lecture_chap5_homework_ver2.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 第5回講義 宿題"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 課題. Tensorflowを用いて, MNISTを多層パーセプトロン(MLP)で学習せよ"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 注意"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- homework関数を完成させて提出してください\n",
    "    - 訓練データはtrain_X, train_y, テストデータはtest_Xで与えられます\n",
    "    - train_Xとtrain_yをtrain_X, train_yとvalid_X, valid_yに分けるなどしてモデルを学習させてください\n",
    "    - test_Xに対して予想ラベルpred_yを作り, homework関数の戻り値としてください\n",
    "- pred_yのtest_yに対する精度(F値)で評価します\n",
    "- 全体の実行時間がiLect上で60分を超えないようにしてください\n",
    "- homework関数の外には何も書かないでください (必要なものは全てhomework関数に入れてください)\n",
    "- 解答提出時には Answer Cell の内容のみを提出してください"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- CNNは使わないでください"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**`tf` の以下のモジュールはこの回では使用できないように制限されています. 注意してください.**\n",
    "```python\n",
    "tf.app\n",
    "tf.compat\n",
    "tf.contrib\n",
    "tf.erros\n",
    "tf.gfile\n",
    "tf.graph_util\n",
    "tf.image\n",
    "tf.layers\n",
    "tf.logging\n",
    "tf.losses\n",
    "tf.metrics\n",
    "tf.python_io\n",
    "tf.resource_loader\n",
    "tf.saved_model\n",
    "tf.sdca\n",
    "tf.sets\n",
    "tf.summary\n",
    "tf.sysconfig\n",
    "tf.test\n",
    "tf.train\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "次のセルのhomework関数を完成させて提出してください"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Answer Cell"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def homework(train_X, train_y, test_X):\n",
    "    # WRITE ME!\n",
    "    import time\n",
    "    import matplotlib.pyplot as plt\n",
    "    %matplotlib inline\n",
    "    \n",
    "    #definition\n",
    "    \n",
    "    eps=0.1\n",
    "    #epoch_num=60\n",
    "    rng = np.random.RandomState(1234)\n",
    "    random_state = 42\n",
    "    \n",
    "    x = tf.placeholder(tf.float32, [None, 784])\n",
    "    t = tf.placeholder(tf.float32, [None, 10])\n",
    "    \n",
    "    w1 = tf.Variable(np.random.uniform(low=-0.08, high=0.08, size=(784, 100)).astype('float32'), name='w1')\n",
    "    b1 = tf.Variable(tf.zeros(100), name='b1')\n",
    "    \n",
    "    w2 = tf.Variable(np.random.uniform(low=-0.08, high=0.08, size=(100, 50)).astype('float32'), name='w2')\n",
    "    b2 = tf.Variable(tf.zeros(50), name='b2')\n",
    "    \n",
    "    w3 = tf.Variable(np.random.uniform(low=-0.08, high=0.08, size=(50, 10)).astype('float32'), name='w3')\n",
    "    b3 = tf.Variable(tf.zeros(10), name='b3')\n",
    "    \n",
    "    params = [w1, b1, w2, b2, w3, b3]\n",
    "    \n",
    "    #forward\n",
    "    \n",
    "    u1 = tf.matmul(x,w1)+b1\n",
    "    z1 = tf.nn.sigmoid(u1)\n",
    "    \n",
    "    u2 = tf.matmul(z1,w2)+b2\n",
    "    z2 = tf.nn.sigmoid(u2)\n",
    "    \n",
    "    u3 = tf.matmul(z2,w3)+b3\n",
    "    z3 = tf.nn.softmax(u3)\n",
    "    \n",
    "    y=z3\n",
    "    \n",
    "    # Step3. 誤差関数の定義\n",
    "    #cost = -tf.reduce_mean(tf.reduce_sum(t*tf.log(y)))\n",
    "    # tf.log(0)によるnanを防ぐ\n",
    "    cost = -tf.reduce_mean(tf.reduce_sum(t*tf.log(tf.clip_by_value(y, 1e-10, 1.0)),axis=1))\n",
    "    #cost = -np.log(y[0,np.nonzero(t)])\n",
    "    \n",
    "    #delta\n",
    "    gw1, gb1, gw2, gb2, gw3, gb3 = tf.gradients(cost, params)\n",
    "    updates = [\n",
    "        w1.assign_add(-eps*gw1), # 勾配降下法\n",
    "        b1.assign_add(-eps*gb1),\n",
    "        w2.assign_add(-eps*gw2), # 勾配降下法\n",
    "        b2.assign_add(-eps*gb2),\n",
    "        w3.assign_add(-eps*gw3), # 勾配降下法\n",
    "        b3.assign_add(-eps*gb3),\n",
    "    ]\n",
    "\n",
    "    train = tf.group(*updates)\n",
    "    \n",
    "    valid = tf.argmax(y, 1)\n",
    "    \n",
    "    start_time=time.time()\n",
    "    \n",
    "    #Online learning\n",
    "    \n",
    "    #cost_plt =[]\n",
    "    #sess = tf.Session()\n",
    "    #sess.run(tf.global_variables_initializer()) # 重みの初期化\n",
    "    #for epoch in range(epoch_num):\n",
    "    #    cost_arry = []\n",
    "    #    for x_in, y_in in zip(train_X, train_y):\n",
    "    #        x_in = x_in.reshape(1,len(x_in))\n",
    "    #        y_tmp = np.zeros(10).reshape(1, 10)\n",
    "    #        y_tmp[0,y_in] = 1.0\n",
    "    #        y_in=y_tmp\n",
    "    #        _cost, _ = sess.run([cost, train], feed_dict={x: x_in, t: y_in})\n",
    "    #        cost_arry.append(_cost)\n",
    "    #        \n",
    "    #    cost_plt.append(np.sum(cost_arry))\n",
    "    #    tttttt = time.time() - start\n",
    "    #    print (\"epoch\",epoch,\"time\",tttttt,\"cost\",cost_plt[len(cost_plt)-1])\n",
    "        #print(cost_plt)\n",
    "    #plt.plot(cost_plt)\n",
    "    \n",
    "    #batch learning\n",
    "    n_epochs = 300\n",
    "    batch_size = 100\n",
    "    n_batches = train_X.shape[0] // batch_size\n",
    "    \n",
    "    nb_classes = 10                                #train_y one-hot\n",
    "    targets = train_y.reshape(-1)\n",
    "    one_hot_targets = np.eye(nb_classes)[targets]\n",
    "    train_y = one_hot_targets\n",
    "    \n",
    "    # Step5. 学習\n",
    "    cost_plt =[]\n",
    "    with tf.Session() as sess:\n",
    "        print(\"start time\",start_time)\n",
    "        sess.run(tf.global_variables_initializer())\n",
    "        for epoch in range(n_epochs):\n",
