LSTM_data_test.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\ericmlyang\\AppData\\Local\\Continuum\\anaconda3\\envs\\tf_cpu\\lib\\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"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import time \n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from __future__ import print_function\n",
    "\n",
    "import tensorflow as tf"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Read Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_input = pd.read_csv('plc_x_reduce.csv')\n",
    "y_label = pd.read_csv('plc_y_reduce.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "x_input size: (52999, 71)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>X*000000</th>\n",
       "      <th>X*000007</th>\n",
       "      <th>X*000009</th>\n",
       "      <th>X*00000B</th>\n",
       "      <th>X*000010</th>\n",
       "      <th>X*000011</th>\n",
       "      <th>X*000012</th>\n",
       "      <th>X*000013</th>\n",
       "      <th>X*000014</th>\n",
       "      <th>X*000015</th>\n",
       "      <th>...</th>\n",
       "      <th>X*0000A5</th>\n",
       "      <th>X*0000A6</th>\n",
       "      <th>X*0000A7</th>\n",
       "      <th>X*0000A8</th>\n",
       "      <th>X*0000A9</th>\n",
       "      <th>X*0000C4</th>\n",
       "      <th>X*0000C7</th>\n",
       "      <th>X*000110</th>\n",
       "      <th>X*000119</th>\n",
       "      <th>X*00011C</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 71 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   X*000000  X*000007  X*000009  X*00000B  X*000010  X*000011  X*000012  \\\n",
       "0         1         1         1         1         1         1         1   \n",
       "1         1         1         1         1         1         1         1   \n",
       "2         1         1         1         1         1         1         1   \n",
       "3         1         1         1         1         1         1         1   \n",
       "4         1         1         1         1         1         1         1   \n",
       "\n",
       "   X*000013  X*000014  X*000015    ...     X*0000A5  X*0000A6  X*0000A7  \\\n",
       "0         0         1         1    ...            0         0         1   \n",
       "1         0         1         1    ...            0         0         1   \n",
       "2         0         1         1    ...            0         0         0   \n",
       "3         0         1         1    ...            0         0         0   \n",
       "4         0         1         1    ...            0         0         0   \n",
       "\n",
       "   X*0000A8  X*0000A9  X*0000C4  X*0000C7  X*000110  X*000119  X*00011C  \n",
       "0         0         0         1         1         1         1         1  \n",
       "1         0         0         1         1         1         1         1  \n",
       "2         0         0         1         1         1         1         1  \n",
       "3         0         0         1         1         1         1         1  \n",
       "4         0         0         1         1         1         1         1  \n",
       "\n",
       "[5 rows x 71 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(\"x_input size:\", x_input.shape)\n",
    "x_input.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "x_label size: (52999, 41)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Y*000040</th>\n",
       "      <th>Y*000044</th>\n",
       "      <th>Y*000047</th>\n",
       "      <th>Y*00004C</th>\n",
       "      <th>Y*000060</th>\n",
       "      <th>Y*000061</th>\n",
       "      <th>Y*000062</th>\n",
       "      <th>Y*000063</th>\n",
       "      <th>Y*000065</th>\n",
       "      <th>Y*000066</th>\n",
       "      <th>...</th>\n",
       "      <th>Y*0000B2</th>\n",
       "      <th>Y*0000B3</th>\n",
       "      <th>Y*0000B5</th>\n",
       "      <th>Y*0000BC</th>\n",
       "      <th>Y*0000BD</th>\n",
       "      <th>Y*0000DA</th>\n",
       "      <th>Y*0000F6</th>\n",
       "      <th>Y*0000FA</th>\n",
       "      <th>Y*000108</th>\n",
       "      <th>Y*00011A</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 41 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   Y*000040  Y*000044  Y*000047  Y*00004C  Y*000060  Y*000061  Y*000062  \\\n",
       "0         1         0         1         1         0         0         0   \n",
       "1         1         0         1         1         0         1         1   \n",
       "2         1         0         1         1         0         1         1   \n",
       "3         1         0         1         1         0         1         1   \n",
       "4         1         0         1         1         0         1         1   \n",
       "\n",
       "   Y*000063  Y*000065  Y*000066    ...     Y*0000B2  Y*0000B3  Y*0000B5  \\\n",
       "0         0         0         0    ...            0         0         1   \n",
       "1         0         0         0    ...            0         0         1   \n",
       "2         0         0         0    ...            0         0         1   \n",
       "3         0         0         0    ...            0         0         1   \n",
       "4         0         0         0    ...            0         0         1   \n",
       "\n",
       "   Y*0000BC  Y*0000BD  Y*0000DA  Y*0000F6  Y*0000FA  Y*000108  Y*00011A  \n",
       "0         0         1         1         1         1         1         1  \n",
       "1         0         1         1         0         1         1         1  \n",
       "2         0         1         1         0         1         1         1  \n",
       "3         0         1         1         1         1         1         1  \n",
       "4         0         1         1         0         1         1         1  \n",
       "\n",
       "[5 rows x 41 columns]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(\"x_label size:\", y_label.shape)\n",
    "y_label.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def batch_generator(x_input,\n",
    "                    y_label,\n",
    "                    batch_size=5, \n",
    "                    seq_len=10):\n",
    "    \n",
    "    batchs = int(y_label.shape[0] / seq_len / batch_size)\n",
    "    print(batchs)\n",
    "    \n",
    "    input_data = []\n",
    "    target_data = []\n",
    "    \n",
    "    \n",
    "    for i in range(batchs):\n",
    "        x = np.zeros(shape=(batch_size, seq_len, x_input.shape[1]))\n",
    "        y = np.zeros(shape=(batch_size, seq_len, y_label.shape[1]))\n",
    "        for b in range(batch_size):\n",
    "\n",
    "            x[b, :, :] = x_input[b*i : b*i+seq_len, :]\n",
    "            y[b, :, :] = y_label[b*i : b*i+seq_len, :]\n",
    "\n",
    "        input_data.append(x)\n",
    "        target_data.append(y)              \n",
    "    return input_data, target_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "class N2N_LSTM:\n",
    "    def __init__(self,\n",
    "                 seq_max_len=10,  \n",
    "                 input_len=71,\n",
    "                 output_len=41,\n",
    "                 batch_size=5,\n",
    "                 lstm_size=[71, 80, 80, 41],\n",
