Day078_HW-訓練神經網路前的注意事項.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Work\n",
    "1. 請嘗試將 preproc_x 替換成以每筆資料的 min/max 進行標準化至 -1 ~ 1 間，再進行訓練\n",
    "2. 請嘗試將 mlp 疊更深 (e.g 5~10 層)，進行訓練後觀察 learning curve 的走勢\n",
    "3. (optional) 請改用 GPU 進行訓練 (如果你有 GPU 的話)，比較使用 CPU 與 GPU 的訓練速度"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "## \n",
    "\"\"\"\n",
    "Your code here (optional)\n",
    "確認硬體資源\n",
    "\"\"\"\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/seanyu/.conda/envs/tf18_keras/lib/python3.6/site-packages/h5py/__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.\n",
      "  from ._conv import register_converters as _register_converters\n",
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import keras\n",
    "\n",
    "# 請嘗試設定 GPU：os.environ\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "train, test = keras.datasets.cifar10.load_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "## 資料前處理\n",
    "\"\"\"\n",
    "Your code here\n",
    "\"\"\"\n",
    "def preproc_x():\n",
    "    pass\n",
    "\n",
    "def preproc_y():\n",
    "    pass"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train, y_train = train\n",
    "x_test, y_test = test\n",
    "\n",
    "# Preproc the inputs\n",
    "x_train = preproc_x(x_train)\n",
    "x_test = preproc_x(x_test)\n",
    "\n",
    "# Preprc the outputs\n",
    "y_train = preproc_y(y_train)\n",
    "y_test = preproc_y(y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\"Code Here\n",
    "建立你的神經網路\n",
    "\"\"\"\n",
    "def build_mlp():\n",
    "    pass\n",
    "\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "## 超參數設定\n",
    "LEARNING_RATE = 0.001\n",
    "EPOCHS = 100\n",
    "BATCH_SIZE = 256"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "optimizer = keras.optimizers.Adam(lr=LEARNING_RATE)\n",
    "model.compile(loss=\"categorical_crossentropy\", metrics=[\"accuracy\"], optimizer=optimizer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 50000 samples, validate on 10000 samples\n",
      "Epoch 1/100\n",
      "50000/50000 [==============================] - 6s 120us/step - loss: 1.6952 - acc: 0.3959 - val_loss: 1.5441 - val_acc: 0.4500\n",
      "Epoch 2/100\n",
      "50000/50000 [==============================] - 2s 48us/step - loss: 1.4660 - acc: 0.4804 - val_loss: 1.4604 - val_acc: 0.4831\n",
      "Epoch 3/100\n",
      "50000/50000 [==============================] - 2s 49us/step - loss: 1.3487 - acc: 0.5232 - val_loss: 1.4204 - val_acc: 0.5011\n",
      "Epoch 4/100\n",
      "50000/50000 [==============================] - 3s 51us/step - loss: 1.2546 - acc: 0.5591 - val_loss: 1.3956 - val_acc: 0.5112\n",
      "Epoch 5/100\n",
      "50000/50000 [==============================] - 3s 59us/step - loss: 1.1757 - acc: 0.5853 - val_loss: 1.3733 - val_acc: 0.5214\n",
      "Epoch 6/100\n",
      "50000/50000 [==============================] - 3s 59us/step - loss: 1.0965 - acc: 0.6112 - val_loss: 1.3694 - val_acc: 0.5204\n",
      "Epoch 7/100\n",
      "50000/50000 [==============================] - 3s 51us/step - loss: 1.0260 - acc: 0.6357 - val_loss: 1.3953 - val_acc: 0.5276\n",
      "Epoch 8/100\n",