    "            train_X, train_y = shuffle(train_X, train_y, random_state=random_state)\n",
    "            cost_arry = []\n",
    "            for i in range(n_batches):\n",
    "                start = i * batch_size\n",
    "                end = start + batch_size\n",
    "                _cost, _ = sess.run([cost,train], feed_dict={x: train_X[start:end], t: train_y[start:end]})\n",
    "                cost_arry.append(_cost)\n",
    "                \n",
    "            cost_plt.append(np.sum(cost_arry))\n",
    "            tttttt = time.time()-start_time\n",
    "            #predict\n",
    "            print (\"epoch\",epoch,\"time\",tttttt,\"cost\",cost_plt[len(cost_plt)-1])\n",
    "        pred_y = sess.run(valid, feed_dict={x: test_X})\n",
    "        print(time.time()-start_time)\n",
    "        plt.plot(cost_plt)\n",
    "    \n",
    "    # Step6. 予測\n",
    "    #pred_y = np.argmax(sess.run(y, feed_dict={x: test_X}),1)\n",
    "\n",
    "    #sess.close()\n",
    "    \n",
    "    \n",
    "    \n",
    "    return pred_y"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- 以下のvalidate_homework関数を用いてエラーが起きないか動作確認をして下さい。\n",
    "- 提出に際して、以下のscore_homework関数で60分で実行が終わることを確認して下さい。\n",
    "- 評価は以下のscore_homework関数で行われますが、random_stateの値は変更されます。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Checker Cell (for student)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.utils import shuffle\n",
    "from sklearn.metrics import f1_score\n",
    "from sklearn.datasets import fetch_mldata\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "\n",
    "\n",
    "def load_mnist():\n",
    "    mnist = fetch_mldata('MNIST original')\n",
    "    mnist_X, mnist_y = shuffle(mnist.data.astype('float32'),\n",
    "                               mnist.target.astype('int32'), random_state=42)\n",
    "\n",
    "    mnist_X = mnist_X / 255.0\n",
    "\n",
    "    return train_test_split(mnist_X, mnist_y,\n",
    "                test_size=0.2,\n",
    "                random_state=42)\n",
    "\n",
    "def validate_homework():\n",
    "    train_X, test_X, train_y, test_y = load_mnist()\n",
    "\n",
    "    # validate for small dataset\n",
    "    train_X_mini = train_X[:1000]\n",
    "    train_y_mini = train_y[:1000]\n",
    "    test_X_mini = test_X[:1000]\n",
    "    test_y_mini = test_y[:1000]\n",
    "\n",
    "    pred_y = homework(train_X_mini, train_y_mini, test_X_mini)\n",
    "    print(f1_score(test_y_mini, pred_y, average='macro'))\n",
    "\n",
    "def score_homework():\n",
    "    train_X, test_X, train_y, test_y = load_mnist()\n",
    "    pred_y = homework(train_X, train_y, test_X)\n",
    "    print(f1_score(test_y, pred_y, average='macro'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "start time 1494686526.321419\n",
      "epoch 0 time 0.2755117416381836 cost 23.049\n",
      "epoch 1 time 0.31070828437805176 cost 22.9837\n",
      "epoch 2 time 0.3263242244720459 cost 22.9917\n",
      "epoch 3 time 0.35721516609191895 cost 22.9766\n",
      "epoch 4 time 0.3788623809814453 cost 22.9721\n",
      "epoch 5 time 0.4261753559112549 cost 22.9841\n",
      "epoch 6 time 0.5052967071533203 cost 22.9738\n",
      "epoch 7 time 0.5353395938873291 cost 22.9618\n",
      "epoch 8 time 0.5951194763183594 cost 22.9671\n",
      "epoch 9 time 0.678790807723999 cost 22.9695\n",
      "epoch 10 time 0.7319478988647461 cost 22.9739\n",
      "epoch 11 time 0.7619924545288086 cost 22.9481\n",
      "epoch 12 time 0.795008659362793 cost 22.9631\n",
      "epoch 13 time 0.8763680458068848 cost 22.9478\n",
      "epoch 14 time 0.9410252571105957 cost 22.9507\n",
      "epoch 15 time 0.9821336269378662 cost 22.9528\n",
      "epoch 16 time 1.0152854919433594 cost 22.9516\n",
      "epoch 17 time 1.0550963878631592 cost 22.9374\n",
      "epoch 18 time 1.0805208683013916 cost 22.942\n",
      "epoch 19 time 1.1273467540740967 cost 22.9341\n",
      "epoch 20 time 1.1661121845245361 cost 22.9051\n",
      "epoch 21 time 1.1898822784423828 cost 22.9181\n",
      "epoch 22 time 1.2229011058807373 cost 22.9294\n",
      "epoch 23 time 1.2469353675842285 cost 22.9071\n",
      "epoch 24 time 1.2729988098144531 cost 22.9117\n",
      "epoch 25 time 1.3107020854949951 cost 22.8942\n",
      "epoch 26 time 1.3757479190826416 cost 22.9011\n",
      "epoch 27 time 1.405245304107666 cost 22.9036\n",
      "epoch 28 time 1.4282793998718262 cost 22.8793\n",
      "epoch 29 time 1.5114147663116455 cost 22.8897\n",
      "epoch 30 time 1.542717456817627 cost 22.8693\n",
      "epoch 31 time 1.583097219467163 cost 22.8784\n",
      "epoch 32 time 1.605130910873413 cost 22.8503\n",
      "epoch 33 time 1.6261615753173828 cost 22.8656\n",
      "epoch 34 time 1.6491963863372803 cost 22.8476\n",
      "epoch 35 time 1.6702375411987305 cost 22.8389\n",
      "epoch 36 time 1.7059483528137207 cost 22.8269\n",
      "epoch 37 time 1.7299845218658447 cost 22.8281\n",
      "epoch 38 time 1.7510154247283936 cost 22.81\n",
      "epoch 39 time 1.7740669250488281 cost 22.8011\n",
      "epoch 40 time 1.794011116027832 cost 22.7893\n",
      "epoch 41 time 1.816044569015503 cost 22.7846\n",
      "epoch 42 time 1.8370766639709473 cost 22.7891\n",
      "epoch 43 time 1.8571069240570068 cost 22.7479\n",
      "epoch 44 time 1.8869197368621826 cost 22.7363\n",
      "epoch 45 time 1.9260151386260986 cost 22.7305\n",