    "                 num_layers=4,\n",
    "                 learning_rate=0.001,\n",
    "                 grad_clip=2,\n",
    "                 keep_prob=1.,\n",
    "                 ):\n",
    "            \n",
    "        self.seq_max_len = seq_max_len\n",
    "        self.input_len = input_len\n",
    "        self.output_len = output_len\n",
    "        self.batch_size = batch_size\n",
    "        self.lstm_size = lstm_size\n",
    "        self.num_layers = num_layers\n",
    "        self.learning_rate = learning_rate\n",
    "        self.grad_clip = grad_clip\n",
    "        self.train_keep_prob = keep_prob\n",
    "        \n",
    "        tf.reset_default_graph() #Clears the default graph stack and resets the global default graph\n",
    "        self.build_inputs()\n",
    "        self.build_lstm()\n",
    "        self.build_loss()\n",
    "        self.build_optimizer()\n",
    "        self.saver = tf.train.Saver() #Saves and restores variables.\n",
    " \n",
    "\n",
    "    \n",
    "    def build_inputs(self):\n",
    "        with tf.name_scope('inputs'):\n",
    "            self.inputs = tf.placeholder(tf.float32, \n",
    "                                         shape=(self.batch_size, self.seq_max_len, self.input_len),\n",
    "                                         name='inputs')\n",
    "            self.targets = tf.placeholder(tf.float32,\n",
    "                                          shape=(self.batch_size, self.seq_max_len, self.output_len),\n",
    "                                          name='targets')\n",
    "            self.keep_prob = tf.placeholder(tf.float32, name='keep_prob')\n",
    "            \n",
    "            \n",
    "            \n",
    "    def build_lstm(self):\n",
    "        def get_a_cell(lstm_size, keep_prop):\n",
    "            lstm = tf.nn.rnn_cell.BasicLSTMCell(lstm_size)\n",
    "            drop = tf.nn.rnn_cell.DropoutWrapper(lstm, output_keep_prob=self.train_keep_prob)\n",
    "            return drop\n",
    "        \n",
    "        # tf.orthogonal_initializer()\n",
    "        # tf.truncated_normal_initializer()\n",
    "        with tf.variable_scope('lstm', initializer=tf.orthogonal_initializer()):\n",
    "            cell = tf.nn.rnn_cell.MultiRNNCell(\n",
    "                         [get_a_cell(size, self.keep_prob) for size in self.lstm_size]\n",
    "                                              )\n",
    "            self.initial_state = cell.zero_state(self.batch_size, tf.float32)\n",
    "            \n",
    "            # 透過dynamic_rnn對cell展開時間維度\n",
    "            self.lstm_outputs, self.final_state = tf.nn.dynamic_rnn(cell, \n",
    "                                                                    self.inputs, \n",
    "                                                                    initial_state=self.initial_state)\n",
    "            #print(self.lstm_outputs)\n",
    "            \n",
    "            \n",
    "            '''\n",
    "            # 透過lstm_outputs得到機率\n",
    "            seq_output = tf.concat(self.lstm_outputs, 1)\n",
    "            x = tf.reshape(seq_output, [-1, self.lstm_size[-1]])\n",
    "\n",
    "            with tf.variable_scope('sigmoid'):\n",
    "                sigmoid_w = tf.Variable(tf.truncated_normal([self.lstm_size[-1],self.output_len], stddev=0.1))\n",
    "                sigmoid_b = tf.Variable(tf.zeros(self.output_len))\n",
    "\n",
    "            self.logits = tf.matmul(x, sigmoid_w) + sigmoid_b\n",
    "            '''\n",
    "            self.proba_prediction = tf.tanh(self.lstm_outputs, name='predictions')\n",
    "\n",
    "            \n",
    "            \n",
    "            \n",
    "    def build_loss(self):\n",
    "        with tf.name_scope('loss'):\n",
    "            self.y_reshaped = tf.reshape(self.targets, self.lstm_outputs.get_shape())\n",
    "            loss =tf.losses.mean_squared_error(predictions=self.lstm_outputs, labels=self.y_reshaped)\n",
    "            #loss = tf.nn.sigmoid_cross_entropy_with_logits(logits=self.logits, labels=self.y_reshaped)\n",
    "            self.loss = tf.reduce_mean(loss)\n",
    "            \n",
    "            \n",
    "            \n",
    "            \n",
    "    def build_optimizer(self):\n",
    "        # 使用clipping gradients\n",
    "        tvars = tf.trainable_variables()\n",
    "        grads, _ = tf.clip_by_global_norm(tf.gradients(self.loss, tvars), self.grad_clip)\n",
    "        train_op = tf.train.AdamOptimizer(self.learning_rate)\n",
    "        self.optimizer = train_op.apply_gradients(zip(grads, tvars))\n",
    "\n",
    "\n",
    "            \n",
    "\n",
    "    def train(self, x, y, iters=10, save_path='./models', save_every_n=200, log_every_n=200):\n",
    "        self.session = tf.Session()\n",
    "        with self.session as sess:\n",
    "            sess.run(tf.global_variables_initializer())\n",
    "            # Train network\n",
    "            \n",
    "            new_state = sess.run(self.initial_state)\n",
    "            for ite in range(iters):\n",
    "                step = 0\n",
    "                print('iters',ite)\n",
    "                for i in range(len(x)):\n",
    "                    step += 1\n",
    "                    start = time.time()\n",
    "                    feed = {self.inputs: x[i], \n",
    "                            self.targets: y[i],\n",
    "                            self.keep_prob: self.train_keep_prob,\n",
    "                            self.initial_state: new_state}\n",
    "                    batch_loss, new_state, _, pred, target = sess.run([self.loss,\n",
    "                                                                       self.final_state,\n",
    "                                                                       self.optimizer,\n",
    "                                                                       self.proba_prediction,\n",
    "                                                                       self.y_reshaped],\n",
    "                                                                       feed_dict=feed)\n",
    "                    # print result\n",
    "                    #self.print_result(x[i], y[i])\n",
    "                    end = time.time()\n",
    "\n",
    "                    # control the print lines\n",
    "                    if step % log_every_n == 0:\n",
    "                        print(\"=======================================================\\n\")\n",
    "                        print('step: {} in iter: {}/{}... '.format(step, ite+1, iters),\n",
    "                              'loss: {:.4f}... '.format(batch_loss),\n",
    "                              '{:.4f} sec/batch'.format((end - start)))\n",
    "                        \n",
    "                    if (step % save_every_n == 0):\n",
    "                        self.saver.save(sess, os.path.join(save_path, 'model'), global_step=step)\n",
    "                        self.jodge(pred, target)\n",
    "                        #print(\"Target: \\n\", target)\n",
    "                        #print(\"PRED: \\n\",pred)\n",
