AP_Triples.ipynb

{
  "cells": [
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {
        "id": "uGfyqu0b1ujL"
      },
      "source": [
        "Create Embedding Network"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "hdULt7ty1jgQ",
        "outputId": "edf199ec-11e5-43e9-a3c9-ca4f130840f1"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Using:\n",
            "\t• TensorFlow version: 2.12.0\n",
            "\t• tf.keras version: 2.12.0\n",
            "\t• Running on GPU\n"
          ]
        }
      ],
      "source": [
        "import warnings\n",
        "warnings.filterwarnings('ignore')\n",
        "\n",
        "import os\n",
        "import time\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "#!pip install tensorflow==2.9.1 # For EfficientNetB0\n",
        "import tensorflow as tf\n",
        "import math\n",
        "\n",
        "import utils\n",
        "\n",
        "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
        "from tensorflow.keras.preprocessing import image\n",
        "from tensorflow.keras.applications import vgg16, imagenet_utils\n",
        "from tensorflow.keras.optimizers import Adam, RMSprop\n",
        "from tensorflow.keras.regularizers import l2\n",
        "from tensorflow.keras.metrics import binary_crossentropy\n",
        "from tensorflow.keras.layers import Dense, Flatten, Input, Lambda, Dropout, BatchNormalization, GlobalAveragePooling2D\n",
        "from tensorflow.keras.models import Model\n",
        "from tensorflow.keras.callbacks import EarlyStopping, ModelCheckpoint, ReduceLROnPlateau\n",
        "\n",
        "from sklearn.metrics import confusion_matrix, ConfusionMatrixDisplay, accuracy_score, precision_score, recall_score\n",
        "import itertools\n",
        "\n",
        "from skimage import exposure\n",
        "\n",
        "import logging\n",
        "from tensorflow.keras.models import load_model\n",
        "\n",
        "logger = tf.get_logger()\n",
        "logger.setLevel(logging.ERROR)\n",
        "\n",
        "print('Using:')\n",
        "print('\\t\\u2022 TensorFlow version:', tf.__version__)\n",
        "print('\\t\\u2022 tf.keras version:', tf.keras.__version__)\n",
        "print('\\t\\u2022 Running on GPU' if tf.test.is_gpu_available() else '\\t\\u2022 GPU device not found. Running on CPU')"
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {
        "id": "UOYYsI6o157S"
      },
      "source": [
        "Load Dataset (Using 150x150)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "zIrlgW2m1tbH",
        "outputId": "500e580f-5652-48be-a49b-3aaf4f228d21"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Found 1372 images belonging to 4 classes.\n",
            "Found 40 images belonging to 4 classes.\n",
            "The dataset contains:\n",
            "• 1372 training images\n",
            "• 40 validation images\n",
            "\n",
            "The training set contains:\n",
            "• 343 glioma images\n",
            "• 343 meningioma images\n",
            "• 343 normal images\n",
            "• 343 pituitary images\n",
            "\n",
            "The validation set contains:\n",
            "• 10 glioma images\n",
            "• 10 meningioma images\n",
            "• 10 normal images\n",
            "• 10 pituitary images\n"
          ]
        }
      ],
      "source": [
        "\"\"\" load the datasets \"\"\"\n",
        "\n",
        "base_dir = os.path.join('dataset', 'pretrain4w')\n",
        "\n",
        "train_dir = os.path.join(base_dir, 'train')\n",
        "val_dir = os.path.join(base_dir, 'validation')\n",
        "test_dir = os.path.join(base_dir, 'test')\n",
        "\n",
        "train_glioma_dir = os.path.join(train_dir, 'glioma_tumor')\n",
        "train_meningioma_dir = os.path.join(train_dir, 'meningioma_tumor')\n",
        "train_normal_dir = os.path.join(train_dir, 'no_tumor')\n",
        "train_pituitary_dir = os.path.join(train_dir, 'pituitary_tumor')\n",
        "\n",
        "val_glioma_dir = os.path.join(val_dir, 'glioma_tumor')\n",
        "val_meningioma_dir = os.path.join(val_dir, 'meningioma_tumor')\n",
        "val_normal_dir = os.path.join(val_dir, 'no_tumor')\n",
        "val_pituitary_dir = os.path.join(val_dir, 'pituitary_tumor')\n",
        "\n",
        "INPUT_SIZE = 150\n",
        "BATCH_SIZE = 16\n",
        "\n",
        "\"\"\" Investigate train - val datasets \"\"\"\n",
        "\n",
        "train_batches = ImageDataGenerator(rescale = 1 / 255.).flow_from_directory(train_dir,\n",
        "                                                         target_size=(INPUT_SIZE,INPUT_SIZE),\n",
        "                                                         class_mode='categorical',\n",
        "                                                         shuffle=True,\n",
        "                                                         seed=42,\n",
        "                                                         batch_size=BATCH_SIZE)\n",
        "\n",
        "val_batches = ImageDataGenerator(rescale = 1 / 255.).flow_from_directory(val_dir,\n",
        "                                                         target_size=(INPUT_SIZE,INPUT_SIZE),\n",
        "                                                         class_mode='categorical',\n",
        "                                                         shuffle=True,\n",
        "                                                         seed=42,\n",
        "                                                         batch_size=BATCH_SIZE)\n",
        "\n",
        "num_glioma_train = int(len(os.listdir(train_glioma_dir)))\n",
        "num_meningioma_train = int(len(os.listdir(train_meningioma_dir)))\n",
        "num_normal_train = int(len(os.listdir(train_normal_dir)))\n",
        "num_pituitary_train = int(len(os.listdir(train_pituitary_dir)))\n",
        "\n",
        "num_glioma_val = int(len(os.listdir(val_glioma_dir)))\n",
        "num_meningioma_val = int(len(os.listdir(val_meningioma_dir)))\n",
        "num_normal_val = int(len(os.listdir(val_normal_dir)))\n",
        "num_pituitary_val = int(len(os.listdir(val_pituitary_dir)))\n",
        "\n",
        "print('The dataset contains:')\n",
        "print('\\u2022 {} training images'.format(num_glioma_train + num_meningioma_train + num_normal_train + num_pituitary_train))\n",
        "print('\\u2022 {} validation images'.format(num_glioma_val + num_meningioma_val + num_normal_val + num_pituitary_val))\n",
        "\n",
        "print('\\nThe training set contains:')\n",
        "print('\\u2022 {} glioma images'.format(num_glioma_train))\n",
        "print('\\u2022 {} meningioma images'.format(num_meningioma_train))\n",
        "print('\\u2022 {} normal images'.format(num_normal_train))\n",
        "print('\\u2022 {} pituitary images'.format(num_pituitary_train))\n",
        "\n",
        "print('\\nThe validation set contains:')\n",
        "print('\\u2022 {} glioma images'.format(num_glioma_val))\n",
        "print('\\u2022 {} meningioma images'.format(num_meningioma_val))\n",
        "print('\\u2022 {} normal images'.format(num_normal_val))\n",
        "print('\\u2022 {} pituitary images'.format(num_pituitary_val))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "moIDBCAz3mP8",
        "outputId": "d0077913-7a6c-4788-9756-9b5f81baca2a"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/vgg16/vgg16_weights_tf_dim_ordering_tf_kernels_notop.h5\n",
            "58889256/58889256 [==============================] - 3s 0us/step\n",
            "Number of layers in the base model:  19\n",
            "Model: \"vgg16\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_1 (InputLayer)        [(None, 150, 150, 3)]     0         \n",
            "                                                                 \n",
            " block1_conv1 (Conv2D)       (None, 150, 150, 64)      1792      \n",
            "                                                                 \n",
            " block1_conv2 (Conv2D)       (None, 150, 150, 64)      36928     \n",
            "                                                                 \n",
            " block1_pool (MaxPooling2D)  (None, 75, 75, 64)        0         \n",
            "                                                                 \n",
            " block2_conv1 (Conv2D)       (None, 75, 75, 128)       73856     \n",
            "                                                                 \n",
            " block2_conv2 (Conv2D)       (None, 75, 75, 128)       147584    \n",
            "                                                                 \n",
            " block2_pool (MaxPooling2D)  (None, 37, 37, 128)       0         \n",
            "                                                                 \n",
            " block3_conv1 (Conv2D)       (None, 37, 37, 256)       295168    \n",
            "                                                                 \n",
            " block3_conv2 (Conv2D)       (None, 37, 37, 256)       590080    \n",
            "                                                                 \n",
            " block3_conv3 (Conv2D)       (None, 37, 37, 256)       590080    \n",
            "                                                                 \n",
            " block3_pool (MaxPooling2D)  (None, 18, 18, 256)       0         \n",
            "                                                                 \n",
            " block4_conv1 (Conv2D)       (None, 18, 18, 512)       1180160   \n",
            "                                                                 \n",
            " block4_conv2 (Conv2D)       (None, 18, 18, 512)       2359808   \n",
            "                                                                 \n",
            " block4_conv3 (Conv2D)       (None, 18, 18, 512)       2359808   \n",
            "                                                                 \n",
            " block4_pool (MaxPooling2D)  (None, 9, 9, 512)         0         \n",
            "                                                                 \n",
            " block5_conv1 (Conv2D)       (None, 9, 9, 512)         2359808   \n",
            "                                                                 \n",
            " block5_conv2 (Conv2D)       (None, 9, 9, 512)         2359808   \n",
            "                                                                 \n",
            " block5_conv3 (Conv2D)       (None, 9, 9, 512)         2359808   \n",
            "                                                                 \n",
            " block5_pool (MaxPooling2D)  (None, 4, 4, 512)         0         \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 14,714,688\n",
            "Trainable params: 14,714,688\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "Model: \"model\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_1 (InputLayer)        [(None, 150, 150, 3)]     0         \n",
            "                                                                 \n",
            " block1_conv1 (Conv2D)       (None, 150, 150, 64)      1792      \n",
            "                                                                 \n",
            " block1_conv2 (Conv2D)       (None, 150, 150, 64)      36928     \n",
            "                                                                 \n",
            " block1_pool (MaxPooling2D)  (None, 75, 75, 64)        0         \n",
            "                                                                 \n",
            " block2_conv1 (Conv2D)       (None, 75, 75, 128)       73856     \n",
            "                                                                 \n",
            " block2_conv2 (Conv2D)       (None, 75, 75, 128)       147584    \n",
            "                                                                 \n",
            " block2_pool (MaxPooling2D)  (None, 37, 37, 128)       0         \n",
            "                                                                 \n",
            " block3_conv1 (Conv2D)       (None, 37, 37, 256)       295168    \n",
            "                                                                 \n",
            " block3_conv2 (Conv2D)       (None, 37, 37, 256)       590080    \n",
            "                                                                 \n",
            " block3_conv3 (Conv2D)       (None, 37, 37, 256)       590080    \n",
            "                                                                 \n",
            " block3_pool (MaxPooling2D)  (None, 18, 18, 256)       0         \n",
            "                                                                 \n",
            " block4_conv1 (Conv2D)       (None, 18, 18, 512)       1180160   \n",
            "                                                                 \n",
            " block4_conv2 (Conv2D)       (None, 18, 18, 512)       2359808   \n",
            "                                                                 \n",
            " block4_conv3 (Conv2D)       (None, 18, 18, 512)       2359808   \n",
            "                                                                 \n",
            " block4_pool (MaxPooling2D)  (None, 9, 9, 512)         0         \n",
            "                                                                 \n",
            " block5_conv1 (Conv2D)       (None, 9, 9, 512)         2359808   \n",
            "                                                                 \n",
            " block5_conv2 (Conv2D)       (None, 9, 9, 512)         2359808   \n",
            "                                                                 \n",
            " block5_conv3 (Conv2D)       (None, 9, 9, 512)         2359808   \n",
            "                                                                 \n",
            " block5_pool (MaxPooling2D)  (None, 4, 4, 512)         0         \n",
            "                                                                 \n",
            " global_average_pooling2d (G  (None, 512)              0         \n",
            " lobalAveragePooling2D)                                          \n",
            "                                                                 \n",
            " dropout (Dropout)           (None, 512)               0         \n",
            "                                                                 \n",
            " flatten (Flatten)           (None, 512)               0         \n",
            "                                                                 \n",
            " dense (Dense)               (None, 4)                 2052      \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 14,716,740\n",
            "Trainable params: 14,716,740\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ],
      "source": [
        "MODEL_FNAME = 'embedding_network.h5'\n",
        "\n",
        "tf.compat.v1.reset_default_graph()\n",
        "\n",
        "tf.random.set_seed(0)\n",
        "tf.keras.utils.set_random_seed(0)\n",
        "\n",
        "\n",
        "\"\"\" Create the model \"\"\"\n",
        "    \n",
        "base_model = vgg16.VGG16(weights='imagenet', include_top=False, input_shape=(INPUT_SIZE,INPUT_SIZE,3)) # loss: 0.0015 - accuracy: 1.0000 - val_loss: 0.1161 - val_accuracy: 0.9750 - Test: 0.94557\n",