      "epoch 46 time 1.9772992134094238 cost 22.7082\n",
      "epoch 47 time 2.0013411045074463 cost 22.6934\n",
      "epoch 48 time 2.0243747234344482 cost 22.672\n",
      "epoch 49 time 2.0464086532592773 cost 22.6515\n",
      "epoch 50 time 2.068449020385742 cost 22.6242\n",
      "epoch 51 time 2.0890161991119385 cost 22.603\n",
      "epoch 52 time 2.1187758445739746 cost 22.5887\n",
      "epoch 53 time 2.1466968059539795 cost 22.5454\n",
      "epoch 54 time 2.1692380905151367 cost 22.5252\n",
      "epoch 55 time 2.198864698410034 cost 22.4782\n",
      "epoch 56 time 2.2307138442993164 cost 22.4554\n",
      "epoch 57 time 2.2547500133514404 cost 22.3999\n",
      "epoch 58 time 2.2752974033355713 cost 22.3523\n",
      "epoch 59 time 2.2974815368652344 cost 22.3069\n",
      "epoch 60 time 2.319514751434326 cost 22.2549\n",
      "epoch 61 time 2.339545488357544 cost 22.1902\n",
      "epoch 62 time 2.3863279819488525 cost 22.1314\n",
      "epoch 63 time 2.4093453884124756 cost 22.0613\n",
      "epoch 64 time 2.429490089416504 cost 21.9621\n",
      "epoch 65 time 2.4636952877044678 cost 21.8929\n",
      "epoch 66 time 2.4864718914031982 cost 21.7887\n",
      "epoch 67 time 2.508504867553711 cost 21.698\n",
      "epoch 68 time 2.538386821746826 cost 21.5676\n",
      "epoch 69 time 2.594276189804077 cost 21.461\n",
      "epoch 70 time 2.6363213062286377 cost 21.3351\n",
      "epoch 71 time 2.6953542232513428 cost 21.1937\n",
      "epoch 72 time 2.7253990173339844 cost 21.0531\n",
      "epoch 73 time 2.745476484298706 cost 20.9009\n",
      "epoch 74 time 2.7806143760681152 cost 20.7407\n",
      "epoch 75 time 2.8126933574676514 cost 20.5614\n",
      "epoch 76 time 2.8357279300689697 cost 20.4075\n",
      "epoch 77 time 2.869371175765991 cost 20.2414\n",
      "epoch 78 time 2.889029026031494 cost 20.0501\n",
      "epoch 79 time 2.9110708236694336 cost 19.8671\n",
      "epoch 80 time 2.9351158142089844 cost 19.6969\n",
      "epoch 81 time 2.956130266189575 cost 19.5245\n",
      "epoch 82 time 2.9788503646850586 cost 19.3274\n",
      "epoch 83 time 3.0003890991210938 cost 19.188\n",
      "epoch 84 time 3.0377299785614014 cost 18.9851\n",
      "epoch 85 time 3.086894989013672 cost 18.8141\n",
      "epoch 86 time 3.1240007877349854 cost 18.6363\n",
      "epoch 87 time 3.150038719177246 cost 18.4539\n",
      "epoch 88 time 3.1700305938720703 cost 18.2752\n",
      "epoch 89 time 3.206972599029541 cost 18.1151\n",
      "epoch 90 time 3.2390031814575195 cost 17.9148\n",
      "epoch 91 time 3.28694224357605 cost 17.74\n",
      "epoch 92 time 3.3358216285705566 cost 17.5467\n",
      "epoch 93 time 3.3578550815582275 cost 17.3336\n",
      "epoch 94 time 3.383039712905884 cost 17.1497\n",
      "epoch 95 time 3.4423553943634033 cost 16.9454\n",
      "epoch 96 time 3.4855127334594727 cost 16.7472\n",
      "epoch 97 time 3.508547306060791 cost 16.5364\n",
      "epoch 98 time 3.5305802822113037 cost 16.3074\n",
      "epoch 99 time 3.553615093231201 cost 16.0809\n",
      "epoch 100 time 3.5771660804748535 cost 15.8842\n",
      "epoch 101 time 3.598186492919922 cost 15.6126\n",
      "epoch 102 time 3.622222423553467 cost 15.363\n",
      "epoch 103 time 3.6442556381225586 cost 15.1269\n",
      "epoch 104 time 3.665287494659424 cost 14.8976\n",
      "epoch 105 time 3.702965497970581 cost 14.612\n",
      "epoch 106 time 3.776646137237549 cost 14.3707\n",
      "epoch 107 time 3.8087546825408936 cost 14.1066\n",
      "epoch 108 time 3.830789089202881 cost 13.8639\n",
      "epoch 109 time 3.8458499908447266 cost 13.6062\n",
      "epoch 110 time 3.913126230239868 cost 13.3551\n",
      "epoch 111 time 3.9782955646514893 cost 13.104\n",
      "epoch 112 time 4.019866704940796 cost 12.875\n",
      "epoch 113 time 4.041899919509888 cost 12.6434\n",
      "epoch 114 time 4.081460952758789 cost 12.4285\n",
      "epoch 115 time 4.1444244384765625 cost 12.2043\n",
      "epoch 116 time 4.201412200927734 cost 11.9798\n",
      "epoch 117 time 4.232455253601074 cost 11.7878\n",
      "epoch 118 time 4.253486633300781 cost 11.5992\n",
      "epoch 119 time 4.349541187286377 cost 11.4169\n",
      "epoch 120 time 4.419092893600464 cost 11.2363\n",
      "epoch 121 time 4.4562859535217285 cost 11.0718\n",
      "epoch 122 time 4.478198289871216 cost 10.8988\n",
      "epoch 123 time 4.495262861251831 cost 10.7487\n",
      "epoch 124 time 4.521108865737915 cost 10.6143\n",
      "epoch 125 time 4.547147274017334 cost 10.4481\n",
      "epoch 126 time 4.568685531616211 cost 10.3233\n",
      "epoch 127 time 4.599913597106934 cost 10.1884\n",
      "epoch 128 time 4.636917352676392 cost 10.0567\n",
      "epoch 129 time 4.658950328826904 cost 9.93757\n",
      "epoch 130 time 4.688447713851929 cost 9.8285\n",
      "epoch 131 time 4.711482763290405 cost 9.72219\n",
      "epoch 132 time 4.7335169315338135 cost 9.61678\n",
      "epoch 133 time 4.761584043502808 cost 9.50497\n",
      "epoch 134 time 4.798199892044067 cost 9.4095\n",
      "epoch 135 time 4.832074880599976 cost 9.30004\n",
      "epoch 136 time 4.857622385025024 cost 9.21855\n",
      "epoch 137 time 4.8791728019714355 cost 9.12508\n",
      "epoch 138 time 4.9023566246032715 cost 9.03674\n",
      "epoch 139 time 4.928672790527344 cost 8.95106\n",
      "epoch 140 time 4.953084707260132 cost 8.88566\n",
