plain_train.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "train.ipynb",
      "version": "0.3.2",
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "metadata": {
        "id": "ikDcT5B__ozM",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "from sklearn import *\n",
        "import tensorflow as tf\n",
        "\n",
        "import os\n",
        "import time"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "_-PHRNbT_ozQ",
        "colab_type": "text"
      },
      "cell_type": "markdown",
      "source": [
        "Load Data"
      ]
    },
    {
      "metadata": {
        "id": "unEQ1ajAAWpX",
        "colab_type": "code",
        "outputId": "8f603ab0-c80c-463e-84b0-29e4ba8faabb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 122
        }
      },
      "cell_type": "code",
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/gdrive')"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Go to this URL in a browser: https://accounts.google.com/o/oauth2/auth?client_id=947318989803-6bn6qk8qdgf4n4g3pfee6491hc0brc4i.apps.googleusercontent.com&redirect_uri=urn%3Aietf%3Awg%3Aoauth%3A2.0%3Aoob&scope=email%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdocs.test%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdrive%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdrive.photos.readonly%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fpeopleapi.readonly&response_type=code\n",
            "\n",
            "Enter your authorization code:\n",
            "··········\n",
            "Mounted at /content/gdrive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "CuVjTolpBXTo",
        "colab_type": "code",
        "outputId": "1eb1d0b5-b434-4092-c5e8-5073f0dde40f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "cell_type": "code",
      "source": [
        "path = '/content/gdrive/My Drive/Plastic_Recyclability_Dataset'\n",
        "\n",
        "batches = []\n",
        "for batch in os.listdir(path):\n",
        "    if batch[0:5] == \"batch\":\n",
        "        batches.append(np.fromfile(path + \"/\" + batch, dtype=np.uint8, count=-1))\n",
        "inputs = np.concatenate(batches)\n",
        "\n",
        "labels = np.fromfile(path + \"/labels.bin\", dtype=np.uint8, count=-1)\n",
        "\n",
        "inputs = inputs.reshape((-1,128,128,3))\n",
        "print(inputs.shape)\n",
        "\n",
        "print(labels.shape)"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(8424, 128, 128, 3)\n",
            "(8424,)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "JQFOBZpr_ozX",
        "colab_type": "code",
        "outputId": "256de0e3-3a72-4274-95a2-1759e396d13f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 50
        }
      },
      "cell_type": "code",
      "source": [
        "print(\"number of recyclable items:\", len(labels[labels[0]==0]))\n",
        "print(\"number of maybe recyclable items:\", len(labels[labels[0]==1]))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "number of recyclable items: 0\n",
            "number of maybe recyclable items: 1\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "9rJqE0Ap_ozc",
        "colab_type": "code",
        "outputId": "8375072d-3034-44d5-af33-0043e9e13c80",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        }
      },
      "cell_type": "code",
      "source": [
        "test_size = 0.15\n",
        "val_size = 0.1\n",
        "\n",
        "X_train_val, X_test, y_train_val, y_test = model_selection.train_test_split(inputs, labels,\n",
        "                                                                              test_size = test_size, random_state = 0)\n",
        "X_train, X_val, y_train, y_val = model_selection.train_test_split(X_train_val, y_train_val,\n",
        "                                                                        test_size = val_size / (1 - test_size), \n",
        "                                                                        random_state = 0)\n",
        "\n",
        "print(\"X train:\", X_train.shape)\n",
        "print(\"Y train:\", y_train.shape)\n",
        "print(\"X validation:\", X_val.shape)\n",
        "print(\"Y validation:\", y_val.shape)\n",
        "print(\"X test:\", X_test.shape)\n",
        "print(\"Y test:\", y_test.shape)"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "X train: (6317, 128, 128, 3)\n",
            "Y train: (6317,)\n",
            "X validation: (843, 128, 128, 3)\n",
            "Y validation: (843,)\n",
            "X test: (1264, 128, 128, 3)\n",
            "Y test: (1264,)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "ZdbsdUvz_ozg",
        "colab_type": "code",
        "outputId": "ed5623fa-e91f-497b-f7e3-16f9d00e6d26",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 366
        }
      },
      "cell_type": "code",
      "source": [
        "num = 1530\n",
        "plt.imshow(inputs[num])\n",
        "if labels[num] == 0:\n",
        "    print(\"recyclable\")\n",
        "elif labels[num] == 1:\n",
        "    print(\"non-recyclable\")\n",
        "else:\n",
        "    print(\"label not 0 or 1\")"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "recyclable\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAVEAAAFNCAYAAAC5YlyiAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "L3t5IfJLiJ1z",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "model = tf.keras.Sequential()\n",
        "\n",
        "for i in range(5):\n",
        "  filters = 2 ** (i+5)\n",
        "  model.add(tf.keras.layers.Conv2D(filters=filters, kernel_size=(3,3), use_bias=False,\n",
        "                                             kernel_regularizer=tf.keras.regularizers.l2(1e-4)))\n",
        "  model.add(tf.keras.layers.BatchNormalization())\n",
        "  model.add(tf.keras.layers.SpatialDropout2D(rate=0.2))\n",
        "  model.add(tf.keras.layers.MaxPooling2D(pool_size=2, strides=[2,2]))\n",
        "  model.add(tf.keras.layers.ReLU())\n",
        "\n",
        "model.add(tf.keras.layers.GlobalAveragePooling2D())\n",
        "\n",
        "\n",
        "model.add(tf.keras.layers.Dense(units=2, activation=tf.nn.softmax, kernel_regularizer=tf.keras.regularizers.l2(1e-4),\n",