        "#base_model = EfficientNetB0(weights='imagenet',include_top=False,input_shape=(INPUT_SIZE,INPUT_SIZE,3)) # loss: 0.0312 - accuracy: 0.9920 - val_loss: 0.1799 - val_accuracy: 0.9250 - Test: 0.933673\n",
        "#base_model = Xception(weights='imagenet', include_top=False, input_shape=(INPUT_SIZE,INPUT_SIZE,3)) # loss: 0.0335 - accuracy: 0.9978 - val_loss: 0.3005 - val_accuracy: 0.9000\n",
        "#base_model = InceptionV3(weights='imagenet', include_top=False, input_shape=(INPUT_SIZE,INPUT_SIZE,3)) # loss: 0.0912 - accuracy: 0.9730 - val_loss: 0.1897 - val_accuracy: 0.9250\n",
        "#base_model = InceptionResNetV2(weights='imagenet', include_top=False, input_shape=(INPUT_SIZE,INPUT_SIZE,3)) # loss: 0.0063 - accuracy: 0.9978 - val_loss: 0.2221 - val_accuracy: 0.9250\n",
        "#base_model = ResNet50(weights='imagenet', include_top=False, input_shape=(INPUT_SIZE,INPUT_SIZE,3)) # loss: 0.0028 - accuracy: 0.9985 - val_loss: 0.1825 - val_accuracy: 0.9250\n",
        "#base_model = vgg19.VGG19(weights='imagenet', include_top=False, input_shape=(INPUT_SIZE,INPUT_SIZE,3)) # loss: 0.0149 - accuracy: 0.9956 - val_loss: 0.1836 - val_accuracy: 0.9750 - Test: 0.930272\n",
        "\n",
        "print(\"Number of layers in the base model: \", len(base_model.layers))\n",
        "    \n",
        "base_model.trainable = True\n",
        "    \n",
        "base_model.summary()\n",
        "    \n",
        "last_output = base_model.output\n",
        "\n",
        "x = GlobalAveragePooling2D()(last_output)\n",
        "x = Dropout(rate=0.5)(x)\n",
        "\n",
        "x = Flatten()(x)\n",
        "x = Dense(4, activation='softmax')(x)\n",
        "    \n",
        "embedding_network = Model(inputs=[base_model.input], outputs=[x])\n",
        "    \n",
        "optimizer = Adam(learning_rate=0.00001) \n",
        "embedding_network.compile(loss='categorical_crossentropy', optimizer=optimizer, metrics=[\"accuracy\"])\n",
        "embedding_network.summary()\n",
        "    \n",
        "\"\"\" callbacks \"\"\"\n",
        "    \n",
        "#early_stopping = EarlyStopping(monitor='val_loss', patience=10)\n",
        "    \n",
        "checkpointer = ModelCheckpoint(filepath='embedding_network.h5', verbose=1, \n",
        "                                   save_best_only=True)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "JkleuxdI3yiH",
        "outputId": "2258c696-08f6-4347-87b6-28b783fd320c"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 1.1564 - accuracy: 0.4796\n",
            "Epoch 1: val_loss improved from inf to 0.76903, saving model to embedding_network.h5\n",
            "86/86 [==============================] - 31s 170ms/step - loss: 1.1564 - accuracy: 0.4796 - val_loss: 0.7690 - val_accuracy: 0.7250\n",
            "Epoch 2/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.7322 - accuracy: 0.7194\n",
            "Epoch 2: val_loss improved from 0.76903 to 0.53816, saving model to embedding_network.h5\n",
            "86/86 [==============================] - 11s 126ms/step - loss: 0.7322 - accuracy: 0.7194 - val_loss: 0.5382 - val_accuracy: 0.8000\n",
            "Epoch 3/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.5347 - accuracy: 0.8017\n",
            "Epoch 3: val_loss improved from 0.53816 to 0.51535, saving model to embedding_network.h5\n",
            "86/86 [==============================] - 11s 125ms/step - loss: 0.5347 - accuracy: 0.8017 - val_loss: 0.5154 - val_accuracy: 0.7750\n",
            "Epoch 4/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.4207 - accuracy: 0.8550\n",
            "Epoch 4: val_loss improved from 0.51535 to 0.37397, saving model to embedding_network.h5\n",
            "86/86 [==============================] - 11s 126ms/step - loss: 0.4207 - accuracy: 0.8550 - val_loss: 0.3740 - val_accuracy: 0.9000\n",
            "Epoch 5/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.3302 - accuracy: 0.8827\n",
            "Epoch 5: val_loss improved from 0.37397 to 0.34895, saving model to embedding_network.h5\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 0.3302 - accuracy: 0.8827 - val_loss: 0.3489 - val_accuracy: 0.8750\n",
            "Epoch 6/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.2469 - accuracy: 0.9103\n",
            "Epoch 6: val_loss improved from 0.34895 to 0.32898, saving model to embedding_network.h5\n",
            "86/86 [==============================] - 11s 129ms/step - loss: 0.2469 - accuracy: 0.9103 - val_loss: 0.3290 - val_accuracy: 0.9000\n",
            "Epoch 7/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.2061 - accuracy: 0.9278\n",
            "Epoch 7: val_loss improved from 0.32898 to 0.29072, saving model to embedding_network.h5\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 0.2061 - accuracy: 0.9278 - val_loss: 0.2907 - val_accuracy: 0.9000\n",
            "Epoch 8/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.1795 - accuracy: 0.9395\n",
            "Epoch 8: val_loss did not improve from 0.29072\n",
            "86/86 [==============================] - 11s 126ms/step - loss: 0.1795 - accuracy: 0.9395 - val_loss: 0.3398 - val_accuracy: 0.9000\n",
            "Epoch 9/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.1576 - accuracy: 0.9475\n",
            "Epoch 9: val_loss did not improve from 0.29072\n",
            "86/86 [==============================] - 11s 129ms/step - loss: 0.1576 - accuracy: 0.9475 - val_loss: 0.3492 - val_accuracy: 0.9000\n",
            "Epoch 10/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.1208 - accuracy: 0.9614\n",
            "Epoch 10: val_loss improved from 0.29072 to 0.20970, saving model to embedding_network.h5\n",
            "86/86 [==============================] - 11s 132ms/step - loss: 0.1208 - accuracy: 0.9614 - val_loss: 0.2097 - val_accuracy: 0.9000\n",
            "Epoch 11/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0850 - accuracy: 0.9716\n",
            "Epoch 11: val_loss did not improve from 0.20970\n",
            "86/86 [==============================] - 11s 130ms/step - loss: 0.0850 - accuracy: 0.9716 - val_loss: 0.4106 - val_accuracy: 0.8750\n",
            "Epoch 12/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0710 - accuracy: 0.9767\n",
            "Epoch 12: val_loss did not improve from 0.20970\n",
            "86/86 [==============================] - 11s 126ms/step - loss: 0.0710 - accuracy: 0.9767 - val_loss: 0.4489 - val_accuracy: 0.9250\n",
            "Epoch 13/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0626 - accuracy: 0.9818\n",
            "Epoch 13: val_loss did not improve from 0.20970\n",
            "86/86 [==============================] - 11s 126ms/step - loss: 0.0626 - accuracy: 0.9818 - val_loss: 0.4935 - val_accuracy: 0.9250\n",
            "Epoch 14/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0616 - accuracy: 0.9810\n",
            "Epoch 14: val_loss did not improve from 0.20970\n",
            "86/86 [==============================] - 11s 129ms/step - loss: 0.0616 - accuracy: 0.9810 - val_loss: 0.2569 - val_accuracy: 0.9000\n",
            "Epoch 15/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0373 - accuracy: 0.9869\n",
            "Epoch 15: val_loss did not improve from 0.20970\n",
            "86/86 [==============================] - 11s 125ms/step - loss: 0.0373 - accuracy: 0.9869 - val_loss: 0.3908 - val_accuracy: 0.9000\n",
            "Epoch 16/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0504 - accuracy: 0.9825\n",
            "Epoch 16: val_loss improved from 0.20970 to 0.20414, saving model to embedding_network.h5\n",
            "86/86 [==============================] - 11s 133ms/step - loss: 0.0504 - accuracy: 0.9825 - val_loss: 0.2041 - val_accuracy: 0.9000\n",
            "Epoch 17/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0397 - accuracy: 0.9920\n",
            "Epoch 17: val_loss did not improve from 0.20414\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 0.0397 - accuracy: 0.9920 - val_loss: 0.2698 - val_accuracy: 0.9250\n",
            "Epoch 18/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0200 - accuracy: 0.9942\n",
            "Epoch 18: val_loss did not improve from 0.20414\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 0.0200 - accuracy: 0.9942 - val_loss: 0.3716 - val_accuracy: 0.9000\n",
            "Epoch 19/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0365 - accuracy: 0.9883\n",
            "Epoch 19: val_loss did not improve from 0.20414\n",
            "86/86 [==============================] - 11s 129ms/step - loss: 0.0365 - accuracy: 0.9883 - val_loss: 0.4161 - val_accuracy: 0.9000\n",
            "Epoch 20/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0295 - accuracy: 0.9920\n",
            "Epoch 20: val_loss did not improve from 0.20414\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 0.0295 - accuracy: 0.9920 - val_loss: 0.3650 - val_accuracy: 0.9250\n",
            "Epoch 21/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0358 - accuracy: 0.9883\n",
            "Epoch 21: val_loss did not improve from 0.20414\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 0.0358 - accuracy: 0.9883 - val_loss: 0.2543 - val_accuracy: 0.9250\n",
            "Epoch 22/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0164 - accuracy: 0.9964\n",
            "Epoch 22: val_loss improved from 0.20414 to 0.16201, saving model to embedding_network.h5\n",
            "86/86 [==============================] - 11s 131ms/step - loss: 0.0164 - accuracy: 0.9964 - val_loss: 0.1620 - val_accuracy: 0.9500\n",
            "Epoch 23/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0208 - accuracy: 0.9942\n",
            "Epoch 23: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 126ms/step - loss: 0.0208 - accuracy: 0.9942 - val_loss: 0.1675 - val_accuracy: 0.9750\n",
            "Epoch 24/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0230 - accuracy: 0.9934\n",
            "Epoch 24: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 0.0230 - accuracy: 0.9934 - val_loss: 0.3495 - val_accuracy: 0.9000\n",
            "Epoch 25/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0110 - accuracy: 0.9978\n",
            "Epoch 25: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 125ms/step - loss: 0.0110 - accuracy: 0.9978 - val_loss: 0.6693 - val_accuracy: 0.9000\n",
            "Epoch 26/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0135 - accuracy: 0.9978\n",
            "Epoch 26: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 129ms/step - loss: 0.0135 - accuracy: 0.9978 - val_loss: 0.3274 - val_accuracy: 0.9000\n",
            "Epoch 27/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0097 - accuracy: 0.9971\n",
            "Epoch 27: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 0.0097 - accuracy: 0.9971 - val_loss: 0.3210 - val_accuracy: 0.9000\n",
            "Epoch 28/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0401 - accuracy: 0.9847\n",
            "Epoch 28: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 126ms/step - loss: 0.0401 - accuracy: 0.9847 - val_loss: 0.4205 - val_accuracy: 0.9250\n",
            "Epoch 29/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0154 - accuracy: 0.9964\n",
            "Epoch 29: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 0.0154 - accuracy: 0.9964 - val_loss: 0.4327 - val_accuracy: 0.9250\n",
            "Epoch 30/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0047 - accuracy: 0.9985\n",
            "Epoch 30: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 0.0047 - accuracy: 0.9985 - val_loss: 0.2540 - val_accuracy: 0.9250\n",
            "Epoch 31/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0182 - accuracy: 0.9905\n",
            "Epoch 31: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 126ms/step - loss: 0.0182 - accuracy: 0.9905 - val_loss: 0.4781 - val_accuracy: 0.9250\n",
            "Epoch 32/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0207 - accuracy: 0.9934\n",
            "Epoch 32: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 0.0207 - accuracy: 0.9934 - val_loss: 0.5501 - val_accuracy: 0.9000\n",
            "Epoch 33/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0057 - accuracy: 0.9993\n",
            "Epoch 33: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 0.0057 - accuracy: 0.9993 - val_loss: 0.6022 - val_accuracy: 0.9000\n",
            "Epoch 34/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0067 - accuracy: 0.9985\n",
            "Epoch 34: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 126ms/step - loss: 0.0067 - accuracy: 0.9985 - val_loss: 0.4664 - val_accuracy: 0.9250\n",
            "Epoch 35/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0225 - accuracy: 0.9920\n",
            "Epoch 35: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 0.0225 - accuracy: 0.9920 - val_loss: 0.2749 - val_accuracy: 0.8500\n",
            "Epoch 36/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0063 - accuracy: 0.9985\n",
            "Epoch 36: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 130ms/step - loss: 0.0063 - accuracy: 0.9985 - val_loss: 0.2968 - val_accuracy: 0.9250\n",
            "Epoch 37/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0310 - accuracy: 0.9927\n",
            "Epoch 37: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 125ms/step - loss: 0.0310 - accuracy: 0.9927 - val_loss: 0.2733 - val_accuracy: 0.9250\n",
            "Epoch 38/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0028 - accuracy: 1.0000\n",
            "Epoch 38: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 130ms/step - loss: 0.0028 - accuracy: 1.0000 - val_loss: 0.3499 - val_accuracy: 0.9250\n",
            "Epoch 39/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0016 - accuracy: 1.0000\n",
            "Epoch 39: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 0.0016 - accuracy: 1.0000 - val_loss: 0.3759 - val_accuracy: 0.9250\n",
            "Epoch 40/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0019 - accuracy: 0.9993\n",
            "Epoch 40: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 125ms/step - loss: 0.0019 - accuracy: 0.9993 - val_loss: 0.3433 - val_accuracy: 0.9000\n",
            "Epoch 41/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 7.0676e-04 - accuracy: 1.0000\n",
            "Epoch 41: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 7.0676e-04 - accuracy: 1.0000 - val_loss: 0.3860 - val_accuracy: 0.9250\n",
            "Epoch 42/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0015 - accuracy: 1.0000\n",
            "Epoch 42: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 129ms/step - loss: 0.0015 - accuracy: 1.0000 - val_loss: 0.3355 - val_accuracy: 0.9000\n",
            "Epoch 43/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 9.6191e-04 - accuracy: 1.0000\n",
            "Epoch 43: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 9.6191e-04 - accuracy: 1.0000 - val_loss: 0.4428 - val_accuracy: 0.9000\n",
            "Epoch 44/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 3.3915e-04 - accuracy: 1.0000\n",