      "50000/50000 [==============================] - 3s 52us/step - loss: 0.9503 - acc: 0.6632 - val_loss: 1.3954 - val_acc: 0.5364\n",
      "Epoch 9/100\n",
      "50000/50000 [==============================] - 3s 55us/step - loss: 0.8871 - acc: 0.6859 - val_loss: 1.4038 - val_acc: 0.5325\n",
      "Epoch 10/100\n",
      "50000/50000 [==============================] - 3s 54us/step - loss: 0.8220 - acc: 0.7085 - val_loss: 1.4546 - val_acc: 0.5347\n",
      "Epoch 11/100\n",
      "50000/50000 [==============================] - 2s 48us/step - loss: 0.7439 - acc: 0.7364 - val_loss: 1.5275 - val_acc: 0.5319\n",
      "Epoch 12/100\n",
      "50000/50000 [==============================] - 2s 44us/step - loss: 0.6957 - acc: 0.7540 - val_loss: 1.5291 - val_acc: 0.5409\n",
      "Epoch 13/100\n",
      "50000/50000 [==============================] - 2s 49us/step - loss: 0.6379 - acc: 0.7749 - val_loss: 1.6498 - val_acc: 0.5312\n",
      "Epoch 14/100\n",
      "50000/50000 [==============================] - 2s 46us/step - loss: 0.5821 - acc: 0.7917 - val_loss: 1.7190 - val_acc: 0.5350\n",
      "Epoch 15/100\n",
      "50000/50000 [==============================] - 2s 49us/step - loss: 0.5361 - acc: 0.8080 - val_loss: 1.7890 - val_acc: 0.5303\n",
      "Epoch 16/100\n",
      "50000/50000 [==============================] - 2s 45us/step - loss: 0.4946 - acc: 0.8256 - val_loss: 1.8268 - val_acc: 0.5391\n",
      "Epoch 17/100\n",
      "50000/50000 [==============================] - 2s 46us/step - loss: 0.4564 - acc: 0.8368 - val_loss: 1.9462 - val_acc: 0.5306\n",
      "Epoch 18/100\n",
      "50000/50000 [==============================] - 3s 60us/step - loss: 0.4299 - acc: 0.8486 - val_loss: 2.0159 - val_acc: 0.5318\n",
      "Epoch 19/100\n",
      "50000/50000 [==============================] - 3s 50us/step - loss: 0.3823 - acc: 0.8671 - val_loss: 2.0511 - val_acc: 0.5306\n",
      "Epoch 20/100\n",
      "50000/50000 [==============================] - 2s 49us/step - loss: 0.3624 - acc: 0.8721 - val_loss: 2.2162 - val_acc: 0.5283\n",
      "Epoch 21/100\n",
      "50000/50000 [==============================] - 2s 49us/step - loss: 0.3331 - acc: 0.8841 - val_loss: 2.2496 - val_acc: 0.5323\n",
      "Epoch 22/100\n",
      "50000/50000 [==============================] - 2s 47us/step - loss: 0.3095 - acc: 0.8910 - val_loss: 2.3870 - val_acc: 0.5346\n",
      "Epoch 23/100\n",
      "50000/50000 [==============================] - 3s 57us/step - loss: 0.2985 - acc: 0.8964 - val_loss: 2.4116 - val_acc: 0.5343\n",
      "Epoch 24/100\n",
      "50000/50000 [==============================] - 3s 53us/step - loss: 0.2711 - acc: 0.9045 - val_loss: 2.5313 - val_acc: 0.5335\n",
      "Epoch 25/100\n",
      "50000/50000 [==============================] - 2s 48us/step - loss: 0.2606 - acc: 0.9087 - val_loss: 2.4853 - val_acc: 0.5256\n",
      "Epoch 26/100\n",
      "50000/50000 [==============================] - 2s 48us/step - loss: 0.2454 - acc: 0.9135 - val_loss: 2.7100 - val_acc: 0.5323\n",
      "Epoch 27/100\n",
      "50000/50000 [==============================] - 3s 56us/step - loss: 0.2490 - acc: 0.9125 - val_loss: 2.6835 - val_acc: 0.5257\n",
      "Epoch 28/100\n",