        "                         bias_regularizer=tf.keras.regularizers.l2(1e-4)))\n",
        "\n",
        "model.compile(optimizer=tf.train.AdamOptimizer(learning_rate=1e-4),\n",
        "                loss='sparse_categorical_crossentropy',\n",
        "                metrics=[\"accuracy\"])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "metadata": {
        "id": "a0xD2nWFzlyg",
        "colab_type": "code",
        "outputId": "fd8a81f6-4031-486d-ff44-9a1e516d8827",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 3451
        }
      },
      "cell_type": "code",
      "source": [
        "start = time.time()\n",
        "history = model.fit(X_train, y_train, epochs=100, batch_size=128, validation_data=(X_val, y_val))\n",
        "end = time.time()\n",
        "print(end-start, \"seconds elapsed\")"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 6317 samples, validate on 843 samples\n",
            "Epoch 1/100\n",
            "6317/6317 [==============================] - 13s 2ms/sample - loss: 0.8883 - acc: 0.5566 - val_loss: 0.8789 - val_acc: 0.5184\n",
            "Epoch 2/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.8150 - acc: 0.5879 - val_loss: 0.7378 - val_acc: 0.5730\n",
            "Epoch 3/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.7876 - acc: 0.6039 - val_loss: 0.6706 - val_acc: 0.6738\n",
            "Epoch 4/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.7794 - acc: 0.6114 - val_loss: 0.6656 - val_acc: 0.6845\n",
            "Epoch 5/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.7618 - acc: 0.6152 - val_loss: 0.6576 - val_acc: 0.6892\n",
            "Epoch 6/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.7530 - acc: 0.6228 - val_loss: 0.6434 - val_acc: 0.6880\n",
            "Epoch 7/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.7421 - acc: 0.6350 - val_loss: 0.6538 - val_acc: 0.6750\n",
            "Epoch 8/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.7346 - acc: 0.6359 - val_loss: 0.6483 - val_acc: 0.6904\n",
            "Epoch 9/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.7166 - acc: 0.6498 - val_loss: 0.6434 - val_acc: 0.6821\n",
            "Epoch 10/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.7179 - acc: 0.6522 - val_loss: 0.6260 - val_acc: 0.7070\n",
            "Epoch 11/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.7108 - acc: 0.6579 - val_loss: 0.6423 - val_acc: 0.6928\n",
            "Epoch 12/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.6991 - acc: 0.6658 - val_loss: 0.6658 - val_acc: 0.6750\n",
            "Epoch 13/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.6919 - acc: 0.6723 - val_loss: 0.6537 - val_acc: 0.6797\n",
            "Epoch 14/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.6889 - acc: 0.6674 - val_loss: 0.6412 - val_acc: 0.6951\n",
            "Epoch 15/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.6864 - acc: 0.6672 - val_loss: 0.6190 - val_acc: 0.6999\n",
            "Epoch 16/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.6779 - acc: 0.6845 - val_loss: 0.6236 - val_acc: 0.7011\n",
            "Epoch 17/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.6713 - acc: 0.6899 - val_loss: 0.6269 - val_acc: 0.6999\n",
            "Epoch 18/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.6696 - acc: 0.6836 - val_loss: 0.6596 - val_acc: 0.6738\n",
            "Epoch 19/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.6551 - acc: 0.6867 - val_loss: 0.6125 - val_acc: 0.7272\n",
            "Epoch 20/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.6645 - acc: 0.6931 - val_loss: 0.6068 - val_acc: 0.7236\n",
            "Epoch 21/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.6594 - acc: 0.6919 - val_loss: 0.6113 - val_acc: 0.7236\n",
            "Epoch 22/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.6495 - acc: 0.6957 - val_loss: 0.6181 - val_acc: 0.7165\n",
            "Epoch 23/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.6377 - acc: 0.7048 - val_loss: 0.6331 - val_acc: 0.7117\n",
            "Epoch 24/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.6371 - acc: 0.7136 - val_loss: 0.6014 - val_acc: 0.7260\n",
            "Epoch 25/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.6357 - acc: 0.7041 - val_loss: 0.6047 - val_acc: 0.7094\n",
            "Epoch 26/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.6334 - acc: 0.7101 - val_loss: 0.5997 - val_acc: 0.7153\n",
            "Epoch 27/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.6267 - acc: 0.7163 - val_loss: 0.6207 - val_acc: 0.7106\n",
            "Epoch 28/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.6308 - acc: 0.7114 - val_loss: 0.6054 - val_acc: 0.7319\n",
            "Epoch 29/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.6131 - acc: 0.7177 - val_loss: 0.5976 - val_acc: 0.7248\n",
            "Epoch 30/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.6090 - acc: 0.7201 - val_loss: 0.6405 - val_acc: 0.7094\n",
            "Epoch 31/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.6069 - acc: 0.7204 - val_loss: 0.7376 - val_acc: 0.6524\n",
            "Epoch 32/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.6087 - acc: 0.7217 - val_loss: 0.5977 - val_acc: 0.7165\n",
            "Epoch 33/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5994 - acc: 0.7290 - val_loss: 0.5858 - val_acc: 0.7307\n",
            "Epoch 34/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5910 - acc: 0.7334 - val_loss: 0.6025 - val_acc: 0.7212\n",
            "Epoch 35/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5911 - acc: 0.7352 - val_loss: 0.5942 - val_acc: 0.7070\n",
            "Epoch 36/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5976 - acc: 0.7317 - val_loss: 0.5882 - val_acc: 0.7284\n",
            "Epoch 37/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5878 - acc: 0.7401 - val_loss: 0.6001 - val_acc: 0.7248\n",
            "Epoch 38/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5812 - acc: 0.7445 - val_loss: 0.6152 - val_acc: 0.7284\n",