      "epoch 141 time 4.977135181427002 cost 8.79737\n",
      "epoch 142 time 4.9984869956970215 cost 8.72315\n",
      "epoch 143 time 5.023902177810669 cost 8.6458\n",
      "epoch 144 time 5.068558931350708 cost 8.55299\n",
      "epoch 145 time 5.092314004898071 cost 8.49838\n",
      "epoch 146 time 5.117351531982422 cost 8.42194\n",
      "epoch 147 time 5.138383626937866 cost 8.35408\n",
      "epoch 148 time 5.159235954284668 cost 8.29458\n",
      "epoch 149 time 5.186925888061523 cost 8.2221\n",
      "epoch 150 time 5.210963010787964 cost 8.15707\n",
      "epoch 151 time 5.227646112442017 cost 8.09542\n",
      "epoch 152 time 5.275028467178345 cost 8.03073\n",
      "epoch 153 time 5.3010780811309814 cost 7.97117\n",
      "epoch 154 time 5.321108102798462 cost 7.91538\n",
      "epoch 155 time 5.3291637897491455 cost 7.85311\n",
      "epoch 156 time 5.398918390274048 cost 7.79502\n",
      "epoch 157 time 5.443181753158569 cost 7.75064\n",
      "epoch 158 time 5.46802830696106 cost 7.69846\n",
      "epoch 159 time 5.492640972137451 cost 7.63949\n",
      "epoch 160 time 5.5156755447387695 cost 7.5752\n",
      "epoch 161 time 5.5397114753723145 cost 7.52249\n",
      "epoch 162 time 5.563747406005859 cost 7.47149\n",
      "epoch 163 time 5.589099168777466 cost 7.42958\n",
      "epoch 164 time 5.612133502960205 cost 7.36927\n",
      "epoch 165 time 5.634166479110718 cost 7.32851\n",
      "epoch 166 time 5.65619969367981 cost 7.27394\n",
      "epoch 167 time 5.679752826690674 cost 7.22241\n",
      "epoch 168 time 5.702061176300049 cost 7.18548\n",
      "epoch 169 time 5.726097106933594 cost 7.13275\n",
      "epoch 170 time 5.748131036758423 cost 7.0905\n",
      "epoch 171 time 5.7701685428619385 cost 7.03981\n",
      "epoch 172 time 5.792219638824463 cost 6.99356\n",
      "epoch 173 time 5.818258285522461 cost 6.95102\n",
      "epoch 174 time 5.843296051025391 cost 6.89352\n",
      "epoch 175 time 5.867332696914673 cost 6.8474\n",
      "epoch 176 time 5.891391277313232 cost 6.79643\n",
      "epoch 177 time 5.9124228954315186 cost 6.77027\n",
      "epoch 178 time 5.935457468032837 cost 6.71713\n",
      "epoch 179 time 5.9584925174713135 cost 6.66823\n",
      "epoch 180 time 5.980544090270996 cost 6.61828\n",
      "epoch 181 time 6.003583908081055 cost 6.5707\n",
      "epoch 182 time 6.025616884231567 cost 6.5352\n",
      "epoch 183 time 6.047650098800659 cost 6.48157\n",
      "epoch 184 time 6.071192979812622 cost 6.45521\n",
      "epoch 185 time 6.091278791427612 cost 6.41961\n",
      "epoch 186 time 6.130127191543579 cost 6.34942\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 187 time 6.164099216461182 cost 6.31889\n",
      "epoch 188 time 6.189266681671143 cost 6.26323\n",
      "epoch 189 time 6.211300611495972 cost 6.22317\n",
      "epoch 190 time 6.2383410930633545 cost 6.20422\n",
      "epoch 191 time 6.260374069213867 cost 6.13344\n",
      "epoch 192 time 6.282928943634033 cost 6.09823\n",
      "epoch 193 time 6.303476333618164 cost 6.05577\n",
      "epoch 194 time 6.325146436691284 cost 6.0174\n",
      "epoch 195 time 6.347179651260376 cost 5.97062\n",
      "epoch 196 time 6.3717241287231445 cost 5.9357\n",
      "epoch 197 time 6.394566059112549 cost 5.87372\n",
      "epoch 198 time 6.423374652862549 cost 5.84207\n",
      "epoch 199 time 6.545874357223511 cost 5.7982\n",
      "epoch 200 time 6.591282844543457 cost 5.76283\n",
      "epoch 201 time 6.616319417953491 cost 5.71797\n",
      "epoch 202 time 6.6526196002960205 cost 5.67566\n",
      "epoch 203 time 6.675163269042969 cost 5.63137\n",
      "epoch 204 time 6.7155773639678955 cost 5.5906\n",
      "epoch 205 time 6.7839601039886475 cost 5.54954\n",
      "epoch 206 time 6.82528281211853 cost 5.52017\n",
      "epoch 207 time 6.845373153686523 cost 5.47266\n",
      "epoch 208 time 6.915632724761963 cost 5.43874\n",
      "epoch 209 time 6.949349880218506 cost 5.37469\n",
      "epoch 210 time 6.976403474807739 cost 5.34296\n",
      "epoch 211 time 6.998602628707886 cost 5.29739\n",
      "epoch 212 time 7.036497592926025 cost 5.27282\n",
      "epoch 213 time 7.05853009223938 cost 5.22097\n",
      "epoch 214 time 7.081076622009277 cost 5.18943\n",
      "epoch 215 time 7.115318775177002 cost 5.14058\n",
      "epoch 216 time 7.146195888519287 cost 5.10231\n",
      "epoch 217 time 7.215599536895752 cost 5.06777\n",
      "epoch 218 time 7.284444808959961 cost 5.01769\n",
      "epoch 219 time 7.30747652053833 cost 4.97919\n",
      "epoch 220 time 7.335591554641724 cost 4.94407\n",
      "epoch 221 time 7.3597252368927 cost 4.90846\n",
      "epoch 222 time 7.455181837081909 cost 4.86754\n",
      "epoch 223 time 7.526494979858398 cost 4.82735\n",
      "epoch 224 time 7.576849937438965 cost 4.79463\n",
      "epoch 225 time 7.6514668464660645 cost 4.76481\n",
      "epoch 226 time 7.73703145980835 cost 4.7169\n",
      "epoch 227 time 7.781628847122192 cost 4.68656\n",
      "epoch 228 time 7.899608135223389 cost 4.65027\n",
      "epoch 229 time 7.964808464050293 cost 4.60803\n",
      "epoch 230 time 7.994974136352539 cost 4.5785\n",
      "epoch 231 time 8.012050151824951 cost 4.54098\n",
      "epoch 232 time 8.078088998794556 cost 4.50025\n",
      "epoch 233 time 8.099753141403198 cost 4.46543\n",
      "epoch 234 time 8.115397930145264 cost 4.43585\n",
