.6811 - lr: 1.2500e-04\n",
            "Epoch 139/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0383 - accuracy: 1.0000 - val_loss: 1.5589 - val_accuracy: 0.6770 - lr: 1.2500e-04\n",
            "Epoch 140/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0382 - accuracy: 1.0000 - val_loss: 1.5494 - val_accuracy: 0.6749 - lr: 1.2500e-04\n",
            "Epoch 141/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0386 - accuracy: 1.0000 - val_loss: 1.5274 - val_accuracy: 0.6790 - lr: 1.2500e-04\n",
            "Epoch 142/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0389 - accuracy: 0.9995 - val_loss: 1.5338 - val_accuracy: 0.6770 - lr: 1.2500e-04\n",
            "Epoch 143/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0381 - accuracy: 1.0000 - val_loss: 1.5531 - val_accuracy: 0.6790 - lr: 1.2500e-04\n",
            "Epoch 144/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0381 - accuracy: 1.0000 - val_loss: 1.4860 - val_accuracy: 0.6728 - lr: 1.2500e-04\n",
            "Epoch 145/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0383 - accuracy: 1.0000 - val_loss: 1.4903 - val_accuracy: 0.6770 - lr: 1.2500e-04\n",
            "Epoch 146/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0366 - accuracy: 1.0000 - val_loss: 1.4861 - val_accuracy: 0.6749 - lr: 1.2500e-04\n"
          ]
        }
      ],
      "source": [
        "history8 = model8.fit(\n",
        "    x = x_train8,\n",
        "    y = y_train8,\n",
        "    batch_size = 64,\n",
        "    epochs = 200,\n",
        "    validation_data = (x_test8, y_test8),\n",
        "    callbacks = [\n",
        "        tfk.callbacks.EarlyStopping(monitor='val_accuracy', mode='max', patience=30, restore_best_weights=True),\n",
        "        tfk.callbacks.ReduceLROnPlateau(monitor='val_accuracy', mode='max', patience=15, factor=0.5, min_lr=1e-5)\n",
        "    ]\n",
        ").history"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 24,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ZEMxC6eDwRCk",
        "outputId": "c75e0cb0-a910-4c9b-cdae-d403f3d27706"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1/200\n",
            "31/31 [==============================] - 2s 26ms/step - loss: 3.5070 - accuracy: 0.3484 - val_loss: 3.7138 - val_accuracy: 0.1091 - lr: 0.0010\n",
            "Epoch 2/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 3.0453 - accuracy: 0.4164 - val_loss: 3.7533 - val_accuracy: 0.1193 - lr: 0.0010\n",
            "Epoch 3/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.7897 - accuracy: 0.4683 - val_loss: 3.6465 - val_accuracy: 0.1379 - lr: 0.0010\n",
            "Epoch 4/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 2.5764 - accuracy: 0.5224 - val_loss: 3.3418 - val_accuracy: 0.1132 - lr: 0.0010\n",
            "Epoch 5/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 2.3723 - accuracy: 0.5409 - val_loss: 3.1829 - val_accuracy: 0.2449 - lr: 0.0010\n",
            "Epoch 6/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 2.2770 - accuracy: 0.5435 - val_loss: 3.1574 - val_accuracy: 0.2243 - lr: 0.0010\n",
            "Epoch 7/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 2.1263 - accuracy: 0.5723 - val_loss: 3.2383 - val_accuracy: 0.2160 - lr: 0.0010\n",
            "Epoch 8/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 2.0264 - accuracy: 0.5795 - val_loss: 2.8480 - val_accuracy: 0.2778 - lr: 0.0010\n",
            "Epoch 9/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.9037 - accuracy: 0.5991 - val_loss: 2.9170 - val_accuracy: 0.2737 - lr: 0.0010\n",
            "Epoch 10/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.7930 - accuracy: 0.6089 - val_loss: 2.2086 - val_accuracy: 0.5226 - lr: 0.0010\n",
            "Epoch 11/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.7126 - accuracy: 0.6222 - val_loss: 2.0382 - val_accuracy: 0.4959 - lr: 0.0010\n",
            "Epoch 12/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.6151 - accuracy: 0.6361 - val_loss: 1.9941 - val_accuracy: 0.4733 - lr: 0.0010\n",
            "Epoch 13/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.5134 - accuracy: 0.6557 - val_loss: 1.9831 - val_accuracy: 0.4506 - lr: 0.0010\n",
            "Epoch 14/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.4686 - accuracy: 0.6583 - val_loss: 2.0290 - val_accuracy: 0.4650 - lr: 0.0010\n",
            "Epoch 15/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.4236 - accuracy: 0.6650 - val_loss: 1.5793 - val_accuracy: 0.6337 - lr: 0.0010\n",
            "Epoch 16/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.3485 - accuracy: 0.6747 - val_loss: 1.6422 - val_accuracy: 0.5864 - lr: 0.0010\n",
            "Epoch 17/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.3274 - accuracy: 0.6680 - val_loss: 1.7948 - val_accuracy: 0.4938 - lr: 0.0010\n",
            "Epoch 18/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.2503 - accuracy: 0.6907 - val_loss: 1.4931 - val_accuracy: 0.5967 - lr: 0.0010\n",
            "Epoch 19/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.1845 - accuracy: 0.6994 - val_loss: 1.3776 - val_accuracy: 0.6461 - lr: 0.0010\n",
            "Epoch 20/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.1110 - accuracy: 0.7169 - val_loss: 1.3976 - val_accuracy: 0.6379 - lr: 0.0010\n",
            "Epoch 21/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.0887 - accuracy: 0.7190 - val_loss: 1.3990 - val_accuracy: 0.6235 - lr: 0.0010\n",
            "Epoch 22/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.0858 - accuracy: 0.7138 - val_loss: 1.2619 - val_accuracy: 0.6481 - lr: 0.0010\n",
            "Epoch 23/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.0295 - accuracy: 0.7205 - val_loss: 1.3297 - val_accuracy: 0.6152 - lr: 0.0010\n",
            "Epoch 24/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.0045 - accuracy: 0.7324 - val_loss: 1.2460 - val_accuracy: 0.6687 - lr: 0.0010\n",
            "Epoch 25/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.9919 - accuracy: 0.7355 - val_loss: 1.5168 - val_accuracy: 0.5617 - lr: 0.0010\n",
            "Epoch 26/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.9723 - accuracy: 0.7319 - val_loss: 1.4343 - val_accuracy: 0.5782 - lr: 0.0010\n",
            "Epoch 27/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.9000 - accuracy: 0.7540 - val_loss: 1.1920 - val_accuracy: 0.6667 - lr: 0.0010\n",
            "Epoch 28/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8924 - accuracy: 0.7509 - val_loss: 1.2768 - val_accuracy: 0.6440 - lr: 0.0010\n",
            "Epoch 29/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.8069 - accuracy: 0.7782 - val_loss: 1.2512 - val_accuracy: 0.6132 - lr: 0.0010\n",
            "Epoch 30/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8586 - accuracy: 0.7524 - val_loss: 1.3242 - val_accuracy: 0.6173 - lr: 0.0010\n",
            "Epoch 31/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8253 - accuracy: 0.7694 - val_loss: 1.2316 - val_accuracy: 0.6584 - lr: 0.0010\n",
            "Epoch 32/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.7690 - accuracy: 0.7869 - val_loss: 1.5842 - val_accuracy: 0.5247 - lr: 0.0010\n",
            "Epoch 33/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.7571 - accuracy: 0.7931 - val_loss: 1.5018 - val_accuracy: 0.5947 - lr: 0.0010\n",
            "Epoch 34/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.7450 - accuracy: 0.7972 - val_loss: 1.1956 - val_accuracy: 0.6626 - lr: 0.0010\n",
            "Epoch 35/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6323 - accuracy: 0.8317 - val_loss: 1.2610 - val_accuracy: 0.6543 - lr: 0.0010\n",
            "Epoch 36/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.5918 - accuracy: 0.8461 - val_loss: 1.2645 - val_accuracy: 0.6337 - lr: 0.0010\n",
            "Epoch 37/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6142 - accuracy: 0.8317 - val_loss: 1.1157 - val_accuracy: 0.6708 - lr: 0.0010\n",
            "Epoch 38/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6012 - accuracy: 0.8441 - val_loss: 1.2949 - val_accuracy: 0.6255 - lr: 0.0010\n",
            "Epoch 39/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.5864 - accuracy: 0.8538 - val_loss: 1.1007 - val_accuracy: 0.7058 - lr: 0.0010\n",
            "Epoch 40/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.5250 - accuracy: 0.8600 - val_loss: 1.2093 - val_accuracy: 0.6049 - lr: 0.0010\n",
            "Epoch 41/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.5184 - accuracy: 0.8734 - val_loss: 1.3467 - val_accuracy: 0.6687 - lr: 0.0010\n",
            "Epoch 42/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.5010 - accuracy: 0.8806 - val_loss: 1.4531 - val_accuracy: 0.6502 - lr: 0.0010\n",
            "Epoch 43/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.5143 - accuracy: 0.8610 - val_loss: 1.2405 - val_accuracy: 0.6420 - lr: 0.0010\n",
            "Epoch 44/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.4489 - accuracy: 0.8924 - val_loss: 1.3948 - val_accuracy: 0.6461 - lr: 0.0010\n",
            "Epoch 45/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.4616 - accuracy: 0.8873 - val_loss: 1.4566 - val_accuracy: 0.6111 - lr: 0.0010\n",
            "Epoch 46/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.4170 - accuracy: 0.9012 - val_loss: 1.3373 - val_accuracy: 0.6481 - lr: 0.0010\n",
            "Epoch 47/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3839 - accuracy: 0.9125 - val_loss: 1.4762 - val_accuracy: 0.6255 - lr: 0.0010\n",
            "Epoch 48/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.4180 - accuracy: 0.8996 - val_loss: 1.2530 - val_accuracy: 0.6523 - lr: 0.0010\n",
            "Epoch 49/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3598 - accuracy: 0.9259 - val_loss: 1.1625 - val_accuracy: 0.6872 - lr: 0.0010\n",
            "Epoch 50/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3604 - accuracy: 0.9223 - val_loss: 1.4468 - val_accuracy: 0.6296 - lr: 0.0010\n",
            "Epoch 51/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3183 - accuracy: 0.9341 - val_loss: 1.3193 - val_accuracy: 0.6687 - lr: 0.0010\n",
            "Epoch 52/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3033 - accuracy: 0.9454 - val_loss: 1.2853 - val_accuracy: 0.7016 - lr: 0.0010\n",
            "Epoch 53/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.2965 - accuracy: 0.9418 - val_loss: 1.3185 - val_accuracy: 0.6337 - lr: 0.0010\n",
            "Epoch 54/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.2684 - accuracy: 0.9552 - val_loss: 1.2294 - val_accuracy: 0.6831 - lr: 0.0010\n",
            "Epoch 55/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2348 - accuracy: 0.9614 - val_loss: 1.2372 - val_accuracy: 0.6749 - lr: 5.0000e-04\n",
            "Epoch 56/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2027 - accuracy: 0.9743 - val_loss: 1.2930 - val_accuracy: 0.6749 - lr: 5.0000e-04\n",
            "Epoch 57/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1788 - accuracy: 0.9789 - val_loss: 1.4500 - val_accuracy: 0.6914 - lr: 5.0000e-04\n",
            "Epoch 58/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1808 - accuracy: 0.9784 - val_loss: 1.4221 - val_accuracy: 0.6852 - lr: 5.0000e-04\n",
            "Epoch 59/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1708 - accuracy: 0.9804 - val_loss: 1.2946 - val_accuracy: 0.6749 - lr: 5.0000e-04\n",
            "Epoch 60/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1640 - accuracy: 0.9804 - val_loss: 1.4007 - val_accuracy: 0.6872 - lr: 5.0000e-04\n",
            "Epoch 61/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1649 - accuracy: 0.9794 - val_loss: 1.3180 - val_accuracy: 0.6955 - lr: 5.0000e-04\n",
            "Epoch 62/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1438 - accuracy: 0.9876 - val_loss: 1.2976 - val_accuracy: 0.6893 - lr: 5.0000e-04\n",
            "Epoch 63/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1390 - accuracy: 0.9846 - val_loss: 1.3239 - val_accuracy: 0.6996 - lr: 5.0000e-04\n",
            "Epoch 64/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.1285 - accuracy: 0.9907 - val_loss: 1.3736 - val_accuracy: 0.6790 - lr: 5.0000e-04\n",
            "Epoch 65/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1229 - accuracy: 0.9902 - val_loss: 1.3305 - val_accuracy: 0.6893 - lr: 5.0000e-04\n",
            "Epoch 66/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1336 - accuracy: 0.9876 - val_loss: 1.4655 - val_accuracy: 0.6358 - lr: 5.0000e-04\n",
            "Epoch 67/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.1250 - accuracy: 0.9897 - val_loss: 1.3036 - val_accuracy: 0.7058 - lr: 5.0000e-04\n",
            "Epoch 68/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1164 - accuracy: 0.9938 - val_loss: 1.3478 - val_accuracy: 0.6934 - lr: 5.0000e-04\n",
            "Epoch 69/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1088 - accuracy: 0.9949 - val_loss: 1.3651 - val_accuracy: 0.6831 - lr: 5.0000e-04\n"
          ]
        }
      ],
      "source": [
        "history9 = model9.fit(\n",
        "    x = x_train9,\n",
        "    y = y_train9,\n",
        "    batch_size = 64,\n",
        "    epochs = 200,\n",
        "    validation_data = (x_test9, y_test9),\n",
        "    callbacks = [\n",
        "        tfk.callbacks.EarlyStopping(monitor='val_accuracy', mode='max', patience=30, restore_best_weights=True),\n",
        "        tfk.callbacks.ReduceLROnPlateau(monitor='val_accuracy', mode='max', patience=15, factor=0.5, min_lr=1e-5)\n",
        "    ]\n",
        ").history"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 25,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "vyk24WNlwRAB",
        "outputId": "ca9a3b10-abaf-4802-9ebf-6cec18544eea"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1/200\n",
            "31/31 [==============================] - 2s 27ms/step - loss: 3.4703 - accuracy: 0.3551 - val_loss: 3.7514 - val_accuracy: 0.1132 - lr: 0.0010\n",
            "Epoch 2/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 3.0645 - accuracy: 0.4364 - val_loss: 3.6190 - val_accuracy: 0.1173 - lr: 0.0010\n",
            "Epoch 3/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 2.7788 - accuracy: 0.4771 - val_loss: 3.4655 - val_accuracy: 0.1955 - lr: 0.0010\n",
            "Epoch 4/200\n",
            "31/31 [==============================] - 0s 15ms/step - loss: 2.5315 - accuracy: 0.5234 - val_loss: 3.2906 - val_accuracy: 0.2078 - lr: 0.0010\n",
            "Epoch 5/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 2.3853 - accuracy: 0.5234 - val_loss: 3.1369 - val_accuracy: 0.2428 - lr: 0.0010\n",
            "Epoch 6/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 2.2416 - accuracy: 0.5419 - val_loss: 2.8881 - val_accuracy: 0.2305 - lr: 0.0010\n",