            "Epoch 39/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5798 - acc: 0.7434 - val_loss: 0.5751 - val_acc: 0.7461\n",
            "Epoch 40/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5798 - acc: 0.7464 - val_loss: 0.6404 - val_acc: 0.7153\n",
            "Epoch 41/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5603 - acc: 0.7499 - val_loss: 0.6208 - val_acc: 0.7295\n",
            "Epoch 42/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5702 - acc: 0.7478 - val_loss: 0.5846 - val_acc: 0.7461\n",
            "Epoch 43/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5676 - acc: 0.7496 - val_loss: 0.5890 - val_acc: 0.7307\n",
            "Epoch 44/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5615 - acc: 0.7595 - val_loss: 0.6085 - val_acc: 0.7212\n",
            "Epoch 45/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5545 - acc: 0.7580 - val_loss: 0.5578 - val_acc: 0.7533\n",
            "Epoch 46/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5469 - acc: 0.7678 - val_loss: 0.5602 - val_acc: 0.7556\n",
            "Epoch 47/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5413 - acc: 0.7637 - val_loss: 0.5859 - val_acc: 0.7355\n",
            "Epoch 48/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5388 - acc: 0.7730 - val_loss: 0.5628 - val_acc: 0.7568\n",
            "Epoch 49/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5348 - acc: 0.7722 - val_loss: 0.5551 - val_acc: 0.7544\n",
            "Epoch 50/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5369 - acc: 0.7690 - val_loss: 0.5874 - val_acc: 0.7485\n",
            "Epoch 51/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5334 - acc: 0.7730 - val_loss: 0.5896 - val_acc: 0.7438\n",
            "Epoch 52/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5313 - acc: 0.7779 - val_loss: 0.5535 - val_acc: 0.7580\n",
            "Epoch 53/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5241 - acc: 0.7784 - val_loss: 0.5553 - val_acc: 0.7521\n",
            "Epoch 54/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5203 - acc: 0.7793 - val_loss: 0.5678 - val_acc: 0.7497\n",
            "Epoch 55/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5020 - acc: 0.7920 - val_loss: 0.5455 - val_acc: 0.7675\n",
            "Epoch 56/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5217 - acc: 0.7815 - val_loss: 0.5995 - val_acc: 0.7402\n",
            "Epoch 57/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.4999 - acc: 0.7948 - val_loss: 0.5568 - val_acc: 0.7580\n",
            "Epoch 58/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5090 - acc: 0.7882 - val_loss: 0.5563 - val_acc: 0.7675\n",
            "Epoch 59/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.5035 - acc: 0.7917 - val_loss: 0.5871 - val_acc: 0.7450\n",
            "Epoch 60/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.4946 - acc: 0.7983 - val_loss: 0.5516 - val_acc: 0.7651\n",
            "Epoch 61/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.4912 - acc: 0.7953 - val_loss: 0.5502 - val_acc: 0.7675\n",
            "Epoch 62/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.4794 - acc: 0.8040 - val_loss: 0.5381 - val_acc: 0.7770\n",
            "Epoch 63/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.4877 - acc: 0.7955 - val_loss: 0.5427 - val_acc: 0.7734\n",
            "Epoch 64/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.4682 - acc: 0.8140 - val_loss: 0.5714 - val_acc: 0.7556\n",
            "Epoch 65/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.4794 - acc: 0.7990 - val_loss: 0.5542 - val_acc: 0.7687\n",
            "Epoch 66/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.4701 - acc: 0.8096 - val_loss: 0.5464 - val_acc: 0.7782\n",
            "Epoch 67/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.4729 - acc: 0.8067 - val_loss: 0.5385 - val_acc: 0.7722\n",
            "Epoch 68/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.4612 - acc: 0.8146 - val_loss: 0.5510 - val_acc: 0.7734\n",
            "Epoch 69/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.4657 - acc: 0.8164 - val_loss: 0.5531 - val_acc: 0.7746\n",
            "Epoch 70/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.4594 - acc: 0.8126 - val_loss: 0.6360 - val_acc: 0.7165\n",
            "Epoch 71/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.4616 - acc: 0.8145 - val_loss: 0.5637 - val_acc: 0.7711\n",
            "Epoch 72/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.4419 - acc: 0.8254 - val_loss: 0.5498 - val_acc: 0.7734\n",
            "Epoch 73/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.4355 - acc: 0.8314 - val_loss: 0.6110 - val_acc: 0.7521\n",
            "Epoch 74/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.4384 - acc: 0.8229 - val_loss: 0.5592 - val_acc: 0.7746\n",
            "Epoch 75/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.4370 - acc: 0.8252 - val_loss: 0.5952 - val_acc: 0.7556\n",
            "Epoch 76/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.4286 - acc: 0.8339 - val_loss: 0.5320 - val_acc: 0.7675\n",
            "Epoch 77/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.4254 - acc: 0.8335 - val_loss: 0.5366 - val_acc: 0.7805\n",
            "Epoch 78/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.4175 - acc: 0.8376 - val_loss: 0.5422 - val_acc: 0.7794\n",
            "Epoch 79/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.4166 - acc: 0.8328 - val_loss: 0.5243 - val_acc: 0.7888\n",
            "Epoch 80/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.4186 - acc: 0.8401 - val_loss: 0.5241 - val_acc: 0.7722\n",
            "Epoch 81/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.4170 - acc: 0.8400 - val_loss: 0.5758 - val_acc: 0.7758\n",
            "Epoch 82/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.4078 - acc: 0.8431 - val_loss: 0.5577 - val_acc: 0.7782\n",
            "Epoch 83/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.4075 - acc: 0.8482 - val_loss: 0.5472 - val_acc: 0.7794\n",
            "Epoch 84/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.3999 - acc: 0.8495 - val_loss: 0.5705 - val_acc: 0.7675\n",