      "epoch 235 time 8.18934965133667 cost 4.39527\n",
      "epoch 236 time 8.220395803451538 cost 4.36873\n",
      "epoch 237 time 8.244466304779053 cost 4.32894\n",
      "epoch 238 time 8.30025601387024 cost 4.30731\n",
      "epoch 239 time 8.347137928009033 cost 4.26378\n",
      "epoch 240 time 8.414960861206055 cost 4.22955\n",
      "epoch 241 time 8.441991090774536 cost 4.19748\n",
      "epoch 242 time 8.462081670761108 cost 4.16262\n",
      "epoch 243 time 8.547355890274048 cost 4.13094\n",
      "epoch 244 time 8.600291013717651 cost 4.1118\n",
      "epoch 245 time 8.648509740829468 cost 4.07542\n",
      "epoch 246 time 8.671550273895264 cost 4.03984\n",
      "epoch 247 time 8.731857299804688 cost 4.01365\n",
      "epoch 248 time 8.8009033203125 cost 3.97957\n",
      "epoch 249 time 8.8711519241333 cost 3.95993\n",
      "epoch 250 time 8.903318405151367 cost 3.92551\n",
      "epoch 251 time 8.91753101348877 cost 3.8909\n",
      "epoch 252 time 8.946562767028809 cost 3.86025\n",
      "epoch 253 time 8.962188243865967 cost 3.84097\n",
      "epoch 254 time 8.991394758224487 cost 3.80241\n",
      "epoch 255 time 9.02513313293457 cost 3.77673\n",
      "epoch 256 time 9.085554599761963 cost 3.74665\n",
      "epoch 257 time 9.1279935836792 cost 3.71279\n",
      "epoch 258 time 9.154032468795776 cost 3.70239\n",
      "epoch 259 time 9.217246055603027 cost 3.65856\n",
      "epoch 260 time 9.317343473434448 cost 3.63899\n",
      "epoch 261 time 9.36182975769043 cost 3.61634\n",
      "epoch 262 time 9.390374898910522 cost 3.59127\n",
      "epoch 263 time 9.44260048866272 cost 3.56893\n",
      "epoch 264 time 9.548592805862427 cost 3.5318\n",
      "epoch 265 time 9.597387790679932 cost 3.51101\n",
      "epoch 266 time 9.625428199768066 cost 3.48707\n",
      "epoch 267 time 9.65787124633789 cost 3.45354\n",
      "epoch 268 time 9.70189619064331 cost 3.44089\n",
      "epoch 269 time 9.733113527297974 cost 3.40909\n",
      "epoch 270 time 9.773765802383423 cost 3.38541\n",
      "epoch 271 time 9.798003435134888 cost 3.36375\n",
      "epoch 272 time 9.822040319442749 cost 3.34277\n",
      "epoch 273 time 9.845122814178467 cost 3.31161\n",
      "epoch 274 time 9.917386531829834 cost 3.27991\n",
      "epoch 275 time 9.94654631614685 cost 3.26761\n",
      "epoch 276 time 9.99472451210022 cost 3.25123\n",
      "epoch 277 time 10.009825468063354 cost 3.22288\n",
      "epoch 278 time 10.038204908370972 cost 3.18914\n",
      "epoch 279 time 10.05823540687561 cost 3.17249\n",
      "epoch 280 time 10.08128571510315 cost 3.15406\n",
      "epoch 281 time 10.117433547973633 cost 3.12425\n",
      "epoch 282 time 10.16429328918457 cost 3.10886\n",
      "epoch 283 time 10.201632738113403 cost 3.07839\n",
      "epoch 284 time 10.232874393463135 cost 3.07417\n",
      "epoch 285 time 10.266363382339478 cost 3.04951\n",
      "epoch 286 time 10.288944005966187 cost 3.03349\n",
      "epoch 287 time 10.375021934509277 cost 2.99768\n",
      "epoch 288 time 10.449198484420776 cost 2.99114\n",
      "epoch 289 time 10.504471778869629 cost 2.96258\n",
      "epoch 290 time 10.52950930595398 cost 2.93278\n",
      "epoch 291 time 10.545180082321167 cost 2.91592\n",
      "epoch 292 time 10.682725429534912 cost 2.89458\n",
      "epoch 293 time 10.745863914489746 cost 2.88611\n",
      "epoch 294 time 10.772407293319702 cost 2.85969\n",
      "epoch 295 time 10.834190130233765 cost 2.84277\n",
      "epoch 296 time 10.865447521209717 cost 2.81394\n",
      "epoch 297 time 10.903369188308716 cost 2.79534\n",
      "epoch 298 time 10.934818983078003 cost 2.78242\n",
      "epoch 299 time 10.958860635757446 cost 2.75676\n",
      "10.965871334075928\n",
      "0.810948399612\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX4AAAD8CAYAAABw1c+bAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1e2617c8cc0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "validate_homework()\n",
    "#score_homework()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "start time 1494686538.3141868\n",
      "epoch 0 time 4.4780333042144775 cost 1285.08\n",
      "epoch 1 time 6.815724849700928 cost 1090.26\n",
      "epoch 2 time 8.374126434326172 cost 595.346\n",
      "epoch 3 time 10.768674612045288 cost 370.348\n",
      "epoch 4 time 13.739470720291138 cost 290.178\n",
      "epoch 5 time 17.729554414749146 cost 248.096\n",
      "epoch 6 time 19.09781527519226 cost 220.673\n",
      "epoch 7 time 21.112181186676025 cost 201.621\n",
      "epoch 8 time 24.434876680374146 cost 187.082\n",
      "epoch 9 time 29.269721746444702 cost 174.951\n",
      "epoch 10 time 33.926270484924316 cost 164.482\n",
      "epoch 11 time 38.162578105926514 cost 155.08\n",
      "epoch 12 time 42.83657884597778 cost 146.297\n",
      "epoch 13 time 47.403119802474976 cost 138.417\n",
      "epoch 14 time 51.60737657546997 cost 130.805\n",
      "epoch 15 time 55.32478046417236 cost 123.975\n",
      "epoch 16 time 57.662930965423584 cost 117.584\n",
      "epoch 17 time 59.79977750778198 cost 111.971\n",
      "epoch 18 time 62.8772873878479 cost 106.451\n",
      "epoch 19 time 67.25136399269104 cost 101.785\n",
      "epoch 20 time 71.13779830932617 cost 97.2867\n",
      "epoch 21 time 75.82672882080078 cost 93.2901\n",
      "epoch 22 time 80.41648840904236 cost 89.5961\n",