            "Epoch 44: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 125ms/step - loss: 3.3915e-04 - accuracy: 1.0000 - val_loss: 0.5878 - val_accuracy: 0.9250\n",
            "Epoch 45/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 3.7843e-04 - accuracy: 1.0000\n",
            "Epoch 45: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 3.7843e-04 - accuracy: 1.0000 - val_loss: 0.5294 - val_accuracy: 0.9250\n",
            "Epoch 46/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 2.5387e-04 - accuracy: 1.0000\n",
            "Epoch 46: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 129ms/step - loss: 2.5387e-04 - accuracy: 1.0000 - val_loss: 0.5535 - val_accuracy: 0.9250\n",
            "Epoch 47/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0042 - accuracy: 0.9985\n",
            "Epoch 47: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 0.0042 - accuracy: 0.9985 - val_loss: 0.5328 - val_accuracy: 0.8750\n",
            "Epoch 48/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0321 - accuracy: 0.9920\n",
            "Epoch 48: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 12s 135ms/step - loss: 0.0321 - accuracy: 0.9920 - val_loss: 0.5831 - val_accuracy: 0.9000\n",
            "Epoch 49/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0401 - accuracy: 0.9825\n",
            "Epoch 49: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 0.0401 - accuracy: 0.9825 - val_loss: 0.3188 - val_accuracy: 0.9000\n",
            "Epoch 50/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0133 - accuracy: 0.9956\n",
            "Epoch 50: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 0.0133 - accuracy: 0.9956 - val_loss: 0.2906 - val_accuracy: 0.9500\n",
            "Epoch 51/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0044 - accuracy: 0.9985\n",
            "Epoch 51: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 129ms/step - loss: 0.0044 - accuracy: 0.9985 - val_loss: 0.1867 - val_accuracy: 0.9250\n",
            "Epoch 52/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0051 - accuracy: 0.9985\n",
            "Epoch 52: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 130ms/step - loss: 0.0051 - accuracy: 0.9985 - val_loss: 0.4260 - val_accuracy: 0.9250\n",
            "Epoch 53/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0017 - accuracy: 0.9993\n",
            "Epoch 53: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 0.0017 - accuracy: 0.9993 - val_loss: 0.1747 - val_accuracy: 0.9500\n",
            "Epoch 54/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0023 - accuracy: 0.9985\n",
            "Epoch 54: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 126ms/step - loss: 0.0023 - accuracy: 0.9985 - val_loss: 0.2465 - val_accuracy: 0.9500\n",
            "Epoch 55/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0030 - accuracy: 0.9985\n",
            "Epoch 55: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 0.0030 - accuracy: 0.9985 - val_loss: 0.2835 - val_accuracy: 0.9250\n",
            "Epoch 56/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0274 - accuracy: 0.9876\n",
            "Epoch 56: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 0.0274 - accuracy: 0.9876 - val_loss: 0.3614 - val_accuracy: 0.9250\n",
            "Epoch 57/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0017 - accuracy: 1.0000\n",
            "Epoch 57: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 0.0017 - accuracy: 1.0000 - val_loss: 0.2825 - val_accuracy: 0.9500\n",
            "Epoch 58/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 5.9256e-04 - accuracy: 1.0000\n",
            "Epoch 58: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 125ms/step - loss: 5.9256e-04 - accuracy: 1.0000 - val_loss: 0.3223 - val_accuracy: 0.9500\n",
            "Epoch 59/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 5.8149e-04 - accuracy: 1.0000\n",
            "Epoch 59: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 5.8149e-04 - accuracy: 1.0000 - val_loss: 0.4041 - val_accuracy: 0.9250\n",
            "Epoch 60/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 8.8090e-04 - accuracy: 1.0000\n",
            "Epoch 60: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 8.8090e-04 - accuracy: 1.0000 - val_loss: 0.4293 - val_accuracy: 0.9250\n",
            "Epoch 61/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 4.7751e-04 - accuracy: 1.0000\n",
            "Epoch 61: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 129ms/step - loss: 4.7751e-04 - accuracy: 1.0000 - val_loss: 0.2876 - val_accuracy: 0.9500\n",
            "Epoch 62/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 1.8182e-04 - accuracy: 1.0000\n",
            "Epoch 62: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 1.8182e-04 - accuracy: 1.0000 - val_loss: 0.4445 - val_accuracy: 0.9250\n",
            "Epoch 63/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 1.8882e-04 - accuracy: 1.0000\n",
            "Epoch 63: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 1.8882e-04 - accuracy: 1.0000 - val_loss: 0.3585 - val_accuracy: 0.9250\n",
            "Epoch 64/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 1.2771e-04 - accuracy: 1.0000\n",
            "Epoch 64: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 1.2771e-04 - accuracy: 1.0000 - val_loss: 0.4125 - val_accuracy: 0.9250\n",
            "Epoch 65/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 1.0878e-04 - accuracy: 1.0000\n",
            "Epoch 65: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 129ms/step - loss: 1.0878e-04 - accuracy: 1.0000 - val_loss: 0.5367 - val_accuracy: 0.9250\n",
            "Epoch 66/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 4.9344e-04 - accuracy: 1.0000\n",
            "Epoch 66: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 130ms/step - loss: 4.9344e-04 - accuracy: 1.0000 - val_loss: 0.6084 - val_accuracy: 0.9250\n",
            "Epoch 67/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 5.9015e-04 - accuracy: 1.0000\n",
            "Epoch 67: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 130ms/step - loss: 5.9015e-04 - accuracy: 1.0000 - val_loss: 0.3901 - val_accuracy: 0.9250\n",
            "Epoch 68/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 1.2170e-04 - accuracy: 1.0000\n",
            "Epoch 68: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 126ms/step - loss: 1.2170e-04 - accuracy: 1.0000 - val_loss: 0.4314 - val_accuracy: 0.9250\n",
            "Epoch 69/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 2.1265e-04 - accuracy: 1.0000\n",
            "Epoch 69: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 2.1265e-04 - accuracy: 1.0000 - val_loss: 0.2509 - val_accuracy: 0.9500\n",
            "Epoch 70/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 1.6506e-04 - accuracy: 1.0000\n",
            "Epoch 70: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 1.6506e-04 - accuracy: 1.0000 - val_loss: 0.5417 - val_accuracy: 0.9250\n",
            "Epoch 71/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 3.8915e-04 - accuracy: 1.0000\n",
            "Epoch 71: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 3.8915e-04 - accuracy: 1.0000 - val_loss: 0.3447 - val_accuracy: 0.9250\n",
            "Epoch 72/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 3.0230e-04 - accuracy: 1.0000\n",
            "Epoch 72: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 3.0230e-04 - accuracy: 1.0000 - val_loss: 0.3818 - val_accuracy: 0.9250\n",
            "Epoch 73/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 6.1076e-04 - accuracy: 1.0000\n",
            "Epoch 73: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 6.1076e-04 - accuracy: 1.0000 - val_loss: 0.4084 - val_accuracy: 0.9500\n",
            "Epoch 74/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0814 - accuracy: 0.9752\n",
            "Epoch 74: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 0.0814 - accuracy: 0.9752 - val_loss: 0.1631 - val_accuracy: 0.9500\n",
            "Epoch 75/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0066 - accuracy: 0.9993\n",
            "Epoch 75: val_loss did not improve from 0.16201\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 0.0066 - accuracy: 0.9993 - val_loss: 0.2237 - val_accuracy: 0.9250\n",
            "Epoch 76/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0078 - accuracy: 0.9971\n",
            "Epoch 76: val_loss improved from 0.16201 to 0.12869, saving model to embedding_network.h5\n",
            "86/86 [==============================] - 11s 130ms/step - loss: 0.0078 - accuracy: 0.9971 - val_loss: 0.1287 - val_accuracy: 0.9500\n",
            "Epoch 77/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0025 - accuracy: 1.0000\n",
            "Epoch 77: val_loss did not improve from 0.12869\n",
            "86/86 [==============================] - 11s 133ms/step - loss: 0.0025 - accuracy: 1.0000 - val_loss: 0.2863 - val_accuracy: 0.9500\n",
            "Epoch 78/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0019 - accuracy: 1.0000\n",
            "Epoch 78: val_loss improved from 0.12869 to 0.12815, saving model to embedding_network.h5\n",
            "86/86 [==============================] - 11s 132ms/step - loss: 0.0019 - accuracy: 1.0000 - val_loss: 0.1281 - val_accuracy: 0.9500\n",
            "Epoch 79/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 7.9549e-04 - accuracy: 0.9993\n",
            "Epoch 79: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 7.9549e-04 - accuracy: 0.9993 - val_loss: 0.4658 - val_accuracy: 0.9250\n",
            "Epoch 80/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0406 - accuracy: 0.9847\n",
            "Epoch 80: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 0.0406 - accuracy: 0.9847 - val_loss: 0.4773 - val_accuracy: 0.9250\n",
            "Epoch 81/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0050 - accuracy: 0.9985\n",
            "Epoch 81: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 126ms/step - loss: 0.0050 - accuracy: 0.9985 - val_loss: 0.5743 - val_accuracy: 0.9250\n",
            "Epoch 82/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0019 - accuracy: 0.9993\n",
            "Epoch 82: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 129ms/step - loss: 0.0019 - accuracy: 0.9993 - val_loss: 0.5346 - val_accuracy: 0.9000\n",
            "Epoch 83/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0017 - accuracy: 1.0000\n",
            "Epoch 83: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 126ms/step - loss: 0.0017 - accuracy: 1.0000 - val_loss: 0.3471 - val_accuracy: 0.9250\n",
            "Epoch 84/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 7.6778e-04 - accuracy: 1.0000\n",
            "Epoch 84: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 7.6778e-04 - accuracy: 1.0000 - val_loss: 0.5219 - val_accuracy: 0.9250\n",
            "Epoch 85/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 2.0567e-04 - accuracy: 1.0000\n",
            "Epoch 85: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 2.0567e-04 - accuracy: 1.0000 - val_loss: 0.4210 - val_accuracy: 0.9250\n",
            "Epoch 86/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 3.7813e-04 - accuracy: 1.0000\n",
            "Epoch 86: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 3.7813e-04 - accuracy: 1.0000 - val_loss: 0.7080 - val_accuracy: 0.9250\n",
            "Epoch 87/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 2.9276e-04 - accuracy: 1.0000\n",
            "Epoch 87: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 129ms/step - loss: 2.9276e-04 - accuracy: 1.0000 - val_loss: 0.3173 - val_accuracy: 0.9500\n",
            "Epoch 88/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 2.3019e-04 - accuracy: 1.0000\n",
            "Epoch 88: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 125ms/step - loss: 2.3019e-04 - accuracy: 1.0000 - val_loss: 0.5295 - val_accuracy: 0.9250\n",
            "Epoch 89/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 1.2351e-04 - accuracy: 1.0000\n",
            "Epoch 89: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 1.2351e-04 - accuracy: 1.0000 - val_loss: 0.4060 - val_accuracy: 0.9500\n",
            "Epoch 90/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 6.7610e-05 - accuracy: 1.0000\n",
            "Epoch 90: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 12s 134ms/step - loss: 6.7610e-05 - accuracy: 1.0000 - val_loss: 0.5056 - val_accuracy: 0.9250\n",
            "Epoch 91/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 4.9831e-05 - accuracy: 1.0000\n",
            "Epoch 91: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 4.9831e-05 - accuracy: 1.0000 - val_loss: 0.4928 - val_accuracy: 0.9250\n",
            "Epoch 92/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 2.5983e-04 - accuracy: 1.0000\n",
            "Epoch 92: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 2.5983e-04 - accuracy: 1.0000 - val_loss: 0.7055 - val_accuracy: 0.9250\n",
            "Epoch 93/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 9.9876e-05 - accuracy: 1.0000\n",
            "Epoch 93: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 9.9876e-05 - accuracy: 1.0000 - val_loss: 0.5776 - val_accuracy: 0.9250\n",
            "Epoch 94/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 6.0020e-05 - accuracy: 1.0000\n",
            "Epoch 94: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 6.0020e-05 - accuracy: 1.0000 - val_loss: 0.5444 - val_accuracy: 0.9250\n",
            "Epoch 95/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 1.1044e-04 - accuracy: 1.0000\n",
            "Epoch 95: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 129ms/step - loss: 1.1044e-04 - accuracy: 1.0000 - val_loss: 0.5708 - val_accuracy: 0.9250\n",
            "Epoch 96/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 4.7336e-05 - accuracy: 1.0000\n",
            "Epoch 96: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 129ms/step - loss: 4.7336e-05 - accuracy: 1.0000 - val_loss: 0.5369 - val_accuracy: 0.9250\n",
            "Epoch 97/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 2.3143e-05 - accuracy: 1.0000\n",
            "Epoch 97: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 130ms/step - loss: 2.3143e-05 - accuracy: 1.0000 - val_loss: 0.5712 - val_accuracy: 0.9250\n",
            "Epoch 98/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 1.5349e-05 - accuracy: 1.0000\n",
            "Epoch 98: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 1.5349e-05 - accuracy: 1.0000 - val_loss: 0.5749 - val_accuracy: 0.9250\n",
            "Epoch 99/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 3.3532e-04 - accuracy: 1.0000\n",
            "Epoch 99: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 3.3532e-04 - accuracy: 1.0000 - val_loss: 0.3094 - val_accuracy: 0.9500\n",
            "Epoch 100/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 2.8359e-05 - accuracy: 1.0000\n",
            "Epoch 100: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 2.8359e-05 - accuracy: 1.0000 - val_loss: 0.3629 - val_accuracy: 0.9250\n",
            "Epoch 101/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 3.6934e-05 - accuracy: 1.0000\n",
