Supervised_Full_Dataset.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "scratchpad",
      "provenance": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/sayakpaul/SimCLR-in-TensorFlow-2/blob/master/Supervised_Full_Dataset.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OJy-Q5dmd85v",
        "colab_type": "text"
      },
      "source": [
        "## Initial setup"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "lIYdn1woOS1n",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "d0a70376-ffe1-49a5-8792-e05d92b6accd"
      },
      "source": [
        "import tensorflow as tf\n",
        "print(tf.__version__)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "2.2.0-rc4\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uckwlGeh4UxH",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 306
        },
        "outputId": "f4423235-3555-4e90-caae-90dbc9d9fb25"
      },
      "source": [
        "!nvidia-smi"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Sat May  9 03:24:29 2020       \n",
            "+-----------------------------------------------------------------------------+\n",
            "| NVIDIA-SMI 440.82       Driver Version: 418.67       CUDA Version: 10.1     |\n",
            "|-------------------------------+----------------------+----------------------+\n",
            "| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |\n",
            "| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |\n",
            "|===============================+======================+======================|\n",
            "|   0  Tesla P100-PCIE...  Off  | 00000000:00:04.0 Off |                    0 |\n",
            "| N/A   35C    P0    26W / 250W |      0MiB / 16280MiB |      0%      Default |\n",
            "+-------------------------------+----------------------+----------------------+\n",
            "                                                                               \n",
            "+-----------------------------------------------------------------------------+\n",
            "| Processes:                                                       GPU Memory |\n",
            "|  GPU       PID   Type   Process name                             Usage      |\n",
            "|=============================================================================|\n",
            "|  No running processes found                                                 |\n",
            "+-----------------------------------------------------------------------------+\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yidv44b_dnj1",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Install wandb for experiment tracking\n",
        "!pip install --upgrade https://github.com/wandb/client/archive/feature/code-save.zip"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JIomoEBgdsse",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import wandb\n",
        "wandb.login()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dtBEZR3b4Zo7",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Other imports\n",
        "from sklearn.preprocessing import LabelEncoder\n",
        "from tensorflow.keras.layers import *\n",
        "from tensorflow.keras.models import *\n",
        "from wandb.keras import WandbCallback\n",
        "import matplotlib.pyplot as plt\n",
        "from imutils import paths\n",
        "from tqdm import tqdm\n",
        "import tensorflow as tf\n",
        "import numpy as np\n",
        "import cv2\n",
        "\n",
        "# Random seed fixation\n",
        "tf.random.set_seed(666)\n",
        "np.random.seed(666)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Umc5UxD6eCln",
        "colab_type": "text"
      },
      "source": [
        "## Dataset gathering and preparation"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9mG6EaYR4XTN",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Gather dataset\n",
        "!git clone https://github.com/thunderInfy/imagenet-5-categories"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XPs164bL4loQ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "3997fddd-c44a-441a-e7d7-a751b707618d"
      },