            "Epoch 7/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 2.0511 - accuracy: 0.5816 - val_loss: 2.7413 - val_accuracy: 0.4300 - lr: 0.0010\n",
            "Epoch 8/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.9673 - accuracy: 0.5975 - val_loss: 2.4704 - val_accuracy: 0.4753 - lr: 0.0010\n",
            "Epoch 9/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 1.8345 - accuracy: 0.6083 - val_loss: 2.3668 - val_accuracy: 0.4609 - lr: 0.0010\n",
            "Epoch 10/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.7842 - accuracy: 0.6022 - val_loss: 2.1254 - val_accuracy: 0.4753 - lr: 0.0010\n",
            "Epoch 11/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.6618 - accuracy: 0.6341 - val_loss: 2.0007 - val_accuracy: 0.5185 - lr: 0.0010\n",
            "Epoch 12/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.5967 - accuracy: 0.6418 - val_loss: 2.1211 - val_accuracy: 0.4239 - lr: 0.0010\n",
            "Epoch 13/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.5295 - accuracy: 0.6500 - val_loss: 1.9773 - val_accuracy: 0.4918 - lr: 0.0010\n",
            "Epoch 14/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.4370 - accuracy: 0.6572 - val_loss: 1.6914 - val_accuracy: 0.6029 - lr: 0.0010\n",
            "Epoch 15/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.3627 - accuracy: 0.6747 - val_loss: 2.0614 - val_accuracy: 0.3868 - lr: 0.0010\n",
            "Epoch 16/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.3567 - accuracy: 0.6706 - val_loss: 1.6527 - val_accuracy: 0.5885 - lr: 0.0010\n",
            "Epoch 17/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.3045 - accuracy: 0.6727 - val_loss: 1.5816 - val_accuracy: 0.5514 - lr: 0.0010\n",
            "Epoch 18/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.2192 - accuracy: 0.7005 - val_loss: 1.6762 - val_accuracy: 0.5247 - lr: 0.0010\n",
            "Epoch 19/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.1521 - accuracy: 0.7169 - val_loss: 1.8164 - val_accuracy: 0.5165 - lr: 0.0010\n",
            "Epoch 20/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.1615 - accuracy: 0.7056 - val_loss: 1.7040 - val_accuracy: 0.5226 - lr: 0.0010\n",
            "Epoch 21/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.0910 - accuracy: 0.7154 - val_loss: 1.6478 - val_accuracy: 0.5494 - lr: 0.0010\n",
            "Epoch 22/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 1.1060 - accuracy: 0.6989 - val_loss: 1.3987 - val_accuracy: 0.6091 - lr: 0.0010\n",
            "Epoch 23/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 1.0206 - accuracy: 0.7344 - val_loss: 1.4854 - val_accuracy: 0.5988 - lr: 0.0010\n",
            "Epoch 24/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.9695 - accuracy: 0.7349 - val_loss: 1.3796 - val_accuracy: 0.5864 - lr: 0.0010\n",
            "Epoch 25/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.9322 - accuracy: 0.7586 - val_loss: 1.5764 - val_accuracy: 0.5782 - lr: 0.0010\n",
            "Epoch 26/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8848 - accuracy: 0.7586 - val_loss: 1.5389 - val_accuracy: 0.5494 - lr: 0.0010\n",
            "Epoch 27/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.8940 - accuracy: 0.7571 - val_loss: 1.3373 - val_accuracy: 0.6523 - lr: 0.0010\n",
            "Epoch 28/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8411 - accuracy: 0.7679 - val_loss: 1.2498 - val_accuracy: 0.6481 - lr: 0.0010\n",
            "Epoch 29/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8358 - accuracy: 0.7710 - val_loss: 1.3110 - val_accuracy: 0.6543 - lr: 0.0010\n",
            "Epoch 30/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.8118 - accuracy: 0.7746 - val_loss: 1.2908 - val_accuracy: 0.6646 - lr: 0.0010\n",
            "Epoch 31/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.7427 - accuracy: 0.8034 - val_loss: 1.2935 - val_accuracy: 0.6358 - lr: 0.0010\n",
            "Epoch 32/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.7527 - accuracy: 0.7895 - val_loss: 1.3986 - val_accuracy: 0.6132 - lr: 0.0010\n",
            "Epoch 33/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6953 - accuracy: 0.8096 - val_loss: 1.2225 - val_accuracy: 0.6893 - lr: 0.0010\n",
            "Epoch 34/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6432 - accuracy: 0.8307 - val_loss: 1.4181 - val_accuracy: 0.5823 - lr: 0.0010\n",
            "Epoch 35/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6167 - accuracy: 0.8389 - val_loss: 1.5242 - val_accuracy: 0.6276 - lr: 0.0010\n",
            "Epoch 36/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6355 - accuracy: 0.8260 - val_loss: 1.2362 - val_accuracy: 0.6564 - lr: 0.0010\n",
            "Epoch 37/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.6030 - accuracy: 0.8353 - val_loss: 1.3372 - val_accuracy: 0.6420 - lr: 0.0010\n",
            "Epoch 38/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.5904 - accuracy: 0.8399 - val_loss: 1.4032 - val_accuracy: 0.6543 - lr: 0.0010\n",
            "Epoch 39/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.6034 - accuracy: 0.8379 - val_loss: 1.3776 - val_accuracy: 0.6070 - lr: 0.0010\n",
            "Epoch 40/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.5757 - accuracy: 0.8564 - val_loss: 1.2228 - val_accuracy: 0.6749 - lr: 0.0010\n",
            "Epoch 41/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.5397 - accuracy: 0.8600 - val_loss: 1.2951 - val_accuracy: 0.6708 - lr: 0.0010\n",
            "Epoch 42/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.5787 - accuracy: 0.8564 - val_loss: 1.3460 - val_accuracy: 0.6523 - lr: 0.0010\n",
            "Epoch 43/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.4703 - accuracy: 0.8868 - val_loss: 1.3654 - val_accuracy: 0.6584 - lr: 0.0010\n",
            "Epoch 44/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.4794 - accuracy: 0.8749 - val_loss: 1.2922 - val_accuracy: 0.6687 - lr: 0.0010\n",
            "Epoch 45/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.4194 - accuracy: 0.8976 - val_loss: 1.2952 - val_accuracy: 0.6852 - lr: 0.0010\n",
            "Epoch 46/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.4303 - accuracy: 0.8960 - val_loss: 1.3048 - val_accuracy: 0.6584 - lr: 0.0010\n",
            "Epoch 47/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.3924 - accuracy: 0.9043 - val_loss: 1.3570 - val_accuracy: 0.6502 - lr: 0.0010\n",
            "Epoch 48/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.3689 - accuracy: 0.9084 - val_loss: 1.3986 - val_accuracy: 0.6708 - lr: 0.0010\n",
            "Epoch 49/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.3041 - accuracy: 0.9388 - val_loss: 1.2882 - val_accuracy: 0.6770 - lr: 5.0000e-04\n",
            "Epoch 50/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2876 - accuracy: 0.9460 - val_loss: 1.2083 - val_accuracy: 0.6934 - lr: 5.0000e-04\n",
            "Epoch 51/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.2410 - accuracy: 0.9614 - val_loss: 1.3548 - val_accuracy: 0.6893 - lr: 5.0000e-04\n",
            "Epoch 52/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2377 - accuracy: 0.9650 - val_loss: 1.3218 - val_accuracy: 0.6872 - lr: 5.0000e-04\n",
            "Epoch 53/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.2262 - accuracy: 0.9640 - val_loss: 1.4329 - val_accuracy: 0.6646 - lr: 5.0000e-04\n",
            "Epoch 54/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.2283 - accuracy: 0.9650 - val_loss: 1.2721 - val_accuracy: 0.6914 - lr: 5.0000e-04\n",
            "Epoch 55/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.2048 - accuracy: 0.9743 - val_loss: 1.5943 - val_accuracy: 0.6481 - lr: 5.0000e-04\n",
            "Epoch 56/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.2017 - accuracy: 0.9701 - val_loss: 1.3265 - val_accuracy: 0.6975 - lr: 5.0000e-04\n",
            "Epoch 57/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1944 - accuracy: 0.9717 - val_loss: 1.2977 - val_accuracy: 0.7099 - lr: 5.0000e-04\n",
            "Epoch 58/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1854 - accuracy: 0.9727 - val_loss: 1.2978 - val_accuracy: 0.6872 - lr: 5.0000e-04\n",
            "Epoch 59/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1859 - accuracy: 0.9722 - val_loss: 1.3411 - val_accuracy: 0.6975 - lr: 5.0000e-04\n",
            "Epoch 60/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1946 - accuracy: 0.9712 - val_loss: 1.4653 - val_accuracy: 0.6914 - lr: 5.0000e-04\n",
            "Epoch 61/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1987 - accuracy: 0.9681 - val_loss: 1.4891 - val_accuracy: 0.6790 - lr: 5.0000e-04\n",
            "Epoch 62/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.1826 - accuracy: 0.9763 - val_loss: 1.4452 - val_accuracy: 0.6955 - lr: 5.0000e-04\n",
            "Epoch 63/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1692 - accuracy: 0.9794 - val_loss: 1.4087 - val_accuracy: 0.7037 - lr: 5.0000e-04\n",
            "Epoch 64/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1520 - accuracy: 0.9815 - val_loss: 1.5045 - val_accuracy: 0.6749 - lr: 5.0000e-04\n",
            "Epoch 65/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1552 - accuracy: 0.9810 - val_loss: 1.4225 - val_accuracy: 0.6975 - lr: 5.0000e-04\n",
            "Epoch 66/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1485 - accuracy: 0.9846 - val_loss: 1.4536 - val_accuracy: 0.7016 - lr: 5.0000e-04\n",
            "Epoch 67/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1530 - accuracy: 0.9810 - val_loss: 1.6622 - val_accuracy: 0.6523 - lr: 5.0000e-04\n",
            "Epoch 68/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1678 - accuracy: 0.9743 - val_loss: 1.4990 - val_accuracy: 0.6728 - lr: 5.0000e-04\n",
            "Epoch 69/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1527 - accuracy: 0.9789 - val_loss: 1.6087 - val_accuracy: 0.6605 - lr: 5.0000e-04\n",
            "Epoch 70/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.1396 - accuracy: 0.9851 - val_loss: 1.5546 - val_accuracy: 0.6770 - lr: 5.0000e-04\n",
            "Epoch 71/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1491 - accuracy: 0.9799 - val_loss: 1.6103 - val_accuracy: 0.6749 - lr: 5.0000e-04\n",
            "Epoch 72/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.1476 - accuracy: 0.9835 - val_loss: 1.4415 - val_accuracy: 0.6996 - lr: 5.0000e-04\n",
            "Epoch 73/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.1169 - accuracy: 0.9918 - val_loss: 1.5046 - val_accuracy: 0.6914 - lr: 2.5000e-04\n",
            "Epoch 74/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1034 - accuracy: 0.9979 - val_loss: 1.5083 - val_accuracy: 0.6914 - lr: 2.5000e-04\n",
            "Epoch 75/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1005 - accuracy: 0.9964 - val_loss: 1.4930 - val_accuracy: 0.6955 - lr: 2.5000e-04\n",
            "Epoch 76/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.1002 - accuracy: 0.9964 - val_loss: 1.5078 - val_accuracy: 0.6934 - lr: 2.5000e-04\n",
            "Epoch 77/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1027 - accuracy: 0.9938 - val_loss: 1.5551 - val_accuracy: 0.6955 - lr: 2.5000e-04\n",
            "Epoch 78/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0963 - accuracy: 0.9969 - val_loss: 1.5214 - val_accuracy: 0.6975 - lr: 2.5000e-04\n",
            "Epoch 79/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0928 - accuracy: 0.9985 - val_loss: 1.5813 - val_accuracy: 0.6893 - lr: 2.5000e-04\n",
            "Epoch 80/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0950 - accuracy: 0.9974 - val_loss: 1.5478 - val_accuracy: 0.7037 - lr: 2.5000e-04\n",
            "Epoch 81/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.1007 - accuracy: 0.9938 - val_loss: 1.4096 - val_accuracy: 0.7140 - lr: 2.5000e-04\n",
            "Epoch 82/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0936 - accuracy: 0.9964 - val_loss: 1.5048 - val_accuracy: 0.6872 - lr: 2.5000e-04\n",
            "Epoch 83/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0966 - accuracy: 0.9949 - val_loss: 1.6009 - val_accuracy: 0.6831 - lr: 2.5000e-04\n",
            "Epoch 84/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0869 - accuracy: 0.9985 - val_loss: 1.5128 - val_accuracy: 0.6975 - lr: 2.5000e-04\n",
            "Epoch 85/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0880 - accuracy: 0.9974 - val_loss: 1.4662 - val_accuracy: 0.6955 - lr: 2.5000e-04\n",
            "Epoch 86/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0860 - accuracy: 0.9985 - val_loss: 1.5014 - val_accuracy: 0.6955 - lr: 2.5000e-04\n",
            "Epoch 87/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0819 - accuracy: 0.9995 - val_loss: 1.4640 - val_accuracy: 0.7058 - lr: 2.5000e-04\n",
            "Epoch 88/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0847 - accuracy: 0.9969 - val_loss: 1.5218 - val_accuracy: 0.6934 - lr: 2.5000e-04\n",
            "Epoch 89/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0806 - accuracy: 0.9985 - val_loss: 1.5289 - val_accuracy: 0.7058 - lr: 2.5000e-04\n",
            "Epoch 90/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0795 - accuracy: 0.9985 - val_loss: 1.5276 - val_accuracy: 0.6893 - lr: 2.5000e-04\n",
            "Epoch 91/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0793 - accuracy: 1.0000 - val_loss: 1.5620 - val_accuracy: 0.6955 - lr: 2.5000e-04\n",
            "Epoch 92/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0762 - accuracy: 1.0000 - val_loss: 1.4609 - val_accuracy: 0.7016 - lr: 2.5000e-04\n",
            "Epoch 93/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0747 - accuracy: 0.9995 - val_loss: 1.5192 - val_accuracy: 0.6893 - lr: 2.5000e-04\n",
            "Epoch 94/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0756 - accuracy: 0.9985 - val_loss: 1.5015 - val_accuracy: 0.6872 - lr: 2.5000e-04\n",
            "Epoch 95/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0771 - accuracy: 0.9979 - val_loss: 1.4348 - val_accuracy: 0.7078 - lr: 2.5000e-04\n",
            "Epoch 96/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0746 - accuracy: 0.9995 - val_loss: 1.4388 - val_accuracy: 0.7037 - lr: 2.5000e-04\n",
            "Epoch 97/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0739 - accuracy: 0.9995 - val_loss: 1.4160 - val_accuracy: 0.7078 - lr: 1.2500e-04\n",
            "Epoch 98/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0724 - accuracy: 0.9995 - val_loss: 1.4120 - val_accuracy: 0.7160 - lr: 1.2500e-04\n",
            "Epoch 99/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0711 - accuracy: 1.0000 - val_loss: 1.4376 - val_accuracy: 0.7078 - lr: 1.2500e-04\n",