      "epoch 23 time 85.35395622253418 cost 86.2403\n",
      "epoch 24 time 89.52286529541016 cost 83.0046\n",
      "epoch 25 time 94.11402249336243 cost 79.959\n",
      "epoch 26 time 99.02065563201904 cost 77.2332\n",
      "epoch 27 time 103.32102751731873 cost 74.6949\n",
      "epoch 28 time 107.62488317489624 cost 72.2103\n",
      "epoch 29 time 112.00797009468079 cost 69.9121\n",
      "epoch 30 time 116.56420850753784 cost 67.8093\n",
      "epoch 31 time 120.22214961051941 cost 65.6811\n",
      "epoch 32 time 124.82837271690369 cost 63.6922\n",
      "epoch 33 time 129.84574151039124 cost 61.8791\n",
      "epoch 34 time 133.14867639541626 cost 60.1188\n",
      "epoch 35 time 135.80514287948608 cost 58.4882\n",
      "epoch 36 time 137.48850870132446 cost 56.8861\n",
      "epoch 37 time 140.67244267463684 cost 55.3793\n",
      "epoch 38 time 142.20406818389893 cost 53.8747\n",
      "epoch 39 time 143.47745990753174 cost 52.5558\n",
      "epoch 40 time 145.72066831588745 cost 51.0992\n",
      "epoch 41 time 147.89032793045044 cost 49.9302\n",
      "epoch 42 time 150.16734147071838 cost 48.6516\n",
      "epoch 43 time 153.8849070072174 cost 47.3831\n",
      "epoch 44 time 158.60259342193604 cost 46.279\n",
      "epoch 45 time 163.17752027511597 cost 45.2678\n",
      "epoch 46 time 167.98442125320435 cost 44.1381\n",
      "epoch 47 time 172.7333037853241 cost 43.1512\n",
      "epoch 48 time 177.21983194351196 cost 42.0976\n",
      "epoch 49 time 181.5937204360962 cost 41.0334\n",
      "epoch 50 time 186.30921339988708 cost 40.2175\n",
      "epoch 51 time 191.23080563545227 cost 39.3484\n",
      "epoch 52 time 195.97440433502197 cost 38.37\n",
      "epoch 53 time 200.62559485435486 cost 37.6199\n",
      "epoch 54 time 204.70622062683105 cost 36.7012\n",
      "epoch 55 time 208.33360195159912 cost 35.918\n",
      "epoch 56 time 212.83678793907166 cost 35.1555\n",
      "epoch 57 time 217.42500185966492 cost 34.4009\n",
      "epoch 58 time 222.3161280155182 cost 33.5743\n",
      "epoch 59 time 226.76459383964539 cost 32.9385\n",
      "epoch 60 time 230.42531776428223 cost 32.2307\n",
      "epoch 61 time 234.72740697860718 cost 31.5484\n",
      "epoch 62 time 238.84765768051147 cost 30.8538\n",
      "epoch 63 time 242.81537652015686 cost 30.2176\n",
      "epoch 64 time 247.45774912834167 cost 29.5029\n",
      "epoch 65 time 249.60667634010315 cost 28.9137\n",
      "epoch 66 time 251.34585547447205 cost 28.2893\n",
      "epoch 67 time 252.92647647857666 cost 27.7445\n",
      "epoch 68 time 256.135370016098 cost 27.2669\n",
      "epoch 69 time 257.9385566711426 cost 26.5605\n",
      "epoch 70 time 259.67621874809265 cost 26.0371\n",
      "epoch 71 time 261.37484788894653 cost 25.5073\n",
      "epoch 72 time 263.9889008998871 cost 24.9787\n",
      "epoch 73 time 268.159690618515 cost 24.4964\n",
      "epoch 74 time 272.38210010528564 cost 23.9998\n",
      "epoch 75 time 276.582731962204 cost 23.503\n",
      "epoch 76 time 281.03948998451233 cost 23.0517\n",
      "epoch 77 time 285.72258377075195 cost 22.5951\n",
      "epoch 78 time 289.30918645858765 cost 22.1732\n",
      "epoch 79 time 293.1398837566376 cost 21.6478\n",
      "epoch 80 time 294.74127888679504 cost 21.3022\n",
      "epoch 81 time 297.3555381298065 cost 20.8857\n",
      "epoch 82 time 299.83962202072144 cost 20.4616\n",
      "epoch 83 time 301.16792607307434 cost 20.0541\n",
      "epoch 84 time 302.42962980270386 cost 19.6292\n",
      "epoch 85 time 303.86214327812195 cost 19.1865\n",
      "epoch 86 time 305.22878670692444 cost 18.8582\n",
      "epoch 87 time 306.5442855358124 cost 18.5013\n",
      "epoch 88 time 310.91678524017334 cost 18.0844\n",
      "epoch 89 time 315.40542697906494 cost 17.7389\n",
      "epoch 90 time 318.65472650527954 cost 17.4148\n",
      "epoch 91 time 322.67327785491943 cost 17.0364\n",
      "epoch 92 time 324.98855447769165 cost 16.7226\n",
      "epoch 93 time 327.539537191391 cost 16.3832\n",
      "epoch 94 time 331.2795481681824 cost 16.1031\n",
      "epoch 95 time 335.4464490413666 cost 15.749\n",
      "epoch 96 time 338.9829180240631 cost 15.4759\n",
      "epoch 97 time 343.4488615989685 cost 15.0973\n",
      "epoch 98 time 346.35341930389404 cost 14.8552\n",
      "epoch 99 time 348.13163328170776 cost 14.5785\n",
      "epoch 100 time 349.447229385376 cost 14.3044\n",
      "epoch 101 time 352.6667504310608 cost 13.9921\n",
      "epoch 102 time 356.6382884979248 cost 13.7611\n",
      "epoch 103 time 360.9265019893646 cost 13.5465\n",
      "epoch 104 time 365.7585587501526 cost 13.2621\n",
      "epoch 105 time 370.2763648033142 cost 12.9997\n",
      "epoch 106 time 374.7946107387543 cost 12.76\n",
      "epoch 107 time 379.9489743709564 cost 12.482\n",
      "epoch 108 time 381.2613482475281 cost 12.2216\n",
      "epoch 109 time 383.61431074142456 cost 11.9973\n",
      "epoch 110 time 386.65294909477234 cost 11.7942\n",
      "epoch 111 time 388.76321959495544 cost 11.5857\n",
      "epoch 112 time 390.65607380867004 cost 11.3496\n",
      "epoch 113 time 392.9194755554199 cost 11.1348\n",
      "epoch 114 time 394.4186038970947 cost 10.8919\n",
      "epoch 115 time 396.27938199043274 cost 10.7594\n",
      "epoch 116 time 398.8538112640381 cost 10.5441\n",
      "epoch 117 time 403.43395805358887 cost 10.3435\n",
      "epoch 118 time 408.30427956581116 cost 10.1585\n",