    " \n",
    "\n",
    "    def jodge(self, pred, target):\n",
    "        pred = np.array(pred)\n",
    "        pred = np.array(pred >= 0.5).astype(int)\n",
    "        \n",
    "        result = np.abs(pred - target)\n",
    "        np.savetxt(\"result.csv\", result, delimiter=',')\n",
    "        return pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1059\n",
      "iters 0\n",
      "=======================================================\n",
      "\n",
      "step: 200 in iter: 1/5...  loss: 0.0827...  0.0040 sec/batch\n",
      "=======================================================\n",
      "\n",
      "step: 400 in iter: 1/5...  loss: 0.0453...  0.0030 sec/batch\n",
      "=======================================================\n",
      "\n",
      "step: 600 in iter: 1/5...  loss: 0.0332...  0.0040 sec/batch\n",
      "=======================================================\n",
      "\n",
      "step: 800 in iter: 1/5...  loss: 0.0367...  0.0030 sec/batch\n",
      "FIRST OUTPUT: \n",
      " [[[0.98700416 0.01051522 0.9810365  ... 0.98107594 0.9860108  0.9818101 ]\n",
      "  [0.9869829  0.01099111 0.9810524  ... 0.98108023 0.9860027  0.98182   ]\n",
      "  [0.98702383 0.01108973 0.9810825  ... 0.98113453 0.986026   0.981858  ]\n",
      "  ...\n",
      "  [0.9870617  0.01105573 0.9811142  ... 0.981161   0.98607296 0.98189074]\n",
      "  [0.98706394 0.01099569 0.98110086 ... 0.98114634 0.98606896 0.9818772 ]\n",
      "  [0.98706424 0.01096025 0.98109365 ... 0.98113656 0.98606616 0.98186994]]\n",
      "\n",
      " [[0.9675998  0.01770059 0.96318066 ... 0.97324985 0.9723766  0.9546489 ]\n",
      "  [0.9675478  0.01879596 0.9630815  ... 0.973305   0.97246253 0.95471895]\n",
      "  [0.96747136 0.01920935 0.9629474  ... 0.9732643  0.97248    0.95470417]\n",
      "  ...\n",
      "  [0.9676652  0.01910154 0.9629559  ... 0.97318465 0.97231907 0.9546717 ]\n",
      "  [0.96768624 0.01906844 0.9629505  ... 0.9731789  0.97231394 0.9546685 ]\n",
      "  [0.9677038  0.01904892 0.9629452  ... 0.97317433 0.97231036 0.9546645 ]]\n",
      "\n",
      " [[0.99065584 0.8446589  0.98952836 ... 0.9869739  0.99249923 0.98996586]\n",
      "  [0.9906879  0.853273   0.98956233 ... 0.98699594 0.9925408  0.99001116]\n",
      "  [0.99072015 0.8597134  0.989595   ... 0.9870208  0.9925682  0.9900422 ]\n",
      "  ...\n",
      "  [0.99081147 0.8761894  0.98968774 ... 0.98710513 0.99264055 0.9901278 ]\n",
      "  [0.9908229  0.8778202  0.9896993  ... 0.98711675 0.9926494  0.990139  ]\n",
      "  [0.9908329  0.87915796 0.9897093  ... 0.9871272  0.9926568  0.99014896]]\n",
      "\n",
      " [[0.9906556  0.8446589  0.98952836 ... 0.9869739  0.99249923 0.98996586]\n",
      "  [0.9906877  0.85327286 0.98956233 ... 0.98699594 0.9925408  0.99001116]\n",
      "  [0.99072015 0.8597134  0.989595   ... 0.9870208  0.9925682  0.99004215]\n",
      "  ...\n",
      "  [0.99081135 0.8761894  0.98968774 ... 0.98710513 0.99264055 0.9901278 ]\n",
      "  [0.9908229  0.8778202  0.9896993  ... 0.98711675 0.9926494  0.990139  ]\n",
      "  [0.9908329  0.87915796 0.9897093  ... 0.9871272  0.9926568  0.99014896]]\n",
      "\n",
      " [[0.9675105  0.02173913 0.9625014  ... 0.97207    0.9735909  0.95420945]\n",
      "  [0.967424   0.02291719 0.9623173  ... 0.97199386 0.97357196 0.9542184 ]\n",
      "  [0.9674225  0.02350502 0.9622484  ... 0.97193444 0.9736077  0.95418173]\n",
      "  ...\n",
      "  [0.9674116  0.02437034 0.96202797 ... 0.9717588  0.9737102  0.9540719 ]\n",
      "  [0.9674102  0.0244355  0.96200174 ... 0.9717384  0.97371954 0.9540559 ]\n",
      "  [0.9674089  0.02448854 0.96197945 ... 0.9717208  0.97372675 0.95404136]]]\n",
      "Target: \n",
      " [[[1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  ...\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]]\n",
      "\n",
      " [[1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  ...\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]]\n",
      "\n",
      " [[1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  ...\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]]\n",
      "\n",
      " [[1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  ...\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]]\n",
      "\n",
      " [[1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  ...\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]]]\n",
      "PRED: \n",
      " [[[0.756082   0.01051483 0.7535143  ... 0.7535313  0.75565624 0.75384843]\n",
      "  [0.7560729  0.01099066 0.75352114 ... 0.7535332  0.7556527  0.7538527 ]\n",
      "  [0.75609046 0.01108928 0.7535342  ... 0.75355667 0.75566274 0.7538691 ]\n",
      "  ...\n",
      "  [0.7561066  0.01105528 0.7535479  ... 0.75356805 0.7556828  0.7538833 ]\n",
      "  [0.75610757 0.01099524 0.75354207 ... 0.75356174 0.75568116 0.7538774 ]\n",
      "  [0.75610775 0.01095981 0.7535389  ... 0.7535575  0.75567997 0.7538743 ]]\n",
      "\n",
      " [[0.7476477  0.01769874 0.7456923  ... 0.750129   0.74974686 0.7418805 ]\n",
      "  [0.7476247  0.01879375 0.74564826 ... 0.7501531  0.7497845  0.74191195]\n",
      "  [0.747591   0.01920699 0.7455888  ... 0.7501353  0.7497921  0.7419054 ]\n",
      "  ...\n",
      "  [0.74767655 0.01909922 0.74559253 ... 0.75010043 0.7497217  0.74189067]\n",
      "  [0.74768585 0.01906613 0.7455901  ... 0.75009793 0.7497193  0.7418893 ]\n",
      "  [0.74769354 0.01904662 0.74558777 ... 0.75009596 0.74971783 0.74188757]]\n",
      "\n",
      " [[0.7576419  0.6882689  0.7571612  ... 0.75606906 0.758426   0.75734776]\n",
      "  [0.7576555  0.69277555 0.7571757  ... 0.7560785  0.75844365 0.75736713]\n",
      "  [0.7576692  0.69611    0.7571896  ... 0.7560891  0.75845534 0.7573803 ]\n",
      "  ...\n",
      "  [0.7577082  0.7045052  0.7572291  ... 0.7561253  0.7584861  0.75741684]\n",
      "  [0.757713   0.7053256  0.7572341  ... 0.7561303  0.7584898  0.7574216 ]\n",
      "  [0.75771725 0.7059973  0.7572384  ... 0.7561347  0.758493   0.75742584]]\n",
      "\n",
      " [[0.7576418  0.6882689  0.7571612  ... 0.75606906 0.758426   0.75734776]\n",
      "  [0.75765544 0.6927754  0.7571757  ... 0.7560785  0.75844365 0.75736713]\n",
      "  [0.7576692  0.69611    0.7571896  ... 0.7560891  0.75845534 0.75738037]\n",
      "  ...\n",
      "  [0.7577081  0.7045052  0.7572291  ... 0.7561253  0.7584861  0.75741684]\n",
      "  [0.757713   0.7053256  0.7572341  ... 0.7561303  0.7584898  0.7574216 ]\n",
      "  [0.75771725 0.7059973  0.7572384  ... 0.7561347  0.758493   0.75742584]]\n",
      "\n",
      " [[0.74760836 0.0217357  0.74539053 ... 0.74961257 0.7502781  0.7416829 ]\n",
      "  [0.74757016 0.02291318 0.74530876 ... 0.7495792  0.7502698  0.7416869 ]\n",
      "  [0.7475695  0.02350069 0.7452781  ... 0.7495532  0.75028545 0.7416704 ]\n",
      "  ...\n",
      "  [0.7475647  0.02436552 0.7451801  ... 0.74947613 0.7503302  0.741621  ]\n",
      "  [0.7475641  0.02443064 0.74516845 ... 0.74946725 0.75033426 0.74161375]\n",
      "  [0.7475635  0.02448364 0.74515855 ... 0.7494595  0.7503375  0.74160725]]]\n",
      "=======================================================\n",