            "Epoch 101: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 3.6934e-05 - accuracy: 1.0000 - val_loss: 0.3907 - val_accuracy: 0.9250\n",
            "Epoch 102/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 1.8887e-04 - accuracy: 1.0000\n",
            "Epoch 102: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 129ms/step - loss: 1.8887e-04 - accuracy: 1.0000 - val_loss: 0.4468 - val_accuracy: 0.9500\n",
            "Epoch 103/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 9.2467e-05 - accuracy: 1.0000\n",
            "Epoch 103: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 130ms/step - loss: 9.2467e-05 - accuracy: 1.0000 - val_loss: 0.5039 - val_accuracy: 0.9250\n",
            "Epoch 104/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 2.0928e-05 - accuracy: 1.0000\n",
            "Epoch 104: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 130ms/step - loss: 2.0928e-05 - accuracy: 1.0000 - val_loss: 0.4326 - val_accuracy: 0.9500\n",
            "Epoch 105/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 2.0782e-05 - accuracy: 1.0000\n",
            "Epoch 105: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 2.0782e-05 - accuracy: 1.0000 - val_loss: 0.4245 - val_accuracy: 0.9250\n",
            "Epoch 106/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 4.7959e-05 - accuracy: 1.0000\n",
            "Epoch 106: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 4.7959e-05 - accuracy: 1.0000 - val_loss: 0.3077 - val_accuracy: 0.9500\n",
            "Epoch 107/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0871 - accuracy: 0.9694\n",
            "Epoch 107: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 0.0871 - accuracy: 0.9694 - val_loss: 0.4807 - val_accuracy: 0.9250\n",
            "Epoch 108/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0090 - accuracy: 0.9964\n",
            "Epoch 108: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 0.0090 - accuracy: 0.9964 - val_loss: 0.6345 - val_accuracy: 0.9000\n",
            "Epoch 109/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0156 - accuracy: 0.9949\n",
            "Epoch 109: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 0.0156 - accuracy: 0.9949 - val_loss: 0.3578 - val_accuracy: 0.9500\n",
            "Epoch 110/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0021 - accuracy: 0.9993\n",
            "Epoch 110: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 0.0021 - accuracy: 0.9993 - val_loss: 0.2720 - val_accuracy: 0.9750\n",
            "Epoch 111/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 9.8944e-04 - accuracy: 1.0000\n",
            "Epoch 111: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 126ms/step - loss: 9.8944e-04 - accuracy: 1.0000 - val_loss: 0.5587 - val_accuracy: 0.9250\n",
            "Epoch 112/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 6.1180e-04 - accuracy: 1.0000\n",
            "Epoch 112: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 6.1180e-04 - accuracy: 1.0000 - val_loss: 0.5117 - val_accuracy: 0.9500\n",
            "Epoch 113/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 3.0213e-04 - accuracy: 1.0000\n",
            "Epoch 113: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 3.0213e-04 - accuracy: 1.0000 - val_loss: 0.4391 - val_accuracy: 0.9500\n",
            "Epoch 114/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 3.7195e-04 - accuracy: 1.0000\n",
            "Epoch 114: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 3.7195e-04 - accuracy: 1.0000 - val_loss: 0.3951 - val_accuracy: 0.9250\n",
            "Epoch 115/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 6.2507e-04 - accuracy: 1.0000\n",
            "Epoch 115: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 125ms/step - loss: 6.2507e-04 - accuracy: 1.0000 - val_loss: 0.4941 - val_accuracy: 0.9500\n",
            "Epoch 116/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 1.0979e-04 - accuracy: 1.0000\n",
            "Epoch 116: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 132ms/step - loss: 1.0979e-04 - accuracy: 1.0000 - val_loss: 0.5504 - val_accuracy: 0.9250\n",
            "Epoch 117/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 2.1156e-04 - accuracy: 1.0000\n",
            "Epoch 117: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 129ms/step - loss: 2.1156e-04 - accuracy: 1.0000 - val_loss: 0.5344 - val_accuracy: 0.9500\n",
            "Epoch 118/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 8.4932e-05 - accuracy: 1.0000\n",
            "Epoch 118: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 126ms/step - loss: 8.4932e-05 - accuracy: 1.0000 - val_loss: 0.4876 - val_accuracy: 0.9500\n",
            "Epoch 119/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 9.4696e-05 - accuracy: 1.0000\n",
            "Epoch 119: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 9.4696e-05 - accuracy: 1.0000 - val_loss: 0.5757 - val_accuracy: 0.9500\n",
            "Epoch 120/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 8.3060e-05 - accuracy: 1.0000\n",
            "Epoch 120: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 8.3060e-05 - accuracy: 1.0000 - val_loss: 0.7770 - val_accuracy: 0.9250\n",
            "Epoch 121/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 1.0827e-04 - accuracy: 1.0000\n",
            "Epoch 121: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 126ms/step - loss: 1.0827e-04 - accuracy: 1.0000 - val_loss: 0.5601 - val_accuracy: 0.9500\n",
            "Epoch 122/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 1.7910e-04 - accuracy: 1.0000\n",
            "Epoch 122: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 1.7910e-04 - accuracy: 1.0000 - val_loss: 0.6403 - val_accuracy: 0.9250\n",
            "Epoch 123/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 3.2860e-05 - accuracy: 1.0000\n",
            "Epoch 123: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 3.2860e-05 - accuracy: 1.0000 - val_loss: 0.5995 - val_accuracy: 0.9500\n",
            "Epoch 124/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 6.3865e-05 - accuracy: 1.0000\n",
            "Epoch 124: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 6.3865e-05 - accuracy: 1.0000 - val_loss: 0.5454 - val_accuracy: 0.9500\n",
            "Epoch 125/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 4.1300e-05 - accuracy: 1.0000\n",
            "Epoch 125: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 129ms/step - loss: 4.1300e-05 - accuracy: 1.0000 - val_loss: 0.5257 - val_accuracy: 0.9500\n",
            "Epoch 126/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 2.4237e-05 - accuracy: 1.0000\n",
            "Epoch 126: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 126ms/step - loss: 2.4237e-05 - accuracy: 1.0000 - val_loss: 0.5316 - val_accuracy: 0.9500\n",
            "Epoch 127/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 4.2707e-05 - accuracy: 1.0000\n",
            "Epoch 127: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 131ms/step - loss: 4.2707e-05 - accuracy: 1.0000 - val_loss: 0.5938 - val_accuracy: 0.9500\n",
            "Epoch 128/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 1.6081e-05 - accuracy: 1.0000\n",
            "Epoch 128: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 1.6081e-05 - accuracy: 1.0000 - val_loss: 0.5835 - val_accuracy: 0.9500\n",
            "Epoch 129/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 2.9633e-05 - accuracy: 1.0000\n",
            "Epoch 129: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 2.9633e-05 - accuracy: 1.0000 - val_loss: 0.5665 - val_accuracy: 0.9500\n",
            "Epoch 130/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 1.3165e-05 - accuracy: 1.0000\n",
            "Epoch 130: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 126ms/step - loss: 1.3165e-05 - accuracy: 1.0000 - val_loss: 0.5490 - val_accuracy: 0.9500\n",
            "Epoch 131/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 7.6262e-05 - accuracy: 1.0000\n",
            "Epoch 131: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 7.6262e-05 - accuracy: 1.0000 - val_loss: 0.5944 - val_accuracy: 0.9250\n",
            "Epoch 132/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 5.8385e-05 - accuracy: 1.0000\n",
            "Epoch 132: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 129ms/step - loss: 5.8385e-05 - accuracy: 1.0000 - val_loss: 0.5457 - val_accuracy: 0.9500\n",
            "Epoch 133/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 1.7771e-05 - accuracy: 1.0000\n",
            "Epoch 133: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 1.7771e-05 - accuracy: 1.0000 - val_loss: 0.5683 - val_accuracy: 0.9500\n",
            "Epoch 134/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 1.6110e-04 - accuracy: 1.0000\n",
            "Epoch 134: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 1.6110e-04 - accuracy: 1.0000 - val_loss: 0.4430 - val_accuracy: 0.9500\n",
            "Epoch 135/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0881 - accuracy: 0.9730\n",
            "Epoch 135: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 130ms/step - loss: 0.0881 - accuracy: 0.9730 - val_loss: 0.2207 - val_accuracy: 0.9500\n",
            "Epoch 136/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0329 - accuracy: 0.9876\n",
            "Epoch 136: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 0.0329 - accuracy: 0.9876 - val_loss: 0.4503 - val_accuracy: 0.9250\n",
            "Epoch 137/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0014 - accuracy: 1.0000\n",
            "Epoch 137: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 0.0014 - accuracy: 1.0000 - val_loss: 0.2347 - val_accuracy: 0.9500\n",
            "Epoch 138/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 4.5349e-04 - accuracy: 1.0000\n",
            "Epoch 138: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 127ms/step - loss: 4.5349e-04 - accuracy: 1.0000 - val_loss: 0.3032 - val_accuracy: 0.9500\n",
            "Epoch 139/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0019 - accuracy: 1.0000\n",
            "Epoch 139: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 14s 159ms/step - loss: 0.0019 - accuracy: 1.0000 - val_loss: 0.2893 - val_accuracy: 0.9250\n",
            "Epoch 140/140\n",
            "86/86 [==============================] - ETA: 0s - loss: 0.0042 - accuracy: 0.9985\n",
            "Epoch 140: val_loss did not improve from 0.12815\n",
            "86/86 [==============================] - 11s 128ms/step - loss: 0.0042 - accuracy: 0.9985 - val_loss: 0.5140 - val_accuracy: 0.8750\n"
          ]
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "End of Training, the model is saved to embedding_network.h5\n"
          ]
        }
      ],
      "source": [
        "\"\"\" Train the model \"\"\"\n",
        "    \n",
        "history = embedding_network.fit(\n",
        "    train_batches,\n",
        "    validation_data = val_batches,\n",
        "    epochs = 140,\n",
        "    verbose = 1,\n",
        "    shuffle = True,\n",
        "    callbacks = [checkpointer]#, early_stopping]\n",
        ")\n",
        "    \n",
        "\"\"\" plot the train and val accuracies \"\"\"\n",
        "# Plot the accuracy\n",
        "utils.plt_metric(history=history.history, metric=\"accuracy\", title=\"Model accuracy\")\n",
        "    \n",
        "# Plot the constrastive loss\n",
        "utils.plt_metric(history=history.history, metric=\"loss\", title=\"Constrastive Loss\")\n",
        "\n",
        "    \n",
        "print(\"End of Training, the model is saved to\", MODEL_FNAME)  "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 10,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "7JLVdDVI179X",
        "outputId": "b598e640-9039-479f-80d3-0cdcc52977f3"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Found 588 images belonging to 4 classes.\n",
            "\n",
            "The test set contains:\n",
            "• 147 glioma images\n",
            "• 147 meningioma images\n",
            "• 147 normal images\n",
            "• 147 pituitary images\n",
            "Model: \"model\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_1 (InputLayer)        [(None, 150, 150, 3)]     0         \n",
            "                                                                 \n",
            " block1_conv1 (Conv2D)       (None, 150, 150, 64)      1792      \n",
            "                                                                 \n",
            " block1_conv2 (Conv2D)       (None, 150, 150, 64)      36928     \n",
            "                                                                 \n",
            " block1_pool (MaxPooling2D)  (None, 75, 75, 64)        0         \n",
            "                                                                 \n",
            " block2_conv1 (Conv2D)       (None, 75, 75, 128)       73856     \n",
            "                                                                 \n",
            " block2_conv2 (Conv2D)       (None, 75, 75, 128)       147584    \n",
            "                                                                 \n",
            " block2_pool (MaxPooling2D)  (None, 37, 37, 128)       0         \n",
            "                                                                 \n",
            " block3_conv1 (Conv2D)       (None, 37, 37, 256)       295168    \n",
            "                                                                 \n",
            " block3_conv2 (Conv2D)       (None, 37, 37, 256)       590080    \n",
            "                                                                 \n",
            " block3_conv3 (Conv2D)       (None, 37, 37, 256)       590080    \n",
            "                                                                 \n",
            " block3_pool (MaxPooling2D)  (None, 18, 18, 256)       0         \n",
            "                                                                 \n",
            " block4_conv1 (Conv2D)       (None, 18, 18, 512)       1180160   \n",
            "                                                                 \n",
            " block4_conv2 (Conv2D)       (None, 18, 18, 512)       2359808   \n",
            "                                                                 \n",
            " block4_conv3 (Conv2D)       (None, 18, 18, 512)       2359808   \n",
            "                                                                 \n",
            " block4_pool (MaxPooling2D)  (None, 9, 9, 512)         0         \n",
            "                                                                 \n",
            " block5_conv1 (Conv2D)       (None, 9, 9, 512)         2359808   \n",
            "                                                                 \n",
            " block5_conv2 (Conv2D)       (None, 9, 9, 512)         2359808   \n",
            "                                                                 \n",
            " block5_conv3 (Conv2D)       (None, 9, 9, 512)         2359808   \n",
            "                                                                 \n",