            "Epoch 85/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.3926 - acc: 0.8544 - val_loss: 0.5310 - val_acc: 0.7912\n",
            "Epoch 86/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.3955 - acc: 0.8496 - val_loss: 0.5566 - val_acc: 0.7734\n",
            "Epoch 87/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.3950 - acc: 0.8485 - val_loss: 0.5418 - val_acc: 0.7805\n",
            "Epoch 88/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.3922 - acc: 0.8520 - val_loss: 0.5632 - val_acc: 0.7770\n",
            "Epoch 89/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.3794 - acc: 0.8585 - val_loss: 0.6278 - val_acc: 0.7473\n",
            "Epoch 90/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.3793 - acc: 0.8604 - val_loss: 0.5461 - val_acc: 0.7746\n",
            "Epoch 91/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.3663 - acc: 0.8647 - val_loss: 0.5594 - val_acc: 0.7699\n",
            "Epoch 92/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.3646 - acc: 0.8653 - val_loss: 0.5657 - val_acc: 0.7746\n",
            "Epoch 93/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.3644 - acc: 0.8666 - val_loss: 0.5518 - val_acc: 0.7770\n",
            "Epoch 94/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.3505 - acc: 0.8770 - val_loss: 0.6418 - val_acc: 0.7521\n",
            "Epoch 95/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.3521 - acc: 0.8783 - val_loss: 0.5827 - val_acc: 0.7865\n",
            "Epoch 96/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.3578 - acc: 0.8715 - val_loss: 0.6407 - val_acc: 0.7533\n",
            "Epoch 97/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.3389 - acc: 0.8827 - val_loss: 0.5678 - val_acc: 0.7746\n",
            "Epoch 98/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.3523 - acc: 0.8779 - val_loss: 0.7062 - val_acc: 0.7343\n",
            "Epoch 99/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.3428 - acc: 0.8776 - val_loss: 0.5608 - val_acc: 0.7746\n",
            "Epoch 100/100\n",
            "6317/6317 [==============================] - 12s 2ms/sample - loss: 0.3400 - acc: 0.8792 - val_loss: 0.6372 - val_acc: 0.7651\n",
            "1170.189365386963 seconds elapsed\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "2TuWCQai_ozw",
        "colab_type": "code",
        "outputId": "8b1619b0-da5a-4cb2-c849-ff1a36a0b7b5",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 4250
        }
      },
      "cell_type": "code",
      "source": [
        "model.summary()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "__________________________________________________________________________________________________\n",
            "Layer (type)                    Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            "input_14 (InputLayer)           (None, 128, 128, 3)  0                                            \n",
            "__________________________________________________________________________________________________\n",
            "conv0 (Conv2D)                  (None, 63, 63, 16)   432         input_14[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm0 (BatchNormalization (None, 63, 63, 16)   64          conv0[0][0]                      \n",
            "__________________________________________________________________________________________________\n",
            "relu0 (ReLU)                    (None, 63, 63, 16)   0           batch_norm0[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "conv1a (Conv2D)                 (None, 63, 63, 16)   2304        relu0[0][0]                      \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm1a (BatchNormalizatio (None, 63, 63, 16)   64          conv1a[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "relu1a (ReLU)                   (None, 63, 63, 16)   0           batch_norm1a[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "conv1b (Conv2D)                 (None, 63, 63, 16)   2304        relu1a[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm1b (BatchNormalizatio (None, 63, 63, 16)   64          conv1b[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "add1 (Add)                      (None, 63, 63, 16)   0           relu0[0][0]                      \n",
            "                                                                 batch_norm1b[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "relu1b (ReLU)                   (None, 63, 63, 16)   0           add1[0][0]                       \n",
            "__________________________________________________________________________________________________\n",
            "conv2a (Conv2D)                 (None, 63, 63, 16)   2304        relu1b[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm2a (BatchNormalizatio (None, 63, 63, 16)   64          conv2a[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "relu2a (ReLU)                   (None, 63, 63, 16)   0           batch_norm2a[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "conv2b (Conv2D)                 (None, 63, 63, 16)   2304        relu2a[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm2b (BatchNormalizatio (None, 63, 63, 16)   64          conv2b[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "add2 (Add)                      (None, 63, 63, 16)   0           relu1b[0][0]                     \n",
            "                                                                 batch_norm2b[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "relu2b (ReLU)                   (None, 63, 63, 16)   0           add2[0][0]                       \n",
            "__________________________________________________________________________________________________\n",