      "\n",
      "step: 1000 in iter: 1/5...  loss: 0.0301...  0.0040 sec/batch\n",
      "iters 1\n",
      "=======================================================\n",
      "\n",
      "step: 200 in iter: 2/5...  loss: 0.0515...  0.0030 sec/batch\n",
      "=======================================================\n",
      "\n",
      "step: 400 in iter: 2/5...  loss: 0.0199...  0.0030 sec/batch\n",
      "=======================================================\n",
      "\n",
      "step: 600 in iter: 2/5...  loss: 0.0127...  0.0050 sec/batch\n",
      "=======================================================\n",
      "\n",
      "step: 800 in iter: 2/5...  loss: 0.0303...  0.0030 sec/batch\n",
      "FIRST OUTPUT: \n",
      " [[[ 0.99851876 -0.00812362  0.9959442  ...  0.99569136  0.99420726\n",
      "    0.99585056]\n",
      "  [ 0.99851876 -0.00841804  0.99595743 ...  0.99572134  0.99423665\n",
      "    0.99587643]\n",
      "  [ 0.9985429  -0.00774365  0.99589765 ...  0.99567044  0.994145\n",
      "    0.99580353]\n",
      "  ...\n",
      "  [ 0.9985557  -0.00736214  0.9958919  ...  0.99565876  0.9941308\n",
      "    0.99579597]\n",
      "  [ 0.9985544  -0.0073142   0.99588567 ...  0.9956508   0.99412304\n",
      "    0.9957904 ]\n",
      "  [ 0.9985543  -0.00729319  0.99588364 ...  0.9956481   0.9941204\n",
      "    0.99578935]]\n",
      "\n",
      " [[ 0.98045087  0.06252036  0.97202444 ...  0.98144156  0.9795194\n",
      "    0.9798712 ]\n",
      "  [ 0.98144567  0.07165478  0.9734239  ...  0.98182076  0.98033386\n",
      "    0.9810534 ]\n",
      "  [ 0.978809    0.06872606  0.97076    ...  0.98081785  0.9792753\n",
      "    0.9791403 ]\n",
      "  ...\n",
      "  [ 0.97685355  0.05441887  0.96831244 ...  0.9800439   0.97793496\n",
      "    0.97679216]\n",
      "  [ 0.9767619   0.05335523  0.96817565 ...  0.97997385  0.97787076\n",
      "    0.97660017]\n",
      "  [ 0.97669214  0.05265709  0.968074   ...  0.97991663  0.9778256\n",
      "    0.97645605]]\n",
      "\n",
      " [[ 0.99694544  0.9051012   0.99490744 ...  0.9944601   0.99648935\n",
      "    0.99666095]\n",
      "  [ 0.99695563  0.9110726   0.9949208  ...  0.99446917  0.99650854\n",
      "    0.9966653 ]\n",
      "  [ 0.99696416  0.9155761   0.9949352  ...  0.9944763   0.9965178\n",
      "    0.9966685 ]\n",
      "  ...\n",
      "  [ 0.99698013  0.9262201   0.9949663  ...  0.99449116  0.99653035\n",
      "    0.99667406]\n",
      "  [ 0.99698144  0.9271314   0.9949687  ...  0.9944928   0.99653107\n",
      "    0.99667466]\n",
      "  [ 0.9969824   0.92784053  0.9949705  ...  0.99449414  0.99653167\n",
      "    0.99667513]]\n",
      "\n",
      " [[ 0.99694544  0.9051012   0.99490744 ...  0.9944601   0.99648935\n",
      "    0.99666095]\n",
      "  [ 0.99695563  0.9110726   0.9949208  ...  0.99446917  0.99650854\n",
      "    0.9966653 ]\n",
      "  [ 0.99696416  0.9155761   0.9949352  ...  0.9944763   0.9965178\n",
      "    0.9966685 ]\n",
      "  ...\n",
      "  [ 0.99698013  0.9262201   0.9949663  ...  0.99449116  0.99653035\n",
      "    0.99667406]\n",
      "  [ 0.99698144  0.9271314   0.9949687  ...  0.9944928   0.99653107\n",
      "    0.99667466]\n",
      "  [ 0.9969824   0.92784053  0.9949705  ...  0.99449414  0.99653167\n",
      "    0.99667513]]\n",
      "\n",
      " [[ 0.9904156  -0.00475996  0.9878445  ...  0.9893855   0.9912254\n",
      "    0.99137384]\n",
      "  [ 0.99052376 -0.0037423   0.9880386  ...  0.98946273  0.9913694\n",
      "    0.991471  ]\n",
      "  [ 0.99056596 -0.00295067  0.98814034 ...  0.9894979   0.99146456\n",
      "    0.9915309 ]\n",
      "  ...\n",
      "  [ 0.99059194  0.00126363  0.98830855 ...  0.98947597  0.9916141\n",
      "    0.9915623 ]\n",
      "  [ 0.9906024   0.00199922  0.9883322  ...  0.98946965  0.99162453\n",
      "    0.99155676]\n",
      "  [ 0.99061334  0.00265389  0.9883531  ...  0.9894645   0.99163276\n",
      "    0.9915514 ]]]\n",
      "Target: \n",
      " [[[1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  ...\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]]\n",
      "\n",
      " [[1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  ...\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]]\n",
      "\n",
      " [[1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  ...\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]]\n",
      "\n",
      " [[1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  ...\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]]\n",
      "\n",
      " [[1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  ...\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]]]\n",
      "PRED: \n",
      " [[[ 0.76097137 -0.00812344  0.7598856  ...  0.7597787   0.7591507\n",
      "    0.759846  ]\n",
      "  [ 0.76097137 -0.00841784  0.7598912  ...  0.7597915   0.759163\n",
      "    0.75985694]\n",
      "  [ 0.7609815  -0.00774349  0.7598659  ...  0.7597699   0.7591242\n",
      "    0.7598261 ]\n",
      "  ...\n",
      "  [ 0.760987   -0.00736201  0.75986356 ...  0.75976497  0.75911826\n",
      "    0.75982285]\n",
      "  [ 0.7609863  -0.00731407  0.75986093 ...  0.7597617   0.759115\n",
      "    0.7598206 ]\n",
      "  [ 0.7609864  -0.00729306  0.75986004 ...  0.7597604   0.75911385\n",
      "    0.75982016]]\n",
      "\n",
      " [[ 0.7532611   0.06243902  0.7495926  ...  0.7536892   0.7528578\n",
      "    0.7530102 ]\n",
      "  [ 0.7536911   0.07153238  0.75020504 ...  0.75385314  0.7532105\n",
      "    0.7535216 ]\n",
      "  [ 0.7525499   0.06861806  0.74903816 ...  0.7534198   0.752752\n",
      "    0.75269353]\n",
      "  ...\n",
      "  [ 0.75170064  0.0543652   0.7479618  ...  0.753085    0.75217056\n",
      "    0.75167394]\n",
      "  [ 0.75166076  0.05330466  0.7479015  ...  0.75305456  0.7521427\n",
      "    0.75159043]\n",
      "  [ 0.7516305   0.05260847  0.7478568  ...  0.7530298   0.7521231\n",
      "    0.75152767]]\n",
      "\n",
      " [[ 0.76030844  0.7187727   0.7594471  ...  0.75925773  0.76011586\n",
      "    0.76018834]\n",
      "  [ 0.7603126   0.7216467   0.7594528  ...  0.75926155  0.76012397\n",
      "    0.7601901 ]\n",
      "  [ 0.76031625  0.72379786  0.7594589  ...  0.75926465  0.76012784\n",
      "    0.7601915 ]\n",
      "  ...\n",
      "  [ 0.760323    0.72882676  0.7594721  ...  0.7592709   0.76013315\n",
      "    0.7601938 ]\n",
      "  [ 0.7603235   0.7292537   0.759473   ...  0.75927156  0.7601335\n",
      "    0.76019406]\n",
      "  [ 0.76032394  0.7295855   0.7594739  ...  0.75927216  0.76013374\n",
      "    0.76019436]]\n",
      "\n",
      " [[ 0.76030844  0.7187727   0.7594471  ...  0.75925773  0.76011586\n",
      "    0.76018834]\n",
      "  [ 0.7603126   0.7216467   0.7594528  ...  0.75926155  0.76012397\n",
      "    0.7601901 ]\n",
      "  [ 0.76031625  0.72379786  0.7594589  ...  0.75926465  0.76012784\n",
      "    0.7601915 ]\n",
      "  ...\n",
      "  [ 0.760323    0.72882676  0.7594721  ...  0.7592709   0.76013315\n",
      "    0.7601938 ]\n",