      "source": [
        "# Train and test image paths\n",
        "train_images = list(paths.list_images(\"imagenet-5-categories/train\"))\n",
        "test_images = list(paths.list_images(\"imagenet-5-categories/test\"))\n",
        "print(len(train_images), len(test_images))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "1250 250\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eOLU3Y224oXl",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def prepare_images(image_paths):\n",
        "    images = []\n",
        "    labels = []\n",
        "\n",
        "    for image in tqdm(image_paths):\n",
        "        image_pixels = plt.imread(image)\n",
        "        image_pixels = cv2.resize(image_pixels, (224, 224))\n",
        "        image_pixels = image_pixels/255.\n",
        "\n",
        "        label = image.split(\"/\")[2].split(\"_\")[0]\n",
        "\n",
        "        images.append(image_pixels)\n",
        "        labels.append(label)\n",
        "\n",
        "    images = np.array(images)\n",
        "    labels = np.array(labels)\n",
        "\n",
        "    print(images.shape, labels.shape)\n",
        "\n",
        "    return images, labels"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cWMYKwiW40f6",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 85
        },
        "outputId": "169bc817-3eeb-4a58-f00c-44b58fd1830c"
      },
      "source": [
        "X_train, y_train = prepare_images(train_images)\n",
        "X_test, y_test = prepare_images(test_images)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "100%|██████████| 1250/1250 [00:05<00:00, 215.45it/s]\n",
            " 10%|█         | 25/250 [00:00<00:00, 241.84it/s]"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "(1250, 224, 224, 3) (1250,)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "100%|██████████| 250/250 [00:01<00:00, 213.37it/s]\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "(250, 224, 224, 3) (250,)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0hbCPZRP45aJ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "le = LabelEncoder()\n",
        "y_train_enc = le.fit_transform(y_train)\n",
        "y_test_enc = le.transform(y_test)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KPT9-NWT6Q5p",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Set up TensorFlow dataset for performance optimization\n",
        "\n",
        "AUTO = tf.data.experimental.AUTOTUNE\n",
        "BATCH_SIZE = 64\n",
        "\n",
        "train_ds = (\n",
        "    tf.data.Dataset.from_tensor_slices((X_train, y_train_enc))\n",
        "    .shuffle(1024)\n",
        "    .batch(BATCH_SIZE)\n",
        "    .prefetch(AUTO)\n",
        ")\n",
        "\n",
        "test_ds = (\n",
        "    tf.data.Dataset.from_tensor_slices((X_test, y_test_enc))\n",
        "    .shuffle(1024)\n",
        "    .batch(BATCH_SIZE)\n",
        "    .prefetch(AUTO)\n",
        ")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "p-bth_5SeUcC",
        "colab_type": "text"
      },
      "source": [
        "## Utilities"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "r3mCu1yI49fT",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Architecture utils\n",
        "def get_resnet():\n",
        "    base_model = tf.keras.applications.ResNet50(include_top=False, weights=None, input_shape=(224, 224, 3))\n",
        "    base_model.trainable = True\n",
        "    \n",
        "    inputs = Input((224, 224, 3))\n",
        "    features = base_model(inputs, training=False)\n",
        "    pooled_features = GlobalAveragePooling2D()(features)\n",
        "\n",
        "    pooled_features = Dense(256)(pooled_features)\n",
        "    pooled_features = Activation(\"relu\")(pooled_features)\n",
        "    outputs = Dense(5, activation=\"softmax\")(pooled_features)\n",
        "\n",
        "    resnet_simclr = Model(inputs, outputs)\n",
        "\n",
        "    return resnet_simclr"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NBrNUWWp6F5Q",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def plot_training(H):\n",
        "\twith plt.xkcd():\n",