            "Epoch 100/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0704 - accuracy: 0.9990 - val_loss: 1.4454 - val_accuracy: 0.7140 - lr: 1.2500e-04\n",
            "Epoch 101/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0695 - accuracy: 0.9995 - val_loss: 1.4333 - val_accuracy: 0.7078 - lr: 1.2500e-04\n",
            "Epoch 102/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0689 - accuracy: 0.9990 - val_loss: 1.4243 - val_accuracy: 0.7058 - lr: 1.2500e-04\n",
            "Epoch 103/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0672 - accuracy: 1.0000 - val_loss: 1.4221 - val_accuracy: 0.7119 - lr: 1.2500e-04\n",
            "Epoch 104/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0704 - accuracy: 0.9990 - val_loss: 1.4501 - val_accuracy: 0.7016 - lr: 1.2500e-04\n",
            "Epoch 105/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0709 - accuracy: 0.9990 - val_loss: 1.4751 - val_accuracy: 0.7016 - lr: 1.2500e-04\n",
            "Epoch 106/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0679 - accuracy: 0.9995 - val_loss: 1.4720 - val_accuracy: 0.7016 - lr: 1.2500e-04\n",
            "Epoch 107/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0663 - accuracy: 0.9995 - val_loss: 1.4585 - val_accuracy: 0.7037 - lr: 1.2500e-04\n",
            "Epoch 108/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0679 - accuracy: 0.9990 - val_loss: 1.4188 - val_accuracy: 0.6996 - lr: 1.2500e-04\n",
            "Epoch 109/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0664 - accuracy: 0.9995 - val_loss: 1.4973 - val_accuracy: 0.6934 - lr: 1.2500e-04\n",
            "Epoch 110/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0646 - accuracy: 1.0000 - val_loss: 1.4353 - val_accuracy: 0.7058 - lr: 1.2500e-04\n",
            "Epoch 111/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0646 - accuracy: 1.0000 - val_loss: 1.4418 - val_accuracy: 0.7058 - lr: 1.2500e-04\n",
            "Epoch 112/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0638 - accuracy: 0.9995 - val_loss: 1.4333 - val_accuracy: 0.6996 - lr: 1.2500e-04\n",
            "Epoch 113/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0631 - accuracy: 0.9995 - val_loss: 1.4500 - val_accuracy: 0.7037 - lr: 1.2500e-04\n",
            "Epoch 114/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0653 - accuracy: 0.9990 - val_loss: 1.4120 - val_accuracy: 0.7037 - lr: 6.2500e-05\n",
            "Epoch 115/200\n",
            "31/31 [==============================] - 0s 12ms/step - loss: 0.0637 - accuracy: 0.9995 - val_loss: 1.4103 - val_accuracy: 0.7058 - lr: 6.2500e-05\n",
            "Epoch 116/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0638 - accuracy: 0.9995 - val_loss: 1.4342 - val_accuracy: 0.7037 - lr: 6.2500e-05\n",
            "Epoch 117/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0635 - accuracy: 0.9995 - val_loss: 1.4376 - val_accuracy: 0.7037 - lr: 6.2500e-05\n",
            "Epoch 118/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0625 - accuracy: 1.0000 - val_loss: 1.4144 - val_accuracy: 0.7099 - lr: 6.2500e-05\n",
            "Epoch 119/200\n",
            "31/31 [==============================] - 0s 14ms/step - loss: 0.0615 - accuracy: 1.0000 - val_loss: 1.4248 - val_accuracy: 0.7099 - lr: 6.2500e-05\n",
            "Epoch 120/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0619 - accuracy: 0.9995 - val_loss: 1.4129 - val_accuracy: 0.7058 - lr: 6.2500e-05\n",
            "Epoch 121/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0630 - accuracy: 0.9995 - val_loss: 1.4354 - val_accuracy: 0.6955 - lr: 6.2500e-05\n",
            "Epoch 122/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0605 - accuracy: 1.0000 - val_loss: 1.4321 - val_accuracy: 0.7016 - lr: 6.2500e-05\n",
            "Epoch 123/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0645 - accuracy: 0.9985 - val_loss: 1.4133 - val_accuracy: 0.7099 - lr: 6.2500e-05\n",
            "Epoch 124/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0638 - accuracy: 0.9985 - val_loss: 1.4370 - val_accuracy: 0.7016 - lr: 6.2500e-05\n",
            "Epoch 125/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0617 - accuracy: 1.0000 - val_loss: 1.4483 - val_accuracy: 0.7016 - lr: 6.2500e-05\n",
            "Epoch 126/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0616 - accuracy: 1.0000 - val_loss: 1.4308 - val_accuracy: 0.6975 - lr: 6.2500e-05\n",
            "Epoch 127/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0602 - accuracy: 1.0000 - val_loss: 1.4375 - val_accuracy: 0.7037 - lr: 6.2500e-05\n",
            "Epoch 128/200\n",
            "31/31 [==============================] - 0s 13ms/step - loss: 0.0593 - accuracy: 1.0000 - val_loss: 1.4466 - val_accuracy: 0.7058 - lr: 6.2500e-05\n"
          ]
        }
      ],
      "source": [
        "history10 = model10.fit(\n",
        "    x = x_train10,\n",
        "    y = y_train10,\n",
        "    batch_size = 64,\n",
        "    epochs = 200,\n",
        "    validation_data = (x_test10, y_test10),\n",
        "    callbacks = [\n",
        "        tfk.callbacks.EarlyStopping(monitor='val_accuracy', mode='max', patience=30, restore_best_weights=True),\n",
        "        tfk.callbacks.ReduceLROnPlateau(monitor='val_accuracy', mode='max', patience=15, factor=0.5, min_lr=1e-5)\n",
        "    ]\n",
        ").history"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4g67kGedwESd"
      },
      "source": [
        "# evaluation"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 26,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 764
        },
        "id": "rzrqDHgijtwW",
        "outputId": "f49f0cae-6511-46fd-d47e-315ba0f369e1"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 1224x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 1224x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAABEMAAADHCAYAAADyMp0/AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3deXxU9dn//9eVPYEECAmJEDEsbizWW9GqKHUBweWL3sWlCioq1L219m6VisoiS6u3+qtAi0tFxSrua0VpVbSCrSDiDUUsCgKWLRACYcn6+f1xZuJkmCQDJHOSzPv5eMzjkHOuz5nrDJ8HJFc+iznnEBERERERERGJFwl+JyAiIiIiIiIiEksqhoiIiIiIiIhIXFExRERERERERETiioohIiIiIiIiIhJXVAwRERERERERkbiiYoiIiIiIiIiIxJUkvxNo6XJyclxhYaHfaey3yspKkpL01y+xof4msaT+JrGk/iax1Fr7296KSgDSklvfs7VkrbW/SfPUlP1t8eLFRc653PDz6t0HqbCwkEWLFvmdxn4rKioiJyfH7zQkTqi/SSypv0ksqb9JLLXW/vbVpi0AHJG3z88q4qPW2t+keWrK/mZm30Y6H9U0GTM71MxeNLMSM9thZi+bWdco26aZ2X1mtsHM9pjZQjMbECEuwczGmNkaM9trZkvNbFiEuKvM7CUz+9bMnJnNque9TzWzBYH33WhmD5hZeoS43mb2rpmVmtlWM3vCzLKjeT4RERERERERaVkaLIaYWQbwHnAUcBVwBXA48L6ZtYniPR4HRgN3A+cDG4B3zOzYsLiJwDhgGnAO8AnwgpmdGxY3AugBzAN21JP3MYGYzYH3HQtcDcwKi+sMfACkAxcBNwEDgTfNTGuqiIiIiIiIiLQy0UyTGQ10B450zq0CMLMvgH8D1wEP1NXQzH4AXA5c45x7InBuPrAcmAAMDZzrBPwPMNU5d3+g+ftm1hOYCvwl5LaDnXPVgXZD6sl7PLAeuNg5VxGILweeNLPfOuc+C8T9CkgG/p9zbnsg7j/AfOBC4OX6Px4RERERETlQmh4jIn6IZuTDUOCTYCEEwDm3GvgYuCCKthXAnJC2lcBzwGAzSw2cHgykALPD2s8G+ppZt5D21Q0lbGbJwBDg+WAhJOB5oDws76HAW8FCSOA9PgTWRvF8IiIiIiIiItLCRDMypDfwWoTzy4GLo2i72jm3O0LbFKBn4M+9gTJgVYQ4gF7A6ihyDeoBpAHLQk865/aa2deB+xFYP6Qb8FiEeywPxrU6C6aRvn0LZGT4nYnEifTdu9XfJGbU3ySW1N/qkJENJ4wCM78zkRZg046dAORlZfqciYjEk2iKIdlAcYTz24AOB9E2eD143O6ccw3ERSsYX9d7B693AKyeuCMj3dzMfgr8FKCgoICioqL9TM9fbb/9FEqLKNdWWRIrlZXqbxI76m8SS+pv+6rci5WXsiOvP65NJ7+zaVVKSkr8TqFJrN3qDdBOLC/zORMJ1Vr7mzRPfvQ3/e99AJxzjwCPAPTr18+1uC2nLnuSoqIi2rS0vKXF2qH+JjGk/iaxpP4WweoP4d276NgmGfTZNLoW931nFLZVeb8PbY3P1tLl5OSwY8cONm/eTEVFRcMNRA5QVVUV5eXlUccnJSWRlpZGbm4uaWlpB/Se0RRDiok8AqSuUR/hbQ+roy18P/KjGGhvZhY2OiQ8LlrBvOrKOzj9Zjvg6onb3/cVERERiW+pWd6xrM5N/0SkhdixYwebNm2iS5cupKenY5r6Jk2koqKC5OTkqGKdc1RWVlJaWsratWvJy8ujXbt2+/2e0SygGlzTI1wv4F9RtO0W2J43vG05368RshxIxVvrIzyOKN4n3Nd4a5DUytvM0vB2xvkXQGAtkzXhcSHvvb/vKyIiIhLfUgPrPpTt9DcPETlomzdvpkuXLmRkZKgQIs2GmZGcnEyHDh0oKChg69atB3SfaIohrwMnmVn3kDcvBPoHrtXnDbxta2sWWjWzJOBS4F3nXHBi4Fy8XWeGh7UfASwL7F4TNedceeCelwTeL+givKJLaN6vA+eZWU0pycxOxRvR0tDziYiIiEiompEhKoZIdBLMSNAP2s1SRUUF6enpfqchUqf09HTKyg5svaFopsk8CtwMvGZmY/GmlUwE1gEzg0FmdhjeiIwJzrkJAM65JWY2B3gosN3tauAGvB1cagofzrnNZvYAMMbMdgKf4RVMzsTb+paQ9+nF9yNG0oHDzOyiwNfznXNbAn8eB3wCPG9m04FC4D7gRefc4pBb3odXdHndzKYA7YDfAf8AXoni8xERERGRoNS23lHFEIlSz05aK6Q504gQac4Opn82WAxxzu0yszOBB4Gn8XZf+Rtwq3OuNDQPIJF9R5tcDUwC7gXaA0uBIc65z8Li7gRKgZ8D+cBK4BLn3JthcZcA94R8fXrgBXAG8EEg78/N7Gzgt8BbQAnwFPCbsOf7zszOAB4AXsKbvvMa8EvnXHUdH4uIiIiIRJKUBonJKoaIiEizFtVuMs65tcCwBmLW4BVEws/vAW4LvOprX4VXMLm3gbhxeKM+GuSc+xA4OYq4/wMGRXNPEREREamHGaS01QKqErUN272+ckj7LJ8zEZF4Es2aISIiIiIi0UvL0sgQidrOsjJ2HuCcf5H9MWvWLMyMVatWNRzcTARzXrNmTUzfd82aNZhZzSsxMZH8/HyGDx/OunXrDuies2bN4k9/+lMjZ3rgohoZIiIiIiIStZRMFUNERBrBeeedx8KFCznkkEN8ef8xY8YwdOhQysvL+eSTTxg/fjwrVqzgH//4R9Rb4QbNmjWLyspKrrnmmibKdv+oGCIiIiIijSstC3ZtaThORCTOVFRUkJSUFPXCn7m5ueTm5jZxVnXr3r07J510EgADBgygoqKCsWPHsnjx4przLZWmyYiIiIhI40ppq5EhItJiPfLII/zgBz8gLS2NnJwcrr32WrZt21YrZtq0aZx88slkZ2fTvn17TjrpJN56661aMcGpJjNmzODXv/41nTt3JjU1le3btzNy5EgKCgpYsmQJp512GhkZGRx++OH88Y9/rHWPSNNkCgsLGTFiBM899xxHH300bdq0oV+/fvz973/f51keeughCgsLSUtL48QTT2TBggUUFhYycuTIA/psjjvuOADWrl1bc27VqlVcccUVdOvWjfT0dLp3784NN9xAcXFxTczpp5/O/Pnz+fjjj2um3px++uk111evXs3w4cPJzc0lNTWVY489lldeadrNXTUyREREREQal9YMkf2QlKDfz0rzcccdd/C///u//OxnP+O+++7ju+++Y+zYsSxbtowFCxaQmJgIeIWOUaNGUVhYSGVlJW+88Qbnn38+b7/9NkOGDKl1z0mTJnHCCSfwyCOPUFVVRVpaGgA7duzg8ssv59Zbb+Xuu+/miSee4IYbbuDII4/kjDPOqDfPjz76iJUrVzJx4kTS0tK46667OP/881mzZg3t27cH4LHHHuMXv/gF1157LRdffDFff/01l19+Odu3bz/gzydYlOnRo0fNuf/85z8ceuihPPTQQ3To0IFvvvmGyZMnc+6557Jw4UIAZsyYwYgRI6iqqmLmzJkAZGV5iyavW7eOU089lU6dOvHggw+Sm5vLnDlzGDZsGK+++ipDhw494Hzro2KIiIiIiDSulLZQvguqqyAh0e9spJnrntvR7xRkfyx4GIr+7W8OOYfDKbc0+m3XrFnDfffdxz333MPdd99dc/6II47g1FNP5Y033uDCCy8E4P7776+5Xl1dzVlnncVXX33FH/7wh32KIXl5ebzyyiv7TI3ZuXMnM2bMqCl8DBgwgHfeeYdnn322wWLIjh07+Pzzz+nQoQMA+fn5nHDCCfzlL3/h8ssvp7q6mvHjx3POOefw2GOP1bTLz89n2LB6N4qtpbq6msrKypo1QyZNmsSwYcM4/vjja2IGDBjAgAEDar4+5ZRT6NmzJ6eddhpLlizhv/7rv+jVqxdZWVlUVlbuM71m3LhxOOeYP38+HTt6/x4MHjyYdevWcffddzdZMURlWBERERFpXGmBLVI1OkREWpB58+ZRXV3N8OHDqaysrHn98Ic/JDMzkw8//LAmdvHixZx//vnk5eWRlJREcnIy8+bNY+XKlfvc98ILL4y4RkhGRkatokdqaipHHHFErSkodTn55JNrCiEAffv2Bb6fvrJ+/XrWr1/PxRdfXKvdBRdcQFJS9GMirrvuOpKTk2nTpg1nnXUWeXl5zJ49u1ZMeXk5kydP5qijjiI9PZ3k5GROO+00gIifR7i5c+cyZMgQ2rVrV+tzHzx4MEuXLmXHjqbZql0jQ0RERESkcaVkeseynZDe3t9cpNn7rrgEgC4d2vmciUSlCUZkNBebN28GoGfPnhGvb926FfCmdZx11ln06tWLhx9+mK5du5KUlMRdd93FihUr9mlX104wocWMoNTUVPbu3dtgrtnZ2fu0A2rabtiwAYBOnTrViktMTCQnJ6fB+weNHTuWCy64gN27d/PCCy8wbdo0brzxxlpb5I4ZM4aHH36Yu+++m1NOOYXMzEzWr1/Pj3/846ieZfPmzcyePXufIkvQ1q1ba6bUNCYVQ0RERESkcWlkiOyHXeXlfqcgAlAzRePdd9+NWKgIXp87dy4lJSU8//zzFBQU1FzfvXt3xPtGu3NMYwoWYIIFnqCqqiqKioqivs9hhx1Gv379AG86zM6dO3niiSe4/vrrOfHEEwF47rnnuPLKKxk7dmxNu9LS0qjfo2PHjvTv358xY8ZEvN65c+eo77U/VAwRERERkcaV0tY7qhgiIi3IoEGDSEhIYO3atQwaNKjOuGDRIzk5uebcV199xccff1yrOOKngoICCgoKeOGFF7j66qtrzr/66qtUVlYe8H2nTp3K888/z/jx42t2z9m9e3etzwLgiSee2KdtamoqO3fu+//CkCFDWLBgAb179yY9Pf2Ac9tfKoaIiIiISOPSyBARacbmzp1Lfn5+rXPt2rVj0KBB3H777dx8882sXLmSH/3oR6SlpbFu3TrmzZvHqFGjOOOMMxg4cCBJSUlceeWV/PKXv2TDhg3cc889dO3alerqap+eqraEhATuueceRo8ezahRo7j44ov55ptvmDp1Ku3atSPhAHdxys/P56abbuL+++9n8eLFHH/88QwZMoQnn3ySvn370rNnT15++WUWLFiwT9tevXoxY8YM5syZQ48ePcjMzOTII49kwoQJnHjiiQwYMICbb76ZwsJCiouLWbZsGd98802tKTmNScUQEREREWlcNSNDmmbROxGRg3HLLfuue9K7d2+WLVvG5MmTOfroo5k+fTrTp0/HzDj00EM566yzOPzww2tin3nmmZqdTnr06MHUqVOZO3cuH3zwQYyfpm6jRo2itLSUBx98kNmzZ9OnTx9mz57N0KFDadfuwNfouf3225k5cyYTJkzgtdde4+GHH8Y5x5133gnAueeey7PPPlszjSa03cqVK2vy+tGPfsQHH3xA165dWbhwIZMmTeI3v/kNW7ZsoWPHjvTp04errrrqoD6D+phzrsluHg/69evnFi1a5Hca+62oqGi/Fs4RORjqbxJL6m8SS+pvdaiqhMfOgn7XwPFN941svGmt/W110TYAuuVkNxApsVRUVMSWLVs4+uij/U5FGtGiRYs44YQTeOqpp7jiiiv8TqdGRUXFPlNtorVixYp6+6mZLXbO9Qs/r5EhIiIiItK4EpMgOUPTZCQqKoKINI3Vq1czffp0TjvtNLKyslixYgWTJ0+mW7duDBs2zO/0fKdiiIiIiIg0vtRMFUNERHyUnp7OsmXLeOqppyguLqZDhw4MHDiQqVOnkpGR4Xd6vlMxREREREQan4ohEqX127YDUJDd3udMRFqX/Px85s6d63cazZaKISIiIiLS+FIztYCqRGV3RYXfKYhIHDqw/XREREREROqjkSEiItKMqRgiIiIiIo0vNUvFEBERabZUDBERERGRxqeRISIi0oxpzRARERERaXypmVBVDpVlkJTqdzbSjKUm6UcSEYk9/csjIiIiIo0vNdM77t0BbXP9zUWatcM6dvA7BRGJQ5omIyIiIiKNL1gMKddUGRERaX5UDBERERGRxpea5R33antdqd+3W4v5dmux32lIHJg1axZmVvNKTEykS5cuXHLJJaxcubJJ3vODDz5g3LhxVFdXN8n9YyX42a1atcrvVBqNiiEiIiIi0vhqRoaU+puHNHtllZWUVVb6nYbEkRdeeIGFCxfy4YcfMmXKFJYsWcJZZ51FSUlJo7/XBx98wPjx41t8MaQ10pohIiIiItL4QtcMERFpRo499lh69uwJQP/+/encuTODBg1iwYIFnHPOOT5nJ7GikSEiIiIi0viCxRBtrysizVxWljetr6Kiotb5pUuXMnToUDp06EB6ejr9+/fno48+qhXz6aefMmjQIDp27Eh6ejrdu3fnxhtvBGDcuHGMHz8egOTk5JrpOfWprKxkypQpHHXUUaSmptK5c2d++ctfsnfv3pqYNWvWYGbMmDGD2267jU6dOpGRkcH555/PmjVrat2voqKCsWPHUlhYSEpKCoWFhYwdO3afZ921axd33HEHPXr0IDU1lfz8fIYNG8amTZtqxRUVFTF8+HCysrLo3LkzP/vZz2rl1pJEVQwxs0PN7EUzKzGzHWb2spl1jbJtmpndZ2YbzGyPmS00swER4hLMbIyZrTGzvWa21MyG1XHP0Wb2pZmVmdlKM7s+7Hqhmbl6Xj8JiR1XR8yr0TyfiIiIiESQ0gYsAco0MkREmpeqqioqKyspKytjxYoV/OY3v6FTp06cfvrpNTGfffYZp5xyCtu2bePRRx/lpZdeomPHjgwcOJDFixcDUFpayuDBg0lMTGTWrFm8/fbb3H333VQGpn2NGjWKa6+9FoC///3vLFy4kIULF9ab24gRI7j33nu5/PLLeeuttxgzZgyPP/44w4cP3yd2ypQp/Pvf/+aJJ55g+vTpLF68mLPPPrtWoeOqq65i6tSpXHnllbz55puMHDmS3/72t1x11VU1MeXl5QwaNIiHH36YkSNH8uabbzJt2jSys7MpLq69ns8VV1xBjx49ePnll7nhhhuYPn06U6ZM2b+/gGaiwWkyZpYBvAeUAVcBDrgXeN/MjnHO7WrgFo8D5wG/Ar4BbgLeMbOTnXOfh8RNBP4HuBNYDPwEeMHMznfO/SUkn9HATGAK8FfgLGCGmZlz7g+BsA3AyRFyuRc4FXgnwrVTgaqQr7c18FwiIiIiUhczSG2rkSHSoIzkZL9TkP301aYt+5zrkJFObmZbqqurWbVl6z7XO7bJoGPbNlRWVfFN0b4/auW2bUOHNhmUV1ayJsKCunmZbWmXkc7eigrSDrLPHHXUUbW+7ty5M2+++WbNCBGAX/3qV3Tt2pX33nuPlJQUAAYPHkyfPn2YOHEir776Kl9++SXFxcX87ne/45hjjqlpO3LkSAAKCgooKCgA4Ic//CFJSfX/+P3RRx8xZ84cnnzySa688koABg4cSHZ2NiNGjODzzz/n2GOPrYnPzMzktddeIyHBG+NwxBFHcOqpp/LUU09x7bXXsmzZMp599lnuuecexo0bB8DZZ59NUlISd911F3fccQfHHHMMs2fPZuHChbz22msMHTq05v4XXXTRPjlefvnlNaNdBg4cyD/+8Q+effbZmnMtSTQjQ0YD3YELnXOvOudeA4YChwHX1dfQzH4AXA78wjn3qHPub8AlwFpgQkhcJ7xCyFTn3P3Oufedc9cB7wNTQ+KSgEnA0865OwNxY4FZwEQzSwZwzpU55z4JfQFfACcCbzjnIi1X/Y+wNl9F8dmIiIiISF1Ss1QMkQYVZLenILu932lIHHnllVf49NNP+ec//8mrr75Kr169OPfcc1mxYgUAe/bsYf78+Vx88cUkJCRQWVlJZWUlzjkGDhzIhx9+CMDhhx9O+/btue6665g9ezbr1q07qLzmzp1LSkoKF110Uc17VlZWcvbZZwPUvG/QRRddVFMIAW/9k4KCgprRJ8H4ESNG1GoX/Hr+/PkAvPvuu+Tn59cqhNTlvPPOq/V13759Wbt27f48ZrMRzQKqQ4FPnHM1e+g451ab2cfABcADDbStAOaEtK00s+eAO8ws1TlXBgwGUoDZYe1nA38ys27OudV4oz1yI8Q9DVyNN7rj/Tpy+TGQCTxZ38OKiIiISCNJzVQxRKQVOiIvt85rCQkJ9V5PSkys93pKUlK91w92VAhAnz59ahZQBW+0xKGHHsq4ceOYM2cO27Zto6qqiokTJzJx4sSI96iurqZdu3a8//77TJw4kRtvvJGdO3fSu3dvxo8fz7BhEVd8qNfmzZspLy+nTZs2Ea9v3Vp7xE1eXt4+MXl5eXz33XcAbNvmjcA55JBDasXk5+fXur5161a6dOkSVY7Z2dm1vk5NTaWsrCyqts1NNMWQ3sBrEc4vBy6Oou1q59zuCG1TgJ6BP/fGm4YTvmnx8sCxF7A6EAewrJ64uoohVwGbgbl1XF8XGKGyHngOGOec21NHrIiIiIg0JDUL9jb+VpXSuqwOTJnolpPdQKRI0wgufPrFF18A0L59exISErjppptqpquEC47IOPbYY3nppZeorKxk0aJFTJkyhUsuuYSlS5fSp0+f/cqjY8eOpKWl7bNIa1Dnzp1rfR2+uGnwXHAqTbBwsXHjRnr06FETs3HjxlrXc3JyWLYs/Efs1i+aaTLZQKRpJduADgfRNng9eNzunHNRxBHhnuFxtZhZF+BM4BnnXPgm5quAO/CKJUOA54FfAK9HupeIiIiIRElrhkgUKqqqqKiqajhQpIns3r2br7/+mtxcb0RKmzZtOO2001i6dCnHHXcc/fr12+cVLikpiZNOOomJEydSXV1dM+UmNTUV8KbeNGTIkCHs3buXkpKSiO8ZXgx58cUXqa6urvn6448/Zv369Zx8srd85oAB3r4lzz33XK12zzzzDEDNgrFnn302Gzdu5I033mgwx9YkmpEhrcEVeIWfWeEXnHPhU27mmdl64CEzG+ic+2t4GzP7KfBT8BbFKSoqavyMm1hJiX5LI7Gj/iaxpP4msaT+Vr/0qkSSS7eyowV+r9Qctdb+VrLde66ixPq3HJXYKikpoaqqap8tWFuyqkDRbdGiRWzcuBHnHBs2bOAPf/gD27Zt44Ybbqh53t/97neceeaZnH322YwcOZJDDjmEoqIilixZQlVVFZMnT+att97iscce44ILLqCwsJBdu3Yxffp0MjMz6devHxUVFRxxxBE19xsyZAiJiYkcf/zxEfPr378/l156KRdddBE///nPOeGEE0hISODbb7/l7bffZvLkyRxxxBE1Oe7cuZOhQ4cyevRoioqKGDt2LD179uSyyy6joqKCI488kksvvZRx48ZRVlbGySefzCeffMLkyZO59NJLOeqoo6ioqODSSy/lkUce4bLLLuPXv/41J554Ijt37mTevHnccsstHHXUUTWfXUVFRa0+EXq+Mf5uDrTtgfxMHk0xpJjII0DqGvUR3vawOtrC9yM6ioH2gR1hXANxBPLZUE9cuCuBz51zXzSQb9CzwEPACXg71tTinHsEeASgX79+LicnJ8rbNi8tNW9pmdTfJJbU3ySW1N/q0SEP1pSRk50NCdEMSJaGtMb+tq3K+/a/NT5bS1deXk5yK9rtJzExEYDLLrus5lxubi59+vRh7ty5DB48uOb8iSeeyKeffsr48eO57bbbKCkpITc3l+OOO47rr7+e5ORkjj76aNq0acPkyZPZsGEDmZmZnHDCCcybN49u3boBcOGFF3LjjTcyc+ZMJk2ahHOOfSdEfO/Pf/4zDz/8MH/605+YOnUqqampFBYWMnjwYAoKCkhOTq75OxkzZgyrVq1i1KhR7Nq1izPOOINp06aRkZFRc7+nnnqKnj178uSTTzJlyhQ6d+7M7bffzj333FNzn+TkZObNm8f48eN5/PHHuffee+nYsSP9+/cnLy+P5OTkms8u9P1DP9PG6CcHeo/ExMQD+vfD6vuLADCz94AU59ypYec/CLT/UT1t7wbGAu1D1w0xs3HAGCDLOVdmZlfiLWx6eOhCrWY2EngC6B5YtHUAMB8YFDpiw8xOx1sr5EznXK01Q8zsBOCfeDvaPFTvw37fphOwCRjjnJtaX2y/fv3cokWLorlts1JUVKT/cCRm1N8kltTfJJbU3xrwxQuwcBqMfNNbTFUOSmvtb8FtWutbNFNir6ioiC1btnD00Uf7nYqEWbNmDd26dePRRx9l1KhRfqfTKCoqKg64GLJixYp6+6mZLXbO7TO3KZoS/evASWbWPeRmhUB/Gl5X4w0gmZCFVgPb414KvBvYSQa8RU0rgOFh7UcAywI7yQAsBIrqiNsGfBwhh6uASuDPDeQaKnj/f+5HGxEREREJldrWO2rdEKlHm5QU2qSk+J2GiMSZaKbJPArcDLxmZmMBB0wE1gEzg0FmdhjwNTDBOTcBwDm3xMzm4K2/kYy3I8wNQDdCChrOuc1m9gAwxsx2Ap/hFUzOxNueNxhXYWZ3ATPM7Du8KSxnAtcAtzjnykMTN7MU4CfA2865zZEezsyWAE8BKwPPNgi4BZjrnHsvis9HRERERCJJzfKOKoZIPbp0aOd3CiIShxoshjjndpnZmcCDwNOAAX8DbnXOlYaEGpDIvqNNrgYmAfcC7YGlwBDn3GdhcXcCpcDPgXy84sQlzrk3w/L5o5k54JfAr4C1wM3OuRkR0j8P6Ig3BacuK/GKPYcEcv8GmAD8rp42IiIiItIQjQwREWl0hYWF9a47ItGJajcZ59xaYFgDMWvwCiLh5/cAtwVe9bWvwiuY3BtFPjMJGZVST9wrkXIKi/lJQ/cRERERkQNQMzJkh795SLP2zZatAHTP7ehzJiIST+Jla10RERERibXgoqkaGSL1qKyu9jsFqYdzDjNteyzN08GMkNEeZyIiIiLSNFQMEWnRkpKSqKys9DsNkTpVVFTUbO+7v1QMEREREZGmkZQKiSkqhoi0UGlpaZSWljYcKOKTHTt2kJl5YFu3qxgiIiIiIk0nLWpjFIgAABsqSURBVEvFEJEWKjc3ly1btrB7924t2CnNhnOO8vJyioqKKC4uJjs7+4DuozVDRERERKTppLTVAqpSr8zUVL9TkDqkpaWRl5fHxo0bKSsr8zsdacWqqqr2a7pLYmIimZmZdO3aldQD/DdExRARERERaToaGSINOKR9lt8pSD3atWtHu3bt/E5DWrmioiJycnJi+p6aJiMiIiIiTSclU8UQERFpdlQMEREREZGmo5Eh0oBVm4tYtbnI7zREJM5omoyIiIiINJ2UtiqGSL2qtTCniPhAI0NEREREpOmkZUHFHqiq8DsTERGRGiqGiIiIiEjTSWnrHTU6REREmhEVQ0RERESk6aQFdgpRMURERJoRrRkiIiIiIk0nJdM7qhgidWiXnuZ3CiISh1QMEREREZGmo5Eh0oC8rEy/UxCROKRpMiIiIiLSdLRmiIiINEMqhoiIiIhI06kZGbLD3zyk2fpq0xa+2rTF7zREJM6oGCIiIiIiTUdrhoiISDOkYoiIiIiINJ2EBG+qjIohIiLSjKgYIiIiIiJNKzVTxRAREWlWVAwRERERkaalYoiIiDQz2lpXRERERJqWiiFSjw4Z6X6nICJxSMUQEREREWlaqZmwS7uFSGS5mW39TkFE4pCmyYiIiIhI09LIEKlHdXU11dXVfqchInFGxRARERERaVqpWV4xxDm/M5FmaNWWrazastXvNEQkzqgYIiIiIiJNKzUTqiuhYo/fmYiIiAAqhoiIiIhIU0vN9I6aKiMiIs2EiiEiIiIi0rRUDBERkWZGxRARERERaVo1xZAd/uYhIiISEFUxxMwONbMXzazEzHaY2ctm1jXKtmlmdp+ZbTCzPWa20MwGRIhLMLMxZrbGzPaa2VIzG1bHPUeb2ZdmVmZmK83s+ggxs8zMRXg9FCH2VDNbEMhvo5k9YGba8FxERESkMaRmeUeNDJEIOrbJoGObDL/TEJE4k9RQgJllAO8BZcBVgAPuBd43s2Occ7sauMXjwHnAr4BvgJuAd8zsZOfc5yFxE4H/Ae4EFgM/AV4ws/Odc38JyWc0MBOYAvwVOAuYYWbmnPtD2HtvAYaGndsQ9nzHAPOAd4DzgW7AfUAX4NIGnk1EREREGqJpMlKPjm3b+J2CiMShBoshwGigO3Ckc24VgJl9AfwbuA54oK6GZvYD4HLgGufcE4Fz84HlwAQChQoz64RXCJnqnLs/0Px9M+sJTAX+EohLAiYBTzvn7gyJ6wxMNLPHnHMVISmUO+c+aeD5xgPrgYuDbc2sHHjSzH7rnPusgfYiIiIiUh8VQ6QelVVVACQlJvqciYjEk2imyQwFPgkWQgCcc6uBj4ELomhbAcwJaVsJPAcMNrPUwOnBQAowO6z9bKCvmXULfH0ykBsh7mmgI3BqFM9Tw8ySgSHA82FFlOeBchp+PhERERFpSHI6JCSqGCIRfVO0jW+KtvmdhojEmWiKIb2BZRHOLwd6RdF2tXNud4S2KUDPkLgyYFWEOELep3fgGJ5PeFxQJzMrMrNKM/vKzG43s9CScw8gLfx+zrm9wNcR7iciIiIi+8vMWzdEC6iKiEgzEc00mWygOML5bUCHg2gbvB48bnfOuSjiiHDP8DiAz/HWHlmOV/D4b7x1Rg4HRjVwv+A9syOcFxEREZH9ldpWI0NERKTZiKYY0iI558J3jfmLmZUCtwbWAvn3gd7bzH4K/BSgoKCAoqKig8jUHyUlJX6nIHFE/U1iSf1NYkn9LXptScWVbGFXC/y+qblorf2tZLv3XEWJ5nMmEqq19jdpnvzob9EUQ4qJPAKkrlEf4W0Pq6MtfD+ioxhoH9gRxjUQRyCfDfXE1eVZ4FagH94CsKH3i5Tj8gjncc49AjwC0K9fP5eTk9PA2zZPLTVvaZnU3ySW1N8kltTfopSVA7u3ka7P66C0xv62rcr79r81PltLp78TiaVY97do1gxZzvdrdYTqBfwrirbdAtvzhrct5/s1QpYDqXhreITHEfI+weJEeD7hcQ0JFly+xlurpNb9zCwNbwedaO8nIiIiIvVJzdI0GYkot20bcrW9rojEWDTFkNeBk8yse/CEmRUC/QPX6vMGkAxcHNI2CbgUeNc5VxY4PRdv15nhYe1HAMsCu9cALASK6ojbhrfDTX2G4xVCPgVwzpUH3vuSQF5BF+EVZxp6PhERERGJhtYMkTp0aJNBhzbhvzsVEWla0UyTeRS4GXjNzMbiFRMmAuuAmcEgMzsMb6TFBOfcBADn3BIzmwM8FNjGdjVwA9CNkIKGc26zmT0AjDGzncBneAWTM/G25w3GVZjZXcAMM/sO+Gsg5hrglkBxI5jL03hb+K7CK2z8NzASmOmc+zrk+cYBnwDPm9l0oBC4D3jRObc4is9HRERERBqSmgXlpVBdDQnR/D5O4kV5ZSUAKUmtdjlDEWmGGvwXxzm3y8zOBB7EKzAY8DfgVudcaUioAYnsO9rkamAScC/QHlgKDHHOfRYWdydQCvwcyAdWApc4594My+ePZuaAXwK/AtYCNzvnZoSE7cQbKXI7kAdUA18CPwNmhN3vczM7G/gt8BZQAjwF/Kahz0ZEREREopSa6R3Ld0JaO39zkWZlzVZvGb8j8nJ9zkRE4klU5Vfn3FpgWAMxa/AKIuHn9wC3BV71ta/CK5jcG0U+MwkZlRLh+jbgwobuExL/IXBytPEiIiIisp9Ss7xjmYohIiLiP41RFBEREZGml9rWO2rdEBERaQZUDBERERGRphc6MkRERMRnKoaIiIiISNOrGRmyw988REREiHLNEBERERGRg6KRIVKHvMy2fqcgInFIxRARERERaXrB3WRUDJEw7TLS/U5BROKQpsmIiIiISNNLTIbkdBVDZB97KyrYW1HhdxoiEmdUDBERERGR2Ehpq2KI7GPttu2s3bbd7zREJM6oGCIiIiIisZGWpWKIiIg0CyqGiIiIiEhspLTVbjIiItIsqBgiIiIiIrGhkSEiItJMqBgiIiIiIrGRkqliiIiINAvaWldEREREYkMjQySC/KxMv1MQkTikYoiIiIiIxEZKW6gsg8pySErxOxtpJrLS0/xOQUTikKbJiIiIiEhspGV5R40OkRC7yyvYXV7hdxoiEmdUDBERERGR2Ehp6x21o4yEWF+8nfXF2/1OQ0TijIohIiIiIhIbae28o0aGiIiIz1QMEREREZHYqBkZomKIiIj4S8UQEREREYkNrRkiIiLNhIohIiIiIhIbGhkiIiLNhLbWFREREZHYSGkLZlpAVWrp3C7L7xREJA6pGCIiIiIisZGQ4BVENDJEQrRNS/U7BRGJQ5omIyIiIiKxk5qlkSFSS+neMkr3lvmdhojEGRVDRERERCR2UjOhrNTvLKQZ+U/JDv5TogKZiMSWiiEiIiIiEjupmRoZIiIivtOaISIiIiISO6mZsOVL+PtDfmfS4qTv3gUZbfxOo/HtDvxIklHpbx5Sy0H3t8MHQV7vxktIpJGpGCIiIiIisdP5WPhuEaz6q9+ZtDjJFRWQnOx3Go2vOsc7JhT5m4fUclD9rWI3lKyD8/63cZMSaUQqhoiIiIhI7PS6wHvJfttRVEROTo7faTS+TVu8Y16uv3lILQfV3z56AP49D6qrvV2kRJoh9UwREREREfFNQYf2FHRo73ca0pjy+nijQ4pX+52JSJ2iKoaY2aFm9qKZlZjZDjN72cy6Rtk2zczuM7MNZrbHzBaa2YAIcQlmNsbM1pjZXjNbambD6rjnaDP70szKzGylmV0fdj3LzO42swVmttXMtgf+fGGEe40zMxfh9Wo0zyciIiIiIgcuIyWZjJRWOP0nnuX39Y4b/8/fPETq0WAxxMwygPeAo4CrgCuAw4H3zSyaFXUeB0YDdwPnAxuAd8zs2LC4icA4YBpwDvAJ8IKZnRuWz2hgJvASMAR4AZhhZjeEhHUFbgTmAyOAS4GvgFfM7KY68jwVODnk9esonk1ERERERA7Cjj172bFnr99pSGPKzIeMbNi03O9MROoUzZoho4HuwJHOuVUAZvYF8G/gOuCBuhqa2Q+Ay4FrnHNPBM7NB5YDE4ChgXOdgP8Bpjrn7g80f9/MegJTgb8E4pKAScDTzrk7Q+I6AxPN7DHnXAWwGujunNsdks47ZnYocDswPUK6/3DOaQlrEREREZEY2rhjJwBZ6Wk+ZyKNxsybKrNpmd+ZiNQpmmkyQ4FPgoUQAOfcauBjoKHVr4YCFcCckLaVwHPAYDNLDZweDKQAs8Pazwb6mlm3wNcnA7kR4p4GOuKN7sA5tyusEBK0COjcQM4iIiIiIiJyMPL7wo7/wO5tfmciElE0xZDeQKSS3nKgVxRtV0coTCzHK370DIkrA1ZFiCPkfYIbVYfnEx5XlwHAl3VcW2dmVWb2rZn91szSG7iXiIiIiIiIRJIX+NFN64ZIMxXNNJlsoDjC+W1Ah4NoG7wePG53zrko4ohwz/C4fZjZT4GT8NYQCbUKuANYAjjgbOAXwHHAoHru9VOAgoICiopa3p7oJSUlfqcgcUT9TWJJ/U1iSf1NYqm19reS7d5zFSWaz5lIqIPvb9m0c0bZN/9gb1bvhsMlrvnx71s0xZAWz8xOB34PPOWceyb0mnMufMrNPDNbDzxkZgOdc38Nv59z7hHgEYB+/fq5lrrfe0vNW1om9TeJJfU3iSX1N4ml1tjftlV5vw9tjc/W0h3038khfUguXUNb/d1KFGL9b0A002SKiTwCpK5RH9G2he9HdBQD7c0svBwcKY4I9wyPq2FmJwCv4+2IM6qBfIOeDRxPiDJeREREREQOQNfs9nTNbu93GtIU8vrAlpVQWe53JiL7iKYYspzv1+oI1Qv4VxRtuwW25w1vW873a4QsB1KBHhHiCHmf4Nog4fmExwFgZn2Bd4DPgWGBnWb2R/i0HRERERERaURpycmkJSf7nYY0hfy+UF0JW+patlHEP9EUQ14HTjKz7sETZlYI9A9cq88bQDJwcUjbJOBS4F3nXFng9Fy8XWeGh7UfASwL7F4DsBAoqiNuG94ON8H3ORyYB3wDnO+c29NArqGC9//nfrQREREREZH9VLJ7DyW79+dbdWkx8gK/s960vP44ER9Es2bIo8DNwGtmNhZvtMREYB0wMxhkZocBXwMTnHMTAJxzS8xsDt76G8nAauAGoBshBQ3n3GYzewAYY2Y7gc/wCiZn4m3PG4yrMLO7gBlm9h3w10DMNcAtzrnyQC6d8AohKcA9QK+wGThLgoUYM1sCPAWsDDzbIOAWYK5z7r0oPh8RERERETlAm3aWAtAuQ5s5tjrpHaBdQWBHmcv8zkaklgaLIc65XWZ2JvAg8DRgwN+AW51zpSGhBiSy72iTq4FJwL1Ae2ApMMQ591lY3J1AKfBzIB+vOHGJc+7NsHz+aGYO+CXwK2AtcLNzbkZIWC/gsMCfa7UP6AasCfx5JV6x55BA7t8AE4DfRWgnIiIiIiIi0crvC98uAOdgnyUiRfwT1W4yzrm1wLAGYtbgFUTCz+8Bbgu86mtfhVcwuTeKfGYSMiolwvUPIuVSR+xPookTERERERGR/ZTXG1a+DSXrof2hfmcjUiOaNUNERERERERE9l9eH++4aZm/eYiEUTFEREREREREmkb7wyA1U8UQaXaimiYjIiIiIiLSFAo7dvA7BWlKCQneVJmNKoZI86KRISIiIiIi4puUpCRSkvQ72lYtrw8Ur4G9O/zORKSGiiEiIiIiIuKb4l27Kd612+80pCnl9faOm1f4m4dICBVDRERERETEN1tKd7GldJffaUhT6nQ0WAJs/MLvTERqqBgiIiIiIiIiTSc5HXIOh03L/c5EpIaKISIiIiIiItK08vp402SqKv3ORARQMURERERERESaWl5vqNwL2772OxMRQMUQERERERERaWr5fb3jJm2xK82DiiEiIiIiIuKb7jnZdM/J9jsNaWptO3mvjSqGSPOgDb1FRERERMQ3SYmJfqcgsZLXRyNDpNnQyBAREREREfHN1tJdbNXWuvEhvw+UbvZeIj5TMURERERERHyzdddutu7a7XcaEgt5fbzjxv/zNw8RVAwRERERERGRWMjuAUlpmiojzYKKISIiIiIiItL0EpOg09GwabnfmYioGCIiIiIiIiIxktcbiv4NFXv8zkTinHaTERERERERkdjIPwbcbFj6HGR18Tsb2R9tcqDLcX5n0WhUDBEREREREd/0zO3odwoSS3m9vHVDFs/yOxPZX4f1VzFERERERESkMSQkaOZ+XEnNhOEvQNlOvzOR/ZWU5ncGjUrFEBERERER8c2WnaUA5Ga29TkTiZm0LO8l4iOVYUVERERExDfFu/dQvFuLaYpIbKkYIiIiIiIiIiJxRcUQEREREREREYkrKoaIiIiIiIiISFxRMURERERERERE4oo55/zOoUUzsy3At37ncQBygCK/k5C4of4msaT+JrGk/iaxpP4msaT+JrHUlP3tMOdcbvhJFUPilJktcs718zsPiQ/qbxJL6m8SS+pvEkvqbxJL6m8SS370N02TEREREREREZG4omKIiIiIiIiIiMQVFUPi1yN+JyBxRf1NYkn9TWJJ/U1iSf1NYkn9TWIp5v1Na4aIiIiIiIiISFzRyBARERERERERiSsqhsQRMzvUzF40sxIz22FmL5tZV7/zkpbNzC4ys5fM7Fsz22NmK81sipllhsV1MLPHzKzIzHaZ2V/NrK9feUvrYWZzzcyZ2b1h59XnpFGY2blm9qGZlQb+/1xkZmeGXFdfk0ZhZv3N7F0z22xmO83sMzO7JiwmzczuM7MNgf93F5rZAL9ylpbBzArM7OFAf9kd+H+zMEJcVP3LzBLMbIyZrTGzvWa21MyGxeJZpPmLpr+ZWT8ze8TMvgzErDWzZ8ysW4T7NUl/UzEkTphZBvAecBRwFXAFcDjwvpm18TM3afH+B6gCfgMMAf4A3ADMM7MEADMz4I3A9VuAYUAyXv8r8CNpaR3M7DLgBxHOq89JozCz64DXgMXAfwMXAy8AGYHr6mvSKMzsGOCveP1nNPBj4FPgcTO7IST08cD1u4HzgQ3AO2Z2bGwzlhamJ3AJUAx8VE9ctP1rIjAOmAacA3wCvGBm5zZu2tJCRdPffgL0Bn6P14fuAI4DFpnZoWGxTdPfnHN6xcEL+DneD6w9Q851AyqB2/zOT6+W+wJyI5y7EnDAmYGvLwh8fUZITDtgG/B7v59Br5b5AjoAG4HLAv3r3pBr6nN6HfQLKAT2ALfWE6O+plejvIDJQDnQNuz8QmBh4M8/CPS3q0OuJwErgdf9fga9mu8LSAj586hAPyoMi4mqfwGdgDJgfFj7vwFf+P2sevn/irK/RfoZ4jCgGpgQcq7J+ptGhsSPocAnzrlVwRPOudXAx3jfyIkcEOfclginPw0cuwSOQ4H/OOfeD2lXgvfbVPU/OVC/BZY5556NcE19ThrDNXjflP2xnhj1NWksKUAFXgEuVAnfj+YeGoiZE7zonKsEngMGm1lqDPKUFsg5Vx1FWLT9azBef50d1n420DfSNAeJL9H0t0g/QzjnvgW28P3PENCE/U3FkPjRG1gW4fxyoFeMc5HW70eB44rAsb7+19XM2sYkK2k1zOxUvBFIN9URoj4njeFU4EvgJ2b2tZlVmtkqMwvtd+pr0lhmBY6/N7POZtbezEYDZwEPBq71BlY753aHtV2O98NCz5hkKq1VtP2rN95v6ldFiAP9bCEHyMyOxhsJsiLkdJP1NxVD4kc23pytcNvwhpqLNAoz6wJMAP7qnFsUOF1f/wP1QdkPZpYCzATud86trCNMfU4aQ2e89bXuA6YCZwPzgGlm9vNAjPqaNArn3DLgdLwRRd/h9avpwPXOuecCYQ31t+wmTlNat2j7Vzaw3QXmKtQTJxI1M0vCG4m5BW/tmqAm629JB9pQRCRc4Degr+GtRXO1z+lI6/VrIB2Y5Hci0uolAJnASOfcy4Fz7wVWxB9jZr/3KzFpfczscOAlvN92Xo83XeYC4I9mttc594yf+YmINLFpwCnAec65SEW5RqdiSPwoJvJvp+qqAIvsFzNLx5sj3x34kXNufcjl+vpf8LpIg8zbDvxOvMW4UsPmx6eaWXtgJ+pz0ji24o0MmRd2/l283WMOQX1NGs9kvPUaznfOVQTO/c3MOgL/n5k9i9efDovQNtjftkW4JhKtaPtXMdDezCzst/Xqh3JAzGwq8FPgKufcu2GXm6y/aZpM/FiON98qXC/gXzHORVoZM0sGXgT6Aec65/4vLKS+/rfWOVfaxClK69EdSMNbNKs45AXeNs/FQF/U56RxLG/gejXqa9J4+gJLQwohQf8EOuLNo18OdDOzjLCYXng70YTPqRfZH9H2r+VAKtAjQhzoZwvZD2Z2J3A78DPn3NMRQpqsv6kYEj9eB04ys+7BE4Fhvv0D10QOiJklAM8AZwIXOuc+iRD2OtDFzH4U0i4L+H+o/8n++Rw4I8ILvALJGXjfrKnPSWN4JXAcHHZ+CLDeObcR9TVpPBuBYwPrIoX6IbAX77efbwDJwMXBi4F59pcC7zrnymKUq7RO0favuXijmIaHtR+Bt8vb6hjkKq2Amf0MuBe40zk3rY6wJutvmiYTPx4FbgZeM7OxeHs9TwTW4S1EKHKgpuP9pzkJ2GVmJ4VcWx+YLvM6sBCYbWa/wvvt/RjAgN/FOF9pwZxz24EPws+bGcC3zrkPAl+rz0lj+AvwPjDTzHKAb/D+vTub79dFUl+TxjINeAF4w8xm4K0ZMhS4DHjQOVcOLDGzOcBDgVGZq4EbgG7s+4OCSC1mdlHgj8cHjueY2RZgi3NuvnMuqv7lnNtsZg/grZ20E/gMr2ByJl6fFWmwv5nZT4CH8Iod74X9DLHDOfcvaNr+ZvsuyiqtVWCu/YPAILxv0v4G3OqcW+NnXtKymdkaIs8vBRjvnBsXiMsG7gcuxJvmsBC4zTm3NAZpSitnZg6Y5JwbG3JOfU4OWmCUxxTgIry1Qb4Epjrn/hwSo74mjcLMzsEbLt4bry99DTwCzHTOVQViggtIXw60B5YCtweLwSJ1CfxfGcl859zpgZio+peZJeIVfkcD+cBKYIJz7sUmSV5anIb6m5nNAq6qLybkXk3S31QMEREREREREZG4ojVDRERERERERCSuqBgiIiIiIiIiInFFxRARERERERERiSsqhoiIiIiIiIhIXFExRERERERERETiioohIiIiIiIiIhJXVAwRERERERERkbiiYoiIiIiIiIiIxBUVQ0REREREREQkrvz/m5kY2INHeSoAAAAASUVORK5CYII=",
            "text/plain": [
              "<Figure size 1296x216 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "best_epoch = np.argmax(history1['val_accuracy'])\n",
        "plt.figure(figsize=(17,4))\n",
        "plt.plot(history1['loss'], label='Training loss', alpha=.8, color='#ff7f0e')\n",
        "plt.plot(history1['val_loss'], label='Validation loss', alpha=.9, color='#5a9aa5')\n",
        "plt.axvline(x=best_epoch, label='Best epoch', alpha=.3, ls='--', color='#5a9aa5')\n",
        "plt.title('Categorical Crossentropy')\n",
        "plt.legend()\n",
        "plt.grid(alpha=.3)\n",
        "plt.show()\n",
        "\n",
        "plt.figure(figsize=(17,4))\n",
        "plt.plot(history1['accuracy'], label='Training accuracy', alpha=.8, color='#ff7f0e')\n",
        "plt.plot(history1['val_accuracy'], label='Validation accuracy', alpha=.9, color='#5a9aa5')\n",
        "plt.axvline(x=best_epoch, label='Best epoch', alpha=.3, ls='--', color='#5a9aa5')\n",
        "plt.title('Accuracy')\n",
        "plt.legend()\n",
        "plt.grid(alpha=.3)\n",
        "plt.show()\n",
        "\n",
        "plt.figure(figsize=(18,3))\n",
        "plt.plot(history1['lr'], label='Learning Rate', alpha=.8, color='#ff7f0e')\n",
        "plt.axvline(x=best_epoch, label='Best epoch', alpha=.3, ls='--', color='#5a9aa5')\n",
        "plt.legend()\n",
        "plt.grid(alpha=.3)\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 27,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "fvKkbmnujtrS",
        "outputId": "1943ec5f-4abf-45fb-d1ea-70bb540287d1"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "16/16 [==============================] - 1s 11ms/step\n"
          ]
        },
        {
          "data": {
            "text/plain": [
              "(486, 12)"
            ]
          },
          "execution_count": 27,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "predictions = model1.predict(x_test1)\n",
        "predictions.shape"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 28,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "KRdpbCqCjtpR",
        "outputId": "7c1f9a33-cc0e-43fa-d9da-ca5b9ad04ffe"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "16/16 [==============================] - 0s 2ms/step - loss: 0.0649 - accuracy: 0.0000e+00\n",
            "{'loss': 0.06492339819669724, 'accuracy': 0.0}\n",
            "16/16 [==============================] - 0s 3ms/step\n",
            "Accuracy: 0.7243\n",
            "Precision: 0.7308\n",
            "Recall: 0.5916\n",
            "F1: 0.6354\n",
            "F1 (per class): [0.72727273 0.48888889 0.95327103 0.58646617 0.25       0.375\n",
            " 0.84210526 0.4        0.93333333 0.75       0.96551724 0.35294118]\n"
          ]
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 864x864 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 720x576 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "metrics = model1.evaluate(x_test1, return_dict=True)\n",
        "print(metrics)\n",
        "\n",
        "labels = y_test1 \n",
        "predictions = model1.predict(x_test1) \n",
        "assert len(labels) == len(predictions)\n",
        "\n",
        "# Compute the confusion matrix\n",
        "cm = confusion_matrix(np.argmax(labels, axis=-1), np.argmax(predictions, axis=-1))\n",
        "\n",
        "# Compute the classification metrics\n",
        "accuracy = accuracy_score(np.argmax(labels, axis=-1), np.argmax(predictions, axis=-1))\n",
        "precision = precision_score(np.argmax(labels, axis=-1), np.argmax(predictions, axis=-1), average='macro')\n",
