neuron.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['batches.meta', 'data_batch_1', 'data_batch_2', 'data_batch_3', 'data_batch_4', 'data_batch_5', 'readme.html', 'test_batch']\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "import pickle\n",
    "import numpy as np\n",
    "import os\n",
    "\n",
    "CIFAR_DIR = './cifar-10-batches-py'\n",
    "print(os.listdir(CIFAR_DIR))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(10000, 3072) (10000,)\n",
      "(2000, 3072) (2000,)\n",
      "[[ 0.63921569  0.6627451   0.60784314 ...  0.6         0.5372549\n",
      "   0.6       ]\n",
      " [ 0.36470588  0.16078431  0.22352941 ... -0.74901961 -0.25490196\n",
      "  -0.17647059]\n",
      " [-0.1372549  -0.14509804 -0.14509804 ... -0.70980392 -0.74901961\n",
      "  -0.70980392]\n",
      " ...\n",
      " [-1.         -1.         -1.         ... -0.3254902  -0.41176471\n",
      "  -0.11372549]\n",
      " [-0.76470588 -0.42745098 -0.2        ... -0.83529412 -0.60784314\n",
      "  -0.12941176]\n",
      " [-0.7254902  -0.73333333 -0.73333333 ...  0.70196078  0.71764706\n",
      "   0.74117647]] [0 1 0 0 1 0 0 1 1 0]\n"
     ]
    }
   ],
   "source": [
    "def load_data(filename):\n",
    "    \"\"\"read data from data file.\"\"\"\n",
    "    with open (filename, \"rb\") as f:\n",
    "        data = pickle.load(f,encoding='bytes')\n",
    "        return data[b'data'], data[b'labels']\n",
    "    \n",
    "class CifarData:\n",
    "    def __init__(self, filenames, need_shuffle):\n",
    "        all_data = []\n",
    "        all_labels = []\n",
    "        for filename in filenames:\n",
    "            data, labels = load_data(filename)\n",
    "            for item, label in zip(data, labels):\n",
    "                if label in [0, 1]:\n",
    "                    all_data.append(item)\n",
    "                    all_labels.append(label)\n",
    "        self._data = np.vstack(all_data)\n",
    "        self._data = self._data / 127.5 -1 # 归一化\n",
    "        self._labels = np.hstack(all_labels)\n",
    "        print(self._data.shape, self._labels.shape)\n",
    "        self._num_examples = self._data.shape[0]\n",
    "        self._need_shuffle = need_shuffle\n",
    "        self._indicator = 0 ## 遍历起始位置\n",
    "        if self._need_shuffle:\n",
    "            self._shuffle_data()\n",
    "    \n",
    "    def _shuffle_data(self):\n",
    "        # [0, 1, 2, 3, 4, 5] -> [5, 3, 2, 4, 0 ,1]\n",
    "        p = np.random.permutation(self._num_examples)\n",
    "        self._data = self._data[p]\n",
    "        self._labels = self._labels[p]\n",
    "    \n",
    "    def next_batch(self, batch_size):\n",
    "        \"\"\"return batch_size examples as a batch.\"\"\"\n",
    "        end_indicator = self._indicator + batch_size\n",
    "        if end_indicator > self._num_examples:\n",
    "            if self._need_shuffle:\n",
    "                self._shuffle_data()\n",
    "                self._indicator = 0 \n",
    "                end_indicator = batch_size\n",
    "            else:\n",
    "                raise Exception(\"have no more example\")\n",
    "        \n",
    "        if end_indicator > self._num_examples:\n",
    "            raise Exception(\"batch size is larger than all examples\")\n",
    "        batch_data = self._data[self._indicator: end_indicator]    \n",