        "\t\tplt.plot(H.history[\"loss\"], label=\"train_loss\")\n",
        "\t\tplt.plot(H.history[\"val_loss\"], label=\"val_loss\")\n",
        "\t\tplt.plot(H.history[\"accuracy\"], label=\"train_acc\")\n",
        "\t\tplt.plot(H.history[\"val_accuracy\"], label=\"val_acc\")\n",
        "\t\tplt.title(\"Training Loss and Accuracy\")\n",
        "\t\tplt.xlabel(\"Epoch #\")\n",
        "\t\tplt.ylabel(\"Loss/Accuracy\")\n",
        "\t\tplt.legend(loc=\"lower left\")\n",
        "\t\tplt.show()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "p1WyaOnoeY2R",
        "colab_type": "text"
      },
      "source": [
        "## Training"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Df8sYNoq6LoS",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Early Stopping to prevent overfitting\n",
        "es = tf.keras.callbacks.EarlyStopping(monitor=\"val_loss\", patience=2, verbose=2, restore_best_weights=True)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fG_OQZ1j55iG",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 357
        },
        "outputId": "be3dc84e-a907-49a0-e7c6-8ea82b3ec046"
      },
      "source": [
        "get_resnet().summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"model\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "input_2 (InputLayer)         [(None, 224, 224, 3)]     0         \n",
            "_________________________________________________________________\n",
            "resnet50 (Model)             (None, 7, 7, 2048)        23587712  \n",
            "_________________________________________________________________\n",
            "global_average_pooling2d (Gl (None, 2048)              0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 256)               524544    \n",
            "_________________________________________________________________\n",
            "activation (Activation)      (None, 256)               0         \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 5)                 1285      \n",
            "=================================================================\n",
            "Total params: 24,113,541\n",
            "Trainable params: 24,060,421\n",
            "Non-trainable params: 53,120\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vHVEJ6Li6Eo8",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "c99e8e1a-211f-4b2a-9d25-25d0d053f62d"
      },
      "source": [
        "# Train the supervised model with full data\n",
        "wandb.init(project=\"simclr\", id=\"supervised-training\")\n",
        "\n",
        "model = get_resnet()\n",
        "model.compile(loss=\"sparse_categorical_crossentropy\", metrics=[\"accuracy\"],\n",
        "              optimizer=tf.keras.optimizers.Adam(1e-3))\n",
        "history = model.fit(train_ds,\n",
        "                    validation_data=test_ds,\n",
        "                    epochs=50,\n",
        "                    callbacks=[es, WandbCallback()])\n",
        "plot_training(history)"
      ],
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/100\n",
            "20/20 [==============================] - 10s 507ms/step - loss: 1.6105 - accuracy: 0.1968 - val_loss: 1.6027 - val_accuracy: 0.2000\n",
            "Epoch 2/100\n",
            "20/20 [==============================] - 9s 457ms/step - loss: 1.6051 - accuracy: 0.2328 - val_loss: 1.5994 - val_accuracy: 0.2040\n",
            "Epoch 3/100\n",
            "20/20 [==============================] - 9s 464ms/step - loss: 1.6029 - accuracy: 0.2032 - val_loss: 1.5950 - val_accuracy: 0.2120\n",
            "Epoch 4/100\n",
            "20/20 [==============================] - 9s 467ms/step - loss: 1.5983 - accuracy: 0.2304 - val_loss: 1.5920 - val_accuracy: 0.2160\n",
            "Epoch 5/100\n",
            "20/20 [==============================] - 9s 468ms/step - loss: 1.5941 - accuracy: 0.2520 - val_loss: 1.5878 - val_accuracy: 0.2480\n",
            "Epoch 6/100\n",
            "20/20 [==============================] - 10s 483ms/step - loss: 1.5921 - accuracy: 0.2304 - val_loss: 1.5839 - val_accuracy: 0.2080\n",
            "Epoch 7/100\n",