            "conv3a (Conv2D)                 (None, 63, 63, 16)   2304        relu2b[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm3a (BatchNormalizatio (None, 63, 63, 16)   64          conv3a[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "relu3a (ReLU)                   (None, 63, 63, 16)   0           batch_norm3a[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "conv3b (Conv2D)                 (None, 63, 63, 16)   2304        relu3a[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm3b (BatchNormalizatio (None, 63, 63, 16)   64          conv3b[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "add3 (Add)                      (None, 63, 63, 16)   0           relu2b[0][0]                     \n",
            "                                                                 batch_norm3b[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "relu3b (ReLU)                   (None, 63, 63, 16)   0           add3[0][0]                       \n",
            "__________________________________________________________________________________________________\n",
            "conv4a (Conv2D)                 (None, 32, 32, 32)   4608        relu3b[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm4a (BatchNormalizatio (None, 32, 32, 32)   128         conv4a[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "relu4a (ReLU)                   (None, 32, 32, 32)   0           batch_norm4a[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "conv4b (Conv2D)                 (None, 32, 32, 32)   9216        relu4a[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "conv_1x14 (Conv2D)              (None, 32, 32, 32)   4640        relu3b[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm4b (BatchNormalizatio (None, 32, 32, 32)   128         conv4b[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "add4 (Add)                      (None, 32, 32, 32)   0           conv_1x14[0][0]                  \n",
            "                                                                 batch_norm4b[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "re_lu_52 (ReLU)                 (None, 32, 32, 32)   0           add4[0][0]                       \n",
            "__________________________________________________________________________________________________\n",
            "conv5a (Conv2D)                 (None, 32, 32, 32)   9216        re_lu_52[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm5a (BatchNormalizatio (None, 32, 32, 32)   128         conv5a[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "relu5a (ReLU)                   (None, 32, 32, 32)   0           batch_norm5a[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "conv5b (Conv2D)                 (None, 32, 32, 32)   9216        relu5a[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm5b (BatchNormalizatio (None, 32, 32, 32)   128         conv5b[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "add5 (Add)                      (None, 32, 32, 32)   0           re_lu_52[0][0]                   \n",
            "                                                                 batch_norm5b[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "relu5b (ReLU)                   (None, 32, 32, 32)   0           add5[0][0]                       \n",
            "__________________________________________________________________________________________________\n",
            "conv6a (Conv2D)                 (None, 32, 32, 32)   9216        relu5b[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm6a (BatchNormalizatio (None, 32, 32, 32)   128         conv6a[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "relu6a (ReLU)                   (None, 32, 32, 32)   0           batch_norm6a[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "conv6b (Conv2D)                 (None, 32, 32, 32)   9216        relu6a[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm6b (BatchNormalizatio (None, 32, 32, 32)   128         conv6b[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "add6 (Add)                      (None, 32, 32, 32)   0           relu5b[0][0]                     \n",
            "                                                                 batch_norm6b[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "relu6b (ReLU)                   (None, 32, 32, 32)   0           add6[0][0]                       \n",
            "__________________________________________________________________________________________________\n",
            "conv7a (Conv2D)                 (None, 16, 16, 64)   18432       relu6b[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm7a (BatchNormalizatio (None, 16, 16, 64)   256         conv7a[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "relu7a (ReLU)                   (None, 16, 16, 64)   0           batch_norm7a[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "conv7b (Conv2D)                 (None, 16, 16, 64)   36864       relu7a[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "conv_1x17 (Conv2D)              (None, 16, 16, 64)   18496       relu6b[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm7b (BatchNormalizatio (None, 16, 16, 64)   256         conv7b[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "add7 (Add)                      (None, 16, 16, 64)   0           conv_1x17[0][0]                  \n",
            "                                                                 batch_norm7b[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "re_lu_53 (ReLU)                 (None, 16, 16, 64)   0           add7[0][0]                       \n",
            "__________________________________________________________________________________________________\n",
            "conv8a (Conv2D)                 (None, 16, 16, 64)   36864       re_lu_53[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm8a (BatchNormalizatio (None, 16, 16, 64)   256         conv8a[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "relu8a (ReLU)                   (None, 16, 16, 64)   0           batch_norm8a[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "conv8b (Conv2D)                 (None, 16, 16, 64)   36864       relu8a[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm8b (BatchNormalizatio (None, 16, 16, 64)   256         conv8b[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "add8 (Add)                      (None, 16, 16, 64)   0           re_lu_53[0][0]                   \n",