            " block5_pool (MaxPooling2D)  (None, 4, 4, 512)         0         \n",
            "                                                                 \n",
            " global_average_pooling2d (G  (None, 512)              0         \n",
            " lobalAveragePooling2D)                                          \n",
            "                                                                 \n",
            " dropout (Dropout)           (None, 512)               0         \n",
            "                                                                 \n",
            " flatten (Flatten)           (None, 512)               0         \n",
            "                                                                 \n",
            " dense (Dense)               (None, 4)                 2052      \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 14,716,740\n",
            "Trainable params: 14,716,740\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "\n",
            "Evaluate on test data\n",
            "Accuracy: 0.9370748299319728\n",
            "Precision: 0.9371511147732052\n",
            "Recall: 0.9370748299319728\n"
          ]
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 2 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "test_batches = ImageDataGenerator(rescale = 1 / 255.).flow_from_directory(test_dir,\n",
        "                                                        target_size=(INPUT_SIZE,INPUT_SIZE),\n",
        "                                                        class_mode='categorical',\n",
        "                                                        shuffle=False,\n",
        "                                                        seed=42,\n",
        "                                                        batch_size=BATCH_SIZE)\n",
        "    \n",
        "\n",
        "test_glioma_dir = os.path.join(test_dir, 'glioma_tumor')\n",
        "test_meningioma_dir = os.path.join(test_dir, 'meningioma_tumor')\n",
        "test_normal_dir = os.path.join(test_dir, 'no_tumor')\n",
        "test_pituitary_dir = os.path.join(test_dir, 'pituitary_tumor')\n",
        "    \n",
        "num_glioma_test = int(len(os.listdir(test_glioma_dir)))\n",
        "num_meningioma_test = int(len(os.listdir(test_meningioma_dir)))\n",
        "num_normal_test = int(len(os.listdir(test_normal_dir)))\n",
        "num_pituitary_test = int(len(os.listdir(test_pituitary_dir)))\n",
        "\n",
        "print('\\nThe test set contains:')\n",
        "print('\\u2022 {} glioma images'.format(num_glioma_test))\n",
        "print('\\u2022 {} meningioma images'.format(num_meningioma_test))\n",
        "print('\\u2022 {} normal images'.format(num_normal_test))\n",
        "print('\\u2022 {} pituitary images'.format(num_pituitary_test))\n",
        "\n",
        "\"\"\" Test the model \"\"\"\n",
        "model = tf.keras.models.load_model(MODEL_FNAME)\n",
        "model.summary()\n",
        "    \n",
        "y_test = test_batches.classes\n",
        "    \n",
        "#Confution Matrix and Classification Report\n",
        "Y_pred = model.predict_generator(test_batches, (num_glioma_test + num_meningioma_test + num_normal_test + num_pituitary_test) // BATCH_SIZE+1)\n",
        "y_pred = np.argmax(Y_pred, axis=1)\n",
        "    \n",
        "class_labels = list(test_batches.class_indices.keys())   \n",
        "    \n",
        "cm = confusion_matrix(test_batches.classes, y_pred)    \n",
        "cm_display = ConfusionMatrixDisplay(confusion_matrix=cm, display_labels=class_labels).plot()\n",
        "    \n",
        "# results = model.evaluate_generator(test_batches)\n",
        "print(\"\\nEvaluate on test data\")\n",
        "print(\"Accuracy:\", accuracy_score(y_test, y_pred))\n",
        "print(\"Precision:\", precision_score(y_test, y_pred, average='weighted'))\n",
        "print(\"Recall:\", recall_score(y_test, y_pred, average='weighted'))\n",
        "    \n",
        "tf.keras.backend.clear_session()"
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {
        "id": "kljGV_L72ja5"
      },
      "source": [
        "Siamese Network"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rkZn3bx9e35o",
        "outputId": "66a5ed9c-790a-4aea-e8b6-5a65d568c83c"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Using:\n",
            "\t• TensorFlow version: 2.12.0\n",
            "\t• tf.keras version: 2.12.0\n",
            "\t• Running on GPU\n"
          ]
        }
      ],
      "source": [
        "# -*- coding: utf-8 -*-\n",
        "import utils\n",
        "\n",
        "import tensorflow as tf\n",
        "from tensorflow.keras.layers import Dense, Input, Flatten, Lambda\n",
        "from tensorflow.keras.models import Model\n",
        "from tensorflow.keras.optimizers import Adam\n",
        "from sklearn.metrics import confusion_matrix, ConfusionMatrixDisplay, accuracy_score, precision_score, recall_score, roc_auc_score, f1_score\n",
        "\n",
        "import numpy as np\n",
        "import os\n",
        "\n",
        "from keras import backend as K\n",
        "from tensorflow.keras.callbacks import EarlyStopping, ModelCheckpoint, ReduceLROnPlateau\n",
        "\n",
        "import logging\n",
        "\n",
        "MARGIN = 0.2\n",
        "\n",
        "tf.random.set_seed(0)\n",
        "tf.keras.utils.set_random_seed(0)\n",
        "\n",
        "logger = tf.get_logger()\n",
        "logger.setLevel(logging.ERROR)\n",
        "\n",
        "print('Using:')\n",
        "print('\\t\\u2022 TensorFlow version:', tf.__version__)\n",
        "print('\\t\\u2022 tf.keras version:', tf.keras.__version__)\n",
        "print('\\t\\u2022 Running on GPU' if tf.test.is_gpu_available() else '\\t\\u2022 GPU device not found. Running on CPU')\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "NLPSgD3se-8y",
        "outputId": "0ca7f209-17e1-4aad-833c-afaa85f451e4"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Found 40 images belonging to 4 classes.\n",
            "The train set contains 40\n",
            "Found 40 images belonging to 4 classes.\n",
            "The valid set contains 40\n",
            "Found 1920 images belonging to 4 classes.\n",
            "The test set contains 1920\n",
            "number of triples for training 40\n",
            "number of triples for validation 40\n",
            "number of triples for test 1920\n"
          ]
        }
      ],
      "source": [
        "INPUT_SIZE = 150\n",
        "\n",
        "basedir = os.path.join(\"dataset\", \"siamese\") \n",
        "\n",
        "train_image_list, train_y_list = utils.load_images(basedir, 'train', (INPUT_SIZE,INPUT_SIZE))\n",
        "print(\"The train set contains\",len(train_image_list)) \n",
        "\n",
        "valid_image_list, valid_y_list = utils.load_images(basedir, 'validation', (INPUT_SIZE,INPUT_SIZE))   \n",
        "print(\"The valid set contains\", len(valid_image_list))  \n",
        "\n",
        "test_image_list, test_y_list = utils.load_images(basedir, 'test', (INPUT_SIZE,INPUT_SIZE))   \n",
        "print(\"The test set contains\", len(test_image_list))  \n",
        "\n",
        "# make train triples\n",
        "triples_train, labels_train = utils.make_triples(train_image_list, train_y_list)\n",
        "\n",
        "# make validation triples\n",
        "triples_val, labels_val = utils.make_triples(valid_image_list, valid_y_list)\n",
        "\n",
        "# make test triples\n",
        "triples_test, labels_test = utils.make_triples(test_image_list, test_y_list)\n",
        "\n",
        "x_train_1 = triples_train[:, 0]  \n",
        "x_train_2 = triples_train[:, 1]\n",
        "x_train_3 = triples_train[:, 2]\n",
        "print(\"number of triples for training\", np.shape(x_train_1)[0]) \n",
        "\n",
        "x_val_1 = triples_val[:, 0] \n",
        "x_val_2 = triples_val[:, 1]\n",
        "x_val_3 = triples_val[:, 2]\n",
        "print(\"number of triples for validation\", np.shape(x_val_1)[0]) \n",
        "\n",
        "x_test_1 = triples_test[:, 0] \n",
        "x_test_2 = triples_test[:, 1]\n",
        "x_test_3 = triples_test[:, 2]\n",
        "print(\"number of triples for test\", np.shape(x_test_1)[0]) \n",
        "\n",
        "# utils.visualize(triples_train[:-1], labels_train[:-1], to_show=4, num_col=4)\n",
        "\n",
        "tf.compat.v1.reset_default_graph()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "z9iioQiRfDfZ",
        "outputId": "3b8152b8-5d3c-43a4-a930-75f52c548562"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Model: \"model\"\n",
            "__________________________________________________________________________________________________\n",
            " Layer (type)                   Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            " anchor_input (InputLayer)      [(None, 150, 150, 3  0           []                               \n",
            "                                )]                                                                \n",
            "                                                                                                  \n",
            " positive_input (InputLayer)    [(None, 150, 150, 3  0           []                               \n",
            "                                )]                                                                \n",
            "                                                                                                  \n",
            " negative_input (InputLayer)    [(None, 150, 150, 3  0           []                               \n",
            "                                )]                                                                \n",
            "                                                                                                  \n",
            " sequential (Sequential)        (None, 5120)         14742340    ['anchor_input[0][0]',           \n",
            "                                                                  'positive_input[0][0]',         \n",
            "                                                                  'negative_input[0][0]']         \n",
            "                                                                                                  \n",
            " lambda (Lambda)                (None,)              0           ['sequential[0][0]',             \n",
            "                                                                  'sequential[1][0]',             \n",
            "                                                                  'sequential[2][0]']             \n",
            "                                                                                                  \n",
            "==================================================================================================\n",
            "Total params: 14,742,340\n",
            "Trainable params: 25,600\n",
            "Non-trainable params: 14,716,740\n",
            "__________________________________________________________________________________________________\n"
          ]
        }
      ],
      "source": [
        "SIAMESE_MODEL_FNAME = 'siamese_network.h5'\n",
        "EMBEDDING_MODEL_FNAME = 'embedding_network.h5'\n",
        "\n",
        "\n",
        "embedding_network = tf.keras.models.load_model(EMBEDDING_MODEL_FNAME)\n",
        "embedding_network.trainable = False\n",
        "\n",
        "model = tf.keras.Sequential() \n",
        "for layer in embedding_network.layers:  \n",
        "    model.add(layer) \n",
        "\n",
        "model.add(Flatten(name='flat'))\n",
        "model.add(Dense(5120, name='den', activation='sigmoid', kernel_regularizer='l2')) \n",
        "\n",
        "\n",
        "\n",
        "# Define the tensors for the triplet of input images\n",
        "anchor_input = tf.keras.layers.Input((INPUT_SIZE,INPUT_SIZE,3), name=\"anchor_input\")\n",
        "positive_input = tf.keras.layers.Input((INPUT_SIZE,INPUT_SIZE,3), name=\"positive_input\")\n",
        "negative_input = tf.keras.layers.Input((INPUT_SIZE,INPUT_SIZE,3), name=\"negative_input\")\n",
        "\n",
        "# Generate the embedding outputs \n",
        "encoded_anchor = model(anchor_input)\n",
        "encoded_positive = model(positive_input)\n",
        "encoded_negative = model(negative_input)\n",
        "\n",
        "inputs = [anchor_input, positive_input, negative_input]\n",
        "outputs = [encoded_anchor, encoded_positive, encoded_negative]\n",
        "#x = tf.keras.layers.Lambda(lambda x: tf.math.l2_normalize(outputs, axis=1))(outputs)\n",
        "\n",
        "\n",
        "loss = Lambda(utils.loss2(MARGIN))(outputs) \n",
        "# Connect the inputs with the outputs\n",
        "siamese = Model(inputs=inputs,outputs=loss)\n",
        "\n",
        "siamese.summary()\n",
        "\n",
        "\"\"\" callbacks \"\"\"\n",
        "\n",
        "reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=5, min_lr=0.00001)\n",
        "\n",
        "early_stopping = EarlyStopping(monitor='val_loss', patience=10, verbose=1, min_delta=0.0001)\n",
        "\n",
        "checkpointer = ModelCheckpoint(filepath=SIAMESE_MODEL_FNAME, verbose=1, \n",
        "                                save_best_only=True)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 16,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "z2FaS_sof4Dk",
        "outputId": "d383a998-4190-49b2-81b3-bc5617511c03"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Model: \"model\"\n",
            "__________________________________________________________________________________________________\n",
            " Layer (type)                   Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            " anchor_input (InputLayer)      [(None, 150, 150, 3  0           []                               \n",
            "                                )]                                                                \n",
            "                                                                                                  \n",
            " positive_input (InputLayer)    [(None, 150, 150, 3  0           []                               \n",
            "                                )]                                                                \n",
            "                                                                                                  \n",
            " negative_input (InputLayer)    [(None, 150, 150, 3  0           []                               \n",
            "                                )]                                                                \n",
            "                                                                                                  \n",
            " sequential_1 (Sequential)      (None, 5120)         14742340    ['anchor_input[0][0]',           \n",
            "                                                                  'positive_input[0][0]',         \n",
            "                                                                  'negative_input[0][0]']         \n",
            "                                                                                                  \n",