      "  [ 0.7603235   0.7292537   0.759473   ...  0.75927156  0.7601335\n",
      "    0.76019406]\n",
      "  [ 0.76032394  0.7295855   0.7594739  ...  0.75927216  0.76013374\n",
      "    0.76019436]]\n",
      "\n",
      " [[ 0.7575395  -0.00475992  0.7564418  ...  0.7571002   0.75788444\n",
      "    0.7579476 ]\n",
      "  [ 0.75758564 -0.00374228  0.7565248  ...  0.7571332   0.7579456\n",
      "    0.75798887]\n",
      "  [ 0.75760365 -0.00295066  0.75656825 ...  0.75714815  0.7579862\n",
      "    0.7580144 ]\n",
      "  ...\n",
      "  [ 0.7576146   0.00126362  0.75664014 ...  0.7571388   0.7580498\n",
      "    0.75802773]\n",
      "  [ 0.75761914  0.00199921  0.7566504  ...  0.75713617  0.7580542\n",
      "    0.75802535]\n",
      "  [ 0.7576238   0.00265389  0.75665927 ...  0.7571339   0.7580578\n",
      "    0.75802314]]]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "=======================================================\n",
      "\n",
      "step: 1000 in iter: 2/5...  loss: 0.0195...  0.0030 sec/batch\n",
      "iters 2\n",
      "=======================================================\n",
      "\n",
      "step: 200 in iter: 3/5...  loss: 0.0216...  0.0040 sec/batch\n",
      "=======================================================\n",
      "\n",
      "step: 400 in iter: 3/5...  loss: 0.0120...  0.0060 sec/batch\n",
      "=======================================================\n",
      "\n",
      "step: 600 in iter: 3/5...  loss: 0.0103...  0.0030 sec/batch\n",
      "=======================================================\n",
      "\n",
      "step: 800 in iter: 3/5...  loss: 0.0110...  0.0030 sec/batch\n",
      "FIRST OUTPUT: \n",
      " [[[ 0.9959428  -0.01735546  0.99606234 ...  0.99164563  0.99524224\n",
      "    0.9951197 ]\n",
      "  [ 0.9971624  -0.01072815  0.994728   ...  0.9914172   0.9934258\n",
      "    0.9946227 ]\n",
      "  [ 0.99666756 -0.00469346  0.992444   ...  0.9899859   0.9918177\n",
      "    0.9923976 ]\n",
      "  ...\n",
      "  [ 0.9966535  -0.00468474  0.9927226  ...  0.9899402   0.9920386\n",
      "    0.99219996]\n",
      "  [ 0.9966383  -0.00466647  0.99276286 ...  0.98996186  0.9921176\n",
      "    0.99219793]\n",
      "  [ 0.9966439  -0.00461177  0.9927787  ...  0.9899715   0.9921428\n",
      "    0.9922057 ]]\n",
      "\n",
      " [[ 0.9977736   0.02172195  0.994823   ...  0.9949844   0.9948934\n",
      "    0.9960802 ]\n",
      "  [ 0.99757606  0.03455879  0.9953428  ...  0.9952296   0.9948787\n",
      "    0.9965495 ]\n",
      "  [ 0.99287236  0.02221965  0.98742175 ...  0.9910691   0.9922418\n",
      "    0.99129444]\n",
      "  ...\n",
      "  [ 0.9938338   0.0045665   0.9861224  ...  0.99085253  0.9921351\n",
      "    0.990298  ]\n",
      "  [ 0.9938595   0.00428423  0.9861638  ...  0.99081016  0.99218947\n",
      "    0.9902553 ]\n",
      "  [ 0.9938744   0.00413941  0.9861944  ...  0.9907828   0.99222726\n",
      "    0.99023145]]\n",
      "\n",
      " [[ 0.99823135  0.98980373  0.99694186 ...  0.9963638   0.9980799\n",
      "    0.9975751 ]\n",
      "  [ 0.9982299   0.98979205  0.9969373  ...  0.9963626   0.99807715\n",
      "    0.9975725 ]\n",
      "  [ 0.99822885  0.98979044  0.9969356  ...  0.9963625   0.99807584\n",
      "    0.99757177]\n",
      "  ...\n",
      "  [ 0.99822587  0.9897792   0.99692947 ...  0.9963619   0.99807173\n",
      "    0.9975696 ]\n",
      "  [ 0.9982256   0.98977846  0.996929   ...  0.9963618   0.99807125\n",
      "    0.9975694 ]\n",
      "  [ 0.9982254   0.98977774  0.9969285  ...  0.9963618   0.998071\n",
      "    0.99756926]]\n",
      "\n",
      " [[ 0.99823135  0.9898038   0.99694186 ...  0.9963638   0.9980799\n",
      "    0.9975751 ]\n",
      "  [ 0.9982299   0.98979205  0.9969373  ...  0.9963626   0.99807715\n",
      "    0.9975725 ]\n",
      "  [ 0.99822885  0.98979014  0.9969356  ...  0.9963625   0.99807584\n",
      "    0.99757177]\n",
      "  ...\n",
      "  [ 0.99822587  0.9897792   0.99692947 ...  0.9963619   0.99807173\n",
      "    0.9975696 ]\n",
      "  [ 0.9982256   0.98977846  0.996929   ...  0.9963618   0.99807125\n",
      "    0.9975694 ]\n",
      "  [ 0.9982254   0.98977774  0.9969285  ...  0.9963618   0.998071\n",
      "    0.99756926]]\n",
      "\n",
      " [[ 0.9954967   0.8801962   0.9920658  ...  0.99174273  0.9957406\n",
      "    0.9921772 ]\n",
      "  [ 0.9955128   0.8840707   0.99211097 ...  0.991787    0.9957525\n",
      "    0.9921888 ]\n",
      "  [ 0.9955176   0.8861669   0.992125   ...  0.99180543  0.9957557\n",
      "    0.9921904 ]\n",
      "  ...\n",
      "  [ 0.99552184  0.88916475  0.9921441  ...  0.9918291   0.99576277\n",
      "    0.99218994]\n",
      "  [ 0.9955217   0.88928366  0.99214506 ...  0.9918298   0.9957633\n",
      "    0.99219006]\n",
      "  [ 0.9955219   0.88935965  0.99214596 ...  0.99183017  0.9957638\n",
      "    0.9921902 ]]]\n",
      "Target: \n",
      " [[[1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  ...\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]]\n",
      "\n",
      " [[1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  ...\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]]\n",
      "\n",
      " [[1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  ...\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]]\n",
      "\n",
      " [[1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  ...\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]]\n",
      "\n",
      " [[1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  ...\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]]]\n",
      "PRED: \n",
      " [[[ 0.759885   -0.01735372  0.75993556 ...  0.7580632   0.7595888\n",
      "    0.759537  ]\n",
      "  [ 0.7603999  -0.01072774  0.75937116 ...  0.757966    0.75881934\n",
      "    0.75932664]\n",
      "  [ 0.76019114 -0.00469342  0.7584025  ...  0.75735635  0.7581364\n",
      "    0.75838286]\n",
      "  ...\n",
      "  [ 0.76018524 -0.0046847   0.75852084 ...  0.75733685  0.7582302\n",
      "    0.7582989 ]\n",
      "  [ 0.7601787  -0.00466643  0.75853795 ...  0.7573461   0.75826377\n",
      "    0.75829804]\n",
      "  [ 0.760181   -0.00461174  0.7585447  ...  0.75735027  0.7582745\n",
      "    0.7583013 ]]\n",
      "\n",
      " [[ 0.7606576   0.02171854  0.7594114  ...  0.7594797   0.7594412\n",
      "    0.75994307]\n",
      "  [ 0.7605743   0.03454503  0.7596314  ...  0.7595835   0.759435\n",
      "    0.76014125]\n",
      "  [ 0.7585845   0.022216    0.75626093 ...  0.7578179   0.7583167\n",
      "    0.7579137 ]\n",
      "  ...\n",
      "  [ 0.7589923   0.00456646  0.7557041  ...  0.7577257   0.7582713\n",
      "    0.7574894 ]\n",
      "  [ 0.7590033   0.0042842   0.7557218  ...  0.7577076   0.7582944\n",
      "    0.7574712 ]\n",
      "  [ 0.7590096   0.00413938  0.7557349  ...  0.75769603  0.75831044\n",
      "    0.7574611 ]]\n",
      "\n",
      " [[ 0.7608504   0.7572787   0.7603069  ...  0.7600628   0.7607866\n",
      "    0.7605739 ]\n",
      "  [ 0.76084983  0.7572736   0.7603049  ...  0.76006234  0.76078546\n",
      "    0.76057285]\n",