            "                                                                 batch_norm8b[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "relu8b (ReLU)                   (None, 16, 16, 64)   0           add8[0][0]                       \n",
            "__________________________________________________________________________________________________\n",
            "conv9a (Conv2D)                 (None, 16, 16, 64)   36864       relu8b[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm9a (BatchNormalizatio (None, 16, 16, 64)   256         conv9a[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "relu9a (ReLU)                   (None, 16, 16, 64)   0           batch_norm9a[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "conv9b (Conv2D)                 (None, 16, 16, 64)   36864       relu9a[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm9b (BatchNormalizatio (None, 16, 16, 64)   256         conv9b[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "add9 (Add)                      (None, 16, 16, 64)   0           relu8b[0][0]                     \n",
            "                                                                 batch_norm9b[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "relu9b (ReLU)                   (None, 16, 16, 64)   0           add9[0][0]                       \n",
            "__________________________________________________________________________________________________\n",
            "conv10a (Conv2D)                (None, 8, 8, 128)    32768       relu9b[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm10a (BatchNormalizati (None, 8, 8, 128)    512         conv10a[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "relu10a (ReLU)                  (None, 8, 8, 128)    0           batch_norm10a[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "conv10b (Conv2D)                (None, 8, 8, 128)    65536       relu10a[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "conv_1x110 (Conv2D)             (None, 8, 8, 128)    32896       relu9b[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm10b (BatchNormalizati (None, 8, 8, 128)    512         conv10b[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "add10 (Add)                     (None, 8, 8, 128)    0           conv_1x110[0][0]                 \n",
            "                                                                 batch_norm10b[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "re_lu_54 (ReLU)                 (None, 8, 8, 128)    0           add10[0][0]                      \n",
            "__________________________________________________________________________________________________\n",
            "conv11a (Conv2D)                (None, 8, 8, 128)    65536       re_lu_54[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm11a (BatchNormalizati (None, 8, 8, 128)    512         conv11a[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "relu11a (ReLU)                  (None, 8, 8, 128)    0           batch_norm11a[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "conv11b (Conv2D)                (None, 8, 8, 128)    65536       relu11a[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm11b (BatchNormalizati (None, 8, 8, 128)    512         conv11b[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "add11 (Add)                     (None, 8, 8, 128)    0           re_lu_54[0][0]                   \n",
            "                                                                 batch_norm11b[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "relu11b (ReLU)                  (None, 8, 8, 128)    0           add11[0][0]                      \n",
            "__________________________________________________________________________________________________\n",
            "conv12a (Conv2D)                (None, 8, 8, 128)    65536       relu11b[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm12a (BatchNormalizati (None, 8, 8, 128)    512         conv12a[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "relu12a (ReLU)                  (None, 8, 8, 128)    0           batch_norm12a[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "conv12b (Conv2D)                (None, 8, 8, 128)    65536       relu12a[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm12b (BatchNormalizati (None, 8, 8, 128)    512         conv12b[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "add12 (Add)                     (None, 8, 8, 128)    0           relu11b[0][0]                    \n",
            "                                                                 batch_norm12b[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "relu12b (ReLU)                  (None, 8, 8, 128)    0           add12[0][0]                      \n",
            "__________________________________________________________________________________________________\n",
            "conv13a (Conv2D)                (None, 4, 4, 256)    131072      relu12b[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm13a (BatchNormalizati (None, 4, 4, 256)    1024        conv13a[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "relu13a (ReLU)                  (None, 4, 4, 256)    0           batch_norm13a[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "conv13b (Conv2D)                (None, 4, 4, 256)    262144      relu13a[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "conv_1x113 (Conv2D)             (None, 4, 4, 256)    131328      relu12b[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm13b (BatchNormalizati (None, 4, 4, 256)    1024        conv13b[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "add13 (Add)                     (None, 4, 4, 256)    0           conv_1x113[0][0]                 \n",
            "                                                                 batch_norm13b[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "re_lu_55 (ReLU)                 (None, 4, 4, 256)    0           add13[0][0]                      \n",
            "__________________________________________________________________________________________________\n",