        "recall = recall_score(np.argmax(labels, axis=-1), np.argmax(predictions, axis=-1), average='macro')\n",
        "f1 = f1_score(np.argmax(labels, axis=-1), np.argmax(predictions, axis=-1), average='macro')\n",
        "f1_classes = f1_score(np.argmax(labels, axis=-1), np.argmax(predictions, axis=-1), average=None)\n",
        "\n",
        "print('Accuracy:',accuracy.round(4))\n",
        "print('Precision:',precision.round(4))\n",
        "print('Recall:',recall.round(4))\n",
        "print('F1:',f1.round(4))\n",
        "print('F1 (per class):' ,f1_classes)\n",
        "\n",
        "# Plot the confusion matrix\n",
        "ig, ax = plt.subplots(figsize=(12,12))\n",
        "ax.matshow(cm, cmap=plt.cm.Blues, alpha=0.3)\n",
        "for i in range(cm.shape[0]):\n",
        "    for j in range(cm.shape[1]):\n",
        "        ax.text(x=j, y=i,s=cm[i, j], va='center', ha='center', size='xx-large')\n",
        "\n",
        "        \n",
        "plt.figure(figsize=(10,8))\n",
        "sns.heatmap(cm.T, cmap='Blues')\n",
        "plt.xlabel('True labels')\n",
        "plt.ylabel('Predicted labels')\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 29,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "46Qu5_8MyaEG",
        "outputId": "82b7bd19-7a68-4be1-c1d8-8b6f6cdd9783"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "16/16 [==============================] - 0s 3ms/step\n",
            "16/16 [==============================] - 0s 3ms/step\n",
            "16/16 [==============================] - 0s 4ms/step\n",
            "16/16 [==============================] - 0s 4ms/step\n",
            "16/16 [==============================] - 0s 4ms/step\n",
            "16/16 [==============================] - 0s 4ms/step\n",
            "16/16 [==============================] - 0s 4ms/step\n",
            "16/16 [==============================] - 0s 4ms/step\n",
            "16/16 [==============================] - 0s 4ms/step\n",
            "16/16 [==============================] - 0s 4ms/step\n",
            "Accuracy: 0.9815\n",
            "Precision: 0.9927\n",
            "Recall: 0.9515\n",
            "F1: 0.9698\n",
            "F1 (per class): [0.92307692 0.98039216 1.         0.98666667 0.90909091 0.98360656\n",
            " 1.         0.88       1.         0.97484277 1.         1.        ]\n"
          ]
        },
        {
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAArYAAAK0CAYAAAD2y37nAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOzdd3wUZf4H8M9DEkjDEBJ6REFDEuv9kigxBbCAikhQD8upFO8URU/g7LSzF+wc6Im9i4d0BRuKEIoEOAsXAiFwXugmKAlkIeX5/ZFkWWB3k01mMjPffN6v17wYdp6d+X54FvabYXZWaa1BREREROR0rawugIiIiIjICGxsiYiIiEgENrZEREREJAIbWyIiIiISgY0tEREREYnAxpaIiIiIRGhRja1S6kSl1Cyl1O9Kqf1KqdlKqe5W10XeKaX+qJT6RCn1X6VUuVIqXyn1hFKqrdW1UcMopRYrpbRS6lGrayHflFIDlVLfKaXKav9tzFVKXWB1XeSdUipDKfWFUmqPUqpUKbVOKXWT1XURoJSKU0r9Qym1Uil1sPbfv5O9jAtVSj2tlNpZ+/62UinVp/krlqfFNLZKqXAASwAkAhgO4EYA8QC+UUpFWFkb+XQ3gCoA4wFcAuBlALcB+FIp1WJeu06llLoOwNlW10H+KaVGAZgHYC2AKwAMBfAvAOFW1kXeKaXOAvAVgBAANwO4EsAaAK8rpW6zsjYCAJwK4GoA+wAs8zPuddTM32QAgwDsBPC5UuoPplconGopX9CglBoD4DkACVrrgtrHegDYDOBerfVzVtZHx1NKddBa7z3msWEA3gZwodZ6iTWVUX2UUtEA8gCMA/ABgMe01hOtrYqOVXsmKQ/AA1rrF6ythhpCKfU4an7ob6+1LvN4fCUAaK3Ps6o2ApRSrbTW1bXrfwHwKoAeWuttHmPOBvBvADdprd+sfSwYwAYA+Vrrwc1euCAt6azXYACr6ppaANBabwWQAyDbsqrIp2Ob2lpran/t1py1UMCeAvCz1vpDqwshv24CUA3gn1YXQg3WGkAFgPJjHv8dLes93Zbqmtp6DEbNHM70eF4lgI8AXKyUamNSeS1CS/pLcDqAn708vgHAac1cCzVe39pf8yytgnxSSmUCGAbgdqtroXplAtgI4Fql1BalVKVSqkApxbmzr7dqf52qlOqqlGqnlLoZwIUAnreuLArA6QC2aq0PHvP4BtT84HJq85ckR7DVBTSj9qi55uVYJQCim7kWagSlVDcADwP4Smuda3U9dDylVGsArwB4Rmudb3U9VK+utcvTqLmWfQtqrrGdppQK1lq/aGVxdDyt9c9KqX4A5gAYXftwBYBbtdYfWVYYBcJfP1K3nRqpJTW25GBKqUjUfMClEsBIi8sh3+4FEAbgMasLoQZpBaAtgBFa69m1jy2pvfb2AaXUVN1SPojhEEqpeACfoObs3q2ouSQhG8A/lVIurfX7VtZHZLWW1Njug/czs75+ciKbUEqFAVgAoCeAvlrrIotLIi9qb503AcBfALQ55jqxNkqpdgBKtdZVlhRI3hSj5u4wXx7z+BeouRNJFwA7mrso8utx1JyhHaS1rqh97GulVAyAF5VSHzbwOk+yzj4AJ3l5vO5MbYmXbdRALeka2w2oua7lWKcB+E8z10INpJQKATALQCqAgVrrnywuiXzrCSAUwHuo+Ye7bgFqPsW9D8CZ1pRGPmyoZzsbJPs5E8APHk1tne8BxADo2PwlUYA2AOhRextST6cBOAyg4PinUEO1pMZ2PoA0pVTPugdq/7sto3Yb2UztvWrfB3ABgCFa61UWl0T+/RvA+V4WoKbZPR/8B9tu5tT+evExj18CoEhrvauZ66H67QLwh9rr2T31BuACz/Y5wQLU3Id4aN0Dtbf7ugbAF1rrQ1YVJkFLuhThVQB3AJinlJoIQAN4BMD/UPNhF7Kf6aj5i/8YgANKqTSPbUW8JMFetNa/Afj22MeVUgDwX631cdvIcp8B+AbAK0qpWACFqPk7NwC8lt2upqHmCzQWKKVeQs01toMBXAfgea31YSuLo5pvzaxdTan99VKl1F4Ae7XWS7XW65VSMwG8UPu/kltR8+VDPQBc3/wVy9JivqABcF8D+DyA/gAUgK8BjPW8cTLZh1JqG7xfhwQAD2mtH2y+aqixlFIa/IIG21JKnQDgCQB/RM3nEDYCeFJr/YGlhZFPSqlLAdyHmsvrQlFzN4sZAF7hNezWq/03z5ulWut+tWPqPmT7JwDtAPwA4D6eAGi6FtXYEhEREZFcLekaWyIiIiISjI0tEREREYnAxpaIiIiIRGBjS0REREQisLElIiIiIhHY2BIRERGRCC22sVVK3WJ1DdRwnC/n4Zw5D+fMeThnzsL5Ml+LbWwB8MXlLJwv5+GcOQ/nzHk4Z87C+TJZS25siYiIiEgQW33zWGxsrD7p5JOb5Vi/7t2L2A4dmuVYqlmOItvevXvRoZnmi4zBOXMezpnzcM6chfNljB9/+mn/4UOHorxtC27uYvw56eSTsWr1GqvLMFwrxdaWiIgksM/JMGPxfdpJOsS23+NrGy9FICIiIiIR2NgSERERkQhsbImIiIhIBDa2RERERCQCG1siIiIiEoGNLRERERGJwMaWiIiIiERgY0tEREREIrCxJSIiIiIR2NgSERERkQhsbImIiIhIBDa2RERERCRCi2tsCwsLMWniBJyTmozOHWNxQmQ4Tu15MgZcdCGeePwxbNu2zeoSA1ZYWIhxY8cgKbEX2kaGo1PHWGRmpmP69GmoqKiwurwmkZqNuZxHajbmch5J2Xbv3o0F8+dj4sQJGDCgP9pHt0NwUCsEB7XCQw89aHV5hpA0X8eyZTattW2W5JQUfbiy2rTliSef0qGhoRqAz+WZZ58z/LiVVdq0Zdas2ToyMtJnnuTkZL1j525Ta2A25pKcS3I25rK+VudlqzZ08fd+PGnyZMOP53uROl8yX4vR0dGbffWSLeaM7aSJE/DA/ffB5XLh/5KT8eLUf2DJN0uxes1aLPx0Ef7+4EM46+yzoZSyutQGy83NxQ03/AllZWWIiorCk09NwfKclVi0+Atcc821AIB169bhiiuyUVlZaXG1gZGajbmclQuQm425nJULkJ0NAHr27ImsPn2sLsMwkufL1tmsPkvbHGdsFyz8zP0TxH33P6APVVT5HFt20OWYM7bpGRkagA4JCdGrVq85bvudY8a6c8+Y8ZrlP90xG3M5LZfkbMzlrFz2yWbsmdKJkybp+QsW6t179urKqmr91ddLxJyxtcd8yXwt+jtja3kza3Zj6zpcqXv27KkB6IEDLzP1UofmbGyX56x0v2hGj77d65jSsoM6JiZGA9BJSUmW/0Vo6dmYy1m5JGdjLmflslc2cxtMKY2tfeZL5muxRV+K8PnixSgsLAQA3P/AeIurMc68eXPd6zcOG+51TFhYGIYOvRoAkJeXh/z8/GapramkZmMuZ+UC5GZjLmflAmRnk0jyfNk9m/jG9pNPZgEAYmNjkXbeee7H9+zZg4KCAuzfv9+q0ppkxYocAEBERASSk5N9jsvKOnK90oqcHNPrMoLUbMzlrFyA3GzM5axcgOxsEkmeL7tnM7SxVUqdqJSapZT6XSm1Xyk1WynV3chjBGrNmu8BAGeccSa01pg+7R9IiD8FcV0747TEXoht3w6pKf+HN994HdXV1VaWGpCNeXkAgPj4eAQFBfkcl5CY6F7P25hnel1GkJqNuZyVC5CbjbmclQuQnU0iyfNl92yGNbZKqXAASwAkAhgO4EYA8QC+UUpFGHWcQFRXVyN/40YAQHT79rh66B8xbuwYbN269ahxP/7wA0bdcjOG/vEqHD582IpSA+JyuVBcXAwA6NYtzu/YuLgj27cXFZlalxGkZmMuZ+UC5GZjLmflAmRnk0jyfDkhm5FnbG8G0BPAEK31XK31PACDAZwEYJSBx2mw33//3X0WdtFnn2Le3DmI79ULs+fMQ/G+31Hy237MnbfA/VPFgvnzMOGB+60oNSClpaXu9YgI/z8zeG4vKyszrSajSM3GXM7KBcjNxlzOygXIziaR5PlyQjYjG9vBAFZprQvqHtBabwWQAyDbwOM02IEDB9zrLpcLnTt3xrdLl2HQ5Zejbdu2iIyMxMDLLsM3336Hrl27AgBeemk6imz+U5PL5XKvt27d2u/YNm3auNfLXeWm1WQUqdmYy1m5ALnZmMtZuQDZ2SSSPF9OyGZkY3s6gJ+9PL4BwGkGHqfBQkNDj/r9XXffgw4dOhw3LjY2Fvfd/wAAoKKiAnPnzG6W+hrLM1d9l04cOnTIvR4WGmZaTUaRmo25nJULkJuNuZyVC5CdTSLJ8+WEbEY2tu0B7PPyeAmAaAOP02Bt27Y96vcX9R/gc2z/ARe713Nzc02ryQieuTzPSnvjuT0yMtK0mowiNRtzOSsXIDcbczkrFyA7m0SS58sJ2Sy/3ZdS6halVK5SKvfXvXsN3XebNm2OOkN74okn+hzruc3oOowWGhqKmJgYAMD27f4vm/C8rKJbnP8Lve1AajbmclYuQG425nJWLkB2Nokkz5cTshnZ2O6D9zOzvs7kAgC01jO01qla69RYL5cJNFVS0pGrIKqqqnyO89wWHBxseB1GS0xKAgAUFBT4zbXJ46bISYlJptdlBKnZmMtZuQC52ZjLWbkA2dkkkjxfds9mZGO7ATXX2R7rNAD/MfA4AcnMynKvb639BjJvtmzZ4l7v2q2rqTUZIT09A0DNJw3Xr1/vc9yyZd8deU5Ghul1GUFqNuZyVi5AbjbmclYuQHY2iSTPl92zGdnYzgeQppTqWfeAUupkABm12ywx5Ior3etz587xOc7zA2PpGZmm1mSE7Owh7vV33n7L6xiXy4WPP54JAEhKSkJCQkJzlNZkUrMxl7NyAXKzMZezcgGys0kkeb5sn01rbcgCIAJAAYCfUHN7r8EAfgBQCCCyIftITknRhyurDV/69x+gAeiIiAi9/oefjtv+w08bdGRkpAagY2Nj9W/7yww9fmWVNmVJz8jQAHRISIhe/X3ucdvvHDNWA9AA9IwZr5lWB7Mxl9RckrMxl7Ny2SdbtanLV18vcWeYNHmy6cc7skidL5mvxejo6M2+eklV25Qaovbrc58H0B+AAvA1gLFa620NeX5KaqpetXqNYfXUycvLQ1bGedi/fz/atWuHu+6+B/36nQ+g5lT501Oewr59NZcBv/Pe+7j22usMPX4rpQzdX53c3Fz065sFl8uFqKgojJ8wEZmZWSgtLcWbb7yOmTM/AgD0TkvD0qXLHHHtcB2p2ZjLWbkAudmYy1m5ALtkM65nAIDly5djS4H79vfYmL8RT0+ZAgAYPDgb2dlH3wZ/+IgRhh7/COPfp+0xX+awOluH2PYFJSUl8V43GnXG1ojFrDO2hyur9ZdfLdGxsbHunyCOXYKDg/ULL0415dhm/tQ0a9Zs99lmb0tycrLesXO35T/dMRtzOTWX5GzMZX2tzstm7JnSYcOG+8zibXHSGVt7zJfM12KznbFtKrPO2NbZs2cPpv1jKhYuXID/btuGyspKdIuLwwUXXIC/3jnWtGtAzDpjW6ewsBBTX3wBixcvQlFREcLDw9ErIQHXXnsdRo26FSEhIaYe30xSszGX80jNxlzOY202Y3uGm0aOxDvvvN3g8ZVV1YYe/wjz3qf5WjSevzO2LaqxtYrZjS0REVHzsE/PYCy+TzuJv8bW8i9oICIiIiIyAhtbIiIiIhKBjS0RERERicDGloiIiIhEYGNLRERERCKwsSUiIiIiEdjYEhEREZEIbGyJiIiISAQ2tkREREQkAhtbIiIiIhKBjS0RERERicDGloiIiIhEYGNLRERERCKwsSUiIiIiEdjYEhEREZEIbGyJiIiISAQ2tkREREQkQrDVBXhSAFopZXUZhis/XGl1CaYJa22rlxAREZlK3ns0ycIztkREREQkAhtbIiIiIhKBjS0RERERicDGloiIiIhEYGNLRERERCKwsSUiIiIiEdjYEhEREZEIbGyJiIiISAQ2tkREREQkAhtbIiIiIhKBjS0RERERicDGloiIiIhEYGNLRERERCKwsSUiIiIiEdjYEhEREZEILaqxLSwsxLixY5CU2AttI8PRqWMsMjPTMX36NFRUVFhd3lHW5q7BE489gssHXoz4Ht3R/oRwdOkQjZQ/nIk777gN/16/rt59PP7ow2gbFtKgZdl3S5shVeCcNGeBYC7nkZqNuZxHajbmch5bZtNa22ZJSUnRlVXalGXWrNk6MjJSA/C6JCcn6x07d5ty7NLyioCW9IxMn3XWLUopfcedY/X+g4d97ueBCZPq3U/d8tnnXwVcZ2l5hWnzZfWcMRdztYRszGV9rczGXE7MZXW26Ojozb56SaW1DqwTNlFqaqpe/X2u4fvNzc1Fv75ZcLlciIqKwgPjJyAzMwulpaV46803MHPmRwCA3mlpWLp0GYKDgw09fvnhyoDGn316IgoLt6BbtzgMufIqpGdkIi4uDocOHcLy5cswbeoLKCkuBgCMu+sePPzo41738/ijD+OJxx4BAKzOXe/3mCed3AMREREB1QkAYa2N/bOqY/WcmYW5nJULkJuNuZyVC5CbjbmclQuwPluH2PYFJSUl8V43Wn2WtjnO2KZnZGgAOiQkRK9avea47XeOGev+CWPGjNcMP36gZ0EvGXiZfvf9j/RvZS6v23/K26Q7duqkAejg4GD9w4aN9Z6xbczZWCvP2Fo9Z8zFXNKzMZezcknOxlzOymWHbP7O2FrezJrd2C7PWen+wx09+navY0rLDuqYmBgNQCclJRlegxkN5XMvTHXnevzJKaIaWzvMGXMxl+RszOWsXJKzMZezctklm7/GVvyHx+bNm+tev3HYcK9jwsLCMHTo1QCAvLw85OfnN0ttTZGZ1ce9vnVroYWVGE/qnDGXs3IBcrMxl7NyAXKzMZezcgH2zya+sV2xIgcAEBERgeTkZJ/jsjwaxRU5OabX1VSHDx92rwe1CrKwEuNJnTPmclYuQG425nJWLkBuNuZyVi7A/tnEN7Yb8/IAAPHx8QgK8t0AJiQmutfzNuaZXldT5Sxf5l73rN2X7EGXokf3rmh/Qjh6dO+KSwdciGefnoJ9+/aZWWajSJ0z5nJWLkBuNuZyVi5AbjbmclYuwP7ZRDe2LpcLxbV3D+jWLc7v2Li4I9u3FxWZWldTlZeX4+Xp0wAArVu3xmWDBtf7nCVff4Vf9+5FRUUFft27F8uXfYcHJ0/AWaf1wsIF880uucGkzhlzOSsXIDcbczkrFyA3G3M5KxfgjGzOubdEI5SWlrrX67uVlef2srIy02oywsMPTsa2bVsBADffciu6dO3qc+zpZ5yBQZdnIyX1HHTp0gUVFRXYvHkTPv7oQ3z91Zf47bffcMN1V+PjT+ZiwMWXNFcEn6TOGXM5KxcgNxtzOSsXIDcbczkrF+CMbKIbW5fL5V5v3bq137Ft2rRxr5e7yk2rqanmzv4E06a+AAA4Nb4XJj/0iM+xo++4E+MnTj7u8XPO7Y0/XX8j3nhtBsb89XZUVVXhjttG4cf/5CM0NNS02htC4pwBzAU4KxcgNxtzOSsXIDcbczkrF+CMbKIvRfBs0jw/bOXNoUOH3OthoWGm1dQUq1etxC1/GQkAaNeuHd7/cCbCw8N9jm/Xrp3f/d30l1swbETN/nbu3IF5c2YbV2wjSZuzOszlrFyA3GzM5axcgNxszOWsXIAzsolubNu2beteP3DggN+xntsjIyNNq6mx8v6zAUOvzEZ5eTnCwsIwc9YcnHb6GU3e701/vtm9vnz5d03eX1NJmjNPzOWsXIDcbMzlrFyA3GzM5axcgDOyiW5sQ0NDERMTAwDYvt3/hctFHhc2d4vzf0F0c9u2bSuyBw3Evn37EBISgnc/mIn0jExD9p2YdJp7fceOHYbssymkzNmxmMtZuQC52ZjLWbkAudmYy1m5AGdkE93YAkBiUhIAoKCgAFVVVT7HbfK4eXBSYpLpdTXU7l27kH3Zpdi5cwdatWqFV19/Cxdfcqlh+1dKGbYvozh9znxhLmflAuRmYy5n5QLkZmMuZ+UC7J9NfGObnp4BoOYTeevXr/c5btmyI/8Nn56RYXpdDVFSUoLBgy5BYeEWAMCL017CVbXf5GGUjXn/ca936dLF0H03lpPnzB/mclYuQG425nJWLkBuNuZyVi7A/tnEN7bZ2UPc6++8/ZbXMS6XCx9/PBMAkJSUhISEhOYoza+ysjJcNeRy/GfDBgDA409OwYiRfzb8OG+8/qp73fNreq3k1DmrD3M5KxcgNxtzOSsXIDcbczkrF+CAbFpr2ywpKSm6skobvqRnZGgAOiQkRK/+Pve47XeOGasBaAB6xozXDD9+aXlFQMuvv5Xpfudf4K7p/vETA97HqjXr9L9/zvM75sVpL7mP0alzZ73r198CPo4Z82WHOWMu5pKejbmclUtyNuZyVi47ZIuOjt7sq5e0vJltjsZ21eo1OjQ0VAPQUVFR+qkpT+ucFav04s+/1Ndcc637D793Wpp2HTK+WQu0Wbx88BB3TZcMvEyvzl3vd/kpb9Nx+3h5xms6KChIn3/BhfrxJ6foeQsX6WUrVutvvsvRr7z2hr7wov7uYwQFBemZs+YEXKeZja3Vc8ZczCU9G3M5K5fkbMzlrFx2yNbiG9vKKq1nzZqtIyMj3X/Yxy7Jycl6x87dphw70GbRV42+lsysPl4b24Y8t31MjP7w408a1dSa2dhaPWfMxVwtIRtzWV8rszGXE3NZnc1fY6u01rCL1NRUvfr7XNP2X1hYiKkvvoDFixehqKgI4eHh6JWQgGuvvQ6jRt2KkJAQU45bfrgyoPFtwwKrIzOrDxZ98fVRj+3dsweLFn2K71evwo8//Bt7du9BSUkxtNaIjm6PM886Cxf1H4DrbxyOE044IaDjeQprbe6X11k1Z2ZjLueRmo25nEdqNuZyHquydYhtX1BSUhLvbVuLamytEmhj6yRmN7ZEREREnvw1tuLvikBERERELQMbWyIiIiISgY0tEREREYnAxpaIiIiIRGBjS0REREQisLElIiIiIhHY2BIRERGRCGxsiYiIiEgENrZEREREJAIbWyIiIiISgY0tEREREYnAxpaIiIiIRGBjS0REREQisLElIiIiIhHY2BIRERGRCGxsiYiIiEgENrZEREREJEKw1QW0BGGt5f4x7yott7oEU3RuG2Z1CURERBQgnrElIiIiIhHY2BIRERGRCGxsiYiIiEgENrZEREREJAIbWyIiIiISgY0tEREREYnAxpaIiIiIRGBjS0REREQisLElIiIiIhHY2BIRERGRCGxsiYiIiEgENrZEREREJAIbWyIiIiISgY0tEREREYnQohrbwsJCjBs7BkmJvdA2MhydOsYiMzMd06dPQ0VFhdXlNZqTcsW1C2/Q8sfLLg5ov3t278LpJ3Vt9PObm5PmLBBScwFyszGX80jNxlzOY8dsSmttyYG9SU1N1au/zzVl33PnzMGIEcNQVlbmdXtycjIWfroIHTt2NOX4ZrE6167S8oDGx7ULb9C4tIwszPr08wbv97aRN2LBnE8a/fxjdW4b1ujn1sfqOTOL1FyA3GzM5axcgNxszOWsXIC12TrEti8oKSmJ97atRTS2ubm56Nc3Cy6XC1FRUXhg/ARkZmahtLQUb735BmbO/AgA0DstDUuXLkNwcLDhNZjBDrka29gO+/PNGPbnW3yOCw+PQPeTT27QPr/+YjGGX30lYjt0xK979wCwb2Nrhzkzg9RcgNxszOWsXIDcbMzlrFyA9dn8NbbQWttmSUlJ0ZVV2vAlPSNDA9AhISF61eo1x22/c8xYDUAD0DNmvGZKDVJzFf12MKClrp5x940P+Lnelk3b9+q4E7trAPqFf77m3n9aRlaT9it5zpiL2ZjLebkkZ2MuZ+WyQ7bo6OjNvnpJy5tZsxvb5Tkr3X+4o0ff7nVMadlBHRMTowHopKQky18wTspldWM76o4xGoA+L7PPUfu3Y2NrlzljLmZjLmflkpyNuZyVyy7Z/DW24j88Nm/eXPf6jcOGex0TFhaGoUOvBgDk5eUhPz+/WWprCqm5ArHhxx/w2svTEBISgseffcHqcuoldc6k5gLkZmMuZ+UC5GZjLmflAuyfTXxju2JFDgAgIiICycnJPsdlZfU58pycHNPraiqpuRqquroa9469A1VVVRj11zGIT0i0uqR6SZ0zqbkAudmYy1m5ALnZmMtZuQD7ZzO0sVVK/VEp9YlS6r9KqXKlVL5S6gmlVFsjjxOIjXl5AID4+HgEBQX5HJeQeKQxytuYZ3pdTeX0XJ/Om4Pzeyfj1C4xSIjriMzkMzH21puR893SBj3/zRn/xA/r1uLE7idhzN33m1ytMZw+Z75IzQXIzcZczsoFyM3GXM7KBdg/m9EfwbsbwC8AxgMoAvB/AB4EcL5SKl1rXW3w8fxyuVwoLi4GAHTrFud3bFzcke3bi4pMraupJOTadMyL/EBZGbYVbsGsj97HxZddjudfmoEToqK8PnfH9iI8/dhDAIBHpjyLsPCG3ULMShLmzBupuQC52ZjLWbkAudmYy1m5AGdkM7qxvVxrvdfj90uVUiUA3gbQD8ASg4/nV2lpqXs9IiLC71jP7b7uyWYXTs4VFh6OAZdehoy+/XBqfAIiIiJQXPwrVuUsw7tvvI59JcX4/NMFuOlPV+PDuQsREhJy3D4m3fM3lJWW4uLLLsdFlwy0IEXgnDxn/kjNBcjNxlzOygXIzcZczsoFOCOboY3tMU1tnTW1v3Yz8lgN4XK53OutW7f2O7ZNmzbu9XJXYPdmbW5OzpX7nwJEtWt33ON9zr8QI2+5DTf+cQh+/vEHrMpZhndefxV/vnX0UeMWLZiHzz9biPCICDz85DPNVXaTOXnO/JGaC5CbjbmclQuQm425nJULcEa25vjwWN/aX5v94pHQ0FD3+uHDh/2OPXTokHs9LNS8b50ygpNzeWtq63To2AmvvPOB+yztmzNePmp7WWkpJt13FwBg3H3j0e3EE80r1GBOnjN/pOYC5GZjLmflAuRmY6K39aYAACAASURBVC5n5QKckc3UxlYp1Q3AwwC+0lqb8125frRte+QzawcOHPA71nN7ZGSkaTUZQWouADjp5B7IOv8CAMC2wi3YtXOHe9tTjzyIXTt2ICHpNNw8+q9WldgoUudMai5AbjbmclYuQG425nJWLsAZ2Uz7/jalVCSAeQAqAYz0M+4WALcAQPfu3Q2tITQ0FDExMSguLsb27f4vXC7yuLC5W5z/C6KtJjVXnfiEJCz5oubrcHft3IHOXbriQFkZ3n7tFQBARp9++HTeHL/7KP51L+Z98i8AwCmnxuOMs/9gbtH1kDpnUnMBcrMxl7NyAXKzMZezcgHOyGbKGVulVBiABQB6ArhYa+0zvdZ6htY6VWud2qFDB8NrSUxKAgAUFBSgqqrK57hNHjcPTkpMMrwOo0nNBQBKqeMeq6ysRHV1zU013njlJdz+5+Felzqb8ze6H/vXh+83W+3+SJ0zqbkAudmYy1m5ALnZmMtZuQD7ZzO8sVVKhQCYBSAVwECt9U9GHyMQ6ekZAGo+kbd+/Xqf45Yt++7IczIyTK+rqaTmAoDN+Ucux+7UuYuFlRhL6pxJzQXIzcZczsoFyM3GXM7KBdg/m9Ff0NAKwPsALgAwRGu9ysj9N0Z29hD3+jtvv+V1jMvlwscfzwQAJCUlISEhoTlKaxKpuX7Ztg3Lvqm5K9xJPXqiS9eam2lEtWuHot8O1rvUScvIcj/20JNPW5LlWFLnTGouQG425nJWLkBuNuZyVi7A/tmMPmM7HcBQAM8COKCUSvNYLLl4JC0tzf2TwquvzsDatWuPGzN+/APuGw6PG3dXs9bXWE7M9eWiT1FZWelz+949u3HLsOvcn7Qc/udbmqu0ZuHEOWsIqbkAudmYy1m5ALnZmMtZuQD7Z1Naa+N2ptQ2ACf52PyQ1vpBf89PTU3Vq783/uYJubm56Nc3Cy6XC1FRURg/YSIyM7NQWlqKN994HTNnfgQA6J2WhqVLlyE42LTP1BnKDrl2lTb83nRpZyaiorICAy8fgpRze+PE7ichNDQUJSXFWLn8O7z35hsoKf4VAHDueen4cO6nR90HryHi2tV8C1laRhZmffp5QM/11LmtObcmscOcmUFqLkBuNuZyVi5AbjbmclYuwPpsHWLbF5SUlMR722ZoY9tUZjW2ADB3zhyMGDHM57dfJCcnY+Gni9CxY0dTjm8Wq3MF2tgW/e+XescNHDwET099ye89b32xe2MLWD9nZpGaC5CbjbmclQuQm425nJULsDYbG9tahYWFmPriC1i8eBGKiooQHh6OXgkJuPba6zBq1K1ev77VCazMFUhju3L5MqzKWYa1a1bjl23bUFJcjLLS/YiIiESXbnFI7d0bQ6+7ASnn9m50PU5obAG+Fp1Iajbmch6p2ZjLeazKxsaWTBNIY+skZje2RERE1Dj+Gtvm+EpdIiIiIiLTsbElIiIiIhHY2BIRERGRCGxsiYiIiEgENrZEREREJAIbWyIiIiISgY0tEREREYnAxpaIiIiIRGBjS0REREQisLElIiIiIhHY2BIRERGRCGxsiYiIiEgENrZEREREJAIbWyIiIiISgY0tEREREYnAxpaIiIiIRGBjS0REREQiBFtdQMugrS7ANJ3bhlldginy95ZZXYJpEjpEWl0CETmW1PczZXUBZBCesSUiIiIiEdjYEhEREZEIbGyJiIiISAQ2tkREREQkAhtbIiIiIhKBjS0RERERicDGloiIiIhEYGNLRERERCKwsSUiIiIiEdjYEhEREZEIbGyJiIiISAQ2tkREREQkAhtbIiIiIhKBjS0RERERidCiGtvCwkKMGzsGSYm90DYyHJ06xiIzMx3Tp09DRUWF1eUFZPfu3Vgwfz4mTpyAAQP6o310OwQHtUJwUCs89NCDVpdnGKfM2YgrLsXpndsGtHyfs6xB+967ZzfOSzjR/bwRV1xqcprGc8p8NYbUbMzlPJKytYT3MknzdSw7ZlNaa0sO7E1qaqpe/X2uKfueO2cORowYhrKyMq/bk5OTsfDTRejYsaMJRzf+zzg4yPfPJJMmT8bf//6g4cf0Tpm2ZyvnLH+v92P6MuKKS7Fm5fIGj2/VqhW+WpuHTl261jv2rluGY/H82e7fn3NeJt6asyig+jwldIhs9HP9sfbvmLmkZmMuZ+UC7JDN2Pcz6e9l1s+XeazM1iG2fUFJSUm8t20t4oxtbm4ubrjhTygrK0NUVBSefGoKluesxKLFX+Caa64FAKxbtw5XXJGNyspKi6sNXM+ePZHVp4/VZRjKaXP26AsvY+43q/0uz77ytnt8Wla/BjW13331ORbPn42Y2A5mlt9kTpuvQEjNxlzOygXIzgbIey+TPF+2zqa1ts2SkpKiK6u04Ut6RoYGoENCQvSq1WuO237nmLEaNT+G6hkzXjOhhmrDl4mTJun5Cxbq3Xv26sqqav3V10vcGSZNnmzKMb0vxs+XHeZsw65Sw5ebbj9S85PTXq13/Jotu3TXuO4agH7iHzPczz3nvMwm1SFxvsxcpGZjLmflsk82vpc5a75kvhajo6M3++olxZ+xXbVqFVbk5AAAbr75FqSmph435rHHHkdMTAwA4Pnnn23W+hrrwQcfwsCBA911SyJxzqqrq/Hp7I8BAOERkbho4OB6nzP96cewo+gXnJOehcFDrzO7xEaTOF91pGZjLmflAuRmk/peJnW+APtnE9/Yzps3171+47DhXseEhYVh6NCrAQB5eXnIz89vltrIO4lztmrZt9i9cwcAYMCgbISFh/sdn/fzj3j31ZcQHBKCSU8+1xwlNprE+aojNRtzOSsXIDubRJLny+7ZxDe2K1bU/FQRERGB5ORkn+Oyso5c11P3kwhZQ+Kczf/Xh+71wUP/5HdsdXU1Hrz7TlRVVWHkbXfilF6JZpfXJBLnq47UbMzlrFyA7GwSSZ4vu2cT39huzMsDAMTHxyMoKMjnuITEI81D3sY80+si36TN2YEDZfjqswUAgK5x3XFuRpbf8R+88Qp+/vdadDvxJIwae29zlNgk0ubLk9RszOWsXIDsbBJJni+7ZxPd2LpcLhQXFwMAunWL8zs2Lu7I9u1FRabWRb5JnLMvF85D+cEDAIBBf7wGSvm+rcyuHdsx9clHAADjH3u63ksWrCZxvupIzcZczsoFyM4mkeT5ckI20Y1taWmpez0iIsLvWM/tvu7JRuaTOGeelyFk1/MhsMfG34UDZaW48NJB6DfAvl/EUEfifNWRmo25nJULkJ1NIsnz5YRsohtbl8vlXm/durXfsW3atHGvl7vKTauJ/JM2Z7t2bMeaFTXfMHZ2yjk4+RSv95MGAHz12XwsWfwpwsIjcP8jU5qrxCaRNl+epGZjLmflAmRnk0jyfDkhm+jGNjQ01L1++PBhv2MPHTrkXg8LDTOtJvJP2pwt+OQjVFdXAwCyr77e57gDZaV4fMI9AIDRd92PrnEnNkt9TSVtvjxJzcZczsoFyM4mkeT5ckI20Y1t27Zt3esHDhzwO9Zze2SkOV85SvWTNmcL/vURAKB1mza4JPtKn+NefOJh7N65A6cmJGHYqDuaq7wmkzZfnqRmYy5n5QJkZ5NI8nw5IVtwsx3JAqGhoYiJiUFxcTG2b/d/4XKRx4XN3eL8XxBN5pE0Zz//ex22bNoIAOh70SWIahftddyBA2X48M0ZAGq+aveLhXO9jqtT/OtefDZ3FgCgxynxSDrzbAOrDoyk+TqW1GzM5axcgOxsEkmeLydkE93YAkBiUhJyli9HQUEBqqqqfN6aYpPHzYOTEpOaqzzyQsqcHfWhsWt837u2qrLSfbnCe6+9jPdee9nvfgs35+OeW0cCAG64ebSljS0gZ768kZqNuZyVC5CdTSLJ82X3bKIvRQCA9PQMADWfyFu/fr3PccuWfXfkORkZptdFvkmYs4qKCiyqPavaPiYWWRcMsLgi80iYL1+kZmMuZ+UCZGeTSPJ82T2b+MY2O3uIe/2dt9/yOsblcuHjj2cCAJKSkpCQkNAcpZEPEuZs2ZIvUFL8KwBg4JVXIzjY93+OnBDVDht2lda71DnnvEz3Yw888pTpWeojYb58kZqNuZyVC5CdTSLJ82X3bOIb27S0NPdPCq++OgNr1649bsz48Q+4bzg8btxdzVofHU/CnM3/uOH3rnU6CfPli9RszOWsXIDsbBJJni+7Z1Na62Y9oD+pqal69fe5hu83NzcX/fpmweVyISoqCuMnTERmZhZKS0vx5huvY+bMmk+u905Lw9Kly/yeXWsc4/+Mly9fji0FBe7fb8zfiKen1Nz7dPDgbGRnZx81fviIEYbXUMP3t2g1hdVzlr+38TeT/v23feh3djwOHzqE+MTTMPfb1YbUdHrnmk+jnnNeJt6as6jR+0noYPynU62eLzNJzcZczsoF2CWbse9nkt/L7DFf5rA6W4fY9gUlJSVebwzfIhpbAJg7Zw5GjBjm89svkpOTsfDTRejYsaMJRzf+z/imkSPxzjtvN3h8ZVW14TXUMKexBayds6Y0tjPffh0P3zcWAHDXpEdw0+1jDanJzo0tYPXfMXNJzcZczsoF2CGbse9n0t/LrJ8v81iZzV9jK/5ShDpDrrgC69b/gDvu+CtOPfVUhIaGon379kg77zy88OJU5KxY5cgXlmROnbP5s2ouQwgKCsKgq66xuJrm49T5agip2ZjLeSRnk0jyfNk1W4s5Y2st+/wZG8+8M7ZWasoZW7sz64wtEbUEUt/PZL6XScUztkREREQkHhtbIiIiIhKBjS0RERERicDGloiIiIhEYGNLRERERCKwsSUiIiIiEdjYEhEREZEIbGyJiIiISAQ2tkREREQkAhtbIiIiIhKBjS0RERERicDGloiIiIhEYGNLRERERCKwsSUiIiIiEdjYEhEREZEIbGyJiIiISAQ2tkREREQkAhtbIiIiIhIh2OoCWgZldQEUoIQOkVaXYJrK6mqrSzBFcCv+nE5kNm11ASbhu7QcfCcgIiIiIhHY2BIRERGRCGxsiYiIiEgENrZEREREJAIbWyIiIiISgY0tEREREYnAxpaIiIiIRGBjS0REREQisLElIiIiIhHY2BIRERGRCGxsiYiIiEgENrZEREREJAIbWyIiIiISgY0tEREREYnQohrbwsJCjBs7BkmJvdA2MhydOsYiMzMd06dPQ0VFhdXlNZrUXIDcbE7KtXv3bixcMB+TJ03EpRcPQMeYaISGBCE0JAiPPPxQQPvSWuO9d9/BJQP6o3u3LoiKDEdC/Cm49Zab8eMPP5iUwBhOmrNAMJfzSMq2Zs0aPPLwQ7h4QH90j+uG8NA2iI46AWeefhpuu3UU1q1bZ3WJTSZpvo5ly2xaa9ssKSkpurJKm7LMmjVbR0ZGagBel+TkZL1j527Tjs9czGaXXK6KqoAWX3UC0BMmTW7wfkp+L9UXXnSRz32FhIToqf+YFnB9dYvkOWMu5rJLtoqqasOWzKwsv/++ANBKKT123Dh9uLLK0GMfu0idL6mvxejo6M2+eknLm9nmaGxXrV6jQ0NDNQAdFRWln3xqil6es1IvWvyFvuaaa92T0DstTbsOVVj+YmnpuSRns0OupjS2PXr21JlZfRrV2F551VXu513Uv7+ePXeeXr5ilZ76j2m6S5cu7jex2XPn2aqxtcOcMRdz2SWbkc3kKaecogHouLg4PWbsWP2vWZ/olau/199+t0w/8uijOiYmxp3pnnvvdVxja4f5kvpabPGNbXpGhvuM0KrVa47bfueYse5JmDHjNctfMC09l+RsdsgVaMM4fuIkPXf+Ar191x7tqqjSn3/1dcCN7cJFi93PueLKK/XBQxVHbd+4eYtu166dBqBP7tFDlx502aaxtcOcMRdz2SWbkc3kZZcN0h/N/Fi7Dld43b6pYIvu1KmTBqCDg4P1xk2bHdXY2mG+pL4WW3RjuzxnpfsPd/To272OKS076P7JMCkpyfIXTEvOJTmbXXI19r/665bGNLb9Bwxwvzlt/aXI65jHn3zKvd/3PvjQFo2tXeaMuVp2LjtlM/Osqbdl6rRp7txTnnnGMY2tXeZL6mvRX2Mr/sNj8+bNda/fOGy41zFhYWEYOvRqAEBeXh7y8/ObpbamkJoLkJtNaq767N+/H0u//RYAcFH//ujSpYvXcdffcKN73fPPykpS54y5nJULkJ3Nnz59+rrXC7dssbCSwEieL7tnE9/YrliRAwCIiIhAcnKyz3FZWX2OPCcnx/S6mkpqLkBuNqm56pObuwaHDx8GAGR6ZDtWp06d0CshAQCwcsWKZqmtPlLnjLmclQuQnc2fun87ACAoKMjCSgIjeb7sns3UxlYptVgppZVSj5p5HH825uUBAOLj4/3+pUhITHSv523MM72uppKaC5CbTWqu+uTnbXSvJ9Q2rr7Ubd9eVIQDBw6YWldDSJ0z5nJWLkB2Nn+WLfvOvZ6YmGRhJYGRPF92z2ZaY6uUug7A2WbtvyFcLheKi4sBAN26xfkdGxd3ZPv2oiJT62oqqbkAudmk5mqI7duPZKgve912rbXl2aXOGXM5KxcgO5s/5eXlmDZ1KgCgdevWGJydbXFFDSN5vpyQzZTGVikVDeB5AH8zY/8NVVpa6l6PiIjwO9Zze1lZmWk1GUFqLkBuNqm5GsKp2Z1ad32Yy1m5ANnZ/Jk8aSK2bt0KALj1ttvQtWtXiytqGMnz5YRsZp2xfQrAz1rrD03af4O4XC73euvWrf2ObdOmjXu93FVuWk1GkJoLkJtNaq6G8MweUk/21p7Zy63NLnXOmMtZuQDZ2Xz5ZNYsvPD88wCAXr164ZFHH7O4ooaTPF9OyBZs9A6VUpkAhsHiyxAAIDQ01L3ueQG6N4cOHXKvh4WGmVaTEaTmAuRmk5qrITyzV9ST/bBn9jBrs0udM+ZyVi5AdjZvVq5ciZEjaj5t365dO8z81yyEh4dbXFXDSZ4vJ2Qz9IytUqo1gFcAPKO1tvy+FW3btnWv1/dBFM/tkZGRptVkBKm5ALnZpOZqCKdmd2rd9WEuZ+UCZGc71oYNG5B9+SCUl5cjLCwMc+bNxxlnnGF1WQGRPF9OyGb0pQj3AggD0OD/M1BK3aKUylVK5e7du9fQYkJDQxETEwPg6A+weFPkcWFztzj/F0RbTWouQG42qbkawvMDBvVlr9uulELXbt1Mras+UueMuZyVC5CdzdPWrVsx8JKLsW/fPoSEhGDmx/9CZmam1WUFTPJ8OSGbYY2tUqo7gAkAJgFoo5Rqp5RqV7u57vfH3RdCaz1Da52qtU7t0KGDUeW4JSbV3B6koKAAVVVVPsdt8rh5cJIDbikiNRcgN5vUXPVJSDpyy5f6btK9adMmADX/CNrh7IXUOWMuZ+UCZGcDgF27duHSiwdgx44daNWqFd56+x1cOnCg1WU1muT5sns2I8/Y9gQQCuA9APs8FgC4u3b9TAOP1yDp6RkAaj6Rt379ep/jPO+Vl56RYXpdTSU1FyA3m9Rc9UlNPcf9IYOc5ct8jtuzZ4/7H8Lz0tObpbb6SJ0z5nJWLkB2tpKSElxy8QBsqf1msZde/ieuvuYai6tqGsnzZfdsRja2/wZwvpcFqGl2zwdQYODxGiQ7e4h7/Z233/I6xuVy4eOPZwIAkpKS6r2JvB1IzQXIzSY1V31OOOEE9O3XDwDw1ZdfYufOnV7HvffuO9BaAzj6z8pKUueMuZyVC5CbraysDJcPugwbfv4ZADDlmWfw57/8xeKqmk7qfAEOyKa1NnUBoAE82pCxKSkpurJKG76kZ2RoADokJESv/j73uO13jhmra+vUM2a8ZkoNzMVsdsnlqqhq0vL5V1+7a5wwaXKDnrNw0WL3c6648kpdfrjyqO0bN2/R7dq10wD0yT166NKDroDrkjxnzMVcdslWUVVt2FJ2sFxfcOGF7ponTppk6P4DWaTOl9TXYnR09GZfvaSqO0NiFqWUBvCY1npifWNTU1P16u9zDa8hNzcX/fpmweVyISoqCuMnTERmZhZKS0vx5huvY+bMjwAAvdPSsHTpMgQHG34XNFNIzQXIzWaHXJXV1QGNz1m+HFu2HPnPlk35+Xjm6SkAgMsHD8blg4/+NqBhw0d43c911wzFnNmzAQAX9e+P2+/4Kzp06Ij169biiccfw44dO6CUwqzZc3DZoMsDqhEAgluZc1tuO8yZGZjLWbkAe2TTMK5nGHrVVZg7dw4A4LLLBuHRxx/3Oz4iIgI9evQw7PieFJTh+7TDfJnF6mwdYtsXlJSUxHvdaPYZ20AWs87YVlZpPWvWbB0ZGen+CeLYJTk5We/Yudvyn4KYS342q3MFeib0hhuH+azV2+JrP8W/7dcXXnSRz+eFhIToqf+Y1ugzyZLnjLmYyy7ZjDxLGsi/KwB0n759HXXG1g7zJfW1aOkZ20CYdca2TmFhIaa++AIWL16EoqIihIeHo1dCAq699jqMGnUrQkJCTDu2maTmAuRmszJXoGds/3LTSLz37jsNHu+q8P0pWa013n/vXbz37rvY8PNP2L9/Pzp36YJ+/c7H7Xf8FWed3fjvdTHrjG0dvhadRWouwNpsRp6xDQkK7O9sn7598fWSbww7viczztjW4WvReP7O2LaoxpaIAm9sncLsxpaIjG1s7cTMxpaM56+x5TsBEREREYnAxpaIiIiIRGBjS0REREQisLElIiIiIhHY2BIRERGRCGxsiYiIiEgENrZEREREJAIbWyIiIiISgY0tEREREYnAxpaIiIiIRGBjS0REREQisLElIiIiIhHY2BIRERGRCGxsiYiIiEgENrZEREREJAIbWyIiIiISgY0tEREREYkQbHUBRNS8glvJ/Hn2wKFKq0swRUQb/jNN9qGsLoCoHjLf4YiIiIioxWFjS0REREQisLElIiIiIhHY2BIRERGRCGxsiYiIiEgENrZEREREJAIbWyIiIiISgY0tEREREYnAxpaIiIiIRGBjS0REREQisLElIiIiIhHY2BIRERGRCGxsiYiIiEgENrZEREREJEKLamwLCwsxbuwYJCX2QtvIcHTqGIvMzHRMnz4NFRUVVpfXaFJzAXKzMZe1SkpKMPPD93Hf3eNw8YV9cdZpvRDXqT06tItArx4nYsjll+LVf76EsrKyBu3vs4ULcM1VQ5DQszs6Rkfi9F49MXLY9chZvszkJE3nlDkLlNRcgKxsu3fvxoL58zFx4gQMGNAf7aPbITioFYKDWuGhhx60ujxDSJqvY9kym9baNktKSoqurNKmLLNmzdaRkZEagNclOTlZ79i527TjMxezMZe5uX4/WNHg5e33P/JZp+cSF3eiXvLdCp/7KSl16Wv/dL3P5yul9F333B9QbccukueMuZyYrdrQxd/fv0mTJxt+PN+L1PmS+VqMjo7e7KuXbBFnbHNzc3HDDX9CWVkZoqKi8ORTU7A8ZyUWLf4C11xzLQBg3bp1uOKKbFRWVlpcbcNJzQXIzcZc9sl1co+euHH4SDzz3Iv4YOYnWLI0B4u+/AavvvE2+g+4BEopFBX9D1cMHogd27d73cfkiQ/gow/eBwAkp6Tirfc+xJLvVuCfr76BU+N7QWuNZ59+Eq/+86XmjNYgTpyzhpCaC5CdDQB69uyJrD59rC7DMJLny9bZrD5L2xxnbNMzMjQAHRISoletXnPc9jvHjHX/hDFjxmuW/xTU0nNJzsZc5uUK5Cxo8f7yesc89cxz7ppH33HncdvXrP9JBwUFaQD63N5pes++sqO2b9u+R590cg8NQEdFRemtRbttdcbWDnPGXE7MZuyZ0omTJun5Cxbq3Xv26sqqav3V10vEnLG1x3zJfC36O2NreTNrdmO7PGflkTen0bd7HVNadlDHxMRoADopKcnyF0xLziU5G3OZm6sp/93vq/mt+2+2P/xf8nHbR/75Znfub5at9LqPt9770D3moUefsE1ja5c5Yy4nZjO3wZTS2NpnvmS+Flv0pQjz5s11r984bLjXMWFhYRg69GoAQF5eHvLz85ultqaQmguQm425nJUrODgYbUJDAQCHDx86apvWGos+XQgASEw6DckpqV73MejybERFRQEAFs6f63WMFaTOmdRcgOxsEkmeL7tnE9/YrliRAwCIiIhAcnKyz3FZWUeu61mRk2N6XU0lNRcgNxtzOSvXd0u/RfGvvwIA4nslHrVt69ZC7Nq1EwCQnpHpcx8hISFIPac3AGDd2lwcPnzYpGoDI3XOpOYCZGeTSPJ82T2b+MZ2Y14eACA+Ph5BQUE+xyUkHnnjytuYZ3pdTSU1FyA3G3PZP1dpaSnyN+bhqccfxfXXXOV+/Lbb7zhq3KaNG93rvRIS/O6zbntVVRW2FGw2sNrGkzRnnqTmAmRnk0jyfNk9W3CzHckCLpcLxcXFAIBu3eL8jo2LO7J9e1GRqXU1ldRcgNxszGXfXFNfeA6Txt/ndVtQUBCeevo5nJd+9FnZ7duP1N+1ntye27dv346k005vQrVNJ2HOvJGaC5CdTSLJ8+WEbKIb29LSUvd6RESE37Ge2xt6U3arSM0FyM3GXM7KBQBZffvh2eenIiEx6bhtnvVHhPvPHR4e7l4/cMD63FLnTGouQHY2iSTPlxOyiW5sXS6Xe71169Z+x7Zp08a9Xu4qN60mI0jNBcjNxlz2zXXDjcNxUf8BAIDygwexaVM+PvzgPSz9ZglG3HgdXpz2Ms7tfd5RzznkkTskgNyucutzS5gzb6TmAmRnk0jyfDkhm+hrbENrP9EMoN4PbRw6dORTz2GhYabVZASpuQC52ZjLvrnax8TgtNPPwGmnn4GUc87FddffiPmffo6/P/wY/rNhAwZd0h9ffr74qOe08chdEUDu0DDrc0uYM2+k5gJkZ5NI8nw5IZvoxrZt27bu9QMHDvgd67k9MjLStJqMIDUXIDcbczkrFwD87e57cc65vXHo0CGMueO2o749nKkEUAAAIABJREFUx7P+Awf95z548KB7PSLC+txS50xqLkB2Nokkz5cTsolubENDQxETEwPg6A97eFPkcWFztzj/F0RbTWouQG425nJWrjqXXnY5gJpsuWtWux/3/NDEjnpye27v1q2bwRUGTuqcSc0FyM4mkeT5ckI20Y0tACQm1Xzwo6CgAFVVVT7HbfK4eXCSlw+L2I3UXIDcbMzlrFwA3P+AA0DR//7nXu/lcRubTfXceHzzpk0Aau6wcMqp8QZX2DhS50xqLkB2Nokkz5fds4lvbNPTMwDUfCJv/fr1PsctW/bdkedkZJheV1NJzQXIzcZczsoFADt37HCve15G0KNHT3Tu3AUAsHKF7xuPV1RUYM33qwAAySmp9X7YorlInTOpuQDZ2SSSPF92zya+sc3OHuJef+ftt7yOcblc+PjjmQCApKQkJNRzw3U7kJoLkJuNuZyVq7q6Ggvmz3H/Pun0I/efVUrh0ssGAQDy/rMB69bmet3HwgXz8PvvvwMABg0e4nWMFaTOmdRcgOxsEkmeL7tnE9/YpqWluX9SePXVGVi7du1xY8aPf8B9w+Fx4+5q1voaS2ouQG425rJPrnfffhMVFRU+t1dXV2PShPux4eefAQDnpWfg5JN7HDVm9B13ur915767xx31CWAA2LdvHx6cNAEAcMIJJ2DYiJuMjNAkTpyzhpCaC5CdTSLJ82X3bEpr3awH9Cc1NVWv/t77mY+myM3NRb++WXC5XIiKisL4CRORmZmF0tJSvPnG65g58yMAQO+0NCxdugzBwc64va/UXIDcbMxlXq4DhyrrH1Sre5dYtGkTiuwrrkTvtPPQ/aSTEB4egd9//x0//rAeH7z3Ln768QcANZ8CXvzVtzjjzLOO28+EB+7FtBefBwCkpJ6DMX+7Gyee2B2b8jfimSlPYvOmmmvMnnnuRdx86+hG5YpoY85rwA5zZgapuQC7ZDO2Z1i+fDm2FBS4f78xfyOenjIFADB4cDays7OPGj98xAhDj3+EMnyP9pgvc1idrUNs+4KSkhLvH1rQWttmSUlJ0ZVV2pRl1qzZOjIyUqPmb+VxS3Jyst6xc7dpx2cuZmMuc3P9frCiwUtUVJTPOj2X+F4Jesl3K3zup6TUpa/90/U+n6+U0nfdc39AtR27SJ4z5nJitmpDl2HDhjfo72LdYvTxjyxS50vmazE6Onqzr16yRZyxrVNYWIipL76AxYsXoaioCOHh4eiVkIBrr70Oo0bdipCQENOObSapuQC52ZjLeIGcsd3w80/44vPFWLN6FbZu3YI9e/bgt337EB4ejk6du+Css/+AQYOzMejy7AZ94OuzhQvw1huv4d/r12HfvhJ06NAR56adh5tH3YaMzKymxDLtjG0dvhadx9psxvYMN40ciXfeebvB4yurqg09/hHGn7Gtw9ei8fydsW1RjS0RyRVIY+skZje2RIGxT89gLPMaWzKev8ZW/IfHiIiIiKhlYGNLRERERCKwsSUiIiIiEdjYEhEREZEIbGyJiIiISAQ2tkREREQkAhtbIiIiIhKBjS0RERERicDGloiIiIhEYGNLRERERCKwsSUiIiIiEdjYEhEREZEIbGyJiIiISAQ2tkREREQkAhtbIiIiIhKBjS0RERERicDGloiIiIhECLa6ACIiI0S0kfnP2X/3HbS6BNOcFB1udQkUMGV1AUR+8YwtEREREYnAxpaIiIiIRGBjS0REREQisLElIiIiIhHY2BIRERGRCGxsiYiIiEgENrZEREREJAIbWyIiIiISgY0tEREREYnAxpaIiIiIRGBjS0REREQisLElIiIiIhHY2BIRERGRCGxsiYiIiEgENrZEREREJEKLamwLCwsxbuwYJCX2QtvIcHTqGIvMzHRMnz4NFRUVVpfXaFJzAXKzMZfzODXbL9u24pnHHsTl55+HlPgTcVpcDLL+kIjrh1yK6c9NQdEv/z3uOevWrMYrU5/F7SOvxyWZqeh9Wg8kdmmHs0/ujEsyUzHhb3/Fj+vWWpCm4Zw6Xw0hNRtzOY8ts2mtbbOkpKToyiptyjJr1mwdGRmpAXhdkpOT9Y6du007PnMxG3M5M5fV2bb8eqDRy31/f1S3CQ31WTcAPfHRp457Xnxikt/n1C033HSL3ryntNH1SZwvya9F5mIuu2SLjo7e7KuXVFrrQHth06SmpurV3+cavt/c3Fz065sFl8uFqKgoPDB+AjIzs1BaWoq33nwDM2d+BADonZaGpUuXITg42PAazCA1FyA3G3M5Kxdgfbb/7jvYqOc989iDePn5pwEAp5/1Bwy9fhgSTzsD4RERKP51L35YtxafL5yLq667ESNH3X7Uc4dclIUToqLwf6m90fPUeHTo1BltT4jCr3t348d1a/Hh269j757dAIC/3D4GDzz0eKNqPCk6vFHP88fq+TKT1GzM5axcgPXZOsS2LygpKYn3utHqs7TNccY2PSNDA9AhISF61eo1x22/c8xY908YM2a8ZvlPQS09l+RszOWsXHbI1pgzoW/MnOOu6bZx9+iCvWU+x+bt2HfcY/m7fve7/7Wb/6e79+ipAejg4GCdu+kX25yxtXq+JL8WmYu57JLN3xlby5tZsxvb5Tkr3X+4o0ff7nVMadlBHRMTowHopKQky18wLTmX5GzM5axcdskWaLO4eU+p7n5yDw1An9//kiZdyuBvuXfyw+4/m1ffn2WLxtYO8yX5tchczGWXbP4aW/EfHps3b657/cZhw72OCQsLw9ChVwMA8vLykJ+f3yy1NYXUXIDcbMzlrFyAM7Mt/foL/LJtKwBg9N/uNe044RGR7vXDhw+ZdpxAOHG+GkpqNuZyVi7A/tnEN7YrVuQAACIiIpCcnOxzXFZWnyPPyckxva6mkpoLkJuNuZyVC3BmtkXz5wAA2sfEIvmc3u7Hf927B9sKt6C0dH+Tj1FdXY1F84+8ufWM79XkfRrBifPVUFKzMZezcgH2z2ZKY6uUGqiU+k6p/2/vzsObKtM3jt/HLnSllLKDC2ppo4DYVul0UdwARwGXQdAB3MWVZZwZFQG3cXdcEEZlcUP9CcMgKEpBHUUoawUUmbKUilpAQIraAqFteH9/lB4KJGnanNOT9+n9ua5cc2hOkvfLC87TkKRGuWEYvxuGUWAYxoV2PFZdNhQWAgCSk5MRFhbm87yU1FTzuHBDoe3rCpbULkBuG7v06gL0bKv5GK4urjOglMJbU15Br/Qz0dPVGRed2x09OrfH5b0yMfOdt3Do0KGA79fj8WDnjh34YmEehl51GVbkfwUAyD7/AnRJPcOWlvrScb8CJbWNXXp1AaHfZvlb8AzDGA5g4uHLY6gennsAsP7tr3Vwu93Ys2cPAKBjx05+z+3U6cj120pKbF1XsKR2AXLb2KVXF6Bn26FDh7Blc/U/+bVIbIk7b7gOCz/+8LjzCr9bhwdG3YnPF3yCl6dNR2RkpM/77HZyG+zft8/rdWd274FnJ06xZvFB0nG/AiW1jV16dQF6tFn6jK1hGKcAeBHA35RSo5VSnyqlFiilnlZKzbPysQJRVlZmHsfGxvo9t/b15eXltq3JClK7ALlt7NKrC9Czrez338xnYb/4NA8LP/4QnU9LxuR3/o1vvv8Z327dianv/QenJacAAD6bPw/PPjqu3o8TFR2NR595Ef+e/1+0bd/e0oaG0nG/AiW1jV16dQF6tFn9jO1NAA4BeNXi+20Qt9ttHvt7RgIAmjVrZh4fcB+wbU1WkNoFyG1jl15dgJ5t+/cf+czbg243Wrdpixkff4qkVq3Nr1/Quy/OSs/A5ednYufPOzB92mu46c4RaN+ho9f7/ODTr+DxeHDI48GeX3Zjef5ivPfGVDz58BhsKdqIBx55EhEREba31UXH/QqU1DZ26dUF6NFm9WtscwBsADDYMIwthmFUGYZRZBjGXXXd0A5RUVHmcUVFhd9zDx488q7e6Kho29ZkBaldgNw2dunVBejZVvv/SADgtntGHzXU1miZ1Ap3jP4bAKCyshIL5s31eZ+nd0lFiutMuLp2R06vi/DXBx/GJ4tXonXbdnhr8iu48ZoBqKqqsjakAXTcr0BJbWOXXl2AHm1WD7YdACQDeBbAUwB6A/gUwETDMEZ6u4FhGLcdfnNZwe7duy1dTHx8vHm8z8drxLxdHxcX5+dM50ntAuS2sUuvLkDPtti4+KN+nXPBRT7Pzb3gYvN43Zqv6/U47dp3wMNP/RMAsGzxIsyc/ma9bm8HHfcrUFLb2KVXF6BHm9WD7QkA4gEMV0pNUUr9Vyl1B4A8AA8YhmEcewOl1GSlVIZSKqN16+OfWQhGVFQUkpKSAADbtvl/4XJJrRc2d+zk/wXRTpPaBchtY5deXYCebc2aNUPLVq3MX7f38+aO2tft2fNLvR8rp9dFiI6pfk9w3rw5dZxtPx33K1BS29ilVxegR5vVg+2ew//76TFfXwigLYBGf5dBqssFACgqKoLH4/F53qZaHx7sSnXZvq5gSe0C5LaxS68uQM+25C5HHv+QnzXXvi48rP5vtwgLC0Pz5gkAgO0h8m5uHfcrUFLb2KVXFxD6bVYPtuvruD7wD020SFZWNoDqd+StWbPG53mLF3915DbZ2bavK1hSuwC5bezSqwvQs+2cPxx5/B+3bvV53o9bi83jtu071PtxKioqsLe0+rmMmDreHd1YdNyvQEltY5deXUDot1k92H5w+H/7HPP1vgBKlFI/W/x4dRow4Arz+O233vR6jtvtxsyZMwAALpcLKSkpjbG0oEjtAuS2sUuvLkDPtj6X9zePF35y/GfY1lgw78h1GZl/qPfjfJ73sfnmkS6u0PgBDTruV6CktrFLry4g9NusHmw/AfAFgNcMw7jdMIzehmFMQfWbyOr/YYkWyMzMNL9TmDJlMr7++vg3SYwZ84D5gcOjR9/bqOtrKKldgNw2dunVBejZdka3s5B7+E1jb742CZs2/O+4c4o2bcDUSS8BqP6EhL6XH/k/qlXL8r3eprYtmzfi0Qf+av76ymuutWLpQdNxvwIltY1denUBod9mKKWsvUPDaA7gSQB/ApCI6o//ekop9V5dt83IyFArVhZYuh4AKCgoQK/zc+F2u5GQkIAxD45FTk4uysrK8Mbr0zBjxvsAgJ6ZmVi0aDHCwy3/gWy2kNoFyG1jl15dgPNtP+zdX/dJxyjatAFX97kA5WW/o3lCC9x2zyhk5pwPAFi5dAlem/A8fvt1LwDghdfeQP+rrzFv+8arE/H4uPvRMzsXvS7uA1fXbmiR2BKHDilsL/kRS778Lz6Y+R7cB6o/l/LKQdfhuUkN++ljJyda/wMpnd4vO0ltY5deXYDzba1btSwqLS1N9nqlUipkLunp6arKo2y5zJo1W8XFxSkAXi9paWlq+46dtj0+u9jGLj27nG7b8su+Bl3enTNftUxq5XPN4eHh6qGn/nnc7cb+42mft6l9CQsLUzffcY/atPP3Bq9R4n5J/rPILnaFSltiYuJmX7Ok5c/YBsOuZ2xrFBcXY8JLLyIvbz5KSkoQExODLikpGDz4WgwffntI/PSchpDaBchtY5d+nGpryDO2NX7ZvQtvTXkFn+d9jG0//ogqTxXate+IrPN64cbhd+HU5C7H3WZfeTkWf/EZli1ehP999y127/wZv+zeBU9VFeITEnDa6Sk4JysbV15zHTqfdnowabY8Y1uDfxb1wy79ONXm7xnbJjXYEhHpJpjBNtTZOdgSkVz+Blur3zxGREREROQIDrZEREREJAIHWyIiIiISgYMtEREREYnAwZaIiIiIROBgS0REREQicLAlIiIiIhE42BIRERGRCBxsiYiIiEgEDrZEREREJAIHWyIiIiISgYMtEREREYnAwZaIiIiIROBgS0REREQicLAlIiIiIhE42BIRERGRCBxsiYiIiEiEcKcXQEREvp2cGOP0EmxT5Tnk9BJsER4m9zkjpZTTS7CFYRhOL4EsIvdvHxERERE1KRxsiYiIiEgEDrZEREREJAIHWyIiIiISgYMtEREREYnAwZaIiIiIROBgS0REREQicLAlIiIiIhE42BIRERGRCBxsiYiIiEgEDrZEREREJAIHWyIiIiISgYMtEREREYnAwZaIiIiIRGhSg21xcTFGjxoJV2oXxMfFoG2bVsjJycKkSRNRWVnp9PIaTGoXILeNXfqR2qZT186dOzHvow8xftxYXNq3N9q0SkRUZBiiIsPw2KOPBHQfVVVVWLNmNSa/9gpuuflGnNXtTEQ3CzfvRwc67Zk/paWlePfddzB69Eicf14uuiSfhpaJCYiNicKJnTrg0r598K9/TUJ5ebnTSw2KlP3yJiTblFIhc0lPT1dVHmXLZdas2SouLk4B8HpJS0tT23fstO3x2cU2dunZJbnN6S53hadeF1/rBKAeHDs+oPsYO+4hv/dT3zV5u0jes8qqQ5Zd3p8x0+9e1FxOPPFEtXTZCksf+9iL1P2S+mcxMTFxs69Zskk8Y1tQUIAhQ65DeXk5EhIS8NTTz2BJ/jLMz1uIQYMGAwBWr16NK68cgKqqKodXGzipXYDcNnbp1QXIbdO9q/OppyIn97x6304pZR7HxsYiJ/c8tGvXzsql2Ub3PfPm1FNPxY033YSXJryM/8z+APlLl+OLLxbhrbeno2/fS2EYBn766Sf88dI+2LZtm9PLrReJ+1UjpNucfpa2MZ6xzcrOVgBURESEWr5i1XHXjxg5yvwOY/LkqY5/F9TUuyS3sUuvLsltodBV32dCxzw4Ts2Z+5HatmOXcld41IJPP6/3M7YfzftETXrlVbVy1Wq170CFcld4VO5552vxjG0o7JmVz5IecFfUec7zL7xoNo0YMVKrZ2xDYb+k/ln094yt48Os3YPtkvxl5m/unXfe5fWcsvL9KikpSQFQLpfL8T8wTblLchu79OqS3BYqXcEOkA0ZbL1ddBhsQ2XP7Hw5gK/ht+afu9PS07UZbENlv6T+WWzSL0WYO3eOeTx02PVez4mOjsbAgdcAAAoLC7Fx48ZGWVswpHYBctvYpVcXILdNapdkTXXPwsPDERUVBQA4ePCgw6sJnOT9CvU28YPt0qX5AKpfS5WWlubzvNxar9Vamp9v+7qCJbULkNvGLr26ALltUrska6p79uUXX+CXX34BAKSmpDq8msBJ3q9QbwtvtEdyyIbCQgBAcnIywsJ8f5RLSuqRvzCFGwptX1ewpHYBctvYpVcXILdNapdkTWnPysrKUFJSgv/M+jdeeOF58+t33zPCwVXVj+T9CvU20YOt2+3Gnj17AAAdO3bye26nTkeu31ZSYuu6giW1C5Dbxi69ugC5bVK7JGsKe/b8P5/Dfff93et1YWFheP6FF5GTk9PIq2oYyfulQ5vowbasrMw8jo2N9Xtu7etD/cOgpXYBctvYpVcXILdNapdkTXnPevW6ABNengiXy+X0UgImeb90aBP9Glu3220eR0ZG+j23WbNm5vEB9wHb1mQFqV2A3DZ26dUFyG2T2iVZU9iz62+4EWvWfos1a79F/tLleP2NN3HhhRfhyy+/wLXXDsKyZcucXmLAJO+XDm2iB9uad1ICQEVFhd9za7/bMjoq2rY1WUFqFyC3jV16dQFy26R2SdYU9iwpKQldu3ZF165dce6552Lo0GFYsPBTPP74E1j/3Xe45OILkTd/vtPLDIjk/dKhTfRgGx8fbx7v27fP77m1r4+Li7NtTVaQ2gXIbWOXXl2A3DapXZI15T37+333o2fPTBw8eBB33DFci5/QJXm/dGgTPdhGRUUhKSkJALBtm/8XLpfUemFzx07+XxDtNKldgNw2dunVBchtk9olWVPfs379+gGobluxYoXDq6mb5P3SoU30YAsAqYdfcF5UVASPx+PzvE21PjzYlRr6L1KX2gXIbWOXXl2A3DapXZI15T1LatXKPP7ppx8dXEngJO9XqLeJH2yzsrIBVL8jb82aNT7PW7z4qyO3yc62fV3BktoFyG1jl15dgNw2qV2SNeU9275tm3kcFxv6/1wPyN6vUG8TP9gOGHCFefz2W296PcftdmPmzBkAAJfLhZSUlMZYWlCkdgFy29ilVxcgt01ql2RNdc8OHTqED+Z8YP76zK5dHVxN4CTvV6i3iR9sMzMzze8UpkyZjK+//vq4c8aMecD8wOHRo+9t1PU1lNQuQG4bu/TqAuS2Se2STOKevfHG66isrPR5/aFDh3D/fX/Hd+vWAQCys3PQuXPnxlpeUCTuV41QbzOUUo36gP5kZGSoFSsLLL/fgoIC9Do/F263GwkJCRjz4Fjk5OSirKwMb7w+DTNmvA8A6JmZiUWLFiM8XI+fWyG1C5Dbxi69ugC5baHQVeU5VK/z8/OXYMuWIvPXmzZsxHPPPQMA6NevP/oNGHDU+cOG3XDcfZSXl2P27FlHfe25Z57Bpk3VrwecPHXaUded1b0HzurRo17rDA+z5zmjUNgzK2eGVkmJiIqKwlVXX40//CELJ598CmJjY/Hbr79izdo1mD79bXyzdi2A6nfjf7loMbp3727Z49dmGIbl9xkK+2UXp9tat2pZVFpamuz1SqVUyFzS09NVlUfZcpk1a7aKi4tTALxe0tLS1PYdO217fHaxjV16dkluc7rLXeGp12XI0GE+1+rt4u0+NmzaUq/7eHDs+HqvU/KeVVYdsuySkJAQ0B6kpKSopctWWPrYx16k7pfUP4uJiYmbfc2STeIZ2xrFxcWY8NKLyMubj5KSEsTExKBLSgoGD74Ww4ffjoiICNse205SuwC5bezSj9Q2J7vq+4ztLTffiHemvx3w+e6K49+xvXXrVqR2OS3g+3hw7HiMG/9QwOcD9j1jW8PJPbNyZli3bh3y5n+C5cuXY0vxFuzauRN79+5FTEwM2rdvjx49zsaAK67AgAFX1PlTroJlxzO2NaT+twNwrs3fM7ZNarAlIqLQUd/BVhd2D7ZOCqWZwUp2DrZkPX+Drdy/fURERETUpHCwJSIiIiIRONgSERERkQgcbImIiIhIBA62RERERCQCB1siIiIiEoGDLRERERGJwMGWiIiIiETgYEtEREREInCwJSIiIiIRONgSERERkQgcbImIiIhIBA62RERERCQCB1siIiIiEoGDLRERERGJwMGWiIiIiETgYEtEREREIoQ7vQAiImqawsNkPrdyoMLj9BJsEx0Z5vQSiPyS+V8VIiIiImpyONgSERERkQgcbImIiIhIBA62RERERCQCB1siIiIiEoGDLRERERGJwMGWiIiIiETgYEtEREREInCwJSIiIiIRONgSERERkQgcbImIiIhIBA62RERERCQCB1siIiIiEoGDLRERERGJ0KQG2+LiYoweNRKu1C6Ij4tB2zatkJOThUmTJqKystLp5TWY1C5Abhu79CO1jV3O+7pgFZ58/FH0+2NvJHc+ES2bR6N96xZI79EVI+6+A2vXrG7wfa//bh0S46MQHx2O+OhwDL/1JgtXbi2d9qw+pHYBIdqmlAqZS3p6uqryKFsus2bNVnFxcQqA10taWpravmOnbY/PLraxS88uyW3ssufxyw5UBXzJys7xuc6ai2EY6u4Ro9Tv+yvrdd+/7atQ55zb86j7um7IsHrdx7EXqXvGLr3aEhMTN/uaJZvEM7YFBQUYMuQ6lJeXIyEhAU89/QyW5C/D/LyFGDRoMABg9erVuPLKAaiqqnJ4tYGT2gXIbWOXXl2A3DZ2hUbXzzt2AAA6duyEu+4ZiXffn4VFS5Zj4eeLMP6Rx9AyKQlKKUyc8CIeGjemXvc9bcprWLVyBVq3aWPH0i2j254FSmoXEOJtTj9L2xjP2GZlZysAKiIiQi1fseq460eMHGV+hzF58lTHvwtq6l2S29ilV5fkNnbZ11WfZ0D7/vEyNf3dGerX8oNer19XuFm1adtWAVDh4eHqm/UbA7rfTVt+VM2bN1cA1GtT3wjpZ2xDYc/YpVebv2dsHR9m7R5sl+QvM39z77zzLq/nlJXvV0lJSQqAcrlcjv+BacpdktvYpVeX5DZ22dsVzODo7fL8iy+bXU889WxAtxlwxVXmIPvdhqKQHWxDZc/YpVdbk34pwty5c8zjocOu93pOdHQ0Bg68BgBQWFiIjRs3NsragiG1C5Dbxi69ugC5bezSqysn9zzz+Pvvi+s8f/4n8zB3zmwkJibiH088befSgiZ1z6R2AaHfJn6wXbo0HwAQGxuLtLQ0n+fl1voPx9L8fNvXFSypXYDcNnbp1QXIbWOXXl0VFRXmcdgJYX7P3bdvH+4dNQIA8PBjT6B169a2ri1YUvdMahcQ+m2WDraGYWQbhrHQMIxdhmGUGYax2jAMRz9bZENhIQAgOTkZYWG+/4OQkppqHhduKLR9XcGS2gXIbWOXXl2A3DZ26dWVv2SxeVx77d489sh4/PTTjzjn3J648aZb7F5a0KTumdQuIPTbLBtsDcPoDuAzABEAbgVwFYBVAKYZhnGHVY9TH263G3v27AFQ/Y5Tfzp1OnL9tpISW9cVLKldgNw2dunVBchtY5deXQcOHMArk14GAERGRuKyy/v7PHftmtV49V8TERYWhhcnTIJhGI21zAaRumdSuwA92sItvK/BAMIA9FNKlR/+2qeHB95hAF6x8LECUlZWZh7Hxsb6Pbf29eXl5X7OdJ7ULkBuG7v06gLktrFLr65HHx6HrVu/BwDcetvtaN+hg9fzPB4P7rnrdng8Htx19wh0P6tHYy6zQaTumdQuQI82K1+KEAmgEsCBY77+m8WPEzC3220eR0ZG+j23WbNm5vEB97EJoUVqFyC3jV16dQFy29ilT9ec2f/BxAkvAgBOT+6C8Y/8w+e5r0x6GWvXrEb79h3w4PhHGmuJQZG4Z4DcLkCPNisHzjcP/+8EwzA6GIbRwjCMWwFcBOAFCx8nYFFRUeZx7Rffe3Pw4EHzODoq2rY1WUEY40KlAAAgAElEQVRqFyC3jV16dQFy29ilR9eK5ctw2y03AABatGiBd/9vJmJiYryeW/LTT3j8sYcBAE8/+zzi4+MbaZXBkbZnNaR2AXq0WfZSBKXUd4Zh9ALwAYA7D3+5EsDtSqn3rXqc+qj9l3vfvn1+z619fVxcnG1rsoLULkBuG7v06gLktrEr9LsK/7ceA6/qjwMHDiA6OhozZs3BGWd29Xn+X0bfg/Lyclx8SW9cefWfGnGlwZG0Z7VJ7QL0aLNssDUMIxnAfwCsB3A7ql+SMADAq4ZhuJVS7/q43W0AbgOAk046yarlAKj+ziIpKQl79uzBtm3+X7hcUuuFzR07+X9BtNOkdgFy29ilVxcgt41dod21dev3GHD5pdi7dy8iIiIw/b0ZyMrO8Xn+hg2FmP/xPADAH7KyMWvmjOPO+eWX3ebxD1u/N8856+yzkZzcxeKCwEnZs2NJ7QL0aLPypQhPoPoZ2suVUvOUUp8rpUYAmAngJcMwvD6WUmqyUipDKZVhx+ftpbpcAICioiJ4PB6f522q9eHBrlSX5euwmtQuQG4bu/TqAuS2sSs0u3b+/DMGXNYXO3ZsxwknnIAp095Cn75/9Hubilr/3PvYIw/hxuv/fNzlb/eOMs/JX7LY/PrCvPm2tQRK9z3zRWoXEPptVg623QB8o5SqPObrKwEkAWhj4WMFLCsrG0D1O/LWrFnj87zFi786cpvsbNvXFSypXYDcNnbp1QXIbWNX6HWVlpai/+V9UVy8BQDw0sRXcPXhn9wkmc575o/ULiD026wcbH8G0MMwjGPfJtcTgBtAqYWPFbABA64wj99+602v57jdbsw8/E8zLpcLKSkpjbG0oEjtAuS2sUuvLkBuG7tCq6u8vBxXX3E5/rf+OwDAE089ixtuvDmg23Y/qwfKDlT5vXy3ocg8/7ohw8yv33XPSFt66kPXPauL1C5AgzallCUXAH8CoAAsQPVra3sDmHj4a88Hch/p6emqyqMsv2RlZysAKiIiQq1YWXDc9SNGjlKH16kmT55qyxrYxTZ26dcluY1d9nWVHagK+PLLr/tUrwsuNNd0/5ix9bp9IJfvNhSZ93/dkGFB3ZfUPWOXXm2JiYmbfc6jVg22h4fbSwF8CWA3gDIAa1H9CQlhTg62y1esUlFRUQqASkhIUE8/86zKX7pc5S34VA0aNNj8ze+ZmancBysd/wPT1Lskt7FLry7Jbeyyr6s+g2K//leYa+r7x8vUioK1fi/rCjeLG2xDYc/YpVdbow22wV7sGmyrPErNmjVbxcXFmb/Zx17S0tLU9h07Hf/Dwi75bexyfq1sY5edXfUZFH2t0dclJ/c8cYNtKOwZu/Rq8zfYGoefaQ0JGRkZasXKAtvuv7i4GBNeehF5efNRUlKCmJgYdElJweDB12L48NsRERFh22PbSWoXILeNXfqR2sYu6x2o8P1O8WPFR9fvUzdzcs/D/IX/rddtfvhhK7qmng6g+jW2r015vV63ry06MqzBt60L/yzqx6m21q1aFpWWliZ7u65JDbZERER2q89gqxs7B1uiQPkbbK38VAQiIiIiIsdwsCUiIiIiETjYEhEREZEIHGyJiIiISAQOtkREREQkAgdbIiIiIhKBgy0RERERicDBloiIiIhE4GBLRERERCJwsCUiIiIiETjYEhEREZEIHGyJiIiISAQOtkREREQkAgdbIiIiIhKBgy0RERERicDBloiIiIhE4GBLRERERCJwsCUiIiIiEcKdXgAREZEk0ZFhTi/BNguWrHN6Cbbok9PN6SWQRfiMLRERERGJwMGWiIiIiETgYEtEREREInCwJSIiIiIRONgSERERkQgcbImIiIhIBA62RERERCQCB1siIiIiEoGDLRERERGJwMGWiIiIiETgYEtEREREInCwJSIiIiIRONgSERERkQgcbImIiIhIhHCnF9CYiouL8fKEl5CXNx8lJSWIiYlBcpcuuPba63DbbcMRERHh9BIbRGoXILeNXfqR2sYu/YRK297SPdhY+C02/m8dNhauw+YN67F/XzkA4Lobbsefb7yzzvt4941/4b03Xw3o8Z58cRq6n32O1+tuHNQXu37eXud9tGnXAW/MyAvo8awSKvtlh1BsM5RSjf6gvmRkZKgVKwtsue85H3yAG24YhvLycq/Xp6WlYd7H89GmTRtbHt8uUrsAuW3s0qsLkNvGLr26AOfbFixZZx5fdn53n+fpNtj2yekW0Brqy+n9spOTba1btSwqLS1N9nZdkxhsCwoK0Ov8XLjdbiQkJOCBMQ8iJycXZWVlePON1zFjxvsAgJ6ZmVi0aDHCw/V4IltqFyC3jV16dQFy29ilVxcQGm2+Btt2HTqhVeu2+O6brwE0bLCd9MZ//J7brn1HREXHeL2uZrDNzLkAQ2++2+d9REREoOOJpxz3dTsG21DYL7s43eZvsNXndzEIf/nLKLjdbkRERGDBws+QkZFhXnfJJZegbbt2mPDSi1ixfDnefust3HTzzQ6uNnBSuwC5bezSqwuQ28YuvbqA0Gu79vrhSDmjO1Jc3dA8oQW+XbMKD4xq+GOecqrXOaVeYuPiLbkfK4TaflkplNvEv3ls+fLlWJqfDwC49dbbjvrNr/H4408gKSkJAPDCC/9s1PU1lNQuQG4bu/TqAuS2sUuvLiA024bcdBfOycxF84QWtj+WbkJxv6wS6m3iB9u5c+eYx0OHXe/1nOjoaAwceA0AoLCwEBs3bmyUtQVDahcgt41denUBctvYpVcXILtNIsn7Fept4gfbpUurv6uIjY1FWlqaz/Nyc887cpvD34mEMqldgNw2dunVBchtY5deXYDsNokk71eot4l/je2GwkIAQHJyMsLCwnyel5Kaah4Xbii0fV3BktoFyG1jl15dgNw2dunVBchuqzH23uEoLtqAfeVliI2Lx0mnnIa0c7Nxaf+BiI9vHtB9fPfN17j75oH4edtPOHToEFoktkQXVzecf9GlyMy5AIZh2FxRTfJ+hXqb6MHW7XZjz549AICOHTv5PbdTpyPXbyspsXVdwZLaBchtY5deXYDcNnbp1QXIbqttTcEy8/i3X/di3doCrFtbgFnvvY7RD/wDf8i5oM772Llj29G//nk7dv68HYu/WIAzup2N+x56Bq1at7V87bVJ3i8d2kQPtmVlZeZxbGys33NrX+/rM9lChdQuQG4bu/TqAuS2sUuvLkB2G1D9aQiZORcixdUVLVu1RlVVFbb9uBVffvYJVq9ain3lZXhi/F/w0BMTkJGZ6/U+wsMj0DO7F9LOycLJnU9HTGwc9pWXYcP6b/DJ3JnYvetn/G/dGoy9dzj++a/piI2Lt61H8n7p0CZ6sHW73eZxZGSk33ObNWtmHh9wH7BtTVaQ2gXIbWOXXl2A3DZ26dUFyG4b8KchXj/vNvWM7riob3/M//DfmPjPx3DI48GEZx/B1Pc+RmStxhovvPou4ry8XKH72efg8quuxZPj78XqVUvx0w/FeO/NV3Hr3X+zpQeQvV86tIl+81hUVJR5XFFR4ffcgwcPmsfRUdG2rckKUrsAuW3s0qsLkNvGLr26ANlt3obR2i7tPxC9L7sSALDnl13I/+qzet9PTEws7n/4WcQ3TwAA5M2bhcrKygauuG6S90uHNtGDbXz8kX9q2Ldvn99za18fFxdn25qsILULkNvGLr26ALlt7NKrC5DdFohL+w00j9etbdhPJ42Ni8d5F/YFALgPHMDmjestWZs3kvdLhzbRg21UVJT5AcHbtvl/4XJJrRc2d+zk/wXRTpPaBchtY5deXYDcNnbp1QXIbgvESaecah7v+WVXEPdz2pH72d3w+6mL5P3SoU30YAsAqS4XAKCoqAgej8fneZtqfXiwK9Vl+7qCJbULkNvGLr26ALlt7NKrC5DdVqdG+oguK0ner1BvEz/YZmVlA6h+R96aNWt8nrd48VdHbpOdbfu6giW1C5Dbxi69ugC5bezSqwuQ3VaXn7YWm8dJSa0bfj8/1LqfVg2/n0BI3q9QbxM/2A4YcIV5/PZbb3o9x+12Y+bMGQAAl8uFlJSUxlhaUKR2AXLb2KVXFyC3jV16dQGy2+oy/8N/m8dde2Q06D72lZfhq//mAQCaRUUhOeVMS9bmi+T9CvU28YNtZmam+Z3ClCmT8fXXXx93zpgxD5gfODx69L2Nur6GktoFyG1jl15dgNw2dunVBchs27plE7aX/Oj3nPkfzsKCj2cDABJbtkJW7oXHnVOwYgkOHnQf9/UaB/bvx1MP/w2///YrAKD3H69CRB0fVRUsiftVI9TbDKVUoz6gPxkZGWrFyoa949GfgoIC9Do/F263GwkJCRjz4Fjk5OSirKwMb7w+DTNmvA8A6JmZiUWLFiM8XI+P95XaBchtY5deXYDcNnbp1QWERtuCJevM4/Xfrsb2bT+Zvy758XvMeu91AEBmzgXIzDl6CL3k0gFH/frT+XMx4dmH0b3HOcjomYNTTk1GfEICPB4PSn743vwBDQBwQlgYxj72Anpm9zpuTfePvAlbizcjK/cinNH9bLTvcCKiomOwf185Cr9bi08+/Dd279wBAOh00il4btJ086O/avTJ6dbw3xQfQmG/7OJ0W+tWLYtKS0uTvV3XJAZbAJjzwQe44YZhPn/6RVpaGuZ9PB9t2rSx5fHtIrULkNvGLr26ALlt7NKrC3C+rfZg+/yTY/F53ocB3/bjRd8e9etP58/Fi0+Nq/N2zRNaYMTfH/H5I3XvH3lTQB8D1q1HBv469kmvP1LXjsEWcH6/7ORkm7/BNuzhhx+2/AEbavLkyQ/feuttttx3qsuFawYNhlIKpaWl2L9/P5o3b46zevTAX//2d/zrlVfRvLn/D4oORVK7ALlt7NKP1DZ26cfpti0/HvmYrGVL/ovvizb6Oftof77xjqN+ndSqNU486VS0aNESxgkGTjjhBHg8HoSFhaFFYhLO7HY2+l11LUbf/xhOPd33azRP7nw62nU8EVHR0QAMAAaqqqoQHRODtu074pw/nIcbh4/C0JvvRmys989TPf2k44ddKzi9X3Zysu3ZZ54uve+++172dl2TecaWiIiIglP7GVtJ7HrGluzh7xlb8W8eIyIiIqKmgYMtEREREYnAwZaIiIiIROBgS0REREQicLAlIiIiIhE42BIRERGRCBxsiYiIiEgEDrZEREREJAIHWyIiIiISgYMtEREREYnAwZaIiIiIROBgS0REREQicLAlIiIiIhE42BIRERGRCBxsiYiIiEgEDrZEREREJAIHWyIiIiISIdzpBRAREZEe+uR0c3oJtqio8ji9BNtEhoc5vYRGxWdsiYiIiEgEDrZEREREJAIHWyIiIiISgYMtEREREYnAwZaIiIiIROBgS0REREQicLAlIiIiIhE42BIRERGRCBxsiYiIiEgEDrZEREREJAIHWyIiIiISgYMtEREREYnAwZaIiIiIROBgS0REREQiNKnBtri4GKNHjYQrtQvi42LQtk0r5ORkYdKkiaisrHR6eQ0mtQuQ28Yu/UhtY5d+pLbp1LVz507M++gjPDR+LC7r2xvtWrdETLNwxDQLxz8eeySg+/jHY4+Yt6nr8tWiL+0NaqCQ3DOlVMhc0tPTVZVH2XKZNWu2iouLUwC8XtLS0tT2HTtte3x2sY1denZJbmOX82tlW2h07T9YVa+Lr3UCUGPGjgvoPsaMHef3fmpf8hZ+Vu811lwk7lliYuJmX7Nkk3jGtqCgAEOGXIfy8nIkJCTgqaefwZL8ZZiftxCDBg0GAKxevRpXXjkAVVVVDq82cFK7ALlt7NKrC5Dbxi69ugC5bbp3de58KnJyc4O6j1Wr1/q9pGecY9FqrRHSe+b0s7SN8YxtVna2AqAiIiLU8hWrjrt+xMhR5ncYkydPdfw716beJbmNXXp1SW5jl15dkttCoau+z4I+8OBYNXvOh+qn7TvV/oNVKm/hZ0E9Y9vQZ2OdesbW6T3z94yt48Os3YPtkvxl5m/unXfe5fWcsvL9KikpSQFQLpfL8b/kTblLchu79OqS3MYuvbokt4VKV7DDY1MabENhz5r0SxHmzp1jHg8ddr3Xc6KjozFw4DUAgMLCQmzcuLFR1hYMqV2A3DZ26dUFyG1jl15dgNw2qV2ShfqeiR9sly7NBwDExsYiLS3N53m5uecduU1+vu3rCpbULkBuG7v06gLktrFLry5AbpvULslCfc8CGmwNw+hkGMbLhmEsMwxjv2EYyjCMU7ycF2UYxrOGYewwDOPA4fPPO/4eG8+GwkIAQHJyMsLCwnyel5Kaah4Xbii0fV3BktoFyG1jl15dgNw2dunVBchtk9pVX/3+2Bcnd2qPhLhonNypPfpcciGee/Zp7N271+mlHSfU9yzQZ2xPB3ANgL0AFvs5bxqAWwGMB3A5gB0AFhiG0SOYRTaU2+3Gnj17AAAdO3bye26nTkeu31ZSYuu6giW1C5Dbxi69ugC5bezSqwuQ2ya1qyE+//wz7N69G5WVldi9ezcWf/UVxo99EGemJuOjDz90enkmHfYsPMDzvlJKtQUAwzBuAdD72BMMwzgLwHUAblJKvXH4a4sArAfwKID+lqy4HsrKyszj2NhYv+fWvr68vNy2NVlBahcgt41denUBctvYpVcXILdNald9dO3aDZf374+MjHPQvkMHVFZWYvOmTZjxf+/hs88+xa+//orrBg/ErNlz0KfvpU4vV4s9C2iwVUodCuC0/gAqAcyodbsqwzDeB3C/YRjNlFIHG7bMhnG73eZxZGSk33ObNWtmHh9wH7BtTVaQ2gXIbWOXXl2A3DZ26dUFyG2T2hWou+8ZibHjHjru6+ee2xN/HjIUU6dMxoi774TH48GddwzH+sJNiIqKcmClR+iwZ1a+eexMAN8rpfYf8/X1ACJR/XKGRlX7D0BFRYXfcw8ePDJzR0dF27YmK0jtAuS2sUuvLkBuG7v06gLktkntClSLFi38Xn/Lrbfh+htvAgDs2L4dcz6Y3RjL8kuHPbNysG2J6tfgHqu01vWNKj4+3jzet2+f33NrXx8XF2fbmqwgtQuQ28YuvboAuW3s0qsLkNsmtctKN99yq3m8+KuvHFxJNR32zPGP+zIM4zbDMAoMwyjYvXu3pfcdFRWFpKQkAMC2bf5fuFxS64XNHTv5f0G006R2AXLb2KVXFyC3jV16dQFy26R2WcnlOsM83r59m4MrqabDnlk52O4FkOjl6zXP1JZ6uQ5KqclKqQylVEbr1q0tXE61VJcLAFBUVASPx+PzvE21PjzYleqyfB1Wk9oFyG1jl15dgNw2dunVBchtk9plFcMwnF7CcUJ9z6wcbNcD6GwYRswxXz8DQAWAIgsfK2BZWdkAqt+Rt2bNGp/nLV585Cn+rOxs29cVLKldgNw2dunVBchtY5deXYDcNqldViks/J953L5DBwdXckSo75mVg+1HACIADKz5gmEY4QAGAVjY2J+IUGPAgCvM47ffetPrOW63GzNnVn+Yg8vlQkpKSmMsLShSuwC5bezSqwuQ28YuvboAuW1Su6wybeoU87j2T/JyUsjvmVIqoAuAPx2+vAJAAbjj8K/Pr3XO+6h+ScItAC4CMAuAG0BaII+Rnp6uqjzK8ktWdrYCoCIiItSKlQXHXT9i5Ch1uElNnjzVljWwi23s0q9Lchu79OqS3BYKXfsPVgV1yVv4mbnGMWPH1Xn+yq/XqHXrN/g95+VJr5j32bZdO7W79LcGrU3iniUmJm72NUsahwfSOhmG4evERUqpXofPiQbwOKp/UEMLAN8AuE8p9WUgj5GRkaFWrCwIaD31UVBQgF7n58LtdiMhIQFjHhyLnJxclJWV4Y3Xp2HGjPcBAD0zM7Fo0WKEhwf6cyucJbULkNvGLr26ALlt7NKrC5DbFgpdFVW+XyvqzdL8JdiyZYv5600bN+Cfzz0LALi8X3/06z/gqPOHDrv+qF9Pf/st3Hn7bTj//F7o3acvzuzaFS2TklBVVYVNGzeaP6ABAMLCwvD+zFm47PJ+DUlDZLjvH3vbUE7vWetWLYtKS0uTvV4Z6DO2jXGx6xnbKo9Ss2bNVnFxceZ3EMde0tLS1PYdOx3/zpVd8tvY5fxa2cYuHbsktzndVd9nQYcMHeZzrd4ux97+tSnTArpdUlKSmvHv2UE9myxxzyx5xrYx2PWMbY3i4mJMeOlF5OXNR0lJCWJiYtAlJQWDB1+L4cNvR0REhG2PbSepXYDcNnbpR2obu/Qjtc3Jrvo+Y3vbLTfhnelvB3z+/oNVR/16165dmP/Jx1ixYjm+WbsWu3btROmePVBKIbFlS3Tr1h29e/fBkGHXo3nz5vVa27HseMa2hlN75u8Z2yY12BIREREdq76DrU7sHGyd4m+wdfwHNBARERERWYGDLRERERGJwMGWiIiIiETgYEtEREREInCwJSIiIiIRONgSERERkQgcbImIiIhIBA62RERERCQCB1siIiIiEoGDLRERERGJwMGWiIiIiETgYEtEREREInCwJSIiIiIRONgSERERkQgcbImIiIhIBA62RERERCQCB1siIiIiEiHc6QUQEREROSkyPMzpJdim6tAhp5dgOeXnOj5jS0REREQicLAlIiIiIhE42BIRERGRCBxsiYiIiEgEDrZEREREJAIHWyIiIiISgYMtEREREYnAwZaIiIiIROBgS0REREQicLAlIiIiIhE42BIRERGRCBxsiYiIiEgEDrZEREREJAIHWyIiIiISoUkNtsXFxRg9aiRcqV0QHxeDtm1aIScnC5MmTURlZaXTy2swqV2A3DZ26UdqG7v0I7WNXc7buXMn5n30IcaPG4tL+/RGm6REREWEISoiDI89+ki97ksphXemv42+vS/BSR3bIyEuBinJp+H2227Ft998Y1PB4QcOlUt6erqq8ihbLrNmzVZxcXEKgNdLWlqa2r5jp22Pzy62sUvPLslt7HJ+rWxjl91d7kpPwBdf6wSgHhw3PuD7Kf2tTF108cU+7ysiIkJNeHlivdZW+9IiMXGzr1nS8WG2MQbb5StWqaioKAVAJSQkqKeefkYtyV+m5uctVIMGDTZ/o3tmZir3wUrH/yI09S7JbezSq0tyG7v06pLcxi57uxo62HY+9VSVk3tegwbbq66+2rzdxZdcombPmauWLF2uJrw8UbVv314BUIZhqNlz5nKwbcglKzvb/A5h+YpVx10/YuQocwMmT57q+F+Gpt4luY1denVJbmOXXl2S29hlb1d9BsYxY8epOR9+pLb9vEu5Kz1qwWef13uwnTc/z7zNlVddpfYfrDzq+g2bt6gWLVooAOqUzp1V2X43B9v6XJbkLzN/g++88y6v55SV71dJSUkKgHK5XI7/ZWjKXZLb2KVXl+Q2dunVJbmNXfZ3NfSf+xs62F7Su7cCoMLDw9X3P5Z4PeeJp5427/ed9/7P0sFW/JvH5s6dYx4PHXa913Oio6MxcOA1AIDCwkJs3LixUdYWDKldgNw2dunVBchtY5deXYDcNnbp1VWX33//HYu+/BIAcPEll6B9+/Zez/vzkKHmce3fKyuIH2yXLs0HAMTGxiItLc3nebm55x25TX6+7esKltQuQG4bu/TqAuS2sUuvLkBuG7v06qpLQcEqVFRUAAByarUdq23btuiSkgIAWLZ0qaVrED/YbigsBAAkJycjLCzM53kpqanmceGGQtvXFSypXYDcNnbp1QXIbWOXXl2A3DZ26dVVl42FG8zjlMODqy81128rKcG+ffssW4PowdbtdmPPnj0AgI4dO/k9t1OnI9dvKymxdV3BktoFyG1jl15dgNw2dunVBchtY5deXYHYtu1IQ13tNdcrpSxtFz3YlpWVmcexsbF+z619fXl5uW1rsoLULkBuG7v06gLktrFLry5Abhu79OoKRCi0ix5s3W63eRwZGen33GbNmpnHB9wHbFuTFaR2AXLb2KVXFyC3jV16dQFy29ilV1cgardH1NEeWbv9gHXtogfbqKgo87jmxcy+HDx40DyOjoq2bU1WkNoFyG1jl15dgNw2dunVBchtY5deXYGo3V5ZR3tF7fZo69pFD7bx8fHmcV0vTK59fVxcnG1rsoLULkBuG7v06gLktrFLry5Abhu79OoKRCi0ix5so6KikJSUBODoFzR7U1LrhcsdO/l/wbPTpHYBctvYpVcXILeNXXp1AXLb2KVXVyBqv2Gsrvaa6w3DQIeOHS1bg+jBFgBSXS4AQFFRETwej8/zNtX6YGRXqsv2dQVLahcgt41denUBctvYpVcXILeNXXp11SXFdeTjy+r6gRObNm0CUD3Q8xnbesjKygZQ/Y67NWvW+Dxv8eKvjtwmO9v2dQVLahcgt41denUBctvYpVcXILeNXXp11SUj4xzzDXP5Sxb7PG/Xrl3mUP+HrCxL1yB+sB0w4Arz+O233vR6jtvtxsyZMwAALperzg8VDgVSuwC5bezSqwuQ28YuvboAuW3s0qurLs2bN8f5vXoBAD779FPs2LHD63nvTH8bSikAR/9eWUIpFTKX9PR0VeVRll+ysrMVABUREaFWrCw47voRI0cpAAqAmjx5qi1rYBfb2KVfl+Q2dunVJbmNXfZ2uSs9Db4s+Oxzc40Pjhsf0G3mzc8zb3PlVVepAxVVR12/YfMW1aJFCwVAndK5syrb7673ulokJm72NUs6Psw2xmC7fMUqFRUVpQCohIQE9fQzz6r8pctV3oJP1aBBg80N6JmZqdwHKx3/y9DUuyS3sUuvLslt7NKrS3Ibu+ztqs/A+PkXi9TkqdPMy1//9ndznf369z/quslTp/m8nyuvusq83cWXXKI+mPuhWrJ0uXp54iTVoUMHBUAZhqH+88GcBg3cTX6wrfIoNWvWbBUXF2f+Rh97SUtLU9t37HT8LwK75Lexy/m1so1dOnZJbmOXfWuoz8A4ZM6PGccAAAH9SURBVOgwn2v1dvF1P3t+/V1ddPHFPm8XERGhJrw8scHPJPsbbI2a1ziEgoyMDLViZYFt919cXIwJL72IvLz5KCkpQUxMDLqkpGDw4GsxfPjtiIiIsO2x7SS1C5Dbxi79SG1jl36ktrHLHlWHDgV87i033Yh3pr8d8PnuSt+f+KCUwrvvTMc706dj/Xfr8Pvvv6Nd+/bo1esC3HX3Peh+1lkBP86x2rVpVbS3tDTZ23VNarAlIiIiakrqM9jqwt9gK/5TEYiIiIioaeBgS0REREQicLAlIiIiIhE42BIRERGRCBxsiYiIiEgEDrZEREREJAIHWyIiIiISgYMtEREREYnAwZaIiIiIROBgS0REREQicLAlIiIiIhE42BIRERGRCBxsiYiIiEgEDrZEREREJAIHWyIiIiISgYMtEREREYnAwZaIiIiIROBgS0REREQiGEopp9dgMgxjN4AfGunhWgH4pZEei4LH/dIP90w/3DP9cM/0wv2yxslKqdbergipwbYxGYZRoJTKcHodFBjul364Z/rhnumHe6YX7pf9+FIEIiIiIhKBgy0RERERidCUB9vJTi+A6oX7pR/umX64Z/rhnumF+2WzJvsaWyIiIiKSpSk/Y0tEREREgnCwJSIiIiIRONgSERERkQgcbImIiIhIBA62RERERCTC/wNGG88Bl2odUwAAAABJRU5ErkJggg==",