      "  [ 0.76084936  0.7572729   0.7603042  ...  0.7600623   0.7607849\n",
      "    0.7605725 ]\n",
      "  ...\n",
      "  [ 0.76084805  0.7572682   0.7603016  ...  0.76006204  0.7607832\n",
      "    0.7605716 ]\n",
      "  [ 0.76084805  0.7572678   0.76030135 ...  0.760062    0.76078296\n",
      "    0.76057154]\n",
      "  [ 0.7608479   0.75726753  0.76030123 ...  0.760062    0.7607829\n",
      "    0.76057154]]\n",
      "\n",
      " [[ 0.7608504   0.7572787   0.7603069  ...  0.7600628   0.7607866\n",
      "    0.7605739 ]\n",
      "  [ 0.76084983  0.7572736   0.7603049  ...  0.76006234  0.76078546\n",
      "    0.76057285]\n",
      "  [ 0.76084936  0.75727284  0.7603042  ...  0.7600623   0.7607849\n",
      "    0.7605725 ]\n",
      "  ...\n",
      "  [ 0.76084805  0.7572682   0.7603016  ...  0.76006204  0.7607832\n",
      "    0.7605716 ]\n",
      "  [ 0.76084805  0.7572678   0.76030135 ...  0.760062    0.76078296\n",
      "    0.76057154]\n",
      "  [ 0.7608479   0.75726753  0.76030123 ...  0.760062    0.7607829\n",
      "    0.76057154]]\n",
      "\n",
      " [[ 0.7596964   0.7065176   0.75824183 ...  0.75810444  0.75979954\n",
      "    0.75828916]\n",
      "  [ 0.75970316  0.7084528   0.7582611  ...  0.7581234   0.75980455\n",
      "    0.7582941 ]\n",
      "  [ 0.75970525  0.70949537  0.758267   ...  0.75813115  0.7598059\n",
      "    0.75829476]\n",
      "  ...\n",
      "  [ 0.7597071   0.7109809   0.75827515 ...  0.7581412   0.7598089\n",
      "    0.75829464]\n",
      "  [ 0.759707    0.7110398   0.75827557 ...  0.7581415   0.75980914\n",
      "    0.75829464]\n",
      "  [ 0.7597071   0.71107733  0.75827587 ...  0.7581417   0.7598093\n",
      "    0.75829476]]]\n",
      "=======================================================\n",
      "\n",
      "step: 1000 in iter: 3/5...  loss: 0.0124...  0.0030 sec/batch\n",
      "iters 3\n",
      "=======================================================\n",
      "\n",
      "step: 200 in iter: 4/5...  loss: 0.0167...  0.0040 sec/batch\n",
      "=======================================================\n",
      "\n",
      "step: 400 in iter: 4/5...  loss: 0.0092...  0.0030 sec/batch\n",
      "=======================================================\n",
      "\n",
      "step: 600 in iter: 4/5...  loss: 0.0093...  0.0040 sec/batch\n",
      "=======================================================\n",
      "\n",
      "step: 800 in iter: 4/5...  loss: 0.0100...  0.0030 sec/batch\n",
      "FIRST OUTPUT: \n",
      " [[[ 9.96570587e-01  6.12869812e-03  9.97576654e-01 ...  9.95207846e-01\n",
      "    9.97431457e-01  9.96872842e-01]\n",
      "  [ 9.97900367e-01  1.01551395e-02  9.97339547e-01 ...  9.94771898e-01\n",
      "    9.96845663e-01  9.97282624e-01]\n",
      "  [ 9.98171210e-01 -4.35459800e-03  9.96114612e-01 ...  9.94874775e-01\n",
      "    9.96243954e-01  9.96656775e-01]\n",
      "  ...\n",
      "  [ 9.94856715e-01 -1.20348632e-02  9.93188024e-01 ...  9.90994036e-01\n",
      "    9.93645310e-01  9.94150460e-01]\n",
      "  [ 9.98296440e-01 -4.59460868e-03  9.96403098e-01 ...  9.95480418e-01\n",
      "    9.96490777e-01  9.96560752e-01]\n",
      "  [ 9.94255126e-01 -1.09024188e-02  9.92562354e-01 ...  9.90965724e-01\n",
      "    9.93599415e-01  9.93782341e-01]]\n",
      "\n",
      " [[ 9.98231590e-01  7.12692132e-03  9.95818436e-01 ...  9.95603800e-01\n",
      "    9.96623218e-01  9.97103035e-01]\n",
      "  [ 9.98285711e-01  1.36377979e-02  9.96379793e-01 ...  9.95978773e-01\n",
      "    9.96449769e-01  9.97688413e-01]\n",
      "  [ 9.91796494e-01  5.07048145e-03  9.82816279e-01 ...  9.89802599e-01\n",
      "    9.92278516e-01  9.90714788e-01]\n",
      "  ...\n",
      "  [ 9.94274914e-01  7.29741121e-04  9.86443758e-01 ...  9.90725040e-01\n",
      "    9.94676054e-01  9.92516339e-01]\n",
      "  [ 9.94163573e-01  5.40204288e-04  9.86286461e-01 ...  9.90416169e-01\n",
      "    9.94672894e-01  9.92406070e-01]\n",
      "  [ 9.94010806e-01  4.19389515e-04  9.85937178e-01 ...  9.89995360e-01\n",
      "    9.94575381e-01  9.92194414e-01]]\n",
      "\n",
      " [[ 9.98538613e-01  9.91514623e-01  9.98112679e-01 ...  9.97224212e-01\n",
      "    9.98502135e-01  9.98467267e-01]\n",
      "  [ 9.98538256e-01  9.91534591e-01  9.98111844e-01 ...  9.97223377e-01\n",
      "    9.98501778e-01  9.98467267e-01]\n",
      "  [ 9.98538375e-01  9.91536200e-01  9.98111606e-01 ...  9.97222900e-01\n",
      "    9.98501539e-01  9.98467386e-01]\n",
      "  ...\n",
      "  [ 9.98538375e-01  9.91538107e-01  9.98111606e-01 ...  9.97222900e-01\n",
      "    9.98501539e-01  9.98467505e-01]\n",
      "  [ 9.98538375e-01  9.91538286e-01  9.98111606e-01 ...  9.97222900e-01\n",
      "    9.98501539e-01  9.98467505e-01]\n",
      "  [ 9.98538375e-01  9.91538346e-01  9.98111606e-01 ...  9.97222900e-01\n",
      "    9.98501539e-01  9.98467505e-01]]\n",
      "\n",
      " [[ 9.98538613e-01  9.91514683e-01  9.98112679e-01 ...  9.97224212e-01\n",
      "    9.98502135e-01  9.98467267e-01]\n",
      "  [ 9.98538256e-01  9.91534591e-01  9.98111844e-01 ...  9.97223377e-01\n",
      "    9.98501778e-01  9.98467267e-01]\n",
      "  [ 9.98538375e-01  9.91536200e-01  9.98111606e-01 ...  9.97222900e-01\n",
      "    9.98501539e-01  9.98467386e-01]\n",
      "  ...\n",
      "  [ 9.98538375e-01  9.91538227e-01  9.98111606e-01 ...  9.97222900e-01\n",
      "    9.98501539e-01  9.98467505e-01]\n",
      "  [ 9.98538375e-01  9.91538227e-01  9.98111606e-01 ...  9.97222900e-01\n",
      "    9.98501539e-01  9.98467505e-01]\n",
      "  [ 9.98538375e-01  9.91538346e-01  9.98111606e-01 ...  9.97222900e-01\n",
      "    9.98501539e-01  9.98467505e-01]]\n",
      "\n",
      " [[ 9.96020257e-01  9.58682418e-01  9.94109213e-01 ...  9.93837416e-01\n",
      "    9.95491683e-01  9.94180858e-01]\n",
      "  [ 9.96019423e-01  9.58911240e-01  9.94112074e-01 ...  9.93831038e-01\n",
      "    9.95487928e-01  9.94174778e-01]\n",
      "  [ 9.96017516e-01  9.59065199e-01  9.94112313e-01 ...  9.93826091e-01\n",
      "    9.95485902e-01  9.94172752e-01]\n",
      "  ...\n",
      "  [ 9.96014118e-01  9.59411502e-01  9.94117022e-01 ...  9.93818939e-01\n",
      "    9.95485961e-01  9.94172990e-01]\n",
      "  [ 9.96013761e-01  9.59436059e-01  9.94117737e-01 ...  9.93818820e-01\n",
      "    9.95486319e-01  9.94173467e-01]\n",
      "  [ 9.96013522e-01  9.59454238e-01  9.94118333e-01 ...  9.93818820e-01\n",
      "    9.95486915e-01  9.94174063e-01]]]\n",
      "Target: \n",
      " [[[1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  ...\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]]\n",
      "\n",
      " [[1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  ...\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]]\n",
      "\n",
      " [[1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  ...\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]]\n",
      "\n",
      " [[1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  ...\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]]\n",
      "\n",