            "conv14a (Conv2D)                (None, 4, 4, 256)    262144      re_lu_55[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm14a (BatchNormalizati (None, 4, 4, 256)    1024        conv14a[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "relu14a (ReLU)                  (None, 4, 4, 256)    0           batch_norm14a[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "conv14b (Conv2D)                (None, 4, 4, 256)    262144      relu14a[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm14b (BatchNormalizati (None, 4, 4, 256)    1024        conv14b[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "add14 (Add)                     (None, 4, 4, 256)    0           re_lu_55[0][0]                   \n",
            "                                                                 batch_norm14b[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "relu14b (ReLU)                  (None, 4, 4, 256)    0           add14[0][0]                      \n",
            "__________________________________________________________________________________________________\n",
            "conv15a (Conv2D)                (None, 4, 4, 256)    262144      relu14b[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm15a (BatchNormalizati (None, 4, 4, 256)    1024        conv15a[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "relu15a (ReLU)                  (None, 4, 4, 256)    0           batch_norm15a[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "conv15b (Conv2D)                (None, 4, 4, 256)    262144      relu15a[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "batch_norm15b (BatchNormalizati (None, 4, 4, 256)    1024        conv15b[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "add15 (Add)                     (None, 4, 4, 256)    0           relu14b[0][0]                    \n",
            "                                                                 batch_norm15b[0][0]              \n",
            "__________________________________________________________________________________________________\n",
            "relu15b (ReLU)                  (None, 4, 4, 256)    0           add15[0][0]                      \n",
            "__________________________________________________________________________________________________\n",
            "avg_pool (GlobalAveragePooling2 (None, 256)          0           relu15b[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "dense_26 (Dense)                (None, 1000)         257000      avg_pool[0][0]                   \n",
            "==================================================================================================\n",
            "Total params: 2,526,264\n",
            "Trainable params: 2,520,280\n",
            "Non-trainable params: 5,984\n",
            "__________________________________________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "uO0V6rKypZhp",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 88
        },
        "outputId": "57a8c281-2b05-4e4a-acea-b5885da9e393"
      },
      "cell_type": "code",
      "source": [
        "#save weights\n",
        "model.save_weights(path + \"/checkpoints/sequential_cp.cpkt\", save_format='tf')"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/network.py:1436: update_checkpoint_state (from tensorflow.python.training.checkpoint_management) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Use tf.train.CheckpointManager to manage checkpoints rather than manually editing the Checkpoint proto.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "qpmlGJU9_oz0",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        },
        "outputId": "64bcc4f9-fa19-476a-ba86-4879186c0535"
      },
      "cell_type": "code",
      "source": [
        "p_train = model.predict(X_train)\n",
        "p_train = np.argmax(p_train, axis=1)\n",
        "\n",
        "p_val = model.predict(X_val)\n",
        "p_val = np.argmax(p_val, axis=1)\n",
        "\n",
        "print(\"train acc:\", metrics.accuracy_score(y_train, p_train))\n",
        "print(\"validation acc:\", metrics.accuracy_score(y_val, p_val))"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "train acc: 0.9012189330378344\n",
            "validation acc: 0.7651245551601423\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "SY43t8uz_oz7",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "8261c512-26f4-4f55-bafe-bfc65d47d119"
      },
      "cell_type": "code",
      "source": [
        "p_test = model.predict(X_test)\n",
        "p_test = np.argmax(p_test, axis=1)\n",
        "\n",
        "print(\"test acc:\", metrics.accuracy_score(y_test, p_test))"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "test acc: 0.7792721518987342\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "thyxUml__o0C",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 122
        },
        "outputId": "93d8d6c1-ab71-4a72-9bef-a8dd81b9f6ad"
      },
      "cell_type": "code",
      "source": [
        "print(history.history.keys())\n",
        "print(history.history['loss'])\n",
        "print(history.history['acc'])\n",
        "print(history.history['val_loss'])\n",
        "print(history.history['val_acc'])"
      ],
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "dict_keys(['loss', 'acc', 'val_loss', 'val_acc'])\n",
            "[0.8882945913179249, 0.8149706688469981, 0.7876414341678984, 0.7793508432457601, 0.7617544475208643, 0.7529747020279246, 0.7420981841618, 0.7346135068773865, 0.7166170499474636, 0.717869973624264, 0.7107583977139333, 0.6991111789207133, 0.6918723310249713, 0.6888844630509516, 0.6863561708639935, 0.6778734511300244, 0.6713033462601107, 0.6696390612736446, 0.6550917834425645, 0.6644767587162638, 0.6593737341528435, 0.6494649164997854, 0.6377225019838569, 0.6370961022882762, 0.6357091718295981, 0.6334218333770302, 0.6266930409958794, 0.6307833886927813, 0.6130813243873364, 0.6089915565548393, 0.6068688454046003, 0.6086766548939515, 0.5993608745740565, 0.5909897378865953, 0.5910543359324244, 0.5975594535194018, 0.5878293480189216, 0.5812261475402569, 0.579810190874507, 0.5797955476635711, 0.560309094635014, 0.5702333919643808, 0.5676217681200647, 0.5615098107666914, 0.5544651476325765, 0.5468718149776377, 