    "        batch_labels = self._labels[self._indicator: end_indicator]\n",
    "        self._indicator = end_indicator\n",
    "        return batch_data, batch_labels\n",
    "    \n",
    "train_filenames = [os.path.join(CIFAR_DIR, 'data_batch_%d' % i) for i in range(1,6)]\n",
    "test_filenames = [os.path.join(CIFAR_DIR, 'test_batch')]\n",
    "\n",
    "train_data = CifarData(train_filenames, True)\n",
    "test_data = CifarData(test_filenames, False)\n",
    "\n",
    "batch_data, batch_labels = train_data.next_batch(10)\n",
    "print(batch_data, batch_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "x = tf.placeholder(tf.float32, [None, 3072])\n",
    "# [None]\n",
    "y = tf.placeholder(tf.float64, [None])\n",
    "\n",
    "# (3072, 1)\n",
    "w = tf.get_variable('w', [x.get_shape()[-1], 1],\n",
    "                   initializer=tf.random_normal_initializer(0, 1))\n",
    "\n",
    "# (1, )\n",
    "b = tf.get_variable('b', [1],\n",
    "                   initializer=tf.constant_initializer(0.0))\n",
    "\n",
    "# [None, 3072] * [3072, 1] = [None, 1]\n",
    "y_ = tf.matmul(x, w) + b\n",
    "\n",
    "# [None, 1]\n",
    "p_y_1 = tf.nn.sigmoid(y_)\n",
    "# [None, 1]\n",
    "y_reshaped = tf.reshape(y, (-1, 1))\n",
    "y_reshaped_float = tf.cast(y_reshaped, tf.float32)\n",
    "loss = tf.reduce_mean(tf.square(y_reshaped_float - p_y_1))\n",
    "\n",
    "# bool\n",
    "predict = p_y_1 > 0.5\n",
    "# [1,0,1,0]\n",
    "correct_prediction = tf.equal(tf.cast(predict, tf.float64), y_reshaped)\n",
    "accuracy= tf.reduce_mean(tf.cast(correct_prediction, tf.float64))\n",
    "\n",
    "with tf.name_scope('train_op'):\n",
    "    train_op = tf.train.AdamOptimizer(1e-3).minimize(loss)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[Train] step: 0, loss: 0.49594, acc:0.50000\n",
      "(2000, 3072) (2000,)\n",
      "[Test] Step: 1, acc: 0.38400\n",
      "[Train] step: 500, loss: 0.19872, acc:0.80000\n",
      "[Train] step: 1000, loss: 0.08715, acc:0.90000\n",
      "[Train] step: 1500, loss: 0.20201, acc:0.80000\n",
      "[Train] step: 2000, loss: 0.15320, acc:0.85000\n",
      "[Train] step: 2500, loss: 0.19967, acc:0.80000\n",
      "[Train] step: 3000, loss: 0.34265, acc:0.65000\n",
      "[Train] step: 3500, loss: 0.20000, acc:0.80000\n",
      "[Train] step: 4000, loss: 0.15003, acc:0.85000\n",
      "[Train] step: 4500, loss: 0.20019, acc:0.80000\n",
      "[Train] step: 5000, loss: 0.20001, acc:0.80000\n",
      "(2000, 3072) (2000,)\n",
      "[Test] Step: 5001, acc: 0.80950\n",
      "[Train] step: 5500, loss: 0.21316, acc:0.75000\n",
      "[Train] step: 6000, loss: 0.33133, acc:0.65000\n",
      "[Train] step: 6500, loss: 0.16976, acc:0.80000\n",
      "[Train] step: 7000, loss: 0.20000, acc:0.80000\n",
      "[Train] step: 7500, loss: 0.18899, acc:0.80000\n",
      "[Train] step: 8000, loss: 0.14741, acc:0.85000\n",
      "[Train] step: 8500, loss: 0.29969, acc:0.70000\n",
      "[Train] step: 9000, loss: 0.13672, acc:0.80000\n",
      "[Train] step: 9500, loss: 0.34972, acc:0.65000\n",
      "[Train] step: 10000, loss: 0.16857, acc:0.80000\n",
      "(2000, 3072) (2000,)\n",
      "[Test] Step: 10001, acc: 0.81400\n",
      "[Train] step: 10500, loss: 0.19619, acc:0.75000\n",
      "[Train] step: 11000, loss: 0.10039, acc:0.90000\n",