            "20/20 [==============================] - 9s 464ms/step - loss: 1.5912 - accuracy: 0.2592 - val_loss: 1.5806 - val_accuracy: 0.2720\n",
            "Epoch 8/100\n",
            "20/20 [==============================] - 9s 465ms/step - loss: 1.5841 - accuracy: 0.2336 - val_loss: 1.5773 - val_accuracy: 0.2400\n",
            "Epoch 9/100\n",
            "20/20 [==============================] - 9s 467ms/step - loss: 1.5813 - accuracy: 0.2792 - val_loss: 1.5737 - val_accuracy: 0.2880\n",
            "Epoch 10/100\n",
            "20/20 [==============================] - 9s 462ms/step - loss: 1.5774 - accuracy: 0.2936 - val_loss: 1.5701 - val_accuracy: 0.3520\n",
            "Epoch 11/100\n",
            "20/20 [==============================] - 9s 469ms/step - loss: 1.5742 - accuracy: 0.2752 - val_loss: 1.5670 - val_accuracy: 0.2760\n",
            "Epoch 12/100\n",
            "20/20 [==============================] - 10s 477ms/step - loss: 1.5719 - accuracy: 0.3432 - val_loss: 1.5635 - val_accuracy: 0.4040\n",
            "Epoch 13/100\n",
            "20/20 [==============================] - 9s 470ms/step - loss: 1.5703 - accuracy: 0.3248 - val_loss: 1.5603 - val_accuracy: 0.3280\n",
            "Epoch 14/100\n",
            "20/20 [==============================] - 9s 464ms/step - loss: 1.5659 - accuracy: 0.3224 - val_loss: 1.5564 - val_accuracy: 0.3520\n",
            "Epoch 15/100\n",
            "20/20 [==============================] - 9s 461ms/step - loss: 1.5622 - accuracy: 0.3400 - val_loss: 1.5533 - val_accuracy: 0.3560\n",
            "Epoch 16/100\n",
            "20/20 [==============================] - 9s 461ms/step - loss: 1.5593 - accuracy: 0.3088 - val_loss: 1.5497 - val_accuracy: 0.3680\n",
            "Epoch 17/100\n",
            "20/20 [==============================] - 9s 464ms/step - loss: 1.5561 - accuracy: 0.4128 - val_loss: 1.5488 - val_accuracy: 0.4640\n",
            "Epoch 18/100\n",
            "20/20 [==============================] - 9s 460ms/step - loss: 1.5527 - accuracy: 0.3856 - val_loss: 1.5433 - val_accuracy: 0.3600\n",
            "Epoch 19/100\n",
            "20/20 [==============================] - 9s 475ms/step - loss: 1.5510 - accuracy: 0.3824 - val_loss: 1.5406 - val_accuracy: 0.3640\n",
            "Epoch 20/100\n",
            "20/20 [==============================] - 9s 466ms/step - loss: 1.5480 - accuracy: 0.3368 - val_loss: 1.5375 - val_accuracy: 0.4040\n",
            "Epoch 21/100\n",
            "20/20 [==============================] - 10s 475ms/step - loss: 1.5451 - accuracy: 0.4224 - val_loss: 1.5348 - val_accuracy: 0.4080\n",
            "Epoch 22/100\n",
            "20/20 [==============================] - 9s 463ms/step - loss: 1.5441 - accuracy: 0.3632 - val_loss: 1.5313 - val_accuracy: 0.3960\n",
            "Epoch 23/100\n",
            "20/20 [==============================] - 9s 461ms/step - loss: 1.5396 - accuracy: 0.4176 - val_loss: 1.5292 - val_accuracy: 0.4480\n",
            "Epoch 24/100\n",
            "20/20 [==============================] - 9s 462ms/step - loss: 1.5367 - accuracy: 0.4088 - val_loss: 1.5252 - val_accuracy: 0.4360\n",
            "Epoch 25/100\n",
            "20/20 [==============================] - 9s 462ms/step - loss: 1.5338 - accuracy: 0.4240 - val_loss: 1.5224 - val_accuracy: 0.4120\n",
            "Epoch 26/100\n",
            "20/20 [==============================] - 9s 467ms/step - loss: 1.5312 - accuracy: 0.4080 - val_loss: 1.5197 - val_accuracy: 0.4760\n",
            "Epoch 27/100\n",
            "20/20 [==============================] - 9s 461ms/step - loss: 1.5283 - accuracy: 0.4152 - val_loss: 1.5169 - val_accuracy: 0.4240\n",
            "Epoch 28/100\n",
            "20/20 [==============================] - 9s 474ms/step - loss: 1.5250 - accuracy: 0.4488 - val_loss: 1.5143 - val_accuracy: 0.4720\n",
            "Epoch 29/100\n",
            "20/20 [==============================] - 9s 475ms/step - loss: 1.5228 - accuracy: 0.4360 - val_loss: 1.5115 - val_accuracy: 0.4640\n",
            "Epoch 30/100\n",
            "20/20 [==============================] - 9s 473ms/step - loss: 1.5204 - accuracy: 0.4520 - val_loss: 1.5082 - val_accuracy: 0.4440\n",