      "50000/50000 [==============================] - 3s 53us/step - loss: 0.2188 - acc: 0.9241 - val_loss: 2.7556 - val_acc: 0.5281\n",
      "Epoch 29/100\n",
      "50000/50000 [==============================] - 2s 50us/step - loss: 0.2184 - acc: 0.9245 - val_loss: 2.7835 - val_acc: 0.5263\n",
      "Epoch 30/100\n",
      "50000/50000 [==============================] - 3s 58us/step - loss: 0.2019 - acc: 0.9303 - val_loss: 2.7758 - val_acc: 0.5251\n",
      "Epoch 31/100\n",
      "50000/50000 [==============================] - 3s 52us/step - loss: 0.2038 - acc: 0.9292 - val_loss: 2.9561 - val_acc: 0.5284\n",
      "Epoch 32/100\n",
      "50000/50000 [==============================] - 2s 49us/step - loss: 0.1757 - acc: 0.9393 - val_loss: 3.0620 - val_acc: 0.5330\n",
      "Epoch 33/100\n",
      "50000/50000 [==============================] - 3s 51us/step - loss: 0.2017 - acc: 0.9305 - val_loss: 3.0542 - val_acc: 0.5307\n",
      "Epoch 34/100\n",
      "50000/50000 [==============================] - 3s 56us/step - loss: 0.1763 - acc: 0.9397 - val_loss: 2.9975 - val_acc: 0.5343\n",
      "Epoch 35/100\n",
      "50000/50000 [==============================] - 3s 56us/step - loss: 0.1643 - acc: 0.9443 - val_loss: 3.0425 - val_acc: 0.5354\n",
      "Epoch 36/100\n",
      "50000/50000 [==============================] - 3s 50us/step - loss: 0.1738 - acc: 0.9403 - val_loss: 3.0115 - val_acc: 0.5252\n",
      "Epoch 37/100\n",
      "50000/50000 [==============================] - 3s 50us/step - loss: 0.1698 - acc: 0.9420 - val_loss: 3.0522 - val_acc: 0.5335\n",
      "Epoch 38/100\n",
      "50000/50000 [==============================] - 2s 49us/step - loss: 0.1655 - acc: 0.9442 - val_loss: 3.1148 - val_acc: 0.5289\n",
      "Epoch 39/100\n",
      "50000/50000 [==============================] - 3s 52us/step - loss: 0.1679 - acc: 0.9427 - val_loss: 3.1031 - val_acc: 0.5327\n",
      "Epoch 40/100\n",
      "50000/50000 [==============================] - 3s 57us/step - loss: 0.1426 - acc: 0.9507 - val_loss: 3.3179 - val_acc: 0.5290\n",
      "Epoch 41/100\n",
      "50000/50000 [==============================] - 3s 50us/step - loss: 0.1391 - acc: 0.9529 - val_loss: 3.3115 - val_acc: 0.5287\n",
      "Epoch 42/100\n",
      "50000/50000 [==============================] - 2s 50us/step - loss: 0.1351 - acc: 0.9531 - val_loss: 3.3435 - val_acc: 0.5325\n",
      "Epoch 43/100\n",
      "50000/50000 [==============================] - 2s 49us/step - loss: 0.1369 - acc: 0.9542 - val_loss: 3.3467 - val_acc: 0.5271\n",
      "Epoch 44/100\n",
      "50000/50000 [==============================] - 3s 60us/step - loss: 0.1438 - acc: 0.9513 - val_loss: 3.1013 - val_acc: 0.5366\n",
      "Epoch 45/100\n",
      "50000/50000 [==============================] - 3s 53us/step - loss: 0.1374 - acc: 0.9540 - val_loss: 3.3741 - val_acc: 0.5322\n",
      "Epoch 46/100\n",
      "50000/50000 [==============================] - 3s 57us/step - loss: 0.1271 - acc: 0.9572 - val_loss: 3.3973 - val_acc: 0.5330\n",
      "Epoch 47/100\n",
      "50000/50000 [==============================] - 3s 51us/step - loss: 0.1227 - acc: 0.9579 - val_loss: 3.3452 - val_acc: 0.5328\n",
      "Epoch 48/100\n",