      "[Train] step: 11500, loss: 0.05329, acc:0.95000\n",
      "[Train] step: 12000, loss: 0.19935, acc:0.80000\n",
      "[Train] step: 12500, loss: 0.27949, acc:0.70000\n",
      "[Train] step: 13000, loss: 0.10005, acc:0.90000\n",
      "[Train] step: 13500, loss: 0.14992, acc:0.85000\n",
      "[Train] step: 14000, loss: 0.22463, acc:0.75000\n",
      "[Train] step: 14500, loss: 0.12399, acc:0.85000\n",
      "[Train] step: 15000, loss: 0.24838, acc:0.75000\n",
      "(2000, 3072) (2000,)\n",
      "[Test] Step: 15001, acc: 0.81650\n",
      "[Train] step: 15500, loss: 0.10012, acc:0.90000\n",
      "[Train] step: 16000, loss: 0.34962, acc:0.65000\n",
      "[Train] step: 16500, loss: 0.15441, acc:0.85000\n",
      "[Train] step: 17000, loss: 0.19095, acc:0.80000\n",
      "[Train] step: 17500, loss: 0.05003, acc:0.95000\n",
      "[Train] step: 18000, loss: 0.20051, acc:0.80000\n",
      "[Train] step: 18500, loss: 0.10050, acc:0.90000\n",
      "[Train] step: 19000, loss: 0.20025, acc:0.80000\n",
      "[Train] step: 19500, loss: 0.30000, acc:0.70000\n",
      "[Train] step: 20000, loss: 0.15010, acc:0.85000\n",
      "(2000, 3072) (2000,)\n",
      "[Test] Step: 20001, acc: 0.81900\n",
      "[Train] step: 20500, loss: 0.20502, acc:0.80000\n",
      "[Train] step: 21000, loss: 0.10184, acc:0.90000\n",
      "[Train] step: 21500, loss: 0.20000, acc:0.80000\n",
      "[Train] step: 22000, loss: 0.20015, acc:0.80000\n",
      "[Train] step: 22500, loss: 0.24988, acc:0.75000\n",
      "[Train] step: 23000, loss: 0.25000, acc:0.75000\n",
      "[Train] step: 23500, loss: 0.00001, acc:1.00000\n",
      "[Train] step: 24000, loss: 0.10001, acc:0.90000\n",
      "[Train] step: 24500, loss: 0.02299, acc:0.95000\n",
      "[Train] step: 25000, loss: 0.10126, acc:0.90000\n",
      "(2000, 3072) (2000,)\n",
      "[Test] Step: 25001, acc: 0.81700\n",
      "[Train] step: 25500, loss: 0.10007, acc:0.90000\n",
      "[Train] step: 26000, loss: 0.24943, acc:0.75000\n",
      "[Train] step: 26500, loss: 0.14995, acc:0.85000\n",
      "[Train] step: 27000, loss: 0.21315, acc:0.75000\n",
      "[Train] step: 27500, loss: 0.10045, acc:0.90000\n",
      "[Train] step: 28000, loss: 0.05005, acc:0.95000\n",
      "[Train] step: 28500, loss: 0.00279, acc:1.00000\n",
      "[Train] step: 29000, loss: 0.20222, acc:0.80000\n",
      "[Train] step: 29500, loss: 0.20000, acc:0.80000\n",
      "[Train] step: 30000, loss: 0.15000, acc:0.85000\n",
      "(2000, 3072) (2000,)\n",
      "[Test] Step: 30001, acc: 0.82050\n",
      "[Train] step: 30500, loss: 0.10000, acc:0.90000\n",
      "[Train] step: 31000, loss: 0.10235, acc:0.90000\n",
      "[Train] step: 31500, loss: 0.14996, acc:0.85000\n",
      "[Train] step: 32000, loss: 0.37168, acc:0.60000\n",
      "[Train] step: 32500, loss: 0.04999, acc:0.95000\n",
      "[Train] step: 33000, loss: 0.15036, acc:0.85000\n",
      "[Train] step: 33500, loss: 0.25226, acc:0.75000\n",
      "[Train] step: 34000, loss: 0.10071, acc:0.90000\n",
      "[Train] step: 34500, loss: 0.24996, acc:0.75000\n",
      "[Train] step: 35000, loss: 0.10086, acc:0.90000\n",
      "(2000, 3072) (2000,)\n",
      "[Test] Step: 35001, acc: 0.82250\n",
      "[Train] step: 35500, loss: 0.10337, acc:0.90000\n",
      "[Train] step: 36000, loss: 0.11855, acc:0.85000\n",
      "[Train] step: 36500, loss: 0.19153, acc:0.80000\n",