            " lambda (Lambda)                (None,)              0           ['sequential_1[0][0]',           \n",
            "                                                                  'sequential_1[1][0]',           \n",
            "                                                                  'sequential_1[2][0]']           \n",
            "                                                                                                  \n",
            "==================================================================================================\n",
            "Total params: 14,742,340\n",
            "Trainable params: 25,600\n",
            "Non-trainable params: 14,716,740\n",
            "__________________________________________________________________________________________________\n",
            "Epoch 1/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.1130\n",
            "Epoch 1: val_loss improved from inf to 0.09323, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 4s 51ms/step - loss: 0.1130 - val_loss: 0.0932 - lr: 1.0000e-04\n",
            "Epoch 2/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.1006\n",
            "Epoch 2: val_loss improved from 0.09323 to 0.08998, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 55ms/step - loss: 0.1006 - val_loss: 0.0900 - lr: 1.0000e-04\n",
            "Epoch 3/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0994\n",
            "Epoch 3: val_loss improved from 0.08998 to 0.08913, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 42ms/step - loss: 0.0994 - val_loss: 0.0891 - lr: 1.0000e-04\n",
            "Epoch 4/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0907\n",
            "Epoch 4: val_loss improved from 0.08913 to 0.08538, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 58ms/step - loss: 0.0907 - val_loss: 0.0854 - lr: 1.0000e-04\n",
            "Epoch 5/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0936\n",
            "Epoch 5: val_loss improved from 0.08538 to 0.08411, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 57ms/step - loss: 0.0936 - val_loss: 0.0841 - lr: 1.0000e-04\n",
            "Epoch 6/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0894\n",
            "Epoch 6: val_loss did not improve from 0.08411\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0894 - val_loss: 0.0842 - lr: 1.0000e-04\n",
            "Epoch 7/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0903\n",
            "Epoch 7: val_loss did not improve from 0.08411\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0903 - val_loss: 0.0844 - lr: 1.0000e-04\n",
            "Epoch 8/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0910\n",
            "Epoch 8: val_loss improved from 0.08411 to 0.08253, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 42ms/step - loss: 0.0910 - val_loss: 0.0825 - lr: 1.0000e-04\n",
            "Epoch 9/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0826\n",
            "Epoch 9: val_loss improved from 0.08253 to 0.08095, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 42ms/step - loss: 0.0826 - val_loss: 0.0809 - lr: 1.0000e-04\n",
            "Epoch 10/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0810\n",
            "Epoch 10: val_loss improved from 0.08095 to 0.08069, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 43ms/step - loss: 0.0810 - val_loss: 0.0807 - lr: 1.0000e-04\n",
            "Epoch 11/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0896\n",
            "Epoch 11: val_loss improved from 0.08069 to 0.08004, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 58ms/step - loss: 0.0896 - val_loss: 0.0800 - lr: 1.0000e-04\n",
            "Epoch 12/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0814\n",
            "Epoch 12: val_loss improved from 0.08004 to 0.07935, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 42ms/step - loss: 0.0814 - val_loss: 0.0794 - lr: 1.0000e-04\n",
            "Epoch 13/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0850\n",
            "Epoch 13: val_loss improved from 0.07935 to 0.07856, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 55ms/step - loss: 0.0850 - val_loss: 0.0786 - lr: 1.0000e-04\n",
            "Epoch 14/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0880\n",
            "Epoch 14: val_loss did not improve from 0.07856\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0880 - val_loss: 0.0790 - lr: 1.0000e-04\n",
            "Epoch 15/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0878\n",
            "Epoch 15: val_loss improved from 0.07856 to 0.07839, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 41ms/step - loss: 0.0878 - val_loss: 0.0784 - lr: 1.0000e-04\n",
            "Epoch 16/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0854\n",
            "Epoch 16: val_loss improved from 0.07839 to 0.07823, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 55ms/step - loss: 0.0854 - val_loss: 0.0782 - lr: 1.0000e-04\n",
            "Epoch 17/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0859\n",
            "Epoch 17: val_loss did not improve from 0.07823\n",
            "40/40 [==============================] - 2s 41ms/step - loss: 0.0859 - val_loss: 0.0794 - lr: 1.0000e-04\n",
            "Epoch 18/175\n",
            "38/40 [===========================>..] - ETA: 0s - loss: 0.0851\n",
            "Epoch 18: val_loss improved from 0.07823 to 0.07811, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 51ms/step - loss: 0.0834 - val_loss: 0.0781 - lr: 1.0000e-04\n",
            "Epoch 19/175\n",
            "38/40 [===========================>..] - ETA: 0s - loss: 0.0862\n",
            "Epoch 19: val_loss improved from 0.07811 to 0.07736, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 56ms/step - loss: 0.0846 - val_loss: 0.0774 - lr: 1.0000e-04\n",
            "Epoch 20/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0830\n",
            "Epoch 20: val_loss did not improve from 0.07736\n",
            "40/40 [==============================] - 2s 51ms/step - loss: 0.0830 - val_loss: 0.0778 - lr: 1.0000e-04\n",
            "Epoch 21/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0818\n",
            "Epoch 21: val_loss did not improve from 0.07736\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0818 - val_loss: 0.0782 - lr: 1.0000e-04\n",
            "Epoch 22/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0814\n",
            "Epoch 22: val_loss did not improve from 0.07736\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0814 - val_loss: 0.0781 - lr: 1.0000e-04\n",
            "Epoch 23/175\n",
            "38/40 [===========================>..] - ETA: 0s - loss: 0.0740\n",
            "Epoch 23: val_loss did not improve from 0.07736\n",
            "40/40 [==============================] - 1s 37ms/step - loss: 0.0814 - val_loss: 0.0783 - lr: 1.0000e-04\n",
            "Epoch 24/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0822\n",
            "Epoch 24: val_loss improved from 0.07736 to 0.07731, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 42ms/step - loss: 0.0822 - val_loss: 0.0773 - lr: 1.0000e-04\n",
            "Epoch 25/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0773\n",
            "Epoch 25: val_loss improved from 0.07731 to 0.07704, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 43ms/step - loss: 0.0773 - val_loss: 0.0770 - lr: 2.0000e-05\n",
            "Epoch 26/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0872\n",
            "Epoch 26: val_loss did not improve from 0.07704\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0872 - val_loss: 0.0771 - lr: 2.0000e-05\n",
            "Epoch 27/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0832\n",
            "Epoch 27: val_loss did not improve from 0.07704\n",
            "40/40 [==============================] - 1s 37ms/step - loss: 0.0832 - val_loss: 0.0772 - lr: 2.0000e-05\n",
            "Epoch 28/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0774\n",
            "Epoch 28: val_loss improved from 0.07704 to 0.07691, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 55ms/step - loss: 0.0774 - val_loss: 0.0769 - lr: 2.0000e-05\n",
            "Epoch 29/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0802\n",
            "Epoch 29: val_loss did not improve from 0.07691\n",
            "40/40 [==============================] - 1s 37ms/step - loss: 0.0802 - val_loss: 0.0771 - lr: 2.0000e-05\n",
            "Epoch 30/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0846\n",
            "Epoch 30: val_loss did not improve from 0.07691\n",
            "40/40 [==============================] - 2s 51ms/step - loss: 0.0846 - val_loss: 0.0772 - lr: 2.0000e-05\n",
            "Epoch 31/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0753\n",
            "Epoch 31: val_loss did not improve from 0.07691\n",
            "40/40 [==============================] - 1s 37ms/step - loss: 0.0753 - val_loss: 0.0770 - lr: 2.0000e-05\n",
            "Epoch 32/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0801\n",
            "Epoch 32: val_loss did not improve from 0.07691\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0801 - val_loss: 0.0769 - lr: 2.0000e-05\n",
            "Epoch 33/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0832\n",
            "Epoch 33: val_loss did not improve from 0.07691\n",
            "40/40 [==============================] - 2s 53ms/step - loss: 0.0832 - val_loss: 0.0770 - lr: 2.0000e-05\n",
            "Epoch 34/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0777\n",
            "Epoch 34: val_loss did not improve from 0.07691\n",
            "40/40 [==============================] - 1s 37ms/step - loss: 0.0777 - val_loss: 0.0769 - lr: 1.0000e-05\n",
            "Epoch 35/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0896\n",
            "Epoch 35: val_loss did not improve from 0.07691\n",
            "40/40 [==============================] - 1s 37ms/step - loss: 0.0896 - val_loss: 0.0770 - lr: 1.0000e-05\n",
            "Epoch 36/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0842\n",
            "Epoch 36: val_loss did not improve from 0.07691\n",
            "40/40 [==============================] - 1s 37ms/step - loss: 0.0842 - val_loss: 0.0770 - lr: 1.0000e-05\n",
            "Epoch 37/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0800\n",
            "Epoch 37: val_loss did not improve from 0.07691\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0800 - val_loss: 0.0770 - lr: 1.0000e-05\n",
            "Epoch 38/175\n",
            "38/40 [===========================>..] - ETA: 0s - loss: 0.0842\n",
            "Epoch 38: val_loss did not improve from 0.07691\n",
            "40/40 [==============================] - 1s 37ms/step - loss: 0.0824 - val_loss: 0.0769 - lr: 1.0000e-05\n",
            "Epoch 39/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0869\n",
            "Epoch 39: val_loss did not improve from 0.07691\n",
            "40/40 [==============================] - 1s 37ms/step - loss: 0.0869 - val_loss: 0.0769 - lr: 1.0000e-05\n",
            "Epoch 40/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0867\n",
            "Epoch 40: val_loss improved from 0.07691 to 0.07689, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 43ms/step - loss: 0.0867 - val_loss: 0.0769 - lr: 1.0000e-05\n",
            "Epoch 41/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0834\n",
            "Epoch 41: val_loss improved from 0.07689 to 0.07688, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 56ms/step - loss: 0.0834 - val_loss: 0.0769 - lr: 1.0000e-05\n",
            "Epoch 42/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0880\n",
            "Epoch 42: val_loss did not improve from 0.07688\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0880 - val_loss: 0.0770 - lr: 1.0000e-05\n",
            "Epoch 43/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0838\n",
            "Epoch 43: val_loss did not improve from 0.07688\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0838 - val_loss: 0.0769 - lr: 1.0000e-05\n",
            "Epoch 44/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0825\n",
            "Epoch 44: val_loss improved from 0.07688 to 0.07683, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 41ms/step - loss: 0.0825 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 45/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0820\n",
            "Epoch 45: val_loss improved from 0.07683 to 0.07683, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 41ms/step - loss: 0.0820 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 46/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0828\n",
            "Epoch 46: val_loss did not improve from 0.07683\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0828 - val_loss: 0.0769 - lr: 1.0000e-05\n",
            "Epoch 47/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0821\n",
            "Epoch 47: val_loss improved from 0.07683 to 0.07682, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 42ms/step - loss: 0.0821 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 48/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0833\n",
            "Epoch 48: val_loss improved from 0.07682 to 0.07680, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 57ms/step - loss: 0.0833 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 49/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0862\n",
            "Epoch 49: val_loss did not improve from 0.07680\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0862 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 50/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0829\n",
            "Epoch 50: val_loss did not improve from 0.07680\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0829 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 51/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0810\n",
            "Epoch 51: val_loss improved from 0.07680 to 0.07679, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 55ms/step - loss: 0.0810 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 52/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0835\n",
            "Epoch 52: val_loss did not improve from 0.07679\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0835 - val_loss: 0.0769 - lr: 1.0000e-05\n",
            "Epoch 53/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0838\n",
            "Epoch 53: val_loss did not improve from 0.07679\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0838 - val_loss: 0.0770 - lr: 1.0000e-05\n",
            "Epoch 54/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0850\n",
            "Epoch 54: val_loss did not improve from 0.07679\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0850 - val_loss: 0.0769 - lr: 1.0000e-05\n",
            "Epoch 55/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0773\n",
            "Epoch 55: val_loss did not improve from 0.07679\n",
            "40/40 [==============================] - 2s 53ms/step - loss: 0.0773 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 56/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0895\n",
            "Epoch 56: val_loss did not improve from 0.07679\n",