      "50000/50000 [==============================] - 2s 48us/step - loss: 0.1188 - acc: 0.9609 - val_loss: 3.3343 - val_acc: 0.5366\n",
      "Epoch 49/100\n",
      "50000/50000 [==============================] - 2s 49us/step - loss: 0.1220 - acc: 0.9599 - val_loss: 3.3797 - val_acc: 0.5302\n",
      "Epoch 50/100\n",
      "50000/50000 [==============================] - 3s 51us/step - loss: 0.1305 - acc: 0.9554 - val_loss: 3.3945 - val_acc: 0.5354\n",
      "Epoch 51/100\n",
      "50000/50000 [==============================] - 3s 55us/step - loss: 0.1325 - acc: 0.9564 - val_loss: 3.4298 - val_acc: 0.5291\n",
      "Epoch 52/100\n",
      "50000/50000 [==============================] - 3s 64us/step - loss: 0.1107 - acc: 0.9627 - val_loss: 3.4772 - val_acc: 0.5373\n",
      "Epoch 53/100\n",
      "50000/50000 [==============================] - 2s 46us/step - loss: 0.1208 - acc: 0.9608 - val_loss: 3.3585 - val_acc: 0.5320\n",
      "Epoch 54/100\n",
      "50000/50000 [==============================] - 2s 46us/step - loss: 0.1009 - acc: 0.9661 - val_loss: 3.5749 - val_acc: 0.5281\n",
      "Epoch 55/100\n",
      "50000/50000 [==============================] - 2s 48us/step - loss: 0.1084 - acc: 0.9636 - val_loss: 3.5843 - val_acc: 0.5305\n",
      "Epoch 56/100\n",
      "50000/50000 [==============================] - 3s 56us/step - loss: 0.1348 - acc: 0.9548 - val_loss: 3.4531 - val_acc: 0.5269\n",
      "Epoch 57/100\n",
      "50000/50000 [==============================] - 3s 53us/step - loss: 0.1158 - acc: 0.9627 - val_loss: 3.5032 - val_acc: 0.5319\n",
      "Epoch 58/100\n",
      "50000/50000 [==============================] - 3s 57us/step - loss: 0.1109 - acc: 0.9634 - val_loss: 3.4944 - val_acc: 0.5309\n",
      "Epoch 59/100\n",
      "50000/50000 [==============================] - 3s 51us/step - loss: 0.1053 - acc: 0.9653 - val_loss: 3.5691 - val_acc: 0.5311\n",
      "Epoch 60/100\n",
      "50000/50000 [==============================] - 2s 49us/step - loss: 0.0932 - acc: 0.9692 - val_loss: 3.6112 - val_acc: 0.5328\n",
      "Epoch 61/100\n",
      "50000/50000 [==============================] - 2s 48us/step - loss: 0.0898 - acc: 0.9700 - val_loss: 3.5790 - val_acc: 0.5310\n",
      "Epoch 62/100\n",
      "50000/50000 [==============================] - 2s 49us/step - loss: 0.1164 - acc: 0.9621 - val_loss: 3.5524 - val_acc: 0.5281\n",
      "Epoch 63/100\n",
      "50000/50000 [==============================] - 3s 51us/step - loss: 0.1049 - acc: 0.9646 - val_loss: 3.6602 - val_acc: 0.5366\n",
      "Epoch 64/100\n",
      "50000/50000 [==============================] - 3s 54us/step - loss: 0.0855 - acc: 0.9719 - val_loss: 3.7403 - val_acc: 0.5270\n",
      "Epoch 65/100\n",
      "50000/50000 [==============================] - 2s 48us/step - loss: 0.1025 - acc: 0.9668 - val_loss: 3.6859 - val_acc: 0.5323\n",
      "Epoch 66/100\n",
      "50000/50000 [==============================] - 3s 51us/step - loss: 0.1023 - acc: 0.9667 - val_loss: 3.5389 - val_acc: 0.5298\n",
      "Epoch 67/100\n",
      "50000/50000 [==============================] - 3s 51us/step - loss: 0.0723 - acc: 0.9762 - val_loss: 3.7416 - val_acc: 0.5270\n",