            "Epoch 31/100\n",
            "20/20 [==============================] - 9s 460ms/step - loss: 1.5184 - accuracy: 0.4152 - val_loss: 1.5055 - val_accuracy: 0.4720\n",
            "Epoch 32/100\n",
            "20/20 [==============================] - 9s 471ms/step - loss: 1.5175 - accuracy: 0.4648 - val_loss: 1.5033 - val_accuracy: 0.4680\n",
            "Epoch 33/100\n",
            "20/20 [==============================] - 9s 462ms/step - loss: 1.5125 - accuracy: 0.4264 - val_loss: 1.4996 - val_accuracy: 0.4600\n",
            "Epoch 34/100\n",
            "20/20 [==============================] - 9s 474ms/step - loss: 1.5101 - accuracy: 0.4208 - val_loss: 1.4973 - val_accuracy: 0.4560\n",
            "Epoch 35/100\n",
            "20/20 [==============================] - 9s 468ms/step - loss: 1.5079 - accuracy: 0.4744 - val_loss: 1.4945 - val_accuracy: 0.4760\n",
            "Epoch 36/100\n",
            "20/20 [==============================] - 10s 476ms/step - loss: 1.5066 - accuracy: 0.4640 - val_loss: 1.4917 - val_accuracy: 0.4560\n",
            "Epoch 37/100\n",
            "20/20 [==============================] - 9s 459ms/step - loss: 1.5040 - accuracy: 0.4312 - val_loss: 1.4887 - val_accuracy: 0.4520\n",
            "Epoch 38/100\n",
            "20/20 [==============================] - 9s 467ms/step - loss: 1.5016 - accuracy: 0.4536 - val_loss: 1.4872 - val_accuracy: 0.4840\n",
            "Epoch 39/100\n",
            "20/20 [==============================] - 9s 462ms/step - loss: 1.4988 - accuracy: 0.4648 - val_loss: 1.4842 - val_accuracy: 0.4600\n",
            "Epoch 40/100\n",
            "20/20 [==============================] - 10s 476ms/step - loss: 1.4967 - accuracy: 0.4664 - val_loss: 1.4823 - val_accuracy: 0.4840\n",
            "Epoch 41/100\n",
            "20/20 [==============================] - 9s 472ms/step - loss: 1.4941 - accuracy: 0.4504 - val_loss: 1.4787 - val_accuracy: 0.4560\n",
            "Epoch 42/100\n",
            "20/20 [==============================] - 9s 471ms/step - loss: 1.4919 - accuracy: 0.4440 - val_loss: 1.4765 - val_accuracy: 0.4560\n",
            "Epoch 43/100\n",
            "20/20 [==============================] - 9s 462ms/step - loss: 1.4904 - accuracy: 0.4656 - val_loss: 1.4751 - val_accuracy: 0.4840\n",
            "Epoch 44/100\n",
            "20/20 [==============================] - 9s 461ms/step - loss: 1.4891 - accuracy: 0.4408 - val_loss: 1.4718 - val_accuracy: 0.4520\n",
            "Epoch 45/100\n",
            "20/20 [==============================] - 9s 463ms/step - loss: 1.4846 - accuracy: 0.4720 - val_loss: 1.4691 - val_accuracy: 0.4560\n",
            "Epoch 46/100\n",
            "20/20 [==============================] - 9s 474ms/step - loss: 1.4839 - accuracy: 0.4488 - val_loss: 1.4675 - val_accuracy: 0.4720\n",
            "Epoch 47/100\n",
            "20/20 [==============================] - 9s 468ms/step - loss: 1.4808 - accuracy: 0.4616 - val_loss: 1.4648 - val_accuracy: 0.4760\n",
            "Epoch 48/100\n",
            "20/20 [==============================] - 10s 475ms/step - loss: 1.4783 - accuracy: 0.4696 - val_loss: 1.4619 - val_accuracy: 0.4600\n",
            "Epoch 49/100\n",
            "20/20 [==============================] - 9s 456ms/step - loss: 1.4766 - accuracy: 0.4688 - val_loss: 1.4597 - val_accuracy: 0.4800\n",
            "Epoch 50/100\n",
            "20/20 [==============================] - 9s 461ms/step - loss: 1.4738 - accuracy: 0.4744 - val_loss: 1.4569 - val_accuracy: 0.4560\n",
            "Epoch 51/100\n",
            "20/20 [==============================] - 9s 466ms/step - loss: 1.4722 - accuracy: 0.4752 - val_loss: 1.4553 - val_accuracy: 0.4840\n",
            "Epoch 52/100\n",
            "20/20 [==============================] - 9s 462ms/step - loss: 1.4706 - accuracy: 0.4824 - val_loss: 1.4534 - val_accuracy: 0.4560\n",
            "Epoch 53/100\n",
            "20/20 [==============================] - 10s 477ms/step - loss: 1.4684 - accuracy: 0.4640 - val_loss: 1.4504 - val_accuracy: 0.4800\n",
            "Epoch 54/100\n",