      " [[1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  ...\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]]]\n",
      "PRED: \n",
      " [[[ 7.6015019e-01  6.1286208e-03  7.6057452e-01 ...  7.5957417e-01\n",
      "    7.6051331e-01  7.6027775e-01]\n",
      "  [ 7.6071095e-01  1.0154790e-02  7.6047462e-01 ...  7.5938976e-01\n",
      "    7.6026624e-01  7.6045054e-01]\n",
      "  [ 7.6082504e-01 -4.3545701e-03  7.5995761e-01 ...  7.5943333e-01\n",
      "    7.6001221e-01  7.6018655e-01]\n",
      "  ...\n",
      "  [ 7.5942564e-01 -1.2034280e-02  7.5871843e-01 ...  7.5778592e-01\n",
      "    7.5891244e-01  7.5912660e-01]\n",
      "  [ 7.6087779e-01 -4.5945761e-03  7.6007938e-01 ...  7.5968951e-01\n",
      "    7.6011646e-01  7.6014602e-01]\n",
      "  [ 7.5917095e-01 -1.0901987e-02  7.5845283e-01 ...  7.5777388e-01\n",
      "    7.5889295e-01  7.5897056e-01]]\n",
      "\n",
      " [[ 7.6085049e-01  7.1268007e-03  7.5983250e-01 ...  7.5974172e-01\n",
      "    7.6017237e-01  7.6037484e-01]\n",
      "  [ 7.6087326e-01  1.3636951e-02  7.6006949e-01 ...  7.5990015e-01\n",
      "    7.6009911e-01  7.6062167e-01]\n",
      "  [ 7.5812733e-01  5.0704377e-03  7.5428247e-01 ...  7.5727814e-01\n",
      "    7.5833231e-01  7.5766706e-01]\n",
      "  ...\n",
      "  [ 7.5917929e-01  7.2974089e-04  7.5584185e-01 ...  7.5767136e-01\n",
      "    7.5934923e-01  7.5843334e-01]\n",
      "  [ 7.5913215e-01  5.4020417e-04  7.5577444e-01 ...  7.5753975e-01\n",
      "    7.5934786e-01  7.5838643e-01]\n",
      "  [ 7.5906736e-01  4.1938946e-04  7.5562459e-01 ...  7.5736040e-01\n",
      "    7.5930661e-01  7.5829655e-01]]\n",
      "\n",
      " [[ 7.6097977e-01  7.5800747e-01  7.6080048e-01 ...  7.6042598e-01\n",
      "    7.6096445e-01  7.6094973e-01]\n",
      "  [ 7.6097959e-01  7.5801593e-01  7.6080000e-01 ...  7.6042557e-01\n",
      "    7.6096427e-01  7.6094973e-01]\n",
      "  [ 7.6097965e-01  7.5801665e-01  7.6080006e-01 ...  7.6042545e-01\n",
      "    7.6096416e-01  7.6094979e-01]\n",
      "  ...\n",
      "  [ 7.6097965e-01  7.5801742e-01  7.6080006e-01 ...  7.6042545e-01\n",
      "    7.6096416e-01  7.6094985e-01]\n",
      "  [ 7.6097965e-01  7.5801760e-01  7.6080006e-01 ...  7.6042545e-01\n",
      "    7.6096416e-01  7.6094985e-01]\n",
      "  [ 7.6097965e-01  7.5801754e-01  7.6080006e-01 ...  7.6042545e-01\n",
      "    7.6096416e-01  7.6094985e-01]]\n",
      "\n",
      " [[ 7.6097977e-01  7.5800741e-01  7.6080048e-01 ...  7.6042598e-01\n",
      "    7.6096445e-01  7.6094973e-01]\n",
      "  [ 7.6097959e-01  7.5801593e-01  7.6080000e-01 ...  7.6042557e-01\n",
      "    7.6096427e-01  7.6094973e-01]\n",
      "  [ 7.6097965e-01  7.5801665e-01  7.6080006e-01 ...  7.6042545e-01\n",
      "    7.6096416e-01  7.6094979e-01]\n",
      "  ...\n",
      "  [ 7.6097965e-01  7.5801754e-01  7.6080006e-01 ...  7.6042545e-01\n",
      "    7.6096416e-01  7.6094985e-01]\n",
      "  [ 7.6097965e-01  7.5801754e-01  7.6080006e-01 ...  7.6042545e-01\n",
      "    7.6096416e-01  7.6094985e-01]\n",
      "  [ 7.6097965e-01  7.5801754e-01  7.6080006e-01 ...  7.6042545e-01\n",
      "    7.6096416e-01  7.6094985e-01]]\n",
      "\n",
      " [[ 7.5991774e-01  7.4368864e-01  7.5910902e-01 ...  7.5899386e-01\n",
      "    7.5969428e-01  7.5913942e-01]\n",
      "  [ 7.5991732e-01  7.4379086e-01  7.5911027e-01 ...  7.5899124e-01\n",
      "    7.5969267e-01  7.5913692e-01]\n",
      "  [ 7.5991660e-01  7.4385965e-01  7.5911039e-01 ...  7.5898910e-01\n",
      "    7.5969189e-01  7.5913602e-01]\n",
      "  ...\n",
      "  [ 7.5991511e-01  7.4401432e-01  7.5911242e-01 ...  7.5898606e-01\n",
      "    7.5969183e-01  7.5913608e-01]\n",
      "  [ 7.5991493e-01  7.4402523e-01  7.5911272e-01 ...  7.5898606e-01\n",
      "    7.5969207e-01  7.5913626e-01]\n",
      "  [ 7.5991488e-01  7.4403334e-01  7.5911295e-01 ...  7.5898606e-01\n",
      "    7.5969231e-01  7.5913650e-01]]]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "=======================================================\n",
      "\n",
      "step: 1000 in iter: 4/5...  loss: 0.0093...  0.0030 sec/batch\n",
      "iters 4\n",
      "=======================================================\n",
      "\n",
      "step: 200 in iter: 5/5...  loss: 0.0137...  0.0030 sec/batch\n",
      "=======================================================\n",
      "\n",
      "step: 400 in iter: 5/5...  loss: 0.0075...  0.0030 sec/batch\n",
      "=======================================================\n",
      "\n",
      "step: 600 in iter: 5/5...  loss: 0.0077...  0.0040 sec/batch\n",
      "=======================================================\n",
      "\n",
      "step: 800 in iter: 5/5...  loss: 0.0087...  0.0030 sec/batch\n",
      "FIRST OUTPUT: \n",
      " [[[ 9.9691486e-01 -1.4468502e-03  9.9820745e-01 ...  9.9645954e-01\n",
      "    9.9844992e-01  9.9760443e-01]\n",
      "  [ 9.9881113e-01 -1.9602948e-03  9.9851936e-01 ...  9.9656677e-01\n",
      "    9.9810261e-01  9.9842346e-01]\n",
      "  [ 9.9896467e-01 -1.2136683e-03  9.9803573e-01 ...  9.9702436e-01\n",
      "    9.9793482e-01  9.9836212e-01]\n",
      "  ...\n",
      "  [ 9.9774051e-01 -1.8705669e-03  9.9711943e-01 ...  9.9528080e-01\n",
      "    9.9674976e-01  9.9727374e-01]\n",
      "  [ 9.9901748e-01 -2.4772869e-04  9.9808794e-01 ...  9.9761188e-01\n",
      "    9.9808085e-01  9.9840337e-01]\n",
      "  [ 9.9257916e-01 -1.4439883e-03  9.9267173e-01 ...  9.9215651e-01\n",
      "    9.9447906e-01  9.9469453e-01]]\n",
      "\n",
      " [[ 9.9911100e-01  6.3529373e-03  9.9817681e-01 ...  9.9780041e-01\n",
      "    9.9828440e-01  9.9848276e-01]\n",
      "  [ 9.9896562e-01  1.3422856e-02  9.9813265e-01 ...  9.9747974e-01\n",
      "    9.9804056e-01  9.9865735e-01]\n",
      "  [ 9.9192697e-01  1.0380424e-02  9.8510206e-01 ...  9.9080324e-01\n",
      "    9.9399334e-01  9.9116242e-01]\n",
      "  ...\n",
      "  [ 9.9617308e-01  1.9955363e-03  9.9197155e-01 ...  9.9342537e-01\n",
      "    9.9666536e-01  9.9383450e-01]\n",
      "  [ 9.9618340e-01  1.9416398e-03  9.9199551e-01 ...  9.9342477e-01\n",
      "    9.9668425e-01  9.9383003e-01]\n",
      "  [ 9.9618626e-01  1.9185315e-03  9.9200207e-01 ...  9.9342513e-01\n",
      "    9.9669176e-01  9.9382460e-01]]\n",
      "\n",
      " [[ 9.9895859e-01  9.9064749e-01  9.9857950e-01 ...  9.9844897e-01\n",
      "    9.9897254e-01  9.9891245e-01]\n",
      "  [ 9.9896169e-01  9.9065590e-01  9.9858201e-01 ...  9.9845183e-01\n",
      "    9.9897361e-01  9.9891424e-01]\n",
      "  [ 9.9896240e-01  9.9065876e-01  9.9858272e-01 ...  9.9845266e-01\n",
      "    9.9897420e-01  9.9891496e-01]\n",
      "  ...\n",
      "  [ 9.9896312e-01  9.9066371e-01  9.9858332e-01 ...  9.9845362e-01\n",
      "    9.9897480e-01  9.9891567e-01]\n",
      "  [ 9.9896312e-01  9.9066406e-01  9.9858332e-01 ...  9.9845362e-01\n",
      "    9.9897480e-01  9.9891567e-01]\n",
      "  [ 9.9896312e-01  9.9066454e-01  9.9858332e-01 ...  9.9845362e-01\n",
      "    9.9897480e-01  9.9891567e-01]]\n",
      "\n",
      " [[ 9.9895859e-01  9.9064749e-01  9.9857950e-01 ...  9.9844897e-01\n",
      "    9.9897254e-01  9.9891245e-01]\n",