      "Epoch 68/100\n",
      "50000/50000 [==============================] - 3s 56us/step - loss: 0.1074 - acc: 0.9637 - val_loss: 3.6225 - val_acc: 0.5361\n",
      "Epoch 69/100\n",
      "50000/50000 [==============================] - 3s 63us/step - loss: 0.1067 - acc: 0.9655 - val_loss: 3.5013 - val_acc: 0.5371\n",
      "Epoch 70/100\n",
      "50000/50000 [==============================] - 3s 52us/step - loss: 0.0800 - acc: 0.9728 - val_loss: 3.6599 - val_acc: 0.5353\n",
      "Epoch 71/100\n",
      "50000/50000 [==============================] - 3s 52us/step - loss: 0.0924 - acc: 0.9708 - val_loss: 3.5541 - val_acc: 0.5312\n",
      "Epoch 72/100\n",
      "50000/50000 [==============================] - 3s 53us/step - loss: 0.0994 - acc: 0.9669 - val_loss: 3.8086 - val_acc: 0.5326\n",
      "Epoch 73/100\n",
      "50000/50000 [==============================] - 2s 48us/step - loss: 0.0873 - acc: 0.9705 - val_loss: 3.6664 - val_acc: 0.5258\n",
      "Epoch 74/100\n",
      "50000/50000 [==============================] - 2s 50us/step - loss: 0.0857 - acc: 0.9721 - val_loss: 3.7556 - val_acc: 0.5257\n",
      "Epoch 75/100\n",
      "50000/50000 [==============================] - 3s 58us/step - loss: 0.0874 - acc: 0.9714 - val_loss: 3.8204 - val_acc: 0.5270\n",
      "Epoch 76/100\n",
      "50000/50000 [==============================] - 3s 50us/step - loss: 0.0806 - acc: 0.9731 - val_loss: 3.6932 - val_acc: 0.5248\n",
      "Epoch 77/100\n",
      "50000/50000 [==============================] - 2s 46us/step - loss: 0.0870 - acc: 0.9712 - val_loss: 3.6347 - val_acc: 0.5275\n",
      "Epoch 78/100\n",
      "50000/50000 [==============================] - 3s 56us/step - loss: 0.0978 - acc: 0.9672 - val_loss: 3.6735 - val_acc: 0.5384\n",
      "Epoch 79/100\n",
      "50000/50000 [==============================] - 2s 49us/step - loss: 0.0712 - acc: 0.9767 - val_loss: 3.8946 - val_acc: 0.5356\n",
      "Epoch 80/100\n",
      "50000/50000 [==============================] - 2s 49us/step - loss: 0.0656 - acc: 0.9785 - val_loss: 3.8528 - val_acc: 0.5396\n",
      "Epoch 81/100\n",
      "50000/50000 [==============================] - 2s 46us/step - loss: 0.0909 - acc: 0.9701 - val_loss: 3.6943 - val_acc: 0.5312\n",
      "Epoch 82/100\n",
      "50000/50000 [==============================] - 2s 50us/step - loss: 0.0832 - acc: 0.9730 - val_loss: 3.7179 - val_acc: 0.5289\n",
      "Epoch 83/100\n",
      "50000/50000 [==============================] - 2s 50us/step - loss: 0.0846 - acc: 0.9726 - val_loss: 3.7447 - val_acc: 0.5328\n",
      "Epoch 84/100\n",
      "50000/50000 [==============================] - 3s 52us/step - loss: 0.0760 - acc: 0.9748 - val_loss: 3.6706 - val_acc: 0.5274\n",
      "Epoch 85/100\n",
      "50000/50000 [==============================] - 2s 49us/step - loss: 0.0667 - acc: 0.9788 - val_loss: 3.7994 - val_acc: 0.5353\n",
      "Epoch 86/100\n",
      "50000/50000 [==============================] - 3s 58us/step - loss: 0.0728 - acc: 0.9766 - val_loss: 3.8816 - val_acc: 0.5318\n",
      "Epoch 87/100\n",
      "50000/50000 [==============================] - 3s 50us/step - loss: 0.1062 - acc: 0.9663 - val_loss: 3.5864 - val_acc: 0.5351\n",