      "[Train] step: 37000, loss: 0.05000, acc:0.95000\n",
      "[Train] step: 37500, loss: 0.09998, acc:0.90000\n",
      "[Train] step: 38000, loss: 0.19836, acc:0.80000\n",
      "[Train] step: 38500, loss: 0.15000, acc:0.85000\n",
      "[Train] step: 39000, loss: 0.24995, acc:0.75000\n",
      "[Train] step: 39500, loss: 0.11454, acc:0.85000\n",
      "[Train] step: 40000, loss: 0.05029, acc:0.95000\n",
      "(2000, 3072) (2000,)\n",
      "[Test] Step: 40001, acc: 0.82350\n",
      "[Train] step: 40500, loss: 0.20002, acc:0.80000\n",
      "[Train] step: 41000, loss: 0.15613, acc:0.85000\n",
      "[Train] step: 41500, loss: 0.10119, acc:0.90000\n",
      "[Train] step: 42000, loss: 0.00113, acc:1.00000\n",
      "[Train] step: 42500, loss: 0.24990, acc:0.75000\n",
      "[Train] step: 43000, loss: 0.25000, acc:0.75000\n",
      "[Train] step: 43500, loss: 0.26126, acc:0.75000\n",
      "[Train] step: 44000, loss: 0.10000, acc:0.90000\n",
      "[Train] step: 44500, loss: 0.25261, acc:0.75000\n",
      "[Train] step: 45000, loss: 0.20013, acc:0.80000\n",
      "(2000, 3072) (2000,)\n",
      "[Test] Step: 45001, acc: 0.82200\n",
      "[Train] step: 45500, loss: 0.05441, acc:0.95000\n",
      "[Train] step: 46000, loss: 0.15000, acc:0.85000\n",
      "[Train] step: 46500, loss: 0.30467, acc:0.70000\n",
      "[Train] step: 47000, loss: 0.20062, acc:0.80000\n",
      "[Train] step: 47500, loss: 0.14383, acc:0.85000\n",
      "[Train] step: 48000, loss: 0.29495, acc:0.70000\n",
      "[Train] step: 48500, loss: 0.11933, acc:0.85000\n",
      "[Train] step: 49000, loss: 0.05004, acc:0.95000\n",
      "[Train] step: 49500, loss: 0.34991, acc:0.65000\n",
      "[Train] step: 50000, loss: 0.05018, acc:0.95000\n",
      "(2000, 3072) (2000,)\n",
      "[Test] Step: 50001, acc: 0.82250\n",
      "[Train] step: 50500, loss: 0.20021, acc:0.80000\n",
      "[Train] step: 51000, loss: 0.24981, acc:0.75000\n",
      "[Train] step: 51500, loss: 0.10112, acc:0.90000\n",
      "[Train] step: 52000, loss: 0.15003, acc:0.85000\n",
      "[Train] step: 52500, loss: 0.15000, acc:0.85000\n",
      "[Train] step: 53000, loss: 0.05002, acc:0.95000\n",
      "[Train] step: 53500, loss: 0.20499, acc:0.80000\n",
      "[Train] step: 54000, loss: 0.30528, acc:0.70000\n",
      "[Train] step: 54500, loss: 0.00000, acc:1.00000\n",
      "[Train] step: 55000, loss: 0.14376, acc:0.85000\n",
      "(2000, 3072) (2000,)\n",
      "[Test] Step: 55001, acc: 0.82250\n",
      "[Train] step: 55500, loss: 0.05568, acc:0.95000\n",
      "[Train] step: 56000, loss: 0.18501, acc:0.80000\n",
      "[Train] step: 56500, loss: 0.06623, acc:0.90000\n",
      "[Train] step: 57000, loss: 0.10021, acc:0.90000\n",
      "[Train] step: 57500, loss: 0.05156, acc:0.95000\n",
      "[Train] step: 58000, loss: 0.20001, acc:0.80000\n",
      "[Train] step: 58500, loss: 0.15482, acc:0.85000\n",
      "[Train] step: 59000, loss: 0.11378, acc:0.85000\n",
      "[Train] step: 59500, loss: 0.07439, acc:0.90000\n",
      "[Train] step: 60000, loss: 0.25000, acc:0.75000\n",
      "(2000, 3072) (2000,)\n",
      "[Test] Step: 60001, acc: 0.81900\n",
      "[Train] step: 60500, loss: 0.15000, acc:0.85000\n",
      "[Train] step: 61000, loss: 0.15151, acc:0.85000\n",
      "[Train] step: 61500, loss: 0.10047, acc:0.90000\n",
      "[Train] step: 62000, loss: 0.04442, acc:0.95000\n",