      "epoch 119 time 412.55870604515076 cost 9.98752\n",
      "epoch 120 time 417.46459770202637 cost 9.75457\n",
      "epoch 121 time 422.6513373851776 cost 9.61066\n",
      "epoch 122 time 427.31033754348755 cost 9.40266\n",
      "epoch 123 time 432.0644118785858 cost 9.24598\n",
      "epoch 124 time 436.9661445617676 cost 9.11259\n",
      "epoch 125 time 440.2286145687103 cost 8.94376\n",
      "epoch 126 time 442.1724896430969 cost 8.79053\n",
      "epoch 127 time 443.8667001724243 cost 8.583\n",
      "epoch 128 time 446.64335346221924 cost 8.44918\n",
      "epoch 129 time 451.3311285972595 cost 8.29546\n",
      "epoch 130 time 454.96908807754517 cost 8.18331\n",
      "epoch 131 time 460.07080078125 cost 8.02946\n",
      "epoch 132 time 464.8447847366333 cost 7.86244\n",
      "epoch 133 time 469.7336401939392 cost 7.70794\n",
      "epoch 134 time 474.0843267440796 cost 7.60106\n",
      "epoch 135 time 478.38698959350586 cost 7.46417\n",
      "epoch 136 time 483.40705156326294 cost 7.35132\n",
      "epoch 137 time 487.76489210128784 cost 7.23197\n",
      "epoch 138 time 492.8675801753998 cost 7.08488\n",
      "epoch 139 time 498.3206741809845 cost 6.96546\n",
      "epoch 140 time 503.0320248603821 cost 6.85024\n",
      "epoch 141 time 507.98752427101135 cost 6.7474\n",
      "epoch 142 time 510.08519077301025 cost 6.62332\n",
      "epoch 143 time 513.0017983913422 cost 6.50574\n",
      "epoch 144 time 517.8427195549011 cost 6.42515\n",
      "epoch 145 time 522.9048337936401 cost 6.28436\n",
      "epoch 146 time 526.9884114265442 cost 6.22095\n",
      "epoch 147 time 528.3075203895569 cost 6.10264\n",
      "epoch 148 time 530.0310456752777 cost 6.00667\n",
      "epoch 149 time 531.5640349388123 cost 5.91184\n",
      "epoch 150 time 533.9137568473816 cost 5.81577\n",
      "epoch 151 time 537.7692832946777 cost 5.73659\n",
      "epoch 152 time 542.9228146076202 cost 5.62048\n",
      "epoch 153 time 547.5765960216522 cost 5.55595\n",
      "epoch 154 time 551.5357496738434 cost 5.4672\n",
      "epoch 155 time 556.1813745498657 cost 5.38503\n",
      "epoch 156 time 561.1229650974274 cost 5.29231\n",
      "epoch 157 time 566.026692867279 cost 5.20583\n",
      "epoch 158 time 570.0302016735077 cost 5.15574\n",
      "epoch 159 time 574.6666207313538 cost 5.08183\n",
      "epoch 160 time 579.5063219070435 cost 5.00779\n",
      "epoch 161 time 584.600382566452 cost 4.90928\n",
      "epoch 162 time 589.3696460723877 cost 4.84826\n",
      "epoch 163 time 593.4182274341583 cost 4.78643\n",
      "epoch 164 time 597.3184399604797 cost 4.71046\n",
      "epoch 165 time 602.1898157596588 cost 4.6425\n",
      "epoch 166 time 607.1318552494049 cost 4.56762\n",
      "epoch 167 time 611.6543085575104 cost 4.51231\n",
      "epoch 168 time 615.8890798091888 cost 4.46336\n",
      "epoch 169 time 620.0381197929382 cost 4.3981\n",
      "epoch 170 time 624.683584690094 cost 4.31647\n",
      "epoch 171 time 629.006281375885 cost 4.26584\n",
      "epoch 172 time 632.3116371631622 cost 4.21667\n",
      "epoch 173 time 633.9945440292358 cost 4.15705\n",
      "epoch 174 time 636.9801073074341 cost 4.08793\n",
      "epoch 175 time 638.6627759933472 cost 4.04277\n",
      "epoch 176 time 642.0988192558289 cost 3.9859\n",
      "epoch 177 time 645.5556335449219 cost 3.94166\n",
      "epoch 178 time 648.28728723526 cost 3.88063\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch 179 time 650.0146250724792 cost 3.83051\n",
      "epoch 180 time 651.4888846874237 cost 3.78114\n",
      "epoch 181 time 653.3437702655792 cost 3.73916\n",
      "epoch 182 time 654.845755815506 cost 3.68839\n",
      "epoch 183 time 657.9144983291626 cost 3.63115\n",
      "epoch 184 time 660.4416308403015 cost 3.59287\n",
      "epoch 185 time 662.8635938167572 cost 3.54752\n",
      "epoch 186 time 665.081960439682 cost 3.50517\n",
      "epoch 187 time 666.998162984848 cost 3.46994\n",
      "epoch 188 time 669.4763793945312 cost 3.41545\n",
      "epoch 189 time 672.3095853328705 cost 3.36876\n",
      "epoch 190 time 675.6917490959167 cost 3.34889\n",
      "epoch 191 time 679.1304092407227 cost 3.29665\n",
      "epoch 192 time 682.668395280838 cost 3.2557\n",
      "epoch 193 time 685.8650722503662 cost 3.22256\n",
      "epoch 194 time 689.6377100944519 cost 3.18342\n",
      "epoch 195 time 693.3410720825195 cost 3.1446\n",
      "epoch 196 time 696.3292427062988 cost 3.11574\n",
      "epoch 197 time 698.4088003635406 cost 3.07596\n",
      "epoch 198 time 701.9495975971222 cost 3.03619\n",
      "epoch 199 time 705.4728584289551 cost 3.00584\n",
      "epoch 200 time 708.291501045227 cost 2.97359\n",
      "epoch 201 time 711.7565400600433 cost 2.93567\n",
      "epoch 202 time 715.3897798061371 cost 2.90402\n",
      "epoch 203 time 718.9665627479553 cost 2.87723\n",
      "epoch 204 time 721.5017004013062 cost 2.8483\n",
      "epoch 205 time 723.3179123401642 cost 2.80527\n",
      "epoch 206 time 725.1995801925659 cost 2.77516\n",
      "epoch 207 time 728.1477808952332 cost 2.75768\n",
      "epoch 208 time 729.8568599224091 cost 2.72029\n",
      "epoch 209 time 731.5553364753723 cost 2.69565\n",
      "epoch 210 time 733.3589763641357 cost 2.6658\n",
      "epoch 211 time 735.2158777713776 cost 2.64029\n",
      "epoch 212 time 736.4639866352081 cost 2.61354\n",