            "40/40 [==============================] - 2s 53ms/step - loss: 0.0895 - val_loss: 0.0770 - lr: 1.0000e-05\n",
            "Epoch 57/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0871\n",
            "Epoch 57: val_loss did not improve from 0.07679\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0871 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 58/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0750\n",
            "Epoch 58: val_loss improved from 0.07679 to 0.07679, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 55ms/step - loss: 0.0750 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 59/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0813\n",
            "Epoch 59: val_loss did not improve from 0.07679\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0813 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 60/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0803\n",
            "Epoch 60: val_loss did not improve from 0.07679\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0803 - val_loss: 0.0769 - lr: 1.0000e-05\n",
            "Epoch 61/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0736\n",
            "Epoch 61: val_loss did not improve from 0.07679\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0736 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 62/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0827\n",
            "Epoch 62: val_loss did not improve from 0.07679\n",
            "40/40 [==============================] - 2s 53ms/step - loss: 0.0827 - val_loss: 0.0769 - lr: 1.0000e-05\n",
            "Epoch 63/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0882\n",
            "Epoch 63: val_loss did not improve from 0.07679\n",
            "40/40 [==============================] - 2s 53ms/step - loss: 0.0882 - val_loss: 0.0770 - lr: 1.0000e-05\n",
            "Epoch 64/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0870\n",
            "Epoch 64: val_loss did not improve from 0.07679\n",
            "40/40 [==============================] - 1s 37ms/step - loss: 0.0870 - val_loss: 0.0769 - lr: 1.0000e-05\n",
            "Epoch 65/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0820\n",
            "Epoch 65: val_loss did not improve from 0.07679\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0820 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 66/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0828\n",
            "Epoch 66: val_loss improved from 0.07679 to 0.07669, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 41ms/step - loss: 0.0828 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 67/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0872\n",
            "Epoch 67: val_loss did not improve from 0.07669\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0872 - val_loss: 0.0769 - lr: 1.0000e-05\n",
            "Epoch 68/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0785\n",
            "Epoch 68: val_loss did not improve from 0.07669\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0785 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 69/175\n",
            "38/40 [===========================>..] - ETA: 0s - loss: 0.0785\n",
            "Epoch 69: val_loss did not improve from 0.07669\n",
            "40/40 [==============================] - 2s 53ms/step - loss: 0.0770 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 70/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0819\n",
            "Epoch 70: val_loss did not improve from 0.07669\n",
            "40/40 [==============================] - 2s 53ms/step - loss: 0.0819 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 71/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0825\n",
            "Epoch 71: val_loss did not improve from 0.07669\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0825 - val_loss: 0.0769 - lr: 1.0000e-05\n",
            "Epoch 72/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0842\n",
            "Epoch 72: val_loss did not improve from 0.07669\n",
            "40/40 [==============================] - 1s 37ms/step - loss: 0.0842 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 73/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0856\n",
            "Epoch 73: val_loss did not improve from 0.07669\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0856 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 74/175\n",
            "38/40 [===========================>..] - ETA: 0s - loss: 0.0797\n",
            "Epoch 74: val_loss did not improve from 0.07669\n",
            "40/40 [==============================] - 1s 37ms/step - loss: 0.0785 - val_loss: 0.0770 - lr: 1.0000e-05\n",
            "Epoch 75/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0839\n",
            "Epoch 75: val_loss did not improve from 0.07669\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0839 - val_loss: 0.0770 - lr: 1.0000e-05\n",
            "Epoch 76/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0768\n",
            "Epoch 76: val_loss did not improve from 0.07669\n",
            "40/40 [==============================] - 2s 38ms/step - loss: 0.0768 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 77/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0766\n",
            "Epoch 77: val_loss did not improve from 0.07669\n",
            "40/40 [==============================] - 2s 53ms/step - loss: 0.0766 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 78/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0831\n",
            "Epoch 78: val_loss did not improve from 0.07669\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0831 - val_loss: 0.0770 - lr: 1.0000e-05\n",
            "Epoch 79/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0866\n",
            "Epoch 79: val_loss did not improve from 0.07669\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0866 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 80/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0875\n",
            "Epoch 80: val_loss did not improve from 0.07669\n",
            "40/40 [==============================] - 2s 39ms/step - loss: 0.0875 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 81/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0885\n",
            "Epoch 81: val_loss did not improve from 0.07669\n",
            "40/40 [==============================] - 2s 39ms/step - loss: 0.0885 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 82/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0829\n",
            "Epoch 82: val_loss did not improve from 0.07669\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0829 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 83/175\n",
            "39/40 [============================>.] - ETA: 0s - loss: 0.0816\n",
            "Epoch 83: val_loss did not improve from 0.07669\n",
            "40/40 [==============================] - 2s 38ms/step - loss: 0.0807 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 84/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0719\n",
            "Epoch 84: val_loss did not improve from 0.07669\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0719 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 85/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0799\n",
            "Epoch 85: val_loss did not improve from 0.07669\n",
            "40/40 [==============================] - 2s 45ms/step - loss: 0.0799 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 86/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0796\n",
            "Epoch 86: val_loss did not improve from 0.07669\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0796 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 87/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0831\n",
            "Epoch 87: val_loss did not improve from 0.07669\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0831 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 88/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0843\n",
            "Epoch 88: val_loss did not improve from 0.07669\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0843 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 89/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0833\n",
            "Epoch 89: val_loss did not improve from 0.07669\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0833 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 90/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0831\n",
            "Epoch 90: val_loss did not improve from 0.07669\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0831 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 91/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0768\n",
            "Epoch 91: val_loss improved from 0.07669 to 0.07668, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 43ms/step - loss: 0.0768 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 92/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0837\n",
            "Epoch 92: val_loss did not improve from 0.07668\n",
            "40/40 [==============================] - 2s 39ms/step - loss: 0.0837 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 93/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0749\n",
            "Epoch 93: val_loss did not improve from 0.07668\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0749 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 94/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0826\n",
            "Epoch 94: val_loss did not improve from 0.07668\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0826 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 95/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0865\n",
            "Epoch 95: val_loss did not improve from 0.07668\n",
            "40/40 [==============================] - 2s 38ms/step - loss: 0.0865 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 96/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0854\n",
            "Epoch 96: val_loss did not improve from 0.07668\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0854 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 97/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0784\n",
            "Epoch 97: val_loss did not improve from 0.07668\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0784 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 98/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0887\n",
            "Epoch 98: val_loss did not improve from 0.07668\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0887 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 99/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0828\n",
            "Epoch 99: val_loss did not improve from 0.07668\n",
            "40/40 [==============================] - 2s 53ms/step - loss: 0.0828 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 100/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0809\n",
            "Epoch 100: val_loss did not improve from 0.07668\n",
            "40/40 [==============================] - 2s 39ms/step - loss: 0.0809 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 101/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0861\n",
            "Epoch 101: val_loss did not improve from 0.07668\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0861 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 102/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0903\n",
            "Epoch 102: val_loss did not improve from 0.07668\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0903 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 103/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0816\n",
            "Epoch 103: val_loss did not improve from 0.07668\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0816 - val_loss: 0.0769 - lr: 1.0000e-05\n",
            "Epoch 104/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0849\n",
            "Epoch 104: val_loss did not improve from 0.07668\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0849 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 105/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0767\n",
            "Epoch 105: val_loss did not improve from 0.07668\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0767 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 106/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0792\n",
            "Epoch 106: val_loss did not improve from 0.07668\n",
            "40/40 [==============================] - 2s 39ms/step - loss: 0.0792 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 107/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0872\n",
            "Epoch 107: val_loss did not improve from 0.07668\n",
            "40/40 [==============================] - 2s 53ms/step - loss: 0.0872 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 108/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0823\n",
            "Epoch 108: val_loss did not improve from 0.07668\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0823 - val_loss: 0.0769 - lr: 1.0000e-05\n",
            "Epoch 109/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0810\n",
            "Epoch 109: val_loss did not improve from 0.07668\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0810 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 110/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0851\n",
            "Epoch 110: val_loss improved from 0.07668 to 0.07665, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 56ms/step - loss: 0.0851 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 111/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0939\n",
            "Epoch 111: val_loss did not improve from 0.07665\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0939 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 112/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0777\n",
            "Epoch 112: val_loss did not improve from 0.07665\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0777 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 113/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0836\n",
            "Epoch 113: val_loss did not improve from 0.07665\n",
            "40/40 [==============================] - 2s 39ms/step - loss: 0.0836 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 114/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0762\n",
            "Epoch 114: val_loss did not improve from 0.07665\n",
            "40/40 [==============================] - 2s 38ms/step - loss: 0.0762 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 115/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0812\n",
            "Epoch 115: val_loss did not improve from 0.07665\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0812 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 116/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0833\n",
            "Epoch 116: val_loss did not improve from 0.07665\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0833 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 117/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0768\n",