            "text/plain": [
              "<Figure size 864x864 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 720x576 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "labels = y_test1 \n",
        "p1 = model1.predict(x_test1)\n",
        "p2 = model2.predict(x_test1)\n",
        "p3 = model3.predict(x_test1)\n",
        "p4 = model4.predict(x_test1)\n",
        "p5 = model5.predict(x_test1)\n",
        "p6 = model6.predict(x_test1)\n",
        "p7 = model7.predict(x_test1)\n",
        "p8 = model8.predict(x_test1)\n",
        "p9 = model9.predict(x_test1)\n",
        "p10 = model10.predict(x_test1)\n",
        "\n",
        "predictions = (p1+p2+p3+p4+p5+p6+p7+p8+p9+p10)/10.0\n",
        "assert len(labels) == len(predictions)\n",
        "\n",
        "# Compute the confusion matrix\n",
        "cm = confusion_matrix(np.argmax(labels, axis=-1), np.argmax(predictions, axis=-1))\n",
        "\n",
        "# Compute the classification metrics\n",
        "accuracy = accuracy_score(np.argmax(labels, axis=-1), np.argmax(predictions, axis=-1))\n",
        "precision = precision_score(np.argmax(labels, axis=-1), np.argmax(predictions, axis=-1), average='macro')\n",
        "recall = recall_score(np.argmax(labels, axis=-1), np.argmax(predictions, axis=-1), average='macro')\n",
        "f1 = f1_score(np.argmax(labels, axis=-1), np.argmax(predictions, axis=-1), average='macro')\n",
        "f1_classes = f1_score(np.argmax(labels, axis=-1), np.argmax(predictions, axis=-1), average=None)\n",
        "\n",
        "print('Accuracy:',accuracy.round(4))\n",
        "print('Precision:',precision.round(4))\n",
        "print('Recall:',recall.round(4))\n",
        "print('F1:',f1.round(4))\n",
        "print('F1 (per class):' ,f1_classes)\n",
        "\n",
        "# Plot the confusion matrix\n",
        "ig, ax = plt.subplots(figsize=(12,12))\n",
        "ax.matshow(cm, cmap=plt.cm.Blues, alpha=0.3)\n",
        "for i in range(cm.shape[0]):\n",
        "    for j in range(cm.shape[1]):\n",
        "        ax.text(x=j, y=i,s=cm[i, j], va='center', ha='center', size='xx-large')\n",
        "\n",
        "        \n",
        "plt.figure(figsize=(10,8))\n",
        "sns.heatmap(cm.T, cmap='Blues')\n",
        "plt.xlabel('True labels')\n",
        "plt.ylabel('Predicted labels')\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "e3luJb6Ev2jU"
      },
      "source": [
        "# save"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 30,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "xofA0OW72DaZ",
        "outputId": "eb7db49c-5bac-4980-8700-6368395d56f0"
      },
      "outputs": [
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:absl:Function `_wrapped_model` contains input name(s) Input with unsupported characters which will be renamed to input in the SavedModel.\n",
            "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op while saving (showing 5 of 6). These functions will not be directly callable after loading.\n",
            "WARNING:absl:Function `_wrapped_model` contains input name(s) Input with unsupported characters which will be renamed to input in the SavedModel.\n",
            "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op while saving (showing 5 of 6). These functions will not be directly callable after loading.\n",
            "WARNING:absl:Function `_wrapped_model` contains input name(s) Input with unsupported characters which will be renamed to input in the SavedModel.\n",
            "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op while saving (showing 5 of 6). These functions will not be directly callable after loading.\n",
            "WARNING:absl:Function `_wrapped_model` contains input name(s) Input with unsupported characters which will be renamed to input in the SavedModel.\n",
            "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op while saving (showing 5 of 6). These functions will not be directly callable after loading.\n",
            "WARNING:absl:Function `_wrapped_model` contains input name(s) Input with unsupported characters which will be renamed to input in the SavedModel.\n",
            "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op while saving (showing 5 of 6). These functions will not be directly callable after loading.\n",
            "WARNING:absl:Function `_wrapped_model` contains input name(s) Input with unsupported characters which will be renamed to input in the SavedModel.\n",
            "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op while saving (showing 5 of 6). These functions will not be directly callable after loading.\n",
            "WARNING:absl:Function `_wrapped_model` contains input name(s) Input with unsupported characters which will be renamed to input in the SavedModel.\n",
            "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op while saving (showing 5 of 6). These functions will not be directly callable after loading.\n",
            "WARNING:absl:Function `_wrapped_model` contains input name(s) Input with unsupported characters which will be renamed to input in the SavedModel.\n",
            "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op while saving (showing 5 of 6). These functions will not be directly callable after loading.\n",
            "WARNING:absl:Function `_wrapped_model` contains input name(s) Input with unsupported characters which will be renamed to input in the SavedModel.\n",
            "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op while saving (showing 5 of 6). These functions will not be directly callable after loading.\n",
            "WARNING:absl:Function `_wrapped_model` contains input name(s) Input with unsupported characters which will be renamed to input in the SavedModel.\n",
            "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op while saving (showing 5 of 6). These functions will not be directly callable after loading.\n"
          ]
        }
      ],
      "source": [
        "model1.save(\"bagging_model1\")\n",
        "model2.save(\"bagging_model2\")\n",
        "model3.save(\"bagging_model3\")\n",
        "model4.save(\"bagging_model4\")\n",
        "model5.save(\"bagging_model5\")\n",
        "model6.save(\"bagging_model6\")\n",
        "model7.save(\"bagging_model7\")\n",
        "model8.save(\"bagging_model8\")\n",
        "model9.save(\"bagging_model9\")\n",
        "model10.save(\"bagging_model10\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 31,
      "metadata": {
        "id": "Ur5dxgd_vuWJ"
      },
      "outputs": [],
      "source": [
        "import pickle \n",
        "\n",
        "with open(\"scaler\", r'wb') as f:  \n",
        "  pickle.dump(scaler, f)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 34,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rZfOrp3P3HXO",
        "outputId": "e7833236-8d46-4d8d-a0e7-3bfa7fd610ea"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "  adding: bagging_model1/ (stored 0%)\n",
            "  adding: bagging_model1/variables/ (stored 0%)\n",
            "  adding: bagging_model1/variables/variables.index (deflated 74%)\n",
            "  adding: bagging_model1/variables/variables.data-00000-of-00001 (deflated 10%)\n",
            "  adding: bagging_model1/assets/ (stored 0%)\n",
            "  adding: bagging_model1/saved_model.pb (deflated 90%)\n",
            "  adding: bagging_model1/keras_metadata.pb (deflated 93%)\n",
            "  adding: bagging_model10/ (stored 0%)\n",
            "  adding: bagging_model10/variables/ (stored 0%)\n",
            "  adding: bagging_model10/variables/variables.index (deflated 74%)\n",
            "  adding: bagging_model10/variables/variables.data-00000-of-00001 (deflated 10%)\n",
            "  adding: bagging_model10/assets/ (stored 0%)\n",
            "  adding: bagging_model10/saved_model.pb (deflated 90%)\n",
            "  adding: bagging_model10/keras_metadata.pb (deflated 93%)\n",
            "  adding: bagging_model2/ (stored 0%)\n",
            "  adding: bagging_model2/variables/ (stored 0%)\n",
            "  adding: bagging_model2/variables/variables.index (deflated 74%)\n",
            "  adding: bagging_model2/variables/variables.data-00000-of-00001 (deflated 10%)\n",
            "  adding: bagging_model2/assets/ (stored 0%)\n",
            "  adding: bagging_model2/keras_metadata.pb (deflated 93%)\n",
            "  adding: bagging_model2/saved_model.pb (deflated 90%)\n",
            "  adding: bagging_model3/ (stored 0%)\n",
            "  adding: bagging_model3/variables/ (stored 0%)\n",
            "  adding: bagging_model3/variables/variables.index (deflated 74%)\n",
            "  adding: bagging_model3/variables/variables.data-00000-of-00001 (deflated 10%)\n",
            "  adding: bagging_model3/assets/ (stored 0%)\n",
            "  adding: bagging_model3/keras_metadata.pb (deflated 93%)\n",
            "  adding: bagging_model3/saved_model.pb (deflated 90%)\n",
            "  adding: bagging_model4/ (stored 0%)\n",
            "  adding: bagging_model4/variables/ (stored 0%)\n",
            "  adding: bagging_model4/variables/variables.index (deflated 74%)\n",
            "  adding: bagging_model4/variables/variables.data-00000-of-00001 (deflated 10%)\n",
            "  adding: bagging_model4/assets/ (stored 0%)\n",
            "  adding: bagging_model4/keras_metadata.pb (deflated 93%)\n",
            "  adding: bagging_model4/saved_model.pb (deflated 89%)\n",
            "  adding: bagging_model5/ (stored 0%)\n",
            "  adding: bagging_model5/variables/ (stored 0%)\n",
            "  adding: bagging_model5/variables/variables.index (deflated 74%)\n",
            "  adding: bagging_model5/variables/variables.data-00000-of-00001 (deflated 9%)\n",
            "  adding: bagging_model5/assets/ (stored 0%)\n",
            "  adding: bagging_model5/saved_model.pb (deflated 90%)\n",
            "  adding: bagging_model5/keras_metadata.pb (deflated 93%)\n",
            "  adding: bagging_model6/ (stored 0%)\n",
            "  adding: bagging_model6/variables/ (stored 0%)\n",
            "  adding: bagging_model6/variables/variables.index (deflated 74%)\n",
            "  adding: bagging_model6/variables/variables.data-00000-of-00001 (deflated 10%)\n",
            "  adding: bagging_model6/assets/ (stored 0%)\n",
            "  adding: bagging_model6/keras_metadata.pb (deflated 93%)\n",
            "  adding: bagging_model6/saved_model.pb (deflated 90%)\n",
            "  adding: bagging_model7/ (stored 0%)\n",
            "  adding: bagging_model7/variables/ (stored 0%)\n",
            "  adding: bagging_model7/variables/variables.index (deflated 74%)\n",
            "  adding: bagging_model7/variables/variables.data-00000-of-00001 (deflated 10%)\n",
            "  adding: bagging_model7/assets/ (stored 0%)\n",
            "  adding: bagging_model7/keras_metadata.pb (deflated 93%)\n",
            "  adding: bagging_model7/saved_model.pb (deflated 90%)\n",
            "  adding: bagging_model8/ (stored 0%)\n",
            "  adding: bagging_model8/variables/ (stored 0%)\n",
            "  adding: bagging_model8/variables/variables.index (deflated 74%)\n",
            "  adding: bagging_model8/variables/variables.data-00000-of-00001 (deflated 10%)\n",
            "  adding: bagging_model8/assets/ (stored 0%)\n",
            "  adding: bagging_model8/keras_metadata.pb (deflated 93%)\n",
            "  adding: bagging_model8/saved_model.pb (deflated 90%)\n",
            "  adding: bagging_model9/ (stored 0%)\n",
            "  adding: bagging_model9/variables/ (stored 0%)\n",
            "  adding: bagging_model9/variables/variables.data-00000-of-00001 (deflated 9%)\n",
            "  adding: bagging_model9/variables/variables.index (deflated 74%)\n",
            "  adding: bagging_model9/assets/ (stored 0%)\n",
            "  adding: bagging_model9/keras_metadata.pb (deflated 93%)\n",
            "  adding: bagging_model9/saved_model.pb (deflated 90%)\n",
            "  adding: scaler (deflated 18%)\n",
            "  adding: metadata (stored 0%)\n",
            "  adding: model.py (deflated 74%)\n"
          ]
        }
      ],
      "source": [
        "! zip -r subEnrico.zip bagging_model* scaler metadata model.py"
      ]
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "provenance": []
    },
    "gpuClass": "standard",
    "kernelspec": {
      "display_name": "Python 3.9.2 64-bit",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "name": "python",
      "version": "3.9.2"
    },
    "vscode": {
      "interpreter": {
        "hash": "916dbcbb3f70747c44a77c7bcd40155683ae19c65e1c03b4aa3499c5328201f1"
      }
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}