            "20/20 [==============================] - 9s 474ms/step - loss: 1.4665 - accuracy: 0.4752 - val_loss: 1.4486 - val_accuracy: 0.4800\n",
            "Epoch 55/100\n",
            "20/20 [==============================] - 9s 469ms/step - loss: 1.4652 - accuracy: 0.4736 - val_loss: 1.4477 - val_accuracy: 0.4960\n",
            "Epoch 56/100\n",
            "20/20 [==============================] - 9s 462ms/step - loss: 1.4615 - accuracy: 0.4704 - val_loss: 1.4442 - val_accuracy: 0.4720\n",
            "Epoch 57/100\n",
            "20/20 [==============================] - 9s 461ms/step - loss: 1.4608 - accuracy: 0.4800 - val_loss: 1.4424 - val_accuracy: 0.4920\n",
            "Epoch 58/100\n",
            "20/20 [==============================] - 9s 470ms/step - loss: 1.4578 - accuracy: 0.4920 - val_loss: 1.4400 - val_accuracy: 0.4880\n",
            "Epoch 59/100\n",
            "20/20 [==============================] - 9s 467ms/step - loss: 1.4559 - accuracy: 0.5008 - val_loss: 1.4376 - val_accuracy: 0.4840\n",
            "Epoch 60/100\n",
            "20/20 [==============================] - 10s 476ms/step - loss: 1.4542 - accuracy: 0.4792 - val_loss: 1.4352 - val_accuracy: 0.4880\n",
            "Epoch 61/100\n",
            "20/20 [==============================] - 9s 463ms/step - loss: 1.4518 - accuracy: 0.4944 - val_loss: 1.4341 - val_accuracy: 0.5000\n",
            "Epoch 62/100\n",
            "20/20 [==============================] - 9s 467ms/step - loss: 1.4504 - accuracy: 0.4840 - val_loss: 1.4310 - val_accuracy: 0.4880\n",
            "Epoch 63/100\n",
            "20/20 [==============================] - 9s 462ms/step - loss: 1.4493 - accuracy: 0.4776 - val_loss: 1.4294 - val_accuracy: 0.4720\n",
            "Epoch 64/100\n",
            "20/20 [==============================] - 9s 473ms/step - loss: 1.4468 - accuracy: 0.4912 - val_loss: 1.4274 - val_accuracy: 0.4920\n",
            "Epoch 65/100\n",
            "20/20 [==============================] - 9s 473ms/step - loss: 1.4449 - accuracy: 0.4824 - val_loss: 1.4251 - val_accuracy: 0.4720\n",
            "Epoch 66/100\n",
            "20/20 [==============================] - 9s 469ms/step - loss: 1.4430 - accuracy: 0.5016 - val_loss: 1.4242 - val_accuracy: 0.5080\n",
            "Epoch 67/100\n",
            "20/20 [==============================] - 9s 459ms/step - loss: 1.4408 - accuracy: 0.5056 - val_loss: 1.4211 - val_accuracy: 0.4760\n",
            "Epoch 68/100\n",
            "20/20 [==============================] - 9s 462ms/step - loss: 1.4406 - accuracy: 0.4584 - val_loss: 1.4189 - val_accuracy: 0.4800\n",
            "Epoch 69/100\n",
            "20/20 [==============================] - 9s 463ms/step - loss: 1.4379 - accuracy: 0.4904 - val_loss: 1.4170 - val_accuracy: 0.4920\n",
            "Epoch 70/100\n",
            "20/20 [==============================] - 9s 466ms/step - loss: 1.4371 - accuracy: 0.5056 - val_loss: 1.4155 - val_accuracy: 0.4960\n",
            "Epoch 71/100\n",
            "20/20 [==============================] - 9s 472ms/step - loss: 1.4338 - accuracy: 0.4896 - val_loss: 1.4140 - val_accuracy: 0.4920\n",
            "Epoch 72/100\n",
            "20/20 [==============================] - 10s 481ms/step - loss: 1.4327 - accuracy: 0.4912 - val_loss: 1.4115 - val_accuracy: 0.4800\n",
            "Epoch 73/100\n",
            "20/20 [==============================] - 9s 462ms/step - loss: 1.4307 - accuracy: 0.5072 - val_loss: 1.4096 - val_accuracy: 0.4840\n",
            "Epoch 74/100\n",
            "20/20 [==============================] - 9s 463ms/step - loss: 1.4303 - accuracy: 0.4840 - val_loss: 1.4076 - val_accuracy: 0.4840\n",
            "Epoch 75/100\n",
            "20/20 [==============================] - 9s 462ms/step - loss: 1.4270 - accuracy: 0.5080 - val_loss: 1.4067 - val_accuracy: 0.4920\n",
            "Epoch 76/100\n",
            "20/20 [==============================] - 9s 465ms/step - loss: 1.4246 - accuracy: 0.5032 - val_loss: 1.4040 - val_accuracy: 0.4840\n",
            "Epoch 77/100\n",
            "20/20 [==============================] - 9s 473ms/step - loss: 1.4235 - accuracy: 0.4984 - val_loss: 1.4022 - val_accuracy: 0.5000\n",