            "Epoch 117: val_loss improved from 0.07665 to 0.07662, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 55ms/step - loss: 0.0768 - val_loss: 0.0766 - lr: 1.0000e-05\n",
            "Epoch 118/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0823\n",
            "Epoch 118: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0823 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 119/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0815\n",
            "Epoch 119: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0815 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 120/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0850\n",
            "Epoch 120: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 53ms/step - loss: 0.0850 - val_loss: 0.0769 - lr: 1.0000e-05\n",
            "Epoch 121/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0800\n",
            "Epoch 121: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0800 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 122/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0826\n",
            "Epoch 122: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0826 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 123/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0950\n",
            "Epoch 123: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0950 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 124/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0829\n",
            "Epoch 124: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0829 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 125/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0853\n",
            "Epoch 125: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0853 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 126/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0810\n",
            "Epoch 126: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 53ms/step - loss: 0.0810 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 127/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0825\n",
            "Epoch 127: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 39ms/step - loss: 0.0825 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 128/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0808\n",
            "Epoch 128: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0808 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 129/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0856\n",
            "Epoch 129: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0856 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 130/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0833\n",
            "Epoch 130: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0833 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 131/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0828\n",
            "Epoch 131: val_loss improved from 0.07662 to 0.07662, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 41ms/step - loss: 0.0828 - val_loss: 0.0766 - lr: 1.0000e-05\n",
            "Epoch 132/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0832\n",
            "Epoch 132: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0832 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 133/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0855\n",
            "Epoch 133: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 38ms/step - loss: 0.0855 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 134/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0796\n",
            "Epoch 134: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 39ms/step - loss: 0.0796 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 135/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0841\n",
            "Epoch 135: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 53ms/step - loss: 0.0841 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 136/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0884\n",
            "Epoch 136: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0884 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 137/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0790\n",
            "Epoch 137: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0790 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 138/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0822\n",
            "Epoch 138: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0822 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 139/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0860\n",
            "Epoch 139: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0860 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 140/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0864\n",
            "Epoch 140: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0864 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 141/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0858\n",
            "Epoch 141: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0858 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 142/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0844\n",
            "Epoch 142: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 53ms/step - loss: 0.0844 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 143/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0840\n",
            "Epoch 143: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0840 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 144/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0831\n",
            "Epoch 144: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0831 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 145/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0856\n",
            "Epoch 145: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0856 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 146/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0851\n",
            "Epoch 146: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0851 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 147/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0797\n",
            "Epoch 147: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0797 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 148/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0801\n",
            "Epoch 148: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0801 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 149/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0767\n",
            "Epoch 149: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 53ms/step - loss: 0.0767 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 150/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0825\n",
            "Epoch 150: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0825 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 151/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0863\n",
            "Epoch 151: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0863 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 152/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0857\n",
            "Epoch 152: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0857 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 153/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0799\n",
            "Epoch 153: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0799 - val_loss: 0.0766 - lr: 1.0000e-05\n",
            "Epoch 154/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0844\n",
            "Epoch 154: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0844 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 155/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0838\n",
            "Epoch 155: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0838 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 156/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0852\n",
            "Epoch 156: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 53ms/step - loss: 0.0852 - val_loss: 0.0769 - lr: 1.0000e-05\n",
            "Epoch 157/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0877\n",
            "Epoch 157: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0877 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 158/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0826\n",
            "Epoch 158: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0826 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 159/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0841\n",
            "Epoch 159: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0841 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 160/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0791\n",
            "Epoch 160: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0791 - val_loss: 0.0766 - lr: 1.0000e-05\n",
            "Epoch 161/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0892\n",
            "Epoch 161: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0892 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 162/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0725\n",
            "Epoch 162: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0725 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 163/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0811\n",
            "Epoch 163: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 53ms/step - loss: 0.0811 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 164/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0842\n",
            "Epoch 164: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 53ms/step - loss: 0.0842 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 165/175\n",
            "38/40 [===========================>..] - ETA: 0s - loss: 0.0867\n",
            "Epoch 165: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 38ms/step - loss: 0.0848 - val_loss: 0.0768 - lr: 1.0000e-05\n",
            "Epoch 166/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0884\n",
            "Epoch 166: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0884 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 167/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0862\n",
            "Epoch 167: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0862 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 168/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0821\n",
            "Epoch 168: val_loss did not improve from 0.07662\n",
            "40/40 [==============================] - 1s 38ms/step - loss: 0.0821 - val_loss: 0.0766 - lr: 1.0000e-05\n",
            "Epoch 169/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0849\n",
            "Epoch 169: val_loss improved from 0.07662 to 0.07659, saving model to siamese_network.h5\n",
            "40/40 [==============================] - 2s 56ms/step - loss: 0.0849 - val_loss: 0.0766 - lr: 1.0000e-05\n",
            "Epoch 170/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0862\n",
            "Epoch 170: val_loss did not improve from 0.07659\n",
            "40/40 [==============================] - 2s 38ms/step - loss: 0.0862 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 171/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0779\n",
            "Epoch 171: val_loss did not improve from 0.07659\n",
            "40/40 [==============================] - 2s 53ms/step - loss: 0.0779 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 172/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0806\n",
            "Epoch 172: val_loss did not improve from 0.07659\n",
            "40/40 [==============================] - 2s 38ms/step - loss: 0.0806 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 173/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0872\n",
            "Epoch 173: val_loss did not improve from 0.07659\n",
            "40/40 [==============================] - 2s 38ms/step - loss: 0.0872 - val_loss: 0.0767 - lr: 1.0000e-05\n",
            "Epoch 174/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0791\n",
            "Epoch 174: val_loss did not improve from 0.07659\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0791 - val_loss: 0.0766 - lr: 1.0000e-05\n",
            "Epoch 175/175\n",
            "40/40 [==============================] - ETA: 0s - loss: 0.0859\n",
            "Epoch 175: val_loss did not improve from 0.07659\n",
            "40/40 [==============================] - 2s 52ms/step - loss: 0.0859 - val_loss: 0.0767 - lr: 1.0000e-05\n"
          ]
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "\"\"\" train the model \"\"\"\n",
        "optimizer = Adam(learning_rate=0.0001)\n",
        "#siamese.compile(loss=utils.loss(1), optimizer=optimizer, metrics=[\"accuracy\"]) # Manhattan [loss: 0.0997 - accuracy: 0.9539] | Euclidean [loss: 0.2267 - accuracy: 0.9487]\n",
        "# siamese.compile(loss=tf.keras.losses.BinaryCrossentropy(from_logits=True), optimizer=optimizer, metrics=[\"accuracy\"]) # Manhattan [loss: 0.4305 - accuracy: 0.9430] | Euclidean [loss: 0.6142 - accuracy: 0.9464]\n",
        "\n",
        "# Compile the model\n",
        "siamese.compile(\n",
        "    optimizer=optimizer,\n",
        "    loss=utils.identity_loss)\n",
        "\n",
        "siamese.summary()\n",
        "history = siamese.fit([x_train_1, x_train_2, x_train_3],\n",
        "    labels_train[:,0].reshape(-1, 1),\n",
        "    validation_data=([x_val_1, x_val_2, x_val_3], labels_val[:,0].reshape(-1, 1)),\n",
        "    batch_size=1,\n",
        "    epochs=175,   # 175 for contrastive 100 for cross ent\n",
        "    callbacks = [checkpointer, reduce_lr]#, early_stopping]\n",
        ")\n",
        "\n",
        "# Plot the Triplet loss\n",
        "utils.plt_metric(history=history.history, metric=\"loss\", title=\"Triplet Loss\")\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "1Ya59cg6gFVJ",
        "outputId": "5a1983c8-1418-4355-9907-9bd4a1db91e7"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "60/60 [==============================] - 24s 210ms/step - loss: 0.0680\n",
            "test loss 0.06801939755678177\n",
            "60/60 [==============================] - 13s 216ms/step\n",
            "\n",
            "Evaluate on test data\n",
            "Accuracy: 0.975\n"
          ]
        }
      ],
      "source": [
        "\"\"\" Test the model \"\"\"\n",
        "siamese = tf.keras.models.load_model(SIAMESE_MODEL_FNAME, custom_objects={'identity_loss': identity_loss})\n",
        "\n",
        "results = siamese.evaluate([x_test_1, x_test_2, x_test_3], labels_test)\n",
        "print(\"test loss\", results)\n",
        "\n",
        "Y_pred = siamese.predict([x_test_1, x_test_2, x_test_3]).squeeze()\n",
        "\n",
        "\n",
        "print(\"\\nEvaluate on test data\")\n",
        "print(\"Accuracy:\", len(np.where(Y_pred <= MARGIN)[0]) / len(Y_pred))\n",
        "\n",
        "tf.keras.backend.clear_session()"
      ]
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "gpuType": "T4",
      "provenance": []
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}