      "[Train] step: 62500, loss: 0.10003, acc:0.90000\n",
      "[Train] step: 63000, loss: 0.10014, acc:0.90000\n",
      "[Train] step: 63500, loss: 0.10372, acc:0.90000\n",
      "[Train] step: 64000, loss: 0.21803, acc:0.80000\n",
      "[Train] step: 64500, loss: 0.05002, acc:0.95000\n",
      "[Train] step: 65000, loss: 0.25000, acc:0.75000\n",
      "(2000, 3072) (2000,)\n",
      "[Test] Step: 65001, acc: 0.82250\n",
      "[Train] step: 65500, loss: 0.10000, acc:0.90000\n",
      "[Train] step: 66000, loss: 0.20002, acc:0.80000\n",
      "[Train] step: 66500, loss: 0.10056, acc:0.90000\n",
      "[Train] step: 67000, loss: 0.15159, acc:0.85000\n",
      "[Train] step: 67500, loss: 0.05285, acc:0.95000\n",
      "[Train] step: 68000, loss: 0.21136, acc:0.80000\n",
      "[Train] step: 68500, loss: 0.15448, acc:0.85000\n",
      "[Train] step: 69000, loss: 0.15799, acc:0.85000\n",
      "[Train] step: 69500, loss: 0.21094, acc:0.80000\n",
      "[Train] step: 70000, loss: 0.10000, acc:0.90000\n",
      "(2000, 3072) (2000,)\n",
      "[Test] Step: 70001, acc: 0.82200\n",
      "[Train] step: 70500, loss: 0.35000, acc:0.65000\n",
      "[Train] step: 71000, loss: 0.10149, acc:0.90000\n",
      "[Train] step: 71500, loss: 0.06793, acc:0.90000\n",
      "[Train] step: 72000, loss: 0.15228, acc:0.85000\n",
      "[Train] step: 72500, loss: 0.10182, acc:0.90000\n",
      "[Train] step: 73000, loss: 0.06443, acc:0.90000\n",
      "[Train] step: 73500, loss: 0.00404, acc:1.00000\n",
      "[Train] step: 74000, loss: 0.20038, acc:0.80000\n",
      "[Train] step: 74500, loss: 0.24877, acc:0.75000\n",
      "[Train] step: 75000, loss: 0.00000, acc:1.00000\n",
      "(2000, 3072) (2000,)\n",
      "[Test] Step: 75001, acc: 0.82250\n",
      "[Train] step: 75500, loss: 0.25919, acc:0.75000\n",
      "[Train] step: 76000, loss: 0.20025, acc:0.80000\n",
      "[Train] step: 76500, loss: 0.15002, acc:0.85000\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[Train] step: 77000, loss: 0.10010, acc:0.90000\n",
      "[Train] step: 77500, loss: 0.10001, acc:0.90000\n",
      "[Train] step: 78000, loss: 0.20212, acc:0.80000\n",
      "[Train] step: 78500, loss: 0.24868, acc:0.75000\n",
      "[Train] step: 79000, loss: 0.05041, acc:0.95000\n",
      "[Train] step: 79500, loss: 0.10125, acc:0.90000\n",
      "[Train] step: 80000, loss: 0.00001, acc:1.00000\n",
      "(2000, 3072) (2000,)\n",
      "[Test] Step: 80001, acc: 0.82350\n",
      "[Train] step: 80500, loss: 0.20030, acc:0.80000\n",
      "[Train] step: 81000, loss: 0.10175, acc:0.90000\n",
      "[Train] step: 81500, loss: 0.05066, acc:0.95000\n",
      "[Train] step: 82000, loss: 0.15048, acc:0.85000\n",
      "[Train] step: 82500, loss: 0.14875, acc:0.85000\n",
      "[Train] step: 83000, loss: 0.05000, acc:0.95000\n",
      "[Train] step: 83500, loss: 0.10409, acc:0.90000\n",
      "[Train] step: 84000, loss: 0.25022, acc:0.75000\n",
      "[Train] step: 84500, loss: 0.17627, acc:0.80000\n",
      "[Train] step: 85000, loss: 0.05008, acc:0.95000\n",
      "(2000, 3072) (2000,)\n",
      "[Test] Step: 85001, acc: 0.82250\n",
      "[Train] step: 85500, loss: 0.19996, acc:0.80000\n",
      "[Train] step: 86000, loss: 0.10016, acc:0.90000\n",
      "[Train] step: 86500, loss: 0.00006, acc:1.00000\n",
      "[Train] step: 87000, loss: 0.10048, acc:0.90000\n",