      "  [ 9.9896169e-01  9.9065590e-01  9.9858201e-01 ...  9.9845183e-01\n",
      "    9.9897361e-01  9.9891424e-01]\n",
      "  [ 9.9896240e-01  9.9065870e-01  9.9858272e-01 ...  9.9845266e-01\n",
      "    9.9897420e-01  9.9891496e-01]\n",
      "  ...\n",
      "  [ 9.9896312e-01  9.9066371e-01  9.9858332e-01 ...  9.9845362e-01\n",
      "    9.9897480e-01  9.9891567e-01]\n",
      "  [ 9.9896312e-01  9.9066406e-01  9.9858332e-01 ...  9.9845362e-01\n",
      "    9.9897480e-01  9.9891567e-01]\n",
      "  [ 9.9896312e-01  9.9066454e-01  9.9858332e-01 ...  9.9845362e-01\n",
      "    9.9897480e-01  9.9891567e-01]]\n",
      "\n",
      " [[ 9.9733883e-01  9.6841371e-01  9.9693668e-01 ...  9.9655342e-01\n",
      "    9.9757820e-01  9.9573153e-01]\n",
      "  [ 9.9733943e-01  9.6878469e-01  9.9693900e-01 ...  9.9655271e-01\n",
      "    9.9757552e-01  9.9572134e-01]\n",
      "  [ 9.9733776e-01  9.6893448e-01  9.9693650e-01 ...  9.9654895e-01\n",
      "    9.9757206e-01  9.9571770e-01]\n",
      "  ...\n",
      "  [ 9.9733055e-01  9.6911198e-01  9.9693495e-01 ...  9.9652869e-01\n",
      "    9.9755704e-01  9.9570930e-01]\n",
      "  [ 9.9732673e-01  9.6903634e-01  9.9693489e-01 ...  9.9652100e-01\n",
      "    9.9755102e-01  9.9570709e-01]\n",
      "  [ 9.9732226e-01  9.6893001e-01  9.9693471e-01 ...  9.9651188e-01\n",
      "    9.9754399e-01  9.9570459e-01]]]\n",
      "Target: \n",
      " [[[1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  ...\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]]\n",
      "\n",
      " [[1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  ...\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]\n",
      "  [1. 0. 1. ... 1. 1. 1.]]\n",
      "\n",
      " [[1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  ...\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]]\n",
      "\n",
      " [[1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  ...\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]]\n",
      "\n",
      " [[1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  ...\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]\n",
      "  [1. 1. 1. ... 1. 1. 1.]]]\n",
      "PRED: \n",
      " [[[ 7.6029545e-01 -1.4468492e-03  7.6084030e-01 ...  7.6010329e-01\n",
      "    7.6094246e-01  7.6058626e-01]\n",
      "  [ 7.6109439e-01 -1.9602920e-03  7.6097167e-01 ...  7.6014847e-01\n",
      "    7.6079619e-01  7.6093137e-01]\n",
      "  [ 7.6115900e-01 -1.2136676e-03  7.6076800e-01 ...  7.6034170e-01\n",
      "    7.6072544e-01  7.6090550e-01]\n",
      "  ...\n",
      "  [ 7.6064360e-01 -1.8705645e-03  7.6038176e-01 ...  7.5960511e-01\n",
      "    7.6022577e-01  7.6044685e-01]\n",
      "  [ 7.6118124e-01 -2.4772866e-04  7.6078999e-01 ...  7.6058942e-01\n",
      "    7.6078701e-01  7.6092285e-01]\n",
      "  [ 7.5845993e-01 -1.4439871e-03  7.5849932e-01 ...  7.5828040e-01\n",
      "    7.5926584e-01  7.5935704e-01]]\n",
      "\n",
      " [[ 7.6122063e-01  6.3528507e-03  7.6082742e-01 ...  7.6066881e-01\n",
      "    7.6087272e-01  7.6095623e-01]\n",
      "  [ 7.6115942e-01  1.3422049e-02  7.6080883e-01 ...  7.6053375e-01\n",
      "    7.6077002e-01  7.6102978e-01]\n",
      "  [ 7.5818282e-01  1.0380049e-02  7.5526607e-01 ...  7.5770473e-01\n",
      "    7.5905991e-01  7.5785750e-01]\n",
      "  ...\n",
      "  [ 7.5998229e-01  1.9955332e-03  7.5820178e-01 ...  7.5881916e-01\n",
      "    7.6019019e-01  7.5899267e-01]\n",
      "  [ 7.5998670e-01  1.9416373e-03  7.5821197e-01 ...  7.5881886e-01\n",
      "    7.6019818e-01  7.5899082e-01]\n",
      "  [ 7.5998789e-01  1.9185289e-03  7.5821471e-01 ...  7.5881904e-01\n",
      "    7.6020128e-01  7.5898844e-01]]\n",
      "\n",
      " [[ 7.6115650e-01  7.5763834e-01  7.6099694e-01 ...  7.6094210e-01\n",
      "    7.6116240e-01  7.6113707e-01]\n",
      "  [ 7.6115781e-01  7.5764185e-01  7.6099801e-01 ...  7.6094323e-01\n",
      "    7.6116276e-01  7.6113784e-01]\n",
      "  [ 7.6115805e-01  7.5764304e-01  7.6099831e-01 ...  7.6094359e-01\n",
      "    7.6116312e-01  7.6113808e-01]\n",
      "  ...\n",
      "  [ 7.6115841e-01  7.5764525e-01  7.6099861e-01 ...  7.6094401e-01\n",
      "    7.6116329e-01  7.6113844e-01]\n",
      "  [ 7.6115841e-01  7.5764537e-01  7.6099861e-01 ...  7.6094401e-01\n",
      "    7.6116329e-01  7.6113844e-01]\n",
      "  [ 7.6115841e-01  7.5764555e-01  7.6099861e-01 ...  7.6094401e-01\n",
      "    7.6116329e-01  7.6113844e-01]]\n",
      "\n",
      " [[ 7.6115650e-01  7.5763834e-01  7.6099694e-01 ...  7.6094210e-01\n",
      "    7.6116240e-01  7.6113707e-01]\n",
      "  [ 7.6115781e-01  7.5764185e-01  7.6099801e-01 ...  7.6094323e-01\n",
      "    7.6116276e-01  7.6113784e-01]\n",
      "  [ 7.6115805e-01  7.5764316e-01  7.6099831e-01 ...  7.6094359e-01\n",
      "    7.6116312e-01  7.6113808e-01]\n",
      "  ...\n",
      "  [ 7.6115841e-01  7.5764525e-01  7.6099861e-01 ...  7.6094401e-01\n",
      "    7.6116329e-01  7.6113844e-01]\n",
      "  [ 7.6115841e-01  7.5764537e-01  7.6099861e-01 ...  7.6094401e-01\n",
      "    7.6116329e-01  7.6113844e-01]\n",
      "  [ 7.6115841e-01  7.5764555e-01  7.6099861e-01 ...  7.6094401e-01\n",
      "    7.6116329e-01  7.6113844e-01]]\n",
      "\n",
      " [[ 7.6047432e-01  7.4800640e-01  7.6030469e-01 ...  7.6014292e-01\n",
      "    7.6057518e-01  7.5979573e-01]\n",
      "  [ 7.6047456e-01  7.4816978e-01  7.6030564e-01 ...  7.6014256e-01\n",
      "    7.6057410e-01  7.5979149e-01]\n",
      "  [ 7.6047385e-01  7.4823570e-01  7.6030463e-01 ...  7.6014107e-01\n",
      "    7.6057261e-01  7.5978988e-01]\n",
      "  ...\n",
      "  [ 7.6047081e-01  7.4831378e-01  7.6030397e-01 ...  7.6013255e-01\n",
      "    7.6056629e-01  7.5978631e-01]\n",
      "  [ 7.6046914e-01  7.4828058e-01  7.6030391e-01 ...  7.6012927e-01\n",
      "    7.6056379e-01  7.5978535e-01]\n",
      "  [ 7.6046735e-01  7.4823374e-01  7.6030380e-01 ...  7.6012546e-01\n",
      "    7.6056075e-01  7.5978422e-01]]]\n",
      "=======================================================\n",
      "\n",
      "step: 1000 in iter: 5/5...  loss: 0.0083...  0.0050 sec/batch\n"
     ]
    }
   ],
   "source": [
    "model_path = './models'\n",
    "if os.path.exists(model_path) is False:\n",
    "    os.makedirs(model_path)\n",
    "input_data, target_data = batch_generator(x_input.values, y_label.values)\n",
    "model = N2N_LSTM()\n",
    "model.train(input_data, \n",
    "            target_data,\n",
    "            iters=2000,\n",
    "            save_path=model_path,\n",
    "            save_every_n=1000,\n",
    "            log_every_n =200\n",
    "            )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(5, 10, 71)"
      ]
     },
     "execution_count": 80,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "input_data[0].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1059"
      ]
     },
     "execution_count": 81,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(input_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "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": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}