            "Epoch 78/100\n",
            "20/20 [==============================] - 10s 478ms/step - loss: 1.4222 - accuracy: 0.5000 - val_loss: 1.4002 - val_accuracy: 0.5080\n",
            "Epoch 79/100\n",
            "20/20 [==============================] - 9s 462ms/step - loss: 1.4198 - accuracy: 0.5040 - val_loss: 1.3993 - val_accuracy: 0.5120\n",
            "Epoch 80/100\n",
            "20/20 [==============================] - 9s 463ms/step - loss: 1.4185 - accuracy: 0.4984 - val_loss: 1.3965 - val_accuracy: 0.4880\n",
            "Epoch 81/100\n",
            "20/20 [==============================] - 9s 465ms/step - loss: 1.4188 - accuracy: 0.4736 - val_loss: 1.3962 - val_accuracy: 0.4960\n",
            "Epoch 82/100\n",
            "20/20 [==============================] - 9s 464ms/step - loss: 1.4150 - accuracy: 0.5120 - val_loss: 1.3950 - val_accuracy: 0.5000\n",
            "Epoch 83/100\n",
            "20/20 [==============================] - 9s 466ms/step - loss: 1.4133 - accuracy: 0.4920 - val_loss: 1.3923 - val_accuracy: 0.5080\n",
            "Epoch 84/100\n",
            "20/20 [==============================] - 10s 477ms/step - loss: 1.4121 - accuracy: 0.5000 - val_loss: 1.3899 - val_accuracy: 0.5040\n",
            "Epoch 85/100\n",
            "20/20 [==============================] - 9s 462ms/step - loss: 1.4109 - accuracy: 0.4992 - val_loss: 1.3890 - val_accuracy: 0.4960\n",
            "Epoch 86/100\n",
            "20/20 [==============================] - 9s 459ms/step - loss: 1.4094 - accuracy: 0.5104 - val_loss: 1.3867 - val_accuracy: 0.5000\n",
            "Epoch 87/100\n",
            "20/20 [==============================] - 9s 464ms/step - loss: 1.4085 - accuracy: 0.4824 - val_loss: 1.3846 - val_accuracy: 0.4920\n",
            "Epoch 88/100\n",
            "20/20 [==============================] - 9s 462ms/step - loss: 1.4063 - accuracy: 0.5112 - val_loss: 1.3846 - val_accuracy: 0.5160\n",
            "Epoch 89/100\n",
            "20/20 [==============================] - 9s 472ms/step - loss: 1.4063 - accuracy: 0.5112 - val_loss: 1.3818 - val_accuracy: 0.5080\n",
            "Epoch 90/100\n",
            "20/20 [==============================] - 10s 477ms/step - loss: 1.4035 - accuracy: 0.4768 - val_loss: 1.3810 - val_accuracy: 0.5280\n",
            "Epoch 91/100\n",
            "20/20 [==============================] - 9s 473ms/step - loss: 1.4021 - accuracy: 0.5040 - val_loss: 1.3790 - val_accuracy: 0.5080\n",
            "Epoch 92/100\n",
            "20/20 [==============================] - 9s 460ms/step - loss: 1.4011 - accuracy: 0.5048 - val_loss: 1.3768 - val_accuracy: 0.4880\n",
            "Epoch 93/100\n",
            "20/20 [==============================] - 9s 462ms/step - loss: 1.3989 - accuracy: 0.5024 - val_loss: 1.3756 - val_accuracy: 0.5040\n",
            "Epoch 94/100\n",
            "20/20 [==============================] - 9s 468ms/step - loss: 1.3979 - accuracy: 0.4952 - val_loss: 1.3737 - val_accuracy: 0.5080\n",
            "Epoch 95/100\n",
            "20/20 [==============================] - 9s 467ms/step - loss: 1.3965 - accuracy: 0.5032 - val_loss: 1.3725 - val_accuracy: 0.5040\n",
            "Epoch 96/100\n",
            "20/20 [==============================] - 9s 472ms/step - loss: 1.3953 - accuracy: 0.5024 - val_loss: 1.3711 - val_accuracy: 0.5120\n",
            "Epoch 97/100\n",
            "20/20 [==============================] - 10s 481ms/step - loss: 1.3930 - accuracy: 0.5048 - val_loss: 1.3695 - val_accuracy: 0.5120\n",
            "Epoch 98/100\n",
            "20/20 [==============================] - 9s 463ms/step - loss: 1.3940 - accuracy: 0.4968 - val_loss: 1.3679 - val_accuracy: 0.4920\n",
            "Epoch 99/100\n",
            "20/20 [==============================] - 9s 461ms/step - loss: 1.3923 - accuracy: 0.4952 - val_loss: 1.3674 - val_accuracy: 0.5080\n",
            "Epoch 100/100\n",
            "20/20 [==============================] - 9s 464ms/step - loss: 1.3902 - accuracy: 0.5088 - val_loss: 1.3649 - val_accuracy: 0.5080\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    }
  ]
}