0.5412562591602705, 0.5387806634331535, 0.5347820114245875, 0.5368603438030453, 0.5333533312477264, 0.5312891687389338, 0.5241117405140345, 0.5202725262407527, 0.5020099333028203, 0.5217250644151067, 0.499942982198695, 0.5089755868711708, 0.5035220652427103, 0.49458784703799247, 0.4911928219245698, 0.4793633517489434, 0.4877446203626597, 0.4681744903670434, 0.4794042774661727, 0.47014530280713174, 0.4729108513439496, 0.461152251581612, 0.4656960254599064, 0.4593543448185117, 0.4616378696777784, 0.44193138132862264, 0.43551032693585734, 0.43842485079532073, 0.43698384104809923, 0.4286062667098852, 0.42536751696896247, 0.41750408530480615, 0.4165852912217355, 0.41860116533940966, 0.4169867394778931, 0.4077910361117118, 0.4075280193835662, 0.39994213779515736, 0.39257005344411705, 0.39547464334079696, 0.3950092378665851, 0.3921619848641146, 0.3793983576474017, 0.379291122314842, 0.3662677077962493, 0.3645611723540836, 0.36438849939185836, 0.3504550947557244, 0.3520720831168798, 0.3578112331328559, 0.33894835415097074, 0.3522878201471697, 0.34276197055047763, 0.34000812022958127]\n",
            "[0.5565933, 0.5879373, 0.60392594, 0.61136615, 0.6151654, 0.622764, 0.6349533, 0.6359031, 0.6498338, 0.6522083, 0.65790725, 0.6658224, 0.6723128, 0.6674054, 0.6672471, 0.6845021, 0.6898844, 0.6835523, 0.6867184, 0.6930505, 0.6919424, 0.69574165, 0.7047649, 0.7136299, 0.7041317, 0.7101472, 0.71632105, 0.7114136, 0.7177458, 0.7201203, 0.72043693, 0.72170335, 0.72898525, 0.73341775, 0.7351591, 0.7316764, 0.74006647, 0.74449897, 0.74339086, 0.7463986, 0.74988127, 0.74782336, 0.74956465, 0.75953776, 0.7579547, 0.7677695, 0.76365364, 0.7729935, 0.772202, 0.76903594, 0.7729935, 0.7779009, 0.7783758, 0.7793256, 0.79198986, 0.7815419, 0.7948393, 0.7881906, 0.79167324, 0.79832196, 0.79531425, 0.8040209, 0.79547256, 0.813994, 0.7989552, 0.8095615, 0.80671203, 0.8146272, 0.8163685, 0.81256926, 0.8144689, 0.8253918, 0.8314073, 0.8228589, 0.8252335, 0.83394015, 0.8334653, 0.83758116, 0.83283204, 0.840114, 0.8399557, 0.8431217, 0.84818745, 0.84945387, 0.85436124, 0.8496122, 0.84850407, 0.8519867, 0.8584771, 0.8603768, 0.86465096, 0.86528414, 0.86655056, 0.8769986, 0.878265, 0.871458, 0.88269746, 0.8779484, 0.8776318, 0.8792148]\n",
            "[0.8789010129930851, 0.7377829428669397, 0.6706231221609693, 0.6655769705206757, 0.6575871973982343, 0.6434113948936281, 0.6538441845106484, 0.6483254523888178, 0.6434026724644545, 0.6259603099466643, 0.6422989342277726, 0.6658127985956547, 0.6537166847311587, 0.64119629659121, 0.619016899262176, 0.6236057987840998, 0.6269459091613019, 0.6595732733023972, 0.6124657944710936, 0.6068116917870364, 0.6112748984764479, 0.6180519933236734, 0.6330869214814753, 0.6014296181968382, 0.6046580360453324, 0.5997108344081458, 0.6206824118415091, 0.605368838262162, 0.5976313945096017, 0.6405300254075128, 0.7375531241845689, 0.5976600354252337, 0.5857677533250404, 0.6024501240974643, 0.5942094835098104, 0.5881836198685715, 0.6001459290124344, 0.6152396195723917, 0.5750506482797444, 0.6403582617905641, 0.6207926943633055, 0.584624095531545, 0.5889721365727281, 0.6085210399412742, 0.5577774715310455, 0.5601779387231964, 0.5858616620337553, 0.5627561316229978, 0.5550871524663724, 0.5874108634668171, 0.5895596949796801, 0.5534572053351577, 0.5553047353686811, 0.5677512066078638, 0.5454617193585188, 0.599539962032653, 0.556779070468984, 0.5563442791908237, 0.5871324093740605, 0.5515620668870557, 0.550162297409483, 0.5381178752666125, 0.5427069324969114, 0.5713896662315016, 0.5541528707297247, 0.5463946606623603, 0.5384550252984568, 0.5510419902003957, 0.5531218378215781, 0.6359633040456444, 0.563744999335754, 0.5497740466116727, 0.610992770885092, 0.5592130627637798, 0.5951570451330318, 0.5320113943635639, 0.536558839491395, 0.5421782378129405, 0.5243333310136309, 0.5241485018888543, 0.5757733805465246, 0.5577433411902522, 0.5472065035700091, 0.5705239878854718, 0.5309921529228011, 0.5565708515058633, 0.5418009117051235, 0.563176963182248, 0.6277513190803438, 0.5461453386402356, 0.5593703913801789, 0.565669498975308, 0.5518141523233371, 0.6418431273275989, 0.5827393526142886, 0.6406766926407955, 0.5678491958661153, 0.7062375806404603, 0.5607859440687005, 0.6371913388954787]\n",
            "[0.5183867, 0.57295376, 0.6737841, 0.6844603, 0.6892052, 0.688019, 0.6749703, 0.6903915, 0.6820878, 0.7069988, 0.6927639, 0.6749703, 0.6797153, 0.6951364, 0.6998814, 0.7010676, 0.6998814, 0.6737841, 0.72716486, 0.72360617, 0.72360617, 0.7164887, 0.7117438, 0.7259787, 0.70937127, 0.71530247, 0.7105575, 0.7319099, 0.7247924, 0.70937127, 0.6524318, 0.7164887, 0.7307236, 0.72123367, 0.7069988, 0.7283511, 0.7247924, 0.7283511, 0.7461447, 0.71530247, 0.72953737, 0.7461447, 0.7307236, 0.72123367, 0.75326216, 0.75563467, 0.73546857, 0.75682086, 0.7544484, 0.7485172, 0.74377227, 0.7580071, 0.7520759, 0.74970347, 0.76749706, 0.7402135, 0.7580071, 0.76749706, 0.74495846, 0.76512456, 0.76749706, 0.77698696, 0.77342826, 0.75563467, 0.76868325, 0.7781732, 0.772242, 0.77342826, 0.77461445, 0.7164887, 0.77105576, 0.77342826, 0.7520759, 0.77461445, 0.75563467, 0.76749706, 0.78054565, 0.77935946, 0.78884935, 0.772242, 0.7758007, 0.7781732, 0.77935946, 0.76749706, 0.7912218, 0.77342826, 0.78054565, 0.77698696, 0.74733096, 0.77461445, 0.7698695, 0.77461445, 0.77698696, 0.7520759, 0.78647685, 0.75326216, 0.77461445, 0.7342823, 0.77461445, 0.76512456]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "RR3qhJbLvXHp",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 170
        },
        "outputId": "a536d807-815f-468c-e11d-b4e31b231226"
      },
      "cell_type": "code",
      "source": [
        "print(metrics.classification_report(y_test, p_test))"
      ],
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.71      0.91      0.80       599\n",
            "           1       0.89      0.66      0.76       665\n",
            "\n",
            "   micro avg       0.78      0.78      0.78      1264\n",
            "   macro avg       0.80      0.79      0.78      1264\n",
            "weighted avg       0.80      0.78      0.78      1264\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}