      "[Train] step: 87500, loss: 0.25586, acc:0.75000\n",
      "[Train] step: 88000, loss: 0.05095, acc:0.95000\n",
      "[Train] step: 88500, loss: 0.10006, acc:0.90000\n",
      "[Train] step: 89000, loss: 0.05045, acc:0.95000\n",
      "[Train] step: 89500, loss: 0.09997, acc:0.90000\n",
      "[Train] step: 90000, loss: 0.16948, acc:0.80000\n",
      "(2000, 3072) (2000,)\n",
      "[Test] Step: 90001, acc: 0.82450\n",
      "[Train] step: 90500, loss: 0.22538, acc:0.75000\n",
      "[Train] step: 91000, loss: 0.05077, acc:0.95000\n",
      "[Train] step: 91500, loss: 0.22023, acc:0.75000\n",
      "[Train] step: 92000, loss: 0.05003, acc:0.95000\n",
      "[Train] step: 92500, loss: 0.25008, acc:0.75000\n",
      "[Train] step: 93000, loss: 0.04964, acc:0.95000\n",
      "[Train] step: 93500, loss: 0.05000, acc:0.95000\n",
      "[Train] step: 94000, loss: 0.25295, acc:0.75000\n",
      "[Train] step: 94500, loss: 0.15023, acc:0.85000\n",
      "[Train] step: 95000, loss: 0.15002, acc:0.85000\n",
      "(2000, 3072) (2000,)\n",
      "[Test] Step: 95001, acc: 0.82450\n",
      "[Train] step: 95500, loss: 0.10048, acc:0.90000\n",
      "[Train] step: 96000, loss: 0.20008, acc:0.80000\n",
      "[Train] step: 96500, loss: 0.20017, acc:0.80000\n",
      "[Train] step: 97000, loss: 0.10674, acc:0.90000\n",
      "[Train] step: 97500, loss: 0.15378, acc:0.85000\n",
      "[Train] step: 98000, loss: 0.05006, acc:0.95000\n",
      "[Train] step: 98500, loss: 0.15381, acc:0.85000\n",
      "[Train] step: 99000, loss: 0.20159, acc:0.80000\n",
      "[Train] step: 99500, loss: 0.10150, acc:0.90000\n"
     ]
    }
   ],
   "source": [
    "init = tf.global_variables_initializer()\n",
    "batch_size = 20\n",
    "train_steps = 100000\n",
    "test_steps =100\n",
    "\n",
    "with tf.Session() as sess:\n",
    "    sess.run(init)\n",
    "    for i in range(train_steps):\n",
    "        batch_data, batch_labels = train_data.next_batch(batch_size)\n",
    "        loss_val, accu_val, _ = sess.run(\n",
    "            [loss, accuracy, train_op], \n",
    "            feed_dict={\n",
    "                x: batch_data,\n",
    "                y: batch_labels})\n",
    "        if i % 500 == 0:\n",
    "            print('[Train] step: %d, loss: %4.5f, acc:%4.5f'% (i, loss_val, accu_val)) \n",
    "        if i % 5000 == 0:\n",
    "            test_data = CifarData(test_filenames, False)\n",
    "            \n",
    "            all_test_acc_val = []\n",
    "            \n",
    "            for j in range(test_steps):\n",
    "                test_batch_data, test_batch_labels = test_data.next_batch(batch_size)\n",
    "                test_acc_val = sess.run([accuracy],\n",
    "                                       feed_dict={\n",
    "                                           x: test_batch_data,\n",
    "                                           y: test_batch_labels\n",
    "                                       })\n",
    "                all_test_acc_val.append(test_acc_val)\n",
    "            test_acc = np.mean(all_test_acc_val)\n",
    "            print('[Test] Step: %d, acc: %4.5f'%(i+1, test_acc)) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:deep_learning]",
   "language": "python",
   "name": "conda-env-deep_learning-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}