      "epoch 213 time 737.7203364372253 cost 2.58334\n",
      "epoch 214 time 739.5951392650604 cost 2.55664\n",
      "epoch 215 time 742.65265417099 cost 2.54024\n",
      "epoch 216 time 744.1500065326691 cost 2.50217\n",
      "epoch 217 time 745.4927830696106 cost 2.47953\n",
      "epoch 218 time 747.0332098007202 cost 2.44842\n",
      "epoch 219 time 748.609570980072 cost 2.42769\n",
      "epoch 220 time 749.8610360622406 cost 2.40403\n",
      "epoch 221 time 751.6666328907013 cost 2.37381\n",
      "epoch 222 time 755.1867077350616 cost 2.35605\n",
      "epoch 223 time 759.4568145275116 cost 2.34044\n",
      "epoch 224 time 764.1029739379883 cost 2.30825\n",
      "epoch 225 time 768.3737487792969 cost 2.29269\n",
      "epoch 226 time 772.554196357727 cost 2.26898\n",
      "epoch 227 time 776.8739743232727 cost 2.25027\n",
      "epoch 228 time 781.6311619281769 cost 2.22928\n",
      "epoch 229 time 786.5319528579712 cost 2.20704\n",
      "epoch 230 time 788.6714811325073 cost 2.18555\n",
      "epoch 231 time 790.2006237506866 cost 2.16736\n",
      "epoch 232 time 792.2941880226135 cost 2.14523\n",
      "epoch 233 time 796.0924582481384 cost 2.12422\n",
      "epoch 234 time 800.1966941356659 cost 2.10402\n",
      "epoch 235 time 803.7550768852234 cost 2.09003\n",
      "epoch 236 time 807.2197666168213 cost 2.06667\n",
      "epoch 237 time 808.6411418914795 cost 2.04587\n",
      "epoch 238 time 811.6046588420868 cost 2.0329\n",
      "epoch 239 time 814.1572887897491 cost 2.00873\n",
      "epoch 240 time 818.0304765701294 cost 1.997\n",
      "epoch 241 time 819.334764957428 cost 1.98135\n",
      "epoch 242 time 820.8499202728271 cost 1.96619\n",
      "epoch 243 time 825.2464530467987 cost 1.94441\n",
      "epoch 244 time 826.6016743183136 cost 1.92812\n",
      "epoch 245 time 828.2691631317139 cost 1.91076\n",
      "epoch 246 time 830.7994043827057 cost 1.89187\n",
      "epoch 247 time 832.2470667362213 cost 1.86958\n",
      "epoch 248 time 836.2425653934479 cost 1.85898\n",
      "epoch 249 time 838.2007527351379 cost 1.84687\n",
      "epoch 250 time 840.9296808242798 cost 1.83181\n",
      "epoch 251 time 846.02991938591 cost 1.81657\n",
      "epoch 252 time 849.2195591926575 cost 1.79887\n",
      "epoch 253 time 850.5035877227783 cost 1.78748\n",
      "epoch 254 time 851.9570162296295 cost 1.77094\n",
      "epoch 255 time 856.6008303165436 cost 1.75534\n",
      "epoch 256 time 861.6527972221375 cost 1.74245\n",
      "epoch 257 time 866.49462890625 cost 1.73109\n",
      "epoch 258 time 871.3622221946716 cost 1.71764\n",
      "epoch 259 time 876.5353000164032 cost 1.70159\n",
      "epoch 260 time 881.8556897640228 cost 1.68649\n",
      "epoch 261 time 887.2295236587524 cost 1.67272\n",
      "epoch 262 time 892.1244204044342 cost 1.66234\n",
      "epoch 263 time 893.5337557792664 cost 1.65177\n",
      "epoch 264 time 895.6405465602875 cost 1.63262\n",
      "epoch 265 time 898.0467302799225 cost 1.6193\n",
      "epoch 266 time 899.9413585662842 cost 1.60564\n",
      "epoch 267 time 901.7051093578339 cost 1.59586\n",
      "epoch 268 time 904.7905328273773 cost 1.58522\n",
      "epoch 269 time 906.4497756958008 cost 1.57406\n",
      "epoch 270 time 910.5628840923309 cost 1.55959\n",
      "epoch 271 time 914.5661234855652 cost 1.54474\n",
      "epoch 272 time 916.0727968215942 cost 1.53333\n",
      "epoch 273 time 918.3489322662354 cost 1.52043\n",
      "epoch 274 time 920.1585230827332 cost 1.50993\n",
      "epoch 275 time 922.992999792099 cost 1.49596\n",
      "epoch 276 time 925.2721123695374 cost 1.48676\n",
      "epoch 277 time 927.5640232563019 cost 1.47458\n",
      "epoch 278 time 931.2524452209473 cost 1.46579\n",
      "epoch 279 time 934.1594676971436 cost 1.45204\n",
      "epoch 280 time 936.2567319869995 cost 1.44516\n",
      "epoch 281 time 939.5432987213135 cost 1.43308\n",
      "epoch 282 time 941.8782715797424 cost 1.42124\n",
      "epoch 283 time 943.4710068702698 cost 1.40921\n",
      "epoch 284 time 945.2100412845612 cost 1.39808\n",
      "epoch 285 time 947.2537534236908 cost 1.39073\n",
      "epoch 286 time 948.7308535575867 cost 1.37822\n",
      "epoch 287 time 952.074319601059 cost 1.36873\n",
      "epoch 288 time 953.441145658493 cost 1.3602\n",
      "epoch 289 time 955.1746165752411 cost 1.34821\n",
      "epoch 290 time 956.8504388332367 cost 1.33813\n",
      "epoch 291 time 958.2358751296997 cost 1.32867\n",
      "epoch 292 time 960.4347848892212 cost 1.31734\n",
      "epoch 293 time 962.9220798015594 cost 1.30898\n",
      "epoch 294 time 966.3081085681915 cost 1.29768\n",
      "epoch 295 time 969.6204407215118 cost 1.29022\n",
      "epoch 296 time 971.0228462219238 cost 1.2819\n",
      "epoch 297 time 974.1200087070465 cost 1.27159\n",
      "epoch 298 time 978.123868227005 cost 1.26233\n",
      "epoch 299 time 982.6268038749695 cost 1.25304\n",
      "982.6801881790161\n",
      "0.977269093313\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYEAAAD8CAYAAACRkhiPAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nzKyFOQTMzFrY/wctjK6M1j3gsgAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1e26a43cfd0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "score_homework()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.5.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}