      "Epoch 88/100\n",
      "50000/50000 [==============================] - 3s 52us/step - loss: 0.0742 - acc: 0.9765 - val_loss: 3.8022 - val_acc: 0.5298\n",
      "Epoch 89/100\n",
      "50000/50000 [==============================] - 3s 51us/step - loss: 0.0681 - acc: 0.9787 - val_loss: 3.6757 - val_acc: 0.5387\n",
      "Epoch 90/100\n",
      "50000/50000 [==============================] - 3s 57us/step - loss: 0.0651 - acc: 0.9791 - val_loss: 3.8219 - val_acc: 0.5369\n",
      "Epoch 91/100\n",
      "50000/50000 [==============================] - 3s 51us/step - loss: 0.0809 - acc: 0.9740 - val_loss: 3.7724 - val_acc: 0.5394\n",
      "Epoch 92/100\n",
      "50000/50000 [==============================] - 2s 50us/step - loss: 0.0709 - acc: 0.9769 - val_loss: 3.8465 - val_acc: 0.5306\n",
      "Epoch 93/100\n",
      "50000/50000 [==============================] - 3s 58us/step - loss: 0.0665 - acc: 0.9787 - val_loss: 3.7989 - val_acc: 0.5300\n",
      "Epoch 94/100\n",
      "50000/50000 [==============================] - 3s 54us/step - loss: 0.0839 - acc: 0.9737 - val_loss: 3.7889 - val_acc: 0.5292\n",
      "Epoch 95/100\n",
      "50000/50000 [==============================] - 2s 49us/step - loss: 0.0737 - acc: 0.9767 - val_loss: 3.7988 - val_acc: 0.5370\n",
      "Epoch 96/100\n",
      "50000/50000 [==============================] - 2s 50us/step - loss: 0.0704 - acc: 0.9769 - val_loss: 3.8652 - val_acc: 0.5324\n",
      "Epoch 97/100\n",
      "50000/50000 [==============================] - 2s 50us/step - loss: 0.0701 - acc: 0.9778 - val_loss: 3.7739 - val_acc: 0.5353\n",
      "Epoch 98/100\n",
      "50000/50000 [==============================] - 2s 48us/step - loss: 0.0789 - acc: 0.9752 - val_loss: 3.6560 - val_acc: 0.5316\n",
      "Epoch 99/100\n",
      "50000/50000 [==============================] - 2s 48us/step - loss: 0.0695 - acc: 0.9782 - val_loss: 3.8526 - val_acc: 0.5285\n",
      "Epoch 100/100\n",
      "50000/50000 [==============================] - 2s 49us/step - loss: 0.0704 - acc: 0.9771 - val_loss: 3.7124 - val_acc: 0.5351\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x7f8f88302588>"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(x_train, y_train, \n",
    "          epochs=EPOCHS, \n",
    "          batch_size=BATCH_SIZE, \n",
    "          validation_data=(x_test, y_test), \n",
    "          shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "\n",
    "train_loss = model.history.history[\"loss\"]\n",
    "valid_loss = model.history.history[\"val_loss\"]\n",
    "\n",
    "train_acc = model.history.history[\"acc\"]\n",
    "valid_acc = model.history.history[\"val_acc\"]\n",
    "\n",
    "plt.plot(range(len(train_loss)), train_loss, label=\"train loss\")\n",
    "plt.plot(range(len(valid_loss)), valid_loss, label=\"valid loss\")\n",
    "plt.legend()\n",
    "plt.title(\"Loss\")\n",
    "plt.show()\n",
    "\n",
    "plt.plot(range(len(train_acc)), train_acc, label=\"train accuracy\")\n",
    "plt.plot(range(len(valid_acc)), valid_acc, label=\"valid accuracy\")\n",
    "plt.legend()\n",
    "plt.title(\"Accuracy\")\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}